Chromatography Encyclopedia of Separation Science Elsevi

2012 III / AMINO ACIDS / Thin-Layer (Planar) Chromatography
acetyl-cysteinyl and with OPA/N-L(D)-isobutyrylcys- See also: II/Chromatography: Liquid: Derivatization;

teinyl AA derivatives gave excellent resolution of en- Mechanisms: Reversed Phase.
antiomers. Consequently, the CDR technique is the
primary importance in a number of practical applica-
tions of the separation of enantiomeric AAs. The
Further Reading
interaction of AAs with the enantiomerically pure Blau K and Halket J (eds) (1993) Handboook of Deriva-
reagents takes place at ambient temperature, without tives for Chromatography. Chichester: John Wiley.
racemization, resulting in the formation of stable BruK ckner H, Langer M, LuK pke M, Westhauser T and Godel
diastereomer derivatives. H (1995) Liquid chromatographic determination of
amino acid enatiomers by derivatization with o-phthal-
Online LC-MS dialdehyde and chiral thiols. Journal of Chromatogra-
In the case of AAs, thermospray ionization has been phy 697: 229.
displaced by the milder techniques of electrospray Deyl Z, Hyanek J and Horakova M (1986) ProRling of
amino acids in body Suids and tissues by means of liquid
(ES) and atmospheric pressure chemical ionization
chromatography. Journal of Chromatography 379: 177.
(APCI), converting analyte molecules without frag- Grunau JA and Swiader JM (1992) Chromatography of 99
mentation into ions. The analyte should contain the amino acids and other ninhydrin reactive compounds in
AAs in a stable form: either in the free condition or in the Pickering lithium gradient system. Journal of
the form of stable derivatives, such as phenyl- Chromatography 594: 165.
thiohydantoins (PTH) or PTCs. SigniRcantly reduced McClung G and Frankenberger WT Jr (1988) Comparison
Sow rates are essential (100}300 nL min\1) for stable of reversed-phase high performance liquid chromato-
ES and APCI operation. In automated Edman micro- graphic methods for precolumn-derivatized amino
sequencing, the ES-MS of PTH derivatives. The pro- acids. Journal of Liquid Chromatography 11: 613.
tonated molecules were measured with a linear re- MolnaH r-Perl I (1998) Amino acids. In: Deyl Z, Tagliaro
sponse in the 50}1000 fmol level. F and Teserova E (eds) Advanced Chromatographic and
Electromigration Methods in BioSciences. Amsterdam:
Elsevier.
Future Trends Snyder LR, Kirkland JJ and Glajch JL (1997) Practical
Efforts are needed to extend the life time, plate HPLC Method Development. New York: Wiley Inter-
number and reproducibility of columns, and to science.
Spackman DH, Stein WH and Moore S (1958) Automatic
standardize testing methods. The extended use of
recording apparatus for use in the chromatography of
thermostated columns is desirable in order to obtain amino acids. Analytical Chemistry 30: 1190.
reproducibility in absolute and relative retention Zhou J, Hefta S and Lee TD (1997) High sensitivity analy-
times. LC-MS will be more widely used in laborator- sis of phenylthiohydantoin amino acid derivatives
ies as the cost of these instruments falls to the level of by electrospray mass spectrometry. Journal of the
GC-MS, and/or an all-purpose interface becomes American Chemical Society of Mass Spectrometry 8:
available. 1165.
Thin-Layer (Planar) Chromatography
R. Bhushan, University of Roorkee, suitable for use with strong corrosive reagents and
Roorkee, India one can perform many kinds of chemical reactions on
J. Martens, Universitat Oldenburg, Oldenburg, the plate, both from the points of view of detecting
Germany and locating the spot and of achieving improved
Copyright ^ 2000 Academic Press separation. Certain groups of interest can be chemic-
ally bonded to the reactive groups of support mater-
ial, e.g. silanization for reversed-phase studies. Im-
pregnation of the adsorbent with a variety of reagents
Introduction adds an additional feature for inSuencing the adsorp-
Thin-layer chromatography (TLC) is a simple and tion characteristics without covalently affecting the
inexpensive technique permitting a number of sam- inert character of the adsorbent. TLC is also success-
ples to be handled simultaneously, thus yielding ful in providing direct resolution of enantiomers of
a higher precision than sequential analysis. The inert a variety of compounds by the proper manipulat-
character of the thin-layer material makes it ideally ion of the support material. The analysis of amino
III / AMINO ACIDS / Thin-Layer (Planar) Chromatography 2013
acids and derivatives and the resolution of enantio- Preparation of Thin-layer Plates
mers of amino acids by TLC techniques using a wide
variety of adsorbents and impregnating agents, Most thin-layer work is done on layers prepared from
the possibility of obtaining relationships between the water-based slurries of the adsorbents. Even with the
chromatographic behaviour and chemical structure same amount and type of binder, the amount of water
and the many practical applications drawn from used for a given slurry varies with kinds and brands of
the literature are described in detail in the following adsorbents. For example, in the case of cellulose the
sections. amount of powder to be mixed with water varies
depending on the supplier: Serva, Camag and What-
Adsorbents and Thin Layers
man recommend the use of 60}80 mL, 65 mL and
A variety of adsorbents such as silica gel, alumina, 25 mL water for 10 g of their cellulose powders,
polyamide and cellulose are available commercially respectively. These slurries may be prepared by shak-
and are used to make thin layers for TLC. Alumina ing a stoppered Sask or by homogenizing for a few
and silica gel are used with or without a suitable seconds with a mechanical mixer. On the other hand,
binder such as gypsum or starch. Mixtures of two for the preparation of an aluminium oxide slurry
adsorbents or adsorbents impregnated with certain (acidic, basic or neutral), it is recommended that 35 g
reagents such as 8-hydroxyquinoline or different of aluminium oxide is used with 40 mL water for
metal ions have also been used successfully to im- spreading equipment, and 6 g of adsorbent in 15 mL
prove resolution. ethanol}water (9:1) mixture for pouring directly on
Cellulose layers have several advantages: they are to the plate without a spreading apparatus. A
stable, they can be used with various speciRc reagents slurry of 120 g of alumina G in 110 mL of water
and they give reproducible data. They are particularly has been used successfully to make 1 mm-thick
recommended for quantitative evaluation by den- layers for preparative TLC. In general, cellulose pow-
sitometry. The drawbacks of cellulose layers are that ders contain impurities that are soluble in water
corrosive reagents cannot be used and the sensitivities or organic solvents, and these should be removed
of detection reactions of certain amino acids are by washing the cellulose several times with acetic
lower than on silica gel layers. acid (0.1 mol L\1), methanol and acetone and dry-
The best known and most widely used adsorbents ing before use. The layer is made by turbo-mixing
for TLC purposes are from Merck, but other products MN (Machery-Nagel) cellulose 300 (15 g) for 10 min
can be used satisfactorily. Pre-coated plates are wide- in distilled water (90 mL) and then spreading it to
ly available and increasingly used for the investiga- give a 0.25 mm thick layer. The layers are left over-
tion of amino acids and their derivatives. For night to dry.
example, ready-made cellulose layers from Macherey- A slurry of silica gel G (50 g) in distilled water
Nagel (Germany) containing MN cellulose-300 in (100 mL) is prepared and spread with the help of
appropriately bound form are one of the best-known a Stahl-type applicator on Rve glass plates of
products. Chiralplate from the same Rrm and Chir 20;20 cm to obtain 0.5 mm thick layers. The plates
from Merck, for the separation of enantiomers of are allowed to set properly at room temperature and
amino acids and their various derivatives, contain then dried (activated) in an oven at an appropriate
a coating of reversed-phase silica gel impregnated temperature (60}903C) for 6 h or overnight. The
with a chiral selector and copper ions. Using plates are cooled to room temperature before ap-
home-made thin-layer plates is possible and it is plying the samples.
recommended that one should not change the The same method has been used successfully to
brand of adsorbent during a particular set of ex- prepare plates with silica gel, silica gel polyamide,
periments. cellulose and these adsorbents impregnated with
Application of mixed layers of cellulose and the a variety of reagents including di-(2-ethylhexyl)
ion exchanger Amberlite CG-120 and a double layer orthophosphoric acid (HDEHP), tri-octyl-phosphine
consisting of a 2 cm band of cellulose#cation ex- oxide (TOPO), 8-hydroxyquinoline, dibenzoyl meth-
changer (45#5 g) in aqueous CM-cellulose (0.05%), ane and several metal salts. Brucine and tartaric acid
with the remaining portion of the layer prepared are also mixed in slurries of silica gel as impregnating
from cellulose SF suspension, have also been effec- reagents to resolve enantiomers of amino acids and
tively used. A newly synthesized support named their PTH derivatives. Mixtures of H2O}EtOH and
aminoplast comparable with that of starch and cellu- other organic solvents can also be used, depending on
lose has been reported. Nevertheless, silica gel con- the nature of the impregnating reagents. Citrate and
tinues to be the most widely used and successful phosphate buffers have also been used for slurrying
material. silica gel in place of water. It is customary to use 0.25
or 0.50 mm thick layers in activated form, but for The same method is adopted for both one- and
preparative purposes 1}2 mm layers are best. two-dimensional modes. The locating reagent is used
after the second run, and a more polar solvent is
Development of Chromatograms
generally used to develop the chromatogram in the
Standard solutions of amino acids are prepared in second dimension.
a suitable solvent such as 70% EtOH or 0.1 mol L\1
HCl in 95% ethanol. These solutions are generally
applied as tight spots, 1}2 cm from the bottom of
Separation of Amino Acids
each layer, using a glass capillary or Hamilton syr- Silica gel and cellulose are the commonest adsorbents
inge. In the beginning a higher concentration, e.g. for one- or two-dimensional resolution of amino
500 ng or more, is applied; however, the detection acids. These have been used as such (untreated) or
limits are determined for the system developed by impregnated with some other reagent employing
repeating the experiment with lower concentrations. a large number of solvents. Some of the successful
The chromatograms are generally developed in rec- systems for one- and two-dimensional resolution of
tangular glass chambers, which should be paper-lined amino acids are given in Table 1 and Table 2.
for good chamber saturation and pre-equilibrated for Table 3 shows a comparative account of the separ-
20}30 min with solvent before use. The time taken ation of amino acids (hRF values) on silica gel,
depends on several factors such as the nature of the cellulose and ion exchange thin layers using n-bu-
adsorbent, the solvent system and the temperature. tanol}acetic acid}water (3 : 1 : 1). The data are of
The developed chromatograms are dried in an oven great value for separating and detecting amino acids
between 60 and 1003C, and the cooled plates are by one-dimensional TLC and based on it the
usually sprayed with ninhydrin reagent. Heating at amino acids have been grouped for the separation
90}1003C for 5}10 min produces blue to purple of 18-component mixtures (separation I) and essen-
zones of all amino acids except proline (yellow spot). tial amino acid mixtures (separation II) by calculating
Table 1 Solvent systems for TLC of amino acids on silica gel
Solvent system Ratio v/v
Silica gel
96% Ethanol}water 7:3
n -Propanol}water 7:3
n -Butanol}acetic acid}water 4:1:1
n -Propanol}34% NH4OH 7:3
n -Propanol}water 1:1
Phenol}water 3:1
Propan-2-ol}water 7:3
Butyl acetate}methanol}acetic acid}pyridine 20 : 20 : 5: 5
n -Butanol}formic acid}ethanol 3:1:1
n -Butanol}acetic acid}chloroform 3:1:1
n -BuOH}HOAc}EtOAc}H2O 50 : 20 : 30 : 20
n -Propanol}H2O 7:3
n -BuOH}H2O}HOAc 40 : 7 : 5
Cellulose
Propan-2-ol}butanone}1 mol L\1 HCl 60 : 15 : 25
2-Methylpropan-2-ol}butanone}acetone}methanol 20 : 1 : 14 : 5
Butanol}acetic acid}H2O 4:1:5
Methanol}H2O}pyridine 20 : 5 : 1
Propanol}8.8% NH3 4:1
Chloroform}MeOH}17% NH3 20 : 20 : 9
Butanol}acetone}Et2NH}H2O 10 : 10 : 2 : 5
Phenol}water 3:1
Ethyl acetate}pyridine}HOAc}H2O 5:5:1:3
n -Butanol}acetic acid}H2O}EtOH 10 : 1 : 3 : 0 : 3 or
4 : 1 : 10 : 1
Ethanol}conc. HCl 30 : 1
n -BuOH}HOAc}H2O 4:1:1
Pyridine}acetone}NH4OH}H2O 26 : 17 : 5 : 12
Propan-2-ol}formic acid}H2O 25 : 3 : 2
Table 2 Solvent systems for two-dimensional TLC
First Second
Silica gel
n -Butanol}HOAc}H2O (4 : 1 : 5, v/v, upper phase) Phenol}water (15 : 1, w/w)
Chloroform}MeOH}17% NH3 (2 : 2 : 1) Phenol}H2O (3 : 1)
n -Butanol}HOAc}H2O (4 : 1 : 5, upper phase) CHCl3}MeOH}17% NH3 (2 : 2 : 1)
Butanone}pyridine}H2O}HOAc (70 : 15 : 15 : 2) CHCl3}MeOH}17% NH3 (2 : 2 : 1)
Cellulose
Propanol}HCOOH}H2O (40 : 2 : 10) t -Butanol}methylethyl ketone}0.88 NH3}H2O (50 : 30 : 10 : 10, v/v)
Propan-2-ol}butan-2-ol}1 mol L\1 HCl (60 : 15 : 25 by vol.) 2-Methyl propanol}butan-2-one}acetone}MeOH}H2O}(0.88)
NH3 (10 : 4 : 2 : 1 : 3 :1) or 2-methylpropanol}butanone}propanone}
methanol}H2O (40 : 20 : 2 : 1 : 14.5, v/v)
the resolution possibilities of each pair of acids been found to be the same. Sorbents with ion ex-
(Table 4). change properties such as diethylaminoethyl (DEAE)}
Amino acids chromatographed in the presence of cellulose have also been used as the stationary phase
trichloroacetic acid (used in deproteinizing serum for TLC separation of the main protein amino acids
samples) show anomalous behaviour, and this inter- with n-butanol}acetic acid}water (5 : 1 : 6, upper
ference can be almost completely removed by phase) and pyridine} water (4 : 1) in one- and two-
predevelopment (twice) in ether saturated with for- dimensional modes.
mic acid. The migration sequences for the separation
of 18 amino acids on reversed-phase thin layers in- Locating the spots of amino acids After drying the
cluding C18 chemically bonded silica gel and on cellu- chromatogram it may be viewed under ultraviolet
lose in n-propanol-H2O (7 : 3, v/v) have generally light if the absorbent had a Suorescent indicator, or
Table 3 hRF (RF;100) values for amino acids on different layers
A B C D E
FXA FXB FXC
Ala 41.9 29.0 32.4 28.8 50.9 51.2 53.6

Ser 26.9 16.1 26.4 24.1 67.1 64.1 67.1
Tyr 50.0 36.1 49.4 45.9 11.9 13.9 15.5
Glu 34.4 22.6 30.0 28.2 34.5 29.4 30.6
Asp 26.3 14.8 25.3 21.8 71.5 68.2 68.6
Arg 25.6 11.0 12.9 10.0 1.8 2.2 2.2
Gly 29.4 14.8 25.9 23.5 55.6 52.4 53.6
Leu 75.0 63.9 51.8 48.8 21.8 17.8 19.4
Ile 73.1 60.0 49.4 47.1 27.8 22.2 23.3
Try 55.6 36.1 54.1 51.8 1.8 2.2 2.2
Met 41.0 22.5 47.3 43.5 28.0 27.2 25.0
Val 63.1 48.4 43.5 41.2 42.5 35.0 34.4
Lys 18.1 7.1 10.0 7.1 7.5 5.0 5.6
His 20.0 7.1 11.7 7.1 10.6 8.9 10.0
Phe 67.5 54.8 52.4 50.0 14.4 11.1 11.7
Thr 32.5 21.3 30.0 27.6 67.1 60.0 57.2
Cys 6.9 3.2 14.1 7.1 55.9 50.0 57.9
Pro 43.8 33.5 24.1 21.2
Time for
17 cm (h) 7 11 4.5 7.5 6.5 6 2
A, Baker Flex cellulose sheets; B, Baker Flex microcrystalline cellulose sheets; C,

Whatman K6 silica gel plates; D, Whatman high performance silica gel plates; E, Fixion
ion exchange sheets (Na# form); FXA, no prior treatment; FXB, layer pre-equilibrated with
equilibration buffer for 16 h; FXC layer pre-equilibrated as for FXB but at 453C. Solvent for
A}D, 2-butanol}acetic acid}water (3 : 1 : 1); solvent for E and run buffer, 84 g citric acid
#16 g NaOH#5.8 g NaCl#54 g ethylene glycol#4 mL conc. HCl (pH 3.3); solvent
equilibriation buffer, run buffer diluted 30 times (pH 3.8).
Table 4 Group separation of amino acids
System as in Group Amino acids resolved

Table 3
A I Leu, Phe, Try, Ala, Glu, Ser, Lys, Cys, Tyr

II Leu, Phe, Try, Thr, Lys
B I Leu, Phe, Tyr, Val, Glu, Asp, Lys
II Leu, Phe, Val, Try, Thr, Lys
C I Try, Ile, Val, Ala, Ser, Cys, Lys
II Try, Ile, Val, Thr, Lys
D I Try, Ile, Val, Ser, Glu, Arg, Lys
II Try, Ile, Val, Thr, Lys
FXA I Thr, Gly, Val, Glu, Met, Leu, Phe, His, Lys, Arg
II Thr, Val, Met, Leu, Phe, His, Lys, Try
FXB I Asp, Thr, Gly, Val, Met, Leu, Thr, His, Lys, Try
FXC I Asp, Thr, Gly, Val, Met, Leu, Thr, His, Lys, Try
Group I, 18-component mixture of amino acids; Group II, Mixture of essential amino acids.
the compounds } such as dansyl amino acids } Suor- (500 mL) with acetone. Portions of this solution
esce. Solvent fronts indicate regularity of solvent are taken and solid ninhydrin is added to give
Sow. Ninhydrin is the most commonly used reagent a Rnal concentration of 0.2% w/v. The chromato-
for the detection of amino acids, and a very large gram is sprayed and heated at 603C for 15 min.
number of ninhydrin reagent compositions have been The results are noted immediately and again after
reported in the literature. The reagent may be made 24 h, at room temperature. Alternatively, the layer
slightly acidic with a weak acid following the use of is impregnated thoroughly with the reagent and
an alkaline solvent and vice versa. Constancy of col- the colours are allowed to develop in the dark at
our formed may be attained by the addition of com- room temperature for 24 h. This reagent gives
plex-forming cations (Cu2#, Cd2# or Ca2#) and spe- permanent colours, mainly red but yellow for pro-
ciRc colours may be produced by the addition of bases line. Sensitivity is 0.5 nmol.
such as collidine or benzylamine. Some of the ninhyd- 5. Ninhydrin (1.0 g) in absolute ethanol (700 mL),
rin compositions and their applications are described 2,4,6-collidine (29 mL), and acetic acid (210 mL)
below. has been used for spraying on solvent-free cellu-
lose layers. The chromatogram is then dried for
1. A solution of ninhydrin (0.2% w/v in acetone) is 20 min at 903C.
prepared with the addition of a few drops of 6. Development of ion exchange resin layers in nin-
collidine or glacial acetic acid. The chromatogram hydrin (1% w/v) in acetone containing collidine
is dipped or sprayed with the solution and dried at (10% w/v) at room temperature for 24 h, or at
603C for about 20 min or at 1003C for 5}10 min. 703C for 10 min has also been recommended.
Excessive heating causes a dark background. Most 7. Spray of ninhydrin (0.1% or 0.2% w/v in acetone
amino acids give a violet colour, while aspartic on chromatograms followed by heating at 60 or
acid gives bluish-red, and proline and hydroxypro- 903C for 10}20 min has also been used.
line give a yellow colour; the sensitivity limit is 1 g. 8. Polychromatic reagent consists of Rrstly, ninhyd-
2. Ninhydrin (0.3 g) in n-butanol (100 mL) contain- rin (0.2% w/v) in ethanol (50 mL)#acetic acid
ing acetic acid (3 mL) is sprayed on a dried, sol- (10 mL)#2,4,5-collidine (2 mL) and secondly,
vent-free layer, which is then heated for 30 min at a solution of copper nitrate (1.0% w/v) in absolute
603C or for 10 min 1103C. Detection limits range ethanol. The two solutions are mixed in a ratio of
from 0.001 g for alanine to 0.1 g for proline and 50 : 3 before use. Replacement of ethanol by
aspartic acid. methanol also gives polychromatic amino acid de-
3. Ninhydrin (0.3 g), glacial acetic acid (20 mL) and tection by joint application of ninhydrin and pri-
collidine (5 mL) are made up to 100 mL with mary, secondary or tertiary amines. The layers are
ethanol or ninhydrin (0.1%, w/v) in acetone}gla- Rrst sprayed with diethylamine, dried for 3 min at
cial acetic acid}collidine (100 : 30 : 4%). 1103C, cooled, and then sprayed with 0.2% w/v
4. A solution of cadmium acetate (0.5 g) in water methanolic ninhydrin and heated for 10 min at
(50 mL) and glacial acetic acid (10 mL) is made up 1103C, when the spots of amino acids appear on
a pale blue background. Use of ninhydrin (0.27 g), Table 5 Detection reactions for specific amino acids
isatin (0.13 g), and triethylamine (2 mL) in meth-
anol (100 mL) gives spots of amino acids on a yel- Amino acid Reagent
low background. Arg 8-Hydroxyquinoline
Arg -Naphthol, urea, Br2
Several other reactions have also been used for the Asp Ninhydrin, borate solution, HCl
detection of speciRc amino acids (Table 5). Oxalic Cys, Met NaN3, iodine
acid (ethanolic 1.25% w/v), dithio-oxamide Gly o -Phthalaldehyde, KOH
(ethanolic-saturated) and dithizone followed by nin- His Sulfanilic acid
hydrin have been used to aid identiRcation and detec- Ser, Thr, Tyr Sodium metaperiodate, Nessler reagent
Try p-Dimethylaminobenzaldehyde
tion of amino acids with various speciRc colours.
Acetylacetone}formaldehyde gives yellow spots
under UV light. Using isatin}ninhydrin (5 : 2) in aq. TLC has been used for the identiRcation of PTH
butanol or modifying ninhydrin detection reagent by amino acids since Edman and Begg used it in their
addition of D-camphor, and various acids improves classical work describing the automatic sequencer.
identiRcation of amino acids. Spraying of layers with Various TLC systems with different kinds of adsor-
1,3-indanedione or o-mercaptobenzoic acid prior to bents, such as alumina, silica gel and polyamide, have
ninhydrin improves sensitivity limits and colour dif- been reported. The results of some TLC systems used
ferentiation. 3,5-Dinitrobenzoyl chloride can be used for resolution and identiRcation of PTH amino acids
to detect amino acids at a 3}4 g level, and synchro- are given below.
nization of timing is achieved by coupling pneumatic Two-dimensional TLC has been carried out using
nebulization with optical Rbre-based detection in plates coated with polyamide containing three Suor-
a chemiluminescence TLC system to detect dansyl- escent additives when all PTH amino acids show
amino acids. Chromatograms sprayed with ninhydrin coloured spots under UV light. About 0.1 nmol of
(0.3 g ninhydrin in 100 mL n-butanol plus 3 mL of PTH amino acid can be detected. Typical results are
glacial acetic acid), air-dried for 5 s, resprayed and given in Table 6. A compilation of solvent mixtures
heated in an oven at 1103C for 10 min gives the best useful in the TLC of PTH amino acids on various
sensitivity, stability and colour differentiation com- supports is given in Table 7.
pared with different recipes of ninhydrin and Suor- Resolution and identiRcation of PTH amino acids
escamine sprays. on silica or polyamide layers, as discussed above, do
not discriminate between derivatives of Leu/Ile and
Separation of Amino Acid Derivatives cannot resolve complex mixtures without two-
dimensional chromatography. DifRculties in resolv-
Separation and identiRcation of derivatives of amino ing combinations of PTH Phe/Val/Met/Thr and PTH
acids such as dinitrophenyl (DNP), PTH, dansyl and di- Asp and Glu are also observed. Use of chloro-
methylamino azobenzene isothiocyanate (DABITC), form}acetic acid (27 : 3, v/v) and chloroform}meth-
is very important, particularly in the primary struc- anol (30 : 4, v/v) has been found to be extremely
ture determination of peptides and proteins. The satisfactory for discriminating between PTH Asp and
preparation of PTH, dansyl, and DNP amino acids, PTH Glu, as the difference in their hRF values is
and the methods for their identiRcation after separ- around 10 units. The difRculties in resolving and
ation from the N-terminal of peptides and proteins, identifying various combinations of PTH amino acids
are available in literature. can be overcome by the use of certain solvent systems,
given in Table 7.
PTH Amino Acids
The PTH amino acids are sensitive to light, and op- Detection of PTH amino acids The methods of de-
tically active derivatives racemize easily. Both manual tection include Rrstly, spraying a dilute solution of
and automated, and liquid-phase and solid-phase Ed- Suorescein on a plain layer of silica gel when the spots
man degradation methods (coupling of the NH2 are visible as dark areas against a yellow background
group of an amino acid at the N-terminal end of in UV light; secondly, exposing the dried chromato-
a polypeptide or a free molecule with phenyl iso- grams to iodine vapours to locate the spots as light
thiocyanate) are currently used for small and large brown compact zones; and thirdly, use of iod-
polypeptides to establish their primary structure. An ine}azide solution when bleached spots on a light
automated sequencer can deliver several PTH amino brown background are observed. The iodine azide
acids in 24 h and these are required to be identiRed method is considered less sensitive and causes difRcul-
rapidly to match the output. ties in demarcating the exact spots and measuring the
Table 6 Characteristic colours of PTH amino acids on polyamide plates containing

mixed fluorescent additive 3
PTH amino acid Colour after
Second treatment Alkaline treatment
Valine Red Red

Proline Red Red
Alanine Red Red
Glycinea Red Brownish red
Serine Red Brownish red (blue)
Asparaginea Red Greenish brown (bluish green)b
Aspartic acid Red Brownish red (dark brown)
Methioninea Red Brownish red
Leucine Red Brownish red
Isoleucine Red Red
Lysine Red Red
Tyrosinea Red Red (bluish green)b
Threoninea Red Bluish green (blue)
Glutaminea Red Greenish brown (white yellow)
Glutamic acid Red Red
Phenylalaninea Red Greenish red (white blue)b
Tryptophana Red Greenish red (white blue)b
Histidinea Red Blue (light blue)b
Argininea Red Purple (blue)b
Cysteic acid Red Brownish red (dark brown)
a
Spots appear yellow, except glycine (pink); bFluorescent. Solvents : toulene-n}pen-
tane}acetic acid (6 : 3 : 2, v/v) and acetic acid}water (1 : 3, v/v) for first and second
dimension, respectively. Alkanine treatment : spray 0.05 mol L\1 NaOH in methanol}
water (1 : 1, v/v), heating at 1503C for 30 min, UV.
correct RF. Characteristic changes in the colours of plate with mixed Suorescent additives is used
some derivatives are observed by heating the plate (Table 6). A rapid colour-coded system due to nin-
after spraying with an alkaline solution when the hydrin spray is mentioned in Table 8; the colours
produced allow easy identiRcation of those amino
Table 7 Various solvent systems for TLC of PTH amino acids acids that have nearly identical RF values, for
example, Lys and Ser degradation products, Ala/Met/
Ratio Phe, and Tyr/Thr. The method is signiRcant because
it gives positive identiRcations of PTH Ser/Lys/
Polyamide
n -Heptane-n -BuOH}HOAc 40 : 30 : 9 Glu/Asp and their respective amides, which cannot be
Toluene}n-pentane}HOAc 60 : 30 : 35 identiRed by gas chromatography (GC).
Ethylene chloride}HOAc 90 : 16
Toluene}n-pentane}HOAc 60 : 30 : 35 Dansyl Amino Acids
EtOAc}n -BuOH}HOAc 35 : 10 : 1
Derivatization of free amino group of amino acids
n -BuOH}MeOH}HOAc (#30 mg butyl 19 : 20 : 1
fluorescent reagent per litre) with 5-methylaminonapthalene-1-sulfonyl (dansyl)
chloride has become increasingly popular for N-ter-
Silica gel
Heptane}CH2Cl2}propionic acid 45 : 25 : 30
minal determinations in proteins and for manual
Xylene}MeOH 80 : 10 Edman degradation. In addition, dansylation has also
CHCl3}EtOH and 98 : 2 been used as one of the most sensitive methods for
CHCl3}EtOH}MeOH (in the same direction) 89.25 : 0.75 : 10 quantitative amino acid analysis.
CHCl3}n -butyl acetate 90 : 10 Two-dimensional TLC on polyamide sheets using
Diisopropyl ether}EtOH 95 : 5
CH2Cl2}EtOH}HOAc (or on cellulose) 90 : 8 : 2
water}formic acid (200 : 3, v/v) for the Rrst-direction
Petroleum ether (60}803)}acetic acid 25 : 3 run and benzene}acetic acid (9 : 1, v/v) for develop-
n -Hexane}n -butanol 29 : 1 ment at right angles to the Rrst run has mostly been
n -Hexane}n -butyl acetate 4:1 employed in conjunction with the Edman dansyl tech-
Pyridine}benzene 2.5 : 20 nique for sequencing peptides. These solvents cannot
MeOH}CCl4 1 : 20
Acetone}dichloromethane 0.3 : 8
resolve Dns-Glu/Asp, Dns-Thr/Ser, and -Dns-Lys/
-Dns-Lys/Arg/His. However, a third run in ethyl
Table 8 Characteristic colours of PTH amino acids following ninhydrin application
PTH derivative Colour properties NH4OH colour change
Proline UV, colourless Light blue after heating

Alanine Purple Deeper colour
Glycine Orange
Serine UV, purple
Serine breakdown Faint orange Weak red
Asparagine Yellow More intense
Carboxymethylcysteine UV, purple
Methioninesulfone Light tan
Methionine Faint tan
Lysine Very faint pink Weak blue after heating
Tyrosine UV, yellow before spray Intense yellow
Threonine Colourless Light tan
Glutamine Dark green Dark blue
Phenylalanine UV, colourless Faint yellow
Tryptophan UV, yellow before spray Deep yellow
Aspartic acid UV, pink Darker
Glutamic acid Grey Dark blue
Silica gel plates, without fluorescent indicator, developed in heptane}CH2Cl2}propionic

acid (45 : 25 : 30) and xylene}MeOH (80 : 10), sprayed with iodine}azide and 1.7%
ninhydrin in MeOH}collidine}HOAc (15 : 2 : 5), heated at 903C for 20 min; colour changes
by blowing a saturated ammonia atmosphere over ninhydrin plate.
acetate}acetic acid}methanol (20 : 1 : 1) in the direc- azobenzene thiohydantoin (DABTH) amino acid via
tion of solvent 2 resolves Dns-Glu/Asp, and Dns- a DABTC derivative, in a manner similar to the Ed-
Thr/Ser. A further run in the direction of solvent man method, where a PTH amino acid is obtained by
2 and 3 using 0.05 mol L\1 trisodium phosphate} the reaction of phenylisothiocyanate (PITC). The
ethanol (3 : 1, v/v) resolves the monosubstituted basic presence of excess free amino acid does not, in any
Dns amino acids. Use of molarity ammonia}ethanol case, interfere with the analysis.
(1 : 1, v/v) as a third solvent for two-dimensional Two-dimensional TLC on polyamide sheets by as-
chromatograms, for the separation of basic dansyl cending solvent Sow is used to identify all DABTH
amino acids in particular, has been effective. Most of amino acids except DABTH-Ile/Leu. No phase equi-
the TLC systems reported up to 1978 required more librium is necessary; H2O}acetic acid (2 : 1, v/v) is
than two runs for complete resolution of all Dns used for the Rrst dimension and toulene}n-
amino acids. A few solvent systems to yield separ- hexane}acetic acid (2 : 1 : 1, v/v) is used for the sec-
ations of basic, acidic and hydroxyl derivatives in the ond. For discrimination between DABTH-Ile/Leu,
presence of other amino acids without resorting to one-dimensional separation on polyamide using for-
the third solvent system and RF values are given in mic acid}ethanol (10 : 9, v/v) or one-dimensional sep-
Table 9. Additionally, a large number of solvent sys- aration on silica gel (Merck) using chloroform}
tems for one- or two-dimensional resolution of dansyl ethanol (100 : 3, v/v) is carried out. The successful
amino acids on silica gel or polyamide have been identiRcation of DABTH amino acids relies on skilful
summarized in Table 10. Bhushan and Reddy re- running of the small polyamide sheet and interpreta-
viewed the TLC of dansyl, and DNP amino acids and tion of the pattern of spots.
evolved several successful and effective solvent sys-
tems for the resolution of almost all the dansyl amino Detection of DABITC derivatives of amino
acids on silica gel plates (Tables 11 and 12). acids The use of DABITC reagent during amino
acid sequencing of proteins has distinct advantages
Detection of dansyl amino acids In all cases, dansyl over the use of dansyl chloride; for example, the
amino acids, being Suorescent, have been detected colour difference between DABITC, DABTC deriva-
under a UV lamp (254 nm). tives and DABTH-amino acids greatly facilitates di-
rect visualization and identiRcation. DABTH amino
DABITC Derivatives of Amino Acids
acids are coloured compounds having absorption
DABITC reacts with the NH2-terminal end of an maxima at 520 nm in acid media ("47 000).
amino acid in basic media to give a 4-dimethylamino Thus, using the visible region, the quantitation and
Table 9 RF values for Dns amino acids in various solvent systems on polyamide sheets
Dns amino acid RF in solvent systems
A B C D E F G H I J
1. Ala 0.53 0.48 0.49 0.69 0.69 0.57 0.81 0.68 0.43 0.76
2. Arg 0.05 0.03 0.03 0.91 0.39 0.09 0.76 0.22 0.01 0.06
3. Asp 0.08 0.07 0.10 0.69 0.38 0.10 0.88 0.37 0.12 0.19
4. Cys 0.03 0.03 0.04 0.19 0.43 0.22 0.78 0.09 0.03 0.06
5. Glu 0.15 0.10 0.15 0.66 0.88 0.02 0.88 0.34 0.05 0.30
6. Gly 0.32 0.21 0.32 0.69 0.63 0.48 0.80 0.48 0.28 0.69
7. His 0.07 0.05 0.13 0.96 0.76 0.32 0.84 0.36 0.06 0.18
8. Ile 0.77 0.54 0.65 0.40 0.57 0.71 0.78 0.76 0.60 0.84
9. Leu 0.70 0.49 0.59 0.34 0.57 0.71 0.78 0.75 0.54 0.80
10. Lys (mono) 0.35 0.21 0.38 0.22 0.09 0.63 0.72 0.58 0.09 0.79
11. Lys (di) 0.53 0.37 0.48 0.78 0.69 0.35 0.82 0.40 0.39 0.76
12. Met 0.52 0.36 0.51 0.43 0.59 0.68 0.80 0.62 0.55 0.81
13. Phe 0.57 0.38 0.53 0.31 0.43 0.68 0.77 0.62 0.51 0.81
14. Pro 0.85 0.66 0.71 0.55 0.74 0.46 0.84 0.75 0.69 0.90
15. Ser 0.12 0.07 0.16 0.81 0.71 0.49 0.82 0.42 0.10 0.44
16. Thr 0.15 0.10 0.26 0.81 0.74 0.57 0.82 0.56 0.16 0.56
17. Tyr 0.63 0.47 0.61 0.00 0.00 0.84 0.73 0.65 0.58 0.91
18. Val 0.72 0.56 0.61 0.47 0.67 0.71 0.81 0.80 0.61 0.88
19. Dns-OH 0.00 0.01 0.00 0.51 0.54 0.16 0.74 0.00 0.04 0.04
20. Dns-NH2 0.51 0.38 0.47 0.71 0.17 0.96 0.49 0.60 0.40 0.91
Solvent systems: A, benzene}acetic acid (9 : 1, v/v) : B, toluene}acetic acid (9 : 1, v/v); C,

toluene}ethanol}acetic acid (17 : 1 : 2, v/v) : D, water}formic acid (200 : 3, v/v); E,
water}ethanol}ammonium hydroxide (17 : 2 : 1, v/v); F, ethyl acetate}ethanol}ammonium
hydroxide (20 : 5 : 1); G, water}ethanol}ammonium hydroxide (14 : 15 : 1, v/v); H,
n-heptane}n-butanol}acetic acid (3 : 3 : 1, v/v); I, chlorobenzene}acetic acid (9 : 1, v/v);
J, ethyl acetate}methanol}acetic acid (20 : 1 : 1).
Table 10 Various solvent systems for TLC of dansyl amino acids
Solvent systems Ratio
1. HCOOH 1.5%
Benzene}acetic acid 9:1
2. Formic acid 1.5%
Benzene}acetic acid 4.5 : 1
3. H2O}pyridine}HCOOH 93 : 35 : 3.5
4. NH4Cl#NH3#ethanol 80 g#22 mL#10 mL
Benzene}pyridine}HOAc 75 : 2 : 6
5. H2O}propanol}formic acid 100 : 5 : 2
6. Ethyl acetate}MeOH}HOAc 20 : 1 : 1
Benzene}HOAc}BuOH 90 : 10 : 5
7. Formic acid 1.5%
8. Benzene}anhydrous HOAc, followed by 9:1
EtOAc}MeOH}anhydrous HOAc in the same direction 10 : 1
Formic acid 1.5%
9. H2O}pyridine}HCOOH 93 : 35 : 3.5
10. Formic acid 3%
11. Me}acetate}iso-PrOH}NH3 9:7:4
CHCl3}MeOH}HOAc 15 : 5 : 1
CHCl3}EtOAc}MeOH}HOAc 45 : 75 : 22.5 : 1
Pet ether}t -BuOH}HOAc 5:2:2
12. CHCl3}MeOH 9:1
13. CCl4}2-methoxyethanol 17 : 3
14. Benzene}pyridine}acetic acid 80 : 20 : 2
Solvents at serial no. 1}8 : two-dimensional TLC on polyamide layers.

Solvents at serial no. 9}14 : one-dimensional TLC on silica gel layers.
Table 11 hRF Values of 10 dansyl amino acids on silica gel thin Table 13 hRF Values of DNP amino acids on silica gel thin
layers (Sl. no."serial number) layers
Sl. Dansyl amino acid Solvent system Sl. N-DNP-L-amino acid Solvent system
no. no.
S1 S2 S3 S4 S5 S1 S2 S3 S4 S5
1. Dansyl-L-alanine 62 61 60 50 27 1. Phenylalanine 53 48 85 70 55
2. Dansyl-L-isoleucine 80 92 85 85 49 2. Isoleucine 68 82 96 97 60
3. Dansyl-L-leucine 83 85 80 89 65 3. Tyrosine 25 30 60 52 36
4. Dansyl-L-methionine 86 64 62 55 31 4. Alanine 40 36 68 50 42
5. Dansyl-L-proline 60 84 72 30 39 5. Glycine 28 17 35 25 27
6. N-O -dansyl-L-tyrosine 55 73 40 60 18 6. Leucine 65 73 93 90 52
7. N--dansyl-L-tryptophan 51 53 46 40 21 7. Tryptophan 48 33 53 47 34
8. Dansyl-L-phenylalanine 77 76 74 52 40 8. Methionine 45 40 75 57 42
9. Dansyl-L-valine 72 88 65 48 35 9. Valine 62 65 90 85 47
10. Dansyl-L-norvaline 75 81 68 45 37 10. Proline 41 45 74 60 38
11. Norvaline 61 62 88 83 45
S1, n -heptane}BuOH}HOAc (20 : 8 : 3, v/v); S2, dichloro-
methane}MeOH}propionic acid (30 : 1 : 0.5, v/v); S3, chloro- Solvent system
form}HOAc}ethylacetate (24 : 5 : 4, v/v); S4, chloroform}MeOH}
ethyl acetate (23 : 8 : 2, v/v); S5 , chloroform}propionic acid}ethyl A1 A2 A3 A4 A5
acetate (23 : 6 : 4, v/v); RF values are average of five determina-
tions. 12. N -DNP-L-serine 51 68 70 70 70
13. N -DNP-lysine 21 26 11 7 27
14. N-S-di-DNP-L-cysteine 82 87 77 85 85
identiRcation of these derivatives become more con- 15. N -DNP-L-glutamic acid 67 80 83 92 82
cyclohexyl-amine salt
venient and sensitive (10 pmol on a polyamide plate). 16. N -DNP-L-aspartic acid 38 70 75 60 60
Exposure to HCl vapours turns all yellow spots to red 17. N -DNP-L-asparagine 30 64 45 38 55
or blue on polyamide sheets. 18. N -DNP-L-arginine 10 6 5 3 18
19. N,N -di-DNP-L-cystine 48 70 55 65 82
DNP Amino Acids
S1, n -heptane-n -butanol-acetic acid (20 : 4 : 1, v/v); S2, chloro-
Use of DNP amino acids, formed by condensation of form}propionic acid (26 : 2, v/v); S3, chloroform}acetic acid
1-Suoro-2,4-dinitrobenzene (FDNB) with the free (21 : 1, v/v); S4, chloroform}ethanol}propionic acid (30 : 2 : 1,
amino group of an amino acid, was Rrst described by v/v); S5, benzene}n -butanol}acetic acid (34 : 1 : 1, v/v); A1,
Sanger in 1945, who identiRed DNP amino acids by chloroform}methanol}acetic acid (25 : 5 : 1, v/v); A2, chloro-
form}propionic acid}methanol (15 : 10 : 1, v/v); A3, n -heptane}
paper chromatography. Since then many modiRca-
butanol}acetic acid (16 : 8 : 4, v/v); A4, n-butanol}ethyl acet-
tions in the methods of obtaining derivatives of ate}acetic acid (20 : 8 : 2, v/v); A5, n-butanol}methanol}propionic
acid (18 : 8 : 2, v/v). RF values are average of five determinations.
Table 12 hRF Values of 10 dansylamino acids on silica gel thin

layers
amino acids for sequence analysis and in identiRca-
Sl. Dansyl amino acid Solvent system tion of such derivatives have been reported, and the
no. use of DNP amino acids for sequencing purposes is
A1 A2 A3 A4 A5 rapidly going out of practice. Nevertheless, the im-
portance of DNP amino acids has not yet disap-
1. N- -dansyl-L-asparagine 56 75 53 30 35
2. Dansyl-L-aspartic acid 66 72 60 64 30 peared.
3. -Dansyl-L-arginine 7 12 3 2 3 Kirchner presented considerable information on
4. N-N-didansyl-L-cystine 84 83 68 85 18 the analysis of DNP amino acids based on the litera-
5. Dansyl-L-cysteic acid 82 80 25 15 11 ture available up to 1970. In one of the earlier
6. Dansyl-L-glutamic acid 80 90 84 74 55
methods, thin-layer plates (20;20 cm;0.25 mm)
7. Dansyl-L-glutamine 62 77 63 41 40
8. N-dansyl-L-lysine 16 20 10 6 8 were prepared from a mixture of 10 g of cellulose
9. N-dansyl-L-serine 72 85 72 58 32 MN-300 and 4 g silica gel H (Merck), homogenized
10. Dansyl-L-threonine 76 88 76 68 45 in 80 mL of water, dried overnight at 373C and
developed in the Rrst dimension with two sol-
A1, Dichlormethane}MeOH}propionic acid (21 : 3 : 2, v/v); A2,
vents successively: iso-propanol}acetic acid}H2O
ethyl acetate}MeOH}propionic acid (22 : 10 : 3, v/v); A3, chloro-
form}MeOH}HOAc (28 : 4 : 2, v/v); A4, chloroform}acetone} (75 : 10 : 15) for 15 min and n-butanol}0.15 mol
HOAc (20 : 8 : 4, v/v); A5, chloroform}acetone}propionic acid L\1 ammonium hydroxide (1 : 1, upper phase). The
(24 : 10 : 5, v/v) RF values are the average of five determinations. dried chromatograms were developed in 1.5 mol
L\1 sodium phosphate buffer (pH 6.0) in the second The impregnating agent participates in various
dimension. mechanisms in the resolution process, including ion-
In almost all methods reported, the separation has pairing, complex formation, ligand exchange, coord-
been carried out in groups of water-soluble and ether- ination bonds, charge transfer, ion exchange and
soluble DNP amino acids, and for each group mostly hydrogen bonding.
two-dimensional TLC has been performed. A few
solvent systems for one-dimensional resolution of DNP Amino acids Resolution of amino acids has been
amino acids on silica gel plates are shown in Table 13. reported to be very rapid and improved by using
copper sulfate, halide ions, zinc, cadmium and mer-
Detection of DNP amino acids The DNP amino cury salts, and alkaline earth metal hydroxides as
acids have been visualized by UV light (360 nm with impregnating materials and some of the results are
dried plates, or 254 nm with wet ones) or by their described in Tables 14}17. The chromatograms de-
yellow colour, which deepens upon exposure to am- veloped in these systems provide compact spots, with-
monia vapours. Thin layers of silica gel usually give out lateral drifting of the solvent front. C18 layers
an intense purple Suorescence for DNP amino acids impregnated with dodecylbenzene sulfonic acid are
under UV light, which masks the presence of very helpful in conRrming the presence of an unknown
faint spots and decreases the colour contrasts. The amino acid in a sample and the migration sequence
cellulose}silica mixed layer gives much lower Suores- on these impregnated plates is reversed, probably due
cence and preserves the colour contrasts between to an ion exchange mechanism. Separation of -
various derivatives. Because of the photosensitivity of amino acids with butan-1-ol}acetic acid}water
these derivatives, it is advisable to carry out their (3 : 1 : 1, v/v), butan-1-ol}acetic acid}chloroform
preparation and chromatography in the absence of (3 : 1 : 1, v/v), and butan-1-ol}acetic acid}ethyl ace-
direct illumination. tate (3 : 1 : 1, v/v), on plain and nickel chloride im-
pregnated plates has been reported; the partition and
Resolution of Amino Acids and adsorption coefRcients for the amino acids under
study were determined on both untreated and Ni2#
Derivatives on Impregnated Layers impregnated silica gel in a batch process and correla-
The technique of incorporating a suitable reagent tions were drawn between TLC separation of amino
with the adsorbent, prior to applying the samples to acids on the impregnated gel with adsorption/parti-
the plates, originated from simple TLC and can be tion characteristics. The results indicate a predomi-
termed impregnated TLC. The reagents and methods nant role of partitioning in the separation. Applica-
used for impregnation are not to be confused with tion of antimony (V) phosphate}silica gel plates in
locating/spray reagents because the latter are required different aqueous, nonaqueous and mixed solvent
for the purpose of identiRcation even on impregnated systems has also been reported. Some impregnated
plates. TLC systems for resolution of amino acids are sum-
marized in Table 18.
Methods for Impregnation
Of the various methods used for impregnation, one is PTH amino acids As mentioned above, certain difR-
mixing of the impregnating reagent with the inert culties in resolving or identifying various PTH amino
support. A second approach is the immersion of the acid combinations have successfully been removed
plates into an appropriate solution of the impregnat- and multicomponent mixtures separated with metal
ing reagent carefully and slowly so as not to impregnated silica gel layers, while other reagents
disturb the thin layer. Alternatively, a solution of such as (#)-tartaric acid and (!)-ascorbic acid have
the impregnating material is allowed to ascend or been used for the resolution of enantiomeric mix-
descend the plate in the normal manner of develop- tures. The methods reported provide very effective
ment; this method is less likely to damage the thin resolution and compact spots (by exposure to iodine
layer. Exposing the layers to the vapours of the im- vapours) and can be applied to the identiRcation of
pregnating reagent or spraying the impregnating re- unknown PTH amino acid; some of these are given in
agent (or its solution) on to the plate have also been Tables 19}21. Some of the successful solvent systems
employed; spraying provides a less uniform disper- for TLC of PTH amino acids on impregnated plates
sion than the other methods. Another approach is are summarized in Table 22.
to have a chemical reaction between the inert support
and a suitable reagent: the support is chemically High performance TLC (HPTLC)/overpressured
modiRed before making the plate, the compounds of TLC (OPTLC) Improvements in the solid-phase
interest are bonded to the reactive groups of the layer. materials for TLC have resulted in an increase in
Table 14 hRF of amino acids in presence of halides
Sl. Amino Control Amino acids pretreated with Plates impregnated with
no. acid plate
Cl\ Br\ I\ Cl\ Br\ I\
1. Gly 07 08 09 12 07 08 09
2. Tyr 30 35 40 47 29 30 31
3. Pro 12 15 19 22 08 09 10
4. Thr 15 14 15 19 13 14 16
5. Cys 22 22 25 27 19 20 22
6. Leu 32 40 47 50T 50T 55T 60T
7. Met 23 35 36 37 22 23 24
8. Ile 30 38 44 44 30 30 31
9. Ala 15 19 13 16T 16T 16 16
10. Try 35T 40 50T 53 30 31 34
11. Phe 36T 41 48 48 365 37 38
12. Val 19 32 25 29 25 26 26
13. Asp 08 13 14 15 08 09 10
14. Ser 09 13T 13 14T 08 08 09
15. His 01 03 04 05 02 02 02
Time (min) 50 64 67 67 50 50 50
Solvent system: n -butanol}acetic acid}chloroform (3 : 1 : 1, v/v); temperature 25$23C.

T"tailing.
separation efRciency, sample detectability limits and not achieved initially. A continuous multiple develop-
reduced analysis time. HPTLC can be used with ad- ment on silica gel was able to separate 18 samples and
vantage for the separation of PTH amino acids but standards simultaneously using Rve development
separation of all 20 common PTH amino acids was steps with four changes in mobile-phase and scanning
densitometry; typical results are given in Table 23.
PTH-Leu/Ile/Pro have been identiRed by HPTLC
Table 15 hRF values for amino acids on copper sulfate and using multiple wavelength detection. OPLC using
polyamide mixed silica gel plates chloroform}ethanol}acetic acid (90 : 10 : 2) for po-
lar, and dichloromethane}ethyl acetate (90 : 10) for
Amino acid A B C
nonpolar PTH amino acids has been successful in
L-Leucine (Leu) 65 63 71 their separation and quantitation; the method is
D,L-Isoleucine (Ile) 66 72 81 claimed to be superior to HPTLC in having relatively
D,L-Tryptophane (Try) 63 68 75 increased migration distance, resulting in the resolu-
D,L-Methionine (Met) 64 64 72
tion of complex mixtures containing a large number
D,L-Valine (Val) 64 60 77
L-Lysine-HCl (Lys) 16T 12 33 of derivatives. OPTLC and HPTLC on RP-8, RP-18,
L-Histidine-HCl (His) 22T 20 39 and home-made ammonium tungstophosphate layers
D,L--Phenylalanine (Phe) 64 65 82 have also been used for the analysis of DNP amino
D,L-Threonine (Thr) 50 51 67 acids.
D,L-Alanine (Ala) 46 45 64
Separation of 18 amino acids on cellulose, silica gel
D,L-Serine (Ser) 40 43 56
L-Tyrosine (Tyr) 58 61 71 and chemically bonded C18 HPTLC plates has been
L-Glutamic acid (Glu) 41 48 58 achieved. All of these plates contain a preadsorbent
D,L-Aspartic acid (Asp) 28 25 44 zone except the cellulose. QuantiRcation is carried
L-Arginine HCl (Arg) 24T 19 39 out by scanning standard and sample zones at
Glycine (Gly) 36 46 49
610 nm. hRF values of amino acid standards on rever-
L-Proline (Pro) 37 36 58
L-Cysteine HCl (Cys) 20T 17 29 sed-phase and on normal-phase layers in different
D,L-2-Aminobutyric acid (Aba) 51 54 61 solvents are given in Tables 24 and 25, respectively.
L-Ornithine 27T 23 35
The values are average of two or more identical runs, 10 cm in Resolution of Enantiomeric Mixtures
35 min. A, untreated silica gel plate; B, copper sulfate-impreg- of Amino Acids and Derivatives
nated silica gel; C, polyamide mixed silica gel layers; T, tailing
Solvent, methanol}butyl acetate}acetic acid}pyridine (20 : 20 : The measurement of speciRc rotation is a common
10 : 5, v/v). and accepted method for evaluating the enantiomeric
Table 16 hRF values of 15 amino acids on silica gel impregnated with Zn, Cd and Hg
salts
A B C D E F G H I J
Thr 25 55 42 41T 35 36 42 33 50 40
Ser 12 38 39 28T 32 29 31T 15 40 31T
Gly 10 35 29 23T 28 25 28 16 35 27T
Lys 03 13 07 05 51 08 05 04 10 05
Ala 30 48 40 31 38 36 38 20 5 35
Tyr 60 60 52 50 48 45 51 62 55 56
Ile 55 67 56 52 50 48 54 50 60 53
Leu 50 65 55 55 52 50 56 47 64 55
Cys 00 00 00 00 00 00 00 00 00 00
Met 45 62 48 48 48 42 48 39 54 45
Glu 18T 43 38 36T 34 27 38T 18 36 34T
Try 57 60 53 51 51 44 54 45 60 47
Phe 54 67 57 55 55 46 57 58 68 52
Val 50 63 45 50 52 42 56 47 57 45
Arg 07 19 13 13 09 11 11 10 15 08
Solvent, butyl acetate}methanol}acetic acid}pyridine (20 : 20 : 5 : 5, v/v). Developing

time, 30 min. Detection limit, 10\4 mol L\1. Solvent front, 10 cm. A, plain silica gel; B, C, D,
0.5%, 0.2%, 0.1% Zn2>C-impregnated, respectively; E, F, G. 0.5, 0.2, 0.1% Cd2>-
impregnated, respectively; H, I, J, 0.5, 0.2, 0.1% Hg2>, respectively. T"tailing.
purity of chiral compounds. The determination of purity, there are few reports on TLC separation
enantiomeric excess (ee) values is inSuenced by the of enantiomers.
presence of impurities and changes in concentration, In general, the following approaches for the resolu-
solvent and temperature, and requires the []D value tion of enantiomers have been used.
for the pure enantiomer. The availability of a reliable
Indirect method
optically pure standard depends on the analytical
method by which it had been resolved from the enan- This method involves reaction of the enantiomeric
tiomeric or racemic mixture of the compound in mixture with a suitable chriral reagent to make the
question. Though TLC provides a direct method for corresponding diastereomeric derivatives prior to
resolution and analytical control of enantiomeric chromatography; the choice of chiral selector is
Table 17 hRF values of amino acids on untreated plates and plates impregnated with
metal sulfates
Aminoacids Unimpregnated Plate impregnated with

plate
Mn 2# Fe 2# Co 2# Ni 2# Cu 2# Zn 2# Cd 2# Hg 2#
Asp 21 51 54 50 62 58 52 59 64
Glu 25 51 63 55 59 61 54 58 65
Phe 45 64 74 67 72 74 69 68 65
Tyr 46 66 73 68 72 70 72 68 71
Lys 7 15 21 22 18 18 16 32 25
Orn 28 15 23 23 19 23 20 28 T
Arg 30 20 25 30 28 28 25 35 33
Ala 30 50 52 53 60 55 49 58 73
Val 48 60 70 58 71 65 59 65 70
Ser 29 45 57 48 57 52 44 56 48
Hypo 26 42 52 47 48 50 43 57 45
Gly 20 43 58 48 52 40 45 54 55
Leu 50 67 72 69 75 71 74 69 SF
Cys SF 24 37 32 34 35 30 34 40
T, Tailing; SF, migrates with solvent front. hRF values are the average of at least five
determinations.
Table 18 TLC of amino acids on impregnated silica gel layers
Solvent system Ratio (v /v) Impregnation
iso-Amyl alcohol}H2O}HOAc 6:5:3 Pyridinium tungstoarsenate

H2O}EtOAc}MeOH 64.3 : 5.7 : 30 Silanized silica and triethanol amine.
SDS, sodium di-octylsulfonate,
dodecyl benzene sulfonic acid
0.1 mol L\1 HOAc in aq. 50% MeOH Dodecyl benzene sulfonic acid
Aq. MeOH#I2 (KCl or HOAc added) Ammonium tungstophosphate and
dodecyl benzene sulfonic acid
Aq. NH4NO3 or HNO3 or Ammonium tungstophosphate
H2O}HOAc}MeOH (79 : 1 : 20)
H2O Polyamide
H2O}butanol}anhyd. HOAc 4:4:2 Kieselguhr or cellulose
n -Butanol}acetic acid}water 4:1:5 Starch}agar (1 : 1)
Propan-2-ol}EtOAc}acetone}methanol}n-pentyl alcohol}aq. 26% 9:3:3:1:1:3:3 Cellulose
NH3}water in first direction; and Butanol}acetone}propan-2-ol} 18 : 8 : 8 : 3 : 6
formic acid}water in second direction
1 mol L\1 NH4NO3}0.1 mol L\1 HNO3 Ammonium tungstophosphate
MeOH}butyl acetate}HOAc}pyridine 4:4:2:1 Copper sulfate and polyamide
n -Butanol}acetic acid}CHCl3 3:1:1 Cl\, Br\, I\
n -Butanol}acetic acid}ethanol 3:1:1 Hydroxides of Mg, Ca, Ba, Sr
Butyl acetate}MeOH}HOAc}pyridine 4:4:1:1 Zn2#, Cd2#, Hg2#
limited due to the feasibility of its reaction with the 3. A suitable chiral reagent is incorporated, such as
analyte. acid, base, an organic compound or a metal com-
plex with the adbsorbent during plate making, or
Direct method
at a stage before developing the chromatogram.
1. This method uses a chiral stationary phase; it may
be due to either natural chirality of the material as DL-amino acids Separation of D,L-tryptophan on
such, like cellulose, or some sort of synthesis of the a crystalline cellulose-coated plate in 1980 seems to
phase. be one of the Rrst TLC reports. Applying the principle
2. Chiral discriminating agents are added to the mo- of ligand exchange, (2S, 4R, 2RS)-4-hydroxyl-1-
bile phase. (2-hydroxy dodecyl)-proline was used as the chiral
Table 19 hRF values of PTH amino acids on Fe2#, Co2#, Ni2# and Zn2# impregnated
silica plates
Sl. PTH-amino Alone Fe 2# Co 2# Ni 2# Zn 2#

no. acid
0.2% 0.3% 0.05% 0.1% 0.1% 0.2% 0.2% 0.3%
1. Alanine 60 42 41 57 51 38 40 50 43
2. Aspartic acid 0 0 0 0 0 0 0 0 0
3. Glycine 39 26 21 44 38 29 30 32 27
4. Glutamic acid 0 0 0 0 0 0 0 0 0
5. Isoleucine 90 84 75 72 90 65 71 81 72
6. Leucine 95 87 71 82 81 70 76 85 76
7. Lysine 23 8 6 15 17 7 4 10 8
8. Methionine 70 54 47 81 62 58 51 57 58
9. Phenylalanine 75 61 49 77 68 52 55 66 58
10. Proline 97 89 89 84 76 83 90 96 89
11. Serine 13 5 5 11 9 11 12 8 5
12. Tyrosine 96 867 69 68 95 85 78 83 78
13. Tryptophan 95 91 70 91 97 77 82 88 81
14. Threonine 86 78 57 94 83 60 63 78 70
15. Valine 85 75 73 96 79 57 58 76 67
Solvent, chloroform}ethyl acetate (29 : 3, v/v); developing time 35 min; solvent front,
10 cm.
Table 20 hRF values of PTH amino acids on untreated plates and plates impregnated with sulfates of Mg, Mn, Fe and Co
PTH amino acid S1 (heptane}butylacetate, 15#5) S2 (heptane}propionic acid, 20#4) S3 (benzene}ethyl acetate,
15#3)a
PS1 M1 M2 M3 M4 PS2 M1 M2 M3 M4 PS3 M4
Methionine 28 30 26 32 31 43 45 30 32 35 62 78
Phenylalanine 30 35 29 37 34 50 52 36 38 40 67 80
Tryptophan 63 61 51 60 57 71 67 55 55 57 82 94
Valine 49 46 40 51 47 66 62 52 50 55 73 85
Isoleucine 62 62 50 62 59 77 72 61 60 62 78 65
Tryosine 66 64 53 64 61 80 74 64 64 65 84 89
Threonine 57 52 45 56 53 63 64 53 53 53 72 83
Alanine 23 25 23 26 25 32 34 27 25 29 50 67
Serine 55 55 42 55 51 48 46 38 49 40 70 44
Leucine 69 65 54 63 56 76 71 62 60 67 80 96
Lysine 06 04 02 03 05 06 10 04 06 07 18 35
Glycine 17 15 13 15 15 17 20 15 15 18 37 55
Glutamic acid 04 06 05 06 06 04 04 06 06 05 0 14
Aspartic acid 05 07 06 07 07 05 08 07 07 06 0 22
Proline 44 33 31 34 32 44 42 35 35 45 79 96
a
Compounds moved to solvent front on plates impregnated with sulfates of Mg, Mn and Fe. PS1, PS2, PS3, untreated plates; M1, M2, M3,
M4, treated with sulfates of Mg, Mn, Fe, and Co, respectively. RF values are the average of at least five determinations. Developed in
30}40 min at 253C$23C, and exposed to iodine vapours to locate the spots.
selector to resolve several racemic -amino acids on methanol}H2O}acetonitrile (50 : 50 : 30, v/v) as the
reversed-phase 18-TLC plates Rrst immersed (1 min) mobile phase and ninhydrin as the detecting reagent
in a 0.25% copper(II) acetate solution (MeOH}H2O, (Figure 1). The RF values for L-Dopa and D-Dopa
1 : 9, v/v), dried, and then immersed in a 0.8% meth- were reported to be 0.47 and 0.61, respectively, and
anolic solution of the chiral selector (1 min); the re- the system is capable of resolving enantiomers in
sults are shown in Table 26. Ready-to-use Chiral- trace amounts, with the lowest level of detection of
plates威 are now marketed by Macherey-Nagel, the D-enantiomer in L-Dopa samples being 0.25%.
Duren, Germany, and Chir威 plates are marketed by The resolution of enantiomers of -substituted -
Merck, Germany. Resolution of DL-methyl Dopa, and amino acids, and racemic mixtures of natural
DL-Dopa is very successful on Chiralplates using and nonnatural amino acids, N-methylated and
Table 21 hRF Values of PTH amino acids on silica plates impregnated with zinc salts
PTH amino acid S1 (heptane}butylacetate, 15#5) S2 (heptane}propionic acid, 20#4) S3 (benzene}ethyl acetate 15#3)
L1 L2 L3 L4 L1 L2 L3 L4 L1 L2 L3 L4
Methionine 17 22 18 25 33 33 29 33 42 57 48 58
Phenylalanine 22 25 25 28 37 38 35 36 47 60 52 60
Tryptophan 36 41 41 51 40 50 40 55 67 74 61 82
Valine 40 35 44 42 52 53 54 52 54 68 64 69
Isoleucine 50 46 55 54 50 62 63 63 62 79 74 74
Tryosine 55 48 59 56 62 64 75 65 64 84 79 78
Threonine 47 40 52 48 53 51 52 56 57 72 72 67
Alanine 23 17 21 22 29 27 23 27 35 44 38 44
Serine 49 45 42 50 38 36 37 39 52 66 60 64
Leucine 55 51 55 59 61 50 67 60 65 81 77 76
Lysine 03 02 03 03 04 05 04 07 6 7 6 8
Glycine 15 10 12 13 16 15 15 15 24 29 25 31
Glutamic acid 0 04 0 04 03 02 02 04 0 0 0 0
Aspartic acid 0 05 0 05 04 03 03 05 0 0 0 0
Proline 38 25 32 30 40 40 40 42 70 77 83 72
L1, L2, L3, L4 plates impregnated with Cl\, SO24\, CH3COO\ and PO34\ of zinc, respectively. Other conditions as in Table 20.
Table 22 TLC of PTH amino acids on impregnated silica gel layers
Solvent system Ratio (v/v) Impregnation a
CHCl3}H2O}EtOAc 28 : 1 : 1 Zn2#, Cd2#, Hg2#

CHCl3}MeOH}Benzene 19 : 1 : 9
CCl4}HOAc 19 : 1
CHCl3}Benzene}EtOAc 25 : 5 : 3 Fe2#, Co2#, Zn2#
CHCl3}EtOAc 29 : 3 Fe2#, Co2#, Ni2#
n -Heptane}n -butyl acetate 15 : 5 Cl\, CH3COO\, PO34\
n -Heptane}n -proprionic acid 20 : 4 of zinc, or SO24\ of
Benzene}EtOAc Mg2#, Mn2#, Fe2#, Co2#
CHCl3}n -butyl acetate 10 : 5
CHCl3}EtOAc 25 : 2
a
Various concentrations of each of the impregnating reagent have been used.
N-formylated amino acids, and various other deriva- have provided enantiomeric separations of amino
tives of amino acids has also been achieved on Chiral- acids and derivatives. Chiral monohalo-s-triazines
plates; typical results are presented in Tables 27 and have been used for the TLC resolution of DL-amino
28. A novel chiral selector from (1R, 3R, 5R)-aza- acids. Racemic dinitropyridyl-, dinitrophenyl- and
bicyclo-[3,3,0]-octan carboxylic acid has been syn- dinitrobenzoyl amino acids are separated on rever-
thesized and used as a copper(II) complex for the sed-phase-TLC plates developed with aqueous-org
impregnation of reversed-phase 18 plates for ligand mobile phases containing bovine serum albumin as
exchange TLC separation of amino acids and the a chiral agent.
results were comparable to those on Chiralplates威.
Chiral selectors such as (!)-brucine and Cu-L-pro- Dansyl-DL-amino acids Reversed-phase TLC
line complex are used to resolve enantiomers of plates from Whatman are developed before applica-
amino acids (Table 29), and (#)-tartaric acid and tion of dansyl amino acids in buffer A (0.3 mol
(!)-ascorbic acid for the resolution of enantiomeric L\1 sodium acetate in 40% acetonitrile, and 60%
PTH amino acids (Table 30). The chiral selectors are water adjusted to pH 7 by acetic acid). After fan-
mixed with silica gel slurry. drying, the plates are immersed in a solution of
Resolution of trytophans and substituted tryp- 8 mmol L\1 N,N-di-n-propyl-L-alanine and 4 mmol
tophans on cellulose layers developed with copper L\1 cupric acetate in 97.5% acetonitrile, 2.5% water
sulfate solutions has shown that excess of Cu2# ions for 1 h or overnight and left to dry in air. After
decreases the chiral discrimination of the system, and applying the samples, the plates are developed in
development with aqueous -cyclodextrin (1}10%) buffer A with or without N,N-di-n-propyl-L-alanine
plus NaCl solutions (0.1, 0.5, 1.0 mol L\1) showed (4 mmol L\1) and cupric acetate (1 mmol L\1) is
the best results with aqueous 4% -cyclo- dissolved in it. The enantiomers are detected by irra-
dextrin}1 mol L\1 NaCl solution; the results are diating with UV light (360 nm) to yield Suorescent
comparable to Chiralplate威. It has been observed that yellow-green spots. Use of 25% acetonitrile is prefer-
chiral effects are essentially additive (for cellulose and red for glutamic and aspartic acids and serine and
-cyclodextrin) and there is a strong temperature de- threonine derivatives. N,N-di-n-propyl alanine can
pendence for the chiral separations. be prepared by the following procedure: L-alanine
- and -cyclodextrins, hydroxypropyl--cyclodex- (17.8 g) is dissolved in ethanol (200 mL) and 10%
trin and bovine serum albumin in the mobile phase palladium on activated charcoal catalyst (3 g) and
Table 23 Optimum experimental conditions for the separation of PTH amino acids by continuous multiple development HPTLC
Step Mobile phase Plate length (cm) Time (min) PTH amino acid identified
1. CH2Cl2 3.5 5 Pro

2. CH2Cl2}propan-2-ol (99 : 1, v/v) 7.5 10 Pro, Leu, Ile, Val, Phe
3. CH2Cl2}propan-2-ol (99 : 1, v/v) 7.5 10 Pro, Leu, Ile, Val, Phe, Met, Ala/Try,
Gly, Lys, Tyr, Thr
4. CH2Cl2}propan-2-ol (97 : 3, v/v) 7.5 10 Pro, Met, Lys, Tyr, Thr, Ser, Glu
5. C2H5COOCH3}CH3CN}CH3COOH 7.5 10 Asn, Glu/Gln, Asp, Cm-Cys, His, Arg
(74.3 : 20 : 0.7, v/v)
Table 24 hRF Values of amino acid standards on reversed- Table 26 Enantiomeric resolution of amino acids by TLC
phase layers
Amino acid RF value (configuration) Mobile
Amino acid TLC system phase
R S
1 2 3 4 5 6
Isoleucine 0.37 (2R, 3R ) 0.44 (2S, 3S ) A
Aspartic acid 30 59 72 60 83 73 Phenylalanine 0.38 0.45 A
Arginine 28 4 35 28 86 82 Tyrosine 0.34 0.26 B
Serine 55 36 69 50 82 73 Tryptophan 0.39 0.45 A
Glycine 50 38 62 45 69 54 Proline 0.40 0.59 B
Tyrosine 91 77 88 68 77 67 Glutamine 0.53 0.37 A
Alanine 78 59 71 63 71 54
Glutamic acid 82 70 86 67 83 69 Development distance: 14 cm; saturated chamber. A, MeOHI
Proline 56 69 64 40 65 46 waterIMeCN (1 : 1 : 4, v/v); B, MeOHIwaterIMeCN (5 : 5 : 3, v/v).
Cystine 11 12 39 33 85 84
Methionine 90 74 75 59 75 61
Lysine 31 84 27 24 74 79
Tryprophan 90 77 85 63 72 63 a sintered glass Rlter and the Rltrate is evaporated to
Valine 90 74 75 59 75 61 dryness. The reaction product (N,N-di-n-propyl-L-
Threonine 78 52 68 50 72 57
alanine) is crystallized from chloroform, and the
Histidine 21 3 29 23 77 68
Phenylalanine 90 76 83 65 72 61 purity may be conRrmed by TLC, and carbon, hydro-
Leucine 90 77 81 62 75 63 gen, nitrogen analysis.
Isoleucine 91 77 81 62 74 61
Layers: 1, 2, Whatman C-18; 3, 5, Merck RP-18; 4, 6, Merck

RP-18W. Mobile phases: 1, 3, 4, n-ButanolIglacial acetic acidI
water (3 : 1 : 1, v/v); 2, 5, 6, n-propanolIwater (7 : 3, v/v).
propionaldehyde (43 mL) is added. The mixture is

hydrogenated for 48 h at 40}503C at an initial hydro-
gen pressure of 50 psi; the catalyst is removed using
Table 25 hRF Values of amino acid standards on normal-phase

layers
Amino acid TLC system
1 2 3 4
Aspartic acid 28 27 26 58
Arginine 18 18 17 2
Serine 26 30 27 40
Glycine 26 32 28 43
Tyrosine 46 58 53 78
Alanine 38 32 32 55
Glutamic acid 69 56 50 64
Proline 45 32 28 50
Cystine 10 11 9 30
Methionine 60 59 51 72
Lysine 15 13 10 4
Tryptophan 55 63 57 82
Valine 60 56 49 68
Threonine 34 32 32 53
Histidine 14 14 11 17
Phenylalanine 68 61 55 80
Leucine 79 61 55 78
Isoleucine 78 59 54 77 Figure 1 Chromatogram showing separation of different D- and
L-dopa samples on Chiralplate. From left to right: 1, L-dopa; 2,
Layers: 1, Cellulose; 2, 4, Whatman silica gel; 3, Merck silica D,L-dopa; 3, D-dopa; 4, 3% L-dopa in D-dopa; 5, 3% D-dopa in
gel. Mobile phases: 1, Butan-2-olIglacial acetic acidIwater L-dopa. Developing solvent, methanol}water}acetonitrile
(3 : 1 : 1, v/v); 2, 3, n -butanolIglacial aceticIwater (3 : 1 : 1, v/v); (5 : 5 : 3, v/v). Developing time 45}60 min. Detection 0.1% nin-
4, n -propanolIwater (7 : 3, v/v). hydrin spray reagent.
Table 27 Enantiomeric resolution of -dialkyl amino acids by TLC
Parent amino acid R1 R2 RF value Configuration Mobile phase
Asp CH2CO2H CH3 0.52 (D) 0.56 (L) A

Glu (CH2)2CO2H CH3 0.58 (L) 0.62 (D) A
Leu CH2CH(CH3)2 CH3 0.48 0.59 A
Phe CH2C6H5 CH3 0.53 (L) 0.66 (D) A
Ser CH2OH CH3 0.56 (L) 0.67 (D) B
Try CH2-3-indolyl CH3 0.54 0.65 A
Tyr CH2-(4-OH-C6H4) CH3 0.63 (D) 0.70 (L) A
Val CH(CH3)2 CH3 0.51 0.56 A
-Amino butyric acid CH2CH3 CH3 0.50 0.60 A
Phe CH2C6H5 CHF2 0.63 0.70 A
Phe CH2C6H5 CH2}CH"CH2 0.57 0.63 A
Phe CH2C6H5 CH2CH2SCH3 0.57 0.62 A
Mobile phase: A, methanol}water}acetonitrile (1 : 1 : 4, v/v); B, methanol}water}acetonitrile (5 : 5 : 3, v/v). Development distance

13 cm; saturated chamber.
In a two-dimensional reversed-phase TLC tech- ated in nonchiral mode using 0.3 mol L\1 sodium
nique for the resolution of complex mixture of dan- acetate in H2O}acetonitrile (80 : 20, pH 6.3) to
syl-dl-amino acids, the Dns-derivatives are Rrst separ- which 0.3 mol L\1 sodium acetate in H2O}aceto-
Table 28 Enantiomeric resolution of racemates by TLC
Racemate RF value Configuration Mobile phase
Valine 0.54(D) 0.62(L) A

Methionine 0.54(D) 0.59(L) A
Allo-isoleucine 0.51(D) 0.61(L) A
Norleucine 0.53(D) 0.62(L) A
2-Aminobutyric acid 0.48 0.52 A
o -Benzylserine 0.54(D) 0.65(L) A
3-Chloralanine 0.57 0.64 A
S -(2-Chlorobenzyl)-cysteine 0.45 0.58 A
S -(3-Thiabutyl)-cysteine 0.53 0.64 A
S -(2-Thiapropyl)-cysteine 0.53 0.64 A
cis-4-Hydroxyproline 0.41(L) 0.59(D) A
Phenylglycine 0.57 0.67 A
3-Cyclopentylalanine 0.46 0.56 A
Homophenylalanine 0.49(D) 0.58(L) A
4-Methoxyphenylalanine 0.52 0.64 A
4-Aminophenylalanine 0.33 0.47 A
4-Bromophenylalanine 0.44 0.58 A
4-Chlorophenylalanine 0.46 0.59 A
2-Fluorophenylalanine 0.55 0.61 A
4-Iodophenylalanine 0.45(D) 0.61(L) A
4-Nitrophenylalanine 0.52 0.61 A
o -Benzyltyrosine 0.48(D) 0.64(L) A
3-Flurotyrosine 0.64 0.71 A
4-Methyltryptophan 0.50 0.58 A
5-Bromotryptophan 0.46 0.58 A
5-Methoxytryptophan 0.55 0.66 A
2-(1-Methylcyclopropyl)-glycine 0.49 0.57 A
N -Methylphenylalanine 0.59(D) 0.61(L) A
N -Formyl-tert-leucine 0.48(#) 0.61(!) A
N -Glycylphenylalanine 0.51(L) 0.57(L) B
A, Methanol}water}acetonitrile (1 : 1 : 4, v/v); B, methanol}water}acetonitrile (5 : 5 : 3,

v/v). Development distance, 13 cm; saturated chamber.
Table 29 Resolution data for enantiomers of amino acids from A macrocyclic antibiotic, vancomycin, has been
brucine-impregnated plates used as a chiral mobile-phase additive for the separ-
ation of 6-aminoquinolyl-N-hydroxy succinimidyl
Sl. no. D-L-amino acid hRF pure L D L
carbamate (AQC) derivatized amino acids and dansyl
1. Alanine 53 18 53 amino acids on chemically bonded diphenyl-Frever-
2. -Aminobutyric acid sed-phase plates. Both the nature of stationary phase
3. Isoleucine 35 16 35 and the composition of the mobile phase have
4. DOPA
5. Leucine a strong inSuence on the enantiomeric resolution;
6. Methionine 29 18 29 typical results are given in Table 32. Another macro-
7. Norleucine cyclic antibiotic, erythromycin, has been used as
8. Phenylalanine 40 27 40 a chiral impregnating reagent for the resolution of 10
9. Serine 50 12 50
10. Threonine 29 16 29
dansyl-DL-amino acids on silica gel plates (Figure 2);
11. Tryptophan 31 17 31 hRF values and solvent combinations are shown
12. Tyrosine 29 22 29 in Table 33. Resolution of dansyl-DL-amino acids
has recently been reported (Table 34) on thin silica
Silica plates impregnated with (!)-brucine, developed in n -
gel plates impregnated with (1R, 3R, 5R)-aza-
butanol}acetic acid}chloroform (3 : 1 : 4, v/v), up to 10 cm.
bicyclo[3,3,0]octan-3-carboxylic acid, which is an
industrial waste material and a proline analogue non-
nitrile (70 : 30) is added to give a Rnal acetonitrile proteinogenic -amino acid.
concentration of 38% or 47%. For the second dimen-
sion, the mobile phase is 8 mol L\1 N,N-di-n-propyl- Quantitation
1
L-alanine and 4 mmol L\ copper(II) acetate dis-
1
solved in 0.3 mol L\ sodium acetate in H2O} TLC is supplemented with spectrophotometric
acetonitrile (70 : 30, pH 7); the plates are developed methods for the quantitation of amino acids and their
in the second dimension using a temperature gradi- PTH and DNP derivatives.
ent. The method is reported to be applicable to the
Amino Acids
resolution of amino acids in a protein hydrolysate
with quantitation by densitometry. The scraped layer corresponding to each spot is ex-
-Cyclodextrin (-CD) plates have been used suc- tracted with 70% ethanol in a known minimum vol-
cessfully for the resolution of enantiomers of dansyl ume, and ninhydrin reaction is performed followed
amino acids and -naphthylamide amino acids. The by spectrophotometry. Six to eight standard dilutions
plates are prepared by mixing 1.5 g of -CD bonded in an appropriate concentration range for each amino
silica gel in 15 mL of 50% methanol (aqueous) with acid are prepared; 2 mL of amino acid solution and
0.002 g of binder and acetate in 50/50 MeOH}1% 2 mL of buffered ninhydrin are mixed in a test tube,
aqueous triethyl ammonium acetate (pH 4.1). Some heated in a boiling-water bath for 15 min, cooled to
of the results are shown in Table 31. room temperature and 3 mL of 50% ethanol is ad-
ded. The extinction is read at 570 nm (or 440 nm for
proline) after 10 min. Standard plots of concentration
Table 30 hRF of pure and resolved enantiomers of PTH amino versus absorbance are drawn for each amino acid.
acids, for tartaric acid-impregnated plate Materials consist of standard solutions of amino
acids, acetate buffer (4 mol L\1, pH 5.5), ethanol
D,L Mixture of hRF of D (resolved) L (resolved) (50%), methyl cellosolve (ethylene glycol mono-
PTH amino acids pure L
methyl ether), and ninhydrin reagent (0.9 g ninhydrin
Met 83 18 83 and 0.12 g hydrantin dissolved in 30 mL methyl
Phe 85 15 85 cellosolve and 10 mL acetate buffer, freshly pre-
Try 95 95 pared). The concentration of the unknown sample is
Val 80 21 80
read from the standard plots. TLC densitometry can
Ile 92 15 92
Tyr 95 16 95 be used to determine 0.5 mg L\1 of phenylalanine in
Thr 85 30 85 blood as an indicator of phenylketonuria.
Ala 55 12 55
Ser 84 10 84 PTH Amino Acids
The quantitation of PTH amino acids is carried out in
Solvent, chloroform}ethyl acetate}water (28 : 1 : 1, v/v). Devel-
opment time, 35 min, solvent front, 10 cm, room temperature, situ or after elution. For in situ determination, the
25$13C. Impregnation with (#)-ascorbic acid resolved D,L mix- Suorescence-quenching areas of PTH derivatives are
tures of PTH-Met, Phe, Val, Ala, Ser. usually measured against the Suorescent background
Table 31 Separation data for enantiomeric compounds on -CD-bonded-phase plates
Compound, D,L mixture RF Mobile Detection

phase a method
D L
Dns-leucine 0.49 0.66 40/66 Fluorescence

Dns-methionine 0.28 0.43 25/75 Fluorescence
Dns-alanine 0.25 0.33 25/75 Fluorescence
Dns-valine 0.31 0.42 25/75 Fluorescence
Alanine--naphthylamide 0.16 0.25 30/70 Ninhydrin
Methionine--naphthylamide 0.16 0.24 30/70 Ninhydrin
a
Volume ratio of methanol to 1% triethylammonium acetate (pH 4.1).
at 254 nm. While using a Turner Suorometer Rtted ation has been carried out. The quantitation of PTH
with a door for scanning chromatoplates, the position amino acids has also been practised as follows: the
of the scanner, the standardization of time between developed chromatograms are exposed to iodine va-
scanning and the end of chromatography, the loading pours and the brownish spots scraped off, eluted with
volume, the developing distance and the layer thick- 95% ethanol or ethyl acetate, and the iodine removed
ness are the important inSuencing factors for repro- by warming the sample tubes in a warm-water bath.
ducibility. The quantitation of PTH amino acids is The optical densities are read at 269 and 245 nm,
also carried out by measuring their UV absorbance appropriate blank determinations are carried out,
after they have been eluted from the layer. The standard plots are drawn, and the concentration of
scraped layer is extracted with methanol overnight, the unknown sample is calculated.
centrifuged for 30 min at 300 rpm, and the spectra of
the extracts are recorded in the range from 320 nm to
about 230 nm. To obtain reproducible UV absorban-
ces the layers must be washed with methanol prior to
development, and with chloroform after the separ-
Table 32 RP-TLC enantiomeric separation using vancomycin

as chiral mobile-phase additive
Compound hRF Vancomycin

concentration
L D (mol L\1)
AQC-allo-iso-leucine 14 21 0.025
AQC-methionine 19 23 0.025
AQC-nor-leucine 13 16 0.025
AQC-nor-valine 21 25 0.025
AQC-valine 23 27 0.025
Dansyl-glumatic acid 21 23 0.04
Dansyl-leucine 03 09 0.04
Dansyl-methionine 05 12 0.04
Dansyl-nor-leucine 03 07 0.04
Dansyl-nor-valine 05 12 0.04
Dansyl-phenylalanine 03 05 0.04
Dansyl-serine 15 20 0.04
Dansyl-threonine 13 17 0.05
Dansyl-tryptophan 01 03 0.04
Dansyl-valine 06 10 0.04
Mobile phase, acetonitrile}0.6 mol L\1 NaCl (2 : 10). Plates de- Figure 2 Chromatogram showing resolution of Dns-DL-
veloped at room temperature (223C) in cylindrical glass cham- phenylalanine, valine and leucine. From left to right: 1, Dns-DL-
bers. Time, 1}3 h for 5;20 cm plates. Visualization, UV. AQC, phenylalanine; 2, Dns-L-phenylalanine; 3, Dns-DL-valine; 4, Dns-
6-Aminoquinolyl-N-hydroxysuccinimidyl carbamate, a fluorescent L-valine; 5, Dns-DL-leucine; 6, Dns-L-leucine. Developing solvent,
tagging agent; reaction mixture of AQC and amino acid was used aq. 0.5 mol L!1 sodium chloride #acetonitrile (15#1). Develop-
without purifying the derivatives. ing time 20}25 min. Detection 254 nm.
Table 33 hRF Values of enantiomers of dansyl amino acids resolved on plates with
erythromycin
Dansyl DL-amino acid Pure L hRF from DL mixture Solvent system

0.5 mol L\1 aq.
D L NaCl}MeCN}MeOH
(v/v)
Serine 64 68 64 10 : 4 : 1
30 36 30 15 : 1 : 1
Glutamic acid 45 56 45 22 : 1 : 0.5
56 65 56 22 : 1 : 0
52 59 52 26 : 1 : 0
Phenylalanine 50 65 50 15 : 2 : 0
20 27 20 15 : 1 : 0
Valine 22 30 22 15 : 1 : 0
Leucine 24 32 24 15 : 1 : 0
Tryptophan 38 47 38 18 : 1 : 0.25
Methionine 56 63 56 25 : 2 : 0.5
Aspartic acid 50 63 50 28 : 1.5 : 0.5
-Amino-n -butyric acid 42 51 42 12 : 1 : 0
Norleucine 63 71 63 16 : 1 : 0 : 0.4 HOAc
Temperature 25$23C. Solvent front, 10 cm. Time, 20}25 min. Visualization, UV,
254 nm.
DNP Amino Acids evaluated by measuring the optical density at 360 nm

or at 385 nm for DNP proline. For a blank, a similar
Use of direct Suorimetric quantitation (Suorescence extract is obtained from a clear spot on the same
quenching) in situ has been recommended. Silica gel layer.
G plates are developed in chloroform}benzyl alco-
hol}acetic acid (70 : 30 : 3 v/v) and n-propanol}am- Future Developments
monia (7 : 3 v/v) and polyamide plates are developed
in benzene}acetic acid (4 : 1 v/v). The spots are scan- TLC and HPLC are often looked at as competitive
ned using a Camag/Turner scanner, after being dried methods, but each has its own advantages. In HPLC,
in a stream of air, at the scanning speed of 20 mm Rnding suitable separation parameters is frequently
min\1 and an excitation wavelength of 254 nm. Al- costly in terms of time and materials; therefore,
ternatively, the layer is scraped off the plate and a combination of the two by Rrst optimizing the
extracted for 5 min, with 1 mL 0.05 mol L\1 Tris particular separation parameter with TLC is a step
buffer, pH 8.6, at room temperature. Then the slurry leading to considerable time-saving and cost for an
is removed by centrifugation and the clear liquid is analysis. TLC is suitable as a pilot technique for the
Table 34 Results from resolution of dansyl DL-amino acids
Dansyl DL-amino acid Pure L hRF from DL mixture Solvent system

0.5 mol L\1 aq.
D L NaCl}MeCN
(v/v)
Phenylalanine 50 65 50 15#2
Valine 38 49 38 15#1.5
Tryptophan 23 34 23 20#0.5
Aspartic acid 55 67 55 15#2
61 70 61 18#2
52 60 52 15#1
30 52 30 20#0.5
Leucine 64 68 64 10#4#1 MeOH
9#3#0.5 MeOH
Norvaline 56 61 56 17#2#0.4 MeOH
16#2#0.25 MeOH
Temperature 25$23C. Solvent front, 10 cm. Time, 25}30 min. Visualization UV, 254 nm.
III / AMINO ACIDS AND DERIVATIVES: CHIRAL SEPARATIONS 2033
investigation of appropriate separation conditions, Bhushan R and Reddy GP (1987) TLC of phenylthiohydan-
particularly because with TLC various phase systems toins of amino acids: a review. Journal of Liquid
can be checked at the same time without expensive Chromatography 10: 3497.
apparatus. TLC will continue to serve as a useful Bhushan R and Reddy GP (1989) TLC of DNP- and dansyl-
method for daily routine control analyses to identify amino acids: a review. Biomedical Chromatography 3:
and determine the purity of a variety of compounds, 233.
Grassini-Straza G, Carunchio V and Girelli M (1989) Flat
including enantiomers, with ease and speed, and can
bed chromatography on impregnated layers: review.
be readily modiRed for new situations. A wide choice Journal of Chromatography 466: 1}35.
for separation conditions will always be available as GuK nther K, Matrens J and Schickendanz M (1984) TLC
various phase systems can be checked simultaneously. enantiomeric resolution via ligand exchange.
Angewante Chemie International Edition in English. 23:
506.
Further Reading Kirchner JG (1978) Thin Layer Chromatography, 2nd edn.
New York: John Wiley.
Bhushan R and Martens J (1996) Amino acids and deriva- Rosmus J and Deyl Z (1972) Chromatography of N-ter-
tives. In: Sherma J and Fried B (eds) Handbook of Thin minal amino acids and derivatives. Journal of
Layer Chromatography, 2nd edn. New York: Marcel Chromatography 70: 221.
Dekker. Sherma J (1976 to 1996) Thin Layer Chromatography or
Bhushan R and Martens J (1997) Direct resolution of enan- Planar Chromatography: Review every two years. Ana-
tiomers by impregnated TLC. Biomedical Chromatogra- lytical Chemistry. Washington, DC: American Chemical
phy 11: 280. Society.
AMINO ACIDS AND DERIVATIVES:

CHIRAL SEPARATIONS
I. D. Wilson, AstraZeneca Pharmaceuticals, D-enantiomer is used to treat arthritis but the L-form
Macclesfield, UK is highly toxic, and the optical purity of the drug
R. P. W. Scott, Avon, CT, USA therefore clearly becomes an issue.
Copyright ^ 2000 Academic Press Another interesting reason for wishing to examine
the ratio of different amino acid enantiomers is that,
as a result of their slow racemization with time, it
provides another means of dating archaeological
Introduction samples. Other applications include the determina-
It is an interesting feature of life that in general its tion of the nature of the amino acids found in micro-
building blocks, whilst often containing chiral bial peptides and polypeptides where D amino acids
centres, are generally composed from optically pure are not uncommon (e.g. as constituents of certain
single enantiomers. An excellent, and well known, antibiotics).
example of this is provided by the amino acids as Chiral separations involve the resolution of indi-
those found in mammals are all of the L-form. This vidual enantiomers that are chemically identical and
being the case, why is there a need to resolve the only differ in the spatial distribution of their indi-
enantiomers of amino acids? vidual atoms or groups. As each isomer will contain
The chiral separation of amino acids is important the same interactive groups, the intermolecular forces
for a number of reasons. Perhaps the major reason for involved in their retention will also be the same.
the pharmaceutical industry is the need for optically Consequently, unless a second retention mechanism
pure amino acids, of the required conRguration, in is invoked, in addition to those involving inter-
order to prepare synthetic peptides, both for testing molecular forces, both enantiomers will exhibit iden-
and as potential new drugs. In this case methods are tical retention times on all stationary phases and
needed to determine optical purity, and measure remain unresolved. A variety of chromatographic
amounts of the unwanted enantiomer at the 0.1% separation strategies have been developed to obtain
level and for large-scale isolation for subsequent syn- the resolution of amino acids. These include gas,
thetic work. Another pharmaceutical example is pro- thin-layer and column liquid approaches. In the
vided by the sulfhydryl drug penicillamine where the case of liquid chromatography these methods have
2034 III / AMINO ACIDS AND DERIVATIVES: CHIRAL SEPARATIONS
included enantiomer separation via chiral stationary uenesulfonyl)-L-phenylalanine, etc.) and copper(II)
phases (CSPs; for a detailed treatment of the various ions to an aqueous/organic solvent. Factors which
stationary phase types the reader is directed to the affect the complex formation include the metal (as
Further Reading and relevant encyclopaedia entries), indicated above, this is usually copper but zinc, nickel
chiral mobile phases (generated by the addition of and mercury have also been used albeit with inferior
a chiral selector to the eluent) or derivatization with resolution), the metal ion/ligand ratio (usually 2 : 1),
a chiral reagent to form diastereoisomers. The meth- the concentration of the metal/ligand complex and
odology used will depend to a large extent on the pH. For practical applications the pH of the mobile
problem to be solved (e.g. analysis or preparative phase would normally be recommended to be in the
isolation) and each of these methodologies for amino range of 7}8 in order to be able to carry out
acids are detailed below. chromatography on conventional reversed-phase col-
umns (this pH preserves the integrity of the columns
Derivatization of Amino Acids to Form and higher pH values cause the precipitation of the
copper complexes).
Diastereoisomers As well as chiral ligand exchange, some use has
One of the earliest strategies to be implemented for been made of the ability of the cyclodextrins to form
the separation of amino acid is the formation of inclusion complexes with amino acid derivatives. The
diastereoisomeric derivatives using an optically pure cyclodextrins are produced by the partial degradation
chiral derivatizing reagent. These can then be separ- of starch followed by the enzymatic coupling of the
ated relatively easily on conventional stationary glucose units into crystalline, homogeneous toroidal
phases with achiral eluents. The major difRculty with structures of different molecular size. The three most
this approach is ensuring that the reagent is indeed widely characterized are the -, - and -cyclodextrins
100% optically pure and that racemization (of either which contain six (cyclohexamylose), seven (cyclo-
reagent or amino acid) does not occur during the heptamylose) and eight (cyclo-octamylose) glucose
derivatization reaction. Clearly, if attempting to de- units, respectively. These cyclic, chiral, torus-shaped
termine optical purity at the 0.1% level, even macromolecules contain the D(#)-glucose residues
a 99.9% pure reagent is not sufRcient. However, if bonded through -(1P4) glycosidic linkages. The
these conditions can be met, the methodology is easy mouth of the torus-shaped cyclodextrin molecule has
and robust. A huge range of chiral derivatizing re- a larger circumference than at the base and is linked
agents have been prepared and many of these can be to secondary hydroxyl groups of the C2 and C3 atoms
used for amino acids. These applications would in- of each glucose unit. The cyclodextrins provide a ubi-
clude the use of, for example, substances such as quitous means of separating enantiomers either as
Marfey’s reagent (1-Suoro-2,4-dinitrophenyl-5-L- mobile-phase additives or when used to make chiral
alanine amide), 2,3,4,6-tetra-O-acetyl-D-glucopyran- stationary phases (see below) and an example of this
osyl isocyanate (GITC) and similar compounds, or would be the use of -cyclodextrin as chiral mobile
the Suorescent 1-(9-Suorenyl)ethylchloroformate phase additive for the resolution of dansylated amino
(FLEC). In addition, it is possible to form highly acids on a conventional reversed-phase column (C8).
Suorescent diastereoisomeric isoindoles from amino
acids using O-phthaldialdehyde and a chiral thiol.
Whilst these examples are among the most common Chiral Stationary Phases for the
chiral derivatizing reagents, there are many others. Separation of Amino Acid Enantiomers
and their Derivatives
Chiral Selectors in the Mobile Phase There are a number of types of chiral stationary phase
An alternative to forming covalent derivatives is to that are used for the separation of amino acids and
employ chiral mobile phase additives that will act as their derivatives and these include ligand exchange
chiral selectors interacting selectivity with the differ- phases, protein-based phases, the Pirkle-type phases,
ent enantiomers of the amino acids to effect a separ- molecular imprints, coated cellulose and amylose
ation. derivatives, macrocyclic glycopeptide phases, and
For amino acids, separation by chiral ligand ex- cyclodextrin-based CSPs.
change has been of considerable importance. In this
Amino Acid Enantioseparation via Chiral Ligand
context a chiral mobile phase can be generated by
Exchange Phases
adding a chiral selector such as L-proline (or another
amino acid such as L-arginine, L-histidine or substan- The separation of amino acids on chiral ligand ex-
ces such as N,N-di-isopropyl-1-alanine or N-(p-tol- change columns represents one of the earliest
methods for the resolution of these compounds, both tween 4.5 and 8.0 and for example, in the case of the
free and as derivatives (e.g. dansylated). The original N-benzoyl-derivatized amino acids, increasing pH re-
work was performed by Rogozhin and Davankov sults in decreased retention. In general the lower the
using resins containing optically active bi- and buffer concentration (from 0 to 0.1 M) the better the
trifunctional -amino acids loaded with a metal ion retention; however, an effect of increased buffer con-
such as copper(II). More recently, more efRcient col- centration (above 0.2 M) has been observed for N-
umns have been prepared by bonding chiral amino benzoyl derivatives. An increase in organic modiRer
acid ligands to silica gel. It is also the case that by concentration reduces the hydrophobic interactions
using a long-chain alkyl-substituted chiral selector of the solutes with the column resulting in shorter
such as N-decyl-1-histidine to the mobile phase a ‘dy- retention times. Whilst very useful for the determina-
namically coated’ CSP can be prepared from a normal tion of, for example, enantiomeric purity, protein
reversed-phase column. In such cases it is still neces- phases tend to have rather limited sample loading
sary to continue to supply a small amount of the capacity.
chiral selector in the mobile phase to ensure that the
The Pirkle-Type Stationary Phases
ligand leached from the stationary phase is constantly
topped up. As with ligand exchangers used as mobile The so-called Pirkle stationary phases (named after
phase additives, the mechanism of retention involves their inventor W. M. Pirkle) consist of relatively small
the formation of complexes between the ligand (gen- molecular weight chiral substances covalently
erally based on L-proline), a metal ion (usually bonded to silica. Each bonded moiety contains a lim-
copper(II) and the amino acid itself. Separations are ited number of chiral sites (sometimes only one).
made using reversed-phase types of eluents. Because Nevertheless, on account of their small size, there will
of the ease of use of ligand exchange chromatography be a larger number of interactive groups bonded to
with chiral mobile phases on standard reversed-phase the silica and thus the probability of the solute inter-
columns, these may be more useful than dedicated acting with a chiral centre is still very high. In addi-
stationary phases. tion, as the interacting molecule is relatively small,
the extra-chiral contributions to retention are also
Amino Acid Enantioseparation via Protein-Based
comparatively small, and consequently the chiral in-
Stationary Phases
teractions themselves represent a higher proportion
The protein-bonded stationary phases were some of of the total. It follows that chiral selectivity becomes
the Rrst to be developed and usually consist of natural a more dominant factor controlling retention with the
proteins (e.g. bovine serum albumin, 1-acid glyco- Pirkle phases.
protein, ovomucoid, etc.) bonded to a silica matrix. The Pirkle phases have also been used very success-
Proteins contain a large number of chiral centres of fully for the separation of many free and derivatized
one conRguration and are known to interact strongly amino acids. The separation of the p-bromophenyl
with small chiral compounds for which they can derivatives of the enantiomers of a number of amino
exhibit strong enantiomeric selectivity. Some speciRc acids is shown in Figure 1. The stationary phase was
interactive sites on the protein provide the chiral a naphthyl urea Pirkle stationary phase multiply-
selectivity, but there are many others that generally bonded to the silica. All of the enantiomers were
contribute to retention. Protein columns based on separated and the analysis time was less than 50 min.
bovine serum albumin have been employed for the Elution was achieved by progressively increasing the
separation of the enantiomers of certain aromatic dispersive character of the mobile phase. Conse-
amino acids and various derivatives, including quently, the chiral selectivity was probably domin-
dansyl, N-(9-Suorenylmethoxycarbonyl)- (FMOC), ated by polar interactions.
N-(Suorescein thiocarbamoyl- (FITC) N-(2,4-dinitro-
Amino Acid Enantioseparation via Coated Cellulose
phenyl) and N-benzoyl. The use of the reagent N-
and Amylose Derivatives
(chloroformyl)carbazole to provide highly Suorescent
derivatives has enabled the resolution of the enantio- Another useful type of chiral stationary phase for
mers of all of the protein amino acids often with high amino acids and their derivatives is based on the
separation factors. Proteins have also been described polymers of cellulose and amylose. Usually the poly-
as showing remarkable enantioselectivity towards N- mers are derivatized to increase chiral selectivity or
acylated amino acids. improve stability. The derivatized cellulose or
The mobile phases employed for this type CSP amylose polymer is coated (not bonded) to a silica
are generally composed of phosphate buffers support. The fact that the chiral material is not
(0.1}0.2 M) modiRed with a limited amount of pro- bonded to the silica makes the material somewhat
pan-1-ol. The pH range normally employed is be- labile with respect to certain solvents.
2036 III / AMINO ACIDS AND DERIVATIVES: CHIRAL SEPARATIONS
phase mode, consequently, retention and selectivity

was again controlled by differential polar interac-
tions.
Amino Acid Enantioseparation via Macrocyclic
Glycopeptide Stationary Phases
There are three commonly used macrocyclic
glycopeptides and they are the antibiotics vanco-
mycin, teicoplanin an avoparcin all of which were
introduced as chiral stationary phases by Armstrong.
They contain a large number of chiral centres,
together with molecular cavities in which solute
molecules can enter and interact with neighbour-
ing groups. Vancomycin, for example, contains
18 chiral centres surrounding three ‘pockets’ or
‘cavities’ which are bridged by Rve aromatic rings.
Strong polar groups are proximate to the ring
structures that can offer strong polar interactions
Figure 1 The separation of a series of amino acid derivatives. with the solutes. This type of stationary phase is
The column was 10-cm long, 6 mm i.d. One mobile phase com- stable in mobile phases containing 100% organic
ponent (A) was 50 mM phosphate buffer (pH 6.0) and the second
component (B) pure acetonitrile. The gradient used was isocratic
solvent.
for 12 min 30% (B), then programmed from 12 to 29 min from The macrocyclic glycopeptides have a higher load-
30% (B) to 47% (B), then from 29 to 49 min 47}67% (B) and, ing capacity than the traditional protein phases and
finally, from 49 to 57 min, 67}93% (B). The flow rate was are more stable. They can also tolerate a much wider
1 mL min\1. 1, L-serine; 2, D-serine; 3, L-threonine; 4, D-threonine; range of solvents than the cellulose and amylose
5, L-alanine; 6, D-alanine; 7, L-valine; 8, D-valine; 9, L-methionine;
10, D-methionine; 11, L-leucine and isoleucine; 12, D-leucine and
phases.
isoleucine; 13, L-tyrosine; 14, L-phenylalanine; 15, D-tyrosine; 16, The macrocyclic glycopeptide stationary phases
D-phenylalanine; 17, L-tryptophan; 18, D-tryptophan; 19, L-lysine; can also be used very effectively for the separation of
20, D-lysosine; 21, L-cystine; 22, D-cystine. (Courtesy of Iwaki K, amino acids and their derivatives. The separation of
Yoshida S, Nimura N and Kinoshita T (1987) Optical resolution of the isomeric bromophenylalanines as their FMOC
enantiomeric amino acid derivatives on a naphthylethylurea mul-
tiple-bonded chiral stationary phase prepared via an activated
derivatives formed by reacting them with 9-Suorinyl-
carbamate intermediate. Journal of Chromatography 404: methylchloroformate is shown in Figure 3. The two
117}122.) enantiomers are very well separated indicating that
the chiral selectivity of the telcoplanin stationary
phase was extremely high. It should be noted, that the
Both cellulose and amylose contain Rve chiral ‘pure’ (S) enantiomer actually contained a signiRcant
centres per unit and thus the polymeric material of- amount of the (R) enantiomer. The macrocyclic
fers a large number of chirally interactive centres and glycopeptide stationary phases often exhibit high
high probability of interaction. There are basically selectivity for chiral substances of biological origin,
two common types of cellulose and amylose deriva- perhaps due to their being biological products them-
tives that are used as stationary phases. The Rrst type selves.
are simple esters usually formed from the acid chlor-
ides such as acetyl chloride or benzoyl chloride. The
more stable, and probably the more popular deriva-
tives, are the carbamates which can be synthesized
from the appropriate isocyanate. The most useful
cellulose- and amylose-based chiral stationary phases
are probably those derivatized with the different sub-
stituted tris(3,5-dimethylphenylcarbamates). An
example of the separation of N-benzyloxycarbonyl
alanine ethyl esters on cellulose tris(3,5-dimethyl-
phenylcarbamate) is shown in Figure 2.
The column was 25-cm long, 4.6-mm i.d., and the
mobile phase was hexane}2-propanol (90 : 10 v/v). Figure 2 The separation of N-benzyloxycarbonyl alanine ethyl
The stationary phase was operated in the normal ester on cellulose tris(3,5-dimethylphenylcarbamate).
removed it leaves a cavity capable of ‘recognizing’

and selectively rebinding the imprinted compound.
This property allows discrimination between enantio-
mers and has been used as the basis for the develop-
ment of CSPs for the highly selective separation of
amino acid derivatives (e.g. dansyl, anilide, BOC-1-
amino acid anilides, etc.) and a number of examples
of this type of separation have been published. A typi-
cal example would be the use of a molecular imprint
to the amino acid derivative L-phenylalanine anilide
for the resolution of a mixture of the two enantiomers
of the print molecule. In this case the more retained
enantiomer is the L-form of the amino acid derivative
as it exhibits a greater afRnity for the stationary
phase. In general the imprinted polymers are most
selective for the particular print molecule used to
prepare them. However, there are examples of the
separation of enantiomers of non-imprint molecules
as well.
Figure 3 The separation of the enantiomers of 2-bro- Conclusions

mophenylalanine and 3-bromophenylalanine. (A) A ‘pure’ sample
of the S enantiomer of FMOC 2-bromophenyl alanine. (B) A ra- As shown above there are various means for sep-
cemic mixture of FMOC 3-bromophenylalanine. The separation arating the enantiomers of amino acids and their
was carried out on a CHIROBOTIC T (teicoplanin) column, 25-cm
long, 4.6-mm i.d., packed with 5 m particles. The mobile phase
was programmed from methanol}1% triethylamine acetate (pH
4.5) (40 : 60 v/v) to methanol}1% triethylamine acetate (pH 4.5)
(60 : 40 v/v) over 20 min. The flow rate was 1.0 mL min\1 and the
sample was injected as a solution in acetone. (Courtesy of
Chirotech Technology Ltd.)
Amino Acid Enantioseparation via

Cyclodextrin-Based Chiral Stationary Phases
In addition to the use of cyclodextrins as mobile
phase additives discussed above, they have also been
widely used for the preparation of CSPs. For this the
three cyclodextrins, , and are bonded to a suit-
able support such as silica. An example of their use
for the separation of three racemic N-t-Boc-amino
acids is shown in Figure 4. It is seen that a very clean
separation of the enantiomers is obtained. Other
examples include the use of -cyclodextrin columns
for the resolution of dansylated amino acid deriva-
tives and -cyclodextrin columns for the separation of
a variety of natural and synthetic amino acids and
their derivatives.
Amino Acid Derivative Enantioseparation via Figure 4 The separation of the enantiomers of three N-t-Boc-
Molecular Imprints amino acids. The column used was 25-cm long packed with
Cyclobond 1 RSP and operated at a mobile phase flow rate of
Molecular imprinted polymers (MIPs) are produced 1.0 mL min\1 at a temperature of !22u( 3C. The mobile phase
by preparing a polymer (usually prepared from was 7% v/v acetonitrile}93% v/v% buffer (1% triethylamine, pH
a methacrylic acid, styrene or 4-vinylpyridine mono- 7.1) and the separation was monitored with a UV detector at
225 nm. (Courtesy of San Chung Chang, Wang LR and
mer template cross-linked with ethylenedimethyl- Armstrong DW (1992) Facile resolution of N-tert-butoxycarbonyl
methacrylate) in the presence of an imprint, or tem- amino acids: the importance of enantiomeric purity in peptide
plate, molecule. When the template is subsequently synthesis, Journal of Liquid Chromatography 15: 1411}1429.)
2038 III / AMINO ACIDS AND PEPTIDES: CAPILLARY ELECTROPHORESIS
derivatives. These range from indirect methods such lary gas chromatography separation of enantiomers.
as the formation of diastereoisomeric derivatives or Analytical Chemistry 62: 914}923.
direct methods that exploit the spatial characteristics Armstrong DW, Tang Y, Chen S, Zhou Y, Bagwill C and
of different enantiomers by making them interact Chen JR (1994) Macrocyclic antibiotics as a new class of
with a chiral stationary or mobile phase. This selec- chiral selectors for liquid chromatography. Analytical
Chemistry 66: 1473}1484.
tively enhances the standard free entropy of distribu-
Beesley TE and Scott RPW (1998) Chiral Chromatography.
tion of one amino acid enantiomer compared to the New York: John Wiley.
other and can provide adequate chiral selectivity to Iwaki K, Yoshida S, Nimura N and Kinoshita T (1987)
permit enantiomeric resolution. By one or other of Optical resolution of enantiomeric amino acid deriva-
these approaches the separation of the enantiomers of tives on a naphthylethylurea multiple-bonded chiral sta-
the majority of naturally occurring amino acids can tionary phase prepared via an activated carbamate inter-
be achieved by liquid chromatography. mediate. Journal of Chromatography 404: 117}122.
Kempe M and Mosbach K (1995) Separation of amino
See also: II/Chromatography: Liquid: Derivatization. acids, peptides and proteins on molecularly imprinted
III/Chiral Separations: Capillary Electrophoresis; Cellu- stationary phases. Journal of Chromatography A 691:
lose and Cellulose Derived Phases; Chiral Derivatization; 317}323.
Countercurrent Chromatography; Crystallization; Cyc- Lam S (1989) Chiral ligand exchange chromatography. In:
lodextrins and Other Inclusion Complexation Approaches;
Lough J (ed.) Chiral Liquid Chromatography, pp.
Gas Chromatography; Ion-Pair Chromatography; Ligand
83}101. Glasgow: Blackie.
Exchange Chromatography; Liquid Chromatography; Mo- Okamato Y (1986) Optical resolution of -blockers by
lecular Imprints as Stationary Phases; Protein Stationary HPLC on cellulose triphenylcarbamate derivative.
Phases; Synthetic Multiple Interaction (‘Pirkle’) Stationary Chemical Letters 1237}1240.
Phases; Supercritical Fluid Chromatography; Thin-Layer Okamato Y, Kaida Y, Aburantani R and Hatada K (1989)
(Planar) Chromatography.
Optical resolution of amino acid derivatives by high-
performance liquid chromatography on tris(phenylcar-
Further Reading bamate)s of cellulose. Journal of Chromatography 477:
Ahnoff M and Einarsson S (1989) Chiral derivatisation. In: 367}376.
J Lough (ed.) Chiral Liquid Chromatography, pp. Pirkle WH and House DW (1979) Chiral high pressure
39}80. Glasgow: Blackie. liquid chromatographic stationary phases. 1. Separation
Allenmark S, Bromgren B and Boren B (1984) Direct liquid of the enantiomers of sulphoxides, amines, amino acids,
chromatographic separation of enantiomers on immobi- alcohols, hydroxyacids, lactones and mercaptans. Jour-
lized protein stationary phases. IV. Molecular interac- nal of Organic Chemistry 44: 1957}1960.
tion forces and retention behaviour in chromatography San Chun Chang, Wang LR and Armstrong DW (1992)
on bovine serum albumin as a stationary phase. Journal Facile resolution of N-tert-butoxycarbonyl amino acids:
of Chromatography 316: 617}624. the importance of enantiomeric purity in peptide syn-
Allenmark S (1991) Chromatographic Enantioseparation: thesis. Journal of Liquid Chromatography 15: 1411}1429.
Methods and Applications, 2nd edn. Chichester: Ellis Skidmore MW (1993) Derivatisation for chromatographic
Horwood. resolution of optically active compounds. In: Blau K and
Armstrong DW, Li W and Chang CD (1990) Polar-liquid, Halket J (eds) Handbook of Derivatives for Chromatog-
derivatised cyclodextrin stationary phases for the capil- raphy, 2nd edn., pp. 215}252. Chichester: John Wiley.
AMINO ACIDS AND PEPTIDES:

CAPILLARY ELECTROPHORESIS
P. Bohn, Institut fu( r Instrumentelle Analytik/ lished by sequencing techniques and the analysis of
Umweltanalytik, Universita( t des Saarlandes, amino acid composition. Irregularities in the struc-
Saarbru( cken, Germany tural characteristics of proteins, e.g. after translation
Copyright ^ 2000 Academic Press of the protein, are determined by fragmentation to
smaller peptides. Progress in the Reld of synthetic
peptides utilizing synthesis based on these partial
Introduction sequences depends on their immunological potential.
Advancement in modern biotechnology is mainly at- The design of new specialized biomolecules such as
tributed to a detailed understanding of the structural hormones or neurotransmitters will have consider-
features of proteins. This is predominantly accomp- able pharmaceutical applications.
III / AMINO ACIDS AND PEPTIDES: CAPILLARY ELECTROPHORESIS 2039
The large number of analytes and the small quant- the anode, whereas lower pH values induce cations
ities present in biological samples have increased the which migrate with the EOF towards the cathode.
demand for speciRc and sensitive analytical tech- Most amino acids lack suitable physical character-
niques. Capillary electrophoresis (CE) is capable of istics that can be exploited for detection. Only few
handling small sample volumes down to microlitre species possess aromatic groups with high absorptiv-
size with only a few nanolitres injected. High efRcien- ity, e.g. try, phe and tyr. In order to analyse native
cies, short analysis time and easy enantiomeric assays amino acids three strategies can be pursued. UV de-
make CE an indispensable tool in the modern analysis tection can be used at low wavelengths. A second
of peptides and amino acids. approach is the application of indirect detection tech-
niques. Detection concepts involving derivatization
technology, especially Suorescence labelling, can also
Amino Acids improve detection sensitivity.
Physicochemical Properties Electrophoretic Systems ^ Separation Strategies
In choosing an electrophoretic system it is important Analysis of native amino acids Direct UV detection
to consider both matrix and the structural features of at wavelengths below 220 nm takes advantage of the
the analytes. Whereas 18 amino acids are found after absorptivity of the carbonyl bond. Detection at such
the hydrolysis of proteins, more than 50 derivatives nonspeciRc wavelengths requires highly transparent
are present in physiological Suids. buffers. Borate and phosphate are convenient electro-
Amino acids are small, highly polar species. The lyte systems. Selectivity is mainly achieved by the
individual species only differ in the residues R. Except optimization of pH because the analysis is performed
for glycine this situation induces a chiral centre at the with the native species.
-C-atom where two enantiomers (R-, S-) can be In order to obtain cationic analytes, pH has to be
distinguished. Classifying these residues R by their adjusted to values lower than the Rrst dissociation
impact on electrophoretic behaviour means a differ- step (pK&2). The stability of fused-silica capillaries
entiation by their polarity or the generation of charge. is restricted to pH values above 2.5. Thus basic condi-
Due to their zwitterionic nature, amino acids possess tions with analytes migrating counter to the EOF are
isoelectric points (pI); pH values equal to pI yield preferred. This separation mode beneRts by prolong-
molecules without net charge and therefore no migra- ing the effective separation distance, keeping the elec-
tion occurs in an electrical Reld. At pH values above trical Reld strength constant so that higher resolution
the pI the molecules are negatively charged and mi- is achieved. Limits of detection are in the range of
grate against the electroosmotic Sow (EOF) towards about 10\4 mol L\1 (Figure 1).
Figure 1 Separation of amino acids and dipeptides in an infusion solution using direct detection at low wavelength. Capillary: fused
silica 75 m i.d., 65/73.5 cm; buffer: borate 40 mmol L\1, pH"11.0; E"408 V cm\1, 191 nm; injection 50 mbar, 5 s. 1, Lys; 2, Pro; 3,
Try; 4, Leu; 5, Ile; 6, Gly-Glu; 7, Val; 8, Phe; 9, His; 10, Met; 11, Ala; 12, Thr; 13, Ser; 14, Gly-Tyr; 15, Glu; 16, Asp.
Indirect UV detection was evolved for the analysis As a consequence of derivatization, amino acids
of small inorganic ions but it is also an efRcient change from small ionic species to large hydrophobic
technique for analysis of a broad range of nonabsor- molecules. Differences in mobilities decrease. A sufR-
bing components. This methodology is performed cient separation selectivity is mainly achieved by
very easily with CE using a UV-absorbing electro- micellar electrokinetic capillary chromatography
lyte. With respect to dissociation behaviour, mobility (MEKC).
and absorptivity the background electrolyte (BGE) Many reagents have been investigated to improve
has to be chosen carefully. As mentioned above, basic sensitivity as well as suitability for Suorescence detec-
conditions should be applied to generate anionic spe- tion. Depending on the separation problem, further
cies of amino acids. Therefore the BGE has to be requirements have to be considered. The reagent must
negatively charged under alkaline conditions. Beside react quantitatively and reproducibly with primary
generating the background signal the nature of the and secondary amines to form stable products. Side
electrolyte used has great inSuence on separation reactions and Suorescence of the tag itself can inter-
selectivity. Best resolution can be achieved with elec- fere with the analysis. The choice of derivatizing
trolytes of moderate mobility, e.g. salicylic acid agent is limited by these prerequisites.
(pH"11.5"!6;10\4 cm2 V\1 s\1). Salicylate at The commonest applied systems are discussed be-
low mmol L\1 concentrations may also be used for low (Figure 3).
indirect Suorescence detection. Concentration limits The classical agent ninhydrin is not used for de-
are in the range of 10\5 mol L\1 (Figure 2). rivatization in CE because the aldehydes formed can-
Another approach to a universal, high sensitivity not be separated.
detection scheme is mass spectrometry (MS). Beside O-phthaldialdehyde (OPA) was one of the Rrst
the very low limits of detection which are achievable, reagents developed for pre-column derivatization in
this technique provides information about molecular liquid chromatography (LC). Strongly absorbing
mass and structure. The compatibility of capillary isoindoles with Suorescence properties are formed in
zone electrophoresis (CZE) to MS can be attributed a rapid reaction. The stability of the derivatives main-
to the low Sow rates in CZE. The main problem in ly depends on the amino acid and the reducing agent,
coupling CZE to MS is the buffer. Further develop- e.g. thiols. Unfortunately, secondary amines are not
ments on suitable volatile buffers and interface types derivatized. An increase in stability and detection
will extend the scope of applications. sensitivity has been achieved by using naphthalene-
2,3-dicarboxaldehyde (NDA) or 3-(4-carboxyben-
Analysis of derivatized amino acids Many of the zoyl)-2-quinolinecarboxaldehyde (CBQCA).
chemical reactions for labelling originate from pep- Phenylthiohydantoins (PTH) of amino acids are
tide synthesis where they were used as protective generated during Edman degradation of peptides.
groups or sequencing agents. Maximum absorbance is found at 254 nm but the
Figure 2 Separation of amino acids and dipeptides using indirect detection. Capillary: fused silica 75 m i.d., 86.5/95 cm; buffer:
salicylic acid 5 mmol L\1; pH"11.5; E"316 V cm\1, 214 nm; injection 50 mbar, 5 s. 1, Lys; 2, Pro; 3, Try; 4, Gly-Glu; 5, Leu; 6, Ile; 7,
Val; 8, His; 9, Met; 10, Ala; 11, Thr; 12, Asn; 13, Ser; 14, Gly; 15, Tyr; 16, Ac-Tyr; 17, Cys-Cys; 18, Ac-Cys; 19, Glu; 20, Asp.
Figure 3 Structures of derivatizing reagents. OPA, o-Phthalaldehyde; NDA, naphthalene-2,3-dicarboxaldehyde; CBQCA, 3-(4-
carboxybenzoyl)-2-quinolinecarboxaldehyde; PITC, phenylisothiocyanate, DNS, 5-dimethylaminonaphthalene-1-sulfonyl chloride;
DABS, dimethylaminoazobenzenesulfonyl chloride; FMOC, 9-fluorenylmethyl chloroformate; FLEC, (R) (S)-1-(fluorenyl) ethyl chloro-
formate.
derivatives lack Suorescence. Analysis is performed chlorides. FMOC amino acids Suoresce strongly and
using phosphate or borate buffers under alkaline con- are stable at room temperature. Detection sensitivi-
ditions. Surfactants such as sodium dodecyl sulfate ties in the nmol L\1 range can be achieved.
(SDS) give a micellar pseudo-stationary phase allow- Beside Suorescence detection, further improve-
ing the partition process. In contrast to cationic sur- ments in sensitivity and speciRcity can be obtained
factants, e.g. dodecyltrimethylammonium bromide with laser-induced Suorescence (LIF) techniques.
(DTAB), analytical systems using anionic surfactants A prerequisite is the match of emission wavelengths
beneRt from a wider migration time window. This of the derivatized analyte with the spectral lines of
can be mainly attributed to their counterosmotic mi- the lasers. Great effort has been invested in the devel-
gration behaviour (Figure 4). opment of new Suorophores such as TRTC, CTSP,
Sulfonyl chlorides can convert primary as well as TBQCA, IDA and CBQ (Table 1).
secondary amines. Well-known representatives are Unfortunately, most of them are not commercially
dansyl (DNS) and dabsyl (DABS) chloride. In order to available.
separate all DNS amino acids, acidic buffers are used Different derivatization techniques are applied: pre-
to reduce the EOF. In addition, neutral surfactants column tagging is the commonest method. Several
such as TWEEN 20 have been applied. The main attempts have been made to transfer post-column
disadvantage is the prolonged analysis time of about methodology from LC to CE. A further promising
70 min. Faster separations can be achieved using SDS technique is derivatization in the capillary because it
with the penalty of a decrease in resolution. In some simpliRes automation. Reagent and sample are injec-
cases resolution can be enhanced by operating at ted in succession. With the tandem mode a plug of
lower temperatures (Figure 5). reagent is injected into the column followed by the
Carbonyl chlorides such as Suorenylmethyl chloro- sample. A second technique is the introduction of an
formate (FMOC) are more reactive than sulfonyl additional plug of reagent after the sample (sandwich
Figure 4 Separation of 20 PTH amino acids by MEKC. Capillary: fused silica 50 m i.d., 59/67.5 cm, buffer: phosphate 25 mmol L\1;
SDS 25 mmol L\1; pH"9.0; E"444 V cm\1, 260 nm; injection 50 mbar, 5 s. 1, Thr; 2, Asn; 3, Ser; 4, Gln; 5, Asp; 6, Gly; 7, Ala; 8, His;
9, Glu; 10, Tyr; 11, Cys; 12, Pro; 13, Val; 14, Met; 15, Leu; 16, Ile; 17, Try; 18, Phe; 19, Lys; 20, Arg.
mode). After a speciRed time for reaction, the separ- The most widely applied cyclodextrins form
ation can be performed. host}guest complexes with one of the enantiomers
preferentially. Compared to migration in the bulk
Chiral analysis Assays of enantiomeric purity are phase, the complexed species possesses a different
easily performed by CE by simply adding the chiral mobility. The separation occurs due to different
selector to the running buffer. Two different method- complex stabilities resulting in different migration
ologies are applied to achieve resolution. First, chiral velocities. Enhancement of enantioselectivity is prim-
distinction can be established by the formation of arily attributed to cavity size (-, -, -cyclodextrin
non-covalently bonded diastereomers. (CD)) and derivatization of the hydroxy moieties of
Figure 5 Separation of DNS amino acids by MEKC in an infusion solution. Capillary: fused silica 50 m i.d., 50/57.5 cm, buffer: borax
20 mmol L\1 SDS 102.5 mmol L\1; pH"9.1; E"435 V cm\1, 214 nm; T"7.53C; injection 50 mbar, 5 s. 1, Thr; 2, Ser; 3, Ala; 4, Gly;
5, Glu; 6, Val; 7, Pro; 8, Met; 9, Ile; 10, Leu; 11, Phe; 12, Try; 13, Arg; 14, His; 15, Tyr; 16, Di D-Lys.
Table 1 Examples of derivatizing reagents and detection wavelengths
(mol L\1 cm\1) abs/ex (nm) em (nm) Remark
OPA 334 455 Presence of reducing agents (thiols), strong absorbance,

strongly fluorescence, unreacted OPA not fluorescent, de-
rivatives lack of stability, reaction rapid
NDA 462 490 Reaction rapid, increased stability compared to OPA,
CBQCA 450 recently commercially available
PITC 254 Peptide sequencing by Edman degradation, cyclic
thiohydantoins; no fluorescence properties
DNS 14 100 254 570 Problems with derivatization by-products
DABS 420-450
FMOC-Cl 265 315 Fluorogenic derivatives with primary/secondary amines,
strong absorbance
FLEC 265 310 Converts enantiomers to diastereomers
TRTC '100 000 540 567 Ex at 540 nm matches with emission line of low cost HE
laser
CTSP 82 000 663 687 Semiconductor laser
TBQCA 465 550
IDA 33 100 409 482
CBQ 466 544
OPA, o-Phthalaldehyde; NDA, naphthalene-2,3-dicarboxaldehyde; CBQCA, 3-(4-carboxybenzoyl)-2-quinolinecarboxaldehyde; PITC,

phenylisothiocyanate; DNS, 5-dimethylaminonaphthalene-1-sulfonyl chloride; DABS, dimethylaminoazobenzenesulfonyl chloride;
FMOC, 9-fluorenylmethyl chloroformate; FLEC, (R) (S)-1-(-fluorenyl)ethyl chloroformate; TRTC, tetramethylrhodamine isothiocyanate;
CTSP, pyronin succinimidyl ester; TBQCA, 3-(4-tetrazolebenzoyl)-2-quinolinecarboxaldehyde; IDA, 1-methoxycarbonylinodolizine-
3,5-dicarbaldehyde; CBQ, 3-(p-carboxybenzoyl) quinoline-2-carboxaldehyde.
the cyclodextrin (methyl-, hydroxypropyl-, sul- osyl isothiocyanate) allow the application of non-
fobutyl-CD). Whereas compounds with a single aro- chiral separation techniques. Detection sensitivity can
matic core Rt into -CDs, -CDs mainly form com- be improved simultaneously by using reagents like
plexes with polynuclear aromates such as tyr or try. FLEC ((R) or (S)-(1-Suorenyl)ethyl chloroformate)
Larger structures are accommodated by -CDs. Most containing chromophores or OPA with a chiral
of the enantiomeric separations are performed using thiol.
phosphate or borate electrolytes with native - or
-CD or mixed MEKC-CD systems which addition- Peptides
ally contain a surfactant, mostly SDS.
Additives like urea or small amounts of organic Peptides are compounds of great medical interest due
solvents can improve the resolution. to their physiological role as hormones and neuro-
Chiral surfactants such as N-dodecanoyl-L-serine transmitters. Furthermore, considering peptides as
(SDVal) or N-dodecanoyl-L-glutamate (SDGlu) have subunits of proteins, peptide mapping after chemical
been investigated. These amino acids with hydropho- or enzymatical cleavage allows characterization of
bic alkyl chains are applied in a mixture with the protein and to reveal metabolic disorders.
nonchiral surfactants, e.g. SDS.
Physicochemical Nature of Peptides
Metal ions of copper(II), zinc(II) or cobalt(III) can
be added to the electrolyte containing an L-isomer of The characteristics of peptides are situated between
an amino acid, e.g. L-proline, L-histidine or a dipep- those of amino acids and high molecular weight pro-
tide, e.g. aspartame. These metal}amino acid or teins. Oligopeptides containing up to 15 amino acids
metal}dipeptide complexes preferentially form a ter- behave similarly to amino acids. Short peptide chains
nary complex with one enantiomer of the amino acid cannot create a complicated conformation. In con-
in the sample. Separation occurred due to different trast, very long polypeptides with chain lengths up to
complex stabilities resulting in different mobilities for approximately 100 units behave like small proteins.
the individual enantiomers. They exhibit characteristic features of secondary and
As a second approach, a racemic mixture of amino tertiary structure.
acids is derivatized with an optically pure reagent Peptides exist in aqueous solution as amphoteric
yielding covalent-bonded diastereoisomers. Reagents ions. Therefore peptides possess isoelectric points
like GITC (2,3,4,6-tetra-O-acetyl--D-glucopyran- (pI). The peptide has net electroneutral properties at
the pI. The zwitterionic characteristics are inSuenced The following equation results from the semi-
predominantly by the acidity of the medium. In acidic empirical approach by Offord relating electro-
media the carboxyl group (pKa&2.7}2.9) is proto- phoretic mobility of peptides with their charge
nated and the peptide behaves as a cation. In alkaline q and their molecular mass M:
media the protonated ammonium group is eliminated
and the zwitterionic form is converted into an anion. "k;q;m\2/3
The degree of dissociation is determined by the dis-
This linear relationship has been validated by experi-
sociation constants of the functional groups yielding
mental research; a large set of analytes covering
different net charges.
collagen fragments, tryptic digest of human growth
A further feature to be considered in elec-
hormone (33 peptides), motilin fragments (24 pep-
trophoretic behaviour is the sequence of the amino
tides) and many additional peptides differ widely in
acids. The dissociation constants of the individual
charge and amino acid sequence. Nowadays several
residues are affected by the arrangement of the amino
computer programs are available which are capable
acids in the chain. Mass-to-charge ratios are altered
of calculating the charge-to-mass ratio just requiring
and the peptide exhibits a different mobility.
the amino acid sequence.
Prediction of Electrophoretic Mobility of Peptides
Electrophoretic Systems ^ Separation Strategies
A theoretical model of electrophoretic migration un-
derlying the experimental approach can be very use- To optimize the separation of peptides, the experi-
ful for the optimization of analytical conditions. It mental conditions have to be adjusted to emphasize
supports predicting peptide mobilities under different differences in the charge-to-mass ratios of the
experimental conditions such as pH. If selectivity analytes.
between closely migrating species has to be imple- Apart from external parameters like electrical Reld,
mented, the model facilitates adjustment of the separ- capillary dimensions (length, inner diameter) and
ation environment. temperature, separation is mostly inSuenced by the
Furthermore, considering technical processes and electrolyte. Intrinsic variables like type of buffer, mo-
purity control of peptide synthesis or enzymatic di- bility, ionic strength, pH and buffer additives deter-
gestion of proteins, a deRned relationship between mine electrophoretic and electroosmotic mobility.
apparent mobility and physicochemical parameters In the Rrst place selectivity in the analysis of pep-
supports the identiRcation of unknown species. Vari- tides is controlled by pH. Altering the acidity of the
ation in the sequence of peptides can also be easily separation medium affects both the charge of the
determined. peptide and the ionization of the capillary wall,
The mathematical description of the migration pro- resulting in the change in EOF.
cess is based on the contribution of two forces. The The hysteresis-like course of the EOF shows the
electrical Reld accelerates an ion with a force propor- greatest variation in the pH range of approximately
tional to its charge. In addition, the ion is inSuenced 5}7, i.e. near the dissociation constant of the silanol
by a retarding force which results from the viscosity groups. For pH values below 3 or greater than 9, the
of the medium and is connected to a size parameter. inSuence of the superimposed EOF can be neglected
For permanently ionized small ions a prediction of and the migration of the peptide is almost indepen-
migration is easily achieved by applying Stoke’s law dent from the EOF.
which correlates mobility with q;r\1 and q;M\1/3 In acidic media (pH&2) both basic and acidic
(q, charge and M, molecular mass). With larger, more residues of the peptide are protonated. Selectivity is
complex aggregates like peptides both the charge and attributed to the number of positive-charged am-
a suitable size parameter has to be ascertained. For monium groups in the chain resulting in different
computing the charge, the sequence of amino acids charge densities. Analytes migrate with the EOF. In
has to be considered as the environment of a residue high pH buffers (pH&10), deprotonation of ter-
affects the extent of ionization, e.g. neighbouring minal and side chain ammonium groups (His) induces
amide bond or acidic/basic residues at terminal negatively charged species (presence of carboxylate
amino or carboxylic groups. This means that the groups) which migrate in the opposite direction to the
ionization constants of the free amino acids have to EOF. At higher pH values the side chain amino
be adjusted. During the development of a theoretical groups of arg and lys are the only ones affected.
understanding of migration phenomena, many ap- Optimization of pH values below 2 and above 12 is
proaches have been made considering mass, surface, difRcult to achieve since the limiting values of mobil-
radius and the number of units in a peptide chain as ity are reached. Furthermore, due to the high conduc-
size parameter. tivities of protons and hydroxyl ions, high currents
accompanied by Joule heating are generated. For crease of Joule heating can give rise to band broaden-
practical purposes selectivity control for peptides ing.
with a majority of acidic moieties is mainly achieved Dispersive effects caused by the interaction with
in the range of pH 3}6 while basic residues are mostly the capillary wall are usually not a problem with
affected at pHs around 10. peptides but larger species can exhibit characteristics
Additionally, isoelectric points of the peptides have similar to proteins in that they tend to adsorb at the
to be included in the optimization strategy. capillary wall.
If peptides are obtained by chemical or enzymatic High ionic strength, extreme pH values and buffer
digests of proteins the cleaving agent has to be con- additives competing in adsorption with the peptides
sidered, e.g. trypsin cuts at the C-terminal side of lys are strategies of optimization which can be adapted
and arg respectively. Thus fragments contain an ex- from protein analysis. At extreme pH values, peptides
cess of acidic residues. Selectivity can be easily affec- and the capillary wall are equally charged so electros-
ted in acidic media. Cleavage at aromatic or aliphatic tatic repulsion diminishes adsorption. Coated capil-
side chains is performed with chymotrypsin or pep- laries have been used to suppress this phenomenon.
sin, yielding fragments with both acidic and basic High salt content in the sample may destroy the
residues and optimization can be extended to the full separation efRciency of the electrophoretic system so
pH range (Figure 6). sample preparation steps must remove the high ionic
Frequently used electrolytes for peptide mapping strength in the sample.
are phosphate, citrate and acetate as acidic buffers Enhancement in selectivity can be attained if an
while borate or TRIS/Tricine are mainly applied un- additional equilibrium is superimposed on to the elec-
der basic conditions. Phosphate and citrate are buf- trophoretic process. Mostly the additives used for this
fers that can be used over a broad pH range due to are complexing agents which interact with speciRc
their multiple association constants. Borate exhibits groups of the peptide.
very low conductivity compared to phosphate and As for amino acids, metal ions can be employed for
other buffers. Buffer concentrations in the range of the separation of peptides and histidine-containing
10 mmol L\1 to approximately 100 mmol L\1 can be peptides especially interact with zinc salts. Separation
used. The electrolytes used should not possess any UV of two histidine dipeptides (L-L, D-L ) can be attributed
absorbance at low wavelengths. to favourable steric arrangement of the histidine resi-
An increase in ionic strength generates sharper dues in one isomer.
peaks (zone focusing) due to the drop of the electrical Cyclodextrins form dynamic inclusion complexes
Reld at the sample}electrolyte boundary and sample with hydrophobic parts of the peptide, e.g. with
loading capacity can be increased. High ionic amino acid residues containing aromatic rings like
strengths induce high electrical currents and the in- phenylalanine. The mass of a complexed analyte is
Figure 6 Tryptic digest of a haemoglobin variant separated by CZE. Capillary: poly(vinyl alcohol) coated fused silica capillary 50 m
i.d., 50/57 cm, buffer: phosphate 50 mmol L\1; pH"2.5; E"526 V cm\1, 214 nm; injection 0.5 psi, 5 s.
increased in this way and lower charge-to-mass ratio or diamino compounds like diamino-pentane, butane
results in decreased mobility. or morpholine can diminish the peptide}wall interac-
Ion-pairing reagents like short chain alkylsulfonic tion. Competing equilibria in the electrostatic attrac-
acids are particularly applied to adjust selectivity for tion between analyte-silanol and amine-silanol
hydrophobic peptides. Concentrations below the groups suppress the adsorption of the peptide. An-
critical micellar concentration are used. The mechan- other approach to reduce adsorption is derivatization
ism is based on the interaction between the hydro- of the silanol groups with an uncharged polymer
phobic surface of the peptides and the hydrophobic (coated capillaries).
alkyl chain. Depending on the hydrophobicity of
Detection Techniques
a peptide, different amounts of alkylsulfonic acid are
attracted. Charge-to-mass ratios of the individual The detection of peptides suffers from the same difR-
peptides are inSuenced to a different extent leading to culties as described for amino acids. Additionally
the separation of the species. only a few amino acids (phe, try, tyr and to a lesser
A second approach to impart selectivity to large extent his, arg, gln, asn) provide residues with strong
peptides with identical mobilities but different hydro- chromophores.
phobicities is the use of ion-pairing reagents above Measuring UV absorbance at low wavelengths
their critical micellar concentration (CMC). This ((220 nm) is the commonest mode of detection
technique may also be used for peptides differing in to give limits of detection of about 1 g mL\1
neutral amino acids such as ala, val, leu or ile. MEKC (&10\5}10\6 mol L\1) which are sufRcient for most
takes advantage of the partitioning of the peptides applications. Spectra obtained by a photodiode array
between the electrolyte and the pseudo-stationary detection support identiRcation of impurities in pept-
phase of the micelles. Hydrophilic moieties of ide synthesis due mainly to the absence of the charac-
the peptide interact with the outer polar sections teristic absorbance of aromatic residues at 220 nm
of the micelle whereas hydrophobic parts are situated (Figure 7).
in the inner hydrophobic sm phere. These peptide} Indirect techniques can be applied as for amino
micelle aggregates possess a different mobility com- acids.
pared to the electrophoretic mobility of the peptide in Detection of trace amounts of peptides requires
free solution. more sensitive methods and sensitivity can be im-
Types of surfactants employed are divided into proved by Suorescence methods.
anionic, cationic and nonionic micelle-forming re- This approach faces the same difRculties as UV
agents. Because of the different charges, different absorbance detection in that only try and, to a lesser
migration directions are obtained. Negatively extent, tyr and phe exhibit native Suorescence when
charged SDS, one of the most frequently used addi- excited at 280 nm (Xe-lamp). However, this ‘natural
tives, migrates counter to the EOF and is used in speciRcity’ facilitates selective identiRcation of try-
concentrations up to approximately 150 mmol L\1. containing peptides. In addition, indirect Suorescence
Common positively charged reagents are cetyl, detection using salicylic acid for anionic charge pep-
dodecyl and hexadecyltrimethylammonium salts. tides (basic buffers) or quinine for the positive mode
These reagents invert the EOF at concentrations be- (acidic buffer) have been applied.
low the CMC so that as a consequence the polarity of To accomplish lower detection limits for a
the applied electrical Reld has to be reversed. broader range of species derivatization techniques
The addition of organic solvents such as methanol, have to be applied and all the agents described for
ethanol, acetonitrile or tetrahydrofuran can provide amino acids can be used for the derivatization of
selectivity for closely migrating peptides. These peptides.
changes can be mainly attributed to solvation of side Increased interest is being paid to mass spectromet-
chains and variations in dissociation of the functional ric techniques for the characterization of peptides,
groups of the peptide. Additionally the EOF is modi- especially soft ionization techniques like electron
Red due to altering the - potential and the increase in spray ionization (ESI). A promising approach to-
buffer viscosity which generates a lower EOF and wards nonfragmented peptides is the matrix-assisted
lower currents. In this way separations have been laser desorption ionization with time-of-Sight mass
established for peptides differing in only a single spectrometers (MALDI-TOF).
neutral amino acid.
Peptides, especially large peptides with protein-like
characteristics, sometimes tend to adsorb at the capil-
Concluding Remarks
lary wall. Beside the possibilities for avoiding disper- CE has proved to be a versatile method for the high
sive effects mentioned above, the addition of amino- efRcient separation of complex mixtures of amino
III / ANAESTHETIC MIXTURES: GAS CHROMATOGRAPHY 2047
Figure 7 CZE separation of a peptide mixture. Capillary: ethylene/vinyl acetate dynamically coated with polyvinyl alcohol 75 m i.d.,
25/45 cm, buffer: phosphate 50 mmol L\1; pH"2.5; E"155 V cm\1, 200 nm; injection 50 mbar, 5 s. 1, Bradykinin; 2, angiotensin II;
3, -MSH; 4, TRH; 5, LH-RH; 6, leucin enkephalin; 7, bombesin; 8, methionin; 9, oxytocin.
acids and peptides due to the manifold separation Further Reading

modes that can be applied. Short analysis times, easy
Bardelmeijer HA, Waterval JCM, Lingeman H et al. (1997)
manipulation of separation conditions and small in-
Pre-, on- and post-column derivatisation in capillary
jection volumes (nanolitres) are further advantages. electrophoresis (review). Electrophoresis 18: 2214.
The Reld of biomedical and clinical amino acid and Blau K and Halket JM (eds) (1993) Handbook of Deriva-
peptide analysis is still under investigation, especially tives for Chromatography, 2nd edn. Chichester: John
as the transfer and adaptation of the separation Wiley.
modes to a broader range of real samples has to be Camilleri P (ed.) (1993) Capillary Electrophoresis } Theory
established. Thus monitoring of in vivo processes, e.g. and Practice. Boca Raton: CRC Press.
analysis of neurotransmitters in cerebrospinal Suid Cifuentes A and Poppe H (1997) Behavior of peptide in
after online microdialysis, could be realized. capillary electrophoresis (review). Electrophoresis 18:
This is directly related to further improvements in 2362.
reproducibility and detection strategies. Landers JP (ed.) (1994) Handbook of Capillary Elec-
trophoresis. Boca Raton: CRC Press.
The most promising techniques that will fulRl the
Novotny MV, Cobb KA and Liu J (1990) Recent advances
demands of trace analysis in biological Suids are in capillary electrophoresis of proteins, peptides and
CE-LIF and CE-MS. amino acids (review). Electrophoresis 11: 732.
Future trends are micro-fabricated CE devices im- Smith JT (1997) Developments in amino acid analysis using
plementing CE technology on a microchip and mul- capillary electrophoresis (review). Electrophoresis 18:
tiple capillary arrays allowing simultaneous analysis 2377.
of up to 96 samples. Thus, a down-scaling of the SzoK koK E (1997) Protein and peptide analysis by capillary
analytical process and the performance of high zone electrophoresis and micellar electrokinetic
throughput analysis could be achieved. chromatography (review). Electrophoresis 18: 74.
ANAESTHETIC MIXTURES:
GAS CHROMATOGRAPHY
A. Uyanek, Ondokuz MayUs University, Introduction
Kampus-Samsun, Turkey
Today, anaesthetists normally use mixtures of ni-
trous oxide and oxygen as a background anaes-
Copyright ^ 2000 Academic Press thetic and carrier to introduce a potent volatile liquid
2048 III / ANAESTHETIC MIXTURES: GAS CHROMATOGRAPHY
Sample Handling and Injection
Sample handling and injection techniques are greatly

inSuenced by the source of the analysed samples such
as liquid samples containing anaesthetics (e.g. blood,
Figure 1 Molecular structures for the volatile liquid anaes- urine, sperm, tissue), low concentration gas samples
thetics (A) halothane, (B) enflurane and (C) isoflurane. (e.g. anaesthetics in pollution studies) and high con-
centration gas samples (e.g. inhaled and exhaled gas
anaesthetic such as halothane (2-bromo-2-chloro-1,1, mixtures).
1-triSuoroethane), isoSurane (1-chloro-2,2,2triSuoro- Direct injection of a liquid sample to the chromato-
ethyl diSuoromethyl ether) or enSurane (2-chloro- graphic column is very simple, but it is a rather crude
1,1,2-triSuoroethyl diSuoromethyl ether: Figure 1) to approach and has serious disadvantages such as con-
produce a state of anaesthesia and analgesia and to tamination of the sample port, column and detector,
sedate a patient. Monitoring the patient’s inhaled and alterations in the baseline characteristics and interfer-
exhaled breath during surgery is very important as ence by water vapour. The problems associated with
a measure of the anaesthetic uptake and the depth of the presence of the liquid in the chromatographic
the anaesthesia. In operating theatres, therefore, system are avoided by the technique of headspace
physical methods of analysis (e.g. dedicated infrared analysis, whereby the vapour above the sample is
analysers) are employed on account of their speed of injected under controlled conditions. Headspace
response and continuous display facilities, though sampling is rapid and is suitable for direct determina-
most can reliably handle only one component at tion of the partial pressure of anaesthetics in blood.
a time. However, there is still the need to analyse such Low concentration samples of liquid anaesthetics
mixtures for all the major components, either in the trapped in an adsorbent-Rlled cartridge (integrated
course of research programmes involving different sampling or passive dosimeter) in pollution studies
agents and different combinations such as inhaled or are introduced into a gas chromatographic system via
exhaled mixture analysis, blood and body Suid analy- a gas sampling valve. Trapped anaesthetics are desor-
sis, anaesthetic pollution studies, thermal decomposi- bed from the adsorption cartridge and transferred by
tion studies or as a back-up to conRrm the perfor- the carrier gas to the main chromatographic column
mance of the dedicated analysers. The major gases by heating the adsorption cartridge rapidly.
present in such mixtures, in addition to air, are car- Low concentration (spot sampling) and high con-
bon dioxide, nitrous oxide and halothane, isoSurane centration samples in the gaseous state may be intro-
or enSurane (or cyclopropane, which is still in use in duced to a gas chromatographic system by a gas-tight
some places). If all the components (gases and va- syringe (0.1}5.0 mL) with the usual septum-type inlet
pours) need to be detected, gas chromatography is system. However, this is not a reproducible sample
extremely powerful in the separation and quantiRca- introduction method and creates problems of reliabil-
tion of the components, in comparison with the other ity where quantiRcation of the components is needed.
techniques available. In addition to this, polymeric material such as rubber
(e.g. on the barrel of a disposable syringe), plastics,
Instrument Requirements and and even glass itself adsorb liquid anaesthetics
(&1}3%) on the contact surface. Adsorption on glass
Procedures surfaces becomes more important when dealing with
There is no rigid boundary separating the basic instru- mixtures at lower concentrations (Figure 2). There-
mental requirements for conventional gas analysis and fore, syringe injection should be avoided in quantita-
anaesthetic mixture analysis by gas chromatography. tive studies.
All the theoretical and practical knowledge and basic If gas samples are to be taken repeatedly to produce
equipment of conventional gas analysis applies to reproducible quantitative data, a gas sampling valve
anaesthetic mixtures and this simpliRes the practice of Rtted with the desired size of sampling loop
the technique in this specialized Reld. A dual-column (0.25}10 mL) should be used at a constant temper-
gas chromatograph Rtted with a gas sampling valve ature and Rlling pressure (usually ambient). It should
(operated at room temperature), and equipped with be noted that, when using a concentration-sensitive
a thermal conductivity detector (TCD) or preferably detector such as TCD, the sample size and column
both TCD and Same ionization detector (FID) is most diameter relationship must be taken into considera-
suitable for all the anaesthetic gas mixture analysis tion to avoid column overloading. Several commer-
encountered. If a septum-type inlet system is also pres- cial gas sampling valves are available in various
ent, it should be placed next to the gas switching valve. conRgurations. Some operate on the slider with the
Synthetic porous polymer beads, which have been

in widespread use as solid adsorbent packing mater-
ial, are available commercially under a variety of
trade names (Chromosorb Century Series, Porapak).
Columns packed with porous polymer beads are
more versatile and capable of separating each of the
individual groups of components such as light gases
and liquid anaesthetics at different temperatures as
well as their complex mixtures with suitable temper-
ature and column arrangements. No special treat-
ment is required to obtain symmetrical peaks as they
are chemically inert to the anaesthetic substances
under the chromatographic conditions employed
(usually 20}2203C). The combined effects of
Figure 2 Adsorption of halothane on glass surface at lower increasing viscosity of the carrier gas and expansion
concentrations. Squares, cylinder preparation; circles, syringe of the stationary phase as the temperature rises result
dilution. in a very marked decrease in the carrier Sow (Figure 3),
e.g. a temperature rise from ambient to 2003C decreases
O-ring principle, while others operate by rotation of the Sow of the carrier from around 50 mL min\1 to
a TeSon威 (polytetraSuoroethylene) or polyimide 20 mL min\1 at 40 psig (2.7 bar) He inlet pressure,
rotor in various Sow paths. The analyst should be with a 2 m, 80}100 mesh Chromosorb 101 column.
aware that some polymeric materials (e.g. silicone Nevertheless, the chromatography remains adequate
rubber O-rings) adsorb anaesthetic vapours to some and gives peaks for the liquid anaesthetics which are
extent (halothane'isoSurane'enSurane). Gas easily integrated. The size of the particles, expressed
switching valves made of a stainless-steel body and in mesh size, is very important in the column efRcien-
TeSon威 rotor or O-rings are the most suitable choice cy as the separation is provided by the surface and
for anaesthetic purposes. It is important to note that structure characteristics of the packing material.
gas sampling valves must not be used with Sow con- When the size of the particles is reduced, the column
trol of the carrier gas, as this restricts the Rlling rate efRciency is increased and so is the inlet pressure
and hence the rate of Sushing of the loop, resulting in because of the high pressure resistance of the column.
tailing peaks, Pressure control is used instead. At the present time, 80/100 mesh is the most widely
used fraction; however, in instances where higher
Choice of Column
efRciency is needed, 100/120 mesh is frequently used.
The column has an essential role in the separation
Column Tubing Materials
process. Optimization of the separation process by
suitable choice of chromatographic column, there- Since most anaesthetic mixtures contain at least one
fore, is the main starting point of any gas chromato- volatile liquid component other than the permanent
graphic analysis. Selection of a column is often made
on the basis of the nature of the samples and the
number of components to be analysed.
Capillary columns have been little used, and mainly
for liquid anaesthetic analysis without gas compo-
nents. The reason for this is the unfavourable reten-
tion factors of low boiling compounds on capillary
columns operated at room temperature.
Packed columns may be subdivided as liquid parti-
tion and solid adsorbent columns. Almost all the
anaesthetic gas analysis reported so far has been per-
formed on packed columns of various lengths, either
single or combined, commonly with 1/8 in and 1/4 in
o.d. Liquid partition columns are generally employed
to separate the high boiling or heavier components
such as liquid anaesthetics, while solid absorbent Figure 3 Relationship of temperature to flow rate of a porous
columns are used for the permanent gases (CO2, polymer packing (80/100 mesh). Flow rate"3.86;10\4 T 2
O2 and N2). !0.283 T#59.6.
used tubing materials for permanent gas analysis at

lower temperatures may not safely be used in anaes-
thetic gas analysis. For example, anaesthetic vapours
(particularly halothane) tend to decompose in contact
with metals (or metal/metal oxide) such as aluminium
(&2003C) and copper ('2503C) at elevated temper-
atures, producing a number of halogenated products.
Relatively inert materials such as glass and stainless
steel may safely be used as column tubing materials
for anaesthetic separation purposes at high operating
temperatures. Since mixtures contain large amounts
of oxygen, heated septum-type injection ports should
have a glass liner to prevent metal}liquid anaesthetic
contact at higher temperature settings.
Figure 4 Variation of the hot-wire TCD responses with detector Choice of Detectors
filament temperature. Circles, halothane; squares, nitrous oxide;
triangles, carbon dioxide; diamonds, air. The most commonly used detectors in anaesthetic gas
analysis are TCD, FID and electron capture detector
(ECD).
gases, operating temperatures with solid adsorbent TCD is concentration-sensitive and has been the
columns are considerably higher (e.g. 150}2203C) most widely used in chromatographic analysis for the
than those required for the separation of the perma- determination of gases, and for any applications in
nent gas alone. Therefore, many of the commonly explosion hazard areas. If inorganic gases, besides
Figure 5 (A) Gas chromatograms for the single-column separation of anaesthetics by temperature programming (linear or non-
linear). A, Air; B, carbon dioxide; C, nitrous oxide; D, halothane; E, isoflurane; F, enflurane. (B) Simple set-up of a temperature-
programmed (linear or nonlinear) dual-column chromatograph.
Figure 6 Chromatograms for dual detector chromatography (A) halothane (left) and isoflurane (right) in atmospheric air. (B) Simple
set-up of a dual detector chromatograph.
liquid anaesthetics, need to be analysed, TCD is the ously removed from the carrier and make-up gases.
detector of choice due to its universal response to Contamination also causes serious interference. The
almost all substances and its very large linear dy- detector must be held at an elevated temperature,
namic range. Because of its relatively poor sensitivity, always with a steady Sow of carrier gas, and must be
it is unsuitable for the determination of low concen- regularly baked out to ensure cleanliness. All these
trations ((40 p.p.m.) without employing extreme factors make ECD a difRcult detector in anaesthetic
detector conditions and large sample volumes. The gas analysis.
nondestructive character of the TCD enables it to be The very widely used FID is a mass-sensitive de-
used in dual-column chromatography by utilizing tector, with the disadvantage compared to the TCD
two channels simultaneously or in series with another that it is destructive. It responds to virtually all or-
detector such as the FID. Sensitivity of the hot-wire ganic components but does not respond to the perma-
TCD depends on the temperature difference between nent gases. In the great majority of studies where only
Rlament and cell wall temperature (Figure 4), and the determination of the volatile liquid anaesthetics is
higher chromatographic responses are obtained at needed (e.g. blood and body Suid analysis), FID is
higher Rlament temperatures. used. If the analysis includes nitrous oxide in addition
The ECD is very sensitive to electrophilic species to liquid anaesthetics, the ECD alone may be chosen.
such as polyhalogenated anaesthetics and also to ni- For low concentration analysis, TCD and FID may be
trous oxide, but its linear dynamic range is limited to connected in series to determine the permanent gases
a range of about 104 and it can easily be saturated. For and the liquid anaesthetics.
this reason, it is generally employed for the low con-
Choice of Carrier Gas
centration determination of liquid anaesthetics and
nitrous oxide. Since oxygen and water inSuence the Choice of the carrier gas depends on the detector
detector sensitivity, these compounds must be rigor- employed. For FID and ECD, carrier gas is not critical
Figure 7 (A) Gas chromatograms for the dual-column separation of A, combined peak; B, air; C, halothane; D, carbon dioxide; E,
nitrous oxide; F, isoflurane; G, enflurane; *, converted peaks. (B) Simple set-up of a temperature-programmed (linear or nonlinear)
single-column chromatography.
and nitrogen may be used for most chromatographic Using temperature as a variable is not, however,
purposes in anaesthetic analysis. For the operation of the only approach. Improved separations can be
the TCD, hydrogen and helium give the highest sensi- achieved by employing mixed column packing mater-
tivities, but helium is preferred on safety grounds. ials in various proportions and column lengths (e.g.
porous polymers and molecular sieves) and multi-
Tactics for the Anaesthetic column (parallel or serially) arrangements operating
in tandem or at different temperatures with single or
Gas Analysis multidetector systems. Utilizing these approaches in
It is usually required to measure a number of the various multicolumn and detector combinations
components in an anaesthetic mixture (e.g. vapours allows the analyst to separate most mixtures of anaes-
and permanent gases), and a single column in a single thetics and permanent gases. Figures 5+7 show the
isothermal run rarely meets this need. Although iso- various arrangements with examples of the chromato-
thermal operation is preferred whenever possible, grams obtained.
temperature programming may be used to improve
the separation process. The magnitude of the temper-
ature range depends on the sample components and
Quantitative Analysis
the nature of the column packing materials. The dis- To be able to carry out quantitative work, the
advantage of temperature programming is that time is gas chromatograph must be calibrated with accu-
required at the end of an analysis to return the initial rately prepared mixtures of known composition.
column temperature. Dynamic methods for calibration such as gas stream
Figure 8 Mixing time for halothane prepared in helium. Squares, 1.1% halothane at 8.5 bar; diamonds, 1.2% halothane at 5.0 bar;
triangles, 1.4% halothane at 3.1 bar.
mixing, permeation, diffusion and evaporation Conclusions

generate continuous Sows of mixtures of known
composition and are generally employed in studies It may be concluded that there is no lack of know-
where large volumes of standards at low concen- ledge, equipment and method to perform gas
trations are needed. Static methods for produc- chromatographic separation and quantitative evalu-
ing standard gas mixtures are appropriate when ation of all types of anaesthetic mixtures from one to
relatively small volumes of mixtures are required multicomponent mixtures (including light gases and
at moderately high concentration levels and have gaseous anaesthetics) in this extensively described
been widely used in calibrating gas chromatographic well-established Reld. Nevertheless, the time required
instruments. The preparation of calibration mixtures for analysis means that gas chromatography is mainly
in gas cylinders involves either volumetric or used for anaesthetic research purposes. Separations
gravimetric mixing. Gravimetric methods in which taking 5}10 min are not acceptable to medical per-
the the concentrations are determined from the sonnel who would require a time scale an order of
mass of each component present in the cylinders magnitude less for analysis of patient’s breath in an
irrespective of the temperature and pressure of the operating theatre. However, there is room for future
mixture represent the nearest approach to an abso- improvements to simplify the column systems, devel-
lute method, provided the mixture is homogenous. oping fast and continuous methods with automated
The mixing rate is inversely proportional to the total samplers to be able to monitor anaesthetic concentra-
pressure and is rapid if thermal or mechanical tions during surgery.
agitation of some kind is introduced to cause tur- See also: II/Chromatography: Gas: Gas-solid Gas
bulence in the gas (usually the cylinder is rolled Chromatography; Headspace Gas Chromatography; De-
in a horizontal position). Without mechanical tectors: General (Flame Ionization Detectors and Thermal
mixing, equilibration is likely to take several days Conductivity Detectors); Detectors: Mass Spectrometry;
(Figure 8). Syringe dilution methods (even with Detectors: Selective; Sampling Systems. III/Gas Analy-
all-glass syringes) are not suitable for calibration sis: Gas Chromatography.
purposes, particularly at lower concentrations, due
to the adsorption of the liquid anaesthetics (see Further Reading
Figure 2). Cowper CJ (1995) The analysis of hydrocarbon gases. In:
Quantitative evaluation may be performed either Adlard ER (ed.) Chromatography in the Petroleum In-
by peak height or by peak area. The most commonly dustry. Amsterdam: Elsevier Science.
used method is based on direct calibration with Cowper CJ and DeRose AJ (1983) The Analysis of Gases by
standard samples which bracket the anticipated Gas Chromatography. Oxford, UK: Pergamon Press.
values in the unknown sample. The correlation Grant WJ (1978) Medical Gases, their Properties and Uses.
Buckinghamshire, England: HM#M.
peak value versus concentration generally exhibits
Hill DW (1980) Physics Applied to Anaesthesia, 4th edn.
a linear plot. The basic condition for successful London: Butterworths.
quantitative analysis is a high degree of constancy of ISO (1981) International Standard 6142. Gas Analysis
operating conditions and the accuracy of the analysis } Preparation of Calibration Gas Mixtures } Weighing
is signiRcantly affected by apparatus parameters, Methods, 1st edn. ref. no: ISO 6142-1981 (E).
characteristics of the detector and the skill of the Stephen CR and Little DM (1961) Halothane. Baltimore,
analyst. MD: Williams & Wilkins.
2054 III / ANALYTICAL APPLICATIONS: DISTILLATION
ANALYTICAL APPLICATIONS:
DISTILLATION
J. D. Green, BP Amoco Chemicals, Hull, UK involatile residue present will remain in the distilla-
tion Sask.
This article is reproduced from Encyclopedia of Analyti-
cal Science, Copyright 1995 Academic Press. Principles
Distillation is a widely used technique in chemical The underlying principles are conveniently illustrated
analysis for characterizing materials by establishing by reference to a vapour}liquid equilibrium diagram
an index of purity and for separating selected compo- (Figure 3). The diagram relates to a binary mixture
nents from a complete matrix. The technique is even containing components P and Q. The lower curve
more widely used in preparative chemistry and gives the composition of the liquid boiling at various
throughout manufacturing industry as a means of temperatures whilst the upper curve gives the com-
purifying products and chemical intermediates. Dis- position of the vapour in equilibrium with the boiling
tillation operations differ enormously in size and liquid. Points x and y therefore give the boiling points
complexity from the semi-micro scale to the ‘thou- of the individual components P and Q respectively.
sands of tonnes per annum’ production operations. For example, point A shows that at X degrees the
For analytical purposes the scale employed is usually vapour has a composition of approximately 90% P,
bench-level. whilst point B shows that the boiling liquid with
Numerous quoted standard speciRcations refer to which it is in equilibrium, has a composition of ap-
distillation ranges as criteria of purity or suitability proximately 80% P. In a continuous distillation pro-
for use, or as indicators of performance. Published cess, such as occurs in a distillation column, liquid of
standards for analytical reagents in the AnalaR range composition C (90% Q, 10% P) vaporizes to vapour
and similar documentation by the American Chem- of composition D which condenses to liquid of com-
ical Society refer to distillation ranges as criteria of position E. Subsequently liquid E becomes vapour
purity for appropriate materials. F and liquid G (composition: 50% Q, 50% P). This
Distillation is the process that occurs when a liquid continuous process of vaporization and condensation
sample is volatilized to produce a vapour that is occurs in the distillation column until a volatile frac-
subsequently condensed to a liquid richer in the more tion leaves the top of the column and is removed from
volatile components of the original sample. The vola- the process by being collected in the collection Sask.
tilization process usually involves heating the liquid At the same time the liquid in the distillation Sask
but it may also be achieved by reducing the pressure becomes progressively more concentrated in the in-
or by a combination of both. This can be demon- volatile component.
strated in a simple laboratory distillation apparatus Distillation techniques may be classiRed into sev-
comprising a Sask, distillation head, condenser and eral different types including:
sample collector (Figure 1). A thermometer is in-
E Distillation at atmospheric pressure
cluded in the apparatus as shown to monitor the
E Distillation under reduced pressure
progress of the operation. In its simplest form this
E Steam distillation
procedure results in a separation into a volatile frac-
E Molecular distillation (short-path distillation)
tion collected in the receiver Sask and a nonvolatile
E Azeotropic distillation
residue in the distillation Sask. When a distillation
E Isopiestic distillation
column is incorporated in the equipment (Figure 2),
the evaporation and condensation processes occur Distillation at atmospheric or reduced pressure
continuously. This results in a progressive fractiona- produces a separation according to the general prin-
tion of the volatiles as they pass up the column. The ciples discussed in the introduction.
most volatile components emerge from the top of the Steam distillation is a means of distilling that part
column initially and the less volatile components of a sample that is volatile in steam at a lower temper-
emerge later. By changing the receivers throughout ature than would otherwise be the case. This method
the course of the distillation a separation or fractiona- is typically used for removing phenols from an aque-
tion is effected. Eventually, all the volatiles will ous sample. A means of introducing steam into the
have passed over into the sample collectors and any distillation Sask must be provided.
III / ANALYTICAL APPLICATIONS: DISTILLATION 2055
Figure 1 Simple distillation apparatus comprising distillation

flask (DF), distillation head (DH), thermometer (T), condenser (C)
and receiver(or collection) flask (RF). (Reproduced by permission
of Longman Scientific & Technical from Furniss et al., 1989.)
Figure 3 Vapour}liquid diagram for a binary mixture of compo-

Molecular distillation, sometimes termed short- nents ‘P’ and ‘Q’, illustrating the principles of distillation (see text
path distillation, is used principally for compounds for details).
normally having high boiling points. In such cases,
very low pressures are needed to achieve the desired
low boiling points. The apparatus is constructed such process results in the toluene}water azeotrope distill-
that the condensing surface is located only a short ing. The distillate can then be examined to determine
distance from the distilling liquid and the pressure is the water content of the original sample.
reduced so that the process is governed to a large Isopiestic distillation is a convenient way of produ-
extent by the mean free path of the molecules in- cing metal-free aqueous samples of volatile acids. The
volved. Hence the terms short-path distillation and ‘crude’ acid is placed in an open container, such as
molecular distillation. a beaker, in a desiccator containing also an open
Azeotropic distillation occurs when a mixture of beaker of pure water. The acid vaporizes and sub-
two materials distils at constant composition. This sequent condensation in the pure water produces an
technique is commonly used to remove water from aqueous sample of the volatile acid without any of the
samples. As an example, toluene may be added to involatile contaminants such as metals.
a complex sample containing water, the distillation The alternative terms ‘Sash’ distillation and ‘frac-
tional’ distillation are sometimes used to describe
some of the above procedures carried out in a particu-
lar way. Flash distillation effects a crude separation
into volatiles and residue, whilst fractional distilla-
tion produces a series of ‘cuts’ of different volatility
(or boiling point) ranges.
Additionally, there are other forms of sample puri-
Rcation and separation that are either a type of distil-
lation or are related to a distillation process:
E Simultaneous distillation/extraction (see applica-
tion section)
E Dean and Stark distillation (see application sec-
tion)
E Simulated distillation (gas chromatographic tech-
nique)
Analytically, distillation is used for two principal
purposes, Rrstly as a criterion of purity and secondly
Figure 2 Distillation apparatus including distillation column as a means of preparing a sample for analysis. Many
(DC). (Reproduced by permission of Longman Scientific & Tech- speciRcation tests include reference to a distillation
nical from Furniss et al., 1989.) range within the limits of which a stated percentage
2056 III / ANALYTICAL APPLICATIONS: DISTILLATION
of the material of interest distils. Alternative distilla- Table 1 Types of distillation column
tion may be used to separate volatiles from a sample
prior to a suitable analytical technique being em- Column type Description/comments
ployed on the distillate or on the residue. Standard Dufton An open tube into which a glass spiral fits
tests are documented that involve distillation as closely
a sample pretreatment method prior to titrimetry, Hempel A simple tube normally filled with a suitable
potentiometry and spectrophotometry. packing (rings/helices) and having a
side-arm near the top
It is of course essential, if meaningful comparative
Oldershaw A column with fixed but perforated plates that
results are to be obtained, that the design and use of maintains a fixed amount of liquid on each
the apparatus are standardized for such determina- plate
tions. Podbielniak A simple tube with a wire packing to provide
large contact area between liquid and
vapour to effect high efficiencies
Apparatus Spinning band A tube fitted with a closely fitting spiral of
PTFE or metal gauze that can be rotated
A wide variety of apparatus is available to satisfy the at typical speeds of 600 to 3000 rev min\1
different distillation techniques. The appropriate de- as the vapour}liquid equilibrium is
sign of apparatus depends upon the type of distilla- maintained in the column
Vigreux A tube having pairs of indentations down its
tion to be performed, considering, for example,
length that slope downwards and provide
whether a vacuum is required or steam is needed. a large and designed surface area to
Descriptions of apparatus are to be found in a num- enhance the liquid}vapour equilibrium
ber of different texts (see the Further Reading). Stan-
dards referring to the design and use of distillation
apparatus have been published by the British Stan- atus by adjustment of the heating rate and by
dards Institute and the American Society for Testing maintaining stable thermal conditions throughout
and Materials. Simulated distillation, which is a gas the apparatus.
chromatographic technique, is dealt with in a [recent]
review by Robillard et al. and referred to in several
standards.
Applications
Apparatus may be discussed in terms of the distilla- Documentation of analytical applications of dis-
tion Sask, the distillation column, the condenser and tillation is widely dispersed. However, there are
the collecting Sask(s). By far the most effort has been numerous references to distillation as a means of
expended in the design and operation of the distilla- characterizing materials and as means of sample pre-
tion column, which is at the heart of the separation treatment in the lists of BSI standards, the ASTM
efRciency. The form of the column, its size and the methods documentation, the analytical methods of
packing used are very inSuential upon the results that the Institute of Petroleum and those of other world-
are achievable. A summary of some different types of wide standards organizations. Table 3 gives a selec-
columns is given in Table 1 and of packings in tion of standards involving distillation originating
Table 2. from various standards organizations.
Once apparatus has been chosen carefully to com-
pare with previously used apparatus or to conform to Table 2 Distillation column packings
standards, the operation of the equipment must be
considered. The following factors are among the most Packing Description
important to be controlled:
Balls Mostly made of glass. Columns have a
tendency to flood easily
E The heating of the distillation Sask must be careful- Helices Made from metal or glass, although metal
ly controlled. may be packed mechanically to produce
E The distillation column must be operated so that it a more uniform column
does not become Sooded. Rings Usually made of glass of an appropriate size
for the column but can be made of
E The reSux ratio, that is the ratio of material return- porcelain, stainless steel, aluminium,
ing via reSux to the distillation column or the copper or nickel. Depending upon design
distillation Sask compared to the amount present- they can be termed Raschig, Lessing or
ed to the condenser in unit time must be carefully Dixon rings
controlled. The higher the reSux ratio, the purer Wire packings Produced as ‘Heli-Grid’ and ‘Heli-Pak’
packings especially for use with
the material collected from the distillation. ReSux Podbielniak columns
ratios are controlled in simple distillation appar-
III / ANALYTICAL APPLICATIONS: DISTILLATION 2057
Table 3 Applications of distillation in analysis Distillation is used widely to determine the moist-
ure or water content of a variety of samples from
Application Standardsa
petroleum products to cereal feeds. The technique
Water/moisture determination used is one of azeotropic distillation using a codistil-
Petroleum products AASHTO T55; ASTM D95; BS 4385; late such as toluene. Table 3 includes a selection of
CNS K6339
Crude oil ASTM D4006
the available methods. Dean and Stark provided
Wool ASTM D2462 a particular design of apparatus that can be used for
Wood/wood products TAPPI T208 OM determining water content following azeotropic dis-
Coal/coke BS 1016
Spices BS 4585; ISO 939
tillation with an immiscible organic solvent. As the
Animal feeds/feedstuffs AACCH 44}50 azeotropic distillate condenses, the water separates
Fats/oils AACCH 44}51; BS 684; ISO 934 from the immiscible organic and can be estimated
Paints and pigments CGSB 1-GP-71 Meth 24-1
Fruits/vegetables SASO 436
directly in a specially graduated collection arm.
Soaps/detergents CGSB 2-GP-d11M Meth 13-2; ISO 4318 Some methods for the determination of water qual-
Tobacco CNS N4133; ISO 6488 ity involve distillation, for example the determination
Pulp and paper CNS P3025
Plastic moulding materials DIN 53713
of a ‘phenol index’, nitrate content or ammonium
content.
Water quality assessment
Phenol index BS 6068 Sect. 2.12; ISO 6439
The determination of nitrogen by the Kjeldahl
Ammonium content BS 6068 Sect. 2.7; ISO 5664 method involves a preliminary distillation of the
Hydrocarbons, purity
sample. Thus methods for the determination of ammo-
Road tars ASTM D20; IP27 niacal and total nitrogen in ammonium nitrate, urea,
Petroleum products AASHTO T115; ASTM D86; BS 7392; sulfuric acid and fertilizers for industrial purposes in-
CNS K6109; IP 123
Creosote/creosote oil AASHTO T62; ASTM D246; CNS K6070
volve a preliminary distillation followed by titrimetry.
Bituminous coatings AASHTO T78 & T110; ASTM D255 Methods for the determination of available Suorine
Aromatic hydrocarbons ASTM D580; CNS K6255 involve distillation prior to a potentiometric or spec-
Volatile organic liquids ASTM D1078
trometric method.
Organic liquids, distillation range and characterization The determination of distillation range is a method
Amyl acetate BS 552
Analytical reagents Anala standards for laboratory chemicals
of establishing the purity of materials. SpeciRc stan-
Butyl acetate BS 551 dard methods are available, for example, for meth-
Chloroform BS 4774 anol, ethylene glycol and propylene glycol. Many
Diethyl ether BS 579
Perchlorethylene BS 1593
unpublished in company methods are used for prod-
Isopropyl acetate BS 1834 ucts and intermediates to validate purity standards
4-Methylpentan-2-one BS 1941 and to establish the suitability of materials for sub-
2-Ethoxyethanol BS 2713
Oil of lime CNS K5089
sequent use.
Citronella oil CNS K6063 As trace analysis of residual compounds in con-
Formic acid ISO 731 Part VII sumables has become more important, methods of
Phenols ISO 1897 Parts 12 & 13
Caprolactam ISO 8661
extracting these compounds have been developed.
A method known as simultaneous distillation extrac-
Miscellaneous application of distillation
Ethyl acetate BS 553
tion developed from the original work of Likens and
White spirit IP 123 Nickerson has been particularly popular and effective
N-determination
for extracting the volatiles from foods and plant ma-
Sulfuric acid/oleum ISO 914 terials, and the herbicide and pesticide residues in
Urea ISO 1592 agricultural products. The method involves steam dis-
Ammonium nitrate ISO 3330, 3331
Fertilizers BS 5551; ISO 5314 & 5315
tilling the compound of interest from an aqueous
Available fluorine in: suspension of the crude sample while the condensed
Hexafluorosilicic acid BS 6445; ISO 6677 steam is continuously extracted with an immiscible
Fluorspar ISO 5439
Arsenic in ores CNS M3094
organic solvent reSuxing within the apparatus. The
Volatiles content design of the apparatus allows the volatiles that are
Aerosols CNS Z6052 extracted from the condensed water to be Sushed into
Fire residues ASTM E1385
the Sask containing the organic solvent. After a pre-
a
Sources: AASHTO, American Association of State Highway Transport viously determined time of extraction, the apparatus
Offices; ASTM, American Society for Testing and Materials; BS, British may be disassembled and the organic solvent re-
Standards Institution; CGSB, Canadian General Standard Board; CNS,
Chinese National Standards; DIN, Deutsche Institut fuK r Normung; ISO,
moved by evaporation from the now concentrated
International Organization for Standardization; SASO, Saudi Arabian Stan- extract. Further analytical techniques can be used to
dards Organization; TAPPI, Technical Association of the Pulp and Paper identify and quantify the components of the residue
Industry; IP, Institute of Petroleum.
according to the particular requirements.
2058 III / ANTIBIOTICS / High Speed Countercurrent Chromatography
A common application of distillation in the separ- Annual Book of American Society for Testing and Mater-
ation sciences is the puriRcation and recovery of sol- ials. Philadelphia: ASTM.
vents especially from HPLC and GPC usage. There is B/R Instruments Corporation } www.brinstruments.com
a range of equipment supplied for recycling of sol- BSI Standards Catalogue. London: British Standards Institute.
vents and useful sources of information can be found Distillation Principles } http://lorien.ncl.ac.uk/ming/distil/
distil0.htm
on the internet, for example the web pages for B/R
Furniss BS, Hannaford AJ, Smith PWG and Tatchell AR
Instruments and Recycling Sciences are included in (1989) Vogel’s Textbook of Practical Organic Chem-
the Further Reading. istry, 5th edn. pp. 168}197. Harlow: Longman ScientiRc
The applications of distillation in analysis are wide- & Technical.
spread, with the technique being used to characterize Godefroot M, Sandra P and Verzele MJ (1981) Chromato-
materials and as a means of preparing samples prior graphy 203: 325.
to analysis. Standard apparatus and methods are de- Likens ST and Nickerson GB (1964) American Society of
scribed for many speciRc applications. Reference to Brewing Chemists, Proceedings, 5.
the general texts and the standards detailed in the Methods for Analysis & Testing (1993) IP Standards for
Further Reading will provide a source of information Petroleum & Related Products, 52nd edn. London:
for future applications. Wiley, Institute of Petroleum.
Perrin DD and Armarego WL (eds.) (1988) PuriTcation of
Laboratory Chemicals, 3rd edn, pp. 5}12. Oxford: Per-
See also: II/Distillation: Energy Management; Historical gamon Press.
Development; Laboratory Scale Distillation; Multicompo- Reagent Chemicals, 8th edn. (1993) American Chemical
nent Distillation; Vapour-Liquid Equilibrium: Theory. Society.
Recycling Sciences Inc. } www.rescience.com
Further Reading Robillard MV, Spock PS and Whitford JH (1991) An
Overview of Methodology and Column Technology for
AnalaR Standards for Laboratory Chemicals. AnalaR Stan- Simulated Distillation Analysis. Bellefonte, PA: Supelco.
dards (1984) (AnalaR is a registered trademark of Stichlmair J and Fair J (1998) Distillation } Principles and
Merck Ltd.). Practice. New York: John Wiley.
ANION EXCHANGERS FOR WATER TREATMENT:

ION EXCHANGE
See III / WATER TREATMENT / Anion Exchangers: Ion Exchange
ANTIBIOTICS
desired compound from a complex matrix such as

High-Speed Countercurrent fermentation broth and crude extract. The puriRca-
Chromatography tion of antibiotics by liquid}liquid partition dates
back to the 1950s when the countercurrent distribu-
tion method (CCD) was used for separation of vari-
H. Oka, Aichi Prefectural Institute ous natural products such as peptide antibiotics,
of Public Health, Nagoya, Japan, aminoglycoside antibiotics and penicillin. However,
Y. Ito, National Institutes of
CCD had serious drawbacks such as bulky fragile
Health, Bethesda, MD, USA
apparatus, long separation times and excessive dilu-
Copyright ^ 2000 Academic Press tion of samples. In the early 1970s an efRcient con-
tinuous countercurrent separation method called
countercurrent chromatography was introduced fol-
Introduction lowed by the advent of high speed countercurrent
Development of antibiotics requires considerable re- chromatography (HSCCC) a decade later. Because of
search effort in isolation and puriRcation of the its high partition efRciency and speedy separation,
III / ANTIBIOTICS / High Speed Countercurrent Chromatography 2059
HSCCC has been widely used for separation and related components that tend to exhibit similar parti-
puriRcation of natural products including a number tion behaviour in a given solvent system. Conse-
of antibiotics as listed in Table 1. Being support-free quently, successful separation necessitates a pains-
chromatographic systems, HSCCC and CCD share taking search for a suitable solvent system, which
important advantages over other chromatographic often requires days, weeks and even months of hard
systems by eliminating complications arising from trial. Once a suitable solvent system is found,
a solid support such as sample loss and decomposition. however, the separation is usually completed within
several hours.
Selection of Two-Phase Solvent HSCCC utilizes two immiscible solvent phases, one
as a stationary phase and the other as a mobile phase.
System Solutes are subjected to a continuous partition pro-
Among the puriRcation of natural products, the cess between these two phases along the column
isolation of antibiotics is one of the most difRcult space free of a solid support, hence the separation is
tasks since the crude sample often contains, in almost entirely governed by the difference between
addition to numerous impurities, a set of closely their partition coefRcients.
Table 1 Separation of antibiotics by HSCCC
Sample Amount Solvent system Mobile phase
Daunorubicin derivatives Chloroform/ethylene chloride/hexane/methanol/ UP

water (1 : 1 : 1 : 3.5 : 1)
Gramicidins A, B, and C 100 mg Benzene/chloroform/methanol/water (15 : 15: 23 : 7)UP
Siderochelin A 400 mg Chloroform/methanol/water (7 : 13 : 8) UP
Efrotomycin 670 mg Carbon tetrachloride/chloroform/methanol/water UP
(5 : 5 : 6 : 4)
Pentalenolactone 50 mg Chloroform/methanol/water (1 : 1 : 1) UP
Bu 2313B 200 mg n-Hexane/dichloromethane/methanol/water LP
(5 : 1 : 1 : 1)
A 201E 350 mg Carbon tetrachloride/chloroform/methanol/water UP
(2 : 5 : 5 : 5)
Tirandamycin A and B 134 mg n-Hexane/ethyl acetate/methanol/water UP
(70 : 30 : 15 : 6)
Actinomycin complex 83 mg Ether/hexane/methanol/water (5 : 1 : 4 : 5) UP
Benzanthrins A and B (quinone antibiotics) 620 mg Carbon tetrachloride/chloroform/methanol/water UP
(4 : 1 : 4 : 1)
Coloradocin 400 mg Chloroform/methanol/water (1 : 1 : 1) UP
Candicidin (polyene macrolide antibiotics) 100 mg Chloroform/methanol/water (4 : 4 : 3) ?
2-Norerythromycins (macrolide antibiotics) 500 mg n-Heptane/benzene/acetone/isopropanol/ UP
0.01 mol L\1 citrate buffer
(pH 6.3) (5 : 10 : 2 : 3 : 5)
Niddamycins (macrolide antibiotics) 200 mg Carbon tetrachloride/methanol/0.01 mol L\1 UP
potassium phosphate buffer (pH 7) (2 : 3 : 2)
Tiacumicins (macrolide antibiotics) 200 mg Carbon tetrachloride/chloroform/methanol/water UP
(7 : 3 : 7 : 3)
Coloradocin (macrolide antibiotics) 400 mg Chloroform/methanol/water (1 : 1 : 1) UP
Sporaviridin complex 100 mg n-Butanol/diethylether/water (10 : 4 : 12) LP
Dunaimycin (macrolide antibiotics) n-Hexane/ethyl acetate/methanol/water UP
(8 : 2 : 10 : 5)/(70: 30 : 15 : 6)
Bacitracin complex 50 mg Chloroform/ethanol/water (5 : 4 : 3) LP
Bacitracin complex 50 mg Chloroform/ethanol/methanol/water (5 : 3 : 3 : 4) LP
Mycinamicins Analytical works n-Hexane/ethyl acetate/methanol/8%aq. LP
ammonia (1 : 1 : 1 : 1)
Colistins Analytical works n-Butanol/0.04 mol L\1 TFA ( 1: 1) containing 1% LP
glycerol
Pristinamycins (macrolide antibiotics) 1 mg Chloroform/ethyl acetate/methanol/water UP
(3 : 1 : 3 : 2)
Pristinamycins (macrolide antibiotics) 1 mg Chloroform/ethyl acetate/methanol/water UP
(2.4 : 1.6 : 3 : 2)
Ivermectin 25 mg n-Hexane/ethyl acetate/methanol/water LP
(19 : 1 : 10 : 10)
Colistin 20 mg n-Butanol/0.04 mol L\1 TFA (1 : 1) LP
Generally speaking, the two-phase solvent system bic groups. Below, we describe the HSCCC separ-
should satisfy the following requirements: ation of selected antibiotics including sporaviridins,
bacitracins, colistins and ivermectins, especially fo-
1. Retention of the stationary phase. Since the cusing on the procedures for optimization of two-
system eliminates the solid support, the retention of phase solvent systems.
the stationary phase in the separation column entirely The apparatus used in the following separations
depends upon the hydrodynamic interaction between was a HSCCC-1A prototype multilayer coil planet
the two solvent phases in the rotating column under centrifuge (Shimadzu Corporation, Kyoto, Japan)
a centrifugal force Reld. While the hydrodynamic with a 10 cm orbital radius which produces a type-J
motion of the two phases is highly complex, the synchronous planetary motion at 800 rpm. The
retention of the stationary phase may be predicted by multilayer coil was prepared by winding about 160 m
the following simple procedure to measure the sett- of PTFE (polytetraSuoroethylene) tubing onto the
ling time of the two phases under gravity: Place 2 mL column holder. Unless otherwise indicated, all separ-
of each phase of the equilibrated two-phase solvent ations were performed under the following condi-
system into a 5 mL capacity graduated cyclinder (al- tions: speed of revolution: 800 rpm; stationary phase:
ternatively, a 13 mm o.d. and 100 mm long glass test organic phase; Sow rate: 3 mL min\1; elution mode:
tube equipped with a plastic cap may also be used) head to tail.
which is then sealed with a stopper. Gently invert the
Sporaviridins
cylinder Rve times to mix the contents and immedi-
ately place it on Sat surface to measure the time Sporaviridins (SVD, Figure 1) are basic water-soluble
required for the mixture to settle into two layers. This antibiotics produced by Kutzneria viridogrisea (for-
settling time should be considerably less than 30 s for merly) Streptosporangium viridogriseum) and they
stable retention of the stationary phase. are active against Gram-positive bacteria, acid-fast
2. Partition coefTcient (K). The partition coefRc- bacteria and trichophyton. As shown in Figure 2,
ient is the key parameter for HSCCC. It is usually they consist of six components each having a 34-
expressed by the analyte concentration in the station- membered lactone ring and seven monosaccharide
ary phase divided by that of the mobile phase. For units, one pentasaccharide (viridopentaose) and two
a successful separation, the K value of an analyte monosaccharides.
should be close to 1. If K;1, the analyte will elute The SDV complex is soluble only in polar solvents
close to the solvent front resulting in loss of peak such as water, methanol and n-butanol, and is extrac-
resolution. On the other hand, if K<1, the analyte ted with n-butanol from the fermentation broth.
will remain in the separation column for a long peri- Therefore, a two-phase solvent system containing n-
od of time, producing an excessively broad peak. In butanol as a major organic solvent was mainly exam-
order to separate two components, the ratio between ined. We found that the SVD sample was entirely
their partition coefRcients, which is called separation partitioned into the upper organic phase in a n-bu-
factor (), should be 1.5 or greater for a standard tanol/water binary two-phase solvent system
semipreparative multilayer coil HSCCC equipment (Table 2). This result indicated that the hydrophobic-
providing a moderate partition efRciency of about ity of the n-butanol phase must be decreased to obtain
800 theoretical plates. a suitable partition coefRcient. A nonpolar solvent
such as n-hexane or diethyl ether was added to the
n-butanol solvent system as a modiRer. Initially, the
HSCCC Separation of Antibiotics volume of n-butanol was Rxed at 10 mL while that of
As mentioned earlier, HSCCC has been successfully the diethyl ether was varied, and a two-phase system
applied to the separation of a variety of antibiotics composed of diethyl ether/n-butanol/water (10 : 4 : 10)
(Table 1). The list includes peptide antibiotics, which was selected. Next, the volumes of n-butanol and
are strongly adsorbed on the silica gel used as the diethyl ether were Rxed while that of water was
stationary phase in column chromatography. Sample varied from 11 to 14. At a solvent ratio of 10 : 4 : 12,
loading capacity of HSCCC widely varies from 1 mg almost evenly dispersed partition coefRcients among
to 10 g, depending on the tube diameter and the the six components were obtained as shown in
length of the multilayer coil used as the separation Table 2. Therefore, this solvent system was selected
column. Two-phase solvent systems may be selected for the separation of the SVD components.
according to the hydrophobicity of the analytes, i.e. The preparative HSCCC separation of six compo-
n-butanol solvent systems for hydrophilic groups, nents from the SVD complex was performed. In this
chloroform systems for moderately hydrophobic experiment the retention of the stationary phase, elu-
groups, and n-hexane systems for the most hydropho- tion time, and elution volume were 75%, 3.5 h and
Figure 1 Structures of sporaviridins. (Reproduced with permission from Oka H et al . (1998).)
500 mL, respectively. The six components were complex. HPLC analyses of the puriRed components
eluted in an increasing order of their partition coefR- are illustrated in Figure 3.
cients yielding high purity of components A1
Bacitracins
(1.4 mg), A2 (0.6 mg), B1 (0.7 mg), B2 (0.5 mg),
C1 (1.1 mg), and C2 (1.4 mg) from 15 mg of the SVD Bacitracins (BCs) are peptide antibiotics produced
by Bacillus subtilis and Bacillus licheniformis. They
exhibit an inhibitory activity against Gram-positive
bacteria and are most commonly used as animal feed
Table 2 Partition coefficients of SVD components with

n-butanol systems
n-Butanol/ Partition coefficients (U L\1)

diethyl ether/water
C2 B2 A2 C1 B1 A1
10 : 0 : 10 2.96 6.41 6.65 4.87 8.81 9.09

10 : 3 : 10 0.96 2.17 2.78 1.84 3.34 4.19
10 : 4 : 10 0.50 1.12 1.59 1.04 1.85 2.91
10 : 5 : 10 0.38 0.78 1.11 0.74 1.25 2.12
10 : 6 : 10 0.39 0.90 1.19 0.81 1.39 1.70
10 : 7 : 10 0.24 0.63 1.10 0.59 1.08 1.82
10 : 4 : 11 0.31 0.89 1.24 0.70 1.50 2.00
10 : 4 : 12 0.38 1.09 1.41 0.80 1.85 2.32
10 : 4 : 13 0.37 1.05 1.51 0.73 1.58 2.10
Figure 2 HPLC separation of sporaviridins. Column, Cosmosil 10 : 4 : 14 0.29 1.09 1.17 0.57 1.22 1.74
5C18 (5 m, 4.6;150 mm); mobile phase, methanol/1mol L\1
ammonium chloride (74 : 26); flow rate, 1 mL min\1; detection, Reproduced with permission from Harada K-I et al . (1990) and
232 nm. (Reproduced with permission from Oka H et al. (1998).) Oka H et al . (1998).
Figure 3 HPLC separation of sporaviridin components. For experimental conditions, see legend to Figure 2. (Reproduced with
permission from Oka H et al . (1998) and Harada K-I et al. (1990).)
additives. Over 20 components are present in the peaks 13}18 whereas those for peaks 20}22 are too
bacitracin complex (Figure 4) among which BC-A large. The ethyl acetate system represented by ethyl
and BC-B are the major antimicrobial components. acetate/ethanol/water showed a long settling time,
BC-F is a degradation product and has nephrotoxic- suggesting poor retention of the stationary phase in
ity. Only the structures of BC-A and -F have been the column. The most promising results were ob-
determined (Figure 5). tained from the chloroform, ethanol and/or meth-
We examined three groups of two-phase solvent anol, water systems as summarized in Table 3.
systems containing n-butanol, ethyl acetate or chloro- Among all combinations for the solvent volume ratio,
form as a major organic solvent, and ethanol and/or chloroform/ethanol/ methanol/water (5 : 3 : 3 : 4) and
methanol as a modiRer against water in each group. chloroform/ethanol/water (5 : 4 : 3) gave the most
The n-butanol system produced suitable K values for desirable K values.
Figure 4 HPLC separation of bacitracins. Column Capcel Pak C18 (5 m, 4.6;150 mm); mobile phase, methanol/0.04 mol L\1
sodium dihydrogen phosphate (6 : 4); flow rate, 1.3 mL min\1; detection, 234 nm. (Reproduced with permission from Oka H et al . (1998)
and Harada K-I et al . (1991).)
Figure 5 Structures of bacitracins A and F. (Reproduced with permission from Harada K-I et al. (1991) and Oka H et al . (1998).)
Figure 6 shows the countercurrent chromatogram We selected a two-phase solvent system composed
of bacitracin components using the chloroform/ of n-hexane, ethyl acetate, methanol and water. This
ethanol/methanol/water (5 : 3 : 3 : 4) system. A 50 mg solvent system is conveniently used for the separation
amount of the bacitracin complex was loaded into the of components with a broad range of hydrophobicity
HSCCC column. The retention of the stationary by modifying the volume ratio between the four sol-
phase was 72.7% and the elution time was about 3 h. vents. In the n-hexane/ethyl acetate/methanol/water
All components were eluted in an increasing order of (8 : 2 : 5 : 5) system Rrst examined, the K values of
their partition coefRcients, yielding 5.5 mg of BC-A the components corresponding to peaks 1}7 were 0,
from peak 18 and 1.5 mg of BC-F from peak 22. 0.46, 0.61, R, 1.86, 3.06, and 4.38, respectively.
Ivermectins
Ivermectins B1 are broad spectrum antiparasitic
agents widely used for food-producing animals such
as cattle and pigs. They are derived from avermectins
B1, the natural fermentation products of Streptomy-
ces avermitilis. Avermectins B1 have double bonds
between carbon atoms at 22 and 23, whereas the
ivermectins B1 have single bonds in these positions
(Figure 7). The ivermectins B1 are a mixture of two
major homologues, ivermectin B1a ('80%) and
ivermectin B1b (;20%), but a crude ivermectin
complex also contains various minor compounds
(Figure 8A).
Table 3 Partition coefficients of the bacitracin components
Chloroform Partition coefficients (U L\1)

ethanol/
methanol/ Peaks 3, 14 17 18 20 21 22
water
5:2:3:4 7.20 2.46 4.17 0.64 0.65 0.48

5:2:1:4 R R 33.27 1.62 1.38 0.75
5:3:3:4 3.35 1.40 2.37 0.57 0.47 0.45
5:3:0:3 11.1 3.20 5.34 0.32 0.35 0.27
5:4:0:2 3.19 1.05 2.00 0.25 0.26 0.21 Figure 6 HSCCC separation of bacitracin components. Appar-
5:4:0:3 5.49 1.46 2.20 0.16 0 0.16 atus, HSCCC-1A; revolution, 800 rpm; solvent system, chloro-
5:4:0:4 6.10 2.04 2.68 0.14 0 0.10 form/ethanol/methanol/water (5 : 3 : 3 : 4); mobile phase, lower or-
ganic phase; flow rate, 3 mL min\1 detection, 254 nm. (Repro-
(Reproduced with permission from Harada K-I et al . (1991) and duced with permission from Harada K-I et al . (1991) and Oka H
Oka H et al . (1998).) et al . (1998).)
Figure 7 Structures of ivermectins and avermectins. (Reproduced with permission from Oka H et al . (1996, 1998).)
This indicates that the component corresponding to A 25 mg quantity of crude ivermectin was separ-
peak 6 (ivermectin B1a) is mostly partitioned in the ated using the above solvent system at a Sow rate of
upper organic phase (Table 4). Although the n- 2 mL min\1. The retention of the stationary phase
hexane/ethyl acetate/methanol/water (9 : 1 : 5 : 5) sys- was 67.6% and the total separation time, 4.0 h. The
tem somewhat improved the K value of peak 6, it was HSCCC elution curve of the ivermectin components
still too large and the value between peaks 6 and 7 is monitored at 245 nm is shown in Figure 9, where all
smaller than 1.5. Finally a slightly less polar solvent components are separated into three peaks, A, B and
mixture at the volume ratio of 19 : 1 : 10 : 10 yielded C. HPLC analysis of each peak fraction and the
the best K value, as indicated in Table 4. The settling column contents revealed that both HPLC and
time of this solvent system was 7 s, promising excellent HSCCC systems elute all components in the same
retention of the stationary phase. In addition, the volume order: HPLC peaks 3, 5, and 6 correspond to HSCCC
ratio between the two phases is nearly 1, indicating that peaks A, B and C, respectively, while HPLC peak
either phase can be used as the mobile phase without 7 was still retained in the HSCCC column. This
wasting the solvents. Therefore, the above solvent was separation yielded 18.7 mg of 99.0% pure ivermectin
selected for separation of ivermectin components. B1a (Figure 8B), 1.0 mg of 96.0% pure ivermectin
Figure 8 HPLC separation of ivermectin components. Column, TSK GEL-80 Ts C18 (5 m, 4.6;150 mm); mobile phase, meth-
anol/water (9 : 1); flow rate, 1 mL min\1; detection, 245 nm. (A) Crude ivermectin; (B) Fraction II (ivermectin B1a); (C) Fraction IV
(ivermectin B1b); (D) Fraction VI (avermectin B1a). (Reproduced with permission from Oka H et al . (1998).)
Table 4 Partition coefficients of the ivermectin components

(K"peak area of upper phase divided by peak area of lower
phase.)
Solvent system Peak no.
1 2 3 4 5 6 7
n-Hexane/ethyl
acetate/methanol/
water (8 : 2 : 5 : 5) 0 0.46 0.61 R 1.86 3.06 4.38
n-Hexane/ethyl
acetate/methanol/
water (9 : 1 : 5 : 5) 0 0.15 0.33 R 1.17 2.31 3.21
n-Hexane/ethyl
acetate/methanol/
water (19 : 1 : 10 : 10)0 0 0.18 0.48 0.79 1.36 2.83
Figure 10 HPLC separation of commercial CL. Column,
(Reproduced with permission from Oka H et al. (1996, 1998).) Chromatorex Ph (5 m, 4.6;250 mm); mobile phase, acetonit-
rile/0.01 mol L\1 TFA aqueous solution (24 : 76); flow rate,
1.0 mL min\1; detection, 210 nm. (Reproduced with permission
B1b (Figure 8C) and 0.3 mg of 98.0% pure avermec-
from Ikai Y et al. (1998) and Oka H et al. (1998).)
tin B1a (precursor of ivermectin) (Figure 8D).
Colistin
tanol and water as a basic solvent system. However,
Colistin (CL) is a peptide antibiotic produced by this combination was not suitable by itself, because
Bacillus polimyxa var. Colistinus that inhibits the the CL components were entirely partitioned into the
growth of Gram-negative organisms. CL is a mixture aqueous phase. In order to partition the CL compo-
of many components (Figure 10) where two main nents partly into the n-butanol phase, various salts
components are colistins A (CL-A) and B (CL-B). As (NaCl and Na2SO4) or acids (HCl, H2SO4 and
shown in Figure 11, CLs-A and -B are linear-ring CF3COOH or TFA) were added as a modiRer. The
peptides that differ only in their N-terminal fatty desired effect was produced from TFA where the
acid. CL is used as a feed additive for domestic ani- partition coefRcients of CL components rose as
mals such as calf and pigs for preventing bacterial the concentration of TFA in the solvent system was
infection and/or improving feed conversion efRcien- increased. This effect may be explained as follows: as
cy. CL is soluble in water, slightly soluble in alcohols, shown in Figure 11, CLs-A and -B have Rve free
but insoluble in nonpolar solvents such as hexane and amino groups in L-diamino-butyric acid (L-Dab), and
chloroform. From this property, we selected n-bu- these amino groups dissociate in the aqueous phase
under neutral to acidic conditions. Since TFA forms
an ion pair with these amino groups, the hydropho-
bicity of the CL components increases with the con-
centration of TFA resulting in their partition toward
the organic phase. In order to determine the optimal
concentration of TFA in the solvent system, K values
of Rve components were measured at various TFA
concentrations. As shown in Figure 12, the K value of
each component increases with the TFA concentra-
tion, and at 40 mmol L\1 TFA concentration,
K values of CL-A and CL-B reached 1.5 and 0.6,
respectively. At this TFA concentration, the values
between the adjacent peaks are all greater than 1.5,
promising a good separation for all components. The
settling time of the solvent system was 28 s, which is
within the acceptable range. Therefore, we selected
Figure 9 HSCCC separation of ivermectin components. Appar- a solvent system of n-butanol/40 mmol L\1 TFA
atus, HSCCC-1A; revolution, 800 rpm; solvent system, n-
aqueous solution (1 : 1) for the HSCCC separation.
hexane/ethyl acetate/methanol/water (19 : 1 : 10 : 10) mobile
phase, lower aqueous phase; flow rate, 2 mL min\1; detection, Using the above solvent system, a 20 mg quantity
245 nm. (Reproduced with permission from Oka H et al . (1996, of commercial CL was separated by HSCCC.
1998).) The retention of the stationary phase was 45%. The
Figure 11 Structures of colistin components. (Reproduced with permission from Ikai Y et al. (1998) and Oka H et al . (1998).)
elution curve monitored at 220 nm is shown in Fig- plications such as adsorptive loss and deactivation of
ure 13. According to the results of HPLC analysis and samples as well as contamination from the solid sup-
the elution curve, all collected fractions were com- port. As shown by our examples, HSCCC can isolate
bined into Rve large fractions as shown in Figure 13. various components from a complex mixture of anti-
The yields of CL-A and CL-B were 9 mg each and biotics by carefully selecting the two-phase solvent
those of other minor components were 0.5}1.0 mg. system to optimize the partition coefRcient (K) of the
HPLC analysis was performed for each fraction; as target component(s). Compared with CCD and other
shown in Figure 14, the fractions of CLs-A and -B countercurrent extraction methods, HSCCC can yield
each produced a peak with a purity of over 90%. higher partition efRciencies in a shorter elution
time. The HSCCC system can also be applied to
microanalytical-scale separations without excessive
Conclusions dilution of samples. We believe that HSCCC is an
Because it is a support-free partition system, HSCCC
has an important advantage over other chromato-
graphic methods in that it eliminates various com-
Figure 13 HSCCC separation of commerical CL. Apparatus,

HSCCC-1A; revolution, 800 rpm; solvent system, n-butanol/
0.04 mol L\1 TFA aqueous solution (1 : 1); mobile phase, lower
Figure 12 Effect of TFA concentration on the partition coeffi- aqueous phase; flow rate, 2.0 mL min\1; detection, 220 nm. (Re-
cients of CL components. (Reproduced with permission from Ikai produced with permission from Ikai Y et al . (1998) and Oka H
Y et al . (1998) and Oka H et al . (1998).) et al . (1998).)
III / ANTIBIOTICS / Liquid Chromatography 2067
Instrumentation. Chromatography: Liquid: Countercur-

rent Liquid Chromatography. III / Antibiotics: Liquid
Chromatography. Supercritical Fluid Chromatography.
Further Reading
Harada K-I, Kimura I, Yoshikawa A et al. (1990) Structural
investigation of the antibiotic Sporaviridin. XV. Prep-
arative-scale preparation of Sporaviridin components by
HSCCC. Journal of Liquid Chromatography 13:
2373}2388.
Harada K-I, Ikai Y, Yamazaki, Y et al. (1991) Isolation of
bacitracins A and F by high-speed counter-current
chromatography. Journal of Chromatography 538:
203}212.
Ikai Y, Oka H, Hayakawa J et al. (1998) Isolation of
colistin A and B using high-speed countercurrent
chromatography. Journal of Liquid Chromatography
21: 143}155.
Ito Y and Conway WD (1996) High-Speed Countercurrent
Chromatography. New York: Wiley.
Oka H, Ikai Y, Kawamura N et al. (1991) Direct interfac-
ing of high speed countercurrent chromatography to frit
electron, chemical ionization, and fast atom bombard-
Figure 14 HPLC analysis of CL components of HSCCC frac- ment mass spectrometry. Analytical Chemistry 63:
tions. (A) CL-A (Fraction 5); (B) CL-B (Fraction 3). (Reproduced 2861}2865.
with permission from Oka H et al . (1998).) Oka H, Ikai Y, Hayakawa J et al. (1996) Separation of
ivermectin components by high-speed counter-current
ideal method for separation and puriRcation of anti- chromatography. Journal of Chromatography A 723:
biotics. 61}68.
Oka H, Harada K-I, Ito Y and Ito Y (1998) Separation of
See also: II/Chromatography: Countercurrent Chromato- antibiotics by countercurrent chromatography. Journal
graphy and High-Speed Countercurrent. Chromatography: of Chromatography A 812: 35}52.
Liquid Chromatography
T. Itoh and H. Yamada, Kitasato University, for the antibiotic of interest, the HPLC conditions
Tokyo, Japan that are able to resolve stereoisomers are described.
Copyright ^ 2000 Academic Press
Aminoglycosides
Aminoglycosides are analysed by reversed-phase
Introduction HPLC. However, derivatization is usually necessary
owing to very poor UV or visible absorption. For
High performance liquid chromatography (HPLC) detection of aminoglycosides without derivatization,
has been widely used for the analysis of antibiotics electrochemical, refractive index or mass spectromet-
because it is superior to conventional microbiological ric detection may be used.
assays in terms of speciRcity, sensitivity and analysis
time. In this article, HPLC conditions for the analysis Amikacin
of a variety of antibiotics are summarized. For analy- For determination of amikacin (Figure 1, structure
sis of biological samples, not only extraction methods 1), the serum sample is loaded onto the silica gel
but also derivatization methods are described, if ne- column, followed by addition of o-phthalaldehyde (a
cessary. Since it is not possible to list HPLC methods derivatizing reagent). The column is eluted with 95%
for all antibiotics in clinical use, only a few have been ethanol (pH 10) and the eluent is heated at 503C.
chosen from each class. Where a stereoisomer exists After cooling, the mixture is injected onto an ODS
2068 III / ANTIBOTICS / Liquid Chromatography
propyl-bonded silica column). Gentamycin is eluted

with a 1 : 1 mixture of acetonitrile}0.2 mol L\1 bor-
ate buffer (pH 10.5) and is derivatized with 9-
Suorenylmethyl chloroformate. Derivatized gen-
tamycin is analysed on an ODS column with a mobile
phase consisting of acetonitrile}water (9 : 1). The de-
rivatives are detected Suorometrically at 260 nm ex-
citation, 315 nm emission, with a detection limit of
less than 50 ng mL\1. Gentamycins C1, C1a and C2
are resolved (Figure 1, structures 2, 3 and 4, re-
spectively).
o-Phthaldialdehyde, dansyl chloride, Suorescamine,
1-Suoro-2,4-dinitrobenzene or 2,4,6-trinitroben-
zenesulfonic acid may also be used as derivatizing
reagents.
Glycopeptides
Various glycopeptide antibiotics are separated with
an ODS column. The mobile phase composition is
either 7}32% acetonitrile (7% for 1 min, then in-
crease to 34% over 13 min) in 0.1 mol L\1 phosphate
buffer (pH 3.2) or 5}35% acetonitrile (5% for 1 min,
then increase to 35% over 13 min) in 0.025 mol L\1
phosphate buffer (pH 6.0). Glycopeptides are detec-
ted at 220 nm.
Vancomycin
Serum is deproteinized with an ice-cold mixture of
Figure 1 Chemical structures of amikacin (1) and gentamycins
10% trichloroacetic acid-acetone (2 : 1) and the
C1 (2), C1a (3) and C2 (4). supernatant is injected into an ODS column. The
mobile phase consists of 50 mmol L\1 sodium dihyd-
rogen phosphate (pH 3.3)}acetonitrile (4 : 1) contain-
column. Mobile phase is methanol}water}aceto- ing 1 mmol L\1 sodium dodecyl sulfate. Vancomycin
nitrile (65 : 30 : 5) containing 0.2% tripotassium is detected at 235 nm with a detection limit of
ethylenediamine tetraacetic acid (EDTA). Amikacin 1 g mL\1.
is detected Suorometrically at 350 nm for excitation
and 420 nm for emission. The detection limit is
1 g mL\1. In order to resolve amikacin from its Macrolides
three isomers, a similar method is used except that the Clarithromycin
mobile phase is methanol}water (7 : 3). Tobramycin
may be analysed with the same method. Clarithromycin (Figure 2, structure 5) and its major
Pre-column derivatization of amikacin is also con- metabolite (14-hydroxyclarithromycin) are analysed
ducted by addition of 1-Suoro-2,4-dinitrobenzene to with a C8 column. The mobile phase consists of
plasma or urine, leading to formation of a stable acetonitrile}methanol}water (39 : 9 : 52) containing
chromophore, which can be detected at 360 nm. An 0.04 mol L\1 sodium dihydrogen phosphate with the
ODS column is used with a mobile phase consisting pH being adjusted to 6.8 using sodium hydroxide.
of acetonitrile}water (68 : 32). The detection limit is The eluent is monitored by electrochemical detection
1 g mL\1. Amikacin may be extracted from serum with a quantiRcation limit of 30 ng mL\1. Plasma
using a cation exchange solid-phase extraction col- and urine are extracted with ethyl acetate}hexane
umn prior to derivatization. (1 : 1).
Gentamycin Erythromycin
Gentamycin in plasma is extracted with a cation Erythromycin A (the major and most active
exchange solid-phase extraction column (carboxy- component, Figure 2, structure 6), erythromycin B,
Penicillins and Cephalosporins

Many penicillins and cephalosporins are chiral, partly
due to the chirality of the side chain. The D-epimers of
ampicillin (see Figure 3, structure 13), cephalexin
(Figure 3, structure 17) and cephaloglycin are more
active than the corresponding L-epimers. Stereo-
isomers also exist for amoxicillin (Figure 3, structure
14), azidocillin, cefamandole, cefsulodin and cef-
tibuten (Figure 3, structure 16). For these -lactams,
commercial preparations contain only a single
isomer.
For some -lactams, commercial preparations con-
tain both epimers. These include carbenicillin (7),
clometocillin, moxalactam (18), phenethicillin (11),
propicillin (12), sulbenicillin (8), temocillin (9) and
ticarcillin (10) (see Figure 3). Epimers of these -
lactams are resolved by reversed-phase HPLC.
Ampicillin
Figure 2 Chemical structures of clarithromycin (5) and eryth-
romycin A (6). In order to separate ampicillin (Figure 3, structure
13) from penicilloic acid, phenylglycine and 6-
aminopenicillanic acid, an ODS column is used with
erythromycin C and related compounds in commercial a mobile phase consisting of 35% acetonitrile in an
preparations are analysed with an ODS column using aqueous solution of 3.5 mmol L\1 sodium dodecyl
a mobile phase consisting of acetonitrile}methanol} sulfate and 0.2 mol L\1 formic acid. Ampicillin and
0.2 mol L\1 ammonium acetate}water (45 : 10 : 10 : other compounds are detected at 254 nm. For separ-
35, pH adjusted to 7.0}7.8). Erythromycins are de- ation of ampicillin from its degradation products, an
tected at 215 nm. ODS column is used with a mobile phase of 22.5%
Erythromycin and its metabolites in biological methanol in an aqueous solution of 5 mmol L\1 tet-
Suids are analysed with an ODS column using a mo- rabutylammonium hydrogen sulfate and 5 mmol L\1
bile phase consisting of acetonitrile}methanol} ammonium sulfate (pH 2.6). Ampicillin and the
0.2 mol L\1 sodium acetate (pH 6.7, 40 : 5 : 55). degradation products are detected at 238 nm.
Erythromycin is detected with a dual-electrode elec- Ampicillin is analysed in biological Suids with an
trochemical detector with a detection limit of ODS column using a mobile phase consisting of
10 ng mL\1 in plasma. Erythromycin is extracted 0.06 mol L\1 phosphate buffer (pH 4.6)}methanol
from plasma with ether, and urine is deproteinized (425 : 75). Ampicillin is detected at 225 nm with
with acetonitrile. Other related erythromycins and a limit of accurate determination of 0.5 g mL\1 in
degradation products are also resolved. urine, plasma or saliva. Samples are deproteinized
with perchloric acid.
Ivermectin
In order to increase sensitivity, ampicillin and its
Ivermectin in tissue is analysed with an ODS column metabolites in urine are subjected to postcolumn al-
using a mobile phase of acetonitrile}water (9 : 1) at kaline degradation following separation with an ODS
653C. Ivermectin is detected Suorometrically at column. Urine is diluted with water and injected
272 nm excitation, 465 nm emission, with a detec- directly. The mobile phase is an aqueous mixture
tion limit of 0.25 ng g\1. Tissue sample is loaded of 5 mmol L\1 sodium heptylsulfonate, 1 mmol L\1
onto a C8 solid-phase extraction column, eluted sodium dihydrogen phosphate and 9 mmol L\1
with acetonitrile, and the eluate dried under a stream phosphoric acid}methanol (1.5 : 1, pH 3.0). Degra-
of N2. The dried residue is dissolved with ethy1 dation of ampicillin and its metabolites is conducted
acetate}hexane (2 : 3), loaded on a silica column, with 0.75 mol L\1 sodium hydroxide, 2 mmol L\1
and eluted with methanol}ethyl acetate (1 : 1). mercuric chloride and 10 mmol L\1 EDTA, and
The eluate is dried under a stream of N2 and the degradation products are detected at 300 nm.
treated with triSuoroacetic anhydride and methyl- The limits of accurate determination are 0.5 g mL\1
imidazole. The analyte thus obtained is injected into for ampicillin and 1}2 g mL\1 for the meta-
an HPLC. bolites.
Figure 3 Chemical structures of carbenicillin (7), sulbenicillin (8), temocillin (9), ticarcillin (10), phenethicillin (11), propicillin (12),
ampicillin (13), amoxicillin (14), cefixime (15), ceftibuten (16), cephalexin (17) and moxalactam (18).
Carbenicillin, Sulbenicillin and Ticarcillin loaded onto an anion exchange solid-phase extrac-
tion column. Carbenicillin epimers are eluted with
These epimeric, di-anionic -lactams are similar in 10% lithium chloride-methanol (3 : 2) and injected
physicochemical properties and are analysed under into an HPLC. Analysis is on an ODS column
very similar conditions. For analysis of carbenicillin with a mobile phase consisting of 0.05 mol L\1
(Figure 3, structure 7), plasma and urine samples are ammonium acetate}methanol (9 : 1). Carbenicillin
Figure 4 Chromatogram of human plasma spiked with carbenicillin. Five hundred microlitres of human plasma was spiked with 20 L
of an aqueous solution of carbenicillin (5.2 mg mL\1). R, R-epimer, S, S-epimer.
epimers are detected at 254 nm. The epimers are HPLC methods for determination of both cis and
resolved to the baseline with the R-epimer being trans isomers have been developed because isomeriz-
eluted prior to the S-epimer (Figure 4). ation is observed in vivo.
Similar methods can be applied for the analysis of Ceftibuten and its trans isomer are separ-
sulbenicillin (Figure 3, structure 8) and ticarcillin ated with an ODS column using a mobile phase
(Figure 3, structure 10). For sulbenicillin, the mobile consisting of 100 mmol L\1 ammonium acetate}
phase consists of 0.05 mol L\1 phosphate buffer (pH methanol (92 : 8). Both isomers are detected at
7.0)}methanol (8 : 1). The epimers are resolved to the 262 nm.
baseline with the S-epimer being eluted faster than the An HPLC method for ceftibuten isomers is also
R-epimer. The detection limit is 0.5 g mL\1 for each developed for plasma and urine samples. Samples
epimer. For ticarcillin, the mobile phase consists of are deproteinized with ethanol and injected into
0.05 mol L\1 phosphate buffer (pH 7.0)}methanol an ODS column with a mobile phase composed of
(12 : 1). The epimers are resolved to the baseline with PIC A (tetrabutylammoniumphosphate)}acetonitrile}
the R-epimer being eluted faster than the S-epimer. methanol (50 : 6 : 3). Both isomers are detected at
256 nm with a detection limit of 1 g mL\1 for each
Ce\xime
isomer.
CeRxime (Figure 3, structure 15) is determined on an
Cephalexin
ODS column with a mobile phase consisting of
acetonitrile}water (2.75 : 7.25) containing 0.01 mol Cephalexin epimers are separated with an ODS col-
L\1 ammonium acetate and 0.01 mol L\1 tetra-N- umn using a mobile phase of 0.1 mol L\1 phosphate
butylammonium bromide. CeRxime is detected at buffer (pH 3.5)}methanol (95 : 5). The epimers are
290 nm. detected at 254 nm. The two epimers are separated to
For analysis of ceRxime in serum, an ODS column the baseline with the L-epimer being eluted prior to
is used with a mobile phase consisting of 0.3% potas- the D-epimer.
sium dihydrogen phosphate}acetonitrile (88.5 : Cephalexin epimers in serum and urine are ana-
11.5). For urine analysis, the mobile phase is a mix- lysed using a TSK-gel ODS-80 TM column after
ture of 0.15% potassium dihydrogen phosphate} deproteinization with methanol. Mobile phase com-
acetonitrile (77 : 23) containing 0.1% phosphoric positions are 10 mmol L\1 ammonium acetate}meth-
acid. CeRxime is detected at 254 nm with a detection anol (4 : 1) for determination of the D-epimer, and
limit of 0.1 g mL\1. 10 mmol L\1 phosphate buffer (pH 3.0)}methanol
(9 : 1) containing 10 mmol L\1 ammonium acetate
Ceftibuten
and 10 mmol L\1 pentanesulfonic acid for deter-
Although commercially available formulations con- mination of the L-epimer. The epimers are detected at
tain only the cis isomer (Figure 3, structure 16), 260 nm.
Other -Lactams Semisynthetic cephalosporins are extracted from

biological Suids and chromatographed with an ODS
Epimers of phenethicillin (PEPC, Figure 3, structure column. Urine samples are merely centrifuged and
11), propicillin (PPPC, Figure 3, structure 12) and diluted with distilled water. Serum samples are mixed
clometocillin are analysed with a Zorbax C8 column. with 0.4 mol L\1 HCl and extracted with CHCl3-n-
The mobile phase is composed of methanol} pentanol (3 : 1). The organic phase is re-extracted
water}5% 0.2 mol L\1 phosphate buffer (pH 7.0) into phosphate buffer (pH 7), which is injected into
and the epimers are detected at 254 nm. Ratios of the HPLC. The mobile phase is 0.01 mol L\1 acetate
methanol in the mobile phase are 37.5, 45 and 50% buffer (pH 4.8)}methanol (15 : 85) with detection
for PEPC, PPPC and clometocillin, respectively. Epi- wavelengths of 254, 245, 234, 275, 270, 240 and
mers are resolved close to the baseline, and the D- 240 nm for cefuroxime, cefoxitin, cefotaxime,
epimers elute faster than the corresponding L-epimers. cefazolin, cefamandole, cephalotin and cefoperazone,
The same HPLC conditions can be used for the respectively.
analysis of ampicillin, amoxicillin and azidocillin, Cephalosporins in serum are also analysed with
except that the methanol content is varied between 10 an octyl column using a mobile phase of meth-
and 40%. The L-epimer elutes faster for ampicillin, anol}12.5 mmol L\1 phosphate buffer (pH 2.6,
whereas the D-epimer elutes faster for amoxicillin and 1 : 4). Cefaclor, cefadroxil, ceRxime, cephalexin and
azidocillin. The less active epimers are not detected in cephradine are simultaneously analysed and detected
the commercial preparations of these penicillins. at 240 nm. The detection limits are 0.1 g mL\1 for
Epimers of PEPC and PPPC are also resolved with ceRxime and 1.0 g mL\1 for other cephalosporins.
an ODS column. The mobile phase consists of Serum is deproteinized with acetonitrile.
100 mmol L\1 ammonium acetate}methanol (62 : 38 Cephalosporins with a tetrazole ring are analysed
and 58 : 42 for PEPC and PPPC, respectively) with from plasma with an ODS column using a mobile
a UV detection at 220 nm. PEPC and PPPC epimers phase consisting of 0.05 mol L\1 phosphate buffer
are baseline separated (Figure 5). (pH 6.6)}methanol with ratios of 3 : 1 and 2 : 1 for
Bacampicillin and cefotiam hexetil are the pro- cefamandole and cefoperazone, respectively. For
drugs of ampicillin and cefotiam, respectively, which cefotiam and cefmetazole, a mixture of phosphate
are commercially available as mixtures of two epi- buffer-tetrahydrofuran (20 : 1) is used as a mobile
mers due to chirality of the prodrug moiety. For phase. Cephalosporins are detected at 254 nm with
separation of bacampicillin isomers, an ODS column a limit of detection of 1 g mL\1 for all cephalos-
is used with a mobile phase consisting of porins.
20 mmol L\1 ammonium acetate}methanol (45 : 55). In order to increase sensitivity, ampicillin,
The isomers are detected at 220 nm. For separation of amoxicillin, cephalexin and cephradine in plasma are
the isomers of cefotiam hexetil, an ODS column is assayed after formation of Suorescent degradation
used with a mobile phase consisting of 50 mmol L\1 products. Plasma is deproteinized with 10% trich-
phosphate buffer (pH 3.0)}acetonitrile (73 : 27). The loroacetic acid and the supernatant is heated under
isomers are detected at 262 nm. Baseline separation various conditions to form degradation products. The
of the isomers of bacampicillin and cefotiam hexetil degradation products are extracted with an organic
are observed. solvent and injected into a Nucleosil C18 column at
Figure 5 Chromatogram of phenethicillin. One hundred microlitres of an aqueous solution of phenethicillin (22 g mL\1) was directly
injected onto HPLC.
Table 1 HPLC conditions for -lactams
-Lactam Stationary phase Mobile phase Detection
Benzylpenicillin ODS Methanol}0.05 mol L\1 ammonium carbonate (1 : 3) UV (254 nm)

Benzylpenicillin ODS Phosphate buffer (pH 6.0)}acetonitrile (4 : 1) UV (225 nm)
Cefsulodin ODS 16.8 mmol L\1 dibasic ammonium phosphate}acetic UV (260 nm)
acid}methanol (100 : 1.68 : 5.98) containing 5 mmol L\1
triethylamine
Cefsulodin ODS Aqueous solution (containing 38.8 mmol L\1 ammonium UV (260 nm)
acetate, 0.292 mmol L\1 dibasic ammonium phosphate
and 9.363 mmol L\1 triethylamine)}acetonitrile}
methanol}dimethylformamide}acetic acid
(1000 : 7.06 : 1.05 : 1.31 : 0.30)
Moxalactam ODS Methanol}0.05 mol L\1 monobasic potassium phosphate UV (254 nm)
(5 : 95) adjusted to pH 6.5
Moxalactam ODS Methanol}0.005 mol L\1 tetra-n-butylammonium UV (254 nm)
phosphate (1 : 3) adjusted to pH 6.0
Moxalactam ODS 0.1 mol L\1 Ammonium acetate}acetonitrile (95 : 5) UV (270 nm)
adjusted to pH 6.5
Moxalactam ODS 0.1 mol L\1 Sodium phosphate}methanol (84 : 16) UV (254 nm)
adjusted to pH 3.2
Temocillin ODS Methanol-0.1 mol L\1 phosphate buffer (pH 7.0, 1 : 9) UV (230 nm)
Temocillin Octyl silane Methanol}0.1 mol L\1 phosphate buffer (pH 7.0, 16 : 84) UV (230 nm)
7-Ureidoacetamido cephalosporins ODS 0.01 mol L\1 Diammonium hydrogen phosphate UV (254 nm)
containing 5}20% methanol
553C. The mobile phase consists of methanol}water Lome]oxacin

(3 : 2) with a Suorescent detection at 345 nm (excita- LomeSoxacin enantiomers are extracted from plasma
tion) and 420 nm (emission) for ampicillin, at pH 7 with a mixture of chloroform}isopentyl alco-
cephalexin and cephradine. For amoxicillin, the mo- hol}diethyl ether (71.25 : 3.75 : 25) and derivatized
bile phase is methanol}water (55 : 45) with a Suor- with S-(#)-(1-naphthyl)ethylisocyanate to form dias-
escent detection at 355 nm (excitation) and 435 nm tereomers. Derivatized diastereomers are analysed
(emission). Detection limits are 0.5 ng mL\1 for am- with a Radial Pak normal-phase column using a
picillin, 2 ng mL\1 for cephalexin and 10 ng mL\1 mobile phase of hexane}chloroform}methanol
for amoxicillin and cephradine. For sensitive deter- (64.5 : 33 : 2.5). Diastereomers are detected
mination of -lactams, pre-column derivatization Suorometrically at 280 and 470 nm for excitation
with imidazole}metal salt reagent or formaldehyde, and emission. The limit of accurate quantiRcation is
or post-column derivatization with o-phthaldial- less than 10 ng mL\1 for each enantiomer.
dehyde or Suorescamine may be applied.
HPLC conditions for several other -lactams are
O]oxacin
summarized in Table 1. Epimers of these -lactams
are separated using the conditions listed in Table 1, Serum and urine samples are diluted with
except for benzylpenicillin. 0.1 mol L\1 phosphate buffer (pH 7.0) and extracted
with dichloromethane. OSoxacin enantiomers in the
extract are reacted with L-leucinamide to form dias-
tereomers. The diastereomers are extracted with
Fluoroquinolones 1 mol L\1 HCl, and injected into an ODS column.
Among Suoroquinolone antibiotics, lomeSoxacin, of- The mobile phase is 0.2 mol L\1 phosphoric acid
loxacin and temaSoxacin are used clinically as the (with the pH adjusted to 1.85 with tetraethylam-
racemates (see Figure 6, structures 19, 20 and 21, monium hydroxide)}acetonitrile (4 : 1) with Suores-
respectively). Therefore, enantiospeciRc HPLC cence detection at 298 nm excitation and 458 nm
methods are described below for these Suoro- emission. The derivative of the S-(!)-enantiomer
quinolones. Non-chiral HPLC conditions for the elutes prior to that of the R-(#)-enantiomer with
chiral as well as other non-chiral Suoroquinolones baseline separation. Detection limits are 3 and
are summarized in Table 2. Detection limits listed 80 ng mL\1 for plasma and urine, respectively.
in Table 2 are mostly those for plasma or serum OSoxacin enantiomers are also analysed using
analysis. a chiral stationary phase (bovine serum albumin
For the second method, temaSoxacin is reacted

with acetic anhydride to form acetylated tema-
Soxacin followed by reaction with isobutylchlorofor-
mate to form cabonylamidated derivatives. These
double-derivatized temaSoxacin enantiomers are
analysed with an ovomucoid conjugated silica gel
column using a mobile phase of 0.02 mol L\1 phos-
phate buffer (pH 7.0)}acetonitrile (92 : 8). The enan-
tiomers are detected at 280 nm. The detection limit is
5 ng mL\1 for each enantiomer with a separation
coefRcient of 1.50, indicating a better resolution by
the second method.
Sulfonamides
Various sulfonamides are analysed with an ODS col-
umn using a mobile phase composed of acetic
acid}triethylamine}water}acetonitrile}methanol
(0.4 : 0.2 : 710 : 100 : 100) and detection at 254 nm.
Sulfonamides in formulations are extracted or dis-
solved using dimethylformamide, methanol or the
mobile phase.
Sulfonamides in body Suids are analysed with an
ODS column using a mobile phase consisting of
acetonitrile-water (1 : 9, changing to 9 : 1 in 10 min).
Detection is either UV at 254 nm or with a mass
spectrometer. Sulfonamides are extracted with
hexane}dichloromethane}ether (1 : 1 : 1) at pH
3.0}3.2.
Sulfamethoxazole
Figure 6 Chemical structures of lomefloxacin (19), ofloxacin
(20) and temafloxacin (21). Sulfamethoxazole and its acetylated metabolites in
body Suids are analysed with an ODS column using
covalently bonded to silica) without derivatization. a mobile phase consisting of methanol}1% acetic
Mobile phase is 0.2 mol L\1 phosphate buffer (pH acid (1 : 4, pH 2.9). Sulfamethoxazole and the metab-
8.0)}2-propanol (97 : 3). Enantiomers are detected olites are detected at 230 nm with a detection limit of
Suorometrically at 298 nm excitation, 458 nm emis- less than 1 g mL\1. Plasma is extracted with ethyl
sion. Resolution and sensitivity are poorer than those acetate, and urine is deproteinized with acetonitrile.
for the above derivatization method. Sulfamethoxazole in body Suids are also analysed
For clinical use, oSoxacin has been changed to with an ODS column using a mobile phase consisting
levoSoxacin which is the pharmacologically active of 0.067 mol L\1 phosphate buffer (pH 6.7)}
S-(!)-enantiomer. methanol (5 : 1). Sulfamethoxazole is detected at
260 nm with a detection limit of 0.5 g mL\1.
Tema]oxacin
Sulfasalazine
TemaSoxacin enantiomers in biological Suids are ex-
tracted with methylene chloride and analysed by two Sulfasalazine is decomposed in the colon to generate
types of HPLC method with derivatization. For the two biologically active drugs, i.e. sulfapyridine and
Rrst method, temaSoxacin enantiomers are reacted 5-aminosalicylic acid. Sulfasalazine in commercial
with S-(!)-N-1-(2-naphthylsulfonyl)-2-pyrrolidine preparations is analysed with an ODS column using
carbonylchloride to form diastereomers, which are a mobile phase consisting of 10}15% 2-propanol in
injected into a silica gel column. The mobile phase is 0.01 mol L\1 phosphate buffer (pH 7.7), and detec-
hexane}methyl acetate}methanol}ammonia water ted at 254 nm. A silica column is also used for analy-
(150 : 100 : 10 : 1) with UV detection at 280 nm. The sis of sulfasalazine and its degradation products in
detection limit is 5 ng mL\1 for each diastereomer commercial preparations with a mobile phase of
with a separation coefRcient of 1.05. chloroform}acetonitrile}n-butanol (4 : 1 : 1).
Table 2 Non-stereospecific HPLC conditions for chiral as well as non-chiral fluoroquinolones
Fluoroquinolone Stationary phase Mobile phase Detection1 Detection limit
Enoxacin ODS 30% Methanol in an aqueous solution UV (265 nm) 3 pmol

of 0.05 mol L\1 potassium dihydrogen
phosphate and 2% acetic acid
Ciprofloxacin ODS An aqueous solution of 18 mmol L\1 Fluorescence 200 ng mL\1
potassium dihydrogen phosphate (ex. 278 nm, em. 475 nm)
and 0.13 mmol L\1 heptane sulfonic
acid}methanol}phosphoric acid
(7 : 3 : 0.01)
Lomefloxacin ODS An aqueous solution of 0.2% sodium Fluorescence 50 ng mL\1
acetate trihydrate, 0.2% citric acid (ex. 280 nm, em. 430 nm)
monohydrate and 0.1% triethylamine
(pH 4.8)}acetonitrile (80 : 23)
Nalidixic acid ODS Water}methanol}cetrimonium bromide (UV 313 nm) 1 g mL\1
(50 : 50 : 0.12)
Nalidixic acid Amino-cyano Methanol}0.1 mol L\1 citrate buffer (UV 254 nm) 0.1 g mL\1
(pH 3, 95 : 15)
Norfloxacin Anion-exchange 0.05 mol L\1 phosphate buffer (UV 273 nm) 0.1 g mL\1
(pH 7)}acetonitrile (4 : 1)
Norfloxacin ODS Acetonitrile}0.01 mol L\1 phosphate (UV 279 nm) 20 ng mL\1
buffer (pH 2.5) containing 1 mmol L\1
triethylamine (11 : 89)
Ofloxacin ODS 0.5% Sodium acetate (pH 2.5)}aceto- (UV 300 nm) 100 ng g\1
nitrile (87 : 13) tissue
Ofloxacin ODS Tetrahydrofuran}0.06 mol L\1 Fluorescence 20 ng mL\1
phosphate buffer (pH 2.6) containing (ex. 282 nm, em. 450 nm)
3% triethylamine (5.5 : 94.5)
Pefloxacin ODS An aqueous solution of 0.2% sodium Fluorescence
acetate, 0.2% citric acid and (ex. 330 nm, em. 440 nm) 50 ng mL\1
0.1% triethylamine-acetonitrile (86 : 14)
Sparfloxacin ODS 5% Acetic acid}acetonitrile}methanol UV 364 nm 5 ng mL\1
(70 : 15 : 15)
Temafloxacin ODS 53% Acetonitrile in an aqueous Fluorescence 10 ng mL\1
solution of 40 mmol L\1 phosphoric (ex. 280 nm, em. 389 nm)
acid, 10 mmol L\1 sodium dihydrogen
phosphate, 0.2% sodium dodecyl sulfate
and 5 mmol L\1 N-acetylhydroxamic acid
Temafloxacin ODS 19% Acetonitrile in an aqueous Fluorescence 10 ng mL\1
solution of 5 mmol L\1 tetra- (ex. 275 nm, em. 450 nm)
butylammonium bromide, 10 mmol L\1
sodium hydrogen phosphate
Fleroxacin ODS 10% Acetonitrile in an aqueous Fluorescence 2.5 ng mL\1
solution of 5 mmol L\1 tetrabutyl- (ex. 277 nm, em. 445 nm)
ammonium bromide, 10 mmol L\1
sodium hydrogen phosphate
Fleroxacin ODS Phosphate buffer (pH 3)}acetonitrile} Fluorescence 5 ng mL\1
methanol (85 : 7.5 : 7.5) containing (ex. 290 nm, em. 470 nm)
tetrabutylammonium hydroxide and
triethylamine
1
ex., excitation; em., emission.
Sulfasalazine in plasma is extracted with isoamyl a methylsilane column using a mobile phase of meth-
acetate and analysed with an ODS column using anol}0.05 mol L\1 phosphate buffer (pH 7.4) con-
a mobile phase of 0.01 mol L\1 phosphate buffer (pH taining 0.1% tetrabutylammonium hydrogen sulfate
7.7)}acetonitrile (83 : 17). Sulfasalazine is detected at (22.5 : 77.5). The eluants are monitored Suorometri-
365 nm with a limit of detection of 5 ng. cally at 320 nm excitation, 389 nm emission, with
Sulfasalazine, sulfapyridine, 5-aminosalicylate and a detection limit of 0.5 g mL\1. Plasma is de-
their metabolites in plasma are analysed with proteinized with methanol.
Tetracyclines ethyl acetate is improved by formation of phenyl-

butazone}tetracycline ion-pairs.
Various tetracyclines are analysed with an octyl col-
umn using a mobile phase of methanol}acetonit-
rile}0.01 mol L\1 aqueous oxalic acid solution (pH Other Antibiotics: Azole Antifungals
adjusted to 2.0 with 28% aqueous ammonia, Itraconazole and its active metabolite (hydroxyitra-
1 : 1.5 : 5) and detection at 360 nm. conazole) in serum are analysed with a Lichrospher
Tetracycline (TC), chlortetracycline (CTC), RP8 column using a mobile phase of acetonit-
doxycycline, minocycline, oxytetracycline (OTC), rile}water (62 : 38) containing 0.05% diethylamine.
impurities of these tetracyclines including 4-epitet- The pH of the mobile phase is adjusted to 6.0 with
racycline, anhydrotetracycline and 4-epianhydrotet- 30% acetic acid. Itraconazole and hydroxyitra-
racycline (a nephrotoxic degradation product) are conazole are detected at 258 nm with detection limits
resolved with an ODS column using a gradient system of 10 and 7 ng mL\1, respectively. Serum is extracted
(see Figure 7 for chemical structures). The mobile with heptane-isoamyl-alcohol (9 : 1).
phase is an aqueous solution of 1 mmol L\1 tetra- Itraconazole and hydroxyitraconazole in plasma
ammonium ethylenediamine tetraacetate and and tissue are also analysed with an ODS column
50 mmol L\1 diethanolamine (pH adjusted to 7.3 using a mobile phase of water}acetonitrile}
with 85% phosphoric acid) containing 2}10% iso- diethylamine (42 : 58 : 0.05). The pH of the mobile
propanol. Tetracyclines are detected at 254 nm. Im- phase is adjusted to 2.45 with 85% phosphoric acid.
purities of tetracycline are also analysed with an ODS Itraconazole and hydroxyitraconazole are detected
column using a mobile phase consisting of meth- Suorometrically at 260 nm excitation and 365 nm
anol}acetonitrile}0.2 mol L\1 aqueous oxalic acid emission. Detection limits of itraconazole are
solution (pH adjusted to 2.0 with 28% aqueous am- 5 ng mg\1 and 5 ng mL\1 for tissue biopsy and
monia, 1 : 1 : 3.5). Tetracycline and impurities are plasma, respectively. Itraconazole in tissue or plasma
detected at 400 nm. is extracted with methanol.
TC, CTC and OTC in plasma and urine are ana- Fluconazole in plasma is analysed with an octyl
lysed with an ODS column using a mobile phase column using a mobile phase of water}acetonitrile
consisting of 0.01 mol L\1 phosphate buffer (pH (72 : 28). Fluconazole is detected at 260 nm with
2.4)}acetonitrile (7 : 3 or 6 : 4). Tetracyclines are de- a detection limit of 0.4 g mL\1. Plasma is de-
tected at 355 nm with a detection limit of 1 g mL\1. proteinized with acetonitrile.
Extraction of tetracyclines from biological Suids into Miconazole in plasma is analysed with an ODS
column using a mobile phase of methanol}aceto-
nitrile}0.01 mol L\1 phosphate buffer (pH 7.0,
36 : 36 : 28). Miconazole is detected at 230 nm with
a detection limit of 5 ng mL\1. Plasma is treated with
an octadecyl solid-phase extraction column prior to
HPLC analysis.
Econazole in serum is determined with an ODS
column using a mobile phase of 0.01 mol L\1 potassi-
um dihydrogen phosphate}methanol (3 : 7), with the
pH being adjusted to 4.5. Econazole is detected at
220 nm with a detection limit of 40 ng mL\1.
Sulconazole in plasma is analysed with an ODS
column using a mobile phase of acetonitrile}
0.01 mol L\1 phosphate buffer (pH 8, 66 : 34). Sul-
conazole is detected at 229 nm with a detection limit
of less than 0.5 g mL\1.
Some of the azole antifungals are used clinically as
the racemates, and the enantio-speciRc HPLC condi-
tions with a chiral stationary phase, tris(chloro-
methylphenylcarbamate)s of cellulose, are reported.
The mobile phase is n-hexane-2-propanol (85 : 15)
Figure 7 Chemical structures of tetracycline (22), chlortet- for separation of enantiomers of enilconazole,
racycline (23), doxycyline (24), minocycline (25) and oxytetracyc- econazole, miconazole and ornidazole, and n-hexane-
line (26). 2-propanol (9 : 1) for bifonazole enantiomers. A similar
III / ANTIBIOTICS / Supercritical Fluid Chromatography 2077
type of cellulosic chiral stationary phase (Chiralcel- atographic assay of lomeSoxacin in human plasma.
OD) with a mobile phase of n-hexane-2-propanol Journal of Pharmaceutical and Biomedical Analysis 13:
(9 : 1) is used for separation of sulconazole enantiomers. 1243}1248.
Griggs DJ and Wise R (1989) A simple isocratic high-
pressure liquid chromatographic assay of quinolones
Conclusion in serum. Journal of Antimicrobial Chemotherapy
Since there is an enormous volume of information on 24: 437}445.
the separation of antibiotics in the literature, readers Itoh T and Yamada H (1995) Diastereomeric -lactam
antibiotics: analytical methods, isomerization and
should be able to Rnd HPLC conditions for almost
stereoselective pharmacokinetics. Journal of Chrom-
any antibiotic of interest. Readers are also encour- atography A 694: 195}208.
aged to consult the ofRcial compendia for analysis of Kirschbaum JL and Aszalos A (1986) High-performance
bulk or formulated drugs. For analysis of bio- liquid chromatography. In: Aszalos A (ed.) Modern
logical samples, the samples may be directly injected Analysis of Antibiotics, pp. 239}322. New York: Mar-
with a column switching technique instead of em- cel Dekker.
ploying liquid}liquid or solid-phase extraction. For Lehr KR and Damm P (1988) QuantiRcation of the enantio-
sensitive detection, drugs may be subjected to pre- or mers of oSoxacin in biological Suids by high-perfor-
post-column derivatization, especially with a Suor- mance liquid chromatography. Journal of Chromatogra-
escent chromophore. Diastereomeric derivatization is phy 425: 153}161.
useful for analysis of chiral drugs. Mass spectrometric Margosis M (1989) HPLC of penicillin antibiotics. In: Gid-
dings JC, Grushka E and Brown PR (eds) Advances in
(MS) detection is another way to increase sensitivity.
Chromatography, pp. 333}362. New York: Marcel
Indeed, cephem and macrolide antibiotics are ana- Dekker.
lysed with HPLC-MS to detect minute amount of Matsuoka M, Banno K and Sato T (1996) Analytical chiral
drugs. For cephem antibiotics, capillary HPLC has separation of a new quinolone compound in biological
been coupled with mass spectrometric detection. Suids by high-performance liquid chromatography.
Journal of Chromatography B 676: 117}124.
See also: II / Chromatography: Liquid: Derivatization; Stead DA and Richards RME (1996) Sensitive Suorimet-
Detectors: Fluorescence Detection; Instrumentation. ric determination of gentamicin sulfate in biological
matrices using solid-phase extraction, pre-column
Further Reading derivatization with 9-Suorenylmethyl chlorofor-
mate and reversed-phase high-performance liquid
Foster RT, Carr RA, Pasutto FM and Longstreth JA chromatography. Journal of Chromatography B 675:
(1995) StereospeciRc high-performance liquid chrom- 295}302.
Supercritical Fluid Chromatography
F. J. Sen oraH ns and K. E. Markides, liquid chromatography (HPLC) is the most com-
Uppsala University, Uppsala, Sweden monly used, followed by thin-layer chromatography
Copyright ^ 2000 Academic Press and gas chromatography (GC), while supercritical
Suid chromatography (SFC) is still being introduced
to this area of application.
In SFC the mobile phase is a Suid subjected to
Introduction pressures and temperatures near or above the critical
point of that Suid, to enhance and control the mobile-
The analysis of antibiotics is of primary importance phase solvating power. This fact determines that the
for drug monitoring in pharmacokinetic and health mobile-phase properties (e.g. diffusivity, density, vis-
studies, as well as for the quality control of drug cosity) are intermediate between those of gases and
production and of numerous food products. As a con- liquids and can be varied and controlled by small
sequence, the demand for new methods of determina- changes in the pressure or temperature of the systems.
tion of antibiotics of very different types is continu- The most common Suid used in SFC is carbon diox-
ously increasing. The main methods employed for ide, which has a critical temperature of 313C, allow-
these analyses include immunoassays and chromatog- ing the separation of thermally labile compounds
raphy, as well as various chemical techniques. Among under mild conditions. In general, antibiotics are
the chromatographic methods, high performance compounds with intermediate to high polarity, while
2078 III / ANTIBIOTICS / Supercritical Fluid Chromatography
supercritical carbon dioxide only has an adequate SFC versus HPLC for
solvating power for nonpolar compounds. For that the Determination of Antibiotics and
reason, the elution of more polar solutes requires the
addition of a more polar organic solvent (for
Related Drugs
example, 5}30% methanol), the so-called modiRer, The main problem posed in the separation of anti-
that increases the polarity of the mobile phase and has biotics is their broad range of structures that
a solvating effect on silica-based packed columns. covers almost the whole range of organic chemistry.
With this technique it is possible to separate anti- This includes carbohydrate, macrocyclic lactones,
biotics in complex samples at lower temperatures quinones, peptides and heterocyclic compounds,
than GC and in shorter times than liquid chromatog- although antibiotics in general are relatively polar,
raphy. The use of low concentrations of additives in nonvolatile and thermally labile drugs. For that rea-
the modiRer (for example, 0.1% triSuoroacetic acid son, liquid chromatography has increasingly been
and/or 0.1% triethylamine) is also used to control the chosen as the method of analysis, and SFC is gaining
separation conditions to an even greater extent, espe- greater acceptance with the extended use of packed
cially retention, peak shape and enantioselectivity. columns combined with organic modiRers and addi-
SFC can be divided into two categories based on tives that allow the separation of the more polar
column type } open tubular and packed } with differ- solutes.
ences in selectivity, detection and need for modiRer GC commonly provides the highest resolution in
addition to the carbon dioxide characterizing the two the shortest analysis time, but it also requires high
types. Both types have been employed in the separ- temperatures and often derivatization of the drugs.
ation of drugs in very different samples. Separations HPLC methods have lower resolution and longer
of antibiotics and related compounds are best per- analysis times. The SFC technique can provide the
formed using packed columns, although there are same resolution as GC and short run times, but with
some applications that do well on open tubular col- the added beneRt that it does not need high temper-
umns. Packed columns can be used with UV detection atures (Figure 1). Typical temperatures are as low as
and a wide range of packing materials, from pure 50}803C in packed-column SFC.
silica, to phenyl, diol, amino, octadecyl-modiRed sil- The packed columns used for SFC of polar drugs
ica or chiral materials such as cyclodextrins, de- are similar to the columns used in HPLC, although
rivatized cellulose or amylose. For these silica-based back-pressure is not a problem in SFC, allowing col-
columns, the peak symmetry is improved and the umns to be coupled in series to achieve high resolu-
retention times of the antibiotics are shortened when tion systems, even for polar analytes. When it comes
a modiRer is added to the carbon dioxide, due to the to detection, the commonest systems for antibiotics
solvating effect on the free silanol groups of the silica determination are UV (Figure 2). or mass spectro-
(cf. end-capping in HPLC). In general, the separation metry (MS). In comparing packed-column SFC and
of antibiotics in SFC is affected by the number, loca- LC, the ultraviolet detector can be operated at lower
tion, nature and conformation of the individual func- wavelengths in SFC, because of the lack of back-
tional groups, which can deRne the need or not, as the ground absorbance from the solvent and the mass
case may be, of a modiRer and additive. Nevertheless, spectrometric ionization techniques work best with
the determination of antibiotics by SFC is not so volatile mobile phases also favouring SFC (Table 1).
straightforward as other pharmaceutical separations
and until now it has not been thoroughly developed. Characteristics of the Separation of
The area that has been developed the most is prob-
ably the chiral separation of antibiotics and related
Antibiotics using Supercritical Fluids
compounds, where the combination of a temperature The main properties of SFC which signiRcantly affect
that is milder and more selective than GC and an antibiotic separation are related to the high solvating
efRciency better than HPLC results in enhanced res- power of supercritical Suids and their low viscosity,
olution, which is especially valuable. which yields high resolution power and throughput.
Another area where supercritical Suids have This fact has two main consequences on this type of
a niche is the monitoring of food products for anti- separation. Firstly, as already pointed out above,
biotic residues. This area is of increasing importance compounds like antibiotics can be analysed at lower
due to the concern for the effect on human health that temperatures than in gas chromatography, and in
abuse of these drugs can have. In this case, the main shorter times than in liquid chromatography, as a
advantage of supercritical Suids is in the sample prep- result of good solvating capacity. Secondly, SFC is
aration from these complex matrices, when super- able to resolve complex mixtures of not very volatile
critical Suid extraction coupled to SFC can be used. compounds, allowing the direct injection of samples
Figure 1 Structures of eight sulfonamides determined by SFC (see chromatogram shown in Figure 2).
that contain antibiotics with little or no sample pre- antibiotics, and presents an alternative approach to
treatment. several LC-MS methods.
Some antibiotics may be degraded or lost during
exposure to light, heat or extreme values of pH. In Online SFE-SFC for the Analysis
SFC, all these factors are avoided, providing separ-
ation under mild conditions that preserves the integ-
of Antibiotics
rity of the sample. Online supercritical Suid extraction (SFE)-SFC can be
Antibiotic determination presents some difRculties used for the analysis of antibiotics in complex sam-
due to the complexity of the sample matrix and the ples, resulting in time savings and less exposure to
relatively low concentration of the antibiotics in these organic solvents. Multidimensional systems take ad-
samples. The whole procedure, including extraction vantage of two orthogonal separation techniques
and fractionation, is not only time-consuming and with complementary characteristics, for example one
prone to error, but may degrade labile antibiotics and extraction and one chromatographic step, where the
create artefacts. Consequently, new approaches have Rrst step is aimed at producing a clean and undiluted
been developed in the last few years that avoid several sample containing the compounds of interest, and the
or all of these questionable sample preparation steps second step provides a high resolution separation of
by using multidimensional systems or even direct the target analytes.
injection in SFC. The main advantages of these online systems is that
The analysis of antibiotics in human serum has a fast and automatic sample preparation reduces or
been performed with both open tubular and packed- avoids the errors of the manual steps. Also, solvent
column SFC. With open tubular columns, it is pos- consumption is less, which reduces toxic hazards and
sible to determine the antibiotic content with a simple disposal costs. As is often the case in chromatogra-
liquid}liquid extraction. The use of a Same ionization phy, the largest source of error in the quantitative
detector has, however, not demonstrated satisfactory analysis of antibiotics and the most time-consuming
detection limits to date. Better results have been ob- steps occur in the sample preparation and extraction
tained with SFC and mass spectrometric detection, stages.
which thus provides a very useful method for the Supercritical Suid techniques have a number of ad-
determination of low levels of impurities in macrolide vantages for use in multidimensional chromatographic
for the analysis of trace levels of residues in bioSuid

samples and also allows class-selective extractions
based on the tuneable polarity of the extracting agent,
which thus represents an additional clean up stage
and further reduces interface from the matrix.
This coupling has to date been applied to separ-
ations in open tubular columns of compounds in
water samples, and may provide a breakthrough de-
velopment for the future, with the use of packed
capillary column SFC-MS for polar analytes. Al-
though this method has not yet been applied to the
determination of antibiotics, it could provide an auto-
matic way of analysing drugs in biological Suids
directly, i.e. with no separate extraction step.
Analysis of Antibiotic Residues in Food

Figure 2 Chromatogram obtained from SFC of a mixture of Antibiotics have been used in animal feed for several
sulfonamides with UV (wavelength 270 nm). Peak identification: decades to control infections and promote growth
A, sulfadoxine; B, sulfamethazine; C, sulfamerazine; D, sulfa- (Figure 3). Recently, increasing concern about anti-
dimethoxine; E, sulfadiazine; F, sulfaquinoxaline; G, sulfachlor- biotic resistance has led to the prohibition of the use
pyridazine; H, sulfathiazole. Chromatographic conditions: column
of some antibiotics for this purpose in several coun-
packed with 5 m particle amino-bonded Spherisorb (100;
4.6 mm i.d.), column temperature 903C, CO2 flow rate of tries. Consequently, there is a growing interest in new
4 mL min\1, pressure 361 bar. Mobile phase was CO2 modified and improved methods of analysis for antibiotics and
initially with 15% methanol and after 4 min with 25% methanol. their residues in food, and LC-MS is a common tech-
Reproduced with permission from Perkins JR, Games DE, Startin nique used to achieve this goal.
JR and Gilbert JJ (1991) Analysis of sulfonamides using super-
In solid and heterogeneous matrices such as food,
critical fluid chromatography and supercritical fluid chromatogra-
phyImass spectrometry. Journal of Chromatography 540: 239. sample preparation is the most time-consuming step
in the routine determination of analytes in trace
amounts, for instance, antibiotic residue levels. It
systems. The commonest multidimensional system, remains the largest source of error in quantitative
LC-GC, is limited to the determination of thermally analytical methods. For this reason, the development
stable and volatile solutes, while SFC can substitute of methods with less sample pretreatment require-
the Rrst fractionation step, as well as the second step ments and with the possibility of automation is
of high resolution chromatography. In the case of desirable.
SFE-SFC, the transfer is performed without changes For this application, SFC in combination with the
in the mobile phase, which minimizes losses of universal Same ionization detector, or with highly
analytes and reduces technical complexity sensitive mass spectrometric detection, shows a good
balance between high resolution and good sample
Analysis of Aqueous Matrices
throughput, that can minimize the need for sample
Recently, new methods for the direct injection of clean-up and be an optimal procedure in speciRc
water samples on to an adsorbent with solvent vent- cases. An example of separation of veterinary anti-
ing and online SFE-SFC of the target solutes have biotics by SFC with UV detection prior to online
been developed. SFC-MS can be seen in Figure 4.
In this procedure, the liquid sample is introduced in An alternative approach is the online coupling of
the SFE cell Rlled with a suitable adsorbent, which SFE and SFC for solid or semi-solid samples, allowing
retains the solutes while the aqueous solvent is vented the extraction of the fraction of interest and the
with an inert gas. The venting of the water improves online transfer of the solutes from the liquid or solid
the performances and Sexibility of both the separ- matrix directly to the chromatograph, reducing sol-
ation and the detection steps. After elimination of vent usage and the need for clean-up.
solvent, the analytes are extracted with supercritical
carbon dioxide, and focused in a cryogenic trap,
providing a solute enrichment before automatic on-
Chiral Separation of Antibiotics
line injection into the SFC column. In addition to its The stereochemistry of an antibiotic is a prominent
speed, this method provides a preconcentration step issue in the development, approval and clinical use of
Table 1 Determination of antibiotics by supercritical fluid chromatography
Column type Sample Comments Detector Reference
Packed (1 mm i.d., 5 m particles)Sulfonamides and 1003C UV Schmidt S, Blomberg, LG

and open tubular (50 m i.d.) tetracyclines 8% Isopropanol as modifier and Campbell ER (1988)
columns Chromatographia 25: 775
Packed column 150;4.6 mm Erythromycin A 653C MS UV Lane SJ (see Markides and
5 m particle amino silica Cephalosporins 2}8% Methanol as modifier Lee, 1988)
Packed column 100;4.6 mm Sulfonamides, veterinary 75}903C MS UV Perkins JR, Games DE,
5 m particle amino-bonded drugs 15}25% Methanol as modifier Startin JR and Gilbert JJ
Spherisorb (1991) Journal of
Chromatography 540: 239
Packed column 250;4.6 mm Lactams 253C Diode array Caude and Macaudiere
10 m particle Chiralcel OB Chiral separation (see Markides and Lee,
1988)
Open tubular 10 m;50 m i.d. Cyclosporine A FK 506 703C FID Wong et al. (1994) Journal
SB-biphenyl-100 (Tacrolimus) Rapamycin Whole blood extracts of Liquid Chromatography
17: 2093
Open tubular 5 m;50 m i.d. Macrolide antibiotic 1003C MS Ramsey et al. (1995)
DB-5 Midecamycin A1 Standard solutions Analytical Proceedings 32:
455
Different packed columns Sulfonamides 50}1003C UV Combs MT, Ashraf-
250;4.6 mm 5 m particles Great influence of temperature Khorassani M and Taylor
on resolution LT (1997) Journal of
Chromatographic Science
35: 176
This summary is not intended to be a comprehensive review of all antibiotic separations by SFC; it aims to provide general information
on the main applications in this field.
these drugs. For the separation of enantiomers, SFC Chiral separations in SFC can be carried out using
with chiral stationary phases is very convenient due open tubular columns with immobilized cyclodex-
to its high resolution and relatively low analysis tem- trins or, more recently, by packed columns with most
perature. of the same phases commonly used in LC, since the
Figure 3 Structures of some veterinary antibiotics analysed by SFC.

will be very valuable in the accurate determination of

high and low concentration of antibiotics in samples
where high resolution and mild conditions are im-
perative.
It is becoming more frequent to use solvents under
subcritical conditions, which is blurring the boundary
between SFC and LC } a fact that is already being
used in chiral separations and will probably become
more frequent in separations of antibiotics and re-
lated drugs.
Another probable source of improvement is the use
of new detectors with higher sensitivity than UV and
Same ionization detection and at the same time com-
patible with the use of modiRers in packed capillary
column SFC. An example is the new amperometric
detectors which avoid the problems observed in the
quantitation of some drugs by SFC.
The development of new generations of commer-
cial equipment for SFE-SFC and SFC-MS that are
more user-friendly and robust than the present ones
would also contribute to a wider use of this tech-
nology in quality control and research of pharma-
Figure 4 Chromatograms obtained from SFC of a mixture of ceutical compounds.
veterinary antibiotics with UV (A, wavelength 215 nm; B,
wavelength 230 nm). Peak identification: A, levamisole; B,
furazolidone; C, chloramphenicol; D, lincomycin. Chromato- Further Reading
graphic conditions: column packed with 5 m particle amino- Agarwal VK (ed.) (1992) Analysis of Antibiotic/Drug Resi-
bonded Spherisorb (100;4.6 mm i.d.), column temperature dues in Food Products of Animal Origin. New York:
753C, CO2 flow rate of 4 mL min\1, pressure 351 bar. Mobile
Plenum Press.
phase was CO2 modified with 15% methanol. Reproduced with
Ahuja S (ed.) (1992) Chromatography of Pharmaceuticals.
permission from Perkins et al. (1991).
Washington, DC: American Chemical Society.
Jinno K (ed.) (1992) Hyphenated Techniques in Supercriti-
chiral selector is covalently bound to the packing cal Fluid Chromatography and Extraction. Amsterdam:
material. The latter method frequently requires the Elsevier.
addition of a modiRer and is more common in the Lee ML and Markides KE (eds) (1990) Analytical Super-
separation of drugs. critical Fluid Chromatography and Extraction. Provo,
Several examples of the SFC of antibiotic enantio- UT: Chromatography Conferences.
mers can be found in the literature, although the Markides KE and Lee ML (1988) SFC Applications. Provo,
UT: Brigham Young University Press.
technique is not as commonly used as LC. Packed
Medvedovici A, Sandra P, Toribio L and David F (1997)
columns with chiral stationary phases are normally Chiral packed column subcritical Suid chromatography
employed with carbon dioxide modiRed with meth- on polysaccharide and macrocyclic antibiotic chiral
anol or ethanol as mobile phase, under supercritical stationary phases. Journal of Chromatography A 785:
or subcritical conditions (i.e. at room temperature). 159.
Further applications can be expected from the use of Perkins JR, Games DE, Startin JR and Gilbert J (1991)
packed capillary columns for chiral separations, Analysis of veterinary drugs using supercritical Suid
which may provide better resolution and shorter chromatography and supercritical Suid chromatogra-
analysis times than the equivalent LC separation. phy}mass spectrometry. Journal of Chromatography
540: 257.
Sen oraH ns FJ and Markides KE (2000) On line SFE-SFC for
Future Developments the analysis of fat soluble vitamins and other lipids from
water matrices. In: Williams JR (ed.) Methods in Mo-
Current developments in new types of columns, lecular Biology: Supercritical Fluid Methods and Proto-
equipment and detectors for SFC show that this tech- cols. Totowa, NJ: Humana Press.
nique has great potential for expansion and will Xie LQ, Markides KE and Lee ML et al. (1993) Bioanalyti-
achieve a broader use in the future with the advent of cal application of multidimensional open tubular col-
new instruments for packed capillary columns and umn supercritical Suid chromatography. Chromatog-
adaptation of the routine use of MS-detectors, which raphia 35: 363.
III / ARCHAEOLOGY: USES OF CHROMATOGRAPHY IN 2083
ARCHAEOLOGY: USES OF
CHROMATOGRAPHY IN
C. Heron and R. Stacey, University of Bradford, UK speciRc identiRcations than was possible before.
GC}IRMS allows the ratios of abundances of stable
Copyright ^ 2000 Academic Press isotopes of elements such as carbon and nitrogen to
be determined for individual compounds introduced
via a gas chromatograph. Stable isotope ratios are of
History particular importance to studies of foodwebs due to
the characteristic isotope signatures of plants utilizing
Although recent advances in analytical methods
different photosynthetic pathways. These distinctive
have accelerated the study of these materials, the
ratios are passed along the food chain to herbivores
analysis and identiRcation of ancient organic residues
and carnivores. The method requires very small sam-
has a long history. An early example, in the 1920s,
ples and is being applied to trace organic residues in
was the use of wet chemical techniques by the chemist
pottery vessels to establish their origin with a high
Alfred Lucas to study organic material from pottery
degree of precision.
and mummiRed human remains from the tomb of
Tutankhamun. Over the last 20 years or so, the
analysis of organic residues has grown into a recog-
nized Reld in its own right. Examples of organic
Methods
residues include the debris associated with the Analysis of archaeological material presents a num-
remains of food and other natural products as a result ber of challenges, including the small amount of
of their manipulation in pottery containers (e.g. sample available, the presence of complex molecular
cooking of food), the balms in the wrappings of mixtures from more than one source, chemical alter-
mummiRed bodies and traces of colouring dyes im- ation due to processing or degradation, and contami-
pregnated in ancient textiles. Given the amorphous nation. Furthermore, every sample is unique. These
character of organic residues, the most effective ap- factors mitigate against simple interpretations of ana-
proach to their identiRcation lies in their chemical lytical results.
composition. Characterization of organic residues Recent developments in instrumental chromato-
generally relies upon the principles of chemotaxon- graphic techniques have enabled trace amounts of
omy, where the presence of a speciRc compound or organic residues to be detected. Hence it is possible to
distribution of compounds in an unknown sample is analyse molecules surviving in an inorganic matrix
matched with its presence in a contemporary natural such as pottery or soil, or surviving in morphological
substance. The use of such molecular markers is not organic remains such as lipids in seeds or bone. Insol-
without its problems, since many compounds are uble or polymeric fractions of residues that are not
widely distributed in a range of natural materials, and themselves volatile enough for conventional analysis
the composition of an ancient residue may have can be broken up by pyrolysis, thereby allowing sep-
changed signiRcantly during burial. In general, mo- aration and identiRcation of the fragments. Pyrolysis-
lecular markers belong to the compound class deRned GC-MS has been successfully applied to the recogni-
as the lipids, a heterogeneous group of molecules tion of biopolymers in fossil and recent higher
which includes fats and oils and molecules with com- plant resins, and to macromolecular debris remaining
mon solubilities, such as the constituents of resins and from the burning of food in archaeological pottery
waxes. vessels.
Early work in this Reld relied heavily on either Preparation of ancient lipids and natural products
thin-layer chromatography (TLC) or gas chromatog- normally involves solvent washing of samples. Pre-
raphy (GC) alone to characterize residues. Today, fractionation of the lipid residue can be undertaken
combined GC}mass spectrometry (GC-MS) and, to using microscale column chromatography or TLC.
a lesser extent, high-performance liquid chromatog- Prior to analysis, unhindered acid functionalities
raphy}MS (LC-MS) are demonstrating considerable are derivatized by treatment with diazomethane.
value in identifying ancient organic matter. The Trimethylsilylation using N,O-bis(trimethylsilyl)
wider availability of these techniques and, in par- triSouroacetamide (BSTFA)#1% trimethylchloro-
ticular, the introduction of GC}isotope ratio mass silane (TMCS) is used for the derivatization of hin-
spectrometry (GC}IRMS), is contributing to more dered carboxyl groups and alcohols. In some cases an
2084 III / ARCHAEOLOGY: USES OF CHROMATOGRAPHY IN
internal standard is added to the sample to quantify or animal source (or indeed whether lipids from
yield. both are present). Odd and branched-chain fatty
GC remains a useful screening technique prior to acids may also be present. These are characteristic
GC-MS and can provide Rngerprint chromatograms, components of bacteria. They are, however, also in-
whereby a complex set of peaks in a mixture can be troduced into ruminant adipose tissue by bacteria in
matched to those in reference samples. Nevertheless, the rumen and migrate throughout the animal’s body,
since molecular alteration is likely, this approach contributing to all the tissues. The presence of ap-
must be exercised with caution. Combined GC-MS preciable levels of these components, both in the free
provides valuable structural information on each of state and as components of the acylglycerol fraction,
the components separated, and permits identiRcation supports the view that the predominant source of
of molecular modiRcation. lipid in the example shown derives from a ruminant
animal.
IdentiRcation of thermal degradation products
Applications may give clues to vessel use. The long chain ketones
in Figure 1 are formed by a high temperature react-
Residues Associated with Pottery
ion of fatty acids that is catalysed by the mineral
Fragments of broken and discarded pottery matrix of the pottery fabric. Thus vessel use may
vessels are one of the most common classes of be further understood by linking the molecular com-
archaeological Rnd. These sherds offer few immediate position of the residues with exposure to high temper-
clues as to their original content and use } a signiR- atures during formation, for example, during
cant point of enquiry in archaeology. During use, cooking. Recent research suggests that animal fats
however, pottery vessels are known to accumulate (such as adipose tissue, dairy products and Rsh/
residues of foods processed in them. If these residues marine mammal oils) and plant tissues (notably the
survive long-term burial then they offer potential waxy compounds coating the surfaces of leaves) have
for determining artefact use. The residues occur the ability, under favourable burial conditions, to
as both charred or burnt deposits, which can be survive.
observed on the surface of the pottery, and as Pottery sherds may also exhibit the remains of
absorbed residues whereby organic components organic surface treatments or sealants preserved as
migrate into the pores of the vessel fabric. The porous surface deposit. These are often resins, waxes or tars.
microstructure of the fabric offers some protection GC analysis of one such deposit, a burnt surface
to the residue from the effects of biodegradation residue on a neolithic potsherd, from Ergolding Fis-
and leaching during burial. The lipid constituents chergasse, Germany (mid 4th millennium BC), led
of these residues preserve rather well, and these to its identiRcation as beeswax (Figure 2). The
can be extracted (by solvent washing of powdered chromatograms shown compare wax ester distribu-
sherd) from excavated sherds and analysed by GC tions in fresh beeswax (Apis mellifera) with the frac-
and GC-MS. tion extracted from the surface deposit removed from
A gas chromatogram from a typical degraded fat the neolithic sherd. The principal wax esters in both
residue recovered from an archaeological sherd of samples are even-carbon-numbered aliphatic chains
Iron Age date (c. 100 BC) is shown in Figure 1B. The of saturated alcohols and fatty carboxylic acids with
residue is rich in acylglycerols and free fatty acids total carbon numbers in the range C40 to C50, with the
and is typical of a partially hydrolysed lipid. This can C46 wax ester the most abundant. The unsaturated
be compared with the composition of fresh mam- wax esters present in the natural beeswax are absent
malian depot fat (Figure 1A), which is dominated from the neolithic residue. This is due to the deleteri-
by intact triacylglycerols. The monoacylglycerols, ous effects of burial, during which the double bond is
diacylglycerols and free fatty acids in the degraded rendered susceptible to oxidation or reduction reac-
fat result from the hydrolytic processes which begin tions. Natural beeswax also contains a considerable
as the pot is used (e.g. during boiling of food) alkane component (in the range C21}C33), yet this was
and continue during burial. Furthermore, lipid severely depleted in the archaeological sample, sug-
residues are depleted in unsaturated fatty acids (such gesting its combustion when the beeswax was
as oleic acid; C18:1). This illustrates the problems burned. The sealing and water-repelling properties of
in making simple comparisons between ancient beeswax suggest that it may have been used to seal the
lipids and fatty acid compositions of modern fats and vessel to enable it to hold liquids. It is possible,
oils. however, that the vessel was used to store the bees-
Fractionation of the lipid to obtain minor constitu- wax for other uses. The identiRcation of this com-
ents, such as sterols, can assist in determining a plant modity also implies the availability of honey to
Figure 1 Partial gas chromatograms showing the compositions of (A) fresh beef fat and (B) the lipid residue extracted from an
Iron-Age cooking vessel from Easingwold, Yorkshire, UK. The peak identities were established by GC-MS and are as follows: F14}F18
denote saturated fatty carboxylic acids with 14}18 carbon atoms respectively; F18:1 denotes a monounsaturated fatty acid with 18
carbon atoms; M16 and M18 are monoacylglycerols with 16 and 18 fatty acyl carbon atoms respectively; K31, K33 and K35 are
mid-chain ketones with 31, 33 and 35 carbon atoms respectively; D34 and D36 represent diacylglycerols with 34 and 36 fatty acyl
carbon atoms respectively. T46}T54 are triacylglycerols with 46}54 fatty acyl carbon atoms respectively. H Internal standard.
Analytical conditions: gas chromatography was carried out on a Hewlett Packard 5890 series II gas chromatography, equipped with
a flame ionization detector. Samples were introduced by on-column injection into a 60 cm;0.32 mm i.d. retention gap of deactivated
polyimide-clad fused silica capillary tubing connected to the analytical column via a capillary connector. The carrier gas was helium at
a constant flow of 1 mL min\1. The temperature of the oven was programmed from 50 to 3403C at a rate of 103C min\1 following
a 2-min isothermal hold at 503C after injection, with the final temperature held for 8 min.
The combined GC-MS was performed using a Hewlett Packard 5972A quadrupole mass selective detector in conjunction with
a Hewlett Packard 5890 series II gas chromatograph. Samples were introduced via a splitless injector at 3403C with a 3-min purge time.
Helium carrier gas was at constant pressure of 10 psi. Mass spectra were recorded over a mass range of 50}700 m. The MSD
interface temperature was 3403C, and the temperature was programmed from 50 to 3403C at a rate of 103C min\1 following a 2-min
isothermal hold at 503C after injection, with the final temperature held for 12 min.
In both cases, the analytical column was a polyimide-clad 12 m;0.22 mm i.d. fused silica capillary coated with BP1 stationary phase
(immobilized poly(dimethylpolysiloxane), OV-1 equivalent, 0.1 m film thickness, SGE, UK).
neolithic communities in Europe. GC-MS has also due type and pottery form and fabric, providing gen-
been used to identify beeswax residues associated eral assessments of use within assemblages.
with medieval ceramics, and in lamps from late
Amorphous Residues and Adhesives
Minoan Crete where the wax was burned as a fuel.
Analysis of a large number of vessels from an Amorphous organic substances can survive in
archaeological site enables correlation between resi- other contexts, such as on stone tools, or as isolated
Many of the early analyses have recently been con-

Rrmed by GC-MS, including lumps with tooth
impressions, interpreted as a very early form of chew-
ing gum.
Figure 3 compares fresh birch bark tar with sam-
ples from the Mesolithic site of Star Carr (Yorkshire,
UK). The triterpenoids of the outer bark of Betula sp.
are derivatives of lup-20(29)-ene and, to a lesser ex-
tent, olean-2-enes. The triterpenoid composition is
modiRed slightly by heating the bark and by some
9000 years of water-logged burial, but the identity
and relative abundance of these biomarkers is sufR-
cient to characterize the archaeological samples. Tars
produced from other bark and wood samples have
a very different molecular composition. For example,
softwoods produce diterpenoid compounds and are
easily distinguishable, while the barks and tars of
other trees such as hazel, rowan and willow produce
triterpenoids but with different carbon skeletons or
relative abundance of the lup-20(29)-enes. Analysis
by GC-MS enables identiRcation of the molecular
markers of the heating of the bark and post-depos-
Figure 2 Partial gas chromatograms comparing the wax ester itional alteration (Figure 4).
distribution in (A) Neolithic deposit on a potsherd from Ergolding Bitumen represents the fraction of sedimentary or-
Fischergasse, Germany with (B) authentic beeswax (Apis melli-
fera ). Peak identities: 1}6 are wax esters in the range C40 (peak 1)
ganic matter which is soluble in organic solvents. The
to C50 (peak 6) comprising mostly hexadecanoic (palmitic) acid liquid or semi-solid varieties of bitumen were widely
esterified with alcohols of increasing chain length (C24 to C34). used in the Near East and Middle East in antiquity,
Peaks 1 to 5 represent co-elution of hydroxymonoester isomers serving as a multipurpose glue and water-prooRng
and are seen in both samples. In contrast, peaks *1 to *5 are only material, a building mortar, medicinal agent and as
present in the authentic sample and represent wax esters com-
prising an unsaturated (octadecanoyl) fatty acid moiety. Their
one of the constituents of the organic preparations
absence in the ancient samples is not unexpected given the applied to mummiRed bodies in Ancient Egypt. Com-
susceptibility of the double bond to oxidation or reduction reac- pounds consistent with a bituminous substance in-
tions. Reproduced with permission from Heron C, Nemcek N, clude saturated hydrocarbons which have linear (al-
Bonfield KM et al . (1994). The chemistry of Neolithic beeswax. kylated alkanes) or cyclic (steranes, terpanes) carbon
Naturwissenschaften 81: 266I269. Courtesy of Springer Verlag.
skeletons. These molecules largely derive from micro-
scopic plants deposited in the sediments as well as
bacterial inputs. It has proved possible to identify
aggregates. An example is birch bark tar, which has molecular and isotopic characteristics of the bitumen,
been used as multipurpose natural product for at least which enables archaeological Rnds to be assigned to
10 000 years, and its use continues to the present day a particular source of bitumen. At the site of Susa,
in some parts of eastern and south-eastern Europe. Iraq (dating from the beginning of the 4th millennium
The tar is obtained by heating fresh birch bark BC), bitumen was deliberately mixed and heated with
(Betula sp.) at temperatures of 250}3503C. Spectro- mineral elements, to produce a substance known as
scopic and chromatographic investigations of the ma- bitumen mastic } a product ideal for fashioning dec-
terial began during the 1960s, and a variety of tech- orative objects by sculpture.
niques such as TLC, infrared and nuclear magnetic
Understanding Archaeological Sites
resonance (NMR) spectroscopy were used to identify
birch bark tar on Sint implements, lithics, ceramic In addition to extant residues, chromatographic ana-
and lumps of tar with human tooth impressions. lyses can be used to identify the remains of ancient
More recent analysis has revealed that the tar human activities that are invisible to the archaeologi-
was used to glue Sint tips to arrows belonging to cal excavator. IdentiRcation of -stanols, which are
OG tzi } the ‘ice man’ discovered in the Tyrolean faecal biomarkers, in soil samples from archaeologi-
Alps in 1991. The tar has also been identiRed on cal sites have enabled identiRcation of speciRc site
potsherds, stone implements and worked bone from features such as cess pits. The approach may also be
Ergolding Fischergasse (mid 4th millennium BC). used on a large scale to look at issues of waste
Figure 3 Partial gas chromatograms obtained by analysis of the solvent-soluble portions of samples of birch bark tar (Betula
pendula) (peak identities were confirmed by GC-MS and are identified in Figure 4). (A) Birch bark tar prepared from fresh bark in the
laboratory (3503C); (B) mesolithic sample from Star Carr (‘resin cake’); (C) mesolithic sample from Star Carr (hafting glue). Reproduced
with permission from Aveling EM and Heron C (1998). Identification of birch bark tar at the mesolithic site of Star Carr. Ancient
Biomolecules 2(1): 69}80. Reproduced with permission of the copyright owners OPA (Overseas Publishers Association) N.V.
disposal and manuring patterns and early results sug- tracing alteration products of purine and pyrimidine
gest that the identiRcation of speciRc sources of faecal bases in nucleic acid extracts of animal and plant
matter may be possible. remains.
Other Applications
These examples cover only a small part of the spec-
Summary
trum of archaeological approaches making use of Today, chromatography is embedded in the battery
chromatographic techniques. It should be emphasized of analytical approaches used to interrogate the sur-
that high performance liquid chromatography has viving materials and residues of past societies. The
been used not only for the separation of amino acids acceleration of research in bimolecular archaeology
and peptides (for the purposes of dating, amino acid in the last decade can largely be attributed to the
racemization studies and isotopic investigations), but availability of increasingly sophisticated analytical
also in the study of ancient wine residues in pottery techniques. GC-MS is becoming the routine approach
containers from the Old World, the analysis of for the characterization of lipids and natural prod-
ancient dyes, the identiRcation of alkaloids (such as ucts, and compound-speciRc carbon isotope deter-
caffeine and theobromine characteristic of cacao in minations are proving their value in identifying the
Mayan archaeological ceramics from Mexico) and in origin of residue components. Chromatographic
Figure 4 Structures of birch bark triterpenoids identified in Figure 3. (Reproduced with permission from Gundel LA and Lane DA
(1999).)
techniques are contributing ever more to our under- graphy; Thin-Layer (Planar) Chromatography. Amino
standing of the relationship between past human Acids and Derivatives: Chiral Separations. Lipids:
populations and their use of plant and animal resources, Gas Chromatography; Liquid Chromatography; Thin-
and of myriad ways in which artefacts were used. Layer (Planar) Chromatography.
See Colour Plate 55.

Further Reading
See also: II/Chromatography: Gas: Derivatization; De-
tectors: Mass Spectrometry; Pyrolysis Gas Chromatogra- Connan J and Deschesne O (1996) Le Bitume a% Suse:
phy. Chromatography: Liquid: Detectors: Mass Spec- Collection du Muse& e du Louvre. Paris, France: DeH parte-
trometry. III/Alkaloids: Liquid Chromatography; Gas ment des AntiquiteH s Orientales, MuseH e du Louvre.
Chromatography; Thin-Layer (Planar) Chromatography. Evershed RP, Dudd SN, Charters S et al. (1999) Lipids
Amino Acids: Gas Chromatography; Liquid Chromato- as carriers of anthropogenic signals from prehistory.
III / ART CONSERVATION: USE OF CHROMATOGRAPHY IN 2089
Philosophical transactions of the Royal Society of Mills JS and White R (1994) The Organic Chemistry of
London B 354: 19}31. Museum Objects. Oxford: Butterworth-Heinemann.
Summary of recent pioneering work in lipid analysis of Orna MV (ed.) (1996) Archaeological Chemistry: Organic,
archaeological materials. Inorganic and Biochemical Analysis. ACS Symposium
Heron C and Evershed RP (1993) The analysis of organic Series 625. Washington, DC: American Chemical So-
residues and the study of pottery use. In: Schiffer MB ciety.
(ed.) Archaeological Method and Theory V, pp. Pollard AM and Heron C (1996) Archaeological Chem-
247}286. Tucson, AZ: University of Arizona Press. istry. Cambridge: Royal Society of Chemistry.
Lambert JB (1997) Traces of the Past: Unravelling the Includes a chapter on the identiTcation of natural prod-
Secrets of Archaeology Through Chemistry. Reading, ucts (resins, pitch and waxes) from European prehistoric
MA: Addison-Wesley. sites.
AROMAS: GAS CHROMATOGRAPHY

See III / FRAGRANCES: GAS CHROMATOGRAPY
ART CONSERVATION:
USE OF CHROMATOGRAPHY IN
S. L. Vallance, Royal Society of Chemistry, oils, gums and proteinaceous materials such as egg,
London, UK milk and collagen glues have all been incorporated
Copyright ^ 2000 Academic Press into paint layers.
Oil painting was popular in northern Europe from
before the 13th century and analytical evidence sug-
Introduction gests that linseed oil was favoured, whilst in Italy,
where oil painting was introduced in the 15th cen-
Analytical science plays a vital role in the conserva-
tury, walnut oil was initially preferred. The oils most
tion of our artistic heritage and chromatography is
widely used in western European art are linseed,
one of the most valuable techniques available to the
walnut and poppy, though the use of other oils, such
conservation scientist.
as tung and safSower, has become more common in
In order to design the optimum safe conserva-
recent years.
tion/restoration treatment plan, which takes account
Plant gums are commonly found as adhesives and
of the nature of the original materials used by the
binders. Gum arabic is primarily used as a paint-
artist, conservators require a detailed knowledge of
ing medium, but others such as gum tragacanth (a
the materials used. The microscopic samples charac-
medium for painting on linen) and cherry gum
teristic of work in this area are notoriously problem-
(which results in an enamel-like effect when mixed
atic to deal with and the sensitivity of the analytical
with egg or casein emulsions) are used less frequently.
technique is paramount.
There is documentary evidence to suggest that
The question why are some painted works in better
gums have been employed as binding media and
condition than others of a similar age? is an impor-
sizing materials for centuries: gum was used as a
tant one for the conservator and speciRc informat-
replacement for sun-dried oil as early as the 12th
ion regarding the nature of the media used in such
century.
works may offer some insight as to why variations in
Proteinaceous media include gelatine, milk and egg
the ageing characteristics of individual paintings
proteins. Animals and Rsh collagen glues are widely
occur.
used as strong adhesives for wood, binders in the
preparation of grounds, size for canvas, and pigment
Paint Media binders in decorative paints. Casein (a mixture of
related phosphoproteins found in milk products), egg
Artists have traditionally used a diverse range of albumin (glair) and egg yolk (tempera) have tradi-
materials as binding media for their pigments: natural tionally found uses as pigment binders, temporary
2090 III / ART CONSERVATION: USE OF CHROMATOGRAPHY IN
varnishes and sealant over primers or grounds: in formed using hexamethyldisilazane as the silylating
addition, casein provides one of the strongest natural agent. Silylation of the amino group usually requires
adhesives known, much used by joiners and cabinet a stronger donor and silylation with either N-
makers in the past. trimethylsilyldiethylamine or N,O-bis(trimethyl-
The choice of binding medium is determined by silyl)acetamide yields the silylamine-silyl ester.
a number of factors, predominantly the nature of the Volatile butyl ester derivatives of amino acids
pigments being used, coupled with the effect desired found in the protein hydrolysate of binding media
by the artist, plus historical factors like location and from paint layers and priming removed from 16th-
the period of the piece. A variety of materials have and 18th-century easel and wall paintings have been
enjoyed periodic popularity due to artistic trends and used for GC analysis. Derivatization is achieved in
scientiRc progress. two stages: the carboxyl functions are Rrst converted
into butyl esters by the addition of butanol (with
dissolved hydrogen chloride), then the amino groups
Analysis of Paint Media are acetylated with triSuoroacetic anhydride. The
samples, dissolved in ethyl acetate, are separated on
The analyses of oil-based media are well documented,
a temperature-programmed capillary column. Calcu-
but any information on the nature of other paint
lation of the relative peak areas of each amino acid
media has been obtained primarily via microscopic
revealed a distinctive proRle for each of the binding
staining methods or crude solubility tests. Existing
media.
methods of analysis provide the basis for the separ-
The existence of an amino acid proRle was estab-
ation of the general categories of binding media (oil,
lished for each of the protein media types, conRrming
gum and protein) by qualitative means: for example,
their identity by the characteristic amino acid ratios
oil and protein layers can be distinguished by
seen for each. Proteinaceous material from the gesso,
the differential staining of cross-sections, whilst
ground and paint layers of a selection of Italian works
colorimetric spot tests can be used for the identiRca-
was hydrolysed under acid conditions, then deionized
tion of polysaccharide gums.
on a small ion exchange column. The samples were
The applicability of these microanalytical tech-
then successfully methylated (carboxyl function) and
niques in situ can be advantageous, but the results can
acetylated (amino function), yielding the highly vol-
be misleading or inaccurate and for this reason these
atile N-acetylmethyl esters of the amino acids, which
techniques are best used in conjunction with other
were separated on a packed column. Results are
analytical methods.
shown in Table 1.
Such simple qualitative techniques, including paper
Loss of analytes is a problem associated with acid
and thin-layer chromatography (TLC), are adequate
hydrolysis. The acid hydrolysis of proteinaceous sam-
where merely the general category of binding medium
ples in the presence of carbohydrates may lead to the
is required, or where the material contains unique
elimination of amino acids as humins, which cause
constituents (e.g. hydroxyproline in gelatine). How-
darkening of the hydrolysate and the formation of
ever, only quantitative chromatographic techniques
insoluble matter. A major contributory factor in the
will enable differentiation between the similar bind-
production of humins (a mixture of coloured com-
ing media where they have no distinctive composi-
pounds which resemble natural melanins) is the
tion, such as in egg and milk proteins.
presence of tryptophan and amino sugars (e.g.
galactosamine) or carbohydrates in the sample, which
degrade during acid hydrolysis.
Gas Chromatography GC has been used to quantify the fatty acid content
Conservation scientists have routinely used gas of eggs. The use of a tempera medium can be con-
chromatographic (GC) methods for the analysis of Rrmed by the absence of azelate in the presence of
proteinaceous, oil and gum media for many years. both palmitate and stearate nondrying oils (i.e., oils
The technique is highly sensitive } an important fac- which thicken but do not dry to a skin). Samples
tor in the analysis of small samples } and a selection removed from aged paint Rlms were saponiRed before
of derivatizing agents is available for use. methylation with diazomethane, then injected dir-
GC analysis of trimethylsilyl derivatives of amino ectly on to a wide-bore column. The presence of the
acids in protein hydrolysate has been widely reported. methyl esters of saturated palmitic and stearic acids,
The carboxyl group of an amino acid is easier to with variable amounts of unsaturated oleic acid, was
silylate than the amino group, which results in the revealed. This method is also applicable for the analy-
formation of two products upon silylation: under sis of oil-based media, the palmitate : stearate ratio
mild conditions, the silyl ester is the major product proving the means of differentiation.
Table 1 Percentage amino acid composition (calculated from peak areas)
Amino acid percentage, peak areas
Laboratory aged samples Samples from paintings
Amino acid Test 1 Test 2 Test 3 Test 4 Sample 1 Sample 2 Sample 3 Sample 4
Alanine 3.9 10.2 5.1 10.0 2.3 10.9 7.5 14.4

Valine 4.4 4.6 6.9 2.4 19.0 1.5 4.8 1.3
Glycine 1.5 4.1 7.4 19.4 3.3 17.5 3.3 27.9
Isoleucine 2.3 3.1 5.2 1.1 4.0 0.6 2.6 0.4
Threonine 4.1 4.9 5.4 2.2 0.1 1.4 4.2 1.4
Leucine 16.4 15.5 11.8 4.0 10.6 3.9 13.0 4.7
Serine 5.8 9.8 7.0 2.9 8.1 2.6 7.9 3.6
Proline 21.0 7.7 5.4 19.6 14.4 20.6 6.3 8.3
Aspartic acid 8.0 15.6 11.1 5.1 7.4 4.6 11.6 7.0
Hydroxyproline 14.2 0.2* 15.7 13.7
Methionine 0.4 0.1 1.1 0.3 0.0 0.2 0.2
Glutamic acid 14.9 11.3 13.2 8.6 18.1 11.6 18.1 13.1
Phenylalanine 5.4 8.0 9.1 3.9 11.8 4.1 13.6 2.8
Lysine 11.2 4.4 10.7 5.6 0.4 4.2 6.1 0.7
Test 1, casein ground; test 2, glair/chalk mix; test 3, egg yolk/chalk mix; test 4, rabbit skin glue/chalk mix; sample 1, upper red layer from
the fac7 ade of San Petronio, Bologna; sample 2, gesso ground from Bellini’s The Madonna of the Meadow ; sample 3, unpigmented
priming from Bellini’s The Madonna of the Meadow ; sample 4, ground layer from Baccafumi’s Tanaquil. Reproduced from White
(1984) with permission.
A mixed medium such as tempera contains both procedure was also employed for the analysis of the
fatty acids and amino acids and GC has been em- paint medium itself and disclosed the use of a mixture
ployed to analyse both components simultaneously. of gum tragacanth and honey.
Samples of mixed proteinaceous and oil media were Samples of media taken from Asian wall paintings
hydrolysed under acid conditions, and derivatized to were also found to contain polysaccharide material
yield the volatile N-(O, S)-ethoxycarbonyl ethyl es- when analysed by GC and TLC. Gum sample hy-
ters, which were then separated by capillary GC. This drolysates were acetylated prior to analysis and char-
method has also been used for the analysis of amino acterization of the media, but the actual method of
acids alone. sample preparation was lengthy and tedious, requir-
Volatile ethyl chloroformate derivatives of amino ing 12 h for hydrolysis and a further 5 h for derivatiz-
acids in the hydrolysate of samples of proteinaceous ation.
media have been analysed to study the effects of Results obtained for the GC analysis of samples
pigments and ageing on the actual analysis/character- of gum from trees of the Acacia genus growing in
ization of proteins from art objects, the results being the vicinity of the Tomb of Nefertari revealed
interpreted via statistical methods. that they were lacking in rhamnose, which usually
GC is the method of choice for the analysis of indicates gum tragacanth, whilst the remainder of
natural gum media from works of art, though to date the sugar content matched that of gum arabic from
there has been relatively little work published in this other sources. When samples of media from paint-
area. An obvious problem associated with the use of ings in the tomb were analysed by GC, the same
many analytical techniques for this type of analysis is sugar proRle was observed: it was therefore con-
the insufRcient sensitivity of the method for the cluded that the paint medium was in fact gum
microscopic samples available to the conservation arabic.
scientist. However, progress has been made in the Twilley’s comprehensive analytical studies of
analysis of gum media by GC, often in conjunction natural gums and their artistic applications em-
with TLC. ployed a variety of techniques, including the GC
Trimethylsilyl derivatives of sugars from the hydro- analysis of trimethylsilyl sugar derivatives. In
lysed samples of surface coating taken from a wooden addition, GC was used for the analysis of reference
Egyptian sarcophagus (dating from the 21st dynasty) samples of aged ink, thus enabling the characteriza-
were analysed using a combination of GC and TLC, tion of ink samples taken from a number of ancient
revealing the presence of gum tragacanth. The same manuscripts.
GC with a mass selective detector following standard gum media samples, whilst Figure 2 shows
a simple ‘one-pot’ hydrolysis and derivatization pro- the chromatograms obtained for samples of priming
cedure was used for the characterization of a number and paint media removed from two works by Blake,
of suspected gum media samples taken from the paint Spiritual Form of Nelson Guiding Leviathan
and ground layers of tempera works by William (1805}9) and Body of Christ Borne to the Tomb
Blake. The results frequently indicated the presence of (1799}1800).
mixed gum media (typically comprising gums arabic, GC has also facilitated the analysis of coatings,
tragacanth and karaya) with the addition of cane such as resins, waxes, lacquers and varnishes re-
sugar. Figure 1 shows the chromatograms of four moved from works of art. Coatings are often applied
Figure 1 Chromatograms of four standard gum media samples: (A) gum arabic; (B) gum tragacanth; (C) karaya gum; (D) cherry
gum.
Figure 2 Chromatograms of samples removed from two works by William Blake: (A) white priming sample from Spiritual Form of
Nelson Guiding Leviathan (1805}9); (B) green paint sample from Body of Christ Borne to the Tomb (1799}1800).
in an attempt to protect them from weathering and Phenylthiocarbamyl derivatives of amino acids in
contamination and, though not classed as media the hydrolysate of proteinaceous media samples have
themselves, their characterization may provide im- been separated on a C18 column using a ternary sol-
portant evidence when determining the reasons for vent system as the mobile phase (water}acetonit-
the ageing/degradation of works of art. rile}acetate buffer).
Varnish samples are commonly saponiRed prior to Following hydrolysis of proteinaceous material re-
methylation, then the components separated on a cap- moved from a series of Italian 15th-century painted
illary column with a linear temperature programme. panels, Halpine used phenyl isothiocyanate (PITC)
Using mass spectrometry as a detection technique, for the derivatization of the amino acids, which were
two major peak groups can be seen, corresponding to then separated on a C18 column using a binary solvent
the diterpenoid and triterpenoid components. system of acetonitrile and acetate buffer (Table 2).
Waxes are stable materials comprising hydrocarbons The addition of nor-leucine as an internal standard
and esters and, because of the virtual absence of polar facilitated the quantiRcation of the amino acid com-
groups, they can be analysed directly by high temper- ponents in the proteins, which in turn resulted in the
ature capillary GC without the need for derivatization. characterization of a number of animal glue and
egg/glue media. However, PITC-amino acid deriva-
Reversed-phase High Performance tives degrade in solution, so must be stored at low
temperature prior to analysis.
Liquid Chromatography PITC derivatives have also been analysed by RP
The speed and sensitivity of reversed-phase high per- HPLC when attempting to identify media samples
formance liquid chromatography (RP-HPLC) has led which had been removed from a variety of French and
to signiRcant developments in the analysis of pro- Italian stone and wooden sculptures, frescoes and
teins: RP-HPLC is now one of the most widely used statues. Proteinaceous material was extracted from
techniques for the analysis of amino acids, since pre- the matrices with sodium hydroxide and the sub-
column derivatization is possible with a selection of sequent analysis indicated the presence of gelatine
derivatizing agents and a variety of detection tech- and egg proteins.
niques can be employed. RP-HPLC lends itself well to 9-Fluorenylmethyl chloroformate (FMOC) is a use-
conservation science, being particularly suitable for ful derivatizing agent for amino acids since it favours
the analysis of the extremely small samples removed mild, aqueous conditions, reacts with both primary
from works of art. and secondary amino acids and is stable at room
Table 2 Percentage amino acid composition
Percentage amino acid composition
Control samples Samples from panels
Amino acid Control 1 Control 2 Control 3 Control 4 Sample 1 Sample 2 Sample 3 Sample 4
Aspartic acid 1.0 9.1 5.6 7.9 5.5 * 10.2 5.2

Glutamic acid 2.3 11.1 9.3 13.0 5.9 6.4 9.8 4.1
Hydroxyproline 12.3 0.6 0.3 * 11.7 9.9 3.8 5.9
Serine 3.8 11.1 10.7 9.4 5.7 5.7 8.4 8.9
Glycine 27.7 10.9 7.8 5.2 26.9 24.7 15.8 18.6
Histamine 0.8 1.3 1.5 1.6 * * * 1.2
Arginine 5.7 5.3 5.9 4.0 4.1 * 4.3 3.0
Threonine 2.6 5.6 6.6 3.9 3.9 * 4.8 6.6
Alanine 11.1 8.6 10.2 8.8 11.8 11.2 10.0 12.8
Proline 16.9 5.4 5.9 4.0 11.6 12.4 6.7 10.3
Tyrosine 1.2 2.7 3.6 2.4 * * 1.1 1.8
Valine 3.3 6.6 7.8 7.5 4.2 5.3 5.5 7.2
Methionine 1.3 1.7 2.0 4.8 * 6.7 1.3 0.3
Isoleucine 1.9 4.3 5.0 5.8 3.6 11.7 4.7 4.6
Leucine 3.9 8.5 10.0 9.0 1.5 6.0 7.2 7.1
Phenylalanine 2.0 3.3 3.7 5.9 3.7 * 3.3 2.4
Lysine 2.0 4.0 4.5 5.7 * * 3.0 *
*Too small to quantify; control 1, rabbit skin glue ground with blue pigment; control 2, egg yolk with red pigment; control 3, egg yolk with
blue pigment; control 4, egg albumin; sample 1, light blue paint; sample 2, dark blue paint; sample 3, light brown paint; sample 4, green
paint. All paint samples removed from Cosimo Tura’s The Annunciation with Saint Francis and Saint Louis of Toulouse
(c. 1475). Reproduced from Halpine (1992) with permission.
temperature for up to 2 weeks. The FMOC moiety is sample size of paint is relatively large when put into
both highly Suorescent and a good UV chromophore, the context of a specimen to be removed from a valu-
so absorption and emission detection techniques can able work of art. Furthermore, museums and galleries
be used. Detection methods are an important concern are notoriously short of both money and space, thus
in conservation science in view of the very small speciRc single-purpose equipment is deemed unaf-
samples available } FMOC is particularly useful since fordable by many institutions.
Suorescence is usually far more sensitive than UV
Thin-layer Chromatography
absorption. Standard proteinaceous media and mu-
seum sample hydrolysates have been characterized as TLC has been used on many occasions, particularly in
FMOC derivatives. the analysis of natural gum media } it is often used in
conjunction with GC for such analyses but can pro-
vide useful information when used alone.
Other Chromatographic Techniques TLC has been used to characterize gum media
taken from a 16th-century manuscript: hydrolysed
Ion Exchange Chromatography
gum samples were separated on silica plates, facilitat-
Ion exchange chromatography was Rrst used for the ing the subsequent identiRcation of gum arabic.
analysis of samples from works of art in 1969 when Samples of binding media from paint layers were
the successful analysis of antique and modern art removed from three ancient Egyptian epitaphal stelae
specimens was reported. The eluted amino acids were on wooden supports, then TLC was used to investi-
detected by optical density with ninhydrin, then the gate the nature of the media, revealing the presence of
percentage amino acid composition was calculated gum tragacanth.
for each sample. Samples of gelatine, casein, glair,
Pyrolysis^Gas Chromatography
tempera and even animal horn were characterized
using this method. Pyrolysis}gas chromatography (Py-GC) has been em-
The use of ion exchange chromatography in this ployed in the analysis of natural gum media from
particular area is problematic: the method of sample works of art. The distinctive pyrograms obtained for
preparation is both lengthy and complex, pH gradi- a series of standard gum samples enabled their identi-
ents are difRcult to control precisely and the required Rcation and it was discovered that by pyrolysing the
complex gum-pigment mixed samples at 4003C, dif- Thin-Layer (Planar) Chromatography. Paints and Coat-
ferences in peak patterns between standard samples ings: Pyrolysis: Gas Chromatography. Pigments: Liquid
and mixed samples were minimized. Sample identi- Chromatography; Thin-Layer (Planar) Chromatography.
Rcation is aided by the use of computational methods Polysaccharides: Liquid Chromatography.
of pattern recognition.
Further Reading
Conclusions Birstein VJ (1975) On the technology of Central Asian wall
paintings: the problem of binding media. Studies in
When preparing to analyse a sample removed from Conservation 20: 8.
a work of art, conservation scientists must select Derrick MR and Stulik DC (1990) IdentiRcation of natural
a technique which gives the maximum amount of gums in works of art using pyrolysis-gas chromatogra-
information for the minimum amount of sample and phy. ICOM Committee for Conservation, 9th Triennial
sample preparation: it appears that RP-HPLC using Meeting, Dresden, 26}31 August 1990: Preprints (ed.
FMOC as the amino acid derivatizing agent is the Grimstad K), p. 9.
optimum analytical technique for the characteriza- Erhardt D, Hopwood W, Baler M and von Endt D (1988)
tion of proteinaceous binding media, whilst GC is A systematic approach to the instrumental analysis of
natural Rnishes and binding media. Preprints of Papers
routinely used for oil-based media.
presented at the 6th Annual Meeting. American Institute
At present, GC analysis of silylated sugar residues for Conservation of Historic and Artistic Works. Wash-
is arguably the best method for the identiRcation of ington, DC: pp. 67}84.
natural gum media, and the use of mass spectrometry Flieder F (1968) Mise au points des techniques d’identiRca-
as the detection technique offers superior sensitivity tion des pigments et des liants inclus dans la couche
and Sexibility for these complex samples. However, picturale des eluminures de manuscrits. Studies in Con-
this seems to be the least investigated area of analysis servation 13: 49.
and signiRcant developments in methodology which Grzywacz CM (1994) IdentiRcation of proteinaceous bind-
will further improve the sensitivity of the technique ing media in paintings by amino acid analysis using
9-Suorenylmethyl chloroformate derivatization and re-
can be anticipated: this is of particular importance in
versed-phase high-performance liquid chromatography.
the analysis of gum media since samples invariably Journal of Chromatography A 676: 177.
contain no more than 10% binding medium, result- Halpine SM (1992) Amino acid analysis of proteinaceous
ing in minute amounts of actual analyte in the sam- media from Cosimo Tura’s The Annunciation with Saint
ples. It is possible that microbore HPLC techniques Francis and Saint Louis of Toulouse. Studies in Conser-
may Rnd a use in conservation science, since they are vation 37: 22.
obviously suited to the minuscule samples routinely Keck S and Peters T (1969) IdentiRcation of protein-con-
provided for analysis. taining paint media by quantitative amino acid analysis.
Studies in Conservation 14: 75.
Simple qualitative techniques such as TLC may be
Kenndler E, Schmidt-Beiwl K, Mairinger F and PoK hm M
sufRcient to indicate the basic media type used in (1992) IdentiRcation of proteinaceous binding media of
works of art, but as more and more works require easel paintings by gas chromatography of the amino acid
some form of conservation or restoration treatment it derivatives after catalytic hydrolysis by a protonated
is becoming increasingly important that the conserva- cation exchanger. Fresenius Journal of Analytical Chem-
tor has as much information as possible relating to istry 342: 135.
the nature of the materials used in the work, in order Masschelein-Kleiner L, Herlan J and Tricot-Marckx F (1968)
to avoid damaging irreplaceable objects of artistic Contribution à l’analyses des liants, adheH sifs et vernis
importance. anciens. Studies in Conservation 13: 105.
Mills JS and White R (1994) The Organic Chemistry of Mu-
Chromatographic techniques provide reliable and
seum Objects, 2nd edn. London: Butterworth Heinemann.
accurate methods of analysis, suitable for use with the Nowik H (1995) Acides amines et acidgras sur un meH me
microscopic samples typically seen in this Reld of chromatogramme un autre regard sur l’analyse de liants
work. Further work should lead to simpliRcation of en peinture. Studies in Conservation 40: 120.
methods of sample preparation } any improvements Twilley JW (1984) The analysis of exudate plant gums in
which mean that the size of samples required for analy- their artistic applications: an interim report. American
sis is reduced and that analyte losses are minimized Chemistry Society Advances in Chemistry Series 205: 357.
would be welcomed by the conservation community. Vallance SL, Singer BW, Hitchen SM and Townsend JH
(1998) The development and initial application of a gas
See Colour Plates 59, 60. chromatographic method for the characterization of
gum media. Journal of the American Institute for Con-
See also: II/Chromatography: Gas: Derivatization; servation 37(3): 294.
Chromatography: Liquid: Derivatization. III/Amino White R (1984) The characterization of proteinaceous bind-
Acids: Gas Chromatography; Liquid Chromatography; ers in art objects. National Gallery Technical Bulletin 8: 5.
2096 III / ATMOSPHERIC ANALYSIS: GAS CHROMATOGRAPHY
ATMOSPHERIC ANALYSIS:
GAS CHROMATOGRAPHY
A. C. Lewis, University of Leeds, Leeds, UK challenging. A number of chromatograms obtained
from atmospheric analysis are also presented in
Figures 1+4. Figures 1 and 2 were obtained from an
urban environment and Figures 3 and 4 came from
Introduction a single Reld campaign held on the west coast of
Ireland to study clean marine air.
The determination of both naturally produced and
anthropogenic organic species in the atmosphere is
a key area in atmospheric chemistry research. Nearest
to the earth’s surface, within the boundary layer of
Sample Acquisition
the troposphere, the determination of hydrocarbon The initial step of sample acquisition is a far from
species is important because of their ability to react trivial task in atmospheric measurements, and segre-
rapidly with NOx in the presence of sunlight to form gates a major grouping of techniques into either in
photochemical smog. At the other extreme of the situ or post-acquisition analysis. The ability to store
atmosphere in the stratosphere, remote measure- an atmospheric sample is critical to the decision on
ments of very long-lived halogenated species are whether analysis may be performed back in the labor-
required because of their critical impact on ozone atory or on-site immediately following acquisition.
destruction. Species such as methane, present Stable species such as methane, carbon monoxide and
throughout the atmosphere, have importance as chlorinated Suorocarbons may be stored successfully
greenhouse gases and inSuence global climate and in sample vessels for many weeks without affecting
temperature change. sample integrity, and widespread measurements of
The wide range of both short- and long-lived spe- these species have therefore been performed. Atmo-
cies that are of interest in atmospheric science spheric degradation products, such as peroxyacetyl
coupled to extremely low concentrations and a re- nitrate (PAN, CH3C(O)2ONO2), are so unstable,
quirement often for in situ automated analysis has led however, that analysis must be performed immediate-
to the development of many novel chromatographic ly. For many other important reactive species such as
techniques and methodologies. While some atmos- alkenes, monoterpenes and dimethyl sulphide, the
pheric species such as organic acids, peroxides and storage of samples has been shown to lead to a
aldehydes have been determined using high perfor- degree of analyte losses due to reaction with
mance liquid chromatography, the majority of species co-sampled pollutants such as ozone and oxides of
are analysed using capillary gas chromatography. The nitrogen. To overcome these problems of reaction
range of compounds that are of interest has resulted during storage, many forms of scrubber (e.g. potass-
in almost every kind of detector Rnding a role within ium iodide and glycerol to remove ozone) have been
atmospheric analysis. tested to remove such species without affecting
This article only deals with the analysis of species sample integrity.
found in the gas phase, although many species present Much atmospheric sampling is performed using
in the atmosphere are bound to particles or are in canister methods, Rlled either by vacuum release or
aerosol form. A vast number of methodologies for by pumping sample into the canister using stainless
analysis of these species exist, although in common steel or TeSon diaphragm pumps. For any sample to
with gas phase species, gas chromatography plays be stored successfully there must be minimal interac-
a vital role in their analysis. tion with the canister walls and coating methods such
The range of techniques that are in use is so broad as electropolishing, silica or TeSon coating are cur-
that a complete review of analytical methods is im- rently in use. The preparation and cleaning of canis-
practical. Many individual methods, however, have ters require careful attention and high vacuums are
common components or key procedural steps and often applied together with elevated temperatures for
these will be discussed. A general outline of a typical periods of hours or even days. A similar method to
atmospheric determination can be broken down into canister samples is the use of collapsible TeSon or
the following steps: sample acquisition, preparation, Tedlar sample bags. These bags may be Rlled by
separation and detection. The Rrst two stages, ac- pump and returned to a laboratory for analysis using
quisition and preparation, often prove to be the most very similar procedures to those used with canister
III / ATMOSPHERIC ANALYSIS: GAS CHROMATOGRAPHY 2097
A huge range of adsorbents is currently available

from high surface area ('1000 m3 g\1) carbon-
based materials with strong retention characteristics
to lower surface area ((50 m3 g\1) polymers such as
Tenax TA. While being relatively low cost compared
to sample canisters, care is often required in the
cleaning and preparation of sample tubes. Samples
may be introduced to the adsorbent tubes either
dynamically over short periods of time (typically
minutes) or via diffusional sampling over longer peri-
ods (typically several days). Carbon-based adsorbents
are suitable for a wide range of species ranging
from volatile hydrocarbons and CFCs to organic
nitrates. Polymeric materials are used mainly for the
concentration of low volatility species such as aro-
matics and monoterpenes in air, although compounds
as large as two and three ring polycyclic aromatics
may also be successfully trapped and desorbed from
the gas phase.
Sulfur species trapping is often performed via
Figure 1 Low molecular weight hydrocarbons including meth- chemisorption on to gold wool traps that provide
ane determined using an online activated charcoal adsorbent trap
a stable matrix for the sample to be stored for reason-
in a programmed temperature vaporization injector. Column
50 m, 0.53 mm i.d. Al2O3 /NaSO4 PLOT (Chrompack) 10 m film able periods of time.
thickness. Desorption temperature was at 163C s\1 from !203C The complete retention of all target analytes on the
to 4003C, and column temperature programmed from 453C to adsorbent bed must be carefully evaluated and calcu-
2003C. Chromatograms showing (a) the separation of C1}C6 com- lation of retention volume (often referred to as break-
ponents of Leeds city-centre air, and (b) a blank following desorp-
through volume) is essential. For sampling very
tion. Peaks: 1"methane, 2"ethane, 3"ethene, 4"propane,
5"propene, 6"2-methylpropane, 7"ethyne, 8"n -butane, volatile species (for example ethane or ethene) the
9"trans-2-butene, 10"1-butene, 11"isobutene, 12"cis -2- retention volume for an adsorbent system is often
butene, 13"2-methylbutane, 14"n -pentane, 15"1,3-bu- the critical factor in determining system sensitivity.
tadiene, 16"pentenes, 17"2-methylpentane, 18"3-methyl- Trapping at sub-ambient temperatures is commonly
pentane, 19"n -hexane, 20"methyl hexanes and hexenes,
employed to increase the maximum sample volume,
21"heptane, 22"methylcyclopentane, 23"benzene, 24"
toluene. Reproduced with permission from Lewis AC and Bartle often using liquid nitrogen or carbon dioxide.
KD 1996. A simplified method for the determination of atmospheric For operation of any instrument in a Reld location,
hydrocarbons. LC}GC International 9: 297}304. the supply of these coolants may be a problem and
modern instruments often now employ Peltier
coolers.
samples. While generally of lower unit cost, the bag
method often produces a greater degree of sample
artefact and analyte losses. Sample Preparation and Injection
Some low volatility species are less suitable for
Removal of Water
sampling using canister methods because of the in-
creased possibility of analyte condensation on to the The inevitable presence of water in atmospheric sam-
walls of the container. For this reason a second ples and its removal prior to sample analysis is a com-
method of sample acquisition is widely used based on plex area. In certain circumstances its presence may
use of a solid-phase adsorbent as an analyte trap. The be both beneRcial, e.g. with canister samples where it
adsorbent used in the trap may be chosen to introduce occupies the active sites on the canister walls, or
an element of selectivity to the trapping mechanism, detrimental, e.g. where it affects analysis in some
although in practice a trap-all approach is commonly way. Alumina porous layer open tubular (PLOT)
used. Adsorbent traps allow for instrument automa- columns are particularly affected by moisture in
tion that is not always possible with canister methods the sample, and large changes in stationary phase
and such traps now form the basis of many national afRnity occur when water is introduced. Detectors
monitoring programmes in the urban environment, as such as mass spectrometers are also extremely
well as automated instruments for research in very sensitive to water introduced with the sample and
clean air. high background noise may result. Even the Same
Figure 2 Aromatic hydrocarbon species determined using an online Tenax TA adsorbent trap in a programmed temperature
vaporization injector. Column 60 m, 0.53 mm i.d., poly(dimethylsiloxane) 3 m film thickness. (Restek RTX-1). Desorption temperature
was at 163C s\1 from 03C to 2203C, and column temperature programmed from 353C to 2403C.
1. unresolved volatile material 13. toluene 25. p -ethyltoluene
2. hexane 14. 2- and 4-methylheptane 26. 1,3,5-trimethylbenzene
3. methyl cyclopentane 15. 3-methylheptane 27. o -ethyltoluene
4. 2,4-dimethylpentane 16. octane 28. 1,2,4-trimethylbenzene
5. benzene 17. ethylbenzene 29. decane
6. 2-methylhexane 18. m - and p -xylene 30. 1,2,3-trimethylbenzene
7. 3-methylhexane 19. styrene 31. indane
8. 2,2,4-trimethylpentane 20. o -xylene 32. 1,4-dimethyl 2-ethylbenzene
9. heptane 21. nonane 33. dimethylethylbenzenes and undecane
10. methyl cyclohexane 22. isopropylbenzene 34. 1,2,3,5-tetramethylbenzene
11. 2,4- and 2,5-dimethylhexane 23. propylbenzene 35. naphthalene
12. 2,2,3-trimethylpentane 24. m -ethyltoluene 36. dodecane
(Reproduced with permission from Lewis AC et al. (1996) Atmospheric monitoring of volatile organic compounds using programmed
temperature vaporisation injection. Journal of High Resolution Chromatography 19: 686}690.)
ionization detector can be affected by injection of tion temperatures. Inorganic adsorbents are also
water, and the Same may be extinguished when water commonly used, notably potassium carbonate and
elutes from the column. magnesium perchlorate. Adsorbents such as these,
Many forms of selective water removal exist, the however, have limited capacity and often require fre-
simplest of which is the use of condensation traps or quent regeneration or replacement. A combination of
stripping coils. Losses of low boiling molecular spe- initial condensation and second stage adsorbent
cies are insigniRcant although condensation of higher scrubber often provides sufRcient capacity to dry
boiling organic material may arise at low condensa- a sample stream of air for many hours or days. Con-
often then performed to remove the remaining com-

ponents from the analytical column. A further
example of direct loop injection is in the analysis of
PAN where, due to thermal decomposition of the
sample, analysis must be performed in situ. It is per-
formed using direct loop injection to a cooled isother-
mal pre-column followed by a short analytical capil-
lary column coupled to an electron capture detector.
An ancillary measurement of carbon tetrachloride is
often gained from using this approach.
Sample concentration methods Whether an atmos-

pheric sample is collected using adsorbent or canister
techniques, several stages of sample preparation are
required prior to injection into the analytical column.
The sheer size of sample } often many litres } pre-
cludes any form of direct injection except in the case
of the high concentration species described earlier,
making sample concentration a vital step prior to
injection.
Canister samples require complete concentration
prior to introduction to the capillary column.
Figure 3 CFCs and halon species in clean marine air deter-

mined using an online Carbosieve micro adsorbent trap and direct
injection to capillary GC. Detection by dual ECD/oxygen-doped
ECD. Column 60 m, 0.33 mm i.d., 1 m film DB-1 (J&W). (A)
Stage 1: standard ECD; (B) Stage 2: oxygen-doped ECD. (Repro-
duced with permission from Bassford M, and Simmonds PG,
University of Bristol, Cantocks Place, Bristol.)
tinuous drying may be achieved using permeation

membranes such as NaRon. These forms of driers
operate by means of a concentration gradient across
a membrane that is permeable only to highly polar
material such as water. A counter-current of dry gas
is passed around the outside of the membrane as
a sheath gas and carries moisture away to waste.
However, this type of drier is unsuitable for
samples where the quantitation of polar materials is
required.
Direct loop injection High concentration atmos-

pheric species require the least amount of sample Figure 4 GC}SIM/MS of natural halocarbons (CH3I, CH2ClI,
preparation, often only the removal of excess water CH2BrI and CH2I2) in marine air. PLOT column trap cooled with
liquid nitrogen. Flash heating to capillary GC. Column 60 m,
from the sample. Methane, carbon monoxide and
0.33 mm i.d., 1.8 m film DB-VRX (J&W). Detection by single ion
N2O analyses are performed simply by Rlling monitoring HP mass selective detector. (Reproduced with per-
a known volume injection loop, followed by direct mission from Carpenter, L and Sturges, WT, School of Environ-
injection into the analytical column. BackSushing is mental Sciences, University of East Anglia, Norwich, UK.)
Analytes are removed from the canister via either With the introduction of Al2O3 PLOT columns, fast
internal canister pressure (for pumped-in samples) or high resolution analysis of even the highest volatility
vacuum pump (for atmospheric pressure samples) species has been made possible and many applica-
over a cryogenic trap. Liquid nitrogen is most com- tions that previously used packed column GC are
monly used but liquid argon has been used to reduce now being performed using capillary columns.
the amount of oxygen retained in the focusing stage. Where only one species is to be determined simple
The concentration zone may consist of a packed tube isothermal separations may be used, often in conjunc-
containing either an absorbent such as Tenax TA or tion with a column backSush step. The analysis of
glass beads or it may simply be empty stainless steel PAN is an example of this where a short backSushing
tubing. Since the concentration stage is at such low pre-column is used prior to a 10 m analytical column.
temperatures, the majority of water vapour in the A simple two-dimensional separation has also been
sample must be removed prior to focusing in order to proposed for PAN using heart-cutting.
stop blockage of lines with ice. Once a sufRcient The wide magnitude of analyte volatilities encoun-
volume has been collected on the focusing trap, the tered imposes limits on the range of species that can
trap is generally Sash heated either electrically or by be successfully separated on a given column. PLOT
using hot water. This results in a very sharp band of columns are used widely for the analysis of C1}C7
compounds being introduced to the head of the ana- hydrocarbons and some high volatility halogenated
lytical column. compounds. The retention characteristics of this type
With adsorbent tube analysis the collected analytes of column are not favourable for the analysis of
are generally thermally desorbed either directly on to oxygenated compounds, and water often plays a sig-
the analytical column (in the case of programmed niRcant role in degrading the quality of any PLOT
temperature vaporization injection), or on to a focus- column separation. Very strong retention of higher
ing cold trap. The desorption temperature is generally boiling species leads to extensive peak broadening
deRned by the maximum temperature that the adsor- coupled with lengthy analysis times.
bing material can support. For polymeric adsorbents Analysis of higher molecular weight species, in-
this may be relatively low ((2503C) while carbon- cluding volatile organic compounds (VOCs), CFCs
based materials may support desorption at temper- and hydrogen-containing chloroSuorocarbon re-
atures of 5003C or higher. placements (hCFCs), aromatic and monoterpene
If a programmed temperature vaporization (PTV) compounds, has commonly been performed with
injector is used, the desorption may be sufRciently nonpolar (methylpolysiloxane) or slightly polar (5%
rapid ('153C s\1) so that initial column focusing is phenyl methylpolysiloxane) 1}5 m-thick Rlm capil-
sufRcient to obtain well-resolved peaks. If the desorp- lary columns, 0.32 mm i.d. by 50 m long. Wide bore,
tion from an adsorbent tube is relatively slow then 0.53 mm i.d. columns are often used where desorp-
a focusing step is used, with a similar concentration tion is direct from a preconcentration trap to the
mechanism to that used with canister samples. Once analytical column. For the analysis of complex hCFC
again, water must be removed from the sample since mixtures in the atmosphere, columns as long as
it may affect the column or the detection system. 100 m have been reported.
While the majority of species are thermally desor- To improve retention of volatile VOCs on thick
bed from the trap to the analytical column, a few Rlm siloxane columns, without use of sub-ambient
types of compounds require solvent extraction prior cooling, operation with columns containing Rlms up
to syringe injection. Organic nitrates fall into this to 15 m thick have been reported. Band broadening
class along with higher molecular weight polycyclic effects through stationary phase diffusion become
aromatic compounds that may suffer from incom- signiRcant with Rlms of this thickness and this ap-
plete or slow thermal release. proach has not been widely adopted. The highest
molecular weight gas phase species such as naphtha-
lene, Suorene and anthracene may be separated
Separation of Atmospheric Samples efRciently only on thinner Rlm non-polar columns
While a number of speciRc applications utilize packed with Rlm thicknesses of typically 0.25}0.5 m.
columns (notably in the analysis of methane, CO and Organic nitrate species in the atmosphere may also
N2O), current methods for the separation of atmos- be determined using capillary GC with either char-
pheric components are performed almost exclusively coal adsorbent traps, extracted with aromatic organic
using capillary column gas chromatography. solvent or via direct cryofocusing from a canister
The few applications remaining where packed col- sample. Lengthy analysis times can result due to the
umns are in use, generally employ molecular sieve necessity of using combinations of columns to achieve
packings for the separation of permanent gas species. full separation of target analytes although commonly
moderate polarity 50% phenyl/50% methyl poly- be used over many years. Some halogen-containing
siloxane columns are used. species of interest (e.g. CH3Cl, CHF2Cl, CH2Cl2) have
a relatively poor ECD response and the use of the
oxygen-doped ECD to enhance their response has
Detection Systems been successful and is demonstrated in Figure 3. The
Even the highest resolution capillary columns often determination of some nitrogen-containing species is
have insufRcient peak capacity to resolve all compo- also performed using ECD, notably in the areas of
nents in a typical atmospheric sample. Since selectiv- organic nitrate analysis and the determination of
ity in trapping is not always possible, selectivity PAN. Organic nitrate analysis using ECD is often
in detection is a useful tool in the simpliRcation of complicated by the co-elution of halogenated com-
atmospheric analysis. pounds, so often a nitrogen-speciRc detector such as
The Same ionization detector (FID) is by far the the chemiluminescence detector is used in parallel.
most commonly used detector in atmospheric analy- Detection of CO is generally performed using the
sis by GC since it offers high sensitivity, extremely reduction gas detector (RGD) where hot HgO reduc-
wide linearity and good reliability. Using well- tion with one CO molecule releases one Hg molecule
cleaned fuel gases coupled to low noise electrometer from a catalytic bed. The Hg molecule released is then
circuitry it is possible to determine amounts down to detected by UV absorption.
1 pg s\1. With a typical sample volume of 1 L, detec- The analysis of sulfur compounds in the atmos-
tion limits for individual species may be in the low phere, in particular dimethyl sulRde (DMS), has often
parts per trillion range (10\12). Calibration can be been performed using a combination of GC with
performed with relative ease but the complexity of sulfur selective detection to overcome problems of
samples can make peak identiRcation difRcult. insufRcient chromatographic resolution. Flame
Methods for alkene/aromatic analysis utilizing selec- photometric detection (FPD) has been used extensive-
tive response from the photoionization detector (PID) ly in the past although signal quenching by co-eluting
have been proposed although they are not in wide- hydrocarbons results in drastically reduced sensitiv-
spread use. ity. The Hall detector or electrolytic conductivity
Mass spectrometry offers obvious solutions to detector (ELCD) has also been used for atmospheric
problems of compound identiRcation but the opera- determinations although it requires regular mainten-
tion of currently available bench-top MS in full scan ance making it unattractive for an automated instru-
mode often has insufRcient sensitivity for trace level ment. Emerging methods are now taking advantage
measurements. Spectral information obtained from of signiRcant advances in bench-top atomic emission
GC/MS of atmospheric hydrocarbons often leads to detectors (AED). The multielemental nature of the
highly similar fragmentation patterns and assists little AED offers signiRcant advantages in atmospheric
in the identiRcation of isomeric species. Similarly, measurements both in terms of sensitivity (sulfur
identiRcation of monoterpene species can only be } 2 pg s\1), and where concurrent emission line
conRrmed through a combination of both spectral measurements for carbon and hydrogen may provide
information and retention time data. information on empirical formulae of unknown spe-
While MS of hydrocarbon species in clean air is cies. The sulfur chemiluminescence detector, which is
often unsuccessful when operating in full scan mode a more recent technique that offers extremely high
there is sufRcient sensitivity and resolution to allow sensitivity and selectivity, may yet Rnd an important
for detection of long-lived CFC species. Long-term role in atmospheric sulfur analysis.
monitoring of these species has been performed by The analysis of oxygenated species in the atmos-
GC}ion trap detection in the worldwide MS}GAGE phere is one of the least studied areas using gas
(global atmospheric gases experiment) network. In chromatography. It is an area of fundamental signiR-
single ion mode, femtogram sensitivities can be cance where species may be present in the atmosphere
achieved and this approach has been used for Reld as direct emissions or as degradation products of
monitoring of naturally produced trace level iodo- oleRns. While most aldehydes and ketones may be
and bromo- compounds. successfully separated using GC, sample preparation
More commonly used for CFC measurements in and analyte detection are the key problem areas.
the atmosphere is the electron-capture detector (ECD) Sample preparation is an area where new adsorbents
that offers high sensitivity to certain species plus high may provide selective concentration of polar species
selectivity over hydrocarbon compounds. GC}ECD simplifying the chromatogram produced. Detection
measurements require careful calibration due to the of oxygenates is a further difRculty due to their often
great variation in response to halogenated species, very low response in both ECD and FID. Use
although their high stability allows gas standards to of elemental speciRc detection such as AED offers
2102 III / BACTERIOPHAGES: SEPARATION OF
some potential in atmospheric oxygenate analysis bined with high resolution gas chromatography}mass
although AED sensitivity for oxygen is only around spectrometry for the analysis of polar and non-polar
100 pg s\1. Detectors such as the helium ionization C4}C14 hydrocarbons involved in photochemical smog
detector (HID), which produce a nonselective high formation. Journal of High Resolution Chromatography
sensitivity response to these types of compounds, may 15: 75}84.
Grimsrud EP (1992) The electron capture detector. In: Hill
in future allow on-line measurements of oxygenates
HH and McMinn DG (eds) Detectors for Capillary
with GC, assuming sufRcient chromatographic res- Chromatography, pp 83}106. New York: John Wiley.
olution or trapping selectivity can be obtained. Helmig D and Greenberg JP (1994) Automated in situ gas
chromatographic}mass spectrometric analysis of ppt
Future Work level volatile organic trace gases using multistage solid-
adsorbent trapping. Journal of Chromatography A 677:
Gas chromatography has an important role to play in 123}132.
monitoring mankind’s emissions into the atmosphere Kruschel BD, Bell RW, Chapman RE, Spencer MJ, and
and exploring the natural balance of biogenically Smith KV (1994) Analysis of ambient polar and non-
released materials. New developments in injection polar volatile organic compounds (VOCs) by thermal
technology and adsorbent materials will allow desorption, high resolution gas chromatography}mass
a greater number of species to be determined auto- spectrometry (TD/HRGC/MS). Journal of High Resolu-
matically in Reld locations. Development in column tion Chromatography 17: 187}190.
Lewis AC, McQuaid JB, Seakins PW, Pilling MJ, Bartle KD
technology to reduce the effect of moisture on
and Ridgeon P (1996) Atmospheric monitoring of vol-
chromatographic separation and broaden the range atile organic compounds using programmed temper-
of volatilities that may be separated on a given col- ature vaporisation injection. Journal of High Resolution
umn will also bring signiRcant beneRts. Improve- Chromatography 19: 686}690.
ments in detector sensitivities and reliability (notably O’Brien JM, Shepson PB, Muthuranu, Hao C, Niki H,
bench-top MS) will determine which of the many Hastie DR, Taylor R and Roussel PB (1995) Measure-
available detectors become standard in the next gen- ment of alkyl and multifunctional organic nitrates at
eration of atmospheric instruments. a rural site in Ontario. Journal of Geophysical Research
100: 22795}22804.
See Colour Plates 56, 57, 58. Penkett SA, Blake NJ, Lightman P, Marsh ARW, Anwyl
P and Butcher G (1993) The seasonal variation of non-
See also: II/Chromatography: Gas: Detectors: General methane hydrocarbons in the free troposphere over the
(Flame Ionization Detectors and Thermal Conductivity De- north Atlantic ocean possible evidence for extensive
tectors); Detectors: Mass Spectrometry; Detectors: Selec- reaction of hydrocarbons with the nitrate radical. Jour-
tive. III/Environmental Applications: Gas Chromatogra- nal of Geophysical Research 98: 2865}2885.
phy-Mass Spectrometry. Simmonds PG, Derwent RG, McCulloch A, O’Doherty
S and Gaudy A (1996) Long-term trends in concentra-
Further Reading tions of halocarbons and radiatively active trace gases in
Atlantic and European air masses monitored at Mace
Cao X-L and Hewitt CN (1993) Passive sampling and gas Head, Ireland 1987}1994. Atmospheric Environment
chromatographic determination of low concentration of 30: 4041}4063.
reactive hydrocarbons in ambient air with reduction gas Swan HB and Ivey JP (1994) Analysis of atmospheric sul-
detector. Journal of Chromatography 648: 191}197. phur gases by capillary gas chromatography with atomic
Ciccioli C, Cecinato A, Brancaleoni A, Frattoni M, and emission detection. Journal of High Resolution
Liberti A (1992) Use of carbon adsorption traps com- Chromatography 17: 814}820.
BACTERIOPHAGES: SEPARATION OF
P. Serwer, The University of Texas Health Science generated by the possibility of using bacteriophages
Center, San Antonio, TX, USA as biological antibiotics. This interest is periodically
revived when difRculties are encountered with the use
of chemical antibiotics. Interest in bacteriophages
was also generated by both the short life cycle and the
Introduction simple (short) genome of bacteriophages. These char-
Bacteriophages are viruses that infect bacteria. acteristics were useful in developing the science of
Historically, interest in bacteriophages was Rrst molecular genetics. Bacteriophages were a favourite
III / BACTERIOPHAGES: SEPARATION OF 2103
for asking questions about transfer of biological in- while in the gel. The cell bursts at the end of the
formation via DNA replication, transcription (copy- infection. Light-scattering (turbidity) decreases when
ing of RNA from DNA) and translation (using the the infected cell bursts. All progeny of a single bac-
information in RNA to assemble proteins). Today, teriophage particle remain in a restricted region of the
bacteriophages are used to carry small pieces of the gel. This region is comparatively clear (nonturbid).
DNA of higher organisms. The purpose is to deter- The clear region is called a plaque. A viable bacterio-
mine the nucleotide sequence of the pieces and, ulti- phage particle is assayed by the formation of a single
mately, the nucleotide sequence of the whole genome plaque on a glass or plastic Petri dish or plate which
of higher organisms. Basic science uses bacterio- holds the gel with plaques. Plaques are counted after
phages as models for understanding how separate incubating viable bacteriophage particles with host
molecules assemble to form a complex structure. The cells. The number of viable bacteriophage particles
questions asked in these studies are fundamental: per plate is small enough (usually 100}2000) so that
How do biological motors work? How is speciRcity overlapping plaques do not cause difRculty in count-
maintained in intermolecular binding? How is accu- ing. The number of plaques is multiplied by a dilution
racy assured during assembly? To what extent is factor, if the bacteriophage particles had been diluted
biological form determined by either pathway of as- before assay. Bacteriophage plaques are illustrated in
sembly or energetics of the Rnal structure? Are errors Figure 1.
corrected during assembly of biological macro- Incubation of a host}bacteriophage mixture in
molecules? Are separate biochemical processes integ- a gel can also be used to prepare large numbers of
rated in the context of the interior of a cell? Pursuit of bacteriophage particles before puriRcation. The bac-
the answers to all these questions requires puriRca- teriophages produced are sometimes called a plate
tion of bacteriophage particles. stock. Preparing of bacteriophage particles via large
PuriRcation of both bacteriophages and bacterio- volume ('100 mL) plate stocks is not efRcient (in
phage-like particles is also needed for other types of time and cost), in comparison to preparing bacteri-
study. Bacteriophage-like particles are present in both ophage particles via infection in liquid culture. How-
lakes and oceans. Some are true bacteriophages. ever, some bacteriophages which do not grow well in
Others are viruses that infect algae, rather than bac- liquid cultures grow comparatively well when pre-
teria. Environmental biologists detect and sometimes pared as a plate stock. Thus, plate stocks are some-
purify these particles without knowing what they are. times used for the growth of bacteriophage particles
PuriRcation helps establish what they are, both bio- before puriRcation. To remove bacteriophage par-
logically and chemically. Finally, bacteriophages are ticles from the gel of a plate stock the gel is minced
model viruses. Study of the evolution of bacterio-
phages helps in the understanding of the evolution of
other viruses, including pathogenic viruses. A major
advantage of studying bacteriophages is their com-
paratively short life cycle. For example, bacterio-
phage T7 has a 13 min life cycle at 373C and a 25 min
life cycle at 303C. The host cell is Escherichia coli.
Most other bacteriophages that infect E. coli have life
cycles less than 60 min at 373C. Thus, studies of evolu-
tion are achieved more quickly than they are with any
other organism. Also, the short life cycle of bacterioph-
ages reduces the time needed to grow them.
Growth of Bacteriophages
Bacteriophages are grown in either gelled or liquid
solutions of nutrients. Gels are made of the polysac-
charide, agar. Gel-embedded infected cells are used to
count viable bacteriophage particles. The procedure
is to embed bacteriophages in the gel, together with Figure 1 Plaques of bacteriophage T7. A Petri plate is filled
with a 1% agar gel in an enriched growth medium. Subsequently,
host cells. The number of host cells is much greater
0.7% molten agar in the same enriched medium is mixed with
than the number of bacteriophages. This gelled mix- both host bacteria and bacteriophage particles. This mixture is
ture is incubated to replicate the cells and infect them. spread on the 1.0% gel. The plate is incubated at 373C. A plaque
A bacteriophage particle infects a growing host cell forms at the position of a single bacteriophage particle.
with either a glass rod or a spatula and the minced gel tant is added before freezing. Glycerol (10%) can be
is incubated with buffer. The bacteriophage particles used as a cryoprotectant. Alternatively, a high mo-
diffuse out of the gel, into the buffer during this lecular weight cryoprotectant can be used. A high
process. Centrifugal pelleting is then used to remove molecular cryoprotectant sometimes used is 10%
the pieces of gel, together with pieces of the host dextran, average molecular weight"10 000. High
bacterium. Some bacteriophages are pelleted with the molecular weight cryoprotectants are less likely to
gel. Thus, the pelleted pieces of gel are resuspended in enter a bacteriophage particle. Therefore, high mo-
buffer and then pelleted a second time. This process is lecular weight cryoprotectants are less likely to cause
called washing. Washing is typically performed two bacteriophage particles to burst from osmotic shock
or three times. The supernatant solutions are pooled during thawing. Osmotic shock during thawing
to produce a clariRed plate stock. occurs because freezing causes nonuniformity in the
Details of plate stock preparation vary slightly concentration of cryoprotectant. If the bacteriophage
among the various bacteriophages. A bacteriophage particle is in a region of comparatively low cryo-
best prepared by plate stock is bacteriophage G. Bac- protectant concentration, an outward osmotic pres-
teriophage G is the largest bacteriophage (known to sure gradient will develop. This is a demonstrated
the author) that can be grown in culture. Bacterio- cause of inactivation of bacteriophage T4 by freeze}
phage G has a double-stranded DNA genome 670 thawing.
kilobase pairs long. Bacteriophage G can be grown
Growth in Liquid Culture
in liquid cultures, but results are erratic and greater
success has been achieved with plate stocks. Typi- A bacteriophage is usually grown in liquid culture,
cally, plate stocks of bacteriophage G are clariRed by when the purpose is either chemical or physical char-
centrifugation at 5000 rpm in a 250 mL bottle (or the acterization of either the bacteriophage or its nucleic
equivalent; centrifugal force at the bottom of the acid. In this case, amounts in the 1}50 mg range may
bottle is 3800 g). This speed is doubled (centrifugal be needed. Either simple, well-deRned media or en-
force is quadrupled) in the case of smaller bacterio- riched media are used. A simple, well-deRned me-
phages that don’t sediment as quickly as bacterio- dium might have glucose as the primary (sometimes
phage G. Pelleting the bacteriophage particles (or only) source of both carbon and energy. The medium
aggregates of them) works against the objective of is typically buffered with phosphate. The medium is
clarifying a plate stock. supplemented with ammonium chloride to provide
Maintaining the stability of the bacteriophage par- nitrogen for proteins. Other salts are also added. Both
ticles is an objective at all stages of puriRcation. magnesium sulfate and calcium chloride are added
Known bacteriophages are stabilized by the presence after sterilization by autoclaving. Apparently, some
of divalent cations. Magnesium is usually used. Thus, requirements, such as iron, are present in sufRcient
magnesium should be present in the buffer. The pres- quantity as contaminants. Alternatively, an enriched,
ence of 0.001 mol L\1 MgCl2 is usually sufRcient for but less well-deRned, medium can be used. The major
stability. Some salt (usually NaCl) should also be components of an enriched medium are often both
present. Some bacteriophages increasingly adsorb to tryptone and an extract of yeast cells.
fragments of host cell, as the salt concentration is Some bacteriophages are more easily (and more
lowered. Thus, salt concentrations are sometimes inexpensively) grown in minimal medium. This is true
raised from the typical 0.1 mol L\ to 0.5 mol L\1 or for some lytic double-stranded DNA bacteriophages,
more. Adsorption to host cell debris can either inacti- for example. A lytic bacteriophage always produces
vate a bacteriophage particle or cause it to pellet with progeny when it infects a cell. The counterpart of
the fragments of gel. a lytic bacteriophage is a lysogenic bacteriophage.
A lysogenic bacteriophage may or may not produce
Storage
progeny. The lysogenic bacteriophage genome both
Plate stocks are also sometimes preferred when a new remains in and replicates with the host, if progeny are
bacteriophage is isolated. The new bacteriophage not produced. This state is called lysogeny. Lysogeny
might have been isolated from the wild. It might also can simplify growth of some bacteriophages, as de-
have been produced by genetic modiRcation of a prescribed below. Growth of a lytic bacteriophage, but
viously isolated bacteriophage. A plate stock made not a lysogenic bacteriophage, encounters the follow-
with a single plate is a rapid and simple way of ing problem: cells are infected at a low ratio of bac-
preserving this new strain. Preservation is completed teriophage particles to host cells, typically 0.01}0.1.
by freezing a clariRed plate stock. Freezing (typically Multiple cycles of infection occur. Therefore, the con-
at !703C) is used to prevent inactivation during centration of cells during the Rnal (and most critical)
storage for periods of years to decades. A cryoprotec- cycle of infection is hard to control. Overgrowth of
cells can result in a suboptimal yield, because of air is an aerator designed for use with tropical Rsh.
inadequate aeration. Undergrowth of cells can result The use of forced air is scalable to at least a 15 L
in suboptimal yield, because of the low concentration culture. Shaking is scalable to roughly a 1 L culture.
of cells. The more rapidly cells grow, the more difR- The control of temperature can be achieved via
cult controlling their concentration during the last several routes. A temperature-regulated fermentor
cycle is. Cells grow more slowly in minimal medium with forced air can be used. The cost and inconven-
(typical doubling time "1.5}2.0 h for E. coli at ience can, however, be dramatically lowered by using
303C) than they do in enriched medium (typical a bottle in a temperature-controlled water bath.
doubling time"30}40 min for E. coli at 303C). Even a bottle in a temperature-controlled, water-
Thus, the infection is more easily controlled in min- Rlled sink can be used. In the latter two cases, a bottle
imal medium. None the less, other factors are also with sterile medium and bubblers is aerated via for-
involved. Some experimentation with growth me- ced air.
dium is required to optimize conditions of growth,
when these conditions have not been previously opti-
First Stage of Puri\cation
mized.
Growing lysogenic bacteriophages is usually easier The targeted degree of puriRcation varies with the
than growing lytic bacteriophages. Lysogenic bacterio- intent of the investigator. The minimal puriRcation is
phages can be made to leave a state of lysogeny and removing large fragments of the host cell. Pelleting in
enter a lytic cycle. Some mutants will do this when the a centrifuge is used. This stabilizes the bacteriophages
temperature is raised. Thus, growth of lysogenic bac- by preventing adsorption of the bacteriophages to
teriophages (bacteriophages and P22, for example) fragments of host cell membrane. Pelleting is typically
can be simpliRed. Host cells in a state of lysogeny are done immediately after lysis for small ((1 L) cul-
grown to an optimal concentration. The temperature tures. Pelleting is sometimes delayed until after
is raised to induce the lytic cycle. The temperature is precipitation of bacteriophage particles for large cul-
then lowered to an optimal temperature for growth. tures. The precipitation is performed by both raising
This process is useful for producing bacteriophage the salt concentration (typically to 0.5 mol L\1) and
particles in large amount. It is also useful for pro- adding a high molecular weight polyethylene glycol.
ducing other components of bacteriophage-infected Details vary among the different bacteriophages. The
cells. For example, some components of bacterio- precipitation concentrates bacteriophage particles for
phage-infected cells retain their metabolic activity subsequent steps in puriRcation. Alternatively, bac-
when infected cells burst (lyse is a frequently used syno- teriophage particles can be concentrated by pelleting,
nym for burst). These activities include packaging without precipitation. The newer centrifuges can pel-
of double-stranded DNA in bacteriophage capsids. let bacteriophages in increased volume and decreased
Thus, a fragment of foreign DNA can be cloned by, time. Bacteriophages in eight 50 mL tubes can now be
Rrst, incorporating the fragment in a bacteriophage pelleted in 1}4 h. The time depends on how rapidly
genome, and, then, packaging the DNA in vitro by the bacteriophage particles sediment at any given
incubating the DNA in an extract of lysed, infected speed of centrifugation.
cells. Extracts of lysed, bacteriophage-infected cells
can sometimes be used for incorporating the foreign Subsequent Stages of Puri\cation
DNA, as well as packaging it. An extract can be made
by use of a lytic bacteriophage, as well as a lysogenic Concentration and partial puriRcation of bacterio-
bacteriophage. The process is, however, less time- phage particles are achieved in the Rrst stage of puriR-
and resource-consuming with a lysogenic bacterio- cation. The subsequent stages can, in theory, be
phage. essentially any procedure of puriRcation used for
other biological macromolecules. However, in prac-
Equipment for Large+Scale Growth of tice, the simplest, highest yielding, highest volume
Bacteriophages in Liquid Culture
procedure has been found to be centrifugation. Three
Small scale (1}1000 mL) liquid cultures are grown in types of centrifugation are used, sometimes in tan-
typical laboratory glassware. A Sask is sometimes dem. The details of procedure vary with both the
used, with aeration by shaking. Alternatively, a bottle properties of the bacteriophage and the purposes of
is used with aeration via a bubbler and forced air. The the investigator.
latter alternative is simpler, less expensive and less
Buoyant Density Centrifugation
space-consuming. A reliable shaker is needed in the
former case; a source of forced air is needed in the Buoyancy is a familiar concept to anyone who has
latter. A reliable and reliably clean source of forced learned to swim. Buoyancy can be used to purify any
macroscopic object by placing the object in solution most linear density gradients. Caesium chloride is the
that continuously varies in density. Gravity will drive most frequently used compound for fractionating
the particle to the region of the density gradient that bacteriophages by buoyant density centrifugation.
has a density equal to the density of the particle The word isodense means same density. But,
(isodense region). This concept also works for bac- what is the density of a bacteriophage particle? Can
teriophages. However, bacteriophages are small. this density vary with position in a density gradient?
They are so small that ultracentrifugation is needed to This density not only can, but does, vary with posi-
drive them to an isodense region (a process called tion in a density gradient. The density varies because
buoyant density centrifugation). Thermally driven the anhydrous components of the bacteriophage are
motion (diffusion) randomizes the position of the not the only components, from the perspective of
bacteriophage particles in the absence of centrifu- buoyant density. The bacteriophage also contains
gation. Even if diffusion did not occur, the bacteri- both water- and density-forming solute. The concen-
ophage particles would take impractically long to Rnd tration of water inside the bacteriophage is not neces-
their isodense region, if centrifugal force were not sarily the same as the concentration outside the
applied. Buoyant density centrifugation is performed bacteriophage. The bacteriophage nucleic acid some-
by mixing bacteriophage particles with a solute that times binds a large amount of solute-free water, for
will form a density gradient when centrifuged (‘be- example. But, this concentration does vary with the
fore’ column of the Rrst row of Figure 2; the sample is concentration of water outside of the bacteriophage.
grey). The density gradient forms because the solute, Analysis of this situation is beyond the scope of this
like the bacteriophage particles, is driven in the direc- article. This analysis is not critical for achieving prac-
tion of the centrifugal force. The motion caused by tical results. However, it is critical for interpreting
centrifugal force eventually achieves equilibrium with densities in terms of a particle’s composition.
thermal motion. The result is a density gradient. The Many studied bacteriophages consist only of a pro-
bacteriophage particles are driven to an isodense re- tein capsid that contains a nucleic acid genome. The
gion of the gradient (‘after’ column of the Rrst row of capsid has a density of roughly 1.3 g mL\1 when
Figure 2; the sample is black). Comparatively small centrifuged to an isodense region in a caesium chlor-
ions (or molecules), like caesium cations, form the ide density gradient. Double-stranded DNA has
a density of roughly 1.7 g mL\1. Both these densities
vary slightly with detailed composition. Double-
stranded DNA bacteriophages are, in general, rough-
ly 50% DNA, 50% protein. The density of these
bacteriophages is 1.5 g mL\1. This density is half-
way between the density of protein and the density of
DNA. This relationship is, however, not an intrinsic
feature of densities. Its source is roughly equal hy-
dration of DNA and protein in caesium chloride
density gradients. These two hydrations are usually
not equal.
Buoyant density centrifugation has the following
limitation. Low molecular weight contaminants
(molecules of RNA, unassembled proteins) diffuse so
rapidly that they contaminate the bacteriophages,
even though their density is different. Thus, bacteri-
ophage puriRcation procedures often have one stage
at which particles are fractionated by size.
Rate Zonal Centrifugation
Rate zonal centrifugation fractionates particles by
both size and shape. The procedure is to layer
a sample in a restricted zone on top of a pre-poured
Figure 2 Purification of bacteriophage particles by ultracen- density gradient. The density gradient is then centri-
trifugation. The three rows illustrate the three types of ultracen-
fuged. All particles migrate into the density gradient,
trifugation described in the text. The appearance of a centrifuge
tube before centrifugation is shown in the column labelled before. because the density gradient has only densities much
The appearance of the same centrifuge tube after centrifugation lower than the densities of the particles being centri-
is indicated in the column labelled after. fuged (illustrated in the second row of Figure 2). The
particles are fractionated primarily by size and shape. The advantages of a caesium chloride step gradi-
The larger a particle is, the more rapidly it sediments. ents make them a method of choice for purifying
The more spherically symmetrical a particle is, the many bacteriophages. A single centrifugation is some-
more rapidly it sediments. Bacteriophages sediment times sufRcient to achieve the purity needed. Alterna-
much more rapidly than unassembled proteins and tively, a buoyant density centrifugation can be per-
RNA. Thus, bacteriophages are separated from these formed after centrifugation in a step gradient.
particles by a single rate zonal centrifugation in a suc- Increase in concentration of the sample is achieved by
rose gradient. Some fragments of host cells co-sedi- combining bacteriophages from several step gradients
ment with bacteriophage particles. However, these in a single buoyant density gradient.
fragments have a variable rate of sedimentation.
The bacteriophage particles have a unique rate of
sedimentation. Therefore, separation from most of
Nondenaturing Gel Electrophoresis
these fragments occurs. Bacteriophage particles can be fractionated by elec-
Rate zonal centrifugation has the following limita- trophoresis through gels. This procedure is called
tion. The sample occupies a small region of the centri- nondenaturing gel electrophoresis because the bac-
fuge tube at the start of fractionation. Furthermore, teriophage particles are intact (infective) during
the sample becomes less concentrated during frac- electrophoresis. In contrast, other procedures of elec-
tionation. Thus, the amount of sample is more limited trophoresis are used to analyse the components of
when compared to the amount fractionated by buoy- disassembled bacteriophage particles. The gels used
ant density centrifugation. During buoyant density for nondenaturing electrophoresis must have pore
centrifugation, the sample occupies the entire centri- sizes large enough to admit bacteriophage particles
fuge tube at the start of fractionation. The sample (30}90 nm in radius). Polyacrylamide gels can be
becomes more concentrated during fractionation. made dilute enough to achieve the needed pore
Thus, rate zonal centrifugation is usually not the sizes. However, these polyacrylamide gels are very
procedure of choice. However, some bacteriophages weak and difRcult to handle. Agarose gels achieve the
(including bacteriophage G) are not stable when cen- needed pore sizes with gels that are easy to handle.
trifuged in caesium chloride density gradients. Rate Agarose has been the gel-forming compound most
zonal centrifugation is a reasonable alternative in this frequently used to fractionate bacteriophages (and
case. Unstable bacteriophages have been stabilized by other viruses) by nondenaturing gel electrophoresis.
including polyethylene glycol in the density gradient Gel electrophoresis fractionates any spherical par-
used for rate zonal centrifugation. ticle by two properties of the particle: the average
electrical surface charge density (), and the radius.
Combined Buoyant Density and Rate Zonal
The value of is the sole determinant of the elec-
Centrifugation
trophoretic mobility ("velocity/electrical Reld) in
Advantages of buoyant density centrifugation are the absence of the gel, within experimental accuracy.
combined with advantages of rate zonal centrifu- For example, a dimer of a bacteriophage capsid has
gation when the following hybrid procedure is used. an electrophoretic mobility indistinguishable from
A comparatively broad zone of sample is layered on that of a capsid monomer, when the electrophoretic
a pre-formed caesium chloride density gradient mobility is extrapolated to a gel concentration of
(before column of the third row of Figure 2). The zero. In theory, the value of will be the sole determi-
caesium chloride density gradient had been formed nant of gel-free electrophoretic mobility, for all con-
before centrifugation, by successfully layering 3}5 ditions likely to be encountered during the analysis of
caesium chloride solutions. These solutions decrease bacteriophage particles. Changing the value of the
in density, from bottom to top. The rapid diffusion of particle’s radius will, however, change its elec-
a caesium cations converts the discontinuous gradient trophoretic mobility during electrophoresis through
(called a step gradient) into a more continuous gradi- a gel. The larger the radius is, the more retarded the
ent. The step gradient with layered sample is centri- particle will be by the Rbres that form the gel.
fuged until bacteriophage particles reach an isodense The fractionation by is independent of fractiona-
position in the gradient (after column of the third row tions achieved by either buoyant density or rate zonal
of Figure 2). Thus, the advantages of buoyant density fractionation. Thus, nondenaturing gel electrophor-
centrifugation are achieved: increase in the sample’s esis can differentiate particles not differentiated by
concentration and fractionation by density. On the any procedure of centrifugation. The result has been
other hand, the advantage of rate zonal centrifugation the Rnding that at least one bacteriophage (T7) exists
is also achieved: separation of bacteriophage particles in more than one state. The different states appear to
from rapidly diffusing components of the cell. have evolved to improve the survivability of T7.
scattering, to avoid stain-induced changing of the

bacteriophage particles.
Large scale fractionation of bacteriophage par-
ticles is almost never done by gel electrophor-
esis. Gel electrophoresis is used primarily for
analysis, not puriRcation. The reasons are the
following:
Figure 3 Nondenaturing gel electrophoresis. A horizontal gel 1. Like rate zonal centrifugation, nondenaturing gel
with samples loaded is illustrated in the before panel. The same electrophoresis begins with the sample in a small
gel with stained, fractionated bacteriophages is shown in the after volume. The sample is diluted during fractiona-
panel. The arrow indicates the direction of electrophoresis.
tion.
2. Recovery of particles after gel electrophoresis is
not as good as recovery after fractionation in solu-
In practice, gel electrophoresis can be performed by tion.
the following procedure: 3. The bacteriophage particles are contaminated
with both agarose and contaminants of agarose,
1. An agarose gel is cast in a horizontal orientation
after fractionation.
(before column in Figure 3). The gel has sample
4. The bacteriophage particles are exposed to prod-
wells for placing the sample (indicated in Fig-
ucts of the electrolysis of water. The result is an
ure 3).
unnecessary source of possible damage of bac-
2. The gel is submerged beneath an electrophoresis
teriophage particles.
buffer. The pH should be close to the pK of the
buffer. The gel chosen should not adhere to the
bacteriophage particles. Agarose can be purchased Concluding Remarks
at several levels of purity. Several agarose deriva-
The puriRcation of bacteriophages by centrifugation
tives and mixtures can be purchased. Some
is a routine procedure. The most difRcult requirement
nonagarose polysaccharides can also be used for
is the obtaining of centrifuges. Challenges are en-
electrophoresis.
countered primarily when a bacteriophage is unstable
3. The sample is placed in a sample well. The sample
during exposure to the conventional conditions of
contains an uncharged compound (glycerol,
centrifugation. A further challenge is encountered
sucrose, a polyethylene glycol, for example),
when bacteriophage particles are puriRed for the
in order to prevent the sample from Soating.
purpose of high resolution analysis of structure.
4. An electrical potential is placed across the gel. The
This latter challenge is heterogeneity of bacteri-
electrical Reld and time of electrophoresis are
ophage particles that appear homogeneous when
chosen, based on the following:
fractionated by centrifugation. Nondenaturing gel
(a) The most rapidly migrating particle should
electrophoresis can reveal heterogeneity. How-
not migrate out of the gel.
ever, nondenaturing gel electrophoresis has not been
(b) The separation between particles should be
useful for preparative fractionation. Thus, a challenge
sufRcient so that each particle forms a separate
for the future is the development of preparative
band.
procedures of nondenaturing gel electrophoresis.
(c) Bacteriophage particles should not disrupt
This last challenge would be met by development of
during electrophoresis.
a procedure of continuous-Sow gel electrophoresis.
(d) The time of electrophoresis should be
Continuous-Sow preparative gel electrophoresis is
short enough so that band broadening is not
possible, in theory. Hopefully, it will be achieved in
a problem.
practice.
(e) The electrical current should be low enough so
that rise in temperature does not destabilize
bacteriophage particles. See also: II/Electrophoresis: Agarose Gels. Centrifu-
gation: Theory of Centrifugation.
The bacteriophage particles are visualized by either
light scattering or staining, after electrophoresis (il-
lustrated in the after column of Figure 3). Ethidium is
Further Reading
a stain used for DNA; Coomassie blue is a stain used Alberts B, Bray D, Lewis J et al. (1994) Molecular Biology
for protein. Preparative fractionation should use light of the Cell, 3rd edn. New York: Garland.
III / BASES: THIN-LAYER (PLANAR) CHROMATOGRAPHY 2109
Proctor LM (1997) Advances in the study of marine viruses. Stent GS and Calendar R (1978) Molecular Genetics: An
Microscopy Research and Techniques 37: 136}161. Introductory Narrative, 2nd edn. San Francisco, CA:
Serwer P, Khan SA and Griess GA (1995) Non-denaturing WH Freeman.
gel electrophoresis of biological nanoparticles: viruses. van Holde KE, Johnson WC and Ho PS (1998) Principles of
Journal of Chromatography A 698: 251}261. Physical Biochemistry. Upper Saddle River, NJ: Prentice
Stafford WF and Schuster TM (1995) Hydrodynamic Hall.
methods. In: Glasel JA and Deutscher MP (eds.) Intro- Yamamoto KR, Alberts BM, Benzinger R et al. (1970)
duction to Biophysical Methods for Protein and Nucleic Rapid bacteriophage sedimentation in the presence of
Acid Research. San Diego, CA: Academic Press, pp. polyethylene glycol and its application to large-scale
111}145. virus puriRcation. Virology 40: 734}744.
BALSAMS AND RESINS: THIN-LAYER

(PLANAR) CHROMATOGRAPHY
See III / ESSENTIAL OILS / Thin Layer (Planar) Chromatography
BASES: THIN-LAYER (PLANAR)

CHROMATOGRAPHY
L. Lepri and A. Cincinelli, University of Florence, Aliphatic Amines

Florence, Italy
Copyright ^ 2000 Academic Press The Rrst attempts to separate aliphatic amines were
performed on silica gel using chloroform}ammonia
Introduction (39 : 1), chloroform}methanol}17% ammonia
According to the Br+nsted}Lowry deRnition (1923) (2 : 2 : 1), butanol}acetic acid}water (4 : 1 : 5) and
a base is a proton acceptor: phenol}water (8 : 3) as eluents. However, highly vol-
atile amines cannot be chromatographed with an
B #H# " BH# ammoniacal solvent. A systematic collection of data
base proton conjugated acid on the chromatographic behaviour of a large number
of aliphatic amine hydrochlorides, with particular
The stronger the base, the larger is its Kb and conse- emphasis on eluents, adsorbents and detection re-
quently the smaller is its pKb and the larger is the pKa agents, was published by Prandi (see hRF values in
of the conjugated acid. Table 1, columns 1, 2 and 3).
There are various types of bases, including natural Silica gel is particularly useful for the adsorption
and synthetic compounds, and they occur in a vast chromatography of amines that have different polar-
array of products, extending from the alkaloids, sulfa ities but does not resolve the fatty amine series. In
drugs (sulfonamides), dyes (azines, indoles), herbi- particular, the RF values increase as the aliphatic
cides (simazine, atrazine, promazine), biogenic chain length increases but this increase becomes
amines to numerous other groups. smaller with increasing chain length.
This chapter includes only aliphatic and aromatic Reversed-phase partition chromatography on
amines and their derivatives, heterocyclic bases and parafRn oil}saturated silica gel is useful for the
miscellaneous compounds (nitrosamines, amides, separation of fatty amines. RM[log(1/RF)\] values
hydrazines). Thin-layer chromatography is used for such amines increase as the length of the aliphatic
extensively for the analysis of bases and can achieve chain increases and there is a linear relationship
separations of complex mixtures comparable to col- between RM and the number of carbon atoms in the
umn liquid chromatography. molecule.
2110 III / BASES: THIN-LAYER (PLANAR) CHROMATOGRAPHY
Table 1 Retention data (hRF)* of aliphatic amine hydrochlorides under different experimental conditionsa
Amine Silica gel b Po-s Kieselguhr c Silica gel Sil C18-50 AWP f
#chlorophenol d #4%HDBS e
Eluent A Eluent B Eluent C Eluent D Eluent E Eluent F
Methylamine 3 6 } 26 92 69
Dimethylamine 4 7 } 31 } }
Trimethylamine } 43 } 35 } }
Ethylamine 7 16 } 46 85 75
Diethylamine 16 32 } 50 } }
Triethylamine } 75 } 53 } }
Ethanolamine 4 10 } } } }
Diethanolamine 5 16 } } } }
Triethanolamine 18 36 } 38 } }
Ethylethanolamine 11 23 } } } }
Ethyldiethanolamine 30 52 } } } }
2-Ethylhexylethanolamine 65 93 } } } }
Propyldiethanolamine 52 69 } } } }
Propylamine 16 35 } 57 } }
Di-n-propylamine 51 80 } } } }
Isopropylamine 17 36 } 63 } }
Diisopropylamine 33 66 } } } }
Propanolamine 4 8 } } } }
Triisopropanolamine 52 85 } } } }
Allylamine } } } 60 } }
Diallylamine } } } 66 } }
Butylamine 22 48 } } 55 65
Di-n-butylamine 63 95 } 80 } }
Tri-n-butylamine } } } 85 } }
Isobutylamine 31 58 } } 62 70
Diisobutylamine 85 99 } } } }
3-Methoxypropylamine 18 43 } } } }
Pentylamine 29 55 } } 34 60
Isoamylamine 30 56 } } 45 62
2-Methylbutylamine 36 68 } } } }
Hexylamine 34 65 86 } 24 53
Cyclohexylamine 33 63 } 76 } }
3-Amino-2,2-dimethylbutane 51 90 } } } }
2-Amino-3-methylpentane 47 78 } } } }
2-Amino-4-methylpentane 42 73 } } } }
Heptylamine 36 70 82 } 14 elongated spots
Octylamine 37 74 78 } 7 3
2-Ethylhexylamine 54 88 } } } }
Di-2-ethylhexylamine 100 100 } } } }
tert-Octylamine 52 87 } } } }
Nonylamine 39 77 74 } } }
Decylamine 40 78 70 } 1 0
Undecylamine 42 79 65 } } }
Dodecylamine 44 79 58 } 0 0
Tridecylamine 47 80 50 } } }
Tetradecylamine 50 82 43 } 0 0
Pentadecylamine 52 83 38 } } }
Hexadecylamine 55 85 30 } } }
Heptadecylamine 58 85 24 } } }
Stearylamine 60 85 18 } } }
1,2-Diaminoethane 2 4 } 22 82 49
1,2-Diaminopropane 3 10 } } 81 58
1,3-Diaminopropane } } } } 84 36
1,4-Diaminobutane } } } } 84 32
1,5-Diaminopentane } } } } 85 26
1,6-Diaminohexane } } } } 79 19
1,7-Diaminoheptane } } } } 72 17
1,8-Diaminooctane } } } } 56 16
N-(3-Aminopropyl)cyclo- 5 18 } } } }
hexylamine
Table 1 Continued
Diethylenetriamine 0 0 } 17 } }
Spermidine } } } } 81 10
Spermine } } } } 72 2
Tetraethylenepentamine 0 0 } } } }
*hRf"Rf;100.
a
Eluents: A and B " chloroform}methanol}17% ammonia in the 82.5 : 15.5 : 2 (A) and 70 : 26 : 4 (B) ratios: C"acetone}17%
ammonia (70 : 30 v/v); D"n-butanol}acetic acid}water (35 : 5 : 10); E"1 M acetic acid # 1 M HCl in 30% methanol; F"2 M
NH4NO3.
b
Silica gel G (Merck); detection reagents: 1% ninhydrin solution in ethanol}acetic acid (95 : 5); 1% potassium permanganate#1%
potassium persulfate (1 : 1); 25% iodine methanolic solution; 5% sodium nitroprusside in acetaldehyde}2% sodium carbonate (1 : 1
v/v) solution. Sample volume; 10 L of a 0.5% water}alchol solution of the amine hydrochloride.
c
Paraffin oil}saturated Kieselguhr G (Merck) layers were prepared by immersing the plates in a 5% solution of the oil in acetone.
d
Home-made plates were prepared by spreading a slurry of 50 g of silica gel G (BDH) in 2% o-chlorophenol solution (100 mL). The
plates were dried for 24 h at 603C before use. Detection agent: 3 g ammonium thiocyanate and 1 g cobalt chloride in 20 mL of distilled
water (blue spots).
e
The Sil C18-50 impregnated layers (Macherey-Nagel) were prepared by immersing the plates in a 4% dodecylbenzenesulfonic acid
(HDBS) solution in 95% ethanol.
f
Home-made plates were prepared by spreading a slurry of 4 g ammonium tungstophosphate (AWP) and 2 g calcium sulfate
hemihydrate in 50 mL of distilled water after stirring 10 min with a magnetic stirrer. Detection agent: 1% ninhydrin solution in a 5 : 1 (v/v)
mixutre of pyridine and glacial acetic acid.
Sources: Adapted from Prandi C (1978) Thin-layer chromatography of aliphatic amines. Journal of Chromatography 155: 149}157;
Srivastava SP, Dua VK and Chauhan LS (1980) Chromatographic behaviour of aliphatic amines on phenol-impregnated thin layers.
Journal of Chromatography 196: 225}235; Lepri L, Desideri PG, Heinler D and Giannessi S (1982) High-performance thin-layer
chromatography of nitrogen compounds on layers of RP-18 and Sil C18-50 untreated or impregnated with dodecylbenzenesulfonic acid
and of ammonium tungstophosphate. Journal of Chromatography 245: 297}308.
Cellulose and aluminium oxide have also been used impregnated with anionic and cationic surfactants.
as adsorbents. The phenomenon of multiple-spot Ion-exchange and/or partition contribute to the re-
formation of amines on cellulose thin layers when tention of the amines depending on the type of sta-
using neutral or weakly acidic eluents is caused by the tionary phase, the percentage of surfactant and the
presence of carboxyl groups in the cellulose. Partial apparent pH of the eluent.
hydrolysis of the amine hydrochloride, volatilization Table 1 (column 5) shows the retention data ob-
of the liberated hydrochloric acid and the presence of tained on Sil C18-50 plates impregnated with a 4%
charged groups in silica gel and alumina layers have dodecylbenzenesulphonic acid solution in 95%
also resulted in double-spot formation for speciRc ethanol and eluted with 1 M acetic acid#1 M hy-
compounds. drochloric acid in 30% methanol. On these plates the
Phosphate and acetate-buffered silica gel and im- retention of polyamines is governed chieSy by an
pregnated plates have been used. Hydrogen bond ion-exchange mechanism while aliphatic mono-
formation between the impregnated plates and amines can be differentiated according to the number
aliphatic amines inSuences their chromatographic be- of carbon atoms in their side chain.
haviour on metal salt-impregnated plates and on Aliphatic amines have been studied on layers of
phenol-impregnated silica-gel layers. weak and strong ion exchangers, Dowex 50-X4
The hRF values of some amines, obtained on silica (Na# and H#), sodium carboxymethylcellulose and
gel impregnated with a 2% solution of o-chloro- Rexyn 102 (Na#), using hydrochloric acid and vari-
phenol using a butanol}acetic acid}water (35 : 5 : 10) ous buffer and salt solutions as eluents.
mixture as eluent are reported in Table 1 (column 4). The use of ammonium tungstophosphate (AWP)
No correlation exists between the pKa value of the as a layer material is particularly promising since on
conjugated acid of an amine and its RM value; it this exchanger different afRnity sequences in com-
therefore seems that the chromatographic behaviour parison with the above-mentioned results are found
of such compounds is due to hydrogen bond forma- (see Table 1, column 6). The behaviour of poly-
tion between the amine and silica gel as well as amines is of interest since it seems to be correlated to
o-chlorophenol. the distance between the protonated amino groups
Reversed-phase thin-layer chromatography of sev- involved in the ion-exchange process and, in the case
eral aliphatic mono- and polyamines has been per- of 1,2-diaminoethane and 1,2-diaminopropane, to
formed on layers of silanized silica gel untreated and the steric hindrance of the methyl group.
Quantitative analysis of the diamine hydrochloride Table 2 (column 1) lists the hRF values of
recovered from acid-hydrolysed copolyamides pre- 19 phenylethylamines on reversed-phase OPTI-
pared from diamine-diacid has been carried out on UPC12 plates eluted with 1 M HCl#3% KCl in
silica gel G eluting with phenol}n-butanol}formic water. The presence of potassium chloride in the
acid}water (5 : 2 : 1 : 2 v/v) or phenol}formic eluent accounts for the compactness of the spots.
acid}water (74 : 1 : 25 v/v). Densitometric scanning Among the diastereoisomers, the differences in the
was performed using a Shimadzu spectrophotometer retention allow the separation of norephedrine from
at 560 nm after spraying the plates with a 0.2% norpseudoephedrine.
solution of ninhydrin in ethanol and heating at 903C The Sil C18-50 plates impregnated with 4% N-DPC
for 15 min. and eluted with 0.5 M Na2CO3 in 30% methanol
SpeciRc procedures have been developed for al- (Table 2, column 2) show considerable differences
kanolamines. The high performance TLC (HPTLC) in the afRnity sequence of phenylethylamines with
behaviour of closely related diethanolamines was respect to the untreated layers and an improvement in
studied on silica-gel layers eluted with binary solvents the separation of both the two diastereoisomers of
(methanol}chloroform, methanol}dichloromethane, norephedrine and the two isomers of phenylethylamine.
methanol}acetone, acetone}chloroform) and on four On layers of RP-18 #4% dodecylbenzenesulfonic
types of reversed-phase, chemically bonded, silica acid (HDBS) (Table 2, column 3), the retention of the
gel with methanol}water as mobile phase. Alkanol- compounds depends on the concentration of hydro-
amines, in particular ethanolamines and iso-pro- chloric acid in the eluent and can be ascribed to the
panolamines, are used extensively in hydraulic brake cation-exchange process between the protonated
Suids and cutting oils as corrosion inhibitors; the amino group and the surfactant adsorbed on the
derivatives with fatty acids are used as emulsiRers and layer.
detergents. Their separation and identiRcation is per- A peculiar behaviour is observed on AWP#
formed on neutral silica gel using methylene chlor- CaSO4 ) H 2O plates (Table 2, column 4) since the

ide}95%}ethanol}ammonia (0.880) in the propor- steric hindrance of the phenyl group on the carbon
tions 43 : 43 : 15 v/v as eluent. A solution of 0.2% atom bound to the amino group allows a complete
ninhydrin, then alizarin in acetone, is employed to separation ("1.86) between the two isomeric
locate the separated alkanolamines. The method has phenylethylamines.
been applied to commercial formulations. The use of o-benzenesulfonamido}p-benzoquinone
in methanol or acetone has been described as a means
to detect and distinguish the 3,4-methylenedioxyam-
Phenylalkylamines phetamines of the ‘Ecstasy’ group on silica gel 60
The structures of alkylamines with an aromatic ring F254 plates with ethyl acetate}acetone}methanol}
in the side chain are shown in Figure 1. The 25% ammonia (20#20#8#2) solution as eluent.
phenylethylamine group comprises a range of natural
and synthetic compounds, some of which are used in
drug formulations, and includes catecholamines and
Derivatized Amines
other biogenic amines which are excreted in the urine. Direct chromatography on thin layers of primary and
The separation of these compounds by TLC and over- secondary amines is frequently difRcult due to the
pressured-layer chromatography (OPLC) is very im- strong adsorption of the NH2 or NH groups to the
portant as shown by the numerous studies dealing adsorbents employed. Derivatives are often used to
with the determination of phenylethylamines in phar- overcome these difRculties. A series of reagents has
maceutical preparations, with their identiRcation as been recommended for the formation of derivatives
drugs of abuse and with the determination of cate- of primary and secondary amines in order to assist in
cholamines, their metabolites and their precursors in separating, identifying and determining such com-
urine. These studies are carried out on silica gel, pounds on thin layers. The formulae of the most
cellulose thin layers or using reversed-phase chromato important reagents are shown in Figure 2. The
graphy on different ready-for-use plates of silanized chromatographic behaviour of derivatized amines
silica gel untreated and impregnated with anionic and with some common reagents is shown in Table 3.
cationic detergents. Most of these derivatives are intensely coloured (i.e.,
The chromatographic behaviour of these com- DADB-, PABS-, DBAB-) or give Suorescent spots
pounds has also been studied on strong and weak (NBD-); in some cases, however, the detection can be
ion-exchangers and on layers of AWP, an inorganic considerably improved by exposure of the plates to
synthetic ion exchanger which has been used in the iodine vapour or by spraying the chromatogram with
separation of other nitrogen compounds. 0.01 M sulphuric acid (DBAB-amides).
Figure 1 Structures of phenylalkylamines.
The procedure used for PABS-amides does not values with respect to the derivative of n-butylamine
allow characterization of the movement of individual taken as 100.
compounds by ordinary RF values, as the chromatog- The dansylated derivatives of ammonia, 1,3-di-
raphy is continued after the solvent front has reached aminopropane, 1,4-diaminobutane, 1,5-diaminopen-
the upper edge of the plate. Therefore, the hRX values tane, spermidine, spermine and histamine were
reported in Table 3 (column 2) represent relative separated on silica-gel 60 plates eluted with
Table 2 Retention data (hRF) of phenylalkylamines under different experimental conditionsa
Compound OPTI-UPC12 b Sil C18- RP-18# AWP e Dowex 50 X4 f Rexyn 102 g

50#4% 4% HDBSd (H#) (Na#)
N-DPC c
Eluent A Eluent B Eluent C Eluent D Eluent E Eluent F
Benzylamine } } } } 28 43
1-Phenylethylamine 28 26 30 42 25 51
2-Phenylethylamine 27 18 24 28 18 38
Tyramine 47 44 66 42 23 31
Dopamine 61 16 75 42 30 21
3-Methoxytyramine 26 46 62 21 } }
3,4-Dimethoxyphenethylamine 10 34 44 6 } }
-Hydroxyphenethylamine 47 38 44 46 } }
Octopamine 74 70 78 56 41 30
Noradrenaline 84 25 87 57 49 20
Normetanephrine 55 70 77 40 } }
Synephrine 54 52 74 48 } }
Adrenaline 68 26 88 } } }
Metanephrine 35 52 72 27 } }
Amphetamine 18 14 24 17 20 43
Norephedrine 28 29 32 41 } }
Norpseudoephedrine 23 21 32 44 } }
Ephedrine 16 16 29 27 } }
Pseudoephedrine 13 15 31 26 } }
Hordenine 16 13 54 17 } }
Histamine } } } } 47 15
Tryptamine } } } } 4 23
a
Eluents: A"1 M HCl#3% KCl in water; B"0.5 M Na2CO3 in water}methanol (30%); C"1 M CH3COOH#1 M HCl in
water}methanol (40%); D"1 M NH4NO3 in water; E"4 M HCl; F"0.2 M acetate buffer.
b
OPTI-UP C12 plates (Antec).
c
The plates were impregnated with 4% N-dodecylpyridinium chloride in ethanol.
d
The plates were impregnated as described in Table 1.
e
4 g Ammonium tungstophosphate#2 g calcium sulfate hemihydrate in 50 mL of distilled water.
f
Home-made Dowex 50-X4 (H#) layers were prepared by mixing 3 g of the resin (200}400 mesh) with 6 g of microcrystalline cellulose
in 40 mL of water.
g
Home-made Rexyn 102 (Na#) layers (Fisher) were prepared by mixing 3 g of the resin (200}400 mesh) with 6 g of microcrystalline
cellulose in 40 mL of water.
Sources: Adapted from Lepri L, Desideri PG and Coas V (1973) Chromatographic and electrophoretic behaviour of primary mono- and
diamines on layers of weak and strong ion exchangers. Journal of Chromatography 79: 129}137; Lepri L, Desideri PG and Heimler
D (1985) High performance thin-layer chromatography of phenylethylamines and phenolic acids on silanized silica and on ammonium
tungstophosphate. Journal of Chromatography 347: 303}309.
chloroform}triethylamine (5 : 1 v/v). Putrescine, ca- Another reagent (NBD-Cl), which itself is not Suor-
daverine, spermidine and spermine can be quantitat- escent but forms Suorescent derivatives with primary
ively determined in human urine; higher amounts of and secondary aliphatic amines, seems to have advant-
the last two amines were found in the urines of cancer ages compared to dansyl chloride since it does not
patients compared to the values of these substances in produce Suorescent products with phenols, thiols and
normal urine. Dansyl amines can be determined by anilines, or as a consequence of hydrolysis reactions.
in situ Suorescence on silica gel and, sometimes, on Two-dimensional (2D) TLC allows the separation
polyamide layers. In favourable cases as little as of nearly every mixture of interest. A comparative
10\12 moles/spot of a DANS-derivative can be visual- study of the chromatographic properties of de-
ized on a normal thin-layer plate. rivatized biogenic amines with dansyl chloride, dab-
The separation and quantiRcation of dansylated syl chloride and 4-chloro-7-nitrobenzoxazole shows
biogenic amines in vegetables have been recently per- that the dansyl chloride should be preferred in terms
formed on 20;20 cm silica-gel HPTLC plates with of both sensitivity and, to a lesser extent, resolution
stepwise gradient elution using the Personal OPLC on silica gel by 2D-TLC. Dansylated and dabsylated
BS50 overpressured-layer chromatography apparatus products were found to have similar TLC character-
(Kovàcs et al., 1998). istics and were adequately resolved by eluting in the
Figure 2 Structures of the reagents used for derivative formation.
Rrst direction with ethyl acetate}cyclohexane (3 : 2 gel G or alumina using organic and aqueous}
v/v) and in the second with benzene}triethylamine organic solutions as eluents.
(5 : 1 v/v). The plates were developed in the dark. Further studies concern the chromatographic be-
haviour of aromatic amines on silica gel impregnated
with silver compounds as -complexing metal and
Aromatic Amines manganese, cadmium and zinc salts as complexing
Table 4 shows the hRF values of several primary agents. On silica gel impregnated with manganese
aromatic amines under different experimental condi- salts, the pKa values of the conjugated acids of
tions. Such compounds have been studied on silica- sixteen isomeric methylanilines and chloroanilines
Table 3 Retention data (hRF or hRXH for eluent B) of derivatized amines under different experimental conditionsa
Amine DADB- b Silica PABS- c Alumina NBD- d Silica DBAB- e Silica DNP- f Silica
gel#Carbowax gel G gel HF254
Eluent A Eluent B Eluent C Eluent D Eluent E
Ammonia } } 45 } 3
Methylamine 15 47 57 17 30
Dimethylamine 44 116 } 19 34
Ethylamine 24 71 63 30 59
Diethylamine 58 143 } 36 69
n-Propylamine 33 87 } 36 75
Di-n-propylamine 65 150 } 44 }
Isopropylamine 39 92 } 40 74
Diisopropylamine } } } 47 }
Allylamine } 81 } 39 62
Diallylamine } 147 } } }
n-Butylamine 42 100 71 42 81
Di-n-butylamine 68 '150 } 48 }
Isobutylamine 50 106 71 43 82
Diisobutylamine } '150 } 48 }
sec-Butylamine } 104 } } }
tert-Butylamine } 105 } } }
n-Amylamine 50 108 } } 85
Di-n-amylamine } '150 } 51 }
Isoamylamine 51 113 } 47 86
n-Hexylamine } 115 } 47 88
Di-n-hexylamine } } } 57 }
Cyclopentylamine } 100 } } }
Cyclohexylamine } 105 } 51 }
Dicyclohexylamine } } } 56 }
Octylamine } 120 } } 90
Di-n-octylamine } } } 64 }
Decylamine } 127 } } 92
Benzylamine } 87 } } 67
Dibenzylamine } 141 } } }
1-Phenylethylamine } 86 } } }
2-Phenylethylamine 26 83 } } 66
Ephedrine } 57 } } }
Amphetamine } 94 } } }
-Phenylisopropylmethylamine } 126 } } }
Mescaline } 15 } } }
Tryptamine 3 } 55 } }
Tyramine 8 } 48 } }
3-Methoxytyramine } } 48 } }
Histamine } } 30 } }
Putrescine } } 31 } }
Cadaverine } } 34 } }
Spermidine } } 27 } }
Spermine } } 30 } }
HhRF"RF;100; hRF"RX values ;100 (relative to the hRX of n-butylamine derivative taken equal to 100).
a
Eluents: A"n-hexane}ethyl acetate (7 : 3); B"25 mL ethyl acetate and 100 mL petroleum ether (62}823 from Shell) saturated with
water; C"toluene}acetic acid (4 : 1 v/v); D"cyclohexane}methylethylketone (70 : 30); E"pentane}benzene}triethylanine
(45 : 45 : 10).
b
4-Dimethylamino-3,5-dinitrobenzoyl-; home-made plates were prepared by spreading a mixture of 10 g of Carbowax 400 (Fluka) and
40 g of silica gel G (Merck) in 80 mL of water.
c
4-(Phenylazo)-benzenesulfonyl-; microchromatoplates (40;76 mm) coated with alumina (Fluka, D5).
d
4-Chloro-7-nitrobenzo-[c]-(1,2,5)-oxadiazole; silica 60 (250 m) TLC plates (20;20 cm) were obtained from Sigma.
e
p-(N,N-Dimethylamino)benzene-p-azobenzoyl.
f
2,4-Dinitrophenyl.
Sources: Adapted from Wirotama IPG and Ney KH (1971) Dunnschicht chromatographie von amines als 4-dimethylamino-3,5-
dinitrobenzoyl amide. Journal of Chromatography 61: 166}168; Jart A and Bigler AJ (1967) Thin-layer chromatographic separation of
primary and secondary amines as 4-(phenylazo) benzene sulfonamides. Journal of Chromatography 29: 255}258; Price NPG and Gray DO
(1993) Mapping of derivatised biogenic amines by two-dimensional thin-layer chromatography. A comparative study. Journal of
Chromatography 635: 165}170; ChuraH c\ ek J (1970) Einige neue reagenzien zur chromatographischen identifizierung von saK uren,
alkoholen und amines. Journal of Chromatography 48: 241}249; Ilert HI and Hartmann T (1972) DuK nnschichromatographie der
2,4-dinitrophenyl derivate wasserdampffluK chtiger amine und ihre anwerdung auf die trennung pflanzlicher amine. Journal of
Chromatography 71: 119}125.
Table 4 Retention data (hRF) of primary aromatic amines under different experimental conditionsa
Amine Silica gel G Silanized silica#4% N-DPC b AG-1 XA(CH3COO\) c pKa
Eluent A Eluent B Eluent C Eluent D Eluent E Eluent F Eluent G
Aniline } } 22 72 92 } } 4.58
o-Toluidine 42 17 16 61 77 } } 4.39
m-Toluidine 29 10 14 68 79 } } 4.69
p-Toluidine 20 5 14 75 81 } } 5.12
2,4-Dimethylaniline } } 7 53 71 } } }
2,6-Dimethylaniline } } 11 21 50 } } }
o-Aminophenol 24 0 } } } 46 83 4.72
m-Aminophenol 13 0 } } } 30 83 4.17
p-Aminophenol 1 0 } } } 70 84 5.49
o-Anisidine 42 15 18 66 78 58 83 4.49
m-Anisidine 30 9 20 53 70 44 83 4.20
p-Anisidine 2 2 26 85 86 74 84 5.29
o-Chloroaniline 75 66 } } } 12 40 2.64
m-Chloroaniline 51 40 } } } 15 64 3.34
p-Chloroaniline 41 22 6 14 44 18 75 3.98
o-Bromoaniline 78 69 6 13 17 10 27 2.60
m-Bromoaniline 58 44 4 10 28 10 55 3.51
p-Bromoaniline 47 27 5 9 37 12 69 3.91
o-Nitroaniline 55 52 5 7 17 4 8 !0.29
m-Nitroaniline 44 36 9 10 28 21 31 2.50
p-Nitroaniline 37 29 6 9 23 2 7 1.02
o-Aminobenzoic acid 47 44 } } } } } }
m-Aminobenzoic acid 28 12 } } } } } }
p-Aminobenzoic acid 37 29 } } } } } }
o-Phenylenediamine 0 0 37 81 86 67 83 4.47
m-Phenylenediamine 0 0 49 96 96 71 84 4.88
p-Phenylenediamine 0 0 60 95 97 79 84 6.08
2,4-Dichloroaniline } } 6 15 44 } } }
2,4-Dinitroaniline } } 3 5 15 0 0 !4.53
2,4-Diaminotoluene } } 34 92 94 72 83 }
2,5-Diaminotoluene } } } } } 79 84 }
2,6-Diaminotoluene } } 50 95 96 73 83 }
3,4-Diaminotoluene } } 10 23 34 67 83 }
2,4-Diaminoanisole } } } } } 72 83 }
2-Amino-4-nitrophenol } } } } } 1 10 }
2-Amino-4,6-dinitrophenol } } } } } 0 0 }
4-Nitro-o-phenylenediamine } } } } } 8 16 }
2-Amino-4-chlorophenol } } } } } 3 41 }
2-Amino-3,4,6-trichlorophenol } } } } } 0 1 }
-Naphthylamine } } 3 9 38 } } }
4-Aminodiphenylamine (DPA) } } 2 31 53 } } }
2-Amino-DPA } } 2 11 23 } } }
3-Methoxy-4-amino-DPA } } 2 29 55 } } }
4-Methoxy-4-amino-DPA } } 2 45 56 } } }
4,4-Diamino-DPA } } 28 97 97 } } }
2,4-Dinitro-4-amino-DPA } } 2 3 26 } } }
Benzidine } } 6 65 81 } } }
o-Tolidine } } 2 26 73 } } }
o-Dianisidine } } 2 7 49 } } }
a
Eluents: A"dibutyl ether ethylacetate}acetic acid (15 : 5 : 1); B"dibutyl ether}acetic acid}n-hexane (20 : 1 : 4); C"0.1 M
CH3COONH4#0.1 M NH4OH in water}methanol (20%) (pH"9.20); D and E"0.1 M and 2 M, respectively, CH3COOH in
water}methanol (20%); F"0.1 M acetate buffer; G"1 M acetic acid.
b
Home-made layers prepared by spreading a mixture of 20 g of silanized silica gel 60HF (Merck) with 4% N-dodecylpyridinium chloride
in 50 mL of 95% ethanol.
c
Home-made AG1-X4 (CH3COO\) plates prepared by mixing 2 g of the resin (200}400 mesh) and 6 g of microcrystalline cellulose in
40 mL of water.
Sources: Adapted from Gillio-Tos M, Previtera SA and Vimercati A (1964) Separation of some aromatic amines by thin-layer
chromatography. Journal of Chromatography 13: 571}572; Lepri L, Desideri PG and Heinler D (1979) Soap thin-layer chromatography
of sulfonamides and aromatic amines. Journal of Chromatography 169: 271}278; Lepri L, Desideri PG and Coas V (1974) Chromato-
graphic and electrophoretic behaviour of primary aromatic amines on anion-exchange thin layers. Journal of Chromatography 90:
331}339.
were correlated with their RM values using ben- a 5 : 1 mixture of ethanol and glacial acetic acid,
zene}ethyl acetate}acetic acid (2#2#1 v/v) as 0.2% chloranil in chlorobenzene or 9-chloroacridine
mobile phase. Separations via charge-transfer in 95% ethanol.
complexes with nitro compounds (picric acid, 2,4,6- Diazotization and coupling can be carried out dir-
trinitrophenyl-N-methylnitramine and 2,4-dinitroch- ectly on the layer, i.e., the plates can be exposed to
lorobenzene) have also been reported. nitrogen dioxide to diazotize the amines and then
Reversed-phase planar chromatography has been sprayed with a solution of 0.1 M -naphthol and
performed on silanized silica gel untreated or impreg- 0.1 M triethylamine in benzene.
nated with cationic and anionic surfactants. The aro- Derivatives of aromatic amines have also been used
matic amines, which are in the free base form at the for separating and identifying these compounds.
pH of the eluent, exhibit a high afRnity towards the Therefore, 2,4-dinitrophenyl derivatives and dansyl
silanized silica gel and are more strongly retained in derivatives have been studied on silica gel with dif-
the presence of N-dodecylpyridinium chloride on the ferent solvent systems.
stationary phase. Fifty-four aromatic amines used as antioxidants
As the pH of the eluent decreases, a sharp increase and/or antiozonants for elastomers have been separ-
in the RF values is observed on the impregnated layers ated on silica gel with a concentrating zone using
(see Table 4, columns 3}5). Such behaviour is corre- benzene}ethyl acetate}acetone (100 : 5 : 1 v/v) and
lated with the protonation of one or more of the benzene}n-hexane (50 : 50 v/v) as eluents. The detec-
amino groups present in the aromatic amines. On the tion reagent is N-chloro-2,6-dichloro-p-benzoquinone
basis of hRF values many interesting separations of monoimine in buffered alkaline medium.
isomers can be carried out.
Primary aromatic amines can be separated, with
difRculty, on polystyrene-based cation exchangers
Heterocyclic Bases
in aqueous}organic solutions and also by elution with Heterocyclic compounds containing one or more ni-
concentrated mineral acids owing to the high trogen atoms have been extensively investigated on
afRnity of such exchangers towards compounds thin-layer chromatography as detailed in Table 5.
which contain one or more aromatic nuclei. There- Almost all the heterocyclics are chromatographed on
fore weak cation exchangers, i.e., carboxymethylcel- polar stationary phases (silica gel, alumina, cellulose,
lulose (H# or Na# form) and alginic acid, or syn- polyamide) and, to a lesser extent, on hydrophobic
thetic inorganic exchangers such as ammonium mo- layers obtained by impregnating polar adsorbents
lybdophosphate and tungstophosphate, have been with nonpolar substances or by using silanized silica
used for separating such compounds. Better results gel plates. Chemically modiRed and impregnated layers
can be achieved using polystyrene-based anion ex- with cationic and anionic surfactants are also used.
changers as shown by the hRF values obtained on Among the various types of ‘simple’ nitrogen het-
AG-1X4 (CH3COO\) plates (see Table 4, columns erocyclics, the separation of pyridines, indoles, quino-
6 and 7). lines and pyrimidines is of interest. The indole group
As regards the inSuence of the pH of the eluent, not of compounds is conventionally divided into the so-
that the protonated forms of the amines exhibit called simple derivatives and the indole alkaloids and
a lower afRnity towards the exchanger than the dyes. A number of simple indole derivatives play
free base forms. important roles in physiological processes. Alkaline
An equation similar to that suggested for alkaloids and acidic systems are employed on both silica gel
can be used for studying quantitatively the inSuence and silanized silica for the separation of these com-
of eluent pH on the chromatographic characteristics pounds (see Table 6).
of aromatic amines: The two-dimensional technique on Sil C18-50
plates can be performed by eluting in the Rrst direc-
(1/RF )!1"(1/RFalk!1)(Ka/Ka#[H#])) tion with n-hexane}ethyl acetate}acetic acid
#(1/RFac!1)([H#]/(Ka#[H#])) (72 : 27 : 1 v/v) and in the second direction with
0.1 M ammonia in 40% methanol. This technique
where Ka is the dissociation constant of conjugated allows the separation of 20 indole derivatives; the
acid of the base and RFac and RFalk are the RF values of spots can be identiRed from their positions on the
the protonated and the free base form of the amines, chromatogram and also from their colours obtained
respectively. after spraying with 1% p-dimethylaminobenzal-
The detection of aromatic amines has been ac- dehyde solution in concentrated hydrochloride acid}
complished with Suorescamine in glacial acetic acid methanol (1 : 1) and heating the plates at 503C for
(1 mg mL\), 5% p-dimethylaminobenzaldehyde in 20 min.
Table 5 Methods used for the separation of various types of nitrogen heterocyclics
Compound Layer Eluent
Azines Alumina Benzene}Chloroform (1 : 1)

Azines Silica gel with or without a fluorescent CCl4}EtOH}Me2CO (50#1#2); n-PrOH-n-hexane
indicator (F254) in different ratios
Azines Cellulose MN-300 F254, Aminoplast CH3OH}H2O}CH3CN (30#20#5);
CH3OH}aqueous ammonia (30#20);
CH3OH}aqueous acetic acid (30#20)
Azines Silica gel F254 impregnated with 5% Water}organic solvents
paraffin oil in n-hexane
Azines RP-2, RP-8, RP-18 H2O}CH3OH in different proportions
Benzodiazepines Silica gel and silica gel F254 CHCl3}Me2O}iPrOH and n-hexane}Me2CO in
different ratios; CHCl3}Me2CO (80#20); benzene;
BuOH}CHCl3}NH4OH (50#50#1);
BuOH}C6H6}NH4OH (50#50#1 and
40#10#30); CHCl3}MeAc (90 : 10); C6H6}
iPrOH}25% NH4OH (85#15#1)
Benzodiazepines RP-18, KC18F Acetate or phosphate buffer#CH3OH or CH3CN in
different proportions
Benzodiazepines Silica gel impregnated with oleyl NaOH}aqueous buffer solutions (pH 9}9.5) saturated
alcohol with oleyl alcohol
Benzoquinoxalinone derivatives Silica gel 60 F254 CH2Cl2#EtAc (17#3)
Benzimidazoles and Benztriazoles Cellulose F254, starch F254 H2O#CH3OH or CH3CN in different proportions
Carbazoles Silica gel Benzene; EtAc}MeOH}HCOOH}Pyridine (7.5}7.5}
7.5}10); EtOH
Carbazoles Alumina Petroleum ether (40}603C)}CHCl3 (10 : 1); petroleum
ether}HAc (10 : 1)
Cyclic amidines Silica gel F254 CH3OH#CHCl3 or benzene in different ratios
Imidazolines Silica gel Benzene}Me2CO}25% NH4OH (4 : 17 : 1)
Imidazoles Silica gel CHCl3}Me2CO}HAc (34 : 4 : 3) EtAc saturated with
NH4OH
Indoles Silica gel BuOH}HAc}H2O (2 : 1 : 1); Me2O}HAc (100 : 1);
phenol}H2O (4 : 1); iPrOH}25% NH4OH}H2O
(20 : 1 : 2) BuOH}EtOH}cyclohexylamine (76 : 3 : 6);
Me2CO}CHCl3}HAc}H2O (8 : 8 : 4 : 1); MeCOEt}
CHCl3}conc. NH4OH (40 : 10 : 1); MeCOEt}CH3OH}
conc. NH4OH; EtAc}iPrOH}conc. NH4OH (45 : 35 : 20)
Indoles Cellulose BuOH}HAc}H2O (12 : 3 : 5); benzene}dioxane}H2O
(1 : 1 : 1); benzene}pyridine}H2O (1 : 1 : 1)
Indoles Polyamide CHCl3}EtAc}HAc (7 : 2 : 1); CHCl3}cyclohexane}
BuOH}HAc}H2O (1 : 1 : 1 : 1 : 0.2)
Indoles Sil C18-50; Sil C18-50 #4% N-DPC H2O}CH3OH}HAc (59 : 40 : 1); 0.1 M NH4OH in 40%
methanol; n-hexane}EtAc}HAc (72 : 27 : 1); 0.1 M
NH4OH#0.1 M NH4Cl in 40% methanol; 0.5 M
NH4OH#0.5 M NH4Cl in 40% methanol; n-hexane
EtAc}HAc (67 : 32 : 1)
Indoles AWP 1 M NH4NO3
Indole esters Silica gel Octanol}petroleum ether (110}1153C) (1 : 5)
Nitroimidazoles Silica gel G F254 impreganted 1 M NH4OH#1 M NH4Cl in water or in various
with a 5% solution of silicone oil in mixtures with methanol
diethyl ether
Piperazines Silica gel EtAc}CH3OH (4 : 1); CHCl3}CH3OH}HAc (14 : 2 : 1)
Piperidines Alumina CHCl3 saturated with NH4OH
Pyrazoles Silica gel EtAc (saturated with H2O); CHCl3}Me2CO (7 : 3);
MeCOEt
Pyrazolones Silica gel Cyclohexane}CHCl3}EtOH (4 : 10 : 1); cyclohexane}
MeCOEt (4 : 5)
Pyrazolidines Alumina
Pyridines Silica gel Benzene}MeOH (25 : 1); EtAc}MeOH}HAc (15 : 4 : 1);
diethylether}dimethylformamide (99 : 1)
Pyridines Silica#Ag2O Me2CO}C6H6 (2 : 3); MeCOEt}iPrOH (4 : 1); CHCl3}
CH3OH (3 : 2)
Table 5 Continued
Pyridine derivatives Silanized silica gel untreated or Phosphate buffer solution}CH3OH (1 : 1) with the buf-
impregnated with sodium dodecyl- fer pH adjusted to the appropriate value (pH 3 to 10)
sulphate as ion-pairing reagent
Pyrimidines Silica gel CHCl3}CH3OH (3 : 1)
Pyrimidines Dowex 50-X4 (H#) 0.25}4 M hydrochloric acid
Pyrimidines Carboxymethylcellulose (Na#), Water; 0.1 and 0.5 M acetate buffer
Dowex 50-X4 (Na#)
Pyrroles Silica gel n-Hexane}CHCl3 (9 : 1); diethylether}2% HAc
in n-hexane (1 : 1)
Pyrrole acids Silica gel CHCl3}96% HAc (1 : 1); benzene}CH3OH}HAc
(45 : 8 : 4)
Quinolines Silica gel EtAc}iPrOH}NH4OH (9 : 6 :4 ); benzene-EtAc (1 : 1)
Quinolines Silica gel 60HF254, aluminium oxide, Binary mixtures of n-heptane with polar modifiers
Florisil (iPrOH; dioxane, EtAc, tetrahydrofuran MeCOEt,
Me2CO, iPr2O, CH2Cl2)
Thioindigoid thiazolidinones Silica gel H EtOH; CH3CN; Et2O; CH3OH
Triazoles Alumina n-Hexane}benzene (1 : 1); 1% adipate in xylene}
formic acid (49 : 1)
Tryptophan and some of its indole metabolites in evaluate the role of the ion-exchange process on the
urine (5-hydroxytryptophan, tryptamine, serotonin, chromatographic behaviour of these compounds
indolyl-3-acetic acid and 5-hydroxyindolyl-3-acetic (Table 8). The hRF sequences of the pyrimidines un-
acid) have been separated by TLC, stained with van der the various conditions are related to solvophobic
Urk’s Salkowski reagent, and determined by scanning effects, their acid}base characteristics, and the associ-
densitometry using indolyl-3-butyric acid as internal ation of the species in solution.
standard. Sep-Pak C18 cartridges are used for extrac- Seven different thin-layer chromatographic systems
tion of metabolites from urine. The detection limits were investigated to separate 19 pairs of the E}Z
are 2 g mL\ 5-hydroxytryptophan, 1.75 g mL\ geometrical isomers of pyrimidine, purine and pyra-
5-hydroxyindolyl-3-acetic acid, 1.5 g mL\ tryp- zole derivatives with potential cytokinin activity.
tophan, 0.8 g mL\ indolyl-3-acetic acid, 0.9 g mL\ These systems employed silica, silanized silica,
indolyl-3-butyric acid, 1.75 g mL\ serotonin and silanized silica/Cu(II) cation, chemically bonded RP-8
1.25 g mL\ tryptamine. and RP-18 as stationary phases and a variety of bi-
The chromatographic behaviour of several pyridine nary mobile phases. The best performance was ob-
derivatives, some diazines and their sulRdes, dimers served on silica gel with n-hexane}ethyl acetate 1#9
and trimers has also been studied on both silica gel v/v as mobile phase.
and silanized silica (see Table 7). The retention of For the location of heterocyclic bases, Dragen-
azines and diazines in adsorption TLC on silica gel dorff’s reagent is usually employed; other detection
with n-propanol}n-hexane eluents can be calculated agents are tetracyanoethylene and iodine}azide solu-
by the relationship adopted by Kowalska for rever- tions.
sed-phase chromatography: Pyrazoles can be visualized with sodium nitroprus-
side; pyrazolones with 1% mercuric nitrate, iodine}
RF"A#B(X1)1/2#CX2 potassium iodide or 10% ferrocyanide}12.5%
hydrochloric acid (1 : 1); and imidazoles with iodo-
where X1 and X2 are the volume fractions of alcohol platinate reagent. Indoles can be detected with
and hydrocarbon, respectively, and A, B, and C the Ehrlich’s, van Urk’s and Prochazka’s reagents, or
equation constants. In the reversed-phase systems on o-phthalaldehyde}sulfuric acid solution (0.15%
RP-2, RP-8 and RP-18 plates, RM values were found w/v). A speciRc and sensitive Suorescent detection
to be linearly dependent on methanol concentration method was proposed for the analysis of nitro-
(binary methanol}water mixtures) showing the de- imidazoles (titanium (III) chloride followed by spray-
pendence of the retention values on the hydrophobic- ing with diazotized sulfanilic acid).
ity and chemical structure of the analysed compounds.
A large number of pyrimidines (nucleobases in-
cluded) have been studied on plates of silanized silica
Miscellaneous Nitrogen Compounds
gel with different characteristics. The layers were also Because of the possible presence of nitroso carcino-
impregnated with anionic surfactants in order to gens in foods, the separation and determination of
Table 6 Retention data (hRF) of indole derivatives under different experimental conditionsa
Indole derivative Silica gel G Sil C18-50/UV254
Eluent A Eluent B Eluent C Eluent D Eluent E
Indole 84 73 16 13 97
Skatole 87 78 } } }
3-Hydroxymethylindole 84 45 } } }
Indole-3-aldehyde 81 20 25 18 60
Indole-3-acetaldehyde 86 46 32 22 96
Indole-3-ethanol } } 29 23 74
Indole-3-acetone } } 24 18 95
Indole-3-acetonitrile 85 46 23 17 95
Indole-2-carboxylic acid } } 34 78 63
Indole-3-carboxylic acid } } } } }
Indole-5-carboxylic acid } } 69 89 75
Indole-3-acetic acid 31 28 62 84 61
Indole-3-propionic acid 38 34 22 77 75
Indole-3-butyric acid 40 38 14 69 81
Indole-3-glyoxylic acid } } 61 82 0
Indole-3-lactic acid } } 59 80 4
Indole-3-acrylic acid 33 29 15 24 64
5-Hydroxyindole-3-acetic acid 19 4 78 92 18
Indole-3-acetamide } } 41 36 34
Indole-3-ethylacetate } } 9 7 98
Indole-3-glyoxylamide } } 25 20 52
Isatin 75 27 42 40 78
Gramine 77 0 } } }
Tryptamine 77 0 47 2 0
Serotonin 65 0 62 8 0
Tryptophan 23 0 64 67 0
5-Hydroxytryptophan 14 0 } } }
a
Eluents: A"methyl acetate}isopropanol}25% ammonium hydroxide (9 : 7 : 4); B"chloroform}96% acetic acid (95 : 5);
C"water}methanol}acetic acid (59 : 40 : 1); D"0.1 M NH4OH in 40% methanol; E"n-hexane}ethyl acetate}acetic acid
(72 : 27 : 1).
Source: Adapted from Stahl E and Kaldeuoey H (1951) Trace analysis of physiologically active, simple indole derivatives. Zeitschrift fu( r
Physiologische Chemie 323: 183}191; Lepri L, Desideri PG and Heinler D (1983) High-performance thin-layer chromatography of
indole derivatives on layers of Sil C18-50 untreated or impregnated with N-dodecylpyridinium chloride and on anmonium tungstophos-
phate. Journal of Chromatography 260: 383}389.
N-nitrosamines are of interest. Silica gel, magne- group and of different substituents under conditions
sium silicate, cyano- and diol-bonded silica have been close to those in physiological systems is highly im-
used as stationary phases to separate these com- portant.
pounds. Alkyl- and arylnitrosamines can be separated The separation of formanilide and ten para-sub-
on silica gel with n}hexane}diethyl ether}dichloro- stituted acetanilides was investigated on starch and
methane (4 : 3 : 2 v/v) and cyclic nitrosamines with cellulose layers, as an alternative to RP-18, using
the same solvent mixture in a ratio of 5 : 7 : 10 aqueous mobile phases with methanol, acetonitrile,
(v/v). acetone and 1-propanol as modiRers. Because
The chromatographic behaviour of 26 com- RM values are related to the partitioning of solute
pounds derived from 1,1-diphenylhydrazine was in- molecules in the given system, they can be regarded as
vestigated by normal and reversed-phase chromato- a measure of solute hydrophobicity.
graphy. The same techniques were used for the The retention behaviour of three series of aromatic
separation of several thiosemicarbazides and 1,2,4- amides was investigated on silica gel with eight binary
triazoline-3-thiones on silica gel, alumina, and C18- solvent mixtures (benzene}CHCl3 20 : 30 v/v; ben-
modiRed silica-gel layers, and nonaqueous and zene}ethyl acetate 45 : 5 v/v; benzene}acetone 45 : 5
aqueous eluents. v/v; benzene}dioxane 45 : 5 v/v; heptane}ethyl acet-
Amides are physiologically active compounds and ate 40 : 10 v/v; carbon tetrachloride with ethyl acet-
a knowledge of possible interactions of the amide ate, acetone or dioxane in a 45 : 5 v/v ratio). The
Table 7 Retention data (hRF) of various heterocylic bases under different chromatographic conditionsa
Compound Silica gel G RP-2 RP-8 RP-18
Eluent A Eluent B Eluent C Eluent D Eluent E Eluent E Eluent E
Pyridine 29 54 20 } 70 55 56
2,2-Bipyridyl } } 40 } 72 53 52
2,2,2-Tripyridyl } } 35 } 66 46 41
2-Methylpyridine 30 54 } } } } }
3-Methylpyridine 35 55 } } } } }
4-Methylpyridine (-picoline) 27 48 0 26 61 49 46
,-Bipicolyl } } 24 20 60 49 43
2,4-Dimethylpyridine 28 49 } } } } }
2,6-Dimethlypyridine 36 59 } } } } }
2,4,6-Trimethylpyridine 26 51 } } } } }
2-Ethylpyridine 42 62 } } } } }
2-n-Propylpyridine 47 64 } } } } }
2-Hydroxypyridine 6 20 } } } } }
2-Aminopyridine 27 50 } } } } }
Pyridine-2-carbinol 18 45 } } } } }
Pyridine-2-aldehyde 51 67 } } } } }
Pyridine-2-carboxylic acid 2 4 } } } } }
Pyridine-2,6-dicarboxylic acid 3 5 } } } } }
2-Acetylpyridine 57 69 } } } } }
2-Benzoylpyridine 62 71 } } } } }
2-Fluoropyridine 62 69 } } } } }
2-Chloropyridine 61 70 } } } } }
2-Bromopyridine 63 72 } } } } }
3-Chloropyridine 56 65 } } } } }
3-Bromopyridine 57 67 } } } } }
3-Iodopyridine 58 70 } } } } }
Pyrazine } } 10 13 73 54 57
2,2-Bipyrazyl } } 24 40 73 52 51
Quinoline } } 33 65 78 58 57
2,2-Biquinolyl } } 70 67 74 53 45
6,6-biquinolyl sulfide } } 5 36 73 36 31
8,8-Biquinolyl sulfide } } 47 78 } } }
Quinoxaline } } 32 54 82 56 56
2,2-Biquinoxalyl } } 54 83 70 43 32
Thieno[2,3-b; 4,5-b]biquinoxalyl } } 8 } 74 51 39
2-Methylquinoxaline } } 20 55 74 55 49
3,3-Dimethyl-2,2-biquinoxalyl } } 34 59 67 41 28
a
Eluents: A"ethylacetate; B"acetone; C"chloroform}ethanol}acetone (50#1#2); D"n-propanol}n-hexane (45#55);
E"methanol}water (90#10).
Source: Adapted from Petrowitz HJ, Pastuska G and Wagner E (1965) Thin layer chromatography of some pyridines and quinolines.
Chemiker-Zeitung 89: 7}12; Baranowski I and Swierczek S (1994) A study of the retention of azines and diazines in RPTLC with
bonded alkyl stationary phases. Journal of Planar Chromatography } Modern TLC 5: 399}405.
anilides used were: N-substituted amides of 2,2- as substituent at the para position. Spots were ob-
dimethylpropanoic acid; N-substituted benzamides served under UV light at "254 nm.
and -phenylacetamides, with }F, }Cl, }Br, }CF3, Dialkylaminoethyl dialkylamido Suorophosphates,
}CH3, }C2H5, }OCH3, }CN, }N(CH3)2 or }N(C2H5)2 which exhibit choline esterase-inhibiting effects, were
Table 8 Retention data (hRF) of pyrimidine derivatives under different experimental conditionsa
Pyrimidine derivative OPTI-UPC12 Sil C18-50#4%HDBS
Eluent A Eluent B Eluent C Eluent D
4-Amino-2-hydroxypyrimidine (cytosine) 54 42 7 24
3-Methylcytosine 53 19 8 25
5-Hydroxymethylcytosine 61 52 11 33
5-Methyl-2,4-dihydroxypyrimidine (thymine) 30 29 78 79
1-Methylthymine 13 19 71 73
2,4-Dihydroxypyrimidine (uracil) 55 51 86 86
4,6-Dihydroxypyrimidine 70 75 83 84
4,6-Dihydroxy-2-methylpyrimidine 63 66 72 74
5-Nitrouracil 57 52 84 87
5-Hydroxymethyluracil 63 60 87 90
6-Methyluracil 31 30 80 83
5,6-Dimethyluracil 14 19 67 76
5-Aminouracil 64 64 73 96
6-Aminouracil 54 58 84 85
Uracil-5-carboxylic acid 87 87 84 94
Uracil-6-carboxylic acid (orotic acid) 68 73 90 89
Orotic acid methyl ester 20 29 77 80
Uracil-6-acetic acid 68 71 90 90
6-Chlorouracil 34 56 69 78
5-Chlorouracil 41 47 68 77
5-Bromouracil 36 40 63 73
5-Iodouracil 27 31 57 66
6-Chloromethyluracil 29 34 62 71
5-Trifluoromethyluracil 40 47 60 66
6-Chloro-1,3-dimethyluracil 3 7 44 48
5-Hydroxyuracil (isobarbituric acid) 66 68 91 93
2-Thiouracil 48 56 70 77
4-Phenyl-2-thiouracil 3 17 23 28
5-Methyl-2-thiouracil 28 34 65 70
6-Hydroxy-2-thiouracil (2-thiobarbituric acid) 68 72 83 85
a
Eluents: A"3% KCl in water; B"0.5 M Na2CO3 in water; C"1 M acetic acid in water}methanol (20%); D"1 M acetic acid
#0.5 M HCl in water}methanol (20%).
Source: Adapted from Lepri L, Coas V, Desideri PG and Zocchi A (1988) Planar chromatography of purines, pyrimidines and
nucleosides on untreated and detergent-impregnated silanized silica plates. Journal of Planar Chromatography } Modern TLC 1:
317}324.
separated on silica-gel plates developed with meth- aminopyrene, aminophenylnaphthalene and amino-
anol}pyridine}formamide, 80#15#5 (v/v) and de- chrysene were found in the sludge.
tected with ninhydrin.
The lipophilicity of fused-ring nitrogen heterocyc-
lic was determined by reversed-phase TLC on paraf-
Concluding Remarks
Rn oil impregnated Silkoplat plates (Labor, MIM) This article demonstrates the current possibilities for
using acetonitrile}aqueous solutions of different pH the separation of selected organic bases. Most atten-
as eluents. tion has been devoted to the advantages of TLC,
Lastly, amino-PAHs have been identiRed in sewage retention mechanisms, structure, sample preparation,
sludge after separation of the DMF extracts on mobile and stationary phases, usual modes of devel-
a silicic acid column into three groups: carbazoles, opment and detection procedures, and also two-
aminoarenes and azaarenes. RP-18 F254 (Merck) dimensional techniques.
layers were developed with acetonitrile#water (9 : 1 QuantiRcation of organic bases mainly by in situ
v/v), observed under UV illumination at "254 and densitometry has been described. The development of
365 nm and then sprayed with speciRc reagents for HPTLC and the potential for other innovations such
detection of the amino group. Aminonaphthalene, as overpressured thin-layer chromatography
aminoquinoline and/or aminoisoquinoline, amino- (OPTLC) with this group of compounds is of con-
Suorene, aminoanthracene and/or aminophenanthrene, siderable interest.
2124 III / BILE ACIDS / Gas Chromatography
See also: II/ Chromatography: Thin-Layer (Planar): FateH r Z, Tasi G, Szabady B and Nyiredy Sz (1998) Identi-
Densitometry and Image Analysis; Layers; Modes of Rcation of amphetamine derivatives by uni-dimensional
Development: Conventional; Modes of Development: multiple development and two-dimensional HPTLC com-
Force Flow, Overpressured Layer, Chromatography and bined with postchromatographic derivatization. Journal
Centrifugal; Spray Reagents. III/Amines: Gas of Planar Chromatography*Modern TLC 11: 225}229.
Chromatography. Impregnation Techniques: Thin- Kovàcs A, Simon-Sarkadi L and Mincsovics E (1998) Step-
Layer (Planar) Chromatography. Pharmaceuticals: wise gradient separation and quantiRcation of dan-
Basic Drugs: Liquid Chromatography. sylated biogenic amines in vegetables using personal
OPLC instrument. Journal of Planar Chromato-
graphy*Modern TLC 11: 43}50.
Klimes J and Kastner P (1993) Thin-layer chromatography
Further Reading of benzodiazepines. Journal of Planar Chromatogra-
Airaudo ChB, Gayte-Sorbier A, Aujoulat P and Mercier phy*Modern TLC 6: 168}180.
V (1988) Thin-layer chromatography of amine anti- Seiler N (1971) IdentiRcation and quantiRcation of amines
oxidants and antiozonants used in elastomers. Journal by thin-layer chromatography. Journal of Chromato-
of Chromatography 437: 59}82. graphy 63: 97}112.
BILE ACIDS
Bile acid conjugates form micelles with phospholipids

Gas Chromatography that solubilize cholesterol in the bile. The primary bile
acids, cholic acid and chenodeoxycholic acid, are
A. K. Batta and G. Salen, effectively reabsorbed from ileum during their en-
UMD-NJ Medical School, Newark, NJ and VA Medical terohepatic circulation, but approximately 5% that
Center, East Orange, NJ, USA escape reabsorption seep into the colon, and are sub-
jected to modiRcation to secondary bile acids by in-
testinal bacteria. These modiRed bile acids, in par-
ticular the 7-dehydroxylated bile acids, lithocholic
Introduction acid and deoxycholic acid, are the major faecal bile
Bile acids, a group of steroidal acids with a carboxyl acids and are also signiRcantly absorbed from the
group in the side chain, are the major end products of colon and circulate in the enterohepatic circulation,
cholesterol catabolism, formed in the liver, con- with deoxycholic acid as one of the major plasma and
jugated with amino acids, glycine and taurine, and biliary bile acids in humans. Whereas only small
secreted into the bile. In most animal species, bile amounts of bile acids are excreted into the urine,
acids contain 24 carbons, with the terminal side chain approximately 500 mg per day is excreted in faeces
carbon in the form of a carboxyl group; however, and forms a major catabolic pathway for the elimina-
certain reptiles have 27-carbon bile acids as the major tion of body cholesterol.
biliary bile acids. Primary bile acids are formed via In hepatobiliary and intestinal diseases, the hepatic
the 5-saturation of cholesterol double bond by hep- synthesis and clearance of bile acids and their intesti-
atic enzymes, epimerization of 3-hydroxyl group to nal absorption are abnormal, which disturbs both
-conRguration and further insertion of 7- and/or cholesterol synthesis and its metabolism, causing in-
12-hydroxyl group, shortening of the side chain by creased plasma, urinary and faecal concentrations of
three carbons and oxidation of the terminal carbon to bile acids. This results in accumulation of precursors
a carboxyl group. Structures of some of the bile acids of cholesterol or bile acids and clinical malformations
found in animal species are shown in Figure 1. Bile ensue. Early diagnosis of such conditions is often
acids facilitate the absorption of dietary lipids, in- possible from bile acid analysis in bile, serum, urine
cluding fat-soluble vitamins and cholesterol, via their and faeces. On the other hand, bile acids have
detergent action. The detergent properties of bile therapeutic applications in conditions of abnormal
acids result from their unique structure with a non- cholesterol biosynthesis and metabolism, and che-
polar steroid skeleton and a polar carboxyl group and nodeoxycholic acid, and its 7-hydroxy epimer,
-oriented hydroxyl groups, further increased by hep- ursodeoxycholic acid, are used for medical treatment
atic conjugation with glycine and taurine (Figure 1). of gallstones while ursodeoxycholic acid is also being
III / BILE ACIDS / Gas Chromatography 2125
Figure 1 Structures of bile acids and their derivatives.
used in clinical trials for a variety of hepatobiliary 1960, GC was performed using metal or glass-packed
diseases, including primary biliary cirrhosis, primary columns. Introduction of capillary or open tubular
sclerosing cholangitis and hepatitis C, and is sugges- columns was a major landmark in the development of
ted to suppress colon polyp formation in experi- GC. The separation of serum bile acids using capil-
mental animals. Thus, bile acid analysis is an impor- lary columns was Rrst demonstrated by Laatikainen
tant diagnostic tool for diseases of the liver and intes- et al. in 1975 (Table 1). Coupled with the fact that
tine, and for monitoring bile acid therapy in such the retention times are highly reproducible, capillary
diseases. Of the several methods employed for bile columns have fast become columns of choice for most
acid analysis in biological Suids, gas chromatography chromatographic applications for bile acids. The
(GC) has proven to be the most versatile and is usual capillary column is 25}50 m long with an inter-
sensitive enough to quantitate nanomole to picomole nal diameter of 0.22}0.25 mm. The inner wall is
amounts present in biological specimens. coated with the liquid phase, which may vary from
nonpolar (e.g. methyl silicones, such as OV-1, CP-
Sil-5 CB and SE-30) to more polar (like phenyl-
Choice of Column methyl- and cyanopropylsilicone, e.g. CP-Sil-19 CB)
For more than a decade, since the Rrst application of depending on the need. Usually peaks are very sharp;
GC for bile acid analysis by Vanden Heuvel et al. in however, the dead volume must be very small and
Table 1 Important advances in gas chromatography of bile acids
1960}61 Vanden Heuvel et al. First application of gas chromatography for bile acid analysis. Bile acid methyl ester-
trifluoroacetates were analysed on packed columns
1965 Grundy et al. Comprehensive method for faecal bile acid analysis using packed columns. Bile acids were
isolated free from faecal fatty acids, sterols and other contaminants before chromatogra-
phy. Method often used as a standard to which other methods are compared
1975 Laatikainen et al. Introduction of capillary columns for serum bile acid analysis
1987 Child et al. Simultaneous quantitation of faecal bile acids, sterols and fatty acids
1998 Batta et al. Bile acid analysis was greatly simplified. Bile acids were derivatized and quantitated directly in
the stool sample and the multiple steps of removal of sterols and fatty acids were avoided
only microgram or nanogram quantities of com- cules, which may interfere in their analysis. It is often
pounds can be injected in order to get appropriate necessary to isolate and at least partially purify bile
peak shapes. acids before preparation for GC. Plasma bile acids
New instrumentation and use of splitless injectors mainly exist as glycine and taurine conjugates and
have greatly reduced the dead volume while increas- unconjugated bile acids are present in only small
ing the sensitivity, so that capillary columns are ideal proportions. Plasma is usually passed through Sep-
for chromatography of trace compounds in biological pak, a reversed-phase C18 cartridge, when proteins
Suids. Quantitation of bile acids in a sample is carried and most of the cholesterol are removed and bile
out by comparison of peak areas with those of known acids and their conjugates are eluted with methanol.
amounts of reference compounds in a range where Other methods of concentrating bile acids include
a linear detector response is obtained. Usually lipophilic anion exchange gel, diethylaminohyd-
a known amount of an internal and/or external stan- roxypropyl Sephadex LH-20 and, more recently, size
dard is added to the sample to facilitate quantitation. exclusion chromatography using Sephadex G-75 gel.
Often a mass spectrometer is used as a detector for Some of these ion exchange resins are also employed
GC, and gas chromatography}mass spectrometry for separation of various conjugated bile acids so that
(GC-MS) has the added advantage that the mass conjugation pattern is determined in the sample. The
fragmentation can give insight to the structure of the bile acid fraction obtained by any of these methods is
compound. The sensitivity of GC-MS is several-fold treated with strong alkali (4 mol L\1 sodium hydrox-
increased by using selected-ion monitoring, when ide, 1153C at 15 psi pressure) to hydrolyse the con-
picomole quantities of bile acids can be detected. jugates; neutral sterols are extracted out with solvents
Selection of the right internal standard is important and the free bile acids are then extracted with ether or
for quantitation of bile acids using the mass spec- ethyl acetate after acidiRcation. Under these condi-
trometer in the selected-ion mode. Often isotope- tions, the glucuronides and sulfate esters are also
labelled bile acids (polydeuterated) are used, the ad- hydrolysed, and a pre-step of solvolysis may not be
vantage being that such compounds mix with the bile necessary. As an alternative to alkaline hydrolysis,
acid to be quantitated and are similarly extracted bile acid conjugates may be hydrolysed with the en-
from the biological source. Otherwise, a bile acid zyme, cholylglycine hydrolase, while -glucuronidase
with different GC retention time is employed, and plus sulfatase is used to hydrolyse glucuronides and
peak heights are calibrated using known amounts of sulfates. In order to correct for losses during extrac-
the compounds to be quantitated. The mass ion se- tion, an internal standard is added before the hydroly-
lected for quantitation is usually a high mass ion and sis step, while an external standard (usually 5-cho-
with signiRcant abundance. lestane) corrects for detector responses of the various
compounds. The appropriate internal standard is
a bile acid that is not present in the bile acid mixture
Extraction of Bile Acids to be quantitated. Nor-deoxycholic acid and nor-
Bile acids are present from 1}2 g mL\1 in plasma cholic acid are usually used as internal standards, but
and urine to signiRcant amounts in the intestinal other compounds like hyodeoxycholic acid and
content and as much as 80 mg mL\1 in the gall blad- 7, 12-dihydroxy-5-cholanoic acid have also been
der bile. They are present in unconjugated form as employed.
well as conjugated with glycine and taurine or as Biliary bile acids are present predominantly as the
sulfate/glucuronides, often in association with pro- glycine and taurine conjugates. Since they are present
teins, sterols and their esters, free or esteriRed fatty in high concentrations in the bile, only a few micro-
acids, bile pigments and water-soluble small mole- litres of bile are usually required for quantitation by
GC. Also, since bile contains only 1}2% of choles-

terol, it is much easier to obtain pure bile acids free
from cholesterol. Urinary bile acids are also present
mainly in conjugated form, and only a few milligrams
per day are excreted. However, in hepatobiliary dis-
eases like primary biliary cirrhosis, urine may become
a major pathway for bile acid excretion and sulfate
conjugation of bile acids is increased. Again, bile
acids need to be deconjugated before GC analysis and
methods similar to those for plasma can be used to
obtain urinary bile acid composition.
Faecal bile acids are mainly found in unconjugated
form and the bile acid pattern is highly complex, due
to bacterial deconjugation and extensive metabolism
during intestinal transit. Most abundant of these sec-
ondary bile acids are deoxycholic acid and lithocholic
acid, which are reabsorbed from the colon and fur-
ther modiRed by hepatic enzymes and circulate in
enterohepatic circulation. Whereas scores of metab-
olites of bile acids are formed in the colon, bile acids
in the jejunum remain predominantly conjugated due
to the absence of bacteria in the small intestine, and
they therefore mirror biliary bile acids. A major difR-
culty in quantitative analysis of faecal bile acids is
their strong binding with the bacterial debris in the
stool, and quantitative extraction is difRcult. Further-
more, in addition to bile acids, stool contains neutral
sterols, including cholesterol and its bacterial metab-
olites and plant sterols and their bacterial metab-
olites, and also fatty acids. Several methods have been
reported for bile acid extraction from faeces, and
most are quite complex. Thus, one method involves
continuous Soxhlet extraction of aliquots of hom-
ogenized stool with chloroform}methanol; the
extracted bile acids are subjected to methyl ester
formation followed by preparative thin-layer chro-
matography. Grundy et al. extracted neutral sterols
from homogenized stool, deconjugated any bile acid
conjugates with alkali and Rnally extracted bile acids Figure 2 Gas chromatography of bile acids. Chromatographic
with chloroform}methanol after acidiRcation of the conditions are described in Table 2; CP-Sil-5 CB capillary column
solution. After evaporation of solvents, the residue was employed. (A) n-Butyl ester-trimethylsilyl ether derivatives of
standard bile acids. Peak identification, derivatives of: 1, nor-
was chromatographed over Florisil and the puriRed
cholic acid; 2, lithocholic acid; 3, deoxycholic acid; 4, cheno-
bile acid fraction was subjected to methyl ester deoxycholic acid; 5, cholic acid; 6, ursodeoxycholic acid. (B)
formation and then to preparative TLC to separate Faecal bile acids in a healthy volunteer. Dried stool (10 mg)
fatty acids. Bands due to bile acids were isolated and containing 20 g nor-cholic acid as internal standard was directly
used for GC. In other methods, stools have been subjected to n-butyl ester followed by trimethylsilyl ether forma-
tion, dissolved in 200 L hexane and 1 L was injected into the
extracted with ammoniacal alcohol, methanol}
GC column. Peak identification: 1}6, same as in (A); a, sitosterol,
hydrochloric acid, acetic acid}toluene, and bile acids 3, isodeoxycholic acid; 7, 3-keto, 12-hydroxy-5-cholanoic acid;
extracted after removal of neutral sterols. 8, 12-ketolithocholic acid (C) Plasma bile acids in a patient with
lipid storage disease, sitosterolemia. Plasma (1 mL) containing
10 g nor-cholic acid as internal standard was used. Bile acids
Derivatization of Bile Acids were enzymatically deconjugated and bile acids obtained by
Derivatization of Carboxyl Group passing through Sep-pak. Bile acids were derivatized as
n-butyl ester-trimethylsilyl ethers, dissolved in 100 L hexane and
The carboxyl group in bile acids is most often 5 L was injected into the GC column. Peak identification: 1}6,
converted into the methyl ester. Treatment with same as in (A); a, sitosterol.
Table 2 GC retention indices of bile acids as their methyl ester- can silylate hydroxyl groups in all positions.
trimethylsilyl ether derivativesa Bis(trimethylsilyl)triSuoroacetamide is equally react-
ive as N,O-bis(trimethylsilyl)acetamide but the re-
Retention indexb
agent and the by-products are more volatile. The
Bile acid methyl ester-trimethylsilyl ether CP-Sil-5 CB CP-Sil-19 CB older silylating reagent, a mixture of 1,1,1,3,3,3-
hexamethyldisilazane, trimethylchlorosilane and
Lithocholic acid 3157 3339 pyridine (3 : 1 : 9) is still commonly used for derivat-
Deoxycholic acid 3221 3373
ization. Retention volumes of several common bile
Chenodeoxycholic acid 3244 3397
Cholic acid 3261 3381
acid methyl ester-silyl ether derivatives on capillary
Ursodeoxycholic acid 3279 3439
Hyodeoxycholic acid 3256 3422
Hyocholic acid 3340 3445
-Muricholic acid 3310 3468
Nor-cholic acid 3140
Nor-deoxycholic acid 3106
Homocholic acid 3346
3,7,12-Trihydroxy-5-cholestanoic 3468
acid
a
A Hewlett-Packard model 6890 gas chromatograph equipped with
a flame ionization detector and an injector with a split/splitless device
for capillary columns was used. A chemically bonded fused silica
CP-Sil-5 CB or CP-Sil-19 CB capillary column (25 m;0.22 mm i.d.)
was employed and helium was used as the carrier gas. The column
temperature was kept at 1003C for 2 min, then increased at a rate of
353C min\1 to a final temperature of 2783C.
b
Retention indices (Kovats values) were determined by previous injec-
tion of an alkane mixture under identical GC conditions.
diazomethane in ether converts bile acids instantly

into their methyl esters. However, small amounts of
methyl ethers are formed as artefacts, so alternative
methods are often employed. Methyl esters are quant-
itatively formed with methanol}sulfuric acid or an-
hydrous methanolic hydrochloric acid. Alternatively,
2,2-dimethoxypropane-hydrochloric acid may be
used for derivatization. In addition to methyl esters,
ethyl, n-propyl, isobutyl and n-butyl esters have all
been employed, the advantage being that the reten-
tion times of the bile acid derivatives are increased
and, sometimes, resolution is improved. In particular,
much better separations from neutral sterols can be
obtained and, as shown in Figure 2, chemical re-
moval of neutral sterols may be completely unnecess-
ary with the use of butyl esters.
Derivatization of Hydroxyl Groups

The derivative of choice for the hydroxyl group is the
Figure 3 Gas chromatography of 6-hydroxylated bile acids.
trimethylsilyl ether and a variety of silylating reagents Chromatographic conditions are described in Table 2. (A) CP-Sil-
are commercially available for quantitative derivatiz- 5 CB capillary column; (B) CP-Sil-19 CB capillary column. Peak
ation of even hindered and tertiary hydroxyl groups. identification, derivatives of: 1, 3,6,7-trihydroxy-5-cholanoic
Thus, N,O-bis(trimethylsilyl)acetamide, in combi- acid; 2, 3,6,7,12-tetrahydroxy-5-cholanoicacid; 3, 3,6,7-
trihydroxy-5-cholanoic acid; 4, 3,6,7,12-tetrahydroxy-5-
nation with trimethylchlorosilane, can convert rela-
cholanoic acid; 5, 3,6,7-trihydroxy-5-cholanoic acid; 6,
tively unhindered hydroxyl groups into their 3,6,7,12-tetrahydroxy-5-cholanoic acid; 7, 3,6,7-tri-
trimethylsilyl ether derivatives and, when mixed with hydroxy-5-cholanoic acid; 8, 3,6,7,12-tetrahydroxy-5-
trimethyliodosilane, the reagent is highly potent and cholanoic acid.
Figure 4 Gas chromatography of acetate-methyl esters and trimethylsilyl ether-methyl esters of bile acids. Chromatographic
conditions are described in Table 2; CP-Sil-5 CB capillary column was employed. (A) Trimethylsilyl ether-methyl esters of bile acids. (B)
Acetate-methyl esters of bile acids. Peak identification: 1, lithocholic acid; 2, deoxycholic acid; 3, chenodeoxycholic acid; 4, cholic acid;
5, 3,6-dihydroxy-5-cholanoic acid; 6, ursodeoxycholic acid; 7, 3,6,7-trihydroxy-5-cholanoic acid; 8, 3,6,7-trihydroxy-5-
cholanoic acid; 9, 3,6,7-trihydroxy-5-cholanoic acid.
columns of low and medium polarity are given in also be protected as acyl derivatives and formate,
Table 2, while the chromatographic resolution of acetate and triSuoroacetate derivatives have all been
these derivatives of a number of bile acids with addi- employed for GC. These derivatives are often more
tional hydroxyl group at C6 (present in several animal stable than the silyl derivative and sometimes show
species and in patients with hepatobiliary diseases) is better resolution, as seen in Figure 4. A one-step
shown in Figure 3. Although trimethylsilyl ethers derivatization of the hydroxyl and carboxyl groups
have gained general applicability in bile acid analysis, with heptaSuorobutyric anhydride, also resulting in
other silylating groups sometimes have advantages. simultaneous hydrolysis of the glycine and taurine
Thus, dimethylethylsilyl and dimethylpropylsilyl ether conjugates of the bile acids, has been reported.
derivatives have longer retention times than the corre-
Derivatization of Oxo Groups
sponding trimethylsilyl ethers, while t-butyldimethyl-
silyl ether derivatives are very stable, and also show Oxo bile acids are formed by bacterial modiRcation
signiRcantly longer retention times, so that often bet- of bile acids during their intestinal transit and are there-
ter resolutions can be obtained. Hydroxyl groups can fore found in the intestinal content. Quantitation of
2130 III / BILE ACIDS / Liquid Chromatography
oxo bile acids is beset with problems. Thus, oxo bile faecal bile acids could be quantitatively derivatized
acids, in particular, 3-oxo bile acids, are vulnerable to and eluted from the faeces and bile acids could be
rigorous alkaline hydrolysis conditions and therefore quantitated in the presence of other faecal compo-
the much milder enzymatic hydrolysis of conjugates is nents. The method is also applicable to plasma bile
preferred when oxo bile acids are suspected. Oxo bile acid measurement. Figure 2 shows plasma and faecal
acid methyl esters can be chromatographed without bile acids in a healthy volunteer determined as the
further derivatization (the hydroxyl groups in the n-butyl ester-trimethylsilyl ether derivatives. Since the
partially oxidized compounds must be derivatized); extra steps of extraction of bile acids and their puriR-
however, sometimes artefacts are produced and it cation are eliminated, the method can easily be
may be better to protect the oxo groups as the O- adapted for routine screening of samples for bile acid
methyloxime or the dimethylhydrazone derivatives. analysis.
See also: II/Chromatography: Gas: Delectors: Mass

Summary Spectometry; Derivatization; Solid Phase Extraction.
Detection and quantitation of bile acids are impor- III/Acids: Gas Chromatography; Liquid Chromatography;
tant in health and disease. Bile acid synthetic defects Thin-layer (Planar) Chromatography. Extraction: Super-
may arise from both defective cholesterol synthesis critical Fluid Extraction.
and metabolism; bile acid therapy may correct some
of these abnormalities. GC has proved to be of vital Further Reading
importance in determination of bile acids in bio-
Batta AK, Salen G, Rapole KR, Batta M, Earnest D and
logical Suids. However, a complete GC analysis gen-
Alberts D (1998) Capillary gas chromatographic analy-
erally takes a long time. By appropriate column and sis of serum bile acids as the n-butyl ester-trimethylsilyl
derivatization selection, analysis time can be signiR- ether derivatives. Journal of Chromatography (B) 706:
cantly reduced, but sample preparation for GC analy- 337}341.
sis is usually time-consuming. Combination of GC Child P, Aloe M and Mee D (1987) Separation and quanti-
and mass spectrometry greatly aids in the character- tation of fatty acids, sterols and bile acids in faeces by
ization of unknown bile acids. In attempts to get gas chromatography as the butyl ester-acetate deriva-
additional information, Child et al. have developed tives. Journal of Chromatography 415: 13}26.
a method to quantitate sterols and fatty acids to- Grundy SM, Ahrens Jr EH and Mietinnen TA (1965)
gether with bile acids in stool samples. Advantage is Quantitative isolation and gas-liquid chromatographic
taken of the greatly increased retention times of the analysis of total faecal bile acids. Journal of Lipid Re-
search 6: 397}410.
n-butyl ester-acetates of bile acids, so that sterol acet-
Kuksis A (1976) Gas chromatography of bile acids. In:
ates are eluted earlier than all bile acids present and Marinetti GV (ed.), Lipid Chromatographic Analysis,
a group separation of faecal fatty acids, sterols and vol. 2, pp. 479}610. New York: Marcel Dekker.
bile acids is achieved. In a further advancement of Sjovall J, Lawson AM and Setchell KDR (1985) Mass
the method, Batta et al. have prepared n-butyl esters spectrometry of bile acids. In: Clayton RB (ed.)
of bile acids directly in faecal samples, followed by Methods in Enzymology 111: 63}113. New York:
trimethylsilyl ether formation. They showed that Academic Press.
K. Saar, Aventis Research & Technology, chemical structure, all bile acids tend to form micelles
Frankfurt, Germany and serve as natural detergents. In addition, bile acids
S. MuK llner, Enzyme Technology, DuK sseldorf, Germany are co-factors of pancreatic lipases and are pivotal for
Copyright ^ 2000 Academic Press the digestion and absorption of fats and fat-soluble
vitamins.
The primary bile acids cholic acid and cheno-
Introduction deoxycholic acid are synthesized in the liver and
Bile acids, the major components of the bile Suid, during their enterohepatic circulation various modiR-
play a signiRcant role in lipid metabolism. After cations occur. Conjugation with the amino acids
formation in the liver and storage in the gall bladder, taurine or glycine as well as deconjugation, sulfat-
the bile is secreted in the duodenum. Due to their ion, glucuronidation and modiRcation by intestinal
III / BILE ACIDS / Liquid Chromatography 2131
bacteria lead to numerous secondary bile acid struc- High Performance Liquid
tures. The biliary and serum bile acid pattern as such Chromatography (HPLC) Analysis
as well as the detection of several secondary bile acids
in serum can serve as disease markers and are of Separation
interest in therapeutic monitoring. The development of sophisticated HPLC instrumen-
In healthy individuals the bile acid concentrations tation and of precise pumps with low pulsation as
in the gall bladder and in duodenal bile are in the well as improved sensitivity of the detection systems
upper micromolar up to millimolar range. The deter- along with new column materials have made this
mination of faecal bile acids is, however, not ham- technique well suited for routine laboratory work.
pered by their overall concentration but by the com- Over the last 20 years, a number of different HPLC
plex pattern of primary and secondary derivatives as methods for determination of bile acids in all kinds of
well as by the nature of the matrix. Bile acid patterns biological Suids have been published, each of them
in serum, which are for healthy individuals in the with special advantages and disadvantages.
upper nanomolar concentration range, are much Early HPLC work was based on polar column
harder to determine. materials like silica gel and others. However, these
A number of gastrointestinal disorders and hepatic were just adaptations from TLC and LC studies to
or biliary diseases, e.g. liver cirrhosis and hepatitis, higher pressure and were again limited in selectivity
lead to increased serum and urinary bile acid levels and reproducibility due to the nature of the matrix as
and result in signiRcant changes in the bile acid pat- well as the solvents used. It took the development of
tern. Since bile acids are the major products of choles- reversed-phase materials like octadecylsiloxane-
terol metabolism, they play an important role in re- modiRed silica to make HPLC the most commonly
search and development of lipid lowering drugs and used analytical method in the bile acid Reld. Using
therapies. nonpolar C18 materials, conjugated as well as non-
This is the reason why simple, reproducible and conjugated bile acids can be separated satisfactorily.
sensitive methods for detection, separation and quan- Reversed-phase materials tolerate most organic sol-
titation of bile acid patterns in all kinds of biological vents and are stable over a pH range from 0 to 7.
matrices and body Suids are of increasing import- They also tolerate high pressures without changes in
ance in the pharmaceutical industry as well as for the column bed, and custom-made packing materials
clinical chemistry. guarantee the required reproducibility of the analyti-
cal results. Therefore it comes as no surprise that
Liquid Chromatographic Methods almost all of the HPLC methods published for bile acid
separation are based on reversed-phase applications.
With enzyme-based assays or other analytical The solvent systems described are normally based
methods like radioimmunoassays either total bile acid on organic solvents, e.g. acetonitrile or methanol,
concentrations or single bile acids can be detected. mixed with a buffer system. Buffers are based either on
Gas}liquid chromatography is a commonly used ammonium salts, ammonium acetate or ammonium
method for bile acid analysis, however, it does not carbamate, or on alkali phosphates.
allow the detection of conjugated bile acids. There-
fore, in order to achieve relevant diagnostic data with Detection
regard to changes in the bile acid pattern, other speci-
Rc separation techniques have to be employed. The major challenge in bile acid analysis is their
In early studies on liquid chromatography of bile detection, particularly of unconjugated bile acids.
acids, silica gel was a commonly used stationary Various methods have been described based on detec-
phase. Other materials such as aluminium oxide or tion systems like UV absorption, refractive index,
neutral polymers were also used for separations Suorescence, electrochemistry and enzymatic or
under gravity pressure. Various solvent systems, e.g. chemical post-column derivatization. The most com-
mixtures of light petroleum, butanol, glacial acetic mon and sensitive methods will be discussed here.
acid and other organic solvents have been described, Electrochemical and refractive index detection, due
mostly in combination with thin-layer chromatogra- to their limitations in sensitivity and reproducibility,
phy (TLC) as a detection method. With the systems play only a marginal role in bile acid analysis.
described, the separation of different groups into con-
jugated/nonconjugated bile acids is possible, however UV detection The amino acids glycine and taurine
poor sensitivity and insufRcient selectivity are major make the whole group of conjugated bile acids detect-
drawbacks and make them unsuitable for routine able by UV absorption at 200 nm (Figure 1). This
diagnostics. method limits the components of the mobile phase to
Fluorescence detection A method to increase sensi-

tivity in determination of bile acids is Suorescence
labelling and Suorescence detection. Two different
systems to generate Suorescent bile acid derivatives
are commonly used.
A relatively old method to quantify total bile acid
concentrations is enzymatic detection by 3-HSD.
Since all natural bile acids possess a hydroxyl group
in the 3-position, by enzymatic dehydrogenation
and via formation of nicotinamide adenine dinucleo-
tide (NADH), spectrometric detection at 340 nm is
possible. Going back to this principle, ‘column reac-
tors’ containing immobilized 3-HSD have been de-
veloped. When connected with the separation unit,
usually a C18 column, each substance is enzymatically
transformed and as one of the reaction products,
NADH can be determined by Suorescence detection
Figure 1 Separation of 10 conjugated bile acids by reversed- at 350}460 nm. This method allows detection of
phase HPLC and UV detection at 200 nm: 1, taurocholic acid; nearly all free bile acids having a 3-hydroxyl group
2, tauroursodeoxycholic acid; 3, glycocholic acid, 4, glycourso- independent from conjugation, but is not suitable for
deoxycholic acid; 5, taurochenodeoxycholic acid; 6, taurodeoxy- derivatives where the 3-hydroxyl group is esteriRed,
cholic acid; 7, glycochenodeoxycholic acid; 8, glycodeoxycholic
e.g. by sulfation or glucuronization. In the analysis of
acid; 9, taurolithocholic acid; 10, glycolithocholic acid.
biological samples, other steroids with free 3-hy-
droxyl group can inSuence the results. Application of
solvents with low UV absorption. As detection at low the post-column derivatization method needs addi-
wavelength is sensitive to all kinds of impurities, tional instrumentation for introduction of the enzy-
efRcient sample preparation especially of biological matic reagent.
material with low bile acid concentration is recom- On the other hand, several Suorescent agents for
mended. Also high purity reagents for the preparation pre-column derivatization have been reported. One of
of buffers and ultra pure organic solvents are re- the most reproducible and sensitive methods is based
quired. Taking this into consideration, all methods on 4-bromomethyl-6,7-dimethoxycoumarin (BMC)
based on RP-HPLC with UV detection allow qualitat- as derivatizing agent BMC reacts with carboxyl
ive and quantitative separation of all conjugated bile groups by formation of Suorescent esters detectable
acids with biological relevance by gradient elution. at 340}460 nm. Therefore, unconjugated and
The detection limits are in the range from 40 to 60 ng glycine-conjugated bile acids can be labelled as well
and show some individual characteristics for the re- as all kinds of bile acid esters at the 3-hydroxyl
spective bile acids. group. Despite the high sensitivity } concentrations of
Lacking a characteristic chromophor and due to less than 5 ng can be detected } the method has
their chemical structure, unconjugated bile acids can- disadvantages. One major problem results from the
not be determined by UV detection with sufRcient fact that taurine-conjugated bile acids cannot be de-
sensitivity under conditions comparable to their con- tected, due to the lack of a carboxyl group necessary
jugated counterparts. One solution to this problem for an ester bond. Samples containing high amounts
might be the addition of an ionic compound like of these conjugates, e.g. gallbladder bile, have to be
hyamine威, tetrabutylammonium phosphate or decyl- treated with choloylglycylhydrolase for deconjuga-
trimethylammonium bromide as counter ion to the tion before the total amount of the free acids can be
mobile phase. Separation and detection are then measured, or an additional method for detection of
based on the formation of ion pairs. Some reports taurine conjugates has to be employed. Though the
suggest that this method increases the sensitivity as BMC reagent will react with all components possess-
well as the selectivity of separation and detection so ing a free carboxyl group, e.g. fatty acids, bile acids
that UV determination of unconjugated bile acids in can be well separated and quantiRed by choosing an
a range of 20}30 ng is possible. In contrast to that, appropriate elution system and adequate sample pre-
detection limits for the conjugated forms are about treatment. The BMC derivatization reaction works
three-fold lower. However, since other detection with dicyclohexanocrown ether (DCCE). Because of
methods provide higher sensitivity and efRciency, ion that, potassium salts of the bile acids have to be
pair chromatography is rarely used in bile acid analysis. prepared so that a further sample preparation step
III / BILE ACIDS / Liquid Chromatography 2133
can be carried out. An alternative, very sensitive

method with a detection limit of less than 1 ng has
been suggested, based on the formation of naphthyl
ester derivatives. This labelling procedure also needs
some careful preparation steps in order to obtain
a good analytical sample for Suorescent detection.
Nevertheless, pre-column Suorescence labelling
seems to be suitable especially for samples containing
low total bile acid concentrations, but also for sam-
ples with unconjugated bile acids as major compo-
nents as in serum (Figure 2).
Detection by mass spectrometry The most sensitive

technique by far for bile acid detection is the coupling
of a suitable HPLC system with a mass spectrometer.
Detection limits range from 1 to 5 pg. However, the
prerequisite for the production of valid and reproduc-
ible data is high quality equipment, resulting in high
initial costs. LC-MS coupling is the method of choice Figure 2 Separation of 10 unconjugated and glycine con-
for the speciRc determination of low bile acid concen- jugated bile acid derivatives (pre-column fluorescence derivatiz-
trations in all biological matrices or tissues. Mass ation with bromomethoxycoumarin) by reversed-phase HPLC and
spectrometry requires very small sample volumes and fluorescence detection: 1, glycoursodeoxycholic acid; 2, glyco-
cholic acid; 3, glycochenodeoxycholic acid; 4, glycodeoxycholic acid;
Sow rates. For the HPLC separation a microsystem
5, ursodeoxycholic acid; 6, cholic acid; 7, glycolithocholic acid; 8,
able to produce constant Sow rates in the range of chenodeoxycholic acid; 9, deoxycholic acid; 10, lithocholic acid.
microlitres per minute is recommended. The use of
standard HPLC systems is possible, but a precise
Sow-splitting device has to be used. This has the the laser, light scattering appears and can be detected.
advantage that the main part of the eluate can be The signal is dependent on the molecular mass of the
utilized for further determinations, e.g. UV detection, separated substance. The ELSD technique seems to be
in order to achieve additional data for quantiRcation. moderately sensitive with respect to UV detection but
A high performance system allowing the baseline detects all bile acids and their derivatives as well as
separation of all bile acids of interest is necessary to other analytes of interest. However, isocratic elution
obtain best results from MS detection. Both isocratic systems are required and, therefore, ELSD detection
and gradient elution are possible, and all kinds of bile is only applicable in niche areas of application.
acids, despite conjugation or other modiRcation, can
Detection of Bile Acids in Biological Samples
be detected. After sample pretreatment and puriRca-
tion to remove interfering substances no further steps Bile The determination of the bile acid pattern in
such as deconjugation, derivatization or enzymatic gallbladder or intestinal bile does not need a time-
treatment are required. consuming sample pretreatment or enrichment. The
Different mass spectrometry systems are reported bile Suid of mammalian individuals consists mainly
to be suitable for LC-MS coupling. HPLC-thermos- of glycine- and taurine-conjugated bile acids in rela-
pray MS methods were reported earlier, but seem to tively high concentrations compared to other body
be less sensitive and reproducible compared to other Suids such as serum or urine samples. Therefore,
methods. Fast atomic bombardment (FAB) MS and HPLC separation with UV detection is the commonly
electrospray ionization (ESI) MS have been interfaced used method for analysis of samples derived from
with HPLC and good results in reproducibility and bile. In the bile Suid of healthy individuals only small
sensitivity have been reported. amounts of protein are present and do not interfere
A special type of a mass detection system is the with bile acid detection. Therefore, no deproteination
evaporative light-scattering mass detector (ELSD). procedure is necessary. However, in order to remove
A special feature of this detection method is that the small amounts of protein as well as cell debris and
eluate coming from the column is evaporated, and larger particles, the addition of methanol followed by
nonvolatile components such as bile acids are deter- a centrifugation step is recommended. For conven-
mined by measurement of the light deSected. The tional analysis, dilution with the mobile phase and
detection unit consists of a photodiode and a laser Rltration through a microRlter is sufRcient as sample
beam. When nonvolatile compounds pass through pretreatment. The use of a guard column in the HPLC
system is recommended when larger sample quantit- are carried out. Bile acids are recovered by elution
ies are to be analysed. with methanol. In most cases, eluted samples are
dried and redissolved in a small volume of the mobile
Faeces Problems in the analysis of bile acids in phase used for HPLC separation. In some reports an
faeces are not caused by the total amount of bile acids additional puriRcation step by anion exchange car-
but by the complexity of the faecal matrix. Since the tridges is described.
primary bile acids in gallbladder bile and in the intes- Despite the fact that this sample preparation
tinum are modiRed by the intestinal microSora during method has been commonly used over a number of
enterohepatic circulation, in faeces mainly secondary years it suffers from several disadvantages. The major
unconjugated bile acids are found. Time-consuming problem is the relatively strong binding of bile acids
extraction procedures have been suggested, e.g. Sox- to serum proteins. Incubation at higher temperatures
hlet extraction or the use of alcoholic alkali solutions. leads to protein denaturation but does not release the
Homogenization of the material in a mixture of bile acid binding sufRciently and reproducibly. An-
chloroform and methanol using an ultraturrax mixer, other approach is the enzymatic digestion of serum
followed by further puriRcation with solid-phase ex- proteins. However, this method seems to be limited,
traction cartridges for sample pretreatment as de- since the products of the protein digestion, small
scribed for serum preparation, seems to be able to peptides and amino acids, interfere with further
extract most faecal bile acids quantitatively. Methods HPLC detection. Recent work describes a deproteina-
for pre-column Suorescence labelling, e.g. with bro- tion procedure by adding serum samples slowly to
momethoxycoumarin derivatives, are reported. For acetonitrile and incubating them in an ultrasonic bath
these derivatization procedures, sample pretreatment for some minutes. This efRciently prevents co-precipi-
with choloylglycylhydrolase for deconjugation is rec- tation of proteins and bile acids and secures reliable
ommended. In addition, detection with an evapor- and validated analytical data. Another point has to be
ative light-scattering detector can be applied, with the considered if solid phase extraction is applied as
advantage that a mass sensitive detection system is a sample pretreatment. C18 column materials are of-
not limited by modiRcations in bile acid structure. fered by several manufacturers and differ in their
However, the method suffers from nonspeciRc matrix quality and their respective features. For instance,
products inSuencing the signal to noise ratio and endcapped and non-endcapped C18 phases are offered
hampering bile acid detection. commercially. These materials can differ in the
quantity and the selectivity of bile acid retention. This
Serum and urinary samples The extraction, detec- leads to remarkable differences in bile acid pattern for
tion and quantiRcation of bile acids in serum samples the sample when different C18 cartridges are used.
of healthy individuals is the most ambitious goal in Therefore, the results obtained can only be validated
the Reld of bile acid analysis. However, since changes when exactly the same material, i.e. same speciRca-
in serum pattern can be characteristic for several tions, batch, etc., is used.
metabolic and cancer diseases, considerable efforts An interesting method for automation of the
have been made to develop methods of high sensitiv- sample pretreatment procedure has been described by
ity and reproducibility as tools for diagnostic pur- Yoshida et al. The online puriRcation, concentration
poses. Due to the nature of the matrix and to total and separation of bile acids in serum samples is
bile acid concentrations in the low micromolar range, performed by an HPLC device equipped with
a sophisticated sample pretreatment procedure is ab- three columns. First samples are injected on to a
solutely necessary. The commonly used method for hydroxyapatite column to remove serum proteins.
extraction and puriRcation of serum samples employs For further puriRcation and concentration, a second
solid-phase extraction as described by Setchell et al. pre-column (Serumout-25威) is used. After washing
Reversed-phase C18 cartridges are activated with and elution from this column the bile acids are injec-
methanol and water as recommended by all manufac- ted via an injection valve to the reversed-phase separ-
turers of these columns. Sodium hydroxide solution is ation column. This strategy of sample pretreatment
added to the serum samples and is followed by might be of value for high throughput assays, e.g. in
a 15}30 min heating step where samples are incu- clinical diagnostics. However, it leaves the problem
bated at 643C to suppress protein binding. The result- of protein binding unsolved and, therefore, lack of
ing solution is passed through the activated C18 car- sensitivity and efRciency is expected.
tridge where the bile acids should be retained while In urinary samples almost the same problems as in
the matrix components are found in the eluted solu- serum bile acid analysis occur. In contrast to
tion. Washing steps with water, in some cases fol- blood samples, the urine from healthy individuals
lowed by nonpolar organic solvents such as n-hexane, does not contain proteins and is available in much
III / BILE COMPOUNDS: THIN-LAYER (PLANAR) CHROMATOGRAPHY 2135
Figure 3 Separation methods for bile acids by liquid chromatography.
higher quantities. For these samples the standard from new column materials and microcolumns (H
puriRcation procedure using C18 solid-phase extrac- 1}2 mm). The future will be in the area of low cost,
tion is suitable. high throughput HPLC-MS devices for routine use.
Due to the low concentration range of the bile acid
See also: II/Chromatography: Liquid: Derivatization.
pattern in serum and urine, the methods of choice for Detectors: Evaporative Light Scattering; Detectors: Flu-
quantitative and reproducible determination are pre- orescence Detection; Detectors: Mass Spectrometry; De-
column Suorescence derivatization and HPLC-MS tectors: Ultraviolet and Visible Detection. III/Bile Acids:
coupling. Gas Chromatography.
Conclusion Further Reading

With regard to the liquid chromatographic methods GuK veli DE and Barry BW (1980) Column and thin-layer
chromatography of cholic, deoxycholic and che-
described for bile acid analysis, especially with regard
nodeoxycholic bile acids and their sodium salts. Journal
to their determination in biological matrices, several of Chromatography 202: 323}331.
approaches for qualitative and quantitative deter- Roda A, Gioacchini AM, Cerre C and Baraldini M (1995)
mination have been described. A critical evaluation of High-performance liquid chromatography-electrospray
the advantages and disadvantages of these methods mass spectrometric analysis of bile acids in biological
results in the conclusion that there are speciRc limita- Suids. Journal of Chromatography B 665: 281}294.
tions for every application (Figure 3). Since cumber- Setchell KD and Worthington J (1982) A rapid method for
some derivatization steps as well as laborious and the quantitative extraction of bile acids and their conju-
unsuitable sample pretreatment play a key role in the gates from serum using commercially available reverse-
value of the analytical results, technologies are phase octadecylsilane bonded silica cartridges. Clinical
needed where direct analysis of the bile acid pattern, Chimica Acta 125: 135}144.
Street JM, Trafford DJH and Makin HLJ (1983) The
especially in serum, is feasible. Only HPLC-MS coup-
quantitative estimation of bile acids and their conjugates
ling presently allows such analysis, but costly equip- in human biological Suids. Journal of Lipid Research
ment and highly skilled personnel requirements limit 24: 491}511.
this method to highly speciRc problems and makes it Yoshida S, Murai H and Hirose S (1988) Rapid and sensi-
unsuitable for routine work. tive on-line puriRcation and high-performance liquid
Presently, HPLC of bile acids is state-of-the-art chromatography assay for bile acids in serum. Journal of
and only incremental improvements can be expected Chromatography 431: 27}36.
BILE COMPOUNDS: THIN-LAYER

K. Saar, Aventis Research & Technology, Introduction
Frankfurt, Germany
S. MuK llner, Enzyme Technology, DuK sseldorf, Germany
Gall bladder bile is, like other physiological Suids,
Copyright ^ 2000 Academic Press of very complex nature. It is produced by liver
2136 III / BILE COMPOUNDS: THIN-LAYER (PLANAR) CHROMATOGRAPHY
hepatocytes of healthy human beings in a volume of

500}1000 mL per day. The major function of the bile
Suid is to support digestion and resorption of lipids
and lipophilic substances during intestinal passage. In
this respect, bile acids and their different conjugates
are the most important components. Bile acids are
steroids of amphophilic nature and, owing to their
chemical structure, tend to form micelles and serve as
physiological detergents. It is this feature that makes
bile acids so important in their role in emulsifying
fatty food components. In addition, bile acids serve as
enzyme cofactors and support fat digestion by activ-
ating lipases and other factors of lipid metabolism.
Humans, together with some other animals, possess
a gall bladder in which bile Suid is stored. The human
gall bladder has a capacity of 15}20 mL of concen-
trated bile. When bile is needed for digestion excre-
tion into the gut is triggered by cholecystokinin,
a gastrointestinal hormone which causes ejection of
gall bladder contents. In the gall bladder bile is con- Figure 1 Thin-layer chromatogram of gall bladder bile from
centrated Rve- to ten-fold, leading to increased con- several animals, and of human T-drainage bile; comparison with
centrations of all major bile components. Total bile selected standard bile acids. C, cholic acid; DC, deoxycholic acid;
production as well as bile composition are highly CDC, Chenodeoxycholic acid; GC, glycocholic acid; GCDC,
glycochenodeoxycholic acid; TC, taurocholic acid; TCDC,
dependent on amount and composition of food. For taurochenodeoxycholic acid; TUDC, tauroursodeoxycholic acid;
example, according to differences in eating, the pH GUDU, glycoursodeoxycholic acid; LC, lithocholic acid; and GLC,
values of human bile range from 6.5 to 8.5. Other glycolithocholic acid. (From MuK ller et al . 1992.)
animals, such as rats, secrete bile Suid directly into
the intestine.
The most important bile components are the bile chromatography (HPLC) and thin-layer chromatog-
acids and their different amino acid conjugates. Bile raphy (TLC), have been described for the analysis of
acids are synthesized in the liver from cholesterol and pharmacologically interesting substances and their
therefore play an additional key role in cholesterol metabolites in bile, but as the analytical conditions
homeostasis. Products of this process are the so-called are always dependent on the chemical nature of the
primary bile acids cholic acid and chenodeoxycholic drug and its concentration, they will not be con-
acid. In the bile Suid mainly glycine and taurine sidered here. Another bile component, cholesterol,
conjugates of these primary bile acids as well as those has become increasingly important during recent dec-
of deoxycholic acid are present. However, after secre- ades. The role of increased serum cholesterol concen-
tion into the gut primary bile acids are modiRed by trations in diseases such as atherosclerosis, leading to
the intestinal microSora. Primary bile acids and all stroke and cardiac infarction, for example, is the
their derivatives produced during intestinal passage subject of debate, but when added to other risk fac-
undergo enterohepatic circulation, by which bile tors its involvement is undisputed. Cholesterol con-
acids are reabsorbed nearly completely and are trans- centrations in gall bladder bile range from 1 to
ported back to the liver. Only a very small part of the 15 g L\1 in humans. Increased bile cholesterol con-
bile acid pool, 2}8%, is excreted with the faeces. centrations in most cases cause the formation of
Nevertheless, the bile of healthy individuals contains gallstones through cholesterol crystallization. The
more primary than secondary bile acids. As changes development of drugs inSuencing either serum or bile
in total bile acid amount as well as in the bile acid cholesterol concentrations is one of the most impor-
pattern are of importance for diagnostic purposes, tant areas in pharmaceutical research. Therefore, also
simple and fast methods for the analysis of these in this Reld fast and simple methods for the detection
biological compounds are required (Figure 1). and quantiRcation of cholesterol in bile are necessary
Many drugs and drug metabolites are secreted into (Figure 2).
the bile, and this, in addition, makes bile analysis Bile Suid also contains high concentrations of
important for the understanding and controlling of phospholipids, especially lecithin, in amounts of
pharmacokinetic mechanisms during drug develop- 2}4 g L\1. However, whether increases or decreases
ment. Several methods using high performance liquid in phospholipid concentration are signiRcant for
monly used for bile acid analysis, TLC is a very

simple and nearly universal method. No highly soph-
isticated and expensive instrumentation is needed for
reliable semiquantitative analysis. If precise quantiR-
cation is essential, a TLC scanner equipped with
a computer software system for fast and easy data
registration and evaluation is recommended. Where-
as HPLC analysis is limited by the fact that only one
sample can be analysed at a time, TLC allows multi-
parallel analysis of probes in combination with an
almost unlimited number of detection and visualiz-
ation methods.
Several solvent systems for the separation of bile
acids in biological Suids are described. With respect
to bile, bile acid concentrations are in the millimolar
range and therefore no detection problems occur,
which makes TLC well suited for bile and bile acid
analysis. In addition, time-consuming sample pre-
treatment procedures can be generally avoided for
standard analysis of bile acids. However, dilution
of the bile samples with phosphate-buffered saline
increases the sensitivity of the selected detection
(Figure 4).
For bile acid pattern analysis a satisfactory separ-
ation of the main components, the glycine- and
taurine-conjugated bile acids, is necessary. In general,
silica gel TLC plates are used for routine analysis
since high quality materials are commercially avail-
able in several sizes and from several suppliers and, in
contrast to reversed-phase plates for example, at
a relatively low price. Our experience in this Reld
leads us to suggest the use of silica gel pre-coated on
glass plates with a concentration zone to improve
Figure 2 Composition of (A) human T-drainage bile (water con- separation, bile acid band shape and overall resolu-
tent 95.5%) and (B) bovine gall bladder bile (water content tion. Solvent systems suitable for the separation of
81.5%). (From MuK ller et al . 1992.) bile acids on silica gel are shown in Table 1. Many
methods for bile acid separation have been described
differential diagnosis of certain enterohepatic dis- in the literature, and only a selection are shown here.
eases is not known. Nevertheless, the analysis of System 2 leads to the most satisfactory results in
phospholipid pools and pattern in bile can be of value resolution and band shape. With this system the
for metabolism and kinetic studies in drug develop- glycine and taurine conjugates of deoxycholic acid
ment. Several TLC-based methods for separation and and of chenodeoxycholic acid can be separated with
detection are known. Other bile components, such as good resolution. The TLC plates have to be de-
bile pigments, proteins (e.g. alkaline phosphatase) veloped six times in the same solvent mixture for bile
and electrolytes, have to be analysed by other analyti- acid separation and this requires more time compared
cal methods (e.g. enzyme assays or spectrometric with the other solvent systems described, but if quan-
methods). In general, biliary protein concentration is tiRcation by TLC scanning is followed system 2 is the
negligible in healthy individuals (Figure 3). method of choice. Solvent system 3 provides a simple
and fast method for the separation of conjugated and
unconjugated bile acids with sufRcient overall resolu-
Thin-Layer Chromatography tion to allow quantiRcation. The advantage of this
system is the simultaneous separation of cholesterol,
Bile Acid Separation
which migrates faster than the bile acids. Chamber
In contrast to other analytical technologies such as saturation before analysis increases resolution and
HPLC and gas chromatography, which are also com- improves band shape. With system 4, a solvent
Figure 3 Bile acid patterns obtained from three samples of human T-drainage bile and from the bile of other animals. Key as for
Figure 1, plus GDC, glycodeoxycholic acid and TDC, taurodeoxycholic acid. (From MuK ller et al . 1992.)
mixture for the separation of conjugated bile acids on In all the separation systems described temperature
reversed-phase TLC plates is provided. TLC condi- plays an important role in resolution.
tions are similar to an HPLC method described for Although room temperature (20}223C) allows sat-
the separation of conjugated bile acids, and may be isfactory separation employing the systems described,
helpful in comparison of sample analysis by different in some cases decreasing the temperature improves
analytical techniques or in cases where simultaneous separation and band shape. Depending on the sample
separation of bile acids and drug metabolites is concentration and composition the inSuence of tem-
required. perature on separation resolution should be taken
agent for a vast amount of biological substances is

anise aldehyde dissolved in sulfuric acid. Spraying or
dipping of the developed TLC plates followed by
incubation at 1253C generates spots of different col-
our which are stable for some time. However, the low
speciRcity of this reagent may cause problems in the
analysis of samples with a high content of other
components or analytes. Spots can also be visualized
under ultraviolet (UV) light, which increases sensitiv-
ity. Individual bile acids can be identiRed by their
different colours (Figure 5).
Molybdatophosphoric acid is another reagent of-
ten used for detection of biological components and
forms blue bands on a yellow ground. Since colour
intensity of the visualized spots is not stable over
a long period of time, quantiRcation must follow
Figure 4 (A) Thin-layer chromatogram of increasing amounts immediately after development of the colour.
(1, 2 and 4 L of 4 mg mL\1 solutions) of bile acid standards TC, The sensitivity of detection of bile acids on TLC
GC, TLC, GCDC, C, GLC, UDC and LC. (B) Fluorescence scan of plates can be increased dramatically by using a re-
the left lane of the chromatogram. Key as for Figure 1. (From
agent system which forms Suorescent bands. HClO4
MuK ller et al . 1992.)
has been described as giving reproducible Suorescent
spots with bile acids on TLC plates, provided a 5%
into account as a critical factor for optimization of solution of this derivatizing agent is used and UV
the chosen solvent system. light of 365 nm excitation wavelength is employed
Numerous detection methods for bile acid visualiz- after spraying and appropriate treatment at higher
ation on TLC plates after separation have been de- temperature.
scribed and two principally different systems are in According to our experience a reagent mixture of
use. Detection methods using reagents which form manganese dichloride and sulfuric acid is preferred
coloured complexes have the advantage that bands for detection, owing to the reproducible results and
can be visualized directly. A universal detection re- its convenience in use. Plates are dipped in the reagent
Table 1 Solvent systems for bile acid separation
System 1
Unconjugated bile acids Isooctane Room temperature
Diethyl ether
n-Butanol
Acetic acid
Mixture 10#2.5#1#1 (v/v)
System 2
Glycine- and taurine-conjugated bile acids Chloroform Room temperature
2-Propanol 2;15 cm followed by
2-Butanol 4;7 cm
Acetic acid
Double-distilled water
Mixture 30#20#10#2#1 (v/v)
System 3
All bile acids Chloroform Room temperature
Methanol Chamber saturation
Acetic acid
Mixture 85#20#9 (v/v)
System 4
Glycine- and taurine-conjugated bile acids Acetonitrile (90%) Room temperature
0.01 M Ammonium carbamate Separation on RP8 plates
pH 7.3
Mixture 55#45 (v/v)
Figure 5 Fluorescence scans of bile specimens from Figure 1: (A) human bile II; (B) human bile I; (C) rabbit bile; (D) pike bile;
(E) monkey bile; (F) carp bile; (G) ox bile. Key as for Figure 1. (From MuK ller et al . 1992.)
and heated to 1103C for 10}15 min. Bile acids react ditions is 2}5 ng and is therefore about 1000-fold
with the reagent with the formation of yellow or higher compared to visible light (Figure 4).
brown coloured spots. Under UV light (365 nm) blue
Cholesterol
or yellow Suorescing bands can be detected. Bile acids
can be differentiated by their respective colour. The Cholesterol and cholesterol esters are important
detection sensitivity for the bile acids under UV con- bile components. Several solvent systems for the
Table 2 Solvent systems for separation of cholesterol and cho- terials or mixtures of puriRed standard compounds.
lesterol esters on silica gel plates Due to the fact that the most abundant phospholipid
in bile samples is lecithin, three solvent systems
System 1 Room temperature
Chloroform Chamber saturation able to separate phospatidylcholine, phosphatidyl-
Methanol ethanolamine, phosphatidylserine and phosphatidyl-
Acetic acid glycerol are shown in Table 3.
Mixture 85#20#9 (v/v) Some of the most commonly used and versatile
detection reagents described in the sections above are
System 2 Room temperature also suitable for phospholipid detection both under
Chloroform UV and visible light. Detection with the 2,7-diSuores-
Acetone
cein reagent increases sensitivity and allows reliable
Mixture 85#15 (v/v) and satisfactory quantiRcation. Detection by a re-
agent mixture containing ammonium molybdate, sul-
System 3 Room temperature
Cyclohexane furic acid and ascorbic acid results in blue bands
Diethyl ether which can be easily detected and quantiRed by a TLC
Mixture 50#50 (v/v)
scanning device at a wavelength of 620 nm.
System 4 Multiple plate development

Methanol
Conclusion
Diethyl ether For decades, TLC has proven its advantages as a fast,
n-Hexane
inexpensive, reliable as well as highly reproducible
technology in the analysis of bile, bile acids and bile
acid derivatives. Especially in bile acid and choles-
separation of these steroids exist, but only some of terol analysis several separation and detection
them are suitable for the analysis of cholesterol in systems have been developed which have shown con-
complex biological Suids. In Table 2 four commonly vincing sensitivity and overall resolution.
used methods are shown. The mixture of chloroform Solvent systems allowing the simultaneous detec-
and acetone is described for the detection of choles- tion of bile acids, cholesterol and drug metabolites in
terol in serum samples and can successfully be a one-step analysis give a signiRcant time advantage
adapted to bile analysis. In system 4 three organic at reduced cost. In addition, the separation system
solvents with decreasing polarity are suggested for and overall conditions can easily be adapted to the
three consecutive development steps and can be used respective analytical problem in most cases.
in automated multiple development devices. Although the quantitative detection of phos-
If simple and fast one-step analysis of cholesterol pholipids in bile does not play a key role in bile
and bile acids is required, solvent system 1 combined
with the above described manganese detection re-
Table 3 Solvent systems for phospholipid separation on silica
agent should be used, since cholesterol can be separ- gel plates
ated and identiRed easily by fast migration and its
orange Suorescence. Other staining reagents contain- System 1
ing either trichloracetic acid or 8-anilinonaphthaline- Chloroform Room temperature
Methanol
sulfonic acid ammonium salt have been shown to
Ammonia (conc.)
react with cholesterol to Suorescent bands, and nu-
merous other universal detection systems are de- Mixture 60#35#3 (v/v)
scribed in the literature. Adaptation to bile analysis System 2
seems to be possible in most cases. Petroleum ether Room temperature
Use of the NCS reagent (1,2-naphthochinone-2- Chloroform
sulfonic acid sodium salt) leads to purple bands Methanol
which can be easily detected and quantiRed with Acetic acid
detection limit at 5 ng cholesterol. Mixture 30#50#15#10 (v/v)
Phospholipids System 3
Cyclohexane Room temperature
As for the other bile components, several solvent Isopropanol
systems for separation of phospholipids are known. Double-distilled water
However, most of them were not developed for the Mixture 30#40#6 (v/v)
analysis of bile samples, but for other biological ma-
2142 III / BIOANALYTICAL APPLICATIONS: SOLID^PHASE EXTRACTION
analysis at the moment, there is still a need to develop bolic Disorders: Detection: Thin-Layer (Planar)
accurate TLC methods for detailed separation of Chromatography.
these lipids. Further investigations on the use of new
TLC materials in the quantitative analysis of bile Further Reading
components are needed, as well as the adaptation of
TLC methods to automated devices. Nevertheless, Jork H, Funk W, Fischer W and Wimmer H (eds) (1989)
compared to other analytical tools TLC is the method Du( nnschichtchromatographie, vol. 1a. Weinheim: VHC
of choice for fast routine use. Verlagsgesellschaft.
MuK llner S, Hofmann R, Saar K and Karbe-ThoK nges B
See also: II/Chromatography: Thin-Layer (Planar): (1992) Economic assay for the evaluation of bile acid
Densitometry; Layers; Mass Spectrometry; Spray Re- sequestrants using ox bile and quantitative TLC analy-
agents. III/Bile Acids: Liquid Chromatography; Gas sis. Journal of Planar Chromatography 5: 408}416.
Chromatography. Clinical Diagnosis: Chromatography. Rivas-Nass A and MuK llner S (1994) The inSuence of critical
Lipids: Liquid Chromatography; Gas Chromatography; parameters on the TLC separation of bile acids. Journal
Thin-Layer (Planar) Chromatrography. Neonatal Meta- of Planar Chromatography 7: 276}285.
BIOANALYTICAL APPLICATIONS:
SOLID-PHASE EXTRACTION
D. A. Wells, Sample Prep Solutions Company, examined, contrasting the use of high-pressure liquid
Maplewood, MN, USA chromatography (HPLC) in the clinical setting and
Copyright ^ 2000 Academic Press the rapid proliferation of HPLC combined with
tandem mass spectrometry (liquid chromatogra-
phy/mass spectrometry/mass spectrometry or
Introduction LC/MS/MS) for speciRc and sensitive detection of
drugs for pharmaceutical bioanalysis.
The quantiRcation of drugs and metabolites in bio-
logical Suids (e.g. plasma, serum and urine) is one of
the most active research Relds in clinical and pharma- Drug Sample Preparation Techniques
ceutical analysis. Drug bioanalysis is important in A reliable analytical method is achieved with efRcient
clinical chemistry to demonstrate optimal drug sample preparation, adequate chromatographic sep-
therapy, because plasma drug concentrations relate to aration, and a sensitive detection technique. Detailed
the therapeutic or toxic effects of a drug. Knowledge and exact guidelines exist for the validation of bi-
of the plasma drug concentration is used to determine oanalytical methods, which meet requirements
why a patient does not respond to drug therapy or agreed to by the Food and Drug Administration. The
why a drug causes an adverse effect. Dosage adjust- sample preparation step is an important component
ment by the physician is then warranted. Drug bi- of each overall bioanalytical method, as it
oanalysis is important in pharmaceutical research to
determine the pharmacokinetics and metabolic bio- E often concentrates an analyte to improve its limits
transformation of new drug molecules. It is a tech- of detection,
nique used throughout the development of all new E removes unwanted matrix components that can
drugs. In particular, during drug discovery, bioanaly- cause interferences upon analysis, thus improving
sis yields essential data that is used in the decision- method speciRcity, and
making process of whether or not a new molecule E frees the analyte from matrix components so that it
should be a candidate for further development. can be placed into a solvent suitable for injection
This chapter discusses the utility of solid-phase into the chromatographic system.
extraction (SPE) in comparison with other drug
Liquid/liquid extraction
sample preparation techniques for bioanalysis. Sev-
eral applications of SPE will be summarized in the A common sample preparation procedure used to
clinical setting for therapeutic drug monitoring, isolate drug analytes from biological matrices is
and in the pharmaceutical research setting for drug liquid/liquid extraction (LLE). It is quite effective at
discovery and development. The separation and removing salts and proteins; water-soluble endogen-
detection techniques used for bioanalysis will be ous components often remain in the aqueous phase.
III / BIOANALYTICAL APPLICATIONS: SOLID^PHASE EXTRACTION 2143
The organic phase containing the extracted analyte is the exclusion of other compounds through a selec-
isolated, evaporated to dryness, and reconstituted in tive wash step. Elution of the analyte is achieved with
liquid chromatographic mobile phase for analysis. an organic solvent that disrupts the attraction to the
One of the beneRts of LLE is that with proper selec- solid sorbent. Solvent exchange is followed by analy-
tion of solvent and pH, very clean extracts can be sis on a chromatographic system. The traditional for-
obtained with good selectivity for the target analytes. mat for SPE has been single disposable columns and
However, the disadvantages of LLE are that cartridges. Its advantages are that multiple samples
can be prepared in parallel, low volumes of solvents
E it is a very labour intensive procedure, are used and procedures can be readily automated.
E it requires large volumes of organic solvents which The technology has been improved in recent years
can be expensive to purchase and dispose, with the introduction of more selective solid sorbent
E exposure of personnel to these solvents can be chemistries, disc-based SPE devices, smaller bed
hazardous to health, mass sorbent loading, on-line SPE techniques
E it cannot easily be automated, and introduction of the 96-well plate format for
E emulsions have been demonstrated to occur, improved productivity.
E evaporative losses can occur upon dry-down with An on-line SPE technique has recently shown great
volatile analytes. utility. A commercial device (ProspektTM (Spark Hol-
land)) combines an autosampler and a solvent deliv-
Despite its drawbacks, LLE continues to be used
ery unit to aliquot liquid samples into the Sow path of
for drug bioanalysis when there is adequate labour
solvent. An SPE cartridge is preconditioned and is
and its associated costs are not prohibitive, and the
in-line with the solvent Sow. The cartridge retains
sample throughput can be adequately met.
target analytes while a weak solvent elutes unretained
Protein precipitation salts and polar matrix components. An optimized
sequence of solvents, each with increasing solvent
A fast and simple method of sample preparation is strength, is used to wash out weakly retained compo-
protein precipitation, also referred to as ‘dilute and nents. A Rnal elution with LC mobile phase elutes the
shoot’. This nonselective technique involves adding analytes of interest from the SPE cartridge and onto
a water-miscible organic solvent (e.g. acetonitrile) an analytical column for chromatographic separation
or inorganic acid (e.g. trichloroacetic acid, 10%) to followed by detection.
the biological matrix (usually in a 3 : 1 or 4 : 1 ratio, The autosampler within this device can be pre-
v/v), centrifuging or Rltering to remove precipitated loaded with up to 160 samples and the entire tray can
proteins and injecting an aliquot of the diluted be analysed in an automated procedure. Its advant-
supernatant. This technique is often performed in ages are: unattended sample prep and analysis, mini-
pharmaceutical drug discovery laboratories as the mized adsorptive losses, since sample transfers are
Rrst attempt to prepare samples for bioanalysis. not performed as in off-line techniques, and trace
Satisfactory analyses have been demonstrated with enrichment of analyte occurs. Some disadvantages are
this rapid sample preparation approach, but it has that analysis is serial (although a sample is always
disadvantages. This technique dilutes the sample by being analysed while another is being extracted and
a factor of four or Rve, so it is useful only when prepared for injection) and sample stability may be an
sample levels are relatively high, typically in the low issue for some drugs as a result of extended storage
g mL\1 range. Also, matrix components are not times in the autosampler.
efRciently removed, and thus may co-elute with the
analyte in the isolated supernatant or Rltrate. When
present, these contaminants have been shown to in- Therapeutic Drug Monitoring
terfere with detection techniques and lower the signal Therapeutic drug monitoring (TDM) is an established
for the analyte of interest. This approach thus lacks clinical specialty in which laboratory specialists
selectivity and problems can arise from column foul- quantify drug concentrations for the purpose of
ing since the precipitation efRciency is not complete. evaluating therapeutic response. Examples of drugs
that are frequently subjected to TDM are antibiotics,
Solid-phase extraction
antiarrhythmics, antiasthmatics, antidepressants,
Solid-phase extraction has, during the last 18 years, antiepileptics, and antineoplastics. These drugs pos-
become recognized as a preferred technique for ex- sess a narrow range of therapeutic and safe plasma
tracting drug analytes from complex bioSuids using concentration. Therapeutic index (TI) is deRned
adsorption chemistry. The attraction of the analyte as the ratio between the maximum and minimum
for a solid phase adsorbent (‘sorbent’) is exploited to plasma concentrations of the drug’s therapeutic
2144 III / BIOANALYTICAL APPLICATIONS: SOLID^PHASE EXTRACTION
range. A narrow range is deRned as a ratio of comes primary, have also been used successfully.
2 to 3. A TI below 2 infers that the dose that yields Methods for these drugs typically involve a solvent
a subtherapeutic response is close to the dose that exchange of organic eluent for aqueous/organic mo-
produces toxicity. Most drugs have a TI of greater bile phase. Evaporative losses are always a concern
than 2. with this step, but can be avoided by use of SPE discs,
The preferred technique for drug analysis is use in which elution is accomplished with a small volume
of an immunoassay analyser (e.g. Suorescence polar- of mobile phase solution.
ization immunoassay, FPIA) because it is relatively
Corticosteroids
quick, can be automated, and requires minimal tech-
nician training for execution. However, newer drugs The measurement of steroids (prednisone, cortisone,
requiring plasma quantiRcation often do not have an prednisolone, cortisol, corticosterone, methylpredni-
immunoassay developed, so chromatography is al- solone) in blood is often inaccurate owing to interfer-
most always used at Rrst. Drugs whose metabolites ence from sample matrix and cross-reactivity with
play a role in their efRcacy and clinical interpretation chemically similar steroids. For example, antiserum
also need to be determined by chromatography, for cortisol is nonspeciRc and cannot differentiate
which analyses multiple components simultaneously. between cortisol, its metabolites and therapeutically
Immunoassays are often not selective enough to dis- administered steroids. Again, SPE is a preferred tech-
tinguish between parent drug and metabolite, and nique for drug sample preparation. An efRcient
interferences can adversely affect results for some method has been reported using C8 sorbent discs,
drugs. Another difRculty facing clinical laboratories which allows for elution in mobile phase compatible
is that more accurate quantitation and detection resolution for direct injection, eliminating the need for
quirements must be satisRed owing to lower dosages a tedious dry-down and reconstitution step. Steroids
being administered. The proliferation of new drugs are released from proteins by incubation at room
also increases the potential for concomitant adminis- temperature with a HCl solution. Neutralization is
tration. When the issue of cost is considered, in addi- accomplished by addition of a sodium borate solu-
tion to the drawbacks listed above for immunoassays, tion. Following centrifugation, the supernatant is
chromatographic techniques can become quite at- loaded onto conditioned C8 discs. A dilute meth-
tractive. Most often for clinical bioanalysis, detection anol}water solution acts as an efRcient wash to re-
methods involve ultraviolet or Suorescence detection move adsorbed proteins. Elution is performed with
coupled with liquid chromatographic separation. acetonitrile, followed by water. The resulting mixture
is compatible with mobile phase for direct injection.
Antidepressants
Cyclosporin
Tricyclic and newer antidepressant drugs are one of
the most frequently monitored classes of therapeutic Cyclosporin, an immunosupressant drug, has many
drugs in the clinical setting. Drug concentrations are metabolites and is commonly monitored for drug
monitored in patients for compliance and to ensure concentrations in the blood of patients who have
that therapeutic blood levels are reached. Also, pa- received an organ transplant. Monoclonal antibody-
tients sometimes take multiple antidepressant drugs, based immunoassays are in use for this assay, as
often from different physicians, which can be deter- well as LC methods. Technology for immunoassay
mined from a LC analysis. Immunoassays frequently detection is constantly being improved, and the
cross-react with these drugs (e.g., imipramine, amit- tests in use now are reliable. However, LC methods
riptyline, desipramine and Suoxetine) and their also function quite well, and the issue of anti-
metabolites. SPE has been used in some immunoassay body versus LC method often rests with cost analy-
kits to measure speciRcally one tricyclic drug. Over- sis for a clinic. LC methods rely on SPE using
all, LC is advantageous because of its ability to moni- whole blood and, when coupled with automat-
tor simultaneously multiple drugs and to resolve po- ion, can be quite cost effective. An advantage of LC
tential interferences from concomitantly adminis- is that it can simultaneously measure several metab-
tered drugs. olites.
Solid-phase extraction methods for antidepressant The extraction procedures for cyclosporin are
drugs abound in the literature. These drugs are basic commonly reversed phase. Note that whole blood
and can be adsorbed to reversed-phase sorbents such is preferred to avoid temperature-dependent cyclo-
as C18 and C8 by both reversed-phase attraction and sporin redistribution. Mixing with acetonitrile}water
secondary interactions via cationic adsorption to haemolyses blood, and aliquots of the supernatant
silanols on the silica surface. Polar sorbents such as are loaded onto conditioned extraction columns.
cyanopropyl, in which the cationic adsorption be- Wash steps may involve a weak concentration of
III / BIOANALYTICAL APPLICATIONS: SOLID^PHASE EXTRACTION 2145
acetonitrile in water and elution is achieved with meth- An application illustrating efRcient sample pre-
anol or ethanol, or with an alcohol}water solution. paration required prior to LC/MS/MS analysis is the
determination of leukotriene LTE4 in urine. Leuko-
Antiepileptics
trienes are biologically important molecules derived
Plasma concentrations of antiepileptic drugs are often from arachidonic acid by the action of the enzyme
monitored during therapy, since a therapeutic range 5-lipoxygenase. LTE4 is difRcult to analyse because it
has been well deRned. These drugs include phenytoin is unstable under a variety of conditions. The unique
and carbamazepine and their metabolites, pheno- capabilities of MS have yielded detailed structural
barbital and newer agents such as lamotrigine. and metabolic information about these compounds.
Solid-phase extraction is commonly performed using The extraction of LTE4 from 5 mL human urine is
reversed-phase sorbents such as C8 and C18. The accomplished by pH adjustment to 4.5 before loading
wash is performed with water, since the more hy- onto a conditioned C18 SPE column or disc. A wash
drophilic drug lamotrigine is removed from the of 5% formic acid is used, followed by elution with
sorbent bed at low organic concentrations in water. a small volume of methanol. The eluate is evaporated
Elution is efRciently accomplished with acetonitrile. and then reconstituted in mobile phase for analysis by
Again, the use of the disc SPE formats can allow LC/MS/MS. This extraction method is simple, yet
elution in volumes small enough to eliminate the selective for LTE4 in human urine. Concentration
evaporation step; a small elution volume of acetonit- range tested was 50 pg mL\1 to 10 ng mL\1.
rile is mixed with water and the resulting solution is A recent advance in improving the throughput of
injected directly onto the liquid chromatograph. drug development, made possible by LC/MS/MS
techniques, is the simultaneous determination of mix-
Pharmaceutical Drug Discovery tures of drug candidates in single analytical samples
as a means to select optimal target drugs. In this case,
and Development animals are dosed with several test compounds at
The discovery and development of new drug entities once; pooled plasma from multiple animals dosed with
can be a perplexing task for analytical chemists. Drug single compounds also has been shown. The speciRcity
molecules that demonstrate activity in receptor assays and sensitivity of the MS detection techniques now
may exhibit structural characteristics that make them available have made this advancement possible. The
poor candidates for absorption in vivo, and other data generated by this approach have been reported
times they may be so rapidly metabolized as to limit to yield meaningful pharmacokinetic data.
their duration of activity in the body. Drugs may be
difRcult to separate on chromatographic columns, High Throughput Applications of SPE
or be so labile that analytical techniques become
a challenge. Once these challenges are overcome, the
in Bioanalysis
determination of drug concentrations in blood and The introduction and utilization of liquid chromato-
urine yields the data used to understand the time graphy interfaced with tandem MS techniques has
course of drug action, or pharmacokinetics, in ani- resulted in a dramatic change in sample prepara-
mals and man. Modern requirements for bioanaly- tion techniques for drug bioanalysis. The speed of
tical assays include speciRcity to determine parent LC/MS/MS, in which run times are typically 1}3 min,
drug from metabolites, sensitivity to detect concen- allows more samples to be analysed per unit time
trations of ng mL\1 and often lower, and speed. than traditional HPLC analysis techniques, which
The mainstay for detection following chromato- require about 10}30 min per sample. The ability of
graphic separation was formerly ultraviolet or Suor- the instrumentation to analyse samples faster than
escent detection. These techniques have served the ever before, combined with the emergence of combi-
analytical needs well over the years, but newer ana- natorial chemistry techniques for drug discovery,
lytical instrumentation, namely the maturing of MS have put more drugs into the drug development pipe-
interfaces to LC, has allowed analytical chemists to line. The greater number of drugs under evaluation
use a more powerful detection technique for their places increased demands on the resources available
routine analyses. The development of more potent for pharmacokinetic drug metabolism support. As
drugs has also placed greater emphasis on the deter- a result, sample preparation has become the rate-limit-
mination of lower concentrations of drugs, down ing step in achieving higher throughput in bioanalysis.
to the pg mL\1 range. The answer to the analytical The pharmaceutical industry has responded to the
need for greater sensitivity, speciRcity and speed has challenge of higher throughput sample preparation
been LC coupled with tandem mass-spectrometry by using a recent advance, SPE in a 96-well microtitre
(LC/MS/MS). plate format. This technique utilizes single blocks or
2146 III / BIOGENIC AMINES: GAS CHROMATOGRAPHY
plates that have 96 wells containing discs or packed See also: II / Extraction: Solid-Phase Extraction.
beds of sorbent particles arranged in an 8-row;12- III / Drugs and Metabolites: Liquid Chromatography }
column rectangular matrix. Although the plates can Mass Spectrometry. Solid-Phase Extraction with Discs.
be processed manually, instrumentation is preferred
for processing liquids through the SPE plates. The Further Reading
advent of high throughput sample preparation for-
mats, in combination with the speciRcity, sensitivity Brewer E and Henion J (1998) Minireview. Atmospheric
and speed of LC/MS/MS analytical techniques, have pressure ionization LC/MS/MS techniques for drug dis-
created a superior combination for the analytical position studies. Journal of Pharmaceutical Science
chemist to meet the demands for faster sample pro- 87(4): 395}402.
Hartmann C, Smeyers-Verbeke J, Massart DL and
cessing and data generation to support the drug devel-
McDowall RD (1998) Review: Validation of bioana-
opment process. lytical chromatographic methods. Journal of Pharma-
ceutical and Biomedical Analysis 17: 193}218.
Conclusion Hoja H, Marquet P, Verneuil B et al. (1997) Applications of
liquid chromatography-mass spectrometry in analytical
The future of drug bioanalysis using SPE is promising toxicology: a review. Journal of Analytical Toxicology
as more and more laboratories increase the usage of 21: 116}126.
LC/MS/MS instrumentation and justify the need for Lensmeyer GL, Darcey BA and Wiebe DA (1991) Application
high throughput SPE. LC/MS/MS instrumentation of a novel form of solid phase sorbent (Empore membrane)
is now common in pharmaceutical laboratories and to the isolation of tricyclic antidepressants from blood.
clinical laboratories are slowly beginning to demon- Journal of Chromatographic Science 29: 444}449.
Lensmeyer GL, Onsager C, Carlson IH and Wiebe DA
strate utility for mass spectrometry for certain drug
(1995) Use of particle-loaded membranes to extract ster-
classes, e.g. tacrolimus (FK506, an immunosuppres- oids for HPLC analyses. Improved analyte stability and
sant). The 96-well plate format is Rnding its way into detection. Journal of Chromatography A 691: 239}246.
many bioanalytical applications; when coupled with McDowall RD, Doyle E, Murkitt GS and Picot VS (1989)
automation, there is currently no faster sample prep- Review: Sample preparation for the HPLC analysis of
aration method. Individual SPE cartridges will con- drugs in biological Suids. Journal of Pharmaceutical and
tinue to have a role in sample preparation, but Biomedical Analysis 7(9): 1087}1096.
96-well plates will proliferate in pharmaceutical ap- Wells DA (1999) 96-Well plate products for solid-phase
plications. More bioanalytical applications will adopt extraction. LC/GC 17(7): 600}610.
smaller sorbent mass products, as the productivity Wong SHY (1989) Review: Advances in liquid chromatog-
gains from reduced solvent volume are realized, espe- raphy and related methodologies for therapeutic drug
monitoring. Journal of Pharmaceutical Biomedical
cially using automation. As the sorbent mass in plates
Analysis 7(9): 1011}1032.
decreases, the number of applications that demon- Wu Y, Lily Y-T, Henion JD and Krol GJ (1996) Determina-
strate elimination of the evaporation step by using tion of LTE4 in human urine by liquid chromatography
small elution volumes of mobile phase compatible coupled with ionspray tandem mass spectrometry. Jour-
solution for direct injection will increase. nal of Mass Spectrometry 31: 987}993.
BIOGENIC AMINES:
GAS CHROMATOGRAPHY
R. Draisci, Laboratorio di Medicina Veterinaria, weight organic bases, produced by the decarboxyla-
Istituto Superiore di Sanita% , Roma, Italy tion of amino acids, that possess biological activity.
P. L. Buldini, CNR Lamel, Bologna, Italy Biogenic amines are receiving increasing interest be-
S. Cavalli, Laboratorio Applicazioni, Dionex Milano, cause they are often present in foods, such as cheese,
Italy
meat, and Rsh where they are used as a useful indi-
Copyright 2000 Academic Press
cator of spoilage and markers of food quality. They
occur naturally in the central nervous system, where
they play an important role as neurotransmitters.
Introduction Their presence in metabolic pathways in health and
The term ‘biogenic amine’ was proposed by Guggen- disease has been studied because of their biological
heim in 1940 in order to deRne the low-molecular- activity as reported by Parvez et al., and they are
III / BIOGENIC AMINES: GAS CHROMATOGRAPHY 2147
Table 1 GC of biogenic amines with FID detection

Analyte Chromato- Carrier, Injector Matrix Sample treatment Reference
graphic type and temper-
column flow rate ature and
program
Aliphatic monoamines, Dual fused He 2603C, Biological fluids Two-phase isobutyloxy Kim et al.
aliphatic polyamines, silica 2603C carbonylation, SPE (1997)
catecholamines, indolyl capillary P43C min\1 purification and deri-
amines, imidazoyl amines DB-5 DB-17, P2803C vatization with N-methyl-
(57 amines) length 30 m, N-(butyldimethylsilyl)-
i.d. 0.25 mm, trifluoroacetamide
film thickness
0.11 m
Histamine, tyramine, Fused silica He, 50 ml 3103C, Foods Liquid}liquid extraction, Laleye et al.
putrescine, cadaverine, capillary min\1 1253CP SPE purification, derivati- (1987)
adrenaline, noradrenaline, SE-54, length 103C min\1 zation with trifluoroacetic
tryptamine, dopamine, 25 m, i.d. P3003C anhydride
agmatine, spermine, 0.31 mm, film
spermidine, thickness
phenylethylamine 0.52 m
Cadaverine, 1,3-diamino- Fused silica He, 0.5 ml 3003C, Biological fluids SPE purification, Muskiet
propane, putreanine, capillary min\1 1203CP derivatization with et al. (1984)
isoputreanine, putrescine, cross-linked 73C min\1 heptafluorobutyric
spermine, spermidine methyl P2603C anhydride
siloxane
deactivated
by silanization,
length 35 m,
i.d. 0.20 mm,
i.d. 0.20 mm,
i.d. 0.20 mm,
film thickness
0.11 m
1,3-Diaminopropane, 0.5% KT-300 N2, 2853C, Plants Ion exchange resin Yamamoto
putrescine, cadaverine, on Uniport 80 ml min\1 1303CP separation, derivatization et al. (1984)
3-aminopropylcadaverine, HP packing, 103C min\1 with ethylchloroformate
spermidine, sym-homo- length 0.5 m, P2803C
spermidine, sym- i.d. 3 mm
norspermidine, spermine,
sym-norspermine,
thermospermine,
canavalamine
Adrenaline, dopamine, 5% OV-17 on N2, 2653C Biological tissues Derivatization with Bock and
noradrenaline, 5- Chromosorb 36 ml min\1 isothermal pentafluorobenzoyl Waser
hydroxytryptamine W HP 80-100 chloride and pyridine (1981)
mesh in acetonitrile
packing,
length 1 m,
i.d. 3 mm
Reproduced with permission by Kim KR et al . (1997).
reported to be responsible for diseases related to their sary: (a) the separation of amines from the matrix
biological activity. and (b) their determination. Pre-column derivatiz-
Muskiet et al. investigated a comparison of the ation is very often required for selectivity and sensi-
popular high performance liquid chromatographic tivity enhancement.
(HPLC) methods, whose chief advantage is a reduced
sample pre-treatment, and other chromatographic
techniques, such as gas chromatography, for the
Sample Preparation
separation and determination of biogenic amines. In The Rrst step is the most critical in terms of obtain-
chromatographic methods, two main steps are neces- ing an adequate recovery for amines, because of the
Figure 1 Dual chromatograms of amines as N(O)-isoBOC, O-TBDMS derivatives separated on DB-5 and DB-17 (both
30 m;0.25 mm I.D., 0.11 m film thickness) dual-capillary column system. GC conditions are described in the text. Peaks: 1"ethyl-
methylamine; 2"tert-butylamine; 3"diethylamine; 4"sec.-butylamine; 5"isobutylamine; 6"diisopropylamine; 7"n-
butylamine; 8"dipropylamine; 9"pyrrolidine; 10"isoamylamine; 11"morpholine; 12"piperidine; 13"n-amylamine; 14"
diisobutylamine; 15"thiazolidine; 16"n-hexylamine; 17"dibutylamine; 18"cyclohexylamine; 19"n-heptylamine;
20"diphenylamine; 21"o-toluidine; 22"benzylamine; 23"n-octylamine; 24"m-toluidine; 25"p-toluidine; 26"-
phenethylamine; 27"dihexylamine; 28"n-decylamine; 29"2,4-dichlorobenzylamine; 30"dicyclohexylamine; 31"1,3-diaminop-
ropane; 32"3,4-dichlorobenzylamine; 33"-hydroxyphenethylamine; 34"norephedrine; 35"ephedrine; 36"putrescine;
37"3,4-dimethoxyphenethylamine; 38"diethanolamine-1; 39"dibenzylamine; 40"cadaverine; 41"diethanolamine-2; 42"
histamine; 43"1,6-diaminohexane; 44"tryptamine; 45"1,7-diaminoheptane; 46"tyramine; 47"3-methoxytyramine; 48"
5-methoxytryptamine; 49"synephrine; 50"octopamine; 51"metanephrine; 52"3,4-dihydroxybenzylamine; 53"normetaneph-
rine; 54"dopamine; 55"spermidine; 56"serotonin; 57"epinephrine; 58"norepinephrine. (Reproduced with permission from
Kim KR et al . (1997).)
complexity of the matrices usually considered. All Chromatography

matrices pose different problems depending on their
complexity so many clean-up techniques have been Separations are mostly performed on capillary col-
proposed making use of different extraction proced- umns whose length ranges from 15 to 30 m. The silica
ures, i.e. liquid}liquid, solid-phase extraction (SPE), capillary columns are coated with different Rlm thick-
etc. and a variety of solvents (perchloric, tri- ness of methylsiloxane, methylphenylsiloxane, or cy-
chloroacetic, and hydrochloric acids and/or both anopropylphenylsiloxane polymers or polyethylene
polar and non-polar organic solvents). glycols. Only a few examples in the past reported the
In the second step, account must Rrst be taken of use of packed columns. Both isothermal and pro-
the amine pre-column derivatization; this is often, if grammed temperature conditions are used.
not always, necessary for enhancing the analyte re-
sponse and selectivity. The other aim of pre-column Detection
derivatization is to obtain derivatives, which are both
volatile and sufRciently stable for subsequent ana- Among the different detectors used in gas chromato-
lyses. Gas chromatographic analysis of the amines graphy for biogenic amine analysis, the four detectors
requires one or more appropriate derivatization pro- most commonly used are: Same ionization detector
cedures to block active protons in amino and other (FID), electron-capture detector (ECD), nitrogen}phos-
polar groups and acylation, silylation, benzoylation, phorus detector (NPD) and mass spectrometry (MS).
sulfonylation, phosphorylation and alkyloxycar-
Flame Ionization Detector (FID)
bonylation have been used. Baker et al. have given an
example of a complete study of the derivatization Some biogenic amines such as aliphatic mono-, di-
process in biogenic amines determination. and polyamines do not exhibit structural features that
permit their speciRc detection, in this case a non- One of the most recent and notable applications
speciRc detector such as Same ionization (FID) can be of FID detector is the determination of 57 amines
used, but the lack of sensitivity and/or selectivity may described in the paper of Kim et al. In their work
require the use of other detectors coupled with FID as they demonstrated the wide applicability of the
shown in Table 1. FID supported by the simultaneous use of two capil-
Bock and Waser acylated, via a single-step reaction, lary columns of different polarity, such as DB-5
some biogenic amines for their speciRc and quantitat- and DB-17 for resolving all the considered amines,
ive gas chromatographic assay. The retention times as shown in Figure 1. The proposed method is based
were found to be inversely proportional to their moon two-phase isobutyloxycarbonylation with a pH
lecular weights and also to the number of the pentaf- shift. The resulting derivatives are separated by SPE
luorobenzoyl groups. FID and ECD detector responses and the remaining hydroxyl groups are derivatized
were compared; FID was selected for nanogram range for dual capillary column gas chromatographic analy-
and ECD for picogram range determinations, in order sis. Response is linear in the 0.2}12-ppm range.
to optimize linearity and accuracy. The structure of
Electron-Capture Detector ECD
the derivatives was conRrmed by GC}MS technique.
Yamamoto et al. proposed a routine method for the Examples of the use of the electron-capture detection
determination of polyamines as their N-ethyloxycar- (ECD) are given in Table 2.
bonyl derivatives. The sample preparation is simple Staruskiewicz and Bond developed a procedure
and reproducible and the derivatives are stable. Also for the quantitative determination of putrescine and
in this case, the identity of the derivatives was con- cadaverine in foods. The amines were extracted with
Rrmed by GC}MS technique. methanol and a dry residue of their hydrochloride
Muskiet et al. prepared extracts of acid-hydrolysed salts was prepared. The salts were derivatized and
biological Suids by pre-puriRcation with silica gel to both the reagent excess and reaction by-products
give recoveries of better than 90%. The isolated com- were removed by means of an alumina column, min-
pounds were derivatized and simultaneously deter- imizing in this way the solvent fronts and interfering
mined by a capillary gas chromatographic method. extraneous peaks, some of which adversely affected
FID and NPD detector responses were compared the ECD response. The recovery of diamines was
showing that NPD is both more selective and sensi- within the range 97 to 102%. Gas chromatographic
tive than FID. separation was performed and the retention time for
Laleye et al. developed triSuoroacetylation for quan- the derivatives of putrescine and cadaverine was
tifying biogenic amines on a microgram scale in foods. found to be 4.3 and 5.7 min, respectively. When
Table 2 GC of biogenic amines with ECD detection
Analyte Chromatographic Carrier, Injector Matrix Sample treatment Reference

column type and temperature
flow rate and program
Histamine, tele- Fused silica capillary He, 2 mL 2503C, Foods Liquid}liquid extraction, Baker et al.
methylhistamine, SE-54, length 15 m, min\1 1053CP derivatization with (1987)
m-, p-tyramine, i.d. 0.25 mm 253C min\1 pentafluorobenzoyl
3-methoxytyramine, P2403C chloride}benzene}
putrescine, cadaverine, acetonitrile
tryptamine, spermine,
spermidine,
2-phenylethylamine,
5-hydroxytryptamine
Cadaverine, putrescine, 3% OV-17 on He, 60 mL 1203CP Biological SPE purification, Fujihara et al.
spermine, spermidine Chromosorb W HP packing min\1 153C min\1 fluids derivatization with (1983)
length 1.5 m, i.d. 3 mm P2803C heptafluorobutyric
anhydride
Cadaverine, putrescine 3% OV-225 on Gas He, 28 mL 2003C, Foods Liquid}liquid extraction, Staruszkiewicz
Chrom Q 100-120 mesh min\1 1803C derivatization with and Bond
packing, length 1.8 m, isothermal pentafluoropropionic (1981)
i.d. 2 mm anhydride, SPE
purification
were conRrmed by GC}MS. The detection limits

(peak/noise ratio"2) of the different types ranged
between 5 and 20 pg on-column and calibration
curves were linear over two orders of magnitude.
A typical chromatogram obtained with this tech-
nique is shown in Figure 2, that is relative to the
determination of perSuoroacetylderivatives in foods.
The method of Baker et al. resulted in a reference
paper for all later authors using the ECD.
Nitrogen^Phosphorus Detector (NPD)
Another very speciRc detector for amines is the nitro-
gen}phosphorus detector (NPD). The most recent
papers relative to the use of this detector for biogenic
amine determination are reported in Table 3.
One of the Rrst comparisons between FID and
another detector (NPD) appeared in the paper of
Muskiet et al. These authors prepared extracts of
acid-hydrolysed biological Suids by pre-puriRcation
with silica gel. The isolated compounds were de-
rivatized and simultaneously determined by capillary
gas chromatography with nitrogen}phosphorus de-
tection. Their comparison between typical chromato-
grams obtained with FID and NPD is shown by an
example in Figure 3.
Figure 2 Gas chromatogram of extracts of bioactive amines Perez-Martin et al. also compared FID and NPD
from (top) cheese sample and (bottom) a solution of authentic results, but contrary to Muskiet et al. they con-
standards. Extraction and derivatization were as described in sidered both detector responses as equivalent. These
the text. Derivatives of 2-phenylethylamine (A); 2-(4-chloro- authors treated seafoods with trichloroacetic acid and
phenyl)ethylamine, internal standard (B); tele-methylhistamine
(C); histamine (D); putrescine (E); tryptamine (F); cadaverine (G);
then extracted the resulting solution with benzene
m-tyramine (H); p-tyramine (I); 3-methoxytyramine (J); 5-hy- under highly basic conditions. Retention times are
droxytryptamine (K); spermidine (L); and spermine (M). Attenu- 1.03 min for dimethylamine and 1.35 min for
ation changes were programmed in to ensure that all peaks trimethylamine.
remained on scale for the purpose of illustration. (Reproduced Hessels et al. developed a capillary gas chromato-
with permission from Baker GB et al . (1987).)
graphic method with NPD for the determination of
the N-acetylspermidine catabolite in biological Suids
compared to a nitrogen-speciRc detector, the ECD and compared it with the GC}MS method of Van den
had greater sensitivity. Less than 1 ppm of diamine in Berg et al., shown in Table 4. The results were within
the food can be quantitated with a coefRcient of a relative standard deviation of 3%.
variation lower than 3.0%. One of the most recent applications of NPD for
Fujihara et al. developed a sensitive and selective polyamines proRling is the work of Dorhout et al.
method for the estimation of polyamines in biological that used capillary gas chromatography with NPD
Suids after their separation from other compounds for the determination of polyamines, N-acetylated
and derivatization. These derivatives are well re- polyamines and the polyamine analogues N,N-
solved within 15 min with detection limit of 0.1 pmol bis(ethyl)norspermine and 1,19-bis(ethylamino)-
for putrescine and cadaverine and 0.02 pmol for sper- 5,10,15-triazanonadecane in biological samples. The
mine and spermidine. suggested method provided for the use of four inter-
Baker et al. proposed a method based on extractive nal standards and pre-puriRcation steps comprising
derivatization of the amines with a perSuoroacylating deproteinization and isolation on silica at pH 9.0
agent under basic aqueous conditions. Polyamine re- with 70}90% recoveries depending on polyamine
covery was virtually quantitative, whilst for the other type. The precision is better than 7%, while the detec-
amines the recoveries ranged from 70 to 85%. Deriv- tion limits (peak/noise ratio"2) of the different
atives showed good chromatographic properties and components ranged from 0.4 to 0.7 pmol. Figure 4
ECD was used for enhancing sensitivity of biogenic shows a typical chromatogram obtained with the
amines in food samples. The derivative structures method described in the Dorhout work.
Table 3 GC of biogenic amines with NPD detection
Analyte Chromatographic Carrier, Injector Matrix Sample treatment Reference

column type and temperature
Putrescine, cadaverine, Fused silica capillary He, 0.6 mL 2603C, Biological Deproteinization, SPE Dorhout et al.
spermine, spermidine, methyl silicone phase, min\1 1203C tissues purification and (1997)
1,3-diaminopropane length 37.5 m, i.d. P53C min\1 derivatization with
0.20 mm, film thickness P2603C heptafluorobutyric
0.11 m P23C min\1 anhydride
P2803C
Polyamines, N -acetylsper- Fused silica capillary 2603C, Biological Alkalization, liquid} Hessels et al.
midine HP 17, length 25 m, 1803CP fluids liquid extraction and SPE (1991)
i.d. 0.20 mm, film 33C min\1 purification
thickness P2703C
0.17 m
Trimethylamine, 4% Carbowax N2, 25 mL 2503C, Seafoods Liquid}liquid extraction Perez-Martin

dimethylamine 20M#0.8% KOH on min\1 1153C et al. (1987)
Carbopack B and isothermal
Chromosorb 103
packing, length 1.8 m,
i.d. 2 mm
Cadaverine, 1,3- Fused silica capillary He, 0.5 mL 3003C, Biological SPE purification, Muskiet et al.
diaminopropane, cross-linked methyl min\1 1203CP fluids derivatization with (1984)
putreanine, isoputreanine, silicone and siloxane 73C min\1 heptafluorobutyric
putrescine, spermine, deactivated, P2603C anhydride
spermidine length 35 m, i.d.
0.20 mm, film thickness
0.11 m
Reproduced with permission from Muskiet FAJ et al . (1984).
Mass Spectrometry (MS) samples. The results provided an opportunity for

qualitative and quantitative comparison of these two
The actual trend in biogenic amine gas chromato- techniques and discussion of their respective merits
graphic analysis is to establish a complete proRle of and limitations.
all the important amines by using a single method, The speciRcity associated with SIM is responsible
thus the widest used detector is mass spectrometry for a signiRcant reduction in the complexity of data
(MS) (Table 4). This technique was formerly used in obtained when using a GC}MS system. The use of
conjunction with other detectors for conRrmation appropriate deuterated internal standards eliminates
purposes only. the inSuence of the original sample matrix, minimiz-
The advantage of MS over other detection systems ing sample pre-treatment, and allows precise multi-
is based on the possibility of identifying compounds component quantitation at low levels.
with a high degree of speciRcity. This can be realized Davis et al. preferred to use GC}MS in the mul-
using high resolution capillary gas chromatography tiple-ion detection mode (MID) after derivatization
combined with the monitoring of characteristic ions with two different reagent systems and precipitation
in the mass spectrum of a compound (GC}MS). of protein with sulfosalicylic acid prior to extraction,
GC}MS can afford characteristic ratios of ion inten- in order to analyse a large number of biogenic amines
sities or the m/z value of a particular ion. These and their metabolites in a single run.
parameters may be changed in a predictable manner Van den Berg et al. described the identiRcation of
by derivatization processes in order to provide addi- a metabolite of spermidine by GC}MS in biological
tional proof of identity. Suids of normal healthy humans and cancer patients.
Duncan et al. compared the results obtained using The quantiRcation was based on stable isotope dilu-
both HPLC with electrochemical detection and GC}MS tion mass fragmentography monitored at m/z 185
in the single-ion monitoring mode (SIM) when ap- and 188.
plied to dopamine and norepinephrine (noradrena- Fujihara et al. measured putrescine metabolism in
line) determination, after a single-step alumina biological Suids. Putrescine was decomposed by
extraction, in a selection of food and beverage oxidative deamination to form ammonia that was
Figure 3 Comparison between typical chromatograms obtained with flame-ionization (A) and nitrogenIphosphorus (B) detectors
after capillary gas chromatography of methyl-heptafluorobutyryl-derivatized extracts of acid-hydrolyzed urine from a normal man.
Abbreviations: DAP, 1,3-diaminopropane; OH Pu, 2-hydroxyputrescine; Pu, putrescine; C, cadaverine; 1, 1,6-diaminohexane; Lys,
lysine; 2, 1,7-diaminoheptane; Isoputr, isoputreanine; Putr, putreanine; 3, N1-methylisoputreanine; DBP, dibutylphthalate (plasticizer,
tentatively identified by gas chromatography-mass spectrometry); 4, bis(3-aminopropyl)amine; Sd, spermidine; 5, N-(3-aminopropyl)-
1,5-diaminopentane; Sp acid 2, N, N ’-bis(2-carboxyethyl)-1,4-diaminobutane; Sp acid 1, N-(3-aminopropyl)-N ’-(2-carboxyethyl)-1,4-
diaminobutane; Sp, spermine; 6, N, N ’-bis(3-aminopropyl)-1,5-diaminopentane; DOP, dioctylphthalate (plasticizer). 1 through 6 are
added internal standards. Time axis, minutes. (Reproduced, with permission, from Muskiet FAJ, van den Bergh GA, Kingma AW,
Fremouw-Ottenvangers DC and Halie R (1984) Total polyamines and their non- amino acid metabolites simultaneously determined in
urine by capillary gas chromatography, with nitrogen}phosphorus detector; and some clinical applications. Clinical Chemistry 30: 687.)
determined by GC}MS. Its quantiRcation was based lysed the phenolic esters and converted the free hy-
on mass fragmentography synchronously monitored droxyl groups to trimethylsilyl esters and analysed
at m/z 195, 211 and 212. The method is inferior to these derivatives by GC}MS}NICI. The molecular
isotope mass spectrometry, but its speed, reproduci- ion of these derivatives (together with the isotope
bility (better than 0.5%) and sensitivity are superior. peaks) carried more than 60% of the ion current,
ShaR et al. applied GC}MS in the negative ion which made the method highly speciRc and gave a po-
chemical ionization mode (NICI) for the determination tential limit of detection below the picogram level.
of biogenic amines. After derivatizing the amines and This technique has the advantage that the silylating
separating them by solvent extraction, they hydro- reagent may be changed in order to shift both the m/z
Table 4 GC}MS of biogenic amines
Analyte Chromatographic column Carrier, Injector Matrix Sample treatment Reference

type and temperature
p-Tyramine, p-octopamine, Fused silica capillary He, 40 mL 2503C, Biological Liquid}liquid extraction Shafi
p-synephrine, dopamine, HP 1, length 12.5 or, min\1 1003C tissues and derivatization with (1995)
noradrenaline, adrenaline, 25 m, i.d. 0.20 mm, film P103C min\1 5-ditrifluoromethylbenzoyl
5-hydroxytryptamine thickness 0.2 m P3003C chloride and isopropyl
dimethylsilyl-N-
methyltrifluoroacetamide
Indole ethylamines Fused silica capillary OV 1, 2603C, Biological Formaldehyde addition, Musshoff
length 12 m, i.d. 0.20 mm, 1003C fluids hyrdrolysis, derivatization et al. (1993)
film thickness 0.33 m P403C min\1 with methyl chloroformate,
P3003C liquid}liquid extraction
and SPE purification
Phenolamines, Fused silica capillary He 2503C, Biological Liquid}liquid MacFarlane

catecholamines, SGE BP1, length 12 m, 1003CP tissues extraction and et al. (1991)
5-hydroxytryptamine, i.d. 0.20 mm, 103C min\1 derivatization with
phenylalanine, film thickness P3003C butyldimethylsilyl
tyrosine, dopamine, 0.25 m chloride and ditri-
p-tyramine, fluoromethylbenzyl
p-synephrine, bromide
p-octopamine,
adrenaline,
noradrenaline
p-tyramine, Fused silica capillary He 2503C, Biological Extraction}derivatization Shafi et al.

m-, p-synephrine, HP1, length 12.5 m 1003CP tissues with difluoromethylbenzoyl (1989)
m- p-octopamine, or 25 m, i.d. 0.20 mm 103C min\1 chloride, liquid}liquid
dopamine, P3003C extraction, derivatization
5-hydroxytryptamine, with bistrimethyl
noradrenaline, silylacetamide (or t-
adrenaline, butyldimethylsilyl
dihydroxybenzylamine chloride
and imidazole) SPE
purification
Putrescine 3% OV-1 on He, 60 mL 2803C, Biological Derivatization of the Fujihara

Chromosorb W AW min\1 2803C fluids ammonia resulting from et al. (1986)
DMCS 80I100 mesh isothermal metabolilsm with
packing, length 2 m, pentafluorobenzoyl
i.d. 3 mm chloride
Spermidine Fused silica capillary He, 1 mL 2503C, Biological Liquid}liquid extraction, Van den
CP-Sil-19, length 25 m, min\1 2003CP fluids derivatization with Bergh et al.
i.d. 0.32 mm, 103C min\1 acetyl chloride (1986)
film thickness P2603C
0.20 m
Dopamine, adrenaline, Fused silica capillary He 1403CP Biological Deproteination and Davis et al.
noradrenaline, DB-1, length 60 m, 103C min\1 fluids derivatization with (1986)
phenylethylamine, i.d. 0.32 mm P2403C methanolic hydrogen
phenylethanolamine, chloride (or
tryptamine, trifluoroethanol) and
m-, p-tyramine, pentafluoropropionic
m-, p-octopamine anhydride
Dopamine, norepinephrine 3% OV-17 packing, He, 30 mL 2223C, Foods, SPE purification, Duncan
(noradrenaline) length 0.7}1 m min\1 1463CP beverages derivatization with et al. (1984)
2003C trifluoroacetic
anhydride and
trifluoroethanol
Figure 4 (A) NICI SIM trace of DTFMB-IPDMS derivatives of deuteriated and undeuteriated biogenic amines (each 20 ng) and (B)
the corresponding endogenous biogenic amines from a single brain of American cockroach after addition of deuteriated internal
standards (20 ng). (C) NICI SIM trace of Pr-PFP derivatives of deuteriated and undeuteriated biogenic amines (each 20 ng) and (D) the
corresponding endogenous biogenic amines from a single brain of American cockroach after addition of deuteriated internal standards
(20 ng). (Reproduced, with permission, from Shafi N (1995) Identification and quantitative determination of biogenic amines and their
metabolites by gas chromatography negative ion chemical ionisation mass spectrometry (GC}NICIMS) Journal of the Chemical Society
of Pakistan 17: 103.)
value of the molecular ion and the retention time of ShaR summarized the use of different derivatizing
the derivative to ensure that its identiRcation is un- agents, whose scope was shifting both the m/z value
equivocal. In the same laboratories MacFarlane et al. and the retention time in order to ensure the un-
used a similar procedure, but they state that the ion equivocal identiRcation of amines (Figure 5). In each
current was generally carried by four or Rve large ions. case, the principal ion in the mass spectrum was the
Musshoff et al. derivatized the compounds formed molecular ion, which carried almost all of the ion
by reaction of formaldehyde with biogenic amines in current under negative ion chemical ionization condi-
biological Suids in order to avoid the simultaneous tions; the sensitivity of this method of derivatization
formation of these compounds via condensation. was 1 pg of biogenic amine on-column.
These derivatization products are stable and can be The possibility of easy determination of biogenic
well puriRed from most of the interfering matrix amines by other chromatographic techniques, such as
compounds by liquid}liquid and solid-phase extrac- HPLC, has reduced the interest in updating the exist-
tion. They have good GC}MS properties, therefore ing gas chromatographic methods.
the retention times, together with the diagnostic mass
fragments (at least three) and the speciRc ion ratios, See also: II/Chromatography: Gas: Column Techno-
can be used for separation and identiRcation. logy; Derivatization; Detectors: General (Flame Ionization
Detectors and Thermal Conductivity Detectors); De- biologically active amines. Jounal of Pharmaceutical
tectors: Mass Spectrometry; Detectors: Selective. Extrac- and Biochemical Analysis 15: 1309.
tion: Solid-Phase Extraction. Laleye LC, Simard RE, Gosselin C, Lee BH and Giroux RN
(1987) Assessment of Cheddar cheese quality by
chromatographic analysis of free amino acids and bio-
Further Reading genic amines. Journal of Food Science 52: 303.
Baker GB, Coutts RT and Holt A (1994) Derivatization MacFarlane RG, Midgley JM and Watson DG (1991) Bio-
with acetic anhydride: applications to the analyses genic amines: their occurrence, biosynthesis and meta-
of biogenic amines and psychiatric drugs by gas bolism in the locust, Schistocerca gregaria, by gas
chromatography and mass spectrometry. Journal of chromatography}negative-ion chemical ionisation
Pharmacological and Toxicological Methods 31: 141. mass spectrometry. Journal of Chromatograpy B 562:
Baker GB, Wong JTF, Coutts RT and Pasutto FM (1987) 585.
Simultaneous extraction and quantitation of several Muskiet FAJ, van den Bergh GA, Kingma AW, Fremouw-
bioactive amines in cheese and chocolate. Journal of Ottenvangers DC and Halie R (1984) Total polyamines
Chromatography 392: 317. and their non--amino acid metabolites simulatneously
Bock UEG and Waser PG (1981) Gas chromatographic determined in urine by capillary gas chromatography,
determination of some biogenic amines as their penta- with nitrogen}phosphorus detector; and some clinical
Suorobenzoyl derivatives in the picogram range and applications. Clinical Chemistry 30: 687.
its applicability to biological materials. Journal of Muskiet FAJ, Dorhout B, van den Bergh GA and Hessels
Chromatography 213: 413. J (1995) Investigation of polyamine metabolism by high-
Davis BA, Durden DA and Boulton AA (1986) Simulta- perfomance liquid chromatographic and gas chromato-
neous analysis of twelve biogenic amine metabolites in graphic proRling methods. Journal of Chromatography
plasma, cerebrospinal Suid and urine by capillary col- B 667: 189.
umn gas chromatography}high-resolution mass spectro- Musshoff F, Daldrup TH and Bonte W (1993) Gas
metry with selected-ion monitoring. Journal of chromatographic}mass spectrometric screening proced-
Chromatography B 374: 227. ure for the identiRcation of formaldehyde-derived
Dorhout B, Kingma AW, de Hoog E and Muskiet FAJ tetrahydro--carbolines in human urine. Journal of
(1997) Simultaneous determination of polyamines N- Chromatography B 614: 1.
acetylated polyamines and the polyamine analogues BE- Parvez S, Nagatsu T and Parvez H (ed.) (1983) Methods of
3-3-3 and BE-4-4-4-4 by capillary gas chromatography Biogenic Amine Research. Amsterdam: Elsevier.
with nitrogen}phosphorus detection. Journal of Perez-Martin RI, Franco JM, Molist P and Gallardo JM
Chromatography B 700: 23 (1987) Gas chromatographic method for the determina-
Duncan MW, Smythe GA, Nicholson MV and Clezy PS tion of volatile amines in seafoods. International Journal
(1984) Comparison of high-performance liquid of Food Science and Technology 22: 509.
chromatography with electrochemical detection and gas ShaR N (1995) IdentiRcation and quantitative determina-
chromatography}mass fragmentography for the assay of tion of biogenic amines and their metabolites by gas
salsolinol, dopamine and dopamine metabolites in food chromatography negative ion chemical ionisation mass
and beverage samples. Journal of Chromatography B spectrometry (GC}NICIMS). Journal of the Chemical
336: 199. Society of Pakistan 17: 103.
Fujihara S, Nakashima T and Kurogochi Y (1983) Deter- ShaR N, Midgley JM, Watson DG, Smail GA, Strang R and
mination of polyamines in human blood by electron- MacFarlane RG (1989) Analysis of biogenic amines in
capture gas}liquid chromatography. Journal of the brain of the American cockroach (Periplaneta ameri-
Chromatography B 277: 53. cana) by gas chromatography negative ion chemical
Fujihara S, Nakashima T and Kurogochi Y (1986) Deter- ionisation mass spectrometry. Journal of Chromatogra-
mination of 15NH3 by gas chromatography}mass spec- phy B 490: 9.
trometry. Application to the measurement of putrescine Staruszkiewicz WF Jr and Bond JF (1981) Gas chromato-
oxidation by human plasma. Journal of Chromatogra- graphic determination of cadaverine, putrescine and his-
phy B 383: 271. tamine in foods. Journal of the Association of OfTcial
Guggenheim M (1940) Die biogenen Amine. Basel: Karger. Analytical Chemists 64: 584.
HalaH sz A, BaraH th A, Simon-Sarkadi L and Holzapfel Van den Berg GA, Kingma AW, Elzinga H and Muskiet
W (1994) Biogenic amines and their production by FAJ (1986) Determination of N-(3-acetamidopropyl)
microorganisms in food. Trends in Food Science and pyrrolidin-2-one, a metabolite of spermidine, in urine by
Technology 5: 42. isotope dilution mass fragmentography. Journal of
Hassels J, Kingma AW, Sturkenboom MCJM, Elzinga H, Chromatography B 383: 251.
van den Bergh GA and Muskiet FAJ (1991) Gas Yamamoto S, Iwado A, Hashimoto Y, Aoyama Y and
chromatographic determination of N-acetylisoput- Makita M (1984) Gas chromatography}mass spectro-
reanine--lactam, a unique catabolite of N1-acetylsper- metry of polyamines as their N-ethyloxycarbonyl deriv-
midine. Journal of Chromatography B 563: 1. atives and identiRcation of sym-homospermidine and
Kim KR, Paik MJ, Kim JH, Dong SW and Jeong DH (1997) sym-norspermine in mosses and ferns. Journal of
Rapid gas chromatographic proRling and screening of Chromatography 303: 99.
2156 III / BIOLOGICAL SYSTEMS: ION EXCHANGE
BIOLOGICAL SYSTEMS: ION EXCHANGE
R. J. P. Williams, Imperial College, bind to surfaces of proteins, other polymers, precipi-

London, UK tates or membranes and exchange rapidly or slowly.
Copyright ^ 2000 Academic Press Consider a magnesium protein such as Mg2# parv-
albumin, where parvalbumin is a common anionic
There are two very different processes which are metal-ion-binding protein in muscle cells. When the
described under the heading of ion exchange within protein, which is magnesium bound in the cell, is
biological systems. The Rrst is exchange in an aque- exposed to calcium, the reaction of exchange occurs:
ous environment of two different ions, MA and MB, Ca2# parvalbumin is formed and Mg2# is freed. If
held by a molecule in free solution, or in a precipitate, the initial protein complex had been Na# parval-
or on a membrane surface: bumin, then replacement by Ca2# would have been
electrogenic, unless of course two Na# ions had been
bound originally.
A #MBX 8 MB #MAX
Mn# n#
[1]
A second example of exchange on the surface of
a polymer is provided by polynucleotides such as
Here, M is a metal cation and X is an anion, often an
RNA and DNA. These polymers have anionic phos-
organic molecule which can be as large as a protein or
phate backbones and their negative charges are bal-
DNA. Exchange of anions is also possible, as in the
anced, not entirely randomly, by exchanging cations,
equilibrium:
including particularly K#, Mg2# and ammonium
derivatives. The positive charge, like the negative
XnA\#MXB 8 XnB\#MXA [2] charge, in an exchanging system can also be carried
on a large molecule and in biology there are many
Both eqns [1] and [2] have been written for exchang- polyamines and positively charged proteins, such as
ing ions of equal charge type n. This is not a necessary histones, which bind DNA. They can bind and ex-
condition, so we can also consider: change anions such as HPO24\ and SO24\.
A further example of biological metal ion exchange
2M#
A #MBX 8 MB #(MA)2X
2#
[3] in one aqueous solution now involving a solid pre-
cipitate is the case of bone, which we can write
A #MXB 8 XB\#M(XA)2 as (Ca2#)2(OH\)PO34\ for simplicity. Bone scavenges
2
2X\ [4]
many other cations by exchange and contains
Again, the charge balance shown so far in the equa- considerable amounts of Mg2# and even Al3# in
tions is not essential, so we must also examine the place of Ca2#. Bone also exchanges Ca2# for
situation of electrogenic exchange: protons (H#). This mineral is additionally an anion-
exchanging solid in which some OH\ is replaced by
M#
A #MBX 8 MB #MAX\
2# F\ (a process used in the protection of teeth). To
some extent, a pair of OH\#PO34\ can be re-
# placed by 2CO23\ anions. Notice that exchange can
A #MXB 8 XB\#MXA
2
X\
be electrogenic when charge is left on the surface of
the bone mineral.
where charge distribution is associated with MX
The last example of this relatively simple exchange
changes. Before considering the second way in which
is at the surface of membranes. All biological mem-
ion exchange can occur across a membrane, we give
branes carry negative charge due to covalently bound
one or two examples of ion exchange to and from
phosphate or carboxylate groups. Since the charges
a bound condition in a single aqueous solution.
are quite close-packed, the membranes have a con-
siderable afRnity for cations and all membranes have
Exchange of Major Cations and associated K# and Mg2# (mainly internally) and
Na# and Ca2# (mainly externally) associated with
Anions in Cells their surfaces. The outermost coats of cell walls in
The major small cations and anions in cells which Gram-negative bacteria are also anionic and are sta-
undergo ready exchange are shown in Table 1. They bilized by exchangeable cation, often Ca2#, incorpo-
may associate with one another in rapid exchange or ration.
III / BIOLOGICAL SYSTEMS: ION EXCHANGE 2157
Table 1 Major small cations and anions of cells
Cation Anion
Na# largely rejected CI\ largely rejected

K# cytoplasmic HPO24\ cytoplasmic
Mg2# cytoplasmic SO24\ cytoplasmic
Ca2# rejected (vesicular) HCO\ 3 balanced
RNH#3 cytoplasmic RCO\ 2 cytoplasmic
ROPO23\ cytoplasmic
H#(pH 7); OH\(pH 7).
Exchange of Trace Elements

Major exchange of trace elements (Table 2) occurs
from many carrier or buffer proteins. For example,
iron is carried in the blood stream of higher animals
by the protein transferrin in the form of Fe3#, in
co-association with the carbonate anion, CO23\. The
Figure 1 An outlined scheme of the variety of ion exchange
uptake into the cell involves the transfer of the whole reactions in the uptake of iron. Iron is captured as Fe3# outside
protein to a vesicle, the lysozyme, which is much the cell by a small organic molecule, ferroxamine (X). FeX is
more acidic than blood. There the protein loses both passed through membranes (TON B) using energy, and is then
iron and CO23\, which are exchanged for bound pro- exchanged in a variety of paths. Some involve proteins, P; some
tons. The iron is processed further into the cell as enzymes catalyse for example the citrate cell. Some iron is ex-
changed into new small organic molecules, porphyrins, to make
Fe2#, only to be reoxidized in a precipitate inside new enzymes. Finally, much is stored in a bound small particle of
a protein matrix, ferritin, where it is stored as Fe(OH)3, called ferritin.
Fe(OH)3. From this small particle, iron exchanges
into the cytoplasm, to give a great variety of com-
pounds (see Figure 1) for the case of bacteria.
A second example of cytoplasmic ion exchange of
Ion Exchange across Membranes
trace elements involves both copper and zinc bound A different form of ion exchange is that which occurs
to sulfur-containing proteins. The proteins release across a membrane separating two aqueous phases
copper or zinc (M) to the cytoplasm in equilibrium (Figure 2). Once again, the simplest exchange is of
with free hydrogen ions: two ions, cations such as Na#/K#, Mg2#/Ca2# or
of anions such as OH\/Cl\, where there is
Pr.M#H# 8 M##PrH equal charge maintenance but change of concen-
tration on both sides of the membrane. There are
where Pr is a protein such as metallothionein. This other possibilities than exchange of equally charged
equilibration ensures constant levels of Zn2# and ions, such as the exchange of 2Na#/Ca2# or
Cu# in the cytoplasm. 2Cl\/HPO24\, when only concentration changes
Trace anions are also carried in the blood stream by are again involved on exchange. The further possi-
special proteins and are then transported into cells. bility is that the exchange process is electrogenic,
The proteins which transfer SO24\, SeO24\ and when charge differences are built up across a mem-
MoO24\ are known to be similar to transferrin. brane. Simple examples are the exchanges of
H#/Ca2# or OH\/HPO24\.
In the case of electrogenic exchange, an electrical
Table 2 Trace cations and anions of cells potential develops across the membrane. Such poten-
tials exist across most membranes of all cells but to
Cation Anion
different degrees. The development of such ion ex-
Mn2# vesicular SeO24\ cytoplasmic change potentials has led to the successive evolution
Fe2#/Fe3# cytoplasmic (precipitate) MoO24\ cytoplasmic from simple cells to nerves and then to the brain. In
Co2# cytoplasmic I\ cytoplasmic these cells Na#/K# exchange carries the electrolytic
Ni2# vesicular NO\3 cytoplasmic
current.
Cu2# rejected
Zn2# cytoplasmic Examples of exchange of ions across membranes
are numerous in biological systems. One of the best
Many of the ions are held in complexes with organic agents. known is the 2Na#/Ca2# exchange in many slow
Figure 2 The distribution of the major anions, CI\ and RPO24\ and cations, Na#, K#, Ca2# and Mg2# across a biological cell
membrane. Filled circles, charged (negative) phospholipids; open circles, uncharged lipids.
muscle cells. This exchange occurs after muscle ac- The Natural Environment and
tion is triggered by the invasion of Ca2# and is used Ion Exchange
to restore the resting state of a muscle such as that of
the heart. An example of electrogenic exchange is To appreciate the value of ion exchange in biological
found in the nerve when 2K# ions enter the cell in systems, we must look at the environment of cells and
exchange for 3Na#, thus creating a membrane po- the nature of cellular organic chemicals. The major
tential (Figure 3). However, this process costs energy Suids surrounding cells are the sea, fresh water and
and was a later development in evolution. We turn to artiRcially maintained body Suids for multicellular
the energetics of such exchange processes below. species such as humans living in air. All of these Suids
are aqueous electrolyte solutions containing elements
as ions, as shown in Figure 3. The composition of the
sea is given in Figure 4. The major feature of the
internal cellular solution, that is, the solution pro-
tected by the cell’s membrane, called the cytoplasm, is
that it contains many organic molecules and anions.
Both the environmental Suids and the cytoplasm have
an osmotic pressure which, unless counter measures
are taken to exclude many of the ions of the environ-
ment, will be greater for the cytoplasm due to the
organic molecules present there. Hence all cells are in
danger of bursting through osmotic stress. Due to the
anionic nature of the organic molecules in the cell and
the fact that the membranes themselves are also ani-
onic due to the phospholipid head groups, they are
also in danger of breakdown due to internal electro-
static repulsion between the anions. Various ways of
overcoming these problems have been observed in
different biological cells and they frequently involve
ion-exchange.
Osmotic stress can be overcome by protecting the
outer membrane by a cell wall. This wall, built from
anionic organic cross-linked polymers, is commonly
found in bacteria and plant cells. The anionic wall is
exposed to the environment and acts as an ion-
exchange resin accepting especially Ca2# and Mg2#
Figure 3 The adenosine triphosphate (ATP) pump in mam-
malian cells. It exchanges 3Na# for 2K# and is electrogenic. The ions, which considerably strengthen the wall by cross-
concentration of ions inside and outside human cells is shown linking the organic polymers in it. These cations free-
below. ly exchange with the environment, so that the wall
Figure 4 The concentrations of elements in various ionic forms in the sea. Surfaces of cells exchange ions with all these species but
clearly those up to atomic number 20 dominate. However, a biological cell exchanges and picks up more than 20 elements, including
molybdate and iodide (from Cox 1995, with permission).
does not have a Rxed chemical composition and acts nonequilibrium conditions, we need to observe that
much like an artiRcial ion exchange resin. cells have to move many substances other than Na#,
All cells, bacterial, plant and animal, also over- K#, Mg2# and Cl\ (Table 3). Firstly, they need to
come osmotic stress by the selective admission and exclude Ca2#, since at the concentration of the envi-
rejection of ions of the environment. The major ions ronment these ions would precipitate cell anions such
of the environment are Na# and Cl\, and in the case as carboxylates and phosphates. Secondly, the cyto-
of the sea they are quite highly concentrated. Both plasmic level of anions such as phosphate is too low
must be prevented from equilibrating between the and that of sulfate is too high in the environment,
environment and the cytoplasm if the cell is not to especially in the sea, so that their cellular concentra-
burst. However, to maintain approximate electrical tions must be controlled. Thirdly, the cell must be
neutrality there has to be counterions to the anionic able to take in anions (or cations) useful as food for
organic constituents of the cell. Hence, all cells admit synthesis of organic molecules and these include ni-
K# and Mg2# ions and reject Na# and Cl\ ions to trate, small organic phosphates and carboxylates and
a greater or lesser degree (Table 3). Of course, this ammonium or small amine cations. Fourthly, there
arrangement of ions across the cell membrane costs are a variety of trace elements in the environment, all
energy, so the forced exchange of ions must use an of which are required in selected amounts in the cell;
energy source internal to the cell. for example, cations such as those of iron and manga-
Before describing the modes of employing energy in nese and anions such as selenate and molydate. Fi-
pumping ions across membranes so as to establish nally, we have not mentioned the universal presence
Table 3 Pumped gradients of metals and their complexes
Metal Inward to cytoplasm Outward to Organelle or vesicle
Na# Free ion Outside

Mg2# Free ion
H# Free ion and many ligands Free ion and many ligands Vesicles, outside
K# Free ion
HPO24\ Free ion Several organic phosphates Mitochondria
CI\ Free ion Free ion Outside
Fe3# Ferroxamines, etc.
Transferrin Citrate Mitochondria
Co2# Vitamin B12
Ca2# Free ion Reticula, outside
of the proton H# which is in constant exchange with Table 4 Examples of ATP-driven ion exchange pumps in cell
many anionic surfaces. Very much of the energy of membranes
biological reactions is expended in ion exchange
Pump Function
across membranes.
K#/Na# K# moves in; Na# moves out. Nerve conduction
H#/Ca2# Ca2# moves out. Fast (skeletal) muscle recovery
Energized Ion Exchange H#/M Trace element input or rejection
There are two major sources of energy open to cells.
One is a gradient of protons, H# (a pH gradient),
developed from the action of light of from the oxida- The use of ATP is that its hydrolysis energy (!G)
tion of organic molecules (Figure 5), while the second can pump ions across membranes:
comes from the hydrolysis of pyrophosphate bound
to nucleotides such as adenosine triphosphate (ATP). ATPPADP#Pi(!G)
The proton gradient can be used in direct ion ex-
!G#XinPXout
change with other ions, X, across the membrane to
accumulate them: !G#XoutPXin
H#
in #X\
outPX\
#
in #Hout It is now known that the action of ATP pumps fre-
quently involves ion exchange so that transfer of one
H# # # #
in #MoutPMin #Hout cation or anion is coupled to transfer of a second ion
in the opposite direction, giving ion exchange while
or to develop unfavourable gradients for protection building energized ion gradients. Examples are given
by removing X or M from the cell to the environment: in Table 4.
H#
in #X\
in PX\
#
out#Hout
Selectivity of Ion Exchange
# # # #
H #M PM #H
in in out out The selectivity of ion exchange interactions depends
on the relative binding strengths of ions to a site,
The Rrst is termed a symport and the second an where the site may be a molecule free in the cyto-
antiport. In every case, movement across the mem- plasm, a carrier molecule in a membrane, a surface
brane can be aided by carriers, M or X in the mem- of a membrane or a precipitated phase, or part of
brane, or by utilizing channels. Channels are control- a channel or a pump for moving ions across mem-
led, gated pores through membranes. branes. The afRnity for the site can usually be ex-
pressed in a very simple form as a binding stability
constant, K, for the equilibrium:
M#X 8 MX
where K"[MX]/[M][X].
The concentration of X is here treated in mass
action equations and takes a similar form for surface
sites or sites in free solution. In the case of surface
sites, the equation takes the form of a Langmuir
isotherm.
Selectivity of binding depends on the size and
charge of an ion in the Rrst instance. It would be
expected that small highly charged ions of opposite
sign would bind together best, but this expectation is
not fully borne out in practice, since competition for
a site also depends on constraints due to hydration of
Figure 5 A simplified picture of the production of a proton the ions:
gradient and then of ATP by the action of light or oxidation of
organic molecules inside a membrane. The gradient or ATP is M(H2O)n#X 8 MX#nH2O
then connected (X) to aqueous phases when ion (M ) exchange
across membranes can be driven, indicated by", the symbol for Since small, highly charged ions are the most strongly
a condenser. hydrated, making for competition between H2O and
Figure 6 The molecule shown at (A) can bind the monovalent ions selectively because of how it wraps around them (C). The hole in
the middle selects the size of the ion, as shown by the binding constants (B). The best binding is for potassium. Such molecules can pick
up cations and transfer them across membranes, exchanging the ion for other ions according to binding strengths and concentrations.
Many antibiotics are based on such exchange possibilities.
X, there is the possibility of matching ions, M, of Together with the fact that the amounts of ions, M, in
different charges and sizes with particular kinds of the cytoplasm varies from 10\1 mol L\1 (Na#, K#)
designed exchange site, X. Controlling factors now to 10\17 mol L\1 (Fe3#), this chemical selectivity
for cations, M, are the charge on X and the steric leads to a virtually complete Rxation of the M distri-
restrictions present in X or induced in MX on bind- bution on different X centres. However, it must not
ing. Real examples illustrate the point that the bind- be forgotten that these associations are not perma-
ing of cations M to sites X can be in almost any order. nent and exchange takes place all the time. Many sites
The case of the preferred selection of K# over the will only be occupied preferentially, not speciRcally,
smaller and larger ions Li#, Na#, Rb# and Cs# by a given cation. This is particularly important when
illustrates the point in Figure 6. In all cells the the environment becomes polluted.
K# channel virtually excludes Na#. Similarly, the The selective uptake of anions follows similar
Ca2# pump of most cells excludes Mg2#. In both properties based on size, charge and steric con-
bases the larger ion is preferred due to the size of the straints. However, covalent attachment is much less
cavity and the accepting anion, X. signiRcant than hydrogen bonding. The ability to
For trace element metal ions such as those of iron form hydrogen bonds by anions appears to be related
(Fe2#, Fe3#), zinc (Zn2#) or copper (Cu#, Cu2#), to surface charge density. Thus, F\ and OH\ readily
facts other than charge and size inSuence selectivity. form H bonds when compared with Cl\, Br\ or SH\.
They are the ability to form covalent bonds depend- Once again, the anions exchange quite rapidly with
ing on the electron afRnity of the cation and organic surfaces.
stereochemical preferences of the metal ions due to their
polarizability. Again, examples illustrate these points.
A very important distinction between the bindings
Ion Exchange and Cell Compartments
of the major metal ions, Na#, K#, Mg2# and Ca2# The selectivity of binding to carriers, to channels
and the trace elements, is that the binding units X for and channel parts of pumps in membranes leads to
the former generally employ organic molecules con-
taining oxygen (O) donor centres only, while for the
trace elements the group X may utilize nitrogen (N) Table 5 Preferred metal/nonmetal ion association
or sulfur (S) donors. Examples are given in Table 5,
Amongst the trace elements the strength of binding Metal Preferred nonmetal association
follows the general series of divalent M2# ions: Na#, K# Low association, preference for O-donor
Mg2#, Ca2# Moderate association with O-donor anions
Cu'Ni5Zn'Co'Fe'Mn'Mg
such as carboxylates and phosphates
The additional selectivity factor arises due to the Fe2#, Co2# Strong association with mixture of N- and
O-donors
stereochemical preferences of these ions.
Cu2#, Zn2# Strong association with S-donor such as thiolates
We can now consider the cytoplasm of a cell as
a solution of ion exchange centres, often proteins, Exchange is fast for Na#, K# but progressively slower down the
which can all bind M but with different afRnities. table.
Figure 7 Some of the distributions of elements in eukaryotic cells. Iron is often in membranes, while CI\, Ca2# and Na# are
concentrated away from the cytoplasm; K# and Mg2# concentrate inside cells, Mn2# is to be found in vesicles, but copper is more
usual outside cells. Zinc is everywhere and cobalt is rare everywhere. Aluminium may be rejected. P, Protein; Ch, chelating agent.
Many movements are due to ion exchange.
separate movement of elements in cells and then posi- trol ion levels throughout their whole body. They
tions them in particular zones of vesicles. Within do this by managing uptake and rejection of ions
these zones, ion exchange of the selected M ions at selectively. Thus it is necessary to ingest sufRcient but
X occurs, so that selectivity of association is manipu- not too much of many ions such as Na#, K#, Cl\
lated by the input of energy. Some cell compartments and HPO24\. The monitoring and rejection organ is
and their selective element contents are shown in the kidney. The kidney membranes act as complic-
Figure 7. ated ion exchange systems. However, the brain ap-
The setting-up of such ion gradients across mem- parently requires special ionic conditions, since cer-
branes has great value, not only in that it allows ebrospinal Suids of quite different composition from
speciRc reactions to occur in particular parts of space blood.
but also that, under stimulus, the ions can be released
from their storage sites into the cytoplasm. There is
then ion exchange signal. A particularly important
Ion Exchange and Pollution
example concerns the storage and subsequent release Many unusual elements enter environmental waters
of calcium ions from vesicles into the cytoplasm, due to mining and industrial processing of minerals.
causing manifold changes in metabolism and mech- Several of these elements are toxic and we note espe-
anical structure. Simple examples are muscle contrac- cially lead and mercury. There is considerable interest
tion and hormone release. today in the removal from cells of poisonous elements
We can look upon a mineral phase as a compartment such as Hg2#, Pb2# and AsO34\. Once again, even in
where minerals such as amorphous silica (often in bacteria, special proteins recognize these ions and
plants), calcium carbonate and calcium hydroxyphos- either transfer them directly across membranes or
phate are the obvious examples. These precipitates, release them to specialized pumps for energized
by exchanging ions, can buffer solutions holding movement out of the cell. While bacteria have evol-
ions in their neighbouring solutions in Rxed amounts. ved detoxifying processes, higher animals have not,
so humans have to take action to remove the offend-
ing elements. The application of a drug may force the
Ion Exchange and Organs offensive element to exchange from the site where it
As well as the local problems of ion exchange, large- causes damage, becoming bound to the drug when it
bodied organisms, such as human beings, must con- is excreted.
III / BIOLOGICALLY ACTIVE COMPOUNDS AND XENOBIOTICS: MAGNETIC AFFINITY SEPARATIONS 2163
Conclusion Further Reading

While ion exchange appears to be an extremely simple Aidley DJ and StanReld PR (1996) Ion Channels. Cam-
idea, it is used in remarkably complicated ways in bridge: Cambridge University Press.
living organisms. Of course, we do not know how life Allen TJA, Noble D and Reuter H (eds) (1989) So-
began but perhaps the earliest step in the direction of dium}Calcium Exchange. Oxford: Oxford Science Pub-
the development of these organized chemical systems lications.
was the formation of an ion gradient across a mem- Birch NJ (1993) Magnesium and the Cell. London: Aca-
demic Press.
brane. Once such a gradient had formed, possibly of
Bockris J O’M and Reddy AKN (1970) Modern Elec-
the proton, ion exchange could be used to move some trochemistry, vols 1 and 2. London: Plenum Press.
elements out of cells and others into cells. These ions Cheung WY (ed.) (1982) Calcium and Cell Function, vols
could be inorganic or organic. The movements became I}IV. New York: Academic Press.
more complicated as more membrane containments Cox PA (1995) The Elements on Earth. Oxford: Oxford
developed in evolution. Within each compartment ions University Press.
could bind to form complexes or precipitates by ex- Frausto da Silva JJR and Williams RJP (1996) The Biolo-
change processes. We know that cells have many dif- gical Chemistry of the Elements. Oxford: Oxford Uni-
ferent kinds of ion gradient through this exchange: versity Press.
some relate to energy storage and some to messages by Gennis RB (1989) Biomembranes. New York: Springer.
release of the gradients. Recently evolved systems are Kaim W and Schwederski B (1994) Bioinorganic Chem-
istry. Inorganic Elements in the Chemistry of Life. Chi-
calcium triggering of muscle and sodium/potassium
chester: Wiley.
currents in nerves. The restoration of the gradients is Michell AR (1995) The Clinical Biology of Sodium. Ox-
very frequently energized by ion exchange across mem- ford: Pergamon Press.
branes. This leads us to the tantalizing problem of the Phillips CGS and Williams RJP (1966) Inorganic Chem-
movement of ions in the brain. Are ion exchange pro- istry, vol. 1, ch. 7, pp. 231}265. Oxford: Oxford Univer-
cesses deeply involved in storage, memory, and in think- sity Press.
ing? We know that the brain is an electrolytic device Stryer L (1988) Biochemistry, 3rd edn. New York: W.H.
and hence ion movements are always active. Clearly, we Freeman.
could speculate at length on this topic, but what is really Townshend A (ed.) (1995) Encyclopedia of Analyti-
required is more experimental evidence concerning cal Science, vol 4, pp. 2261}2365. London: Academic
ion exchange in organisms and especially in the brain. Press.
Walton HF and Rocklin RD (1990) Ion Exchange in
See also: II/Ion Exchange: Inorganic Ion Exchangers; Analytical Chemistry. Boca Raton, FL, USA: CRC
Theory of Ion Exch]ange. Press.
BIOLOGICALLY ACTIVE COMPOUNDS

AND XENOBIOTICS: MAGNETIC
AFFINITY SEPARATIONS
I. S[ afar\ eH k and M. S[ afar\ eH kovaH , target molecules in the presence of vast amounts of
Institute of Landscape Ecology, Academy of accompanying compounds in both small and large-
Sciences, C[ eske& Bude\ jovice, Czech Republic scale processes, even in the presence of particulate
Copyright ^ 2000 Academic Press matter, are necessary.
In the area of biosciences, isolation of biologically
active compounds and xenobiotics is usually per-
formed using a variety of chromatography proced-
Introduction ures, afRnity chromatography being one of the most
Isolation and separation of speciRc molecules is used important. AfRnity ligand techniques currently rep-
in almost all areas of biosciences and biotechnologies. resent the most powerful tool available for down-
Separation technology is thus one of the most impor- stream processing both in terms of their selectivity
tant areas for further study and development. New and recovery. The strength of column afRnity
separation techniques, capable of treating dilute solu- chromatography has been shown in thousands of
tions or solutions containing only minute amounts of successful applications, especially on a laboratory
2164 III / BIOLOGICALLY ACTIVE COMPOUNDS AND XENOBIOTICS: MAGNETIC AFFINITY SEPARATIONS
scale. The disadvantage of standard column proced- Magnetic carriers and adsorbents are commercially
ures is the impossibility of such systems to cope with available and can also be prepared in the laboratory.
samples containing particulate material so they are Such materials are usually available in the form of
not suitable for use in the early stages of the isola- magnetic particles prepared from various synthetic
tion/puriRcation process where suspended solid and polymers, biopolymers or porous glass, or magnetic
fouling components are present. In this case mag- particles based on inorganic magnetic materials such
netic afRnity batch adsorption, applications of as surface-modiRed magnetite can be used. In fact,
magnetically stabilized Suidized beds or magneti- many of the particles behave like paramagnetic or
cally modiRed two-phase systems have shown their superparamagnetic ones responding to an external
usefulness. magnetic Reld, but not interacting themselves in the
The basic principle of magnetic afRnity separation absence of a magnetic Reld. This is important due to
is very simple. Magnetic carriers bearing an immobi- the fact that magnetic particles can be easily resus-
lized afRnity ligand or magnetic biopolymer particles pended and remain in suspension for a long time. The
are mixed with a sample containing target com- diameter of the particles is from ca. 50 nm to ca.
pound(s). Samples may be crude cells lysates, whole 10 m. Magnetic particles having a diameter larger
blood, plasma, urine, cultivation media, water, soil than ca. 1 m can be easily separated using simple
extracts and many others. Following an incubation magnetic separators, while separation of smaller par-
period when the target compound(s) bind to the mag- ticles (magnetic colloids with a particle size ranging
netic afRnity particles, the whole magnetic complex is between tens and hundreds of nanometers) may re-
easily and rapidly removed from the sample using an quire the use of high-gradient magnetic separators.
appropriate magnetic separator. After washing, the Commercially available magnetic particles can be
isolated target compound(s) can be eluted and used obtained from a variety of companies. In most cases
for further work. polystyrene is used as a matrix, but carriers based on
Magnetic separation techniques have several ad- cellulose, agarose, silica, porous glass or silanized
vantages in comparison with standard separation magnetic particles are also available. Particles with
procedures. The separation process can be performed immobilized afRnity ligands are available, oligo-
directly in crude samples containing suspended solid deoxythymidine, streptavidin, antibodies, protein A
material. Due to the magnetic properties of the mag- and protein G being used most often. Magnetic par-
netic afRnity particles (and the diamagnetic proper- ticles with such immobilized ligands can serve as
ties of the majority of the contaminating molecules generic solid phases to which native or modiRed afRn-
and particles present in the treated sample) they can ity ligands can be immobilized (e.g. antibodies in the
be relatively easily and selectively removed from the case of immobilized protein A, protein G or second-
sample. In fact, magnetic separation is the only feas- ary antibodies, biotinylated molecules in the case of
ible method for recovery of small magnetic particles immobilized streptavidin or adenylated molecules in
(diameter ca. 0.1}1 m) in the presence of biological the case of immobilized oligodeoxythymidine). In ex-
debris and other fouling material of similar size. ceptional cases, enzyme activity may decrease as a re-
Moreover, the power and efRciency of magnetic sep- sult of usage of magnetic particles with exposed iron
aration procedures is especially useful for large-scale oxides. In this case encapsulated microspheres, hav-
operations. The magnetic separation techniques are ing an outer layer of pure polymer, are safer. In
also the basis of various automated procedures, espe- Table 1 is given a list of companies producing and
cially magnetic particle-based immunoassay systems selling magnetic particles of various types.
for the determination of a variety of analytes. The In the laboratory, magnetite (or similar magnetic
basic equipment for laboratory experiments is very materials such as maghemite or ferrites) particles are
simple. Magnetic particles of various types are easily usually surface modiRed by silanization. This process
available, as well as magnetic separators. A short modiRes the surface of the inorganic particles so that
description is given below. appropriate function groups become available, which
enable easy immobilization of afRnity ligands.
Biopolymers such as agarose, chitosan, -car-
Equipment and Materials rageenan and alginate can be easily prepared in
Magnetic carriers with immobilized afRnity ligand or a magnetic form. In the simplest way, the biopolymer
magnetic particles prepared from a biopolymer exhi- solution is mixed with magnetic particles and after
biting afRnity for the target compound(s) are used to bulk gel formation the magnetic gel formed is broken
perform the isolation procedure. Magnetic separators into Rne particles. Alternatively, biopolymer solution
are necessary to recover magnetic particles from the containing dispersed magnetite is dropped into a
system. mixed hardening solution or a water-in-oil suspension
Table 1 Examples of commerically available magnetic particles suitable for magnetic affinity separations
Manufacturer/supplier Name of particles Diameter Polymer End groups/ Immobilized ligands

( m) composition/surface activation
modification possibility
Advanced Biotechnologies, XM200 Microsphere 3.5 Polystyrene }COOH Oligo (dT), antibodies,
Epsom, UK streptavidin, protein A
Magnacil 0.5}3.5 Silica Silica
Bangs Labs, Fishers, IN, USA Magnetic &1 Styrene}divinyl }COOH, }NH2 Streptavidin, protein
Microspheres benzene copolymer A, antibodies
Cortex Biochem, San Leandro, MagaCell Cellulose }OH Streptavidin, protein
CA, USA A, protein G, oligo
(dT), DEAE, CM, PEI
CPG, Lincoln Park, NJ, MPG 5 Porous glass }SiOH, glyceryl, Streptavidin, avidin
USA }NH2, hydrazide
Dynal, Oslo, Norway Dynabeads M-280 2.8 Polystyrene Tosyl activated Streptavidin, oligo
Dynabeads M-450 4.5 (dT), antibodies
Dynabeads M-500 5.0
Immunotech, Marseille, Iobeads &1 Antibodies, avidin
France
Merck, Darmstadt, Germany Biobeads 1 Polystyrene Streptavidin
1}3 Silica
Novagen, USA Magnetight Oligo (dT)
PerSeptive Biosystems, BioMag 0.5}1.5 Silanized iron oxides }COOH, }NH2 Antibodies, protein A,
Farmingham, MA, USA protein G,
streptavidin, biotin
Prolabo, Fontenay-Sous-Bois, Estapor &1 Polystyrene }COOH, }OH,
France }NH2
Promega, Madison, WI, MagneSphere Streptavidin
USA Paramagnetic
Particles
ProZyme, San Leandro, Magnetic beads 1.4 Styrene}divinyl- Streptavidin, protein
CA, USA 2.2 benzene copolymer A, protein A/G, protein
2.3 G
Qiagen Ni-NTA Magnetic 20}70 Agarose Nitrilotriacetic acid
agarose beads
Quantum Magnetics, USA Magnetic particles 0.05 Streptavidin, DEAE,
0.25 CM, C18, protein A,
silica
Scigen, Sittingbourne, UK M 100 1}10 Cellulose }OH Streptavidin, biotin,
M 104 oligo (dT)
M 108
Magnetic agarose 1}5 Agarose
Seradyn, Indianopolis, IN, Sera-Mag 1 }COOH Streptavidin, oligo
USA (dT)
Spherotech, Libertyville, SPHERO magnetic 1}4.5 Polystyrene }COOH, }NH2 Streptavidin, biotin,
IL, USA particles antibodies
technique is used to prepare spherical particles. Basi- available carriers (see Table 1). To immobilize other
cally the same procedures can be used to prepare ligands to both commercial and laboratory-made
magnetic particles from synthetic polymers such as magnetic particles, standard procedures used in afRn-
polyacrylamide or poly(vinylalcohol). ity chromatography can be employed. Usually func-
In one of the approaches used, standard afRnity tional groups available on the surface of magnetic
chromatography material is post-magnetized by particles such as }COOH, }OH or }NH2 are used for
pumping the water-based ferroSuid through the col- immobilization; in some cases, magnetic particles are
umn packed with the sorbent. Magnetic material already available in the activated form (e.g. tosyl
accumulates within the afRnity adsorbent pores thus activated).
modifying the chromatography material into mag- Magnetic separators are necessary to separate the
netic form. magnetic particles from the system. In the simplest
Some afRnity ligands (usually general binding approach, a small permanent magnet can be used,
ligands) are already immobilized to commercially but various magnetic separators employing strong
rare-earth magnets can be obtained at reasonable

prices. Commercial laboratory scale batch magnetic
separators are usually made from magnets embedded
in disinfectant-proof material. The racks are con-
structed for separations in Eppendorf microtubes,
standard test tubes or centrifugation cuvettes. Some
have a removable magnetic plate to facilitate washing
of separated magnetic particles (Figure 1). Other
types of separators enable separations from the wells
of microtitration plates and the Sat magnetic separ-
ators are useful for separation from larger volumes of
suspensions (up to ca. 500}1000 mL).
Flow-through magnetic separators are usually
more expensive and more complicated, and high-
gradient magnetic separators (HGMS) are typical Figure 2 (See Colour Plate 62). A typical example of laborat-
examples (Figure 2). Laboratory-scale HGMS are ory-scale high-gradient magnetic separators. OctoMACS Separ-
constructed from a column packed with Rne mag- ator (Miltenyi Biotec, Germany) can be used for simultaneous
netic-grade stainless-steel wool or small steel balls isolation of mRNA. Courtesy of Miltenyi Biotec, Germany.
placed between the poles of an appropriate magnet.
The suspension is pumped through the column, and
magnetic particles are retained within the matrix. magnetic particles. In case antibodies are used as free
After removing the column from the magnetic Reld, afRnity ligands, magnetic particles with immobilized
the particles are retrieved by Sow and usually by secondary antibodies, protein A or protein G are used
gentle vibration of the column. for capturing the complex. Alternatively, the free
afRnity ligands can be biotinylated and magnetic par-
Basic Principles of Magnetic ticles with immobilized streptavidin or avidin are
used to capture the complexes formed. In both
Af\nity Separations methods magnetic particles with isolated target com-
In general, magnetic afRnity separations can be per- pound(s) are magnetically separated and then a series
formed in two different modes. In the direct method of washing steps are performed to remove the major-
an appropriate afRnity ligand is directly coupled to ity of contaminating compounds and particles. The
the magnetic particles or biopolymer exhibiting afRn- target compound is then usually eluted, but for speci-
ity towards target compound(s) is used in the course Rc applications (especially in molecular biology, bi-
of preparation of magnetic afRnity particles. These oanalytical chemistry or environmental chemistry)
particles are added to the sample and target com- they can be used still attached to the particles, such as
pounds then bind to them. In the indirect method the in the case of polymerase chain reaction, magnetic
free afRnity ligand (in most cases an appropriate ELISA, etc.
antibody) is added to the solution or suspension to The two methods perform equally well, but, in
enable the interaction with the target compound. The general, the direct technique is more easily controlled.
resulting complex is then captured by appropriate The indirect procedure may perform better if afRnity
ligands have poor afRnity for the target compound.
In some cases, nonspeciRc binding of accompany-
ing compounds can be observed due to the speciRc
properties of the magnetic particle material. In this
case, pretreatment with the magnetic carrier without
immobilized afRnity ligand or with immobilized non-
speciRc molecules will usually help. The nonspeciRc
binding can be also minimized by adding a nonionic
detergent both in the sample and in the washing
buffers after isolation of the target.
In most cases, magnetic batch afTnity adsorption is
used to perform the separation step. This approach
Figure 1 (See Colour Plate 61). Examples of test-tube mag- represents the simplest procedure available, enabling
netic separators (Dynal, Norway). Left, Dynal MPC-6; right, Dynal the whole separation to be performed in one test-tube
MPC-1. Courtesy of Dynal, Oslo, Norway. or Sask. If larger magnetic particles (with diameters
ca.'1 m) are used, simple magnetic separators can At present, magnetic afRnity separation techniques
be employed. If magnetic colloids (diameters ranging are used especially in molecular biology for the isola-
between tens and hundreds of nanometers) are used tion of RNA, DNA and oligonucleotides. Almost all
as afRnity adsorbents, high-gradient magnetic separ- the procedures employ the same basic principle,
ators have usually to be used to remove the magnetic based on the hybridization of immobilized oligo-
particles from the system. nucleotides and target structures. There are several
Alternatively, magnetically stabilized Uuidized companies offering oligodeoxythymidine immobi-
beds (MSFB), which allow continuous separation, lized on magnetic particles, which can be effectively
can be used. The use of MSFB is an alternative to used for rapid isolation of highly puriRed, intact
conventional column operation, such as packed bed poly A#mRNA from eukaryotic total RNA. Poly
or Suidized bed, especially for large-scale puriRcation A#mRNA has been successfully isolated from vari-
of biological products. Magnetic stabilization enables ous samples, such as cells, animal and plant tissues,
the expansion of a packed bed without mixing blood, cells isolated by immunomagnetic separation,
of solid particles. High column efRciency, low pres- parafRn-embedded tissues, etc. Also cells and tissues
sure drop and elimination of clogging can be attained. containing high RNase activities can be used for
Biocompatible two-phase systems, composed for mRNA isolation. The separated mRNA can be eluted
example from dextran and polyethylene glycol, are from the beads by lowering the ionic strength of the
often used for isolation of biologically active com- elution buffer and used for further applications
pounds, subcellular organelles and cells. The separ- (Northern blotting, dot}blot hybridization, hybrid-
ation of the phases can be accelerated by the addition ization probes) or used still bound to magnetic par-
of Rne magnetic particles or ferroSuids to the system ticles (cDNA synthesis, construction of solid-phase
followed by the application of a magnetic Reld. Mag- cDNA library, etc.). Enzymatic downstream applica-
netically enhanced phase separation usually increases tions are usually not inhibited by the presence of
the speed of phase separation by a factor of about 10 magnetic particles. The covalent binding of oligodeo-
in easy systems, but it may increase by a factor of xythymidine to magnetic particles makes it possible
many thousands in difRcult systems. to regenerate the speciRc adsorbent and to reuse it up
to four times.
Examples of Magnetic Af\nity Isolation of DNA and RNA can be simply per-
Separations of Biologically Active formed using biotinylated speciRc nucleic acids or
oligonucleotides immobilized on magnetic particles
Compounds and Xenobiotics with immobilized streptavidin. Usually large binding
Magnetic afRnity separations have been successfully capacity can be achieved resulting in excellent reac-
used in various areas, such as molecular biology, tion kinetics and high efRciency of the procedure. In
biochemistry, immunochemistry, enzymology, ana- addition, magnetic silica particles have been used for
lytical chemistry and environmental chemistry. simple isolation of DNA and RNA from various bio-
Tables 2+4 show some selected applications of these logical samples and also to purify DNA fragments
techniques. after agarose gel electrophoresis.
Table 2 Typical examples of magnetic separations of nucleic acids
Nucleic acid Magnetic system used Typical examples
RNA Magnetic particles with immobilized oligo (dT)25 Eukaryotic poly A# mRNA, viral poly
A# RNA
Magnetic particles with immobilized specific oligonucleotides tRNA
DNA Dynabeads DNA DIRECT PCR-ready DNA

Biotinylated cloned genomic DNA immobilized on Dynabeads cDNA
Streptavidin
Dynabeads M-280 Streptavidin with immobilized biotinylated M13 single-stranded DNA
oligonucleotide complementary to the lacZ region
Magnetic particles with immobilized pyrimidine oligonucleotide Double-stranded target DNA (triple
helix formation)
COOH-terminated magnetic beads (under specific concentrations Double-stranded DNA, PCR products,
of polyethylene glycol and salt) M13 single-stranded DNA
Magnetic silica particles DNA
Table 3 Selected examples of magnetic affinity separation of proteins
Isolated protein Magnetic system used Typical examples
Enzymes Sub-micron ferrite particles with immobilized soybean trypsin Trypsin

inhibitor
Ferrofluid-modified 5-AMP-Sepharose 4B Alcohol dehydrogenase
Magnetic agarose beads with immobilized dye Lactate dehydrogenase
Magnetic chitin Lysozyme
Dynabeads with immobilized polyclonal antibodies Angiotensin-converting enzyme
Iminodiacetic acid coupled to magnetic agarose and charged AngioI-TEM--lactamase
with Zn2#
Alginate-magnetite beads Pectinase
Magnetic affinity aqueous two-phase system Hexokinase
Antibodies Magnetic particles with immobilized antibodies against human Human IgG antibodies
IgG
Magnetic particles with immobilized protein A or protein G Antibodies
Human serum albumin immobilized onto ferromagnetic dacron Antibodies against human serum albumin
Lectins Magnetic cross-linked chitosan Solanum tuberosum lectin
Receptors Dynabeads M-450 sheep anti-mouse IgG1 with immobilized Human transferrin receptor
monoclonal antibody
Magnetic particles with immobilized oligonucleotide containing Ecdysteroid receptor (EcdR) from
EcdR binding sequence Drosophila melanogaster
Other proteins Magnetic particles with immobilized DNA/RNA fragment DNA/RNA binding proteins
containing the specific binding sequence
Magnetic particles with immobilized m-aminophenylboronic acid Glycated haemoglobin
Organomercurial-agarose magnetic beads Transcriptionally active chromatin restriction
fragments with accessible histone H3 thiols
Ni-NTA Magnetic agarose beads 6xHis-tagged proteins
Table 4 Selected examples of magnetic separation of low-molecular-weight biologically active compounds and organic and inorganic
xenobiotics
Type of compound Magnetic system used Typical examples
Biologically active compounds Magnetic particles with immobilized Aldosterone

aldosterone antiserum
Molecularly imprinted polymer containing (S)-propranolol
magnetic iron oxide
Magnetic charcoal Separation of free antigens in
radioimmunoassays
Organic xenobiotics Magnetic particles with immobilized Cu- Polyaromatic hydrocarbons, triphenylmethane
phthalocyanine dyes
Magnetic polyethyleneimine microcapsules Carcinogens
Magnetic particles with immobilized specific Pesticides, polyaromatic hydrocarbons, TNT,
antibodies PCBs
Bacterial cells adsorbed to magnetite Chlorinated hydrocarbons and pesticides
Magnetic charcoal Water-soluble dyes, pesticides
Inorganic xenobiotics Magnetic chitosan Cu2# ions

Cells of Enterobacter spp. immobilized on Ni2# ions
magnetite
Magnetic cross-linked cell walls of Ions of heavy metals
Saccharomyces cerevisiae
Magnetotactic bacteria Ions of heavy metals
Solvent extractants on magnetic Radionuclides
microparticles
In the case of protein separation no simple strategy under development. Due to the commercial availabil-
for magnetic afRnity separations exists. Various afRn- ity of magnetic afRnity particles and kits these tech-
ity ligands have been immobilized on magnetic par- niques are currently used mainly in molecular biology
ticles, or magnetic particles have been prepared from (especially for separation of nucleic acids) and as
biopolymers exhibiting afRnity for target enzymes or parts of the kits for the determination of selected
lectins, as shown in Table 3. Immunomagnetic par- analytes using magnetic ELISA and related tech-
ticles, i.e. magnetic particles with immobilized speci- niques (especially determination of clinical markers
Rc antibodies against the target structures, have been and environmental contaminants). Up to now small-
used for the isolation of various antigens, both mol- scale separations prevail and thus the full potential of
ecules and cells and can thus be used for the separ- these techniques has not been fully exploited.
ation of speciRc proteins. Enzyme isolation is usually It can be expected that further development will be
performed using immobilized inhibitors, cofactors, focused on two areas. The Rrst one will be oriented to
dyes or other suitable ligands, or magnetic beads the laboratory-scale application of magnetic afRnity
prepared from afRnity biopolymers are used. A gen- separation techniques in biochemistry and related
eral procedure, especially from the point of view of areas (isolation of a variety of both low- and high-
recombinant oligohistidine-tagged proteins, is based molecular weight substances of various origins dir-
on the application of metal chelate magnetic adsor- ectly from crude samples thus reducing the number of
bents. Another general procedure employs immobi- puriRcation steps) and in biochemical and environ-
lized protein A or protein G for the speciRc separation mental analysis (application of immunomagnetic par-
of immunoglobulins from ascites, serum and tissue ticles for separation of target analytes from a mixture
culture supernatants. followed by their detection using ELISA and related
Magnetic separation of low-molecular-weight bio- principles). Such a type of analysis enables portable
logically active compounds has been used in the assay systems to be constructed for the detection and
course of their determination by various types of determination of environmental contaminants dir-
immunoassays. Usually immunomagnetic particles ectly on site or for near-patient analysis of various
directly capture the target analyte, or magnetic par- disease markers. Alternatively, fully automated sys-
ticles with immobilized streptavidin are used to cap- tems for the detection of clinical markers will be
ture the complex of biotinylated primary antibody constructed.
and the analyte. The separated analyte is then deter- In the second area, larger-scale (industrial) systems
mined using an appropriate method. will be developed and used for the isolation of biolo-
Isolation and separation of organic and inorganic gically active compounds directly from the crude cul-
xenobiotics from environmental and clinical samples ture media, wastes from food industry, etc. It is not
using magnetic techniques may Rnd useful ap- expected that large amounts of low-cost products will
plications in the near future. Immobilized copper be isolated using magnetic techniques, but attention
phthalocyanine dye has an afRnity for planar organic will be directed to the isolation of minor, but highly
compounds, such as polyaromatic hydrocarbons with valuable components present in raw materials. Of
three and more fused aromatic rings in their molecu- course, prices of magnetic carriers have to be
les, and for triphenylmethane dyes, both groups rep- lowered, and special types of low-cost, biotechnol-
resenting real or potential carcinogens and mutagens. ogy-applicable magnetic carriers prepared by simple
Immunomagnetic separation of xenobiotics such as and cheap procedures have to become available.
pesticides, TNT or PCBs is used as a Rrst step in the Magnetic separations could thus be the technique for
course of their immunoassay. the 21st century.
Magnetic solid-phase extraction (MSPE) enables
preconcentration of target analytes (e.g. environ- See Colour Plates 61, 62.
mental contaminants) from large volumes of solu-
tions or suspensions using relatively small amount of See also: I/Affinity Separation. II/Affinity Separation:
magnetic speciRc adsorbent. This procedure can sub- Immunoaffinity Chromatography. Extraction: Solid-
stitute the standard liquid}liquid and solid-phase ex- Phase Extraction; Solvent Based Separation. III/Catalyst
traction procedures. Studies: Chromatography: Isolation Magnetic Tech-
niques; DNA. Immunoaffinity Extraction: RNA. Appendix
1/Essential Guides for Isolation/Purification of Cells:
Future Perspectives
Essential Guides for Isolation/Purification of Enzymes
The isolation and separation of biologically active and Proteins: Essential Guides for Isolation/Purifica-
compounds and xenobiotics using magnetic afRnity tion of Nucleic Acids. Appendix 2/Essential Guides to
techniques are a relatively new approach and still Method Development in Affinity Chromatography.
2170 III / BIOMEDICAL APPLICATIONS / Gas Chromatography ^ Mass Spectrometry
Further Reading processes using magnetic carrier technology. Minerals

Engineering 7: 1039}1056.
Biomagnetic Techniques in Molecular Biology (1998) S[ afar\ mH k I and S[ afar\ mH kovaH M (1997) Overview of magnetic
Technical Handbook, 3rd edn, 184 pp. Oslo: Dynal. separations used in biochemical and biotechnological
Cell Separation and Protein PuriTcation (1996) Technical applications. In: HaK feli U, SchuK tt W, Teller J and
Handbook, 2nd edn, 165 pp. Oslo: Dynal. Zborowski M (eds) ScientiTc and Clinical Applications
HaK feli U, SchuK tt W, Teller J and Zborowski M (eds) (1997) of Magnetic Carriers, pp. 323}340. New York: Plenum
ScientiTc and Clinical Applications of Magnetic Car- Press.
riers, 628 pp. New York: Plenum Press. S[ afar\ mH k I and S[ afar\ mH kovaH M (1999) Use of magnetic tech-
Lundeberg J and Larsen F (1995) Solid-phase technology: niques for the isolation of cells. Journal of Chromatog-
magnetic beads to improve nucleic acid detection and raphy B 722: 33}53.
analysis. Biotechnology Annual Review 1: 373}401. Sinclair B (1998) To bead or not to bead. Applications of
Moffat G, Williams RA, Webb C and Stirling R (1994) magnetic bead technology. The Scientist 12(13):
Selective separations in environmental and industrial 17}19.
BIOMEDICAL APPLICATIONS
and occupational hygiene are illustrated below.

Gas Chromatography ^ Mass Others are exempliRed under the headings sport, en-
vironment, food and forensics. The analysis of bio-
Spectrometry macromolecules and whole cells is another rapidly
expanding Reld but using other mass spectrometric
V. Garner, Stockport, UK and separatory techniques (electrospray, atmospheric
Copyright ^ 2000 Academic Press pressure and matrix assisted laser desorption ioniz-
ation; see also LC}MS, CE}MS). Gas chromatogra-
phy}mass spectrometry (GC}MS) of pyrolysates can
Introduction provide information about otherwise intractible ma-
The tremendous technological developments that terials.
have followed the initial interfacing of gas chromato- The elucidation of mechanisms and biochemical
graphs with mass spectrometers together with the pathways using tracer and labelling techniques is an
phenomenal advances in computerized data handling example of the third category of applications which is
have provided an analytical technique that Rnds prac- also a rapidly expanding area with the wider avail-
tically universal application. The biomedical sciences ability of stable as opposed to radioactive labelled
afford an enormous range of applications of this compounds. The ability to separate and distinguish
instrumentation where its full potential as a primary between components in a complex mixture that
method for the separations and identiRcation of ex- differ solely in their isotopic composition allows
tremely complex mixtures is clearly demonstrated. exogenous materials to be distinguished from endo-
The applications can be grouped into three broad genous species; it also provides a means of improving
categories based upon the nature of the analytes: quantitation (via isotope dilution) and allows dy-
namic studies of metabolism or dysfunction to be
E Small volatile molecules, e.g. metabolites, xeno- undertaken. It is this latter area of GC}SIRMS
biotics (drugs, toxins, etc), food components (GC}stable isotope ratio MS) that will be empha-
E Large labile molecular species, e.g. biomacro- sized.
molecules and even whole cells
E Isotopomers (molecules differing only in isotopic
composition), e.g. tracer studies, isotopic labell- Analysis of Small Volatile Molecules
ing, breath gas diagnostics, natural abundance Instrumentation
studies
The majority of instruments utilize capillary columns
The analysis of small volatile molecules, perhaps after thereby allowing relatively simple connection to the
derivatization, is the major application area; parti- mass spectrometer. Earlier systems used packed col-
cular biomedical applications in clinical chemistry umns that required some form of separator in order
III / BIOMEDICAL APPLICATIONS / Gas Chromatography ^ Mass Spectrometry 2171
to reduce the amount of carrier gas entering the erated in positive ion mode with electron impact
mass spectrometer ionization chamber. A 5% ionization at 70 eV spectra generated can be com-
poly(diphenyldimethylsiloxane) stationary phase is pared with those in the extensive databases organized
frequently used because of its wide general applicabil- by NIST and Wiley. Using computerized Rle-handling
ity and stability. Samples can be introduced using an techniques, spectra can be compared very rapidly and
autoinjector or manually by syringe; headspace gases, a list of possible compounds can be compiled. The
solutions in volatile solvents, solid-phase microex- mass spectrometer can also be operated in negative
traction (SPME) systems and thermal desorption with ion mode which allows improved sensitivity with
cryofocussing can all be used. In some cases it is particular analytes especially with chemical ionization.
necessary or advisable to derivatize the sample in
Applications
order to enhance its thermal stability.
Mass spectrometers of any type can be coupled, Figure 1 shows typical results from a GC}MS study
from huge magnetic sector machines to time-of-Sight using a bench-top quadrupole system. The sample
systems and the small ‘bench-top’ quadrupole or ion was obtained by extraction of serum taken after
trap instruments. When the mass spectrometer is op- administration of an oestradiol prodrug. The upper
Figure 1 GC}MS of serum extract following oestradiol prodrug administration: (top) total ion chromatogram; (middle) mass
chromatogram for m/z"272; (bottom) mass spectrum for peak at Rt"28.76 min and NIST library spectrum for oestradiol.
Figure 2 Mass chromatogram for m/z"74 identifying FAMEs.
Figure 4 Urinary amine concentrations of paired samples.
trace shows the total ion chromatogram (TIC), the

second trace shows the mass chromatogram for traction and derivatization. The graph emphasizes the
m/z"272 which is diagnostic for oestradiol. The difference between the amounts of amine in the
mass spectrum obtained for this component and the paired samples; although the values were well below
NIST library match are the lower traces. regulatory limits, those from the Rfth pair led to a
Use of mass chromatograms of diagnostic ions change in working practice for that donor.
allows facile recognition of homologous series such as Amines can be converted into perSuoroacyl deriva-
fatty acid methyl esters (FAMEs). Figure 2 shows the tives which are more stable thermally and chemically;
mass chromatogram for m/z"74 for the FAME de- such derivatives give improved sensitivity in electron-
rivatives prepared from an archeological sample; this capture detectors and this is also manifest in negative
valuable information assisted in the interpretation of ion chemical ionization mass spectrometry. This ap-
the pottery artifact. proach can be adopted for the analysis of materials
The signal at m/z"74 is due to methyl ethanoate that form relatively stable gas phase anions in the
formed in the mass spectrometer by a McLafferty mass spectrometer. Figure 5 shows data from another
rearrangement shown in Figure 3. This process can typical example: prostanoids such as PGF2 are con-
?
occur with any FAME having an available hydrogen verted into the t-butyldimethylsilyl ether/penta-
atom at position 4 and thus provides a useful diagnos- Suorobenzoyl ester derivatives. Under negative ion
tic ion. chemical ionization conditions using either methane
Selected ion recording (SIR) measures the ion cur- or ammonia as reagent gas, the ester function is lost in
rent for a restricted range of ions instead of the whole a fragmentation reaction to form a fragment ion at
spectrum. This is of particular use in biomonitoring m/z"695 corresponding to the silyl ether car-
studies for example where the analyte is well charac- boxylate anion shown in the Rgure. This ion shows
terized from the chromatographic and spectrometric satellites due to silicon and carbon isotopes at 696
point of view, affording improved sensitivity and and 697 that are in accord with calculated distribu-
precision. tions. Detection levels for this analyte are in the low
Figure 4 is a graphical representation of the data pg L\1 (ppm) range for full scan data and fg L\1
from an occupational health study of urinary amines: (ppb) range with selected ion recording.
paired samples of urine taken at the beginning and The Rnal example of this Rrst group of applications
end of a working day were analysed for a particular involving small volatile species crosses the boundary
aromatic amine. A protocol was used that freed into analysis of large intractible materials and con-
the amine from excreted conjugates followed by ex- cerns a pyrolysis study. Occasionally there may be
insufRcient sample to carry out a normal extraction
prior to GC}MS analysis, this is particularly so with
conserved archeological material and microscopic bi-
opsy samples. In such circumstances, microscale
sealed vessel pyrolysis GC}MS can be applied: Fig-
ure 6 shows the TIC obtained from a hair sample
Figure 3 McLafferty rearrangement in a FAME M# to form (2 mg) taken from an Egyptian mummy. The indi-
m/z"74. vidual components of the pyrolysate can be identiRed
Figure 5 Mass spectrum of derivatized PGF2 in negative ion

?
chemical ionization and structure of anion corresponding to
m/z"695.
by comparison with library spectra and interpreted in

terms of the mummiRcation process.
This pyrolysis method can be applied to other com-
plex analytes including whole cells in order to in-
vestigate occluded materials or for the identiRcation
of chemical modiRcation of biopolymers such as starch.
The quantitative analysis of halogenated dibenzo-
dioxins and -furans in biological (and environmental)
matrices is a good example of the combination of
high resolution GC (HRGC) and MS (HRMS) tech- Figure 7 Protocol for pretreatment of dioxin/furan samples
prior to HRGC}HRMS.
nologies using the isotopes of chlorine and carbon
to facilitate identiRcation and quantiRcation. Prior to

any instrumental analysis, the Sesh, vegetation or
other material has to be prepared according to a qual-
ity assured protocol that is summarized in Figure 7.
Intrinsic to the procedure is the use of standards:
the Rrst standard to be added is 13C12-2,3,7,8-
tetrachlorodibenzodioxin upon which quantitation
is based. Another standard, 13C12-1,2,3,4-tetra-
chlorodibenzodioxin is added after the chemical ma-
nipulation of the sample is complete and immediately
prior to GC}MS analysis. Comparison of the signals
from the two standards then allows the efRciency of
the extraction process to be assessed. Typical data
Figure 6 Total ion chromatogram from GC}MS of pyrolysis sets are shown in Figures 8 and 9: both show four
products from archaeological hair sample. traces, the upper two traces are the high resolution
Figure 8 HRGC}HRMS SIR data from dioxin/furan analysis of meat extract.
SIR data for m/z"319.8965 and 321.8936 second group of applications follows a different ap-
corresponding to the molecular ions of tetrach- proach in which isotopomeric components are ini-
lorodibenzodioxins (TCDD), i.e. C12H4(35Cl4)O2 and tially separated then converted into light gases, car-
C12H4(35Cl337Cl)O2 respectively. The lower two traces bon dioxide, hydrogen and nitrogen. It is these latter
in each case correspond to the ions of m/z" materials that are investigated in the isotope ratioing
331.9368 and 333.9339 for the 13C-isotopomers, i.e. mass spectrometer whereby the ratios of carbon, hy-
13
C12H4(35Cl4)O2 and 13C12H4(35Cl337Cl)O2 for the drogen, oxygen and nitrogen isotopes are determined
two standards. with much greater precision.
The data shown in Figure 8 represents TCDD
Instrumentation
levels below regulatory limits whereas those in Fig-
ure 9 (from an environmental sample) were signiR- The design of the mass spectrometer is extremely
cantly higher. simple comprising an electron impact ionization
source, a very stable magnetic analyser and triple
Faraday cup collector (Figure 10A).
Analysis of Isotopomers The combustion interface incorporates a furnace
The above example shows the dual beneRts of containing heated copper oxide which converts or-
high resolution instrumentation and isotope dilution ganic chemicals in the column efSuent into carbon
to effect precise quantiRcation and identiRcation. dioxide and water; a cold trap is used to remove the
There are many other examples of this type of ap- water vapour. Conversion of the individual compo-
plication that relies upon analysis of the intact nents into the same chemical species such as carbon
analyte molecule in the mass spectrometer. The dioxide removes some of the variables that arise from
Figure 9 HRGC}HRMS SIR data from dioxin/furan analysis of an environmental sample.
different sample behaviour and isotopomeric dis- Applications

tribution commonly observed in electron impact The use of stable isotopes such as 2H, 13C, 15N or
ionisation. Simultaneous recording of the ion 18
O instead of the radioactive analogues removes all
beams at m/z"44, 45 and 46 (corresponding to risks associated with radiation but the introduction
12 16
C O213C16O2 and 12C16O18O respectively for car- of an isotopic substitution at the site of a rate-limiting
bon dioxide) using three separate Faraday collectors reaction can introduce kinetic isotope effects. Thus an
minimizes background signal Suctuation. These dif- artiRcial change to the natural isotopic distribution
ferences result in a dramatic improvement in the could in theory alter the kinetics of the biochemical
stability of the signal allowing determination of iso- process and thereby affect the overall metabolic rate.
topic ratios at about 10\5 at%. Figure 10B shows Many theoretical calculations and experimental ob-
a graphical output from the three collectors for three servations have been made: the greatest differences
reference gas pulses and one sample. occur with hydrogen replacement with a maximum
The units used to express the relative difference in relative rate ratio (kH/kD) of 18. The differences with
isotopic abundances are either atoms% (at%) or other isotopic replacements are much smaller: 12C/13C
delta (, per mil or ). up to 1.25; 14N/15N, 1.14; 16O/18O, 1.119 and 32S/34S,
1.05. This particular risk is however even smaller

no. of minor atoms than that with the corresponding radiolabels whose
At%" ;100 values are 1H/3H, 60 (max) and 12C/14C, 1.5.
no. of major atoms
13
(sample ratio!reference ratio) Breath Gas Testing Using CO2
" ;1000
reference ratio The basic principle is illustrated in Figure 11.
or blood, gastric Suid or muscle. These latter requir-

ing clinical intervention.
The 13C-urea breath test for Helicobacter pylori A

characteristic of this bacterium is its high urease ac-
tivity; this is exploited by administering a solution of
13
C-urea (1 mg kg\1 body weight, ca. 75 mg for an
average adult) and investigating the effect on the level
of 13CO2 in the breath after a short time interval of
about 20}30 min. If the bacterium is present, the urea
is converted into bicarbonate which appears in the
breath as 13CO2; in its absence the isotopic distribu-
tion remains unchanged (Figure 12).
Typical results from a group of untreated and
treated patients are shown in Figure 13: differences
between the pre- and post-breath samples (referring
to 13C-urea administration) of more than 3 are
taken to indicate infection.
Other metabolic breath tests Introduction of a var-

iety of 13C-labelled substrates into a test meal that is
Figure 10 (A) Schematic representation of a GC}SIRMS sys- ingested by the patient can be used to monitor par-
tem (B) Graphical output from SIRMS of two reference gas
ticular metabolic functions in vivo without the need
pulses, sample, reference gas pulse.
for invasive surgery and the collection of biopsy sam-
ples and subsequent in vitro biochemical investiga-
Oral administration of either a solution of the tion. Thus gastric emptying rates can be measured
13
C-labelled substrate in water or fruit juice or a sus- using octanoic acid labelled at position 1 with 13C;
pension in a Savoured colloidal preparation is usually abnormalities in the cytochrome P450 pathway in
preferred to intravenous injection to emphasize the liver metabolism can be determined using a variety of
‘non-invasive’ approach. Breath gas is the easiest substrates shown in Figure 14 with typical data
means of sample collection. However, there are shown in Figure 15.
others including saliva, urine, faeces, milk, hair, nails These data show that measurements taken after
a time interval of one hour (after dosing with the
labelled substrate) allow unambiguous detection of
abnormality. A similar distinction is possible for lipid
malabsorption using a range of 13C-labelled triacyl
glycerides. The results summarized in Figure 16 are
from a study of absorption of 13C-dodecanoic acid
introduced intraduodenally.
Tracer studies
The facile determination of carbon isotope ratios in
speciRc compounds extends the scope of tracer stud-
ies to include assessment of metabolic rates for a wide
variety of substrates. Thus administration of speciR-
cally labelled lipids, amino acids and carbohydrates
followed by GC}SIRMS analysis of fractionated
serum samples allows not only the identiRcation of
Figure 11 Schematic diagram of procedure for breath gas Figure 12 Chemical basis of the Helicobacter pylori breath
testing. test.
Figure 13 Typical results from Helicobacter pylori breath tests.
metabolic products but also the measurement of turn- detect adulteration and authenticity of foodstuffs and
over rates of those substrates. The sensitivity of the identify migration patterns of insects.
system is such that measurements can be made using The variation in 13C content of a variety of natural
enriched substrates (as opposed to fully labelled com- materials is summarized in Figure 18. Plants convert
pounds). Figure 17 summarizes the results of a study carbon dioxide into carbohydrates by two main
using 13C-enriched maize sugar administered at photosynthetic mechanisms: the Calvin}Benson cycle
1 g kg\1 body weight then timed blood samples puri- or the Hatch}Slack pathway. Plants such as rice,
Red and analysed for enrichment of plasma glucose wheat, potatoes, beet or brassicas utilize C3 inter-
by GC}SIRMS. mediates and result in depleted 13C content (i.e. delta
The data demonstrated that even at relatively low values are more negative) whereas maize, sugar cane
levels of enrichment, the appearance rate could be and tropical fruits use C4 intermediates that demon-
measured. strate higher 13C content. The differences are large
enough to be translated through the food web and be
Natural abundance measurements
measured by GC}SIRMS.
Primary and secondary kinetic isotope effects, al- The average -13C value of honey, produced from
though small, lead to fractionation of isotopomers Sowers of C3 plants by bees, is signiRcantly lower
and the observed natural variations in isotopic than that of high fructose corn syrup (HFCS, produc-
abundances. These temporal and geographical varied from maize, a C4 plant). Hence, it is possible not
ations can be measured using SIRMS and used to only to identify adulteration of honey with HFCS but
also to determine its extent.
Many natural products, e.g. vanillin, are also avail-
able as synthetic products from chemical or bio-
chemical syntheses; the distinction between natural,
nature-identical and synthetic can be determined
using GC-SIRMS. The natural product obtained by
extraction has a generally higher -13C value (ca.
!20) compared to fossil fuel derived material (ca.
!30) or fermentation product (ca. !35).
The method can be used to distinguish exogenous
materials from their endogenous counterparts; this is
particularly useful in the detection of anabolic ster-
oids’ administration in meat products, animals,
Figure 14 Substrates for breath tests of liver metabolism. racehorses or athletes where conventional GC}MS
Figure 15 Comparison of data from patient and control for liver metabolism.
Figure 16 Appearance data for breath gas analysis following 13C-dodecanoic acid administration in a study of lipid malabsorption.
Figure 17 Appearance rates from a GC}SIRMS study of plasma glucose.
protocols have been shown to be inadequate. When m/z"3 due to H# 3 arising from an ion molecule
the carbon isotope ratios for the major metabolites of reaction in the mass spectrometer source. This latter
testosterone, for example, are normalized to an endo- problem is overcome by application of a correction
genous reference compound such as cholesterol, factor calculated from data from reference gas pulses.
the differences between testosterone-treated and un- The sample (e.g. insect wings, ca. 1 mg) is pyro-
treated animals is statistically signiRcant. lysed in a helium stream in a furnace containing
It is not only carbon that shows variability in its quartz chips and nickel carbide at about 10003C.
isotopic distribution; hydrogen and oxygen also show The pyrolysis products, hydrogen, carbon monoxide
geographical variability that is correlated to rainfall and nitrogen, are separated using a packed column
latitude. Continuous-Sow pyrolysis measurements of (1.5 m, molecular sieve 5 A> ) then introduced into the
D and 18O have been used to study animal migra- mass spectrometer.
tion patterns using a development of the technology Reference pulses of hydrogen and carbon monox-
described above. The use of helium as a carrier gas ide are introduced prior to the sample peaks and used
means that there is a large signal generated for He# to calculate the D and 18O values for m/z 3/2 and
at m/z"4 and this needs to be totally resolved from m/z 30/28 and 29/28 (for carbon monoxide). A typi-
the relatively small signal at m/z"3 due to HD#. cal data set is:
This is further complicated by the presence of ions at
Sample 1 Sample 2 Sample 3
Latitude 303 403 503

D precipitation !15 !33 !80
D sample !78 !92 !125
18O sample #5.7 #3.8 #0.3
Conclusion
GC}MS will continue to provide much high quality
and fundamental analytical information that will
allow the biomedical sciences to develop along the
avenues described above for some time. Improve-
Figure 18 Natural variation in 13C content () of natural mater- ments in the technology of extraction and sampling
ials. will allow progressively smaller samples to be used.
2180 III / BITUMENS: LIQUID CHROMATOGRAPHY
The availability of relatively inexpensive and safe See also: II/Chromatography: Gas: Detectors: Mass
isotopically labelled substrates and precursors will Spectrometry. III/Drugs of Abuse: Solid-Phase Extrac-
beneRt tracer studies and allow improved precision in tion.
quantitative measurements of a wide variety of
analytes. GC}SIRMS will facilitate dynamic studies
of metabolism and migration. The capacity to distin-
Further Reading
guish exogenous from endogenous materials is bound Chapman JR (1993) Practical Organic Mass Spectrometry.
to be a signiRcant and expanding area of application A Guide for Chemical and Biochemical Analysis, 2nd
that will Rnd great use in the identiRcation of abuse, edn. Chichester: John Wiley.
adulteration and authenticity. Johnstone AW and Rose ME (1996) Mass Spectrometry for
Chemists and Biochemists. 2nd edn. Cambridge: Cam-
bridge University Press.
Acknowledgements Krumbiegel P (1991) Stable Isotope Pharmaceuticals for
Clinical Research and Diagnosis. (Drug Development
The following are acknowledged: Nick Ordsmith and and Evaluation), vol.18. Jena: Fischer.
Keith Hall from Hall Analytical Laboratories Ltd. McLafferty FW and Turecek F (1993) Interpretation of
Andy Phillips from Micromass Ltd. Nick Bukowski Mass Spectra. 4th edn. Sausalito: University Science
from ThermoQuest. Books.

See III / CLINICAL CHEMISTRY: THIN LAYER (PLANAR) CHROMATOGRAPHY
BITUMENS: LIQUID CHROMATOGRAPHY
K. Dunn, G. V. Chilingarian and T. F. Yen, Carbon and hydrogen constitute 80}95% of the resin
University of Southern California, Los Angeles, and asphaltene molecules, with oxygen always pres-
CA, USA ent. Sulfur, nitrogen and metals are also usually pres-
Copyright ^ 2000 Academic Press ent. The content of resins in crude oil (2}40%)
exceeds by far (from 3 to 40 times) the content of
asphaltenes (trace to 6%; usually (1%). Carboids
Introduction and carbenes, which resemble asphaltenes, differ
Great confusion exists in the literature on the deRni- from them by having a higher oxygen content. How-
tion of the term bitumen. It occurs in the earth’s crust ever, they are absent in crude oils and distillate cuts.
in various forms: Rrstly, in a dispersed state, in trace In small quantities, they are found in residues from
quantities; and secondly, as accumulations, where vacuum distillation, in cracked tars, and in native and
bitumen either impregnates the rock or occurs in petroleum asphalts.
a pure or nearly pure form. In the Rrst half of the Bitumens and the related heat-treated asphalt are
20th century, the term bitumen was applied to a very complex mixture of compounds with a wide
crude oil and its natural derivatives (maltha, asphalt, range of molecular weights, which are difRcult to
ozokerite). In this original meaning, bitumens are isolate into classes of pure compounds. Resins are
called naphthides. The term bitumen is also used for semi-liquid (sometimes almost solid) dark brown to
asphalts and asphalt-like manufactured substances black substances which have a speciRc gravity around
utilized in road construction, including products from 1 and a molecular weight of 500}2000 (usually
the processing of coals and/or oil shales. In crude oils, 600}1000). Asphaltenes, on the other hand, are dark,
in distillate cuts beginning with kerosene, and in amorphous powders and have a speciRc gravity
distillation residues, there is a group of high molecu- greater than 1 and molecular weights of 1000}10 000
lar weight heteroorganic compounds that are lumped (average 5000). Using various methods, the observed
under the name of resins and asphaltenes. Their molecular weights suggest that asphaltenes form
content may be as large as 40% in heavy crude oils. molecular aggregates or colloids, even in dilute
III / BITUMENS: LIQUID CHROMATOGRAPHY 2181
Table 1 Unit molecular weight of asphaltenes
Method Particle weight Researcher
Viscosity 900}4000 Fischer and Schrem (1959)

High angle X-ray 805}3360 Yen et al. (1961)
Light absorption 1000}4000 Markhasin et al. (1966)
Electron microscopy 3440}5430 Dickie et al. (1969)
VPO 2950}8130 Dickie and Yen (1967)
VPO 2000}4600 Dickie and Yen (1967)
VPO 1220}2160 Speight et al. (1985)
Cryoscopic, naphthalene 1700 Hilman and Barnett (1937)
Cryoscopic, phenathrene 2500 Boyd and Montgomery (1962)
Cryoscopic, benzene 5000}6000 Katz (1934)
solutions. The results are inSuenced by solvent polar- particular physical and chemical properties. Depend-
ity, asphaltene concentration and the temperature ing on the solubility in certain solvents, asphalt or
used. Different molecular weights of asphaltenes can bitumen is generally fractionated into four main frac-
be obtained by different instruments. For example, tions: saturates, aromatics, resins and asphaltenes.
the data obtained from solution viscosity and cryo- The polarity of these fractions increases in the follow-
scopic method are low, whereas those obtained from ing order: saturates(aromatics(resins(asphal-
the ultracentrifuge and electron microscope are high. tenes. A detailed scheme for bitumen fractionation is
Vapour pressure osmometry (VPO) is considered to shown in Figure 1 and explained as follows:
be the most accurate method at present when a good
1. Gas oil is propane-soluble and n-pentane-insoluble.
solvent is used to disperse the asphaltenes. Because
2. Resins are n-pentane-soluble but propane-insol-
the solvent affects the degree of asphaltene aggrega-
uble.
tion, the true molecular weights of asphaltenes are
3. Asphaltenes are toluene-soluble but n-pentane in-
generally much lower than those measured. Some of
soluble.
these results are listed in Table 1.
4. Preasphaltenes are insoluble in both n-pentane
The asphaltene content is a variable value depend-
and toluene.
ing on the relative amounts and characteristics of the
source material and on the procedure adopted for Asphaltenes are multipolymer systems containing
separation. The physical and chemical characteristics a great variety of building blocks. The statistically
are of considerable importance: the higher boiling average molecule contains a Sat sheet of condensed
point components of crude oil can precipitate in cer- aromatic systems that may be interconnected by sul-
tain solvents, whereas the lower boiling point compo- Rde, ether, aliphatic chains or naphthenic ring link-
nents (cyclic and aromatic compounds) are soluble in ages. Gaps and holes appear as defect centres in the
solvents. Most of the studies on bitumens and as- aromatic systems most likely involving free radicals.
phalts are concentrated on groups of compounds with Heterocyclic atoms may be coordinated to transition
Figure 1 Fractionation scheme of bitumen or asphalt.

metals such as vanadium, nickel and iron. Approx- Resins are considered smaller analogues of asphal-
imately Rve of these sheets are associated by } tenes with a much lower molecular weight. The resins
interaction. They are stacked one above the other: the contain aromatic compounds substituted with longer
distance between the sheets ranges from 0.36 to alkyl chains and a greater number of side chains
0.38 nm, giving an overall height for a stack of attached to the rings than asphaltenes. The combina-
1.6}2.0 nm; the average sheet diameter appears to be tion of the saturated and aromatic characteristics of
0.85}1.5 nm. The hypothetical structure of an as- resins stabilizes the colloidal nature of asphaltenes
phaltene is shown in Figure 2. present in the oil.
Figure 2 (A) The classification of asphaltene structure. (B) The structure of asphaltene stack. (C) The structrure of one sheet of
asphaltene stack.
The gas oils constitute the lowest molecular frac- bitumen or asphalt separation (SARA method, centri-
tion of the asphalt and serve as the dispersion medium fugal TLC, TLC-FID and the combination of silica gel
for the peptized asphaltenes. Structurally, gas oils chromatography and ion exchange) are described in
consist mostly of naphthenic}aromatic nuclei with detail in the following sections.
a greater proportion of side chains than the resins.
Alkyl naphthenes predominate and straight chain al-
kanes are rarely present. The naphthenic content is
The SARA Method
15}50%, with naphthenics containing two to Rve Background
nuclei per molecule.
The fractionation techniques for crude oil involve The SARA method enables the separation of bitumen
solvent fractionation and precipitation, column into groups. The method is an extension of column
chromatography, ion exchange and complexation chromatography in order to permit more standard-
methods. No matter what method is used, the impor- ized separation. Two packed columns are employed:
tant step is the isolation of asphaltenes. Of course, all Firstly, a SARA column (ion exchange resins) and
these separation processes are carried out on the secondly, an activated silica gel column. The resin
heavy fractions of petroleum only; usually the volatile fractions are irreversibly retained in the SARA col-
fractions have been removed by distillation. umn, whereas the gas oil (saturates and aromatics)
The most popular method of column chromatogra- fraction remains in the solvent after elution of this
phy is the fractionation into saturates, aromatics, column. The gas oil fraction can be further separated
resins and asphaltenes (SARA). Solvent fractionations into saturates and aromatics by the activated silica gel
have been used in the industry since 1930, becoming column. A Sow diagram of the SARA method is given
widely accepted between 1950 and 1970. Recently, in Figure 3. Although the SARA method has been
centrifugal thin-layer chromatography (TLC) has considered as a standard separation technique for
been used to separate crude oil successfully. This asphaltenes, there are some drawbacks to prevent it
method reduces the separation time and cost, as well from becoming popular: it is time-consuming (it takes
as preserving each fraction for further study. TLC at least 1 week) and there is incomplete recovery
with Same ionization detector (TLC-FID) can achieve (resin fraction is difRcult to recover).
quantitative analysis quickly and simply. The combi-
Separation Procedure
nation of solvent fractionation by silica gel and ion
exchange chromatography can achieve a more de- Separation of asphaltenes The asphaltene fraction is
tailed separation showing the complexity of petro- separated by a precipitation process. The bitumen
leum. The polar fraction of the crude oil is separated and asphalt samples can be dissolved Rrst in toluene
by ion exchange. For chromatographic methods for to form dispersions for precipitation by another
Figure 3 Flow chart of SARA separation method.

liquid. Asphaltene precipitation occurs when an ex- colourless. Liquid must always be present in the top
cess of n-pentane is added to the toluene dispersion. reservoir. All operations are carried out under nitro-
A toluene}n-pentane volumetric ratio of 1 to 50 is gen. The result of the separation is that the gas oils
required to ensure the precipitation of asphaltenes. (saturates and aromatics) collect in the bottom Sask,
After the precipitaion process, the toluene dispersion whereas the resins remain in the column. The gas oils
can be separated into two fractions: pentane-insol- can be further fractionated into saturates and aro-
uble (asphaltenes) and pentane-soluble (maltenes). matics on a silica gel column using n-hexane as the
To purify asphaltenes, the precipitation process is eluent, with the saturates being eluted Rrst. The cut-
followed by Soxhlet extraction. n-Pentane is used off point of elution between saturates and aromatics
during Soxhlet extraction until it becomes clear. This can be distinguished by UV examination. Under the
requires over 48 h. The maltene fraction can be fur- UV light, the saturates are colourless, whereas the
ther separated by column chromatography. aromatics exhibit a yellow-green colour.
Separation of saturates and resins The SARA col- Separation of resins In the analytical SARA separ-
umn is dry-packed under vibration with various ation, the resin content is obtained from the mass
packings, purged with nitrogen and equilibrated with balance, by deducting the weights of asphaltenes and
n-pentane. The column contains four discretely gas oils from the original weight of the sample. This
packed zones with a sequence of H# cation exchange method is easy and rapid to quantify the resin content
resin, OH\ ion exchange resin, clay-coated with fer- in the bitumen or asphalt sample; however, the fact
ric chloride, and OH\ ion exchange resin, as illus- that resins cannot be collected for further study is
trated in Figure 4. The maltenes present in the sample a drawback. The resin fraction retained in the
are dissolved by the n-pentane, placed in the top SARA column may be semiquantitatively recovered
reservoir and carried into the column. The leaching (approximately 90%) by exhaustive elution of the
process is continued by recycling until the eluate is packing with chloroform. For complete recovery of
Figure 4 Preparative SARA column. (A) Recycle column for separating resins and oils; (B) SARA column packing sequence. All
dimensions are in millimetres. (Modified after Atgelt et al., 1979.)
Figure 5 Stepwise separation of resins. HAcids I are stronger than the corresponding acids II. (Reproduced with modification from
Jewell, 1979.)
resins, individual cation/anion columns for stepwise ent. Petroleum residues are commonly extracted over-
separation are used. This separation process, shown night, whereas samples of tar sands, shale oils and
in Figure 5, includes cation exchange, HCO\ 3 anion coal liquids may require 3}4 days.
exchange, OH\ anion exchange, and clay-FeCl3 com-
plexation. This scheme provides 10 groups of resins,
and this demonstrates the great complexity of petro- Centrifugal Preparative Thin-layer
leum composition. Chromatography
Separation using the SARA method is a time-con-
Background
suming process. The extraction time varies with the
amount and nature of the sample and especially with Hydrocarbon-type fractions from bitumen, heavy
the content of asphaltenes and other insolubles pres- crude and distillation residues can be separated by
TLC. This is a rapid, effective and inexpensive method dissolve the sample is very important. Details are
which has the potential to become the major tool for discussed in the following sections.
semi-quantitative analysis of composition of the heavy
hydrocarbon mixtures. The application of TLC to the Rotor preparation The glass rotor should meet cer-
separation of petroleum products has been successfully tain requirements: chemical resistance to solvents and
employed by many investigators (e.g., Lian et al., developing reagents, ability to withstand temper-
1992). A conventional TLC method can differentiate atures needed for reaction in the rotor, mechanical
(and characterize) among heavy and residue fractions strength, thickness uniformity and low cost. Silica gel
from petroleum, coal liquids, tar sands, bitumens and 60 HF254#366 (with a binder) can be used as an
asphalts. A preparative, centrifugally accelerated, adsorbent coated on the rotor as a thin layer. The
radial, thin-layer chromatograph (the Chromatotron) thickness of the adsorbent layer determines the
is suitable not only for quantitative determination of proper Sow rate of mobile phase and solvent volume.
bitumens and asphalts, but also for the collection of Table 2 shows the interrelationships of adsorbent
various fractions. It is very important to select the layer thickness, solvent volume and Sow rate.
appropriate solvents or solvent mixtures for separation
into four fractions. Figure 6 shows the solubility Sample preparation and application The process of
parameters of common solvents in relation to those adding the sample to the Chromatotron rotor is one
required to dissolve the various asphalt fractions. of the most important steps to ensure the success of
the separation. The solvent used must be highly vol-
atile and nonpolar because solvents with a high
boiling point or high polarity are difRcult to remove
One type of centrifugal TLC equipment for analysis from the adsorbent during sample introduction.
of the constituents of hydrocarbon mixtures and ex- Sample injection is performed by means of a pump
pecially for determination of the composition of bitu- operated at a low Sow rate in order to add a large
men is illustrated in Figure 7. The preliminary prep- volume of sample to the rotor and to achieve vapor-
aration (such as rotor coating, sample preparation ization of the solvent. Sample introduction as a nar-
and application to the rotor) is important in order to row initial band can enhance the resolution greatly. It
achieve good results. Proper preparation of the is necessary to wait for the solvent to vaporize com-
sample and the appropriate selection of solvent to pletely before starting the elution process. If a small
Figure 6 Solubility parameters of common solvents in relation to the values required for dissolution of asphalt fractions. a 95 : 5 (v/v);
b
tetrahydrofuran; c estimated, mixture; d calculated value.
Figure 7 Schematic diagram of the Chromatotron. (Courtesy of Harrison Research.)
amount of solvent is retained in the adsorbent after rotor at a constant speed, which establishes radial
introduction of the sample, it will adversely affect the elution of solvent and sample, forming concentric
separation process due to spreading of the sample in bands of separated substances that move to the edge
the rotor. of the rotor and are channelled by an output tube for
collection. The four steps used in the procedure are:
Chromatotron The Chromatotron is a glass rotor
1. Petroleum ether elution of saturated and mono-
coated with a thin layer of adsorbent. The mobile
and diaromatic hydrocarbons.
phase and solutions of compounds are introduced to
2. Petroleum ether}ether mixture (80/20, v/v) elu-
the adsorbent through an inlet, which delivers them
tion of polyaromatic compounds and resin-1 frac-
close to the centre of the rotor. A motor drives the
tion. This solvent is most suitable for ensuring that
Table 2 Interrelationships among the thickness of the adsor-
the aromatic hydrocarbons form a separate con-
bent layer, solvent volume and flow rate using the Chromatotron centric band.
3. Ethyl acetate elution of resin-2 fraction.
Thickness of adsorbent Flow rate Solvent volume 4. Tetrahydrofuran elution of asphaltenes.
layer (mm) (mL min\1) (mL)
The method can be used for both qualitative
1 2}4 5 and quantitative purposes; the complete process is
2 6}8 10 summarized in Figure 8. It is very important to deter-
4 8}10 20
6 '10 30
mine the cutoff points for the separation process for
collecting four high purity fractions. There are two
Figure 8 Characterization of bitumen or asphalt by preparative method using the Chromatotron.
methods which can be used to identify the cutoff points: thereby succesively vaporized/pyrolysed and the
ionizable carbon is detected at the collector electrode.
1. UV examination: this method is useful for estima-
The FID signals from each fraction are ampliRed and
ting the cutoff point between the saturates and
recorded as separate peaks.
aromatics. To identify this point, the rotor can be
In as much as the bitumens comprise a complex
examined by illumination with UV light at
mixture of various substances which are impossible
254 nm. The aromatics will exhibit a yellow-green
to isolate, a compositional analysis technique is
colour, whereas the saturates are colourless.
generally used whereby a sample is separated into
2. Spot test: a spot test can be used to identify the
four fractions by performing multistage development
cutoff points of saturates, aromatics and resins.
with eluting solvents of varying polarities. The sol-
Spots obtained using a freshly prepared sulfuric
vent elution sequence for asphalt separation is Rrstly,
acid}formaldehde solution (98#2, v/v) have the
hexane for separating saturates; secondly, toluene for
following colours: saturated compounds, colour-
separating aromatics; and thirdly, dichloro-
less; monocyclic aromatics, reddish brown; dicyclic
methane}methanol (95/5, v/v) for separating resins.
aromatics and polycyclic aromatics, dark green.
The order of eluted components from the top of the
rods is: saturates, aromatics, resins and asphaltenes.
Thin-layer Chromatography with
Analysis
Flame Ionization Detection
TLC-FID is a fast, multiple-analysis method which
Background
can be performed simultaneously on 10 samples. The
The Rngerprinting of asphalt and its components can total elapsed time for the analysis is less than 2 h
be made by TLC-FID much more rapidly and simply (compared to about 2 days per sample with the SARA
than by classical column chromatography and prep- procedure). It is, however, destructive and does not
arative TLC. TLC-FID involves separation of solvent- yield fractions for further examination.
extractable organics on silica-coated quartz rods Linearity of response versus concentration is
called Chromarods into saturates, aromatics and important in obtaining quantitative results. The
polar component classes combined with semiquan- sample size, the Sow rate of hydrogen fed to the
titative detection by automated FID. FID and the speed of scanning all have an effect on
the linearity of response versus concentration, base-
Separation
line stability of the FID signal and reproducibility of
The working principle of a commercially available response values.
instrument, the Iatroscan TH-10, designed to scan the
adsorbent-coated Chromarods, is illustrated in Fig-
ure 9. A set of up to 10 coated rods is mounted on rod
Silica Gel and Ion Exchange
holders used for both chromatography and sub- Chromatography + Separation of
sequent scanning. The samples to be analysed are Polar Fractions
dissolved in a solvent and applied near one end of the
Background
coated rods, which are then developed using suitable
solvents, as in conventional TLC. Thereafter the de- The separation and study of polar compounds in
veloping solvent is removed by evaporation, and crude oil and bitumens is of considerable importance
Rnally the coated rods are scanned in the FID device. in the Relds of petroleum recovery and processing.
The fractions resolved on each of the Chromarods are Numerous methods for separation of polar compounds
Figure 9 Construction of a coated-rod TLC scanner, latroscan TH-10 Mk III. (Modified from Mukherjee, 1991.)
from crude oil and bitumens have been developed. chromatography. The Rrst stage separates the crude
Among these, the most widely used is the method oil or bitumen into four fractions (I}IV: Figure 10).
developed by the Bureau of Mines and American The polar fraction obtained from the Rrst stage is
Petroleum Institute. The acid and base fractions of further split into subfractions by the second-stage
heavy petroleum distillates or residues are isolated by chromatography.
ion exchange chromatography, the neutral nitrogen
compounds by complexation chromatography on Silica gel chromatography Fractionation of crude
ferric chloride and saturates, and mono-, di- and oil or bitumen is carried out by solvents with increas-
polyaromatics by adsorption chromatography on ac- ing polarity through a column using Baker reagent
tivated alumina or on a combined alumina}silica gel grade, 60}200 mesh silica gel. The silica gel is ther-
column. The deRnition of acids or bases by ion ex- mally activated before use. The sample (10}20 g) is
change methods is based on the hydrogen donating or dissolved in 200 mL tetrahydrofuran (THF) and then
accepting tendency of the molecules. In as much as Rltered. The volumetric ratio of the sample to the
many polar compounds are amphoteric, their deRni- adsorbent is about 1:35. The columns are eluted with
tion as acids or bases depends on the analytical se- n-hexane, toluene, 4:1 (v/v) toluene}methanol and
quence employed. Although this method is rather 2:1 (v/v) toluene}methanol, to obtain fractions I, II,
complex and tedious, it is used to separate the polar III and IV, respectively. Figure 10 shows the overall
fractions from crude oil and bitumens. separation scheme. All samples must be evaporated to
a constant weight by (rotary) evaporation and using
a vacuum oven. Recovery of fraction IV is generally
There are two stages in the separation: Rrstly, silica lower than 5% and, therefore, this fraction has not
gel chromatography and secondly, ion exchange been considered for further study. The most polar
Figure 10 Solvent fractionation scheme using a silica gel column for crude oil and shale oil.
Figure 11 Separation of fraction III (see Figure 10) to subfractions by ion exchange chromatography: A, acid; B, base; and
N, neutral.
groups are present in fraction III because of the high used. Two neutral fractions can be obtained at this
interfacial tension compared with those for fractions last stage of separation process. A detailed diagram of
I and II. Fraction III is further fractionated by ion the scheme is given in Figure 11.
exchange chromatography.
Ion exchange chromatography The polar material

Conclusion
(fraction III) obtained using the silica gel column is The SARA separation method is at present the most
mixed with cyclohexane. There are three further comprehensive and most widely used method for the
stages of separation: anion exchange, cation ex- study of heavy oil fractions. Despite its shortcomings
change and clay}FeCl3 complexation. In order to ob- (tedious, time-consuming and possible losses), it can
tain acidic fractions from the polar material, a strong- be viewed as a standard by which other methods
ly basic anion exchange resin is used and an eluting should be evaluated. Preparative, centrifugal, TLC
scheme is employed, based on increasing polarity of with the Chromatotron is capable of isolating crude
the solvent. This scheme includes four elution steps: oil and bitumen in the amounts required for further
cyclohexane; toluene; a 3:2 mixture of toluene}meth- identiRcation by other instrumental techniques.
anol; and a 3:2 mixture of toluene}methanol TLC-FID has a great potential of becoming a stan-
saturated with CO2. The Rrst fraction eluted by cyclo- dard method for quantitative evaluation of crude oil
hexane from the anion exchange resin is further frac- and bitumen. It is rapid, accurate, requires only
tionated by passing it through a strongly acidic cation small samples and does not involve tedious proced-
exchange resin. The same elution scheme as in ures in comparison to the SARA method. Solvent
the Rrst elution step is used, with a slight modiRcation fractionation using silica gel rods simply separates
for the cation exchange elution in that the fourth step crude oil and bitumen into four fractions. With auto-
uses a 5.4: 3.6:1.0 mixture of toluene}methanol} matic instrumentation, the method can be used
isopropylamine. Three basic fractions result from routinely for industrial purposes. The combination,
these fractionations. of fractionation by solvents and ion exchange fol-
A column packed with clay}FeCl3 is employed to lowed by complexation, is a good method for study-
separate further the material obtained from the ing the composition of crude oil and bitumen. Ion
cyclohexane elution in the second stage. The solvents exchange processes rely on the great selectivity for
used to obtain the neutral fractions are cyclohexane acidic and basic components in the crude oil. The
and dichloromethane. The volumetric ratio of sample complexation process enables the neutral nitrogen
to adsorbent is around 1:10 for the three adsorbents compounds to be separated. This scheme provides 10
III / CARBAMATE INSECTICIDES IN FOODSTUFFS: CHROMATOGRAPHY AND IMMUNOASSAY 2191
resin fractions, which demonstrates the great com- separation of heavy-end petroleum distillates. Analytical
plexity of crude oil composition. Chemistry 44: 1391.
The choice of a procedure primarily depends on the Katz M (1934) Alberta bitumen. Canadian Journal of Re-
information required. Combining different separ- search 10: 435}451.
ation steps and/or short-cuts to achieve a speciRc Lian H, Lee CZH, Wang YY and Yen TF (1992) Character-
ization of asphalt with the preparative Chromatotron.
purpose is possible.
Journal of Planar Chromatography 5: 263}266.
See also: II/Chromatography: Liquid: Mechanisms: Mukherjee KD (1991) Handbook of Thin-layer Chrom-
Ion Chromatography. Chromatography: Thin-layer atography, pp. 339}350. New York: Marcel Dekker.
(Planar): Modes of Development: Conventional. III/Crude Sadeghi KM, Sadeghi MA, Wu WH and Yen TF (1989)
Oil: Liquid Chromatography. Flame Ionization Detec- Fractionation of various heavy oils and bitumen for
tion: Thin-Layer (Planar) Chromatography. Flash characterization based on polarity. Fuel 68: 782}787.
Chromatography. Metal Complexes. Petroleum Prod- Shue FF and Yen TF (1981) Concentration and selective
ucts: Liquid Chromatography. identiRcation of nitrogen and oxygen-containing com-
pounds in shale oil. Analytical Chemistry 53:
Further Reading 2081}2084.
Suatoni JC and Swab RE (1976) Preparative hydrocarbon
Atgelt KH, Jewell DM, Latham DR and Selucky ML (1979) compound type analysis by high performance liquid
Chromatography in Petroleum Analysis, pp. 186}214. chromatography. Journal of Chromatographic Science
New York: Marcel Dekker. 14: 535.
Boyd ML and Montgomery DS (1962) Structural-group Speight JG, Wernick DL, Gould KA et al. (1985) Molecular
analysis of the asphaltene and resin components of weight and association of asphaltene, a critical review.
Athabasca bitumen. Fuel 41: 335}350. Rev. Inst. Fr. Pe& t. 40: 51}62.
Dickie JP and Yen TF (1967) Microstructures of the asphal- Wang YY and Yen TF (1990) Rapid separation and charac-
tic fractions by various instrumental methods. Analyti- terization of fossil fuels by thin-layer chromatography.
cal Chemistry 39: 1847}1852. Journal of Planar Chromatography 3: 376}380.
Dickie JP, Haller MN and Yen TF (1969) Electron micro- Weinberg VA, White JL and Yen TF (1983) Solvent frac-
scopic investigation on the nature of petroleum asphal- tionation of petroleum pitch for mesophase formation.
tics. Journal of Colloid Interface Science 29: 375}484. Fuel 62: 1503}1510.
Eremenko NA (1990) Petroleum Geology Handbook. Los Yen TF (1990) Asphaltenic materials. In: Encyclopedia of
Angeles, CA: OSI. Polymer Science and Engineering, 2nd edn, pp. 1}10.
Hilman ES and Barnett B (1937) The composition of New York: John Wiley.
cracked and uncracked asphalts. Proc. Am. Soc. Testing Yen TF and Chilingarian GV (1994) Asphaltenes and as-
Materials, 37: 558. phalts 1. In: Developments in Petroleum Science 40A.
Hirsch DE, Hopkins RC, Coleman HJ et al. (1972) Separ- New York: Elsevier Science.
ation of high-boiling petroleum distillates using gradient Yen TF, Erdman JG and Pollack SS (1961) Investigation of
elution through dual-packed (silica gel}alumina gel) ad- the sturcture of petroleum asphaltenes by X-ray diffrac-
sorption columns. Analytical Chemistry 44: 915. tion. Analytical Chemistry 33: 1587}1594.
Jewell DM (1979) Chromatography in Petroleum Analysis, Yen TF, Shue FF, Wu WH and Tzeng D (1983) Ferric
pp. 284}285. New York: Marcel Dekker. chloride-clay complexation method. Removal of nitro-
Jewell DM, Weber JH, Bunger JW et al. (1972) Ion-ex- gen-containing components from shale oil and related
change, coordination and adsorption chromatographic fossil fuels. ACS Symposium Series 230: 457}466.
CARBAMATE INSECTICIDES IN
FOODSTUFFS: CHROMATOGRAPHY
AND IMMUNOASSAY
G. S. Nunes, Federal University of Maranha o/ UFMA, Introduction
SaJ o LuU& s-Ma, Brazil
D. BarceloH , CID/CSIC, Barcelona, Spain Pesticides have received special attention over the
years, due mainly to the problems of environmental
and food contamination. Analytical methods for de-
termining pesticide residues have their main applica-
Copyright ^ 2000 Academic Press tion in the control of food for human consumption,
2192 III / CARBAMATE INSECTICIDES IN FOODSTUFFS: CHROMATOGRAPHY AND IMMUNOASSAY
especially in the control of fruit and vegetables taken place in extraction/clean-up procedures, and in
produced using direct applications of pesticides. the Rnal determination of pesticide residues in food-
Moreover, the determination of pesticide residues in stuffs.
food is fundamental in monitoring and regulatory This article examines recent progress, focusing
programmes. Pesticide residue levels higher than primarily on simpliRed and miniaturized analytical
the maximum residue level (MRL) are monitored methods for determining carbamate insecticides in
through two different but complementary ap- foodstuffs, including fast and simple extraction/clean
proaches: regulatory monitoring focusing on raw ag- up strategies, and the use of IA techniques for detec-
ricultural commodities that measure the levels in indi- tion prior to laboratory analysis.
vidual lots for compliance with legal tolerances, and
total diet study, in which dietary intakes of pesticides
are determined by analysis of fruit and vegetables. Chromatographic Methods for
Carbamates constitute a family of pesticides regis- Carbamate Analysis
tered for use on several crops in South America,
LC Methods
Europe and the USA. Their use for pest control has
progressively increased in recent years, along with Standard LC methods for carbamate determination
organophosphates (OPs), as alternatives to organo- based on Krause’s method generally consist of rever-
chlorine (OC) insecticides. Owing to their broad sed-phase HPLC with gradient elution followed by
spectrum of biological activity, carbamates can be two post-column reactions to yield Suorescent species
used as insecticides, miticides, fungicides, nematicides (Figure 1). The carbamates are hydrolysed with an
and molluscicides. Table 1 shows the structures of alkaline solution (usually sodium hydroxide) to
the most extensively used carbamates, including some a methylamine that is sequentially derivatized in the
N-methylcarbamate (NMC) insecticides. Carbamate presence of o-phthaldehyde (OPA) and mercap-
residues are of concern because of the acute toxicity toethanol (MERC) to create the Suorescent product.
of some compounds } aldicarb and carbofuran ex- Two high pressure pumps are used to introduce the
hibit an LD50 (the dose of a compound that causes post-column reagents. To separate the main NMCs
death in 50% of the organisms to which it has been and their metabolites, a cycle time of 45}60 min is
administered) in rats of 1 and 8 mg kg\1, respective- required. Limits of detection (LODs) in water analy-
ly. Some carbamate residues are suspected carcino- sis are in the range of 1 to 4 ng, which is equivalent
gens and mutagens. Such insecticides act as inhibitors to 2.5 to 10 p.p.b. in the water injected. However,
of the acetylcholinesterase enzymes, and a number of if a preconcentration step is carried out before chro-
adverse effects have been reported in the literature. matographic separation, the LOD value is lower.
Several analytical methods have been proposed for The sensitivity and selectivity of Krause’s method
the separation and quantiRcation of carbamate resi- have allowed its use for the determination of residues
dues in food samples. The carbamate pesticides are of NMCs in fruits and vegetables with great accuracy.
thermally labile and not readily amenable to gas Unfortunately, such methods involves extraction with
chromatography (GC), making the use of the liquid methanol, and large amounts of sample and solvent
chromatographic (LC) techniques preferable. In gen- are used. The clean-up, starting with successive
eral, the time and expenses involved in classical ana- liquid}liquid partition (LLP) steps, and Rnishing with
lytical methods (i.e. sampling, sample preparation elution of the target compounds on a Celite威/char-
and laboratory analysis) have substantially limited coal adsorbent column, is mainly responsible for the
the number of samples that can be analysed in food slowness and tediousness of the method, making the
research and surveys. In addition, the quantity of analysis of a large number of samples impractical.
chemicals and toxic solvents that are used sometimes Usually, it is satisfactorily used as a reference method
offer a risk factor considerably greater than that of in collaborative studies and also to carry out valida-
the pesticide residue to be determined. These disad- tion of new methods. Different procedures for clean-
vantages have emphasized the need for developing up of crop extracts by employing either glass adsor-
fast, easy, robust, sensitive and cost-effective methods bent columns and commercially available solid-phase
capable of being used in the Reld. Instrumental tech- extraction (SPE) cartridges have been compared and
niques that combine these characteristics are slowly a speciRc solid-phase elution protocol employing
starting to appear in pesticide residue analysis. They a solvent polarity gradient proposed. A schematic
include the immunoassay (IA) techniques, immunoaf- outline of the method is illustrated in Figure 2. The
Rnity chromatography and electrochemical/optical extraction is carried out with methanol, and a LLP
biosensors. To avoid the general drawbacks of the step is only necessary if a UV detector is to be used. In
classical methods, signiRcant developments have most cases, recoveries of the more polar compounds
Table 1 Carbamates commonly used in crop protection and some of their breakdown products
Compound name Chemical structure Class IUPAC name
Aldicarb Insecticide 2-Methyl-2-(methylthio)propionaldehyde

o-methylcarbamoyloxime
Aldicarb sulfone Insecticide (metabolite 2-Methyl-2-(methylsulfonyl)propanal

of aldicarb) o-[(methylamino)carbonyl]-oxime
Aldicarb sulfoxide Insecticide (metabolite 2-Methyl-2-(methylsulfinyl)propanal

of aldicarb) o-[(methylamino)carbonyl]-oxime
Bendiocarb Insecticide 2,3-Isopropylidenedioxyphenyl

methylcarbamate
Carbaryl Insecticide 1-Naphthyl methylcarbamate
Carbendazim Insecticide 2-Methyl benzimidazol-2-ylcarbamate
Carbofuran Insecticide 2,3-Dihydro-2,2-dimethylbenzofuran-

7-yl-methylcarbamate
Methiocarb Insecticide, acaricide, 4-Methylthio-3,5-xylyl methylcarbamate

molluscicide
Methomyl Insecticide S-Methyl N(methylcarbomoyloxy)

thioacetimidate
1-Naphthol Insecticide (metabolite 1-Naphthalenol

of carbaryl)
Table 1 Continued
Compound Name Chemical structure Class IUPAC name
Oxamyl Insecticide, nematicide, N,N-Dimethyl-2-methylcarbamoyloxime-

acaricide 2-(methylthio)acetamide
Pirimicarb Insecticide 2-Dimethylamine-5,6-dimethylpirimidine-

4-dimethylcarbamate
Propoxur Insecticide 2-Isopropoxyphenyl-N-methylcarbamate
are higher if the partitioning step is omitted and three other pesticides in a sample matrix of rice grain.
Suorimetric detection is employed. The method performance was comparable to conven-
A series of tests to assess the feasibility of using tional clean-up procedures.
activated carbon membranes in the clean-up of veg- Following the general tendency toward miniaturiz-
etables (green peppers) for the analysis of NMC pesti- ation, a new method for NMC determination in fruits
cides by HPLC with post-column derivatization and and vegetables has been proposed. This method elim-
Suorescence detection have been carried out. These inates the need for delivery pumps to introduce the
tests showed that the membranes were effective in hydrolysis/derivatization reagents, since they are al-
retaining sample interferences in both ofSine (with ready present in the mobile phase. For this purpose,
a 22 cm diameter activated carbon membrane) and a Kromasil-100 C18 column (which tolerates basic
online (extract injected directly in the HPLC system) conditions) and a borate buffer mobile phase were
methods. Solid-phase extraction with bonded-silica chosen. A simpliRed LC}Suorescence (FS) method for
adsorbents, including octyl (C8), ethyl (C2), octadecyl the determination of traces of carbamates in grains,
(C18) and cyclohexyl (CH), has also been successfully fruits and vegetables has been described. Here, the
developed as an alternative to LLP clean-up. Elution sample size and solvent volumes were considerably
patterns and recovery from various adsorbent mater- reduced, and a clean-up on aminopropyl-bonded Sep
ials were determined for 17 OPs, nine carbamates and Pak威 cartridge was performed in order to separate the
target compounds from the coextractives. Recovery
of the pesticides and limits of detection in different
food matrices varied from 60 to 103% and from 1 to
4 ppb. A complete analytical methodology for the
determination of some NMCs in vegetable, fruit and
feed crop samples has been developed in which the
pesticides are extracted, cleaned on an aminopropyl-
bonded SPE column, and then determined by LC-FS.
Depending on the detection technique, not only the
method selectivity, but also the method sensitivity,
can be changed considerably. For example, by using
LC with UV detection, the presence of coextractives
can severely limit the analyte determination, resulting
in poor method sensitivity when analysing real sam-
Figure 1 Liquid chromatography system with post-column re- ples compared with standard solutions. Despite this,
actions and fluorimetric detection for determination of N-methyl- and considering that the maximum residue levels for
carbamate pesticides. foodstuffs are higher than those for water samples,
Figure 2 Block diagram of a scheme for extract preparation for HPLC analysis N-methylcarbamate pesticides in fruit and vegetables.
If fluorimetric detection is performed, the LLP step can be omitted. Elution through an SPE column was carried out according to the
method described by Nunes GS, Ribeiro ML, Polese L and BarceloH D (1998) Comparison of different clean up procedures for the
determination of N-methylcarbamate insecticides in vegetable matrices by high-performance liquid chromatography with UV detection.
Journal of Chromatography 795: 43}51.
a simple and rapid method has been proposed to ture charcoal}Celite clean-up column have been car-
analyse eight carbamate insecticides and 10 of their ried out. Good recovery data obtained by spiking
main metabolites in apples, pears and lettuce. The pears, carrots and bananas at the 0.1}0.5 p.p.m. dem-
clean-up procedure is based on that of the De Kok onstrated the excellent performance of this method.
method, which uses solid-phase cartridges, and the Various types of detectors have been evaluated
amounts of solvents and sample are substantially re- for the analysis for the carbamate insecticides by
duced. It was found that matrix interferences could GC. Among these, the nitrogen-phosphorus detector
be minimized by diluting the Rnal extract, but the (NPD), the Same photometric detector (FPD) with
method sensitivity is also reduced. either a phosphorus or sulfur Rlter, the electron-cap-
ture detector (ECD) and the mass spectrometric de-
GC Methods
tector (MS), either in electron impact mode (EI) or in
The relationship between the mode of detection and positive chemical ionization (PCI) mode, have been
the required extraction procedure for this mode has tested. Sensitivity, linearity and selectivity of the dif-
resulted in the emergence of numerous sample prep- ferent detectors have been detailed and application to
aration procedures. Using GC}mass spectrometry real samples, including foodstuffs, has also been pre-
(MS) and LC-FS techniques, the extraction of 199 sented. Although the thermolability of the carba-
pesticides from fruits and vegetables with acetonit- mates is considered the major limitation as regards to
rile, and removal of the coextractives with a minia- the use of GC techniques, it has been shown that
thermal degradation of some carbamates and metab- employing LLP procedures using organic solvent of
olites does not occur under certain conditions, as was limited water capacity, to achieve the removal of the
demonstrated by the fragmentation patterns of the coextractives present in the sample extract and/or
studied compounds. GC-NPD methods for the solid-phase clean-up with silica or Florisil. Finally,
monitoring of residues of some carbamates in real analyte determination is performed by GC or HPLC
samples have been also described. A comparison has with selective detectors.
been made of GC and LC techniques for the analysis More than 300 pesticides and pesticide-related
of the most popular (in terms of amount produced compounds can be determined by the well-known
and applied) pesticide classes, such as carbamates, MRMs described in the ofRcial literature, such as the
phenylureas, triazines, phenoxy acid herbicides and AOAC method, the German MRM S19 (Deutsche
chlorinated phenols. Sample concentration and detec- Forschungsgemeinschaft, DFG), and the method
tion have been discussed in relation to their inSuence adopted by the National Food Administration of
on the performance of the particular separation tech- Sweden. They all have several disadvantages, such as
nique. GC methods, when applicable, have the ad- their inefRciency as screening methods, since they are
vantages of greater separation efRciency, higher speed too time-consuming and labour-intensive, the large
of analysis and the availability of a wide range of amount of solvents used, and, in addition, newly de-
highly sensitive detectors. On the other hand, LC is veloped groups of pesticides are becoming more polar
often the choice when polar, nonvolatile and/or ther- and/or thermodegradable, making difRcult their analy-
molabile compounds need to be analysed, as it is in sis by conventional chromatographic techniques.
the case of carbamates. During the past two decades, research has gone in
the direction of reduction of organic solvent toxicity,
Supercritical Fluid elimination of the partition step and elimination of
the column clean-up. A rapid and efRcient multi-
Chromatography/Extraction residue extraction procedure has been reported using
Supercritical Suid chromatography (SFC) is a tech- ethyl acetate and sodium sulfate, followed by gel
nique that in many ways is a hybrid of GC and HPLC. permeation chromatography (GPC) on an SX-3 col-
It is recognized as a valuable technique for the analy- umn. Its effectiveness to analyse OC and OP insecti-
sis of thermally labile compounds such as carba- cides was conRrmed. Several other methods based on it
mates. A few applications have been reported for SFC have arisen using speciRc detectors that have decreased
in the Reld of pesticide determination in food. The the number of interfering chromatographic peaks.
versatility in separation, the possibility of using dif-
LC-MS Techniques
ferent detectors (LC or GC detectors), and the pros-
pect of directly coupling with supercritical Suid ex- ConRrmation of the presence of carbamates in real
traction (SFE), make this analytical technique very samples has been performed by HPLC-MS with vari-
attractive. SFE using CO2 has been examined for ous interfaces, but in general most of these studies
separating carbamates from interfering coextractives have been performed with standard solutions.
prior to analysis either by LC-FS or by GC with ion Table 2 summarizes the development of LC-MS tech-
trap mass spectrometry (ITMS) detection. Pre-extrac- niques for the analysis of different pesticides, includ-
tion of ground meat with acetonitrile before SFE left ing carbamate insecticides. Up to the present, few
behind over 99% of the fat and Rbre. SFE of the food analysis applications have been reported. Atmo-
acetonitrile extracts with pelletized diatomaceous spheric pressure chemical ionization (APCI) tech-
earth further reduced the amount of coextractives niques for the mass spectral analysis of several NMCs
10-fold, removing the interferences that would have have been evaluated and the results compared with
appeared in the Suorescence mode. those obtained using ionspray (ISP) and thermospray
(TSP) interfaces. The results were also compared
Simpli\cation of Multiresidue with EI ionization and methane CI spectra obtained
with a particle beam (PB) interface. These methods
Methods were applied to the conRrmatory analysis of three
OfRcial laboratories that investigate and analyse pes- representative carbamate pesticides, spiked at the
ticide residues usually utilize the established multi- 0.1 ppm level in green peppers.
residue methods (MRM) of analysis. One of the most Multiresidue conRrmations of pesticides from dif-
commonly used MRM for pesticide analysis in fruit ferent classes using APCI techniques have also been
and vegetables samples is the AOAC (Association of performed. It was observed that the fragmentation of
OfRcial Analytical Chemists) method. It involves an the carbamates studied was highly voltage-depen-
aqueous acetone extraction and laborious clean-up dent. An increase in the potential of the sampling
Table 2 MS interfaces most commonly used for LC analysis of pesticides, including carbamate insecticides
Spectrometry interface Pesticides studied Sample type LOD (ng)
APCI, ISP, TSP and CI NMCs Aqueous solutions *

Green peppers
APCI (in positive- and negative- 17 pesticides in five chemical classes Ground water Full-scan
ion modes), TSP, and PB-MS (triazines, phenylureas, carbamates, mode: 0.8}10
organophosphorus and miscellaneous) SIM mode:
0.01}1
TSP Anilides, carbamates, N-heterocyclic, Aqueous solutions *
organophosphorus, and phenylureas)
TSP (with online and offline SPE Nitrogen- and phosphorus-containing pesticides Aqueous solutions SIM mode:
procedures) 0.04}0.6
TSP 3 N-methylcarbamates, 3 N-methylcarbamoyl- Aqueous solutions *
oximes, 2 substituted urea pesticides, and 1
ester of a substituted carbamic acid
TSP (with either filament- and 19 carbamates and 12 of their known degradation Aqueous solutions *
discharge-assisted ionization modes products
TSP 16 carbamates from four chemical subclasses Aqueous solutions *
(oxime-NMCs, aryl-NMCs, N-phenylcarbamates
and methyl esters of substituted carbamic acids)
FIA-PB-PCI 14 carbamates Aqueous solutions *
FIA-PB-EII-CI (with either negative- 33 carbamates and 14 of their degradation products Surface water *
and positive-ion modes)
APCI 11 carbamates Food samples 0.3}10
ES 20 carbamates Fruits and vegetables *
ES N-Heterocyclic compounds, phenylureas and Fruits and vegetables *
carbamates
APCI, atmospheric pressure chemical ionization; CI, chemical ionization; DCI, desorption chemical ionization; EII, electron impact
ionization; ES, electrospray interface; FIA-PB-PCI, flow injection analysis-particle beam-positive chemical ionization; ISP, pneumati-
cally assisted electrospray ionization; LOD, limit of detection; SIM, selected-ion monitoring; SPE, solid-phase extraction; TSP,
thermospray interface.
cone strongly affected the formation of diagnostic carbamate analysis have also been compared, and the
daughter ions. The dependence of the ion abundances so-called suppression effects on the ion formation
in the TSP mass spectra of several pesticides has been have been studied in coeluted compounds. Thermally
studied with regard to the vaporizer and the gas- labile carbamates gave unsatisfactory results with re-
phase temperatures and under collision-activated dis- gard to spectral compatibility between the interfaces.
sociation conditions. Fragmentation pathways under These differences are due to thermally assisted hy-
certain experimental conditions were investigated for drolysis reactions that occur in the various vaporizer
some of the carbamate pesticides. Both vaporizer and designs. Different approaches using desorption chem-
source jet temperatures were monitored. APCI, ESP, ical ionization (DCI) and Sow injection (FIA)}par-
fast-atom bombardment, Cf-252 plasma desorption ticle beam (PB)}ammonia PCI}MS were further
and collision-activated dissociation spectra were then evaluated. The mass spectra using FIA-PB-PCI-MS
performed for the pesticides to conRrm proposed exhibit higher relative abundances for fragment ions,
pathways and to gain additional information. and the ion intensities are strongly dependent on the
Through an interlaboratory study involving nine ion source pressure. Another study, employing EI
laboratories, it was concluded that, in TSP-LC-MS ionization and ammonia/methane with positive/nega-
systems, the thermospray tip temperature plays a ma- tive chemical ionization, has shown ammonia to be
jor role in adduct formation and ion fragmentation of the best reagent gas. Using ammonia gave less frag-
thermally labile carbamate pesticides. As a result, this mentation and better quantitative results than meth-
temperature needs to be carefully controlled. This ane for the analysis of 33 carbamates and 14 of their
effect has been conRrmed by investigating the inSu- degradation products.
ence of three LC eluent additives (ammonium acetate, LC-APCI-MS and LC-post-column Suorimetry
ammonium formate and nicotinic acid) and the va- for the determination of carbamates in foodstuffs
porizer temperature on ion formation. The perfor- were compared. It was observed that, despite its
mances of different thermospray interfaces (which great potential in detection and conRrmation, in gen-
exhibit wide differences in source geometry) used in eral LC-APCI-MS is less sensitive for quantitation
purposes, which agrees with previous reports. The slower oxidation to the sulfone. It is interesting to
feasibility of using reversed-phase LC-MS with an consider the rapid degradation of aldicarb after its
electrospray interface for measuring traces of NMC application, because in some cases the parent com-
insecticides in 10 different types of fruit and veg- pound is not present in the studied matrix. Since that
etables has been evaluated. Extraction with meth- the toxicity could be effectively more acute if the
anol, followed by clean-up on a Carbograph 1 extrac- metabolites were present in more elevated concentra-
tion cartridge, provided an average recovery higher tions, their conRrmation in real samples by LC-MS is
than 80% for all compounds. It was noticed that of crucial importance. Such compounds have been
changing from methanol to acetonitrile as the organic analysed by LC-APCI-MS using the selected ion
modiRer resulted in a signiRcant decrease in ion signal monitoring (SIM) mode based on Rve channels.
for the carbamates. The presence of coextractives Figure 4 shows a typical total ion chromatogram and
derived from the crop samples in the electrosprayed the Rve selected ions used to identify the compounds
solution did not interfere signiRcantly with the in orange extracts. Here, the fragmentation was as-
ionization process of the compounds studied. The sisted by the addition of ammonium formate in the
photochemical behaviour of pesticides in a photolysis mobile phase (acetonitrile/water) and ion adducts did
reactor coupled online with an LC-ESP-MS system not appear in the mass spectra, resulting in a sensitive
has been investigated this system has been used suc- method free from interferences.
cessfully to trace conRrmatory analysis of carbamates
Thin Layer Chromatography Methods
in foodstuffs.
Recently an analytical method for conRrmation of Thin-layer chromatography (TLC) is used for the
aldicarb and its metabolites in fruit and vegetable qualitative and quantitative analysis of a wide variety
samples has been optimized. Aldicarb is one of the of compounds, including pesticides. Among the car-
NMCs most commonly used for insect control in bamate insecticides analysed by TLC can be found
tropical countries in different agricultural Relds. Fig- those most commonly used in crop protection, such
ure 3 illustrates schematically the metabolic pathway as carbaryl, carbofuran, methomyl, bendiocarb,
of aldicarb under environmental conditions, and propoxur, aldicarb and carbendazim. Since the
shows the formation of the mass fragments used to 1980s, only a few papers concerning carbamate
identify the parent and metabolic compounds. Initial analysis by TLC have been published. Thin-layer
metabolic attack is rapid and complete oxidative con- chromatographic procedures have been developed
version to aldicarb sulfoxide is followed by a much for the separation of several carbamates, such as
Figure 3 Metabolic pathway of aldicarb to its degradation products under environmental conditions, and mass fragments used for the
LC-MS confirmation of the compounds.
and analysis time while still maintaining reliability

and sensitivity. Moreover, IA can be used as a screen-
ing method in order to detect food contamination.
The relatively short analysis time allows for the
screening of a large number of samples which is
a major advantage.
IA has been shown to have potential as a screening
method for the detection of pesticides in food sam-
ples. A rapid bioluminescence method was used for
screening several OP and NMC insecticides in pro-
cessed baby food. Among the 155 samples tested,
there were 23 suspected positives (14.2%) that were
further analysed by HPLC; this resulted in conRrma-
tion of the presence of carbaryl in 18 of the samples.
The utility and applicability of an analytical
method depends in great part on the absence of
matrix interferences. In this regard, ELISA is not
Figure 4 Mass spectra for (1) aldicarb sulfoxide (2), aldicarb
different from other detection techniques, and the
sulfone and (3) aldicarb present in a spiked orange extract. Peak sample preparation prior to analysis is still a critical
confirmation by SIM mode of five selected channels on the total point for pesticide residue determination by IA
ion chromatogram (TIC). Final extract 20-fold concentrated (see methods. A competitive ELISA has been developed
sample preparation in Figure 2). Chromatrographic separation of for quantiRcation of methyl 2-benzimidazolecarba-
the compounds was carried out using a C18 RP-LC column and
a water/acetonitrile mixture, both containing 0.1% ammonium
mate in fruit juices. Matrix effects are minimized by
formate as mobile phase. diluting the samples before IA. Rapid methods based
on water or acetonitrile extraction have been evalu-
carbaryl, bendiocarb, carbofuran, 2-isopropyl- ated for screening carbofuran and aldicarb sulfone in
phenyl-N-methylcarbamate (MIPC) and 2-sec-butyl- mean and liver using commercial ELISA kits. The
N-methylcarbamate (BPMC). Suitable schemes using Rnal extracts are diluted in order to eliminate the
plates coated with silica gel containing 1% zinc acet- effect of the solvent, which is more pronounced than
ate as support, and benzene/ethyl acetate (50:10) as the natural compound effects in some cases. Attempts
solvent have been developed for carbamate analysis have been made to minimize the matrix and organic
in various matrices, such as aqueous biological Suids solvent effects in ELISA for carbaryl by diluting the
and food samples, mainly fruit juices. Residues of crop extracts in the assay buffer, but it was observed
carbofuran and its two metabolites have been extrac- that the effect of the solvent residue in the diluted
ted with HCl, partitioned into CH2Cl2, chromato- extracts on the assay was still higher than the matrix
graphed on silica gel and detected with KOH-p- effect. The direct application of IA to vegetable/fruit
nitrobenzenediazonium Suoroborate. Sequential juices diluted in the assay buffer is possible, without
TLC has been used for the detection and determina- large losses in the method sensitivity. As an example,
tion of carbaryl in water samples. More recently, the Figure 5 shows some calibration curves for carbaryl
chromatographic behaviour of carbamate pesticides in assay buffer and in lemon juices diluted in assay
and related compounds has been examined on thin buffer. In this case the absence of matrix effect, after
layers of alumina, barium, sulfate, calcium carbon- sample dilution and pH adjustment, indicates good
ate, calcium phosphate, calcium sulfate, cellulose and IA performance. Unfortunately, in most cases the
silica Gel G. recoveries of IA-based methods have proved to be
lower than those of methods that do not employ any
Immunoassay (IA) Methods for prior sample treatment.
Different experimental approaches to ELISA quan-
Carbamate Determination tiRcation of NMC insecticides in fruit juices, without
In the last few years, the number of enzyme-linked any sample pre-treatment other than dilution, have
immunosorbent assay (ELISA) methods for the deter- also been developed. For carbaryl and carbofuran
mination of pesticides has increased, but there is still residue determination in fruit juices, high afRnity
a lack of IA methods for pesticide residue determina- monoclonal antibodies (MABs) have been used. Off-
tion in food and crop samples. IA techniques can line SFE and ELISA have been used for the determina-
provide complementary and/or alternative ap- tion of nine pesticides, including four carbamates, in
proaches in reducing the use of expensive equipment foodstuffs consisting of baby food and Food and
a recent development. In the analysis of pesti-

cides in food matrices, the use of IC is still more
limited. IC is based on the highly selective interaction
of antigens with their antibodies, which are immobi-
lized on a support material called an immunosorbent.
The production of antibodies against pesticides is
based on the conjugation of hapten to a large im-
munogenic carrier molecule (typically bovine serum
albumin or keyhole limpet haemocyanin), and
subsequently the complex is injected into a suitable
vertebrate (rabbit, mouse, rat, sheep). Since anti-
body}antigen interactions occur over short distances,
steric effects are involved in the coupling reaction.
These steric effects are what make antibody}antigen
interactions so selective, and only the antigen that
produced the immune response, or very closely re-
lated molecules, will be able to blind to the antibody.
Thus, theoretically, when the sample is run through
the immunosorbent the analytes are selectively re-
tained and subsequently eluted free of the coex-
tratives (Figure 6). Once the analytes have been sep-
arated from interferences, they can be determined by
conventional chromatographic techniques. The use of
Figure 5 Matrix effect of lemon juice on the enzyme-linked
immunosorbent assay (ELISA) for carbaryl analysis. Calibration IC as a separation tool before pesticide analysis is
curves were constructed with PBST buffer and PBST-diluted thus extremely attractive.
lemon juices. (A) Natural pH of the diluted samples; (B) pH Among the carbamate pesticides, only carbofuran
adjusted with an alkaline solution. PBST, (Reproduced with per- has been separated from natural components of crop
mission from Nunes GS (1999) Analysis of N-methylcarbamate
samples by IC and a highly sensitive online IC with
insecticides by chromatographic techniques, immunoassay
(ELISA) and amperometric biosensors. Institute of Chemistry coupled-column LC-MS was used to analyse some
IUNESP, Anaraquara, Sa o Paulo, 230 pp [doctorate thesis]. fruit and vegetable samples.
Drug Administration (FDA) Total Diet Study (TDS)

samples. The beneRts of SFE-ELISA included replace-
Current Trends and Conclusions
ment of harmful organic extraction solvents, rapid A tremendous amount of work has been done and
extractions with a relatively inexpensive extractant, much more is under way in the Reld of pesticide
and a reduced number of steps in the determination of analysis is food samples. In the chromatographic Reld
the target compounds. several new techniques for pesticide analysis, such
In contrast to conventional chromatographic tech- as immunoafRnity chromatography and LC-MS with
niques for pesticide analysis in foods, IA methods various interfaces, have appeared. These have un-
have not yet been extensively characterized. The vali- doubtedly contributed to increased separation efR-
dation of the proposed ELISA against another ciency and improved sensitivity. It is expected that
validated method has been the primary objective in the bioanalytical techniques for pesticide analysis
only a few published papers. As for any analytical will become a common analytical alternative because
method, quality control and assessment of material of their demonstrated advantages, especially in the
and equipment stability are required. In addition, IA analysis of carbamates in food samples. The capacity
evaluation involves deRning working ranges, sensitiv- that these compounds show in inhibiting a certain
ity, precision, accuracy, linearity, speciRcity and class of enzymes (the cholinesterases) must be further
matrix effects. explored. Development of approaches based on the
coupling of the chromatographic separation with bi-
Immunoaf\nity Chromatography
odetection systems is a promising alternative.
ImmunoafRnity chromatography (IC) has been wide- Sample handling is the bottleneck in the analysis
ly used for the determination of various analytes in of carbamate insecticides in foodstuffs, since in many
the medical Reld; however, the use of antibodies im- cases a complex clean-up step is needed before
mobilized on an appropriate support to pre-concen- chromatographic separation. The use of biological
trate pesticides from environmental samples is only techniques, such as immunoassays and biosensors,
III / CARBOHYDRATES / Electrophoresis 2201
Figure 6 Schematic separation of analytes by immunoaffinity chromatrography for pesticide determination in food matrices.
can overcome some of these limitations. In the LC- FAO (1993) Agriculture Towards 2010. C 93/24, Docu-
MS techniques, the use of an atmospheric pressure ment of the 27th Session of FAO Conference, Rome.
chemical ionization interface is at present the Harlow ELD Antibodies: A Laboratory Manual, ch. 5,
best alternative since it offers high selectivity and Cold Spring Harbor, NY: Cold Spring Harbor Labo-
sensitivity for the trace determination of carbamates. ratory.
Hassal KA (1983) The Chemistry of Pesticides: Their Meta-
See also: II/Affinity Separation: Immunoaffinity bolism, Mode of Action and Uses in Crop Protection.
Chromatography. Chromatography: Gas: Detectors: New York: Macmillan.
Mass Spectrometry. Chromatography: Liquid: De- Lopez-Avila V, Charan C and Van Emon J (1995) Im-
tectors: Mass Spectrometry. Extraction: Supercritical munoassays for residue analysis. In: Beier RC and
Fluid Extraction. III/Immunoaffinity Extraction. Multi- Stanker LH (eds) Food Safety, ACS Symposium Series
residue Methods: Extraction. Pesticides: Extraction 621, p. 438. American Chemical Society.
from Water; Gas Chromatography; Supercritical Fluid National Library of Medicine (1992) Carbaryl. Hazardous
Chromatography; Thin-Layer (Planar) Chromatography. Substances Databank 4: 293}297.
Sawer LD, McMahon BM, Newsome WH and Parker GA
Further Reading (1990) In: Helrich K (ed.) OfTcial Methods of Analysis,
Association of OfTcial Analytical Chemists, Agricultural
BarceloH D and Hennion M-C (eds) (1997) Trace determina- Chemicals, Contaminants, Drugs, vol. 1, p. 274. Arlin-
tion of pesticides and their degradation products in gton, VA: Association of OfRcial Analytical Chemists.
water. In: Techniques and Instrumentation in Analytical Sherma J (1989) Analytical Methods for Pesticides and Plant
Chemistry, vol. 19. Amsterdam: Elsevier Science. Regulations, vol. 17, San Diego, CA: Academic Press.
CARBOHYDRATES
Introduction
Electrophoresis Electrophoresis has been an important tool for carbo-
hydrate analysis since its early stages of development.
O. Grosche, Universita( t des Saarlandes, Moving with time from paper electrophoresis to
Saarbru( cken, Germany polyacrylamide slab gel electrophoresis and then
Copyright ^ 2000 Academic Press to the sophisticated high performance capillary
2202 III / CARBOHYDRATES / Electrophoresis
Figure 1 Boric acid as a Lewis acid forms borate in water. Borate reacts with polyol compounds to mono- or bisdiol complexes.
electrophoresis (HPCE), its inSuence in many chem- media acts as a Lewis acid to form the tetrahedral
ical, biochemical and biomedical Relds has grown anion B(OH)\ 4 . At pH 8}10, the formation of mono-
constantly. New detection techniques and buffer sys- and diesters with the carbohydrates is most effective,
tems have been developed to make use of the advant- but depends largely on the structure of that particular
ages offered by HPCE, such as low sample and buffer sugar (Figure 1).
volumes, fast separations and high efRciencies. To- The complexation of metal cations also depends
day, sample volumes in the range of single cell liquids on the structure of the carbohydrate. In addition,
can be analysed. The development of highly sensitive the ionic radius is crucial for the stability of the
detectors and new derivatization procedures has con- complex. Complexation is possible with alkaline
stantly lowered the detection limit. In the light of metals, Cu2#, Pb2#, Zn2#, Ni2# and Cd2#. For alka-
these advantages, the following article will focus on line earth metal buffers, the separation efRciencies
capillary electrophoresis. decrease in the following order: Ba2#'Sr2#'
Ca2#Mg2#. Another way to introduce an ionic
charge is by the deprotonation of the hydroxyl func-
The Electrophoretic System tions of the sugars. For this purpose, high pH
values (sodium hydroxide solutions, pH'12) are
Electrolyte Systems
necessary.
The presence of an ionic charge is the usual prerequi- Carbohydrates can also be separated by micellar
site for migration in an electrophoresis system. Only electrokinetic chromatography (MEKC). In this case,
some saccharides, such as aldonic acids, uronic acids, the separation is caused by the different distribution
sialic acids, aminosugars or sulfated sugars of chon- coefRcients of the sugars between the free solution
droitin, dermatan, keratan and heparin, possess and the interior of the migrating micelles of a charged
charged functional groups in their structures. The detergent (sodium dodecyl sulfate).
separation of uncharged molecules requires the con-
Detection Systems
version into charged species by complex formation or
derivatization. Precolumn derivatization Most carbohydrate spe-
The most commonly used systems in planar elec- cies do not posses strong chromophores in their
trophoresis are those with borate buffers, but other structures. This condition does not allow their direct
inorganic acids can form complexes with uncharged sensitive detection by UV absorption. To overcome
saccharides. In HPCE, systems are based mainly on this difRculty, carbohydrates can be tagged with suit-
sugar}borate complexes and to a lesser extent on able Suorophores or chromophores. If the tag also
sugar}metal cation complexes. Boric acid in aqueous provides the charge necessary for electrophoresis,
uncharged carbohydrates can also be separated with- tions introduce several negative charges to the mol-
out the use of complexing buffer systems as well. In ecules, which results in high mobility and low adsorp-
principle, derivatization of the hydroxyl groups tion of the derivatives to the capillary surface. All that
should be a good approach for the tagging of carbo- these compounds have in common, is that they con-
hydrates, but the derivatization of sugars with differ- tain a single amino function, which allows the attach-
ent reactivities would lead to multiple tagging of the ment to the carbonyl function of the reducing sugar
molecule. This would consequently cause a distribu- by reductive amination. In this one-pot reaction, the
tion of different products rather than a single deriva- open chain form of the sugar reacts with the amine to
tive. Therefore, other functional groups of the sugar a Schiff’s base. To accelerate the reaction and to shift
molecule must be considered. The carbonyl group in the equilibrium to the product side, the imino func-
the open chain form of reducing sugars is the most tion is reduced by sodium cyanoborohydride to the
widely used functional group for the attachment of respective secondary amine (Figure 2). An exception
the label. In amino sugars or acidic compounds, the to this is the reaction of reducing carbohydrates with
amino group or the carboxyl group provides potential 1-phenyl-3-methyl-pyrazolone (PMP). In this case,
for derivatization. the acidic hydrogens of PMP and the aldehyde func-
Since it is required to produce only one species of tionality condense under slightly basic conditions.
derivative in precolumn derivatization, the label must A disadvantage of precolumn derivatization is the
not possess more than one reactive function for higher structural similarity of the compounds and the
the atttachment to the carbohydrate. Another re- higher complexity of sample preparation. Another
quirement is the high molecular absorptivity (UV problem is the varying reactivity of carbohydrates,
detection) or photoluminescence efRciency (Suores- especially of monosaccharides, with the derivatizing
cence detection) of the tag. Although laser-induced agent. Optimized reaction conditions have to be de-
Suorescence detection (LIF) exhibits the lowest detec- termined for every sample mixture. The reproducibil-
tion limits, it is difRcult to Rnd a suitable reagent for ity of the sample preparation is crucial for the
the respective laser system. The label has to be match- quantiRcation of the compounds.
ed to the laser system used with respect to excitation
wavelength and emission intensity. Table 1 compares Indirect detection Indirect detection modes have
the detection limits of the different systems after been applied for the analysis of compounds whose
derivatization. structures lack the necessary physical properties for
The different labelling agents can be divided into direct detection. The key element is the use of a back-
three groups according to their set charge in the ground electrolyte, which provides a high, continuous
separation system: positive, negative or uncharged. signal. The analyte ion displaces the background elec-
Table 2 gives examples of some commercially avail- trolyte ion and leads to a change in the UV absorption
able derivatization agents and the wavelengths used signal. Therefore, the background electrolyte needs to
in different detection systems. Recently, they have have the same type of ionic charge as the analyte. The
found their way into standard chemical catalogues attainable detection limit is given by
(Fluka, Sigma, Aldrich, ICN). The positively charged
compounds usually contain heterocyclic nitrogen CM
Clim"
functions, e.g. 2-aminopyridine or 6-aminoquinoline. DR ) TR
They can be protonated at lower pH values and are
commonly used for separation in phosphate buffer CM represents the concentration of the background
solutions (pH 2.5}4). Depending on their pK values, electrolyte, DR is the dynamic reverse and TR is the
these amino compounds can also be used as un- transfer ratio. DR is deRned as the ability to measure
charged labels in high pH buffers. Chromophores a small change on top of a large signal and is equal to
with strongly acidic groups, like aminonaphthalene the signal/noise ratio of the background signal. TR is
trisulfonic acid (ANTS), remain negatively charged deRned as the number of molecules of the background
over a wide pH range. Multiple sulfonic acid func- ions displaced by the charged analyte. It can be seen,
Table 1 Detection limits of monosaccharides after derivatization
Detection Absolute amount (mol) Concentration (mol L\1) Weight concentration
UV derivatization 10\13}10\11 10\6}10\4 ca 10 ppm

Fluorescence derivatization 10\16}10\11 10\9}10\4 ca 10 ppb
Derivatization for LIF 10\21}10\17 10\13}10\9 ca 1 ppt
Table 2 Examples for derivatization agents and used detection wavelengths
Structure Name ex em
2-Aminopyridine (2-AP) 240, 320 400
6-Aminoquinoline (6-AQ) 355 550
1-Phenyl-3-methyl-5-pyrazolone (PMP) 245 }
5-Aminofluoresceine 257 471
1-Aminonaphthalene-7-sulfonic acid 229 482
7-Aminonaphthalene-1,3-disulfonic acid (ANDS) 325 520
8-Aminonaphthalene-1,3,6-trisulfonic acid (ANTS) 223, 325 520
8-Aminopyrene-1,3,6-trisulfonic acid (APTS) 325 520
2-Aminoacridone (AMAC) 488 520

Figure 2 The open chain of the reducing carbohydrate reacts with the amino compound to a Schiff base, which is reduced to
a secondary amine by sodium cyanoborohydride.
that CM should be as low as possible while still Amperometric detection Another method is pulsed
generating a high DR. The TR should be close to unity. amperometric detection (PAD) with gold or platinum
Either indirect laser-induced Suorescence detection electrodes, using a strongly alkaline buffer, which has
or UV detection can be applied to the separation of been adapted from the HPLC-PAD systems. These
underivatized carbohydrates. Coumarin (LIF) and systems require a specialized pulse sequence and
sorbic acid (UV), for example, have been used as therefore expensive instrumentation. Other systems
background electrolytes with deprotonated carbohy- have used ultramicroelectrodes, equipped with a 25-
drates at high pH values. In indirect detection, the m copper wire at a constant potential. These systems
concentration of the background electrolyte should permit the realization of a linear range over three
be low. Thus, the use of a borate buffer (required magnitudes and a limit of detection for mono- and
concentrations between 0.1 and 0.25 mol L\1) is ex- disaccharides in the femtomole range. However, us-
cluded. Furthermore, the deprotonation of the carbo- ing PAD systems, the running buffer must not contain
hydrates at high pH values is limited by the increasing any electroactive species that might oxidize on the
concentration of hydroxide anions, which compete working electrode and cause a strong background
with the background electrolyte ions in the displace- signal or poison the electrode surface. Thus, am-
ment mechanism. perometric detection excludes the use of borate buf-
fers, which are essential for the separation of many
Direct detection Different modes of direct detection neutral and native saccharides.
have been applied to HPCE separations of un-
derivatized carbohydrates. First, low wavelength UV Refractive index detection The determination of the
(below 200 nm) can be used to detect carbohydrates refractive index (RI) has shown its potential in HPLC,
which have a sufRcient molar absorption coefRcient, although it is not very sensitive and its detection limit
especially those compounds having carboxyl or other falls below that of low wavelength UV. To overcome
UV-absorbing groups. Mixed oligosaccharides of the problems with the low volume Sow in capillary
heparin and heparan sulfate, chondroitin sulfate and electrophoresis (CE), a sub-nanolitre laser-based re-
dermatan sulfate can be detected at 232 nm. The poor fractive index detector has been developed. The de-
absorption coefRcients of many carbohydrates in the tection is based on the change of interferences caused
range 190 to 280 nm, limits this detection mode. by the change of the refraction index. However, the
adaptation of RI detection of HPCE gives some prob- Several approaches have been made to inSuence
lems, caused by the joule heating, which introduces the resolution of the HPCE separation of mono-
changes in the temperature of the liquid and, with saccharides.
this, changes in the refractive index. The most common buffer system is based on the
complexation of the carbohydrates with an alkaline
Capillaries with partially increased inner diameter borate buffer. Figure 3 demonstrates the separation
Theoretically, the limit of detection is reduced with of a mixture of several aldopentoses and aldohexoses.
the increasing inner diameter of the capillary, due to The sugars were tagged by reductive amination with
the increased lightpath through the solution. This is 2-aminoanthracene. A counter-electroosmotic separ-
limited by the increasing current with constant Reld ation with normal polarity and Suorescence detection
strength, on moving to greater diameters. High cur- was performed. In this case, the intrinsic mobilities of
rents lead to joule heating effects and contribute to the analytes are lower than the mobility of the elec-
additional band broadening. Capillaries, which are troosmotic Sow (EOF). The analyte molecules are
widened locally at the detection window, can help to transported against their migration direction to the
decrease the detection limit. In this case, the capillary cathodic end of the capillary. The fastest compound is
is partially etched with hydroSuoric acid at the detec- detected last. Five monosaccharides were baseline
tion window. The reaction is controlled by temper- separated, while one pair (arabinose/mannose) co-
ature. The diameter widens drastically at the ‘hot migrate. The stability of the carbohydrate complex
spot’ only. The reaction rate with the capillary wall at with the borate ion, which depends on the structure
other locations is relatively low. Using this procedure, of the monosaccharide, has the dominating inSuence
25-m capillaries can be widened up to 75 or 150 m, on the separation. Due to these differences, fucose
easily. Recently, detection cells with elongated light- (desoxyhexose) is separated from the unresolved pair
path (high sensitivity cells, z cells) have become com- mannose (hexose) and arabinose.
mercially available. To underline the differences in the migration mech-
anisms, Figure 4 shows a co-electroosmotic separ-
ation of the same mixture in an acidic medium. At
Applications and Separation Methods a pH of 2.0, the amino function of the analyte is
protonated and leads to a co-electroosmotic migra-
Monosaccharides
tion, i.e. in the same direction as the electroosmotic
The analysis of monosaccharides, the basic units of Sow. The fastest compound is detected the Rrst. Due
carbohydrates, is crucial for many areas of biochem- to the low dissociation of the silanol groups at the
istry, pharmacology, biotechnology and food science. capillary surface, the EOF is very low at this pH.
Figure 3 Counter-electroosmotic separation of 2-AA-labelled monosaccharides rhamnose (Rha), xylose (Xyl), glucose (Glu),
arabinose (Ara), mannose (Man), fucose (Fuc), galactose (Gal) and 2-aminoanthracene (2-AA). Buffer: borate 250 mM, pH 10.5;
U"30 kV; capillary-fused silica, i.d."50 m, l"88}100 cm; inj. 6s 100 mbar.
Figure 4 Co-electroosmotic separation of 2-AA-labelled monosaccharides rhamnose (Rha), xylose (Xyl), glucose (Glu), arabinose
(Ara), mannose (Man), fucose (Fuc), galactose (Gal), maltotriose as standard (M3ose) and 2-aminoanthracene (2-AA). Buffer:
phosphate 100 mM, 20% (v/v) iPropOH, pH 2.0; capillary fused silica, l"49}60 cm, i.d."50 m; U"ramp 2.5 kV min\1 from 5 to
30 kV; inj. 6s 100 mbar.
To enhance resolution, an organic modiRer (iso- they consist of one type or more than one type of
propanol) was added. The co-electroosmotic separ- monosaccharide unit that alternate in the repetitive
ation leads to a different electropherogram. Here, sequence. High molecular weight polysaccharides are
other parameters, such as the hydrodynamic radius usually analysed through their degradation products
of the analytes, have the dominating inSuence (chemical or enzymatic cleavage). The systems de-
on the migration mechanism. The differences in mo- scribed for the separation of monosaccharides can be
bility of mannose, fucose and arabinose are too small easily applied to the separation of oligosaccharides.
to lead to a separation, while the resolution of the With few exceptions, the mobilities decrease with
epimers glucose and mannose is increased compared increasing molecular weight of the analytes. The use
to the borate buffer system. of divalent metal ions for complexation is not favour-
It is also possible to separate a monosaccharide able for analysing carbohydrates, as it leads to insufR-
mixture by micellar electrokinetic chromatography cient separation efRciency.
(MEKC). In this system, the dominating mechanism The composition of oligosaccharide mixtures
is the dynamic distribution of the uncharged analytes should be considered when selecting the separation
between free solution and the migrating micelles of system. The use of a borate buffer focuses on the
the detergent (often SDS). Here, other properties of structural differences of the oligosaccharides and
the analytes, such as differences in their hydrophobic- should be used for separation of hetergeneous
ity, effect the separation. oligosaccharides with the same degree of polymeri-
Figure 5 shows the separation of a monosaccharide zation. For homologous oligosaccharides, the use of
mixture in a trisborate/SDS buffer. Unlike the results permanently charged labels in a co-electroosmotic
of the previous separation systems, the peaks of man- system is favourable. Here, the charge-to-mass ratio
nose and arabinose show a sufRcient resolution. is the dominant parameter responsible for the migra-
Table 3 gives some examples of the buffer systems tion differences. The inSuence of the electroosmotic
and tags used so far to analyse monosaccharide mix- Sow should be reduced to a minimum to ensure good
tures. The use of borate buffers dominates. Other resolution of the compounds with higher molecular
systems have also been used, but have not yet found mass.
their way into standard applications. Figure 6 shows the counter-electroosmotic separ-
ation of 2-AA labelled maltooligosaccharides in the
Linear Oligo- and Polysaccharides
borate buffer system. As the molecular weight in-
Oligo- and polysaccharides can be classiRed either as creases, the differences in migration time decrease
homo- or heteropolymers, depending on whether very fast. This can be explained by a lower degree of
Figure 5 MEKC separation of 2-AA-labelled monosaccharides. 1, rhamnose; 2, cellobiose; 3, xylose; 4, ribose; 5, glucose; 6,
mannose; 7, arabinose; 8, galactose; 2-AA, 2-aminoanthracene. Buffer: trisborate 100 mM/urea 4 M/SDS 100 mM, pH 8.3; capillary-
fused silica, l"48}60 cm, i.d."25 m; U"30 kV.
structural change, going step by step to higher degrees differences decrease drastically and lead to insufR-
of polymerization. Also, the ratio of analyte/EOF cient resolution.
decreases with increasing degree of polymerization. For higher degrees of polymerization ('7), the
With a higher degree of polymerization, the mobility use of charged labels at low EOF is to be preferred.
Table 3 Examples of the separation mode, buffer systems and detection methods for different carbohydrates
Analytes Detection Separation mode Derivatization Buffers
Monosaccharides Direct UV CZE 2-AP 200 mM borate, pH 10.5

Monosaccharides Direct Fluorescence CZE 2-AA 200 mM borate, pH 10.5
Monosaccharides Indirect UV CZE } 6 mM sorbate, pH 12.1
Monosaccharides Indirect LIF CZE } 50 M fluoresceine, pH 12.2
Monosaccharides Direct LIF CZE APTS 25 mM borate, pH 10.0
Monosaccharides Amperometric CZE } 100 mM NaOH, pH 13
detection
Monosaccharides RI detection CZE 100 mM borate, pH 9.0
Maltooligosaccharides Direct UV CZE 2-AP 100 mM phosphate, pH 2.5
DPmax(25
Maltooligosaccharides Direct LIF CZE ANTS 50 mM phosphate, pH 2.5
Branched xyloglucans Direct UV CZE 2-AP 100 mM phosphate, 50 mM
tetrabutylammonium bromide,
pH 4.75
Dextrans DPmax(25 Direct fluorescence CZE 2-AA 100 mM phosphate, pH 2.0
Low molecular weight heparin Direct UV CZE } 10 mM borate, 50 mM sodium dodecyl
sulfate
Polygalacturonic acid Direct LIF CGE ANTS 100 mM Tris, 25 mM borate, pH 8.5
DPmax(70 Polyacrylamide gel, 18% T, 3% C
Hyaluronic acid DPmax(400 Direct LIF CGE APTS 25 mM citric acid, 12.5 mM tris, 0.03%
aminodextran (mol. wt 10 000)
Glycoforms of ribonuclease A Direct UV CZE 20 mM phosphate, 50 mM sodium
dodecyl sulfate, 5 mM borate, pH 7.2
DPmax, maximum degree of polymerization (baseline separation); CZE, capillary zone electrophoresis; CGE, capillary gel electro-
phoresis.
Figure 6 Counter-electroosmotic separation of the starch hydrolysate Dextrin 10 (numbers indicate the degree of polymerization).
Buffer: borate 300 mM, pH 10.5; capillary fused silica, l"108}120 cm, i.d."50 m; field"30 kV; inj. 6s 100 mbar.
It can be seen in Figure 7 that protonation of the To improve the resolution of homologue oligosac-
amino function of the 2-aminoanthracene-labelled charides with a higher degree of polymerization, two
aminoglycans leads to a higher resolution of the different approaches have been attempted. First, the
oligosaccharide peaks. With a higher number of use of coated capillaries (polyether, polyvinyl alco-
monosaccharide units, the mobility differences de- hol) showing very low or virtually no electroosmotic
crease less than those in the borate system. Increasing Sow can lead to an increase in resolution. The disad-
charge of the label can increase the mobility of the vantages of these systems are the limited pH range in
molecules. This has proved to be the case with naph- which they can be used and the high cost of the
thalene di- and trisulfonic acid derivatives especially. capillary material.
By derivatizing with aminonaphthalene trisulfonic A similar approach to that for the separation of
acid, it is possible to separate more than 30 malto- oligonucleotides involves using gel-Rlled capillaries.
oligomers within 10 min. Gel-Rlled capillaries can have a sieving effect on the
Figure 7 Co-electroosmotic separation of the starch hydrolysate Dextrin 10 (numbers indicate the degree of polymerization). Buffer:
phosphate 100 mM, pH 2.0; capillary-fused silica, l"49}60 cm, i.d."50 m; U"25 kV; inj. 5s 100 mbar.
analytes. This size discrimination can be used Glycoconjugates

to increase the differences in mobility. Unfortu- Glycoproteins Glycoproteins function as enzymes,
nately, the use of gel-Rlled capillaries is accompanied transport proteins, receptors, hormones and struc-
by many difRculties such as poor gel-to-gel reproduci- tural proteins. The carbohydrate content of glycopro-
bility, bubble formation, quenching effects by the teins can vary from less than 1% to more than 60%
matrix, and the collapse of the gel matrix buffer. by weight. Protein glycosylation can occur at two or
Another problem is the relatively low mobility of more positions in the amino acid sequence. The
the formerly uncharged molecules, which are tagged glycans at a single position may be heterogeneous or
by a charged label. The use of gel-Rlled matrices can may be missing from some molecules. This leads to
increase the migration time to an unacceptable populations of glycosylated species of a single
extent. protein (glycoforms). The relative proportions of
For structures consisting of charged monosacchar- glycoforms are found to be reproducible, depend-
ide species, such as polyuronic acids, the use of a mixing on the glycosylation conditions (environment
ture of aminodextran and linear polyacrylamide of the reaction, physical state and type of organism),
(LPAA) as buffer additive is successful. In this case, the manufacturing process and the isolation
the retention mechanism cannot be due to the sieving procedures.
effect of the matrix only. The system is, rather, based On the one hand, glycoform separation and map-
on size-dependent electrostatic interactions between ping has to cope with the problems introduced by the
the protonated amino functions of the aminodextran protein moiety, such as interactions of the protein
and the dissociated carboxylic groups of the uronic with the capillary wall in low pH electrolyte systems.
acids. In this case, a hydrophilic coating of the capillary
Table 3 gives some examples of the systems used surface can aviod adsorption phenomena. On the
so far. other hand, the UV absorption coefRcients at 200 nm
Branched Oligosaccharides are sufRcient for the detection of the compounds
without precolumn derivatization. Common buffer
Branched oligosaccharides can be produced by enzy- systems are phosphate, TRIS/boric acid or borate
matic digestion of branched heteropolysaccharides, buffers, depending on the separation mode and the
for example xyloglucan polysaccharides. In combina- pH range.
tion with a charged label, some information about the
branching of the compounds can be achieved. If only
Glucosaminoglycans Glucosaminoglycans (muco-
the label is charged, the compounds migrate in order
polysaccharides) are unbranched polysaccharides of
of increasing size, as shown before. With similar
alternating uronic acid and hexosamine residues, for
molecular weight, the less branched oligosaccharides
example, heparin, chondroitin sulfate, dermatan
migrate faster than the more branched ones. For
sulfate or hyaluronic acid. After exhaustive treatment
branched xyloglucan oligosaccharides, a mobility
with polysaccharide lyases, disaccharides can be ob-
index system has been introduced to describe the
tained bearing unsaturated uronic acids. These
migration behaviour of the branched compounds
compounds can be detected directly at 232 nm. To en-
with respect to their linear homologues. The equation
hance sensitivity, they can also be labelled. Since the
below shows how the mobility index (MI) can be
mucopolysaccharides are already charged over a wide
calculated:
pH range, it is not necessary to introduce a charge by
a tag or complexation. The use of SDS micelles can

log s!log n#1
MI"100n#100 increase the resolution.
log n!log n#1
s is the electrophoretic mobility of the branched Glycolipids Some lipids contain oligosaccharide
analyte, and n and n#1 are the electrophoretic mo- moieties as an integral part of their structure. Since
bilities of the two homologues with n and n#1 these contain a hydrophilic head and a hydrophobic
repetitive units, which migrated before and after tail, they form micellar systems. Temperature,
the branched fragment. This index system can only be concentration of the analyte and ionic strength of the
applied if the following requirements are fulRlled. electrolyte system are crucial for the size and mobility
First, the system must consist of homopolymers or of the micelles. Thus, the separation of some
heteropolymers with a strictly repeating sequence. glycolipids as monomeric species is impeded. The
Second, the logarithmic mobilities of the homologues investigations made so far have been based on
must show a logarithmic dependence on the degree of phosphate buffer systems and low wavelength UV
polymerization, which is not always the case. detection.
III / CARBOHYDRATES / Gas Chromatography and Gas Chromatography^Mass Spectrometry 2211
Further Directions conjugates. Journal of Chromatography Library 58:

267}360.
Capillary electrophoresis has proven potential for Hase S (1996) Precolum derivatization for chromato-
carbohydrate analysis. High efRciencies and short graphic and electrophoretic analyses of carbohy-
analysis times are its advantages compared with other drates A. Journal of Chromatography A 720 (1#2):
analytical separation methods. A variety of different 173}182.
Kakehi K and Honda S (1996) Analysis of glycoproteins,
separation modes can be applied to separate complex
glycopeptides and glycoprotein-derived oligosaccharides
carbohydrate mixtures. It has been shown that, depend- by capillary electrophoresis. Journal of Chromato-
ing on the properties of the sample, different separation graphy A 720 (1#2): 377}393.
methodologies can be applied and optimized. Lee KB, Loganathan D, Merchant ZM and Linhardt RJ
The electrophoretic analysis of carbohydrates is (1990) Carbohydrate analysis of glycoproteins. A
still under development. In particular the analysis of review. Applied Biochemistry and Biotechnology 23 (1):
biochemical compounds, such as glycoconjugates, is 53}80.
still a challenge and can give new information about Linhardt RJ and Pervin A (1996) Separation of nega-
cell mechanisms, structure of antibodies, etc. This tively charged carbohydrates by capillary electro-
article has focused on the basic aspects of capillary phoresis. Journal of Chromatography A 720 (1#2):
electrophoresis of carbohydrates. The carbohydrates 323}335.
Oefner P, Chiesa C, Bonn G and Horvath C (1994) Devel-
described were conRned to simple model compounds.
opments in capillary electrophoresis of carbohydrates.
For more speciRc information, the Further Reading Journal of Capillary Electrophoresis 1 (1): 5}26.
list should be consulted. Olechno JD and Nolan JA (1997) Carbohydrate analysis by
capillary electrophoresis In: Handbook of Capillary
See also: III/Ion Analyses: Capillary Electrophoresis. Electrophoresis, 2nd edn, pp. 297}345. Cleveland, OH:
CRC Press.
Further Reading Paulus A and Klockow A (1996) Detection of carbohy-
drates in capillary electrophoresis. Journal of Chromato-
El Rassi Z (1994) Capillary electrophoresis of carbohy- graphy A 720 (1#2): 353}376.
drates. Advances in Chromatography 34: 177}250. Voegel PD and Baldwin RP (1997) Electrochemical detec-
El Rassi Z and Nashabeh W (1995) High performance tion in capillary electrophoresis. Electrophoresis 18
capillary electrophoresis of carbohydrates and glyco- (12}13): 2267}2278.
Gas Chromatography and Gas Chromatography}Mass

Spectrometry
A. Fox, M. P. Kozar and P. A. Steinberg, thought developed, relating to whether the anomeric
University of South Carolina, centre should be retained or destroyed in derivatiz-
Columbia, SC, USA ation. Both persist to this day. Sugars exist in equilib-
Copyright ^ 2000 Academic Press rium betweeen ring and straight chain forms. If the
anomeric centre is not eliminated, derivatization Rxes
the sugars in the anomeric ring forms. Thus, two to
Overview of Derivatization of Sugars four peaks will be produced from each L or D sugar
for GC, GC-MS or GC-MS-MS Analysis (two from furanose and two from pyranose anomers).
Table 1 overviews many of the important events in Interpretation of chromatograms becomes complic-
chromatographic and mass spectrometric analysis of ated and quantitation difRcult. Acidic sugars (i.e.
sugar monomers. These will be discussed at appropri- generally with carboxyl groups) require additional
ate points in this review. It was not until the years derivatization steps. The carboxyl moiety may be
1961}1963 that gas chromatography (GC) was converted to a lactone, ester or reduced to an alditol.
applied to the quantitative analysis of mixtures of
The Alditol Acetate Procedure
neutral and amino sugar monomers, even though
individual sugars had been derivatized earlier. Carbo- The alditol acetate procedure was the Rrst developed
hydrate analysis by GC lagged behind the analysis in which the anomeric centre is eliminated. The
of many other compounds because of the difRculty anomeric centre is converted by borohydride reduc-
in producing volatile derivatives. Two schools of tion (although later borodeuteride was introduced)
2212 III / CARBOHYDRATES / Gas Chromatography and Gas Chromatography^Mass Spectrometry
Table 1 Important advances in analysis of carbohydrates
1960}1965 Fundamental articles on preparation of sugar derivatives for gas chromatography (GC)
1970s Introduction of gas chromatography}mass spectrometry (GC-MS) for structure analysis
of carbohydrates
Late 1970s}early 1980s Introduction of selected ion monitoring (SIM) GC-MS for trace analysis
of derivatized carbohydrates
1980s Development of anion exchange liquid chromatography/pulsed amperometric detection
of native sugars
1990s Introduction of liquid chromatography}electrospray mass spectrometry for analysis
of underivatized sugars
1995}1996 Introduction of gas chromatography}tandem mass spectrometry for trace analysis
of derivatized sugars
1995}1996 Introduction of liquid chromatography}tandem mass spectrometry for identification
of native sugars
to an alcohol prior to acetylation (to produce a and/or borate, acetylation can be somewhat un-
volatile derivative). Unfortunately, borate derived predictable. When using sodium acetate as a catalyst
from the added borohydride, or borodeuteride, in- following removal of borate and moisture, these
hibits the subsequent acylation step. The classical problems do no occur but unfortunately, the process
procedure employs a multi-step evaporation with to remove borate and moisture is tedious to perform.
methanol}acetic acid to remove borate as tetramethyl An automated evaporator has been developed to per-
borate gas in between the reduction and acetylation form the multiple cycles of methanol}acetic acid ad-
steps (Figure 1). dition/evaporation for borate removal in the alditol
In the 1960s, sodium acetate and pyridine were acetate procedure. More recently, an automated
introduced as catalysts for acetylation, pyridine being derivatization instrument has been developed to
more efRcient than sodium acetate. However, in both automate the entire alditol acetate procedure.
cases borate must be removed prior to acetylation.
The Trimethylsilyl Procedure
Subsequently, several catalysts (including methyl-
imidazole) were described that allowed acetylation Also in the early 1960s, other workers described
without removal or borate or moisture. Unfortunate- a simple one-step trimethylsilyl derivatization of hy-
ly, many catalysts (including both pyridine and droxyl groups. This remains one of the most common
methylimidizole) often generate side-reaction prod- procedures used for GC analysis. Derivatization is
ucts (from reaction with acetic anhydride) that pro- readily achieved. However, complex chromatograms
duce chromatograms contaminated with extraneous are produced due to anomer formation noted above.
peaks. Furthermore, in the presence of moisture Also, analysis must be prompt since the derivatives
Figure 1 An example of an alditol acetate derivatization reaction employing sodium borodeuteride in the reduction step. Differences
in the mass spectra of ribose and ribitol illustrate the distinctness of mass spectra of alditols and aldoses.
decompose in the presence of moisture, complicating Choice of Derivatives for Instrumental Detection
clean-up of complex samples. Most reports in the 1960s and 1970s employed the
The Aldononitrile Acetate and O -Methyloxime Same ionization detector (FID). While the FID is
Acetate Procedures useful as a GC detector, it lacks speciRcity. The use of
the mass spectrometer (MS) or tandem mass spec-
Procedures developed in the 1970s involving destruc- trometer (MS-MS) as a GC detector dramatically
tion of the anomeric centre were particularly con- expands the capability of the analysis. GC-MS, using
cerned with proceeding directly to the Rnal acylation electron impact ionization (EI) or chemical ionization
step. Such procedures generally employed acetylation (CI) followed by positive ion detection, employs the
as the Rnal acylation step due to the stability of same derivatives as in FID but provides speciRcity
acetylated sugar derivatives. Such methods included that the FID lacks.
the aldononitrile acetate and the O-methyloxime There have been a few reports using the electron-
acetate procedures. To generate the former, the al- capture detector (ECD) offering the possibility of
dehyde is reacted with hydroxylamine to produce an increased sensitivity if a halogenated (electron-
oxime which is converted to a nitrile on acetylation. capturing) derivative such as triSuoroacetyl,
O-methyloxime acetates, on the other hand, are pre- pentaSuoropropyl or heptaSuorobutyl is employed.
pared by destruction of the anomeric centre by reac- Unfortunately, such derivatives are unstable in the
tion with O-methyl-hydroxylamine, also followed by presence of moisture. Furthermore, compounds other
acetylation. Unfortunately, O-methyloxime acetates than sugars may be converted to halogenated deriva-
display a new isomeric centre and two products, syn- tives, resulting in increased background. Thus, it is
and anti, are generated for each sugar. unclear whether the increased sensitivity can be
An advantage of both aldononitrile and O-methyl- utilized. An unusual derivative not widely used is
oxime acetate methods is that aldoses produce dis- the O-pentaSuorobenzyl oxime acetate derivative.
tinct peaks from alditols. However, substitution of For this derivative, pentaSuorobenzyl oxime replaces
borodeuteride for borohydride reduction (in alditol the aldehyde as the electron-capturing group prior to
acetate formation) labels the aldose group, whilst acetylation. Unlike other halogenated derivatives, the
alditols remain unlabelled. Thus, aldoses and alditols O-pentaSuorobenzyl oxime acetate is stable to moist-
can be distinguished by gas chromatography}mass ure. Alternatively, such derivatives might be appro-
spectrometry (GC-MS) analysis. Alternatively, al- priate for use with MS in the negative ion mode. Like
ditols do not contain an anomeric centre and can be ECD, negative ion chemical ionization (NI-CI) GC-
acetylated without prior derivatization steps (e.g. re- MS also necessitates an electron-capturing derivative.
duction) and are simply analysed. This also allows Use of GC-MS-MS might take further advantage of
discrimination of alditols from aldoses. For further the increased sensitivity of NI-CI since background is
details see the section on GC-MS analysis of alditol essentially eliminated.
acetates, below. The Rrst benchtop GC-MS instruments were intro-
The Tri]uoroacetyl Procedure duced in the late 1970s and had only EI capability.
Later instruments could perform both EI and CI with
TriSuoroacetyls, like trimethylsilyl derivatives, are both positive and negative ion detection capabilities.
also formed by a simple one-step derivatization pro- In the past 3}4 years, comparably priced benchtop
cedure. However, as with trimethylsilyl derivatives, GC-MS-MS instruments, also with EI-CI and posit-
multiple peaks are generated for each sugar and are ive/negative ionization capability, have been intro-
unstable on exposure to moisture. Therefore, samples duced. Most GC-MS and GC-MS-MS analyses of
must be analysed soon after derivatization. As noted sugars are still performed with EI in positive ion
above, acetate derivatives are generally stable indeR- detection mode. These instruments are simple to
nitely. operate and maintain and are run by Windows-
Chiral Derivatives based PCs.
Methods developed have followed two approaches:
after conventional derivatization (e.g. permethyla- Preparation of Alditol Acetate
tion) enantiomers (L and D isomers) can be separated Derivatives of Sugars Present
on chiral (e.g. cyclodextrin) GC columns; alterna-
tively, glycosidation with chiral reagents (e.g. an
in Complex Matrices
optically active alcohol such as butanol) produces Steps in the analysis of complex sugar mixtures in-
diastereoisomers that can be resolved on conven- clude release of the sugar by hydrolysis, addition of
tional nonchiral capillary columns. internal standard, derivatization and instrumental
analysis. Examples of analyses performed with the aminosugar (e.g. methylglucamine) as an internal
alditol acetate procedure will be used here to illus- standard for amoinosugars dramatically increases
trate these steps. precision. However, on less polar columns, such as
the DB-5ms, certain high molecular weight neutral
Hydrolysis
sugars (e.g. heptoses) tend to elute in the same region
There have been numerous articles on the selection of as aminosugars (e.g. aminohexoses). In the analysis of
hydrolysis conditions. Parameters requiring consid- heptoses, when using arabinose as the internal stan-
eration include temperature, duration and the type dard (by comparing relative peak areas), the RSD for
and strength of acid (primarily hydrochloric, sulfuric D-glycero-D-mannoheptose and L-glycero-D-manno-
and triSuoroacetic acids). Optimization is highly de- heptose was found to be 25.3% and 30.7%, respec-
pendent on the sugar(s) of interest to be released from tively. Using methylglucamine, RSD was lowered to
a particular polymeric matrix. If the hydrolysis condi- 11.0 and 7.2%. This suggests that, when selecting an
tions are too gentle, there is an inadequate release of internal standard, the relative retention time of the
the sugar monomers. Conversely, if the conditions are eluting sugars, rather than similarity in structure, may
too harsh, then destruction of certain sugars occurs. be more critical. Heptoses are highly characteristic of
Neutral sugars are often easier to release and destroy Gram-negative bacteria.
than aminosugars. Invariably, a compromise must be The aldoses L-glycero-D-mannoheptose and D-gly-
made in the analysis of a mixture of neutral and ceroheptose are not available commercially. On
aminosugars in a complex matrix. It must also be reduction they are converted into their respective
recognized that it is not the absolute amount of poly- heptitols. However, the ketose heptulose, which is
meric sugar present in the matrix that is being deter- available commercially, on reduction generates both
mined, but the amount released as monomers under L-glycero-D-mannoheptitol and D-glycero-D-manno-
the selected hydrolysis conditions. heptitol. Hydrolysates of Escherichia coli contain
Following hydrolysis and prior to derivatization, L-glycero-D-mannoheptose but not D-glycero-D-
the acid must be removed. TriSuoroacetic and hydro- mannoheptose. Thus, by comparing chromatograms
chloric acids are generally removed by evaporation. of standards containing heptulose and hydrolysates
Under these conditions further destruction of sugars of E. coli (as alditol acetates) it is possible to identify
is possible. Sulfuric acid can be removed by neutral- the two heptose peaks. Figure 2 shows a typical sep-
ization with a solution of barium hydroxide or a aration of a mixture of neutral and amino sugar
suspension of barium carbonate. Unfortunately, on standards (including both heptose peaks) on a DB-
neutralization with a barium hydroxide solution, it 5ms fused silica capillary column and a hydrolysate
is difRcult to remove the sugar from the precipitate of of E. coli (containing L-glycero-D-mannoheptose but
barium sulfate. The use of barium carbonate is not not D-glycero-D-mannoheptose).
practical because a coat of barium sulfate forms,
protecting a large portion of the particulate barium The Automated Derivatization Instrument
for Preparation of Alditol Acetates
carbonate from reacting. Thus, large quantities of
barium carbonate are needed for neutralization. A The manually performed alditol acetate derivatiz-
simple alternative involves neutralization with a solu- ation procedure described elsewhere has been used
tion of an organic base (N,N-dioctylmethylamine) in for many years. This multi-step procedure can now
chloroform. Sugars remain in the aqueous phase and be performed sequentially by a computer-controlled
sulfate is removed in the organic phase. instrument. The procedure, which previously took
212 working days, is performed in an automated
Choice of Internal Standard
fashion, requiring a total of 90 min manual work.
There is a great deal of variability of sugar monomers The samples are processed in four stages:
that can be present in complex biological matrices.
1. evacuation/hydrolysis (3 h 15 min, automated);
Quantitation of each sugar present in a proRle is
2. prederivatization clean-up (1 h, manual);
desirable, but it is not practical to select one internal
3. alditol acetate derivatization (23 h, automated);
standard for each sugar. Generally, internal standards
4. post-derivatization clean-up (30 min, manual).
are selected for groups of sugars based on structural
similarities. On SP-2330 columns, neutral sugars The core of the machine, where chemical manip-
elute much earlier than aminosugars. Relative stan- ulations and reactions are performed, consists of
dard deviations (RSD) for multiple samples on GC a custom-built manifold with 21 glass chambers, to
analysis for late-eluting aminosugars are high if an each of which a test tube is attached. The manifold is
early eluting neutral sugar (e.g. arabinose) is used seated in a movable heating block. A series of electri-
as the internal standard. Selection of a late-eluting cally driven solenoid valves are attached in-line with
Figure 2 Selected ion monitoring (SIM) GC-MS chromatograms of alditol acetates (A) sugar standards and (B) hydrolysate of
Escherichia coli. The following m/z were monitored: 160 (deoxyribose); 171 (rhamnose and fucose); 145 (ribose, arabinose and
xylose); 199 (inositol, 199); 290 (mannose, glucose and galactose); 318 (glucosamine, galactosamine, mannosamine); 362 (D,D-
heptose and L,D-heptose); 168 (muramic acid); 327 (methylglucamine).
the manifold. A set of solvent valves control the input Following hydrolysis, internal standards are added,
of solvent and/or nitrogen gas to each sample cham- the samples are removed from the instrument, neu-
ber. A set of gas valves controls output to atmosphere tralized with 2 mL 50% N,N-dioctylmethylamine
or vacuum. Additionally, closure of all valves allows (Fluka, Buchs, Switzerland) in chloroform, and then
the samples to be sealed in a closed chamber. centrifuged. The aqueous phase containing the sugar
Computer control of the individual stages of the monomers is removed and passed through C18 col-
derivatization process is a major feature of the sys- umns (J&W, Folsom, CA) into 21 new sample tubes
tem. Ten mg of each sample, in 1 mL of 2 mol L\1 via the evacuated 21-chamber manifold described
sulfuric acid, is placed in each of the custom test above. Aqueous sodium borodeuteride 200 L
tubes. The samples are attached to the manifold and (25 mg mL\1) is then added to each sample.
the program started. Oxygen is evacuated by repeated The derivatization procedure is entirely under
alternate exposure of nitrogen and vacuum. After computer control. After a 2 h delay, in which
evacuation, the program sets the heating block to sample reduction occurs at room temperature, meth-
1003C for hydrolysis. Heating continues for 3 h anol}acetic acid (200 : 1 v/v) is added by activation of
under nitrogen. a solenoid valve connected to a reagent reservoir. The
program then sets the heating block to 603C, and can be chosen for selected ion monitoring (SIM).
evaporation under N2 occurs for 30 min. This step Once retention time has been established, SIM
(solvent addition and evaporation) is automatically chromatograms allow for clean, easily quantiRable
repeated servaral times to remove borodeuteride as peaks for each monomer. Figures 3 and 4 compare
tetramethyl borate gas. After the last addition, the total ion and SIM GC-MS analysis of two bacterial
system is evacuated by activation of the attached hydrolysates (Bacillus subtilis strains 168 and W23).
vacuum pump, and the samples are dried for 4 h at In the total ion chromatograms there are a number
room temperature. Acetic anhydride is then added to of tiny background peaks (e.g. the glucose and
the samples from another reservoir, and the samples methylglucamine peaks both have shoulders). The
acetylated for 13 h at 1003C. Finally, the samples are extraneous peaks are eliminated in the SIM
evaporated to dryness under N2, and chloroform is chromatograms. Furthermore, sensitivity is dramati-
added from a third reservoir. cally increased in SIM analysis. This is illustrated in
The Rnal post-derivatization clean-up (taking that galactosamine is readily detected in the SIM }
30 min) is performed manually but also uses the 21- but not the total ion chromatogram } for strain 168.
sample manifold, alleviating the necessity for addi- In such analyses, generally g amounts of sugars are
tional equipment. Samples are passed through a pair present but, in our experience, SIM GC-MS can be
of connected Chem-Elut columns (Varian, Walnut used to detect as little as 100}250 ng in complex
Creek, CA), the Rrst pre-treated with 2 mol L\1 samples (10}20 mg of starting sample). However,
acetic acid and the second with 14.8 mol L\1 am- background peaks become increasingly more of
monium hydroxide. The chloroform eluent is evapor- a problem as the concentration of sugar is decreased.
ated under N2, and samples reconstituted for analysis. As noted above, a native sugar forms 2}4 anomers
upon acylation, thus creating multiple peaks from
a single sugar and complicating chromatograms.
Instrumental Analysis of Alditol The anomeric centre is usually Rrst destroyed. If
Acetate Derivatives hydroxylamine and O-methylhydroxylamine are
Columns of GC Analysis used in the acylation of carbohydrate monomers,
aldononitrile acetates and O-methyloxime acetates,
There was a great deal of work in the early days in the respectively, generate distinct chromatographic
selection of packed columns for separation of a com- peaks for aldoses and alditols. This is not the case for
plex mixture of neutral and amino sugars by GC. This alditol acetates.
work was readily extrapolated to the vastly improved Sugars are generally reduced using sodium
fused silica columns introduced later. As an example, borohydride or borodeuteride prior to acylation to
excellent capillary GC separation of neutral and eliminate anomer formation. For example, during
aminosugar mixtures is obtained on relatively polar reduction of aldoses the C1 aldehyde is converted to
SP-2330 columns. However, aminosugars require an alcohol (i.e. aldose to an alditol), whereas alditols
high Rnal temperatures and/or extended run times for remain chemically unchanged. Using sodium
elution and this column tends to display poor temper- borohydride, in the formation of alditol acetates,
ature stability under such conditions. Furthermore, aldoses and alditols cannot be differentiated. How-
irreversible adsorption of aminosugars is a signiRcant ever, when using sodium borodeuteride, two
problem, causing poor sensitivity of the aminosugars deuteriums are added to the aldehyde moiety, one of
relative to neutral sugars. More recently, nonpolar which remains after acylation. Thus, there is a one
DB-5ms columns have been used which have not mass unit shift in ions containing C1. Fragments lack-
exhibited these problems. In complex mixtures, ing C1 generate ions of the same m/z for deuterated
sugars are observed as sharp peaks with almost base- and nondeuterated samples.
line resolution. It would be highly desirable if a com- As an example, B. subtilis W23 is readily dis-
mercial column were developed that displayed the criminated from B. subtilis 168 by the presence of
stability of the DB-5ms column but had the resolving a ribitol containing polysaccharide (teichoic acid).
capacity of the SP-2330 column. Total ion and SIM chromatograms of carbohydrates
derived from these two bacterial strains are shown in
GC-MS Analysis of Alditol Acetates
Figures 3 and 4 respectively. During hydrolysis,
(Total Ion Spectra and SIM)
ribitol is released from the teichoic acid. Unfortu-
Using total ion GC-MS, carbohydrate monomers nately, RNA releases ribose upon hydrolysis which
can readily be identiRed by their characteristic ion is converted to ribitol upon reduction, thus hindering
spectra. Once the sugar has been identiRed, a particu- the detection of ribitol originating from teichoic acid.
lar ion or set of ions characteristic of that monomer However, ribitol from teichoic acid is also partially
Figure 3 Total ion GC-MS chromatograms of alditol acetates of hydrolysates of Bacillus subtilis (A) strain W23 and (B) strain 168.
Ribose is present at 1.13% of the dry weight of the sample of W23 analysed (total 10 mg), i.e. 113 g.
converted to anhydroribitol during hydrolysis. Strain C1 end labelled with deuterium and the other, un-
W23 can therefore be readily distinguished from labelled end. For alditols, C1 is not labelled, resulting
strain 168 by the presence of anhydroribitol. The in single peaks. Thus, the mass spectrum of ribose
mass spectrum of anhydroribitol and its structure contains pairs of ions of nearly equal abundance (i.e
are shown in Figure 5. The molecular weight of an- m/z 115/116, 145/146, 187/188 and 217/218) gener-
hydroribitol is 260; m/z 187 (M-73, breakage be- ated from the two ends of the molecule, while ribitol
tween C4 and C5, which leaves the ring intact), m/z would be dominated by single ions (i.e. m/z 115, 145,
127 (loss of acetic acid, 60) and m/z 85 (loss of 187 and 217). Therefore, for strain W23 where
ketene, 42). ribitol, in addition to ribose, is present, the mass
On examination of SIM chromatograms, the spectrum contains a greater abundance of the lower
ribitol/ribose peak cannot be used to distinguish the mass ion of each pair than for strain 168 (where
two organisms. However, this can be accomplished ribose alone is present). Figure 6 clearly demonstrates
by full scan (GC-MS) analysis. As noted above, on that standard ribose has a mass spectrum indistin-
borodeuteride reduction, aldoses (e.g. ribose) gain guishable from the ribose peak derived from the hy-
two deuteriums, one of which remains after acyla- drolysate of strain 168. However, the peak for
tion, while alditols (e.g. ribitol) remain unchanged. strain W23, containing a mixture of ribose and
For aldoses, pairs of peaks are generated, from the ribitol, is readily distinguished by the dominance of
Figure 4 Selected ion monitoring GC-MS chromatograms of alditol acetates of hydrolysates of Bacillus subtilis (A) strain W23 and
(B) strain 168. Same ions as Figure 2 plus 187 (anhydroribitol).
115 over 116, 145 over 146, 187 over 188 and 217 which can be quite striking. Aminodideoxyhexoses,
over 218. quinovosamine and fucosamine (found in legionel-
The base peak in EI mass spectra of alditol acetates lae), have been noted to display distinct mass spectra.
is generally dominated by the acetylinium ion (m/z Differences in mass spectra among isomers are
43). Many primary fragments are produced by cleav- accentuated by the use of borodeuteride. Aldoses are
age between sequential carbon atoms. Secondary asymmetric since there is an aldehyde on C1. Asym-
fragmentation results from losses of acetic acid (m/z metry is retained after borodeuteride but not
60), acetoxyl groups (m/z 59) and ketene (m/z 42). borohydride reduction since C1 is labelled. It has been
Generally, mass spectra of aminosugars are relatively proven that all eight hexoses can be differentiated by
simple since cleavage preferentially occurs between a combination of distinct mass spectra and/or reten-
the carbon with attached acetamido group and adjac- tion times.
ent acetylated carbons. Muramic acid, 3-O-lactyl glucosamine, is an un-
Mass spectra of stereoisomers of alditol acetates usual sugar which additionally contains a carboxyl
contain ions of the same m/z. On casual observation group in ether linkage. A lactam (a cyclic amide) is
the mass spectra appear similar. However, certain formed by internal dehydration between its carboxyl
isomers display differences in relative ion abundances and amino groups on derivatization. In contrast to
Figure 5 Mass spectrum of alditol acetate of anhydroribitol derived from hydrolysate of Bacillus subtilis W23.
acetylation of other aminosugars, which produce am- provides even greater speciRcity in detecting trace
ides, muramicitol pentaacetate (acetylated muramic amounts of chemical markers in complex matrices.
acid) has an imido group in which two acyl groups, Tandem mass spectrometry has the added advantage
lactyl and acetyl respectively, are linked to the nitro- of generating a total ion spectrum from a selected
gen atom. Formation of the imido moiety requires precursor ion (product ion spectrum). The resulting
harsh conditions (higher temperatures and longer product ion spectrum can be used for a deRnitive
heating times). identiRcation of the compound of interest at trace
Naturally occurring O-methylated sugars exist in levels.
bacteria and in eukaryotes. The fragmentation Multiple reaction monitoring (MRM) and genera-
pattern of methylated sugars is distinctive. Fragmen- tion of product ion spectra both involve three discrete
tation between the O-methylated carbon and the ad- mass analysis steps. The Rrst stage involves selection
jacent acetylated carbon atoms dominates the spectra. of a precursor ion. This instrumental clean-up re-
Additional secondary ions can be produced by loss of moves other ions. The precursor ion is then frag-
methanol (m/z 32) and formaldehyde (m/z 30). mented by collision-induced dissociation (CID) using
an inert gas. In the third stage, all precursor ions can
GC-MS-MS Analysis of Alditol Acetates (Multiple
be collected (product ion spectrum) or a single prod-
Reaction Monitoring and Product Ion Spectra)
uct ion is selected for monitoring (MRM). Both SIM
High resolution chromatographic separations coupled GC-MS and MRM GC-MS-MS analysis allow excel-
with selective clean-up steps are important in improv- lent quantitation of such chemical markers, but the
ing the speciRcity of the detection of chemical latter provides much greater conRdence in trace
markers (e.g. muramic acid as a marker for bacterial analysis.
infection) in complex matrices. However, chromato- Two types of GC-MS-MS instruments are prim-
graphic separation is not sufRcient to eliminate ex- arily used in such analysis: ion traps and triple quad-
traneous peaks when nonselective detectors are em- rupoles. In triple quadrupole instruments, the three
ployed. The use of the mass spectrometer as a selec- stages of analysis are performed using three distinct
tive GC detector (i.e. GC-MS analysis in SIM), helps quadrupole mass analysers. There is some decrease in
greatly in diminishing background noise by focusing sensitivity due to loss of ions in transmission through
only on ions that are present in the compound of the three quadrupoles. In ion trap tandem mass spec-
interest. However, even when using SIM, it is not trometers, the three stages occur in the same mass
uncommon to Rnd extraneous background peaks. analyser. This dramatically simpliRes the instrument
The tandem mass spectrometer, as a GC detector, and its cost. Furthermore, sensitivity of MS-MS
Figure 6 Mass spectra of alditol acetates (A) ribose standard (B) ribose present in hydrolysate of strain 168 (C) a mixture of ribitol
and ribose present in hydrolysates of Bacillus subtilis W23. Ions derived from the C1 end are labelled with deuterium (e.g. m/z 146). Ions
generated from the C6 end are not labelled and thus m/z is one less (e.g. m/z 145).
analysis is improved, particularly in product ion been used as internal standards in the analysis of
spectrum mode. However, in trace quantitative anal- muramic acid (a unique chemical marker for bacteria
ysis of carbohydrates, in MRM mode, it has been not found elsewhere in nature). While quantitation is
observed that the ion trap is less precise than the triple better with the latter, there is a possibility of contami-
quadrupole. However, the low cost, ease of use of the nation of the 13C-labelled muramic acid with non-
ion trap and its power for absolute identiRcation labelled muramic acid. Thus, at the detection limit
(product ion spectrum) make its use extremely at- of the procedure, where the major issue is to prove
tractive for diagnostic applications. the presence or absence of muramic acid, methyl-
It is important to note that heavy isotope-labelled glucamine is preferred. At higher levels, where quan-
internal standards for many sugars are unavailable as titation is the major issue, 13C-labelled muramic acid
pure compounds. Whole 13C-labelled bacterial cell is preferred. Obviously an internal standard with
hydrolysates may be used as a cheap alternative both characteristics would be optimal and this is
source. As an example, in quantitative studies, both currently under investigation. Isomuramic acid has
methylglucamine and 13C-labelled muramic acid have been described as a natural component of the
the ion trap is the utility of the product ion spectrum

for absolute identiRcation of trace levels in complex
matrices. This is likely to have particular utility for
studies of clinical specimens to determine the presence
of bacteria (that are difRcult to culture) or their nonvi-
able cell wall components (not detectable by culture).
Sterile body Suids and tissues from healthy humans or
animals do not generally contain muramic acid.
Figure 7 shows GC-MS-MS chromatograms of cer-
ebrospinal Suid from a patient with bacterial menin-
gitis. The full scan (precursor ion m/z 403 from
muramic acid and m/z 327 for, the internal standard,
methylglucamine) would be similar in appearance to
a SIM GC-MS chromatogram (m/z 403 and m/z 327
respectively). Muramic acid is readily detected, but
there are clearly numerous background peaks, parti-
cularly in the m/z 327 window. For comparison,
using multiple reaction monitoring (m/z 403 to m/z
198 transition) background peaks are eliminated. The
sample contained a total of 13.6 ng of muramic acid.
Figure 8, for comparison, shows full scan and MRM
Figure 7 GC-MS-MS shows chromatograms of cerebrospinal

fluid from a patient with bacterial meningitis. (A) Full scans (pre-
cursor ion m/z 403 for muramic acid and m/z 327 for methyl-
glucamine) would have a similar appearance to a SIM GC-MS
chromatogram (m/z 403 and m/z 327 respectively); (B) multiple
reaction monitoring (m/z 403 to m/z 198 transition). 250 L of
cerebrospinal fluid was analysed and found to contain a total of
13.6 ng muramic acid. Methylglucamine 500 ng (internal stan-
dard) was present in the sample.
lipopolysaccharide of certain bacteria and also has

been chemically synthesized. Whether muramic acid
and isomuramic acid can be adequately chromato-
graphically resolved has not yet been addressed.
Application of GC-MS-MS in Trace Analysis
Ion trap GC-MS-MS has been used for absolute iden-
tiRcation of trace levels of muramic acid in human
body Suids. This is the only report to date using
GC-MS-MS to detect muramic acid or any other
marker for bacteria in a human/animal body Suid or
tissue. Product ion mass spectra (upon MS-MS analy-
sis) of muramic acid peaks (530 ng mL\1) found in
infected human body Suids were identical to those of
Figure 8 GC-MS-MS chromatograms of uninfected cerebrospi-
pure muramic acid. An illustration of the detection of nal fluid from a patient with otitis media. (A) Full scan of precursor
muramic acid in human cerebrospinal Suid is shown m/z 403 (equivalent in appearance to SIM GC-MS); (B) multiple
(approximately 5 ng mL\1). A powerful feature of reaction monitoring (m/z 403 to m/z 198 transition).
Figure 9 Product ion spectrum (GC-MS-MS) of m/z 403 (A) muramic acid standard (250 ng total) and (B) cerebrospinal fluid from
patient with bacterial meningitis (2.7 ng in 400 L of cerebrospinal fluid analysed).
of cerebrospinal Suid from a patient with otitis media MRM analysis has great utility for determining the
(inner-ear infection). In this instance, culture in- levels of bacterial contamination for clinical and
dicated the absence of bacteria in the sample. As environmental analyses. For example, muramic acid
expected, there is no peak at the retention time of levels have been demonstrated to serve as a useful
muramic acid; this serves as a negative control. measure of biocontamination of air. This has rel-
A product ion spectrum of muramic acid present in evance to our understanding of the ‘sick building
a cerebrospinal Suid (2.7 ng total) and a muramic phenomenon’ and assessing the quality of indoor air.
acid standard (250 ng) are compared in Figure 9. The product spectrum is of limited use in the analysis
Muramic acid is clearly categorically identiRed from of environmental samples, since bacteria are inva-
the spectrum of the patient’s sample, although back- riably present. However, as noted above, it is
ground ions are obvious. Clearly, this represents an a powerful tool for absolute detection of bacteria in
analysis close to the detection limit, for GC-MS-MS human body Suids and tissues which are sterile in the
of this type of complex biological sample. absence of infection.
Gas Chromatography and Liquid Chromatography ation is usually eliminated. The limit of detection is
In the 1980s successful analysis of underivatized still currently much lower with GC-based procedures.
sugars using high performance liquid chromato-
graphy was introduced. Eliminating derivatization See also: II/Chromatography: Gas: Detectors: Mass
Spectrometry. III/Carbohydrates: Electrophoresis; Liquid
dramatically reduces sample handling and chemical
Chromatography; Thin-Layer (Planar) Chromatography.
manipulation. After separation on anion exchange
Polysaccharides: Liquid Chromatography.
columns (particularly the PA1 column) sugars are
detected with a pulsed amperometric detector (PAD).
This LC system has been successfully used for the Further Reading
separation and identiRcation of amino, neutral and
Biermann C and McGinnis G (eds) (1989) Analysis of
acidic sugars. Chromatography is performed in con- Carbohydrates by GLC and MS. Boca Raton: CRC Press.
centrated NaOH. At high pH, interaction of ionized Conboy JJ and Henion J (1992) High performance
hydroxyl groups with the anion exchange resin pro- anion-exchange chromatography coupled with mass
duces excellent sugar separations. The electrochemical spectrometry for the determination of carbohydrates.
(PAD) detector also detects carbohydrates with excel- Biological Mass Spectrometry 21: 397.
lent sensitivity at alkaline pH. Unfortunately, in the Fox A and Black G (1994) IdentiRcation and detection of
analysis of sugars in complex matrices, the nonselec- carbohydrate markers for bacteria: derivatization and
tive PAD detector often has inadequate speciRcity to gas chromatography}mass spectrometry. In: Fenselau
discriminate sugars of interest from background noise. C (ed.) Mass Spectrometry for the Characterization of
When chromatography is performed in conjunc- Microorganisms, p. 107. Washington, DC: American
Chemical Society.
tion with the mass spectrometer (e.g. GC-MS or LC-
Fox A, Schwab JH and Cochran T (1980) Muramic
MS), the increased selectivity of detection allows acid detection in mammalian tissues by gas}liquid
analysis of less puriRed samples. For example, whole chromatography}mass spectrometry. Infection and Im-
cell hydrolysates can be analysed. With LC-PAD munity 29: 526.
analysis, structural components such as glycoproteins Fox A, Wright L and Fox K (1995) Gas chromatography
or polysaccharides must be puriRed prior to analysis. tandem mass spectrometry for trace detection of
LC-MS-MS has been used for quantitation of sugars muramic acid, a peptidoglycan marker in organic dust.
in complex environmental matrices. LC-MS and LC- Journal of Microbiological Methods 22: 11.
MS-MS for the analysis of carbohydrates show great Gunner SW, Jones JKN and Perry MB (1961) Analysis of
promise but are still in the developmental stage. Cur- sugar mixtures by gas}liquid partition chromatography.
rently, GC-MS and GC-MS-MS have considerably Chemistry and Industry (London) 255.
lower detection limits than LC-MS and LC-MS-MS. Hardy MR, Townsend RR and Lee YC (1988) Monosac-
charide analysis of glycoconjugates by anion exchange
Automation of sugar derivatization for GC eliminates
chromatography with pulsed amperometric detection.
many of the advantages of LC-based techniques over Analytical Biochemistry 170: 54.
the GC-based ones but the only derivatization pro- LoK nngren J and Svenssen S (1974) Mass spectrometry in
cedure for carbohydrate analysis by GC that has been structural analysis of natural carbohydrates. In: Tipson
automated, at this time, is the alditol acetate procedure. R and Horton D (eds) Advances in Carbohydrate Chem-
istry and Biochemistry, Vol. 29, p. 41. Amsterdam:
Conclusions Academic Press.
Rocklin RD and Pohl CA (1983) Determination of carbo-
GC-MS and GM-MS-MS are mature techniques hydrates by anion exchange chromatography with pul-
which may be used in the analysis of carbohydrate sed amperometric detection. Journal of Liquid
monomers in complex samples. In SIM GC-MS and Chromatography 6: 1577.
MRM GC-MS-MS, sugars are readily detected and Sawardeker JS, Sloneker JH and Jeanes A (1965) Quantita-
quantitated in complex samples, although use of tive determination of monosaccharides as their alditol
MRM dramatically lowers the detection limit. Using acetates by gas}liquid chromatography. Analytical
total ion (MS) and product ion spectra (MS-MS), Chemistry 37: 1602.
identiRcation at trace levels is readily performed. The Simpson RC, Fenselau CC, Hardy MR et al. (1990) Ad-
aptation of a thermospray liquid chromatography/mass
detection limit is much lower for the latter when the
spectrometry interface for use with alkaline anion
ion trap GC-MS-MS is used. Substantial improve- exchange liquid chromatography of carbohydrates.
ments have been made to sample preparation, includ- Analytical Chemistry 62: 248.
ing simpliRcation and computer-controlled automa- Sweeley CC, Bentley RS, Makita M and Wells WW (1963)
tion of derivatization reactions for GC analysis. This Gas}liquid chromatography of trimethylsilyl derivatives
is likely to help GC-based techniques to remain com- of sugars and related substances. Journal of the Ameri-
petitive with their LC competitors where derivatiz- can Chemical Society 85: 2497.
2224 III / CARBOHYDRATES / Liquid Chromatography
C. Corradini, Istituto di Cromatografia det CNR, graphic methods driven by polar interactions, which
Roma, Italy involve partitioning between the more hydrophobic
Copyright ^ 2000 Academic Press mobile phase and a layer of mobile phase enriched
with water and partially immobilized on the station-
ary phase. Since 1975 aminopropylsiloxane-bonded
silica columns have been used widely for this type of
Introduction
chromatography. The fundamental mechanism gov-
The development of selective and sensitive methods erning separation of carbohydrates using amino-
for analyses for carbohydrates is one of the most propylsiloxane-bonded phases is partition between
challenging areas of analytical chemistry. The main the mobile phase and the water-enriched solvent
difRculties in the analysis of carbohydrates arise associated with the stationary phase.
from their considerable number of isomeric forms The most common bonded phases are manufac-
due to the various possible conRgurations of the tured by reaction of microparticulate silica (having
monosaccharides. Oligo- and polysaccharides are an average particle diameter of 3 or 5 m) with
composed of different combinations of them, organosilanes (for example, 3-aminopropyltri-
forming linear or branched polymeric species, many methoxysilane), to form siloxane bonds. Improve-
of which may differ only in the position of at- ments in column efRciency can be achieved using
tachment and anomeric conRguration of the spherical particles of silica gel, which form more
glycosidic linkages. Moreover, complex carbohy- homogeneous beds than those obtained employing
drates can also be characterized by the presence of irregularly shaped particles.
various nonglycosyl substituents, which include For separations on aminoalkylsiloxane-bonded sil-
acetyl, pyruvyl, methyl and sulfate esters and ether, ica columns, acetonitrile}water mixtures are usually
amine and phosphate groups. Finally, methods of employed as the mobile phase. The proportion of
analysis for carbohydrates require special attention as acetonitrile ranges from 80 to 90% by volume for
they play an important role in many diverse research chromatography of sugars and other carbohydrates
and industrial domains such as biochemistry, clinical of low molecular weight. For example, a mobile
chemistry, biology, pharmacy, biotechnology and phase consisting of 85% acetonitrile in water is useful
food science. for the separation of monosaccharides. A low per-
Over the last 25 years, high performance liquid centage of water in the mobile phase is necessary to
chromatography (HPLC) has been the method of increase the interactions between sugars and the
choice for the determination of carbohydrate species, amino groups bonded to the stationary phase, im-
and as a result a large number of HPLC methods have proving selectivity and monosaccharide separation.
been developed to determine a wide variety of carbo- On the other hand, 80% acetonitrile in water allows
hydrate samples. good separation of di- and trisaccharides differ-
Alkyl or aminoalkyl bonded silica and polymeric ing sufRciently in structure, such as sucrose, mal-
phases, protonated or metal-loaded cation exchange tose, lactose and rafRnose. For chromatographic
resins in conjunction with refractive index (RI) or low analysis of oligosaccharides of degree of polymeriz-
wavelength UV detection, as well as high-perfor- ation (dp) above 3, mobile phases containing
mance anion exchange chromatography at high pH acetonitrile in lower proportions are required. Under
(HPAEC), coupled with pulsed amperometric detec- these conditions separations are currently carried out
tion (PAD) are the methods commonly used. These at room temperature and with Sow rates from 1 to
different approaches are considered in more de- 2 mL min\1. An example of the effective separ-
tail later. ation of oligosaccharides obtained by lowering the
acetonitrile content in the eluent is shown in Figure 1,
where about 30 distinct peaks of a partial hydrolysate
Polar Phases of (1P4)--D-glucan from amylose are separated us-
Polar silica-based packings, as well as unmodiRed ing a mixture of acetonitrile}water (57 : 43, v/v).
silica are suitable stationary phases for the separation Although acetonitrile is the most common organic
of sugars and other carbohydrates by hydrophilic component of the mobile phase, the use of ethyl
interaction chromatography (HILIC), a name pro- acetate, together with acetone or methanol have also
posed in 1990 by Alpert to indicate all chromato- been proposed.
III / CARBOHYDRATES / Liquid Chromatography 2225
and do not form Shiff’s bases with reducing

sugars. Polymer-based columns have also been pro-
posed, increasing application in HILIC of carbohy-
drates. Although diol-bonded silica, as well as vinyl-
pyridinium polymers have proven to be capable of
greater selectivity in carbohydrate separations,
aminopropylsiloxane-bonded columns are the most
frequently used in HILIC of carbohydrates. A list of
some HPLC columns commonly employed in the
HILIC of carbohydrates is given in Table 1.
Cation Exchangers
HPLC analyses of carbohydrates are most frequently
carried out on cation exchange stationary phases con-
Figure 1 Chromatogram of partial hydrolysate of (1P4)--D- sisting of sulfonated polymer-based materials, which
glucan (amylose). Chromatographic conditions: column, ERC- can be obtained in various degrees of cross-linking
NH-1171 (200;6 mm ID); eluent, acetonitrile}water (57 : 53); flow and various particle sizes.
rate, 1 mL min\1, detector, Shodex RI SE-32 at 1;10\5 RI units
Cation exchangers based on poly(styrene-divinyl-
full scale; temperature ambient. (Reproduced from Journal of
Chromatography (1985), 321: 151, with the permission of benzene) (PS-DVB) copolymers are the most common
Elsevier Science Publishers.) stationary phases used for carbohydrate separations.
Copolymer-based columns offer many advant-
Aminopropylsiloxane-bonded silica packings have ages for the analysis of carbohydrates and alditols.
been widely used in HILIC of sugars and other carbo- The PS-DVB copolymers exhibit excellent physical
hydrates. However, they have the drawback of lim- strength, pH stability over a wide range and are not
ited life, due to the formation of glycosylamines by easily subjected to degradation by oxidation, hy-
reaction of the amino groups bonded to the stationary drolysis or elevated temperature. Carbohydrate sep-
phase with reducing sugars (formation of Shiff’s arations on column packed with a cation exchanger
bases). The formation of glycosylamines results in PS-DVB resin can be affected by the degree of
both deactivation of the column and loss of the sugar cross-linking and type of counterion.
analytes with an undesirable effect on quantita- PS-DVB resins are relatively rigid gel-type media
tive analysis. An alternative to the use of these col- and separation can take place when the anayltes pen-
umns is the addition of a small amount (0.01}0.02%, etrate at least partially into the matrix. The lower the
v/v) of a diamino or polyamino compound to the cross-linking, the more open the structure and more
eluent used in HPLC on a silica column, which results permeable it is to higher molecular weight carbohy-
in modiRcation of the silica to give it characteristics of drates. For example, a 4% cross-linked resin is suit-
an amino propylsiloxane-bonded phase. This ap- able to resolve oligosaccharides of high molecular
proach can be accomplished with various amine weight, whereas smaller oligosaccharides can be well
modiRers, which must carry at least two amino resolved on an 8% cross-linked resin.
groups, one being required to bond the modiRer to As mentioned earlier, the selectivity of cation ex-
the silica gel via hydrogen bond formation, and one change columns depends to a large extent on the ionic
which has to be free to interact with the carbohydrate form of the packing material. Loading of sulfonated
analytes. Usually in situ modiRed silica columns resins with various cations produces substantial cha-
show a longer life than aminopropylsiloxane- nges in retention of neutral carbohydrates in several
bonded columns, because the amine adsorbed on the ways, such as the formation of coordination com-
silica surface is continuously regenerated by the plexes, ion}dipole interactions, ionic hydration and
amine added to the mobile phase. Various polyfunc- hydrogen bonding. When the resins are loaded with
tional amines have been proposed as modiRers, in- Ca2#, Ag#, Pb2# La3#, marked differences in
cluding ethylenediamine, 1,4-diaminobutane, tet- the retention and selectivity are observed and the
raethylenepentamine (TEPA) and piperazine. mechanism of retention involves the formation of
Another approach is to replace aminopropyl- coordination complexes between the carbohydrates
siloxane-bonded columns with a silica-based packing and the Rxed-metal ions. However, the separation of
carrying bonded amide, cyano, diol, polyol or oligosaccharides is predominantly governed by size-
cyclodextrins as alternative stationary phases. These exclusion mechanisms and oligomers elute in order of
phases have similar selectivity to amino-type phases decreasing molecular weight. The Rrst systematic
Table 1 Examples of silica-based columns and chromatographic conditions employed in carbohydrate separation
Column Functional length;ID Mobile phase Temperature Flow rate Detector Applications
(particle size, group (mm) (3C) (mL min\1)
m)
(Supplier)
Adsorbosil Amino 250;4.6 Acetonitrile} Ambient 1.0 ELSD Glucose sucrose lactose
(5 m) water melezitose maltotriose
(Alltech) (85 : 15)
Adsorbsphere Amino 250;4.6 Acetonitrile} Ambient 1.5 RI or ELSD Fructose glucose sucrose
(5 m) water maltose lactose
(Alltech) (85 : 15)
Alltima NH2 Amino 250;4.6 Acetonitrile} Ambient 1.5 ELSD Maltooligosaccharides
(5 m) water
(Alltech) (gradient)
Amino, Spheri-5 Amino 250;4.6 Acetonitrile} Ambient 1.0 RI Monodisaccharides
(5 m) water
(Brownlee) (75 : 25)
Nucleosil Amino 250;4.6 Acetonitrile} 1.0 RI Monooligosaccharides
carbohydrate water
(5 m) (79 : 21)
(Macherey-
Nagel)
Hypersil APS-2 Amino 100;3.0 Acetonitrile} Ambient 0.5 RI Monodisaccharides
(5 m) water
(ThermoQuest) (80 : 20)
Supelcosil LC-NH2 Amino 250;4.6 Acetonitrile} Ambient 1.0 RI Ribose arabinose
(5 m) water galactoste sucrose
(Sigma) (75 : 25) maltose isomaltose
melezitose raffinose
Ultrasil NH2 Amino 250;4.6 Acetonitrile} Ambient 1.0 RI Mono-, di-, trisaccharides
(10 m) water
(75 : 25)
(Beckman)
Lichrospher DIOL Diol 250;7.0 Acetonitrile} Ambient 1.0 ELSD Monosaccharides
(10 m) water dextrins
(Merck) (gradient)
Zorbax NH2 Amino 250;4.6 Acetonitrile} Ambient 1.0 RI Fructose glucose sucrose
(5 m) water maltose lactose
(Du Pont) (75 : 25)
Zorbax OH Diol 250;6.0 Acetonitrile} Ambient 1.0 ELSD Monosaccharides
(10 m) water
(Du Pont) (90 : 10)
Zorbax ODS C18 250;4.6 Water 60}703C 1.0 ELSD dp 2}5
(10 m)
(Du Pont)
Spheri-5 ODS C18 100;4.6 Water Ambient 0.5 RI glucose dp 2}5

(5 m)
(Brownlee)
study of the chromatography of sugars and alditols can be attributed to coordination between the metal
on a cation exchange resin was published in 1975 by ions and -OH groups of sugars and alditols and their
Goulding, who showed that the complexing cations retention can be correlated with the ability to form
Ca2#, Ag#, Sr2# Ba2# and La3# gave longer reten- relatively strong chelate complexes of the adjacent
tion times than the alkali-metal cations. This effect hydroxyl groups of sugars and alditols with the Rxed
Figure 2 Separation of maltooligosaccharides from dp 14 to dp

2. Chromatographic conditions: column, Rezex RSO-Oligosac- Figure 3 Separation of monosaccharides and alditols.
charide (200;10 mm ID); mobile phase, water; flow rate, 0.4 mL Chromatographic conditions: column, Rezex RPM-Monosacchar-
min-1, temperature, 753C; detector, RI (with permission of Phe- ide (300;7.0 mm ID); mobile phase, water; flow rate,
nomenex. Torrance, CA, USA). 0.6 mL min\1, temperature, 753C; detector, RI (with permission of
Phenomenex. Torrance, CA, USA).
cation of the resin. For example, stronger binding will

occur in sugars where the conRguration of the hy- resin column in the hydrogen form, the reducing
droxyl groups at C-1, C-2 and C-3 of the pyranose monosaccharide is eluted as a single peak (Fig-
ring have an axial}equatorial}axial sequence that can ure 5A), whereas with the same column in the cal-
form relatively strong tridentate complexes than the cium form, almost baseline separation of and
binding of sugars, which form complexes with only anomers is achieved, indicating that a ligand ex-
a pair of axial-equatorial hydroxyl groups. In sul- change mechanism is also involved (Figure 5B).
fonated cation-exchange resins carrying Li#, Na#, The most common approach to avoid the forma-
and H#, there is no evidence that these counterions tion of doublets is to increase column temperature. By
interact directly with carbohydrates, and their ef- heating the HPLC column in the calcium form to
fects can be interpreted simply by ionic hydration. 853C, the rate of interconversion between the and
Cation exchange resins carrying calcium ions are the anomers of glucose is accelerated, and it is eluted as
most commonly used for chromatography of sugars a single peak. Separations are usually performed with
allowing good separations of monomeric sugars (pen- a column temperature ranging from 80 to 853C,
toses, hexoses), alditols and di-, trisaccharides though lower temperatures (653C) may be used.
(Figure 2). Cation exchange resins carrying silver ions Higher temperatures give faster diffusion and
give better separation of maltooligosaccharides up to narrower peaks. Conversely, increasing temperature
a degree of polymerization of 13 glucose units (see is normally accompanied by a decrease in retention.
Figure 3). Organic modiRers such as acetonitrile may be added
Carbohydrate separations on cation exchange to the eluent up to a concentration of 30% (v/v)
resins loaded with various metal ions can be carried without causing any damage to the PS-DVB matrix.
out with deionized water as the mobile phase. How- For example, the addition of acetonitrile increases
ever, when chromatographic runs are performed at retention of the analytes and may improve resolution
room temperature, peaks are broadened or eluted as of sugar alcohols.
a doublet owing to the separation of the two Sulfonated resins in the protonated form are useful
anomeric forms of reducing carbohydrates. As is well for proRling not only monosaccharides and sugar
known, for each of the ring modiRcations of a free alcohols, but may be useful for analysis of a wide
sugar, two isomers ( and isomers or anomers) can range of organic acids, alone and in combination with
exist, because a new asymmetric centre is created carbohydrates, short-chain alcohols, aldehydes and
by ring closure at the reducing carbon atom (see ketones. Selectivity can be alterated by changing the
Figure 4). The two anomers can be selectively re- pH as well as the type of mineral acid (H2SO4 or
tained on cation exchange resins loaded with various HCl at mmol concentration), used as the eluent. Em-
metal ions. For example, when glucose is chromatog- ploying these columns, amino sugars may also be
raphed at room temperature on a sulfonated PS-DVB separated. Sugars of food interest and sugar alcohols,
Figure 4 Anomeric configurations of D-glucose.
including sorbitol and mannitol may be separated Anion Exchangers

using pure water as the mobile phase. An example is
shown is Figure 6, where a mixture of lactose, galac- Anion exchange chromatography is not a technique
tose, mannitol, sorbitol and xylitol are separated on commonly associated with analysing neutral carbo-
a sulfonated cation exchange resin in protonated hydrates. It has been shown that the hydroxyl groups
form. Furthermore, a separation of lactose, glucose in carbohydrates can form reversible anionic com-
and galactose in a lactose-hydrolysed milk sample is plexes with borate, which can be separated on an
shown in Figure 7. Both separations were performed anion exchange column. However, this technique is
by isocratic elution with water at room temperature. not currently applied to HPLC of mono- and
Under these conditions sugar alcohols, as well as oligosaccharides.
reducing sugars are eluted as single sharp peaks. Anion exchange chromatography is basic media
Tables 2 and 3 list some of the commercially avail- allows the separation of carbohydrates without addi-
able columns containing sulfonated styrene-divinyl- tion of any borate to the mobile phase. The pKa of
benzene resins in various ionic forms. neutral carbohydrates usually fall in the range of
12}14, and at high pH their hydroxyl groups are
either partially or completed ionized, enabling this
class of compounds to be separated as anions by high
performance anion exchange chromatography
(HPAEC). Moreover, alkaline conditions allow the
detection of carbohydrates by PAD at a gold elec-
trode. The compatibility of PAD with gradient elu-
tion coupled with the high selectivity of speciRcally
tailored anion exchange stationary phases, allows
mixtures of underivatized simple sugars, oligo- and
polysaccharides to be separated with good resolution
in a single run, and quantiRed down to picomole
levels. Under these conditions, separations cannot be
performed using classical silica-based columns due to
their poor stability at high pH levels. The columns
commercially available are packed with polymer-
based anion exchangers produced by proprietary
manufacturing processes. These packings include
Figure 5 Elution of D-glucose at room temperature. Chromato- electrostatically latex-coated polymer-based anion
graphic conditions of chromatogram (A) column, PL Hi-Plex exchangers and macroporous polymer-based resins
H (hydrogen form); dimensions, 300;7.7 mm; mobile phase, functionalized with quaternary ammonium groups,
water; flow rate, 0.6 mL min\1; temperature, 253C; detection,
which have unique selectivity for sugar alcohols.
refractive index detection. Chromatographic conditions of
chromatogram (B) column, PL Hi-Plex Ca (calcium form); other Monosaccharide separations are usually per-
conditions as chromatogram A. Both columns from Polymer formed under isocratic conditions and the elution
Laboratories (Shropshire, UK). order is inversely correlated with pKa . Alditols are
Figure 6 Separation of (1) lactose, (2) galactose, (3) mannitol, (4) sorbitol and (5) xylitol on a sulfonated cation-exchange resin in the
hydrogen form. Chromatographic conditions: column, PL Hi-Plex H (8 m, 300;7.7 mm ID Polymer Laboratories Ltd, Shropshire, UK);
mobile phase, water; flow rate, 0.4 mL min\1 at room temperature; detector, refractive index detection.
less retained than the corresponding reducing sugars linkage position of a single sugar. Correlation of
as can be readily understood on the basis of acidity. retention times with the structure of different
Since the anomeric hydroxyl group is the most acidic oligosaccharides may suggest that at least two factors
among all the OH groups, replacing it with an ordi- are involved in determining the superior resolution of
nary OH group (which is less acidic) would naturally oligosaccharides by HPAEC: the relative acidity of
result in less retention. For oligosaccharides, reten- the hydroxyl groups and the accessibility of oxy-
tion increases dramatically with increasing molecular anions of the saccharides to the functional groups of
mass. HPAEC, followed by PAD detection, has been the stationary phase.
found to resolve positional isomers of neutral Addition of a modiRer, such as sodium acetate,
oligosaccharides, which are deRned as having the allows the elution of oligosaccharides in more reason-
same number, type, sequence, and anomeric conRg- able times. Acetate anion is used as a strong ‘pusher’
urations as monosaccharides, but differing in the because it offers rapid equilibration, allowing a
Figure 7 Separation of (1) lactose, (2) glucose and (3) galactose in an UHT lactose-hydrolysed milk. Chromatographic conditions as
in Figure 6.
Table 2 Examples of polymeric-based ion-exchanger columns and chromatographic conditions employed in carbohydrate separ-
ation
Column Matrix Length;ID Mobile Temperature Flow rate Detector Applications

(Supplier) (PS-DVB)* (mm) phase (3C) (mL min)\1
Cation Sulfonated 300;6.5 Water 90 0.5 RI Maltotriose, sucrose,

exchange (calcium form) glucose, galactose,
(Alltech) fructose, mannitol,
arabitol, sorbitol
Aminex HPX-87N (PS-DVB) 300;7.8 Water 85 0.6 RI Monosaccharides
(Bio-Rad) sulfonated
(sodium form)
Aminex HPX-87K Sulfonated 300;7.8 Water 85 0.6 RI Monosaccharides
(Bio-Rad) (potassium
form)
Aminex HPX-87C Sulfonated 300;7.8 Water 85 0.6 RI Monosaccharides, sugar
(Bio-Rad) (calcium form) alcohols
Aminex HPX-87P Sulfonated (lead 300;7.8 Water 85 0.6 RI Cellobiose, glucose,
(Bio-Rad) form) xylose, galactose
arabinose mannose
Aminex HPX-42A Sulfonated (silver 300;7.8 Water 85 0.4 RI Oligosaccharides to
(Bio-Rad) form) dp 11
Aminex HPX-87H Sulfonated 300;7.8 0.005 N 50!65 0.6 RI Glucose, ribose, fructose,
(Bio-Rad) (hydrogen form) sulfuric organic acids
acid
SupelcogelC-611 Ion-exchange 300;7.8 10\4 N NaOH 60 0.5 RI Sucrose, glucose,
(Supelco) resin containing fructose, mannitol,
two different sorbitol, ribose
cations
Nucleogel sugar Sulfonated 300;6.5 Water 80 0.4 RI Sucrose, maltose,
Pb (calcium form) glucose, xylose,
(Macherey-Nagel) galactose, arabinose,
mannose
Nucleogel ION Sulfonated 300;7.8 Sulfuric acid 30 0.6 RI Sucrose, glucose, citric
300 OA (hydrogen form) 0.01 N acid, fructose, tartaric
(Macherey-Nagel) acid, malic acid,
glycerol acetic acid,
lactic acid, methanol,
ethanol
Nucleogel Sugar PS-DVB 300;7.8 Water 85 0.4 RI Mono oligosaccharides
Na sulfonated (lead
(Macherey-Nagel) form)
IOA-1000 Sulfonated 300;7.8 0.005 65 0.3!0.4 RI Maltotriose, maltose,
(Alltech) (hydrogen form) sulfuric acid glucose, fructose,
organic acids
*PS-DVB"polystyrene}divinylbenzene copolymer.
faster elution of strongly retained components with- elution, alditols and mono- and disaccharides can be
out compromizing selectivity and does not interfere separated in a single run (Figure 9).
with amperometric detection. Figure 8 is the As has been shown by several authors in compara-
chromatogram of a sample of starch hydrolysate con- tive studies, the use of HPAEC-PAD for quantiRca-
taining over 20 linear homologous maltooligosac- tion of neutral monosaccharides, aminosaccharides,
charides. The separation was obtained by increasing glucuronic acids, linear and branched oligosacchar-
sodium acetate content in the mobile phase by linear ides in complex matrices, is usually superior
gradient elution. On the other hand, under isocratic to HPLC-RI and HPLC-UV in terms of sample
Table 3 Columns with a selectivity comparable to Aminex columns
Aminex Sarasep Nucleogel Polyspher Sugar-Pak Rezex Supelcogel

column (length;ID mm) (length;ID mm) (length;ID mm) (length;ID mm) (length;ID mm) (length;ID mm)
(Interchim) (Macherey-Nagel) (Merck) (waters) (Phenomenex) (Sigma)
HPX-87N } Sugar Na CH NA } RNM }

(300;7.8) (300;7.8) (300;7.8)
HPX-87K } } } } RKP K
(300;7.8) (300;7.8)
HPX-87C Car-Ca Sugar Ca CH CA SC-1011 RCM Ca

(300;7.8) (300;7.8) (300;6.5) SC 1821 (300;8.0) (300;7.8) (300;7.8)
HPX-87P Car-Pb Sugar Pb CH PB SP-0810 RPM Pb

(300;7.8) (300;7.8) (300;7.8) (300;7.8) (300;7.8) (300;7.8)
HPX-42A } } } } RSO Ag
(300;7.8) (300;7.8)
HPX-87H Car-H ION 300 OA OA HY } RHM C-610

WA-1 (300;7.8) (300;6.5) (300;7.8) H
(300;7.8) (300;7.8)
preparation and sensitivity. In the last Rve years many nonpolar stationary phases having relatively strong
significant developments in carbohydrate analysis by hydrophobic character. The choice of an RP column
HPAEC-PAD have been reported. with pure water as the eluent is an easy method for
HPLC analysis of carbohydrates. However, the mech-
anism governing retention of carbohydrates is not
Alkylsilica Packings reversed-phase partition but a form of hydrophobic
Octadecylsiloxane bonded silica materials, usually chromatography, involving van der Waals interac-
designed for reversed-phase (RP) chromatography are tions between carbohydrates and the alkyl chains of
Figure 8 Separation of linear maltooligosaccharide oligomers of up to 20 residues. Chromatographic conditions: column, CarboPac
PA 100 (250;4.0 mm ID) connected to a CarboPac PA 100 guard column (50;4.0 mm ID), all from Dionex, Sunnyvale, CA, USA;
eluent A, 50 mM sodium acetate in 120 mM sodium hydroxide; eluent B, 300 mM sodium acetate in 120 mM sodium hydroxide; flow
rate, 1.0 mL min\1 at room temperature; detector, PED in pulsed amperometric mode. Maltooligosaccharides were eluted by a linear
gradient from eluent A to eluent B in 35 minutes. Numbers over peaks indicate dp values.
Figure 9 Separation of glycerol (1) xylitol, (2) sorbitol, (3) mannitol, (4) lactitol, (5) maltitol, (6) glucose, (7) fructose, (8) and sucrose,
(9) by isocratic elution. Chromatographic conditions: column, CarboPac PA 10 (250;4.0 mm ID) connected to a CarboPac PA 10
guard column (50;4.0 mm ID), all from Dionex, Sunnyvale, CA, USA; mobile phase, 65 mM sodium hydroxide; flow rate, 1 mL min\1
at room temperature; detection, PED in pulsed amperometric mode.
the bonded phase. The Rrst HPLC separation of tadecyl groups into hydrophilic porous polymers hav-
carbohydrates employing octadecylsiloxane-bonded ing hydroxyl groups on the surface. Koizumi et al.
silica columns was carried out in 1981 using a Dex- have described the use of an AsahipakTM ODP-50
tropak2+ (Waters) column speciRcally developed for column packed with C18-bonded vinyl alcohol
the separation of underivatized oligosaccharides. copolymer gel for the separation of glucooligomers,
However, separation of monosaccharides was poor. which were selectively eluted as single peaks using
Furthermore, a drawback of the reversed-phase col- sodium hydroxide at pH 11 as the mobile phase.
umn in carbohydrate separation is the undesired res- Under these conditions, the rate of interconversion
olution between and anomers of the saccharides, between the and anomers of saccharides was
which complicates the chromatograms. accelerated satisfactorily, avoiding the elution of
Different approaches have been proposed to anomeric doublets. The same polymeric column has
avoid unwanted double peaks. The most widely used been used to separate cyclodextrins and branched
approach is to reduce the oligosaccharide samples cyclodextrins. Columns packed with C18-bonded
with sodium borohydride. In this way for each reduc- phases have also found application in ion-pair
ing saccharide the corresponding alditol is eluted as chromatography with tetrabutylammonium in a buf-
a single peak. As reported earlier, another approach fered medium, for analysis of oligosaccharides con-
to avoid the formation of doublets is to increase taining an acid or sulfate group.
column temperature, which increases the rate of in-
terconversion between the and anomers. Chee-
tham and Teng reported that modiRcation of a Dex-
Detection
tropak C18 column by adding nonionic Triton X-100 One shortcoming in the HPLC of carbohydrates is the
detergent or tetramethylurea to the eluent, avoids the often inadequate sensitivity and selectivity in the de-
elution of anomeric doublets. The elution of broad tection process, particularly when they are at a much
peaks or doublets due to resolution of anomers can be lower concentration than other compounds present in
precluded by the addition of triethylamine to the complex matrices. The analysis of neutral carbohy-
eluent. However, under these conditions, some risk of drates is difRcult due to their absence of a chro-
silica degradation is likely. mophore. RI detection is the most popular detection
This drawback may be overcome by the use of technique for carbohydrates in HPLC but a serious
reversed-phase packings obtained by introducing oc- drawback is that it can only be used to monitor
Figure 10 Detection of sugars in fruit juices by ELSD. Chromatographic conditions: column, LiChrosorb Diol (5 m, 250;4.6 mm ID,
from Merck); mobile phase, dichloromethane}methanol (82 : 18 v/v); flow rate, 1 mL min\1 at room temperature; detector, ELSD;
evaporator, 503C; air inlet pressure, 20 psi. Chromatogram A: fructose (tr"5.95), glucose (tr"7.85), sucrose (tr"11.39) in a blood
orange juice. Chromatogram B: fructose (tr"5.93), glucose (tr"7.82), sucrose (tr"11.30) in a grapefruit juice.
isocratic separations. As a substitute for RI, the evap- suitable for solutes which are characterized by
orative light-scattering detector (ELSD) can be used a lower volatility than the eluent used for the
with gradient elution, thus performing better chromatographic separation. A wide variety of sol-
chromatographic separations. In addition, ELSD pro- vents may be used as eluents; highly volatile mobile
vides rapid stabilization before operation and im- phases are preferred either as pure solvents or as
proves baseline stability and better sensitivity. The a combination of two or more. Nevertheless, carbo-
direct detection of analytes is achieved through three hydrates are more soluble in water than in organic
consecutive stages, which are independent of each solvents and therefore water is frequently a constitu-
other. The Rrst stage consists in the nebulization of ent of the mobile phase.
the column efSuent with air or nitrogen to pro- By coupling the HPLC system with an ELSD de-
duce a Rnely divided spray, which passes through tector, carbohydrate separations on bare silica, as
a heated chamber to vaporize the eluent. During the well as diol and polyol, bonded phases can be per-
second stage, solutes less volatile than the eluent formed by isocratic or gradient elution from mixtures
create a particle stream that intersects a collimated of dichloromethane}methanol. With these organic
light beam (mono- or polychromatic, depending on eluents, separation of mono-, di- and oligosacchar-
the apparatus design), producing light scattering. The ides can be achieved by gradient elution without
scattered light is then detected by a photomultiplier baseline drift and with a good sensitivity; an example
(or photodiode), where the output is proportional to of the analysis of sugars using a diol silica-based
the amount of solute present over a wide concentra- column coupled with ELSD detection is reported in
tion range. Description of the technical performance Figure 10. Fructose, glucose and sucrose are separ-
and conRguration of the current commercial ELSD ated from a blood orange juice and a grapefruit juice
detectors for both HPLC and supercritical Suid using a mixture of dichloromethane}methanol
chromatography (SFC), have recently been reported (82 : 18 v/v) as the mobile phase. The detection limit
by Henry, Dreux et al. and Lafosse et al. Owing to the is 20 ng for all analysed sugars. Furthermore, the
nature of light scattering detection, ELSD is especially stationary phase employed avoided Schiff’s base
formation with reducing sugars and periodic regen- detection in HPLC will always be preferred whenever
eration of the column was not needed. The high possible.
chemical stability of the column is demonstrated In HPLC, polar aliphatic compounds containing
comparing the reproducibility of the retention times electroactive groups can be directly detected by pul-
reported in chromatogram A with those reported in sed electrochemical detection (PED), a generic term
chromatogram B, which was performed 100 injec- introduced by LaCourse to indicate all detection
tions apart. The use of ELSD detection is less favour- strategies based on the application of multi-step
able when employing metal-bearing cation exchange potential-time waveforms at noble metal electrodes
columns, where carbohydrates are eluted with plain for electrochemical detection.
water at high temperature (80}953C). Since these Amperometric detection is used to measure the
cationic resin}water systems cause an increase in de- current or charge generated by the oxidation or re-
tector noise at the temperature of the separation. On duction of analytes at the surface of a working elec-
the other hand, Dreux and Lafosse demonstrated the trode in a Sow cell. For several classes of molecules,
usefulness of ELSD detection in oligosaccharide and which include carbohydrates and aliphatic alcohols
polysaccharide analysis obtained on octadecyl- and amines, amperometric detection can only be per-
siloxane-bonded silica columns with gradient elution formed if a repeating sequence of three potentials is
at increasing methanol content in water. applied for speciRc times to the working electrode.
Only few carbohydrates exhibit signiRcant absorb- Amperometric detection under the control of a three-
ance in the low UV, such as acidic disaccharides step potential-time waveform is known as PAD.
released from glycosaminoglycans by enzymatic di- This detection mechanism was Rrst applied in 1981
gestion, which carry an unsaturated uronic acid resi- for the detection of aliphatic alcohols at Pt electrodes
due at the nonreducing end that enables their UV and then proposed for the direct determination of
detection at an absorbance maximum of 232 nm. carbohydrates when in 1983 Rocklin and Pohl
Neutral carbohydrates can be analysed by UV detec- coupled the Rrst commercial detector dedicated to
tion at wavelengths lower than 200 nm but under PAD with HPAEC.
these conditions they are subject to interference In pulsed amperometric detection mode, the work-
from other compounds with similar absorption ing electrode is held at an analytical potential E1 for
properties. Furthermore, in this UV region acetonit- a time t1. At this potential the CHOH groups of
rile also absorbs strongly and therefore the UV de- carbohydrates are oxidized. The current is measured
tector is unsuitable for use in HILIC of underivatized at the end of the E1 pulse to allow the charging
carbohydrates, where separations are usually per- current to decay, resulting in more accurate measure-
formed employing mobile phases rich in this solvent. ment of the electroactive species. If only a single
This drawback can be overcome by either pre- or potential is applied, the detector response would
postcolumn derivatization. The most common chem- steadily decrease as the electrode surface became
ical derivatization methods allow the conversion of fouled. The electrode surface is oxidatively cleaned
carbohydrates into derivatives that can be detected by a positive potential E2 (E2'E1) for a time t2. The
with an online ultraviolet, visible, or Suorimetric de- voltage on the electrode is then reversed to a strongly
tector. Besides an improvement in the detection prop- reducing potential E3 for a time t3 to convert the gold
erties, the chemical reaction can also enhance the oxide layer at the surface of the electrode back to
selectivity of the total analytical method. For native gold, thus renewing the surface.
example, pre-column derivatization of oligosacchar- Selection of the potentials for detection, cleaning
ides with a hydrophobic chromophore is the most and conditioning are based on current (i)-potential
common approach to enhance the resolution and (E) response from cyclic voltammetry (CV) recorded
detection of oligosaccharides into their components for triangular waveforms (E!i). Extensive studies
by RP-HPLC. on the optimization of waveforms for pulsed am-
A wide variety of methods have been developed for perometric detection of carbohydrates have been
pre- and postcolumn derivatization of carbohydrates. published recently by LaCourse. Carbohydrates can
Comparative advantages and disadvantages of these be detected by PAD at Au electrodes in highly alka-
two approaches have been reviewed by Giese and line media. For separations which are not carried out
Honda (see Further Reading). under adequately alkaline conditions, as in the analy-
Although prechromatographic and postcolumn sis of monosaccharides, the detection requirement of
chemical derivatization methods allow the conver- a strong alkaline medium can be satisRed by post-
sion of carbohydrates into either photometrically or column addition of NaOH (generally in the range
Suorimetrically active adducts that can be detected 300}500 mM) by means of a pump, a hydraulic
with high sensitivity, the simplicity of sensitive direct device, or membrane reactor.
III / CARBOHYDRATES / Thin-Layer (Planar) Chromatography 2235
Conclusions: Summary of Main Trends tectors; Instrumentation; Mechanisms: Reversed Phases.

III / Polysaccharides: Liquid Chromatography.
A wide variety of HPLC methods are available to
analyse carbohydrates from various complex ma- Further Reading
trices. Aminopropylsiloxane-bonded silica of small
particle diameter (3 m), amine-bonded vinyl alcohol Alpert AJ (1990) Hydrophilic-interaction chromatography
copolymer packings, diol- and cyclodextrin-bonded for the separation of peptides, nucleic acids and other
silica, as well as macroporous cross-linked vinyl- polar compounds. Journal of Chromatography 499: 177.
Cheetham NHW and Teng G (1984) Some applications of
pyridinium polymers, are some of the columns of
reversed-phase high-performance liquid chromatogra-
choice for high selectivity in the HILIC of carbohy- phy to oligosaccharide separation. Journal of Chrom-
drates. Furthermore, a large number of columns may atography 336: 161.
be selected to separate carbohydrates on cation ex- Churms SC (1990) Recent developments in the chromato-
change resins carrying different counterions. Em- graphic analysis of carbohydrates. Journal of Chrom-
ploying these columns the chromatographic separ- atography 500: 555.
ation is governed by a combination of size exclusion Dreux M and Lafosse M (1995) Evaporative light scatter-
and ligand exchange mechanisms. In oligosaccharide ing detection of carbohydrates in HPLC. In: El Rassi
separations, size exclusion is the primary mechanism Z (ed.) Carbohydrate Analysis } High Performance
which involves the binding of hydroxyl groups of the Liquid Chromatography and Capillary Electrophoresis.
sugars with the Rxed counterion of the resin, thus Amsterdam: Elsevier.
El Rassi Z (1995) Carbohydrate Analysis } High Perfor-
they elute last.
mance Liquid Chromatography and Capillary Elec-
On the other hand, RP-HPLC with plain water as trophoresis. Amsterdam: Elsevier.
the eluent may be useful for the separation of various Giese R and Honda S (1996) Chromatographic and elec-
underivatized oligosaccharides. The mechanism gov- trophoretic analyses of carbohydrates (Special Issue).
erning this HPLC method produces separations that Journal of Chromatography 720.
complement those given by cation exchange columns Goulding RW (1975) Liquid chromatography of sugars and
and therefore the two techniques should be used in related polyhydric alcohols on cation exchangers } The
conjunction in isolation and analysis of carbohy- effect of cation variation. Journal of Chromatogra-
drates. phy 103: 229.
Column instability and low detection sensitivity are Herbreteau B (1992) Review and state of sugar analysis by
the major disadvantages using these HPLC methods. high performance liquid chromatography. Analusis 20:
355.
However, column and detection technologies to ana-
Honda S (1984) High-performance liquid chromatography
lyse carbohydrates are in continuous development. of mono- and oligosaccharides. Analytical Biochemistry
Mono- and oligosaccharides are detected without 140: 1.
any derivatization by PAD, which is sensitive and Hurst WJ (1997) Seminars in Food Analysis } HPAEC-
selective for detection of these analytes. The use of PAD in Food and Beverage Analysis. London: Chapman
HPAEC is now widely accepted as a powerful analyti- and Hall.
cal tool for carbohydrate analysis. However, HPAEC LaCourse WR (1993) Pulsed electrochemical detection at
methods require dedicated columns and instrumenta- noble metal electrodes in high performance liquid
tion. chromatography. Analusis 21: 181.
Lee YC (1990) High-performance anion-exchange chrom-
See also: II / Chromatography: Liquid: Detectors: Evap- atography for carbohydrate analysis. Analytical Bio-
orative Light Scattering; Detectors: Refractive Index De- chemistry 189: 151.
J. F. Robyt, Laboratory of Carbohydrate Chemistry differ in the number of carbon atoms, the conRgura-
and Enzymology, Iowa State University, Ames, IA, USA tion of the chiral centres and the size of the molecules,
Copyright ^ 2000 Academic Press that is to say whether they are di-, tri- and oligo-
saccharides. If two carbohydrates have any one of the
three characteristics in common, they can be difRcult
Introduction to separate. There are some carbohydrates that differ
Carbohydrates are some of the more difRcult com- by having a special functional group, for example
pounds to separate from each other. They primarily a carboxyl group (giving uronic, onic and aric acids),
2236 III / CARBOHYDRATES / Thin-Layer (Planar) Chromatography
an amino or acetyl-amino group (giving amino and gel, a stable and inert substance. The silica gel
N-acetyl amino sugars), or a phosphoric acid ester thin layer is usually uniformly 250 m thick over the
group (giving carbohydrate phosphates). Neverthe- TLC plate. The resolution of carbohydrates is
less, because carbohydrates play such an important achieved by the selection of a speciRc solvent system.
role in living systems and comprise a large number of Many solvent systems have been empirically de-
important naturally occurring products, methods for veloped. The particular solvent used depends on
separating them were some of the earliest chromato- whether the carbohydrates are monosaccharides, dis-
graphic procedures developed. accharides, trisaccharides or oligosaccharides, or
Early chromatographic methods in the 1950s for whether they contain a charged group such as
separating carbohydrates used paper as the solid sup- a uronic acid, an amino sugar or a carbohydrate
port. In the early 1960s Thoma introduced thin-layer phosphate.
chromatography (TLC) for separating carbohydrates. A relatively simple solvent that has found valuable
The use of TLC, however, was not particularly easy use in the separation of a number of mono-, di- and
and convenient at that time, as each laboratory had to trisaccharides is acetonitrile}water (85 : 15, v/v).
prepare its own thin-layer plates. This led to a wide This solvent is frequently the Rrst choice when separ-
variation in the quality of the plates and in the results ating unknown carbohydrates in a sample. Solvents
that were obtained, even within a single laboratory. for the separation of more complex mixtures can then
The results were difRcult to reproduce from day to be developed using other organic compounds in com-
day and particularly from laboratory to laboratory. It bination with the acetonitrile}water system. Many
was not until reproducible thin-layer plates were solvent systems contain three components: water,
commercially developed and produced at a reason- a water-soluble component and a water-insoluble
able price, primarily by Whatman and Merck in the component. The three components are combined in
late 1970s, that TLC became a viable and valuable various proportions so as to give a homogeneous
procedure. solution. For maximum resolution, the technique of
TLC has now become the principal method of multiple ascents is frequently employed. In this tech-
choice for the separation of carbohydrates. High per- nique, the solvent is allowed to ascend to the top of
formance liquid chromatography (HPLC) has dis- the TLC plate, whereupon the plate is removed from
placed TLC in many laboratories, although for carbo- the developing tank and the solvent is completely
hydrates and other materials, TLC is much preferred evaporated from the plate. The plate is then put back
over HPLC. When compared with HPLC, TLC is into the TLC tank for a second, third and/or
much less expensive, easier to perform, uses simple fourth ascent. Figure 1 illustrates the separation of
equipment, uses much less sample, and the detection relatively complex mixtures of carbohydrates using
of the carbohydrates directly on the plate is more this technique with acetonitrile}water (85 : 15, v/v)
sensitive. TLC is also more reproducibile. With solvent.
HPLC, once a sample has been added to a column, A very sensitive detection system has recently been
the column is irreversibly modiRed. Because of the developed in which most carbohydrates can be detec-
high cost of the HPLC columns, many hundreds of ted in the 50}2000 ng range, directly on the TLC
samples must be separated and analysed in series on plate. The method uses a detecting reagent, consisting
the same column to make the procedure economically of 0.3% (w/v) of N-(1-naphthyl)ethylenediamine and
feasible; whereas sample placed on to a TLC plate is 5% (v/v) concentrated sulfuric acid in methanol. The
separated in its own little column or lane, which is TLC plate is removed from the developing tank, dried
not affected by other compounds that have previously and rapidly dipped into the reagent. The plate is dried
been adsorbed and separated on the support material. and placed in an oven at 1203C for 10 min. Carbohy-
Further, approximately 18 samples can be simulta- drates give blue-black spots on a white background.
neously separated on a single 20;20 cm TLC Most carbohydrates can be detected by using this
plate, which can include standards. A simple com- reagent, with the exception of 2-amino-2-deoxy
parison of the migration positions with the migration sugars, 2-N-acetylamino-2-deoxy sugars and sugar
positions of the standards can be used for qualitative alcohols. This is primarily due to the requirement of
identiRcation. forming furfurals from the carbohydrates by dehy-
dration with the sulfuric acid. An older method used
TLC Support Material, Solvents and sulfuric acid charring, in which a 25% (v/v) solution
of sulfuric acid in methanol was sprayed on to the
Detection Methods TLC plate, dried and then heated at 1203C for
The principal support material found most useful for 10 min. This is a much less sensitive method,
valuable in the separation of carbohydrates is silica with detection in the 10}100 g range. Further, the
magnitude more sensitive and relatively easy to use

quantitatively.
TLC Separation of Homologous Series

of Oligosaccharides
The separation of oligosaccharides is important in
a wide variety of applications, such as analysis of
carbohydrates in foods, analysis of oligosaccharides
from glycoproteins and glycolipids, the analysis of
enzymatic synthesis of oligosaccharides and the
analysis of the hydrolysis of polysaccharides. Many
of these analyses involve the separation of a homolo-
gous series of oligosaccharides containing one or pos-
sibly two kinds of glycosidic linkages and as many as
20}30 monosaccharide residues of the same type in
the oligosaccharides. The separation of a homologous
series of saccharides, containing 2}30 monosacchar-
ides residues, linked -1P2, -1P3, -1P4,
and -1P6, has been reported, using silica-gel plates
developed with butanol-1}ethanol}water. The devel-
opment time was however, several hours for one
ascent.
Much faster systems, using Whatman K5 plates or
Merck silica gel 60 plates and multiple ascents with
acetonitrile}water systems, have been developed for
the separation of maltodextrins (-1P4 linked
glucosaccharides) and isomaltodextrins (-1P6
linked glucosaccharides). The separation of malto-
dextrins was achieved using acetonitrile}ethyl acet-
ate}propanol-1}water in volume proportions of
85 : 20 : 50 : x. The amount of water (x) can be var-
ied from 50 to 70 parts by volume, depending on the
desired number of saccharides to be separated. The
largest saccharides can have 20 D-glucose residues.
Likewise, using up to four ascents can increase the
number of saccharides separated. Figure 2 illustrates
the separation of maltodextrins using one to four
Figure 1 Separation of various mono-, di- and trisaccharides
ascents of 85 : 20 : 50 : 60 volume proportions at
on Whatman K5 TLC plates (18.5 cm solvent path length for each 20}223C.
ascent), using four ascents of acetonitrile}water in volume pro- Isomaltodextrins migrate more slowly than an
portions of 85 : 15 at 20}223C. The carbohydrates were detected equivalent maltodextrin mixture, primarily because
by using the N-(1-naphthyl)ethylenediamine}sulfuric acid}meth- of differences caused by the -1P6 glycosidic link-
anol dipping reagent.
age vs the -1P4 glycosidic linkage. To obtain an
equivalent separation for the isomaltodextrins, the
water content (x) of the solvent system has to be
spraying technique is much less environmentally increased to 90 or 100 volume proportions. Figure 3
friendly than a dipping technique. illustrates the separation of isomaltodextrins using
The separation of radiolabelled carbohydrates per- one to four ascents of acetonitrile}ethyl acetate}pro-
mits several sensitive techniques to be used to detect panol-1}water in the volume proportions of
and quantify the carbohydrates directly on the 85 : 20 : 50 : 90 at 20}223C.
TLC plate, such as a TLC radioactive scanner The power of TLC separations of saccharides, us-
or a PhosphoImager. Both can be used to detect ing these solvents, is illustrated in Figure 4 in which
and quantitate 14C, 35S, or 32P radioactive isotopes, complex saccharides are separated from mixtures
although the PhosphoImager is several orders of produced by the action of two different enzymes
contain six, seven and eight D-glucose residues, re-

spectively having one -1P6 glycosidic linkage and
Rve, six and seven -1P4 glycosidic linkages. These
saccharides migrate in between the maltodextrins, for
example, BB6 migrates behind G6 and faster than
G7, B7 migrates behind G7 and faster than G8,
and so forth. Saccharides marked BB9, BB10 and
BB11 contain nine, 10 and 11 D-glucose residues,
respectively, with each having two -1P6 glycosidic
linkages and seven, eight and nine -1P4 glycosidic
linkages. Saccharides containing one D-galactose
residue will usually migrate more slowly than an
equivalent saccharide containing all D-glucose
Figure 2 Separation of D-glucose and maltodextrins on What-

man K5 TLC plates (18.5 cm solvent path length for each ascent),
using 1}4 ascents of acetonitrile}ethyl acetate}propanol-1}water
in volume proportions of 85 : 20 : 50 : 60 at 20}223C. The num-
bers at the top of each chromatogram indicate the number of
solvent ascents. The carbohydrates were detected using the
N-(1-naphthyl)ethylenediamine}sulfuric acid}methanol dipping
reagent. Reproduced with permission from Robyt and Mukerjea
(1994).
acting on polysaccharides. The saccharides were sep-

arated by using four ascents of acetonitrile}ethyl acet-
ate}propanol-1}water in volume proportions of
85 : 20 : 50 : 50 at 20}223C on a Whatman K-5 plate.
Lane 1 on the plate is a set of maltodextrin standards,
containing D-glucose, maltose, maltotriose and so Figure 3 Separation of D-glucose and isomaltodextrins on
forth down to maltodextrins with 14 D-glucose resi- Whatman K5 TLC plates (18.5 cm solvent path length for
dues; lane 2 shows an equivalent set of maltodextrin each ascent), using 1}4 ascents of acetonitrile}ethyl acetate}pro-
saccharides that were obtained by the action of panol-1}water in volume proportions of 85 : 20 : 50 : 90. The
numbers at the top of each chromatogram indicate the number of
isoamylase acting on shellRsh glycogen; lane 3 shows
solvent ascents. The carbohydrates were detected using N-(1-
the separation of products that resulted from the naphthyl)ethylenediamine}sulfuric acid}methanol dipping re-
action of porcine pancreatic -amylase acting on po- agent. Reproduced with permission from Robyt and Mukerjea
tato amylopectin. Saccharides marked B6, B7 and B8 (1994).
residues. Compare the migration of lactose with cel-

lobiose in Figure 1.
To illustrate the tremendous power that TLC has in
separating carbohydrate oligosaccharides with very
similar structures, Figure 5A shows the chromato-
graphic separation of saccharides formed by the ac-
tion of Leuconostoc mesenteroides B-512FM dex-
transucrase transfer of D-glucose from sucrose to
malto-oligosaccharide acceptors having two to eight
D-glucose residues. Two different products were for-
med by the transfer of D-glucose to the reducing end
of the acceptor saccharide to form an -1P6 linkage
(product-1a) and transfer of D-glucose to the non-
reducing end of the acceptor saccharide to form -
1P6 linkage (product-1b). Products 2a, 2b, 3a, 3b
and so forth are products resulting from the transfer
to C6 of reducing-end glucose residue and to C6 of the
nonreducing-end glucose residue, and so forth of the
Rrst, second and third acceptor products. These com-
plex products were separated using four ascents of
ethyl acetate}ethanol}water (2 : 2 : 1, v/v/v) on
a Whatman K5 TLC plate.
Figure 5B shows the separation of an even more
complex mixture of products, containing the same
number of D-glucose residues, but differing from each
other by the location of the transferred D-glucose
residue and the types of linkages, either -1P6 or
-1P3. D-Glucose was transferred from sucrose to
maltotriose by the action of three enzymes (F, I and
S). Enzyme I gave four initial products, 1a}1d; 1a had
D-glucose transferred to C6 of the reducing-end glu-
cose residue of maltotriose; 1b had D-glucose trans-
ferred to C6 of the nonreducing-end glucose residue of
maltotriose; 1c had D-glucose transferred to C3 of the
reducing-end glucose residue of maltotriose; and 1d
had D-glucose transferred to C3 of the nonreducing-
end glucose residue of maltotriose. Enzymes F and
S only gave two initial products, 1a and 1b. All three
enzymes gave higher products: 2a, 2b, 3a, and so forth.
Separation and Detection of

Sugar Alcohols on TLC
One of the more difRcult separations of carbohy-
drates is that of separating aldoses from their corre-
sponding alditols (sugar alcohols). Alditols can be
Figure 4 Separation of maltodextrins and maltodextrins, con-
separated from their aldoses on TLC by using two
taining one and two -1P6 linkages. Lane 1, glucose and mal-
todextrin standards; lane 2, maltodextrins resulting from the ac- ascents of acetonitrile}ethyl acetate}propanol-1}
tion of isoamylase reaction with shellfish glycogen; lane 3, - water in volume proportions of 85 : 20 : 20 : 15. As
amylase action on amylopectin. Whatman K5 TLC plate (18.5 cm previously mentioned, alditols are not readily detec-
solvent path length for each ascent) was irrigated four times with ted by the sulfuric acid-based reagents described
acetonitrile}ethyl acetate}propanol-1}water in volume propor-
above. The best detection system is an alkaline silver
tions of 85 : 20 : 50 : 50 at 20}223C. The carbohydrates were
detected using the N-(1-naphthyl)ethylenediamine}sulfuric acid} nitrate}sodium thiosulfate dipping system that will
methanol dipping reagent. Reproduced with permission from detect alditols in the range of 500 ng to 1 g. The
Robyt and Mukerjea (1994). dry TLC plate is dipped into a silver nitrate}acetone
Figure 5 (A) TLC separation of a series of products produced by the action of Leuconostoc mesenteroides B-512FM dextransucrase
catalysed transfer of D-glucose from sucrose to maltodextrin acceptors, maltose (G2) to maltooctaose (G8). With G3 and higher, the
enzyme gave two products of the same size, transferring D-glucose to the C6 position of the reducing-end glucose residue of the
acceptor (product 1a) and the transfer of D-glucose to C6 position of the nonreducing-end glucose residue of the acceptor (product 1b).
Products 2a, 2b, 3a, 3b and so forth represent transfer to C6 position of the reducing-end and the nonreducing-end glucose residues of
the first, second, and so forth acceptor products. The saccharides were separated on Whatman K5 TLC plates using four ascents of
ethyl acetate}ethanol}water in volume proportions of 2 : 2 : 1 (18.5 cm solvent path length for each ascent). The compounds were
labelled with 14C and detected by autoradiography. Reproduced with permission from Fu and Robyt (1990). (B) Separation of the
products from the transfer of D-glucose from sucrose to maltotriose by the action of three enzymes, Leuconostoc mesenteroides
B-512FM dextransucrase (F), Streptococcus mutans mutansucrase (I), and S. mutans dextransucrase (S). Whatman K5 TLC plate was
irrigated with four ascents of ethyl acetate}ethanol}water, in volume proportions of 2 : 2 : 1. The compounds were labelled with 14C and
detected by autoradiography. Reproduced with permission from Fu and Robyt (1991).
solution for 5 min. The reagent is prepared by adding a white background. Figure 6 illustrates the separ-
1 mL of saturated aqueous silver nitrate to 200 mL of ation of seven different alditols from their corre-
acetone, followed by the dropwise addition of distil- sponding aldoses.
led water until the silver nitrate is dissolved. The plate
is dried and then dipped into an alkaline methanol Separation of Methylated
solution containing 0.4% NaOH (w/v) for 30 min.
Brown to black spots appear for the alditols. The
Monosaccharides by TLC
plate is then rapidly dipped into a 1.5 mol L\1 solu- The separation of O-methylated monosaccharides
tion of Na2S2O3 containing 0.08 mol L\1 Na2SO3 and is especially important. O-Methylated mono-
0.25 mol L\1 NaHSO3, and then the plate is washed saccharides are produced from the methylation of
for 1 min in running water, giving black spots on oligosaccharides and polysaccharides followed by
Figure 6 Separation of alditols (sugar alcohols) from their corresponding aldoses on Whatman K5 TLC plates (18.5 cm solvent path
length for each ascent) using two ascents of acetonitrile}ethyl acetate}propanol-1}water in volume proportions of 85 : 20 : 20 : 15 at
20}223C. The carbohydrates were detected by using the alkaline silver nitrate dipping reagents. Reproduced with permission from Han
and Robyt (1998).
acid hydrolysis. The number of methyl groups and

their location on the resulting monosaccharides indi-
cate the positions of the linkages to the monosacchar-
ides in the oligosaccharides or polysaccharides and
whether or not the saccharides are branched. For
example, if a gluco-polysaccharide contains 1P4
glycosidic linkages with 1P6 branch linkages, there
would be three kinds of methylated monosaccharides:
2,3,6-tri-O-methyl-D-glucose (major), 2,3,4,6-tetra-O-
methyl-D-glucose (minor from the nonreducing-end D-
glucose residues) and 2,3-di-O-methyl-D-glucose (mi-
nor from the branched D-glucose residues). The use of
TLC to separate and analyse these methylated mono-
saccharides has greatly simpliRed methylation analysis
of oligo- and polysaccharides. Further, because of the
sensitivity of detecting the methylated monosacchar-
ides (lower limits of 25}50 ng range) directly on the
TLC plate, micro amounts of saccharide (10 g) can be
methylated and analysed. Figure 7 illustrates the
chromatography of methylated D-glucoses on What-
man K6 TLC plates, using two ascents of acetonitrile}
chloroform}methanol in volume proportions of 3 : 9 : 1
at 20}223C. The methylated carbohydrates are detected
by dipping the plate into the N-(1-naphthyl) ethylene-
diamine}sulfuric acid}methanol reagent described
Figure 7 Separation of O-methylated D-glucoses. Lanes 1 and
above, followed by heating for 10 min at 1203C.
3 are standards; lane 2 is the analysis of a complex, highly
branched D-glucan. Whatman K6 plates were irrigated (18.5 cm
solvent path length for each ascent) at 20}223C, using two as- Quantitative Determination
cents of acetonitrile}chloroform}methanol in volume proportions of Carbohydrates
of 3 : 9 : 1. The carbohydrates were detected by using the N-(1-
naphthyl)ethylenediamine}sulfuric acid}methanol dipping re- The easiest method of detecting and quantitating
agent. Reproduced with permission from Mukerjea et al. (1996). carbohydrates on a TLC plate is to use 14C-labelled
Figure 8 Densitometric scan of TLC number 3 shown in Figure 2, using a Bio-Rad imaging densitometer. The relative weight per
cents for each of the carbohydrates are shown at the top of each peak. The first peak is D-glucose, followed by maltodextrins: maltose,
maltotriose, maltotetraose, and so forth, down to a maltodextrin with 14 D-glucose residues.
carbohydrates. The radioactivity can be detected and sucrose, and so forth. However, for many purposes,
quantitated directly on the plate using a phos- the relative amounts of the individual saccharides in
phoimager. The use of radioactively labelled carbo- a mixture are sufRcient. In this instance, the indi-
hydrates, however, is limited to carbohydrates that vidual saccharides that are separated in a mixture can
can be obtained in radioactive form and therefore be scanned, the densities of the compounds summed
limited in the kinds of experiments that can be per- and the percentage of the individual compounds com-
formed. The more usual case is the chromatography puted by dividing the individual densities by the sum
of nonlabelled carbohydrates. The detection of carbo- of the densities. Figure 8 shows a density scan of
hydrates on a TLC plate, using a sulfuric acid-based chromatogram no. 3 in Figure 2.
reagent, such as N-(1-naphthyl)ethylenediamine re-
agent described above, can be quantiRed using Preparative Separation of
a scanning densitometer. Absolute, quantitative
amounts can be determined using various amounts of
Carbohydrates Using TLC
the monosaccharide involved in the saccharides as By using the same solvent systems as is used in ana-
standards, for example, D-glucose, D-mannose, or lytical TLC separations, but substituting TLC plates
that have a much thicker silica gel layer

(1000}2000 m), preparative separations of carbo-
hydrates can be obtained. Approximately 25}50 L
of a 20}30 mg sample of a mixture of carbohydrates
is streaked along a sample application line, 15 mm
above the bottom of a 20;20 cm TLC plate. The
plate is developed in the usual way. After develop-
ment, 15 mm strips are cut off from each end of the
chromatogram and dipped into the detecting reagent.
The strips are lined up with the plate and the loca-
tions of the desired compounds are marked. The silica
gel area containing the compound is scraped from the
plate into approximately 2 mL of water. The suspen-
sion is mixed and the silica gel removed by Rltration
or centrifugation. The aqueous extract can then be
concentrated to dryness by evaporation, and the com-
pound redissolved if necessary.
Points to Consider to Obtain Good

Separations
1. A relatively small (1}25 L) measured amount of
sample should be carefully added on to the TLC
plate, keeping the size of the spot as small as pos-
sible. The use of a microlitre syringe pipette, such as
a Hamilton syringe, is desirable. When many sam-
ples are to be added to the TLC plate, a multi-
spotting instrument can conveniently be used to
make very small, uniform spots with a minimum of
effort. The multi-spotter is also convenient when
a relatively large volume, for example 10}25 L, of
sample is to be added.
2. Be sure that the developing tank is tightly sealed
during development (vacuum grease can be used
with glass-to-glass surfaces). The solvent and va- Figure 9 Glass chamber and holder used to dip the TLC plates
pour should be in equilibrium before beginning into the detecting reagent.
the development.
3. Allow the solvent to ascend to the top of the
TLC plate to give the maximum path length
for the solvent. The old idea that the TLC plate a hair-dryer is a fast and convenient method for
should be removed from the solvent before the drying plates.
solvent has reached the top or just as the solvent 6. In developing the chromatogram, rapidly and
reaches the top of the plate is not valid, if the carefully dip the irrigated plate into the detecting
chamber is tightly sealed. In fact, it is best to allow reagent to obtain a uniform amount of reagent
5}10 min extra, after the solvent reaches the top of over the entire plate. The use of a narrow 1.5;
the TLC plate, before removing the plate. This 22;22 cm glass chamber is convenient for dip-
ensures that the plate has been completely saturated ping. A stand and a lid have to be fabricated from
with the solvent over the entire surface of the plate. polypropylene or similar material to hold the
4. Use no more than 15 mm margin from the bottom chamber (Figure 9).
of the plate to the point of sample application, so
as to give the maximum path length for the sol-
vent. Also, be sure only to use pencil to mark the Conclusions and Future Prospects
TLC plates.
5. In multiple ascending chromatography, dry the TLC has become a powerful analytical tool for separ-
plates thoroughly between each ascent. The use of ating compounds with very similar chemical and
physical properties, such as carbohydrates. The devel- Chromatography-Mass Spectrometry; Liquid Chromato-
opment of reliable and reproducible commercial TLC graphy.
plates has placed it at the forefront of separation
methods that can be used for both qualitative and Further Reading
quantitative determinations. Further, the use of rela-
Block RJ, Durrum EL and Zweig G (1958) A Manual of
tively simple and inexpensive materials permits
Paper Chromatography and Paper Electrophoresis, 2nd
a wide range of laboratories, from the high-school edn, pp. 170}214. New York: Academic Press.
teaching laboratory to quality control and research Churms SC (1982) In: Zweig G and Sherma J (eds). Hand-
laboratories, to analyse and compare compounds eas- book of Chromatography: Carbohydrates, vol. I, pp.
ily, quickly and with high sensitivity. The use of 137}153. Boca Raton, FL: CRC Press.
relatively inert materials, such as silica gel on glass, Fu D and Robyt JF (1990) Acceptor reactions of maltodex-
allows chemical reactions to be conducted directly on trins with Leuconostoc mesenteroides B-512FM dex-
the plate where the compounds have been separated. transucrase. Archives of Biochemistry and Biophysics
These reactions provide qualitative identiRcation of 321: 379}387.
the compounds by speciRc colour production and the Fu D and Robyt JF (1991) Maltodextrin acceptor reactions
quantitative determination of the compounds by the of Streptococcus mutans 6715 gluocosyltransferases.
Carbohydrate Research 217: 201}211.
use of scanning densitometry. The latter is parti-
Gauch R, Leuenberger U and Baumgartner E (1979)
cularly important for carbohydrates as they do not Quantitative determination of mono-, di-, and trisac-
lend themselves to the usual physical detection and charides by thin-layer chromatography. Journal of
quantifying methods, such as ultraviolet or visible Chromatography 174: 195}210.
absorbance, light scattering or refractive index. The Han NS and Robyt JF (1998) Separation and detection of
development of a wide variety of solvents has also sugars and alditols on thin-layer chromatograms.
permitted the separation of carbohydrates that differ Carbohydrate Research 313: 135}137.
from each other in subtle aspects of conRguration of Huber CN, Scobell HD, Tai H and Fisher EE (1968)
chiral centres, differences in the kinds of glycosidic Thin-layer chromatography of the malto-oligo and
linkages present, and differences in the size of the megalosaccharides with mixed support and multiple
compounds. irrigations. Analytical Chemistry 40: 207}209.
Jeanes A, Wise CS and Dimler RJ (1951) Improved tech-
Future developments in TLC of carbohydrates may
niques in paper chromatography of carbohydrates. Ana-
come slowly as much progress has been made in the lytical Chemistry 23: 415}418.
last 50 years. But, as in most areas of endeavour, need Koizumi K, Utamura T and Okada Y (1985) Analysis of
is the necessity of invention. Developments will un- homogeneous D-gluco-oligosaccharides and -polysac-
doubtedly be made in new solvents that will separate charides (degree of polymerization up to about 35) by
new kinds of mixtures of carbohydrates obtained high performance liquid chromatography and thin-layer
from natural products and from chemical and enzy- chromatography. Journal of Chromatography 321:
matic modiRcation reactions. There will probably be 145}157.
detection methods developed that will give greater Mukerjea R, Kim D and Robyt JF (1996) SimpliRed and
sensitivity in the submicro range of carbohydrates on improved methylation analysis of saccharides, using
the plate. The latter will then also provide for the a modiRed procedure and thin-layer chromatography.
Carbohydrate Research 292: 11}20.
development of new physical instrumentation using
Robyt JF and Mukerjea R (1994) Separation and quantitat-
lasers for scanning density, reSectance and Suorescence ive determination of nanogram quantities of maltodex-
in quantifying carbohydrates directly on the plate. trins and isomaltodextrins by thin-layer chromatogra-
phy. Carbohydrate Research 251: 187}202.
See also: II / Chromatography: Thin-Layer (Planar): Su D and Robyt JF (1993) Control of the synthesis of
Layers; Preparative Thin-Layer (Planar) Chromatography; dextran and acceptor-products by Leuconostoc mesen-
Radioactivity Detection; Spray Reagents. III / Carbohydr- teroides B-512FM dextransucrase. Carbohydrate Re-
ates: Electrophoresis; Gas Chromatography and Gas search 248: 339}348.
III / CAROTENOID PIGMENTS: SUPERCRITICAL FLUID CHROMATOGRAPHY 2245
CAROTENOID PIGMENTS: SUPERCRITICAL

FLUID CHROMATOGRAPHY
V. Sewram, Programme on Mycotoxins and ness. Feeding studies have also shown that cis isomers
Experimental Carcinogenesis (PROMEC), of -carotene have lower pro-vitamin A activities
Medical Research Council, Tygerberg, South Africa when compared to the all-trans form. -Carotene,
M. W. Raynor, Advanced Technology Center, besides having the highest pro-vitamin A activity of
Matheson Gas Products, Longmont, CO, USA the carotenoids, has also been reported to have anti-
Copyright ^ 2000 Academic Press neoplastic activity, perhaps due to their antioxidant
and free radical quenching activity, not only at the
stage of onset of the disease but also on existing
Introduction tumours. Hence the importance of the constituents, in
Carotenoids are one of the main classes of natural terms not only of colour but also nutrition, explains
pigments and are found in a large number of fruits why attempts have been made to characterize and
and vegetables (oranges, tomatoes, carrots) and in determine these pigments which occur together in
leaves where their presence is masked by chlorophylls mixtures that can be resolved only with the mildest
until autumn. They are also found in some animal and most selective analytical methods. While the
products (eggs, milk) and seafoods. Typically, caro- existence of carotenes and xanthophylls was known
tenoids contain eight isoprenoid units bonded such before 1906, it was not until Tswett developed col-
that the units are reversed at the centre of the mole- umn chromatography that much was known about
cule. With this arrangement, many carotenoids are the carotenoids. Subsequently, chromatographic
symmetrical in nature. The pigmentation of these methods have improved drastically from countercur-
tetraterpenes is a result of the chromophore created rent distribution, gas chromatography (GC), thin-
by the series of conjugated double bonds. The struc- layer chromatography, gravity column chromatogra-
tural formula of all carotenoids is derived from that phy through to high performance liquid chromatog-
of lycopene, starting with different structural modiR- raphy (HPLC).
cations which include hydrogenation, cyclization, Recently, a number of studies have been performed
oxidation or dehydrogenation. The oxygenated deriv- involving the separation of carotenoids by either
atives are known as xanthophylls and bare either the supercritical (SFC) or subcritical Suid chromatogra-
epoxy, carbonyl or hydroxy ester functions on their phy, showing that SFC with carbon dioxide as the
extremity or terminal ring, while the hydrocarbon mobile phase can provide an alternative to the tradi-
carotenoids are referred to as carotenes. The struc- tional HPLC methods used. A brief history of the SFC
tures of some basic carotenoids are given in Figure 1. of carotenoids up to the present is summarized in
Carotenoids have been used for many years in the Table 1. Supercritical Suids have lower viscosities
food industry as colouring material. These com- than liquids and thus solute diffusion coefRcients are
pounds also have many biochemical roles, including higher than in conventional solvents. Furthermore,
the important functions of light harvesting and the low critical temperature of some Suids enables
photoprotection in photosynthesis. In humans, the many heat-sensitive compounds to be separated with-
primary use of carotenoids, whether taken as caro- out degradation. Manipulation of various parameters
tenoid-containing food or as dietary supplements, is such as temperature, pressure, mobile phase, modi-
the prevention or correction of vitamin A deRciency. Rers and stationary-phase type makes complex separ-
However individual carotenoids are capable of form- ations possible. These aspects are considered in more
ing different cis/trans geometrical isomers and this in detail in subsequent sections.
turn is known to affect their biochemistry in certain
situations. In fresh plant tissue, all the double bonds Effect of Mobile Phase and
have the all-trans conRguration (most stable struc-
tural form); however, isomerization to the cis conRg-
Modi\ers on Carotene Separations
uration results in a loss of nutritional value. Hence Supercritical CO2 has been the most commonly used
the determination of these isomers is necessary mobile phase in SFC. It is nontoxic and also has a low
for the quality control of fresh foodstuffs in order to critical temperature (313C), which enables the separ-
assess the pro-vitamin A activity of foods and for the ation of thermally labile compounds such as caro-
evaluation of the effects on food-processing on fresh- tenoids at low column temperatures. Furthermore, it
2246 III / CAROTENOID PIGMENTS: SUPERCRITICAL FLUID CHROMATOGRAPHY
Figure 1 Chemical structures of some basic carotenoids.

Table 1 Advances in SFC of carotenoid pigments a mixture of tetrahydrofuran}methanol in CO2

whereas, for more polar modiRers, the line curves
Date Development slightly downwards due to increased solvent polarity,
1968 Separation of - and -carotene by packed-column SFC which indicates that solute}solvent interactions are
(Giddings et al.) important. On the other hand, an exponential curve
1983 Separation of lycopene and - and -carotene (Gere) is observed for the nonpolar modiRers, suggesting
1989 Separation of geometrical isomers of - and -carotene that the solubility of the carotenes is enhanced by an
by open tubular SFC (Schmitz et al.)
amount independent of the concentration of the
1991 Understanding of the effect of temperature, pressure,
modifier and stationary-phase type on carotene separ- modiRer.
ations (Lesellier et al., Aubert et al.) The selectivity between the trans and cis isomers is
1994 Investigation of spectral shifts of carotenoids in super- unaffected by the modiRers, whereas the selectivity
critical CO2 (Hui et al.) between the - and -all-trans compounds diminishes
with increasing percentage of modiRer. With
a CO2}methanol}acetonitrile mixture, the selectivity
is compatible with a wide range of HPLC and GC between trans and cis isomers falls as the proportion
detectors. Nitrous oxide exhibits a polarity similar to of acetonitrile is increased, for both the - and
CO2 but has only received limited use because of its -carotenes. This phenomenon has been used to ad-
oxidizing properties. Nevertheless, both of these mo- vantage in optimizing the separation of the compo-
bile phases have been investigated with -carotene as nents of a carrot extract, as shown in Figure 2. With
a probe molecule. Interestingly, the retentive proper- methanol}carbon dioxide, the peaks were not well
ties of the polar amino stationary phase for -caro- resolved; however, replacing some of the methanol
tene are largely affected by the CO2 and N2O mobile with acetonitrile allowed on additional component in
phases, in comparison to the nonpolar octa- the extract to be resolved and eluted before all-trans
decylsilyl stationary phase, which is not appreciably -carotene.
affected by these mobile phases.
In cases where increased solubilizing power is re- Effect of Stationary Phase Type on
quired to elute components of interest, modiRers have Retention and Selectivity of
been successfully added to the supercritical Suid. The
resulting effect is reduced solute retention, improve-
Carotenoids
ment of chromatographic efRciency and, in some Both open tubular and packed columns have been
cases, altered elution order. One of the advantages of employed in the separation of carotenoids. Several
supercritical CO2 is that it is chemically compatible -carotene isomers have been separated on an
and miscible with a large number of modiRers. The SB-Cyanopropyl-25 (25% cyanopropyl}75% methyl-
effects of a range of modiRers } methanol, acetonit- polysiloxane) stationary phase employing supercriti-
rile, tetrahydrofuran, dichloromethane and trich- cal CO2 with 1% ethanol as the mobile phase. Separ-
lorotriSuoroethane } on the SFC of carotenes have ation of the isomers of -carotene however requires
been studied using binary and ternary mixtures with the use of a SB-cyanopropyl-50 (50% cyanopro-
CO2 containing 3}20% (v/v) of the modiRers. Addi- pyl}50% methyl-polysiloxane) stationary phase. A
tion of each of these modiRers produces a concentra- mixture of -carotene, echinenone, canthaxanthin,
tion-dependent decrease in the capacity factors (k) of astacene and fucoxanthin has also been separated on
the carotenes. The effectiveness of the modiRers has open tubular columns with cross-linked poly(cyanop-
not been directly correlated with their densities nor ropyl)methylsiloxane and Carbowax stationary phases.
with their polarities, and this suggests that speciRc However, the more polar carotenoids did not elute.
interactions between the solutes and the modiRers are One of the problems with immobilized polar sta-
important. tionary phases in open tubular columns is their insta-
Minimal retention is obtained with the strongest bility towards higher concentrations of polar modi-
modiRer, dichloromethane, which is closest to the Rers. Consequently, both polar and nonpolar packed
dipole}dipole interaction of Snyder’s triangle. Fur- columns have been investigated for SFC of caro-
thermore, increasing the dichloromethane concentra- tenoids. The retention mechanism in normal-phase
tion in the CO2 results in an overall decrease in columns is based on the lattice-Suid model where the
retention of the carotenes and also a decrease in important property of the mobile phase in -carotene
selectivity, probably due to reduced interactions be- retention is absorptivity as well as solubilizing power.
tween the solutes and the stationary phase, as all the Elution of carotenoids from normal-phase alumina
carotenes have high afRnities for this solvent. A plot and silica particles can be compared to that of
of 1/k against modiRer concentration is linear for normal-phase LC; however, the materials are strong
The effect of the stationary phase can be evaluated

from overall variations in retention factors or from
variations in the extent of separation of different
pairs or groups of compounds, such as Rrst, - and
-carotenes (the isomers not being separated) and
second, cis and trans isomers of -carotene, or
cis/trans - and -carotenes.
The relationship between the retention factor in
SFC and carotene solubility in the mobile phase can
be given by the following equations (as long as the
same kind of Suid is used as a mobile phase):
(0ss!0st)
ln kR"ln #ln C0st# !ln S [1]
RT
S"asat
m Cm
0
[2]
where is the volume ratio of the stationary and

mobile phases, C0st is the standard (surface or volume)
concentration in the stationary phase, 0ss is the chem-
ical potential of the pure solid solute at standard
pressure, 0st is the chemical potential of the solute in
the stationary phase as it moves toward inRnite dilu-
tion and standard concentration and pressure, asat m is
the activity of the solute at saturation in the solvent
and C0m is the standard concentration of the solute in
the mobile phase. The Rrst, second and third terms on
the right-hand side of eqn [1] are constant for a par-
ticular column, solute and temperature. If the prop-
erty of the stationary phase is independent of the
Figure 2 Separation of a carrot extract by supercritical fluid nature and pressure of the mobile phase, the solute
chromatography using an Ultrabase UB 225 column capacity factor is a function of only the solute solubil-
(250;4.6 mm i.d.); mobile phase, CO2}methanol}acetonitrile
(85 : 0.75 : 14.25, v/v/v); pressure, 15 MPa; temperature, 223C;
ity in the mobile phase, regardless of the kind of
flow rate, 3.0 mL min\1; detection, 450 nm. Peak identification: 1, solvent.
or -all-trans-carotene; 2, all-trans--carotene; 3, cis--caro- The effect of the nature of the stationary phase on
tenes; 4, all-trans--carotene; 5, cis--carotene; 6, cis--caro- carotene retention should be more signiRcant in SFC
tene. Reproduced with permission from Aubert et al. (1991). compared with LC because the interaction between
a solute and a supercritical Suid is generally small. In
adsorbents and can prove too adsorptive for CO2 to fact, some researchers have reported that solute reten-
elute the more polar analytes. The reversed-phase tion in SFC is sensitive to the properties of the station-
packings modiRed with C8 and C18 exhibit dipole} ary phase. Where the stationary phase is prepared
dipole interactions from the surface silanols in addi- with a monofunctional alkylsilane, there is one-to-
tion to the desired dispersion characteristics and have one bonding between the reagent and the silanol
been successfully applied to carotenoid analysis. groups, giving a ‘brush-type’ structure. Di- and
The retention of -carotene on octadecylsilyl trifunctional silanes can bond to more than one
(ODS) columns is well correlated to its solubility in silanol group on the silica support, to give essentially
the mobile phase. Both the type and the percentage the same type of brush-type stationary phases as
carbon loading of the stationary phase inSuence the monofunctional silanes. However, they can also poly-
separation of the carotenes by SFC. Among factors merize in the presence of traces of water. Under
that determine the performance of a column, such as suitable conditions, a stationary phase can also be
the shape and size of the particles, the pore size, the prepared in which each alkylsilane that is bonded to
speciRc surface area and the percentage surface cover- the surface of the silica gives rise to an arbores-
age, the kind of function that is bonded to the support cent}polymeric structure that is not brush-like and
is particularly important. differs from column packings in which the support
Table 2 Effect of different stationary phases on the selectivity and resolution of carotene isomers
Column L V0 Porosity Type of tr - kR - Separation Selectivity Resolution Separation

(mm) (%) bonding carotene carotene of - and of trans/cis of trans/cis of - and -
(min) -carotenes -carotene -carotene carotene
isomers isomers trans/cis
isomers
Ultrabase UB 225 250 2.463 59 M 11.22 12.58 Yes '1 '1.5 Yes
Spheri-5 ODS-5A 250 2.456 59 P(A) 11.40 12.85 Yes '1 '1.5 Yes
LiChrospher 100
RP 18 250 2.150 68 D 8.35 10.57 Yes '1 '1.5 Yes
LiChrospher 100
PR 18e 250 2.187 70 D 9.60 12.05 Yes '1 '1.5 Yes
Nucleosil C18 250 2.866 69 P(A) 7.38 6.66 Yes '1 '1.5 Yes
ChromTech CT-Sil
C18 150 1.545 62 ! 4.89 8.36 Yes '1 '1.5 Yes
Superspher 100
RP18 250 2.106 67 D 8.16 8.02 Yes '1 '1.5 Yes
Spherisorb ODS-2 100 1.029 62 P(A) 3.75 9.73 Yes '1 '1.5 Yes
Supelcosil LC-PAH 150 1.768 71 P(A) 2.05 2.37 Yes '1 (1.5 No
Erbasil C18 150 1.656 66 P(A) 3.60 5.40 Yes '1 (1.5 No
Suplex pKb-100 150 1.623 65 ! 2.38 3.27 Yes '1 '1.5 No
Ultracarb 5-ODS 20 150 1.789 72 ! 10.32 16.20 Yes '1 '1.5 No
Zorbax ODS 250 2.678 64 M 4.98 4.50 Yes "1 (1.5 No
Ultrasphere ODS 250 2.520 61 M 2.48 1.86 No "1 ! No
Vydac 201 HS 150 1.760 70 M 5.43 8.13 No "1 ! No
Partisil 5 ODS-3 250 2.803 67 M 7.70 7.17 No "1 ! No
Bondapak C18 300 2.908 70 ! 6.29 5.42 No "1 ! No
Vydac 210 TP 150 1.680 67 P(A) 1.60 1.73 Yes '1 (1.5 No
Perkin-Elmer
HC-ODS/PAH 250 0.327 88 P(A) 1.25 2.05 Yes '1 (1.5 No
Vydac 218 Tp 250 2.898 70 P(A) 3.28 2.32 Yes '1 '1.5 No
Hypersil 15C18 150 2.635 63 M 7.01 6.90 No "1 ! No
Synchropak
SCD-100 150 1.874 75 M 1.68 1.59 Yes '1 (1.5 No
M, monofunctional C18 groups; D, difunctional C18 groups; P(A), arborescent}polymeric C18 groups; ! exact nature unknown.
itself is polymeric. Arborescent}polymeric C18- Effect of Temperature and

bonded stationary phases appear to be particularly Pressure on Carotenoid Separations
suitable for separating closely related compounds
that differ in the degree of planarity of their structures The solubility of a substance in a supercritical Suid is
and have successfully been applied to SFC separation the sum of two factors: the volatility of the substance
of cis and trans - and -carotenes. Table 2 lists the (which is a function of temperature) and the solvating
different columns that have been evaluated for their effect of the supercritical Suid (which is a function of
separating capabilities of the different classes of caro- Suid density). Hence, solubility is controlled experi-
tenes. Particular attention should be paid to the es- mentally by selecting appropriate temperatures and
timation of the retention (R) factor (k) for the pressures which are important for controlling reten-
analytes because this parameter characterizes the station in SFC. These parameters inSuence the solvating
tionary phase independently of factors such as size power, efRciency and selectivity. At constant pres-
and porosity of the column. It has been found that on sure, an increase in temperature decreases the density
arborescent}polymeric columns, the separation of - and consequently the solvating power. The temper-
and -carotenes is incomplete if the retention (R) ature also inSuences the diffusion coefRcients of the
factor for -carotene is below 6. One also needs to be solutes in the supercritical Suid. With increasing tem-
aware that, unlike a polar stationary phase, an ODS perature, diffusion coefRcients increase and higher
stationary phase is not appreciably affected by the chromatographic efRciency results. However, one
nature of the mobile phase and that -carotene, being needs to exercise caution in this regard as carotenoids
much larger than the alkyl chains grafted on the silica are thermally labile and will decompose at high tem-
surface, is less affected by the residual silanol groups peratures. With pure carbon dioxide, the capacity
on the silica surfaces. factor (kR) increases in proportion to the temperature
Figure 4 Dependence of k R values of carotenes on pressure

R
Figure 3 Dependence of k values of carotenes on temper- ( P ) at constant temperature (223C). Column, Spheri-5 ODS-5A;
ature at constant pressure (25 MPa). Column, Spheri-5 ODS-5A; mobile phase, CO2Imethanol (80 : 20 v/v); flow rate,
mobile phase, CO2-methanol (80 : 20 v/v); flow rate, 3.0 mL min\1; detection, 450 nm. Open squares, all-trans -caro-
3.0 mL min\1; detection, 450 nm. Squares, all-trans -carotene, tene; circles, cis -carotenes; triangles, all-trans -carotene; filled
circles, all-trans -carotene. Reproduced with permission from squares, cis -carotenes. Reproduced with permission from
Aubert et al. (1991). Aubert et al. (1991).
(at constant pressure) and can be explained by the where the Suid is very compressible, as the pressure
resulting decrease in the density of the mobile phase. drop along the column is associated with a density
On the other hand, as shown in Figure 3, when the gradient and reduced efRciency.
carotenes are eluted with carbon dioxide containing
12% methanol, kR decreases with increasing temper- Detectors in SFC with Reference to
ature between 22 and 553C, while the resolution
between the all-trans - and -carotenes decreases as
Carotenoid Analysis
the temperature is increased. Although the optimum SFC is compatible with a wide range of detection
temperature for this separation is between 22 and methods. The two basic types of detectors are the
253C, and the mobile phase is therefore subcritical, ionization detectors and optical detectors. Most com-
this is of little consequence because there is no discon- mercial open tubular SFC systems provide a Same
tinuity in the physical properties of the Suid at the ionization detector (FID) as standard and it is therefore
critical point. the most widely used detector for carotenoids. How-
At constant temperature, an increase in pressure ever, detection of carotenoids in supercritical Suids
produces an increase in density. Increasing the pres- following chromatographic separation has also been
sure increases the mobile-phase viscosity, thus de- achieved using mass spectrometry (MS) and UV/Vis
creasing the mass transfer term C and the diffusion detectors. SFC-MS provides detailed information on
coefRcient. In SFC, density (or pressure program- the molecular structure of eluted carotenoids and
ming) is the primary method for developing separ- greatly aids peak identiRcations. Due to the thermal
ation. It is analogous to temperature programming in instability of these compounds, ‘soft’ chemical ioniz-
GC, and eluent composition programming in LC. ation is recommended. Use of methane as the chemical
Increasing the pressure at constant temperature leads ionization reagent gas and a low source temperature of
to decreased capacity factors, which can be explained 100}1203C results in minimal fragmentation, even for
by the enhanced solubility of the solutes with increas- thermally labile carotenoids. For more routine analy-
ing density of the mobile phase. Figure 4 illustrates sis, online UV-Vis monitoring (especially using photo-
the dependence of kR values of carotenes on pressure diode array detection) is now the standard procedure
at constant temperature. In this work, the pressure for the analysis of carotenoids by HPLC and is the
was varied between 100 and 250 bar, at 223C and the preferred detection system for the SFC analysis of
capacity factors were observed to decrease less rap- carotenoids, with a number of high pressure, temper-
idly at pressures above 200 bar. This is probably due ature-regulated Sow cells being available. The effects
to the nonlinearity of the P}T curve of carbon diox- of pressure and temperature on the absorption spectra
ide, which at 223C becomes less compressible. In of carotenoids dissolved in supercritical carbon diox-
practice, it is preferable to avoid working at pressures ide are discussed in the following section.
Behaviour of Electronic Absorption Table 3 The maximum absorbance wavelength of carotenoids

Spectra in Supercritical CO2 in hexane and in supercritical CO2 (6000 psi at 353C)
When using UV-Vis detection for SFC of carotenoids, Carotenoid max (nm)
attention must be paid to spectral shifts that occur in
Hexane CO2
supercritical Suids, in comparison to those measured
in liquid solvents. This is particularly relevant as the All-trans--carotene 450.0 437.0
effect can inSuence qualitative and quantitative re- Lycopene 472.0 456.0
sults. The shape of the absorption spectra of all-trans- Astaxanthin 474.0 453.0
-carotene, 15-cis--carotene and the xanthophylls, Canthaxanthin 464.0 452.0
Zeaxanthin 448.0 436.0
zeaxanthin, cathaxanthin and astaxanthin, which are
nonacidic oxygen derivatives of the carotenes, are
similar in both supercritical CO2 and hexane. The
max of the Rve carotenoids however shifts to a shorter pressure of 3000 psi and over this range only
wavelength (hypsochromic shift) in supercritical a 2.0 nm shift is observed. Thus, changes to both
CO2. Figure 5 shows the absorption spectra of all- pressure and temperature can induce shifts in the
trans--carotene in supercritical CO2 and in hexane. (1Bu#) max of all-trans--carotene in supercritical
Table 3 lists the maximum absorbance wavelength of CO2, although the former is much more signiRcant. In
the selected carotenoids in supercritical CO2 and in addition to the main absorption peak in the visible
hexane. Due to the relatively low solubilities of the region, cis-isomers of carotenoids exhibit a peak in
xanthophylls in supercritical CO2 below 5000 psi the UV region, termed the cis-peak, which originates
pressure, comparison of the spectra has been carried from the 1Ag# transition. The electronic absorption
out at 6000 psi and 353C. While solvent-induced spectrum of 15-cis--carotene in supercritical CO2
shifts of the visible (1Bu#) spectra of carotenoids in shows a clear cis-peak in the region 250}350 nm.
a range of polar and nonpolar organic solvents com- Unlike the behaviour of the main max in the visible
monly used for HPLC are well established, changes in region, the position of the cis-peak maxima does
the supercritical CO2 density have also been observed not alter as a function of temperature and pressure,
to shift the max. Density is related to both pressure but remains Rxed, suggesting that the energies
and temperature, hence a change in either of the two of the 1Bu# and 1Ag# states of carotenoids may
variables is known to affect the absorption maxima of not respond in the same manner to alternations in
carotenoids. An increase in pressure increases the visible density.
max of all-trans--carotene. Over the pressure range of
1500}6000 psi (at a constant temperature of 353C), the
position of this max shifts 7.0 nm to longer wavelength Future Developments
(430.0 to 437.0 nm) in a nonlinear manner.
As SFC continues to evolve, it will probably be in-
The effect of variation in temperature on max has
creasingly applied to carotenoid analysis, particularly
also been assessed in the range 25}503C at a constant
in view of the thermal instability of these compounds.
Development of more selective stationary phases and
application of micro-packed columns for SFC are key
areas where improvements are likely. Micro-packed
columns with internal diameters of 200}400 m
offer increased efRciencies, shorter run times and in-
creased ease of interfacing with ionization detectors.
Further, mobile phases such as the Suoro-
carbons have shown potential in similar applications
and are therefore likely to be investigated for SFC of
carotenoids.
Further Reading
Aubert M-C, Lee CR, Krstulovic AM et al. (1991) Separ-
Figure 5 Absorption spectra of all-trans--carotene in super- ation of trans/cis- - and -carotenes by supercritical
critical CO2 (continuous line) and in hexane (dashed line). Super- Suid chromatography. (I) Effects of temperature, pres-
critical conditions: mobile-phase CO2, temperature 353C, pressure and organic modiRers on the retention of carotenes.
sure 3000 psi. Reproduced with permission from Hui et al. (1994). Journal of Chromatography 557: 47.
2252 III / CATALYST STUDIES: CHROMATOGRAPHY
Giddings JC, McLaren L and Myers MN (1968) Dense-gas chromatography and supercritical Suid chromatogra-
chromatography of nonvolatile substances of high mo- phy. Journal of Chromatography 633: 9.
lecular weight. Science 159: 197. O’Neil C and Schwartz SJ (1992) Chromatographic analy-
Hui B, Young AJ, Booth LA et al. (1994) Detection of sis of cis/trans carotenoid isomers. Journal of Chrom-
carotenoids on supercritical Suid chromatography atography 624: 235.
(SFC). A preliminary investigation on the spectral Shifts Sakaki K, Shinbo T and Kawamura M (1994) Retention
of carotenoids in supercritical carbon dioxide. behaviour of -carotene on polar and nonpolar station-
Chromatographia 39: 549. ary phases in supercritical Suid chromatography.
Lesellier E, Tchapla A, PeH chard M-R et al. (1991) Journal of Chromatographic Science 32: 172.
Separation of trans/cis - and -carotenes by super- Schmitz HH, Artz WE, Poor CL et al. (1989) High-per-
critical Suid chromatography. (II) Effect of the type formance liquid chromatography and capillary super-
of octadecyl-bonded stationary phase on retention and critical Suid chromatography separation of vegetable
selectivity of carotenes. Journal of Chromatography carotenoids and carotenoid isomers. Journal of Chrom-
557: 59. atography 479: 261.
Lesellier E, Tchapla A, Marty C and Lebert A (1993) Tee E-S and Lim C-L (1991) The analysis of carotenoids
Analysis of carotenoids by high-performance liquid and retinoids: a review. Food Chemistry 41: 147.
CATALYST STUDIES:
CHROMATOGRAPHY
B. Mile, Llandaff, Cardiff, Wales, UK scientists and engineers. Tables 1 and 2, respectively,
contain the important concepts and terms and the
Copyright ^ 2000 Academic Press major technical advances in catalysis.
Catalysis has a much longer history than GC } in-
deed life itself may originate from the catalysed con-
Introduction version of the primitive atmosphere into bioorganic
Catalysis is a branch of chemical kinetics of great molecules on the surfaces of clay minerals. The
industrial and commercial importance. Heterogen- ancients used enzymatic catalysts unknowingly in
eous catalysts are certain particulate solids of fermentation to make wine and vinegar. It is salutary
high surface area (1}300 m2 g\1) that increase the that these natural catalysts far outperform any of
rates of attaining equilibria. This is achieved by our synthetic catalysts at ambient pressures and
the temporary attachment of reactant molecules temperatures. Silver, regarded for at least three mil-
by moderate chemical bonds to active sites on lennia as having preservative and curative powers
their surfaces. Catalysts themselves are not consumed and long used for storing wines and wound treat-
during the process, although their activity is eventual- ments, has now re-emerged in the sterile surface coat-
ly lost by surface degradation. Over 90% of the ing, Amenitrop, whose active antibacterial and
world’s manufactured chemicals involve catalysis at antiviral ingredients are silver thiosulfate com-
one or more stages. The market is about C7;109 per plexes.
annum, with C200 return on each pound spent on In catalysis empirical technology and industrial in-
a catalyst. novation have almost always outpaced scientiRc
theory and understanding, but this is less so today;
exceptions include the Haber process for ammonia
Background to Heterogeneous production. Nevertheless, it has always been true that
Catalysis and Contributions from basic studies provide the intellectual framework and
accelerate technical developments, i.e. they ‘catalyse’
Gas Chromatography progress!
The invention of gas chromatography (GC) in 1952 The scientiRc study of catalysis started soon after
and its development since then have been widely the beginnings of chemistry and at the time when
reviewed. Before looking at the applications of GC phlogiston theory still held some sway. In 1835 Ber-
to catalysis, it is necessary to summarize the zelius coined the term ‘catalysis’ (which is Greek for
major stages in the concepts of catalysis and the often to loosen) after realizing that research Rndings in the
independent technological achievements of industrial Rrst decades of the nineteenth century demonstrated
III / CATALYST STUDIES: CHROMATOGRAPHY 2253
Table 1 Terms and characteristics used in catalysis
Term Characteristics
Heterogeneous catalysts Large surface area solids that accelerate the rate of attaining equilibirum; they are
themselves chemically unchanged by the reaction and therefore do not affect the final
equilibrium yields
Active sites Those unique sites on the surface where the catalytic event takes place. They are usually
the steps, edges and kinks of the flat unreactive terraces, and often represent only a few
percent of the total surface atoms. Interestingly, catalysis is chemistry at imperfections
and discontinuities.
Specific activity Rate of reaction per unit area of catalyst under standard reactant concentrations. Activity
is sometimes defined empirically by the temperature needed for a reaction to reach an
arbitrary rate of conversion.
Turnover frequency (TOF) Number of product molecules formed per second per active site (10\2 to 102 s\1 for
synthetic catalysts, compared with 103 to 107 for enzymes).
Structure sensitive/insensitive reactions Rate dependent on particle size and crystal face/rate invariant to size and face.
Selectivity The fraction of reacted molecules converted to a desired product
Durability How long the catalyst remains active
the accelerating effects of small amounts of materials pheres at 4503C. The process is a tour de force of
that themselves were chemically unchanged by the chemical engineering linked with outstanding science.
reaction. He is sometimes erroneously quoted as sug- The molecular understanding of catalysis really
gesting an almost occult catalytic force, whereas in begins in the twentieth century, as have the majority
fact he clearly stated the new ‘force’ is ‘a manifesta- of the catalysts and processes now in use. This is
tion of the electrochemical afRnities of matter’ especially true of the petroleum and petrochemical
Michael Faraday had a clear concept of catalysis in industry, which in the 1960s and 1970s had annual
1834 even before Berzelius’ presentation of the term growth rates of 15%. The outstanding development
to the Swedish Academy of Sciences. It is tempting to in the interwar years was the Houdry process for the
associate the discovery of catalysts with the alchem- catalytic cracking of heavy petroleum fractions into
ists vain search for the philosopher’s stone to trans- volatile gasoline components (C2}C13). The immedi-
mute base metals into gold; in some respects the ate postwar period saw developments such as: (1)
traces of ‘catalyst stone’, which transform ‘useless’ platforming catalysts for upgrading gasoline research
raw materials into useful products, are more valuable octane numbers (RONs) by isomerization of straight
than a stone for making ornamental gold. The appel- into branched alkanes and the dehydrogenation of
lation ‘black art’ still lingers, perhaps with some justi- cyclohexanes to aromatics; and (2) steam reforming
Rcation, since most industrial catalysts are still for- of methane, naphtha and fuel oil feedstocks into syn-
mulated by recipes that are incompletely understood thesis gas, a carbon monoxide/hydrogen mixture
and retain an aura of mystery. important for ammonia and methane production.
The burgeoning nineteenth century chemical indus- The recent development of vehicle exhaust catalytic
try quickly appreciated the utility and exploitability converters based on rhodium-promoted platinum}
of catalysts and the lead chamber and contact pro- alumina for converting carbon monoxide, nitrogen
cesses developed rapidly, as did the hydrogenated oxides and unburnt hydrocarbons into harmless
hardening of vegetable oils and the synthesis of in- products is a remarkable achievement for catalysis
digo. Perhaps the most important catalyst ever de- science, especially in view of the varying gas
veloped, because it averted world famine, was that compositions and temperatures that have to be
for the high pressure synthesis of ammonia by Fritz sustained.
Haber in 1909, and its commercialization by The realization by Langmuir that catalysis involved
Badische Aniline and Sodafabrik. The technical de- short-range chemical bonds, which dictated that reac-
velopment of the iron catalyst, which remains largely tions occurred in surface monolayers, and Taylor’s
unchanged, exempliRes the rewards of painstaking suggestion that catalysis occurred only at unique
research, a full appreciation of thermodynamics and active sites, together with the discovery of the import-
the courage to build large scale industrial plants oper- ance of atomic clusters with speciRc geometric ar-
ating at the then unprecedented pressure of 300 atmos- rangements by Balandin, were seminal to the growing
Table 2 Important stages and concepts in heterogeneous detailed understanding, as were the applications of
catalysis the absolute theory of reaction rates.
Date Event
Outline of Conventional GC in
The beginning! Life originates on the surfaces of clay minerals
Prehistory Enzymes for making wine and vinegar, sterile
Catalysis Research
silver Since many of the developments in catalysis listed in
1834/1835 Berzelius coins the term catalysis
Table 2 predate the invention of GC in 1952, its
Faraday appreciates the occurrence of
catalysis contributions have inevitably been since then. How-
1830}80 Industrial: H2SO4 by lead chamber, contact ever, all the important characteristics of catalysts
catalysts; synthetic dyes by mercury catalysis; deRned in Table 1 require analysis of complex gas
hardening of oils mixtures. Thus inevitably the use of this powerful
1900 Start of modern kinetics by Bodenstein and
analytical technique in catalyst research is now en-
Ostwald
1900}14 Gasification of coal; syn-gas manufacture demic and pervasive; so much so that it is now used
1909 Haber’s catalytic synthesis of ammonia almost routinely in catalyst laboratories and on cata-
1918 Langmuir’s monolayer and Rideal Ely lyst testing rigs. It is the technique sine qua non for
mechanisms multicomponent analysis, especially when allied to
1920sP Petroleum industry: continuous processing;
mass spectrometry and spectroscopic methods. It has
catalytic cracking; platforming catalysts
1923 Fischer}Tropsch hydrocarbons from syn-gas revolutionized the analysis of complex volatiles in the
1925 Taylor proposes catalysis occurs only at active same way that liquid and paper chromatography
sites have caused a sea change in the analysis of complex
1925}36 Houdry catalytic cracking process to increase involatile bioorganic molecules. It has replaced pre-
octane number of gasoline
vious tedious and time-consuming methods such as
1929 Balandin’s proposal of catalysis on multiplet sites
1938 Brunauer}Emmet}Teller (BET) adsorption fractional distillation, which have largely been aban-
isotherm doned since its advent (but see later for a discussion of
1939}45 Synthetic substitutes for natural products like the online infrared methods now in an advanced state
rubber of development). The complex nature of catalysis
1941 Use of oriented metal films by Beeck and
product streams is illustrated by the chromatograph
Schwab
1941 Fluidized-bed technology of a gasoline in Figure 1.
1950}1970s Explosive growth of the petrochemical industry The acceleration of reaction rates by catalysts
1952 Gas chromatography introduced arises because of the lowering of the energy barriers
1955 Zeigler Natta stereoregular polymerization through the formation of surface compounds. The
catalysts
energy released by the making of surface bonds facil-
1955 Sasol gasification of coal in South Africa
1956 Gas chromatographic reactors introduced itates the scission or rearrangement of bonds in the
1960 Steam reforming of methane and naphtha to substrate molecule, as depicted in Figure 2. The sur-
synthesize gas for ammonia and methanol face bonds must be of intermediate strength; if they
manufacture are too strong permanent bonding to the surface
1964 Oxychlorination to make vinyl chloride
occurs, and if they are too weak the perturbations are
monomer
1965 Microcatalytic chromatographic reactor too small. This gives rise to an optimum value of the
introduced enthalpy of formation of the surface intermediate and
1965 Sohio process for acrylonitrile and acrylic acid the so-called ‘volcano’ plots when catalyst activity is
1970P Surface science studies of surface}substrate plotted against bond enthalpy.
intermediates and surface atom arrangements
GC cannot give direct information about these
and restructuring
1976P Catalytic converters for exhausts emission crucial surface intermediates, although ingenious ex-
control periments using radiolabelling experiments and nor-
1985P1990s Control of and blending of refinery steams by GC mal kinetic analysis and numerical modelling can
Improvements in exhaust catalytic converters lead to information about their nature. Major ad-
Development of catalytic power generators
vances in surface spectroscopic techniques have oc-
Hygiene catalysts: sterile coatings, oven
cleaning curred since the 1960s allow the observation and
ICI Hydcat process for hypochlorite removal quantitation of minuscule amounts (10\3 mono-
from waste streams layers) of stable and metastable surface species.
Immobilized enzymes I pharmaceutical However, there has sometimes been a short-sighted
production
overemphasis of the results of these powerful and
Zeolites and aluminium phosphate (ALPO)
catalysts developed expensive techniques and a neglect of those from
cheaper methods of Rnal product analysis such as GC.
Figure 1 Chromatogram of a gasoline fraction (courtesy of Dr R. Malpas, Shell Research Ltd, Thornton Research and Technology,
Centre).
There is the danger that the moieties seen by surface the multifarious reactor types in use, namely: batch,
spectroscopies are mere spectators and not the active continuous stirred tank slurry, Rxed and Suid bed,
participants. Prudent researchers interlink surface trickle bed, catalytic gauze and spinning basket.
science results with those from stable Rnal product The high sensitivity of GC (detection limits of 0.1
time proRles and conventional rate studies on the to 1 ppm) enables very low conversions to be studied.
same catalyst. This problem has been highlighted in This reduces the problems of poisoning by products
an elegant demonstration that the rates of ethene of secondary reactions. GC is used routinely and
hydrogenation were invariant to the levels of the extensively in laboratory research to Rnd new cata-
much studied surface ethylidenes. This underlines the lysts and processes for existing and new products. It is
essential importance of stable product analytical also used in all the scale-up stages to full plant opera-
methods in catalyst research and evaluation. These tion. The output of the plant is constantly monitored
are integral components for the study of catalysis in by GC and other methods of analysis and the results
used to trim operating conditions to meet product
speciRcations when there are changes in feedstock
composition, Sow rates, and reactor temperatures
and pressures.
Conventional reactor studies can be readily auto-
mated by using multiple sample loop valves activated
by signals from minicomputers, which also acquire
and process the compositional data and GC condi-
tions in digital modes. A semiautomatic pulsed small
reactor system with a simple timing device for samp-
ling at prescribed times and early data acquisition and
processing methods was described as early as 1960.
A commercial laboratory rig incorporating GC is
now marketed for catalytic studies at pressures of
several hundred atmospheres and temperatures up to
3503C (Chemical Data Systems, Pittsburgh, USA).
The pulse of products from the high pressure micro-
reactor is depressurized in a multiple loop valve so
Figure 2 Energy diagram for a reaction occurring homogene- that its pressure is compatible with that of the carrier
ously and heterogeneously. gas stream into which it is injected. The unit is
semiautomatic and data acquisition, processing and

control is by a personal computer (PC).
The very ease of using GC routinely often means
that the technique is not mentioned in the title of
research papers in the open literature, in much the
same way that the use of NMR is often assumed and
not noted in papers on synthetic organic chemistry.
Perusal of recent issues of journals such as the Journal
of Catalysis and the Journal of Applied Catalysis
indicate that about 40% have used GC for product
analysis, while a BIDS search of the years 1991}97
shows the linking of GC with catalysis in about 90
papers each year.
It is worth emphasizing that process improvement
is more common than purposeful or serendipitous
invention and discovery. This is because large com-
panies prefer to expand and advance by steady or
even marginal changes associated with low risk rather
than to initiate new high risk processes requiring
large capital investment and many years of negative
cash Sows and long payback periods. An example is
hydrocracking, which was not considered worth-
while by most of the oil companies until Chevron Figure 3 Schematic of a microcatalytic chromatographic reac-
tor. TCD, thermal conductivity detector.
(California, USA) brought the Rrst hydrocracking
plant on stream in the mid-1960s } competition is as
much the mother of invention as necessity! The scale
of operations provides the motivation for marginal for catalyst research since it was Rrst described in
improvements; a catalyst that increases the yields of 1955. A schematic of the system is shown in Figure 3.
a product by 1% can earn C1 million per annum if The principle of this ‘one shot’ technique is simple.
product turnover is C100 million p.a. It is in such A small catalytic reactor is placed as near as possible
marginal improvements that GC can prove so valu- to the entrance of a GC column. Small pulses of
able because of the ease of measuring such small reactant are injected into an inert or reactive carrier
changes in product yields. gas stream just before the reactor. Reaction takes
GC has played a role in the development of devices place in the catalyst bed and the unconverted reactant
to protect the environment from the complex chem- pulse together with volatile products are swept into
ical wastes from the chemical semiconductor and the GC column where the components are separated
electroplating industries. It was used in the develop- in the usual way and then detected by a rapid re-
ment of the remarkable ‘three-way’ catalytic conver- sponse microthermal conductivity detector. A useful
ter Rtted into motor vehicle exhausts. The simulta- adjunct, employed even in the earliest studies, is to
neous oxidation of carbon monoxide and unburnt place a Geiger detector in the exit stream from the
hydrocarbons alongside the reduction of nitric oxide thermal conductivity detector so that the radioactiv-
in the same catalyst bed exposed to widely different ity of each peak can be determined. This extends the
exhaust compositions and temperatures from ambi- method to include tracer experiments using 14C- and
ent to 10003C) is a major chemical and engineering tritium-labelled compounds. The advantages and dis-
achievement. Even the humble kitchen stove has been advantages of the microreactor are listed below.
rendered self-cleaning by incorporating manganese
Advantages
dioxide/zeolites/ferrites into its glass enamel coatings.
The application of catalysis and GC in environmental 1. Rapid surveys of new or modiRed catalyst are
protection is increasing in importance. possible, enabling rapid assessment of whether the
changes in catalyst formulations have been beneR-
Microcatalytic Chromatographic cial or detrimental (but see item 6 below). The rate
of acquisition of results ultimately depends on the
Technique speed of analysis; because of its relatively slow
The microreactor is probably the most direct link speed, the interval between successive pulses can
between GC and catalysis and has been widely used be tens of minutes in packed columns but less
in the high speed wall coated open-tubular col- rate:

umns (WOTC) now available.
2. The high sensitivity of GC enables very small reac- 2aL PrT0 (Pi/Pr)3/2!1
tr"
tant pulses to be used with a consequent reduction 3V P Tr (Pi /Pr)2!1

in the risk of poisoning active sites by trace impu-
rities. where L is the reactor length and a its cross-
3. Radioactive tracer experiments are straightfor- section; V0, P0 and T0 are the volume Sow rate,
ward, enabling the path of individual reaction sites pressure and temperature at the end of the GC
in reactant molecules to be tracked into the prod- column; Pi and Pr are the pressures at the inlet and
uct molecules. outlet of the reactor.
4. By careful analysis of the changes in product yields Nonlinear Rrst-order plots result if the uncorrec-
with changes in Sow rate (contact time) and ted residence time is used, whereas excellent lin-
sample size, the rate expression of the reaction can earity is obtained by using corrected residence
be established and rate constants and Arrhenius times.
parameters estimated (but see Disadvantage 2). 3. If the carrier gas is itself a reactant, its depletion in
5. The technique is particularly well suited for kin- the gas pulse leaving the reactor can result in
etic isotope experiments. Alternate pulses of artefacts because of changes in detector sensitivity
a deuterated and protonated reactant can be injec- when the pulse components pass through the de-
ted into the microreactor so that their reactions tector. This can be avoided by split stream tech-
are compared on exactly the same catalyst sample niques. In microreactor studies of the dehydroch-
under identical conditions. Very accurate mea- lorination of 1,2-dichloroethane, air was used as
sures of primary and secondary kinetic isotope the carrier gas to eliminate the builid-up of car-
effects (the ratio kD/kH) can obtained and subtle bonaceous poisons when inert carrier gases were
mechanistic detail unravelled. used. The problem of different amounts of oxygen
6. Poisoning effects can sometimes be monitored if feeding into the Same ionization detector (FID)
the activity of the catalyst is found to diminish Same at different Sow rates was overcome by
with successive pulses, an aspect that can be en- splitting the stream leaving the GC column, so that
hanced by using longer contact times or larger the air supply to the Same could be held constant
pulses. by venting different amounts of excess air through
7. The developments in data acquisition, storage and a control valve in the column exit stream. Peak
processing, coupled together with the use of highly broadening and overlapping occurs in the catalyst
reproducible and readily automated loop sampling bed and in the dead volume between the end of the
valves, beneRt and facilitate the use of the microreactor and the beginning of the GC column, lead-
catalytic reactor for mechanistic and kinetic stud- ing inevitably to some loss of resolution.
ies and for catalyst screening. 4. The data obtained are inherently integral despite
the small contact times used, so that the technique
Disadvantages
is not readily adapted to give differential reaction
1. Steady-state conditions are often not achieved in kinetics.
the short contact times used in microreactors. This 5. Most catalyst laboratories will probably use
can be useful for kinetic studies, but is a disadvan- microcatalytic chromatographic reactors in the
tage for screening tests of a new catalyst or a new near to medium-term future, especially since high
formulation since the catalyst will usually operate speed numerical computer methods are now avail-
commercially in a continuous Sow conRguration, able to model the differential equations of Sow
where steady-state or pseudo steady-state condi- and reaction in these reactors. These results place
tions prevail. rigorous and stringent tests on the reaction mecha-
2. Accurate measures of the residence time of the nisms proposed and reliable estimates of rate
reactant pulse in the catalyst bed are essential for parameters for some of the rate-determining steps
reliable kinetic measurements. The time cannot be can result. These numerical modelling techniques
calculated simply from the outlet gas volumetric are much better than the use of the approximate
Sow rate and the bed volume because of the con- though complicated rate expressions, which con-
siderable pressure drop between the reactor and tain functions of many rate parameters often con-
the end of the GC column, which varies with Sow joined and requiring much ingenuity and chemical
rate. The corrected residence tine tr is determined intuition to untwine. However, the necessity for
from the following equation derived by applica- intricate valving, pressure and temperature correc-
tion of the gas laws to the exiting carrier gas Sow tions, and computer control and data acquisition
and processing, leads to a loss of the essential maintain the gas Sow and peak separation. By CCR it
simplicity of the microreactor technique. These has been possible to achieve virtually complete de-
elaborations may lead researchers into using con- hydrogenation of cylcohexane to benzene at temper-
ventional catalytic reactors with proscribed atures where the equilibrium constant was
switching of the reacted gas streams into multi- 10\3 mol L\1. The yields of useful primary products
loop sampling valves linked to GC and GC-MS such as alcohols and epoxides are often low because
analysers. of their rapid secondary reactions. The secondary
reactions can be reduced chromatographically by
Chromatographic Catalytic Reactors holding back the primary product on the catalyst
surface while the reactant pulse rapidly sweeps down
(CCR): the Use of Fluid Logic Modules the column because of its lower retention. Figure 5
A GC catalytic reactor deliberately combines the illustrates the simpliRed situation for the reaction
chromatographic separation of components with sequence given below:
a catalytic function in the same reactor bed. The
technique, Rrst described in 1956, is a variant of the RH#O2PROH [I]
microcatalytic reactor just discussed. The reactor, in
conjunction with pulse techniques, can be used for ROH#ÒaPCO2#H2O [II]
yield enhancement by displacing equilibria and for
preserving primary products from secondary reac- The occurrence of reaction [II] can be reduced by
tions with one of the reactants. The yields of primary rapid separation of the reactant pulse containing oxy-
products from a conventional catalytic reactor are gen from the ROH formed and retained in the cata-
ultimately limited, either thermodynamically through lytic chromatographic column. By using a silica-sup-
the value of the equilibrium constant, or kinetically ported silver oxide CCR, large quantities of crotonal-
by consumption of primary products, which often dehyde have been preserved in the catalytic oxidation
react more rapidly with one of the reactants, such as of but-1-ene. The lifting of both the thermodynamic
oxygen, than does the other reactant, such as an and kinetic limitations is enhanced by using narrow
alkene or alkane. Both these limitations can be avoid- injected pulses. Injectors based on Suid logic mod-
ed in a chromatographic catalytic reactor (CCR). The ules, which have no moving parts and none of the
equilibrium limitation arises because the catalyst ac- intrinsic delays of actuating the opening and closing
celerates the second-order back-reactions as much as of solenoid valves, are ideal for this task and have
it does the forward reaction. The back-reactions may previously been used for fast GC analysis. Pulses
be minimized by chromatographically separating the down to 10 ms have been generated and further re-
products from each other. Figure 4 illustrate the case ductions are possible.
for the simple equilibrium, A"B#C. The back-
reaction can only occur in the initial regions of the
column where there is overlap between the B and
Process Monitoring and Control
C peaks. If the bed is long enough, then complete The wealth of detail given by GC about the composi-
conversion of A into B and C is possible. The laws of tion of streams leaving reactors and puriRcation units
thermodynamics are not broken because the equilib- such as distillation columns can, in principle, be used
rium limitation is circumvented by an energy input to for manual, semiautomatic or fully automatic process
Figure 4 Illustration of how the equilibrium, APB#C, is displaced in a GCC reactor.

Figure 5 Illustration of how a primary product, ROH, is preserved from secondary reactions with oxygen in a GCR.
control. Automatic GC has been applied to determine forward by posing a challenge to the analyst to speed
gasoline components and a linear algorithm has been up chromatographic methods. If the challenge can be
developed relating Reid vapour pressures, boiling met, then the closing of control loops for catalytic
points and ASTM distillation curves to component reactors based on GC will follow.
indices with good correlations and repeatability. The
method has been developed further to enable the
RON of a gasoline to be determined automatically by
Current Trends and Future Prospects
GC analysis with a standard deviation of only 0.09 Table 3 lists some of the recent developments that are
for RON values in the range 90 to 100. The method close to commercial realization together with more
has not yet been taken to the point of closing a con- speculative, but possible, advances in the new millen-
trol loop so that a reactor or a distillation column is nium. GC will play an important part in the catalyst
controlled by GC, although this closure is feasible.
GC suffers as the control element because of the
Table 3 Current trends and future prospects
time needed for analysis. The response time of the
plant must be longer than the time for analysis, which Gas turbines for electricity generation powered by catalytic
is usually tens of minutes, and there is a clear need to combustion of natural gas and biomass
develop faster GC based on short narrow-bore open- Better electrochemical catalysts with high durability for fuel cells
tubular columns and variants. It is relevant that based on methaonl and eventually hydrocarbons
Production of propene oxide by the catalytic partial oxidation of
a method called the ‘optical octane dipstick’ is now propene
being used in many reRneries. This uses a nondisper- Home and hospital hygiene based on catalyst-containing surface
sive near-infrared analyser (NIR) to determine the coatings
absorbence at Rve wavelengths. Apparently a very Treatment of commercial and domestic waste
good measure of RON is obtained from an algorithm Search for power generation from renewable resources and
sunlight-activated catalytic dissociation of water into hydrogen
linking these absorbencies to ASTM/RONs deter- and oxygen
mined in a standard Petter engine. Clearly this spec- Enzymes immobilized on rugged supports may provide ways of
troscopic method is faster than GC but can only be fixing nitrogen at ambient temperatures and also of achieving
used for a narrow range of RONs and composition. commercial chiral synthesis
The more detailed compositional description by GC Closing of control loops for catalytic reactors, distillation columns,
etc., based on GC analysis of product streams and neural
analysis could outweigh its present relative slowness. control networks
The NIR control units already in use pave the way
2260 III / CATALYST STUDIES: CHROMATOGRAPHY / Isolation: Magnetic Techniques
theories and technologies that will be needed to bring Hall WK, MacIver DS and Weber HP (1979) Industrial and
these ideas to fruition. The tasks are challenging, even Engineering Chemistry 52: 425.
formidable, and require the commitment of scientists Guichon G and Gonnard F (1979) Analytical Chemistry
and engineers as well as the provision of vast funds 51: 379.
for research, development and capital investment by Mile B, Adlard ER, Roberts GM and Sewell PA (1988)
Catalysis Today 2: 685.
governments and industry.
Mile B, Ryan TA, Tribeck TD, Zammitt MA and Hughes
GA (1994) Topics in Catalysis 1: 153.
Further Reading Mile B, Howard JA, Tomietto M, Joly HA and Sayari
(1996) Journal of Materials Science 31: 3080.
Carberry JJ (1976) Chemical and Catalytic Reaction Engin- Petroff N, Hoscheitt A and Durand D (1993) Hydrocarbon
eering. New York: McGraw-Hill. Processing (International Edition) 72: 103.
Choudry VR and Doraiswamy R (1971) Industrial Engine- Rideal EK and Taylor HS (1919) Catalysis: Theory and
ering and Chemical Product Research and Development Practice. London: Macmillan.
10: 218. Rooney JJ (1985) Journal of Molecular Catalysis 31:
Durand JP, Boscher Y and Dorbon M (1990) Journal of 147.
Chromatography 404: 49. Somorjai GA (1984) Chemical Society Review 13: 321.
Emmett PH (1957) Advances in Catalysis, vol. 9, p. 645. Zelter MS (1993) Hydrocarbon Processing (International
New York: Academic Press. Edition) 72: 103.
CATALYST STUDIES: CHROMATOGRAPHY
efRcient procedure and the whole separation process

Isolation: Magnetic Techniques can be performed in the same tube. Large differences
between magnetic permeabilities of the magnetic and
I. S[ afar\ eH k and M. S[ afar\ eH kovaH , Institute of Landscape
nonmagnetic materials can be exploited in developing
Ecology, Academy of Sciences, highly selective separation methods. The separation
C[ eske& Bude\ jovice, Czech Republic procedure can easily be scaled up if large quantities of
living cells are required.
Principles of Magnetic Separation

The Rrst experiments with magnetic separation of
cells date from the 1950s when lymphocytes were
of Cells
magnetically separated after in vitro phagocytosis of Two types of magnetic separation can be distin-
iron granules. The real boom in the application of guished when working with cells. In the Rrst type,
magnetic labels for cell isolation came in the 1980s cells to be separated demonstrate sufRcient intrinsic
and since then an enormous amount of work has been magnetic moment so that magnetic separations can
done in both the development and application of this be performed without any modiRcation. There are
technique. only two types of such cells in nature: magnetotactic
Magnetic separation of cells has several advantages bacteria containing small magnetic particles within
in comparison with other techniques. In general, the their cells and red blood cells (erythrocytes) contain-
magnetic separation procedure is gentle, facilitating ing high concentrations of paramagnetic haemoglo-
the rapid handling of delicate cells in an unfriendly bin. In the second type, cells of interest have to be
environment. It permits the cells of interest to be tagged by a magnetic label to achieve the required
isolated directly from crude samples such as blood, contrast in magnetic susceptibility between the label-
bone marrow, tissue homogenates, stools, food, culti- led and unlabelled cells. The attachment of magnetic
vation media, soil and water. The cells isolated labels is usually attained by the use of afRnity ligands
by magnetic separation are usually pure, viable and of various types, which can interact with target
unaltered. Magnetic separation is a simple, fast and structures on the cell surface. Usually antibodies
III / CATALYST STUDIES: CHROMATOGRAPHY / Isolation: Magnetic Techniques 2261
against speciRc cells surface epitopes are used, but manence the particles are not attracted to each other
other speciRc ligands can also be employed. The and therefore they can be easily suspended into a ho-
newly formed complexes have magnetic properties mogeneous mixture in the absence of any external
and can be manipulated using an appropriate mag- magnetic Reld. Magnetic particles typically comprise
netic separator. Rne grains of iron oxides dispersed throughout the
The magnetic separation process for the puriRca- interior of a polymer particle (in many cases of
tion of target cells usually consists of the following a monosized type), the surface chemistry of which can
three fundamentals steps: be modiRed to provide a range of different linking
methods. Alternatively, silanized particles of mag-
1. The suspension containing cells of interest is netic iron oxides or magnetic porous glass can be used
mixed with magnetic labels. Incubation time is for the same purpose.
usually not longer than 30}60 min. Then the mag- A number of particulate magnetic labels can be
netic complex formed is separated using a mag- purchased commercially. Up to now, in most applica-
netic separator and the supernatant is discarded or tions, monosized polymer particles marketed as
used for another application. Dynabeads (Dynal, Oslo, Norway) in various forms
2. The magnetic complex is washed several times to have been used. Dynabeads are prepared from mono-
remove unwanted contaminants. The selected cells sized macroporous polystyrene particles which are
with attached magnetic labels can be used directly, magnetized by an in situ formation of ferromagnetic
e.g. for cultivation experiments. Alternatively, material inside the pores. Other commercially avail-
cells can be disrupted and the cell content analysed able magnetic particles can be successfully used for
using various methods. cell separation. A selection of these products can be
3. If necessary, a variety of detachment procedures found in Table 1.
can be used to remove the magnetic labels from
the separated cells. After detachment, the mag- Colloidal magnetic labels Colloidal magnetic labels
netic label is removed from the suspension in (typical size c. 50}200 nm) are prepared by a variety
a separator and free cells are ready for further of methods which result in Socks composed of poly-
applications and analyses. mer (typically dextran, starch or protein) and mag-
netite and/or other iron oxide crystals. A standard
procedure for the synthesis of superparamagnetic
Necessary Equipment dextran nanospheres is performed by precipitation
of iron oxide in the presence of the polysaccharide.
Magnetic labels and magnetic separators are neces- To such materials, ligands (usually antibodies, lec-
sary to be able to perform efRcient cell separation. tins, streptavidin or biotin) are coupled so they can
Typical examples are given below. be used for cell separation. Using high gradient
magnetic columns, the labelled cells are easily separ-
Magnetic Labels ated. Magnetic particles isolated from magnetotactic
With the exception of erythrocytes and magnetotactic bacteria (50}100 nm) composed of magnetite
bacteria, the cells to be isolated have to be magneti- covered by a stable lipid membrane can also be used
cally labelled in order to be amenable to magnetic successfully.
treatment. Magnetic labelling can be performed with
magnetic and superparamagnetic particles (c. 1 m Magnetoliposomes Magnetoliposomes are mag-
and more in diameter), magnetic colloids (c. netic derivatives of ordinary liposomes prepared by
50}200 nm), magnetoliposomes or with molecular incorporation of colloidal magnetic particles into the
magnetic labels. In most cases the magnetic properties lipid vesicles. When magnetoliposomes are associated
of the labels are caused by the presence of small with antibodies they can label and/or selectively con-
particles of magnetite (Fe3O4) or maghemite (- centrate the target cells.
Fe2O3); in some cases, chromium dioxide particles or
ferrite particles have been used. Molecular magnetic labels Molecular magnetic
labels are usually lanthanides (especially erbium), fer-
ritin and magnetoferritin. Erbium in the form of er-
Magnetic and superparamagnetic particles Mag- bium chloride (ErCl3) has been used for magnetic
netic and superparamagnetic particles (typical dia- labelling of a variety of cells. Different Er3# binding
meter c. 1}5 m) attached to the target cells can easily sites, such as carboxyl groups in glycoproteins or the
be removed from a suspension with a simple magnetic Ca2# receptor sites on the cell wall are responsible for
separator. Since there is usually no magnetic re- the binding of erbium ions.
Table 1 Selected examples of commercially available magnetic particles and colloidal magnetic labels used or usable for magnetic
separation of cells
Name Diameter (m) Polymer End groups and Immobilized Manufacturer/supplier

composition/surface activation possibility compounds
modification
BioMag &1 Silanized iron oxides }COOH Secondary Abs, Perseptive

}NH2 anti-CD Abs, Biosystems,
anti-fluorescein Ab, Framingham, MA, USA
protein A, protein G,
streptavidin, biotin
Dynabeads M-280 2.8 Polystyrene Tosyl-activated Secondary Abs, Dynal, Oslo, Norway
Dynabeads M-450 4.5 anti-CD Abs, Abs
Dynabeads M-500 5 against Escherichia
coli O157, Salmonella,
Listeria, Crypto-
sporidium, streptavidin,
oligo(dT)
Estapor &1 Polystyrene }COOH Prolabo,
}NH2 Fontenay-sous-Bois,
France
Ferrofluids 0.135 Modified hydrophilic }COOH Secondary Abs, Immunicon, Huntingdon
0.175 protein }NH2 protein A, Valley, PA, USA
streptavidin
M 100 1}10 Cellulose }OH Scigen, Sittingbourne,
M 104 UK
M 108
MACS 0.05 Dextran }OH Secondary Abs, Miltenyi Biotec,
Microbeads anti-CD Abs, Bergisch Gladbach,
streptavidin, biotin Germany
Magnetic 1}2 Polystyrene, }COOH Secondary Abs, Polysciences,
microparticles cellulose, }NH2 protein A, Warrington, PA, USA
polyacrolein protein G,
streptavidin
Magnetic 0.09}0.6 Starch, dextran, }OH, }COOH Streptavidin, Micro caps, Rostock,
nanoparticles chitosan biotin, protein A Germany
MagNIM 0.05 }COOH Secondary Abs, Cardinal Associates,
0.25 }NH2 Ab against E. coli Santa Fe, NM, USA
0.5 O157, streptavidin,
protein A
Magnetic &1 Polystyrene }COOH Secondary Abs, Bangs Laboratories,
particles }NH2 Protein A, Fishers, IN, USA
streptavidin,
MPG 5 Porous glass }NH2, hydrazide, Streptavidin, CPG, Lincoln Park, NJ,
glyceryl avidin USA
XM200 3.5 Polystyrene }COOH Secondary Abs, Advanced
Microsphere protein A Biotechnologies,
Epsom, UK
Ferritin is a naturally occurring, soluble iron stor- cell membrane. Magnetic derivatives of ferritin called
age protein in mammals. For magnetic modiRcation magnetoferritin, prepared by controlled reconstitu-
of cells cationized horse spleen ferritin (ferritin tion conditions, have also been used for cell labelling.
coupled with N,N-dimethyl-1,3-propanediamine) ex-
Magnetic Separators
hibiting a net positive charge at pH 7.5 is usually
used. Under these conditions the cationized ferritin A variety of magnetic separators is available on the
readily forms ionic bonds with the anionic sites on the market, starting with very simple concentrators for
one test tube and ending with complicated fully auto-

mated devices. In many cases, especially when work-
ing with larger labels, very cheap home-made
magnetic separators can be used successfully.
Batch magnetic separators Batch magnetic separ-

ators are usually made from strong rare-earth per-
manent magnets embedded in disinfectant-proof
material. The racks are designed to hold various sizes
and numbers of tubes. Some of the separators have
a removable magnet plate to facilitate easy washing
of magnetic particles (Figure 1). Test tube magnetic
separators enable separation of magnetic particles
from volumes ranging between about 5 L and
50 mL. It is also possible to separate cells from the
wells of standard microtitration plates. Magnetic
complexes from larger volumes of suspensions (up to
approximately 500}1000 mL) can be separated using
Sat magnetic separators. More sophisticated mag- Figure 2 (See Colour Plate 64). Examples of laboratory-scale
netic separators are available, e.g. those based on the high gradient magnetic separators. Left: MiniMACS separation
quadrupole and hexapole magnetic conRguration. unit; right: MidiMACS separation unit, both with inserted columns.
Courtesy of Miltenyi Biotec, Bergisch Gladbach, Germany.
Flow-through magnetic separators Flow-through
magnetic separators are characterized by the Sow of
ment of computer-controlled magnetic cell sorters,
the liquid and suspended cells through the separation
such as CliniMACS (Figure 3) and AutoMACS, both
system. These systems are usually more expensive and
produced by Miltenyi Biotec, Germany.
more complicated in comparison with batch separ-
ators, but for preliminary experiments simple devices
can also be used.
Laboratory-scale high gradient magnetic separtors
(HGMS; Figure 2) are composed of small columns
loosely packed with Rne magnetic-grade stainless
steel wool which are placed between the poles of
strong permanent magnets or electromagnets. Mag-
netically labelled cells are pumped through the col-
umn, labelled cells are retained on the steel wool, the
Reld is removed and cells are retrieved by Sow and
usually by gentle vibration of the column. Automa-
tion of the separation process has led to the develop-
Figure 1 (See Colour Plate 63). Example of a magnetic separ-

ator (Dynal MPC-M) for work with microcentrifuge tubes of the Figure 3 (See Colour Plate 65). Computer-controlled magnetic
Eppendorf type, with a removable magnet plate to facilitate easy cell sorter CliniMACS. Courtesy of Miltenyi Biotec, Bergisch Glad-
washing of magnetic particles. Courtesy of Dynal, Oslo, Norway. bach, Germany.
Another differentiation of magnetic separation

techniques is based on the selection of the magneti-
cally labelled cells.
1. Negative selection. Negative selection is a method

by which a cellular subset is puriRed by removing
all other cell types from the sample. Both the direct
and indirect method are applied for negative selec-
tion. The advantage is that the puriRcation process
does not involve any direct contact of magnetic
labels with the cells to be isolated.
2. Positive selection. The target cells are isolated
from the cell suspension. Both the direct and in-
direct method can be used. The separated magneti-
cally labelled cell complexes can be further charac-
terized directly, but in many cases it is necessary to
remove larger magnetic particles from the posit-
ively selected cells after their isolation.
3. Depletion of cells. Depletion is a method by which
one or more unwanted cellular subsets is removed
Figure 4 (See Colour Plate 66). The Isolex 300i Cell Selection
System. Courtesy of Nexell Therapeutics, USA.
from a cell suspension. Both the direct and the
indirect procedure can be applied for this purpose.
Immunomagnetic Separation
A continuous magnetic sorter based on an elec-
trophoresis counter-Sow chamber has been de- Immunomagnetic separation (IMS) is the most often
veloped. The injected magnetic particles are deviated used approach for the isolation of cells. Most often,
in the inhomogeneous magnetic Reld and focused into a monoclonal antibody is used for IMS, but also
a stream that is completely separated from the polyclonal antibodies are used successfully. IMS can
streams of the undeviated particles. be performed in all the formats mentioned above.
The Isolex 300i Magnetic Cell Separator, produced In the direct method the appropriate antibody is
by Nexell Therapeutics, USA, is intended for the coupled to the magnetic particles and colloids, which
isolation of stem cells from blood or bone marrow. are then added directly to the sample. Ideally, the
The whole process is done automatically and the antibody should be oriented with its Fc part towards
device represents a Sexible platform for future ap- the magnetic particle so that the Fab region is point-
plications in cell separation (Figure 4). ing outwards from the particle. Several procedures
are available for direct binding of antibodies
Procedures for Performing Magnetic (Table 2).
The indirect method is also used very often. The
Separation of Cells cell suspension is Rrst incubated with primary anti-
Magnetic separation of cells is usually performed in bodies which bind to the target cells. Not only puri-
one of the following formats: Red primary antibodies have to be used, crude
antibody preparations or serum can also be used.
1. Direct method. The afRnity ligand is coupled to After incubation, the unbound antibodies are usually
the magnetic particles, which are then added dir- removed by washing. Then magnetic particles with
ectly to the cell sample. During incubation the immobilized secondary antibody are added to bind
magnetic particles will bind the target cells which the labelled cells. Target cells } primary antibody
can be then recovered using a magnet. complexes } can be also captured by protein A or
2. Indirect method. Target cells are Rrst sensitized protein G immobilized on magnetic carriers. Alterna-
with a suitable primary afRnity ligand. After incu- tively, biotinylated or Suorescein-labelled primary
bation, excess unbound afRnity ligand is removed antibodies and magnetic particles with immobilized
by washing the cells and then magnetic particles streptavidin or anti-Suorescein antibody are used to
with an immobilized secondary afRnity ligand capture the target cells.
with afRnity for the Rrst ligand are added. The The indirect method is generally more efRcient in
magnetic particles will bind the target cells, which removing target cells from a suspension because free
can then be recovered using a magnetic separator. antibodies will Rnd their target antigen more easily
Table 2 Selected procedures for binding of antibodies on magnetic particles and colloids
Adsorption of antibodies on hydrophobic magnetic particles (especially those made of polystyrene)

Covalent binding of antibodies on activated magnetic particles (e.g. tosyl-activated), or on magnetic particles carrying appropriate
functional groups (e.g. carboxy, amino, hydroxy, hydrazide) using standard immobilization procedures
Immobilization of secondary antibodies (i.e. antibodies against primary antibodies) on magnetic particles followed by binding of primary
antibodies
Immobilization of biotinylated antibodies on magnetic carriers with immobilized streptavidin
Immobilization of antibodies on magnetic particles with immobilized protein A and protein G
Immobilization of antibodies tagged with oligo dA on magnetic particles with immobilized oligo dT
Immobilization of antibodies on magnetic carriers with immobilized boronic acid derivative via their carbohydrate units on the Fc part
than antibodies bound to magnetic particles. The interactions can be partially eliminated using bovine
indirect technique is recommended when the target or human serum albumin, casein and nonionic ten-
cell has a low surface antigen density or a cocktail of sides such as Tween 20.
monoclonal antibodies is used. The direct method is
Magnetic Separations using other Labels
usually faster and requires less antibody than the
indirect method. Also, the direct method is advant- Antigens Antigens immobilized on magnetic par-
ageous when one does not want to cover all antigen ticles can be used for the isolation of antibody ex-
sites with antibody. pressing or antigen-speciRc cells. This approach has
Typically, 95}99% viability and purity of the pos- been successfully used for selection of antigen-speciRc
itively isolated cells can be achieved with a typical hybridoma cells or human antibody-producing cell
yield of 60}99%. Depletion efRciency often reaches lines.
99.9% and leaves remaining cells untouched. Sequen-
tial depletions are markedly more efRcient. Lectins Lectins immobilized on magnetic carriers
Magnetic particles usually do not have any nega- can interact with saccharide residues on the cell surfa-
tive effect on the viability of the attached cells. Many ces. A typical example of this approach is the applica-
types of magnetic particles are usually compatible tion of immobilized Ulex europaeus I lectin which
with subsequent analytical techniques such as Sow binds to terminal L-fucosyl residues present on the
cytometry, electron and Suorescence microscopy, surface of human endothelial cells. Magnetic beads
polymerase chain reaction (PCR), Suorescence in situ can be released from the isolated cells using a free
hybridization (FISH) or cultivation in appropriate competing sugar.
nutrient media. In some cases, however, it is neces-
sary to remove larger immunomagnetic particles from Oligosaccharides Oligosaccharides immobilized on
the cells after their isolation. The detachment process magnetic particles can be used for the rapid isolation
can be performed in several ways (Table 3). of speciRc lectin-expressing cells. Target cells bound
Incubation time for cell separation is usually to the magnetic particles can be released using a free
5}60 min while the binding of primary antibodies to competing saccharide structure.
secondary coated magnetic particles usually takes 30 min
or less. In positive isolation, the purity of cells gener- Bacteriophage Salmonella-speciRc bacteriophage
ally decreases with time, although the yield increases. immobilized to a magnetic solid phase has been used
NonspeciRc interactions of nontarget cells with hy- for the separation and concentration of Salmonella
drophobic magnetic particles can be expected. These from food materials.
Table 3 Selected typical procedures for detachment of cells after immunomagnetic separation
Incubation of rosetted cells overnight in cell culture medium with subsequent mechanical forces such as firm pipetting, flushing the
suspension 5}10 times through a narrow-tipped pipette
Trypsin, chymotrypsin and pronase have general applicability for proteolytic detachment of isolated cells
Detachment with a polyclonal antibody that reacts with the Fab fragments of primary monoclonal antibodies on magnetic beads. This
principle is commercialized by Dynal, Norway (DETACHaBEAD)
Using synthetic peptides which bind specifically to the antigen-binding site of primary antibodies (Baxter Healthcare, Deerfield, IL, USA)
Antibodies immobilized via carbohydrate units on the Fc part to the magnetic particles with immobilized }B(OH)3 groups are dissociated
with sorbitol
A complex primary antibody}DNA linker can be split enzymatically using DNase
Cryptosporidium oocysts were successfully released from the immunomagnetic particles by decreasing the pH of the suspension
(adding HCl)
Erbium ions, ferritin and magnetoferritin have tation requires an effective removal of T cells from
been used for magnetic labelling of both prokaryotic the bone marrow of the donor. A direct method
and eukaryotic cells. Magnetotactic bacteria can be enabled a 103 times depletion of T cells.
introduced into granulocytes and monocytes by Magnetic particles are being increasingly used for
phagocytosis which enables their magnetic separ- isolation of human cell subsets directly from blood
ation. Submicron magnetic particles of -Fe2O3 and other cell sources. B lymphocytes, endothelial
adhere to the surface of Saccharomyces cerevisiae, cells, granulocytes, haematopoietic progenitor cells,
making the cells magnetic and amenable to magnetic Langerhans cells, leukocytes, monocytes, natural
separation. killer cells, reticulocytes, T lymphocytes, spermato-
Magnetotactic bacteria, due to the presence of fer- zoa and many others may serve as examples. Cells
romagnetic material in their cells, can be magnetically from other animal and plant species have been suc-
separated without any labelling. Erythrocytes can be cessfully separated, too.
separated by the high gradient magnetic separation Not only whole cells, but also cell organelles can be
technique after conversion of diamagnetic eryth- isolated from crude cellular fractions. Dynal (Oslo,
rocytes containing oxyferrohaemoglobin into para- Norway) has developed Dynabeads M-500 Subcellu-
magnetic red blood cells by the oxidation of the iron lar, which are able to isolate rapidly more than 99%
atoms in the cell haemoglobin to the ferric state of target organelles.
(methaemoglobin). Erythrocytes, infected by Plas- In the area of parasitology Cryptosporidium and
modium, contain paramagnetic hemozoin, that is Giardia are the parasites where IMS is of interest.
a component of malarial pigment. The paramagnetic Two commercially available kits can be used for this
moment of hemozoin is of sufRcient magnitude to purpose. Both products are used in the method 1622:
enable the separation of malaria-infected (hemozoin- Cryptosporidium in Water by Filtration/IMS/FA
bearing) erythrocytes. (December 1997 Draft) of the US Environmental
Protection Agency. In very low turbidity samples
Magnetic Separations in Microbiology, (clean waters), IMS has demonstrated signiRcantly
Cell Biology, Medicine and better results than the standard procedures. When
water samples were turbid, the recovery efRciency of
Parasitology IMS diminished.
IMS and, in some cases, lectin-magnetic separations
are often used in the above-mentioned disciplines. In
microbiology they are especially used for the detec-
Future Developments
tion of pathogenic microorganisms. IMS enables the Magnetic separation of cells is a simple, rapid, speci-
time necessary for detection of the target pathogen to Rc and relatively inexpensive procedure, which en-
be shortened. Target cells are magnetically separated ables the target cells to be isolated directly from crude
directly from the sample or the pre-enrichment me- samples containing a large amount of nontarget cells
dium. Isolated cells can than be identiRed by stan- or cell fragments. Many ready-to-use products are
dard, speciRc microbiological procedures. IMS is not available and the basic equipment for standard work
only faster but also usually gives a higher number of is relatively inexpensive. The separation process can
positive samples. Also sublethally injured and stressed be relatively easily scaled up and thus large amount of
microbial cells can be very efRciently isolated using cells can be isolated. New processes for detachment
IMS. The most important microbial pathogens can be of larger magnetic particles from isolated cells enable
detected using commercially available speciRc im- use of free cells for in vivo applications. Modern
munomagnetic particles; they are used for the detec- instrumentation is available on the market, enabling
tion of Salmonella, Listeria and Escherichia coli O157. all the process to run automatically. Such devices
New immunomagnetic particles for the detection of represent a Sexible platform for future applications in
other microbial pathogens are under development. cell separation.
Removal of cancer cells is one of the most impor- IMS play a dominant role at present but other
tant applications of IMS in the area of cell biology speciRc afRnity ligands such as lectins, carbohydrates
and medicine. The Rrst experiments were performed or antigens will probably be used more often in the
in the 1970s and since then an enormous number of near future. There are also many possibilities to
applications have been described. Cancer cells are combine the process of cell magnetic separation
usually removed from bone marrow prior to its with other techniques, such as PCR, enabling the
autologous transplantation and using IMS they are elimination of compounds possibly inhibiting DNA
detected in blood. Elimination of graft-versus-host polymerase. New applications can be expected, espe-
disease (GvHD) in allogenic bone marrow transplan- cially in microbiology (isolation and detection of
III / CELLS AND CELL ORGANELLES: FIELD FLOW FRACTIONATION 2267
microbial pathogens) and parasitology (isolation and Recktenwald D and Radbruch A (eds) (1998) Cell Separ-
detection of protozoan parasites). No doubt many ation Methods and Applications. New York: Marcel
new processes and applications in other Relds of bio- Dekker. 352 pp.
sciences and biotechnologies will be developed in the S[ afar\ mH k I and S[ afar\ mH kovaH M (1999) Use of magnetic tech-
near future. niques for the isolation of cells. Journal of Chromatog-
raphy B 722: 33d53.
See Colour Plates 63, 64, 65, 66. S[ afar\ mH k I, S[ afar\ mH kovaH M and Forsythe SJ (1995) The applica-
tion of magnetic separations in applied microbiology.
See also: II/Centrifugation: Analytical Centrifugation;
Journal of Applied Bacteriology 78: 575d585.
Large-Scale Centrifugation. III / Cells and Cell Or-
Ugelstad J, Olsvik ", Schmid R et al. (1993) ImmunoafRn-
ganelles: Field Flow Fractionation.
ity separation of cells using monosized magnetic poly-
mer beads. In: Ngo TT (eds) Molecular Interactions
Further Reading in Bioseparations, pp. 229d244. New York: Plenum
Cell Separation and Protein PuriTcation (1996) Oslo, Press.
Norway: Dynal. 165 pp. Ugelstad J, Prestvik WS, Stenstad P et al. (1998) Selective
HaK feli U, SchuK tt W, Teller J and Zborowski M (eds) (1997) cell separation with monosized magnetizable polymer
ScientiTc and Clinical Applications of Magnetic Car- beads. In: AndraK W and Nowak H (eds) Magnetism in
riers. New York: Plenum Press. Medicine: A Handbook, pp. 471d488. Berlin: Wiley-
Olsvik ", Popovic T, Skjerve E et al. (1994) Magnetic VCH Verlag.
separation techniques in diagnostic microbiology. Clini- UhleH n M, Hornes E and Olsvik " (eds) (1994) Advances in
cal Microbiology Reviews 7: 43d54. Biomagnetic Separations. Natick: Eaton Publishing.
CELLS AND CELL ORGANELLES:

FIELD FLOW FRACTIONATION
P. J. P. Cardot, S. Battu, Speci\c Cell Characteristics
T. Chianea and S. Rasouli,
Universite& de Limoges, Limoges, France Cellular materials range in size from 1 m to 50 m.
Cell populations are classiRed by a set of morphologi-
Copyright ^ 2000 Academic Press cal, functional and biophysical characteristics. The
biophysical characteristics are of particular interest
Introduction in FFF. Usually, separations in FFF are inSuenced
(but not directed) by surface properties of the sample
Analysis and sorting of living cells and puriRcation of components (to avoid particle}particle or particle}
cell organelles are important procedures in the life separator interactions). These properties can be
sciences. There is a wide range of techniques and modulated by the use of appropriate carrier-phase
methodologies available, which can be divided into modiRers (surfactants). In terms of FFF separations,
three main groups. The techniques in the Rrst groups mass, size and density appear to be the major Rrst
are based on physical criteria such as species size, order parameters. However, size is generally deRned
density and shape, and include centrifugation, elutri- by the radius or the diameter of a sphere whose
ation and Reld Sow fractionation. Those in the sec- volume is identical to that of the cell. Size can there-
ond group are linked to cell surface characteristics, fore be deduced accurately if the cell of interest is
while Sow cytometry techniques make up the third perfectly spherical. However, this is not usually the
group. At a fundamental level, Reld Sow fractiona- case, and the sphericity index, I, is then used:
tion (FFF) exploits the physical characteristics of the
cells or cell organelles. However, cells or cell or- 4.84(V 2/3 )
ganelles exhibit some speciRc characteristics that can I"
S
be described by a multipolydispersity matrix. The
different physical characteristics of these biological In this equation V is the cell volume and S is its
materials require different FFF techniques and modes surface area can be difRcult to determine. In terms of
of operation. Special care must be taken if biological cell population, these dimensions are averages and
integrity and viability are to be preserved. should be associated with a variance. These general
2268 III / CELLS AND CELL ORGANELLES: FIELD FLOW FRACTIONATION
considerations are also valid for cell density. Each versa. The interactions between the environmental
measurable cell characteristic is therefore associated material and the cells can now be assessed by the
with an average value and its related variance. general &biocompatibility’ rules.
A probe of this ‘diversity’ is of great signiRcance, even However the HNRBC population found in circula-
in a highly puriRed cell population. Polydispersity is ting blood is not a homogenous, and contains red
historically deRned in polymer chemistry as the ratio blood cells (RBCs) of different ages. This suggests
of the percentage of the standard deviation to the that a morphologically homogenous population can
average value. be considered to contain different sub-populations.
Any cell population can therefore be described by An HNRBC suspension contains cells of equal vol-
a set of values (average, variance) that leads to a poly- ume, and cells of equal density, shape or mass. This
dispersity index. Cells are different from polymers in complex deRnition of ‘population’ characteristics is
that each measurable characteristic of a cell can be described in Figure 1A. If a set of common para-
associated with polydispersity, so polydispersity can meters is deRned, the HNRBCs that meet these limita-
apply to mass, size, volume or density. This matrix of tions may represent only a minor proportion of the
parameters (average, variance) can be used to deRne whole population. By considering some aspects of
a &multipolydispersity matrix’. The dimensions of the this multipolydispersity matrix, rules for cell separ-
multipolydispersity index depend on the information ation based on biophysical characteristics can be
available. For example, the human normal red blood found. Figure 1B shows a three-dimensional graphi-
cell (HNRBC) is the best known cellular material. Its cal representation of the multipolydispersity of the
average volume has been measured as 95 $ 5 fL, circulating blood cells (platelets, HNRBC, lympho-
and its surface area has been calculated as cytes, monocytes, granulocytes) based on the charac-
138.0$5 m2. The HNRBC is known for its discoid teristics of size and density. It is obvious that if a size
shape, which generates other measurable dimensions. driven cell separation technology could be used, all
Its average diameter is 8.1$0.43 m, and minimum these populations could be isolated. An isolation
and maximum thicknesses of 1.0 $ 0.3 m and technique based only on density would be more com-
2.4$0.15 m have been measured. An average plex, as some blood cell population densities overlap.
sphericity index of 0.77 has been calculated. The However a size and density driven separation (whose
density was found to vary from 1.035 to 1.102, with balance has not so far been assessed) would allow
an average value of 1.051. Using these data, it is selective isolation.
possible to describe the HNRBC population using
a preliminary multipolydispersity matrix, as shown in
Table 1. FFF Techniques for Cell Separations
More complex properties such as cell/nucleus ratio
The Different Techniques
or surface electric charge density ( potential, used for
dielectric cell puriRcation) can later be added to this The requirements of different Relds have led to a
basic matrix. In FFF, the sample (i.e. cell surface) wide variety of FFF techniques. Four major technolo-
characteristics are most important in the development gies have emerged. The Rrst uses Relds generated by
of any separation process. The separator material is gravity, either simple earth gravity or centrifugally
chosen so as to avoid or limit particle}separator inter- generated multigravitational Relds. It is now generi-
actions, which can lead to limited recovery and viabil- cally described as sedimentation FFF (SdFFF for
ity, separator ageing or poisoning. In this domain multigravity Relds and GFFF for gravity induced
a general set of empirical rules emerged. If the cell Relds). The basic technology is the same, although
surface is hydrophilic (blood and yeast cells) a hydro- a more complex instrumental design is required for
phobic material should be used for the separator. The SdFFF, as shown in Figure 2A and B. However, the
lower the surface energy of the cell, the less non- very simple GFFF designs and procedures can be
speciRc cell}separator interactions occurred, and vice upgraded for SdFFF separations. This group was his-
torically the most successful in biological applica-
tions. Channel cleaning, decontamination and steril-
Table 1 Multipolydispersity matrix for HNRBC
ization procedures are common and can be per-
Average Standard Polydispersity formed simply, as described later. There are many
deviation (%) channel wall materials available, and a wide diversity
of sample injection technologies or procedures.
Volume (fL) 95 5 10 The second group encompasses Relds generated by
Density 1.051 0.08 3
Sphericity 0.77 0.15 11
a Sow, where the accumulation wall is made of a
semipermeable membrane of adapted cut-off. This
III / CELLS AND CELL ORGANELLES: FIELD FLOW FRACTIONATION 2269
Figure 1 Multipolydispersity matrix. (A) The definition of a homogenous population is complex. A cell population with homogenous
morpological characteristics is actually a complex mixture of sub-populations with different biophysical characteristics. (B) Three-
dimensional plot of circulating blood cells (normalized dimensions) as a function of size and density (from bibliographic data). Red blood
cells and lymphocytes have the closest characteristics.
Figure 2 The different FFF techniques. (A) Gravitational FFF device; (B) Sedimentation FFF device; (C) symmetrical flow FFF
device; (D) asymmetrical flow FFF device; (E) experimental layout of DEP-GFFF device; (F) schematic view of SPLITT-FFF device. 1,
Channel wall (accumulation wall); 2, Mylar spacers; 3, semipermeable membrane (accumulation wall); 4, frit; 5, interdigitated
electrode; 6, power amplifier and source of signal; a, inlet (sample and carrier phase); b, outlet (to detector and fraction collection); a’
and b’, secondary inlet and outlet.
2270 III / CELLS AND CELL ORGANELLES: FIELD FLOW FRACTIONATION
group is of interest because of the possibility of separ- limit the risk. However, Sow injections may enhance
ating species by buoyancy (particle suspension and selectivity because of the different speeds at which
carrier phase with identical density), such as cell or- sample components reach the steric hyperlayer fo-
ganelles or macromolecules. Two Sow-generated FFF cused position in the channel thickness. Elution of
techniques were developed. Symmetrical Sow FFF cellular material in a steric mode is associated with
(FFFF) uses two independent Sows, one to create the reduced recovery and possibly with reduced selectiv-
external Reld and the second to sweep the species ity. Examples of the steric hyperlayer elution mode of
along the channel (Figure 2C). Asymmetrical Sow HNRBCs and duracytes are given in Figure 3.
FFF (AFFFF) uses the same Sow along and across the
channel, as shown in Figure 2D. However, sample
injection procedures are more limited and complex Cell Elution Methodology
membrane}sample interactions may occur. Fortu- and Detection
nately, the wide range of membranes available has
Decontamination and Sterilization of FFF Separator
made the technique very versatile. The in-channel
pressure needed to create the transverse Sow-gener- This is necessary to prevent bacterial contamination,
ated Reld by means of off-channel restrictors has led but the use of sodium azide should be avoided as it is
to rather sophisticated fraction collection devices. a cellular toxic. Safe and effective decontamination pro-
The third group uses electric Relds either as the cesses can be considered as an additional step in the
main transverse Reld (electric FFF) or as dielectric cleaning procedure.
Relds combined with gravity (DEP-GFFF) to obtain
a hyperlayer focusing elution mode, as shown in
Figure 2E. The fourth group encompasses continuous
separation devices based on the general SPLITT con-
cept, as shown in Figure 2F.
General Features Linked to Cell Elution Process
The osmotic and pH properties of the biological
media must be preserved. Buffer solutions of known
pH and ionic strength are used. For example the
classical isotonic phosphate buffered saline (PBS)
solution. Compared with other FFF elution carried
phases, only limited surfactants can be used. Sucrose
solution and albumin are both possibilities, the latter
being the most versatile carrier phase modiRer, usu-
ally at a concentration of 0.1% (v/w). The surfactant
effect of albumin limits particle}particle interactions
as well as particle}wall ones. Albumin adsorption
occurs, on relatively hydrophobic channel wall ma-
terials, leading to channel ageing or poisoning. Rigor-
ous cleaning and channel regeneration procedures are
therefore recommended.
As the general purpose of cell elution in FFF is to
limit particle}wall interactions, the commonly used
‘stop Sow injection procedure’ can cause problems.
The procedure involves injecting sample at the inlet of
the channel in the absence of a Sowing stream but
while an external Reld is applied. The species are
Figure 3 Red blood cells with different elution characteristics in
concentrated near the accumulation wall at the begin- sedimentation field flow fractionation. (A) Fractograms (elution
ning of the channel. The procedure is essential if signal) of fresh RBCs. Channel walls made of hydrophobic mater-
species are eluted according to the ‘Brownian’ elution ials, channel dimensions 0.5 cm wide, 0.025 cm thick, 58 cm long.
mode, but this is not the case for cells. With cells the The channel distance from rotor axis, 10.7 cm. Photometric de-
tection at 254 nm. Sample injection in the established flow (flow
risk of generating particle}particle or particle}wall
rate of 0.5 mL min\1). (B) Fractograms showing retention differ-
interactions is increased. SpeciRc instrumental modi- ences between fresh RBCs and duracytes. RBCs and duracytes
Rcations of the inlet geometry (directly on the accu- are similar in size, but duracytes are rigid fixed RBCs of higher
mulation wall) or low Sow-injection procedures will density. FC, fresh red blood cell; DC, duracytes.
III / CHELATING ION EXCHANGE RESINS 2271
The cleaning procedure involves Sushing the chan- should be chosen. For large sample volume prepara-
nel (ten times the void volume) with a hypo-osmotic tions, SPLITT techniques are useful, whatever the
carrier phase to destroy the cells, then with a classical external Reld. The major technical goal in cell separ-
deproteinization agent (ten times the void volume) to ation is not the set-up of the separator, but the con-
desorb and destroy the cell and surfactant proteins struction of a chain of knowledge. The separator is
adsorbed in the FFF separator. Finally the system is designed according to the sample characteristics as
Sushed again with the hypo-osmostic carrier phase. well as the mode of operation. Cleaning and steriliz-
Decontamination is performed by allowing a hypo- ation procedures, and detection and viability proced-
chloride solution (3}63C) to Sow through the whole ures, are linked to the nature of the material to be
FFF system. The volume used should be three to four puriRed. This ‘biotechnological’ process is of major
times the void volume of the FFF device. Prior to injec- importance for FFF separations in the life sciences.
tion the channel should be Sushed with ethanol and
rinsed with the sterile mobile phase. Decontamination See also: II/Particle Size Separation: Field Flow Frac-
and sterilization can be simultaneous if a 53C hypo- tionation: Electric Fields; Theory and Instrumentation of
chloride solution is used with a 703C ethanol solution. Field Flow Fractionation. III/Colloids: Field Flow Frac-
tionation. Polymers: Field Flow Fractionation. Proteins:
Cell Detection, Viability and Recovery Field Flow Fractionation. III / Catalyst studies: Chromato-
graphy: Isolation: Magnetic Techniques.
At the outlet of the channel it is necessary to obtain
eluted material that retains its integrity, that is the
diagnosis of the whole particle. Photometric devices Further Reading
operated in the light scattering mode can be used to Bernard A, Paulet B, Colin V and Cardot PhJP (1995) Red
follow elution. After fraction collection off-line ana- blood cell separations by gravitational Reld Sow frac-
lyses methods such as microscopy, granulometric or tionation: instrumentation and applications. Trends
Sow cytometry analyses are recommended. Cell spe- Anal. Chem. 14(6): 266d273.
ciRc staining is also possible. Cell viability can be Caldwell KD, Cheng ZQ and Hradecky P (1984) Separ-
diagnosed by means of speciRc tests, and recovery ation of human and animal cells by steric Reld Sow
must also be clearly deRned. fractionation. Cell Biophysics 6: 233d251.
Giddings JC (1993) Field Sow fractionation analysis of
macromolecular, colloidal and particulate materials.
Conclusions Science 260: 1456d1465.
Nilson M, Kallio PT, Bailey JE, Bulow L and Wahlund KG
The choice of FFF separation techniques for puriRca- (1999) Expression of Vitreoscilla hemoglobin in Es-
tion of cell or cell organelle populations is inSuenced cherichia coli enhances ribosome and tRNA levels:
by their possible elution modes. Micron sized species a Sow Reld Sow fractionation study. Biotechnology Pro-
eluted under the steric hyperlayer model can be separ- gress 15(2): 158d163.
ated using sedimentation techniques if their density Williams PS, Zborowski M and Chalmers JJ (1999) Flow
rate optimization for the quadrupole magnetic cell
differs from that of the carrier phase medium. This is
sorter. Analytical Chemistry 71: 3799}3807.
also possible for ‘nonbuoyant’ sub-micron sized cell
Yang J, Huang Y, Wang XB, Becker FF and Gascoyne PRC
organelle species eluted according to the ‘Brownian’ (1999) Cell separation on microfabricated electrode us-
elution mode. ing dielectrophoretic/gravitational Reld-Sow fractiona-
If species are to be separated according to their size, tion. Analytical Chemistry 71: 911d918.
FFF techniques that use a Sow generated external Yue V, Kowal R, Neargarder L et al. (1994) Miniature
Reld are preferred. If differences in surface character- Reld-Sow fractionation system for analysis of blood
istics are important, electrical-based FFF techniques cells. Clinical Chemistry 40(9): 1810d1814.
CHELATING ION EXCHANGE RESINS
R. J. Eldridge, CSIRO Molecular Science, Introduction

Clayton South, Victoria, Australia
The useful role of ion exchange is to separate a solu-
Copyright ^ 2000 Academic Press tion at a low initial concentration Ci into a small
2272 III / CHELATING ION EXCHANGE RESINS
volume of concentrate and a much larger volume of one cation from a mixture to be selectively concen-
depleted solution. In favourable cases the concentra- trated by a large factor.
tion Ce of the target species in the regeneration Weak acid resins tend to be more selective than
efSuent approaches the regenerant concentration Cr. strong acid. Table 1 also lists some selectivity coefR-
The concentration factor achieved is then given ap- cients (not constants), deRned by:
proximately by:
K"MM ) m2Na /M2Na ) mM [6]
R,Ce/Ci+Cr/Ci [1] n#
for adsorption of divalent cations M on a carboxyl
One usually wants to separate a single valuable resin initially in the Na# form. Here MM is the molal
species from the other constituents of the initial solu- concentration of Mn# in the resin phase and mM its
tion, and the ion exchanger must then be selective for molality in the solution phase. Although K and K are
the target species. (Separation can also be achieved only roughly comparable, the carboxyl resin is clearly
by selective desorption, but selective adsorption is more selective among these four divalent cations.
preferable because more of the available resin capa-
city is occupied by the target ion.) The afRnity of
Chelating Resins
a resin for a single ion M, under given conditions,
can be represented by a distribution coefRcient, Ion exchangers incorporating chelating groups have
deRned by: higher selectivity among cations than conventional
strong or weak acid resins. (In one sense, adsorption
amount of M on unit mass of resin of a di- or trivalent cation by any resin is a chelation
D, [2]
amount of M in unit volume of solution process, since the ion is coordinated by two or more
ligands linked by a chain of covalent bonds, but the
Selectivity between two ions can be expressed by term chelating resin is more usually reserved for
a separation factor, deRned by: resins containing discrete chelating groups, each at-
,CTe ) CdU/CTd ) CeU [3] tached to a single monomer unit. A given cation may
be coordinated by one or more of these chelating
where CT represents the concentration of the target units.) Technical aspects of their use were outlined by
and CU that of an unwanted species, while Ce refers to Waitz in 1979. A review and a monograph empha-
the regeneration efSuent and Cd to the depleted solu- sizing analytical applications were published by
tion. Warshawsky and Marhol, respectively, in 1982.
Common strong acid resins show a small degree A comprehensive review by Sahni and Reedijk ap-
of selectivity between singly and doubly charged peared in 1984, and a shorter update by Beauvais and
cations, and even within these categories. Table 1 Alexandratos in 1998.
shows some selectivity values from the literature for Resins with an enormous variety of chelating
a typical strong acid cation exchanger. These are ligands have been synthesized in the laboratory, but
rational equilibrium constants for the reaction: few have been manufactured commercially. Essential-
nRSO3Na#Mn#P(RSO\ n#
#n Na# [4] ly any complexing agent used in analytical chemistry
3 )nM
can be coupled to a resin, although the challenge to
deRned by: the synthetic chemist is to avoid sacriRcing ligand
groups such as thiol or primary amino groups in
K,AM ) anNa/AnNa ) aM [5]
the coupling reaction. Many published syntheses
where AM is the activity (on the rational scale) of have been too complex to be economic on an indus-
Mn# bound to the resin phase and aM the activity trial scale or have yielded resins of low stability, but
(molal scale) of the salt MXn in the solution phase. new resin structures continue to be reported. In
A and a are the activities of the analogous sodium this review we can consider only a few illustrative
, ,
species. The observed values are too small to allow examples.
Table 1 Selectivities of strong and weak acid cation exchangersa
Ion Na# K# Mg 2# Ca 2# Zn 2# Ni 2# Cu 2# Pb 2#
K 1.00 1.5 4.3 11 4.8 6.1 5.9 39

K 1.00 14 98 38 550
a
Rational selectivities K of a sulfonated polystyrene resin (8% cross-linked); molal selectivities K of a methacrylic acid resin (5%
cross-linked, degree of ionization 0.85, background electrolyte 1 mol L\1 NaNO3), relative to Na#.
Table 2 Some commercial chelating resins
Chelating group Examples
Amidoxime Duolite ES 346a, Diaion CR-50b

Aminophosphonate Duolite ES 467, Lewatit OC 1060c
Iminodiacetate Dowex A-1d, Diaion CR-10, Figure 1 Chelating functionalities: (A) iminodiacetic acid and
Duolite ES 466, Lewatit TP 208 (B) aminophosphonic acid.
Diphosphonate Diphonixe
Bis(2-picolyl)amine Dow 3N (Dowex XFS 4195)
2-Picolyl-2- Dow 2N (Dowex XFS 43084) resins. Adsorption of all metal cations decreases with
hydroxypropylamine decreasing pH, and uptake of moderately preferred
Oligoamine Diaion CR-20, Lewatit E 304, ions such as Ni2# and Zn2# becomes negligible at
Sumichelate MC10f pH 1}2, but Hg2# and Cu2# are still strongly adsor-
a
bed. For Hg2# in sulfuric acid, D+2 L g\1 at pH 1.
Rohm and Haas; b Mitsubishi Chemical; c Bayer; d Dow Chem-
ical; e Eichrom Industries; f Sumitomo Chemical. The advantage of this high selectivity is illustrated
by some results for treatment of a highly acidic,
mercury-containing smelter wastewater. Simulated
Early chelating resins based on phenolic polymers wastewater containing Hg(II), Zn(II), Fe(III), Cd(II)
were largely displaced by styrene-divinylbenzene and Pb(II) salts at pH 1.5 was treated with an
resins, functionalized via chloromethylation, because iminodiacetate resin in bench-scale column runs.
of their greater stability. More recently, acrylic poly- Table 4 and Figure 2 show the composition of the
mers have gained ground, with glycidyl methacrylate column efSuent over 157 bed volumes. Uptake of
being a favoured monomer because ligands are easily Hg(II) was close to 100%, while most of the other
coupled by reaction with its epoxide group, although cations were not signiRcantly adsorbed. Adsorption
possible hydrolysis of the ester link at extremes of pH of mercury can be represented by:
limits the applicability of these resins. Sherrington,
Driessen and their co-workers have used this strategy RN(CH2COOH)2#Hg2#
to prepare a suite of chelating resins containing nitro- PRN(CH2COO\)2Hg2##2H# [7]
gen heterocycles. Similarly, Chanda and Rempel have
coupled a wide range of chelating groups to polyben- It appears that a small amount of Pb(II) is adsorbed
zimidazole beads activated with epichlorohydrin. initially, then displaced by Hg(II). Fe(III) is adsorbed
Table 2 lists some commercially available chelat- more strongly than Pb(II), but still less strongly than
ing resins. The order of preference for iminodiacetate Hg(II). Uptake of iron can essentially be eliminated
resins is typically: by reducing Fe(III) to Fe(II), which is not signiRcantly
adsorbed at this low pH.
Univalent cations: Ag'Li'Na'K'Rb'Cs
Divalent cations: Hg'U(VI)'Cu'Pb'Ni
'Cd+Zn'Co'Fe'Mn Regeneration with Complexation
'Ca'Mg'Ba'Sr
Chelating resins are commonly regenerated with min-
Trivalent cations: Cr'In'Fe'Ce'Al'La
eral acid, but in some cases the resin’s afRnity for the
Structures for the iminodiacetate and amino- target metal is too great for this to succeed. In the
phosphonate groups are shown in Figure 1. Some Hg(II)-iminodiacetate system desorption with nitric
selectivity coefRcients (no deRnition given) for acid failed because of the very strong complexation of
iminodiacetate resins, recalculated from Waitz, are Hg(II). Concentrated NaCl solution was an effective
collected in Table 3. regenerant because Cl\ ions convert Hg2# to a series
of chlorocomplexes (Figure 3). In other words, Hg2#
is adsorbed more strongly than H#, but can be dis-
Iminodiacetate Resins: Mercury placed even by Na# because it is simultaneously
Iminodiacetate resins are clearly more selective complexed by Cl\. In this example, both adsorption
among the transition metals than are weak acid and desorption steps are selective for Hg(II) over
Table 3 Selectivities at pH 4 of an iminodiacetate chelating resin, relative to Ca2#
Ion Ca 2# Fe 2# Co 2# Cd 2# Zn 2# Ni 2# Pb 2# Cu 2# Hg 2#
K 1.00 4.0 6.7 15 17 57 1200 2300 2800

Table 4 Selective removal of Hg(II) on a chelating resina
Hg(II) Zn(II) Fe(III) Cd(II) Pb(II)
mg L\1 % mg L\1 % mg L\1 % mg L\1 % mg L\1 %
Feed 11.8 100 78 100 37 100 1.7 100 4.6 100

Bed volumes
23 (0.015 (0.1 73 94 1 3 1.6 94 4.0 87
47 0.015 0.1 74 95 7 19 1.7 100 5.5 120
77 0.015 0.1 78 100 14 38 1.7 100 5.0 109
111 0.030 0.3 78 100 17 46 1.7 100 4.7 102
157 0.035 0.3 76 97 28 76 1.7 100 4.9 107
a
Metal concentrations before and after treatment of a simulated smelter wastewater (pH 1.5) with an iminodiacetate resin.
Reproduced with permission from Becker NSC and Eldridge RJ (1993) Reactive Polymers 21: 5.
most elements, although Fe(III) shares the ability of for aminophosphonate resins Mg Na +1000 and
Hg(II) to adsorb as a cation and desorb as a chloro- Ca
Na +20 000. These resins are very effective in soften-
complex. The regeneration reaction can be written: ing applications, including, as an extreme case, the
removal of low levels of unwanted divalent
RN(CH2COO\)2Hg2##2Na##2Cl\
cations from brine in caustic chlorine plants, despite
PRN(CH2COO\Na#)2#HgCl2 [8] the great excess of sodium ions.
Alkaline Earths Amidoxime Resins: Uranium from

The formation of chelate complexes is often regarded
Seawater
as an exclusive property of transition metal cations, The possibility of recovering valuable metals from the
but both iminodiacetate and aminophosphonic resins oceans presents a challenge to designers of chelating
chelate alkaline earth cations and consequently resins. Uranium is present in seawater at about
are highly selective for them relative to the alkali 14 nmol L\1 (3 g L\1), predominantly as the tricar-
metal cations. Separation factors achievable with bonatouranate(VI) anion UO2(CO3)43\. Recovery
iminodiacetate resins are reported to be 50}60 by ion exchange would require an extremely selec-
for Mg/Na and 500}600 for Ca/Na at pH 7}10; tive sorbent, since seawater also contains Na# at
Figure 2 Separation of metal cations on an iminodiacetate resin at pH 1.5. Triangles, Pb(II); open diamonds, Cd(II); squares,
Zn(II); circles, Fe(III); filled diamonds, Hg(II). (Reproduced with permission from Becker NSC and Eldridge RJ (1993) Reactive
Polymers 21: 5.)
Figure 3 Regeneration of a chelating resin may require a complexing eluant: desorption of Hg(II) from an iminodiacetate resin with
3 mol L\1 sodium chloride (circles) or 2 mol L\1 nitric acid (squares). (Reproduced with permission from Becker NSC and Eldridge RJ
(1993) Reactive Polymers 21: 5.)
0.6 mol L\1, Mg2# at 0.4 mol L\1 and transition In one experiment with unspiked seawater, an ex-
metal ions such as Fe3#, Cu2# and Zn2#, at 10} perimental amidoxime resin adsorbed 1 mg g\1 ura-
100 nmol L\1. Iminodiacetate resins fail to adsorb nium (D+300 L g\1). Fe3#, Cu2# and Zn2# were
signiRcant amounts of uranium from seawater, since concentrated by similar factors. Isolation of pure ura-
their UO2#2 /Mg
2#
selectivity coefRcient is only nium would therefore need selective regeneration or
&600. During the 1980s, amidoxime resins derived post-treatment of the regeneration efSuent.
from polyacrylonitrile (Figure 4) were extensively in- Amidoxime resins are also highly selective for
vestigated by several research groups, particularly in Ga(III) at high pH and have been used to recover
Europe and Japan, and shown to adsorb up to about gallium from Bayer liquor. Again, capacity loss is
1 mg U g\1 resin from spiked seawater. Distribution a problem.
coefRcients decrease in the order:
Fe(III)+U(VI)'Cu(II)'Ni(II)'Co(II)'Zn(II) Oligoamine Resins
Ca(II)'Mg(II) Many of the transition metal cations are strongly
coordinated by uncharged nitrogen-containing
from &300 L g\1 for Fe(III) and U(VI) to ligands. Consequently, chelating resins possessing
&20 mL g\1 for Mg(II). UO2# 2 cations are believed strongly electron-donating nitrogens will selectively
to be complexed by two deprotonated amidoxime remove Cu2#, for example, from solutions contain-
groups, which displace the carbonate ligands. Ad- ing alkali and alkaline earth cations. Since the resin is
sorption rates are higher for resins incorporating hy- uncharged, the adsorbed cations must be accom-
drophilic cross-linkers or comonomers. Hydrochloric panied by mobile counterions to maintain elec-
acid at 0.5}1 mol L\1 is an effective regenerant, but troneutrality. SenGupta has exploited this effect to
causes some hydrolysis of the amidoxime groups. adsorb anions such as phosphate, selenite, oxalate
and phthalate on copper-loaded picolylamine resins
(Dow 2N or 3N). These coordinating anions displace
anions, such as sulfate, that are associated with
adsorbed Cu(II) only by electrostatic attraction. Loss
of copper from the resin during ligand exchange is
extremely small.
Even conventional weak base resins } usually used
Figure 4 Synthesis of an amidoxime resin from polyacrylo- as anion exchangers } are capable of separating
nitrile. transition metals from alkaline earths. Uptakes of
Cu(II) have been demonstrated at pH 5 on the phen-

olic weak base resin Duolite A7 and the acrylic weak
base Amberlite IRA35 comparable to their anion ex-
change capacities (1}2 meq g\1). Uptake of Ni(II) or
Zn(II) was smaller. Regeneration with mineral acid
was efRcient. An aminophenol resin of similar struc-
ture to Duolite A7 was found to adsorb Cu(II), pre-
sumably accompanied by sulfate ions, from CuSO4
solutions at pH 4}7. Other transition metal ions Figure 6 Structure of Diphonix威.
(Fe2#, Co2#, Ni2#, Zn2#, Cd2# and Hg2#) were
scarcely adsorbed in this pH range, but in the proto-
nated form (pH+2) the resin became highly selective In pH 1 nitric acid distribution coefRcients on
for Fe3#. Diphonix decrease in the order:
The transition metal afRnity of resins having
Cu(II)'Zn(II)'Co(II)+Ag(I)
aliphatic oligoamine substituents increases with chain
length (number of amino groups in each chelating from &3 L g\1 to &500 mL g\1. For these ions
group). It has been reported that acrylic oligoamine a double-logarithmic plot of the distribution coefRc-
resins (Figure 5) can adsorb signiRcant amounts of ient against [H#] has slope !n, where n is the cation
Cu2#, Ni2#, Zn2# or Cd2# from solutions con- charge. This indicates that adsorption occurs by
taining ethylenediaminetetraacetic acid (EDTA), simple exchange of Mn# for H#. In 1 mol L\1 nitric
provided alkaline earth cations are also present. acid, the order of selectivity is:
Transition metal uptakes were highest on resins
derived from tetraethylenepentamine or penta- Th(IV)'Fe(III)+U(VI)'Np(VI)+Pu(IV)'Bi(III)
ethylenehexamine. +Eu(III)'Am(III)Al(III)'Cr(III)
with D as high as 900 L g\1 for Th(IV) and 200 L g\1
Diphonix for Fe(III) and U(VI), decreasing to 400 mL g\1 for
A similar coordination mechanism operates in the Al(III) and 80 mL g\1 for Cr(III). Uptake of the
adsorption of some metal ions from highly acidic strongly preferred ions shows little pH dependence,
solution on phosphonic acid resins, for example ac- indicating adsorption without displacement of
tinide ions on Diphonix威, one of very few new chelat- H# on phosphonic acid groups that are not signiR-
ing resins to be introduced commercially in the cantly ionized under these conditions. To maintain
1990s. The chelating groups in this resin are contrib- electroneutrality, the complexed metal ions must be
uted by vinylidene-1,1-diphosphonic acid monomer associated with anions from solution, either free or in
units, but the resin also contains acrylic acid and mixed-ligand complexes containing phosphonic acid
styrenesulfonic acid units (Figure 6). As with the groups and small ion ligands such as Cl\. The less
amidoxime resins described above, hydrophilic preferred ions show simple ion exchange behaviour
monomers are essential for high reaction rates. over a limited pH range, but uptake increases
Diphonix is the outcome of work at the Argonne at high acid concentrations, indicating a change in
National Laboratory on 1,1-diphosphonic acid mechanism.
ligands and at the University of Tennessee on ion The very high afRnity of Diphonix for Fe(III) makes
exchangers possessing two or more types of func- it very suitable for removing iron from acidic copper
tional group, operating by different mechanisms. It electrolytes in copper production by electrowinning.
was developed primarily for actinide separations in The process has been demonstrated in pilot-scale
highly acidic media, but also strongly adsorbs trials at copper reRneries, with no decrease in resin
divalent transition metal cations at pH 1}2 and performance over hundreds of cycles. Fe(III) is desor-
trivalent cations at pH 0}2. bed by reducing it to Fe(II).
Circumventing Limitations to
Selectivity
The selectivity between two cations of a chelating
resin should in principle be equal to that of the chelat-
ing ligand itself, but this is seldom true in practice.
Figure 5 Acrylic oligoamine resin. For example, iminodiacetate resins are very much less
Table 5 Comparison of an iminodiacetate resin with the iminodiacetate iona
Ion Ca 2# Fe 2# Co 2# Cd 2# Zn 2# Ni 2# Pb 2# Cu 2# Hg 2#
K 1.00 4.0 6.7 15 17 57 1200 2300 2800

Km 1.00 1600 2.3;104 600 2.8;104 4.7;105 7;104 9;107 1.5;109
a
Selectivities for divalent cations (relative to Ca2#) of an iminodiacetate chelating resin (K ) and the iminodiacetate ion (Km).
selective than the free iminodiacetate ion, as shown in riamine (DETA), for example) can attach to the pre-
Table 5, which compares equilibrium constants for cursor resin (chloromethylated polystyrene-co-
the reaction: divinylbenzene) through either primary or secondary
nitrogens, and can react with multiple chloromethyl
HN(CH2COO\)2Ca2##M2# groups, introducing additional covalent cross-links.
0 HN(CH2COO\)2M2##Ca2# [9] This effect was eliminated in experimental resins: the
primary nitrogens of DETA were blocked by forming
with the corresponding values from Table 3 for Schiff base adducts, followed by amination exclus-
iminodiacetate resins. There are several reasons for ively through the secondary nitrogen and subsequent
the discrepancy. In the Rrst place the resin is hetero- hydrolysis of the Schiff base groups to recover the
geneous on a molecular scale. Its chelating groups are primary amino groups (Figure 7). The latter were
located in environments that may vary in polarity, then converted by further reaction to iminodiacetate,
dielectric constant or steric crowding, depending for aminophosphonate or other chelating groups. The
example on their closeness to cross-links. This vari- selectivity sequence of the bis(iminodiacetate) deriva-
ation in chemical environment affects the binding tive is:
constant for different metal ions to differing extents,
Fe(III)'In(III)'Ga(III)+Cu(II)'Zn(II)'Al(III)
degrading the selectivity of the resin. Secondly, the
high local concentration of ligands in the resin may with distribution coefRcients decreasing from
lead to a mixture of 1 : 1, 2 : 1 and higher complexes. &10 L g\1 at pH 1 for Fe(III) to &10 mL g\1 for
At loadings less than 100%, the number of ligands Al(III) at pH 2. The resin readily separates Ga(III) and
coordinated to each metal ion is almost impossible In(III) from Zn(II) and Al(III) in acid solution, but
either to control or to measure. Thirdly, the exchange Fe(III) and Cu(II) must Rrst be removed (by precipita-
reaction itself changes the properties of the resin. tion as sulRdes, for example). For most trivalent rare
Selectivities of simple ion exchangers are well known earth cations, D+100}500 mL g\1 at pH 2, with the
to vary with the extent of loading, and the same will highest values being observed for the medium atomic
be true of chelating resins, especially when adsorbed
metal ions are coordinated by ligands attached to
different polymer chains, reversibly creating addi-
tional cross-links. The result of these factors is not
merely reduced selectivity of polymeric resins relative
to low molecular weight ligands, but in some cases
the order of selectivity is reversed (Table 5).
The mixed composition and additional cross-link
effects can be eliminated by designing the resin to
accommodate the maximum coordination number of
the cation with a single multidentate ligand. Resins
containing EDTA moieties are an example. These
were synthesized by Takeda et al. at Asahi Chemical
in 1985, with m-divinylbenzene as the essential build-
ing block. Addition of bromine across one double
bond before polymerization allowed the introduction
of an iminodiacetate group on each carbon of the
pendant C2 chains. However, even in these resins,
ligands still experience a range of chemical envi-
ronments. Figure 7 Synthesis of diethylenetriamine resin coupled exclus-
Oligoamine weak base resins suffer an additional ively through the secondary nitrogen, and conversion to a
source of heterogeneity since the amine (diethylenet- bis(iminodiacetic acid) derivative.
Figure 8 Lysine-N ?,N ?-diacetic acid resin.
weight lanthanides (Sm-Dy). The selectivity is great achievable in the gel resin, and adsorption then in-
enough to permit chromatographic separation of volves both electrostatic and complexing interac-
pairs of lanthanides (La/Pr or Nd/Sm). tions.
The effect of inhomogeneity can be minimized by Isothiuronium resins have been used in laboratory-
introducing spacer groups between the chelating and pilot-scale PGM separations from HCl leach
groups and the polymer backbone. A highly selective solutions, with thiourea or thiocyanate as regenerant.
nitrilotriacetate-like resin was synthesized by func- However, new sulfur}nitrogen resins are still being
tionalizing a polystyrene matrix with lysine-N?,N?- reported for heavy metal and PGM applications.
diacetic acid (Figure 8). This resin shows high selec- Polybenzimidazole functionalized with dithiooxam-
tivity for trivalent ions such as Ga(III) and In(III) over ide was found to be strongly selective for Pd(II) and
divalent cations. When loaded with Fe(III) it can be Pt(IV) over Rrst-row transition metals in acidic
used to adsorb As(III) and As(V) oxyanions by ligand solution. Other researchers report similar behaviour
exchange. with resins containing, for example, thiosemicar-
bazide, dithizone, thiadiazole or thiotetrazolium
substituents.
Sulfur Ligands
In the quest for high afRnity and high selectivity,
many sulfur-containing ligands have been investi-
Kinetics
gated, including thiol, thioether and thiocarbonyl Adsorption and desorption rates are an important
groups, as well as sulfur}nitrogen heterocycles. Thiol consideration in ion exchange, as well as equilibrium
resins are available from several manufacturers. Al- behaviour. Complexation rates beneRt from small
though falling outside the above deRnition of chelat- particle size, high porosity and, as noted above, high
ing resins, they can be regarded as the prototype of hydrophilicity. Capturing the rate advantage of small
resins containing thiocarbonyl, thioether or dithiocar- size requires a solution to the problem of handling
boxylate groups, or multiple ligands such as thiol/amino Rne particles. Incorporating a magnetic component is
combinations. Thiol resins have been used for adsorp- one strategy that has been demonstrated. Resins can
tion of mercury(II) and are effective in the presence of also be made in nonconventional formats, such as
competing ligands such as Cl\. However, regeneration graft copolymers or Rne Rbres, to keep diffusion paths
is difRcult, requiring concentrated acid, and the thiol short. A range of Rbrous sorbents was developed
group is prone to oxidation. under the name Fiban, derived from polyacrylonitrile
Sulfur-containing ligands tend to have high afRnity Rbres or polypropylene Rbres grafted with styrene,
for platinum group metals (PGM). Isothiuronium including a chelating material containing imidazoline
resins (Figure 9), Rrst prepared in the 1960s by Koster groups derived from polyacrylonitrile. These sorbents
and Schmuckler and subsequently commercialized in could be used to form Rlaments or nonwoven fabrics.
both gel (SraRon) and macroporous (Monivex) ver- Similarly, others have reported amidoxime sorbents
sions, adsorb PGM from strongly acidic solution and made from polyacrylonitrile Rbre and woven into
require complexing agents for regeneration. Adsorp- cloth for use in uranium recovery, and polyacrylonit-
tion is believed to be by simple ion exchange when the rile-g-glycidyl methacrylate Rbres (Tiopan) function-
functional groups are fully protonated, but this not alized with ligands including thiosemicarbazide, 8-
mercaptoquinoline and 2-aminothiazole.
Conclusion
Ingenious syntheses of chelating resins continue to be
reported. However, large scale applications for such
Figure 9 Isothiuronium group. materials are limited, and few are likely to be produced
III / CHEMICAL WARFARE AGENTS: CHROMATOGRAPHY 2279
industrially. The main markets appear to be in the amidoxime polymers from seawater. Industrial and
production of low volume, high value elements: Engineering Chemistry Research 26: 1970}1977.
precious metals, rare earths with valuable optical prop- Hoffmann H and Martinola F (1988) Selective resins
erties and raw materials for electronic devices. Applica- and special processes for softening water and solutions:
tions in lower value Relds such as waste management or a review. Reactive Polymers 7: 263}272.
Marhol M (1982) Chelating resins and inorganic ion
by-product recovery should beneRt from novel resin
exchangers. In Svehla G (ed.)Wilson and Wilson’s
formats designed for improved resin/liquid contacting. Comprehensive Analytical Chemistry, Vol. XIV, ch. 6,
pp. 377}399. Oxford: Elsevier.
See also: II/Ion Exchange: Historical Development; Naden D and Streat M (eds) (1984) Ion Exchange Techno-
Novel Layered Materials: Non-Phosphates; Novel Layered logy. Chichester: Ellis Horwood.
Materials: Phosphates; Organic Ion Exchanger; Organic Sahni SK and Reedijk J (1984) Coordination chemistry of
Membranes; Theory of Ion Exchange. chelating resins and ion exchangers. Coordination
Chemistry Reviews 59: 1}139.
Further Reading Suzuki TM and Matsunaga H (1991) Metal selective poly-
mer resins for the separation and concentration of rare
Beauvais RA and Alexandratos SD (1998) Polymer-sup- metals. Trends in Inorganic Chemistry 2: 33}47.
ported reagents for the selective complexation of metal Warshawsky A (1982) Selective ion exchange polymers.
ions: an overview. Reactive and Functional Polymers Die Angewandte Makromolekulare Chemie 109/110:
36: 113}123. 171}196.
Chanda M and Rempel GL (1990) Polybenzimidazole resin van Berkel PM, Punt M, Koolhaas GJAA, Driessen WL,
based new chelating agents. Palladium(II) and plati- Reedijk J and Sherrington DC (1997) Highly copper(II)-
num(IV) sorption on resin with immobilized selective chelating ion-exchange resins based on
dithiooxamide. Reactive Polymers 12: 83}94. bis(imidazole)-modiRed glycidyl methacrylate co-
Chiarizia R, Horwitz EP, Alexandratos SD and Gula MJ polymers. Reactive and Functional Polymers 32: 139}151.
(1997) Diphonix威 resin: a review of its properties and Zhao D, SenGupta AK and Zhu Y (1995) Trace con-
applications. Separation Science and Technology 32: 1}35. taminant sorption through polymeric ligand exchange.
Hirotsu T, Katoh S, Sugasaka K, Takai N, Seno M and Industrial and Engineering Chemistry Research 34:
Itagaki T (1987) Adsorption of uranium on cross-linked 2676}2684.
CHEMICAL WARFARE AGENTS:

CHROMATOGRAPHY
P. A. D’Agostino, Medicine Hat, Alberta, ance-monitoring regime involving site inspections
Canada was built into the CWC to ensure that the treaty
Copyright ^ 2000 Academic Press remains veriRable. The Organisation for the Prohibi-
tion of Chemical Weapons, or OPCW, based in The
Hague, has responsibility for implementation of the
Introduction treaty. Routine OPCW inspections have or will take
The development and application of analytical place at declared sites, including small scale produc-
methods for the analysis of chemical warfare agents tion, storage and destruction sites, and challenge
have received increased attention, due in large part to inspections will take place at sites suspected of non-
the recently ratiRed Convention on the Prohibition of compliance.
the Development, Production, Stockpiling and Use of An analytical capability will be required to help
Chemical Weapons and their Destruction (commonly verify compliance with the treaty, since inspectors
referred to as the Chemical Weapons Convention or will have the option to take and analyse suspect
CWC). After considerable effort the CWC was samples to help establish compliance or noncom-
opened to signature in 1993, with the treaty coming pliance. Gas chromatography is the current method
into force on 29 April 1997. The treaty has been of choice for the separation and analysis of chemical
ratiRed by 120 states, all of which agree not to devel- warfare agents, with this technique being employed in
op, produce, stockpile, transfer or use chemical the Reld-portable gas chromatographic}mass spectro-
weapons and agree to destroy their own chemical metric (GC-MS) instrumentation in use by the OPCW
weapons and production facilities. A strong compli- inspectorate.
2280 III / CHEMICAL WARFARE AGENTS: CHROMATOGRAPHY
Chemical warfare agents are a group of toxic high casualties, remains in riot control and for the
chemicals that have been deRned in the CWC as ‘any training of military personnel in chemical defence.
chemical which through its chemical effect on life Incapacitating agents have been included as the USA
processes can cause death, temporary incapacitation did develop an agent in this category.
or permanent harm to humans or animals2’. The compounds listed in Table 1 represent the
Poisonous or toxic compounds have been utilized in most common chemical warfare agents, grouped by
an effort to gain military superiority throughout his- category with their Chemical Abstracts registry num-
tory but it is only during the past century that chem- bers, and is not intended to be exhaustive. It has been
ical warfare agents have been produced and used on estimated that more than 7000 compounds are con-
a large scale. Tear gas grenades were used in 1914 by trolled under the CWC, although in practical terms
the French at the outbreak of World War I, but it was the actual number of chemical warfare agents, pre-
not until the Germans Rrst used chlorine near Ypres in cursors, degradation products and related com-
1915 that the world entered the modern era of chem- pounds that are contained in the OPCW GC-MS
ical warfare. Other chemical warfare agents such as database is in the hundreds. The structures of some
phosgene and mustard were used by both sides in common nerve and blister chemical warfare agents
World War I. are illustrated in Figure 1.
The use and development of chemical warfare
agents continued after World War I despite the sign- Gas Chromatographic Methods
ing of the 1925 Geneva Protocol, which bans the Rrst
use of chemical weapons. Mustard was used by the Chemical warfare agents have often been referred to
Italians against the Abyssinians (Ethiopia) during as warfare gases and, in the military, the phrase ‘gas,
1936}1937, and just prior to World War II, the gas, gas’ has become synonymous with attack by
Germans discovered and produced the Rrst nerve chemical warfare agents. In fact, many chemical war-
agent, tabun. Nerve agents were made into weapons fare agents exist as liquids at ambient temperatures
by the Germans but neither side made use of their but have varying degrees of volatility and pose a sig-
chemical weapons stock. More effective nerve agents, niRcant vapour hazard as well as a liquid contact
such as VX, were developed in the 1950s, mustard hazard. This physical characteristic has made the
was used in the Yemen War (1963}1967), and allega- analysis of chemical warfare agents amenable to gas
tions of chemical warfare agent use were reported in chromatography (GC) with a variety of detectors
South-East Asian conSicts. Nerve and mustard agents including mass spectrometry (MS).
were used by Iraq in the 1980s war between Iran and The OPCW, an important end-user of analytical
Iraq, and were considered a real threat to United techniques for chemical warfare agents, requires the
Nations armed forces during their action against Iraq use of two or more spectrometric techniques and the
(1990}1991). More recently, sarin was used against availability of authentic reference standards for the
the population of a Kurdish village in 1993 and again unambiguous identiRcation of these controlled com-
in 1995 by terrorists in the Tokyo underground tran- pounds. For this reason, the combined use of
sit system. Proliferation of chemical weapons and GC}Fourier transform infrared spectrometry (FTIR)
their use will hopefully decrease over the coming has received increased attention as newer technolo-
years as the CWC proceeds towards its goal of world- gies have led to detection limits approaching those
wide chemical weapons destruction. routinely reported during GC-MS analysis. For ana-
lyses involving low levels of chemical warfare agents
in the presence of high levels of interfering chemical
Chemical Warfare Agent Categories background, tandem mass spectrometry (MS-MS) is
Chemical warfare agents have been classiRed into receiving increased attention.
nerve, blister, choking, vomiting, blood, tear and
GC Retention Behaviour
incapacitating agent categories based on their effect
on humans. The most signiRcant chemical warfare Packed column GC was routinely employed in the
agents in terms of military capacity and past use are past for the analysis of chemical warfare agents, but
the nerve and blister agents. For these reasons the with the advent of fused silica capillaries this techno-
analysis of these compounds will be emphasized over logy has been used less frequently. Capillary column
the other groups. The choking, blood and vomiting GC has become the most frequently employed
agents are for the most part obsolete chemical agents analytical separation method for the screening of
that were employed during World War I. The tear samples contaminated with chemical warfare agents.
agents were used during the Vietnam War but their Separation of chemical warfare agents may be
primary use, because of their inability to produce achieved with fused silica columns coated with poly-
Table 1 Common chemical warfare agents
Full name (trival name(s)) Chemical Abstracts no.
(a) Nerve (reacts irreversibly with cholinesterase; this results in acetylcholine accumulation,
continual stimulation of the body’s nervous system and eventual death)
O-Isopropyl methylphosphonofluoridate (sarin, GB) 107-44-8
O-Pinacolyl methylphosphonofluoridate (soman, GD) 96-64-0
O-Cyclohexyl methylphosphonofluoridate (GF) 329-99-7
O-Ethyl N,N-dimethylphosphoramidocyanidate (tabun, GA) 77-81-6
O-Ethyl S-2-diisopropylaminoethyl methylphosphonothiolate (VX) 50782-69-9
(b) Blister (affects the lungs, eyes and produces skin blistering)
Bis(2-chloroethyl)sulfide (mustard, H) 505-60-2
Bis(2-chloroethylthio)ethane (sesquimustard, Q) 3563-36-8
Bis(2-chloroethylthioethyl)ether (T) 63918-89-8
Tris(2-chloroethyl)amine (HN-3) 555-77-1
2-Chlorovinyldichloroarsine (lewisite, L) 541-25-3
(c) Choking (affects respiratory tract and lungs)

Chlorine 7782-50-5
Carbonyl dichloride (phosgene, CG) 75-44-5
(d) Vomiting (causes acute pain, nausea and vomiting in victims)

Diphenylarsinous chloride (DA) 712-48-1
10-Chloro-5,10-dihydrophenarsazine (adamsite, DM) 578-94-9
Diphenylarsinous cyanide (DC) 23525-22-6
(e) Blood (prevents transfer of oxygen to the body’s tissues)

Hydrogen cyanide (HCN, AC) 74-90-8
(F) Tear (Causes tearing and irritation of the skin)

[(2-chlorophenyl)methylene]propanedinitrile (CS) 2698-41-1
2-Chloro-1-phenylethanone (CN) 532-27-4
Dibenz[b,f][1,4]oxazepin (CR) 257-07-8
(g) Incapacitating (prevents normal activity by producing mental or physiological effects)

3-Quinuclidinyl benzilate (BZ) 6581-06-2
siloxane or other Rlms and retention index data rela-

tive to n-alkanes and n-alkylbis(triSuoromethyl)phos-
phine sulRdes (M-series) have been reported for many
chemical warfare agents and related compounds un-
der temperature programming conditions. Retention
indices relative to n-alkanes have been reported
for 100% dimethyl-polysiloxane (e.g. J&W DB-1),
(95%)-methyl-(5%)-diphenyl-polysiloxane (e.g. J&W
DB-5), (86%)-dimethyl-(14%)-cyanopropylphenyl-
polysiloxane (e.g. J&W DB-1701) and other Rlms. In
general, the best separations have been achieved with
a moderately polar Rlm such as (86%)-dimethyl-
(14%)-cyanopropylphenyl-polysiloxane. Table 2 lists
typical retention index data for several common
chemical warfare agents on three different liquid
phases under temperature programming conditions.
The use of retention indices relative to n-alkanes by
the OPCW during on-site inspections is anticipated in
the near future to differentiate between controlled
compounds that exhibit similar electron impact mass
Figure 1 Structures of common chemical warfare agents. spectrometric (EI-MS) data during GC-MS analysis,
Table 2 GC retention index data for common chemical warfare thioethyl)ether (T) by weight, while HQ is usually
agents (relative to n-alkanes) 75% distilled mustard and 25% sesquimustard (Q)
by weight. HS is crude mustard-containing 15% car-
Compound name GC retention index a
bon tetrachloride. Munitions grade samples typically
DB-1 DB-5 DB-1701 contain additional sample components that may pro-
vide synthetic procedure or source information.
O-Isopropyl methylphosphono- These samples contained several other sulfur-contain-
fluoridate (sarin, GB) 792 824 966
ing impurities, including 1,4-thioxane, 1,4-dithiane
O-Pinacolyl methylphosphono- 1008 1045 1188
fluoridate (soman, GD)b 1013 1049 1193 (two common mustard degradation products) and
several longer chain blister agents (chromatographic
O-Ethyl N,N-dimethylphosphor- peaks 8, 9 and 10 in Figure 2).
amidocyanidate (tabun, GA) 1078 1132 1340 Figure 3 illustrates the application of capillary GC-
O-Ethyl S-2-diisopropylamino- FID for the characterization of tabun and related
ethyl methylphosphono-
thiolate (VX) 1664 1710 1881 impurities in a munitions grade sample used for mili-
Bis(2-chloroethyl)sulfide tary chemical agent detector evaluation. The volatile
(mustard, H) 1124 1173 1326 organic content of this munitions grade tabun
Bis(2-chloroethylthio)ethane was estimated on the basis of the FID response of
(sesquimustard, Q) 1623 1689 1923
the individual components. Tabun accounted for
Bis(2-chloroethylthioethyl)-
ether (T) 1910 1983 2241 81% of the sample. The impurities, isopropyl N,N-
dimethylphosphoramidocyanidate (isopropyl analogue
a
GC retention indices determined with three J&W 0.25 m films of tabun), ethyl N,N,N,N-tetramethylphosphoro-
with the following temperature program: 503C (2 min), then diamidate, N,N,N,N-tetramethylphosphorodiamidic
103C min\1 to 3003C (5 min).
b
cyanide and the cluster of pyrophosphates, represent-
Retention data for both enantiomer pairs.
ed approximately 5%, 4%, 2% and 5%.
but different GC retention behaviour. Application of

GC retention indices during the analysis of VX-con-
taminated samples is likely since the EI-MS data for
VX and a number of VX-related compounds are
remarkably similar. The EI data for these compounds
lack a molecular ion and contain a base ion at m/z
114 due to (CH2N(iPr)2)# and additional ions related
to the !SC2H4N(iPr)2 substituent.
Chiral stationary phases have been developed for
the resolution of stereoisomers of several chiral nerve
agents, most notably soman. The use of multiple
columns of differing polarity during one analysis has
also been successfully employed during chemical war-
fare agent analysis and the term ‘retention spectro-
metry’ was coined to describe this technique.
Conventional GC Detectors
Most of the GC detectors commonly applied to pesti-
cide residue analysis have also been applied to the
screening of samples for chemical warfare agents.
Figure 2 Capillary column GC-FID chromatograms of three
Detection limits are typically in the nanogram to
munitions grade mustard samples: (A) HT, (B) HS and (C) HQ.
picogram range. Flame ionization detection (FID) is Identified compounds include: 1, 1,4-thioxane; 2, 1,4-dithiane; 3,
routinely used for preliminary analyses, as this tech- mustard (H); 4, bis(2-chloroethyl)disulfide; 5, 2-chloroethyl (2-
nique provides a good indication of the complexity chloroethoxy)ethyl sulfide; 6, sesquimustard (Q); 7, bis (2-chloro-
of a sample extract. Figure 2 illustrates GC-FID ethylthioethyl)ether (T); 8, 1,14-dichloro-3,9-dithia-6,12-dioxatet-
radecane; 9, 1,14-dichloro-3,6,12-trithia-9-oxatetradecane; and
chromatograms obtained using a J&W DB-1 capillary
10, 1,16-dichloro-3,9,15-trithia-6,12-dioxaheptadecane. (GC col-
column for three different munitions grade mustard umn: 15 m;0.32 mm ID J&W DB-1, temperature programme:
formulations, HT, HS and HQ. HT is typically 60% 503C (2 min), then 103C min\1 to 2803C (5 min), 2;10\10 A full
distilled mustard (H) and 40% bis(2-chloroethyl- scale.)
Figure 3 Capillary column GC-FID chromatogram of munitions grade tabun (GA) sample. Compounds identified include: 1, tabun
(GA); 2, ethyl isopropyl N,N-dimethylphosphoramidocyanidate; 3, ethyl N,N,N ,N -tetramethylphosphorodiamidate; 4, N,N,N ,N -
tetramethylphosphorodiamidiccyanide; and 5, pyrophosphates. (GC-column: 15 m;0.32 mm ID J&W DB-5, temperature programme:
503C (2 min), then 103C min\1 to 2803C (5 min), 5;10\11 A full scale.)
The need for higher speciRcity and sensitivity has depends on the amount of preparation required to
led to the application of element-speciRc detectors obtain a suitable sample or extract for GC analysis. In
such as Same photometric detection (FPD), thermi- the simplest case where neat liquid can be obtained
onic detection (TID), atomic emission detection the sample requires dilution with a suitable solvent
(AED) and electron capture detection (ECD). The prior to GC-MS analysis. Environmental and other
simultaneous use of FID with one or more element- samples generally require (at a minimum) solvent
speciRc detectors has also been demonstrated during extraction and concentration prior to analysis, with
dual or tri-channel GC analysis using conventional an exception being direct thermal desorption analysis
and thermal desorption sample introduction. While of samples using a GC equipped with a thermal de-
these detectors may provide strong collaborative sorption device that may be loaded with the actual
evidence for the presence of chemical warfare agents, sample. Recommended methods for sample prepara-
they cannot be used for full conRrmation. Use of GC tion have been published by The Ministry of Foreign
with one or more spectrometric technique such as MS Affairs of Finland as part of their contribution to
is required to conRrm the presence of chemical war- Chemical Disarmament.
fare agents. Figure 4 illustrates the capillary column GC-MS
chromatogram obtained for the extract of a soil
Mass Spectrometry
sample containing VX and several related com-
Mass spectrometry is the method of choice for the pounds. Seven components were identiRed in the
detection and characterization of chemical warfare sample following ultrasonic extraction of 1 g of soil
agents, their precursors, degradation products and with dichloromethane. Extraction of chemical war-
related compounds. Extensive use has been made of fare agents from biological samples is generally more
GC-MS and the mass spectra of numerous chemical difRcult. Bonierbale, Debordes and Coppet recently
warfare agents and related compounds have been demonstrated a method for the extraction of VX from
published, with the most common chemical warfare rat blood as part of a study designed to follow the
agent mass spectra being available in the OPCW, hydrolysis of VX. Figure 5 illustrates a typical mass
commercial or defence community databases. chromatogram for the blood extract containing VX,
Samples collected for chemical warfare agent analy- malathion and diisopropylaminoethanethiol.
sis typically fall into one of the following general The data acquired in Figures 4 and 5 and most MS
categories: (a) munitions or munition fragments (e.g. data obtained by OPCW inspectors during inspec-
neat liquid or artillery shell casing); (b) environmental tions have been obtained under EI-MS conditions.
(e.g. soil, water, vegetation or air samples); (c) artiR- However, many of the chemical warfare agents in
cial materials (e.g. paint or rubber); and (d) biological particular the organophosphorus nerve agents such as
media (e.g. blood or urine). The ease of analysis VX, and the longer chain blister agents related to
molecular ion data, but also affords a high degree of

speciRcity as less basic sample components are not
ionized by the ammonium ion.
Capillary column GC-MS-MS offers the analyst
the potential for highly speciRc, sensitive detection of
chemical warfare agents as this technique signiR-
cantly reduces the chemical noise associated with
complex sample extracts. The speciRcity of product
scanning with moderate sector resolution, as well as
the speciRcity of ammonia CI, were demonstrated
with a hybrid tandem mass spectrometer during anal-
ysis of painted panel samples circulated during an
international round robin veriRcation exercise.
The painted panel extract was contaminated with
numerous hydrocarbons. Only two of the three lon-
ger chain blister agents, sesquimustard (Q) and bis(2-
chloroethylthioethyl)ether (T), could be identiRed
during capillary column GC-MS (EI) analysis (Fig-
ure 6A). The arrow indicates the chromatographic
retention time of the third blister agent, 2-chloroethyl
(2-chloroethoxy)ethyl sulRde (O). The speciRcity of
Figure 4 Capillary column GC-MS (El) chromatogram of dich- ammonia CI (Figure 6B) was clearly demonstrated
loromethane extract of soil sample contaminated with VX. VX and during this analysis. All three longer chain blister
six other compounds were identified: 1, diethyl methylphosphon-
agents were determined in the presence of high levels
ate; 2, N,N-diisopropylethylamine; 3, 2-(diisopropylamino)ethyl
methyl sulfide; 4, O,S-ethyl methyl methylphosphonothiolate;
5, O,S-diethyl methylphosphonothiolate; 6, VX; and 7, bis[2-
(diisopropylamino)ethyl] disulfide. (GC column: 15 m;0.32 mm
ID J&W DB-1701, temperature programme: 503C (2 min) then
103C min\1 to 2803C (5 min).)
mustard, such as Q and T, do not provide molecular

ion information during EI-MS. This hinders con-
Rrmation of these compounds and makes identiRca-
tion of novel chemical warfare agents or related im-
purities difRcult.
Considerable effort has been devoted to the use of
chemical ionization (CI) as a complementary ioniz-
ation technique. This milder form of ionization gener-
ally affords molecular ion information for the chem-
ical warfare agents and has been used extensively for
the identiRcation of related compounds or impurities
in chemical warfare agent munitions samples and
environmental sample extracts. The identity of these
related compounds is important because the origin of
samples, synthetic process information or degree of
degradation (weathering) may be deduced based on
this ‘Rngerprint’ data.
Isobutane, ethylene and methane gases were ini-
tially demonstrated as suitable CI gases for the
acquisition of organophosphorus nerve agent CI-MS Figure 5 Capillary column GC-MS (El) chromatogram of
data. More recently, the efRcacy of ammonia CI-MS 374 mol L\1 VX incubated for 2 h in the presence of rat plasma.
Extracted sample components: 1, 3 mol L\1 malathion; 2, VX;
for organophosphorus nerve agents and related com-
and 3, diisopropylaminoethanethiol.(GC column: 25 m;0.32 mm
pounds has been demonstrated and many laborator- ID CP 8CB; temperature programme: 1003C (0.5 min), then
ies now employ this complementary conRrmation 303C min\1 to 2103C (1 min) followed by 303C min\1 to 3003C
technique. Ammonia CI not only offers abundant (3 min).)
Figure 6 Capillary column (A) GC-MS (El), (B) GC-MS (ammonia Cl) and (C) GC-MS/MS (El) chromatograms obtained during
analysis of international round robin painted panel extracts. Sequimustard (Q) and bis(2-chloroethylthioethyl)ether (T) were detected
during El analysis. The downward arrow in (A) indicates the retention time of 2-chloroethyl (2-chloroethoxy)ethyl sulfide (O). This
compound (O) was masked by the organic content of the sample during El analysis and only detected following (B) ammonia Cl and (C)
MS-MS analysis. (GC column: 15 m;0.32 mm ID J&W DB-1701, temperature programme: 403C (2 min), then 103C min\1 to 2803C
(5 min).)
of interfering hydrocarbons, as the hydrocarbons The three longer chain blister agents were well re-
were not sufRciently basic to ionize. Similarly, it was solved with the J&W DB-1701 capillary column, with
possible to use the resolution of hybrid tandem mass all three components exhibiting similar product
spectrometry to discriminate between ions at m/z 123 spectra during GC-MS-MS analysis.
arising from the longer chain blister agents from
Hydrolysis Products of Chemical Warfare Agents
those ions at m/z 123 arising from the hydrocarbon
background. The resultant GC-MS-MS chromato- Both the nerve and blister agents undergo hydrolysis
gram (Figure 6C), where only m/z 123 ions due to the in the environment and methods are required for
blister agents were transmitted into the collisional retrospective detection and conRrmation of these hy-
activated dissociation cell, was virtually free of chem- drolysis products. Hydrolysis products are signiRcant
ical noise and all three components were detected. as they are generally compounds that would not be
routinely detected in environmental samples and their information for the longer chain hydrolysis products,
presence strongly suggests the prior presence of chem- were obtained under ammonia CI conditions.
ical warfare agents. The degradation products of the Use of thermospray mass spectrometry and, more
chemical warfare agents, in particular the nerve recently, the use of atmospheric pressure ionization
agents, are nonvolatile hydrolysis products that must techniques (e.g. electrospray (ESI), ionspray and at-
be derivatized prior to GC analysis. A variety of mospheric pressure CI), has enabled the direct mass
derivatization techniques including methylation, t- spectrometric analysis of the hydrolysis products of
butyldimethylsilylation and trimethylsilylation have chemical warfare agents. Both techniques may be
been investigated to allow GC analysis of, in particu- interfaced to liquid chromatography for component
lar, the organophosphorus acids related to the nerve separation, with thermospray having been largely
agents (e.g. alkyl methylphosphonic acids and superceded by atmospheric pressure ionization (API)
methylphosphonic acid). for most LC-MS applications. Examples of LC-ESI-
Mustard and longer chain blister agents hydrolyse MS methods for the direct analysis of chemical war-
to their corresponding diols, with thiodiglycol being fare agent hydrolysis products are relatively few, and
the product formed following hydrolysis of mustard. only recently has this technique been demonstrated
These compounds may be analysed by GC-MS dir- for the analysis of the nerve agents as well. These new
ectly provided the sample is loaded onto the column LC-ESI-MS methods complement existing GC-MS
using ‘cool’ on-column injection. Figure 7 illustrates methods for the analysis of chemical warfare agents
a typical GC-MS(EI) chromatogram obtained for and their hydrolysis products and will likely replace
products formed following hydrolysis of a sample some GC-MS methods currently in use for the analy-
containing both HT and HQ. The principal hydrolysis of contaminated aqueous samples.
sis products of H, Q and T } thiodiglycol, bis(2-
hydroxyethylthio)ethane, and bis[(2-hydroxyethyl-
thio)ethyl] ether, respectively } are well resolved with
Safety and Disposal
the J&W DB-1701 capillary column. Molecular ion Chemical warfare agents are extremely hazardous
and lethal compounds. They can only be used in
designated laboratories by personnel trained in safe-
handling and decontamination procedures and with
immediate access to medical support. Safety and stan-
dard operating procedures must be developed and
approved before any chemical warfare agents are
handled. Chemical warfare agents can only be used in
laboratory chemical hoods with a minimum face
velocity of 100 linear feet per minute equipped with
emission control devices that limit exhaust concentra-
tion to below 0.0001 mg m\3. Personnel handling
chemical warfare agents should wear rubber gloves,
lab coats and full faceshields, and a respirator (gas
mask) must be kept within easy reach. SufRcient de-
contaminants to destroy all chemical warfare agent
being handled must be on hand before commencing
operations.
Blister agents such as mustard can be destroyed
with 10% hypochlorite or strong bleach solutions.
The nerve agents can be destroyed using saturated
methanolic solutions of sodium or potassium hydrox-
ide. Decontaminated chemical warfare agents must
Figure 7 Capillary column GC-MS(El) chromatogram of the
hydrolysis products of a munitions grade mustard sample contain- be disposed of in an environmentally approved
ing HT and HQ. Eight compounds were identified: 1, 1,4-dithiane; method according to local legislation.
2, mustard (H); 3, hemisulfur mustard; 4, thiodiglycol (hydrolysis
product of H); 5, 2-chloroethyl (2-hydroxyethylthio)ethyl ether;
6, 2-chloroethyl (2-hydroxyethylthio)ethyl sulfide; 7, bis(2-hy- Conclusion
droxyethylthio)ethane (hydrolysis product of Q); and 8, bis[(2-
hydroxyethylthio)ethyl] ether (hydrolysis product of T). (GC Capillary column GC is the chromatographic method
column: 15 m;0.32 mm ID J&W DB-1701, temperature of choice for the separation and analysis of chemical
programme: 403C (2 min), then 103C min\1 to 2803C (5 min).) warfare agents. This technology is employed in the
III / CHIRAL SEPARATIONS / Capillary Electrophoresis 2287
Reld-portable GC-MS instrumentation currently in hydrolysis of the nerve agent VX in rat plasma. Journal
use by the OPCW inspectorate. Use of GC with one of Chromatography B 688: 255d264.
or more spectrometric technique such as mass spec- Compton JAF (1988) Military Chemical and Biological
trometry is required to conRrm the presence of chem- Agents. Caldwell, NJ: The Telford Press.
ical warfare agents. For this reason many analyses are D’Agostino PA (1995) Chemical warfare agents d analysis
and characterization. In: Townshend A and Worsfold PJ
carried out by GC-MS under electron impact or
(eds.), Encyclopedia of Analytical Science, pp. 599d608.
chemical ionization conditions. For analyses involv- London: Academic Press.
ing low levels of chemical warfare agents in the pres- D’Agostino PA, Hancock JR and Provost LR (1999) Packed
ence of high levels of interfering chemical back- capillary liquid chromatography-electrospray-mass
ground, GC-MS-MS is receiving increased attention. spectrometry analysis of organophosphorus chemical
Atmospheric pressure ionization (e.g. electrospray warfare agents. Journal of Chromatography A 840:
(ESI), ionspray and atmospheric pressure CI) tech- 289d294.
niques may be used for the direct mass spectrometric Gander TJ (1996) Jane’s NBC Protective Equipment. Lon-
analysis of the hydrolysis products of chemical war- don: Butler and Tanner Inc.
fare agents in aqueous samples. Examples of LC-ESI- Ivarsson U, Nilsson H and Santesson J (1992) A BrieTng
MS methods for these analyses are relatively few, and Book on Chemical Weapons. Ljungforeytagen Oregro.
Kaipainen A, Kostiainen O and Riekkola M-L (1992) Iden-
only recently has this technique been demonstrated
tiRcation of chemical warfare agents in air samples using
for the analysis of the nerve agents as well. These new capillary column gas chromatography with three simul-
LC-ESI-MS methods complement existing GC-MS taneous detectors. Journal of Microcolumn Separations
methods for the analysis of chemical warfare agents 4: 245d251.
and their hydrolysis products and will likely replace Kientz ChE (1998) Review: Chromatography and mass
some GC-MS methods currently in use for the analy- spectrometry of chemical warfare agents, toxins and
sis of contaminated aqueous samples. related compounds: state of the art and future prospects.
Journal of Chromatography 814: 1d23.
See also: II/Chromatography: Gas: Column Techno- Methodology and Instrumentation for Sampling and Anal-
logy; Detectors: Mass Spectrometry; Detectors: Selective. ysis in the VeriRcation of Chemical Disarmament
Chromatography: Liquid: Detectors: Mass Spectro- (1977d1994) Helsinki: The Ministry of Foreign Affairs
metry. of Finland.
Somani SM (1992) Chemical Warfare Agents. New York:
Further Reading Academic Press.
Witkiewicz Z, Mazurek M and Szulc J (1990) Review:
Bonierbale E, Debordes L and Coppet L (1997) Application Chromatographic analysis of chemical warfare agents.
of capillary column gas chromatography to the study of Journal of Chromatography 503: 293d357.
CHIRAL SEPARATIONS
Amino Acids and Derivatives

See III / AMINO ACIDS AND DERIVATIVES: CHIRAL SEPARATIONS
Capillary Electrophoresis
B. J. Clark, University of Bradford, Bradford, UK ation scientist a very wide range of application
(biopolymers to cations), with ultra-high efRciency. It
Copyright ^ 2000 Academic Press is generally cost-effective, easy in use, with low
sample and buffer requirements and has the capabil-
ity for automation. For these reasons CE has
Background expanded considerably and is now used as an
Since its commercial inception in 1988 capillary elec- orthogonal and complementary technique alongside
trophoresis (CE) has increasingly offered the separ- well-established analytical procedures. Of all its
2288 III / CHIRAL SEPARATIONS / Capillary Electrophoresis
application Relds, the separation of compounds with acts in free solution, some success with constraining
one or more chiral centres has been particularly suc- the chiral selector within the capillary has been
cessful. Analytical chiral separations increased in im- achieved. Two approaches have been considered } at-
portance throughout the 1980s: considerable interest taching the selector to the capillary wall or trapping it
was generated in the resolution of stereoisomers of within a gel (in capillary gel electrophoresis), al-
food additives, agrochemicals, petrochemicals and though in practice, limited applications have been
pharmaceuticals. During this time the analyst was reported so far. The most recent proposal for optical
presented with numerous procedures to examine separations is through electrochromatography, where
chiral compounds. Chromatography and, parti- a chiral stationary phase is packed into the silica
cularly, high performance liquid chromatography capillary and the mobile phase moves under elec-
(HPLC) with immobilized chiral stationary phases troosmotic Sow (EOF).
used as column packings was the most successful for
separation of enantiomeric mixtures, albeit at con-
Chiral Selector in the Buffer Phase
siderable cost.
However, chiral separations by chromatography Ligand Exchange
do not give a unique solution to all difRculties relating
to the resolution of chiral compounds. Thus, with CE CE enantiomeric resolution was Rrst established,
analysts have another tool at their disposal, with all through a procedure named ligand exchange. This is
the beneRts of the technique to be exploited. Two based on metal chelate complexation with copper or
major processes of separation are undertaken in en- zinc at the centre of the complex. It was directed
antiomeric separations. The commonest procedure is towards amino acids, which as dansyl-labelled deriv-
where the enantiomers are resolved directly as atives were resolved through initially interacting cop-
diastereoisomers in the conditions of a chiral environ- per (Cu(II)) and L-histidine in the run buffer and then
ment present in the capillary. The transient dia- forming a ternary metal complex with an enantiomer
stereoisomers are formed in situ and resolved on the of a chiral amino acid. In an enantiomeric mixture the
basis of their different physicochemical properties. metal complexes can have differential stability, and
Alternatively, conventional CE separations follow therefore mobility differences between the enantio-
after off-line diastereomeric derivatives are produced mers result. In the case above, a neutral pH buffer
through reaction of the analyte with optically active was used, which gives a positively charged metal}
reagents, in a procedure which is commonly referred histidine interaction and the enantiomer forming the
to as an indirect method. An example is the resolution higher stability complex migrates faster. One draw-
of the diastereomeric pairs of D, L-tryptophan with back of this procedure is the requirement that the
(#)-diacetyl-L-tartaric anhydride which are resolved amino acid in the metal}amino acid interaction has to
on a polyacrylamide-coated, conventional silica cap- be of very high purity to stop enantiomeric impurities
illary. Certain precautions are necessary within this being introduced into the assay. Following the initial
procedure; primarily, a reagent of very high optical research, the same group impressively resolved 14
purity (100%) is required for speciRc reaction with racemic amino acids by this procedure where the
the individual enantiomers. Variability in purity of complex was Cu(II)}aspartame. At present the pro-
a reagent sample would result in diastereomeric inter- cedure is limited to amino acids, with some extension
ference and errors in the determination of the to peptides and additional separations, and most ap-
analytes. Incomplete reaction and differences in the plications appeared in the early years of CE.
speed of reaction are also problems which must be
Cyclodextrins
addressed when carrying out the derivatization.
Over the last 10 years, a number of CE modes have The interest in cyclodextrins (CDs) in the chemical,
been examined for separation of stereoisomers. Of cosmetic, food and drug industries has grown con-
these, capillary zone electrophoresis (CZE) in considerably. In the pharmaceutical Reld incorporation
junction with chiral additives in the buffer phase is within the CD has led to improvements in bioavaila-
the most frequently used for charged chiral com- bility, pharmacokinetics and stability. However, to
pounds. For neutral chiral analytes (and charged the analyst, their major inSuence has been in enan-
compounds), micellar electrokinetic chromatography tiomer stereoselectivity, which originates from the
(MEKC) has been applied through a chiral surfactant chirality of the glucose units and involves stereochem-
or a chiral additive and achiral surfactant, as addi- ical interaction through hydrophobic inclusion, hy-
tives to the buffer in a conventional silica capillary, drogen bonding and often steric repulsion, when the
when operated at the critical micelle concentration. CD is used as an additive in an aqueous CE buffer.
In contrast to these applications where the selector The CDs, produced enzymatically from starch, are
comprised of 6}8 D-(#)-glycopyranose units as -, - that CDs are neutral molecules and that the difference
and -CD. The molecules are shaped as truncated in electrophoretic mobility in mixtures has to occur
cones, with relatively hydrophilic exteriors, due to from the charge on the chiral analytes. In method
secondary C2 and C3 hydroxy groups at the top and development it is important to optimize the enan-
primary C6 hydroxy groups at the bottom of the tiomeric separation, and there are three main areas of
cone. The rigid structure presents an internal hydro- interest, two of which revolve around solvation ef-
phobic cavity, which will accommodate ring-struc- fects, inSuenced through varying the pH and the
tured enantiomers. Therefore, for chiral recognition organic additive concentrations. The CD concentra-
of enantiomers, the most favourable chemical struction is also important: a theoretical model, has been
ture is where the hydrophobic centre, such as a cyc- described which indicates a maximum concentration
loalkane or aromatic group (which generally interacts for enantiomeric resolution. Other researchers have
most closely with -CD), gives a close cavity Rt. But conRrmed this and extended discussions to chiral
one of the enantiomers gives a poorer steric Rt, which separation models which include the function of pH
results in differential binding constants between the and organic additives in the buffer. Other parameters
enantiomers. Typically, a small molecule with shown to have an inSuence on resolution include
a single aromatic ring will include with a tight Rt into buffer concentration and internal diameter of the
the -CD cavity. Additionally, hydrogen-bonding capillary.
sites are required for interaction with the secondary It was clear, however, in method development and
hydroxyls on the rim of the cavity and at least one in stereoselective HPLC assays with chiral additives
repulsive interaction with the CD. Generally, for in- that the solubility of the native CDs was a limiting
teraction with the CDs the chiral molecule’s hydro- factor, particularly with -CD. As a result a large
phobic centre should be close to the chiral centre number of CD derivatives have been synthesized and
(Figure 1). tested and are now commercially available for use in
CDs are generally reasonably straightforward to CE. These range from the electrophoretically neutral
work with in CE. They are commercially available, methylated -CD and hydroxypropyl -CD to
ultraviolet transparent and, during method develop- charged species, such as methylamino -CD and sul-
ment, the concentration and the buffer conditions, fobutylated -CD. In the former case solubility is
such as pH, can easily be changed, without the need improved manifold and, in the latter case, in addition
for long equilibration times. One area of slight difR- to improved solubility the charge on the CD gives
culty is solubility which, although adequate in water electrophoretic mobility to uncharged and neutral
or organic solvent for and (14.5 and 23.2 g per chiral compounds. Apart from these advantages the
100 mL in water), is only 1.85 g per 100 mL with the derivatized CDs often have different enantioselectivi-
commonly used -CD. Other general properties are ties over the native CDs (Figure 2). The Rrst enan-
tioseparations with CDs were carried out in 1985 as
an extension to isotachophoresis, where the CD was
added to the leading buffer. However Fanali is credi-
ted with one of the Rrst uses of -CD for the resolu-
tion of ephedrine alkaloids by CZE.
With a better understanding of the implications of
chiral compounds in everything from the chemical to
food and pharmaceutical industries, stereospeciRc
synthesis has developed to a point where very few
racemic compounds will be used in the future. How-
ever, the analyst still has a role to play in checking the
enantiopurity of the synthetic product. Detection
levels are regularly at (0.5% m/m and the order of
migration can often dictate whether low enantiomer
levels are detected. Most favourable conditions gen-
erally exist when the minor enantiomer precedes the
Figure 1 Capillary-zone electrophoresis resolution with -cyc- major component. One option to control the migra-
lodextrin (-CD) of the drug nomifensine maleate (65 mg L\1) tion order is to operate with anionic, cationic or
which has the hydrophobic group close to the chiral centre. The
neutral coated silica capillaries. A typical example is
buffer was 20 mmol L\1 sodium tetraborate (pH 12.5) containing
20 mmol L\1 -CD. The capillary was 37 cm (30 cm to de- shown in Figure 3 of a cationic coating, a polyamine
tector);50 m i.d. The applied voltage was 20 kV, detection on the surface and -CD, which gives a limit of
wavelength 284 nm and temperature 303C. quantitation of (0.05% m/m for D-tryptophan, in
Crown Ethers
Crown ethers act in a similar manner of enantiomer
inclusion as CDs and contain a central cavity, al-
though the mechanism is based on ionic and hydro-
gen bonding. They are macrocyclic polyethers, sol-
uble in both aqueous and organic solvents, which
form stable complexes with enantiomers, which have
a primary amine or alkylamine functionality. For
resolution, differential stability of the host}guest
complex is required. This is reliant on the spatial
arrangement between the amine and its hydrogen
bonds with the ether oxygens. 18-Crown-6-tetracar-
boxylic acid is the easiest to obtain commercially and
has been used to resolve a number of primary amines
and racemic amino acids on a silica capillary, by
addition of the crown ether into the buffer phase at
low concentration, under CZE conditions. By this
means a few chiral separations have been achieved
which have not been fully successful by other means,
such as chiral peptides. Crown ethers have also been
used in combination with CDs for complex mixtures.
In general, however, crown ethers have not been
Figure 2 The resolution of oxamniquine by capillary-zone elec-

trophoresis with neutral cyclodextrins. Resolution was at pH 2.25
with hydroxypropyl -CD, but there was no resolution with -CD at
this pH. The -CD gives resolution at pH 12, but with reversed
migration order. (A) 50 mmol L\1 disodium hydrogen phosphate
(pH 12) with 25 mmol L\1 -CD; applied voltage 15 kV; (B)
50 mmol L\1 sodium dihydrogen phosphate (pH 2.25) containing
40 mmol L\1 hydroxypropyl -CD; applied voltage 20 kV. The
temperature was 303C and detection wavelength 246 nm.
the presence of the L-enantiomer. During these runs

the polarity is reversed over the conventional direc-
tion of operation. The coated capillaries mentioned
above are commercially available and have been Figure 3 Detection of D-tryptophan at 0.05% m/m in L-tryp-
shown to be stable if used within the suggested limits. tophan (LOD (3;s/n) 0.01% m/m) with a polyamine coated
The anionic (sulfonic acid) coated capillaries are of capillary (eCAPTM with polyamine regeneration solution) to give
a positive charge on the capillary. Capillary was 37 cm (30 cm to
particular interest in chiral and conventional separ-
detector);50 m fused silica and buffer 40 mmol L\1 tris-phos-
ations as they give a consistent EOF over the range phoric acid containing 75 mmol L\1 -CD. The applied voltage
pH 3}9 and therefore are more controlled in opera- was !10 kV, detection wavelength 241 nm, injection time 2 s
tion. and operating temperature 303C.
extensively used because they need to be applied been cross-linked with glutaraldehyde to form a poly-
under controlled conditions as they are highly toxic mer matrix. However, BSA has a high ultraviolet
and they are limited to amines and alkylamines. background and as such needs to be used in indirect
However, in the applications reported they give high- detection mode or where the gel is only partially
ly efRcient resolutions of certain chiral compounds. loaded on to the capillary and stops just before the
detection window. AGP, ovosomucoid (egg white)
Other Chiral Additives
and fungal cellulose have all been used, with some
Apart from the CDs, linear oligosaccharides are po- success. The major problem with the fungal cellulose
tential chiral discriminators in CE, and maltodextrin is absorption on to the capillary wall if used with
mixtures, corn syrups and pure malto-oligosacchar- uncoated capillaries, although adding materials such
ides have all been used to resolve non-steroidal anti- as polyethylene glycol or methyl cellulose into the
inSammatory drugs (NSAIDs). Polysaccharides have electrolyte does alleviate wall interaction.
also been discussed in these reports, with applications
of heparin, a naturally occurring mammalian mu- Micellar Electrokinetic Capillary
copolysaccharide (anticoagulant), in the resolution of
chiral drugs. The molecule is chiral, highly anionic
Chromatography
and helical and has been used in the chiral discrimina- Neutral compounds present a problem in CE since
tion of antimalarials and antihistaminics. In Figure 4, they have no differential mobility because of their
oxamniquine is resolved with a large resolution factor lack of charge and are carried without separation by
with 2% w/v heparin. Thus heparin and some of the the EOF. One solution is to operate with surfactants,
other polysaccharides with their high separation ef- which form micelles above a certain concentration,
Rciencies appeared to be good alternatives to CDs. and then differentiate between the compounds on the
There is one drawback: because of their hetero- basis of both electrophoretic migration and micellar
geneous character, batch-to-batch and different partitioning. This technique has been so successful
source material can result in changes in the resolution that it is now one of the commonest procedures used
and therefore method robustness would be question- in CE. Initially the emphasis was on achiral com-
able for regular assay for industry. pounds, but it was quickly extended to chiral com-
Proteins provide another alternative based on suc- pounds by the introduction of either chiral surfac-
cesses in HPLC and particularly separations with tants or chiral additives, such as CDs. Surfactants are
1-acid glycoprotein (AGP) and bovine serum al- long chain molecules, characterized by a hydrophobic
bumin (BSA) for chiral drug separations. BSA has tail pointing inwards and a hydrophilic head pointing
outwards into the aqueous buffer. Micelles are am-
phiphilic aggregates of surfactant molecules which
form above a surfactant concentration, known as the
critical micelle concentration (CMC) (Table 1).
There are four classes of surfactants } anionic,
Table 1 The critical micelle concentrations of chiral surfactants

and achiral surfactants
Surfactant Critical micelle

concentration
(mmol L\1)
Chiral
Sodium cholate 14
Sodium taurocholate 13
Sodium deoxycholate 5
Sodium taurodeoxycholate 4
Sodium N-dodecanoyl-L-valinate 6 (403C)
Achiral
Sodium dodecyl sulfate (SDS) 8
Figure 4 The resolution of oxamniquine by capillary-zone elec- Sodium decane sulfate 40
trophoresis with heparin in the electrolyte solution. The electrolyte Tetradecyltrimethylammonium bromide
was 50 mmol L\1 sodium dihydrogen phosphate (pH 3.0) contain- (TTAB) 4
ing 2% m/v heparin. The capillary was 71 cm (50 cm to de- Cetyltrimethylammonium bromide
tector);50 m i.d., the applied voltage 20 kV, temperature 303C (CTAB) 1
and the detection wavelength 246 nm.
cationic, zwitterionic and nonionic } which can be

both synthetic and naturally occurring. The synthetic
group includes the anionic sodium dodecyl sulfate
(SDS) and cationic cetyltrimethylammonium bromide
(CTAB).
A slightly different process, microemulsion elec-
trokinetic chromatography, has recently been de-
veloped. In this process a combination of oil}
water}surfactant and co-surfactant such as an alkyl
chain alcohol is used as a pseudo-stationary phase.
Chiral Surfactants
These can be synthetic optically active amino acid
derivatives and natural surfactants, such as glycyrrhic
acid, -escin, bile salts and digitonin. Digitonin is
nonionic, but has been used as a mixed micelle with
SDS to give the electrophoretic mobility, to resolve
chiral amino acids. These materials are generally
water-soluble and can be added easily at micelle con-
Figure 5 The use of bile salts to resolve the enantiomers of
centrations to the electrolytes in MEKC, although diltiazem and related substances by chiral micellar electrokinetic
viscosity increases should be noted. Initially enan- chromatography (MEKC). The electrolyte was 20 mmol L\1 dis-
tiomeric separations were directed towards neutral odium hydrogen phosphate}sodium tetraborate (pH 7.0) contain-
chiral compounds such as benzoin, and warfarin, ing 50 mmol L\1 sodium taurodeoxycholate. The capillary was
a heart drug, which were selectively resolved with 65 cm (50 cm to detector);50 m i.d. Applied voltage was 20 kV,
detection wavelength was 210 nm and temperature was 253C.
materials such as sodium N-dodecanoyl-L-valinate
(SDVal) and N-dodecanoyl-L-glutamate (SDGlu). But
it was soon realized that charged chiral compounds
separation of chiral analytes has become quite popu-
could also be resolved. Sodium taurocholate (STC) or
lar, particularly as there are many derivatized CDs to
the taurodeoxycholate (STDC) has been used under
choose from, and many examples have been pub-
acidic conditions for the resolution of dansylated-DL-
lished. Of the native CDs, -CD has shown the
amino acids and for carboline derivatives and 2,2-
greatest success, which is the opposite of the case with
dihydroxy-1,1-dinaphthyl used under neutral condi-
the conventional addition of CDs to the buffer phase.
tions for the drug diltiazem (Figure 5).
In this case the larger inner diameter of the cavity of
the -CD allows not only the chiral molecules to be
Chiral Selector with an Achiral Surfactant
included but also the surfactant.
This mode can give added selectivity to resolution of
both neutral and charged chiral molecules, as the Capillary Electrochromatography
solutes are distributed between chiral selector such as
a CD, the aqueous phase and the micelle (Figure 6).
(CEC)
As a result, the chiral analytes will move in and out of There is much current interest in electrochromatogra-
the micelle/CD, with the CD moving with the EOF phy, which is a hybrid technique between HPLC and
and the micelle (depending on polarity) moving ac- CE, operating with packed capillaries. It is based on
cording to charge. Racemic amino acids, -blocker partitioning the analytes between the stationary
drugs, Trogers base and the drug terbutaline have all phase (in packed capillaries) and the mobile phase
been resolved with this procedure. In the case of (electrolyte) under CE conditions with a high applied
terbutaline either 5 mmol L\1 di-O-methyl-- voltage. The procedure beneRts from the large range
cyclodextrin or 15 mmol L\1 -cyclodextrin was of HPLC stationary phases and some speciRc phases
successfully used with 15 mmol L\1 SDS. Other ap- which are beginning to be developed for CEC. It links
proaches have been to carry out competetive separ- the high efRciency of CE with the range of separations
ations with CD-MEKC, where an additive, such as possible with reversed-phase HPLC. For transport of
D-camphor-10-sulfonate, is introduced into the buffer the analytes in CEC, EOF is still present, but now the
and competes for inclusion with the chiral analytes. effective EOF is mainly generated from the packed
With this procedure an improved resolution of bed of stationary phase rather than at the walls of the
Trogers base was obtained. Overall, this form of capillary. This results from the electrical double layer
Figure 6 Schematic of the chiral analyte 5-alkyl-5-phenyl hydantoin distributed between the cyclodextrin, the aqueous phase and the
micelle. SDS, Sodium dodecyl sulfate.
on the surface of the particles in contact with the towards the cathode, but with triethylammonium
mobile phase and produces a potential. The poten- acetate the Sow is reversed. This can be beneRcial in
tial of CEC was Rrst demonstrated in 1987, showing controlling the order of retention in CEC. The pH
the presence of the partitioning effects and, for change in the electroosmotic mobility can be used to
charged analytes, the continued effect of electromig- obtain a suitable overall retention time. Organic
ration. In terms of the separation conditions in CEC modiRers such as methanol and acetonitrile in the
and HPLC, the stationary and mobile phases are very electrolyte have an effect on both the retention and
similar, but the separation efRciency can be much enantioselectivity. The results from studies with CD-
improved. Many applications of electrochromatogra- bonded silica phases indicate that a range of para-
phy are currently being studied and the range of meters and additives can affect the retention and
capillary packing materials continues to expand. enantioresolution in CEC, as they do with chiral
Chiral stationary phases are being examined by stationary phases in HPLC.
a number of researchers. In one of the Rrst reports of Other stationary phases which have been shown to
chiral separations by CEC 5 m AGP, immobilized give enantioseparations are the cellulose and amylose
on to silica, was packed into a 50 m internal dia- derivatives which have been coated on to silica. These
meter capillary. Cationic and neutral enantiomers stationary phases are known as Chiralcel and Chiral-
were resolved by this method, which included ben- pak phases (Daicel, Japan), respectively. These mater-
zoin and the drugs cyclophosphamide and hexobar- ials have been successful in HPLC for enantioselectiv-
bital. This early work did not illustrate large improve- ity with a wide range of chiral compounds and have
ments in peak efRciency over other procedures as the recently been utilized as packing materials for CEC
capillary packing methods required improvement, (Figure 7). Generally in CEC good resolution has
but it did illustrate the potential. been achieved for a number of chiral compounds
CDs have also been used as immobilized silica- resolved by HPLC, with improvements in separation
based materials and the neutral CD hydroxypropyl-- efRciencies in many cases.
CD silica was used in resolution of the basic drug
mianserin. In method design with the CDs, the choice
of the background electrolyte has played an impor-
Future Developments
tant part and pH, organic modiRer and differences in Capillary electrophoresis has had a major impact on
Reld strength have all been considered. The direction the resolution of chiral compounds by the methods
of Sow past the detector was shown to be dependent discussed above. The method is now regularly used to
on the background electrolyte used. For -cyclodex- assay enantiomeric compounds in many Relds with
trin and a phosphate buffer the direction of Sow is generally better efRciency than in HPLC and with
2294 III / CHIRAL SEPARATIONS / Cellulose and Cellulose Derived Phases
shorter overall migration times. Many different areas

are showing considerable research interest, such
as electrochromatography, where in the future com-
mercially available columns may prevail over in
house methods of packing capillaries with chiral sta-
tionary phases. Generally, commercial interest in the
area of capillary packings is likely to increase as more
‘tailor-made’ phases are introduced.
Further Reading
Chankvetadze B, Endresz G and Blaschke B (1996)
Charged cyclodextrin derivatives as chiral selectors in
capillary electrophoresis. Chemical Society Reviews
141}153.
Clark BJ (1994) Separations of enantiomeric compounds by
chiral selectors in the mobile or solvent phase. In:
Subramaninan G (ed) A Practical Approach to Chiral
Separations, p. 311. Weinheim, Germany: VCH.
Kuhn R and Hoffstetter-Khun S (1993) Capillary Elec-
trophoresis: Principles and Practice. Berlin: Springer-
Verlag.
Lelievre F, Yan C and Zare RN (1996) Capillary electroch-
romatography: operating characteristics and enan-
tiomeric separations. Journal of Chromatography
A 723: 145}156.
Nishi H and Terabe S (1995) Optical resolution of drugs by
capillary electrophoretic techniques. Journal of Chro-
matography A 694: 245}276.
Rabel SR and Stobaugh JF (1993) Applications of capillary
electrophoresis in pharmaceutical analysis. Pharmaceut-
ical Research 10: 171}175.
Schmitt T and Engelhardt H (1995) Optimisation of enan-
tiomeric separations in capillary electrophoresis by re-
Figure 7 Electrochromatographic resolution of methylphenyl
versal of the migration order and using different de-
barbitone by a Chiralcel OJ (Daicel, Japan) chiral stationary
phase. The cellulose ester derivative commercially coated on to rivatized cyclodextrins. Journal of Chromatography
silica was packed into a 30 cm (21 cm packed and 22 cm to A 697: 561}570.
detection window);75 m i.d. fused silica capillary. The mobile Wren SAC (1993) Theoretical aspects of chiral separation
phase was acetonitrile }water (70 : 30 v/v), applied voltage 15 kV in capillary electrophoresis. Journal of Chromatography
and temperature 253C. A 603: 235}241; 609: 363}367; 636: 57}62.
Cellulose and Cellulose Derived Phases
J. Dingenen, Janssen Research Foundation, microcrystalline cellulose tribenzoate became com-

Belgium mercially available. Another milestone in the devel-
Copyright ^ 2000 Academic Press opment of cellulose- and amylose-based phases was
the work of Okamoto and co-workers who in 1984
introduced polysaccharide derivatives coated on
Introduction a macroporous silica. Currently such materials are
The properties of cellulose derivatives for the separ- certainly the most universally applicable type of com-
ation of enantiomers were recognized in 1966 by mercially available chiral stationary phases. Then, in
LuK tringhaus and Peters. The full potential of micro- 1988 Francotte and co-workers introduced cellulose
crystalline cellulose triacetate (CTA) was deRnitely esters of aromatic acids in the form of almost spheri-
established by Hesse and Hagel in 1973. Meanwhile, cal, partially crystalline particles with a relatively
III / CHIRAL SEPARATIONS / Cellulose and Cellulose Derived Phases 2295
good mechanical stability. The last few years have propanol, different amounts of water, or is com-
also seen a lot of research in grafting cellulose and pletely replaced by methanol or 2-propanol.
amylose derivatives on to a silica matrix. If compounds bearing an ionizable group have to
Recently, different types of polysaccharide phase be analysed, buffer systems should be tried. CTA
(Chiralcel AD, Chiralcel AS, Chiralcel OD and seems to withstand the use of buffers in the range
Chiralcel OJ) wall-coated open tubular (WCOT) col- between pH 5 and 10 for a long time without any
umns have also been used for supercritical Suid signiRcant loss of chromatographic properties. Elu-
chromatography applications. ents with a high water content at high or low pH
Here, the properties and possibilities of different values should be avoided because CTA can be hydro-
types of polysaccharide phases used for enantiomer lysed under these conditions.
separation are discussed. Empirically, it has been found that, on CTA suc-
cessful enantiomer separations can be expected if the
Microcrystalline Cellulose Triacetate analytes possess an aromatic or nonaromatic ring
close to the chiral centre or if the products have an
By a heterogeneous acetylation procedure, cellulose asymmetric atom on a rigid ring structure.
can be acetylated so that the mutual arrangement of Also it is known from experience that compounds
the carbohydrate chains within their crystalline re- bearing an ionizable group, like for example, a hy-
gions is maintained. CTA obtained by this acetylation droxyl, carboxyl or amino group, are generally poor-
technique has only weak chiral recognition abilities. ly resolved on CTA.
Only in its swollen state is triacetyl cellulose capable Derivatization of these functional groups into the
of separating enantiomers. An ethanol}water (95 : 5, corresponding ester, amide or carbamate derivative
v/v) mixture is the most frequently used swelling often improves the selectivity. For alcohol, an esteriR-
agent for CTA. When dry CTA is boiled for 15 min in cation reaction with a P-substituted benzoic acid
the ethanol}water mixture, a volume increase of chloride (F, Cl, Br, CN, NO2, OCH3) has been suc-
about 40% is observed. cessfully applied to solve a number of different separ-
Microcrystalline CTA is a material that can be used ation problems.
to separate a broad variety of enantiomers belonging CTA is attractive for chiral separations because it
to many different classes of organic compounds can be synthesized from a low cost natural product.
(thiazoline-2 thiones, barbiturates, hydrocarbons, The wide range of products that can be separated,
oxiranes, Savanones, - and -lactones, metal com- together with its high loading capacity, certainly
plexes, etc.) Its high loading capacity makes it an makes it a material that should be considered as a
extremely useful material for preparative chromato- valuable tool in the development of chiral separation
graphic purposes. methods. Besides the usefulness of CTA in the Reld of
enantiomer separations, this material is also suitable
Mobile-Phase Design
for the separation of positional isomers.
A lot of different solvents, e.g. acetone, chlorinated
alkanes and acetonitrile, cannot be used because they Microcrystalline Cellulose
dissolve CTA more or less completely. In tetra-
hydrofyran, CTA forms a gel. In some other solvents,
Tribenzoate
like for example, 1,4 dioxane and dimethoxyethane, As early as 1969, Safanova and Klenkova had de-
CTA swells too strongly. Different investigators scribed the heterogeneous benzoylation of microcrys-
found as a general rule that the retention factors of talline cellulose. RimboK ck et al. later described
the investigated analytes increased when, respective- a method for preparing this material using an ultra-
ly, methanol, ethanol or propanol was used as the sonic Reld to accelerate the reaction. Nowadays,
eluent. However, the highest selectivity factors are a methanol}water solution containing about 40% of
often found with ethanol as the eluent. Successful the pre-swollen material is commercially available.
separations are also obtained by using other low Comparable with CTA, different solvents or sol-
molecular weight alcohols, ethers, hydrocarbons and vent combinations also cause different degrees of
mixtures of these eluents. Examples are methanol-2- swelling of this material. Dichloromethane or tet-
propanol (80 : 20; v/v); ethanol}tert-butyl methyl- rahydrofuran cannot be used as the eluent because
ether}water (86 : 10 : 4, v/v) and n-hexane}2-propanol} tribenzoylcellulose is soluble in these solvents.
water (70 : 27 : 3, v/v). With microcrystalline cellulose tribenzoate, a
In general, a signiRcant inSuence on the retention stationary phase for chromatographic enantiomer
factor and enantioselectivity is observed when separations has been introduced, which allows the
ethanol as the eluent is modiRed with methanol, 2- resolving of different classes of organic compounds.
Cellulose tribenzoate, compared to some other

types of chiral stationary phases, offers the advant-
age that, besides analytical separations, preparative
work can also be performed without a great deal of
effort.
Benzoyl Cellulose Beads

(in the Pure Polymeric Form)
Francotte and co-workers developed this type of Figure 1 Benzoyl cellulose beads in the pure polymeric form:
chiral stationary phase. These materials are cellulose effect of polar modifier concentration on the retention factor.
esters of aromatic acids in the form of almost spheri- Column: 125;4 mm i.d. filled with 10 m p-methylbenzoyl cellu-
cal, partially crystalline particles with a good mech- lose beads. Mobile phase: n-hexane}2-propanol in different
ratios. Flow rate: 1 mL min\1. Sample: Mesuximid.
anical stability.
To characterize these phases, a broad variety of selectivity factor in the studied range of polar modi-
product classes on experimental batches of benzoyl- Rer content remains rather small.
and p-methylbenzoyl cellulose beads have been The benzoyl cellulose derivatives in the pure poly-
examined. Methanol, ethanol and n-hexane-2-pro- meric form allow the separation of a broad variety of
panol were used as the eluent and the results obtained compounds. In many cases, results are obtained
were compared with data measured under the same which are comparable with the values measured on
experimental conditions on the corresponding phys- the corresponding physically coated phases. Further-
ically coated materials (Chiralcel OB and Chiralcel more, these materials are very useful as stationary
OJ). phases for preparative chromatographic applications.
In general, much higher retention values are ob-
served on the pure cellulose derivatives compared to Physically Coated Cellulose and
the values measured on the corresponding coated
stationary phases. This observation is logical, because
Amylose Derivatives
the physically coated phases only contain about 20 This type of chiral stationary phase, developed by
weight percent of the cellulose derivative. For the Okamoto and commercialized by Daicel Chemical
investigated product series, in most cases smaller re- Industries, can be considered to be the most univer-
tention values were measured for methanol than for sally applicable type of chiral stationary phase for
ethanol as the mobile phase. Furthermore, most prod- both analytical and preparative chromatographic
ucts also demonstrated the highest resolution values applications. Many different column types are avail-
when methanol was used as the eluent. able on the market. However, experience suggests
For some of the substances tested, practically no that, with four different types of phases, a wide range
difference in resolution was observed between the of racemates, belonging to a broad variety of product
pure cellulose material and the coated phases, while classes, can be separated. The most interesting phases
other product classes were always better separated on in our view are:
the commercially available columns.
1. Cellulose derivatives:
Mobile-Phase Effects (a) Chiralcel OD (3,5 dimethylphenyl carbamate
derivative)
A chromatogram of the Mesuximid enantiomer sep-
(b) Chiralcel OJ (para-methylbenzoyl derivative)
aration that clearly illustrates the separation power of
2. Amylose derivatives:
this type of materials is depicted in Figure 1.
(a) Chiralpak AD (3,5 dimethylphenyl carbamate
For Mesuximid, using n-hexane}2 propanol as the
derivative)
mobile phase, the investigation of the effect of the
(b) Chiralpak AS [(S)--methylbenzyl carbamate
polar modiRer concentration on the retention factor
derivative)]
demonstrate that, below a certain polar modiRer con-
centration, the retention factor strongly increases. The analytical phases are coated on a 10 m wide
This can be an indication that competition between pore silica. However, some types of phases are also
the polar modiRer and the solute for hydrogen-bind- available on a 5 m silica matrix. Furthermore,
ing sites on the stationary phase might be the deter- two phases, which are speciRcally designed for rever-
mining factor in the retention process. sed-phase applications, have been brought on to the
Although the retention factor is strongly inSuenced market. An overview of the different types of com-
by the polar modiRer concentration, the effect on the mercially available coated phases is given in Table 1.
Table 1 Commercially available silica-coated phases for discouraged the use of other solvent combinat-
enantioseparation ions than those recommended by the manufac-
turer. However, in the last few years, other aprotic
Cellulose esters Type of absorbent
solvents, e.g., acetonitrile, methyl-tertiary-butyl ether
Chiralcel OJ p-Methylbenzoyl and ethyl acetate have been applied on certain types
Chiralcel OJ-R For reversed-phase application of the derivatized polysaccharide columns. Certainly
Chiralcel OB Benzoyl in the Reld of preparative chromatographic enan-
Chiralcel OB-H Coated on a 5 m silica matrix
tiomer separations. These alternative solvents widen
Chiralcel OA Acetyl
Chiralcel OK Cinnamoyl the application range of this type of stationary
Chiralcel CA-1 Acetyl phase.
For method development on this type of phase
Cellulose carbamates screening experiments on 50;4.6 mm ID pre-col-
Chiralcel OD 3,5-Dimethylphenyl
umns Rlled with respectively Chiralcel OJ, Chiralcel
Chiralcel OD-H Coated on a 5 m silica matrix
Chiralcel OD-R For reversed-phase applications OD, Chiralpak AD and Chiralpak AS are convenient.
Chiralcel OD-RH For reversed-phase applications In use such column are rapidly equilibrated, give fast
Chiralcel OC Phenyl separations and have in many cases more than sufR-
Chiralcel OG p-Methylphenyl cient separation power for initial method develop-
Chiralcel OF p-Chlorophyenyl ment work. As a general rule, pure ethanol at a Sow
Amylose carbamates rate of 0.5 mL min\1 is used as the eluent of choice to
Chiralpak AD 3,5 Dimethylphenyl perform the Rrst experiments.
Chiralpak AS (S)-Methylbenzyl Pure ethanol is chosen based on the experience
that, for a lot of products, below a certain amount of
polar modiRer in an n-hexane or n-heptane based
mobile phase, the retention factors strongly increase.
Mobile-Phase Design and Parameter Optimization In general, the effect of the polar modiRer on the
Different investigators have suggested that the mech- retention factor decreases upon increasing modiRer
anism of chiral recognition on cellulose and amylose content. An indication that the competition between
derivatives is based on: the solute and the polar modiRer in the eluent for the
hydrogen-bonding sites of the stationary phase is
1. Hydrogen bonding
a saturable process and a maximum effect on the
2. Dipole}dipole interaction
retention factor will be reached within a certain range
3. Charge transfer complex formation (} interac-
of polar modiRer concentration (for our type of com-
tions)
pounds, mostly situated between 15 and 20%). If, in
4. Possible inclusion into chiral cavities or channels
the screening experiments, the analytes are insufR-
of the chiral stationary phase
ciently retained or no separation is observed, ethanol
Whatever the type of interaction involved, the mobile is mixed with n-heptane or n-hexane in different
phase must be considered as a dynamic part of the ratios. Once a suitable n-heptane}ethanol ratio has
system, capable of interacting with both the enan- been found to reach a k value between 3 and 6,
tiomeric solute and the chiral stationary phase. For ethanol is replaced by the same molar amount of one
solutes where hydrogen bonding plays an important of the other lower aliphatic alcohols (1- or 2-pro-
role in the selective chiral interaction process, proton- panol, primary, secondary or tertiary butanol). In
donating polar modiRers can compete with the solute many cases, large effects on the resolution values can
for the hydrogen-bonding sites of the stationary be observed when ethanol as polar modiRer is re-
phase. In other cases, } interactions between an placed by another alcohol.
aromatic moiety on the solute and the chiral station- Experience shows that pure methanol or mixtures
ary phase seems to be the most important interaction of ethanol and methanol or ethanol and 2-propanol
force. are often very useful to improve a separation.
Most of the separations on physically coated cellu- On these phases it has been observed that the
lose and amylose columns have been performed with retention factors of a range of 2,3-dihydro-1,4-
mobile phases consisting of n-hexane or n-heptane as benzodioxin-2-carboxylic acid esters increased with
the major component, mixed with various aliphatic increased chain length of the alcohol used for
alcohols. The choice of solvent combination is princi- chromatography on a Chiracel OB column. This ef-
pally based on recommendations for use by the fect might be based on a reduced capability of the
manufacturer. Concerns about the stability of these larger alcohols to compete with the solute for hydro-
columns combined with the relatively high cost gen-binding sites on the stationary phase.
The higher retention values measured for the bran- tiomer, because it is easier to quantify a small peak in
ched alcohols compared to their corresponding linear front of a large one than in the opposite situation.
analogues may be due to steric inSuences, which Based on experience, it is also important to mention
result in a reduced tendency of these alcohols to that sometimes even small variations in experimental
interact by hydrogen bonding with the polysacchar- conditions can have a tremendous effect on the res-
ide phase. olution of enantiomers on this type of stationary
The observed decrease in retention factor with in- phases.
creasing chain length of the ester group could be an The often observed strong inSuence of small differ-
indication that hydrophobic effects also contribute to ences in experimental conditions on the resolution,
the interaction mechanism between the solute and the certainly has to be considered when robust and repro-
chiral stationary phase. ducible methods have to be developed on this type of
For the lower members of the homologue series of phases, for regulatory (Good Laboratory (GLP),
the 2,3-dihydro-1,4-benzodioxin-2-carboxylic acid Good Manufacturing (GMP)) purposes.
esters, practically no difference in resolution values An example of the use of aprotic modiRers, which
can be observed between the different polar modi- can also affect the separation on these phases, is as
Rers. Depending on the type of alcohol used, the follows. About 40 g of a diastereomeric mixture had
resolution rapidly decreases for the pentyl ester and to be separated into its four enantiomers. Very good
the higher homologues. Only for n-butanol as polar separation was obtained on an amylose 3,5-dimethyl-
modiRer, is the decrease in resolution value rather phenyl carbamate (Chiralpak AD) column using pure
limited. For ethanol the lowest resolution values are ethanol, pure methanol or a mixture of ethanol and
observed for the whole product series. There are 2-propanol in a 90 : 10 volume ratio. Unfortunately,
indications that, for the investigated solutes, hydro- the solubility of the diastereomeric mixture in the
gen-bonding forces certainly play an important role alcohol used was very poor. We therefore decided to
in the chiral recognition process. study the behaviour of ethanol}acetonitrile mixtures,
This type of experiment clearly demonstrates that, because the solubility of the product was a lot better
when n-hexane or n-heptane}alcohol mixtures are in acetonitrile.
used, testing different alcohols as polar modiRer dur- The results of these experiments are summarized in
ing method development should be performed, be- Table 2 and graphically represented in Figure 3.
cause in many cases, large differences in enantioselec- As illustrated in Figure 3, the addition of an aprotic
tivity may be observed. modiRer to the polar organic solvent has a clear effect
It is also interesting to note that, by changing the on the retention behaviour and the enantioselectivity,
polar modiRer phase, an inversion of the elution order especially for the most retained enantiomers. The
can occur, as illustrated in the following example. On investigated solute molecule contains different hydro-
a Chiralcel OD column, it was possible to separate gen-bonding groups, which can strongly interact with
the investigated compound (R89439) in its two enan- the carbamate functionality of the stationary phase.
tiomers using a mixture of n-hexane}ethanol in an Combined with poor solubility of the substance in the
80 : 20 volume ratio. The desired enantiomer eluted lower molecular weight alcohols, this makes these
as the Rrst peak under these experimental conditions, solvents less suitable to displace the solute from
as illustrated in Figure 2A. When the ethanol as polar the stationary phase and probably explains the high
modiRer was partially replaced by methanol, a rever- retention values observed for pure methanol or
sal of elution order was observed, as illustrated in ethanol.
Figure 2B. The addition of acetonitrile improves the solubility
Some investigators have suggested that binding of the product. This at Rrst simpliRes the transfer of
a polar modiRer to sites near the chiral cavities might the solute into the mobile phase and furthermore
alter the steric environment of these cavities. If the gives the protic solvent a better chance to compete for
environment of these cavities changes, it can certainly the hydrogen-bonding sides of the stationary phase,
have an inSuence on the steric Rt of the chiral solutes which may explain the observed systematic decrease
in these cavities, which may in part be responsible for in retention factors with increasing acetonitrile con-
the observed phenomenon of reversal in elution centration. Below a certain concentration of protic
order. It is furthermore interesting to note that a re- solvent in the eluent, the strong hydrogen-bonding
versal of elution order can often be achieved by the properties of the solute predominate again, resulting
changeover of a carbamate-type phase towards a ben- in an increase in retention factor. Using a mixture of
zoate-type phase. This knowledge can be very useful ethanol and acetonitrile in a 30 : 70 volume ratio
when a small amount of one enantiomer has to be allowed the separation of the diastereomeric mixture
determined besides a large excess of the other enan- in its four enantiomers without difRculty.
Figure 2 Effect of mobile-phase composition on the elution order. Column: 250;4.6 mm. Chiralcel OD (cellulose 3,5 dimethylphenyl
carbamate). Mobile phase: (A) n-hexane}ethanol (80 : 20; v/v); (B) n-hexane}ethanol}methanol (80 : 5 : 15; v/v). Flow rate:
1 mL min\1. Sample: R89439 (racemate plus pure enantiomer).
Basic or acidic additives can also have major effects

Table 2 Effect of the addition of acetonitrile on retention behav-
on the separations. Thus, for some basic or acidic
iour and enatioselectivity
substances, it is often necessary to add respectively
Acetonitrile k 1 k 2 k 3 k 4 (1}2) (3}4) a base or an acid to improve the peak shape. Based on
(%) the manufacturer’s recommendations, it is possible to
use up to 1% of di- or triethylamine to reduce the
0 2.94 6.43 11.97 28.24 2.187 2.359 tailing of basic substances. Acetic or triSuoroacetic
5 1.80 3.6 8.67 14.30 2 1.649 acid can be used for the analysis of acidic substances.
10 1.25 2.7 7.55 10.25 2.16 1.358 The usefulness of these mobile-phase additives, is
15 0.95 2.08 6.34 8.03 2.189 1.267 shown in the following example. For the enantiomer
20 0.66 1.59 5.47 6.62 2.409 1.210 separation of [($) 2,6-dichloro--(4-chlorophenyl)-
30 0.50 1.25 5.26 5.26 2.50 1.0
4-(4,5-dihydro-3,5-dioxo-1,2,4-triazin-2-(3H)-yl)
40 0.46 1.1 5.05 5.05 2.391 1.0
benzene acetonitrile] (diclazuril) initially a method
50 0.45 1.04 5.72 5.72 2.311 1.0
70 0.90 1.73 9.38 12.05 1.922 1.285
was used in which the acidic NH-group in the 3,5-
80 1.31 2.67 16.86 27.0 2.038 1.601 dioxo-1,2,4 triazin part of the molecule was methyl-
ated with diazomethane. The derivatized product
Figure 3 Effect of acetonitrile on (A) the retention factor; (B) enantioselectivity. Column: 250;4.6 mm Chiralcel AD (amylose 3,5
dimethylphenyl carbamate). Mobile phase: ethanol}acetonitrile in different ratios. Flow rate: 1 mL min\1. Filled circles, k 1; triangles,
k 2; open circles, k 3; squares, k 4.
could easily be analysed on an amylose 3,5 dimethyl- experiment on a micro-LC column is shown in Fig-
phenyl carbamate (Chiralpak AD) column, using ure 4B.
ethanol}n-hexane in an 80 : 20 volume ratio. The use of an amine as tailing reducer is illustrated
A chromatogram of this separation is illustrated in in the following example. A few grams of a chiral
Figure 4A. amino alcohol had to be separated in its two enantio-
As can be seen from Figure 4A, reaction with dia- mers. Only partial separation could be obtained on
zomethane results in different reaction products, with a Chiralcel OD (cellulose 3,5-dimethylphenyl carba-
both nitrogen alkylated and oxygen alkylated com- mate) column using a mixture of n-hexane and 2-
pounds observed. Because direct analysis of this prod- propanol in a 70 : 30 volume ratio. Because a severe
uct on the cellulose- or amylose-based stationary tailing was observed, we investigated the effect on the
phases, using the classical eluents, was not possible, addition of triethylamine as mobile-phase additive.
due to the presence of the acidic NH-group in the A small quantity of 0.1 vol% of triethylamine im-
3,5-dioxo-1,2,4-triazin part of the molecule (which proved the peak shape and the resolution. However,
resulted in a retardation of the substance on the a much better result was observed when the
stationary phase), the effect of the addition of triSu- triethylamine content was increased to 0.5 vol%.
oroacetic acid was examined. The result of such an Thereafter, triethylamine was replaced for the same
Figure 4 (A) Separation of diclazuril after derivatization with diazomethane. Column: 250;4.6 mm i.d. Chiralpak AD (amylose 3,5
dimethylphenyl carbamate). Mobile phase: ethanol}n-hexane (80 : 20; v/v). Flow rate: 0.5 mL min\1. Detection: UV (290 mm). Injection
volume: 10 L. Temperature: ambient. Peak 1: Enantiomers of N-methylated product. Peaks 2 and 3: Enantiomers of O-methylated
product. (B) Separation of diclazuril using trifluoroacetic acid as mobile-phase additive. Column: 150;0.32 mm i.d. Chiralpak AD.
Mobile phase: ethanol}1% trifluoroacetic acid. Flow rate: 5 L min\1. Detection: UV (280 nm). Injection volume: 60 nL. Temperature:
ambient.
amount of diethylamine. The addition of 0.5% die- maximum resolution on the preparative column, the
thylamine resulted in the highest resolution value. amount of 2-propanol had to be reduced from 30%
Therefore, the Rrst experiment on the preparative to 10 vol%, while the triethylamine concentration
column was performed with a mobile phase contain- had to be increased to 2 vol%.
ing 0.5 vol% of diethylamine. The obtained result Using this method, it was possible to inject 250 mg
was rather poor. The enantiomers eluted as relatively of the product. The optimized method enabled sufR-
broad peaks, which were only partially resolved. Be- cient amount of the pure enantiomers to be prepared
cause for preparative chromatographic application in a reasonable amount of time.
we prefer to work with triethylamine instead of di- The solute structure will also clearly have a direct
ethylamide, the method was therefore further opti- effect on the separation. Thus, the chiral recogni-
mized using triethylamine as tailing reducer. To reach tion process on polysaccharide phases results from
differences in the summation of binding energies ori- an important role in the chiral recognition process,
ginating from: which of course makes it not always that easy to
predict whether a new product in a series of compara-
1. hydrogen bonding
ble structures will be separated or not on a particular
2. dipole}dipole interactions
type of stationary phase.
3. charge transfer (}) complex formation
The next example also demonstrates that it is not
4. steric interactions
always easy to predict whether a product within
It is not possible to generalize which type of interac- a series will be separated on a certain type of polysac-
tion forces plays the key role in the solute}chiral charide phase.
stationary phase complex formation. Hydrogen Some tetramisol derivatives were investigated on
bonding certainly has a strong role in the selective three different carbamate-type phases (Chiralcel OC,
chiral interaction process. Chiralcel OF and Chiralcel OD) using n-hexane}2-
Based on the knowledge that on polysaccharide propanol in a 70 : 30 volume ratio. For the investi-
phases different mechanisms play a role in the chiral gated compounds, the highest retention is measured
recognition process and small variations in experi- on the Chiralcel OF (p-chlorophenyl carbamate) col-
mental conditions or solute structure can strongly umn, while the smallest values are observed on the
affect the enantioselectivity, it is clear that small cha- Chiralcel OD column. However, on the three differ-
nges in the molecular structure of a speciRc type of ent column types the meta-substituted derivative has
compounds that are in general well separated can always the most strongly retained.
often be a challenge to Rnd an acceptable separation Although the products were most strongly re-
method for some of the members of such a product tained on the Chiralcel OF column, not one of the
series. 10 investigated substances was fully baseline-re-
The effect of small structural changes on the reten- solved. Two of the ortho-substituted compounds and
tion behaviour and the enantioselectivity is illustrated the unsubstituted tetramisol were best resolved on
in the following examples. the Chiralcel OC column, while for the meta- and
In the Rrst example, a few imidazole derivatives para-substituted derivatives the highest resolution
bearing an ester function on the imidazole ring were value was in general measured on the Chiralcel OD
investigated on a Chiralcel OC (cellulose phenyl column.
carbamate) as well as on a Chiralcel OD (cellulose It is quite clear therefore from the different exam-
3,5-dimethylphenyl carbamate) column, using n- ples that small changes in solute structure or experi-
hexane}2-propanol as the mobile phase. On the mental conditions can have a strong inSuence on the
Chiralcel OC column a mixture of n-hexane}2-pro- chromatographic behaviour. This phenomenon
panol in a 90 : 10 volume ratio was used. Because makes it often difRcult to predict whether a product
with this mobile-phase composition the products will be separated on a certain type of polysaccharide
were not sufRciently retained on the Chiralcel OD phase or not. Therefore, even when a good mobile-
column, the amount of polar modiRer had to be phase composition has been found on a particular
reduced to 5 vol% on this column type. On both column, it is always advisable to test this solvent
column types the retention factor strongly decreased mixture on another column of the same type (carba-
with increasing chain length of the eater group at- mate or benzoate).
tached to the imidazole ring. Also steric effects Achiral derivatization can be exploited to improve
seem to play a role, because on both columns the resolution as it is well known from experience that
smallest retention value is measured for the 2-propyl compounds bearing an ionizable group, e.g. a car-
ester. boxyl, hydroxyl or amino group, are often poorly
When the resolution values are considered, a clear resolved on the polysaccharide type of phases. De-
difference was to be observed between both column rivatization of these functional groups to the corre-
types. For the 1-propyl and 2-propyl esters, baseline sponding ester, carbamate or amide derivatives fre-
resolution is obtained on both stationary phases. quently improves the separation.
However, the observed differences for the ethyl and For alcohol, for example, an esteriRcation reaction
methyl ester are striking. On the Chiralcel OC col- with a para-substituted benzoic acid chloride has
umn the ethyl ester is still baseline-resolved, while the proven to be effective in solving difRcult separation
methyl ester is only partially separated, whereas on problems. The effect of an esteriRcation reaction on
the Chiralcel OD column, the methyl ester is very well the retention behaviour and the enantioselectivity is
separated and the ethyl ester did not display any illustrated by means of the next examples.
separation at all. Certainly there is an indication that The hydroxyl function of two completely different
small but nevertheless inSuential contributions play compounds was derivatized to yield the following
types of esters: cases, these compounds are easier to separate than the
native alcohol. For very difRcult separations we
1. Benzoyl

eventually had to synthesize the naphthylsulfonyl
2. p-Fluorine
derivative.
3. p-Chlorine
Temperature is well known to affect chiral iso-
4. p-Bromine
lation. Thus, on chiral stationary phases, speciRc in-
5. p-TriSuoromethyl Benzoyl
teraction mechanisms are involved in the separation
6. p-Methyl
process. Therefore, this type of phase often displays
7. p-Methoxy
slow mass transfer characteristics.
8. p-Nitro
Temperature, together with the mobile-phase velo-
9. p-Cyano
city, certainly has to be considered as a factor that can
10. 1-Naphthoyl
be used to inSuence this mass transfer process.
11. 2-Naphthoyl
In the next example, we examined the combined
12. 9-Antracoyl
effects of temperature and Sow velocity variations on
The different esters were respectively investigated enantioselectivity, column efRciency and resolution.
on a Chiralcel OD, Chiralcel OJ, Chiralpak AD and The enantiomers of the investigated racemate were
Chiralpak AS column using pure ethanol as the elu- difRcult to separate. Only partial separation could be
ent. achieved with n-hexane}ethanol in a 90 : 10 volume
These compounds were retained more on the ben- ratio on a Chiralcel OJ column, using our standard
zoate type of phase than on the carbamate type of experimental conditions of Sow rate and temper-
stationary phase. Also, the resolution is signiRcantly ature. To investigate this separation problem further,
higher on the Chiralcel OJ column than on the other the temperature was varied between 5 and 403C and
columns. The favourable results on the benzoate type Sow rates between 0.25 and 2 mL min\1 were tested.
of column inspired us to investigate the difference A good linear relationship between the logarithm
between ethanol and methanol as the mobile phase of the value and the temperature has been observed.
for this type of compound. For methanol much higher The value steadily increased with decreasing tem-
retention factors are measured than when ethanol is perature. However, the most signiRcant parameter to
the eluent. In general, about twice as high resolution indicate the separation between two products is the
values are measured for methanol. resolution value. This parameter is determined by
In the foregoing example, the best results for the both thermodynamic and kinetic contributions.
different esters were obtained on a benzoate type of Although two completely different measuring prin-
column. This is certainly not a general rule. ciples have been used, comparable patterns are ob-
Another alcohol [4,4-dimethoxy-1-(phenylmethyl)-3 served for the two resolution values. Both Rgures
piperidinol] was derivatized to yield a similar series of indicate that only for temperatures below 153C and
esters. This homologue series was also investigated on a Sow rate of 0.25 mL min\1 can a nearly baseline
the four different types of polysaccharide phases, separation be obtained. Although it was not possible
using ethanol as the eluent. to obtain a full baseline separation of the enantio-
On the Chiralcel OJ column only the 2-naphthoyl mers, the optimized analysis method was sufRciently
derivative was partially separated. On the Chiralcel accurate to follow up the investigations that were
OD column only the 2-naphthoyl and the 9-an- performed to develop a stereospeciRc synthesis
thracoyl ester were partially resolved, while on the method.
Chiralpak AS column only the 9-anthracoyl ester As this example indicates, during method develop-
showed partial resolution. However, on the Chiral- ment and optimization experiments, the factored
pak AD column most of the products were separated. temperature certainly has to be investigated. Besides
What is striking in this particular example is the essential information for analytical purposes, it also
partial separation of the native alcohol, while the gives chromatographers performing preparative chro-
benzoyl- and p-Suorobenzoyl esters do not show any matographic separations a good idea of whether it is
separation at all. possible to work at higher temperatures without los-
For further synthesis applications, alcohols are fre- ing efRciency. Due to the limited choice of possible
quently converted to the mesylated and tosylated mobile-phase compositions, the major problem one
derivative. When we are confronted with the prepara- has to deal with in preparative chromatographic
tive chromatographic separation of a racemic alco- work on polysaccharide phases is the solubility of the
hol, we always investigate the mesylated or tosylated product to be separated in the eluent used. Because
alcohol before carrying out other derivatization solubility in most cases increases with increasing tem-
work, because we have experienced that in many perature, the ability to work at higher temperatures
often improves throughput in preparative chromato- eluent, the diastereomers were very well separated.
graphic separations. Because with pure ethanol the retention factors were
The experiments performed also demonstrate that relatively high, the effect of the addition of acetonit-
the Sow velocity is a parameter that can be used to rile was investigated. A few chromatograms of these
optimize a separation process. experiments are depicted in Figure 5. Using a mixture
of 95 vol% ethanol and 5 vol% acetonitrile, it was
easy to isolate a large pure amount of both dias-
Separation of Diastereomers tereomers.
Although diastereomers are in general easy to separ-
Reversed-Phase Applications
ate under normal or reversed-phase chromatographic
conditions, it often happens, especially when the As mentioned before, two types of polysaccharide
chiral centres are located far away from each other, phases which are speciRcally designed for reversed-
that it is not always easy to separate these com- phase applications are currently on the market. This
pounds. When we are confronted with such a prob- type of phase is extremely interesting for the direct
lem, we always investigate the possibilities of micro- injection of aqueous solutions (ionic products,
crystalline cellulose triacetate or tribenzoate and the plasma samples, etc.). Furthermore, the columns are
physically coated derivatized polysaccharides. We useful in column-coupling techniques (for example,
have experienced that this approach can be a solution an octadecylsilica column followed by a polysacchar-
in many cases. An example of such a separation is ide column) to solve difRcult separation problems.
described in the next example. The manufacturer recommends on this type of col-
A racemic alcohol was derivatized with optically umn to use a mobile phase consisting of an aqueous
pure (1S)-camphor sulfonic acid chloride. It was not solution of a sodium, ammonium or potassium salt in
possible to separate the obtained diastereomers under combination with methanol, ethanol or acetonitrile
normal-phase conditions on bare silica or amino- as the organic modiRer. The choice of the cationic
modiRed silica. Under the reversed-phase conditions part of the salt seems not to have a signiRcant effect
which we generally apply in our laboratories it was on the separation. However, a signiRcant difference
not possible to separate the isomers. We did some in separation has been observed with various anions.
experiments on different derivatized polysaccharide Anions, such as ClO\ 4 and PF\6 in general show good
phases. On a 50;4.6 mm i.d. Chiralpak AD (amylose separation. Also the salt concentration can strongly
3,5-dimethylphenyl carbamate) using ethanol as the affect the retention time and the resolution.
Figure 5 Diastereomer separation. Column: 50;4.6 mm i.d. Chiralpak AD (amylose 3,5 dimethylphenyl carbamate). Flow rate:
1 mL min\1. Mobile phase: 1, ethanol; 2, ethanol}acetonitrile (95 : 5; v/v); 3, ethanol}acetonitrile (90 : 10; v/v); 4, acetonitrile.
Mobile-phase design and parameter optimization OD-R column, the retention values for phosphate or
Triethylamine salts are very popular as tailing reduc- sulfate anion are markedly higher than the values
ers in reversed-phase applications. We have therefore measured for the other anions. The lowest values
investigated the usefulness of these salts on the poly- were in most cases measured for chloride as the
saccharide-type phases. Experience with cyclodextrin counterion.
and derivatized cyclodextrin columns taught us that On both column types, the anionic part of the
a salt concentration of 50 mmol L\1 in general is high tailing reducer has a clear effect on enantioselectivity
enough to obtain good peak shapes for basic substan- (Figure 6). In general, the best results are obtained for
ces. Experimental work was started with this salt phosphate or sulfate as the anion, although for some
concentration. Afterwards, speciRc experiments per- products better results were obtained using another
formed to investigate the effect of the tailing reducer type of anion.
concentration on the chromatographic parameters The Chiralcel OJ-R column is much more suitable
conRrmed that 50 mmol L\1 was a good choice. In for the separation of the investigated compounds
a concentration range between 10 and 50 mmol L\1 than the Chiralcel OD-R column.
the highest resolution values were always measured
for the highest salt concentration. However, this con- Type of polar modiTer In experiments using a mix-
centration also resulted in the strongest retardation of ture of methanol and triethylamine in a 70 : 30 vol-
the compounds. ume ratio, high retention times were observed for
In a Rrst set of experiments, the difference in some compounds. Therefore, it was interesting to
chromatographic behaviour of a few products under investigate the effect of the addition of acetonitrile on
normal-phase and reversed-phase conditions was in- retention time and enantioselectivity. To perform
vestigated on a Chiralcel OJ-R and Chiralcel OD-R these experiments, a mixture of 70 vol% methanol
column. The columns were Rrst equilibrated with and 30 vol% triethylamine adjusted to pH 2.5 with
ethanol. Thereafter, a mixture of n-hexane}2-pro- phosphoric acid was prepared and thereafter mixed
panol in a 70 : 30 volume ratio was pumped through with different amounts of acetonitrile. The different
the column until equilibrium was reached and the chromatograms obtained during these experiments
different products were analysed. After rinsing the are shown in Figure 7.
columns with ethanol, the column was equilibrated As Figure 7 clearly illustrates, acetonitrile strongly
with a mixture consisting of 70 vol% methanol and affects retention of the investigated compound. Al-
30 vol% of a 50 mmol L\1 triethylamine solution in though the resolution steadily decreases with increas-
water adjusted with sulfuric acid to a pH value of 2.5. ing acetonitrile content, the k value of the second
In general the products are better retained when eluting peak could be reduced by a factor of 38 with
reversed-phase conditions are applied. It was also a full baseline resolution of the enantiomers as a result.
striking that for most of the investigated products Because the elution power of acetonitrile is signiR-
there was a large difference in resolution values be- cantly higher than for methanol, we generally prefer
tween the two modes of operation. to start new experiments with a methanol}water-tail-
ing reducer mixture. If the compounds of interest are
Type of tailing reducer Experiments on cyclodex- too strongly retained with this eluent, methanol is
trin and derivatized cyclodextrin columns have systematically replaced by acetonitrile until an ac-
shown us that, for the analysis of basic compounds, ceptable compromise between the retention factor
the anionic part of the tailing reducer has a clear and the resolution value is found.
effect on enantioselectivity. To investigate this effect,
we selected some imidazole derivatives. pH In the analysis of ionizable compounds, pH can
To perform the experiments, a 55 mmol L\1 solu- certainly have a strong effect on the chromatographic
tion of triethylamine in analytical grade water was behaviour. We therefore investigated for the product
prepared and 2 L portions of this solution were ad- series summarized in Table 2 the effect of pH vari-
justed to a pH value of 2.5 with respectively hydro- ations in the range between 2.5 and 4.5 on a Chiralcel
chloric, triSuoroacetic, methanesulfonic, camphor- OJ-R using a mobile phase composed of meth-
sulfonic, sulfuric and phosphoric acid. After pH ad- anol}acetonitrile and a 50 mmol L\1 triethylamine
justment the solution was further diluted to obtain solution in water adjusted to the desired pH value
a solution which contains exactly 50 mmol L\1 with phosphoric acid. For one of the products investi-
triethylamine. gated, the different chromatograms obtained during
As eluent a mixture composed of 70 vol% of meth- these experiments are depicted in Figure 8.
anol and 30 vol% of the triethylamine solution was Figure 8 shows that the retention factors stead-
used. On the Chiralcel OJ-R and on the Chiralcel ily increase with increasing pH value, reaching
Figure 6 Effect of the anionic part of the tailing reducer on the resolution. Column: 250;4.6 mm i.d. Chiralcel OJ-R ( p-methyl-
benzoyl cellulose). Flow rate: 1 mL min\1. Mobile phase: 50 mmol L\1 triethylamine adjusted to a pH value of 2.5 with different
acids}methanol (30 : 70; v/v).
maximum at pH 4 and decreasing again above this ide-based columns (Chirose-bond C1 and Chirose-
pH value. A similar pattern was also observed for the bond C3) on the market.
resolution value. We were able to test a few experimental chemically
The different experiments that have been per- bonded polysaccharide phases. At Rrst we compared
formed clearly indicate that, for ionizable com- a chemically bonded p-methylbenzoyl cellulose col-
pounds, pH can have an important effect on the umn with the corresponding physically coated mater-
enantioselectivity, which of course makes it a para- ial (Chiralcel OJ) using the classical alcohol-based
meter to be investigated thoroughly during method mobile phases. Whereas on the coated phases most of
development and optimization. the investigated compounds were partially or com-
pletely resolved with pure ethanol as the eluent, this
Chemically Bonded Polysaccharide Phases
was not the case on the chemically bonded material.
Over the last few years, various attempts have been For most of the compounds investigated it was neces-
made to graft derivatized polysaccharides on to a sil- sary to dilute ethanol with n-hexane to obtain an
ica matrix, without losing the chiral recognition acceptable resolution value. The experiments per-
properties of these materials. One of the major prob- formed certainly prove that the chiral properties
lems encountered was the limited amount of polysac- of the derivatized polysaccharide were not lost dur-
charide that could be chemically bonded. For the time ing the grafting process. However, compared to
being, only the French company Chiralsep (La the coated phases, smaller retention factors were
Fresnaye) has a few chemically bonded polysacchar- measured on the chemically bonded material. This is
Figure 7 Effect of type of polar modifier on chromatographic behaviour. Column: 250;4.6 mm i.d. Chiralcel OJ-R ( p-methylbenzoyl
cellulose). Flow rate: 1 mL min\1. Mobile phase: 50 mmol L\1 triethylamine adjusted to a pH value of 2.5 with phosphoric
acid}methanol (30 : 70; v/v)#acetonitrile in different ratios.
possibly an indication that only a small quantity of For all the investigated compounds, the smallest
the chiral moiety was chemically bonded on to the retention and selectivity values were measured
silica matrix. Also, some differences in enantioselec- for ethanol as the organic modiRer. For some prod-
tive properties could be observed between both ucts, extreme differences in resolution could be
phases. Phenoperidine, for example, was not resolved observed between the different experimental condi-
with pure ethanol on the coated phase while on the tions applied. For an amino alcohol, the chro-
chemically bonded column this product was very well matograms obtained using methanol and ethanol
separated, as illustrated in Figure 9. as organic modiRer are graphically compared in
A few experiments under reversed-phase condi- Figure 10.
tions were also performed on the chemically bonded Although only a limited number of experiments
p-methylbenzoyl cellulose column. Methanol or were performed, we can conclude that the tested
ethanol in combination with a 0.5% ammonium acet- chemically bonded polysaccharide phase has a high
ate solution in water was used as the mobile phase. potential for reversed-phase applications.
Figure 8 Effect of pH variations on the chromatographic behaviour. Column: 250;4.6 mm i.d. Chiralcal OJ-R ( p-methylbenzoyl
cellulose). Flow rate: 1 mL min\1. Mobile phase: 50 mmol L\1 TEA adjusted to the indicated pH values with phosphoric acid}meth-
anol}acetonitrile (30 : 40 : 30; v/v).
Besides these experiments, a series of 37 racemates ly Chiralcel OJ, Chiralcel OD, Chiralpak AD and
belonging to different product classes were investi- Chiralpak AS originating from a different source to
gated on chemically bonded equivalents of respective- the column used in the foregoing experiments.
For the Rrst tests, pure ethanol was used as the
mobile phase. Using this eluent, the largest number of
products were separated on the chemically bonded
equivalent of Chiralcel OJ. Twenty compounds were
partially or completely resolved on this column, com-
pared to six products on the cellulose 3,5-dimethyl-
phenyl carbamate column, 14 products on the
amylose 3,5-dimethylphenyl carbamate column and
eight compounds on the chemically bonded equiva-
lent of Chiralpak AS. However, the number of prod-
ucts that were separated on the chemically bonded
p-methylbenzoyl cellulose column was only 69% of
the number of products resolved on the physically
coated Chiralcel OJ column, using the same eluent.
Because for the same eluent composition similar re-
tention factors were measured on the chemically
bonded columns and their physically coated equiva-
lents, we may conclude that the amount of cellulose
or amylose grafted on to the silica matrix has to be
the same order of magnitude as the amount present
on the physically coated phases.
Other solvents were also used on these columns.
After extensive use of dichloromethane, which nor-
mally dissolves cellulose and amylose derivatives, the
Figure 9 Separation of phenoperidine. Column: 250;4.6 mm properties of the columns were tested again and com-
i.d. chemically bonded p-methyl benzoyl cellulose. Flow rate: pared with the results obtained on a fresh column.
1 mL min\1. Mobile phase: ethanol. Practically no difference in retention or resolution
Figure 10 Separation of an amino alcohol under reversed-phase conditions. Column: 250;4.6 mm i.d. chemically bonded p-methyl-
benzoyl cellulose. Flow rate: 1 mL min\1. Mobile phase: continuous line, 0.5% ammoniumm acetate in HPLC-grade water}ethanol
(50 : 50; v/v); dotted line, 0.5% ammonium acetate in HPLC-grade water}methanol (50 : 50; v/v).
values could be observed before and after the use of Some of the chemically bonded polysaccharide
dichloromethane. This is an indication that the poly- phases which have been tested certainly have a high
saccharide derivative was perfectly bonded on to the potential. The possibility of using a broader pallet of
silica matrix. solvents or solvent combinations especially widens
Some of our products were also investigated under for preparative chromatographic work the Reld of
normal-phase conditions on the Chirose-bond C1 application.
column of Chiralsep. Most compounds eluted as rela-
tively broad peaks with a severe tailing. From the 22
investigated compounds, only three products were Conclusions
fully baseline-resolved. The addition of a small Different types of derivatized cellulose and amylose
amount of triethylamine did not solve the tailing stationary phases are nowadays commercially avail-
problem. For only one of the investigated compounds able. With a limited number of these phases it is
was a very good resolution observed on the Chirose- possible to separate a broad variety of products. Fur-
bond C1. The results of these experiments are sum- thermore, these phases are also extremely useful for
marized in Table 3. preparative chromatographic applications.
The ability to use dichloromethane for this com- Although differences in chromatographic behav-
pound is interesting for preparative chromatographic iour between the tested chemically bonded phases
purposes, because the solubility of this product in the and their physically coated equivalents have been
commonly used solvents (alcohol, or alcohol}n- observed, we might expect that, after further optim-
hexane mixtures) is rather poor. ization of the grafting process, these phases will cer-
tainly enlarge the Reld of application of the de-
Table 3 Use of dichloromethane on chemically bonded cellu-
rivatized cellulose and amylose materials.
lose derivative
See also: II / Chromatography: Liquid: Chiral Separ-
Mobile-phase composition k 2 Resolution ations in Liquid Chromatography: Mechanisms. III / Chiral
Separations: Amino Acids and Derivatives; Capillary
n-Hexane}2-propanol (45 : 55; v/v) 3.93 2.15 5.88
Electrophoresis; Chiral Derivatization; Countercurrent
n-Hexane}2-propanol}dichloro-
methane (40 : 50 : 10; v/v) 3.14 1.97 4.63
Chromatography; Crystallization; Cyclodextrins and Other
Inclusion Complexation Approaches; Gas Chromatography;
2310 III / CHIRAL SEPARATIONS / Chiral Derivatization
Ion-pair Chromatography; Ligand Exchange Chromatog- Beesley TE and Scott RPW (1998) Chiral Chromatography.
raphy; Liquid Chromatography; Molecular Imprints as Chichester: Wiley.
Stationary Phases; Protein Stationary Phases; Super- Hesse G and Hagel R (1973) A complete separation of
critical Fluid Chromatography; Synthetic Multiple Interac- a racemic mixture by elution chromatography on cellu-
tion (‘Pirkle’) Stationary Phases; Thin-Layer (Planar) lose triacetate. Chromatographia 6: 277}283.
Chromatography. Johns DM (1989) Binding to cellulose derivatives. In
Lough WJ (ed.) Chiral Liquid Chromatography,
Further Reading pp. 166}175. Glasgow: Blackie.
Shibata T and Mori K (1989) In KrstulovicH (ed.) Polysac-
Allenmark S (1991) Chromatographic Enantioseparation, charide Phases in Chiral Separations by HPLC, Applica-
Methods and Applications, 2nd edn. London: Ellis tions to Pharmaceutical Compounds, pp. 336}398.
Horwood. Chichester: Ellis Horwood.
Chiral Derivatization
S. GoK roK g, Chemical Works of Gedeon, Richter mance liquid chromatography (HPLC) and gas
Ltd, Budapest, Hungary chromatography (GC) columns, thin-layer chromato-
Copyright ^ 2000 Academic Press graphy (TLC) plates, capillary electrophoresis (CE)
capillaries, etc., with ordinary mobile phases } are
used. The main possibilities for the separation of
Introduction enantiomers are:
The Importance of Enantiomeric Separations
E Transformation of the enantiomers to covalently
The separation of enantiomers of chiral compounds bonded diastereomeric derivatives by reacting
by chromatographic methods and related techniques them with homochiral derivatizing reagents prior
is one of the important tasks in modern analytical to their chromatographic separation using achiral
chemistry, especially in the analysis of compounds of stationary phases or and mobile phases. The de-
biological and pharmaceutical interest. However, tailed description of this general method, which is
analysis of this kind is also required in food analy- often referred to as an indirect method, is described
sis and the analysis of pesticides, Savours and here.
fragrances. The reasons for this are as follows: E Incorporation of the chiral reagent in the mobile
phase for the dynamic formation of diastereomeric
E As a consequence of the existing or potential differ- adducts, ion pairs or complexes with the enantio-
ences between the biological}pharmacological ac- mers to be separated during the chromatographic
tivities of the antipodes of racemic drugs, analytical run. In this case also, achiral stationary phases are
methods are required for their simultaneous deter- used.
mination in biological samples, thus enabling one E Separation of the enantiomers on chiral stationary
to follow the fate of the enantiomers of the admin- phases (HPLC, GC and TLC). Although in prin-
istered drug (candidate) in the animal or human ciple this general method does not require derivat-
organism. ization, the separation can be improved in many
E Asymmetrical syntheses are in the focus of interest cases by modifying the enantiomers using pre-
in various Relds of organic chemistry especially in column achiral derivatization. These aspects are
the synthesis of drugs administered as the pure also brieSy discussed here.
enantiomer. In these cases and also if the prepara-
tion of the enantiomers is carried out by classical
resolution techniques, analytical methods are Covalent Chiral Derivatization of
necessary for the determination of their enantio- Enantiomers and Separation of the
meric purity. Diastereomeric Derivatives on
Possibilities for Enantiomeric Separations
Achiral Columns
Introductory Remarks: the Role of Covalent Chiral
Since in achiral environment the physicochemical
Derivatization in Enantiomeric Separations
properties of the antipodes of racemates are identical,
their separation is not possible if the generally used, The derivatization of enantiomers using homochiral
achiral separation systems } ordinary high-perfor- reagents to form their diastereomeric derivatives
III / CHIRAL SEPARATIONS / Chiral Derivatization 2311
separable on achiral GC or HPLC columns or TLC vent composition and column selection, but in this
plates was the Rrst, widely used general method in the case it is even more important to select a proper
chiral analysis of drugs and related materials. After derivatizating reagent enabling diastereomers to be
the introduction of newer chromatographic and cap- formed with sufRciently different molecular Rne
illary electrophoretic techniques brieSy mentioned structures for their chromatographic separation. If
above, the importance of enantioseparations based the number of the chiral centres in the analyte is more
on covalent chiral derivatization has naturally de- than one, the reagent should enable the separation of
creased to some extent. However, this general all stereoisomers. An example for this is the separ-
method is still a method of choice widely used espe- ation of (S,R), (R,S), (S,S) and (R,R) nadolol in
cially in HPLC. The reasons for this are the large Figure 1.
number of commercially available homochiral re- The structural features which inSuence the separ-
agents and well-established reactions enabling dias- ation of the diastereomers are:
tereomeric derivatives with excellent separation and
detection possibilities to be formed and the possibility E The distance between the two chiral centres. In the
of tailor-made separations using inexpensive, achiral majority of cases this is two to four atoms but there
columns, i.e. being in possession of the R- and S- are many exceptions to this rule.
forms of the reagent, it is possible to have the peak of E Conformational rigidity of the diastereomers fa-
the enantiomeric impurity eluting before the main vours resolution. Bulky groups in the vicinity of
peak. one of the chiral centres or the incorporation of one
The importance of covalent chiral derivatization of them into a ring system are especially advantage-
can be characterized by the large number of publica- ous. For example, in the case of one of the most
tions (above 300) from the early 1970s up to the widely used chiral derivatizing reagent, GITC
present time. References to the most important publi- (2,3,4,6-tetra-O-acetyl--D-glucopyrasonyl iso-
cations from this Reld can be found in chapters in thiocyanate, see later) the chiral centres of the
books and review papers listed in Further Reading reagent are in the pyrane ring. The exchange of the
which cover the literature until 1993. The relatively acetyl groups to the bulkier benzoyl groups im-
large number of papers published after 1993 describ- proves the resolution. The position and bulkiness
ing new applications of previously described derivat- of the remote functional groups also inSuences the
ization reagents, moreover the introduction of new separation of the diastereomers. In these instances,
reagents is an indication that this branch of chiral however, the direction of the inSuence is difRcult
analysis has retained some of its importance to the to predict. As an example, the investigation lead-
present day (see Rgure legends). ing to the highly efRcient derivatizing agent for the
amino group (#)-2-methyl-2-naphthyl-1,3-ben-
Important Features of Chiral Derivatization zodioxole-4-carboxylic acid chloride is mentioned.
Reactions and Reagents
It was found that the difference between the bulki-
The presence of a suitable functional group in the ness of the two substituents at C-2 favourably
molecule of the analyte The prerequisite of the use inSuences the separation and with the isomeric
of any kind of enantioseparation based on covalent, 5-carboxylic acids only poor separation is achiev-
chiral derivatization is the presence of at least one able.
functional group (amino, hydroxyl, carboxyl, epoxy, E The formation of hydrogen bonds. For example, in
thiol, etc.) in the molecule of the chiral compound the case of propranolol the secondary amino group
which is capable of reaction with the reactive group of the molecule is transformed by chiral isocyan-
of the derivatizing reagent. ates or isothiocyanates to diastereomeric urea or
thiourea derivatives. The formation of a hydrogen
Good chromatographic properties of the derivatives bond between the carbonyl or thiocarbonyl group
The optimization of the chromatographic parameters of the latter and the free hydroxyl group of pro-
of retention time, peak shape and separability of the pranolol plays an important role in the separation:
enantiomers from other components of the complex the resolution deteriorates with etheriRcation of
mixture by proper selection of the column (usually the hydroxyl group.
reversed-phase but in some cases normal phase col-
umns), and the composition and pH of the eluent is Selection of the elution order This is especially im-
done in the usual ways for achiral HPLC. portant if the aim of the analysis is the determination
of trace level enantiomeric impurity in a drug which is
SufVcient separation of the diastereomers formed administered as the pure enantiomer. In order to have
This is also inSuenced by the above-mentioned sol- the impurity peak appear before the main peak the
Figure 1 Resolution of the four diastereomers of the two racemates of nadolol as urea derivatives after reaction with (R)-(!)-1-
(naphthyl)ethyl isocyanate and their determination in dog plasma extracts. Column YMC-AM-303 ODS(250;4.6 mm, 5-m); mobile
phase, water}acetonitrile (60:40, v/v); flow rate, 1 mL min\1; temperature, 403C; UV detection at 285 nm. (A) Blank control dog plasma;
(B) plasma spiked with 50 ng ml\1 of each diastereomer; (C) plasma obtained 2 h after oral administration of 1 mg kg\1 of racemic
nadolol. Peaks: 1, (S,R )-nadolol; 2, (R,S )-nadolol; 3, (R,R )-nadolol; 4 (S,S )-nadolol. Reproduced with permission from Hoshino M,
Yajima K, Suzuki Y and Okahira A (1994) Journal of Chromatography B 661: 281, copyright Elsevier.
proper chromatographic system (normal- or reversed- example, N-triSuoroacetyl-(S)-(!)-prolyl chloride or

phase) has to be selected; it is even more advantage- anhydride, were found to racemize upon storage.
ous to select a derivatizing agent that is available in Reagents which are available commercially at the
both the R and S forms. Curves a and b in Figure 2 present time fulRl this requirement.
demonstrate this: the elution order of amino acids
changes upon changing from R to S reagent in their
The absence of kinetic resolution and racemization
derivatization with the o-phthalaldehyde-N-butyryl-
The absence of kinetic resolution, e.g. difference be-
cysteine reagent.
tween the reaction rates of the two enantiomers with
the reagent and the enantiomeric stability of the
Unidirectional derivatization reaction taking place analyte and its diastereomeric derivative are impor-
under mild conditions The most widely used reactant prerequisites of the applicability of a reagent to
tions are completed at room temperature within a given purpose. These can be checked by comparing
1 h and the reaction mixture can be injected directly the peak areas of the diastereomers during and after
into the chromatograph. The necessity of heating or the reaction with a racemate. The peak area should be
extraction of the reaction mixture does not preclude close to unity.
a reaction being used and nor does the occurrence of
side reactions, provided that their products do not Good chromophoric or Wuorophoric properties of
interfere with the detection and quantiRcation of the the reagent Although the primary aim of derivatiz-
peaks of the main products and the side reactions do ation in chiral chromatography is the formation of
not show stereospeciRcity. easy to separate diastereomeric derivatives it is
advantageous if, at the same time, the reagent im-
Enantiomeric purity and stability of the derivatizing proves the detectability of the separated enantiomers
reagent The enantiomeric purity of the reagent is by introducing chromophoric or Suorophoric groups
one of the most important factors determining the into their molecules. A typical example for such
success of the determination of the enantiomeric ‘dual-purpose’ Suorophoric reagents is (!)-2-[4-(1-
purity of the analyte. It is evident that when using aminoethyl)phenyl]-6-methoxybenzoxazole. The de-
a homochiral reagent containing its antipode as an tection limit for the enantiomers of 2-phenyl-
impurity, the latter also reacts with the main compon- propionic acid after derivatization with this reagent is
ent of the analyte. This results in a diastereomeric as low as 10 fmol (1.5 pg). Many more examples of
derivative which has the same retention time as that this type are presented in the next section.
originating from the reaction of the main component It is important to note that the UV or Suorescence
of the reagent and the impurity of the analyte. It is characteristics of the diastereomers formed are not
therefore difRcult to estimate if the satellite peak necessarily equal and therefore have to be checked
originates from the impurity of the reagent or from during the validation of a new method by compar-
that of the analyte. If the aim of the study is the ing the spectra and band intensities of the dias-
determination of the enantiomeric purity of a drug tereomers.
administered as pure enantiomer and the test limit for
the antipode is 0.5%, the enantiomeric purity of the Covalent Enantiomeric Derivatization of Some
reagent should be at least 99.9%. If the require- Important Functional Groups
ment for the enantiomeric purity of the drug is higher,
Derivatization of amines The reagents suitable for
the purity of the reagent should be even higher. If
the chiral derivatization of amines can be categorized
the aim is the determination of commensurable
as follows:
amounts of enantiomers (e.g. in biological samples),
1}2% of the enantiomeric impurity in the reagent is
tolerable. Activated carboxylic acids These are usually car-
The enantiomeric purity of the reagent can be boxylic chlorides and the reaction with primary and
checked if the enantiomers of the analyte (or at least secondary amines leads to diastereomeric carbox-
one of them) are available in enantiomerically pure amides. The classical reagent R(#)--methoxy-(tri-
form. The relative peak area of the diastereomeric Suoromethyl)phenylacetyl chloride is still in use for
impurity after the reaction with the reagent will be the preparation of gas chromatographically separable
characteristic of enantiomeric impurity of the re- diastereomers. Some others (‘dual-purpose’ reagents
agent. with strong UV absorption or Suorescence) include
The enantiomeric stability of the reagent is also an 1-(4-nitrophenylsulfonyl)-L-prolyl chloride, dansyl-
important prerequisite to obtain reliable results. For L-proline activated by triethylamine/diethyl phos-
Figure 2 High-performance liquid chromatographic (HPLC) elution profile of the mixture of L and D-amino acids (L : D"2 : 1), glycine
and L-homo-arginine (internal standard) derivatized as the isoindoles with (A) o -phthalaldehyde-N-isobutyryl-D-cysteine and (B)
o -phthalaldehyde-N-isobutyryl-L-cysteine; (C) amino acids from an ethanolic extract of Lactobacillus acidophilus, derivatization with
o -phthalaldehyde-N-isobutyryl-L-cysteine. Column, Hypersil ODS (250;4 mm, 5 m); mobile phase, gradient elution; A"23 mM
sodium acetate (pH 5.95); B"methanol}acetonitrile (600 : 50 v/v), linear gradient from 0% B to 53.5% B in 75 min; flow rate
1 mL min\1; fluorescence detection, 230 nm excitation, 445 nm emission. Reproduced with permission from BruK ckner H, Haasmann S,
Langer M, Westhauser T and Godel H (1994) Journal of Chromatography A 666: 259, copyright Elsevier.
phorocyanidate, N-[4-(6-methoxy-2-benzoxazolyl)] formate is that it is suitable for the derivatization of

benzoyl-L-phenylalanine or -proline, activated by the tertiary amine promethazine via demethylation of
2,2Y-dipyridyl disulRde/triethylphosphine, N-ben- the dimethylamino moiety.
zyloxycarbonyl-L-phenylalanine, activated by acetic
anhydride(#)-2-methyl-2-naphthyl-1,3-benzo-
dioxole-4-carboxylic acid chloride, (S)-(#)-naproxen Isocyanates These reagents form diastereomeric
chloride, (S)-(#)-Sunoxaprofen chloride, (S)-(#)-be- urea derivatives with chiral primary and secondary
noxaprofen chloride, etc. Their strong Suorescence amines. (R)-(!)- and (S)-(#)-1-(1-naphthyl)ethyl
enables the enantiomers of chiral drugs such as isocyanate and (R)--methylbenzyl isocyanate are
tranylcypromine, tocainide, carvedilol, baclofen and among the classical chiral derivatizing reagents.
propranolol to be determined. (The isocyanate, Isocyanate derivatives of the drugs mentioned in one
isothiocyanate and chloroformate derivative of these of the previous sections as the carboxylic chlorides
compounds have also been introduced as chiral de- and (R)-N-3,5-dinitrobenzoyl)phenyl glycine have
rivatizing agents as shown in the subsequent sections.) also been used for the derivatization of -blockers
An on-line solid-phase derivatization reagent and other amines. Eqn [1] shows the reaction be-
is Suorenylmethyloxycarbonyl-L-proline (FMOC-L- tween nadolol and (R)-(!)-1-(1-naphthyl)ethyl
proline), bonded to beads of a styrene-divinylbenzene isocyanate, while the separation of the four dias-
copolymer as the active ester of a 4-hydroxy-3- tereomers of the two racemates is depicted in
nitrobenzophenone moiety. By positioning the HPLC Figure 1. It is remarkable that by selecting a suit-
column after the reaction column, the transformation able reagent and proper chromatographic conditions
of chiral amines (e.g. amphetamine) to their highly not only can the four diastereomers be separated but
Suorescent diastereomeric FMOC-L-prolyl deriva- the interference from dog plasma can also be elimi-
tives and their separation was achieved. nated.
Isothiocyanates Of the isothiocyanates forming

Chloroformates The above-mentioned FMOC thiourea derivatives with primary and secondary
group has been incorporated into another types of amines, GITC is most widely used. If the four acetyl
chiral reagents: (#)-1-(9-Suorenyl)ethyl chlorofor- groups are replaced by benzoyl groups the sensitivity
mate is one of the most widely used derivatization of the detection is greatly improved. Other reagents
reagents for the chiral HPLC of amino acids, - of this type leading to highly Suorescent derivatives
blockers, to form the corresponding carbamate include (R)-(!)- and (S)-(#)-4-(3-isothiocyanatopyr-
derivatives. rolidin-1-yl)-7-nitro-2,1,3-benzoxadiazole and its
An interesting feature of another widely used re- 7-(N,N-dimethylaminosulfonyl) analogue as well
agent of the chloroformate type, (!)-menthyl chloro- as DDITC ((1R,2R)- and (1S,2S)-N-[(2-isothio-
cyanato)cyclohexyl]-3,5-dinitrobenzoylamide, which valinamide analogue of the Marfey reagent gives even

excels with the high chemical stability, high UV better resolution.
activity and the excellent separability of the dia-
stereomeric derivatives. The equation of the reaction o-Phthalaldehyde#chiral thiols This dual derivat-
of GITC and DDITC with the -blocker metoprolol ization reaction leading to Suorimetrically highly ac-
and the separation of the derivatives are depicted in tive isoindole derivatives is another generally used
eqn [2] and Figure 3, respectively. method in the chiral analysis of amino acids (see eqn
N-Haloarylamino acid derivatives Marfey’s reagent [3]). As the chiral thiol N-acetyl-L-cysteine is most
(1-Suoro-2,4-dinitrophenyl-5-L-alaninamide) is one widely used but the use of 2,3,4,6-tetra-O-acetyl-1-
of the most generally used reagents in the analytical thio--D-glucopyranoside and N-isobutyryl-L-cys-
control of racemization during peptide synthesis. The teine (and D-cysteine) have been found to be advant-
peptides are split either by hydrochloric acid or en- ageous in the resolution of the diastereomers. The
zymatically. The nucleophyllic attack of the -amino reaction of amino acids with o-phthalaldehyde and
group of the amino acids on the C}F bond activated the latter reagent is shown in eqn [3], while in Figure 1
by the two nitro groups on the aromatic ring results in the separation of as many as 36 enantiomers of 18
a smooth reaction to form diastereomeric aniline de- amino acids and amino acid analogues as well as glycine
rivatives with good UV detectability (see eqn [4]). The is depicted together with a practical applicaion.
Enzymatic deamination Chiral HPLC of a mixture UV or Suorometric properties are also in use,
of amino acid enantiomers with and without selective e.g. (R)--methyl-4-nitrobenzylamine, (R)-(#)-1-(1-
oxidative deamination of D-amino acids with the aid naphthyl)ethylamine, (!)-1-(1-anthryl)ethylamine,
of the enzymes D-amino acid oxidase and catalase R-(!)- and (S)-(#)-amphetamine, (1R,2R)-(!)- or
enables D-amino acids to be identiRed in the mixture. (1S,2S)-(#)-2-amino-(4-nitrophenyl)-1,3-pro-
panediol, L-leucinamide, L-alanine--naphthylamide,
Derivatization of the carboxyl group
L- or D-O-(4-nitrobenzyl)tyrosine methyl ester, (!)-
EsteriTcation with chiral alcohols (#)- or (!)-2- 2-[4-(1-aminoethyl)-phenyl]-6-methoxybenzoxazole
octanol, (#)-1-phenylethanol, (!)-menthol, (#)- or and other related derivatives where the 1-aminoethyl
(!)-2-butanol are the classical reagents for the chiral group is replaced by L-leucyl or D-phenylglycyl
analysis of carboxylic acids. The reactions usually groups, drug-related amines (Sunoxaprofen amine,
require harsh conditions and for this reason the benoxaprofen amine and naproxen amine), (R)- and
danger of racemization should be taken into consid- (S)-1-(4-dansylaminophenyl)ethylamine, etc. The re-
eration. action of ibuprofen and pranoprofen with the last
reagent and the separation of the diastereomeric car-
Amidation with chiral amines Prior to their reac- boxamide derivatives are shown in eqn [4] and
tions with chiral amines the carboxyl group should Figure 4, respectively.
be activated. Possibilities for this are, e.g. reaction Derivatization of the alcoholic and phenolic hydroxyl
with thionyl chloride to form carboxylic chlor- groups The most frequently used general method
ides, with chloroformates to form mixed anhydrides, for the derivatization of the hydroxyl group of chiral
with 1,1-carbonyldiimidazole to form reactive N- alcohols and phenols is esteriRcation. A great variety
acylimidazoles and with the classical coupling agent of chiral carboxylic acids have been used for this
dicyclohexylcarbodiimide to form the reactive N- purpose such as R(#)- and S(!)--methoxy-(tri-
acylurea derivatives. The classical but still widely Suoromethyl) phenylacetic acid (Mosher’s acid),
used amine reagent is (S)-(!)--methylbenzylamine, R(#)-trans-chrysanthemic acid, (!)-menthenylo-
but several others with excellent separation power, xyacetic acid for the gas chromatographic or HPLC
separation with UV detection and (!)-(1S,2R,4R)-

endo-1,4, 6,7,7-hexachlorobicyclo[2.2.1]-hept-5-
ene-2-carboxylic acid and (S)-(#)-2-tert-butyl-2-
methyl-1,3-benzodioxolo-4-carboxylic acid for
HPLC separation with Suorimetric detection, etc.
Dicyclohexylcarbodiimide can be used as the coup-
ling agent, but the use of in situ transformation of the
acids to their chlorides by the addition of thionyl
chloride is more widespread. The direct enzymatic
D-(#)-glucuronidation reaction of phenols has also
been described.
Acyl chlorides, anhydrides and acyl cyanides can be
used directly. For example (!)-camphanic acid,
R(#)--methoxy-(triSuoromethyl)phenylacetyl
chloride, -naphthylsulfonyl-L-prolyl chloride,
Sunoxaprofen chloride, (R,R)-O,O-diacetyl- (or di-
p-toluoyl-) tartaric anhydride and (!)-2-methyl (or
methoxy)-1,1-binaphthalene-2-carbonyl cyanide
lead to well-separable, UV or Suorimetrically
active diastereomeric ester derivatives with chiral
alcohols.
As an illustration, the reaction of delmopinol
and its analogue used as the internal standard in
the determination of the drug in plasma with the
Figure 3 Resolution of (R,S )-metoprolol derivatized as (R,R )-
DDITC- and GITC-thioureas. Column, Hypersil ODS (125; reagent (R,R)-O,O-di-p-toluoyl-tartaric anhydride
4 mm, 5 m); mobile phase, acetonitrile-20 mM ammonium acet- is shown in eqn [5]. The separation is depicted in
ate 55:45, v/v; flow rate 1 mL min\1 detection at 254 nm. Repro- Figure 5.
duced with permission from Kleidernigg OP, Posch K and Lindner
W (1996) Journal of Chromatography A 729: 33, copyright
Elsevier.
Figure 4 Resolution of (A) (R,S )-ibuprofen and (B) (R,S )-pranoprofen as the carboxamides formed with (S )-1-(4-dan-
sylaminophenyl)ethylamine. Column ODS-80TM (150;4.6 mm, 5 m); mobile phase: 50 mM sodium acetate (pH 6.5)Iacetonitrile (A)
30 : 70, v/v and (B) 45 : 55, v/v, and flow rate 1 mL min\1; fluorescence detection, 338 nm excitation, 535 nm emission. Reproduced
with permission from Iwaki K, Bunrin T, Kameda Y and Yamazaki M (1994) Journal of Chromatography A 662: 87, copyright Elsevier.
Further derivatization reactions include the use amino group can be overcome by using a suitable
of (R)-1-(1-naphthyl)ethyl isocyanate for the catalyst (4-pyrrolidinopyridine). (!)-Menthyl chloro-
derivatization of, e.g. diacylglycerol derivatives to formate, which has already been mentioned, has also
form the corresponding carbamates. The lower reac- found application here as a reagent for amines; the
tivity of the hydroxyl group compared with the reaction products with chiral alcohols are carbonates.
Figure 5 Resolution of (!)- and (#)-delmopinol (peaks 1 and 2) and (!)- and (#)-internal standard (peaks 3 and 4) and their
determination in human plasma extracts. Derivatization with (R,R )-O,O-di-p-toluoyl tartaric acid anhydride to form the esters. (A) Blank
plasma; (B) plasma spiked with 6 ng of each enantiomer; (C) authentic plasma sample. Column, Hypersil ODS (125;4 mm, 5 m);
mobile phase, 100 mM ammonium acetate}acetonitrile 35 : 65, v/v, pH 5.7; flow rate, 0.8 mL min\1; electrochemical detection.
Reproduced with permission from Egginger G, Blaschke E, Lindner W and Olsson A-M (1994) Journal of Chromatography A 666: 275,
copyright Elsevier.
Derivatization of the aldehyde group The import- Pirkle-Type Chiral Columns

ance of this kind of chiral derivatization react-
The purposes of using pre-column derivatization are:
ions is much smaller than those described for amines,
carboxylic acids and alcohols. For example,
E Blocking of the polar groups of the analyte (amino,
aldoses can be transformed with L-cysteine methyl
carboxyl, hydroxyl groups) to avoid strong interac-
ester to diastereomeric thiazolidine derivatives
tions with the polar groups of the chiral selector
which can be separated by GC after trimethyl-
thus improving the separation.
silylation. The two aldehyde groups of gossypol
E The introduction of aromatic rings into the analyte
were transformed with (R)-(!)-2-amino-1-propanol
to enable } interactions to take place between
to the diastereomeric Schiff’s bases separable by
the aromatic moieties of the chiral selector and the
HPLC.
analyte which play a dominant role in the resolu-
tion of the enantiomers.
E Improve UV detectability.
Derivatization of epoxides The oxirane ring of
chiral epoxides is opened by sodium sulRde to form To fulRl these requirements amino groups in the
a vicinal hydroxythiol derivative. In the second step analyte are usually transformed to aromatic acyl de-
of the derivatization reaction the thiol group reacts rivatives by 4-nitro- or 3,5-dinitrobenzoyl chloride or
with o-phthalaldehyde and a chiral amino acid to to aromatic urea or thiourea derivatives by means of
form the diastereomeric isoindole derivative men- phenyl-, 4-Suorophenyl-, 4-methoxyphenyl-, 3,5-
tioned in the section dealing with the derivatization of dinitrophenyl- or 1-naphthyl-isocyanates and iso-
the amino group. Another ring-opening reagent is thiocyanates. Carboxyl derivatives are derivatized
2-propylamine. The secondary amino group of the after activation with thionyl chloride or carbodiim-
1,2-amino alcohol is then reacted with 2,3,4,6-tetra- ides to form anilides, 3,5-dimethylanilides, 3,5-dinit-
O-acetyl (or benzoyl)--D-glucopyranosyl iso- roanilides, naphthylmethylamides, etc. The derivatiz-
thiocyanate to form the diastereomeric thiourea de- ing agents for the hydroxyl group are benzoyl chlor-
rivatives. ide, 3,5-dinitrobenzoyl chloride, 1-naphthyl chloride
to form ester derivatives and 1-naphthyl- or 3,5-dinit-
rophenyl isocyanates to form the corresponding car-
Derivatization of the thiol group The formation of bamates.
diastereomeric isoindole derivatives from chiral
amino acids, chiral thiols and o-phthalaldehyde (al- Cyclodextrin-Bonded Phases
ready mentioned in the section dealing with the de- In the case of enantiomeric HPLC separations on
rivatization of the amino group; see eqn [4]) can also cyclodextrin-bonded stationary phases the necessity
be used for the enantioseparation of thiols. Reagents of pre-column achiral derivatization is not as general
of other types are (R,R)-dinitrobenzoyldiamino- as with the Pirkle-type phases. The introduction
cyclohexyl isothiocyanate and N-[(2-isothiocyanato)- of the aromatic ring into analytes often improves
cyclohexyl]-pivalinoyl amide which transform the their separation. Amino acids are usually derivatized
thiol compound to their diastereomeric dithiocarba- with dansyl chloride, 3,4-dinitrobenzoyl chloride,
mate derivatives. o-phthalaldehyde/2-mercaptoethanol, etc., or with
highly Suorescent reagents such as 9-Suorenylmethyl
chloroformate, 9-Suorenylmethoxycarbonylglycine
chloride, 6-aminoquinolyl-N-hydroxysuccinimidyl
Derivatization of Enantiomers
carbamate. etc.
With Achiral Reagents to Improve
Their Chromatographic Properties 1-Acid Glycoprotein Column
and Their Separation on Chiral In most instances no derivatization is necessary to
Columns obtain good enantioseparation. In the case of car-
boxylic acids, separation is improved by the addition
Introductory Remarks
of ion pairing agents to the eluent. For some basic
Chiral columns enable the direct separation of enan- drugs anionic reagents in the eluent also improves the
tiomers. In many cases the chromatographic proper- column efRciency, peak symmetry and resolution.
ties of the enantiomers can be improved by trans- Covalent derivatization is only seldom used.
forming them to suitable derivatives using achiral Acetylation or formylation of the amino and hy-
reagents. droxyl groups, esteriRcation of the carboxyl group
III / CHIRAL SEPARATIONS / Countercurrent Chromatography 2321
with 2-propanol have been found in some instances to Capillary Electrophoresis; Cellulose and Cellulose Derived
improve the separation. Phases; Chiral Derivatization; Countercurrent Chromato-
graphy; Cyclodextrins and Other Inclusion Complexation
Functionalized Cellulose Phases Approaches; Gas Chromatography; Ion-Pair Chromato-
graphy; Ligand Exchange Chromatography; Liquid
In the overwhelming majority of cases enantiomers Chromatography; Molecular Imprints as Stationary
can be separated on these columns without derivatiz- Phases; Protein Stationary Phases; Supercritical Fluid
ation. In a few instances esteriRcation of the carboxyl Chromatography; Synthetic Multiple Interaction (‘Pirkle’)
group and the introduction of a benzoyl group into Stationary Phases; Thin-Layer (Planar) Chromatography.
the molecule have been found advantageous.
Further Reading
Conclusions and Future Trends
Ahnoff M and Einarsson S (1989) Chiral derivatization. In:
It is indisputable that the direct chromatographic Lough WJ (ed.), Chiral Liquid Chromatography, pp.
enantioseparations based on chiral stationary phases 39}80. Glasgow: Blackie.
has greatly surpassed the importance of indirect sep- Allenmark SG (1989) Chromatographic Enantioseparation:
arations based on covalent chiral derivatization. The Methods and Application. Chichester: Ellis Horwood.
fact that the overwhelming majority of papers report- Arvidsson E, Jansson SO and Schill G (1991) Enantiomeric
ing on the development and application of chiral derivatization. In: Ahuja S (ed.) Chiral Separations by
separations are from these Relds does not necessarily Liquid Chromatography, pp. 126}140. Washington:
reSect the real of the contribution of covalent chiral Americal Chemical Society.
derivatization in the solution of practical problems in Gal J (1993) Indirect methods for the chromatographic
separation of drug enantiomers. In: Wainer IW (ed.),
pharmaceutical and biomedical analysis. In some
Drug Stereochemistry. Analytical Methods and Pharma-
areas (e.g. the determination of small concentrations cology, pp. 65}106. New York: Marcel Dekker.
of D-amino acids in various biological samples, etc.) GoK roK g S (1990) Enantiomeric derivatization. In: Lingeman
the well-established classical methods are still in H and Underberg WJM (eds), Detection-Oriented
use. It is also remarkable that the combination of Derivatization Techniques in Liquid Chromatography,
chiral derivatization with the use of chiral stationary pp. 193}216. New York: Marcel Dekker.
phases has proved to be suitable for the solution of GoK roK g S and Gazdag M (1994) Enantiomeric derivatization
very delicate problems (separation of several enantio- for biomedical chromatography. Journal of Chromato-
mers and diastereomers of drugs with more than one graphy B 659, 51}84.
chiral centre). It should also be noted that the combi- Skidmore MW (1993) Derivatization for chromatographic
nation of chiral chromatography with Suorescence resolution of optically active compounds. In: Blau K
and Halket J (eds), Handbook of Derivatives for
derivatization enables the limit of detection of enan-
Chromatography, pp. 215}252. Chichester: Wiley.
tiomeric impurities to be decreased to the 10 ppm. Souter RW (1985) Chromatographic Separation of
level. These tendencies are expected to continue in the Stereoisomers. Boca Raton: CRC Press.
near future. Stevenson D and Wilson ID (eds) (1989) Chiral Separ-
ations. New York: Plenum Press.
See also: II/Chromatography: Liquid: Derivatization. Zief M and Crane LJ (eds) (1987) Chromatographic Chiral
III/Chiral Separations: Amino Acids and Derivatives; Separations. New York: Marcel Dekker.
Countercurrent Chromatography
Y. Ma and Y. Ito, matography, which can be used for the separation of

National Institutes of Health, a variety of enantiomers by dissolving a chiral selec-
Bethesda, MD, USA tor (CS) in the liquid stationary phase. The method
Copyright ^ 2000 Academic Press eliminates the time-consuming procedure which in-
volves chemically bonding the chiral selector to
a solid support for chiral chromatography. It also
Introduction provides an important advantage over the conven-
Countercurrent chromatography (CCC) is a generic tional methods in that the same column can be used
term for support-free liquid/liquid partition chro- for the separation of various enantiomers simply by
2322 III / CHIRAL SEPARATIONS / Countercurrent Chromatography
dissolving the suitable chiral selector in the liquid analyte has a partition coefRcient ranging from 0.5 to
stationary phase. 1 in the CS-free solvent system. When the organic
In the past various hydrostatic CCC systems, such phase is used as the stationary phase, the chiral selec-
as droplet CCC, rotation locular CCC (RLCCC) and tor may be made hydrophobic by attaching a long
centrifugal partition chromatography (CPC), have hydrocarbon chain to enhance both solubility and
been used for the separation of chiral compounds. retention in the stationary phase.
None of those techniques, however, is considered In each separation, the column is Rrst Rlled about
satisfactory for preparative purposes in terms of half way with the CS-free stationary phase. This is
sample size, resolution and/or separation time. Re- followed by introduction of the CS-containing sta-
cently, the hydrodynamic CCC system termed high tionary phase (about 60% of the total column capa-
speed CCC has remarkably improved the technique city) by discharging the excess amount of CS-free
so that highly efRcient separations can be achieved in stationary phase from the other end of the column. In
a short elution time. This technique has been success- this way some amount of CS-free stationary phase
fully applied to the separation of racemates using (about 10% of the stationary phase retained in the
a Pirkle-type chiral selector. In this method analytical column) remains at the end of the column during
(milligram) and preparative (gram) separations can separation to absorb carried-over CS from the mobile
be performed simply by adjusting the amount of phase which would contaminate the eluted fractions.
chiral selector in the liquid stationary phase in the After the sample solution is injected through the
standard separation column. Gram quantities of en- sample port, the mobile phase is pumped into the
antiomers can be obtained by pH-zone-reRning CCC, column while the column is rotated at the required
a new preparative separation technique recently de- speed, usually 800}1000 rpm. If desired separation
veloped for the separation of ionizable compounds can be carried out by successive injection of samples
(see pH-zone-reRning CCC). In addition, the present without replenishing the CS-containing stationary
method allows computation of the formation con- phase.
stant of the chiral-selector complex, one of the most Figure 1 shows the separation of four pairs of di-
important parameters for studies on the mechanism nitrobenzoyl (DNB)}amino acid enantiomers by
of enantioselectivity. the standard CCC technique using a two-phase sol-
vent system composed of hexane/ethyl acetate/
Standard High speed CCC Technique methanol/10 mM HCl with N-dodecanoyl-L-3,5-
dimethylanilide as a CS in the organic stationary
for Chiral Separation phase. Because of its high hydrophobicity (K'100),
The separations are performed using a commercial this CS has high solubility in the organic stationary
high speed CCC centrifuge equipped with a set of phase and is almost entirely partitioned into the sta-
multilayer coil separation columns. The two-phase tionary phase. All analytes were well resolved in
solvent system is selected in such a way that the target 1}3 h. This chiral selector is similar to the ‘Pirkle’
Figure 1 Separation of four racemic pairs of DNB-amino acids by the standard high speed CCC technique. Experimental conditions:
apparatus: multilayer coil high speed CCC centrifuge with a semipreparative column of 1.6 mm i.d. and 330 mL capacity; solvent
system: hexane/ethyl acetate/methanol 10 mM HCl (8 : 2 : 5 : 5, v/v/v/v), N-dodecanoyl-L-proline-3,5-dimethylanilide (CS) (2 g) was
added to the organic stationary phase (200 mL) as a chiral selector; samples: from left to right, ($)-DNB-phenylglycine, ($)-DNB-
phenylalanine, ($)-DNB-valine, and ($)-DNB-leucine, 5}10 mg of each dissolved in 5 mL of solvent consisting of equal volumes of
each phase; flow rate: 3.3 mL min\1; revolution: 800 rpm; analysis of fractions: optical rotation and circular dichroism; stationary phase
retention: 65% of the total column capacity.
resolution is attained by using a saturated solution of

the CS in the stationary phase in a given column, and
the resolution is further improved by using a longer
and/or wider-bore coiled column which can hold
a greater amount of CS in the stationary phase.
The preparative capability of the system was
investigated on the separation of DNB-leucine enan-
tiomers by varying the CS concentration in the sta-
tionary phase. The results shown in Figure 3 indicate
that the sample loading capacity is mainly determined
by the CS concentration in the stationary phase, i.e.
the higher the CS concentration, the greater the peak
resolution and sample loading capacity. As men-
tioned earlier, the standard HSCCC column (typically
300 mL in capacity) can be used for both analytical
and preparative separation simply by adjusting the
CS concentration in the stationary phase.
pH-Zone-re\ning CCC for Chiral

Separation
pH-Zone-reRning CCC is a powerful preparative
technique that yields a succession of highly concen-
trated rectangular solute peaks with minimum over-
lap where impurities are concentrated at the peak
boundaries. This technique has been applied to the
resolution of DNB-amino acid racemates using a bi-
nary two-phase solvent system where the retainer
acid and chiral selector were added to the organic
stationary phase and the eluent base to the aqueous
mobile phase. Figure 4 shows a typical chromato-
Figure 2 Effects of the amount or concentration of CS on the
separation of DNB}amino acid racemates. In all resolved
gram of a DNB-leucine racemate obtained by pH-
chromatograms, the first peak represents (!)-enantiomer and zone-reRning CCC. The pH of the fraction (dotted
the second peak (#)-enantiomer. Experimental conditions: ap- line) reveals that the peak is evenly divided into two
paratus and column: see the Figure 1 caption; sample: racemic pH zones with a sharp transition. Compared with the
DNB}amino acid mixture consisting of DNB-valine and DNB- standard CCC technique, the pH-zone-reRning CCC
leucine, each 5}10 mg dissolved in 2 mL of solvent (1 mL of each
phase): solvent system: hexane/ethyl acetate/methanol/10 mM
technique allows separation of large amounts of
HCl (8 : 2 : 5 : 5, v/v/v/v); stationary phase: upper organic phase analyte in a shorter elution time.
with CS ranging from 0 to 4 g in 200 mL as indicated; mobile
phase, lower aqueous phase; flow rate: 3 mL min\1; revolution: Determination of Formation Constant
800 rpm. of CS}Enantiomer Complex
The present technique has an advantage over conven-
chiral stationary phase which has been introduced for tional HPLC by allowing the determination of the
the HPLC separation of racemic DNB}amino acid formation constant (Kf) for the CS}enantiomer com-
t-butylamides. For CCC separation an N-dodecanoyl plex which is useful for developing the CS for chiral
group was covalently attached to the CS molecule to chromatography.
increase its hydrophobicity so that it is almost entirely In a schematic view of the portion of the separation
partitioned into the organic stationary phase. column in Figure 5, enantiomers (A# and A ) are
The effect of the CS concentration in the stationary \
partitioned between the organic stationary phase (up-
phase on the peak resolution was investigated by per half) and the aqueous mobile phase (lower half).
a series of experiments as shown in Figure 2. As the In the organic stationary phase enantiomers form
CS concentration was increased, the separation factor a CS complex [CSA ] according to their formation
and peak resolution were increased as indicated by !
constants (Kf ). In this situation:
eqn [6]. The results clearly imply an important tech- !
nical strategy for the present method: the best peak D "([A ]org#[CSA ]org)/[A ]aq [1]
! ! ! !
2324 III / CHIRAL SEPARATIONS / Countercurrent Chromatography
Figure 3 Preparative separation of DNB-leucine racemate by high speed CCC. Experimental conditions: solvent system:
hexane/ethyl acetate/methanol/10 mM HCl (6 : 4 :5 : 5) where the organic stationary phase containing CS at 10 to 60 mM as indicated;
samples: ($)-DNB-leucine 125}1000 mg dissolved in 10}45 mL of solvent. (For other conditions, see the Figure 1 caption.)
D0"[A ]org/[A ]aq [2] tion ratio) and D is the partition coefRcient in a CS-
! ! !
containing solvent system. From these equations, the
Kf "[CSA ]org/[CS]org[A ]org [3] partition coefRcient of the analyte is given by the
! ! !
following equation:
where D0 is the partition coefRcient of the analyte in
a CS-free solvent system (this is also called the parti- D "D0(1#Kf [CS]org) [4]
! !
Figure 4 Separation of DNB-leucine racemate by pH-zone-refining CCC. Experimental conditions: apparatus: see the Figure 1
caption; solvent system: methyl t-butyl ether/water; stationary phase: upper organic phase to which trifluoroacetic acid (40 mM) and CS
(40 mM) were added; mobile phase: lower aqueous phase to which aqueous ammonia was added to 20 mM; sample: ($)-DNB-leucine
2 g; flow rate: 3.3 mL min\1; revolution: 800 rpm; analysis: chirality by analytical HSCCC (see Figure 1) and pH by a portable pH meter.
Note that the analysis of the fraction from the mixing zone (middle chromatogram) shows three peaks corresponding to (!)-
DNB}leucine, impurity and (#)-DNB}leucine from left to right.
In Figure 5, which shows the chemodynamic equi-

librium in a portion of the separation column, if the
analytes are ionizable (e.g. acids) they will be par-
tially dissociated to form anions [A\]aq which are
!
almost insoluble in the organic phase.
In this equilibrium state, the following set of equa-
tions is given for each racemate:
Figure 5 Schematic diagram of simple chemohydrodynamic
D "([A H]org#[CSA H]org)/([A H]aq#[A\]aq)
equilibrium between the racemates (A ) and chiral selector (CS)
! ! ! ! ! !
in the separation column.
[8]
D0"[A H]org/[A H]aq [9]

Here, [CS]org is the difference between the initial con- ! !
centration of the CS and the concentration of the Ka"[A H]aq/[A\]aq[H#]aq [10]
! !
CSA complex, i.e.:
!
Kf "[CSA H]org/[CS]org[A H]org [11]
! ! !
[CS]org"[CS]initial![CSA ]org [5]
! where D , D0, Ka, and Kf represent the partition
! !
coefRcient, the partition ratio, the dissociation con-
When [A ]org[CS]initial, [CS]org approaches [CS]initial
! stant, and the CSdcomplex formation constant for
hence, eqn [4] may be rewritten:
each racemate, respectively. From these equations,
we obtain:
D +D0(1#Kf [CS]initial) [6]
! !
pHZ "pKa#logD0/D )(1#[CS]orgKf )!1
The validity of eqn [6] has been examined by ! ! !
a series of experiments, the results of which indicated [12]
that the separation factor (D#/D ) is increased as
\ In pH-zone-reRning CCC, the peak resolution is
expected by increasing the concentration of chiral
mainly determined by the difference in pH between
selector in the stationary phase (Figure 2).
For computation of Kf, eqn [6] can be modiRed
into a more convenient form:
(D !D0)/D0+Kf [CS]initial [7]

! !
where D and D0 can be computed from the
!
chromatograms obtained with and without the chiral
selector in the stationary phase.
Using eqn [7] the formation constant (Kf ) has
!
been determined by a series of experiments where
small amounts (0.1}0.2 mg) of enantiomers were
separated at various CS concentrations in the
stationary phase. Figure 6 is drawn by plotting the
(D !D0)/D0 values from each enantiomer against
!
the initial CS concentration in the stationary phase
where the formation constant is computed from the
slope of the straight line. These results indicate that
the method is useful for computing the formation
constants of various analyte}CS pairs.
General Chemodynamic Model in Chiral CCC

This second model deals with more generalized con- Figure 6 Determination of formation constant of CS-DNB-
amino acids by the standard HSCCC technique. The diagram was
dition where the ionic analytes are dissociated in the
produced by plotting (D !D0)/D0 (eqn [7]) against the initial CS
aqueous mobile phase. This approach can be useful !
concentration in the organic stationary phase. The slope of each
for predicting the feasibility of chiral resolution by line indicates the formation constant (Kf) of the corresponding
pH-zone-reRning CCC. enantiomer.
2326 III / CHIRAL SEPARATIONS / Crystallization
the two zones, i.e.: 4. The method is very useful for investigation of the
enantioselectivity of the chiral selector including
pHZ "log([CS]org/Z#Kf##1!D0/Kr)/
! determination of formation constant and separ-
([CS]org/Z Kf #1!D0/Kr) [13] ation factor.
\ \
5. pH-Zone-reRning CCC can be applied to chiral
where [CS]org/Z# and [CS]org/Z are the free CS con-
\ separation allowing a large scale separation in
centrations in the A# and A zones, respectively.
\ a shorter separation time.
When Kf#'Kf , [CS]org/Z#([CS]org/Z ([CS]initial.
\ \
Eqn [13] indicates that chiral resolution can be im-
See also: II/Chromatography: Chromatography: Instru-
proved by increasing D0/Kr and/or choosing the CS mentation. Chromatography: Liquid: Column Techonol-
with a large Kf#/Kf value. It also implies that in-
\ ogy. III/Chiral Separations: Amino Acids and Deriva-
creasing the CS concentration will yield higher peak tives; Capillary Electrophoresis; Cellulose and Cellulose
resolution. Derived Phases; Countercurrent Chromatography; Crys-
tallization; Cyclodextrins ad Other Inclusion Complexation
Advantages of Chiral CCC
Approaches; Gas Chromatography; Ion-Pair Chromatog-
Countercurrent chromatography can be applied to raphy; Ligand Exchange Chromatography; Liquid Chro-
the separation of enantiomers by dissolving a suitable matography; Molecular Imprints as Stationary Phases;
chiral selector in the liquid stationary phase in anal- Protein Stationary Phases; Supercritical Fluid Chromatog-
raphy; Synthetic Multiple Interaction (‘Pirkle’) Stationary
ogy to binding the CS to the solid support. The
Phases; Thin-Layer (Planar) Chromatography. Zone
HSCCC technique has the following advantages over
Refining Countercurrent Chromatography.
the conventional chromatographic technique using
a solid stationary phase:
Further Reading
1. The method permits repetitive use of the same
column for a variety of chiral separations by Ma Y and Ito Y (1995) Chiral separation by high-speed
choosing appropriate chiral selectors. countercurrent chromatography. Analytical Chemistry
2. Both analytical and preparative separations can be 67 (17): 3069-3074.
Ma Y, Ito Y and Foucault A (1995) Resolution of gram
performed in a standard CCC column by adjusting
quantities of racemates by high-speed countercurrent
the amount of CS in the liquid stationary phase.
chromatography. Journal of Chromatography A 704:
The method is cost effective especially for large 75-81.
scale preparative separations. Ma Y, Ito Y and Berthod A (1998) Chiral separation by
3. The separation factor and peak resolution can be high-speed countercurrent chromatography. In: Menet
improved by increasing the concentration of CS in JM and Thiebaut D (eds), Countercurrent Chromato-
the stationary phase. graphy. New York: Marcel Dekker.
Crystallization
A. Collet, ED cole Normale SupeH rieure de Lyon, The crystallization methods comprise several vari-
Lyon, France ants: (1) direct crystallization of enantiomer mix-
Copyright ^ 2000 Academic Press tures, (2) separation of diastereoisomer mixtures
} the so-called classical resolution } and (3) crystalli-
zation-induced asymmetric transformation. The Rrst
Methods for obtaining optically active compounds in two were discovered by Louis Pasteur in 1848 and
enantiopure form are commonly classiRed into three 1853, respectively. The third was Rrst reported in
categories: utilization of chiral pool starting materials 1913. At the turn of the 21st century, these methods
(stereoselective multistep synthesis), creation of are in wide use in laboratory-scale separations as well
chirality from achiral precursors (asymmetric syn- as in industry for the preparation of pharmaceutical
thesis) and separation of racemates into their enan- and agrochemical active principles. To cite but a few
tiomer constituents (resolution). This last method can examples, hundreds to thousands of tonnes of com-
be carried out in a variety of ways: crystallization mercially important materials such as (!)-menthol,
processes, chromatography of racemates on chiral (S)-(#)-naproxen, L--methyldopa, D-phenylglycine
stationary phases and kinetic resolution mediated by and the pyrethroid insecticide deltamethrin are pro-
chiral reagents or enzymes. duced by such methods.
III / CHIRAL SEPARATIONS / Crystallization 2327
In the context of enantiomer separations crystalli-

zation techniques are attractive for several reasons.
Firstly, in many instances they are more straightfor-
ward and more economical than any other method.
Secondly, these methods, once considered old-
fashioned, have, during the past two decades, been
greatly improved in their rationale and efRciency, as
a consequence of a better knowledge of the properties
upon which separations by crystallization are based:
identiRcation of racemate types, utilization of phase
diagrams describing solid}liquid equilibria. Thirdly,
these methods apply not only to the resolution of
racemates (i.e. 1 : 1 mixtures of d and l enantiomers),
they can also be used for obtaining pure enantiomers
from nonracemic (partially resolved) mixtures, re-
gardless of their origin. And Rnally, the so-called
crystallization-induced asymmetric transformation
overcomes the inherent 50% yield limitation of a res-
olution, permitting up to 100% of a racemic material
to be converted into one enantiomer in a single stage.
Enantiomer Mixtures
This section deals with crystallization methods allow-
ing pure enantiomers to be prepared from partially
resolved mixtures or from racemates without the help
of any chiral auxiliary reagent.
Phase Diagrams of Enantiomer

Systems
We call enantiomer systems mixtures containing the
two mirror-image d and l components in any ratio,
either without solvent (binary (d, l) mixtures), or in
the presence of a solvent (ternary (d, l, ) mixtures).
Depending on the nature of the crystal phases which
may coexist with the liquid (melt or solution),
three categories of enantiomer systems have been
identiRed.
In the Rrst category, which represents 5}10% of
cases, the enantiomers crystallize separately from one
Figure 1 Binary melting point phase diagram for enantiomer
another (homochiral crystallization, spontaneous res- systems. (A) Conglomerate; (b) racemic compound with
olution). A solid (d, l) system then consists of a mech- TR(TA, eutectic at x"0.3 and 0.7; (c) racemic compound with
anical mixture of d and l crystals in a ratio corre- TR'TA, eutectic at x"0.1 and 0.9. On recrystallization, a solid
sponding to their respective mole fractions (x; 1!x). mixture M in which the mole fraction of the major enantiomer
(here d) is 0.8 (enantiomeric excess ee"60%) can give this enan-
The melting point phase diagram shows a simple
tiomer in pure form in cases (A) and (B), the maximum yield being
eutectic at x"0.5 (racemate composition). Such given by the ratio NA/EA; in case (C) only the racemic compound
a (d, l) system is said to be a conglomerate of enantio- can be obtained (see text).
mers (Figure 1A). In a conglomerate, the racemate
has a melting point about 30 K lower than those of its
pure enantiomers. Its solid-state infrared spectrum is behave ideally in the liquid state, the liquidus curves
identical with the solid-state spectrum of a pure enan- of a conglomerate phase diagram can generally be
tiomer, because the infrared spectrometer does not accurately calculated by means of the SchroK der}van
make a difference between right- and left-handed Laar equation, where TA and HA are the temper-
crystals. Since most binary mixtures of enantiomers ature and enthalpy of fusion, respectively, of a pure
enantiomer, and R is the gas constant. This equation behaviour of enantiomer mixtures on crystallization
gives the melting temperature T of a mixture in can be simply derived from the binary phase diagrams
which x represents the mole fraction of the major depicted in Figure 1. The component which can be
enantiomer: obtained upon crystallization will be either the ra-
cemic compound, or the enantiomer in excess, de-

HA 1 1 pending on the composition of the starting mixture
ln x" !
R TA T with respect to the closest eutectic of the phase dia-
In the second, main category of enantiomer sys- gram. Thus, recrystallization of mixture M, having
tems, representing 90}95% of cases, the d and l enan- x"0.8 (ee"60%) can give the major enantiomer in
tiomers crystallize together to form an ordered 1:1 pure form in cases (A) and (B). The maximum pos-
compound, called a racemic compound (heterochiral sible yield is given by NE/EA } 60% in case (A), and
crystallization). Typical binary phase diagrams corre- only 33% in case (B). In case (C), recrystallization of
sponding to this case are shown Figure 1(B) and (C). M can only afford the racemate in a maximum yield
The melting point TR of the racemic compound can NE/RE. In this case the enantiomeric enrichment
be either lower or higher than that of the pure enan- takes place in the liquid, but under equilibrium condi-
tiomers (however, on average, the difference between tions the highest attainable purity is that of the eutec-
TR and TA rarely exceeds $20 K). The solid-state tic. In other terms, in a partially resolved conglomer-
infrared spectrum of a racemic compound is different ate, it is always possible to recover the major enan-
from that of a pure enantiomer. This is a simple way tiomer in a yield equal to the ee of the starting mater-
of distinguishing between a racemic compound and ial, whereas in a racemic compound the outcome of
a conglomerate (in addition to the melting point cri- the crystallization will depend on the enantiomeric
terion). In such systems, a partially resolved solid composition of the starting material and on that of
consists of a mechanical mixture of two crystal eutectic; even in relatively favourable cases such as
phases, the racemic compound, in amount that depicted in Figure 1(B), the maximum possible
r"2(1!x), and the enantiomer in excess, in amount yield will be lower than the ee of the starting material.
ee"2x!1. In Figure 1(B) and (C) the liquidus
curves for the enantiomer branches are calculated by Ternary Phase Diagrams
means of the SchroK der}van Laar equation, while the
For applications in which accurate quantitative data
racemic compound branch is calculated by a similar
are required (e.g. optimization of large-scale recrys-
equation Rrst given by Prigogine and Defay, where
tallizations), it is preferable to utilize ternary (solubil-
TR and HR are the temperature and enthalpy of
ity) phase diagrams (d, l, ). Solubility phase dia-
fusion, respectively, of the racemic compound:
grams are 3D triangular prisms where each face is
a binary melting point phase diagram, (d, l), (d, )

2HR 1 1
ln 4x(1!x)" ! and (l, ). It is customary to convert this 3D repres-
R TR T
entation into a 2D one by keeping one of the variables
In the last category of enantiomer systems, called Rxed, most often the temperature. Accordingly,
pseudoracemates, there is almost no chiral discrim- a horizontal slice of the prism generates the usual
ination between the d and l species which co-crystal- triangular isotherm describing the solid}liquid phase
lize more or less at random within the same lattice to equilibria at given T. This is illustrated in Fig-
form a solid solution. In a pseudoracemate, a par- ure 2(A), for a conglomerate. In order to facilitate the
tially resolved sample consists of a single crystal con- use of such triangular isotherms, it is convenient to
taining the d and l enantiomers in a ratio (x; 1!x). adopt a system of cartesian axes (C, E), where C rep-
Such systems are, fortunately, not common. They resents the concentration and E the excess of enan-
may occur with rod-shaped or quasi-spherical mol- tiomer, expressed in the same dimensionless unit
ecules (e.g. camphor). They do not lend them- (generally in weight %); C and E are deRned as
selves easily to separation by crystallization tech- follows:
niques, and for this reason they will no longer be
considered here. d#l d!l
C" ;100 E" ;100
d#l# d#l#
Solubility Rules Derived from
Typical ternary isotherms for a conglomerate and
Binary-phase Diagrams a racemic compound are shown in Figure 2(B) and
In the absence of temperature- or solvent-dependent (C), respectively. In a conglomerate the solubility of
polymorphism, the solubility rules that govern the the racemate is normally twice the solubility of each
Figure 2 (A) 3D-representation of a d, l, system and the corresponding 2D-isotherm, for a conglomerate. The composition of
a ternary system X is conveniently defined by its cartesian coordinates (C, E ). Note that E/C"ee, the usual enantiomeric excess,
defined as ee"(d!l )/(d#l ); (B) and (C) represent ternary isotherms that correspond roughly to the binary phase diagrams (A) and
(B) of Figure 1. On recrystallization, mixture M will only afford the major enatiomer in pure form if the composition of the ternary system
is comprised between N and P, the maximum yield being obtained for system N (see text). In diagram (C), e represents the ee of the
ternary eutectic E.
individual enantiomer, unless special circumstances, signiRcantly with temperature, except in the case of
such as common ion effects or aggregation phe- polymorphism or of solvation.
nomena, decrease or increase the solubility ratio with This is why information on the type of enantiomer
respect to its normal value. Figure 2(B) and (C) system and, in particular, on the location of the eutec-
roughly correspond to the binary phase diagrams of tics in racemic compound phase diagrams is of great
Figure 1(A) and (B), respectively. On recrystallization importance when the puriRcation of partially resolved
of mixture M, the major enantiomer d can be ob- mixtures by crystallization techniques is under-
tained between N and P, for instance from system O. taken. When the phase diagram of the considered
However, the best yield Y in pure d will be obtained system is not favourable (e.g. as in Figure 1C), recrys-
from system N, at concentration CN (in weight %): tallization should be avoided and it may be advisable
to postpone the puriRcation to a next step, or to seek
NE 100 derivatives having more favourable phase diagrams.
Y" ;
AE CN These concepts are particularly useful for the Rnal
puriRcation of enantiomers prepared by asymmetric
For a conglomerate the above calculated yield is the synthesis or chiral chromatography.
same as that derived from the binary phase diagram,
Y"NE/AE. For a racemic compound, Y calculated
from the ternary isotherm is usually very close to that
Direct Crystallization of Racemates
derived from the binary phase diagram because in Separation of enantiomers by direct crystallization of
such systems the ee of the eutectic does not change their racemate is an attractive method, because no
costly chiral auxiliary is required. Such resolutions

only apply to conglomerates. Essentially two main
types of methods are then available. In the Rrst of
these the two enantiomers are allowed to crystallize
simultaneously from a solution or melt which is al-
ways close to the racemic composition. In the second,
called entrainment, the crystallization of a single en-
antiomer is promoted from a supersaturated solution
or supercooled melt which is not allowed to come to
equilibrium.
Simultaneous Crystallization
Although hand-sorting of the d and l crystals (as
originally performed by Pasteur on sodium am-
monium tartrate) can occasionally be a useful tech-
nique, the localization of the crystallization of each
enantiomer on suitably disposed d and l seeds repres-
ents an improvement of greater practical interest. In
small-scale preparations, one can utilize a small num-
ber of d and l seeds sufRciently separated from one
another in the same crystallizer, and the large crystals
which are eventually formed are collected manually.
Large-scale applications utilize pairs of crystallization
chambers, each one being loaded with a large amount
of seeds of one enantiomer. The system is then con-
tinuously fed with a circulating, slightly super- Figure 3 Principle of the entrainment process. In the cycle
saturated, racemic solution. Such processes have been MNPQ, MN and PQ correspond to the crystallization of enantio-
proven to be valuable for continuous large-scale pro- mers l and d, respectively; NP and QM correspond the loading of
racemic material. In commercial processes, 20}90 crystallizations
ductions such as those developed by Haarmann
are commonly performed.
& Reimer for (!)-menthol (resolved as its benzoate),
and Merck for L--methyldopa (resolved as its nitrile
precursor).
weight of racemic material is added to the Rltrate and
dissolved by heating. This results in a new super-
Entrainment saturated systems of composition P, symmetrical to
The entrainment method takes its origin in experi- M, where the d enantiomer is now in the same excess
ments performed by Gernez in 1866, showing that as was the l in the previous experiment. Seeding with
a supersaturated solution of sodium ammonium tar- the d form and crystallization up to Q then yields
trate, when seeded with a particle of (#) salt, only +2E g% of d, and addition of the same weight of
yielded crystals of that salt. The resolution by entrain- racemate allows the return to M, and so forth.
ment is a batch process, which rests on the control of In practice, the economics of the process depends
the crystallization rates of the two enantiomers, and on the amount of material collected after each crystal-
implies the utilization of ternary phase diagrams, lization, which should represent at least 10}15% of
such as that shown in Figure 3. A solution M, super- the solute, and on the number of cycles which can be
saturated with respect of both enantiomers, and con- performed; this number is limited by the build-up of
taining a small excess (E g%) of one of them (here, l) impurities which follows the addition of fresh ra-
is seeded with crystals of that enantiomer. Crystalli- cemate at each step, and which may eventually dis-
zation is then allowed to proceed until the solution turb the crystallization kinetics. The Roussel-Uclaf
has reached composition N. At this point, the rota- and Zambon processes for the manufacture of the
tion of the solution is approximately equal and oppo- chloramphenicol and thiamphenicol intermediates
site in sign to that of the starting solution, and the shown in Figure 3 are well-known industrial applica-
l enantiomer that has crystallized amounts to twice its tions of resolution by entrainment. The method is
excess in the original solution M (i.e. +2E g%). At also of great value for laboratory-scale resolutions,
this stage, the crystals are separated off, and the same especially at the 100 g to kg scale.
For low-melting conglomerates, the entrainment lows that all physical properties involving the crystal
can also be effected without solvent, in a supercooled themselves and the crystal/liquid or crystal/gas equi-
melt. Such processes are easily understood by means libria, such as melting points, solubilities, sublima-
of the melting point phase diagrams, by considering tion properties, crystal densities, etc., are different
the metastable extension of the liquidus curves below for the p and n species of a given pair. However,
the eutectic temperature. the possibility of separating p and n diastereoisomers
by crystallization of their 1 : 1 mixture, resulting
from the reaction of a racemate with a resolving
Diastereoisomers agent, does not depend only on the properties of
The most widely used method for the separation of the pure p and n compounds: it rests primarily on
enantiomers, often called classical resolution, rests on the existence of favourable solid}liquid phase equi-
the crystallization of diastereoisomers formed from a libria for the binary (p, n) or ternary (p, n, )
racemate (dl) and an enantiopure reagent (say, D), systems. The phase diagrams of diastereoisomer
called a resolving agent: systems (Figure 4) are basically similar to those of
enantiomer systems. There exist, however, several
dl#DPdD#lD important differences between them: (1) the phase
diagrams of diastereomer mixtures are not symmetri-
It is convenient to designate with letters p (positive) cal; (2) in contrast to enantiomers, diastereoisomers
and n (negative) the diastereoisomers resulting from preferentially form eutectics or solid solutions (Fig-
reaction of the two constituents of like sign and ure 4(A) and (C), respectively); (3) the occurrence
opposite signs, respectively. In this convention, no of 1 : 1 (pn) addition compounds depicted in Fig-
account is taken of the sign of rotation of the dia- ure 4(B) is by far less than that of (dl) racemic
stereoisomers themselves, which, if needed, can be compounds.
speciRed by adding a # or ! subscript to the p and Only diastereoisomer systems forming eutectic
n descriptor. Accordingly, the above reaction yields phase diagrams are suitable for resolution by crystal-
a mixture of p (dD) and n (lD) diastereoisomers, lization. In the solubility isotherm of Figure 4(A),
which, for instance, can both be dextrorotatory, i.e. 1 : 1 (p, n) mixtures will afford the pure less soluble
p# and n#. The opposite enantiomer (L) of the diastereoisomer (here p) if the crystallization is car-
resolving agent would then afford with the same ried out between N and P, for instance from solution
racemate the diastereoisomer pair p (lL) and O. The best yield however will be obtained from
\
n (dL), mirror image of p# and n# (Marckwald solution N. This yield is given by:
\
principle). Note that the reciprocal resolution of DL
by, for instance, d, would yield a mixture of p# (dD) NE 100
Y" ;
and n (dL). Ep CN
\
Diastereoisomeric salts, formed from simple acid}
base reactions, are central to such resolutions, al- where CN is the concentration of solution N. Since the
though covalent diastereoisomers (esters, amides, p/n ratio in the ternary eutectic is usually close to that
etc.) are also occasionally resolvable by crystalliza- of the binary eutectic, Y is not very different from
tion. To cite but a single example, the DSM company RE/AE in the melting point phase diagram. The best
resolves DL-phenylglycine by crystallization of its dia- systems are those in which the eutectic is close to the
stereoisomeric salts with (#)-10-camphorsulfonic edge of the phase diagram, and the maximum value
acid, to produce the D-enantiomer required for the of Y"50% is approached when E is closed to one of
manufacture of the antibiotic ampicillin, at a scale of the pure components. The occurrence of such good
more than 1000 tonnes per year. systems has been estimated at 20}25% of dia-
Diastereoisomeric inclusion complexes, in which stereoisomer mixtures. It is important to recognize
the resolving agent is a chiral crystalline host lattice, that in such cases one or two crystallizations are
represent an interesting alternative for the resolution normally sufRcient to obtain the less soluble dias-
of substances that cannot form salts, or do not pos- tereoisomer in pure form.
sess functional groups suitable for formation of Systems in which a 1 : 1 [p, n] compound exists, as
covalent diastereoisomers. in Figure 4(B), are totally unsuitable for resolution
because crystallization of a 1 : 1 mixture will afford
the 1 : 1 compound. Some resolution is possible with
Phase Diagrams and Solubility Rules systems forming a solid solution, as shown in Fig-
Diastereoisomers p and n are distinct compounds and ure 4(C), providing that the solubility difference be-
exhibit different structures in the crystal state. It fol- tween the pure components is sufRciently large. Most
Figure 4 Melting point (left) and solubility (right) phase diagrams for diastereoisomer systems. (A) eutectic; (B) 1 : 1 addition
compound; (C) solid solution. The arrows indicate the nature of the solid obtained from crystallization of a system of composition O.
often, however, the enrichment is very modest and Resolving Agents

utilization of systematic fractional crystallization
techniques is required. These techniques are very tedi- In order to set up practical process for the separation
ous and time-consuming and this is why resolution of of diastereoisomeric salts, initially a system forming
(p, n) systems forming solid solutions is not recom- a eutectic must be selected. This leads to the choice of
mended unless alternative methods cannot be found. the resolving agent and of the solvent. Next the crystal-
Note that diastereoisomeric lattice inclusion com- lization conditions are optimized by means of phase
plexes are more prone to form solid solutions than the diagrams along the lines discussed above. A classical
other types of diastereoisomers and often need re- resolving agent is a chiral acid or base available in bulk
peated crystallizations to reach good enrichments. In quantities at low price, and which has a propensity to
such cases rather than purify the diastereoisomers, it form crystalline diastereoisomers when combined with
may be advisable to complete the puriRcation on the racemic bases or acids. Table 1 lists some of the most
partially resolved enantiomers, along the lines in- common resolving agents for salt formation. Other,
dicated above. more specialized, resolving agents for the separation of
Table 1 Common resolving agents
Acids Bases
Tartaric acid (#), (!) -Methylbenzylamine (#), (!)

Dibenzoyltartaric acid (#), (!) Ephedrine (#), (!)
Mandelic acid (#), (!) 2-Amino-1-butanol (#), (!)
Camphoric acid (#) Quinine (!)
Malic acid (#), (!) Quinidine (#)
10-Camphorsulfonic acid (#), (!) Cinchonine (#)
-Bromo--camphorsulfonic acid (#), (!) Cinchonidine (!)
Glutamic acid (#), (!) Brucine (!)
Aspartic acid (#), (!) Yohimbine (#)
-Camphanic acid (#), (!) Dehydroabietylamine (#)
1,1-Binaphthyl-2,2-diyl-hydrogen N-Methyl-D-glucamine (!)
phosphate (#), (!)
diastereoisomers by chromatography or for analytical tion (i.e. dD & lD) at a rate faster than the crystalli-
purposes (nuclear magnetic resonance spectroscopy) zation of the least soluble one. Such a phenomenon
are not mentioned here. For important commercial was Rrst reported by Leuchs in 1913; he isolated in
applications, it may be appropriate to design new 94% yield a single diastereoisomeric salt after reac-
resolving agents: for example, the N-alkyl-D- tion of an easily racemizable racemic acid with
glucamine family (prepared from D-glucose) has been brucine. In addition to the requirement that the
developed by Syntex as a substitute for cinchonidine diastereoisomers must epimerize easily, processes
for the large-scale resolution of naproxen (over 1000 of this type can only be successful if the dias-
tonnes per year). There are no clear guidelines allow- tereoisomer system belongs to the eutectic type. In
ing the prediction of a good resolving agent for Figure 5, the starting material has composition
a given substrate. This is not really a serious problem O (overall concentration C0). In the liquid phase, the
because the number of available resolving agents, composition of the diastereoisomer mixture in chem-
being very limited, generally allows systematized pro- ical equilibrium is indicated by L, which in the case
tocols to identify the best ones very quickly. shown is slightly shifted towards the n species. The
Salts usually need polar solvents to crystallize, and Rnal equilibrium state is featured by N, representing
a statistical survey of solvents used in over 800 such the overall composition of the system at the end of the
resolutions indicate that anhydrous or aqueous process. System N consists of pure solid p in equilib-
acetone and alcohols (ethanol, methanol, 1- and 2- rium with liquid L. The yield in diastereoisomer p is
propanol, 1-butanol), and water feature in 80% proportional to LN/Lp, and can be derived
of cases. The presence of water may be necessary from C0 and the composition of the liquid in equilib-
whenever the salts crystallize as hydrates. This is why rium (Ceq):
95% ethanol (rather than absolute ethanol) is to be
preferred during the selection of a resolving agent. (C0!Ceq) 100
Y" ;
Interesting achievements in the area of chiral host (100!Ceq) C0
lattices have been reported during the last decade.
A small number of chiral substances that form dia- Accordingly, yields approaching 100% can only be
stereoisomeric crystalline inclusion complexes with obtained in poor solvents, in which Ceq is very small.
a variety of guests have been identiRed. In addition to Typically, Ceq"2 g% and C0"20 g% would give
the alkaloid brucine, which has been utilized for re- a Y of nearly 92%. In contrast to the case of a
solving, at least partially, chiral halogenoalkanes (in- classical resolution, the yield does not depend directly
cluding CHFClBr) and, more recently, acetylenic on the location of the eutectic. However, it is the
alcohols and cyanohydrins, several new hosts have distance between the eutectic E and the equilibrium
been discovered, among which readily available diols solution L which provides the driving force for the
prepared from tartaric acid seem to have the greatest process, and this is why, again, a eutectic close to the
potential as resolving agents. edges of the phase diagram is preferable.
There are many examples of such processes in
Crystallization-induced Asymmetric industry. They can be performed on diastereo-
isomeric salts (e.g. phenylglycine camphorsulfonate)
Transformations as well as on covalent diastereoisomers (e.g. in the
The 50% yield limit of a resolution can be exceeded if Roussel-Uclaf synthesis of the insecticide deltameth-
the diastereoisomers can be equilibrated in the solu- rin).
Figure 5 Crystallization-induced asymmetric transformation of diastereoisomers. In the solubility isotherm shown, L represents the
composition of the liquid in which the two diastereoisomers are in chemical equilibrium (here with a slight excess of n). This liquid is itself
in physical equilibrium with the solid, consisting of pure p. After completion of the chemical and physical equilibrations, the starting
system O will be converted into the final system N, consisting of solid p in equilibrium with liquid L.
Crystallization-induced asymmetric transforma- Collet A (1989) Optical resolution by crystallization

tions of enantiomers forming conglomerates can also methods. In: Krstulovic AM (ed.) Chiral Separations by
be carried out by combining entrainment techniques HPLC. Chichester, UK: Ellis Horwood.
with simultaneous racemization of the substrate in Collet A (1996) Optical resolution. In: Reinhoudt DN (ed.)
Comprehensive Supramolecular Chemistry, vol. 10.
the solution. Although a number of examples have
Oxford: Pergamon.
been described, and a number of patents have been Collet A, Brienne M-J and Jacques J (1980) Optical resolu-
Rled, only a few processes of this type have reached tion by direct crystallization of enantiomer mixtures.
commercial application. Chem. Rev. 80: 215.
Collet A, Ziminski L, Garcia C and VigneH -Maeder F (1995)
Future Prospects Chiral discrimination in crystalline enantiomer systems:
facts, interpretations, and speculations. In: Siegel JS (ed.)
The understanding of phase equilibria in enantiomer Supramolecular Stereochemistry. NATO ASI Series.
and diastereoisomer systems is a central question, not Dordrecht: Kluwer.
only in view of its practical relevance to separation Collins AN, Sheldrake GN and Crosby J (eds) (1994)
processes, but also because it may shed light on some Chirality in Industry. Chichester, UK: Wiley. Ibid.
of the mechanisms governing molecular recognition (1997) Chirality in Industry II.
in condensed matter. Among the numerous unsolved Eliel EL, Wilen SH and Mander N (1994) Stereochemistry
questions, that of the prediction of racemate types is of Organic Compounds. New York: John Wiley.
perhaps one of the most challenging. Recently, it has Jacques J, Collet A and Wilen SH (1981) Enantiomers,
Racemates, and Resolutions. New York: John Wiley.
been shown that, in theory, conversion of racemic
Ibid. (1994) Reissue with corrections. Malabar, Florida:
compounds into conglomerates could be achieved for Krieger.
some racemates by crystallization under high pres- Toda F (1996) Diol, bisphenol, and diamide host com-
sure. An experimental veriRcation of this prediction pounds. In: MacNicol DD, Toda F and Bishop R (eds)
would be of great importance in extending the possi- Comprehensive Supramolecular Chemistry, vol. 6.
bilities of resolving racemates by direct crystallization. Chichester, UK: Pergamon.
Weber E (1996) Shape and symmetry in design of new
See Colour Plate 67. hosts. In: MacNicol DD, Toda F and Bishop R (eds).
Comprehensive Supramolecular Chemistry, vol. 6.
Further Reading Chichester, UK: Pergamon.
Ahuja S (ed.) (1997) Chiral Separations, Applications and Wilen SH, Collet A and Jacques J (1977) Strategies in
Technology. Washington, DC: American Chemical Society. optical resolutions. Tetrahedron 33: 2725.
III / CHIRAL SEPARATIONS / Cyclodextrins and Other Inclusion Complexation Approaches 2335
Cyclodextrins and Other Inclusion Complexation Approaches

J. Dingenen, Janssen Research Foundation, Belgium One of the Rrst effective uses of cyclodextrins in
chromatography was as mobile phase additive in
thin-layer chromatography. In the mid-1980’s a pro-
cess was developed to produce stable cyclodextrin
high performance liquid chromatographic (HPLC)
Introduction phases. Nowadays, native -, - and -cyclodextrin,
as well as a variety of derivatized cyclodextrin HPLC
Cyclodextrins are cyclic nonreducing oligosacchar-
columns are commercially available. Also many cyc-
ides containing from six to twelve glycose units in
lodextrin-based capillary gas chromatography (GC)
a C-1 chair conformation, bonded through -(1,4)
columns are on the market. With the growing import-
linkages (Figure 1). The glycopyranose units are
ance of capillary electrophoresis in chiral separations,
arranged in the shape of a hollow truncated cone. The
the use of native cyclodextrin and cyclodextrin deriv-
larger opening of the molecule is surrounded by
atives as an electrolyte additive steadily increases.
the secondary (C-2 and C-3) hydroxyl groups, while
the primary (C-6) hydroxyl groups constitute the
smaller end of the cone. Cyclodextrins in HPLC Applications
Since all the primary and secondary hydroxyl Native cyclodextrin HPLC columns were deliberately
groups are located at the outside of the molecule, the designed to be used in the reserved-phase mode of
exterior faces are hydrophilic. The interior cavity, operation, in order to take full advantage of the
essentially comprised of methylene linkages and host}guest complexation capabilities of the molecule.
glycosidic oxygen bridges, is relatively hydrophobic In a more recent and somewhat different experi-
in comparison with polar solvents such as water. mental approach, the inclusion properties are sup-
Furthermore, the glycosidic oxygen bridges produce pressed by using a non-hydrogen bonding, polar or-
a high electron density, giving the interior of the ganic solvent (e.g., acetonitrile) as the main compon-
cavity a slightly Lewis-base character. ent of the mobile phase. Acetonitrile has the tendency
The three smallest cyclodextrin homologues are to occupy the cavity and seems to enhance hydrogen
readily commercial available: bonding between the hydroxyl groups on the cyc-
E -cyclodextrin (cyclohexamylose); lodextrin and hydrogen bonding groups on the chiral
E -cyclodextrin (cycloheptamylose); analyte.
E -cyclodextrin (cyclooctamylose); In this so-called polar organic mode of operation,
the addition of small amounts of glacial acetic acid
The basic property of cyclodextrins is their ability to and triethylamine is used as a tool to enhance enan-
form selective inclusion complexes with a broad var- tioselectivity. On the other hand, the addition of
iety of organic and inorganic molecules. a hydrogen bonding solvent such as methanol allows
The formation of inclusion complexes is in general reduced retention of strongly retained molecules. This
determined by the ability of the guest molecule to technique produces some unusual enantioselectivities
closely Rt the cavity of the cyclodextrin. However, the that certainly enhance the usefulness of native cyclo-
polarity of the guest molecule also plays an important dextrin phases.
role. Furthermore, a variety of cylcodextrin derivatives
Inclusion complexes are usually formed in the pres- has been immobilized on a chromatographic support,
ence of water or in water mixed with organic modi- which can be used under normal as well as under
Rers. reversed-phase conditions. The most popular com-
mercially available derivatized cyclodextrin chiral
selectors are listed in Table 1.
Mobile Phase Design and Parameter Optimization
Because complex stability constants usually have
greater values in water or water}organic organic sol-
vent mixtures than in a pure organic medium, native
cyclodextrin columns are predominantly used in
Figure 1 Numbering of carbon atoms in the cyclodextrin ring the reserved-phase mode. Ethanol, propanol, 1,4-
structure. dioxane, dimethyl sulfoxide, dimethyl formamide,
2336 III / CHIRAL SEPARATIONS / Cyclodextrins and Other Inclusion Complexation Approaches
Table 1 Commercially available derivatized cyclodextrin phases
R Trade name, Advanced Separation Technologies
Cyclobond威 I 2000 Ac Acylated
Cyclobond威 I 2000 RSP Hydroxypropyl ether

Cyclobond威 I 2000 SP
Cyclobond威 I 2000 RN Naphtylethyl carbamate

Cyclobond威 I 2000 SN
Cyclobond威 I 2000 DMP 3,5-Dimethylphenyl carbamate
Cyclobond威 I 2000 PT p-Toluoyl
methanol and acetonitrile have been used as organic In these initial experiments, especially on the
modiRers. However, methanol and acetonitrile are Cyclobond威 column, a poor peak shape for different
most commonly used. It is difRcult to predict in ad- products was observed. This effect was less pro-
vance which of these two modiRers will produce the nounced on the Chiradex威 column but it also occur-
best separation in any given case. In many cases, pH red for some products. The origin of the poor peak
and ionic strength of the aqueous part of the mobile shapes can have different causes, as for example an
phase are even more important than the choice of the insufRcient shielding of residual silanol groups.
organic modiRer. Therefore, the effect of the tailing reducer concentra-
The following factors inSuence enantioselectivity. tion was investigated.
Figure 2 illustrates the effect of the tailing reducer
Ionic strength With an eluent composed of a mix- concentration on the retention factor and the resolu-
ture of 5 millimolar tetrabutylammonium hydrogen- tion of miconazole (R14889). In this Rgure, the res-
sulfate (TBAHS) in water and methanol in an 80}20 olution factor based on the location of the valley
volume ratio, experiments were performed on a point between two peaks is used. The resolution para-
-cyclodextrin Astec column (Cyclobond威) and a meter (AuSoK sung ()) is deRned to evaluate the degree
similar Merck column Merck (Chiradex威). The only of separation between two partially resolved enantio-
difference between both stationary phases is the mers. This measuring principle is based on the loca-
chemistry used to attach the -cyclodextrin to the tion of the valley point between two adjacent peaks.
silica matrix. It is a useful and, from a measurement viewpoint,
the Cyclobond威 column. Also, the ionic strength re-

quired to reach the highest resolution value is differ-
ent for both columns.
Further experiments dealing with the effect of ionic
strength on peak shape and chromatographic behav-
iour have demonstrated that for each individual col-
umn type}tailing reducer combination it can be very
helpful to investigate this parameter thoroughly.
However, to save time, we nowadays start our chiral
method development work on cyclodextrin-based
columns with tailing reducer concentrations between
30 and 50 mmol, because we have experienced that
with these values there is in general a good chance of
being successful.
Type of tailing reducer The acetic acid salt of

triethylamine is popular as tailing reducer in reversed-
phase chromatography. Instead of using acetic acid to
adjust the pH value of an aqueous triethylamine solu-
tion, we investigated the usefulness of some other
organic and inorganic acids. At Rrst we did not expect
any inSuence of the type of counterion on enan-
tioselectivity, but some preliminary experiments
showed a distinct effect, worthwhile to investigate
Figure 2 Influence or tailing reducer concentration on the re- further. Therefore, we examined about 30 products
tention factor and valley point resolution 䉱, Cyclobond] ; belonging to different product classes. As mobile
, Chiradex] . Experimental conditions: column: 25 cm;4.5 mm phase, a mixture of 20 vol% of methanol and 80
ID chemically bonded cyclodextrin phase; mobile phase: TBAHS vol% of 50 mmol aqueous triethylamine solution ad-
in water at different concentrations}methanol (80}20, v/v); flow
rate: 1 mL min\1. Solute:
justed to a pH value of 2.5 with respectively
hydrochloric, hydrobromic, phosphoric, perchloric,
sulfuric, triSuoroacetic and oxalic acids was used.
For all the investigated products, the largest reten-
tion factors were observed for the triethylamine solu-
tion adjusted to the desire pH value with sulfuric and
oxalic acids. Also with sulfate as counterion, the
largest number of investigated products was partially
or completely resolved. Compared to the other acids,
with triSuoroacetic and perchloric acid a smaller
number of products were separated. To further inves-
tigate the observed effect, we repeated the same ex-
periments for a series of azoles, which differed only
slightly in structure. For these substances, the effect of
the counterion on resolution is illustrated in Figure 3.
easy evaluation technique for characterizing the See also Table 2.
separation degree between two peaks. For this product series, the results conRrm the
If we analyse Figure 2, it is easy to conclude that initial observations. Furthermore, Figure 3 clearly
ionic strength has a strong effect on the chromato- demonstrates that even small change in molecular
graphic behaviour of the investigated chiral molecule. structure, as for example the number or position of
It is furthermore perfectly clear that the inSuence of the chorine atoms on one of the phenyl groups of the
ionic strength on the retention factor is more pro- molecule, can have a tremendous effect on the enan-
nounced on the Chiradex威 than the Cyclobond威 col- tioselectivity.
umn. However, for both types of stationary phase Although for the investigated series of azoles pH
a constant value is observed above a tailing reducer adjustment with sulfuric acid in general resulted in
concentration of 20 mmol. At low TBAHS concentra- good peak shapes and high resolution values, one
tion, the smallest resolution values are measured on peculiarity was observed. Under these experimental
Figure 3 Azoles: effect of the counterion on resolution. Experimental conditions: column: 25 cm;4.6 mm ID Cyclobond威 (Astec);
mobile phase: 50 mM triethylamine in water adjusted to pH 2.5 with different acids}methanol (80}20, v/v); flow rate: 1 mL min\1,
solutes; see Table 2.
conditions, R78575 was strongly retained and both trated previously. The inSuence of the counterion on
enantiomers eluted as broad peaks with an irregular the retention factor and resolution of the investigated
shape (Figure 4). On the other hand, pH adjustment compounds can possibly be related to a difference in
of the triethlmaine solution with the other acids ability of the counterion to compete with the organic
resulted in normal peak shapes (Figure 5). This is guest molecule for the hydrogen bonding sites of the
an indication that minor differences in structure, or cyclodextrin host.
small variations in experimental conditions, can have To complete the experiments concerning the inSu-
a tremendous effect on the chromatographic behav- ence of the type of tailing reducer, we also found it
iour of enantiomers. necessary to do some tests in which the anionic part
A comparable set of experiments was subsequently of the tailing reducer was kept identical while the
performed using tetrabutylammonium hydroxide cationic part was varied. In these experiments,
instead of triethylamine as tailing reducer. In this set a 30 mmol solution of tetramethyl-, tetraethyl-, tetra-
of experiments, oxalic acid was excluded and per- propyl- and tetrabutylammonium hydroxide solution
chloric acid could not be used owing to the formation was adjusted with sulfuric acid to a pH value of 2.5
of an insoluble salt with tetrabutylammonium hy- and used as mobile phase in an 80}20 volume ratio
droxide in the water}methanol mixture. For the with methanol as organic modiRer.
whole test series of 24 different substances, sulfuric The smallest resolution values are observed for
acid could be identiRed as the counterion that gener- tetramethylammonium hydroxide and the highest
ated the highest resolution values. From a practical values for tetrabutylammonium hydroxide, but com-
point of view, it is certainly an advantage that sulfate pared with the inSuence of the anionic part of the
is the anion of choice. For temperatures between tailing reducer the differences are far less pro-
room temperature and 503C the corrosion of stainless nounced. However, for the investigated test series, the
steel (316 or equivalent) is negligible for dilute sulfur- tailing factor measured at 10% of the peak height
ic acid solutions, while the chemical resistance of this gradually decreases with increasing chain length of
material for chloride ions (even in low concentra- the alkylammonium hydroxide.
tions), is rather limited.
In conclusion, the anionic part of the tailing re- pH Whatever type of silica-based reversed phases
ducer has a clear effect on enantioselectivity, as illus- are used, these materials always display some acidic
Table 2 Structures of the investigated azole derivatives Figure 6 clearly illustrates that the retention factor
of the investigated products follows a typical rever-
sed-phase pattern. At higher pH values the degree of
protonation of the analyte diminishes. As a result the
chance for hydrophobic interactions with the station-
ary phase increases and higher retention values are
measured. When the pH is equal to the pKa value
Azoles of the sample, the number of protonated and
nonprotonated molecules is equal. Small changes in
Product R
pH around this value will immediately have an effect
T824 H on the retention factor. Even the effect of small struc-
tural changes, causing some differences, in hydropho-
R14827 (econazole) bic nature of the nonprotonated solute molecules, can
be clearly observed.
R8110 has the smallest retention factor owing
to the presence of a Suorine atom on the phenyl
R14889 (miconazole) group, giving the molecule a more polar character
than the other two members of the test series. The
largest retention factors are measured for R7405
bearing an ethyl group on the ester function attached
R24095 (isoconozole) to the imidazole ring instead of a methyl group for
R7315.
To better visualize the effect of pH variations
on enantioselectivity we often use for the graphical
R78575
R78690
surface area properties owing to residual hydroxyl

groups on the silica surface. During the analysis of
basic substances not only hydrophobic interactions
take place, but also acid}base interactions between
these acidic groups and basic functions in the analyte
can be expected to occur. This type of interaction
often results in increased retention combined with
peak tailing. One of the possibilities to solve this
problem is to adjust the pH below the pKa value of
the sample that has to be analysed. At pH values
lower than pKa!2 the basic function is protonated
and a salt is formed which no longer has basic proper-
ties.
Because the cyclodextrin columns are used under
Figure 4 Peak shape of the different azoles. Experimental con-
reversed-phase conditions, it is certainly worthwhile
ditions: column: 25 cm;4.6 mm ID Cyclobond威 (Astec); mobile
to investigate the effect of pH on enantioselectivity. phase: 50 mM triethylamine in water adjusted to pH 2.5
For some monobasic molecules, the effect of pH on with sulfuric acid}methanol (80}20, v/v); flow rate: 1 mL min\1;
the retention factor is depicted in Figure 6. solutes: see Table 2.
[R}NH2] ) [H#]
Ka"
[R}NH# 3 ]
[R}NH#3 ]
pKa"pH#log
[R}NH2]
(Henderson}Hasselbalch equation)
[R}NH#3 ]
pKa!pH"log
[R}NH2]
From:
[R}NH#3 ]
10[pKa\pH]"
[R}NH2]
and:
[R}NH#
3 ]#[R}NH2]"100
Figure 5 Peak shape of R78575. Experimental conditions: col-

umn: 25 cm;4.6 mm ID Cyclobond威 (Astec); mobile phase:
50 mM triethylamine in water adjusted to pH 2.5 with different
Figure 6 Effect of pH on rentention factor. 䉱, R7315;
acids}methanol (80}20, v/v); flow rate: 1 mL min\1; solute:
, R7405; 䡵, R8110. Experimental conditions: column:
25 cm;4 mm ID Chiradex威 (Merck); mobile phase: 50 mM
triethylamine in water adjusted to different pH values with sulfuric
acid}methanol (80}20, v/v); flow rate: 1 mL min\1; solutes:
representation of experimental data the degree of

protonation of the solute molecules instead of pH
values. For monobasic substances the ratio between
protonated and nonprotonated molecules at a certain
pH value can be easily calculated:
R}NH#
3 R}NH2#H
#
tivity at higher pH values. Besides investigations on

the effect of pH using triethylamine as basic sub-
stance in the aqueous part of the mobile phase, ex-
periments have also been performed with different
tetraalkylammonium hydroxides adjusted to the
desired pH value with sulfuric acid. Within the
same pH range, tetrabutylammonium hydroxide and
triethylamine displayed comparable patterns when
the valley point resolution was plotted against the
percentage of protonation of the solute molecules.
For a dibasic substance (R60844) with respectively
pKa values of 5.4 and 6.7 the protonation pattern
Figure 7 Effect of degree of protonation of the analyte on res- versus pH, together with the valley point resolution
olution. 䉬, R7315; , R7405; 䉱, R8110. Experimental conditions: for two types of mobile phase additives, is given in
column: 25 cm;4 mm ID Chiradex威 (Merck); mobile phase: Figure 8.
50 mM triethylamine in water adjusted to different pH values
with sulfuric acid}methanol (80}20, v/v); flow rate: 1 mL min\1;
Although for the dibasic product R60844 both
solutes: basic functions remain fully protonated up to a pH
value of 4, the resolution continuously increases
between pH 2 and 4. However, above pH 4 the
enantioselectivity rapidly decreases.
For the different product series that have been
examined we have observed that under the same
experimental conditions, in general, higher resolution
values are obtained with triethylamine than with
it is possible to easily calculate for each pH value the

ratio between protonated and free base. For R7315,
R7405 and R8110 the valley point resolution versus
the percentage protonation is depicted in Figure 7. Figure 8 Effect of pH on the degree of protonation and resolu-
For the investigated product series, it is clear that tion of a dibasic substance. N, Alpha 0; } } }, alpha 1; - - - -, alpha
under the experimental conditions applied above a 2; 䡵, TBA-OH; , TEA. Experimental conditions: column:
25 cm;4 mm ID Chiradex威 (Merck); mobile phase: 50 mM
certain degree of protonation of the solute molecules
triethylamine in water adjusted to different pH values with sulfuric
the enantioselectivity strongly decreases. Probably acid}methanol (80}20, v/v); and 30 mM tetrabutylammonium hy-
due to an increase in hydrophilic character of the droxide in water adjusted to different pH values with sulfuric
protonated molecules, the strength of the hydropho- acid}methanol (80}20, v/v); flow rate: 1 mL min\1; solute;
bic interactions between the analyte and the cavity of
the cyclodextrin molecule diminishes, with a reduced
enantioselectivity as a result. Within the investigated
pH range, triethylamine (pKa"10.72) always re-
mains fully protonated. Therefore, competition for
hydrophobic interactions with the cyclodextrin cavity
between the tailing reducer and the solute molecules
has to be considered as nonexistent and can be ex-
cluded as a possible reason for reduced enantioselec-
tetrabutylammonium hydroxide as mobile phase

additive.
In conclusion, the stability of the inclusion com-
plexes formed between the solute molecule and the
cyclodextrin seem to be dependent on the charge of
the guest molecule. Therefore, the retention as well as
the degree of separation of molecules bearing an
ionizable acidic or basic group can be affected by
changing the pH. However, the effect of pH vari-
ations is not so easy to predict.
In the experiments performed, some products
(R7315, R7405, T824, etc.) display the highest res-
olution values when they are present as free base or as Figure 9 Influence of temperature on the retention factor.
a partially charged molecule. Other products 䉬, R7315; ), R7405; 䉱, R60844; , R23979. Experimental
conditions: column: 25 cm;4 mm ID Chiradex威 (Merck); mobile
(R60844, most of the azoles) reach maximum resolu-
phase: 30 mM triethylamine hydroxide in water adjusted to differ-
tion when they are completely or nearly completely ent pH values with sulfuric acid}methanol (80}20, v/v); flow rate:
protonated. For some products the effect of pH vari- 1 mL min\1; temperature range: 1}403C; solutes:
ations within a rage of 2}3 pH units is small, while
for others extreme effects can be observed. Further-
more, for different types of tailing reducers the res-
olution versus pH proRles can have a different shape.
Therefore, the different experiments that have been
performed to investigate the effect of pH on enan-
tioselectivity clearly demonstrate that besides host}
guest complexation interactions between the solute
molecule and the cyclodextrin cavity, the hydroxyl
groups at the outside of the cyclodextrin molecule
together with reversed phase and other less predict-
able types of interactions certainly play an important
role in the chiral recognition process.
Because small changes in pH can have a tremen-
dous inSuence on the enantioselectivity, it is certainly
advisable to thoroughly investigate this parameter
during method development and optimization work
on cyclodextrin columns.
Temperature In equilibrium-based processes, tem- the resolution value with increasing temperature ob-
perature plays an important role. For all investigated served for R7405 can be considered as a logical be-
compounds, a very good linear relationship between haviour, because for R7405 and R7315 the smallest
the natural logarithm of the retention factor and the retention factors are measures } an indication that the
inverse of temperature is observed. For the products binding forces between the stationary phase and these
R7315 and R7405 only small differences in slope are solutes are smaller than for the other products investi-
measured. Indications that for both products the en- gated. As a result, the effect of a temperature increase
thalpy values for solute}stationary phase transfer are on the resolution is more pronounced for these com-
very similar. The effect of temperature on the valley pounds. However, we did not expect to observe a dif-
point resolution is given in Figure 9. ferent effect of temperature variations for R7315 and
Only for R23979 was a valley point resolution of R7405, because for both products only minor differ-
one measured for the whole temperature range be- ences in enthalpy values were measured. Therefore,
tween 1 and 403C meaning that for this product the temperature variations probably have the same inSu-
effect of temperature variation is of the same magni- ence on both enantiomers of R7405, while the enan-
tude for both enantiomers. For R60844 and R7315 tiomers of R7315 are affected in a different way.
practically no inSuence on the resolution was ob- In conclusion, temperature variations have an in-
served between 1 and 153C. Above that temperature, Suence on the retention factor as well as on the
for both products the resolution value starts to de- resolution value. However, the effects observed differ
crease in a similar way. The continuous decrease of from one product to another. Therefore, this parameter
has to be investigated on an individual basis during processes, chiral stationary phases often display slow
method development and optimization work. mass transfer characteristics. Therefore, on chiral sta-
tionary phases, Sow rate can have a strong effect on
Type and concentration of organic modiVer It is the enantioselectivity. For difRcult separations,
known that for the reversed-phase analysis of several lowering the Sow rate certainly has to be considered
alkylbenzenes on chemically bonded cyclodextrin as a tool to enhance enantioselectivity.
columns the type of alcohol, together with its concen-
Derivatized Cyclodextrins
tration in the mobile phase, strongly inSuences the
retention behaviour of these substances. Native cyclodextrin columns cannot be used
To investigate the effect of type of alcohol on effectively for the separation of enantiomers under
enantioselectivity, we performed some experiments in normal phase chromatographic conditions. On the
which the normally used modiRer (methanol) was other hand, different naturally occurring chiral mol-
replaced by the same amount of ethanol 1-propanol. ecules that have been derivatized are extensively and
For all the investigated substances, the retention fac- very successfully used in the normal phase mode of
tor strongly decreased with increasing chain length of operation.
the alcohol used. Triacetylcellulose, obtained by heterogeneous
The retention factors measured for an ethanol- acylation of cellulose, was one of the Rrst commer-
based mobile phase are about 30}50% lower than the cially available derivatives. However, the later de-
values observed for methanol as organic modiRer. veloped and commercialized aromatic cellulose and
A decrease of approximately the same magnitude amylose derivatives (benzoates, carbamates), com-
could be observed when ethanol was replaced by pared with triacetylcellulose, are much more univer-
1-propanol. Therefore, in extreme cases the retention sally applicable. Owing to the broad applicability
factor drops to about 20% of the value measured for of the cellulose and amylose derivatives similar
methanol, when this solvent is replaced by the same cyclodextrin-based stationary phases have been
amount of 1-propanol. With increasing chain length developed.
the hydrophobicity of the alcohol increase, which In our laboratories we investigated the possibili-
enhances the chance for competition between the ties of the derivatized cyclodextrin columns under
solvent and the analyte molecules to interact with the normal as well as reversed-phase chromatographic
hydrophobic cyclodextrin cavity. conditions. We also compared these phases with the
Because a strong decrease in retention time of the corresponding cellulose derivatives.
solutes could be observed when ethanol or 1-pro- Our Rrst experiments on the functionalized cyclo-
panol was used as organic modiRer, these solvents dextrin phases were performed under normal phase
seem to have a greater afRnity for the cyclodextrin conditions. A series of 42 products covering a broad
cavity than methanol. Therefore, they will be more range of organic chemistry were investigated on the
effective for displacing strongly retained substances S-naphthylethyl carbamate, the para-toluoyl and 3,5-
from a cyclodextrin column. The manufacturers of dimethylphenyl carbamate cyclodextrin derivative,
cyclodextrin columns in fact use this property, be- using n-hexane-2-propanol in different ratios as the
cause they recommend regenerating their columns by mobile phase.
passing several column volumes of pure ethanol The results obtained were rather poor. Of the
through the column, followed by pure water and then whole test series only six products were partially or
methanol. completely resolved on the 3,5-dimethylphenyl car-
In general, an increasing water content in the mobile bamate column. The situation was even worse on the
phase increases both the retention and the enan- two other columns tested. Therefore, we decided to
tioselectivity. However, for practical applications the switch immediately to the reversed-phase mode of
retention factors are generally too long when mobile operation. The initially used test series of 42 products
phases with a high water content are used, and a com- was Rrst investigated on the three above-mentioned
promise has to be found. For our product classes we cyclodextrin columns with a mobile phase consisting
therefore use a mixture of 80 vol% of aqueous phase of 0.5% ammonium acetate in water and methanol in
and 20 vol% of methanol as a typical starting mobile a 30}70 volume ratio. This mobile phase composition
phase composition. If the products are too strongly was chosen after some preliminary experiments,
retained or do not elute at all, the amount of meth- which demonstrated that higher water content caused
anol is systematically increased. a tremendous increase in retention times. Compared
with the results under normal phase conditions, the
Flow rate Due to the very speciRc type of interac- number of products separated and the degree of
tions, which play a major role in chiral recognition separation were much better on all the investigated
column types. Because earlier experiments on the tested were partially resolved when (d)-camphorsul-
native cylcodextrin phases have demonstrated that fonic acid was used to adjust the pH of the tet-
an acidic pH generally results in better separations, ramethylammonium hydroxide solution.
a 20 mM tetrabutylammonium hydrogensulfate
solution (pH 2.3) was thereafter used as tailing Comparison of derivatized cyclodextrins and the
reducer. corresponding cellulose derivatives Because de-
Owing to the well-known reversed-phase effect of rivatized cellulose and amylose columns are exten-
retention decrease with lowering pH values, the sively used in our laboratories for both analytical and
methanol content in the mobile phase had to be preparative chromatographic applications, it seemed
reduced to 40 vol% instead of the 80 vol% used in worthwhile to compare these phases with the corre-
the experiments with ammonium acetate as tailing sponding cyclodextrin derivatives. At present only
reducer. Under these experimental conditions, the two cellulose phases are commercially available
largest number of products was separated on the which can be used equally well under reversed-phase
S-naphthylethyl carbamate column, although the and normal phase conditions, namely Chiralcel威
results on the other two columns only differed slight- OD-R (3,5-dimethlyphenyl carbamate and Chiralcel威
ly. It was also interesting to observe that some prod- OJ-R (para-methylbenzoate) (Daicel, Japan). We
ucts, which could not be separated on native cyc- compared these phases with Cyclobond威-DMP and
lodextrin, were completely resolved on one of the Cyclobond威-PT columns (Advanced Separation
derivatized phases. In the next set of experiments, Technologies), respectively.
a 20 mM solution of respectively tetramethyl-, The Rrst experiments were performed under nor-
tetraethyl- and tetrabutylammonium hydroxide was mal phase conditions, using n-hexane}2-propanol in
adjusted with sulfuric acid to a pH value of 2.5 and a 70}30 ratio as the mobile phase. If products eluted
used in combination with 70 vol% of methanol as too fast with this mobile phase composition, the
the mobile phase. The effect of the cationic part of the amount of 2-propanol was reduced to respectively 20
tailing reducer is not clear. With tetraethylam- or 10 vol%. A total of 21 different products were
monium hydroxide the largest number of products examined. The results of these experiments are sum-
are partially or completely resolved, but the resolu- marized in Table 3.
tion values are in general somewhat higher with From this data it is clear that for the investigated
tetramethylammonium hydroxide, although the dif- product classes the cellulose derivatives are far su-
ferences with the two other alkylammonium hydrox- perior compared to the corresponding cyclodextrin
ides are minimal. Only for one member of the test phases when normal phase conditions are applied.
series was tetramethylammonium hydroxide required We thereafter examined the same product series
to obtain partial resolution. under reversed-phase conditions using a mixture of
On native cyclodextrin columns, we could clearly 70 vol% of methanol and 30% of a 50 mM
demonstrate the inSuence of the anionic part of the triethylamine solution adjusted to pH 2.5 with sulfuric
tailing reducer. Therefore, a similar test was done acid. The results of these experiments are summarized
on the 3,5-dimethylphenyl carbamate cyclodextrin in Table 4.
derivative, using 20 mM tetramethylammonium When we compare this data with the results ob-
hydroxide adjusted to pH 2.5 with respectively triSu- tained under normal phase conditions, we have to
oroacetic, hydrochloric, phosphoric, (d)-camphor- conclude that in the reversed phase mode of
sulfonic and sulfuric acids, (d)-Camphorsulfonic acid
has been deliberately chosen to investigate eventual
Table 3 Derivatized cyclodextrins versus the corresponding
additional effects by introducing chirality in the mo- cellulose derivatives under normal phase conditons
bile phase. After pH adjustment, the aqueous phase
was mixed with methanol in a 30}70 volume ratio. Column type Good separa- Partially Total
Some products were also tested with a 50}50 mixture tion '0.90 resolved a
of aqueous phase and methanol. Comparable with
Number % Number % Number %
the observations on the native cylcodextrin column,
and also on the functionalized cyclodextrin, pH ad- Chiralcel OD-R 9 42.9 6 28.6 15 71.4
justment with sulfuric acid resulted in the largest Cyclobond I-DMP 2 9.5 3 14.3 5 23.8
number of separations. For all the other acids, the Chiralcel OJ-R 13 61.9 5 23.8 18 85.7
Cyclobond I-PT } } 2 9.5 2 9.5
number of partially or completely resolved products
dropped to about 50% or less of the value observed a
For the partially resolved peaks, the resolution on the cellulose
for sulfuric acid. However, three products which derivatives is always much higher than on the cyclodextrin deriva-
could not be separated with one of the different acids tives.
Table 4 Derivatized cyclodextrins versus the corresponding For a series of products which under comparable
cellulose derivatives under reversed phase conditons experimental conditions are all very well separated on
the native cyclodextrin column, the results on the
Column type Good separa- Partially Total
tion '0.90 resolved a different cyclodextrin and cellulose derivatives using
50 mM triethylamine adjusted to pH 2.5 with sulfur-
Number % Number % Number % ic acid and methanol in a 30}70 volume ratio as
mobile phase are illustrated in Figure 10.
Chiralcel OD-R 2 9.5 10 47.6 12 57.1
Because only one substance of the test series is
Cyclobond I-DMP 2 9.5 5 23.8 7 33.3
Chiralcel OJ-R 14b 66.7 } } 14 66.7 separated on the Cyclobond-PT column, and most
Cyclobond I-PT } } 3 14.3 3 14.3 of the products are only partially resolved on the
Cyclobond-DMP column, while all these products
a
For the partially resolved peaks, the resolution on the cellulose are perfectly baseline resolved on native cyclodex-
derivatives is always much higher than on the cyclodextrin deriva-
trine, it is clear that other parameters must play a role
tives.
b
All products are fully baseline resolved ("1.00). in the chiral recognition process on the derivatized
phases.
As a general conclusion it can be stated that for the
type of substances investigated the derivatized cyclo-
operation fewer products are separated on the cellu- dextrin columns are, in both modes of operation
lose derivatives, although on the Chiralcel威 OJ-R (normal as well as reversed phase), less universally
column all separated products are fully baseline re- applicable than the corresponding cellulose deriva-
solved. The smallest resolution value equals 2.5 while tives.
the largest value is greater than 12.5, while under
Hydroxypropyl--Cyclodextrin Derivative
normal phase conditions the highest resolution value
observed equals 6.3. A hydroxypropyl--cyclodextrin column (experi-
On the cyclodextrin derivatives a few more prod- mental phase of the Chromatography Research group
ucts are separated but the increase in number is of Merck Darmstadt) in the reversed-phase mode of
certainly not spectacular. Furthermore, in many cases operation has been extensively investigated. The
where partial resolution has been indicated in the result on this type of material were completely
table the chromatograms only showed a small devi- comparable with the data obtained on the native
ation in the peak shape, indicating the early beginning cyclodextrin columns. However, for the whole range
of separation. of products the degree of separation was in general
Figure 10 Derivatized cylcodextrin columns versus the corresponding cellulose derivatives. Experimental conditions: column:
250 mm;4.6 mm ID (Cyclobond威-DMP, Cyclobond威-PT, Chiralcel威 OD-R, Chiralcel威 OJ-R); mobile phase: 50 mM triethylamine
adjusted to pH 2.5 with sulfuric acid}methanol (30}70, v/v); flow rate: 1 mL min\1.
better on the hydroxypropyl column. To investigate Cyclodextrin Phases in Preparative

new products, we therefore always start our experi- Chromatographic Applications
ments on a hydroxypropyl--cyclodextrin column The importance of preparative chromatographic
instead of using the classical native -cyclodextrin enantiomer separations in industry is continuously
type of material. growing. Therefore it was very interesting to investi-
gate the usefulness of cyclodextrin phases in prepara-
-Cyclodextrin tive chromatographic applications.
A test series of 28 products, which also have been On an experimental hydroxypropyl--cyclodextrin
investigated on -cylcodextrin, have been examined phase (Merck Darmstadt), several products which
on a -cyclodextrin column using a mobile phase showed a good resolution on the corresponding ana-
consisting of 80 vol% of a 50 mM triethylamine lytical material were investigated on a preparative
solution in water adjusted to pH 2.5 with sulfuric scale. An example of a preparative chromatographic
acid and 20 vol% of methanol. On the -cyclodextrin separation is illustrated in Figure 11.
column, 22 products were partially or completely On the hydroxypropyl--cyclodextrin phase, in
resolved. On the -derivative only nine products general a loading capacity of 2 mg g\1 of pack-
could be resolved. ing material was used. However, in some other cases
Compared with the results obtained on the -cyclo- we were able to load up to 4 mg of product per
dextrin column, it is clear that from a general point of gram of stationary phase, which from a preparative
view the derivative is less suitable for the separation chromatographic point of view certainly can be
of the investigated product series. Nevertheless, it considered as a reasonable value for this type of
remains an additional tool that might help to solve application.
a separation problem when experiments on other
types of cyclodextrin columns fail.
Cyclodextrins as Stationary Phases
Dynamically Generated Cyclodextrin Phases in Gas Chromatography
Instead of using the commercially available chemic- For the preparation of cyclodextrin-based capillary
ally bonded cyclodextrin phases, it is also possible to columns for gas chromatographic applications, two
perform enantiomer separations on a reversed-phase complementary methods have been developed and
column after addition of cyclodextrin or cyclodextrin are commercially used. In the Rrst method, alkylated
derivates to the mobile phase. A number of experi- cyclodextrins are diluted with polysiloxanes and im-
ments have been performed with hydroxypropyl-- mobilized on the inside wall of a fused silica capillary.
cyclodextrin as mobile phase additive. This derivative In the other method, pentyl and hydroxyalkyl-
was initially chosen because it is readily soluble in dimethylcyclodextrins are coated on the inside wall
water. of a fused silica capillary. Different well-known chro-
To investigate the possibilities of a dynamically matography companies offer cyclodextrin-based cap-
generated cyclodextrin phase, a test series of 22 illary columns:
products was examined on three different types of
reversed-phase packing material. RP Select E Chrompack: diluted cyclodextrins;
B (Merck), Hypersil BDS (Shandon) and Aluspher E Macherey-Nagel: undiluted n-pentylated or
RP Select B (Merck) were selected as stationary acylated cyclodextrins;
phases. An aqueous solution containing 50 mM E Advanced Separation Technologies: a broad var-
triethylamine and 50 mM of hydroxypropyl-- iety of undiluted cyclodextrin derivatives (per-
cyclodextrin adjusted to pH 3 with sulfuric acid methylated, hydroxypropyl, triSuoroacylated,
combined with methanol in an 80}20 volume ratio butyrylated, dialkylated).
was used as the mobile phase.
Compared with the data obtained on a chemically Different types of compounds (amines, epoxides, al-
bonded hydroxypropyl--cyclodextrin column using kanes, alcohols, lactones, sugars, etc.) can be separ-
the same eluent, the chemically bonded column gives, ated on cyclodextrin-based capillary columns.
in general, better results than the dynamically coated The usefulness of these materials is illustrated by
reversed-phase materials. Furthermore, alumina as means of the separation of the four isomers of
stationary phase matrix seems to be less effective. a piperidine derivative. Gas chromatography on cyclo-
However, in a few exceptional cases the Aluspher dextrin-based columns was tried after acetylation
Select B column gives as good or even better results with triSuoracetic anhydride using the following
than the silica-based materials. procedure:
An advantage of this dervatization procedure was The Rrst experiments were done on a 15 m;
that all types of salts (used to investigate the possibili- 0.32 mm ID Chiralsil-Dex威 column (Chrompack).
ties of diastereomeric salt formation as a stereoselec- On this type of column, it was only possible to separ-
tive synthesis method) could be acylated without ate the cis and trans isomers.
prior liberation of the free base. Thereafter, chromatographic experiments were
performed on four different cyclodextrin phases from
Advanced Separation Technologies:
E Chiraldex威 B-PH: (S)-2-hydroxypropyl per-

methylated -cyclodextrin;
E Chiraldex威 B-DA: (2,6-di-O-n-pentyl)--cyclodex-
trin;
E Chiraldex威 B-TA: (2,6-di-O-n-pentyl-3-O-tri-
Suoroacetyl)--cyclodextrin;
E Chiraldex威 G-TA: (2,6-di-O-n-pentyl-3-O-tri-
Suoroacetyl)--cyclodextrin.
Chromatograms of the different experiments are il-

lustrated in Figure 12 and Figure 13. The largest dif-
ferentiation between cis and trans isomers was ob-
served on the permethylated hydroxypropyl--
cyclodextrin column (Chiraldex威 B-PH). However,
the best separation of all isomers individually was
observed on the triSuoroacylated -cyclodextrin col-
umn (Chiraldex威 B-TA).
Figure 11 Preparative chromatographic separation on a
Noticeable is the time required to perform an
-cyclodextrin columns. Experimental condition: column: 80 mm
ID dynamic axial compression column (Prochrom); stationary analysis. Compared with the analysis method on the
phase: 500 g 10 m chemically bonded hydroxypropyl--cyc- crown ether column, a GC analysis is more than six
lodextrin (experimental phase Merck Darmstadt; packing pres- times faster, although one has to take into account
sure 80 bar; mobile phase: 50 mM triethylamine adjusted with that on the crown ether column the diastereomeric
50 mM sulfuric acid to pH 2.5}methanol (80}20, v/v); flow rate:
salt could be analysed as such, while for the gas
150 mL min\1; detection: UV, wavelength 220 nm, range 2.56
AUFS; sample size: 1 g dissolved in 50 mL of concentrated sulfur- chromatographic method the product has to be de-
ic acid; solute: rivatized prior to chromatography.
Cyclodextrins in Capillary
Electrophoresis
In the pharmaceutical industry the importance of
capillary electrophoresis is continuously growing. It is
a technique which is increasingly used for determina-
tion of the optical purity of intermediates and Rnal
products. Many different optically pure compounds
The different types of cyclodextrins, which are

most frequently used are:
E -cyclodextrin;
E -cyclodextrin;
E -cyclodextrin;
E (2-hydroxy)propylated -cyclodextrin;
E (2-hydroxy)propylated -cyclodextrin;
E Heptakis (2,6-di-O-methyl) -cyclodextrin;
E Heptakis (2,3,6-tri-O-methyl) -cyclodextrin;
E carboxymethylated -cyclodextrin.
Figure 12 Capillary GC analysis on cyclodextrin stationary

phases. Experimental conditions: column: (A) 15 m;0.32 mm ID
Chiraldex威 B-PH ((S )-hydroxypropyl--cyclodextrin (Astec)); (B)
15 mm;0.32 mm ID Chiraldex威 B-DA (dipentylated -cyclo-
dextrin (Astec)); carrier gas: helium (linear velocity 25 cm s\1);
temperatures: column 150}2003C (33C min\1), injector 2103C,
detector 2103C; detection: FID; injection: 1 L split (split ratio Figure 13 Capillary GC analysis on cyclodextrin stationary
1/100). phases. Experimental conditions: column: (A) 15 m;0.32 mm ID
Chiraldex威 B-TA (trifluoroacetyl--cyclodextrin (Astec));
(B) 15 m;0.32 m ID Chiraldex威 G-TA (trifluoroacetyl--cyclodex-
can be used to generate the required chiral environ- trin (Astec)); carrier gas: helium (linear velocity 25 cm s\1); tem-
ment. Cyclodextrins of course are also very suitable peratures: column 150}2003C (23C min\1), injector 2103C, de-
as an electrolyte additive. tector 2103C; detection: FID; injection: 1 L split (split ratio 1/100).
III / CHIRAL SEPARATIONS / Gas Chromatography 2349
Figure 14 Capillary electrophoresis using a derivatized cyclodextrin as electrolyte additive. Experimental conditions: equipment:
P/Ace system 5500 (Beckman); capillary: 50 m ID uncoated fused silica; total length: 57 cm; length to detector: 50 cm; electrolyte:
20 mM Heptakis (2,3,6-tri-O-methyl)-cyclodextrin 10 mM disodium hydrogen phosphate solution adjusted to pH 2.2 with phosphoric
acid; analysis: temperature 253C, voltage #20 kV, inject sample 2 s, detection UV (220 nm).
The usefulness of cyclodextrins as an electrolyte addi- Interaction (‘Pirkle’) Stationary Phases; Thin-Layer
tive is illustrated in the following example. A sub- (Planar) Chromatography.
stance containing three optical centres, which means
eight possible isomers, had to be separated. HPLC Further Reading
experiments on different types of chiral stationary
phases did not succeed in a complete resolution of the Allenmark S (1991) Chromatographic Enantioseparations,
mixture. The result of a capillary electrophoresis ex- Methods and Applications. London: Ellis Horwood.
periment using Heptakis (2,3,6-tri-O-methyl) -cyc- Beesley TE and Scott RPW (1998) Chiral Chromatography.
Chichester: Wiley.
lodextrin as electrolyte additive is illustrated in Fig-
Bender ML and Komiyama M (1978) Cyclodextrin Chem-
ure 14. istry. Berlin: Springer.
Compared with HPLC, in capillary electrophoresis Chiraldex GC Handbook, 5th edn (1996) Whippany, NJ:
many more parameters can be varied to improve Advanced Separation Technologies.
separation. Therefore, most of the methods de- Coventry L (1989) Cyclodextrin inclusion complexation.
veloped on one of the commonly used chiral station- In Lough WJ (ed.) Chiral Liquid Chromatography.
ary phases can be replaced by a capillary electrophor- Glasgow: Blackie.
esis methods, using cyclodextrins or another chiral Cyclobond Handbook (1996) Whippany, NJ: Advanced
auxiliary as electrolyte additive. Separation Technologies.
Han SM and Armstrong DW (1989) Separation of enantio-
See also: II/Chromatography: Liquid: Ion Pair Liquid mers and other isomers with cyclodextrin-bonded phases:
Chromatography; Mechanisms: Chiral; III/Chiral Separ- rules for chiral recognition. In KrstulovicH (ed.) Chiral
ations: Amino Acids and Derivatives; Capillary Elec- Separations by HPLC, Applications to Pharmaceutical
trophoresis; Cellulose and Cellulose Derived Phases; Compounds, pp. 208}286. Chichester: Ellis Horwood.
Chiral Derivatization; Gas Chromatogrpahy; Ion-Pair Hinze WL (1981) In: van Oss CJ (ed.) Separation and
Chromatography; Ligand Exchange Chromatography; PuriTcation Methods, vol. 10, p. 159.
Liquid Chromatography; Molecular Imprints as Stationary Kaiser R (1965) Chromatographie in der Gasphase I. Mann-
Phases; Protein Stationary Phases; Synthetic Multiple heim: Bibliographisches Intitut.
Gas Chromatography
V. Schurig, University of Tu( bingen, Germany phase (CSP) was discovered by Gil-Av and co-
workers at the Weizmann Institute of Science, Israel,
in 1966. At the outset of this work, according to
Gil-Av,
Introduction
The separation of enantiomers (optical isomers) by this topic was in a ‘state of frustration’. Nobody
capillary gas chromatography on a chiral stationary believed it could be done. In fact, people were
2350 III / CHIRAL SEPARATIONS / Gas Chromatography
convinced that there could not possibly be a large ical reaction with an enantiomerically pure resolv-
enough difference in the interaction between the D- ing agent and subsequent gas chromatographic
and L-solute with an asymmetric solvent. This was separation of the diastereomers is achieved using
the feeling people had, even those known as unor- a conventional achiral stationary phase.
thodox thinkers. This view had also some experi- 2. Direct method. Gas chromatographic separation
mental basis, because a number of communications of the enantiomers is achieved using a chiral sta-
had been published, in which it was claimed that tionary phase (CSP) containing a resolving agent
such resolutions could be effected, but nobody was of high (but not necessarily 100%) enantiomeric
able to reproduce these results, and some of them purity.
were shown to be deRnitely wrong.
While the Rrst method entails the formation of dia-
Today, almost a reversal of this situation exists.
stereomers before separation, the second involves the
According to the GC Chirbase data bank, for most
rapid and reversible diastereomeric association be-
volatile racemic compounds of a variety of different
tween the CSP (selector) and the racemic, or non-
chemical classes, ranging from apolar to polar, an
racemic, analyte (selectand). Since diastereomers
appropriate CSP is available and 22 200 chiral separ-
display different physical properties, discrimination
ations by gas chromatography (GC) of 7637 analytes
by incomplete recovery, decomposition and losses
on 684 CSPs (120 are commercialized) have been
may occur during work-up, isolation and sample
reported up to the end of 1997.
handling in method (1). Also the detector response
Three principal CSPs are currently employed; these
can be, in principle, different for diastereomers. Be-
undergo hydrogen bonding, coordination and inclu-
cause an achiral detection device does not discrimi-
sion. ModiRed -, - and -cylodextrins have proved
nate between enantiomers, the comparison of relative
to be the most versatile and universal CSPs in GC.
peak areas employing method (2) provides a direct
Anchoring the CSPs to a polysiloxane backbone leads
measure of the enantiomeric composition, provided
to Chirasil-type stationary phases with improved
the detector response is linear over a wide concentra-
temperature stability, efRciency and robustness. Im-
tion range. Consequently, method (2) is preferred for
mobilization of Chirasil-type stationary phases on the
the determination of enantiomeric excess. This ap-
inner column wall furnishes chemically bonded CSPs.
proach requires an efRcient selector}selectand system
Because of the high efRciency, sensitivity and speed,
displaying chiral recognition.
chiral separation by high resolution capillary GC
Enantioselectivity is governed by the biased dia-
represents a versatile and attractive method for enan-
stereomeric association between the enantiomers of
tiomer analysis. However the prerequisites of the
the selectand and the chiral selector. Fortunately, by
method, i.e. volatility, thermal stability and resolv-
employing capillary columns, efRciency is mostly
ability of the chiral analytes, restrict its universal use.
high enough to resolve racemates having a difference
The main application of chiral separations by GC is
of the Gibbs free energy of diastereomeric association
concerned with the precise determination of enan-
as little as !R,S(G)"0.025 kJ mol\1 (at 253C),
tiomeric compositions of chiral research chemicals,
corresponding to a separation factor, , of 1.01
intermediates, auxiliaries, metabolites, precursors,
("kR/kS, with k"the retention factor and the sub-
drugs, pesticides, fungicides, herbicides, pheromones,
script R referring to the second eluted enantiomer and
Savours and fragrances. As the insight into chiral-
S to the Rrst eluted enantiomer). The measured enan-
ity}activity relationships steadily improves and, as
tiomeric composition of the analyte determined by
a consequence, legislation on chiral compounds be-
method (2) is independent of the enantiomeric purity
comes more stringent, the development of reliable
of the CSP. Lowering the enantiomeric purity of the
methods for the determination of the enantiomeric
CSP, however, results in a decrease of and it is unity
excess up to 99.9% is of great importance (% enan-
when the chiral selector in the stationary phase is
tiomeric excess"100(R!S)/(R#S), where R is the
racemic. Method (2) is especially useful for the deter-
major enantiomer and S the minor enantiomer). This
mination of enantiomeric excess when no sample
goal is readily met by enantioselective GC.
derivatization is required. Owing to the enormous
separation power of capillary GC, contaminants and
Methodology impurities are usually separated from the enantiomers
and the simultaneous analysis of the enantiomers of
The separation of enantiomers by GC can be per-
different compounds (e.g. all proteinogenic amino
formed in two modes.
acids) is feasible in one analytical run (cf. Figure 1).
1. Indirect method. Enantiomers are converted off- Temperature programming and established ancillary
column into diastereomeric derivatives by chem- techniques such as multidimensional chromatography
Figure 1 Simulationeous enantiomer separation of 20 proteinogenic amino acids as N, O, S-pentafluoropropanoate-isopropylester

(histidine as Nim-ethoxycarbonyl) derivatives by GC on Chirasil-Val [IV] between 85 and 1853C at 0.35 bar (gauge) hydrogen. Column,
50 m;0.27 mm (i.d.) glass capillary. D-Enantiomers are eluted before L-enantiomers. (From Bayer E (1983) Chiral recognition of
( r Naturforschung 38b: 1281}1291.)
natural products on optically active polysiloxanes. Zeitschrift fu
and the use of interfacing and coupling methods E chiral metal chelates via coordination (complexa-
can readily be adapted to chiral separations. Sensi- tion GC);
tivity can be extended to the picogram level by E cyclodextrin derivatives via inclusion.
electron capture detection or by the combination
of gas chromatography with mass spectrometry Initially, the chiral selectors were used as involatile
(GC-MS). GC-MS-selected ion monitoring (SIM) neat liquids or as solutions in squalane or poly-
can detect trace amounts of enantiomers in complex siloxanes, respectively. Subsequently, a number of
matrices. chiral selectors have been chemically linked to poly-
When the racemate is highly volatile and the chiral siloxanes (Chirasil-type stationary phases).
separation factor is large ('1.3), preparative enan-
Chiral Stationary Phases Based on
tiomer separation by GC with packed columns is
Hydrogen Bonding
possible. The development of simulated moving bed
(SMB) approaches can be expected in the future. The Rrst successful separation of racemic N-tri-
Semipreparative separations can already be carried Suoroacetyl amino acid alkyl esters on glass capillary
out at low values. Recovery from the gaseous car- columns coated with involatile N-triSuoroacetyl-L-
rier is straightforward and pure enantiomers can be isoleucine lauryl ester [I] (cf. Scheme 1) was achieved
obtained even on enantiomerically impure CSPs. For by Gil-Av and co-workers in 1966 and a semi-
analytical purposes it may be important to differenti- preparative version of this method was reported later.
ate a separation of a chiral compound into enantio- Since then the great potential of this fundamental
mers from the common separation of two achiral approach has stimulated continuing research on en-
analytes. A racemate, when resolved, should produce antiomer separation not only by GC, but also by
an exact 1 : 1 peak ratio of the enantiomers. Clearly, other chromatographic techniques such as HPLC.
only one peak (peak coalescence, "1) is expected It was recognized that in the dipeptide phase [II]
when the racemic CSP is employed. When two CSPs (cf. Scheme 1) the C-terminal amino acid was not
of opposite conRgurations are applied in different essential to chiral recognition while the additional
columns, peak inversion (peak switching) can be ob- amide function was important for additional hydro-
served for a chiral analyte in which one enantiomer is gen bonding. The second chiral centre was therefore
in excess. sacriRced by preparing the diamide [III], e.g. derived
from valine. This chiral selector was subsequently
coupled via the amino function to a statistical
Classi\cation of Chiral Stationary copolymer of dimethylsiloxane and (2-carboxy-
Phases propyl)methylsiloxane. The resulting polymeric CSP,
Enantiomer separation by GC is mainly performed on Chirasil-Val [IV], combining enantioselectivity and
three types of CSPs: efRciency of silicones, exhibits excellent GC proper-
ties for the enantiomer separation of chiral com-
E chiral amino acid derivatives via hydrogen bond- pounds undergoing hydrogen bonding. Chirasil-
ing; Val [IV] is commercially available in both enantiomeric
Scheme 1 Hydrogen-bonding-type chiral stationary phases.
forms. The temperature-programmed simultaneous belonging to the smallest chiral molecules. A limit-
enantiomer separation of all proteinogenic amino ing factor of coordination-type CSPs [VIII, IX] is the
acids in less than 25 min is illustrated in Figure 1. low temperature range of operation (25}1203C).
A straightforward approach to polymeric CSPs is The thermal stability has been increased by the
based on the modiRcation of cyanoalkyl-substituted
polysiloxanes (XE-60, OV-225). For instance the dia-
mide [III] was chemically linked to the polysiloxane
to give (L)-[V]. The diastereomeric selectors (L, R, and
L, S)-[VI] contain two chiral centres that enhance
enantioselectivity (matched case) or compensate en-
antioselectivity (mismatched case).
Enantiomer separation by hydrogen-bonding CSPs
generally requires derivatization of the analyte in
order to increase volatility and/or to introduce suit-
able functions for additional hydrogen-bonding asso-
ciation.
Chiral Stationary Phases Based on Coordination
The chiral metal coordination compound dicarbonyl-
rhodium(I)-3-triSuoroacetyl-(1R)-camphorate [VII]
(cf. Scheme 2) was used for the enantiomeric separ-
ation of the chiral oleRn 3-methylcyclopentene by
complexation GC in 1977. The scope of enantiomer
separation by complexation GC was later extended to
oxygen-, nitrogen- and sulfur-functionalized com-
pounds using chiral ketoenolate-bis-chelates of,
among others, manganese(II) and nickel(II) ions de-
rived from terpene ketones such as camphor [VIII,
IX], menthone, carvone, pulegone, and others after
perSuoroacylation.
Figure 2 illustrates the enantiomer separation
by complexation GC of simple aliphatic oxiranes, Scheme 2 Coordination-type chiral stationary phases.
(CDs) can be employed in high resolution capillary

columns for enantiomer analysis. Thus, neat per-
methylated -cyclodextrin, [XI] (cf. Scheme 3), was
used above its melting point and in a supercooled
state. Per-n-pentylated and 3-acyl-2,6-n-pentylated
CDs are liquids at room temperature. The CD deriva-
tives, [XIII]}[XVII], have been used in the undiluted
form for the separation of enantiomers of many
classes of compounds on deactivated Pyrex glass cap-
illary columns. The more polar CD derivatives,
[XX]}[XXIII], have been coated on fused silica capil-
lary columns.
Figure 2 Enantiomer separation of monoalkyl-substituted
To combine the enantioselectivity of CDs with the
oxiranes by complexation GC on manganese (II) bis [3-(hepta- excellent coating properties and efRciency of poly-
fluorobutanoyl)-(1R )-camphorate] [IX] (0.05 molal in squalane) at siloxanes, alkylated CDs have been preferentially dis-
603C. C1"Methane. Column, 160 m;0.4 mm (i.d.) stainless solved in moderately polar polysiloxanes (silicones)
steel capillary. (From Schurig V and Weber R (1981) Journal of such as OV-1701. Thus, the CD derivatives can be
Chromatography 217: 51}70.)
employed for GC enantiomer separation irrespective
of their melting point and phase transitions. The
preparation of immobilized polymeric CSPs (Chirasil- simultaneous separation of a test mixture of enantio-
Metal; [X] in Scheme 2). In Figure 3 the GC mers of different classes of compounds is depicted in
enantiomer separation of homofuran at 953C on Figure 4.
Chirasil-Nickel [X] is shown. The presence of three hydroxyl groups that can be
regioselectively alkylated and acylated offers an enor-
Chiral Stationary Phases Based on Inclusion
mous number of possible -, - and -cyclodextrin
The Rrst enantiomer separation using an inclusion- derivatives, which are not always readily accessible
type CSP in GC was reported in 1983 for - and and may require tedious puriRcation steps. Occa-
-pinene and cis- and trans-pinane on packed col- sionally, CD derivatives such as octakis(3-O-
umns containing native -cyclodextrin in formamide. butanoyl-2,6-di-O-n-pentyl)--cyclodextrin [XVII]
Later it was recognized that alkylated cyclodextrins are highly enantioselective for the GC enantiomer
Figure 3 Enantiomer separation of homofuran at 953C and transient elution profiles at higer temperatures on Chirasil-Nickel [X].
(A) Experimental gas chromatograms (column 10 m;0.1 mm (i.d.) fused silica capillary; film thickness, 0.25 m). (B) Simulated
chromatograms. (From Schurig V, Jung M, Schleimer M and KlaK rner F-G (1992) Chem. Ber. 125: 1301}1303.)
Heptakis(2,3,6-tri-O-methyl)--cyclodextrin [XI]
Heptakis(2,6-di-O-methyl-3-O-trifluoroacetyl)--cyclodextrin [XII]
Hexakis(2,3,6-tri-O-n-pentyl)--cyclodextrin (Lipodex A) [XIII]
Hexakis(3-O-acetyl-2,6-di-O-n-pentyl)--cyclodextrin (Lipodex B) [XIV]
Heptakis(2,3,6-tri-O-n-pentyl)--cyclodextrin (Lipodex C) [XV]
Heptakis(3-O-acetyl-2,6-di-O-n-pentyl)--cyclodextrin (Lipodex D) [XVI]
Octakis(3-O-butanoyl-2,6-di-O-n -pentyl)--cyclodextrin (Lipodex E) [XVII]
Hepatakis(2,3-di-O-acetyl-6-O-t -butyldimethylsilyl)--cyclodextrin [XVIII]
Hepatakis(6-O-t-butyldimethylsilyl-2,3-di-O-methyl)--cyclodextrin [XIX]
Heptakis(O-(S-2-hydroxypropyl)-per-O-methyl)--cyclodextrin (PMHP--CD, mixture) [XX]
Hexakis(2,6-di-O-n-pentyl)--cyclodextrin (Dipentyl--CD) [XXI]
Heptakis(2,6-di-O-n-pentyl)--cyclodextrin (Dipentyl--CD) [XXII]
Heptakis(3-O-trifluoroacetyl-2,6-di-O-n-pentyl)--cyclodextrin (DPTFA--CD) [XXIII]
Scheme 3 Cyclodextrin-type chiral stationary phases.
separation of certain racemates (cf. Figure 5). Also E long-term stability with absence of droplet forma-
derivatives containing the bulky butyldimethylsilyl tion leading to loss of efRciency;
substituent at the lower rim of the CD ([XVIII] and E immobilization by crosslinking and/or surface
[XIX]) represent useful complementary CSPs. bonding;
A superior class of CSP has been obtained by chemic- E compatibility with all injection techniques.
ally linking the CD derivatives to the polysiloxane back-
bone furnishing Chirasil-Dex [XXIV] (cf. Scheme 4). The rationalization of chiral recognition involving
Fused silica columns coated with Chirasil-Dex CD derivatives is difRcult since almost all classes of
have advantages such as: chiral compounds, ranging from apolar to highly po-
lar, are susceptible to enantiomer separation on a cer-
E use of a nonpolar polysiloxane matrix (in which tain CD-derived CSP, often with no logical depend-
CD derivatives cannot be physically diluted) result- ence on molecular shape, size and functionalities of
ing in low elution temperatures for polar analytes; the selectand and the selector (, , ). Clearly, multi-
E high degree of inertness allowing analysis of polar modal recognition processes are important which
compounds without prior derivatization;
E higher CD concentration resulting in increased sep-
aration factors;
Figure 4 Enantiomer separation of the test mixture -pinene

(1, 2), trans-pinane (3, 4), cis -pinane (5, 6), 2,3-butanediol (rac)
(7, 8), 2,3-butanediol (meso) (9), -valerolactone (10, 11), 1-
phenylethylamine (12, 13), 1-phenylethanol (14, 15) and 2-ethyl- Figure 5 Enantiomer separation of the inhalation anaesthetics
hexanoic acid (16, 17) by GC on heptakis(2,3,6-tri-O -methyl)-- desflurane, isoflurane and enflurane by GC on immobilized
cyclodextrin [XI] (10% (w/w) in OV-1701) at 503C and 0.7 bar polysiloxane-bonded octakis(3-O-butanoyl-2,6-di-O -n-pentyl)--
(gauge) helium. Column, 50 m;0.25 mm (i.d.) fused silica capil- cyclodextrin [XVIII] at 283C. Column, 10 m;0.25 mm (i.d.) fused
lary; film thickness, 0.25 m. (Courtesy Chrompack International, silica capillary; film thickness, 0.18 m. (From Grosenick H and
Middelburg, The Netherlands.) Schurig V (1997) Journal of Chromatography A 761: 181}193.)
Above Tiso the sign of enantioselectivity changes,

leading to peak inversion (second kind). Below the
coalescence temperature, the sign of enantioselectiv-
ity R,S(G) is governed by !R,S(H) and above it,
by R,S(S). A rare example of the temperature-de-
pendent reversal of enantioselectivity in GC enan-
tiomer separation on a single CSP is demonstrated in
Figure 6. Usually, even at high temperatures, enan-
tioselectivity is dominated by enthalpy control and
separation factors increase with decreasing temper-
ature. Therefore, it is recommended that the lowest
possible temperature is used for enantiomer separ-
ation by GC.
Scheme 4 Chirasil-Dex-type chiral stationary phase. For undiluted CSPs the quantity !R,S(G) can
easily be obtained from the separation factor undil,
may involve inclusion, hydrogen bonding, dipole}
dipole interactions and dispersion forces. Since enan-
tiomer separations have also been observed with
per-n-pentylated amylose, inclusion may not be a pre-
requisite for chiral recognition using CDs. Mechanis-
tic investigations, some of which include molecular
modelling studies, have been carried out although no
clear-cut rationale for chiral recognition has emerged
thus far.
Thermodynamics of Enantiomer
Separation
Enantiomer separation by GC is brought about by the
difference in the Gibbs free energy !R,S(G) of the
diastereomeric association equilibria between the en-
antiomers (selectand) and the CSP (selector). An
important prerequisite is a fast and reversible associ-
ation equilibrium (fast kinetics). The chemical associ-
ation equilibria in the stationary phase are described
by KR and KS, with R referring to the second eluted
enantiomer and S to the Rrst eluted enantiomer. For
enantiomer separation, the Gibbs}Helmholtz equa-
tion [1] applies, where R is the universal gas constant,
T is the temperature (K), H is the enthalpy and S is the
entropy.
KR
!R,S(G)"RT ln "!R,S(H)#T R,S(S)
KS
[1]
For a 1 : 1 molecular association, the quantities

R,S(S) and R,S(H) display an opposing effect on
Figure 6 (A) Temperature-dependent peak inversion (second
!R,S(G). At the isoenantioselective temperature kind) at 55 and 1103C and peak coalescence (second kind) at
Tiso, given by eqn [2], peak coalescence (second kind) 703C during GC enantiomer separation of isopropyloxirane on
occurs (R,S(G)"0, KR"KS; no enantiomer separ- nickel (II) bis[3-(heptafluorobutanoyl)-(1R )-8-methylene-cam-
ation). phorate] (a derivative of [IX]) (0.126 molal in OV-101). Column,
22 m;0.25 mm (i.d.) glass capillary. (B) Linear van’t Hoff plot and
determination of Tiso. (From Schurig V (1997) In: Jinno K (ed.)
R,S(H)
Tiso" [2] Chromatographic Separations Based on Molecular Recognition,
R,S(S) ch. 7, pp. 371}418. New York: Wiley-VCH.)
according to eqn [3]. conRguration and the order of elution for enantio-
mers belonging to homologous series of compounds.
!R,S(G)"RT ln undil [3] Although consistent relationships between the order
of elution and absolute conRguration of congeners
Although it is occasionally used, eqn [3] is not valid have been observed in many instances, remarkable
for diluted CSPs because dil is concentration depen- inconsistencies are also known. As a rule, such com-
dent. parisons, if any, should be restricted to measurements
at the same temperature since peak inversion (second
Enantiomerization kind) may occur at different temperatures as the re-
sult of enthalpy versus entropy compensation or
The conRgurational integrity of the enantiomers dur- multimodal chiral recognition mechanisms. There-
ing the GC process of separation is essential for a cor- fore, the assignment of absolute conRgurations by GC
rect enantiomer analysis. When enantiomers invert can be ambiguous.
the conRguration (or conformation) during separ-
ation, transient elution proRles are obtained that are
characterized by plateau formation between the ter- Method of Enantiomer Labelling
minal peaks of the enantiomers. The barrier of enan- Enantiomers can be quantiRed in complex matrices
tiomerization (GS) can be determined by dynamic when a known amount of the pure enantiomer is
GC via peak form analysis of interconversion proRles added as an internal standard. The pure enantiomer is
and the comparison of experimental and simulated an ideal internal standard as the enantiomeric excess
chromatograms (cf. Figure 2). Only minute amounts is not inSuenced by sample manipulations in diluted
of the easily available racemic compound are re- systems (achiral derivatization, dilution, injection,
quired. If enantiomerization is fast within the detection, chemical and physical losses). The method
chromatographic timescale, peak coalescence (third of enantiomer labelling presupposses the precise
kind) occurs (cf. Figure 2 at 1303C). knowledge of the enantiomeric excess of the sample
Another on-column method for determining inter- and the standards.
conversion kinetics is based on the ‘stopped-Sow’ Simultaneous enantiomer and isotopic labelling in
technique. In the Rrst part of the column enantiomers enantiomer analysis can also be carried in the GC-
are quantitatively separated. Afterwards, the Sow is MS-SIM mode.
stopped and the column is heated, whereby enan-
tiomerization in the separated fractions commences.
After cooling, the Sow is restored and the enan- Precision and Sources of Error
tiomerized fractions are separated in the second part The precision of enantiomeric excess determined by
of the column. From the reaction time and enan- GC is high over the whole range from 0 (racemic) to
tiomeric compositions the rate constant can be cal- 99.9% (nearly enantiomerically pure). At a high
culated. Using a combination of three columns, i.e. enantiomeric excess, the minor enantiomer should
a separation column, an empty reactor column and preferentially elute as the Rrst peak in order to facili-
another separation column, connected via switching tate correct integration.
valves for peak-cutting, enantiomerization can be Despite the great success of GC for determining
carried out in the gas phase and in the absence of the enantiomeric excess, potential sources of error should
CSP in the reactor column (enantioselective multi- be considered:
dimensional stopped-Sow GC).
E decomposition of the sample during chromatogra-
phy (the enantiomer which spends a longer time in
Assignment of Absolute the column will be lost preferentially, causing an
Con\gurations by error in enantiomeric excess);
Enantioselective GC E coelution of impurities accidentally increasing
peak areas;
The determination of absolute conRgurations of
E enantiomerization causing peak distortions (pla-
chiral analytes is an important task in enantiomer
teau formation);
analysis. Absolute conRgurations of minute amounts
E peak distortions caused by inadequate instrumen-
of chiral samples may be determined directly, and free
tation;
of chiroptical evidence, by GC via co-injection of
E nonlinear detector response.
reference compounds with known stereochemistry.
Absolute conRgurations may also be predicted in- In general, the error in enantiomeric excess due to
directly by empirical rules that correlate the absolute decomposition of the analyte can be reduced if the
difference of the residence time in the column is internal diameter of the columns to 0.1 and 0.05 mm
minimized for both enantiomers. This goal may be (i.d.) requires thinner Rlms of the stationary phase in
realized by using short columns, high pressure drops, order to keep the phase ratio constant. The reduced
elevated temperatures and CSPs exhibiting only amount of stationary phase decreases the sample ca-
small separation factors . A rather frequent cause for pacity. The reduced signal-to-noise ratio may become
the deviation from the expected 1 : 1 ratio for the critical in regard to the precision of the enantiomeric
racemic mixture consists of the coelution of im- excess determination.
purities. This interference can be recognized by deter- Unfortunately, there is no universal CSP available
mining the enantiomeric excess on two columns and column selection is a matter of trial and error.
coated with CSPs of opposite chirality. The veriRca- Chirasil-Val [IV], Chirasil-Dex [XXIV] and per-
tion of the ideal 1 : 1 ratio of a racemic mixture is methylated -cyclodextrin dissolved in polysiloxane
always recommended in enantiomer analysis by [XI] represent the most popular CSPs. On a given
chromatography. It may also be used to test integra- enantioselective column, the parameters column
tion devices. length, temperature, Rlm thickness, concentration of
CSP, mobile phase velocity and their inSuences on the
resolution factor Rs (which is governed by the reten-
Practical Considerations tion factor k, separation factor and efRciency N)
The merit of GC enantiomer separation is the great have to be carefully balanced. Some variables cannot
range of resolvable classes of compounds. With a few be freely selected when commercial columns are
exceptions, enantiomer separation by GC is charac- used.
terized by low separation factors , and, as a (beneR- While Chirasil-Val [IV] and Chirasil-Metal [X] are
cial) consequence, reduced separation times. The use available in both enantiomeric forms, Chirasil-Dex
of highly efRcient capillary columns is recommended. [XXIV] and other modiRed cyclodextrins, [XI]}
Chirasil-Val [IV], Chirasil-Dex [XXIV], and most [XXIII], occur only in the D form. The strategy of
cyclodextrin derivatives, [XI]}[XXIII], coated onto peak inversion (Rrst kind), employed to elute the
fused silica capillary columns, are commercially minor enantiomer as the Rrst peak, is precluded with
available. Factors such as availability, price, perfor- carbohydrate-based CSPs.
mance and reproducibility should guide the analyst
when selecting chiral stationary phases for GC. Im- See also: II/Chromatography: Gas: Derivatization.
mobilized chiral stationary phases, such as Chirasil- III/Chiral Separations: Chiral Derivatization; Cyclodex-
Dex [XXIV], have the advantage of solvent compati- trins and Other Inclusion Complexation Approaches;
bility, resistance to temperature shock and longevity. Ion-Pair Chromatography; Ligand Exchange Chromatog-
Enantiomer separation on Chirasil-type stationary raphy; Liquid Chromatography; Supercritical Fluid Chrom-
atography; Synthetic Multiple Interaction (‘Pirkle’) Station-
phases can be performed in the usual temperature
ary Phase.
range 25}2203C. For special applications it is also
possible to use cryogenic temperatures down to
!203C. Further Reading
The dimensions of commercial columns are typi-
Gil-Av E (1975) Present status of enantiomeric analysis by
cally 10}25 m;0.25 mm (i.d.) and the Rlm thickness gas chromatography. Journal of Molecular Evolution 6:
of the chiral stationary phase is 0.25 m. Column 131}144.
miniaturization has important merits. Since enan- Helmchen G, Hoffmann RW, Mulzer J and Schaumann E
tiomer separation represents a binary separation sys- (eds) (1995) Houben-Weyl, vol E 21a Stereoselective
tem, the whole elution window required for multi- Synthesis, Schurig V Gas Chromatography, Stuttgart:
component mixtures need not be exploited unless Thieme, 168}187.
enantiomers are detected in complex matrices. With KoK nig WA (1987) The Practice of Enantiomer Separation
shorter columns, the elution temperature can be de- by Capillary Gas Chromatography. Heidelberg: HuK thig.
creased, so that the chiral separation factor is in- KoK nig WA (1992) Enantioselective Gas Chromatography
creased in the common enthalpy-controlled region of with ModiTed Cyclodextrins. Heidelberg: HuK thig.
KoK nig WA (1993) Enantioselective gas chromatography.
enantioselectivity. The loss of efRciency is compen-
Trends in Analytical Chemistry 12: 130}137.
sated by the gain of selectivity leading to com- Koppenhoefer B, Epperlein U and Schwierskott M (1997)
parable resolution factors Rs. The shorter analysis Fresenius Journal of Analytical Chemistry 359: 107}114.
times increase the sharpness of peaks and hence the Schreier P, Bernreuther A and Huffer M (1995) Analysis
detectability of the enantiomers. Recommended are of Chiral Organic Molecules}Methodology and Ap-
2 m;0.25 mm (i.d.) columns with a Rlm thickness of plications, Ch 3.5, pp. 132}233. Berlin: Walter de
0.25 m. Further miniaturization via reduction of the Gruyter.
2358 III / CHIRAL SEPARATIONS / Ion-Pair Chromatography
Schurig V (1984) Gas chromatographic separation of enan- Angewarde Chemie International Edition English 29:
tiomers on optically active stationary phases. Angewarde 939}957.
Chemie International Edition English 23: 747}765. Snopek J, SmolkovaH -KeulemansovaH E, CserhaH ti T, Gahm K
Schurig V (1994) Enantiomer separation by gas chromato- and Stalcup A (1996) Cyclodextrins in analytical separ-
graphy on chiral stationary phases. Journal of ation methods. In: Atwood JL, Davies JED, MacNicol
Chromatography 666: 111}129. DD and VoK gtle F (eds) Comprehensive Supramolecular
Schurig V and Nowotny H-P (1990) Gas chromatographic Chemistry, vol. 3, ch. 18, pp. 515}571. Oxford:
separation of enantiomers on cyclodextrin derivatives. Pergamon.
Ion-Pair Chromatography
E. Heldin, Uppsala University, Uppsala, Sweden The nonpolar environment may be a liquid (mobile or
stationary), a surface or a micelle in the liquid. For
simplicity the principle for a liquid}liquid chromato-
graphic system with aqueous mobile phase will be
presented. Here the retention factor for the solute,
Introduction kHB#, is equal to:
In ion pair chromatography, the solute ion is distrib-
uted between the mobile and the stationary phase kHB#"DHB#;(Vs/Vm) [2]
together with an ion of opposite charge (a so-called
counterion). The technique is often used in reversed- where DHB# is the distribution coefRcient of the sol-
phase chromatography as a convenient method to ute between the aqueous mobile phase and the or-
control the retention of solutes. The principle may, ganic stationary phase and Vs and Vm are the volumes
for example, be applied to direct chiral separations of the two phases.
using either chiral or achiral counterions. When using The distribution coefRcient is deRned as:
an achiral counterion, the chromatographic system
has to contain a chiral selector, e.g., a chiral station- Corg
D" [3]
ary phase. The purpose of an achiral counterion then Caq
is to control the retention of the solute. However, the
counterion may also inSuence the stereoselectivity by i.e., the ratio of the total concentration of solute in
interactions with the chiral selector molecules. A chiral organic phase over the total concentration in aqueous
counterion may, on the other hand, be used with non- phase. For the ion pair, DHB# may be expressed as:
chiral stationary phases to promote chiral separation.
[HBC]org
DHB#" "Kex;[C\]aq [4]
[HB#]aq
Ion Pairs: Principles
In order to distribute the solute molecules in a nonpo- Thus, the retention factor will depend on the extrac-
lar environment, it has to be uncharged. However, if tion constant of the ion pair and the concentration of
an ion of opposite charge is present in enough concen- counterion in the aqueous mobile phase. The magni-
tration the two ions may be distributed as a pair in the tude of the extraction constants depends on the hy-
same way. An ion being double charged may be drophobicity of the solute and the counterion, the
distributed together with two ions of single charge or
one double charged, the only prerequisite being
electroneutrality of the pair (Figure 1).
The equilibrium for a simple 1 : 1 ion pair may be
given as:
HB##C\ HBC
and is characterized by an extraction constant:
[HBC]org
Kex" # [1] Figure 1 Distribution of charged compounds to a nonpolar
[HB ]aq;[C\]aq phase together with a counterion.
III / CHIRAL SEPARATIONS / Ion-Pair Chromatography 2359
kind of interaction forces between the two ions, but the models otherwise resemble each other to
and the physiochemical properties of the organic a large extent.
phase.
For protolytic solutes, the retention will also de- Principles of Chiral Separation
pend on the pH of the aqueous phase, as the charge of
the solute is pH dependent. At a pH where an amine
Using Achiral Counterions
is uncharged, its distribution coefRcient is: The achiral counterion itself does not provide
stereochemical interactions and thus it is expected to
[B]org only inSuence the retention and not the separation of
DB " "KD(B) [5] an enantiomeric pair. Different counterions have
[B]aq
been added to the aqueous mobile phase in a
where KD(B) is the distribution constant of B. liquid}liquid chromatographic system with a nonpo-
Considering a more general expression for the dis- lar chiral stationary phase (a tartaric acid ester). The
tribution coefRcient for a basic compound over the enantioselectivity was about the same for all counter-
entire pH range in the presence of a counterion, the ions studied and, as expected, the type of ion and its
equation is as follows: concentration could be used to control the retention
(Table 1).
Corg [B]org#[HBC]org Several of the chiral selectors used are, however,
D" " complex biomolecules, e.g., proteins which cannot be
Caq [B]aq#[HB#]aq
treated as simple organic phases. The addition of
KD(B)K(HB#) Kex(HBX)[C\]aqaH# charged (and uncharged) compounds to the mobile
" # [6] phase, when a protein is used as the stationary phase,
K(HB#)#aH# K(HB#)#aH#
Uncharged form Ion pair
may change both the retention and the enantioselec-
tivity. The changes in enantioselectivity may be dra-
matic; the two enantiomers in a pair may even elute in
where K(HB#) is the acid dissociation constant of HB#
an opposite order.
deRned as:
[B] ) aH# Principles of Chiral Separation

K[HB#]" [7] Using Chiral Counterions
[HB#]
The chiral counterion may be the only chiral com-
Equation [6] means that the retention of a solute may pound needed for separation of enantiomers in
be made up of two parts: the retention as an ion pair a chromatographic system. The solute enantiomers
and the retention in an uncharged form: form diastereomeric ion pairs with the chiral counter-
ion, also called the selector (Figure 2). In order to
k"kuncharged#kion pair [8] achieve enantioselectivity, the ion pairs formed
should have different stabilities and/or different dis-
In chromatographic systems with solid stationary tribution properties with the stationary phase. If the
phases, the retention involves distribution with difference in distribution properties with the station-
a solid surface. The limited capacity of the surface ary phase is responsible for the stereoselectivity, it is
and the possibility to have competition for the limited important that the retention as a diastereomeric ion
adsorption sites have to be considered in the equation pair (kchiral) is more pronounced than the retention of
Table 1 Influence of counterion structure on retention and enantioselectivity
Solute ! ClO\4 PF\6 Heptane SO\

3 Hexane SO\
3 Hexane SO\
3
(45 mM) (45 mM) (45 mM) (45 mM) (90 mM)
k k k k k k
N-methylephedrine 1.14 1.1 1.13 2.2 1.13 4.6 1.11 7.3 1.12 3.4 1.10 6.2
Ephedrine 1.15 1.1 1.15 2.2 1.15 4.0 1.13 7.7 1.41 3.4 1.13 6.9
Norephedrine 1.16 1.2 1.16 2.5 1.16 4.2 1.14 9.6 1.15 4.1 1.14 8.6
Solid phase: Novapak Phenyl (4 m-particles); liquid stationary phase, (2R,3R )-di-n-butyltartrate; mobile phase, counterion in phos-
6 "hexafluorophosphate, Heptane SO\
phate buffer pH 2.8 (ionic strength 0.1). PF\ 3 "heptanesulfonate, Hexane SO\ 3 "hexanesul-
fonate.
Figure 2 Ion pair formation in mobile phase and possible distribution to stationary phase.
the solute alone (kachiral) (Figures 2 and 3). Referring K*RR"adsorption constant of the diastereomeric
to eqn [8] the retention factor may be described as: ion pair (R)}Solute : (R)}Selector.
K* "adsorption constant of the diastereomeric
k"kachiral#kchiral [9] SR
ion pair (S)}Solute : (R)}Selector.
and the observed value (R/S) compared to the maxi- [(R)}Selector]mob"concentration of chiral selector
mal (only chiral retention) as: in the mobile phase.
kR max#(kachiral/kS(chiral)) The chiral selector does not have to be optically pure,
R/S" " [10] as the interactions are noncovalent. An enantiomeri-
kS 1#(kachiral/kS(chiral))
cally impure selector will affect the stereoselectivity
where the S-form is eluted Rrst. Therefore, the con- but only an impurity of exactly 50% may ruin it
centration of selector has to reach a certain level to completely. The expected stereoselectivity, HS/R when
assure maximal stereoselectivities ( values). using enantiomerically impure selectors in the mobile
In order to promote the formation of an ion pair in phase may be calculated from the following expression:
the mobile phase, solvents of low polarity have often
been used, e.g., hexane and dichloromethane. The S/R[(R)}Selector]mob#[(S)}Selector]mob]
retention and stereoselectivity is controlled by the H
S/R"
[(R)}Selector]mob#S/R[(S)}Selector]mob]
temperature, the mobile phase composition (i.e. type
and concentration of counterion, competing substan- S/RP#[100!P]mob
" [12]
ces and polar modiRers) as well as the choice of P#S/R[100!P]mob
stationary phase.
The chiral counterion and its concentration in- where S/R is the stereoselectivity obtained when the
Suences the equilibrias outlined in Figure 2. Equation selector is present in one pure enantiomeric form
[11] expresses the stereoselectivity when varying the (here the R-form) and P is the fraction of selector
counterion concentration. From eqn [11] it is obvious present in the R-form. Experiments have shown good
that the inSuence of selector concentration on enan- agreement with theory (Table 2).
tioselectivity is difRcult to predict. The effect of
a changed selector concentration will depend on the Chiral Counterions Used
relative magnitudes of ion pair formation constants in
the mobile phase and stereoselective distribution to In order to obtain chiral discrimination of the solutes,
the stationary phase: the counterion and solute ions have to establish
(K1#K1KRR[(R)}Selector]mob#KSRKHSR[(R)}Selector]mob#KSRKRRKHSR[(R)}Selector]2mob)
S/R" [11]
(K1#K1KSR[(R)}Selector]mob#KRRKHRR[(R)}Selector]mob#KSRKRRKHRR[(R)}Selector]2mob)
K1"adsorption constant for achiral adsorption of multipoint interactions where the stability of the in-
solute. teractions differ between the enantiomers. For the
KRR"formation constant of (R)}Solute : (R)}Se- stability of the diastereomeric ion pair, it is advant-
lector in mobile phase. ageous if the molecules involved have a rigid struc-
KSR"formation constant of (S)}Solute : (R)}Selec- ture, and most of the chiral counterions used are
tor in mobile phase. rigid. At least three points of interaction for one of the
symmetrical peaks. Uncoated silica has in some cases

been used for the purpose of retaining hydrophilic
compounds. As the solid phase, i.e., the particle sur-
face, plays an important role in the chiral recognition
process it should be considered in the optimization
process. The introduction of porous graphitic carbon
(PGC) as packing material thus opened new possibili-
ties and has broadened the use of chiral mobile phase
additives. PGC is a very hydrophobic material whose
unique three-dimensional structure is suitable for the
recognition of isomers. Another advantage is that it
may be used with more polar mobile phases, e.g.,
methanol}water mixtures. The use of polar mobile
phases has several advantages in the separation pro-
cess, among them the possibility of direct injection of
biological Suids.
(ⴙ)-10-Camphorsulfonic Acid and Analogues

Figure 3 Influence of achiral retention on the separation factor
for the enantiomers. (Reprinted from Heldin E. (1991) Tartaric (#)-10-Camphorsulfonic acid [I] has been used to
acid derivatives as chiral selectors in liquid chromatography. separate the enantiomers of amino alcohols, mainly
Comprehensive Summaries of Uppsala Dissertations from the -blocking agents (derivatives of 1-aryl-2-amino al-
Faculty of Pharmacy 74: 1}37, with permission from Acta Univer-
sitas Upsaliensis.)
cohols). These solutes have an aromatic part to which
a side chain is connected. The selectivity has been
found to be dependent on the distance between the
enantiomers are necessary for recognition of the hydroxyl and the amine groups, implying that both
three-dimensional structure. All these points do not, these structural features play an important role in
however, have to be only between selector and solute chiral recognition. (#)-10-Camphorsulfonic acid is
but may also involve solvent molecules and the an aprotic acid and it also contains an oxo group
achiral packing material. which may take part in hydrogen bonding. A model
Most of the early studies with chiral ion-pairing for the interaction between (#)-10-camphorsulfonic
selectors in the mobile phase were performed with acid and -blocking agents has been suggested and is
a silica-based DIOL stationary phase. The DIOL shown in Figure 4. Several chiral separations have
functionality was chosen after comparing several been performed using (#)-10-camphorsulfonic acid
phases (nitrile, nitro, C2, silica) as it gave the most in dichloromethane : 1-pentanol (199 : 1) as mobile
phase and a DIOL stationary phase. When using
Table 2 Influence of enantiomeric purity of counterion on
a DIOL functionalized packing material and sulfonic
stereoselectivity acid in the mobile phase, it is of vital importance to
choose a solvent with a low content of polar compo-
D-form of counterion (%) H
S/R nents as the mobile phase in order to promote ion pair
Calculated Found
0 } 1.38
5 1.34 1.33
10 1.29 1.28
15 1.23 1.26
50 1.00 1.01
100 0.72a 0.72a
a H
R/S"1.38. Solid phase: LiChrosorb DIOL. Mobile phase:
2.5 mM mixtures of benzoxycarbonylglycyl-L- and D-proline and
0.25 mM triethylamine in dichloromethane (80 ppm water).
HS/R"k for (S )-propranolol/k for (R )-propranolol. Results re- Figure 4 Model for interaction between the (#)-10-camphor-
produced from Pettersson C, Karlsson A and Gioeli C (1987). sulfonic acid and amino alcohols. (Reprinted from Pettersson
Influence of enantiomeric purity of a chiral selector on C and Schill G (1986) Separation of enantiomers in ion-pair
stereoselectivity. Journal of Chromatography 407: 217}229, with chromatographic systems. Journal of Liquid Chromatography 9:
permission from Elsevier Science. 269}290, by courtesy of Marcel Dekker, Inc.)
formation. Attempts to use Br-3-(#)-10-camphorsul- with a mobile phase consisting of 5;10\3 M (#)-10-
fonic acid in a system with a DIOL stationary phase, camphorsulfonic acid in dichloromethane : methanol
in order to separate -blocking agents, have not been (25 : 75) (Figure 5). As stated above, PGC may be
successful. With these solutes and in that particular used together with polar mobile phases. A fairly high
system, the Br group was believed to sterically pre- content of dichloromethane is needed in order to
vent the three-point contact between the selector and elute the dihydropyridines from the highly retentive
the solute. Moving the sulfonate group away from the surface of PGC. Exchanging (#)-10-camphorsulfonic
oxy group to position 8 (Br-3-(#)-8-camphorsul- acid for Br-3-(#)-10-camphorsulfonic acid still sep-
phonic acid) restored the stereoselectivity to some arated the enantiomers although with somewhat
extent. lower selectivities.
Cinchona Alkaloids
The cinchona alkaloids are rigid molecules contain-
ing four chiral carbons [II] and their use as chiral
selectors have been studied. Quinine has the absolute
conRguration 3(R),4(S),8(S),9(R) while quinidine has
3(R),4(S),8(R),9(S). It is at the positions 8 and 9 that
the chiral recognition is believed to take place due to
hydrogen bonding groups, and the exchange of quin-
R1 R2 R3
ine for quinidine reverses the retention order of some
enantiomers, e.g., 10-camphorsulfonic acid and O-
(#)-10-Camphorsulfonic acid SO3H H H
(#)-3-Br-10-Camphorsulfonic acid SO3H H Br
methylmandelic acid (Table 3). The mechanism of
(#)-3-Br-8-Camphorsulfonic acid H SO3H Br chiral recognition is, however, not simple as the effect
of an exchange of the hydroxyl for an ethylcarbonate
group at position 9 reduces the enantioselectivities
The selector has been used in a system with porous but not to the same extent for all solutes. Further-
graphitic carbon (PGC) as support in order to separ- more, the exchange of an ethoxy group for hydrogen
ate some dihydropyridines. The enantiomers of am- at the aromatic moiety (cinchonidine) also had an
lodipine and an analogue, UK52.829, were separated inSuence on the stereoselectivity. The results in
Figure 5 Separation of racemic amlodipine and UK52.829 on Hypercarb-STM. Column temperature: 303C. Mobile phase: 5 mM (1S )-
(#)-10-camphorsulfonic acid in dichloromethane}ethanol (25 : 75, v/v). Flow-rate: 1.5 mL min\1. (Reprinted from Josefsson M,
Carlsson B and Norlander B (1994) Chiral ion-pair chromatographic separation of two dihydropyridines with camphorsulfonic acids on
porous graphitic carbon. Journal of Chromatography 684: 23}27, with permission from Elsevier Science.)
Table 3 Influence of counterion structure on retention and enantioselectivity
Solutes Quinine ethyl Quinine Quninidine Cinchonidine

carbonate
k1 k1 k1 k1
10-Camphorsulfonic acid 1.28 1.7(#) 1.47 5.2(!) 1.30 3.0(#) 1.24 6.7(!)
N-(1-Phenylethyl)phthalamic acid 1.01 12.9a 1.14 6.7(#) 1.15 4.2(!) 1.18 9.6(#)
O-methylmandelic acid 1.12 7.8(!) 1.18 3.8(#) 1.02 8.7b
-Methoxy--trifluoromethylphenylacetic acid 1.09 1.7a 1.16 2.4(!) 1.10 1.1(#) 1.0 2.6
Solid phase: LiChrosorb DIOL. Mobile phase: 0.35 mM chiral amine and 0.35 mM acetic acid in dichloromethane (dry) : 1-pentanol
(99 : 1). "k 2/k 1.
a
Mobile phase consisted of 0.35 mM quinine ethyl carbonate and 0.35 mM acetic acid in dichloromethane (dry) : n-hexane : 1-pentanol
(49 : 50 : 1). Results from Pettersson C and No K (1983) Journal of Chromatography 282: 671}684 and Pettersson C
(1984) Journal of Chromatography 316: 553}567 with permission from Elsevier Science.
b
Retention order uncertain.
Table 3 come from two different mobile phases. the resolution of D- and L-dansyl-valine is shown in
Stereoselectivities should be comparable, as n-hexane Figure 6.
does not have a signiRcant effect on enatioselectivity.
Peptides
A high content of n-hexane is needed to retain hy-
drophilic compounds. The Rrst peptide used for a chiral separation was
L-leucine-L-leucine-L-leucine, a zwitterionic com-
pound. Some amino acids, including, D,L-tryptophan
and glycyl-D,L-phenylalanine, have been partially
resolved by reversed-phase chromatography. The
N-protected dipeptide N-benzyloxycarbonyl-glycyl-
L-proline (L-ZGP), shown in Table 4, has been used
successfully as a counterion in the separation of
enantiomers of amines. When bare silica or modiRed
silicas are used as packing material, an enantiomeric
pair generally has to be a -amino alcohol in order
to be separable. In order to separate the enantiomers
of amines (which lack a strong hydrogen-bonding
The complexity of the systems is high as the solutes group in the vicinity of the asymmetric carbon atom)
may form ion pairs in the mobile phase, as well as be or rather hydrophilic amino alcohols, PGC is a
distributed to the stationary phase both as ion pairs good choice of stationary phase. Most recent separ-
and as uncharged compounds (see Figure 2). In addi- ations have been performed on PGC material and
tion, the selectors may be distributed to the stationary a large number of different enantiomers have been
phase alone or together with mobile phase ions. In the shown to be separable, of which some examples are
cinchona systems, the mobile phase often contains an given in Table 5. The inSuence of the structure
acid to control the retention and the stereoselectivity of the chiral selector on the enantioselectivity of
of the acidic solutes. It is not possible to provide alprenolol has been studied using both DIOL func-
a simple guideline to know which acid to use as tionalized particles and PGC as packing material.
the effect on enantioselectivity differs between the In the DIOL system, one analogue of ZGP with a
solutes studied. The distribution to the packing ma- esteriRed acid function (L-ZGP methyl ester) and
terial is important and inSuences both the retention one lacking the glycyl group (Z-L-proline) have
and the enantioselectivity. been tested. It was found that the acidic function
When optimizing chiral ion pair separations it is is needed and that this function, together with the
therefore important to carefully consider both the keto group in the benzyloxycarbonyl group, was
mobile phase and the packing material. Enantiomers enough for enantioselectivity. Exchanging the glycyl
of N-blocked amino acids, sulfonic acids and car- in L-ZGP for L-alanyl will, however, completely
boxylic acids have been separated with the cinchona ruin the stereoselectivity and the methyl group in
alkaloids as mobile phase additives. An example of alanyl is believed to block the hydrogen-accepting
garded as three supplementary selectors. Although all

three show enantioselectivity for metoprolol, L-ZGP
is not ‘usable’ for metoprolol analogues having a dis-
tance larger than two carbons between the hydroxyl
and amine function. In order to match the increase in
the distance between possible points of interaction,
introduction of a hydrogen-bonding group further
apart in the selector promotes the enantioselectivity.
Recently, two new N-derivatized dipeptides, N-
benzyloxycarbonyl-L-glutamyl-L-proline (Z-L-Glu-L-
Pro) and N-benzyloxycarbonyl-L-glutamyl-D-proline
(Z-L-Glu-D-Pro), have been reported. Only one of the
dipeptides, Z-L-Glu-L-Pro, shows enantioselectivity
for the amino alcohols studied (metoprolol and ana-
logues), although ion pairing occurred with both pep-
tides. The possibility of using Z-L-Glu-L-Pro at tem-
peratures above ambient has been investigated. In
contrast to using L-ZGP, the enantioselectivity is,
however, only slightly affected when increasing the
temperature. Although the retention is decreasing at
elevated temperatures, the enantiomers of, for
example, metoprolol, may be separated within a wide
temperature range, 10 to 453C. An enantiomeric sep-
aration of metoprolol is given in Figure 7. It may also
be worth noting that the elution order of the enantio-
mers were opposite using Z-L-Glu-L-Pro and L-ZGP
as selectors, even though in both cases the terminal
peptide was L-proline. The relative retention order of
the studied amino alcohols was independent of the
selector used, L-ZGP or Z-L-Glu-L-Pro.
Figure 6 Resolution of D,L-dansyl-valine. Solid phase: LiCh- ( ⴚ )-2,3:4,6-Di-O-isopropylidene-2-keto-L-
rosorb DIOL. Mobile phase: Quinine and acetic acid 0.35 mM in
gulonic Acid ((ⴚ)DIKGA)
dichloromethane (dry) #1-pentanol (99#1). (Reprinted from
Pettersson C (1984) Chromatographic separation of enantiomers (!)-DIKGA [III], has been used to resolve enantio-
of acids with quinine as chiral counterion. Journal of Chromato-
graphy 316: 553}567, with permission from Elsevier Science.)
mers by fractional crystallization and has also been
Table 4 Counterion structure and stereoselectivity

keto group. Reversing the positions of alanine and
proline restores the selectivity to some extent
(Table 4).
The aromatic function in L-ZGP provided by the
N-blocking benzyloxycarbonyl group seems not to be
a prerequisite for chiral recognition as a low enan-
tioselectivity has been observed for some solutes us-
ing N-cyclopentylpropionyl-L-proline as chiral selec- Counterion R1 R2 kRb S/R
tor. Thus, when using a DIOL stationary phase, Z-glycyl-L-proline (L-ZGP) HNCH2CO H 4.4 1.24
chiral recognition is believed to involve an electro- Z-L-proline } H 3.2 1.15
static attraction to the acidic function in L-proline and Z-L-proline methylestera } CH3 1.9 1.00
Z-L-alanyl-L-proline HNCH(CH3)CO H 2.6 1.00
hydrogen bonding to the carbonyl group.
Z-L-prolyl-L-alanine 6.4 1.10
Several selectors structurally related to L-ZGP have
been studied in systems with PGC as support. In Solid phase: LiChrosorb DIOL. Mobile phase: 2.5 mM counterion
Table 6, the results from one of these studies is and 1.0 mM triethylamine in dichloromethane (80 p.p.m. H2O).
a
Triethylamine 4.0;10\5 M in mobile phase. Reprinted from
presented. L-ZGP, N-benzyloxycarbonyl-glycyl- Pettersson C and Josefsson M (1986) Chromatographia 21: 321
glycyl-L-proline (L-ZGGP) and 1-[(2S)-3-mercapto-2- with permission from the authors.
methylpropionyl)-L-proline (captopril) may be re- b
Solute: R- and S-alprenol.
used as chiral selector in liquid chromatography. Sev- enantiomers when (!)-DIKGA is present in a polar
eral compounds including -blocking agents, local mobile and PGC is used as support. An example is
anaesthetics and neuroleptics are resolved into their given in Figure 8. The enantiomers of p-hydro-
Table 5 Examples of different solutes separated in systems with L-ZGP as chiral counterion
Solute Structure k1
Ephedrine 7.06a 1.34
N-Isopropyl- 1.63a 1.57

2-hydroxy-
2-(spirocyclo-
pentane-
1,1-inden)-
3-ylethylamine
8-Hydroxy-2- 2.24a (3.5c) 1.12 (2.04c)

(di-n-propyl-
amino) tetraline
Metoprolol 18b (2.4c) 1.09 (1.25c)
Alprenolol 22b 1.17
Sotalol 9.8b 1.13
1-Phenyl- 3.8c 1.15

2-aminopropane
Promethazine 1.0c 1.15

Table 5 Continued
Solute Structure k1
Mepivacaine 9.6d 1.35
Prilocaine 0.91c 1.12
a
Solid phase: Nucleosil CN. Mobile phase: 2.5 mM L-ZGP and 0.25 mM triethylamine in dichloromethane (80 ppm water). Results from
Pettersson C, Karlsson A and Gioeli C (1987) Journal of Chromatography 407: 217I229 with permission from Elsevier Science.
b
Solid phase: PGC. Mobile phase: 5 mM L-ZGP and 4.5 mM NaOH in methanol. Results from Huynh NG, Karlsson A and Pettersson
C (1995) Journal of Chromatography 705: 275I287 with permission from Elsevier Science.
c
Solid phase: PGC. Mobile phase: 10 mM L-ZGP in dichloromethane (80 ppm water). Results from Karlsson A and Pettersson C (1991)
Journal of Chromatography 543: 287I297 with permission from Elsevier Science and Karlsson A, Luthman K, Pettersson A and Hacksell
U (1993) Acta Chemica Scandinavica 47: 469I481.
d
Solid phase: PGC. Mobile phase: 5 mM L-ZGP in dichloromethane : hexane (1 : 1). Results from Karlsson A and Pettersson C (1992)
Chirality 4: 323I332, reprinted by permission WileyILiss, Inc., a subsidiary of John Wiley and Sons, Inc.
xyephedrine are separated within 4 min. sulfonic acid, be as low as possible in order to pro-
mote the chiral separations. The equilibration time
needed to obtain a stable system with dry solvent is,
however, very long and therefore it is convenient to
work with higher water contents although not a
water-saturated mobile phase. The water content
does not inSuence the enantioselectivity when L-ZGP
is used as counterion. Longer retention times are,
however, obtained when more water is present in the
solvent. Water-saturated injection solutions can be
General Mobile Phase Considerations injected into the system without inSuence on either
retention or stereoselectivity.
Beside the counterion and its concentration, the com-
As the polarity of the mobile phase controls the
position of the mobile phase will depend on the col-
retention of the solutes, modiRers like alcohols,
umn packing material used and the type of solutes to
acetonitrile or tetrahydrofuran have sometimes to be
analyse. Uncharged or charged modiRers may be
used in order to elute solutes in a reasonable time.
needed in the mobile phase in order to elute the
Several of these modiRers also have an effect on the
solutes in a reasonable time and also to resolve them
enantioselectivities of these systems. Some modiRers
into their enantiomers.
have been studied in a system with (#)-10-camphor-
sulfonate in dichloromethane as mobile phase. (In
Mobile Phase Composition when using DIOL
these cases the mobile phases were prepared from dry
Functionalized Silica as Stationary Phase
dichloromethane.) It was found that the stereoselec-
When DIOL functionalized silica particles are used as tivity was lower with hydrogen donating modiRers
stationary phase, the mobile phase has to be a non- than with hydrogen accepting ones (Table 7). One
polar solvent. The water concentration in the non- hypothesis is that the modiRers may interact with the
polar solvent may then be of vital importance for the hydrogen-bonding groups in the selectors. It should
success of the separation and should, together with be noted that 1-pentanol is preferred as a modiRer
some of the chiral counterions, quinine and camphor- since the peaks obtained are more symmetrical than
Table 6 Counterion structure and enantioselective retention

Solutes:
Counterions:
Figure 7 Separation of (R )- and (S )-metoprolol. Solid phase:

Hypercarb. Mobile phase: 3.4 mM Z-L-Glu-L-Pro and 5.5 mM
NaOH in methanol. Column temperature: 353C. (Reprinted from
Karlsson A and Karlsson O (1997) Enantiomeric separation of
amino alcohols using Z-L-Glu-L-Pro or Z-L-Glu-D-Pro as chiral
counterions and Hypercarb as the solid phase. Chirality 9:
650}655, with permission from Wiley-Liss, Inc., a subsidiary of
John Wiley & Sons, Inc.)
Solute Counterion
(n) phases resemble those used with the DIOL stationary
L-ZGP L-ZGGP Captopril phase. The addition of a nonchiral hydrophobic acid
may, however, be needed in order to elute some of the
k1 k1 k1
hydrophobic acids from the PGC. Naproxen was
1 1.8 1.32 4.9 1.05 9.0 1.07 found to be too strongly retained using a mobile
2 2.2 1.0 3.9 1.09 7.0 1.0 phase consisting of quinine in dichloromethane and
3 1.2 1.06 4.0 1.44 } } hexane. When adding -naphthalenecarboxylic acid
to the mobile phase, naproxen was eluted within
Solid phase: PGC. Mobile phase: 7.2 mM of counterion in di-
a reasonable time and the enantiomers were separ-
chloromethane (80 ppm H2O). Reprinted from Karlsson A and
Pettersson C (1992) Chirality 4: 323 with permission from Wiley- ated (Figure 9). The effect of addition of an acid to
Liss, Inc., a subsidiary of John Wiley and Sons, Inc. the mobile phase is not easy to predict as the added
acid may compete for both the distribution to limited
adsorption sites on the support and also for ion pair
when using, for example, tetrahydrofuran. The im- formation with the chiral selector. Furthermore, if the
proved peak shape is believed to be due to deactiva- counterion added to the polar mobile phase is ex-
tion of strong adsorption sites on the silica surface. pected to remain in an uncharged form, an acid or
Peak shapes may also be improved by adding to the base has to be added in order to obtain a charged
mobile phase a compound with the same protolytic counterion. L-ZGP has a pKa of 7}9 in methanol. The
character as the solute enantiomers. Thus, for the amount of base added (e.g., sodium hydroxide) will
separation of amines, triethylamine is often used to determine the degree of protolysis of the L-ZGP.
improve peak symmetry. Acidic compounds to be However, an excess of base will result in an alkaline
analysed with, for example, quinine, may need addi- solution where the solutes may be uncharged. The
tion of acid to the mobile phase in order to effect added base may thereby inSuence the retention and
elution. The added acid is believed to compete for the stereoselectivity, in other words inSuence both
adsorption to the solid phase and for ion pair forma- k and the ratio kachiral/kchiral (Figure 3).
tion with quinine.
Conclusions
Mobile Phase Composition when using PGC
The addition of a counterion to the mobile phase is
as Stationary Phase
a technique often used to control the retention of
When PGC is used as stationary phase, both nonpolar solutes. If the added counterion is chiral, chiral separ-
and polar mobile phases may be used. The nonpolar ations may be achieved provided that appropriate
Figure 9 Resolution of (R,S)-naproxene. Solid phase: PGC. Mo-

bile phase: 0.35 mM quinine and 0.10 mM -dinaphthalenecar-
bonic acid in dichloromethane : hexane (1 : 1) (Reprinted from
Karlsson A and Pettersson C (1992) Separation of enantiomeric
amines and acids using chiral ion-pair chromatography on porous
graphitic carbon. Chirality 5: 323}332, with permission from
Wiley-Liss, Inc., a subsidiary of John Wiley and Sons, Inc.)
available in both its enatiomeric forms, the retention

order of a pair of enantiomers may be controlled. In
contrast to derivatization with chiral reagents, there
is no need for optically pure counterions or counter-
ions with exactly known purities. An optically impure
selector will, in chiral ion pair chromatography, still
give enantiomerically pure peaks, although eluting
Figure 8 Separation of rac-p-hydroxyephedrine. Solid phase: closer to each other than they would if an optically
Hypercarb. Mobile phase: 20 mM (!)-DIKGA and 15 mM NaOH pure selector were used.
in isopropanol : acetonitrile (9 : 1). Flow rate: 1.0 mL min\1.
(Reprinted from Pettersson C and Gioeli C (1993) Chiral separ- See also: II/Chromatography: Liquid: Ion Pair Liquid
ation of amines using reversed-phased ion-pair chromatography. Chromatography. Chromatography: Thin-Layer
Chirality 5: 241}245, with permission from Wiley-Liss, Inc., a sub-
(Planar): Ion Pair Thin-Layer (Planar) Chromatography.
sidiary of John Wiley & Sons, Inc.)
III/Chiral Separations: Capillary Electrophoresis; Chiral
Derivatization; Cyclodextrins and Other Inclusion Com-
solvents and solid supports are chosen. Using mobile plexation Approaches; Ligand Exchange Chromtography;
phase additives is an easy way to screen for new chiral Liquid Chromatography; Molecular Imprints as Stationary
selectors. Furthermore, if the chiral counterion is Phases; Protein Stationary Phases; Synthetic Multiple In-
teraction (‘Pinkle’) Stationary Phases.
Table 7 Influence of polar components in the mobile phase
Polar solvent Content (%) k1 #/

\
Asf a Further Reading
Pentanol 0.5 20.3 1.08 1.5 Davankov VA, Meyer VR and Rais MA (1990) Vivid model
1 10.8 1.06 1.1 illustrating chiral recognition induced by achiral struc-
5 2.02 1.02 1.1 tures. Chirality 2: 208}210.
Isopropanol 0.5 13.5 1.07 2.6
Huynh NG, Karlsson A and Pettersson C (1995) Enan-
Acetonitrile 1 25.1 1.09 2.2
Tetrahydrofuran 1 11.9 1.10 3.7 tiomeric separation of basic drugs using N-benzyloxycar-
Ethyl acetate 1 29.0 1.10 3.6 bonylglycyl-L-proline as counter ion in methanol. Jour-
nal of Chromatography 705: 275}287.
Sample: Alprenolol (#and!forms). Mobile phase: (#)-10-cam- Josefsson M, Carlsson B and Norlander B (1994) Chiral
phorsulfonate, 2.1 mM in dichloromethane}polar solvent. ion-pair chromatographic separation of two dihyd-
a
Asf"asymmetry factor measured at baseline. Back part of
peak/front part of peak. Results from Pettersson C and Schill ropyridines with camphorsulfonic acids on porous
G (1981) Journal of Chromatography 204: 179I183, with per- graphitic carbon. Journal of Chromatography 684:
mission from Elsevier Science. 23}27.
III / CHIRAL SEPARATIONS / Ligand Exchange Chromatography 2369
Karlsson A and Karlsson O (1997) Enantiomeric separation Petterson C and Heldin E (1994) A practical approach
of amino alcohols using Z-L-Glu-L-Pro or Z-L-Glu-D-Pro to chiral separations by liquid chromatography. In:
as chiral counter ions and Hypercarb as the solid phase. Subramanian G (ed.) Ion-Pair Chromatography in
Chirality 9: 650}655. Enantiomer Separations, pp. 279}310. Weinheim:
Knox JH and Jurand J (1982) Separation of optical isomers VCH.
by zwitterion pair chromatography. Journal of Chroma- Schill G, Wainer IW and Barkan SA (1986) Chiral separ-
tography 234: 222}224. ation of cationic drugs on an 1-acid glycoprotein
Knox JH and Ross P (1997) Carbon-based packing mater- bonded stationary phase (Enantiopac威). Journal of
ials for liquid chromatography. Advances in Chromato- Chromatography 365: 73}78.
graphy, 74}162.
Ligand Exchange Chromatography

V. A. Davankov, Nesmeyanov-Institute of Organo- time its leading positions in developing novel chiral
Element Compounds, Russian Academy of Sciences, chromatographic systems and evaluating mechanisms
Moscow, Russia of chiral recognition and discrimination. This tech-
Copyright ^ 2000 Academic Press nique contributed much to the successful develop-
ment of chiral silica-bonded stationary phases, both
monomeric and polymeric, chiral coatings on col-
umn-packing materials and chiral mobile phase addi-
tives. In this last technique, it transpired that
Introduction the central metal ion does not need to reside in the
Ligand exchange chromatography (LEC) was Rrst stationary phase, but, instead, can be a part of a
introduced by Helfferich in 1961 as a general chro- chiral complex that is doped into the mobile phase to
matographic technique to separate compounds which interact there with the analytes. This widens signiR-
are able to form labile complexes with transition cantly the deRnition of LEC to a chromatographic
metal cations. The basic idea was to immobilize such process in which the formation and breaking of labile
ions as Cu(II) or Ni(II) on a stationary phase, in coordinate bonds to a central metal cation are re-
particular, a cation exchanger with sulfonic or sponsible for the separation of complex forming
carboxylic functional groups, and then let the ions analytes.
form labile coordination compounds with analytes Ligand exchange has been realized in all known
which possess electron-donating functional groups modes of enantioselective chromatography, including
and can enter the coordination sphere of the metal liquid chromatography, thin layer chromatography,
ion, thus acting as ligands. Those analytes that form gas chromatography, capillary electrophoresis and
stronger complexes with the central ion, are retained countercurrent liquid chromatography. In gas
longer on such a metal ion-incorporating stationary chromatography the principle of ligand exchange is
phase. more commonly known under the name ‘complexa-
Starting with the idea that, in the densely packed tion chromatography’.
coordination sphere, ligands enter multipoint interac-
tions with each other and should therefore mu- Theoretical Aspects of Chiral
tually recognize the spatial conRguration of neigh-
bours, Davankov and Rogozhin (1968}71) syn-
Discrimination in LEC
thesized chiral complex-forming resins and further Being built of protons, neutrons and electrons, which
developed LEC into a powerful chiral chromato- basically are chiral elementary species, all atoms
graphic method. This technique, for the Rrst time in possess inherent chirality. Therefore, two enan-
liquid chromatography, resulted in a complete and tiomeric molecules resulting from a mirror-symmetric
reliable resolution of a racemate into constituent en- arrangement of a certain ensemble of atoms in
antiomers. Typical analytes for enantioseletive LEC space, can differ in their thermodynamic stability.
are members of such important classes of chiral com- However, this difference may only amount to about
pounds as amino acids, hydroxy acids and amino 10\14 J mol\1, which practically makes the enantio-
alcohols. Having soon become one of the most exten- mers energetically identical and indistinguishable in
sively investigated methods for the direct resolution any achiral environment. Using any chromatographic
of enantiomers, LEC maintained for a long period of technique, two enantiomers can be recognized and
2370 III / CHIRAL SEPARATIONS / Ligand Exchange Chromatography
discriminated by the action of a chiral selector only. raising the column temperature to about 503C, which
The latter has to be involved in interaction with the speeds up the ligand exchange. Ions of Cr(III),
enantiomers, resulting in the formation of adducts Co(III), Pd(II) and the like, form kinetically inert,
which, being diastereomeric species, may differ in stable complexes unsuitable for LEC. However, the
their stability. These adducts in LEC are ternary com- second coordination sphere of these complexes,
plexes composed of the molecule of the chiral selec- which actually is their solvation shell, is well organ-
tor, the complexing metal ion, and the enantiomer to ized, too, and it does exchange its ligands readily, so
be recognized. Thermodynamic enantioselectivity of that an enantioselective chromatography process can
the system, i.e. the difference in stability of the above also be based on outer coordination sphere ligand
two diastereomeric ternary complexes which incor- exchange.
porate (R) or (S) enantiomers of the analyte, repres- Zn(II) would usually coordinate four heteroatoms
ents the sole source of chiral discrimination of the with lone electron pairs in the form of a tetrahedron.
enantiomers in LEC. For an analytical scale resolu- The coordination number of the Ni(II) ion is six and
tion of enantiomers, the enantioselectivity does not its ligands are usually arranged in an octahedron.
need to be high, since, in principle, enantioselectivity Copper binds four ligands in its main coordination
of G0"25 J mol\1 would correspond to a separ- square, but offers two additional, more remote
ation factor "1.01, which is fully sufRcient for two axial positions, so that a complex with the coord-
peaks to be resolved in a gas chromatographic capil- ination number six adopts a distorted octahedral
lary column generating about 105 theoretical plates. structure.
For a preparative scale liquid chromatography, one It is worth noting, that the optical purity of a com-
would prefer to deal with a separation factor of at plex-forming chiral selector employed in an LEC sys-
least "1.5, which would correspond to enan- tem does not need to be necessarily 100%. It is
tioselectivity of about 1 kJ mol\1. Ligand exchange a distinguishing feature of all direct chromatographic
often generates a discrimination effect of this order of separations of enantiomers that even a selector con-
magnitude. With the G0 increasing further, the taminated by its enantiomeric form can provide
column selectivity, , rises exponentially, according a complete resolution of racemic analytes. Enan-
to the general relation G0"RT ln . In LEC sys- tiomeric impurities in the chiral selector do not di-
tems with polydentate ligands, separation selectivities minish column efRciency but merely result in the two
of up to "10 are known. sharp enantiomeric peaks of the analyte approaching
As a matter of principle, a chiral selector has to each other and completely coalescing into one sharp
enter into at least a three-point interaction with the peak when the enantiomeric excess (e.e.) of the selec-
enantiomers, in order to be able to recognize the tor falls to zero. This phenomenon (which is not an
difference in the spatial structure of the latter. This issue, at all, with proteins or natural carbohydrates
requirement is easily met with tridentate ligands as the chiral selectors, since these selectors are avail-
forming ternary complexes with metal ions which able in only one enantiomeric form) has been repeat-
have a coordination number of six. Among practic- edly proven experimentally in chiral exchanging sys-
ally important classes of organic compounds, how- tems involving amino acids of known e.e. as the
ever, are molecules with two electron-donating selector.
heteroatoms, N and/or O, situated in a chain of
carbon atoms in positions 1,2 or 1,3, e.g. in 1,2- and Chiral-Bonded Ligand Exchanging
1,3-diamines, 1,2- and 1,3-amino alcohols, - and -
amino acids and - and -hydroxy acids. When form-
Stationary Phases
ing complexes with metal ions, these compounds The Rrst chiral ligand exchanging polystyrene-type
usually function as bidentate ligands with both het- stationary phases were synthesized and tested by
eroatoms coordinated to the same metal ion in the Rogozhin and Davankov as early as 1968, while
form of a Rve- or six-membered chelate ring. analogous silica-bonded phases emerged a decade
It is very important that formation and dissociation later. Even nowadays, cross-linked polystyrene resins
of the bonds in the coordination sphere of the metal remain popular in preparative chromatography,
ion be fast. Otherwise, a slow rate of ligand exchange mainly due to their excellent chemical durability and
would degrade the efRciency of the chromatographic high loading capacity. To prepare such chiral pack-
system. Therefore, only metals forming kinetically ings, spherical particles of styrene-divinylbenzene
labile complexes can be employed in LEC, preferably copolymers are usually subjected to chloromethyla-
Cu(II), Ni(II) and Zn(II). Even with the above metal tion and the active chlorine atoms are then replaced
ions, it has been repeatedly reported that the efRcien- by amino groups of chiral natural -L-amino acids,
cy of ligand exchanging systems can be enhanced by in particular (S)-proline and (S)-hydroxyproline.
(According to modern stereochemical nomenclature, tate amino acid-type ligands, like histidine or allo-
L- and D--amino acids belong to R and S conRgura- hydroxyproline, the opposite elution order of the
tion series, respectively, with a few exceptions, how- enantiomers is observed, thus giving values smaller
ever.) The structure of a chiral selector immobilized than 1. Many other types of chiral selectors have been
in this way as well as its complex with a Cu(II) ion similarly immobilized on chloromethylated cross-
and an amino acid analyte, both of which bind to the linked polystyrene beads, with the result that cyclic
selector during the process of LEC, can be represent- amino acids as well as benzyl-substituted chiral pro-
ed by the following scheme: pan-1,2,-diamine display the highest selectivity and
largest application range. Typical examples for re-
solving racemates of common -amino acids on these
phases are presented in Table 1.
Two additional polymeric matrixes, cross linked
polyacrylamide and poly(glycidylmethacrylate), have
also been shown to be useful for immobilizing chiral
amino acid selectors. The mode of functionalizing of
these resins, when they incorporate (S)-phenylalanine
and (S)-proline as typical chiral selectors, can be rep-
resented by the following structures:
Here, the most important features of the ter- The Rrst resin demonstrates higher afRnity to (R)-
nary mixed-ligand bis(amino acidato)copper com- isomers of all amino acids, without any exception,
plex are: resolving all racemates with a selectivity of at least
"1.3.
E the formation of two Rve-membered chelate rings; All the above polymeric ligand exchangers have
E the occupation of the four positions of the Cu(II) been used successfully for both analytical and prep-
ion main coordination square by four electron- arative resolution, e.g. for obtaining highly radioac-
donating atoms of the ligands; tive tritium-labelled optically active amino acids. In
E the trans-arrangement of the two negatively order to remove trace copper ions from enantiomers
charged carboxyl groups and the bi-substituted ni- resolved in preparative experiments, the fractions of
trogen atoms, respectively, which minimizes both interest can be gravity Rltered through a small addi-
the electrostatic and sterical repulsion of the tional column packed with silica, any chelating resin
ligands in the complex. or even the same chiral resin, with a blue band of
Cu(II) gradually forming at the top of the subsidiary
The resins incorporating (S)-proline and (S)-hy- column.
droxyproline, when loaded with Cu(II) ions and Silica-bonded chiral ligand exchangers were ex-
eluted with ammonia solution, are usually found to pected to demonstrate in analytical-scale separations
retain -amino acids of the opposite, (R)-conRgura- still better column efRciency than polymer gels. Three
tion more effectively. In this case, the enantioselectiv- types of bonded phases were suggested independently
ity of the system, when expressed as the ratio of and simultaneously in 1979 by the groups of
retention factors of the (R)- and (S)-enantiomers of Foucault, Davankov and Guebitz, based on three
the analyte, "kR/kS, is larger than 1. With triden- different ways of activating the initial macroporous
Table 1 Enantioselectivity, "kD/kL, of resolution of racemic amino acids on the Cu(II) forms of polystyrene resins which incorporate
residues of L-proline, L-hydroxyproline, L-allo-hydroxyproline, L-azetidine carboxylic acid and N1-benzyl-(R)-propane-1,2-diamine.
Mobile phases: aqueous ammonia solutions. Ambient temperature
Racemic amino acid RPro RHyp RaHyp RAzCA RBzpn
Alanine 1.08 1.04 1.04 1.06 0.70

Aminobutyric acid 1.17 1.22 1.18 1.29 0.49
Norvaline 1.34 1.65 1.42 1.24 0.48
Norleucine 1.54 2.20 1.46 1.40 0.49
Valine 1.29 1.61 1.58 1.76 0.31
Isovaline } 1.25 } } }
Leucine 1.27 1.70 1.54 1.24 0.49
Isoleucine 1.50 1.89 1.74 1.68 0.62
Serine 1.09 1.29 1.24 2.15 0.62
Threonine 1.38 1.52 1.48 0.78 0.44
Allothreonine 1.55 1.45 } } }
Homoserine } 1.25 } } }
Methionine 1.04 1.22 1.52 1.29 0.60
Asparagine 1.18 1.17 1.20 1.44 0.70
Glutamine 1.20 1.50 1.40 1.25 }
Phenylglycine 1.67 2.22 1.78 1.38 0.49
Phenylalanine 1.61 2.89 3.10 1.86 0.52
-Phenyl -alanine } 1.07 } } 0.38
Tyrosine 2.46 2.23 2.36 1.78 }
Phenylserine } 1.82 } } }
Proline 4.05 3.95 1.84 2.48 0.47
Hydroxyproline 3.85 3.17 1.63 2.25 }
Allo-hydroxyproline 0.43 0.61 1.48 1.46 }
Azetidine carboxylic acid } 2.25 } } }
Ornithine 1.0 1.0 1.20 1.0 1.0
Lysine 1.10 1.22 1.33 1.06 1.0
Histidine 0.37 0.36 1.32 0.56 0.85
Tryptophan 1.40 1.77 1.10 1.13 }
Aspartic acid 0.91 1.0 0.81 0.88 1.0
Glutamic acid 0.62 0.82 0.69 0.77 }
microparticulate silica gel, namely, by Rrst grafting acids, dipeptides, some -amino acids, etc. Amino
silica with aminopropyl, chloroalkyl, and 3- alcohols do not form stable chelates with Cu(II) ions,
glycidoxypropyl groups, respectively: but are easily resolved after conversion into Schiff
bases or, better still, after reaction with bromoacetic
acid, which converts the amino group into a chelating
glycine moiety.
The silica-bonded phases, however, still require
elevated temperatures (usually 503C) to generate suf-
Rcient column plate numbers. In addition only neu-
tral or weakly acidic aqueous}organic eluents, in
particular, in the pH range between 6.0 and 4.0 can
be applied. Therefore, a much more hydrolytically
Whereas the aminopropyl spacer binds a chiral amino stable packing was introduced by multiple-point
acid-type selector via the carboxyl group, the other binding of chloromethylated polystyrene chains to
two functions easily react with the amino group of the silica matrix, followed by substituting the active
the selector. The epoxy-activated and L-proline incor- chlorine atoms of the polymer for chiral selectors.
porating silica was the Rrst chiral-bonded stationary This type of polymer-bonded phase can safely be used
phase on the market (Chiral ProCu, Serva, Heidel- at elevated temperatures, and display high resolving
berg, Germany). Silica-bonded ligand exchangers power towards numerous racemates, as exempliRed
have been used successfully to resolve racemates of in Figure 1. In LEC, the elevated temperature does
numerous amino acids, hydroxy acids, N-dansyl- not noticeably affect the enantioselectivity of the
and N-alkyl-amino acids, -triSuoromethyl--amino column.
Chiralpak WH and WM (Daicel, Japan), TSK gel

Enantio L1 (Toso, Japan), MCI gel CRS 10W
(Japan), Chirosolve L-proline, Chirosolve L-valine
(JPS Chimie, Switzerland). Any scientiRc evaluation
of results obtained by using these phases is, however,
complicated, since the exact structure of immobilized
chiral ligands on many commercially available chiral
stationary phases is not speciRed by the manufacturers.
Chiral Dynamically Coated Stationary

Phases
Ligand exchanging columns of an exceptional efR-
ciency and enantioselectivity can be easily prepared
from commercially available reversed-phase columns
by dynamically coating the column with a hydropho-
bic chiral selector. According to the technique sugges-
ted by Davankov and Unger in 1980, percolating
a solution of N-alkyl-L-hydroxyproline in aqueous
methanol through the column results in depositing
the chiral selector on the bonded alkyl stationary
phase. After a subsequent saturation of the retained
ligand with copper ions, the column is ready for
racemate resolutions in water-rich eluents which do
not desorb the chiral selector. Depending on the
length of the N-alkyl group (heptyl, decyl or hexa-
decyl) and conditions of coating, the amount of the
chiral selector deposited can be easily regulated over
a very broad range. Such columns (also commercially
available from Regis Technologies, USA, as ‘Davan-
Figure 1 Enantiomeric resolution of a mixture of eight DL-amino
acids with a chiral stationary phase, polystyrene grafted on silica kov columns’) display a high resolving power toward
gel and substituted with L-Hypro. Conditions: column, 250;4 mm numerous amino acids (Figure 2) and their deriva-
i.d, dp 5 m: eluent, 0.5 mM copper acetate, 10 mM ammonium tives, racemic glycyl-di- and tripeptides, 3-amino-
acetate, pH 4.5, acetonitrile 30%; flow rate, 0.7 mL min\1; tem- -caprolactam, as well as norephedrine and its
perature, 753C; UV 254 nm. (Reprinted with permission from
analogous amino alcohols. Interestingly, the N-alkyl-
Davankov VA (1989) Chapter 15, Figure 5, p. 464. In: Krstuloviz
L-hydroxyproline-coated columns, under severe
AM (ed.) Chiral Separations by HPLC. Applications to Pharma-
ceutical Compounds, pp. 446I475. Chichester, England: Ellis overloading conditions, show a peculiar distortion of
Horwood.) elution proRles of the enantiomers in that the Rrst
peak has no tail whereas the second peak shows
When dealing with complicated analyte matrixes, a very sharp front, which enormously increased the
it is often useful to combine in sequence a chiral column loadability, implying high preparative poten-
column with a conventional achiral one, in order to tial of the system.
supplement the resolution of enantiomeric pairs in Coating reversed-phase columns with N,N-dioctyl-
the chiral column with the selectivity towards differ- L-alanine has proved especially efRcient in resolving
ent analytes of the second column. Thus, chiral ligand racemic -hydroxy acids. Chiral coatings prepared
exchanging bonded phases are easily compatible with with N?-decyl-L-histidine, Nim-decyl-L-histidine, N2-
achiral reversed phase columns, since both columns octyl-(L)-phenylalaninamide, N,S-dioctyl-N-methyl-
can be operated in a gradient mode with aque- (D)-pencillamine, hydrophobic Schiff bases of chiral
ous}organic eluents which contain about 0.1 mM amino alcohols and N-dodecyl-N-carboxymethyl-
copper acetate and about 0.25 M ammonium acetate. (1S,2R)-norephedrine show high efRciency and selec-
Several silica-bonded and polymeric chiral ligand tivity. Reversed-phases coated with N,S-dioctyl-(D)-
exchangers are commercially available, e.g. Chiral-Si penicillamine and (R,R)-tartaric acid mono-(R)-1-(-
100 L-ProCu, Chiral-Si 100 L-ValCu, Chiral-Si 100 naphthyl)ethylamide are commercially available un-
L-HyProCu (Serva, Heidelberg, Germany), Nucleosil der the name Sumichiral OA-5000 and 6000 (Sumica
Chiral-1 (Macherey-Nagel, Dueren, Germany), Chemical Analysis Service, Japan).
Figure 2 Enantiomeric resolution of a mixture of seven DL-amino acids with a chiral coating of N-decyl-L-Hypro on reverse phase
material. Conditions: column, 100;4.2 mm, LiChrosorb RP 18, dp 5 m; eluent, methanol}water (15 : 85), 0.1 mM Cu(II) acetate, pH
5.0; flow rate 2 mL min\1; temperature 203C, UV 254 nm; elution sequence; 1, L-Ala; 2, D-Ala; 3, L-Val; 4, L-Arg; 5, D-Arg; 6, L-Leu; 7,
L-Nleu; 8, D-Val; 9, L-Phe; 10, D-Leu; 11, L-Trp; 12, D-Nleu; 13, D-Phe; 14, D-Trp. (From Davankov VA, Bochkov AS, Kurganov AA,
Roumeliotis P and Unger KK (1980) Separation of unmodified -amino acid enantiomers by reverse phase HPLC. Chromatographia 13:
677}685, Figure 2, p. 680.)
Whereas long N-alkyl substituents of the above should be ascribed to ligand exchange in the outer,
chiral selectors are best suitable to bring about the solvation shell of this complex.
intercalation of the latter between the surface alkyl Finally, polymeric chains of polyacrylamide graf-
groups of the reversed phase support, porous graphite ted with L-proline can also be permanently adsorbed
offers a Sat and rigid surface for the chiral selector to onto the surface of silica gel, to give a chiral ligand
stick to. In this case, polyaromatic substituents show exchanger which resolves amino acids in polar media.
better retention and resolving power of the coated The ease of preparation of chiral-coated ligand
chiral selector, as is the case with N-naphthylmethyl- exchanging phases, their exceptional efRciency, dura-
L-proline or N-(2-naphthalenesulfonyl)phenylalanine bility, wide application range and the opportunity of
(see Figure 3). regulating the elution order of the enantiomers of
Similarly, hydroxylated macroporous silica gel can interest by simply changing the conRguration of the
be coated with hydrophilic chiral selectors such as selector applied, make the technique increasingly
silver(I)-d-camphor-10-sulfonate or (#)-cobalt(III)- popular.
tris-ethylenediamine trichloride. In non-aqueous
media, these phases easily resolve racemates of some Chiral Ligand Exchanging Mobile
chiral diene and cyclopentadienylrhodium(I)-nor-
bornadiene complexes. Since in the latter case
Phases
ethylenediamine ligands of the inert cobalt complex In the technique described in the previous section, the
are not exchanged for any new ligands, the separation chiral selector permanently resides on the surface of
Figure 3 Enantiomeric separation of racemic amino acids and hydroxy acids with a chiral coating of N-(2-naphthalenesulfonyl)-L-
phenylalanine on porous graphite. Conditions: column PGC 94F (5 m, 50;4.6 mm), eluent 2.0 mM Cu(CH3COO)2, pH 5.6, flow rate
0.5 mL min\1, UV 254 nm, temperature 303C. (Reprinted with permission from Knox JH and Wan Q-H (1995) Chromatographia 40:
9}14.)
the column packing. The coating is deposited prior to tector cell with the chiral eluent in the form of ternary
the chromatographic experiments and no eluents complexes. These complexes are diastereomeric and
should be used which cause a leakage of the selector. can signiRcantly differ in molar extinction. Indeed,
A different approach, which was independently sug- a case of enantioselective Suorescence quenching of
gested in 1979 by the groups of Karger and Lindner dansylamino acids by a chiral Cu(II) complex of L-
and Hare and Gil-Av, consists in using chiral selectors phenylalanylamide in aqueous solutions has been re-
which predominantly reside in the mobile phase. In ported, thus requiring two calibration curves for
this case, the selector complex has to be continuously a quantitative determination of the two enantiomers.
introduced into the column. Analytes to be resolved In combination with Cu(II), Zn(II), Ni(II) and some
in such a chiral eluant form mixed-ligand complexes other transition metal ions, many chiral selectors
with the selector, which are diastereomeric in their have been shown to function successfully in reversed
structure, and, therefore, may be differently retained phase systems as chiral dopants. They mainly belong
through the interaction with the column packing. to the following classes of chelating compounds:
A great advantage of this technique is that many
different chiral selectors can be easily tested with E unsubstituted L--amino acids (proline,
respect to the analyte to be resolved. Another advant- phenylalanine, isoleucine, histidine, etc.);
age is that smaller ligands are usually applied as the E N-alky-L-amino acids (N-benzyl-proline, N,N-di-
eluent dopants, which results in enhanced rates of propyl-alanine, N,N-dimethyl-valine);
ligand exchange and better efRciency of the columns. E L--amino acid amides (valine amide, phenyl-
A disadvantage is the loss of the chiral selector and alanine amide, prolyloctylamide, aspartyl-
problems of separating the selector from the enantio- alkylamide);
mers resolved in the case of preparative experiments. E esters of L--amino acids (histidine methyl ester);
It should be emphasized also that the two enantio- E peptides (prolyl-valine, aspartylphenylalanine
mers separated in the achiral column enter the de- methyl ester or aspartame);
E chiral diamine (N,N,N,N-tetramethyl-(R)-pro- mers. This approach is useful in analysing enan-

panediamine-1,2-( L )-alkyl-4-octyl-diethylene- tiomeric impurities in chiral compounds, where it is
triamine); always advisable to have the trace component eluting
E mandelic acid; before the major enantiomer.
E derivatives of (R,R)-tartaric acid (tartaric acid Finally, it is worth mentioning that the chiral mo-
mono-n-octylamide); bile phase mode of LEC is very convenient to prove
E nucleotides (adenosine diphosphate, -nicotinam- the reciprocity relations in chiral recognition, where
ide, adenine dinucleotide and, in particular, Savine the selector is exchanged for the analyte and vice
adenine dinucleotide). versa. The enantioselectivity of the system should
remain the same, unless some additional factors inter-
Mainly racemic -amino acids and their numerous fere with the essential interactions within the dia-
derivatives, -hydroxy acids and amino alcohols are stereomeric sorption complexes.
successfully resolved. Less common, but nevertheless
successful, are resolution of substituted pterins,
phenylhydantoins and ornithine analogues. Mechanisms of Chiral Discrimination
Depending on the hydrophobicity of the analyte in LEC
and chiral selector applied, the content of the organic
modiRer in the chiral mobile phase has to be adjusted Of many enantioselective chromatographic systems
in such a manner that the retention of enantiomers developed thus far, ligand exchanging systems are the
remains in the desired time window. In addition to most examined. Understanding of the mechanism of
this parameter, retention (and enantioselectivity) in chiral discrimination is greatly facilitated here by the
LEC would slightly decrease with the rise in the ionic fact that, of the three interactions needed for the
strength of the eluent and temperature of the column. chiral selector to recognize the solute enantiomers,
Conversely, rising pH and concentration of the chiral two are well deRned as the trans-coordination bonds
additive in the mobile phase cause a clear increase in of pairs of carboxylic and amino groups in the main
retention and a less marked increase in enantiomeric coordination square of the Cu(II) ion. The third inter-
resolution. action should be looked for in the out-of-plane steri-
Under normal phase conditions, with bare silica gel cal interactions or coordination in remote axial posi-
as the packing, copper complexes of N,N,N,N- tions of the coordination sphere. Thus with L-proline
tetramethyl-(R)-propane-1,2-diamine can be used to or L-hydroxyproline incorporating polystyrene-type
resolve amino acids and mandelic acid. Bis(L- resins, retention of L-isomers of amino acids is dimin-
prolinato)copper is efRcient even in combination with ished by sterical interactions with the water molecule
sulfonated polystyrene cation exchangers, which ad- coordinated in one of the axial positions. On the
ditionally demonstrates the versatility of the chiral other hand, the binding of D-isomers is facilitated by
mobile phase approach. favourable hydrophobic interaction between the sol-
The high efRciency of the method is exempliRed by ute -substituent and the polystyrene matrix, as
Figure 4. shown below:
It is easy to imagine that, by using a mobile phase Tridentate amino acids, L-His, L-Asp, L-Glu, L-
with a chiral selector of the opposite conRguration, aHyp, manage to replace the water molecule from the
we reverse the elution order of the resolved enantio- axial position and form one additional coordination
bond, which logically explains the inverted elution shown above for the Rrst resin. Conversely, with
order of these solutes: an L-phenylalanine-type chiral selector in a poly-
With tridentate L-allo-hydroxyproline as the chiral acrylamide matrix, the balance of steric interactions
selector, both bifunctional and trifunctional solutes results in a reduction in the retention of L-amino acids
can form two coordinate bonds only and elute in the (see above).
order L before D without exception: The decisive role of interactions with the
achiral sorbent matrix in the enantiomer recogni-
tion was discovered by Davankov in chiral-coated
and chiral mobile phase reversed phase systems.
Here, only one of two solute enantiomers can real-
ize favourable hydrophobic interactions with the
silica immobilized alkyl groups, thus forming
the stronger retained diastereomeric sorption
complex.
With the L-Pro type selectors immobilized on

polyglycidylmethacrylate, polyacrylamide, and poly- The mechanism of chiral discrimination is less clear
vinylpyridine matrixes, which occupy three coord- with selectors and/or analytes other than -amino or
ination positions of Cu(II), the steric repulsion -hydroxy acids, and metal ions other than Cu(II),
interactions diminish retention of the D-amino acid since the coordination mode of ligands in these com-
isomers as compared to L-isomers, as this was plexes is less well known.
Other Chromatographic and Related

Techniques of Resolving Enantiomers
in Ligand Exchanging Systems
In gas chromatography, ligand exchange has acquired
great importance since Schurig introduced metal
complexes of chiral terpeneketonate (acylated cam-
phor, carvone, pulegone, menthone) as additives to
stationary phase in capillary GC in 1977. Ni(II),
Mn(II), Cu(II), Zn(II), and Rh(I) proved especially
useful in this ‘complexation’ chromatography. The
resolution mechanism is ascribed in terms of the in-
teraction between the above sterically or electroni-
cally unsaturated metal chelates with volatile ligands
which offer lone electron pairs to form a labile mixed-
ligand sorption complex. Enantioselectivity of this
process is small, but, due to the extremely high plate
numbers attainable with capillary GC columns, ex-
cellent resolution is achieved for many classes of
alkenes, alcohols, ethers, oxirane, sulfur-containing
compounds, aziridine, etc. It is important that even
compounds containing one single electron-donating
atom (O, N, S) are successfully resolved by GC, con-
trary to LC, where the presence of two donors to
form a chelate with the central metal ion is highly
desirable.
Similar bonded chiral selectors, e.g. polysiloxane-
anchored Ni(II)-bis-[(3-heptaSuorobutanoyl)-(1R)-
camphorate], Chirasil-Nickel, can also be used in the
supercritical Suid chromatography (SFC) mode. Al-
though SFC cannot compete with GC with regard to
efRciency, a signiRcant increase in selectivity, due to
a substantially lower temperature of the column, can
compensate for the loss in plate number. Because of
the high solvation strength of supercritical carbon
dioxide, a number of low-volatility racemic alcohols,
diols and esters have been resolved by ligand ex-
change SFC at temperatures as low as 40}703C.
In capillary electrophoresis (CE) and other elec-
tromigration techniques, chiral ligand exchange can
be involved in several ways. Cu(L-His)2 has just been
added to the ammonium acetate electrolyte to induce
chiral resolution of dansylated amino acids. Analyte
enantiomers which are strongly involved in the
formation of positively charged mixed complexes
with Cu(L-His) move toward the cathode at a higher
rate than the uncomplexed ligands. Since the enan-
tioselectivity of the ligand exchange reaction in the
Figure 4 Enantiomeric separation of R,S-hydroxy acids with bulk solution is small, the resolution of enantiomers
a chiral mobile phase. Conditions: 8 mM L-PheNH2, 4 mM is also small. It is only slightly better with copper-
Cu(CH3COO)2, pH 6.0, ambient temperature, column Spherisorb aspartame as chiral additive. A signiRcant improve-
3 ODS-2 (3 m, 150;4.6 mm), flow rate 1 mL min\1, postcolumn
ment in selectivity can be obtained by adding
derivatization with Fe(CIO4)3, UV 436 nm. (Reprinted with per-
mission from Galaverna G, Panto F, Dossena A, Marchelli R and tetradecyl sulfate, as a micelle-forming surfactant.
Bigi F (1995) Chirality 7: 331}336. Wiley-Liss, Inc., a subsidiary of Obviously, the enantioselectivity increases through
John Wiley & Sons, Inc.) the interaction of the diastereomeric ternary complexes
nique, ligand exchange opened enormous possibilities

for chiral thin-layer chromatography. Macherey-
Nagel in cooperation with Degussa (Germany) de-
veloped reversed-phase plates coated with copper
complexes of N-(2-hydroxydodecyl)-L-hydroxypro-
line in the manner described earlier by Davankov for
chiral coating of RP-columns. The Chiralplate威 has
proved to be extremely versatile in the resolution of
racemic -amino acids, their N-methyl-, N-formyl-,
-alkyl-, and halogenated amino acids, dipeptides,
-hydroxy acids, thiazolidine derivatives, anomers of
several nucleobases, etc. Copper complexes of N,N-
diallky amino acids have also been tested as chiral
coatings of plates, but so far have not found broad
application.
Finally, enantiomeric separation of phenylalanine
and lactic acid by diffusion through a liquid mem-
brane was studied, using a solution of N-decyl-L-
hydroxyproline copper complexes in hexanol}decane
mixtures as the membrane. The rate of migration of
the D-Phe was found to be about 2.4 times faster as
compared to those of the L-isomer. Though the pro-
ductivity potential of the technique is high, the above
value of the kinetic enantioselectivity is still insufR-
cient for practical use.
Applications
Two principal advantages of chiral LEC should be
mentioned with respect to practical application of the
method. Firstly, the detection of compounds which
do not absorb in the near UV region is greatly facilit-
Figure 5 Electrophoregram of the enantiomer separation of
ated by the fact that they partially elute from the LEC
DL-Phe, DL-Trp and DL-MeTrp with and without SDS. Conditions: column in the form of metal complexes. Even having
(a) 80 mM L-Hypro, 40 mM Cu(II) sulfate, pH 4.0; (b) 50 mM no strong chromophore in the molecule, they strongly
L-Hypro, 25 mM Cu(II) sulfate, 15 mM SDS, 3 M urea, pH 4.0. absorb at 254 nm in the complexed form. When com-
Capillary: fused silica, 75 cm;75 m i.d. (effective length, 66 cm) plexed to Fe(III), hydroxy acids can be selectively
U"27 kV, UV 208 nm, ambient temperature. (Reproduced with
permission from Schmid MG and Guebitz G (1996) Enantiomer 1:
detected at 420 nm. The second great advantage of
23}27. Gordon and Breach Publishers.) ligand exchange is that enantioselectivity of the separ-
ation is largely unaffected by the temperature of the
column and the presence of buffer salts and organic
with the surface of the pseudo-stationary micelle modiRer in the eluent. This makes the method easily
phase. Good resolution for dansyl (DNS) amino acids compatible with numerous other separation tech-
has been obtained with mixed-micelle solutions con- niques, including multidimensional and multicolumn
taining sodium dodecyl sulfate (SDS) and copper chromatography.
complexes of N,N-didecyl-L-alanine. Figure 5 shows In general, the efRciency and easiness of the direct
an example of CE resolution of underivatized racemic resolution of amino acids and hydroxy acids, without
amino acids with Cu(L-Hyp)2 in the support electro- any prior derivatization of the analyte, make LEC
lyte, which is facilitated by SDS micelles. Note the a technique of choice for serial determinations of
inversion of the elution order of the enantiomers and enantiomeric compositions. This is the case, for
solutes on transition from the homogeneous support example in fossil dating where, depending on the
electrolyte solution to the pseudo-heterogeneous rate of racemization of the amino acid under invest-
micellar system. igation, the dating can be performed in the time
In addition to paper chromatography which repre- range from several hundred years to one million
sents the oldest chiral planar chromatography tech- years. Another application area of this kind is the
determination of D-amino acids in food products ligand exchanging enantioselective column liquid
caused by partial racemization through heat or chromatography.
microwave treatment, and to fermentation processes.
Occurrence of D-amino acids in biological Suids See also: II/Chromatography: Liquid: Chiral Sep-
(e.g. serum, cerebrospinal Suid, urine), bone and tis- arations in Liquid Chromatography: Mechanisms.
sues is also of great importance from a diagnostic II/Chromatography: Thin-Layer (Planar): Ion Pair Thin-
point of view. Rapid enantiomeric analysis tech- Layer (Planar) Chromatography. III/Amino Acids and De-
rivatives: Chiral Separations. Chiral Separations: Cap-
niques are needed when asymmetric synthesis or
illary Electrophoresis; Cellulose and Cellulose Derived
chemical or enzymic racemate resolution processes
Phases; Chiral Derivatization; Cyclodextrins and Other
are being developed. Examination of the enan- Inclusion Complexation Approaches; Ion-Pair Chromatog-
tiomeric composition of hydroxy acids, in particular raphy; Liquid Chromatography; Molecular Imprints as Sta-
malic acid, helps to investigate adulteration of apple tionary Phases; Protein Stationary Phases; Synthetic
juice and other soft drinks. Rapid preparation of Multiple Interaction (‘Pirkle’) Stationary Phases; Thin-
enantiomers of compounds labeled with highly radio- Layer (Planar) Chromatography.
active atoms, where only dilute solutions can be han-
dled, is another excellent application area for chiral Further Reading
LEC.
Three decades of the existence and intensive devel- Davankov VA (1980) Resolution of racemates by ligand-
opment of enantioselective LEC have shown this exchange chromatography. Advances in Chromatogra-
technique to be an extremely versatile and productive phy 18: 139}195.
general approach to resolving racemates of com- Davankov V (1989) Ligand exchange phases. In: Krstuloviz
AM (ed.) Chiral Separations by HPLC. Applications to
pounds that form complexes with metal ions. Other
Pharamaceutical Compounds, pp. 446}475. Chichester,
types of molecular interactions have been later found England: Ellis Horwood.
to provide alternative successful ways of organizing Davankov VA (1992) Ligand-exchange chromatography of
the formation of labile diastereomeric adducts with chiral compounds. In: Cagniant D (ed.) Complexation
appropriate chiral selectors, e.g. formation of inclu- Chromatography, pp. 197}245. New York: Marcel
sion complexes of analytes with cyclodextrins and Dekker.
cyclic antibiotics or formation of charge-transfer Davankov VA (1994) Chiral selectors with chelating prop-
complexes between electron-donating and electron- erties in liquid chromatography: fundamental reSections
deRcient aromatic groups in Pirkle-type chiral sta- and selective review of recent developments. Journal of
tionary phases. Designing novel chiral selectors Chromatography A 666: 55}76.
which simultaneously offer to the analyte several dif- Davankov VA and Kurganov AA (1983) The role of achiral
sorbent matrix in chiral recognition of amino acid
ferent modes of interactions could probably result in
enantiomers in ligand-exchange chromatography.
developing chiral chromatographic systems with Chromatographia 17: 686}690.
wider application areas or higher enantioselectivity Davankov VA, Kurganov AA and Bochkov AS (1983) Res-
with respect to particular important racemates. Thus olution of racemates by high-performance liquid
Rrst attempts of providing cyclodextrins with a cop- chromatography. Advances in Chromatography 22:
per complexed residue moiety have recently been 71}166.
reported in the literature. Another practically unex- Davankov VA, Navratil JD and Walton HF (1988) Ligand
plored Reld for further development could be ligand Exchange Chromatography. Boca Raton, Florida: CRC
exchange in nonaqueous and even nonpolar media. Press.
With the solvent molecules not competing for the Gil-Av E, Tishbee A and Hare PE (1980) Resolution of
coordination positions of the metal ion, much weaker underivatized amino acids by reversed phase chromatog-
raphy. Journal of the American Chemical Society 102:
electron donors than carboxylic or amino functional
5115}5117.
group of the analyte, should sufRce for the complex Guenther K (1988) Thin-layer chromatographic enan-
formation. Ether, sulRde and carbon}carbon double tiomeric resolution via ligand exchange. Journal of
bonds should thus be suitable for complexation. Chromatography 448: 11}30.
Achievements of complexing gas chromatography Schurig V (1988) Enantiomer analysis by complexation gas
and ‘argentation’ thin-layer chromatography cor- chromatography: scope, merits and limitations. Journal
roborate the feasibility of this development of of Chromatography 441: 135}153.
III / CHIRAL SEPARATIONS / Liquid Chromatography 2381
J. Haginaka, Mukogawa Women’s University, III and IV produced with (!)-B overlap with the
Nishinomiya, Japan peaks of II and I, respectively, on achiral stationary
Copyright ^ 2000 Academic Press phases, because II and III, and I and IV are enan-
tiomeric pairs. Therefore, if the optical purity of the
A pair of enantiomers only differ in their optical derivatization reagent is not known or not taken into
rotation, but their physical properties such as melting consideration, the optical purity of the target com-
point, boiling point, refractive index and solubility pound will not be determined accurately. Further,
are identical. On the other hand, the physical and this is the case when the separation of enantiomers is
chemical properties of a pair of diastereomers are carried out by the use of chiral additives to the eluent
different. For the separation of enantiomers by liquid on achiral stationary phases.
chromatography, it is essential to form a dia- In contrast, direct resolution of enantiomers using
stereomeric complex in the mobile phase or in the chiral stationary phases does not have the drawbacks
stationary phase or to convert to diastereomers. The described above. We can easily determine 0.1%
former is called the direct method, and is based on the or 0.05% of the antipode using chiral stationary
addition of chiral additives to the eluent or the use of phases.
chiral stationary phases, resulting in the formation of
a diastereomeric complex in the mobile or stationary Indirect Methods
phase. The second indirect method is based on the
The indirect method, involving reaction with a
reaction of the enantiomers with a homochiral re-
homochiral reagent, is an efRcient technique for the
agent, resulting in the formation of a pair of dia-
separation of many enantiomers. It is essential that
stereomers. This article deals with both the indirect
the chiral derivatization proceeds completely in both
and direct methods using liquid chromatography.
enantiomers, and that racemization does not occur.
The indirect methods are unsuitable for analysis of
Purity of Chiral Derivatization enantiomers in a standard sample and pharmaceut-
Reagent or Chiral Selector ical preparations, because the low amount of the
antipode level should be determined, and are unsuit-
For the chiral derivatization method, there is one point able for preparative purposes. However, they are
which necessitates careful interpretations of the results. suitable for trace analysis of enantiomers in complex
When the enantiomeric mixture of (#)-A and (!)-A matrices such as biological samples and environ-
are derivatized with the chiral derivatizing reagent of mental samples because of the introduction of highly
(#)-B, it often includes (!)-B as a chiral impurity. If the sensitive tags. These include UV-visible, Suorescent
reagent B is 100% optically pure, two diastereomers and electrochemical tags. Fluorescence derivatization
(#)-A-(#)-B (I) and (!)-A-(#)B (II) are formed. If is the most effective to determine the target com-
reagent B is not optically pure, additional diastereomers pound in complex matrices in terms of sensitivity
(#)-A-(!)-B (III) and (!)-A-(!)B (IV) are formed. and/or selectivity. Table 1 shows the Suorescence
derivatization reagents (Figures 1 and 2) used for the
[(#)-A#(!)-A]# [(#)-B#(!)-B] P separation of enantiomers bearing amino, keto,
Enantiomers Reagents hydroxyl and carboxyl groups.
(#)-A-(#)-B# (!)-A-(#)B# Direct Methods

I II
Direct methods using chiral mobile-phase additives
(#)-A-(!)-B# (!)-A-(!)B can separate many enantiomers by addition of chiral
III IV selectors to an eluent on achiral stationary phases.
However, for preparative applications, the additives
Reaction products must be removed. Separations on chiral stationary
phases can be used for both analytical and prepara-
The enantiomers of A to be separated and deter- tive purposes. At the present time, the trend is to
mined as their diastereomeric derivatives (I and II) use chiral stationary phases for the separation of
can be resolved as the respective peaks on an achiral enantiomers by liquid chromatography whenever
stationary phase. However, the peaks of the products possible.
2382 III / CHIRAL SEPARATIONS / Liquid Chromatography
Table 1 Fluorescence derivatizating reagents used for the separation of enantiomers
Reagent Enantiomer Separation mode
OPA#N-protected L-cystein Amino acids, amino alcohols, baclofen, tranylcypromine Reversed-phase

FLEC Amino acids, methamphetamine, ephedrine, atenolol Reversed-phase
DANE Amino acids, ibuprofen, indoprofen, naproxen, loxoprofen Normal-phase
NEIC Propranolol, nadolol, prenylamine, betaxolol, acebutolol Reversed-phase and normal-phase
Methyl-BNCC Hydroxyls, -hydroxy acids, propranolol, penbutolol Normal-phase
NBD-Apy N -acetyl amino acids, antiinflammatory drugs Reversed-phase
NBD-Pro-COCl Alcohols, amine Reversed-phase
NBD-ProCZ Ketones Reversed-phase
NBD-PyNCS Amines, -blockers, amino acids, peptides Reversed-phase
FLOPA Ibuprofen, -phenylcyclopentyl acid Reversed-phase and normal-phase
FLOP-Cl Amino acids, peptides Reversed-phase and normal-phase
MNE-OTf -Methoxyphenylacetic acid, propranolol Reversed-phase
See Figures 1 and 2 for definitions of abbreviations.
Chiral Stationary Phases E Type II: the primary mechanism for the formation
of the solute and chiral stationary phase complex is
Many chiral selectors are adsorbed or immobilized through attractive interactions, but inclusion com-
covalently on to liquid chromatography supports. plexes also play an important role.
The chiral stationary phases obtained are classiRed E Type III: the solute enters into chiral cavities within
into two types with respect to their general structure. the chiral stationary phase to form inclusion com-
One type is based on synthetic or natural polymers, plexes.
which are totally or intrinsically chiral; the other type E Type IV: the solute is a part of a diastereomeric
is based on a chiral selector of low molecular weight. metal complex; this is called chiral ligand exchange
Chiral stationary phases can be further classiRed into chromatography.
Rve types based on the solute and chiral stationary- E Type V: the chiral stationary phase is a protein and
phase interactions: the solute and chiral stationary phase complexes
are based on combinations of hydrophobic, elec-
E Type I: the diastereomeric complexes of the solute trostatic and hydrogen-bonding interactions.
and chiral stationary phase are formed by attract-
ive interactions such as hydrogen-bonding, }, Table 2 shows commercially available chiral sta-
dipole stacking, etc., between the solute and chiral tionary phases which are classiRed into these Rve
stationary phase. types.
Figure 1 Structure of reagents (see Table 1).

Figure 2 Structure of reagents (see Table 1).
Type I Chiral Stationary Phases gen-bonding interactions in the nonpolar solvent used
as the mobile phase play an important role in chiral
These stationary phases were based on aromatic - recognition, as described above, these stationary
acid (3,5-dinitrobenzene) and -base (a naphthalene phases are mainly used in the normal-phase mode,
moiety) derivatives. In addition to } interaction although it is possible to separate some enantiomers
sites, they have hydrogen-bonding and dipole}dipole in the reversed-phase mode.
interaction sites provided by an amide, urea or ester Recently, chiral stationary phases based on the
moiety. By nuclear magnetic resonance measure- macrocyclic antibiotics, vancomycin and teicoplanin,
ments in solution, three interactions are proposed to have been developed. These stationary phases can
occur in the more favoured diastereomeric complex separate many enantiomers in both normal- and
between an N-(3,5-dinitrobenzoyl)--amino amide reversed-phase modes.
and an N-(2-naphthyl)--amino ester: a }donor}ac-
ceptor interaction and two hydrogen-bonding inter-
actions, as shown in Figure 3. In the development of
Type II Chiral Stationary Phases
these stationary phases, the reciprocality concept was Underivatized saccharides such as cellulose and
introduced as follows: if optically active A resolves starch have been used as chiral stationary phases.
the enantiomers B, then optically active B resolves the Cellulose, which contains c. 200 glucose units (micro-
enantiomers of A. As a result, }acid and }base crystalline cellulose), has been extensively used for
compounds have been separated using chiral station- the chiral resolution of highly polar compounds such
ary phases based on }base and }acid derivatives. as amino acids, amino acid derivatives and dia-
Since a combination of simultaneous } and hydro- minodicarboxylic acids. The chiral recognition ability
Table 2 Chiral stationary phases for liquid chromatography
Type Ligand Trade name
Type I N-(3,5-Dinitrobenzoyl)-D-, D-,L-Phenylglycine

N-(3,5-Dinitrobenzoyl)-L-phenylglycine
N-(3,5-Dinitrobenzoyl)-D-, D-,L-Leucine
N-(3,5-Dinitrobenzoyl)-L-leucine
Naphthyl-D-, naphthyl-L-alanine D-, L-Naphthylalanine
Naphthyl-D-, naphthyl-L-leucine D-,L-Naphthylleucine
N-(3,5-Dinitrobenzoyl)-(R )-1-naphthylglycine Sumichiral OA-2500
N-(3,5-Dinitrobenzoyl)-aminocarbonyl-L-valine Sumichiral OA-3100
N-[(S )-(1-Naphthyl) ethylaminocarbonyl]-L-valine Sumichiral OA-4000
(R )-, (S )-1-Naphthylethylamine LC-(R )-, LC-(S )-Naphthyl urea
Teicoplanin Chirobiotic T
Vancomycin Chirobiotic V
Type II Microcrystalline cellulose triacetate Chiralcel CA-1

Cellulose triacetate
Cellulose tris(4-methylbenzoate) Chiralcel OJ, OJ-R
Cellulose tribenzoate Chiralcel OB, OB-H
Cellulose triacetate Chiralcel OA
Cellulose tricinnamate Chiralcel OK
Cellulose tris(3,5-dimethylphenylcarbamate) Chiralcel OD, OD-H, OD-R
Cellulose tris-phenylcarbamate Chiralcel OC
Cellulose tris(4-methylphenylcarbamate) Chiralcel OG
Cellulose tris(4-chlorophenylcarbamate) Chiralcel OF
Amylose tris(3,5-dimethylphenylcarbamate) Chiralpak AD
Amylose tris((S )-1-phenylethylcarbamate) Chiralpak AS
Poly-N-acryloyl-(S )-phenylalanine ethylester ChiraSpher
(#)-Poly(triphenylmethyl methacrylate) Chiralpak OT(#)
(#)-Poly(diphenyl-2-pyridylmethyl methacrylate) Chiralpak OP(#)
Type III 2,2-Diphenyl-1,1-binaphthol derivatives of 18-crown-6 Crownpak CR(#)

Crownpak CR(!)
-, -, -Cyclodextrin Cyclobond III, I, II
-, -Cyclodextrin ChiraDex, ChiraDex Gamma
-Cyclodextrin derivatives Cyclobond I Ac, SP, RSP, SN, RN, DMP, PTa ;
Ultron ES-PhCDb ; Nucleosil -PMc
-, -Cyclodextrin derivatives Cyclobond III Ac, II Ac
Type IV L-Hydroxyproline Nucleosil Chiral-1

2-Amino-1,2-diphenylethanol Chiralpak WE
N,S -Dioctyl-D-penicillamine Sumichiral OA-5000
Type V Bovine serum albumin Resolvosil BSA-7, BSA-7PX

Ultron ES-BSA
Chiral-BSA
Human serum albumin Chiral-HSA
Chiral-HSA
1-Acid glycoprotein Chiral-AGP
Ovomucoid Ultron ES-OVM
Avidin Bioptic AV-1
Cellulase Chiral-CBH
Pepsin Ultron ES-Pepsin
a
Ac, Acetylate; SP, (S )-2-hydroxypropyl ether; RSP, racemic 2-hydroxylpropyl ether; SN, (S )-naphthylethylcarbamate; RN, (R )-
naphthylethylcarbamate; DMP, 2,6-dimethylphenylcarbamate; PT, para-toluoyl ester.
b
PhCD, Phenylcarbamate.
c
-PM, Permethylate.
of the cellulose is based on the microcrystallinity It was found that microcrystalline cellulose triacet-
because, when treated with dilute alkali, cellulose ate preserved microcrystallinity and had excellent
loses its chiral recognition ability, resulting in a stable chiral recognition ability. In contrast, microcrystal-
amorphous form. line cellulose triacetate precipitated from a solution
Type III Chiral Stationary Phases

Type III chiral stationary phases include cyclodextrin
(CD), polymethacrylate and crown ether stationary
phases. The solute enters into chiral cavities to form
inclusion complexes and the relative stability con-
stants of the resulting diastereomeric complexes
are different. The cavities of CD and polymetha-
crylate are hydrophobic, while those of crown ethers
are hydrophilic.
-, - and -CD are cyclic oligosaccharides contain-
ing 6, 7 and 8 -1, 4-D-glucoside units, respectively.
Because an aromatic portion of a molecule can enter
into the chiral cavity, solutes having an aromatic
moiety at, or adjacent to, the chiral centre are well
resolved. Acetyl, hydroxypropyl, naphthylethyl car-
bamate and phenyl carbamate derivatives of CDs
have been prepared and used for chiral resolution of
many solutes which cannot be separated using native
CDs. Chiral stationary phases based on CDs and
Figure 3 The more favoured diastereomeric complex between derivatized CDs have been predominantly used in
an N-(3,5-dinitrobenzoyl)--amino amide and an N-(2-naphthyl)- reversed-phase mode.
-amino ester. A }donor}acceptor interaction and two hydrogen- Chiral stationary phases based on polymetha-
bonding interactions are indicated by arrows. (Reproduced with crylate have been prepared using a chiral monomer
permission from Pirkle WH and Pochapsky TC (1987) Advances such as (S)-acryloylphenylalanine, and achiral mono-
in Chromatography, vol. 27, p. 116. New York: Marcel Dekker.)
mers such as triphenylmethyl methacrylate and
diphenyl-2-pyridylmethyl methacrylate together with
chiral cation catalysts. In normal-phase mode, chiral
has another morphology, and different chiral recogni- stationary phases based on these polymers can separ-
tion properties, compared with microcrystalline cellu- ate many racemic solutes that are difRcult to resolve
lose triacetate. The chiral stationary phases based on by other methods because of a lack of functionality.
the cellulose triacetate precipitated from a solution are Crown ethers, which are synthetic macrocyclic
prepared by coating the polymer on a silica gel matrix. polyethers, can form selective complexes with various
Many chiral stationary phases prepared by this tech- cations. Chiral crown ethers covalently bound or
nique are commercially available as triacetate, triben- adsorbed on to silica supports can separate enan-
zoate, trisphenylcarbamate, tribenzyl ether and tricin- tiomeric ammonium compounds such as amino acid
namate derivatives of cellulose. esters, amines and amino alcohols. The multiple hy-
Chiral stationary phases based on cellulose and drogen-bonding interactions between the ammonium
amylose derivatives could separate 78% of racemates group and the ether oxygens play an important role in
examined. However, the disadvantage of stationary the chiral recognition.
phases based on the cellulose and amylose derivatives
coated on silica gels is the restriction of the eluents
used; the coated polymer is soluble in some eluents
Type IV Chiral Stationary Phases
and removed. Polysaccharide derivatives chemically Chiral ligand exchange chromatography is based on
bound to silica gel overcome this problem. The the formation of diastereomeric ternary complexes
coated and the chemically bound polymers showed that involve a transition metal ion (M), chiral ligand
different chiral recognition properties. (L), and the enantiomers of the racemic solute (R and
With regard to the chiral recognition mechanism, S). Of all the transition metals examined [Cu(II),
the interaction of the solute and the chiral stationary Ni(II), Zn(II), Hg(II), Co(III), Fe(III), etc.], Cu(II)
phase based on cellulose phenylcarbamate derivatives forms the most stable complexes, and cyclic amino
has been investigated by computational chemistry acids such as L-proline and L-hydroxyproline are the
and nuclear magnetic resonance measurements of the best chiral selectors, when bound to a polymer
complex. It was found that } and hydrogen-bond- or silica support. The diastereomeric mixed chelate
ing interactions play an important role in chiral rec- complexes formed in this system are represented
ognition of the solute. by the formulas L-M-R and L-M-S. When these
complexes have different stabilities, the less stable and ovotransferrin from egg whites, and Savoprotein
complex is eluted Rrst, and the enantiomeric solutes (riboSavin-binding protein) from egg whites and
are separated. The enantiomers resolved include yolks. The advantages of protein-based stationary
amino acids, amino acid derivatives, 2-amino alco- phases generally include the use in reversed-phase
hols, barbiturates and hydantoins. mode, enantioselectivity for a wide range of com-
pounds and direct analysis without derivatization.
The disadvantages are low capacity, lack of column
Type V Chiral Stationary Phases ruggedness and limited understanding of the chiral
Chiral stationary phases based on a protein are of recognition mechanism. These stationary phases are
special interest because of their unique properties of mainly used for analytical purposes.
stereoselectivity and because they are suited for separ- With regard to the chiral recognition mechanism,
ating a wide range of enantiomeric mixtures. Protein- hydrophobic, electrostatic and hydrogen-bonding in-
based stationary phases so far developed have interactions take place between the chiral stationary
cluded albumins such as bovine and human serum phase and the solute.
albumin, enzymes such as trypsin, -chymotrypsin,
lysozyme and pepsin, and glycoproteins such as
1-acid glycoprotein from human or bovine serum,
Chiral Mobile Phase Additives
cellobiohydrolase I from the fungus Trichoderma There are no fundamental differences between the
reesei, ovomucoid (in fact, ovoglycoprotein), avidin techniques using chiral stationary phases and chiral
Table 3 Chiral selectors used as mobile-phase additives
Chiral selector Enantiomers Stationary phase
Metal complex
L- or D-Proline#Cu(II), Amino acids Cation exchanger,
L-Phenylalanine#Cu(II), ODS, OS
N-Methyl-L-phenylalanine#Cu(II),
N,N-Dimethyl-L-phenylalanine#Cu(II),
(R, R )-Tartaric acid mono-1-octylamide#Cu(II)
L-Propyl-n-octylamide#Ni(II), Dansyl amino acids ODS, OS
L-Proline#Cu(II),
L-Arginine#Cu(II),
L-Histidine#Cu(II)
(R, R )-Tartaric acid mono-1-octylamide#Cu(II) -Amino alchols ODS
Uncharged additives
1,1-Binaphthyl derivatives of 18-crown-6 Amino acids ODS
Cyclodextrins
-Cyclodextrin -, -Pinene ES
-Cyclodextrin Propranolol, pseudoephedrine, salsolinol, Porous graphite carbon,
thalidomide, dansyl amino acids CN, ODS
-Cyclodextrin Norgesterol ODS, CN
TM--Cyclodextrin Benzoin, ethyl mandelate Si
5-Butyl-1-methyl-5-phenylbarbituric acid, ODS
Dansyl phenylalanine
CM-, CE--Cyclodextrin Aminoethylbenzodioxane derivatives, hexobarbital BS
Chiral acid
10-Caphorsulfonic acid Amino alcohols Diol
Z-Glycyl-L-proline Amino alcohols, N-alkylated-2-aminotetralines Diol
Chiral amine
Alprenolol, quinine, quinidine, cinchonidine 10-Caphorsulfonic acid, N-(1-phenylethyl) Diol
phthalamic acid, O-methylmandelic acid,
O-methoxy--trifluoromethyl-phenylacetic
acid, 2-phenylacetic acid, naproxen
ODS, octadecylsilyl; OS, octylsilyl; ES, ethylsilyl; CN, cyanopropylsilyl; TM, heptakis(2,3,6-tri-O-methyl); Si, silica gels; CM, car-
boxymethyl; CE, carboxyethyl; BS, butylsilyl; Z, N-benzoxycarbonyl.
III / CHIRAL SEPARATIONS / Molecular Imprints as Stationary Phases 2387
mobile phase additives. This means that all chiral Further Reading
selectors covalently bound to supports can be used for
Allenmark S (1991) Chromatographic Enantioseparation:
the separation of enantiomers by addition to the mo- Methods and Applications, 2nd edn. Chichester: Ellis
bile phase. With chiral mobile phase additive tech- Horwood.
niques, there are at least two possible mechanisms; Krstulovic AM (ed.) (1989) Chiral Separations by HPLC:
one is that the chiral mobile phase additive and the Applications to Pharmaceutical Compounds. Chiches-
enantiomers may form diastereomers in the mobile ter: Ellis Horwood.
phase. Another is that the stationary phase may be Lipkowitz KB (1995) Theoretical studies of type II}V chiral
coated with the chiral mobile phase additive, result- stationary phases. Journal of Chromatography A, 694:
ing in diastereomeric interactions with the enan- 15}37.
tiomeric pairs during chromatography. It is thought Lough WJ (ed.) (1989) Chiral Liquid Chromatography.
that both mechanisms occur depending on both the London: Blackie.
Pirkle WH and Pochapsky TC (1987) Chiral stationary
stationary and mobile phases used.
phases for direct LC separation of enantiomers. Ad-
The techniques using chiral mobile phase additives vances in Chromatography 27: 73}127.
can be divided into three categories: metal complexa- Stevenson D and Wilson ID (eds) (1988) Chiral Separ-
tion used in chiral ligand exchange chromatography, ations. New York: Plenum Press.
the use of various uncharged additives, and the ion- Taylor DR and Maher K (1992) Chiral separations by
pairing techniques used for charged analytes. Table 3 high-performance liquid chromatography. Journal of
shows chiral selectors used as mobile phase additives. Chromatographic Science, 30: 67}85.
Toyo’oka T (1996) Recent progress in liquid chromato-
See also: II/Chromatography: Liquid: Derivatization; graphic enantioseparation based upon diastereomer
Mechanisms: Chiral; III/Chiral Separations: Cellulose formation with Suorescent chiral derivatization re-
and Cellulose Derived Phases; Chiral Derivatization; Cyc- agents. Biomedical Chromatography, 10: 265.
lodextrins and Other Inclusion Complexation Approaches; Wainer IW (1993) Drug Stereochemistry: Analytical
Ligand Exchange Chromatography; Ion-Pair Chromatog- Methods and Pharmacology, 2nd edn. New York:
raphy; Protein Stationary Phases; Synthetic Multiple Inter- Marcel Dekker.
action (‘Pirkle’) Stationary Phases. Inclusion Complexa- Zief M and Crane LJ (eds) (1988) Chromatographic Chiral
tion: Liquid Chromatography. Separations. New York: Marcel Dekker.
Molecular Imprints as Stationary Phases

M. Kempe, Lund University, Lund, Sweden until the 1970s that the Reld of molecular imprint-
ing started to mature to its present form. Wulff
introduced a new approach that allowed the intro-
duction of functional groups at deRned positions in
synthetic network polymers. This approach is often
In 1949, Frank Dickey published what can be con- referred to as covalent molecular imprinting, to
sidered the Rrst paper on a molecularly imprinted distinguish it from the noncovalent approach de-
synthetic material. The work was inspired by the veloped by Mosbach and his co-workers in the early
theories of Dickey’s mentor Linus Pauling, who had 1980s.
suggested that the primary structures of all polypep- The Reld of molecular imprinting is growing rap-
tides constituting the antibodies are the same and that idly. Molecularly imprinted polymers (MIPs) have
the diversity originates from folding directed by phys- found application as stationary phases for chiral sep-
ical contact with the antigens. Even if Pauling’s arations and solid-phase extractions, as synthetic
hypothesis on antibodies later turned out to be incor- antibodies in competitive ligand-binding assays, as
rect, his ideas laid the foundation for the concept of recognition elements in sensors and as catalysts of
molecular imprinting. Consistent with these early chemical reactions. The concepts of molecular im-
studies, molecular imprinting can be deRned as printing will be described brieSy here. For a more
a method in which the selectivity of a material for detailed description of the imprinting principle and
a chosen molecule is induced by the presence of the the preparation of molecularly imprinted polymers,
molecule during the preparation, assembling or re- see ‘AfRnity Separation: Imprint Polymers’ in this
arrangement of the material. encylopedia. The remainder of this paper focuses on
Dickey’s studies were followed by several other the use of molecularly imprinted polymers as chiral
investigations in the same direction, but it was not stationary phases (CSPs).
2388 III / CHIRAL SEPARATIONS / Molecular Imprints as Stationary Phases
The Concepts of Molecular Imprinting ive as stationary phases for chromatography. Even if
several examples of covalently imprinted stationary
Molecular imprinting, sometimes referred to as tem- phases have been reported, it was not until the devel-
plate polymerization, is a technique for preparing opment of the noncovalent approach that the tech-
synthetic polymers of predetermined selectivity. Re- nique became competitive for the preparation of
ceptor-like binding sites are tailor-made in situ by the CSPs. This review will therefore focus on noncovalent
copolymerization of cross-linkers and functional molecular imprinting. The covalent approach has
monomers, which are interacting covalently or non- been covered in several excellent reviews by Wulff.
covalently with print molecules (or templates). After Even if the general understanding has been that the
polymerization, the print molecules are removed selectivity of MIPs is due to the formation of speciRc
from the polymer, either by extraction or by chemical recognition sites complementary to the print mole-
cleavage, leaving recognition sites complementary to cules, early critics of the technique argued that the
the print molecules in the shape and positioning of recognition could come from binding to print mol-
functional groups. The polymer is subsequently able ecules entrapped in the highly cross-linked polymers.
to rebind the print molecules. The noncovalent A small percentage of the print molecules are inac-
approach of molecular imprinting is exempliRed in cessible and remain in the polymer network after
Figure 1. extraction (normally less than 1%). However, this
The association/dissociation kinetics of non- was ruled out as the cause of the selectivity by experi-
covalent MIPs is in general faster than is observed ments showing that a polymer containing covalently
with polymers prepared by the covalent approach. bound print molecules did not exhibit any selective
For this reason, the former polymers are more attract- binding of the print molecule.
Figure 1 Schematic representation of the concept of noncovalent molecular imprinting. Functional monomers, in this case meth-
acrylic acid, interact with the print molecule, Z-Glu-OH. Cross-linker is added and the polymerization is initiated. The interactions are
maintained in the resulting polymer. The print molecule is removed from the polymer by extraction, leaving a specific recognition site
complementary to the print molecule in shape and positioning of the functional groups. The polymer has attained a memory of the print
molecule and is able selectively to rebind it.
Molecular imprinting produce recognition sites lene glycol dimethacrylate) imprinted with one of the
with a distribution of binding strengths; the sites are enantiomers (Figure 2).
heterogeneous. Some sites have a highly selective af- The speciRcity and selectivity of MIPs can be Rne-
Rnity for the template, whereas others are less selec- tuned by careful choice of monomers and solvent,
tive. When used for chromatographic applications, and by optimizing the molar ratios of the components
the heterogeneity is reSected in band broadening and in the polymerization mixture. The recognition relies
asymmetric peaks. on multiple interaction points. The more fun-
tionalized the print molecule is, the more interactions
are possible. An example of a highly selective polymer
Chiral Separations is poly(4-vinylpyridine-co-EDMA) imprinted with
Initial studies on noncovalent MIPs focused mainly Z-L-aspartic acid (1). The chromatogram in Fig-
on the preparation of materials selective for amino ure 3A shows the separation of racemic Z-aspartic
acid derivatives. The polymers possessed not only acid on this CSP. Aspartic acid and glutamic acid
a selectivity for the amino acid used as the print differ by only one methylene unit, but despite this
molecule, but were also found to be enantioselective; small difference, the Z-aspartic acid-imprinted CSP
the enantiomer present during the polymerization was not able to resolve racemic Z-glutamic acid (2)
was preferentially bound. These Rndings, together (Figure 3B). The side chain of Z-L-glutamic acid can-
with observations that the polymers were mechan- not be accommodated into the recognition site in
ically and chemically stable, made them attractive as a way that allows speciRc interaction between the
CSPs and spurred further research. carboxy functionality and the positioned pyridine
The imprinting effects of MIPs prepared with op- group of the polymer. The same type of polymer,
tically active compounds as the print molecules are imprinted with Z-L-glutamic acid, was able to resolve
readily demonstrated by chromatographic evalu- racemic Z-glutamic acid, but not racemic Z-aspartic
ation. For example, when the L-enantiomer of an acid. Similar observations have been done on
amino acid derivative is used as the print species, poly(methacrylic acid-co-EDMA) imprinted with
a column packed with the resulting polymer will these print molecules.
retain the L-enantiomer longer than the D-enantiomer,
and vice versa when the D-enantiomer is used as the
print molecule. Reference polymers prepared with the
racemate or without print molecule will not be able to
resolve the racemate. A stereoselective memory is
hence induced in the polymers by the print molecules.
Molecularly imprinted CSPs have been prepared
for applications in high performance liquid
chromatography (HPLC), thin-layer chromatography
(TLC), capillary electrophoresis (CE) and capillary
electrochromatography (CEC). CSPs for HPLC are
by far the most studied.
HPLC
A large number of chiral amino acids and peptides
have been imprinted. Several MIPs selective for phar-
maceuticals have also been described. The most wide- Polymers imprinted with Z-L-phenylalanine (3a)
ly used method has been bulk polymerization fol- were able to separate racemic Z-phenylalanine efR-
lowed by grinding, sieving and packing into HPLC ciently (Table 2). Racemic Z-alanine (3b) could also
columns. Alternatively, the polymers can be prepared be separated on these CSPs, even if lower separation
by any of the methods discussed in ‘AfRnity Separ- factors were observed. When the amino-group of the
ation: Imprint Polymers’ in this encyclopedia. Some racemate was protected with tert-butyloxycarbonyl
examples of MIP CSPs are shown in Table 1. (Boc) (3c) or 9-Suorenylmethyloxycarbonyl (Fmoc)
The selectivities are in many cases comparable to (3d), the separations were poorer than with the ra-
those of commercially available CSPs. For example, cemate of the print molecule, which was protected
a separation factor () of 17.8 was found for the with benzyloxycarbonyl (Z). In contrast to the CSPs
separation of the two enantiomers of a dipeptide on described above, which were unable to separate the
poly(methacrylic acid-co-EDMA) (EDMA"ethy- racemate of a molecule which only differed from the
Table 1 A selection of molecularly imprinted CSPs for HPLCa
Print molecule Polymer b c RSd f/ge
Amino acids
H-L-Phe-OHf poly (CuVBIDA-co -EDMA) 1.45 n.d. n.d.
H-L-Phe-NHPhg poly (MAA-co -EDMA) 13 n.d. n.d.
H-D-p -NH2Phe-NHPhh poly (MAA-co -EDMA) 15 n.d. n.d.
Ac-D-Trp-OMei poly (MAA-co -EDMA) 3.92 2.2 1.0
Ac-L-Trp-OHj poly (AA-co -EDMA) 3.24 2.02 n.d.
Boc-L-Trp-OHi poly (MAA-co -2VPy-co -EDMA) 4.35 1.9 1.0
Fmoc-L-Phe-OHk poly (MAA-co -EDMA) 1.36 n.d. n.d.
Z-L-Asp-OHl poly (4VPy-co -EDMA) 2.81 1.22 0.81
Z-L-Phe-OHm poly (MAA-co -TRIM) 2.29 3.14 1.00
Z-L-Tyr-OHm poly (MAA-co -PETRA) 2.86 5.47 1.00
Dansyl-L-Phe-OHi poly (MAA-co -2VPy-co -EDMA) 3.15 1.6 0.96
Small peptides
H-L-Phe-Gly-NHPhn poly (MAA-co -EDMA) 5.1 n.d. n.d.
Boc-L-Phe-Gly-OEtm poly (MAA-co -TRIM) 3.04 3.44 1.00
Z-L-Ala-L-Ala-OMem poly (MAA-co -TRIM) 3.19 4.50 1.00
Ac-L-Phe-L-Trp-OMeo poly (MAA-co -EDMA) 17.8 n.d. 1.00
Z-L-Ala-Gly-L-Phe-OMem poly (MAA-co -TRIM) 3.60 4.15 1.00
Pharmaceuticals
(S )-Timololp poly (MAA-co -EDMA) 2.9 2.0 n.d.
(S )-Naproxenq poly (4VPy-co -EDMA) 1.65 n.d. n.d.
(S ,R )-Ephedriner poly (MAA-co -TRIM) 3.42 1.6 n.d.
(S ,S )-Pseudoephedriner poly (MAA-co -TRIM) 3.19 1.8 n.d.
a
The print molecules and their optical antipodes were separated.
b
AA, Acrylamide; CuVBIDA, Cu(II)[N -(4-vinylbenzyl)]iminodiacetate; EDMA, ethylene glycol dimethacrylate; Ita, itaconic acid; MAA,
methacrylic acid; PETRA, pentaerythritol triacrylate; TRIM, trimethylolpropane trimethacrylate; 2VPy, 2-vinlypyridine; 4VPy, 4-vin-
lypyridine.
c
The separation factor were calculated as "k print molecule/k optical antipode; k "(t!t 0)/t 0 ; t is the retention time of the analyte and t0 is the
retention time of unretained compound (the void).
d
The resolution factors (R S) were calculated according to Wulff G, Poll HG and MinaH rik M (1986) Enzyme-analogue built polymers. XIX.
Racemic resolution on polymers containing chiral cavities. Journal of Liquid Chromatography 9: 385}405.
e
The resolution factors (f /g ) were calculated according to Meyer VR (1987) Some aspects of the preparative separation of enantiomers
on chiral stationary phases. Chromatographia 24: 639}645.
f
Data from Vidyasankar S, Ru M and Arnold FH (1997) Molecularly imprinted ligand-exchange adsorbents for the chiral separation of
underivatized amino acids. Journal of Chromatography A 775: 51}63.
g
Data from Sellergren B and Shea KJ (1993) Chiral ion-exchange chromatography. Correlation between solute retention and
a theoretical ion-exchange model using imprinted polymers. Journal of Chromatography A 654: 17}28.
h
Data from Sellergren B and Nilsson KGI (1989) Molecular imprinting by multiple noncovalent host-guest interactions: Synthetic
polymers with induced specificity. Methods in Molecular and Cellular Biology 1: 59}62.
i
Data from RamstroK m O, Andersson LI and Mosbach K (1993) Recognition sites incorporating both pyridinyl and carboxy functionalities
prepared by molecular imprinting. Journal of Organic Chemistry 58: 7562}7564.
j
Data from Yu C and Mosbach K (1997) Molecular imprinting utilizing an amide functional group for hydrogen bonding leading to highly
efficient polymers. Journal of Organic Chemistry 62: 4057}4064.
k
Data from Kempe M and Mosbach K (1994) Chiral recognition of N?-protected amino acids and derivatives in non-covalently
molecularly imprinted polymers. International Journal of Peptide and Protein Research 44: 603}606.
l
Data from Kempe M, Fischer L and Mosbach K (1993) Chiral separation using molecularly imprinted heteroaromatic polymers. Journal
of Molecular Recognition 6: 25}29.
m
Data from Kempe M (1996) Antibody-mimicking polymers as chiral stationary phases in HPLC. Analytical Chemistry 68: 1948}1953.
n
Data from Andersson LI, O’Shannessy DJ and Mosbach K (1990) Molecular recognition in synthetic polymers: preparation of chiral
stationary phases by molecular imprinting of amino acid amides. Journal of Chromatography 513: 167}179.
o
Data from RamstroK m O, Nicholls IA and Mosbach K (1994) Synthetic peptide receptor mimics: Highly stereoselective recognition in
non-covalent molecularly imprinted polymers. Tetrahedron : Asymmetry 5: 649}656.
p
Data from Fischer L, MuK ller R, Ekberg B and Mosbach K (1991) Direct enantioseparation of -adrenergic blockers using a chiral
stationary phase prepared by molecular imprinting. Journal of the American Chemistry Society 113: 9358}9360.
q
Data from Kempe M and Mosbach K (1994) Direct resolution of naproxen on a non-covalently molecularly imprinted chiral stationary
phase. Journal of Chromatography A 664: 276}279.
r
Data from RamstroK m O, Yu C and Mosbach K (1996) Chiral recognition in adrenergic receptor binding mimics prepared by molecular
imprinting. Journal of Molecular Recognition 9: 691}696.
Figure 2 Separation of 10 g of a mixture of Ac-L-Phe-L-Trp-OMe and Ac-D-Phe-D-Trp-OMe on a poly (methacrylic acid-co -EDMA)
CSP (4.6;200 mm column) imprinted with Ac-L-Phe-L-Trp-OMe. Isocratic elution at 1 mL min\1 with CHCl3-HOAc (99 : 1). Attentu-
ation was increased 10-fold at 10 min. (Adapted from RamstroK m O, Nicholls IA and Mosbach K (1994) Tetrahedron : Asymmetry 5:
649}656, 1994, with permission from Elsevier Science, UK.)
print molecule by one methylene unit, these polymers the amino-protecting group and the side chain, but
were able to separate all of the tested structurally that an exact Rt is not necessary for enantioseparation
related racemates, even if the separations were not as in these cases.
good as with the print molecule and its optical an-
tipode. This shows that the polymer recognizes both
Figure 3 Poly (4-vinylpyridine-co -EDMA) CSP (4.6;200 mm

column) imprinted with Z-L-aspartic acid. (A) 20 g of racemic Z-
aspartic acid was applied. Isocratic elution at 0.5 mL min\1 with
tetrahydrofuranIHOAc (24 : 1) and detection at 260 nm. (B) 20 g
of racemic Z-glumatic acid was applied. Isocratic elution at
0.5 mL min\1 with tetrahydrofuran} HOAc (199 : 1) and detection
at 260 nm. (Adapted from Kempe M, Fischer L and Mosbach K In a study on -adrenergic blockers, (S)-timolol
(1993) Journal of Molecular Recognition 6: 25}29, 1993, with was imprinted in EDMA-based polymers. When the
permission from John Wiley & Sons, UK.) functional monomer was methacrylic acid, the
Table 2 Separation of racemic amino acid derivatives on Z-L-Phe-OH-imprinted CSPs
Racemate poly (MAA-co-EDMA) a ,b ,c poly (MAA-co-TRIM) a,b,d

Z-Phe-OH 1.84 2.49

Boc-Phe-OH 1.31 1.78
Fmoc-Phe-OH 1.21 1.66
Z-Ala-OH 1.28 1.59
a
EDMA, Ethylene glycol dimethacrylate; MAA, methacrylic acid; TRIM, trimethylolpropane
trimethacrylate.
b
The separation factors were calculated as "k L / k D; k"(t!t 0)/t 0; t is the retention time
of the analyte and t 0 is the retention time of unretained compound (the void).
c
Data from Kempe M and Mosbach K (1994) Chiral recognition of N?-protected amino
acids and derivatives in non-covalently molecularly imprinted polymers. International
Journal of Peptide and Protein Research 44: 603}606.
d
Data from Kempe M (1996) Antibody-mimicking polymers as chiral stationary phases in
HPLC. Analytical Chemistry 68: 1948}1953.
polymer was able to resolve not only racemic timolol It is noteworthy that the load capacity and the
(4), but also racemic propanolol (5). In contrast to resolving capability increased when the trifunctional
this, a timolol-imprinted polymer prepared with cross-linker pentaerythritol triacrylate (PETRA) was
itaconic acid as the functional monomer instead of used instead of EDMA. The same features have been
methacrylic acid could only resolve racemic timolol seen with polymers prepared with trimethylol-
out of a number of racemates of structurally related propane trimethacrylate (TRIM), another trifunc-
-blockers (Figure 4). This clearly demonstrates that tional cross-linker. Poly(methacrylic acid-co-TRIM)
the selectivity of MIPs can be highly dependent on the imprinted with a dipeptide was able to resolve 1 mg
functional monomer used. of the racemate with almost baseline separation (ana-
lytical column: 4.6;250 mm) (Figure 5).
(S)-Naproxen (6), a nonsteroidal anti-inSammat-
ory drug, has been imprinted in poly(4-vinylpyridine-
co-EDMA) by two different approaches. Bulk polym-
erization followed by grinding and sieving resulted in
highly irregular particles and a two-step swelling and
polymerization method gave uniformly sized beads.
Both materials, used in the chromatographic mode,
were able to separate naproxen from the related ibup-
rofen (7) and ketoprofen (8). The polymers were also
able to resolve racemic naproxen, but not the ra-
cemates of ibuprofen and ketoprofen (Figure 6). Even
A comparison of six different CSPs, all imprinted

with the same print molecule (Z-L-phenylalanine),
conRrms that the choice of monomers is important
for the selectivity of the resulting polymers (Table 3).
The selectivity of EDMA-based polymers was higher
when vinylpyridines were used, either alone or to-
gether with methacrylic acid, than when methacrylic
acid was used alone. Methacrylic acid interacts with
the print molecule through hydrogen bonds. The
beneRcial effect of vinylpyridine is attributed to
strong ionic interactions between the carboxy groups Figure 4 Separation of 20 g of racemic timolol on a
of the print molecule and the pyridinyl groups. The poly (itaconic acid-co -EDMA) CSP (4.6;200 mm column) im-
printed with (S )-timolol. Isocratic elution at 1 mL min\1 with
polymer prepared with acrylamide also showed
EtOH}tetrahydrofuran}HOAc (5 : 4 : 1). (Adapted from Fischer L,
a higher selectivity than the one prepared with meth- MuK ller R, Ekberg B and Mosbach K (1991) Journal of the Ameri-
acrylic acid. Acrylamide forms strong hydrogen can Chemical Society 113: 9358}9360, 1991, with permission
bonds even in a polar solvent such as acetonitrile. from the American Chemical Society, USA.)
Table 3 Chiral separation of racemic Z-Tyr-OH on molecularly imprinted CSPs
Polymer a Separated amount ( g) b RSc f /g d
poly (AA-co -EDMA)e 40 3.62 2.52 n.d.

poly (MAA-co -EDMA)f 10 1.82 n.d. 0.50
poly (4VPy-co -EDMA)g 20 4.00 1.53 0.94
poly (2VPy-co -EDMA)f 10 3.81 1.90 0.95
poly (MAA-co -2VPy-co -EDMA)f 10 4.32 1.90 0.97
poly (MAA-co -PETRA)h 100 2.86 5.47 1.00
poly (MAA-co -PETRA)h 1000 2.06 n.d. 0.93
a
AA, Acrylamide; EDMA, ethylene glycol dimethacrylate; MAA, methacrylic acid; 2VPy,
2-vinylpyridine; 4VPy, 4-vinylpyridine; PETRA, pentaerythritol triacrylate.
b
The separation factors were calculated as "k L/k D; k "(t !t 0)/t 0; t is the retention
time of the analyte and t 0 is the retention time of unretained compound (the void).
c
The resolution factors (RS) were calculated according to Wulff G, Poll HG and MinaH rik
M (1986) Enzyme-analogue built polymers. XIX. Racemic resolution on polymers contain-
ing chiral cavities. Journal of Liquid Chromatography 9: 385}405.
d
The resolution factors (f /g ) were calculated according to Meyer VR (1987) Some aspects
of the preparative separation of enantiomers on chiral stationary phases. Chromato-
graphia 24: 639}645.
e
Data from Yu C and Mosbach K (1997) Molecular imprinting utilizing an amide functional
group for hydrogen bonding leading to highly efficient polymers. Journal of Organic
Chemistry 62: 4057}4064.
f
Data from RamstroK m O, Andersson LI and Mosbach K (1993) Recognition sites incorpor-
ating both pyridinyl and carboxy functionalities prepared by molecular imprinting. Journal
of Organic Chemistry 58: 7562}7564.
g
Data from Kempe M, Fischer L and Mosbach K (1993) Chiral separation using molecu-
larly imprinted heteroaromatic polymers. Journal of Molecular Recognition 6: 25}29.
h
Data from Kempe M (1996) Antibody-mimicking polymers as chiral stationary phases in
HPLC. Analytical Chemistry 68: 1948}1953.
Figure 5 Separation of mixtures of Z-L-Ala-L-Ala-OMe and Z-D-Ala-D-Ala-OMe on a poly (methacrylic acid-co -TRIM) CSP
(4.6;250 mm column) imprinted with Z-L-Ala-L-Ala-OMe. (A) 100 g was applied. Gradient elution at 1 mL min\1 with CHCl3-HOAc
(99.75 : 0.25) and CHCl3-HOAc (4 : 1) ("B). Gradient: 0}10 min, 0% B; 10}18 min, 0}5% B; 18}22 min, 5% B; 22}24 min 5}0% B.
Detection at 260 nm. (B) 1 mg was applied. Isocratic elution at 1 mL min\1 with CHCl3-HOAc (99.75 : 0.25). Detection at 260 nm.
(Adapted from Kempe M (1996) Analytical Chemistry 68: 1948}1953, 1996, with permission from the American Chemical Society,
USA.)
Figure 6 Separation of racemic mixtures of naproxen, ibuprofen and ketoprofen on poly (4-vinylpyridine-co-EDMA) CSPs
(4.6;100 mm columns) imprinted with (S )-naproxen. (A) The column was packed with particles prepared by grinding and sieving
a bulk polymer. Isocratic elution at 0.1 mL min\1 with tetrahydrofuran}heptane}HOAc (250 : 250 : 1) and detection at 260 nm.
(Adapted from Kempe M and Mosbach K (1994) Journal of Chromatography A 664: 276}279, 1994, with permission from Elsevier
Science, UK.) (B) The column was packed with beads prepared by a two}step swelling and polymerization method. Isocratic elution at
1.0 mL min\1 with CH3CN}phosphate buffer (20 mmol L\1, pH 4.0) (1 : 1) and detection at 254 nm. (Adapted from Haginaka J,
Takehira H, Hosoya K and Tanaka N (1997) Chemistry Letters, 555}556, 1997, with permission from the Chemical Society of Japan.)
if comparisons of the two chromatograms cannot be In general, the polymerizations in noncovalent mo-
done because of differing Sow rates, it is not obvious lecular imprinting have to be done in nonaqueous
that the chromatographic efRciency was better with solutions to prevent water molecules from interfering
the uniformly sized beads (Figure 6B) than with the with the interactions between the monomers and the
irregular particles (Figure 6A). This may be due to templates, as previously discussed. Several reports,
impairment on the selectivity by water interfering however, show that the chromatography can be
with the monomer}print molecule complex, since performed efRciently with buffered aqueous eluents.
water was used as the suspension medium in the An approach has been developed which allows
two-step swelling and polymerization method. both the imprinting and the chiral separation of free
amino acids to be carried out in aqueous solutions.
The recognition was based on metal coordination}
chelation interactions using N-(4-vinylbenzyl)imino-
diacetic acid as the functional monomer. The method
worked best for aromatic amino acids (Figure 7).
TLC
MIPs selective for phenylalanine anilide have been
evaluated as stationary phases in TLC. Glass backing
plates were coated with mixtures of Rnely ground
MIP particles and binders. The plates showed prefer-
ential retardation of the enantiomer used as the print
molecule (Figure 8). The RF values of both enantio-
mers on nonimprinted polymers were higher than
those observed on the imprinted polymers. The separ-
ations suffered from spot broadening, which was at-
tributed to the heterogeneity of the recognition sites.
Figure 9A shows the separation of racemic prop-

ranolol by CEC on a capillary column Rlled with
poly(methacrylic acid-co-TRIM), polymerized in situ
using (R)-propanolol as the print molecule. Several
attractive features make this system look promising
for the future: very low consumption of the print
molecule (in this case only 10 g), fast preparation of
the capillaries (3 h) and fast separation (less than
2 min). In Figure 9B, MIP particles selective for (S)-
propanolol were added to the mobile phase in a CE
separation.
Future Developments
Molecular imprinting is a technique which has great
potential. MIPs have found many applications, and
many more are likely to be developed. One of the Rrst
applications investigated was CSPs, the subject of this
chapter. A number of polymer systems have been
developed and these have been used to imprint differ-
ent classes of compounds.
To be able to attribute the binding of an MIP to an
imprinting effect it is of the utmost importance to
show that speciRc recognition sites were formed due
to the presence of the print molecules during the
polymerization. This is done by comparison with
Figure 7 Separation of 17 g of racemic phenylalanine on
appropriate reference polymers. Polymers prepared
poly (Cu(II)[N -(4-vinylbenzyl)]iminodiacetate-co-EDMA) grafted without print molecules are not always the best
to silica particles (4.6;50 mm column). The polymer was im- choice, since the physical properties (surface area,
printed with D-phenylalanine. Isocratic elution at 1 mL min\1,
503C with 1.5 mmol L\1 glycine in water. (Adapted from
Vidyasankar S, Ru M and Arnold FH (1997) Journal of
Chromatography A 775: 51}63, 1997, with permission from
Elsevier Science, UK.)
The technique is attractive because it is simple, fast

and allows multiple analyses in the same run.
CE and CEC
Chiral separation by CE and CEC is achieved using
a chiral selector which is either free in the mobile
phase or immobilized to the stationary phase. MIPs
can be used as the selector in both approaches. Mo-
lecularly imprinted capillary columns have been pre-
pared by various approaches:
1. packing performed MIP particles into the capillar-
ies
2. dispersion polymerization in situ in the capillaries
3. incorporating preformed MIP particles into poly- Figure 8 Separation of L- and D-phenylalanine anilide on TLC
acrylamide gels plates covered with poly (methacrylic acid-co -EDMA) imprinted
with (A) L-phenylalanine anilide; (B) D-phenylalanine anilide and
4. in situ polymerization on the inner walls of the
(C) no print molecule. Elution with CH3CN}HOAc (99 : 5).
capillaries (Adapted from Kriz D, Berggren Kriz C, Andersson LI and Mos-
5. Rlling the capillaries with a monolithic polymer bach K (1994) Analytical Chemistry 66: 2636}2639, 1994, with
with continuous pores by in situ polymerization. permission from the American Chemical Society, USA.)
Figure 9 (A) Capillary electrochromatography (CEC). Separation of racemic propranolol on a poly (methacrylic acid-co -TRIM) CSP
(75 m;350 mm capillary column) imprinted with (R )-propranolol. The sample was injected electrokinetically (5 kV, 3 s) and was
separated at a constant voltage of 30 kV. The electrolyte was CH3CN}acetate buffer (4 mol L\1, pH 3.0) (8 : 2). Detection at 214 nm.
The capillary was thermostated to 603C and an overpressure of 7 bar was applied. (Adapted from Schweitz L, Andersson LI and
Nilsson S (1997) Analytical Chemistry 69: 1179}1183, 1997, with permission from the American Chemical Society, USA.) (B)
Capillary electrophoresis (CE). Separation of racemic propranolol using 0.05% (w/v) poly (N -acryloylalanine-co-EDMA) particles
imprinted with (S )-propranolol as a chiral additive in the background electrolyte (100 m;470 mm capillary column). The sample was
injected by a 3 s pressure injection and was separated at a constatnt voltage of 15 kV. The electrolyte was 5 mmol L\1 phosphate
buffer, pH 7.0. Detection at 210 nm. Temperature: 253C. (Adapted from Walshe M, Garcia E, Howarth J, Smyth MR and Kelly MT
(1997) Analytical Communications 34: 119}122, 1997, with permission from the Royal Society of Chemistry, UK.)
porosity, etc.) of these polymers are often different and PETRA instead of the bifunctional EDMA. These
from those of imprinted polymers. Reference poly- Rndings look promising for future developments of
mers prepared with the optical antipode or a racemic MIPs for semipreparative and preparative puriRca-
mixture as the print species are preferred. The selec- tions.
tivity will be reversed when using the optical antipo- Molecular imprinting is an expanding area attract-
de, and a racemic mixture will give a polymer in- ing an increasing number of scientists, in both indus-
capable of separating the two enantiomers (unless the try and academia. This is evidenced by the fact that
monomers are chiral). three-quarters of all papers on molecular imprinting
The goal of an endeavour involving chromato- were published during the last decade and one-third
graphic separation is to achieve the best possible during 1996}1997. The rapid development of this
performance with respect to selectivity, resolution, technique is likely to result in many breakthroughs
load capacity and analysis time. Much research effort within the next few years.
on MIPs has therefore focused on improving the
chromatographic performance. The use of monodis- See also: II/Affinity Separation: Imprint Polymers.
perse spherical beads instead of irregular particles
improves the efRciency and investigations in this di-
rection with MIPs have already given promising re- Further Reading
sults. Another issue that needs to be investigated Ansell RJ and Mosbach K (1996) Molecularly imprinted
further is the heterogeneity of the binding sites. polymers: new tools for biomedical science. Pharma-
A more homogeneous population of sites would im- ceutical. News 3: 16}20.
prove the chromatographic performance. The load Dickey FH (1949) The preparation of speciRc adsorbents.
capacity of MIPs has been shown to be improved by Proceedings of the National Academy of Science of USA
the use of trifunctional cross-linkers such as TRIM 35: 227}229. (seminal paper)
III / CHIRAL SEPARATIONS / Protein Stationary Phases 2397
Kempe M and Mosbach K (1995) Separation of amino ramanian G (ed.) A Practical Approach to Chiral Separ-
acids, peptides and proteins on molecularly imprinted ations by Liquid Chromatography, pp. 69}93. Wein-
stationary phases. Journal of Chromatography A 691: heim: VCH.
317}323. Vidyasankar S and Arnold FH (1995) Molecular
Kempe M and Mosbach K (1995) Molecular imprinting imprinting: selective materials for separations, sensors
used for chiral separations. Journal of Chromatography and catalysis. Current Opinions in Biotechnology 6:
A 694: 3}13. 218}224.
Mayes A and Mosbach K (1997) Molecularly imprinted Wulff G (1986) Molecular recognition in polymers pre-
polymers: useful materials for analytical chemistry. pared by imprinting with templates. In: Ford WT (ed.)
Trends in Analytical Chemistry 16: 321}331. Polymeric Reagents and Catalysts, pp. 186}230. Wash-
Mosbach K and RamstroK m O (1996) The emerging tech- ington, DC. American Chemical Society.
nique of molecular imprinting and its future impact on Wulff G (1995) Molecular imprinting in cross-linked ma-
biotechnology. BioTechnology 14: 163}169. terials with the aid of molecular templates } a way
Sellergren B (1994) Enantiomer separation using tailor- toward artiRcial antibodies. Angewandte Chemie Inter-
made phases prepared by molecular imprinting. In: Sub- national Edition English 34: 1812}1832.
Protein Stationary Phases

J. Haginaka, Mukogawa Women’s University, phases based on BSA, HSA, pepsin, AGP, CBH I,
Nishinomiya, Japan ovomucoid and avidin are now commercially avail-
able. Among those, AGP and ovomucoid-based sta-
tionary phases can separate a wide range of weakly
acidic, weakly basic and neutral racemates.
In the early 1950s, it was reported that the binding of The advantages of protein-based stationary phases
the enantiomers of an anionic azo-dye to bovine generally include the use of an aqueous mobile phase,
serum albumin (BSA) or human serum albumin enantioselectivity for a wide range of compounds and
(HSA) was different. Further, it was reported that direct analysis without derivatization. The disadvan-
L-tryptophan was bound to serum albumin (bovine
tages included low capacity, lack of ruggedness and
mercaptoalbumin) more strongly than the D-form. limited understanding of the chiral recognition mech-
These studies clearly indicated the possibility of enan- anism. Thus, the protein-based stationary phases
tioselective binding of ligands to proteins. In 1973, are useful for analytical purposes, but are not gener-
BSA-Sepharose was used for the separation of tryp- ally applicable to preparative isolation. To stabilize
tophan enantiomers; this is the Rrst report on the use the protein-based stationary phases, chemical
of a protein stationary phase for chiral resolution modiRcation of the side chains of the amino acids
purposes. With this stationary phase, D- and L-tryp- in the protein has been tried. Further, chiral
tophan were clearly resolved, and the D-form was stationary phases based on a protein fragment or
eluted Rrst, as shown in the previous binding studies protein domain have been prepared. These can be
in solution. In the following years, high performance of higher capacity because only the active protein
liquid chromatography (HPLC) chiral stationary mass is used. Also, it is possible to understand
phases based on a protein were developed and used the chiral recognition sites of protein-based station-
to separate a variety of enantiomers. HPLC chiral ary phases by investigating whether or not indepen-
stationary phases based on a protein are of special dent chiral binding sites exist on each fragment or
interest because of their unique properties of domain.
stereoselectivity and because they are suited for separ- This article deals with the preparation of HPLC
ating a wide range of enantiomeric mixtures. Protein- chiral stationary phases based on a protein, their
based stationary phases developed so far have in- chiral recognition properties and the chiral recogni-
cluded albumins such as BSA and HSA, enzymes such tion mechanisms of these stationary phases.
as trypsin, -chymotrypsin, lysozyme and pepsin, and
glycoproteins such as 1-acid glycoprotein (AGP)
from human or bovine serum, cellobiohydrolase Preparation of HPLC Chiral Stationary
I (CBH I), ovomucoid (in fact, ovoglycoprotein),
avidin, ovotransferrin and Savoprotein (riboSavin-
Phases Based on a Protein
binding protein). The physical properties of these Generally, a protein is bound to derivatized silica
proteins are shown in Table 1. Chiral stationary gels. The disadvantage of silica-based stationary
2398 III / CHIRAL SEPARATIONS / Protein Stationary Phases
Table 1 Physical properties of proteins
Protein Molecular mass Carbohydrate Isoelectric Origin

(kDa) composition (%) point
Albumins
BSA 66 4.7 Bovine serum
HSA 66 4.7 Human serum
Enzymes
Trypsin 24 10.1 Bovine pancreas
-Chymotrypsin 25 8.1}8.6 Bovine pancreas
Pepsin 34 (1 Porcine stomach
Lysozyme 14 10.5}11.0 Egg white
Glycoproteins
1-Acid glycoprotein 44 45 2.7 Human or bovine
serum
Cellobiohydrolase I 60 6 3.6 Fungus
Ovoglycoprotein 30 25 4.1 Egg white
Avidin 68 7 10.0 Egg white
Ovotransferrin 77 2.6 6.1 Egg white
Flavoprotein 32}36 14 4 Egg white or yolk
phases is that the eluent pH is limited to the ranges adsorbed protein can be eluted, and it is better to bind
2}8. However, in a strong acidic or alkaline solution, a protein covalently to the base materials in order to
a protein sometimes suffers from denaturation. The avoid losses. The chiral recognition properties of
separation of enantiomers on a protein-based station- a bound or adsorbed protein may be different from
ary phase is generally attained using an eluent whose those of the protein in solution because of blocking of
pH is between 3 and 8. Thus, the limitation of eluent functional groups and/or conformational changes.
pH ranges originated from silica-based materials is no The bound protein is often more stable to the changes
problem for the use of protein-based stationary of eluent pH and eluent composition compared to the
phases. Figure 1 shows the typical preparation protein in solution.
method for protein-based HPLC chiral stationary
phases; in part A the method includes activation of
porous aminopropylsilica gels by N,N-disuc- Retention and Enantioselectivity of
cinimidylcarbonate (DSC), binding of a protein and Solutes on Protein-based Stationary
blocking of the activated amino groups. In this case, Phases, and Optimization of
a side chain amino group(s) of a protein such as lysine
and arginine and/or an N-terminal amino group
Resolution
could be used for binding the protein to the activated Table 2 shows the inSuence of eluent pH on the
gel. On the other hand, using water-soluble carbo- retention and enantioselectivity of various solutes on
diimide and N-hydroxysulfosuccinimide (HSSI), the AGP-based chiral stationary phases. The retention
carboxyl group of a protein can be bound to aminop- factor of a basic solute, metoprolol, increased with an
ropylsilica gels, as shown in Figure 1B. increase in the eluent pH. The decrease in the reten-
Further, proteins can be bound to aminopropyl- tion factor of an acidic solute, 2-phenoxypropionic
silica gels using glutaraldehyde as a cross-linker, re- acid, is ascribable to ion exclusion in addition to ionic
sulting in cross-linking by Schiff-base formation. The repulsion between the carboxyl group of the 2-
resulted imino functions are reduced by using sodium phenoxypropionic acid and the negatively charged
cyanoborohydride. Glycerylpropylsilica gels ac- AGP with an increase in eluent pH. The retention of
tivated with 1,1-carbonyldiimidazole have been used basic solutes should be due to electrostatic interac-
for the preparation of chiral stationary phases. Ac- tions with the positively charged solutes and the nega-
cording to the two methods described above, an tively charged protein, in addition to hydrophobic
amino group of a protein is used to bind to the interactions. Although the retention factor of an
derivatized silica gels. In addition, a protein can uncharged solute (ethotoin, hexobarbital) shows al-
be physically adsorbed on to porous silica gels. The most no pH dependence, a slight increase is observed
disadvantage of the adsorption method is that the with increasing eluent pH. This increase might be
Figure 1 Synthesis scheme for the preparation of protein-based stationary phases: (A) via an amino group of a protein; (B) via
a carboxyl group of a protein.
due to changes in the binding properties of the pro- retention and enantioselectivity of various solutes
tein resulting from conformational changes. Table 3 on AGP-based chiral stationary phases. With an in-
shows the inSuence of the 2-propanol content on crease in the 2-propanol content, the retention and
Table 2 Influence of eluent pH on retention of enantioselectivity of various solutes on

AGP-based chiral stationary phase
Solute pH 4.5 pH 5.5 pH 6.5 pH 7.5
k1 k1 k1 k1
2-Phenoxypropionic acid 8.55 1.59 1.77 1.57 0.32 1.48

Ethotoin 4.06 3.82 3.87 4.19 3.82 4.59 4.13 5.06
Metoprolol 0.40 1.25 2.20 1.29 9.23 1.42 22.5 1.48
Hexobarbital 9.39 1.44 9.47 1.47 10.3 1.66 11.6 2.10
Mobile phase, 0.01 mol L\1 phosphate buffer; k1 is the retention factor of the first eluted
enantiomer; is enantio separation factor"k2/k1, where k2 is the retention factor of the
second eluted enantiomer. (Reproduced with permission from Hermansson J (1989)
Enantiomeric separation of drugs and related compounds based on their interaction with
1-acid glycoprotein. Trends in Analytical Chemistry 8: 251.)
Table 3 Influence of 2-propanol on the retentivity and enantioselectivity of various

solutes on AGP-based chiral stationary phase
Solute 2-PrOH (%)
1 2 4 6 8
k1 k1 k1 k1 k1
Disopyramide 8.51 3.70 3.62 3.37 1.77 3.20

Chlorpheniramine 11.2 2.34 7.35 1.71 4.59 1.38
Mepensolate 6.35 1.54 2.65 1.40 1.42 1.38 1.00 1.21
Mepivacaine 26.0 1.36 10.7 1.31 4.42 1.33 2.48 1.36 1.58 1.35
Bupivacaine 18.6 1.70 8.84 1.72 5.01 1.74
Mobile phase, 2-propanol in phosphate buffer, pH 7.2; k1, retention factor of the first eluted
enamtiomer; , enantioseparation factor"k2/k1, where k2 is the retention factor of the
second eluted enantiomer. (Reproduced with permission from Hermansson J (1989)
Enantiomeric separation of drugs and related compounds based on their interaction with
1-acid glycoprotein. Trends in Analytical Chemistry 8: 251.)
enantioselectivity of solutes are decreased. These re- phases. As described above, hydrophobic, electro-
sults suggest that hydrophobic and electrostatic inter- static and hydrogen bonding interactions play an im-
actions play an important role in the retention and portant role in chiral recognition of solutes on these
enantioselectivity of racemic solutes on AGP-based phases. Thus, enantioseparations of solutes can be
columns. Further, the hydrogen bonding properties of optimized by changing eluent pH, and the type and
the organic modiRer inSuence enantioselectivity to content of the uncharged organic modiRer. Some-
a large extent. As shown in Figure 2, verapamil enan- times, charged modiRers such as N,N-dimethyl-
tiomers are not resolved on the AGP-based column octylamine and octanoic acid are used for the enan-
using 1-propanol as an organic modiRer, but are tioseparation of a charged solute. Figure 3 shows a
resolved using acetonitrile. scheme for the optimization procedure for
Similar retentive and enantioselective properties ovomucoid-based stationary phases.
are observed with other protein-based stationary
Albumin-based Stationary Phases
BSA and HSA are closely related proteins and, conse-
quently, the chromatographic properties of the chiral
stationary phases based on these proteins are similar.
Sometimes the elution order is reversed between
chiral stationary phases based on these proteins; on
the HSA-based phases (S)-warfarin elutes before (R)-
warfarin, whereas on the BSA-based phases the oppo-
site elution order is observed.
A variety of weakly acidic and neutral compounds
are resolved on chiral stationary phases based on BSA
and HSA. 2-Arylpropionic acid derivatives such as
naproxen, Surbiprofen, ibuprofen, ketoprofen and
fenoprofen, reduced folates such as leucovorin and
5-methyltetrahydrofolate, and benzodiazepines such
as oxazepam, lorazepam and temazepam are separ-
Figure 2 Influence of the nature of the organic modifier on the ated. Figure 4 shows enantioseparations of leuco-
enantioselectivity of verapamil on an AGP-based column. HPLC vorin, lorazepam hemisuccinate and N-benzoyl-
conditions: column, Chiral-AGP (4.0 mm i.d.;100 mm); eluent: phenylalanine on an HSA-based column. However,
(A) 10% acetonitrile in 0.01 mol L\1 phosphate buffer, pH 7.0; (B) cationic compounds are not resolved on BSA and
4% 1-propanol in 0.01 mol L\1 phosphate buffer, pH 7.0. (Repro-
HSA phases. The structure-binding relationship for
duced with permission from Hermansson J (1989) Enantiomeric
separation of drugs and related compounds based on their inter- benzodiazepines using HSA stationary phases reveals
action with 1-acid glycoprotein. Trends in Analytical Chemistry 8: that the binding of benzodiazepines occurs at a site
251. that contains both a hydrophobic pocket and an area
Figure 3 Scheme for the optimization procedure for ovomucoid-based stationary phases. ACN, acetonitrile. (Reproduced with
permission from Kirkland KM and McCombs DA (1994) Changes in chiral selectivity with temperature with an ovomucoid protein-based
column. Journal of Chromatography A 666: 211.)
Figure 4 Enantioseparations of (A) leucovorin, (B) lorazepam hemisuccinate and (C) N-benzoyl-phenylalanine on an HSA-based
column. HPLC conditions: column, 4.6 mm i.d.;150 mm; eluent, 50 mmol L\1 phosphate buffer (pH 7.0): 1-propanol (94 : 6, v/v); flow
rate, 0.8 mL min\1. Peaks: 1,(6S )-leucovorin; 2, (6R )-leucovorin; 3, (!)-(R )-lorazepam hemisuccinate; 4, (#)-(S )-lorazepam
hemisuccinate; 5, N-benzoyl-D-phenylalanine; 6, N-benzoyl-L-phenylalanine. (Reproduced with permission from Domenici E, Bertucci
C, Salvadori P et al. (1990) Synthesis and chromatographic properties of an HPLC chiral stationary phase based upon human serum
albumin. Chromatographia 29: 170.)
of cationic charge, and that the chiral recognition

occurs in this binding site.
Enantioselectivity of stationary phases based on
BSA produced with isolated protein fragments has
been investigated. The BSA fragment following peptic
digest of BSA has molecular weights of about 35 kDa
which is an N-terminal half of amino acid residues
1}307. The BSA fragment phases give longer reten-
tions for benzoin and benzodiazepines, and higher
enantioselectivity for lorazepam, benzoin and feno-
profen because of a higher density of chiral recogni-
tion site(s), compared with native BSA phases.
Figure 5 shows chromatograms of lorazepam enan-
tiomers on BSA and BSA fragment-based columns.
However, it is plausible that the conformation of the
BSA fragment might be different from that of the
native BSA.
Enzyme-based Stationary Phase

Trypsin and -chymotrypsin are a family of serine
proteases. Trypsin-based stationary phases can re-
solve O-, N,O-derivatized amino acids which are
substrates of the enzyme. This means that chiral sep-
arations are due to the activity of the enzyme, and
that the chiral recognition site is on the enzyme activ-
ity site. -Chymotrypsin stationary phases can resolve
amino acids and amino acid derivatives.
When the eluent of pH 7 is continuously delivered,
the pepsin-based stationary phases lose their chiral
recognition properties. This result reveals that the
immobilized pepsin is irreversibly denatured above
pH 7. Thus, the use of an eluent with pH less than 6 is
Figure 6 Enantioseparations of (A) homochlorcyclizine,

(B) verapamil, (C) alprenolol and (D) oxazepam on a pepsin-
based column. HPLC conditions: column, 4.6 mm i.d.;100 mm;
eluent, 20 mmol L\1 phosphate buffer (pH 5.1) containing (A)
5 and (B) 10% acetonitrile and (C) and (D) 5% ethanol; flow rate,
0.8 mL min\1. (Reproduced with permission from Haginaka J,
Miyano Y, Saizen Y et al. (1995) Separation of enantiomers on
a pepsin-bonded column. Journal of Chromatography A 708:
Figure 5 Chromatograms of lorazepam enantiomers on (A) 161.)
BSA-based and (B) BSA-fragment-based columns. HPLC condi-
tions: column, 2.1 mm i.d.;100 mm; eluent, 50 mmol L\1 phos- recommended. By using a mixture of phosphate buf-
phate buffer (pH 7.5) containing 4% 1-propanol; flow rate,
fer and organic modiRer as an eluent, cationic (espe-
0.2 mL min\1. (Reproduced with permission from Haginaka J and
Kanasugi K (1995) Enantioselectivity of bovine serum albumin- cially -blocking agents) and neutral enantiomers are
bonded columns produced with isolated protein fragments. Jour- resolved, while no resolution of acidic enantiomers
nal of Chromatography A 694: 71.) is observed. Figure 6 shows enantioseparations of
based stationary phase, where the retention factor of

the second-eluted enantiomer is increased drastically.
This is due to an allosteric interaction in which the
afRnity of the protein for the enantiomer is increased
by the addition of the modiRer.
AGP consists of a protein domain and sugar moie-
ties, both of which have chiral components. It was
thought that drug binding to AGP occurred at a single
hydrophobic pocket or cleft within the protein
domain of the molecule. However, the role of sugar
moieties on enantioselective binding by AGP has not
been investigated. It has been reported that sialic acid
residues inSuence the enantioselective binding of
basic drugs in different ways. They are not involved
in the enantioselective verapamil}AGP binding. On
the other hand, they participate in the binding of
(S)-propranolol but not of (R)-propranolol. Further
Figure 7 Enantioseparations of (A) metoprolol and (B) pindolol
studies are required to clarify the role of sugar moie-
on an AGP-based column. HPLC conditions: column, Chiral-AGP
(4.0 mm i.d.;100 mm); eluent (A) 3.8% ethanol in 0.01 mol L\1 ties in chiral recognition and the chiral recognition
phosphate buffer, pH 7.0; (B) 15% methanol in 0.01 mol L\1 mechanism of AGP.
phosphate buffer, pH 7.0. (Reproduced with permission from
Hermansson J (1989) Enantiomeric separation of drugs and re-
lated compounds based on their interaction with 1-acid glycopro- Cellobiohydrolase-based Stationary
tein. Trends in Analytical Chemistry 8: 251.) Phases
Chiral stationary phases based on CBH I can resolve
homochlorcyclizine, verapamil, alprenolol and acidic and basic racemates into their enantiomers.
oxazepam on a pepsin-based column. Also, lysozyme- Higher enantioselectivity is especially obtained for
based stationary phases showed chiral recognition the separation of -blocking agents such as pro-
ability to cationic and neutral solutes but no resolu- pranolol, oxprenolol and metoprolol. Figure 8 shows
tion of anionic solutes. a chromatogram of propranolol enantiomers on a
CBH I-based column.
1-Acid Glycoprotein-based Stationary CBH I has a structural organization with a ter-
minal, 36 residue-long binding domain connected to
Phase the rest of the enzyme (i.e. the core) through a Sexible
AGP is the major plasma protein responsible for the arm. The interconnecting region is rich in serine,
protein binding of basic drugs. A variety of cationic, threonine and proline residues and is highly
anionic and uncharged compounds can be resolved glycosylated. The core is enzymatically active. CBH
using this stationary phase. Cationic compounds re- I is enzymatically degraded into two fragments, core
solved include -blockers such as alprenolol, ox-
prenolol, propranolol, metoprolol and pindolol and
histamine antagonists such as chlorpheniramine and
dimethindene. Figure 7 shows enantioseparations of
metoprolol and pindolol on an AGP-based column.
Anionic compounds resolved include 2-arylpropionic
acid derivatives such as fenoprofen, ibuprofen and
naproxen. As described above, separation of enantio-
mers is optimized by changing eluent pH, and the
type and content of the uncharged organic modiRer.
However, cationic modiRers such as N,N-dimethyl- Figure 8 Chromatogram of propranolol enantiomers on a CBH
octylamine and tetrabutylammonium bromide and I-based column. HPLC conditions: column, 5.0 mm i.d.;250 mm;
anionic modiRers such as octanoic acid and butylic eluent, sodium-acetate buffer pH 4.7, I"0.01, containing 0.5%
2-propanol; flow rate, 0.3 mL min\1; detection, 254 nm. (Repro-
acid can be used effectively with AGP-based mater-
duced with permission from Erlandsson P, Marle I, Hansson L et
ials. An ion-pairing modiRer, N,N-dimethyloctyl- al. (1990) Immobilized cellulase (CBH I) as a chiral stationary
amine, in the eluent gives a large increase in enan- phase for direct resolution of enantiomers. Journal of the Ameri-
tioselectivity of racemic acid, naproxen, on an AGP- can Chemical Society 112: 4573.)
and binding domain. Each fragment has been shown

to contain at least one enantioselective site for prop-
ranolol. The dominating enantioselective site for
propranolol and other solutes is located on the core,
the main part of the enzyme. The three-dimensional
structure of the active site of CBH I has been elucid-
ated by X-ray crystallography, and it has been shown
that the binding site is a tunnel with the dimensions
0.4;0.7;4 nm. There are seven acidic amino acid
residues, four tryptophan residues and also, tyrosine,
serine, threonine, arginine and histidine lining the Figure 9 Structure of U-80 413.
tunnel. This gives the prerequisite for obtaining stereo-
selective binding of a broad range of chiral solutes. combination of the Rrst and second domains, or the
second domain of the OMCHI gives chiral recogni-
Ovoglycoprotein-based Stationary tion ability.
These results suggest that three domains may be
Phases needed to work in concert for chiral recognition of
Chiral stationary phases based on chicken ovomucoid various solutes, because columns made with the
(OMCHI) from egg whites have been prepared, whole, intact OMTKY and OMCHI can resolve
which show chiral recognition abilities for a wide a wide range of weakly acidic, weakly basic and
range of cationic, anionic and uncharged compounds, neutral racemates. Recently, a new protein from
similar to AGP phases. However, the chiral recogni- chicken egg whites has been isolated and character-
tion ability of OMCHI comes from other glycopro- ized. It is termed ovoglycoprotein (OGCHI, which
teins, as described below. means ovoglycoprotein from chicken egg whites). It is
Ovomucoid from turkey egg whites (OMTKY) and found that 10% of OGCHI is included in crude
OMCHI, which exist as three tandem, independent OMCHI preparations. OMCHI and OGCHI col-
domains, have been isolated, puriRed and umns made from isolated pure proteins have been
characterized, and columns based on OMTKY and compared with regard to their chiral recognition abil-
OMCHI domains have been made to test their chiral ities. It is found that the pure OMCHI gives no chiral
recognition properties. The third domain of OMTKY recognition abilities, and that the pure OGCHI gives
and OMCHI consists of glycosylated (OMTKY3S better chiral recognition than those of the impure
and OMCHI3S) and unglycosylated domains OMCHI reported previously, as shown in Table 4.
(OMTKY3 and OMCHI3). The OMTKY3 and
OMTKY3S, and OMCHI3 and OMCHI3S are enan-
Table 4 Comparison of retention factor (k1), enantioselectivity
tioselective to at least two classes of compounds,
() and resolution (Rs) of various solutes on columns made with
benzodiazepines and 2-arylpropionic acid deriva- crude OMCHI and isolated OGCHI
tives. Glycosylation of the third domain does not
affect chiral recognition. The chiral recognition Compound Column
mechanism of the OMTKY3 has been elucidated us-
Crude OMCHI OGCHI
ing nuclear magnetic resonance measurements, mo-
lecular modelling and computational chemistry. The k1 Rs k1 Rs
selected binding model for each of the (R)- and (S)-
enantiomers of U-80 413 (whose structure is illus- Benzoin 2.50 2.71 6.06 11.4 3.18 10.1
Hexobarbital 0.35 1.00 1.52 1.29 0.83
trated in Figure 9), a 2-arylpropionic acid derivative, Alprenolol 2.53 1.12 0.31 15.9 1.13 0.84
with OMTKY3 shows similarities and differences in Propranolol 7.49 1.12 0.44 42.6 1.18 0.78
orientation and intermolecular interactions between Chlorpheniramine 1.03 2.05 3.00 5.42 2.27 5.89
the (R)- and (S)-enantiomers. The carboxyl groups of Ibuprofen 4.05 1.18 0.88 9.03 1.39 2.58
Ketoprofen 7.69 1.11 0.82 23.5 1.20 1.97
each enantiomer engage in electrostatic interactions
with the positive charge on arginine-21. The carbonyl All values were averages of three replicates. HPLC conditions:
group on U-80 413’s central ring shares a hydrogen column, 2.0 mm i.d. ; 100 mm; eluent, 20 mmol L\1 phosphate
bond with the NH# 3 group of lysine-34. The distin- buffer (pH 5.1)-ethanol 90 : 10 (v/v); column temperature, 253C;
guishing difference between the enantiomers is the flow rate, 0.2 mL min\1; detection, 220 nm. (Reproduced with
permission from Haginaka J, Seyama C and Kanasugi N (1995)
proximity of the phenyl group of the (R)-enantiomer The absence of chiral recognition ability in ovomucoid: ovoglycop-
and phenylalanine-53. However, neither the Rrst rotein-bonded HPLC stationary phases for chiral recognition.
nor the second domain of the OMTKY and a Analytical Chemistry 67: 2539.)
Figure 10 Enantioseparations of (A) ibuprofen, (B) ketoprofen and (C) flurbiprofen on an avidin-based column. HPLC conditions:
column, 4.6 mm i.d.;150 mm; eluent, 20 mmol L\1 potassium phosphate (pH 6.5) containing 6% ethanol; flow rate, 1.2 mL min\1.
(Reproduced with permission from Miwa T, Miyakawa T and Miyake Y (1988) Characteristics of an avidin-conjugated column in direct
liquid chromatographic resolution of racemic compounds. Journal of Chromatography 457: 227.)
Though only 10% OGCHI was included in the im- amino acids shorter. There are differences in the
pure OMCHI, the impure OMCHI-based column carbohydrate links between Savoproteins from egg
gave moderate chiral recognition (Table 4). This is yolks and whites. Chiral stationary phases based on
due to the fact that OGCHI is preferentially bound to Savoproteins from egg whites and yolks exhibit chiral
DSC-activated aminopropyl-silica gels compared recognition abilities for uncharged, anionic and
with OMCHI, despite similarity in their average mo- cationic compounds. There is no evidence whether
lecular masses (30 and 27 kDa, respectively). the drugs interact with the riboSavin-binding site or
not.
Other Protein-based Stationary
Phases Future Trends
A number of further minor protein-based stationary Chiral recognition sites of protein-based stationary
phases for HPLC, based on avidin, ovotransferrin and phases will be located and their chiral recognition
Savoprotein, have been described. Avidin, a basic mechanisms will be elucidated using X-ray crystallog-
protein, strongly binds biotin with an association raphy, 1H-nuclear magnetic resonance spectroscopy
constant of 1015 mol L\1. Avidin-based stationary and computational chemistry. Based on these Rnd-
phase shows excellent chiral recognition ability for ings, a protein, protein fragment or protein domain
2-arylpropionic acid derivatives such as ibuprofen, having chiral recognition abilities or a point-mutated
ketoprofen, Surbiprofen, pranoprofen and feno- protein can be overexpressed by genetic technologies.
profen. Figure 10 shows enantioseparations of In the future, we could make protein-based stationary
ibuprofen, ketoprofen and Surbiprofen on an avidin- phases, which have better chiral recognition abilities
based column. A biotin-bound avidin stationary and higher loadability, and are more stable than those
phase does not exhibit chiral recognition ability, so far prepared.
because the biotin modiRes the structure of the
avidin. See also: III/Chiral Separations: Cellulose and Cellu-
Ovotransferrin is labile to heat and acid, but it lose Derived Phases; Cyclodextrins and Other Inclusion
seems that it is stable to heat when combined with Complexation Approaches; Ion-Pair Chromatography;
metal ions such as iron, copper, manganese and Liquid Chromatography; Liquid Exchange Chromatogra-
zinc. Ovotransferrin is further stabilized by conjuga- phy; Molecular Imprints as Stationary Phases.
tion to silica gel as a chiral stationary phase, and
ovo-transferrin-based stationary phases have been Further Reading
used for the separation of a basic compound,
azelastine. Allenmark S (1991) Chromatographic Enantioseparation:
Methods and Applications, 2nd edn. Chichester: Ellis
Flavoprotein, the riboSavin-binding protein, in egg
Horwood.
whites and yolks has been introduced as a chiral Allenmark SG and Andersson S (1994) Proteins and pep-
stationary phase for HPLC. Egg white and yolk tides as chiral selectors in liquid chromatography. Jour-
Savoproteins appear to be the product of the same nal of Chromatography A 666: 167.
gene but to have undergone different post-transla- Haginaka J (1997) HPLC chiral stationary phases based on
tional modiRcations. The amino acid sequence is the a glycoprotein. Trends in Glycoscience and Glycotech-
same but the yolk Savoprotein is between 11 and 13 nology 9: 399.
2406 III / CHIRAL SEPARATIONS / Supercritical Fluid Chromatography
Krstulovic AM (ed.) (1989) Chiral Separations by HPLC: Stevenson D and Wilson ID (eds) (1988) Chiral Separ-
Applications to Pharmaceutical Compounds. Chiches- ations. New York: Plenum Press.
ter: Ellis Horwood. Wainer IW (1993) Drug Stereochemistry: Analytical
Lough WJ (ed.) (1989) Chiral Liquid Chromatography. Methods and Pharmacology, 2nd edn. New York: Mar-
London: Blackie. cel Dekker.
Narayanan SR (1992) Immobilized proteins as chromato- Zief M and Crane LJ (eds) (1988) Chromatographic Chiral
graphic supports for chiral resolution. Journal of Phar- Separations. New York: Marcel Dekker.
maceutical and Biomedical Analysis 10: 251.

N. Bargmann-Leyder and M. Caude, Laboratoire de problem represents the major limitation of this
Chimie Analytique (unite& de recherche associe& e au method. Moreover, this method has some disadvan-
CNRS) de l’Ecole Supe& rieure de Physique et Chimie tages: (1) the chiral reagent must be optically pure, or
Industrielles de Paris, Paris, France its optical purity has to be well known (otherwise,
A. TambuteH , Centre d’Etudes du Bouchet, Le Bouchet, poor accuracy will be achieved); (2) the derivatizing
France
reaction must be quick and quantitative; and (3) the
Copyright ^ 2000 Academic Press chromatographic behaviour of the derived dia-
stereomers should be suitable (easy separation, good
stability under the chromatographic conditions, ease
of recovery with absence of racemization during the
Introduction step leading to the initial enantiomers).
Great emphasis is currently placed on differences This was the method of choice before the develop-
in biological activities, potencies and toxicities of ment of chiral stationary phases (CSPs). It is still
enantiomeric pharmaceutical compounds. The applied, but usually in order to improve detection
US Food and Drug Administration (FDA) has re- limits. The method is not commonly used with
cently implemented regulations for the enantiomeric SFC although (S)-trolox methyl ether has been used
purity of enantiomeric drugs and chemicals. This has to derivatize chiral alcohols for attempted separat-
led to the development of chromatographic methods ion by GC and SFC with achiral systems. Using
for the enantiomeric resolution of racemates includ- this derivatizing method, several compounds were
ing gas chromatography, liquid chromatography, and successfully resolved by SFC but GC failed be-
more recently supercritical Suid chromatography cause of thermal decomposition of the ester
(SFC). derivatives.
The physicochemical properties of enantiomers are
the same except when they are placed in an asymmet-
ric environment. This can be obtained before the
Formation of Labile Diastereomers in
chromatographic column or within the column by the Mobile Phase
using a chiral derivatizing agent in the mobile phase This method generally consumes chiral reagent.
or by using a chiral stationary phase. Moreover, the major limitation concerns detection,
which must be compatible with the nature of the
Formation of Diastereomers by Using chiral reagent contained in the mobile phase. In
the case of preparative applications the limitation
a Pre-Column Derivatization is related to the recovery of the sample, which
In this method, the racemate is reacted with an must be separated from the chiral reagent. Although,
optically pure compound leading to formation of the optical purity of the reagent has no effect on
diastereomers. Owing to their different physioco- the accuracy of the results, it decreases the selectiv-
chemical properties, diastereomers can be resolved by ity of the method. One example of the use of SFC
using classical achiral mobile and stationary phases. in this way is the chiral separation of amino alcohols
This method can only be applied to molecules bearing using chiral ion pairing (Figure 1). In this case
reactive functions such as amines, acids and alcohols. SFC analysis time was signiRcantly less than that for
For preparative purposes, partial racemization can high performance liquid chromatography (HPLC)
occur when recovering the initial enantiomer. This separation.
III / CHIRAL SEPARATIONS / Supercritical Fluid Chromatography 2407
lator is somewhat difRcult to adjust precisely with

a capillary column. Packed columns provide greater
scope for applications, mainly due to the greater
number of chiral stationary phases commercially
available and their ease of use. Consequently only
SFC on chiral packed columns will be described here.
Packed Columns
CSPs designed for HPLC are widely used in SFC. This
is because separations are performed at room temper-
ature so that there is greater interaction energy
between CSP-racemate (and therefore higher selectiv-
ities) with fewer racemization problems. Packed-col-
umn SFC has the same advantages and this is why,
since 1985, this technique has been successfully ap-
plied to chiral separations. However, the lack of com-
mercial equipment for packed-column SFC has long
been a major problem in the development of the
technique; this handicap is now being overcome. The
advantages of SFC over HPLC include: faster analy-
sis, faster column equilibration, faster method devel-
opment and also reduced generation of hazardous
waste. It must also be underlined that SFC sometime
exhibits thermodynamic advantages over HPLC by
providing greater selectivity values (particularly with
natural polymer CSPs).
Chiral Stationary Phases

The Rrst commercial LC-CSP was described in 1981
Figure 1 Chiral separation of propranolol (A) and DPI 101}106 and more than 100 CSPs are now commercially avail-
(B) using ion pair SFC. Operating conditions: 100;4.6 mm i.d. able (Table 1). These CSPs can be divided into four
column packed with 5 m cyanopropyl-grafted silica (Brownlee
GS}GU); mobile phase: carbon dioxide/acetonitrile (80 : 20, v/v)
groups, depending on their chemical structure and the
containing 5;10\3 mol L\1 of triethylamine and 3.5 ; chiral recognition mechanisms involved.
10\2 mol L\1 of N-benzoxycarbonylglycyl-1-proline;pressure 250
bar; temperature: 213C. (Reproduced from Steuer W, Schindler Group I Group I CSPs are divided into two sub-
M, Schill G and Erni F (1988) Supercritical fluid chromatography
with ion-pairing modifiers. Separation of enantiomeric 1,2- groups. Brush-type CSPs (Pirkle-type and analogues,
aminoalcohols as diastereomeric ion pairs. Journal of constitute the Rrst subgroup (IA in Table 1). They are
Chromatography 447: 287}296, with permission from Elsevier the most amenable to scientiRc investigation because
Science.) they work as independent CSPs, since each chiral
graft operates independently in distinguishing the
solute enantiomers.
Use of Chiral Stationary Phases Ligand-exchange CSPs (subgroup IB in Table 1)
cannot be used in SFC because the formation of the
As in HPLC, SFC with chiral stationary phases, ternary complex, chiral selector}CuII}solute, takes
which was described for the Rrst time by Mourier and place almost exclusively in an aqueous medium.
colleagues, is the most powerful technique for the
separation of enantiomers. Group II This group contains cyclodextrin CSPs
and crown ether CSPs. Only the cyclodextrins have
Capillary Columns
been applied in SFC (Table 1).
The coupling of SFC-CSPs can be performed either
with capillary or packed columns. However, few ap- Group III The chiral selector is here a polymer,
plications have been described using capillary col- natural (as with amylose and cellulose) or synthetic
umns since the number of commercially available (as with polyacrylamide) bearing a lot of stereogenic
GC-CSPs is low and setting up the back-pressure regu- centres and asymmetric cavities. The formation of the
Table 1 Commercially available CSPs
Chiral selector Commercial name Supplier
Type IA
(R )- or (S )-(3,5-Dinitrobenzoyl)phenylglycine DNBPG B, R
ChiralDNBPG-C Ser
Sumichiral OA-2000 Sum
Sumichiral OA-2000S Sum
(R )- or (S )-N-(3,5-Dinitrobenzoyl)tyrosine n-butylamide ChyRoSine-A Sed
(S )-(S )-N-(3,5-dinitrobenzoyl)tyrosine [1-(1-naphthyl)-ethyl]amide ChyRoSine-AD Sed
(S )-(3,5-Dinitrobenzoyl)leucine DNBLeu B, R
ChiralDNBL-C Ser
(S )-(3,5-Dinitro-benzoyl)phenylalanine Chiraline SFCC
(R )--Methylbenzyl urea Supelcosil-LC-(R)-urea Sup
(R )- or (S )-N-(2-Naphthyl)alanine R
(S )--(1-Naphthyl)ethylamine Sumichiral OA-1000 Sum
(R )-Phenylglycine amide derivative and (S)-(4-(4-Chlorophenyl) isovaleric acid derivative Sumichiral OA-2100 Sum
(R )-Phenylglycine amide derivative
(1R-3R )-Chrysanthemic acid derivative Sumichiral OA-2200 Sum
(R )- or (S )-1-(3,5-Dinitrobenzoyl)naphthyl glycine Sumichiral OA-2500 Sum
Sumichiral OA-2500S Sum
(S )-Valine t-butyl urea Sumichiral OA-3000 Sum
(S )-(3,5-Dinitrobenzylurea)valine Sumichiral OA-3100 Sum
(S )-(3,5-Dinitrobenzylurea)tert-leucine Sumichiral OA-3200 Sum
(S )-Valine-(S )-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4000 Sum
(S )-Valine-(R )-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4100 Sum
(R )-Phenylglycine-(R)-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4200 Sum
(R )-Phenylglycine-(S)-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4300 Sum
(S )-Proline-(S )-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4400 Sum
(S )-Proline-(R )-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4500 Sum
(S )-t -Leucine-(S )-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4600 Sum
(S )-t -Leucine-(R )-[1-(1-naphthyl)ethyl]urea Sumichiral OA-4700 Sum
Tartric acid and 3,5-dinitrobenzylphenylethylamine Nucleosil Chiral-2 MN
Dimethyl N -3,5-dinitrobenzoyl--amino-2,2-dimethyl-4-pentyl phosphonate (R )--Burke 1 B, R
(S ,S )- or (R ,R )-1-[(3,5-Dinitrobenzoyl)amino) 2-allyl-1,2,3,4-tetrahydrophenanthrene (S ,S ) or (R ,R ) Whelk-O 1 B, R
N -3,5-Dinitrobenzoyl-3-amino-3 phenyl-2-(1,1-dimethylethyl)propanoate -GEM 1 B, R
Type IB
Silica-grafted amino acids (proline, valine, hydroxyproline) Chiral hydroxyCu Ser
Chiral proCu Ser
Chiral valCu Ser
Nucleosil Chiral-1 MN
Chiralgel L-prolinamide MN
Chiralgel L-valinamide MN
Chiralgel L-phenylalinamide MN
Chiralpak WM/WE D
Chiralpak MA (#) D
Accusphere JW
1,2-(2-Carboxymethylamino)-diphenyl ethanol Chiralpak WE D
Type IIA
-Cyclodextrin Cyclobond III A
-Cyclodextrin Cyclobond I A
Chiradex M
Chiral -dex Ser
-Cyclodextrin Cyclobond II A
Acetylated -cyclodextrin Cyclobond III Ac A
Acetylated -cyclodextrin Cyclobond I Ac A
-Cyclodextrin derived (S )-2-hydroxy-propyl Cyclobond I SP A
-Cyclodextrin derived 2-hydroxy-propyl (racemic) Cyclobond I RSP A
-Cyclodextrin derived (S )-[1-(1-naphthyl)ethyl]carbamate Cyclobond I SN A
-Cyclodextrin derived (R )-[1-(1-naphthyl)ethyl]carbamate Cyclobond I RN A
-Cyclodextrin derived [1-(1-naphthyl)ethyl]carbamate (rac) Cyclobond I RSN A
-Cyclodextrin derived 3,5-dimethylphenylcarbamate Cyclobond I DMP A
-Cyclodextrin derived 4-methylphenylcarbamate Cyclobond I PT A
Table 1 Continued
Chiral selector Commercial name Supplier
Type IIB
Grafted silica crown ether Crownpak CR(#) D
Type IIIA
Triacetylated microcrystalline cellulose (raw polymer) Cellulose triacetate M
Chiral triacel MN
Chiralcel CA-1 D
Cellulose triacetate Chiralcel OA D
Tribenzoate cellulose Chiralcel OB; OB-H D
Triphenylcarbamate cellulose Chiralcel OC D
Tri(3,5-dimethylphenyl)carbamate cellulose Chiralcel OD; OD-H D
Chiralcel OD-R D
(reversed-phase)
Type IIIA
Tri(4-chlorophenyl)carbamate cellulose Chiralcel OF D
Tri(4-methylphenyl)carbamate cellulose Chiralcel OG D
Tri(4-methylbenzoate)cellulose Chiralcel OJ D
Tricinnamate cellulose Chiralcel OK D
Tri(3,5-dimethylphenyl)carbamate amylose Chiralpak AD D
Tri-(R )-(1-phenylethyl)]carbamate amylose Chiralpak AS D
Type IIIB
Poly(N -1-acryloylphenylalanine ethylester) Chiraspher M
Poly(triphenylmethylmethacrylate) Chiralpak OT(#) D
Poly(2-pyridyl-diphenylmethylmethacrylate) Chiralpak OP(#) D
Type IV
Bovine serum albumin Resolvosil-BSA-7 MN
1-Glycoproteic acid Enantiopac LKB
Chiral-AGP CT
Human serum albumin Chiral protein 2 SFCC
Ovomucoide Ultron ES-OVM MM
Vancomycin Chirobiotic V A
Teicoplanin (macrocyclic antibiotics) Chirobiotic T A
Cellobiohydrolase (stable enzyme) Chiral CBH A
Suppliers : A, Astec; B, Baker; CT, ChromTech AB; D, Daicel; JW, J&W Scientific; LKB; M, Merck; MM, MAC-MOD Analytical; MN,
Macherey-Nagel; R, Regis; Sed, SEDERE; Ser, Serva; Sum, Sumitomo; Sup, Supelco.
solute-CSP complex involves inclusion of the solute in ter. This is because many compounds of pharmaceut-
the chiral cavities acting cooperatively. Group III ical interest contain a -donor group.
CSPs (Table 1) can be applied in SFC. CSPs derived from N-(3,5-dinitrobenzoyl)amino
acids are among the most widely used for enan-
Group IV Group IV contains protein and antibiotic- tiomeric separations of numerous compounds. The
grafted silica. As for the phases in sub-group Ib, these early commercialization of the well-known (R)-N-
CSPs cannot be used in SFC. (3,5-dinitrobenzoyl)phenylglycine-derived CSP ((R)-
DNBPG), designed by Pirkle and co-workers in 1980,
Applications and the easy and inexpensive preparation of this type
of CSP, has prompted many researchers to design new
The most interesting applications of SFC concern the -acid CSPs. Although the scope of applications of
type IA and III CSPs, and to a lesser extent type II these CSPs does not vary very much, all workers agree
CSPs. that small structural changes in the phases can have
signiRcant effects on the chromatographic behaviour.
Group IA
Our laboratories have been involved in the develop-
As a general rule, most applications concern the ment of CSPs derived from tyrosine. Among them,
brush-type CSPs having a -electron acceptor charac- a ‘broad-spectrum’ CSP has been marketed under the
registered name ChyRoSine-A and an improved ver- Valine-diamide phases have been used in SFC for
sion of this has been described. Their enantiomer- the enantioseparation of racemic N-4-nitroben-
recognition abilities have been evaluated both by LC zoylamino acid isopropyl esters. The enantioselectiv-
and SFC and the scope of applications including nu- ity in SFC was comparable to that in LC using the
merous racemates such as benzodiazepines, sulfox- mixture 2-propanol/n-hexane as mobile phase but
ides, phosphine oxides, lactams and -blockers dem- the time required for analysis was less than 5 min by
onstrated. SFC.
An anthrylamine derivative adsorbed onto porous As a general rule, the use of SFC does not improve
graphitic carbon has been used to separate two enantioselectivity for type I CSPs. The selectivities
commercial anti-inSammatory agents (ibuprofen and obtained in LC and SFC are identical, showing that
Surbiprofen) and a series of racemic tropic acid the chiral recognition mechanisms are the same for
derivatives. The enantioselective properties of this hexane and carbon dioxide. In this case, the advant-
material were compared with the corresponding sil- age of SFC over LC is of a kinetic nature, giving
ica-based CSP and it was concluded that the former higher efRciency per unit time and therefore faster
was more efRcient. analysis. Figure 2 illustrates the kinetic advantage of
-Basic CSPs, deriving from tyrosine and bearing SFC over LC by showing the separation of the enan-
two stereogenic centres, were designed and success- tiomers of Oxazepam on ChyRoSine-A in both LC
fully applied to the enantioseparation of pharmaceut- and SFC. At constant resolution, the analysis time by
ical compounds using SFC. Warfarin and ICI 176334 SFC is 6 min, and 24 min by LC. However, it must be
(a potential nonsteroidal antiandrogen used in the emphasized that in some cases different selectivities
treatment of prostate cancer) were baseline-resolved between LC and SFC are encountered.
on these phases without any prior derivatization step The Rrst of these cases concerns the noncon-
into 3,5-dinitrobenzoyl derivatives. Several -donor ventional separation of -acceptor solutes on -ac-
CSPs, with (R)-N-pivaloylnaphthylethylamide as the ceptor CSPs. In such a case, } charge transfer
chiral selector group, have been applied to SFC. interactions cannot take place during the chiral
Figure 2 LC (A) and SFC (B) separations of the enantiomers of oxazepam using a ChyRoSine-A CSP: a comparison of analysis time
at constant resolution (Rs"3.5). Operating conditions: 150;4.6 mm i.d. column packed with 5 m ChyRoSine-A CSP. LC: mobile
phase, hexane/ethanol (90 : 10); flow rate, 2 mL min\1. SubSFC: mobile phase, carbon dioxide/ethanol (92 : 8); flow rate,
4.5 mL min\1 at 03C; outlet pressure 200 bar. Temperature, 253C; UV detection at 229 nm. (Reproduced from Bargmann-Leyder N,
TambuteH A and Caude M (1992) ChiraliteH et chromatographie en phase supercritique. A review. Analysis 20: 189}200.)
Figure 3 Influence of the nature of the mobile phase on the resolution of N-(3,5-dinitrobenzoyl)phenylglycinol on (R )-DNBPG. LC
conditions: mobile phase, hexane/ethanol (85 : 15, v/v) (k (r )"3.9, k (s)"5.6) or hexane/chloroform (10 : 90, v/v) (k (r )"4.3,
k (s)"16.3); flow rate 2 mL min\1; temperature, 253C; UV detection at 254 nm. SubSFC conditions: mobile phase, carbon diox-
ide/ethanol (93 : 7, w/w); flow rate 4.5 mL min\1 at 03C; average column pressure, 200 bar; temperature, 253C; UV detection at
254 nm. (Reproduced from Macaudière P, Lienne M, Caude M, Rosset R and TambuteH A (1989) Resolution of -acid racemates on
-acid chiral stationary phases in normal-phase liquid and subcritical fluid chromatographic modes. A unique reversal of elution order on
changing the nature of the achiral modifier. Journal of Chromatography 467:357}372, with permission from Elsevier Science.)
recognition mechanism. This is why the main mecha- blockers was achieved on commercially available
nism may vary depending on the mobile phase com- ChyRoSine-A CSP and on its improved version,
position, sometimes resulting in a reversal of the whereas these solutes appear to be unresolved or
elution order. As shown in Figure 3, important dis- poorly resolved by normal-phase liquid chromatogra-
crepancies in the selectivity values are noted between phy (Figure 4; Table 2). The chromatographic behav-
hexane/ethanol in LC and the supercritical mobile iour (both in SFC and LC) of various propranolol
phase carbon dioxide/ethanol. The chromatographic analogues has been thoroughly studied and further
behaviour observed in SFC is somewhat similar to spectroscopic investigations carried out. Starting
that observed in LC with hexane/methylene chlor- from these data, detailed chiral recognition mecha-
ide/chloroform mobile phases. nisms have been proposed, based on molecular mod-
The second major exception concerns the separ- elling. The solute conformations are selected by tak-
ation of -blockers using ChyRoSine-A as CSP. Sur- ing into account the information provided by the
prisingly, the direct separation of a series of - 1
H NMR spectra and it appears that the solvating
a short packed column and a high mobile phase Sow

rate. The use of high speed chiral separations allows
a decrease in solvent consumption (CO2 and polar
modiRer), and by minimizing band broadening in the
column gives better detectability. As an example,
Figure 8 shows the enantioseparation of propranolol
and pindolol. These results again demonstrate the
kinetic superiority of the SFC over LC. Moreover, in
the case of -blockers, the better kinetics of SFC is
combined with enhanced thermodynamics owing to
the favoured chiral recognition provided by the con-
formation of the molecules in the presence of carbon
dioxide.
Finally, it should be noted that type I CSPs have
been successfully applied to preparative SFC.
Group II
The scope of applications of a given cyclodextrin is
determined by the goodness of the Rt between the
chiral cavity and the size of the solute to be resolved.
If the cavity is too large compared with the solute,
Figure 4 Comparative chromatograms of the resolution of there is no preferred orientation (and therefore no
propranolol on ChyRoSine-A CSP by LC (A) and SFC (B). Oper- selectivity); on the other hand, if the cavity is too
ating conditions: 150;4.6 mm i.d. column packed with 5 m
small, there is no solute inclusion. Most separations
ChyRoSine-A CSP. LC. mobile phase/hexane/ethanol containing
1% v/v of n-propylamine (95 : 5, v/v); flow rate 1 mL min\1. SFC: are therefore achieved using -cyclodextrin for which
mobile phase, carbon dioxide/ethanol containing 1% v/v of n- the internal diameter of the cavity (0.78 nm) is well
propylamine (90 : 10); flow rate, 4 mL min\1 at 03C, outlet pres- suited to naphthyl, biphenyl, benzoyl or cyclohexyl
sure 200 bar. Room temperature; UV detection at 224 nm. (Re- moieties present in numerous molecules of pharma-
produced with permission from Siret L, Bargmann N, TambuteH
ceutical interest. -Cyclodextrin is well adapted to
A and Caude M (1992) Direct enantiomeric separation of -
blockers on ChyRoSine-A by supercritical fluid chromatography: molecules bearing large substituents (such as pheno-
supercritical carbon dioxide as transient in situ derivatizing agent. barbital); -cyclodextrin is preferentially used for
Chirality 4: 252}262.) smaller molecules bearing a single aromatic group or
a small aliphatic chain.
Cyclodextrin CSPs have been used for SFC, al-
though they are, a priori, better adapted to the
effect of carbon dioxide induces a change in confor- separation of enantiomers in reversed-phase LC.
mation of propranolol (Figure 5). This change occurs In fact, in normal-phase liquid chromatography,
in the presence of carbon dioxide but only if the the hydrophobic solvent, e.g. hexane, chloroform,
solute bears both an amino proton and an ether etc., occupies the cyclodextrin cavity and cannot
function separated by three carbon atoms. Without easily be displaced by solutes. The average behaviour
carbon dioxide, (R)- and (S)-propranolol conformers of the column is then somewhat similar to that
have geometrical structures such that the chiral recog- of a diol column (almost no chiral resolution has
nition process is poor: the chiral centre of the solute been obtained using cyclodextrin phases in normal-
cannot develop stereoselective interactions with the phase liquid chromatography). The small size of
CSP and the interactions involved are the same for the carbon dioxide molecule means that it can be
both enantiomers (Figure 6). On the other hand, the displaced more easily than other apolar solvents such
conformation of propranolol in the presence of car- as hexane from the cyclodextrin cavities. Moreover,
bon dioxide is geometrically favourable to the chiral the carbon dioxide molecule exhibits an induced
discrimination. The conformations of the chiral sta- dipole moment, giving it a higher polarity than
tionary phase, (R)-solute, (S)-solute and their respect- hexane.
ive associations are shown in Figure 7. In this case, The use of polar modiRers induces a decrease in
the (R)-propranolol conformer involves higher energy retention (polar modiRer competes with the solute
interactions with (S)-CSP than the (S) conformer. and increases the solubility of the solute). In terms
High speed chiral separations (analysis duration of selectivity, all the polar modiRers that have been
(1.5 min) of -blockers have been achieved using used (methanol, ethanol, 1-butanol, 2-butanol,
Table 2 Chromatographic data for the resolution of propranolol and some analogues on ChyRoSine-A CSP by LC and SFC
Compounds LC SFC
% polar k2 % polar k2
modifier modifier
1 5 11.7 1.14 12 19.8 2.07
2 5 15.5 1 12 12.8 1.07
3 5 13.2 1 12 11.3 1
4 2.5 10.7 1 12 10.9 1.07
5 5 9.72 1.32 12 24.7 2.27
6 5 9.2 1 12 13.2 1.08
7 5 1.7 1.01 12 13.9 1.47

2.5 2.3 1.03
Operating conditions: column 150;4.6 mm i.d., UV detection 224 nm. LC: mobile phase hexane/ethanol containing 1% (v/v) of
n-propylamine, the percentage (v/v) of polar modifie r in hexane is indicated in the table; room temperature; flow rate 2 mL min\1. SFC:
mobile phase carbon dioxide/methanol containing 1% (v/v) of n-propylamine, the percentage (v/v) of polar modifier in CO2 is indicated
in the table; temperature 253C; average column pressure 180 bar; flow rate at 03C 4 mL min\1.
Figure 5 Change of the propranolol conformation induced by carbon dioxide. (A) Optimized structures of (R )- and (S )-propranolol
without CO2. The intramolecular hydrogen bonding is by an arrow. (B) Optimized structures of (R )- and (S )-propranolol with CO2. In
order to simplify the figure, only two molecules of carbon dioxide are illustrated. (Reproduced with permission from Bargmann-Leyder
N, Sella C, Bauer D, TambuteH A and Caude M (1995) Separation of -blockers using supercritical fluid chromatography: investigation of
the chiral recognition mechanism using molecular modelling. Analytical Chemistry 67: 952}958.)
2-propanol) are almost equivalent. Water should not

be used since it decreases selectivity signiRcantly.
As reported by Macaudière and co-workers, the
use of cyclodextrin CSPs in SFC allows particular
selectivities to be obtained. A comparison between
solute behaviour in reversed-phase and in normal-
phase liquid chromatography clearly demonstrated
that SFC and reversed-phase liquid chromatography
are two complementary techniques; this result has
widened the range of cyclodextrin phase applications.
Figure 9 shows the comparison of the separation of
the 2-naphthyl and o-anisyl phosphine oxides on Cyc-
lobond I in normal-phase LC and SFC. No or very
weak enantiomeric resolution is achieved using
normal-phase LC.
Enantiomeric separation of a variety of drugs
and related compounds (ancymidol, coumachlor,
ibuprofen, mephenytoin, tropicamide, verapamil,
etc.) on an (S)-naphthylethylcarbamoylated--cyclo-
dextrin phase using sub- and supercritical Suid
Figure 6 Optimized association between the optimized struc- chromatography has been accomplished by Williams
tures of (R )-propranolol (without carbon dioxide) and ChyRoSine- and co-workers. Compounds previously resolved on
A CSP. (Reproduced with permission from Bargmann-Leyder N,
native or derivatized cyclodextrin CSPs in LC using
Sella C, Bauer D, TambuteH , A and Caude M (1995) Separation of
-blockers using supercritical fluid chromatography: investigation reversed-phase or polar organic mobile phases could
of the chiral recognition mechanism using molecular modelling. be resolved in SFC using a simple carbon diox-
Analytical Chemistry 67: 952}958.) ide/methanol eluent. Resolution of cromakalim was
Figure 7 Optimized associations between the optimized structures of (R )- and (S )-propranolol (with carbon dioxide) and ChyRoSine-
A CSP. (Reproduced with permission from Bargmann-Leyder N, Sella C, Bauer D, TambuteH A and Caude M (1995) Separation of -
blockers using supercritical fluid chromatography: investigation of the chiral recognition mechanism using molecular modelling.
Analytical Chemistry 67: 952}958.)
not obtained on the (S)-naphthylethylcarbamoylated-

-cyclodextrin CSP using LC, but was readily ac-
complished using SFC (Figure 10). The separation of
the enantiomers of N-(3,5-dinitrobenzoyl)valine
methyl ester, ancymidol and proglumide was also
obtained in a single run using carbon dioxide/meth-
anol eluent, whereas the same separations in LC re-
quired three different mobile phases.
Group III
Chiralcel-OD CSP tris(3,5-dimethylphenyl carba-
mate cellulose) has been used successfully for the SFC
enantioseparation of -blockers, potassium channel
activator analogues and other compounds. A Chiral-
pak-AD column, tris(3,5-dimethylphenyl carbamate
amylose), has been used to resolve enantiomeric mix-
tures of nonsteroidal anti-inSammatories.
Other CSPs derived from cellulose have been suc-
cessfully applied to the SFC enantioseparation of
compounds of pharmaceutical interest. For example,
an intermediate in the synthesis of a drug targeted for
cardiac arrhythmia was separated on Chiralcel-OB;
the four optical isomers of a new calcium channel
Figure 8 High-speed enantiomeric separation of (A) prop-
ranolol and (B) pindolol on ChyRoSine-A CSP. Operating condi- blocker, LF 2.0254, were resolved on Chiralcel-OJ;
tions: column, 50;3.2 mm i.d.; mobile phase, carbon diox- and some CSPs have been applied to the SFC separ-
ide/(ethanol containing 1% (v/v) of n-propylamine) (80 : 20, (v/v); ation of various frequently used drug racemates such
flow rate 7.5 mL min\1 at 03C; temperature, 273C; pressure, 220 as profens and barbiturate derivatives, benzo-
bar; UV detection at 224 nm. (Reproduced with permission from
diazepines, etc.
Bargmann-Leyder N, Thiebaut D and Vergne F et al. (1995) High
speed chiral separation of -blockers by supercritical fluid A Chiralcel-OD-H column and an achiral amino-
chromatography on ChyRoSine-A. Chromatographia 39: propyl column have been employed for the analysis of
673}681.) products formed in rat liver microsomal metabolism
Figure 10 SFC separation of the cromakalim enantiomers on

the (S)-naphthylethylcarbamoylated--cyclodextrin CSP. Operat-
ing conditions: mobile phase, carbon dioxide/methanol (96 : 4);
flow rate, 2 mL min\1; temperature 303C; pressure, 15 MPa; UV
detection at 254 nm. (Reproduced from Williams KL, Sander LC,
and Wise SA (1996) Comparison of liquid and supercritical fluid
chromatography using naphthylethylcarbamoylated--cyclodex-
trin chiral stationary phases. Journal of Chromatography A 746:
91}101, with permission from Elsevier Science.)
characteristics)). This coupling allowed the authors

to achieve baseline separations with all solutes
investigated, basic (-blockers, benzodiazepines)
Figure 9 Comparison of the separation of the 2-naphthyl and

o-anisyl phosphine oxides on Cyclobond I using LC and SFC. LC
conditions: mobile phase, hexane/ethanol (85 : 15, v/v); flow rate,
1 mL min\1; UV detection at 234 nm. SFC conditions: mobile
phase, carbon dioxide/methanol (94 : 6, w/w); flow rate,
4.5 mL min\1; temperature 253C; pressure, 150 bar; UV detection
at 234 nm. (Reproduced from Macaudière P, Caude M, Rosset
R and TambuteH A (1987) Resolution of racemic amides and
phosphine oxides on a -cyclodextrin-bonded stationary phase by
subcritical fluid chromatography. Journal of Chromatography
405:135}143, with permission from Elsevier Science.)
Figure 11 SFC separation of ibuprofen (1), fenoprofen (2),
clenbuterol (3), propranolol (4) and lorazepam (5) using the serial
coupling of different CSP columns. Operating conditions: col-
of racemic camazepam (a hypnotic/anxiolytic drug in umns, Chiralpak AD-Chiralcel OD-Chirex 3022; mobile phase,
clinical use) and the fast chiral separation of different carbon dioxide/methanol (0.5% triethylamine#0.5% trifluoro-
compounds (oxprenolol, pindolol, warfarin) has been acetic acid) with methanol programmed from 4% (5 min) to 30%
achieved by microbore SFC using a Chiralcel-OD at 5% min\1; flow rate 2 mL min\1; temperature, 253C; pressure,
200 bar. (Reproduced with permission from Kot A, Sandra P and
type stationary phase.
Venema A (1994) Sub- and supercritical fluid chromatography on
Kot and co-workers proposed the serial coupling packed columns a versatile tool for the enantioselective separ-
of different CSP columns (Chiralpak-AD, Chiralcel- ation of basic and acidic drugs. Journal of Chromatographic
OD and Chirex 3022 (brush-type with -donor Science 32: 423}448.)
and acidic (nonsteroidal anti-inSammatory drugs, performed. It appears that, contrary to what occurs
-agonists). As an example, Figure 11 shows the sep- for type I CSPs, important discrepancies in selectivity
aration of ibuprofen, fenoprofen, clenbuterol, prop- values may exist between LC and SFC. The system-
ranolol and lorazepam in a modiRer-programmed atic comparison of LC and SFC for Chiralcel-OD and
run. Chiralpak-AD CSPs demonstrates clearly that the
Systematic comparison of the chiral recognition presence of polar functional groups such as primary
mechanisms in LC and SFC for type III CSPs has been or secondary hydroxyl or amine functions may cause
Figure 12 Comparison of LC and SFC for the separation of mefloquine (A), viloxazine (B) and temazepam (C) using Chiralcel OD
CSP. Operating conditions: column, Chiralcel OD. LC, mobile phase hexane/ethanol containing 1% (v/v) of n-propylamine (90 : 10, v/v)
for (A) and (C), 50 : 50 (v/v) for (B); flow rate 1 mL min\1; room temperature. SFC: mobile phase, carbon dioxide/ethanol containing 1%
(v/v) of n-propylamine 90 : 10 (v/v) for (A) and (B) 95 : 5 (v/v) for (C); flow rate, 2 mL min\1; temperature, 253C; pressure: 200 bar. UV
detection. Separations are optimized for selectivity. (Reproduced with permission from Bargmann-Leyder N, TambuteH A and Caude
M (1995) A comparison LC-SFC for cellulose and amylose-derived chiral stationary phases. Chirality 7: 311}325.)
2418 III / CHIRAL SEPARATIONS / Synthetic Multiple Interaction (‘Pirkle’) Stationary Phases
large discrepancies in selectivity between LC and SFC instrumentation (full control over many
SFC. This result is peculiar to cellulose and amylose- chromatographic parameters and particularly full
derived CSPs, for which the interactions involved in control of the pressure). Analysts are looking for the
chiral recognition are not always well balanced. chiral column that is best able to achieve racemate
Therefore, in the case of chiral resolution of polar separation easily and in a single run. This objective
solutes, the analyst should try both LC and SFC so will probably never be achieved, but we can expect
that the more stereoselective one can be chosen. that the serial coupling of chiral columns (two or
Figure 12A}C show some examples of the different three) will allow some progress in this direction
selectivities that may exist between LC and SFC for owing to the kinetic advantage exhibited by SFC
polymer-type CSPs. over LC.
Other polymer-type CSPs have been used in SFC,
such as those based on polymethacrylates of helical
conformation and a polysiloxane CSP (polyWhelk- Further Reading
O), the ‘polymeric version’ of the commercially avail- ArieK ns EJ (1989) Racemates } an impediment in the use of
able brush-type CSP, Whelk-O 1. For the latter, the drugs and agrochemicals. In: AM Krstulovic (ed.) Chiral
comparison was performed between the polymeric Separations by HPLC: Applications to Pharmaceutical
CSPs and its brush-type analogue, and it appeared that Compounds. Chichester: Ellis Horwood. pp. 31}68.
the polyWhelk-O CSP affords greater enantioselectiv- Kot A, Sandra P and Venema A (1994) Sub- and supercriti-
ity and shorter retention under the same conditions. cal Suid chromatography on packed columns a versatile
tool for the enantioselective separation of basic and
acidic drugs. Journal of Chromatographic Science 32:
Conclusion 423}448.
Chiral separation is one of the Relds where SFC is Mourier P, Eliot E, Caude M, Rosset R and TambuteH A
recognized to have better characteristics than HPLC, (1985) Supercritical and subcritical Suid chromatogra-
both from a kinetic and sometimes thermodynamic phy on a chiral stationary phase for the resolution of
phosphine oxide enantiomers. Analytical Chemistry 57:
point of view.
2819}2823.
In general, SFC offers faster separations than LC Pirkle WH, House DW and Finn JM (1980) Broad spec-
and often better selectivity values (particularly with trum resolution of optical isomers using chiral high-
cellulosic and amylosic polymer-type chiral station- performance liquid chromatographic bonded phases.
ary phases, and also with brush-type CSPs in particu- Journal of Chromatography 192: 143}158.
lar cases). Consequently, SFC should be considered as Siret L, TambuteH A, Caude M and Rosset R (1991) Chiral
a powerful analytical tool for the separation of basic recognition mechanisms on chiral stationary phases de-
and acidic drugs. rived from tyrosine } SpeciRc inSuence of the nature of
Capillary columns should be chosen for the analy- the asymmetric centre vicinal functional group. Journal
sis of chiral compounds having a low or medium of Chromatography 540: 129}143.
polarity. On the other hand, packed columns are Wainer IW and Drayer DE (eds) (1988) Drug Stereochemis-
try: Analytical Methods and Pharmacology. New York:
preferred for analytes of high polarity for which a po-
Marcel Dekker.
lar modiRer must be added to the supercritical carbon Williams KL, Sander LC and Wise SA (1996) Use of naph-
dioxide mobile phase. thylethylcarbamoylated--cyclodextrin chiral stationary
Currently, to meet the requirements of quality con- phase for the separation of drug enantiomers and related
trol laboratories, most analyses are performed with compounds by sub- and supercritical Suid chromatog-
packed columns. This is mainly due to the progress in raphy. Chirality 8: 325}331.
Synthetic Multiple Interaction (‘Pirkle’) Stationary Phases
C. J. Welch, Merck & Co. Inc., Rahway, NJ, USA interaction (Pirkle-type, or brush-type) CSPs have
Copyright ^ 2000 Academic Press proven to be among the most useful for many liquid
chromatographic enantiomer separations. These
CSPs consist of an enantioenriched small molecule
selector immobilized on an inert chromatographic
Introduction support, typically silica gel. Separation is achieved
Among the many types of chiral stationary phases when the two enantiomers of the analyte are differen-
(CSPs) that have been developed, synthetic multiple tially adsorbed by the CSP (Figure 1). A combination
III / CHIRAL SEPARATIONS / Synthetic Multiple Interaction (‘Pirkle’) Stationary Phases 2419
Advantages of Brush-Type CSPs

In general, brush-type CSPs possess a number of ad-
vantages; since the selector is a small molecule, which
is often completely synthetic, a structure which con-
tains no labile or reactive components can usually be
developed and the mode of attachment of the selector
to the chromatographic support can be chosen for
durability. Brush-type CSPs are typically covalently
attached to the chromatographic support. Thus, most
brush-type phases are chemically robust and are gen-
erally quite long-lived. Longevity is desirable for an
analytical CSP, but truly essential for a preparative
Figure 1 Enantiomer separation on CSPs is made possible by CSP, where continuous operation for several years
formation of transient diastereomeric adsorbates with differing may be required. The chemical robustness of brush-
free energies. In this illustration, the analyte enantiomers are type CSPs results in the ability of these materials to be
depicted as right and left hands, and the CSP is depicted as used with a wide variety of mobile phases, which
immobilized right-handed gloves.
provides greater Sexibility in method development,
especially when poorly soluble analytes are being
of simultaneous, geometrically constrained, inter- investigated. The ability to utilize a variety of mobile
molecular interactions utilizing forces such as hydro- phases is of even greater importance in preparative
gen bonding, - attraction, ionic interactions, and chromatography, where compound solubility can
steric repulsion can result in diastereomeric adsor- drastically limit enantioseparation productivity.
bates with differing free energies, the prerequisite for An additional advantage of brush-type phases
enantioseparation. stems from the fact that the selectors which are used
Brush-type CSPs are just one of the several types of are typically small molecules with molecular mass of
CSPs that have been developed to date. Other types less than 11000 Da. Consequently, the selectors can
include CSPs based on natural polymers such as poly- be very densely arrayed on the chromatographic
saccharides or proteins, artiRcial polymers, and ‘im- surface, resulting in a phase which is highly resistant
printed’ polymers. Exactly what constitutes a brush- to sample overload and which has a very high pre-
type CSP is a matter of some debate. In the broadest parative capacity.
deRnition, CSPs based on immobilized cyclodextrins, Finally, most synthetic CSPs are available in either
antibiotics, etc. should perhaps be included in this enantiomeric form. Consequently, either elution
category. However, this article will deal primarily order (# before !, or ! before #) can be chosen.
with the synthetic CSPs developed by Professor Elution of the minor enantiomer before the major is
William H Pirkle. generally preferred in analysis, while elution of the
desired component before the undesired can greatly
increase productivity in preparative HPLC.
Background
Although Pirkle’s name is strongly associated with
the development of synthetic multiple interaction Survey of Some Important Pirkle-Type
CSPs, several important materials of this type were CSPs
developed by earlier researchers. Landmark events
The Beginnings: A Carbinol CSP
from the 1960s include the development of TAPA
CSPs and chiral GC stationary phases. The 1970s The Work on CSPs in the Pirkle laboratories grew out
brought a number of further advances, including of studies on 1H NMR chiral solvating agents. Immo-
Davankov’s development of ligand exchange CSPs, bilization of an analogue of the enantiopure chiral
Baczuk’s pioneering development of a synthetic CSP resolving reagent, triSuoromethyl-9-anthryl carbinol,
designed speciRcally for the chromatographic separ- afforded a CSP capable of separating the enantiomers
ation of the enantiomers of a particular analyte mol- of electron-deRcient aromatic sulfoxides and many
ecule (DOPA), and Cram’s development of chiral other racemates. Further studies showed that 3,5-
crown ether CSPs. During this same decade, Pirkle’s dinitrobenzamide (DNB) derivatives of the enantio-
investigations into the development of 1H NMR mers of a variety of amines, amino alcohols, amino
chiral solvating agents led to the development of acids and related compounds were separated with the
several useful brush-type CSPs. following CSP structure [I].
Structure [III] (DNB PG-Ionic)

Structure [I]
Principle of Reciprocity, and its Importance in CSP

Development: Development of DNB Amino Acid
CSPs
The observation that the enantiomers of some DNB
derivatives are well resolved on CSP structure [I]
prompted an investigation of the reciprocal situation.
As a general rule, if a CSP is useful for separating the
two enantiomers of an analyte, then a second CSP Structure [IV] (DNB Leu)
derived from one of the enantiomers of that analyte
should be useful for separating the enantiomers of
compounds which are structurally related to the
chiral selector of the original CSP. This concept is
known as the ‘principle of reciprocity’, and has been
of great use in the development of new CSPs by Pirkle
and his co-workers. The principle holds true gener-
ally, although the manner in which the chiral selector
is tethered to the silica surface can inSuence chiral Structure [V] (DNB Leu-Ionic)
recognition in either a beneRcial or detrimental
fashion. After Rnding that several DNB phenyl- subsequently prepared and also found to be useful for
glycine-derived analytes were particularly well re- the resolution of the enantiomers of a variety of
solved on CSP structure [I], Pirkle and co-workers arylalkylcarbinols, with the phenylglycine-derived
prepared and evaluated several phases derived from CSP structure [II] generally performing better than
DNB phenylglycine and related compounds. The the leucine-derived CSP structure [IV]. CSP structure
ready availability of enantiopure (R)-phenylglycine, [III] was found to be widely useful for the separation
which is utilized in the commercial production of an of the enantiomers of racemates from a variety of
antibiotic, suggested that phenylglycine-derived CSPs functional group classes, and was shown to be useful
might provide an economical and convenient method for the gram-scale preparative resolution of ra-
for resolving the enantiomers of the highly useful cemates using automated preparative chrom-
chiral solvating agent, triSuoromethyl-9-anthryl car- atography. The general ability of CSP structure [III]
binol, and perhaps other alcohols as well. to resolve the enantiomers of a variety of analytes led
As expected, CSP structure [II] afforded resolution to its introduction as the Pirkle 1A CSP by Regis
of the enantiomers of the triSuoromethyl-9-anthryl Chemical Company in 1980 } the Rrst commercially
carbinol chiral solvating reagent as well as some re- available HPLC chiral stationary phase. Somewhat
lated compounds. The related ionically bonded later, CSP structures [II], [IV] and [V] were also made
amino acid-derived CSPs structures [III] and [V] were commercially available, and today a wide variety of
CSPs incorporating the highly useful DNB moiety
have been developed. Commercialization of these
early Pirkle phases provided researchers in a number
of Relds with tools for separating the enantiomers of
structurally diverse racemates. Surprisingly, the ioni-
cally tethered DNB amino acid CSP structures [III]
and [V], are quite long lived when used with relatively
non-polar mobile phases, although they are readily
degraded when highly polar eluents such as methanol
Structure [II] (DNB PG) or water are used.
CSPs based on DNB derivatives of the amino acid, N-Aryl Amino Acid-Derived CSPs
tyrosine, were subsequently developed by a team of The development of second generation Pirkle phases
French researchers including Caude, Rosset, and with electron-rich aromatic groups reached it’s zenith
TambuteH , and a CSP based on DNB naphthylglycine with the investigations by Pirkle and Pochapsky of
was developed by O) i and co-workers in Japan. The compounds derived from the enantiomers of the read-
3,5-dinitrobenzamide group has proven useful in the ily prepared electron-rich N-aryl amino acid deriva-
development of a number of additional phases, as will tives. This is perhaps the best understood chiral rec-
be described later. ognition system yet studied. In some preliminary
Second Generation Pirkle CSPs with Electron Rich studies, it was shown that N-aryl amino acid deriva-
Aromatic Groups tives were well resolved using the DNB Leu-Ionic CSP
structure [V]. Subsequently, the structural require-
Studies in the Pirkle laboratories and elsewhere ments for chiral recognition were deRned, and
showed that the DNB amino acid-derived CSPs were a chiral recognition rationale was proposed. The high
capable of separating the enantiomers of a wide var- degree of enantioselectivity observed in this system
iety of analytes possessing electron-rich aromatic ('10 in some cases) can be attributed to the inti-
groups. At this point, it was only natural to again turn mate association of the two components of the more
to the principle of reciprocity and prepare and evalu- stable diastereomeric adsorbate as illustrated later.
ate ‘reciprocal’ materials based on some of these This chiral recognition system has been studied at
structures. A number of reciprocal phases were pre- great length using spectroscopic as well as chromato-
pared and evaluated, although they lacked the requi- graphic techniques. An X-ray structure of a 1 : 1 com-
site generality required of a commercial CSP. Never- plex of soluble analogues of the compounds shows
theless, these studies revealed many important design essentially the same adsorbate structure as that ini-
features that would prove useful later. tially proposed based upon chromatographic and
The enantiomers of aryl hydantoins are well re- spectroscopic studies (structure [VI]):
solved on DNB amino acid-derived CSPs, and several A reciprocal N-(2-naphthyl)valine-derived station-
hydantoin-based CSPs were prepared to study the ary phase was prepared and shown to provide
mechanisms for chiral recognition of these types of unprecedented levels of resolution for a variety of
analytes and to probe the ability of this type of phase racemates containing electron-deRcient aromatic
to afford separation of various racemates. groups. The related alanine-derived CSP structure
Separation of the enantiomers of a number of aryl- [VII] was subsequently reported, and commercial-
substituted phthalides was studied using DNB amino ized, and has proven to be a valuable and widely used
acid-derived CSPs such as CSP structures [III] and research tool:
[V], and a detailed study of the effect of analyte Subsequent structural reRnements led to the prep-
structure on chromatographic performance led to aration and commercialization of the leucine-derived
some optimized structures for chiral recognition. One CSP structure [VIII], which generally affords signiR-
such compound was used to prepare a reciprocal aryl cantly greater enantioselectivity in the separation of
phthalide CSP which, as expected, showed a general DNB-containing enantiomers than does CSP struc-
ability to resolve a number of racemates containing ture [VI]. The predominant effect seems to be a great-
electron-deRcient aromatic systems. ly diminished retention of the initially eluted enan-
Another class of compounds which are well re- tiomer, presumably owing to the ability of the more
solved on the DNB amino acid-derived CSPs are
general amide structures bearing at least one aromatic
ring. In an extensive series of studies, Pirkle, Hyun
and co-workers prepared and evaluated more than 10
CSPs derived from various analogues of (l-naph-
thyl)ethylamine (-NEA) in which a variety of
different tether geometries and substituent patterns
were explored. Study of this group revealed a number
of interesting and useful principles of CSP design.
However, the CSPs themselves were somewhat
limited in that they were useful primarily for
separation of the enantiomers of analytes bear-
ing electron-deRcient aromatic groups. Similar
CSPs were prepared and studied by O) i and co-
workers. Structure [VI]
Structure [VII] (N2N-Ala CSP)
Structure [VIII] (N1N-Leu CSP)
bulky leucine sidechain to inhibit approach of the representing improvements upon the widely used
least retained enantiomer from the undesired face of DNB phenylglycine-derived CSP structure [II] have
the aromatic ring of the CSP. been developed in the Pirkle laboratories and sub-
These phases have proven to be widely useful for sequantly commercialized. The Rrst of these, the
separating the enantiomers of chiral amines, amino GEM 1 CSP (CSP structure [IX]), grew out of
alcohols, amino acids, etc., and are sometimes also studies of the enantiopurity analysis of some chiral
useful for separating the enantiomers of chiral alco- -lactams. The Rnding that some DNB -amino acid
hols, thiols, etc. However, analysis with these phases derivatives were well resolved led to the preparation
generally requires that the analyte be derivatized so as and evaluation of the DNB -amino acid-derived
to incorporate an electron-deRcient aromatic group. structure [IX]. An X-ray crystal structure of the selec-
Typically, derivatization reagents such as 3,5-di- tor used in CSP structure [IX] shows the t-butyl and
nitrobenzoyl chloride or 3,5-dinitrophenylisocyanate phenyl substituents projecting from opposite faces of
are used. Despite this somewhat bothersome need to the stationary phase. The bulky t-butyl substituent at
make derivatives, these CSPs remain quite useful and the 2-position is thought to play a major role in
show a tremendous generality. controlling the conformation of this phase, preven-
Recently, a group of CSPs based on proline anilides ting intramolecular hydrogen bonding, and restrict-
have been developed in the Pirkle laboratories. These ing access to one face. The GEM CSP is useful for
exhibit remarkable enantioselectivities for some DNB the separation of the enantiomers of anilide deriva-
derivatives. tives of aromatic carboxylic acids and N-protected
amino acids.
Improved DNB CSPs
As will be related later, the most general CSP yet
The design of CSPs is an evolutionary process so that developed in the Pirkle laboratories, the Whelk-O 1
continuing reRnements and improvements in both the (CSP structure [X]), was actually designed to separate
understanding of chiral recognition and the design of the enantiomers of one particular compound, nap-
such phases are constantly being made. Several CSPs, roxen. In addition to resolving the enantiomers of
Structure [IX] (-GEM I CSP)

A research programme directed at developing a

phase capable of separating the enantiomers of un-
derivatized propranolol enantiomers was under-
taken. A group of three CSPs were designed,
prepared, evaluated, and shown to separate the
enantiomers of propranolol with the elution order
predicted by the chiral recognition model. Several
other -blockers were similarly resolved, however,
the poor chromatographic efRciency of these
materials limited their practical utility.
In an effort to overcome this obstacle, a group of
Structure [X] (Whelk-O 1 CSP)
amino phosphonic acid-derived CSPs were prepared
and shown to provide improved resolution of prop-
naproxen and nearly all related nonsteroidal anti- ranolol. Consequently, CSP structure [XII] was pre-
inSammatory drugs (NSAIDs) this CSP is useful for pared, evaluated, and found to provide very good
resolving the enantiomers of analytes from a host of resolution for the enantiomers of propranolol and
functional group classes. a number of related -blockers. This stationary phase
A very useful DNB-containing CSP based upon was commercialized as the -Burke I CSP. Subsequent
diphenylethylenediamine has been prepared and in- minor improvements in the tethering chemistry led to
vestigated by Uray, Lindner and Maier. This CSP, an improved version commercialized as the -Burke II
commercialized as the ULMO CSP (structure [XI]), CSP (structure [XII]).
shows a general ability to separate the enantiomers of
Naproxen-Speci\c CSPs Using the Immobilized
a number of analytes containing electron-rich aro-
Guest Method
matic groups. In addition, it shows an ability to
separate the enantiomers of aryl carbinols which is As mentioned earlier, one of the most general Pirkle-
unsurpassed among Pirkle-type CSPs. type CSPs resulted from a study aimed at developing
Another improved DNB-containing CSP based on something which would be useful for separating the
the relatively inexpensive trans 1,2-diamino- enantiomers of the drug, naproxen. At the time of the
cyclohexane (DACH) was developed by the research inception of this project, separation of the enantio-
team of Gasparrini, Misiti, Villani, and co-workers; mers of naproxen and related NSAIDs with Pirkle-
this phase has never been commercialized, but has type CSPs generally required formation of deriva-
been studied extensively. tives. The Rnding that underivatized naproxen enan-
tiomers could be marginally resolved on the -GEM
-Blocker-Speci\c CSPs
1 CSP prompted an investigation into the develop-
In the mid-1980s, some of the principles of CSP ment of improved separations. Drawing on the famil-
design were becoming fairly well understood in the iar principle of reciprocity, a new approach termed
Pirkle laboratories, and attempts were made to design the ‘immobilized guest method’ was utilized in this
selectors for particular target racemates of signiRcant study. Two CSPs were prepared from enantiopure
scientiRc and economic importance. naproxen, and these were used to investigate the
For example, the enantiomers of the -blocker drug enantioseparation of a variety of different chiral
propranolol had been shown to be marginally re- analytes. This study led to an understanding of the
solved on the DNB phenylglycine-derived CSP struc- structural requirements for enantioselective naproxen
ture [II] as any of a variety of N-acylated derivatives. recognition, and a new chiral selector incorporating
Structure [XI] (ULMO CSP)

Structure [XII] (-Burke I CSP)
many of these key structural features was proposed. been developed at Regis Technologies. In this
This new selector was synthesized and found to be technique, libraries of CSPs are prepared on a 50
well separated on the immobilized guest naproxen milligram scale by solid phase synthesis on silica
CSPs. Larger scale synthesis, resolution, and immo- particles. The resulting libraries are then rapidly
bilization of the selector afforded a CSP which re- evaluated using a process which does not require
solved naproxen enantiomers with a very high degree packing into a column. This technique provides a use-
of enantioselectivity ("2.25). In addition, struc- ful tool for rapidly determining which CSP from
turally related NSAIDs such as ibuprofen, ketopro- a library of many hundreds shows the greatest enan-
fen, Surbiprofen, etc. are also resolved. Subsequent tioselectivity for a particular analyte. In addition,
mechanism-based structural modiRcations led to the some of the necessary structural requirements for
development and commercialization of the Whelk-O chiral recognition can be determined by comparing
1 CSP (CSP structure [X]) which affords improved the relative performance of various CSPs in the li-
resolution of NSAID enantiomers. In addition to re- brary. This information can, in turn, suggest further
solving the enantiomers of NSAIDs, the Whelk-O CSP libraries, which can be prepared and evaluated.
1 CSP has proven to be the most general CSP de- Ultimately, the process can lead to discovering CSPs
veloped to date in the Pirkle laboratories, as it is which display preparatively useful enantioselectivity
capable of resolving the enantiomers of racemates for given target racemates.
from a host of functional group classes.
A number of variations on the Whelk-O structure
have been developed in the Pirkle laboratories and at
Regis Technologies. For example, the same selector
immobilized via a trifunctional silane linkage has
been commercialized as the Whelk-O II CSP (struc-
ture [XIV]). This CSP is more resistant to selector
cleavage under harsh mobile phase conditions.
Another analogue of the Whelk-O CSP has proven Structure [XIII] (-Burke II CSP)
useful for separations utilizing supercritical or sub-
critical carbon dioxide as an eluent. This phase, com-
mercialized as the PolyWhelk-O CSP, is prepared by
Rrst immobilizing the Whelk-O chiral selector onto
a polysiloxane polymer, then covalently bonding the
resulting polymer to silica gel. The resulting material
often shows improved efRciency and enantioselectiv-
ity when operated in the SFC mode.
New Trends in Design and Development of
Pirkle-Type CSPs
The design and development of new CSPs continues
at an active pace in the Pirkle laboratories, with an Structure [XIV] (Whelk-O II CSP)
emphasis on using an array of analytical tools to
develop a mechanistic understanding of CSP behav-
iour. Developing CSPs which will be useful in the
preparative chromatographic separation of pharma-
ceutical enantiomers is another major research area.
A new method for microscale synthesis and evalu-
ation of libraries of Pirkle-type phases has recently Structure [XV]
Figure 2 Relative enantioselectivity seen by a library of 50 dipeptide DNB CSPs for the enantiomers of a test racemate.
For example, preparation of a library of 50 di- See also: III/Chiral Separations: Amino Acids and Deriv-
peptide DNB CSPs and screening for the ability to atives; Capillary Electrophoresis; Cellulose and Cellulose
separate the enantiomers of the test analyte shown Derived Phases; Chiral Derivatization; Countercurrent
(structure [XV]) afforded the results presented in Fig- Chromatography; Crystallization; Gas Chromatography;
Ion-Pair Chromatography; Ligand Exchange Chromatog-
ure 2, which suggest that bulky groups in the aa 2
raphy; Liquid Chromatography; Molecular Imprints as
position and hydrogen bonding groups in the aa 1 Stationary Phases; Protein Stationary Phases; Super-
position are advantageous. critical Fluid Chromatography; Thin-Layer (Planar)
Subsequent preparation and evaluation of such a Chromatography.
‘focused’ library revealed a number of new CSPs with
very high enantioselectivity for the target analyte.
One of the better performing members of this library Further Reading
was prepared on a larger scale, packed into a column,
and shown to be able to afford exceptionally high Allenmark S (1991) Chromatographic Enantioseparation:
productivity for the preparative separation of the Methods and Applications, 2nd edn, New York: Ellis
enantiomers of the racemic test analyte. This new Horwood.
approach to CSP development promises to be very Pirkle WH and Pochapsky TC (1986) A new, easily
accessible reciprocal chiral stationary phase for
useful for large-scale preparative chromatographic
the chromatographic separation of enantiomers.
enantioseparation, and a variety of structurally di- Journal of the American Chemistry Society 108:
verse CSP libraries are now being investigated by 352}354.
a number of different research groups. Pirkle WH, Finn JM, Hamper BC and Schreiner JL (1982)
American Chemistry Society Symposium Series No. 185.
Eliel and Otsuba (eds) ch. 18, 245.
Conclusion Pirkle WH, Welch CJ, Burke JA and Lamm B (1992) Ana-
A number of Pirkle-type CSPs are of great use in the lytical Proceedings 29: 225.
chromatographic separation of enantiomers. Several Pirkle WH, Welch CJ and Lamm B (1992) Design, synthesis
of these materials are commercially available and and evaluation of an improved enantioselective nap-
widely used in diverse research disciplines. An ever- roxen selector. Journal of Organic Chemistry 57:
3854}3860.
increasing understanding of the requirements for
Welch CJ (1994) Evolution of chiral stationary phase de-
chiral recognition and the design of useful CSPs sign in the Pirkle laboratories. Journal of Chromatogra-
is evident from a survey of Pirkle’s work over the phy 666: 3}26.
past two decades. The entry of other research Welch CJ, Bhat GA and Protopopova MN, Enantiomer 3:
groups into the area of CSP design suggests that there 463.
will be continued developments in this area in coming Welch CJ, Protopopova MN and Bhat GA, Enantiomer, 3,
years. 471.
2426 III / CHIRAL SEPARATIONS / Thin-Layer (Planar) Chromatography
L. Lepri and M. Del Bubba, University of Florence, away from each other and only a bimolecular interac-
Florence, France tion is generally possible with the optical antipodes to
Copyright ^ 2000 Academic Press be separated.
Introduction Cellulose
In chiral chromatography, two diastereomeric ad- The linear polysaccharide cellulose is composed of
ducts with different physicochemical properties are (#)-D-glucose units and its relative molecular mass
formed during elution. The adducts differ in their ranges from 2.5;105 to 1;106 or more. The long
stability (in chiral stationary phases, CSPs, or chiral chains are arranged on a partially crystalline Rbre
coated phases, CCPs) and/or in their interphase dis- structure and held together by numerous hydrogen
tribution ratio (in chiral mobile phases, CMPs). bonds between the hydroxyl groups. The hydrolysis
According to Dalgliesh, three active positions of of cellulose with 2.5 mol L\1 hydrochloric acid at c.
the selector must interact simultaneously with the 1003C removes amorphous material and yields a more
active positions of the enantiomer to reveal differ- crystalline polymer called ‘microcrystalline cellulose’
ences between optical antipodes. This is a sufRcient and marketed as Avicel by several companies.
condition for resolution to occur but it is not essen- The mechanism of chiral recognition is not yet
tial. Chiral discrimination may happen as a completely clariRed even though a signiRcant role is
result of hydrogen bonding and steric interactions, attributed to the cellulose structure and to the hydr-
making only one attractive force necessary in this oxyl groups, the protection of which with BrCN result-
type of chromatography. Moreover, the creation of ed in the loss of chiral recognition. The optical antipo-
speciRc chiral cavities in a polymer network (as in des resolved are highly polar, such as amino acids,
‘molecular imprinting’ techniques) could make it with multiple sites for hydrogen bond formation.
possible to base enantiomeric separations entirely
on steric Rt. Cellulose Derivatives
Among derivatized polysaccharides, cellulose triacet-
Chiral Stationary Phases and Chiral ate (CTA) is the most used stationary phase for the
resolution of racemic compounds by TLC.
Coated Phases The different Rt of the two enantiomers into the
Few chiral phases are used in TLC; one of the main laminae of the polymer leads to separation of the
reasons for this is that chiral stationary phases with optical antipodes which is mainly governed by the
very high UV background can only be used only with shape of the solutes (Sat molecules showing a better
Suorescent or coloured solutes. For example, amino- permeation into the cavities) and only to a minor
modiRed ready-to-use layers bonded or coated with extent by electrostatic interactions involving the func-
Pirkle-type selectors, such as N-(3,5-dinitrobenzoyl)- tional groups of the molecules. Hence the name ‘in-
L-leucine or R(!)--phenylglycine, are pale yellow clusion chromatography’. In addition, the chiral rec-
and strongly adsorb UV light. Another reason is the ognition of CTA depends strongly on its structure and
high price of most CSPs. the type of eluent and increases as the crystallinity of
The most widely used CSPs or CCPs are polysac- the polysaccharide is increased. Microcrystalline cel-
charides and their derivatives and silanized silica gel lulose triacetate (MCTA) can be prepared from
impregnated with an optically active copper (II) com- microcrystalline cellulose with almost complete
plex of derivatized hydroxyproline. The use of silica preservation of microcrystallinity. Usually, the type of
gel impregnated with chiral polar selectors, such as eluent and its composition are important for chiral
D-galacturonic acid, (#)-tartaric acid, (!)-brucine, recognition because these produce different swelling of
L-aspartic acid or a complex of copper (II) with L- MCTA, which in turns enables the separation of sol-
proline, should also be mentioned. utes of different sizes and characteristics.
In CSPs, owing to the nature of the polymeric The use of n-hexane}isopropanol mixtures resulted
structure, the simultaneous participation of several in unsatisfactory separations since extremely elon-
chiral sites or several polymeric chains is conceivable. gated spots are generally obtained. Aqueous}alcohol-
In CCPs, the chiral sites are distributed at the surface ic solutions have the opposite effect, giving rise to
or in the network of the achiral matrix relatively far round and compact spots.
III / CHIRAL SEPARATIONS / Thin-Layer (Planar) Chromatography 2427
This CSP is able to resolve a broad range of struc- serum albumin (BSA) and the macrocyclic antibiotic
turally different racemates. In general, more polar vancomycin as chiral agents.
molecules require a higher percentage of water than
hydrophobic compounds. Unmodi\ed and Modi\ed -Cyclodextrins
as Mobile-Phase Additives
Hydrophobic Silica Gel Impregnated with Copper (II)
Complex of (2S,4R,2RS )-N-(2-hydroxydodecyl)-4- Among the three cyclodextrins (, , ), only -CD
hydroxyproline and its derivatives have been used for successful res-
olution of various racemates by TLC. In an aqueous
The structure of the selector is shown in Figure 1.
solution -CD is represented as a truncated cone
Chiralplate and HPTLC-CHIR are the only precoated
with different sized mouths (0.60}0.65 nm); the
plates built up from such material. The chiral layer on
height of the cavity is 0.78 nm. The 2- and 3-hydroxyl
the latter plates is combined with a so-called ‘concen-
groups are oriented towards the outside and are re-
trating zone’. The sample to be separated is applied to
sponsible for the aqueous solubility properties of this
this small band and is transported with the solvent
oligosaccharide. The hydrogen atoms and glycosidic
front, forming a narrow band at the interface of the
oxygen groups are located inside the molecule, form-
two sections; consequently, a higher efRciency of the
ing the relatively hydrophobic cavity that interacts
separation process is obtained.
with organic optical isomers to form diastereomeric
Many racemates have been resolved on both layers
inclusion complexes. Under reversed-phase condi-
by ligand-exchange chromatography (LEC). The sep-
tions, the combination of hydrophobic and steric in-
arated enantiomers are those capable of forming
teractions with hydrogen bonding between the chiral
diastereomeric complexes of different stability with
solutes and the 2- and/or 3-hydroxyl groups may be
the metal ion and the chiral selector. The requirement
the cause of enantioselectivity.
of sufRcient stability of the ternary complex involves
Aqueous}organic solutions (i.e. methanol}water or
Rve-membered ring formation and compounds such
acetonitrile}water mixtures) are usually used as
as -amino and -hydroxy acids are the most suitable.
eluents. -CD is enantioselective in TLC only at
the high concentrations reached by adding large
Chiral Mobile Phases amounts of urea, which increases the aqueous solubil-
ity of -CD more than ten-fold. However, urea
CMPs permit the use of conventional stationary
tends to compete with solutes for the preferred loca-
phases and show fewer detection problems than CSPs
tion in the hydrophobic cavity, thus decreasing the
or CCPs. However, high cost chiral selectors (i.e.
separation factor. Chemical modiRcation with hy-
-cyclodextrin) are certainly not advisable for TLC.
droxypropyl, hydroxyethyl or methyl groups has
Enantiomer separations have been achieved using
been used to increase the solubility of -CD and its
chiral mobile phases in both normal and reversed-
complexes in water, eliminating the need to use urea.
phase chromatography. The Rrst technique employs
Optimization of the enantioselectivity can be
silica gel and, mostly, Diol F254 HPTLC plates
achieved by modifying the concentration and nature
(Merck) and, as chiral selectors, D-galacturonic acid,
of organic solvent, pH and buffer concentration of
N-carbobenzoxy(CBZ)-L-amino acids or peptides,
the eluent.
1R-(!)-ammonium-10-camphorsulfonate and 2-O-
[(R)-2-hydroxypropyl)]--cyclodextrin.
Bovine Serum Albumin as Mobile-Phase Modi\er
Most separations have been obtained by reversed-
phase chromatography on hydrophobic silica gel with BSA is a protein of relative molecular mass 66 210,
-cyclodextrin (-CD) and its derivatives, bovine consisting of 581 amino acids in a single chain. It is
a relatively acidic protein (isoelectric point 4.7), high-
ly soluble in water, but precipitates from high salt
solutions. At pH 7.0 its net charge is !18. Hydro-
phobic interactions strongly contribute to the afRnity
of organic solutes for BSA; simultaneous contribu-
tions exist from electrostatic interactions, steric ef-
fects, hydrogen bonding and charge-transfer pro-
cesses.
Mobile phases containing this chiral selector have
been employed for the resolution of a broad variety of
Figure 1 Structure of (2S,4R,2RS )-N-(2-hydroxydodecyl)-4- racemic compounds on silanized silica plates. Eluents
hydroxyproline. were prepared by dissolving BSA (Serva, Heidelberg,
FRG), fraction V, pH 5.2 or 7.0, in different buffer Thus far, 84 proteinogenic and nonproteinogenic
systems and then adding the desired amount of amino acids have been separated without derivatiz-
2-propanol. The results suggest the use of acidic ation using mostly methanol}water}acetonitrile
eluents to separate N-derivatized amino acids on (50 : 50 : 200, v/v/v) as eluent. Usually the D enan-
wettable RP-18W/UV254 (Macherey-Nagel) or tiomer is the more retained. Racemic serine shows
RP-18W/F254 (Merck) plates. On the other hand, low resolution, while threonine and basic amino acids
the resolution of free amino acids increases with have not been resolved as yet.
increasing eluent pH; in particular, the useful pH The separations of enantiomeric amino acids re-
range on SIL C18-50/UV254 layers (Macherey-Nagel) ported in Table 2 using a variety of chiral selectors
is 9}10. are very interesting since they also include the un-
A prerequisite for optical discrimination is the pres- resolved compounds mentioned above. Round, com-
ence of aromatic as well as polar groups in the solute. pact spots are generally obtained on silica gel plates
The existence of stereospeciRc binding sites on al- eluted with 2-O-[(R)-2-hydroxypropyl]--cyclodex-
bumin is well known (i.e. tryptophan and warfarin) trin solutions as deduced from R values which are

and it is believed that this binding occurs at a number equal to or higher than the value for all the amino
of relatively deRned regions. Two independent non- acids with the only exception of DL-citrulline
cooperative types of sites (chiral and achiral) coexist (R "0.94). Visualization is performed by spraying

on the protein; the retention mechanism is partially in the plates with a salicyladehyde solution (1.5 g) in
accord with that proposed on columns packed with 100 mL toluene and then heating at 503C for 10 min
immobilized BSA since in TLC albumin is used as (yellow spots).
a mobile-phase additive. Table 3 gives the performance of cellulose plates,
which are very effective in resolving racemates of
aromatic and basic amino acids. Home-made layers
Retention and Resolution Data of microcrystalline cellulose powder (commercially
In this article the contribution of chiral TLC to enan- available from Merck, and Fluka) can be obtained
tiomeric separations is surveyed, emphasizing the ver- with optimal homogeneity by spreading an aqueous
satility of the method but without discussing the suspension with about 25% chiral material. The
chiral recognition mechanism; this aspect has been plates are dried at room temperature and do not
examined elsewhere and in other parts of this Ency- require activation before use. Polar mixtures (i.e.
clopedia. ethanol}pyridine}water) are the best eluents since
they separate enantiomers as efRciently as an aqueous
solvent (i.e. 0.1 mol L\1 NaCl) but give rise to more
Amino Acids and Their Derivatives
compact spots.
A broad variety of racemic amino acids has already Chiralplates were very effective in resolving ra-
been resolved by chiral TLC. Table 1 summarizes cemates of N-alkyl, N-carbamyl and N-formyl amino
the analytical separations achieved in this Reld on acids and of several dipeptides (Table 4). It is worth
20 cm;20 cm Chiralplates (Cat. No. 811058, Ma- noting that the dipeptide with the C-terminal L-con-
cherey-Nagel; thickness 0.25 mm). The separations Rguration always has a higher retention than the one
can be easily transferred to the 10 cm;10 cm with the C-terminal D-conRguration. Some racemic
HPTLC-CHIR plates with concentrating zone dipeptides were also resolved on microcrystalline cel-
since they are precoated with the same chiral selector. lulose with pyridine}water (2 : 1 or 4 : 1) and on SIL
With eluents A, B and C, 2 L of a 1% solution of C18-50/UV254 plates using BSA as chiral mobile phase
the racemates in methanol or methanol}water and additive.
with eluent D, 2 L of a 0.5% solution of the ra- Derivatization of amino acids may be used to im-
cemates in 0.1 mol L\1 hydrochloric acid}methanol prove chiral recognition, detectability and sensitivity
1 : 1 were applied to the plates. Migration time in- of the method and to label amino acids residues of
creases from 0.5 h (eluent A) to 1.5 h (eluent C). peptides and proteins, especially the N-terminal
Detection was performed by dipping the plates for 3 s amino acid. Dimethylaminonaphthalenesulfonyl
in a 0.3% ninhydrin solution in acetone and then (dansyl) amino acids form when primary and second-
heating at 1103C for c. 5 min. Red spots appear on ary amino acids react with dansyl chloride, generat-
a white background. ing strongly Suorescent compounds. The best
The amount of solute applied to the plates chromatographic conditions for their separation are
(10}20 g) is an order of magnitude greater than that reported in Table 5. The most complete study,
generally employed in TLC; the use of HPTLC-CHIR performed on KC18 F plates (Whatman) using -CD
layers improves the sensitivity of the method. as chiral agent, concerns the racemates of 26
Table 1 Enantiomeric separation of proteinogenic and nonproteinogenic amino acids on Chiralplatesa
Racemate hRF1 b hRF2 c Eluent d
Ala 69(D) 73 1.22 D

Ser 73(D) 76 1.17 D
Abu 48 52 1.17 A
Val 54(D) 62 1.39 A
Nva 49(D) 56 1.32 A
Leu 53(D) 63 1.51 C
Ile 47(D) 58 1.55 A
Nle 53(D) 62 1.44 A
allo-Ile 51(D) 61 1.50 A
t-Leucine 40(D) 51 1.56 A
Met 54(D) 59 1.23 A
Eth 52(D) 59 1.32 A
Pro 41(D) 47 1.27 A
cis-Hype 41(L) 59 2.08 A
allo-Hyp 41(L) 59 2.08 A
Pipecolic acid 51 58 1.32 D
Phe 49(D) 59 1.49 A
Homophenylalanine 49(D) 58 1.43 A
Tyr 58(D) 66 1.40 A
Dopa 47(L) 58 1.55 B
Trp 51(D) 61 1.50 A
Asp 50(D) 55 1.22 A
Glu 54(D) 59 1.22 A
Gln 41(L) 55 1.76 A
Thyroxine 38(D) 49 1.56 A
PhenylGly 57(D) 67 1.53 A
CyclopentylGly 43 50 1.32 A
(1-Methylcyclopropyl)-Gly 49 57 1.38 A
(2-Thienyl)-Gly 55 66 1.58 A
3-Cyclopentyl-Ala 46 56 1.49 A
3-(1-Naphthyl)-Ala 49(D) 56 1.32 A
3-(2-Naphthyl)-Ala 44(D) 59 1.82 A
Met(O2)f 62(D) 66 1.18 A
Eth(O2) 55 59 1.18 A
Seleno-Met 53(D) 61 1.38 A
S-Methylthio-Cys 47(D) 55 1.38 A
S-Methylthio-HomoCys 44 52 1.38 A
S-(2-Chlorobenzyl)-Cys 45 58 1.68 A
S-(3-Thiobutyl)-Cys 53 64 1.58 A
S-(2-Thiopropyl)-Cys 53 64 1.58 A
O-Benzyl-Ser 54(D) 65 1.58 A
O-Benzyl-Tyr 48(D) 64 1.92 A
3,3-Dimethyl-Nva 40(D) 56 1.91 A
4-Methyl-Trp 50 58 1.38 A
4-Methoxy-Phe 52 64 1.64 A
5-Methoxy-Trp 55 66 1.58 A
3-Chloro-Ala 57 64 1.34 A
4-Amino-Phe 33 47 1.80 A
4-Bromo-Phe 44 58 1.75 A
4-Chloro-Phe 46 59 1.68 A
2-Fluoro-Phe 55 61 1.28 A
4-Iodo-Phe 45(D) 61 1.91 A
4-Nitro-Phe 52 61 1.44 A
3-Fluoro-Tyr 64 71 1.37 A
5-Bromo-Trp 46 58 1.61 A
-Methyl-Ser 56(L) 67 1.59 B
-Methyl-Abu 50 60 1.50 A
-Methyl-Val 51 56 1.22 A
Table 1 Continued
Racemate hR F1b hRF2 ac Eluent d
-Methyl-Leu 48 59 1.55 A
-Methyl-Met 56(D) 64 1.39 A
-Methyl-Phe 53(L) 66 1.72 A
-Methyl-Tyr 63(D) 70 1.37 A
-Methyl-Dopa 46(L) 66 2.27 B
-Methyl-Trp 54 65 1.58 A
-Methyl-Asp 52(D) 56 1.17 A
-Methyl}Glu 58(L) 62 1.18 A
-Ethyl-Ala 55 61 1.28 A
-Propyl-Ala 55 63 1.39 A
-Butyl-Ala 51 63 1.63 A
-Difluoromethyl-Phe 63 70 1.37 A
-Propenyl-Phe 57 63 1.28 A
2-Methyl-Phe 43(D) 54 1.55 A
-Methyl-Phe 36(R, R ) 56(S, S ) 2.26 A
2-Methyl--methyl-Phe 48(S, R ) 55(R, S ) 1.32 A
2-6-Dimethyl-Phe 38(D) 52 1.76 A
-Methyl-p-nitro-Phe 43(R, R ) 62(S, S ) 2.16 A
52(S, R ) 60(R, S ) 1.38 A
2,5-Dimethyl-4-methoxy-Phe 45(D) 57 1.61 A
-Hydroxy-Phe 49(R, R ) 63(S, S ) 1.77 A
-Methyl-Tyr 52(R, R ) 67(S, S ) 1.87 A
55(S, R ) 67(R, S ) 1.66 A
2-Methyl-Tyr 54(D) 62 1.39 A
2,5-Dimethyl-Tyr 56(D) 67 1.59 A
a
Migration distance 13 cm; chamber saturation.
b
hRF"RF;100.
c
"(1%RF1!1)/(1%RF2!1).
d
A"methanol}water}acetonitrile 50 : 50 : 200 (v/v/v); B"methanol}water}acetonitrile 50 : 50 : 30 (v/v/v); C"methanol}water
10 : 80 (v/v); D"acetone}methanol}water 10 : 2 : 2 (v/v/v).
e
Hyp"hydroxyproline.
f
Met(O2)"methionine sulfone.
common and uncommon dansyl (Dns) amino acids. time optical isomers have been separated with eluents
Similar results can be obtained on 10 cm;10 cm SIL containing BSA in the presence of very high levels
C18-50/UV254 layers with the same eluents but with (12}36%) of organic modiRer. The resulting spots
lower migration times. Some enantiomeric Dns- have the shape of a reversed triangle. FMOC-DL-Asn
amino acids such as Lys, Met, Nva, Pro and aromatic and FMOC-DL-Gln are not resolved. The
compounds show low selectivity coefRcients with - order of retention of the D and L forms of the
CD. Therefore, it can be useful to resolve these ra- different compounds is variable. The D forms of
cemates on RP-18W/UV254 plates with eluents con- FMOC-Pro, FMOC-Trp and FMOC-Met are more
taining BSA since very high values have been retained than the L forms, whereas the opposite is
achieved. noted for the other amino acids. This behaviour is
Other N- and C-terminal substituents studied by also shown from DNP-amino acids and other N-
chiral TLC include 2,4-dinitrophenyl (DNP), 3,5- derivatives.
dinitro-2-pyridyl (DNPy), 3,5-dinitrobenzoyl (DNB), Most DNP, DNPy and DNB-DL-amino acids are
o-nitrophenylsulfenyl (o-NPS), 9-Suorenyl- resolved on RP-18W/UV254 plates with 0.1 mol L\1
methoxycarbonyl (FMOC), methylthiohydanthoin acetate buffer solutions containing 2% iso-
(MTH), phenylthiohydanthoin (PTH), t-butoxycar- propanol and different BSA concentrations (2}6%)
bonyl (t-BOC), carbobenzoxy (CBZ), phthalyl, but few of them show chiral separation with phos-
acetyl, p-nitroanilide (pNA) and -naphthylamide phate buffer (0.05 mol L\1 potassium dihydrogen
(NA) (Table 6). phosphate#0.05 mol L\1 disodium hydrogen phos-
The maximum RF for the enantiomers of FMOC phate), an eluent of higher pH (6.86) than that pre-
amino acids was obtained at different concentrations viously used. Enantiomeric DNPy-Ala, DNPy-Nva
of 2-propanol. It is worth noting that this is the Rrst and DNP-Eth(O2) are completely separated at low
Table 2 Resolution of racemic amino acids by chiral TLC
Racemate hRF1 a hRF2 b Separation technique
Ala 18(D) 53 5.13 Slurry of silica gel (Merck) and (!)-brucine brought to pH 7.1
Ser 12(D) 50 7.33 with 0.1 mol L\1 NaOH and spread on 20 cm;20 cm plates.
Thr 16(D) 29 2.15 Eluent: butanol}acetic acid}water 3 : 1 : 4 (v/v/v). Migration dis-
Ile 16(D) 35 2.82 tance, 10 cm; development time 0.5 h.
Met 18(D) 29 1.86 Visualization: ninhydrin.
Phe 27(D) 40 1.80
Tyr 22(D) 29 1.45
Trp 17(D) 31 2.19
Trp 59(D) 72 1.77 SIL C18-50/UV254 plates (Cat. No. 711308, MN) 10 cm;10 cm,
Trp-NH2c 31(L) 40 1.48 thickness 0.20 mm.
4-Methyl-Trp 42 65 2.56 Eluent: 0.05 mol L\1 NaHCO3#0.05 mol L\1 Na2CO3 contain-
5-Methyl-Trp 37 61 2.53 ing 6% BSA and 6% isopropanol (pH 9.8); for the resolution
6-Methyl-Trp 66 78 1.82 of 7-methylTrp, 5-methoxyTrp and Kynurenine, 0.05 mol L\1
7-Methyl-Trp 41 50 1.43 sodium tetraborate was used. Migration distance, 8 cm;
5-Methoxy-Trp 42 49 1.32 development time 1 h 50 min.
4-Fluoro-Trp 51 66 1.86 Visualization: p-dimethylaminobenzaldehyde.
5-Fluoro-Trp 43 63 2.23
6-Fluoro-Trp 42 54 1.62
Kynurenine 69(D) 80 1.80
3-(1-Naphthyl)-Ala 34(D) 40 1.29
Val 62(D) 68 1.30 DC plastikfolien, Kieselgel 60 F254 (Merck), 20 cm;20 cm,

Gln 59(D) 66 1.35 thickness 0.2 mm.
Arg 50(D) 59 1.44 Eluent: acetonitrile}water 1 : 2.5 for Arg, His and Lys and 1.5 : 2
Cit 65(D) 69 1.20 for the others; the water containing 6.5 ) 10\3 mol L\1 2-O-[(R )-
His 46(D) 55 1.43 2-hydroxypropyl]--CD.
Lys 49(D) 60 1.56 Migration distance, 18 cm at 193C.
a
hRF"RF;100.
b
"(1%RF1!1)/(1%RF2!1).
c
Tryptophanamide.
temperature (103C) and pH values (0.5 mol L\1 mobile phase modiRer while pNA derivatives show
acetic acid) where their retention by the layer is sufR- discordant results. In fact, DL-Leu-pNA is fully re-
cient. The unresolved racemates include DNPy-Ser, solved but DL-Ala-pNA failed since the latter optical
DNP-Asp, DNP-Glu and DNPy-Trp. The Rrst three antipodes do not form inclusion complexes of
amino acids are markedly polar and are only slightly sufRcient stability with -CD. In addition, BSA seems
retained by silanized silica gel plates, even when efRcient in the enantioseparation of pNA derivatives.
eluted with acidic solution; this may be the reason for
their not being resolved. In general, planar
-Hydroxycarboxylic Acids
chromatography clearly separates the enantiomers of
N-derivatized hydrophobic amino acids. Table 7 reports the separation and resolution data for
The complete resolution of DNPy-DL-Trp is ob- aliphatic and aromatic DL--hydroxycarboxylic acids
tained on layers of SIL C18-50/UV254 with -CD as on HPTLC-CHIR plates with concentrating zone
chiral agent. where the selectivity coefRcients appear to be higher
The optical isomers of PTH-amino acids are sensi- than those obtained on Chiralplates using the same
tive to light and readily racemize. Racemization of eluent (dichloromethane}methanol, 45 : 5 v/v).
these optical active derivatives is observed on Vanadium pentoxide may be used for detection of
silanized silica gel plates with acidic eluents. MCTA aromatic and aliphatic compounds. The oxide
plates may be useful since they are able to separate (1.82 g) is dissolved in 30 mL of 1 mol L\1 sodium
enantiomeric MTH-Phe, MTH-Tyr, MTH-Pro and carbonate by ultrasonic bath and, after cooling,
PTH-Pro with neutral aqueous}alcoholic eluents. 46 mL of 2.5 mol L\1 sulfuric acid and acetonitrile to
Among C-terminal substituents, the enantio- 100 mL are added. Plates dipped in this solution and
meric NA derivatives of amino acids were well sep- allowed to stand at room temperature give blue spots
arated on silanized silica gel plates with -CD as on a yellow background. All racemates studied were
Table 3 Retention and resolution data for racemic amino acids on home-made and precoated cellulose plates
Racemate hRF1a hRF2 b Eluent Remarks
Trp 50(D) 53 1.12 A Home-made microcrystalline cellulose plates

5-HydroxyTrp 25(D) 31 1.35 A (Avicel SF, Funakoshi, Japan) 20 cm;20 cm.
Kynurenine 54(D) 61 1.33 A Development time 2.3 h; visualization: UV365.
3-Hydroxykynurenine 47(D) 53 1.27 A A"methanol}butanol}benzene}water
5-Hydroxykynurenine 20(D) 26 1.40 A 2 : 1 : 1 : 1 (v/v/v/v).
3-Methoxykynurenine 55(D) 62 1.33 A
N--acetylkynurenine 74(D) 82 1.60 A
Diaminoadipic acid 23(D, D) 28 1.30 B Cellulose plates (Merck); B"methanol}

Diaminopimelic acid 25(D, D) 37 1.76 B water}acetic acid 40 : 10 : 2 (v/v/v).
Trp 46(L) 52 1.27 C DC-Alufolien Cellulose F254 plates (Merck)

5-HydroxyTrp 34(L) 41 1.34 C (20 cm;20 cm;0.1 mm). Development time
Kynurenine 38(D) 47 1.44 C 1.8}4.5 h; migration distance 15 cm; visualiz-
4-AminoPhe 40(L) 45 1.22 D ation: ninhydrin. Eluents: methanol}water 3 : 1
Phe-4-sulfonic acid 70(L) 73 1.15 Ec (D), 7 : 3 (F), 3 : 2 (C); n-butanol}acetic
o-Tyr 57 61 1.17 F acid}water 1 : 1 : 1 (E).
m-Tyr 55 59 1.17 F Layers heated at 1103C for 5 min before use.
p-Tyr 81(L) 83 1.14 Fc
p-Tyr-3-sulfonic acid 30(L) 40 1.55 E
Dopa 53(L) 57 1.17 C
Trp 40(L) 49 1.44 G Avicel SF plates 20 cm;20 cm, Lot 8390,

His 11(D) 13 1.20 G Funakoshi, Japan. Development time 11.5 h at
Phe 55(L) 59 1.17 G 03C. Visualization: ninhydrin.
Tyr 53(L) 60 1.33 G G"ethanol}pyridine}water 2 : 3 : 1 (v/v/v) or
Dopa 43(L) 50 1.32 G 1 : 1 : 1 (v/v/v).
Cys 6(D) 8 1.35 G
Thr 51(D) 56 1.22 G
Tyr 75(L) 81 1.42 H DC-Plastikfolien Cellulose plates (Merck, Cat.
Trp 57(L) 62 1.23 H No. 5577), 20 cm;20 cm;0.1 mm.
4-MethylTrp 29(L) 36 1.37 H Migration distance 10 cm.
42(L) 52 1.50 I Visualization: iodine vapour.
5-MethylTrp 37(L) 46 1.45 H Eluents: 0.1 mol L\1 NaCl (H); ethanol}
48(L) 54 1.27 I pyridine}water 1 : 1 : 1 (I).
6-MethylTrp 32(L) 39 1.35 H
47(L) 55 1.37 I
7-MethylTrp 34(L) 41 1.34 H
4-FluoroTrp 38(L) 44 1.28 H
53(L) 60 1.33 I
59(L) 64 1.23 I
61(L) 65 1.18 I
5-HydroxyTrp 31(L) 36 1.25 H
41(L) 48 1.33 I
Kynurenine 48(L) 55 1.32 H
43(L) 51 1.38 I
a
hRF"RF;100.
b
"(1%RF1!1)/(1%RF2!1).
c
Two successive developments with the same eluent.
completely resolved, the D forms being the most isocyanate in dichloromethane, has been performed
retained. on HPTLC-NH2 F254 plates chemically bonded with
N-(3,5-dinitrobenzoyl)-R-(!)--phenylglycine
(DNBPG) and eluted with different mixtures of n-
Acidic and Basic Drugs
hexane/isopropanol.
The enantiomers of basic -blocking drugs can be Interesting separations of racemates with a -
separated on HPTLC DIOL F254 plates (Merck) with aminoalcohol structure (i.e. ephedrine and noreph-
N-CBZ-Gly-L-Pro (or similar chiral agents) in the edrine, and -blockers) can be achieved on MCTA
mobile phase, while the separation of the same plates after their cyclization with phosgene to form
drugs, derivatized with (R)-(!)-1-(1-naphthyl)ethyl 5-substituted oxazolidinones.
Table 4 Enantiomeric separation of N-alkyl, N-carbamyl and N-formyl amino acids and of dipeptides on Chiralplates
Racemate hRF1a hRF2 Eluent Remarks
N-Formyl-t-Leu 48(#) 61(!) A A"methanol}water}acetic acid 50 : 50 : 200 (v/v/v).

N-Methyl-Abu 65 73 D B"methanol}water}acetic acid 50 : 50 : 30 (v/v/v).
N-Ethyl-Abu 69 72 D D"acetone}methanol}water 10 : 2 : 2 (v/v/v).
N-Methyl-Ala 64 70 D M"1 mmol L\ copper (II) acetate, 5% methanol (pH 5.8).
N-Methyl-Asp 58(L) 67(D) B Migration distance 13 cm; chamber saturation.
N-Methyl-Leu 49(L) 57(D) A Visualization; Ehrlich’s reagent for N-carbamylTrp, iodine for
N-Methyl-Nle 68 77 D N,N-dimethyl-Phe, and ninhydrin for the others.
N-Methyl-Nva 67 76 D
N-Ethyl-Nva 70 74 D
N-Methyl-Phe 50(D) 61(L) A
N,N-Dimethyl-Phe 55(D) 61(L) B
N-Methyl-m-Tyr 36 52 B
N-Methyl-Val 65(L) 70(D) B
N-Carbamyl-Trp 44(L) 55(D) M
Gly-Phe 57(L) 63(D) B
Gly-Leu 53(L) 60(D) B
Gly-Ile 54(L) 61(D) B
Gly-Val 58(L) 62(D) B
Gly-Trp 48(L) 55(D) B
Leu-Leu 19(D, L) 26(L, D) A
48(D, L) 57(L, D) B
Ala-Phe 21(D, L) 26(L, D) A
59(D, L) 65(L, D) B
Met-Met 29(D, L) 33(L, D) A
64(D, L) 71(L, D) B
Asp-Phe-OCH3 50(L, L) 62(D, D) A
50(L, D) 62(D, L) A
a
hRF"RF;100.
Many acidic drugs (Figure 2) are resolved as 3,5- enantiomers of the tested compounds can be separ-
dinitroanilyl (DNAn) derivatives on precoated ated by at least one of the chiral phases reported in
HPTLC-NH2 F254 plates derivatized with R-(!)-1-(1- Table 10.
naphthyl)ethyl isocyanate (Table 8). Although the In the series of Savanones no chiral discrimination
naphthylethyl chromophore has a high UV adsorptiv- was observed on MCTA plates for racemic 2-hy-
ity, the detection problems found on plates bonded droxy-, 4-hydroxy- and 4-methoxySavanone in
with DNBPG were not observed. contrast to polysubstituted compounds. Partial res-
High selectivity coefRcients are obtained for un- olution was obtained for Savanone, 6-methoxy- and
derivatized acidic drugs on MCTA and diphenyl-F 6-hydroxySavanone. Two successive developments
plates eluted with aqueous}organic solutions contain- with the same eluent (ethanol}water, 80 : 20 v/v)
ing, in the latter case, a chiral macrocyclic antibiotic effectively improves the separation of these racemates
(vancomycin). on MCTA layers.
Other pharmaceuticals resolved include bendro- The addition of -CD to the mobile phase permits
Sumethiazide, coumachlor, mephenytoin, oxindanac separation of enantiomeric Savanone and its 2-hy-
benzyl ester, warfarin, chlorowarfarin, hexobarbital, droxy-, 4-hydroxy- and 4-methoxy derivatives.
oxazepan, lorazepan, norphenylephedrine, hyos- Albumin is able to resolve racemic polysubstituted
cyamine and colchicine. Savanones and 2-hydroxySavanone. Alkaline mobile
phases must be used for their separation.
Flavanones
Flavanones occur in nature and have been isolated in
Miscellaneous Compounds
an optically active form. They contain only hydroxyl
and methoxy groups and differ from one another in The chiral NMR solvating agent 1-(9-anthryl)-2,2,2-
the number and/or position of such substituents triSuoroethanol (TFAE) has been separated by a var-
(Table 9). iety of chromatographic techniques and has became
With the exception of glycosides, 5-methoxy-, 7- a reference compound for testing new optically active
hydroxy- and 5-hydroxy-7-methoxySavanone, the selectors. For example, values of 2.02 and 2.34
Table 5 Enantioseparation of racemic Dns-amino acids with chiral mobile phases
Dns-Amino acid hRF1a hRF2 b Eluent Remarks
Abu 42(L) 47 1.22 C Reversed-phase plates: 5 cm;20 cm and 20 cm;20 cm,

Ala 40(L) 47 1.33 G KC18F, Whatman, USA. Development time 6}8 h.
Arg 55(L) 65 1.52 H Eluents: acetonitrile}0.133 mol L\1 -CD, 25 : 75 (A);
Asn 60(L) 69 1.48 H methanol}0.163 mol L\1 -CD, 35 : 65 (B);
Asp 64(L) 70 1.31 A acetonitrile}0.151 mol L\1 -CD, 30 : 70 (C);
Cit 54(L) 63 1.45 H acetonitrile}0.133 mol L\1 -CD, 20 : 80 (D);
Cys 37(L) 42 1.23 I methanol}0.151 mol L\1 -CD, 30 : 70 (E);
Gln 57(L) 66 1.46 H acetonitrile}0.231 mol L\1 -CD, 35 : 65 (F);
Glu 65(L) 72 1.38 B methanol}0.2 mol L\1 -CD, 35 : 65 (G);
His 58(L) 64 1.28 H acetonitrile}0.2 mol L\1 -CD, 20 : 80 (H);
Ile 33(L) 40 1.35 L methanol}0.2 mol L\1 -CD, 55 : 45 (I);
allo-Ile 30(L) 38 1.43 L acetonitrile}0.2 mol L\1 -CD, 32 : 68 (L);
Leu 30(L) 35 1.25 C methanol-saturated -CD, 60 : 40 (M);
Lys 35(L) 39 1.18 M methanol}0.2 mol L\1 -CD, 50 : 50 (N).
Met 34(L) 38 1.19 C Aqueous solutions of -CD also contain urea (saturated
Nle 24(L) 28 1.23 C solution) and 3.5% sodium chloride.
Nva 32(L) 34 1.09 C Visualization: UV254.
Orn 35(L) 40 1.23 M
Phe 35(L) 39 1.18 C
Pro 39(L) 41 1.08 N
Ser 41(L) 47 1.27 D
Thr 42(L) 51 1.43 E
Trp 43(L) 45 1.08 F
Tyr 23(L) 26 1.17 M
Val 36(L) 43 1.34 C
N-Methyl-Val 24(L) 28 1.23 N
Abu 34(L) 56 2.47 O RP-18W/UV254 plates (Art. 811075, Macherey-Nagel).

Asp 68(D) 79 1.77 O Migration distance 7 cm.
Glu 45(D) 65 2.27 O Eluents: 5% BSA in 0.1 mol L\1 acetate buffer (O);
Leu 6(D) 15 2.76 P 5% BSA and 1% NaCl in 0.5 mol L\1 acetic acid (P);
Met 32(L) 50 2.12 Q 6% BSA in 0.1 mol L\1 acetate buffer (Q);
Nle 38 50 1.63 Q 7% BSA in 0.5 mol L\1 acetic acid (R).
Nva 25(L) 73 8.13 O Eluents also contain 2% isopropanol.
Phe 24(L) 45 2.59 Q Visualization: UV254.
Ser 39(D) 46 1.33 R
Thr 34(L) 43 1.46 Q
25(D) 32 1.41 R
Trp 37(D) 62 2.77 O
Val 20(L) 33 1.97 O
hRF"RF;100.
a
"(1%RF1!1)/(1%RF2!1).
b
were obtained, respectively, on OPTI-TAC phases containing BSA for the separation of ($)-1,1-
F254 (Antec) plates eluted with ethanol}water 80 : 20 bi-2-naphthol ("2.15) and ($)-binaphthyl-2,2-
(v/v) and on SIL C18-50/UV254 layers using 6% BSA in diyl-hydrogen phosphate ("4.65).
0.05 mol L\1 sodium tetraborate containing 20% The enantiomeric separations of synthetic pyreth-
isopropanol (pH 9.75) as mobile phase. roids, such as alfamethrin and fenpropathrin, on
($)-1-(9-Fluorenyl)ethanol an analogue of TFAE, home-made MCTA plates with ethanol}water
was also resolved on home-made MCTA plates elut- 80 : 20 ("1.37 and 1.20, respectively) should be
ing with 2-propanol}water 80 : 20 (v/v) ("2.24). noted since their optical antipodes have different
The separation of chiral compounds with restricted rates of degradation and biological activity towards
rotation, as in the case of binaphthyl type of sub- animals and plants.
stances, can be effected both on CSPs and with The resolution of racemic fenoxaprop-ethyl on the
CMPs. The Rrst technique requires the use of MCTA above-mentioned CSP ("1.52, isopropanol}water
plates to resolve ($)-1,1-binaphthyl-2,2-diamine 80 : 20) in interesting since chlorophenoxyalkyl car-
("1.99) while the latter involves chiral mobile boxylic acids and esters are widely used herbicides.
Table 6 Enantioseparation of derivatized amino acids by chiral TLC
Derivative hRF1a hRF2 b R sc Plate Eluent e

Fmoc-Ala 89(L) 37 1.43 1.1 SIL C18-50/UV254 A
Fmoc-Chad 17(L) 30 2.09 2.3 SIL C18-50/UV254 B
Fmoc-Leu 29(L) 36 1.37 1.3 SIL C18-50/UV254 C
Fmoc-Met 20(D) 28 1.55 1.1 SIL C18-50/UV254 D
Fmoc-Nle 18(L) 27 1.68 1.4 SIL C18-50/UV254 B
Fmoc-Nva 19(L) 27 1.57 1.6 SIL C18-50/UV254 B
Fmoc-Phe 41(L) 50 1.43 1.2 SIL C18-50/UV254 C
Fmoc-Pro 47(D) 77 3.78 1.6 SIL C18-50/UV254 B
34(D) 40 1.29 2.0 MCTA E
Fmoc-Trp 20(D) 41 2.79 2.7 SIL C18-50/UV254 D
Fmoc-Val 36(L) 43 1.35 2.3 SIL C18-50/UV254 C
DNP-Abu 59 89 5.75 4.0 RP-18W/UV254 F
DNP-Cit 34 41 1.35 1.7 RP-18W/UV254 G
DNP-Eth 45 61 1.94 2.0 RP-18W/UV254 G
DNP-Eth(O2) 26 35 1.53 2.0 RP-18W/UV254 H
DNP-Leu 28(L) 54 3.02 3.6 RP-18W/UV254 I
24(D) 31 1.42 1.8 SIL C18-50/UV254 L
DNP-Met 28 61 4.02 3.3 RP-18W/UV254 M
DNP-Met(O2) 44(D) 50 1.27 2.0 RP-18W/UV254 N
DNP-Met(O) 27 34 1.39 2.3 RP-18W/UV254 N
DNP-Nle 31 63 3.78 5.1 RP-18W/UV254 I
DNP-Nva 40 89 12.19 5.0 RP-18W/UV254 O
DNP-Pip 45 56 1.56 1.3 RP-18W/UV254 F
DNPy-Ala 47 53 1.27 1.6 RP-18W/UV254 H
DNPy-Leu 45 70 2.85 4.0 RP-18W/UV254 N
38 43 1.23 1.5 SIL C18-50/UV254 L
23 31 1.50 2.5 SIL C18-50/UV254 P
DNPy-Met 31 56 2.61 4.2 RP-18W/UV254 Q
DNPy-Nle 25 63 5.11 5.7 RP-18W/UV254 N
DNPy-Nva 21 31 1.69 2.3 RP-18W/UV254 H
DNPy-Phe 28 61 4.02 5.0 RP-18W/UV254 Q
47 51 1.17 1.2 SIL C18-50/UV254 L
18 26 1.60 2.0 SIL C18-50/UV254 P
DNPy-Trp 30 49 2.24 5.0 SIL C18-50/UV254 L
DNB-Leu 34(L) 51 2.02 3.0 RP-18W/UV254 N
DNB-PhenylGly 30(L) 67 4.75 7.3 RP-18W/F254S N
MTH-Pro 12 16 1.39 1.3 RP-18W/UV254 R
33 37 1.19 1.0 MCTA E
MTH-Phe 43 49 1.27 1.7 MCTA S
MTH-Tyr 42 45 1.13 1.0 MCTA E
PTH-Pro 13 25 2.23 2.5 MCTA E
N-[1-(1-Naphthyl)ethyl]phthalamic acid 54(R) 58 1.17 1.6 MCTA T
N-Benzylproline ethyl ester 19(D) 22 1.20 1.0 MCTA U
Amethopterin 9(L) 19 2.34 2.3 RP-18W/UV254 V
N-Acetyl-5-methyl-Trp 33 76 6.44 3.0 RP-18W/UV254 M
N-CBZ-Trp 44(D) 88 9.33 3.0 RP-18W/UV254 Z
N-t-BOC-Trp 16(L) 23 1.57 1.2 RP-18W/UV254 V
N-t-BOC-p-nitro-Phe 20(D) 30 1.71 2.3 SIL C18-50/UV254 W
Ala--NA 68(L) 76 1.49 1.5 SIL C18-50/UV254 Y
59 66 1.35 } KC18F J
Leu--NA 54(L) 64 1.50 2.0 SIL C18-50/UV254 L
Met--NA 45(L) 55 1.50 2.3 SIL C18-50/UV254 L
Ala-p-NA 12(L) 14 1.19 0.4 RP-18W/UV254 K
Leu-p-NA 4(L) 7 1.81 1.2 RP-18W/UV254 K1
19(L) 23 1.27 1.0 SIL C18-50/UV254 K2
56(L) 63 1.33 1.7 SIL C18-50/UV254 L
a
hRF"RF;100.
b
"(1%RF1!1)/(1%RF2!1).
c
Rs"2;(distance between the centres of two adjacent spots)/(sum of the width of the two spots in the direction of development).
d
Cha"-Cyclohexylalanine.
e
Eluents: A"6% BSA, 23% 2-propanol, 0.1 mol L\1 acetate buffer; B"5% BSA, 23% 2-propanol, 0.1 mol L\1 acetate buffer; C"5% BSA,
36% 2-propanol, 0.1 mol L\1 acetate buffer; D"6% BSA, 12% 2-propanol, 0.1 mol L\1 acetate buffer; E"2-propanol/water 60 : 40 (v/v);
F"4% BSA, 2% 2-propanol, 0.1 mol L\1 acetate buffer (103C); G"6% BSA, 2% 2-propanol, 0.1 mol L\1 acetate buffer; H"4% BSA, 1%
NaCl, 2% 2-propanol, 0.5 mol L\1 acetic acid (103C); I"4% BSA, 2% 2-propanol, 0.05 mol L\1 phosphate buffer; L"0.15 mol L\1 -CD in
a water}acetonitrile solution (80 : 20, v/v) containing 26% urea and 3% NaCl; M"3% BSA, 2% 2-propanol, 0.05 mol L\1 phosphate buffer;
N"5% BSA, 2% 2-propanol, 0.1 mol L\1 acetate buffer; O"4% BSA, 2% 2-propanol, 0.1 mol L\1 acetate buffer; P"Hydroxypropyl--CD
(13.8 g) in water}acetonitrile}acetic acid (45 : 4 : 1, v/v/v, 100 mL); Q"2% BSA, 2% 2-propanol, 0.1 mol L\1 acetate buffer; R"9% BSA, 2%
2-propanol, 0.1 mol L\1 acetate buffer; S"2-propanol}water 80 : 20 (v/v); T"ethanol}water 70 : 30 (v/v); U"2-propanol}water 40 : 60 (v/v);
V"8% BSA, 2% 2-propanol, 0.5 mol L\1 acetic acid; Z"5% BSA, 2% 2-propanol, 0.5 mol L\1 acetic acid; W"6% BSA, 6% 2-propanol,
0.05 mol L\1 NaHCO3#0.05 mol L\1 Na2CO3; Y"0.1 mol L\1 -CD in a water}acetonitrile solution (80 : 20, v/v) containing 20% urea and 3%
NaCl; J"0.163 mol L\1 -CD in a water}methanol solution (65 : 35, v/v) containing 3.5% NaCl and saturated with urea; K"8% BSA, 3%
2-propanol, 0.1 mol L\1 acetate buffer; K1"6% BSA, 8% 2-propanol, 0.1 mol L\1 acetate buffer; K2"8% BSA, 20% 2-propanol, 0.1 mol L\1
acetate buffer.
Table 7 Separation of enantiomeric -hydroxycarboxylic acids Quantitative Analysis of

on HPTLC-CHIR platesa
TLC-Separated Enantiomers
Racemate hRF1b hRF2 c Eluent d
TLC is generally coupled with spectrophotometric
Mandelic acid 36 48 1.64 A methods for quantitative analysis. QuantiRcation can
4-Bromo-mandelic acid 33 44 1.59 B be achieved by in situ densitometry or after extraction
4-Chloro-mandelic acid 35 42 1.34 B of solutes from the scraped layer. The evaluation of
3-Hydroxy-mandelic acid 47 59 1.62 B
detection limits for separated enantiomers is essential
4-Hydroxy-mandelic acid 45 57 1.61 C
3,4-Dihydroxy-mandelic acid 33 44 1.59 C because precise determinations of trace levels of D- or
4-Hydroxy-3-methoxy- 24 33 1.55 A L-enantiomer in an excess of the other is becoming
mandelic acid more and more important.
2-Hydroxy-2-phenyl- 38 47 1.44 A On Chiralplates and HPTLC-CHIR layers, densito
propanoic acid
metry can be performed after postchromato-
2-Hydroxy-3-phenyl- 39 51 1.62 A
propanoic acid graphic derivatization of compounds with ninhy-
Lactic acid 70(D) 76(L) 1.38 D drin or vanadium pentoxide. Successful separation
2-Hydroxy-butanoic acid 27 37 1.58 A of amino acids on Chiralplates depends on the
2-Hydroxy-3-methoxy- 33 46 1.73 A hydrochloric acid content of the applied solu-
butanoic acid
tion (usually a methanol}0.1 mol L\1 HCl 1 : 1 (v/v)
2-Hydroxy-4-methylthio- 33 45 1.66 A
butanoic acid mixture).
2-Hydroxy-pentanoic acid 25 39 1.91 A Remission}location curves of DL--hydroxycar-
2-Hydroxy-3-methyl- 34 49 1.86 A boxylic acids, achieved in reSectance mode with
pentanoic acid a Shimadzu CS930 scanner or a Desaga CD60 den-
2-Hydroxy-4-methyl- 35 47 1.64 A
sitometer, show that only enantiomers with high RF
pentanoic acid
2-Hydroxy-hexanoic acid 62(D) 69(L) 1.36 D values (*0.10) can be baseline resolved on
2-Hydroxy-octanoic acid 36 50 1.78 A 10 cm;10 cm HPTLC-CHIR plates (Figure 3). On
2-Hydroxy-tetradecanoic 34 49 1.86 A such plates, L-2-hydroxy-3-phenylpropionic acid
acid spiked with 1% D enantiomer (RF"0.12) gives rise
2-Hydroxy-hexadecanoic 39 56 1.99 A
to partially resolved peaks but the D isomer is still
acid
2-Hydroxy- 39 56 1.99 A visible. With respect to small particle size HPTLC-
docosahexanoic acid CHIR layers, higher Rs values have been obtained on
20 cm;20 cm Chiralplates owing to migration dis-
a
Migration distance, measured from concentrating zone, 13 cm; tances being twice as long ( values being equal).
visualization: (a) the plates were dipped in MnCl2}sulfuric acid
The remission}location curves of Figure 4 and the
heating up to 1203C for 30 min for aromatic -hydroxycarboxylic
acids; (b) the plates were dipped for 2 s in vanadium (V)}sulfuric calibration line for L-phenylalanine (Figure 5) dem-
acid solution and dried at room temperature for c. 45 min for onstrate that quantitative determinations of L-isomer
aromatic and aliphatic -hydroxycarboxylic acids. in D-phenylalanine on Chiralplates (RF"0.10) are
b
hRF"RF;100. possible in a working range of 0.04}0.4 g/spot, that
c
"(1%RF1!1)/(1%RF2!1).
d
is 0.1}1%. Further determinations include 0.1% D-t-
Eluents: A"dichloromethane/methanol 45 : 5 (v/v); B"
0.05 mol L\1 KH2PO4 in a methanol}acetonitrile}water Leu in L-t-Leu (RF"0.11), 0.1% L-5,5-dimethyl-
50 : 50 : 200 (v/v/v) mixture; C"0.1 mol L\1 LiCl in a dichloro- thiazolidine-4-carboxylic acid in the D-enantiomer
methane}ethanol 85 : 15 (v/v) mixture; D"acetonitrile}water (RF"0.14) and 1% D-hydroxyphenylalanine in the
3 : 2 (v/v). L-enantiomer.
The peak of 1% Dns-D-Glu in L-enantiomer is vis-
ible on 20 cm;20 cm RP-18 plates (Merck) impreg-
The use of mobile phases containing -CD seems nated with a solution of 8 mol L\1 N,N-di-n-propyl-
1
to be particularly appropriate for the resolution of L-alanine and 4 mmol L\ cupric acetate.
racemic S-(1-ferrocenyl-2-methylpropyl)thioethanol On 10 cm;20 cm cellulose plates L-tryptophan
and S-(1-ferrocenylethyl)thioethanol ("1.43 and spiked with 5% D-enantiomer gives rise to partially
1.18, respectively). resolved peaks owing to the small RF value (0.06).
Many noncharged solutes with a carbonyl group The use of MCTA allows the determination of
close to the stereogenic centre can be resolved on enantiomeric mixtures in the ratios 100 : 1 and
MCTA plates (benzoin, benzoin methyl ether, 200 : 1. (S)-2,2,2-TriSuoro-1-(9-anthryl)ethanol can
2-phenylbutyrophenone, 2- and 3-methylindanone, be detected at 1% level in (R) enantiomer on OPTI-
2-phyenylcyclohexanone, 2-phenylcycloheptanone TAC F254 plates eluted with ethanol}water 80 : 20
and 2-oxazolidone derivatives). (RF"0.17; length of run 10 cm). Baseline-resolved
Figure 2 Structure of acidic drugs.
peaks (RF"0.10) were obtained for the two Conclusions

atropisomers of 1,1-binaphthyl-2,2-diamine on
20 cm;20 cm home-made MCTA plates at 100 : 1 Chiral TLC plays a signiRcant role both in economi-
ratio. Partial resolution only was observed at a cal routine analyses and in determination of optical
ratio of 200 : 1, but the S isomer is still visible purity of individual antipodes. Detection limits of
(Figure 6). *0.1% D- or L-isomer can be currently achieved.
Table 8 Retention and resolution data for derivatized and free acidic drugs by chiral TLC
Drug hRF1a hRF2 b Eluent c Plates and remarks
DNAn-ibuprofen 28 (S) 45 (R) 2.10 A Precoated 10 cm;10 cm HPTLC-NH2 F254s plates

DNAn-naproxen 15 (S) 24 (R) 1.79 A (Altech, Deerfield, IL, USA), derivatized with (R )-(!)-1-
DNAn-fenoprofen 23 33 1.65 A (1-naphthyl)ethyl isocyanate.
DNAn-flurbiprofen 23 33 1.65 A Visualization: UV254 and UV360.
DNAn-benoxaprofen 20 30 1.71 A
Flurbiprofen 18 24 1.44 B MCTA plates.
Carprofen 36 41 1.23 C Visualization: UV.
Indoprofen 58 63 1.22 D 5 cm;20 cm chemically-bonded diphenyl-F plates.
a
hRF"RF;100.
b
"(1%RF1!1)/(1%RF2!1).
c
Eluents: A"n-hexane}isopropanol}acetonitrile 20 : 8 : 1 (v/v/v); B"ethanol}water 40 : 60 (v/v); C"isopropanol}water 60 : 40
(v/v); D"acetonitrile}0.6 mol L\1 NaCl}1% triethylammonium acetate buffer (pH 4.1) containing vancomycin.
Table 9 The structure of racemic flavanones
R3 R5 R6 R7 R2 R3 R4 Name
Y Y Y
H H H H H H H Flavanone
H OCH3 H H H H H 5-Methoxyflavanone
H H OH H H H H 6-Hydroxyflavanone
H H OCH3 H H H H 6-Methoxyflavanone
H H H OH H H H 7-Hydroxyflavanone
H H H H OH H H 2-Hydroxyflavanone
H H H H H H OH 4-Hydroxyflavanone
H H H H H H OCH3 4-Methoxyflavanone
H OH H OH H H H Pinocembrin
H OH H OCH3 H H H Pinocembrin-7-methylether
H OH H OH H H OH Naringenin
H OH H OH H H OCH3 Isosakuranetin
H OH H OCH3 H H OH Sakuranetin
H OH H Gla H H OH Naringenin-7-glucoside
H OH H Rh-Glb H H OH Naringin
H OH H OH H OH OH Eriodictyol
H OH H OH H OCH3 OH Homoeriodictyol
H OH H OH H OH OCH3 Hesperetin
OH OH H OH H OH OH Taxifolin
a
Gl"Glucoside.
b
Rh-Gl"Rhamnosidoglucoside.
Table 10 Retention and resolution data for racemic flavanones by chiral TLC
Racemate hRF1a hRF2 b Rsc Plate Eluent d
Flavanone (F) 16 20 1.31 1.6 SIL C18-50/UV254 A

22 24 1.12 0.4 MCTA B
6-Hydroxy-F 36 39 1.14 0.8 MCTA B
6-Methoxy-F 24 27 1.17 0.8 MCTA B
2-Hydroxy-F 10 16 1.71 2.0 SIL C18-50/UV254 C
19 24 1.35 1.6 SIL C18-50/UV254 A
4-Hydroxy-F 38 42 1.18 1.2 SIL C18-50/UV254 A
4-Methoxy-F 13 19 1.57 2.0 SIL C18-50/UV254 A
5,7-Dihydroxy-F 54 60 1.27 1.8 MCTA D
4,5,7-Trihydroxy-F 23 28 1.30 1.6 MCTA E
5,7-Dihydroxy-4-methoxy-F 18 21 1.21 1.3 MCTA E
4,5-Dihydroxy-7-methoxy-F 43 48 1.22 1.2 MCTA D
3,4,5,7-Tetrahydroxy-F 26 30 1.21 1.5 MCTA E
4,5,7-Trihydroxy-3-methoxy-F 23 26 1.17 0.8 MCTA E
3,5,7-Trihydroxy-4-methoxy-F 23 27 1.24 1.5 MCTA E
3,3,4,5,7-Pentahydroxy-F 44 48 1.17 1.3 MCTA E
a
hRF"RF;100.
b
"(1%RF1!1)/(1%RF2!1).
c
Rs"2;(distance between the centres of two adjacent spots)/(sum of the width of the two spots in the direction of development).
d
Eluents: A"0.15 mol L\1 -CD aqueous solution with urea (32%) and NaCl (2%)}acetonitrile 80 : 20 (v/v), migration distance
8.5 cm; B"ethanol}water 80 : 20 (v/v), migration distance 12 cm; C"0.05 mol L\1 sodium bicarbonate#0.05 mol L\1 sodium
carbonate solution containing 6% BSA and 12% isopropanol, migration distance 8 cm; D"ethanol}water 70 : 30 (v/v), migration
distance 14 cm; E"methanol}water 80 : 20 (v/v), migration distance 16 cm.
Figure 3 Remission}location curves recorded on 10 cm;

10 cm HPTLC-CHIR plates. (A) D,L-Lactic acid (RF"
0.05); (B) D,L-2-hydroxybutanoic acid (RF"0.10); (C) D,L-2-
hydroxyoctanoic acid (RF"0.14).
Figure 4 Remission}location curves recorded on 20 cm;

20 cm Chiralplates. (A) D-Phe spiked with 0.1% L-Phe; (B) 0.1%
L-Phe.
Figure 6 Densitograms of (R ) and (S ) - 1,1-binaphthyl-2,2-

diamine mixtures in the ratios 50 : 1, 100 : 1 and 200 : 1 on MCTA
layers, eluted with ethanol}water 80 : 20 (v/v). Migration distance
17 cm. (A) (R )"10 g, (S )"0.2 g; (B) (R )"20 g;
(S )"0.2 g; (C) (R )"40 g; (S )"0.2 g.
Less work is being carried out on chiral TLC than

on column chromatography, even though the two
techniques may give complementary results and TLC
has advantages such as low cost and easy evaluation
of the tests.
Future possibilities of chiral TLC include:
1. the synthesis of enantiomeric derivatives that are
easier to resolve and more sensitively detected
than those so far investigated;
2. the availability of layers prepared from new cellu-
Figure 5 Calibration line for L-phenylalanine. IE, integration lose derivatives and, in addition, the availability of
units; y"!463#16 349x ; r"0.9992; Sxo"0.0038 g per more versatile MCTA plates using highly crystal-
spot; "540 nm. line and homogeneously sized material;
3. more extensive application of normal-phase Further Reading

chromatography with a chiral mobile phase addi-
tive (DIOL plates are particularly advisable); Armstrong DW, He FY and Han SM (1988) Planar
chromatographic separation of enantiomers and dia-
4. the use of eluents containing new chiral selectors
stereomers with cyclodextrin mobile phase additives.
in reversed-phase systems, which is the technique Journal of Chromatography 448: 345}354.
most widely used for enantioseparations. Dalgliesh CE (1952) The optical resolution of aromatic
amino acids on paper chromatograms. Journal of the
See also: II/Chromatography: Thin-Layer (Planar): Chemical Society III: 3940}3942.
Densitometry and Image Analysis; Layers; Spray Re- Gunther R and MoK ller K (1996) Enantiomer separations.
agents. III/Amino Acids and Derivatives: Chiral Separ- In: Sherma J and Fried B (eds) Handbook of Thin Layer
ations: Chiral Separations: Capillary Electrophoresis; Chromatography, pp. 621}682. New York: Marcel
Cellulose and Cellulose Derived Phases; Chiral Derivatiz- Dekker.
ation; Cyclodextrins and Other Inclusion Complexation Lepri L (1997) Enantiomer separation by TLC. Journal of
Approaches; Ion-Pair Chromatography; Ligand Exchange Planar Chromatography, Modern TLC 10: 320}331.
Chromatography; Liquid Chromatography; Molecular Im- Lepri L, Coas V and Desideri PG (1992) Planar chromatog-
prints as Stationary Phases; Protein Stationary Phases; raphy of optical isomers with bovine serum albumin in
Supercritical Fluid Chromatography; Synthetic Multiple In- the mobile phase. Journal of Planar Chromatography,
teraction (‘Pirkle’) Stationary Phases. Modern TLC 5: 175}178.
III / CITRUS OILS: LIQUID CHROMATOGRAPHY 2441
CITRUS OILS :
LIQUID CHROMATOGRAPHY
P. Dugo, Universita% di Messina, Messina, Italy components. Often these components were not iden-
L. Mondello and G. Dugo, Facoltà di Farmacia, tiRed before and were responsible for ultraviolet (UV)
Messina, Italy absorption. Knowledge of the chromatographic char-
Copyright ^ 2000 Academic Press acteristics and the chemical structures of these com-
pounds was considered very important in order to
determine the authenticity of the oils. In fact, valuable
Citrus essential oils are very complex matrices which cold-pressed oils may be adulterated with less valu-
contain numerous compounds of different chemical able cold-pressed or distilled oils. In these cases the
classes. These compounds are generally divided into presence and/or the content of some oxygen hetero-
two fractions: the volatile fraction, which is the cyclic compounds may be useful to determine the
most representative, and ranges between 85 and kind and also the degree of adulteration.
99% in the different cold-pressed citrus oils, and the Many methods developed for the TLC analysis of
non-volatile residue, which ranges between 1 and oxygen heterocyclic compounds of citrus oils used
15%. silica gel as a stationary phase, and mixtures of
The development of new instrumental analytical hexane or cyclohexane with variable amounts of
techniques, mainly chromatographic, has allowed the ethyl acetate as mobile phases. For example, a paper
characterization of citrus essential oils to become of 1965 reports the separation of coumarins of many
more precise. citrus oils by TLC, obtained on silica gel plates, using
Gas chromatography is an essential tool for the mixtures of hexane with variable amounts (25}70%)
study of the volatile fraction, while liquid chromatog- of ethyl acetate, according to the different polarity of
raphy (thin-layer chromatography (TLC) and high- the components. The spots were detected by UV ab-
performance liquid chromatography (HPLC) com- sorbance at 254 and 366 nm. The components were
bined with spectral absorption and Suorescence also isolated from the plates, extracted from silica gel
measurements) is widely used for the study of the and analysed by UV spectroscopy. Table 2 provides
composition of the non-volatile residue. This fraction chromatographic and spectroscopic data for the oxy-
consists largely of oxygen heterocyclic compounds gen heterocyclic compounds of lemon, bergamot,
(coumarins, psoralens (furanocoumarins) and poly- mandarin, sweet orange and bitter orange. Identi-
methoxylated Savones) that exhibit strong absorp- Rcation was carried out by comparison of Rf
tion in the ultraviolet region (max about 315 nm). values and spectroscopic data with those of standard
The presence of coumarin compounds is wide- compounds.
spread in plants of the Rutaceae family to which the
citrus species belong. Their presence is qualitatively
and quantitatively different in the different citrus oils,
so knowledge of the oxygen heterocyclic fraction may
be useful to assess authenticity, the geographical ori-
gin and the possible adulteration of the oils.
Figure 1 shows the basic structures of the oxygen
heterocyclic compounds present in citrus oils. In the
numbered position, coumarins and psoralens may con-
tain hydroxyl, methoxyl, isopentenyl, isopentenyloxy,
geranyloxy groups; polymethoxylated Savones con-
tain methoxyl groups. Table 1 lists the oxygen het-
erocyclic compounds identiRed in citrus oils.
TLC Separations
The literature from the 1930s to the end of 1970s
reports numerous TLC separations of non-volatile Figure 1 Structures of oxygen heterocyclic compounds pres-
residue of citrus oils with the aim of isolating the ent in citrus oil.
2442 III / CITRUS OILS: LIQUID CHROMATOGRAPHY
Table 1 Oxygen heterocyclic compounds identified in citrus essential oils
Bitter Sweet Lemon Lime Bergamot Mandarin Grape

orange orange fruit
Aurapten (7-geranyloxycoumarin) X X X
Auraptenol (7-(2-hydroxy-isopent-3-enyloxy)coumarin X
Marmin (7-(6,7-dihydroxygeranyloxy)coumarin) X
Umbelliferone (7-hydroxycoumarin) X
Herniarin (7-methoxycoumarin) X X
Epoxyaurapten (7-(6,7-epoxygeranyloxy)coumarin) X
7-Isopentenyloxycoumarin X
Meranzin (7-methoxy-8(2,3-epoxy)isopentenylcoumarin) X X
Osthol (7-methoxy-8-isopentenylcoumarin) X X
Meranzin hydrate (7-methoxy-8(2,3-dihydroxy)isopentyloxy- X X
coumarin)
Isomeranzin (7-methoxy-8-(2-one-isopentyl)coumarin) X X
7-(Methoxy-8(2-formyl-2-methylpropyl)coumarin) X
Citropten (5,7-dimethoxycoumarin) X X X X X X
5-Isopentenyloxy-7-methoxycoumarin X X X
5-(2,3-Epoxy-isopentyloxy)-7-methoxycoumarin X
5-Geranyloxy-7-methoxycoumarin X X X
5-(2,3-Dihydroxy-isopentyloxy)-7-methoxycoumarin) X
Bergapten (5-methoxypsoralen) X X X X X
Epoxybergamottin (5-(6,7-epoxy-geranyloxypsoralen) X X
Epoxybergamottin hydrate (5-(6,7-dihydroxy-geranyloxypsoralen) X X
Bergamottin (5-geranyloxypsoralen) X X X X
Bergaptol (5-hydroxypsoralen) X X X X X
Oxypeucedanin (5-(2,3-epoxy-isopentyloxy)psoralen) X X X
Oxypeucedanin hydrate (5-(2,3-dihydroxy-isopentyloxy)psoralen) X X X
Pabulenol/Gosferol (5-(2-hydroxy-3-methylbut-3-enyloxy)psoralen) X
Isoimperatorin (5-isopentenyloxypsorelen) X X
8-Geranyloxypsoralen X X
Heraclenin (8-(2,3-epoxy-isopentyloxy)psoralen) X X
Heraclenol (8-(2,3-dihydroxyisopentyloxy)psoralen) X
Imperatorin (8-isopentenyloxypsoralen) X X
8-(6,7-Epoxygeranyloxy)psoralen X
5-Methoxy-8(2,3-epoxy-isopentyloxy)psoralen X
5-Geranyloxy-8-methoxypsoralen X X
Byakangelicin (8-(2,3-dihydroxy-isopentyloxy)-5-methoxypsoralen) X X
Isobyakangelicol (5-methoxy-8-(2-one-isopentyl)psoralen) X
5-Isopent-2-enyloxy-8-(2,3-epoxy-isopentyloxypsoralen) X
Phellopterin (5-methoxy-8-isopentenyloxypsoralen) X
Byakangelicol (5-methoxy-8-(2,3-epoxy-isopentyloxypsoralen) X X X
Isopimpinellin (5,8-dimethoxypsoralen) X
Cnidilin (5-Isopentenyloxy-8-methoxypsoralen) X
Neobyakangelicol (5-methoxy-8-(2-hydroxy-3-methylbut-3- X
enyloxy)psoralen
5-Isopentenyloxy-8-(2,3-dihydroxy-isopentyloxy)psoralen X
Cnidicin (5,8-diisopentenyloxy)psoralen X
5-Methoxy-8-geranyloxypsoralen X
3,6,7,8,4-pentamethoxyflavone X
Tangeretin (4,5,6,7,8-pentamethoxyflavone) X X X X
Nobiletin (3,4,5,6,7,8-hexamethoxyflavone) X X X X
3,3,4,5,6,7,8-Heptamethoxyflavone X X X X
Sinensetin (3,4,5,6,7-pentamethoxyflavone) X X X
3,3,4,5,6,7-Hexamethoxyflavone X
Tetra-O-methylscutellarein (4,5,6,7-tetramethoxyflavone) X X X X
Isopentenyloxy"3-methylbut-2-enyloxy; Geranyloxy"37-dimethyloct-2,6-enyloxy.
Figure 2 shows another example of a TLC separ- 75 : 25 (B). Detection was by UV absorbance at
ation of coumarins of a cold-pressed lemon oil, ob- 254 nm. As can be seen, 21 components have been
tained using butyl acetate instead of ethyl acetate, in separated, but only the main components were identi-
particular, hexane}butyl acetate 65 : 35 (A) and Red: (1) bergamottin; (3) 5-geranyloxy-7-methoxy-
Table 2 Chromatographic and spectroscopic data of the oxygen heterocyclic compounds of lemon, bergamot, mandarin, sweet
orange and bitter orange oils
Fluorescence Rf25H Rf30H Rf40H Rf50H Rf60H Rf70H max min shoulder
Lemon oil
1. yellow 0 0 0 315 280 }
2. yellow (bergaptol) 0.02 0.04 0.05 265, 310 260, 280 250
3. violet } 0.08 0.15 315 280 230, 245, 255
4. red (byakangelicin) 0.14 0.16 0.20 240, 265, 310 235, 255, 285 250
5. yellow (5, 8 -... psoralen) 0.20 0.26 0.27 250, 307 235, 275 260
6. blue (citropten) 0.33 0.36 0.38 245, 325 240, 265 255
7. yellow (8-geranyloxypsoralen)
8. blue 0.39 0.41 0.49 315 280 245, 270
9. blue (5-geranyloxy-7- 0.44 0.46 0.55 323 277 245, 270
methoxycoumarin)
10. yellow (bergamottin) 0.50 0.52 0.59 250, 310 240, 278 270
Bergamot oil
1. Yellow 0 0 0 0 0 0 } } }
2. Red 0 0 0 0 0 0.02 } } }
3. Blue 0 0 0 0.03 0.03 0.04 } } }
4. Blue 0 0 0 0.03 0.06 0.12 } } }
5. Violet 0.01 0.02 0.04 0.06 0.12 0.17 325 285 270
6. Yellow/red } 0.03 0.09 0.14 0.20 0.26 270, 320 285 270
7. Blue } 0.08 0.16 0.25 0.29 0.32 } } 265
8. Green } 0.08 0.22 0.29 0.40 0.44 270, 325 260, 305 }
9. Blue 0.14 0.17 0.28 0.32 } } 324 275 270
10. Yellow (bergapten) 0.20 0.22 0.29 0.37 0.50 0.55 250, 260, 310 235, 255, 280 }
11. Blue (citropten) 0.33 0.36 0.38 0.42 0.53 0.59 245, 255, 270, 325 240, 250, 265, 275 }
12. Blue (5-geranyloxy-7- 0.44 0.46 0.52 0.56 0.67 0.68 323 278 250, 270
methoxycoumarin)
13. Yellow (bergamottin) 0.50 0.52 0.59 0.60 0.67 0.68 240, 310 235, 280 270
14. Red } } 0.62 0.64 0.72 0.73 } } }
Mandarin oil
1. Yellow 0 0 0 0 325 } }
2. Blue 0 0 0.05 0.07 } } }
3. Blue 0.03 0.06 0.12 0.16 325 285 270
4. Yellow/green (nobiletin) 0.06 0.10 0.19 0.22 270, 333 260, 290 245
5. Yellow (tangeretin) 0.09 0.15 0.27 0.32 270, 325 245, 290 }
6. Red (a polymethoxyflavone) 0.13 0.18 0.32 0.34 270, 325 245, 290 }
7. Yellow/green (a flavanone) 0.18 0.26 0.35 0.38 275, 322 260, 300 }
8. Brown (a flavanone) 0.23 0.29 0.38 0.47 275, 332 265, 310 }
9. Red (a flavanone) 0.30 0.31 0.46 0.51 265, 275 260, 270 }
10. Blue (citropten) 0.37 0.41 0.54 0.56 } } }
11. Pale blue 0.53 } } } } } }
12. Blue (methyl anthranilate) 0.56 0.61 0.63 0.66 } } }
13. Blue (N-methyl 0.64 0.65 0.70 0.70 253, 355 240, 290 }
methylanthranilate)
Sweet orange oil

1. Yellow 0 0 0 0 } } }
2. Blue 0 0.03 0.05 0.07 } } }
3. Blue 0.03 0.06 0.12 0.16 265, 325 260, 295 }
4. Yellow/green (nobiletin) 0.05 0.10 0.19 0.22 250, 270, 333 240, 260, 290 }
5. Yellow/red (tangeretin) 0.10 0.17 0.26 0.33 270, 324 250, 285 }
6. Green (5,8-dihydroxy-3,7,3,4- 0.18 0.27 0.37 0.48 255, 270, 330 245, 260, 290 }
tetramethoxyflavone)
7. Blue (citropten) 0.38 0.44 0.53 0.59 } } }
8. Blue (methyl 0.56 0.61 0.63 0.66 } } }
anthranylate)
9. Yellow 0.67 0.70 0.74 0.87 } } }
Table 2 continued
Fluorescence Rf25H Rf30H Rf40H Rf50H Rf60H Rf70H max min shoulder
Bitter Orange oil

1. Yellow 0 0 0 0 0 } } }
2. Blue 0 0.02 0.03 0.05 0.07 320 268 260
3. Blue/yellow 0 0.04 0.06 0.13 0.18 313 280 270
4. Yellow/green (nobiletin) 0.02 0.05 0.11 0.18 0.22 270, 329 265, 285 250
5. Yellow/red (tangeretin) 0.03 0.10 0.17 0.26 0.33 270, 324 250, 290 }
6. Blue (aurapten) 0.10 0.18 0.25 0.34 0.44 255, 322 250, 265 }
7. Violet (umbelliferone) 0.14 0.20 0.29 0.38 0.49 245, 255,325 240, 250, 265 320
8. Blue 0.16 0.24 0.33 0.43 0.50 320 265 255
9. Yellow (bergapten) 0.23 0.29 0.38 0.48 0.55 250, 260, 310 245, 255, 280 235, 250
10. Blue (citropten) 0.23 0.35 0.48 0.51 0.59 } } }
11. Yellow (isoimperatorin) 0.29 0.39 0.47 } } 250, 310 245, 280 240
12. Blue 0.29 0.39 0.47 } } 260, 270, 320 245, 265, 275 }
13. Violet 0.32 0.39 0.47 } } 258, 320 240, 270 }
14. Blue (methyl anthranilate) 0.42 0.48 } } } } } }
15. Blue 0.49 0.48 } } } 235 } }
16. Yellow 0.54 0.59 0.62 0.70 0.73 } } 230, 310
17. Blue 0.66 } } } } 270 265 }
HThe subscript number represents the % amount of ethyl acetate in the eluent mixture. (Reproduced with permission from D’Amore G
and Calapaj R (1965) Rassegna Chimica 6: 264}269.)
psoralen; (7) 8-geranyloxypsoralen; (9) citropten; TLC separation of citrus oils has also been coupled
(12) oxypeucedanin; (14) byakangelicol. These main with detection by in situ Suorimetry. This method
components were isolated by preparative column shows the advantages of both the techniques, and
chromatography, crystallized and analysed by spec- a good selectivity and sensibility. As an example, the
troscopic methods [infrared (IR), UV, nuclear mag- quantitative determination of 5-geranyloxy-7-meth-
netic resonance (NMR)]. oxycoumarin and 5,7-dimethoxycoumarin (citrop-
ten) present in bergamot, lime and lemon oils, and the
qualitative proRle of citrus oils have been obtained by
measuring the Suorescence and the Suorescence
quenching proRles directly on the TLC plates. Emis-
sion monochromator wavelength settings of 403, 440
and 490 nm were used to obtain the various Suores-
cence emission proRles. An excitation wavelength of
272 nm and an emission wavelength of 520 nm were
used to obtain the Suorescence quenching proRle.
These values correspond to the maximum of excita-
tion and emission of the Suorescent indicator in the
adsorbent layer.
Analytical and preparative TLC of coumarins and
psoralens have been widely used. Some of the advant-
ages of this technique are its simplicity and low cost.
Disadvantages may be the difRculty of controlling the
Sow rate, the low resolution, the low reproducibility
and the long time required for development. In recent
years various modern planar chromatographic
methods have been reported for the analysis of
coumarins. Some of these methods use a new forced-
Sow TLC technique, developed by TyihaH ck and co-
workers between 1979 and 1981, called OPLC (over-
Figure 2 TLC separation of coumarins and psoralens of lemon
pressured layer chromatography). This new planar
oil. Eluent: hexane}butyl acetate, 65 : 35 (A) and 75 : 25 (B).
(Reproduced with permission from Glandian R, Corneteau H, technique combines the advantages of classical TLC
Drouet S and Rouzet M (1978) Plantes, Medicinales et and HPLC: shorter analysis times, lower solvent con-
Phytoterapie, 12 (2), 112}122.) sumption, simultaneous analysis of a large number of
Figure 3 (A) One-dimensional OPLC development and (B) two-dimensional OPLC development of 16 closely related coumarins.
(1) Umbelliferone: (2) herniarin; (3) psoralen; (4) osthol; (5) apterin; (6) angelicin; (7) bergapten; (8) oxypeucedanin; (9) isobergapten;
(10) scopoletin; (11) sphondin; (12) xanthotoxin; (13) imperatorin; (14) pimpinellin; (15) isopimpinellin; (16) new archangelicin deri-
vate. (Reproduced with permission from Harmala P, Botz L, Sticher O and Hiltunen R (1990) Journal of Planar Chromatography 3:
515}520.)
samples, the possibility of performing isocratic or dimensional high performance TLC (2D TLC).
gradient elution, control of the Sow rate and higher Figure 3 shows the one-dimensional (A) and the two-
efRciency. The literature reports some examples of dimensional (B) OPLC development of 16 coumarins.
the applications of OPLC technique to the analysis of Another way to increase the efRciency of TLC
coumarins and psoralens. separation is by using the stepwise gradient tech-
As shown by HaK rmaK laK et al. in 1990, TLC separ- nique. By increasing the strength of the mobile phase,
ation methods can be improved by use of two- the separation of compounds with similar Rf values is
Figure 4 Chromatograms of plant extracts and references coumarins corresponding to the following stepwise gradient programs: (A)
10}50% methyl ethyl ketone in n-heptane; (B) 30}70% methyl ethyl ketone in n-heptane; (C) 2.5}15% ethyl acetate in chloroform.
P"Pastinaca sativa; H"Heracleum sphondylium; S"Sium sisarum; L"Libanotis intermedia. (1) Osthol; (2) isopimpinellin; (3)
imperatorin; (4) bergapten; (5) xanthotoxin; (6) umbelliferone. (Reproduced with permission from Glowniak K, Matysik G, Bieganowska
M and Soczewinski E (1986) Chromatographia 22: 307}310.)
Table 3 R f values obtained using chloroform}n-butyl acetate}hexane, 9 : 1 : 15 as eluent
Compound Fluorescence Sweet Bitter Mandarin Grape- Lemon Bergamot Mexican

at 366 nm orange orange fruit lime
Bergamottin Y 0.52 0.52 0.52 0.52

Aurapten V 0.41
5-Geranyloxy-7-methoxycoumarin B 0.39 0.38 0.38
8-Geranyloxypsoralen Y 0.34
Osthol V 0.38 0.38
Bergapten#Epoxybergamottin Y 0.30 0.30
Bergapten Y 0.30 0.30
Citropten B 0.26 0.25 0.25
Meranzin#Isomeranzin V 0.25 0.25
Herniarin B 0.23
Epoxyaurapten V 0.18
? B 0.18
? P 0.17
Oxypeucedanin Y 0.10 0.11
Epoxybergamottin hydrate Y 0.10 0.10
Meranzin hydrate V 0.08 0.08
Byakangelicol Y 0.04 0.04
Polymethoxylated flavones Y 0.04 0.04 0.04 0.04
Polymethoxylated flavones B 0.03 0.03 0.03 0.03
B: Blue; Y: Yellow; V: Violet; P: Pink. (Reproduced with permission from Dugo P, Mondello L, Lamonica G and Dugo G (1996) Journal of
Planar Chromatography 9: 120}125.)
Table 4 Rf values obtained using n-butyl acetate}hexane, 80 : 20 as eluent
Compound Fluorescence Sweet Bitter Mandarin Grape- Lemon Bergamot Mexican

at 366 nm orange orange fruit lime
Bergamottin Y 0.96
Bergamottin#5-Geranyloxy-7- Y#B 0.97 0.97 0.97
methoxycoumarin
8-Geranyloxypsoralen Y 0.92 0.92
Aurapten V 0.90
? B 0.82
? B 0.82
Osthol V 0.74 0.74
Citropten B 0.71 0.71 0.71
? B 0.71
Epoxybergamottin Y 0.70 0.68
Bergapten Y 0.65 0.65 0.66
Bergapten#epoxyayrapten Y#V 0.65
Herniarin V 0.62
Oxypeucedanin Y 0.61 0.60
? Y 0.60
? B 0.56
Byakangelicol Y 0.49 0.49
Isomeranzin V 0.49 0.49
Meranzin V 0.45 0.45
? B 0.41 0.40
Tangeretin P 0.25 0.25 0.25 0.25
Heptamethoxyflavone Y 0.21 0.21 0.21 0.21
Tetra-O-methylscutellarein P 0.20 0.20
Hexamethoxyflavone SB 0.16
Nobiletin Y 0.14 0.14 0.14 0.14
Sinensetin SB 0.10 0.10
Epoxybergamottin hydrate Y 0.10 0.10
Meranzin hydrate V 0.04 0.04
B: Blue; SB: Sky-blue; Y: Yellow; V: Violet; P: Pink. (Reproduced with permission from Dugo P, Mondello L, Lamonica G and Dugo G
(1996) Journal of Planar Chromatography 9: 120}125.)
Figure 6 HPLC chromatogram of oxygen heterocyclic compo-

nents of cold-pressed sweet orange oil. (1) Tangeretin; (2) hep-
Figure 5 HPLC chromatogram of oxygen heterocyclic compo- tamethoxyflavone; (3) nobiletin; (4) tetra-O-methylscutellarein; (5)
nents of genuine cold-pressed lemon oil. Column 25 cm;4.6 mm 3,3,4,5,6,7-hexamethoxyflavone; (6) sinensetin. (Reproduced
with permission from McHale D and Sheridan JB (1989) Journal
internal diameter Zorbax 6 m spherical silica; gradient elution,
solvent A hexane}ethyl acetate (9 : 1), solvent B hexane}ethyl of Essential Oil Research 1: 139}149.)
alcohol (9 : 1), 2}95% B over 25 min; flow rate, 1.5 mL min\1;
sample volume 20 L (20% solution of lemon oil in dichloro-
methane); detection UV absorbance at 315 nm. For peak assign- HPLC Separations
ment, see Table 5. (Reproduced with permission from McHale D
and Sheridan JB (1988) Flavour Fragrance Journal 3: 127}133.) As well as TLC methods the literature reports numer-
ous qualitative and quantitative methods for the anal-
ysis of coumarins and related compounds by both
improved. Figure 4 shows an example of gradient normal- and reversed-phase HPLC. Detection is com-
TLC of coumarins and psoralens of some plant ex- monly performed by UV absorbance, but methods
tracts. which use the highly selective and sensible Suores-
Tables 3 and 4 provide results obtained for the cence detector are also reported.
OPLC analysis applied to seven citrus oils, using silica Two important papers are those of McHale and
gel 60 F254 HPTLC plates with impregnated edges Sheridan of 1988 and 1989 in which the authors
(Sow rate: 0.7 mL min\1). Detection was by UV at developed normal-phase HPLC methods for the ana-
366 nm; time of analysis: 10 min. Because of its ad- lysis of the most common citrus oils, making
vantages, this method has been proposed as a rapid, huge progress in the identiRcation and in the quantit-
preliminary check to evaluate the authenticity of ative determination of oxygen heterocyclic com-
citrus oils. pounds. The Rrst paper refers on the composition of
Table 5 Composition of oxygen heterocyclic fraction of cold-pressed Sicilian lemon oil

reported by McHale and Sheridan (1988)
Peak no. Component Concentration mg L\1 (ppm)
1 Bergamottin 2200
2 5-Geranyloxy-7-methoxycoumarin 1600
3 Isoimperatorin 180
4 5-Isopentenyloxy-7-methoxycoumarin 80
5 Unidentified (UV 7-substituted coumarin) 10
6 Citropten 650
7 8-Geranyloxypsoralen 750
8 Phellopterin # Imperatorin 90#60
9 5-Isopentenyloxy-8-epoxyisopentyloxypsoralen 220
10 Unidentified (UV ill-defined) }
11 Oxypeucedanin 1100
12 Byakangelicol 450
13 Oxypeucedanin hydrate 260
14 Byakangelicin 70
(Reproduced with permission from McHale D and Sheridan JB (1988) Flavour Fragance
Journal 3: 127}133.)
Figure 7 HPLC chromatogram of oxygen heterocyclic compo- Figure 9 HPLC chromatogram of oxygen heterocyclic compo-
nents of cold-pressed bitter orange oil. (1) Osthol; (2) epoxyber- nents of cold-pressed lime oil. (1) Bergamottin; (2) 5-geranyloxy-
7-methoxycoumarin; (3) 5-geranyloxy-8-methoxypsoralen; (4)
gamottin; (3) bergapten; (4) tangeretin; (5) heptamethoxyflavone;
(6) meranzin; (7) isomeranzin#nobiletin. (Reproduced with per- 5-isopentenyloxy-7-methoxycoumarin; (5) 5-isopentenyloxy-8-
mission from McHale D and Sheridan JB (1989) Journal of Essen- methoxypsoralen; (6) citropten; (7) 8-geranyloxypsoralen; (8) her-
niarin; (9) bergapten; (10) isopimpinellin; (11) oxypeucedanin;
tial Oil Research 1: 139}149.)
(12) isobyakangelicol; (13) byakangelicol#heraclenin. (Repro-
duced with permission from McHale D and Sheridan JB (1989)
Journal of Essential Oil Research 1: 139}149.)
cold-pressed lemon oil. Figure 5 shows the HPLC

chromatogram in which 14 components were detec- gamot, sweet orange, bitter orange, grapefruit and
ted and quantitatively determined. The experimental lime oils, as illustrated in Figures 6+9.
conditions are shown in the Rgure legend. Table 5 Table 6 reports the experimental conditions used
reports the quantitative results. for these analyses. The paper reported quantitative
These workers identiRed and quantiRed not only data for all the oils analysed, according to their geo-
the main components, but also those present in lower graphical origin.
amounts. Moreover, they analysed commercial sam- Figure 10 shows a reversed-phase HPLC chrom-
ples of lemon oils, and demonstrated the validity of atogram obtained in 1992 by Ziegler and Spiteller for
the method to detect some adulterations practised to a lemon oil under the following experimental condi-
increase the UV absorbance of oils previously diluted tions: column, Spherisorb ODS2 (C18 } particle size
with distilled ones. The method allows for the detec- 5 m), 250;4.6 mm internal diameter; Sow,
tion of p-dimethylaminobenzoate, grapefruit oil 1 mL min\1; solvents: A, methanol}water}acetonit-
and/or lime oil, that are the most common substances rile (1 : 1.35 : 0.5) B, acetonitrile. Program: 10% B to
used to increase UV absorbance. The paper of 1989
shows normal-phase HPLC chromatograms of ber-
Figure 8 HPLC chromatogram of oxygen heterocyclic compo-

nents of cold-pressed grapefruit oil. (1) Bergamottin; (2) aurapten;
(3) osthol; (4) epoxybergamottin; (5) epoxyaurapten; (6) tanger- Figure 10 Reversed phase HPLC chromatogram of oxygen
etin; (7) heptamethoxyflavone; (8) meranzin; (9) isomeranzin heterocyclic components of cold-pressed lemon oil. For peak
# nobiletin. (Reproduced with permission from McHale D and assignment, see Table 7. (Reproduced with permission from
Sheridan JB (1989) Journal of Essential Oil Research 1: Ziegler H and Spiteller G (1992) Flavour Fragrance Journal 7:
139}149.) 129}139.)
Table 6 HPLC conditions reported by McHale and Sheridan (1989) for the analysis of oxygen heterocyclic compounds of citrus oils
Lemon Lime Mandarin Grape fruit Bitter orange Sweet orange
Eluent (A) hexane}ethyl acetate, 9 : 1 Hexane}ethyl alcohol, 19 : 1 Hexane}ethyl alcohol, 9 : 1

(B) hexane}ethyl alcohol, 9 : 1
Programme From 98 A : 2 B to 5 A : 95 B over 25 min Isocratic Isocratic
Column 6 m Zorbax SIL spherical, 25 cm;4.6 mm internal diameter
Injection volume 20 L of a 20% solution of oil in dichloromethane
Detection UV absorbance at 315 nm
Flow rate 1.5 mL min\1
40% B in 20 min; 40% B to 70% B in 15 min; 70% Citrus oils that show a quite complex composition
B to 90% B in 2 min. Injection: 100 L of a 0.2% of the oxygen heterocyclic fraction are difRcult to
solution of original lemon oil in 10% B and 90% A. analyse either by normal- or reversed-phase HPLC
Detection: UV at 220 and 310 nm. with a single column. In these cases the ‘column
Table 7 lists the 25 components identiRed by switching’ technique can be useful to improve the
HPLC, MS, GC}MS and NMR, 11 of which identi- separation of those critical peaks. This technique has
Red for the Rrst time in a cold-pressed lemon oil as been applied successfully to the analysis of bitter
trace constituents. It is noteworthy that on the orange and grapefruit oils, which show a very similar
basis of the spectroscopic data obtained, bergap- composition, to separate 17 components and, in par-
ten does not result to be present in genuine lemon ticular, to the couple meranzin}isomeranzin. Fig-
oil, in contrast with data previously reported in ures 11 and 12 show the results obtained, together
literature. with the experimental conditions and peak identiRca-
tion.
Table 7 Components identified in the oxygen heterocyclic frac- Most of the data found in the literature refer to the
tion of cold-pressed Sicilian lemon oil by Ziegler and Spiteller
characterization of bergamot oil, and in particular to
Peak no. in Components the determination of 5-methoxypsoralen (bergapten),
HPLC run which is known to have a higher phototoxic action
than other psoralens found in citrus oils. Bergamot oil
1 5-(2,3-Dihydroxyisopentyloxy)-7- is widely used in the cosmetic and pharmaceutical
methoxycoumarinH
HeraclenolH
2 Oxypeucedanin hydrate
3 Byakangelicin
4 Citropten
5 Heraclenin
6 Pabulenol/GosferenolH
7 NeobyakangelicolH
8 Oxypeucedanin
Byakangelicol
5-(2,3-Epoxyisopentyloxy)-7-methoxycoumarinH
9 5-Isopentenyloxy-8-
(2,3dihydroxyisopentyloxy)psoralenH
10 Imperatorin
11 7-IsopentenyloxycoumarinH
12 8-(6,7-Epoxygeranyloxy)psoralenH
13 Phellopterin
14 Isoimperatorin
15 5-Isopentenyloxy-7-methoxycoumarin
16 5-Isopentenyloxy-8-(2,3-
epoxyisopentyloxy)psoralen
17 CnidicinH
18 8-Geranyloxypsoralen
19 AuraptenH Figure 11 HPLC chromatogram of oxygen heterocyclic com-
5-Methoxy-8-geranyloxypsoralenH ponents of cold-pressed bitter orange oil. For peak assignment
20 Bergamottin and experimental conditions see Figure 12. (Reproduced with
21 5-Geranyloxy-7-methoxycoumarin permission from Dugo P, Mondello L, Stagno d’Alcontres I,
Cavazza A and Dugo G (1997) Perfumer and Flavorist 22:
HCompounds previously unknown in cold-pressed lemon oil. 25}30.)
The API technique can use two different interfaces,

the electrospray (ES) or the atmospheric pressure
chemical ionization (APCI), and can give different
information than those obtained with conventional
LC}MS interfaces. Both the techniques are classiRed
as ‘soft’ ionization methods. By varying the voltage of
the sample cone it is possible to obtain different
degrees of fragmentation.
The HPLC}MS technique with the APCI interface
has been applied to the analysis of coumarins of citrus
oils to conRrm the identiRcation of some components
or to obtain more information for those not identiRed
yet. As an example, Figure 13 shows the HPLC}UV
chromatogram of a cold-pressed bergamot oil, com-
pared to the full scan HPLC}MS chromatogram ac-
quired at different cone voltage values.
Figure 14 shows the MS spectra obtained at differ-
Figure 12 HPLC chromatogram of oxygen heterocyclic com- ent cone voltage values for one of the components of
ponents of cold-pressed grapefruit oil obtained in the following bergamot oil (bergamottin). As can be seen, at the
conditions: -porasil column 30 cm;3.9 mm internal diameter lower cone voltage value the (M#H)# ion is visible,
(10 m) for the first 12 min, then the flow was switched to a sec-
while at higher values additional fragmentation
ond column, Zorbax silica 25 cm;4.6 mm internal diameter
(7 m). Eluent A, hexane}ethyl acetate, 9 : 1; eluent B, occurs.
hexane}ethyl alcohol, 9 : 1. 2}95% B over 23 min (2}25 min) with The HPLC}MS technique allowed the conRrma-
a concave gradient, then 20 min isocratic 95% B, flow rate tion of the presence of oxygen heterocyclic com-
1.6 mL min\1; sample volume 20 L (5% solution of oil in pounds previously not identiRed in bergamot oil, such
hexane}ethyl acetate, 75 : 25); detection by UV absorbance at
as tetra-O-methylscutellarein and sinensetin. Fig-
315 nm. Peak assignment: (1) bergamottin; (2) aurapten; (3)
osthol; (4) bergapten; (5) epoxybergamottin; (6) epoxyaurapten; ure 15 shows the HPLC}UV chromatogram at cone
(7) unknown coumarin; (8) meranzin; (9) isomeranzin; (10) un- voltage value of 20 V, and the extracted chromato-
known coumarin 2; (11) tangeretin; (12) 3,3,4,5,6,7,8-hepta- gram at m/z 243 and 273, corresponding to the
methoxyflavone; (13) nobiletin; (14) tetra-O-methylscutellarein; (M#H)# ions of tetra-O-methylscutellarein and
(15) unknown coumarin; (16) epoxybergamottin hydrate; (17)
sinensetin.
meranzin hydrate; (i.s.) internal standard, coumarin. (Reproduced
with permission from Dugo P, Mondello L, Stagno d’Alcontres I,
Cavazza A and Dugo G (1997) Perfumer and Flavorist 22: Supercritical Fluid Chromatography
25}30.)
(SFC)
A few applications of supercritical Suid chromatogra-
industries. Usually, genuine cold-pressed bergamot phy (SFC) to the analysis of citrus coumarins have
oil contains about 2000}3000 ppm of bergapten, but been reported. Even though this technique is less
many industrial processes have been developed with popular in the analysis of citrus coumarins, it is inter-
the aim to reduce its concentration to values of esting to compare the results obtained by SFC to
only a few parts per million. The oils so obtained are those obtained by HPLC. Figure 16 shows a normal
known as ‘bergapten-free’ oils. Table 8 summarizes phase HPLC chromatogram of the polymethoxylated
some of the TLC and HPLC methods proposed for Savones of sweet orange oil (A), compared to the
the quantitative determination of bergapten in be- packed SFC chromatogram obtained for the same oil
rgamot oil. (B).
Table 9 provides the chromatographic conditions
under which the two analyses have been carried out.
HPLC+MS Both the analyses allowed a complete separation of
To obtain more information on the nature and the the six polymethoxylated Savones (PMFs) known to
structure of the components analysed by HPLC, an be present in sweet orange oil. The SFC analysis was
MS detector can be coupled on-line to the HPLC completed in less than 6 min, while the HPLC analy-
system. An innovative interface for the HPLC}MS sis took more than 25 min. This represents a reduc-
coupling is the API (atmospheric pressure ionization), tion in the analysis time by a factor of four. Quantit-
that differs from the traditional interfaces because the ative results obtained with the SFC method compared
ionization takes place at atmospheric pressure. well with those obtained with the HPLC method.
Table 8 TLC and HPLC methods for the analysis of bergapten in bergamot oil.
Technique Stationary phase Mobile phase Detection method Flow rate Sample
HPLC 2 cm;2.3 mm internal Hexane}chloroform, UV abs. at 254 nm 0.25 mL min\1 10 L of a solution of

diameter stainless 75 : 25 P "100 psi 0.05 g of oil in 10 mL of
steel tubing packed CHCl3
with 37-50 Corasil II
TLC Silica gel plates F-254 Isoctane}ethyl acetate, UV abs. at 254 nm Oil diluted 1 : 10 in CHCl3
83 : 17
TLC Silica gel plates F-254 Cyclohexane} UV abs. at 254 and
(20 m) ethyl acetate} 366 nm
acetic acid,
80 : 20 : 2
TLC 10;10 cm Silica gel Petroleum ether or Densitometer at 254 or
HPTLC plates 60 benzene followed by 308 nm or Fluore-
F-254 (5 m) cyclohexane}ethyl scence (exc 330 nm,
acetate,75 : 25 em 450 nm)
TLC 10;10 cm RP 18 TLC Methanol}water, 80 : 20 Densitometer at 254 or Oil diluted in methanol
plates (7 m) 308 nm
HPLC Lichrosorb Si 60 (5 m) Heptane}isopropanol, UV abs at 254 nm 0.85 mL min\1,
23 cm;4.35 mm 93 : 7 P"23 bar
internal diameter
HPLC Lichrosorb RP 18 (5 m) Methanol}water, 9 : 1 UV abs at 254 nm 1.1 mL min\1,
17 cm;4.35 mm P"24 bar
internal diameter
HPLC Lichrosorb Si 60 (5 m) Hexane}ethyl acetate} UV absorbance at 1 mL min\1 Undiluted for oil with less
25 cm;4.6 mm propan-2-ol, 88 : 10 : 2 305 nm than 40 mg L\1 of
internal diameter (isocratic) bergapten, or diluted in
the range 20}40 mg L\1
with CHCl3 ) 20 L inj
Figure 13 (A) HPLC}UV chromatogram and (B), (C) and (D) total ion current (TIC) HPLC chromatograms acquired at cone voltage
values of 60, 40 and 20 V, respectively, of coumarin fraction of a genuine bergamot essential oil. (1) Bergamottin; (2) 5-geranyloxy-7-
methoxycoumarin; (3) citropten; (4) bergapten; (a) tetra-O-methylscutellarein; (b) sinensetin; (i.s.) internal standard.
Figure 14 Cone voltage fragmentation of bergamottin using APCI ionization.
Conclusion thermore, the achievement of structural information

on unknown constituents of a complex mixture is
Citrus essential oils show a characteristic composi- a strategic element for guiding an efRcient and selec-
tion of their oxygen heterocyclic fraction. These com- tive isolation procedure.
ponents make it possible to differentiate the indi- A big advantage can be achieved by using hyphen-
vidual oils and to detect mixtures or mutual contami- ated techniques. In the last few years, LC-MS is be-
nation. Quantitative analysis is often necessary to coming more and more popular, because if the intro-
assess authenticity or geographical origin of an oil. duction of API (atmospheric pressure ionization)
Sometimes, quantitative analysis has to be preceded techniques as a means for mass spectrometric sample
by isolation of single components for structure elu- introduction. This interface permits a highly selective
cidation and also to obtain pure standard compounds and sensitive detection method, to be obtained and
to be used for preparation of standard solutions, since use of HPLC-MS in routine analysis. Recently, LC-
they are not always commercially available. More- NMR has been introduced as another powerful com-
over, these components possess numerous pharmaco- plementary technique for on-line structural identiRca-
logical activities, so isolation may be necessary for tion, even though it is much less sensitive than LC-
testing speciRc biological activities. MS. Application of planar chromatography coupled
Both analytical and preparative analyses can be with mass spectrometric (MS) or Fourier transformed
carried out with planar and column liquid chromato- infrared (FT-IR) have also been developed.
graphic methods (TLC, OPLC, HPLC). Usually, the Looking to the future, it is reasonable to expect
preparative separations, that may be long and labori- a much wider use of hyphenated techniques that will
ous, are followed by further puriRcation steps and by allow the rapid structural determination of constitu-
spectroscopic analyses for identiRcation. ents of complex matrices requiring only a small
Working with natural products, efRcient detection amount of samples, and the use of shorter HPLC
and rapid characterization are often essential. Fur- columns packed with smaller particles that will
Figure 15 (A) HPLC}UV chromatogram, (C) TIC HPLC chromatogram acquired at cone voltage of 20 V and (B) TIC HPLC
chromatogram extracted at m/z 343#373, of coumarin fraction of a genuine bergamot oil.
Figure 16 (A) HPLC chromatogram and (B) packed SFC chromatogram of polymethoxylated flavones of sweet orange oil. 1"(A)
tangeretin; 2"(B) heptamethoxyflavone; 3"(C) nobiletin; 4"(D) tetra-O-methylscutellarein; 5"(E) 3,3,4,5,6,7-hexamethoxy-
flavone; 6"(F) sinensetin. ((A) Reproduced with permission from Dugo P, Mondello L, Cogliandro E, Stagno d’Alcontres I and
Controneo A (1994) Flavour and Fragrance Journal 9: 105}111 and (B) reproduced with permission from Dugo P, Mondello L, Dugo G,
et al. (1996) Journal of Agricultural and Food Chemistry 44: 3900}3905.)
2454 III / CLINICAL APPLICATIONS / Capillary Electrophoresis
Table 9 Experimental conditions of the HPLC and of the SFC analyses of polymethoxylated flavones of sweet orange oil
HPLC SFC
Column Zorbax silica, 25 cm;4.6 mm internal diameter S5W uncoated silica, 25 cm;4.6 mm internal diameter
(7 m) (5 m)
Eluent Hexane}ethyl acetate, 95 : 5 CO2 modified with small amounts of methanol
Flow rate 1.6 mL min\1 2 mL min\1 for 4 min, then gradient of 2 up to 5 mL min\1
thereafter held constant
Programme Isocratic P"100 atm, T"403C; modifier, 1.5% min\1 from 10% to
30% thereafter held constant
Injected amount 20 L of a 5% solution of oil in ethyl acetate 100 L of a solution obtained by diluting 0.71 g of oil to
20 mL of ethyl acetate
Detection UV absorbance at 315 nm UV absorbance at 315 nm
provide faster separations with the same resolution as Di Giacomo A and Mincione B (1994) Gli Olii Essenziali
that observed in longer columns packed with particles Agrumari in Italia. Reggio Calabria: La Ruffa.
of larger diameter. Dugo P, Mondello L, Stagno d’Alcontres I, Cavazza A and
Dugo G (1997) Oxygen heterocyclic compounds of cit-
rus essential oils. Perfumer and Flavorist 22: 25}30.
See also: II/Chromatography: Liquid: Detectors:
McHale D and Sheridan JB (1988) Detection of adulter-
Mass Spectrometry. Chromatography: Thin-Layer
ation of cold-pressed lemon oil. Flavour and Fragrance
(Planar): Densitometry and Image Analysis; Modes of
Journal 3: 127}133.
Development: Forced Flow, Over Pressured Layer
McHale D and Sheridan JB (1989) The oxygen heterocyclic
Chromatography and Centrifugal; Preparative Thin-Layer
compounds of citrus peel oils. Journal of Essential Oil
(Planar) Chromatography. III/Essential Oils: Gas
Research 1: 139}149.
Chromatography; Thin-Layer (Planar) Chromatography;
Murray RDH, Mendez J and Brown SA (1982) The Natural
Distillation.
Coumarins, Occurrence, Chemistry and Biochemistry.
Chichester: John Wiley.
Further Reading Proceedings of the Symposium Cumarine: Ricerca ed Ap-
plicazioni, Padova, Italy 20}22 September 1990.
Di Giacomo A and Calvarano M (1978) Il Contenuto di Padova, Italy: Imprimitur.
Bergaptene nell’Essenza di Bergamotto Estratta Sherma J and Fried B (1996) Handbook of Thin-Layer
a Freddo. Essenze Derivati Agrumari 48: 51}83. Chromatography. New York: Marcel Dekker.
CLINICAL APPLICATIONS
literature. Such reviews can be divided into:

Capillary Electrophoresis 1. Broad-coverage reviews, such as those of Leh-
mann et al. (1997), Perrett (1999; CZE in clinical
P. G. Righetti, University of Verona, Verona, Italy chemistry) and Guzman et al. (1997; dedicated
C. Gelfi, ITBA, CNR, Milan, Italy also to on-line analyte concentration and micro-
reaction). Also of interest are special issues of the
Copyright ^ 2000 Academic Press Journal of Chromatography B dedicated to CZE
in the life sciences (Krstulovic 1997) and of Elec-
trophoresis devoted to CZE in the clinical sciences
(Landers 1997) and in forensic science (McCord,
Introduction 1998).
The area of clinical applications of capillary elec- 2. Specialized reviews, such as those of Thormann
trophoresis (CZE) is such a rapidly growing Reld that et al. (1996, 1997; drug analysis in body Suids);
it would be impossible here to cover it in detail. We Lurie (1996; analysis of seized drugs), Hong and
thus offer a list of major reviews to which the reader Baldwin (1997; metabolite proRling in human
is referred for a more comprehensive coverage of the urine), Righetti and GelR (1997a,b, 1998; CZE of
III / CLINICAL APPLICATIONS / Capillary Electrophoresis 2455
DNA for molecular diagnostics), Jellum et al. ductivity of such a solution. Some areas or interest
(1996, analysis of urinary diagnostic metabolites where CZE offers unique resolution and sample
and serum proteins), Lazaruk et al. (1998; quantiRcation are discussed below.
genotyping of forensic short tandem repeat
systems). Drug Analysis
With the more efRcient therapeutic application of
CZE: Some Basic Concepts Related to various drugs and the necessity for screening and
Clinical Chemistry conRrmation of drugs in body Suids for diagnostic
and research purposes, there has evolved a need for
An important aspect of CZE relevant to clinical
reliable analytical procedures. CZE is becoming the
chemistry is the paradox by which one of the noted
method of choice for drug monitoring in body Suids,
advantages of CZE, namely the small volume of the
including plasma, serum, saliva and urine. Some
capillary, also leads to a signiRcant drawback, i.e. the
relevant data are given in Tables 1+3.
minute amount (3}10 nL) of sample introduced,
which results in poor concentration limits of detec-
Pro\ling Clinically Important Metabolites in Urines
tion, one to two orders of magnitude lower than in
high-performance liquid chromatography (HPLC). There are approximately 300 known metabolic dis-
Thus, preconcentration techniques are often required orders many of which, if not treated, have serious and
for compounds present in very low concentrations in sometimes life-threatening consequences. Generally,
biological Suids. These include transient isotacho- the diagnosis of these disorders is made or conRrmed
phoresis, analyte stacking, Reld-ampliRed sample by identiRcation and quantiRcation of characteristic
injection (all on-capillary techniques) or off- metabolite(s) occurring in body Suids as a result of
column preconcentration techniques, such as liquid} the particular enzyme deRciency involved in each
liquid or liquid}solid extraction. Alternative disease. In addition, once the diagnosis has been
on-column methods include miniaturized solid-phase made, the effectiveness of subsequent therapy can
extraction with cartridges containing reversed-phase also be evaluated by analogous monitoring of the
HPLC packing materials or with impregnated mem- levels of the same metabolic markers. Recently, CZE
branes. When analysing small analytes or drugs in has become an attractive tool for monitoring clinic-
sera, it is often necessary to deproteinize the sample. ally important metabolic markers, including alditols,
This can be efRciently achieved by extracting sera carbohydrates and amino acids, which can be proRled
with 60% acetonitrile, which accomplishes two directly in urine with electrochemical (EC) detection
tasks: protein precipitation and analyte stacking upon at a copper electrode. EC detection allows sensitivity
electrokinetic injection because of the very low con- down to the femtomolar level and can be performed
Table 1 Selected validated CZE/MEKCa assays for drugsb
Drug Body fluid Therapeutic range Calibration range CE method Sample preparation/ Reference
(g mL\1) (g m L\1) detection c assay
Theophylline Serum 8}20 2.0}28.3 MEKC DSI/275 EMIT

Phenobarbital Serum 15}40 5.3}52.6 MEKC DSI/245 EMIT
Ethosuximide Serum 40}100 19.8}99.2 MEKC DSI/220 FPIA
Flucytosine Serum 20}80 20}123 MEKC DSI/210 Bioassay
Antipyrine Plasma * 1.0}40 MEKC DSI/240 HPLC
Antipyrine Saliva * 1.9}66 MEKC DSI/260 HPLC
Felbamate Serum 15}135 5.0}160 MEKC DSI/214 HPLC
Pentobarbital Serum * 10}100 CZE PPI/253 HPLC
Thiopental Serum * 2.0}60 MEKC EXI/290 HPLC
Bupivacaine Drain * 0.5}20 CZE EXI/200 GC
Cicletanine Plasma * 0.01}1.0 MEKC EXI/214 HPLC
Retinol Serum * 0.003}0.035 CZE UF/LIF HPLC
a
MEKC, micellar electrokinetic capillary chromatography; CZE, capillary zone electrophoresis; DSI, direct sample injection; EXI, extract
injection; PPI, injection of supernatant after protein precipitation; EMIT, enzyme-multiplied immunoassay technique; FPIA, fluores-
cence polarization immunoassay; GC, gas chromatography; HPLC, high-performance liquid chromatography; UF, ultrafiltration; LIF,
laser-induced fluorescence detection with 325 nm excitation and 465 nm emission.
b
Reprinted from Thormann W, Zhang CX and Schmutz A (1996) Therapeutic Drug Monitor 18, 506}520, with permission.
c
The number represents the detection wavelength.
Table 2 Selected CZE/MEKCa screening confirmation assays for illicit, abused and banned drugsb
Drugs, drug classes Body fluid, tissue CZE method Sample preparation Reference assay
Barbiturates Urine, serum MEKC DSI, EXI EMIT

Salicylate, paracetamol, Urine, serum MEKC, CZE DSI, EXI FPIA, EMIT, UF
antiepileptics
11-Nor--tetrahydro Urine MEKC EXI FPIA
cannabinol-9-carboxylic acid
Methadone and its primary Urine CZE DSI, EXI EMIT, GC-MS
metabolite
Benzodiazepines Urine MEKC EXI EMIT, GC-MS
Benzoylecgonine, opioids, Urine MEKC EXI EMIT
methaqualone, amphetamines
Cocaine and all above Urine MEKC, CZE EXI EMIT, FPIA, GC-MS
mentioned classes
Cocaine, morphine Hair MEKC EXI HPLC
-Blockers Serum MEKC EXI *
Diuretics Urine, serum CZE EXI GC-MS
a
For abbreviations, see Table 1.
b
Reprinted from Thormann W, Zhang CX and Schmutz A (1996) Therapeutic Drug Moniter 18, 506}520, with permission.
directly on urine without extensive sample clean-up following metabolites: orotic acid, pyroglutamate,
or analyte derivatization. Some examples are given in adenylosuccinate and propionic acid, for the follow-
Table 4. Other interesting data on proRling the ing diseases: HHH-syndrome (hyperornithinemia-
hyperammonemia-homocitrullinuria), glutathione
Table 3 Retention of anabolic steroids relative to testosteronea deRciency, adenylosuccinase deRciency and propionyl
CoA carboxylase deRciency, respectively, can be
Compound b MEKC c HPLC GC found in Jellum et al. (1997).
Fluoxymesterone 0.925 0.78 1.50
Boldenone 0.964 0.74 1.05 Pro\ling Proteins in Biological Matrices
Nandrolone 0.979 0.84 0.91
Methandrostenolone 0.985 0.86 1.12 Separation and quantiRcation of distinct proteins
Testosterone 1.00 1.00 1.00 from biological matrices is another goal now being
Methyltestosterone 1.02 1.17 1.05 accomplished by CZE. A list of some major proteins
Methandriol 1.06 1.25 0.89 of importance for clinical diagnosis is given in
Stanolone 1.07 1.25 0.89
Table 5. In some cases, immunosubtraction can be an
Boldenone acetate 1.12 1.46 1.27
Stanozolol 1.16 1.69 1.68 efRcient way of quantifying some protein families by
Testosterone acetate 1.17 1.76 1.21 CZE. A typical example is the quantiRcation of speci-
Nandrolone propionate 1.22 1.88 1.29 Rc immunoglobulin subclasses by sequential
Danazol 1.23 1.52 * immunosubtraction. The sample is exposed to Rve
Clostebol acetate 1.24 1.90 *
different Sepharose supports, each containing an im-
Testosterone propionate 1.26 2.01 1.43
Methandriol 3 acetate 1.26 2.13 1.10 munoglobulin-speciRc binder. Three of them are spe-
Testosterone isobutyrate 1.35 2.17 1.54 ciRc for the heavy chains IgG, IgA and IgM and two
Nandrolone phenylpropionate 1.44 2.25 2.28 are speciRc for the light chains or . After incuba-
Testosterone cypionate 1.64 2.63 2.19 tion and sedimentation, the treated samples and an
Testosterone enanthate 1.69 2.60 1.92
untreated control are separated by CZE. Six elec-
Methandriol dipropionate 1.81 2.98 1.70
Nandrolone decanoate 2.06 2.87 2.26 tropherograms are generated per sample. The class
Boldenone undecylenate 2.20 2.73 2.62 and type of monoclonal component can be deter-
Testosterone undecanoate 2.36 3.18 2.56 mined by overlaying electropherograms from
Oxymetholone * * 1.28 before and after immunosubtraction.
Oxandrolone * * 1.17
Testosterone isocaproate * * 1.77
Testosterone decanoate * * 2.36
CZE Separations of Clinically Relevant
a
Reprinted from Lurie IS (1996) International Laboratory, with Diagnostic DNA
permission.
b
In order of increasing retention times. A number of applications of CZE in sieving liquid
c
For abbreviations, see Table 1. polymers (notably linear polyacrylamides and
Table 4 Normal constituents of human urine and expected response for CZE at a Cu electrodea
Compound Related metabolic Normal concentrations (M)b Migration time in Detection limit (M)
disorder 0.1 N NaOH (min)
Alditols
Erythritol * 608 12.6 0.5
Inositol Diabetes, renal failure 357 12.6 0.3
Ribitol * 35 12.8 *
Xylitol * 35 12.8 *
Arabitol * 195 12.8 *
Glucitol * 35 12.9 0.5
Mannitol Diabetes 104 14.6 0.5
Carbohydrates
Sucrose * 43 16.9 1
Lactose * 15 19.6 1
Fucose * 97 19.6 *
Galactose Galactosaemia 29 21.3 1
Glucose Diabetes 262 22.4 1
Rhamnose * 115 22.8 *
Arabinose Pentosuria 89 23.4 *
Fructose Fructosuria 72 23.4 1
Xylose Pentosuria 53 25.4 *
Ribose Pentosuria 31 26.2 1
Amino acids
Lysine Hyperlysinaemia 328 31.6 4
Threonine Aminoaciduria 183 35.4 2
Histidine Histidinaemia 860 43.1 1
Others
Creatinine Muscle and renal disease 8800 12.8 80
Uric acid Gout 2093 '60 1.6
a
Reprinted from Hong J and Baldwin RP (1997) Journal of Capillary Electrophoresis, 4, 65}71, with permission.
b
Calculated assuming that the average urine volume for a 24-h period is 1.5 L.
celluloses) for the analysis of polymerase chain reac- applications, divided into four classes: human gen-
tion (PCR) products of clinically relevant, diagnostic etics, quantitative gene dosage, microbiology/virol-
DNA have been reported. Table 6 lists some major ogy and forensic medicine.
Table 5 Survey of CZE separations of proteins in biological matricesa
Proteins Matrix CZEb mode Detection mode
Immunoglobulin G, transferrin, albumin, CSF CZE UV 185 nm

prealbumin, -trace proteins
Apolipoprotein A-I, A-II, B100, B48, C-III Serum MECC UV 190 nm
and E
Lipoprotein subfractions (HDL, VLDL, Serum CITP Vis 570 nm
IDL, LDL)
Leucine aminopeptidase Serum, urine CZE LIF
Cerebrospinal fluid proteins CSF CZE UF 200 nm
Myoglobin Urine, tissue CZE Vis 405 nm
Albumin, 1-acidic glycoprotein, transferrin, Urine CZE UV 200 nm
-microglobulin, immunoglobulin light chains
Monoclonal antibodies (anti-TNF, anti-CEA) Serum containing culture medium CZE, CIEF, CGE UV 200 nm
Imidodipeptides (prolidase deficiency) Urine CZE UV 269 nm
Cathepsin D Breast tissue CZE UV 214 nm
Hemoglobin variants Plasma CZE, CIEF UV 210 nm
36 low side Mr proteins, cut-off 30 and 5 kDa Seminal, vaginal fluids, serum, saliva CZE UV 214 nm
a
Reprinted from Lehmann R, Voelter W and Liebich HM (1997) Journal of Chromatography B 697, 37}66, with permission.
b
Abbreviations: CIEF, capillary isoelectric focusing; CGE, capillary gel electrophoresis; CITP, capillary isotachophoresis; CSF,
cerebrospinal fluid; CEA, carcino-embryonic antigen; TNF, tumour necrosis factor, HDL, high density lipoproteins; LDL, low density
lipoproteins; VLDL, very low density lipoproteins.
Table 6 Survey of selected capillary gel electrophoretic separations of clinically relevant diagnostic DNAa
Disease DNA amplification Sieving polymers Detection
Human genetics
Cystic fibrosis Allele specific PCR and restriction 6% linear PAAb UV 254 nm
digest of PCR products (deletion)
Cystic fibrosis PCR, F508 6% linear PAA UV 254 nm
Cystic fibrosis PCR/GATT microsatellites 6% linear PAA UV 254 nm
Cystic fibrosis Point mutants, TGCE 8% linear poly(AAEE) UV 254 nm
Duchenne/Becker muscular dystrophy PCR multiplex reaction 6}10% linear PAA UV 260 nm
Dystrophin gene RFLP 0.5% HPMC UV 254 nm
Thalassaemia Point mutants, TGCE 4% poly(AAP), 1.5% HEC UV 254 nm
Congenital adrenal hyperplasia PCR deletion 6% linear PAA UV 254 nm
Androgen insensitivity syndrome CAG triplet analysis 6% linear PAA UV 254 nm
Kennedy’s disease CAG triplet expansion 8% poly(AAEE) UV 254 nm
N-ras gene (human cancer) Point mutants SSCP 8% PAA UV 260 nm
ERBB2 oncogene RFLP 0.5% HPMC UV 260 nm
TX gene PCR 3% PAA LIF
P-53 SSCP 4% PAA UV 260 nm
P-53 SSCP 2% PAA LIF
Cancer (microsatellites instability) PCR Bio-Rad sieving polymer UV 260 nm
Apolipoprotein B gene PCR/VNTR 0.7% MC UV 260 nm
Apolipoprotein E gene PCR/RFLP 3% T PAA UV 260 nm
Apolipoprotein E gene PCR/RFLP Beckman e/CAP LIF
Medium chain AcylCoA PCR/allele specific Polyacrylamide gel LIF
dehydrogenase deficiency
von Willebrand Factor gene PCR/VNTR 1% HEC LIF
Fetal DNA (Y-chromosome) PCR Beckman dsDNA 1000 gel buffer LIF
Quantitative gene dosage

Down’s Syndrome Quantitative PCR 8% PAA UV 254 nm
Rh D/d genotyping Quantitative PCR 8% PAA UV 254 nm
Follicular lymphomas Competitive PCR 4% PAA UV 260 nm
Basic fibroblast growth factor Competitive RT-PCR 6% PAA UV 254 nm
Microbiology/virology
Mycobacterium tuberculosis SSCP and ddF 1% HEC or 3% T, 0.5%C PAA gel LIF
Hepatitis C virus RT-PCR 1% HEC LIF
Polio Virus RT-PCR 3% T linear PAA UV 254 nm
HIV-1 RT-PCR 3% linear PAA LIF
Mitochondrial DNA PCR 1% HEC LIF
Mitochondrial DNA PCR 0.5% MC LIF
VNTRs at locus D1S80 PCR/VNTR 0.5% HEC LIF
VNTRs at locus D1S80 PCR/VNTR 0.5% MC LIF
VNTRs at locus HUMTH01 PCR/VNTR 1% HEC LIF
VNTRs at locus HUMTH01 PCR/VNTR 3% T, 3% C gel UV 260 nm
Therapeutic DNA
Antisense oligonucleotides 18% PAA MALDI-MS LIF
Antisense oligonucleotides 10% PAA, isoelectric His UV 254 nm
Antisense oligonucleotides 10% T PAA, pH gradient UV 254 nm
a
Reprinted from Righetti PG and Gelfi C (1997), with permission.
b
Abbreviations: PAA, polyacrylamide; VNMTR, variable number of tandem repeats; HEC, hydroxyethyl cellulose; MC, methyl cellulose;
SSCP, single strand chain polymorphism; RT, reverse transcription; HPMS, hydroxypropyl methyl cellulose; AAP, acryloyl amino
propanol; AAEE, acryloyl amino ethoxy ethanol, TGCE, temperature gradient capillary electrophoresis; RFLP, restriction fragment
length polymorphism; LIF, laser induced fluorescence.
Examples of Some Separations tions of CZE. From our experience in DNA separ-
ations, we offer here a few examples pertaining to the
It is quite difRcult to compress in such a few pages the screening for human genetic diseases. Figure 1 dis-
vast literature in the Reld of clinico-chemical applica- plays the CZE analysis of a multiplex PCR for the
Figure 1 Simultaneous detection of F508, G542X, N1303K and 1717-1GPA mutations in cystic fibrosis by CZE in polymer
networks. Traces: A, patient carrying the mutations 1717-1GPA and F508; B, patient affected by the G542X/N1303K mutations;
A#B, artificial mixture of the amplified DNA fragments of patients A and B. Conditions: 100 m ID, 37 cm long capillary, filled with
a viscous solution of linear 6% T polyacrylamide in 100 mmol L\1 TBE (Tris-borate-EDTA) buffer, 10 mol L\1 ethidium bromide, pH
8.3; Run at 165 V cm\1 with detection at 254 nm; electrophoretic sample injection at 165 V cm\1 for 8 s. Reprinted from Gelfi C,
Righetti PG, Magnani C, Cremonesi L and Ferrari M (1994) Clinica Chimica Acta 229, 181}189, with permission from Elsevier Science.
simultaneous detection of four mutations, F508, provide an insight on the unique resolving power and
G542X, N1303K and 1717-1GPA in cystic Rbrosis capability of CZE as applied to problem solving in
(CF). This is an interesting example, in that it offers clinical chemistry.
a fast and reliable method for the simultaneous detec-
tion of mutations which are predominant in a given
population. In Italy, a survey of 391 CF patients,
Conclusions and Future Horizons
originating from all geographical regions, revealed Compared with HPLC and gas chromatography,
that F508 (53% of CF chromosomes), G542X CZE has some distinct advantages, such as small
(4%), 1717-1GPA (4%) and N1303K (4%) are the sample size, minimal sample preparation, use of very
most frequent mutations, accounting for 65% of all small amounts of organic solvents and inexpensive
molecular defects pertaining to CF. Figure 2 gives the chemicals, ease of buffer change and method develop-
CZE analysis for the Duchenne (DMD) and Becker ment and low cost of capillary columns. Elec-
(BMD) muscular dystrophies, which represent the trokinetic capillary assays are complementary to the
two most common myopathies described to date. widely employed immunoassays. For the widespread
They are given the names of Chamberlain and Beggs adoption of CE in routine laboratories, a number of
since these two scientists proposed two PCR assays improvements should be made. With the availability
(each based on co-ampliRcation of nine dystrophin of instrumentation comprising multiple capillaries in
gene exons) allowing for the detection of over 98% parallel, sample throughputs comparable to those ob-
DMD/BMD deletions. Thus, a method attempting tained in automated immunoassays should be poss-
simultaneous analysis of DMD/BMD should offer ible. The same goal will be reached with the
unambiguous resolution and identiRcation of 18 frag- availability of chip-based instrumentation, i.e. CZE
ments ranging in size from c. 100}500 bp. This is in on a glass chip on which separation channels,
fact achieved in the lower trace of Figure 2 (repres- a picolitre sample injector and solute detection are
enting a healthy individual). The upper trace shows combined on an area of a few cm2. In this approach,
a patient affected by muscular dystrophy, in which Suid Sow is driven electrokinetically (thus via a plug,
four fragments (196, 202, 331 and 357 bp) are miss- not a laminar Sow) through a network of intersecting
ing. We hope that these two examples, albeit limited, small channels fabricated on planar glass substrates
Figure 2 Screening for Duchenne (DMD) and Becker (BMD) muscular dystrophies by CZE in sieving liquid polymers. The upper
trace represents the separation of 14 exons of modified deleted Chamberlains’ and Beggs’ mixed multiplex. The lower electrophero-
gram shows the separation of 18 exons of modified nondeleted Chamberlains’ and Beggs’ multiplex. All runs were in a 32 cm long,
75 m internal diameter capillary, filled with short-chain polyacrylamide, obtained by chain transfer at 703C, in 89 mM TBE buffer, pH
8.3. Run: 165 V cm\1 with detection at 254 nm. Sample injection: 100 V cm\1 for 25 s. Reprinted from Gelfi C, Orsi A, Leoncini F,
Righetti PG, Spiga I, Carrera P and Ferrari M (1995) BioTechniques 19, 254}263, with permission from Elsevier Science.
by photolithographic masking and chemical etching Krstulovic AM (guest ed.) (1997) Capillary electrophoresis
techniques and formed by bonding the etched sub- in the clinical sciences. Journal of Chromatography
strate to a plain glass plate. Capillaries 30}70 m B 697: 1}289.
wide, about 10 m high and a few centimetres long Landers JP (guest ed.) (1997) Capillary electrophoresis in
have been shown to provide analytical runs in a few the clinical sciences. Electrophoresis 18: 1707}1906.
Lazaruk K, Walsh PS, Oaks F, Gilbert D, Rosenblum BB,
seconds.
Menchen S, Scheibler D, Wenz HM, Holt C and Wallin
J (1998) Genotyping of forensic short tandem repeat
Further Reading (STR) systems based on sizing precision in a capil-
lary electrophoresis instrument. Electrophoresis 19:
Guzman NA, Park SS, Schaufelberger D, Hernandez L, 86}93.
Paez X, Rada P, Tomlison AJ and Naylor S (1997) New Lehmann R, Voelter W and Liebich HM (1997) Capillary
approaches in clinical chemistry: on-line analyte concen- zone electrophoresis in clinical chemistry. Journal of
tration and microreaction capillary electrophoresis for Chromatography B 697: 37}66.
determination of drugs, metabolic intermediates and Lurie IS (1996) Applications of capillary zone electrophor-
biopolymers in biological Suids. Journal of Chromato- esis to the analysis of seized drugs. International Labor-
graphy B 697: 37}66. atory 24: 21}29.
Hong J and Baldwin RP (1997) ProRling clinically impor- McCord BR (1998) Capillary Electrophoresis in Forensic
tant metabolites in human urine by capillary elec- Science. Electrophoresis 19: 1}126.
trophoresis and electrochemical detection. Journal of Perrett D (1999) Capillary zone electrophoresis in clinical
Capillary Electrophoresis 4: 65}71. chemistry. Annals of Clinical Biochemistry 36:
Jellum E, Dollekamp H and Blessum C (1997) Capillary 133}150.
electrophoresis for clinical problem solving: analysis of Righetti PG and GelR C (1997a) Capillary zone elec-
urinary diagnostic metabolites and serum proteins. trophoresis of DNA for molecular diagnostics. Elec-
Journal of Chromatography B 683: 55}65. trophoresis 18: 1709}1714.
III / CLINICAL APPLICATIONS / Electrophoresis 2461
Righetti PG and GelR C (1997b) Non-isocratic capillary Thormann W (1997) Drug monitoring by capillary elec-
electrophoresis for detection of DNA point mutations. trophoresis. In Wong SHY, Sunshine I (Eds) Analytical
Journal of Chromatography B 697: 195}205. Therapeutic Drug Monitoring, pp. 1}19. Boca Raton:
Righetti PG and GelR C (1998) Analysis of clinically- CRC Press.
relevant, diagnostic DNA by capillary zone and Thormann W, Zhang CX and Schmutz A (1996) Capillary
double-gradient gel slab electrophoresis. Journal of zone electrophoresis for drug analysis in body Suids.
Chromatography A 806: 97}112. Therapeutic Drug Monitor 18: 506}520.
Electrophoresis
J.-D. Tissot, A. Layer and P. Schneider, automate. Serum protein electrophoresis is widely
Fondation CRS, Lausanne, Switzerland used in clinical laboratories, especially for the evalu-
H. Henry, Centre Hospitalier Universitaire Vaudois, ation of changes in proteins associated with inSam-
Lausanne, Switzerland mation, liver or kidney diseases as well as for the
detection and identiRcation of paraproteins. Tradi-
Copyright ^ 2000 Academic Press tional clinical electrophoretic procedures are manual
methods that use agarose gels or cellulose acetate
membranes as the separation bed. Quantitation of the
Introduction Rve major serum fractions is done by densitometric
scanning of the gel or the membrane. Clinical inter-
It is somewhat arbitrary to outline electrophoretic
pretation is based on the alteration of the content of
applications that are used in a routine clinical labor-
one or more of the Rve fractions. Agarose, as support-
atory because they are highly dependent on the speci-
ing medium for protein electrophoresis, has been re-
Rcity of each application. Nowadays, a multitude of
ported to give better resolution as well as to allow
electrophoretic methods are routinely used in various
better detection of paraproteins than is cellulose
clinical laboratories, according to the speciRc re-
acetate. Semiautomated agarose electrophoresis and
search being developed. More and more sophisticated
immunoRxation can be performed with various
methods are needed to resolve speciRc clinical prob-
commercially available systems. No differences be-
lems, the reason why specialization of laboratories is
tween manual and semiautomated methods have
mandatory in order to provide accurate results at the
been seen with respect to paraprotein identiRcation.
lowest possible cost. In addition, quality control is of
Over the last few years, capillary zone electrophor-
major importance. Therefore, automation is pro-
esis (CZE) has emerged as a powerful new tool for
gressively introduced in all the steps involved in
rapid separation of various biopolymers, including
the analytical process, and only a few manual
proteins. Separation by this technique depends on the
methods will survive in the future. However, highly
electrophoretic mobility of the analyte and the elec-
sophisticated electrophoretic techniques should be
troosmotic Sow of the bulk solution, and can be
maintained and developed in a limited number of
easily automated. Direct quantitation of proteins via
specialized laboratories, in order to resolve the differ-
peptide bonds is possible using UV detection at
ent problems that are encountered in clinical
214 nm. Separation patterns obtained by CZE are
medicine. Here, we present selected examples of
similar to those obtained after densitometric scanning
electrophoretic techniques that are employed in the
of cellulose acetate membrane electrophoresis or
clinical laboratory.
agarose gel electrophoresis. Recently, dedicated auto-
mated systems for the routine analysis of human
serum proteins in clinical laboratories have become
Serum Protein Electrophoresis commercially available. High sample throughput is
Electrophoretic techniques for the separation of attained due to the presence of several fused-silica
human serum proteins have been used for Rfty years. capillaries, which allow the simultaneous analysis
Resolution has been improved by the use of support of different samples. Several studies have clearly
media such as paper, starch gel, cellulose acetate, shown that the electrophoretic patterns and the clini-
agarose, and polyacrylamide gels, which have ren- cal information obtained by CZE are comparable
dered electrophoretic methods very popular in the with the data obtained by classical methods. In addi-
diagnostic area. However, many of those methods tion, the method is also suited to detect monoclonal
have remained labour intensive, being difRcult to gammopathies. Multicapillary instruments have been
2462 III / CLINICAL APPLICATIONS / Electrophoresis
designed for automation of both routine serum pro- with acute myocardial infarction, unstable angina,
tein electrophoresis and for monoclonal component and thrombogenic carotid atherosclerosis. The
typing by subtraction (immunoRxation electrophor- method of study of LDL, which uses ion exchange
esis } immunosubtraction). In this latter technique, chromotography of LDL isolated by ultracentrifuga-
CZE is performed on the supernatant of serum sam- tion, requires particular care to avoid artefacts due to
ples that have reacted with Sepharose beads coated in vitro auto-oxidation of the LDL and is time con-
with an immunospeciRc binder (IgG, IgA, IgM, and suming. HDL are a family of protein}lipid complexes
). The low detection limit of the technique (0.5 g L\1 that play a central role in cholesterol transport. HDL
for IgG, 0.75 g L\1 for IgA and IgM) contributes to collectively contain apolipoproteins (apo) A-I and
the ability of CZE to detect small monoclonal gam- A-II as major protein components, together with
mopathies. apoC, E, and A-IV. Changes in HDL composition
occur during normal metabolism, as the result of
a particular genetic proRle, or as a consequence of
Lipoprotein Analysis disease. The protein composition of lipoproteins can
The cardiac risk proRle that is routinely measured in be analysed by chromatography, electrophoresis, or
the clinical laboratory is an important indicator of immunoassay. SDS-PAGE resolves proteins accord-
susceptibility to the development of atherosclerosis. ing to molecular size, while charge separation can be
There is clearly a need to develop a more compre- achieved by zone electrophoresis or isoelectric focus-
hensive cardiovascular risk proRle that includes more ing. QuantiRcation of protein bands after elec-
factors than cholesterol and triglycerides. Triglycer- trophoresis is achieved by immunoblotting or by
ides and cholesterol are transported in blood in asso- staining with Coomassie Blue. However, due to dif-
ciation with lipoproteins, that can be separated by ferences in the chromogenicity of different apolipop-
ultracentrifugation into four main classes according roteins, these staining methods are only semiquan-
to their density: chylomicrons, very low-density titative. Immunoassays offer good sensitivity and pre-
lipoproteins (VLDL), low-density-lipoproteins cision, but results can vary with different antisera.
(LDL), and high-density lipoproteins (HDL). The Furthermore, immunoassays do not distinguish be-
current lipid proRle measured focuses on the lipid tween different apo isoforms. The availability of high
components of lipoproteins. Because proteins asso- performance capillary electrophoresis systems has
ciated with lipoproteins, apoproteins, play a central created the possibility of applying this technology to
role in lipid homeostasis, important research efforts lipoprotein analysis. Protein separation is performed
have been directed to the study of apoproteins. at high Reld strengths in micropore capillaries, with
Free-Sow isotachophoresis (ITP) and capillary ITP direct monitoring by online UV detection. The separ-
(cITP) has been utilized for many years to study ation is analogous to that achieved with slab gel
plasma lipoproteins. The discriminating principle is systems, except that no support media are used, elec-
based on the net charge of lipoproteins. cITP is trophoresis being carried out in a free solution, and
a high-resolution electrophoretic technique, by which results are obtained in minutes rather than hours.
ionic sample components are separated according Separations based on differences in molecular size are
their net electric charge and without molecular sieve performed in a SDS-containing UV transparent poly-
effects. With cITP, lipoprotein analysis can be per- mer network. This mode, usually termed ‘capillary
formed directly from whole serum and offers the SDS gel electrophoresis’ (though it actually uses a non-
potential of a reliable and automated quantitation of gel matrix), has been found to give similar results to
lipoprotein subpopulations. Sample pretreatment is those obtained with SDS slab gel electrophoresis.
negligible and only few nanolitres of sample are ne-
cessary for the analysis. LDL are a heterogeneous
population of particles that varies in size, density,
Determination of Serum Protein
composition, and electric charge. One signiRcant as- Phenotypes and Microheterogeneities
pect of this variability, relevant to atherosclerosis, is The determination of the phenotype of particular
the phenotypic pattern of LDL density and size distri- serum proteins may have clinical relevance.
bution among individuals with and without coronary Hereditary deRciencies of the proteinase inhibitor 1-
artery disease. Using nondenaturing gradient gel elec- antitrypsin, one of the most common inborn
trophoresis, various LDL subgroups have been identi- metabolic errors in Europeans, lead to pulmonary
Red on the basis of size. Individuals with lipoprotein emphysemia in young adults and to liver cirrhosis in
proRles enriched in small dense LDL were found to be children. Many distinct subtypes have been identiRed
predisposed to coronary artery disease. Higher using isoelectric focusing, the most common normal
plasma levels of LDL have been found in patients phenotype being MM, whereas the major deRcient
phenotypes are termed MS, MZ, SS, SZ and ZZ. The mutations. Apolipoprotein E (apoE) is a secreted gly-
last form, ZZ, is associated with low concentration of coprotein with a molecular mass of 35 kDa which is
the protein in plasma and with severe clinical mani- synthesized primarily in the liver and brain. ApoE
festations. Isoelectric focusing in immobilized pH plays a critical role in lipid metabolism through its
gradients represents a major improvement, and the function of redistributing lipids amongst the cells of
method can be utilized for subtyping many different various organs. It is a constituent of VLDL and of
proteins with high reproducibility. a subclass of HDL. The APOE gene is polymorphic
The microheterogeneity of serum glycoproteins in and its three common alleles 2, 3, and 4 code for
patients with chronic alcohol abuse as well as in isoforms E2, E3 and E4, respectively. The iso-forms
patients with carbohydrate-deRcient glycoprotein E3 and E4 differ from E2 by single arginine (R)
syndrome can be globally studied by two-dimensional substitution at one or both cysteine (C) at position
electrophoresis, whereas SDS-PAGE and Western 130 and 176 of the apoE amino acid sequence (acces-
blotting allows detailed comparison and character- sion number P02649; SwissProt database). Com-
ization of speciRc proteins. pared with allele 3, reduced LDL cholestrol (LDL-c)
and apoB levels frequently accompany the 2 allele,
Diagnosis of Haemoglobinopathies higher LDL-c levels occurring with the 4 allele, and
both 2 and 4 alleles are associated with hypertrig-
and Haemoglobin A1c lyceridaemia. In addition to the dyslipidaemic tend-
Disorders of haemoglobin synthesis such as sickle cell ency, the relative odds for prevalent coronary heart
disease, - and -thalassaemias or haemoglobin vari- disease are found to be increased with the 4 allele.
ants } grouped under the term ‘haemoglobinopathies’ A compelling association between risk of Alzheimer’s
} are frequently observed, and up to 45% of the disease in both late-onset familial Alzhemer’s disease
newborns from regions at risk present an abnormal and sporadic Alzheimer’s disease with the 4 allele
haemoglobin. Therefore, it is of importance to have has been also demonstrated. APOE genotyping is
cost-effective methods to screen large populations. usually determined by a blood test using DNA iso-
Isoelectric focusing has been used for many years, and lated from leukocytes, embedded tissues, Guthrie
when performed correctly, produces excellent hae- spots, buccal epithelial cells and restriction isotyping.
moglobin separation, with very little band overlap The method is based on the existence of a GCGC
when bands are measured to 0.1 mm against controls. recognition site for the endonuclease HhaI overlap-
It has been shown that high-power liquid chromato- ping the coding sequences present for amino acids
graphy also allows accurate diagnosis of the haemo- 130 and 170 in the 4 allele, absent in position 130 in
globinopathies, and that both approaches can be used the 3 allele, and absent in both position 130 and 170
for a universal screening. Another interesting ap- in the 2 allele. A 244-bp sequence including the two
proach to study haemoglobin variants is capillary polymorphic sites is ampliRed by using PCR and is
isoelectric focusing. The technique allows high-efR- digested by HhaI (Figure 1). The resulting fragments
ciency separation and precise quantitation of haemo- are separated by electrophoresis on polyacrylamide
globin variants over a wide range of concentrations. gels. After electrophoresis, the gels are either treated
Haemoglobin A1c (Hb A1c) is the analyte of choice with ethidium bromide or are silver stained. The
for monitoring metabolic control in patient with dia- method cannot detect rare APOE variants unless nu-
betes mellitus. Hb A1c can be measured using capil- cleotide substitution alter the HhaI-restriction sites
lary electrophoresis, without apparent interference within the PCR-ampliRed region. In that case the rare
by other haemoglobin variants such as Hb F, Hb S variants are identiRed only by using DNA sequencing.
or Hb C. PCR and restriction isotyping for APOE genotyping
must be preferred to isoelectric focusing (IEF). Be-
cause the post-translational modiRcations of apoE
Genotyping of Proteins such as glycosylation or desialylation alter the electric
Many complex biological questions, encountered in charges of the isoforms, misclassiRcation of geno-
all Relds of medicine, have been solved using DNA- types currently occurs when using IEF.
based technologies. It is now possible to make many
different prenatal diagnoses using genetic testing with Small Molecules (Drugs, Steroids)
one technique or a combination of the three basic
molecular testing techniques } nucleotide sequencing,
Monitoring
restriction fragment length polymorphism mapping Drugs, licit or illicit, play a pivotal role in almost all
and molecular DNA analysis using the polymerase aspects of life. Monitoring of drugs is of major
chain reaction (PCR) } routinely used to assess gene importance for regulatory authorities, in clinical
Figure 1 Application of PCR to the analysis of apolipoprotein E. (A) Map of the PCR-amplified region encoding common APOE
isoforms and location of HhaI cleavage sites. The distance in basepairs between the Hha I sites that distinguish isoforms are shown
for each genotype. (B) Silver-stained PAGE after the electrophoretic separation of Hha l fragments from different patients.
settings, in forensic science, for drug testing at the gen can be studied using mixed micellar electrokinetic
workplace, and in the pharmaceutical industry. With capillary chromatography (MEKC), using borate buf-
the advent of fused-silica capillary instrumentation, fer and SDS. The separation is rapid and quantitation
drug analysis by capillary electrophoresis has made is efRcient.
important progress. Capillary electrophoresis of
drugs, as well as their metabolites in almost all biolo-
gical samples, has been shown to provide high-quality
Cerebrospinal Fluid Analysis
data that can be used for diagnostic drug monitoring. Analysis of cerebrospinal Suid (CSF) proteins using
Corticosteroid hormones are usually analysed by electrophoretic techniques is particularly useful in the
immunological techniques or HPLC. Immunoassays diagnosis and management of neurological diseases
are prone to interference from various compounds, particularly in the detection of immune response
a disadvantage not observed with HPLC. However, within the central nervous system (CNS). In healthy
the latter technique is hampered by the large sample individuals, all CSF immunoglobulins are derived
and eluent volumes needed. Corticosteroid hormones from plasma. However, in most of the inSammatory
such as cortisone, cortisol, progesterone and oestro- conditions within the CNS, activated B-lymphocytes
secrete their immunoglobulins directly into the brain been evaluated in samples from 1007 patients with
extracellular space. This local immune response oc- neurological diseases. It was found that IEF has a sensi-
curs typically in demyelinating disorders such as mul- tivity of 95% for multiple sclerosis against 67% for the
tiple sclerosis, in chronic CNS infections as well as in quantitative tests and a speciRcity with a false-posit-
autoimmune disorders with neurological involve- ive rate of 0% versus 3.5% for the quantitative tests.
ment. An intrathecal synthesis of immunoglobulins The diagnosis and the treatment of the discharge of
can be detected either by quantitative methods and CSF (liquorrhea) from the subarachnoid space into
the results are expressed using various indexes the nasal or aural mucosa (CSF rhinorrhea and otor-
or by qualitative methods such as IEF and Western rhea) is a critical clinical problem. The most common
blotting. Because the locally synthesized immuno- cause of liquorrhea is traumatic fracture: in a large
globulins are the product of a limited number of series of pediatric patients with temporal bone frac-
clones, they currently produce an oligoclonal pattern tures, liquorrhea was noted in 25% of the patients.
superposed to the plasma pattern (Figure 2). IEF is Other causes are neurosurgical procedures, intra- and
now considered as a more sensitive test than the extracranial tumours, as well as primary CSF rhinor-
quantitative index for the detection of an intrathecal rhea through congenital bony dehiscences. The diag-
immune response. The diagnostic sensitivity and spe- nosis of liquorrhea is important in view of developing
ciRcity of both quantitative and qualitative tests has a potentially fatal meaningitis. The use of prophylac-
tic antibiotics does not appear to be beneRcal in
reducing the incidence of post-traumatic meningitis.
Surgical intervention is therefore necessary. Detection
of 2-transferrin is now considered as the most re-
liable test in the identiRcation of CSF rhinorrhea and
otorrhea. 2-transferrin is a transferrin isoform which
is locally synthesized in the central nervous system.
This isoform is characterized by the presence of trun-
cated N-glycans devoid of terminal galactose and
sialic acid. The serum transferrin and 2-transferrin
isoforms found in CSF are separated by native elec-
trophoresis using agarose and detected by immuno-
blotting. Because there is no gold standard for the
diagnosis of liquorrhea, the sensitivity and the speciR-
city of the test are unknown.
Urine Analysis
Electrophoretic methods have been widely used for
the clinical analysis of the proteins contained in urine
(proteinuria). Normally, the renal glomeruli restrict
Rltration of plasma proteins, and only traces are
present in the urine. In several renal diseases and
particularly in patients with glomerulonephritis,
nephrosis, glomerulosclerosis and IgA nephropathy,
large amounts of proteins may be present in the urine.
This is the reason why so many different elec-
trophoretic techniques, including SDS-PAGE, isoelec-
tric focusing, and high-resolution two-dimensional
polyacrylamide gel electrophoresis have been applied
over the years to study in detail the protein composi-
tion of the urine of patients with kidney diseases.
Figure 2 Analysis of the immunoglobulins of the cerebrospinal The Diagnosis of Infectious Diseases
fluid. Typical IgG patterns of sera (S) and of cerebrospinal fluid by Western Blotting
(CSF) separated by isoelectric focusing and revealed by Western
blot. (A) Normal control. (B) Patient with multiple sclerosis. The Despite the analytical power of gel electrophoresis,
arrow heads indicate the oligoclonol bands found only in the CSF. direct speciRc identiRcation of separated components
by ligands such as antibodies, lectin or enzymes was found that the 14-3-3 immunoassay had an overall
generally hindered by the pore size of the gel matrix. sensitivity of 96% (68 true-positive results and
Thus the development of protein blotting techniques 3 false-negative results) and an overall speciRcity of
that enabled the separate components to be transfer- 88% (164 true-negative and 22 false-positive results).
red from gels onto membranes where they were Interestingly, the speciRcity of the immunoassay
bound and available for participation in a range of amongst all patients with dementia was 96% (90
reactions was of enormous practical signiRcance. Ba- true-negative and 4 false-positive results). It was also
sically, with these techniques, viral proteins, either observed that the immunoassay may become positive
from culture Suid or obtained by recombinant mo- early in the course of CJD. However the immuno-
lecular technologies, are separated by SDS-PAGE, assay does not allow a quantitation of the 14-3-3
then blotted from the polyacrylamide gel matrix onto proteins and needs further validation with larger
membranes by an electrophoretic transfer, and Rnally numbers of patients.
identiRed using enzyme immunoassay. The Western
blot method has been used in key epidemiological
studies that reported the unambiguous association of
Identi\cation of Microorganisms
human immunodeRciency virus (HIV) with AIDS, Pulsed-Reld gel electrophoresis (PFGE) is a method
and is still an essential tool for conRrming the pres- widely used to separate fragments of DNA as long as
ence of antibodies to HIV. Western blot has been used several million bases by subjecting the gel to an elec-
as the ‘gold standard’ conRrmatory test for specimens trical current alternately delivered from two angles
found to be reactive in screening assays. Protein blot- in timed intervals, which minimizes diffusion of large
ting can be applied to a large variety of infectious molecules. PFGE has many important clinical ap-
diseases, to study allergens in patients with allergy as plications, particularly in the Reld of infectious dis-
well as to identify autoantigens in patients suffering eases. PFGE allows typing bacteria such as group
from autoimmune disorders. A streptococci, Staphylococcus aureus, Escherichia
The transmissible spongiform encephalopathies or coli or Salmonella enterica. PFGE can be particularly
prion diseases constitute a group of progressive and useful for assisting epidemiological investigations of
fatal neurodegenerative diseases that affect both hu- illnesses caused by a common-source of pathogen
man and animals. The diseases affecting humans in- such as Escherichia coli O157 : H7 in food poisoning
clude Creutzfeldt}Jakob disease (CJD) for which or to demonstrate that recurrent episodes of staphy-
three main forms have been recognized: inherited, lococcal bacteraemia are primarily related to relapses
sporadic, and acquired (iatrogenic). Patients with rather than to new infections. The technique is com-
CJD present a progressive dementia with myo- monly known as DNA Rngerprinting.
clonus usually associated with the presence of sharp-
wave complexes on their electroencephalogram. The
gold standard for a deRnitive diagnosis of CJD re-
quire a histological examination of the brain and
immunostaining for the protease-resistant prion
protein (PrPres). Several CSF proteins such as neurone-
speciRc enolase, S-100b, CK-BB, ubiquitin and
protein 14-3-3 have been proposed as potential
pre-mortem markers for CJD. Amongst them, only
protein 14-3-3 proved useful diagnostically. The dif-
ferential expression of CSF proteins in patients with
CJD has been evaluated by using two-dimensional
electrophoresis, and two protein spots with CJD has
been evaluated by using two-dimensional elec-
trophoresis, and two protein spots (proteins 130 and
131) were identiRed. N-terminal sequencing of pro-
tein 130 matches the sequence of 14-3-3, a brain-
speciRc protein of 28 kDa which is believed to have Figure 3 Application of Western blot to the diagnosis of Creutz-
a regulatory function in monoamine biosynthesis. feldt}Jakob disease. Immunoblot by using anti-14-3-3 after
SDS-PAGE separation of (A) human brain proteins, (B) cerebro-
A Western blot assay has been developed, using
spinal fluid proteins from a patient with Creutzfeldt}Jakob disease
a commercially available antibody directed against and (C) normal control cerebrospinal fluid. Note that anti-14-3-3
14-3-3 (Santa Cruz Biotechnology) which cross- immunoglobulins bound nonspecifically to a cerebrospinal fluid
reacts with protein 130 and 131 (Figure 3). It was protein with a molecular mass close to 55 kDa.
Summary mann C (1999) Microheterogeneity of serum glyco-

proteins in patients with chronic alcohol abuse com-
Electrophoretic techniques will certainly continue to pared to carbohydrate-deRcient glycoprotein syndrome
be applied in the clinical laboratory for many years to type I. Clinical Chemistry 45: 1408}1413.
come. However, only highly reproducible methods Hixson JE and Vernier DT (1990) Restriction isotyping of
such as those based on capillary electrophoresis will human apolipoprotein E by gene ampliRcation and
survive on the condition that they can be applied to cleavage with HhaI. Journal of Lipid Research 31:
resolve heterogeneous clinical problems and that they 545}548.
can be fully automated with electronic handling of Hoffmann A, Nimtz M, Getzlaff R and Conradt HS (1995)
the data. The evolution of electrophoretic techniques ‘Brain-type’ N-glycosylation of asialotransferrin from
achieved over the last few years has allowed a revol- human cerebrospinal Suid. FEBS Letters 359: 164}168.
Hsich G, Kenney K, Gibbs CJ, Lee KH and Harrington MG
ution in medical science. However, in terms of man-
(1996) The 14-3-3 brain protein in cerebrospinal Suid as
ual operations, the good old time of ‘blue Rngers’ has a marker for transmissible spongiform encephalopa-
probably Rnished. thies. New England Journal of Medicine 335: 924}930.
See Colour Plate 68. James RW, Hochstrasser DF, Tissot JD, Funk M, Appel
RD, Barja F, Pellegrini C, Muller AF and Pometta
Further Reading D (1988) Protein heterogeneity of lipoprotein particles
containing apolipoprotein A-I without apolipoprotein
Bienvenu J, Graziani MS, Arpin F, Bernon H, Blessum C, A-II and apolipoprotein A-I with apolipoprotein A-II
Marchetti C, Righetti G, Somenzini M, Verga G and isolated from human plasma. Journal of Lipid Research
Aguzzi F (1998) Multicenter evaluation of the Paragon 29: 1557}1571.
CZETM 2000 capillary zone electrophoresis system for Keir G, Zeman A, Brokes G, Porter M and Thompson EJ
serum protein electrophoresis and monoclonal compo- (1992) Immunoblotting of transferrin in the identiRca-
nent typing. Clinical Chemistry 44: 599}605. tion of cerebrospinal Suid otorrhoea and rhinorrhea.
Bossuyt X, Schiettekatte G, Bogaerts A and Blanckaert Annals of Clinical Biochemistry 29: 210}213.
N (1998) Serum protein electrophoresis by CZE 2000 Marshall T and Williams KM (1998) Clinical analysis of
clinical capillary electrophoresis system. Clinical Chem- human urinary proteins using high resolution elec-
istry 44: 749}759. trophoretic methods. Electrophoresis 19: 1752}1770.
Campbell M, Henthorn JS and Davies SC (1999) Evalu- Mayeux R, Saunders AM, Shea SS, Evans D, Roses AD,
ation of cation-exchange HPLC compared with isoelec- Hyman BT, Crain B, Tang MX and Phelps CH (1998)
tric focusing for neonatal hemoglobinopathy screening. Utility of the apolipoprotein E genotype in the diagnosis
Clinical Chemistry 45: 969}975. of Alzheimer disease. New England Journal of Medicine
Clark R, Katzmann JA, Kyle RA, Fleisher M and Landers JP 338: 506}511.
(1988) Differential diagnosis of gammopathies by capil- McLean BN, Luxton RW and Thompson EJ (1990) A study
lary electrophoresis and immunosubtraction: analysis of of immunoglobulin G in the cerebrospinal Suid of 1007
serum samples problematic by agarose gel electrophor- patients with suspected neurological disease using
esis. Electrophoresis 19: 2479}2484. isoelectric focusing and the log IgG-index. Brain 113:
Conti M, GelR C, Bosisio AB and Righetti PG (1996) 1269}1289.
Quantitation of glycated hemoglobins in human adult Righetti PG and Bossi A (1997) Isoelectric focusing in
blood by capillary isoelectric focusing. Electrophoresis immobilized pH gradients. Journal of Chromatography
17: 1590}1596. B, Biomedical Sciences and Applications 699: 77}89.
Doelman CJ, Siebelder CW, Nijhof WA, Weykamp CW, Schmitz G, MoK ller C and Richter V (1997) Analytical
Janssens J and Penders TJ (1997) Capillary electrophor- capillary isotachophoresis of human serum lipoproteins.
esis system for hemoglobin A1c determinations evalu- Electrophoresis 18: 1807}1813.
ated. Clinical Chemistry 43: 644}648. Shihabi ZK and Frieberg MA (1997) Analysis of small
Fowler VG Jr, Kong LK, Corey GR, Gottlieb GS, McClel- molecules for clinical diagnosis by capillary electrophor-
land RS, Sexton DJ, Gesty-Palmer D and Harrell LJ esis. Electrophoresis 18: 1724}1732.
(1999) Recurrent Staphylococcus aureus bacteremia: Stocks J, Nazeem Nanjee M and Miller NE (1998). Analy-
pulsed-Reld gel electrophoresis Rndings in 29 patients. sis of high density lipoprotein apolipoproteins by capil-
Journal of Infectious Disease 179: 1157}1161. lary zone and capillary SDS gel electrophoresis. Journal
Harrington MG, Merril CR, Asher DM and Gajdusek DC of Lipid Research 39: 218}227.
(1986) Abnormal proteins in the cerebrospinal Suid of Wilson PW, Myers RH, Larson MG, Ordovas JM, Wolf PA
patients with Creutzfeldt}Jakob disease. New England and Schaefer EJ (1994) Apolipoprotein E alleles, dys-
Journal of Medicine 315: 279}283. lipidemia, and coronary heart disease. The Framingham
Hempe JM, Granger JN and Craver (1997) Capillary offspring study. Journal of the American Medical Asso-
isoelectric focusing of hemoglobin variants in the pedia- ciation 272: 1666}1671.
tric clinical laboratory. Electrophoresis 18: 1785}1795. Zerr I, Bodemer M and Weber T (1997) The 14-3-3 brain
Henry H, Froehlich F, Perret R, Tissot JD, Eilers-Messerli protein and transmissible spongiform encephalopathy.
B, Lavanchy D, Dionisi-Vici C, Gonvers JJ and Bach- New England Journal of Medicine 336: 874.
2468 III / CLINICAL APPLICATIONS / Gel Electrophoresis
Gel Electrophoresis
J.-D. Tissot, A. Layer and P. Schneider, Foundation trophoretic techniques applied to sera from healthy
CRS, Lausanne Switzerland adults cannot separate individual clonal products.
F. Forestier, Institut de PueH riculture de Paris, Paris, Immunoglobulin produced by an expanding B cell
France clone to a level permitting detection by electro-
H. Henry, Centre Hospitalier Universitaire Vaudois, phoretic techniques is known as ‘monoclonal gam-
Lausanne, Switzerland
mopathy’ (MG), and has been observed in a wide
Copyright ^ 2000 Academic Press variety of disease. MG can be also observed in hu-
mans without overt disease and is termed ‘benign
MG’, or ‘monoclonal gammopathy of undetermined
Introduction signiRcance’. The frequency of benign MG has been
The separation of polypeptides by electrophoresis al- increasing in parallel with the reRnement of the tech-
lowed Arne Tiselius to describe protein fractions cor- niques. IdentiRcation of monoclonal immunoglobu-
responding to albumin, -, -, and -globulins in lins requires sensitive and rapid screening methods.
serum with the Rrst published diagram of human Electrophoresis on cellulose acetate membrane is sat-
serum protein electrophoresis in 1939. The number isfactory for initial screening.
of fractions slowly expanded into electrophoretic High-resolution agarose gel electrophoresis (HRE)
subfractions identiRed as 1, 2, 1, 2, 1 and 2. is more sensitive for the detection of rare monoclonal
The mobility characteristics of these fractions are proteins (Figure 1). ConRrmation of the presence of
still used to denote serum proteins such as 1- an MG should be performed using methods such as
macroglobulin, 2-antiplasmin or 2-microglobulin. immunoelectrophoresis (IEP) or immunoRxation
Sophisticated new electrophoretic techniques for electrophoresis (IFE). IEP was described in 1953, and,
identifying many proteins simultaneously and relat- until recently, was considered as the method of refer-
ing them to diseases have been developed over ence. As a routine procedure, it is time-consuming,
the years, many of them being described in other and requires a high level of technical expertise. The
studies. Almost all body Suids have been studied consequence of the long diffusion process is that
by electrophoresis, serum, urine and cerebrospinal small amounts of protein become too dilute to be
Suid being evaluated in great detail by different tech- detected, resulting in low sensitivity. The most impor-
niques. However, despite the major developments tant drawback of IEP is the so-called ‘umbrella ef-
and progress achieved in protein separation, only fect’. Furthermore, the demonstration of and
a restricted number of methods are routinely used in monoclonal light chains belonging to IgA or IgM
the clinical laboratory. Nowadays, serum protein isotypes is often masked by the presence of polyclonal
electrophoresis is mainly used to study major serum IgG, that, because of faster diffusion, migrates ahead
protein alterations such as those observed in patients to an area situated between IgA and IgM. IFE, like
with inSammatory, liver or kidney diseases, as well as IEP, includes two steps, but without the diffusion
in patients presenting lymphoproliferative disorders process. In the Rrst step, protein fractions are separ-
and alterations of immunoglobulin (Ig) production. ated by agarose gel electrophoresis; in the second
Electrophoretic techniques, in association with ultra- step, they are revealed by overlaying of antiserum
centrifugation, are also used to study lipoproteins and speciRc to the different immunoglobulin heavy and
to classify hyperlipidaemic disorders. light chains. The technique is considered as the pro-
Here, we provide selected clinical situations cedure of choice in the routine diagnosis of MG,
studied by electrophoretic methods, with a particular because of its simplicity, speed, and sensitivity. Other
emphasis on 2D-PAGE, in order to illustrate how advantages of the method are ease of interpretation
electrophoresis can be used to gain insight into par- and the absence of diffusion. Furthermore, the en-
ticular clinical problems. hanced sensitivity of IFE also yields more frequent
detection of low concentration MG and, what is more
important, of multiple immunoglobulin bands or
Monoclonal Immunoglobulins subtle bands of restricted heterogeneity. In such
The large diversity of antibodies in an individual cases, further assessment of the clonality of immuno-
results from highly complex mechanisms inSuencing globulins depends on additional techniques such as
B-cell development, expansion, and immunoglobulin immunoblotting, immuno-isoelectric focusing or
(Ig) secretion. This diversity is so complex that elec- high-resolution two-dimensional polyacrylamide gel
III / CLINICAL APPLICATIONS / Gel Electrophoresis 2469
electrophoresis (2D-PAGE). With 2D-PAGE, , ,

and chains appear in nonoverlapping regions of
protein maps (Figure 2). Monoclonal heavy chains
are easily differentiated from polyclonal heavy
chains, according to their different two-dimensional
electrophoretic patterns. Polyclonal heavy chains, ac-
cording to their different two-dimensional electro-
phoretic patterns. Polyclonal heavy chains are highly
heterogeneous and are resolved as ‘unspotted’ diffuse
zones whereas monoclonal heavy chains show charge
and, to a lesser degree, size microheterogeneity.
Polyclonal light chains appear as clustered nondis-
crete spots following 2D-PAGE, and form cloudy
zones with unevenly distributed densities. In about
two-thirds of patients with MG, monoclonal light
chains are detected as a dominant and well-deRned
spot. In the remaining one-third of the patients,
examined monoclonal light chains disseminate in
more than one spot. In patients with monoclonal
heavy chain disease, only fragments of monoclonal
heavy chains are synthesized and released into the
circulation by malignant lymphoplasmocytic cells.
Such fragments are identiRed with immunoRxation
electrophoresis as proteins bands that only bind spe-
ciRc anti-heavy chain antibodies, but not speciRc anti-
or anti- light chain antibodies (Figure 3).
However, the abnormal characteristics of such
monoclonal heavy chains are best studied using 2D-
PAGE of protein-G puriRed fractions, as demon-
strated in Figure 4. A set of spots () is observed at
the basic side of the gel, in an area corresponding to
pI from 5.8 to 6.8, and Mr of 32 to 38 kDa. Spots
corresponding to polyclonal IgG chains and poly-
clonal immunoglobulin light and chains are
also identiRed. No spot corresponding to a mon-
oclonal immunoglobulin light chain is observed.
The 32 to 38 kDa Mr of the monoclonal chain
indicates that it is a fragment of a normal chain, and
the microheterogeneities of the pI suggest that it is
glycosylated. Finally, the binding of the molecule to
protein G implies that the Fc region of the immunog-
lobulin is intact. In contrast to what was seen in the
serum of another patient presenting the same disease,
abnormal monoclonal chains of higher Mr are not
observed.
Cold Agglutinins and Cryoglobulins

Figure 1 High resolution serum protein agarose electrophor-
esis profiles. Serum protein electrophoresis, and protein den- Cold agglutinins (CAs) are immunoglobulins, most
sitometry of serum samples from patients presenting monoclonal frequently IgM, that agglutinate erythrocytes at tem-
IgG-, IgM- and IgA- gammopathies were performed by using peratures below 373C. Immunohaemolytic anaemia
commercially available kits. The isotype of the monoclonal heavy
related to CAs can be observed in a wide variety of
chains as well as the type of the monoclonal light chains were
identified using immunofixation electrophoresis. The protein diseases (Figure 5).
peaks corresponding to the monoclonal components are shown Monoclonal CAs, generally belonging to the IgM
by arrowheads. isotype, are observed in patients with lymphopro-
Figure 2 High-resolution two-dimensional polyacrylamide gel electrophoresis in a case of monoclonal IgD-. (A) Serum and (B)
purified IgD (anti- Sepharose). a, albumin; , polyclonal heavy chains of IgM; , polyclonal heavy chains of IgA; , polyclonal heavy
chains of IgG; ,, polyclonal immunoglobulin light chains; , monoclonal -chain monoclonal; , monoclonal chain; t, transferrin. The
gels were silver stained, and presented with the higher molecular weights at the top, and the acidic side on the left. (Reproduced from
Tissot JD, Schneider P, Hohlfeld P et al. (1993) Two-dimensional electrophoresis as an aid in the analysis of the clonality of
immunoglobulins. Electrophoresis 14: 1366, with permission from Wiley-VCH, Weinheim.)
liferative disorders or suffering from the ‘idiopathic’ concentrations in serum, can be easily performed
chronic cold agglutinin disease, whereas polyclonal using 2D-PAGE. CAs are isolated from serum by cold
CAs, also belonging to the IgM isotype, may be found absorption of immunoglobulins on red bloods cells.
in patients with various infectious diseases, but more After several cold washes, red blood cells coated with
particularly in patients with Epstein}Barr virus or CAs are rewarmed to 373C. After centrifugation, the
Mycoplasma pneumoniae infection. The evaluation supernatant is collected and studied with 2D-PAGE.
of the clonality of CAs, that are frequently at very low As mentioned in the previous section, polyclonal IgM
are quite easily differentiated from monoclonal IgM
according to their different electrophoretic patterns
(Figure 6).
Cryoproteins are deRned as proteins precipitating
at low temperature. Most frequently, the precipitate
contains immunoglobulins, and are therefore called
cryoglobulins. Three types of cryoglobulins have been
described: type I contains a single monoclonal
immunoglobulin, whereas type II is a mixture of
a monoclonal immunoglobulin with polyclonal
immunoglobulins, and type III is a mixture of poly-
clonal immunoglobulins of different isotypes, most
frequently IgG and IgM (Figure 7). Type II and type
III are also called mixed cryoglobulins. A new type
Figure 3 Immunofixation electrophoresis in a case of -chain II}III class of cryoglobulins, containing polyclonal
disease. A serum sample from a 56-year-old patient was ana- IgG associated with a mixture of polyclonal and mon-
lysed by immunofixation electrophoresis. A protein band (arrow-
oclonal IgM has been recently described through 2D-
heads) was shown to react with anti-human immunoglobulin chain
antiserum, but not with anti-human or light-chain antibodies. PAGE. In a series of 265 cryoglobulins studied with
SPE, serum, protein electrophoresis; G, A, M, , , immunodetec- 2D-PAGE, we identiRed type I, II, II}III, and III in
tion of protein bands with corresponding antisera. 3.4%, 26, 22.6 and 43.8% of the cases, respectively.
ledge of normal fetal physiology, including some as-

pects of protein biology. However, the amount of
pure fetal blood which can be derived for research
purposes from samples obtained for prenatal diag-
nosis is obviously limited. Thus, electrophoretic tech-
niques that allow studies of plasma and serum pro-
teins contained in minute samples (usually less than
1 L) must be applied. Many proteins detected in the
plasma and serum of adults are already present in the
blood of normal embryos at a very early stage of
gestation. Not all proteins are produced by the con-
ceptus but many are the result of transfer from the
mother across the placenta. Some proteins remain at
a fraction of their adult levels throughout intrauterine
life, others are progressively produced in increasing
amounts when term approaches, and Rnally, others
are present in higher concentrations during fetal life
than after delivery. ModiRcations of plasma and se-
rum protein concentrations during early extrauterine
life have been well documented. The plasma concen-
tration of a given protein is governed not only by fetal
synthesis and degradation, but also by the result of
the exchange between mother and fetus through the
Figure 4 High resolution two-dimensional polyacrylamide gel
electrophoresis in a case of -chain disease. Protein G purified
placenta. Materno-fetal transfer of proteins involves
protein fraction from a 56-year-old patients presenting -chain several different mechanisms such as a Rrst-order
disease was analysed using 2D-PAGE. , polyclonal heavy process or active transport. The concentration of each
chains of IgG; , spots corresponding to monoclonal chain fetal plasma protein results of a balance between
fragments; -, polyclonal immunoglobulin light chains. The gel opposing dynamic metabolic and physiological pro-
was silver stained, and presented with the higher molecular
weights at the top, and the acidic side on the left.
cesses which proceed simultaneously. The relative
impact of these factors contributing to a plasma pro-
tein concentration continuously shifts during devel-
They were composed of oligoclonal immunog- opment, and not always in the same manner. 2D-
lobulins mixed with traces of polyclonal IgG in 4.2% PAGE is an ideal tool to study such a dynamic pro-
of the cryoglobulins studied. These cryoproteins were cess. Fetal protein maps reveal spots which are always
tentatively named type II}IIIvariant cryoglobulins. absent in those of adults. A readily evident set of
spots, located at the acid side of albumin is found in
such fetal samples and corresponds to -fetoprotein
Fetal Proteins (Figure 8). This protein is less apparent as gestational
Gene expression and regulation are dependent on age increases. A second ‘fetal’ polypeptide, character-
highly sophisticated mechanisms. The genome- ized by an apparent Mr of 46 kD and a pI of 5.0
related modiRcations of protein synthesis and expres- is observed as a small spot, located under those of
sion are most evident during fetal development. 1-antitrypsin. N-terminal microsequencing (25-
The development of a safe and easy technique for EDPQ) and immunoblotting using speciRc anti-1-
fetal blood sampling has greatly improved our know- antitryspin antibodies reveal that this spot most likely
Figure 5 Classification of cold agglutinins.

Figure 6 Characterization of cold agglutinins using high resolution two-dimensional polyacrylamide gel electrophoresis. Details of
electrophoretograms of plasma/serum samples (A}C) and their purified cold agglutinins fractions (A}C). Purified polyclonal (D) and
monoclonal (E) IgM heavy chains are shown as controls. A}A, serum and purified CAs from a patient with chronic virus C hepatitis;
B}B, plasma and purified CAs from a patient with Mycoplasma pneumoniae infection; C}C, plasma and purified CA from a patient with
chronic cold agglutinin disease. a, albumin; R, reference protein used to highlight the relatively acidic pI of the monoclonal -chain found
in patient C; mu, polyclonal chain; MU, monoclonal chains. The charge microheterogeneity of monoclonal chains is highlighted by
arrowheads. The gels were silver stained, and presented with the higher molecular weights at the top, and the acidic side on the left.
(Reproduced from Tissot JD and Spertini F (1995) Analysis of immunoglobulin by two-dimensional gel electrophoresis. Journal of
Chromatography A 698: 225}250, with permission from Elsevier Science.)
corresponds to a fetal-speciRc form of 1-antitryspin,

already identiRed in mouse plasma. This polypeptide
is observed in all fetal samples and in all samples
obtained from infants of under two years of age but is
either undetectable or appears as a shaded spot in
adults.
Genetic polymorphism of some serum proteins
such as Gc-globulin, Apo A-IV, or Apo E (Figure 9)
can be also studied using 2D-PAGE. As shown in
Table 1, the technique allows determination of the
frequency of the Apo-E phenotypes, as well as the
calculation of the gene frequencies.
During fetal development, 2D-PAGE reveals a pro-
gressively more and more complex protein pattern
and an increase in the size of many spots. Three major
protein pattern modiRcations are observed on 2D gels
of fetuses at different gestational ages: (a) the pro-
gressive appearance of the protease inhibitor 1-an-
tichymotrypsin; (b) the progressive increase of poly-
clonal IgG, which is particularly evident during the
last weeks of pregnancy; (c) the gradual diminution of
Figure 7 High-resolution two-dimensional polyacrylamide gel -fetoprotein which is undetectable on protein maps
electrophoresis of mixed type III cryoglobulins. The cryoprecipi- of term newborns. On the other hand, before the 38th
tate is characterized by the presence of a mixture of polyclonal
IgM and polyclonal IgG. , polyclonal IgM chains; , polyclonal
week of gestation, polyclonal IgA or IgM heavy
IgG chains; -, immunoglobulin light chains. The gel was silver chains, as well as the two main proteins involved in
stained, and presented with the higher molecular weights at the free haemoglobin transport and catabolism * hap-
top, and the acidic side on the left. toglobin and hemopexin * are not detectable.
Figure 9 Genetic polymorphism of apolipoprotein E in fetuses.

Phenotype E3/4 (A, fetus at the age of 22 weeks, A fetus at the
age of 30 weeks), phenotype E2/3 (B, fetus at the age of 22
weeks, B fetus at the age of 30 weeks), phenotype E3/3 (C, fetus
Figure 8 Fetal proteins. Details of electrophoretograms of at the age of 22 weeks). M, 1-microglobulin; C, chain of
plasma samples obtained from fetuses at 22 (A), 28 (B), and complement factor C4; A-I, apolipoprotein A-I, A-I, pre-apolipo-
protein A-I; and , immunoglobulin light chains. The gel were
32 weeks (C) of gestation. A, albumin; , immunoglobulin
gamma chains; 1, dimers of fibrin ( chains), only observed when silver stained, and presented with the higher weights at the top,
small clots were present in the sample; 2, -fetoprotein; 3, fetal and the acidic side on the left.
form of 1-antitrypsin. The gel was silver stained, and presented
with the higher molecular weights at the top, and the acidic side on phomannose isomerase. Both these defects alter the
the left.
metabolism of mannose in cells and result in a de-
crease of GDP-mannose for N-glycan synthesis. The
diagnosis is usually made by isoelectric focusing of
Carbohydrate-de\cient Glycoprotein serum transferrin showing a different pattern of cath-
Syndrome (CGDS) odal shift due to the loss of terminal sialic acids.
Another way of diagnosis is to separate the transferrin
Carbohydrate-deRcient glycoprotein syndrome isoforms lacking their N-glycans by using SDS-PAGE.
(CDGS) is a group of autosomal recessive disorders Other more complex diagnostic methods have been
affecting multiple organ systems. All of the affected
patients present moderate or severe brain disorders.
Table 1 Distribution of ApoE in fetuses and adults determined
In the Rrst years of life, CDGS is characterized by by 2D-PAGE
hypotonia with failure to thrive, dysmorphism and
coagulation abnormalities. These clinical manifesta- Phenotypes Fetuses Adults Total a Expected b
tions are the direct embryologic and physiological
E2/2 0 2 2 1
consequences of an underglycosylation of protein.
E2/3 10 21 31 31
The alterations are related to a reduced number of E2/4 1 1 2 4
entire N-glycans on glycoproteins and they have been E3/3 76 152 228 225
described for serum transferrin, 1-antitrypsin, 1- E3/4 18 32 50 57
acid glycoprotein, 1-antichymotrypsin, fetuin, ceru- E4/4 2 6 8 4
Total 107 214 321 322
loplasmin and antithrombin III, C3a and C4a. In
patients with CDGS type 1a and type 1b, the post- a
Allele frequencies: Apo EH2: 0.058, Apo EH3: 0.836, Apo EH4:
translational defect is due either to a deRciency of 0.106.
phosphomannomutase or a deRciency of phos- b
Hardy-Weinberg equilibrium.
Figure 10 Comparison of the specificity of antibodies raised against transferrin. Transferrin isoforms identified by antibodies directed
against multiple epitopes (polyclonal Ab) and transferrin isoforms that presented the N-432 transferrin epitope recognized by a targeted
antibody (targeted Ab). Lane 1, normal serum control; lane 2, serum from a patient with alcoholism; lane 3, serum from a patient with
type 1 carbohydrate-deficient glycoprotein syndrome (CDGS).
reported such as 2D-PAGE and matrix-assisted laser 2D-PAGE is increasingly used as an important tool
desorption ionization}time of Sight (MALDI}TOF) for biological research. However, despite the fact that
mass spectrometry of serum transferrin. these techniques are particularly useful for studying
We have described the characterization of a new a clinical problem, routine clinical analysis is fre-
antibody which is speciRcally directed against the quently incompatible with the use of such sophisti-
N-glycan binding site localized on the asparagine-432 cated techniques. As illustrated by several examples
of transferrin and which allows immunodetection of presented in this article, 2D-PAGE can be chosen as
transferrin devoid of N-glycan (Figure 10). This anti- a complementary technique to gain insight into sev-
body has the potential to provide a new immuno- eral speciRc clinical problems. The combination of
chemical tool for the CDGS diagnostic. different electrophoretic techniques can be very use-
ful to resolve complex problems.
Conclusion
Further Reading
Numerous diseases can be diagnosed using elec-
trophoretic analyses of body Suid proteins. However, Henry H, Tissot JD, Messerli B et al. (1997) Micro-
heterogeneity of serum glycoproteins and their liver
many different electrophoretic techniques are avail-
precursors in patients with carbohydrate-deRcient
able. Electrophoresis on cellulose acetate is employed glycoprotein syndrome type I: apparent deRciencies
to study many different enzymes such as serum in apolipoprotein J and serum amyloid P. Journal of
amylase, pancreatic isolipase, and pyruvate kinase. Laboratory and Clinical Medicine 129: 412}421.
Agarose gel electrophoresis is extensively used to gain Keren DF (1999) Procedures for the evaluation of mon-
insight into the understanding of lipoprotein biology. oclonal immunoglobulins. Archives of Pathology and
Isotachophoresis or ‘displacement’ electrophoresis Laboratory Medicine 123: 126}132.
allows the simultaneous concentration and effective Krasnewich D and Gahl WA (1997) Carbohydrate-deRcient
separation of different charged substances, including glycoprotein syndrome. Advances in Pediatrics 44:
biological macromolecules. AfRnity electrophoresis is 109}140.
used to evaluate cross-reactivity of antihapten anti- Levinson SS and Keren DF (1994) Free light chains
of immunoglobulins: clinical laboratory analysis. Clini-
bodies whereas lectin afRnity electrophoresis is best
cal Chemistry 40: 1869}1878.
used to characterize serum glycoproteins. Capillary Marshall T and Williams KM (1998) Clinical analysis of
electrophoresis offers the possibility of rapid and au- human urinary proteins using high resolution elec-
tomated analysis of different kinds of molecules, with trophoretic methods. Electrophoresis 19: 1752}1770.
high reproducibility and improved quantiRcation. Nathoo SA and Finlay TA (1987) Fetal-speciRc forms of
Isoelectric focusing as well as SDS-PAGE are corner- alpha 1-protease inhibitors in mouse plasma. Pediatric
stone techniques used to characterize polypeptides. Research 22: 1}5.
III / CLINICAL CHEMISTRY: THIN-LAYER (PLANAR) CHROMATOGRAPHY 2475
Tissot JD, Hohlfeld P, Forestier F et al. (1993) Tissot JD, Invernizzi F, Schifferli J, Spertini F and Schneider
Plasma/serum protein patterns in human fetuses and P (1999) Two-dimensional electrophoretic analysis of
infants: a study by high-resolution two-dimensional cryproteins: a report of 335 samples. Electrophoresis 20:
polyacrylamide gel electrophoresis. Applied and Theor- 606}613.
etical Electrophoresis 3: 183}190. Young DS and Tracy RP (1995) Clinical applications of
Tissot JD and Spertini F (1995) Analysis of immuno- two-dimensional electrophoresis. Journal of Chromato-
globulins by two-dimensional gel electrophoresis. Jour- graphy A 698: 163}179.
nal of Chromatography A 698: 225}250.
CLINICAL CHEMISTRY: THIN-LAYER

J. BZa9 dek, Military University of Technology, General Principles of Clinical

Warsaw, Poland Medicine Analyses by TLC
A. Zdrojewski, Institute of Chemistry, Warsaw,
Poland TLC is a most economical and cost-effective tech-
nique and ideally suited for performing clinical ana-
lyses of biological samples, where the sample load is
high. The method is characterized by high selectivity,
Introduction enabling separation of the analyte from interfering
Clinical medicine is mainly based on the results of substances. At least two basic groups of applications
chemical analysis and these results are very impor- of TLC in clinical medicine analysis can be distin-
tant. It is estimated that more than 50% of diagnoses guished. The most important of them are where TLC
are based on laboratory data. The analyses are usu- has been applied for analytical and screening pur-
ally done using immunoassays, chromatographic and poses where qualitative or semi-quantitative results
spectroscopic methods. The combined techniques, are required. The second group concerns work, where
e.g. liquid chromatography}mass spectrometry TLC performs the function of a clean-up technique
(LC/MS) gas chromatography}tandem mass spectro- for initial sample preparation.
metry (GC-MS-MS) are also applied. The develop- Clinical analysis presents a difRcult analytical
ment of sophisticated methods of analyses reduced problem. The proteins, lipids, or carbohydrates, usu-
interest in simple and rapid methods such as thin- ally present in biological samples, may interfere with
layer chromatography (TLC). Starting from paper analytes. Analyses are also made difRcult by the fact
chromatography and improved over the years, TLC that many substances of interest are only slightly
has become very useful also for clinical medicine. soluble in organic solvents.
TLC has been used to solve analytical, and sometimes
Sample Preparation
preparative, problems.
A large number of publications on the use of There is a generally accepted view in laboratory prac-
TLC for clinical medicine analysis can be found in tice, that the TLC stage of sample preparation is of
the literature. An excellent compilation concern- little importance because it is merely a clean-up tech-
ing different applications of analytical methods nique. This is incorrect, especially in the case of clini-
(including TLC) for isolation and identiRcation of cal medicine analysis where analytes are to be found
drugs in pharmaceutical, body Suids and post- in complex matrices such as plasma, serum or urine,
mortem materials is the book edited by Moffat. whole blood, faeces, saliva, cerebrospinal Suid, gas-
This work, prepared in the Department of Pharm- tric Suid or body tissues. These matrices are very com-
aceutical Sciences of the Pharmaceutical Society plex multicomponent mixtures and therefore sample
of Great Britain, is a comprehensive and clear preparation prior to analysis cannot be omitted.
account of clinical analytical chemistry up to 1986. The process of separation from biological samples
An excellent supplement, covering the early 1990s, and puriRcation of analytes is usually realized by
is the monograph of Jain. More recent works protein precipitation, dialysis, hydrolysis, ultraRltra-
more commonly refer to applications of instrumental tion, dilution, liquid}liquid or solid-phase extraction.
TLC. Originally, the most common extraction method was
2476 III / CLINICAL CHEMISTRY: THIN-LAYER (PLANAR) CHROMATOGRAPHY
liquid}liquid extraction but in recent years solid- tion reagents are listed in Table 2. IdentiRcation can
phase extraction has become an increasingly popular also be made using spectroscopic techniques such as
method for isolation of compounds from biological TLC/FABMS or FABMS-MS (fast atom bombardment
matrices. Lyophilization, saponiRcation, microwave mass spectrometry or tandem mass spectrometry).
processing and supercritical Suid extraction are used Densitometric measurement of the in situ Suores-
less commonly. cence or absorbance provides information about the
quantity of an analyte in a chromatographic band.
Chromatogram Development Techniques Such measurements are suitable for the quantiRcation
of drugs such as chlorpromazine, tricyclic anti-
Chromatographic separation of clinical samples is
depressant and anticonvulsant in blood serum and
carried out mainly in normal phase systems. High
plasma. Quantitative analyses are also performed by
performance silica gel, aluminium oxide, polyamide
combination of TLC with other analytical tech-
and amino-bonded silica gel are commonly used as
niques, e.g. UV-densitometry.
stationary phases for the separation of basic and
acidic drugs and other polar substances such as amino
acids. It is better to separate neutral drugs and other Applications
nonpolar compounds in reversed-phase (RP) systems. In a view of the complexity of clinical analysis, it is
The commonest stationary phases for RP systems are not possible to cover all applications extensively.
silica gel with chemically bonded octyl and octadecyl Therefore, only the most important applications of
groups. Numerous solvent systems consisting of mix- TLC in clinical medicine analyses are presented. The
tures of organic solvents such as hexane, methanol, criteria of aims of analysis are: (1) the diagnosis and
chloroform, ethyl acetate or mixtures of polar sol- treatment of hard intoxication, (2) screening re-
vents with water, organic acids or ammonia have search, (3) monitoring of therapeutic drugs concen-
been used. It is sometimes advantageous to introduce tration, and (4) evaluation of the efRciency of new
modifying agents to the mobile phase to improve methods of therapy. In such applications, the drugs
selectivity. and their metabolites, carbohydrates, amino acids,
One-dimensional ascending or horizontal tech- bile acids, lipids, porphyrins and other compounds of
niques have usually been applied for the development clinical interest are analysed. The results of such work
of chromatograms in a closed chamber. For clinical are collected in libraries (databases) that signiRcantly
medicine needs, there are standard TLC kits (Toxi- facilitate diagnosis and therapy.
Prep type and Toxi-Lab). These are commercially
available and have been used above all for extraction, The Diagnosis and Hard Intoxication Treatment
concentration and analysis of the acidic, basic, and
The term hard intoxication is applied to the deter-
neutral drugs. Recently, multiple gradient develop-
mination of harmful effects, arising in a short time
ment (AMD), two-dimensional techniques and over-
after introducing a large dose of poison to the body.
pressure layer chromatography have been more often
These effects are mainly observed after drugs or alco-
applied to the analyses listed in Table 1.
hol overdose or the introduction of organic solvents,
pesticides, etc., into body. The symptoms of intoxica-
tion can be similar to those caused by disease so that
Visualization and Quanti\cation diagnosis based only on clinic symptoms in insufR-
IdentiRcation of analytes is achieved by co- cient; in such situations analytical data are very im-
chromatography of reference standards with the un- portant. Only laboratory results, can determine the
known and comparison of RF values. TLC retention type and concentration of poison in a biological ma-
indices are less reliable than for other chromato- terial or a functional disorder caused by overdose
graphic methods so that additional information from intoxication or poisoning. The biological samples,
individual spots is needed. Post-chromatographic de- where poisons and/or their metabolites are found are
rivatization is the basic method of visualization espe- mainly blood or urine. The most important features
cially when there exists a speciRc reaction that can of such analyses are the necessity to determine the
conRrm the presence of particular analytes in the spot type of poison as quickly as possible. The basic re-
or band. Physical methods such as Suorescence under quirement (especially in analysis followed by diag-
ultraviolet light are also often applied. The possibility nosis) is for information about the poison, from
of conRrmation of identity based on at least two simple, rapid and reliable qualitative or semiquan-
criteria is the essential advantage of TLC. The titative methods. The common use of TLC in clinical
colorimetric reactions of drugs, narcotics, and other medicine analyses of hard intoxication is due to its
substances of clinical interest with different visualiza- ability to provide such data.
Table 1 The applications of thin-layer chromatography (TLC) in clinical medicine research. The representative examples of
separations
No. Matrix and analyte Chromatographic system Type of development
Analyte Body fluid Stationary phase Mobile phase
1 Theophyline Serum Aminopropylsiloxane Ethanol}5% aqueous diethylamine One-dimensional, isocratic

modified silica solution (95 : 5)
2 Tetrahydro- Urine HPTLC silica gel SPE elution Toxi-Prep system: incorporate
cannabinol solution: ethyl acetate}concentrated SPE with direct spotting onto
(THC) ammonium hydroxide (98 : 2) TLC plate, one-dimentional,
metabolites Developing solution: heptane}ethyl isocratic elution
acetate}glacial acetic acid (7 : 3 : 10)
3 Bile acid Bile Silica gel Chloroform}methanol}acetic acid One-dimensional, isocratic

(85 : 20 : 9)
4 Porphyrins Urine, RP-HPTLC (C-18 Acetonitrile}ion-pair reagent One-dimensional, isocratic

faeces bonded wettable (IPR)}acetate buffer pH4,1 (9 : 5 : 5)
phase) IPR " N-cetyl-N,N,N-
trimethylammonium bromide
5 Amino acids Urine Cellulose (I) Pyridine}acetone}aqueous Two-dimensional, isocratic

ammonium hydroxide}water
(13 : 8.5 : 2.5 : 6)
(II) Isopropanol}formic acid}water
(25 : 3 : 2)
6 Amphetamine Urine Aminopropylsiloxane (I) Ethanol}triethylamine}hexane Two-dimensional, isocratic

derivatives modified silica (15 : 9 : 76)
(II) Acetone}triethylamine}hexane
(23 : 9 : 68)
7 Lipids Sebaceous Silica gel with Hexane}diethyl ether}acetic acid One-dimensional, isocratic
gland concentrating (80 : 20 : 1)
zone
8 Prostaglandins Serum HPTLC silica gel Ethyl acetate}diethyl OPLC one-dimensional

ether}benzene}dioxane}hexane
(45 : 12 : 5 : 8 : 30)
9 Amino acid Urine Silica gel Chloroform}methanol}ethyl AMD gradient elution

derivatives acetate}acetic acid
10 Barbiturate Serum, HPTLC silica gel (I) Ethyl acetate}ethanol}hexane OPLC two-dimensional
derivatives urine (2.5 : 2.5 : 95)
(II) Pyridine}hexane (2 : 8)
11 Tetrahydro- Urine HPTLC (I) Dichloromethane}hexane}methanol One-dimensional, isocratic

cannabinol I-silica gel (7 : 2 : 1)
(THC) WRFs (II) Dichloromethane}hexane}methanol
II-silica gel F (7 : 1 : 2)
12 Ecdysteroids Plasma RP-TLC (C-2,C-8, Methanol}water (9 : 1) One-dimensional, isocratic

C-12, C-18,
aminopropyl,
cyanopropyl
bonded phases)
The majority of this work is aimed at the deter- TLC, identiRcation is based mainly on comparison of
mination of analytical parameters of xenobiotics RF values and on visualization with speciRc colour
and/or their metabolites. In diagnostic tests using reactions. Tests for the identiRcation of alkaloids,
Table 2 Examples of visualization and quantification
Analyte Visualization method Quantification
Yes No
Amino acids
Common protein amino acids 3-Phenyl-2-thiohydantoin (PTH) derivative, 270 nm absorption X
Urinary amino acids (incl. proline Ninhydrin, isatin, isatin-p-dimethylaminobenzaldehyde X
and hydroxyproline)
Bile acids Manganese dichloride in a mixture of water, methanol and sulfuric acid X
and fluorescent measurement
Cholesterol and its esters Aniline blue, bromophenol blue, helasol green and alkaline blue X
Drugs
Antibiotics
Doxycycline, oxytetracycline and Fluorescent measurement, 365 nm /'440 nm X
tetracycline
Aminoglycoside streptomycin and Dansylation and fluorescent measurement X
neomycin
Baclofen Dansylation and fluorescent measurement, X
Barbiturates Mercuric chloride-diphenylcarbazone reagent; fluorescein solution; X
mercurous nitrate spray
Benzodiazepines UV radiation 254/366 nm, dipped in concentrated sulfuric acid and X
observed under UV (366 nm); diazotization and coupling with 1-naphthol X
or Bratton}Marshall reagent and next fluorescent measurement
-Blocking drugs
Atenolol, celiprolol, metoprolol, Brillant Blue B, UV light absorption (254 nm) X
propafenone, propranolol and talinolol
Halofantrine UV light absorption (256 nm) X
Nonsteroidal anti-inflammatory drugs UV light absorption (254 nm) X
(e.g. ketoprofen, piroxicam, diclofenac,
ibuprofen, paracetamol)
Narcotics
Amphetamine metabolites UV light absorption (278, 283 nm) and by fluorescence after X
derivatization with fluorescamine 365 nm/'440 nm
Heroin (diacetylmorphine) HgCl2}K3Fe(CN)6 and light absorption (580 nm) X
9-Tetrahydrocannabinol (THC) UV light absorption (210 nm) X
Tricyclic antidepressive drugs Amelinka’s reagent or tested by UV irradiation X
Fluoxetine, imipramine, doxepine,
opipramol
Phospholipids Iodine vapour X
Free porphyrins Fluorescent measurement X
Prostaglandins -Bromo-2-acetonaphthone (BAN) and fluorescent measurement X
barbituric acid derivatives, benzodiazepine deriva- RF values by a graphical standardization procedure)

tives and other drugs from other pharmacological in a possibly large group of substances. In case of
groups have been known for many years. clinical analyses there are usually hundreds of drugs
Mathematical methods have recently been applied and their metabolites. A mathematical function (e.g.
in the diagnosis of hard intoxication. The Rrst step of ‘discriminating power’ or ‘principal component anal-
these methods is to evaluate the RF values in two, ysis’) for calculation of analytical data is then chosen.
three and sometimes four chromatographic systems. The data set obtained allows for relatively rapid iden-
These systems must be carefully selected to ensure an tiRcation of investigated (unknown) xenobiotics.
appropriate distribution of RF values across the A full description of the application of such methods
plates, the reproducibility of the measurement of can be found in the work by de Zeeuw et al.
these parameters and the correlation of chromato-
TLC as a Screening Method
graphic properties between systems. Next, the sets of
correlated RF values are determined (experimentally Screening methods reduce the cost and time of ana-
determined RF values are converted into correlated lysis of numerous groups of samples. In clinical
medicine analysis, this term is referred to a set of seven specimens. The method is based on TLC and
biological samples (laboratory screening methods) or involves Rve major steps: solid phase extraction, con-
to a set of individuals endangered by contact with centration, spotting, development of chromatograms
harmful substances (metabolic screening). In both and detection.
cases, the aim of analysis is the exclusion of a particu- The kits have been found to be particularly useful
lar sample (for an individual) from the set of samples for analysis of basic and neutral drugs. The compari-
(individuals), which should be examined with great son of the usability of three kits (TL-A, TL and TP) in
care. Application of TLC to such aims is attractive the monitoring of basic drugs is presented in Table 3.
as this technique enables simultaneous analysis of Another example of screening methods is the detec-
several samples. tion of an unknown, potentially toxic xenobiotic in
Laboratory screening methods are similar to the presence of a number of endogenous substances.
diagnostic analyses, but the object of investigation is Such deRned screening methods are focused on limit-
usually known. During the screening other properties ing the number of expensive methods of analyses.
such as changes in the body under xenobiotics inSu- Metabolic screening is based on observation of
ence, e.g. changes in enzyme activity, creation of changes in biochemical proRle of patients within
methemoglobine, increased porphyrin expulsion, etc. a relatively short time. The commonest analyses are
are measured. Because drugs or their metabolites are for cholesterol and lipids or amino acids and por-
excreted and concentrated in urine, this is the matrix phyrins. An example of such an application of TLC is
most commonly used for such examination. Espe- the work of Lai et al. They proposed a simple test for
cially useful for screening investigations are kits such the determination of porphyrins (porphyrins play
as the Toxi-Prep (TP) kit, proposed by Steinberg and a crucial role in the biological processes of haem
coworkers which can simultaneously extract up to synthesis; in the case of defects in the biosynthetic
Table 3 Overall urine basic drug analysisa
Drug Occurrences on
Both TP and TL TP only TL-A only
Amitryptyline and metabolites 5 4 2

Caffeine 0(1)
Cimetidine 1(1)
Clindamycin and metabolites 2 4
Codeine 1
Cyclobenzaprine 1
Desipramine and metabolites 1 1
Diphenhydramine 5(3) 1(7)
Doxepin 1
High migrator 13 3 5(3"cocaine;
2"lidocaine)
Imipramine 1
Low migrator 0(1)
Meperidine and metabolites 1(1) 1
Methadone and metabolites 16(1) 5
Metoprolol 1
Morphine 0(1) 0(2) 6
Nicotine and metabolites 17 2 34
Nortriptyline and metabolites 16 10(2) 2
Oxycodone 1
Phenotiazine 3 1
Quinidine/quinine 46 8(5)
Ranitidine 2
Sympathomimetic amines 2(2) 0(1)
Trazodone metabolites 2(1)
Verapamil 2 2
Unidentified substance 1 3
Totals 141 59 64
(33 distinct drugs) [133(8)]b [38(21)]b [64]b
a
Reprinted from Steinberg DM et al. (1997) Clinical Chemistry 43(11): 2099}2105, with permission from the American Association of
Clinical Chemistry.
b
Number of times definitively identified (questionable identifications).
pathway of haem an increase of porphyrin excreted is

observed). Numerous groups of people are occupa-
tionally exposed to the substances causing por-
phyries, and these have to be periodically screened.
The urinary porphyrin-free acids are separated on
a RP-HPTLC plate coated with C18 bonded silica gel.
The mobile phase is buffered (pH 4.1) with a ternary
mixture (acetonitrile, ammonium acetate buffer, ion-
pairing reagent). Porphyrins, separated by TLC, cre-
ate spots with a characteristic proRle (Rngerprints),
and these proRles (differing in the quantity of ex-
creted porphyrins) are observed in porphyria (Fig-
ure 1). By observing chromatograms in UV light, the
basic criteria of metabolic screening can be fulRlled.
TLC has the advantage over other methods in that
it can screen for many drugs simultaneously, with
a relatively high sensitivity, and can handle several
samples per plate. Figure 2 Representative TLC chromatograms of (A) a blank
plasma sample, (B) derivatization product of baclofen after ex-
Drug Concentration Monitoring traction from volunteer plasma (70 ng baclofen mL\1), (C) de-
rivatization product of the fluoro analogue after extraction from
The pharmacological action of many drugs depends
volunteer plasma (320 ng of the fluoro analogue mL\1). Re-
not on the amount taken but on the concentration in printed from Krauss D, Spahn H and Mutschler E (1988)
the blood. Relations between these two values (dose Arzneimittel Forschung Research 38(II)(10): 1533}1536, with
and therapeutic concentration) depend on the drug permission.
and often have individual characteristics. It was
found that after taking the same dose of phenytoin and rapid methods, which indicates the special suit-
(drug in anticonvulsants) its concentration in patients ability of TLC for such analysis. Drug concentration
blood differed more than twentyfold. Therefore, one monitoring requires (in contrast to diagnostic and
of the tasks of clinical medicine analyses is to estab- screening methods) quantitative analysis.
lish relationships between the concentration of the A good example of the application of TLC for drug
drug in blood and its dosage. Such analyses are called concentration monitoring is the work by Krauss et al.
drug concentration monitoring. They are performed on baclofen and its Suoro analogue (these are central-
mainly on the dosage of drugs of little therapeutic ly acting muscle relaxants, which are used for the
value, combined therapy, etc. Drug concentration treatment of spastic disorders). Gas chromatography
monitoring requires use of relatively simple, cheap with electron-capture detection requires a time-
Figure 1 (See Colour Plate 69). RP-HPTLC chromatograms of fecal porphyrins of porphyria cutanea tarda (PCT) and variegate
porphyria (VP) patients. Left panel: lane 1, VP feces of patient 1; lane 2, VP feces of patient 2; lane 3, external quality control sample of
VP; lane 4, mixed calibrators of free porphyrins; lane 5, PCT feces; d, deuteroporphyrin. Middle panel: lane 1, mixed calibrators of free
porphyrins; lane 2, i-urobilin standard; lane 3, stercobilin standard. Right panel: S, mixed calibrators of free porphyrins; NU, commercial
urine control with a negative Urobilistix result; PU, urine obtained from a patient with hepatitis and a positive Urobilistix result; 8,
uroporphyrin; 7, heptacarboxylic porphyrin; 6, hexacarboxylic porphyrin; 5, pentacarboxylic porphyrin; 4, coproporphyrin, 3, tricar-
boxylic porphyrin; 2, protoporphyrin. Reprinted from Lam C-W, Lai C-K and Chan Y-W (1998) Clinical Chemistry 44 (2): 345}346, with
permission from the American Association of Clinical Chemistry.
consuming separation and derivatization step; HPLC were separated from constituents of the biological
does not require derivatization but the detection limit material (Figure 2). The Suorescence intensity of the
is insufRcient for such studies. Several enantiospeciRc derivatization products is sufRcient to determine
methods have also been developed but their applica- plasma concentrations over a suitable period of
bility to biological material has not been shown. The time after single dose administration of both drugs.
method proposed by Krauss is based on Suorescent The detection limit (10 ng mL\1), linearity (60 ng
derivatization of the drug with benoxaprofen chlor- spot\1) and method deviation have been estimated.
ide, TLC separation of the resulting amides and their The applicability of the method has been proved by
quantiRcation by direct measurement of the Suores- investigating plasma (and urine) samples of two vol-
cence. This method requires extensive puriRcation of unteers after oral administration of 20 mg baclofen as
the samples (e.g. solid-phase extraction) in order a single dose. Full suitability of the method for phar-
to remove interfering endogenous amino acids. macokinetic routine analyses was demonstrated.
Chromatograms are obtained with a stationary phase
Evaluation of the Ef\ciency of New Methods
of silica gel 60 without a Suorescence indicator
of Therapy
(plates with concentrating zones) and a mobile phase
of diisopropyl ether}2-propanol}tetrahydrofuran, Evaluation of the efRciency of new methods of ther-
90 : 6 : 5, v/v. After development, both compounds apy mainly concerns drugs newly introduced into
Figure 3 (See Colour Plate 70). OPLC}DAR profile of dog urine extracts. Sample, dog urine samples after 10 mg kg\1 oral dosing of
14
C-daramciclane. The first and fourth tracks are standards (St); tracks 2 and 3, 0}24-h and 24}48-h urine fractions, respectively. Single
OPLC development eluent A; external pressure, 5.0 Mpa; flow rate, 250 L min\1, rapid volume, 250 L; eluent volume, 4100 L; time,
994 s; DAR run time, 60 min. Reprinted from SzuH nyog J, Mincsovics E, Hazai I and Klebovich I (1998) Journal of Planar Chromatogra-
phy}Modern TLC 11 (1): 25}29, with permission.
clinical medicine. From an analytical point of view it introduce the fractions obtained to other analytical
is a more complicated process than drug concentra- instruments. From the analytical point of view,
tion monitoring since the object of investigation is not in these methods TLC has the role of a clean-up
only concentration of the drug in blood but also its technique.
assimilation and excretion, harmfulness, metabolism, In analytical practice two basic methods of TLC
etc. Such investigations are performed mainly on combined with other analytical methods are applied.
blood and urine samples. The older method of off-line coupling is similar to
An interesting example of abilities of TLC in preparative cleaning of the sample. Eluate obtained
the above applications is the work of SzuH nyog et al. after chromatographic separation and sorbent separ-
They applied a combination of overpressure-layer ation is introduced into the analytical instrument.
chromatography (OPLC) with the relatively high sen- The technique should be used when the recovery of
sitive and rapidity of digital autoradiography (DAR). an analyte is of secondary importance. An advantage
Biological samples were analysed after one- or of off-line methods is the possibility of performing
two-dimensional separation. Using the example of analysis with optional detectors. In on-line methods
deramciclane (a new anxiolytic compound) they the TLC plate is introduced to the detector step-by-
demonstrated, that the proposed method could be step. Analytes are removed from the adsorbent by
successfully used for study of in vivo metabolism in cathode sputtering or laser desorption. An important
different phases of studies. The main advantages of advantage of such methods is the full quantitative
the method were the extremely rapid separation and analysis of practically all substances of clinical inter-
puriRcation by OPLC, the possibility of the quantitat- est. Unfortunately, these techniques are expensive
ive evaluation of metabolites over a wide linear range, and require very speciRc injectors that are not usually
visual and numeric comparison of metabolite proRles commercially available.
in various biological matrices and cost-effective meta- Martin et al. applied TLC coupled with MS}MS to
bolism studies (Figure 3). the analysis of antipyrine and its metabolites in ex-
tracts of human urine. The urine samples were ob-
tained from a volunteer who had received a single
TLC as a Clean-up Technique oral dose of antipyrine. After 12 h, an aliquot was
The basic method of applications of TLC as a clean- enzymatic hydrolysed to liberate the analytes
up technique consists of the preparative separ- from glucuronide conjugates. Next, analytes
ation of analytes and removal from the TLC plate were extracted (LLE), concentrated and separated
together with adsorbent and elution with an appro- by TLC. A good separation (Figure 4) was ob-
priate solvent. This technique is used mainly in tained on silica gel HPTLC plates with a mobile
pharmacological work. In clinical medical analyses phase of chloroform}methanol}triSuoro-acetic acid,
the method is used to separate mixtures and then 95 : 5 : 1, v/v.
Figure 4 Chromatograms obtained from (A) a urine sample following enzymic hydrolysis and solvent extraction and (B) a standard
mixture of the analytes and the internal standard: o, origin; sf, solvent front; 3OHA, 3-hydroxyantipyrine; DMAA, dimethylaminoan-
tipyrine; A, antipyrine; 4OHA, 4-hydroxyantipyrine; NORA, norantipyrine. Reprinted from Martin P, Morden W, Wall P and Wilson
I (1992) Journal of Planar Chromatography}Modern TLC 5(4): 255}258, with permission.
Figure 5 TLC}FABMS spectra obtained directly from the silica gel for (A) antipyrine and (B) material co-chromatographing with
antipyrine from the urine extract (insets: structure and UV spectra). Reprinted from Martin P, Morden W, Wall P and Wilson I (1992)
Journal of Planar Chromatography}Modern TLC 5(4): 255}258, with permission.
The appropriate areas of the plate were then re- in the number of such analyses is about 15}20%
moved for further analysis by FAB MS and FAB with systematic increases in the number of different
MS}MS. The authors demonstrated (Figure 5) that clinical chemical parameters. However, not all
this relatively simple technique would appear to make clinical analyses can be performed using TLC. TLC
TLC}FAB MS}MS eminently suitable for conRrming is relatively simple, comes with low equipment
the identity of drug metabolites in body Suids for the cost, and has both a high efRciency and sensitivity.
study of drug metabolism or the detection of drugs Some disadvantages of TLC include only fair speciR-
abuse, etc. city and the requirement of skill in accurately re-
cognizing drug patterns by interpreting the coloured
spots visualized with selective reagents, etc. It should
Conclusions also be emphasized that TLC is a comparative
The analytical tasks connected with clinical needs method; more complicated tasks in clinical medi-
are very large. It is estimated that the annual increase cine (investigation of metabolism and metabolite
2484 III / CLINICAL DIAGNOSIS: CHROMATOGRAPHY
structure) can be performed by this method but only Jork H, Funk W, Fischer W and Wimmer H (1994) Thin-
in combination with other, mainly spectroscopic, Layer Chromatography: Reagents and Detection
analytical techniques. Methods } Physical and Chemical Detection Methods;
Activation Reactions, Reagent Sequences, Reagents II,
See Colour Plates 69, 70. vol. 1b. Weinheim: VCH.
Krauss D, Spahn H and Mutschler E (1988) QuantiRcation
See also: II/Chromatography: Paper Chromato-
of Baclofen and its Suoro analogue in plasma and
graphy. Chromatography: Gas: Detectors: Selective;
urine after Suorescent derivatisation with Benoxa-
Gas Chromatography}Mass Spectrometry. Chromato-
profen chloride and thin-layer chromatographic
graphy: Liquid: Detectors: Mass Spectrometry.
separation. Arzneim.-Forschung/Drug Res. 38(II):
Chromatography: Thin-Layer (Planar): Densitometry
1533}1536.
and Image Analysis; Layers; Mass Spectrometry; Modes
Lai CK, Lam CW and Chan YW (1994) High-performance
of Development: Conventional; Modes of Developement:
thin-layer chromatography of free porphyrins for
Forced Flow, Over Pressured Layer Chromatography and
diagnosis of porphyria. Clinical Chemistry 40:
Centrifugal; Spray Reagents. Extraction: Analytical Ex-
2026}2029.
tractions; Solid-Phase Extraction; Solvent Based Separ-
Martin P, Morden W, Wall P and Wilson I (1992) TLC
ation; Supercritical Fluid Extraction. III/Amino Acids:
combined with tandem mass spectrometry: application
Thin-Layer (Planar) Chromatography. Bases: Thin-Layer
to the analysis of antipyrine and its metabolites in ex-
(Planar) Chromatography. Bile Compounds: Thin
tracts of human urine. Journal of Planar Chromatogra-
Layer (Planar) Chromatography. Biomedical Applica-
phy } Modern TLC 5: 255}258.
tions: Gas Chromatography-Mass Spectrometry; Thin-
Moffat AC (ed.) (1986) Clarke’s Isolation and IdentiTca-
Layer (Planar) Chromatography. Carbohydrates: Thin-
tion of Drugs. London: Pharmaceutical Press.
Layer (Planar) Chromatography. In-Born Metabolic
Steinberg DM, Sokoll LJ, Bowles KC, Nichols JH, Roberts
Disorders: Thin-layer (Planar) Chromatography.
R, Schultheis SK and O’Donnell CM (1997) Clinical
Lipids: Thin-Layer (Planar) Chromatography. Proteins:
evaluation of Toxi-Prep: a semi-automated solid-phase
Thin-Layer (Planar) Chromatography.
extraction system for screening of drugs in urine. Clini-
cal Chemistry 43: 2099}2105.
Further Reading SzuH nyog J, Mincsovics E, Hazai I and Klebovich I (1998)
A new tool in planar chromatography: combination of
Jain R (1996) Thin-layer chromatography in clinical chem- OPLC and DAR for fast separation and detection of
istry. In: Fried B and Sherma J (eds) Practical Thin-layer metabolites in biological samples. Journal of Planar
Chromatography. A Multidisciplinary Approach, ch. 7, Chromatography } Modern TLC 11: 25}29.
pp. 131}152. Boca Raton: CRC Press. de Zeeuw RA, Franke JP, Degel F, Machbert G, ShuK tz
Jork H, Funk W, Fischer W and Wimmer H (1994) Thin- H and Wijsbeck J (eds) (1992) Thin Layer Chroma-
Layer Chromatography: Reagents and Detection tographic Rf Values of Toxicologically Relevant
Methods } Physical and Chemical Detection Methods; Substances on Standardized Systems. Weinheim:
Fundamentals, Reagent I, vol. 1a. Weinheim: VCH. VCH.
CLINICAL DIAGNOSIS:
CHROMATOGRAPHY
I. D. Watson, University Hospital, Liverpool, UK ically; immunology } assessment of antibody status in

disease; molecular biology and cytogenetics } special-
ist services looking at genetic disease. Of these depart-
ments, the clinical biochemistry repertoire is the most
Clinical laboratories within the UK are responsible amenable to chromatography.
for providing services for the diagnosis and monitor- Clinical biochemistry laboratories in an average
ing of disease. There are several specialities within district general hospital perform over a million ana-
clinical laboratories or pathology: haematology } ex- lyses per annum. The large volume of samples (aver-
amining blood cells and factors relating to blood cell age request/analysis ratio &1 : 4) received each day,
production; biochemistry } measurement of metab- the clinical demand for rapid turnaround plus the
olites, hormones, drugs, proteins; histopathology r need for analytical imprecision of less than 5% with
examination of tissues and cells, usually microscop- acceptable relative accuracy means that high levels
III / CLINICAL DIAGNOSIS: CHROMATOGRAPHY 2485
Table 1 Analytes, methodology, volume and cost: Illustration of a ‘typical’ clinical biochemistry laboratory workload,
ranked in order of cost
Analyte Tests per day Method Cost /test a (approx)

(approx)
Urea and electrolytes (Na#, K#) 4000 Automated wet chemistries and C1.00 each, i.e. C4.00 profile
ion-selective electrodes
C-reactive protein 50 Automated nephelometry C4.00
Thyroid function 200 Automated non-isotopic immunoassay C5.00
(thyroid stimulating hormone, thyroxine)
Bedside device for a single drug of abuse 1 Immunochromatography C5.00
Serum proteins 30 Electrophoresis C15.00
Drugs of abuse 30 Automated immunoassay and C15.00
chromatography
Haemoglobin A1c 50 Automated chromatography C10.00
a
Cost includes consumables, reagents, labour overheads and equipment depreciation.
of automation are not only desirable but essential. throughput, yet surprisingly this potential has never
Most chemistries used are either wet chemistry or been utilized apart from a few aRcionados. Tradi-
dry-Rlm technology. The nature of certain analytes tional TLC lacked chromatographic efRciency with
may make them less amenable to the high volume consequent loss of resolution and lack of sensitivity.
chemistries or the volumes of the rarer analytes make Widely used in many formative studies over 30 years
the development of automated chemistries more ex- ago, having replaced paper chromatography, it is still
pensive or time consuming. Chromatography Rts this used for analytes such as testing drugs of abuse. To
last scenario in clinical laboratories. An illustrative ensure reproducibility from plate to plate, commer-
workload pattern is presented in Table 1. cial plates are used, often with a Suorescent indicator
Clearly chromatography is utilized for those tests to aid detection although Rnal detection relies on
that are low volume and for which, generally, rapid location reagents.
turnaround is not required. Table 2 illustrates the Nearly 20 years ago two developments still in cur-
analytes, the mode of chromatography and reasons rent use greatly improved the utility of TLC: the
for use. advent and provision of HPLC column packing ma-
terials led to their introduction in TLC giving the
anticipated high performance, i.e. high performance
Thin Layer Chromatography (TLC) thin-layer chromatography (HPTLC). The improved
Of all chromatographic techniques this is the best resolution and sensitivity when coupled to signiRcant
for parallel analyses enabling cost-effective high improvements in densitometers and location re-
Table 2 Common current clinical chromatographic applications
Analyte class Chromatographic mode a Reason for use
Haemoglobin A1c LC Most specific, rapid

Haemoglobin variants LC Specific for certain variants
Metabolites associated with LC/GCMS Specificity/flexibility
inborn errors of metabolism LC}MS}MS Introduced to selected centres as
cost-effective and high specificity
Drugs
Therapeutic drug monitoring LC/GC Cost/flexibility
Drugs of abuse
Screening TLC/HPTLC/GC Specificity/flexibility/cost
Confirmation GC/GCMS/LC}MS Specificity
Toxicology TLC/GC/LC Flexibility
Markers of bone turnover LC Specificity, being superseded by
immunoassay
Near patient testing devices, Immunochromatography Convenience
e.g. drugs of abuse ‘stick’ tests
a
LC, liquid chromatography; GC, gas chromatography; MS, linked mass spectrometer; TLC, thin layer chromatography; HPTLC, high
performance thin layer chromatography.
agents, meant that for the Rrst time quantitative TLC ed method in clinical laboratories performing sub-
with performance equivalent to GC or LC was pos- stance abuse testing (Figure 1).
sible. While some laboratories, particularly in Ger- However, the other development in TLC was the
many, enthusiastically adopted this technique it development of a cellulose-based commercial system
gained very few adherents elsewhere, particularly in with a stylized Marquis reaction with reference to
clinical laboratories where the demand and skill base RF and sequential colour changes collected in a com-
were inadequate. pendium. This system was developed for clinical toxi-
Since the mid-1980s there has been an increasing cology work allowing it to be performed by the non-
problem of drug abuse. The initial assays developed specialist. In overdoses it works well despite its lack
for screening in the late 1970s used enzyme-multi- of chromatographic efRciency. Unfortunately the sys-
plied immunoassay technique (EMIT), followed tem was inappropriately applied to drugs of abuse
a little later by Suorescence polarization immunoas- testing which has greater sensitivity and speciRcity
say (FPIA). These techniques were good for screening requirements and external quality assurance data
large numbers of individuals for a range of substances consistently demonstrate poor performance.
or classes of substances, e.g. cannabis, cocaine, opi- In clinical practice, knowledge of which drug
ates, benzodiazepines, barbiturates. The test for opi- has been taken as determined by laboratory studies is
ates was designed to detect heroin abuse but in fact it useful for only a very few drugs and the concept of
detects other opiates, i.e. codeine and dihydro- screening, usually using chromatography, is no
codeine, which are legitimately available. Clearly, longer commonly performed in the UK; in difRcult
therefore, any positive for this assay system, and also cases, however, this is the preferred method of
for others, requires conRrmation. Immunoassay- investigation.
based methods inherently depend on the speciRcity of TLC has a further unique property in clinical in-
their antibody and may not detect subtle structural vestigation. The plate complete with separation can
differences between compounds leading to false pos- be sent from the original investigating laboratory to
itives. This clearly has implications for the subject a reference centre which will be able to investigate
tested, whether for clinical, forensic or employment directly (perhaps using TLC-MS-MS) (Figure 2) any
purposes. It is now a Rrmly established principle, difRculty to identify/conRrm spots. Currently this
regretably not always adhered to, that before taking type of approach is a neglected area.
action conRrmation with a non-correlated technique
should be performed. Chromatography is the tech-
nique of choice.
Gas Chromatography (GC)
In clinical testing, HPTLC with appropriate loca- The heyday of GC in clinical laboratories was in the
tion reagents and visual inspection is adequate as 1970s. Biological Suids contain proteins in high con-
a conRrmation technique, although this would usu- centration (&75 g L\1) in a complex mixture of
ally be supported by other chromatogaphic modes. endogenous and exogenous metabolites and the com-
HPTLC also enables screening for many of these pound of interest may be nonvolatile and present in
compounds not detected by the immunoassay low concentrations. The trick therefore was to extract
screens; consequently HPTLC is a dynamically utiliz- the compound of interest quantitatively, make it vol-
Figure 1 HPTLC of extracted urine from drug abusers. Morphine, the principal metabolite of heroin, is indicated with an arrow.
Figure 2 HPTLC } fast atom bombardment spectra of nitrazepam (100 ng on plate). Daughter on 282, delta of 46 (MW 236)
characteristic of nitrazepam. (Reproduced with permission from Watson ID (1998) Therapeutic Drug Monitoring 20: 490}497.)
atile and chromatograph it. Method development molecular biology, many cases are sporadic muta-
could take many man-hours and while some auto- tions requiring resolution for the optimal care of that
mation pre- and post-chromatography could be individual. Failure to achieve this results not only in
performed, these were labour-intensive methods incapacity for the individuals and their family, but
best suited to low molecular weight metabolites. also in signiRcant costs to the health service.
In the 1980s with increasingly better LC systems, Capillary GC is satisfactory for a clinical service for
GC usage in clinical laboratories went into drugs of abuse although GC-MS is essential in any
decline. forensic or employment issues, in which some clinical
Continued use of GC is necessary for a few laboratories are involved.
analytes, however. Methanol, ethylene glycol, pro-
panol and ethanol poisoning present overlapping
clinical pictures; knowledge of which alcohol has
been taken and how much is vital. GC with
Same ionization detection is the ideal method
giving rapid, reliable results and can be utilized
in an emergency situation (Figure 3). Some laborator-
ies still use GC for drug analysis.
Capillary GC with Same ionization and/or
nitrogen}phophorus detectors offers high sensitivity
and in the latter case good speciRcity for determining
biological analytes. Applications cover substance
abuse, therapeutic drug monitoring and intermediate
endogenous metabolites. However, GC-MS as bench-
top analysers, either ion-trap or quadrupole, are be-
coming more common, though they are still rare, in
clinical laboratories. The main reasons for the poor
uptake of this combination are cost, demand and
staff skills required.
As noted in Table 2 intermediate metabolites
found in inborn errors of metabolism which can lead
to signiRcant morbidity and mortality, are currently
detected by capillary GC. Early detection is vital, and Figure 3 Gas chromatography of alcohols in blood. (Repro-
although family screening can now be performed by duced with permission from Tames F.)
Liquid Chromatography (LC)

Liquid chromatography, with its wide range of modes
and better biocompatibility than other chromato-
graphic methods, has been utilized in clinical laborat-
ories. Classical column chromatography was used in
sample preparation of many of the standard wet
chemistry methods in the past. Protein isolation using
afRnity chromatography enabled isolation of anti-
bodies and subsequent labelled antibodies to pro-
vide the Rrst radioimmunoassays in the 1950s and
1960s. Moore and Stein in 1954 developed an amino
acid analyser which was reRned over the years. Al-
though primitive by today’s standards, it enabled the
Rrst reliable measurement of the common amino
acids allowing investigation into their metabolism,
role in nutrition and relevance in inborn errors of
metabolism. Amino acid analysers were automated
Figure 4 Isocratic liquid chromatography of anticonvulsants in
with post-column reaction for detection and to a lim-
serum. Analysed on Gilson ASTED system with sample pre-dialy-
ited extent speciRcity. Much of the basic knowledge sis and trace enrichment prior to chromatography. (Reproduced
learnt through this pioneering work formed the basis with permission from O’Connell, D.)
for subsequent biological applications of LC.
Ten to Rfteen years ago LC was making a signiR- core still use LC, though whenever possible using an
cant impact in research-orientated clinical laborator- automated system. Surveys through the UK National
ies. Initially work focused on reversed-phase mater- External Quality Assurance Programme have consis-
ials with many publications on the separation of tently shown LC methods to have the best accuracy
a wide variety of drugs, endogenous metabolites and with acceptable imprecision. The introduction of new
steroid hormones. A major difRculty was sensitivity anticonvulsants, e.g. lamotrigine, has led to a demand
and there was no true universal detector, nonetheless for analysis. This has meant development of LC as-
therapeutic drug monitoring already expanding due says demonstrating the role of chromatography in
to the availability of EMIT technology consolidated development of clinical investigations.
using LC procedures for previously difRcult analytes. This is a common scenario in method development
A classic example was the common anticonvulsant in that research or reference centres develop an LC
carbamazepine which suffered thermal degradation assay for a new analyte to detect or monitor disease.
on GC leading to signiRcant imprecision. This was As the utility of this determinand is demonstrated,
resolved by using LC. Additionally, it became readily demand rises, causing processing difRculties for the
possible to measure several drugs simultaneously. reference centre, parallel to an increasing demand
While this had been done using temperature-pro- from less specialist centres. Some will invest in an LC
grammed GC it was frequently necessary to derivatize solution but the reagent manufacturers are alerted to
compounds to obtain satisfactory volatility and po- the developing interest in the analyte. Knowing that
larity and this could affect the selectivity. LC enabled clinical laboratories have a predilection for rapid,
the same approach, initially using solvent gradient automated large single or multiple similar medium
programming on underivatized samples, often with analysers, they develop wet chemistry or immunoas-
minimal sample preparation. This approach enabled say methods. This approach will be reagent expensive
separation of a full range of anticonvulsants: ethosuc- but require minimal labour utilizing pre-existing
cimide, primidone, phenytoin, phenobarbitone and equipment. Once this is marketed and adopted the
carbamazepine and their metabolites (Figure 4). LC procedure, albeit more accurate, declines and may
This meant that efRcient processing, all samples be dropped altogether.
followed the same analytical track, reduced costs per A further illustration of this is the measurement of
analyte and the occasional detection of inappropriate bone turnover; the original analyte hydroxyproline
medication. In this particular area the wide and in- was determined by a cumbersome, manual, wet
creasing range of nonisotopic immunoassays compat- chemistry assay. As interest grew in hormone replace-
ible with automation has meant many clinical labor- ment therapy effects in osteoporosis much effort was
atories Rnd it more organizationally efRcient to per- expended in developing a gold standard LC assay for
form these assays by automated immunoassay; a hard pyrodinoline and deoxypyrodinoline. There were
particular problems, eventually resolved, with ob-

taining a satisfactory internal standard. This has be-
come established as the reference method, but is now
being supplanted by immunoassay measurement of
deoxypyrodinoline.
Screening methods using LC have suffered from the
lack of reproducibility between reversed phases. The
REMEDI system for drug abuse screening is however
an established screening technique linked to a linear
diode array with spectral library and has been shown
to provide satisfactory identiRcations on screening for
drugs.
The detection of inborn errors of metabolism by
reference paediatric biochemistry laboratories has
hitherto relied heavily on GC-MS with all the in-
herent problems plus the need for separate assay
condition for different compound classes, e.g. organic
acids and amino acids. Recently the NHS Research
and Development Programme } Health Technology
Assessment } after an evidence-based medicine re-
search exercise, has indicated that LC}MS}MS
would deliver appreciably greater beneRts than cur-
rent systems for the detection of a variety of different
forms of inborn error of metabolism. Hitherto, fund-
ing for such equipment had been stalled on the capital
cost, but demonstration of the economies delivered
by LC tandem MS argued in favour of the system
which is now being introduced in selected centres. It
is such economic analyses that may progress and
sustain chromatographic methods in the face of a de-
creasing skill base and drive for consolidation in
clinical laboratories.
An example of where the speciRcity of LC is valued
nd is allied to an improvement in efRciency is the
analysis of haemoglobin A1c. (Hb A1c). Hb A1c is
glycated haemoglobin A, and is a long-term measure
of control in diabetes mellitus. Recent international
advice has called for close control of Hb A1c in
diabetics and requires an accurate and precise Figure 5 Ion exchange chromatography of haemoglobin A1c,
method. Early studes on Hb A1c used classical col- an indicator of glucose control in diabetics. Hb A1c retention 1.75
umn chromatography on mini-columns, this was la- min.
bour intensive and commercial ‘kit’ LC solutions be-
gan to be offered. Soon an automated wet chemistry
analyser compatible methods were developed and As diabetic assessment is performed regularly and
embraced. many organ systems are examined, it is not uncommon
However, signiRcant proportions of the popula- for blood to be drawn from a patient and the Hb A1c
tion, depending on ethnic mix, do not have haemo- to be measured in the clinic using LC while they wait.
globin A as their sole haemoglobin and to assess This provides the clinician with the result when seeing
glycaemic control one must look at the glycation of the patient and improves the patient throughput.
the variant haemoglobin; the wet chemistry method
cannot do this. There has therefore been a resurgence
of interest in the ‘kit’ LC solutions; popular ones use
Immunochromatography
anion exchange chromatography which have run Immunochromatography is the preserve of research
times of 4 min providing full sample automation for laboratories as a laboratory technique. Near-patient
over 250 samples per day (Figure 5). testing, however, utilizes commercially produced
2490 III / COAL: FLOTATION
positive and negative controls. When the antigen and

antibody combine they develop a visible colour spot
or band which conRrms the presence of the com-
pound of interest (Figure 6). While these devices are
expensive and inaccurate they have the beneRt of
immediacy which may be clinically acceptable pro-
vided they are used appropriately.
The Future
Chromatography has maintained its role in certain
niches in clinical laboratories. Interest in manufac-
turer-supplied solutions for chromatography, par-
ticularly LC, exists and compensates for the lack
of skill base. For difRcult low throughput analyses
this may be how developments will be consolidated.
Capillary zone electrophoresis could impact on much
current LC work but again skill and capital costs
militate against this. If accuracy rather than impreci-
sion becomes a major clinical laboratory issue, as it
may, then the inherent accuracy of chromatography
probably linked to mass spectrometry will provide
a role for deRnitive methods and may provide a role
for methods used in routine laboratories.
See also: II/Chromatography: Thin-Layer (Planar):

Mass Spectrometry. III/Toxicological Analysis: Liquid
Chromatography.
Further Reading
Baselt RC (1987) Analytical Procedures for Therapeutic
Figure 6 Immunochromatography slide for once-only use near Drug Monitoring and Emergency Toxicology, 2nd edn.
the subject testing to detect abused drugs. C"control, Massachussetts: PSG Publishing.
THC"tetrahydrocannabinol, a metabolite of cannabis. Bowers LD, Ullman MD and Burtis CA (1994) Chromatog-
raphy in: Burtis CA and Ashwood ER (eds) Tietz Text-
book of Clinical Chemistry, 2nd edn. Philadelphia: WB
‘sticks’ which use immunochromatography. The prin- Saunders.
ciple is that the sample, e.g. urine, is applied to the Kuenigsberger RU (1988) High performance liquid
stick which is then developed, e.g. by capillary attrac- chromatography. In: Williams DL and Marks V (eds)
tion, the analyte of interest binding at a zone where Principles of Clinical Biochemistry, 2nd edn. London:
there are antibodies. There are sometimes built-in Heinemann.
COAL: FLOTATION
B.K. Parekh, University of Kentucky, Lexington, pyrite) has received increased attention since the
KY, USA 1960s. The froth Sotation process is typically used for
Copyright ^ 2000 Academic Press treating (0.5-mm size coal and is currently the only
technique both effective and economical to clean coal
Introduction on a commercial scale. In the USA, the majority of
The process of froth Sotation for upgrading the qual- coal preparation plants discard the (0.5-mm coal
ity of coal by removing mineral matter (ash and owing to the high cost of processing of the Rne coal.
III / COAL: FLOTATION 2491
Recovering Rne coal offers important economic and Hydrophobicity of Coal

environmental functions. In economy terms, the plant
The hydrophobicity of coal varies with the rank of the
recovers an extra amount of clean coal that would
coal and oxygen functional group present in the coal.
have otherwise been discarded to the impoundment.
The high volatile bituminous coals are the most hy-
Recovering the clean coal reduces the amount of Rnes
drophobic, whereas lignite is the least hydrophobic.
to the ponds, and improves the quality of recycled
One technique to quantify the hydrophobicity of coal
water.
is through measuring contact angles of water on coal
The basic coal Sotation technology has been de-
surfaces. Figure 1 shows the contact angle data of
rived from ore Sotation, where the technology has
water on various coals. Note that the maximum con-
been extensively used. The Rrst froth Sotation plant
tact angle or the hydrophobicity occurs at &88% C,
for coal was established in the United Kingdom in
and the high-carbon content anthracite coals are less
1920 and the Rrst US plant was established in 1930.
hydrophobic.
Froth Sotation technology has made substantial pro-
gress over the last Rfty years.
Coal is a solid combustible material and exists in Zeta Potential of Coal
the ground along with impurities. Coal, being com-
posed of carbon elements, is hydrophobic in nature The zeta potential of various coals with respect to
and thus is a good candidate for the froth Sotation pH is shown in Figure 2. Note that the point of
technique. The impurities present in coal basically zero charge (PZC) is highly variable and depends
consist of clays, quartz, calcite, dolomite, pyrite, on the source and type of coal. The sub-bituminous
chlorite, etc. which are hydrophilic in nature and and lignite have a PZC of &2, whereas the high
thus, can be eaily removed in an aqueous medium. volatile (hvA) bituminous and anthracite coals
Pyrite minerals present in coal have an ambivalent have a PZC of &5. In general, a reduction in
character and are sometimes difRcult to remove by the PZC value indicates a reduction in hydropho-
the Sotation technique. bicity.
Figure 1 Hydrophobicity of various coals as a function of per- Figure 2 Zeta potential and point-of-zero charge (PZC) of coals
centage carbon based on the contact angle technique. (Aplan of various ranks. (Aplan and Arnold (1991), courtesy of SME,
(1989), courtesy of SME, Littleton, Colorado, USA.) Littleton, Colorado, USA.)
Reagents
The purpose of Sotation reagents is to provide
a strong hydrophobic surface and to create small
relatively stable bubbles. For coal Sotation, in
general, only the collector and frother reagents are
used.
Collectors
Theoretically speaking, the highly hydrophobic coals
(containing 85}90%C) should not require any collec-
tor. In practice, a small amount of either No. 2 fuel oil
or kerosene is used as collector. The amount of a col-
lector required varies from a low (0.11}0.2 kg t\1)
for a high rank coal to a high (1.0}2.0 kg t\1) for
a low rank coal.
Frothers
The primary function of the frother is to produce
a large quantity of Rne size bubbles. The bubbles
should be able to carry the coal to the surface without
breaking, and once out of the Sotation machine, it
should break down. The most commonly used
frothers for coal Sotation are either pure alcohol, e.g.
Figure 3 A line diagram of a column flotation cell.
MIBC (methyl isobutyl carbinol) or a mixture of
various alcohols and the polypropylene glycol-based
frothers. The amount of frother required varies from Mechanical
0.2 to 0.5 kg t\1.
The most commonly used Sotation machine is one in
which a mechanically driven impeller agitates in the
Depressant
pulp and disperses air into it. The major development
The function of depressant is to suppress the Sota- in mechanical Sotation cells has been in the design of
tion of one component of the mixture of solids by
adding a speciRc chemical. In coal Sotation,
pyrite usually Soats along with the coal. Many
papers have been published on the subject; however,
Chaudhari and Aplan, and Xu and Aplan, have con-
ducted a detailed investigation of various reagents
for depressing Sotation of pyrite. They concluded
that there is no universal reagent for depressing
pyrite. In the coal industry, pyrite depression
is not practised; however, in the future this might
become an important step for the coal industry to
survive.
There are some other variables for Sotation, such
as pH, dispersing reagents, percentage solids, particle
size, etc. However, in the coal industry, very little or
no attention is given to these factors. Readers inter-
ested should refer to Aplan’s work.
Flotation Machines
The coal industry uses either mechanical or column Figure 4 Schematic diagram of the MicrocellTM. (Yoon et al.
cells. (1992), courtesy of Gordon and Breach Science Publishers.)
columns were installed in the USA at the Powell analysis. An ash content reduction of about 40}45%
Mountain Coal Company, Virginia. was achieved while recovering 77% of the combust-
The ‘Mocrocel’ column Sotation developed at the ible material, which equates to an ash rejection of
Virginia Polytechnique and State University uses an about 95%.
inline mixture to generate Rne bubbles in the column. Mohanty and Honaker published a comparative
In this column, a part of the reject stream is passed evaluation of the three leading column Sotation tech-
through the inline mixture along with the frother and nologies. According to them, the packed column pro-
air to generate Rne bubbles. Figure 4 shows a dia- duced the best separation performance owing to its
gram of the MicrocellTM column. ability to support a deep froth zone. However, be-
The Static Tube Sotation system developed at the cause of the absence of an air-sparging system and
Michigan Technology University uses corrugated consequently larger bubbles, the solids-carrying capa-
plates packed inside the column to break up large city of the froth was minimal. On the other hand, the
bubbles into a smaller size bubble. The machine does solids-carrying capacity or solids throughput
not utilize any special bubble-generating device. achieved with the Jameson cell, was found to be
The Jamison cell is a column cell that is much maximal. The MicrocellTM achieved maximum carry-
shorter than any of the columns described earlier. As ing capacity while providing a high energy recovery
shown in Figure 5, the Jamison cell utilizes a pressur- with a reasonably low amount of reagents. Figure 7
ized feed stream injected through a long tube called shows the combustible recovery (amount of combust-
the downcomer to draw atmospheric air into the ible material) as a function of product ash and total
device. The resulting jet formed is discharged into sulfur obtained for a coal using the three column
a short column where coal Soats to the surface and technologies. The AFW (advanced Sotation washa-
tailings are discharged at the bottom. Wash water is bility) is a limiting curve indicating the possible ash
generally added through a tray located above the and sulfur that could be achieved using froth Sotation
froth phase. Currently, quite a few Jamison cells technology. Note that the packed column provided
are being used in Australia and a few coal preparation both low ash and sulfur; however, the MicrocellTM
plants in the USA. Figure 6 compares the separa- and Jameson cell both provided high combustible
tion performance as a function of ash content for recovery.
a coal using the laboratory-, pilot- and full-scale In all work on coal Sotation, a high ash rejection
Jamison cell units. The L-shape of the release curve has been reported using column Sotation technology.
indicates that the impurities in the coal are well liber- However, removal of pyritic sulfur has been mar-
ated. The laboratory- and pilot-scale units were near- ginal. The main reason for this is attributed to the
ly able to produce the performance of the release ambivalent nature of coal pyrite; some pyrites are
Figure 8 Schematic drawing of the combined advanced flotation/enhanced gravity separation circuit. (Luttrell et al. (1998), courtesy
of Gordon and Breach Publishers.)
III / COAL: FLOTATION 2495
hydrophobic and some are hydrophilic in nature. In

the case of coal, pyrite is always associated with some
carbon which affords it hydrophobic and makes it
Soat with coal. Luttrell et al. conducted studies on
combining column Sotation with an advanced gravity
separation technology to remove pyrite from coal.
The process Sowsheet for the two-stage circuit is
shown in Figure 8. A MicrocellTM Sotation column
froth product after derating was sent to a multi-
gravity separator (MGS) developed by the Mosley
Ltd, UK. The MGS is similar to the shaking table,
except that the particles in the Sowing Rlm are also
subjected to centrifugal forces. Figure 9 compares the
performance of the Sotation column alone to that
obtained using the combined Sotation column/MGS
circuit. The rejection of pyritic sulfur improved from
60.5% to 83.6% and ash rejection improved from
83.8% to 87.7% using the combined techniques.
Using the MGS unit, the loss of coal (energy) was
very low, in the order of 2}3% points.
Coal Sotation is a manually operated process. Re-
cently, process optimization has been achieved
Figure 10 Optoelectronic tailings detector. (Meenan (1999),
through more efRcient circuit designs and innovative
courtesy of SME, Littleton, Colorado, USA.)
sensor development. Some new approaches have been
developed in the sensor area, which help in designing
a controlled reagent delivery system. The Consol Co. tent, the more light reSected back to the photocon-
has developed an inexpensive optical-based system ductor and the lower the resistance.
shown in Figure 10. The main components of the A machine-vision or ‘video-based’ analyser system
detector are a glass tube, a photoconductor, an has also been successfully tested by Virginia Polytech-
opaque barrier, and a light-emitting diode (LED). The nique Institute and State University. The video-based
light from the LED is reSected from the slurry to the system could detect changes in slurries over an oper-
main photoconductor. The photoconductor changes ating range of 60}90% ash.
resistance in proportion to the light reSected from the
tailing stream. The higher the tailings slurry ash con- Conclusion
In summary, coal Sotation, column Sotation has
shown signiRcant advantages from the technical and
economic points of view. A combination of column
with advanced gravity separators provides a much
cleaner coal with low ash and pyritic sulfur contents.
In the future, all new coal preparation plants will
utilize column technology along with process optim-
ization and sensors for the economic recovery of
ultra-Rne coal particles.
Further Reading
Adel GT, Luttrell GH, Cruz EB and Dunn PL (1997) In-
plant testing of a video-based coal slurry ash analyzer.
Presented at the 14th International Coal Preparation
Conference, pp. 39}156.
Aplan FF (1989) Coal Sotation } the promise and the
problems. In: Chander S and Klimpel RR (eds) Advances
Figure 9 Separation performance obtained using the flotation in Coal and Mineral Processing Using Flotation. pp.
column and combined flotation column/MGS circuit. (Luttrell et al. 95}104. Littleton, CO: Society of Mining, Metallurgy,
(1998), courtesy of Gordon and Breach Publishers.) and Exploration, Inc. (SME).
2496 III / COLLOIDS: FIELD FLOW FRACTIONATION
Aplan FF and Arnold BJ (1991) Flotation. In: Leonard JW Horsley RM and Smith HG (1951) Principles of coal Sota-
(ed.) Coal Preparation, pp. 450}485. Littleton, CO: tion. Fuel 30: 54}63.
Society of Mining, Metallurgy, and Exploration, Inc. Luttrell GH, Venkatraman P and Yoon RH (1998) Re-
(SME). moval of hazardous air pollutant precursors by ad-
Arnold BJ and Aplan FF. (1989) The hydrophobicity of coal vanced coal preparation. Coal Preparation 19:
macerals. Fuel 68: 651}658. 2243}2255.
Bury E. and Bicknell A (1921) PuriRcation of coal by froth Meenan, GF (1999) Modern coal Sotation practices. In:
Sotation. Colliery Guardian 21: 337. Parekh BK and Miller JD (eds) Advances in Flotation
Bury E, Broadbridge W and Hutchinson A (1920) Froth Technology, pp. 309}319. Littleton, CO: Society of
Sotation as applied to the washing of industrial coal. Mining, Metallurgy, and Exploration. Inc. (SME).
Institute of Mining Engineers 60: 243}253. Mohanty MK and Honaker RQ (1999) A comparative
Chaudhari V and Aplan FF (1992) Pyrite depression during evaluation of the leading advanced Sotation technolo-
coal Sotation } Part I. Inorganic. Mining and Mineral gies. Minerals Engineering 12(1): 1}13.
Processing Journal 9: 51}56. Parekh BK, Bland AE and Groppo JG (1990) A para-
Davis DH (1948) Froth Sotation of minus 48 mesh bitumi- metric study of column Sotation. Coal Preparation 8:
nous coal slurries. Transactions AIME 177: 320}337. 49}60.
Gutierez JA, Purcell RJ and Aplan FF (1984) Estimating the Yang DC (1986) Column froth Uotation. US Patent No.
hydrophobicity of coals. Colloids and Surfaces 12: 1}25. 4,592,834 (January 1986).
Honaker RQ, Patwardhan A, Mohanty MK and Bhaskar Yoon RH, Luttrell GH, Adel GT and Mankosa MJ (1992)
K (1999) Fine coal cleaning using the Jamison cell: The The application of microcell column to Rne coal clean-
North American Experience. In: Parekh BK and Miller ing. Coal Preparation 10: 177}188.
JD (eds) Advances in Flotation Technology, Xu DD and Aplan FF Joint use of metal ion hydroxy
pp. 331}341. Littleton, CO: Society of Mining, Metal- compounds and organic polymers to depress pyrite and
lurgy, and Exploration, Inc (SME). ash during coal Sotation.
COBALT ORES: FLOTATION

See III / NICKEL AND COBALT ORES: FLOTATION
COLLOIDS: FIELD FLOW FRACTIONATION
G. Karaiskakis, University of Patras, Patras, Greece hydrodynamic chromatography and Reld-Sow frac-
tionation (FFF).
FFF is a family of separation methods introduced
Introduction by Giddings in 1966, but several groups all around
the world have contributed signiRcantly to the devel-
When dealing with colloidal materials, the para- opment. The various subtechniques of FFF are best
meters of size and size distributions are needed, suited to the separation and characterization of collo-
because many physicochemical processes, like ag- idal materials and macromolecules, including biolo-
gregation and deposition on solid surfaces, are in- gical components ranging in size from proteins to
Suenced by the size and size distributions of these living cells, environmental colloidal particles, as well
materials. as industrial polymers and colloids, powders, latexes
For the analysis of submicron or supramicron and emulsions.
colloidal particles, a sedimentation process, either
under gravity or with centrifugal force, is necessary to
sort different diameter particles into classes, and
then to obtain an average size and a size distri-
Principles of FFF
bution. Among the successful techniques used for FFF is a Sow elution method with retention achieved
this purpose, are electrophoretic mobility and by using applied Relds to drive the solutes into quiet
the methods of size exclusion chromatography, Sow regions. The applied Relds are able to achieve
Aplan FF and Arnold BJ (1991) Flotation. In: Leonard JW Horsley RM and Smith HG (1951) Principles of coal Sota-
(ed.) Coal Preparation, pp. 450}485. Littleton, CO: tion. Fuel 30: 54}63.
Society of Mining, Metallurgy, and Exploration, Inc. Luttrell GH, Venkatraman P and Yoon RH (1998) Re-
(SME). moval of hazardous air pollutant precursors by ad-
Arnold BJ and Aplan FF. (1989) The hydrophobicity of coal vanced coal preparation. Coal Preparation 19:
macerals. Fuel 68: 651}658. 2243}2255.
Bury E. and Bicknell A (1921) PuriRcation of coal by froth Meenan, GF (1999) Modern coal Sotation practices. In:
Sotation. Colliery Guardian 21: 337. Parekh BK and Miller JD (eds) Advances in Flotation
Bury E, Broadbridge W and Hutchinson A (1920) Froth Technology, pp. 309}319. Littleton, CO: Society of
Sotation as applied to the washing of industrial coal. Mining, Metallurgy, and Exploration. Inc. (SME).
Institute of Mining Engineers 60: 243}253. Mohanty MK and Honaker RQ (1999) A comparative
Chaudhari V and Aplan FF (1992) Pyrite depression during evaluation of the leading advanced Sotation technolo-
coal Sotation } Part I. Inorganic. Mining and Mineral gies. Minerals Engineering 12(1): 1}13.
Processing Journal 9: 51}56. Parekh BK, Bland AE and Groppo JG (1990) A para-
Davis DH (1948) Froth Sotation of minus 48 mesh bitumi- metric study of column Sotation. Coal Preparation 8:
nous coal slurries. Transactions AIME 177: 320}337. 49}60.
Gutierez JA, Purcell RJ and Aplan FF (1984) Estimating the Yang DC (1986) Column froth Uotation. US Patent No.
hydrophobicity of coals. Colloids and Surfaces 12: 1}25. 4,592,834 (January 1986).
Honaker RQ, Patwardhan A, Mohanty MK and Bhaskar Yoon RH, Luttrell GH, Adel GT and Mankosa MJ (1992)
K (1999) Fine coal cleaning using the Jamison cell: The The application of microcell column to Rne coal clean-
North American Experience. In: Parekh BK and Miller ing. Coal Preparation 10: 177}188.
JD (eds) Advances in Flotation Technology, Xu DD and Aplan FF Joint use of metal ion hydroxy
pp. 331}341. Littleton, CO: Society of Mining, Metal- compounds and organic polymers to depress pyrite and
lurgy, and Exploration, Inc (SME). ash during coal Sotation.
COBALT ORES: FLOTATION

See III / NICKEL AND COBALT ORES: FLOTATION
COLLOIDS: FIELD FLOW FRACTIONATION
G. Karaiskakis, University of Patras, Patras, Greece hydrodynamic chromatography and Reld-Sow frac-
tionation (FFF).
FFF is a family of separation methods introduced
Introduction by Giddings in 1966, but several groups all around
the world have contributed signiRcantly to the devel-
When dealing with colloidal materials, the para- opment. The various subtechniques of FFF are best
meters of size and size distributions are needed, suited to the separation and characterization of collo-
because many physicochemical processes, like ag- idal materials and macromolecules, including biolo-
gregation and deposition on solid surfaces, are in- gical components ranging in size from proteins to
Suenced by the size and size distributions of these living cells, environmental colloidal particles, as well
materials. as industrial polymers and colloids, powders, latexes
For the analysis of submicron or supramicron and emulsions.
colloidal particles, a sedimentation process, either
under gravity or with centrifugal force, is necessary to
sort different diameter particles into classes, and
then to obtain an average size and a size distri-
Principles of FFF
bution. Among the successful techniques used for FFF is a Sow elution method with retention achieved
this purpose, are electrophoretic mobility and by using applied Relds to drive the solutes into quiet
the methods of size exclusion chromatography, Sow regions. The applied Relds are able to achieve
III / COLLOIDS: FIELD FLOW FRACTIONATION 2497
the task of separation more gently and with more such that the protrusion of the particle out of the
accurate control than the two-phase distribution for- Sow system is determined by the particle’s size rather
ces used in chromatography. than by its Brownian motion.
Just as its name suggests, fractionation in FFF is the
combined effect of an applied cross-Reld and the
laminar parabolic Sow proRle of the carrier mobile Sub-techniques of FFF
phase. The separation process in FFF is carried out in Among the various Relds and gradients used to drive
a thin, unobstructed, ribbon-like channel with a species to the channel wall, four speciRc kinds have
rectangular cross-section and elongated extremities been most developed, yielding a different subtech-
connected to capillary tubes allowing the continuous nique of FFF. Among those used are centrifugal or
Sow of the carrier liquid (Figure 1). Flow velocity gravitational forces (sedimentation FFF"SdFFF),
is highest at the middle of the channel and lowest thermal gradients (thermal FFF"ThFFF), cross-Sow
near the two channel walls. As the carrier stream (Sow FFF"FlFFF) and electrical Relds (electrical
moves through the channel, an external force Reld FFF"ElFFF). All these subtechniques can be worked
or gradient is applied perpendicularly to the Sow, under the normal mode of operation, in which the
which pushes the components of the sample into concentration distribution along the axis which is
the slower moving streams near the outer wall, so perpendicular to the carrier Sow is exponential as
that they form a thin, steady-state layer against that a consequence of the balance between Reld-induced
wall. displacements and diffusion, or under the steric mode
The mean thickness of the layer is different for each of operation in which the elution order is inverted
distinct chemical or particulate species, and depends relative to that observed for normal FFF. They can
on the strength of the interaction between the Reld also be used in the hyperlayer and potential barrier
and the species and on the diffusion coefRcient corre- modes of operation, which will be described in detail
sponding to that species. Larger colloidal species and later.
macromolecules are usually forced more vigorously
toward the wall than are smaller ones. This is the
basis for selective retention in normal FFF, which Retention Theory
produces in general an elution spectrum in which
The peak retention in FFF can be expressed by the
small particles are eluted Rrst and large particles last,
retention ratio, R, deRned as the ratio of the
until steric effects dominate. At this transition point
void volume, Vo, to the component retention volume,
a foldback in elution order occurs, and the larger
VR :
particles elute from the system before the smaller ones
} a trend opposite to that observed in normal FFF.
Steric FFF (StFFF) predominates when the particle Vo
R" [1]
radius exceeds the mean layer thickness of the solute, VR
The retention ratio R in normal FFF is related to the

dimensionless retention parameter ("l/w, where
l is the mean layer thickness and w is the channel
thickness) by the equation:
R"6[coth(1/2)!2]+6 [2]
The last approximation is valid for highly retained

components. Eqn [2] is general for all FFF subtech-
niques, but as the parameter varies with the applied
external Reld or gradient, the experimental parameter
R is related to different physicochemical quantities of
the sample under study, thus leading to the sample’s
separation depending on its interactions with the
external Reld:
Figure 1 Schematic representation of the FFF channel with the
parabolic flow profile in detail. Reproduced with permission from
Giddings JC (1979) Field-flow fractionation of polymers: one- 6kNATs 36kT
R" " 3 (SdFFF) [3]
phase chromatography. Pure and Applied Chemistry 51: 1459. MG d wG
6T 6D Instrumentation
R" " (ThFFF) [4]
aw(dT/dx) DTw(dT/dx)
Sedimentation FFF
The apparatus of SdFFF consists of a channel with
6DVo highly polished walls and rectangular cross-section
R" (FlFFF) [5]
VQ cw2 which is Rtted to the inside of a centrifuge basket.
Solvent is pumped into and out of the system with the
6D aid of special low volume rotor seal and subsequently
R" (ElFFF) [6] to a UV detector. The associated parts of the system
Ew
(e.g. pumps, detectors, recorders, sample injectors)
are readily available liquid chromatography compo-
3d nents. At present, the SdFFF rotor can be operated at
R" (StFFF) [7] a maximum speed of 32 000 rpm, which corresponds
w
to an applied Reld of approximately 85 000 g for the
common rotors used. With such a rotor, colloidal
where M is the solute molecular weight, d is the particles as small as 0.005 m can be retained and
Stokes diameter, "s! is the density differ- analysed, when their density is between 1.5 and
ence between the particle (s) and the carrier 3.0 g cm\3. The upper limit of particle size is reached
liquid (), G is the Reld strength expressed in acceler- when the mean layer thickness becomes equal to or
ation, T is the absolute temperature, k the Boltz- less than the particle radius, and steric factors deter-
mann’s constant, NA Avogadro’s number, D is the mine the retention ratio.
solute}solvent diffusion coefRcient, DT the coefRc- In gravitational FFF (GrFFF) the earth’s gravi-
ient of thermal diffusion, a the dimensionless tational Reld is used and the apparatus consists of
thermal diffusion factor, dT/dx is the temperature a channel formed between two glass plates separated
gradient applied at right angles to Sow, VQ c is the by a Mylar spacer.
volumetric cross-Sow rate, is the particle’s
electrophoretic mobility, E is the electrical Reld Flow FFF
strength, and is a dimensionless factor that
accounts for the drag-induced reduction in velocity, In Sow FFF the external Reld is simply the Sow of
and for the increase in velocity due to the activity of a second solute stream perpendicular to the carrier
lift forces. stream. The walls of the FFF channel are constructed
The above equations show that the solute para- of semipermeable membranes, so as to accommodate
meters controlling retention, and thus characterized the cross-Sow and, at the same time, to retain solute
by retention, are M, d and s in SdFFF, DT and a in particles in the channel. The semipermeable mem-
ThFFF, D and d in FlFFF, and D in ElFFF, and d, brane material is the limiting factor in the perfor-
in StFFF. As in all FFF techniques, the retention ratio mance of the Sow FFF system.
is related to the particle size and the fractionation is
based on particle size differences or on the D and
DT coefRcients. Applications
We note that all R values in the normal FFF mode SdFFF and FlFFF can be applied to the separation and
are inversely proportional to the strength of the Reld characterization of spherical or irregular, monodis-
or gradient employed: G, dT/dx, V c and E } therefore perse or polydisperse colloidal particles. The particle
making possible the manipulation of the retention size analysis of colloids can be carried out successfully
ratios by changes in Reld strength. This makes by working in the four modes of FFF.
FFF applicable to many different sample types over
a wide particle size distribution range. A wide variety
Normal SdFFF
of components can be handled within a run by
applying Reld programming methodologies in which Normal SdFFF has been applied to many systems
the Reld strength is gradually reduced during the involving polymers, biological macromolecules (such
separation. as DNAs, virus particles and protein aggregates),
In the present work, we focus on sedimentation subcellular particles, emulsions and a great variety of
FFF (centrifugal and gravitational) and Sow FFF, natural and industrial colloids. An example of SdFFF
which are the most usable and accurate for the is shown in Figure 2, which illustrates the separation
separation and characterization of colloidal of colloidal polystyrene latex microspheres. No other
materials. method can yield comparable resolution for submic-
perse colloids and emulsions, and the separation

and characterization of narrow populations or sub-
sets within a polydisperse population of colloidal
particles.
A wide variety of inorganic colloids of industrial
importance, including ceramics, carbon black,
quartz, clay silica, alumina, titania, haematite and
hydroxyapatite, have been separated and character-
ized by SdFFF with high resolution in less than
15 min, under optimum conditions. SdFFF also per-
mits analytical separations of submilligram quantit-
ies, and the separated components are easily collected
for subsequent work. SdFFF can be used on a larger
scale to separate quantities of the order of a gram of
puriRed components. Fractionation and characteriza-
tion by SdFFF of the submicron colloidal particles
contained in natural waters have also been carried
out. With the aid of an on-channel concentration
procedure, SdFFF has been shown to be applicable to
an area of analysis where the particle system in natu-
ral water is complex and the particles are present in
much lower concentrations than in samples which
can be analysed without the preconcentration pro-
cedure. Although particle diameter can be obtained
Figure 2 Separation by SdFFF of four polystyrene (PS) by SdFFF only if the particle density is known,
latex beads with the diameters indicated in the scheme. The this apparent weakness can be converted into a
experimental conditions were as follows: field strength 193.7 g; major strength. By changing the density of the
flow rate 12 cm3 h\1; void volume of the channel 2.0 cm3. Re- suspending medium, the difference between it and
produced from Giddings JC, Myers MN, Caldwell KD and Fisher
the density of the solute particles can be changed
SR (1980) Analysis of biological macromolecules and particles by
field-flow fractionation. Methods of Biochemical Analysis 26: 79, by known increments, and it becomes simple to split
with permission from Wiley and Sons, Inc. up the d parameter and terms of eqn [3]. There-
fore, both particle size and density can be found by
SdFFF.
Particle size distribution analyses have been carried
ron particles. One of the advantages of SdFFF in out on a wide variety of suspended organic and inor-
characterizing colloids is the fact that they are separ- ganic particulates (such as polychloroprene and poly-
ated into sharp fractions and that certain properties methyl methacrylate lattices, water-based titania and
of each fraction can be deduced by the measurement carbon black dispersions, phthalocyanine blue dyes)
of the fraction’s retention time. Accurate character- using the methodology of time-delayed exponential
ization is possible without calibration because of the force-Reld sedimentation Reld-Sow fractionation
simplicity and theoretical tractability of the SdFFF (TDE-SdFFF). Particle retention time in TDE-SdFFF
system (cf. eqn [3]), when interactions between the correlates simply and accurately with the logarithm
particles and the channel wall are absent. Particle of particle size (diameter or mass). Relative to con-
parameters subject to characterization include mass, stant-Reld SdFFF, the TDE-SdFFF method provides
density, polydispersity, diffusivity and various par- the advantages of decreased analysis time and im-
ticle dimensions (number and weight average dia- proved detection sensitivity, while maintaining ad-
meters, differential and cumulative size distribu- equate resolution for accurate differential and cumu-
tions). lative particle size distributions.
A series of papers relating to colloid characteriza- SdFFF has been also used for the kinetic study of
tion by SdFFF has been published by Giddings. The aggregation of various inorganic colloids, including
Rrst paper showed that relatively monodisperse hydroxyapatite and sulRde samples. From the vari-
colloidal populations can be characterized with ation of the number average diameter obtained by
respect to particle mass, diameter, density and SdFFF with time, rate constants for the bimolecular
polydispersity. Subsequent papers described methods process of aggregation and stability factors of the
for obtaining particle size distributions for polydis- colloids have been determined.
Normal FlFFF mined by GrFFF with various parameters, such as

solution pH and ionic strength, kind of detergent
FlFFF has been applied to a variety of species in-
and time, in combination with microelectropho-
cluding silica gel, latex and virus particles, as well
resis measurements, useful conclusions about the
as proteins, protein clusters and water-soluble
stability and consequently the aggregation and
polymers. FlFFF has also been applied to humic
deposition phenomena of these sulRdes can be
materials, pushing the lower molecular weight
deduced.
limit to a few hundred. An asymmetric version of
In the normal mode of SdFFF operation the rela-
FlFFF has been introduced in which only one wall,
tionship between VR and d can be established using
the accumulation wall, is permeable to the liquid,
eqn [3], without empirical calibration. In the steric
whereas the other wall is solid. Thus, the cross-
mode of SdFFF (Sd/StFFF), calibration must be
Sow is generated by the fraction of the inlet Sow that
established empirically, as the steric correction factor
exists through the membrane. With this asymmetric
in eqn [7] is unknown, increasing with the Sow
system, various biological macromolecules and
velocity and decreasing with the increase of Reld
water-soluble polymers have been successfully
strength. As retention is based on particle density as
analysed.
well as size, a purely size-based calibration is not
generally valid. By examining the balance bet-
Steric FFF
ween driving and lift forces, it is observed that an
StFFF represents the upper limit of the Reld strength equal retention ratio is obtained for equal size
applied. Although any effective Reld may be app- particles subject to equal driving forces independently
lied to the StFFF mode, the gravitational Reld repres- of the particle density. Thus, by adjusting the Reld
ents the most practical means of utilizing the principle strength G to compensate exactly for the density
of StFFF to the analysis of particles in the diameter difference , so that the product G in eqn [3] is
range 1}100 m. The Rrst experimental evidence the same for the sample and the standard poly-
for the applicability of StFFF was presented in 1978 styrene latexes, linear calibration plots of log(reten-
by Giddings for the fractionation of glass beads tion time) versus log(diameter) for the latexes can be
having diameters 10}32 m. Later the technique used to characterize the sample under study. This
was applied to the separation of a mixture of spheri- calibration procedure for Sd/StFFF has been success-
cal silica beads in the diameter range 5}14 m, fully applied to the rapid size separation and
to the separation of a number of large particle sam- measurement of particle size distribution of various
ples, including chromatographic support materials starch granules.
and Rne-ground coals, as well as to useful applica-
Hyperlayer FFF (HyFFF)
tions in biochemistry and biology (yeast cells, whole
blood). The sedimentation hyperlayer Reld-Sow fractionation
Recently, the StFFF technique has been applied to (SdHyFFF) technique differs from conventional
particle size analysis (number and weight average FFF methods in that the solute zone is focused
particle diameters, as well as differential and cumu- into a thin layer by the opposing forces of centrifugal
lative particle size distributions) of polydisperse acceleration and buoyancy in a density gradient,
supramicron particles of strengite (FePO4 ) 2H2O with which is formed while the carrier is Sowing through
number average diameter of 3.5 m), which is of the channel. So an equilibrium distribution is
great signiRcance in natural waters, and mixed sul- established by the time the Suid reaches the down-
Rdes (CuxZn(1 x)S, 0(x(1, with number average stream injection size. Solutes with low molecular
\
diameter in the range 5.7}7.6 m), which are of weights require impractically high Reld strengths to
paramount importance in industrial chemistry. The form substantial gradients, whereas particulate
particle size distributions, which correspond to the density modiRers establish the gradient both rapidly
longest axis of the needle-shaped FePO4-2H2O and effectively. Ideally each particle type in
particles, found under various experimental condi- SdHyFFF is focused into a very thin layer at its own
tions by StFFF, are in good agreement with those characteristic elevation. Here it occupies a thin
obtained by transmission electron microscopy Suid layer Sowing at a Rxed velocity. The difference
(TEM). This conRrms the feasibility of the StFFF of SdHyFFF from the normal SdFFF is that it is
technique for characterizing polydisperse, irregular, the only FFF methodology for which each distinct
inorganic colloids, and veriRes the general features of species is narrowly conRned to its own unique velo-
the theory. city state.
From the variation of the number and weight The applications of SdHyFFF include the fraction-
average diameter of the CuxZn(1 x)S particles deter- ation of polystyrene latex and biological particles
\
(human and animal cells). With the introduction Potential Barrier FFF (PBFFF)
of denser particles its effective density range has
been extended to silica, thus making the tech- Potential barrier Reld-Sow fractionation (PBFFF),
nique applicable to environmental particles. developed by Karaiskakis, is a combination of poten-
Another application of SdHyFFF technique is tial barrier chromatography and FFF. Potential bar-
the fractionation of coal and limestone par- rier chromatography can be applied to separate par-
ticles. ticles based on differences in size or in any of the
In general, the SdHyFFF method is applicable to physicochemical parameters involved in the potential
larger particles than SdFFF, where (wall-induced) energy of interaction between the particles and the
steric effects become important for particle sizes packing of the chromatographic column. The particle
around 1 m. The focusing of sample zones at loca- size analysis of colloidal materials by PBFFF is based
tions removed from the wall in SdHyFFF has the either on particle size differences or on Hamaker
advantage of eliminating interface-related nonideali- constant, surface potential and Debye}Huckel recip-
ties, such as sample adhesion and zone broadening rocal differences. The method, which is based on the
caused by surface roughness. existence of a surmountable potential barrier between
Important developments include the so-called the colloidal particles and the FFF channel wall, is
hyperlayer Sow FFF, which has extended the app- classiRed as an FFF method rather than a chromato-
lications to particles larger than 1 m in dia- graphic method, because the selective interaction is
meter. Also, the introduction of various program- experienced in one phase. Thus, by combining poten-
ming procedures has been of importance. A number tial barrier chromatography with normal or steric
of reports of the combined use of FlFFF and FFF one could separate according to two mecha-
multi-angle laser light scattering (MALLS) instru- nisms, one governed by the depth of the potential
ments have also been published, illustrating the use- energy well for the different particles and the other
fulness of this hyphenated technique in characterizing determined by the interactive force between the par-
colloidal particles. ticles and the external Reld. In its simplest form the
Figure 3 Fractionation of haematite-I ( -Fe2O3(I) with nominal diameter 0.148 m) and haematite-II ( -Fe2O3(II) with nominal
diameter 0.248 m) colloidal particles by the potential barrier SdFFF technique. The experimental conditions, expect for those given in
the scheme, were as follows: field strength 15.5 g; flow rate 150 cm3 h\1; void volume of the channel 2.06 cm3. Reproduced from
Koliadima A and Karaiskakis G (1990) Potential-barrier field-flow fractionation, a versatile new separation method. Journal of
Chromatography 517: 345, with permission from Elsevier Science.
technique consists of changing the ionic strength of is reasonable to expect serious efforts by the com-
the carrier solution from a high value, where only one panies producing FFF equipments, to improve their
of the colloidal materials of the binary mixture to be resolution and expand the particle size range to lower
separated totally adheres at the beginning of the FFF limits of analysis. As far as FFF applications are
channel wall, to a lower value, where all the adhered concerned, it is believed that future efforts will be
particles are released. focused on the size separation and characterization of
Figure 3 shows the fractionation of two haematite polydisperse, irregular colloidal particles, as well as
samples ( -Fe2O3(I) with nominal particle diameter on the aggregation and deposition phenomena of
0.148 m and -Fe2O3(II) with nominal particle dia- colloids on solid surfaces.
meter 0.248 m) with monodisperse spherical par-
ticles, by the PBSdFFF technique. The mixture was See also: II/Particle Size Separation: Field Flow frac-
introduced into the channel with a carrier solution tionation: Electric Fields; Field Flow Fractionation: Ther-
containing 0.5% v/v detergent FL-70, 0.02% w/w mal; Theory and Instrumentation of Field Flow Fractiona-
NaN3 and 3;10\2 mol L\1 KNO3. At this high elec- tion. III/Cells and Cell Organelles: Field Flow Frac-
trolyte concentration all of the -Fe2O3(II) particles tionation. Colloids: Field Flow Fractionation. Poly-
adhered at the beginning of the SdFFF Hastelloy-C mers: Fixed Flow Fractionation. Proteins: Field Flow
channel wall, whereas all of the -Fe2O3(I) particles Fractionation.
were eluted from the channel. The average diameter
of the eluted -Fe2O3(I) particles was found from eqn
[3] to be 0.151 m, in good agreement with that
Further Reading
obtained by normal SdFFF (0.145 m) or determined Beckett R and Hart BT (1993) Use of Reld-Sow fractiona-
by TEM (0.148 m). Variation of the carrier solution tion techniques to characterize aquatic particles, colloids
to one containing only 0.5% v/v detergent FL-70 and and macromolecules. In: BufSe J and van Leeuwen HP
0.02% w/w NaN3 without any KNO3 released all the (eds) Environmental Particles, vol. 2, pp. 165}205.
adhered -Fe2O3(II) particles and gave a particle dia- Boca Raton, FL: Lewis Publishers.
meter (0.244 m) in good agreement with that found Dondi F and Guiochon G (eds) (1992) Theoretical Ad-
vancements in Chromatography and Related Separation
by normal SdFFF (0.237 m) or obtained by TEM
Techniques, vol. 383. NATO ASI series. Dordrecht:
(0.248 m). Kluwer Academic.
A general methodology for the analysis of a collo- Giddings JC (1966) A new separation concept based on
idal mixture by PB SdFFF consists of injecting into the a coupling of concentration and Sow nonuniformities.
column the mixture with a carrier solution in Separation Science 1: 123}125.
which the ionic strength is too high to ensure total Giddings JC (1991) UniTed Separation Science. New York:
adhesion of all the components of the mixture, except Wiley.
for one with the lower attractive force with the chan- Giddings JC, Martin MN, Moon MH and Barman BN
nel wall. Then a programmed variation (decrease) of (1991) Particles separation and size characterization by
the ionic strength of the carrier solution is applied to sedimentation Reld-Sow fractionation. In: Provder
release, in time, the adherent particles according to J (ed.) Particle Size Distribution II: Assessment and
Characterization, American Chemical Society, Series no.
their size and/or surface characteristics. As the
472, Ch. 13, pp. 198}216. Washington, DC: ACS Sym-
PBSdFFF technique is based on particle}wall interac- posium.
tions, its applications can be extended by using differ- Hiemenz PC (1977) Principles of Colloid and Surface
ent materials, such as stainless steel, TeSon and Chemistry. New York: Marcel Dekker.
polyimide. PBSdFFF is also a convenient and accurate Janca J (1988) Field-Flow Fractionation: Analysis of
method for the concentration and analysis of dilute Macromolecules and Particles. New York: Marcel
colloidal samples. This makes PBSdFFF a highly Dekker.
attractive technique for the characterization of sam- Karaiskakis G and Cazes J (eds) (1997) Journal of Liquid
ples where even particles of the same size but of Chromatography & Related Technologies, Special Issue
different chemical composition are present in low on Field-Flow Fractionation (Vol. 20, Nos 16 & 17).
concentration. New York: Marcel Dekker.
Lloyd PJ (ed.) (1987) Particle Size Analysis 1985. New
York: Wiley.
Future Developments Martin M and Williams PS (1992) Theoretical basis of
Reld-Sow fractionation. In: Dondi F and Guiochon
From the detailed description of the FFF techniques G (eds) Theoretical Advancement in Chromato-
presented above, it is concluded that the most usable graphy and Related Separation Techniques, vol. 383.
FFF methods for particle size analysis of colloids NATO ASI series, pp. 513}580. Dordrecht: Kluwer
are SdFFF and FlFFF. Looking to the future, it Academic.
III / COMPUTER DATABASES FOR TWO-DIMENSIONAL ELECTROPHORESIS 2503
COMPUTER DATABASES FOR

TWO-DIMENSIONAL ELECTROPHORESIS
T. Toda, Tokyo Metropolitan Institute of Gerontology, the gel. The protein proRle on the 2-D gel reports
Tokyo, Japan useful information about qualitative and quantitative
Copyright ^ 2000 Academic Press states of proteins in the specimen. Most computer
databases for 2-D electrophoresis have generally been
constructed on such visualized 2-D gel images.
Two-dimensional (2-D) electrophoresis is an unri-
valled technique of the highest capacity to separate
many thousands of proteins in complex specimens 2-D Gel Image Analysis
such as a crude extract of cells. The development of Computer analysis of the 2-D gel images was de-
computer systems for 2-D gel image analysis has veloped after O’Farrell’s paper because 2-D gel im-
promoted the construction of comprehensive ages were too complicated to be analysed manually.
databases for 2-D electrophoresis. The recent ad- Lipkin and Lemkin of the National Institutes of
vance in the Internet computer network has allowed Health in the USA made their GELLAB system on
us to share useful pieces of information located in a DEC minicomputer in 1980, and independently of
many 2-D electrophoresis databases on the world- this, Anderson et al. constructed their original TY-
wide web (WWW) via the Internet. And a new Reld of CHO system in 1981. Garrels and Franza prepared
research named proteomics has been growing based a prototype software for 2-D gel image analysis on
on 2-D electrophoresis databases. a compact Hewlett Packard desktop computer in
1979, and then developed it into the QUEST system
2-D Electrophoresis for an UNIX-based minicomputer in 1983. The
PDQUEST, which is a commercially available soft-
2-D electrophoresis, established by O’Farrell in 1975, ware package for SUN UNIX workstations, was a re-
is a combined technique of isoelectric focusing (IEF) vision of the QUEST system. Many other software
and sodium dodecyl sulfate (SDS)-polyacrylamide gel packages, such as Melanie II and GELLAB II, have
electrophoresis (PAGE) for achieving the Rnest res- been made commercially available. Those Macin-
olution of proteins in crude specimens. Protein mol- tosh- and Windows-based software packages have
ecules of different isoelectric points (pI) are separated allowed 2-D gel image analysis to be performed easily
at the Rrst stage of IEF, and components of different for database construction.
molecular masses are resolved in the second-dimen-
sional SDS-PAGE.
The IEF in a mobile pH gradient, which is auto- Western Blotting
matically generated by lining up carrier ampholytes in Western blotting is an immunochemical technique for
the order of pI, is performed in the original protocol detecting speciRc proteins. Protein spots separated by
of O’Farrell’s 2-D electrophoresis. However, there 2-D electrophoresis are transferred on to a nitrocel-
is still a problem with the conventional IEF method lulose or polyvinylidene diSuoride (PVDF) membrane
in the drift of the pH gradient toward the by electrotransfer blotting. After blocking, the mem-
cathode. Righetti and colleagues have devised an branes are treated with a speciRc antibody and fol-
improved method of IEF which runs on an immo- lowed by a second enzyme-linked antibody. Localiza-
bilized pH gradient (IPG) to prevent cathodic drift. tion of speciRc protein on the membrane is visualized
The IPG-IEF is adopted in most current protocols of by an enzyme reaction. The results of spot protein
2-D electrophoresis for constructing computer identiRcation by Western blotting are also included in
databases. 2-D electrophoresis databases.
Protein Detection Microsequencing

Proteins separated on a gel plate are detected by The N-terminal peptide sequences of proteins are
autoradiography if they are labelled with a radioiso- directly determined with an automatic peptide
tope. Nonradioactive proteins are visualized by silver sequencer after electrotransfer blotting on to a PVDF
or dye staining depending on the protein amount on membrane, if the N-terminal -amino group is free.
2504 III / COMPUTER DATABASES FOR TWO-DIMENSIONAL ELECTROPHORESIS
Proteins that are blocked at the N-terminus are often Mass Spectrometry
cleaved with endopeptidase. The digests are separ-
ated by reversed-phase high performance liquid Mass spectrometry is another powerful technique
chromatography (HPLC) and then subjected to for identiRcation of proteins separated by 2-D
sequencing. Protein databases on the Internet may electrophoresis. The peptide mass Rngerprinting of
answer a query on the homology of the amino acid endopeptidase digests is extensively used for primary
sequence for identiRcation of the protein if the data- assignment of the protein. The ExPASy Molecular
base includes an entry of the protein itself or family Biology server (http://www.expasy.ch/) offers useful
gene products. The sequence data and the results of proteomic tools such as PeptIdent, which helps us in
homology search are both valuable information for peptide mass Rngerprinting works for protein identi-
constructing 2-D electrophoresis databases. Rcation (Figure 1).
Figure 1 Proteomic analysis tool PetIdent in the ExPASy Molecular Biology server. The URL is http://www.expasy.ch/tools/
peptident.html.
History of 2-D Electrophoresis Anderson and his co-workers in 1981, was also only
Databases for personal use on an ofSine computer. Human
keratinocyte 2-D electrophoresis databases, made
The 2-D electrophoresis database, reported by Lipkin by Celis et al. on a UNIX workstaion with
and Lemkin in 1980, was only for multiple 2-D gel PDQUEST software, were offered to many research
image analyses on a stand-alone minicomputer. In groups commercially, but not made accessible on the
their GELLAB system, a composite gel (CGEL) Internet.
database was made by extracting spot data from Appel and his co-workers constructed their SWISS-
multiple gels and merging them into a representative 2DPAGE database (Figure 2) on the ExPASy molecu-
gel. The primary database consisted of the lists of lar biology server of the Swiss Geneva University
corresponding spots, their associated properties and Hospital in 1993. It was the Rrst regular 2-D elec-
interrelations. The human serum protein 2-D electrophoresis database on the WWW, and it has been
trophoresis database for clinical use, reported by accessible on the Internet.
Figure 2 The SWISS-2DPAGE top page in the ExPASy Molecular Biology server of the Swiss Geneva University Hospital. The URL is
http://www.expasy.ch/ch2d/.
Figure 3 The WORLD-2DPAGE home page for indexing to 2-D PAGE databases and services throughout the world. The URL is
http://www.expasy.ch/ch2d/2d-index.html.
WORLD-2DPAGE home page (Figure 3) present- World-Wide Web for

ed at http://www.expasy.ch/ch2d/2d-index.html in 2-D Electrophoresis Database
the ExPASy server offers a convenient link to many
2-DE databases in the world. The WWW is a database server system which was
The databases listed in Table 1 were indexed initiated in the European Centre for Nuclear Re-
in the WORLD-2DPAGE page for link as of 15 June, search on the basis of Internet protocol technology.
1999. The WWW allows browsers on a computer to
Some of the databases have been built follow- access information stored on remote servers
ing the guidelines for the federated 2-D electro- through the Internet. One of the main features of
phoresis database recommended by Appel et al. Rules WWW documents is the hypertext structure; another
for the database federation are summarized in signiRcant feature is the function of a clickable image
Table 2. map.
Table 1 A partial list of 2-D electrophoresis databases indexed in the WORLD-2DPAGE home page
SWISS-2DPAGE (Geneva University Hospital) http://www.expasy.ch/ch2d/

SIENA-2DPAGE: 2D-PAGE database at Department of Molecular Biology,
University of Siena, Italy http://www.bio-mol.unisi.it/2d/2d.html
2-DE maps at LSB Corp http://WWW.LSBC.COM/2dmaps/patterns.htm
Human and mouse 2-D PAGE databases:
Danish Centre for Human Genome Research http://biobase.dk/cgi-bin/celis
RAT HEART-2DPAGE (German Heart Institute, Berlin) http://gelmatching.inf.fu-berlin.de/&pleiss/2d/
HEART-2DPAGE (German Heart Institute, Berlin) http://userpage.chemie.fu-berlin.de/&pleiss/dhzb.html
HSC-2DPAGE (Heart Science Centre, Harefield Hospital) http://www.harefield.nthames.nhs.uk/nhli/protein/
PDD: Protein Disease Database (NIMH}NCI) http://www-lecb.ncifcrf.gov/PDD/
PPDB: PhosphoProtein DataBase http://www-lecb.ncifcrf.gov/phosphoDB/
Molecular Anatomy Laboratory at Indiana University, Columbus http://iupucbio1.iupui.edu/frankw/molan.htm
Human Colon Carcinoma Protein Database at JPSL,
Ludwig Institute for Cancer Research, Melbourne, Australia http://www.ludwig.edu.au/jpsl/jpslhome.html
TMIG-2DPAGE: Age-related Protein Database
at Tokyo Metropolitan Institute of Gerontology http://proteome.tmig.or.jp/2D/
UCSF 2D PAGE (A375 cell line) http://rafael.ucsf.edu/2DPAGE/home.html
NIMH-NCI, National Institute of Mental Health and National Cancer Institute.
Hypertext Cross-Reference GenBank

Links from the 2-D electrophoresis database to re- (A HREF"`http://www3.ncbi.nlm.nih.gov/
lated databases, such as GenBank DNA database in htbin-post/Entrez/ query?db"n&form"
NCBI Entrez, SWISS-PROT protein database or 6&dopt"g&uid"X75598a TARGET"
SWISS-3DIMAGE three-dimensional structure `}blanka' X75598(/A'
databases, through hypertext cross-references is
a common function of most databases for 2-D elec- Clickable Image Map
trophoresis. The function is achieved with anchor
tags in hypertext mark-up language (HTML) and Clickable image map is a function of CGI installed in
executive scripts for common gateway interface the WWW server software for achieving an active
(CGI). For example, in TMIG-2DPAGE, the data link from a location on an image to a corresponding
entry for human nm-23 carries the following hyper- Rle. All federated 2-D electrophoresis databases on
text cross-reference to the corresponding SWISS- the WWW have 2-D gel images for the clickable
PROT (P15531) and NCBI GenBank (X75598) image map. The content of each protein data entry is
entries respectively. displayed on the monitor screen when the mouse
button is clicked on the corresponding spot on the
SWISS-PROT 2-D gel image.
(A HREF"`http://www.expasy.ch/cgi-bin/
sprot-search-ac?P15531a TARGET"`}blanka' Setting up a 2-D Electrophoresis
P15531(/A' Database on the WWW
Table 2 Guidelines for building a federated 2-DE database
To set up a 2-D electrophoresis database, software
for WWW server, such as Apache httpd, must be
1. Individual entries in the database must be accessible from working on a UNIX or LINUX server computer con-
remote by keyword search nected to the Internet. The function of clickable im-
2. The database must be linked to other databases through
active hypertext cross-references
age map is included in the Apache modules if they are
3. A main index has to be supplied that provides a means of properly installed. To set up a new 2-D electrophor-
querying all databases through one unique entry point esis database, prepare the following four types of Rles
4. Individual protein entries must be accessible through clickable Rrst, according to the Apache Reference Manual,
images which is available in the Apache directory.
5. 2-DE analysis software, been designed for use with federated
2-DE database, must be able to access directly individual 1. A hypertext Rle for the image map top page
entries in any federated 2-DE database
(***.html).
For full details, see Appel RD et al. (1996) in Electrophoresis 17: 2. A 2-D gel image Rle in the GIF format for the
540}546. clickable image map (***.gif).
3. A map Rle for directing the jumping destination home to Rve locally maintained databases (SWISS-
(***.map). PROT, SWISS-2DPAGE, PROSITE, SWISS-3DIM-
4. Text Rles of protein data entries as destinations for AGE and ENZYME). Besides the database service,
links (***.html). the server offers special tools for proteomic analysis,
such as PeptIdent for peptide mass Rngerprinting.
When these Rles are placed in the appropriate di- SWISS-PROT serves as a searchable main index, pre-
rectories in the WWW server, the basal activity of the pared according to rule 3, and has cross-references to
2-D electrophoresis database starts running. the other four databases. SWISS-2DPAGE contains
data on protein identiRed on various 2-D PAGE refer-
ence maps. The hypertext cross-references as of
2-D Electrophoresis Databases on the 15 June 1999 include links to SWISS-PROT, YPD
WWW database of yeast Saccharomyces and the
ECO2DBASE E. coli database. The X-ray crystallog-
SWISS-2DPAGE
raphy Protein Data Bank (PDB), the Mendelian
The SWISS-2DPAGE is a fully implemented 2-DE Inheritance in Man data bank (OMIM), the G-
federated database. The ExPASy WWW server is protein-coupled receptor database (GCRDb) and the
Figure 4 Human keratinocytes 2-D PAGE database in the Danish Centre for Human Genome Research. The URL is http://
biosun.biobase.dk/cgi-bin/make}gel.start?msau00032#0#0#0#0.
database of yeast (Saccharomyces cerevisiae) genes teins identiRed on various reference maps. Databases
coding for proteins (LISTA) are also linked through of human keratinocytes (Figure 4) are served for the
SWISS-PROT. study of skin biology, and those of transitional cell
carcinomas are for the study of bladder cancer.
Human 2-D PAGE Databases of the Danish Centre
For Human Genome Research Heart-2DPAGE
The human 2-D PAGE databases at the University of The Heart-2DPAGE at the German Heart Institute
Aarhus, which were developed for functional genome in Berlin is a human myocardial 2-D protein data-
analysis in health and disease, contain data on pro- base that implements the Rrst four rules of the
Figure 5 TMIG-2DPAGE clickable map of TIG-3 human fibroblast line. The URL is http://proteome.tmig.or.jp/2D/Fibro/
humfb}menu.html.
federated 2-D database. It contains data on the hu- (Multi-Agent Retrieval Vagabond on Information
man heart ventricle and atrium. The data are access- Networks). Although some irrelevant sites may some-
ible both by keyword search on the protein name and times be hit, the search engine is helpful for many
by mouse-clicking on two images of 2-D. Entries researchers to Rnd speciRc 2-D electrophoresis
provide data related to the isoelectric point, the mo- databases of their own interests.
lecular mass and the amino acid sequence of each
spot protein.
Further Reading
PDD Protein Disease Database Anderson NL, Taylor J, Scandore A et al. (1981) The
The PDD Protein Disease Database on the WWW TYCHO system for computer analysis of two-dimen-
server is a part of the National Institute of Mental sional gel electrophoresis patterns. Clinical Chemistry
27(11): 1807}1820.
Health and National Cancer Institute (NIMH-NCI)
Appel RD, Hoogland C, Bairoch A and Hochstrasser DF
Protein}Disease Database Project for correlating dis- (1999) Constructing a 2-D database for the World Wide
eases with proteins observed in serum, cerebrospinal Web. Methods in Molecular Biology 112: 411}416.
Suid, urine and other common human body Suids Celis JE, Gromov P, Stergaard M et al. (1996) Human 2-D
based on biomedical literature. The PDD database PAGE databases for proteome analysis in health and
includes the data of quantitative and qualitative pro- disease: http://biobase.dk/cgi-bin/celis. FEBS Letters
tein variations with disease states, answering ques- 398(2}3): 129}134.
tions on protein patterns found in common body Garrels JI and Franza Jr BRF (1989) The REF52 protein
Suids with respect to disease conditions in the litera- database. Journal of Biological Chemistry 264(9):
ture. 5283}5298.
Hoogland C, Sanchez JC, Tonella L et al. (1998) Current
TMIG-2DPAGE status of the SWISS-2DPAGE database. Nucleic Acids
Research 26(1): 332}333.
TMIG-2DPAGE is a database of human proteins in- Lemkin PF (1997) The 2DWG meta-database of two-di-
volved in the mechanisms of cell ageing. It includes mensional electrophoretic gel images on the Internet.
two clickable image maps for data entries of proteins Electrophoresis 18(15): 2759}2773.
in normal cell ageing, and for those in the disease Lipkin LE and Lemkin PF (1980) Data-base techniques for
state of Werner’s syndrome patients. On the TMIG- multiple two-dimensional polyacrylamide gel elec-
2DPAGE clickable map of TIG-3 human Rbroblast trophoresis analyses. Clinical Chemistry 26(10):
line, quantitative variations of proteins observed in 1403}1412.
the process of replicative cell ageing are demonstrated O’Farrell PH (1975) High resolution two-dimensional elec-
trophoresis of proteins. Journal of Biological Chemistry
with coloured crosses (Figure 5).
250(10): 4007}4021.
As a mouse button is clicked on a protein spot, the Righetti PG, Gianazza E and Bjellqvist B (1983) Modern
corresponding data entry is displayed on the browser. aspects of isoelectric focusing: two-dimensional maps
Each data entry contains information on age-related and immobilized pH gradients. Biochemical and Bio-
protein variation, physicochemical properties, refer- physical Methods 8(2): 89}108.
ences and active links to both SWISS-PROT and Toda T and Kimura N (1998) TMIG-2DPAGE: a new
NCBI Entrez nucleotide sequence databases. concept of two-dimensional gel protein database for
research on aging. Electrophoresis 18(2): 344}348.
Toda T and Ohashi M (1995) Review: Current status and
Keyword Search of perspectives of computer-aided two-dimensional
2-D Electrophoresis Databases densitometry. Journal of Chromatography A 698(1):
41}54.
The 2DHunt on the ExPASY server (http://www. Vesterberg O and Svensson H (1966) Isoelectric fractiona-
expasy.ch/ch2d/2DHunt/) is a convenient search en- tion, analysis, and characterization of ampholytes in
gine for WWW sites of 2-D electrophoresis databases. natural pH gradients. IV. Further studies on the resolv-
The site list for 2DHunt search is periodically created ing power in connection with separation of myoglobins.
by the site retrieval robot supplied from the Marvin Acta Chemica Scandinavica 20(3): 820}834.
CONTINUOUS ION EXCHANGE USING

POWDERED RESINS
See III / POWDERED RESINS: CONTINUOUS ION EXCHANGE
III / COSMETICS AND TOILETRIES: CHROMATOGRAPHY 2511
COPRECIPITATION: TRACE ELEMENTS:

EXTRACTION
See III / TRACE ELEMENTS BY COPRECIPITATION: EXTRACTION
COSMETICS AND TOILETRIES:

CHROMATOGRAPHY
(uncharged substances) which do not dissociate in
M. Carini and R. M. Facino, Istituto Chimico
Farmaceutico Tossicologico, University of
aqueous solution, and ionic (charged substances)
Milan, Italy which dissociate and form ions, one of which be-
comes the actual surface-active agent. Ionic surfac-
This article is reproduced from Encyclopedia of Analyti- tants are classiRed by the nature of their charges in
cal Science, Copyright ^ 1995 Academic Press solution: anionic (negatively charged), cationic (pos-
itively charged), amphoteric (both positively and
Toiletries, used by millions of consumers for the daily negatively charged).
care and hygiene of the body, include several prod-
ucts (mainly rinse products) with different formula-
tive bases: soaps, shampoos, bath foams, toothpastes,
Anionic Surfactants
deodorants. Anionic surfactants constitute about 65% of all sur-
Since surfactants are the basic materials used in factants manufactured and it is not surprising that the
toiletries (in soaps and shampoos they increase the bulk of literature on surfactants deals with the analy-
washing properties of water; in shaving products they sis of these compounds. Table 1 reports the main
act as wetting and foaming agents; in bath oils they types of anionic surfactants used in toiletries.
make the perfume water-soluble; in hair products The quality control of anionics in raw materials
they are used as conditioners), this article will deal and in Rnished products is mainly quantitative and
with the techniques used for the analysis of these several methods that give a total estimate of active
important constituents. ingredients have been developed. The two-phase
Leaving aside the rough, but still used, organoleptic mixed indicator titration and other titrimetric ana-
testing (odour, colour, clarity, opalescence), analyti- lyses, such as direct colorimetric titration and precipi-
cal procedures for quality control may range from tation titration are the simplest procedures, since they
physical evaluations (speciRc gravity, refractive in- are sufRciently reliable, require little and inexpensive
dex, optical rotation, viscosity) to chemical analysis equipment, and can be used for both product devel-
by standard volumetric and gravimetric methods, and opment and quality control applications. The two-
instrumental analysis by chromatographic techniques phase mixed indicator titration is based on the
(thin-layer, gas and liquid chromatography) and extraction of an anionic-indicator or cationic-indi-
spectroscopic techniques (ultraviolet, infrared and cator complex into a nonaqueous solvent (usually
nuclear magnetic resonance spectroscopy and mass chloroform) in equilibrium with an aqueous solution
spectrometry). Chromatographic and spectroscopic of the unknown anionic surfactant or the titrating
methods now Rnd wide application, since toiletries cationic surfactant. The method is not quantitative
and raw materials are complex mixtures, and there for compounds containing fewer than 12 carbon
is a constant need to distinguish subtle structural atoms when chloroform is the organic phase (in this
differences in composition and determine impurities case a mixed organic phase of 2 : 3 (v/v) chloro-
even if present in trace amounts. form}1-nitropropane must be used). Sodium, magne-
The term surfactant (a contraction for surface- sium or sulfate ions at concentrations up to
active agent) is used to describe organic chemicals 0.4 mol L\1 do not interfere, while chloride interferes
that, when added to a liquid, change the inter- above 5;10\3 mol L\1.
facial properties of that liquid. Chemically surfac- All the spectrophotometric procedures are based
tants can be divided into two major classes: nonionic on the formation of a solvent-extractable compound
2512 III / COSMETICS AND TOILETRIES: CHROMATOGRAPHY
Table 1 Anionic surfactants
Type General structure
Alkyl carboxylates (soaps) R}COO\X#

C8}C18 fatty acids; salts with NaOH, KOH, NH4OH, monoethanolamine (MEA),
diethanolamine (DEA), triethanolamine (TEA)
Alkylethoxylated carboxylates R}(OCH2CH2)nOCH2COO\X#
R"C8}C18 fatty alcohols
X"Na, K, NH4, MEA, DEA, TEA
#
Alkyl sulfates R}O}SO\ 3 X
X"Na, K, Mg, NH4, MEA, DEA, TEA
Alkylethoxylated sulfates (AES) R}(OCH2CH2)n}O}SO\ 3 X
#

X"Na, K, Mg, NH4, MEA, DEA, TEA
Alkylarylsulfonates R}C6H5}SO\ 3 X
#
R"C10}C12
#
-Olefin sulfonates R}SO\ 3 Na
R"C14}C16
Isethionates R}CO}O}CH2CH2}SO\ 3 Na
#
R}CO"C12}C18 fatty acids
between the anionic surfactant and a coloured One-dimensional chromatography

cationic species: Methylene blue, Methyl green, Tol- Alumina 60 F254 with isopropanol.
uidine blue, Rosaniline, Brilliant green and Methyl Silica gel 60 with propanol}chloroform}
violet being the most widely used. These cationic methanol}10 mol L\1 ammonia (10 : 10 : 5 : 2); ethyl-
compounds are not extractable as such by organic acetate}acetic acid}methanol}10 mol L\1 ammonia
solvents, but in the presence of anionic species they (45 : 5 : 2.5); ethanol}acetic acid (9 : 1).
form a stable, stoichiometric ion-association complex Silica gel G containing 10% ammonium sulfate
that is poorly soluble in water, because it has no nett with chloroform}methanol}0.05 mol L\1 sulfuric
charge. The complex is extracted into the organic acid (80 : 19 : 1).
solvent and the absorbance gives a direct measure of
the surfactant present. Two-dimensional chromatography
These techniques are not suited for establishing the Silica gel with acetone}tetrahydrofuran (9 : 1)
qualitative composition of these surfactants, e.g. for followed by propanol}chloroform}methanol}
differentiating homologues and oligomers, for char- 10 mol L\1 ammonia (10 : 10 : 5 : 2).
acterization of single components of a surfactant mix-
ture, or for detection and quantiRcation of impurities Several spray reagents are used for detection
(unreacted materials, by-products), and when such and identiRcation: Pinakryptol Yellow, Dragendorff,
determinations are required for Rnished detergent ninhydrin, leucomalachite green and iodine vapour.
formulations. This method can be applied to all classes of tensides
Hence, more speciRc and sensitive chromato- as it distinguishes anionics from nonionics (fatty di-
graphic techniques such as thin-layer chromatogra- ethanolamides and ethoxylates) in shampoos, bath
phy (TLC), gas chromatography (GC) and liquid foams and soaps. The aqueous samples can be freeze-
chromatography (LC) have been developed and are dried and extracted with a suitable solvent
now widely accepted in the surfactant industry. Mass (ethanol}water) to minimize foaming.
spectrometry (MS) and tandem mass spectrometry Basically, anionics are analysed by reversed-phase
(MS-MS), although considered the techniques of LC using an ion-pairing agent and an organic solvent
choice for rapid characterization of surfactant mix- (acetonitrile, methanol, tetrahydrofuran)}water
tures, have not yet gained general acceptance as rou- gradient. In reversed-phase ion pair chromatography,
tine analytical techniques. the addition of an appropriate ion}pairing reagent
TLC, because of its rapidity and low cost, is par- (tetrabutylammonium hydrogensulfate) to the mobile
ticularly useful. Anionics (sulfates, sulfonates, soaps) phase suppresses the ionic nature of the sample while
can easily be separated under the following condi- introducing some charge to the nonpolar surface of
tions (all ratios given are volume ratios). the stationary phase. The retention of the resulting
ion pair is then controlled by pH, counterion con- and sulfates can also be analysed using sodium
centration and mobile phase polarity. Linear and naphthalenedisulfonate}acetonitrile as the mobile
branched-chain (C4}C18) alkylbenzenesulfonates are phase on a polymeric Suorocarbon}amine cross-
separated by this method according to the length and linked weak anion exchange silica column with
structure of the alkyl chain, using as the mobile phase either indirect conductivity or photometric detec-
0.1 mol L\1 tetrabutylammonium hydrogensulfate tion modes.
(pH 5)}water}acetonitrile (gradient elution). At pH The solid-phase reagent (SPR) procedure has been
5, the sulfonates and pairing reagent are completely introduced recently as a new method of postcolumn
ionized, as are the carboxylated surfactants, whose conductivity detection of alkylsulfates and alkyl-
pKa values are somewhat higher (pKa&4). The struc- sulfonates. SPR, an aqueous suspension of sub-
ture and concentration of the pairing agent also micrometre particles of a polymeric cation exchange
inSuences the retention behaviour: increasing the material in the hydrogen form, is pumped into the
lipophilicity increases the retention of the surfactant eluent stream coming from the column (silica-based
ion pair by enhancing its afRnity for the nonpolar reversed-phase). The postcolumn reaction transforms
stationary phase. The increase in the concentration of the tetrabutylammonium alkyl sulfate or sulfonate
the pairing agent will lead to greater coverage of the into the corresponding free acid. This changes the
stationary phase surface and consequently to longer analytes into more conductive species and tet-
sample retention. rabutylammonium borate eluent to the low conduct-
A limitation is that only the chromophoric compo- ivity boric acid. The conductivity detection method
nents of such mixtures can be monitored by ultra- with SPR makes it possible to employ gradient elution
violet (UV) absorption detection. Nonchromophoric for separation of complex mixtures.
anionic surfactants such as alkyl sulfates or their GC and GC-MS cannot be applied directly to the
corresponding alcohols and acids can be determined analysis of anionic surfactants since these compounds
by LC after derivatization to 3,5-dinitrobenzoate are too polar and nonvolatile to be amenable either to
esters or p-(methylthio)benzoate esters. The acidic GC or to conventional electron-impact MS (desul-
forms of -oleRn sulfonates, alkyl sulfates, alkyl- fonation with acids, alkali fusion sulfochlorination,
ethoxylated sulfates and alkyl phosphates react with methylation, pyrolysis-GC are well known methods
4-diazomethyl-N,N-dimethylbenzenesulfonamide to of prederivatization for GC analysis). The use of soft
produce UV-absorbing derivatives that can be ionization techniques such as Reld desorption (FD)
separated by reversed-phase LC. Direct detection (no and fast atom bombardment (FAB) is highly suited
derivatization) can be achieved by the use of a for characterization of these polar compounds by
spectroSuorimetric detector operating at 225 nm (ex- MS.
citation) and 295 nm (emission) (reversed-phase FAB-MS (in positive or negative ion mode) can be
C18 (RP-18) column; mobile phase: 0.1 mol L\1 so- used to analyse complex anionic mixtures without
dium perchlorate in 80 : 20 methanol}water) or by prior separation of the components, since it gives
a differential refractometer (refractive index) de- abundant deprotonated (negative) or cationized (pos-
tector. Long}chain alkane sulfonates (C12}C20) are itive) molecular ions, and no fragmentation. As is
separated by this method using a phenyl column with shown in the case of a mixture of ethoxylated alcohol
a 75% methanol}25% 0.1 mol L\1 sodium nitrate (C12/C14) sulfates (Figure 1), the technique not only
mobile phase. gives the complete pattern of oligomer distribution
An alternative approach to separating and detect- (length of the alkyl and/or of the ethoxylate chain),
ing aliphatic anionic surfactants is ion interaction but also information about the purity of the raw
chromatography (reversed-phase column) with an material (the c and d series in Figure 1B are the
aromatic ion-pairing agent also acting as chromo- cationized molecular ions of unreacted materials, the
phore for UV detection at 254 nm (cetylpyridinium unsulfated ethoxylated fatty alcohols). Ethoxylated
chloride, phenethylammonium ion). alcohol sulfates are easily detected by this technique
Anion exchange chromatography with indirect in Rnished detergent formulations, even in the pres-
detection is not a common approach for aliphatic ence of other surfactant types (i.e. amphoteric ten-
sulfonates, although it is simpler than ion pair sides), as has been demonstrated for shampoos.
chromatography. Using hydrogenphthalate, sulfo-
salicylic acid or m-sulfobenzoic acid in 60 : 40
acetonitrile}water as mobile phase, C2}C8 sulfonates
Nonionic Surfactants
can be separated on a strong cation exchange column Nonionic surfactants constitute the second most im-
with indirect UV absorbance detection at 297, 320 portant class of tensides: although their foaming
and 298 nm, respectively. C6}C12 aliphatic sulfonates properties are low in respect to those of anionics, they
Figure 1 (A) Negative-ion and (B) positive-ion FAB mass spectra of ethoxylated alcohol sulfates (anionic surfactants). (From Maffei
Facino et al., 1989.)
are widely used in detergent formulations (especially acid, molybdophosphoric acid, picric acid, Malachite
in bath foams) to increase viscosity and as foam green, potassium tetracyanatozincate and ammonium
boosters. Table 2 reports the chemical classiRcation tetrathiocyanatocobaltate (III). This last reagent
of the most important nonionic surfactants. gives better results than tungstophosphoric acid and
Spectrophotometric methods for the quantitative than the potentiometric method with Dragendorff’s
analysis of nonionic surfactants are popular: they are reagent. It can be applied for determination of
based on complex formation with tungstophosphoric ethoxylate compounds in detergent solutions: the sur-
Table 2 Nonionic surfactants
Ethoxylated alcohols R}O(CH2CH2O)nH

Reaction product between ethylene oxide and fatty alcohols
Ethoxylated alkylphenols R}C6H4}O(CH2CH2O)nH
Reaction product between ethylene oxide and alkylphenol (R"C8/C9)
Alkanolamides R}CO}N}(CH2CH2OH)n n"1,2
Reaction product of fatty acids (C10}C18) with mono- or diethanolamine
Alkylglycosides (glucose ethers) R}O}(Gluc)n
Reaction product between glucose and fatty alcohols (R"C8}C14)
factant is extracted from the aqueous solution with The NMR approach involves the calculation of
chloroform and the extract is treated with the reagent absolute integrals of three types of hydrogen atoms:
to form a blue complex that is then analysed spectro- from these integration values it is possible to calculate
photometrically. The absorbance is not affected by the degree of condensation and the alkyl residue com-
temperature, electrolytes or dilution. position, as each surfactant molecule contains four
Several electrochemical and potentiometric tech- aromatic hydrogen atoms that can be used as an
niques are also available for quantitative analysis of internal reference to determine the number of hydro-
nonionic tensides: for example, using a barium ion gen atoms corresponding to the remainder of the
selective electrode it is possible directly to quantitate peaks.
the surfactant in the range of 2;10\5 to 1; TLC with a Same ionization detector (FID) has
10\3 mol L\1. A typical potentiometric titration is been applied for the separation and quantitative de-
based on formation of insoluble complexes with mo- termination of nonionic surfactants containing an
lybdophosphoric acid: the sample dissolved in aque- average number of oxyethylene units not higher than
ous ethanol containing barium chloride is treated 8.0. The oligomers are separated on Chromarod S-II
with an excess of the complexant and the unreacted (silica gel-coated rods) with double development;
molybdophosphoric acid is titrated potentiometri- (a) benzene}ethyl acetate (6 : 4) for 10 cm from the
cally with diantipyrylmethane, using a platinum indi- start; (b) ethyl acetate}acetic acid}water (8 : 1 : 1) up
cator cathode. to a distance of 8 cm. After development, the rods are
In the quality control of ethoxylated compounds, it passed through a FID operating with hydrogen
is important to determine their composition, since (160 mL min\1) and air (2 L min\1). The major ad-
they are manufactured as mixtures of homologous vantage of LC in the analysis of nonionic surfactants
compounds that differ in the length of the ethoxylate lies in its ability to separate and quantitate alcohol or
and/or the alkyl chain. Several spectroscopic nuclear alkylphenol ethoxylate oligomers that differ in the
magnetic resonance (NMR) and infrared (IR) and length of the ethoxylate chain. While alkylphenol
chromatographic techniques (TLC, GC, LC) and ethoxylates can readily be identiRed by UV detection,
supercritical Suid chromatography (SFC) are avail- aliphatic compounds, since they do not posses signiR-
able to determine the degree of ethoxylation and the cant UV absorption, must be derivatized prior to LC
distribution of homologues in nonionic surfactant (for example by esterRcation with 3,5-dinitrobenzoyl
mixtures. chloride). Reversed-phase LC with refractive index
Both IR spectroscopy and NMR spectroscopy may detection has been proposed to establish the retention
be used for the determination of the average molecu- behaviour of a wide range of ethoxylated and/or
lar mass (Mr), the average degree of condensation (x), propoxylated adducts: there is a linear relationship
and the hydrophilic/lipophilic balance (HLB) of between the logarithm of the capacity factor and the
nonylphenol ethylene oxide condensates. The IR degree of polymerization of the ethoxylated and/or
method is based on the regression between the logar- propoxylated C12, C16, C18 alcohols, ethylene ox-
ithm of the surfactant properties and the logarithm of ide}propylene oxide copolymers, poly(ethylene
the ratio of the heights of the bands at 840 and glycol)s and poly(propylene glycol)s. This can be used
960 cm\1, corresponding to aromatic C}H vibra- not only for the prediction of chromatographic separ-
tions: the Rrst band is greater than the second for ation, but also for the estimation of the degree of
compounds of smaller degree of condensation, but polymerization and of the length of the alkyl chain.
this relationship becomes inverted as the degree of Recently, the evaporative light-scattering (ELS)
polymerization increases. detector, also known as the mass detector, was
introduced as a universal detector for separation and

quantiRcation of all surfactant species. The detector
measures light refracted by the nonvolatile particles
after the efSuent from the LC is nebulized and the
carrier solvent is evaporated. The detector gives an
equal and linear response factor for each class of
surfactant that is independent of molecular mass (the
amount of refracted light is proportional to the con-
centration of the analyte species).
Alcohol ethoxylates have been characterized by
GC as acetate derivatives on a packed column or as
silylated derivatives using a fused silica capillary col-
umn. However GC gives only a partial Rngerprint,
since only low molecular mass components can
be detected (the free alcohols and short-chain
ethoxylated homologues, up to approximately 12
ethylene oxide oligomers).
High temperature capillary gas chromatography
and SFC are new alternative procedures for the analy-
sis of these compounds that are thermally unstable or
have low volatility. The alcohol and ethoxylate distri-
butions, mean molecular mass and average number of
moles of ethylene oxide can be calculated rapidly
with both the methods (polyglycols with average mo-
lecular masses of 2000}2500 Da have been success-
fully analysed by SFC). Advantages and limitations of
the SFC and high temperature capillary GC proced-
ures can be summarized as follows: (1) for routine
quality control analyses of known alcohol ethoxy-
lates, both techniques appear to be equally suited;
(2) SFC is time-saving because derivatization is not
required, although for complex mixtures derivatiz-
ation improves resolution (acetylation by means of
acetic anhydride and pyridine or silylation with
bis(trimethylsilyl)triSuoroacetamide and pyridine);
(3) the GC technique is able to resolve C12}C18 alco-
hol ethoxylate oligomers, thus avoiding ambiguous
identiRcation of components. Figure 2 shows the
chromatographic proRles of a C12/C13 alcohol eth-
Figure 2 Analysis of ethoxylated alcohols (nonionic surfac-
oxylate with an average of 6.6 moles of ethylene tants) by GC, high temperature GC and SFC. (A) Capillary GC of
oxide obtained by SFC and by high temperature cap- a silylated C12/C13 alcohol ethoxylate with an average of 6.6 moles
illary GC after silylation. of ethylene oxide. (B) Capillary SFC of the same mixture (un-
Among the mass spectrometric methods, the use of derivatized). (C) Capillary high temperature GC of the same
mixture (after silylation). (From Silver AH and Kalinoski HT (1992)
conventional GC-MS electron impact (EI) ionization
Journal of the American Oil Chemist’s Society 67: 599}608.)
is limited to nonionic surfactants with a low degree of
ethoxylation. Compounds with a high degree of
ethoxylation (20}25 units) cab be identiRed directly negative ion mode) gives direct characterization of
in raw materials and in Rnished detergent formula- alkylpolyglycosides, a new generation of highly polar
tions by soft ionization techniques such as direct nonionic tensides not amenable to analysis by con-
chemical ionization (DCI), FD and FAB. This last, in ventional chromatographic methods. The method,
positive ion mode, furnishes the complete pattern of which is based on unambiguous molecular mass de-
oligomer distribution of ethoxylated compounds, termination of the single components (protonated or
since it gives a series of ions at 44-Da intervals (proto- deprotonated molecular ions), allows the deRnition of
nated and/or cationized molecular ions only, with no length of both the alkyl and the glucosidic chains (up
fragmentation). In addition FAB-MS (in positive or to 10 glucose units).
Free poly(ethylene glycol)s (PEGs) are the main rats and mice. 1,4-Dioxane is formed by dimerization
contaminants of ethoxylated derivatives and are fre- of ethylene oxide during the process of alcohol/phen-
quently found in the products obtained from them, ol ethoxylation and might be found in the Rnal deter-
because they can be formed as side-products in the gent formulations via the use of ethoxylated fatty
reaction of ethylene oxide with the hydrophobic com- alcohol sulfates as cleansing agents.
ponent (in which case they are present as a mixture of 1,4-Dioxane is commonly determined by GC.
homologous polymeric derivatives with a molecular A simple method applied to shampoos, requiring
mass distribution that depends on the reaction condi- minimal sample preparation (dilution with water
tions); they can be added intentionally to obtain spe- containing the internal standard isobutanol and
ciRc performances of the Rnal product; and they can direct injection), is carried out with packed column
arise from the decomposition of adducts in the syn- (15% OV-1 on 100/120 Chromosorb WHP) and FID
thetic reaction or during the processing. Hence deter- (injection temperature 1853C; detector temperature
mination of free PEGs is important not only from the 3253C; temperature programme 853C (2 min),
viewpoint of routine quality control of the manufac- 53C min\1 to 953C, followed by clean-up step). Lin-
turing process, but also for the determination of the earity is in the concentration range 1}250 mg kg\1;
suitability of surfactants for speciRc purposes. Among limit of detection 1 mg kg\1.
the procedures used for the separation of PEGs from An alternative technique than can be applied to
adducts and the unreacted starting material, a simple different cosmetic matrices is GC-MS with selected
method involves extraction of an ethyl acetate solu- ion monitoring (SIM); prior to analysis, rapid and
tion of surfactants with 5 mol L\1 sodium chloride, efRcient puriRcation from the interfering materials of
followed by extraction of the aqueous phase with the cosmetic products is achieved by use of combined
chloroform, evaporation of the solvent and gravimet- silica/octadecylsilica cartridges (limit of detection
ric determination of PEGs (accurate temperature con- 3 mg kg\1).
trol is required). Scheme 1 shows a procedure useful for separation
Column LC is faster and more reproducible for the and quantitation of a hypothetical detergent product
separation of free PEGs from the other components of (liquid or powdered) formulated with different types
the mixture. Silica, hydrophobized with dichloro- of active ingredients: amine oxide, ethoxylated alco-
dimethylsilane, with chlorobenzene as the stationary hols (nonionic), alcohol sulfate (AS), ethoxylated
phase, separates PEGs from their adducts using alcohol sulfates (AES) and linear alkyl sulfonates
acetone}water}acetic acid as the mobile phase. Utiliz- (LAS). Under basic and neutral conditions, amine
ing reversed-phase chromatography on silanized oxide behaves as a nonionic material, while under
silica gel, and 30% aqueous isopropanol as mobile acidic conditions it acts as a cationic agent.
phase, PEGs are eluted, while adducts are desorbed A sample (&5 g) of liquid detergent or of the
with 96% ethanol. Partition chromatography, with alcohol}soluble material (1}2 g) from a powdered
ethyl acetate as the mobile phase and cellulose as detergent is dissolved in a minimum volume of
support for the stationary phase (30% sodium chlor- ethanol}water (1 : 1) and passed through a strong
ide solution), is used for the determination of PEGs in cationic ion exchange column (Dowex 50WX4,
adducts of fatty alcohols, alkylphenols, fatty acids 200}400 mesh, sulfonic acid form). Elution with
and alkanolamides. ethanol}water (1 : 1) separates anionic and nonionic
By applying hexane}isopropanol}water mobile surfactants from amine oxide selectively absorbed on
phases of controlled composition (different ratios of the resin. The amine oxide is eluted from the column
hexane to isopropanol), either ethylene oxide adduct with 1 mol L\1 ethanolic hydrochloric acid and the
(EOA) or PEG oligomers can be separated on a eluate, after neutralization, is extracted with carbon
bonded diol phase, and their distributions evaluated tetrachloride. The isolated amine oxide fraction can
(refractive index detection). The PEG or EOA be further characterized by NMR, IR or GC and
oligomers can easily be separated up to the 30-mer quantiRed by a titration method: under acidic condi-
even without gradient elution, and ethoxylated sur- tions amine oxides are determined as quaternaries
factants (fatty alcohols, fatty acids, fatty acid mono- with a standard alkylbenzene sulfonate and methyl-
ethanolamides and alkylphenols) up to an ethoxyla- ene blue indicator (see Cationic Surfactants). This
tion degree of 20. method does not distinguish amine oxides from their
Another important contaminant of ethoxylated de- precursor alkyldimethylamines: the latter can be ana-
rivatives (both anionic and nonionic) is 1,4-dioxane: lysed by gas liquid chromatography (GLC).
according to the European Economic Community If the amine oxide distribution and average mo-
Directive on Cosmetics, commercial products must lecular mass are unknown, they can be determined by
be free from this compound, since it is carcinogenic in packed column (Apiezon L on 60/80 Chromosorb W.
and then hydrolysed with 1 mol L\1 sulfuric acid.

Hydrolysis converts all the sulfated anionic material
to ethoxylated alcohols or fatty alcohols (the sul-
fonated anionic fraction is not affected by acid
hydrolysis), which, after neutralization, can be re-
covered by carbon tetrachloride extraction.
The remaining ethanol}water phase containing
sulfonated species is evaporated and the sulfonate
is recovered and weighed by a salting out pro-
cedure; alternatively, it can be qualitatively and
quantitatively analysed by the methods previously
described.
Cationic Surfactants
Cationic surfactants are devoid of detergent or foam-
ing properties, but are excellent hair conditioners: for
these reasons their use in toiletries is limited to formu-
lations of speciRc shampoos. Table 3 shows the main
types of cationic surfactants used in cosmetics.
The ion pair extraction technique has proved suited
for the determination of cationic surfactants by two-
phase titrations and/or by spectrophotometry (the
method is based on extraction of an ion pair between
surfactant and dye, which is the basis of the well
known Epton Methylene blue and The Cosmetic,
Toiletry and Fragrance Association (CTFA) mixed
indicator method). In the two-phase titration of
cationics by lauryl sulfate in the presence of a suitable
indicator dye (Methylene blue, Thymol blue, Bromo-
phenol green, disulRne blue}dimidium bromide), the
dye}surfactant ion pair is extracted almost com-
pletely by the organic solvent chloroform or methyl-
Scheme 1 Separation of different types of surfactants. AS" ene chloride.
alcohol sulfates; AES"ethoxylated alcohol sulfates; LAS"lin- When the titrant (an oppositely charged surfactant)
ear alkyl sulfonates. is added, surfactant}surfactant ion pair formation
takes place. The end point is indicated when enough
titrant is added so that the small amount of the dye
HMDS) GC: these compounds pyrolyse to 1-oleRns present is displaced from the dye}surfactant ion pair
(column temperature 2803C; injection temperature and returns to the aqueous phase. Alternatively, the
2403C; detector temperature 3303C) and pyrolysis is dye}surfactant ion pair can be spectrophotometri-
essentially complete over the range of C12 to C18 alkyl cally determined after chloroform extraction from the
chains. Alkyldimethylamines do not decompose in aqueous solution.
these conditions and their peaks are well separated Using Bromophenol blue as dye indicator, it is
from oleRns: therefore determination of the precur- possible to quantitate cationic surfactants and the
sors should be possible by the use of a suitable inter- corresponding amines when both are present in
nal standard. a detergent mixture. It has been shown that with
The aqueous alcohol phase containing free non- long-chain quaternary ammonium compounds (cetyl-
ionic ethoxylated alcohols, sulfated anionic and sul- trimethylammonium bromide), Bromophenol blue
fonated anionic material is extracted with carbon forms two different compounds: in alkaline solutions
tetrachloride to separate nonionic surfactants, which a blue di(cetyltrimethylammonium) salt, but in acid
can then be analysed according to one of the methods solution a yellow mono(cetyltrimethylammonium)
mentioned previously. The aqueous alcohol residue salt.
containing only sulfated and sulfonated anionic ma- Hence cationics can be estimated spectrophotomet-
terials is concentrated in vacuo to remove the alcohol rically in two different ways, as the blue di-salt in
Table 3 Cationic surfactants
Alkyltrimethylammonium halides
Alkylethoxylated ammonium halides
Dialkyldimethylammonium halides/saccharinates
Alkylbenzyldimethylammonium halides
Alkylpyridinium halides
Alkylisoquinolinium halides/saccharinates
alkaline solutions (absorbance maximum at 606 nm) termination of cationic surfactants with Orange II as
and as the yellow mono-salt in acid solutions (absorb- dye indicator has the same kind of applications: dye
ance maximum at 416 nm). salts are determined at 490 nm after chloroform ex-
Separate estimations of the quaternaries (which do traction from aqueous solutions of surfactants and
not hydrolyse) and the amine salts (which can hydro- excess of Orange II dye: Orange II reacts with a 1 : 1
lyse easily in alkaline solutions) can be carried out stoichiometry with cationic tensides and the molar
working at different pH values: a higher pH will absorptivity and the wavelength of maximum absorb-
decrease the contribution of the amine even more ance for the dye salts in chloroform are independent
because the higher the pH, the greater the hydrolysis of the reacting surfactant. Isolation of dye salt in
of the amine salts into the amine and removal by the chloroform can be also used as a means of estimating
organic solvent. Determination of the amine salts in average equivalent weights of commercial cationic
the presence of cationics can be carried out by estima- surfactants.
ting the total cationics in acid solution and the quat- By selective changing of the pH, the method might
ernaries only in alkaline solution: the amine content is be applied for quantiRcation of cationic precursors
obtained by difference. The spectrophotometric de- such as amines and amine oxides of amphoteric sur-
factants and of mixtures of amine and quaternary dimethylammonium and alkylpyridinium halides
ammonium compounds. with C10}C18 alkyl groups can be separated by em-
The prerequisite for the extraction method is the ploying porous microspherical poly(styrene}divinyl-
formation of a lipophilic surfactant}dye ion pair benzene) gel as the stationary phase and 0.5 mol L\1
which is then extracted into chloroform or methylene perchloric acid in methanol as the mobile phase (the
chloride. However, there are many cationic polymers logarithm of the capacity factor for each homologous
used as hair conditioners that do not form lipophilic series is directly proportional to the alkyl chain
ion pairs, such as cationic polypeptides, and in addi- length). Figure 3 shows the reversed-phase liquid
tion many surfactants form an emulsion during chromatograms of a homologous series (C12}C18) of
extraction with lipophilic solvents, causing problems n-alkylbenzyldimethylammonium chlorides obtained
in determining the end point. In all these cases, quant- under different experimental conditions.
itative analysis of cationic surfactants can be The mass spectrometric soft ionization techniques
performed by a potentiometric method using a (FD, FAB) allow a rapid and unequivocal structure
‘surfactant’ electrode and sodium dodecyl sulfate as elucidation of the components of a mixture of
titrant. cationic surfactants. FAB in the positive ion mode
Quaternary ammonium salts are not amenable as gives unambiguous spectra, with abundant molecular
such to GC because of their low volatility and limited ions and no fragmentation, furnishing detailed in-
thermal stability. Long chain quaternary ammonium formation on the length of both the alkyl and the
compounds undergo extensive but reproducible ethoxylate chains in polyethoxylated derivatives
decomposition in a classical gas chromatographic (these last compounds are frequently used as hair
system, to tertiary amines and alkyl halides. This conditioners).
chromatographic behaviour has led to the develop-
ment of an analytical approach carried out with dedi-
cated instruments such as a Curie point pyrolyser or
Amphoteric Surfactants
a Rlament pyrolyser coupled to a GC-MS system, By deRnition, amphoterics are surfactants that have
which has been applied both for structure elucidation anionic or cationic properties depending on the pH
(distribution of homologous compounds) and for and that have an isoelectric point. Because of the
quantitative determination of cationic surfactants in highly nucleophilic character of oxygen, amine ox-
various matrices. ides also have salt formation potential, and for this
Long chain N-alkylpyridinium (alkyl"C10}C18) reason their analysis is similar to that of amphoterics.
salts can be determined by GLC of the reduction Table 4 shows the main amphoteric types produc-
products obtained by treatment with sodium ed today: alkylamido and alkyl betaines (and their
tetrahydroborate and nickel(II) chloride. The pro- respective amine oxides), alkylamido- and alkylsul-
cedure is useful for routine analysis of N-alkyl- fobetaines, amphoglycinates (formerly imidazolines).
pyridinium salts, as the reduction to perhydrogenated Among the surfactants, amphoterics are those
derivatives takes place quantitatively and cleanly in more prone to contamination from intermediates,
aqueous media at room temperature with easily han- since their synthesis involves several reaction steps.
dled reagents. Alkylamidobetaines are synthesized from the inter-
The most promising and convenient approach is mediate amidoamines, which in turn are obtained
LC, although its application is limited to UV-absorb- by reaction of fatty acids with amines (mainly
ing quaternaries (quantiRcation of both UV- and non- dimethylaminopropylamine); the amidoamines react
UV-absorbing quaternaries can be achieved with with sodium monochloroacetate in aqueous solution
a LC system coupled to a conductivity detector). Both (alkaline medium) to give betaine derivatives
normal-phase ion pair LC and reversed-phase LC (eqn [I]):
have been used for analysis of cationics: reversed-
phase chromatography is common but problematic, R}COOH#H2N}C3H6}N(CH3)2P
since these compounds mostly elute from octadecyl-
R}CONH}C3H6}N(CH3)2
silica columns as badly tailing peaks.
The addition of ion-pairing agents and/or quater- #Cl}CH2}COO\Na#Pbetaines [I]
nary amines to the mobile phase generally does not
eliminate this unwanted phenomenon. The substitu- The corresponding amine oxides are prepared by
tion of an octadecylsilica by a polymeric polysty- oxidation of the intermediates amidoamines with
rene}divinylbenzene column was found to afford hydrogen peroxide in aqueous solution.
a considerable improvement in the peak shapes. In a similar way, alkylbetaines are prepared by
For example, a homologous series of alkylbenzyl- carboxylation (with sodium chloroacetate) of the
Figure 3 Reversed-phase LC peaks of a homologous series of n-alkylbenzyldimethylammonium chlorides (cationic surfactants). LC

conditions: all mobile phases contain 0.1 mol L\1 sodium perchlorate (pH 3); (A) acetonitrile}water (9 : 1); (B) acetonitrile}water (1 : 1);
(C) acetonitrile}water (7 : 3); (D) methanol}water (9 : 1); (E) methanol}water (3 : 2); (F) methanol}water (17 : 3); (G) THF}water
(3 : 2); (H) THF}water (1 : 1); (I) THF}water (3 : 2) (THF"tetrahydrofuran). Stationary phases: octadecylsilica (A, D, G); cyanopropyl-
silica (B, E, H); phenylpropylsilica (C, F, I). (From Abidi SL (1985) Journal of Chromatography 324: 209}230.)
alkyldimethylamines according to eqn [II]: Sulfobetaines are synthesized according to eqn [III]:
R}COOH#H2N}C3H6}N(CH3)2P
R}COOH#NH3 P R}C,N R}CO}NH}C3H6}N(CH3)2#Cl}CH2}CH"CH2P
\2H2O
[R}CO}NH}C3H6}N(CH3)2}CH2}CH"CH2]#Cl\
P R}CH2}NH2 P R}CH2}N(CH3)2 [II]
#2H2 #CH3X #NaHSO3Psulfobetaine [III]
Table 4 Amphoteric surfactants
Alkylbetaines R}N#(CH3)2}CH2COO\ R"C12}C18

Alkylamine oxides R}N(CH3)2PO
Alkylamidobetaines R}CO}NH}(CH2)n}N#(CH3)2}CH2}COO\ R}CO"fatty acids
Alkylamidoamine oxides R}CO}NH}(CH2)n}N(CH3)2PO
Amphoglycinates R}CO"fatty acids
Sulfobetaines R}CO}NH}(CH2)n}N#(CH3)2}C3H6}SO\
3 R}CO"fatty acids
The position of the sulfo group is not certain, and the The alkyl distribution in alkylamidobetaines,
resulting amphoteric surfactants are thought to be amidoamine oxides and amphoglycinates can be
a mixture of the 2- and 3-sulfopropylated quaternary evaluated by GC, while alkylbetaines, sulfobetaines
ammonium compounds. or amine oxides are preferably analysed by LC. The
Amphoglycinates are prepared by reaction of fatty determination of alkyl distribution in amide deriva-
acids with aminoethylethanolamine to give inter- tives is carried out after hydrolysis (with concentrated
mediate cyclic compounds, imidazolines (it is com- hydrochloric acid) of the amide bonds: the free fatty
mon knowledge that this Rrst step does not produce acids are then converted into the corresponding
the linear amides). By carboxylation with sodium methyl esters by derivatization with conventional
chloroacetate in aqueous solution, ring opening methods (such as methanol}sulfuric acid).
occurs with formation of amphoglycinates that are Reversed-phase LC (RP-18 column) is successfully
not of uniform composition. used for characterization of all amphoterics, both in
Isoelectric point determination is the Rrst measure raw materials and in cosmetic formulations, using
to identify amphoteric surfactants: this can be carried methanol}water (80 : 20) (Figure 4) or methanol}
out by conductivity titration, potentiometric titration aqueous sodium hypochlorite as mobile phase.
or electrophoresis (isoelectric focusing). By poten- Alkylamido products (including alkylamidoamine
tiometric titration, the following isoelectric points oxides) can be detected by absorbance at 215 nm,
have been determined: alkylamidobetaines &7.0; while for alkyl distribution in alkylamine compounds
alkylbetaines &6.0; alkylamidoamine oxides &8.5; a refractive index detector is recommended.
alkylamine oxides &9.0. Direct analysis of amphoterics (sulfobetaines) in
TLC on silica gel plates with chloroform}meth- combination with coconut and tallow soaps can be
anol}ammonia (30 : 50 : 2) or ethanol}chloroform} carried out by reversed-phase LC (detection by differ-
ammonia (45 : 40 : 15) as mobile phases rapidly dis- ential refractometry) using methanol}water (85 : 15)
tinguishes and identiRes different amphoterics, even as the mobile phase containing 0.2% (v/v) acetic acid
when present in detergent formulations; detection is (pH&4). At this pH value, tallow and coconut soap
with 0.1% Bromophenol blue followed by treatment mixtures are analysed as fatty acids and are easily
with 0.1% sodium periodate in aqueous solution. separated from the sulfobetaine components.
IR spectroscopy is an alternative method for identi- Ionic and amphoteric surfactants can also be separ-
Rcation of amphoterics: in the case of amino oxides, ated on reversed-phase columns with 2-naphthalene-
special precautions must be taken during preparation sulfonic acid as counterion in the mobile phase (aque-
of the samples (freeze-drying and not drying at 1053C ous methanol) and detected by UV absorbance and
must be used to remove the water, otherwise the N}O differential refractometry. The simultaneous use of
bond will be broken). The typical N}O bands are at both UV and refractive index detectors allows ion
approximately 960 and 930 cm\1 for both alkyl- pairing (ionic) and nonpairing (amphoteric) compo-
amido and alkylamine oxides; for the alkylamido nents in a mixture to be distinguished.
derivatives, the secondary amide bands at c. 3300, In the case of sulfobetaines, it is possible to separ-
1640 and 1550 cm\1 are diagnostic; the character- ate the Rnal product from reagents and intermediates:
istic bands of betaines are at 1605, 1402, 1340 cm\1 neither amphoterics nor long-chain fatty acids form
(carboxylate bands), 890 cm\1 (quaternary N band), ion pairs with the counterion in the mobile phase and
and 3275, 1633 and 1549 cm\1 (secondary amide they are detected by differential refractometry only.
bands, only present in alkylamidobetaines). The intermediates amidoamine and long chain allyl
Unlike alkyl dimethylamines, whose presence in

alkylbetaines and alkylamino oxides is undesirable,
residual levels of long chain amidoamines in alkyl-
amidobetaines are ‘cosmetically’ acceptable (they in-
crease viscosity and have foam booster properties).
Titration methods for the determination of amine
residues are primarily used for alkylbetaines or
alkylamine oxides. In the method of Metcalfe, the
total amine content is determined by titration (in
isopropanol as solvent) with hydrochloric acid in
isopropanol. After quaternarization of the residual
tertiary amines with methyl iodide at 503C, and
repeated titration with acid, the total amine oxide
content is determined; the tertiary amine content can
be determined by subtraction (limit of detection ap-
proximately 0.5%).
LC furnishes a more selective and sensitive deter-
mination of these amine residues in all classes of
amphoterics. The separation is carried out on a
reversed-phase column (C18) with hexane}isop-
ropanol (60 : 40) as mobile phase containing
2 mmol L\1 octanosulfonic acid (for the ionic pairing
of the amines); detection is by UV absorbance at
215 nm for amidoamines and refractive index for
alkylamines.
Alternatively, postcolumn detection can be em-
ployed for primary, secondary and tertiary amines,
but not for quaternaries: the compounds separated by
the LC column are Rrst converted into the corre-
sponding N-chloramines with hypochlorite; the N-
chloramines are then treated with iodide to form
Figure 4 Reversed-phase LC peaks of (A) tallow-derived sul- triiodide, which can be monitored by its absorbance
fobetaines and (B) coconut oil-derived sulfobetaines (moblie at 355 nm.
phase methanol}water, 80:20; refractive index detection). (From: Chloroacetic acid residues can be evaluated by tit-
Parris N et al. (1977) Analytical Chemistry 49: 2228}2231.)
ration or by chromatographic methods. In the titra-
tion method, the Rrst step involves estimation of total
chlorine content by silver nitrate, after sodium hy-
quaternary ammonium chloride are detected as ion droxide hydrolysis of the sample (2 h under reSux:
pairs by both UV absorbance and refractive index under these conditions chloroacetic acid is hydrolysed
detection. Detection of ionizable surfactants as to glycolic acid and chloride). The titration of an
UV-absorbing ion pairs improves detection limits unsaponiRed sample gives the chloride content.
100-fold over those obtained by differential refracto- The amount of ‘organic chloride’ corresponding to
metry. chloroacetic acid is obtained by subtracting the chlor-
Amphoterics are frequently contaminated with ide content from the total chlorine. The detection
various by-products: free fatty acids; free amines limit of the method lies at 0.03% organic chloride,
(long chain amines and long chain amidoamines); and equivalent to 0.08% (800 mg kg\1) chloroacetic
free chloroacetic acid. The determination of free fatty acid. Using ion chromatography with a conductivity
acids is limited to amide products, and especially to detector (amino exchange column with a hydroxide
alkylamidobetaines (in this case the presence of resid- gradient elution), the limit of detection reduces to
ual amounts of free fatty acids is not a drawback, approximately 20 mg kg\1.
since these compounds positively affect the viscosity The low volatility of alkylbetaines hampers the use
characteristics of the alkylamidobetaine in combina- of conventional EI and chemical ionization (CI) MS
tion with anionics). Fatty acids can be determined as for structure determination. The pyrolytic behaviour
methyl esters by GC after extraction with diethyl of this class of compounds has been studied under
ether of the dried product. EI conditions: the most important pyrolytic process
is the intermolecular isomerization to tertiary An alternative approach to structure determination

aminoesters (CH3)2N}CH2}COOCH3. Although is FD-MS, which gives as prominent ions the proto-
pyrolysis EI spectra are useful for structure con- nated species [M#H]#; intermolecular alkyl
Rrmation of pure compounds, they have limited or transfer also occurs during Reld desorption, resulting
no utility for the analysis of mixtures of constituents in mass spectra containing structurally diagnostic
of unknown chain length, since the spectra are adduct ions (methyl, ethyl, propyl groups linked to
dominated by the ions generated by C}N cleavage nitrogen readily undergo intermolecular transfer to
and the intensities of the molecular ions of the esters give [M#CH3]#, [M#C2H5]# and [M#C3H7]#).
are low. The presence in the mass spectra of several other
Figure 5 (A) Positive ion and (B) negative ion FAB mass spectra of cocamidopropylbetaine (amphoteric surfactants). (From Maffei
Facino et al., 1989.)
adduct and fragment ions (whose relative intensities a range of individual AE concentration from 60 ppt
strictly depend on emitter current) complicates the to 2.17 ppb. The method is able to distinguish
analysis of complex mixtures by this technique. highly branched propylene-based alcohol ethoxy-
FAB-MS in the positive or negative ion mode is lates from isomeric linear ethylene-based alcohol
more promising, since gives not only an immediate ethoxylates.
Rngerprint of the alkyl distribution in a mixture of Positive ion atmospheric pressure chemical ioniz-
amphoteric surfactants, but also direct information ation mass spectrometry (APCI-MS) has been suc-
on the presence of contaminants. cessfully applied to the determination of the oligomer
As is shown with a commercial sample of co- distribution of alkylphenol polyethoxylates and
camidopropylbetaine, in the positive ion mode (Fig- fatty alcohol polyethoxylates. Positive ion and nega-
ure 5A) the protonated (a series) and cationized (b tive ion API-MS techniques have been used for qual-
series) molecular ions of the propylamidobetaine de- ity control of the individual steps of the manufactur-
rivatives of coconut fatty acids (C12}C18) can easily be ing process (intermediates and Rnal products) of new
detected; the mass spectra also contain a few ions (at classes of anionic surfactants, the alkylpolyglucoside
m/z 238, 240, 268, 296, 322, 324) that are due to esters of sulfosuccinic, citric and tartaric acid. With
fragmentation reactions (loss of a dimethylamino- both techniques, the complex mixtures can be injec-
acetic acid residue and loss of the carboxylic group) ted directly into the ion source without prior
and some ions that, as determined by MS-MS (parent chromatographic separation, and the constituents are
scan mode) correspond to the protonated molecular identiRed on the basis of quasi-molecular ions:
species of the dimethylaminopropylamide derivatives cationized ions or solute}solute cluster ions in posit-
[R}CO}NH}(CH2)3}N(CH3)2#H]# of fatty acids ive ion mode and deprotonated ions in negative ion
present in the mixture as unreacted materials (ions at mode.
m/z 283, 285, 313).
Negative FAB ionization cannot be used for identi- See also: II/Chromatography: Gas: Derivatization; De-
Rcation of amphoteric surfactants because these com- tectors: Mass Spectrometry. Chromatography: Liquid:
pounds do not give [M}H]\ ions, but it is an excellent Derivatization; Detectors: Refractive Index Detectors;
tool for a rapid detection of unreacted fatty acids, Ion Pair Liquid Chromatography. Chromatography:
which under these conditions give abundant de- Thin-Layer (Planar): Spray Reagents. Extraction: Solid-
Phase Extraction. III/Fatty Acids: Gas Chromato-
protonated molecular ions [M}H]\ (m/z 143 cap-
graphy. Flame Ionization Detection: Thin Layer
rylic; m/z 171 capric; m/z 197 laurylic; m/z 199 (Planar) Chromatography. Surfactants: Chrom-
lauric; m/z 227 myristic; m/z 255 palmitic; m/z 281 atography; Liquid Chromatography.
oleic; m/z 283 stearic acid) (Figure 5B). Where alkyl
(C12}C14) betaines and cocamidopropylbetaine have
been identiRed, this approach can also be successfully Further Reading
applied for the rapid detection of amphoteric surfac-
tants in Rnished detergent formulations. Borè P (1985) Cosmetic Analysis. Selective Methods and
Techniques, Cosmetic Science and Technology Series,
vol. 4. New York: Marcel Dekker.
Recent Developments Cross J (1977) Anionic Surfactants } Chemical Analysis,
Surfactant Science Series, vol. 8. New York: Marcel
The more recent developments in the Reld of surfac- Dekker.
tants analysis, in raw materials, in detergent formula- Cross J (1987) Nonionic Surfactants } Chemical Analysis.
tions, or in environmental samples, are all based on Surfactant Science Series, vol. 19. New York: Marcel
the application of new mass spectrometric soft ioniz- Dekker.
ation techniques (thermospray and atmospheric pres- Evans KA, Dubey ST, Kravetz L, Dzidic I, Gumulka J,
sure ionization (API)). These techniques are more Mueller R and Stock JR (1994) Quantitative determina-
rapid and versatile than conventional FAB-MS, which tion of linear primary alcohol ethoxylate surfactants
is dependent upon the surface activity of the sample in environmental samples by thermospray LC/MS.
in a given viscous liquid matrix and requires time- Analytical Chemistry 66: 699}705.
Hummel DO (1996) Analysis of Surfactants. Munich:
consuming screening of the matrix compounds to
Hanser Publishers.
achieve maximal ionization response. Maffei Facino R, Carini M, Minghetti P, Moneti G, Arlan-
Thermospray mass spectrometry coupled to rever- dini E and Melis S (1989) Direct analysis of different
sed-phase LC has been applied for the quantitative classes of surfactants in raw materials and in Rnished
determination of linear primary alcohol ethoxylate detergent formulations by fast atom bombardment mass
(AE) surfactants in environmental samples at levels spectrometry. Biomedical and Environmental Mass
from 25 to 102 ppb (total AE), corresponding to Spectrometry 18: 673}689.
2526 III / CRUDE OIL: LIQUID CHROMATOGRAPHY
Maffei Facino R, Carini M, Depta G, et al. (1995) Atmo- Porter RM (ed.) (1991) Critical Reports on Applied Chem-
spheric pressure ionization mass spectrometric analysis of istry, vol. 32: Recent Developments in the Analysis of
new anionic surfactants: the alkylpolyglucoside esters. Surfactants. London: Elsevier.
Journal of the American Oil Chemists’ Society 72: 1}9. Rieger MM (1997) Surfactants in Cosmetics. Surfactant
Metcalfe LD (1962) Potentiometric titration of long chain Science Series, vol. 68. New York: Marcel Dekker.
amine oxides using alkyl halide to remove tertiary amine Rosen MJ and Goldsmith HA (1972) Systematic Analysis
interference. Analytical Chemistry 34: 1849. of Surface Active Agents. New York: Wiley-Interscience.
Milwidsky BM and Gabriel DM (1982) Detergent Analy- Schmitt TM (1992) Analysis of Surfactants. Surfactant
sis. New York: Halsted-Wiley. Science Series, vol. 40. New York: Marcel Dekker.
CRUDE OIL: LIQUID CHROMATOGRAPHY
B. N. Barman, Equilon Enterprises, LLC, Houston, (and other monoaromatics) to polycyclic aromatic
TX, USA hydrocarbons (PAHs). The polars are usually aro-
Copyright ^ 2000 Academic Press matic in nature and consist of compounds that may
contain nitrogen, sulfur and oxygen as heteroatoms.
Asphaltenes are highly condensed aromatic struc-
Introduction tures.
Conventional TLC with silica and alumina adsor-
Chromatographic methods that utilize liquid mobile bents provides separation of components from crude
phases include open-column liquid chromatography, oils based on their polarity. TLC with Same ioniz-
high performance liquid chromatography (HPLC), ation detection (TLC-FID) has been applied for the
size exclusion chromatography (SEC) and thin-layer determination of hydrocarbon types. SEC has been
chromatography (TLC). These techniques have been particularly useful for the characterization of heavy
widely applied for the evaluation of crude oils (as crude oil fractions.
well as their subfractions) for their quality, processa-
bility or hazards. This overview covers various ap-
proaches to the characterization of crude oils by
Open-Column Liquid Chromatography
these techniques. SpeciRc applications, operational Crude oils have been fractionated into saturates, aro-
advantages and limitations of these methods are also matics, resins and asphaltenes using open-column
highlighted. liquid chromatography. Asphaltenes are n-pentane,
The major applications of open-column liquid n-hexane- or n-heptane-insolubles depending on the
chromatography and HPLC to the characterization of n-alkane used. The n-alkane-soluble materials,
crude oils and related materials including residua, termed maltenes, are usually fractionated on a silica
topped crude oils, coal liquids or shale oils involve or alumina column using appropriate solvents. In
preparative fractionation for the determination of general, saturates are extracted with an n-alkane
hydrocarbon types or class separation to be followed (such as n-hexane) followed by elution of aromatic
by the determination of important subgroups and and polar fractions with solvents or solvent mixtures
individual components. There are also numerous of higher eluotropic strengths. Quantitative data are
reports where analytical HPLC with various detec- obtained by the gravimetric determination of each
tion schemes has been applied to the quantitative fraction after evaporation of solvent or solvent mix-
characterization of crude oils as well as other fossil ture. Rotary evaporation under mild vacuum is
fuels. a common practice for the concentration of the col-
Crude oils are usually fractionated into several lected fractions.
compound classes according to their molecular struc- A crude oil separation scheme is shown in Figure 1.
tures. A majority of class separations have dealt with Maltenes are obtained by precipitation of asphaltenes
the determination of saturates, aromatics, resins (or from the crude oil using n-heptane. Using column
polars) and asphaltenes (SARA). Saturates consist of liquid chromatography on alumina, and solvents or
parafRnic and naphthenic compounds. If oleRns solvent mixtures as indicated in Figure 1, fractions
are present in the sample, they are usually grouped enriched with saturates, aromatics I and II and polars
with saturates. Aromatics range from alkylbenzenes can be obtained. The aromatics I fraction contains
III / CRUDE OIL: LIQUID CHROMATOGRAPHY 2527
Figure 1 Fractionation of crude oil by open-column liquid chromatography. Column void volume V3"35 mL.
monoaromatics, and aromatics II fraction is com- and aromatics on the silica. After removing residual
posed of diaromatics and polycyclic aromatic com- aromatics from the clay adsorbent by washing with
pounds. n-pentane, polars are desorbed with a 1 : 1 tol-
Many special separation schemes have been ap- uene}acetone mixture. Saturates and polars are deter-
plied to obtain fractions enriched with speciRc types mined gravimetrically by complete evaporation of
of compounds by tailoring the adsorbents as well as solvents from the n-pentane and toluene}acetone
the eluting solvents. Beside enrichment and type sep- fractions, respectively. The amount of aromatics is
aration of hydrocarbons, the following have been calculated by difference or, if desired, aromatics can
typical separation schemes for the characterization of be recovered by Soxhlet extraction using toluene.
crude oil. Both small (mg or less) and large (multigram) scale
fractionations of crude oil can be carried out by
1. Solubility fractionation of crude oil with solvents
open-column solid}liquid chromatography to obtain
of increasing eluotropic strength, for example, in
fractions for further evaluation. These fractions are
the order of n-pentane, cyclohexane, toluene and
analysed by a host of analytical techniques such as gas
methylene chloride.
chromatography (GC), pyrolysis GC, HPLC, infrared
2. Crude oil fractionation of sulfur and nitrogen
(IR) spectroscopy, and nuclear magnetic resonance
compounds for their characterization or identiRca-
(NMR) spectroscopy. Elemental compositions for C,
tion by other methods.
H, N, S, O, V and Ni can also be determined.
3. Fractionation of PAHs for their determination,
Class separation by open-column liquid chromato-
ring number distribution and degree of alkylation.
graphy is labour-intensive and often suffers from in-
4. Fractionation of saturates into n-alkanes and cyc-
accuracies due primarily to overloading effects that
lic plus iso-alkanes using a 0.5 nm molecular sieve
result in cross-contamination of hydrocarbon types.
column.
Moreover, there is possible loss of light components
5. Fractionation of acidic and basic materials from
during evaporation of solvent from a collected frac-
crude oil using anion exchange and cation ex-
tion that can add signiRcant uncertainties in the
change resins, respectively.
quantitative data. There can also be recovery prob-
ASTM D2007 clay-gel method (approved by the lems arising from irreversible adsorption of some
American Society for Testing and Materials for the compounds on the adsorbents.
determination of characteristic groups in rubber ex-
tender and processing oils and other petroleum- High Performance Liquid
derived oils) has been applied to hydrocarbon-type
determination of crude oils. In this method, two glass
Chromatography
percolation columns are connected in series with the Modern HPLC provides both preparative and ana-
upper column containing clay and the lower one lytical scale separation of hydrocarbon types from
having clay at the top and silica gel at the bottom. The crude oils. Preparative separation is carried out prim-
sample solution in pentane is added at the top of the arily for gravimetric determination of hydrocarbon
upper column. n-Pentane is then used to elute satu- types after the removal of solvents from the collected
rates from clay and silica, leaving polars on the clay fractions. HPLC offers Sexibility to allow separation
of different compound types based on their polarity tector at a speciRc wavelength varies with the number
and afRnity using various solvents and adsorbents. An of aromatic rings, or with the isomeric compounds
HPLC system can also be automated. Since rapid having the same number of aromatic rings. The very
fractionation of materials for collection is possible by small difference in the RI of n-hexane (or n-heptane)
HPLC, the method offers a much desired alternative and saturates is a serious limitation of an RI detector
to lengthy and multistep open-column liquid to be highly effective for the determination of satu-
chromatography. rates. The RI responses from aromatics and polar
Crude oils are complex materials consisting of compounds are much lower compared to those from
hundreds of individual compounds of different sizes UV detection. The response factors of different hy-
and molecular structures. The resolution level ob- drocarbon types, deRned as peak area per unit con-
tained during initial separation of crude oil for com- centration for both RI and UV detectors, show vari-
pound classes by HPLC is often marginal. Therefore, ations with many factors, including the crude oil,
when determination of individual compounds or column type, column dimensions and mobile phase.
functionalities is involved, fractions from the initial Therefore, experimental response factors are often
HPLC separation are subjected to ofSine or online determined by preparative separation and collection
analysis by high resolution chromatographic or spec- of different hydrocarbon groups followed by ana-
troscopic techniques. lytical separation and detection of the collected
fractions. For hydrocarbon-type analysis, each
Columns and Solvents calibration is only valid for samples of the same
source of crude oil.
Both analytical and preparative scale separations of
Some special HPLC detectors, including FID, mass
hydrocarbon types by HPLC have been carried out on
spectrometric detector (MSD) and evaporative light-
commercially available silica, aminopropylsiloxane-
scattering detector (ELSD) have been used in conjunc-
bonded silica, 2,4-dinitroanilinopropylsiloxane-
tion with HPLC separation of crude oils or their high
bonded silica or cyanopropylsiloxane-bonded silica
boiling residua. These detectors have some attractive
columns. Amino- and cyanopropylsiloxane columns
features. FID and ELSD are more sensitive than RI
have been used extensively in the class separation of
detectors and provide reasonably uniform response
crude oils and other fossil fuels.
factors for different hydrocarbon types. MSD pro-
n-Hexane and n-heptane have been the most com-
vides detection of materials with high sensitivity and
monly used mobile phases for the separation of satu-
speciRcity, and affords valuable information on mo-
rates from other hydrocarbon classes. Mobile-phase
lecular weight, structure and functionality of the mol-
modiRers such as methylene chloride, chloroform,
ecules.
tetrahydrofuran, methanol or 2-propanol have also
Although both FID and MSD have been successful
been used to facilitate the elution of aromatics and
as universal detectors in gas chromatography, the use
polars.
of these detectors with HPLC has been limited, prim-
Preparative columns can be much larger in size
arily because of the lack of effective interfaces to
(with typical column dimensions 80;0.6 cm com-
remove the mobile phase and to transport the sample
pared to 25;0.46 cm or less for the analytical col-
to the detection system. A rotating disc FID has been
umn) to allow the separation of a few tenths of
demonstrated with samples boiling above 3403C to
a gram or more of sample in each run. The particle
obtain hydrocarbon-type data covering a wide range
size of the porous packing materials can also be much
of compositions. HPLC-MSD with thermospray or
larger than that for the analytical column (50}100
moving belt interface has been applied to the charac-
versus 5}10 m). Typical Sow rates for preparative
terization of heavy hydrocarbons.
separations are 20 mL min\1 or higher, compared to
The ELSD has been applied to the hydrocarbon-
2 mL min\1 or less for analytical separations. Separ-
type determination of fossil fuels boiling above 3153C
ation times in both cases can be 30 min or less.
(6003F). With the ELSD, mobile phase containing the
analyte is nebulized with an inert gas (such as nitro-
Detectors
gen) and sprayed into a heated drift tube where the
The choice of detectors for the quantitative deter- mobile phase is vaporized, leaving behind a Rne mist
mination of various compound classes has been a of dry micrometre-sized droplets in solvent vapour.
major concern in HPLC. Conventional diode-array As the sample particles pass through a Sow cell, they
UV detector and differential refractive index (RI) scatter light from a laser beam to produce a signal.
detector are commonly used with HPLC. A UV The mass versus signal from ELSD is usually non-
detector is not suitable for saturates as they do not linear. However, the signal can be linearized using
have chromophores. The response from a UV de- a power}law model c"msb, where c and s are mass
and signal, and m and b refer to the proportionality Both resolution and selectivity of hydrocarbon type
constant and power}law exponent, respectively. separations can be enhanced using different column
Element-speciRc detectors such as graphite furnace types and multiple mobile phases during a single run.
atomic absorption (GFAA) detector and inductively A solvent with higher eluotropic strength than an
coupled plasma}atomic emission spectrometry (ICP- n-alkane or a multisolvent step gradient has been
AES) detector have been interfaced with HPLC for found to be more effective for total recovery of aro-
the speciation of metal-containing compounds, in- matic and polar compounds.
cluding metal porphyrins. Electrochemical detectors A separation scheme B (reported by Pearson and
have been applied to the detection of electroactive Gharfeh, 1986) utilizes two pumps, two six-port
compounds such as phenols. valves, one cyanopropyl silica column and two
aminopropyl and cyanopropyl silica columns. Using
n-hexane from the Rrst pump, the sample is allowed
Separation Schemes
to pass through the cyanopropyl silica column where
There have been a few efforts to optimize separation polar compounds are strongly retained. The saturates
between hydrocarbon groups or subgroups. Most in- and aromatics pass on to the two aminopropyl and
volve trials with the separation and detection of cyanopropyl silica columns. The cyanopropyl silica
model compounds expected to be present in the column is then isolated using the Rrst switching valve.
sample. In almost all previous HPLC studies of crude As soon as the saturates elute, the second valve is used
oils and related materials, instruments have been to backSush the aminopropyl and cyanopropyl silica
automated for separation, detection and collection columns for the elution of aromatics. The polar com-
of fractions. pounds are then eluted by back Sushing the cyano-
A simple HPLC separation scheme (scheme A) is propyl silica column with methyl tert-butyl ether
shown in Figure 2 where a single six-port valve from the second pump. Figure 3 shows a chromato-
and a single column (or a series of columns) have gram of a crude oil residue obtained by the separation
been used. Usually, deasphalted crude oil solution scheme B using FID for detection. Since a crude oil
in hexane is injected for analysis by HPLC. The residue meets the boiling point requirements for an
saturates and neutral aromatics (monoaromatics, ELSD to be quantitative, ELSD can also be applied to
diaromatics, triaromatics and tetraromatics) are detect and determine its components separated by
separated Rrst with the forward Sow of hexane. HPLC.
After this, the valve-switching causes a reversal
Special HPLC Applications
of mobile-phase Sow through the column, allowing
the backSushing of the column for the elution Both preparative HPLC and analytical HPLC have
of polar compounds. The cut point between saturates been utilized for sample clean-up and separation of
and aromatics is determined by an RI detector while speciRc groups of compounds for online or ofSine
cut points between aromatic ring subfractions as characterization by capillary GC, GC-mass spectro-
well as polar compounds are determined by a UV metry (MS) or GC-element-speciRc detection. Col-
detector. umn-liquid chromatography or semipreparative
Figure 2 A scheme for hydrocarbon-type separation from crude oil by HPLC.

common detectors are RI, ELSD and UV. Typical

separation times are 40 min or less.
Usually molecular weight distributions of fossil
samples are determined relative to narrow polysty-
rene standards. Although both RI and ELSD provide
mass versus molecular weight distribution of the
sample, a UV detector can be used for similar distri-
bution for speciRc compound type, using a com-
pound-speciRc wavelength. For example, distribution
of petroporphyrins in polar fractions can be obtained
Figure 3 HPLC chromatogram of a crude oil residue. (Modified by monitoring sample elution with a UV detector at
with permission from Pearson and Gharfeh, 1986. Copyright 1986 around 400 nm. Petroporphyrin distributions have
American Chemical Society.) also been obtained by SEC-GFAA and SEC-ICP-AES
detections.
HPLC fractionation followed by analytical HPLC
have also been very effective for the determination of
certain compound types, including PAHs, petropor-
Thin-Layer Chromatography
phyrins, sulfur heterocycles, phenols and nitrogen TLC using silica or alumina plates and appropriate
bases such as azarenes. solvents can provide quick class separation of a crude
oil primarily for collection of its fractions. The detec-
tion methods for the separated components include
Size Exclusion Chromatography visual examination of coloured spots, observation by
SEC is usually applied to the analysis of high molecu- irradiation with UV light, UV-Suorescence scanning
lar weight materials. Unless there are speciRc interac- and spraying of chromogenic or Suorogenic reagents.
tions between the column packing and the sample Rapid and direct determination of hydrocarbon
components, the elution of molecules in SEC is types of a crude oil can be achieved by TLC coupled
bounded by limits representing total exclusion and with FID. In TLC-FID (which is also known as Iatro-
total permeation through macroporous particles scan), thin reusable quartz rods (Chromarods) sin-
packed into a column. In this method, the high mo- tered with micrometre-sized silica or aluina particles
lecular weight materials elute Rrst, followed by small- are used as the stationary phase. After spotting a few
er molecules, which is the basis for the determination micrograms of the sample, the Chromarods are de-
of molecular weight distribution by SEC. veloped sequentially with several solvents or their
Typical crude oils contain compounds having low mixtures of decreasing eluotropic strengths to achieve
((100 Da) to high molecular weights ('2000 Da). desired separation. For example, the Chromarods can
The average molecular weight can be around be developed Rrst with toluene for 5 min to separate
1000}1500 Da. Since 30% or more of the materials saturates plus aromatics from polar compounds.
in the crude oil may have molecular weights similar to After this, they can again be developed with an n-
that of the mobile phase and therefore, falling close to alkane (such as n-heptane) for 30 min, for the separ-
the total permeation limit of typical SEC columns, ation of saturates from aromatics. During the second
SEC is more suitable for the determination of molecu- step, the position of polar compounds remains un-
lar weights of heavy petroleum fractions derived from changed, and aromatics are distributed broadly ac-
crude oil. Atmospheric or vacuum residua and polars cording to their polarity or the number of aromatic
and asphaltene fractions from crude oil have been rings. As an option, polar compounds can also be
analysed by SEC to monitor changes during their subdivided into two resin types by developing the
processing or to fractionate them into narrower cuts Chromarods with a 9:1 chloroform}methanol mix-
for further analysis. SEC can also be used as a sample ture for 3.5 min. The Chromarods are dried in an
clean-up technique to remove small molecules from oven after each development. Finally, each
high molecular weight materials or vice versa. Chromarod is scanned by passing it through an oxy-
The analysis of crude oil or its fractions by SEC is gen}hydrogen Same for the detection of separated
carried out using a series of small pore size (usually components.
4100 nm) poly(styrene-divinylbenzene) columns and Due to limitations in the design of the commercial-
mobile phases of tetrahydrofuran, toluene, chloro- ly available instruments, the TLC-FID method is only
form, N-methyl-2-pyrrolidone or pyridine. The quantitative for samples boiling above 3003C. The
sample size can be 5 mg or less. Almost all detectors low boiling materials present in the whole crude oil
used with HPLC can be coupled with SEC. The most evaporate during development and drying. Such
evaporation may also occur when Chromarods are A cursory examination of chromatograms of a topped
exposed to the Same during scanning of the adjacent crude oil in Figure 4B and of a vacuum residue in
rod. Therefore, TLC-FID is more suitable for high Figure 4C suggests a marked compositional differ-
boiling residua or fractions derived from crude oils ence between these two samples. Note that the vac-
than for whole crude oils. uum residue was derived from the topped crude oil as
The three TLC-FID chromatograms in Figure 4 vacuum tower bottoms after vacuum distillation. The
have been obtained by developing Chromarods with vacuum residue contains much more polar and large
toluene for 5 min followed by n-heptane for 30 min. ring aromatic compounds than the topped crude oil.
Figure 4A demonstrates the capability of TLC-FID SpeciRcally, the amounts of saturates, aromatics and
for the resolution of saturates and aromatics, and of polars are 28.6, 56.8 and 14.6% (w/w) in the topped
the ring-based separation of aromatic compounds. crude oil compared to 9.0, 58.5 and 32.5% (w/w),
Peaks 1}5 in this chromatogram are for Nujol, respectively, in the vacuum residue.
n-heptadecylbenzene, 2,3-dimethylnaphthalene, 9- Unlike model aromatic compounds in Figure 4A,
methylanthracene and 9,10-dimethyl-1,2-benzan- aromatics in both topped crude oils and vacuum
thracene, respectively. The unidentiRed peaks (right- residua do not show distinct peaks for different ring
hand side of Figure 4A) are for polar impurities. types. This is primarily due to chromatographic over-
Figure 4 TLC-FID chromatograms. (A) Separation of hydrocarbons by class and number of aromatic rings using a model mixture
described in the text; (B) topped crude oil; (C) vacuum residue.
2532 III / DECANTER CENTRIFUGES IN PHARMACEUTICAL APPLICATIONS
lap arising from degree of substitution as well as Further Reading

substituents of diverse chemical structures attached to
Ali MA and Nofal WA (1994) Application of high perfor-
the monoaromatic as well as polycyclic aromatic mance liquid chromatography for hydrocarbon group
compounds. type analysis of crude oils. Fuel Science and Technology
International 12: 21}33.
Future Possibilities Altgelt KH and Gouw TH (eds) (1979) Chromatography in
Petroleum Analysis. Chromatographic Science, vol. 11.
At present, relevant analytical data on crude oil com- New York: Marcel Dekker.
position are very useful in crude oil processing as Barman BN (1996) Hydrocarbon type analysis of base oils
petroleum reRners are increasingly using blended and other heavy distillates by thin-layer chromato-
feeds instead of limiting the plant operation to graphy with Same-ionization detection. Journal of
a single type of crude oil. The compositional data will Chromatographic Science 34: 219}225.
be more important as the depletion of light and sweet Grizzle PL and Sablotny DM (1986) Automated liquid
crude oils continues and reRneries process more chromatographic compound class group-type separation
heavy and sour crude oils. For process control, prod- of crude oils and bitumens using chemically bonded
aminosilane. Analytical Chemistry 58: 2389}2396.
uct quality, catalyst performance and environmental
Hsu CS and Qian K (1993) High boiling aromatic hydrocar-
compliance, faster and better analytical approaches bons characterized by liquid chromatography}thermos-
to crude oil characterization will become a vital part pray}mass spectrometry. Energy & Fuels 7: 268}272.
of the reRnery operation. As far as liquid chromato- Neal AC (1995) HPLC and column liquid chromatography.
graphic methods are concerned, multistep column- In: Adlard ER (ed.) Chromatography in the Petroleum
liquid chromatography will be less attractive due to Industry, Ch. 12, pp. 247}394. Amsterdam: Elsevier.
limitations such as lengthy turnaround times, high Padlo DM, Subramanian RB and Kugler EL (1996) Hydro-
level of uncertainties in the compositional data, ex- carbon class analysis of coal-derived liquids using high
cessive solvent consumption and costs involved in performance liquid chromatography. Fuel Processing
their disposal. Modern HPLC, SEC and TLC-FID Technology 49: 247}258.
methods will continue to provide valuable informa- Pearson CD and Gharfeh SG (1986) Automated high-per-
formance liquid chromatography determination of hy-
tion on crude oils. Most likely, faster and better
drocarbon types in crude oil residues using a Same
results will be achieved through major developments ionization detector. Analytical Chemistry 58: 307}311.
in system automation for coupling these techniques Sinninghe DansteH JS, Rijpstra WIC, de Leeuw JW and
with other chromatographic and spectroscopic Lijmbach GWM (1994) Molecular characterization of
methods. Such hyphenated techniques are already organically-bound sulfur in crude oils. Journal of High
being pursued to Rnd better approaches to the charac- Resolution Chromatography 17: 489}500.
terization of fossil fuels, including crude oils. Willsch H, Clegg H, HorsReld B et al. (1997) Liquid
chromatographic separation of sediment, rock and coal
See Colour Plate 71. extracts and crude oil into compound classes. Analytical
See also: II/Chromatography: Liquid: Detectors: Ultra- Chemistry 69: 4203}4209.
violet and Visible Detection; Mechanisms: Size Exclusion Xu H and Lesage S (1992) Separation of vanadyl and nickel
Chromatography. III/Flame Ionization Detection: Thin- petroporphyrins on an aminopropyl column by high-
Layer (Planar) Chromatography. Flash Chromatogra- performance liquid chromatography. Journal of
phy. Chromatography 607: 139}144.
DECANTER CENTRIFUGES IN
PHARMACEUTICAL APPLICATIONS
J. V. McKenna, Alfa Laval, Warminster, PA, USA areas of application. Growing demands for higher
yields and dryer cell cake, punctuated by the pursuit
of lower cost solutions, continue to keep high perfor-
mance centrifuges at centre stage in the quest for
Centrifugation continues to perform a vital role in better separation procedures. In addition, tighter re-
pharmaceutical process operations. In the 1990s new gulations and inspection procedures continue to
enhancements to the technology attracted many new act as catalysts for the development of fail-safe
III / DECANTER CENTRIFUGES IN PHARMACEUTICAL APPLICATIONS 2533
Figure 1 Basic decanter.

Figure 2 (See Colour Plate 72). General decanter operation.
equipment for sterility and containment to meet more

stringent user demands. Within this environment, charge ports, while clariRed liquid or centrate over-
a decanter (Figure 1), which is the most compact type Sows the plate dams situated at the opposite end of
of separation equipment available, is often the centri- the bowl. While decanters can handle a wide range of
fuge of choice for producing pharmaceuticals such particle sizes, dryer cake is always produced when
as erythromycin, insulin (human and animal) and particle size is maximized } preferably to 2}10 m
many other large bulk products. These workhorse depending upon G-force. It is important to note that
machines perform a wide range of separation duties, in biotech lysed Escherichia coli separations, cakes
including liquid}solid, liquid}liquid, and that are 30}40% dry have been achieved with particle
liquid}liquid}solid separations and liquid}liquid ex- sizes of 1 m or less.
tractions. Decanters have become most useful for In practise, decanting centrifuges provide excellent
many pharmaceutical operations including recovery performance levels when handling slurries with sig-
and solids purity. These reliable and versatile machin- niRcant amounts of solids (generally '10% by vol-
es are active in many of industry’s typical process ume), and are often the only choice for solids concen-
steps, e.g. cell harvesting, concentration and washing, trations above 40% by volume. With appropriate
cell debris removal, inclusion body recovery and conveyor modiRcations they can also be used to sep-
puriRcation, solvent extraction, dewatering and arate two liquids from each other or from a solid
recovery. phase. The scroll of the decanter provides a continu-
Currently, the trend in solid}liquid separation ous discharge of solids. By interposing nozzles near
is towards higher solids fractions in the fermentation the beach section it is possible to provide a washing
process. This translates into greater productivity step before the solids are discharged. This method can
and lower capital costs per pound of product produc- be used to wash various crystals. In contrast to basket
ed. Typical fermentation processes involve amino or disc-stack centrifuges,where liquid Sow is laminar
acids, antibiotics, bacteria, enzymes, ethanol, fungi, and Sow patterns are relatively simple and contained,
inorganic salts, organic acids, polypeptides, poly- the Sow patterns and consequent design calculations
saccharides, steroids, vaccines, vitamins and yeasts. are more complex for decanters. Separation perfor-
Decanter centrifuges are also used in extraction mance is more subject to analysis by experience and
and fractionation processes for hormones, alk- pilot testing.
aloids, blood separation (animal) and natural product
isolation.
Vertical Units
Most decanter centrifuges are horizontal, but vertical
General Operation orientations (Figure 3) are also available. The same
Normally decanters handle two- or three-phase sep- centrifugal operating principles apply to both types,
aration of high-solids slurries (Figure 2). The slurry and process performance is not affected by whether
enters the centre of the bowl, and denser solid par- the axis of rotation is vertical or horizontal. The main
ticles are sedimented against a rotating bowl wall. difference between the two types lies in the seal de-
The less dense liquid forms a concentric inner layer. sign. A vertical machine has only one high speed
Centrifugal force compacts the solids as they are rotating seal plus one low speed (42}52 rpm) or static
’ploughed’ out of the pond and up the conical ‘beach’ seal, while a horizontal unit has at least three high
by a helical screw conveyor. Solids exit through dis- speed seals. For this reason, a vertical conRguration is
disc-stack centrifuges were primarily used for ap-

plications, from pilot to full production. Today
decanters are capable of 10 000 G operation, making
the machines competitive with disc-stack centrifuges
for speciRc pharmaceutical applications.
Continuous operation is another reason why the
Figure 3 (See Colour Plate 73). Vertical decanter. industry often opts for decanters. The feed and dis-
charge operations (solids and liquids) never cease.
There is less downtime and maintenance is substan-
tially lower than with other separation alternatives.
easier to seal. With only one sealing surface this unit At the same time decanters are less affected by
is often the preferred choice when handling hazard- feed variations, such as varying particle size,
ous materials or toxic materials. Vertical centrifuges than Rlter systems which require changing media
are often used in high temperature, high pressure mesh sizes. The centrifuges also more efRciently sep-
applications for handling pressures from vacuum up arate adhesive, slimy, high viscosity or Suffy mater-
to 10 bar with process temperatures from !10 to ials, which often blind or impair Rlters. They also
7003F. provide dryer cake results for these materials than do
If Soor space is a concern, vertical decanters offer Rlters.
distinct advantages over horizontal machines as they Minimal liquid loss is also associated with de-
require 60}80% less Soor area. They also provide canters. Permeable Rlters lose liquid through absorp-
more production per square foot with less investment tion as well as losing Suids via Rlter disposals. This
in piping, foundations and real estate. loss becomes critical if a high value is attached to the
liquid phase. Also, Rlters usually separate only two
phases simultaneously, requiring another pass for
Factors of Acceptance a third separation, while centrifuges are capable of
For small batch fermentations (100 L or less) mem- one-step, three-phase liquid}liquid}solid separations
brane technology and mechanical Rltration remain (e.g. oil, water, solids).
the prevalent technologies for processing diverse
stream arrays. However, for large volume runs re- Improvements to Decanters
quiring environmental measures and control, de-
canters are normally speciRed for several reasons. Several changes to decanter technology have affec-
First, the enclosed seal design of this type of cen- ted the performance of the machines in the last
trifuge type offers inherent advantages. Decanters 10 years.
can be sealed hermetically for the protection of
Settling Vanes
operating personnel, and offer clean-in-place opera-
tion and sterilization. Second, the efRcient operation New conveyors have been Rtted with unique incline
of a decanter can produce a dry continually dis- vanes to enhance the collection area of the disc unit.
charged solids phase without contamination of Settling vanes (Figure 4) reduce turbulence and strat-
either the product stream or the centrate. Also ify feed Sow in the clariRcation zone to promote
advancing work on new metallic Rnishes continues settling and prevent re-entrainment of solids. This
to offer promise for new sterility solutions and
safeguards.
Additional factors also work in favour of these
machines. Possibly foremost is the growing need to
produce dryer cake. As expected, when there is
less water solids require less energy for drying.
Depending upon the application, decanters can
produce dry solids as high as 30}95% for dry solids.
The high G-force capabilities of the units can separate
out even the Rnest particles, resulting in cleaner
centrate.
The ability to operate at higher G-forces remains
a goal. Indeed, high G-force separation continues to
expand. Five years ago decanter centrifuges could
only generate forces up to 5000 G. During that period Figure 4 Conveyor with settling vanes.
III / DECANTER CENTRIFUGES IN PHARMACEUTICAL APPLICATIONS 2535
Figure 5 (See Colour Plate 74). Performance-enhancing features of a decanter centrifuge.
results in more efRcient solids recovery and a 50% innovation acts as a catalyst to force newer solids to
improvement in centrate clarity. move old solids out of the bowl.
Hydraulic Disc (Baf]e Disc) Ef\cient Feed Zone
This feature shown in Figure 5 enhances the separ- Redesign of the feed zone as shown in Figure 5 has
ation of soft, gelatinous materials such as proteins, helped to reduce the shear forces on the feed material
enzymes, and fermented products. The disc is used as much as possible. Shear forces on the liquid
to increase the separation efRciency on hard-to-scroll stream can cause problems, especially when liquid
materials (materials that tend to slip on the conical goes from rest to bowl speed. Upgrade designs now
section of a decanter). Acting like a dam, solids build allow for gentle acceleration and entrance into the
up around the disc, holding in the liquid centrate to bowl by means of a low-shear feed zone. This im-
provide ultra-deep pond settlings above the level of provement has eliminated the need for acceleration
the solids discharge. The resulting hydrodynamic blades.
pressure compacts solids, assists scrolling, and ex- Variable Speed and Differential Control
trudes the compacted solids past the disc and out of
Programmable Logic Controller (PDC) and digital
the centrifuge.
controllers are now used to control the large speed
swings of a decanter’s backdrive motor that are
Dryer Cake: Plough Tile/Conveyor Innovations
associated with differential control. These systems
New designs now provide purer and dryer cake. provide the necessary control features to monitor
Recent innovations permit the centrifugal retention and stabilize a decanter’s eddy current brakes, direct
of solids for longer periods of time under maximum current motor, and alternating current variable
G-force. A new conveyor handling compressible
solids, such as Rbrous materials, has been able to
produce up to 35% dryer solids in vitamin, gluten,
and cornstarch processing applications.
A newly developed plough tile conveyor (Figure 6)
uses scoop-shaped tiles to reduce the torque produced
by the solids moving through the centrifuge. By lifting
and turning the solids from the bowl wall these tiles
reduce the torque, produce dryer cake, and increase
capacity. The addition of this innovation enables the
conveyor to achieve up to twice the throughput of
standard conveyors.
Another additional improvement is the ‘Kiwi con-
veyor’ (Figure 7), a patented Sightless conveyor that
achieves dryer cake to improve clarity. This unique Figure 6 Plough tile conveyor.
available from some manufacturers. New software

eliminates second guessing and enables users to
analyse machines on a continuous basis, e.g. monthly,
weekly, etc., by checking for new vibrational patterns
due to changes in unit load, balance, or the hydraulic
dynamics of the bearing. Conditioning monitoring
provides real-time information about the working
operation of the centrifuge. Accompanying instru-
mentation and probes also analyse critical areas such
as bearing patterns to note changes, predict failure
and prevent unexpected costly shutdowns. Data ac-
quired and stored in the program can also be used to
determine what spare parts to keep in stock.
Figure 7 ‘Kiwi’ conveyor.
frequency drives to within 0.1 rpm of the desired set Conclusion

points. By maintaining lower differential controls, The development of pharmaceutical products re-
users achieve dryer cake. quires separation equipment that delivers high re-
Centripetal Pumps and Containment
coveries and high solids purity. During the 1990s
decanter centrifuges closed the gap on or, depending
Optional centripetal pumps can be added to minimize upon the application, even surpassed the advantages
foaming and aerosols, and gently discharge delicate once held by alternative separation devices. In centri-
pharmaceuticals without the risk of oxidation or fugation, operating costs are notably lower than with
contamination to the product stream. Such pumps Rltration devices with capital investment of hardware
help control liquid cresting and the separation efR- as the only signiRcant cost. Decanters also require less
ciency of the unit. auxiliary equipment such as secondary pumps, tanks,
and extra ventilation machines. Thus large-scale
Vapour Sealant
pharmaceutical operations rely on continuous centri-
For applications requiring biohazard-safe, vapour- fuges to produce maximum possible amounts of dry
sealed operation, casing seals are available to contain cake and decanters are well suited to this task.
pressure up to 14.4 kPa inches of water. Lip seals,
labyrinth seals, and gas-purged mechanical seals are See Colour Plates 72, 73, 74.
typically used. If required, decanters can be equipped
with steam sterilization for aseptic steam processing.
Further Reading
Clean-In-Place Operation
Alfa Laval (1984) Complete Range of Centrifuges for the
Self-cleaning units are also available for faster and Fermentation Industry. Technical Brochure IB 41045E.
safer pharmaceutical processing. Built-in motor con- Tumba, Sweden: Alfa Laval.
trols safely reduce rotation to 1 G and automatically Amber CM (1961) Centrifugation equipment theory.
clean the decanter’s bowl and casing. This eliminates Industrial Engineering Chemistry 53: 430}433.
the need to tear down or dismantle the centrifuge. Arkinson B and Mavituna F (1983) Biochemical Engineer-
ing and Biotechnology Handbook. New York: Nature
Press.
Conditioning Monitoring Axelsson HAC (1998) Centrifugation. Technical Brochure.
Tumba, Sweden: Alfa Laval.
Together with the decanter improvements, the capa- Davies E (1965) Selection of equipment for solid}liquid
bility to maintain product integrity has been further separation. Transactions of the Institution of Chemical
enhanced by conditioning monitoring programs Engineers 43: T256}T259.
III / DE-INKING OF WASTE PAPER: FLOTATION 2537
DE-INKING OF WASTE PAPER:

FLOTATION
C. Jiang and J. Ma, Enzymatic Deinking Technologies, the Rbre surface and removing them by washing,
Norcross, GA, USA Sotation, cleaning and screening. Common con-
taminants include ink, staples, paper clips, sand, plas-
tics and stickies. The most important and widely used
de-inking process to date is the froth Sotation pro-
Introduction cess. This process removes the widest range of ink
The earliest work on utilization of froth Sotation for particles from wastepapers. Flotation alone or in
de-inking of wastepapers dates back to the 1930s. combination with other processes can remove almost
The Rrst Sotation de-inking patent was Rled by Hines all types of ink particles and other contaminants from
in 1933. However, it was not until 1952 that the Rrst the slurry of wastepaper.
commercial Sotation de-inking system was installed
at a paper mill in the USA, and the Rrst European
installation was at a tissue mill in Greece in 1959. Up Flotation Process and Equipment
to 1970, the growth of Sotation de-inking technology Flotation Process
was relatively slow. However, in the past 20 years,
the market has grown extremely rapidly. The world- Pulping The Rrst step in de-inking wastepaper is
wide Sotation capacity for de-inking of wastepapers pulping. Because of the nature of the chemicals and
has increased from 0.2 million tons in 1965 to about equipment used to pulp wastepaper, the pulping pro-
25 million tons in 1995. cess is analogous to sulRte ‘cooking’. Chemicals, to-
De-inking is a separation process to remove inks gether with heat and mechanical energy, are used to
and other nonRbrous contaminants from waste- detach the ink particles and other contaminants from
papers. Different types of units are required to separ- the Rbres in a pulper. Pulping can be either a batch or
ate inks from Rbres, and this mainly includes wash- continuous process. Newsprint mills typically use
ing, Sotation, cleaning and screening. The selection continuous pulping at a consistency of 4}8%. Other
and operation of these units are based on the types mills usually use batch pulping at a higher consistency
of wastepapers and the requirements of the Rnished of 8}18%.
de-inked pulp. Wastepaper is commonly grouped
into Rve categories, which include mixed paper, Flotation Flotation de-inking is a selective separ-
old newspapers, old corrugated containers, pulp sub- ation process that utilizes the difference of surface
stitutes and high grade de-inked. Table 1 shows physicochemical properties between the ink and Rbre.
a typical wastepaper classiRcation and the Rnished Flotation chemicals are fed to the wastepaper slurry
products obtained from different kinds of waste- to render the ink particles selectively more hydropho-
papers. bic and hence to increase the Soatability. When the
De-inking is a two-stage process which involves air bubbles are sparged into the Sotation cell contain-
dislodging the ink and nonRbrous contaminants from ing the wastepaper slurry, the ink particles get at-
Table 1 Typical wastepapers and their classification
Category Compositions and abbreviations Finished products
Old newspapers Old newspapers (ONP), white blank newspapers Newsprint, boxboard, tissue, paper towels
Old corrugated containers Old corrugated containers (OCC), double Linerboard, boxboard, corrugating medium,
linerboard kraft corrugated clippings (DLK) kraft towels
Mixed paper Mixed office waste (MOW), old magazines Printing and writing paper, paperboard, tissue,
(OMG), sorted ledgers, computer printout (CPO), paper towels, magazine newsprint
manifold white ledger (MWL), sorted office
waste (SOW)
Pulp substitutes Unprinted paper and board, boxboard Fine paper, tissue, envelopes
cuttings, printer trims, envelope cuttings
High grade de-inked Fine paper, tissue, printing and writing papers
2538 III / DE-INKING OF WASTE PAPER: FLOTATION
tached to the air bubbles due to their relatively high Evolution of Flotation Cells
hydrophobicity and are Soated to the surface of sus-
pension, and the hydrophilic Rbres remain in the Froth Sotation is the most widely used separation
water phase. process in modern paper mills. During the last 10
years, the development of Sotation de-inking cells
Flotation De-Inking System
has been pursued more aggressively than the tech-
Flotation is traditionally the standard European nologies of any other segment of the pulp and paper
de-inking system for old newspapers. The stocks or industry.
wastepaper slurry after pulping generally go through Initially, Denver Sotation cells used in the mineral
a soaking stage in a dump chest to swell the Rbres and industry were installed in paper mills. These cells are
to improve ink detachment from the Rbre. open, rectangular vats, with mechanical removal of
The stock is subsequently aerated at 0.7}1.5% con- Sotation froth by a rotating paddle and mechanical
sistency in a series of Sotation cells. Typically, six to mixing of air and pulp suspension at the bottom.
10 Sotation cells in series (primary Sotation) are However, these cells are not currently in use.
required for efRcient ink removal. The froth from Although the development of Sotation cells for
primary Sotation is subsequently cleaned in a second- de-inking was less dramatic in the early years, there
ary or recovery stage (usually two cells) to further have been many changes in cell design in the last two
recover food Rbres and to decrease Rbre loss. A typi- decades. Table 2 lists the historical development of
cal and representative example of a Sotation system is Sotation de-inking cells.
illustrated in Figure 1. The major driving forces of the evolution of mod-
Most technical advances made during the past 10 ern Sotation cells are the reduction in energy and
years involved utilization of a combination of Sota- water consumption, lower footprint space and an
tion and washing stages to remove inks from the more increase in efRciency and capacity. Although many
complex wastepaper. The concept of the post- changes in Sotation cell design have been made, the
Sotation system is to add a disperger or kneader improvements in Sotation de-inking performance are
between two standard Sotation stages. The dispersion not always obvious. More recently, because of the
or kneading stage further helps the detachment and great advances in printing, coating and modiRcation
size reduction of ink and other nonRbrous particles, of paper by converters to impart special properties,
and hence improves the overall Sotation perfor- Sotation de-inking has evolved from removing ink
mance. particles only to removing an ever-increasing variety
Figure 1 Flow diagram of a typical flotation de-inking mill for mixed paper.
Table 2 Historical evolution of flotation de-inking cells
Year Flotation cell name Comments Year Flotation cell name Comments
1952 Denver cell First commercial de-inking 1986 Lamort DA Verticel Double aeration
flotation cell
1959 Voith paddle cell First cell designed 1987 Beloit PDM cell First cell to use pressure to
specifically for de-inking improve efficiency and
with improved air mixing pressurized rejects removal
First cell totally sealed
1972 Escher Wyss First cell to use compressed 1990 Voith elliptical cell Tubular shape compressed from
FZ-1 cell air cylindrical to elliptical to
promote faster air removal
1975 Swemac Hellberg First cell without agitator
cell and with controlled air bubble
sizes
First cylindrical cell 1990 Eshcer Wyss Designed for speck removal
CFS cell
1976 Escher Wyess No agitator compressed air 1991 Black Clawson Turbine acts as an agitator
FZ-U cell enters bottom of cell IHI/BC cell and distributes air through
cell
1978 Lamort Verticel First cell with vacuum to 1991 Kamyr gas- Cyclone body with porous
remove rejects sparged cyclone media
First cell with multi-feed 1992 Eshcer Wyss Stacked cell units
and aeration stages with CFC cell
no agitator
1981 Black Clawson First cell with gravity feed 1993 Comer Spidercel First cell with agitation
ULTRACELL between stages provided by reactor
blades
1981 Voith tubular First tubular cell with air 1994 Thermo/Black Similar to Lamort Veticel
injector cell drawn through injectors Clawson Verticel with multiple aeration
Mac cell closed
1981 Shinhama HIFLO Hydrocyclone body 1995 Voith Sulzer Similar to Voith elliptical
FLOTATORS EcoCell cell with Escher CF cell
air injector
1983 Beloit Lineacell First cell with separate 1996 Kvaerner Hymac First cell to use two types
aeration, mixing and column flotation of air spargers to provide
separation stages cell different size air bubbles
1984 Esher Wyss CF Step diffuser improves
mixing
of objectionable, noncellulosic materials. However, and soaps used as ink collectors are only effective at
in terms of ink removal efRciency, older Sotation cells alkaline pHs. Neutral pH Sotation de-inking requires
perform satisfactorily. the use of more effective and selective nonionic sur-
factants.
Factors Affecting Pulping and Flotation
Temperature Pulping and Sotation temperatures
The performance of Sotation de-inking is affected by
are typically maintained at 40}553C and 35}453C,
many factors such as pH, consistency, temperature,
respectively. In general, elevated temperatures are
ink/Rbre particle size, chemical types, water hardness
used to improve the deRbrization of special waste-
and air bubble size. Properly controlling these factors
paper such as wet strength paper. Lower temperature
ensures the efRcient removal of ink.
is beneRcial to high stickies-containing wastepaper,
such as magazines.
pH pH signiRcantly affects both the pulping and
Sotation processes by altering the physical and chem- Consistency Consistency or dry Rbre per cent
ical properties of Rbres and inks. High pH helps weight in water has a direct impact on the ink particle
swelling of Rbres and promotes the detachment of size distribution. Depending on the types of waste-
inks from Rbre. However, if the pH is too high paper, different pulping consistencies are used. In
(e.g.'10), it may cause yellowing or darkening of newsprint mills, low-consistency (4}8%) pulping is
lignin-containing pulps. The conventional fatty acids generally used. However, in ofRce paper de-inking
mills, medium (8}12%) or high consistency To evaluate the de-inking efRciency, both the ink
(12}18%) pulping is widely applied. In general, the removal of different size particles and brightness
higher the pulping consistency and the longer the should be reported together with reject rate. For
pulping time, the smaller the ink particles liberated. newsprint mills, brightness and ERIC measurements
Typical Rbre consistency in Sotation operation is are usually employed and, for white grades, ink count
0.7}1.2%. measurement is very common.
Particle size Flotation is most effective for removing

De-Inking Chemicals and Recipes
ink particles ranging from 10 to 150 m. In general,
particles smaller than 10 m and larger than 150 m The use of chemicals is involved in almost every
cannot be efRciently removed by Sotation process. aspect of the key processes in de-inking. Chemistry is
of great importance in Sotation de-inking in terms of
Water hardness In newsprint Sotation, fatty acids Rbre swelling, ink detachment, dispersion, anti-re-
or soaps are used as ink collectors, and a moderate deposition, ink agglomeration, ink collection and re-
amount of calcium ions (100}300 p.p.m.) is required moval. Most of the chemicals used for de-inking are
to make the ink Soatable. No additional hardness or fairly standard commodity products, such as sodium
calcium ions are added when nonionic surfactants are hydroxide and sodium silicate. On the other hand,
used. some other chemicals are relatively complex and have
multiple functions. In general, most de-inking
chemicals are added in the pulper. Commonly used
Ash or Vller content There is a strong relationship
de-inking chemicals in pulping and Sotation, their
between the amount of clay or Rllers in the waste-
functions and addition points are listed in Table 3.
paper and the ink removed in a Sotation stage. Flota-
tion de-inking becomes much more effective when the Pulper Chemicals
wastepaper has signiRcant ash content. An 8}10%
ash is considered a minimum requirement, and The chemicals used in the pulper depend strongly on
12}14% is preferable. the types of wastepaper processed. The principal
chemicals used for pulping and Sotation are sodium
Evaluation of Flotation Performance hydroxide, sodium silicate, chelating agents, hydro-
gen peroxide, surfactants and solvents. The roles of
Three major parameters are widely used in the paper
these major de-inking chemicals are brieSy discussed
industry to characterize the Sotation deinking efR-
below.
ciency: brightness, ink removal and reject rate.
Brightness is the percentage of reSectance measure-
ment of pulp or paper products at a wavelength of Sodium hydroxide Sodium hydroxide is used to
457 nm. It is originally developed to evaluate bleach- promote Rbre swelling and to saponify or hydrolyse
ing efRciency. In general, pulp brightness increases as the ink resins by increasing pH and alkalinity. The
ink is removed. The brightness increase is usually in type and amount of alkali required in the pulper
the order of 10}l5 units for newsprint de-inking and depend on the type of mechanical treatment, temper-
of 5}10 units for white grades. ature and pulping time. However, the addition of
Reject rate is deRned as the mass rate of reject to excessive caustic soda to ground wood-containing
total stock fed into Sotation. A main objective of furnishes will cause the pulp to yellow or darken.
Sotation de-inking is to obtain the maximum ink This is termed as ‘alkali darkening or yellowing’.
removal at a minimum reject rate. For mixed ofRce Sodium carbonate is rarely used in modern de-inking
wastepaper (MOW), reject rate is commonly 10% mills.
and for ONP/OMG, it is about 15%.
Ink removal is calculated based on the ink concen- Sodium silicate Sodium silicate or water glass is
tration of the pulp before and after Sotation. Image frequently used in conjunction with sodium hydrox-
analysis techniques are used to measure ink particle ide, especially in the de-inking of groundwood pa-
size ('3 m), total counts and the total surface area. pers. It not only serves as an alkali to swell Rbre and
Effective residual ink concentration (ERIC) measure- as a dispersant of ink particles, but also buffers the
ment was developed to determine the visual effect of pulp to a pH range which is favourable to the action
residual inks on the de-inked pulp. The visual effect is of hydrogen peroxide.
primarily dependent on the presence of small size ink
particles ((3 m) rather than the total ink content Hydrogen peroxide Hydrogen peroxide is one of
of the paper. the most commonly used pulping chemicals in the
Table 3 Chemicals used in flotation de-inking and their functions
Chemical Structure/formula Function Furnish type Dosage Addition point

( % of fibre)
Sodium NaOH Fibre swelling } ink break-up All grades 0}5 Pulper
hydroxide Saponification
Ink dispersion
Sodium Na2SiO3 Wetting Groundwood 0.5}5 Pulper
silicate Ink dispersion grades
Peroxide stabilization Lightly inked
Alkalinity and buffering ledger
Sodium Na2CO3 Alkalinity Groundwood 0.25}5 Pulper
carbonate Buffering grades
Water softening Lightly inked
ledger
Hydrogen H2O2 Prevention of fibre yellowing Groundwood 0.5}2.5 Pulper
peroxide grades
Coloured
ledgers
Sodium Na2S2O4 Bleach, colour stripping Groundwood 0.5}1.5 Pulper
hydrosulfite grades
Coloured Pulp storage
ledgers
Sodium or Hexametaphosphate Metal ion sequestrant All grades 0.2}1 Pulper
potassium Tripolyphosphate Ink dispersion
phosphate Alkalinity and buffering
Detergency
Chelating EDTA Metal ion chelation All grades 0}0.5 Pulper
agents DTPA Peroxide stabilizer
Calcium CaCl2 Fatty acid/soap collector aid Groundwood 90}300 p.p.m. Flotation
ions grades Pulper
Hydrophilic Polyacrylate Ink anti-redeposition All grades 0.1}0.5 Pulper
polymers Carboxymethylcellulose Anti-redeposition
Nonionic Ethoxylated alcohol Ink collector
surfactants Ethoxylated alkyl phenols Flotation frother All grades 0.1}2 Pulper
Wetting Flotation
Emulsification
Fatty acids Stearic acid Ink collector All grades 0.5}3 Pulper
or soaps Oleic acid Flotation frother Flotation
Fatty acid mixtures
Solvents C1}C14 aliphatic Ink softening Wood-free 0.5}2 Pulper
saturated Solvation of wax grades
hydrocarbons
recycling of groundwood wastepaper. It is also widely paper recycling industry. Compounds like DTPA and
used as a bleaching chemical. The addition of hydro- EDTA have been banned in some countries, for
gen peroxide in the pulper is to offset the formation of example, Sweden and Norway.
chromophores created by high alkaline pH.
Dispersants Sodium tripolyphosphate and tetra-
Sodium hydrosulVte HydrosulRte is mainly used as sodium pyrophosphate are sometimes added to the
a reductive bleaching agent to bleach recycled pulp pulper to provide multiple functions such as ink dis-
and to decolourize the coloured Rbres. persion and metal chelating. Use of laundering anti-
redeposition agents such as carboxymethylcellulose
Chelating/sequestering agent Chelating compounds and sodium polyacrylate can also help disperse the
are commonly added in the pulper to form complexes ink particles, prevent redeposition of ink on the Rbre,
with multivalent metal ions to prevent peroxide and increase de-inked pulp brightness.
decomposition. Diethylenetriaminepentaacetic acid
(DTPA) and ethyelenediaminetetraacetic acid Solvents Organic solvents were once widely used
(EDTA) are the most common chelates used in the to dissolve waxes and varnishes, but environmental
concern has curtailed the use of these chemicals. ing. To function effectively as ink collectors, fatty
Solvents used in wastepaper deinking include C12}C14 acids require the presence of moderate concentration
hydrocarbons and glycol ethers. of calcium ions such as at least 12 degree German
hardness (dH) or approximately 200 p.p.m. as cal-
Flotation Chemicals
cium carbonate. The calcium ions can be sourced
Surfactants are probably the most important chem- from the paper Rllers such as calcium carbonate, or
icals in Sotation de-inking. They consist of two prin- from the addition of calcium chloride or oxide. To
cipal components } a hydrophilic component and maximize the function of any source of calcium ions,
a hydrophobic component. It is assumed that the it is important to maintain the Sotation in alkaline
hydrophilic end of the molecule attaches to the ink conditions ('8}8.5), otherwise fatty acids precipi-
particle, leaving the treated surface state hydropho- tate. However, the presence of excess calcium ions in
bic. Most surfactants used in Sotation de-inking play the system may cause Rbre loss.
two important roles. Firstly, they function as ink
collectors that selectively render the ink particle sur- Nonionic surfactants Nonionic surfactants en-
face more hydrophobic and facilitate ink particle}air compass a large number of synthetic chemicals of
bubble attachment. Secondly, they serve as Sotation varied types and structures. Major types of nonionic
frothers that generate moderate foaming. Surfactants surfactants include fatty ethoxylate, alkyl phenol
used in Sotation de-inking can be cationic, anionic, ethoxylate and fatty acid alkoxylate. Cloud point and
nonionic or amphoteric, and are added either at the hydrophilic/lipophilic balance (HLB) value are two
pulper or just before the Sotation cells. important terms used to describe a given nonionic
The most frequently used surfactants are fatty acids surfactant. Cloud point is the temperature at which
and their soaps, as well as nonionic surfactants. nonionic surfactants become separated from the solu-
Cationic surfactants are not currently used in Sota- tion. Below the cloud point, surfactant has higher
tion cells. Commonly used Sotation de-inking surfac- foaming and above the cloud point, the foaming of
tants and their formulas are shown in Table 4. surfactants decreases dramatically. HLB value is the
ratio of weight percentages of hydrophilic to hydro-
Fatty acids and soaps Fatty acids and their soaps are phobic groups in the structure. Generally, for the
early Sotation de-inking surfactants, and are com- same surfactant, the higher the HLB value, the higher
monly used in Europe than in North America. Mix- the foaming ability, and the lower the ink collection
tures of fatty acids with carbon chain lengths of ability. An effective nonionic Sotation surfactant usu-
16}18, such as stearic, oleic, palmitic and linoleic ally possesses the properties of good ink collection
acids, are commonly used as ink collectors. Saturated and adequate foaming.
fatty acid soaps usually have better ink collection Some of the most common nonionic surfactants
while unsaturated fatty acid soaps have higher foam- used in Sotation deinking are EO/PO copolymers, in
Table 4 Typical flotation de-inking surfactants
Types Name Formula and structure
Anionic Fatty acid and emulsion CH3(CH2)nCOOH

Fatty acid soap (stearic/palmitic/oleic acid and soap) CH3(CH2)nCOONa
CH3(CH2)nCOONa
Alkylbenzene sulfonate R}(C6H4)}SO3Na
Fatty alcohol sulfate R}OSO3Na
Fatty alcohol ether sulfate R}O}(CH2CH2O)nSO3Na
Cationic Dodecyltrimethylammonium bromide CH3(CH2)12NH4Br
Amphoteric Sulfobetaine RN#(CH3)2(CH2)xSO\

3
Nonionic Fatty alcohol ethoxylate R}O}(CH2CH2O)xH

Ethoxylated alkyl phenol R}(C6H4)}O}(CH2CH2O)xH, n"8}9
EO/PO copolymers HO}(EO)x }(PO)y }(EO)z }H
Fatty acid alkoxylate RCOO}(CH2CH2O)xH
Alkyl phosphate ester (RO(CH2CH2O)n)2POONa
Fuel oil CH3(CH2)nCH3
Fatty oil alkyleneoxide derivative
EO, Ethylene oxide; PO, propylene oxide.

which the hydrophilic part is EO (ethylene oxide) and ONP/OMG mixture Ash plays a very important
the hydrophobic part is PO (propylene oxide). Al- role in the Sotation de-inking of newsprint. It is
koxylates of fatty alcohols or fatty acids containing a common practice to include a certain per cent of old
both EO and PO units have been used as Sotation magazines (OMG) in old newsprint (ONP) for better
surfactants. Sotation de-inking efRcacy. Since OMG and mixed
As environmental regulations become tougher, de- ofRce waste (MOW) contain Rllers such as calcium
inking mills tend to prefer to use surfactants that are carbonate, the presence of these Rllers in the waste-
easily biodegradable. Because of the difRculty of paper promotes ink Sotation. This is because the
breaking down alkyl phenols in wastewaters, they are Rllers can provide the calcium ions needed by the
gradually being replaced by readily biodegradable fatty acids. Traditionally, the mixtures of ONP and
products such as alcohol ethoxylates. OMG, usually in the ratios of 70 : 30 and 50 : 50, are
de-inked by Sotation at an alkaline pH using fatty
Flotation Recipes
acids or soaps. A signiRcant improvement in ink re-
The selection of optimal Sotation chemistry recipes moval of newsprint can also be achieved by adding
depends strongly on both ink properties and types of clay to the pulper.
wastepaper. The Sotation recipes that are commonly
employed are summarized in Table 5. Mixed paper Most of the early plants used alkaline
conditions to de-ink mixed wastepaper. However,
Old newspapers Newsprint (100%) is often de-ink- there is a trend for modern de-inking mills to switch
ed by washing. However, Sotation can also be effec- to neutral conditions. Fatty acids are commonly used
tive to remove newsprint inks with fatty acid/calcium in Europe and Asia and nonionic surfactants alone
ion chemistry. In the absence of calcium ions, the ink are commonly used in North America for de-inking
particles do not Soat using fatty acid collectors. The mixed paper. However, fatty acids in combination
addition of calcium is indispensable for a fatty acid or with nonionic surfactants are found to be the most
a soap to function properly as a collector. Positively effective in removing both large and small ink par-
charged calcium ions are bonded to the fatty acid and ticles.
to the negatively charged ink particles, and thereby
promote ink Sotation. Calcium ions (often as calcium Flexographic inks Flexographic inks are water-
chloride) should be added to the pulp simultaneously based inks and are difRcult to de-ink using Sotation
or before the fatty acid at a level of above due to the hydrophilic nature and very small size
100}150 p.p.m. as calcium carbonate. ((5 m) of the ink particles. Flexographic inks tend
Table 5 Typical flotation recipes of waste paper de-inking
Medium Pulping Flotation
Alkali and bleaching Chelate or pH Collector and frother Furnishes suitable for the
agent dispersant flotation chemistry
Alkaline de-inking NaOH EDTA 8.5}11.5 Fatty acids or soaps with 100% ONP, 100% flexographic
Na2SiO3 DTPA calcium ions ONP/OMG
H2O2 Phosphate Fatty acids or soaps ONP/OMG, ONP/SOW, OMG
trimming/MOW
Fatty acids or soaps and ONP/OMG, ONP/SOW, OMG
nonionic surfactants trimming/MOW
Nonionic surfactants Sorted ledger, MOW, CPO
Fuel oil and nonionic 100% flexographic ONP
surfactants and OMG
Neutral de-inking NaOH (optional) EDTA 5.5}8.0 Fatty acids or soaps MOW, sorted ledger, OMG
DTPA (optional) Fatty acids or soaps trimming/MOW
and nonionic surfactants MOW, sorted ledger,
OMG/MOW, CPO, manifold
Nonionic surfactants MOW, sorted ledger,
OMG/MOW, toners
Fuel oil and nonionic Flexographic ONP/OMG
surfactants
For abbreviations, see Table 1.

to redeposit on the Rbres and may cause a dramatic been conducted to use enzymes to improve de-inking
brightness drop of the pulp. Similar to 100% news- efRciency. The enzymes used included primarily
print, fatty acid soaps and calcium ions are found to cellulases, hemicellulases, amylase, lipase or resinase.
be effective in removing Sexographic inks. Commercial application of enzymes to the Sotation
de-inking of wastepapers showed enhanced ink re-
Toner inks Xeroxgraphic paper and laser computer moval efRciency. Neutral pH de-inking further
printout (LCPO) are frequently found in government beneRts the use of enzymes since it can improve
publications and general ofRce waste. The inks in ink detachment from the Rbres and repulping efR-
these papers are thermoplastic powders or toners, ciency. Since most enzymes work at acidic pH
and are Rrmly bonded to the Rbres with a heat and lower temperature, enzyme manufacturers
fusion printing process. The toner ink particles are have to develop thermophilic and alkaline-stable
hydrophobic, but their removal efRciency by Sotation enzymes to lower usage and enhance their effec-
is poor compared with conventional inks since the tiveness.
toner inks are large in size and cannot be sufRciently
detached from Rbres in the pulper. Effective removal
of toner inks can be realized using high consistency Further Reading
mechanical dispersion of the pulp with kneader and Borchardt JK (1997) An introduction to deinking chem-
disperger followed by Sotation using nonionic istry. In: Doshi MR and Dyer JM (eds) Paper Recycling
surfactants. Kneader and disperger assist Sotation Challenge, vol. II. Deinking and Bleaching, pp. 18}30.
by improving toner}Rbre detachment and by re- Appleton, Wisconsin: Doshi.
ducing the ink size distribution to a more Soatable Borchardt JK and Ferguson LD (1998) Deinking chemistry.
range. In: Doshi RM and Dyer JM (eds) Paper Recycling Chal-
lenge, vol. III. Process Technology, pp. 71}81. Apple-
ton, Wisconsin: Doshi.
Future Trends Ferguson LD (1992) Deinking chemistry: part 1. Tappi
Flotation de-inking is a complex separation process Journal 75(7): 75}83.
Ferguson LD (1992) Deinking chemistry: part 2. Tappi
of inks and other contaminants from Rbres. Due to
Journal 75(8): 49}58.
economical reasons and strict environmental regula- Hines PR (1933) US patent no. 2,005,742.
tions, new materials used by paper manufacturers Jelks JW (1954) Development in Sotation deinking of waste
and new printing technologies, paper mills require paper. Tappi Journal 37(10): 176A}180A.
environmentally benign de-inking technologies which Larsson A, Stenius P and OG dberg L (1984) Surface
can easily Rt into the current de-inking system with- chemistry in Sotation deinking, part 2. The importance
out extra capital investment. of ink particle size. Svensk Papperstidning 87(18):
In recent years, great progress has been made in R165}R169.
Sotation de-inking technologies with respect to Sota- McCool M (1993) Flotation deinking. In: Spangenberg RJ
tion cell design, utilization of new surfactants and the (ed.) Secondary Fiber Recycling, pp. 141}162. Atlanta,
understanding of de-inking chemistry. However, the Georgia: TAPPI Press.
McKinney R (1998) Flotation deinking overview. In:
rapid advances in printing, coating and other modiR-
Doshi RM and Dyer JM (eds), Paper Recycling Chal-
cations of paper make de-inking more difRcult. More lenge, vol. III. Process Technology, pp. 99}114.
effort is needed to understand the Sotation behaviour Appleton, Wisconsin: Doshi.
of new types of wastepapers. Shrinath A, Szewczak JT and Bowen IJ (1991) A review of
Flotation chemistry plays the most important role ink-removal techniques in current deinking technology.
in determining the ink removal efRciency. Neutral Tappi Journal 74(7): 85}93.
Sotation de-inking has become increasingly popular Smook GA (1994) Handbook for Pulp and Paper Techno-
in the last 10 years. This is mainly because neutral logists, 2nd edn. Vancouver: Angus Wilde.
de-inking has great potential to lower chemical usage Somasundaran P and Zhang L (1998) Fundamentals of
and cost, to reduce water treatment cost, to improve Sotation deinking. In: Doshi RM and Dyer JM (eds),
product quality and paper machine runnability. Mills Paper Recycling Challenge, vol. III. Process Technology,
pp. 83}98. Appleton, Wisconsin: Doshi.
will beneRt from switching from alkaline to neutral
Welt T and Dinus RJ (1997) Enzymatic deinking. In: Doshi
Sotation de-inking. Since no caustic or silicate is ad- MR and Dyer JM (eds), Paper Recycling Challenge, vol.
ded in the pulper, Rbres are not yellowed or darkened. II. Deinking and Bleaching, pp. 235}246. Appleton,
As a result, bleaching chemicals such as peroxide may Wisconsin: Doshi.
not be required. Woodward TM (1986) Appropriate chemical additives are
Enzymatic de-inking represents a new approach to key to improved deinking operations. Pulp Paper
modern paper-recycling mills. Extensive research has 60(11): 59.
III / DEOXYRIBONUCLEIC ACID PROFILING / Overview 2545
DEOXYRIBONUCLEIC ACID PROFILING
in their sequence. However, a more extensive and

Overview DNA-speciRc polymorphism exists in the noncoding
DNA: the tandemly repeated sequences. These poly-
morphic loci are DNA fragments, made up of a nu-
R. Coquoz, Institut de Police Scientifique et de
Criminologie, Lausanne, Switzerland
cleotide sequence (the repeat unit), tandemly repeated
like a series of identical beads on a necklace (Fig-
Copyright ^ 2000 Academic Press ure 1). The repeat unit can be from two to more than
This article is reproduced from Encyclopedia of Analyti-
cal Science, Copyright ^ 1995 Academic Press 100 nucleotides long. The polymorphism comes from
the extreme variation in the number of repeats. The
alleles differ then by their size (length polymor-
Introduction phism). This kind of repetitive sequence has been
The forensic biologist has traditionally used a large called minisatellites or also variable number tandem
set of markers including blood group antigens and repeat (VNTR). There is a huge reservoir of polymor-
serum proteins to establish links between evidence phism in these sequences and it is their potential use
and comparison samples. These markers are poly- which has stimulated the development of the various
morphic and genetically determined. In 1985, a new DNA typing procedures.
technique was introduced which opened the door
to the forensic use of deoxyribonucleic acid (DNA) DNA Typing Procedures
technology: its inventor coined the term ‘DNA Rnger-
printing’. Since then, DNA typing in this and many Any biological material containing DNA may be use-
other forms has Sourished and the forensic biologist ful, including blood, sperm, saliva, hair, autopsy tis-
now has access to a large variety of powerful sue, bone. All DNA typing procedures involve a DNA
techniques. isolation step, which allows the elimination of sub-
stances such as proteins which can interfere in the
sometimes delicate enzymatic steps involved in any
DNA and its Polymorphism DNA typing method.
The traditional way to prepare samples is to dis-
DNA is an extremely large linear molecule which
solve them in a sodium dodecylsulfate (SDS) pro-
stores genetic information in the form of a sequence
teinase K solution which lyses the cells and digests
of its constituent elements: the nucleotides. The DNA
the proteins. The samples are then extracted with a
molecule is made up of two complementary chains
phenol/chloroform mixture and the DNA is Rnally
or strands of nucleotides (Figure 1). There are four
ethanol-precipitated and dissolved in the adequate
different nucleotides (symbolized by the letters
buffer. Polymerase chain reaction (PCR) analysis
A, C, T, G) whose chemical afRnity determines
requires minimal amounts of DNA so that it is
the complementarity of the nucleotide sequences
possible to start with samples so small that they are
of the two strands of the DNA molecule. Nucleo-
unlikely to contain substantial amounts of interfering
tides A and T, and G and C always face each other;
molecules. Simple boiling of the samples in the pres-
as a consequence, the length of DNA fragments is
ence of a heavy-metal chelator (Chelex resin) is then
usually expressed in ‘base pair’ (bp) units. This chem-
usually sufRcient as a preparation method.
ical afRnity allows single strands of complement-
ary DNA to hybridize to each other. Nucleotide
chains are oriented, having so called 5 and 3 ends,
and the two strands of DNA have an opposite ori-
entation. The DNA synthesis occurs in the direction
5P3.
The polymorphism within DNA resides partly in
the existence of minor nucleotide sequence variations
among different individuals at various locations
(loci) in the DNA molecule, which has its counterpart
in the polymorphism of the amino acid sequence of Figure 1 Schematic description at four increasing levels of
the proteins. The gene variants (the alleles) differ magnification for DNA and VNTRs.
2546 III / DEOXYRIBONUCLEIC ACID PROFILING / Overview
Restriction Fragment Length Polymorphism (RFLP) The probe can be designed so as to detect a single
Typing VNTR locus; it is then a single locus probe (SLP). If
the individual has inherited DNA from his parents
The isolated DNA is reproducively cut into fragments with a different number of repeats at the ana-
by the action of a sequence speciRc DNA cutting lysed VNTR, the result is in the form of two bands
enzyme, a restriction enzyme (hence RFLP). The frag- (heterozygote); otherwise, there will be a single band
ments, which can be more than 20 000 bp long, (homozygote). But since the repeat units of dif-
are then separated according to their sizes by elec- ferent VNTRs often have sequence similarities to
trophoresis on agarose gel. The DNA fragments are each other, it is possible to detect a whole family of
transferred to a nylon membrane to which they will VNTRs at the same time provided a probe is used
remain attached but are still accessible for hybridiza- under conditions allowing hybridization with par-
tion with a probe. This membrane will provide a rep- tially complementary DNA fragments. In this case we
lica of the electrophoresis product. Labelled pieces of speak of a multilocus probe (MLP) and the result
DNA having a sequence complementary to the repeti- appears in the form of a set of bands with a very
tive sequence to be detected (the probe) are added to individual pattern similar to a bar code.
the nylon membrane and hybridize to the appropriate
DNA fragments. These can Rnally be detected by Polymerase Chain Reaction (PCR)
autoradiography with a Rlm sensitive to the labelled The advent of the PCR has dramatically enlarged the
probe (Figure 2). The DNA fragments to be detected spectrum of methods available for DNA typing. The
will appear on the Rlm as dark bands (Figures 2 and PCR is a DNA ampliRcation method which cyclically
3). Their position will be a measure of their size, reproduces the natural DNA replication. Each cycle
which is itself proportional to the number of repeti- consists of (1) a denaturation step where the two
tive elements in the VNTR. A comparison with DNA DNA strands are separated by heating, (2) an anneal-
fragments of known sizes allows an accurate size ing step where, after cooling, short synthetic DNA
determination. This Rrst probe can be stripped from strands (the primers) are hybridized on both sides of
the membrane allowing for reprobing with a second the DNA fragment to be ampliRed, and (3) an elonga-
probe, etc. There is a large choice of VNTRs available tion step where a DNA polymerase adds nucleotides
for RFLP typing (Table 1). It is the same for restric- to the primers to synthesize the DNA strands com-
tion enzymes: but a few have found wider use because plementary to the template DNA. The result of many
of their robustness and their ability to cut DNA cycles is millions of copies of DNA fragment de-
into small fragments. These enzymes are Hae III, limited by the primers. The use of a heat-stable poly-
Hinf I, Pst I, Alu I and Pvu II, the Rrst two being the merase has rendered the whole process very easy and
most used. The labelling of the probe is traditionally allowed it to be automated. The nature of a PCR
done through the use of 32P-labelled nucleotides. makes it naturally extremely sensitive, with a theoret-
However, probes coupled to enzymes catalysing the ical detection threshold of one molecule. The primers
production of a chemiluminescent substance allow allow the ampliRcation to be highly speciRc and the
detection limits as low as those obtained by radioac- ampliRcation product can then frequently be detected
tive means. using a simple nonspeciRc detection method, which
Figure 2 Schematic description of the main steps in RFLP typing.

Rc probes, using a reverse dot}blot format (Figure 4):

the various probes are attached to a membrane strip,
arranged as a series of invisible dots to which the
ampliRcation products can hybridize depending upon
thier sequence. Through the use of labelled nucleo-
tides during the PCR, the genotype of the sample can
be read through the appearance or not of coloured
dots on the strip, at the position of each probe (Fig-
ure 4). In a way, the format is not very different
from that of the strips for urine analysis so widely
used in clinical medicine.
Ampli\cation Fragment Length Polymorphism
(AMP-FLP) Typing
Another use of a PCR has been the ampliRcation of
VNTRs to analyse their length polymorphism. How-
ever, the quite poor performance of PCRs in the
ampliRcation of large DNA fragments, such as those
of the VNTR loci analysed through RFLP typing, has
led to the search for smaller VNTRs (Table 1). By
designing primers located on each side of a VNTR,
it can be ampliRed by a PCR and the ampliRed frag-
Figure 3 Typical examples of RFLP typing of forensic cases. ments can be analysed by electrophoresis on agarose
On the left: sexual assault case; St, size markers, V, victim, VS, or polyacrylamide gels and compared to standards
vaginal swab (containing material from the victim and her assail- (Figure 4). While RFLP typing required the use of
ant), S1, suspect 1, S2, suspect 2. On the right: paternity test with
probes to detect speciRc VNTR-DNA fragments
M, mother, C, child, F, putative father.
among thousands of other fragments, the ampliRca-
tion process here allows the use of nonspeciRc detec-
makes the whole analytical process convenient and tion methods (ethidium bromide staining, silver stain-
rapid. ing) to make the bands visible. This kind of PCR-
analysed VNTR polymorphism has been called ampli-
PCR and Dot+Blot Analysis Rcation fragment length polymorphism (AMP-FLP).
Many methodologies have been designed which make
Short Tandem Repeats (STR) Typing
use of the main principle of a PCR. In forensic
science, PCR was Rrst used to analyse sequence poly- The particularly high efRciency of PCR in the
morphisms such as the polymorphism of the HLA- ampliRcation of very small fragments has attracted
DQ locus. In that case, the alleles, which differ attention toward a class of VNTRs made of very
only by a few nucleotide changes, are identiRed after small repeats (di-, tri-, tetranucleotide repeats)
use of a PCR through hybridization with allele-speci- (Table 1). They are sometimes called microsatellites
Table 1 Selection of some of the main DNA loci analysed by the various DNA typing methods
Locus Corresponding Chromosome Repeat unit length Typing method used Heterozygosity b (%)
probe location (bp)
* 33.6 1a 37a multilocus RFLP *

D1S 7 MS1 1 9 monolocus RFLP '99
D7S21 MS31 7 20 monolocus RFLP 98
D2S44 YNH24 2 31 monolocus RFLP 94
D10S28 TBQ7 10 33 monolocus RFLP 97
HLA-DQ * 6 * PCR#dot blotting 79
D1S80 MCT 118 1 16 AMP-FLP 80
D17S5 YNZ22 17 70 AMP-FLP 82
APOC2 Mdf 5 19 2 STR 80
HUMTH01 * 11 4 STR 80
a
These characteristics refer to the VNTR used as the multilocus probe.
b
Heterozygosity is calculated here as the percentage of apparent heterozygotes in the population (Caucasian in this case).
Figure 4 A PCR amplification for AMP-FLP typing (left) and reverse dotIblot analysis (right).
or short tandem repeats (STRs) and they are probably (Figure 5). The PCR is carried out in parallel in two
the richest class of polymorphic loci available. Their tubes in a mixture containing a primer complement-
ampliRcation conditions are simple and the size range ary to a DNA sequence outside the VNTR and a low
of their alleles is usually narrow. They are then well concentration of primer complementary to the repeat
suited to multiplexing: it is indeed easy to choose unit variant a. This set of two primers allows the
microsatellite loci and adequate primers so as to ob- multiplication of an array of all the possible fragment
tain ampliRed DNA fragments situated in well separ- going from one end of the VNTR to every repeat a. In
ated size ranges for each locus. Consequently a whole the second tube, the use of a primer complementary
group of STRs can be ampliRed at the same time to the repeat unit variant b allows the multiplication
and analysed on the same gel. The small size of their of another array of all the possible fragments going
repeat unit, however, is also a disadvantage. The from one end of the VNTR to every repeat b. An
various alleles only differ in length by two, three unequal ampliRcation of the short versus long frag-
or four nucleotides. Such a separation power is only ments is avoided through the use of a tag attached to
possible through the more complex technology of the two alternative primers. This tag is a short strand
sequencing gel electrophoresis. However, the se- of DNA. A fourth primer complementary to this tag is
quencing technology is undergoing constant improve- then used to allow the further ampliRcation of the
ment due to the challenge of the human genome arrays of fragments. The content of the two tubes is
sequencing project, and STR typing has beneRted then loaded on two adjacent lanes of an agarose gel.
from the developments in that Reld such as Suor- After electrophoresis, Southern blotting, hybridiza-
escent labelling and automatic sequencing. tion with a probe for the VNTR and autoradiogra-
phy, the result appears in the form of two parallel
Minisatellite Variant Repeat (MVR)-PCR
ladders of bands with fainter or missing rungs. For an
The polymerase chain reaction has made possible yet individual, the result is indeed the sum of the results
another powerful DNA typing method: MVR-PCR. given by two VNTR fragments, those inherited from
This has been designed to reveal a polymorphism the mother and the father: at each repeat position in
consisting of variations in the sequence of the repeat the VNTR the subject can have variant a twice, vari-
unit within some VNTRs. Instead of being a pure ant b twice, or a once and b once. These three possi-
repetition of one repeat unit, some VNTRs are com- bilities can easily be coded by the numbers 1, 2, 3,
posed of combinations of two or more almost identi- allowing a simple digitization of the result (Figure 5).
cal repeat units. It is easy to calculate that even with It is clear that the existence of more than two repeat
only two repeat variants a and b, the number of variants makes the coding more complex and the
possible combinations of repeats is huge. The se- system more informative but the principle remains
quence of repeats is revealed through a process some- the same. With MVR-PCR, there is no need to have
hat similar to the Sanger DNA sequencing method carefully standardized electrophoretic conditions. Each
Figure 5 Schematic description of MVR-PCR.
result is literally read by itself, and different samples polymorphic base pair, a ligation assay is performed
are simply compared at the level of their codes. using an antigen-labelled primer and one of two
biotin-labelled primers each being speciRc for one of
Other Methods
the two possible alleles. In the presence of the cor-
Other DNA typing methods have been developed or responding sample allele, the biotin-labelled primer is
proposed, and still others are yet to appear. Two ligated to the antigen-labelled primer. After capture
methods are worth mentioning. One is mitochondrial of the product in microtitre plates coated with strep-
DNA (mtDNA) sequence analysis. A DNA sequence tavidin (a binding protein for biotin), the presence
analysis after PCR ampliRcation is a natural way to of the antigen can be detected using a traditional
examine DNA polymorphism. However, because it is immunoenzymatic method (ELISA procedure). The
unable to reveal more than a few hundred base pairs combined analysis of more than 15 such two-allele
at a time, and because of its complexity, it is not as sites can reveal genotypes with frequencies below
attractive as the above-mentioned methods. How- 10\6. The most interesting features of this method
ever, it has been a precious tool in the analysis of are its technical simplicity and its straightforward
mitochondrial DNA, which is a small piece of DNA, interpretation (positive or negative readout) which
maternally inherited, present in all mitochondria. make it particularly amenable to automation. It could
Part of it (the D-loop) contains a highly polymorphic reach its highest performance if all the loci could be
sequence and its sequence determination is therefore analysed simultaneously within the same tube.
very informative. Owing to the presence of more than
1000 copies in each cell. mtDNA analysis is very
sensitive. Moreover, hair shafts, which do not contain
Applications
nuclear DNA, are rich in mtDNA and are therefore Deoxyribonucleic acid typing gained prominence
amenable to genetic analysis. from the start because of its identiRcation power. It is
The second method is polymorphic sequence- the most powerful identiRcation technique after Rn-
tagged sites (pSTS) analysis. This is a method de- gerprint examination. It represents a big jump from
signed to reveal single base pair changes. After a PCR the traditional protein polymorphism analysis where
ampliRcation of the DNA fragment containing the powerful identiRcation could only be reached by the
sequential analysis of more than 20 polymorphic sys- respective allele frequencies). The frequency of the
tems, each requiring its own analytical process. The combined genotypes at four loci is then 2a1a2;
improvement is particularly evident in semen analysis 2b1b2;2c1c2;2d1d2. However, these multiplications
where the number of available polymorphic systems can only be done under the assumption of statistical
was extremely low. In RFLP typing, the analysed independence of the multiplied factors. This trans-
VNTR loci are so polymorphic that most of the lates into certain assumptions related to the genetic
possible genotypes have frequencies below 1%. It is independence of the analysed VNTR loci and to the
then easy to calculate that the combined analysis of genetic structure of the populations (Hardy-Weinberg
four such loci reveals genotypes with frequencies of equilibrium).
the order of 10\8. Both AMP-FLP typing and STR Controversy has arisen over the possibility of the
typing use smaller, less polymorphic and therefore existence of genetic substructuring within the popula-
less informative VNTRs, but a similarly high in- tions. Some geneticists expressed concern that the
formation yield can be reached by increasing the frequency of a genotype A1A2 might have an apparent
number of loci analysed. value using the general population database, but that
the true frequency might be much higher owing to the
Interpretation
high frequency of these speciRc alleles in a genetically
Forensic identiRcation is achieved through the dem- distinct population subgroup. At the extreme, if
onstration of matching biological characteristics be- alleles A1A2 only appear in this subpopulation where
tween evidence and comparison material. The match- they have a frequency of 50% each, the detection of
ing is straightforward when discrete and well-identiR- allele A1 in a sample would make the detection of
able alleles are analysed such as in the various allele A2 almost certain. Because of such effects, com-
methods derived from a PCR. However, in RFLP bined genotype frequencies could be underestimated
typing, a match is not always easy to establish. First by several orders of magnitude. The accumulating
there is the intrinsic difRculty of having to deal with data suggest, however, that there is not much reason
alleles which cannot each be differentiated by elec- to worry about such effects. The VNTR loci are
trophoresis. Indeed, because of the size range of the extremely polymorphic in every population studied
DNA fragments, the separation power of elec- and, for any given genotype constellation, there is an
trophoresis is insufRcient to separate alleles differing extremely low probability of having substantial fre-
in length by only one repeat unit. Two samples hav- quency differences when calculating with one popula-
ing bands at an identical position may in fact possess tion data or another. Moreover, within the major
different alleles. Secondly, there is the possibility that races, the frequency distributions are very similar
a DNA sample originating from the same person as between different populations. And Rnally, forensic
the comparison sample contains interfering molecu- laboratories use very conservative approaches which
les, leading to a different electrophoretic behaviour always favour the accused at each interpretation step.
and shifted band positions. This bandshifting can
Paternity Testing
only be demonstrated and evaluated through the use
of proper controls (internal standards). The identiRcation power of DNA typing has naturally
The next step in the interpretation is the evaluation been used in paternity testing (Figure 3). But here the
of the power of the evidence. It is usually expressed as question relates to the transmission of genetic mater-
a ratio of the probability of the evidence under the ial from one generation to the other. Mutations
hypothesis that the evidential material has the same which can cause apparent exclusions of true fathers
source as the comparison sample and the probability then have to be taken into account. The polymor-
of the evidence under the hypothesis that somebody phism of the VNTRs is linked to an exceptionally
else is the source of the evidence material. This ratio is high mutation rate. This mutation rate can be below
normally inversely proportional to the frequency of 10\4 to more than 0.05 mutations, per meiosis for the
the genetic characteristics detected in the adequate most polymorphic loci such as D1S7. This has to be
reference populaton. The determination of these fre- compared to the mutation rate of coding DNA se-
quencies is a difRcult task. The extreme variability of quences which is estimated to be around 10\5 for
the VNTR loci makes the gathering of genotype fre- a 1000 bp gene. The calculation of probability of
quency data unrealistic. Statistically trustworthy fre- paternity has to integrate these mutation rates. But it
quencies would require the analysis of a huge number remains true that the increased power of DNA typing
of individuals. It is thus necessary to rely on allele has made paternity testing much easier and has al-
frequencies to calculate the genotype frequencies by lowed the solution of complex situations, such as the
multiplication. The frequency of genotype A1A2 at unavailability of the putative father, which some-
locus A is the product 2a1a2 (with a1 and a2 being the times remained unsolved with traditional typing.
Practical Use a minielectrophoresis gel followed by nonspeciRc

DNA staining. The examination of the intensity of the
In the real world of forensic science, there are practi- staining and the average size of the DNA fragments
cal contingencies which can be the source of difRcul- allows both quantitative and qualitative evaluation of
ties or even interpretation errors. An important con- the available DNA. The quality of the material can
tingency is the quantity of the evidential material. also affect RFLP typing through the inhibition of the
Any living cell contains about 6 pg of DNA, and the restriction enzyme. One of the expected DNA frag-
main forensic DNA sources contain about 30 ng ments may appear at a position corresponding to a lar-
L\1 (blood), 400 ng L\1 (semen), 2 ng L\1 (saliva) ger size than it should, because the enzyme has failed
and 0}250 ng per hair root. As indicated in Table 2. to cut the restriction site closest to the VNTR efRciently.
RFLP typing would require at least a few microlitres of The experienced analyst usually has no problems
blood and the various PCR methods require about avoiding interpretation errors arising from the above-
0.05 L. The theoretical PCR sensitivity of one mol- mentioned situations or others (extra bands due to
ecule would suggest that less than 1 nL of blood should unsufRciently stringent hybridization, incomplete
be sufRcient. However, this does not take into account stripping, internal restriction site or bacterial con-
the potential loss during DNA isolation, or, more tamination, missing bands due to blotting problems,
importantly, that one of the alleles may be absent or etc.). However, without extensive controls, it may be
underepresented in small samples containing only difRcult to prove the tentative explanation. And the
a few molecules, for simple stochastic reasons. As limited amount of evidence material usually pre-
a consequence, it will not appear at the end of the cludes repeating an analysis.
process: a heterozygote will appear as a homozygote. With PCR-derived methods, there are a few speciRc
In other words, there may be an allele drop-out. potential sources of error worth mentioning. The Rrst
Another important factor is the quality of the ma- is contamination. A PCR is so senstitive that inadver-
terial. Although DNA is a robust molecule and usable tent transfer of DNA from one sample to another,
DNA fragments have been recovered from thousand from laboratory personnel to the samples, etc., can
year old mummiRed bodies, it can very quickly be happen and be unnoticed. The large-scale use of nega-
degraded into smaller fragments when exposed to tive controls should reveal any such contamination.
humidity, heat or sunlight. With such degraded sam- Moreover, the establishment of adequate laboratory
ples, DNA typing methods like STR typing, which design and use of pertinent working guidelines and
analyse very short DNA fragments, may still be ap- decontamination procedures should prevent the oc-
plied successfully. However, RFLP typing may be no currence of such damaging episodes. The second
longer possible. Or even worse, a DNA sample pos- source of error is differential ampliRcation. The trust-
sessing a large and a small allele at a VNTR locus may ful ampliRcation of a heterozygote requires that the
have larger fragments which are too degraded to be two alleles are ampliRed with the same efRciency,
detected while the smaller ones will have survived otherwise the result is an allele drop-out. Conse-
enough to give a detectable band. Here again, the quently, the conditions which lead to such differential
result is an allele drop-out. That is why, in the usual ampliRcation have to be established and the proper
procedure, a small portion of the sample is loaded on precautions must be taken.
Table 2 Main characteristics of the various techniques available for DNA typing
Amount of genomic Ability to use Probability of matching Degree of automation

DNA required damaged DNA per locus analysed a possible
RFLP typing with single locus probes '20 ng ! (1% !

RFLP typing with multilocus probes '500 ng ! (10\7b !
PCR#dot}blot analysis
(HLA-DQ typing) (1 ng ## 7% #
AMP-FLP typing (1 ng ## 5}10% ##
STR typing (1 ng ### 2}10% ###
MVR-PCR '1 ng # (10\6c #
mtDNA sequencing ;1 ng ## 10\4c ##
pSTS analysis (1 ng ### 40% ####
a
The probability of matching (Pm ) is the probability that two individuals taken at random have the same genotype. Pm "ni"1 P 2i
where P i is the frequency of the genotype for the analysed locus.
b
Pm is that for the whole set of loci detected by this probe.
c
These numbers are estimates based on limited population samples.
2552 III / DEOXYRIBONUCLEIC ACID PROFILING / Capillary Electrophoresis
The potent identiRcation power of DNA typing See Colour Plate 75.
makes the establishment of databases for the storage See also: II/Electrophoresis: Blotting; Deoxyribonucleic
of DNA proRles possible and desirable. New proRles Acid, Theory of Techniques for Separation; One-dimen-
can be compared with stored proRles to identify crim- sional Polyacrylamide Gel Electrophoresis; One-dimen-
inals, link unsolved cases, etc. This necessitates a high sional Sodium Dodecyl Sulphate Polyacrylamide Gel
level of standardization among the laboratories parti- Electrophoresis.
cipating in the information exchange networks. They
must at least analyse the same polymorphic loci and, Further Reading
for RFLP typing, they must use the same enzymes and Ballantyne J, Sensabaugh G and Witkowski J (1989) DNA
similar analytical protocols. Because of the large Technology and Forensic Science, Banbury report 32.
choice of polymorphic loci and analytical methods, Cold Spring Harbor: Cold Spring Harbor Laboratory
Press.
the minimum degree of coherence is not at all war-
Burke T, Dolf G, Jeffreys AJ and Wolff R (eds) (1991) DNA
ranted unless a strong effort is made between the Fingerprinting: Approaches and Applications. Basel: Bi-
laboratories to reach a consensus. And a consensus is rkhauser Verlag.
difRcult to reach in a fast moving Reld where new DNA Fingerprinting (1991) Symposium paper. Elec-
technologies are constantly being developed. trophoresis 12: 101}232.
There is no doubt that DNA typing is the ‘safest’ Farley MA and Harrington JJ (1991) Forensic DNA Tech-
identiRcation technique ever used in forensic biology. nology. Chelsea, MI: Lewis Publishers.
But, because of its power and its consequent impact Kirby LT (1990) DNA Fingerprinting: An Introduction.
in a courtroom, it is and will remain under intense New York: Stockholm Press.
Proceedings of the Second International Symposium on
scrutiny by the forensic and legal communities. Con-
Human IdentiTcation (1991) Madison. WI: Promega
sequently, a lot of attention has been paid to quality Corporation.
control and proRciency testing. There is certainly Rittner C and Schneider PM (eds) (1992) Advances in
the opportunity for future improvements and devel- Forensic Haemogenetics, vol. 4. Berlin, Heidelberg:
opments in this Reld. Springer-Verlag.
Capillary Electrophoresis
M. Chiari and L. Ceriotti, Institute of Biocatalysis and base resolution for fragments ranging from 15 to
Molecular Recognition, CNR, Milan, Italy more than 500 bases were reported), the difRculties of
Copyright ^ 2000 Academic Press producing adequate gel-Rlled capillaries hindered
their greater application. Heiger et al. proposed, as
Introduction early as 1990, the use of polyacrylamide cross-linked
with a very low or zero concentration of N,N-methyl-
One of the earliest reports on DNA capillary elec- enebisacrylamide. Strangely enough, the revolution
trophoresis (CE) was by Cohen et al. who demon- brought about by the use of polymer solutions as
strated the high resolution of nucleosides and oligo- sieving matrices was only evident in 1991 as a result
nucleotides, in the absence of a sieving gel, by simply of the work of Guttman and Cooke and Grossman
trapping the analyte into sodium dodecyl sulfate and Soane. Since then, hundreds of reports have dem-
(SDS) micelles. One year later, in 1988, the same onstrated the advantages of performing DNA separ-
author discussed the separation of DNA restriction ations in a narrow fused silica capillary. Due to its
fragments in a gel-free environment. The Rrst reports high resolving power and quantitative capability, CE
on the use of polyacrylamide gel capillary columns to has been successfully applied to different kinds of
separate DNA with remarkable efRciency and resolu- DNA analysis, including the following: DNA se-
tion were presented the same year by Guttman et al. quencing, separation of restriction fragments, poly-
and Cohen et al. By 1989, gel-Rlled CE was already merase chain reaction (PCR) products and synthetic
a well-established technique. These preliminary re- oligonucleotides.
ports were then followed by a Surry of articles de-
scribing applications of capillary gel electrophoresis Advantages of Capillary over Slab Gel
(CGE) to the separation of DNA restriction frag-
ments. In spite of the high resolving power and
Electrophoresis
efRciency offered by CGE in the analysis of nucleic For many years electrophoretic separations of
acids (15}30 million plates per metre and a single- DNA were carried out in slab gels. One of the main
III / DEOXYRIBONUCLEIC ACID PROFILING / Capillary Electrophoresis 2553
Table 1 Typical field strengths applied in different DNA elec- the capillary with the detector located at the anode
trophoretic systems and analysis time end. In hydrodynamic injection, the capillary is
Separation system Typical field Time placed in the sample vial and a sample plug is forced
strength (V cm\1) to enter into the capillary by applying, for a given
amount of time, a differential pressure across the
Agarose slab gel 5 Hours
capillary. Both injection systems present advantages
Conventional polyacrylamide
slab gel 8 Hours and disadvantages. Hydrodynamic injection avoids
Sequencing in thin bias in the amount of the sample injected with the
polyacrylamide DNA slab gel 40 Hours faster analytes injected in greater quantities and is
Capillary electrophoresis 300}700 Minutes or more suitable for quantitation.
seconds
DNA Sieving Matrices

problems in electrophoresis is the dissipation of Joule
heat created by the passage of current through the gel At neutral pH, DNA molecules being negatively
and the buffer solution. When the gel is not sufR- charged migrate towards the anode when placed in
ciently and uniformly cooled, signiRcant Joule heat- a potential Reld. Their electrophoretic mobility is
ing occurs and temperature gradients are formed in given by this simple expression:
the gel. The existence of a thermal gradient within the q
gel can cause band broadening and, eventually, DNA e" [1]
f
denaturation. An efRcient way to overcome this prob-
lem is to carry out the separation in a fused silica where q is the net charge on the molecule and f is the
capillary. At present, a typical capillary used for DNA molecular friction coefRcient. DNA molecules in free
electrophoresis has a 50}100 m internal diameter solution adopt an open, extended conformation that
(i.d.), a fused silica wall about 150 m thick and allows solvent molecules to stream around each seg-
a thin exterior coating (10 m) of polyimide to pro- ment of the biopolymer equally. This implies that
vide Sexibility. The high surface-to-volume ratio of electrophoretic mobility of double-stranded DNA in
the capillary leads to an efRcient dissipation of Joule free solution is independent of molecular size as both
heat, which eliminates thermal and gravitational con- the net charge on the DNA molecule and its frictional
vection and allows the use of electric Relds up to coefRcient increase linearly with the number of base
700 V cm\1. Table 1 shows typical values of Reld pairs. A size-dependent separation of DNA requires
strengths applied in slab gel and CE and analysis the addition of a sieving medium to the running
times required by the different systems. As a result of buffer. In slab gel electrophoresis, cross-linked poly-
the high Reld strengths typically used in CE, DNA acrylamide and agarose became commonly adopted
separation 25 times faster than in slab gel elec- in DNA work as support matrices due to their ability
trophoresis is easily achieved. to reduce thermal and gravitational convection which
Besides high speeds, another great advantage of CE would suppress the resolution. In addition, porous
is that DNA samples can be automatically loaded and gels allow the passage of analytes providing the size-
the separated zones can be detected online by Suores- based separation of free-draining polyelectrolytes with
cence or UV absorbance. Furthermore, the ease of nearly equal free solution electrophoretic mobility.
online detection renders CE suitable for automation. In a fused silica capillary the anticonvective role
The amount of samples typically loaded in CE is and the DNA-separating role of the electrophoresis
about three orders of magnitude lower (2}100 nL) matrix can be decoupled. In CE, even in the absence
than that used in slab gel electrophoresis, making the of a dense gel matrix, only very minimal thermal
technique useful as a tool for the analysis of molecu- convection and diffusion of analyte molecules occur
lar biology products, which are often available only during electrophoresis due to the high electrical resist-
in minute amounts. Two types of online injection ance of fused silica, resulting in low current genera-
modes are performed in CE: electrokinetic injection tion. This feature allows the replacement of the gel
and hydrodynamic injection. Electrokinetic injection with a solution of a linear neutral polymer. Most
is accomplished by placing one end of the capillary in of the problems resulting from difRculties in the
the sample vial (which may contain as little as 2 L of production of good quality gel-Rlled columns and
sample) along with a platinum wire electrode. The with their limited lifetime have simply been overcome
application of an electric Reld for a short time forces with the introduction of polymer solutions. Capillar-
charged analytes to enter into the capillary through ies Rlled with linear polymers are relatively easy to
electrophoretic migration. DNA molecules, being prepare and use. Due to their low viscosity, solutions
negatively charged, are injected at the cathode end of of polymers can be automatically pumped into the
concentrations and Rnding the point of deviation

from linearity on the viscosity vs. concentration plot.
In an entangled polymer solution the DNA migration
mechanism is similar to that observed in chemical gels
(cross-linked gels), complicated by the dynamic na-
ture of the DNA}polymer and polymer}polymer in-
teractions. A high resolution can only be achieved in a
rather narrow DNA size range (from 100 to 2000 bp).
When a polymer solution is far below the entangle-
ment threshold concentration, polymer chains remain
isolated in solution and, even if they transiently inter-
act with DNA molecules, they do not form a porous
network. However, DNA molecules larger than
2 kbp can be separated in ultradilute hydroxyethyl
cellulose (HEC) solutions, although with a lower res-
olution. A new separation mechanism, transient en-
tanglement coupling mechanism, was proposed to
explain why the separation occurs in the absence of
a sieving effect. When DNA and isolated polymer
chains encounter each other, they become transiently
entangled. DNA molecules, more stiff and extended
in solution than polymer chains, drag the latter until
Figure 1 Schematic representation of flexible polymer in solu- they escape, slowing down their electrophoretic mo-
tion. (A) dilute solution; (B) c"c *; (C) semidiluted solution; (D) bility. Large DNA fragments experience stronger en-
cross-linked chains. (Reproduced with permission from Heller tanglement interactions, therefore the time required
C (1997) Analysis of Nucleic Acids by Capillary Electrophoresis.
to escape is size-dependent and DNA mobility be-
Viesbaden: Vieweg.)
comes size-dependent. Below c*, high molecular
weight polymer chains provide a high speed separ-
capillary before each run and removed at the end of ation of large molecules with a moderate resolution.
the separation. The automatic replacement of the A polymer, in order to be a good sieving matrix for
sieving matrix increases the capillary lifetime drama- DNA in capillary electrophoresis, should meet the
tically while reducing the need for sample clean-up. following requirements:
1. high water solubility;
Polymer Solutions as DNA Sieving 2. low viscosity when used above its entanglement
Matrices threshold concentration;
3. high chemical stability;
A number of polymers with different properties (such
4. good sieving capacity;
as molecular weight, hydrophilicity, chain stiffness)
5. UV transparency.
have been tested for their ability to sieve DNA mol-
ecules. Cellulose derivatives (hydroxyethyl, hy- Most studies have demonstrated that the molecular
droxypropyl and hydroxypropylmethyl cellulose), mass of the polymer dissolved in the electrophoresis
uncross-linked polyacrylamide and polyacrylamide buffer inSuences the resolution and the interval of
derivatives, dextran, polyvinyl alcohol, polyethylene DNA sizes which can be separated. Polyacrylamide
oxide and polyethylene glycol represent some exam- has, for a long time, been the matrix of choice for
ples of sieving matrices used in CE with varying DNA sequencing and for dsDNA fragments up to
degrees of success. 600 bp. Recently, several authors have replaced poly-
According to Grossman and Soane, as size-depen- acrylamide with polymers obtained by radical polym-
dent DNA electrophoretic separation requires the use erization of N-mono and di-substituted acrylamido
of an entangled polymer solution. Coils of linear derivatives. These polymers offer the advantage of
polymers, above a critical concentration, c*, deRned being more resistant to alkaline hydrolysis than poly-
as the entanglement threshold concentration, begin to acrylamide. One of the most promising derivatives of
interact physically and form a physical gel with tran- polyacrylamide that has found application in CE is
sient pores (Figure 1). The c* concentration, charac- the N,N-dimethylacrylamide (DMA). Short chain
teristic for each polymer, can be experimentally esti- poly(DMA) is produced by carrying out polymeriz-
mated by measuring viscosity at different polymer ation in organic solvents such as dioxane and t-
butanol or in the presence of a chain transfer agent.

Besides optimal sieving capacity, this polymer pos-
sesses the unique feature of suppressing electroos-
motic Sow (EOF) even at 0.001% w/v concentration,
allowing DNA separation to be carried out in un-
coated capillaries. For large DNA (600 bp to 23 kbp)
cellulose ethers, hydroxyethyl, hydroxypropyl, hy-
droxypropylmethyl (HPMC) and methyl cellulose
(MC) have been used as separation media. Chemic-
ally modiRed cellulose derivatives are typically used
to separate DNA fragments at concentrations of
0.2}1% w/v. Due to the fact that stiff and extended
chains become entangled more easily than Sexible
random coil polymers, the operative concentration of
cellulosic polymers is considerably lower than that
used with polyacrylamides.
In]uence of Electroosmotic Flow on

DNA Separation
The most important characteristic of CZE capillary
columns is associated with the chemical structure of Figure 2 Synthetic scheme for a polyacrylamide coating.
the fused silica. The presence of several types of
silanol groups (SiOH) on the inner surface of the
capillary column which are weakly acid in character
dramatically inSuences the CZE separation mechan- of the conducting medium. Most of these methods
ism by inducing an electroosmotic component to the cannot be used for DNA since the operative pH and
transport of ions during the analysis. Negative immo- buffer ionic strengths that are typically used in the
bile charges on the wall surface generate an electric separation are chosen to maintain DNA molecules in
double layer at the silica}buffer interface consisting their native, double-stranded helical conformation.
of an immobile layer of positive ions strongly adsor- Therefore there are severe limitations in the choice of
bed on to the surface and a diffuse layer extending buffer composition.
into the liquid, which is made up of hydrated ions An effective way to suppress EOF which is fully
loosely and aspeciRcally bonded to the surface. When compatible with DNA separation conditions is by
an electric Reld is applied tangentially to the surface, coating the capillary inner surface with a neutral
electrical forces acting on the mobile part of the polymer. This can be accomplished in several ways:
diffuse double layer cause the movement of ions to-
ward the oppositely charged electrode. The migration 1. by dynamic adsorption of water-soluble neutral
of these ions drags the surrounding solvent molecules; polymers. These modiRers, generally contained in
this overall movement of solution is known as EOF. the running buffer, are strongly adsorbed to the
The control of the EOF and the suppression of wall wall by hydrogen bonding or hydrophobic interac-
adsorption is a key aspect in the achievement of tions. Methylcellulose and poly(vinyl alcohol), for
successful DNA separations for two reasons: Rrstly, instance, have been used to produce a layer of high
EOF can cause the extrusion of the sieving matrix viscosity on the capillary wall which suppresses
during electrophoresis and secondly, immobile EOF.
charges on the wall can electrostatically interact with 2. permanent coating with small or large molecules
ions which are oppositely charged. DNA molecules, chemically bonded to the surface silanols or im-
being negatively charged under typical separation mobilized as Rlms of various thicknesses on the
conditions, e.g. pH 8.0, 100 mmol L\1 Tris}borate or capillary wall. The simplest way to prepare these
acetate}EDTA buffer, are generally repelled by the coatings is by bonding an organosilane bearing
wall. However, DNA sample contaminants and a reactive group that can be used, in a second step,
cationic intercalating dyes adsorb onto the wall, caus- to incorporate the organosilane into the polymer
ing peak broadening and a reduction in efRciency. (Figure 2). Polyacrylamide, polyvinylpyrrolidone,
There are a number of approaches to eliminate and epoxy polymer coatings have been success-
EOF and wall adsorption that act on the composition fully prepared in this way.
Independently of the type of neutral polymer and detector, its remarkable sensitivity renders this system
of the mechanism which the polymer uses to bond to the most popular method for DNA detection. The use
the wall, the viscosity in the thick layer of the double of efRcient Suorophores and laser excitation gives
layer appears to govern EOF suppression. However, sensitivities up to six orders of magnitude greater
the nature of the interaction with the wall and the than that typically achieved with UV absorbance. It
structure of the polymer play an important role on the has been demonstrated that bands containing 1 pg of
coating stability. DNA can be detected using ethidium bromide as
Suorescent label. Several different labels with distinct
spectral properties are available (FAM, JOE,
Detection TAMRA and ROX are four Suorophores used for
One of the major advantages of CE is that the separ- DNA sequencing which has relatively widely spaced
ated zones can be detected online. A short stretch absorbance and emission spectra). These Suorophores
(1}2 mm) of the polyimide coating is removed from are excited by the 488 nm and 514 nm lines of a com-
the outside of the capillary and the UV transparent mon argon laser. Their ex,max varies from 550 to
(UV cut-off of 170 nm) fused silica ‘window’ is dir- 610 nm whereas their em,max are between 520 and
ectly placed inside the detector. Online detection pre- 605 nm.
vents diffusion of separated zones resulting from Most of the efforts in developing LIF detection
Sowing through the connections required by an off- were carried out in the area of DNA sequencing
column detection system. Nucleic acids are detected by CE. The availability of good labelling strategies
by UV absorbance at 260 nm or Suorescence. UV is at the heart of LIF detection development.
absorbance is simple and inexpensive; furthermore, Recently, energy transfer primers have been de-
this detection system allows the detection of nucleic veloped to increase sensitivity. These primers contain
acids based on their intrinsic UV absorbance without a common donor dye at the 5 end and an acceptor
any need for intercalating dyes that can alter DNA dye about 8}10 nucleotides away. The presence
electrophoretic mobility. The main drawback of UV of a common donor allows the use of a single laser
absorbance detection in CE is its limited sensitivity at 488 nm to excite all four Suorophore pairs efR-
resulting from the short path length available across ciently.
the 50 m capillaries typically used for DNA separ-
ation. Because of this limitation, laser-induced Su-
orescence (LIF) is currently employed for applications
Applications
that require a higher sensitivity. An example is DNA The advantages of DNA CE over slab gel are such
sequencing where separated zones contain 1}10 at- that an increasing number of molecular biologists
tomoles of material. around the world are beginning to apply this rapid
Although an LIF detection system is more expen- and efRcient technique to DNA mapping and sizing
sive and complicated to build than a UV absorbance separations of medical and genetic interest.
Figure 3 Electropherogram demonstrating the efficient separation of a DNA molecular mass marker containing a mixture of
fragments from cleavage of plasmid pBR322 with restriction endonuclease HAE III. Buffer 100 mmol L\1 Tris}borate, 2 mmol L\1
EDTA, pH 8.2. Temperature: 253C; injection: 5 s at 1 kV; separation 200 V cm\1.
Restriction Endonucleases and DNA Digests ucts that may result from the use of nonoptimized
conditions or from overampliRcation. CE has been
Most work in CE literature has involved separating
successfully used for quantiRcation and sizing of am-
DNA restriction digests. The discovery of enzymes
pliRed products. Given its small sample requirements,
capable of recognizing and cutting a particular short
CE can detect PCR products at a low cycle number
sequence of base pairs (usually 4}8 bp long) in
using LIF, providing an efRcient tool to evaluate PCR
a double-stranded DNA molecule has been of funda-
reaction parameters. Many of the separations which
mental importance in molecular biology. Restriction
can be carried out on a slab gel have been successfully
enzymes are used to compare near-similar DNA mol-
performed in CE. Some of the most signiRcant papers
ecules by cutting them into smaller fragments which
on separation of PCR products by CE are reported in
differ in length or sequence. Restriction digests such
Table 2.
as 174-HaeI-II, pBR322 HaeIII and pBR322 MspI,
containing fragments in the size range from 50 to Capillary Sequencing
1550 bp, have been separated by CE. Larger DNA
The Rrst separations of DNA sequencing fragments
fragments, from 125 to 23 130 bp, are usually gener-
by CE were reported in 1990 by groups at DuPont,
ated with -HindIII. These restriction digests are rela-
Utah, Wisconsin, Northeastern and Alberta. DNA
tively inexpensive, can be obtained at high concentra-
sequencing involves the separation of a set of single-
tion in a puriRed form and contain fragments of
stranded fragments differing by one nucleotide in
known size. They are often used in CE to test the
length which share a common starting point and
separation efRciency of a given separation system
terminate randomly at a particular base. Since 1977,
(Figure 3).
two methods have been developed and reRned to
sequence DNA fragments up to 500 bp. They are
Polymerase Chain Reaction
based on nucleotide-speciRc chemical cleavage or on
Polymerase chain reaction (PCR) is a molecular copy- the use of chain-terminating inhibitors of DNA poly-
ing process that allows the ampliRcation of the merases (dideoxynucleotides -ddNTP-). Almost all
quantity of DNA available for a given test. Using DNA sequencing reactions nowadays are carried out
a three-step temperature cycle, PCR allows speciRc according to the enzymatic protocol of Sanger. DNA
regions of DNA to be ampliRed to a detectable level. sequencing provides the ultimate tool in the detection
The analysis of PCR products requires the ability to of DNA sequence variation. However, for measuring
separate the target sequence from nonspeciRc prod- variations in large DNA regions it is necessary Rrst to
Table 2 Application of PCR product analysis by CE
Type System examined PCR product size (bp)

(resolution needed)
Diagnostic Androgen insensitivity syndrome 136, 139, 160 (3 bp)

Diagnostic Congenital adrenal hyperplasia 127, 135 (8 bp)
Diagnostic Cystic fibrosis F508 95, 98 (3 bp)
Diagnostic Cystic fibrosis, GATT microsatellites (for linkage) 111, 115 (4 bp)
Diagnostic Down’s syndrome, D21S11 220, 224 (4 bp)
Diagnostic Duchenne and Becker muscular dystrophy (18) exon 88, 547 (3 bp)
Diagnostic Dystrophin: DXS 164 locus 740 bpP220, 520 bp
Diagnostic ERBB2 oncogene 1.1 kb bpP500, 520 bp
Diagnostic Factor V mutation 115, 138, 202
Diagnostic Hepatitis C virus 380, 187, 289
Diagnostic HIV-1 virus (gag gene) 115
gag, pol and env genes 142, 394, 442
Diagnostic Kennedy’s disease 480, 540
Diagnostic Medium-chain acyl-coenzyme A dehydrogenase deficiency 175, 202
Diagnostic (SSCP) p53 gene mutation clusters A and B 372
Diagnostic Polio virus 53, 71, 97, 163
Diagnostic VNTR: aboB locus 600}1000 (16 bp)
Diagnostic ZFY gene, Y-chromosome 307
Forensic HLA-DQa 242
Forensic Mitochondrial DNA 402, 437, 1021
Forensic STR: HUMTH01 locus 179}203 (4 bp)
Forensic VNTR: D1S80 401}801 (16 bp)
Genotyping Soybean plant simple sequence repeats (SSRs) 401}801 (16 bp)
Figure 4 (See Colour Plate 76). Four-colour M13mp18 sequencing separation at 423C. The separation of DNA sequencing fragments
was performed in a 6.5% poly(dimethylacrylamide) (98 kDa) solution (75 cP) in 100 mmol L\1 TAPS buffer (pH 8.0) containing 8 mol L\1
urea. A 50 m i.d. bare silica capillary, 51 cm long (40 cm to the window) was used. DNA was electrokinetically injected for 20 s at
60 V cm\1, and the fragments were separated for 125 min at 160 V cm\1. Reproduced from Madabhushi (1998), Electrophoresis 19:
224}230, with permission Wiley}VCH.
subdivide the region into smaller sections of about have suggested the association of a particular dye to
400 bp. Hence, large numbers of sequencing reac- a speciRc ddNTP chain terminator. In one approach,
tions must be performed in each sample. To become Suorescence was excited with two lasers, collected
applicable for screening purposes, DNA sequencing with a single microscope objective and discriminated
requires dramatic improvements in speed. CE pro- with a Rlter wheel. Alternatively, a single laser-excit-
vides a tool to increase sequencing rates signiRcantly; ed Suorescence and a two-channel directly reading
at present, over 700 bases of sequence are routinely Suorescence detector were used to discriminate
generated in a 2 h separation, and occasionally runs Suorescence between the dyes.
of 1.000 in 135 min have been generated. Figure 4 A number of capillary array electrophoresis instru-
shows a four-colour M13mp18 DNA sequencing ments with UF detection have recently become com-
separation. mercially available. These instruments have great po-
Unlike in slab gel, where the four sets of A, C, tential for all those involved in the human genome
G and T terminated fragments run in parallel in four project. In capillary array electrophoresis, many ca-
different lines, in CE it is more convenient to run the pillaries (typically six arrays of 16 capillaries) are set
four sets of fragments in the same capillary because in parallel allowing parallel separations to be run as
this eliminates capillary-to-capillary variation in the in slab gel but offering the advantages of analysis
migration velocity of different fragments. Different speed and automation typical of capillary electro-
strategies have been developed to accomplish this phoresis.
goal, making use of four different labels with distinct
Mutational Analysis
spectral properties. Several labels, such as FAM, JOE,
TAMRA and ROX, are available for sequencing. Detection and localization of single-base differences
Some authors have reported the use of primers in speciRc regions of genomic DNA are of primary
marked with different Suorophores whereas others importance in the analysis of mutations associated
Figure 5 CZE analysis of mutant MV1 in exon 1 of the CFTR gene. The panel shows a run at a constant temperature plateau of 653C.
Inset: run in a 65}673C temperature gradient, with a slope of 0.13C min. Sample injection: electrokinetic, 3 s at 4 kV. Reproduced with
permission from Heller (1997) Analysis of Nucleic Acids by Capillary Electrophoresis. Viesbaden: Vieweg.)
with human disease. The methods used to detect temperature on voltage ramps. A voltage gradient is
point mutations exploit sequence-dependent vari- then generated during the run, taking into account
ations in base-pairing which cause differences in con- the melting proRles of the ampliRed fragments that
formation and lead to altered electrophoretic mobili- can be predicted by the dedicated software of Lerman
ties. Differences in melting behaviour of duplex mol- and Silverstein.
ecules are the bases for separation of DNA sequence
variants by electrophoresis. By running the separation
under a temperature-programmed gradient (TGCE or
Future Developments
thermal gradient capillary electrophoresis), a local Looking to the future, we can expect new develop-
loss of interstrand base paring is induced, generating ments in instrumentation and technology. Another
DNA molecules with a conformation very different promising area is micromachining technology which
from the normal worm-like rods. This allows the has found widespread use in electronic and mechanic
separation of molecules according to their melting engineering, and is now increasingly applied to ana-
characteristics and the detection of deviations from lytical chemistry and biotechnology. Photolithogra-
a wild-type sequence (Figure 5). In CE, the denatur- phy and chemical etching techniques have been com-
ing temperature gradients can be generated internally bined to create the Reld of micromachining in which
via ohmic heat produced by voltage ramps. The tem- three-dimensional microstructures have been fab-
perature increase linked to voltage ramps inside the ricated on a micrometer scale. Electrophoresis on
capillary can be predicted as it depends on capillary a chip has been successfully used for the separation of
diameter, its total length, the electric current values oligonucleotides, DNA restriction fragments and
linked to a given applied voltage, the buffer electric PCR products. In these microchip systems, Suid Sow
conductivity and its coefRcient of thermal conductiv- and reagent mixing are achieved using electrokinetic
ity. With the help of appropriate software, given these transport phenomena. Future developments can also
input parameters it is easy to assess dependence of be expected in this last area.
2560 III / DETERGENT FORMULATIONS: ION EXCHANGE
See Colour Plate 76. Barron AE and Blanch HW (1995) DNA separations by
slab gel and capillary electrophoresis: theory and prac-
See also: II/Electrophoresis: Capillary Electrophoresis;
tice. Separation and PuriTcation Methods 24: 1}118.
Capillary Electrophoresis}Mass Spectrometry; Capillary
De Gennes (ed.) (1979) Scaling Concepts in Polymer Phys-
Electrophoresis-Nuclear Magnetic Resonance. III/
ics. Ithaka, NY: Cornell University Press.
Deoxyribonucleic Acid Profiling: Overview.
Dovich NJ (1997) DNA sequencing by capillary elec-
trophoresis. Electrophoresis 18: 2393}2399.
Further Reading Heller C (ed.) (1997) Analysis of Nucleic Acids by Capillary
Andrews AT (ed.) (1986) Electrophoresis: Theory, Tech- Electrophoresis. Viesbaden: Vieweg.
niques and Biochemical Clinical Applications. Oxford: Righetti PG (ed.) (1995) Capillary Electrophoresis in Ana-
Clarendon Press. lytical Biotechnology. Boca Raton: CRC.
DETERGENT FORMULATIONS:
ION EXCHANGE
S. P. Chopade and K. Nagarajan, sulfonation. Although ABS had very good cleaning
Michigan State University, East Lansing, MI, USA properties, they were non-biodegradable and their
accumulation in the environment caused foaming in
sewage treatment plants and rivers. Hence ABS were
replaced by their straight chain analogues, linear al-
Introduction kylarylsulfonates (LAS) in 1965. Surfactants can be
broadly classiRed as anionic, cationic and non-ionic.
The primary ingredients of a detergent are the surfac- Today’s detergent formulations contain mainly the
tants, whereas builders provide the necessary back- anionic surfactants and a lesser amount of non-ionic
bone. The maximum efRciency of surfactants is surfactants. Cationic surfactants have very small mar-
achieved when the hardness in water is removed. ket share compared to the anionic and non-ionic
Builders provide this essential function of water sof- materials. The use of non-ionics is increasing in liquid
tening, primarily by means of ion exchange. Although detergents as they offer greater stability and formula-
the performance of surfactants is directly dependent tion Sexibility. Whereas the main function of builders
on the efRcacy of the builder, the consumer rarely is to provide water softening capability and alkalin-
notices the importance of the latter. The development ity, they also serve other important functions as dis-
in detergent technology during the last decade has persants, antiredeposition agents, and anticorrosion,
been to improve the property of ion exchange that bleach stabilization and processing aid. Figure 1 illus-
can lead to enhanced washing power of the detergent, trates the various important functions played by
while considering other factors such as environmental zeolites in the washing process. Hard water, if not
concerns and cost. This has been largely due to the
development of ion exchange materials and molecular Table 1 Typical detergent composition
sieve type materials that have been used in detergent
formulations as builders. Laundry detergents have Ingredient Powder Heavy-duty
a yearly $4.4 billion market alone in the United States, detergent liquid detergent
shared equally by the liquid and powder detergents. Surfactants
A typical detergent composition is shown in Table 1. Anionic 15}20 10}40
In today’s detergents, builders constitute about Non-ionic 0}3 0}10
6}25 wt% of liquid detergents and about 20}55 wt% Builders
of powder detergent formulations. Thus builders play Zeolite 20}30 0}25
a signiRcant role in the detergents market. Citrate 0}5 0}10
Polycarboxylates 0}3 }
Until the early twentieth century, cleaning products Carbonate 8}12 0}25
were essentially soaps, i.e. sodium salts of natural Sodium silicates 1}3 }
fatty acids. The surfactants in the Rrst synthetic Sodium sulfate 20}25 }
detergents were short chain alkylnaphthalenesulfon- Enzymes 0}2 0}1.5
ates, which were followed by long chain alkylben- Perborate bleach 0}5 0}10
Polymer stabilizer } 0}1
zenesulfonates (ABS). ABS were prepared by alkyla- Enzyme stabilizer } 0}5
tion of benzene with propylene tetramer followed by
III / DETERGENT FORMULATIONS: ION EXCHANGE 2561
Figure 1 Function of various detergent ingredients in the washing process.
softened by builders, would react with the anionic acid was introduced in some countries as an STPP
surfactants and make them insoluble, thereby render- substitute, but was withdrawn after concern about its
ing them inefRcient. toxicological properties. In the 1970s, the
aluminosilicate zeolite A, was introduced. The
aluminosilicates differ from the STPP builders in
Development of Today’s Builders that they are water insoluble and the mechanism of
The earliest builders consisted of sodium carbonate removal of the calcium and magnesium ions from the
and sodium silicate. Although inexpensive, the disad- water is ion exchange instead of complexation.
vantage of using these as water softeners is that they Usually they are used in conjunction with sodium
form sparingly soluble precipitates that deposit on
fabrics and cause incrustation. In the 1950s, these
builders were replaced by a complexing agent,
sodium diphosphate. However, it was found that
diphosphates also had similar shortcomings of form-
ing precipitates. These drawbacks were overcome
by introduction of sodium triphosphate (STP) and
sodium tripolyphosphate (STPP) in the early 1960s.
STP and STPP act as sequestering agents, which trap
calcium and magnesium ions in the water very efR-
ciently, forming water-soluble complexes. They also
provide other functions like loosening the bond be-
tween soil and fabrics and buffering capacity. How-
ever, due to their nutrient value, the phosphates
caused excessive fertilization and growth of algae in Figure 2 Water hardness distributions. CaCO3 (ppm):
lakes and slow-Sowing rivers, a phenomenon more , (90; , 90}270; , (270. (After Showell, with per-
commonly known as eutrophication. Nitrilotriacetic mission from Marcel Dekker.)
carbonate and polycarboxylates or citrates. Today in The overall stability constant, K, is the product of the
countries like the United States, Japan and some Cen- stability constants for each step (K1 ) K2). The log
tral European countries, zeolites have almost com- K values for stability constants of polyphosphates for
pletely replaced phosphates in detergents. binding of Ca2# and Mg2# ions are fairly high (5.2
and 5.7, respectively) at 'pH 9. Hence, under nor-
Builders for Water Softening mal washing conditions, STPP shows excellent se-
The percentage of builders in detergents depends on questering capacity for both calcium and magnesium
factors like water hardness, wash temperature and ions in solution. The binding capacity of STPP is
wash time. The dissolved inorganic salts of calcium fairly high, 158 mg CaO g\1 at 203C and 113 mg
and magnesium impart hardness to water. For the CaO g \1 at 903C. As the phosphates and the calcium
washing process to proceed effectively, it is important and magnesium chelates formed are water soluble,
to remove the hardness ions in the water and replace the removal is very fast. In addition to water soften-
them with sodium ions, rendering softness to the ing, after removal of the soil, they offer the function
water. Thus the builder must exhibit high capacity for of suspending soil by electrostatic repulsion. They do
selective removal of these ions from water. The not provide as high buffering capacity/alkalinity as
hardness level of water is different in different coun- sodium carbonate or sodium silicate. The low capa-
tries, leading to slight differences in detergent formu- city of STPP to absorb non-ionic surfactants limits its
lations. For example, water in Japan has relatively use in the new compact detergents where consider-
low hardness, typically less than 90 ppm CaCO3, able amounts of the latter are used. EfRciency and
whereas in the United States it is of low-to-intermedi- low cost are the biggest advantages of STPP. In many
ate hardness. The water in Western Europe has very countries, where phosphates are not banned for en-
high hardness (greater than 270 ppm CaCO3 in 42% vironmental reasons, STPP is still the major player in
of the region) (Figure 2). the detergent builder market.
Phosphates ruled the detergent world until the in-
Zeolite A
troduction of inorganic builders. They are still widely
used in a signiRcant part of the world, e.g. Eastern The need for a substitute for STPP led to development
Europe, Southern Asia, Africa, Australia and Latin of zeolites as detergent builders. Among the
America. We will discuss the phosphates brieSy before aluminosilicates, zeolite A is most preferred because
going into detail on zeolites as detergent builders. We of its high ion exchange capacity (170 mg CaO g\1)
will also include some other inorganic silicate builders and low cost of manufacturing. As opposed to chela-
that may have impact on the detergent industry of the tion by STPP, zeolite A removes hardness of water by
future. The main focus will be on the ion exchange means of ion exchange, a process we will discuss in
properties of the materials, particularly zeolites. more detail in further sections. Another important
difference that zeolite A exhibits from STPP is the
Phosphates selectivity for removal of divalent ions. STPP simply
Phosphates are excellent chelating agents and work chelates both the Ca2# and Mg2# ions non-selective-
by a sequestering mechanism, i.e. as previously men- ly from water. Zeolite A being a molecular sieve,
tioned, by trapping the calcium and magnesium ions shows higher selectivity toward Ca2# ions over
and forming water-soluble complexes. A chelating Mg2# ions due to high hydration of Mg2# ions. This
agent possesses multiple sites with lone-pair electrons is advantageous, as studies have indicated that the
that are available to interact with corresponding elec- presence of a certain level of Mg2# ions in fact im-
tron-deRcient coordination sites on metal ions like proves the detergent performance. The ion exchange
Ca2# and Mg2# ions. The chelation is reversible and kinetics are strongly dependent on the wash temper-
the strength of chelation is indicated by the equilib- ature, and hence zeolites tend to be more effective at
rium constant, also known as a stability constant. higher wash temperatures.
The chelation proceeds in a stepwise manner between In addition to these properties, zeolite A offers
metal ions, I, and chelating agent, C. For a divalent other functionality to the detergents. It shows adsorp-
cation it is a two step process given as: tion characteristics that are pH dependent. At pH 10,
which is the pH during a normal wash, there is
minimal adsorption of anionic or cationic surfac-
[CI]
C#I 0 CI K1" [1] tants, leaving them available for the soil removal. As
[C][I] the pH drops during the rinse cycle, the adsorption
tendency of zeolites increases, leading to adsorption
[C2I] of colloidally dispersed soil particles by hetero-
CI#C 0 C2I K2" [2]
[CI][C] coagulation. Thus zeolites act as anti-redeposition
agents, minimizing fabric incrustation. In a mixed -Disilicate

wash of coloured and white clothes, zeolite A sup-
presses the dye transfer from coloured to whites pro- Amorphous sodium silicates were among the earliest
hibiting staining. Zeolite A will ion-exchange trace compounds that were used as builders. These com-
metal ions such as Cu2#, Zn2#, etc. as well as remove pounds remove the water hardness by sequestering
them by the surface deposition of basic salts. As these Ca2# and Mg2# ions, but form insoluble precipi-
metal ions are often responsible for decomposition of tates. Therefore amorphous silicates are used only in
bleaches, use of zeolite A leads to an increase in the small proportions (0}3 wt%) mostly as a co-builder
bleach performance. The only major shortcoming of in the zeolite A builder systems. Here they provide
zeolite A over STPP is the slower rate of ion exchange a source of alkalinity and show selectivity towards
at low temperatures. This is often overcome by use of Mg2# ions. In contrast with amorphous silicates,
a soluble complexation agent, like citrate or a layered silicates are crystalline and work by an ion
polycarboxylate. The complexation agent acts exchange mechanism. These are polymeric crystalline
catalytically. The soluble complexation agent binds materials, which have a layered structure. The mono-
reversibly with the Ca2# ions of the precipitate valent metal ions (Na#) in the interstices can be
(either by ion exchange or by chelation), making easily exchanged with divalent ions in solution and
them soluble. These solubilized Ca2# ions are then are responsible for the ion exchange capacity of the
readily exchanged by zeolite, making the complexa- material. -Disilicate, which has the empirical for-
tion agent free for more transport. Usually about mula Na2Si2O5, is an anhydrous alkali silicate that is
2}8% of polycarboxylates are used as co-builders to most important for its ion exchange properties. The
enhance the performance of zeolite A. formation of -disilicate is not thermodynamically
favoured over the corresponding - and -phases;
hence production requires a specialized high-temper-
Zeolite MAP
ature crystallization process. The -disilicate has
Other than zeolite A, zeolite P and in a few cases a very regular structure (Figure 3) as opposed to the
zeolite X are the only important zeolites that have amorphous silicates.
been considered for detergent applications. Zeolite The -disilicate does not have a thermodynamically
MAP, the maximum aluminium P which is a synthetic stable structure in aqueous media. It disintegrates
gismondine, is an improved version of zeolite P. In slowly into monomeric sodium silicate, commonly
addition to the high binding capacity (&160 mg known as water glass solution but this process is
CaO g\1), zeolite MAP has several distinct advant- much slower than the ion exchange kinetics. The
ages over zeolite A. First, it enhances the stability of hardness ions of water replace the sodium ions in the
the sensitive and environment friendly bleaching -disilicate interstices and give stability to the frame-
agent, sodium percarbonate. Sodium percarbonate is work. During the wash cycle, the pH of the solution is
sensitive to moisture present in the detergent formula- generally maintained at a high level, and the ion-
tion. The low mobility of water in the zeolite MAP exchanged disilicate is stable. As the pH drops in the
structure imparts more stability to the water-sensitive rinsing cycle, the Ca2# and Mg2# ions are released as
bleach. Secondly, it exchanges calcium ions more silicates and the disilicate disintegrates into the wash
rapidly and more selectively. Especially at low tem- water. The size of these insoluble calcium and magne-
peratures, the ion exchange kinetics are considerably sium silicates formed is small and does not cause
faster than those of zeolite A. Finally, zeolite P ex- incrustation.
hibits higher liquid carrying capacity, e.g. for carrying The -disilicate is characterized further by its
non-ionic liquid surfactants, allowing free-Sowing multi-functionality in detergent building. It is a very
powder at higher liquid loading. With all these ad- good source of alkalinity and has the high buffering
vantages, zeolite MAP is replacing zeolite A in more capacity required for the washing process. It exhibits
and more detergent formulations. high adsorption capacity for the non-ionic surfac-
tants. Its mild desiccant properties and ability to
combine with heavy metal ions make it compatible
with unstable bleaches like sodium percarbonate.
This imparts good storage stability to the detergent
formulation and allows the use of environmentally
friendly bleaches. Once again, due to cost consider-
ations, -disilicate cannot completely replace zeolite
Figure 3 Polymetric structure of -disilicate. , sodium; A at this point, but it certainly has great potential to
, SiO4 tetrahedra. act as a supplement to the zeolite A systems.
Zeolites: Synthesis, Structure and

Ion Exchange
Aluminosilicates are denoted by the general formula
Mx/n[(AlO2)x ) (SiO2)y] ) wH2O where M is the cation
of valence n, w is the number of water molecules and
y/x usually has values of 1}5 depending upon the
structure. The sum (x#y) represents the total num-
ber of tetrahedra in a unit cell and the formula within
parentheses represents the framework composition.
The cation M, which functions as a charge balance in
various solids, can be exchanged depending upon the
equilibrium constant and this property Rnds use in Figure 4 Structure of zeolite A.
several applications including catalysis, wastewater
treatment, treatment of radioactive waste and in our interconnecting system is a double four-ring or an
context, detergents. octahedron. The structure can be described in terms
There are around 80 species of natural zeolitic of two types of polyhedra; one is a simple cubic
minerals and about 150 types of synthetic zeolites. arrangement of octahedron or -cage and the other is
Of these, only a few have practical signiRcance a truncated octahedron of 24-hedron or -cage as
and a very few among them have importance in shown in Figure 4. The cubic -cages (Al4Si4O16) are
detergents. The synthetic zeolites, mainly zeolite A, placed in the centres of the edges of a cube in the
and to some extent zeolite P and zeolite X are used in truncated octahedra. These -cages connect the
detergent formulations, mainly as ion exchangers. -cages, creating a three-dimensional structure hav-
Synthesis of Zeolites
ing pores of size 4.2 A> . Each corner of the cube
(-cage) is occupied by the truncated octahedra (-
Zeolites are normally synthesized in the sodium form. cage) enclosing a cavity with a free diameter of 6.6 A> .
Most of the crystalline sodium zeolite is produced by The centre of the unit cell is a large cavity, which has
crystallization from sodium aluminosilicate gels at a free diameter of 11.4 A> . A unit cell of zeolite A con-
a temperature below 1503C. The properties of the tains 24 tetrahedra; 12 each of AlO4 and SiO4. A fully
product zeolite depend greatly on the gel preparation hydrated zeolite A contains 27 water molecules per
and it is therefore a key step in zeolite preparation. unit cell. The theoretical Si/Al ratio in zeolite A is
In a typical process for the synthesis of zeolite A, 1 but in many preparations the Si/Al ratio is slightly
sodium silicate and sodium aluminate solutions are less than 1.
Rrst mixed together. The resulting aluminosilicate is
precipitated and crystallized hydrothermally to ob- Structure of Zeolite MAP (Synthetic Gismondine)
tain the zeolite. Seed crystals of zeolite A are often Zeolite MAP is a type 1 zeolite, meaning that the
used to obtain crystalline zeolite A in high selectivity. interconnecting units are single four-ring units (also
The crystallized zeolite is then washed, Rltered and known as S4R). The eight-membered rings are inter-
dried. The usual gel preparation temperature is connected by S4R units as shown in Figure 5. The
50}803C and the temperature during crystallization is interconnecting chains run parallel to both X- and Y-
slightly higher 80}903C. The time required for gel axes, creating a three-dimensional channel system. The
preparation may vary from 0.5 to 1.5 h, whereas mean free openings of the pores are 3.1;4.5 A> and
1}2 h is necessary for the crystallization. Optimizing 2.8;4.8 A> . The structure is somewhat Sexible, as the
the following variables controls the product quality: connecting S4R units are two-dimensional. Due to this
mixing sequence, speed of agitation, seed crystals, Sexibility, slight structural differences are observed
aging time and temperature. The desired properties by dehydration of the zeolite. MAP has a Si/Al ratio
for optimum detergent application are small particle of 1. This high aluminium content leads to a higher
size (typically (10 m), narrow particle size range, negative charge in the crystal lattice and therefore it
poor adhesive capacity (achieved by having crystal- has the highest theoretical ion exchange capacity.
line material), high calcium binding power and good
wettability. Theory of Ion Exchange in Zeolites
Structure of Zeolite A The ion exchange process may be represented by the

following equation:
The structure of the crystallized zeolite A is quite
(z) #yA(s) xB(s) #yA(z)
xBy# x# y# x#
unique. It is classiRed as a type 3 zeolite, for the [3]
If the ions are equal valence (x"y) then:

KAB"AB [8]
If xOy, then:

Az x
\y
[AB]x"KAB [9]
As
The corrected selectivity coefRcient KYBA includes

a correction for the activity coefRcient of the ions in
the equilibrium:
Ayz Bxs xB
KYBA" y y y [10]
Bz A s A
where A and B are mean ionic activity coefRcients of
the ions in solution.
Figure 5 Structure of zeolite MAP (synthetic gismondine). Ion Exchange by Zeolites

The cation exchange properties of zeolites were ob-
where x and y are the charges of cations A and B served some one hundred years ago. It was this prop-
and the subscripts z and s refer to the zeolite and erty and the ease with which it was happening that
solution, respectively. The equivalent fractions of the led to an early interest in ion exchange materials for
exchanging cation in the solution and zeolite are use as water-softening agents. The cation exchange
deRned by: behaviour of zeolites depends upon the following:
xmAs E Si/Al ratio;
As" A [4] E nature of the cation species, cation size (both an-
xms #ymBs
hydrous and hydrated) and cation charge;
number equivalent of exchanging cation A E temperature;
Az"
total equivalent of cations in the zeolite E concentration of the cation species in solution;
[5] E anion species associated with the cation in solu-
tion;
where mAs and mBs are the molalities of the ions A and E solvent (most of the exchange is carried out in
B respectively, in the equilibrium solution; also aqueous solution although some work has been
(Az#Bz)"1 and (As#Bs)"1. The ion exchange done in organic solvents); and the
isotherm is a plot of Az as a function of As at a given E structural characteristic of the particular zeolite.
total concentration in the equilibrium solution and at
constant temperature. The preference of a zeolite for The cations that contribute towards the hardness
one of the two ions is expressed by the separation of water are calcium and magnesium ions. It is there-
factor, AB, deRned by: fore evident that the property of zeolite should be
such that it has a large ion exchange capacity and
AzBs should be selective towards the exchange of these
AB" [6]
BzAs ions. Zeolite A, zeolite MAP, and to a certain extent
zeolite X, have been shown to exhibit this property
If ion A is preferred, then AB is greater than and consequently are the favourable candidates of
unity. The separation factor depends on the total choice towards being incorporated in the detergent
concentration of the solution, the temperature and As. formulation.
It is not affected by the choice of the concentration Ion exchange in zeolite A seems to be possible only
units. with univalent and divalent counterions. Exchange of
If AB"1, the exchange is ideal and obeys the higher valence ions fails with zeolite A. The order of
Law of Mass Action. However, ideal behaviour is decreasing selectivity for univalent ions in zeolite A is
seldom observed and preference for one of the two as follows: Ag'Ti'Na'K'NH4'Rb'Li'Cs.
ions is usually shown. The rational selectivity coefRc- For divalent ions, the order of decreasing selectivity is:
ient is: Zn'Sr'Ba'Ca 'Co'Ni'Cd'Hg'Mg.
Ayz Bxs The ion exchange kinetics depends on the zeolite
KAB" x y [7]
Bz As structure and cation properties. As zeolite A has
reason for the dominance of zeolite A among other

zeolites as far as builder applications are concerned.
The table indicates that the exchange capacity of
zeolite MAP approaches that of zeolite A as the Si/Al
ratio approaches unity. The exchange kinetics and
total exchange capacity are also dependent on the
Si/Al ratio. Table 2 shows comparison of time re-
quired to reduce concentration of 2 mmol L\1 Ca2#
ions in a solution to the level speciRed. Although
zeolite MAP with a Si/Al ratio approaching unity
shows high capacity and faster kinetics, it becomes
increasingly difRcult to prepare an aluminosilicate
material with a zeolite MAP structure at lower Si/Al
ratios. In many cases the measured exchange capaci-
Figure 6 Kielland plots for zeolite MAP and zeolite A cal-
cium}sodium exchange, 253C, 0.1 mol L\1 solution. ties deviate from theoretical values due to impurities
or variation in chemical composition. The speciRc
exchange capacity also varies with the exchange cation.
a small pore diameter of 4.1 A> , the exchange of ions
like Mg2# is slow due to its high hydration at low Environmental Aspects of Detergent
temperatures. At higher temperatures, the hydration
shell is reduced, making the effective ion size smaller.
Building
Thus the diffusion kinetics and in effect, exchange STPP is a very efRcient and cost-effective builder. As
kinetics of Mg2#, are enhanced at higher temper- it is water soluble, and the hardness removal is by
ature. Ions like Ca2# diffuse easily at room temper- chelation, the process is very fast even at low temper-
atures as they exhibit smaller hydration. This sieving atures. Irrespective of these advantages, many coun-
action is responsible for the selectivity of zeolite A to- tries have stopped using STPP in detergents as it
wards Ca2# ions as compared to Mg2# ions. causes environmental problems. Phosphates, being
Zeolite MAP is more selective than zeolite A to- essential nutrients, cause excessive fertilization in
wards calcium exchange. The plot of calcium selectiv- stagnant waters and slow-Sowing rivers, which leads
ity constant log K versus the calcium loading (Kiel- to excessive growth of algae. These problems can be
land plot, Figure 6) shows peculiar behaviour for avoided by employing a wastewater treatment system
zeolite MAP. For zeolite A, as expected, the selectiv- that removes the phosphorus. However, as alumino-
ity constant decreases with increase in calcium load- silicates made an entry, it was preferred to limit the
ing. In case of zeolite MAP, the opposite behaviour is use of phosphorus compounds in detergents. Alumino-
observed, which indicates a cooperative exchange silicates are environment friendly materials. Alumino-
process. The insertion of some calcium ions in the silicates are produced by combining silica and
structure causes a structural change in zeolite MAP, alumina (from bauxite ore). After use in detergents,
predisposing the framework to accept further calcium they are returned to the environment, where they
ions more readily. Its Sexible structure, which was decompose back to silica and alumina. The only con-
explained earlier, allows this phenomenon. cern about the aluminosilicates arises from their in-
The ultimate base exchange capacity of a zeolite soluble nature. There are some reports of zeolites
depends on the Si/Al ratio; it increases as the Si/Al leading to enhanced sludge in the wastewater. As long
ratio approaches unity (Table 2). This is the primary as the particles are larger than 1 m, they can be
Table 2 Ion exchange capacity and rate of removal of Ca2# from aqueous solutions by zeolites
a a
Zeolite Si/Al Effective ion Time required Time required
exchange capacity to achieve to achieve
(mg CaO g\1 zeolite) 0.5 mmol L\1 Ca 2# (s) 0.01 mmol L\1 (s)
Zeolite MAP 1.46 123 1 Ineffective

1.21 146 1 5
1.12 155 1 4
1.005 159 2 11.5
Zeolite A 1.0 152 14 95
a
Initial Ca2# concentration 2 mmol L\1, 1.48 g dm\3 zeolite, 253C.
III / DRUGS AND METABOLITES / Liquid Chromatography^Mass Spectrometry 2567
easily removed by sedimentation. The -disilicate, efRcient builder and detergent formulations, the chal-
though insoluble during the wash process, dissolves lenge is to achieve compatibility, environment friend-
when the solution becomes dilute during the rinse cycle. liness, and low production cost, all at the same time.
See also: II/Ion Exchange: Inorganic Ion Exchangers;

Future Trends Novel Layered Materials: Phosphates; Novel Layered
The builder development for detergents is now more Materials: Non-Phosphates.
and more driven by environmental issues. Zeolite
A was a major breakthrough over the conventional Further Reading
phosphates. An ideal builder would be one combining
Adams CJ, Araya A, Carr SW et al. (1997) Zeolite MAP:
the efRcacy of STPP and the environment friendly
The new detergent zeolite. Studies in Surface Science and
nature of (alumino) silicates. At this point, combina- Catalysis 105: 1667.
tion of builders and co-builders is the way detergents Breck DW (1974) Zeolite Molecular Sieves: Structure,
are being formulated. The new materials zeolite MAP Chemistry and Use. New York: John Wiley.
and -disilicate, are the future; though more efRcient, Cutler WG and Kissa E (eds) (1987) Detergency Theory
they still have to cross the cost barrier to become and Technology. New York: Marcel Dekker.
major players in the detergent industry. As efforts are Showell MS (ed.) (1998) Powdered Detergents. New York:
constantly underway to come up with more and more Marcel Dekker.
DNA
See III / DEOXYRIBONUCLEIC ACID PROFILING: Overview; Capillary Electrophoresis
DRUGS AND METABOLITES
high sensitivity is effectively supplied by the ion

Liquid Chromatography}Mass multiplier of the mass spectrometer sensor. Thirdly,
in order to identify the individual metabolites, struc-
Spectrometry tural information is required which must be highly
speciRc and contain adequate detail at high resolu-
R. P. W. Scott, Avon, CT, USA tion. Such data is readily provided by the mass spec-
Copyright ^ 2000 Academic Press trum of each eluted solute, which can either be com-
pared with standard spectra or identiRed by using
The tandem analytical instrument, comprising a well-established interpretation procedures. In some
liquid chromatograph and a mass spectrometer is cases, the liquid chromatograph can be coupled to
ideally suited for identifying drug metabolites and for a MS/MS instrument that will provide mass spectra
following their metabolic pathways. The segregation of each fragment ion, from each eluted solute, should
of each drug metabolite from the sample matrix, and more detailed information be needed.
its subsequent identiRcation, Rrstly requires a very For the effective use of LC/MS, special sample
efRcient separation technique. As the metabolites are preparation techniques are necessary and suitable
often chemically very similar to the parent drug, this LC/MS interfaces should be employed. Both these
exigency is adroitly furnished by the liquid subjects are discussed in detail in other parts of the
chromatography (LC) column (usually microbore) Encyclopedia, and it will be sufRcient here to give the
packed with very small particles. Secondly, the mater- sample preparation details of each typical application
ials of interest are inevitably present in the biological that is discussed. In addition, the general interface
system in very small quantities and thus, despite the systems that are used, such as thermospray, electro-
use of sample concentrating techniques, a very sensi- spray, atmospheric pressure chemical ionization devi-
tive detection technique is essential. The required ces, transport interfaces etc., are also described in
2568 III / DRUGS AND METABOLITES / Liquid Chromatography^Mass Spectrometry
detail under LC interface devices and can be referred ous layer was discarded and the solvent extract care-
to when appropriate. Drug and drug metabolites Rnd fully evaporated to dryness and then dissolved in
their way to many parts of the body in a variety of a mixture of 2 mL of dichloromethane and 6 mL of
ways: via the alimentary canal, via the blood stream, hexane. The solution was then extracted by passing
or through the lymph system, and sometimes by it through a prepared Bond-Elut NH2 extraction
direct diffusion through speciRc tissues. In animal cartridge, which was then washed with 5 mL of
tests, drug and drug metabolites can be found, extrac- a hexane/dichloromethane mixture (1 : 1) and 2 mL
ted and assayed in many specialized tissues such as of a hexane/chloroform mixture (1 : 1). The cartridge
the liver, kidneys, heart muscle, brain, etc. and are was then extracted with 5 mL of a mixture of chloro-
usually extracted using standard procedures after form and methanol (7 : 3) and the extract evaporated
homogenizing the tissue. In human tests, the drug and to dryness. The residue was then redissolved in
drug metabolites are usually measured in biological 100 L of mobile phase, which consisted of 0.1 mol
Suids, such as whole blood, blood plasma, gastric L\1 ammonium acetate/acetonitrile (3 : 1). The
Suids, urine, saliva, etc., although occasionally sample was injected onto a reversed phase column,
samples of tissue from biopsies are also examined. 12.5 cm long and 4 mm i.d., containing RP18 station-
In simple cases of drug monitoring, the concentration ary phase bonded to 5-m particles. Chromatograms
of a particular drug and its metabolites may be showing the elution of the Furazolidone by single ion
monitored during treatment merely to ensure that monitoring are shown in Figure 1. It is seen that as
the drug level is kept at an optimum for prime re- a result of the use of single ion monitoring, the pro-
sponse. In more complicated procedures, the drug cedure can be made highly selective. The recovery of
and its metabolites may be monitored in a number of the drug ranged between 65 and 70%. The minimum
different Suids and tissues to determine not only the detectable level of contamination was about 1 g
metabolic pathway, but also those tissues and sites kg\1 of tissue, which was quite adequate to conRrm
where speciRc types of metabolism or drug break- that a sample did not contain the drug in excess of the
down occur. tolerance level.
The unique advantages of the various techniques Another example of the use of the thermospray
and procedures used in LC/MS for monitoring drugs interface for monitoring drugs in cellular matter from
and their metabolites are best illustrated by means of animals is given in the work of Cannavan et al. who
a number of carefully selected applications. There is developed a technique for measuring levamisole in
a wide range of LC/MS applications available for pig and sheep tissue. Levamisole L-(!)-2,3,5,6-
drug and drug metabolite analysis and the following tetrahydro-6-phenylimidazole(1,1-b)thiazole, the
have been chosen as typical examples of the use of laevorotatory isomer of tetramisole, is an anthelmin-
the technique in pharmacology. Furazolidone (N-(5- tic drug used to control gastrointestinal parasites in
nitro-3-furfurylidene-3-amino)-2-oxazolodinone) is cattle, pigs and sheep. In a similar way to that used in
a 5-nitrofuran antibiotic that is added to animal feeds the assay of furazolidone, a somewhat complicated
to help prevent such infections as Escherichia coil sample preparation procedure was necessary. 3 g of
and Salmonella in cattle, pigs and poultry. It follows the tissue sample was mixed with 2 g of anhydrous
that it would be important to know the amount of sodium sulfate, 9 mL of ethyl acetate and 0.5 mL of
residues (if any) remaining after slaughter in any 50% (w/v) potassium hydroxide solution, and the
meat used for human consumption. In Europe, for mixture homogenized. The mixture was then centri-
example, the maximum amount of furazolidone fuged and 6 mL of n-hexane was added to 6 mL of the
that is tolerated is 5 g kg\1 of animal foodstuff, supernatant liquid, and then passed through a Baker
Furazolidone is light sensitive and so operations Bond CN cartridge column. The column was washed
must be carried out under artiRcial yellow light. An with 5 mL of chloroform/n-hexane mixture (1 : 1) and
example of the measurement of furazolidone in some air-dried. The levamisole was eluted with two 5-mL
pharmacokinetic studies is afforded by the work of aliquots of methanol, evaporated to dryness, and
McCracken et al. who used thermospray ionization the residue taken up in 200 L of the mobile
LC/MS for its determination. The procedure that was phase consisting of acetonitrile}tetrahydrofuran}
used is as follows. 2-g samples of liver and muscle triethylamine}water (350 : 50 : 2 : 598 v/v) containing
tissue were homogenized with 40 mL of a mixture of ammonium acetate (0.1 mol L\1). This solution was
McLlvaine buffer and methanol (7 : 3) and then used for the LC/MS analysis. 50-L samples
then centrifuged for 15 min. The supernatant liquid were placed on a Li-Chrospher 60RP-select B column
was then removed and evaporated down to 15 mL at (12.5 cm long, 4 mm i.d), packed with 5-m particles.
403C. 25 mL of dichloromethane was then added and Then mass spectrometer was the Hewlett-Packard
the mixture shaken for about 1 min. The upper aque- 5989A MS engine with thermospray. Single ion
Figure 1 Single ion chromatograms (m/z"243) of furazolidone demonstrating the sensitivity levels obtainable. (Reproduced with
permission from McCracken et al., 1995.)
chromatograms obtained from the analysis are shown ded to the homogenate, which was sonicated and
in Figure 2. They again emphasize the advantages of Rnally centrifuged. A portion of the supernatant
single ion monitoring; the peak of interest is selected liquid was then treated with phosphoric acid mixed
from a complex mixture of peaks with only one other with dichloromethane and again centrifuged.
appearing in the chromatogram. The limit of detec- Acetonitrile and n-hexane were added, the mixture
tion was found to be about 5 ng g\1 of tissue and shaken and again centrifuged. The lower layer was
the calculated mean recovery for liver, kidney and separated and then washed with water. The solvent
muscle tissue was found to be 93, 85 and 79%, mixture was then extracted with phosphate buffer,
respectively. centrifuged once again and the lower layer treated
Many penicillins, particularly penicillin G, are ex- with tetrabutylammonium hydrogen sulfate. The
tensively used in veterinary medicine. It follows that solution was then extracted with dichloromethane,
to prevent antibiotic residues from entering the hu- the extract evaporated to dryness and dissolved in an
man food chain, their levels in animal products that acetonitrile water mixture. Separations were per-
have been prepared for human consumption need to formed on an Intersil ODS-2 reversed-phase column
be carefully monitored. BlanchSower et al. developed (15 cm long and 4.6 mm i.d). The outlet from the
a procedure for simultaneously monitoring Rve peni- column was coupled to a Megabore probe of a VG
cillins, including oxacillin, cloxacillin and dicloxicil- Platform ES-MS, which was operated in the negative
lin in both muscle and kidney tissue and also in milk. ion mode. The source was maintained at 1203C and
BlanchSower et al. employed an LC/MS tandem in- the Sow rate of the drying and nebulizing gas was
strument Rtted with an electrospray interface, and 10 L h\1. The extraction and focus voltages were
utilized single ion monitoring to selectively locate and about 17 and 24 V, respectively. It was found that the
measure each antibiotic. The extraction process was fragmentation pattern was signiRcantly changed by
exceedingly complicated. An appropriate tissue adjusting the voltage on the extraction cone. As the
sample was pulverized, spiked with an appropriate voltage was increased, the degree of fragmentation
standard and homogenized. Acetonitrile was then ad- increased. The effect of extraction cone voltage is
Figure 2 Single Ion chromatograms of levamisole extracted from liver tissue. (A) Levamisole standard (1 m mL\1), (B) negative
liver sample, and (C) liver samples#levamisole (34.8 ng g\1). (Reproduced with permission from Cannavan et al., 1995.)
demonstrated in Figure 3. It would appear that a po- 20 V, the peak at an m/z value of 293 has markedly
tential of 5 V on the extraction cone produced just increased and has been joined by a signiRcant peak at
two ions above an m/z value of 160, i.e. m/z"434 m/z"295. At the same time, the original major peak
and 436. Increasing the potential to 10 V produced at m/z"434 had shrunk to a minor peak and peak at
another peak at an m/z value of 293. At a potential of m/z"436 was barely visible. It is obvious that the
operating conditions of the interface can be critical in
order to achieve the desired selectivity and sensitivity.
This effect is typical for electrospray interfaces. In the
antibiotic assays, the extraction potential was main-
tained between 15 and 17 V. Examples of the
chromatograms obtained for cloroxacillin and peni-
cillin G contained in muscle tissue and obtained by
single ion monitoring at different m/z values are
shown in Figure 4. It shows that monitoring at m/z
values of 293, 390 and 434 selects the cloxacillin
from the accompanying unresolved substances almost
exclusively. It would also appear that the best signal-
to-noise ratio was achieved employing m/z values of
293 and 434. The apparent magnitude of the signal-
to-noise ratio indicates that the assay could detect
levels of cloxacillin as low as 40 ng g\1. The optimum
extraction voltage for assaying penicillin G in milk
appears to be far more critical, the best selectivity
being obtained at a m/z value of 333. The lower limit
of detection, however, is much smaller and it would
appear that an antibiotic level of 1 ng g\1 might be
detectable.
Figure 3 The effect of changing the extraction voltage on the
fragmentation pattern. Extraction potential 5 V(A), 10 V (B), 15 V Although the technique of LC/MS is ideal for the
(C) and 20 V (D). (Reproduced with permission from Blanchflower detection and identiRcation of drugs and their meta-
et al., 1994.) bolites in samples of biological origin, it is often
Figure 4 Chromatogram from the assay of cloxacillin and penicillin G monitored at different m/z values. (A) Muscle spiked with
200 ng g\1 cloxacillin, (B) milk spiked with 4 ng g\1 penicillin G. (Reproduced with permission from Blanchflower et al., 1994.)
necessary to use quite complicated sampling proced- were then again displaced onto the LC column for
ures. For example, Cai and Henion developed an separation. The column eluent was then passed
intricate combination of sampling techniques to de- through an atmospheric pressure chemical ionization
termine LSD and its analogues in urine. The urine interface to the mass spectrometer. A diagram of their
sample was Rrst subjected to afRnity chromatogra- apparatus is shown in Figure 5. The analytical pro-
phy, which selectively removed the LSD and its cedure was quite complicated. The immuno-afRnity
metabolites from the urine. The materials isolated on column that was employed was an HiPac protein
the afRnity column were then displaced and collected G column (3.3 mm;2.1 mm) packed with 30-m
in a special trap, from which the materials of interest particles. Prior to use, the column was equilibrated
Figure 5 Diagram of the sampling arrangement for the analysis of LSD in urine by an LC/MS. (Reproduced with permission from Cai
and Henion, 1996.)
Figure 6 (A) Total ion current chromatogram of a tryptic digest sample of human growth hormone. (B) Spectrum of a product from the
tryptic digest of human growth hormone obtained from low dead volume atmospheric ionization interface.
with phosphate buffered saline (PBS). 30 L of PBS- rates of about 80 to 100 L min\1 were used, and
diluted antibody solution (10% antibody/90% PBS) about 3 L min\1 of the Sow was split from the
was then injected onto the column. Then, the sample mainstream and passed to the capillary column
of human urine, diluted with PBS (50% urine/50% via the capillary interface. It can be seen from Fig-
PBS), was pumped through the protein G column and ure 6 that an excellent separation is obtained and
immediately Sushed with PBS to remove the weakly apparently little resolution is lost in the capillary
bound impurities. While this process was taking interface. The mass spectrum of the peak marked T2
place, the trapping column (1.5 cm long, 1 mm i.d.; in the chromatogram is shown in Figure 6B. It is clear
packed with 5-m C18 particles) and the LC column that good quality spectra can be obtained for ion
(15 cm long, 0.3 mm i.d.; packed with 3-m C18 masses of up to least 900. Such a combination of
particles) were equilibrated with the mobile phase. techniques can be invaluable for the structural elu-
The PBS was then pumped through the afRnity col- cidation of compounds generated in biochemical re-
umn and the trap, desorbing the materials from the search.
afRnity column and re-adsorbing them on the trap.
The trap was then back-Sushed and the desorbed See also: II/Chromatography: Liquid: Detectors: Mass
materials eluted through the LC column into an API Spectrometry. Extraction: Solid-Phase Extraction.
interface and thence into the mass spectrometer. Four
metabolic products in addition to the unchanged LSD Further Reading
itself were separated and identiRed from their mass
spectra. The concentration of LSD in the original BlanchSower WJ, Hewitt SA and Kennedy DG (1994) Con-
urine sample was 0.9 ng mL\1. The results of Cai and Rrmatory assay for the simultaneous detection of Rve
Henion demonstrate how, by utilizing the selective penicillins in muscle, kidney and milk using liquid
chromatography}electrospray mass spectrometry. Ana-
extraction that is provided by afRnity chromatogra-
lyst 119: 2595d2601.
phy, very high sensitivities can be obtained. Cai J and Henion J (1996) On-line immunoafRnity extra-
Recently Thomson et al. described a modiRcation tion-coupled column capillary liquid chromatogra-
of the atmospheric pressure chemical ionization tech- phy/tandem mass spectrometry: trace analysis of LSD
nique involving a special low dead volume interface. analogs and metabolites in human urine. Analytical
Thomson et al. employed packed microbore columns Chemistry 68(1): 72d78.
(170, 320 and 500 m i.d., with lengths ranging from Cannavan A, BlanchSower WJ and Kennedy DG (1995)
5 to 15 cm) in conjunction with a low-volume, wall- Determination of levamisole in animal tissues using
coated capillary column as an interface. The total ion liquid chromatography}thermospray mass spectro-
current chromatogram of the tryptic digest sample metry. Analyst 120: 331d333.
from about 1 pmol of human growth hormone is McCracken RJ, BlanchSower WJ, Rowen C, McCoy MA
and Kennedy DG (1995) The determination of
shown in Figure 6A. The column was packed with an
furazolidone in porcine tissue using thermospray liquid
octadecyl bonded-phase having a mean pore size of chromatography}mass spectrometry and study of the
300 A> and a particle diameter of 7 m. The separ- pharmacokinetics and stability of its residues. Analyst
ation was developed by employing a gradient of 120: 2347d2351.
from 20% solvent (0.1% TFA in water, Figure 6A) Thomson B, Covey T, Shushanm B, Allen M and Sakuma
to 80% solvent (75% of 0.1% TFA and 25% T (private communication) A low dead volume atmo-
acetonitrile, Figure 6B) over a period of 1 h. Flow spheric ionization interface, Perkin Elmer Corporation.
III / DRUGS AND METABOLITES / LC`NMR`MS 2573
Liquid Chromatography ^Nuclear Magnetic Resonance ^

Mass Spectrometry
R. Plumb, G. Dear, I. Ismail and B. Sweatman, A Systematic Approach to Drug
Glaxo Wellcome Research and Development, Metabolite Identi\cation
Ware, UK
This section describes an approach to drug metabolite
identiRcation which utilizes semi-preparative HPLC,
LC/MSn, NMR and HPLC/NMR. The potential of
Introduction coupling NMR and tandem mass spectrometry (MSn)
in the form of an ion trap mass spectrometer is also
As part of the development process of any new chem- highlighted. The advantage of using an ion trap mass
ical entity as a drug substance, evidence is required spectrometer over a quadrupole instrument is prim-
that its metabolism in humans is similar to its meta- arily the ability to generate MSn spectra by repetitive-
bolism in the animal species selected for toxicological ly isolating and fragmenting stored ions, and thereby
evaluation. In order to obtain this information a num- producing second, third and subsequent generation
ber of studies are carried out using both radiolabelled product ions. This enables advantage to be taken of
and unlabelled material. Xenobiotics are generally both structural and molecular weight information
metabolized to generate more polar compounds that generated by the MS to complement the NMR data.
are then more readily eliminated from the body. The The power of the use of these techniques in combina-
metabolism of drugs commonly occurs in two main tion with each other is illustrated, using the example
phases: phase I, functionalization type reactions such of novel non-nucleoside reverse transcriptase inhibi-
as oxidation, reduction and hydrolysis and phase II tor, GW420867 (see Figure 1) which serves to dem-
where metabolism reactions are generally conjugative onstrate the approach currently taken to drug metab-
reactions such as glucuronidation, sulfation, methyla- olite characterization in our laboratory.
tion, and acetylation whereby an endogenous group
is attached to the molecule.
Historically drug metabolite identiRcation has usu- Sample Fractionation
ally been based on the comparison of UV detected
high-performance liquid chromatography (HPLC) re- Urine samples collected following oral administration
tention times of isolated ‘unknown’ metabolites with of GW420867 to animal species and man were
authentic standards. This method of detecting drug pooled to provide 100 mL of sample in each case. The
metabolites, and subsequently characterizing, is not pooled urine samples were separately freeze-dried
only a time-consuming process, and hence expensive, and then reconstituted in distilled water to give a Rnal
but yields no or very limited structural information. volume equal to one tenth of the initial volume. The
The application and use of mass spectrometry in resulting concentrated urine samples were centri-
drug metabolism studies has contributed signiRcantly fuged and the supernatants removed and stored at
to the improved structural characterization of novel 0}83C prior to isolation by preparative HPLC. Injec-
drug metabolites. More recently, tandem mass spec- tions of the concentrated urine samples (10 mL) were
trometry (MS/MS) has been increasingly used in both made onto a preparative HPLC column, and analytes
the characterization and quantitation of metabolites separated with reversed-phase gradient elution. The
in microsomal incubates, urine, plasma and faecal column eluent was collected into 96 well microtitre
extracts derived from both in vitro and in vivo sources. plates at 15 s fraction intervals. The microtitre
Even though mass spectrometry is an exquisitely plates were transferred to a 96 well dry down station
sensitive and speciRc detector capable of providing
drug metabolite information from complex biological
matrices, it does not always provide unequivocal struc-
tural identiRcation, and in these instances NMR spec-
troscopy is often needed to provide conclusive site
speciRc structural characterization. Increasingly the
complementary information provided by MS and NMR
is used in conjunction with HPLC separation techniques
and indeed reports exist where all three techniques
have been linked to generate on-line LC/NMR/MS. Figure 1 Structure of GW420867.
2574 III / DRUGS AND METABOLITES / LC`NMR`MS
Figure 2 1H-NMR spectrum of a 1 mg mL\1 solution of authentic GW420867X in 50 : 50 acetonitrile I deuterium oxide, obtained on
a Bruker DRX-600 spectrometer over 256 scans with dual solvent suppression.
and evaporated to dryness under a constant Sow of molecule contained a Suorine atom and there are
nitrogen at 253C. The dried fractions were re-dis- little or no endogenous Suorinated background sig-
solved in 50 : 50 acetonitrile d deuterium oxide. The nals in urine. However comparison of the 1H NMR
microtitre plates were sonicated for 10 min in an spectra obtained for the isolated fractions with that
ultrasonic bath to aid the resolvation process and the acquired for the authentic parent compound may also
resulting dissolved fractions transferred individually be used for this purpose. The 1H NMR spectrum of
to 5 mm NMR tubes for analysis by 1H and 19F NMR the parent drug is shown in Figure 2. The aromatic
spectroscopy for the presence of drug-related material. protons at 6.78 and 6.94 ppm and the aliphatic pro-
tons at 0.85 and 1.29 ppm were used to identify
Fraction Screening by 1H and 19F NMR possible drug metabolites and also to indicate struc-
tural changes where appropriate. By contrast the
In this particular example the 19F NMR spectra 19
F NMR spectra only provides limited structural
may be used to ascertain the distribution of drug- information on changes in the immediate vicinity of
related material in the processed fractions as the drug the Suorine atom.
Figure 3 1H-NMR spectrum of the H-5 hydroxyl glucuronide metabolite of GW420867X inIacetonitrile I deuterium oxide (1 : 1)
obtained on a Bruker DRX-600 spectrometer over 64 scans with dual solvent suppression, following isolation from rabbit urine.
Figure 4 LC/MS/MS analysis of the H-5 hydroxyl glucuronide metabolite of GW420867X following injection of a rabbit urine fraction
on to a 250;4.6 mm Zorbax RX-C8 5 m column. Analytes were eluted with a 0}100% aqueous 0.1% formic acid/acetonitrile gradient
over 30 min at 1 mL min\1. The column eluent was monitored using the Esquire ion trap operating in positive ion electrospray ‘auto’
MS/MS, with a resonance fragmentation amplitude of 1.2 V.
A 1H NMR spectrum from an isolated rabbit urine Analysis of Metabolite Containing

fraction acquired using only 64 scans, is given in Fraction by Mass Spectrometry
Figure 3 and illustrates that metabolites can poten-
tially be isolated with high degree of purity. Inspec- Fractions highlighted using 19F and 1H NMR spectro-
tion of the 1H NMR spectrum readily enabled the scopy as containing drug-related material were fur-
metabolite to be identiRed as a glucuronic acid conju- ther investigated using electrospray LC/MS/MS oper-
gate of the 5-hydroxylated species (as shown in Fig- ating in data dependent MS/MS mode. Small injec-
ure 3); the anomeric H1 (5.1 ppm) and H2dH5 tions enabled discernible MH# or [M } H]\ ions for
(3.4}3.6 ppm) glucuronyl protons being clearly visible. all metabolites to be obtained to provide complemen-
Figure 5 Rationalization of the product ions resulting from MS/MS of the isolated H-5 hydroxyl glucuronide of GW420867.
tary molecular weight and structural information to The example cited above is an extremely common
supplement. LC/MS/MS chromatograms and spectra and representative example where the precise posi-
corresponding to the NMR spectrum in Figure 3 are tion of aromatic protons are established unequivo-
given in Figure 4. This metabolite could be rationaliz- cally through the use of 1H NMR spectroscopy. The
ed as a hydroxyl glucuronide (see Figure 5), and MS data provided added conRdence by providing
therefore complemented the NMR data already ac- both molecular weight and some structural informa-
quired. In common with many glucuronide conju- tion. Thus by combining the NMR and MS data full
gates, the loss of the glucuronyl moiety in the MS/MS structural elucidation is more effectively and rapidly
spectrum (!176 amu) is clearly observed. MS/MS achieved. Thus for GW4208657 using NMR and MS
data interpretation of all spectra was assisted by com- as described above, a total of 17 metabolites were
parison with the MS/MS spectrum of an authentic unequivocally identiRed following extraction from
standard. various matrices.
Figure 6 The LC/NMR/MSn system was configured as depicted by connecting the Bruker DRX-600 NMR spectrometer and the
Bruker}Esquire mass spectrometer in series. In this configuration the mass spectrometer was operated just outside the 5 Gauss line of
the NMR magnet.
Further Analysis Using Directly impure mixtures of metabolites required further

Coupled LC/NMR/MSn separation and or longer data acquisition using
LC/NMR/MSn.
In most cases metabolites isolated directly from LC/NMR probes are generally more sensitive than
the preparative HPLC fractions alone gave satis- standard NMR probes as the sample is concentrated
factory 1H NMR and MS spectra and could therefore into a smaller volume by the chromatographic pro-
be readily characterized without further investi- cess and the superior Rlling factors (this is the proxim-
gation. However some fractions which had either ity of the NMR RF coils to the sample) obtained in
a low concentration of metabolite or which were LC/NMR probes.
Figure 7 Chromatogram, mass spectrum and 1H NMR spectrum acquired following the injection of a human urine fraction onto the
LC/NMR/MSn system illustrated in Figure 6. The peak at 13.3 min was trapped in the NMR flow probe, and a 1H NMR spectrum
acquired over 256 scans. The peak was then pumped into the ion trap through the electrospray source and the subsequent mass
spectrum acquired in positive ion mode using deuterated solvents. Chromatography was performed on a 250;3.2 mm Phenomenex
Magellen C18 column, eluted with a 0}80% aqueous 0.1% formic acid/acetonitrile gradient, over 35 min.
Additionally the combination of LC/NMR/MS deuterium ‘back exchange’, resulting in undeuterated

allows an impure fraction to become further puriRed MH# ions.
through the use of an analytical chromatographic An UV chromatogram acquired from an on-line
system step following the initial semipreparative iso- LC/NMR/MSn analytical run from a human urine
lation. An example of the use of LC/NMR/MS for the fraction is shown in Figure 7 and indicates the pres-
identiRcation of a novel human urine metabolite of ence of three main peaks. The 1H NMR spectrum and
GW420867 is given below. the resulting MS spectrum acquired for peak III, elut-
For the application of LC/NMR/MS the instru- ing at 13.3 min is also shown in Figure 7. Inspection
ments are conRgured in series as shown in Figure 6. of the 1H NMR spectrum indicates that aromatic
This arrangement allows easy operation of peak stor- substitution has occurred in the position meta to the
age followed by transfer to the NMR spectrometer Suorine-bearing carbon, resulting in the loss of an
and stop Sow analysis, and it avoids difRculties with aromatic proton. The remaining 2 aromatic protons
synchronization of the NMR and MS data capture; at 6.84 ppm (doublet of doublets, J"10.6 Hz and
NMR data acquisition occurs Rrst. 1.47 Hz) and 7.15 ppm (broad due to restricted rota-
Following acquisition of the 1H-NMR spectrum, tion at the amide bond) are consistent with substitu-
the metabolite was transferred from the NMR tion in the 5-hydroxy position.
probe to the ion trap mass spectrometer, where MS The presence of glucuronic acid anomeric H1 pro-
and MSn spectra were obtained. The system was oper- ton at 5.04 ppm suggests the presence of a glucuron-
ated in one of two ways; in one approach, the Sow ide. Calculation of the signal integrals in the aliphatic
from the NMR probe was split directly into the elec- region at 1.26 ppm indicated the loss of one of the
trospray interface. In this mode, the presence of isopropyl methyl groups. Hydroxylation at the iso-
deuterium oxide in the mobile phase produced MD# propyl group would generate a CH2OH group whose
ions in which all labile hydrogens had been ex- protons will be evident in the NMR spectrum in the
changed with deuterium. In the second mode, an region of 3.5 ppm. Although these signals could not
aqueous make up Sow was employed to effect be clearly observed, as the signals from the H2dH4 of
the glucuronic acid were obscuring this region, the cidation of the structure and fragmentation pathway,
sum total of integral values of all the protons in this as shown in the case of this novel human urinary
region indicated the presence of Rve protons. This metabolite. In this example, comparison of the prod-
was in support of the hypothesis of hydroxylation at uct ions at m/z 183 and 255 to their equivalents
the isopropyl methyl. formed from parent drug (m/z 167 and 239 respec-
The MS spectrum shown in Figure 7 was acquired tively, data not shown), indicates aromatic hy-
by directly coupling the LC/NMR probe outlet to the droxylation the position of which was conRrmed by
MS. As NMR spectra are acquired in a mixture of the 1H NMR spectrum. Loss of the isopropyl group to
D2O/ACN the resulting ion at m/z 496 from this generate m/z 255 from m/z 313 suggested that the site
directly coupled arrangement is in a deuterated form of secondary hydroxylation was the isopropyl group,
(MD#). In order to simplify matters an MS spectrum and this is supported by the NMR data. The use of an
of the same peak was acquired, using the ‘back ex- ion trap mass spectrometer providing molecular
change’ conRguration, which allows all the deuterium weight and structural fragmentation information,
atoms to be exchanged with hydrogens. This is given from MS and MSn experiments respectively, when
in Figure 8. In this spectrum the MH# ion can be combined with NMR provides an extremely powerful
observed at m/z 489, which indicates that there are structure elucidation tool.
six exchangeable hydrogens on the original metab-
olite. Subsequent MS4 experiments on m/z 489 were Conclusions
conducted as the peak eluted into the ion trap and
these MS/MS spectra are also given in Figure 8. MSn The importance of combining data from different
spectra enhance the interpretation of the original analytical techniques has been demonstrated. NMR
MS/MS spectrum and can be used to assist the elu- and MSn have been coupled with preparative and
Figure 8 MS and MSn spectra acquired following the injection of human urine fraction onto the LC/NMR/MSn system operating in
‘back-exchange’ mode. Fragmentation was induced using a resonant excitation amplitude of 1.2 V, following mass isolation.
2580 III / DRUGS OF ABUSE: SOLID-PHASE EXTRACTION
analytical HPLC to allow rapid and effective identi- technique, which can lead to particularly signiRcant
Rcation of metabolites in urine, following the oral cost-savings when dealing with the expensive
administration of GW420867. The combination of deuterated solvents used in NMR spectroscopy. Re-
these techniques enables structural and molecular sidual protonated solvent suppression in the NMR
weight information to be interpreted with greater spectrum then becomes an easier task.
efRciency and accuracy. Capillary LC and capillary electrochromatography
(CEC) coupled to mass spectrometry are already in
widespread use within the pharmaceutical industry
Future Prospects and therefore the connection of capillary LC or CEC
The advantages of capillary HPLC and nanospray to both NMR and MS would be a natural progres-
MS, both ‘miniaturized’ versions of their prede- sion.
cessors, HPLC and electrospray, are well documented
and there is also a move towards miniaturization in See also: II/Chromatography: Liquid: Detectors: Mass
most other modern analytical techniques. The en- Spectrometry; Large-Scale Liquid Chromatography; Nu-
hanced sensitivity obtained by capillary LC using clear Magnetic Resonance Detectors.
a capillary UV Sow cell may be extended to NMR,
with the development of capillary NMR Sow cells.
The higher efRciency separations and reduced band Further Reading
broadening that result from capillary LC has the
Chervet JP, Ursen M and Salzmann JP (1996) Anal. Chem.
effect of producing a more concentrated sample elut-
68: 1507.
ing from the column and therefore a higher sample Olson DL, Lacey ME and Sweedler JV (1998) Anal. Chem-
concentration into the NMR Sow cell. This poten- istry News and Features, April.
tially leads to a better signal to noise ratio at lower Vanhoutte K, Van Dongen W and Esmans L (1998) Rapid
analyte concentrations in the acquisition of NMR Commun. Mass Spectrom., 12: 15.
data. A further advantage of capillary LC is the Vanhoutte K, Van Dongen W, Hoes I et al. (1997) Anal.
reduced solvent consumption associated with this Chem. 69: 3161.
DRUGS OF ABUSE:
SOLID-PHASE EXTRACTION
F. Musshoff, Institute of Legal Medicine, Bonn, The drug screening process can generally be
Germany divided into two stages, sample preparation and anal-
ysis of the drugs. Some forms of sample work-up
} isolation and concentration } are required for most
analytical techniques, such as thin-layer chromato-
graphy (TLC), high performance liquid chromatogra-
Introduction phy (HPLC) and gas chromatography (GC) coupled
In toxicological analysis, two basic approaches can be to various detector systems. The samples available for
distinguished: Rrst, a directed search, geared to a lim- analysis are complex biological matrices, in which
ited number of substances, such as in workplace test- toxicologically relevant substances are present in
ing or the analysis of alcohol or special drugs in traf- trace amounts compared to the endogenous com-
Rc offences; and second, an undirected search, also pounds present. Therefore, work-up procedures
called systematic toxicological analysis (STA). STA should retain all relevant substances, at the same time
can be deRned as an undirected search for potentially removing all irrelevant substances and interferences.
harmful substances whose presence are uncertain Liquid}liquid extraction (LLE), often combined
and whose identities are unknown. STA is required if with sample pretreatment procedures such as conju-
little or no information is available in so-called gate hydrolysis, digestion or protein removal, was
general unknown cases. However, if one toxicant the standard method in the past. Although LLE pro-
is known, the analyst is required to establish whether ved to be suitable in a great number of cases, there are
other compounds of toxicological relevance are many disadvantages of this technique, e.g. matrix
present. interferences, emulsion formation or the use of large
III / DRUGS OF ABUSE: SOLID-PHASE EXTRACTION 2581
Table 1 Various solid-phase extraction adsorption phases and solvents
Solubility of Water-soluble Water-insoluble

the sample
Nonionic Ionic
Solvent Aqueous Aqueous Aqueous Organic Organic
Polarity of the Nonpolar Middle polar Polar Cationic Anionic Nonpolar Middle polar Polar
sample
Recommendable C18ec SiOH CN SA SB C18ec SiOH CN

adsorption phase C18 NH2 OH NH2 C18 NH2 OH
C8 PA DMA C8 PA
C4 DMA C4 DMA
C2 NH2 C2 NH2
Phenyl Phenyl
CN CN
Selection of Hexane CHCI3 CHCI3 Acids Bases Hexane CHCI3 CHCI3

recommendable CH2CI2 CH2CI2 CH2CI2 Salt Salt CH2CI2 CH2CI2 CH2CI2
elution solvents Acetonitrile Ethyl Ethyl solutions solutions Acetonitrile Ethyl Ethyl
Alcohols acetate acetate Buffers Buffers Alcohols acetate acetate
Alcohols Alcohols Alcohols Alcohols
Water Water
volumes of organic solvents. In recent years, sample Pretreatment of Biological Fluids

preparation by solid-phase extraction (SPE) has re- and Tissues
ceived widespread interest and today many types of
SPE materials are commercially available (Table 1). The main purposes of sample pretreatment are Rrst,
Most publications have been geared towards the release of drugs from the biological matrix; second,
isolation of one compound or a limited number of removal of proteins and other compounds which
substances, i.e. directed analysis. However, in STA (un- could interfere with further analysis; and third, ad-
directed analysis), a compromise between acceptable justment of the pH, ionic strength and concentration
recovery of many substances and adequate removal of of the sample to allow optimum extraction.
matrix compounds should be reached. In order to devel- For the analysis of plasma/serum samples, buffer
op a SPE method, each step of the procedure should be solutions are widely used for dilution. Protein precipi-
evaluated very carefully, including the selection of tation is a common method to obtain deproteinized
a suitable sorbent, the pH of the sample and the extrac- samples. Proteins can be precipitated by organic sol-
tion system, the clean-up step, the properties and the vents, inorganic salts, metallic ions or acids (Table 2).
volume of the eluent, the Sow rate of the sample and the However, it must be emphasized that co-precipitation
eluent passed through column. often results in losses of the relevant drugs. Urine
Table 2 Relative efficiencies of various protein precipitants towards removing proteins
Precipitant pH of supernatant Volume of precipitant (mL) to precipitate

'98% protein in 0.5 mL plasma
Trichloroacetic acid, 10% (w/v) 1.4}2.0 0.2

Perchloric acid, 6% (w/v) (1.5 0.4
Tungstic acid 2.2}3.9 0.6
Metaphosphoric acid, 5% 1.6}2.7 0.6
Copper sulfate}sodium tungstate 5.7}7.3 1.5
Zinc sulfate}sodium hydroxide 6.5}7.5 2.0
Zinc sulfate}barium hydroxide 6.6}8.3 2.0
Ammonium sulfate (saturated) 7.0}7.7 2.0
Acetonitrile 8.5}9.5 1.5
Acetone 9.0}10.0 1.5
Ethanol 9.0}10.0 2.0
Methanol 8.5}9.5 2.0
Reprinted from Blanchard J (1981) Evaluation of the relative efficacy of various techniques for deproteinizing plasma samples prior to
high-performance liquid chromatographic analysis. Journal of Chromatography 226: 455, with permission form Elsevier Science.
samples can usually be diluted like plasma/serum

samples. However, since many relevant substances or
metabolites will be excreted in a conjugated form, a
deconjugation step prior to SPE is recommended.
Because of the highly variable salt content, the urine
samples should be diluted with at least an equal
volume of water or buffer before applying the sample
on to the SPE column. This is particularly important
when ion exchange is the preferred extraction
mode; otherwise counterions may compete with the
relevant drugs during sample application and result in
losses of the latter. Sonication combined with buffer
dilution is a useful technique for the pretreatment of
whole blood samples. For pretreating tissue samples,
homogenization followed by enzyme digestion (i.e.
papain, subtilisin-A, neutrase, collagenase or trypsin)
and/or protein precipitation and centrifugation of
samples such as liver, kidney or intestine prior to
application on to SPE cartridges are useful. Brain
tissue with a high content of lipids can be used after
incubation with lipase prior to extraction. High Sow
SPE columns are available for use with more viscous
Suids.
As in LLE, pH is an important factor in SPE. The
optimal pH values of the sample and the extraction
system depend on the properties of the relevant drugs
and the sorbent and the interaction between the drugs
and the functional groups of the sorbent. When
a nonpolar sorbent, for example, octadecyl-bonded
silica (C18), is used, the main interactions are van der
Waals forces/hydrogen bonding. Therefore, the pH of
both sample and the column should be adjusted to
a value so that the relevant drugs are in their un-
charged forms. With ion exchange as the underlying
principle, the pH of the sample must be adjusted to
such a value that most of the drugs are charged, so
that they can be retained on the column by the
opposite charge of the functional groups of the sor-
bent. Additionally, total ionic strength of the sample
is important in ion exchange SPE. A low ionic
strength, often obtained by diluting the sample with
water or a low ionic strength buffer solution
((0.1 mol L\1) is preferred, because any species
that can act as counterions reduces the retention of
ionic drugs. Figure 1 Solid-phase extraction process.
The Principles of Solid-phase

phy are to separate each compound with a good peak
Extraction shape and with relatively short retention times. To-
SPE is a physical extraction process that involves day, various sorbent materials are commercially
a solid phase and a liquid phase. It is based on the available. Figure 1 illustrates the steps of SPE. The
principle of liquid chromatography, but with differ- column is Rrst conditioned with an appropriate sol-
ent purposes. The aims of SPE are to isolate the vent to solvate the functional groups of the sorbent.
relevant compounds from a sample matrix and to After the solvent is further conditioned with the
concentrate them, while those of liquid chromatogra- sample matrix buffer, the pretreated sample is forced
through the sorbent by aspiration or positive pres- cleaned very carefully, otherwise interfering substan-
sure. The column containing retained analytes is sub- ces originating from the resin will appear in the ex-
sequently washed with an appropriate solvent that tracts. SDB extractions in columns have now been
selectively elutes the impurities but leaves the analytes largely replaced by SPE using silica-based columns.
on the column. The puriRed analytes are Rnally eluted Recently, new SDB-based SPE columns (e.g. Bond
with a solvent strong enough to displace the analytes Elut ENV, Varian) have become available, with
from the sorbent. which the drawbacks may be overcome. A typical
procedure with this type of material is as follows: the
biological sample is diluted and the pH is adjusted to
Basic Types of Solid Phases the desired value. The resin is washed with four
Diatomaceous Earth (Extrelut威, Chemelut威) column volumes of acetone, three column volumes of
methanol and three times with three column volumes
The principle of SPE using diatomaceous earth is of distilled water. The diluted sample is passed
closely related to conventional LLE. The aqueous through the column where the analytes are absorbed.
phase is absorbed on to the diatomaceous earth, After the resin is washed with water, the analytes
a porous material which acts as support for the aque- are eluted with an organic solvent (e.g. methanol,
ous phase. This provides a large surface area for ethyl acetate, methanol}chloroform, acetone}diethyl
partition into an elution solvent, which Sows through ether, etc.).
the immobilized specimen on the column under
gravity. The elution is a continuous process and may Octadecyl-bonded Silica
give superior recoveries in a shorter time compared to
LLE. Other advantages are the elimination of Octadecyl-bonded silica absorption phases (e.g. C18-
centrifugation, aspiration and Rltration steps and the end-capped) are often used for the directed search to
prevention of emulsion formation. However, rela- a limited number of substances, such as in testing for
tively large volumes of organic solvents are still re- special drugs of abuse in trafRc incidents. The nonpo-
quired. For screening purposes this type of SPE must lar phase retains, at a suitable pH, substances by
be carried out with at least two columns: one for hydrophobic interactions with the alkyl chains. For
acidic and neutral substances and one for basic and example, the simultaneous analysis of tetrahydro-
neutral compounds. A typical procedure with this cannabinol (THC) and its metabolites, 11-hydroxy-
type of material is as follows: the biological sample is THC (11-OH-THC) and 11-nor-9-THC carboxylic
diluted with an appropriate buffer and poured on to acid (THC-COOH) in serum samples is possible as
the column. The bed mass of the column and the follows: the biological sample is diluted with
sample volume must be in agreement with each other. 0.01 mol L\1 acetic acid and the pH is adjusted to 4.
After a 10}15 min equilibration period, twice the An organic solvent (methanol) is used to solvate the
volume of the diluted aqueous sample is used for bonded functional groups and to remove organic resi-
elution from an organic solvent, which is waterim- dues from the sorbent. Buffer is added afterwards to
miscible. exchange the organic solvent with an aqueous solu-
tion. The diluted sample is passed through the col-
Styrene-divinylbenzene Resin (SDB) umn where analytes absorb. After the column is
washed with water followed by a solution of 40%
Polystyrene-divinylbenzene copolymer (e.g. XAD-2)
acetonitrile in water, the analytes are eluted with
is a hydrophobic resin that can absorb many water-
acetonitrile (Figure 2).
soluble organic compounds, principally by van der
Waals forces and additionally by hydrophobic bond-
Mixed-mode Bonded Silica
ing and dipole}dipole interactions. For binding to
the resin the substances must be in a hydrophobic The most widely used SPE materials are bonded silica
state. Therefore, usually two columns are needed: gels, in which end silanol groups have been de-
one for acidic and neutral substances and one rivatized with organic moieties consisting of alkyl
for basic and neutral compounds. Generally the ex- chains with and without a variety of functional
tracts are clean enough to allow GC or TLC groups, such as }OH, }C6H5, }NH2, }CN, }SO3H
determinations at therapeutic and toxic concentra- and }COOH. Based on the modes of the interaction
tions. SDB resin is especially interesting for analysing mechanisms between the functional groups of the SPE
urine samples since glucuronide and sulfate con- materials and the relevant compounds, the extraction
jugates can be isolated. However, the extraction can be divided into three types: nonpolar, polar and
yields of drugs isolated from different biological sam- ion exchange. However, there is no bonded silica SPE
ples may vary considerably. The resin has to be column which only contains one type of functional
Figure 2 GC-MS chromatogram recorded in the selected ion monitoring (SIM) mode. The serum sample was worked up using
a C18-end-capped extraction column and after derivatization of the dried extract using MSTFA after cannabinoids had been determined:
THC (5.2 ng mL\1), 11-hydroxy-THC (4.5 ng mL\1) and THC-COOH (105.6 ng mL\1).
group. Multiple modes of interactions will happen tion step. When a serum or plasma sample is added to
during the extraction process and the inSuence of 0.1 mol L\1 phosphoric acid, this coagulation can be
secondary interactions should be kept in mind. Addi- avoided.
tionally, in so-called mixed-mode silica-bonded SPE
columns, the silanol groups are partially derivatized Column preconditioning The dried sorbents in SPE
with medium length alkyl chains and partially with columns are not in a state to interact with analytes
cation exchange substituents, which can exert at least and appropriate conditioning is required prior to
two types of interactions. Screening procedures using sample application. For nonpolar and multiple-
this type of SPE material have been of increasing interaction phases (mixed-mode), the sorbent must
interest and SPE columns with mixed-mode phases be preconditioned with suitable solvents, for
are available from a number of manufacturers, e.g. example methanol, followed by water or a buffer
Bond Elut Certify (Varian Sample Preparation Prod- wash. The organic solvent used is to solvate the
ucts), Clean Screen DAU (Worldwide Monitoring bonded functional groups and to remove organic
Corp.), Isolute HCX (International Sorbent Techno- residues from the sorbent. Water or buffer, of which
logy) and TSC (Merck). Mixed-mode bonded silica the pH, ionic strength and polarity have been
can retain, at a suitable pH, acidic and neutral sub- adjusted, is added afterwards to remove the organic
stances by hydrophobic interactions with the alkyl solvent with an aqueous solution to prepare the
chains and the basic substances by interactions with SPE column to receive an aqueous sample. For a
the cation exchange groups. Differential elution can polar phase, for example aminopropylsiloxane-
take place by a suitable adjustment of the pH and the bonded silica, the column needs to be treated with
choice of solvents. a nonpolar solvent, such as hexane, to activate the
A typical extraction procedure is described here in surface.
more detail.
Sample application After column preconditioning,
Sample pretreatment Dilution with a 0.1 mol L\1 the pretreated sample is transferred onto the SPE
phosphate buffer at pH 6.0 is most widely used. At column and is drawn through it by applying a light
this pH the weakly basic, the neutral and the weakly vacuum. Normally the Sow rate of sample passing
acidic compounds, such as barbiturates, are in the through the SPE cartridge should be kept to
nonionized form and retained by the octadecyl 1}2 mL min\1.
substituent of the sorbent. However, strongly
acidic compounds like many nonsteroidal anti-in- Column wash and pH adjustment Usually the
Sammatory drugs are deprotonated, ionized and column is washed with 1}2 mL deionized water or an
therefore not retained. When blood samples are appropriate solvent (20% methanol in water) selec-
brought to lower pH values coagulation of proteins tively to remove the impurities which may interfere
occurs, resulting in difRculties in the sample applica- with the analysis. To Rnd a wash solvent which is
able to clean the column effectively without losses ume. Furthermore, the eluent should be selective, so
of the drugs, the analyst must compromise be- that interfering compounds will not be eluted to-
tween acceptable recoveries of a great number of gether with the relevant drugs. Theoretically, eluent
different substances and adequate removal of selection may be achieved by considering the polarity
matrix compounds. In many cases, pH adjustment is index (P), the solvent selectivity and the eluotropic
introduced into this stage to bring the pH of the strength (0) of the solution solvents. The strength of
column to a given value for selective, pH-dependent a solvent is its ability preferentially to dissolve com-
elution of the drugs. In order to get a reproducible pounds according to polarity, while the selectivity is
differential elution using mixed-mode bonded its ability selectively to dissolve one compound as
silica, the pH of the column has to be adjusted opposed to another. Solvents have been classiRed into
to about pH 3. At higher pH values a large number eight selectivity groups according to their proton do-
of basic compounds will elute in the Rrst frac- nor, proton acceptor and dipole interaction charac-
tion (acidic and neutral substances). Lower pH teristics. Figure 3 represents the properties of various
values can cause deterioration of the extraction col- solvent groups with different selectivities. For
umn. To adjust pH, 0.5}1.0 mL diluted acetic acid is example, solvents in group I are strong proton accep-
sufRcient. tors/weak donors with intermediate dipole moments
and solvents in group VIII are relatively strong do-
Column drying Drying of the columns is necessary nors/weak proton acceptors with virtually no dipole
when no water is allowed in the analysing step (GC). interactions.
Drying is carried out by applying vacuum to the The P values and the selectivities of common sol-
column for about 5 min or centrifugation of the col- vents used in SPE are listed in Table 3. The desired P
umn. Further drying can be carried out by applying value can be obtained by using a binary mixture and
a small volume of methanol (50 L) or a larger vol- can be calculated by the following equation:
ume of hexane (1 mL) followed by vacuum. A dry
column is easily obtained, but there is a risk of par- P" aPa# bPb [1]
tially eluting hydrophobic substances such as ben-
zodiazepines in this wash process. where a and b are the volume fractions of solvents
A and B in the mixture; Pa and Pb are the P values of
Elution of relevant drugs For drug elution, the elu- the pure solvents A and B. In practice, both P and
ent should be strong enough so that the drugs can be selectivity of the solvents should be considered when
eluted completely with a reasonably small eluent vol- selecting the best eluent system to elute drugs but not
the impurities. Data shown in Table 3 indicate that
the P values of 2-propanol, chloroform and ethyl
acetate are very similar, yet they belong to different
selectivity groups.
The eluotropic strength 0, which deRnes solvent
strength quantitatively for a given adsorbent, is an-
other helpful parameter for chosing a suitable eluent.
The eluotropic strength is the adsorption energy per
unit area of the solvent and Table 3 lists the 0 values
of some common solvents and binary mixtures. This
knowledge can be helpful in the development of
a new SPE method. If a certain eluent system gives
high recoveries of test drugs but also elutes many
impurities, the use of another eluent mixture with
a similar 0 value could be helpful.
The volume of a selected eluent is another impor-
tant factor in the development of an extraction pro-
cedure. Generally, the volume of the eluent should be
as small as possible. Increasing volumes will prolong
Figure 3 Classification of solvent selectivities (the Roman the extraction period, may elute more impurities and
numbers represent various groups of solvent selectivities). e,
may lose more volatile drugs (such as amphetamines)
d and n represent the fraction of P contributed by interactions
associated with ethanol (acceptor), dioxane (donor) and nitro- when an evaporation step is required after column
methane (polar). Modified with permission from Snyder LR (1974) extraction. The Sow rate of the eluent passing
Journal of Chromatography 92: 223. through the column should allow adequate interac-
Table 3 Solvent eluotropic strength on silica (0), polarity indices (P ) (measure of solvent’s ability to interact as proton donor, proton
acceptor or dipole) and selectivities of some common solvents
Solvent Eluotropic strength (0) Polarity (P) Selectivity group
Water '0.73 10.20 VIII

Acetic acid '0.73 6.20 IV
Methanol '0.73 6.60 II
Methanol}acetonitrile (40 : 60) 0.67
Methanol}diethyl ether (20 : 80) 0.65
2-Propanol 0.63 4.30 II
Methanol}methylene chloride (20 : 80) 0.63
Pyridine 0.55 5.30 III
Isobutyl alcohol 0.54 3.00
Acetonitrile 0.50 6.20 VI
Diethyl ether}acetonitrile (80 : 20) 0.45
Ethyl acetate 0.45 4.30 VI
Acetone 0.43 5.40 VI
Methyl ethyl ketone 0.39 4.50 VI
Tetrahydrofuran 0.35 4.20 III
Methylene chloride 0.32 3.40 V
Chloroform 0.31 4.40 VIII
Diethyl ether 0.29 2.90 I
Benzene 0.27 3.00 VI
Toluene 0.22 2.40 VI
Pentane}diethyl ether (80 : 20) 0.20
Cyclohexane 0.03 0.00
Pentane 0.00 0.00
n-Hexane 0.00 0.06
n-Heptane 0.00 0.20
tion. When ion exchange is the main interaction, the may be washed away. Therefore, an additional LLE
Sow rate should be(2 mL min\1 since ion exchange on the sample coming from the column and column
interactions occur at a slower rate than polar and wash could be introduced in a general screening pro-
nonpolar interactions. cedure.
Using mixed-mode bonded silica extraction col-
umns in the Rrst fraction, fraction A, the analytes Conclusions and Further
retained by the hydrophobic groups of the sorbent are
eluted using a moderately polar solvent like dich-
Developments
loromethane or combinations, e.g. acetone}chloro- The use of SPE for the toxicological analysis of
form (1 : 1), hexane}diethyl ether (40 : 60), hexane} drugs of abuse in biological samples has increased
ethyl acetate (75 : 25). To avoid dirty extracts in the rapidly. Due to the different properties of the drugs
second fraction, fraction B (basic compounds), an of interest, mixed-mode SPE columns are better
in-between polar washing step, for instance with suited for screening purposes than single-mode col-
methanol, may be needed. umns. However, although the same type of SPE ma-
The basic substances retained by the cation ex- terial can be obtained from different manufacturers,
change groups of the sorbent in their protonated form the results using different materials, and even results
are eluted by an organic solvent mixture, usually with obtained from different batches from the same
2% ammonia. Ammoniated ethyl acetate or ammoni- manufacturer, may show signiRcant differences in
ated dichloromethane}2-propanol (80 : 20) for the behaviour, i.e. in particle size distribution and
elution of more polar substances is widely used. Sow velocities. Today, chemically modiRed silica
Table 4 gives an overview of extraction methods and SDB sorbents are also available in extraction
using mixed-mode SPE phases for drug screening. discs. These materials are very promising, since
Amphetamines and other relatively volatile substan- samples can be processed faster using smaller vol-
ces often show lower recoveries, probably caused umes of organic solvents while still allowing rela-
by evaporation in the Rnal step of the SPE pro- tively large sample volumes. Furthermore, extrac-
cedure. Polar drugs like acids and acetaminophenes tions can be performed outside working hours by
are scarcely retained under the condition used and automation of manual SPE methods with the addi-
Table 4 Overview of mixed-mode SPE methods for drug screening
Sample type SPE column Sample Drug conc. Elution method (fraction A Detection Extraction Relative
type volume ( g mL\1 or and fraction B) yield (%) (%) standard
g g\1) deviation
Urine (U) CS DAU A: 4 mL U 0.5}2 A: 10 mL DCM TLC

Plasma (P) B: 5 mL U B: DCM}2PrOH}25% NH3 GC-MS 61}88a (9
(147 : 49 : 4)
BEC 2 mL U/P 10 A: 4 mL Clf}Ac (1 : 1) GC-FID 97}104 (6
B: 2 mL EtAc}33% NH3 (98 : 2)
BEC 1 mL U 0.05 A: 1 mL Hex}ETAC (8 : 2) GC-MS
B: 2 mL DCM}2PrOH}25%
NH3 (80 : 20 : 2)
1: BEC 5 mL U 0.4}1 A: 3 mL Hex}EtAc (75 : 25) GC-MS 1: 60}88b 1: (10
2: Isolute B: 3 mL EtAc}28% NH3 (98 : 2) 2: 48}88b 2: (8
BEC 1 mL U/P 0.1}0.2 A: 4 mL Clf}Ac (1 : 1) GC-NPD U: 82}105 U: (8
B: 2 mL EtAc}33% NH3 (98 : 2) P: 77}103 P: (7
Whole blood BEC 1 mL 0.05}5 A: 4 mL DCM GC-FID 25}104c (14
B: 4 mL EtAc}25% NH3 (98 : 2)
BEC 1 mL 2 A: 4 mL Clf}Ac (1 : 1) GC-FID 81}103 (8
B: 2 mL EtAc}33% NH3 (98 : 2)
BEC 1 mL 0.2}4 A: 2 mL 60% acetoned GC-NPD 50}100 (8
NH3 (80 : 20 : 2)
1: BEC 1 mL 1 A: 3 mL Hex}EtAc (1 : 1) GC-NPD 1: 73}112 1 : 9.7e
2: CS DAU B: 3 mL DCM}2PrOH}28% 2: 59}115 2 : 7.8
NH3 (78 : 20 : 2)
BEC 1 mL 0.5 A: 4 mL Clf}Ac (1 : 1) GC-MSf
B: 2 mL EtAc}25% NH3 (98 : 2)
C: 2 mL DCM}2PrOH}25%
NH3 (80 : 20 : 2)
BEC 1 mL 0.05}0.5 A: 2 mL Clf}Ac (1 : 1) GC-NPD 58}107 d (11
B: 3 mL EtAc}33% NH3 (98 : 2) GC-MSf 26}117 (16
Tissue XTRACT 1.25 g A: 2 mL DCM; 2 mL GC-MS
Hex}Eth (4 : 6)
NH3 (80 : 20 : 2); 4 mL EtAc
BEC 0.1 g 20 4 mL Clf}Ac (1 : 1) GC-NPD 45}101 (9
B: 2 mL EtAc, 33% NH3 (98 : 2) GC-FID
SPE column materials: CS DAU: Clean screen DAU, Worldwide Monitoring, Horsham, PA.
BEC: Bond Elut Certify, Varian Sample Preparation Products, Harbor City, CA.
Isolute: Isolute HCX, International Sorbent Technology, Hengoed Mid Glamorgan, UK.
XTRACT: Worldwide Monitoring, Horsham, PA.
Abbreviations AC, acetone; Clf, chloroform; DCM, dichloromethane; EtAc, ethyl acetate; Eth, diethyl ether; Hex, hexane; NH3,
concentrated ammonia; 2PrOH, 2-propyl alcohol.
a
One SPE column is used for acidic and neutral drugs and one for basic drugs.
b
Low recoveries for barbital and ephedrine.
c
Morphine and amphetamine are hardly recovered.
d
Basic fractions of SPE were cleaned up by liquid}liquid extraction with butyl acetate; recovery of paracetamol is low.
e
Mean values.
f
TMS derivatization.
g
Extraction yields at a spiked concentration of respectively 0.1 and 0.25 g mL\1.
Reproduced from Franke and de Zeeuw (1998) with permission.
tional intention of improving the reproducibility, of- Further Reading

fering high throughput, and reducing labour costs.
Chen XH, Franke JP and de Zeeuw RA (1992) Solid-phase
Various automated SPE systems are commercially extraction for systematic toxicological analysis. Forensic
available. Science Review 4: 147.
Chen XH, Wijsbeek J, Franke JP and de Zeeuw RA (1992)
See also: II/Extraction: Solid-Phase Extractions. III/ A single-column procedure on Bond Elut Certify for
Solid-Phase Extraction with Cartridges. Sorbent systematic toxicological analysis of drugs in plasma and
Selection for Solid-Phase Extraction. urine. Journal of Forensic Sciences 37: 61.
2588 III / DYES / High-Speed Countercurrent Chromatography
Ferrara SD, Tedeschi L, Frison G and Castagna F (1992) Scheurer J and Moore CM (1992) Solid-phase extraction of
Solid-phase extraction and HPLC-UV conRrmation of drugs from biological tissues } a review. Journal of
drugs of abuse in urine. Journal of Analytical Toxicol- Analytical Toxicology 16: 264.
ogy 16: 217. Van Horne KC (1985) Sorbent Extraction Technology.
Franke JP and de Zeeuw RA (1998) Solid-phase extraction Harbor City, CA: Analytichem International.
procedures in systematic toxicological analysis (review). Zief M and Kiser R (1988) Solid Phase Extraction for
Journal of Chromatography 713: 51. Sample Preparation. Phillipsburg, NJ: JT Baker.
DYES
CCC was applied from the outset to the separation of

High-Speed Countercurrent multi-gram quantities of dye mixtures such as xan-
thene and Suoran dyes used as colour additives in
Chromatography food, drugs or cosmetics and as biological stains.
Using a modiRed procedure, the applications of this
A. Weisz, Food and Drug Administration, Washington technique have been extended to the separation of
DC, USA gram quantities of the highly polar mono-, di- and
Y. Ito, National Institutes of Health, Bethesda, trisulfonated components of D&C Yellow No. 10
MD, USA and Yellow No. 203 (both Quinoline Yellow, CI
Copyright ^ 2000 Academic Press 47005) and of other sulfonated dyes such as FD&C
Yellow No. 6 (Sunset Yellow, CI 15985) and D&C
Green No. 8 (Pyranine Concentrated, CI 59040).
Countercurrent chromatography is one of the A general approach to the separation of dyes by
liquid}liquid partition chromatographic techniques HSCCC is presented in Figure 1.
that do not use a solid support. High-speed counter-
current chromatography (HSCCC) uses centrifugal
force to retain one of the liquid phases in an lto
Instrumentation
multilayered coil column while the second liquid For the separation of dyes, both by conventional and
phase is pumped through the column. The principles pH-zone-reRning CCC, commercially available high
of this technique have been discussed by Ito. speed CCC centrifuges are used. Such instruments
HSCCC in its two forms, conventional and pH- are described in the HSCCC entry in this volume.
zone-reRning CCC, is relatively new among the prep- The separations described below were performed
arative techniques used for the separation of dyes. with HSCCC systems from PC, Potomac, MD, USA
Conventional HSCCC has been applied to this pur- and Pharma-Tech Research, Baltimore, MD, USA.
pose since the mid-1980s when Fales et al. separated A schematic diagram of a HSCCC system used for
various components present in a sample of the separation of dyes is shown in Figure 2.
triphenylmethane biological stain, Methyl Violet 2B,
and when Freeman and co-workers puriRed azo tex- Conventional HSCCC
tile and ink dyes (i.e. acid, direct and disperse azo-
dyes). This technique was subsequently used for the Conventional HSCCC may be applied only to the
separation and puriRcation of components from separation of relatively small amounts (up to several
other colours, such as D&C Red No. 28 (Phloxine B, hundred milligrams when the common 325 mL vol-
CI 45410), Sulforhodamine B (CI 45100) and Gar- ume column is used) of dye mixtures. In contrast to
denia Yellow. It was also effectively implemented as pH-zone-reRning CCC, conventional HSCCC also
a complement to preparative high performance liquid permits separation of nonionic components. Analyti-
chromatography (HPLC) for the separation of a com- cal size separations (less than 10}20 mg of an ionic or
plex synthetic mixture of brominated tetrachlorof- nonionic component of interest) should be performed
luorescein dyes. Conventional HSCCC was applied to by conventional HSCCC.
the separation of quantities of dyes of up to several
Selection of the Solvent System
hundred milligrams when the common 1.6 mm
i.d./325 mL volume column was used. By contrast, The solvent system used for a conventional HSCCC
the more recently developed (1993) pH-zone-reRning separation is selected according to the hydrophobicity
III / DYES / High-Speed Countercurrent Chromatography 2589
Figure 1 General approach to the preparative separation of dyes by HSCCC.
of the targeted components so that the respective chromatography pump (Figure 2). The sample to be
partition coefRcients (K) of these components are separated, dissolved in a minimum volume of the
close to 1. Table 1 summarizes the solvent systems solvent system (equal volumes of each phase), is then
used for speciRc dye separations. Also listed in loaded into the column through the sample injection
Table 1 are the general requirements for the selection valve (which facilitates the elimination of air bubbles)
of a solvent system. The procedures for a systematic by syringe or with pressurized nitrogen (60 psi). The
search for a suitable solvent system have been dis- mobile phase is then pumped into the column while
cussed previously. the column is rotated (usually at 3 mL min\1 and
800 rpm). The efSuent is passed through a UV-visible
detector, and fractions of 3 or 6 mL per test tube are
The separation is initiated by Rlling the column with collected with the aid of a fraction collector. The
the stationary phase (usually the upper organic phase fractions obtained from the HSCCC separation are
of the biphasic solvent system) using the liquid then further analysed (e.g. by HPLC).
Figure 2 Schematic presentation of a HSCCC system.

Table 1 Selection of a two-phase solvent system for conventional HSCCC of dyes
Reported solvent systems

Methyl Violet 2B (CHCl3}AcOH}0.1 mol L\1 HCl (2 : 2 : 1))
Sulforhodamine B (n-BuOH}0.01 mol L\1 aq trifluoroacetic acid (1 : 1))
Gardenia Yellow (EtOAc}n-BuOH}H2O (2 : 3 : 5))
Azo-dyes (acid, direct and disperse azo-dyes)
Phloxine B (Cl 45410) (EtOAc}n-BuOH}0.01 mol L\1 aq NH4OAc (1 : 1 : 2))
Brominated tetrachlorofluoresceins (EtOAc}n-BuOH}0.01 mol L\1 aq NH4OAc (1 : 1 : 2))
New solvent systems: general requirements

Settling time shorter than 30 s
Partition coefficient (K ) close to 1 and should be different for each component of the dye
mixture
Solvent system should provide nearly equal volumes of upper and lower phases
Applications principles of this technique are discussed in this vol-

ume. Ideally, the sample components should be stable
Table 1 summarizes the applications of conventional
over a wide pH range, from 1 to 10 for dyes contain-
HSCCC to the separation of components from syn-
ing a carboxylic acid group and from 0.5 to 13.5 for
thetic and natural dye mixtures. Figures 3 and 4 show
sulfonated dyes. The quantity of each targeted com-
two examples in which this technique was success-
ponent in the sample mixture must be at least
fully applied to the separation of dye mixtures.
0.1 mmol and preferably over 1 mmol. Depending on
Oka et al. employed conventional HSCCC to sep-
the solubility of the sample mixture in the solvent
arate the main components from the gardenia fruit
system, multi-gram quantities of dyes can be separ-
extract (Gardenia jasminoides Ellis) used extensively
ated in one experiment using a typical preparative
in Japan as a food colour additive under the name
column of approximately 325 mL capacity. A general
Gardenia Yellow (Figure 3). The solvent system used
approach to the separation of dyes by pH-zone-reRn-
for the separation, ethyl acetate}n-butanol}water
ing CCC is presented in Figure 5. This method is
(2 : 3 : 5), was chosen based on the partition coefRc-
applied to the separation of dyes containing car-
ient (K) of each of the 14 components, as determined
boxylic acid groups (e.g. xanthene dyes or their lac-
by HPLC. Figure 3A shows the HPLC analysis of the
tone analogues, Suoran dyes) by using an organic acid
original mixture. The chromatogram obtained for the
(e.g. triSuoroacetic acid, TFA) as a retainer in the
separation of the three main components (geniposide,
organic stationary phase and a base (e.g. ammonia)
trans-crocin and 13-cis-crocin) from a 25 mg portion
as an eluter in the aqueous mobile phase. Com-
of Gardenia Yellow by conventional HSCCC, using
pounds containing one or more amine groups (e.g.
the upper organic layer as the stationary phase, is
Methyl Violet 2B) may also be separated by pH-zone-
shown in Figure 3B.
reRning CCC by using an organic base as a retainer
Figure 4 shows the chromatogram obtained by
(e.g. triethylamine) and an inorganic acid as an
Fales et al. for the HSCCC separation of the compo-
eluter (e.g. HCl), as described elsewhere in the present
nents of a 6 mg portion of the triphenylmethane dye
volume.
Methyl Violet 2B, composed of a mixture of N-
The more polar sulfonated dyes can be separated
methylated forms of pararosaniline. The solvent sys-
by afRnity-ligand pH-zone-reRning CCC (right col-
tem used in this case, CHCl3}acetic acid}0.1 mol
umn in Figure 5), in which case a ligand is added to
L\1 HCl (2 : 2 : 1), succeeded in separating the various
enhance the partitioning of the dye components into
methylated homologues and several contaminants.
the organic stationary phase.
For this separation, the aqueous layer was used as the
stationary phase. The proposed structures of the sep- Selection of the Solvent System
arated components and of one of the isolated con-
The process of selecting a solvent system for pH-
taminants (mol.wt 491; Figure 4) are based on their
252 zone-reRning CCC separations is very different from
Cf plasma desorption mass spectra.
that conducted when selecting a solvent system for
conventional HSCCC separations. For most pH-
zone-reRning CCC separations, a suitable solvent sys-
pH-Zone-Re\ning CCC tem can be found by testing various volume ratios of
pH-zone-reRning CCC can only be applied to the ether (diethyl or methyl-tert-butyl, MTBE)}acetonit-
separation of ionic or ionizable components. The rile (CH3CN)}water. To the chosen system, an or-
Figure 3 Separation by conventional HSCCC of components from a sample of the Japanese food colour additive Gardenia Yellow.
(A) High performance liquid chromatograms at 254 nm and 435 nm of the original sample. (B) Conventional HSCCC for the separation
of a 25 mg portion of the sample and the HPLC chromatograms of the separated components. Experimental conditions: solvent system:
ethyl acetate}n-butanol}water (2 : 3 : 5 by volume). The aqueous phase was used as mobile phase. Sample: 25 mg Gardenia Yellow
dissolved in 2 mL solvent (1 mL of each phase). Flow rate: 2 mL min\1 in the head-to-tail elution mode. Detection: 254 (continuous line)
and 435 nm (dashed line). Speed of revolution: 800 rpm. Stationary-phase retention: 65.8%. SF, Solvent front. (Oka et al., 1995 with
modifications.)
ganic acid (TFA) is added to the organic layer as For the separation of dyes containing amino groups,
a retainer and an inorganic base such as ammonia is an organic base (such as triethylamine) is added to
added to the aqueous layer as an eluter if dyes con- the organic stationary phase as a retainer and
taining carboxylic acid groups are to be separated. an inorganic acid (HCl) is added to the aqueous
Figure 4 Separation by conventional HSCCC of components from a sample of Methyl Violet 2B. Experimental conditions: Solvent
system: chloroform}acetic acid}0.1 mol L\1 HCl (2 : 2 : 1 by volume). The organic phase was used as mobile phase. Sample: 6 mg
Methyl Violet 2B dissolved in 5 mL organic phase. Flow rate: 4 mL min\1 in the head-to-tail elution mode. Speed of revolution: 800 rpm.
(Fales et al., 1985 with modifications.)
mobile phase as an eluter. For afRnity-ligand priate two-phase solvent system for the separation
pH-zone-reRning CCC separations (for sulfonated by standard pH-zone-reRning CCC of a dye con-
dyes), a ligand is added to the stationary phase taining a carboxylic acid group (summarized in
to retain the sulfonated dyes in the column by enhanc- Figure 6):
ing their partitioning into the organic stationary
phase. 1. Prepare a two-phase solvent system by thor-
oughly equilibrating water and either MTBE or
Standard pH-zone-reVning CCC The following diethyl ether in a separatory funnel at room tem-
steps are recommended for the selection of an appro- perature.
(e.g. methanol) and measure the absorbance at

an appropriate wavelength using a spectro-
photometer. Obtain the partition coefRcient,
Kbase(U/L), by dividing the absorbance of the dye
in the upper phase by that in the lower phase.
4. If Kbase50.5, the above test should be repeated
with a less polar (more hydrophobic) solvent sys-
tem such as n-hexane}ethyl acetate}methanol}
water (1 : 1 : 1 : 1).
5. If Kbase1, add retainer acid TFA (approximately
20 mmol L\1) to the mixture to bring the pH to
approximately 2, and re-equilibrate the mixture
by agitation. Using procedure 3, obtain Kacid. If
Kacid1, the solvent system can be effectively used
to separate the sample components.
6. If Kacid42, the above tests should be repeated
with a more polar (more hydrophilic) solvent sys-
tem such as n-butanol}water.
For amine-containing dyes, substitute HCl for am-

monia in step 3 above to obtain Kacid1 and substi-
Figure 5 General approach to the separation of dyes by pH-
zone-refining CCC. (Reproduced from Fales et al., 1985 with tute triethylamine for TFA in step 5 above to obtain
permission from the American Chemical Society.) Kbase1.
2. Deliver a 2 mL aliquot of the upper (U) and of the AfVnity-ligand separations Sulfonated dyes are
lower (L) phase into a test tube. Add a very small much more polar (hydrophilic) compounds than the
amount of dye and agitate to equilibrate the con- carboxylic acid dyes. They tend to distribute pre-
tents. dominantly in the aqueous phase of a conventional
3. Add a small amount of aqueous ammonia two-phase solvent system, even if the aqueous phase
(+28%; eluter) to the mixture (to give a base is highly acidic. For the separation of sulfonated dyes
concentration of approximately 12 mmol L\1, by pH-zone-reRning CCC, the addition of a hydro-
pH 10) and equilibrate the mixture. If almost phobic ligand is necessary in order to retain the dyes
all the colour visibly partitions into the lower in the organic stationary phase. Two kinds of
aqueous phase, then the partition coefRcient ligands were used successfully for the separation of
Kbase(U/L)1. If visual assessment is unclear, sulfonated dyes: ion exchange reagents (e.g.
Kbase may be determined by spectrophotometry. dodecylamine, tridodecylamine), which are always
Dilute an aliquot from each phase with solvent retained in the organic stationary phase, and
Figure 6 Conditions required for the separation of dyes containing carboxylic acid groups by pH-zone-refining CCC. For conditions
required for the separation of dyes containing amino groups, see text.
Figure 7 Conditions required for the separation of sulfonated dyes by affinity-ligand pH-zone-refining CCC in the ion exchange
mode.
ion-pairing reagents (e.g. tetrabutylammonium hy- brate the contents. Measure the Kbase as above. If
droxide), which partition into either phase. These Kbase1, the solvent composition is adequate for
ligands are dissolved in the organic phase (stationary separation.
phase and/or sample solution) at a concentration de- 7. If Kbase50.2, add more base and retest.
termined in preliminary experiments.
A concentration of 5% ligand (dodecylamine) in
the stationary phase was used for the separation of
Ion exchange mode separations (Figure 7) The fol- the monosulfonted components of D&C Yellow No.
lowing steps are recommended for the selection of an 10, while a higher concentration (up to 20%) of
appropriate two-phase solvent system for the separ- ligand was required for the separation of the di- and
ation of sulfonated dyes by afRnity-ligand pH-zone- trisulfonated components from Quinoline Yellow (in-
reRning CCC in the ion exchange mode: cluding Yellow No. 203).
1. Prepare a two-phase solvent system composed of
Ion-pairing mode separations (Figure 8) The fol-
either MTBE}CH3CN}water at a volume ratio of
lowing steps are recommended for the selection of an
2:2:3 (used for the separation of the main compon-
appropriate two-phase solvent system for the separ-
ent from FD&C Yellow No. 6, Sunset Yellow,
ation of sulfonated dyes by afRnity-ligand pH-zone-
CI 15985) or iso-amyl alcohol (iAA)}MTBE-
reRning CCC in the ion-pairing mode:
CH3CN}water at volume ratios from 3 : 1 : 1 : 5 to
3 : 5 : 1 : 7 (used for the separation of mono-, 1. Follow steps 1 and 2 as described above for the ion
di- and trisulfonated components of Quinoline exchange mode separations.
Yellow, CI 47005). 2. Add a small amount of ion-pairing reagent
2. Deliver a 2 mL aliquot of the upper (U) and of the (tetrabutylammonium hydroxide, TBAOH, 40%
lower (L) phase into a test tube. Add a small weight solution in water, 0.4 mmol) to give a con-
amount of dye and agitate to equilibrate the concentration of approximately 100 mmol L\1 and
tents. a pH of 12.8. Equilibrate the mixture by agitation.
3. Add a small amount of ion exchange reagent (e.g. Measure the partition coefRcient Kbase, as de-
dodecylamine, tridodecylamine at 5}20% concen- scribed above.
tration) and equilibrate the mixture by agitation. 3. If Kbase45, add more ion-pairing reagent and re-
4. Add a small amount of a strong inorganic acid test.
(e.g. HCl, H2SO4 at 3}4%, pH 0.8). Measure the 4. If Kbase1, add an organic acid, TFA (eluter), to
Kacid as above. give an acid concentration of approximately
5. If Kacid45, add more acid and retest. 80 mmol L\1 and a pH of 1.6. Re-equilibrate
6. If Kacid1, add eluter NH3 (28% aq NH3) to the the mixture by agitation. Measure the partition
mixture to give a base concentration of approxim- coefRcient K as above. If Kacid1, the solvent
ately 110 mmol L\1 and a pH of 11. Re-equili- composition is adequate for separation.
Figure 8 Conditions required for the separation of sulfonated dyes by affinity-ligand pH-zone-refining CCC in the ion-pairing mode.
5. If Kacid is not small enough (greater than 0.2), add AfVnity-ligand pH-zone-reVning CCC Ion ex-
more TFA and retest. change mode A separation is initiated by completely
Rlling the column with ligand-free stationary phase
(upper organic phase) by using the liquid chromato-
graphy pump. Approximately 100 mL of ligand-con-
Standard pH-zone-reVning CCC A standard separ- taining stationary phase is pumped into the column,
ation is initiated by completely Rlling the column with thereby displacing part of the column contents. Then
the acidiRed (or basiRed) organic stationary phase the sample solution is loaded into the column through
using the liquid chromatography pump (Figure 2). the sample injection valve by syringe or with pressur-
Then the sample solution is loaded into the column ized nitrogen (60}80 psi). The sample solution is pre-
through the sample injection valve by syringe or with pared as follows: the sample is dissolved in lower
pressurized nitrogen (60}80 psi). When dyes that aqueous phase (e.g. 60 mL for 5 g of Quinoline Yel-
contain a carboxylic group are separated, the sample low) and mixed with stationary phase (60 mL) con-
solution is prepared as follows: the sample is dis- taining the ligand. The sulfonated dyes are brought
solved in the organic stationary phase containing the into the upper phase of the sample solution by the
retainer (TFA) and a smaller amount of aqueous cautious addition of H2SO4 (95}98%, 2.2 mL), at
lower phase (free of eluter). The sample solution which point the pH of the sample solution is approx-
should have a low pH, thus ensuring that the target imately 0.8.
dye is distributed into the upper (stationary) phase After the sample solution is loaded into the col-
of the sample solution. It is desirable that the umn, the aqueous mobile phase containing the basic
sample be completely dissolved in the sample solu- eluter is pumped through the rotating column as
tion. If the sample does not dissolve completely in the described above for performing standard pH-zone-
sample solution, it can be loaded into the column if it reRning CCC. If carryover of the stationary phase
can be homogenized into a Rne suspension by sonica- occurs, the elution is stopped (after approximately
tion for a short time. Ideally, the volume of the 60 mL of stationary phase has been eluted) while the
sample solution should not exceed 100}140 mL for rotation of the column is continued. After 3}4 h of
a preparative separation column of 325 mL capacity. rotation, the elution of the mobile phase is restarted,
After the sample solution is loaded into the column, and usually no more carryover of the stationary phase
the mobile phase containing the eluter (basiRed aque- is observed.
ous phase) is then pumped (usually at 3 mL min\1)
into the column while the column is rotated Ion-pairing mode A separation is initiated by com-
(800}1000 rpm). The efSuent is passed through pletely Rlling the column with ligand-free stationary
a UV-visible detector Sow cell that is further connec- phase (upper organic phase) by using the liquid
ted to a pH Sow cell (for continuous pH monitoring; chromatography pump. The sample solution is then
alternatively, the pH of each collected fraction is loaded into the column through the sample injection
manually recorded with a pH meter) and fractions valve by syringe or with pressurized nitrogen
} usually 3 mL per test tube } are collected with (60}80 psi). The sample solution is prepared as fol-
a fraction collector. lows: the sample is dissolved in lower aqueous phase.
Figure 9 Separation by standard pH-zone-refining CCC of components from the colour additive D&C Orange No. 5 (Cl 45370 : 1).
(A) HPLC analysis of the original sample. (B) pH-zone-refining CCC of the separation and the HPLC chromatograms of the separated
components. Continuous line, absorbance (206 nm); dotted line, pH. Experimental conditions: solvent system: diethyl ether}acetonit-
rile}0.01 mol L\1 aqueous ammonium acetate adjusted to pH 9 with aqueous NH3 (4 : 1 : 5 by volume). The aqueous phase was used as
mobile phase. Sample: 5 g D&C Orange No. 5 suspended in 80 mL solvent system (40 mL each of the upper and lower phases). TFA
200 l was added to the sample solution as a retainer. Flow rate: 3 mL min\1 in the head-to-tail elution mode. Speed of revolution:
800 rpm. Detection: 206 nm. (Weisz et al., 1994 with modifications.)
To this solution an equal volume of upper phase that stationary phase occurs, follow the directions given
contains approximately 100 mmol L\1 of the ion- above for the ion exchange mode separation.
pairing reagent (TBAOH) is added. The pH of the
Applications
sample solution is approximately 12.8, and the dye
partitions mostly into the upper phase. After the Figures 9+13 show successful applications of pH-
sample solution is loaded into the column, the mobile zone-reRning CCC to the preparative separation of
phase, consisting of the aqueous lower phase and components of Suorescein and sulfonated dyes.
80 mmol L\1 TFA as an acid eluter, is pumped
through the rotating column as described above for Standard pH-zone-reVning CCC separations Fig-
standard pH-zone-reRning CCC. The pH of the mo- ure 9 shows the separation of the three brominated
bile phase is approximately 1.6. If carryover of the homologues contained in the colour additive D&C
Figure 10 Separation by standard pH-zone-refining CCC of components from the colour additive FD&C Red No. 3 (erythrosine, Cl
45430). (A) HPLC analysis of the original sample. (B) pH-zone-refining CCC of the separation and HPLC analyses of the separated
components. Continuous line, absorbance (206 nm); dotted line, pH. Experimental conditions: solvent system: diethyl ether}acetonit-
rile}0.01 mol L\1 aqueous ammonium acetate (4 : 1 : 5 by volume). The aqueous phase, adjusted to pH 7.53 with aqueous NH3, was
used as mobile phase. The organic phase (500 mL), to which was added TFA (400 L) as a retainer, was used as stationary phase.
Sample: 3 g FD&C Red No. 3 suspended in 40 mL solvent system (20 mL of the lower phase and 20 mL of the unacidified upper
phase). Flow rate: 3 mL min\1 in the head-to-tail elution mode. Speed of revolution: 800 rpm. Detection: 206 nm. (Weisz, 1996 with
modifications.)
Orange No. 5 (CI 45370:1) by standard pH-zone- ents elution of a pure component of the mixture, as
reRning CCC. The HPLC chromatogram of the illustrated by the associated HPLC chromatograms of
original sample is shown in Figure 9A. The pH-zone- aliquots of the combined fractions from the three
reRning CCC chromatogram of a suspension contain- hatched regions. The recoveries of the three dye com-
ing 5 g of this mixture is shown in Figure 9B. The ponents were good (a: 82%; b: 90.3%; c: 77%).
three broad absorbance plateaux (solid line) corres- Figure 10 shows the use of standard pH-zone-reRn-
pond to the three pH plateaux. Each plateau repres- ing CCC for the separation of iodinated homologues
Figure 11 Separation by standard pH-zone-refining CCC of components from a sample of tetrachlorofluorescein (TCF). (A) HPLC
analysis of the original mixture. (B) pH-zone-refining CCC of the separation and the assigned structures of the isolated components.
Continuous line, absorbance (206 nm); dotted line, pH. Experimental conditions: solvent system: diethyl ether}acetonitrile}0.01 mol
L\1 aqueous ammonium acetate (4 : 1 : 5 by volume). The aqueous phase adjusted to pH 9.06 with aqueous NH3 (0.1%, c. 14 mmol
L\1) was used as mobile phase. The organic phase was used as stationary phase. Sample: 350 mg TCF suspended in 5.5 mL of each
of upper and lower phases. TFA (200 L) was added as a retainer in the sample solution. Flow rate: 3 mL min\1 in the head-to-tail
elution mode. Speed of revolution: 800 rpm. Detection: 254 nm. Retention of the stationary phase: 74.8%. (Weisz et al., 1995 with
modifications.)
and positional isomers from a 3 g sample of the Figure 11 illustrates the separation of the main
colour additive FD&C Red No. 3 (erythrosine, CI component and Rve contaminants from a sample of
45430). All three components were well-separated 350 mg of commercial tetrachloroSuorescein (TCF).
with small mixing zones between them. TCF is an important intermediate in the preparation
Figure 12 Separation by affinity-ligand pH-zone-refining CCC in the ion exchange mode of components from a sample of the
Japanese colour additive Yellow No. 203 (Quinoline Yellow, Cl 47005). Top: HPLC analysis of the original mixture. Bottom: recon-
structed pH-zone-refining CCC elution profile (a constant amount from each CCC collected fraction was analysed by HPLC and the
area of the peak obtained at 516 nm was plotted). Experimental conditions: solvent system: iso-amyl alcohol}methyl-tert-butyl ether}
acetonitrile}water (3 : 5 : 1 : 7 by volume). Stationary phase: 100 mL upper phase, to which was added 20% dodecylamine as a retainer
(pH 11.7) and 220 mL of ligand-free upper phase (see text). Mobile phase: lower phase, to which was added aqueous NH3 (163 mmol
L\1, pH 11.7). Sample solution: 5 g of Yellow No. 203 dissolved in 60 mL each of lower phase and ligand-containing upper phase. To
this mixture was added conc. H2SO4 (40 mmol). The pH of the sample solution (light line, top) became 0.8 and the dye partitioned
mostly in the upper phase. Flow rate: 3 mL min\1 (see text). Speed of revolution: 800 rpm. Detection: 516 nm (bold line). Retention of
the stationary phase: 36.1%. Time of the separation: equilibration 12 h; actual separation time: 5 h. For more details, see text.
Reproduced from Weisz et al. (1995) with permission from the American Chemical Society.
Figure 13 Separation by affinity-ligand pH-zone-refining CCC in the ion-pairing mode of the two components of Yellow No. 203 that
elute together in the separation shown in Figure 12. Top: HPLC analysis of the combined fractions 220}246 from the separation in
Figure 12. Bottom: reconstructed pH-zone-refining CCC of the separation and HPLC analyses of the separated components.
Experimental conditions: solvent system: iso-amyl alcohol}methyl-tert-butyl ether}acetonitrile}water (3 : 4 : 1 : 7 by volume). Stationary
phase: upper phase. Mobile phase: lower phase to which was added TFA (80.7 mmol L\1, pH 1.6). Sample solution: 0.55 g of residue
from combined fractions 220}246 from the separation in Figure 12 dissolved in 30 mL of solvent system (15 mL each of lower and upper
phases), to which was added tetrabutylammonium hydroxide (100 mmol L\1). The pH of the sample solution (light line, top) became
12.8 and the dye partitioned mostly in the upper phase. Flow rate: 3 mL min\1. Speed of revolution: 825 rpm. Detection: 516 nm (bold
line). Retention of the stationary phase: 75%. Time of the separation: 2.5 h.
of higher halogenated dyes (e.g. Phloxine B, Rose See also: II / Chromatography: Countercurrent Chroma-
Bengal), and its contaminants can be carried over tography and High-Speed Countercurrent Chromatogra-
during the manufacturing process. Two of the iso- phy Instumentation. III / Dyes: High-Speed Countercurrent
lated contaminants (e and f) had not been reported Chromatography.
previously.
AfVnity-ligand pH-zone-reVning CCC separations

Further Reading
Ion exchange mode Figure 12 shows the separation Fales HM, Pannell LK, Sokoloski EA and Carmeci, P.
of a 5 g portion of the Japanese colour additive Yel- (1985) Separation of methyl violet 2B by high-speed
low No. 203 (Quinoline Yellow, CI 47005) with the countercurrent chromatography and identiRcation by
ion exchange reagent dodecylamine as the retainer in Californium-252 plasma desorption mass spectrometry.
the stationary phase at a concentration of 20%. To Analytical Chemistry 57: 376}378.
retain the stationary phase in the column, at the point Freeman HS and Williard CS (1986) PuriRcation proced-
in the separation when the column contains approx- ures for synthetic dyes: 2. Countercurrent chromatogra-
imately 50% of the mobile and 50% of the stationary phy. Dyes and Pigments 7: 407}417.
Freeman HS, Hao Z, McIntosh SA and Mills KP (1988)
phase, reduce the Sow rate of the mobile phase to
PuriRcation of some water soluble azo dyes by high-
one-Rfth of the original Sow rate (i.e. 0.6 mL min\1 speed countercurrent chromatography. Journal of
instead of 3 mL min\1) for several hours, and then Liquid Chromatography 11: 251}266.
the Sow rate is increased to its original value. This Ito Y and Weisz A (1994) pH-Zone-ReRning Countercur-
procedure results in a satisfactory retention of the rent Chromatography. US Patent Number 5,332, 504.
stationary phase of 36.1% (measured at the end of Ito Y (1996) Principle, apparatus, and methodology of
the separation). A good separation was obtained high-speed countercurrent chromatography. In: Ito
for two major components, c and b. The Rrst few Y and Conway WD (eds) High-speed Countercurrent
fractions containing c were contaminated with com- Chromatography, pp. 3}44. New York: John Wiley.
ponent a (not shown in Figure 12). Two other Ito Y and Ma Y (1996) Review: pH-Zone-reRning counter-
components (d and e) eluted together, as shown by current chromatography. Journal of Chromatography
A 753: 1}36.
the associated HPLC chromatograms in Figure 12.
Lyon HO, De Leenheer AP, Horobin RW et al. (1994)
Standardization of reagents and methods used in
Ion-pairing mode By using an ion-pairing reagent cytological and histological practice with emphasis on
(TBAOH) as the retainer in the sample solution dyes, stains and chromogenic reagents. Histochemical.
containing the two components of Yellow No. 203 Journal 26: 533}544.
(0.55 g) that eluted together in Figure 12, d and Oka H, Ikai Y and Kawamura N et al. (1991) PuriRcation
e were well-separated, as shown in Figure 13. Com- of food color red no. 106 (Acid Red) using high-speed
pound d is a newly identiRed disulfonated positional counter-current chromatography. Journal of
isomer of Quinoline Yellow. The results presented in Chromatography 538: 149}156.
Figures 12 and 13 for the separation of the compo- Oka H, Ikai Y and Yamada S et al. (1995) Separation of
nents of Yellow No. 203 demonstrate that the ion Gardenia Yellow components by high-speed countercur-
rent chromatography. In: Conway WD and Petroski RJ
exchange and ion-pairing modes can complement
(eds) Modern Countercurrent Chromatography. ACS
each other. Symposium Series, vol. 593, pp. 92}106. Washington,
DC: American Chemical Society.
Conclusion Weisz A (1996) Separation and puriRcation of dyes by con-
ventional high-speed countercurrent chromatography
HSCCC has been used to accomplish the difRcult task and pH-zone-reRning countercurrent chromatography. In:
of separating and/or purifying multi-gram amounts Ito Y and Conway WD (eds) High-speed Countercurrent
of dyes containing either sulfonic or carboxylic acid Chromatography, pp. 337}384. New York: John Wiley.
groups. No other preparative-scale chromatographic Weisz A, Langowski AL, Meyers MB, Thieken MA and Ito
separation of multi-gram quantities of these dyes has Y (1991) Preparative puriRcation of tetrabromotetrach-
been previously reported. It is envisioned that loroSuorescein and Phloxine B by centrifugal counter-
current chromatography. Journal of Chromatography
HSCCC could be scaled up to separate kilogram
538: 157}164.
quantities of dyes through modiRcations of the Weisz A, Scher AL, Andrzejewski D, Shibusawa Y and Ito
instrumentation. The availability of large quantities Y (1992) Complementary use of counter-current
of highly puriRed dyes will have many beneRts, chromatography and preparative reversed-phase high-
such as enabling standardization of biological stains performance liquid chromatography in the separation of
and lowering the cost of the ultra-pure laser-quality a synthetic mixture of brominated tetrachloroSuores-
dyes. ceins. Journal of Chromatography 607: 47}53.
2602 III / DYES / Liquid Chromatography
Weisz A, Scher AL, Shinomiya K, et al. (1994) A new Weisz A, Andrzejewski D, Highet RJ and Ito Y (1998)
preparative-scale puriRcation technique: pH-zone-reRn- Separation of a newly-identiRed contaminant from
ing countercurrent chromatography. Journal of the commercial 4,5,6,7-tetrachloroSuorescein by pH-zone-
American Chemical Society 116: 704}708. reRning countercurrent chromatography. Journal of
Weisz A, Andrzejewski D, Shinomiya K and Ito Y (1995) Liquid Chromatography & Related Technologies 21:
Preparative separation of components of commercial 183}193.
4,5,6,7-tetrachloroSuorescein by pH-zone-reRning Weisz A, Mazzola EP, Matusik JE and Ito Y (1999) Separ-
countercurrent chromatography. In: Conway WD and ation of components of the color additive D&C Yellow
Petroski RJ (eds) Modern Countercurrent Chromatogra- No. 10 (Quinoline Yellow) by pH-zone-reRning
phy. ACS Symposium Series no. 593, pp. 203}217. countercurrent chromatography. Journal of Chromato-
Washington, DC: American Chemical Society. graphy (in press).
W. Nowik, Laboratoire de Recherche des Monuments depends on the generic, basic structure. There are
Historiques, Champs-sur-Marne, France over 20 family structures which many hundreds of
Copyright ^ 2000 Academic Press compounds belong to. Examples of the most fre-
quently encountered structures are shown in Table 1.
The most important application of dyes, from
Introduction a historical point of view, is textile dyeing. For this
reason, the most common classiRcation of dyestuffs is
Natural dyes are organic matter made up of coloured determined by their dyeing properties and dyeing
compounds originating from natural living sources technology. According to their dyeing properties,
such as plants and animals. These compounds can be natural dyes belong to four principal groups: direct
found either directly in the extracts or become col- dyes (e.g. curcumin, a diaryloylmethane from Cur-
oured by hydrolysis, oxidation, condensation, etc., cuma longa), mordant dyes (e.g. alizarin and other
from extracted colourless precursors. Some of them anthraquinones from various madders), reactive dyes
change colour following such reactions or by com- (e.g. depsides and depsidones from lichens) and vat
plexation with certain metallic cations. dyes (e.g. indigotine from indigo, woad, dyer’s knot-
These dyestuffs are generally employed for dyeing weed and others).
Rbres and fabrics (usually protein and cellulosic), Membership to one of the dye families is determined
staining several organic and mineral materials (his- by the type of ‘connection’ between colorant and sup-
tological preparations, wood, hair, feathers, Easter port. However, many dyestuffs may show more than
egg shells), producing organic pigments (painting, one dyeing mechanism, simultaneously or according
printing) and colouring of food, beverages, cosmetics to applied dyeing conditions. The type of connection
and pharmaceutical products. They are used in ana- determines the method of liberation of dyestuff prior
lytical chemistry as well as acid}base indicators or to analysis. IdentiRcation of type and origin of dyes-
complexation agents. A large number of naturally tuffs requires an understanding of composition.
occurring coloured substances are chemically labile.
This excludes their direct use principally because of
poor light-fastness. Some dyestuffs, in addition to Extraction
their colouring qualities, have therapeutic properties
Direct Raw Material Extraction
like regulation of metabolism, anti-inSammatory
treatment and anti-cancer prevention among others. Many natural dyestuffs or their precursors (e.g. indi-
Used frequently around the world in traditional can and isatan B for indigotine) are water-soluble.
societies, starting probably from the Stone Age, natu- Their extraction from raw plant or animal material
ral dyes were rapidly replaced by synthetic dyes from can be achieved by water, sometimes heated to facilit-
the second half of the nineteenth century onwards. ate penetration into the cells. This process, along with
Nowadays, natural dyes have a growing importance possible cell destruction, can also be accelerated by
in our lives since almost all of them are hypoaller- the use of an ultrasound bath. In several cases, the
genic and nontoxic for humans, which is a signiRcant addition of water-miscible solvents (e.g. methanol,
advantage over many synthetic dyes. ethanol or acetonitrile) improves the recovery of rela-
Chemical classiRcation of dyestuffs follows the tively hydrophobic compounds, such as curcumin,
general arrangement of organic compounds and certain antraquinons, etc. Moderately polar solvents
III / DYES / Liquid Chromatography 2603
Table 1 Some important chemical families of dyestuffs
Generic name Generic structure Example
Name Usual name
Carotenoid Carotenoid-dicarboxylic acid Crocetin

(R1, R2"COOH)
Diaryloylmethane Diferuoylmethane (4,4"IOH, Curcumine

5,5"IOMe)
Gallotannin Digaloylhamamelose Hamamelitannin
Flavane 3,3,4,5,7-Pentahydroxyflavane Catechin
Flavanone 3,3,4,7-Tetrahydroxyflavanone Fustin
Flavone 4,5,7-Trihydroxyflavone Apigenin
Flavonol 3,4,5,7-Tetrahydroxyflavone Kaempferol
Isoflavone 3,4,5-Trihydroxy-7- Santal

methoxyisoflavone
Table 1 Continued
Name Usual name
Homoisoflavone 3,4,5,7- Sappanon A

Tetrahydroxyhomoisoflavone
Chalcone 3,4,4,5,6-Pentahydroxychalcone Robtein
Aurone 3,4,6-Trihydroxyaurone-6- Sulfurein

glucoside
Benzoquinone 3,6-Dihydroxy-2,5-di(p- Atromentine

hydroxyphenyl) benzoquinone
Naphthoquinone 2-Hydroxynaphtoquinone Lawsone
Coumarin 7,8-Dihydroxycoumarin-7--D- Daphnin

glucoside
Anthraquinone 1,3-Hydroxy-2- Munjistin

carboxyanthraquinone
Xanthone 1,7-Dihydroxy-9H-xanthen-9-on- Euxanthic acid

7-D-glucuronide
Table 1 Continued
Name Usual name
Phenoxazone -Aminoorcein (R1"Me, Orceina

R2"NH2, R3"orcinol)
Indole 6,6-Dibromoindigotine Purplea
Berbine 5,6-Dihydro-3-hydroxy-2,9,10- Berberine

trimethoxy dibenzo[a,g]
quinolizinium hydroxide
a
The name of a natural mixture of related compounds. It’s applied here when the particular usual name of the compound does not exist.
such as ethyl acetate have been proposed for general anol to prevent possible precipitation of some com-
extraction purposes. The choice of extraction solvent pounds. This hydrolysis destroys chemical ionic and
and conditions affects the quantitative aspects of coordination links between dye molecules and sup-
dyestuff analysis. port or mordant complexes. Fixed extraction condi-
Even the use of water, and especially warm water tions are the goal of repetitive results, because of
} traditionally used for the recovery of colouring possible partial structural change or destruction of
matter } may also introduce composition changes. several compounds. In applied acid solutions a de-
Many natural dyestuffs are heterosides (gluco-, polymerization of oligomeric and polymeric molecu-
rhamno-, primeverosides and other) and can become les (e.g. gallotanins to gallic acid, lacmus to orceines)
aglycone forms through the action of enzymes (e.g. can be observed. The amount of acid usually em-
glucosidases) present in living organisms. Their activ- ployed in the extraction medium is also enough to
ity can be inhibited either by excessive heating catalyse the oxidation of homoisoSavonoids (e.g.
(boiling) or by the addition of denaturation factors haematein to haematoxylin). The major transforma-
(alcohols, strong electrolytes). tion remains, nevertheless, the aglyconeization of the
For quantiRcation purposes, especially in pharma- heterosides. Decomplexation with ligands stronger
ceutical and cosmetic laboratories, extracts are ana- than the dyestuffs has been proposed to improve the
lysed after acid hydrolysis. The hydrolysis of hetero- recovery of complex dyes by avoiding hydrolysis of
sides transforms them into related aglycones (e.g. heterosides.
ruberythric acid to alizarin). This treatment simpliRes In the special case of a non-soluble matrix, as for
the chromatogram, because one aglycone can be ob- organic pigments in drying oil layers from paintings,
tained from several monosaccharide or disaccharide the direct derivatization with m-(triSuoromethyl)
derivatives and the number of chromatographic phenyltrimethylammonium hydroxide (TMTFTH),
peaks decreases (Figure 1). followed by methylation of paint medium with boron
triSuoride/methanol (BF3/MeOH) is employed. This
Coloured Matter Samples Extraction
method provides good recovery of dyestuffs, princi-
Acid hydrolysis has remained a quasi-universal pally anthraquinoids. Its main inconvenience is the
method of natural dyestuff extraction from dyed and formation of multiple methylated derivatives for
stained supports. The extraction medium is usually some compounds.
composed of an aqueous mineral acid solution (e.g. Vat dye constituents are very sparingly soluble in
HCl or H2SO4), sometimes with the addition of meth- aqueous solutions. Some indigoids are soluble in
Figure 1 Influence of hydrolysis on composition of dyestuff extracted from root of Madder (Rubia tinctorium L.). (A) Chromatogram
of fresh aqueous extract; (B) the same extract after acid hydrolysis. Compounds: 1 ruberythric acid (-2-alizarin primeveroside),
2 alizarin, 3 purpurin.
chloroform, but one of them } 6,6-dibromoindigo- for a satisfactory extraction of all compounds from
tine from so called ‘true’ or ‘Tyrian’ purple } is well vat dyes, heated pyridine, dimethylformamide (DMF)
known as a ‘soluble-in-nothing’ compound. In fact, or dimethyl sulfoxide (DMSO) is necessary. Pyridine
Figure 2 Influence of aqueous mobile phase composition and RP-18 stationary phase brand on separation of dyestuff from Dyers’
Broom (Genista tinctoria L.). (A) Hypersil BDS, 3 m, 100;4.6 mm; 80% H2O}10% MeOH}10% of 0.5% H3PO4, from 1 to 40 mn
gradient MeOH 90%. (B) Hypersil BDS, 3 m, 100;4.6 mm, 85% H2O}5% MeCN}10% of 1% MSA, from 1 to 40 mn gradient MeCN
50%. (C) Adsorbosphere HS, 3 m, 100;4.6 mm; 85% H2O}5% MeCN}10% of 1% MSA, from 1 to 40 mn gradient MeCN 50%.
Compounds: 1}5 nonidentified, 6 luteolin, 7 genistein, 8 apigenin.
Figure 2 Continued
is quite toxic and apparently destroys alkyl-bonded Such behaviour of natural dyes determines the type
silica stationary phase when extracts are separated. of chromatographic system used. Apart from the ten-
The indigotins in DMF solutions isomerize to related tative applications of a normal phase (silica) and
indirubins with the help of ultraviolet light. DMSO is a nonpolar eluent, reversed-phase conditions are
very viscous and displays mixing problems (known as most often applied.
‘viscous Rngering’) when the aqueous mobile phase is The solvent system is commonly composed of
employed and affects stationary phase efRciency water, an organic modiRer and an ion suppressor or
(ranging from poor peak shape to multiple peaks). It counterion.
appears that DMF is the best medium, under condi- Many ion suppression reagents are used to control
tions of time- and temperature-controlled extraction, pH. At pH around 2.5 most dyestuffs are unionized
even if its viscosity makes large volume injections but some basic compounds become ionized (e.g. be-
difRcult. rberine, which is fortunately rather rarely encoun-
tered). The most popular acids are o-phosphoric
(H3PO4) and acetic (CH3COOH, AcH), but in the
Chromatography case of mass spectrometry (MS) on-line detection they
should be replaced by volatile triSuoroacetic acid
Mobile Phases
(TFA). To avoid the formation of poorly soluble salts
The solubility of natural dyestuffs covers practically of basic compounds and/or to improve peak shapes,
the entire range of solvent polarities, from water to methanesulfonic acid (MSA) has been proposed
n-hexane. Still, many of them are soluble in both (Figure 2, compare A and B).
moderately polar organic solvents (e.g. chloroform) The use of counterions is less habitual owing to
and water. These properties are linked to the com- the considerable equilibrium stabilization time
pounds’ structures. In addition to their generic, low needed for the stationary phase in gradient ana-
polar, aromatic or isoprene structures, they usually lysis.
contain polar groups, especially proton-releasing ones Analysis time in both isocratic and gradient elution
such as hydroxyl and carboxyl. Practice shows that the is usually between 15 to 45 min, depending on
use of water with some organic water-miscible sol- the type and the number of compounds requiring
vents solubilizes almost all classes of dyestuffs. separation.
Figure 3 General and selective UV-vis channel detection of violet dyed wool sample (Coptic textile, 6th century). (A) General
detection at 285 nm; (B) specific detection of red components corresponding to Madder at 450 nm; (C) specific detection of indigo blue
at 615 nm. Compounds: 1 anthragallol, 2 pseudopurpurin, 3 alizarin, 4 indigotin, 5 purpurin.
Figure 3 Continued
Stationary Phases time and, if possible, on absorption at characteristic

wavelengths, usually in the visible (e.g. 360 nm for
Among the stationary phases used in natural dye-
yellows, 485 nm for reds, 650 nm for blues, etc., with
stuffs analysis, the major part of applications is
a suitably large bandwidth) (Figure 3). The photo-
achieved on octadecyl (called RP 18, or C 18) bonded
diode array (PDA) gives the possibility of conRrma-
silica. This type of phase is the most versatile; how-
tion of compound identity by the corresponding peak
ever, different brands bring about important differ-
spectra (Figure 4).
ences in separation characteristics (Figure 2, compare
The principal inconvenience of UV-vis detection is
B and C). The octyl (RP 8) and phenyl (Ph) phases
the recognition of only those substances entered as
have some speciRc applications and can offer other,
references in the database. Even the PDA gives only
sometimes more satisfactory separations than RP 18
partial structure recognition because of limited UV-
in speciRc cases.
vis information contained in the spectra and their
Advances in natural dyes chromatography follow
possible alteration in different mobile phase composi-
general trends: a growing efRciency of separations
tions. New compounds have to be collected at the
and reduced separation time as well as increased
detector exit and analysed by other, usually spectro-
sensitivity of detection. This is the reason for the
metric, methods. This problem can be avoided by
constant testing of the types of stationary phases, and
application of on-line mass spectrometry, giving more
of the rising use of smaller and more uniform particles
precise information about the molecular structure.
and also of shorter and narrower columns.
An interesting approach to detection can be in-line
tandem of PDA and a mass-selective detector (MSD)
Detection and Identi\cation permitting pre-selection of coloured compounds and
thus their detailed analysis.
Detectors
Sources Recognition Criteria
Dyestuff components show strong selective absorp-
tion in the UV-vis spectrum. This effect is responsible High-performance liquid chromatography analysis of
for their coloration and makes absorption detection natural dyes gives an identiRcation of compounds.
simple. In the case of single or multichannel detectors, From this information, as a Rrst step, a chemical class
the recognition of compounds is based on retention or dyeing group can be deduced. Moreover, the
Figure 4 Representation of a fragment of data collected from PAD during the analysis of extract from females of Kermes (Kermes
vermilio PLANCH). Kermesic acid (higher absorption and flavokermesic acid (lower absorption). (A) General view (axes: x"time,
y"wavelength, z"absorption); (B) Chromatograms cut display (x, z). (C) Spectra cut display ( y, z).
qualitative and quantitative information about the of minor peaks on chromatograms recorded at the
compounds detected in a chromatogram can help in colour-characteristic wavelengths. The correspond-
dyestuff source recognition. This is the aim of ing characteristic compounds are adequate for the
chemotaxonomy, employed for many purposes: in identiRcation of botanical or zoological phylem,
biology, in the history of civilizations, in forensic family, subfamily, group or genus, but can also be
sciences, etc. In general, analysis of colorants from common for several, quite different species (Fig-
plant and animal material give a few major and a lot ure 5). Species identiRcation is based on the relative
Figure 4 Continued
quantity and identity of marker compounds. Their in the sample. In the Rrst approach this can be evalu-
presence or absence is characteristic for only one ated as a function of colour intensity. Of course,
species. Given the complexity of colorants originating precise evaluation is more complex and should con-
from natural sources, as is the case for other natural sider the substrate (or matrix) material properties
substances, sometimes it should be satisfactory to such as surface-to-mass ratio, parameters of penetra-
obtain and to compare just their Rngerprints. tion of dyestuff in mass of matrix and others. Substra-
The chromatographic Rngerprints appearance de- te physicochemical behaviour can inSuence the
pends on sample preparation, separation conditions sample preparation method prior to analysis and
and detection principles. That is why the constitution extraction efRciency. Preparatory treatment may
of one’s own database containing the results of analy- modify dyestuff component proportions.
sis of reference samples according to established, in- By diminishing sample size, the analytical informa-
variable and repeatable procedure is required. tion about dyestuff composition is progressively lost,
starting with ‘disappearance’ from chromatograms of
Extra-Analysis Factors Affecting trace substances and continuing through the vague
appearance of minor components until the identiRca-
Dyes Composition tion of principal compounds is Rnally uncertain. As
The analytical results obtained from real, coloured ob- a result source recognition precision declines and
jects sometimes show important differences in dye com- becomes more difRcult to perform. Still, this effect
position in comparison with reference samples from does not affect the relative quantity of dyestuff com-
one’s own database or published data. These differences ponents. The lower limit of detection depends on
limit a clear interpretation and source identiRcation. performances of the chromatographic system and
Such a situation is a result of one or more factors, of detector used.
which the most important are quantity of sample, Another modifying factor, dyestuff preparation, af-
method of dyestuff preparation and ageing of the dye. fects many stages, from raw material extraction to
The quantity of coloured sample necessary for deposition on substrate. For each step, a multitude of
analysis depends on the total contents of the dyestuff recipes were invented using diverse plant parts and
Figure 5 Some reference samples dyed with Eurpoean plant extracts containing both luteolin (1) and apigenin (2) (detection at
345 nm). (A) Saw-wort (Serratula tinctoria L.); (B) Dyer’s Broom (Genista tinctoria L.); (C) Weld (Reseda luteola L.); (D) Daphne
gnidium L.
Figure 5 Continued
varying operation procedures, as well as the addition recent systematic study of the inSuence of different
of different organic and mineral substances to the operations and additives on dye composition restricts
dyestuffs extracts and dyebaths (Figure 6). Lack of interpretation.
Figure 6 Four contemporary reference samples of wool dyed with Madder (Rubia tinctorium L.) obtained from different dyers
(detection at 254 nm). Compounds: 1 pseudopurpurin, 2 alizarin, 3 xanthopurpurin, 4 purpurin, 5 unidentified anthraquinone.
A third possible dyestuff composition change deter- folds the conjugated impacts of many physical and
minant, which is very important in historical samples, chemical factors. Light is considered as the most
is ageing (Figure 7). The generic name ‘ageing’ en- important of them in dyestuffs degradation. In
Figure 6 Continued
comparison with other colouring matter (e.g. mineral tion is poor. Other important factors are water (even
pigments), the stability of dyes’ molecular structures as air humidity), air constituents (e.g. ozone), air
under visible light, and especially ultraviolet excita- pollutants (above all sulfur and nitrogen oxides) and,
Figure 7 Accelerated ageing of dyestuff extracted from Brazil (Caesalpinia sappan L.) (detection at 285 nm) (A) Freshly dyed wool;
(B) Sample after standardized accelerated ageing. Compound 1: brazilein derivative formed during acid hydrolysis.
for archaeological excavated objects, mineral salts global process. Determination of their particular roles
solutions giving dangerous pH values. Each of the is possible but does not allow exact modelling or
ageing elements presented represents just a part of the reproduction because of probable synergy between
them. In any case, the methods of ‘artiRcial’ ageing ation) and to their properties (stability, transforma-
(also called ‘accelerated’ ageing) give a good approxi- tion and degradation). They are the principal phe-
mation, sufRcient for comparative colour fastness nomena affecting chromatogram interpretation.
testing. Many of the compounds separated have not yet been
For textile fabrics it is important to mention the identiRed. Their identiRcation will thus certainly be
inSuence of washing and solvent cleaning on the the objective of future works.
presence of dyestuffs. Their resistance to wet cleaning
methods (wash-fastness) is principally related to the See also: II/Chromatography: Liquid: Detectors: Ultra-
dyeing group and washing conditions. Some dyestuffs violet and Visible Detection. Extraction: Solvent Based
or their components are sensitive to washing. Such Separation.
textile handling also belongs to ‘ageing’, but the ad-
jective ‘natural’ is in this case probably less adequate. Further Reading
During ageing two effects are generally observed:
hue change and fading. Change of hue is a result of Dzido TH, Soczewinski E and Gudej J (1991) Computer-
partial dyestuff fading, because of different degrada- aided optimization of high performance liquid chrom-
atographic analysis of Savonoids from some species of
tion rates of dye components. Chromatograms ob-
the genus Althea. Journal of chromatography 550: 71.
tained from such samples show qualitative and Evans KP and Truslove NJ (1993) Advances in chromatog-
quantitative differences from the original dyestuff. raphy for dyestuffs. Review of Progress in Coloration
Real sample composition can be ambiguous and sim- 23: 36.
ilar to several sources, especially if the markers are Fischer Ch-H, Bischof M and Rabe JG (1990) IdentiRcation
absent: in extremis it can be so far from any known of natural and early synthetic textile dyes with HPLC
reference that the question of a ‘new’, not yet identi- and UV/Vis spectroscopy by diode array detection. Jour-
Red source can appear. This kind of hypothesis al- nal of Liquid Chromatography 13(2): 319.
ways merits scrupulous veriRcation. Halpine SM (1995) An improved dye and lake pigment
Fading usually signiRes transformation of coloured analysis method for high-performance liquid chrom-
compounds into colourless products. Recent develop- atography and diode-array detector. Studies in Conser-
vation 41: 76.
ments in HPLC-MS provides a way to identify orig-
Justensen U, Knutsen P and Leth T (1998) Quantitative ana-
inal dyestuffs in discoloured samples from their lysis of Savonols, Savones, and Savanones in fruits, veg-
degradation products, even in small amounts. This etables and beverages by high-performance liquid
approach corresponds to expectations of specialists in chromatography with photo-diode and mass spectromet-
various domains and will certainly be developed in ric detection. Journal of Chromatography A 799: 101.
the future. Kirby J and White (1996) The identiRcation of red lake
pigment dyestuffs and a discussion of their use. National
Gallery Technical Bulletin 17: 56.
Conclusion Koren ZC (1994) HPLC analysis of the natural scale insect,
madder and indigoid dyes. Journal of the Society of
In spite of many historic and recent scientiRc works
Dyers and Colourists 110: 237.
concerning the chemistry and analysis of natural Nogata Y, Ohta H, Yoza K-I, Berhow M and Hasegawa
dyestuffs, our knowledge of this subject still seems to S (1994) High performance liquid chromatographic de-
be far from exhaustive. The contribution of HPLC in termination of naturally occurring Savonoids in Citrus
advances of chemotaxonomy of dyestuffs of plant with a photodiode-array detector. Journal of Chrom-
and animal origin has attracted growing interest in atography A 667: 59.
recent years. This technique is relatively easy to per- Schweppe H (1992) Hanbuch der Naturfarbstoffe. Land-
form and does not need any complex sample prepara- sberg am Lech: Ecomed.
tion. Its performances are sufRcient for good separ- Wouters J (1998) Qualitative and quantitative analysis of
ation of many dyestuff components in a reasonable tannins extracted from new and aged lethers, by high
time. The HPLC technique provides the possibility of performance liquid chromatography. Bulletin de l’In-
stitut Royal du Patrimoine Artistique 26: 199.
both qualitative and quantitative analysis of separ-
Wouters J and Verhecken A (1989) The coccid insect dyes:
ated compounds by employing detectors considered HPLC and computerized diode-array analysis of dyed
as standard in liquid chromatography (UV, UV-Vis, yarns. Studies in Conservation 34: 189.
PDA) or recently adapted, powerful techniques of Wouters J and Verhecken A (1991) High-performance
detection (MSD). liquid chromatography of blue and purple indigoid
Problems still remaining are directly linked to the natural dyes. Journal of the Society of Dyers and Col-
biosynthesis of natural dyes (possible pathway vari- ourists 107: 266.
III / DYES / Thin-Layer (Planar) Chromatography 2619
P. E. Wall, Merck Limited, Poole, Dorset, UK that dyes are easily visualized on a chromatographic
layer by their colour. Often slight differences in hue
are more clearly seen on the layer than in solution and
hence are easily distinguishable. It is therefore rarely
necessary to employ detection reagents unless the
Introduction area of interest is dye intermediates which may lack
Synthetic dyes comprise a large group of organic, the conjugation needed in their molecular structure to
organic salt and organometallic compounds which be coloured in visible light. Of course, there are
number in the thousands. Most individual dyes are a large number of dyes which either exhibit Suores-
assigned colour index (CI) numbers which helps to cence quenching in short wavelength ultraviolet (UV)
identify structure and properties as well as identity light (254 nm) or naturally Suoresce by excitation in
where a series of names have been used by the manu- long wavelength UV light (usually 366 nm). Where
facturers for the same dye. Pigments are also listed separated dyes on the chromatographic layer do Su-
and can be either organic or inorganic in nature. The oresce, the limit of sensitivity of detection is often in
value of planar chromatography in the identiRcation the low nanogram or high picogram level. In the
of dyes and separation of impurities and their quan- commercial environment, as dye quality can vary
tiRcation is the main subject of this article. Synthetic from batch to batch and colour can be matched by
dyes will be split into nine groups, the Rrst three being using different dyes, the planar chromatographic
the major ones. For all these groups, the development technique allows the analysis of many samples
of modern thin-layer chromatographic (TLC) against references or certiRed standards on the same
methods will be compared where possible with older layer under the same conditions in one development
TLC procedures. run. Hence the analysis time and the cost per sample
are substantially lower when compared with liquid
chromatography.
In most forms of chromatography, as in many
Use of TLC for the Separation of Dyes spectroscopic analyses, the extraction of the dye
Synthetic dyes and pigments are used in a wide range from whatever matrix is being considered and the
of industries. They are the colours of printing inks, subsequent sample pretreatment are often time-
textile dyes, cosmetics, plastics, histological and consuming, but usually necessary as impure dyes
cytological stains, and some are permitted as food notoriously contain many impurities that are easily
and drink colorants. In some of these applications and Rrmly retained on the chromatographic adsor-
purity is important, in others it is the presence or bents. Liquid chromatographic columns and pre-
absence of certain impurities or intermediates, and columns quickly become ‘poisoned’ and then either
in still others it is the identiRcation and uniformity have to be discarded or require extensive solvent
that is important. In all these instances planar clean-up before re-use. In TLC only minimal clean-up
chromatography has proved to be the ideal solution or extraction is required as most polar impurities
with the separation power and spot/zone capacity will remain at or near the origin of application and
necessary to resolve closely related dyes and inter- more nonpolar ones will migrate with the solvent
mediates. This has been made possible by the wide front (normal-phase separation). As the thin-layer
selection of stationary phases and the almost un- plate is not re-used, the unseparated material at the
limited variations possible with mobile-phase origin or at the solvent front is of no consequence.
mixtures, sometimes composed of quite aggressive However, the chromatograms obtained are a source
solvents that would not be used in column liquid of extensive data about the dye separation. Not
chromatography. The only limiting criteria for only can the chromatographic tracks be scanned
the solvent mixtures are adverse effects on the ad- spectrodensitometrically, but individual UV/
sorbent binder, reaction with the bonded phases visible spectra of the separated chromatographic
and high viscosity and surface tension of the zones can be recorded, and a number of other spec-
solvents. troscopic techniques such as mass spectrometry (MS),
There are a number of other advantages to the use nuclear magnetic resonance (NMR), Fourier trans-
of TLC for the analysis of dyes compared with other form infrared (FTIR) and Raman spectra can be
chromatographic techniques. The most obvious is applied.
2620 III / DYES / Thin-Layer (Planar) Chromatography
For use in histology, many reference standards are Table 1 Individual permitted food dyes (depending on country)
available, as tested and approved by the Biological
Permitted dye E number CI number
Stains Commission. Such dyes and stains are labelled
accordingly, usually giving the dye content for the Amaranth E123 16185
batch as numbered. Most such dyes are single compo- Brilliant black (black PN) E151 28440
Brilliant blue FCF E133 42090
nents, but where it is known to be composed of two
Carmoisine (azorubine) E122 14720
or more components (e.g. methylene violet (3)), azure Erythosine E127 45430
A or C (up to 6), methyl violet (4), aniline blue (3) and Ponceau 4R E124 16255
fast green FCF (3)), the dye content will be based on Indigo carmine E132 73015
the named dye (usually the major component). The Quinoline yellow E104 47005
Red 2G E128 18050
testing of dyes suitable for histology and cytology has
Sunset yellow FCF E110 15985
always relied on a number of wet chemistry tests, Tartrazine E102 19140
UV/visible spectral analysis, TLC analysis and the Yellow 2G (food yellow 5) E107
actual performance as a stain. The wet chemistry tests
include titanous chloride assay, sulfated ash and
moisture content. The total dye content has been also be a useful technique, as only a limited number of
determined by titration with standard titanous chlor- dyes are permitted for food use (Tables 1}4). These
ide solution and measurement of absorbance (UV). are indicated on the label of the food or drink product
High performance TLC (HPTLC) not only can be and can easily be identiRed by TLC. The presence and
used to determine accurately the total dye content, identiRcation of nonpermitted food dyes is one of the
but also the concentration of the individual compo- strengths of TLC.
nents if, as so often happens, the sample is either Many of the early planar chromatography methods
a mixture of dyes or there are impurities present for dye analysis in the 1960s used paper or cellulose
that are closely related structurally to the parent dye thin layers. Later, more methods, some still currently
(e.g. determination of crystal violet (hexamethyl- used, were based on silica gel G and aluminium oxide
pararosanilin) in methyl violet 2B, 3B, 6B or 10B thin layers. In more recent times, commercially avail-
(a mixture of various proportions of hexamethyl-, able pre-coated silica gel and cellulose layers with
pentamethyl-, tetramethyl-, and trimethyl-para- better reproducibility have become the preferred
rosanilin)). adsorbents for dye separations. This has become par-
In most industrial uses of dyes (textiles, printing ticularly the case where quantitative determinations
and plastics) and in forensic work, TLC is used for are required for dye content on HPTLC plates.
identiRcation and on some occasions to quantify
these results. In the commercial environment textile
dyes are not normally marketed in the pure state. Table 2 Recommended solvents/solvent mixture for the separ-
ation of lipophilic solvent (fat) dyes on silica gel 60 pre-coated
Both the end user and the manufacturer of the dyes
TLC plates
are mainly interested in a dye of standard reproduc-
ible quality in their application rather than chemical Solvent dye CI numbers Solvents for
purity. For this reason the TLC analysis of the dyes separation a
for textile, cosmetic or printing purposes presents Sudan black B 26150 Dichloromethane
different problems to that for staining purposes in Fat red 7B 26050 Toluene
histology. A commercial dye will often contain impu- Sudan I 12055 Heptane}ethyl
rities composed of by-products of the dye synthesis acetate (80#10 v/v)
Sudan II 12140 Cyclohexane}ethyl
and starting materials. Inorganic salts are usually
acetate (90#10 v/v)
present: these have been used to ‘salt out’ or precipi- Sudan III 26100
tate the dye during manufacture or they may have Sudan IV 26105
been added as extenders, so that different batches Butter yellow 11020
have the same dyeing potential. Variations in colour Sudan red G 12150
Indophenol blue 49700
hue are sometimes adjusted by the addition of an-
Diethyl yellow 11021
other dye. Oil blue APS 61551
In histology and cytology, although the identity of Waxoline green G-FW 61565
the dye is important, the purity plays an important Waxoline red MP-FW 60505
part. It has also been realized that other dye impu- Waxoline yellow E 47000
Oil red O 26125
rities need to be present to give a superior quality
stain. Both Giemsa and Leishman’s stain are classic a
These recommended solvents and solvent mixtures apply to all
examples of this. In food and drinks, TLC of dyes can the solvent dyes.
Table 3 RF values for a number of solvent (fat) dyes developed on silica gel 60 pre-coated plates
Solvent dye RF values (dichloromethane) RF values (n-hexane}ethyl acetate (90#10 v/v))
Diethyl yellow 0.72

Oil blue APS 0.34
Waxoline green G-FW 0.64
Waxoline red MP-FW 0.54
Waxoline yellow E 0.12
Sudan black B 0.10 (f), 0.18 (f), 0.22 (m),
0.61 (m), 0.85 (f)
Waxoline blue 0.43 (m), 0.13, 0.79
Sudan I 0.68
Sudan II 0.2
Sudan III 0.56
Sudan IV 0.56
Sudan orange G 0.14
Sudan R 0.18
f, Faint zone; m, main zone.
For the purposes of planar chromatography, separ- of solvents employed for the TLC separations of dyes
ations of synthetic dyes can be split into a number of within each group is similar and notably different
groups depending on the ionic nature and molecular from those used for other groups.
structure of the dye. These are:
1. Solvent (fat) dyes.

Solvent (Fat) Dyes
2. Basic dyes. As their name suggests, solvent dyes are dyes soluble
3. Acid dyes. in nonpolar solvents (mostly water-immiscible),
4. Reactive dyes. and are used to colour mineral oils, waxes, fats and
5. Disperse dyes. plastics. They are lipophilic in nature and do not form
6. Metal complex dyes. organic salts. Structurally the dyes are characterized
7. Direct dyes. by the azo, quinone-imine (indophenol dyes), or
8. Organic pigments. anthraquinone (waxoline dyes) groups they contain.
9. Food colorants. By far the largest group is the azo dyes. These are
mainly monoazo (sudan dyes) and a few disazo dyes
As mixtures of dyes generally fall into these re- (e.g. Sudan red B (CI 26110), oil red O (CI 26125;
spective groups and it is rare, if ever, that a dye from Figure 1).
one of the major structural groups is found mixed Although TLC separations of solvent dyes have
with another, the TLC will be examined in detail for been reported on alumina layers, silica gel is by far
each group in turn. This is important as the polarity the preferred adsorbent. Separation of solvent dyes
Figure 1 Solvent dyes. Structures of characteristic groups.

Table 4 RF values for Sudan III and IV on silica gel 60 RP8 solvents to those used on silica gel G. However, the
HPTLC pre-coated plates. Plate developed in a saturated chamber separated chromatographic zones on pre-coated
Solvent dye RF value Developing solvent plates are much sharper, noticeably less diffuse and
misshapen, leading to good overall resolution
Sudan III 0.32 Methanol}acetic acid}water (Figures 2 and 3). This should be expected as the
Sudan IV 0.22 (90#5#5 v/v)
silica gel will have a much narrower particle size
distribution over the particle size range, and a mean
particle size (10}12 m) less than that of a ‘home-
on standardized silica gel G layers was reported in the made’ plate.
early 1960s by Stahl. This was one of the Rrst re- The resolution can be further improved for Sudan
corded separations on silica gel and described the dyes by using a reversed-phase adsorbent. In fact, the
resolution of butter yellow (CI 11020), Sudan R (CI resolution is so good that results can be quantiRed on
12150) and indophenol (CI 49700) using a mobile an HPTLC silica gel 60 RP8 plate (Merck) by scann-
phase of pure benzene. In the years that followed, ing the developed plate at 540 nm (Figure 4). In this
a variety of similar-polarity solvents were reported as example, samples were applied as spots of 100 nL
suitable with silica gel G. Chloroform and benzene with a manual nanoapplicator (CAMAG). The rela-
and mixtures of petroleum spirits, and single aliphatic tive standard deviation was 4.5%.
hydrocarbons with ethyl acetate, diethyl ether, ace-
tone and methanol were used to separate butter yel-
low, Sudan orange R (CI 12055), Sudan III (CI
26100) and Sudan red BB (CI 26105). The anthra-
quinone series of dyes have been separated with tol-
uene}cyclohexane (50#50% v/v) on silica gel G.
This readily resolves solvent blue 36, waxoline purple
A (CI 60725) and waxoline green G (CI 61565).
Dimethylaminoazobenzene and related dyes are sep-
arated using the solvent chloroform}methanol
(95#5% v/v) on the same stationary phase.
Most solvent dyes will readily migrate and separate
from each other on commercially pre-coated silica gel
60 TLC layers using the same or similar polarity
Figure 2 Separation of two main components of Sudan black

B on silica gel 60 pre-coated TLC plates. Developed in
a saturated tank with dichloromethane as solvent. Scanned with Figure 3 Separation of fat red 7B from one main impurity on
a spectrodensitometer at 540 nm. silica gel 60 pre-coated TLC plates.
Figure 5 Acid and basic dyes. Structures of characteristic

groups.
Figure 4 Pure Sudan IV resolved on silica gel 60 TLC plates Table 5 Commonly named basic dyes according to their struc-
using n-hexane}ethyl acetate (90#10 v/v) as mobile phase in tural groups
a saturated tank.
Basic dye groups Commonly named dyes
Triaminotriarylmethane Pararosanilin
Basic Dyes New fuchsin
Rosanilin
Basic dyes form one of the larger areas of interest Diaminotriarylmethane Malachite green
when it comes to TLC. As their name suggests, they Brilliant green
are bases which can form salts which are soluble in Victoria blue
Monoaminotriarylmethane Methyl violet
water. Their applications vary widely in industry,
Crystal violet
including colouring paper, some textiles and plastics, Ethyl violet
and are used as the basis of many biological stains. Acridine Acridine red
Basic dyes include the following structural groups: Acridine orange
triarylmethanes (mostly triphenylmethanes), quinon- Xanthene Pyronin B
Pyronin G
eimines, azo, acridine and xanthene compounds (Fig-
Rhodamine B
ure 5). These can be further split: the triarylmethanes Rhodamine 6G
into amino and hydroxy derivatives and the quinone- Azin Safranine O
imines into indamins, azins, thiazins and oxazines. Oxazine Brilliant cresyl blue
Table 5 shows some of the commonly named dyes Orcein
Resazurin
which Rt into these groups.
Gallocyanine
As with solvent dyes, much of the early separation Cresyl fast violet
work was performed on silica gel G. Table 6 lists Nile blue
a number of solvent mixtures which have proved Cellestine blue
useful in the separation of the general classes of basic Thiazin Azure A, B, C
Methylene blue
dyes. With the advent of commercial TLC and
Toluidine blue
HPTLC pre-coated silica gel 60 and bonded phases, New methylene blue
the separation methods have improved, allowing res- Thionine
olution of dyes with closely related structures, and the Methylene violet
determination of concentrations of individual dyes by Azo Janus green
Chrysoidin
in situ scanning using a spectrodensitometer (Table 7).
Bismarck brown R and Y
Hence thiazin dyes are now easily resolved, including
Table 6 Solvent systems recommended for the separation of basic dyes on silica gel 60 pre-coated TLC plates and sheets
Basic dyes Solvent mixtures
Malachite green, methyl violet Butan-1-ol}ethanol}water (90#10#10 v/v)

Basic fuchsin, rhodamine B, rhodamine 6G Butan-1-ol}acetic acid}water (40#10#50 v/v)
Methylene blue, victoria blue B, crystal violet, malachite green, rhodamine B Butanone}acetic acid}propan-2-ol (40#40#20 v/v)
Pyronine G, acridine red, rhodamine B Propan-1-ol}formic acid (80#20 v/v)
Methylene blue, malachite green Ethyl acetate}acetic acid}water (30#10#20 v/v)
Crystal violet, methylene blue, malachite green, basic fuchsin Butan-1-ol}acetic acid}water (20#5#10 v/v)
thionine, azure A, B, C and methylene blue on shown in Figure 7 (separation of the blue dye compo-
HPTLC silica gel 60 using a mobile phase of butan-1- nents of Wright’s stain) on a normal-phase pre-coated
ol}butanone}acetic acid}water (30#20#10#10 silica gel 60 plate (Merck).
v/v). Methylene blue can easily be distinguished from For most basic dyes, TLC and HPTLC silica gel 60
new methylene blue even though their visible/UV layers normally give sufRcient resolution and have
spectra are almost identical (Figure 6). proved in some cases to be suitable even for the
Many of these dyes are the basis of the biological separation of dyes from closely related impurities
Romanowski stains (Giemsa, Leishman, Wright, Jen- (Figure 8). In special cases, either reversed-phase
ner and May-Grunwald). An assessment of the dye layers or other bonded layers have solved particularly
quality of the stain and its individual dye components difRcult separation problems. One of these has been
can be made by the above chromatographic methods. the separation of pararosanilin from rosanilin, new
Usually the ‘ripened’ or oxidized versions of these fuchsin and magenta II. Pure pararosanilin is used in
stains produce the better histological staining proper- the preparation of Feulgen and Schiff’s reagent for
ties. Oxidation produces more azures and thionine, aldehyde and ketone detection and in biological stain-
all of which can be determined, if required, by ing. All four compounds are triaminotriphenyl-
HPTLC procedures. The complete resolution of methane dyes, differing consecutively by one methyl
methylene blue and its oxidation products including group. Although silica gel 60 RP8 HPTLC layers are
azures and thionine presents a difRcult problem even able to resolve just the four components, tailing prob-
for HPTLC as these dyes are all closely structurally lems make it difRcult to scan the zones quantitatively.
related. A typical example of such a separation is However, by introducing the ion-pairing reagent
Table 7 Solvent systems for individual basic dyes on silica gel 60 TLC and HPTLC pre-coated plates. RF values obtained and
recommended scanning wavelength for detection
Basic dye Solvent mixture RF value Detection wavelength
Brilliant cresyl blue Ethyl acetate}methanol}ammonia (0.88)}water 0.56 605 nm

(35#11#5#5 v/v)
Crystal violet Butan-1-ol}acetic acid}water (50#5#10 v/v) 0.39 620 nm
Ethyl violet 0.58 585 nm
Acridine orange Ethyl acetate}methanol}ammonia (0.88)}water 0.64 360 nm (fluorescence)
(33#11#5#5 v/v)
Azure B Butan-1-ol}acetic acid}water (50#5#10 v/v) 0.12 (m), 0.19 (m), 0.29 (m), 620 nm
0.33, 0.40
Brilliant green 0.46 620 nm
Malachite green 0.36 620 nm
Methyl violet 6B 0.39 (m), 0.46, 0.51 620 nm
Methylene blue Butan-1-ol}acetic acid}water (50#10#20 v/v) 0.29 (m), 0.34 620 nm
New methylene blue Butan-1-ol}acetic acid}water (50#5#10 v/v) 0.32 610 nm
Nile blue 0.48 620 nm
Orcein Ethyl acetate}methanol}ammonia (0.88)}water 0.44 (m), origin, 0.19, 0.26, Visual
(35#11#5#5 v/v) 0.37, 0.39, 0.41, 0.42, 0.47,
0.48, 0.49, 0.54
Pyronin G 0.27 360 nm (fluorescence)
Rhodamine B Butan-1-ol}acetic acid}water (50#5#10 v/v) 0.48 540 nm
Safranine O 0.45 (m), 0.53 540 nm
Thionine Butan-1-ol}acetic acid}water (50#10#20 v/v) 0.50 620 nm
m, Main zone.
Figure 6 Comparison of methylene blue separation with new methylene blue. Both separated on silica gel 60 pre-coated TLC plates
with butan-1-ol}acetic acid}water (50#5#10 v/v) as mobile phase. Overlaid spectra also shown.
pentanesulfonic acid sodium salt (2%) into the devel- stains and are used for colouring wool, polymer
oping solvent mixture (methanol}water}formic acid: Rbres, leather and paper. They are mostly sulfonic
75#20#5 v/v), the effect of tailing is dramatically and carboxylic acid compounds, often prepared as
reduced, allowing quantitative determination of all their sodium salts to enable good water solubility.
four dyes. Another difRcult separation problem arises They can be split into a range of structural groups:
with Bismarck brown Y and R. Resolving the dyes azo, triarylmethane, xanthene, anthraquinone and
from their impurities is not easily achieved. In this quinone-imine. The separation of these groups on
instance silica gel 60 NH2-bonded HPTLC layers silica gel G and cellulose using a variety of solvent
have proved successful (Figure 9). mixtures for development is given in a number of the
older TLC books found in the Further Reading list.
Commercial pre-coated silica gel 60 TLC and
Acid Dyes HPTLC plates, where the adsorbent has a carefully
Like basic dyes, acid dyes are used widely throughout controlled pore size and particle size, give improved
industry. They are the basis of inks and biological resolution of acid dyes compared with that obtained
deviation. It is also possible to determine where rogue

dyes have been used to try and replace the genuine
dye. Examples of Congo red and brilliant cresyl blue
are shown in Plates 1 and 2.
More recently, reversed-phase silica gel layers have
proved useful for acid dyes where separation has
not been possible on normal-phase silica gel. An
example of this is eosin bluish shade, where reversed-
phase plates have proved satisfactory, as shown in
Table 8.
Reactive Dyes
Reactive dyes attach themselves to cellulose Rbres in
wool and cotton. They are structurally deRned into
the groups azo, anthraquinone and phthalocyanine
derivatives. One well-known commercial group is the
procion dyes which can be separated on silica gel
G with solvent mixtures } butan-1-ol}propan-1-
ol}ethyl acetate}water (20#40#10#30 v/v) and
Figure 7 Separation of blue components of Wright’s stain. Sta- dioxane}acetone (50#50 v/v).
tionary phase is silica gel 60 pre-coated TLC plate. Mobile phase Hydrolysis products of reactive dyes have also been
is ethyl acetate}methanol}ammonia solution (0.88)}water examined by TLC on silica gel using the solvent
(35#11#5#5 v/v). Mixture of methylene blue, azures A,
mixture butan-1-ol}pyridine}water}ammonia (0.88
B and C, and thionine and other oxidation products.
solution: 15#5#3#2 v/v), and with solvent mix-
ture chloroform}methanol}acetone (18#3#2 v/v)
for example with silica gel G. It has been possible to with detection at 310 and 614 nm using a spectroden-
quantify results by spectrodensitometric scanning in sitometer. The TLC of the reaction product and hydro-
a comparable way to basic dyes (Figure 10). Purity lytic by-products of terminal ring isomers of reactive
can be determined with a high degree of accuracy as is blue 2 (an anthraquinone dye) have been studied on
shown with the commercial samples of light green silica gel using butan-1-ol}propan-2-ol}ethyl acetate}
(Figure 11). Typically, Rve standards of certiRed dye water (20#40#10#30) as developing solvent.
are developed on the same plate alongside the sam-
ples. Peak measurements are made by reSectance and
an x-power curve plotted against concentration. Un-
Disperse Dyes
knowns are determined with about $1% standard Similar in structure to fat-soluble dyes, disperse dyes
are mainly used to colour cellulose acetate and poly-
mer Rbres (principally polyester). They exhibit low
solubility in water, but are soluble in organic sol-
vents. Disperse dyes are types of nitrodiphenylamine,
azo and anthraquinone dyes, but not those associated
with either sulfonic or carboxylic acid groups. Com-
mercially they belong to the Celliton group of dyes
which can be separated by TLC on silica gel G adsor-
bent using chloroform}methanol (95#5 v/v) as mo-
bile phase. Better separations of amino derivatives of
anthraquinones have been obtained using chloro-
form}acetone (90#10 v/v). Mixtures of chloro-
form}ethyl acetate (50#50 v/v) or various dilutions
of toluene with acetone can be employed on pre-
coated silica gel 60 layers. When analysis of the dyes
Figure 8 Separation of crystal violet from other closely related
is required in natural and synthetic polymers, di-
less methylated pararosanilins. Stationary phase: silica gel 60
pre-coated HPTLC plates. Mobile phase: butan-1-ol}acetic chloromethane, acetone or dimethylformamide can
acid}water (50#5#10 v/v). Peak 1: hexamethylpararosanilin be used to extract them ready for application to the
(crystal violet), peak 2: pentamethylpararosanilin. TLC plate.
Figure 9 Separation of Bismarck brown Y and R on silica gel 60 NH2 pre-coated HPTLC plates. Mobile phase: industrial methylated
spirit. (A) The two components of Bismarck brown Y; (B) Bismarck brown R.
Metal Complex Dyes 1 : 2 Complexes

These are acid dyes, mostly belonging to the azo These are water-insoluble, but solvent-soluble dyes.
group. They can be divided into two distinct types: They are metal complexes of chromium and cobalt.
1 : 1 and 1 : 2 complexes. The best adsorbent for separation on thin layers is
polyamide using a solvent mixture of methanol}
1 : 1 Complexes water}ammonia (0.88 solution: 80#16#4 v/v).
These are water-soluble dyes of sulfonic acid type
which can be separated by TLC on silica gel using the
solvent systems described under the section on acid
Direct Dyes
dyes. For the most part they are chromium complex Direct dyes are used for dyeing cotton Rbres and
dyes. leather. They are mainly polyazo dyes which can be
Figure 10 Quantification of Congo red dye. Single component dye. Typical x -power fit for reflectance data against concentration for
a batch of certified dye. Stationary phase: silica gel 60 pre-coated HPTLC plates. Mobile phase: ethyl acetate}methanol}ammonia
solution (0.88)}water (35#11#5#5 v/v).
Figure 11 Quantification of light green on silica gel 60 pre-coated HPTLC plates. Six standards applied as spots of nL loadings using
a certified light green dye. The pure dye content of a number of commercial light green batches are listed as determined using this
chromatographic procedure.
well separated on silica gel layers. Some, like Congo Organic Pigments
red, Evan’s blue and trypan blue, have already been
These are the colours of artist’s paints which are also
discussed under the acid dye section, but others can
used for pigmenting rubbers and plastics. Included
be resolved using the same solvent mixtures as recom-
mended for the above dyes in Table 9. Slight modiR-
cations to the ratio of the solvents can often further
optimize the solvent mixture for the speciRc dye.
Figure 13 (See Colour Plate 79). Separation of brilliant cresyl

blue on silica gel 60 HPTLC plate. Mobile phase: ethyl acetate}
Figure 12 (See Colour Plate 78). Set of certified Congo red methanol}ammonia solution (0.88)}water (35#11#5#5 v/v),
standards developed on a silica gel 60 HPTLC layer. Mobile saturated tank. Six standards applied at increasing concentration
phase: ethyl acetate}methanol}ammonia solution (0.88)}water (from 50 ng) on the right side of the plate. The six samples
(35#11#5#5 v/v). A minor impurity is visible ar RF value 0.07. developed on the left, all of equal concentration, were claimed to
The two dye substances alongside these are of a dye that was be brilliant cresyl blue, but in fact were a mixture of toluidine blue
claimed to be Congo red. and Nile blue sulfate.
Table 8 Commonly named acid dyes according to their struc- of many of these dyes is listed in Table 10. These dyes
tural groups are all water-soluble and separate readily on thin
Acid dye groups Commonly Colour index layers of silica gel 60. As they are ionic in nature, it is
named dyes numbers important that either acid or base modiRers are used
in the solvent mixture (e.g. acetic or formic acids or
Azo Methyl red 13020
ammonia solution). This reduces ‘tailing’ or ‘streak-
Methyl orange 13025
Tropaeolin O 14270 ing’ of chromatographic zones by either suppressing
Orange II 15510 ionization or totally ionizing the components of the
Ponceau 2R 16150 sample.
Orange G 16230 Commercial dyes for food use are often blends of
Chromotrope 2R 16570
permitted acid dyes, e.g. green food dye is usually
Tartrazine 19140
Congo red 22120 a mixture of brilliant blue FCF (E133) and tartrazine
Trypan blue 23850 (E102); red food dye can be a mixture of red 2G
Evans blue 23860 (E128) and tartrazine (E102). The TLC methods
Ponceau S 27195 must therefore have the ability not only to separate
Triarylmethane Acid fuchsin 42685
impurities, but also to resolve the parent dyes.
Fast green FCF 42053
Light green SF 42095 Most food dyes are prepared for use in a glycerine
Alkali blue 5B 42750 base. Although little pre-chromatographic prepara-
Aniline blue 42755 tion of samples is required, it is important where this
Xylene cyanol FF 43535 base is used to solubilize these dyes further with
Xylenol orange
water}methanol (50 : 50 v/v) so that when the sample
Sulfonphthaleins
Xanthene Fluorescein 45350 is applied to the plate, the sample solution will
Eosin Y 45380 properly ‘wet’ the layer. Solutions can be Rltered if
Phloxine 45405 required. As the dyes are so readily soluble in water,
Erythrosine B 45430 their extraction from the food matrix does not usually
Rose bengal 45440
present a problem. For colours in drink, the sample
Quinone-imine Azocarmine G 50085
Azocarmine B 50090 normally only requires dilution with methanol before
Anthraquinone Alizarin red S 58005 application as a spot or band to the chromatographic
Indigo carmine 73015 layer.
here are the phthalocyanine dyes along with some azo,

Future prospects
vat and anthraquinone dyes. They are for the most Future prospects of dye analysis using TLC/HPTLC
part insoluble in both organic solvents and water. are likely to be closely linked with the use of video
Hence chromatographic separations have proved dif- imaging techniques (charged coupled devices). Usu-
Rcult. However, some chromatographic work has ally dye analysis involves many samples in a commer-
been described. Some pigments will dissolve in hot cial environment. Hence a technique that has the
dimethylformamide. They can then be applied as such ability to take a video image of an entire developed
to silica gel layers. Thorough drying is recommended plate which could contain many chromatograms, and
before development with benzene or benzene modi- has the software capabilities to determine concentra-
Red with chloroform. The use of paper chromatogra- tions with some degree of accuracy from the image
phy has also been described with propan-1-ol}ammo- density of the chromatographic zones, has a real
nia (0.88 solution: 20#10 v/v) as mobile phase. beneRt to the analyst. Hard-copy pictures of the
chromatograms can be produced for entry into ana-
lytical reports. The speed of such analysis will un-
Food Colorants doubtedly improve as the speed of computation be-
Acid dyes are used as colorants in food and drink. In comes ever faster and further software improvements
the past, the recommended stationary phase for food are made.
dye separations was either cellulose or modiRed cellu- The identiRcation and possible quantiRcation of
loses (diethylaminoethyl (DEAE), PAB). However, as minor dye impurities will also become more feasible
in other areas of modern TLC, silica gel 60 has taken as hyphenated techniques such as FTIR, MS and
over as the preferred stationary phase. The optimum Raman spectroscopy are linked to HPTLC and be-
chromatographic conditions have therefore to a large come more widely used. Manufacturers of pre-coated
extent already been described in the acid dye section. plates are now commercially producing smaller par-
However, for completeness the optimum resolution ticle size adsorbents (3}5 m), some based on spheri-
Table 9 Solvent systems for individual acid dyes on silica gel 60 TLC and HPTLC pre-coated plates. RF values obtained and
recommended scanning wavelength for detection
Acid dye Solvent mixture RF values Detection

wavelength
Alizarin red S Butan-1-ol}acetic acid}water (50#10#20 v/v) 0.29 (m), 0.22, 0.46 Visual
Aniline blue 0.22 (m), 0.39 (m), 0.08, 620 nm
0.27, 0.32, 0.49, 0.64
Fast green FCF 0.31 620 nm
Light green SF 0.19 620 nm
Orange G 0.28 470 nm
Alkali blue 5B, 6B Ethyl acetate}methanol}ammonia (0.88)}water 0.28, 0.35, 0.41, 0.47, 0.52, 540 nm
(35#11#5#5 v/v) 0.55, 0.87, 0.91
Congo red 0.47 470 nm
Crystal ponceau Butan-1-ol}IMS}water}ammonia (0.88) 0.45 470 nm
(50#25#25#10 v/v)
Ponceau S 0.37 470 nm
Indigo carmine 0.53 620 nm
Acid fuchsin 0.46 (m), 0.33, 0.42 540 nm
Eosin Y Ethyl acetate}methanol}ammonia (0.88)}water 0.42 (m), 0.38 470 nm
(33#11#5#5 v/v)
Erythrosin B 0.43 540 nm
Naphthol yellow S 0.33 (m), 0.44 (f) 400 nm
Evans blue Ethyl acetate}methanol}ammonia (0.88)}water 0.16 (m), 0.42 (f) 620 nm
(35#11#5#6 v/v)
Fast garnet GBC salt Butan-1-ol}acetic acid}water (50#5#5 v/v) 0.82 (m), 0.48 (f) 400 nm
Trypan blue Ethyl acetate}methanol}ammonia (0.88)}water 0.11 (m), 0.32, 0.42 620 nm
(35#11#5#6 v/v)
Azocarmine G Ethyl acetate}methanol}ammonia (0.88)}water 0.28 540 nm
(35#11#6#4 v/v)
Phloxine Ethyl acetate}methanol}ammonia (0.88)}water 0.38 540 nm
(33#11#2#2 v/v)
m, Main zone; f, faint zone.
cal shaped particles. These will have the advantage of will be more concentrated with less diffusion than
better Sow characteristics for the migrating solvent, that presently seen on an HPTLC plate.
although slower Sow rates and, most importantly for
See Colour Plates 77, 78, 79.
dye impurity analysis, improvement in sensitivity.
The smaller average particle size and spherical shape See also: II / Chromatography: Thin-Layer (Planar):
will mean that such minor components on the layer Densetometry and Image Analysis; Mass Spectrometry.
III / Thin-Layer Chromatography-Vibration Spectro-
Table 10 Solvent systems and RF values for food dyes separ- scopy. Dyes: High-Speed Countercurrent Chromatogra-
ated on silica gel 60 pre-coated TLC plates phy; Liquid Chromatography.
Food dye Solvent system RF value

Further Reading
Erythrosin Ethyl acetate}methanol} 0.43
ammonia (0.88 solution)}water The Colour Index, 3rd edn (1982) Bradford, UK: Society of
(33#11#5#5 v/v) Dyers and Colourists.
Sunset yellow 0.33 Gupta VK (1990) Synthetic dyes. In: Sherma J and Fried
Tartrazine 0.11 B (eds), Handbook of Thin-layer Chromatography, pp.
Indigo carmine Butan-1-ol}IMS}water} 0.53 939}969. New York: Marcel Dekker.
ammonia (0.88 solution: Liedekerke BM, Leenheer AP and De Spiegeleer BM (1991)
50#25#25#10 v/v)
High-performance thin-layer chromatographic analysis
Red 2G Ethyl
of thiazine dyes on silica, cyano, and reversed-phase
acetate}methanol}water}
ammonia (0.88 solution: C18 layers. Journal of Chromatographic Science 29:
30#15#5#10 v/v) 49}53.
Brilliant blue FCF Loach KW (1971) Thin-layer chromatographic separation
of methylene blue and related thiazine dyes. Journal of
IMS, industrial methylated spirits. Chromatography 60: 119}126.
III / ECDYSTEROIDS: CHROMATOGRAPHY 2631
Marshall PN and Lewis SM (1974) A rapid thin-layer Recent Advances in Thin-Layer Chromatography,
chromatographic system for Romanowsky blood stains. pp. 207}210. New York: Plenum Press.
Stain Technology 49: 235}240. Wall PE (1989) HPTLC as a quantitative method for the
Randerath K (1963) Thin-Layer Chromatography, determination of the purity of dyes of histological impor-
pp. 211}214. London: Academic Press. tance. Journal of Planar Chromatography 2: 246}247.
Stahl E (ed) (1969) Thin-Layer Chromatography: Wall PE (1991) Thin layer chromatographic separation of
A Laboratory Handbook. Berlin: Springer- thiazins: problems and solutions. Journal of Planar
Verlag. Chromatography 4: 365}369.
Wall PE (1988) Separation and quantiRcation of Fuchsin Wall PE (1993) The value of planar chromatography for the
Basic using reversed-phase thin-layer chromatography. analysis of triphenylmethane dyes. Journal of Planar
In: Dallas FAA, Read H, Ruane RJ and Wilson ID (eds) Chromatography 6: 394}403.
ECDYSTEROIDS: CHROMATOGRAPHY
R. Lafont and C. Blais, Ecole Normale Supe& rieure Solvent Partitioning

et Universite& Pierre et Marie Curie, Paris,
France Partition between n-butanol and water can be used to
J. Harmatha, Academy of Sciences of the Czech remove polar contaminants, whereas partition be-
Republic, Prague, Czech Republic tween aqueous methanol and hexane removes non-
I. D. Wilson, AstraZeneca Pharmaceuticals Ltd, polar materials. Lipids can also be removed from
Macclesfield, Cheshire, UK aqueous extracts with hexane}methanol (7 : 3, v/v),
light petroleum or n-propanol}hexane (3 : 1, v/v).
The nature of the contaminants to be removed and
that of the ecdysteroids to be isolated govern the
Introduction choice for a given partition system. The number of
Ecdysteroids are present both in animals (mainly free -OH groups signiRcantly affects partition coefR-
Arthropods) and plants and comprise about 300 cients (Table 2).
different molecules related to ecdysone (Figure 1). The combination of two successive partition steps
Structural variation in the number of carbons allows the elimination of both polar and apolar con-
on the side-chain and of substituents at various taminants. It is thus possible to combine (1) chloro-
positions (Table 1) results in the presence of form/water and (2) water/butanol. This results in
compounds displaying very different polarities. a butanol-containing fraction that is signiRcantly
Most available chromatographic techniques have enriched. It is possible to select a narrower range of
been applied to the isolation and analysis of polarity by replacing chloroform with a more polar
ecdysteroids. Paper chromatography is now obsolete organic solvent, e.g. isobutyl acetate, that neverthe-
and no longer used. Gas chromatography (GC) is of less allows ecdysteroids to remain in the water phase.
limited use, as the derivatization procedures neces-
sary to make volatile derivatives require careful
control. Currently, the most widely used techni-
ques are high performance liquid chromatography
(HPLC) and thin-layer chromatography (TLC),
with the former providing the major analytical
methods.
Liquid^Liquid Partitions
The simplest separation method concerns partition-
ing between two non-miscible solvents, and it is
currently used for clean-up of biological samples. On
this basis, several procedures have been designed,
which allow the preparation of almost pure com- Figure 1 The structure of ecdysone, the first ecdysteroid iso-
pounds in the gram scale. lated from Bombyx mori pupae.
2632 III / ECDYSTEROIDS: CHROMATOGRAPHY
Table 1 Variations on the 20-hydroxyecdysone molecule Table 2 Partition coefficients (K ) of ecdysteroids (data from
Lafont et al ., 1994b)
Type Positions on the molecule
Ecdysteroid K
Hydroxyl groups
Additional IOH 1, 5, 11, 16, 18, 19, 23, 24, 26 CyclohexaneIn-butanolIwater (5 : 5 : 10)
Missing IOH 2, 20, 22, 25 Ecdysone 3.54
Makisterone A 1.27
Oxidation 20-Hydroxyecdysone 0.52
('CHOHP'C"O) 3, 22 3-Epi-20-hydroxyecdysone 0.52
(ICH2OHPICOOH) 26 26-Hydroxyecdysone 0.39
(Pepoxide) 22I23 20,26-Dihydroxyecdysone 0.06
Epimerization 3/3, 5/5 ChloroformImethanolIwater (2 : 1 : 1)
Alkyl substitution 24 (methyl, methylene, ethyl, etc.) 2,22-Dideoxyecdysone 13.0
2-Deoxyecdysone 2.7
Esterification Ecdysone 0.4
Acetates 2, 3, 22, 25 20-Hydroxyecdysone 0.1
Fatty acyls 22
Benzoates 20, 22, 25 concentration in the non-polar phase
K"
Cinnamates 2 concentration in the polar phase
Coumarates 3
Phosphates 2, 22, 26
Sulfates 22
Lactone ring formation Concerns mainly C-28 or C-29 mobile droplets and the stationary phase is the
ecdysteroids rate-limiting process. High-speed counter-current
chromatography (HSCCC) overcomes this drawback
Etherification
Intramolecular Between C-22 and C-25 and separations are performed within a few hours.
Methoxy ether 25 This technique has so far only been applied to the
ecdysteroids in a small number of cases.
Ketal/acetal formation
Acetonides 2I3, 20I22
Benzylidene acetals 20I22
Thin-Layer Chromatography (TLC)
Glycosylation
Galactosides 3, 22 Normal-phase (absorption) chromatography on silica
Glucosides 3, 22, 25, 26 gel has been used extensively in the isolation of ecdys-
Dehydration 9(11), 14(15), 24(25), 25(26) teroids and for metabolic work. Despite the advent
Side-chain cleavage C-20/ C-22, C-17/C-20
of HPLC, the low expense, simplicity and speed of
TLC ensures a continuing role for this technique in
ecdysone research.
Counter-current Distribution Chromatographic Procedures
Counter-current distribution (CCD) is a multi-tube Normal-phase systems Many solvent systems have
extension of the above partition procedure. Bu- been used for TLC of ecdysone and related com-
tenandt and Karlson (1954) puriRed the Rrst ecdys- pounds, and these are summarized in Table 4. A wide
teroid (ecdysone) from Bombyx pupae by CCD with range of RF values on silica gel have been reported.
butanol}cyclohexane}water (4 : 6 : 1). This technique
is presently of limited use, and it has been replaced by
the more convenient droplet counter-current chro- Table 3 Solvent systems for droplet counter-current chrom-
matography technique described below. atography (DCCC)
Solvent system Mode

Droplet Counter-current Chromatography
CHCl3 /MeOH/H2O (13 : 7 : 4) Ascending
Droplet counter-current chromatography (DCCC) CHCl3 /C6H6 /EtOAc/MeOH/H2O Descending
allows an efRcient puriRcation of crude samples up (45 : 2 : 3 : 60 : 40)
to the gram range (Table 3). DCCC enables the C6H6 /CHCl3 /MeOH/H2O (5 : 5 : 7 : 2) Descending
preparation of reasonably, although not absolutely,
CHCl3 /MeOH/H2O (65 : 20 : 20) Descending
pure compounds (a subsequent HPLC step may be
required to get pure ecdysteroids). One DCCC sepa- CHCl3, chloroform; MeOH, methanol; C6H6, benzene; EtOAc,
ration usually lasts several days: exchanges between ethyl acetate.
Table 4 Some representative solvent systems for TLC of

ecdysteroids on silica gel
Solvent system Composition [RF]
E 20E
CHCl3/95% EtOH 7:3 0.39 0.34

CHCl3/MeOH 9:1 0.10 0.07
CHCl3/Pr}1}OH 9:5 0.21 0.12
CH2Cl2/Me2CO/MeOH 2:1:1 0.69 0.62
CH2Cl2/Me2CO/EtOH 16 : 4 : 5 0.32 0.10
CH2Cl2/MeOH/H2O 79 : 15 : 1 0.32 0.19
CH2Cl2/MeOH/ 77 : 20 : 2 : 1 0.47 0.40
25% NH3IH2O/H2O
EtOAc/EtOH 4:1 0.49 0.46
E, ecdysone; 20E, 20-hydroxyecdysone. CHCl3, chloroform;

EtOH, ethanol; MeOH, methanol; PrI1IOH, n-propanol; CH2Cl2,
dichloromethane; Me2CO, acetone; EtOAc, ethyl acetate.
For consistent results, plates should be heated at

1203C for 1 hr, then deactivated to constant activity Figure 2 Normal-phase TLC on a silica gel high-performance
over saturated saline. An example of the type of TLC plate using chloroformIethanol (4 : 1) to separate 20-hydro-
xyecdysone (20E) metabolites formed in an insect. 20E3A2P,
separation that can be achieved in metabolic studies 20-hydroxyecdysone 3-acetate 2-phosphate; 20Eoic, 20-hydro-
in insects is shown in Figure 2. xyecdysonoic acid; 20,26E, 20,26-dihydroxyecdysone; 20E2A,
20-hydroxyecdysone 2-acetate; 20E3A, 20-hydroxyecdysone
Reversed-phase systems An alternative to normal- 3-acetate. After Wilson ID and Lafont R (1986). Thin-layer
phase TLC (NP-TLC) is reversed-phase TLC (RP- chromatography and high-performance thin-layer chromatogra-
phy of [3H] metabolites of 20-hydroxyecdysone. Insect Biochem-
TLC) on silica bound to alkyl chains (2 to 18 car- istry 16: 33I40, reprinted with permission.
bons). Methanol}water, ethanol}water, isopro-
panol}water, acetonitrile}water and acetone}water
systems have been used as mobile phases, with meth-
anol}water solvents providing the most general solvent over the range 0 to 100% increases the RF
vent system. The order of migration in RP-TLC is values of ecdysone and 20-hydroxyecdysone, and
roughly opposite of that seen using NP-TLC (Table 5). this shows a quite linear relationship between the
Results vary signiRcantly with plate manufacturers. percentage of methanol in the solvent and RF
Changing the proportion of methanol in the sol- (Figure 3).
Table 5 R F for representative ecdysteroids in TLC
Compound System 1 System 2 System 3
Calonysterone 0.42 0.20 0.37

Cyasterone 0.33 0.40 0.51
2-Deoxy-20-hydroxyecdysone 0.31 0.21 0.29
2-Deoxyecdysone 0.38 0.15 0.17
Ecdysone 0.21 0.29 0.28
20-Hydroxyecdysone 0.15 0.44 0.38
20-Hydroxyecdysone 2-cinnamate 0.53 0.04 0.03
Inokosterone 0.17 0.44 0.37
Kaladasterone 0.49 0.17 0.30
Makisterone A 0.20 0.31 0.40
Muristerone A 0.27 0.32 0.31
Polypodine B 0.22 0.42 0.44
Ponasterone A 0.42 0.16 0.18
Ponasterone C 0.38 0.29 0.37
Ponasterone C 2-cinnamate 0.65 0 0
Poststerone 0.32 0.37 0.38
System 1, silicagel plates, solvent CHCl3IMeOH (4 : 1); System 2, Merck C18-bonded plates, solvent MeOHIH2O (1 : 1); System 3,
Whatman C18-bonded plates, solvent MeOHIH2O (1 : 1).
acid or ammonia give Suorescent reactions, with the

former being slightly more speciRc.
Evolved TLC Techniques
Automatic multiple development (AMD) In AMD,
the plate is repeatedly developed with the same
solvent, which migrates more and more with each
development. This method allows a reconcentration
of ecdysteroids at each run, in particular by sup-
pressing tailing, and this Rnally results in sharper
bands and improved resolution.
Over-pressure layer chromatography (OPLC) In

OPLC, the solvent is forced through the layer by an
HPLC pump and the plate can be developed within
minutes. The use of high-performance TLC plates
Figure 3 Influence of the solvent composition (MeOH-water) rather than conventional TLC plates is recommended
on the RF of ecdysone and 20-hydroxyecdysone analysed by for optimal results. Developing the plates under these
RP-TLC on C18-bonded plates. 䊐, 20-hydroxyecdysone; ,
ecdysone.
conditions will minimize diffusion, while allowing
mass transfer to proceed.
Neither AMD nor OPLC have been used to any
Detection of Ecdysteroids after TLC signiRcant extent for the separation of ecdysteroids,
although where they have been employed for these
Detection of ecdysteroids on the TLC plate can be compounds good results have been obtained.
accomplished using a set of techniques of varying
speciRcities (Table 6). Non-speciRc techniques in-
clude iodine vapour, heating in the presence of am- Low-Pressure Column
monium carbonate (which produces Suorescent Chromatography
spots), or Suorescence quenching if a Suorescing
Preparative Columns
agent is incorporated into the silica. More speciRc
reagents such as the vanillin}sulfuric acid spray can These systems are used for the isolation and puriRca-
be used to give spots of characteristic colour. Sulfuric tion of ecdysteroids from crude extracts. Columns are
Table 6 Various methods for the visualization of ecdysteroids after TLC
Method Operating mode
UV absorbance Direct visualization under UV light: poorly sensitive method

Use of a scanner and obtention of UV spectra
1
Fluorescence quenching Use of silica plates containing a luminescent agent (ZnSe)
Non-specific colour reactions I2 vapours

Phosphotungstic acid gives blue colour
Anisaldehyde
Fluorescence induction H2SO4

(NH4)2CO3
1
Vanillin spray reagent Spray with vanillin/95% EtOH/H2SO4 (5 : 70 : 25, w/v/v), then heat at 100I1203C for 10 min
Reactions for 3-oxoecdysteroids FolinICiocalteu gives blue colour

2,4-Diphenylhydrazine gives yellow colour (#K3Fe(CN)6 gives orange colour)
Triphenyltetrazolium chloride gives red colour
Radioactivity Scanner or autoradiography (metabolic studies)
Mass spectrometry Direct introduction of the plate, or FAB-MS on scraped silica

1
Classical methods.
Table 7 Some low pressure chromatographic systems for medium- or large-scale

purification of ecdysteroids
Stationary phase Solvent system
Normal phases
Silica (silica gel, silicic acid or celite) CHCl3/MeOH (100 : 3; 95 : 5; 80 : 20; or SG1)
CHCl3/EtOH (19 : 1)
CH2Cl2/EtOH (SG)
EtOAc/MeOH (SG)
Alumina CHCl3/MeOH (2 : 1; or SG)
CHCl3/EtOH (SG)
CH2Cl2/EtOH (9 : 1; or SG)
EtOAc/MeOH (1 : 1)
EtOAc/EtOH (2 : 1; 1 : 1; or SG)
Me2CO/CH2Cl2/H2O (62.5 : 15 : 10)
Sephadex LH20 CHCl3/EtOH (88 : 12)
CH2Cl2/MeOH (SG)
CH2Cl2/Me2CO
Reversed-phases
Amberlite XAD-2 H2O/MeOH (SG)
H2O, then EtOH
Amberlite XAD-16 H2O, then EtOH
Sephadex LH20 EtOH/H2O (7 : 3)
MeOH
Polyamide H2O
Ion-exchange
DEAE-Sephadex Step-gradient of NaCl in H2O
1
SG, step-gradient.
CHCl3, chloroform; MeOH, methanol; EtOH, ethanol; CH2Cl2, dichloromethane; EtOAc,
ethyl acetate; Me2CO, acetone.
Rlled with either normal (polar) phases (silica or volatile buffer is used. Normal-phase SPE cartridges
alumina) eluted with organic solvents, or non-polar have also been used for the fractionation of biological
phases (Amberlite XAD-2, polyamide or Sephadex extracts.
LH20) eluted with aqueous mixtures (Table 7). Ion-
exchange phases (e.g. DEAE-Sephadex) eluted with
buffers can be used for polar anionic ecdysteroids
(Figure 4). The size of the column has to be adapted
to that of the sample, with a sorbent-to-sample
ratio higher than 50 (w : w), and these methods can
be used with very large samples. They represent
a rather cheap and reasonably efRcient procedure
for getting fractions from which ecdysteroids can
be crystallized (if present in large amounts) or fur-
ther puriRed by HPLC (see below). Step-gradient
elution with solvents of increasing strength allows
the separation of ecdysteroids over a wide range of
polarity.
Disposable Cartridges
Small solid phase extraction (SPE) cartridges contain- Figure 4 Separation of acidic ecdysteroids from Manduca
ing 0.2}1 g of non-polar HPLC phase allow the clean- sexta using a DEAE-Sephadex column (6.5 cm long, 2 cm i.d.).
up of small samples with a good recovery (Figure 5). Elution was performed with a step-gradient of NaCl. Peak 2
contains ecdysonoic acids, peak 3 contains phosphate conju-
They can also be used to adsorb ecdysteroids from
gates. Redrawn with permission from Lozano R, Thompson MJ,
aqueous media, thus allowing an easy and quan- Svoboda JA and Lusby WR (1988) Isolation of acidic and con-
titative recovery of ecdysteroids from organ/cell jugated ecdysteroid fractions from Manduca sexta pupae. Insect
culture media or from HPLC fractions when a non- Biochemistry 18: 163I168.
High Performance Liquid Diode-array detectors provide information about

Chromatography the absorbance spectrum of all eluted peaks. It
can thus be directly checked whether a compound
High performance liquid chromatography (HPLC) co-migrating with a reference ecdysteroid has a suit-
offers a wide range of techniques for analytical and able UV spectrum. Good spectra can be obtained
preparative purposes. Co-migration of a compound with very small amounts (less than 100 ng) of
with a reference ecdysteroid in one (or several) sol- ecdysteroids and such data provide an additional
vent system(s) represents the usual way for identiRca- criterion to assess the identity of UV-absorbing
tion of the compound. Some common HPLC systems peaks.
are listed in Table 8.
Fluorescence detectors require the preparation of
Chromatographic Procedures Suorescent derivatives of ecdysteroids, which may
increase the sensitivity of detection by two orders of
Ecdysteroid detection
magnitude when compared with UV. Phenanthrene
UV detectors are well-suited to the detection of boronic acid, a reagent speciRc for -diols (here the
ecdysteroids, as most ecdysteroids possess a con- 20,22-diol), and 1-anthroyl nitrile, which reacts with
jugated 7-en-6-one moiety which provides a strongly alcohols (here the 2-OH of ecdysteroids), have been
absorbing chromophore (max 242 nm, log ca. 4). used. These reactions, however, are not speciRc
This allows the easy detection of less than 10 pmol enough for ecdysteroids and they have not been
amounts. widely adopted.
Figure 5 Utilization of reversed-phase cartridges for ecdysteroid purification. The cartridge must be rinsed with 5 mL MeOH then
5 mL water prior to use. Redrawn from Lafont R, Morgan ED and Wilson ID (1994) Chromatographic procedures for phytoecdysteroids.
Journal of Chromatography 658: 31I53, with permission from Elsevier Science.
Table 8 Chromatographic systems commonly used for the HPLC analysis of ecdysteroids
Mode of chromatography Polar Medium Apolar
Ionic Nonionic
Normal-phase (silica, diol, APS, TMS)

Chloroform/95% ethanol N # #
Chloroform/methanol N # #
Cyclohexane/isopropanol/water N N # #
Dichloroethane/isopropanol/water N # #
Dichloromethane/tetrahydrofuran/methanol N # #
Dichloromethane/ethanol/water N # #
Dichloromethane/isopropanol/methanol N # #
Dichloromethane/isopropanol/water N # # #
Dichloromethane/methanol N # #
Dichloromethane/methanol/water/acetic acid N # #
Hexane/ethanol/methanol/acetonitrile N N # #
Isooctane/isopropanol/water N N # #
Reverse-phase (C18, C8, phenyl, etc.)
Acetonitrile/isopropanol #
Acetonitrile/isopropanol/water # # #
Acetonitrile/water # # #
Acetonitrile/Tris-HClO4, Tris-HCl, Na citrate, TFA 0.1% # # #
Isopropanol/water # # #
Methanol/water, Na acetate, Na phosphate # # #
Methanol #
Ion-pair
Acetonitrile/cetrimide-phosphate #
Methanol/tetrabutylammonium #
Ion-exchange
Ammonium acetate # N N N
N, does not apply. APS, aminopropyl silane; TFA, trifluoroacetic acid; TMS, trimethylsilane.
Radioactivity monitoring provides a direct com- Off-line procedures may be used to improve the
parison with UV absorbance, and this easily allows sensitivity and/or selectivity of ecdysteroid detection
one to make correspondence between the radioactive (or to identify ecdysteroids after preparative HPLC
peaks and unlabelled reference compounds added in puriRcation). These analytical methods include chief-
the sample before injection. They are currently used ly immunoassays (RIA, EIA) and also several recently
for metabolic studies. designed in vitro bioassays.
Normal-phase systems Normal-phase HPLC systems

Mass spectrometry (MS) gives important structural
generally use silica columns (sometimes polar-bonded
information. The interfacing problems between
columns) and, for example, dichloromethane}
HPLC and MS have been overcome in recent years
isopropanol}water mixtures. Non-polar ecdysteroids
and this technique will undoubtedly develop in
(such as esters or precursors) can be separated with
the future. An example of HPLC-MS applied to
a 125 : 15 : 1 (v/v/v) mixture, medium-polarity com-
an ecdysteroid-containing plant extract is given in
pounds (such as ecdysone and 20-hydroxyecdysone)
Figure 6.
with a 125 : 30 : 2 mixture, and more polar (but non-
ionic) ecdysteroids (such as 26-hydroxyecdysteroids
Nuclear magnetic resonance spectrometry can also and glucosides) with 125 : 40 : 3 or 100 : 40 : 3 mix-
be used on-line to identify ecdysteroids. This re- tures.
quires rather expensive deuterated HPLC solvents Dichloromethane-based solvents strongly absorb
and the method is only suitable for plant extracts UV and do not allow UV spectra to be obtained with
where ecdysteroid concentrations are high enough. diode-array detectors, nor are they well suited to
An example of the use of this emerging technology in in-line radioactivity monitoring (due to their quench-
a plant extract is shown in Figure 7. ing properties). These problems do not exist with
mixtures provides the most widely used system.

Acetonitrile}water (acetonitrile}buffer) mixtures are
more efRcient, especially when polar conjugates
and/or ecdysonoic acids are present. Adequate sys-
tems have been designed for polar and apolar metab-
olites, which may both be present within the same
sample. Surprisingly, apolar fatty acyl esters of ecdys-
teroids are not eluted with pure acetonitrile, whereas
they are with methanol, despite the fact it is a more
polar solvent (the same is true for cholesterol).
In the case of polar (ionizable) metabolites, it may
be of interest to use different pHs, which will result in
modiRed retention times, while uncharged ecdys-
teroids will retain the same elution time. Moreover,
this gives an easy access to the pKa value of ionizable
groups, which is of interest for the characterization of
conjugates.
Ion-exchange chromatography Anion-exchange

columns are used for the puriRcation of polar conju-
gates. They represent an efRcient method, com-
plementary to reversed-phase HPLC. However, using
two different pHs (e.g. pH 7.5, then 2.5) with HPLC,
provides an equivalent opportunity for obtaining
pure ionizable conjugates, e.g. phosphate esters.
Various Aims of HPLC
Figure 6 Reversed-phase HPLC-MS total ion current trace
(upper) of an extract of Silene otites . Peak 1, integristerone A; Analytical and preparative HPLC Columns of dif-
2, 20-hydroxyecdysone; 3, 2-deoxy-20-hydroxyecdysone; 4, 2- ferent sizes with the same packing are available,
deoxyecdysone. The mass spectra (lower) are of 20-hydroxyec- therefore it is very easy to scale up any chromato-
dysone (left) and integristerone A (right). Structures in insets to
mass spectra. After Wilson ID, Lafont R, Shockcor JP et al. (1999)
graphic separation. Maximum sample load is related
High-performance liquid chromatography coupled to nuclear to the amount of stationary phase in the column. In
magnetic resonance spectroscopy and mass spectrometry ap-
plied to plant products: Identification of ecdysteroids from Silene
otites. Chromatographia 49: 374}378, reprinted with permission.
cyclohexane-based mixtures, however, the poor solu-

bility of ecdysteroids in these mixtures causes some
problems for the analysis of polar ecdysteroids and
also for preparative purposes.
Polar-bonded columns (e.g. -diol, -polyol or
-aminopropylsilane, APS) can also be used. Diol-
bonded columns used with extracts of the phasmid
Carausius morosus allowed the separation of a wide
array of polar and non-polar metabolites, whereas
the APS phase has provided efRcient separations of
mixtures containing 3-OH, 3-OH and 3-oxo ec-
dysteroids. One major interest in such polar-bonded Figure 7 Stopped-flow HPLC-NMR spectrum of 20-hydroxyec-
columns is that solvent gradients can be used, where- dysone following reversed-phase HPLC on a C18-bonded column
as lengthy re-equilibration times would be the case with D2O acetonitrile as mobile phase. (Reproduced with per-
mission from Wilson ID, Morgan ED, Lafont R and Wright
with silica columns.
B (1998) High-performance liquid chromatography coupled to
nuclear magnetic resonance spectroscopy. Applications to the
Reversed-phase systems Reversed-phase HPLC on ecdysteroids of Silene otites. Journal of Chromatography A 799:
C18-bonded columns eluted with methanol}water 333}336).
favourable cases, samples up to the milligram range present and adequate sample clean-up. The most re-
can be run on analytical columns. Larger amounts, of liable quantiRcation is obtained if an internal stan-
course, require wide-bore columns. It may happen dard is added before sample puriRcation. Many
that some compounds that are baseline resolved when phytoecdysteroids can be used as internal standards.
loading is low, are not well separated with a larger The choice must be made after a preliminary run of
load. This is, for instance, observed with the inokos- a representative sample in order to select a compound
terone/20-hydroxyecdysone pair run on a semi-prep- that does not co-migrate with major impurities and of
arative column (Zorbax威-SIL, 9.6 mm i.d.), even course differs from the ecdysteroids present in the
when less than 0.5 mg is injected. biological extract.
Selectivity in HPLC
Quantitative analyses HPLC has been used for the
direct quantiRcation of individual ecdysteroids in bio- Both isocratic reversed-phase and normal-phase sys-
logical samples. For animal extracts, this requires tems can separate rather complex ecdysteroid mix-
a high sensitivity because of the low concentrations tures, and the use of solvent gradients increases the
Table 9 HPLC data on a set of ecdysteroids using several chromatographic systems
Compound DIW CIW AW MW IW
2-Deoxy-20-hydroxyecdysone 13.9 11.7 12.5 11.8 14.8

2-Deoxyecdysone 9.5 8.9 31.9 22.6 44.1
3-Dehydro-20-hydroxyecdysone 9.2 16.2 6.5 5.8
3-Epi-20-hydroxyecdysone 25.7 19.6 5.6 5.4
5-2-Deoxy-20-hydroxyecdysone 13.6 13.6 11.6 10.7 13.0
5-2-Deoxyecdysone 9.2 10.0 30.2 21.7 38.3
5-20-hydroxyecdysone 24.9 24.1 4.9 4.8 5.0
5-Ecdysone 15.0 16.2 9.3 7.3 9.9
20-Hydroxyecdysone 28.9 21.3 5.2 5.4 6.0
20-Hydroxyecdysone 2-acetate 10.7 13.8 14.6 11.3
20,26-Dihydroxyecdysone 89.9 47.1 3.7 3.7 4.3
22-Deoxy-20-hydroxyecdysone 21.7 15.3 13.5 11.7 18.5
22-Epi-20-hydroxyecdysone 50.0 25.8 4.6 4.7
22-Oxo-20-hydroxyecdysone 16.8 15.1 12.4 9.8
24-Epi-makisterone A 16.8 15.5 7.8 6.9
25-Deoxyecdysone 6.1 6.4 102.9 32.7
Abutasterone 35.5 27.1 4.4 4.8 5.0
Ajugalactone 7.2 20.2 14.5 7.5
Cyasterone 9.9 14.5 8.1 5.8
Ecdysone 18.8 15.1 9.7 8.4 13.0
Gerardiasterone 40.6 27.9 5.5 6.0
Integristerone A 38.7 28.8 4.2 4.4 5.0
Makisterone A 20.5 17.7 7.0 6.9
Makisterone C 11.3 11.3 16.8 12.9
Polypodine B 18.8 21.5 5.4 5.4 5.5
Ponasterone A 6.6 7.9 32.8 17.5 48.8
Poststerone 9.0 13.3 9.3 6.4
Rubrosterone 8.7 13.3 5.6 4.3
Sidisterone 9.6 26.0 12.2 6.3
Stachysterone C 7.2 8.1 22.7 13.2
Turkesterone 89.9 36.4 4.4 4.1 3.7
䉭24,25-25-Deoxyecdysone 6.4 6.8 24.3
䉭24,28-Makisterone A 13.6 13.6 8.0 7.5
䉭25,26-25-Deoxyecdysone 6.4 6.8 21.1
䉭9,11-20-Hydroxyecdysone 7.2 7.7 14.1
䉭9,11-Ecdysone 22.5 16.8 7.2
Analytical columns were either a Zorbax-Sil column (DuPont), 25 cm;4.6 mm i.d., eluted with DIW (dichloromethane}iso-
propanol}water, 125 : 30 : 2) or CIW (cyclohexane}isopropanol}water, 100 : 40 : 3) or a Spherisorb 5ODS2 (Biochrom), 25 cm;
4.6 mm i.d., eluted with water #1 (final concentration) trifluoroacetic acid and either 23% acetonitrile (AW), 50% methanol (MW), or
18% isopropanol (IW). Flow rate was 1 mL min\1 in every case.
power of these systems. In order to draw up general Selectivity in NP-HPLC

rules, it is necessary to analyse the HPLC behaviour
of numerous ecdysteroids differing by single or mul- Columns NP-HPLC columns are packed with either
tiple modiRcations, and to use a set of different HPLC silica or polar-bonded silica. The use of diol-bonded
systems. A set of such data are given in Table 9, and material instead of silica does not introduce large
the corresponding conclusions are given in Table 10. changes; retention times vary, but the elution order
Table 10 Effects of individual changes (with reference to the 20E molecule) on the chromatographic behaviour of ecdysteroids as
analysed using various chromatographic systems
Change Effect on capacity factor (k)
Normal-phase Reversed-phase
DIW CIW MW AW
Change of the number

of -OH groups
#1, 11, 24, 26 ##(1(24(11"26) #(24(1(11(26) !(26(11(1(24)! !(26(1(11"24)
#5 ! " (!) (#)
!2, 20, 22, 25 !(25(2(20(22) !(25(2(20"22) #(20(2"22(25) #(20(2(22(
(25)
Change of
stereochemistry
at C-5 (5P5)
If 2-OH present ! # (#) (!)
If 2-OH absent " # " (!)
Change of ## # ! !
stereochemistry
at C-22
Change at C-3
3-OHP3-oxo !! ! (#) #
3-OHP3-OH ! ! " (#)
Presence of
substituents at C-24
24-Me ! ! # #
24-Et !! !! ## ##
Conjugation of 20E
Monoacetates !!(25(2"3(22) !(25(2(3(22) #(22"3(25(2) ##(3(22(2(25)
Monoglucosides ##(22(25(2"3) Not tested #(22(25(2"3) !(22(25(2"3)
Presence of
double bonds
9, 11 (#) (#) ! !
25, 26 # ! ! !
24, 25 # # ! !
Presence of a !! (!)
lactone on the
side-chain
(cyasterone vs. 20E)
Side-chain
cleavage products
Poststerone (C21) !! !! # #
Rubrosterone (C19) !! !! ! !
Changes of the k value are described as follows:!!, strong reduction;!, reduction; (!), weak reduction;", almost no change;
(#), weak increase; #, increase; ##, large increase. Numbers in parentheses refer to the positions concerned, and the molecules
are classified according to increasing k values. DIW, dichloromethane}isopropanol}water; CIW, cyclohexane}isopropanol}water;
MW, methanol}water; AW, acetonitrile}water.
usually remains the same. Aminopropyl (APS) col- NP-HPLC is rather inefRcient for the separation of
umns may interact speciRcally with 3-oxoecdys- compounds with or without extra double bonds,
teroids and therefore introduce a different selectivity. whereas RP-HPLC allows their easy resolution.
C2-bonded phases behave like weakly active silicas
and give very symmetrical peaks; they may be of Selectivity in RP-HPLC
interest for very polar non-ionic ecdysteroids.
The retention times obtained with silica columns Columns Many different types of column are avail-
may decrease upon prolonged use, because water- able, which differ by the type (C6, C8, C18, C22,
containing solvents slowly deactivate the column. phenyl, CN, etc.) and extent (percentage of the car-
A reactivation cycle with anhydrous solvents: alco- bon load) of bonding, and also by the porosity
hol, dichloromethane, hexane, dichloromethane (6}30 nm) of the silica used. All these parameters
(50}100 mL each) allows an almost complete recov- inSuence the selectivity of the column, thus, ensuring
ery of initial retention times. reproduction of a separation described in the litera-
ture requires the same type of column. C18 (or ODS)
Solvents UV monitoring of the HPLC efSuent pre- bonded silicas are the most widely used column
cludes the use of solvents such as ethyl acetate, packings.
benzene or acetone (such limitations were not en-
countered with TLC). The solvent is usually based on Solvents The most common RP-HPLC solvents con-
a chlorinated hydrocarbon (dichloro(m)ethane, tain water and either methanol or acetonitrile, al-
chloroform) modiRed with an alcohol (methanol, though other organic modiRers (ethanol, isopro-
ethanol or isopropanol). Water added just below panol, butanol, etc.) can be used. Using a buffer
saturation reduces peak tailing. instead of water, or adding 0.1% (v/v) triSuoroacetic
Dichloromethane- and cyclohexane-based mix- acid often results in much improved separations, es-
tures display a different selectivity, as shown in pecially when ionizable ecdysteroids are present.
Figure 8. Selectivity changes are particularly impres- The relative retention of ecdysteroids differing by
sive when considering the pair 5/5, or compounds a single -OH group varies with the organic solvent of
bearing a lactone ring on the side-chain. As cyc- the mobile phase. Extra -OH groups generally in-
lohexane-based mixtures are highly viscous, the crease the polarity; their effect depends both on their
working pressures may exceed 100 bar with analyti- position in the molecule (Figure 9) and on the solvent
cal columns run at a Sow-rate of 1 mL min\1. system used.
Increasing temperatures to 503C overcomes this Isopropanol}water mixtures provide the best sep-
drawback and results in about a 40% decrease of arations for 5/5 pairs. Methanol is particularly
working pressure without affecting the separation. efRcient towards extra double bonds and gives a base-
Figure 8 An example of selectivity in NP-HPLC. E, ecdysone; 20E, 20-hydroxyecdysone; 20,26E, 20,26-dihydroxyecdysone; PolB,
polypodine B; Turk, turkesterone. CIW, cyclohexaneIisopropanolIwater; DIW, dichloromethaneIisopropanolIwater.
Figure 9 Separation of reference ecdysteroids by RP-HPLC

(column Spherisorb 5ODS2, 250 mm;4.6 mm, solvent 35%
methanol in water, isocratic, flow-rate 1 mL min\1). The numbers
indicate the position of the additional }OH group by reference to
20-hydroxyecdysone (20E).
line separation of 24,25-ene and 25,26-ene pairs.

Higher temperatures increase efRciency, decrease
Figure 10 SFC of ecdysteroids on 5 m-cyanopropyl-bonded
pressure and result in a reduction of k with only silica gel with carbon dioxide}methanol (9 : 1) as mobile phase
limited effects on selectivity. at 3 mL min\1, 603C and 290 bar (sample from Silene otites).
(Reproduced from Raynor MW, Kithinji JP, Bartle K et al. (1989)
Packed column supercritical-fluid chromatography and linked
Supercritical Fluid Chromatography supercritical-fluid chromatographyImass spectrometry for the
(SFC) analysis of phytoecdysteroids from Silene nutans and Silene
otites. Journal of Chromatography 467: 292}298, with permission
Supercritical carbon dioxide is a non-polar Suid and from Elsevier Science.)
therefore SFC is more or less equivalent to normal-
phase HPLC. The eluting power is increased by
adding methanol to the Suid. Various types of packed tion (the universal detection used with GC) and
columns can be used (Table 11), including silica or mass spectrometry (interfacing causes far fewer prob-
bonded silica; bonded silica is less retentive as is lems than with HPLC). No derivatization is required,
observed with NP-HPLC. Alternatively, fused silica and this represents an important advantage over GC.
capillary columns can be used. SFC appears, therefore, to be an interesting compro-
A major advantage of SFC over HPLC is shorter mise, but its use has been up to now limited to
retention times. The compounds elute as very sharp plant extracts containing particularly high concen-
peaks and the sensitivity of detection is increased trations of ecdysteroids. The easy removal of carbon
accordingly. A second advantage is the high transpar- dioxide after fraction collection represents an
ency of carbon dioxide in the infrared (IR), which advantage for preparative uses. An example of
allows the use of FT-IR detectors. Furthermore, the separation of ecdysteroids by SFC is given in
SFC is compatible with both Same ionization detec- Figure 10.
Table 11 Some SFC systems used for ecdysteroid analysis
Column Solvent system
Normal phase (silica)

Hypersil 5 m (10 cm;4.6 mm i.d.) CO2IMeOH (4 : 1) 300 bar, 803C, 4 mL min\1
Bonded phases
Spherisorb-CN 5 m CO2 or CO2IMeOH (9 : 1)
Spherisorb-ODS2 5 m CO2 or CO2IMeOH (9 : 1)
See Lafont et al . (1994b) for more details.

Gas Chromatography See Colour Plate 80.
Gas chromatography (GC) was Rrst developed for See also: II / Chromatography: Gas: Derivatization; De-
ecdysteroids during the early 1970s. Despite the tectors: Selective. Chromatography: Liquid: Counter-
advantages of GC (especially in combination with the current Liquid Chromatography; Derivatization; Detectors:
electron capture detector), concerning sensitivity and Mass Spectrometry; Detectors: Ultraviolet and Visible
speciRcity compared with other techniques, its use for Detection; Fluorescence Detectors in Liquid Chro-
matography; Nuclear Magnetic Resonance Detectors.
ecdysteroids has been limited. As a majority of bio-
Chromatography: Thin-Layer (Planar): Densitometry
synthetic intermediates of the ecdysteroids are in- and Image Analysis; Layers; Mass Spectrometry; Modes
volatile, it is necessary to convert them into volatile of Development: Conventional; Modes of Development:
trimethylsilyl ether derivatives. Derivatization is usu- Forced Flow Over Pressured Layer Chromatography and
ally performed with either trimethylsilylimidazole Centrifugal; Spray Reagents. Extraction: Analytical Ex-
(TMSI) or N,O-bis-trimethylsilyltriSuoroacetamide, tractions; Multistage Countercurrent Distribution. III / Im-
although other more specialized derivatives have mobilised Boronic Acids: Extraction. Natural Prod-
been used. Complete derivatization is not always ob- ucts: Liquid Chromatography I Nuclear Magnetic Reson-
tained and a single ecdysteroid can give rise to several ance. Steroids: Liquid Chromatography and Thin-Layer
chromatographic peaks. (Planar) Chromatography; Gas Chromatography; Super-
Not all ecdysteroids are suitable for GC even after critical Fluid Chromatography.
silylation. Silylation of compounds such as cyasterone
(which contains a lactone in the side-chain) can result
in a variety of derivatives characterized either by long
Further Reading
retention times or poor peak shapes. Some ecdys- Horn DHS and Bergamasco R (1985) Chemistry of ecdys-
teroid conjugates such as coumarate esters, appear to teroids. In: Kerkut GA and Gilbert LI (eds) Comprehens-
break down to the free ecdysteroid during the silyla- ive Insect Physiology, Biochemistry and Pharmacology,
tion procedure. vol. 7, pp. 185}248. Oxford: Pergamon Press.
Lafont R (1988) HPLC analysis of ecdysteroids in plants
Chromatographic Systems and animals. In: Kalasz H and Ettre S (eds) Chromato-
graphy ’87, pp. 1}15. Budapest: Akademiai Kiado.
Chromatography was Rrst performed with packed Lafont R and Beydon P (1990) Methods for ecdysteroid
columns (1}3 m long, 2 mm i.d.) containing non- analysis. In: Gupta AP (ed.) Morphogenetic Hormones
polar stationary phases such as OV-1 and OV-101 of Arthropods. Discoveries, Syntheses, Metabolism,
coated on Gas Chrom Q and Gas Chrom P. Fused- Evolution, Mode of Action and Techniques, vol.1,
silica columns (10}25 m long, 0.22 mm i.d.) now pp. 485}512. New Brunswick: Rutgers University Press.
provide much better separations. A typical operation Lafont R and Wilson ID (1996) The Ecdysone Handbook,
temperature for the columns is 2803C. 2nd edn. Nottingham: The Chromatographic Society
Press.
Flame ionization detectors (FID) allow detection
Lafont R, Kaouadji N, Morgan ED and Wilson ID (1994)
limits in the nanogram range. Electron-capture detec- Selectivity in the HPLC analysis of ecdysteroids. Journal
tion (ECD) provides the best available sensitivity of Chromatography 658: 55d67.
among chromatographic methods (5 pg), and GC Lafont R, Morgan ED and Wilson ID (1994) Chromato-
coupled with MS gives structural information with as graphic procedures for phytoecdysteroids. Journal of
few as 0.1 ng of ecdysteroids (particularly important Chromatography 658: 31d53.
with animal extracts). However, the need for derivat- McCaffery AR and Wilson ID (eds) (1990) Chromatogra-
ization is such a drawback that this powerful method phy and Isolation of Insect Hormones and Pheromones.
is seldom used for ecdysteroids nowadays. London: Plenum Press.
Morgan ED and Marco MP (1990) Advances in techniques
for ecdysteroid analysis. Invertebrate Reproduction and
Conclusion Development 18: 55d66.
Koolman J (ed.) (1989) Ecdysone, From Chemistry to
HPLC is the most widely used method for the separ-
Mode of Action. Stuttgart: Georg Thieme Verlag.
ation of ecdysteroids. At the present time, hyphenated Thompson MJ, Svoboda JA and Feldlaufer MF (1989)
techniques (HPLC-MS and HPLC-NMR) are devel- Analysis of free and conjugated ecdysteroids and polar
oping and will provide particularly powerful tools for metabolites of insects. In: Nes WD and Parish EJ (eds)
the analysis of ecdysteroid-rich (plant) samples in the Analysis of Sterols and Other Biologically SigniTcant
future. Steroids, pp. 81}105. San Diego: Academic Press.
2644 III / ECOLOGICALLY SAFE ION EXCHANGE TECHNOLOGIES
ECOLOGICALLY SAFE ION EXCHANGE

TECHNOLOGIES
D. Muraviev, Autonomous University of Barcelona, exchange isothermal supersaturation or IXISS)
Barcelona, Spain allows, in certain instances, for additional elimina-
tion of operations 4 and 5. Several examples of the
tailored application of dual-temperature IE and IXISS
effects to design ecologically safe IE technologies are
The industrial application of ion exchange (IE) is given below.
growing. In many instances IE can successfully re-
place existing large scale industrial processes which IXISS-based IE Processes1
do not satisfy modern ecological standards. To be
competitive, IE technology should be highly efRcient The tailored application of IXISS to design highly
and ecologically safe. The general scheme of standard efRcient IE technologies is based on the use of the
IE processes comprises several auxiliary operations IXISS-active stripping agents, which must meet the
(besides IE treatment), listed in Table 1. Several ap- following requirements:
proaches have been applied to eliminate some of these
auxiliary operations and to improve the efRciency 1. The IXISS-active eluent must on the one hand bear
and ecological safety of the process through signiR- the desired counterion to combine the desorption
cant savings of chemicals and energy and reduction of of the product with the regeneration of the ion
waste. exchanger. On the other hand, it must also contain
One of these approaches is based on the ap- an appropriate co-ion to provide the formation of a
plication of dual-temperature IE processes which stable SS of a low solubility compound (the prod-
exploit the different afRnities of the resin towards uct) and to shift the IE equilibrium in the system to
ions to be separated at different temperatures. the desired direction. For example, the IE reaction
The main advantage of dual-temperature IE comes of displacement of a divalent metal ion, M2# 1 (e.g.
from avoiding the use of auxiliary reagents which Mg2# or Ca2#), by a monovalent one, M# 2 (e.g.
are conventionally required to displace (strip) the Na#), from a carboxylic cation exchanger can be
ions to be separated from the resin phase and to written as follows:
regenerate the ion exchanger. This last stage is
# #
1 #2M2 02R-COO\M2 #M1
(R-COO )2M2# 2#
known to be the main source of waste in IE techno-
\
logy (Table 1). In dual-temperature IE, both stages
[1]
can be combined in one through the use of a thermo-
stripping procedure. This allows the exclusion of
auxiliary operations 4 and 5. Hence, essentially The equilibrium in reaction 1 is characterized by
reagentless and wasteless separation processes can be the equilibrium coefRcient, KM
M2:
1
designed.
Another route to avoid auxiliary operations (e.g. QM2(CM1)1/2
M2"
KM [2]
1
numbers 2 and 3; Table 1) is based on a combination (QM1)1/2CM2

of IE conversion and the product concentration pro-
cesses into one stage. Frontal IE chromatography and where C and Q are the concentrations of ions in
reverse frontal separation can be applied for this the solution and resin phases, respectively. For
purpose. In certain instances, both of these IE separ- M2 is usually 1. If reaction 1
chloride media, KM 1
ation techniques allow the concentration of the target proceeds in a carbonate medium, for example,
substance to a level exceeding its solubility at a given when the resin in the M1 form is treated with
temperature. Moreover, this supersaturated solution
(SS) remains stable for a long period, while after
leaving the column it crystallizes spontaneously. This
1
allows the design of a nearly ideal process where Adapted with permission from Muraviev D, Khamizov R,
Tikhonov N et al. (1998). Clean ion-exchange technologies. I.
a crystalline product is obtained directly after the IE Synthesis of chlorine-free potassium fertilizers by an ion-exchange
treatment. The tailored application of this phenom- isothermal supersaturation technique. Industrial Engineering and
enon (discovered by Muraviev and known as ion Chemistry Research 37: 1950}1955.
III / ECOLOGICALLY SAFE ION EXCHANGE TECHNOLOGIES 2645
Table 1 Basic auxiliary operations of standard ion exchange process
No. Operation Consumption of Wastes (approx.

% of total)
Energy Chemicals
1 Preparation of stock solution Low Low Up to 5

2 Concentration of solution after IE treatment (e.g. by evaporation) High Low
3 Recovery of purified product (e.g. by crystallization) Medium Low Up to 5
4 Regeneration of ion exchanger and auxiliary chemicals for reuse Low/medium High Up to 80
5 Neutralization of waste before their disposal Low High/medium Up to 10
M2 carbonate solution (Co, mol dm\3), it is SS in the interstitial space of the column during
coupled with the formation of M1 carbonate, the IE treatment cycle and, on the other hand,
which can be described by the solubility product fast decomposition (crystallization) of this
of M1CO3 (LM1CO3), where: solution after its removal from the column. In the
case of inorganic substances a uniRed interpreta-
LM1CO3"CM21#CCO23\ [3] tion of the IXISS phenomenon must be based on
general principles of the aggregative stability of
If M1CO3 forms a stable SS, where it exists in an dispersion systems, adapted to the particular IE
associated (molecular) form at a concentration system.
CM mol dm\3 exceeding times its solubility, CS, 3. The following main factors may inSuence the stab-
at a given temperature, eqn [3] takes the following ility of dispersions of precrystalline molecular ag-
form: gregates in the interstitial space of a column: Rrst-
ly, an effective charge of the polymolecular ag-
LM1CO3"CM2#(Co!CM) [4]
1
gregate (micelle), which is due to the sorption of
either counter- or co-ions on the particle surface;
By introducing CM"CS, and after substitution of
and secondly, the ionic strength of the medium,
CM21# from [4] into [2], one obtains:
which may strongly inSuence the coagulation
(crystallization) conditions. If, for example, an
QM2(LM1CO3)1/2 excess of co-ions exists in the interstitial space, the
M2"
KM [5]
1
(QM1)1/2Co(Co!CS) charge of the micelles will be the same as that of

the functional groups of the ion exchanger. In this
As follows from eqn [5], at constant Co, CS and case, the sorption of the micelles on the surface of
LM1CO3, KMM2 increases with and may reach sufR-
1
the ion exchanger beads becomes impossible and
ciently high value (1), as PCo/CS. Equation a stabilizing action of the resin bed towards SS can
[5] can be rewritten in a more general form for the be expected. In contrast, in the presence of an
displacement of the divalent metal ion from the excess of counterions, the charge sign of the prec-
resin with an IXISS-active stripping agent bearing rystalline aggregates will be opposite to that of the
a monovalent counterion as follows: functional groups and fast decomposition of SS
must be observed due to the sorption of micelles
1/2
QDisLSss on the surface of the ion exchanger, followed by
Sss "
KDis [6]
(QSss) Co(Co!CS)
1/2
crystallization of the component under super-
saturation.
Here ‘Dis’ and ‘Sss’ superscript and subscript de-
Manufacture of Chlorine-free Potassium
note the displacer and the substance under super-
Fertilizers2
saturation (the product), respectively; Co is the
concentration of the displacer solution, L is the The production of chlorine-free potassium salts (with
solubility product of the target compound, CS is minimum Cl\ admixture) such as K2SO4, and others,
the solubility of the product at a given temper- is of particular interest since it deals with the problem
ature, and is the degree of supersaturation for the
product solution. Relationship [6] is the funda- 2
Adapted with permission from Muraviev D, Khamizov R,
mental equation describing the shift of IE equilib- Tikhonov N et al. (1998). Clean ion-exchange technologies. I.
rium in IXISS systems of different types. Synthesis of chlorine-free potassium fertilizers by an ion-exchange
2. The successful application of the IXISS effect re- isothermal supersaturation technique. Industrial Engineering and
quires, on the one hand, maximum stability of the Chemistry Research 37: 1950}1955.
of effectively cultivating some chlorophobic plants eqns [9] and [10]) appears to be more efRcient when
(e.g. citruses, vegetables and herbs) which are ad- carried out at 282 K to minimize the solubility of
versely affected by high Cl\ concentration. Potassi- K2SO4 in the solution collected.
um sulfate is produced in substantial quantities in The IE equilibrium in both systems is shifted to the
Europe by the Mannheim process from K2CO3 and right when using dilute KCl and Na2SO4 solutions at
H2SO4 or by reaction of H2SO4 with KCl. Both ver- the Rrst stages of both processes. Application of IXISS
sions of the Mannheim process are complicated by effect in the Sowsheets of both processes allows the
problems of utilizing gaseous wastes (CO2 and HCl). product formation stage to be improved due to the
In the USA and some other countries, K2SO4 is manu- shift of IE equilibrium in reactions 8 and 10 to the
factured by exchange reactions between potassium, right. This clearly follows from eqn [5], which can be
sodium and magnesium salts by their dissolution rewritten for, e.g. SO24\}Cl\ exchange (see reaction
and fractional crystallization. This latter process re- [9]), in terms of the equilibrium separation factor
quires utilization of large volumes of liquid waste. (usually written using either equivalent or equiva-
The IE synthesis of K2SO4 from KCl and lent fraction concentration scales):
Na2SO4 ) 10H2O is based either on cation or anion
exchange reactions: qCl KDCS
SO4"
Cl [11]
qSO4 Co(Co!CS)
Cation exchange synthesis
Here KD is the dissociation constant of K2SO4. The
R-SO3Na#KClNR-SO3K#NaCl [7]
same reasoning is applicable to interpret the selectiv-
ity reversal in the cation exchange system.
2R-SO3K#Na2SO4N2R-SO3Na#K2SO4 [8] Another advantage of the IXISS-based synthesis of
K2SO4 deals with the possibility of reusing the super-
Anion exchange synthesis natants obtained after crystallization of K2SO4 as
a displacer in the subsequent stripping cycles. For
R-N(CH3)3Cl#Na2SO4 example, after separation of K2SO4 crystals, the
N(R-N(CH3)3)2SO4#2NaCl [9] supernatant obtained in the Rrst process is fortiRed
with Na2SO4 up to the desired concentration of sul-
(R-N(CH3)3)2SO4#2KCl fate ions (2 mol dm\3) and is then directed to the next
N2R-N(CH3)3Cl#K2SO4 [10] stripping cycle. A diagram of the cation exchange
version of the process for the synthesis of chlorine-
In the Rrst process, a sulfonate cation exchanger is free potassium sulfate is given in Figure 1. The unit
Rrst converted from the Na> into the K> form with comprises two ion exchange columns operating inter-
dilute KCl solution (0.1 mol dm\3), followed by de- mittently in a loading (see eqn [7]) or displacement
sorption (stripping) of the product (K2SO4) with con- (see eqn [8]) mode of operation. The second stage
centrated Na2SO4 solution (2 mol dm\3). The second (displacement) is carried out using an Na2SO4 (or
process starts with the conversion of a strong base a Na2SO4}K2SO4 mixture) solution at 308 K. The
anion exchanger from the Cl\ into the SO24\ form rinsing water produced after each stage is returned to
with dilute Na2SO4 solution (&0.25 mol dm\3). the process and is used to dissolve either KCl (rinsing
K2SO4 is then produced during the stripping of sul- after loading) or Na2SO4 (rinsing after displacement).
fate ions with concentrated KCl solution The NaCl efSuent obtained after the loading stage is
(&3}4 mol dm\3). The stripping of K2SO4 from the directed into the reverse osmosis unit, which pro-
resins in both cases leads to the formation of an SS of duces desalinated water and NaCl concentrate, used
K2SO4 with a degree of supersaturation, , of approx- to manufacture crystalline NaCl. The desalinated
imately 2. Nevertheless, K2SO4 does not precipitate in water obtained is returned to the process. Hence,
the column and remains as a stable SS at least for the process is essentially wasteless and ecologically
a period of several hours. At the same time, this SS clean.
crystallizes spontaneously following its removal from
Decalcination of Mineralized Waters
the column. The maximum efRciency of the Rrst pro-
cess (see eqns [7] and [8]) is achieved when the strip- A number of modern technologies include water
ping of the product is carried out at 308 K, followed treatment processes, which in many instances involve
by crystallization of K2SO4 at 293 K to provide the a calcium removal stage. IE methods are widely
highest difference in sodium and potassium sulfate applied to solve this problem for low mineralized
solubilities inside the column. The second process (see surface waters. The problem of processing highly
Figure 1 Flowsheet of process for cation exchange synthesis of chlorine-free potassium sulfate. (Reproduced from Muraviev D,
Khamizov R, Tikhonov N et al. (1998). Clean ion-exchange technologies. I. Synthesis of chlorine-free potassium fertilizers by an
ion-exchange isothermal supersaturation technique. Ind. Eng. Chem. Res. 37: 1950}1955, with permission from the American
Chemical Society.
mineralized waters is far more complicated. For after the loading with Ca2#. The most efRcient de-
example, preliminary treatment of seawater prior to sorption of Ca2# from zeolite is achieved by using an
its further desalination requires extensive decalcina- NaCl}Na2SO4 mixture. The reaction of Ca2#}Na#
tion to solve the problem of gypsum formation on exchange is coupled in this case with the reaction of
heater surfaces of distillers and clogging of mem- CaSO4 formation and, as a result, the equilibrium in
branes in reverse osmosis or electrodialysis units. the system is shifted to the right. The overall desorp-
Modern seawater-processing technologies, such as IE tion process is described by the following equation:
recovery of magnesium, also require preliminary re-

moval of calcium. y
The problem of calcium removal from seawater x# R2-Ca#xNa2SO4#yNaCl
2
(and highly mineralized waters and brines) can be
successfully solved by using an IXISS-based process y
"(2x#y)R-Na#xCaSO4 # CaCl2 [12]
on specially modiRed conventional (low cost) 2
sorbents with enhanced selectivity for Ca2#. The Rrst
requirement is dictated by the necessity for the efR- The optimal molar ratio of Na2SO4 to NaCl (x/y in
cient use of the sorbent capacity towards Ca2# at eqn [12]) in the regenerating solution is 0.2. In this
&Rve-fold magnesium over calcium excess in the case the regeneration of zeolite requires a close to
seawater under treatment. The second arises from the stoichiometric amount of the regenerating agent. The
need to process 1000 m3 of seawater to produce 1 ton eluate obtained during the stripping stage (Figure 2)
of magnesium. For example, several successive treat- is supersaturated (+5). Nevertheless, it coexists
ments of zeolites (e.g. of A type) with seawater and with the granulated sorbent phase for a long period
a concentrated NaCl solution result in their stable (&24 h at 293 K) as a stable SS. This solution spon-
modiRcation due to irreversible sorption of Mg2#. taneously crystallizes following removal from the col-
An average Ca Mg value for the modiRed zeolite (from umn with the formation of gypsum, which is the only
seawater) rises to &27 in comparison with 4.5 for waste product in the process. After removal of the
the unmodiRed sorbent. The sorbent removes vir- CaSO4 precipitate by Rltration, the supernatant can
tually no Mg2# from seawater, whereas Ca2# uptake be fortiRed with the desired amount of NaCl}Na2SO4
appears to be nearly equal to its capacity mixture and returned to the process for reuse. The
(&4 mmol kg\1). At the same time, Ca Na for the modi- process is carried out in a two-Rxed bed column
Red zeolite remains at a sufRciently low level (&3.6) set-up. The columns operate intermittently in a cal-
to simplify its regeneration with sodium salt solutions cium removal and regeneration mode.
Figure 3 Concentration}time history of desorption of Mg2#

with 1.5 mol L\1 Na2CO3#0.60 mol L\1 NaHCO3 (circles) from
carboxylic resin pre-loaded with decalcinated natural seawater at
293 K. Triangles, Mg2# concentration in supernatant after short-
term heating (during 10 min from 290 to 310 K) followed by crys-
tallization of supersaturated solution samples. (Reproduced from
Figure 2 Desorption of Ca2# from modified zeolite A with Khamizov R, Muraviev D, Tikhonov N et al. (1998). Clean ion-
1.25 mol L\1 NaCl#0.25 mol L\1 Na2SO4 mixture (circles). Tri- exchange technologies. II. Recovery of high-purity magnesium
angles, calcium concentration in supernatant after crystallization compounds from seawater by an ion-exchange, isothermal super-
of supersaturated solution samples. (Modified with permission saturation technique. Ind. Eng. Chem. Res. 37: 2496}2501, with
from Muraviev et al., 1998). permission from the American Chemical Society.
Recovery of High Purity Magnesium Compounds (MgCO3), nesquegonite crystals are calcium-free. The
from Seawater 3 yield of MgCO3 3H2O depends on the conditions of
The traditional magnesium-from-seawater tech- crystallization of the SS collected. Thus, crystalliza-
nology includes mixing the raw seawater with ‘milk tion at ambient temperature over several hours gives
of lime’ Rltration of Mg(OH)2 slurry, followed by its &70% yield of the product in one desorption cycle.
treatment with HCl, evaporation, drying and electro- A rapid increase of temperature in the crystallizer
lysis. The process does not allow for producing sufR- from &290 K to &310 K over 10 min) allows for
ciently pure Mg. The possibility of designing an alter- a substantial increase in the rate of crystallization,
native IXISS-based process for recovery of high purity which results in the rise of the product yield to
Mg compounds from seawater has appeared from the '90%.
discovery of IXISS of MgCO3 in the resin bed. The The block scheme of the pilot unit for recovery
IXISS effect is observed during elution of Mg2# from of high purity MgCO3 from seawater is shown in
carboxylic resin pre-loaded with decalcinated sea- Figure 4. Ca-free seawater (see previous section)
water with a solution of Na2CO3}NaHCO3 mixture. passes from the top to the bottom through two of the
Magnesium carbonate does not precipitate in the three columns C1}C3, loaded with carboxylic resin in
column and remains as a stable SS (with +5) at the Na form. At the same time the third column is
least over a period of 72 h. After removal of this working in the regeneration (magnesium-stripping)
solution from the column (Figure 3), the desorbed mode of operation. After conversion of resin in the
magnesium spontaneously crystallizes in the form of Mg form the columns are treated from the bottom to
well-shaped nesquegonite (MgCO3 3H2O) crystals. the top with a stripping solution of 1.5 mol L\1
The purity of magnesium compound obtained ap- Na2CO3#0.6 mol L\1 NaHCO3 mixture (also con-
pears to be '99.9% since, unlike magnesite taining a residual MgCO3 from the recycled stripping
solution). Ca-free seawater, displaced from the col-
umns, is directed to tank T1 until the appearance of
3
supersaturated eluate which, in turn, is directed to
Adapted with permission from Muraviev D, Khamizov R,
tanks T2 and T3 (crystallizers supplied with heating
Tikhonov N et al. (1998). Clean ion-exchange technologies. II.
Recovery of high-purity magnesium compounds from seawater by and Rltration facilities to collect the crystalline prod-
an ion-exchange isothermal supersaturation technique. Industrial uct) by reswitching the automatic valve V1. After
Engineering and Chemistry Research 37: 2496}2501. crystallization and removal of magnesium carbonate,
Figure 4 Schematic diagram of experimental pilot unit for recovery of high purity magnesium compounds from seawater: ion
exchange columns C1}C3; solution tanks T1}T6; solution pumps P1 and P2; automatic valves V1 and V2. (Reproduced from Khamizov R,
Muraviev D, Tikhonov N et al. (1998). Clean ion-exchange technologies. II. Recovery of high-purity magnesium compounds from
seawater by an ion-exchange isothermal supersaturation technique. Ind. Eng. Chem. Res. 37: 2496}2501, with permission from the
American Chemical Society.
the stripping solution is returned to tanks T4 and Gibbs}Helmholtz equation:

T5 for fortiRcation with the desired amount of the

Na2CO3}NaHCO3 mixture from T6 and reuse. Then G
G"H#T [13]
the sorption cycle is repeated. The stripping solution T P
displaced from the columns is also returned to T4
and T5 until the appearance of treated seawater in the and the isotherm of an IE reaction by:
line that is controlled by the automatic valve V2.

1/zA 1/zBz
Hence, the process shown in Figure 4 is totally free of a\
AR aBX 6
waste. GP,T"GoT#RT ln 1/zB 1/zAzX [14]
a\
BR aAX
where Go"!RT ln K, and K is the thermodyn-

amic equilibrium constant of the IE reaction corre-
Dual-temperature IE Processes
sponding to the condition GP,T"0.
The efRciency of the dual-temperature IE process is Eqns [13] and [14] lead to the van’t Hoff equation:
primarily determined by the temperature sensitivity

of the IE system which, in turn, depends on the value ln K H
" [15]
of the heat effect of a given IE reaction. The thermo- T P RT 2
dynamics of a reversible IE reaction in the system
including, for example, counterions AzA# and BzB#, Under standard conditions (when K depends only on
a co-ion XzX\ and a cation exchanger R bearing temperature), eqn [15] can be rewritten in conven-
univalent functional groups can be described by the tional derivatives and then integrated from T1 to T2.
In the simplest case, when H is independent of operation, e.g. thermosorption or thermostripping.

temperature, one obtains: Respective concentration}volume histories are shown
in Figure 6 for the Mg concentration process. The

KT2 H 1 1 resulting eluate is either collected in tank Conc. 1 (Mg
ln "! ! [16]
KT1 R T1 T2 or Br concentrate after thermostripping) or returned
to the sea (after thermosorption). The concentrate
The resulting relationship [16] is widely used to obtained after the Rrst column is subjected to the
describe dual-temperature IE processes. same sequence of operations on the second column to
produce the second concentrate, and so forth.
Concentration of Magnesium and Bromine
In a subsequent treatment of cold and hot seawater
from Seawater
in a Rxed-bed IE column, the concentration of Br\ in
At present, around 25% of overall world production the hot stripping solution increases by a factor of 2,
of magnesium and 70% of that of bromine is pro- while the concentration of Cl\ and SO24\ decreases.
vided from the sea and other hydromineral resources. The multistage process (using AV-17 anion exchange
Traditional methods for producing Mg (see previous resin, Reakhim, Moscow, Russia, Russian analogue
section) and Br (air-stripping technique) by process- of strong base anion exchangers such as, e.g. Dowex-
ing seawater, despite their proRtability, do not satisfy 1, Amberlite IRA 400, Purolite A 400 and Lewatit
increasingly stringent ecological standards. Conse- M 500) enriches Br\ concentration in the Rnal
quently, new alternative ecologically clean technolo- concentrate up to an acceptable level for further pro-
gies, based on IE separation methods, have to be cessing ('5 g L\1). Four dual-temperature sorption-
developed. The dual-temperature IE concentration of stripping cycles (using Lewatit R 250 K resin, Bayer,
Mg and Br from seawater is based on the strong Germany) allow for concentrating Mg up to
temperature dependence of the separation coefRcients &0.4 mol dm\3 (versus 0.11 mol dm\3 in the initial
Na and Cl of weak acid (for Mg) and strong base (for
Mg Br
seawater), as shown in Figure 7. The Rnal concentrate
Br) IE resins, respectively. For example, the Br
Cl value can be decalcinated and used for producing high purity
decreases by a factor of &2, while Mg
Na value increases magnesium (see above). The resins used in both pro-
by a factor of &1.2 when the temperature of sea- cesses do not require any regeneration. Hence, the
water rises from 283 to 363 K. The principle of the processes are completely free of waste products.
dual-temperature concentration of Mg and Br by us- Reduction in energy expenditure for heating the
ing a cascade of Rxed-bed columns is shown in Fig- seawater can be successfully solved by using con-
ure 5. The Rrst column is intermittently treated with ventional, or concentrated, sunlight (in areas with
hot and cold seawater depending on the mode of a high level of solar radiation) as the principle and
Figure 5 Flowsheet of reagentless dual-temperature ion exchange processing of seawater. (Reproduced from Muraviev D, Noguerol
J and Valiente M (1997) Seawater as auxilliary reagent in dual-temperature ion-exchange processing of acidic mine waters. Environ.
Sci. Technol. 31: 379}383, with permission from the American Chemical Society.
Figure 7 Concentration of Ca2# (circles), Mg2# (squares) and

Na# (triangles) obtained in consecutive thermostripping}sorption
cycles vs number of cycles. (Reproduced from Muraviev D,
Noguerol J and Valiente M (1996) Separation and concentration
of calcium and magnesium from seawater by carboxylic resins
with temperature-induced selectivity. Reactive and Functional
Polymers 28: 111}126, with permission from Elsevier Science.
for recovering minerals (e.g. Mg and Br). In this way,

the power costs in heating seawater and moving it
through the mineral recovery process (which substan-
tially exceeds 50% of the overall expenditures for
electricity) could be written off.
Concentration of Copper from Acidic Mine Waters4

The treatment of acidic mine waters (AMW), repres-
enting natural efSuents from pyritic ore deposits, is of
great economic and ecological importance. The
Figure 6 (A) Thermostripping (T"263 K) and (B) thermosorp- AMW are characterized by low pH (&2) and rela-
tion (T"353 K) breakthrough curves obtained from natural sea-
water (Mediterranean sea) on polyacrylic Lewatit R 250-K resin.
tively high concentrations of metal ions such as Fe3#,
Circles, calcium; squares, magnesium; triangles, sodium. (Repro- Zn2# and Cu2# (Table 2). The initial AMW requires
duced from Muraviev D, Noguero J and Valiente M (1996). Separ- pre-conditioning by selective removal of iron prior to
ation and concentration of calcium and magnesium from sea- IE treatment. This conditioning is carried out by
water by carboxylic resins with temperature-induced selectivity. adjusting the pH to 3.4}3.5 with alkali followed by
Reactive and Functional Polymers 28: 111}126, with permission
from Elsevier Science.
either conventional or biooxidation of Fe(II) to
Fe(III). The Rnal removal of the Fe(OH)3 precipitate is
carried out by Rltration.
ecologically clean energy source (sun-boiler systems). Two commercial IE resins, namely a polyacrylic
An alternative solution can be the use of seawater in (e.g. Lewatit R 250-K) and an iminodiacetic (e.g.
the cooling cycles of steam power stations. The
amount of seawater pumped through the cooling 4
Adapted with permission from Muraviev D, Noguerol J and
cycles of these stations is approximately 20 000}
Valiente M (1997). Application of the reagentless dual-temper-
30 000 m3 h\1. This hot seawater is currently pum- ature ion-exchange technique to a selective separation and concen-
ped back into the sea, which means that billions of tration of copper versus aluminium from acidic mine waters.
joules of heat are wasted. This seawater could be used Hydrometallurgy 44: 331}346.
Table 2 Composition of initial and conditioned samples of native acidic mine waters (AMW) from Rio Tinto area (Huelva, Spain)a
AMW sample Concentration (mg L\1) pH
SO 24\ Fe Cu Zn Al Mn Mg Ca
1 (initial) 16 450 5050 239 912 399 75 751 326 1.9

1 (conditioned) 16 300 3 235 890 386 73 735 319 3.5
2 (conditioned) 17 350 0.3 115 1275 530 90 950 475 3.5
a
The AMW samples have been collected from the natural generic metal-bearing effluents originated from the pyritic ore deposits typical
for the southern provinces of Spain and Portugal.
Lewatit TP-207) can be used for the dual-temperature produce magnesium or bromine concentrates in
IE concentration of copper from AMW. Acrylic resin a continuous mode of operation.
is selective for Al3# (over other AMW metal ions)
and iminodiacetic resin manifests high selectivity to-
wards Cu2#. The uptake of Al3# increases remark- Concluding Remarks
ably while that of Cu2# decreases with temperature
The number of large scale industrial applications of
for both resins and that for the rest of the metals
both dual-temperature IE and IXISS-based IE pro-
depends weakly on temperature. This effect gives the
cesses is still very limited. For example, the dual-
possibility for the dual-temperature IE concentration
temperature partial demineralization of surface
of copper from AMW by using the same set-up and
waters using specially synthesized polyampholyte
the mode of operation as shown in Figure 5. The
resins (so-called Sirotherm process) so far remains the
resins are Rrst equilibrated with cold AMW (at
only industrial application of dual-temperature IE. At
293 K). Then a selective thermostripping of Cu2# is
present, a large scale pilot plant using the counter-
carried out using hot AMW (at 353 K), leading to an
current version of an IXISS-based process for the
increase in Cu2# concentration in the eluate by a fac-
recovery of more than 300 tons of high purity magne-
tor of 1.3 (for acrylic resin) or 1.2 (for iminodiacetic
sium carbonate from seawater per year has started
resin). The concentration of Al3# in the same eluate
operation in the Vladivostok region of Russia. The
drops to 50% in the Rrst case and to 70% in the
unit shown in Figure 4 adequately imitates the basic
second. The efRciency of both thermosorption and
Sowsheet of Vladivostok plant in the Rxed-bed mode
thermostripping processes (at constant solution Sow
of operation.
rates and resin bed heights) depends on the interval of
The prospects for wider implementations of eco-
working temperatures (see eqn [11] and is higher the
logically clean IE processes in different Relds of indus-
greater this interval (Figure 8).
try are primarily determined by their obvious advant-
The concentration of Cu2# achieved after the
ages and relative simplicity. For example, the use of
fourth thermosorption}thermostripping cycle in-
IXISS effect in the design of highly efRcient and eco-
creases (in comparison with the initial AMW) by
logically safe IE technology does not require any
a factor of &3, while that of Al3# decreases by one
speciRc IE equipment and can be easily realized using
order of magnitude (Figure 9). The substantial in-
standard IE columns. At the same time, a deeper
crease in the efRciency of the process (higher degree of
insight into the IXISS phenomenon can substantially
concentration) can be achieved using either higher
widen the area of practical application of this effect.
Rxed resin bed columns (see Figure 5) or a cascade of
In this regard, the solution of the following problems
countercurrent columns, as shown in Figure 10. The
seems to be of particular importance:
unit comprises several two-section countercurrent
columns which are operated at different temperatures
to provide thermosorption/ thermostripping condi- 1. IdentiRcation of chemical compounds (both or-
tions in the sections. The Rrst column is fed by the ganic and inorganic) exhibiting IXISS effect
native AMW and produces the Rrst concentrate, (IXISS-active compounds)
which is collected from the boundary between sec- 2. Evaluation of stabilizing efRciency of commercial-
tions and then directed to the bottom section of the ly available IE materials towards SS of IXISS-
second column and so on. The resin in all columns active compounds of different types (electrolytes,
circulates in a closed cycle and does not require regen- nonelectrolytes, polyampholytes and zwitterlytes)
eration. The unit shown in Figure 10 can also be used 3. Development of theoretical fundamentals of IXISS
for the dual-temperature IE processing of seawater to effect and some others
certain instances the design of special IE equipment

such as, for example, two-sectional jacketed IE col-
umns providing the dual-temperature mode of opera-
tion. At the same time, it seems useful to emphasize
that the shift of chemical equilibrium to the desired
direction due to the modulation of temperature in the
system represents one of the basic physicochemical
concepts which are widely used in different Relds of
chemical technology. From this viewpoint the dual-
temperature IE cannot be considered as a somewhat
exotic separation method as in many instances the
experience (both theoretical and experimental) accu-
mulated in other areas of chemical science and engin-
eering can be successfully applied for the further
development of this fractionation technique. On the
other hand, it seems useful to distinguish the follow-
ing problems, the solution of which can help to widen
the application of dual-temperature IE in industry:
1. the evaluation of temperature sensitivity of com-

mercially available ion exchangers towards
different ionic systems of practical interest to esti-
mate their potential use in dual-temperature IE
processes
2. the tailored design and synthesis of temperature-
responsive IE resins tuned for the dual-temper-
ature IE fractionation of certain ion mixtures. Ion
exchangers of this type are not so far commercially
Figure 8 (A) Thermostripping and (B) thermosorption break-

through curves for Cu2# and Al3# obtained from native AMW
(see Table 2) on iminodiacetic Lewatit TP-207 resin at various
thermostripping temperatures: 313 K (circles); 333 K (squares);
353 K (triangles). Thermostripping is carried out at indicated tem-
perature after loading of resin at 293 K. Thermosorption is carried
out at 293 K after finishing the thermostripping cycle at the in-
dicated temperature. (Reproduced from Muraviev D, Noguerol
J and Valiente M (1997) Application of the reagentless dual-
temperature ion-exchange technique to a selective separation
and concentration of copper versus aluminium from acidic mine
waters. Hydrometallurgy 44: 331}346, with permission from Else- Figure 9 Concentrations of Cu (circles) and Al (squares) ob-
vier Science. tained in consecutive thermosorption}stripping cycles from AMW
on iminodiacetic Lewatit TP-207 resin vs number of cycles. (Re-
produced from Muraviev D, Noguerol J and Valiente M (1997)
Several recent publications and reviews by the author
Application of the reagentless dual-temperature ion-exchange
are recommended to those interested in this subject. technique to a selective separation and concentration of copper
Unlike IXISS-based IE processes, large scale indus- versus aluminium from acidic mine waters. Hydrometallurgy 44:
trial applications of dual-temperature IE requires in 331}346, with permission from Elsevier Science.
Figure 10 Flowsheet of continuous process for dual-temperature ion exchange treatment of acidic mine waters. (Reproduced from
Muraviev D, Noguerol J and Valiente M (1997) Application of the reagentless dual-temperature ion-exchange technique to a selective
separation and concentration of copper versus aluminium from acidic mine waters. Hydrometallurgy 44: 331}346, with permission from
Elsevier Science.
available. Their appearance can dramatically organic, and biological substances. Solvent Extraction
stimulate the further development and wider ap- and Ion Exchange 16: 1}74.
plication of dual-temperature IE techniques Grevillot G (1986) Principles of parametric pumping. In:
3. the combination of dual-temperature-based and Cheremisinoff NP (ed.) Handbook of Heat and Mass
conventional IE processes within one technolo- Transfer, vol. 2, Mass Transfer and Reactor Design,
pp. 1429d1474. Houston: Gulf Publishers.
gical Sowsheet will increase the efRciency of the
Khamizov RKh, Muraviev D and Warshawsky A (1995)
whole process Recovery of valuable mineral components from sea-
Interested readers can Rnd more detail on the sub- water by ion exchange and sorption methods. In:
ject in the recent review by Muraviev et al. Marinsky A and Marcus Y (eds) Ion Exchange and
Solvent Extraction, vol. 12, pp. 93}148. New York:
See also: II/Ion Exchange: Organic Ion Exchangers; His- Marcel Dekker.
torical Development; Theory of Ion Exchange. Muraviev D, Khamizov RKh and Tikhonov NA (1998)
Ion-exchange isothermal supersaturation. Solvent Ex-
Further Reading traction and Ion Exchange 16: 151}222.
Muraviev D, Noguerol J and Valiente M (1999) Dual-
Arkhangelsky LK and Belinskaya FA (1982) Ion Exchange in temperature ion-exchange fractionation. Solvent Ex-
Chemical Technology (in Russian). Leningrad: Khimiya. traction and Ion Exchange 17: 767d850.
Bobleter O and Bonn G (1991) Ion exchange chromatogra- Streat M (1991) General ion exchange technology. In: Dor-
phy. In: Dorfner K (ed.) Ion Exchangers, pp. 1187}1242. fner K (ed.) Ion Exchangers, pp. 685}716. Berlin: Wal-
Berlin: Walter de Gruyter. ter de Gruyter.
Gorshkov VI (1995) Ion exchange in countercurrent col- Tondeur D and Grevillot G (1986) Parametric ion exchange
umns. In: Marinsky A and Marcus Y (eds) Ion Exchange processes (parametric pumping and allied techniques).
and Solvent Extraction, vol. 12, pp. 29}92. New York: In: Rodrigues AE (ed.) Ion Exchange Science and Tech-
Marcel Dekker. nology. NATO ACI Series, vol. 107, pp. 369d399. Dor-
Gorshkov VI and Ivanov VA (1999) Reagent-free ion-ex- drecht: Martinus Nijhoff.
change separation. Solvent Extraction and Ion Ex- Wankat PC (1981) Cyclic separation techniques. In: Rod-
change 17: 695}766. rigues AE and Tondeur D (eds) Percolation Processes.
Gorshkov VI, Muraviev D and Warshawsky A (1998) Ion- Theory and Applications, pp. 443d515. Rockville:
exchange methods for ultra puriRcation of inorganic, Sijthoff and Noordhoff.
III / ELECTROCHEMICAL ION EXCHANGE 2655
ELECTROCHEMICAL ION EXCHANGE
J. P. H. Sukamto, S. D. Rassat, R. J. Orth and ‘EIX’ is used more broadly here to denote a set of
M. A. Lilga, Pacific Northwest National separation techniques that use electrochemistry and
Laboratory, Richland, WA, USA IX, not just the process shown in Figure 2.) Here,
Copyright ^ 2000 Academic Press a cathodic potential is applied to the electrode mater-
ial assembly for the uptake of cations. The applied
potential serves to generate an electric Reld that
Introduction increases the transport rate of cations toward
Electrochemical ion exchange (EIX) is a process the electrode material assembly and to electrolyse
where electrochemistry and ion exchange (IX) are water to produce hydroxyl ions that activate the IX
combined to effect the separation of ions more efR- material. To elute the sorbed cations, an anodic
ciently than the use of either technique alone, espe- potential is applied to the electrode/cationic ion
cially in minimizing secondary waste generation. The exchange material assembly. This results in the local
varieties of EIX, beginning with those using electro-
chemically inactive IX materials and proceeding to
those using electrochemically active ion exchange
(EaIX) materials, are discussed here.
EIX Processes Using

Electrochemically Inactive Ion
Exchange Materials
Some materials used in the EIX process are not active
electrochemically, that is, they do not contain func-
tional groups that can be reduced or oxidized. Exam-
ples of commonly used separation techniques of this
type include electrodialysis and electrophoresis/elec-
trochromatography. In electrodialysis, the permselec-
tivity of IX membranes is used in combination with
an electric Reld to separate anions and cations. In
a variation of conventional electrodialysis called elec-
trodiaresis polishing, IX materials are packed be-
tween the IX membranes; the IX material serves as
an immobile electrolyte. The schematic of the
process shown in Figure 1 shows that the water is
processed through the IX materials. Due to the high
concentration of ionic sites in the IX material, fairly
resistive (up to 15 M) water can be produced at
the outlet. In electrophoresis/electrochromatography
where IX packing materials are used, IX inter-
actions between the analyte and the column material
enhance the separation capability over that available
if differences in electrophoretic mobilities alone were
used.
Another process in this category, a process termed Figure 1 Cationic and anionic IX resins are depicted as
EIX by the original researchers (Allen et al.), uses an hatched and open areas, respectively. Only cations are mobile in
the cationic IX material, and only anions are mobile in the anionic
assembly consisting of an electrochemically inactive
IX material. Cations in the feed water are expelled from the water
IX material attached to an electrode as the separation while flowing through the cationic IX material. Similarly, anions
agent. This approach, for the separation of cations, is are expelled from the feed water while flowing through the anionic
shown in Figure 2. (It should be noted that the term IX material.
2656 III / ELECTROCHEMICAL ION EXCHANGE
of cations. For anions, anionic IX materials are

used; an anodic potential is applied to the assem-
bly during uptake and a cathodic potential is applied
during elution. The advantages of this approach
over conventional IX processes are increased up-
take rate due to the applied electric Reld and more
efRcient use of hydrogen or hydroxyl ions for
both uptake and elution. Although the thermo-
dynamics of the elution process have not changed
as compared to conventional IX, the close mity of
the generated hydrogen or hydroxyl ions to the ex-
change sites increases the overall process rate by re-
ducing the time required for uptake and elution. In
addition, fewer excess hydrogen or hydroxyl ions are
required.
Figure 2 Cation uptake (top) and elution (bottom) at the EIX
electrodes. Water reduction leads to production of hydroxyl ions
resulting in the de-protonation of sites in the cationic IX resin. In EIX Processes Using Electroactive
turn, the IX resin will uptake cations to maintain electroneutrality. Ion Exchange Materials
The electric field, E, present during the water-reduction process,
serves to enhance cation transport to/within the cationic IX resin. Figures 3(A) to (D) show conceptually the electroac-
The reverse reactions occur during cation elution. tive ion exchange (EaIX) approach, a subset of EIX
techniques. Here, EIX processes, where EaIX mater-
ials are used, will be denoted as EIX/EaIX. Fig-
production of hydrogen ions from the splitting of ure 3(A) and (B) illustrate cation separation. The
water, which in turn displaces the sorbed cations as in reduction of a generic EaIX material, X, results in
conventional IX processes. In addition, an electric a net negative charge that has to be compensated by
Reld is generated which accelerates the transport rate a cation (see Figure 3A). Provided that X is selective
Figure 3 (A) Cation intercalation/uptake during film reduction. (B) Cation de-intercalation/elution during film oxidation. (C) Anion
intercalation/uptake during film oxidation. (D) Anion de-intercalation/elution during film reduction.
for a speciRc cation, then the cation uptake shown in ions, however, still relies on the speciRc interactions
Figure 3(A) results in the selective separation of a ca- between the stationary phase and the ions.
tion, M# # #
1 , from, say M2 . Elution of M1 is simply EaIX materials may also be used for preparative-
achieved by oxidizing X\ back to X, which is neces- scale separations. Two modes of operation are pos-
sarily accompanied by the expulsion of M# 1 (see sible: packed bed and membrane. The sequence of
Figure 3B). The analogous uptake and elution of an- operational events in the packed bed mode in
ions are shown in Figure 3(C) and (D), respectively. EIX/EaIX is a combination of the sequences used in
For anions, uptake occurs during oxidation, and elu- electrochemical chromatography and conventional
tion accompanies reduction. It should be noted that IX processes: Rrstly, analyte is sorbed into the EaIX
the same sequence of oxidation and reduction is used material during the uptake cycle (with or without the
for the EIX/EaIX approach and the process shown in application of a potential) and secondly the sorbed
Figure 2. More importantly, however, the two ap- ions are eluted from the EaIX material by the applica-
proaches are very different with respect to the speciRc tion of a potential (see Figure 4). It should be noted
electrochemical reactions that take place during oxida- that EaIX materials can be used as conventional IX
tion and reduction. In the process shown in Figure 2, materials during the sorption cycle; the distinguishing
both oxidation and reduction reactions are water-split- feature of the EaIX material is the ability to elute
ting. The reductive water-splitting results in hydrogen sorbed ions by simply applying the appropriate po-
gas and hydroxyl ion production, whereas the oxida- tential (Figure 3B and D). In the second mode of
tive water-splitting results in oxygen gas and hydrogen operation, the EaIX materials are made as mem-
ion production. While this approach results in relative- branes (see Figure 5). This mode of operation is
ly efRcient use of the protons and hydroxyls, the ex- superior to the packed-bed approach since solution
change ratio is not one to one. That is, more than one switching is not required. This in turn implies that
equivalent of protons is required to displace one equiv- continuous operation is possible and that neither the
alent of cations (or uptake one equivalent of anions). process nor waste stream will be diluted, which neces-
In addition, in the processing of radioactive materials, sarily occurs when solution switching takes place in
hydrogen gas generation is undesirable because of the packed-bed mode. On the other hand, the packed-
safety issues. On the other hand, with the use of EaIX bed mode is more advantageous in treating very
materials (as shown in Figure 3A to D), only one
equivalent of electrons is needed to elute one equiva-
lent of ions. Therefore, using EaIX materials further
reduces the amount of secondary waste generated. The
electric Reld generated within EaIX materials has two
functions: Rrstly to increase the transport rate of ions
through the EaIX material and secondly to attract (or
expel) the ions to the binding sites since the electric
Reld is strongest near an oxidized or reduced site. The
last point is particularly noteworthy since it implies
that highly selective materials can be used. Although
the requirements of high selectivity and ease of elution
are typically in conSict, the two requirements are not
mutually exclusive if the elution is purely electrostatic
in nature.
EaIX materials have been applied for analytical-
scale separation in a process called electrochemical
chromatography. This process is very similar to con-
ventional liquid chromatography except that the sta-
tionary phase is electroactive. As in conventional
chromatographic processes, the retention times for dif-
ferent ions are largely controlled by interactions with
the stationary phase. However, the redox state of the
stationary phase allows additional control of the inter- Figure 4 EIX/EaIX columns consist of the EaIX electrodes and
counterelectrodes. Connections to the elution solutions are
actions between the analytes and the stationary phase.
closed during sorption of target ions from the process waste
For example, anions can be easily separated from stream. Sorption can be with or without an applied potential (see
cations by applying a positive or anodic potential to eqns [2] and [3] in the text). Elution is achieved by applying
the stationary phase. Separation of the different an- a potential (see Figure 3B and D) and flowing the elution solution.
Figure 5 The EaIX membrane is always an electrode when used in the membrane mode. (A) Reduction of EaIX membrane (negative
electrode) results in uptake of cations from the treated stream. (B) Oxidation of EaIX membrane (positive electrode) results in elution of
cations to the waste stream. The treated and waste streams never come in contact.
dilute streams since more intimate contact between Electrochemical Methods

the process stream and the EaIX materials is possible.
In addition to standard bulk contact methods typi-
In addition, there are possibly more difRculties asso-
cally used to characterize IX materials, electrochemi-
ciated with preparing EaIX materials as membranes
cal methods are well suited for evaluating EaIX
than as packed-bed materials.
materials, especially for evaluating their expected
performance when an applied potential is used in the
Characterization of EaIX Materials uptake cycle. Two conventional electrochemical
methods } cyclic voltammetry and chronoam-
In the development and testing of an EIX/EaIX pro- perometry } are particularly useful in evaluating the
cess, the ion-loading capacity and speciRc ion selec- ion-loading capacity of EaIX materials. Cyclic vol-
tivity of the EaIX materials are of particular interest. tammetry is best suited for a planar geometry since
Selectivity is deRned in terms of a separation factor 21,
which describes the selectivity for species ‘2’ over
species ‘1’:
x2/x1
21" [1]
x2/x1
The numerator is the ratio of ion mole fractions

within the EaIX material, and the denominator is the
mole fraction (or concentration) ratio in the bulk
binary solution contacting the EaIX material. The
separation factor is especially useful for estimating
process performance. For example, Figure 6 shows
the fraction of Na# recovered as a function of
K# removed for various values of KNa. The plots show
that if a minimum of 90% Na# recovery is required
while expelling '70% of the K#, then an KNa of Figure 6 Fraction of Na# recovered as a function of K# re-
approximately 30 is needed. moved for various separation factor, Na
K .
the slowly ramped applied potential is less intrusive analysis is complicated, but not prohibitively, by the
on the samples. Chronoamperometry, on the other simultaneous transport of solvent (e.g., hydration
hand, is more suitable for determining the capacity of water) and other species with the ions of interest.
EaIX materials deposited on a high-surface area elec-
trode because mass transfer limitations affect mea- Flow Cell Methods
sured cyclic voltammograms (CVs). Flow cell methods for EIX/EaIX characterization dif-
A CV is obtained by measuring the current passed fer from conventional IX methodologies only in the
at the electrode as the applied potential is swept material preparation. To achieve high volumetric ion-
linearly at a Rxed scan rate between two potential loading capacity, EaIX materials are typically coated
limits. When a suitable potential window is selected, on high surface-area electrodes (e.g., porous metal
cations (anions) are loaded into the EaIX material foam or mesh). These EaIX electrodes can be used in
during the cathodic (anodic) sweep and the ions are any of the modes represented in eqns [2] and [3]. To
eluted during the reverse sweep. The total charge create an EaIX bed, multiple porous electrodes are
passes in a load or unload cycle, as determined by used in series (see Figure 7A and B). When used in the
integrating the current over any linear sweep, is a di- conventional mode in the uptake cycle, the process
rect measure of the ion-loading capacity when the stream simply Sows through the electrodes. The efSu-
charge transfer is only a result of a known electro- ent solution composition is analysed and break-
chemical reaction (see eqn [2] below). In chronoam- through curves are obtained.
perometry, the potential is stepped from one potential
limit to the other. The current is measured while the
potential is held at a limit for a Rxed period of time,
Properties of EaIX Materials
and again, the integrated charge passed is a measure Functional requirements of EaIX materials are elec-
of the ions loaded into or eluted from an EaIX electronic conductivity, ionic conductivity, selectivity for
trode assembly. the ion of interest, and reasonable stability (physical
and chemical). Of the numerous inorganic and or-
ganic materials that fulRl all or some of the listed
Electrochemical and Gravimetric Methods
requirements, discussions will be limited to nickel
The two electrochemical methods described above hexacyanoferrate (NiHCF) as an example of
yield only the ion-loading capacity. In order to deter- a cationic EaIX material and polyvinylferrocene
mine the separation factor, an additional independent (PVF) as an example for an anionic EaIX material.
measurement is required. This is accomplished by Hexacyanoferrate materials (including NiHCF)
combining the electrochemical measurements with are known to be selective for alkali metals and, in
gravimetric measurements. A common term for this
combined method is electrochemical quartz crystal
microgravimetry (EQCM). As the name suggests,
mass accumulation within the EaIX material is
monitored in addition to the total charge passed.
Typically, quartz crystals are coated with an elec-
trode on which a layer of an EaIX material is depos-
ited. The fundamental frequency of the crystal is very
sensitive to the mass loaded in the EaIX material;
monitoring this fundamental frequency, the mass
loaded during uptake and elution can be determined.
For a crystal oscillating at 5.9 MHz, a 1 Hz decrease
in the frequency corresponds to a mass increase of
&4.0;10\9 g. In EQCM, the oscillation frequency is
monitored at the same time that the potential is swept
to obtain a CV. This combination of measurements
gives simultaneous information on electrochemical
(ion) capacity and mass changes during load and
elution cycles. Since mass change differs according to
the molecular weight of the ions transported in and
out of the EaIX Rlm, the selectivity of the EaIX
material for an ion in a binary mixture of known Figure 7 (A) Single electrode with the associated spacers and
composition can be determined using EQCM. The gaskets. (B) Eight-stack cell used for flow-through studies.
Figure 7 Continued
particular, are extremely selective for cesium (Cs#) centre in NiHCF, an alkali cation associates with the
over sodium (Na#) and potassium (K#). More re- ferrocyanide moiety to maintain charge neutrality
cently, the preference of polyvinylferrocene for per- through the following reaction, where M# is an al-
rhenate (nonradioactive chemical analogue of per- kali metal cation:
technetate) anions over nitrate anions has been dem-
onstrated. MNiIIFeIII(CN)6#M##e\M2NiIIFeII(CN)6 [2]
Nickel Hexacyanoferrate
The reverse reaction, upon oxidation of the FeII
/
Nickel hexacyanoferrate [Ni Fe (CN) \ \]
II II/III 2
6 centre, results in the dissociation of a single alkali
(NiHCF), an electroactive material, is known to com- metal cation per molecule of hexacyanoferrate.
plex reversibly with the alkali metal cations such as Oxidation of NiHCF Rlm deposited on a substrate
Na#, K#, and Cs#. Upon reduction of the iron (FeIII) electrode, therefore, leads to the expulsion or
deintercalation of alkali cations from the Rlm into equipment capital costs scale roughly with the elec-
a contacting solution, while the reduction of depos- trode costs, it is necessary to minimize electrode area
ited NiHCF Rlm leads to the uptake or intercalation to make the EaIX process Rnancially attractive. The
of alkali cations from solution into the Rlm (see Fig- larger the separation factor KNa, the smaller the EaIX
ure 3B and A, respectively). The selectivity for alkali electrode area necessary to remove a given amount of
cations M# by NiHCF increases with molecular K#. Electrode area is also reduced if the ion capacity
weights, Cs#K#'Na#. Therefore, K# or/and of the EaIX material per unit electrode area is in-
Na# is readily exchanged for Cs#; for example: creased.
The selectivity of NiHCF for K# in preference to
Na2NiIIFeII(CN)6#2Cs#Cs2NiIIFeII(CN)6#2Na# Na# was quantiRed by EQCM and by bulk-contact
experiments. Separation factors KNa ranged from
[3] about 5 using EQCM to a maximum of 24 in bulk-
contact tests; these differences will be discussed fur-
Equation [3] shows that the EaIX material can be ther below. Figure 8(A) and (B) show the results of
used in conventional IX. The Cs# is bound so strong- EQCM experiments for a NiHCF-coated quartz crys-
ly that elution is only possible through oxidation of tal in contact with potassium and sodium sulfate
the Fe centre from II to III. Chemical oxidation has solutions and mixtures. The CVs in Figure 8(A) show
been demonstrated with NiHCF as well as for the
copper and zinc analogues. Approximately Rve col-
umn volumes of 8 mol L\1 nitric acid are required for
effective elution of all the sorbed Cs#. The cost and
hazard associated with this eluent are signiRcant. Use
of the EIX/EaIX approach, therefore, provides an
attractive alternative since the oxidation can be done
more efRciently via the electrochemical approach.
It is imperative for the EIX/EaIX process that there
is intimate contact between the EaIX material and the
electronically conducting substrate. NiHCF can be
conveniently deposited onto a conducting substrate
electrochemically. A nickel surface corroded in a
solution containing hexacyanoferrate ions results in
the precipitation/deposition of NiHCF on the surface.
The electrochemical route is particularly advantage-
ous over other methods (e.g., precipitation and sol-
gel) since deposition within the pores of a porous
electrode can be carried out readily.
Applications The selectivity of NiHCF for alkali

cations with an afRnity order Cs#K#'Na# is
attributed to the relative sizes of the ions, both hy-
drated and not, and the NiHCF cubic lattice structure
that the ions must penetrate and then occupy. Be-
cause NiHCF is both electronically and ion conduct-
ing, is readily deposited as a Rlm on conductive elec-
trode substrates, and is alkali cation-speciRc, it is an
ideal EaIX material for K# and Cs# separation ap-
plications.
Figure 8 EQCM results for a series of 0.5 mol L\1 Na2SO4 and
Potassium separation The forest products industry 0.5 mol L\1 K2SO4 solution mixtures demonstrating selectivity of
requires selective removal of K# and recovery of a NiHCF film for K# over Na#: (A) Cyclic voltammograms indi-
Na#. As the plots in Figure 6 show, the separation cate sensitivity to K# in solutions '25 times more concentrated
in Na#; and (B) QCM mass data, normalized by the ion-loading
factor is critical in determining the extent to which capacity and converted to units of apparent molar weight, indicate
Na# can be recovered for a required removal of K#. greater mass changes as solutions become more concentrated in
Therefore, quantiRcation of KNa is essential for scale- K# and relatively more K# is loaded into the film. (Adapted, with
up purposes and capital cost estimation. Because permission from Rassat et al. (1999), Elsevier Science.)
Table 1 Apparent molar masses and separation factors in mixtures of 0.5 mol L\1 sodium and potassium sulfate solutions. (Adapted,
with permission from Rassat et al. (1999), Elsevier Science)
Solution xNa : xK Experiment 1 Experiment 2 Experiment 3
M (g mol\1) Na
K
M (g mol\1) Na
K
M (g mol\1) Na
K
1:0 29.0$0.5 25.3$0.4 22.8$0.4

25 : 1 31.7$0.5 3.8$1.1 30.2$0.6 5.3$0.8 28.4$0.4 5.9$0.7
10 : 1 35.4$0.5 4.6$0.6 35.2$0.6 5.6$0.5 32.7$0.5 5.1$0.4
5:1 39.2$0.6 5.0$0.7 39.4$0.6 5.2$0.5 36.9$0.6 4.6$0.4
0:1 49.4$0.9 53.0$0.9 52.3$0.8
the reversibility of the cation uptake (negative cur- aration factor, KNa, determined by this bulk-contact
rents) and elution (positive currents). In addition, the method, is very sensitive to the total capacity value.
area between the abscissa and either the negative or Since there are more uncertainties in determining the
positive currents is proportional to the net ionic load- total capacity for the foam electrodes (in comparison
ing. Combined with the apparent molar weights to the small planar EQCM electrodes), the variability
shown in Figure 8(B), separation factors ranging and uncertainty of bulk-contact separation factors
from 3.8 to 5.9 were calculated (see Table 1). In obtained are greater. Despite the uncertainties asso-
addition to providing quantitative estimates of the ciated with the total capacity, the batch-contact tests
separation factors, Figure 8(A) and (B) qualitatively are in agreement with the EQCM results in that
show the preference of K# over Na#. As the mix- NiHCF materials are selective for K# over Na#. The
tures become more concentrated in K#, the peaks reason for the difference in the magnitude of the
shift to higher potential, more in line with that of pure separation factors determined by the two techniques
K2SO4 solution. Even in solutions Rve times more is presently unclear. Three possibilities are (1) differ-
concentrated in Na#, the peaks shift substantially ences in NiHCF Rlm preparations resulting from dif-
toward those for pure K2SO4 solution, indicating the ferences in the electrode substrate on which they were
relative selectivity of NiHCF for K#. The shift to- deposited, (2) differences in solution ionic strengths
wards higher apparent molecular weights in the mix- (&1 mol L\1 alkali for EQCM and (0.1 mol L\1
tures also indicates the preference for K#. alkali for bulk contacting), and (3) the potential was
Bulk-contact tests of NiHCF, a more direct applied to NiHCF in the EQCM experiments but not
measure of selectivity, resulted in separation factors in the bulk-contact tests.
ranging from 14 to 24. Representative experimental
details and results are shown in Table 2. In all the Cesium separation The radioactive isotope 137Cs,
tests shown in Table 2, the Na# : K# molar concen- a Rssion product of nuclear fuel processing and cor-
tration ratio was &12, which is the approximate rosion of fuel rods in commercial nuclear reactors, is
ratio of the ions in pulp mill application. The amount a trace component of several process and waste
of K# taken up by the NiHCF without any applied streams in the nuclear industry. Because of the strong
potential (IX mode, eqn [3]) was determined from afRnity of NiHCF for Cs#, separation of this ion by
the total solution volume and the difference in EIX/EaIX is ideal. Figure 9(A) and (B) show the
K# concentration before and after contact. The sep- EQCM results for dilute Cs# in competition with
excess Na#. The experiments are analogous to those
Table 2 NiHCF bulk-contact separation factors and experi- described for K# separation above, but the shifts
mental conditions. (NiHCF-coated circular disc electrodes toward pure Cs# solutions observed in the mixtures
&5-cm diameter by &0.6-cm thick, 80 pores inch\1 porosity, and are more pronounced.
&60 cm2 cm\3 specific volume contacted with 18 mL of mixed Table 3 summarizes the apparent molar masses
#
ion solution)
and separation factors Cs
Na for a series of Na : Cs#
Test [K#] (mM) [Na#] (mM) Capacity (C) Na
K mixtures. The separation factors range from 178 to
593, clearly demonstrating the enhanced selectivity of
Initial Final Initial Final NiHCF for Cs# relative to Na# and K# (compare to
Table 1). Neglecting the results for the 81
1 2.19 1.30 28.0 29.1 2.06 14
Na# : 1 Cs# mixture, which had a higher estimated
2 0.97 0.30 13.0 13.7 2.31 15
3 4.81 3.44 56.8 54.1 1.97 24 uncertainty, there appears to be a trend of increasing
Cs# selectivity with decreasing Cs# concentration.
Table 3 Apparent molar masses and separation factors in

mixtures of 1.0 mol L\1 sodium and cesium nitrate solutions.
Rassat et al. (1999), courtesy of Elsevier Scientific.
Solution xNa : xCs M (g mol\1) Na

Cs
1:0 14.4$0.2
H2390 : 1 31.1$0.9 593$40
H910 : 1 37.3$2.1 341$36
442 : 1 46.1$1.0 268$15
155 : 1 59.3$1.0 178$12
81 : 1 83.0$1.3 361$60
0:1 98.4$2.2
HGravimetric and/or electrochemical measurements not at steady

state.
Regeneration cycles were carried out by potential

cycling in a solution of sodium nitrate. The reasons
for the diminished breakthrough volume after the
Rrst cycle are presently unclear. It is speculated that
this is a ‘plugging’ issue. One hypothesis is that be-
cause of the great afRnity of NiHCF for Cs#
and strong binding in the cubic lattice, migration of
Cs# to all sites within the Rlm is hindered, and
therefore, some Cs# are permanently bound in the
NiHCF. The most noteworthy point is that regenera-
tion of the EaIX material is reversible (starting with
the second cycle) without the need for strong oxidiz-
Figure 9 EQCM results for a series of 1.0 mol L\1 NaNO3 and
1.0 mol L\1 CsNO3 solution mixtures demonstrating selectivity of
NiHCF film for Cs# over Na#: (A) cyclic voltammograms indi-
cate sensitivity to Cs# in solutions &2400 times more concen-
trated in Na#; and (B) QCM mass data, normalized by the ion-
loading capacity and converted to units of apparent molar weight,
indicate greater mass changes as solutions become more con-
centrated in Cs# and relatively more Cs# is loaded into the film.
(Adapted, with permission from Rassat et al. (1999), Elsevier
Science.)
In the limited range of K# and Na# mixtures tested,

such a trend was not detected.
Flow-through EIX/EaIX experiments demonstrate
the regenerability of the EaIX material without the
use of highly oxidizing solutions. Breakthrough
curves for a 0.2-ppm Cs# feed stream are shown in
Figure 10. Here, the EIX/EaIX system was operated
in the conventional IX mode in the uptake cycle. The
breakthrough curve for the Rrst Sow test shows that
the breakthrough point (where the concentration of Figure 10 Breakthrough curves for a feed stream of 0.2-ppm
Cs# in the efSuent stream is one-half the feed concen- Cs in a EaIX bed consisting of eight NiHCF-coated porous nickel
tration) occurs after &110 bed volumes were passed foam electrodes operated in IX mode. The electrodes were regen-
erated electrochemically in concentrated NaNO3 solution be-
in the Rrst Sow test. In subsequent Sow tests, each
tween each test. Experimental conditions: 80 pores inch\1 or
following regeneration of the NiHCF electrodes, the &60 cm2 cm\3 nickel foam; CsNO3 solution flowed at
breakthrough capacity was reduced to &40 bed vol- 24 mL min\1; bed volume of &39 mL; and maximum ion capa-
umes and the breakthrough proRles were consistent. city &2.0 C.
and electrochemical polymerizations should be suit-

able for depositing PVF within porous substrate.
Applications The preference of polyvinylferrocene

for perrhenate (nonradioactive chemical analogue of
pertechnetate) anions over nitrate anions has been
demonstrated. Nitrates are the main competing an-
ions for separation of pertechnetate in radioactive
tank wastes. The current and the frequency responses
of a PVF-coated EQCM as a function of a cyclic
potential scan are shown in Figure 11(A) and (B).
(The frequency response is shown rather than the
normalized mass change because of complications
due to the less rigid PVF Rlms compared to the
NiHCF Rlms.) The more negative (cathodic) potential
peaks observed in the pure ReO\ 4 solution and in the
mixture, compared to a pure NO\ 3 solution, indicate
the preference of ReO\ 4 over NO\ 3 . In addition, the
frequency responses shown in Figure 11(B) indicate
a substantial mass gained in the PVF Rlm upon oxida-
tion in a solution containing both NO\ 3 and ReO\ 4 ,
more mass than can be attributed to NO\ 3 alone. This
supports the contention that ReO\ 4 ions (which are
heavier than NO\ 3 ) are preferentially taken up by the
PVF. The data shown in Figure 11(A) and (B) corre-
spond to a separation factor of 30.
Figure 11 EQCM results for a series of 0.5 mol L\1 NaNO3 and Future Developments
0.5 mol L\1 NaReO4 solution mixtures demonstrating selectivity
of a PVF film for ReO\ 4 over NO\ 3 : (A) cyclic voltammograms
EIX processes (using EaIX and nonelectroactive IX
indicate sensitivity to ReO\
4 in solutions nine times more concen- materials) are very promising methods for ion separ-
trated in NO\ 3 ; and (B) QCM frequency shifts indicate greater ation due to the potential savings resulting from
mass changes as solutions become more concentrated in ReO\ 4
minimization of secondary waste generation. Better
and relatively more ReO\ 4 is loaded into the film.
understanding of system performance through large-
scale (e.g., pilot-scale) studies still needs to be carried
ing solutions (i.e., 8 mol L\1 nitric acid used by pre- out as well as the development of new materials. For
vious researchers). example, EaIX materials selective for strontium
Polyvinylferrocene (Sr2#) are of interest to the nuclear industry. Calcium
(Ca2#) selective materials are valuable for preventing
Polyvinylferrocene [!(FeII/III(C5H5) (C5H4CH2 scale formation in many industries.
1#/0
CH2) )}], or PVF, is a well-studied organometallic Finally, effective removal of NO\ and arsenate an-
3
polymer. In contrast to NiHCF, oxidation of PVF to ions is critical for safe drinking water.
the 1# state requires the uptake of anions to main-
tain electroneutrality (see Figure 3C), making PVF
See also: II/Ion Exchange: Historical Development; Inor-
a suitable anionic EaIX material candidate. Recently, ganic Ion Exchangers; Organic Ion Exchangers; Theory of
the preference of PVF for perrhenate (ReO\ 4 ) over Ion Exchange.
nitrate (NO\ 3 ) was demonstrated. The uptake and
elution reactions are analogues to that shown in
eqn [2]. As with NiHCF, PVF can also be used as Further Reading
conventional IX materials. SpeciRcally, the NO\ 3 in
Genders D and Weinberg NL (eds) (1992) Electrochemistry
(PVF#)(NO\ 3 ) is readily exchanged for ReO\4 . Other for a Cleaner Environment. East Amherst, NY: The
possible applications for PVF include the extraction Electrosynthesis Company Inc.
of arsenates and chromates. Krishnan R and Ibanez J (1997) Environmental Elec-
PVF has been prepared through chemical, electro- trochemistry. Fundamentals and Applications in Pollu-
chemical, and plasma polymerization. Both plasma tion Abatement. San Diego: Academic Press.
III / ELECTRODIALYSIS: ION EXCHANGE 2665
Lewis TM, Wallace GG and Smyth MR (1999) Elec- Rose TL, Rudd E, Murphy O and Conway BE (eds) (1994)
trofunctional polymers: their role in the development of Proceedings of the Symposium on Water PuriTcation by
new analytical systems. Analyst 124: 213. Photocatalytic, Photoelectrochemical, and Electro-
Rassat SD, Sukamto JH, Orth RJ, Lilga MA and Hallen chemical Processes. Pennington, NJ: The Electro-
RT (1999) Development of an electrically switched chemical Society.
ion exchange process for selective ion separations. Tsuda T (ed.) (1995) Electric Field Applications in
Separation and PuriTcation Technology 15: Chromatography, Industrial and Chemical Processes.
207. Weinheim: VCH.
ELECTRODIALYSIS: ION EXCHANGE
G. Pourcelly, Laboratory of Materials and bearing Rxed charges of the same sign and bipolar
Membrane Processes, Montpellier, France membranes bearing positive and negative Rxed
Copyright ^ 2000 Academic Press charges located on each side of the membrane.
IEMs are sheet-shaped materials through which a
selective ion transport can be established under a
Introduction driving force, generally an electric Reld and/or a con-
Separations with synthetic membranes have become centration gradient.
increasingly important; today membrane processes Most of them are of a polymeric nature. They are
are used in a wide range of applications and their constituted of reticulated macromolecular chains
number will certainly increase. forming a tridimensional structure. In this network,
A membrane is a permselective polymer, inorganic ionizable functionalized groups are attached to the
or metal phase which restricts the motion of certain polymeric matrix and are at the origin of the mem-
species. By controlling the relative rates of transport brane selectivity. For example cation exchange mem-
of various species it gives one product depleted in branes (CEMs) contain Rxed negative charges and
certain components and a second product concen- mobile cations which can be exchanged with other
trated in these components. Membrane performance cations present in an external phase in contact with
is characterized by two terms: Sux and selectivity. the membrane. The ions balancing the Rxed exchange
Flux (or permeation rate) is the volumetric mass Sow sites are called counterions. The concentration of
of Suid passing through the membrane per unit area counterions is relatively high and therefore counter-
of membrane and unit mass time. Selectivity is ions carry most of the electric current through the
a measure of the relative permeation rates of different membrane. The Rxed charges attached to the polymer
components through the membrane. matrix repel ions of the same charges (co-ions). This
Processes may be classiRed according to the driving exclusion, which is a result of electrostatic repulsion,
force used: (1) a pressure differential leads to micro-, is called Donnan exclusion, named after F. G. Don-
ultra- and nanoRltration and reverse osmosis; (2) nan, who Rrst reported the phenomenon in 1910.
a concentration difference across the membrane leads However, as the membrane selectivity is never ideal,
to diffusion of a species between two solutions (dialy- the membrane material can be penetrated by a non-
sis); (3) a potential Reld applied to an ion exchange negligible amount of electrolyte. A schematic
membrane leads to migration of ions through the structure of such a homopolar CEM is depicted in
membrane (electrodialysis, membrane electrolysis Figure 1. Under an applied electric Reld, the CEM
and electrochemical devices). This last category and bearing sulfonic exchange groups (}SO\ 3 ) mainly
more speciRcally electrodialysis is the subject of this allows the transport of counterions.
section. This electrically driven process uses ion ex- A bipolar membrane (BPM) is composed of two
change membranes, a description of which follows. layers of ion exchangers joined by a hydrophilic junc-
tion. The diffusion of water from both sides of the
BPM allows it to dissociate under the electrical Reld
Ion Exchange Membranes to generate protons and hydroxyl ions, which further
Electrodialysis (ED) uses membranes containing Rxed migrate from the junction layer through the cation
charged groups attached to the polymer backbone of exchange and anion exchange layers of the bipolar
its membrane. Two kinds of ion exchange mem- membrane. Generally, the cation exchange groups
branes (IEMs) are used in ED: homopolar membranes are sulfonic and the anions are trimethyl quaternary
III / ENVIRONMENTAL APPLICATIONS / Flotation 2675
the second step, the concentrated sodium lactate is Further Reading

split in a two-compartment bipolar ED with a CEM
DauRn G, ReneH F and Aimar P (eds) (1998) Les Se& parations
(conRguration of Figure 10A) to generate lactic acid par Membrane dans les Proce& de& s de l’lndustrie Alimen-
and NaOH. The acid stream, still containing Na# taire, Collection: Sciences et Techniques Alimentaires.
ions, is then puriRed by a cation exchange resin, while Paris: Lavoisier.
caustic soda is recycled to the fermenter for pH con- Davies TA, Genders JD and Pletcher D (1997) lon Per-
trol. For economical reasons in the bipolar ED step, meable Membranes. Alresford, UK: Electrochemical
the conversion rate of sodium lactacte is kept at 95%, Consultancy/Alresford Press.
but almost 100% could be easily achieved. A simpli- Gavach C, Bribes JL, Chapotot A et al. (1994) Improve-
Red schema of this process is reported in Figure 17. ments of the selectivity of ionic transport through elec-
trodialysis membranes in relation with the performances
of separation electromembrane processes. Journal de
Conclusion Physique IV, Colloque C1 4: 233.
ED with homopolar membranes was Rrst developed Mani KN (1991) Electrodialysis water splitting technology.
several years ago, essentially for desalting brackish Journal of Membrane Science 58: 117}138.
Mani KN, Chandla FP and Byszewski CH (1988) Aquatech
waters and reconcentrating brine for seawater. Con-
membrane technology for recovery of acid/base values
ventional ED is also widely used on a large industrial from salt stream. Desalination 68: 149.
scale in the dairy industry for demineralization of Mulder M (1991) Basic Principles of Membrane Techno-
whey. The technical feasibility of applying ED with logy. Dordrecht: Kluwer.
BPMs to a variety of commercially interesting pro- Sandeaux J, Sandeaux R, Gavach C et al. (1998) Extraction
cesses has been demonstrated. ED techniques are very of amino acids from protein hydrolysates by elec-
promising because they can be applied to environ- trodialysis. Journal of Chemical Technology and Bio-
mental protection (depollution and recycling of chem- technology 71: 267.
icals) and to bioindustries (food, pharmaceuticals and Scott K (1995) Handbook of Industrial Membranes, 1st
biotechnology). For all these kinds of applications, edn. Oxford: Elsevier.
special membranes have been elaborated showing Sistat P, Huguet P, Resbeut S et al. (1999) Polymeric
ion-exchange membranes: material, characterization,
adapted selectivity in ion transport under an electric
transport analysis, applications. In Recent Research
driving force. For improvement of most of the other Developments in Electroanalytical Chemistry. India:
applications of ion exchange membranes, research is Transworld Research Network.
mainly focused on the membrane processes themsel- Strathmann H (1996) Electromembrane Processes, Mem-
ves rather than on the synthesis of new membranes. brane and Science Technology Series. Oxford: Elsevier.
ENVIRONMENTAL APPLICATIONS
between rising bubbles and matter suspended in the

Flotation liquid, followed by adhesion of the particle to the
bubble surface and separation of the resulting
bubble}particle aggregate from that liquid. Environ-
M. A. Burstein, NPACI Edcenter on Computational
Science and Engineering, San Diego State University,
mental application of this technology includes the
San Diego, CA, USA selective separation of speciRc solids or liquids from
I. M. Flint, CPTI, Vancouver, BC, Canada solid suspensions, liquids from liquid suspensions or
certain dissolved species from solutions.
Copyright ^ 2000 Academic Press Separations which are based essentially on the ad-
sorption from solution include colloidal suspensions
of solid or liquid products, and the separation of
certain dissolved substances and ions. In order to
Introduction make a separation, there must be a drop in free energy
Flotation is the selective separation of solid particles, for the removed substance when it attaches to the
liquid droplets, chemicals or ions, or biological enti- bubble. For selective separations this drop must
ties from a bulk liquid, based on their surface proper- signiRcantly exceed those of the competing ions or
ties. In the process two actions occur: the collision colloids. In general, this results when the substance to
2676 III / ENVIRONMENTAL APPLICATIONS / Flotation
be removed is hydrophobic (if it is separate phase) or tions, and also at machinery plants where oil testing
surface-active (if it is dissolved) and the substances to of production is used. On offshore platforms, natural
remain with the carrying liquid are hydrophilic. gas is supplied instead of air to Soat hydrocarbon
There are many variations of Sotation vessel that droplets and remove them to an oil pad.
attempt to perform these actions, including induced Typically, for oil}water separation, a froth layer is
air machines like Sotation columns, agitated tanks not formed and oil is removed in a form of oil pad
and turbulent contact vessels; dissolved air Sotation continuously or in batch mode by rising liquid level in
units; or electroSotation units where electrolytic bub- the vessel.
bles are the carrier. The principal factor which in-
Suences the design is the ability of the unit to generate
bubbles of a size that will maximize the likelihood of Rendering By-Product Recovery
attachment of the dispersed particles and removal of In the food industry, the processing of Rsh, fowl, wool
bubble}particle aggregate from suspension. or slaughterhouses produces a stream of liquid that
carries animal oils and/or suspended solids. These
Hydrocarbon Removal from Water streams have a high oxygen demand and often cause
odour problems. The organic contaminants can be
Flotation itself is not restricted by the concentration removed by Sotation. This can be done by Rrst screen-
of the input contaminates, but the equipment used ing and/or settling the process stream to remove lar-
may be limited in its fractional removal and the purity ger products. The remainder is Soated in a quiescent
of its products. Liquid hydrocarbons are generally vessel to remove the larger products, followed by
highly hydrophobic and form very stable bubble more intense Sotation contactors to remove remnant
aggregates. However, low density differences be- oils. A properly designed circuit should recover all
tween the hydrocarbon and water mean that the but approximately 10 p.p.m. of the organic wastes.
droplet velocity relative to carrying water Sow is low.
As the relative velocity between the rising bubble and
the settling droplet in a still suspension is low, Sota- Reprocessing of Existing Mineral
tion columns, with their quiescent collection environ- Waste Dumps
ment are generally used to separate larger droplets,
whereas mechanical cells can operate on smaller sizes Flotation is used in the processing of secondary ma-
and high intensity contact devices on yet smaller sized terials in mining industry. As high grade mineral
droplets. However, the high turbulence devices pro- deposits are exhausted, reprocessing of old tailings
duce a large population of very small bubbles which, dams and ponds, stockpiles of low grade and oxidized
because of their low rise velocity, cannot easily be ores as well as metallurgical slags becomes economi-
separated from the transport water. Depending on cally feasible with technical improvements in process-
the droplet size distribution, with conventional units ing. It will also extend the lifetime cycle of mines and
the residual hydrocarbon concentration in the water concentrators. Although this reprocessing of waste
will be limited to purity levels between 10 and dumps and tailings dams will in turn produce new
30 p.p.m. dumps and dams, they will be of a lower metal con-
Recent designs speciRcally engineered to overcome tent so that acid damage will be reduced. In many
the small bubble problem have produced aqueous cases, the grade of waste material from 50}100-year-
underSow with less than 6 p.p.m. hydrocarbon con- old mines is higher than that of ores mined today, for
tent. This improvement trend is expected to continue example, copper content in old tailings dams is often
as the principles behind Rne bubble separation are over 0.8%, whereas its typical content in run-of-mine
better understood. porphiry ores is 0.4}0.6%. Reprocessing of mine tail-
Some products that can be separated from water ings usually does not require substantial expense be-
using this technology are diesel, motor oils and other cause the particle size has already been reduced to the
automotive products, crude oil and tar sands prod- point that different mineral crystals are liberated
ucts, creosote, polyaromatic hydrocarbon and poly- from each other.
aromatic phenol groups, chlorinated hydrocarbons,
plant and animal oils, waxes, and many paints and Site Run-off Treatment and
organic solvents.
Usually, Sotation is used after settling tanks for
Soil Decontamination
oil}water separation and prior to granular media Many sites, such as pole treatment yards, truck and
Rlters. Typical applications are offshore platforms, bus-washing facilities, and others which are still in
oil reRneries, large garages and vehicle service sta- use may be producing surface run-off or underground
III / ENVIRONMENTAL APPLICATIONS / Flotation 2677
plumes of contaminants. Most of these steams con- vessel. Similarly, microbubble dispersions can be used
tain contaminants that are Soatable. Also, Sotation is to Soat colloidal solids, although it is usual to add
becoming widely used for soil treatment at industrial coagulants to increase the size of the colloidal aggreg-
and military sites, where soil contains substantial ates.
concentrations of oil or other chemical poisonous Flotation systems can operate under externally sup-
products. plied electrical potentials which, by altering the sur-
Run-off water is collected by ditching and the con- face charge of the particles in the Reld, will optimize
taminated feed is pumped to a processing facility. The bubble}particle attachment. As bacteria and other
Rrst stage of separation is usually a gravity mixer microorganisms such as microalgae, are normally
settler, in which the resident liquid contains about hydrophobic, they potentially can be removed from
50% of organic matter, which effectively coalesces water by microbubble Sotation.
the Rner organic droplets. The residence time of the Flotation is used as the main method for de-inking
suspension in the settling unit has to be sufRcient to of recycled paper. The aim of the process is to remove
ensure that the coalesced droplets report to the or- only the ink from wastepaper Rbres suspended in
ganic-rich product stream. The required residence a slurry. Under current practice using conventional
time is dependent on the hydrocarbons to be re- mineral-processing mechanical cells, it is not possible
moved. The organic-rich stream is bled off and may to get a Rbre-free ink-rich overSow product. The
be burned or shipped off site after passing through disposal of this as landRll is both expensive and envir-
a further aqueous coalescing device. The aqueous onmentally objectionable. For these reasons, and be-
stream, or underSow of the gravity separator, then cause the partially de-linked product only has a lim-
passes through a hydrocyclone (high capacity streams ited use in low quality products, there is a need for the
only) followed by a Sotation cell. The organic prod- replacement of the existing cells by units speciRcally
ucts of both the hydrocyclone and Sotation device are designed for this application.
recycled to the mixer settler for further processing.
The aqueous stream, or underSow of the Sotation
device, is the Rnal product to be returned to the Conclusions
environment. The number of environmental applications of the
The system for processing plume water is the same Sotation process has increased dramatically over the
as run-off water, with the exception that a series of last 10 years. This trend will continue in the foresee-
wells must be made to lower the water table locally, able future, as increased environmental concerns are
thus preventing water from leaving the contaminated manifested with respect to the treatment of liquid
site. This well water is then processed with the site efSuents and solid waste materials.
run-off water. The treatment of these efSuents will lead to the
Flotation systems can have very high capacities and development of unconventional devices and methods
produce a high purity Rnal aqueous stream. However, for Sotation of ultra-Rne particles, ions and aggres-
they are more expensive than gravity units at low sive media.
throughputs. The relatively low capital and operational costs of
Sotation make it attractive for industrial use as an
Possible Ef]uent Treatments integral part of the Sow sheet.
Ion Sotation is widely used to extract ions from

See also: I/Flotation. II/Flotation: Cyclones for Oil/
aqueous solutions. Bubbles can be stabilized by sur- Water Separations; Historical Development; Oil and water
factants of various types. The bubble surface charge Separation. III/De Inking of Waste Paper: Flotation.
can thus be tailored to affect the preferential removal
of a speciRc ion or ionic complex. Normally, some
reagent (collector) is added to improve ion sorption at Further Reading
the gas}liquid interface. As ion Sotation is a mass
transfer process of dissolved substances, its rate is Clift R, Grace JR and Weber ME (1987) Bubbles: Drops
and Particles. New York: Academic Press.
proportional to a speciRc bubble surface area. There-
Finch JA and Dobby GS (1990) Column Flotation. New
fore, minimizing the average bubble size is critically York: Pergamon.
important; dissolved air, electroSotation, cavitation, Levenspiel O (1972) Chemical Reaction Engineering, 2nd
or high turbulence are used to generate microbubbles. edn. New York: Wiley.
It includes saturation of feed stream with dissolved Lynch AJ, Johnson NW, Manlapig EV and Thorne CG
air at high pressure, and then releasing the pressure to (1981) Mineral and Coal Flotation Circuits, Their Simu-
atmospheric and discharging into a clariRer-type lation and Control. New York: Elsevier.
2678 III / ENVIRONMENTAL APPLICATIONS / Gas Chromatography^Mass Spectrometry
Pinfold TA (1972) Chapter 4: Ion Sotation. Chapter 5: Seba F (1962) Ion Flotation. New York: American
Precipitate Sotation. In: Lemlich R and Arod J (eds) Elsevier.
Adsorptive Bubble Separation Techniques, pp. 53}90. Zhou ZA, Xu Z and Finch JA (1994) On the role
New York: Academic Press. of cavitation in particle collection during Sotation }
Rubinstein JB (1995) Column Flotation, Processes, Designs a critical review. Mineral Engineering 7:
and Practices. Basel, Switzerland: Gordon and Breach. 1073}1084.
Gas Chromatography^Mass Spectrometry

N. Scott and G. Gutnikov, California State brought the technique within the reach of most labor-
Polytechnic University, Pomona, CA, USA atories. Computer-assisted operation of these instru-
ments, including automated sample injection, data
acquisition, online searches of mass spectral libraries
Introduction and quantiRcation by selected ion monitoring (SIM)
has led to rapid, efRcient and convenient analytical
Increasing public concern over environmental pollu- systems. These advances have made compact, rugged,
tion reSects the heightened awareness of the toxicity portable GC-MS instruments available for Reld ap-
of a large number of chemicals that Rnd extensive plications, thereby rendering GC-MS a vital tech-
application in wide-ranging Relds. The consequent nique in environmental analysis.
impetus to reduce chemical contamination of the en-
vironment has generated a growing body of legislation
and mechanisms for enforcement via speciRc regula-
tory agencies. For example, following the discovery
Sample Pre-Treatment
of trihalomethanes in chlorinated drinking water in The low concentrations of pollutants and the com-
the 1970s the Safe Drinking Water Act (SDWA) plex matrices in which they frequently occur gener-
was passed that directed the United States Envi- ally preclude direct injection of environmental sam-
ronmental Protection Agency (US EPA) to undertake ples into a chromatograph. Sample pretreatment is
a comprehensive study of the contaminants present then necessary to remove components that would
in drinking water. The pre-eminent technique em- interfere in the analysis and to concentrate target
ployed in these investigations was gas chromatogra- analytes that are present at extremely low levels.
phy with mass spectrometric detection (GC-MS). The Pretreatment varies considerably with the nature of
distinctiveness of the mass spectra of the target the sample and the information sought. Pollutants are
analytes and their volatility were two vital charac- dispersed in air, water and soil.
teristics required for GC-MS. As the studies initiated Atmospheric samples may contain numerous
by the SWDA yielded results, GC-MS has been species ranging from gaseous to nonvolatile sub-
recommended for analysis of many environmental stances adsorbed on particles. They are usually col-
contaminants. lected in highly polished SUMMA canisters and
Since many pollutants occurring at trace levels in Tedlar bags or on sorbents or in an impinger
complex matrices are either volatile or amenable to solution. SpeciRcally, the EPA Compendium Method
derivatization to volatile products, GC-MS has TO-14 mandates SUMMA passivated canister samp-
proved to be an effective means for veriRcation of ling. Particles are commonly retained on membrane
compliance with environmental regulations. GC and Rlters or impactors. Gaseous components and vol-
MS play complementary roles in the analysis of mix- atile organic compounds (VOCs) are collected by
tures. Volatile constituents of complex mixtures may trapping, either cryogenically for the most volatile
be conveniently separated by GC but not identiRed or on sorbents of increasing retentivity for the less
unambiguously with conventional detectors. MS volatile.
provides much more deRnitive structural informa- The constituents of particulates are separated into
tion that permits identiRcation. The sample sizes approximately organic and inorganic substances by
necessary for GC and MS are also comparable and Soxhlet extraction or ultrasonication with an appro-
sample volatility is necessary for both. priate organic solvent (methanol, dichloromethane,
The utility of GC-MS for environmental analysis cyclohexane, etc.) or by supercritical Suid extraction
has been further enhanced by the development of (SFE) with CO2. SFE is becoming increasingly accep-
relatively inexpensive table-top instruments that has ted for the extraction of analytes because it is
III / ENVIRONMENTAL APPLICATIONS / Gas Chromatography^Mass Spectrometry 2679
more rapid, reliable and efRcient for a variety of a weak one to adsorb the heavier organics, followed
matrices than the older techniques and is environ- by strong one to retain all the remaining. The analytes
mentally friendly in reducing the need for organic are desorbed by Sash-heating from weak adsorbents
solvents. (primarily graphitized carbon black and Tenax); by
The soluble fraction comprises a complex mixture solvent extraction from strong ones (activated char-
of various organic compounds that may require fur- coals), using carbon disulRde, dichloromethane, etc.
ther fractionation prior to GC-MS analysis. This has (Figure 1).
entailed separation into fractions of different polar- For aqueous samples preconcentration and clean-
ities by such techniques as liquid}liquid extraction or up are often achieved concurrently. In the purge-
column chromatography following preliminary re- and-trap (dynamic headspace sampling) tech-
moval of the acidic and basic components. Solid- nique, the volatile species that are present in water are
phase extraction (SPE) has become a popular alterna- swept out of solution by a stream of nitrogen or
tive because of its simplicity, speed, improved sample helium into a sorbent trap consisting of activated
clean-up, easy automation and the capacity to handle carbon, Tenax or silica gel. The analytes are desorbed
multiple samples simultaneously. More recently Rlter thermally and transferred along with the carrier
discs have been introduced to perform the same func- gas to the GC-MS instrument. If the number of
tions more conveniently. volatile compounds is few, analysis of the head-
VOCs are generally preconcentrated on adsorbent space gases is possible directly but with the drawback
traps. Different traps are often necessary as no single that water vapour is likely to be injected into the
sorbent performs satisfactorily over the entire range instrument.
of organic volatilities and polarities encountered. Semi-volatile species present in water must be Rrst
Two (or more) sorbents are used in tandem, Rrst extracted into a suitable organic solvent or retained
Figure 1 Schematic diagram for thermal desorption}gas chromatography}mass spectrometry system. (Reproduced from Ma
C (1997) Performance evaluation of a thermal desorption/gas chromatographic/mass spectrometric method for the characterization of
waste tank headspace samples. Environmental Science and Technology 31(3): 853}859, with permission from the American Chemical
Society.)
on an SPE cartridge and then eluted for injection. The and Technology and John Wiley & Sons) are compila-
polarity of the organic stationary phase, which is tions of EI-mass spectra. However, negative chemical
bonded to a silica support, determines the retention ionization mass spectrometry (NCIMS) is used some-
selectivity. times, e.g. for the analysis of positional isomers of
For solid samples, VOCs are Rrst extracted into nitropolycyclic aromatic hydrocarbons via high res-
methanol, which is then added to water for purge- olution mass spectrometry (HRMS). Organophos-
and-trap. For extraction of semivolatiles from solid phates can be analysed by either EI or CI.
samples SCF is particularly advantageous. The ex- QuantiRcation of very low concentrations necessi-
tracted analytes collect in an appropriate solvent on tates operating in the SIM mode. SIM involves ac-
reducing the pressure to atmospheric level. quiring ion currents only at a few speciRed m/z values
Sample pretreatment is beset with a serious prob- that are characteristic of the analyte(s). Data acquisi-
lem } the potential loss of analytes } which necessi- tion in this mode avoids sampling the barren regions
tates monitoring solute recovery by employing of a full scan spectrum, thereby improving the ion
internal standards with similar properties. Also accu- counting and the sensitivity of detection. Internal
mulation of water during preconcentration can affect standards are usually isotopically labelled com-
the GC-MS instrument, especially the pumps neces- pounds that may be purchased individually or as
sary to maintain the vacuum. A remedy that has been mixtures from agencies such as the National Institute
attempted is the use of NaRon diffusion dryers, but of Science and Technology (NIST) or the EPA.
this may cause the loss of small, polar organic com- Despite enhanced sensitivity, SIM limits the detect-
pounds. able analytes to those producing ions of the speciRed
m/z values. Hence, for identiRcation, scanning the
entire mass spectrum of each compound may be ne-
Instrumentation cessary. But this compromises the detection sensitiv-
Capillary columns (0.32 mm i.d.) of appropriate po- ity. Therefore one strategy is to use GC with a multi-
larity are routinely used in view of their greater efR- detector system (GC/MD) in conjunction with GC-
ciency. However, for relatively high concentrations MS. The multidetector system enables sensitive quan-
megabore capillary columns (0.53 mm i.d.) are pre- tiRcation while the mass spectrometer operated in the
ferred with connection to the mass spectrometer ion full-scan mode enables identiRcation of some of the
source being achieved via a glass jet separator for unknown pollutants. The primary quantiRcation sys-
whose optimum functioning a stream of make-up gas tem is thus the GC/MD system that comprises con-
is provided. This approach has been widely used for ventional GC detectors such as FID, PID (photoioniz-
the analysis of ambient air samples; waste and solid ation detector) and ECD (electron-capture detector).
samples; purge-and-trap analysis; as well as thermal
desorption analysis of sorbent tubes.
The mass spectrometer of a GC-MS system must be
Applications
capable of rapid response to monitor the solutes that Environmental analysis serves different objectives,
are eluted from the chromatographic column in quick including routine monitoring, quality assurance,
succession. Hence ion trap and quadrupole instru- litigation and research. The list of target pollutants
ments are the most widely used mass analysers. The and the desired detection limits undergo continual
compact ion trap detector (ITD) also enhances the revision in light of toxicological research. For the
portability of GC-MS instruments for Reld analysis. determination of a particular pollutant at the desired
Calibration of a mass spectrometer is usually based level of sensitivity, the same analytical technique is
on the diagnostic ions of either perSuorotributyl- not adopted by all environmental agencies. GC-MS is
amine (PFTBA) or perSuorokerosene (PFK). How- sometimes mandated, sometimes suggested as an al-
ever the US EPA has introduced m/z abundance cali- ternative method. Hence some examples that are pre-
bration for checking the performance of the entire sented in this article may not (yet) be approved by
GC-MS system, not merely of the mass spectrometer. regulatory agencies but are nevertheless acceptable to
Two compounds mandated for this purpose are bro- environmental chemists as representing the wide ap-
moSuorobenzene (BFB) for volatile analytes and plicability of GC-MS in this Reld.
decaSuorotriphenylphosphine (DFTPP) for semi- The US EPA classiRes organic pollutants into two
volatiles. classes: volatiles (compounds that exist as gases at
Electron impact (EI) ionization is the most widely room temperature and are easily removed from the
used mode of ionization mainly because the databases sample matrix); and semivolatiles (compounds
available for mass spectral library searches (such as that can also exist as gases but need some form of
those provided by the National Institute of Science extraction to be removed from the matrix). Both
classes of compounds occur in outdoor, indoor and In order to avoid the cumbersome process of deter-
workplace atmospheres as well as in soil and water, mining individual compounds, more convenient
and their concentrations may range from 100 ppm in methods have been accepted as regulatory bench-
the vicinity of emission points to a few parts per marks. These include the GC-MS or GC-FID analysis
trillion (ppt) in pristine environments. The compo- of total petroleum hydrocarbons (TPH) or mixtures
nents of a particular mixture are likely to occur at of benzene, toluene, ethylbenzene and xylene iso-
different concentrations, with the more toxic target mers (BTEX). In the TPH method the total area of
compounds being at times at much lower levels than unresolved and resolved chromatographic peaks is
the less hazardous ones. measured.
VOCs for which US EPA approved GC-MS methods Interest in monitoring the indoor environment has
of analysis exist cover a broad range of compounds, been prompted by the well-known ‘sick building syn-
including highly volatile organics (carbon tetrachlor- drome’. The environment indoors can be worse than
ide, chloroform, acrylonitrile, allyl chloride, etc.); that outdoors because of the accumulation of volatile
semivolatile organics (benzene, nitrobenzene, chloro- pollutants owing to poor air circulation. These pollu-
benzenes, toluene, trichloroethane, etc.); N-nit- tants not only enter from the air outdoors but are also
rosodimethylamines; polychlorinated dibenzo-p-di- released from building materials and furnishings,
oxins (PCDDs) and polychlorinated biphenyls cleaners, air fresheners, gas-burning stoves, etc. GC-
(PCBs); and a variety of miscellaneous compounds MS has been invaluable for their analysis following
including chlorinated compounds and aromatics. preconcentraton on Tenax and thermal desorption.
A major fraction of VOCs are hydrocarbons, espe- Separation has been achieved on a bonded SE-54
cially by roadsides where trafRc is heavy. For capillary column with temperature programming
example, in West Los Angeles, petroleum residues are (Figure 3).
the main solvent-soluble organic fraction of carbon- Polycyclic aromatic hydrocarbons (PAHs) are ubi-
aceous aerosols. For analysis, these particles are quitous environmental contaminants with carcino-
collected on quartz Rlters and extracted by ultrasoni- genic properties and present an analytical challenge
cation Rrst with hexane and then with benzene/ because of the complexity of the mixtures in which
isopropanol. The extracted constituents are deter- they exist as many different isomers of the parent
mined by high resolution gas chromatography-MS. compounds. Hence a separation step is indispensable
A bonded OV-1701 (86% dimethyl/14% cyano- but volatility requirements restrict the applicability of
propylphenyl polysiloxane) column and a quadrupole GC to compounds of low and moderate molecular
mass spectrometer operated in the EI mode have been mass, the heavier ones being analysed by HPLC with
employed in such studies. Suorescence detection. PAHs are isolated from air-
Hydrocarbon contamination of ground water and borne samples by a combination of a TeSon Rlter and
soil often occurs through leaking fuel tanks. In the polyurethane foam sampling; from solid matrices
California Leaking Underground Fuel Tank (LUFT) either by extraction with SCF employing carbon di-
method GC-MS is preferred to GC-FID for monitor- oxide, or by microwave-assisted extraction (MAE)
ing such leakage without interference by other into an organic solvent. For concentration and
semivolatile species. The most abundant ions in the sample clean-up SPE with combined C18 amino (NH2)
mass spectra result from the fragmentation of C10}C23 or cyano (CN) solid phases can be employed with
n-alkanes and occur at m/z values of 43 and 57. They subsequent elution being carried out with CH2Cl2.
correspond to C3H# #
7 and C4H9 , respectively, and are Deuterated PAHs are available from US EPA to serve
regarded as qualiRer and target ions for monitoring as internal standards. More than 60 compounds en-
purposes. compassing parent PAHs, their alkylated derivatives
GC-MS is an invaluable technique for chemical and heterocyclic analogues have been analysed at
Rngerprinting of crude oil spills in land or marine 5}100 ppb levels with a DB-5 capillary column em-
environments. Initial screening by Suorescence spec- ploying both the scanning and SIM modes. Of even
troscopy and GC is followed by GC-MS identiRcation greater toxicological interest are the nitro-PAHs that
of speciRc compounds such as steranes, triterpanes, have been ranked by the International Agency for
phytanes, pristanes, etc., that are most resistant to Research on Cancer (IARC) as probable human car-
weathering. Similar analysis of tarballs, which are cinogens. Their analysis is even more difRcult because
formed by highly weathered oils, may enable identi- of the much lower concentrations and the presence
Rcation of the petroleum source (Figure 2). When of many positional isomers. Moreover, unlike the
considerably more components are monitored, multi- parent PAHs, nitro-PAHs are not Suorescent, thereby
variate and pattern recognition statistical analysis limiting the sensitivity attainable by HPLC.
methods become necessary for data interpretation. Hence GC-NCI-MS is the technique of choice for the
Figure 2 GC-MS n-alkane distribution patterns (m/z 85) for tarball samples BC-1, BC-2, CA-1 and ANS reference oil. (Reproduced
from Wang Z et al., 1998, with permission from Wiley}VCH.)
determination of nitro-PAHs. Polar or moderately Chlorinated herbicides and pesticides are routinely
polar capillary columns have been utilized for their analysed by capillary GC with MS or ECD. In view of
separation with d9-1-nitopyrene or d5-dinitropyrene the volatility of organophosphates GC is the tech-
serving as internal standards for quantiRcation. nique of choice for analysis via phosphorus-speciRc
Organochloro and organophosphate pesticides detectors. However, MS detectors operated in the
constitute another major category of toxic pollutants. SIM mode are gaining popularity, especially with
Figure 3 Indoor air at a Swedish preschool. (Reproduced from Subramanian, 1995, with permission from Wiley}VCH.)
electron impact (EI) or chemical (CI) ionization for polar column such as DB-5 enables determination of
detection at low ppb. For characterization, CI is pre- homologous groups of the dioxins but a polar column
ferred because when CH2Cl2 is used as reagent gas the such as SP-2331 is necessary for separation of most of
(M#Cl)\ ion is formed that gives a peak at a higher the 2,3,7,8-congeners. EI is the usual method of ioniz-
m/z value, thereby enhancing selectivity. ation and 13C-2,3,7,8-TCDD is widely used as an
Insecticidal carbamates readily undergo thermal internal standard in quantiRcation. A mass spectrom-
degradation and hence must be derivatized with hep- eter resolving power of even 10 000 is insufRcient to
taSuorobutyric anhydride (HFBA) prior to GC-MS separate a TCDD (m/z"321.8936) that is coeluted
analysis. Ethylenethiourea, an environmental metab- with heptachlorobiphenyl (m/z"321.8678). This
olite of carbamate fungicides and an accelerator used difRculty underscores the importance of high resolu-
in synthetic rubber production, has been classiRed by tion GC-high resolution MS. Despite this difRculty,
the IARC as a potential human carcinogen. Its high by the mid-1980s GC-MS analytical methods had
polarity and water solubility require derivatization to been developed for the separation of all TCDD iso-
a volatile compound and extraction into an organic mers and quantiRcation at the 10\15 level (Figure 4).
solvent. This has been achieved by conversion to an PCBs and polychloroterphenyls, which number
S-alkyl derivative with m-triSuoromethylbenzyl over 200, are stable compounds that can cause seri-
chloride or 3,5-bis(triSuoromethyl)benzyl bromide. ous toxicity through bioaccumulation. PCBs have
GC-NCI-MS on a D-1701 capillary column and the been used in transformers, hydraulic Suids, etc. The
SIM mode has permitted sub-ppb detection limits to method of choice for their analysis is GC-MS. In the
be realized. EI mode, a PCB molecule can lose an ortho- chlorine
Other organochloro compounds of great interest atom giving a (M!35)# ion, which facilitates dis-
include the highly toxic, and environmentally persist- tinguishing between two coeluted PCBs one of which
ent, polychloro-dibenzo-p-dioxins (PCDDs), -furans lacks an ortho-chlorine atom. NCI, with methane as
(PCDFs) and -biphenyls (PCBs), which are generated reagent gas, has improved detection sensitivities 10 to
mainly during various combustion and manufactur- 100-fold not only for PCBs, but also for other or-
ing processes. They usually occur as very complex ganochlorine compounds such as toxaphene, chlor-
mixtures of a large number of compounds. For dane, etc. In NCI, the base peak is usually either
example, there are 22 isomers of tetrachlorodibenzo- M\ or (M } H)\.
p-dioxin (TCDD) alone, the 2,3,7,8-isomer being the Large volume (100 L) on-column injection capil-
most toxic. The very similar masses of these com- lary GC-MS has been applied to the determination of
pounds demand high mass spectrometric resolving aliphatic and aromatic organochlorine compounds
powers for peak separation. Hence much attention present in process water at low ppt levels. In the
has been focused on simplifying the composition of extraction of these solutes into hexane, the tedium of
the mixtures by appropriate sample extraction and manual operation has been obviated by an automated
clean-up prior to GC-MS. They are extracted from procedure involving a laboratory robot. This has led
solids such as soil, sediments, dust, etc., into an or- to signiRcant savings in extracting solvent, sample
ganic solvent and the extracts are cleaned up via size, labour and time of analysis.
column chromatography on silica gel or alumina. In recent years concern over the possible use of
Analysis is invariably carried out by GC-MS. A nonchemical warfare agents has generated interest in
their sensitive and reliable determination. The most

common include 1,1-thiobis(2-chloroethane) (sulfur-
mustard gas); O-ethyl-N,N-dimethylphosphoroami-
docyanidate (tabun); isopropylmethylphosphonoSu-
oridate (sarin) and pinacolyl methylphosphonoSuori-
date (soman). Their degradation products include
(C) 50 m OV-101 HRGC columns. (Reproduced from Buser HR (1980) High resolution gas chromatography of the 22 tetrachlorodibenzo-p-dioxin isomers.
Figure 4 Mass fragmentogram (m/z 320) of a composite sampling showing elution of all 22 TCDD isomers on (A) 55 m Silar 10C, (B) 50 m OV-17, and
methylphosphoric acid (MPA); iP MPA; pinacolyl
MPA; cyclohexyl MPA and dithioglycol (DTG). GC-
MS is advantageous for their analysis since their mass
spectra are well known and available for comparison.
From samples such as clothing, soil, exhumed skel-
etons, etc., the parent compounds are extracted into
CH2Cl2; the degradation products are extracted into
water and converted to t-butyldimethylsilyl (TBDMS)
derivatives. After preliminary screening by low res-
olution EI GC-MS, conRrmatory evidence is obtained
by one of two methods: (1) GC-MS with EI or CI with
NH3 as reagent gas or (2) GC-tandem mass spectro-
metry (GC-MS/MS), where a precursor ion further
fragments in a collisional deactivation chamber. For
conRrming the presence of sarin at low ppb to ppm
level, GC-MS/MS using both EI and NH3 CI has been
carried out, employing two columns of different po-
Analytical Chemistry 52: 2257}2262, with permission from the American Chemical Society.)
larity (Figure 5).

GC-MS has been extended to the analysis of highly
polar compounds such as aliphatic glycols that are
used as antifreeze in automobiles and in aircraft de-
icing. Interest in their determination stems from the
Figure 5 Reconstructed ion chromatograms for (A) nerve

agents GA, GB, GD, GF and sulfur mustard (SM) (1 ng injected)
and (B) TBDMS derivatives of: 1, ethyl MPA; 2, iPMPA; 3,
pinacolyl MPA; 4, cyclohexyl MPA; 5, 5-cyclohexyl MPA; and
6,TDG (500 pg injected). (Reproduced from Black et al., 1994,
with permission from Elsevier Science.)
acute toxicity of their metabolic products. Glycols A knowledge of the total content of an element is of
form volatile n-butylboronate derivatives with the limited toxicological value since the toxicity is species
characteristic 10B : 11B isotopic ratio (1 : 4) assisting in dependent. Organometallic compounds of lead and
identiRcation. Employing a DB-1 column and an ion tin have been reduced to their hydrides and sorbed on
trap detector operated in the SIM mode, sub-ppm the GC column of a GC-ICP-MS unit and detection
detection limits have been attained. Haloacids that limits of 0.3}2 ng mL\1 for tin have been attained.
are formed during disinfection of drinking water con- Developing a convenient method for the analysis of
stitute another group of polar compounds that have mixtures of arsenic(III), arsenic(V), monomethylar-
been analysed by GC-MS. These acids are readily sonic and dimethylarsinic acids continues to be a for-
derivatized to their methyl esters with diazomethane. midable analytical challenge. In a recent low resolu-
GC-MS interfaced to inductively coupled plasma} tion EI GC-MS method, which has employed hexa-
mass spectrometry (ICP-MS) enables speciation of chlorobenzene as internal standard, ppb detection
elements at trace levels and has been applied, in limits have been attained following derivatization to
conjunction with isotope dilution, to the analysis of methyl thioglycolates. However, arsenic(III) and ar-
some environmentally signiRcant inorganic species. senic(V) have been left unresolved (Figure 6).
Figure 6 Representative chromatograms of HCB and the TGM derivatives of DMAA, MMAA and As(III). (A) Calibration standard
containing 10 ppb DMAA and MMAA and 20 ppb As(III). (B) River water spiked with 5 ppb DMAA and MMAA and 40 ppb total inorganic
arsenic (As(III) and As(V)). SIM program: m/z 195 from 3.0 to 5.0 min, m/z 282 from 5.0 to 6.1 min, m/z 195 from 6.1 to 8.0 min, and m/z
285 from 8.0 to 9.3 min. (Reproduced from Claussen, 1997, with permission from Elsevier Science.)
Conclusion entire system being under computer control. A para-

mount consideration is the ability of the system to
Future developments of GC-MS in environmental withstand shocks and vibrations during transport to
analysis will encompass both sampling and instru- and from the site(s). The greater speed, sensitivity and
mentation. Environmentally friendly extraction resolution needed for analysis of toxic pollutants pro-
methods such as SFE and MAE that reduce or elimin- duced in fast processes such as Rres are continually
ate the need for organic solvents are likely to gain the focus of developments in HRGC/HRMS. For fas-
greater acceptance. A striking development in this ter analysis, short (1 m) capillary columns called
trend is the solventless extraction technique of solid- ‘transfer lines’ and supersonic molecular beams for
phase microextraction (SPME), which concentrates sampling and ionization have been employed. The
the analytes on a stationary phase that is bonded or low sample capacity associated with very short col-
coated onto a fused silica Rbre. Subsequent thermal umns has been overcome via multicapillary columns.
desorption introduces the analytes into the injection For monitoring solutes that are rapidly eluted from
port of the GC-MS unit. This method has been ap- columns, mass spectrometer scan speeds must be in-
plied to the analysis of volatile organic solvents, or- creased signiRcantly without compromising resolu-
ganochlorine and organophosphate compounds and tion. Alternatively, all masses must be scanned at the
sub-ppb to ppt detection levels have been attained same time as is done in Fourier transform-MS
with ion trap mass spectrometry. (FTMS) and ITD. However, both these techniques
There is tremendous interest in developing minia- operate in the pulse mode, making sample utilization
turized, Reld-portable GC-MS units to facilitate on- inefRcient. Hence other modes of array detection
site, real-time monitoring. For unattended operation have been attempted, an interesting example being the
and reduced labour, the sampling, extraction and development of an electro-optical ion detector (EOID)
injection steps should be carried out by a robot, the for microbore capillary column chromatography. In
Figure 7 Total ion chromatogram obtained from a mixture of compounds consisting of air (11), dichlorofluoromethane (16),
chloromethane (19), bromoethane (28), chloroethane (30), dichloromethane (59), 1,1,1-trichloroethane (126), chloroform (162),
benzene (188), and trichloroethylene (270). Each compound in the mixture has a concentration of 1 ppmv. A sample volume of 0.5 L
was injected, and a signal integration time of 250 ms was used for each frame. (Reproduced from Gutnikov G. (1991) Development of
a miniaturized gas chromatograph-mass spectrometer with a microbore column and an array detector. Analytical Chemistry 63(18):
2012}2016, with permission from the American Chemical Society.)
III / ENVIRONMENTAL APPLICATIONS / Pressurized Fluid Extraction 2687
EOID the electrons generated by a microchannel elec- to the analysis of chemical warfare samples, found to
tron multiplier array are converted to photons by contain residues of the nerve agent sarin, sulphur mus-
a phosphor screen and detected via an array of photo- tard and their degradation products. Journal of Chrom-
diodes (Figure 7). Very high sensitivites have been atography A 662: 301}321.
reported, with benzene being detected at Bruner F (1993) Gas Chromatographic Environmental
Analysis. New York: VCH Publishers.
7.5;10\14 g levels.
Claussen FA (1997) Arsenic speciation of aqueous environ-
Multidimensional gas chromatography (MDGC) is mental samples by derivatization with thioglycolic acid
another technique that seeks to overcome the prob- methyl ester and capillary gas}liquid chromatography}
lems involved in the analysis of multicomponent mix- mass spectrometry. Journal of Chromatographic Science
tures of volatile constituents. It is based on the separ- 35: 568}572.
ation of the constituents via one column and isolating Jones FE (1994) Toxic Organic Vapors in the Workplace.
the coeluted species in one or more cryotraps to be Boca Raton, FL: Lewis Publishers.
separated by a second column of different selectivity. Karasek FW, Hutzinger O and Safe S (eds.) (1985) Mass
MDGC is usually coupled with detectors such as IR Spectrometry in the Environmental Sciences. New York:
and MS for increased speciRcity. Plenum Press.
GC-MS has emerged as one of the premier methods Keith LH (ed.) (1992) Compilation of E.P.A.’s Sampling
and Analysis Methods. Grand Rapids, MI: Lewis Pub-
for the rapid, convenient and sensitive analysis of
lishers.
pollutants. This development is partly a result of the Sinha MP and Gutnikov G (1991) Development of a Minia-
public concern and legislative pressure for reliable turized Gas Chromatograph}Mass Spectrometer with
monitoring of environmental quality. The commer- a Microbore Capillary Column and an Array Detector.
cial availability of relatively inexpensive, computer- Analytical Chemistry 63, 2012}2016.
interfaced, bench-top instruments has rendered the Subramanian G (ed.) (1995) Quality Assurance in Environ-
technique almost routine for cost-effective analysis. mental Monitoring, Instrumental Methods. Weinheim:
Even greater speeds and resolving powers of the fu- VCH Publishers.
ture generations of instruments will augment the sig- Tamilarasan R, Morabito PL, Lamparski L, Hazel-
niRcance of GC-MS in the environmental Reld. wood P and Butt A (1994) Determination of neutral
chlorinated extractable organic compounds in water
samples using large volume on-column injection
See also: II/Chromatography: Gas: Detectors: Mass
capillary gas chromatography}mass spectrometry.
Spectrometry. Extraction: Solid-Phase Extraction. III/
Journal of High Resolution Chromatography 17:
Fungicides: Gas Chromatography. Herbicides: Gas
689}694.
Chromatography. Pesticides: Gas Chromatography. Poly-
Wang Z, Fingas M, Landriault M et al. (1998) IdentiRca-
chlorinated Biphenyls: Gas Chromatography. Poly-
tion and linkage of tarballs from the coasts of Vancouver
cylic Aromatic Hydrocarbons: Gas Chromatography.
Island and Northern California using GC/MS and iso-
topic techniques. Journal of High Resolution Chrom-
Further Reading atography 21(7): 383}395.
Wilkins CL (1994) Multidimensional GC for qualitative IR
Berezkin VG and Drugov YS (1991) Gas Chromatography and MS of mixtures. Analytical Chemistry 66(5):
in Air Pollution Analysis. Amsterdam: Elsevier. 295A}301A.
Black RM, Clarke RJ, Read RW and Reid TJ (1994) Ap- Winegar ED and Keith LH (eds.) (1993) Sampling and
plication of gas chromatography}mass spectrometry Analysis of Airborne Pollutants. Boca Raton, FL: Lewis
and gas chromatography}tandem mass spectrometry Publishers.
Pressurized Fluid Extraction

S. R. Sumpter, DuPont Agricultural Products, also frequently extracted. Sludge, chimney brick, Sy
Wilmington, DE, USA ash and urban dust comprise other materials that are
Copyright ^ 2000 Academic Press extracted using PFE. There are several reasons to
use pressurized Suid extraction for environmental
Introduction samples. PFE methods are:
Pressurized Suid extraction (PFE) is used to extract E automated
a wide variety of compounds from solid and semi- E easy to transfer
solid environmental samples. Soil is the predominant E environmentally friendly, consuming little solvent
environmental solid extracted. Sediment samples are E relatively fast
2688 III / ENVIRONMENTAL APPLICATIONS / Pressurized Fluid Extraction
E simple, requiring little expertise Table 1 Compound classes of environmental importance

E acceptable to regulatory agencies. extracted using pressurized fluid extraction
Compound class
Several PFE methods have been developed to extract
compounds from environmental samples. These are Aliphatic hydrocarbons
listed in Table 1. These methods are summarized in Alkylphenols
the sections that follow. One promulgated method in BTEX
a
Chlorinated herbicides
particular has extensive applicability. This method is
Chlorinated hydrocarbons
United States Environmental Protection Agency Ethoxylates
(EPA) Method 3545. It is suitable for the extraction Explosives (HMX, RDX, TNT, DNT)
of several compound classes from environmental Gasoline
solid and semisolid samples. Compound classes listed Linear alkylbenzenesulfonates (LASs)
a
Organochlorine pesticides (OCPs)
in Table 1 that are extracted using EPA Method 3545 a
Organophosphorus pesticides (OPPs)
are noted. Phenoxyacid herbicides
The authors of EPA Method 3545 found the PFE a
Polychlorinated bipheyls (PCBs)
method gives recoveries comparable to those ob- Polychlorinated dibenzofurans (PCDFs)
tained by rigorous Soxhlet extraction and other tech- Polychlorinated dibenzo-p-dioxins (PCDDs)
a
Polycyclic aromatic hydrocarbons (PAHs)
niques, such as shaking, supercritical Suid and a
Semi-volatile base/neutral/acid (BNAs)
sonication extraction methods. Compared to these Acids
techniques, PFE takes less time ((30 min) and con- Alcohols
sumes less solvent (15}30 mL). Using PFE, the Amides
method is applicable to the extraction of poly- Aromatic amines
Aromatic chloroethers
chlorinated biphenyls (PCBs), semi-volatile base/
Azobenzenes
neutral/acids (BNAs), organophosphorous pesticides Benzidines
(OPPs) organochlorine pesticides (OCPs), polycyclic Chloroanalines
aromatic hydrocarbons (PAHs) and chlorinated Chlorobenzenes
herbicides from soil. Of all of the extraction condi- Chlorophenols
Methylphenols
tions, only the extraction solvent is changed when
Nitroanalines’
extracting different compound classes. For example, Nitrobenzenes
acetone}hexane is used to extract a sample when Nitrophenols
analysing for OCPs and PCBs, methylene chlor- Nitrosamines
ide}acetone is used to extract a sample when analys- Phenyl hydrazines
Phthalates
ing BNAs and OPPs, and acetone}methylene chloride
Toluidines
acidiRed with phosphoric acid is used to extract Total petroleum hydrocarbons (TPHs)
a sample when analysing for chlorinated herbicides.
Extraction conditions for Method 3545 are shown in a
Compounds extracted using U.S. EPA Method 3545.
Table 2.
Relative recoveries of the PFE method (Method
3545) compared to other methods such as shaking or undecane, tetradecane, and pentadecane are efR-
sonication extraction are summarized in Table 3. In ciently extracted using the conditions listed in
this work, relative recoveries are the quotient of PFE Table 4. Due to its volatility, pentane recoveries
recoveries divided by shaking, sonication, or Soxhlet are the lowest of these analytes. Pentane boils at
extraction recoveries. 363C, so special precautions are required to keep
The following sections summarize details of critical it from evaporating from the extraction cell before
sample preparation steps that are necessary before the extraction is performed. To prevent loss of pen-
PFE, as well as the conditions used to extract the tane, extraction cells are cooled and aluminium reten-
compounds listed in Table 1. tion discs are added into the tops of the extraction
cells.
Water in the form of super-heated steam may also
Applications be used to extract aliphatic hydrocarbons. Dodecane,
pentadecane, octadecane, heneicosane, tetracosane,
Aliphatic Hydrocarbons
heptacosane, triacontane, and tritriacontane are efR-
Aliphatic hydrocarbons are readily extracted by PFE ciently extracted from solid and semi-solid envi-
using aqueous and organic extraction solvents. Using ronmental samples using the conditions listed in
organic extraction solvent, pentane, nonane, decane, Table 5.
Table 2 Summary of sample preparation and extraction conditions used in EPA Method 3454 to extract PCBs, semi-volatile BNAs,
OPPs, OCPs and chlorinated herbicides from solid and semisolid environmental samples
Extraction conditions Extraction analysis conditions
Sample preparation Grind sample to 100}200 mesh (150}75 m particle size)

Dispersing or drying agent Mix sample with anhydrous sodium sulfate or diatomaceous earth
Sample size Up to 30 g if a dispersing agent is not used
Extraction cell volume 11, 22, or 33 mL cell, depending on sample size
Extraction solvent OCPs and PCBs: 1 : 1 acetone : hexane
BNAs and OPPs: 1 : 1 methylene chloride : acetone
Chlorinated herbicides: 2 : 1 acetone : methylene chloride acidified with phosphoric acid
Temperature 1003C
Heat step 5 min
Static time 5 min
Flow type Static, dynamic, or mix of both
Number of cycles 1}5
Extraction pressure 2000 psi
Flush volume 60% of cell volume
Purge time 60 s
Extraction conditions from USEPA SW-846, 3rd edn., Update III (July 1995) Test methods for evaluating solid waste, Method 3545,
Accelerated solvent extraction. U.S. GPO: Washington, DC.
Alkylphenols acetone, or dioxan may be used as the extraction

solvent. Since the solvent is typically evaporated be-
Alkylphenols are metabolites of alkylphenol ethoxyl-
fore analysis, the most volatile solvent should be
ates: a class of non-ionic surfactants that were used as
used.
cleaning agents and still Rnd use today. Alkylphenols
are monitored in the environment as they pose poten- BTEX (Benzene, Toluene, Ethylbenzene, and
tial risk to animals. Time, and the amount of organic Xylene)
solvent, are saved in the pressurized Suid extraction
of alkylphenols from sediment. For the extraction, BTEX (benzene, toluene, ethylbenzene, and xylene)
CO2 extraction solvent is modiRed with organic sol- mixtures are efRciently extracted from solid and
vent. Either methanol, ethanol, 1-butanol, 2-pro- semisolid environmental samples by PFE using or-
panol, acetone, or dioxane may be used. Sediment ganic and aqueous solvents. BTEX mixtures are ex-
spiked with heptylphenol and nonylphenol is extrac- tracted from sand spiked with BTEX by PFE using
ted with 100% recovery after 15 min extraction organic solvent and standard conditions described in
times. Longer times are required to extract 100% Table 4. BTEX mixtures are also quantitatively ex-
nonylphenol from the sediment. A dynamic extrac- tracted by subcritical water using the conditions listed
tion time of 60 min with 27.5% methanol does not in Table 5.
completely extract aged nonylphenol from soil.
Ethoxylates
Table 6 shows a summary of sample preparation and
extraction conditions for alkylphenols. Alkylphenol ethoxylates are non-ionic surfactants
If the extractor does not allow the use of CO2, used for cleaning that are monitored in the environ-
100% methanol, ethanol, 1-butanol, 2-propanol, ment. Ethoxylates are extracted from sediment by
Table 3 Summary of pressurized fluid extraction validation data for EPA Method 3545. Reported relative recoveries for the various
compounds are the results of the PFE method used divided by the U.S. EPA reference methods listed
Compound class U.S. EPA reference method % Relative recovery
Chlorinated herbicides 8150A (shake method) 113

Organochlorine pesticides 3541 (automated Soxhlet) 97
Organophosphorous pesticides 3540 (Soxhlet) 99
Polychlorinated biphenyls 3540 (Soxhlet) 98
Polycyclic aromatic hydrocarbons 3540 (Soxhlet) 105
Semivolatile base/neutral/acids 3541 (automated Soxhlet) 99
Data compiled from the following sources: Ezzell (1998) American Environmental Laboratory January/February 24; Ezzell et al. (1995)
LC.GC 13: 390; and Richter et al. (1995) American Laboratory February, 24.
Table 4 Summary of standard sample preparation and extrac- Table 6 Summary of sample preparation and extraction condi-
tion conditions used to extract aliphatic hydrocarbons, BTEX, tions used to extract alkylphenols, ethoxylates, and linear alkyl-
gasoline and TPH from soil benzene sulfonates from environmental solid and semi-solid sam-
ples
Extraction conditions Extraction/analysis conditions
Dispersing or drying agent None used
Sample size 10 g Sample preparation Water was separated from
Extraction cell volume 11 mL sediment by centrifugation.
Extraction solvent Methylene chloride for aliphatic Sediment was dried at ambient
hydrocarbons and BTEX temperature and ground to
Temperature 1003C a particle size of less than
Heat step 5 min 10 m and water content was
Static time 5 min adjusted to 10% (w/w). After
Flow type Mix of both static and dynamic filling the cells, empty space in
extraction the extraction cell was filled with
Number of cycles 1 3 mm diameter glass beads.
Extraction pressure 1500 psi Any remaining space, 2 mL was
filled with modifier solvent.
Extraction conditions from Ezzell JL and Richter BE (1996) Ameri- Dispersing or drying agent None used
can Environmental Laboratory February, 16. Sample size 1g
Extraction cell volume 5 mL
Extraction solvent CO2, modified with methanol,
PFE. Extraction conditions for octylphenol-9,5- ethanol, 1-butanol, 2-propanol,
ethoxylate, nonylphenol-13-ethoxylate and decyl- acetone, or dioxane.
phenol monoethoxylate are the same as those for Temperature 1003C
Heat step None
alkylphenols (see Table 6). Using the modiRed CO2 as
Static time 10 min
the extraction solvent with dynamic extraction, Flow type Mix of static (10 min) and dynamic
60 min longer are required. (5}60 min) extraction
Number of cycles 1
Explosives Extraction pressure 2200 and 2940 psi
Explosives are another important class of compounds
Krei{elmeier and Durbeck (1996) Fresenius Journal of Analytical
that are monitored in the environment. They are Chemistry 354: 921.
monitored in military site or decommissioned site soil
Table 5 Summary of sample preparation and extraction condi-
since the explosives can potentially contaminate
tions used to extract aliphatic hydrocarbons, BTEX, PCBs, and
PAHs from soil, sediment, sludge, and urban dust using water as water supplies. PFE extracts HMX (octohydro-
the extraction solvent 1,3,5,7-tetranitro-1,3,5,7-tetrazocine), RDX (hexa-
hydro-1,3,5-trinitro-1,3,5-triazine), TNT (2,4,6-tri-
Extraction conditions Extraction analysis conditions nitrotoluene), and DNT (2,4-dinitrotoluene) using
the conditions shown in Table 7.
Sample preparation Soil is homogenized
Dispersing or drying agent None used The extraction solvent used depends on the analyti-
Sample size 0.2}0.5 g, cell void volume filled cal technique preferred. Acetone is the best choice for
with sand GC and methanol is the best solvent for LC analysis.
Extraction cell volume 0.5}0.8 mL (4.6 mm i.d.;30 mm For the results shown in Figure 1, methanol was used
stainless steel HPLC column or
as the extraction solvent. The extraction step takes
4.6 mm i.d.;50 mm stainless
steel SFE cell) 12 min per sample, resulting in 45 mL of extract.
Extraction solvent HPLC grade water purged two
hours with nitrogen to remove Gasoline
dissolved oxygen
Temperature 2503C Gasoline is extracted from spiked sand using the
Extraction time 15 min standard PFE conditions listed in Table 4. The aver-
Flow type Dynamic, 1 mL\1 min\1 age recovery of gasoline using these conditions is
Number of cycles 1
94.4%, as determined by IR detection.
Extraction pressure 73.5 psi for aliphatic hydrocarbons
and 735 psi for BTEX, PCBs,
and PAHs Linear Alkylbenzenesulfonates
Extraction conditions from Yang et al. (1997) Environmental Anionic surfactants are widely used for industrial as
Science and Technology 31: 430. Reproduced with permission well as household cleaning and for pesticide formula-
from the American Chemical Society. tions. Of the anionic surfactants, biodegradable

tions used to extract explosives from soil
Sample preparation Soil is homogenized

Dispersing or drying agent None used
Sample size 30 g
Extraction cell volume 33 mL
Extraction solvent Methanol or acetone
Temperature 1003C
Heat step 5 min
Static time 5 min
Flow type Mix of both static and dynamic
Number of cycles 1
Extraction pressure 1500 psi
Flush volume 60% of cell volume
Purge time 90 s Figure 2 Average percent relative recovery data of herbicides
extracted from three soil types. Soils were fortified with 50}500
Extraction conditions from Ezzell (1998) American Environmental and 500}5000 g kg\1. (Data from Ezzell et al. 1995.)
Laboratory January/February 24.
generate greater harvests and to protect food sup-

linear alkylbenzenesulfonates (LAS) are the most plies. As pesticides are present in soil, sediment and
common and can be found in waste water systems water, they are an important class of environmental
and river water. From these water sources, anionic compounds. Pesticides are extracted by PFE using
surfactants partition to sediment. LAS are selectively organic and aqueous solvents.
extracted from sediment by PFE allowing simple Sample preparation and extraction conditions for
quantitation by high performance liquid chromatog- chlorinated herbicides, OCPs, OPPs in soil using EPA
raphy. Extraction conditions for linear LAS [linear Method 3545 are shown in Table 2. The method
decyl- up to tridecylbenzenesulfonate (LAS-10 to (Method 3545) speciRes the use of mixtures of or-
LAS-13)] are the same as those for alkylphenols listed ganic solvents to extract a wide variety of organo-
in Table 6. Methanol is the best CO2 modiRer (at chlorine and organophosphate pesticides and chlorin-
2200 psi) for these compounds. ated herbicides.
At least eight chlorinated herbicides (2,4-D; 2,4-
Pesticides
DB; 2,4,5-T; 2,4,5-TP; dalapon, dicamba, dichlorop-
Pesticides including herbicides, insecticides and fungi- rop and dinoseb) are extracted using this method.
cides are broadly applied throughout the world to Extraction takes 12 min per sample, generating 15
mL of extract. Average relative recoveries for the

tions used to extract pyrithiobac sodium from soil
Sample preparation Remove sticks and rocks and

break up clumps of soil
Dispersing or drying agent Mix 10 g soil with 7 g of silica gel
Sample size 10 g
Extraction cell volume 22 mL cell
Extraction solvent Water
Temperature 1003C
Heat step 5 min
Static time 5 min
Flow type Mix of both static and dynamic
extraction
Number of cycles 1
Figure 1 Average percent relative recovery data of explosives Extraction pressure 200 psi
extracted from soil spiked at 3 mg kg\1 level. (Reprinted from Flush volume 60% of cell volume
Ezzell 1998. Copyright 1998 by International Scientific Com- Purge time 60 s
munications, Inc.)
Figure 3 Average percent relative recovery data of organo- Figure 4 Average percent relative recovery data of organo-
chlorine pesticides extracted from three soil types. Soils were phosphorous pesticides spiked at low and high levels on three soil
types and then extracted using EPA Method 3545. Soils were
fortified at 5, 50 and 250 g kg\1. (Data from Richter et al., 1995.
Copyright 1995 by International Scientific Communications, Inc.) fortified with 250 and 2500 g kg\1. (Data from Ezzell et al., 1995.)
eight chlorinated herbicides are shown in Figure 2. Polychlorinated Biphenyls

Relative recoveries are the quotient of PFE recoveries
divided by shaker extraction recoveries. Polychlorinated biphenyls (PCBs) are signiRcant en-
Organic acids are extracted using organic solvents vironmental pollutants that are routinely monitored
and by subcritical water. The advantages of using in soil, sediment and sludge. PCBs are extracted from
aqueous extraction solvent are (1) purchase and dis- these matrices using organic and aqueous extraction
posal costs are low and (2) it is environmentally solvents. PFE conditions using organic solvents are
friendly. Conditions used to selectively extract py- listed in Table 2; PFE conditions used for aqueous
rithiobac sodium (the active ingredient in Staple威 solvents are listed in Table 5. Although not listed in
herbicide, used to control broadleaf weeds in cotton) Table 2, acetonitrile and methylene chloride also pro-
are listed in Table 8. vide adequate extraction of PAHs in environmental
Organochlorine pesticides (OCPs) are extracted us- solid and semi-solid samples. Concentrations of PCBs
ing the conditions listed in Table 2. Average relative extracted from soil, sediment, urban dust and sludge
recoveries for the 20 OCPs are shown in Figure 3. using PFE and exhaustive Soxhlet extraction are sim-
Relative recoveries are the quotient of PFE recoveries ilar. This is true for both organic and aqueous extrac-
divided by automated Soxhlet extraction recoveries. tion solvents in PFE.
Organophosphate pesticides (OPPs) listed in
Table 9 are extracted from soil using the conditions
Table 10 Summary of sample preparation and extraction con-
listed in Table 2. Average relative recoveries for the
ditions used to extract dioxins and furans from chimney brick,
24 OPPs are shown in Figure 4. Relative recoveries urban dust, fly ash and sediments
are the quotient of PFE recoveries divided by Soxhlet
extraction recoveries. Extraction conditions Extraction analysis conditions
Sample preparation Grind soil to 100}200 mesh

Table 9 Twenty-four organophosphate pesticides extracted by (150}75 m particle size)
EPA Method 3545 Dispersing or drying agent Mix with anhydrous sodium
sulfate or HydromatrixTM
Azinphos methyl Mevinphos Sample size 4}10 g
Chlorpyrifos Monocrotphos Extraction cell volume 11, 22 or 33 mL cell, depending
Coumaphos Naled on sample size
Demeton-O,S Parathion ethyl Extraction solvent Toluene, 15 mL
Diazinon Parathion methyl Temperature 1803C
Dichlorovos Phorate Heat step 9 min
Dimethoate Ronnel Static time 5 min
Disulfoton Sulfotepp Flow type Static, dynamic, or mix of both
EPN Sulprofos Number of cycles 1}2
Ethoprop TEPP Extraction pressure 2000 psi
Fensulfothion Tetrachlorvinphos
Fenthion Tokuthion Extraction condition reproduced from Richter et al. (1997)
Chemosphere 34: 975, with permission from Elsevier Science.
Polychlorinated Dioxins and Furans terials indicate that PFE efRciently extracts PAHs
from these matrices. Figure 5 shows a comparison of
Polychlorinated dibenzo-p-dioxins (PCDDs) and
the amounts of PAHs found in PFE extracts and
polychlorinated dibenzofurans (PCDFs) are environ-
certiRed values.
mental pollutants that are extracted from chimney
The following organic solvents also provide ad-
brick, urban dust, Sy ash, soil and sediment using
equate extraction of PAHs in environmental samples:
PFE. Conditions for the extraction of PCDDs and
toluene/methanol 1 : 1 (v : v), toluene, methylene
PCDFs are listed in Table 10.
chloride, acetonitrile, hexane/acetone 1 : 1 (v : v) and
PFE and Soxhlet extraction results of PCDDs and
water. Figure 6 shows the amounts of PAHs found in
PCDFs levels found are essentially equivalent for the
urban dust (SRM 1649) extracts comparing various
matrices tested. PFE required less time and solvent
organic and aqueous extraction solvents.
than Soxhlet extraction.
Polycyclic Aromatic Hydrocarbons Semi-Volatile Base/Neutral/Acid Compounds

Polycyclic aromatic hydrocarbons (PAHs) are impor- Semi-volatile base/neutral/acid compounds (BNAs)
tant environmental compounds being carcinogenic are extracted from solid and semi-solid environ-
and mutogenic. They are extracted by organic sol- mental samples using the conditions listed in Table 2.
vents and by subcritical water. Using organic sol- Average relative recoveries for the 56 US EPA priority
vents, PAHs are extracted from urban dust, sediment pollutants list (PPL) and target compound list (TCL)
and soil using the PFE conditions listed in Table 2. are shown in Figure 7. Relative recoveries are the
Subcritical water extraction requires the conditions quotients of PFE recoveries divided by Soxhlet extrac-
listed in Table 5. Results from certiRed reference mation recoveries.
Figure 5 Comparison of amounts of PAHs found in PFE extracts and certified values. (Data from Richter, Jones, Ezzell et al. 1996.)
Figure 6 Recovery of PAHs from urban dust (certified reference NIST SRM 1649) extracted by PFE. Extraction solvents were water
(2503C, 725 psi) and methylene chloride/acetone 1 : 1. (Data for water extraction from Hawthorne et al., 1994.) Data for methylene
chloride extraction (DCM 1) from Richter et al., 1995. Data for methylene chloride extraction (DCM 2) from Schantz et al., 1997.)
Total Petroleum Hydrocarbons environmental samples. In comparing PFE to Soxhlet,

sonication and shaking extraction methods, re-
Total petroleum hydrocarbons (TPHs) are extracted
searchers Rnd that the amounts of analyte extracted
from soil samples using the standard PFE conditions
from solid and semi-solid samples are similar, but
listed in Table 4. Levels of TPHs extracted by PFE
PFE requires less extraction time and less extraction
and Soxhlet extraction are typically similar.
solvent. PFE methods are easy to develop and to
transfer. PFE provides automated sample extraction,
Future Applications allowing increased productivity of laboratory person-
There are several reasons why analysts use PFE to nel. Finally, PFE methods are being accepted by regu-
extract analytes of interest from solid and semisolid latory agencies. Given these advantages, researchers
III / ENVIRONMENTAL APPLICATIONS / Solid-Phase Microextraction 2695
with conventional solvent extraction for organophos-

phorus pesticides and hebicides. LC.GC 13: 390}398.
Hawthorne SB, Yang Y and Miller DJ (1994) Extraction of
organic pollutants from environmental solids with sub-
and supercritical water. Analytical Chemistry. 66:
2912}2920.
Krei{elmeier A and Durbeck H-W (1996) Determination of
alkylphenols and linear alkylbenzene sulfonates in sedi-
ments applying accelerated solvent extraction (ASE).
Fresenius Journal of Analytical Chemistry. 354:
921}924.
Richter BE, Ezzell JL, Felix D, Roberts KA and Later DW
(1995) An accelerated solvent extraction system for the
Figure 7 Average percent relative recovery data of 56 semi- rapid preparation of environmental organic compounds
volatile base/neutral/acid components extracted from three soil in soil. American Laboratory 27: 24}28.
types using PFE. Soils were fortified at 250, 2500, and Richter BE, Jones BA, Ezzell JL, Porter NL, Avdalovic
12 500 g kg\1. (Data from Richter et al. (1995) reproduced with N and Pohl C (1996) Accelerated solvent extraction:
permission from International Scientific Communications Inc.)
a technique for sample preparation. Analytical Chem-
istry 68: 1033}1039.
will continue to use PFE to develop many more envir- Richter BE, Ezzell JL, Knowles DE, HoeSer F, Mattulat
onmental sample applications. AKR, Scheutwinkel M, Waddell DS, Gobran T and
Khurana V (1997) Chemosphere 34: 975}978.
See also: II /Extraction: Supercritical Fluid Extraction. Schantz MM, Nichols JJ and Wise SA (1997) Evaluation of
III/Environmental Applications: Supercritical Fluid Ex- pressurised Suid extraction for the extraction of envir-
traction; Soxhlet Extraction. Superheated Water Mobile onmental matrix reference materials. Analytical Chem-
Phases: Liquid Chromatography. istry 69: 4210}4219.
USEPA SW-846, 3rd Edn., Update III (July 1995) Test
Further Reading Methods for Evaluating Solid Waste, Method 3545,
Accelerated Solvent Extraction. Washington. U.S. GPO.
Ezzell J (1998) The Use of SW-846 Method 3545 for Yang Y, Bowadt S, Hawthorn SB and Miller DJ (1995)
automated extraction of environmental samples. Ameri- Subcritical water extraction of polychlorinated bi-
can Environmental Laboratory. 10: 24}25. phenyls from soil and sediment. Analytical Chemistry.
Ezzell JL and Richter BE (1996) Automated sample prep- 67: 4571}4576.
aration for environmental laboratories using accelerated Yang Y, Hawthorne SB and Miller DJ (1997) Class-selec-
solvent extraction. American Environmental Laborat- tive extraction of polar, moderately polar, and nonpolar
ory. 16}18. organics from hydrocarbon wastes using subcritical
Ezzell J, Richter BE, Felix WD, Black SR and Meikle JE water. Environmental Science and Technology. 31:
(1995) A comparison of accelerated solvent extraction 430}437.
Solid-Phase Microextraction
T. Nilsson, Technical University of Denmark, various environmental applications are discussed.
Lyngby, Denmark Traditionally, SPME has been combined with analy-
Copyright ^ 2000 Academic Press sis by gas chromatography (GC), and mainly aqueous
samples have been analysed. This combination has
proved to be sensitive, accurate and precise for the
Solid-phase microextraction (SPME) is a technique quantitative analysis of volatile organic compounds
for the extraction of organic compounds from gas- and different classes of pesticides. Solid samples can
eous, aqueous and solid matrices such as many environ- also be analysed by SPME in spite of the stronger
mental samples. It is rapid and simple, which makes it matrix effects, and recently SPME has been coupled
ideal for automation and in situ measurements, and with liquid chromatography (LC) for the analysis of
no harmful solvents are used. The principle of SPME polar pesticides.
is equilibration of the analytes between an organic
polymeric phase coated on to a fused-silica Rbre and Principle
the sample matrix. The parameters of importance for The principle of SPME is that a fused-silica Rbre is
the equilibration process are described below and coated with an organic polymer and exposed to the
2696 III / ENVIRONMENTAL APPLICATIONS / Solid-Phase Microextraction
sample. The Rbre is mounted inside a steel syringe a theoretical treatment because of unpredictable
needle for protection in order to be able to penetrate matrix effects.
the septum of the sample vial and the GC injector
without damaging the Rbre. Subsequently, the Rbre Adsorption Kinetics
can be pushed out of the needle for exposure to the Basically, the kinetics of adsorption are governed by
sample. The analytes will then diffuse into the Rbre diffusion. A number of models have been reported for
coating until equilibrium has been established. different situations, but they provide only approxim-
ate descriptions under ideal conditions. In practice,
Extraction Ef\ciency
a number of factors will cause deviations from these
Basically, the extraction efRciency is determined by conditions, thus a simple, empirical model has been
the extraction time, the sample concentration and the described for practical purposes:
distribution constant of the analyte between the Rbre
coating and the sample. The classical situation is n1"n
1 [1!exp(!t/)] [3]
extraction with the Rbre immersed in a water sample.
where is a measure of the equilibration velocity, and
The amount of analyte extracted by the Rbre coat-
n1 is the amount adsorbed on the Rbre coating at the
ing equilibrium n1 is determined by the expression:
time t.
KV1V2C02
1 "
n [1]
KV1#V2 Extraction
The extraction conditions are optimized in order to
where K is the distribution constant, V1 is the volume
obtain a rapid and sensitive analysis. It is important
of the Rbre coating, V2 is the sample volume, and C02 is
to remember that SPME depends on diffusion and
the initial sample concentration.
distribution. Thus, the required extraction time can
Another SPME approach is sampling from a head-
be reduced by increasing the diffusion rates, and the
space above the sample in the vial. In this case, the
extraction efRciency can be improved by increasing
amount of analyte adsorbed after inRnite time n 1 is
the distribution constant.
given by the equation:
Choice of Fibre Coating
C02V1V2k
1 "
n [2] Various SPME Rbres are commercially available
kV1#kV3#V2
(Table 1). The polydimethylsiloxane (PDMS) coating
where V3 is the volume of the headspace, k is the Rbre is recommended for the extraction of nonpolar com-
coating}gas distribution constant, and k is the pounds. Three different PDMS Rbres exist with
gas}water distribution constant of the analyte. k is a coating thickness of 7, 30 and 100 m, respectively.
directly proportional to Henry’s constant and is usu- Usually, 100 m coating is used due to its higher
ally relatively low for the compounds analysed by extraction capacity. However, for higher boiling
SPME. Thus, only if V3 is considerable compared to compounds with high distribution constants and long
V2, a lower sensitivity will be observed with head- equilibration times, the thinner coatings should be
space}SPME (HS-SPME). The advantage of HS- used. The 7 m coating has the advantage that it is
SPME is that equilibration times will be much shorter chemically bonded and can be used at temperatures
due to the fact that the diffusion is several orders of up to 3403C.
magnitude faster in the gas phase than in liquids.
Extraction Parameters
Another advantage of HS-SPME is that it can be
applied for the analysis of solid and dirty samples. Reliable quantitative analysis can be performed under
However, such applications are not so amenable to nonequilibrium conditions, but the sensitivity will be
Table 1 SPME fibres and their recommended use
Fibre coating material Abbreviation Recommended use
Polydimethylsiloxane PDMS Nonpolar compounds

Carboxen/polydimethylsiloxane Carboxen/PDMS Very volatile compounds
Polyacrylate PA General
Polydimethylsiloxane/divinylbenzene PDMS/DVB General
Carbowax/divinylbenzene CW/DVB Polar compounds
Carbowax/templated resin CW/TPR Polar compounds
better when the extraction time is sufRcient to reach optimize the desorption parameters in order to
near-equilibrium. The equilibration time can be guarantee that the Rbre can be used for subsequent
shortened by agitation or heating of the sample which analysis without intermediate cleaning. This is essen-
increase the diffusion rates. Normally, an extraction tial for automation purposes and for trace analysis.
time comparable to the time of the chromatographic For GC analysis, desorption should be as rapid as
analysis is chosen in order to maximize sample possible. The best injection is obtained when the
throughput. Rapid stirring using a magnetic bar is desorption temperature is sufRciently high to ensure
efRcient, but may not always be very reproducible; an almost complete desorption within 1 min or less.
vibration of the Rbre is a valid alternative for small However, a longer desorption time is often required
samples. At higher temperatures the equilibration to avoid carry-over, in which case cryogenic focusing
will proceed faster due to the higher diffusion rates; may be necessary. For LC analysis, desorption using
however, the amount adsorbed at equilibrium will be the mobile phase is the best solution and can even be
lower. performed in the Sowing mobile phase (dynamic
An internally cooled SPME device has been de- mode) if the desorption is rapid. However, a higher
veloped for the purpose of maintaining favourable content of an organic solvent is often needed to ob-
distribution constants while extracting from a heated tain a satisfactory desorption. In this case, the desorp-
sample. Sample heating may be necessary to release tion chamber is Rlled with the appropriate solvent
analytes that are adsorbed on solid matrices. Addi- mixture and desorption takes place (static mode) be-
tion of a salt normally has a positive inSuence on the fore the injection is performed by switching the valve.
extraction efRciency due to the increased ionic A high content of organic solvent may adversely af-
strength of the solution. When acidic compounds are fect the chromatography if the initial mobile phase is
analysed, the pH should be below the lowest pKa much weaker. Furthermore, the SPME Rbres are not
value, because ionic compounds are not extracted but very resistant to organic solvents, so the coupling of
the lifetime of the Rbre is reduced at low pH values. SPME with LC is still at the experimental stage.
A methanol content of less than 1% in spiked samples
does not affect the SPME process signiRcantly. It
must always be borne in mind that relatively large
Analysis of Aqueous Samples
amounts of other compounds in a complex matrix Numerous successful applications of SPME for the
may have a signiRcant effect on the distribution con- analysis of aqueous samples have been reported. The
stant. analytical conditions are summarized in Table 2, and
the results for the different classes of compounds are
discussed below. In most of the studies, only spiked
Desorption samples have been analysed; however, considering
In case of analysis by GC, thermal desorption is the limited effect of suspended solids and humic
performed in the injector. For analysis by LC, the substances at levels typical for lake, river and
injection loop is replaced by a special SPME desorp- groundwater, such environmental samples can also
tion chamber and the desorption is performed in the be analysed. In more complex sample matrices,
mobile phase or a solvent mixture. It is important to SPME can be used to measure the freely dissolved
Table 2 Applications of SPME for the analysis of aqueous samples
Compound Fibre coating Analysis (alternative detector for

some compounds)
Volatile organic hydrocarbons PDMS GC-MS (FID)

Halogenated volatile organic compounds PDMS GC-MS (ECD)
Polychlorinated biphenyls, polyaromatic hydrocarbons and
heteroaromatic compounds PDMS or PA GC-MS (ECD, FID)
Phenols and nitro-compounds PA or more polar GC-MS (ECD)
Organochlorine, organonitrogen and organophosphorus pesticides PA or more polar GC-MS (ECD, AED, NPD)
Fatty acids, phenoxy acid herbicides and amines PA or PDMS/DVB Derivatization/GC-MS (ECD, FID)
Organometallics and inorganic metal ions PDMS Derivatization/GC-MS
Phenoxy acid, sulfonylurea, phenylurea, carbamate and other polar
herbicides CW/TPR LC-ESI/APCI-MS (UV, DAD)
Abbreviations: MS, mass spectrometry; FID, flame ionization detection; ECD, electron capture detection; AED, atomic emission
detection; NPD, nitrogen and phosphorus detection; ESI, electrospray ionization; APCI, atmospheric pressure chemical ionization; UV,
ultraviolet absorption; DAD, diode array detection.
compounds. While the traditional techniques extract ciency is higher at low pH values. Typically, detection
the total amount, only a small amount is extracted by limits are in the low g L\1 range and the relative
SPME, so the equilibrium with the matrix is not standard deviations are from 5 to 12%. The sensitiv-
perturbed. By addition of an internal standard. e.g. ity and chromatography can be improved for most of
a deuterated surrogate, the total concentration and the phenols by aqueous-phase acetylation prior to
the distribution of the analytes can be determined. extraction. Nitrotoluenes, nitroanilines and nitroben-
Unless an isotope-labelled analogue of the analyte is zenes have also been analysed successfully by SPME.
used, the beneRt of an internal standard in SPME
analyses is very limited because even similar com- Organochlorine, Organonitrogen and
pounds may behave differently in the SPME process. Organophosphorus Pesticides
In several studies, the analysis of organochlorine
Volatile Organic Hydrocarbons
pesticides has been examined. Generally, equilibra-
The analysis of benzene, toluene, ethylbenzene and tion times range from 30 to 90 min, detection limits
m-, o- and p-xylene (BTEX compounds) by SPME has are in the low ng L\1 range with electron-capture
been studied extensively. Many other gasoline and detection (ECD) and mass spectrometry (MS), and
fuel-related hydrocarbons have also been analysed. standard deviations vary from 5 to 20%. For the
Generally, the standard deviation of replicates is organonitrogen and organophosphorus pesticides,
around 5% and detection limits are in the low g L\1 similar precision and equilibration times have been
range for the lightest compounds down to low ng L\1 observed, and the detection limits are at very low
level for the higher boiling analytes with the PDMS ng L\1 level with MS and nitrogen and phosphorus
Rbre coating. detectors.
Halogenated Volatile Organic Compounds Fatty Acids, Phenoxy Acid Herbicides and Amines
Numerous applications of SPME for the analysis of Fatty acids can be analysed directly from aqueous
halogenated volatile organic compounds have been samples by SPME. However, for short chain fatty
reported. The PDMS Rbre coating performs well acids the detection limits are in the high g L\1 range
for these compounds. The precision and sensitivity with the polyacrylate Rbre coating and worse with
are similar to those reported for the volatile organic other Rbre coatings. However, the sensitivity can be
hydrocarbons. considerably improved by derivatization. Different
derivatization procedures have been examined and
Polychlorinated Biphenyls, Polycyclic Aromatic detection limits below g L\1 can be obtained in the
Hydrocarbons and Hetero-Aromatic Compounds best cases. Similar detection limits are obtained for
Polychlorinated biphenyls (PCBs) and polycyclic aro- phenoxy acid herbicides and amines by derivatiz-
matic hydrocarbons (PAHs) have been analysed in ation-SPME.
spiked water samples and in groundwater. The equili-
Organometallics and Inorganic Metal Ions
bration times were approximately 60 min with the
PDMS Rbre coating. However, detection limits in the SPME has mainly been applied for organic trace anal-
low ng L\1 range can be obtained with an extraction ysis. However, a few applications for inorganic metal
time of only 10 min. The relative standard deviations ions and organometallics have been reported: bis-
of these analyses are around 20% for the PCBs and muth was extracted using an experimental SPME
10% for the PAHs. Possibly, better precision could be Rbre coated with a liquid ion exchanger; aqueous-
achieved by increasing the extraction time in order to phase derivatization with tetraethylborate followed
approach equilibrium. Pyrazines and several other het- by SPME has been applied to analyse methylmercury
eroaromatic compounds have been analysed success- and labile Hg2# in river water, and the same derivat-
fully with detection limits from low g L\1 levels for ization approach can be used for the analysis of tin
the volatile analytes down to low ng L\1 levels for the and lead. The detection limits are in the low ng L\1
semivolatile analytes. The precision is in the range range.
from 3 to 14% relative standard deviation. The extrac-
tion efRciency is strongly enhanced by salt addition. Phenoxy Acid, Sulfonylurea, Phenylurea,
Carbamate and Other Polar Herbicides
Phenols and Nitro-Compounds
SPME coupled with LC-electron spray ionization
Usually, salt addition has a positive effect on the (ESI)/atmospheric pressure chemical ionization
extraction of phenols and nitrophenols, and for (APCI)-MS has been applied for the trace analysis of
analytes with pKa values below 7 the extraction efR- polar pesticides in spiked water samples and lake
water. Acidic, as well as neutral, priority pesticides from wells. These methods require pumping of the
representing all major pesticide classes can be ana- groundwater to the surface, sampling into appropri-
lysed successfully with single ion monitoring (SIM) ate containers, and transport to the analytical labor-
detection limits in the ng L\1 range. Detection limits atory. Sample loss and sample composition variation
in the low g L\1 range and standard deviations be- may occur during these steps. Thus, a number of
low 10% were reported when UV detection was used. downhole sampling devices have been developed.
Finally, SPME}Sow injection}MS-MS has been de- However, each has limitations with respect to Sexi-
veloped for the purpose of rapid, target-oriented bility of sample type and sample size, maximum oper-
screening analysis. ating pressures and depth, portability and adaptabil-
ity to difRcult Reld conditions. The characteristics of
Validation of Standard Methods for SPME make it ideal for Reld sampling, i.e. it is simple,
robust, portable, independent of sample volume and
Routine Analysis instrument conRguration, and it is impossible to plug
In order for SPME to be applied for routine analysis, the Rbre with particulate matter. Thus, SPME samp-
two issues are very important: automation and qual- ling probes have been developed for use in monitor-
ity assurance. Thus, an autosampler has been de- ing wells (Figure 1) and for Rtting to the head of
veloped for SPME-GC analysis, and three inter- a cone penetrometer. They were tested in a mobile
laboratory studies have been performed to validate laboratory for on-site measurements of volatile or-
the quantitative performance of SPME. One study ganic compounds in groundwater and soil gas. Com-
addressed the analysis of 12 organochlorine, or- parison of results obtained by in situ SPME and
ganonitrogen and organophosphate pesticides at low SPME after traditional sampling from the same
g L\1 level in aqueous samples. In the other two groundwater wells conRrmed the feasibility of the in
studies, standard methods for the analysis of volatile situ sampling approach. Slightly lower results were
organic compounds and triazine herbicides in aque- obtained in most cases after traditional sampling.
ous samples were validated at low g L\1 levels and
around the European limit of 0.1 g L\1 for indi-
Solid Samples
vidual pesticides in drinking water. Subsequently,
both methods were presented by the Italian Standard- The major complication in the analysis of soil
ization Organization, Unichim. The validations were samples is the strong sorption of the analytes to the
performed in accordance with the International Stan- matrix. A nearly complete exhaustive extraction
dardization standard method 5725-1994 concerning could be achieved for the BTEX compounds from
interlaboratory statistics. The results regarding accu- sand and clay matrices by heating the sample and
racy and precision are given in Table 3. The 95% using an internally cooled SPME Rbre for the extrac-
conRdence interval of the gross average of the re- tion. However, in the case of less volatile and more
ported results always included the true concentration polar analytes, the sorption is stronger and the recov-
of the test sample, i.e. the accuracy of the methods ery is very dependent on the organic carbon content
was very good. The precision obtained would meet of the soil. Thus, calibration using a model matrix
the requirements for most routine analyses. would only be acceptable for screening purposes,
while calibration by standard addition is needed for
reliable quantitative analysis. Addition of water to
In Situ Measurements the sample helps to release the analytes from the
Many well-established extraction techniques have sample matrix and improves the recoveries drasti-
been applied for the analysis of groundwater samples cally. A nearly complete extraction of polyaromatic
Table 3 Interlaboratory validation of SPME for quantitative analysis
Analytes Number of Concentration of Accuracy Average Average

participating the test sample repeatability reproducibility
laboratories (g L\1) (%) (%)
Volatile organic 12 3 True values within 9.3 28.7

compounds confidence intervals
Organochlorine, 11 2}25 True values within 11.5 28.3
organonitrogen and confidence intervals
organophosphorus
pesticides
Triazine herbicides 8 0.05}0.12 True values within 9.6 13.6
confidence intervals
Figure 1 SPME probe for in situ groundwater sampling from wells. (Reproduced with permission from Nilsson et al., 1998.)
hydrocarbons from different soils has been achieved organic compounds. However, further development
by high temperature or subcritical water extraction. of these methods is needed before they can be applied
Finally, the leachability of pesticides from soils has for routine analysis. Especially, further research on
been studied by SPME. the coupling of SPME and LC-MS may allow many
new environmental applications. Inorganic metal
ions and organometallics have been analysed as well,
Conclusion and the use of an ion exchange Rbre coating may
SPME has successfully been applied for the quantita- allow more applications in this Reld. The small vol-
tive analysis of most of the organic, environmental ume and the noninterfering character of SPME are
priority compounds, which can be analysed by GC, in important factors for numerous applications. Finally,
aqueous samples. In more complex sample matrices, the easy handling and simple design make SPME
such as wastewater and soils samples, quantitative a good choice for in-Reld analytical work.
analysis by SPME may be difRcult because matrix
effects inSuence the distribution constants signiR- See also: II/Extraction: Solid-Phase Microextraction.
cantly. Standard methods have been developed and
validated regarding sensitivity, precision and accu-
racy for the trace analysis of volatile organic com-
Further Reading
pounds and several classes of pesticides in aqueous Ai J (1997) Solid phase microextraction for quantitative
samples. Derivatization/SPME-GC and SPME-LC analysis in nonequilibrium situations. Analytical Chem-
have been applied for the analysis of more polar istry 69: 1230.
III / ENVIRONMENTAL APPLICATIONS / Soxhlet Extraction 2701
Chen J and Pawliszyn JB (1995) Solid-phase microextrac- Nilsson T, Pelusio F, Montanarella L et al. (1995) An
tion coupled to high-performance liquid chromatogra- evaluation of solid-phase microextraction for analysis of
phy. Analytical Chemistry 67: 2530. volatile organic compounds in drinking water. Journal
Daimon H and Pawliszyn J (1996) High temperature water of High Resolution Chromatography 18: 617.
extraction combined with solid phase mictroextraction. Nilsson T, Ferrari R and Facchetti S (1997) Inter-laborat-
Analytical Communications 33: 421. ory studies for the validation of solid-phase microextrac-
Eisert R and Levsen K (1996) Solid-phase microextraction tion for the quantitative analysis of volatile organic
coupled to gas chromatography: a new method for the compounds in aqueous samples. Analytica Chimica Acta
analysis of organics in water. Journal of Chromatogra- 356: 113.
phy A 733: 143. Nilsson T, Montanarella L, Baglio D et al. (1998) Analysis
Eisert R and Pawliszyn J (1997) New trends in solid-phase of volatile organic compounds in environmental water
microextraction. Critical Reviews in Analytical Chem- samples and soil gas by solid-phase microextraction.
istry 27: 103. International Journal of Environmental Analytical
Fromberg A, Nilsson T, Larsen BR et al. (1996) Analysis of Chemistry 69: 217.
chloro- and nitroanilines and -benzenes in soils by head- Pawliszyn J (1997) Solid Phase Microextraction, Theory
space solid-phase microextraction. Journal of Chrom- and Practice. New York: Wiley-VCH.
atography A 746: 71. Pawliszyn J (ed.) (1999) Applications of Solid Phase Micro-
Hageman KJ, Mazeas L, Grabanski CB et al. (1996) extraction. Cambridge: Chromatography Monographs
Coupled subcritical water extraction with solid-phase Series, Royal Society of Chemistry.
microextraction for determining semivolatile organics in Zhang Z and Pawliszyn J (1993) Headspace solid-
enviornmental solids. Analytical Chemistry 68: 3892. phase microextraction. Analytical Chemistry 65:
Louch D, Motlagh S and Pawliszyn J (1992) Dynamics of 1843.
organic compounds extraction from water using liquid- Zhang Z, Yang MJ and Pawliszyn J (1994) Solid-phase
coated fused silica Rbres. Analytical Chemistry 64: microextraction, a solvent-free alternative for sample
1187. preparation. Analytical Chemistry 66: 844A.
Soxhlet Extraction
M. D. Luque de Castro and L. E. GarceH a Ayuso, of some of these has not been as great as desirable.
University of CoH rdoba, CoH rdoba, Spain These steps constitute ‘critical points’ of an analytical
Copyright ^ 2000 Academic Press method and, consequently, the main source of errors.
The pretreatment step (including separation tech-
niques) can be considered as a ‘critical step’. Conven-
The health of our environment is now a matter of tional Soxhlet extraction is currently one of the most
great concern. This has stimulated an intensive search frequently used pretreatment techniques, not only in
for an understanding of both the ways in which the environmental analysis, but also in many other Relds.
natural environment works and the anthropogenic Its principles, performance, environmental applica-
actions that bring about environmental changes. tions and improvements are considered in more detail
A large number of studies have been, or are in the below.
process of being, developed in order to increase our
knowledge of the processes causing environmental Principles of Conventional Soxhlet
pollution and to propose clean analytical methods for
monitoring and subsequent control. Thus, a high per-
Extraction
centage of the studies developed so far fall within the Soxhlet extraction is a very useful tool for preparative
Reld of analytical chemistry. There are a number of purposes in which the analyte is concentrated from
stages involved in any analytical method: deRnition of the matrix as a whole or separated from particu-
the aim, selection and establishment of an appropri- lar interfering substances. Sample preparation of
ate method, sampling plan, sample collection, sample environmental samples has been developed for dec-
handling and pretreatment, Rnal measurements (de- ades using a wide variety of techniques. Solvent ex-
tection/determination), method validation, assess- traction of solid samples, which is commonly known
ment and interpretation of the results and, Rnally, as solid}liquid extraction (also referred to as leaching
safety. or Lixiviation in a more correct use of the
In spite of the evolution of analytical techniques physicochemical terminology), is one of the oldest
involved in some of the above mentioned stages (par- methods for solid sample pretreatment. Conventional
ticularly detection/determination), the development Soxhlet extraction remains as one of the most
2702 III / ENVIRONMENTAL APPLICATIONS / Soxhlet Extraction
and the role of Soxhlet extraction in the overall ana-

lytical process.
Table 1 summarizes the main characteristics of Sox-
hlet extraction, clean up and detection/determination
for each group of pollutants discussed below.
Use of Conventional Soxhlet

Extraction for Leaching
of Organic Pollutants
Organic pollutants present in solid environmental
samples from natural or man-made sources are most
of the time properly extracted with organic solvents
in a Soxhlet device. The following sections show how
conventional Soxhlet extraction is used as a technique
for separation of the analytes from the solid matrix.
The pollutants are classiRed into three groups de-
pending on the matrices in which they are present:
air, solid and liquid samples. It is worth noting that,
in spite of the solid character of particulates, the
samples are classiRed as a function of the medium in
which they are present. Thus, airborne and water
particulates are classiRed as air and liquid samples,
respectively.
Figure 1 Soxhlet extraction apparatus. (Reproduced from
Reeves RN (1994) Environmental Analysis. New York: John Organic Pollutants in Air
Wiley.) Organic pollutants are in the gas phase of the atmos-
phere and are associated with airborne particulate
relevant techniques in the environmental extraction matter. Sampling of organic pollutants is usually car-
Reld. This assertion is supported by the double use of ried out by Rlters through which a large volume of air
conventional Soxhlet: (1) as an extraction step in is circulated. After Rltration, the Rlter is extracted
a given method, and/or (2) as a well-established with an appropriate solvent and the target analytes
model for comparison of new extraction alternatives. are removed in the extract. When the extract is not
In conventional Soxhlet, the sample is placed in suitable for direct individual separation/determina-
a thimble-holder and during operation is gradually tion a prior clean-up stage must be applied. These and
Rlled with condensed fresh solvent from a distillation other aspects are discussed below.
Sask. When the liquid reaches an overSow level,
a siphon aspirates the whole contents of the thimble- Polycyclic aromatic hydrocarbons Polycyclic aro-
holder and unloads it back into the distillation Sask, matic hydrocarbons (PAHs) and their derivatives are
carrying the extracted analytes in the bulk liquid. among the most studied environmental pollutants.
This operation is repeated until complete extraction is This is due to both their continuous emissions from
achieved. This performance makes Soxhlet a hybrid combustion and their biological activities, such as
continuous}discontinuous technique. Inasmuch as toxicity, mutagenicity and carcinogenicity. Deter-
the solvent acts stepwise, the assembly can be con- mination of PAHs from air requires a suitable samp-
sidered as a batch system; however, since the solvent ling device, including an aspiration pump, which
is recirculated through the sample, the system also retains the air particulates on glass or quartz Rbre,
bears a continuous character. Figure 1 shows cellulose or paper Rlters. However, the Rlter is not the
a scheme of a conventional Soxhlet device. As can be sole collection medium for PAH-containing small-
seen, Soxhlet extraction is a very simple technique. size aerosols. Chemicals such as Tenax GC, PUFs (see
This simplicity makes the procedures for different Table 2 for abbreviations) or Amberlite XAD-2
samples very similar. For this reason, an overview of resins should provide a back up to these Rlters by
its extensive application in the environmental Reld retaining PAHs not trapped by the Rlter. A layer of
during the last two decades is presented here, which activated charcoal between two Rlters is used for
cover the kind of samples, analytes, solvents, etc. used collection of gaseous components.
Table 1 Selection of the most representative characteristics of methods for determination of organic pollutants with Soxhlet extraction as pretreatment step
Compound Matrix Sampling Soxhlet features Clean up Detect/Determ.
Solvent Time
PAHs Solid samples (soil, Homogenization Polar species From 8 to 48 h HSorbents Techniques
sediment, sludge, (mix, ground and Dimethyl formamide depending on the Alumina (non-polar analytes GC
biological tissue, etc.) sieve) Tetrahydrofuran matrix, analyte and with polar interferents) HPLC
Moisture control Acetonitrile solvent Silica gel (strongly polar TLC
PCDDs/PCDFs Acetone species) SFC
Methanol Florisil (aromatic and Spectroscopy
Ethanol aliphatic interferents)
Activated carbon (non-polar Detectors
PCBs/OCPs Non-polar species analytes from aqueous FID (hydrocarbons)
Hexane samples) NPD (nitrogen and
Iso-octane phosphorous)
Phenols Ethers HSolid phase extraction FPD (sulfur and
Particulates, vapour, Filters (glass or Light petroleum }(non-polar species) phosphorus)
smoke, liquid samples quartz fibre) C18 ECD (halogenated)
Benzidines and PUFs, Aromatics C8 UV/VIS and DAD
XAD resins, Benzene Cyclohexyl MS (universal)
Tenax GC, etc. Toluene Phenyl Flourescence
Cyano
Nitrosamines General use }(polar species)
Dichloromethane Amino
Ethyl acetate Diol
Isopropyl alcohol HChromatography
Phthalate esters Chloroform HDistillation
HLLE
Etc.
III / ENVIRONMENTAL APPLICATIONS / Soxhlet Extraction

2703
Table 2 Abbreviations particulate phases before extraction. The Rlter and

foam are then extracted by Soxhlet with toluene (ben-
ASE Accelerated solvent extraction
zene/ethanol and dichloromethane have also been
CE Capillary electrophoresis
CIMS Chemical ionization mass spectrometry used but less frequently than toluene), followed by an
ECD Electron capture detector acid}base clean-up step, then drawing of the extract
EIMS Electron impact mass spectrometry through micro-columns of silica gel, alumina and
EPA Environmental protection agency carbon (sometimes a chromatographic clean up is
FID Flame ionization detector
applied before). Individual separation of the analytes
FMASE Focused microwave-assisted Soxhlet extraction
GC Gas chromatography and determination are performed by high-resolution
HPLC High performance liquid chromatography GC}EIMS-SIM.
HPS High pressure Soxhlet extraction
LC Liquid chromatography Polychlorinated biphenyls (PCBs) and other chlorin-
LLE Liquid}liquid extraction
ated compounds A wide variety of pollutants other
MAE Microwave assisted extraction
MS Mass spectrometry than PCDDs and PCDFs contain chlorine. This
NPD Nitrogen phosphorus selective detector makes them very harmful due to their activity in the
OCPs Organochlorine pesticides atmosphere. PCBs and other compounds (chlorinated
PAHs Polycyclic aromatic hydrocarbons solvents, halogenated methoxybenzenes (halogenated
PCBs Polychlorinated biphenyls
anisoles), polychlorinated naphthalenes, etc.) are
PCDDs Polychlorinated dibenzo-dioxins
PCDFs Polychlorinated dibenzo-furans the most representative chlorinated compounds in
PUFs Polyurethane foam plugs addition to OCPs (alpha and gamma hexachloro-
SEC Size-exclusion chromatography cyclohexane (HCHs), chlorothalonil, dichloSuanid,
SFC Supercritical fluid chromatography toxaphene, chlorpyrifos, alachlor, etc.).
SFE Supercritical fluid extraction
Sampling of these compounds is similar to that
SIM Selected ion monitoring
TLC Thin-layer chromatography described in the previous two sections and either the
UV Ultraviolet adsorbent used or the extraction solvent and post-
VOCs Volatile organic compounds extraction procedures applied are widely documented
in the literature. Thus, the performance of different
sorbents, such as Chromosorb 102, Porapak R,
The adsorbed compounds are extracted by Soxhlet Supelpak-2, Amberlite XAD-2, Amberlite XAD-4,
into different organic solvents such as toluene, ben- Carbonaceous Ambersorb XE-340 and polyurethane
zene, cyclohexane, tetrahydrofuran, dichloromethane foam, has been evaluated with use of atmospheres
and liquid CO2. Clean-up extract procedures are containing known concentrations of OCPs. The most
mainly performed by the use of silica gel (SFC, LLE efRcient trap for HCH and PCBs was found to be
and LC are also used), and the individual separ- two cartridges containing PUFs-Tenax-GC sandwich
ation/determination stage involves HPLC-UV and traps, connected in series. Halogenated anisoles and
GC-MS. A less common technique for PAH deter- hexachlorobenzene (HCB) have been sampled from
mination is TLC-UV. air using a high volume sampling technique in which
air was pumped through two layers of solid sorbent:
Polychlorinated dibenzo-dioxins (PCDDs) and the upper sorbent layer was a mixture of silica gel
polychlorinated dibenzo-furans (PCDFs) Analysis 60/ENVI-Carb or ANGI-Sorb 5B/ANGI-Sorb 10B
of dioxins has become an issue of major importance and the lower sorbent layer was silica gel 60/ENVI-
because of their carcinogenic nature. The main source Carb or ANGI-Sorb 2.5B. These solid sorbents are
of dioxin emission is combustion (in waste inciner- Soxhlet extracted with organic solvents, mainly di-
ation and the steel industry). In comparison with the chloromethane (but also with others such as petrol-
wide variety of methods used for the determination of eum ether, ethyl ether, hexane, acetone and mixtures)
PAHs from air, the method for determination of and, after the appropriate clean up, the detection/
PCDDs and PCDFs is well established and only some determination step is developed mainly by GC-ECD.
minor modiRcations are found in the literature. These
compounds are sampled by splitting the air into two Other pollutants usually determined in air In addi-
phases. The split involves glass Rbre Rlters, which tion to PAHs and chlorinated pollutants from the
retain airborne particulates, and PUF solid sorbents, atmosphere, there are some other families of
which retain vapours. The PUF sorbents are fortiRed pollutants that have a signiRcant effect on the envi-
with a range of 13C-labelled PCDDs and PCDFs be- ronment. Thus, alcohols and phenols, organophos-
fore sampling (sometimes spiking precedes the samp- phorous compounds, anilines and benzidines,
ling step). The same spiked procedure is applied to aliphatic hydrocarbons, etc., can be included in this
classiRcation. The procedures for determination and istics of these compounds makes it mandatory to
the general characteristics of the chemicals and in- study them apart from the rest of the chlorinated
struments used (solid sorbents, extractant, extraction pollutants. Despite the existence of papers on deter-
time, clean up and detection/determination) are sim- mination of dioxins and furans from solid matrices
ilar to those discussed in previous sections. such as soil or sediment, the main source of emission
in the literature is combustion processes. Here, the
Organic pollutants in solid samples Organic pollu- most important solid matrix is Sy ash from inciner-
tants are present in a wide range of environmental ators. However, the procedure followed for Sy ash
solid samples. Thus, soil, sediment, sewage sludge could be applied to other matrices with minor
and ash are the most commonly studied matrices in changes of the sample preparation step. Fly ash sam-
the environmental Reld. All these matrices can be ples are collected at the bottom of the electrostatic
leached by conventional Soxhlet extraction. There precipitators of the incinerator (this is why they are
are two criteria that solid samples must meet before considered as solid pollutants; however, when ashes
extraction: (1) they must be Rnely divided (in order to are in the atmosphere they are considered as air
improve the sample}solvent contact), and (2) sample pollutants). The Sy ash is treated with 1 N HCl,
moisture must be carefully controlled. After the ex- Rltered, washed with H2O and air-dried at room
traction step by an organic solvent, a clean-up pro- temperature. After addition of 13C-labelled PCDD
cedure is necessary due to the co-extraction of both and PCDF internal standards, the sample is Soxhlet-
fat residues and other interferent substances. The extracted with toluene (benezene and hexane/acetone
usual procedures for analytical processing of the most have also been used) and the extract is concentrated,
representative organic pollutants are as outlined then subjected to clean up involving partitioning
below. with concentrated H2SO4 and sequential LC on
multi-layered silica gel (containing acid-modiRed,
Polycyclic aromatic hydrocarbons Soil or sediment base-modiRed, AgNO3-modiRed and neutral forms
samples are dried to constant weight, ground, sieved of silica gel), acid alumina and Celite/Carbopack
and mixed with anhydrous Na2SO4. Soxhlet extrac- stationary phases as described in US EPA Method
tion is performed with dichloromethane (chloro- 8280A. A portion of the resulting solution is analysed
benzene, benzene, cyclohexane and hexane/acetone by high resolution GC-EIMS-SIM.
are less common for this purpose) and the extract is
reduced in a Kuderna}Danish concentrator, a rotary- Polychlorinated biphenyls and other chlorinated
evaporator or under a N2 stream. Clean up (if needed) compounds The large number of chlorinated pollu-
involves: (1) passing the extract through one or more tants in solid samples present in the environment
microcolumns containing silica gel, alumina, C18, makes their classiRcation and study very difRcult.
Florisil, Amberlite XAD-2, etc., and/or (2) fractiona- However, it has been clearly demonstrated that there
tion by semi-preparative normal-phase LC. The are three groups (PCBs, OCPs and polychlorinated
PAHs are determined either by HPLC (with UV, MS phenols), which are the subject of most of the papers
or Suorimetric detection) or GC (with FID or MS found in the literature. Despite that the following
detection). paragraphs are devoted to these groups of organic
Fly ash is Soxhlet extracted with toluene (or pollutants, the procedures described are also applic-
benzene) and the extract is cleaned up by TLC on able to the rest of the chlorinated compounds (parti-
a plate coated with silica gel. Detection/determina- cularly with regard to Soxhlet extraction).
tion involves the same techniques as those used in The Rrst step of the analysis of chlorinated phenolic
soil. compounds in polluted soils is wetting of the sample
Sewage sludge is Soxhlet extracted with toluene. with H2SO4, followed by Soxhlet extraction with
The extract is evaporated and then, liquid}liquid par- hexane/acetone or diethyl ether/petroleum ether. Dif-
titioned between cyclohexane, dichloromethane and ferent techniques can be used for subsequent treat-
H2O. The cyclohexane extract is puriRed on a silica ment of the extract, the most representative being
column. Separation of neutral and polar PAHs is acetylation and LLE. After evaporation, the acetates
performed by HPLC-UV with individual separ- are cleaned up on a deactivated silica gel column and
ation/determination by GC-FID. subjected to GC (with ECD or MS detection). The
chlorophenols are also individually separated/deter-
Polychlorinated dibenzo-dioxins and polychlorinated mined by LC-atmospheric-pressure-CIMS.
dibenzo-furans As could be seen in the previous PCBs are among the most studied pollutants in
section, the extraction of PCDDs and PCDFs is usu- solid samples in recent years. Their properties (resis-
ally performed in a single step. The special character- tance to oxidation, acids, bases, and other chemicals,
high thermal stability, dielectric behaviour, etc.) biochemistry and toxicology and, therefore, beyond
make them the most frequently used organic com- the scope of this article.
pounds and, consequently, the greatest environmental It is obvious that Soxhlet extraction is not a suit-
pollutant. able method for direct separation of analytes from
As the overall procedures followed for the deter- liquid sample, but that it must be used after a pre-
mination of PCBs and OCPs are similar, both groups vious Rltration step. This procedure is very similar to
are usually determined together. Solid sample prep- that followed for air samples. The two phases existing
aration is similar to that shown for the determination in water within suspension matter samples (liquid
of pollutants from solid samples (see PAHs from and particulate) are split by Rlters (paper, glass-Rbre
soils). Soxhlet extraction is performed with either or quartz-Rbre) and solid sorbents (PUFs, granular
a single solvent (toluene or dichloromethane) or activated carbon, Aquapak 440A, Separon SE, ODS
a mixture of solvents such as, hexane/acetone, pen- gel and Amberlite XAD-2, -4 or -7 resins). The solid
tane/dichloromethane, etc. Clean up usually involves sorbents and the particulate matter constitute the
sulfur removal by reaction with Cu or tetrabutyl- solid sample to be subsequently extracted by Soxhlet.
ammonium sulRte and the use of silica gel, Florisil Collection of the solid matter by centrifugation is an
and fractionation by SEC if required. The most usual alternative to Rltration. Some of the procedures found
way for individual separation/determination of PCBs in the literature are discussed below.
and OCPs is GC-ECD. PCBs and OCPs are Soxhlet extracted from solid
particulate matter with dichloromethane. After silica
Other pollutants in solid samples There are some gel or Florisil clean up in the presence of activated Cu,
other pollutants that are frequently determined in analyses are developed by GC-ECD (or GC-MS). The
solid samples. These compounds (in general, the same determination of organophosphorus and organoni-
as those covered in Organic Pollutants in Air) are trogen pesticides in solid particulate matter from sur-
usually in trace amounts in the environment. Thus, face water is developed by dichloromethane Soxhlet
organic pollutants such as aliphatic hydrocarbons, extraction and GC-NPD. The determination of PAHs
aldehydes, non-chlorinated phenols, cresols, anilines from water samples requires hexane/acetone Soxhlet
and benzidines, VOCs (mainly organic solvents), extraction, cleanup on both an Al2O3}Na2SO3 col-
phthalates, amines, non-chlorinated pesticides, etc., umn and silica column, and Rnally, individual separ-
are also determined due to their importance in the ation by HPLC and Suorimetric detection.
environment. Soxhlet extraction has shown its suit- Many aquatic organisms accumulate pollutants in-
ability as a technique for separation of these pollu- side their tissues by bioaccumulation. This behaviour
tants from solid environmental matrices. is used in pollution surveillance programmes, due to
the following advantages it provides: (1) the analyses
Organic Pollutants in Water
of soil, air, water, etc. yield levels of pollutants pres-
Water is the most important liquid in the environ- ent at the time the samples were taken, whereas those
ment. A wide range of human activities, such as observed in bioaccumulator species reSect the level
mining, coal and fuel combustion, industrial and agri- over a period of time; and (2) pollutants concentrate
cultural processes, domestic sewage, etc, contribute in biological species at high levels, and can therefore
to the pollution of the aquatic environment. be monitored by analytical techniques with relatively
However, the key contribution to water pollution is high detection limits.
the solubilization of pollutants. Thus, the study The most relevant Reld in which biological samples
of pollutants in water can be divided into pollutants are used is in the analysis of the aquatic environment.
that are solubilized in water (mainly inorganic pollu- Fish, mussels and a number of other species are
tants) and those present as the solid phases of the studied in order to evaluate pollution levels in the
aquatic medium, such as sludge, sediment, partic- surrounding environment. The pollutants studied
ulates and biological species (organic and inorganic are the same as in previous sections, and Soxhlet
pollutants). extraction has great relevance as a pretreatment tech-
The way in which sludge and sediment are Soxhlet nique due to the solid character of the matrix. In spite
extracted in order to isolate the analytes from the of this, one of the most relevant drawbacks in Soxhlet
solid matrix has been discussed in previous sections. extraction of these samples is their high water content
This section is therefore devoted to the analysis of (the presence of water in samples subjected to Soxhlet
organic pollutants (mainly PCBs and pesticides) from extraction constitutes a shortcoming whose impor-
water and associated particulates. In spite of the fact tance is a function of the amount of water). Thus, it is
that biological samples are also a part of the environ- necessary to macerate the sample initially with an-
ment, they are considered to be a matter concerning hydrous Na2SO4. After this, a conventional Soxhlet
extraction of the mixture with an organic solvent autoclave or by the use of either commercial or labor-
(depending on the nature of the analytes) is per- atory-made supercritical Suid-Soxhlet extractors.
formed. Thus, PAHs and PCBs and be removed from environ-
mental samples using HPS with liquid CO2 in work-
ing conditions close to those corresponding to the
Use of Conventional Soxhlet supercritical state of the extractant. The main draw-
Extraction for the Leaching of back of this alternative is the change from supercriti-
Inorganic Pollutants cal to liquid state of the extractant, which affects
Soxhlet performance.
The application of conventional Soxhlet extraction in
the Reld of inorganic pollutants has been developed to
Automated Soxhlet (Soxtec HT and BuK chi B811)
a much less extent than that of organic pollutants.
There are in the literature only a few applications of Commercial automated Soxhlet devices perform the
Soxhlet extraction to inorganic compounds. Soxhlet extraction with similar precision to conventional Sox-
extraction is used mainly as a clean up step prior to hlet but with a signiRcant saving of time and extrac-
the determination of inorganic pollutants with the tant. The most relevant characteristic of a Soxtec
aim of removing organic substances in the extract. As system is the possibility of developing three different
an exception to this, metals have been determined in steps, namely, boiling, rinsing and recovery of the
the extract after a Soxhlet step. This is possible either solvent, by switching a lever. The B811 extractor is
when metals are forming organometal compounds or able to perform the same steps as a Soxtec device but
are concentrated by sorption on PUFs impregnated it can also work as a conventional Soxhlet. The over-
with an organic substance. However, the suitability all performance of the B811 extractor is computer
of Soxhlet extraction for the isolation of inorganic controlled. The analysis of organic pollutants from
pollutants is poor and some other techniques such as soil and sediments is an example of the methods
hydrolysis, digestion, distillation, etc. are recommen- developed for use with this device.
ded for this purpose.
Focused microwave-assisted Soxhlet extraction
(Soxwave and FMASE)
Improvements in Soxhlet Extraction
The Soxwave is a commercial system with an opera-
Soxhlet extraction has been the most frequently used tional performance similar to that of Soxtec HT. The
technique for isolation of organic pollutants from system performs the same three steps but with two
environmental samples for the last twenty years. signiRcant differences: (1) a different heating source
However, the use of new extraction techniques that (microwave instead of electricity) is used; and (2) the
overcome the drawbacks associated with Soxhlet is, sample is also irradiated with microwave energy,
today, one of the most promising research lines in the making it easier to rupture the analyte}matrix bonds.
Reld of solid sample treatment. The main drawback of Soxwave is its dependence on
The most signiRcant drawbacks of Soxhlet extrac- the extractant dielectric constant. Thus, efRcient ex-
tion are the long time required for the extraction and tractions are only obtained with polar solvents (due
the large amount of organic solvent wasted, which is to the characteristics of microwave irradiation) and,
not only expensive to dispose of but which can cause consequently, this device is not as universal as a
environmental pollution itself. Moreover, the con- conventional Soxhlet is. Moreover, the energy the
ventional device is not easily automated. There are analytes receive is at least as high as that necessary to
two different ways in which to circumvent the draw- reach the boiling point of the solvent, which can
backs of conventional Soxhlet extraction, namely: cause degradation of thermolabile analytes. Some ap-
(1) the use of one of the new alternatives (such as SFE, plications of Soxwave have been developed in the
MAE, ASE, etc.); and (2) the improvement of conven- environmental Reld and, due to its suitability for
tional Soxhlet. As this article is devoted to Soxhlet water-based extractions, quantitative extractions
extraction, the following sections discuss only the have been achieved in the isolation of metals from
latter way, that is, improving the shortcomings of solid samples such as coal or soils.
the conventional device while keeping its positive FMASE is the last of the improvements carried out
characteristics. on the conventional Soxhlet extractor. In fact, the
FMASE device works as a conventional Soxhlet, but
High-pressure Soxhlet extraction
the cartridge zone, where the extractant and sample
High pressure in Soxhlet devices is achieved by plac- are in contact, is placed in the microwave cavity
ing the extractor in a cylindrical stainless-steel of a specially designed focused microwave oven.
Figure 2 Scheme of the FMASE device. (Reproduced with permission from GarcmH a-Ayuso LE and Luque de Castro MD (1999)
Analytica Chimica Acta 382: 309.)
Figure 2 shows a scheme of the FMASE device. The extractant due to its design, as both thermal insula-
similar performance with respect to its conventional tion and shortening of the present distillation
counterpart makes FMASE a suitable alternative for device is mandatory for reception of water vapour on
almost all the applications developed in a con- the sample-cartridge vessel, condensation there and
ventional Soxhlet, that can be developed without dropping on the sample. A comparison between con-
changes, except in the time required for quantita- ventional Soxhlet and FMASE for leaching of PAHs,
tive extraction. herbicides and n-alkanes from soil shows the advant-
FMASE maintains the advantages of conventional ages of the latter when compared with the con-
Soxhlet extraction while overcoming the limitations, ventional design. FMASE provides efRciencies similar
such as the long extraction time, non-quantitative to those obtained by conventional Soxhlet with ex-
extraction of strongly retained analytes and unsuita- traction times at least eight times shorter.
bility for automation. Solvent distillation in FMASE
is achieved by electrical heating, which is independent
of the extractant polarity, and recycling saves
Further Reading
75}85% of the total extractant volume. The main Anderson R (1987) Sample Pre-treatment and Separation.
drawback in FMASE is the difRculty of using water as New York: John Wiley.
III / ENVIRONMENTAL APPLICATIONS / Supercritical Fluid Extraction 2709
Ballschmeter K (1993) Sample treatment techniques for or- Poole CF and Poole SK (1996) Trends in extraction of
ganic trace analysis. Pure Applied Chemistry 55: 1943. semivolatile compounds from solids for environmental
FiReld FW and Haines PJ (1995) Environmental Analytical analysis. Analytical Communications 33: 11H.
Chemistry. London: Chapman and Hall. Poole SK, Dean TA, Oudsema JW and Poole CF (1990)
Luque de Castro MD and Da Silva MP (1997) Strategies for Sample preparation for chromatographic separations:
solid sample pretreatment. Trends in Analytical Chem- an overview. Analytica Chimica Acta 236: 3.
istry 16: 16. Prichard E, MacKay GM and Points J (1996) Trace Analy-
Luque de Castro MD and Garcia-Ayuso LE (1998) Soxhlet sis: A Structured Approach to Obtaining Reliable Re-
extraction of solid materials: an outdated technique with sults. Cambridge: The Royal Society of Chemistry.
a promising innovative future. Analytica Chimica Acta Warner PO (1976) Analysis of Air Pollutants. New York:
369: 1. John Wiley.
Supercritical Fluid Extraction

V. Camel, Institut National Agronomique water can be detrimental to the extraction of nonpo-
Paris-Grignon, Paris, France lar compounds. Consequently, matrices with a high
Copyright ^ 2000 Academic Press water content (typically 75%) require the addition of
a drying agent to the sample (e.g. hydromatrix, a pel-
There has been growing interest in supercritical Suid letized diatomaceous earth, magnesium or sodium
extraction (SFE) in the past few years due to its sulfate). This also enlarges the surface area of the
numerous advantages over liquid extraction (rapid- sample. However, the presence of residual water usu-
ity, low solvent volumes, nontoxicity of carbon diox- ally favours the extraction of polar compounds.
ide, great selectivity by modifying the Suid density, Grinding the sample should also enhance the ex-
low dilution of the extracts, possibility of online traction but, excessive grinding may lead to a pres-
coupling with chromatographic techniques and auto- sure drop within the extraction cell, thereby decreas-
mation). ing the solubility of the analyte at the bottom of the
Analytical applications of SFE began in the late cell. The pressure drop problem may be overcome by
1980s, with particular focus on environmental sam- mixing the Rnely ground sample with a coarse dis-
ples. While early reports were on spiked matrices persing agent.
and/or highly contaminated samples, recent applica- Finally, the presence of sulfur in some matrices (e.g.
tions deal with samples containing low levels of in- sediments or sewage sludges) can cause lack of repro-
curred contaminants. It was soon found that extrac- ducibility and restrictor blockages. To overcome
tion conditions are strongly dependent on both the these problems, it is suggested to mix the sample with
solutes and the matrix, so that parameters need to be copper prior to extraction to act as sulfur scavenger.
adjusted for every new application.
This article will focus on the main pollutants ex- Liquids A few studies have been performed on the
tracted, showing the important parameters that inSu- direct SFE of aqueous matrices using a special extrac-
ence extraction recoveries, and illustrating the great tion vessel. However, such analytes are mainly pre-
potential of this technique together with its limitations. concentrated on to a solid-phase extraction (SPE)
disk or cartridge, before being eluted with the super-
critical Suid. This SPE-SFE combination offers
Sample Preparation Prior to a greater selectivity compared to elution with an
Extraction organic solvent (e.g. CO2 at low density selectively
To ensure better desorption of analytes from the extracts organochlorine pesticides from C18 disks,
matrix, several sample treatments can be performed, while extraction at a higher density in the presence of
either physical (e.g. grinding) or chemical (e.g. addi- methanol is required to elute organophosphorus
tion of derivatization reagents). pesticides).
Pretreatment of the Sample Derivatization Reactions
This step is of prime importance, as it may greatly Extraction of highly polar compounds may be im-
enhance the extraction efRciency. proved by coupling derivatization reactions with SFE,
to convert polar functions into less polar groups for
Solids The moderate water solubility in supercriti- better solubility in the Suid. This procedure affords
cal CO2 may lead to restrictor plugging; in addition, extracted compounds that are readily amenable to
2710 III / ENVIRONMENTAL APPLICATIONS / Supercritical Fluid Extraction
gas chromatography. Besides, the derivatizing agent PAH solubility in supercritical CO2 decreases with
may react with active sites of the matrix, leading to increasing number of fused aromatic rings, so addi-
better desorption of solutes. tion of a modiRer is recommended to achieve accept-
The three main reactions are alkylation (with able recoveries. Methanol is the most common modi-
acidic methanol, alkyl halides, tetraalkylammonium Rer, but satisfactory results can be obtained with
salts or Grignard reagents), acylation (mostly with other modiRers. For example, toluene-modiRed CO2
acetic anhydride, in the presence of organic bases is efRcient in extracting two to six fused aromatic ring
such as pyridine) and silylation (with hexamethyl- PAHs from soil with high carbon content (50%);
disilazane and trimethylchlorosilane). As the latter addition of toluene to the sample also improves the
requires relatively anhydrous conditions, matrices extraction of nitro-PAHs from diesel and air partic-
with moisture contents greater than 0.4% may reduce ulates. A combination of toluene, triSuoroacetic acid
derivatization efRciency. and triethylamine is an even better modiRer for PAHs
Derivatization may be performed prior to extrac- and nitro-PAHs; the additives are thought to block
tion or under supercritical Suid conditions (in situ the matrix active sites, thus preventing possible read-
derivatization). The latter approach is most common sorption of solute molecules. Extraction of PAHs
as it reduces sample handling. Pre-extraction derivat- from air particulate matter is also improved in the
ization is used for particular applications (e.g. alkyla- presence of diethylamine or acetonitrile, as illustrated
tion with Grignard reagents due to their low solubil- in Figure 2.
ity in CO2). As complex environmental matrices con- The efRcacy of the modiRer is highly dependent on
tain many potential interferences that can be de- matrix characteristics. For example, the addition of
rivatized, excess quantities of reagent should be used. methylene chloride as a static modiRer allowed the
quantitative extraction of PAHs from soil with CO2;
Ion Pairing this modiRer solubilizes the soil aggregates, thus in-
SFE of ionic compounds may be possible by forma- creasing the contact between soil particles and super-
tion of an ion pair, which is soluble in the Suid. The critical CO2.
ion-pairing reagent may also react with the matrix The effect of increasing the temperature is always
active sites, thus favouring the desorption of solute advantageous at constant density. ModiRer and tem-
molecules. perature effects are additive, so that extraction using
CO2 with modiRers at high temperature is usually the
most rigorous SFE method for the extraction of parti-
Common Pollutants Extracted by SFE cularly difRcult samples such as urban air partic-
ulates, as shown in Figure 3.
SFE has been successfully applied to the determina- Another approach has been the use of in situ chem-
tion of several pollutants from different matrices. The ical derivatization to determine PAHs from a harbour
strong solute}matrix interactions usually impose sediment. The derivatizing agent (Tri-Sil, a 2:1 (v/v)
proper modiRer selection and elevated temperature. mixture of hexamethyldisilazane and trimethyl-
Typical extraction conditions for the main pollutants chlorosilane) was added to the extraction vessel prior
are given in Table 1. to the extraction step. As shown in Figure 4, results
were improved compared with extraction with CO2
Polynuclear Aromatic Hydrocarbons
or 10% methanol-modiRed CO2. As PAHs cannot be
Polynuclear aromatic hydrocarbons (PAHs) have derivatized, the effect of the reagent was to help
commonly been extracted from environmental ma- displacement of the analytes from the matrix.
trices, and SFE has recently been adopted as an ofR- Although nitrous oxide modiRed with 5% meth-
cial method (US Environmental Protection Agency anol may be considered to be the most efRcient Suid
method 3561). for extracting PAHs, its use should be avoided due to
These analytes are relatively nonpolar and should the possibility of explosion with this combination.
be extracted with neat CO2. However, the delocalized DiSuorochloromethane is also efRcient but environ-
-electron system of PAHs can cause strong interac- mental concern discourages its common use. Subcriti-
tions with the active sites of the matrix surface, hin- cal water (2503C, 50 bar) seems a more viable alter-
dering their extraction. Extraction of high molecular native to CO2 for the SFE of organic compounds with
weight PAHs from real samples therefore requires a wide range of polarities. The dielectric constant of
high pressure and temperature, as illustrated in Fig- water decreases as the temperature increases so that
ure 1 for urban dust particulates. Elevated temper- at moderate temperatures (50}1003C) polar com-
ature are suspected to favour analyte desorption from pounds are extracted (e.g. phenols, amines), while
the active sites of the matrix. nonpolar to moderately polar organics (including
Table 1 Typical applications of SFE to solid environmental matrices
Compounds Matrices Reagent added to Fluid Observations

the matrix
Polyaromatic Soils, sediments, None CO2}CH3OH (10%) Strong interactions

hydrocarbons urban dust, fly ash CO2}toluene (10%) with the matrix
CO2}diethylamine High temperatures
(10%) recommended,
Subcritical H2O as well as
with addition of
modifier
CH2Cl2 CO2 Solubilization of soil
aggregates by
CH2Cl2
Tri-Sil CO2 Tri-Sil displaces the
solutes from the
matrix
Polychlorinated Soils, sediments, None CO2 Moderate
biphenyls sewage sludge CO2}CH3OH (1}2%) temperatures
CO2}diethylamine (70}1003C)
Subcritical H2O
Dioxins Fly ash, sediments None CO2}CH3OH (2%) Strong interactions
with the matrix
Better recoveries
with a dry matrix
Strong acid CO2 Destruction of the
matrix by the acid
Phenols Soils, house dust None CO2}CH3OH (2I20%) High temperatures
CO2}CH3OH (32%)I recommended
H2O (8%)
Acetic anhydride and CO2 In situ acetylation of
pyridine phenols
Pesticides Organochlorine Soils, sediments None CO2}toluene Toluene disrupts
solute}matrix
interactions
Organophosphorus None CO2}CH3OH (5%)
Triazines None CO2}CH3OH (10}30%) Matrix moisture
CO2}[CH3OH#2%H2O] enhances the
(10%) extraction
Phenoxyacetic None CO2}CH3OH (20%)
acids
TMPA CO2 Ion pairing and
methylation
BF3/CH3OH CO2 Methylation
Surfactants Nonionic Soils, sediments, None CO2}CH3OH (27.5%)
sewage sludges H2O CO2
Anionic None CO2}CH3OH (40%)
TAA salts CO2 Ion pairing
Methylation reagent CO2 Methylation
Cationic None CO2}CH3OH (30%)
Metallic species Organometallics Sediments None CO2}CH3OH (10}20%)
Organic liganda CO2}CH3OH (5%)
Derivatization reagentb
Metal ions Sediments Organic ligandc CO2}CH3OH (5%) Formation of a metal
chelate
Methanol increases
the chelate solubility
TMPA, trimethylphenylammonium hydroxide; TAA, tetraalkylammonium.

a
Organic ligands: dithiocarbamates (mainly sodium diethyldithiocarbamate and diethylammonium diethyldithiocarbamate).
b
Derivatization reagents: hexylmagnesium bromide, thioglycolic acid methylester.
c
Organic ligands: dithiocarbamates, -diketones, crown ethers, organophosphorus compounds.
Figure 1 Recoveries of PAHs from urban air particulates (standard reference material SRM 1649) using supercritical CO2a extraction
at different pressures and temperatures. a40-min extractions. (From Langenfeld JJ, Hawthorne SB, Miller DJ and Pawliszyn J (1993)
Effects of temperature and pressure on supercritical fluid extraction efficiencies of polycyclic aromatic hydrocarbons and poly-
chlorinated biphenyls. Analytical Chemistry 65: 338I344. Copyright 1993 American Chemical Society.)
Figure 2 Influence of the presence of a modifier in supercritical CO2 on the recoveries of PAHs from air particulate matter (standard
reference material SRM 1649)a. a400 bar, 803C, 250 L (10% v/v) modifier added to the sample, 5 min static/10 min dynamic. (From
Langenfeld JJ, Hawthorne SB, Miller DJ and Pawliszyn J (1994) Role of modifiers for analytical-scale supercritical fluid extraction of
environmental samples. Analytical Chemistry 66: 909I916. Copyright 1994 American Chemical Society.)
Figure 3 Temperature effect on the recoveries of PAHs from air particulate matter (standard reference material SRM 1649) using
methanol modified CO2a. a400 bar, 80 L (10% v/v) methanol added to the sample, 15 min static/15 min dynamic. (From Yang Y,
Gharaibeh A, Hawthorne SB and Miller DJ (1995) Combined temperature/modifier effects on supercritical CO2 extraction efficiencies of
polycyclic aromatic hydrocarbons from environmental samples. Analytical Chemistry 67: 641I646. Copyright 1995 American Chemical
Society.)
Figure 4 Recoveries of PAHs from a harbour sediment (HS-3) using either supercritical CO2, 10% methanol modified CO2 , or in situ
derivatization with Tri-Sila followed by CO2 extraction. Three sequential extractions were conducted (each at 350 bar and 603C, 15 min
static/15 min dynamic). a Tri-Sil is a mixture of hexamethyldisilane and trimethylchlorosilane 2 : 1 (v/v); 0.5 mL of this reagent was added
to the cell prior to the extraction. The derivatizing agent was added just prior to each static step. (From Hills JW and Hill HH (1993)
Carbon dioxide supercritical fluid extraction with a reactive solvent modifier for the determination of polycyclic aromatic hydrocarbons.
Journal of Chromatographic Science 31: 6I12. With permission of Preston Publications, a Division of Preston Industries Inc.)
PAHs) are extracted at higher temperatures Recently, a single SFE method for Reld extraction
(200}2503C), as illustrated in Figure 5. of PCBs and PAHs in soils has been developed with
PAHs have also been determined in water samples neat CO2 (1503C, 400 bar) to avoid co-extraction of
after their preconcentration on to C18 disks and their matrix material, allowing direct gas chromatography.
further elution with supercritical CO2. The simultaneous extraction and clean-up of mussel
samples can be achieved by adding Florisil on top of
Polychlorinated Biphenyls
the sample, enabling the direct determination of 11
Polychlorinated biphenyls (PCBs) are lipophilic and PCBs (as well as 15 organochlorine pesticides).
thereby highly soluble in CO2. Hence, CO2 alone or
Dioxins
modiRed with a small amount of methanol (typically
1}2%) is efRcient for their extraction. Figure 6 illus- Dioxins (polychlorinated dibenzo-p-dioxins (PCDDs)
trates the effects of both pressure and temperature on and polychlorinated dibenzofurans (PCDFs)) are of
the recovery of PCBs from river sediment using neat great environmental concern owing to their acute
CO2. Best recoveries are obtained at high temper- toxicity. Like PCBs, they are readily amenable to
ature, whatever the pressure. Surprisingly, high mo- extraction by SFE. As these pollutants have been
lecular weight PCBs are more efRciently extracted, mainly detected in emissions from municipal inciner-
despite their expected reduced solubilities; in fact, ators, their extraction from Sy ash matrices has been
this is in agreement with the tighter binding of low investigated. SFE gave satisfactory results, as com-
molecular weight PCBs to the sediment matrix. pared to Soxhlet extraction. Despite the high solubil-
Recovery rates may be improved by addition of ity of these compounds in pure CO2, it gave almost no
modiRers, especially methanol, as illustrated in Fig- extraction due to the strong matrix adsorption of the
ure 7. Thus, methanol-modiRed CO2 allowed the SFE dioxins. Upon addition of 2% methanol to the CO2,
of PCBs and organochlorine pesticides at the part- 2,3,7,8-tetrachlorodibenzo-p-dioxin was efRciently
per-trillion level in marine sediments. Water under extracted from a dry sediment; the presence of water
subcritical conditions is also an effective extractant in the matrix hindered its extraction. The suitability
for PCBs. of nitrous oxide for the SFE of PCDDs and PCDFs
Figure 5 Recoveries of PAHs from urban air particulate (National Institute for Standards and Technology NIST 1649) using either
supercritical CO2a (2003C, 659 bar), 10% toluene modified CO2b (803C, 405 bar) or subcritical water (2503C, 50 bar). a 40-min
extractions. b 20-min extractions (10 min static/10 min dynamic). (From Hawthorne SB, Yang Y and Miller DJ (1994) Extraction of
organic pollutants from environmental solids with sub- and supercritical water. Analytical Chemistry 66: 2912I2920. Copyright 1994
American Chemical Society.)
Figure 6 Recoveries of PCBs from river sediment (standard reference material SRM 1939, containing 3% water and 10% organic
matter) using supercritical CO2a extraction at different pressures and temperatures. a 40-min extractions. (From Langenfeld JJ,
Hawthorne SB, Miller DJ and Pawliszyn J (1993) Effects of temperature and pressure on supercritical fluid extraction efficiencies of
polycyclic aromatic hydrocarbons and polychlorinated biphenyls. Analytical Chemistry 65: 338I344. Copyright 1993 American
Chemical Society.)
Figure 7 Influence of the presence of a modifier in supercritical CO2 on the recoveries of PCBs from river sediment (standard
reference material SRM 1941)a. a400 bar, 803C, 250 L (10% v/v) modifier added to the sample, 5 min static/10 min dynamic. (From
Langenfeld JJ, Hawthorne SB, Miller DJ and Pawliszyn J (1994) Role of modifiers for analytical-scale supercritical fluid extraction of
environmental samples. Analytical Chemistry 66: 909I916. Copyright 1994 American Chemical Society.)
has also been described. Alternatively, the matrix the soil increases due to stronger solute}matrix inter-
may be destroyed by exposure to a strong acid, and actions, improvement upon derivatization is evident.
further extracted with neat CO2. In addition, extracts are cleaner because of milder
extraction conditions.
Phenols
Similarly, the SFE of pentachlorophenol and re-
Phenols are moderate to highly polar compounds. lated compounds from soil samples can be achieved
Thus, several approaches have been used for their using in situ acetylation (with acetic anhydride and
SFE: direct addition of a polar modiRer (e.g. triethylamine at 803C) followed by CO2 extraction.
water, acetonitrile, methanol) to the matrix, dynamic Increasing the extraction temperature from 50 to
addition of the modiRer to the CO2, or in situ acety- 2003C resulted in higher recoveries of chlorophenols
lation. from an industrial soil.
For example, addition of 2% methanol improved
Pesticides
their extraction from soil. Enhanced-Suidity liquid
extraction (CO2 with 20% methanol) improved the Pesticides have a broad range of physical properties
recovery of phenols from house dust. Further im- and chemical structures. Their solubility in pure CO2
provements could be achieved using a methanol} may be evaluated from their octanol}water partition
water}CO2 mixture (32.1/7.9/60 mol %), as water is coefRcients. Organochlorine pesticides are highly sol-
supposed to swell the matrix material, allowing more uble in pure CO2, while organophosphorous com-
efRcient penetration and interruption of matrix} pounds require a modiRer; addition of a polar modi-
analyte interactions. Similar results have been ob- Rer becomes crucial for triazines. In the case of
tained with sediments. phenoxyacetic acids, an ion-pairing or derivatization
As illustrated in Figure 8, phenols are efRciently reagent needs to be added.
acetylated during SFE by means of direct reaction Another useful parameter is the soil}water parti-
with acetic anhydride. Even though recoveries of tion coefRcient, as it is indicative of the pesticide’s soil
phenols decrease as the activated charcoal content of adsorption; recoveries have been shown to decrease
Figure 8 Recoveries of phenolic compounds from three garden soils with 2, 5 and 10% activated charcoal content, using SFE alonea
or SFE-derivatizationb. aCO2, 903C, 382 bar, 0.77 g mL\1, addition of 100 L methanol to the sample, 10 min static/15 min dynamicb.
CO2, 1153C, 0.4 g mL\1, addition of 20 L pyridine and 115 L acetic anhydride to the sample, 5 min static/15 min dynamic. (From
Llompart MP, Lorenzo RA, Cela R, Li K, Belanger JMR and Pare JRJ (1997) Evaluation of supercritical fluid extraction, microwave-
assisted extraction and sonication in the determination of some phenolic compounds from various soil matrices. Journal of Chromatog-
raphy A 774:243I251. With permission from Elsevier Science.)
as the soil organic content is increased, due to strong

analyte}matrix interactions.
Nonpolar pesticides Despite their high solubility in

CO2, the solute}matrix interactions may yield lower
recoveries than expected from solubility alone. Thus,
extractions of organochlorine pesticides from spiked
soils were unsatisfactory, especially for soils with
a high organic content. ModiRers have been tested to
enhance the SFE of organochlorine pesticides from
different matrices. For example, toluene added to
a contaminated soil improved the extraction with
CO2 of hexachlorocyclohexane isomers.
Several organochlorine pesticides were also efR-
ciently extracted from aqueous matrices using a com- Figure 9 Extraction of native pesticides from an agricultural soil
using either classical extraction (two sequential extractions with
bination of SPE (on to C18 disks and cartridges) and 0.5 mol L\1 KOH in 10% KCl/water) or SFE (with TMPA (three
pure CO2 SFE. sequential extractions (each 15 min static/15 min dynamic), CO2,
400 bar, 803C) or BF3}methanol (a single extraction (15 min
Moderately polar and polar pesticides Methanol- static/15 min dynamic), CO2, 400 bar, 803C) as derivatization
modiRed (5%) CO2 is a common extractant for or- reagents). Open bars, 2,4-D; filled bars, dicamba. (Reproduced
with permission from Hawthorne et al., 1992. Copyright 1992
ganophosphorus pesticides. The polar triazine herbi- American Chemical Society.)
cides require a higher percentage of methanol (10%)
in CO2 to increase their solubility and disrupt sol-
ute}matrix interactions. Methanol (10%) containing The ion-pairing methylating reagent TMPA facilit-
2% water may also be used. The ternary mixture ated the CO2 extraction of 2,4-D and dicamba from
acetone}water}triethylamine (90/10/1.5 v/v/v) is also sediments. The presence of methyl iodide improved
efRcient. Water is suspected of increasing the surface the recoveries of 2,4-D and 2,4,5-T from soil. Alkyla-
area of clay containing soils by swelling. Thus, a dition with methanol and BF3 as a catalyst is also
rect correlation between diuron extraction from mon- promising for the preferential extraction of 2,4-D
tmorillonite clay and the percentage swelling of the over dicamba, as illustrated in Figure 9.
matrix (due to the modiRer) was observed at different
Surfactants
pressures and temperatures. In contrast, triethyl-
amine should compete with solute molecules on to Nonionic surfactants (e.g. alkylphenolethoxylates)
the active sites of the soil. Also, soil moisture has been have been extracted from sediments with 27.5%
reported to enhance triazine extraction, as well as methanol-modiRed CO2 (450 bar and 1003C). Water-
addition of a surfactant (Triton X-100), which prob- modiRed CO2 (350 bar and 803C) was also efRcient in
ably leads to the matrix swelling and the formation of extracting nonylphenol polyethoxylates from dried
nonionic reverse micelles. sewage sludge, yielding recoveries higher than tradi-
Extraction of bound pesticide residues in soils may tional techniques.
require more severe conditions (for example, extrac- Quantitative extraction of anionic surfactants (lin-
tion of triazine from a mineral soil entailed 30% ear alkylbenzenesulfonates) from soil, sediment and
methanol, 350 bar and 1253C). sludge can be obtained using 40% methanol-modiRed
CO2 (380 bar and 1253C). Also, TAA salts have been
Ionic pesticides SpeciRc SFE conditions are required used as ion-pairing reagents to extract linear alkyl-
to improve their solubility and/or to overcome strong benzenesulfonates and linear alkylsulfonates from
solute}matrix interactions. Addition of 20% methanol sewage sludge. Finally, derivatization of anionic sur-
to CO2 allowed the extraction of four herbicides factants into their methyl esters may also be per-
(dicamba, 2,4-dichlorophenoxyacetic acid (2,4-D), 2- formed to enhance their extraction.
(2,4,5-trichlorophenoxy)propionic acid (2,4,5-TP) and The ditallowdimethylammonium cation was ex-
2,4,5-trichlorophenoxyacetic acid (2,4,5-T)) from tracted from anaerobically stabilized sewage sludge
house dust (440 bar and 100 or 1503C). Pre-extraction and marine sediment using 30% methanol-modiRed
of extraneous matrix material was achieved with CO2 CO2 (380 bar, 1003C). No improvement could be
after hexane was added to the sample. The mixture obtained in the presence of ion pair reagents. Due to
acetone}water}triethylamine (90/10/1.5 v/v/v) also the ionic character of this surfactant, it was assumed
enhanced extraction of 2,4-D from soil. that high concentrations of anionic surfactants ini-
tially present in the matrix allowed the formation of

ion pairs with the cationic surfactant, thereby en-
abling its extraction.
Surfactants (e.g. alcohol phenol ethoxylate) were
also extracted from aqueous samples with either
direct SFE (by means of a modiRed extraction cell)
or SPE-SFE. In the latter case, methanol-modiRed
CO2 was required for efRcient elution from the
C18 discs.
Metallic Compounds
Metals exist in the environment as organometallic
compounds, ionic species or inorganic compounds.
Organometallic compounds are usually soluble in
supercritical Suids and may be extracted directly. Figure 10 Comparison of extraction efficiencies of organotin
On the other hand, ionic species require the addition compounds from a reference sediment (PACS-1) using SFE with
or without the addition of sodium diethyldithiocarbamate
of a ligand to be extracted. Consequently, speciation
(NaDCC). Open bars, certified value; grey bars, no NaDCC; black
can be achieved by SFE using sequential extractions bars, with NaDCC. (Reproduced with permission from Chau et al.,
(with proper selection of ligands). As an example, 1995, and with permission from Elsevier Science.)
methylmercuric chloride and dimethylmercury could
be extracted with neat CO2 and 100 bar (503C)
from solid materials; a dithiocarbamate reagent was Organotins in aqueous matrices were ethylated
further introduced into the matrix to extract Hg2# with sodium tetraethylborate, enriched on C18 disks
ions. and further extracted with acid-modiRed supercritical
CO2. Alternatively, aqueous matrices containing
Organometallic compounds Several studies have in- butyl-, phenyl- and cyclohexyltin compounds were
vestigated the use of SFE for extracting organometal- collected on a C18 disk, before being derivatized via
lic compounds from environmental samples. Thus, Grignard ethylation and extracted using supercritical
tributyltin has been successfully extracted from sedi- CO2.
ments using methanol (20% v/v)-modiRed supercriti-
cal CO2. Methanol as a modiRer provided the most Metal ions Extraction of free metal ions by super-
favourable recovery of trimethyllead, triethyllead and critical CO2 requires charge neutralization. This can
diethyllead from sediment and urban dust, as com- be achieved by binding the metal ion to organic
pared to water and acetone (446 bar and 803C, 10% ligands, thereby resulting in neutral stable complexes
modiRer). that are soluble in CO2. Obviously, rapid complexa-
Other approaches have been tested: binding to an tion kinetics and a high stability constant for the
organic ligand, formation of an ion pair and in situ neutral complex will enhance the extraction process.
derivatization. Thus, with the addition of diethylam- A key factor is the solubility of the complex in the
monium diethyldithiocarbamate as a ligand, di-, tri- supercritical Suid.
and tetra-substituted organotin species could be ex- Different organic ligands (dithiocarbamates, -
tracted from soils and sediments using 5% methanol- diketones, crown ethers and organophosphorus re-
modiRed CO2 (while recoveries of monoalkyltins reagents) have been tested for their ability to extract
mained low). Monobutyltin was efRciently extracted heavy metals, lanthanides and actinides from several
from a reference sediment on addition of sodium matrices. In particular, Suorinated ligands yield metal
diethyldithiocarbamate in the extraction vessel (Fig- complexes with higher solubility in supercritical CO2,
ure 10); increased extraction efRciencies of trimethyl- making them more effective for the extraction of
lead, trimethyltin, dibutyltin and tributyltin were also metal ions. In addition, alkyl substitutions in ligands
observed in this way. may enhance the solubility of metal complexes in
Hexylmagnesium bromide as a derivatizing agent CO2. As an example, diethyldithiocarbamates and
assisted the CO2 extraction of monophenyltin, Suorinated -diketones are effective chelating agents
diphenyltin and triphenyltin from sediment. Di- for extracting transition metal ions from solid ma-
methylarsenic acid and monomethylarsenic acid trices, as shown in Figure 11 for Cd2#. Recoveries
could be extracted from a solid sample by supercriti- are improved with methanol (5%)-modiRed CO2, as
cal CO2 after in situ derivatization with thioglycolic the solubility of the metal chelate is enhanced. As
acid methyl ester. dithiocarbamates tend to decompose in supercritical
Addition of a proton-ionizable crown ether (tert-

butyl-substituted dibenzobistriazolocrown ether) in
methanol (5%)-modiRed CO2 allowed the selective
extraction of Hg2# from sand if a small amount of
water was present in the matrix (200 bar and 603C).
Other divalent metal ions (Cd2#, Pb2#, Co2#, Mn2#,
Ni2#) remained in the sand under these conditions.
Very recently, Suorinated hydroxamic acids have
been used for the SFE of Fe(III) with unmodiRed CO2.
Toxic metals (As, Cd, Cr, Cu, Pb) have been extrac-
ted from real contaminated soil and wood samples
using the Cyanex reagent (bis(2,4,4-trimethylpentyl)-
monothiophosphinic acid) as an extractant in super-
critical CO2 modiRed with 5% methanol.
Up to the present, most of the experiments conduc-
ted have focused on spiked samples and future studies
need to be conducted with real environmental sam-
Figure 11 Recoveries of Cd2# from spiked sand samples us- ples. In such samples, the active sites and natural
ing SFE with dithiocarbamates or -diketones as chelating ligands present may bind strongly to certain metal
agents. Open bars, CO2; filled bars, 5% methanol-modified CO2; ions, thereby hindering their complexation with ad-
453C, 250 bar; 15 min static/15 min dynamic. LiFDDC; bis(tri-
ded ligands. Native metals can also be in highly insol-
fluoroethyl)dithiocarbamate; Et2NH2DDC; diethylammonium di-
ethyldithiocarbamate; NaDDC; sodium diethyldithiocarbamate; uble forms (such as oxides and sulRdes), leading to
APDC, ammonium pyrrolidinedithiocarbamate; TFA, trifluoro- a fraction of the metals that may not be extractable by
acetylacetone; TTA, thenoyltrifluoroacetone; HFA, hexa- SFE. It seems that SFE may be used to evaluate the
fluoroacetylacetone. (Reproduced with permission from Wai amounts of leachable metals in solid matrices.
et al., 1996, and with permission from Elsevier Science.) Metal ions have also been directly extracted from
aqueous samples. First, the supercritical CO2 is
passed through a vessel Rlled with the ligand. Next,
CO2 in the presence of water, an excess of reagent is the Suid saturated with the ligand passes through
recommended to achieve good metal extraction ef- the aqueous phase. For example, CO2 containing
Rciencies. thenoyltriSuoroacetone and tributyl phosphate
Figure 12 Comparison of extraction techniques for the determination of PAHs from contaminated soil: Soxhlet extraction (10 g
sample mixed with 10 g anhydrous sodium sulfate; 150 mL dichloromethane; heating 24 h), SFE (2 g sample; 20% methanol-modified
CO2; 703C, 250 kg cm\2, 30 min), atomospheric microwave-assisted extraction (MAE) (2 g sample; 70 mL dichloromethane; heat
297 W; 20 min) and accelerated solvent extraction (ASE) (7 g sample; dichloromethane}acetone 1:1 (v/v); 1003C, 2000 psi; 10 min).
(Reproduced with permission from Saim et al., 1997, and with permission of Elsevier Science.)
extracted lanthanides (La3#, Eu3# and Lu3#) from Hawthorne SB, Yang Y and Miller DJ (1994) Extraction of
a buffered acetate solution. organic pollutants from environmental solids with sub-
and supercritical water. Analytical Chemistry 66:
2912}2920.
Future Trends Hills JW and Hill HH (1993) Carbon dioxide super-
Despite rapid growth in the past few years, SFE is still critical Suid extraction with a reactive solvent modiRer
rarely used for routine applications. This is mainly for the determination of polycyclic aromatic hydro-
carbons. Journal of Chromatographic Science 31:
because of the large number of parameters to control,
6}12.
as well as the inSuence of the matrix. The strong Langenfeld JJ, Hawthorne SB, Miller DJ and Pawliszyn
matrix}analyte interactions that may occur in envir- J (1993) Effects of temperature and pressure on super-
onmental matrices frequently make the development critical Suid extraction efRciencies of polycyclic aro-
of quantitative extraction conditions based solely on matic hydrocarbons and polychlorinated biphenyls.
solubility considerations and spike recoveries invalid Analytical Chemistry 65: 338}344.
for real samples. In addition, its development is also Langenfeld JJ, Hawthorne SB, Miller DJ and Pawliszyn
limited by the high capital cost required. J (1994) Role of modiRers for analytical-scale supercriti-
Yet SFE has several advantages over other tech- cal Suid extraction of environmental samples. Analytical
niques (especially rapidity and low solvent volumes, Chemistry 66: 909}916.
as shown in Figure 12). It is successful in extracting Lee ML and Markides KE (eds) (1990) Analytical Super-
critical Fluid Chromatography and Extraction. Provo,
a broad range of pollutants from numerous matrices.
UT: Chromatography Conferences.
In particular, polar compounds and ionic species can Llompart MP, Lorenzo RA, Cela R et al. (1997) Evaluation
be extracted through addition of a polar modiRer, of supercritical Suid extraction, microwave-assisted ex-
a derivatization reagent, an ion-pairing reagent or traction and sonication in the determination of some
a ligand. These recent applications will be more thor- phenolic compounds from various soil matrices. Journal
oughly studied in the next few years, especially the of Chromatography A 774: 243}251.
possible speciation, due to selective extraction, of Luque de Castro MD, Valcarcel M and Tena MT (1994)
metallic compounds. Analytical Supercritical Fluid Extraction. New York,
Subcritical water appears to be a very promising Springer-Verlag.
Suid, as it offers the opportunity to extract polar to McHugh M and Krukonis V (1994) Supercritical Fluid
nonpolar compounds by simply increasing the tem- Extraction, 2nd edn. Boston, MA: Butterworths.
Saim N, Dean JR, Abdullah MDP and Zakaria Z (1997)
perature. No doubt this Suid will be more common in
Extraction of polycyclic aromatic hydrocarbons from
future SFE applications. contaminated soil using Soxhlet extraction, pressur-
Finally, extraction conditions will be optimized for ised and atmospheric microwave-assisted extraction,
numerous real environmental samples, including cer- supercritical Suid extraction and accelerated solvent
tiRed reference materials, thereby leading to wider extraction. Journal of Chromatography A 791:
use of this technique. 361}366.
Taylor LT (1996) Supercritical Fluid Extraction. New
See also: II/Chromatography: Gas: Derivatization. York: Wiley-Interscience.
Chromatography: Liquid: Ion Pair Liquid Chromatogra- Wai CM, Wang S, Liu Y, Lopez-Avila V and Beckert WF
phy; Mechanisms: Ion Chromatography. Extraction: (1996) Evaluation of dithiocarbamates and -diketones
Solid-Phase Extraction; Solvent Based Separation. as chelating agents in supercritical Suid extraction of
III/Metal Complexes: Ion Chromatography. Cd, Pb, and Hg from solid samples. Talanta 43:
2083}2091.
Westwood SA (ed.) (1992) Supercritical Fluid Extrac-
Further Reading tion and its use in Chromatographic Sample Pre-
Chau YK, Yang F and Brown M (1995) Supercritical Suid paration. Glasgow: Blackie Academic and Pro-
extraction of butyltin compounds from sediment. Ana- fessional.
lytica Chimica Acta 304: 85}89. Yang Y, Gharaibeh A, Hawthorne SB and Miller DJ (1995)
Hawthorne SB, Miller DJ, Nivens DE and White DC (1992) Combined temperature/modiRer effects on supercritical
Supercritical Suid extraction of polar analytes using in CO2 extraction efRciencies of polycyclic aromatic hy-
situ chemical derivatization. Analytical Chemistry 64: drocarbons from environmental samples. Analytical
405}412. Chemistry 67: 641}646.
III / ENZYMES / Chromatography 2721
ENZYMES
1. Preparation of reaction mixture and enzyme,

Chromatography where the reaction mixture usually consists of a
controlled substrate solution with the correct
S. Nilsson and S. Santesson,
temperature, pH and any cofactors needed for
University of Lund, Lund, Sweden catalysis. Enzymes often demand a more complex
preparation methodology than substrates and
This article is reproduced from Encyclopedia of Analyti- the reaction mixture. This will be discussed
cal Science, Copyright ^ 1995 Academic Press later.
Copyright ^ 2000 Academic Press 2. Initiation and incubation, which are usually
Separation and Determination in started by adding the proper enzyme preparation
Physiological Samples to the reaction mixture or vice versa. All sub-
sequent time measurements are related to this
The determination of an absolute enzyme concentra-
initial time.
tion in a physiological sample is principally straight-
3. Termination, which can be achieved in various
forward; the main problem is the need for a puriRed
ways. Normally, this means inactivation of the
sample to use as a standard.
catalytic activity of the enzyme.
Enzymes are proteins found in nature in complex
4. Separation of the enzyme products from the en-
mixtures, usually in cells which perhaps contain sev-
zyme and its substrates.
eral hundreds of different enzymes. In order to under-
5. Detection and indetiRcation of the enzymatically
stand and interpret enzyme data from complex biolo-
formed product(s) during speciRc incubation inter-
gical systems in, for instance, a subcellular organelle
vals.
(such as a mitochondrion), a cell or whole organism,
6. Under certain conditions enzyme activity can be
we must try to understand its properties in as simple
followed dynamically, i.e. by rate measurement
a system as possible. From studies of an isolated
(f/t, where f is temperature, absorbance, Su-
enzyme we can learn about its speciRcity for certain
orescence, etc., and t is time).
substrates, the kinetic parameters for the reaction and
7. Interpretation of the produced data.
the possible means of regulation. All these parameters
are useful for understanding the role of the enzyme in
Handling of Specimens and Samples
more complex systems. The ready availability of iso-
(the Preanalytical Phase)
lated enzymes has been of considerable value in
a number of medical and industrial applications. For all biological material (tissue, urine, cerebrospi-
To study a given enzyme properly in physiological nal Suid, cell cultures, etc.) the same basic sequence of
samples it must be puriRed. Maintenance of biolo- procedures applies.
gical activity is the goal throughout the whole puriR-
cation scheme. Extracellular enzymes usually with- 1. Preparation of subject to be investigated.
stand the variety of stresses they are exposed to dur- 2. Collection of specimen.
ing the puriRcation. In contrast, when released from 3. Separation of sample from specimen.
their natural protective environment, intracellular 4. Transport of specimen and/or sample.
enzymes are very sensitive to various steps in the 5. Storage of specimen and/or sample.
puriRcation scheme. Integral membrane enzymes are 6. Pretreatment of samples for enzymatic analysis.
especially vulnerable during solubilization.
SpeciRc examples, together with trends in enzyme
determination in physiological samples, will be dis- The specimen is deRned as that part of the subject
cussed in this article. which is taken as representative for the analysis.
The sample is the material that is actually analysed.
Preparation of Enzymes with Biological The sample can be derived from, prepared from or be
a part of the specimen which is homogenized for
Activity from Physiological Samples measurement of enzyme activities. Consequently the
Enzyme Activity Measurement Process aliquot of homogenate that is analysed is the sample.
To elucidate enzymatic activity from physiological Only under certain conditions is the sample identical
samples certain steps have to be performed. to the specimen.
2722 III / ENZYMES / Chromatography
Sample Preparation Strategy different cell types within the cellular compartments
being separated form each other. Group II samples
Two factors should be considered during preparation are thus homogeneous populations of each type of
of enzymes with biological activity from physiolo- cell. This becomes the starting material for samples
gical samples. The Rrst factor for consideration is the where cell-surface activities can be directly assayed or
selection of the biological sample that is to be used as the cells can be lysed (broken up), thus providing
the starting material for the puriRcation. The samples access to the activities in intracellular organelles and
can be subdivided into three groups, depending on on cytoplasmic fragments.
their complexity (see Figure 1). The third group (III) consists of subcellular frag-
The Rrst group (I) includes samples containing dif- ments liberated by lysis of group II samples. These
ferent cell types and extracellular compartments, e.g. fragments include organelles such as mitochondria as
samples containing organs, tissue, biological Suids, well as those operationally deRned as the membrane
microbial cells, and unicellular organisms from a cul- fraction or a fraction containing soluble components.
ture medium or fermentation broth. Initially cellular The initial steps within this group will be separation
compartments have to be separated from noncellular of different organelles and separation of soluble from
compartments. The second group (II) consists of insoluble material. This is followed by solubilization
Figure 1 Enzyme purification scheme. Samples can be divided into three groups, I, II and III, depending on their complexity. Samples
in the different groups enter and leave the purification process at different points. The samples in group I are the most complex,
consisting of both extracellular and cellular enzyme-containing material which must first be separated. Group II samples contain
different types of cells, one of which contains the enzyme of interest. In group III the enzyme source is from only one type of cell, from
which the ‘right’ subcellular fraction gives the ‘right’ enzyme. (Adapted from Rossomando, 1987.)
of membrane fractions and Rnally separation of mo- Samples Obtained from Cell Culture
lecular species form each other.
For cells grown in liquid culture, including mam-
The second factor for consideration during the
malian cells, fungi, protozoa and bacteria, noncellu-
preparation of biologically active enzymes concerns
lar compounds should be separated form the cells
to what extent the sample should be puriRed. The tra-
before analysis takes place. Even here, a low-speed
ditional end point of any puriRcation scheme would
centrifugation is sufRcient for separation of cells from
be a homogeneous protein. The main goal should be
culture media. The supernatant should be assayed for
to assay a single enzymatic activity without interfer-
enzymatic activity and the cell-pellet material could
ence form other activities. However, for some studies
be set aside for later assay or lysis.
it is advantageous or even necessary to assay the
activity of interest in the presence of other activities.
Sample Obtained from Tissue or Organ Enzyme Activity Determination
Tissues and organs (e.g. skin, liver) can be divided Extracellular Enzymes
into at least two compartments, the cellular compart- The extracellular Suid around tissues or the growth
ment and the extracellular compartment. The activity medium around mammalian cells, bacteria, yeast or
of interest could be localized in either of the compart- fungi cells often contains the enzyme activity of inter-
ments. Cell-sorting techniques should be used where est. Several factors have to be considered before activ-
whole undamaged cells are separated from the stable ity measurements can be started. Proteolytic enzymes
Rbrillar matrix. Some cell damage is unavoidable must be inhibited early in the determination process,
with the harsh methods often used for disrupting because otherwise they will degrade and destroy the
the matrix, e.g. cutting or dicing with scissors, enzymatic activity of interest. Another factor for con-
shearing in a blender, or grinding. SpeciRc disruption sideration is the interference of small molecules,
of the matrix can be performed using puriRed en- which could be erroneously measured as enzyme sub-
zymes, e.g. collagenase for mammalian tissues. Tryp- strate or product. A special case is inhibitors which
sin and other proteolytic enzymes have also been used diminish enzyme activity. If a serum-free medium is
with success. Such procedures result in a mixture of not used during mammalian cell propagation, special
cells extracellular compartments including some in- care concerning serum enzymatic activity measure-
soluble fragments, and added enzymes (including re- ments and puriRcation has to be taken.
agents).
Within the Cellular Compartment
Samples Obtained from Tissue or Organ Culture After sample preparation, cells are separated from
noncellular material in different manners depending
Cultured samples are usually treated in the manner on the complexity of the original sample (i.e. from
described previously. Precautions must be taken to which organ the sample originates). Because a num-
avoid errors due to the additional extracellular com- ber of different types of cells could remain, an assay
partments and the culture medium, which may con- of any complex sample should begin with the prep-
tain enzymatic activities originating from the medium aration of only one cell type.
itself or produced during sample growth. Many separation methods utilizing different prop-
erties of the cells have been used. In sucrose gradient
Samples Obtained from Biological Fluids
centrifugation, density differences between cell types
Body Suids such as blood, cerebrospinal Suid and are used and each cell type Rnds its equilibrium posi-
saliva contain cells as a normal component. However, tion in the sucrose gradient. Field Sow fractionation
in other Suids cells are a contamination. Cells in urine can also be used for separating cells according to their
may indicate a disease process. The study of en- size and shape. Adipose tissue cells will separate with-
zymes in such Suids also requires separation of the out centrifugation; they just Soat. Antibodies raised
two compartments (cells from biological Suid). against any special marker on the cell surface could
Because biological Suids do not contain Rbrillar be explored for selection of that special cell type with
matrix material, the separation will, for instance, be various techniques. For example, using metallic iron
a simple centrifugation step (5000 g for 15 min) coupled to antibodies a strong neodymium perma-
which will produce a pellet containing most of the nent magnet could be used for selecting a speciRc cell
cellular material. The supernatant produced by cen- type. Selective chemical lysis of cell types that are not
trifugation can be assayed for enzymatically active of interest followed by mild centrifugation will pro-
proteins. vide the cells containing the enzyme of interest. Even
homogeneous mammalian cells can differ in enzy- Table 1 Marker enzymes for different subcellular fractions
matic activity if the age or nutritional status of the
animals is not identical. Subcellular fraction Enzyme
Intact Cells Nuclei DNA nucleotidyltransferase

Nuclei Nicotinamide-nucleotide
When a reaction mixture consists only of one type of adenyltransferase
cell any assay of enzymes on the cell surface will only Mitochondria Succinate dehydrogenase
Mitochondria Cytochrome c oxidase
be straightforward if the cells are not disrupted dur-
Endoplasmic reticulum Glucose-6-phosphatase
ing determination, because the presence of intracellu- Lysosomes Acid phosphatase
lar components could give rise to false results. For Lysosomes Ribonuclease
example, if cell-surface adenosine triphosphatase Peroxisomes Catalase
(ATPase) activity is to be measured it is suitable to Peroxisomes Urate oxidase
Plasma membrane 5-Nucleotidase
monitor the product ADP (adenosine diphosphate).
Cytosol Glucose 6-phosphate dehydrogenase
However, if the cells are lysed, intracellular ADP will Cytosol Lactate dehydrogenase
affect the enzyme activity determination. Cytosol 6-Phosphofructokinase
Subcellular Samples From Price and Stevens (1989), p. 369.
Depending on the localization of the enzyme of inter-
est, different strategies for the cell lysis have to be A subcellular fractionation scheme based on centrifu-
considered. A subcellular fractionation could be quite gation for mammalian cells is outlined in Figure 2.
rewarding later in the enzyme puriRcation route. The enzymatic activity of interest can be followed,
Methods such as sonication, use of a French press, together with the activity of marker enzymes, through
blending or homogenization are useful for the lysis of the subcellular scheme (Table 1). The subcellular
different types of cells. localization is then predicted and further puriRca-
For example, a French press is needed for the lysis tion can be performed.
of bacterial cells with rigid cell walls. The use of
a Potter}Elvehjem homogenizer (PTFE (poly(tetra Enzyme Puri\cation Methods
Suoroethylene)) pestle in a glass mortar) for the hom-
ogenization of cells with fragile cell membranes is Maintenance of Biological Activity
a procedure that will merely break the outer cell Soluble proteins, either cytoplasmic or inside or-
membrane, leaving most of the cell organelles intact. ganelles, are present in highly concentrated soups
Figure 2 A subcellular fractionation scheme for mammalian cells by differential centrifugation. (From Price and Stevens (1989), p.
368.)
of proteins, with concentrations ranging from the sulRnic acid or irreversibly oxidized to the sul-
100 mg mL\1 to as high as 400 mg mL\1 inside the fonic acid.
mitochondrial matrix. Oxygen tension is low and These reactions can be suppressed by using differ-
different natural reducing compounds such as ent additives such as ethylenediaminetetraacetic acid
glutathione are present to maintain a high reducing (EDTA), which masks by complexation cations that
potential. Other stabilizing agents are also present, would otherwise catalyse the formation of disulRde
such as different substrates and products. When the bonds with cysteine residues at the catalytic site.
tissue is disrupted during puriRcation proteins are Other commonly used reducing agents are sulfhydryl-
released from their protective environment and pro- containing reagents such as -mercaptoethanol, 2,3-
teolytic enzymes which are held in separate compart- dimercaptopropanol, thioglycolate, glutathione, cys-
ments are also released. As a consequence the enzyme teine and the dithio-analogues of the reduced C4
of interest has to be protected from oxidation, pro- sugars, threitol and erythritol (DTT, DTE). EDTA
teolytic degradation and irreversible unfolding of and other complexing agents often cannot be used for
their tertiary structure. This can be achieved in many enzymes that have an essential metal ion in the active
different ways and should be tailor-made for each site.
puriRcation scheme and enzyme. Proteinases or proteolytic enzymes are contained
Denaturation during puriRcation can be minimized inside living cells. In mammalian cells they are packed
if precautions are taken according to extremes of pH, in lysosomes. In microorganisms they are often found
temperature and organic solvents. The natural pH between the plasma membrane and the cell wall. Dur-
inside a cell is normally in the range 6}8. Using ing preparation of an enzyme extract by cell homogen-
buffers within this pH range at appropriate ionic ization, release of proteolytic enzymes will occur and
strength should protect against pH denaturation. Re- their activity has to be inhibited. Depending on the
ducing the temperature by 15}253C decreases many wide scale of different proteinases present in the cell
degradation processes three- to Rve}fold. Reducing homogenate, many types of inhibitors have to be used.
the temperature also slows down the processes in- Diisopropyl Suorophosphate (DFP) inhibits serine
volved in the separation method, for instance in size proteases. Note that DFP is dangerous to handle
exclusion chromatography. because it is volatile and attacks human acetyl-
Enzymes in dilute preparations may denature due cholinesterase, vital in nerve conduction. Phenyl-
to adsorption onto the wall of the container or onto methylsulfonyl Suoride (PMSF) is a nonvolatile serine
the chromatography matrix used. Enzymes with protease inhibitor and does not attack acetylcholines-
quarternary structure will dissociate and the activity terase. It inhibits some thiolproteases and some car-
of interest will be lost. boxypeptidases, but it has to be dissolved in acetone
Adsorption and denaturation of dilute enzyme or isopropanol.
samples can be circumvented by using a carrier pro- Pepstatin, leupeptin and antipain are peptide-based
tein such as bovine serum albumin (BSA), or even inhibitors that are very potent against acid proteases
better a commercial synthetic carrier protein with such as pepsin, cathepsin D and yeast protease A.
a simple and known structure at a concentration as Normal concentrations of inhibitors used in freshly
high as 1 mg mL\1. Care must be taken to avoid prepared extracts are 5 mmol L\1 EDTA, 1 mmol L\1
interaction between the carrier protein and the PMSF, 10 mol L\1 pepstatin, 10 mol L\1 leupep-
enzymes of interest. The molecular size of the carrier tin and 10 mol L\1 antipain.
protein should also differ by at least a factor of Other stabilizing factors during enzyme puriRca-
two from the enzyme of interest, for easy removal by tion are some nonaqueous hydrophilic molecules.
size exclusion chromatography if the pure enzyme Owing to the high nonaqueous content in cell cyto-
protein is needed, for instance in amino acid plasm (10}15% w/v), the water around the protein
sequencing. molecules is not freely mobile and thus stabilizes the
Catalytic site inactivation by speciRc reactions is protein structure. To mimic this situation in prepared
hard to avoid. Loss of cofactors can be prevented by enzyme extracts glycerol is added at 10}50% (w/v).
including them in the puriRcation buffer. Covalent Below 30% the viscosity is not harmful for most
modiRcation of the active site which contains reactive methods used during enzyme puriRcation except for
amino acid residues responsible for the catalysis is ultraRltration and centrifugation and in some ‘salt-
common. The most troublesome amino acid is cys- ing-out’ experiments because hydrogen bonding and
teine, which is very susceptible to modiRcation. Sul- hydrophobic forces decrease. Sugar or sugar alcohol
fhydryl residues at the active site may be in the solutions such as glucose, fructose, lactose and
ionized form, which is prone to oxidation and can sorbitol and be used instead of glycerol. The mech-
form disulRde bonds, or may be partially oxidized to anism of protection is similar.
Separation Methods Based on Size, Shape, Mass, amount of enzyme that converts 1 mol of substrate
Charge, Hydrophobicity, Solubility and per min under deRned reaction conditions) can be
Biological Recognition followed through the puriRcation scheme d it should
The different physicochemical properties of the en- rise and then reach a plateau. Crude extracts should
zyme that should be utilized during a puriRcation be concentrated by fractional precipitation with am-
scheme include size, shape, charge, hydrophobicity, monium sulfate or poly(ethylene glycol) (PEG) or
solubility and biological recognition. The salient adsorbed and desorbed from a chromatographic
points of various separation methods are listed in matrix as soon as possible. Precipitated proteins, after
Table 2. dissolution in a small volume, are more stable be-
After each step in a puriRcation scheme, a proper cause they are more concentrated. Centrifugation
enzyme activity assay should be carried out and the with Reld strength from 5000 to 50 000 g is widely
amount of protein determined. If correctly done used for subcellular fractionation and for ammonium
the speciRc activity (in units per milligram) (1 U is the sulfate and PEG-precipitated enzymes.
Table 2 Enzyme separation methods
Physicochemical Method Chracteristic Scale Use Enzyme activity

property recovery
Size, shape or mass Centrifugation Moderate resolution; Large or small Partial fractionation Good
slow
Gel filtration Moderate resolution; Small Desalting, size Good
slow determination,
fractionation
Field flow fractionation Good resolution; fast Small Size determination, Good
fractionation
Ultrafiltration Bad resolution; Large or small Desalting, concentration Good
slow/medium
Dialysis Bad resolution; slow Large or small Desalting, concentration Good
Polarity
(a) Charge Ion exchange High resolution; fast Large or small Fractionation, Good
chromatography concentration
Chromatofocusing Excellent resolution; Medium Fractionation, Poor
fast concentration
Electrophoresis High resolution; Small/medium Fractionation, Medium/poor
medium/fast visualization
Isoelectric focusing Excellent resolution; Small/medium Fractionation, Poor
medium visualization
Capillary Excellent resolution; Extremely small Fractionation, size Good
electrophoresis fast determination possible
(b) Hydrophobic Hydrophobic Good resolution; fast Large or small Fractionation, Good
character interaction concentration
chromatography
Reversed-phase Excellent resolution; Large or small Fractionation, Poor
chromatography fast concentration
Solubility Change in pH Medium resolution; Large or small Concentration, Medium
fast fractionation
Change in ionic Medium resolution; Large or small Concentration, Medium/good
strength fast fractionation
Decrease in dielectric Medium resolution; Large or small Concentration, Medium
constant fast fractionation
Two-phase separation Medium/good Large or small Fractionation, Good
resolution; medium concentration
Biological activity Affinity Excellent resolution Generally small Fractionation, Medium/good
chromatography concentration
Specific binding or DyeIligand Good resolution; Large or small Fractionation, Medium/good
structure features chromatography fast concentration
Immuno- Excellent resolution; Generally small Fractionation Medium/good
chromatography fast
Covalent Medium/good Medium/small Fractionation Medium
chromatography resolution; fast
Field Wow fractionation Field Sow fractionation A summary of the steps taken in the puriRcation of
(FFF) is a chromatography-like separation technique HSL from rat epididymal adipose tissue is given in
which is designed for fractionation of macro- Figure 3.
molecules, colloids and particles. The principle is Step 1. Fat pads frozen in 0.25 mol L\1 sucrose,
simple. A laminar Sow of carrier liquid between two 1 mmol L\1 EDTA, 1 mmol L\1 DTE and
walls, separated by c. 0.1 mm, creates a parabolic 10 mol L\1 of leupeptin and antipain in liquid nitro-
velocity proRle. The sample is injected into the carrier gen were homogenized (PotterdElvehjem homogen-
stream at the inlet of the channel and exits through izer) in 30% (w/v) 0.25 mol L\1 sucrose and partially
the outlet end which is connected to a detector. delipidated by removing the fat Soating after centrifu-
Sample retention is achieved when molecules are gation at 5000 g for 10 min at 43C.
pushed to the accumulation wall (an ultraRltration Step 2. The supernatant was further delipidated
membrane) by an external Reld force (a cross-Sow), so and separated from the pelleted material by centrifug-
that they obtain different average distances from the ing at 100 000 g for 45 min (referred to as the
wall and are placed at different heights in the para- ‘100 000 g supernatant’).
bolic Sow proRle. The different sample molecules are Step 3. The pH was lowered to 5.2 with acetic acid.
consequently transported down the channel at differ- HSL was precipitated over 30 min on ice and the
ent velocities. Thus separation can be achieved. pellet collected after 30 min of 10 000 g centrifu-
The size range embraced by FFF is from small gation. The pellet was resuspended in 20 mmol L\1
proteins (c. 10 kDa), up to organelles and cells with Tris-HCl (2-amino-2-hydroxymethylpropane}1,3-
a diameter of several micrometers. FFF does not rely diol hydrochloride), pH 7.0, containing sucrose as
on a stationary phase, which makes it very useful for before (this fraction is referred to as ‘the pH 5.2 ppt
separation of labile enzyme molecules. Separation fraction’). This fraction contains practically all the
times are very short (3 d 10 min) and selectivity ac- HSL and about 25% of the contaminating proteins.
cording to size is better than for gel Rltration. Loada- The preparation is stable for several months at
bility is so far limited to c. 200 g per separation run. !803C.
Step 4. Further solubilization of the pH 5.2 ppt
fraction was by sonication at 103C in the nonionic
Examples of Enzyme Determination detergent C13E12 (heterogeneous alkyl polyoxyethy-
in Physiological Samples lene glycol type). The solubilized HSL was frac-
Hormone-Sensitive Lipase from Adipose Tissue
tionated by gradient sievorptive chromatography on
quaternary aminoethyl (QAE)-Sephadex, for a separ-
Hormone-sensitive lipase (HSL, EC 3.1.1.3) is an ation time of 12 h at 103C. 70% of the recovered
amphiphilic enzyme and the key control of energy enzyme was pooled and concentrated 15-fold by
substrate Sow in mammals. Its activity in adipose ultraRltration and referred to as the QAE-Sephadex
tissue determines the rate of hydrolysis of stored fraction.
triacylglycerols and thereby the production of fatty Step 5. This fraction was dialysed and concentrated
acids for release as free fatty acids (FFAs) into the three-fold further against 20 mmol L\1 Trisdacetic
circulation. acid, pH 7.50, containing 20% (w/v) glycerol,
The following parameters were considered during 15% PEG, 1 mmol L\1 DTE, 0.2% C13E12 and
puriRcation. 10 mol L\1 leupeptin for 8 h at 43C, immediately
followed by chromatography on a Mono Q column
1. Development of a suitable assay procedure
(polymer-based strong anion exchanger for liquid
2. Selection of the best source from which the mol-
chromatography from Pharmacia) as described in de-
ecule could be puriRed.
tail in Figure 4. The enzyme peak fractions collected
3. Solubilization of the desired molecule.
were immediately brought to pH 7.0 by addition of
4. Development of a series of isolation and concen-
a potassium phosphate buffer to give a Rnal concen-
tration procedures which includes stabilizing the
tration of 30 mmol L\1 and the fractions were stored
molecule at each stage.
at !803C in 50% (w/v) glycerol. This enzyme prep-
Lipase activity was measured against emulsiRed [3H]- aration is referred to as the ‘Mono Q enzyme’.
oleic acid-labelled monooleoylglycerol (a diacyl- Step 6. The last step of the puriRcation scheme
glycerol ether analogue). An enzyme activity of 1 U consisted of Mono S chromatography (polymer-
corresponds to the release of 1 mol of fatty acids per based strong cation exchanger for liquid chromato-
minute at 373C. The assay was performed between graphy from pharmacia). Before chromatography the
each puriRcation step. The enzyme source was rat Mono Q enzyme was dialysed and concentrated
adipose tissue (epididymal fat pads). three-fold for 3 h against 10 mmol L\1 potassium
2728
III / ENZYMES / Chromatography
Figure 3 Purification scheme for HSL from rat epididymal adipose tissue. Enzyme activity was monitored at each step. (Reproduced with permission form Nilsson
and Belfrage (1986).)
Figure 4 Fractionation of partially purified HSL by high-performance Mono Q anion exchange chromatography. (A) The sample,
7.5 mL of concentrated QAE-Sephadex enzyme (about 17 mg of protein), representing half of the preparation from adipose tissue of
500 rats, was applied to an 8 mL Mono Q column (fitted with a precolumn, flow rate 1.0 mL min\1 at a back-pressure of about 1.5 MPa)
pre-equilibrated in 50 mmol L\1 Tris-acetate, pH 7.5, containing 1 mmol L\1 DTE, 20% (w/v) glycerol and 0.2% (w/v) of the nonionic
detergent C13E12. (B) After adsorption the lipase was eluted (4.0 mL min\1 at a back-pressure of about 4.2 MPa) by the increasing salt
and decreasing pH gradient obtained by addition of 0.3 mol L\1 sodium acetate, pH 7.0, as indicated by the conductivity and pH profile
in the figure. HSL activity (shaded area) was measured towards an emulsified lipid substrate. One unit (U) of enzyme activity
III / ENZYMES / Chromatography
corresponds to 1 mol of fatty acid released per minute at 373C. The protein composition of the QAE-Sephadex enzyme sample applied
to the column (lane a), and the indicated column fractions (lanes b}o), analysed by SDS-PAGE and Coomassie blue staining. Lanes to
the extreme left and right are reference proteins; values are in kiloDaltons. Arrow labelled HSL signifies the HSL M r"84 000 subunit.
2729
Enzyme peak fractions corresponding to 70% of the total enzyme eluted were pooled for the next step. (Reproduced with permission
from Nilsson and Belfrage (1986).)
Table 3 Purification of HSL from rat adipose tissuea
Purification step Volume (mL) Protein (mg) Enzyme activity Specific activity Purification Yield (%)
(mol fatty acids ( mol fatty acids (-fold)
per min) per min per mg
protein)
100 000 g supernatant 790 4900 314 0.06 1 100

pH 5.2 ppt 50 1204 289 0.24 4 92
QAE-Sephadex 46 33 112 3.40 57 36
Mon Q LCb 16 2.1 58 27 450 18
Mono S LCb 8 0.2 34 154 2567 11
a
Enzyme was purified from about 600 g of epididymal fat pads from 500 rats. Enzyme activity was measured with monoacylalkyl-
glycerol substrate at 373C.
b
Combined enzyme from two identical chromatographic treatments of half the initial batch.
Reproduced with permission from Nilsson and Belfrage (1986).
phosphate buffer, pH 7.0, followed by 3 h against the A single band on SDS-PAGE with Coomassie blue,
same buffer, pH 6.5, both containing the same con- or even silver staining together with maximum en-
centration of glycerol, PEG, C13E12, DTE and leupep- zyme activity, are not conclusive evidence for identi-
tin as used for the Mono Q chromatography. The Rcation.
enzyme peak fractions (70% of the enzyme activity
Plasminogen Activator in Gingival Crevicular Fluid
recovered) were brought to pH 7.0 and glycerol was
added to 50% (w/v). The results of the puriRcation There is a correlation between plasminogen activator
are illustrated in Table 3. (PA) concentration in gingival crevicular Suid (GCF),
The obstacles encountered in the determination of which is an extracelluar exudate occurring in the
enzymes in biological samples are well illustrated in gingival crevice, and gingival inSammation. The con-
this puriRcation scheme for the enzyme HSL, which centration of plasminogen activator inhibitor (PAI) in
has been notoriously recalcitrant to puriRcation be- GCF also plays an important role. The Rbrinolytic
cause of its low tissue abundance, amphiphilicity and system is activated by PAs, which are serine proteases
general lability. What are the necessary precautions that catalyse the conversion of the inactive proen-
that have to be fulRlled during the process? After zyme plasminogen to the active enzyme plasmin
every solubilization, fractionation or concentration which then activates collagenase and thereby partici-
step, the biological activity has to be estimated to pates in the tissue destruction seen at inSammatory
detect any inhibition, destruction or loss of the en- lesions.
zyme of interest. To obtain optimum enzyme activity, Sampling of GCF was performed by placing small
certain precautions have to be taken in between every discs (Millipore GWVP-Rlter 0.22 m, calibrated in
step as discussed previously, by adding reducing size to absorb a determined volume) in the gingival
agents or stabilizers, lowering the temperature or crevice.
speeding up the separations where possible. To eluci- PA determination is performed by two different
date the effectiveness of different fractionation steps methods.
used, protein purity must also be examined with
a nonchromatographic method such as sodium 1. Enzyme-linked immunosorbent assay (ELISA),
dodecyl sulfate}polyacrylamide gel electrophoresis where the amount of protein is determined by
(SDS-PAGE) or capillary electrophoresis. Another, placing the small discs in the wells of the micro-
not easy, problem of importance is the identiRcation titre plates used for ELISA (further details of
of the protein band on SDS-PAGE that corresponds ELISA are given elsewhere). This can be done
to the enzymatic activity. providing monoclonal or polyclonal antibodies
In the case of HSL, it was possible to carry out that have been raised against the enzyme, which in
the identiRcation in rather crude preparations turn demands a relatively pure enzyme for immu-
because the enzyme was activated through covalent nization.
modiRcation, i.e. phosphorylation. Then it was 2. Gel lysis can be used for the determination of
possible to ‘tag’ HSL with 32P and identify the SDS- enzymatic activity of PA. The Rlter discs are placed
PAGE band by autoradiography. After the Mono on plasminogen-rich Rbrin plates and incubated
Q puriRcation step there was only one phos- for 18 h. PA activity can then be derived from the
phorylated band. size of the gel lysis.
Figure 5 Instrumental set-up for levitation single cell analysis.
Single Cell Analysis for example been performed on freshly prepared,

intact and living primary adipocytes. The instrumen-
The key step in future enzyme determination is minia- tal set-up is shown in Figure 5. A single adipose cell in
turization, combined with highly selective separation a 500 nL droplet is acoustically levitated. Stimulation
and detection methods. Separation and determination of adipocytes with -adrenergic agonists results in
should be done simultaneously. The goal is to be able activation of adenylate cyclase, production of cAMP
to study the molecular processes of life at the level of and activation of cAMP dependent protein kinase
a single cell and its subcellular compartments, if (PKA). PKA phosphorylates HSL, leading to activa-
possible without destroying the cell’s integrity, and tion of enzyme activity and increased lipolysis, result-
analysing for a speciRc enzyme. Capillary liquid ing in FFA release and a consequent pH decrease in
chromatography and capillary electrophoresis are the surrounding buffer droplet. Addition of insulin
good candidates for single-cell analysis. In fact, small antagonizes this effect and hence also the decrease in
molecules such as catecholamines, dipeptides and pH. The change in pH, i.e. the cell response in the
proteins have already been determined in single viable droplet, is followed by a pH-dependent Suorophore
cells. This can be achieved either by sucking a single continuously monitored by Suorescence imaging de-
cell into the capillary and causing cell lysis by use of tection. Additions to the levitated droplet are
a high ionic strength buffer, or penetrating the outer achieved using continuous Sow-through droplet dis-
cell membrane with an ultrathin capillary (2 m i.d.) pensers. To counteract droplet evaporation, which
and sucking in some of the cell contents. affects the Suorescence intensities, a dispenser is used
By introducing a substrate for a speciRc enzyme to continually add water, thus keeping the droplet
during the separation, it is possible to follow its acti- volume constant. An image analysis computer pro-
vity directly after cell lysis by studying the decrease of gram is employed to calculate droplet Suorescence
the substrate and increase of the product concentra- intensities during experiments. The method is parti-
tions. In some cases the enzyme itself can be moni- cularly useful for studying of dynamic events in natu-
tored during the separation if the amount is high ral cellular environments at the single cell level, e.g.
enough. Detection techniques sensitive enough to de- for the screening of new drug candidates or for study-
tect substrates, products and enzymes at the concentra- ing side effects and reactions between cells.
tion levels derived from a single cell experiment are Subcellular fractionation at the single-cell level us-
laser-induced Suorescence or amperometric detection. ing a two-phase levitated droplet system is under
Single cell analysis can also be achieved using other development at the author’s laboratory.
miniaturized analysis systems. A very suitable method
for studying living cells and biochemical reactions is See also: I/Affinity Separation. II/Affinity Separation:
acoustically levitated microdroplets. Cell studies have Theory and Development of Affinity Chromatography.
2732 III / ENZYMES / Liquid Chromatography
Centrifugation: Analytical Centrifugation. Chromatogra- Nilsson S and Belfrage P (1986) PuriRcation of hormone-
phy: Protein Separation. III/Enzymes: Liquid Chromato- sensitive lipase by high-performance ion exchange
graphy. Proteins: Capillary Electrophoresis; Centrifu- chromatography. Analytical Biochemistry 158:
gation; Crystallization; Electrophoresis; Glycoproteins: 399}407.
Liquid Chromatography; High-Speed Countercurrent Prince NC and Stevens L (1989) Fundamentals of Enzymol-
Chromatography; Ion Exchange; Metalloproteins: Chrom- ogy, 2nd edn. Oxford: Oxford University Press.
atography. Appendix 1/Essential Guides for Isolation/ Rossomando FE (1987) High Performance Liquid
Purification of Enzymes and Proteins. Chromatography in Enzymatic Analysis. New York:
Wiley-Interscience.
Scopes KR (1987) In: Cantor RC (ed.) Protein PuriTcation
Further Reading Principles and Practice, 2nd edn. New York: Springer-
Bergmeyer HU, Bergmeyer J and Grass M (1986) Methods Verlag.
of Enzymatic Analysis, 3rd edn, vols. IdXII. Weinheim: Suelter HC (1985) A Practical Guide to Enzymology and
VCH. Biochemistry. New York: John Wiley & Sons.
D. Shekhawat and N. Kirthivasan, Michigan State tant factors for understanding their separation. The
University, East Lansing, MI, USA size of a protein depends upon the number of amino
Copyright ^ 2000 Academic Press acid units in the protein, whereas the polarity de-
pends on the hydrophilic and hydrophobic units
present.
Introduction A protein molecule contains one of the three
groups: uncharged polar, potentially positively
Enzymes Rnd applications in food, pharmaceutical
charged (basic side chain) or a potentially negatively
and biochemical industries. They are found in combi-
charged (acidic side chain). These side chains are
nation with other macromolecules or various small
normally ionizable and this leads to proteins having
molecules. Their uses require identiRcation and puri-
characteristic isoelectric points. Since there are other
Rcation of the enzymes. The nature, quality, and
issues governing the protein molecule, such as size,
quantity of the desired enzyme are determined by its
shape and nature of the solution (pH), the overall net
intended use. For example, the food industry needs
charge and polarity depends upon the combination of
enzymes in large quantities and the pharmaceutical
these factors. In general, the chromatographic pro-
industry requires ultra pure enzymes. High-perfor-
cesses associated with the properties of enzymes can
mance liquid chromatography (HPLC) is widely used
be summed up as indicated in Table 1.
for the separation or puriRcation of enzymes on
a preparative or analytical scale. It is also used for the
analysis of the enzymatic activity. High-Performance Liquid
Chromatographic Techniques
Properties of Proteins and Practical Size-Exclusion Chromatography
Implications Size-exclusive chromatography (SEC) is primarily
All enzymes are proteins. All proteins are macro- used as a Rrst step in puriRcation when molecules
molecules with molecular weights ranging from hun-
dreds to several thousands. The sequence of amino Table 1 Chromatographic processes associated with enzyme
acid in a protein is speciRc and this gives each protein properties
unique properties. A protein with just amino acids as
a building block is a simple protein; those that con- Enzyme property Chromatographic method
tain additional units, such as a nucleic acid, a lipid or Net charge Ion-exchange chromatogrphy
a metal etc., are called conjugated proteins. Size Size-exclusion chromatography
There are 20 naturally occurring amino acids Substrate affinity and Affinity chromatography
found in proteins that vary in structure; thus, it is the conformation
amino acid sequence and composition that determine Polarity Reversed-phase chromatography and
hydrophobic-interaction
the properties of enzymes. Two distinct properties are chromatography
the size and polarity of the protein, which are impor-
III / ENZYMES / Liquid Chromatography 2733
differ signiRcantly in size. This technique is used ex-

tensively in biochemistry for fractionation and mo-
lecular weight determination of proteins and en-
zymes. The basis of separation in SEC, as the name
suggests, is the size of the molecules to be separated.
Spherical beads made of a cross-linked gel of a poly-
mer such as silica, agarose, or polyacrylamide are
used as column packings. Small molecules can enter
all pores and elute with a characteristic volume equiv-
alent to the column hold-up volume. Large molecules
are excluded from those pores with a smaller cross-
section than the solute and elute in a smaller volume
than the small molecules. Consequently, molecules
passing through the column separate on the basis of
their size and elute in order of decreasing molecular
weight.
The resolution depends on gel bead size, pore size,
column size, sample size, and Sow rate of the mobile
phase. Low Sow rates of the mobile phase, long and
narrow columns, and small gel bead sizes give the
highest resolution. The pore size of gel beads is
designed according to the size of the molecules of
interest.
The nature of the mobile phase in SEC is very
important in enzymatic separation as the protein (en-
zyme) conformation can be changed due to solvent
polarity, pH, ionic strength, and salt concentration.
Conformational change can change the whole
chromatographic behaviour of an enzyme. In SEC,
solute}stationary phase interaction is completely pro-
hibited for better separation. However, the mobile Figure 1 Gel permeation of Bacillus circulans proteins on acryl-
phase may produce (ionic or hydrophobic or both) ate gel (300;7.8 mm) at flow-rate of 0.5 mL min\ using 2 mg (A)
solute}stationary phase interactions. It is necessary to and 20 mg (B) of cell extract proteins. Pooled active peaks
design an ideal mobile phase for chromatography to (0.1 mg) from (B) were reinjected (C). Reproduced from Hamada
JS (1995) Journal of Chromatography A 702, 163}172, with
avoid the above problems. Most enzymes are stable in
permission from Elsevier Science.
the pH range of 5}8. The desired pH of mobile
phase in SEC is obtained by using buffers suitable for
both the enzyme to be analysed and the stationary (PGase) from Bacillus circulan cell extract (1).
phase. Tris(hydroxymethyl)aminomethane salt PGase sample loading was 2}30 mg in 0.05 M sodium
solution or phosphate buffers are widely used for phosphate buffer (pH 8.0) and the eluent was 0.05 M
enzymes. Denaturing solvents are sometimes also sodium phosphate at Sow rate of 0.3 mL min\
employed for the chromatographic separation of (Figure 1). The pooled PGase peak from multiple
enzymes. Protein denaturants, detergents in mobile injections was then further puriRed by anion-ex-
phase, change the original proteins to random change chromatography.
coil conRrmations.
Ion Exchange Chromatography
Organic solvents such as acetonitrile can be used
for SEC of enzymes. Using acetonitrile is very advant- Ion exchange chromatography (IEC) is a widely used
ageous because it can be evaporated after elution to technique for enzyme puriRcation because of the net
concentrate the enzyme solution. Acetonitrile is an charge characteristics of enzymes. In IEC, the charged
ideal organic solvent if a UV detection method is functional groups are covalently bound to the solid
used. However, the solubility of enzymes in acetonit- surface of the matrix. Cellulose, silica or styrene-
rile solution limits its use in the mobile phase. divinylbenzene is used as a matrix. Cation exchanger
A prepacked column of cross-linked methacrylate resins contain immobilized negatively charged func-
gel (Ultrahydrogel from Waters) was used for pre- tional groups (i.e. RSO\ 3 , RCO\
2 , RPO\ 4 and RO\)
liminary isolation of fractions of peptidoglutaminase and anion exchangers contain immobilized positively
charged functional groups (i.e. quaternary am- pH of the solvent. Ionization of ammonium groups
monium groups). Resins containing sulfonyl and occurs at any pH below the isoelectric point (pI) and
quaternary ammonium groups are strong ion ex- contributes a positive charge on the enzyme. Sim-
changers and ionize at any pH, whereas weak ion ilarly, a negative charge on the enzyme is obtained by
exchangers containing functional groups like car- a pH above its pI. The retention on ion exchange
boxyl and secondary or primary amines ionize within columns therefore depends on the net charge carried
a certain range of pHs. When an ionized solute passes by the enzyme molecule to be separated and the pH of
through an ion exchange column, the sample ions the mobile phase is chosen accordingly. The pH of the
adsorb and displace counter ions on the surface. The mobile phase should be slightly above the pI of the
adsorption process is reversible and adsorbed ions are enzymes to be separated for anion-exchange columns
eluted by a salt solution. Using a suitable mobile and vice versa for cation-exchange columns. The na-
phase regenerates the resins. ture of displacing counterion in the salt used also
The choice of the appropriate ion exchange resin affects the enzyme retention on the column. Higher
for a particular enzyme separation depends on the valent ions are stronger displacers than lower valent
isoelectric point of the enzyme to be separated. The ions and thus give lower retention. The smaller size of
isoelectric point of any molecule is the pH value at ions also favours lower retention if the charge on the
which the molecule has no net charge, i.e. an equal counterion is the same. Gradient elution based on
number of negative and positive charges. Enzymes are ionic strength variation or pH changes is used in IEC.
acidic at a pH above the isoelectric point and anion- Chromatographic separation of most enzymes is car-
exchange resins are used for separation of such en- ried out at a low temperature to preserve enzyme
zymes. Similarly, cation-exchange resins are used for stability.
basic enzymes. The quaternary methylamine resin from Waters
Column packing material, particle size and pore (Accel Plus QMA) has been used for the separation of
diameter of the support, column length, mobile peptidoglutaminase from B. circulam cell extract. En-
phase, and temperature are some important para- zyme load was 1.0 mg in 20 L of 0.02 M phosphate
meters which affect separation of enzymes in IEC. In buffer (pH 8.0) and the eluent used was 0.05 M so-
a column with smaller packing particle sizes, large dium phosphate buffer and 0.1}0.8 M KCl at a
enzyme molecules diffuse at a slower rate and this Sow rate of 0.5 mL min\1 for analytical separation
results in enhanced resolution and lower elution time. and 1.5}10.0 mL min\1 for preparative separation
Most of the surface area of a support is conRned (Figure 2).
within the pores and the diameter of pores affects the
penetration of enzyme molecules into the column
Reversed-Phase Chromatography
matrix and hence the mass transfer and loading capa-
city. Pore diameters of +300 A> are ideal for most Reverse-phase chromatography (RPC) is the most
enzymes with molecular weight up to 100 000, pro- popular chromatographic method for the puriRca-
viding loading capacity and good resolution. Larger tion, separation, and analysis of the biological mol-
pore diameters are needed for higher molecular ecules because of its high resolution and ease of hand-
weight enzymes. The column length is not an impor- ling. Column packings are usually prepared from
tant factor in the resolution of enzymes in IEC. Using silica particles and hydrophobic long-chain alkylsilyl
short columns has many advantages, e.g. concen- ligands. n-Butyl (C4), n-octyl (C8), n-octadecyl
trated eluents, lower pressure and lower column cost. (C18), and alkylphenyl groups are used for separating
Lower loading capacity is a disadvantage of using enzymes. The nonhydrophobic molecules in the
a short column. sample do not strongly interact with the hydrophobic
The pH, ionic strength and salt composition of stationary phase of the column and elute earlier,
mobile phase are also important factors in IEC separ- while hydrophobic molecules in the sample interact
ation of enzymes. Aqueous organic solvents are used with the hydrophobic stationary phase of the column
as mobile phases. The amount of organic solvent in and elute later.
the mobile phase is determined by trial and error and The column packing material, particle size and
depends completely on the nature of the molecules to pore diameter of support, column dimension, mobile
be separated. Excessive organic solvent should be phase, and length of hydrophobic ligands, determine
avoided because it can destroy the stability of enzyme the effectiveness of an RPC procedure. Silica is the
molecules. The isoelectric point (pI) of enzymes deter- most widely used support because of its mechanical
mines the type of column used for separation (dis- stability, efRciency, and ability to be bonded with
cussed earlier) as well as the pH of the mobile phase. hydrophobic ligands. However, silica supports are
The net charge on enzyme molecules depends on the not stable under basic conditions (pH'8). Small
agents or buffers are added to the mobile phase to

achieve reasonable retention times. The effectiveness
of these organic modiRers depends on solvent polarity
and increases with decrease in polarity. Isopropanol
is a very good solvent for highly hydrophobic en-
zymes and methanol is better for hydrophilic en-
zymes. Acetonitrile is the most suitable organic modi-
Rer because it has intermediate polarity, low viscos-
ity, and low UV adsorption. It is volatile and can be
easily removed from the eluent. The function of ad-
ded solvents is to decrease the interaction between the
stationary phase and highly hydrophobic molecules
and thus reduce the retention time. Ion-pairing agents
or buffers set the eluent pH and interact with the
enzyme to enhance the separation. TriSuoroacetic
acid (TFA) is widely used as an ion pairing agent.
Buffers such as phosphate or hydrochloric acid are
also used. The mobile phase for RPC typically con-
sists of water, organic solvent, and triSuoroacetic
acid (0.1%) or phosphoric acid.
Capillary columns packed with nonporous (pellicu-
lar) supports have been used for fast separation of
enzymes or proteins at high temperatures and at high
Sow rates. Packed capillary RP-HPLC columns have
several advantages over conventional columns } fast
separation, reduction in solvent usage and the ability
to work with small samples. The mass transfer be-
tween the stationary phase and the mobile phase is
fast with pellicular packings because the diffusional
distances in the stationary phase are short owing to
Figure 2 Anion-exchange separation of Bacillus circulans pro- limited chromatographic interaction at the outer sur-
teins using (A) QM anion exchange (150;3.9 mm) column at face. Capillary columns are stable at higher temper-
1.0 mg load and 0.5 mL min\1 flow rate, (B) 150;19 mm QM atures and at higher pressures because of the solid,
anion exchange column at 30 mg load and 5.0 mL min\1 flow
rate, and (C) DEAE anion exchange (150;21.5 mm) column at
Suid-impervious core of the micropellicular packing.
5 mg load and 5 mL min\1 flow rate. Reproduced from Hamada Rapid mass transfer resulting from the pellicular con-
JS (1995) Journal of Chromatography A 702, 163}172, with Rguration and higher temperature is mainly respon-
permission from Elsevier Science. sible for the fast separation of enzymes. Higher tem-
peratures may not be appropriate for the stability of
support particles favour higher resolution, but high some enzymes.
column backpressures and the large size of enzyme Fast separation of a mixture of four proteins was
molecules to be separated do not favour the smaller performed in 6 s at 1203C on a 3 cm column packed
particle sizes. A particle size of +20 m is optimal with 2 m pellicular ODS-silica (Figure 3).
for RPC column for enzymatic HPLC. A pore size of The biological activity of enzymes is sometimes lost
300 A> is the most commonly available size in com- due to high backpressures, mobile phase (low pH and
mercial RPC columns. For large enzyme molecules organic modiRers) and a strong hydrophobic station-
use of a large pore size (+1000 or greater) is sugges- ary phase. Hydrophobic interaction chromatography
ted to avoid any diffusional problems. (HIC) has less harsh chromatographic conditions
The hydrophobicity of n-alkyl group attached to than RPC and can be used to preserve the biological
the silica decreases with decreasing chain length of activity of enzymes.
n-alkyl group (C18'C8'C4'C2). Smaller n-al-
Hydrophobic Interaction Chromatography
kyl group chains are favoured for highly hydrophobic
samples and vice versa. More hydrophobic columns The basis of separation in hydrophobic interaction
(e.g. C18) require a stronger mobile phase (higher chromatography (HIC) is the same as for RPC. These
amount of organic solvent). Organic solvents such as methods differ in the properties of the mobile and
acetonitrile, methanol or isopropanol and ion-pairing stationary phases. HIC is carried out with an aqueous
enhances the binding between enzymes and weakens

the hydrophobic stationary phase.
Most separation variables in HIC behave in the
same way as in RPC but the nature of the mobile
phases and stationary phases differs in these two
HPLC methods. The HIC performs separation under
nondenaturing conditions whereas RPC denatures
enzymes during separation because of the mobile
phase conditions (organic solvent and highly acidic)
and the highly hydrophobic stationary phase. The mo-
bile phase in HIC is neutral and nonorganic and pro-
tects enzymes from denaturation. Salts such as sodium
or potassium phosphate are added into the mobile
phase to buffer it at pH+7. The bonded phase in
HPHIC is an aryl or smaller alkyl (n(5) group,
weak hydrophobic ligands, attached to silica support.
The puriRcation of Chromobacterium viscosum
lipase has been studied using hydrophobic interaction
chromatography. The stationary phase was prepared
Figure 3 Fast separation of standard mixture of proteins: by covalent immobilization of polyethylene glycol on
1"ribonuclease A; 2"cytochrome c; 3"lysozyme; 4 "lac- Sepharose gel (Sepharose CL-6B, Pharmacia). The
toglobulin B. Column (30;4.6 mm) packed with 2 m pellicular extent of lipase was affected by the salt used and
ODS-silica; 12 s linear gradient from 10 to 90% (v/v) acetonitrile
increases with increasing ionic strength in the eluent
(ACN) in water containing 0.1% trifluoroacetic acid (TFA), tem-
perature"1203C, flow rate"5 mL min\1 and column inlet pres- buffer and with higher pH value. The best recovery of
sure"240 bar. Reproduced from Chen H and Horvath CS (1995) lipase was observed when potassium phosphate was
Journal of Chromatography A, 705, 3}20, with permission from used as a salt compared to NaCl, Na2SO4 and
Elsevier Science. (NH4)2SO4 (Figure 4).
Af\nity Chromatography
solution of higher salt concentration at neutral condi-
tions and uses a weak hydrophobic stationary phase. The basis of afRnity chromatography (AC) is the
The higher concentration of salt in the mobile phase selective adsorption of the molecule to be separated
Figure 4 HIC on PEG 10 000-Sepharose CL-6B column. Buffer: (A) 15% (w/w) K3PO4; (B) 20% (w/v) (NH4)2SO4; (C) 15% (w/v)
Na2SO4; (D) 4 M NaCl in 10 mM phosphate (pH 7). Desorption with 10 mM phosphate buffer (pH 7). From Queiroz JA, Garcia FAP and
Cabral JMS (1996) Journal of Chromatography A 734, 213}219, with permission.
from mixture on the matrix of the column. A speciRc

ligand for a speciRc biological molecule is chosen
and it is covalently bound to the matrix of the col-
umn. For example, a ligand such as adenosine will
bind only enzyme adenosine deaminase and not any
other molecules. When a mixture is applied to the
afRnity chromatography column then the molecule
that has speciRc afRnity for the ligand will stay in the
column and all other unbound molecules will migrate
through the column. The interaction between adsor-
bed molecule and ligand is reversible. Changing the
pH or other conditions of the mobile phase can de-
sorb the bound molecule. A molecule that has more
afRnity for the ligand than the bound molecule can be
Figure 5 Separation of a protein mixture: a"transferrin,
included in the mobile phase to elute the desired b"myoglobin, c"lysozyme, and d"cytochrome c. Packing:
bound molecule. This chromatographic method is Nucleosil 1000-5 DIOL CaP-HA 2.5% (100;6 mm). Linear gradi-
carried out under nondenaturing conditions during ent of sodium phosphate (pH 6.8), 1}350 mM (60 min); flow rate
the separation of enzymes or proteins. " 1mL min\1, temperature"253C. Reproduced from Bruno G,
The type of ligand and its support, state of mobile Gasparrini F, Misiti D, Arrigonimartelli E, and Bronzetti M (1990)
Journal of Chromatography A 504, 319}333, with permission from
phase at each stage of separation and Sow rate deter- Elsevier Science.
mine the resolution in AC. Ligands can be speciRc for
a molecule or a group of molecules. The desired between amine groups of the enzymes and phosphate
ligand must be highly speciRc for enzyme molecule(s) groups on the surface of the hydroxyapatite and, also,
to be separated, be stable under applied conditions, calcium coordination complex formation between
have reversible binding with applied sample and pos- calcium groups in the hydroxyapatite and carboxyl
sess an appropriate functional group to couple with groups in enzymes. Low concentration phosphate
the support. Cross-linked agarose or other pressure- buffers are used to elute acidic and neutral enzymes
stable polymer is used as a support for the ligand. and high concentration buffers are used to elute basic
Buffers at each step of separation must be non- enzymes.
denaturing to maintain speciRcity of ligand and elut- A protein mixture containing transferrin, myo-
ing enzyme(s). Low pH buffers are employed in the globin, lysozyme, and cytochrome c was separated
desorption step to break the solute}ligand inter- using hydroxyapatite as a support. Protein solution
action. SpeciRc desorbing agents, which compete (10}50 L; 1 g L\ protein) was loaded onto the
with adsorbed solute molecule(s) for the same bind- column and eluted with a linear gradient of sodium
ing site, are sometimes also used. The Sow rate of phosphate buffer (pH 6.8) (Figure 5).
mobile phase also affects the retention time and peak
Perfusion Chromatography
shape.
Alhama et al. have applied AC technique for the Perfusion chromatography is a new chromatographic
puriRcation of glutathione reductase and glucose-6- technique, introduced by Afeyan and coworkers in
phosphate dehydrogenase from cell-free extract of 1989}1991, for reducing resistance to stagnant mo-
baker’s yeast, Rsh liver, and rabbit hemolysates with bile phase mass transfer in liquid chromatography. It
high recovery. They used an epoxy-activated silica may be used for both rapid analysis and preparative
column derivatized with the ligand 8-[(6- chromatography of large molecules such as enzymes.
aminohexyl)amino]-2-phosphoadenosine-5-diphos- A new chromatographic packing material (POROS,
phoribose. The bound ligand concentration was Perseptive Biosystems) which contains two sets of
11.4 mol g\1 of dry silica and the loading capacity interconnecting bimodal pores has been employed in
was 2}3 mg of glutathione reductase. perfusion chromatography. The members of one pore
set having a mean diameter in the range 6000}8000A>
are called throughpores. The high surface area needed
Hydroxyapatite High-Performance Liquid
for adequate sample capacity is achieved by the small-
Chromatography
er diffusive pores (dpore+1000 A> ). The mobile phase
Hydroxyapatite, Ca10(PO4)6(OH)2, is a form of cal- Sows through the through-pores. In this manner, sol-
cium phosphate which has been used, in particular, utes enter the interior of the particles convectively by
as a packing material for enzymes and proteins separ- the through-pores and then diffuse into the diffusive
ations. The basis of separation is ionic interactions pores.
simple to use and relatively inexpensive. The analysis

is also nondestructive and hence suitable for prepara-
tive work. Furthermore the solvents best suited for
liquid chromatography are transparent to UV-vis.
Refractive index detectors are nondestructive, con-
centration sensitive and universal in that they respond
to virtually all compounds with the proper choice of
mobile phase but are of low sensitivity. The speciRcity
of analysis of enzymes by Suorescence detection
arises because many parameters related to the Suores-
cence intensity can be exploited. However, this tech-
Figure 6 Separations of hybridoma cell cultural supernatant on nique requires the use of very selective regents that
protein A POROS M. 0.5 mL injection; 30;2.1 mm column; react with speciRc functional groups of the analyte to
10 mM phosphate pH 7.4, 0.15 M NaCl; elution with 0.3 M acetic
produce Suorescent derivatives.
acid (2%, v/v), 0.3 M MgCl2; 2 mL min\1 flow-rate. Reproduced
from Afeyan NB, Fulton SP and Regnier FE (1991) Journal of
Chromatography A 544, 267}279, with permission from Elsevier Future Trends
Science.
HPLC will continue to be the important tool for
POROS-based chromatographic packing material separation of enzymes. The new capillary columns
can be used in any chromatographic mode such as ion packed with nonporous support and microsporous
exchange, reversed-phase, hydrophobic interaction, support in perfusion chromatography will be helpful
and afRnity chromatography. These supports separ- in fast analysis of enzymes or proteins. Separation is
ate proteins rapidly compared with conventional faster and more selective when HPLC is carried out at
HPLC using higher mobile phase velocities. higher temperatures. A heat exchanger, which can
The separation of immunoglobulin G(IgG) from bring the eluent rapidly to column temperature, will
hybridoma cell culture supernatant has been com- increase separation reliability at higher temperatures.
pleted in 80 s using a POROS protein A (aldehyde- Conventionally, HPLC is used for the analytical sep-
coupled) column. The sample was loaded in a 0.1 M aration as well as for preparative separation of en-
phosphate buffer pH 7.4 with 0.15 M NaCl and zymes. Discontinuity of the HPLC process and the
eluted with 0.3 M acetic acid (2%, v/v) with 0.3 M dilution of the products after elution are two major
MgCl2. The column was loaded with 0.5 mL of hy- disadvantages. The simulated moving bed (SMB)
bridoma cell culture supernatant with the Sow rate of technique can make HPLC a continuous process.
2 mL min\ (Figure 6). A column packing material should be designed for
a higher sample loading and for fast HPLC. Thus,
a large-scale separation should be fully automated
Assay of Enzymatic Activity and continuously operating, loading samples, collect-
An important aspect of the separation of enzymes by ing fractions, regenerating the column and with
HPLC is the assay of enzyme activity. Enzyme- various fail-safe devices to protect the column and
catalysed reactions can be monitored spectro- product.
photometrically and many of the substrates or
products absorb visible or UV light. It allows deter- See also: II/Affinity Separation: Theory and Develop-
mination of the progress of a reaction by direct and ment of Affinity Chromatography. Chromatography:
continuous monitoring. While other methods of dis- Liquid: Mechanisms: Ion Chromatography; Mechanisms:
continous assay focus on monitoring one of the Reversed Phases; Mechanisms: Size Exclusion
compounds of the reaction, the HPLC technique Chromatography. III/Peptides and Proteins: Liquid
Chromatography.
offers the simultaneous determination of several sub-
strates of the reaction. This method is probably the
best as it offers a complete mass balance of the reac- Further Reading
tion being analysed.
Afeyan NB, Fulton SP, and Regnier FE (1991) Journal of
Chromatography A 544: 267}279.
Detectors for Enzyme Analysis Alhama J, Lopezbarea J and Toribio F (1991) Journal of
Ultraviolet-visible (UV-vis) detectors are the most Bruno G, Gasparrini F, Misiti D, Arrigonimartelli E and
commonly encountered detectors in enzyme analysis Bronzetti M (1990) Journal of Chromatography A 504:
because enzymes are UV-active and UV detectors are 319}333.
III / ESSENTIAL OILS / Distillation 2739
Chen H and Horvath CS (1995) Journal of Chromatogra- Queiroz JA, Garcia FAP, and Cabral JMS (1996) Journal of
phy A 705: 3}20. Chromatography A 734: 213}219.
Hamada JS (1995) Journal of Chromatography A 702: Rossomando EF (1991) High Performance Liquid
163}172. Chromatography in Enzymatic Analysis. John Wiley
Mant CT and Hodges RS (eds) (1991) High-Performance and Sons, Inc., New York.
Liquid Chromatography of Peptides and Proteins: Sep- Wiseman A (ed.) (1985) Handbook of Enzyme Biotechnol-
aration, Analysis and Conformation. Boca Raton: CRC ogy. New York: John Wiley and Sons, Inc.
Press, Inc.
ESSENTIAL OILS
century. In 1507, Hieronymus Brunschwig’s book on

Distillation distillation, Liber De Arte Distillandi, described dis-
tillation techniques for four essential oils, namely,
turpentine (known since antiquity), juniper wood,
E. Hernandez, Texas A&M University, rosemary and spike.
College Station, TX, USA
Before the ninth century it was still widely believed
Copyright ^ 2000 Academic Press that most essential oils had strong curative properties.
Therefore it was chieSy pharmacists who developed
and improved methods of distillation for the recovery
Introduction and puriRcation of natural essential oils.
Essential oils are generally understood to be volatile Eventually, with the development of the Relds of
compounds which are freely soluble in alcohol, ether medicine and pharmacology and the dispelling of
and vegetable and mineral oils and are usually as- some medicinal myths, the use of essential oils in
sumed to be the result of distillation or a steam- pharmaceutical products lost importance and their
stripping process. The use and processing of essential use became restricted to perfumes, beverages and
oils began in the East more than 2500 years ago. The foodstuffs.
process of distillation, which is the technical basis of
the essential oil industry, was also conceived and Rrst
employed in the Orient, especially in Egypt, Persia
Applications of Essential Oils
and India. Turpentine and camphor appear to be the Attractive aromas which leave a pleasant memory
Rrst documented essential oils prepared by distillation association are used as marketing devices to sell ed-
in Greece by Herodotus (484}425 BC). ible and cosmetic products, including unlikely mater-
The use of essential oils in ancient times consisted of ials such as detergents. The producer is counting on
preparing ointments by mixing oils from Sowers with the consumer preferring a product that left a pleasant
fatty oils; this was done by placing Sowers and roots aromatic memory.
with the oil in glass bottles which were then allowed Current speciRc uses of essential oils are to add
to sit for periods of time. Sometimes the Sowers or Savour to foodstuffs and beverages and to scent per-
roots were macerated with wine before the fatty oil fumes, lotions, soaps, detergents and household
was added, and the product obtained by digestion cleaners. For example, d-limonine from citrus peel is
Rltered and boiled down to a thicker consistency. a very strong solvent and it is used in a wide variety of
Medieval alchemists laboured for many years to cleaning products. Essential oils are a major part of
extract from materials found in nature what they carbonated beverage Savourings; the most common
called the quinta essentia or the Rfth essence. They Savours include lemon, lime, orange, cassia, cinna-
believed that a combination of earth, air, Rre and mon and nutmeg. Essential oils are also used to Sa-
water existed in some form or other from which quint- vour many foods such as sweets and candies, cookies,
essential materials could be extracted from some snacks and chewing gum.
plants. These quintessential extracts derived from The Reld of aromatherapy constitutes a small part
plants were believed to be remedies for a wide variety of the essential oils industry but it is a fast-growing
of diseases. area and requires a wide variety of essences. Not
The production and use of essential oils did not much scientiRc work has been done in this area to
become widespread until the second half of the 16th support any of the medicinal and psychosomatic
2740 III / ESSENTIAL OILS / Distillation
claims. Practitioners suggest that aromatherapy goes Table 1 Classification of components in essential oils
beyond the effect of simply imparting an agreeable
sensation and psychological state of well-being. Some Component Examples
speculate that inhaling certain essential aromas can Hydrocarbons d-Limonene in lemon oil
affect the limbic system, producing a meas- Alcohols Borneol in camphor
urable physiological response. More research is Esters Methyl salicylate in oil of wintergreen
certainly needed to document the beneRts of this Aldehydes Benzaldehyde, decanals
Ketones Menthone in oil of peppermint
application.
Lactones Coumarin from Tonka beans
Other properties of essential oils with commercial
potential include antimicrobial effects. The inhibition
of 25 different bacteria using an essential oil of mar-
joram has been reported. Similar effects are noted for proRles of many aromas and fragrances which permit
other volatiles and essences derived from plant mater- the establishment of composition standards for trade
ials. It was found that the short chain volatiles such as regulations and for the synthesis of aromas using less
5}8 carbon aldehydes and ketones resulting from the costly starting raw materials.
distillation of vegetable oils had antimicrobial prop- Essential oils are commonly grouped into six
erties against bacteria such as Staphylococcus aureus classes according to their chemical nature. Tables 1
and Escherichia coli. In fact, this antimicrobial effect and 2 list some of the most commonly utilized essen-
is believed to be a defence mechanism in plants tial oils worldwide.
against microbial pathogens. Another function of es- The most common essential oil used as a food
sential oils in plants is reported to be as an attractant aroma and Savour is orange essence (Table 2). Over
of insects, enabling plants to use the insects as pollen 50% of the commercial essential oils and natural
carriers for plant reproduction. extracts are obtained from cultivated plants. Exam-
The production of essential oils on a larger scale ples of these include mint aroma and Sower essences
was started in the USA in the earlier 19th century. such as rose, geranium, mints, coriander, lavender
Three indigenous plants, sassafras, American worm- and jasmine.
seed, and wintergreen, as well as turpentine, were the Citrus oils such as orange, lemon and grapefruit
Rrst oils to be produced in the USA in large amounts aromas are considered by-products of juice extraction
for export worldwide. and concentration. Orange oil, for example, is mar-
Many aromatic plants for essential oils grow wild keted both as a Savour and as a material for use in
or are cultivated in small scale family-oriented busi- cleaning products. In this case the distinction has to
nesses. Today only a few essential oils are produced be made that orange oil and citrus oils in general are
by modern or centralized methods. An example of recovered in two ways; one way is as d-limonine, in
these is the cultivation and harvesting of aromatic which the oil resulting from the peel of the fruit as the
Sowers in the Grasse region of southern France,
where essential oil distillation units are placed near Table 2 Main essential oils produced in the world
the Relds to extract and recover the essential oils on
site. In the case of citrus oils, for example, much Essential oil Main components
larger scale distillation systems are set in place at Orange Limonene, terpeniols
juice-processing plants. These aroma distillation units Mint Linalool, linalil acetate
are set up next to the evaporators where the concen- Eucalyptus Cineole, pinene, limonene
tration of juices is taking place. These systems are Citronella Geraniol, citronellal, citronelol
Clove Eugenol, caryophylene
common in the USA as well as in other large citrus-
Lime Limonene, terpeniols
producing countries such as Brazil and China. It Spearmint Mentheol, menthone, pulegone
should be noted that these aroma distillation units are Lavander Pinene, lemonene, caphene, octanone
used not only in citrus-processing plants but in any Marjoram Tojuene, pinene, sabinene
plant that concentrates fruit juices by evaporation. Camphor Bisabolol, cadinol cubenol
Coriander Terpinene, p-cymen, pinene
In addition to improving production and puriRca-
Patchouli Patchoulool, sesquiterpenes
tion processes for the recovery of essential oils, the Rose Citronellol, geraniol, linalool
essential oil industry has been active in developing Cinnamon Fenchene, cinnamaldehyde, pinene
synthetic aromas. Essential oils are some of the most Sandalwood Santalene, curcumene
studied chemical compounds as regards their com- Lemongrass Citral, linaloo, geraniol
Jasmin Benzyl acetate, linalool, benzyl benzoate
position and physical and chemical properties. Ad-
Ginger Terpenio, neral, geraneal
vances in organic chemistry have allowed for the Anis Trans anethol, chavicol
establishment of techniques to deRne the component
juice is extracted. This has a lower value and is sold as in the recovery of essential oils are Sowers, roots,
Savouring and also as a cleaning agent. The true seeds, leaves and twigs of aromatic plants.
volatiles from citrus are what results from volatiliz- Once the plant materials have been prepared, the
ation during the evaporation of the juice and this is method for removing the essential oil depends on the
a product that commands a higher value and has type of product being handled. For example, a combi-
a wider range of aroma components such as al- nation of boiling water and steam injection is used
dehydes, ketones and terpenes. Other oils such as with Sower petals in order to avoid agglutination of
clove or pepper oil may come from spices, or as oils, the materials in the distillation unit. Figure 1 illus-
oleoresins, extracts or Savours. trates this simple method of distilling essential oils.
Other raw materials for essential oil production are This type of distillation is commonly set up in the
harvested in the wild. Wild thyme and rosemary, for Relds where the raw materials are being harvested in
example, are common in Spain and grow back abun- order to prevent spoilage and deterioration of aromas
dantly following harvest. Sumatran cinnamon trees during transportation or storage.
are similarly self-rejuvenating. However, due to high A simple distillation apparatus consists of a retort or
demand, some countries have resorted to overhar- still, a condenser and a receiver. This system is com-
vesting plants for essential oil production from the monly used in the Relds for processing lavender. Typi-
wild. This has created local problems of forest de- cal conditions for this method are to heat a mixture of
struction and the sustainability of the industry has raw material with water to boiling point (1003C)
become a serious concern for some underdeveloped while injecting steam. The resulting vapours are then
countries. As a result, some countries, such as Brazil, condensed and recovered. This usually results in two
have implemented harvest moratoriums or banned phases: a heavy phase, mostly composed of water,
the harvesting of some species in order to avoid wip- and a light phase, which contains the essential oil.
ing out some species of aromatic plants. These con- A variation of this method is to apply vacuum
servation efforts and replanting programmes have (10}25 in Hg) to the system to allow boiling to take
meant that some oils are now available again com- place at a lower temperature and thus prevent the
mercially. thermal decomposition of the essential oil.
Some aroma systems involve the use of only steam
in order to obtain a more concentrated essential oil in
Recovery of Essential Oils the condenser. This system is applicable where there
The aroma and fragrance industry has been estimated is no agglomeration or agglutination of raw mater-
at approximately $2 billion per year, with a growth ials. When the raw material is liquid, such as fruit
rate of 3.5% per year. Orange oil is both the largest juice or macerated materials in water, the fresh liquid
volume oil and, as a by-product of the orange juice Sows in countercurrent with vapour. Vapour}liquid-
industry, relatively inexpensive, ranging from $0.75 contacting devices such as a sieve plate column or
to $1.40 a pound. Although the US production of a packed column are sometimes used. Under ideal
citrus oil is declining, with South America and China conditions, it is advisable to have a rectiRcation col-
growing in importance as a major producer, the USA umn next to the distillation unit in order to obtain
continues to be a major producer of mint and cedar a more concentrated and pure essential oil. This is the
oils. The mint oils as a group } peppermint, spearmint most efRcient way from the standpoint of steam con-
(both grown in the USA) and cornmint (produced in sumption as well. Such a scheme is illustrated in
India and China and as a by-product of menthol Figure 2, with distillation/rectiRcation as the concen-
production) } are the highest value oils, with pepper- tration step.
mint selling for about $12}15 a pound. As mentioned above, essential aromas from citrus
Distilled essential oils are generally recovered by are recovered as a by-product from the production of
three methods, classiRed according to the way heat is citrus juice concentrates. These juices are usually con-
applied to materials: boiling water, steam distillation centrated by multiple effect evaporation. There are
or a combination of both. Conditions of temperature, four or more effects where the Rrst effect is usually
pressure or vacuum, and processing time will depend heated by live steam and then subsequent effects are
on the characteristics of the essential oil, particularly heated by the steam generated in the preceding effect.
as regards susceptibility to oxidation and heat de- The vapours generated in the Rrst effect are the
composition. richest in the essential oils and the essence is conden-
The basic process consists of macerating or com- sed and recovered in a distillation column. This is the
minuting the plant materials to rupture the oil sacs. most important source of essential oils with regard to
This allows the essential oil to be exposed and carried volume, estimated worldwide at 25 000 metric tons
by the steam or water used. Common materials used and more than 60 million US dollars.
2742 III / ESSENTIAL OILS / Distillation
Figure 1 Aroma recovery unit by hydrodistillation.
There is another method of distillation used by very Other methods for recovering aromas and fragran-
specialized Savour companies called molecular distil- ces are solvent extraction (hexane, alcohols) and
lation. In this evaporation system the liquid introduc- supercritical CO2 extraction. Strictly speaking, the
ed is spread on the walls of a heated vertical cylinder resulting oils from these extraction methods are not
under high vacuum (less than 50 mmHg) by wiper essential oils since no evaporation or distillation takes
blades forming a thin layer. This produces highly place. However, recovery of high priced aromas and
efRcient mixing and heat transfer of the Suid, reduc- fragrances by supercritical extraction is growing
ing the residence time to a minimum. Variable tem- worldwide and extraction of some low price essential
perature allows for the fractionation of aroma com- oils is also done using hexane or alcohol. Supercritical
ponents according to their molecular weight. Heat- extraction has the advantage that it does not involve
sensitive aromas and specialized aromas such as dairy the application of high heat and addition of steam or
Savours are manufactured by the industry using this water. Also, the extractability properties of CO2,
type of distillation system. namely afRnity for hydrophobic or hydrophilic com-
Figure 2 Aroma recovery unit with distillation column.

pounds, can be manipulated with pressure and tem- vour, are more water-soluble and are easier to blend
perature. Extraction methods by steam distillation, in beverages and foodstuffs.
solvent extraction and supercritical CO2 usually pro- Well-known proRles of components of essential
duce different results with regards to yield and com- oils are instrumental in setting standards for the as-
position proRles of the extracted materials. Steam sessment of quality and prevention of adulteration.
distillation usually gives the lowest yield of re- Knowing the chemistry of essential oils has allowed
covered aroma but produces a concentrate that is the successful manufacture of synthetic essential oils;
a true essential oil. Solvent extraction with hexane however, some essential oils are so complex that
or alcohols produces the highest yield but there is the odour and Savour characteristics just cannot be
always the possibility that unwanted involatile mater- duplicated.
ials might end up in the Rnal product. Also, solvent
See Colour Plates 81, 82.
extraction uses high temperatures. Supercritical
CO2 extraction produces a lower yield than con- See also: II/Chromatography: Gas: Headspace Gas
ventional solvent extraction but higher than steam Chromatography. Distillation: Extractive Distillation.
distillation and it also has the advantage of using Extraction: Solvent Based Separation. III/Citrus Oils:
little or no heat and no steam or moisture in the Liquid Chromatography.
process. However, this method has the disadvant-
age of being expensive since it involves high pressure
and sophisticated equipment and controls. In some Further Reading
cases the quality of the aroma extracts from CO2 Abdallah MA, Foda YH, Saleh M, Saki MSA and Mostafa
extraction is superior to aromas recovered by steam MM (1975) IdentiRcation of the volatile constituents of
distillation. the Egyptian lemmongrass oil. I. Gas chromatographic
analysis Nahrung 19: 195}200.
Azzouz MA, Reineccius and Moshonas MG (1976) Com-
parison between cold-pressed and sitilled lime oils
Industry Standards and Trends through the application of gas chromatography and
mass spectrometry. Journal of Food Science 41:
With the exception of large companies that have
324}328.
access to sophisticated analytical means, the assess- Bednarczki AA and Kramer A (1975) IdentiRcation and
ment of quality of essential oils is difRcult. However, evaluation of the Savour-signiRcant components of gin-
the composition of many essential oils is well known ger essential oil. Chemical Senses Flavour 1: 377}386.
and the industry can adopt standards that can be Deans SG and Svoda KP (1990) The antimicrobial proper-
applied to essential oils. Oils used in medical applica- ties of marjoram (Origanum majorana L.). Journal of
tions, for example, camphor oil, must meet the stan- Flavour and Fragrances 5: 187.
dards set forth in the US pharmacopeia. Oils used in Garnero J, Guichard G and Buil J (1976) L’huile essentiele
food products must meet the Food Chemical Codex et le concrete de rose de Turquie Parfumes, Cosmetiques
standards and the Fragrance Manufacturers Asso- et Savons 8: 33.
ciation (formerly the Essential Oil Association, or Gill LS, Lawrance BM and Morton JK (1973) Variation in
Mentha arvensis L. (Labiatae). The North American
EOA) is in the process of updating a speciRcations
Populations. Botanical Journal of the Linnaean Society
book. 67: 213}232.
It is common practice for large producers of essen- Gunther E (1972) The Essential Oils. Huntington, NY: RE
tial oils to process their oils further following pressing Krieger Publ. Co.
or distillation, either to produce a standard pro- Hernandez E and Rathbone SJ (1998) Properties of de-
duct year after year, or to change the characteristics odorizer distillate byproducts recovered in a molecular
of the oil. For example, peppermint oil is commonly still. Annual Report-Food Protein R&D Center. College
redistilled to remove some of the front-end compo- Station, Texas: Texas A&M University.
nents that give the oil an unwanted Savour or aroma Ikeda RM, Stanley WI, Vaniere SH and Spittler EM (1962)
note. Monoterpene hydrocarbons of some essential oils. Jour-
Another common practice is folding of oils, or nal of Food Science 27: 455}458.
Moyler DA (1996) Commerical extraction of Savours and
concentrating them by removing certain components.
perfumes. Journal of Chemical Technology and Biotech-
It is common practice to redistil citrus essential oils, nology 65: 296.
in some cases removing up to 90% of the original Penfold AR and Willis JL (1961) The Eucalyptus. London:
volume in order to remove most of the unwanted Leonard Harris Ed.
terpenes. This is known as terpeneless citrus oils, Saravacos GD and Moyer JC (1968) Volatility of some
which carry a higher price in the essence markets. aroma compouns during vaccum-drying of fruit juices.
These folded oils are more stable, have a better Sa- Food Technology 22: 623}628.
2744 III / ESSENTIAL OILS / Gas Chromatography
Shaw P and Coleman RL (1975) Compositions and Savour Teisseire P, Maupetit P and Corbier B (1974) Contri-
evaluation of a volatile fraction from cold pressed val- bution to the knowledge of Patchouli oil. Reserches 19:
encia orange oil. International Flavours Food Additives 8}35.
6: 190}193. Thijssen HAC (1970) Concentration processes for liquid
Steltenkamp RJ and Cazzasa WT (1967) Composition of food containing volatile Savors and aromas. Journal of
the essential oil of lavandin. Journal of Agriculture and Food Technology 5: 211}229.
Food Chemistry 15: 1063}1069. Urdang G (1943) Pharmacy in Ancient Greece and
Tabacchi R, Garnero J and Buil P (1974) Contribution Rome. Amercian Journal of Pharmaceutical Education
a l’etude de la composition de l’huile de fruits d’anise de 7: 169.
Turque. Revista Italiana 56: 683. Van der Gen A (1972) Corps olfactifs a l’odeau de jasmin.
Takaoka D, Takaoka A, Ohshita T and Hiroi M (1976) Parfumes, Cosmetiques et Savon 2: 356}370.
Sesquiterpene alcohols in camphor oil. Phytochemistry Virmani OP and Datta SC (1971) Essential oil of Cym-
15: 425}426. bopogon winteranius (Oil of Citronella, Java). Flavour
Taskinen J (1974) Composition of the essential oil of sweet Industry 3: 595}602.
marjoram obtained by distillation with steam and by Walker GT (1968) Sandalwood oil. The chemistry of oil of
extraction and distillation with alcohol-water mixture. sandalwood. Perfumes and Essential Oil Research 59:
Acta Chemica Scandinaria B28: 1221}1228. 778}785.
Gas Chromatography
C. Bicchi, University of Turin, Turin, Italy separation of enantiomers; multidimensional GC;
identiRcation of essential oil components through GC
and/or combined techniques (GC-MS, GC-FTIR);
GC-Isotope ratio mass spectrometry and authenticity
An essential oil is internationally deRned as the prod- of an essential oil; GC-snifRng for sensory evaluation;
uct obtained by steam distillation, hydrodistillation and statistical analysis applied to GC proRles.
or expression (for citrus fruits) of a plant or of a part
of it. This deRnition is now less strictly applied, and
the fractions resulting from several other techniques Sample Preparation
that sample the volatile fraction of a plant are now Steam Distillation and Hydrodistillation
erroneously classiRed as essential oils. In general it
would be more correct to call them volatile fractions An essential oil is classically obtained by steam or
of a vegetable matrix, and to use the term essential oil hydrodistillation via equipment based on the circula-
more speciRcally for samples obtained by distillation tory distillation apparatus introduced by Clevenger in
or expression. In addition to distillation or expres- 1928. Apparatus and operation modes are now well
sion, the volatile fraction of a vegetable matrix can be established. Several pharmacopoeias give diagrams
obtained through static or dynamic headspace gas and instructions of how to obtain essential oils. Fig-
chromatography (HS-GC), solid-phase microextrac- ure 2 is taken from the European Pharmacopoeia.
tion (SPME-GC), simultaneous distillation}extrac- On the other hand, sampling techniques for the
tion (SDE), solvent extraction or supercritical Suid volatile material are under constant evolution. The
extraction (SFE). most used techniques are static or dynamic HS-GC,
Components of an essential oil are generally me- SPME/GC, SDE and SFE.
dium-to-highly volatile with medium-to-low polarity,
Headspace Sampling (HS-GC)
and as a consequence GC is the technique of choice
for their analysis. Figure 1 shows the structure of Static HS-GC, dynamic HS-GC HS is a sampling
some typical essential oil components. These charac- technique applied to the determination of volatiles in
teristics also facilitate their identiRcation, which in the gaseous phase in equilibrium with the matrix to
general can be achieved by combining chromato- be sampled.
graphic (retention indices) data with mass spectro- HS-GC sampling is generally classiRed as static or
metry (GC-MS) and Fourier transform infrared dynamic HS. In static HS-GC, the analyte is sampled
spectroscopy (GC-FTIR). from a hermetically sealed vial after the matrix
This article aims to cover the main aspects related has reached equilibrium with its vapour at a pre-
to the analysis of essential oils, in particular with determined temperature. Figure 3A shows the static
sample preparation techniques related to GC; GC HS-GC pattern of a sage sample. The sample was
III / ESSENTIAL OILS / Gas Chromatography 2745
Figure 1 Structure of some typical essential oil components. 1, Myrcene; 2, t-ocimene; 3, geraniol; 4, nerol; 5, linalol; 6, linalyl
acetate; 7, limonene; 8, -terpineol; 9, terpinen-4-ol; 10, terpinyl acetate; 11, 1,8-cineole; 12, menthone; 13, menthol; 14, menthyl
acetate; 15, menthofurane; 16, carvone; 17, camphor; 18, borneol; 19, bornyl acetate; 20, i -borneol; 21, t--farnesene; 22, (Z,Z )--
farnesene; 23, (Z,E )--farnesene; 24, (E,Z )--farnesene; 25, germacrene D; 26, chamazulene; 27, (!)--bisabolol; 28, bisabolol
oxide A; 29, bisabolol oxide B; 30, spiroether; 31, anethole; 32, estragole; 33, eugenol; 34, methyl eugenol.
ax TA cartridge through a nitrogen Sowstream of

30 mL min\1 for 2 min.
Solid-phase microextraction SPME is a sampling

technique based on absorption developed by Arthur
and Pawliszyn. With SPME, the analytes are ab-
sorbed from the liquid or gaseous sample on to an
absorbent coated fused silica Rbre, which is part of
the syringe needle, for a Rxed time. The Rbre is then
inserted directly into a GC injection port for thermal
desorption. SPME is a solvent-free technique which is
sensitive because of the concentration factor achieved
by the Rbre, and selective because of the different
coating materials which can be used. One of the
advantages of SPME is the possibility to sample dir-
ectly the vapour phase in equilibrium with the matrix
(headspace (HS)-SPME), or the matrix extract or
solution (liquid sampling-SPME) directly, provided
that suitable Rbres are used. Figure 3C shows the
SPME-GC pattern of the same sage sample already
analysed. The dried sage leaves are equilibrated as for
static headspace sampling for 1 hour with a 100 m
polymethylsiloxane-coated Rbre.
All these headspace techniques are easy to auto-
mate and to standardize. This is particularly true for
static HS-GC and SPME-GC which can be used for
fully automatic routine analysis. Static HS-GC is
highly reliable for quantitative analysis, when asso-
ciated with the multiple headspace extraction method
developed by Kolb. Dynamic HS-GC is also quite
easy to standardize, now that automatic purge-and-
trap systems are commercially available. However,
reproducible dynamic HS sampling is conditioned by
Figure 2 Apparatus for the determination of essential oils in
vegetable drugs (European Pharmacopoeia (2000) 3rd edn, a large number of parameters (volume to be sampled,
Copyright Council of Europe). Plant material suspended in water volume of the headspace system, sampling time and
is heated to boiling; the resulting vapour, consisting of a homo- speed, carrier Sow rate, trapping material, including
geneous mixture of essential oil and steam, is then condensed in batch and producer, kinetics of component release in
the refrigerator (F,G) and recovered in the collecting bubble (J); different matrices) that make it quite difRcult to com-
two layers are formed, the upper with the essential oil and the
lower with the aqueous phase, the latter being continuously recir- pare results from different laboratories.
culated through the (M}B) tubing. The different HS sampling techniques are normally
used for different applications: in general, static HS-
equilibrated for 1 h at 603C and 1 mL of the gas GC is suitable for the analysis of highly volatile frac-
phase in equilibrium with the vegetable matrix was tions, HS SPME-GC is suitable for the analysis of
automatically injected and analysed by GC. medium-volatile fractions, while dynamic HS-GC is
In dynamic HS-GC, the sample is obtained by cap- used for trace analysis or for very diluted headspace.
turing the volatiles in a gaseous efSuent passed
through or over the matrix on to a suitable trapping
system, such as cryotraps, solid adsorbents, liquid The high selectivity of supercritical Suids, together
stationary phases or selective reagents for a given with the low polarity and molecular weight of most
class (or classes) of compounds, coated on a solid of the volatile fraction components, permits low ex-
support. The trapped volatiles are then recovered traction pressure and temperature to be used, thus
through heat or solvent elution either on-line or off- limiting the classes of the extracted components to
line to the gas chromatograph. Figure 3B shows the those that characterize an essential oil (mono- and
GC pattern of the same sage sample as in Figure 3A. sesqui}terpenoids, phenylpropanoids and aliphatic
The volatile fraction was transferred to a 50 mg Ten- oxygenated compounds). This often makes the com-
position of SFE extracts quite similar to that of the GC Analysis

corresponding essential oil obtained through hydro-
Classical GC Analysis
or steam distillation. In addition, several organolepti-
cally important components that are water-soluble Essential oils are generally analysed by capillary GC.
and which are generally lost in the water phase during The most popular stationary phases used for essential
the steam distillation are quantitatively recovered by oil analysis are methylpolysiloxanes (SE-30, OV-1,
SFE. Typical is the case of phenylethanol, which is the OV-101, DB-1, PS-347.5) and methylphenyl-
main component in a rose SFE extract, while it is polysiloxanes (SE-52, SE-54, PS-086, DB-5) as
a minor component in the corresponding essential oil. apolar stationary phases; and polyethylene glycol
Figure 3 Capillary GC patterns of (A) the static HS-GC; (B) dynamic HS-GC and (C) HS SPME-GC of a sample of dried sage leaves.
Analysis conditions: column 15 m, 0.25 mm i.d. OV-1, df: 0.3 m; temperature programme: (A) from !103C (10 min) to 303C at
303C min\1 then to 1503C at 33C min\1 and to 2003C at 53C min\1; (B) and (C) from 303C to 1503C at 33C min\1 and to 2003C
at 53C min\1. Peak identification: 1, -pinene; 2, camphene; 3, -pinene; 4, myrcene; 5, limonene; 6, 1,8-cineole; 7, -thujone;
8, -thujone; 9, camphor; 10, iso-borneol; 11, borneol; 12, bornyl acetate; 13, -caryophyllene; 14, -humulene.
Figure 3 Continued
(Carbowax 20M) as a polar stationary phase. GC cy of columns prepared by high temperature silyla-
data are also very useful to identify most of the tion; the possibilities of tuning column polarity by
components in an essential oil: an effective approach using different diluting phases; the small CD amounts
is to combine the retention data from two different- necessary to prepare columns; shorter analysis times;
polarity stationary phases (see below). and the possibility of measuring the thermodynamic
parameters involved in enantiomer discrimination.
Enantiomer GC Analysis
Almost all the essential oil components can now be
One of the most important successes of the last 10 separated on CD stationary phases without derivatiz-
years has been enantioselective GC recognition of ation. This is particularly true for the so-called sec-
chiral essential oil components with cyclodextrin de- ond-generation CDs, developed especially for GC,
rivatives (CDs). The importance of enantiomer separ- which show high enantioselectivity, and afford highly
ation and of determining enantiomer excess is well reliable column performance. The most successful
known. Biosynthetic and geographical origins, as CDs are symmetrically and asymmetrically alkylated
well as technological treatments and/or authenticity CDs, acylated CDs and CDs asymmetrically sub-
of most of the essential oils, can now also be evalu- stituted in position 6 with the groups tert-butyl-
ated through the enantiomeric composition of their dimethylsilyl or thexyldimethylsilyl, and in positions
underivatized optically active components. This is 2 and 3 with methyl, ethyl or acetyl groups. In gen-
also important because optical isomers can have dif- eral, the most popular matrices for the CDs are poly-
ferent sensory properties, such as the well-known phenylcyanospropylsiloxes (including various OV-
cases of the different smells of both carvone and 1701 types), polyphenylsiloxanes (including SE-52 or
limonene enantiomers. The Rrst GC separations PS-086) and methyl polysiloxanes (including SE-30,
of enantiomers through CDs were obtained by OV-1 and PS-347.5). The latest generation of CDs
Koscielski and Sibilska in 1983; they separated - and makes it possible to characterize an essential oil by
-pinene, the corresponding pinanes and -3-carene determining the enantiomer abundances of several of
with a column packed with underivatized -CDs. The its optically active components simultaneously, very
Rrst capillary column applications were in 1987 with often in a single GC run. Figure 4 shows the simulta-
the almost contemporary work of Juvancz and neous enantiomer separation of optically
Schurig. CDs are generally carried in apolar to mod- active components characterizing lavender oil:
erately polar polysiloxanes, as Rrst proposed by linalyl oxides, linalol, linalyl acetate, camphor,
Schurig. The chief reasons for this are the wider range borneol, bornyl acetate, -terpineol and cis- and
of operating temperatures; the inertness and efRcien- trans-nerolidols are successfully and simultaneously
Figure 4 Simultaneous enantiomer GC separation of optically active components characterizing lavender oil: 1, cis-linalyl oxide;
2, trans-linalyl oxide; 3, linalol; 4, linalyl acetate; 5, borneol; 6, bornyl acetate; 7, -terpineol; 8, cis-nerolidol; 9, trans-nerolidol. Column:
25 m, 0.25 mm 30% 2,3-diethyl-6-t-butyl-dimethylsilyl--CD/PS-086, film thickness 0.15 m. Analysis conditions: from 703C (1 min) to
1903C (10 min) at 23C min\1.
separated with a 30% 2,3-diethyl-6-t-butyl-dimethyl- Multidimensional GC

silyl--CD in PS-086.
Koenig and Joulain have made a very important Multidimensional GC (MDGC) is a very useful tech-
contribution to this Reld: they have identiRed and nique to analyse a complex mixture such as an essen-
structurally characterized about 330 sesquiterpene hy- tial oil. In MDGC, groups of components not separ-
drocarbons, including the enantiomer recognition of ated on the Rrst column can automatically be trans-
the optically active ones. Figure 5 shows the enantio- ferred on-line to a second column coated with a dif-
selective GC pattern of a group of sesquiterpene hy- ferent stationary phase. The possibilities of MDGC
drocarbons (-elemene, -copaene, ar-curcumene, - are still not fully appreciated: it is true that early
bisabolene and (E)--bisabolene) separated on a 20% systems were difRcult to tune, inSexible and above all
2,6-di-O-methyl-3-O-pentyl--CD/OV-1701 column. very expensive; however, most of the present systems
Figure 5 Enantioselective GC pattern of a -elemene, -copaene, ar-curcumene, -bisabolene and (E )--bisabolene separated on
a 20% 2,6-di-O -methyl-3-O -pentyl--CD/OV-1701 column. Analysis conditions: from 603C (1 min) to 1903C (10 min) at 0.63C min\1.
(Courtesy of Professor W. Koenig, University of Hamburg.)
are fully automatic and not too expensive. Above all, three columns. The last two percentages are close to
they consist of two independent GC units connected that obtainable with MS, which is generally around
through a transfer interface, which can be used inde- 90%. Since a GC system affording simultaneous in-
pendently when MDGC is not necessary. jection into two columns is simple to assemble, and
MDGC is particularly useful with enantiomer GC today’s processing systems can easily handle two de-
analysis, which may double the number of peaks of tector signals, a manual or automatic crossed-identi-
the optically active components, making the Rcation procedure is not difRcult to set up. This is
chromatogram resulting from the analysis of an es- particularly true with the latest-generation instru-
sential oil even more complex, and increasing the ments: the development of GC instruments with elec-
probability of peak overlap, thus interfering with tronic pressure control of the mobile phase and of GC
a correct determination of enantiomeric ratios. In ovens in which the temperature is strictly controlled
these cases MDGC operates a sort of clean-up on the and evenly distributed, has overcome several prob-
Rrst column, so that only selected peaks are transfer- lems with instrumentation. These give rigorous con-
red to the chiral column. trol of mobile-phase parameters (Sow rate, pressure
and average linear velocity) and of temperature para-
Identi\cation meters over the whole GC run. Thus the chromato-
graphic process, and hence retention, becomes highly
Essential oil components are generally identiRed reproducible: as an example, under Rxed conditions
through GC or GC-MS or, better, through their com- a retention index precision of 1 unit was maintained
bination. The safest way to identify an essential oil over 1 month for some of the most signiRcant essen-
component, and in particular a sesquiterpene, is to tial oil components of both Matricaria chamomilla
combine dual-column GC data and mass spectro- (OV-1 column) and Tagetes lucida (CW-20M col-
metry (MS) data with IR data, because of the high umn; Table 1).
structure-related speciRcity of infrared spectroscopy Retention indices are fundamental in making reten-
(IR) signals. It is important to remember that identi- tion a reliable identiRcation tool for GC, although
Rcation and structure elucidation are totally different many problems still exist; in particular, the variation
things: identiRcation can only be by comparison with of stationary-phase polarity and mobile-phase char-
reference data. It is risky to propose a new structure acteristics as a function of temperature in pro-
or, worse, a new skeleton, from results obtained only grammed analysis. As a consequence, a comparison
by combined techniques (GC, GC-MS, GC-FTIR) of data from different laboratories can only be made
without parallel isolation and spectroscopic invest- with analyses run under carefully controlled operat-
igation (in particular, nuclear magnetic resonance) of ing conditions.
the new compound.
As in all Relds of analytical chemistry, the introduc- Identi\cation by GC-MS
tion of data systems has revolutionized the approach Mass spectral data are often } and perhaps to some
to identiRcation. Operators can now build their own extent erroneously } considered the key for compon-
personal libraries with retention indices and mass ent identiRcation. Many people give too much prior-
spectra obtained with their own instruments, and ity to mass spectrometry (MS) data over chromato-
combine the GC and GC-MS data for crossed identi- graphic data, and seldom give due weight to the
Rcation of essential oil components. complementarity of GC and MS data. This is prob-
ably because many manufacturers and operators do
Identi\cation through Chromatographic Data
not yet consider GC-MS as a technique in its own
Since essential oils are generally very complex mix- right, but a simple coupling between GC and MS.
tures, reliable GC location and identiRcation of their Nowadays, identiRcation is generally made
components can only be through retention indices, through commercially available mass spectra libraries
calculated by the KovaH ts method or with the van den (NBS, NIST, Wiley, TNO); these are nonspecialized
Dool algorithm; these make retention values indepen- collections of spectra mainly taken from the litera-
dent of GC conditions. IdentiRcation through reten- ture. As a consequence, the identiRcation of a com-
tion indices is in general only considered signiRcant ponent must be carefully conRrmed, since the mass
when two successful matches are obtained from dif- spectra are from different origins and have been re-
ferent-polarity stationary phases. When a suitable corded under different operative conditions. A classi-
reference database is available, the percentage of cor- cal example is the differences in the spectra produced
rect identiRcations obtained through retention data is by different mass analysers: ion trap, quadrupole or
generally around 65% with one column, 80% with magnetic sector instruments. Most operators over-
two different-polarity columns, and above 90% with come this problem by building dedicated libraries
Table 1 Reproductivity over time of reference index of Matricaria chamomilla L. essential oil component (OV-1) and of Tagetes lucida
Cav. essential oil components (CW-20M)
Matricaria chamomilla L Tangetes lucida Cav
Compound RI a RI b Compound RI a RI b
1 Trans--farnesene 1442 1441 1 Myrcene 1159 1157

2 Bisabolol oxide B 1619 1618 2 Trans--ocimene 1247 1246
3 -Bisabolone oxide A 1637 1637 3 Linalol 1553 1553
4 -Bisabolol 1649 1649 4 Estragole 1656 1656
5 Chamazulene 1674 1674 5 Anethole 1807 1805
6 Bisabolol oxide A 1702 1701 6 Methyl eugenol 2006 2005
7 Spiroether 1805 1805 7 -Caryophyllene 1566 1566
8 Germacrene D 1675 1673
RIa Reference initial index; RIb Reference index calculated 1 month later under the same conditions. GC analysis: columns: 25 m,
0.25 mm i.d. OV-1 column, df: 0.3 m; 25 m, 0.25 mm i.d. CW-20 m column, df: 0.25 m. Analysis conditions: injection: split, split ratio
1 : 20, temperature 2303C; detector: FID, temperature 2503C; temperature programme: from 503C (1 min) to 2203C (10 min) at
33C min\1; carrier gas: hydrogen, constant flow: 1.5 mL min\1.
consisting of spectra recorded with their own GC-MS different polarity columns) are then used to identify
systems. Libraries dedicated to the essential oil Reld are each component. Figure 6 shows retention indices
also available, as is the case of Adams library for ion and mass spectra of cis and trans--irones.
trap mass spectra, or the Joulain and Koenig collec- The use of retention indices, as a further active
tion of sesquiterpene hydrocarbon spectral data. identiRcation key in combination with mass spectra
Chromatographic data used either actively or pass- or within the classical library search procedure, can
ively in a library search can play a fundamental role be extremely useful. The ideal procedure should in-
in the successful identiRcation of essential oil compo- clude simultaneous and/or sequential searches with
nents. Several compounds, in particular sesquiter- retention indices from two different stationary
penoids, have low resolution mass spectra that are phases, and mass spectra in which Sexible and select-
almost indistinguishable. In this case, mass spectra able priorities can be actuated.
can mainly be used to locate the spectra in the total Unfortunately, identiRcation by retention indices
chromatogram; retention indices (better if from two associated with mass spectra is not absolutely risk-
Figure 6 Retention indices on OV-1 and CW-20M and mass spectra of cis and trans--irones.
Figure 7 Retention indices on OV-101 and DB-Wax and mass spectra of (A) eugenol (RI(OV-101): 1323; RI(DB-Wax: 2158) and
(B) 2-methoxy-3-hydroxy-allylbenzene (RI(OV-101): 1325; RI(DB-Wax): 2160). (Courtesy of Professor K.-H. Kubeczka, University of
Hamburg.)
free: there are some rare exceptions in which pairs of Fourier transform infrared spectroscopy (FTIR) com-
compounds have almost identical mass spectra, and bined with GC has emerged as a powerful technique
also have retention indices which fall within the in- and as the ideal complement to MS for component
strumental and analytical limit on two different sta- identiRcation in complex mixtures, thanks to its
tionary phases. This is the case of mass spectra and ability to distinguish geometric and positional
retention indices of 2-methoxy-3-hydroxy-allylben- isomers and to characterize organic functions; more-
zene and eugenol on OV-101 and DB-Wax (Figure 7). over, the identiRcation of a compound through its
A last approach, no less important, is GC/single ion FTIR spectrum is very reliable. To exploit in full the
monitoring (SIM)-MS, which is very selective and as complementarity between FTIR and MS data, sys-
a consequence the most reliable procedure for tems combining online GC, FTIR and MS or FID
quantitative analysis. In particular, GC-SIM-MS have also been assembled. In spite of this, FTIR, as
can be an alternative to MDGC for direct deter- a detector for GC, is not as popular as MS, because
mination of enantiomeric ratios of optically active of the lack of sensitivity for many compounds
components in the essential oil. A suitable choice when compared to GC or GC-MS systems. Another
of speciRc diagnostic ions can overcome the inter- reason is the lack of extensive libraries. The cheapest
ferences to a correct determination of enantiomeric and most widely adopted GC-FTIR system is based
ratio due to peaks coeluting with the two (or more) on the so-called light}pipe interface, which pro-
enantiomers. duces vapour-phase spectra: this makes existing
collections of spectra useless for automatic iden-
Identi\cation by GC-FTIR
tiRcation, because they are mainly recorded in
Although MS is the preeminent technique to identify the liquid or solid phase. Some relatively small
a component in a complex mixture analysed by GC, it collections of vapour-phase spectra of compounds in
has some drawbacks, e.g. in the differentiation of the essential oils and Savour Relds are now commer-
structural isomers giving identical mass spectra. cially available.
GC-Isotope Ratio Mass Spectrometry relevant to the overall smell while others, present in
trace amounts but with low detection thresholds, are
The stable isotope ratio is an important parameter in
not revealed by FID, nitrogen}phosphorous detector
biochemistry, nutrition and drug research, and in
(NPD) or Same photometric detector (FPD) but are
origin assignment and authenticity control of essen-
detected by GC-snifRng. Sensory methods are there-
tial oils. This ratio has gained in importance with the
fore needed to detect trace components responsible
introduction of on-line coupled GC-isotope ratio MS
for the smell of an essential oil. A snifRng device is
systems, where the analytes eluting from the GC
very simple and inexpensive to assemble. Several have
column are combusted to carbon dioxide in an oven
been described: in a typical one, the Sow of the gas
and analysed in an isotope ratio mass spectrometer,
eluting from the analytical column is split through
adjusted for the simultaneous determination of
a T piece on one side to an FID and on the other to
mass 44 (12C16O2), 45 (13C16O2, 12C16O17O) and 46
the snifRng port, which consists of a shaped glass
(12C16O18O) in the nmol range and with high pre-
funnel. A stream of air (or nitrogen) saturated with
cision (40.3). The actual ratio is obtained from the
water is sent coaxially with the mobile phase to the
ratio of the areas of two isotope peaks; this value is
snifRng port to avoid dehydration of the nasal tissues.
then compared to a standard value by applying the
The results of olfactory measurements can be quali-
following expression:
tative, giving a description of the odour of each
peak corresponding to an odour-active component;
"(Rsa/Rst!1);1000 on the basis of these qualitative results, a semiquan-
titative evaluation is also possible. Several ap-
where Rsa is the isotope ratio of the sample and proaches have been developed for semiquantitative
Rst that of the standard; -C13 is given in parts per sensory evaluation: the best known are Charm analy-
thousand. sis developed by Accree and AEDA (aroma extract
The -C13 value is particularly effective when com- dilution analysis), developed by Grosch. Charm anal-
bined with enantiomeric recognition of the chiral ysis is based on snifRng a series of decreasing dilutions
component(s) characterizing an essential oil. Enan- of the components eluting in the GC odour-active
tiomer GC analysis may fail in authenticity deter- zones characterized with a speciRc sensorial descrip-
mination when recemates of natural origin are pres- tor. The beginning and end of each particular odour
ent, or when racemization occurs during processing perception are Rxed. Charm values are calculated
or storage of natural products, and chiral essential oil through the formula c"dn\1, where n is the number
components are blended with the corresponding syn- of coincident responses and d is the dilution factor.
thetic chiral compounds. AEDA is similar but it uses a dilution factor equal to
Enantiomeric GC separation combined with iso- the last dilution in which an odour-active component
tope ratio MS is a very effective method of evaluating is detected.
the authenticity of an essential oil since the mass
spectrometer detects enantiomers of the same natural GC Pro\le Analysis
source which have identical -C13 values. As a conse-
quence, identical -C13 ratios are expected for enan- IdentiRcation and quantitation of the characterizing
tiomers from genuine compounds, even if the chiral components are not always sufRcient to discriminate
molecules to be analysed are partially racemized: it between essential oils from a single species, or to
seems improbable that racemic compounds would be classify them, evaluate their quality or origin, or de-
synthesized through different biochemical pathways tect adulterations. On the other hand, although sens-
in the same organism. Enantiomer GC-isotope ratio ory analysis (snifRng) is of prime importance in over-
MS or, even better, multidimensional GC-isotope ra- all evaluation, it is also sometimes insufRcient and,
tio MS of enantiomers can therefore detect blends of above all, it is generally considered insufRciently ob-
optically pure chiral essential oil components with jective. This is particularly true when series of sam-
synthetic racemates. ples of different origins have to be evaluated at the
same time. In these cases, the overall GC proRles of
the essential oils under investigation (or the proRle of
Sensory Analysis the volatile fraction obtained through related samp-
ling techniques) can be a successful marker to charac-
(GC-Snif\ng Detection) terize and/or discriminate between them objectively.
The GC proRle of an essential oil does not necessarily However, essential oil proRles are so complex that
reSect the sensory properties of its components. Some multivariate statistics is needed to obtain an exhaust-
of the components, present in large quantities, are not ive evaluation. Several statistical approaches have
Figure 8 (A) Scatterplot of principal components and (B) distrbution of the loadings considered for PCA of a set of peppermint
essential oils of different origins (Piedmont (Italy), France and India.)
been proposed (cluster analysis, fuzzy clustering, in particular for routine purposes. The success of
linear discrimination analysis, neural network, prin- PCA is strictly related to a correct selection of vari-
cipal component analysis (PCA), and principal ables (the peak areas corresponding to speciRc essen-
component similarity analysis). PCA is successful in tial oil components, generally chosen among those
analysing multivariate data, since it can investi- whose peak areas are detectable and reproducibly
gate relationships between large numbers of vari- measurable in all samples under investigation).
ables, and it is useful for reducing the numbers of Figure 8 shows the distribution of the loadings con-
variables in a data set by Rnding linear combinations sidered for PCA and the scatterplot of principal
of those variables that explain most of the variability. components of a set of peppermint essential oils of
These characteristics make PCA successful in different origins (Piedmont (Italy), France and
comparing and discriminating groups of essential India): PCA clearly distinguishes the origins of the
oils, for instance versus a series of reference samples, samples.
III / ESSENTIAL OILS / Thin-Layer (Planar) Chromatography 2755
Conclusions and Sowers. Flavour and Fragrance Journal 5:

131}145.
Although essential oil analysis is now a well-estab- Bicchi C, Manzin V, D’Amato A and Rubiolo P (1995)
lished Reld, further work is needed, not only to Cyclodextrin derivatives in GC separation of enan-
improve sample preparation and analysis techniques, tiomers of essential oil, aroma and Savour com-
but also to deal with one of the main aims of this pounds. Flavour and Fragrance Journal 10:
research Reld: to isolate and elucidate the structure 127}137.
of new odorous compounds. These studies evolve Bicchi C, D’Amato A and Rubiolo P (1999) Cyclodextrin
along two main lines. The Rrst and classical one derivatives as chiral selector for direct gas chro-
combines isolation and spectroscopic techniques matographic separations of enantiomers. Journal of
and mainly concerns new mono- and sesquiter- Chromatography A 843: 99}121.
Blank I (1996) Gas chromatography}olfactometry in food
penoids. The second mainly involves the so-called
aroma analysis. In: Marsili R (ed.) Techniques for
supervolatile fraction and perfumed trace com- Analysing Food Aroma. New York: Dekker.
pounds, two fractions that play a fundamental role Grosch W (1993) Detection of potent odorants in foods by
in odour impact. For the supervolatile fraction, aroma extract dilution analysis. Trends in Food Science
some topics requiring further study are HS com- Technology 4: 68}73.
bined with effective cryotrapping techniques, systems Ishibara M, Tsumeya T, Shiga M et al. (1992) New py-
for direct GC injection of large volumes of gas sam- ridine derivatives and basic components in spearmint oil
ples and GC columns with a high retention capacity. (Mentha gentilis f. cardiaca) and peppermint oil
For compounds present in the essential oil at the (Mentha piperita). Journal of Agricultural and Food
p.p.m. level (e.g. pyridine derivatives in peppermint Chemistry 40: 1647}1655.
and orange oils), a number of points would beneRt Joulain D and Koenig W (1998) The Atlas of Spectral
Data of Sesquiterpene Hydrocarbons. Hamburg:
from further investigation. These include increased
E.B. Verlag.
selectivity of sample preparation techniques and Kolb B and Ettre L (1997) Static Headspace}Gas
increased sensitivity and selectivity of analysis Chromatography. New York: Wiley-VCH.
techniques. Mosandl A (1995) Enantioselective capillary gas chromat-
rography and stable isotope ratio mass spectrometry in
See also: II/Chromatography: Gas: Column Techno- the authenticity of Savours and essential oils. Food Re-
logy; Detectors: Mass Spectrometry; Detectors: Selective; view International. 7: 597}664.
Headspace Gas Chromatography; Sampling Systems; Pawliszyn J (1997) Solid Phase Microextraction: Theory
III/Essentials Oils: Distillation. Terpenoids: Liquid and Practice. New York: Wiley-VCH.
Chromatography. Sandra P and Bicchi C (eds) (1987) Capillary Gas Chrom-
atography in Essential Oil Analysis. Heidelberg:
Huethig.
Further Reading Schurig V and Novotny H-P (1990) Gas chromatographic
separation of enantiomers on cyclodextrin derivatives.
Accree T (1997) GC/Olfactometry with a sense of smell. Angewandte Chemie, International Edition English 29:
Analytical Chemistry 69: 170}175A. 939}957.
Accree T and Teranishi R (1993) Sensible principle and Tarja n G Nyiredy Sc, Gyo r M et al. (1989) Thirtieth
techniques. In Flavor Science. Washington, DC: Ameri- anniversary of the retention index according to Kova tz
can Chemical Society. in gas-liquid chromatrography. Journal of Chromatog-
Adams RP (1989) IdentiTcation of Essential Oils by Ion raphy 472: 1}92.
Trap Mass Spectroscopy. New York: Academic Press. Thomas AF and Bassols F (1992) Occurence of pyridines
Bicchi C and Joulain D (1990) Headspace}gas chroma- and other bases in orange oil. Journal of Agricultural
tographic analysis of medicinal and aromatic plants and Food Chemistry 40: 2236}2243.

P. Dugo, L. Mondello and G. Dugo, Introduction
Universita% di Messina, Messina, Italy
Essential oils are mixtures of mainly volatile com-
Copyright ^ 2000 Academic Press ponents belonging to different chemical classes. They
2756 III / ESSENTIAL OILS / Thin-Layer (Planar) Chromatography
are characterized by a pleasant smell and are gener- ponents of essential oils do not absorb in the UV-
ally obtained by steam distillation of aromatic plants, visible region, and UV detectors are the most popular
with the exception of citrus peel essential oils, which in HPLC. Thin-layer chromatography (TLC) is a very
are produced by cold-pressing the peel of the fruits. widely used chromatographic technique, and modern
This process involves the abrasion of peel and the HPTLC can be advantageously used instead of HPLC
removal of the oil in an aqueous emulsion that is or GC in many analytical situations.
subsequently separated in a centrifuge. Other Some of the advantages of TLC are its simplicity,
methods may be water distillation or extraction with economy in materials, simultaneous analysis of
sub- or supercritical Suids. Essential oils are not sol- a large number of samples and the use of com-
uble in water, but are quite soluble in alcohol. plementary post-chromatographic universal and
Resins can be either natural or prepared: natural selective detection methods. Some disadvantages of
resins are exudations from trees or plants, and are TLC, such as the long times required for development
formed in nature by the oxidation of terpenes; pre- or the difRculties in controlling the speed of solvent
pared resins are oleoresins from which the essential migration, may be overcome by OPLC (overpres-
oils have been removed. Resins are mixtures of many sured layer chromatography). OPLC, developed by
components; they are solid, amorphous, more or less Tyihak and co-workers at the end of the 1970s, is
coloured, nonvolatile, with a characteristic smell, in- a planar liquid chromatographic technique with ad-
soluble in water, but soluble in alcohol or other vantages over classical TLC and HPLC: the station-
organic solvents. ary phase is covered completely by a Sexible mem-
Balsams are natural raw materials exuded from brane under external pressure and, because the eluent
a tree or a plant; they have a high content of benzoic is introduced to the layer by means of a pump, the
acid, benzoates or cinnamates. solvent Sow rate may be controlled. Improved separ-
The main constituents of essential oils, balsams and ation efRciency, shorter time requirement, better res-
resins are terpene or aromatic hydrocarbons, and olution, and lower solvent consumption than classical
their oxygenated derivatives (alcohols, aldehydes, TLC or HPLC can thus be obtained. Moreover,
esters, ketones, oxides, etc.). OPLC can be used for analytical or preparative pur-
The physiological role of oils and resins in plants poses and maintains all the advantages already men-
and trees is not well understood. It is likely that they tioned for classical TLC.
play a role as lures for insects. They may also serve to Another advance in the development of TLC is
protect plants from parasites, increase the rate of chromatography on permanent rod-shaped layers
transpiration and act as a seal for wounds. They are (Chromarods), whose mechanical and chemical
largely used in perfumery, food or pharmaceutical properties permit detection of the separated com-
industries, as Savouring agents or because of their ponents in an ionization detector; this technique
different pharmacological actions. was developed in the late 1960s and can be success-
fully used for quantitative separations of substances
Characterization of Essential Oils, that cannot be analysed by GC because of their low
volatility.
Balsams and Resins There are many papers on planar chromatography
Essential oils may be characterized by the determina- that well illustrate the most recent developments of
tion of physicochemical properties such as boiling this technique: the use of multidimensional develop-
point, freezing point, solubility, density, optical rota- ment, the coupling with particular detectors such as
tion, refractive index, etc. These parameters can also MS or FT-IR detectors, or the use of fully com-
help in the detection of adulteration. For the study of puterized image processing instruments. All these
the qualitative and quantitative composition of essen- developments are accompanied by improvements in
tial oils and resins, chromatographic methods are the the performance and in the reproducibility of pre-
techniques of choice. Since the main part of the oil coated layers for planar chromatography.
consists of volatile components, gas chromatography The literature reports some applications of planar
(GC) equipped with FID (Same ionization) or MS chromatography to the analysis of essential oils, bal-
(mass spectrometer) detectors is the most used sams and resins, to obtain different information.
technique. Methods have been developed that use both conven-
High performance liquid chromatography (HPLC) tional TLC, HPTLC and OPLC for analytical and/or
is also widely used for separation of semi-volatile or preparative separations. Only a few papers have been
nonvolatile components, for preparative purposes, or published on the separation of terpenes and related
for the analysis of thermally labile components. The substances on Chromarods. One of the reasons
limitation of HPLC is detection, because many com- for this is probably the volatility of terpenes of low
relative molecules mass, leading to low, irre-

producible FID response. Resins have been separated
from arnica (benzene/chloroform, 67 : 33) and
Tolu balsam (benzene/chloroform/formic acid,
69 : 29 : 2), but the results obtained were not satisfac-
tory. The number of applications is limited, if com-
pared with the applications of GC or HPLC to the
analysis of essential oils and plant extracts. However,
often TLC is essential as a simple analytical and
preparative technique. A recent review summarized
the chromatographic methods reported by the Euro-
pean Pharmacopoeia (2nd edn) for the analysis of
products from medicinal plants, including essential
oils, balsams and resins. Over half of the methods
reported for medicinal plant products are chromato-
graphic methods, of which TLC represents 82%.
Preparative TLC
Essential oils and resins are complex mixtures con- Figure 1 TLC chromatograms of tetraploid camomile (CFH)
taining numerous components that belong to differ- essential oil obtained by steam distillation (A), compounds iso-
ent chemical classes present in different concentra- lated from CFH oil (B) and separation of trans-farnesene and
tion. Often it is necessary to fractionate the mixture chamazulene (C). Mobile phase: (A) and (B) dichloro-
methane/toluene/ethyl acetate (70 : 28 : 2, v/v/v); (C) cyclo-
into single chemical classes or to isolate single compo-
hexane. Detection by spraying the plates with 1% vanillin solu-
nents, particularly useful for the determination of tion. (Reproduced with permission from Zekovic Z, Pekii B,
properties such as authenticity, geographical origin or Lepojevic Z and Petrovic L (1994) Chromatographia, 39:
pharmacological activity. 587I590.)
This fractionation can be carried out by prepara-
tive TLC. For example, for detecting illegal adulter- Preparative separations have also been carried out
ation of pharmacognostic mint (Mentha arvensis) or by OPLC to isolate antifungal compounds of the
peppermint (M. piperita) oils with racemic menthyl essential oil obtained by water distillation of the
acetate, samples were separated by TLC (polygram fresh bark of Ocotea usambarensis from Rwanda.
SIL G/UV plates, CH2Cl2 as mobile phase) and the The strategy followed for characterization of essen-
zone of menthyl acetate extracted from the plate and tial oil constituents is illustrated in Figure 2. This
analysed by chiral capillary GC. Since natural mint scheme shows how a combination of chromato-
oils contain 100% of the (!)-isomer of menthyl acet- graphic techniques, including TLC, can be useful to
ate, the presence of the (#)-isomer can be used to characterize bioactive constituents of medicinal
quantify this kind of adulteration. In this case, plants.
preseparation is necessary to obtain a reliable stereo-
differentiation without problems of peak overlap in
a direct GC analysis.
Quantitative Analysis by TLC
TLC has been used as preparative tool to obtain Some methods have been developed for the quanti-
pure components to be used for further characteriza- tative analysis of the main volatile components of
tion. For example, components of the sesquiterpene essential oils by TLC as a rapid and easy alternative to
fraction of camomile oil were separated by semip- other chromatographic determinations, in particular
reparative TLC. The components so obtained } trans- GC and HPLC which are more expensive and time-
farnesene, chamazulene, cis-en-in-dicycloether, trans- consuming.
en-in-dicycloether, (!)--bisabolol, (!)--bisabolol It is difRcult to Rnd applications of TLC
oxide A, and (!)--bisabolol oxide B } were analysed densitometry to problems in the Reld of essential oils,
by GC-MS, and the MS spectra were used to build e.g. for the quantiRcation of individual components,
a database since mass spectra of these components probably because of the volatility of the various com-
are not included in some commercial MS libraries. ponents. One example is the quantitative determina-
Figure 1 shows the TLC separation obtained using tion of linalool, linalyl acetate and terpinen-4-ol in
silica gel TLC plates. Table 1 reports the RF values lavender oil. The Rrst two compounds are the main
and identiRcation of components. components of the oil.
Table 1 hRF values of camomile oil components separated in Figure 1
N Compounds hRF (%) Colour
A, B C
1 Trans-farnesene 93.22 64.61 Dark green

2 Chamazulene 46.15 Dark red
3 Cis-en-in-dicycloether 63.13 } Light brown
4 Trans-en-in-dicycloether }
5 (!)--Bisabolol 41.53 } Violet
6 (!)--Bisabolol oxide A 20.76 } Yellow
7 (!)--Bisabolol oxide B 16.94 } Yellow
A,B, mobile phase dichloromethane/toluol/ethyl acetate (70 : 28 : 2, v/v/v). C, mobile

phase cyclohexane. (Reproduced with permission from (1994) Chromatographia, 39:
587I590.)
Figure 3 shows the TLC and GC analysis of a lav- qualitative and quantitative analysis of the main com-
ender oil. The quantitative results obtained with the ponents of essential oils. It can be useful in the identi-
two techniques are comparable. This result shows Rcation of an oil, and can simultaneously also detect
that TLC densitometry is a good technique for both less volatile components.
Other examples are the quantitative determination
of citral in citrus oils and 1,8-cineole in eucalyptus
oils. The determination of citral (a mixture of two
terpene aldehydes, neral and geranial) is of particular
importance for citrus oils, mainly for lemon oils. TLC
determination of citral has been carried out on silica
gel plates, developed with hexane/chloroform
(70 : 30) and measured at 250 nm with a TLC scan-
ner. The results were compared with those obtained
by GC, and the ratio between TLC and GC values
was constant at 0.8.
Eucalyptus essential oil contains a high amount of
1,8-cineole, an oxygenated monoterpene. In recent
years, interest in developing new uses for cineole-rich
eucalyptus oils has been renewed. The determination
of 1,8-cineole by TLC has been carried out using
silica gel plates and a mixture of light petro-
leum/chloroform (70 : 30) as mobile phase. Visualiz-
ation is with 4-dimethylaminobenzaldehyde-S re-
agent (4-DMAB) and quantitation is with a scanner.
Quantitative results compared with GC-MS data
were found to be essentially identical.
TLC for Detection of Authenticity and

Botanical Origin
TLC can be used for screening samples to establish
their authenticity and botanical origin. An example is
the determination of the quality of cinnamon by
TLC densitometry. Cinnamon is one of the oldest
Figure 2 Strategy for the characterization of essential oil con-
stituents from the bark of Ocotea usambarensis (Laureaceae).
known spices; the cinnamons of commerce are de-
(Reproduced with permission from Hostettmann K., Terreaux, C, rived from the dried inner bark of several species of
Marston, A and Petterat O, (1997) Journal of Planar Chromatog- Cinnamomum. Cinnamon oil derived from the inner
raphy, 10: 251I257. bark of Cinnamomum zeylanicum Nees is generally
Figure 3 TLC (middle), densitometry (upper) and GC (lower) of lavender oil. TLC conditions: 20;20 cm plates, coated with 0.25 mm
silica gel 60. Mobile phase: dichloromethane/methanol, 10 : 1.
Detection: anysaldehyde/sulfuric acid ("680 nm); copper-acetate/phosphoric acid (white light); copper-sulfate/phosphoric acid
followed by sulfuric acid spray (white light).
Sample volume: 15 mL of a 5% dichloromethane solution of oil or pure standard components. GC conditions: 25 m;0.32 mm i.d.
HP-5 fused silica capillary column (0.17 m film thickness); carrier gas: H2; detector (FID) and injector (split) at 2503C. Temperature
program: 60}2403C, 63C min\1. (Reproduced from (1989) Mikrochim. Acta (Wien), 3: 1I6, with permission from Springer-Verlag,
Wien.)
considered to have a better Savour, and commands mon and cassia. As can be seen, cinnamon and cassia
the highest price. oils can be easily distinguished because of the pres-
This oil can be adulterated with cinnamon leaf ence of higher amount of coumarin and the absence
or root bark oils, that have a different composi- of eugenol in cassia oil.
tion from inner bark oil and are less valuable. Another example is the method developed to dis-
Moreover, some countries consider oil obtained cover the poisonous Japanese star anise or shikimi
from C. cassia Blume as cassia oil, and it is un- fruits (Illicium anisatum), when they are mixed with
lawful to present cassia as cinnamon, or to pre- those of the Chinese anise or star anise (I. verum).
pare mixtures of the two and present the result as Miristicin only occurs in the essential oil of shikimi,
cinnamon. albeit in small amounts, but it is absent in star anise.
Figure 4 shows the TLC densitometry separation TLC allows the detection of an admixture of only 5%
of polar aromatic semivolatile compounds of cinna- shikimi, as shown in Figure 5.
grown wild in Greece is studied. The composition has

been studied by GC-MS (ITD) and carvacrol found
to be the main components. A variety of oregano,
O. vulgare subsp. hirtum, has a high thymol content,
while the other plant species contain less than 1%
of thymol. The antioxidant activity of essential
oils, pure carvacrol, thymol and butylated hy-
droxytoluene (BHT) was tested on silica gel TLC
plates. The plates were sprayed with a solution of
-carotene and linoleic acid and then exposed to
daylight until the background colour disappeared.
Spots of components having antioxidant activity
remained yellow for a longer period.
Figure 4 Separation of cinnamon and cassia oils by silica gel TLC for the Analysis of Phenolic
HPTLC plates. (A) Authentic sample of essential oil from the inner
bark of Cinnamomum zeylanicum Nees; (B) cinnamon oil adulter- Compounds
ated with leaf oil; (C) authentic sample of cassia oil from C. cassia
Blume. Peaks: 1, cinnamyl alcohol; 2, coumarin; 3, eugenol; 4,
TLC has been used to analyse less volatile compo-
2-methoxy-cinnamaldehyde; 5, trans-cinnamaldehyde; 6, cinna- nents such as Savonoids, phenolic acids or
myl acetate. TLC conditions: TLC plates predeveloped in coumarins. These classes of components are very im-
hexane/triethylamine (46#4, v/v) and separation in unsaturated portant for the characterization of plant materials,
chamber with hexane/chloroform/triethylamine (90#6#4, and can have speciRc pharmacological activities. For
v/v/v). The separation was scanned at "255 nm. (Reproduced
with permission from Poole SK, Kiridena W, Miller KG and Poole
example, the spasmolitic activity of camomile is
CF (1995) Journal of Planar Chromatography, 8: 257I268. mainly due to the presence of Savonoids apigenin,
apigenin-7-O--glucoside and its acetylated deriva-
tives. Figure 6 show the HPTLC chromatogram of
camomile Savonoids. HPTLC is the fastest chromato-
Determination of the Antioxidant graphic method for qualitative identiRcation of
Activity of Essential Oils apigenin and its glucosides in camomile. Quantitative
The versatility of TLC is also illustrated in the follow-
ing example, where the composition and the anti-
oxidant activity of essential oils from oregano plants
Figure 6 TLC chromatogram of camomile flavonoids. Key: C,

dry extract of camomile ligulata flowers (CLF); AC, dry extract of
autofermented CLF; A, apigenin; L, luteolin; ADG#AMG, mix-
ture of apigenin-7-O--diacetylglucoside and apigenin-7-O--
monoacetylglucoside; AG, apigenin-7-O--glucoside. Experi-
Figure 5 TLC chromatogram of (1) mixture of shikimi fruit and mental details: silica gel 60 F254 HPTLC plates precoated with
star anise fruit oils; (2) mixture of standard compounds; and (3) concentrating zone; eluent:benzol/ethylmethylketone/methanol
reference mixture. Key: A, anethole; F, foeniculin; M, myristicin; (5 : 5.3 : 1.5); detection, "254 nm. Yellow spots of separated
S, safrol; r, red; b, blue. (Reproduced from (1990) Deutsche flavonoids appear after spraying the plate with 96% sulfuric acid
Apotheker Zeitung, 130: 1194I1201, with permission from Deut- and heating briefly to 1203C. (Reproduced with permission from
scher Apotheker, Verlag.) (1994) Chromatography, 39: 587I590.
compounds in citrus oils. An OPLC method has been

developed to separate these components in seven cit-
rus oils: sweet orange, bitter orange, mandarin,
grapefruit, lemon, bergamot and Mexican lime.
The OPLC separation is fast (10 min) and allows the
differentiation of the various oils and the determina-
tion of possible contamination or sophistication of
the oils.
In particular, many methods for the determination
of bergapten (5-methoxypsoralen) have been re-
ported, because of the problems linked with its
phototoxicity. The European Pharmacopoeia (3rd
edn) reports a TLC method to detect the presence of
bergapten in bitter orange Sower oil. Bergapten
shows a greenish-yellow Suorescence at 365 nm. The
Figure 7 Densitogram of a camomile sample. Peaks: 1, presence of this psoralen may be an indication of the
apigenin-7-O-glucoside; 2, caffeic acid; 3, apigenin; 4, umbel- presence of bitter orange peel oil.
liferone; 5, ferulic acid; 6, herniarin. (Reproduced from Menziani
E, Tosi, B, Bonora A, Reschiglian, P and Lodi G (1990) Journal of
Chromatography, 511: 396I401, with permission from Elsevier Conclusion
Science.)
The examples described above show that both ana-
lytical and preparative, planar chromatography can
determination is possible using appropriate instru- be used successfully in a very wide range of applica-
ments. tions to essential oils, balsams and resins. Thin-layer
With the development of the automated multiple chromatography has the ability to separate mixtures
development (AMD) technique, planar chromatogra- of substances of similar structure and the Sexibility of
phy may be applied successfully to the analysis of a variety of methods of detection. It also has the
complex matrices. AMD-HPTLC gives the opportun- advantages of being cheap and easy to perform. As
ity to carry out separation processes using a gradient recently stated by J. Sherma in a review that appeared
development mode. An optimized AMD-HPTLC in Analytical Chemistry in 1996, ‘fully instrumental
procedure has been applied to the separation of phen- planar chromatography, performed properly is the
olic compounds (Savonoids, coumarin, phenylcar- most economic, powerful, and accurate quantitative
boxylic acids) of Chamomilla recutita Sower ex- analytical method for mixtures of soluble substances
tracts. Figure 7 shows the densitogram obtained un- of low vapour pressure’.
der optimized conditions: HPTLC plates and stepwise
gradient development in an enclosed chamber. Fif-
teen steps were used, with drying times 6 min for the Further Reading
Rrst four steps, then 4 min for the next 13 steps, and European Pharmacopoeia (3rd edn) (1997}1999). Stras-
10 min for the last step. As preconditioning, nitrogen bourg: Council of Europe.
was bubbled through water; the preconditioning time Poole CF and Poole SK (1995) Multidimensionality in
was 15 s for each step. planar chromatography. Journal of Chromatography
Cold-pressed citrus essential oils contain about A 703: 573}612.
90}99% of volatile components, with a nonvolatile Ranny M (1987) Thin Layer Chromatography with Flame
residue that ranges from approximately 1 to 10% in Ionisation Detection. Dordrecht Holland D. Reidel Pub-
the different oils and consists, in large part, of many lishing Company.
oxygen containing heterocyclic compounds, mainly Sherma J (1996) Planar chromatography. Analytical Chem-
istry 68: 1Rd19R.
the coumarins, psoralens and polymethoxylated
Sherma J and Fried B (1996) Handbook of Thin-Layer
Savones. The qualitative and quantitative composi- Chromatography: New York: Marcel Dekker.
tion of the nonvolatile residue characterizes the dif- Somsen GW, Morden W and Wilson ID (1995) Planar
ferent citrus oils, and play an important role in identi- Chromatography coupled with spectroscopic tech-
Rcation, quality control and authenticity. The litera- niques. Journal of Chromatography A 703: 613}665.
ture reports numerous TLC methods for the analysis Touchstone JC (1993) Column watch. LC-GC 11:
or preparative isolation of oxygen heterocyclic 404d411.
2762 III / EXPLOSIVES / Gas Chromatography
EXPLOSIVES
Extraction of Explosives from Water

Gas Chromatography 1. Liquid}liquid extraction is carried out with a sep-
aration funnel using 1 L of water (containing TNT
J. Yinon, Weizmann Institute of Science, and other nitroaromatic compounds) and shaking
Rehovot, Israel three times with 30 mL of methylene chloride. The
Copyright ^ 2000 Academic Press combined organic phases are dried over anhyd-
rous sodium sulfate and reduced in volume to
1 mL in a rotary evaporator at 403C, after the
Introduction exchange of the methylene chloride with meth-
anol. Other solvents can be used, such as benzene
Although some explosives are thermally labile and or toluene for nitroaromatic compounds, or
others are not volatile enough, gas chromatography isoamyl acetate for nitramines. Extraction can
(GC), with a variety of detectors, has been found to also be carried out using 100 mL of water. In that
be a good method for separation and analysis of case only about 6 mL solvent are required.
a certain number of organic explosives. This can be 2. Solid-phase extraction is carried out using Amber-
achieved when using the GC under controlled experi- lite XAD-2, XAD-4, XAD-8 resins (1 : 1 : 1), C18
mental conditions, such as the temperature of the phases, phenyl phases and cyano phases, 2.5 g of
column, injector and detector, type and length of the XAD resin is placed in a 15;1 cm i.d. glass col-
column, special injection techniques and the use of umn plugged with silanized glass wool and Sushed
selective detectors. with methanol and water. 1 L of water, containing
In this article we will describe some of the GC the explosives, is forced through the column at
methods used for the analysis of explosives } the a Sow rate of 30 mL min\1, using nitrogen pres-
preferred columns, the injection techniques and the sure. The column is then dried with a stream of
preferred detectors. In order to be able to evaluate nitrogen and eluted twice with 15 mL of methyl-
GC as a method for the analysis of explosives, it is ene chloride. Drying, concentration and solvent
necessary to present a short overview of the main exchange are carried out as in the liquid}liquid
organic explosives (Figure 1). These explosives can be extraction method. Recovery for TNT is 95%.
divided into the three following groups:
Extraction of Explosives from Soil
1. Nitroaromatic compounds
2. Nitrate esters 1. 10 g of undried, homogenized soil is extracted
3. Nitramines with 25 mL acetone in an ultrasonic bath for
30 min. The extract, after passing through Rlter
The most widely used nitroaromatic explosive is paper, is ready for analysis. Recovery for RDX is
2,4,6-trinitrotoluene (TNT). Nitrate esters include 95%.
ethylene glycol dinitrate (EGDN), glycerol trinitrate 2. Supercritical Suid extraction (SFE) can be used for
(nitroglycerin, NG) and pentaerythritol tetranitrate the extraction of explosives from soil. Extraction
(PETN). Nitramine explosives include 1,3,5-trinitro- is made with supercritical CO2 at 5000 psi and
1,3,5-triazacyclohexane (RDX), 1,3,5,7-tetranitro- 503C. The dynamic SFE mode is being used, where
1,3,5,7-tetrazacyclooctane (HMX) and 2,4,6-N-tetra- fresh supercritical Suid is Sushed continuously
nitro-N-methylaniline (tetryl). through the sample matrix and then passes
Additional nitroaromatic compounds encountered through a trap in which the analytes of interest are
are 2-, 3- and 4-nitrotoluene, 2,4- and 2,6-dinitro- collected. This mode of operation gives a better
toluene (DNT) and degradation products of TNT recovery than the static mode.
such as 2-amino-4,6-dinitrotoluene and 4-amino-2,6-
dinitrotoluene.
Capillary Columns
Low polarity columns are used, because the polar
Extraction Procedures interaction of the nitro groups can produce irrevers-
Various extraction procedures can be used. Several of ible adsorption on the stationary phase or decomposi-
them are described as follows. tion of the explosives at higher temperatures.
III / EXPLOSIVES / Gas Chromatography 2763
programmed from !53C to 2503C at a rate of

2003C min\1. A temperature-programmed injector is
much better suited for GC analysis of thermally labile
compounds, as it will minimize decomposition in the
injector.
Detectors
Electron-Capture Detector (ECD)
An electron-capture detector is an ionization chamber
in which electrons are produced from a radioactive
source (usually tritium or nickel-63). These electrons
are injected into a stream of inert carrier gas (helium
or nitrogen), where they reach thermal energy
equilibrium through collisions with the carrier gas.
The thermal electrons are collected at an anode,
thus producing a standing current. When an
electron-capturing compound (the sample), such as
a halogen- or a nitro-containing compound, is intro-
duced into the carrier gas, the standing current
is reduced. The reduction in current is proportional
to the concentration of the sample. Electron-capture
detectors have a fast response and are highly sensi-
Figure 1 Structure of commonly used explosives (see text for
abbreviations used).
tive for most electron-capturing compounds.
However, their speciRcity for explosives is low. De-
tection limits are in the 5}50 pg range for the various
Columns commonly used for the separation and explosives.
analysis of explosives include DB-1 (BP-1 or CP-Sil An example is presented in Figure 2, which
5CB) (100% dimethylpolysiloxane) and DB-5 (5% shows the GC-ECD chromatogram of a soil sample
phenylmethylpolysiloxane). Other columns, such as taken from a former explosives storage bunker. The
DB-17 (50% phenylmethylpolysiloxane) and OV- column used was a 30 m, DB-5 capillary column,
225 (50% cyanopropylphenylmethylpolysiloxane) held at 1803C isothermal. The dinitrotoluenes were
have also been used, but only for separation of by-products of TNT, while the aminodinitrotoluenes
nitroaromatic compounds. were microbial degradation products of TNT.
Another important factor is the column length.
Thermal Energy Analyser (TEA) Detector
Compounds which evaporate at higher temperatures,
such as RDX and PETN, should be eluted from the The TEA detector, also known as a chemilumines-
column as fast as possible in order to minimize their cence detector, is a nitrogen-speciRc detector. In the
decomposition. This can be done by either increasing TEA detector, nitro compounds are pyrolysed to
the Sow rate of the carrier gas or decreasing the form NO ) radicals, which pass into a reaction
length of the column. Columns as short as 1.5 m have chamber where they are oxidized by ozone to form
been used although short columns will, of course, electronically excited nitrogen dioxide (NOH 2 ). The
have a poorer separation capability. excited nitrogen dioxide decays back to its ground
Table 1 shows a summary of GC columns and state with emission of chemiluminescent light in the
conditions which have been used for the separation of near-infrared region ( 0.6}2.8 m). The inten-
a variety of explosives. sity of the emitted light is proportional to the NO
concentration and hence to the nitro compound
concentration. The TEA detector, although more
speciRc for explosives than the ECD, is less sensitive
Injectors by one to two orders of magnitude. An example is
Injectors used include split}splitless injectors at presented in Figure 3, which shows the GC-TEA
temperatures ranging from 1703C to 2503C, a Sash chromatograms of pure PETN and of a sample taken
vaporizing injector at 2703C and a temperature- from the debris of a bombing, containing traces of
programmable injector (cooled with liquid CO2), PETN.
Table 1 GC columns used for separation of explosives
Column type Column dimensions Temperature programme Carrier gas Detector /s used Explosives analysed
DB-1 30 m;0.32 mm i.d. 70}2503C at He MS DNT, TNT

0.25 m film 33C min\1
DB-1 5 m;0.20 mm i.d. 75}2253C at He ECD, MS Semtexa, PETN,
0.33 m film 203C min\1 RDX
DB-1 15;0.25 mm i.d. 80}2503C at He MS TNT. Tetryl, RDX,
0.25 m film 253C min\1 HMX, PETN
BP-1 12.5 m;0.22 mm i.d. 60}2403C at He TEA NG, TNT, RDX
BP-1 12 m;0.25 mm i.d. 60}2503C at He ECD EGDN, NG, PETN,
0.25 m film 403C min\1 DNT, TNT, RDX
DB-5 10 m;0.32 mm i.d. 50}2503C at He TEA EGDN, NG, PETN,
0.25 m film 103C min\1 DNT, TNT
DB-5 15 m;0.25 mm i.d. 50}2603C at He MS DNT isomers
DB-5 1.5 m;0.25 mm i.d. 60}2203C at He MS NG, PETN
CP-Sil 5CB 20 m;0.25 mm i.d. 50}903C at H2 ECD DNT, TNT
90}2503C at
3.53C min\1
DB-17 30 m;0.32 mm i.d. 70}2503C at He ECD, TEA DNT, TNT
OV-225 30 m;0.32 mm i.d. 70}2503C at He ECD, TEA DNT, TNT
OV-101 25 m;0.25 mm i.d. 50}2403C at He ECD EGDN, NG, PETN,
0.1 m film 403C min\1 DNT, TNT, RDX,
HMX, Tetryl
DB-1301 4.5 m wide bore 128}2253C at He or ECD DNT, TNT, RDX,
1.0 m film 303C min\1 H2 HMX
a
Semtex is a plastic explosive containing RDX and PETN.
The chromatograms were recorded at two different 2503C were done with a split}splitless injector, in the
injection temperatures: 1703C and 2503C. At 2503C splitless mode.
PETN decomposes, therefore a decomposition peak
appears in the chromatogram. The column used was Gas Chromatography+Mass
a 10 m, DB-5 capillary. Injections at 1703C and
Spectrometry (GC-MS)
The good separation capability of capillary column
GC, together with the high sensitivity and identiRca-
tion capability of the mass spectrometer, have made
GC-MS a powerful method in analytical chemistry.
The use of GC-MS for the analysis of explosives is
limited by the thermal decomposition characteristics
of some of the explosives. Precautions to be taken
when analysing explosives by GC-MS are similar to
those taken when using GC with any other detector.
GC-MS for the analysis of explosives has been used
in three different ionization modes: electron ioniz-
ation (EI), chemical ionization (CI) and negative-ion
chemical ionization (NCI). The produced ions are
Figure 2 GC-ECD chromatogram of a sample taken from a for- mass separated by a mass analyser (magnetic sector,
mer explosives storage bunker. Peak identity: 4: 2,6-dinitro-
quadrupole or ion trap), detected by an electron
toluene; 5: 2,4-dinitrotoluene; 6: TNT; 7: 4-amino-2,6-dinitro-
toluene; 8: 2-amino-4,6-dinitrotoluene; 9: RDX. (Reproduced from multiplier, recorded by a data system and stored in
Haas R et al. (1990) Fresenius Journal of Analytical Chemistry the computer as a mass spectrum. There are different
338: 41}45, by permission of Springer-Verlag.) ways to display the results: (i) as a total ion
III / EXPLOSIVES / Gas Chromatography 2765
Figure 3 GC-TEA chromatograms. (A) Pure PETN. (B) A sample from a bombing scene, containing PETN. Chromatograms were
run at injection temperatures of 1703C and 2503C. Peak d is a decomposition product of PETN. (Reproduced from Kolla P (1994)
Journal of Chromatography A 674: 309}318, by permission of Elsevier Science Publishers.)
chromatogram (TIC) or reconstructed ion chromato- Electron Ionization (EI)

gram (RIC), which is equivalent to a GC chromato-
gram } a mass spectrum of each one of the GC peaks The basic form of ionization in mass spectrometry is
can be displayed; (ii) as mass chromatograms, which electron ionization (EI), where an electron beam, usu-
are GC chromatograms including only preselected ally at an energy of 70 eV, collides with the molecules
masses. Each one of the GC separated compounds of the sample to transform them into positively
can be represented by one or several masses which are charged ions. In addition, extensive fragmentation of
characteristic of the full mass spectrum. the ions occurs, resulting in a mass spectrum which
Figure 4 GC-MS EI mass chromatograms of a mixture of explosives (10 ppb each), extracted from water by liquid}liquid extraction.
(Reproduced from Yinon J (1996) Journal of Chromatography A 742: 205}209, by permission of Elsevier Science Publishers.)
does not always contain a molecular ion. The frag- containing abundant ions at m/z 30 (NO#), m/z
mentation patterns thus obtained can be correlated 46 (NO# #
2 ) and m/z 76 (CH2ONO2 ). However, they
with speciRc functional groups, enabling recognition have different GC retention times, but this means that
of many structural features in the analysed molecule. their identiRcation is based on chromatographic
An EI mass spectrum of a molecule can, therefore, be properties. In the CI mass spectra of NG and PETN,
considered a ‘Rngerprint’ of that molecule and can be MH# ions are observed, at m/z 228 and m/z 317,
used as an identiRcation tool. respectively.
An example is presented in Figure 4 which shows
the GC-MS EI mass chromatograms of a mixture of Negative-ion Chemical Ionization (NCI)
explosives (10 ppb each), extracted from water by
In negative-ion chemical ionization (NCI) a reagent
liquid}liquid extraction with methylene chloride.
gas at 0.1 to 1.0 Torr is introduced in the ion source
Each of the explosives could be identiRed by at least
of the mass spectrometer. This reagent gas acts prim-
one characteristic ion (in most compounds a fragment
arily as a moderator in producing high concentrations
ion). The column used was a 15 m;0.255 mm i.d.
of low energy electrons. Negative ions are formed in
DB-1 capillary column. Column temperature pro-
the analysed sample by electron capture. These ions
gramme was 803C to 2503C at 253C min\1. Injector
are either molecular ions or (M}H)\ions. Fragment
temperature was programmed from !53C to 2503C
ions are also formed by dissociation of part of the
at 2003C min\1. The mass spectrometer was an ion
molecular ions. Special reagents can be introduced
trap operated in the EI mode.
into the ion source causing ion}molecule reactions,
Chemical Ionization (CI) thus forming characteristic adduct ions. For example,
the NCI mass spectrum of TNT will contain abun-
In chemical ionization (CI) ions are formed by reac-
dant anions at m/z 227, M\, m/z 210, (M}OH)\,
tion of the sample molecules with a known pre-
m/z 197, (M}NO)\ and m/z 181, (M}NO2)\. The
selected set of reagent ions. These reagent ions are
detection limit of explosives in the NCI mode is, in
produced by ion}molecule reactions in a reagent
general, lower by one order of magnitude then in the
gas introduced in the ion source of the mass spec-
positive CI mode.
trometer at a pressure of 0.1 to 1.0 Torr. Common
reagent gases are methane and isobutane.
CI mass spectra contain usually an MH# ion and
little fragment ions. It is, therefore, suitable for mo-
Conclusions
lecular weight identiRcation. In many cases, in the CI While GC is a chromatographic method and needs
mass spectrum of an explosive an MH# ion might be comparison of retention times with standards, mass
observed, while in the EI mass spectrum of the same spectrometry is an identiRcation method and pro-
compound there is no molecular ion. For example, vides a ‘Rngerprint’ of the investigated compound.
the EI mass spectra of NG and PETN are similar, The combination of GC with MS incorporates both
III / EXPLOSIVES / Liquid Chromatography 2767
separation and identiRcation capabilities, and is mination of nitroaromatics in water. Journal of

therefore superior to GC alone. Chromatography 518: 21}40.
Both GC and GC-MS, although being suitable Francis ES, Wu M, Farnsworth PB and Lee ML (1995)
techniques for the separation and analysis of ex- Supercritical Suid extraction/gas chromatography
plosives, have a limitation in that the injector and with thermal desorption modulator interface and
nitro-speciRc detection for the analysis of explo-
column have to be heated. This fact necessitates
sives. Journal of Microcolumn Separations 7:
taking special precautions when dealing with the 23}28.
more thermally labile compounds. Liquid Haas R, Schreiber I, v. Low E and Stork G (1990) Concep-
chromatography}mass spectrometry (LC-MS), where tion for the investigation of contaminated munition
the injector and column are at room temperature, plants. 2. Investigation of former RDX-plants and Rlling
does not have these limitations, and is therefore a bet- stations. Fresenius Journal of Analytical Chemistry 338:
ter choice when analysing the more thermally labile 41}45.
explosives. However, GC-MS is readily available in Hable M, Stern C, Asowata C and Williams K (1991) The
most analytical laboratories, while LC-MS is not. determination of nitroaromatics and nitramines in
This situation is expected to change in the next 5 to ground and drinking water by wide-bore capillary gas
10 years, which will place LC-MS as the method of chromatography. Journal of Chromatographic Science
29: 131}135.
choice for the separation and analysis of explosives.
Kolla P (1994) Gas chromatography, liquid chromatogra-
In both GC-MS and LC-MS, the addition of tan- phy and ion chromatography adapted to the trace analy-
dem mass spectrometry (MS-MS) provides an extra sis of explosives. Journal of Chromatography A 674:
dimension for improved selectivity and therefore im- 309}318.
proved identiRcation. Slack GC, McNair HM and Wasserzug L (1992) Character-
ization of Semtex by supercritical Suid extraction and
off-line GC-ECD and GC-MS. Journal of High Resolu-
See also: II/Chromatography: Gas: Detectors: Mass tion Chromatography 15: 102}104.
Spectrometry; Detectors: Selective. Extraction: Analyti- Tamiri T, Zitrin S, Abramovich-Bar S, Bamberger Y and
cal Extractions; Solid-Phase Extraction; Solid-Phase Sterling J (1992) GC/MS Analysis of PETN and NG in
Microextraction. Explosives: Liquid Chromatography; Post-Explosion Residues. In: Yinon J (ed.) Advances in
Thin-Layer (Planar) Chromatography. Analysis and Detection of Explosives, pp. 323}334.
Dordrecht: Kluwer Academic Publishers.
Welsch T and Block H (1997) Separation and enrich-
Further Reading ment of traces of explosives and their by-products
from water by multiple micro liquid extraction for their
Douse JMF (1987) Improved method for the trace analysis determination by capillary gas chromatography.
of explosives by silica capillary column gas chromatog- Fresenius Journal of Analytical Chemistry 357:
raphy with thermal energy analysis detection. Journal of 904}908.
Chromatography 410: 181}189. Yinon J (1996) Trace analysis of explosives in water by gas
Douse JMF and Smith RN (1986) Trace analysis of explos- chromatography}mass spectrometry with a temper-
ives and Rrearm discharge residues in the Metropolitan ature-programmed injector. Journal of Chromatography
Police Forensic Science Laboratory. Journal of Energetic A 742: 205}209.
Materials 4: 169}186. Yinon J and Zitrin S (1993) Modern Methods and Ap-
Feltes J, Levsen K, Volmer D and Spiekermann M (1990) plications in Analysis of Explosives. Chichester:
Gas chromatographic and mass spectrometric deter- John Wiley.
U. Lewin-Kretzschmar, J. Efer and forensic science and toxicology (investigation of ex-
W. Engewald, University of Leipzig, Leipzig, plosions or of criminal actions) and environmental
Germany monitoring (water and soil analysis at sites intensively
used for military purposes). Figure 1 shows some of
the more common explosives.
In the literature various methods and procedures
Introduction have been described for analysing these compounds.
Explosive analysis is important in different areas: ex- In the last few years, the investigation of explosives in
plosive manufacture (quality and wastewater control), the water and soil around former ammunition plants,
2768 III / EXPLOSIVES / Liquid Chromatography
Figure 1 Structures of common explosives.
munition depots or dumps dating back to World explosive-related compounds in complex samples, in-
War II has become increasingly important. In these cluding separation, detection and sample prepara-
samples there are not only explosives but also tion, focusing in particular on compounds occurring
their by-products and metabolites. For example, in samples around former ammunition plants
water samples from around the former ammunition (Table 1).
plant at Elsnig (Saxony, Germany) contained many
explosive-related compounds of various classes
(Table 1). Sample Preparation
These constituents, which range in concentra- Water Samples
tions from ng L\1 to mg L\1, make it difRcult to
analyse such samples. However, to assess the toxic Brown glass bottles should be Rlled up to the brim
potential, reliable analysis of all the compounds with water samples and the bottles made gas-tight,
down to the trace amounts (about 0.1 g L\1) is for example, using TeSon packings, and stored at
necessary, as some are highly toxic, carcinogenic or 43C. To prevent adsorption losses or degradation
mutagenic. at the glass surfaces, the addition of methanol or
Capillary gas chromatography offers advantages of acetonitrile or the use of silanized glass vessels is
separation efRciency and favourable detection limits; recommended. To prevent bacterial degradation, so-
however, because of the thermal instability and high dium azide (about 0.5 g L\1) can be added to the
polarity of some compounds, high performance samples.
liquid chromatography (HPLC) determination is of- In principle, for extraction and enrichment, various
ten the method of choice. This requires thorough methods such as liquid}liquid extraction (LLE), solid-
optimization of HPLC conditions (stationary and phase extraction (SPE) or solid-phase microextrac-
mobile-phase) as well as selective and sensitive detection (SPME) can be used.
tion systems and, in some cases, selective sample
preparation or pre-separation of the sample into dif- Neutral compounds An effective and reproducible
ferent fractions. liquid}liquid extraction of neutral explosive com-
The aim of this article is to provide an overview of pounds such as the nitroaromatics can take place
the possibilities given by HPLC methods to determine with various solvents such as dichloromethane, ethyl
Table 1 Compounds in water samples in the neighbourhood (stirring or shaking in an Erlenmeyer Sask). In order
of the former ammunition plant in Elsnig to obtain good recovery for the relatively highly
volatile mononitrated aromatics and chloroaro-
Compounds Abbreviations
matics, great care should be taken when concentrat-
Nitroaromatics and nitramines ing or redissolving extracts. To extract polar com-
2,4,6-Trinitrotoluene 2,4,6-TNT pounds like hexogen, octogen or nitroguanidine, con-
Hexahydro-1,3,5-trinitro-1,3,5-triazine RDX/Hexogen tinuous LLE in rotary perforators is more effective
2,2’,4,4’,6,6’-Hexanitrodiphenylamine Hexyl
(Figure 2).
1,3-Dinitrobenzene 1,3,-DNB
2,4-Dinitrotoluene 2,4-DNT Good results are also obtained with SPE. For neu-
2,6-Dinitrotoluene 2,6-DNT tral nitroaromatics recoveries of between 70 and
3,4-Dinitrotoluene 3,4-DNT 100% are reached with octadecylsiloxane-bonded sil-
Nitrobenzene NB ica materials (RP-18). Increasingly, the highly porous
2-Nitrotoluene 2-NT
(speciRc surface '1200 m2 g\1) and high purity ad-
3-Nitrotoluene 3-NT
4-Nitrotoluene 4-NT sorbents based on styrene-divinylbenzene (SDVB)
Octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine Octogen/HMX copolymer are used and these are particularly suitable
1,3,5-Trinitrobenzene 1,3,5-TNB for the more water soluble compounds such as
Chloroaromatics nitramines.
Chlorobenzene CIB The US Environmental Protection Agency (EPA)
1-Chloro-2,4-dinitrobenzene 1-Cl-2,4-DNB method makes use of salting-out-effects in the extrac-
1-Chloro-4-nitrobenzene 1-Cl-4-NB tion. This enables nitramines, nitroaromatics and ni-
1,2-Dichlorobenzene 1,2-DClB
1,4-Dichlorobenzene 1,4-DClB
trate esters to be extracted with solvents freely
2,3-Dichloronitrobenzene 2,3-DClNB miscible with water, such as acetonitrile. Here, good
2,5-Dichloronitrobenzene 2,5-DClNB recoveries are obtained, but reproducibility is not as
1,2,4-Trichlorobenzene 1,2,4-TClB good as with the methods mentioned above.
Amino- and aminonitroaromatics
2-Amino-4,6-dinitrotoluene 2-A-4,6-DNT Acidic compounds The extraction of acidic com-
4-Amino-2,6-dinitrotoluene 4-A-2,6-DNT pounds such as nitrophenols and nitrobenzoic acids
2-Amino-4-nitrotoluene 2-A-4-NT
should take place at low pH in order to ensure the
2-Amino-6-nitrotoluene 2-A-6-NT
4-Amino-2-nitrotoluene 4-A-2-NT presence of these compounds in their nondissociated
2,6-Diamino-4-nitrotoluene 2,6-DA-4-NT form. For this purpose a pH value of 2 has proved
2,3-Diaminotoluene 2,3-DAT successful.
2,4-Diaminotoluene 2,4-DAT Also, continuous extraction in a rotary extractor
2,6-Diaminotoluene 2,6-DAT
leads to a higher yield than is the case in discontinu-
3,5-Dinitroaniline 3,5-DNA
ous extraction (Figure 3). Suitable solvents are dich-
Nitrophenols loromethane and ethyl acetate.
2,4-Dinitrophenol 2,4-DNP
3,4-Dinitrophenol 3,4-DNP
In general, lower recoveries will be obtained for the
3,5-Dinitrophenol 3,5-DNP volatile ortho-substituted nitrophenols if a concentra-
2-Methyl-4,6-dinitrophenol 2-M-4,6-DNP tion step is necessary after the extraction.
4-Methyl-2,6-dinitrophenol 4-M-2,6-DNP High recoveries are obtained on SDVB copolymers.
3-Methyl-2-nitrophenol 3-M-2-NP In SPE with RP-18 materials, recoveries '70% for
3-Methyl-4-nitrophenol 3-M-4-NP
4-Methyl-2-nitrophenol 4-M-2-NP
the mononitrophenols are only reached if large
5-Methyl-2-nitrophenol 5-M-2-NP amounts of salt (300 g NaCl L\1) are added.
3-Nitrophenol 3-NP An efRcient enrichment of acidic compounds is also
4-Nitrophenol 4-NP possible after ion pair formation with tetrabutylam-
2,4,6-Trinitrophenol Picric acid/PA monium chloride in a neutral to basic medium or by
Nitrobenzoic acids extractive derivatization by means of acetic anhydr-
2-Amino-4-nitrobenzoic acid 2-A-4-NBA ide or pentaSuorobenzoyl chloride in the presence of
2,4-Dinitrobenzoic acid 2,4-DNBA
a phase transfer catalyst with dichloromethane as
2-Nitrobenzoic acid 2-NBA
3-Nitrobenzoic acid 3-NBA solvent.
4-Nitrobenzoic acid 4-NBA
Basic compounds An efRcient enrichment of
aminoaromatics and especially of diaminoaromatics
acetate, methyl isobutylketone and methyl tert-butyl can be reached at pH 12 with continuous LLE
ether at different pH values. As a rule, recovery with dichloromethane or SPE on SDVB copolymer
of '70% can be achieved with triple extraction materials (Figure 4). Discontinuous extraction with
Figure 2 Comparison of various extraction techniques for neutral compounds. Samples of 0.5 L (water spiked with 2 g L\1 for each
component and adjusted to pH 9 with 0.1 mol L\1 sodium hydroxide were enriched as follows: (A) discontinuous LLE (open columns):
stirring three times with 25 mL dichloromethane for 30 min in an Erlenmeyer flask; (B) continuous LLE (hatched columns): 0.5 L
extraction with 150 mL dichloromethane in a heavy-phase rotary perforator (Normag); (C) SPE-RP-18 (dotted columns): enriched on
2 g PolarPlus (Baker), conditioned with 3 mL acetone, methanol and water, and eluted with 6 mL methanol; (D) SPE-SDVB (filled
columns): enriched on 200 mg LiChrolut EN (Merck), conditioned with 3 mL acetonitrile, methanol and 9 mL water, and eluted with 2 mL
methanol acetonitrile mixture (1 : 1).
dichloromethane is not effective. The use of RP-18 Soil Samples

materials or ethyl acetate as a solvent is out of the
question because of the high pH. The basic condition for reliable analysis of soil sam-
ples is representive sampling and good homogeniz-
Fractionation Where there are very complex sam- ation of the samples.
ples, for example, the water samples from Elsnig, The method most frequently used to prepare
precise qualitative and quantitative analysis is only the soil samples to analyse explosives is extraction in
possible after pre-separation of the components. an ultrasonic bath. Compared with Sohxlet extrac-
A suitable method is class fractionation based on LLE tion it has several advantages: careful treatment
(Figure 5) at various pH values. of thermolabile compounds, easy handling, mini-
Here, by means of the discontinuous dichloro- mum apparatus expenditure and low consumption of
methane extraction at pH 9, the nitro and mono- solvents.
aminoaromatics } which in most cases are the main Suitable solvents are, in principle, acetone, meth-
contaminants } can be almost completely separated anol and acetonitrile. However, with a view to the
from the acidic and basic compounds. Only the more subsequent HPLC determination, methanol and
polar nitramines are not completely extracted and are acetonitrile should rank Rrst.
partially included in the other two fractions. Using Nitroaromatics are quantitatively extracted by
a more efRcient extraction technique in the Rrst ex- both solvents. But for the more polar compounds like
traction step (for example, continuous extraction or octogen, hexogen and hexyl, acetonitrile is preferred
SPE on SDVB), selective pre-separation is not (Table 2).
possible since acidic and basic compounds would Ultrasonic extraction is described in detail in
also be partially extracted. US-EPA-8330. In this case, 2 g of soil is extracted
Figure 3 Optimization of extraction for acidic compounds. Samples of 0.5 L (water spiked with 2 g L\1 for each component and
adjusted to pH 2 with 0.1 mol L\1 hydrochloric acid) were extracted with 150 mL dichloromethane in a heavy-phase rotary preforator or
with 200 mL ethyl acetate in a light-phase rotary perforator, respectively. Open columns, MeCl 3;30 min; hatched columns, MeCl 4 h;
dotted columns, MeCl 10 h; filled columns, Etac 4 h.
with 10 mL acetonitrile over 18 h at room tem- carefully optimized separation conditions. Therefore,
perature. before separation in the case of complex samples,
A new efRcient method for extraction of soil sam- pre-separation of the components into different frac-
ples is accelerated solvent extraction (ASE), which tions by HPLC is useful.
has proved suitable for nitroaromatics.
Neutral Compounds
RP-18 phases are very suitable for separation of com-
HPLC Separation Systems plex mixtures of nitro- and nitroaminoaromatics,
In general, RP-18 materials are used to separate ex- nitramines and nitrate esters with methanol}water
plosive-related compounds. According to sample or methanol}buffer mobile phases.
composition and detector selection, the composition In general, at a methanol}water ratio of about 1:1,
of the mobile phase (type and amount of organic the retention of compounds on RP-18 phases will
modiRer, buffer additives) varies considerably. The increase as follows: nitroguanidine ( octogen (
most common mobile phases are buffered or unbuf- hexogen ( EGDN ( DEGN ( 1,3-DNB ( 2,4,
fered methanol}water and acetonitrile}water mix- 6-TNT ( 4-A-2,6-DNT ( 2-A-4,6-DNT ( 2,6-
tures in isocratic or gradient operation. DNT ( 2,4-DNT ( 2-NT ( 4-NT ( 3-NT (
Ethanol, n-propanol and dioxane as modiRers or PETN ( diphenylamine (Figure 6).
ternary solvent mixtures such as water}methanol} For some compounds, however, various RP-18 col-
acetonitrile or water}methanol}tetrahydrofuran are umns show different selectivities, which result in co-
of little practical importance and result in too small elution and retention reversal of various pairs of sub-
a selectivity change compared with binary mobile stances, as shown in Table 3. Two columns of com-
phases. plementary selectivity can be used to verify the separ-
Because of the generally limited separation perfor- ation results or to reduce peak overlapping.
mance of HPLC, complete separation of all explosive- If chlorinated aromatics additionally occur in real
related compounds cannot be achieved on any one samples (Table 1) these can be determined under the
column in one chromatographic run, not even under same conditions as the nitroaromatics. However, for
Figure 4 Extraction of basic compounds. Samples of 0.5 L (water spiked with 2 g L\1 for each component and adjusted to pH 12
with 1 mol L\1 sodium hydroxide) enriched (A) by continuous LLE with 150 mL dichloromethane and (B) by SPE on 500 mg LiChrolut
EN (Merck), conditioned with 3 mL acetonitrile, methanol and 9 mL water, and eluted with a 2 mL methanol/acetonitrile mixture (1:1).
Open columns, LLE 4 h; hatched columns, LLE 10 h; filled columns, SPE-SVDB.
Figure 5 Optimized fractioned extraction procedure. For description of the extraction steps see Figures 2}4. (Reproduced with
permission from Lewin U et al. (1997) Chromatographia 45: 91.)
Table 2 Extraction of spiked soil samples with different siloxane and aminopropylsiloxane-bonded silica
solvents sorbents. As the normal-phase mode has considerable
disadvantages (disturbance by traces of water, no
Compound Acetonitrile Methanol
gradient elution), these separation systems are used
Recovery (%) RSD (%) Recovery (%) RSD (%) only rarely and in most cases only with detectors
which are incompatible with aqueous mobile phases,
Hexogen 97 2.5 82 2.4 such as the thermal energy analyser (TEA) and the
Hexyl 86 2.2 73 1.5
electron-capture detector (ECD).
Octogen 95 3.4 69 4.3
2,4,6-TNT 98 2.1 98 2.1 Large selectivity differences in RP-18 phases are
also obtained on nitrophenyl-modiRed silica gel and
2 g soil spiked with 10 mg kg\1 for each component, extracted on porous graphitic carbon (PGC) (Table 4).
with 10 ml solvent for 15 min in an ultrasonic bath. RSD, relative Thus, retention on these phases will increase with
standard deviation (n"3).
the growing number of nitro groups. Furthermore, in
contrast to the RP-18 columns, large retention differ-
ences are observed for the isomeric dinitro and
the more retained dichlorocompounds, gradient aminodinitro compounds. In addition, the PGC
elution is recommended. phase, due to its high hydrophobicity, shows gener-
To clarify peak overlapping, other phases in ally higher retentions, which require a higher meth-
addition to RP-18 phases have proven their value. anol content ('85%) and the separation perform-
US-EPA, for example, recommends a cyanopropyl ance is not satisfactory, due to the low efRciency of
column as a second column showing a clearly differ- such columns.
ent selectivity: nitroguanidine(NB(toluol(2- In addition to the commercially obtainable col-
NT(4-N ....T(3-NT(EGDN(1,3-DNB(2,6- umns, special materials such as the charge transfer
DNT(2,4-DNT(TNT(4-A-2,6-DNT(2-A- phases like arylpropylether, N-propylaniline and saf-
4,6-DNT(hexogen(tetryl(diphenylamine( rol phases or a two-dimensional coupling of RP-18
octogen(PETN. phases with these columns for increasing the selectiv-
Similar retention orders are also observed under ity have been tested but offer no great advantage over
normal-phase conditions on silica gel, cyanopropyl- the common materials.
Figure 6 Chromatogram of a standard mixture of explosive-related compounds. 51% methanol}49% water (v/v); Spherisorb ODS
2 column, 5 m, 250;4 mm (HP); 1 mL min\1; 253C; UV 254 nm.
Table 3 Retention behaviour of neutral explosive-related compounds on different RP-18 columns
Column Spherisorb ODS 2 Eurospher 100 C18 UltraSep Es EX

(HP) (Knauer) (Sepserv)
Compound Peak order ka Peak order kb Peak order kc
NG 1 0.22 1 0.14 1 0.07

Octogen 2 0.60 2 0.63 2 0.51
Hexogen 3 1.55 3 1.67 3 1.82
1,3,5-TNB 4 2.21 5 2.26 4 2.51
2-A-6-NT 5 2.23 4 2.00 5 2.71
2-A-4-NT 6 2.74 6 2.59 6 3.03
Tetryl 7 3.03 8 3.84 7 3.61
1,3-DNB 8 3.48 7 3.47 8 4.37
2,4,6-TNT 9 3.90 11 4.98 10 4.56
NB 10 4.00 9 4.14 8 4.14
4-A-2,6-DNT 11 4.36 12 5.63 11 5.36
3,4-DNT 12 4.68 10 4.36 12 5.41
2-A-4,6-DNT 13 4.72 13 5.85 13 6.18
2,6-DNT 14 5.55 14 6.09 14 6.67
2,4-DNT 15 6.06 15 6.28 15 7.63
2-NT 16 7.67 16 7.60 16 8.48
4-NT 17 8.36 17 8.32 17 9.46
3-NT 18 9.05 18 8.98 18 10.16
Dimension of each column: 250;4 mm; 5 m; a hold-up time 1.57 min; b hold-up time 1.70 min; c hold-up time 1.18 min. 51%
methanol}49% water (v/v); 1 ml min\1; 273C; 10 g mL\1 per component.
Acidic Compounds The separation of the underivatized compounds

at acidic pH between approximately pH 2 and 4
The separation of the acidic explosives hexyl and is relatively simple. For acidiRcation of the
picric acid and the metabolites nitrophenols and mobile phase, many acids and buffers can be used.
nitrobenzoic acids is, in principle, possible on RP- For example, to separate nitrophenols, in addi-
materials under the following conditions: at acidic tion to octogen and hexogen, a 0.01 mol L\1
pH value; with the addition of ion pair reagents like sodium dihydrogen phosphate/phosphoric acid
hexadecyltrimethylammonium chloride at neutral or buffer (pH 3) on RP-18-columns is suitable
basic pH values; and after methylation. (Table 5).
Table 4 Retention behaviour of neutral explosive-related compounds on different RP phases
Column RP-18 phase a Nitrophenyl phase b PGC phase c

(Eurospher 100; Knauer) (Cosmosil 5 NPE) (Hypercarb; Shandon)
Mobile phase 51% MeOH/49% H2O 62% MeOH/38% H2O 95% MeOH/5% H2O
Compound Peak order k Peak order k Peak order k
Octogen 1 0.63 13 18.94 3 3.18

Hexogen 2 1.67 7 8.21 1 1.22
1,3,5-TNB 3 2.26 5 6.95 12 41.23
Tetryl 4 3.84 8 8.28 4 4.39
1,3-DNB 5 3.47 4 6.23 9 18.39
2,4,6-TNT 6 4.98 11 13.93 10 28.22
4-A-2,6-DNT 7 5.63 9 9.95 8 17.26
2-A-4,6-DNT 8 5.85 12 17.83 13 41.40
2,6-DNT 9 6.09 6 7.31 5 5.19
2,4-DNT 10 6.28 10 13.35 11 28.47
2-NT 11 7.60 1 4.64 2 3.00
4-NT 12 8.32 2 5.42 7 6.71
3-NT 13 8.98 3 5.62 6 5.24
a
Column: 250;4 mm; 5 m, hold-up time 1.73 min. b Column: 150;4.6 mm; 5 m, hold-up time 1.09 min. c Column: 100;4.6 mm;
7 m, hold-up time 1.18 min. 1 mL min\1; 273C; 10 g mL\1 per component.
Table 5 Retention of nitrophenols and nitramines on RP-18 retained or only poorly separated should be separated
columns at pH 3 on porous polymer or PGC phases, because these
phases have relatively homogeneous nonpolar surfa-
Column Eurospher 100 C18 Spherisorb-ODS 2
ces and high stability over the full pH range.
Compound Peak order ka Peak order kb Even using various mobile phases and buffer sys-
tems, these columns do not show any satisfactory
Octogen 1 0.60 1 0.53 separation of the compounds of interest; this is due to
2,6-DNP 2 1.39 2 0.66
the low efRciency of commercial columns.
Hexogen 3 1.67 3 0.98
PA 4 2.06 4 1.07
2,4-DNP 5 2.31 5 1.13
4-NP 6 2.46 6 1.84
Detection
3-NP 7 2.76 7 2.03
UV Detection
2,5-DNP 8 3.22 8 2.30
3,4-DNP 9 3.25 10 2.96 To detect explosive-related compounds, UV is mainly
4-M-2,6-DNP 10 3.38 9 2.27
3-M-2-NP 11 3.56 11 2.98
used. Aromatic nitrocompounds are UV-absorbing
2-NP 12 3.73 12 3.10 and can, as a rule, be sensitively detected at 254 nm
3-M-4-NP 13 3.83 13 3.78 (Figure 9). At this wavelength it is also possible to
2-M-4,6-DNP 14 6.20 15 5.45 detect nitramines and aminoaromatics. In addition to
3,5-DNP 15 6.78 14 4.58 sensitive detection, selectivity against matrix compo-
4-M-2-NP 16 8.37 16 7.79
5-M-2-NP 16 8.37 16 7.79
nents and eluent impurities is reached at this
wavelength, as most interferents absorb at lower
Dimension of both columns: 250;4 mm; 5 m; a hold-up time wavelengths.
1.73 min; b hold-up time 1.59 min. 51% methanol}49% 0.01 mol As a rule, nitrate esters and chlorobenzenes do not
L\1 sodium dihydrogen phosphate}phosphoric acid. buffer pH show a maximum at wavelengths higher than 200 nm.
3(v/v); 1 mL min\1; 273C; 10 g mL\1 per component.
Therefore, these compounds should be detected at the
lowest possible wavelength. For methanol-containing
To separate nitrobenzoic acids a further decrease in eluents the minimum practicable wavelength is around
pH is necessary. Good separation of these com- 210 nm.
pounds in the presence of nitrophenols and nitra- The UV detector shows a high linearity for the
mines can be obtained at pH 2 (addition of 0.005 mol explosive-related compounds in the concentration
L\1 sulfuric acid) and a methanol content of 47% range of about 0.01 to 100 g mL\1.
(Figure 7). For aromatic nitro-compounds, limits of detection
Under these conditions for the determination of (LODs) reach 5}50 ng mL\1 (0.1}1 ng absolutely for
hexyl a gradient after 20 min to 85% methanol with- an injection volume of 20 L) in the sample solution.
in 20 min is used. For nitrate esters and nonnitrated chlorobenzenes the
LODs are around 100}250 ng mL\1 (or 2}5 ng abso-
Basic Compounds lutely).
The RP-18 phases based on silica are, in principle, Using variable wavelength detectors, multichannel
also suitable for the determination of the metabolites wavelength detectors or photodiode array detectors,
diaminotoluenes, diaminonitrotoluenes and nitroani- it is possible to optimize the selectivities and LODs
lines. For this, however, particularly inert materials for certain purposes, by measuring at the absorption
with a low silanol group activity are necessary be- maximum or at several wavelengths in one chromato-
cause otherwise the peaks show marked tailing and graphic run. For example, hexyl can be selectively
large peak widths. For example, a Eurospher column detected at high and sensitively at 420 nm.
showed good properties in this respect (Figure 8). Additionally, the photodiode array detector en-
We have not been able to observe an increase in ables compounds to be identiRed with marked ab-
separation performance with falling pH improving sorption maxima and minima (e.g. nitrophenols,
peak shapes, as is often described in the literature. On nitrobenzoic acids and nitroaromatics) by means
the other hand there is, as expected a reduction in the of the simultaneously recordable spectra.
retention under these conditions. An increase in re-
Electrochemical Detection
tention by restraining the protonation of the basic
compounds at pH values '10 is not to be recom- In principle, all the nitrocompounds can be deter-
mended because of the instability of the phases used. mined by means of the electrochemical detector in the
Alternatives to the separation of the diamino com- reduction mode. Phenols and amino compounds can
pounds, which on RP-18 phases are insigniRcantly be detected in the oxidation mode.
Figure 7 Chromatogram of a standard mixture of nitrophenols, nitro benzoic acids, nitramines and hexyl. Eurospher
100 RP 18 column, 5 m, 250;4 mm; 47% methanol}53% 0.01 mol L\1 sulfuric acid (pH 2) (v/v), after 20 min linear gradient to 85%
methanol within 20 min, 1.0 mL min\1, 273C; UV 254 nm. (Reproduced with permission from Lewin U et al. (1997) Chromatographia
45: 91.)
The method mainly used in HPLC for electro-

chemical detection is amperometry: a constant
optimized potential difference (working potential)
is applied between the working and reference elec-
trode, which has previously been determined from
cyclovoltammograms or hydrodynamic voltammo-
grams.
In commercially available electrochemical de-
tectors, solid electrodes are used as they are easier to
handle, although nitroaromatics can be detected sen-
sitively with liquid mercury electrodes and in particu-
lar with the hanging mercury drop electrode.
To determine explosive-related compounds, the
standard electrode material, glassy carbon, has
proved its value, because it can be used over a wide
potential range of about !1.3 to about #1.3 V.
In addition, amalgamated gold electrodes or mer-
Figure 8 Chromatogram of a standard mixture of amino
cury Rlm electrodes have been used for reductive
and diaminoaromatics Eurospher 100 RP 18 column, 5 m, detection.
250;4 mm (Knauer); 40% methanol}60% 0.01 mol L\1 water The selectivity of detection can be inSuenced by the
(v/v), 1.0 mL min\1, 273C; UV 254 nm. choice of working potential. In general, substances
Figure 9 Spectra of explosive-related compounds. Detected with HP 1050 variable wavelength detector in the scanning mode with
51% methanol}49% water or 0.01 mol L\1 phosphate buffer (pH 3) (v/v) for the acidic compounds, respectively.
whose half-wave potential is at least 150 mV larger by measuring simultaneously at various potentials or
than the working potential are not detected. at each optimal potential of the compounds.
Dual cells or electrode arrays may result in an
increase in the informational content of a chromato- Reduction mode Nitro compounds have very differ-
graphic run or in a reduction in the limit of detection ent half-wave potentials. They depend on the type,
detection, the aminoaromatics and nitrophenols have

very different half-wave potentials, depending on the
type, number and position of the substituents
(Table 6).
For example, nitro groups will increase half-wave
potentials, whereas they are reduced by amino and
methyl groups, which can be utilized for selective
detection.
However, the greatest selectivity advantages of
anodic detection in analysing explosive-related com-
pounds is that aminoaromatics become selective in the
presence of nitroaromatics and nitramines (Figure 11).
Likewise, phenols and hexyl can be detected selec-
tively, in addition to nitrobenzoic acids and nitra-
mines (Figure 12).
In this way, the electrochemical detector, especially
coupled with the UV detector, which is almost univer-
sal for explosive-related compounds, leads to a valu-
able gain in information in real samples.
Furthermore, for most nitrophenols and aminoaro-
matics, lower LODs are reached by anodic detection
compared with UV detection. In particular, the detec-
tion limits for diamino compounds are lower by a fac-
tor of up to 100 (up to 0.05 ng mL\1 or 1 pg abso-
lute) without enrichment.
Figure 10 Electrochemical detector chromatograms of the Table 6 Half-wave and optimal working potentials for oxidative
neutral fraction of a groundwater sample from Elsnig at different detection
potentials. Eurospher 100 RP 18 column, 5 m, 250;4 mm
(Knauer); 51% methanol}49% 0.01 mol L\1 phosphate buffer Compound E1/2 (V) E opt. (V)
(pH 3) (v/v). (Reproduced from Lewin U et al. (1996) Journal of
2-A-4,6-DNT 1.05 '1.20
Chromatography A 730: 161, with permission from Elsevier
4-A-2,6-DNT 1.05 '1.20
Science.)
2-A-3-NT 0.95 1.20
2-A-4-NT 0.95 1.20
number and position of the substituents and rise, for 2-A-6-NT 0.95 1.20
example in the following way: 4-A-2-NT 0.90 1.10
2,3-DAT 0.08 0.30
trinitro( dinitro( mononitroaromatics 2,4-DAT 0.25 0.50
2,6-DAT 0.28 0.50
4 nitroanilines 2-M-4,6-DNP 1.05 1.20
4-M-2,6-DNP 1.00 1.20
2,6-DA-4-NT 0.60 0.75
In this way a limited selective detection of certain 2-M-3-NP 0.65 0.80
compounds is possible (Figure 10). The optimum po- 3-M-2-NP 0.65 0.80
tential for the detection of all components is in per- 3-M-4-NP 0.90 1.25
4-M-2-NP 0.85 1.15
chlorate eluent (pH 5.5) at around !1.2 V.
4-M-3-NP 0.65 0.80
General problems in reductive detection are caused 5-M-2-NP 0.85 1.15
by the difRculty of complete removal of oxygen dis- 2-NA 1.15 '1.20
solved in the mobile phase or in the sample, which is 4-NA 1.05 1.20
reduced at potentials of about !0.5 V and leads to 2-NP 0.90 1.20
3-NP 0.80 1.00
greater disturbances by system peaks and high resid-
4-NP 0.95 1.20
ual currents. In most cases it is not possible to reach PA '1.20 '1.20
much lower LODs than with UV detection.
Conditions: ELCD HP 1049 A with glassy carbon thin-layer
working electrode and Ag/AgCl reference electrode; 51% meth-
Oxidation mode A great advantage of oxidative de- anol}49% 0.01 mol L\1 sodium dihydrogen phosphate}phos-
tection is that, unlike the reductive mode, oxygen has phoric acid buffer pH 3 (v/v) or sodium perchlorate solution (pH
no negative inSuence on the detection. Like reductive 5.5) for diaminoaromatics, respectively; 1 mL min\1; 273C.
Figure 11 Chromatograms of the basic fraction of a groundwater sample from Elsnig (upper level). (A) UV (254 nm); (B) electro-
chemical detector (#0.7 V). Eurospher 100 RP 18 column, 5 m, 250;4 mm; 40% methanol}60% 0.01 mol L\1 sodium perchlorate
solution (v/v), 1.0 mL min\1, 273C. (Reproduced with permission from Lewin U et al. (1997) Chromatographia 45: 91.)
The electrochemical detector shows high linearity oxidized on a glassy carbon electrode. For this, LODs
over a concentration range of 104 but, compared with of 120}250 pg have been found. The expense of the
UV detection, reproducibility is somewhat lower. apparatus, however, is relatively high and the yield of
Furthermore, there is often a decrease in the response the photolysis is very different for the various com-
over a long measuring period: this can be attributed pounds.
to passivation of the electrode surface and requires its
Mass Spectrometric Detection
regeneration.
Determination of the nitroaromatics in the easier In the last few years the development of mass selective
oxidation mode takes place after photolysis in a post- detectors for coupling with HPLC has made good
column reactor via the nitrite produced, which is progress. In addition to thermospray ionization (TSI),
Figure 12 Chromatograms of the acidic fraction of a groundwater sample from Elsnig (lower level). (A) UV (254 nm); (B) electro-
chemical detector (#1.2 V). Eurospher 100 RP 18 column, 5 m, 250;4 mm (Knauer, Berlin); 51% methanol}49% 0.01 mol
L\1 phosphate buffer (pH 3) (v/v), after 20 min linear gradient to 80% methanol within 20 min, 1.0 mL min\1 (Reproduced with
permission from Lewin U et al. (1997) Chromatographia 45: 91.)
the atmospheric pressure ionization (API) techniques, droxyl and amino groups). Thus, reliable identiRca-
electrospray ionization (ESI) and atmospheric pres- tion in very complex samples is possible by the
sure chemical ionization (APCI) are increasingly gain- HPLC-NMR coupling which has been developed
ing in importance. over the last few years.
Explosives can be detected with TSI as negative In the continuous Sow mode at low Sow rates
ions (mostly [M#CH3COO]\ or [M!H]\). Nitra- ((0.02 mL min\1) and large injection volumes (ap-
mines and amino compounds can be registered as proximately 400 L), determination of explosives in
positive ions (mostly [M#NH4]#). With negative the lower microgram range is possible. The reproduc-
ionization in the full-scan mode, LODs are in the ng ibility of about 2% in the upper microgram range is
mL\1 range. With selected ion monitoring in the acceptable.
Rlament-on negative ion mode, the LODs are a factor
of 100 lower.
As a result of the electron-withdrawing effect of the
Further Detection Techniques
nitro groups, nitroaromatics can be detected with API Fluorescence detection is not suitable for the deter-
as negative ions [M!H]\ (Figure 13). mination of nitro compounds, as nitro groups will
The sensitivity of detection depends on the type, diminish Suorescence intensity. However amino com-
number and position of the substituents. It increases pounds can be determined very selectively and sensi-
with the number of nitro groups and is considerably tively.
higher for nitrophenols and nitrobenzoic acids than As nitro compounds form nitrogen monoxide by
for the corresponding neutral nitrobenzenes and pyrolysis, they can also be detected by a thermal
nitrotoluenes. energy analyser. The sensitivity depends on the sub-
Furthermore, sensitivity depends on the composi- stance class and is lower for nitroaromatics than for
tion of the mobile phase such as the type and quantity nitramines and nitrate esters. An advantage of the
of the organic modiRer, buffer additives and pH detector is its speciRcity for compounds carrying oxy-
value. nitrogen functional groups. Because of the complex
As with TSI, hexogen and octogen form clusters apparatus and the restriction to the normal-phase
with acetate ions [M#CH3COO]\. A cluster forma- separation mode, the detector has not achieved wide
tion of these compounds with ammonium ions usage.
[M#NH4]# should, in principle, also be possible in Occasionally, reports have been published on fur-
the positive mode. ther pre-column and post-column derivatization tech-
In addition to the molecular ions and molecular niques. For example, for the amino compounds the
cluster ions, fragment ions are also formed, and elim- diazotization and coupling with N-(1-naphtyl)-
ination of oxygen or reductions and rearrangements ethylenediammonium chloride into azo dyes is suit-
of the nitro group are observed. Nitrophenols frag- able. This reaction is also suitable for the determina-
ment with the loss of the hydroxyl and the nitro tion of nitroaromatics after conversion with tita-
group. This can be used to obtain structural informa- nium(III) chloride into the amines or after photolysis
tion. However, the assignment of position isomers is and diazotization of the nitrite formed with sul-
difRcult, so that to investigate complex samples, care- fanilamide.
fully optimized HPLC conditions are needed and the The electron-capture detector, widely used in
additional use of a UV detector is an advantage. gas chromatography, was not successful in prac-
The LODs largely depend on the particular com- tice because of incompatibility with polar mobile
pound and are between 0.1 and 10 g mL\1 for the phases.
neutral nitroaromatics and nitramines and around
10}100 ng mL\1 for the nitrophenols. Reproducibil-
ity is acceptable (about 2}5%).
Conclusions
To determine the thermally unstable explosive-
related compounds, the method of choice is HPLC.
Nuclear Magnetic Resonance Spectroscopy
However, because of the limited separation perfor-
Proton nuclear magnetic resonance spectroscopy (1H- mance, where there are very complex samples, HPLC
NMR) is well suited for the determination of explos- determination is only possible after optimization of
ives because these analytes are small aromatic the method and pre-separation of the samples into
molecules which offer good manageable spectra; their different fractions.
aromatic protons appear over a relatively wide fre- RP-18 phases have proved their value as standard
quency range (approximately 3 ppm) because of the columns. In special separation problems the use of
various substituents (nitro, carboxyl, methyl, hy- a second column is useful.
Figure 13 TIC and several mass spectra of an extract of drainage water from contaminated soil. PE-SCIEX API 100 LC/MS-System;
Heated NebulizerTM (PE), Ultrasep ES RP 18 column, 5 m, 250;4 mm (Sepsev, Berlin); 41% methanol}59% water (v/v) adjusted to
pH 5.0 by means of acetic acid, 203C.
In most cases, UV detection, which is almost uni- NMR coupling, which has been recently commercial-
versal for explosive-related compounds, is used. Be- ly introduced, are of great advantage.
cause of the complexity of samples, to clarify signal In general, the available techniques (includ-
overlapping, and last but not least for identiRcation, ing sample preparation) enable explosives, their
the use of selective detectors such as the electrochemi- by-products and metabolites to be determined down
cal and mass spectrometric detector as well as HPLC- to a range of 0.1 g L\1.
2782 III / EXPLOSIVES / Thin-Layer (Planar) Chromatography
See also: II/Chromatography: Liquid: Detectors: Mass Hubball J (1992) The use of chromatography in forensic
Spectrometry; Detectors: Ultraviolet and Visible Detection; science. Advances in Chromatography 32: 154}172.
Nuclear Magnetic Resonance Detectors. Extraction: Jenkins TF, Leggett DC, Grant CL and Bauer CF (1986)
Solid-Phase Extraction; Solvent Based Separation; Ultra- Reverse-phase high-performance liquid chromato-
sound Extractions. III/Solid Phase Extraction with Car- graphic determination of nitroorganics in munitions
tridges. wastewater. Analytical Chemistry 58: 170.
Levsen K, Preiss A and Feltes J (1995) Analysenmethoden
Further Reading fuK r Explosivstoffe. In: Rippen G (ed.) Handbuch der
Umweltchemikalien vol. 3, 31st edn, pp. 36}75. Land-
Berberich DW, Yost RA and Fetterolf DD (1988) Analysis sberg: Ecomod.
of explosives by liquid chromatography/thermospray/ Lewin U, Wennrich L, Efer J and Engewald W (1997)
mass spectrometry. Journal of Forensic Science 33: 946. Determination of highly polar compounds in water sam-
Bratin K, Kissinger PT, Briner RC and Bruntlett CS (1981) ples around former ammunition plants. Chromato-
Determination of nitro aromatic, nitramine and nitrate graphia 45: 91.
ester explosive compounds in explosive mixtures and Lloyd JBF (1992) HPLC of explosive materials. Advances
gunshot residue by liquid chromatography and reductive in Chromatography 32: 173}261.
electrochemical detection. Analytical Chimica Acta 130: US Environmental Protection Agency (1992) Method 8330,
295. revision 1. Washington, DC: EPA.
Fedoroff BT (ed.) (1960}83) Encyclopedia of Explos- Yinon J and Zitrin S (1993) Modern Methods and App-
ives and Related Items, vols 1}10. Dover: Picatinny lications in Analysis of Explosives. New York:
Arsenal. Wiley.

J. Bladek, Military University of Technology, Warsaw, during: (i) testing of environmental pollutants, (ii)
Poland forensic investigations, and (iii) checking technolo-
Copyright ^ 2000 Academic Press gical processes and service conditions of munitions
manufacture. From an analytical point of view, at
least three groups of TLC applications in explosive
Introduction
investigations can be distinguished. The Rrst of them,
The term explosive has two meanings. It is used for most often represented in the research literature, con-
individual chemical compounds among which cerns qualitative (including screening methods) and
trinitrotoluene (TNT), 1,3,4-trinitro-1,3,5-triazacyc- quantitative analysis of explosives. The second is
lohexane (hexogen-RDX), 1,3,5,7-tetranitro-1,3,5,7- where TLC is applied as a clean-up technique. In this
tetrazacyclooctane (octagen-HMX), pentaerythriol case analysis is completed by other analytical tech-
tetranitrate (pentrik-PETN) nitroglycerine (NG) and niques. The last group of applications mainly covers
nitrocellulose (NC) are commonly known. This term the evaluation of LE stability.
is also used for mixtures of the above individual Although, recently, more sophisticated methods
compounds and their mixtures with other, non-ex- such as thermal analyses and gas or liquid chromatog-
plosive substances; the type and amount of compo- raphy are commonly used in many laboratories, TLC
nents in a mixture determines its properties (brisance, is still in use. Starting from paper chromatography
melting, plasticity and the like). Explosives have been and improved over many years, TLC has become very
classiRed in many ways according to different cri- effective in the analysis of explosives. Apart from
teria. Most important is the kind of addition, the NO2 high performance adsorbents, and the ability to per-
group and type, and the velocity of the reaction inform separations in both normal and reversed-phase
volved. The Rrst are divided into the following systems, the real advances are due to the development
groups; nitro compounds containing the C-NO2 of densitometry and the spray-on technique of samp-
group, nitrate esters with C-O-NO2 group and nit- ling.
ramines with C-N-NO2 group. The second criterion
divides explosives into high and low explosives (HEs
Analyses of Explosives
and LEs respectively).
The identiRcation and quantiRcation of the HEs or Early TLC analysis was performed on home-
LEs are very valid and also present difRcult analytical made chromatographic plates and involved the
problems. These problems become evident especially separation of classical HEs such as TNT, RDX,
III / EXPLOSIVES / Thin-Layer (Planar) Chromatography 2783
Table 1 Examples of TLC separation of explosives
Analyes Chromatographic system Type of elution
Stationary phase Moblie phase
TNT, RDX, PETN, tetryl, NC, NG Silica gel TrichloroethyleneIacetone (4 : 1) One-dimensional, isocratic
TNT, RDX, HMX, PETN, tetryl Silica gel HPTLC F254 Carbon tetrachlorideIaceto- One-dimensional, isocratic
nitrile (8 : 1)
HMX, RDX, 2,4-DNT, TNT, NG, Silica gel HPTLC F254 I petrol ether } acetone (3 : 1) Two-dimensional, isocratic
PETN, tetryl II petrol ether } ethyl acetate
(8 : 1)
Trinitrobenzene, 2,4-DNT, TNT tetryl, Silica gel HPTLC F254 Benzene - petrol etherI One-dimensional, isocratic
NG, PETN, RDX methanol (8 : 6 : 1)
TNT m-dinitrobenzene, trinitro- Silica gel G-magnesium Xylene One-dimensional, isocratic
benzene, tetryl, 2,4-dinitrochloro- silicate (1 : 1)
benzene, picramide, heksyl
2,3,4-TNT, 2,4,6-TNT: Silica gel HF254 Benzene}hexane}pentane One-dimensional, isocratic
1,2-DNB: RDX (5 : 4 : 1)
RDX, HMX, tetryl, PETN, TNT, Silica gel with chemical AMD, gradient elution
2,4-DNT, 2,6-DNT bonded octadecyl,
RP-C-18
Hexyl, picric acid, RDX, HMX, Silica gel HPTLC AMD, gradient elution
2,4,6-N-tetranitro-N-methyl-
aniline, 2,4- and 2,6-dinitro-
toluenes, TNT, 1,3-dinitrobenzene,
2-amino-4,6-dinitrotoluene, 4-amino-
2,6-dinitrotoluene
HMX and PETN (see Yinon and Zitrin in Further RF and colour reaction. The majority of more or less
Reading). speciRc explosive visualization schemes have been
One-dimensional ascending or horizontal tech- worked out long ago (Table 2). Exhaustive reviews of
niques have usually been applied for the development those methods can be found in the publication by
of chromatograms in a closed chamber (Table 1). Yinon and Zitrin, mentioned above, and in York
Multiple and two-dimensional development tech- et al. (see Further Reading).
niques have seldom been used but automated mul- Most investigations on the analysis of explosives by
tiple development (AMD) and gradient development TLC have been conRned to environmental protection.
are becoming more frequently used. In such research, classical TLC has been applied
In TLC much attention is paid to the methods of mainly as a screening method and has served to pro-
analyte visualization, especially to search for speciRc vide essentially qualitative analysis; instrumental
reactions which can conRrm the presence of an TLC provides both qualitative (screening) and quant-
analyte in a spot or band. ConRrmation of the ident- itative determination. Environmental analysis of ex-
ity of an analyte is thus based on at least two criteria, plosives is justiRed by the fact that nitrotoluenes and
Table 2 Examples of visualization of explosives
Visualization method Analyte Quantification
Yes No
UV absorption 2,3,4-TNT, 2,4,6-TNT, 1,2-DNB, RDX x

NaOH in methanol acetone (1 : 1) Nitroaromatic compounds x
complex formation with o-toluidine, N,N-dimethyl- m-DNB, TNB, TNT, tetryl, 2.4-dinitrochloro x
aniline, m-chloroaniline or anthracene benzene, picramide, heksyl
Griess reagent RDX, HMX x
UV followed by 5% DPA in methanol acidified RDX and nitrates x
by H2SO4
5% DPA in methanol Nitroaromatic compounds x
NaOH/ Griess reagent NG, NC x
0.2% WuK rster’s salt in methanol NC, NG, PETN x
especially trinitrotoluenes are highly toxic com- these are not speciRc reactions and require other
pounds. The aromatic amines formed by their biode- techniques for conRrmation. The work describes the
gradation are suspected to be carcinogenic. Due to separation techniques for nitroaromatics, nit-
careless handling during the manufacture, loading, roamines and nitrate esters, classical, colour visuali-
and packing etc., of explosives, groundwater, surface zation methods and also the costs of analysis. Labor-
water and soil may be contaminated with these com- atory analytical methods for the most commonly
pounds. Occasional plant accidents, and residues found components of explosives, and environmental
from World War II are also sources of pollution. transformation of these substances in a soil matrix,
Screening methods, which most often use have been developed in the work of Jenkins et al.
colorimetric visualization reactions, serve to lower More recently gradient automated multiple devel-
the cost and are less time-consuming for the analysis opment (AMD) HPTLC using a normal phase system
of a large number of samples. An excellent example has been applied to the determination of explosives
which illustrates the advantages of TLC for such and their biodegradation products in contaminated
applications is provided by Haas and Stork. The most soil and water. The subjects of examination were
interesting information in this work lies in the hexyl, picric acid, RDX, HMX, 2,4,6-N-tetranitro-
method of sample preparation (liquid}liquid extrac- N-methylaniline, 2,4- and 2,6-dinitrotoluenes, TNT,
tion) and the technique of separation of large num- and other by-products such as 1,3-dinitrobenzene,
bers of analytes. Water samples taken from waste- 2-amino-4,6-dinitrotoluene, and 4-amino-2,6-dinit-
heaps, containing post-production residues of TNT, rotoluene found in former ammunition site waste in
were extracted with isopropyl ether under a range of Germany. The use of optimized gradient elution and
pHs. At pH 4 aromatic amines were retained in the an AMD system allowed the separation of analytes
aqueous phase and phenols and nitro compounds from environmental samples and quantiRcation at the
were extracted into the organic phase (ether extract 5}20 ng level; humic substances presented in water
A). To enable extraction of amines into the organic samples do not interfere. In the case of samples of soil
phase, the pH was adjusted to 8 (ether extract B). with a high concentration of humic substances,
Phenols were isolated from extract A by treating with a clean-up technique is necessary. The superiority of
1M NaOH (nitrotoluenes remained in the extract). TLC is displayed by the fact that similar analyses
The aqueous phase, of extract A was then re-adjusted using GC or HPLC require (even for water samples)
to pH 4 and re-extracted to transfer phenols to the very careful sample clean-up or the application of size
organic phase (extract C). After drying and concen- exclusion chromatography.
tration, the three extracts are analysed by normal Modern TLC equipment has also been applied to
phase chromatography using two-dimensional iso- the quantitative determination of HE residues in soil
cratic elution. Analytes are identiRed by the quench- and water using a reversed-phase system with auto-
ing of Suorescence or by colour reactions. In extract mated multiple development gradient elution after
A (nitrotoluenes) 25 compounds were found and 2,4-, liquid}liquid and solid-phase extraction. QuantiRca-
2,6-dinitrotoluene and TNT were identiRed. In ex- tion was carried out by UV or visible absorption light
tract B (nitroamines) 26 compounds were found. In measurement in situ. The work demonstrated that the
extract C, containing phenols, 28 compounds were application of AMD was effective for the isola-
found and 2,4-dinitro-6-methylophenol and 2,6- tion and separation of relative complex HE mixtures
dinitro-4-methylphenol were identiRed. ConRrma- in one analytical process (Figure 1). Most of the
tion of TLC results and quantiRcation were per- examined HEs absorb UV light, which allows their
formed using spectrophotometry and gas chromato- quantiRcation by densitometry (Table 3).
graphy. During the measurement of PETN, Wurster’s salts
On-site TLC screening methods for explosives in were applied (organic nitrate esters oxidize Wurster’s
soil have been reviewed by Nam who suggested that salts yielding the so-called ‘Wurster’s cations’ which
the results obtained using two colorimetric-based are intensely coloured). It has been proved that this
methods (TNT and RDX-methods), commonly used reaction can also be used for quantitative analysis.
for on-site analysis of explosives in the USA, are not The extraction of HEs from water samples, using
perfect. The Rrst of the methods, is based on the solid-phase extraction (SPE) cartridges Rlled with
reaction of an acetone soil extract of TNT with base, SDB-1 phase (recoveries +90%), was shown
to produce reddish coloured Jankowsky ions. For the to be more effective than liquid}liquid extrac-
RDX method, the soil extract is Rrst acidiRed and tion. Slightly lower values of recoveries during extrac-
reacted with zinc to reduce the RDX to nitrous acid; tion from soil (77.0}84.4%) were obtained but these
the solution obtained is reacted with Griess reagent to result from the more complex procedure for their
produce a reddish coloured azo dye. Unfortunately, preparation for analysis (necessity of puriRcation).
Figure 1 Chromatogram of high explosives: 1"octagen, 2"hexogen, 3"tetryl, 4"pentrit, 5"trotyl, 6"2,4-DNT, 7"2,6-
DNT: (A) scanning at wavelength 220 nm, (B) pentrit, scanning at 600 nm. Abscissa } absorbance, ordinate } range of the spots in mm.
(Reproduced with permission from BBa9 dek J et al. 1998.)
The proposed technique for the elimination of the Post-blast analyses, or identiRcation of explosive
majority of non-polar impurities from soil extracts residues in forensic investigations, are directly asso-
using SPE is very effective for the identiRcation ciated with combat, and criminal or terrorist activity.
and quantiRcation of the analytes examined Results of such analyses may give information about
(Figure 2). the type, and sometimes also about the source of
Table 3 Parameters of quantification
High explosive max, nm Calibration curve Detection limits,

ng
A*.10\3" Correlation coefficient Max, range of linearity, ng
Octagen 215 8.8c!0.1 0.9861 1600 200

Hexogen 220 10.7c#0.6 0.9858 1600 200
Tetryl 220 28.5c#3.1 0.9989 1000 50
Pentrit 600 30.9c#0.3 0.9899 1600 200
TNT 220 12.5c#50.5c!0.6 0.9989 - 50
2.4-DNT 240 29.7c!1.0 0.9378 1000 100
2.6-DNT 220 10.7c!0.7 0.9728 2000 200
*A, densitometric peak area; c, amount of the pesticide in the band.

explosives or about possible contact with explosives.

From an analytical point of view, these methods are
similar to investigations of explosives in environ-
mental matrices; the difference concerns isola-
tion from the matrix (debris, bodies and surfaces).
These matrices are usually extracted in two ways: by
water and by organic solvent. The water extract is
then analysed for inorganic ions and the organic ex-
tract for the explosives. Yinon and Zitrin have de-
scribed the complexity of such analysis (see Further
Reading).
TLC as a Clean-up Method

Forensic and environmental analyses are usually per-
formed in several stages, one of the most important
being sample preparation. This stage covers the isola-
tion, concentration, and puriRcation of analytes for
further analysis. Three basic methods of puriRcation
of analytes may be distinguished: liquid}liquid ex-
traction, solid-phase extraction and preparative thin
layer chromatography. For preparative TLC, the
analytes are spotted as a long band on the start line.
After developing chromatograms the plates are dried,
and cut, if plastic or aluminium, or scored, if glass, so
that they contain the standard lanes and part of the
Figure 2 Chromatograms of soil sample: (A) before purifica- sample. Classic, preparative TLC has been rarely used
tion, (B) after purification by SPE. Notation as in Figure 1. (Repro- for the puriRcation of explosives. Easy access and
duced with permission from BBa9 dek J et al. 1998.)
a variety of stationary phases for SPE make it more
useful than preparative TLC in this application.
Table 4 Examples of TLC separation and visualization of stabilizers
Stabilizer Chromatographic system Type of elution Visualization
Stationary phase Mobile phase
DPA and its derivatives Silica gel HPTLC 1 step: petrol etherIbenzene (1 : 1) One-dimensional, UV or VIS absorption
2 step: benzene two-steps, isocratic
1 dimension: benzene-1,2-
dichloroethaneI carbon
tetrachloride (10 : 5 : 6)
DPA and its derivatives Silica gel 2 dimension: ethyl acetateIpetrol Two-dimensional, UV or VIS absorption
ether (2 : 8) isocratic
EC and its derivatives Silica gel 1 dimension: benzene Two-dimensional, UV or VIS absorption
2 dimension: benzeneIdiisopropyl isocratic
ether (3 : 1)
DPA and its derivatives Silica gel with zinc 1 dimension: petrol Two-dimensional, 0.25% ethanol solution of
powder etherIbenzeneI acetone isocratic p-diethylaminebenz-
(99 : 99 : 2) aldehyde in 0.25 M HCI
2 dimension: petrol otherIethyl
acetate (4 : 1)
EC and its derivatives Silica gel with zinc 1 dimension: 1,2-dichloroethane Two-dimensional, EC: 0.003%
powder 2 dimension: petrol etherIethyl isocratic dichlorofluorosceine in
acetate (4 : 1) ethanol followed by UV;
derivatives: 0.25%
ethanol solution of p-
diethylaminebenz-
aldehyde in 0.25 M HCI
TLC in Research lizers others contribute to stabilizer depletion. Exam-

ples of the application of TLC to stabilizer measure-
This work mainly concerns LEs (propellants). Nitrate ment are presented in Table 4.
ester-based propellant compositions decompose un- One example of the application of TLC for propel-
der normal storage conditions producing nitrogen lant stability research uses three chromatographic
dioxide as a primary decomposition product. This systems for the isolation of stabilizers (Figure 3)
NO2 is the source of further, autocatalytic decompo- from single-base, free emulsion double base and
sition of propellants, limiting their safe storage life. In emulsion propellants (Table 5). Samples are
order to reduce the rate of deterioration of propel- divided into 2 parts. One is heated for 1 h at 1203C
lants, stabilizers (usually diphenylamine } DPA) or in an open vessel and the other left unheated. All
centralite I-(N,N-diethyl-N,N-diphenylurea) }EC, samples are then dissolved in acetone (this procedure
which react with the primary decomposition prod- is more effective then the extraction of stabi-
ucts, are usually incorporated, to inhibit degradation. lizers) and analysed to observe the differences
The amount of stabilizers added varies, but is typi- between the state of chemical change of the stabilizer
cally less than 1% (in the case of DPA) and 3% (in the and its reaction products before and after ageing.
case of centralite). The stabilizers react with nitrogen Analyses are performed using a liquid}crystal
oxides forming different degradation products detector, and conRrmation was done using
and whilst some of these products also act as stabi- densitometry.
Figure 3 Fragments of propellant chromatograms: (A) single-base, (B) double base, (C) emulsion propellant. Abscissa } absorb-
ance, ordinate } range of the spots in mm. Peaks notation: 1"4-nitrodiphenylamine, 2"N-NO-diphenylamine, 3"2-nitro-
diphenylamine, 4"diphenylamine, 5"4-nitro centralite, 6"centralite, 7"2, 4-dinitrotoluene, 8"dibutylphthalate, (c) 9"central-
ite, 10"2,4-dinitrotoluene, 11"diphenylamine. (Reproduced with permission from BBa9 dek J et al. 1993.)
Table 5 Characteristic compositions of propellant samples
Propellant Concentration [%]
NC EC NG DPA DNT DBP Others
Single-base 96I99 0.7I2.0 0.0I1.8

Emulsion-free double- base 56I66 20I45 0.7I3.0 7I12 0.0I5.0 0.0I1.5
Emulsion 56I70 10I20 0.5I1.0 3I4 10I15 0.0I1.5
In three lots of single-base propellants, stored for ives, can be observed in the 1990s and is connected
48, 47 and 41 years consumption of DPA in the range with the intensive development of the method.
75}100% was determined (before heating). The main TLC, like all other chromatographic techniques, is
products of chemical change are N-NO-DPA, 4-NO2- a comparative method. It is impossible to obtain
DPA, 2-NO2-DPA and also traces of dinitro-DPA. information about the structure or identiRcation of
The most often observed differences in the con- unknown compounds or mixtures. In such cases there
centrations of DPA before and after heating were in exits the possibility of combining TLC with other,
the range 0.2}0.6% (m/m). In most stable double- mainly spectroscopic, analytical techniques (e.g.
base propellants during storage of up to 50 years, TLC-MS). Such combinations will probably start
a small decrease of 5}10% EC content was observed. widening the application of TLC in the analysis of
Rarely, traces of 4-NO2-EC were found. This work explosives.
also found that in some unstable double-base propel-
lants (stored for 40 years) and mortar propellants See also: II/Chromatography: Thin-layer (Planar):
(stored for 38 and 23 years), 90}100% of the EC had Densitomery and Image Analysis; Instrumentation;
Layers; Mass Spectrometry; Modes of Development:
changed into its reaction products mostly into mono-
Conventional; Preparative Thin-layer (Planar) Chromatog-
nitro-EC. Studying chemical changes of EC stable raphy; Spray Reagents; Extraction: Solvent Based
propellants before and after heating, there were deter- Separation; Solid-Phase Extraction. III/Explosives: Gas
mined differences in EC concentrations in the Chromatography; Liquid Chromatography. Humic Sub-
range 0.2}0.8% (m/m). As in single-base propellants, stances: Liquid Chromatography.
the dynamics of stabilizer changes in double-base
propellants, both under natural and accelerated age- Further Reading
ing varies up to ten times under the same conditions
of ageing. It means, that the quality, especially of B"a9 dek J, Miszczak M and SD liwakowski M (1993) Applica-
basic components of the propellants, has greater in- tion of liquid}crystalline detectors to the quantitation of
Suence on their decomposition than the ‘age’. The use propellant’s stabilizers by thin layer chromatography.
of TLC methods to determine the stabilizers and their Chemical Analysis, 38: 339.
B"a9 dek J, PaplinH ski A, Neffe S, and Rostkowski
reaction products in the propellants seems to be suit-
A (1998) Application of TLC for Determination of High
able and reliable. Explosive Residues in Water and Soil Samples. Chemical
Analysis, 43:711.
Conclusions Crockett AB, Jenkins TF, Craig HD and Sisk WE (1998)
Overview of On-Site Analytical Methods for Explosives
There are a large set of analytical needs connected in Soil. US Army Corps of Engineers. Cold Regions
with the analysis of explosives. The importance of Research and Engineering Laboratory. Special Report
this problem can be appreciated by looking at the 98: 4.
statistics on recent bombings in the world. Besides the Hass R and Stork G (1989) Konzept zur Untersuchung von
tasks mentioned above, there is a need for vehicles or RuK stungsaltlasten. 1. Untersuchung chemaliger TNT-
mail screening, bomb search, protection of special Fabriken und FuK llstellen. Fresenius Z. Analytical Chem-
istry 335: 839.
infrastructure features, and so on. Of course, not all
Kirchner JG (1978) Thin Layer Chromatography. Chiches-
analysis can be performed using TLC. ter: John Wiley & Sons.
In the history of TLC for analysis of explosives the Nam SI (1997) On-Site Analysis of Explosives in Soil:
1960s and 1970s was a time of very intensive use. The Evaluation of Thin Layer Chromatography for Con-
development of modern analytical methods has Trmation of Analyte Identity. US Army Corps of Engin-
caused a decrease in interest in TLC. An increase of eers. Cold Regions Research and Engineering Laborat-
interest in TLC, as applied to the analysis of explos- ory. Special Report 97: 21.
III / FATS / Crystallization 2789
Steuckart C. Berger-Preiss E and Levsen K (1994) Deter- York H, Funk W, Fischer W and Wimmer H (1994)
mination of explosives and their biodegradation prod- Thin Layer Chromatography: Reagents and Detec-
ucts in contaminated soil and water from former ammu- tion Methods } Physical and Chemical Detection
nition plants by automated multiple development high- Methods: Fundamentals. Reagent 1. Volume la.
performance thin-layer chromatography. Analytical Weinheim: VCH.
Chemistry 66: 2570. York H, Funk W, Fischer W and Wimmer H (1994) Thin
Yinon J and Zitrin S (1981). The Analysis of Explosives. Layer Chromatography: Reagents and Detection
Oxford: Pergamon Press. Methods } Physical and Chemical Detection Methods:
Yinon J and Zitrin S (1993). Modern Methods and Applica- Activation Reactions, Reagent Sequences, Reagents II.
tions in Analysis of Explosives. Chichester: John Wiley Volume 1b. Weinheim: VCH.
& Sons.
EXTRACTION: PRESSURIZED FLUID EXTRACTION

See III / ENVIRONMENTAL APPLICATIONS: Pressurized Fluid Extraction;
PRESSURIZED FLUID EXTRACTION: NON-ENVIRONMENTAL APPLICATIONS
FATS
vegetable oils compared to conventional materials,

Crystallization etc. This review highlights the basic information on
recent research on the crystallization of fats, with
a speciRc emphasis on the separation phenomena.
K. Sato, Hiroshima University, Higashi-Hiroshima, The speciRc characteristics of the crystallization of
Japan
fats are, on the one hand, polymorphism, and mo-
Copyright ^ 2000 Academic Press lecular interactions on the other. No long-chain com-
pound without polymorphic modiRcations is present,
Introduction and this property is more enhanced in triacylglycerol
(TAG) crystals. The molecular interactions in TAG
Fats, as mostly represented by triacylglycerols, are crystals are operative mostly through van der Waals
employed in foods, cosmetics, pharmaceuticals, etc. forces between hydrophobic aliphatic chains, in
as the main bodies of end products, or as the matrices which geometrical and steric hindrance is critical be-
in which Rne chemicals are dispersed. The crystalliza- tween the glycerol groups.
tion behaviour of fats has two major industrial im-
plications: (a) processing of the end products made of
fat crystals, such as chocolate, margarine and
Polymorphism in Fats
shortening, whipping cream, etc. and (b) separation Polymorphism is deRned as the ability for a chemical
of speciRc fats and lipids materials from natural re- compound to form different crystalline or mesophase
sources, mostly from vegetable or animal fats and (liquid crystalline) structures. The melting and crys-
oils, which contain various molecular species having tallization behaviour differs from one polymorph to
different chemical and physical properties. As for the others, since the different crystal structures corres-
separation technology of crude fat resources, such as pond to different Gibbs energies.
palm oil, milk fats, hydrogenated vegetable oils, etc. Polymorphic crystallization is primarily deter-
it may be worth noting that current market demands mined by the rate of nucleation, as determined by
raise a great necessity to develop the fractionation of thermodynamic and kinetic factors. If the Rrst crystal-
high-melting and low-melting fats and lipids through lizing form is less stable, it converts to more stable
dry fractionation. The main causes for this are the polymorphs through the solid matrix or liquid medi-
replacement of hydrogenation by dry fractionation, ation. As a consequence, morphology of fat crystals
a new regulation of usage of fat materials for confec- is a function of the crystal structure itself, and the
tionery fats, better functionality of physically reRned thermal processes of the crystallization and sub-
2790 III / FATS / Crystallization
Polymorphism and Macroscopic Physical

Properties
The microscopic properties of the fat polymorphs
eventually result in complex macroscopic physical
properties of mixing, melting/crystallization and the
structure/morphology of the crystallized materials. In
particular, the rate of crystallization and the crystal
morphology are directly inSuenced by polymorphism.
Mixing of fats The mixing properties of different

TAG materials, is highly relevant in fat materials,
because many fats present in real systems are multi-
component in two ways: (a) a fat phase containing
many different types of TAGs and (b) each TAG
molecule involving different types of fatty acid moie-
ties, namely mixed-acid TAGs. Three typical phases
occur in binary solid mixtures, when the two compo-
nents are miscible in all proportions in a liquid state:
solid solution phase, eutectic phase, and compound
formation. The general tendency in the relationship
between the molecular interactions and the phase
Figure 1 Relationship of molecular properties of fat crystals behaviour may be summarized as follows: structural
and macroscopic physical properties of fats.
afRnity results in solid solutions, poor interactions
form eutectic phases and speciRc interactions give rise
sequent transformation. The interrelationships to molecular compounds.
among fat molecule structures, polymorphic crystal- In addition, polymorphism makes the mixing be-
line structures, phase behaviour of the mixing of haviour more complicated. For example, in the mix-
different fats and other ingredients, melting and sol- ture of mono-saturated acid TAGs, a eutectic phase
idiRcation, and transformation and crystal morpho- with a limited region of solid solution is the stable
logy are illustrated in Figure 1. form when the chain length difference is no larger
Molecule Structures and Polymorphism of Fats than two carbon atoms. However, metastable forms
of and exhibit solid solution phases. Furthermore,
This section brieSy describes the molecular structural the chain length structure also affects the mixing
properties of fat polymorphs. behaviour, even if the polymorphic properties are
Typically, three polymorphs, , and , are usu- similar between the component TAGs: for example,
ally observed in TAG crystals. However, two or even the double chain length and triple chain length struc-
three more forms are present in TAGs if their fatty tures are not miscible with each other.
acid structures are rather complicated, in the sense As for the formation of compound systems,
that the three fatty acid moieties (R1, R2 and R3 compound formation occurs in particular sets of
in Figure 1) differ in their chemical structures (length, TAG mixtures, in such mixtures as SOS/SSO (1,2-
saturated/unsaturated, even-chain or odd-chain, distearoyl-3-oleoylglycerol) and POP/OPO (P"
etc.). The three forms are basically characterized by palmitoyl), where the compound is formed at a con-
the lateral chain packing mode, expressed in subcell centration ratio around 1 : 1. The mixing behaviour
structures: hexagonal (H) packing for , orthorhom- and the crystallization properties in binary fat mix-
bic perpendicular (ON ) packing for , and triclinic tures will be discussed in a later section.
parallel (T ) packing for . Since the lateral packing
mostly determines the crystalline density, it follows Melting behaviour The melting behaviour in rela-
that is the least dense packing, intermediate and tion to fatty acid composition of TAGs is straightfor-
the densest packing. As for the chain length struc- ward: the longer the fatty acid chain, the higher the
ture, the double chain length is revealed in the TAGs melting point, and saturated-acid TAGs have higher
containing similar types of three fatty acid moieties, melting points than unsaturated fatty acid TAGs.
and, by contrast, the triple chain length structures are The next concern is the effects of polymorphism of
revealed in the TAGs where one or two of the fatty fats (which are mostly of a monotropic nature) on
acid moieties are largely different in their shapes from the melting and crystallization behaviour. As an
the others. example, we take three polymorphic forms of
unique for the speciRc polymorph of a speciRc TAG.

The morphology of crystals, however, is a result of
combined interactions of the crystal structure itself,
and rate of crystal growth of the speciRc crystalline
faces which are inSuenced by the growth conditions.
For simplicity, however, it may safely be referred
to the polymorph-dependent crystal morphology;
reveals irregular shaped tiny crystals, bulky-
shaped tiny crystals, and lozenge-shaped needle
crystals.
Polymorphic Crystallization
Polymorphic crystallization is primarily determined
by the rate of nucleation, being governed by thermo-
dynamic and kinetic factors. A primary concern is the
polymorphic nucleation, in which the so-called Ost-
wald step rule is of great interest. This predicts that
a phase change may occur step by step by way of
successively more stable phases. Thus, the metastable
forms nucleate Rrst, prior to the most stable form,
when nucleation is induced under large kinetic fac-
tors, e.g. supercooling or supersaturation. When the
kinetic factors are minimized, the step rule is broken,
and the more stable forms are nucleated at very re-
duced rates.
Figure 2 (A) Structure models and Gibbs energy (G )}temper- This tendency was conRrmed in the crystallization
ature (T ) relationship of three polymorphs of PPP, and (B) rate of of three forms of PPP. For PPP, induction time (), the
nucleation of PPP polymorphs. *, form; 䢇, form; 䊐, form. time until the occurrence of the Rrst-appearing crys-
Tc, crystallization temperature. tals are detectable, is shortest for , intermediate for
and longest for as shown in Figure 2(B) for the
tripalmitoylglycerol (PPP), whose structural and crys- case of simple cooling of the liquid. However, when
tallization properties have been well elucidated, as the temperature of cooling Suctuates around the
illustrated in Figure 2. melting points of or , the nucleation of more stable
Figure 2(A) shows , and forms of PPP. Mo- forms is markedly accelerated, because of melt-me-
lecular structures of the three forms are brieSy speci- diated crystallization.
Red by the following: disordered conformation in , Melt-mediated nucleation is ascribed to the mono-
intermediate packing in and most dense packing in tropic nature of polymorphism, as revealed in the
. Therefore, the Gibbs free energy (G) values are following processes: (a) melting of the less stable
highest in , intermediate in and lowest in , result- form, (b) nucleation and growth of more stable
ing in the lowest melting temperature for , etc. As forms, and (c) mass transfer in the liquid formed by
a result of the combined effects of the molecular melting of the less stable form. It often occurs that the
structures and the rate of crystallization, the crystal rate of melt-mediated transformation is considerably
morphology is of amorphous irregularity for , tiny higher than the rates of crystallization of the stable
bulky shape for and needle shape for . polymorph without passage through metastable
The polymorphic properties of PPP are basically forms. Dynamic X-ray diffraction studied using syn-
common to those in the other fats. However, more chrotron radiation (SR-XRD) has enabled in situ ob-
complexity is revealed when the fatty acid composi- servation of the melt-mediated transformation over
tions of the three moieties become heterogeneous, in a time scale of tens of seconds.
particular for TAGs involving saturated and un- In regard to the SR-XRD studies, deeper insights of
saturated fatty acid moieties as present in many veg- the polymorphic crystallization have been unveiled
etable fats. by the time-resolved analyses of the crystallization
from neat liquid. The result has shown that: (a) the
Structure and morphology The crystal structures as formation of lamella ordering, both for double or
determined by X-ray analysis using single crystals, are triple chain length structures occurs more rapidly
2792 III / FATS / Crystallization
than that of the subcell packing; (b) the rate of the

} melt-mediated transformation is remarkably
high when the lamella structures of the pre-existing
form () remains. So-called ‘melt-memory effect’,
meaning the effect of solid structures present in the
liquid phase, may be clariRed by the time-resolved
SR-XRD under temperature variation.
The control of the polymorphic crystallization may
be affected by the addition of crystal seeding and the
application of shear stresses. As to the former effect,
it was found that speciRc molecular interactions be-
tween the crystal seed materials and the polymorphic
forms of speciRc fats are a prerequisite for the seeding
effect, in polymorphic correspondence, structural
similarity and thermal stability. As a typical demon-
stration, a quite interesting seeding effect has been
observed in cocoa butter crystallization, which has
been applied to solidiRcation control of chocolate
production processes.
Cocoa butter has six polymorphs of form I through
to VI, in which form V is the most desirable. For
optimally obtaining form V in the solidiRcation pro-
cess, a tempering process is applied using the follow-
ing temperature proRle: Rrst cooling the molten choc- Figure 3 Relative crystallization time (tr) of cocoa butter
olate to 26}273C, reheating to around 303C, and with the addition of three crystal seeding materials (SSS,
tristearoylglycerol).
subsequent cooling for bulk solidiRcation. During the
Rrst cooling process, rapid crystallization causes the
formation of unstable polymorphic forms of cocoa thermodynamic stability of the seed crystal, which is
butter crystals due to Ostwald’s Law of stages. The related to its solubility in the molten cocoa butter,
occurrence of the unstable forms are not favourable, and (c) the relationship of the acyl chain length of
because they cause poor moulding and induce fat TAG between the seed material and cocoa butter. As
bloom. Therefore, the Rrst-cooled chocolate is heated to (a), it was shown that the 2 of BOB or SOS and
to around 303C, so that the unstable forms melt and form V show identical structural properties to each
quickly change to form V by the melt-mediated other. However, the melting point of SOS form,
transition. After the tempering, chocolate is poured around 433C, is too low for it to be put into molten
into a mould, and cooled for complete solidiRcation. cocoa butter at 303C, and thereby no effect of crystal
The seed crystallization technique aims at achiev- seeding is operative. As to the form of SSS (tris-
ing two major advantages, compared to the temper- tearoyl-glycerol), its melting point, 703C, is optimal
ing technique: (a) it accelerates the crystallization of as far as the thermodynamic stability is concerned.
form V of cocoa butter, and (b) it does not use the However, the molecular structural afRnity between
cooling/heating process of tempering. For this pur- the of SSS and 2 of BOB is poorest, because of the
pose, a technique using BOB (1,3-dibehenoyl-2-sn- difference in the chain length structure: double chain
oleoylglycerol) 2 polymorph has been developed. length for of SSS and triple chain length for 2 of
Figure 3 shows the relative crystallization time (tr) BOB. Therefore, a Rnal solution of the Rnest seeding
of seeded dark chocolate examined at 303C. The material is the 2 of BOB.
crystallization time was measured with a rotational As to the shear effect, the newest Rnding of the
viscometer and tr is deRned as the ratio of the crystal- shear stress application on cocoa butter crystalliza-
lization times with and without seeding. Figure 3 tion indicates a by-passing effect to crystallize -type
clearly shows that the stable form 1 of SOS, which is polymorph (form V), which otherwise does not cry-
a major component TAG in cocoa butter, is most stallize directly from the quenched liquid. A time-
effective, stable form 2 of BOB is in-between and resolved SR-XRD spectra revealed the transforma-
of SSS is least effective. From this result, one may tion from -type to -type polymorphs during the
argue that there are three factors in the crystal seeding initial stages of crystallization under shear stresses.
effect: (a) similarity in the crystal structure between The molecular mechanism for this conversion is not
the seed crystal and form V of cocoa butter, (b) fully understood.
Phase Behaviour and Crystallization Therefore, the eutectic phase is revealed in the com-
Kinetics in Binary Fat Mixtures ponent and compound materials.
The formation of the molecular compound has two
Fats present in natural sources are mixtures of differ- implications for the fat separation problems: Rrstly,
ent types of TAGs. Therefore, the complicated behav- no easy separation is possible by dry fractionation for
iour of melting, crystallization and transformation, the molecular-compound mixtures, and secondly
morphology, etc. of real fat systems is partly due to polymorphic transformation is affected by the pres-
the physical properties of the component TAGs, and ence of the molecular compounds. More interesting-
partly or more importantly due to the phase behav- ly, the rate of nucleation of the molecular compound
iour of the mixtures. To resolve these complexities in is much higher than those of the component TAGs.
mixed systems, a fundamental study has been made The diverse properties in fat mixing behaviour, as
on binary and ternary mixtures of speciRc TAG com- illustrated above for a typical model substance, might
ponents. Recent studies carried out on the phase be- also be revealed in natural fats having high melting
haviour and crystallization kinetics of the binary mix- fractions, such as palm oil, shea fats, etc. Therefore,
tures of principal TAGs, are given below. for the purpose of the separation of the high-melting
In general, three typical phases may occur in binary and low-melting fractions, it is pertinent to examine
solid mixtures, when the two components are miscible physical analyses of the mixing behaviour of the prin-
in all proportions in a liquid state: a solid solution cipal TAG components, as basic research on the one
phase, a eutectic phase and compound formation. For hand, and to extend the main results to more multi-
TAG mixtures, two factors are concurrently domina- component systems tightly mimicking real systems.
ting: polymorphism and chain}chain interactions.
The polymorphic inSuence is clearly seen for the
PPP}SSS mixture system, where the carbon numbers See also: II/Crystallization: Melt Crystallization; Polymo-
(nc) of the fatty acid moieties of the two TAGs differ rphism; Control of Crystallizers and Dynamic Behaviour.
by two. Solid solutions are formed in the and
forms, yet the eutectic phase was formed in the most Further Reading
stable form. This contrast is ascribed to the less
Hachiya I, Koyano T and Sato K (1989) Seeding effects on
condensed molecular packing in and and to the solidiRcation behavior of cocoa butter and dark choc-
most dense packing for . It is reported, however, that olate. I. Kinetics of solidiRcation. Journal of the Ameri-
the formation of the solid solution in the metastable can Oil Chemists’ Society 66: 1757d1762.
forms is hindered when the difference in nc between Hagemann JW (1988) Thermal behaviour and polymor-
the component TAGs exceeds four, and that eutectic phism of acylglycerols. In Garti N and Sato K (eds)
phases become predominant. Crystallization and Polymorphism of Fats and Fatty
It has recently been found that the phenomena Acids, pp. 9}95. New York: Marcel Dekker.
observed in the PPP}SSS mixture above does not Hui YH (ed.) (1996) Bailey’s Industrial Oil and Fat Prod-
simply apply to the mixtures of saturated}un- ucts, Vol. 3, 5th edn. New York: John Wiley.
saturated mixed-acid TAGs such as POP}OPO, Kellens M, Meeussen W, Riekel C and Reynaers H (1990)
Time resolved X-ray diffraction studies of the polymor-
POP}PPO and many other similar mixtures, where
phic behaviour of tripalmitin using synchrotron radi-
O is oleic acid moiety, and PPO, for example, is ation. Chemistry and Physics of Lipids 52: 79}98.
a stereoisomer of POP where the fatty acids are Padley FD (1997) Chocolate and confectionery fats. In Gun-
connected to different glycerol carbon atoms. Despite stone FD and Padley FB (eds) Lipid Technologies and
the quite diversiRed phase behaviour exhibited in Applications, pp. 391}432. New York: Marcel Dekker.
these combinations, one may make two important Sato K (1986) Polymorphism of pure triacylglycerols and
points: (a) steric hindrance between the saturated acid natural fats. In Padley FD (ed.) Advances in Applied
and oleic acid moieties induces the formation of eu- Lipid Research, pp. 213}268. London: Jai Press.
tectic phases for the , and polymorphs for the Sato K and Kuroda T (1987) Kinetics of melt crystallization
mixture of POP}PPP, and (b) attractive interactions and transformation of tripalmitin polymorphs. Journal
through the saturated and oleic acid moieties cause of the American Oil Chemists’ Society 64: 124}127.
Sato K, Ueno S and Yano J (1999) Molecular interactions
the formation of molecular compound at a 50 : 50
and kinetic properties of fats. Progress in Lipid Research
concentration ratio of the two components. Quite 38: 91}116.
interestingly, the chain length structure of the mo- Ueno S, Minato A, Seto H, Amemiya Y and Sato K (1997)
lecular compound is largely deformed from those of Synchrotron radiation X-ray diffraction study of liquid
the component molecules, e.g. a triple chain length crystal formation and polymorphic crystallization of
structure in POP and PPO, but a double chain length SOS (sn-1,3-distearoyl-2-oleoyl glycerol). The Journal
structure in the molecular compound of POP}PPO. of Physical Chemistry B 101: 6847}6854.
2794 III / FATS / Extraction by Solvent Based Methods
Extraction by Solvent Based Methods

E. J. Birch, University of Otago, Dunedin, New Zealand Solvent extraction of fats and oils from seeds be-
came possible from the middle of the 19th century. In
1948 the Rrst commercial solvent extraction plant
was built. Since then, over 200 extractors have been
Along with proteins and carbohydrates, fats and oils supplied worldwide for capacities up to 6000 tons per
make up the major classes of food components. Ed- day. Solvent processes found favour in the vegetable
ible fat and oil usage falls into four major product oilseed reRning industry, with continuous miscella
categories: cooking and salad oils, shortenings, mar- (solvent plus pressed cake) reRning and continuous
garines and specialty products. Tissues from animals solvent fractionation (e.g. winterization) becoming
liberate oils on being boiled and oils can be pressed commercial reality in the early 1950s.
from fruits, vegetables and seeds, such as olives,
soybean and sesame. Solvent extraction is a viable
alternative to pressing and can recover nearly all the Solvent Extraction Methodology
oil from the seeds. Figures for world production of
Extraction Theory
fats and oils have only been kept since 1942 and,
although growth in the use of fats and oils since then Solvent extraction is usually used to recover a com-
has outstripped population increases, there has been a ponent from either a solid or liquid. The sample is
signiRcant shift from animal to vegetable fats. contacted with a solvent that will dissolve the solutes
Table 1 shows the relative changes in world produc- of interest. Solvent extraction is of major commercial
tion Rgures for selected major fats and oils since importance to the chemical and biochemical indus-
1935. tries, as it is often the most efRcient method of separ-
Traditional methods employing liquid extraction ation of valuable products from complex feedstocks
rely on the use of simple processing equipment and or reaction products. Some extraction techniques in-
low pressure applications. They fall into two main volve partition between two immiscible liquids;
categories of water Sotation and traditional pestle others involve either continuous extractions or batch
and mortar extraction procedures. The water Sota- extraction of solid. Because of environmental con-
tion method involves heating in water followed by cerns, many common liquid}liquid processes have
size reduction (pestle and mortar-type equipment) been modiRed either to utilize benign solvents, or
and skimming off of the oil followed by heating to replaced by processes such as solid-phase extraction.
remove the moisture. Coconut and palm oil extrac- The solvent can be a vapour, supercritical Suid or
tion efRciencies range from 40 to 60%, though free liquid, and the sample can be a gas, liquid or solid.
fatty acid contents are high. The Ghani mill (typically
powered by animals) has been used in India for 3000 Table 1 World production of commercially important edible fats
years and at the beginning of this century approxim- and oils
ately 97% of oilseeds were processed this way, utiliz-
ing over half a million ghanis. This Rgure has dropped Fat/oil source World production (MMT)
to 4% in modern times, owing to the introduction of 1935 1955 1975 1995
screw and hydraulic presses, and solvent extraction.
The hydraulic press was invented by Joseph Bramah Soybean 0.5 2 8.5 16
in 1795 and continued to be used in the American Palm 0.5 1 2 14.5
Rapeseed (canola) 1.3 1.5 2 9.5
oilseed crushing industry until the 1940s. By then,
Butter 4 3.5 5.3 8
new options were replacing the labour-intensive Sunflower 0.4 1.5 3.5 7.5
batch-processing presses, with expellers (continuous Tallow 1.7 3 5.5 7
screw presses) and direct solvent extraction with the Lard 3 3.5 4.2 5.5
two unit operations often occurring together as Cottonseed 1.5 1.8 2.8 4.8
Peanut 1.8 2 3.2 3.9
a two-stage process. Mechanical expression can only
Coconut 1.7 2 2.5 3.3
reduce oilseed oil content to 2}3% whereas solvent Sesame 0.8 1 1.7 1.9
extraction will reduce this Rgure to about 0.5%. Low Fish 0.5 0.5 1.2 1.5
temperatures during solvent extraction can produce Olive 0.6 0.6 0.7 0.8
a higher quality oil than higher temperature screw Total 18.3 23.9 43.1 84.2
pressing but may also extract nontriglyceride mater-
ial, making the oil inferior to pressed oil. MMT, million metric tons.
III / FATS / Extraction by Solvent Based Methods 2795
Fats and oils are hydrophobic and hence insoluble Supercritical Suids have been investigated since
in polar solvents. Use is made of their afRnity for the last century, with the strongest commercial inter-
nonpolar media in oil extraction, separation of oil est initially focusing on the use of supercritical
from bleaching earths and oil-reRning technology. toluene in petroleum and shale oil reRning during
Fatty acids accompanying oil extraction are also sol- the 1970s and latterly supercritical carbon dioxide
uble in polar solvents and use can be made of this in for fats and oils. Carbon dioxide cannot be used
reRning methods to separate the two groups. How- as a simple substitute for organic solvents, however.
ever, difRculty is experienced in that fatty acids and Phosphatides (e.g. lecithin) are selected against
other liquids may be more soluble in the oils themsel- compared with hexane when extracting vegetable
ves than in the selected separation solvent. Solubility and Rsh oils, for example. As a solvent, dense
covers a wide range and is inSuenced by a rise in carbon dioxide tends to be selective for lower molecu-
temperature (which increases solubility) and an in- lar weight lipophilic compounds. This can be utilized
crease in chain length (lower solubility) of fatty acids for partial fractionation of free fatty acids from
making up the triglycerides. triglycerides, decreasing cholesterol levels and in-
creasing -carotene content by selective control of
Choice of Solvent
solvent temperature and pressure to exploit differ-
The ideal solvent for oil extraction must possess sev- ences in solubility, vapour pressure and molecular
eral features that are impossible to Rnd in any one weight.
solvent. These properties include the ability to
General Processes
solubilize the oil at low temperatures, selectivity to-
wards triglycerides, chemical inertness, immiscibility The main disadvantage of solvent extraction is the
with water, nonSammable, low viscosity and surface high equipment cost and plants tend to be large,
tension, nonexplosive, noncaustic, low boiling point, processing hundreds to thousands of tons per day.
nonirritant and nonpoisonous. The extracted miscella (solution of solvent plus oil)
Hexane is the almost exclusively chosen solvent contains Rnes, which need to be separated and well
due to its solvent power, volatility, low and nontoxic washed. Solvent and oil will also be held up in the
residue levels and immiscibility with water. However, solids (marc).
care in handling is required due to high Sammability.
Before hexane, carbon disulRde and trichloroethylene Pretreatment Before oil can be extracted from fruits
were used; however, the former has since been ban- and vegetables, the seeds must be prepared. Seed
ned and the latter declared undesirable for preparing preparation for extraction involves cleaning, dehull-
animal feeds. The greater solvency and nonSamma- ing, cooking to denature proteins, adjusting moisture
bility of trichloromethane renders it of use for extrac- content to the right level and then crushing or rolling
tion of tallow from meat and bone but, due to desol- into particles or Sakes of uniform size and thickness.
ventizing problems, it is not used with oilseeds. Re- For solvent extraction, the optimum moisture level
placement with dichloromethane is an option but not allows Saking of the seeds whilst minimizing crum-
favoured due to the risk of hydrochloric acid forma- bling. Ready penetration of the solvent is enhanced
tion and the use of ethanol, either as a solvent or as an without blocking the extractor so that the miscella
azeotrope with hexane, has been studied but not which forms can be easily separated from the cake.
commercialized. Improving pretreatment of press-cakes allows a re-
Recovery of solvent from the extraction of fats is duction in solvent-to-solids ratios and a reduction in
a major consideration in solvent choice. The accept- solvent hold-up in the desolventization process. The
able recovery standard is 99.92%, which can be French Enhanser Press is an extruder used to pelletize
represented by a loss of 1.135 litres of hexane per oil-bearing seeds or pre-press cake to provide an ideal
processed tonne of soya. Losses may occur from medium for solvent extraction. As pellets discharge
desolventizing the meal, from stripping the Rnal oil from the die plate of the Enhanser Press, they expand
product and from air and water discharges to waste. and Sash off moisture. This creates dense pellets with
In the US the Environmental Protection Agency is a vast matrix of open-structured, internal solvent
charged with controlling emission levels with respect passages. These pellets yield much better extraction
to hexane, although required levels are still to be set results than Sakes or pre-press cake. This results in
under the Federal Clean Air Act. A likely example for savings in distillation energy (steam required).
soybean processing is 0.2}0.25 kg for every 100 kg of
beans processed. To date, the industry relies on self- Extraction The extraction operation typically fol-
regulation, required because of the high cost of lost lows a countercurrent Sow process where the solids
solvent. move in the opposite direction to the solvent}oil
miscella, which meets the oil-rich Sakes at high oil works similarly for drainage of miscella and operates
concentration. The Sakes are sequentially extracted in continuous mode where the Sakes meet solvent in
with solvent of lesser oil content through the different a countercurrent direction in different zones of the
stages until the almost entirely extracted meal is extractor. The percolation extractor employs solvent
Rnally met with pure solvent to complete the extrac- being pumped over and percolating down through
tion efRciently. The Rrst miscella wash leaves the a bed of Sakes or a cake and leaving via a perforated
system for distillation and oil recovery while the Rnal plate at the bottom. Immersion extractors are claimed
extracted Sakes go to a desolventizing process. to allow better extraction from Rne cake particles,
which may block the bed of a percolation extractor.
Batch extractors. The process involves sequential Examples of immersion and percolation extractors
washing of oil-bearing material with progressively are shown in Figures 1 and 2.
leaner miscellas until the Rnal wash is with solvent
alone. The vessels are loaded one at a time and are no Desolventizing The extracted Sake material may
longer used for any other than specialty runs. contain 25}35% residual solvent. In the desolven-
tizer/toaster, steam is used to evaporate the volatile
Total immersion extractors. The material to be pro- solvent countercurrently. The vapour phase is con-
cessed travels through a pool of solvent. These are densed and collected. The meal is toasted in steam-
early designs (1930}1950s) and suffered from excess jacketed trays, then dried and cooled. For edible
Rnes carried along in the solvent. Sour, the process may be replaced with a Sash desol-
ventizing system.
Percolation extractors. These may be either batch Solvent stripping of the extracted oil is carried out
or continuous and differ from the immersion extrac- by evaporation. Removal of Savour components is
tor in that the solvent passes through the solids, likely at this stage also. The miscella is normally
dissolving out the oil. Five main types are in use: separated into oil and vapour through a series of
basket, rotary, perforated belt, sliding-bed and rec- falling Rlm evaporators and stills with the miscella
tangular loop extractors. on the tube side and the vapours on the shell side.
The Rrst-effect evaporator uses steam and solvent
Combined plants. Percolation and immersion ex- vapour from the desolventizer/toaster to concentrate
traction may be combined sequentially to advantage the miscella to about 80% oil content. The second
(e.g. C.M.B. Percolimm). Flaked seed, which has been stage operates atmospherically or under vacuum to
percolation-extracted, is immersion-extracted and bring the concentration up to 95}98% oil, and the
then desolventized. The immersion miscella is used as remaining volatiles are stripped in the still operating
a solvent in the percolation extractor for the original under vacuum. ReRnements to the process are used
seed Sakes. The extractors are of two main types: depending on the oil being extracted. For cottonseed
deep- or shallow-bed. The deep-bed-type (or rotary oil, caustic may be added during the Rrst- and second-
extractor) is semi-continuous with a number of bask- effect evaporators and the mix centrifuged to remove
ets supported on a drainage screen, designed to allow colour bodies before they set during distillation.
the miscella to pass. The screens can be rotating or Recovery of the solvent from immersion extractors
Rxed, as can the baskets and washing manifolds. The traditionally took place in a series of interconnected
extractor moves slowly and miscella drains through cylindrical vessels where the marc progressed by
the screen until the basket reaches the Rnal position dropping from one vessel to the next in a zigzag
when the solids are released, the screen reclosed and fashion. These early models were often named
a fresh load of solids deposited. The shallow-bed-type schneckens (winding staircase) desolventizers. With
Figure 1 Oil solvent extraction apparatus (immersion type). Courtesy of Europa Crown Ltd., Hessle, UK.
Figure 2 Oil solvent extraction apparatus (percolation type). Courtesy of Europa Crown Ltd., Hessle, UK.
the advent of percolation extractors, the desolven- passed to a second vessel and the temperature
tizer/toaster appeared in the 1950s. This consists of reduced to 103C. Filtration using rotary drums
an upright cylinder divided into horizontal sections or separates out the Rrst stearin fraction and the oil
trays. Material is agitated via a sweep arm which also plus solvent is carried through a further series of
opens a door, allowing product to fall to the tray cooling steps at 73C, 43C and 23C respectively. The
below. Heat is supplied to vaporize the solvent, which second stearin fraction is then removed through drum
is Sashed off the topmost trays with sparge steam, Rlters. All three fractions are freed from solvent by
which also partially cooks (toasts) the solids. The distillation.
sparge steam can be absorbed by the solids on con- Beef tallow and hydrogenated vegetable oils can
densing, which prevents carry-over of dust to the also be similarly doubly fractionated. The Rrst stearin
condenser. A variation to the basic models allows fraction can be used for shortenings, the second for
Sash desolventizing, where recirculating steam under confectionery butters and the third (or olein fraction)
high velocity strips the solvent at low temperature to as a frying oil which is liquid at room temperature.
avoid denaturing the protein. Another variation is Without solvent, the conventional dry fractionation
countercurrent desolventizer/toasters, where the yields a stearin fraction which contains too much
steam is introduced at the bottom and travels through entrained oil to be of use as a confectioner’s hard
to the top to a steam-jacketed desolventizing tray, butter.
where indirect steam Sashes off surface solvent. Flash Winterization is beneRcial for rice bran oil where
desolventizers are used to prepare high nitrogen soy the oil is frequently cloudy at room temperature.
protein for meat analogues and the desolventizer/ Solvent winterization separates the high and low
toasters recover solids as animal feed. In the solvent melting triglycerides. Solvents used include hexane,
extraction industry, the term DTDC stands for desol- acetone and isopropyl acetate. Although fractional
ventizing}toasting}drying}cooling. These four opera- crystallization from miscella has had limited accept-
tions can be performed in a single DTDC machine, ance, it is potentially less labour-intensive than the
or split between two separate DT and DC machines. conventional batch winterizing process with faster
Both types of conRgurations have speciRc advant- throughput and yields greater quantities of winterized
ages which may make one or the other preferable oil. In addition, manufacturers could produce modi-
in certain applications. Figure 3 shows a DTDC Red fats without the formation of trans isomers by
system. fractionally crystallizing mixtures of vegetable and
animal oils from miscella. Other options involve ad-
Solvent reVning of fats and oils Fractionation using aptation of the technology for continuous miscella
solvents involves forming a miscella of oil in the hydrogenation and miscella bleaching employing sil-
solvent, which is then cooled to produce crystalliza- ica gel to remove colour.
tion. Filtration of the stearin fraction follows and Solvent extraction of oils from used bleaching
both fractions are then heated to recover the solvent. earths is a logical method of recovery but only for
The Bernardini process uses hexane as solvent and large volumes to make it economical. Enclosed Rlters
produces two stearin and one olein fractions. Acetone allow hexane to extract the oil from the earth in
and 2-nitropropane have also been utilized as sol- several stages before recovery of oil by evaporation of
vents. The process for palm oil involves cooling an the miscella. Chlorinated solvents are also effective
equal mixture of oil and hexane to 30}333C and solvents depending on the end use of the recovered
pumping to the chiller where the mixture is held at oil. Supercritical Suids are being researched as
203C, when crystallization begins. The cooled mass is a cheaper alternative.
Figure 3 A desolventizing}toasting}drying}cooling system. Courtesy of Europa Crown Ltd., Hessle, UK.
Selected Applications in the Fats as in the development of processes to extract olive,

and Oils Processing Industry canola and coconut oils. Hexane may also be added
to aid the separation.
Table 2 shows a range of applications arising from Isopropanol and ethanol were used to extract cot-
patents during the 1970s. Other examples include tonseed oil since the combination made it possible to
separation of fat-free protein and oil from peanuts, extract the gossypol and render the solids more suit-
rice bran and soybean (hexane), cottonseed (hexane able as animal feed. However the oil needs more
plus acetic acid), starch plus oil from corn grits, germ, reRning due to phosphatides and carbohydrates
hulls and gluten (hexane or isopropyl alcohol) and which are also extracted.
separation of beef tallow into Rve fractions with dis- Two-phase extraction employing polar and non-
tinctive thermal characteristics by multistep crystalli- polar solvents has been successfully used in rapemeal
zation. Direct solvent extraction is used for low con- processing. An extraction with methanol followed by
tent ((20%) oilseeds such as soya, rice bran and washing with hexane and phase separation lowers the
milled corn germ. Pre-pressing followed by solvent glucosinate content.
extraction is utilized with high oil content seeds,
Soybeans
although some manufacturers claim high efRciency
with direct methods. Soybean oil had very rapid growth to become the
Using water as a solvent for oil extraction has dominant edible oil in the world, partly due to the
commercial applications for palm, olive and coconut versatility of the oil which enables it to be used in
oils but not for oilseeds due to high residual oil in the a wide variety of products and applications.
extracted meal and extra energy needed in the separ- Solvent extraction is the preferential method for
ation and drying steps. Adding enzymes to the aque- soybeans, the others being hydraulic pressing and
ous medium to digest cell walls can be advantageous, expeller pressing. Soybeans form stable Sakes,
Table 2 Patents issued for solvent extraction applications from 1973 to 1977
Solvent system Oil source Equipment Patent
Hexane}ethanol (2}30%) Soybean, cottonseed, Countercurrent extraction, AE Staley

safflower, sunflower, centrifugation, steam stripping
peanut, sesame
2-Propanol}H2O2 (0.1}1.0%) Oats Soxhlet, centrifugation, Du Pont

evaporation
Hexane Yeast powder Continuous solvent extraction Simon}Rosedowns

(Rotocel)
Water (1 part) plus 2 parts acetone Palm fruit, olives Multistage countercurrent RA Gouche
or ethyl alcohol}ethyl acetate} disintegrator}extractor
acetone (1 : 1 : 1) basket centrifuge
or ethyl alcohol}ethyl acetate} Buchner filter
isopropyl ether (4 : 2 : 1) vaccuum stripping
Heptane, ethylene dichloride, trichlorethylene, Animals, fish, Upward fluidized bed French Oil
perchlorethylene, hexane vegetables exchange, azeo-extraction
unlike cottonseed, peanut and other high oil- extraction process is only applied for technical
bearing seeds, where meals need to be processed purposes.
through slotted wall extruders before being solvent- Using an immersion system (e.g. Kurd), raw mate-
extracted. rial from animal processing is Rrst heated and broken
into small pieces before being mixed with the solvent
Palm (usually perchlorethylene) in an autoclave. The mis-
For extraction of palm oil it is common practice to cella is drained and may be used again for a second
use screw presses and not employ wet methods. Sim- extraction, when the solvent is evaporated and the
ilarly, solvent fractionation processes are available solids returned to the extractor for desolventizing
for fractionation but the comparative operational with steam. The solids, free of solvent, contain
cost of miscella crystallization and Rltration restricts 40}60% water and are then dried and milled for
the process to production of 2-oleodipalmitin-rich meal. Continuous percolation plants may also be util-
fractions for use in cocoa butter substitutes. Yield of ized (De Smet extractors). The raw material is broken
olein from the miscella is about 80%. up in a cooking step and excess tallow drips through
a perforated plate, leaving the residue with 30}35%
Canola (Rapeseed) and Sun]ower fat. This is ground and extracted (usually with
hexane) in a similar fashion to the system described
Canola and sunSower are high oil content seeds and for immersion extractors earlier and the solvent sep-
are generally processed by a pre-press solvent extrac- arated from the miscella by evaporation. The solids
tion. This removes the oil from the seed in two steps, are toasted to desolventize them.
which maximizes oil yields while minimizing residual Solvent extraction may be combined with press
oil in the meal. The canola presscake, which yields extraction, either by using solvent techniques for fur-
approximately 60% oil, may be mechanically ex- ther oil recovery after pressing or by pressing miscella
truded to improve the solvent extraction process. The out of the solid residue rather than recovery by de-
industry increasingly relies on DTDC equipment for cantation. These options reduce the amount of sol-
desolventizing. vent that has to be removed, lowering cost and saving
energy.
Animal Fats
Fish Oils
Screw presses are the method of choice for most
renderers, although some animal fats are solvent- Conventional Rsh oil extraction requires relatively
extracted. Use of solvents such as acetone to remove high temperatures and solvent extraction can provide
unwanted components such as cholesterol from milk a low temperature alternative, but solvent choice is
fat has proven effective. The potential for solvent limited to food grade-approved cases. The use of
residues in the product does not meet with regulatory alcohol, an approved solvent, has proven uneconomi-
or consumer approval, however, hence the solvent cal due to poor extraction efRciencies. Supercritical
carbon dioxide extraction is considered safe but, al- then reacted with oleic anhydride to give the 2-oleo-
though the product is of good quality, possible disaturated product.
removal of antioxidants, such as tocopherols and
phospholipids, and the retention of residual trace Recent Developments
components is of concern. Advances in solvent extraction technology have in-
volved the areas of energy conservation, inSuences of
Cottonseed
increasing size of extraction plants, adaptation of
Cottonseeds contain about 30% oil. Screw pressing is conventional extraction plants to produce edible
commonly used for cottonseed oil pressing, but direct meals, percolation versus immersion extractors and
solvent (hexane) extraction yields 11.5% more oil, direct solvent extraction versus pre-pressing followed
leaving less than 1% in the meal. Pre-pressing fol- by solvent extraction.
lowed by solvent extraction is the most economical Commercial developments are attractive for alter-
alternative due to the cost of solvent. ReRning is native or specialty oils compared with traditional
necessary to remove the gossypol and related oil products and for overcoming the costs and con-
pigments. straints of traditional solvent extraction systems
for minimizing industrial wastes. Utilization of va-
Saf]ower pour contactors to conserve heat during process-
Oil extraction in the safSower industry has shifted ing and dual-stage stripping columns in removing the
from a largely screw press expeller base to less costly last traces of solvent from the oil also contribute to
expander-extruders which are capable of extracting efRciency.
two-thirds of the oil and preparing collets ideal for The biggest interest in the last decade has been the
solvent extraction. Solvent is able to move naturally applications of supercritical carbon dioxide, because
through the Rbre channels and the bed acts as a natu- it has a near-ambient critical temperature (313C),
ral Rlter medium. thus biological materials can be processed at temper-
atures around 353C. The density of the supercritical
Coconut CO2 at around 200 bar pressure is close to that of
Copra is processed using a dry process com- hexane, and the solvation characteristics are also sim-
prising crushing or expelling and optional further ilar to hexane, thus it acts as a nonpolar solvent.
solvent extraction to recover the residual oil. This Around the supercritical region, CO2 can dissolve
contrasts with the wet method for the fresh kernel triglycerides at concentrations up to 1% mass. The
which separates oil from the coconut milk by centri- major advantage is that a small reduction in temper-
fugation. ature, or a slightly larger reduction in pressure, will
result in almost all of the solute precipitating out as
Olives the supercritical conditions are changed or made sub-
critical. Supercritical Suids can produce a product
Solvent recovery of oil from olives is limited to po- with no solvent residues. A wide range of fats and oils
mace processing. Superior olive oil is produced by have been extracted employing supercritical Suid ex-
pressing. The solvent extracts minor components at traction from sources including Rsh, vegetable oils,
higher levels than physical methods and requires re- nuts, cereals, citrus peel, egg yolk, wormwood and
Rning before use. yeast extract. Examples of pilot and production-scale
Cocoa Butter products include decaffeinated coffee, choles-
terol-free butter, low fat meat, evening primrose oil
Cocoa beans possess a chocolate aroma which devel- and squalene from shark liver oil.
ops during roasting of the beans. However, this Processes for the selective extraction of fats and oils
aroma is lost if solvent extraction is employed during employing propane and a mixture of propane with up
processing. The yield of cocoa butter is higher but the to 50% carbon dioxide in the subcritical state have
value may be less if the odour is desired. been described (European patents 0-591-981, 1993
The major triglyceride found in cocoa butter is and 0-721-980, 1995) for the extraction of fats and
2-oleopalmitostearin. Cocoa butter substitutes can be oils from vegetable, animal and microbial materials.
manufactured using solvent systems based on meth- The low pressures involved allow milder extraction
anol and hexane to prepare this triglyceride via isola- conditions than conventional processing, providing
tion of saturated 1,3-diglycerides from the reaction of a good yield of high grade products.
palm oil (hydrogenated soybean or cottonseed oils
have also been employed) and glycerine using sodium See also: II/Extraction: Supercritical Fluid Extraction.
methoxide catalysts. The isolated diglycerides are III/Food Technology: Supercritical Fluid Extraction.
III / FLAME IONIZATION DETECTION: THIN-LAYER (PLANAR) CHROMATOGRAPHY 2801
Further Reading Gutcho M (ed.) (1979) Edible Oils and Fats: Recent Devel-
opments. Food Technology Review No. 49. New Jersey:
Achaya KT (1994) Ghani: a traditional method of oil pro- Noyes Data Corporation.
cessing in India. Food, Nutrition and Agriculture 4(11): Head S and Sweeten T (1999) Traditional methods for
23. processing oilseeds. Inform 10(2): 151.
Bockisch M (1998) Fats and Oils Handbook. Illinois: Keeper TG (1996) Minimising solvent loss. Grasas-y-
AOCS Press. Aceites 6(24): 373.
Cavanagh GC (1997) Looking back: AOCS and vegetable Palmer MV and Ting SST (1995) Applications for super-
oil processing. Inform 8(7): 762. critical Suid technology in food processing. Food Chem-
Davie J and Vincent L (1980) Extraction of vegetable oils istry 52: 345.
and fats. In: Hamilton RJ and Bhati A (eds) Fats and Uh YH (ed.) (1996) Bailey’s Industrial Oil and Fat Prod-
Oils: Chemistry and Technology, p. 217. London: Ap- ucts, 5th edn, vols 1}5. New York: Wiley.
plied Science Publishers. Weiss TJ (ed.) (1983) Food Oils and their Uses, 2nd edn.
Gunstone FD (ed.) (1987) Palm Oil. Critical Reports in Westport, CT: AVI.
Applied Chemistry, vol. 15. New York: Society of
Chemical Industry/Wiley.
SUPERCRITICAL FLUID CHROMATOGRAPHY

See III / OILS, FATS AND WAXES: SUPERCRITICAL FLUID CHROMATOGRAPHY
FATTY ACIDS: GAS CHROMATOGRAPHY

See III / LIPIDS: Gas Chromatography
FLAME IONIZATION DETECTION:

THIN-LAYER (PLANAR)
CHROMATOGRAPHY
R. G. Ackman, Canadian Institute of Fisheries, silica gel (10 m particles) held in place by a soft glass
Halifax, Nova Scotia, Canada frit. Ten Chromarods are conveniently held in a stain-
less steel rack for application of samples and sub-
sequent development in a covered solvent tank, exact-
The Iatroscan is a British invention brought to fru- ly as for planar TLC. The removal of solvent takes
ition in Japan by Iatron Laboratories of Tokyo, which only a few minutes and the rack can then be dropped
is basically a hospital equipment company. It has into a holding frame in the Iatroscan proper for
become unexpectedly popular in such diverse analyti- scanning. This process can be controlled for max-
cal areas as marine lipids and heavy petroleum frac- imum sensitivity but usually takes less than 10 min.
tions. The combination of the resolving power of A virtue of the 10 Chromarods is that 10 different
thin-layer chromatography (TLC), itself only some- samples can be quickly compared or any combination
what more than 40 years old, with the simplicity and can be replicated or compared to calibration stan-
sensitivity of the hydrogen Same ionization detector dards run at the same time. The basic mechanism for
(FID), developed about that time as a superb detector passing the rod through the Same is fully automated.
for gas}liquid chromatography (GC), was a happy In the popular Mark III Iatroscan, the frame holding
marriage, usually summarized as TLC-FID. The basic the development rack of up to 10 Chromarods was
separation technology of the Chromarod-SIII is con- inclined. This has been replaced in the Mark V unit
ducted on a quartz rod 0.9 mm in diameter and (Figure 1) with a horizontal frame. In the Mark IV
152 mm in length, coated with 75 m thickness of Iatroscan the TLC-FID principles remained the same
2802 III / FLAME IONIZATION DETECTION: THIN-LAYER (PLANAR) CHROMATOGRAPHY
process. Generally, the Iatroscan has not found wide

application in the food industry. The response of
carbohydrates in the FID is low because of the high
oxygen content of the molecules.
The Rrst problem in taking up TLC-FID is that
those familiar with planar TLC often think in mg,
and must adapt to g } usually not more than 20 g
total per Chromarod. The second is that the applica-
tion of a few micrograms of nonvolatile material in
1 l of solvent can be automated or manual, but
always results in some band spreading at the point of
application. Solvent focusing has been found to over-
come this usually minor problem and to narrow the
sample band mixtures prior to actual development.
Usually the choice is of a poor solvent for the mater-
ials in question, and for focusing, a development of
the solvent front of less than 1 cm is adequate. An
example is presented in Figure 2.
It is rare to Rnd any unburnt organic material after
analysis but it is good practice to clean the silica gel
Chromarods regularly overnight in strong sulfuric
acid, rinsing thoroughly in water, and passing
through the scan cycle prior to use. If the previous
samples generate any residue problems, such as from
the calcium, magnesium and zinc of phytic acid, it
will show up in this conditioning scan.
Early Chromarods showed variations in thickness
and polarity that were mostly overcome with the
Figure 1 Top view of Mark V Iatroscan with horizontal rack introduction of the machine-produced silica gel
holding 10 Chromarods in position for automatic scanning in the Chromarods S-III. Alumina rods are also available
FID. The flame jet on the right is visible below the ion collector. but the literature does not indicate their wide use.
The Chromarods pass between the two FID parts as the frame
holding the development rack cycles for scanning. After each
Although there is a tendency to regard many solids
scan the frame moves sideways and returns between the and liquids as nonvolatile, this can be a tricky subject.
Chromarods, bypassing the FID and moving sideways to start the Polar groups such as those of fatty acids and esters
next Chromarod. The right-hand rod is reflected in part of the ion
collector.
but some improvements in quantitation of lipids were

found in a new detector design, and that development
led to an improved FID arrangement installed in the
Mark V. It has not yet been rigorously evaluated for
quantitation, for example in conjunction with hydro-
genation of complex lipid extracts, but should be an
improvement over the robust Mark III as regards
quantitation.
General Considerations
GC and high performance liquid chromatography Figure 2 (A) Caffeine deposited on a Chromarod S-III from an
may frequently require an hour for each analysis. queous solution and then developed and scanned. (B) Benefit of
With several sets of Chromarods at hand, an analyst solvent focusing with methanol prior to development. (Repro-
duced with permission from Ackman RG and Heras H (1997)
can conduct several types or sets of analyses per hour,
Recent applications of Iatroscan TLC-FID methodology. In
since the development times (40}50 min) and scan McDonald RE and Mossoba MM (eds) New Techniques and
times (&10 min) can overlap. Tanks with different Applications in Lipid Analysis, pp. 325I340. Champaign, IL:
solvent systems can be ready to participate in this AOCS Press.
adhere to the silica gel quite well. Sterols are polar Marine Lipids
(R-OH) molecules of reasonably high molecular
weight (387 for cholesterol), but the planar molecule The Rrst installation of an Iatroscan in North Amer-
may make hydrogen bonding difRcult, and erratic ica was in 1976 in a marine lipids laboratory. The
calibration factors have been reported. It is assumed resulting publications on analyses of various complex
that the radiant heating of the approaching Same materials attracted much attention among lipid
can sometimes vaporize part of the sterol band before chemists and biochemists, leading to a special issue of
it can be combusted to form ions. Squalene (molecu- the journal Lipids in August 1985. Lipids of indi-
lar weight 411) had practically no binding capability vidual small marine organisms could be analysed for
and can lose half its apparent mass for similar rea- the Rrst time and the sensitivity enabled extraction of
sons, but we have found that it is easily made less water-soluble lipids to be modiRed to collect and
volatile and gives a full response if the Chromarod is extract less sample, and thus conserve on solvent use.
exposed to iodine vapour for a few minutes prior to The Iatroscan was quickly adopted in many countries
scanning. with marine research programmes.
The Chromarod-Iatroscan technology for analysis It is not often recognized that many human body
of nonvolatile materials is especially useful for high- lipids, especially those of muscle, liver and the blood,
molecular-weight polymeric oxygenated materials have fatty acid compositions spanning the same range
such as are found in oxidized oils. These are usually as are found in Rsh oils and lipids. The latter include
not easy to move along the Chromarod with develop- all varieties of lipids found in ourselves and other
ing solvents, whereas simple dimeric and trimeric animals, and can be good materials to train with.
triacylglycerols can be resolved by development. Some will be featured in the few following examples
With the use of a nitrogen-speciRc attachment, the of separations as demonstrations. In Rsh muscle lipids
FID has greatly augmented sensitivity in the N-sensi- the fatty acid extremes in all lipid classes are the
tive mode. This thermionic detector mode has long relatively short chain myristic acid (14 : 0) and pal-
been available in GC, and is notoriously temperamen- mitic acid (16 : 0) on the one hand, and the long
tal. It can extend TLC-FID into the selective analysis chain, highly unsaturated eicosapentaenoic acid
of many shellRsh toxins, many of which contain a few (20:5n-3) and docosahexaenoic acid (22:6n-3) on the
atoms of nitrogen in very large molecules (e.g. mol. other hand. A superefRcient separation is shown in
wt 301, 7;N, for saxitoxin). Figure 3. In the A and B chromatograms the free fatty
For brevity this review will focus on two materials, acids and sterol esters are split into two respective
marine lipids and heavy hydrocarbon fractions, but subclasses, one with 14:0 and 16:0 as the principal
the possibilities for analysing reasonably large mol- fatty acids and the other with 20:5#22:6 as
ecules are almost unlimited. the dominant fatty acids. After hydrogenation,
Figure 3 Iatroscan TLC-FID chromatograms of a fraction enriched with neutral lipids isolated from cod flesh lipids. (A) Neutral lipid
(NL) fraction from cod flesh stored on ice for 3 days after being caught; (B) NL spiked with authentic 1-0-palmityl glyceryl ether
dipalmitate (GE) coinciding with highly unsaturated free fatty acid; (C) Hydrogenated NL spiked with GE. Solvent system hexane:diethyl
ether:formic acid; 97 : 3 :1. FFA, Free fatty acid; PL, phospholipids; SE, steryl ester; SF, solvent front; ST, free sterol; TAG,
triacylglycerol. (Reproduced with permission from Ohshima T, Ratnayake WMN and Ackman RG (1987) Cod lipids, solvent systems
and the effect of fatty acid chain length and unsaturation on lipid class analysis by Iatroscan TLC-FID. Journal of American Oil
Chemists’ Society 64: 219I223.)
Figure 4 Iatroscan TLC-FID showing the effect of the degree of unsaturation on the separation of C22 free fatty acid standards on
Chromarods-SII. Experimental conditions are development in hexane:diethyl ether:formic acid (97:3:1, v:v:v) for 40 min. O, Origin; SF,
solvent front. Shorthand gives chain length and number of methylene interrupted ethylenic bonds.
chromatogram C shows that the pairs have collapsed only in having the 1, 2 or 3 ethylenic bonds.
into single peaks as the Chromarod behaviours of the Chromarods dipped in silver nitrate can resolve such
resulting 14:0, 16:0, 20:0 and 22:0 are not very differ- mixtures as well as handle some types of cis-trans
ent. This is shown by the behaviours of selected sets separations, but these simple vegetable lipid cases are
of fatty acids and triacylglycerols (Figures 4 and 5). usually best handled by GC.
Hydrogenation is not possible with many classes of It is possible to develop one or more classes of
organic compounds: it is not only feasible in analyses lipids along the Chromarod, while ‘parking’ the rest
of lipid classes, but it has a unique advantage. The at or near the point of application, scanning partway
hydrogenated lipid fatty acids, unlike the natural down the Chromarod to determine the most mobile
highly unsaturated fatty acids, are stable to oxidation class, then redeveloping the balance of the material
and can be studied and analysed at leisure, or with applied to whatever extent is desired into the clean
different solvent systems. Peaks are also sharper, im- space thus presented by the Rrst scan. This means that
proving sensitivity limits slightly. multiple scans are always of the same original sample
For most simple lipid classes such as are found in and conducted on the same Chromarod.
vegetable oil products and mixtures, separation by A good example of multiple development is pro-
lipid classes is facilitated by the fact that the common vided by a lipid class analysis of the total lipids of the
fatty acids are palmitic (16:0), stearic (18:0), oleic muscle of the Rsh silver hake (Figure 6). The actual
(18:1), linoleic (18:2n-6) and -linolenic (18:3n-6). separation of the lipid classes was conducted with
Except for palmitic acid, these are all identical in a development sequence of three different solvent
chain length (C18), and the unsaturated acids differ systems. The extracted lipids were dissolved in
Figure 5 Iatroscan TLC-FID showing the effect of the degree of unsaturation and chain length on the separation of triacylglycerol
standards on Chromarods-SII. Experimental conditions and abbreviations are the same as in Figure 4. TAG-16:0, tripalmitin;
TAG-12:0, trilaurin; TAG-18:0, tristearin; TAG-18:3, trilinolenin. (Reproduced with permission from Ohshima T and Ackman RG (1991)
New developments in Chromarod/Iatroscan TLC-FID: analysis of lipid class composition. Journal of Planar Chromatography 4: 27I34.)
Figure 6 Sequential TLC-FID profiles of partial chromatograms of the lipid classes extracted from silver hake muscle tissue. I, II, and
III represent the three-stage development sequence to separate total lipids on a silica gel Chromarod-SIII as described in the text.
(Reproduced with permission from Zhou S and Ackman RG (1996) Interference of polar lipids with the alkali metric determination of free
fatty acids in fish lipids. Journal of the American Oil Chemists’ Society 73: 1019}1023.)
chloroform at an appropriate concentration, and this achieve relative movement of the rest of the neutral
solution was then spotted on to Chromarods-SIII in lipids, which usually develop in the order wax/sterol
1 L volumes from glass Microcap 1 L disposable esters, triacylglycerols, cholesterol, and di- and
pipettes. The Chromarods were then conditioned in monoacylglycerols, to positions where there is no
a constant humidity chamber for 5 min. The Rrst conSict with the free fatty acids.
development was carried out for 55 min in Solving such problems with TLC-FID may be com-
hexane:chloroform:propan-2-ol:formic acid; 80:14: pared with GC with only one column, and changes in
1:0.2, by vol. The Chromarods were then dried at temperature programming may be the only variable
1003C for 1.5 min and partially scanned from the top possible. With the Chromarod an unlimited choice of
to a point just below the diacylglycerol peak solvent systems is available and, when combined with
(Figure 6I). The Chromarods were then redeveloped scan and redevelopment, almost any lipid class separ-
in acetone for 15 min, dried at 1003C for 1.5 min and ation is possible.
partially scanned to below the acetone}mobile polar Figure 7 is of a commercial animal lipid mixture.
lipid position (Figure 6II). Finally, the Chromarods The A chromatogram appears to show that the domi-
were again developed in chloroform:methanol:water nant triacylglycerol is accompanied by two peaks
(70:30:3, by volume) for 60 min, dried at 1003C for matching exactly 1,3-diacylglycerols and 1,2-diacyl-
3 min and completely scanned to reveal different glycerols. This was considered to be an unusual com-
phospholipids (Figure 6III). position. To verify it, hydrogenation of 10 mg of the
In this example the free fatty acids are clearly sample (a simple process carried out by stirring in
separated from triacylglycerols. This is sometimes methanol:hexane : : 3:2 under hydrogen for 1 h, with
difRcult to achieve in a single development of a mix- a few mg of PtO2), gave the materials of the B
ture of animal lipids with one of the common lipid chromatogram. The triacylglycerol peak is sharper
class solvent systems such as hexane:diethyl and the supposed 1,2-diacylglycerol is now added to
ether:formic acid 85:9:1 (by volume). The problem the original 1,3-diacylglycerol peak. Clearly, the sup-
can be clariRed by considering the free fatty acids as posed 1,2-diacylglycerol component consisted of two
having a key position on the silica gel of the highly unsaturated fatty acids, probably a mixture of
Chromarod, and adjusting the solvent polarity to arachidonic acid (20:4n-6) and docosahexaenoic acid
Figure 7 A commercial lipid product developed (A) in a solvent mixture of hexane:ethyl acetate:formic acid; 94:6:1 (v:v:v) hydro-
genated, and (B) reanalysed. The smaller peaks, ostensibly 1,3-diacylglycerols (1,3 DAG) and 1,2-diacylglycerols (1,2 DAG), were
shown to be two types of 1,3-diacylglycerol. TAG, Triacylglycerol.
(22:6n-3), materials currently of interest in infant The problem in crude petroleum analyses was basi-
nutrition. cally the lack of natural standards, so that the quanti-
tation of the FID response would reSect the mass of
the particular complex fraction and pure chemicals
Heavy Hydrocarbon Fractions representative of a fraction were unsatisfactory refer-
At one time coal provided a variety of liquid and ence materials. For North Sea crude oils this difRculty
semisolid materials, the latter usually referred to as has been overcome by preparation of appropriate
pitch. The high-molecular-weight materials con- standard fractions from typical crude oils, so that
sisted of polycyclic aromatic hydrocarbons that could TLC-FID can provide data reliable for interlabora-
be individually deRned with some difRculty, and tory comparisons. In the petroleum laboratory par-
more complex materials that were deRned, mostly ticular difRculty is found with methods for heavy
by solubility, as maltenes, asphaltenes and pre-as- aromatic fractions and the more polar classes of ma-
phaltenes. The use of TLC-FID in their analysis terials. The latter often contain sulfur and nitrogen
has been investigated for more than a decade and it molecules and this makes some reporting technolo-
promises to reduce solvent use and speed up analysis gies of little use, but the impact on TLC-FID response
time enormously. The trend to coal liquiRcation is not very signiRcant.
to produce fuel fractions competing with petroleum One reason for industry laboratory problems is the
fractions makes new analytical technology even obsolescence of standard methods, a problem not
more useful to that industry, and at the same limited to the petroleum industry alone. When very
time the petroleum industry is turning to heavy large volumes of commodities are bought and sold
crude oils and raw materials recovered from tar there must be standards (and applicable methods)
sands. agreed on by all parties. Many have been around for
The products recovered from crude petroleum decades with no changes. Meanwhile the internal
range all the way from hydrocarbon gases to alkanes combustion engine has been progressively Rne-tuned
of chain lengths up to C100, polycyclic aromatics to conserve energy, and even the robust diesel engine
ranging from naphthalene upward, and other very needs higher standards for volatile distillates. Com-
complex high-molecular-weight materials, often in- plex reRning and cracking steps produce even more
corporating nitrogen or sulfur. In the petroleum in- heavy residues which must be investigated and utiliz-
dustry, standard methods tend to be time-consuming ed. The TLC-FID, introduced in about 1976, was
and complex. To make the life of the petroleum immediately seized on by the petroleum industry and
analyst even more difRcult, ‘cracking’ to produce offers distinct advantages. The example given in Fig-
more valuable volatile fractions leaves residues of ure 8 is taken from a recent paper on the subject. The
heavy materials such as asphaltenes. The application Chromarod scans illustrate the weaknesses of the
of TLC-FID to the latter has shown superiority to ASTM method D2007-91, based on rather lengthy
conventional methods, and has gradually been accep- and cumbersome open column chromatography on
ted, as shown by numerous publications. clay and silica gel columns in series.
Figure 8 Superiority of Iatroscan TLC-FID over ASTM D2007 method in hydrocarbon analyses as exemplified with an aromatic
petroleum extract and its fractions from the ASTM method. Chromatograms are for (A) TLC-FID of aromatic extract; (B) saturates by
ASTM D2007, (C) aromatics by ASTM D2007, (D) polars by ASTM D2007; (E) residual polars from clay. Two-step Chromarod
development of n-heptane for 30 min, followed by development with toluene for 5 min. (Reproduced with permission from Barman BN
(1996) Hydrocarbon-type analysis of base oils and other heavy distillates by thin-layer chromatography with flame ionization detection
and by the clay-gel method. Journal of Chromatography Science 34: 219I225.)
Conclusion deRned by decades of patient work by others, but the

adaptation to rapid analysis by thin-layer silica gel
A recent paper on supercritical Suid chromatography chromatography on a microgram scale may require
suggested that often attempts to replace older and a combination of imagination, knowledge and perse-
proven GC and HPLC methods with novel techno- verence. The TLC-FID is a system that offers the
logy can be disappointing. That the TLC-FID system challenge that makes science enjoyable!
has been popular in only a few analytical Relds may
be due to the need for close interaction between the See also: III/Geochemical Analysis: Gas Chromatog-
raphy. Lipids: Gas Chromatography; Liquid Chromatogra-
analyst and the method } almost a lost art in the face
phy; Thin-Layer (Planar) Chromatography. Oils, Fats and
of contemporary automated equipment. Waxes: Supercritical Fluid Chromatography. Petroleum
One exception to this is the TLC-FID of the Products: Thin-Layer (Planar) Chromatography.
Chromarod-Iatroscan combination, mostly used in
research laboratories. As long as researchers have
relatively nonvolatile organic materials to analyse,
Further Reading
their resolution and determination by TLC is often Lipids
a challenge for which the Sexibility of the Iatroscan is Ackman RG and Heras H (1997) Recent applications of
ideally suited. Their chemical nature may have been Iatroscan TLC-FID methodology. In: MacDonald RE
2808 III / FLASH CHROMATOGRAPHY
and Mossaba MM (eds) New Techniques and Applications Sebedio J-L, Farquharson TE and Ackman RG (1985)
in Lipid Analysis, pp. 325}340. Champaign: AOCS Press. Quantitative analyses of methyl esters of fatty acid geo-
Hara K, Cho S-Y and Fujimoto K (1989) Measurement of metrical isomers, and of triglycerides differing in unsat-
polymer and polar material content for assessment of uration, by the Iatroscan TLC-FID technique using
the deterioration of soybean oil due to heat cooking. AgNO3 impregnated rods. Lipids 20: 555}560.
Journal of the Japan Oil Chemists’ Society 38: 463}470. Shantha NC and Ackman RG (1990) Advantages of total
Kaitaranta JK and Ke PJ (1981) TLC-FID assessment of lipid hydrogenation prior to lipid class determination on
lipid oxidation as applied to Rsh lipids rich in triglycer- Chromarods S-III. Lipids 25: 570}574.
ides. Journal of the American Oil Chemists’ Society 58:
710}713. Hydrocarbons
Ohshima T and Ackman RG (1991) New developments in
Chromarod/Iatroscan TLC-FID: analysis of lipid class Barman BN (1996) Hydrocarbon-type analysis of base oils
composition. Journal of Planar Chromatography 4: and other heavy distillates by thin-layer chromatogra-
27}34. phy with Same ionization detection and by the clay-gel
Ohshima T, Ratnayake WMN and Ackman RG (1987) method. Journal of Chromatographic Science 34:
Cod lipids, solvent systems and the effect of fatty acid 219}225.
chain length and unsaturation on lipid class analysis by Bharati S, Patience R, Mills N and Hanesand T (1997)
Iatroscan TLC-FID. Journal of the American Oil Chem- A new North Sea oil-based standard for Iatroscan analy-
ists’ Society 64: 219}223. sis. Organic Geochemistry 26: 49}57.
Parrish CC (1987) Separation of aquatic lipid classes by Poirier MA, Rahimi P and Ahmed SM (1984) Quantitative
Chromarod thin-layer chromatography with measure- analysis of coal-derived liquids residues by TLC with
ment by Iatroscan Same ionization detection. Canadian Same ionization detection. Journal of Chromatographic
Journal of Fisheries and Aquatic Science 44: 722}731. Science 22: 116}119.
FLASH CHROMATOGRAPHY
C. F. Poole, Wayne State University, Detroit, MI, complex mixtures into simpler groups for analysis. Its
USA primary virtue is low cost, since virtually no special
equipment is required, and the stationary and mobile
phases are inexpensive enough to be discarded after
a single use, or can be recycled. Resolution is less than
Flash chromatography and related techniques are that obtained by medium and high pressure liquid
widely used for laboratory-scale fractionation of mix- chromatography but the operational costs and equip-
tures from organic synthesis or for analysis when only ment needs are greater for these techniques. Flash
a modest increase in resolution over conventional chromatography is often employed as a pre-separ-
column liquid chromatography is required. These ation technique to remove particulate matter and
techniques employ short columns packed with par- sample components that are either weakly or strongly
ticles of an intermediate size (typically 40}60 m) retained on the separation column in medium and
combined with accelerated solvent Sow achieved high pressure liquid chromatography. This allows
through modest pressure or suction. Compared to higher sample loads to be separated under more selec-
conventional column liquid chromatography, separ- tive separation conditions and avoids column con-
ations are obtained in less time; isolated compounds tamination and regeneration problems. The produc-
are often purer because resolution between bands is tion costs of the isolated products are thus rendered
increased and band tailing is reduced; and com- more favourable.
pounds that are degraded or altered during
chromatography are recovered in higher purity be-
cause of the shorter contact time with the chromato-
Dry-column Chromatography
graphic system. Dry-column chromatography (Figure 1) is a variant
The main applications of Sash chromatography are of preparative thin-layer chromatography with sim-
puriRcation of synthetic products, isolation of target ilar resolution but a higher sample loading capacity.
compounds from natural products, the simpliRcation A glass column or nylon tube is packed with thin-
of mixtures prior to high resolution preparative (usu- layer chromatographic grade sorbent, usually silica
ally) liquid chromatography and the fractionation of gel, to a height of 10}15 cm. Sample is added as
III / FLASH CHROMATOGRAPHY 2809
Dry-column chromatography is not a widely used

technique. Preparative thin-layer chromatography or
Sash chromatography has generally been preferred.
Although separations are fast, the recovery of separ-
ated zones is slow and labour-intensive compared to
elution methods.
Vacuum Chromatography
Vacuum chromatography can be taken to mean the
operation of a short column under suction to acceler-
ate solvent migration. Either a short column or
a BuK chner Rlter funnel Rtted with a glass frit is dry-
packed with sorbent. The sorbent bed is consolidated
initially by tapping the side of the column during
Rlling and pressing the top layer of the sorbent bed
with a Sat object, such as a stopper, while suction is
applied at the other end. Consolidation is completed
by releasing the vacuum and pouring a solvent of low
polarity over the surface of the sorbent bed followed
Figure 1 Apparatus for dry-column chromatography.
by restitution of the vacuum. If the column is packed
correctly the solvent front will descend the column in
a horizontal line; otherwise the column should be
a concentrated solution or preadsorbed on to a small sucked dry, repacked and tested again. When all the
amount of sorbent. Separation is achieved by devel- solvent has passed through the column, residual sol-
oping with a suitable volume of solvent to reach the vent trapped between particles is removed by suction.
lower end of the bed. Suction at the bottom of the A solution of the sample in a suitable (weak) solvent
column and/or slight overpressure at the top may be or preadsorbed on to a small amount of sorbent or
required to supplement capillary forces in moving the inert material, such as Celite, is applied to the top of
mobile phase down the column. Separated bands the column (Figure 2). The sample solvent, if used, is
are removed by extrusion, slicing (if a nylon column sucked gently into the column packing. A piece of
is used) or by digging out, and the products freed Rlter paper with the same diameter as the inside
from the sorbent by solvent extraction. The separ- diameter of the column or funnel is placed on top
ation is fast, requires very little solvent and provides of the sorbent bed to prevent disruption of the
higher resolution than classical column techniques bed during addition of solvent. The column is
due to the use of sorbents with a smaller average then eluted with appropriate solvent mixtures of
particle size. It is suitable for the recovery of small gradually increasing solvent strength. Between sol-
quantities of material since the loading capacity is vent applications the column is sucked dry and
only about 0.2}1.0% w/w of the sorbent used, de- the eluent collected in test tubes or round-bottom
pending on the difRculty of separating the bands of Sasks. Using a multiport manifold (similar to a pig
interest. adapter for distillation) or a separatory funnel
Thin-layer chromatography provides a suitable allows sequential fraction collection without having
technique for method development in most cases, to disassemble the apparatus after each fraction is
although signiRcant differences in separations can collected.
arise for mixed solvents, particularly when the sol- Vacuum chromatography is simple, rapid and con-
vent components differ in polarity and/or volatility. venient. Optimum sample loads are similar to Sash
These differences result from the absence of pre- chromatography. However, it is not unusual to use
equilibrium with a vapour phase in the dry-column sample overload conditions to separate simple mix-
technique. Nylon columns can be more difRcult to tures by stepwise gradient elution or to simplify
pack than glass columns, particularly when longer mixtures for further separation. Under these con-
lengths are used, but nylon columns are easier to ditions the sample loads may reach 10% (w/w),
section and allow colourless bands to be observed or even higher, of the bed mass. Compared to
with a UV lamp. Glass columns built up of segments Sash chromatography, solvent changes are easy
connected by ground-glass joints can be useful for because the head of the column is at atmospheric
simplifying the extrusion process. pressure.
Figure 2 Apparatus for vacuum chromatography. (Reproduced from Pelletier SW, Chokshi HP and Desai HK (1986) Separation of
diterpenoid alkaloid mixtures using vacuum liquid chromatography. Journal of Natural Products 49: 892, with permission from the
American Chemical Society.)
Flash Chromatography valve. Pressures employed are typically less than

1}2 atm, with the various parts of the apparatus
A glass column of a suitable length containing a small (Figure 3) held in place by springs, clamps or screw-
glass-wool plug and a layer of acid-washed sand or thread connectors. It is a wise precaution to use
glass frit at its base is partially Rlled with sorbent plastic-coated glass columns or a safety shield to
using the dry-packing or slurry-packing technique. minimize the possibility of accidents. The column
Incremental addition of the sorbent followed by tap- should not be allowed to run dry during the elution
ping of the column with a hard object generally gives sequence.
etter results for dry packing than bulk Rlling of the Flash chromatography is simple to perform and is
column. After packing, the column is freed from widely used in many laboratories. The main disad-
trapped air and further consolidated by forcing sev- vantage is that the apparatus requires constant
eral column volumes of a weak solvent through the disassembly and reassembly of the air pressure
sorbent bed until no further air bubbles are seen inlet adapter in order to introduce new solvent into
exiting the column and the bed is stable. It is difRcult the column. Potter described an apparatus with a
to pack wide-diameter columns ('5 cm) by dry- lateral solvent reservoir to overcome this problem.
packing procedures and in this case slurry packing is The essential feature of this apparatus is that a sol-
nearly always used. In this case, the column is par- vent reservoir with tap at the reservoir-to-column
tially Rlled with a small volume of weak solvent and inlet is attached to the side of the column with the
a dilute suspension of the sorbent in the same solvent inlet situated above the height of the sorbent bed.
is added slowly in increments, with excess solvent A second tap at the top of the column allows air
intermittently drained away. Periodic pressurization pressures to be equalized for rapid solvent addition to
of the sorbent bed is used to aid consolidation. Tap- the column, without having to disassemble the appar-
ping the sides of the column is not normally em- atus. Radial compression columns have also been
ployed. The sample is added to the column in a small used with some commercial Sash chromatography
volume of solvent or adsorbed to a small amount of apparatus.
packing material.
Finally, a thin layer of glass beads, acid-washed
sand or other inert material is added to the top of the
Stationary Phases
column to prevent disturbance of the column bed by Silica, and to a lesser extent alumina, are the most
solvent added for elution. The amount of free space common stationary phases used for the separation of
above the sorbent bed must be sufRcient to hold the low molecular weight organic compounds. Chemi-
volume of solvent equivalent to the fraction size col- cally bonded silica sorbents are used for the separ-
lected, or a solvent reservoir must be inserted between ation of polar organic compounds in the normal
the column and the pressure regulation valve. The and reversed-phase modes. Wide-pore, chemically
Sow rate is adjusted to about 5 cm min\1 by applica- bonded phases are used for the separation of bio-
tion of gas pressure and controlled by the regulation polymers. There is no technical reason why any
sizes is about 5 cm min\1. At this velocity well-

packed columns are expected to provide about 5}20
theoretical plates per centimetre of bed height, de-
pending on the column packing density and the qual-
ity of the sorbent material.
Some separations demand specially prepared sta-
tionary phases. A method has been described for
impregnating silica with silver nitrate for the isolation
of compounds with unsaturated groups capable of
forming charge transfer complexes with silver. Silica
was impregnated with phosphoric acid and the cal-
cium salt of ethylenediaminetetraacetic acid to isolate
microtoxins that were either unstable or produced
tailing bands on normal silica gel. Thin-layer
chromatography is generally a suitable technique to
identify suitable additives for improving chromato-
graphic properties in Sash chromatography. Silica
and chemically bonded phases coated with cellulose
tris(3,5-dimethylphenyl carbamate) were used to iso-
late 10}100 mg amounts of pure enantiomers from
racemic mixtures. The selection of mobile phases for
this application is conveniently optimized using high
pressure liquid chromatography.
Sample Loading
The sample is usually added to the column in a small
volume of a weak solvent and the solution forced into
the sorbent bed, forming a narrow sample application
zone. For samples of low solubility in weak solvents,
the sample is taken up in a strong solvent and added
Figure 3 Apparatus for flash chromatography.
to a small amount of column packing material or
other inert support. The solvent is then stripped from
moderately rigid chromatographic sorbent, stable to the slurry under vacuum to produce a dry free-Sow-
solvent changes and available in the required particle ing powder that can be added to the top of the
size range, could not be used. In practice, the cost of column. Sorbent (1}2 g) may be required for each
the sorbent has to be set against the value of the gram of sample. It is important that the sample is
product isolated, since the sorbent is often used for completely dry (high vacuum is used to remove the
a single sample application and regeneration may be last traces of solvent) and free of lumps to obtain
impossible, or tedious, costly and uncertain. Chemic- symmetrical separated zones. If the sample layer is
ally bonded phases are more expensive than silica, or relatively long compared to the column bed length,
have to be synthesized form silica prior to use, and for then a stepwise solvent gradient must be used for
this reason are less popular. For low molecular elution to minimize zone broadening.
weight neutral organic compounds, small-pore silicas There are no simple relationships between the
with a high surface area and high loading capacity are sample amount that can be separated, the dimensions
preferred, in particle size ranges of 20}40 or of the sorbent bed and the volume and number of
40}63 m. The smaller particle size materials provide collected fractions. The loading capacity depends on
higher resolution per unit length but generate greater the ease of separation of neighbouring zones, the
back-pressure. Since longer columns can be used for sorption capacity of the sorbent and the method of
the larger particle size sorbent, differences in resolu- sample elution. It can be increased by using wider
tion are often not great. Because of the limited operat- columns and sorbents with a larger speciRc surface
ing pressure, columns are rarely longer than 30 cm, area. A rough empirical guide is presented in Table 1.
and 10}15 cm is recommended, unless longer col- For stepwise gradient elution it has been assumed that
umns are required to provide additional resolution. the sample can be separated into fractions of different
The optimum mobile-phase velocity for these particle polarity when estimating the typical sample load.
Even for difRcult samples it is often more productive For samples of wide polarity a useful gradient is to
to use column overload conditions, combining frac- start from a weak solvent, such as hexane, and add to
tions containing pure materials, and recycle those this various volume increments of a strong dipolar
containing mixtures. Flash chromatography may lack solvent (such as ethyl acetate, dichloromethane,
the resolving power needed to separate the compo- chloroform or acetone), terminating with the strong
nents of interest. In this case a higher resolution dipolar solvent. Then continue adding volume in-
technique, such as medium or high pressure liquid crements of a strong hydrogen-bond solvent (meth-
chromatography, would be a better choice, perhaps anol, ethanol, 2-propanol) to the strong dipolar
using Sash chromatography to isolate fractions con- solvent, terminating with the strong hydrogen-bond
taining the components of interest from other sample solvent. Monitoring the separation by thin-layer
components. chromatography allows the solvent gradient to be
trimmed and optimized to suit the requirements for
individual separations. Predicting the number of frac-
Method Development tions required at each step remains quite arbitrary
Thin-layer chromatography is widely used to opti- and is best conducted by monitoring the composition
mize the separation conditions for silica gel Sash of each fraction as it is collected. When adding
chromatography. For isocratic separations a mobile a strong solvent in a binary mobile phase for a silica
phase which provides an RF of about 0.35 for the gel sorbent, it is important to note that the solvent
zone of interest is chosen. If several zones are to be strength for the mixture has a steep curved proRle.
separated, then the solvent strength is adjusted such For compositions containing low volume fractions of
that the centre zone has an RF of 0.35. If all zones of strong solvent, the volume fraction of strong solvent
interest are well separated from each other and from should be incremented by small changes, resulting in
impurities (RF50.2), then the solvent strength can relatively large changes in retention, for example,
be adjusted so that the most retained zone of interest 1%, 3%, 5%, 10% (v/v). At higher volume fractions
has a RF+0.35. For fractionation and large sample of strong solvent, the changes in volume fraction
loads it is critical that the most selective solvent com- should be larger to produce a signiRcant change in
position for the separation is used. This can be quick- retention, for example, 30%, 40%, 60%, 80%,
ly identiRed using the PRISMA model, a guided trial- 100% (v/v).
and-error procedure using thin-layer chromatogra- Silica gel (or alumina) is the most suitable sorbent
phy and parallel separations with different solvents. for the separation of low molecular weight organic
The same process can be used to identify the composi- compounds soluble in organic solvents and for separ-
tion of solvents suitable for the recovery of individual ations of geometric isomers and diastereomers. For
sample zones in order of increasing polarity by step- compounds at the extreme end of the general adsorp-
wise gradient elution. tion scale (Figure 4), separations are difRcult because
Table 1 Approximate sample-loading conditions for flash chromatography

(density of silica+0.45 g mL\1)
Column Amount of Sample loading for a Sample Typical fraction

diameter silica gel particular TLC resolution (g) loading volume (ml)
(cm) (g) (g)
RF50.2 RF50.1
Isocratic elution (bed heightⴝ15 cm)

1 5 0.1 0.04 5
2 20 0.4 0.16 10
3 45 0.9 0.36 20
4 80 1.6 0.6 30
5 130 2.5 1.0 50
Stepwise gradient elution (bed heightⴝ10 cm)

3 30 1}3 50}100
4 55 3}8 100}200
6 125 8}35 200}300
8 250 35}60 200}300
10 350 60}80 300}500
14 700 80}150 300}500
For neutral organic compounds, thin-layer chro-

Difficult to separate because Alkanes Weak
solvent strength is too high Aromatics matography is widely used. Microscope slide-sized
Halogenated compounds plates are suitable to screen individual fractions as

Ethers they are obtained and larger plates for the grouping of
Nitro compounds multiple fractions. A wide range of selective and uni-

Nitriles versal visualizing reagents are available to meet most
Carbonyl compounds
detection requirements. Compounds with UV absorp-
Alcohols
Phenols tion can be visualized by Suorescence diminution

Amines using layers containing an inorganic Suorescent indi-
Amides
cator. But most of all, thin-layer chromatography is
Difficult to separate because Carboxylic acids
used because it is quick, portable, inexpensive and
solvent selectivity is too low Sulfonic acids Strong
generally adequate for the task.
Figure 4 General adsorption scale for silica gel
chromatography. Future Developments
Flash chromatography and related laboratory-scale
of inadequate selectivity. Water-soluble compounds, techniques are already widely used for preparative
including biopolymers and easily ionized compounds, chromatography when only modest resolution is re-
are generally better handled by reversed-phase quired. The virtues of these techniques are favourable
chromatography. Compounds of low polarity that cost considerations and minimal instrumentation re-
are weakly retained on silica gel with hexane as a sol- quirements. They are not a substitute for high resolu-
vent can be separated on chemically bonded phases in tion, preparative-scale techniques but a complement
the normal or reversed-phase modes. For reversed- to them. Consequently, radical changes in how Sash
phase separations, chemically bonded phases with chromatography is carried out are not expected. The
water as the weak solvent are used, and the solvent most likely future development is the wider use of
strength and selectivity of the eluting solvent changed sorbents other than silica gel in generally optimized
by adding different volumes of water-miscible or- separation schemes, made possible by the declining
ganic solvents, such as acetone, methanol, acetonit- cost of chemically modiRed and other selective
rile, tetrahydrofuran, etc. Optimization of solvent sorbents.
composition by thin-layer chromatography is pos-
sible but predictions may be unreliable due to differ- See also: II/Chromatogrphy: Thin-Layer (Planar):
ences in sorption properties between the column and Modes of Development: Conventional; Spray Reagents.
layers. A better solution is to pack a short (10 cm)
metal column with the sorbent for Sash chromatogra- Further Reading
phy and use high pressure liquid chromatography to
optimize separation conditions. Ideally, for isocratic Chappell I and Baines PE (1991) Bio-Sash chromatography.
Rapid, low-cost, puriRcation of peptides. Biochromato-
elution a solvent composition should be chosen that
graphy 10: 236}238.
provides a retention factor of 2}3 for the component Claeson P, Tuchinda F and Reutrakul V (1993) Some em-
of interest or those components of a mixture that are pirical aspects on the practical use of Sash chromatogra-
the most difRcult to separate. For mixtures of wide phy and medium pressure liquid chromatography for the
polarity, stepwise solvent gradients are easily con- isolation of biologically active compounds from plants.
structed and optimized by the same approach. Journal of the ScientiTc Society of Thailand 19: 73}86.
Conway WD, Bachert EL, Sarlo AM and Chan CW (1998)
Comparison of countercurrent chromatography with
Detection Sash chromatography, Journal of Liquid Chromatogra-
Monitoring separations by Sash chromatography can phy & Related Technologies 21: 53}63.
be online and continuous using standard liquid Edwards C, Lawton LA, Coyle SM and Ross P (1996)
Laboratory-scale puriRcation of microcystins using Sash
chromatographic detectors (e.g. UV-visible, refractive
chromatography and reversed-phase liquid chromatog-
index, or evaporative light scattering) but is more raphy. Journal of Chromatography A 734: 163}173.
commonly done ofSine by collecting fractions that Gogou AI, Apostolaki M and Stephanou EG (1998) Deter-
are subsequently combined, based on the similarity of mination of organic molecular markers in marine aero-
their composition. Suitable monitoring techniques sols and sediments: one-step Sash chromatography
are thin-layer, gas or liquid chromatography, elec- compound class fractionation and capillary gas chro-
trophoresis, bioassays, immunoassays and spectro- matographic analysis. Journal of Chromatography
scopy (e.g. infrared and nuclear magnetic resonance). A 799: 215}231.
2814 III / FLAVOURS: GAS CHROMATOGRAPHY
Grieb SJ, Matlin SA and Belenguer AM (1996) Flash Milat M-L and Blein J-P (1995) Cercospora beticola toxins
chromatography with cellulose tris(3,5-dimethyl- III. PuriRcation and thin-layer and high-pressure liquid
phenylcarbamate)-coated phases. Improved resolution chromatographic analyses. Journal of Chromatography
of basic analytes. Journal of Chromatography A 728: A 699: 277}283.
195}199. Poole CF and Poole SK (1991) Chromatography Today.
Hostettmann K, Marston A and Hostettmann M Amsterdam: Elsevier.
(1997) Preparative Chromatography Techniques. Potter GA (1994) New lateral reservoir Sash chromatogra-
Applications in Natural Product Isolation. Berlin: phy system for the expeditious preparative puriRcation
Springer. of organic compounds. Journal of Chromatography A
Li T-S, Li J-T and Li H-Z (1995) ModiRed and conve- 675: 237}239.
nient preparation of silica impregnated with silver ni- Still WC, Kahn M and Mitra A (1978) Rapid chromato-
trate and its application to the separation of steroids and graphic technique for preparative separations with
triterpenes. Journal of Chromatography A 715: moderate resolution. Journal of Organic Chemistry 43:
372}375. 2923}2926.
FLAVOURS: GAS CHROMATOGRAPHY
F. P. Scanlan, Quest International, Naarden, Weurman Flavour Research Symposia. Many GC

The Netherlands methods have been used for a variety of purposes,
Copyright ^ 2000 Academic Press including Savour and raw material quality control,
Savour stability, identiRcation of off-Savours and
taints, studies of Savour biogenesis and metabolic
Flavours are composed from a wide variety of mater-
pathways of plant volatiles, identiRcation of new Sa-
ials such as essential oils, extracts of natural products,
vour molecules, consumer product development, and
individual aroma chemicals and many other materials
process optimization. The Savour analysis techniques
having an organoleptic impact producing a desired
covered here are: headspace GC including solid-
effect. Flavour analysis by gas chromatography (GC)
phase microextraction (SPME) combined with GC
is only associated with volatile and semi-volatile com-
thermal desorption techniques, pyrolysis-GC-MS,
pounds. From an early stage in the development of
multidimensional GC, GC-MS and GC with selective
GC, the analysis of Savours has played an important
detectors, chiral separations, GC-O and the recently
part. Up until the 1980s, the emphasis was on using
developed fast GC process. Of course the quality of
GC to identify individual Savour molecules at trace
an analysis depends on the extraction techniques and
levels (parts per million (ppm) to parts per trillion
sample preparation procedures. These areas are
(ppt) range and sometimes less). This is still an impor-
covered in other chapters in this encyclopedia. Head-
tant part of the job for Savour analysts, although
space analysis and pyrolysis are also considered to be
today there is a greater tendency to correlate molecu-
sampling techniques, but they are included here as
lar structures with sensory attributes. GC methods
they can be used in a coupled mode.
that have been developed for Savour analysis include
temperature-programmed capillary GC and GC com-
bined with mass spectrometry (MS). Likewise, a tech- Headspace Gas Chromatography and
nique such as GC-Olfactometry (GC-O), which was
once the realm of the Savour and fragrance industry,
Thermal Desorption Techniques
is currently enjoying applications in the food and Since the early application of headspace GC analysis
beverage, cosmetics, packaging, plastics, and phar- to Savours the technique has undergone a consider-
maceuticals industries. able degree of automation. Now it is possible to
This article provides a comprehensive review of the perform high throughput analysis, and reduce varia-
most important GC applications in Savour analysis. bility by automating the sampling and injection pro-
The Further Reading section lists some important cess. Basically there are two forms of headspace anal-
literature sources that provide a useful overview of ysis: static and dynamic. In both forms volatiles that
GC as applied to Savour analysis. In addition there could be a source of interferences for the GC separ-
is the Food Science & Technology Series from Elsevier ation are removed from a complex sample matrix. It
Science Ltd. This series contains the proceedings is important to note that the headspace contains the
from the International Flavor Conferences and the part of the Savour that one perceives Rrst. Static
III / FLAVOURS: GAS CHROMATOGRAPHY 2815
headspace analysis is performed in a closed vessel in for 15 min at room temperature. A 50 : 50 v/v
which the volatiles reach an equilibrium between two (100 mL) mixture of pentane and ether was used for
phases. The results depend on the partition coefR- the solvent extraction. To obtain a representative
cients of the individual molecules between these extract, a combination of polar and apolar solvents is
phases. In dynamic systems an inert carrier gas is often used in Savour analysis. GC analysis was run on
swept over the sample and the volatiles are trapped a 50 m;0.32 mm i.d.;1 m HP-5 apolar phase col-
onto a support. Static headspace analysis has been umn. The Rrst observation to be made is that the
automated in combination with cryofocusing devices. solvent peaks at the beginning of Figure 1 are absent
Dynamic headspace analysis can handle higher on Figure 2. Also absent in Figure 2 is the very broad
sample throughputs using automated thermal desorp- peak at 39}45 min. This peak is benzoic acid, which
tion devices in combination with cyrofocusing to unfortunately also creates interferences with other
treat a series of sampling tubes containing the trapped compounds eluting over the same time period. As
volatiles. A variety of adsorbent phases are used for expected, solvent extraction also shows some smaller
trapping volatiles according to their polarity. In re- broad peaks belonging to the polar acids eluting be-
cent years the solid phase micro-extraction (SPME) fore 24 min. The absence of these peaks with SPME is
technique has been developed for sampling volatiles. expected as PDMS is an apolar material. This absence
It is based on the principle of using a stationary phase can also be an advantage in that interferences can be
coated onto a Rbre that traps the volatiles in contact reduced. Relative concentrations of extracted com-
with the surface. The Rbre can be placed in the head- pounds can be lower with SPME as compared with
space above the sample or indeed can be plunged into solvent extraction. This is because of the limited sur-
a liquid sample. After a predetermined time, the Rbre face area available for adsorption on the SPME Rbre.
is removed and inserted directly into the GC injection Quantitative results are difRcult to compare, but
port. Commonly used phases are polydimethyl- qualitatively both techniques reveal nearly all the
siloxane (PDMS) and polyacrylate (PA) as well as the same apolar Savour compounds. The most abundant
more conventional octadecylsilyl C18. The simplicity apolar compound after just 31 min is limonene. An-
and cost effectiveness of this technique has led to its other difference between the two techniques is the
widespread application for Savour analysis. It is also molecular weight range extracted. Solvent extraction
possible to use this technique with some GC autosam- allows the largest molecular weight range to be ex-
plers. The GC chromatograms in Figures 1 and tracted. While there are limitations to SPME, this can
2 allow a comparison of results obtained from solvent be advantageous for some applications in which it is
extraction and SPME, respectively. The sample was necessary to avoid interfering compounds and pro-
a lemon Savour. For SPME extraction, a PDMS Rbre duce a ‘cleaner’ extract. Thermal desorption tech-
was placed over the sample in a closed conical Sask niques involve the extraction of volatiles from
Figure 1 GC trace obtained after solvent extraction of a lemon flavour.

Figure 2 GC trace obtained after SPME of a lemon flavour for 15 min.
a sample contained in a glass tube, or as mentioned (NMR) or infrared (IR) analysis. Another variation
previously from a trapping material, directly into on the single oven is the dual-oven system. This has
a GC injection port. A form of cryofocusing or cooled the advantage of controlling the temperature of both
injection is required for reliable quantitative analysis. columns. Multidimensional systems can be used for
performing semipreparative GC isolations. They have
Multidimensional Gas also been used with a isotope ratio mass spectrometer
for authenticity analysis. Multidimensional systems
Chromatography (MDGC) are very Sexible and are extremely useful in the Sa-
Flavours and especially natural Savour extracts can vour laboratory for ultratrace analysis.
be extremely complex, containing hundreds of com-
pounds that are not always completely resolved. The
problems of separating overlapping peaks can be
Gas Chromatography+
overcome by the use of very high resolution columns Mass Spectrometry and Gas
and also by using two columns instead of one to Chromatography+Selective Detection
improve the peak capacity. The use of a switching
Gas Chromatography+Mass Spectrometry (GC-MS)
mechanism between the columns makes it possible to
select a segment of the chromatogram obtained with GC analysis beneRted from the advances in more
the Rrst column and transfer it to a second column for stable and higher resolution columns and became
further separation (heart-cutting). The second col- more powerful when used with a mass spectrometer.
umn may be of another polarity or performance or Without underestimating other analytical techniques,
indeed may be a chiral column. Such a system can GC-MS is probably by far the technique that has
also be used in connection with a collection device to contributed the most to the analysis of Savours over
recover the isolated component. The collection device the last 30 years. The literature is plentiful on GC-MS
may be a cryogenic system or glass tubes Rlled with applications to Savour analysis. The combination of
a suitable trapping materials. An elegant possibility the separation power of the gas chromatograph with
exists whereby the glass collection tubes Rt into the the selectivity of the mass spectrometer has been the
injection port of a GC column Rtted with a septum- major analytical tool for revealing components of
less cooled injection system, facilitating the transfer essential oils and natural products as well as all vol-
of the isolated volatiles into a GC-MS or GC-O usu- atile Savouring materials. GC-MS applications en-
ally Rtted with a pre-column. Equally, using a simple abled the VCF (volatile compounds in food) list of
accessory the isolated volatiles can be transferred to known Savour molecules to grow from about 500 in
a suitable support for nuclear magnetic resonance the early 1960s to nearly 7000 today.
Gas Chromatography+Atomic Emission Detection Gas Chromatography+Electron-Capture Detection

(GC-AED) (GC-ECD)
The atomic emission detector is a multi-elemental The analysis of some classes of Savour compounds,
detector based on the principle of scanning the atomic such as the highly volatile aldehydes or volatile
emission bands of several elements as compounds fatty acids, is facilitated by employing derivatization
elute from the GC and enter a microwave plasma. For techniques using halogen-containing reagents, e.g.
the selected element, AED is generally more sensitive pentaSuorobenzylhydroxylamine (PFBOA). The
than the Same ionization detector (FID) (at low parts high sensitivity of the ECD technique is therefore
per million (ppm) levels) and there is the possibility of advantageous for detecting trace-level derivatized
acquiring data for several elemental wavelengths carbonyl compounds. This approach has also been
simultaneously. AED can be used as a low-level applied in Savour studies of lipid oxidation com-
screening tool for organohalogens, organosulfur, or- pounds.
ganometallics and also some isotopes useful for
isotopic labelling studies. With respect to essen-
tial oils and Savours, this technique has been used Chiral Separations
for the detection of trace contaminants such as pes-
ticides. One of the advantages of AED is that complex Many Savour molecules have one or more chiral
matrices such as a citrus oil may be injected neat carbons and can exist as enantiomers. The separation
or as a simple dilute solution to avoid the need of enantiomers by GC can be used as a method for
for tedious extraction techniques. GC-AED is studying their individual odours or for the purposes
a useful complementary technique to benchtop GC- of authenticity analysis. Most natural biosynthetic
MS. pathways produce Savour molecules with one pre-
dominant enantiomer. This can be exploited for the
control of food and beverage adulteration. The chiral
Gas Chromatography+Sulfur Detection GC chromatograms shown in Figures 3 and 4 are
It was only in the 1980s that attention was given to analyses of Savour extracts made from strawberry
the relevance of sulfur volatiles and semi-volatiles to fruit preparations. The chiral column was 25 m long
the character of Savours. Previously, sulfur molecules and contained a diacetyl tert-butyldimethylsilyl -
were not considered very important and were cyclodextrin stationary phase. Figure 3 shows the
more often associated with off-odours and contami- separation of both the (R)- and (S)-enantiomers of
nations. Studies carried out on products such as ethyl-2-methylbutyrate at 9 min while only the (S)-
cheese, trufSes and fresh strawberries have revealed enantiomer is present in Figure 4. Both enantiomers
the importance and Savour impact of sulfur com- in Figure 3 are present in almost equal abundance
pounds. Both sulfur chemiluminescence detection which is indicative of the addition of a ‘nature identi-
(SCD) and Same photometric detection (FPD) are cal’ ester. The development of chiral stationary
used for selective sulfur detection. These techniques phases (mainly modiRed -cyclodextrin phases) has
have also been used in combination with snifRng port allowed the resolution of many enantiomers. When
detection and headspace analyses. Generally, the enantiomer separations have been combined with
identiRcation of sulfur compounds has been difRcult CG-O to differentiate the odour of each enantiomer
because their olfactory detection threshold is ex- they have revealed three main classes of enantiomers:
tremely low. Thus, they may be detected by the nose enantiomers of equal odour; enantiomers having the
but the analyst must carry out several extractions and same odour but different intensities or secondary
concentrations of the product to obtain a sample notes; and enantiomers with quite different odours.
large enough to cause a peak to appear on the Examples of enantiomers with different odours are
chromatogram. (#)-nootkatone, which has the aroma of grape-fruit
peel, and (!)-nootkatone, which has stronger
woody, spicy notes and only minor grapefruit peel
Gas Chromatography+Nitrogen Detection
character. Also, (#)-(S)-carvone has a typical
Nitrogen}phosphorus detection (NPD) has been caraway oil smell while (!)-(R)-carvone has a
applied to the analysis of nitrogenous compounds minty odour. An interesting compound is 5-an-
in cheese, meat and yeast extracts for Savouring drost}16-en-3-one, where the (!) enantiomer is
specialities. Amino compounds and breakdown prod- odourless and the (#) enantiomer is characterized as
ucts of proteins are often associated with bitterness either musk-like or urine-like. Human genetics
and are important when considering the taste of a throws in a further complication with this molecule,
Savouring. in that a third of the population perceive it as odour-
Figure 3 Separation of (R)- and (S )-enantiomers of ethyl-2-methylbutyrate at 9 min.
less, another third as musk-like and the remaining analyst’s arsenal that correlates analytical and sens-
third as urine-like. ory data. The use of a snifRng port as well as a
physical detector to sniff GC efSuents dates back to
Gas Chromatography+Olfactometry the early 1960s. The technique has since improved,
because of technological advances in GC instrumen-
(GC-O) tation, snifRng port design and the use of computer
The most valuable detector in Savour analysis is the tools for data processing. The strength of GC-O lies
human nose. GC-O is the technique in the Savour in the fact that the odorous compounds in a product’s
Figure 4 Separation of the (S )-enantiomer of ethyl-2-methylbutyrate.

extract can be detected from among all the other space analysis carried out on a commercial American
non-odorous compounds in the sample. GC-O roast coffee purchased in a supermarket. Coffee
involves measuring the perceived odour and its reten- aroma is very complex, containing over 800 volatile
tion time. The odour of a molecule can be described compounds. To facilitate explanation, the example
in terms of its aromatic quality, intensity and po- shown is the 30}39 min segment of the chromato-
tency. Intensity pertains to the perception one has at gram. The FID signal is in blue and the electronic
a given concentration, whereas potency refers to push-button signal generated while snifRng is over-
a comparison of concentration amounts with respect laid in red. Flavour descriptors as stated by the ob-
to other molecules at the same intensity level. As for server are shown. The retention indices for the odor-
all GC analysis, identiRcation is helped by using com- ous peaks are indicated at the baseline of the red
pilations and databases of relative retention times. signal. It is interesting to note that many peaks do not
There are a few variations of the GC-O technique: necessarily have an odour and many odorous molecu-
Charm (combined hedonistic response method), les are too low in concentration to have a detector
AEDA (aroma extract dilution analysis) and OSME response. Sulfur-containing molecules (e.g. the peak
(Greek for smell). Charm and AEDA are techniques at retention index 1236) often have extremely low
for measuring potency and OSME quantiRes inten- odour thresholds and never show up on a chromato-
sity. In its simplest form the nose sniffs efSuents from gram unless the sample has undergone some pretreat-
the GC column and the retention times of interesting ment. In these cases, GC-O in combination with a re-
compounds can be noted along with an odour de- tention index database can be very useful. As the nose
scriptor. A snifRng port can be mounted on top of the is generally more sensitive than any physical detector,
GC or in the GC oven door, or can be connected via the retention times of odorous molecules often corres-
a heated transfer line as is the case with the recently pond with the absence of any chromatographic peak.
commercialized Sexible snifRng port. By making it To resolve this, more sample work-up, column liquid
possible to use an electronic push-button device or chromatography and/or MDGC are possible solu-
joystick an electronic signal can be generated that tions. There are some limitations to the technique.
represents the nose response, see Figure 5. Overlaying Unfortunately, human noses are not standardized and
the signals for the nose and for a physical GC detector some suffer from anosmia, the inability to perceive
(often an FID device) results in a chromatogram dis- a certain odour, and also sensitivities are different.
playing the retention times of the odorous molecules Some people are more sensitive to particular odours
on top of the normal FID peaks of the mixture. This than others. Another difRculty is in attributing
result can be termed an ‘aromagram’ and an example a meaningful and consistent Savour descriptor. To
is displayed by Figure 6. This is an example of head- overcome this, subjects can be tested for anosmia and
Figure 5 GC-O instrumentation.

trained in Savour language. Also, it is known that of Savour compounds, and the use of a pyrolysis
some molecules exert a synergistic effect when in the probe mounted on a GC column that is coupled to
Savour but do not show this when eluted separately a mass spectrometer is a pre-requisite for their identi-
on the column. Thus, the odour impact can be differ- Rcation. For Maillard studies, the pyrolysis chamber
ent. Finally, a subject must be very experienced to be is used as a reactor. A limiting parameter of commer-
able to relate the snifRng of individual compounds cial pyrolysis devices is the small sample capacity
back to the organoleptic impact of the entire Savour (generally milligrams). Thus the concentrations of
mixture. What is important is that the subject’s per- volatiles generated are equally low and can limit the
ception is reproducible. The most potent odour mol- analysis to the study of the major compounds formed.
ecules contributing to a Savour can be determined by However, this may yield useful information. Another
performing GC-O with successive dilutions of the way to exploit this technique is by using model mix-
injected sample. GC-O can be made more powerful tures of Savour components to investigate the reac-
for routine Savour analysis by combining it directly tions that control or favour the development of Sa-
with MS detection. Thus retention indices, mass vours from their thermal precursors.
spectra and Savour descriptors can be obtained. The
combination of retention indices with mass spectra
are especially useful when identifying some terpenes
Fast Gas Chromatography
that have very similar mass spectra. Recent advances in GC instrumentation, notably in
electronic pressure control of column and split Sows,
Pyrolysis+Gas Chromatography+Mass faster ovens and the use of narrow bore columns,
have made it possible to increase the speed of analysis
Spectrometry without loss of resolution. Generally, GC runs of
Pyrolysis-GC-MS has found some limited applica- more than 2}3 h are common for determining the
tions in Savour analysis, primarily in the study of quality of an essential oil. It has recently been shown
process or reaction Savours. In fact, the very Rrst use that analyses of nutmeg and lemon essential oils that
of a glass capillary column was on the pyrolysis once took 80 min can now be achieved in less than
products of tobacco, to study cigarette aroma and 20 min. Fast GC has reduced run times considerably
Rltration. The area of most interest has been the study and is consequently very advantageous for routine
of Savour volatiles formed from Maillard reactions quality control laboratories under pressure to achieve
between amino acids and reducing sugars. The Mail- a greater sample throughput per shift. Figures 7 and
lard reaction is responsible for the brown colour and 8 illustrate the fast GC analysis of a lemon oil. The
the taste of bread crusts and meat and is essential in 80 min chromatogram (Figure 7) was obtained on
most savoury Savour systems. The applications a standard 60 m;0.32 mm i.d.;1.2 m RSL-200
studied by pyrolysis generally involve a great variety column (apolar phase). The 8 min chromatogram
Figure 6 An example of an aromagram of an American toast coffee.

Figure 7 Fast GC analysis of a lemon oil (80 min chromatogram).
(Figure 8) was obtained with a column of the same to obtain similar resolutions. As shown, peak elution
phase but with the dimensions 15 m;0.1 mm order was not changed and relative abundances are
i.d.;0.25 m. The analyses were performed in satisfactory in spite of a ten-fold reduction in analysis
constant pressure mode. The hydrogen carrier gas time.
velocity was changed from 27 cm s\1 to 46 cm s\1.
Pressures were 48 kPa and 227 kPa respectively.
Likewise the split ratio was modiRed from 1/25 to
Future Developments
1/1000. In both cases the injection volume was 1 L. Advances in GC instrumentation, column technology
The oven temperature programme was also modiRed and application of computer tools will have positive
Figure 8 Fast GC analysis of a lemon oil (8 min chromatogram).

2822 III / FOAM COUNTERCURRENT CHROMATOGRAPHY
consequences for Savour analysis. There will most Further Reading

likely be a development of larger capacity sampling
Charalambous G, ed. (1978) Analysis of Foods and Beverages
devices using adsorbant tubes and SPME Rbres for } Headspace Techniques. New York: Academic Press.
headspace analysis. The desire to analyse unstable David F, Gere GR, Scanlan F and Sandra P (1999) Instru-
volatiles will lead the analyst to develop derivatiz- mentation and applications of fast high-resolution capil-
ation techniques. GC-O will become better known lary gas chromatography. Journal of Chromatography
and will not be used only for Savour and fragrance A 842: 309}319.
analysis. The evolution of this technique may see the KoK nig WA (1987) The Practice of Enantiomer Separation
development of expert systems, such as voice recogni- by Capillary Gas Chromatography. Heidelberg:
tion software to automatically annotate aromagrams Huethig.
and software to help with interpretation against Kondjoyan N and BerdagueH J-L (1996) A Compilation of
known chemical data and correlation with sensory Relative Retention Indices for the Analysis of Aromatic
Compounds. Saint Genes Chapanelle: Laboratoire
data. The development of Savour databases that
Flaveur INRA.
combine chromatographic and spectroscopic data Marsili R, ed. (1997) Techniques for Analyzing Food
will continue. Increasing the speed of analysis will see Aroma. Food Science & Technology Series 79. New
fast GC develop for Savour quality control analysis. York: Marcel Dekker.
An interesting development in the task of comparing Mussinan CJ and Morello MJ, eds (1998) Flavor Analysis
data from different instruments with different de- } Developments in Isolation and Characterization. ACS
tectors is retention-time (RT) locking. This has al- Symposium Series 705. Washington: American Chem-
ready been successfully applied and will be aided by ical Society.
the development of speciRc RT lock Savour and fra- Nijssen LM, Visscher CA, Maarse H, Willemsens LC and
grance libraries. Boelens MH, eds (1996) Volatile Compounds in Food,
ualitative and Quantitative Data. 7th edn (including
supplement 1, 1997). Zeist: TNO Nutrition and Food
See also: II/Chromatography: Gas: Derivatization; De- Research Institute.
tectors: General (Flame Ionization Detectors and Thermal Pawliszyn J (1997) Solid Phase Microextraction } Theory
Conductivity Detectors); Detectors: Mass Spectrometry; and Practice. New York: Wiley-VCH.
Detectors: Selective; Headspace Gas Chromatography; Sandra P and Bicchi C (1987) Capillary Gas Chromatogra-
Large-Scale Gas Chromatography; Multidimensional phy in Essential Oil Analysis. Heidelberg: Heuthig.
Gas Chromatography; Extraction: Solid-Phase Micro- Schreier P, ed. (1984) Analysis of Volatiles}Methods}
extraction. III/Allergens in Perfumes: Gas Chromato- Applications. Berlin: W. de Gruyter & Co.
graphy-Mass Spectrometry. Chiral Separations: Gas Werkhoff P, Brennecke S, Bretschneider W, et al. Chiros-
Chromatography. Fragrances: Gas Chromatography. peciRc analysis in essential oil, fragrance and Savour
Pheromones: Gas Chromatography. Solid-Phase research. Zeitschrift fu( r Lebensmittel Untersuchung und
Micro-Extraction: Overview. Forschung 196: 307}328.
FOAM COUNTERCURRENT
CHROMATOGRAPHY
H. Oka, Aichi Prefectural Institute of Public Health, afRnity of samples in aqueous solution and it has
Nagoya, Japan a great potential for application to biological sam-
Y. Ito, National Institute of Health, Bethesda, ples. However, the use of this method in research
MD, USA laboratories has been extremely limited, mainly due
Copyright ^ 2000 Academic Press to a lack of efRcient instruments. Foam separation
instruments generally consist of a single tubular col-
umn where the foam is generated by introducing the
Introduction gas phase at the bottom of the column (Figure 1).
Under the gravitational Reld, the foam moves up-
Foam separation methods have long been used for the wards towards the top of the column to collect foam-
separation of various samples ranging from metal active materials. Although various mixing devices
ions to mineral particles. The separation is based on such as bafSes, solid beads and rotary mixers are used
a unique parameter of foaming capacity or foam to improve contact, the use of a short column under
III / FOAM COUNTERCURRENT CHROMATOGRAPHY 2823
frame around the central axis of the centrifuge. The

rotary frame holds a coiled separation column and
a counterweight symmetrically at a distance of 20 cm
from the central axis of the centrifuge. A set of gears
and pulleys produces synchronous planetary motion
of the coiled column. This planetary motion induces
a countercurrent movement between foam and the
mother liquid through a long, narrow, coiled tube.
Introduction of a sample mixture into the coil results
in the separation of sample components. Foam-active
components are quickly carried with the foaming
stream and are collected from one end of the coil
while the rest move with the liquid stream in the
opposite direction and are collected from the other
end of the coil (Figure 2).
The column design for foam CCC is shown in
Figure 4. The coil consists of a 10 m long, 2.6 mm i.d.
TeSon tube with a 50 mL capacity. The column is
equipped with Rve Sow channels. The liquid is
fed from the liquid feed line at the tail and collected
from the liquid collection line at the head. Nitrogen
gas is fed from the gas feed line at the head and
discharged through the foam collection line at the tail
while the sample solution is introduced through the
sample feed line at the middle of the coil. The
head}tail relationship of the rotating coil is conven-
tionally deRned by an Archimedean screw force
where all objects of different density are driven to-
wards the head. Liquid feed rate and sample injec-
tion rate are each separately regulated using needle
valves while the foam collection line is left open to
the air.
Application
Foam CCC can be applied to a variety of samples
having foam afRnity. Foam afRnity can be classiRed
Figure 1 Conventional foam separation column. into two categories: (i) the afRnity to the foam-produ-
cing carrier; and (ii) the direct afRnity to the
gas}liquid interface. Samples which lack direct
a gravitational Reld limits the efRciency of these sys- afRnity to the gas}liquid interface can be in-
tems and consequently, the separation is inefRcient. directly absorbed to the foam if they have an afRnity
In 1976, foam countercurrent chromatography to the foam-producing agents such as a surfactant.
(CCC) was developed to improve the foam separation Samples, such as detergents and other foam-produ-
technology. In this foam CCC method, foam and cing substances, can be separated without special
liquid undergo countercurrent movement through treatment because they have afRnity to the gas}liquid-
a long, Rne, TeSon tube (10 m;2.6 mm i.d.) under interface.
a strong centrifugal force Reld as illustrated in Fig-
ure 2. This article describes the foam CCC tech-
Foam Separation Using Surfactants
nology and its application to a variety of samples.
This technique is applied to the sample having afRnity
to the foam-producing carrier. Sodium dodecyl sul-
Apparatus of Foam CCC fate (SDS) and cetyl pyridinium chloride (CPC) were
Figure 3 illustrates the design of the foam countercur- used as carriers to study the effects of electric charges
rent chromatograph. The motor drives the rotary on the foam afRnity of various compounds. Figure 5
Figure 2 Foam CCC scheme.
illustrates two sets of foam chromatograms obtained for methylene blue. When the sample mixture was
from a mixture of methylene blue and DNP}leucine introduced with the anionic SDS surfactant, the posit-
mixture using SDS (top) and CPC (bottom) as carrier ively charged methylene blue was adsorbed on to the
reagents. In each chromatogram, the ordinate indi- foam and quickly eluted through the foam collection
cates absorbance values measured at two line (top, right) while the negatively charged
wavelengths, 430 nm for DNP}leucine and 620 nm DNP}leucine was carried with the liquid stream in
Figure 3 Design of foam CCC centrifuge. (Reproduced from Ito (1985) with permission from Marcel Dekker Inc.)
Figure 4 Column design for foam CCC.
the opposite direction and eluted through the liquid was totally eluted through the foam collection
collection line (top, left). Similarly, when the same line (bottom, right) and positively charged methyl-
sample mixture was eluted with the cationic CPC ene blue through the liquid collection line (bottom,
surfactant, the negatively charged DNP}leucine left).
Figure 5 Separation of methylene blue and DNP}leucine by foam CCC. (Reproduced from Bhatnagar and Ito (1988) with permission
from Marcel Dekker Inc.)
Foam Separation Without Surfactants

Using ionic surfactant as carriers, the samples which
lack direct afRnity to the gas}liquid phase interface
can be separated if they have opposite electric
charges. However, in this case complicated proced-
ures are required to remove the surfactants after frac-
tionation. On the other hand, many natural products
have foaming capacity, and therefore foam CCC may
be performed without such surfactants.
In order to demonstrate this possibility, bacitracin
complex (BC) was selected as a test sample, because it
has a strong foaming capacity. Foam CCC for separ-
ation and enrichment of BC components has been
conducted using nitrogen gas and distilled water en-
tirely free of surfactant or other additives. The fol-
lowing sections describe chromatographic fractiona-
tion of BC with batch sample loading and enrichment
of foam-active compounds from a bulk liquid on
continuous sample feeding.
Batch sample loading BC is a basic cyclic peptide

antibiotic commonly used as a feed additive for live- Figure 6 Elution curve of bacitracin components from foam
stock worldwide. It consists of more than 20 compo- line. Foam CCC conditions: liquid flow rate, 3.2 mL min\1; needle
nents, but chemical structures of these components valve, 0.8 turn open; standing time ater sample injection, 5 min;
are still unknown except for BCs-A and -F. N2 gas pressure, 80 psi; revolution speed, 500 rpm; sample size,
5 mg per 0.5 mL in H2O; fractionation, 15 s per tube.
Foam separation of BC components was initiated
by simultaneous introduction of distilled water from
the tail and nitrogen gas from the head into the longest retention time in HPLC analysis were col-
rotating column while the needle valve at the liquid lected in the Rrst foam fraction with a small amount
collection line was fully opened. After a steady-state of less hydrophobic compounds (peaks 11 and 13).
hydrodynamic equilibrium was reached, the pump Peak 15 is hardly visible in the HPLC chromatogram
was stopped and the sample solution was injected of the original sample due to its low concentration,
through the sample port. After the desired standing but the same peak is clearly observed in the
time, the needle valve opening was adjusted to the chromatogram of the Rrst foam fraction. BC-A was
desired level and pumping was started again. EfSu- almost completely isolated in peak 11 from other
ents from both outlets were collected at 15 second components eluted in the tenth fraction. In the twen-
intervals. tieth fraction, peak 7 appeared in the HPLC
Figure 6 shows the elution curve of BC compo- chromatogram. Components with lower hydropho-
nents from the foam outlet. The vertical axis indicates bicity than peak 7 did not appear in the foam frac-
the absorbance at 234 nm and the horizontal line, the tions. These results clearly indicate that the bacitracin
fraction number. This elution curve shows three ma- components are separated in the order of hydropho-
jor peaks as indicated by arrows. The fractions corre- bicity of the molecule in the foam fractions with the
sponding to these peaks were subjected to HPLC most hydrophobic compounds being eluted Rrst.
analysis. As described above, BC components were separa-
HPLC chromatograms of BC components in the ted according to their hydrophobicity using foam CCC
foam fractions are shown in Figure 7. Under rever- without surfactant. This method can also be applied
sed-phase HPLC conditions, BC was separated into to continuous sample feeding as described below.
more than Rfteen peaks. Generally, hydrophilic com-
pounds elute earlier than hydrophobic compounds Continuous sample feeding On the basis of the pre-
under these conditions. In this study, the HPLC elu- liminary experimental results, the following foam CCC
tion time is used to indicate the polarity of the BC conditions were chosen for large scale sample feeding.
components. Thus, BC-A which elutes earlier in The conditions were as follows: needle valve, 2.0 turns
HPLC is more hydrophilic than BC-F. The most hy- open; sample concentration, 50 p.p.m.; sample size,
drophobic compounds (peaks 14 and 15) with the 2.5 L; sample feed rate, 1.5 mL min\1 at 40 psi; ni-
Figure 7 High performance liquid chromatographic analyses of bacitracin components in foam fractions. HPLC conditions: column,
Capcell Pak C18 (5 m, 150;4.6 mm, i.d.); mobile phase, CH3OH/0.04 M Na2HPO4 (62/38); flow rate: 1 mL min\1; detection, 234 nm.
trogen gas feed pressure, 80 psi; liquid Sow rate, 0; immediately mixed with the N2 stream and the gener-
sample collection, pooling the foam and the liquid ated foam moves towards the foam outlet at the tail.
efSuents separately; and revolution speed, 500 rpm. Since the coiled column consists of a 10-m-long tube,
Figure 8 shows the results of HPLC analyses of the foam must travel through a 5-m-long narrow
bacitracin in the foam and liquid fractions obtained coiled path before it reaches the foam outlet. As the
by large scale continuous foam CCC. The concentra- foam travels through the decreasing pressure gradient
tion in the foam fraction increases with the hydro- along the coil, every bubble is expanded while the
phobicity of the components. Peak 3 was enriched 22 excess Suid is removed by the centrifugal force.
times; peak 7, 31 times; peak 11, 1400 times; peak Therefore, we assume that the foam must be sub-
12, 1070 times; peak 13, 1380 times; and peak 14, jected to a repetitive process of coalescence, eruption
2260 times. In the liquid fraction, peaks 3 and 7 were and regeneration before reaching the foam outlet at
barely detected. Thus, continuous enrichment and the tail. Consequently, successful foam CCC using
concentration in foam CCC is quite effective for the nitrogen gas and distilled water free of surfactant
detection and isolation of a small amount of natural requires strong foam-producing capability and foam
product with a foaming capacity. stability of analytes. A lack of either property would
result in its failure.
Estimation of applicability of the sample to foam For this purpose, two parameters were selected, i.e.
CCC In order to apply the foam CCC technique to ‘foaming power’ and ‘foam stability’, which can be
various natural products, it is necessary to establish determined by one simple test. In each test, the
a set of physicochemical parameters which reliably sample solution (20 mL) is delivered into a 100 mL
indicate their suitability for foam CCC. In foam CCC graduated cylinder with a ground stopper and the
the sample solution is introduced from the sample cylinder vigorously shaken for 10 s. The foaming
inlet at the middle of the coiled column where it is power is expressed by the volume ratio of the result-
Figure 8 High performance liquid chromatographic analyses of bacitracin components in foam and liquid fractions. HPLC conditions:
column, Capcell Pak C18 (5 m, 150;4.6 mm, i.d.); mobile phase, CH3OH/0.04 M Na2HPO4 (62/38); flow rate, 1 mL min\1; detection,
234 nm.
ing foam to the remaining solution, and the foam enriched by foam CCC. These minimum require-
stability by the duration of the foam. ments of foaming parameters tentatively determined
In order to correlate the foaming parameters mea- by the bacitracin experiment were found to be consis-
sured by this simple test method to the productivity in tent with those obtained from the other four samples.
foam CCC, the following Rve samples were selected The above simple test has been applied to enrich-
because of their strong foaming capacities: bacitracin, ment of microcystins, hepatotoxic cyclic peptides
gardenia yellow, rose bengal, phloxine B, and senega produced by cyanobacteria. Microcystins was extrac-
methanol extract. The results of our studies indicated ted from the bloom sample 917S with distilled water
that a sample having a foaming power greater than 1.0 to obtain two extracts with different foaming capaci-
and a foam stability over 250 min could be effectively ties. The Rrst extract had foaming power of 1.88 and
Figure 9 High performance liquid chromatographic analyses of bloom sample 917S extract that satisfies the foaming parameters:
A, original sample; B, foam fraction; C, liquid fraction.
III / FOOD ADDITIVES / Liquid Chromatography 2829
foam stability of 93 min which satisRed only the for- See also: II/Chromatography: Liquid: Countercurrent
mer requirement. The second extract had a foaming Liquid Chromatography. III/Antibiotics: High-Speed
power of 1.32 and the foam stability of 720 min Countercurrent Chromatography; Liquid Chromatography;
which satisRed both requirements for foam CCC. Supercritical Fluid Chromatography.
Then, both samples were subjected to foam CCC.
Figure 9A shows a typical HPLC chromatogram of Further Reading
an extract from cyanobacteria bloom sample 917S
Bhatnagar M and Ito Y (1988) Foam countercurrent
containing microcystins. As indicated in the chromatography on various test samples and the effects
chromatogram, peaks 1 (microcystin RR), 2 (micro- of additives on foam afRnity. Journal of Liquid
cystin YR), 3 (microcystin LR), and 4 (microcystin Chromatography 11: 21.
LR-s) were chosen to evaluate their foam enrichment. Ito Y (1976) Foam countercurrent chromatography: New
In the Rrst extract, the enriched concentrations of the foam separation technique with Sow-through coil planet
components are only 3}4 times and polar compo- centrifuge. Separation Science 11: 201.
nents with retention times shorter than that of micro- Ito Y (1985) Foam countercurrent chromatography based
cystin RR were still present in the foam fraction. The on dual countercurrent system. Journal of Liquid
HPLC analysis of foam fraction and liquid fraction of Chromatography 8: 2131.
the second extract is shown in Figures 9B and C. The Ito Y (1987) Foam countercurrent chromatography with
the cross-axis synchronous Sow-through coil planet cen-
enrichment reached 10}30 times and polar compo-
trifuge. Journal of Chromatography 403: 77.
nents are eliminated from the foam fraction indicat- Oka H (1996) Foam countercurrent chromatography of
ing that the target compounds are selectively en- bacitracin complex. In Ito Y and Conway WD (eds)
riched. The HPLC analysis of liquid fraction of both High-Speed Countercurrent Chromatography, pp.
extracts showed similar proRles. These results indi- 107}120. New York: Wiley.
cate that these foaming parameters can be effectively Oka H, Harada K-I, Suzuki M, Nakazawa H and Ito
applied to a crude mixture containing a large amount Y (1989) Foam countercurrent chromatography of ba-
of impurities. citracin with nitrogen and additive-free water. Analyti-
cal Chemistry 61: 1998.
Oka H, Harada K-I, Suzuki M, Nakazawa H and Ito
Conclusions Y (1989) Foam countercurrent chromatography of ba-
citracin I. Batch separation with nitrogen and water free
Foam CCC can be successfully applied to a variety of of additives. Journal of Chromatography 482: 197.
samples having foam afRnity with or without surfac- Oka H, Harada K-I, Suzuki M, Nakazawa H and Ito
tants. The present method offers important advant- Y (1991) Foam countercurrent chromatography of ba-
ages over the conventional foam separation methods citracin II. Continuous removal and concentration of
by allowing efRcient chromatographic separation of hydrophobic components with nitrogen gas and distilled
water free of surfactants or other additives. Journal of
sample in both batch loading and continuous feeding.
Chromatography 538: 213.
We believe that the foam CCC technique has a great Oka H, Iwaya M, Harada K-I, Muarata H, Suzuki M, Ikai
potential in enrichment, stripping and isolation of Y, Hayakawa J and Ito Y (1997) Effect of foaming
foam-active components from various natural and power and foam stability on continuous concentration
synthetic products in both research laboratories and with foam countercurrent chromatography. Journal of
industrial plants. Chromatography A 791: 53.
FOOD ADDITIVES
Introduction
Liquid Chromatography Food is a complex heterogeneous mixture of a wide
range of chemical constituents such as moisture,
V. D. Sattigeri, L. R. Gowda and carbohydrates, proteins, Rbres, vitamins, etc. Besides
P. R. Ramasarma, Central Food Technological these, processed foods contain a wide array of addi-
Research Institute, Mysore, India tives and contaminants. Analysis of product composi-
Copyright ^ 2000 Academic Press tion is a prerequisite for ascertaining product quality,
2830 III / FOOD ADDITIVES / Liquid Chromatography
implementing regulatory enforcements, checking com- On account of the wide range of different classes
pliance with national and international food stan- of chemical compounds and matrices encountered,
dards, contracting speciRcations and nutrient labelling it is not possible to give a universally applicable
requirements and providing quality assurance for use analytical scheme for food additives. Figure 1 illus-
of the product for the supplementation of other foods. trates the various steps involved in the analysis of
Food preservatives form an important class of food antioxidants in potato chips by HPLC as a typical
additives. They are primarily used to prevent microbi- example.
al growth, to improve or maintain the nutritional
Preservatives
value of food, to maintain palatability and whole-
someness and to enchance Savour and colour. Other Preservatives are chemicals added to food products to
food additives used include colours, colour modiRers, prevent or inhibit the growth of microbes. Benzoic
Savours, Savour enhancers, humectants, non-nutri- acid, sorbic acid, propionic acid and methyl-, ethyl-
tive sweeteners, pH control agents, thickeners, stabi- and propyl-esters of p-hydroxybenzoic acid (para-
lizers and emulsiRers. Food additives are regulated bens) are the most commonly used preservatives.
and speciRed by law in most countries to ensure safety Preliminary extraction from the food matrix before
for the consumer and prevent deception and fraudu- analysis is required. Steam distillation, solvent and
lent practices. Labelling regulations require that in- solid phase extractions are the most commonly used
formation be provided on the kind of food, its pro- methods. Hild and Gertz have reviewed the analytical
cessing and the additives contained in it. methods available for the quantitative determination
The complex heterogeneous nature of foods de- of preservatives in food.
mands effective separation techniques such as High For the HPLC determination of benzoates and
Performance Liquid Chromatography (HPLC) with sorbates, many methods have been reported using
its wide array of column materials, and detectors. either isocratic or gradient techniques with RP-
Food additives are usually present in small quantities columns and UV detection at ambient temperatures.
in processed food items. Their separation from food LC has been used to separate the homologous esters
constituents therefore requires a thorough under- of p-hydroxybenzoic acid (parabens) including
standing of the chemistry and physics of the food methyl- and propyl-hydroxybenzoates. This is of
constituents and additives in order to select the best speciRc importance as the homologous esters of
analytical procedures. Increased automation has p-hydroxybenzoic acid (parabens) are a group of sim-
gained universal acceptance for the effective separ- ilar compounds having closely related properties. The
ation and analysis of nearly all food components and method described can efRciently determine parabens
food additives. along with BHA, TBHQ, PG, NDGA and Ionox-100
on reverse phase systems with electrochemical detec-
tion. The typical linear range extends from 10\11 to
Typical HPLC Analytical Systems 10\6 mole of injected analyte. Recovery of parabens
In HPLC analysis of food additives, a single solvent is about 80%. Detector potential for parabens is
(or solvent mixture) is often not sufRcient to carry #1.10 V and the electrochemical detector provides
out the separation under isocratic conditions. Hence, low limits of detectability.
solvent systems of varying proportions are generally Normal as well as reversed phase methods have
used for gradient elution. Abundant literature is been used to determine the esters of p-hydroxyben-
available on convenient, versatile and precise liquid zoic acid by extraction with acetonitrile. For normal
chromatography (LC) separations of complex food phase HPLC, the use of LiChrosorb Si 60 columns
constituents and additives. Problem areas such as with a mobile phase of iso-octane#diethyl ether#
band tailing, trace analysis, preparative separations acetonitrile (500#35#0.3) and for RP HPLC, RP-
and so on have received considerable attention and 18 columns with methanol}water (80 : 20) and UV
these particular problems can now be tackled in rela- detection have been suggested. Recoveries are
tively systematic and simple ways. 95}104% with 1}2% RSD. A method for the analysis
LC is ideally suited for the separation of macro- of parabens in meat products has been developed in
molecules and ionic species, heat-labile natural prod- which the samples are extracted with acetonitrile,
ucts and high molecular weight or less stable com- Rltered and analysed on a C18 column. The peaks are
pounds such as proteins, nucleic acids, amino acids, detected and quantitated using UV detector at
dyes, synthetic polymers and food additives. 254 nm. Average recoveries are 92% for methyl para-
HPLC is well suited for the quantitative determina- ben and 94% for propyl paraben.
tion of food additives in one step and it has largely HPLC has been used to determine sorbic acid,
replaced other analytical methods. benzoic acid and p-hydroxybenzoic acid (PHB) esters
Figure 1 Stages of typical HPLC analysis of antioxidants in potato chips.
in foods. A mixture of acetonitrile, 2-propanol, Gradient elution methods have been compared for
ethanol and oxalic acid was used for extraction. After the simultaneous determination of benzoic acid,
refrigerating and separating the interfering materials sorbic acid and parabens in ground beef, non-meat
by centrifugation, the extract was analysed without products and pork sausages. These preservatives were
further cleanup, using Spherisorb ODS II (3 m) and extracted with 70% ethanol and analysed on
methanol}water}phosphoric acid}tetrahydrofuran as a Novapak C-18 column with a linear gradient mo-
eluant and detection at 230 nm and 245 nm. The bile phase consisting of 10}70% methanol in 1.5%
detection limits were 0.5 mg/kg, 2 mg/kg and aqueous ammonium acetate and 1.5% aqueous acetic
10 ng/kg for sorbic acid, benzoic acid and esters of acid over 10 minutes, with 10 minute hold. Recove-
p-hydroxybenzoic acid, respectively. ries of benzoic acid, sorbic acid, methyl-, ethyl-, and
Using C-18 silica with methanol and phosphate propyl parabens range from 99}103%. Seven pre-
buffer (1 : 9, v/v) a lower limit of detections ranging servatives have been simultaneously measured in 28
from 5 mg/kg to 1 mg/kg can be obtained for benzoic food samples. Better selectivity and sensitivity for
and sorbic esters of PHB. HPLC compared to other procedures have been re-
The sorbate content of commercial yoghurt sample ported.
following ion-pair extraction of sorbic acid and A method applicable to many liquid and solid
benzoic acid in tri-n-octylamine has been reported. foods has been described. A C-18 column is used with
It uses a RP-18 column with methanol}phosphate methanol}phosphate buffer (5 : 95, v/v) as a mobile
buffer (40 : 60, pH 4.5, ionic strength 0.1). Mean phase. This work was carried out to check the speciR-
recoveries are 70}88% with a precision of 1.1 to city of the isocratic-LC method for common food
3.3% RSD. Isocratic HPLC is suitable for the deter- additivies such as L-ascorbic acid, caffeine, artiRcial
mination of the benzoic and sorbic acid in beer. sweeteners, antioxidants and synthetic colours. The
Steam distillation and direct extraction as sample method is applicable for determining benzoic acid
pretreatment methods for analysis of benzoic acid and sorbic acid in a wide variety of foods such as
and sorbic acids in salad dressing mayonnaise have beverages, fruits, seafoods, vegetables, sauces, dairy
been compared. Benzoic acid in chilli sauce can be products, bakery products and confectionery prod-
determined by using RP-HPLC with detection at ucts. 4-Hydroxyacetanilide is used as internal stan-
254 nm. dard and detection is at 227 nm. Mean recoveries of
90}105% with a precision of 1}6% and detection stationary phase and 6 mM H2SO4 as mobile phase
limit of 20 mg/kg have been achieved. RP-HPLC for with electrochemcial detectors. Average recoveries
quantitative and simultaneous determination of were 81}103% with standard deviation of 4.6%.
benzoic acid, sorbic acid, PHB, salicylic acid, 5-nitro- Other food products analysed by different workers by
furylacrylic acid, p-chlorobenzoic acid and PHB es- this method include apple, avocado mix, broccoli,
ters in wines and beverages has also been reported. cabbage, ketchup, grape juice, mushrooms, onions
SulRtes and sulRting agents permitted in the food and raisins.
industry include sodium sulRte, sodium hydrogen sul- Table 1 gives details of some methods for the anal-
Rte, sodium metabisulRte, potassium metabisulRte, ysis of preservatives in foods and food products.
calcium sulRte and potassium hydrogen sulRte. In the
1980s, an ion-pair method was introduced for the Antioxidants
determination of sulRtes in fruit juices, dried bread, During storage, oils and fats undergo various reac-
salad dressing, ground beef, liquid caramel, fruit tions that reduce their nutritive value and also pro-
cake, apple pie, raisin, apple juice, apple sauce, dried duce volatile compounds, to give unpleasant smells
onions and white wine using high pressure ion-ex- and tastes; the phenomenon is referred to as rancid-
change (I-E) polystyrene with }NH# 3 and }NR# 3 ity. In many cases the presence of antioxidants can
groups as stationary phase and 0.0024 M Na2CO3 inhibit the onset of rancidity.
and 0.003 M Na2HCO3 as mobile phase with con- Synthetic antioxidants permitted to be added to
ductivity detector. It has a very good detection limit food are:
of 1 ppm and an analysis time of 25 minutes. The
results were not affected by the presence of volatile BHA } 1(or 3)-(t-butyl)-4-hydroxy anisole
acids or even organic sulfur compounds. Average PG } Propyl gallate
recoveries of 98.3% with standard deviation of 1.88 TBHQ } t-Butyl hydroquinone
were reported with a detection limit of 1 p.p.m. NDGA } Nor dihydroguaiaretic acid
Electrochemical detectors are now preferred for OG } Octyl gallate
sulRte analysis due to their better sensitivity. DG } Dodecyl gallate
SulRtes have been analysed in lemon juice, beer, (BHT } Butylated hydroxytoluene is allowed in
mashed potato and white wines using anion exchange a few countries)
Table 1 Details of some methods for the analysis of preservatives in food and food products
Food Products Preservative analysed Analytical details
Stationary phase/ Mobile phase Detection system

column
Meat samples Parabens -Bondapak C-18 45% acetonitrile in UV at 254 nm

water
Yoghurt Sorbic acid, RP-18 Methanol}phosphate UV at 254 nm
benzoic acid buffer (pH 4.5, ionic
strength 0.1)
(40 : 60)
Fruit ready to serve, Benzoic acid, sorbic acid, Novapak C-18 Gradient elution of UV at 254 nm
beverages, jams, parabens 10}79% methanol in
jellies, meat 1.5% aqueous
products ammonium acetate
and 1.5% aqueous
acetic acid over
10 min
Beverages, fruits, Benzoic acid, sorbic acid C-18 Methanol : phosphate UV at 227 nm
seafoods, buffer (0.03 M, pH
vegetables, sauces, 6.5) (5#95)
dairy, bakery and
confectionery
products
Salad dressing, Sulfites (free and total) I-E polystyrene column 0.003 M NaHO3, Dionex conductivity
caramel, fruit cake, 0.024 M, Na2CO3, in detector
apple pie, raisins, pure water
onions, sauce, white
wine, fruit cocktail
Satisfactory and complete extraction of anti- dapak C-18 column with a mobile phase of acetonit-
oxidants from a food matrix into various organic rile}water (55 : 45, v/v). Antioxidants and anti-
solvents is not always easy because of co-extraction microbials (Parabens) have been analysed in a variety
of interfering substances. Antioxidants such as BHA, of commercial products, such as cereals, snacks and
BHT, TBHQ and Ionox-100 are susceptible to losses shortenings using amperometric detection. The typi-
due to evaporation and utmost care needs to be exer- cal linear range is from 10\11 to 10\6 mole of injected
cised during concentration under vacuum. NDGA, analyte.
PG, OG and DG are relatively polar nonvolatile Seven antioxidants have been determined using
compounds and their recovery is usually satisfactory. a linear gradient from 30% solution B (acetonit-
HPLC produces good separation between chemically rile}acetic acid 95 : 5, v/v) in solution A (water}acetic
similar compounds in mixtures to be analysed and acid 95 : 5, v/v) to 100% solution B over 10 minutes
enables the determination of up to 15 different anti- with detection at 280 nm. Fifteen antioxidants have
oxidants in one single run. been measured in dried foods as well as fats and oils.
The general analysis protocols for antioxidants in The antioxidants were separated by isocratic elution
foods comprise extraction in solvents and determina- with Suorescence and UV detection. Recoveries
tion by reversed phase HPLC. The best solvents for ranged from 80}106.7%.
extracting antioxidants from fats are acetonitrile and The antioxidants diphenylamine and ethoxyquin
water-alcohol mixtures. The fat is usually dissolved in were estimated using methanol}0.01 M ammonium
hexane or petroleum ether and the antioxidant is then acetate (60 : 30, v/v) with Suorescence and UV detec-
extracted into the polar solvent. The literature indi- tion. This method has been used successfully for the
cates the use of a variety of chromatographic proced- separation of fungicide residues and antioxidants in
ures with UV detection at 280 nm as most commonly fresh fruits. BHA, BHT, PG, OG, DG and TBHQ in
used. Mobile phases are acetonitrile, acetic acid, corn oil, cottonseed oil and beef fat have been deter-
methanol and water. mined. A procedure for the determination of anti-
A HPLC method for the simultaneous determina- oxidants in vegetable oils without prior extraction
tion of phenolic antioxidants in vegetable oils, did not resolve BHT from neutral lipids and suffered
lard and shortening has been reported. It was con- from interference due to co-eluting materials. PG,
cluded that nine antioxidants, viz, BHA, TBHQ, trihydroxybutyrophenone, TBHQ, BHA, BHT,
IONOX-100 and THBP, PG, OG, DH and NDGA in NDGA and 3,5-di-tert-butyl-4-hydroxy-methyl-
vegetable oils, lards and shortening could be separ- phenol have been determined in fats, oils and dry
ated by gradient elution with water}acetonitrile foods. Antioxidants in dried foods such as potato
plus 5% acetic acid as mobile phase. The recoveries Sakes, dry coffee, whiteners and dessert topping
ranged from 96 to 103%. A rapid and speciRc mixes were isolated after rehydration and extraction
HPLC method for analysis of TBHQ in vegetable in acetonitrile and subsequent separation on a C-18
oils is also documented. A HPLC method was investi- column. The overall recoveries ranged from 64.3 to
gated with amperometric detection to analyse BHA, 103.6%. The method is highly accurate and hence
BHT and TBHQ in edible oils. The antioxidants was adopted as an ofRcial method (AOAC).
were well separated, identiRed and quantiRed Tocopherols in vegetable oils have been separated
with high sensitivity. Recoveries ranged from 98 to by both reversed-phase and normal-phase LC. A
101%. method using a Radial PAK cartridge has been used
The use of RP-HPLC to quantitatively determine for analysing individual tocopherols in eleven sam-
Rve antioxidants } BHA, BHT, PG, OG and DG } in ples of lupine oil. The results showed the presence of
fats has been described. HPLC enables the determina- -tocopherol (42}69 mg/100 g oil), and -tocopherol
tion of the full range of antioxidants from polar in traces (0.1}0.7 mg/100 g oil). This method is su-
compounds to the non-polar substances in a perior to GC in which up to 30% tocopherol losses
single chromatogram using gradient elution. Sensi- occur during pretreatment of the sample. The simul-
tive detection wavelengths are at 280 nm for UV taneous determination of -tocopheryl acetate, toc-
and at 315 nm for Suorescence emission measure- opherols and tocotrienols in food involving extrac-
ments. tion in hexane, separation on Lichrosorb Si-60 with
Amperometric detection, which is both sensitive hexane}di-isopropyl ether (93 : 3) as mobile phase
and speciRc has been used. Determination of BHA and Suorescence detection at 290 nm, 330 nm, has
and BHT in chewing gums after extraction in hexane been reported. Recoveries are 95}100% with a detec-
and with a second extraction into dimethyl sulfoxide tion limit of 420 ng.
has been reported. The resulting extract was acidiRed Most methods for the analysis of antioxidants use
with hydrochloric acid and separated on a -Bon- C18 columns with detection at 280 nm. However,
electrochemical or Suorimetric detection or simulta- 18 column with acetic acid}water}acetonitrile as mo-

neous detection by two or more techniques has also bile phase and UV detector at 280 nm has also been
been used. Mobile phases are usually composed of documented.
aqueous acid (acetic/phosphoric acid), buffers or salts Murakita (1992) and Klein & Leubolt (1993) have
together with methanol or acetonitrile. In many cases reviewed the analysis of various antioxidants by
results are improved by gradient elution. HPLC.
A method using a C-18 column for -tocopheryl Table 2 shows details of major liquid chromato-
acetate and tocopherols has been described which graphic methods for the analysis of antioxidants in
allows separation of nine synthetic phenolic anti- foods and food products.
oxidants along with natural antioxidants. Gradient
Non-Nutritive Sweeteners
elution is with water-acetonitrile-methanol-iso-
propanol. This method not only allows simultaneous Saccharin, cyclamates, aspartame, acesulfame-K are
detection of antioxidants and triglycerides but is also some of the widely used non-nutritive sweeteners.
useful in studying inhibition effects of antioxidants in Soft drinks containing saccharin are readily ana-
oil. lysed with minimal sample treatment. For juice,
BHA, BHT, TBHQ, NDGA and gallates have been sweets, jams or desserts, an additional extraction step
resolved and quantitatively determined on a Lich- has to be performed. A method for the separation and
rosorb RP-18 column with gradient elution using detection of saccharin, sodium benzoate and caffeine
acetonitrile}water}phosphoric acid and detection at has been reported involving the use of 5% acetic acid
280 nm. Fluorimetric detection can also be used. The as mobile phase and UV detetion at 254 nm. The
analyisis of BHA, BHT, TBHQ and gallates in carrot resolution factor was '2.0 between saccharin and
juice, powdered milk, appetizers and cake using elec- sodium benzoate and between benzoate and caffeine.
trochemical detection has also been reported. It was The detection limits were 0.14, 0.05 and 0.024 g for
suggested that as many as twelve antioxidants could saccharin, benzoate and caffeine respectively. Analy-
be detected by a single isocratic HPLC analysis. The sis of non-artiRcially sweetened soft drinks gave no
quantitation of BHA, BHT, TBHQ, NDGA, gallates interfering peaks with these additives. This method
and other antioxidants in foods using Supelcosil LC- has been adopted by the AOAC because of its accu-
Table 2 Liquid chromatographic methods for the analysis of antioxidants in foods and food products
Food products Antioxidant analysed Analytical details

column
Potato flakes BHA, BHT C-18 Reversed phase gradient UV, 280 nm
elution by acetonitrile
with 5% acetic acid and
5% acetic acid in water
Coffee whiteners TBHQ, BHA C-18 Reversed phase gradient UV, 280 nm
elution by acetonitrile
Dessert topping PG, C-18 Reversed phase gradient UV, 280 nm
DG, OG mixes elution by acetonitrile
with 5% accetic acid and
Cheese, snacks, cake BHA, BHT, TBHQ, PG, C-18 Reversed phase gradient UV, 280 nm
mix OD, DG elution by acetonitrile
Oils, lards, shortenings BHA, BHT, TBHQ, THBP, C-18 Reversed phase gradient UV, 280 nm
Ionox-100, NDGA elution.
Water-acetonitrile with
5% acetic acid
Instant cereals, snacks, BHA, TBHQ, PG, NDGA -Bondapak C-18 Methanol}0.1 M Amperometric
gelatin desserts, and Parabens ammonium acetate (or detection
hydrogenated fats 0.01 M phosphate) buffer
(1 : 1, v/v)
racy. An isocratic HPLC method using a cation ex- reported. Analysis of acesulfame-K, alitame, aspar-
change column, a 0.1 M ammonium dihydrogen- tame, caffeine, sorbic acid, theobromine, theophyl-
phosphate mobile phase and UV detection at 214 nm line and vanillin in table-top sweeteners, candy,
has been reported for the detection of saccharin, liquid beverages and other foods using a -Bondapak
aspartame, benzoic acid and caffeine in soft drinks. C-18, column and a mobile phase of acetonit-
Base-line separations of these four additives were rile}0.0125 M potassium dihydrogen phosphate
achieved. Changing the wavelength of detection from (10 : 90, v/v) at pH 3.5 and UV detection at 220 nm
254 nm to 214 nm led to an increase in the detection has been advocated. This method allows for the si-
response of aspartame. At all levels of addition the multaneous determination of theobromine, theophyl-
recovery for aspartame was 100%. Analysis time line, caffeine, vanillin, dulcin, sorbic acid, saccharin,
could be reduced by increasing the Sow rates without alitame, aspartame and their degradation products in
sacriRcing resolution. a single run of 60 min duration. Table-top sweetener,
A gradient method for the separation of saccharin, candy, soft drink, fruit juice, fruit nectar, yoghurt,
aspartame, benzoic acid and some colours in soft cream, custard, chocolate and biscuits have been ana-
drinks using a detection wavelength of 214 nm has lysed by simple extraction or by just dilution using
been reported. The mobile phase was methanol (10% this method.
increasing to 60%) with 50 mM phosphate buffer at Some of the simpler LC methods for sweetener
pH 3.6. For aspartame, either isocratic or gradient analysis are given in Table 3.
elution was used. A method for the determination of
aspartame, cyclamate, dulcin and saccharin using an
Food Colours
ion-pair separation with indirect photometric detec-
tion has also been reported. A method for determin- Colour is a prime sensory quality by which foods are
ing acesulfame-K using UV detection at 237 nm and judged and food quality and Savour are closely
a mobile phase of water}methanol (9 : 1, v/v) contain- associated with colour. Consumers are conditioned
ing 10 mM tetrabutylammonium hydrogensulfate to expect foods of certain colours and to reject
has been reported. The absence of appreciable any deviation from these expectations. Colourants
absorption above 200 nm by cyclamate has led to the also play a signiRcant role in enhancing the aes-
advent of special methods for its detection. Post- thetic appeal of food. Colourants are very impor-
column ion-pairing of cyclamate with either methyl tant ingredients in many convenience foods such
violet or crystal violet renders it easily detectable. as confectionery products, desserts, snacks and
Pre-column derivatization agents used are sodium beverages.
hypochlorite or o-phthalaldehyde. An ion pair HPLC The regulatory status of colourants used in differ-
method with indirect photometric detection of cycla- ent countries throughout the world is in a constant
mate has been used for thick yoghurt samples and state of Sux due to the toxicological considerations.
solid foods such as biscuits. An extensive review of genotoxicity of food, drug and
A method has been described for the detection cosmetic colours and other azo triphenylmethane and
of acesulfame-K, saccharin, dulcin, benzoic acid, xanthan dyes has been published by Combes and
caffeine and vanillin in ready-to-serve beverages, Haveland-Smith (1982).
dry beverage mix samples and other food products. Synthetic colours can be classiRed by their chem-
The separation was carried out on a -Bondapak ical structure as azo (mono, di and tris),
C-18 column using methanol}acetic acid}water indol, triphenylmethane and methin dyes. They
(35 : 5 : 60, v/v/v) as mobile phase and with UV de- are mostly acidic or anionic and acidic groups
tection at 254 nm. This method is advantageous be- like sulfuric acid, carboxylic acid or hydroxy
cause all the additives can be detected in a single step, groups form a negatively charged coloured ion.
which renders it useful in routine food analysis. Basic or cationic dyes contain substituted amino
Biemer analysed acesulfame-K in candy gum using groups.
an anion exchange column, sodium carbonate The dyes have to be Rrst extracted from the com-
(300 mg/L) as mobile phase and conductometric de- plex food matrix; adsorbents like wool Rbres, pow-
tection. dered polyamide, cellulose ion exchange resins or RP
A method for the determination of aspartame, sac- cartridges (Sep-Pak C18) are frequently used. Ion-
charin, benzoic acid, sorbic acid and caffeine in cola pair chromatography has been used for the quantitat-
drinks, table-top sweeteners, soft drinks and complex ive analysis of twelve primary food colours in grape
foods on a LiChrosorb C-18, column using acetonit- beverages with a mobile phase of 45 : 55 methanol}
rile}0.1 M sodium dihydrogenphosphate (15 : 85, water, the useful detection wavelengths were 610 nm
v/v) at pH 4.5 and UV detection at 215 nm has been for blues and greens and 480 nm for reds, oranges
Table 3 Simpler methods for the analysis of food products for non-nutritive sweeteners
Food products Sweetener analysed Analytical details

column
Coffee, carbonated Sucralose RadialPak C-18 Water}methanol 70 : 30 RI

cola, lemon
beverages
Fruit drinks, cherry Cyclamate Nucleosil C-18 Methanol}water 80 : 20 UV, 313 nm
nectar, mayonnaise,
chocolate
Fruit juice, yoghurt, Acesulfam-K Lichrosorb RP-18 Methanol}water 9 : 1 UV, 237 nm
Cola with 1 mM tetra butyl
ammonium hydrogen
sulfate
Cola, pudding, Saccharin, cyclamate, Bondapak C-18 or Phosphate buffer RI, UV 200 nm
chocolate Alitame Supelcosil LC-18 20 mM (pH 3.5) :
acetonitrile 97 : 3
Ready-to-drink and dry Saccharin, Bondapak C-18 Methanol}acetic acid} UV, 254 nm
mixes of beverages, Acesulfam-K water, 35 : 5 : 60
tomato sauce
Candy, chewing gum Saccharin, AS4A, anion exchange Sodium carbonate Conductivity
Acesulfam-K resin (140 mg/L)
Candy, beverages, Rebausides A and C, Lichrosorb NH2 Acetonitrile}water UV, 210 nm
pickles, soy sauce Stevioside
Diet cola Saccharin, Hypersil ODS Phosphate buffer (pH UV, 216 nm
Acesulfam-K 3.5)}acetonitrile
85 : 15
Soft drinks, candy, Rebausides A, Finepak SIL NH2 Acetonitrile}water UV, 210 nm
pickle Stevioside 80 : 18 with
tetrabutyl-ammonium
phosphate
Fruit yoghurts Aspartame Bondapak C-18 Phosphate buffer UV, array 400
125 mM (pH 3.5)}
acetonitrile 90 : 10
Soft drinks, table-top Acesulfam-K, Sucralose, Supelcosil LC-18 Acetonitrile (20 mM) : Vis 585 nM (after
sweeteners Saccharin, cyclamate phosphate buffer (pH post column
3.5)}gradient of 97 : 3 ion-pairing)
to 85 : 15
Candy, soft drinks, Aspartame and its Bondapak C-18 Phosphate buffer UV, 220 nm
yoghurt, custard, fruit decomposition products, 125 mM (pH 3.5)}
juice, nectar, biscuit, Saccharin, Alitame, acetonitrile 90 : 10,
chocolate Acesulfam-K 85 : 15, 98 : 2
Table-top sweeteners Acesulfam-K Lichrosorb RP-18 Potassium phosphate UV, 227 nm
buffer
Shrimp Saccharin IonPac AS-5 Sodium carbonate Conductivity
(7.7 mM) acetonitrile,
sodium hydroxide
(33 mM)
Cyanophenol
and yellow. Ponceau, Fast red-E, Benzyl violet 4B, using Nucleosil, C-18 or Lichrosorb RP-8 columns
erythrosine and some non-permitted synthetic col- and detection at 505 nm. The mobile phase con-
ours were separated. The procedure has been re- sisted of water}acetone mixtures (80 : 20) with tetra-
ported to be a viable and quicker alternative to TLC- butylammonium chloride added as ion-pair agent
spectrophotometric techniques. A method for the de- (0.2 g/L).
termination of L-orange, Sunset Yellow FCF and Pon- Carotenoids in red bell peppers were separated
ceau 4R by means of ion-pair chromatography has without saponiRcation using a C18 column and meth-
also been described. It has been used for analysis anol}ethyl acetate as mobile phase with detection at
of food dyes E 110, E 111 and E 12 in Rsh samples 475 nm.
Table 4 Simpler methods for the analysis of colours in food and food products
Food products Colour analysed Analytical details

column
Fruit juices, grape Ponceau, Fastred E, Ion-pair HPLC Methanol}water 480 nm, 610 nm
beverages, benzyl violet 4B,
confectionery erythrosine
Fish, bakery, meat E-110-orange II Lichrosorb RP-8 Water } acetonitrile 505 nm
products, E-III-orange I and E-124 (80 : 20)
miscellaneous Ponceau 4R
products
Olive oil Chlorophylls, carotenoids C-18 Methanol } acetone 410 nm
Red bell pepper Carotenoids ODS Methanol } ethylacetates 410 nm
(1 : 1)
Strawberry Anthocyanins ODS 10}30% aq. acetonitrile Photodiode
with 0.5% TFA array detector
A rapid method has been reported for quantitation E ethylene or propylene glycol fatty acid esters
of chlorophylls and carotenoids in virgin olive oil, by E ethoxylated derivatives of monoglycerides.
solid phase extraction on a C-18 column. The fat free
pigments were separated and concentrated. A total of Quantitative analysis of emulsiRers is difRcult
17 pigments were separated and quantitated with as most of them are similar in structure, their
a C-18 column and gradient elution of water}ionic commercial sources are quite heterogeneous and
pairing reagent}methanol and methanol}acetone their extraction from starchy foods is very difR-
(1 : 1). Detection was at 410 nm and 430 nm. - cult. A key problem is the quantitative extraction of
Carotene and other hydrocarbon carotenoids emulsiRers and the exclusion of interfering substan-
have been determined in red grape fruit cultivars ces. This problem is further complicated by the
with non-aqueous eluents using a C-18 column presence of food ingredients such as proteins, and
and isocratic mobile phase consisting of acetonitrile, the innate heterogenicity of most of the emulsiRers
methylene chloride and methanol (65 : 25 : 1, as well as the wide variation in their composition.
v/v/v). The schemes of analysis for lecithin, monogly-
Anthocyanins have been extracted from cultured cerides, TEMS, acetylated monoglycerides, partial
cells of strawberry plants using a 35% solution of polyglycerol esters, propylene glycol esters, poly-
acetic acid}acetonitrile}water (20 : 25 : 55), contain- sorbates, lactic acid esters, ethoxylated mono-
ing 0.1% triSuoroacetic acid; using a C-18 column, glycerides and sugar esters have been discussed.
and using 10 to 30% aqueous acetonitrile containing Baur has recommended solvents for extraction of
0.5% of triSuoroacetic acid as eluent in 30 min at emulsiRers.
403C with photodiode array detection. The method A method for the separation of monoglyceride (E
yielded higher concentration of anthocyanin than 471), sodium stearoyllactylate (E 481), calcium
other methods. stearoyllactilate (E 482), diacetyltartaric acid esters
Table 4 gives details of methods for the analysis of of mono- and diglycerides (E 472e) and mixed acetic-
colours in foods and food products. and tartaric acid esters of mono- and diglycer-
ides (E 472f) on a semi-preparative column has been
Emulsi\ers and Wetting Agents described. Various emulsiRers were identiRed by off-
Food emulsiRers assist the stabilization and forma- line high resolution mass spectrometer. Analysis of
tion of emulsions by reducing surface tension at the sodium or calcium stearoyllactrylate showed that the
oil}water interface. Common food emulsiRers used major components were 2-stearoyl and 2-palmitoyl
are: lactic acid and their salts.
Sodium dioctylsulfosuccinate, a wetting agent,
E lecithin and lecithin derivatives has been permitted in a variety of food products
E glycerol fatty acid esters including dry beverage bases. A post-column ion-
E hydroxycarboxylic acid and fatty acid esters pair extraction method was employed using methyl-
E lactylate fatty acid esters ene blue as counterion. Then the compound was
E polyglycerol fatty acid esters extracted into chloroform from the aqueous phase
2838 III / FOOD ADDITIVES / Thin-Layer (Planar) Chromatography
For analysis, a CN column was used with acetone} Macrae R (1982) HPLC in Food Analysis. London: Aca-
0.01 M KH2PO4 (1 : 5, v/v) as a mobile phase. demic Press.
Malissek R and Wittkowski R (1993) High Performance
Liquid Chromatography in Food Control and Research.
See also: II/Chromatography: Liquid: Mechanisms:
Lancaster: Technomic.
Reversed Phases. III/Food Additives: Thin-Layer
Nollet LM (ed.) (1992) Food Analysis by HPLC. New
(Planar) Chromatography.
York: Marcel Dekker.
Nollet LM (1996) Handbook of Food Analysis. New York:
Further Reading Marcel Dekker.
O’Brien Nabors L and Gelardi RC (1992) Alternative
Baur FJ (1973) J. Am. Oil. Chem. Soc. 50: 85. Sweeteners. New York: Marcel Dekker.
Combes RD and Haveland-Smith RB (1982) Mutat. Res. Prodolliet J and Bruehart M (1993) Determination of
98: 101. Acesulfam-K in foods. J. AOAC International, 76: 268.
Grendy TH (1991) Intense Sweeteners for Food Industry: Sardesai VM and Waldsham TH (1991) Natural and Syn-
An Overview. Trends in Food Sci. Tech., 2: 1. thetic Sweeteners. J. Nutr. Biochem. 2: 236.
Klein H and Leubolt R (1993) J. Chromatogr. 640: 259. Walters DE, Orthoefer FT and Dubois GE (1991)
Kurihara Y and Nirasawa S (1994) Sweet, antisweet and Sweeteners: Discovery, Molecular Design and
sweetness-inducing substances. Trends in Food Sci. Chemoreception, ACS Symposium Series 450. Washing-
Tech. 5: 37. ton: American Chemical Society.

M. Vega, Faculty of Pharmacy, University of
Concepcio& n, Concepcio& n, Chile organoleptic) purpose in the manufacture,
Copyright ^ 2000 Academic Press processing, preparation, treatment, packing,
packaging, transport or holding of such foods,
results or may reasonably expect to result (directly
Introduction or indirectly), in it or its by-products becoming
a component of or otherwise affecting the charac-
Thin-layer chromatography (TLC) is a relatively old teristics of such foods. The term does not
technique among the other chromatographic separ- include ‘contaminant’ or substances added to the
ation methods. In food additive analysis, this simple food for maintaining or improving nutritional
technique is the tool of choice, mainly because the qualities.
high throughput of samples that it can manage in
parallel and the wide range of compounds that can be
analysed simultaneously. Others deRnitions include:
Substance with non-nutritive properties, known

Food Additives chemical composition, intentionally added to food;
Anything added to food is not necessarily a food generally in small amounts, with the aim of im-
additive. Generally, a food additive is a substance or proving presentation (appearance, Savour, texture)
a mixture of substances different to the bulk of the and conservation properties of foods.
food and present as a result of any aspect of produc-
tion, processing or packing. This deRnition does not In other countries, such as Spain, additives are all
include hazardous contaminants. substances that can be added intentionally to food
The Codex Alimentarius Commission for Food Ad- and drink, without the purpose of changing the nutri-
ditives deRnes these as follows: tive value, to modify processing and conservation
characteristics, as well as to improve their adaptation
to the use for which it is produced.
Food additive means any substance not normally
consumed as a food by itself and normally used as
Classi\cation of Food Additives
a typical ingredient of the food, whether or not
it has nutritive value, the intentional addition Many methods have been used to classify food
of which to food for a technological (including additives. The majority imply functional grouping.
III / FOOD ADDITIVES / Thin-Layer (Planar) Chromatography 2839
Chemical-type grouping is convenient because it puts It is necessary to extract the additive compound
together moieties of similar structures and chemical from the food matrix and to apply additional puri-
properties in comparative categories. Toxicological Rcation procedures where necessary. Chromato-
and metabolic studies can also be correlated with graphic methods, which involve a separation pro-
chemical grouping. However, compounds belonging cess, allow the isolation of the compound(s) to
to the same chemical family have different functions be analysed. High performance liquid chromatog-
in the food industry. In spite of the fact that a com- raphy (HPLC), gas chromatography (GC) and
pound can have two or more different functional TLC have all been used extensively for the Rnal
groups, this classiRcation is more practical in the food analysis.
industry. Table 1 shows a typical classiRcation of For large numbers of samples, the comparative
food additives. advantage of TLC is that it is a completely instru-
Table 1 shows the diversity of compounds in- mental technique that can deal with many samples
cluded in the different classes generated. Grouping by simultaneously, and with samples of a diverse
functional group type can include chemical substan- nature.
ces both naturally and structurally quite different. In the past few years, many reviews have been
This is an additional problem for the analysis of published with the aim of featuring the relevant char-
substances considered, classiRed or included in lists of acteristics of instrumental planar chromatography,
food additives. or high performance thin-layer chromatography
In modern quality control, analysis is required (HPTLC).
at every step and not just in the Rnal product. This is In the Rrst place, it is necessary to describe the
to prevent possible defects directly at the critical advances obtained in the preparation of sorbents for
points but it produces a signiRcant increase in the stationary phases.
number of samples and the number of analyses to be Silica still represents the most frequently used
carried out. material for the stationary phase. Approximately
On the other hand, it is important to consider that 90% of separations by TLC are still carried out using
additives can be applied exclusively to those foods silica. Modern sorbents are characterized by smaller
where regulation points out speciRcally that they and more uniform particle size, implying a reduction
must be used and normally they must be declared on of equivalent plate height. This results in a higher
labels attached to the food. number of theoretical plates for a given run length
In spite of tolerance limits for some additives, the compared with traditional phases and, in turn, this
amount added should not exceed the amount ad- allows separations in shorter distances with corre-
equate to attain the objective, using the appropriate sponding time savings and reduction of diffusion
manufacturing procedure. This justiRes the necessity problems that appear when the mobile phase is re-
to detect and quantify food additives. Today there are tarded excessively.
many analytical procedures applied to these substan- Stationary phases of polarity from intermediate
ces. Obviously the method used will depend on the to reversed phase, have been developed. Most of
analytes, and their characteristic and/or physico- them are obtained by chemical modiRcation of silica
chemical properties. gel (Table 2).
Recently, sorbents with spherical particles such
as Lichrospher Si 60 F254s have appeared. They
Table 1 Permitted classes of food additives in Australia
offer shorter analysis times, improved separation
Class of additive Property of food influenced
Preservatives Shelf-life
Table 2 HPTLC stationary phases used in food additives
Colourings Appearance analysis
Flavouring and flavour enhancers Flavour
Antioxidants Shelf-life Stationary phase Food additives
Artificial sweetening substances Flavour, energy value
Vitamins and minerals Nutritive value 3-Aminopropyl Sugars, carboxylic acids,
preservatives
Modifying agents Reversed-phase, C18 Preservatives, antioxidants
Vegetable gums Texture, appearance Aluminium oxide Lipophilic food dyes
Mineral salts Texture, appearance Silica normal Antioxidants, sweeteners,
Food acids Shelf-life, flavour, texture surfactants, dyes
Emulsifiers Texture, appearance Cellulose Artificial sweeteners,
Humectants Texture, shelf-life carboxylic acids
Thickeners Texture Cellulose MN 300 Food dyes
efRciency, more compact spots, higher spot capacity Finally, video scanning not only allows informa-
and lower detection limits. All these new plates are tion to be saved digitally but also gives quantitative
applicable to the analysis of many analytes including results from image integration.
those used as food additives. Unlike HPLC, TLC needs very little sample puriR-
Advances in instrumentation in the whole cation and can be used with raw or dirty materials,
chromatographic process, ranging from application thereby saving time and additional expense.
devices through automatic developing chambers, and
automated multiple development (AMD) chambers Applications of HPTLC in Food
with computer control, all deserve special considera-
tion. Computer-controlled AMD using polarity
Additive Analysis
gradients increases efRciency in separation to limits TLC has been used for many years in the analysis of
comparable to HPLC, and retains the high through- food additives, such as food dyes, preservatives
put of samples to be analysed in the same chromato- (Table 3), antioxidants and sweeteners.
graphic run. Food colourants are in many cases fundamental
Major developments in densitometers have food additives because consumers judge product
meant improvements in the quantitative analytical quality by its colour. On the other hand, before a
ability of planar chromatography. These, coupled dye (natural or synthetic) is permitted for use on
to software, allow quantitative analysis of sub- food it has to be shown that it is nontoxic and
stances at very low concentrations thanks to the noncarcinogenic. The current list in western Euro-
high sensitivity of detectors that allow measure- pean countries comprises about 30 natural or artiR-
ments over the whole UV-visible spectrum as well as cial substances permitted as food dyes, and is very
Suorescence. small compared to the vast number of known dyes.
Another important aspect of HPTLC in food addi- This makes it necessary to have easy and fast methods
tive analysis, is that there are some compounds with to detect forbidden or unapproved food dyes
difRcult detection characteristics. The variety of re- (Table 4).
agents available, overcomes this difRculty and they Because the importance of colourings in foods,
can be used as pre- or postchromatographic derivatiz- a variety of separation procedures are still being
ing agents. examined with the aim of improving performance.
Table 3 Analysis of preservatives by HPTLC
Preservatives Layer Eluent Detection Reference
p-Hydroxybenzoates Silanized silica C18 Methanol}water (7#3, UV "270 nm Volkmann D (1980) HRC
n-propyl, ethyl, methyl 6#4, 5#5, 4#6 v/v) and CC, 3: 189
8 different preservatives Mixed layers of silica Petroleum ether}CCl4} UV "254 nm GosseleH JAW (1971)
and cellulose, F254 CHCl3}formic acid} Journal of Chromatography
glacial acetic acid 63: 433
(50#40#20#8#2 v/v)
Propyl, ethyl, methyl, RP18 W/F254 Acetone}water UV "254 nm Machery-Nagel (1990) TLC
hydroxybenzoates, (40#60 v/v) department
4- hydroxybenzoic acid
Parabens Silicagel 60, F254 1) pentane} UV "255 and Zimmermann A 8th
(hydroxybenzoic dichloromethane}acetic 310 nm Symposium of German
acid esters) acid (25#25#3 v/v) Association of Scientific
2) petroleum ether}diethyl and Applied Cosmetics,
ketone}acetic acid Hamburg, November 1989
(88#5#12, v/v)
Benzoic acid, sorbic Polyamide/cellulose Toluene}petroleum ether} UV "230 Duden R, Frikers R,
acid and parabens CHCl3}acetic acid benzoic acid Calverley KH, Park,
(30#15#10#1 v/v) "260 others Rios VM, Lebensm
Z (1973) Unters,
und Forsch 151: 23
III / FOOD ADDITIVES / Thin-Layer (Planar) Chromatography 2841
Table 4 HPTLC systems for food dye analysis
Food dye Layer Eluent Detection Reference
Seven dyes: erythrosine, Cellulose MN300 Sodium citrate 2.5%} Coloured substances Machery-Nagel (1990)
brilliant black NN, fast 25% ammonia} TLC Department
red E, naphtol red, methanol (20#5#3,
yellow orange S, v/v/v)
ponceau 4R, tartrazine
Tartrazine, Amaranth Silica-RP18 (A) methanol}acetonitrile} Scanning at different Oka H et al. (1987)
Indigo Carmine, New 5% sodium sulfate wavelengths Journal of
coccine, Sunset (3#3#10, v/v/v) Chromatography 411:
yellow FCF, Allura Red (B) methanol}MEK}5% 437}444
Ac, Fast green FCF, sodium sulfate
Brilliant blue FCF, (1#1#1, v/v/v)
R-106, R-103, R-3,
R-105, and R-104
Sulfonated dyes RP18; ion pair (A) Methanol}water Visual comparison Korner A (1993) Journal
ponceau, optimization (8#2, v/v) of Planar
tartrazine, (B) Methanol} Chromatography 8:
azorubin etc. water #20 mM solution 138}143
of tetrabutylammonium
bromide
Unlawful food dyes Silica-RP18 (A) methanol}acetonitrile} TLC}FAB}MS Oka H et al. (1994)
detection 5% sodium sulfate Journal of
(3#3#10, v/v/v) Chromatography A 674:
(B) Methanol}MEK}5% 301}307
sodium sulfate
(1#1#1, v/v/v)
Quinoline Yellow, Silica 60, (A) NH3}methanol}ethyl Different wavelengths RoH zylo JK and
Sunset F254 OPLC acetate (1#3#6, v/v/v) Siembida KR (1997)
Yellow, Cochineal (B) NH3}MEK}n-butanol Proceedings of the 9th
Red A, Indigo (2#3#5, v/v/v) International
Carmine, Tartrazine, Symposium on
Amaranth, Erythrozine Instrumental Planar
Chromatography
The use of surfactants, (hexadecyltrimethylam- of a variety of food additives, like antioxidants,

monium bromide) incorporated in the mobile natural food colourings, preservatives and water-
phase for the separation of acids and alkaline food soluble vitamins with silica plates. In this Reld, AMD
colourings, new polymer coatings for plates and new has also been successful in the separation, identiRca-
adsorbents such as Scolecite (corresponding to tion and quantiRcation of a diverse range of anti-
a natural zeolite) are recent approaches. Some col- oxidants.
ourants such as indigo carmine, cochineal red, acid Other food additives that requires adequate control
amaranth and tartrazine G, have been separated on are artiRcial sweeteners and antioxidants, because of
thin magnesium oxide layers with mixtures of 15% the carcinogenic properties attributed to some of
sodium citrate and methanol. Reversed-phase plates, them. Analysis of these is frequently carried out
obtained by impregnation of silica plates with 10% by HPLC, but, HPTLC shows the comparative
liquid parafRn in petrol ether are used for separation advantages formerly mentioned. Table 5 summa-
of different food dyes with advantage. rizes some HPTLC systems for analysis of these food
For food preservatives like benzoic and sorbic additives.
acid, the use of methods such as solid phase extrac-
tion (SPE) allow a better separation of these food See also: II/Chromatography: Thin-Layer (Planar):
additives from natural ingredients present in bever- Layers; Modes of Development: Forced Flow; Over Pres-
ages. sured Layer Chromatography and Centrifugal; Spray
Overpressured thin layer chromatography (OPTLC) Reagents. Dyes: High-speed Countercurrent Chroma-
has been shown to be a good tool for the separation tography; Liquid Chromatography; Thin-Layer (Planar)
Table 5 HPTLC systems for antioxidants and sweeteners
Additive Layer Eluent* Detection Reference
BHA, BHT, NDGA, Silicagel 60, F254 OPLC CHCl3}HAc CHCl3} Spraying with 0.5% Siembida R (1997)
Gallic acid esters Methanol}HAc solution of 2,6 Proceedings of the
Benzene-Methanol} dichloroquinone} 9th International
acetone}HAc 4 chlorimide and Symposium on
Methanol}acetone}water heating to 1053C Instrumental
Planar
Chromatography
Gallic acid esters, BHA, Silicagel-G25 HR Petroleum ether} Spraying with 0.5% Machery-Nagel (1990)
BHT, DBH, TBH benzene}HAc solution of 2,6 TLC Department
(2#2#1, v/v/v) dichloroquinone}
4 chlorimide and
heating to 1053C
BHA and dodecylgallate Silicagel Xylene}CHCl3}propanol} 10% Phosphomolybdate Sherma J and Fried
formic acid}HAc or vainillin in sulfuric B (1991)
(45#45#10#1#1, acid Handbook of Thin
v/v/v/v/v) Layer
Chromatography.
Marcel Dekker,
Inc.: 702
Saccharin, cyclamate Laboratory made mixed Xylene}HAc} Spray of developed plate Wooldich et al. (1969)
layers n}propanol}formic acid with ethanolic Z Lebenm. Unters.
Cellulose (45#7#6#2 v/v) dichlorofluorescein Forsch. 139: 142
MN300Ipolyamide; solution
(3#2)
Aspartam, acesulfam, Silicagel G-25, UV254 Xylene}HAc} Scanner dual Machery-Nagel (1990)
saccharin n}propanol}formic acid wavelength TLC Department
(45#7#6#2 v/v) 215/370 nm
* HAc"acetic acid
Chromatography. Food Additives: Liquid Chromatogra- Furia TE (ed.) (1972) Handbook of Food Additives, 2nd
phy. Impregnation Techniques: Thin Layer (Planar) edn. Boca Raton: CRC Press, Inc.
Chromatography. Pigments: Liquid Chromatography; Jork H, Funk W, Fisher W and Wimmer H (1994) Thin
Thin-Layer (Planar) Chromatography. Layer Chromatography: Reagents and Detection
Methods, vol. IB. Weinheim: VCH.
Sherma J and Fried B (eds) (1990) Handbook of Thin-Layer
Further Reading Chromatography. New York: Marcel Dekker, Inc.
CserhaH ti T and FogaH cs E (1997) Trend in thin } layer Stahl E (ed.) (1969) Thin}Layer Chromatography. A La-
chromatography. Journal of Chromatography Science boratory Handbook, 2nd edn. Heildelberg, New York:
35: 383}391. Springer-Verlag Berlin.
Fried B and Sherma J (1994) In: Fried B and Sherma J, Touchtone JC (ed.) (1992) Practice of Thin Layer
Thin-Layer Chromatography: Techniques and Applica- Chromatography, 3rd edn. New York: John Wiley and
tions, 3rd edn. New York: Marcel Dekker Inc. Sons. Inc.
III / FOOD MICROORGANISMS: BUOYANT DENSITY CENTRIFUGATION 2843
FOOD MICROORGANISMS:
BUOYANT DENSITY CENTRIFUGATION
R. Lindqvist, National Food Administration, to design a procedure for the separation of microor-
Uppsala, Sweden ganisms from a speciRc food. This approach has been
Copyright ^ 2000 Academic Press successful for several food/microorganism combina-
tions and it has been possible to separate and concen-
trate bacteria from food and to remove inhibitors
Introduction sufRciently to allow detection of bacteria by both
PCR and NASBA. However, in some cases, especially
Density gradient centrifugation is an established tech- when the food contains denser components, there can
nique for the separation and puriRcation of eu- be limitations which may be overcome by use of
karyotic and prokaryotic cells, viruses and subcellular a two-layer technique also described here. The proto-
components such as plasmids, mitochondria and nu- cols presented are based on work described in Lind-
cleic acids. Using this technique, components may be qvist et al. (1997), Lindqvist (1997) and Anonymous
separated based on their differences in density or size (1995) (see Further Reading).
during centrifugation in a gradient medium. Gradient
media that have been used include caesium chloride,
sodium metrizoate, sucrose, Ficoll, Ludox威, Percoll威 General Theory and Methodology
and BactXtractor2+. One of the limitations with this
technique has often been the properties of the gradi- Principles of Centrifugation
ent medium used. For instance, it is crucial that the Equation [1] describes the sedimentation of a sphere
gradient medium is at physiological ionic strength to in a centrifugal Reld:
avoid cell lysis or dehydration effects. Further, the
gradient medium should be nontoxic and not affect
the viability of the cells. v"[d 2(!1)/18];g [1]
Density gradient centrifugation using different
gradient media has been used to separate a wide where v"sedimentation rate, d"diameter of the
range of microorganisms from various types of sam- particle (hydrodynamically equivalent sphere),
ples. For example, bacteria have been recovered from "particle density, 1"liquid density, "viscos-
soil using sucrose gradients and different types of ity of the medium and g"centrifugal force.
protozoa have been separated from river water, From this relationship it can be seen that the sedi-
faeces, etc., using various gradient media. Further, mentation rate of a particle is proportional to its size
gradient centrifugation in Percoll威 has been used to and to the density difference between the particle and
distinguish subpopulations of pathogenic bacteria the medium. Also, the sedimentation rate is zero
and to separate live from dead eukaryotic cells. when the density of the particle is equal to the density
To some extent this technique has also been used to of the surrounding medium. Further, the sedimenta-
separate microorganisms from food, e.g. in the separ- tion rate decreases with increasing viscosity of the
ation of bacteria from milk and natural yoghurt. The medium and increases with increasing centrifugal
main beneRts of using density centrifugation is its force applied. From this it also follows that separ-
simplicity and speed in separating and concentrating ation depending on the conditions chosen may be
intact organisms from foods while at the same time carried out based on either the size (rate zonal centri-
removing compounds that might interfere with or fugation) or the density differences (isopycnic centri-
inhibit the detection method. The ability to remove fugation) between particles. In the latter case, each
inhibiting or interfering material present in food has particle will sediment to an equilibrium position in
been evaluated by subsequent detection of prepared the gradient where the gradient density is equal to the
microorganisms through methods of varying sensitiv- density of the particle. In the present work, the
ity such as traditional plate count procedures, the isopycnic technique is discussed.
polymerase chain reaction (PCR), nucleic acid se- The speciRc cell density, i.e. cell weight/cell volume
quence based ampliRcation (NASBA), and ATP when measured by buoyancy in a given medium
measurements. capable of forming density gradients, is referred to as
The present work describes how to use buoyant the buoyant density. Consequently, anything that
density centrifugation and gives an example of how affects the size of the microorganism, e.g. osmotic
2844 III / FOOD MICROORGANISMS: BUOYANT DENSITY CENTRIFUGATION
conditions and growth phase, may also affect their ately linearly to the g force and time of the centrifu-
buoyant densities and, thus, separation. This stresses gation. Rotor geometry and the size of the tubes also
the importance of the properties of the gradient me- have a marked effect on gradient shape. In contrast to
dium and of using standardized conditions when de- the self-generated continuous gradient, for some ap-
veloping and using the separation protocol. plications a uniform density centrifugation using one
(cushion) or several layers (step gradient) of the gradi-
Gradient Media
ent medium is preferred. The latter approaches are
Two gradient media with favourable properties for discussed here for the separation of bacteria from
work with microorganisms are Percoll威 and Bact- food.
Xtractor2+. The composition and properties of these
Overview of Methodology
gradient media are similar (see below). The most
important difference is that BactXtractor may be au- When designing a separation protocol, the same pro-
toclaved after NaCl and peptone have been added to cedure may be followed independent of the sub-
prepare a standard isotonic medium (SIM). Percoll sequent detection method. This procedure includes
and BactXtractor are nontoxic, have a low osmotic the following steps (Figure 1): (1) determination of
pressure and viscosity. These media consist of col- the buoyant density of the microorganism; (2) deter-
loidal silica particles of 15}30 nm diameter coated mination of the buoyant density of the food; (3)
with polyvinylpyrrolidone (PVP). They can form self- selection of the concentration of the gradient medium
generated gradients in the range of 1.0}1.3 g mL\1, to be used in the separation of the microorganism
which correspond to the cell densities of many micro- from food; (4) evaluation of the separation protocol
organisms. When a solution of Percoll威 (or Bact- with the desired detection method; and (5) optimiza-
Xtractor2+) is centrifuged at '10 000;g in an tion of the protocol if necessary.
angle-head rotor, the coated and hydrated silica par-
ticles will sediment resulting in an uneven distribution
of particles and the formation of a self-generated
Preparation of the Gradient Medium
density gradient. The gradient is formed isometrically The gradient media, Percoll威 (Pharmacia Biotech,
around the initial density of the gradient medium and Sweden) or BactXtractor2+ (QRAB, Uppsala,
becomes steeper with centrifugation time. The shape Sweden) have a density of around 1.130 g mL\1. Be-
of the gradient can be visualized by the use of col- fore use, the medium is made isotonic with physiolo-
oured density marker beads and is related approxim- gical saline by aseptically adding 8.5 g L\1 NaCl, and
Figure 1 Schematic overview of the sequential steps involved in designing a protocol for separation of microorganisms from food.
its suitability for maintaining microorganisms is im- points:

proved by addition of 1.0 g L\1 peptone. This stock
solution is termed 100% standard isotonic medium sy"[(g!p)/(100)];Cs#p
(100% SIM) and may be autoclaved if prepared with
BactXtractor2+. The solution of 100% SIM is diluted "[(1.1395!1.0095)/100]Cs#1.0095
to the required concentration with the appropriate
volume of peptone-water (8.5 g NaCl and 1.0 g pep- "0.0013;Cs#1.0095 [3]
tone in 1 litre of Millipore water).
The density of the 100% SIM solution described where sy"the density of SIM of concentration
above is calculated by the following formula: Cs (g mL\1), p"density of peptone water, esti-
mated here to be 1.0095 g mL\1, Cs"concentration
of SIM (%) and g"the actual density of the gradi-
s"[Vg;g#mNaCl#mp]/Vg
ent medium, assumed here to be 1.130 g mL\1.
"[100;1.130#0.85#0.1]/100"1.1395 [2] The line described by eqn [3] and shown in
Figure 2 is valid only for the properties of the gradi-
where s"the density of 100% SIM (g mL\1), ent medium and diluent assumed in this work but
Vg"the volume of gradient medium to be prepared similar graphs can easily be constructed for other
(mL), assumed here to be 100 mL (the volume change experimental conditions. The graph can then be used
by addition of solutes is negligible), g"the actual to determine what concentration of SIM is required
density of the gradient medium, assumed here to be to produce a gradient medium of a particular density.
1.130 g mL\1, mNaCl"the amount of NaCl added
(g), assumed here to be 0.85 g and mp"the amount Determination of the Buoyant
of peptone added (g), assumed here to be 0.1 g.
By using eqn [2], the density of 100% SIM was
Densities of Microorganisms
calculated to be 1.1395 g mL\1, assuming that Since the buoyant densities of microorganisms may
100 mL of gradient medium with a density of be affected by a number of factors, it is important that
1.130 g mL\1 was mixed with NaCl and peptone. they are handled and separated from food under
The relationship between density and concentration standardized conditions. For instance, the buoyant
of SIM can be determined by plotting the calculated densities of bacteria in Percoll威 gradients have been
density of a 100% solution and the density of a 0% shown to vary with growth rate and during the cell
solution, i.e. the density of the diluent, as a function cycle for some bacteria but not for others. Further,
of percent SIM concentration and then determining efforts to inactivate bacteria by, for instance, boiling,
the equation for the straight line between these two heating, and low pH treatment have shown that,
Figure 2 The relationship between density and concentration of the standard isotonic medium (SIM). The relationship described by
this line is only an example and was calculated based on the conditions described in this work. The exact relationship must be
calculated based on the density of the gradient medium and the diluent used.
depending on the treatment and the strain used, this this must be tested empirically. The best SIM concen-
may or may not affect their buoyant densities. Some tration to use depends of course on the microorgan-
variation in buoyant densities of a given strain may be isms. In our work, concentrations between 50% and
expected depending on the culture media used, stor- 80% and between 40% and 70% have been used
age time after growth, etc., but ideally this is in with microorganisms and food, respectively. How-
a density range where separation is not affected. ever, in addition to buoyant density, speciRc require-
However, the presence of different subpopulations in ments on the amount of sample and the shape of the
the sample, e.g. log-phase and stationary cells, may gradient may inSuence the scale of the experiment,
result in a wide continuous distribution, or even sep- i.e. rotor geometry and size of the tubes. Figure 3
arate bands, of microorganisms within the centrifuge shows examples of a large scale and a small scale
tube. protocol.
Generation and Reading of the Gradient
Preparation of Microorganisms
The gradient is generated by centrifugation and is
A sufRcient number of strains are cultured and
visualized by substituting the sample volume with
treated under relevant conditions to collect data on
a solution of colour-coded density marker beads
the buoyant density of the microorganism and its
(Pharmacia Biotech, Sweden) on top of the gradient
variation. The number of cells loaded on to the gradi-
medium. A volume of marker solution equivalent to
ent must be large enough to form a visible band in the
the sample volume is prepared by adding approxim-
centrifuge tube after centrifugation. In most cases
ately 5}10 L of each marker bead to a physiological
a solution containing 108}109 cfu mL\1 is sufRcient.
saline solution. During centrifugation, the density
The solution is prepared by centrifuging an appropri-
beads equilibrate at positions in the gradient corre-
ate volume of culture, washing, and resuspending in
sponding to their densities. The distance of the differ-
physiological saline or peptone-water. The washing
ently coloured density beads from the bottom of the
and resuspension of cells in fresh physio-
tube is recorded and the densities of the beads are
logical saline solution serves to suppress variations
plotted versus the position to generate a calibration
in osmotic pressure introduced by the presence of
curve (Figure 4). Similarly, the position of the micro-
metabolites, etc.
organisms in the centrifuge tube is recorded and the
corresponding buoyant density is read from the cali-
Sampling Loading
bration curves. The best resolution is obtained in the
The appropriate volume of the microorganism sus- steep part of the curves where a large distance in the
pension or solution containing the density marker centrifuge tubes corresponds to a small difference in
beads is carefully layered on the standard isotonic density (Figure 4). By comparing the shapes and
medium. There are no deRnite rules on how much locations of the three gradients in Figure 4 it can be
sample it is possible to load on to the gradient and seen that the resolution in the 60% SIM gradient is
Figure 3 Example of two protocols for the determination of the buoyant density of microorganisms and food.
of gradient medium of a uniform density (cushion

centrifugation). During the previous steps the buoy-
ant densities of the microorganism and of the food
have been determined under the relevant conditions
which will indicate if separation is possible, i.e. if
there is a difference in buoyant densities that may be
exploited. The food is generally less dense than the
microorganisms and the density of the gradient me-
dium is chosen so that it lies between the densities of
the food and the microorganism. Thus, the microor-
ganisms will be found in the bottom of the tube after
centrifugation. The optimum concentration of SIM to
use to obtain the cutoff density between food and
microorganisms can be determined from the relation-
Figure 4 Illustration of how to construct calibration curves for ship between density and SIM concentration de-
the determination of the buoyant densities of microorganisms and scribed in eqn [3] and Figure 2.
food. The gradients were generated by layering 0.5 mL of density
marker beads on 6.5 mL of 60%, 70% or 80% SIM and centrifug-
ing at 25 000;g for 25 min. Left: the positions of the different Separation by a Uniform Density
bands of colour-coded density marker beads in 60% SIM. Right:
the curves were constructed by plotting the position and the Centrifugation
density of the density marker beads in the diagram. The best
A quick way to test if the correct concentration of
resolution is obtained in the steep part of the curves, i.e. large
difference in position corresponds to a small difference in density. SIM has been selected is to run a uniform density
centrifugation (see below) of a cell suspension and
food homogenate, respectively. After centrifugation,
better at lower densities and vice versa for the 80%
food should remain on top of the gradient medium
SIM. However, in addition to the selection of the SIM
and cells should be visible in the lower part of the
concentration, it is sometimes useful to vary the cen-
tube. Based on the exact position of microorganisms,
trifugation time and speed when optimizing the gradi-
the volume of sample that needs to be retrieved is
ent for a speciRc microorganism or food. In a given
determined. The sample volume is the amount of
gradient, the width of the more or less well-deRned
medium contained in the centrifuge tube from the
band of microorganisms following centrifugation de-
upper band limit to the bottom of the tube.
pends on the number of microorganisms and the
In order to run a uniform density centrifugation,
range of buoyant densities present in the population
gradient formation during centrifugation should be
under study.
avoided. This may be achieved either by centrifu-
gation in a swing-out rotor or, if using an angle-head
Determination of the Buoyant rotor, by centrifuging under conditions where gradi-
ent formation is negligible (e.g. low g forces or short
Densities of Food centrifugation times).
The food homogenate to be loaded on the gradient
medium must be sufRciently concentrated to be vis- Evaluation of Protocol
ible in the centrifuge tube. Thus, a more concentrated
To evaluate the separation procedure, inoculated
homogenate is often used for this step than will be
food homogenates are analysed with the desired
used in the Rnal separation protocol. Typically
method of detection, e.g. plate counts or PCR. The
a 1 : 1 to 1 : 10 (w/v) homogenate is suitable. The
size of the centrifuge tubes to use depends on
determination of the buoyant density of food is then
the amount of sample needed for detection as well as
carried out by centrifugation in a self-generated
the number of samples to be analysed in a given time.
density gradient as described above for the
In Figure 5, examples of a large scale (larger
microorganism.
sample capacity) and a small scale (larger sample
throughput) protocol for separation of microorgan-
Selecting the Concentration of isms from food are offered to assist in the selection of
which centrifugation conditions to be used. Similar,
the Gradient Medium or identical, protocols have been used to separate
The simplest technique to separate microorganisms microorganisms, e.g. Shigella spp., E. coli O157:H7,
from food is to centrifuge the sample on a single layer Yersinia enterocolitica, Campylobacter spp. and
Figure 5 Example of two protocols for the separation of microorganisms from food by a uniform density centrifugation.
Zygosaccharomyces rouxii from different foods, e.g. the homogenate and/or preparing the homogenate in
raw beef, ground beef, different vegetables, shrimps, a stomacher bag with a Rlter bag. If this does not help,
chicken, blackcurrant syrup, and milk, prior to detec- or if the food contains particles denser than the
tion by methods such as PCR, NASBA and plate microorganism, it is possible to use two layers of
count procedures. gradient medium of different but uniform concentra-
tions (step gradient). The density of the second layer
is chosen to lie between that of the microorganisms of
Optimization of Protocol
interest and the denser components of the food. This
If, due to incomplete separation, the detection limit technique has been used to separate pathogenic bac-
is not satisfactory there are some key separation teria in a blue cheese from the lighter cheese particles
parameters that can be changed. Initially it can and from denser fungal mycelia. The sample is re-
be helpful to add density marker beads corresponding trieved at the interface between the gradient layers
to densities similar to those of the microorganisms and this position can be identiRed by centrifugation
during the uniform density centrifugation. The of an identical tube where the sample is replaced with
location of the beads will indicate where in the tube density marker beads.
the microorganisms will be found and if the optimal Another possibility, if the exact buoyant density of
SIM concentration has been used. The beads may also the microorganism is known such as in a well-deRned
indicate if a gradient has formed during centrifu- and constant experimental system, is to perform
gation. the separation step using a continuous gradient
Instead of sampling from above, as suggested in centrifugation.
Figure 5, microorganisms can be retrieved by inser-
tion of a syringe through the bottom of the centrifuge
tube to avoid the mixing of inhibitory compounds or
Future Developments
particles from the supernatant during cell removal. Buoyant density centrifugation is a general method
If particles from the food end up in the treated which recovers nonattached microorganisms over
sample, one can try to decrease the particle content in a speciRc buoyant density. This suggests two possible
the sample volume prior to centrifugation by diluting areas for development and improvement. The Rrst
III / FOOD TECHNOLOGY / Membrane Separations 2849
would be to increase the fraction of nonattached cells Chromatium warmingii and Chromatium vinosum.
by optimization of the homogenization or stomach- Archives in Microbiology 137: 350}356.
ing process. The second area is the exciting possibility Guerrero R, Pedros-Alio C, Schmidt TM and Mas J (1985)
of developing a separation protocol speciRc for single A survey of buoyant density of microorganisms in pure
types of microorganisms, or a systematic or metabolic cultures and natural samples. Microbiologia 1: 53}65.
Kubitschek HE (1987) Buoyant density variation during the
group of microorganisms. This may be achieved by
cell cycle in microorganisms. Critical Reviews in Micro-
manipulating the buoyant density of an organism biology 14: 73}97.
through a selective uptake of a speciRc compound. Lindqvist R (1997) Preparation of PCR samples from food
Further, since it is possible to perform the separation by a rapid and simple centrifugation technique evaluated
on a micro scale, it may be feasible to design auto- by detection of Escherichia coli O157:H7. International
mated systems for sample preparation and analysis Journal of Food Microbiology 37: 73}82.
with a high sample capacity. Lindqvist R, Norling B and Thisted Lambertz S (1997)
A rapid sample preparation method for PCR detection
See also: II/Centrifugation: Theory of Centrifugation. of food pathogens based on buoyant density centrifu-
gation. Letters in Applied Microbiology 24: 306}310.
Payne MJ and Kroll RG (1991) Methods for the separation
Further Reading and concentration of bacteria from foods. Trends in Food
Anonymous (1992) Percoll威 Reference List, 2nd edn. Science & Technology 5: 384}389.
Sweden: Pharmacia P-L Biochemicals Inc. Pertoft H and Laurent TC (1968) The use of gradients of
Anonymous (1995) Percoll威 Methodology and Applica- colloidal silica for the separation of cells and subcellular
tions, 2nd edn. Sweden: Pharmacia Biotech Inc. particles. In: Gerritsen T (ed.) Modern Separation
Basel RM, Richter ER and Banwart GJ (1983) Monitor- Methods of Macromolecules and Particles, vol. 2,
ing microbial numbers in food by density centrifu- pp. 71}90. New York: Wiley Interscience.
gation. Applied and Environmental Microbiology 45: Scherer P (1983) Separation of bacteria from a meth-
1156}1159. anogenic wastewater population by utilizing a self-
Guerrero R, Mas J and Pedros-Alio C (1984) Buoyant generating percoll gradient. Journal of Applied
density changes due to intracellular content of sulphur in Bacteriology 55: 481}486.
FOOD TECHNOLOGY
partial replacement for ion exchange. To date, per-

Membrane Separations vaporation applications are few in the food industry,
although it could be used for puriRcation of volatile
M. Cheryan, University of Illinois, Urbana, Illinois, USA aroma compounds partially to replace distillation.
This article focuses on selected food products with
varying physical properties and chemical composi-
tion and will illustrate the general applicability of
One of the earliest successful industrial applications
membrane technology in the food industry.
of membrane technology was in the food industry. In
1972, a dairy plant in New York began processing
cheese whey by reverse osmosis. Membrane separ- Dairy Industry
ations are now ubiquitous in the food industry, as
Milk
shown in Table 1. The main use of reverse osmosis is
the concentration of liquid foods, to complement or The dairy industry probably accounts for the largest
replace evaporation. NanoRltration is used for desalt- share of installed membrane capacity among food-
ing and de-acidiRcation with partial concentration, processing applications. Figure 1 is a general sche-
while ultraRltration is used for fractionation, concen- matic of possible applications of membranes in the
tration and puriRcation of food streams. MicroRltra- processing of milk. Reverse osmosis (RO) is mostly
tion is used for clariRcation and removal of suspended used to preconcentrate milk prior to evaporation (al-
matter to replace centrifuges and Rlter presses. It is though there are RO techniques that could concen-
also used for pasteurizing and sterilizing liquids in- trate skim milk up to 45% solids, as with the Fresh-
stead of using heat. Electrodialysis is Rnding use for note process, described later). This not only saves
demineralization and de-acidiRcation, as a possible sufRcient energy to justify the technology, but it also
2850 III / FOOD TECHNOLOGY / Membrane Separations
Table 1 Food industry applications of membrane technology
Dairy
RO: Preconcentration of milk and whey prior to evaporation; bulk transport; specialty fluid milk products
NF: Partial demineralization and concentration of whey
UF: Fractionation of milk for cheese manufacture; fractionation of whey for whey protein concentrates; specialty fluid milk products
MF: Clarification of cheese whey; defatting and reducing microbial load of milk
ED: Demineralization of milk and whey
Fruits and vegetables
Juices: apple (UF,RO), apricot, citrus (MF, UF, RO, ED), cranberry, grape (UF, RO), kiwi, peach (UF, RO), pear, pineapple (MF, UF,
RO), tomato (RO)
Pigments: anthocyanins, betanins (UF, RO)
Wastewater: apple, pineapple, potato (UF, RO)
Animal products
Gelatin: concentration and de-ashing (UF)
Eggs: concentration and reduction of glucose (UF, RO)
Animal by-products: blood, wastewater treatment (UF)
Beverages
MF, UF: Wine, beer, vinegar } clarification
RO: Low-alcohol beer
Sugar refining
Beet and cane extracts, maple syrup, candy wastewaters } clarification (MF, UF), desalting (ED), preconcentration (RO)
Oilseeds, cereals, legumes
Soybean processing: Protein concentrates and isolates (UF); protein hydrolysates (CMR); oil degumming and refining (UF, NF);
recovery of soy whey proteins (UF, RO); wastewater treatment (MF, UF, NF, RO)
Corn refining: Steepwater clarification and concentration (MF, UF, RO); light-middlings treatment: water recycle (RO); saccharification
of liquefied starch (CMR); purification of dextrose streams (MF, UF); fermentation of glucose to ethanol (CMR); downstream processing
of fermentation broths (MF, UF, NF, RO, ED, PV); wastewater treatment (MF, UF, NF, RO)
RO, reverse osmosis; NF, nanofiltration; UF, ultrafiltration; MF, microfiltration; ED, electrodialysis; CMR, continuous membrane
reactor; PV, pervaporation.
exposes milk to less heat during the concentration to be the preferred technique in these applications. UF
process, which minimizes protein denaturation and allows the passage of the lactose and soluble salts
development of the ‘cooked’ Savour and other heat- while retaining the protein and fat and some of the
damaging effects on the constituents of milk. insoluble or bound salts. There is considerable poten-
Perhaps the greatest potential for RO and/or ultra- tial in the manufacture of specialty milk-based bever-
Rltration (UF) in the dairy industry is in bulk milk ages such as lactose-reduced and calcium-enhanced
transport, especially in those countries which have Suid milk products. Total milk protein isolates are
large distances between producing and consumption usually manufactured by co-precipitation using a
areas. Considering that milk is more than 85% water, combination of heat, acid and/or calcium salts. This
preconcentration of the milk prior to shipment to generally results in low protein solubility, which re-
central dairies should result in considerable savings stricts its use as a functional food ingredient. UF of
in transportation costs, as well as reducing chilling milk in combination with diaRltration can produce
and storage costs. RO milk products, when recon- 90% protein isolates from milk with lactose concen-
stituted with good-quality water, are indistinguish- trations of less than 0.1%.
able from unconcentrated milks in Savour and other The principal use of milk UF in the dairy industry
quality attributes. On-farm ultraRltration of milk is today is the manufacture of cheese. From a mem-
technically feasible for large dairy herds if the concen- brane technologist’s point of view, cheese can be
trated milk is used for the manufacture of cheese. deRned as a fractionation process whereby protein
Stability of ultraRltered milk is satisfactory with (casein) and fat are concentrated in the curd, while
proper pretreatment such as thermalization at lactose, soluble proteins, minerals and other minor
65}703C for 10}20 s to minimize lipase activity, and components are lost in the whey. In the UF cheese-
if health and safety requirements are met on the farm. making process, milk is Rrst concentrated to a ‘pre-
cheese’ which will have the same protein, fat and/or
UltraVltration of milk RO milk could be used in the solid levels normally found in cheese. This pre-cheese
manufacture of several other products, such as cul- is then converted to cheese by conventional or modi-
tured milk products and cheese. However, UF seems Red cheese-making methods. Some of the beneRts of
Figure 1 Membrane processing of milk. (Reprinted with permission from Cheryan and Alvarez (1995).)
using UF in cheese-making are: other hand, there is no signiRcant change in the con-
centration of other components, so the permeate is
E There is an increase in yield of 10}30% with soft essentially bacteria-free skim milk.
and semi-soft cheeses due to the inclusion of the This process became commercial in 1989 to pro-
whey proteins duce more stable pasteurized and refrigerated milk
E The amount of enzyme (rennet) required is some- products. It could also be useful in subtropical and
times lower tropical countries, where inadequate refrigeration
E There is a reduced volume of milk to handle and transportation facilities result in high microbial
E Fewer cheese-making vats are required loads in the milk coming into dairy plants. Such
E Plant space is used better a membrane system in the bulk-milk holding station
E There is little or no whey production because most or on the receiving dock of the milk-processing
of the water and lactose has already been removed plant could lower the microbial load signiRcantly
and improve the quality of milk products in these
Considerable research has been conducted with countries.
a variety of cheeses such as feta, quarg, white cheeses Enriched casein fractions (i.e. separated from whey
from Turkey, Egypt (domiati, kariesh), Greece (te- proteins and other soluble milk components without
leme) and South America (queso fresco), goat’s milk isoelectric precipitation) can be produced using MF
cheese, Camembert, ricotta, mozzarella, cheddar and membranes. In addition, -casein can be isolated
processed cheese. from casein micelles if the temperature is lowered to
less than 53C, which causes -casein to dissociate
MicroVltration of milk The main applications of from the micelle and be removed in the permeate (a
microRltration (MF) in milk processing are fat separ- loose UF membrane of 100 000 (molecular weight
ation and bacterial removal. This concept has been cut-off) can also be used to isolate -casein from
put into commercial practice as the uniform trans- milk). This protein has biological activity potential in
membrane pressure (UTP) or co-current permeate pharmotherapeutic applications.
#ow (CPF) process. Tubular ceramic membranes with
Cheese Whey
1.4 m pores are operated in a double-loop constant-
pressure operation. Because of the uniform and low Whey is a by-product of the cheese industry. During
pressure proRles in the membrane module, fouling is the manufacture of cheese, most of the milk protein
low and Sux is very high (700}1000 L m\2 h\1 at (the casein) and fat is concentrated in the curd, which
10-fold concentration of skim milk). Bacterial reten- eventually becomes the cheese, while other constitu-
tion is 99% and the microbial load usually found in ents go into the water phase and become the whey.
milk and fat is also substantially rejected. On the Every 100 kg of milk will give about 10}20 kg of
Figure 2 Membrane processing of cheese whey. (Reprinted with permission from Cheryan (1998).)
cheese depending on the variety, and about 80}90 kg There are three primary areas where membranes can
of liquid whey. The disposal of whey is a problem: be applied in this application: Rrstly, clariRcation, e.g.
its biological oxygen demand (BOD) is 32 000} in the production of sparkling clear beverages using
60 000 p.p.m. It has low solid content and a very microRltration or ultraRltration; secondly, concentra-
unfavourable ratio of lactose to protein, which makes tion, e.g. using reverse osmosis to produce fruit juice
it difRcult to utilize in food products without chang- concentrates of greater than 423 Brix (a measure of
ing its composition. Prior to membrane technology, sugar concentration); and thirdly, de-acidiRcation,
as much as 60}70% of the whey produced was dis- e.g. electrodialysis or nanoRltration to reduce the
posed as sewage, with the rest being used primarily acidity in citrus juices.
for animal feed or human food. World production in
Clari\cation of Fruit Juice
1996 was estimated at 80}130 million tons per year:
the USA produced about 30 million tons per year. Fruit juices are prepared by extraction followed by
Membrane technology, and UF in particular, has a series of Rltration and clariRcation steps to yield
provided a valuable means of upgrading cheese whey clear single-strength juices (Figure 3). These opera-
and increasing its utilization as a human food. tions are usually labour- and-time-consuming. Mem-
The appropriate membrane can simultaneously frac- brane Rltration can replace the holding, Rltration and
tionate, purify and concentrate whey components decantation steps. The properties of the membrane,
(Figure 2), enhancing their market value and reduc- especially its pore size distribution, affect Sux and
ing the pollution problem. Today, whey protein con- capacity, as well as juice properties such as clarity,
centrates (WPC) produced by UF are well established browning compounds and total phenolics in the Rn-
in the food and dairy industries. Owing to the rela- ished product.
tively mild process conditions of temperature and pH, Membrane Rltration has several advantages over
the functionality of the whey proteins remains good, traditional methods. It eliminates Rning agents (be-
giving rise to a wide range of applications. The initial ntonite, gelatin, etc.), most enzymes (pectinase,
protein content of 10}12% (dry basis) can be in- amylase), centrifugation and diatomaceous earth Rl-
creased by UF, to result in 35, 50 or 80% protein tration. Process times are reduced from 12}36 to
products, with a concomitant decrease in lactose and 2}4 h. Juice yields are higher, by 2}15%, and product
some salts. WPC can be further fractionated into quality is better. The largest application is apple juice,
-lactoglobulin and -lactalbumin fractions as but the following have also received considerable
shown, or be used for the manufacture of caseino- attention: apricot, carrot, cherry, cranberry (which
macropeptide, a compound which may have a phar- was one of the earliest applications of ceramic mem-
motherapeutic value. branes), blackcurrant, grape, guava, kiwi, lemon,
lime, maple sap, melon, orange, passion fruit, peach,
pear, pineapple, plum, raspberry, strawberry and
Fruit Juices tomato.
Next to the dairy industry, fruit and vegetable juices Citrus juices (grapefruit, orange, lemon, lime) are
have beneRted the most from membrane technology. being upgraded by combining UF and adsorbent resin
Figure 3 Processing fruit juices by conventional and membrane technology. (Reprinted with permission from Cheryan (1998).)
technology to remove bitter compounds such as However, conventional RO is limited by osmotic

limonin, naringin, hesperidin, polyphenols and many pressure and viscosity considerations to less than 303
other off-Savour compounds. These compounds are Brix. Therefore, RO can be used as a preconcentra-
in the aqueous phase of the juice. Fresh or recon- tion step, with thermal evaporation completing the
stituted citrus juice which has been de-oiled and pas- required concentration to 42}453 Brix. Adding RO
teurized is Rrst ultraRltered to separate the pulp. The ahead of the evaporators can increase evaporator
clariRed permeate containing the sugars and bitter capacity and reduce thermal treatment.
compounds enters the absorption column which con- A signiRcant development in the 1980s was the
tains an adsorbent resin speciRcally designed to redevelopment of the FreshNote process by Du Pont
move these compounds. The debittered juice is then and FMC. It allowed the production of highly con-
recombined with the pulp (the UF retentate) to give centrated (42}703 Brix) fruit juices using a combina-
a product with less than 5 p.p.m. limonin, which is tion of high and low retention RO membranes. UF is
its apparent taste threshold (400}500 p.p.m. for Rrst used to separate the pulpy solids from the serum.
naringin). The UF retentate, about 1/10th}1/20th the feed vol-
In recent years, several fruit juice installations have ume, is subjected to a pasteurization treatment that
incorporated ceramic membranes. The higher cost destroys spoilage microorganisms and improves stab-
has been justiRed by the higher Sux, much longer life ility of the Rnished product when blended back with
and their resistance to aggressive processing and the concentrated UF permeate. The serum (UF per-
cleaning conditions. The ability to backSush to un- meate), which amounts to about 90}95% of the feed
block feed channels and back-pulsing during opera- volume, is concentrated by RO using hollow Rne
tion are other advantages. Rbres made of aromatic polyamide. Pressures are typ-
ically 1000}2000 psi. A multistage system is used
Concentration of Fruit Juice
with high rejection membranes in the early stages and
Orange and other citrus juice concentrates are mostly low rejection membranes in later stages. Permeates
produced by conventional multi-stage evaporation. with signiRcant sugar or Savour compounds are re-
RO with the appropriate polyamide composite mem- turned to stages containing high rejection mem-
brane can concentrate juice without a signiRcant loss branes. Fruit juice concentrates of 45}553 Brix have
of aroma, sugar or acids. The low temperatures avoid been obtained commercially, and up to 703 Brix
thermal damage of delicate aroma components. has been obtained in pilot trials. Careful control of
operating conditions is necessary. For example, the tion step. The MF or UF pretreatment is well suited
freshly extracted juice is blanketed with nitrogen and for subsequent ion exchange softening and chromato-
its temperature is controlled below 103C throughout graphic puriRcation. Another advantage is that a high
the remainder of the process. The Savour compounds quality soft cane molasses is obtained and this can go
in the serum are not subjected to any heat during directly to chromatographic separation to recover
processing, which also explains the high Savour sucrose and fructose, or to MF to remove some of the
scores for this product. Flavour and cost comparisons monovalent salts.
indicate very good market potential for this process. Another place in the sugar industry for UF or MF is
Commercial installations to date include tangerine to clarify thick juice (after evaporation), reducing
juice and apple juice concentrates. bacterial counts and storage losses. Treating thick
Concentration of tomato juice presents a difRcult juice has the advantage of handling lower volumes,
problem, because it has a high pulp content (25% but this is partially compensated for by higher viscos-
Rbre) and a high viscosity (which behaves in a non- ity and lower Sux.
Newtonian manner). Because of this, tubular mod-
ules are probably best. The colour of the Rnal tomato
concentrate is very good, and shows none of the
Vegetable Proteins
browning normally associated with evaporation. Most of the work in this area has been done with soya
The interest in RO of maple syrup grew in the beans. Once the oil has been removed from soya
1970s in response to increasing energy prices. RO is beans, the resulting meal is mostly used for animal
able to remove about 60% of water from the maple feed, with perhaps only about 3}5% being used di-
sap, resulting in a decrease of 33% in the processing rectly as human food. UF has been successfully used
cost compared to the all-thermal process. The con- to upgrade the quality of the soy protein by selectively
centrate is then boiled in a conventional open-pan removing undesirable components such as oligosac-
evaporator to develop the characteristic colour and charides (implicated in gastrointestinal stress when
Savour. consuming soya beans) and phytic acid (which forms
insoluble chelates with minerals and can form com-
plexes with proteins that reduce their bioavailability).
Sugar Re\ning To produce soy protein concentrates, the raw mater-
The most appropriate application of membranes in ial is defatted soy Sour which is extracted with dilute
this industry is for clariRcation and puriRcation of the alkali. The extract is then ultraRltered. The Rnal com-
extraction juices. If UF or MF were used at the mill to position will approximate to a soy protein concen-
remove the colloidal and macromolecular impurities, trate of 70% protein, dry basis. To produce isolates
a clear decolorized thin juice would be obtained with (90% protein), the Rbre and insoluble carbohydrate
little or no need for addition of lime, carbon dioxide are removed by centrifugation or Rltration prior to
or sulRte. If ion exchange is done immediately after UF. The UF technique usually results in higher yields
UF, lime could be eliminated completely. An added because of the inclusion of soy whey proteins
beneRt of membrane technology is that, with no mac- that are normally lost in conventional manufac-
romolecules and reduced lime levels, fouling and scal- turing methods. These whey proteins could also be
ing of the evaporator is reduced, which in turn reduc- contributing to the superior functional properties of
es down time and cleaning costs. Higher yield of the UF soy products, in addition to the beneRts of
sugar and better crystallization are also possible: the nonthermal and nonchemical nature of the UF
near-white sugar could be made in a single crystalliza- process.
Figure 4 Edible oil processing. Membrane technology can be used in the four unit operations within the enclosed box.
III / FOOD TECHNOLOGY / Supercritical Fluid Chromatography 2855
Vegetable Oils stable in organic solvents, the applicability of this

technology will widen considerably in the industry,
The basic unit operations in vegetable oil processing
especially in the production of ‘neutraceuticals’
are shown in Figure 4. Oil is extracted from plant
(minor compounds in plants that are thought to have
material (oilseeds) using a solvent, usually hexane.
considerable health beneRts) and grain processing
Published research indicates that about 50}70% of
(e.g., corn, soyabeans, wheat).
the hexane can be recovered and recycled using
nanoRltration membranes instead of the evaporators
used today, thus reducing energy consumption sub- Further Reading
stantially. The extracted crude oil is mostly triglycer-
ides, but it also contains small amounts of free fatty Bhave RR (ed.) (1991) Inorganic Membranes. Synthesis,
Characteristics and Applications. New York: Van Nos-
acids, phosphatides (lecithins/gums) and waxes,
trand Reinhold.
among other impurities. In place of physical or chem- Cheryan M (1992) Concentration of liquid foods by reverse
ical reRning, it is possible to use UF membranes for osmosis. In: Lund DB and Heldman DR (eds) Handbook
the degumming step, thereby producing a substan- of Food Engineering, p. 393. New York: Marcel
tially oil-free lecithin. Dekker.
In membrane reRning, the crude oil is treated with Cheryan M (1998) UltraTltration and MicroTltration
a solvent such as methanol to extract the free fatty Handbook. Lancaster, PA: Technomic.
acids. After phase separation, the methanol layer is Cheryan M and Alvarez J (1995) Membranes in food pro-
subjected to nanoRltration to recycle the methanol cessing. In: Noble RD and Stern SA (eds) Membrane
while producing a free fatty acid concentrate. This Separations. Technology, Principles and Applications,
avoids the traditional alkali-reRning process which p. 415. Amsterdam: Elsevier.
Cheryan M and Nicholas DJ (1992) Modelling of mem-
results in soapstock formation and oil losses. Dewax-
brane processes. In: Thorne S (ed.) Mathematical Mod-
ing with microRltration membranes can also be done. elling of Food Processes, p. 49. London: Elsevier.
In this process, the oil is cooled to a temperature Ho WSW and Sirkar KK (eds) (1992) Membrane Hand-
below the wax crystallization temperature before be- book. New York: Chapman and Hall.
ing microRltered to produce a stable edible oil. The Lloyd DR (ed.) (1985) Materials Science of Synthetic
application of membrane technology in the edible oil Membranes. Washington, DC: American Chemical
industry is expected to reduce energy consumption, Society.
reduce losses of oil, reduce the usage of chemicals and Matsuura T (1994) Synthetic Membranes and Membrane
water and reduce the discharge of contaminated Separation Processes. Boca Raton, FL: CRC Press.
efSuents. Raman LP, Cheryan M and Rajagopalan N (1994) Con-
sider nanoRltration for membrane separations. Chem-
ical Engineering Progress 90: 68.
Conclusions Renner E and El-Salam MH (1991) Application of
UltraTltration in the Dairy Industry. New York:
The food industry is one of the largest users of mem- Elsevier.
brane technology. With new developments in low- Singh N and Cheryan M (1997) Membrane applications in
fouling membranes and modules, and membranes corn wet milling. Cereal Foods World 42: 520.
J. W. King, National Center for Agricultural Utilization high performance liquid chromatography (HPLC)
Research, Agricultural Research Service/USDA, and hence is often ignored or relegated to a minor role
Peoria, IL, USA by food analysts. Despite these difRculties, SFC has
Copyright ^ 2000 Academic Press been applied to a variety of applications for the detec-
tion and quantiRcation of analytes, that are at least
The role of supercritical Suid chromatography (SFC) soluble to even a minor extent in supercritical carbon
in the analysis of foods and agriculturally derived dioxide (SC-CO2) } by far the most popular mobile
products has been somewhat moderated by uncer- phase utilized in the technique.
tainties in the availability of required instrumentation The application of SFC to food matrices came
for the past 15 years. In addition, SFC competes for naturally due in part to the early application of
the same analytical opportunities as gas (GC) and SC-CO2 extraction in the food industry, i.e. for the
2856 III / FOOD TECHNOLOGY / Supercritical Fluid Chromatography
extraction of coffee, hops and similar food items used by food analysts, due to the lack of an interface that
routinely by the consuming public. SFC is particularly permits routine coupling and use of the SFE/SFC
applicable to the analysis of lipid-containing mater- mode. However, preparative and even production
ials, due to relative high solubilities exhibited by these scale SFC has been utilized for specialized applica-
solutes (analytes) in SC-CO2. Analysis and detection tions in the food production industries, and probably
of ultra-trace components in foodstuffs, e.g. pestici- will see increased application due to the current inter-
des or drugs, has not been generally successful be- est in producing high value nutraceutical compo-
cause of the problems in routinely interfacing and nents, in a natural and environmentally benign
using sensitive detectors, such as the electron-capture manner.
detector (ECD) with SFC, due to the change in mo- Since SFC is perceived as a niche technique in the
bile-phase characteristics with respect to time during food industry, it is important to recognize when and
the analysis. However, the ability routinely to use the where it can be used to advantage relative to what can
Same ionization detector (FID) with SFC has probe achieved using GC and HPLC. Some of these
vided the analyst with a useful technique to detect an opportunities are as follows:
array of components, or a speciRc moiety, in complex
food matrices. 1. Reduction of the use of organic solvents relative to
With respect to the chromatographic technique HPLC;
utilized, it is capillary SFC which has been cited more 2. Direct analysis of samples avoiding sample prep-
often then packed-column SFC in the analysis of aration steps;
foods. This is somewhat unfortunate since the 3. Deformulation of commercial food products;
packed-column mode also offers interesting possibili- 4. Detection of product adulteration or deteriora-
ties, particularly when interfaced with a ultraviolet tion;
(UV) detector or evaporative light-scattering detector 5. Support of food engineering extraction/reaction
(ELSD). The recent use of this technique for the process development.
analysis of chiral compounds may also create some
opportunities in food analysis, where the knowledge With respect to nonpolar solutes, pressure-or density-
of the chirality of certain compounds (e.g. Savour programmed SFC provides the capability to analyse
esters) is of importance. In general, the coupling of compounds having molecular weights approaching
analytical supercritical Suid extraction (SFE) with 1200 amu in one chromatographic analysis. Sep-
SFC has not been adopted to any considerable extent aration of these compounds is a function of their
Figure 1 Supercritical fluid chromatography analysis of deodorizer distillate with an SB-octyl-50 column using flame ionization
detection (FID).
solubility in the mobile phase, their respective vapour analysis of food components, both changes in the rate
pressures and miscibility of the solute in the Suid of increase of the Suid pressure or density with re-
phase. For example, in Figure 1, a number of compo- spect to time, or in some cases changeover to an
nents have been separated using a capillary SFC exponentially-based Suid programme, will sufRce to
method that traditionally would have required the optimize the SFC run.
use of both GC and HPLC, and derivatization of Because of the molecular complexity exhibited by
some of the analytes. Utilizing SFC allows the analyst many food ingredients and compositions, it is not
to avoid the above approaches, and to analyse dir- unusual to have a temperature gradient with respect
ectly the sample, obtaining a snapshot of the entire to time superimposed on the mobile-phase pressure
molecular composition. These characteristic elution programme during the SFC run. For example, separ-
patterns produced by SFC can be used to identify the ation of the like-carbon number triglycerides in
presence or absence of a particular molecular con- soybean oil is not possible by pressure or density
stituent in a food sample, thereby providing valuable programming alone, but by superimposing a temper-
information for the food product formulator, to ature gradient during the analysis, these oil compo-
match or alter in developing new and competitive nents can be well separated. SFC analysis under iso-
products. baric conditions is limited in application when
Increasing concerns about minimizing or eliminat- analysing foods; however, it should not be over-
ing the use of hazardous organic solvents in the labor- looked since it can often yield the most precise and
atory also bodes well for the application of SFC. accurate results.
Incorporating SFC for the separation and detection of The FID is the most commonly used detector for
food-related solutes, eliminates not only most of the SFC. FID sensitivity to food components is lower
traditional solvent needs associated with HPLC, but than that obtained with GC since expansion of the
any solvents utilized in the extraction or sample mobile phase dilutes the detector signal substantially.
work-up steps prior to analysis. In this regard, SFC is However, the FID signal can be ampliRed to permit
an excellent tool for monitoring the end-result of an analysis down to the p.p.m. level, provided any shift
extraction or reaction of a food component using in the baseline can be compensated for. Analytes with
supercritical Suid media. Also, by using SFC, food- chromaphoric properties are amenable to UV detec-
related analytes that are thermally labile or suscep- tion in conjunction with SFC. The absorption maxi-
tible to degradation via oxidation are not exposed ma of components shift as a function of Suid density
to the harsh conditions that often accompany their or pressure, but most detector units constructed for
analysis by GC or HPLC. This advantage can be operation at these elevated pressures allows for stop
attributed to the protective action of CO2 which Sow or in situ, on-the-Sy scanning of peaks to deter-
excludes oxygen, and the low temperatures used mine absorbance maxima shifts. For example,
when separating components via SFC. bathochromic shifts of 15}20 nm have been recorded
for carotenoids over a 250 atm pressure interval in
Selecting and Optimizing Separation SC-CO2. The mating of the ELSD detector with SFC
has been reported by several investigators; however,
Conditions for Food Components day-to-day stability is inferior to that experienced
For the SFC analysis of food-related samples, the with HPLC-ELSD couplings when applied to food
analyst will undoubtly want to start with a general analysis. Hetero-element-speciRc detectors, such as
Suid-programming sequence to interrogate the ECD or Same photometric detector, have mostly been
sample matrix as to its components. These pro- utilized in research studies using SFC and have not
grammes are executed for an extended time to ensure seen serious adoption for routine analysis. Again,
optimum resolution and detection of the unknown or detector sensitivity and stability under SFC condi-
target analyte(s). Run times of 90 min in length are tions limit their sensitivity at best to parts per million
not unusual in this initial stage of method develop- range. The use of SFC with mass spectroscopy has
ment. After the target analytes have been identiRed by remained mainly an academic art, and commercial
retention time-matching with standards, or via an instrumentation development has been limited to
independent method such as mass spectrometry (MS), date.
the original programme can be modiRed to reduce the Most of the promising applications of capillary
analysis time or improve the resolution within SFC have utilized nonpolar bonded phases such as
the chromatogram. However, changes in the methyl, octyl, phenyl and biphenyl silicas. The weak
mobile-phase programme will usually be done to elutropic strength of neat SC-CO2 has favoured the
hasten the elution of early or late eluting components use of short chain length; monomeric silane-modiRed
that are of no importance in the analysis. In the columns have C1, C4, C18, phenyl, amino and diol
2858 III / FOOD TECHNOLOGY / Supercritical Fluid Chromatography
phases for packed-column SFC. The choice of these reaction as well as in the quality control of food raw
phases is not so much related to their selective inter- products and ingredients.
action with food-related solutes, but to their surface- Triglyceride-based oils/fats are also readily amen-
modifying properties which reduce peak tailing and able to analysis by SFC. Separation of the individual
solute interaction with the silica matrix. Resin col- components is once again governed by molecular
umns have also been utilized, but they are susceptible weight considerations, thereby allowing SFC to facil-
to voiding unless speciRcally packed for use under itate the separation of the major triglyceride species,
supercritical Suid conditions. i.e. T50, T52, T54, etc. For some oils, such as coconut
oil, will-resolved chromatograms result, while for
Types of Food Components Analysed other oils, e.g. soybean oil, there is overlap between
the saturated and unsaturated triglyceride species,
by SFC making superimposition of temperature gradient
A myriad of food-related components and matrices along with the pressure gradient programme for the
have been analysed by SFC, as indicated by the partial mobile phase necessary to achieve adequate re-
listing in Table 1. These include naturally occurring solution. However, even without ideal resolution,
ingredients such as fats/oils, spices, etc.; minor un- the rapid analysis afforded by SFC can be used to
wanted constituents like pesticides, antibiotic drugs considerable advantage for quality control, where
and mycotoxins; and speciRc food components, in- speed, rather than optimal separation, is often
cluding nutraceuticals and Savouring aids. Inspection desired.
of Table 1 indicates a preponderance of applications Detection of minor components in foods is limited
in the lipid analysis area. Indeed, SFC is tailor-made by the detector stability problems noted previously;
for lipid analysis, although somewhat lacking in the however, those components which can be detected by
high resolution capabilities demonstrated by high using FID, UV or ELSD are often analysed more
temperature GC. The retention pattern for lipid sol- rapidly by SFC, due to the time savings afforded by
utes in SFC, as shown in Figure 1, follows a distinct avoiding elaborate preparation of the sample prior to
pattern governed approximately by the solute’s mo- analysis. SFC analysis provides a more detailed pro-
lecular weight/volatility characteristics. Elution of the Rle of the entire sample in addition to detecting the
following classes of lipids is in the order: fatty acid target analyte. This allows a more accurate assess-
methyl esters, free fatty acids, hydrocarbons, vit- ment of the total contribution of the minor constitu-
amins, sterols, wax esters, mono- followed by ent to the entire ingredient proRle, e.g. the presence of
diglycerides, and then triglycerides/steryl esters. Al- sterol esters in sawtooth palmetto berry extracts,
though there is some overlap between individual where fatty acids and triglycerides are the major
classes of the above solutes, due to the overlapping constituents.
molecular weights ranges (e.g. triglycerides and steryl Other food sample types that are readily analysed
esters), this separation pattern has proven very useful by SFC are the fat-soluble vitamins, essential and
in tracking conversion of lipid species undergoing Savour oil ingredients, spice materials, hop compo-
Table 1 Food components separated and analysed by SFC
Carbohydrates Derivatized corn syrups, mannose glycans

Chiral compounds Monoterpenes, pyrazines, clenbuterol
Drugs/antibiotics Caffeine, erythromycin, polycyclic ether antibiotics, sulfonamides, assorted steroids
Hydrocarbons Sesquiterpenes, squalene, waxes and wax esters
Lipids Fatty acids, fatty acid esters, monoglycerides, diglycerides, triglycerides, sterol esters, sterols
(cholesterol), fat-soluble vitamins, tocopherols, phospholipids (lecithin), lipid hydroperoxides,
glycolipids
Nutraceuticals Valeriana, gingolides, sawtooth palmetto berry
Oils/fats Celery oil, coconut, fish, soybean, wheat germ, palm oil, rice oil, milk/cheese triglycerides
Packaging/film components Polypropylene oligomers, polyvinyl chloride, phenolic antioxidants, low molecular weight poly-
styrene
Pesticides Halogenated, organophosphorus, carbamate, pyrethrins, acidic phenoxy herbicides, sulfonyl
ureas
Pigments Carotenoids, xanthophylls
Speciality ingredients Hops components
Spices/flavours Capsicum, cardoman, coumarin, curry, garlic components, majoran, rosemary, vanillin
Terpenes/essential and fruit oils Grapefruit oil, limonenes, mint, lemon
Other toxicants Mycotoxins, nitrosamines, polycyclic aromatic hydrocarbons
nents and nutraceutical formulations. Some SFC- tween the -tocopherol and cholesterol was achieved
based separations require the use of a co-solvent (usu- using the lengthy density programme. However, it is
ally 5}20 vol%) in addition to the SC-CO2 for the perhaps more important that, by adjusting the elution
mobile phase. For example, phospolipids are only conditions, the background components (Rsh oil tri-
sparingly soluble in neat SC-CO2, but these polar glycerides) that were of no interest in this analysis can
lipid compounds can be chromatographed success- be programmed off the column without resorting to
fully on packed silica columns by incorporating a pre-fractionation of the sample prior to SFC analy-
ethanol and/or water as a modiRer into the mobile sis or derivatization of the sample matrix.
phase. Likewise, carbohydrate moieties, which ex- Not all applications of SFC require the above con-
hibit limited or no solubility in SC-CO2 or SC- ditions for high resolution separations. For example,
CO2/co-solvent mobile phases, can be derivatized to packed-column SFC (5 m, C8 Deltabond) has been
allow their analysis by SFC. used to clean up samples prior to other types of
chromatographic analysis (GC). In this case, or-
Selected Applications of SFC in ganochlorine and organophosphorus pesticides were
extracted by SFE with SC-CO2 from a meat sample,
Food Analysis and the pesticides separated from the co-extracted fat
In this section, several brief examples will be given to moieties using the packed SFC column. Hence, by
illustrate the utility and potential of SFC in food ‘heart-cutting’ the appropriate elution fraction,
analysis. Figure 2 illustrates the SFC separation and a lipid-free, pesticide-containing fraction was pro-
detection of -tocopherol and cholesterol in a Rsh oil vided for GC residue analysis.
capsule. This was achieved on a capillary SB-methyl SFC is an excellent technique to monitor reaction
column at 1203C using the density programme noted chemistry between lipid species, since it avoids the
on the horizontal axis. Although this analysis took need to employ more than one analytical technique or
over 90 min to perform, it illustrates some of the sample derivatization. Further, it permits the success-
beneRts that can be achieved using SFC. For example, ful chromatography of all of the relevant reactants
the chromatogram in Figure 2 was achieved with no and products in one chromatographic analysis.
sample preparation other than to dilute the oil in Examples where SFC has been applied are in the
a small quantity of solvent and to inject it into the esteriRcation or transesteriRcation of lipids, glycer-
chromatograph. In addition, no derivatization of the olysis reactions and randomization of fats/oils.
sample was required and adequate resolution be- Table 2 shows the analysis of the glyceride content of
Figure 2 Determination of cholesterol and -tocopherol in a fish oil capsule by capillary SFC.
2860 III / FOOD TECHNOLOGY / Supercritical Fluid Extraction
Table 2 Analysis of glycerides in a randomized fat sample based on chromatographic fractionations initially de-
veloped using analytical scale packed SFC columns.
Methods of analysis MG DG TG Time of analysis
SFC-FID 0.2 9.6 90.1 25 min Further Reading

GC-FID 0.1 6.9 92.9 30 min
HPLC-FID 13.5 86.5 1h Anton C and Berger C (eds) (1997) Supercritical Fluid
HPLC-ELSD 8.0 92.0 30 min Chromatography with Packed Columns. New York:
LC-silica column 1.0 7.7 93.1 2h Marcel Dekker.
TLC 2.0 11.0 87.0 30 min Caude M and Thiebaut D (eds) (1999) Practical Supercriti-
cal Fluid Chromatography and Extraction. Amsterdam:
MG, Monoglyceride; DG, diglyceride; TG, triglyceride. Harwood Academic.
Dean JR (ed.) (1993) Applications of Supercritical Fluids
in Industrila Analysis. Boca Raton, Florida: CRC
a randomized fat sample using six different analysis Press.
methods. The results given in Table 2 suggest that King JW (1990) Applications of capillary supercritical Suid
SFC-FID analysis yields comparable data for an chromatography } supercritical Suid extraction to natu-
equivalent analysis time to that obtained using the ral products. Journal of Chromatographic Science 28: 9.
GC-FID and HPLC-ELSD methods. However, the King JW (ed.) (1996) Supercritical Suid extraction and
SFC method does not require the time and effort for chromatography. Seminars in Food Analysis 1:
sample preparation associated with the alternative 101}116, 133}144, 163}165.
King JW and List GR (eds) (1996) Supercritical Fluid Tech-
techniques and, in addition, saves on the cost of
nology in Oil and Lipid Chemistry. Champaign, Illinois:
solvents and chemical reagents. A further illustration AOCS Press.
of the cost- and time-saving advantages of SFC is Lee ML and Markides KE (eds) (1990) Analytical Super-
noted by its ability to monitor free and methylated critical Fluid Chromatography and Extraction. Provo,
fatty acids, thereby providing a reasonably quick and Utah: Chromatography Conferences.
accurate assay for these compounds in foodstuffs to McDonald RE and Mossoba MM (eds) (1997) New Tech-
support nutritional analysis claims and the detection niques and Applications in Lipid Analysis. Champaign,
of frying oil deterioration as a function on time. Illinois: AOCS Press.
Preparative or production scale SFC is now being Nam KS and King JW (1994) Coupled SFE/SFC/GC for the
used as a separation technique in the food industry. trace analysis of pesticide residues in food samples. Jour-
Fractionation and isolation of higher value food com- nal of High Resolution Chromatography 17: 577.
Saito M, Yamauchi Y and Okuyama T (eds) (1994) Frac-
ponents, such as tocopherols and phospholipids, or
tionation by Packed Column SFC and SFE. New York:
the -fatty acids/esters from Rsh oils, have been cited VCH.
in the literature. Recently, a production plant for the Smith RM (ed) (1988) Supercritical Fluid Chromatography.
separation of Rsh oil ethyl esters has been constructed London: Royal Society of Chemistry.
in Spain to produce 595% pure polyunsaturated Wenclawiak B (ed.) (1992) Analysis with Supercritical
fatty acids for the nutraceutical market. The basic Fluids: Extraction and Chromatography. Berlin:
separation design of this production scale plant is Springer-Verlag.
S. S. H. Rizvi, Institute of Food Science, nonpolluting. The solubility and selectivity properties
Cornell University, Ithaca, New York, NY, USA of SC-CO2 has been compared with hexane. While
Copyright ^ 2000 Academic Press both are nonpolar solvents, the selectivity of SC-CO2
is enhanced in the presence of modiRers (entrainers).
In the food industry, supercritical Suid extraction For example, in the absence of such polar modiRers as
(SFE) is currently being used in a number of areas, water and ethanol, SC-CO2 alone is a poor solvent
shown in Table 1. The most attractive features of SFE for extraction of caffeine from coffee beans or nic-
in food processing is the fact that separation can be otine from tobacco. While the selectivity of SC-CO2
carried out at relatively low temperatures (40}603C) and the mechanism of modiRer action are not com-
using benign solvents. The solvent most widely used pletely understood, studies in the area of supercritical
thus far is carbon dioxide, which is inexpensive, Suid chromatography (SFC) have indicated that
nontoxic, nonSammable, easily recoverable, and Lewis acid}base pairing, induced-dipole interactions
III / FOOD TECHNOLOGY / Supercritical Fluid Extraction 2861
Table 1 Selected applications of supercritical fluid extraction in food processing
Process Commercial manufacturers Literature source
Coffee decaffeination KaffeHAG AG, Germany; General Foods, Texas USA; Zosel (1978)
SKW-Trotsberg, Pozzillo, Italy Williams (1981)
Hops and spices extraction SKW-Trotsburg, Munchmuenstar, Germany; Hubert and Vitzthum (1978)
Paul and While Beigat, UK; Pfizer, Sydney, Nerbraska; Vollbrecht (1982)
J.I. Hass, Yakima, Washington
Flavours and fragrances Flavax GmbH, Rehlingen Germany; Calame and Steiner (1982)
Canilli Albert and Louis, Grasse, France Caragay and Little (1981)
Vegetable oils and fatty acids Mohi Oil Mills, Japan; Marbert GmbH, Dusseldorf, Germany Stahl et al. (1980)
Friedreich (1984)
and hydrogen bonding play an important role in hundred patents on SFE of biomaterials in the United
determining the selectivity of SC-CO2. States alone. A potential user of these processes is
The principal disadvantage of SFE is that relatively likely to be faced with the involved task of determin-
high pressures (typically 50}100 atm or more) are ing if patent infringements exist or identify sources of
required. Even though the energy savings make SFE legitimate licensing agreements. While SFE develop-
attractive, the initial capital cost of high-pressure ments are growing globally, further research is
equipment overrides these considerations, especially needed both in terms of fundamental studies and
at current energy prices. While the overall cost of SFE applications.
is dictated by such other factors as volume, price and The feasibility of extraction of a number of food
the continuous or batch nature of the process, eco- materials using supercritical Suids has been investi-
nomic considerations have slowed its commerciali- gated over the past two decades. In particular, much
zation. Generally, SFE is best suited for difRcult sep- activity has focussed on extracting and reRning fats,
arations, not attainable by conventional processes. In oils and their derivatives. The equilibrium solubility
situations where SFE can produce a new product or values of some of these are shown in Table 2. A num-
when environmental or regulatory concerns make its ber of advantages have been cited for the use of SFE in
use more attractive, the application may more than the processing of food-grade fats and oils from both
justify the cost. The hazards of high pressure and the animal and plant sources. These include:
use of Sammable solvents are also perceived un-
favourably by many not experienced in these areas. E Low temperature processing reduces degradation
While common in the petroleum industry, the food of temperature and oxygen-sensitive components.
industry also uses a number of processes like homo- E Both extract and rafRnate are free of solvent and
genization, extrusion and compression routinely and can be used in food.
thus should be able to deal with moderate pressures. E Extraction and fractionation into various cuts
Another frequently overlooked problem associated of different physicochemical properties can be per-
with SFE is patent infringement. There are over one formed simultaneously.
Table 2 Solubilities of selected food materials in supercritical fluids
Food material SC solvent Solubility Extraction Conditions Co-solvent Reference

(%wt/wt) (3C) (bar)
}
Beta-carotene Ethylene
0.23
0.17 50
70
374
374 }
Chang and Randolph (1989)
}
Cholesterol CO2
0.33
0.37
60
60
270
270 MeOH
Wong and Johnston (1986)
Coumarin CO2 0.7 40 100 } King and Friedrick (1990)

}
Butterfat
Palm oil triglycerides
CO2
2.2
1.1*
40
75
248
300 }
Yu et al. (1992)
Brunner (1994)
Ethylene 2.2* 70 300 }
}
Soybean triglycerides CO2
0.6*
3.0*
50
50
300
600 }
Friedrich (1984)
Hop extract CO2 9.4 80 400 } Stahl et al. (1987)

Evening primrose oil CO2 8.0 40 300 } Lee et al. (1994)
*Estimated from graphs.

2862 III / FORENSIC SCIENCES / Capillary Electrophoresis
E A continuous and large-scale process can be Further Reading

economically competitive to hexane-based opera-
Brunner G (1994) Gas Extraction. New York: Springer.
tions. Chang AD and Randolph AD (1989) Precipitation of
Exploiting the commonality of high pressure be- microsize organic particles from supercritical Suids.
tween supercritical Suid and extrusion processing op- AICHE Journal 35: 1876}1882.
erations, a hybrid unit operation called supercritical Charpentier BA and Sevenantes MR (eds) (1988) Tech-
niques and Applications. Supercritical Fluid Extraction
Suid extrusion (SCFX) has been recently developed.
and Chromatography. ACS Symposium Series 366.
This new process permits generation of microcellular Washington DC: American Chemical Society.
structure at low temperature by using SC-CO2 as Friedrich, JP (1984) Supercritical CO2 extraction of lipids
a blowing agent instead of steam to puff the extru- from lipids containing materials. U.S. Patent 4,466,923.
date, thus decoupling the conventional dual role of Lee BC, Kim JD, Hwang KY and Lee YY (1994) In: Rizvi
water, which otherwise serves both as a blowing SSH (ed.) Supercritical Fluid Processing of Food and
agent as well as a plasticizer. The use of super- Biomaterials. New York: Chapman and Hall.
critical Suid also permits deposition of solute into the King JW and Friedrich JP (1990) Quantitative correlations
extrudate matrix. between solute molecular structure and solubility in
SigniRcant progress has also been made in the anal- supercritical Suids. Journal of Chromatography 517:
ysis of food and related materials using SFE with SFC. 449}458.
McHugh M and Krukonis VA (1994) Supercritical Fluid
Sample preparation for analysis often requries orders
Extraction. Boston: Butterworth-Heinemann.
of magnitude more time than the analysis itself and Rizvi SSH, Mulvaney SJ and Sokey AS (1995) The com-
the use of supercritical Suids obviates the need for bined application of supercritical Suid and extrusion
hazardous organic solvents with no additional treat- technology. Trends in Food Science Technology 6(7):
ment prior to identiRcation of the analyte by other 232}240.
techniques such as GC, GC-MS, FTIR, etc. The solu- Rizvi SSH (ed.) (1984) Supercritical Fluid Processing of
bility of lipid-like materials in SC-CO2 ranges from Biomaterial. New York: Blackie Academic and Profes-
1 to 30 wt%, depending on the density of the Suid sional.
used, and therefore, SFE has become a method of Stahl E, Quirin KW and Gerard D (1987) Dense Gases for
choice for rapid extraction of fats and oils from Extraction and ReTning. New York: Springer-Verlag.
a variety of food matrices such as animal, vegetable, Taylor LT (1996) Supercritical Fluid Extraction. New
York: John Wiley.
grain and seafood products. Other successful applica-
Williams DF (1981) Extraction with supercritical gases.
tions include extraction of fat-soluble vitamins, pesti- Chemical Engineering Science 36(11): 1769.
cides, sterols, and fatty acids. As an analytical tool, Wong JM and Johnston KP (1986) Solubilization of bi-
SFC has also made signiRcant progress over the past omolecules in carbon dioxide based supercritical Suids.
decades but has yet to prove its superiority over the Biotechnology Progress 2: 29}39.
more conventional techniques. Yu ZR, Rizvi SSH and Zollweg JA (1992) Extraction of oil
from evening primrose seed with supercritical carbon
See also: II/Extraction: Supercritical Fluid Extraction. dioxide. Journal of Supercritical Fluids 5: 114.
III/Food Technology: Supercritical Fluid Chromatogra- Zosel K (1978) Separation with supercritical gases: practi-
phy. On-Line Sample Preparation: Supercritical Fluid cal applications. Angewandte Chemical International
Extraction. Edition English 17: 702.
FORENSIC SCIENCES
enzyme markers as a means of identiRcation in crim-

Capillary Electrophoresis inal cases over many years. In 1991 capillary elec-
trophoresis (CE) was introduced to forensic analysis.
J. Sadecka, Slovak Technical University, The separation of bulk heroin, heroin impurities and
Bratislava, Slovak Republic degradation products using micellar electrokinetic
capillary chromatography (MEKC), the determina-
Copyright ^ 2000 Academic Press tion of drugs of abuse in urine and also the determina-
Several slab electrophoretic techniques have fre- tion of benzodiazepines and sulfonamides in urine by
quently been used to discriminate between red cell CE}mass spectrometry were described in the same
III / FORENSIC SCIENCES / Capillary Electrophoresis 2863
year. Today, forensic applications of CE include anal- pounds; and (iii) MEKC to analyse neutral and/or
ysis of drugs of abuse, gunshot residues, explosives, charged compounds.
pen inks and toxins as well as polymerase chain Most abused drugs are bases which are generally
reaction (PCR) ampliRed DNA. water-soluble and ionized as cations at low pH. The
use of simple electrolyte solutions such as phosphate,
citrate and formate at pH values of 2}3 gives a useful
Drugs of Abuse initial separation. Basic drugs can be analysed by
TLC and HPLC, but interactions with the stationary
One of the major tasks for forensic laboratories is the phases can lead to peak tailing. This problem does not
analysis of illicit and controlled drugs, in both the occur so frequently in CE. In addition, these simple
seizure and biological samples. electrolytes have low background UV absorbance and
can be operated at low wavelengths of 190}200 nm,
Seizure Samples
where many drugs have signiRcantly enhanced UV
Seizure samples are analysed in order to identify the absorbance coefRcient. CE at low pH can be used to
major compounds. In addition, the determination of detect by-products in puriRed codeine, to investigate
trace compounds permits the samples to be allocated amphetamine derivatives in Ecstasy tablets, and for
to the source and production procedures. Seizure assay for various pharmaceutical formulations which
samples may consist of a mixture of acidic, neutral contain 1,4-benzodiazepines and phenothiazines.
and basic compounds that may be nonpolar and/or At high pH the migration direction of acidic com-
polar. At least two independent analytical parameters pounds is against the electroosmotic Sow, which
should be used to establish the identity of the drug, maximizes mobility differences. Operation with
and infrared spectroscopy and thin-layer chromatog- simple electrolytes such as phosphate pH 7 or borate
raphy (TLC) are widely used for this purpose. Quan- pH 9.5 often leads to useful initial separation for
titation is usually carried out by gas chromatography acidic compounds.
(GC) and high performance liquid chromatography MEKC can be used when dealing with uncharged
(HPLC). GC is a high resolution technique, but prob- solutes or mixtures of charged and neutral species.
lems can arise for thermally degradable, polar and This approach may also be considered when
nonvolatile substances. HPLC is less suited to drug simple mobility differences prove insufRcient in capil-
proRling, because it is a relatively low resolution lary zone electrophoresis (CZE). Both anionic and
technique compared with GC. cationic surfactants have been used as micelle modi-
CE is a relatively new technique is forensic drug Rers, which, furthermore, are complementary ap-
analysis. The three main separation mechanisms have proaches. MEKC has been applied to a wide range of
been used for seizure samples: (i) low pH to analyse controlled substances, including heroin, cocaine,
basic compounds; (ii) high pH to analyse acidic com- opium alkaloids, amphetamines, hallucinogens,
Figure 1 Typical example of a MEKC separation of the components of a drug mixture. Buffer: 25 mmol L\1 borate (pH 9.24)}20%
methanol}100 mmol L\1 SDS. Capillary: bare fused silica, i.d. 50 m, total length 55 cm (35 cm to detector). Potential 20 kV. UV
detection at 200 nm. Peak identification: a, caffeine; b, barbital; c, pentobarbitone; d, morphine; e, narceine; f, 6-monoacetylmorphine;
g, codeine; h, nalorphine; i, lidocaine; j, procaine; k, heroin; l, flunitrazepam; m, acetylcodeine; n, thebaine; o, papaverine;
p, amphetamine; q, narcotine; r, cocaine; s, diazepam; t, tetracaine. (Reprinted with kind permission of Elsevier Science from Tagliaro F
et al. (1996) Journal of Chromatography A 735: 227}235.)
barbiturates, benzodiazepines and cannabinoids. An The important problem faced by CE in the Reld of
electropherogram of a complex mixture of 20 drugs biological sample analysis is still its relatively low
(acidic, neutral and basic) is shown in Figure 1. concentration sensitivity. Increased sensitivity can be
It is clear that MEKC represents an excellent tech- obtained both instrumentally and by processing the
nique for drug screening. In addition, photodiode- sample before analysis. Liquid}liquid extraction
array UV, laser-induced Suorescence (LIF) and mass (LLE) and solid-phase extraction (SPE) methods have
spectrometry (MS) detection can greatly increase spe- been used as sample pretreatment for CE. After LLE
ciRcity of the analysis. Greater speciRcity of screening and SPE, extracted mixtures can be dried and re-
could be obtained by using two complementary sep- solved with a small volume of solvent, thus achieving
aration techniques, e.g. MEKC with either GC or detection limits of about 10 ng mL\1. The sensitivity
HPLC. The complementary nature of MEKC and can be gained from employing more sensitive Suores-
CZE for the identiRcation of 17 illicit drugs and cence detection. When LIF can be applied, the sensi-
related compounds ionized at pH 2.35 has also been tivity limit of CE can be improved by a factor of
demonstrated. MEKC with sodium dodecyl sulfate about 1000 or more over UV absorbance detection.
(SDS) at pH 9.2 gave a highly noncorrelated separ- Unfortunately, the choice of wavelength emitted by
ation compared to that obtained on a CZE system at laser is limited and this is the main limitation of LIF
pH 2.35. MEKC was found to be signiRcantly, but application to drug analysis. Concentrating the
inversely, correlated with a CZE system at pH 9.2. sample on the capillary } ‘stacking’ } is a simple
The reproducibility of migration times or relative technique that overcomes the poor detection limits of
migration times in MEKC is most important for CE. Three general stacking methods are used in CE:
screening applications. Migration time precision of (i) low ionic strength buffer in the sample; (ii) stack-
1% relative standard deviation (RSD) for repeated ing by including acetonitrile in the sample; and (iii)
injection has been shown; this is essential to allow isotachophoresis (ITP).
conRrmation of the identity of each individual com- For forensic purposes it is often more appropriate
pounds present. Relative migration times generally to identify the metabolite rather than the parent drug,
give better repeatability, with RSD values of less than since additional information yielded by the full meta-
1%. bolic proRle of a drug may be important in ascertain-
The results generated by MEKC are often coming the route of administration. Drug metabolites are
pared with those of HPLC and/or GC. GC affords most often studied by HPLC; however, phase II
higher resolution than MEKC; however, derivatiz- metabolites, e.g. glucuronide and sulfate conjugates,
ations are commonly required. MEKC offers signiR- are acidic and highly polar and elute with little resolu-
cantly greater efRciency, selectivity, peak symmetry tion in reversed-phase HPLC systems. Phase II metab-
and speed compared to HPLC. In addition, the drugs olites are ideal for direct analysis by CE without the
that are poorly chromatographed by HPLC or not at need for previous derivatization or hydrolysis. MEKC
all by GC exhibit good electrophoretic behaviour with diode array detection has been used for the
using MEKC. A recognized deRciency of MEKC is determination of morphine, morphine-3-glucuronide,
sensitivity, which is below that for HPLC-UV. morphine-6-glucuronide, normorphine, meclofena-
mic acid and its metabolites in equine urine. SPE
Biological Samples
procedures were developed to concentrate and purify
The analysis scheme of drugs of abuse in biological the analytes from post-administration urines. The
samples involves screening using an immunoassay low concentration sensitivity of MEKC in compari-
test. This does not enable positive identiRcation to be son with HPLC and GC-MS can be overcome by
made, but permits negative samples to be detected using a suitable sample preparation procedure, in
and discarded. Subsequently, the results must be con- particular ofSine SPE.
Rrmed by a more speciRc method. Without doubt, Hair analysis is a tool to prove drug abuse in
GC-MS is the reference method to conRrm positive questions of drug-related fatalities, revocation and
screening tests. At present, blood and urine represent restoration of driver licences, criminal responsibility,
most samples analysed for abused drugs and toxi- prenatal drug exposure and offences of narcotics law.
cants in most laboratories. If analyte concentrations The concentration of drugs in the hair are in the
are high enough, biological Suids, even those contain- ng mL\1 range, at least in cases of chronic abuse. CE
ing high concentrations of ions and proteins, can be applications in hair analysis are still in an early stage
directly injected on to a CE system, after simple of development. The few reports published until now
Rltration/centrifugation of the sample. Urine analysis come from Tagliaro’s group. The use of CE with
can be very fast and simple, while SDS additive must UV detection, at 238 nm for cocaine and 214 nm
be used with plasma to solubilize protein. for morphine, has permitted the achievement of
moderate sensitivity 0.2 ng mL\1 for both analytes. lodextrin (-CD) as a chiral selector. Different -CD
Later, stacking techniques were developed in order to derivatives have been used with success to separate
increase sensitivity (about Rve times). MEKC has also ephedrine and analogous compounds, amphetamines,
been applied to hair analysis. The sensitivity was methamphetamine and methylenedioxy-derivatives
slightly worse } 0.4 ng mL\1 } than with CZE, but of amphetamines.
selectivity was much higher. Good resolution and The versatility of modiRed uncharged and charged
efRciency were obtained with both methods. The -CDs in the direct resolution of -agonists, -antag-
same-day RSDs of migration time were (1% in onists, phenylethylamines, alcohol stimulants and
CZE and (2% in MEKC. Same-day precision RSD thalidomide and its metabolites by CE was shown.
was 3}5%. A total of 42 compounds were optically resolved
Among banned pharmaceutical substances, those using hydroxypropyl--CD and 20 with sodium
which are of greatest relevance to sport medicine are sulfobutyl ether--CD. The preliminary analysis of
anabolic agents, stimulants, diuretics, narcotics and ephedrine, amphetamine, methamphetamine and
-blockers. In addition to several methods for screen- methylenedioxy-derivatives of amphetamine in urine
ing (immunoassays, GC and HPLC), conRrmation by (Figure 2) and hair (Figure 3) showed that after
MS, if needed, is used. Also, CE is a useful technique a liquid}liquid extraction, urine samples could be
for the simultaneous screening of different types of analysed with a sensitivity below 500 ng mL\1. For
drugs, e.g. -blockers, anabolic steroids, diuretics and hair analysis, it is necessary to increase sensitivity
narcotics. LIF detection provides ultimate sensitivity (0.1 ng mL\1) by applying a stacking procedure.
while maintaining the extreme separation efRciency
of CE. Both CE and MEKC have been applied. The
main advantage of MEKC over CZE is that neutral
Forensic DNA Samples
and charged compounds as well as compounds which DNA polymorphism analysis has recently been recog-
are insoluble in water can be separated in a single run. nized as a source of identiRcation for individuals in
criminal cases and unidentiRed human remains. The
conventional technique for DNA typing based on
Chiral Separations restriction fragment length polymorphism (RFLP) has
The chiral resolution of drugs is of forensic signiR- been replaced by more accurate, sensitive and faster
cance for legal and intelligence purpose. In many PCR procedures. In contrast to RFLP, the PCR pro-
instances, only one enantiomer is controlled under cedures require less DNA and can be used on DNA
legal status, and proper identiRcation is therefore which is degraded. There are several PCR-based pro-
critical. For example, only the (#)-enantiomer of cedures under development; however, short tandem
nerpseudoephedrine and the (!)-enantiomer of repeat (STR) sequences are currently of major im-
propoxyphene are controlled. In addition, the enan- portance in the Reld of identiRcation of individuals in
tiomeric composition of seized samples can provide forensic cases. STRs are DNA segments, typically
information on the possible different synthetic routes. found in noncoding regions, which are composed of
The enantiomeric purity of drug detected in urine or repeating units of 2}5 base pairs (bp). Co-ampliRca-
other biological samples may exclude the possibility tion of two or more of these loci in one PCR provides
that the subject has taken that drug in a racemic an efRcient mechanism for typing multiple genetic
preparation. loci simultaneously. The detection of STRs is based
CE is a technique that has been shown to be ideally on the variation in the length of STR-containing PCR
suited for chiral separations. When compared to products. These PCR-ampliRed STRs must be separ-
other techniques, such as liquid chromatography, CE ated to determine the size, quantity and/or sequence
has the advantages of efRciency, resolution, selectiv- of each fragment.
ity, speed and direct chiral separation. Therefore, CE DNA restriction fragments and PCR products have
has become a method of choice in the chiral analysis traditionally been separated by slab gel electrophor-
of therapeutic drugs, but the attention paid to con- esis.
trolled or illicit drugs is still low. As for chromatogra- Recently, CE has emerged as a novel, high perfor-
phy, it is possible to separate enantiomers directly or mance DNA analysis tool. Early work on DNA separ-
after a pre-derivatization step. ation was done on cross-linked polyacrylamide gel-
In the separation of amphetamine analogues, Rlled capillaries. However, gel-Rlled capillaries are
2,3,4,4-tetra-O-acetyl--D-glucopyranosyl isothio- difRcult to prepare and have a short life time. The
cyanate as a derivatization reagent was used and utility of CE has been greatly enhanced by using
enantiomers were separated using MEKC. However, noncross-linked polymer networks instead of rigid gel
more applications are conducted by CE with -cyc- media. These polymer solutions offer low-to-medium
Figure 2 Typical electropherograms of: (A) blank human urine extract; (B) extract from blank human urine spiked with: 1, racemic
ephedrine; 2, amphetamine; 3, methamphetamine; 4, 3,4-methylenedioxyamphetamine; 5, 3,4-methylenedioxymethamphetamine; 6,
3,4-methylenedioxyethylamphetamine, at concentrations of 1 g mL\1 for every racemic analyte. Conditions: buffer, 100 mmol L\1
phosphate, pH 2.5, containing 10 mmol L\1 -cyclodextrin. Capillary, uncoated fused silica, 45 cm;50 m i.d. Potential, 10 kV.
Detection, UV absorbance at 200 nm. (Reprinted with kind permission of Wiley-VCH from Tagliaro F et al. (1998) Electrophoresis 19:
42}50.)
viscosity, which makes replacement of separation me- ecule at the end(s) of the DNA fragments could lead
dium possible after each electrophoresis run. The to efRcient separation of relatively large DNA frag-
polymer solution also has a broader effective DNA ments (100}900 bp) in free solution. Contrary to
size range due to its Sexible and larger effective pore current electrophoretic methods, this method requires
size structure. no sieving matrix, provides better results at high
The CE system produced results which were com- voltage and leads to shorter preparation time and
parable to those obtained on slab gel electrophoresis, faster separations.
with a level of precision of $0.1% bp (between Detection of PCR products has been achieved in
instruments). This comparison is very important if three ways: (i) UV absorbance by the DNA fragment;
a comparison is to be made of results obtained by (ii) LIF using intercalating dyes; and (iii) Suorescence
different laboratories and to standardize available of primers tagged with Suorescence dyes.
procedures. In automated Suorescence analysis the alleles from
DNA fragments cannot normally be separated in an STR locus are PCR-ampliRed from human
free solution. However, the Rrst clinical experimental genomic DNA using an unlabelled primer and one
results demonstrated that adding an uncharged mol- primer labelled at the 5-end with a Suorescent dye.
Figure 3 Typical electropherograms of: (A) blank human hair extract; (B) extract from blank human hair spiked with: 1, racemic
ephedrine; 2, amphetamine; 3, methamphetamine; 4, 3,4-methylenedioxyamphetamine; 5, 3,4-methylenedioxymethamphetamine;
6, 3,4-methylenedioxyethylamphetamine, at concentrations of 1 ng mg\1 for every racemic analyte. Conditions: Buffer, 100 mmol L\1
phosphate, pH 2.5, containing 15 mmol L\ -cyclodextrin. Capillary, uncoated fused silica, 45 cm;50 m i.d. Potential, 10 kV.
Detection, UV absorbance at 200 nm. (Reprinted with kind permission of Wiley-VCH from Tagliaro F et al. (1998) Electrophoresis 19:
42}50.)
Denatured PCR products are then analysed by CE Recently, a new CE instrument capable of simulta-
with in-lane size standard (DNA-fragments of known neous multicolour detection and high resolution of
size labelled in a different colour dye) on slab gel or DNA fragments was developed. This instrument, the
CE capable of real-time multicolour Suorescence de- ABI Prism 310 Genetic Analyser, is highly automated.
tection. The collected data are then analysed by soft- Multiplex STR products are sequentially injected into
ware which automatically determine allele size based a single capillary and detected by LIF. LIF is detected
on a standard curve for the in-line size standard. STR on a charged coupled device camera, which simulta-
loci which overlap in size can be distinguished using neously detects all wavelengths from 525 to 680 nm.
different dyes that Suoresce at different wavelengths. Ninety-six samples in a single tray can be analysed by
Results have indicated that the sizes obtained for STR the instrument. The polymer used on the instrument
alleles can differ depending on the gel and elec- has many performance features that are critical to the
trophoresis conditions and depending on the instru- success of STR analysis in forensic work: alleles dif-
ment used, however, high precision can be obtained fering in size by a single base (up to 250 bp in length)
in multiplex PCR analysis by using an in-line internal can be detected; sizing precision (between alleles of
standard ((0.16 nucleotide SD). the same length) of less than 0.15 nucleotide SD is
possible; analysis time per sample is less than 30 min; and CE for separation and detection can be recom-
capillary life is at least 100 injections; and the run mended in order to avoid false-positive results.
temperature is set at 603C to provide a highly de-
naturing environment for the DNA samples.
The disadvantage of CE is that it is a serial tech-
Other
nique, making its total throughput no better than the The analysis of inks as part of the detection of fraudu-
long run times and parallel separations in convential lent documents is a small but important part in the
slab gel electrophoresis. Several attempts to obtain operation of a forensic laboratory. TLC and HPLC
faster and higher throughput separations have been have been extensively used to separate and distin-
reported. These include capillary array electrophor- guish inks during the last decades. In comparison, CE
esis in ultra thin slab gels. These two techniques are has been applied only rarely. UV-Vis, Suorescence
limited by difRculty in assembling the separation sys- and particle-induced X-ray emission (PIXE) detection
tem and in carrying out sample introduction. The use of electrophoretically separated diluted original inks
of CE to provide continuous automated loading of and ink extracts (Figure 4) from substrate material
PCR products on to ultra thin slab gels shows new provide sufRcient information for the comparison of
potential for increasing sample throughput in STR different inks (Table 1). The possibility of compari-
analysis, although separation resolution still needs to son of 50 forensic inks by MEKC has also been
be improved. investigated. The separation patterns of individual
Over the years CE has become widely used as dyes were compared with those obtained by HPLC
a power tool in post-PCR analysis. However, it is and TLC, showing a much higher separation efRcien-
difRcult to introduce PCR to routine laboratories, cy for MEKC. Some inks, which cannot be dis-
because of the possibility of false-positive results. criminated by applying the HPLC and TLC method,
These false positives may be caused by sample-to- can deRnitely be distinguished using MEKC.
sample contamination or by the carry-over of pre- Nicotine, nornicotine and anabasine, the active
viously ampliRed PCR product. The online coupling principal components in all tobacco products, have
of fused silica capillary as the microreactor for PCR been separated by CE and the potential for CE to
Figure 4 Electropherograms of extracts from dried blue and black inks see (see Table 1) and original inks diluted 100-fold in
5 mmol L\1 borate buffer, pH 8.25. Capillary, fused silica 45 cm;50 m i.d. Potential, 25 kV. Detection, laser-induced fluorescence at
exc/em"320/436 nm. (A) Extract ink 1; (B) original ink 1 (diluted 1 : 100); (C) extract ink 2; (D) original ink 2 (diluted 1 : 100).
Table 1 Listing of fountain pen inks investigated molecules based not only on charge, but also on size,
hydrophobicity and stereospeciRcity. CE offers cer-
Ink Colour Manufacturer Country tain advantages for forensic analysis:
number of origin
1 Blue Cross USA 1. higher theoretical plate number than HPLC;

2 Black Cross USA 2. in many instances CE is faster than GC and HPLC;
3 Royal blue Pelikan Germany
4 Brilliant black Pelikan Germany
3. with regard to sample preparation, CE is easier
5 Blue Pilot Japan than GC and HPLC. In many instances the sample
6 Black Pilot Japan can be injected directly with little or no prepara-
7 Blue Lamy Germany tion;
8 Black Lamy Germany 4. lower cost per analysis;
9 Royal blue Geha Germany
10 Brilliant black Geha Germany
5. full automation;
11 Blue-black Parker USA 6. CE is complementary to GC and HPLC;
12 Washable blue Parker USA 7. two complementary techniques such as CE and
13 Permanent black Parker USA MECC can be carried out with the same instru-
14 Royal blue washable Parker France ment.
15 Permanent blue Parker France
16 Black Waterman France
17 Royal blue Mont Blanc Germany Despite these features, the technique has not yet
(Reprinted with kind permission of Wiley-VCH from Rohde et al.
been widely accepted in the forensic community. This
(1998) Electrophoresis 19: 31}41.) may be in part due to the legal system. Different
countries have different standards to achieve legal
defensibility of analytical results in court and forensic
characterize tobacco products on their alkaloid pro-
laboratories rely to a large extent on commercial
Rles for classiRcation purposes has also been demon-
instruments which are specially built and approved
strated.
by a governmental agency for speciRc analysis. Even
Marine phytotoxines present a major public health
so, the US Drug Enforcement Agency is now using CE
problem because they can contaminate seafood. CE
for general drug screening to quantitate heroin sam-
and MEKC enable okadaic acid, microcystins and
ples.
maitotoxin to be detected in the picogram range.
Two drawbacks of CE are often stated: low repro-
Since any case of mushroom intoxication may have
ducibility and low sensitivity. However, due to sev-
legal consequences, the accurate determination of
eral recently presented results (e.g. detection limits in
mushroom toxins is of primary importance for foren-
the region of ng mL\1, the precision of migration
sic pathologists and toxicologists. The analysis of
times (1% RSD, same-day and day-to-day repeata-
amatoxins by CE instead of radioimmunoassay has
bility characterized by RSD values in the range of
several advantages: analysis is faster, less costly and it
1}4%, when peak area ratios were used), these draw-
requires smaller amounts of sample. Rapid and sensi-
backs seem not to be so critical.
tive CE method for the separation and determination
As stated by Kuffner et al.: ‘The legal criteria of
of the psilocybin and baeocystin in hallucinogenic
Daubert, as long as they are met by the scientiRc
mushrooms has also been reported.
community, will allow CE into evidence as acceptable
Many types of explosives consisting of inorganic
expert testimony’.
and organic components have been used in criminal
cases. It has been demonstrated that the original com-
See also: II/Electrophoresis: Capillary Electrophoresis;
position of some explosive devices can be derived
Capillary Electrophoresis-Mass Spectrometry; Capillary
from the components of the post-blast residue. In Electrophoresis-Nuclear Magnetic Resonance. III/Clini-
short, CE offers a powerful tool which is suitable for cal Chemistry: Thin-Layer (Planar) Chromatography.
both high and low explosive and gunshot residue Forensic Sciences: Liquid Chromatography.
analysis. CE is suitable for the determination of both
inorganic and organic components, showing greater
versatility than the traditional methods such as Further Reading
atomic absorption spectrometry and PIXE. Kuffner CA Jr, Marchi E, Morgado JM and Rubio CR
(1996) Capillary electrophoresis and Daubert: time for
Conclusion admission. Analytical Chemistry 68 (7): 241A.
Lurie IS (1997) Application of micellar electrokinetic capil-
CE is a new technique in the forensic laboratory for lary chromatography to the analysis of illicit drug seiz-
the separation and quantitation of a wide variety of ures. Journal of Chromatography A 780: 265.
2870 III / FORENSIC SCIENCES / Liquid Chromatography
McCord BR (ed.) (1998) Volume symposium capillary conRrmation of results in forensic drug analysis. Journal
electrophoresis in forensic science. Electrophoresis of Chromatography A 735: 227.
19(1): 11. Tagliaro F, Turrina S, Pisi P et al. (1998) Determination of
Tagliaro F and Smith FP (1996) Forensic capillary elec- illicit and/or abused drugs and compounds of forensic
trophoresis. Trends in Analytical Chemistry 15 (10): interest in biosamples by capillary electrophoretic/elec-
513. trokinetic methods. Journal of Chromatography B 713:
Tagliaro F, Turrina S and Smith FP (1996) Capillary 27.
electrophoresis: principles and applications in illicit Thormann W, Molteni S, Caslavska J and Schmutz A
drug analysis. Forensic Science International 77: (1994) Clinical and forensic applications of capillary
211. electrophoresis. Electrophoresis 15: 3.
Tagliaro F, Smith FP, Turrina S et al. (1996) Complement- von Heeren F and Thormann W (1997) Capillary elec-
ary use of capillary zone electrophoresis and micellar trophoresis in clinical and forensic analysis. Electro-
electrokinetic capillary chromatography for mutual phoresis 18: 2415.
L. A. Kaine, C. L. Flurer and K. A. Wolnik, presence or absence of known compounds or class of

Forensic Chemistry Center, US Food and compounds; (ii) screen for unspeciRed analytes; (iii)
Drug Administration, Cincinnati, perform a comparative analysis; and (iv) verify
OH, USA and/or quantify substances in a sample. Many quali-
Copyright ^ 2000 Academic Press tative tests are used in forensic laboratories for their
speed and ability to identify unknowns. In some cases
Forensic science is the application of the sciences to qualitative results (identity) sufRce, while in others
the court of law. Consequently, forensic science and quantiRcation is important. Like gas chromatography
the legal system are intimately intertwined. Results (GC), high performance liquid chromatography
obtained from the examination and analysis of foren- (HPLC) can be used for both qualitative and quantit-
sic samples and the forensic samples themselves com- ative analyses. However, 80}85% of all known com-
prise evidence of a crime. It is the individualization of pounds are not amenable to GC. The stationary and
the sample, i.e. the singular association between the mobile-phase combinations available and the many
samples(s) and an illegal act, that is unique to forensic detection modes possible make HPLC a universal
science. Because of the legal consequences, results separation scheme. Unlike GC, it is not limited by the
require a high degree of certainty and the techniques volatility or thermal stability of an analyte. HPLC can
used must be admissible in court. The Daubert rule, analyse solutes encompassing a wide molecular
a 1993 decision upheld by the US Supreme Court, weight range, from monatomic species to proteins.
assigns to the judge the role of determining admissi- A range of solute hydrophobicities and polarities can
bility of scientiRc evidence. Among the factors con- be accommodated, from acidic and basic species that
sidered by the judge are: (i) whether the technique incorporate many drugs of abuse, pharmaceuticals,
has been tested and subjected to peer review; (ii) dyes and food colourings, to neutral and/or hydro-
whether error rates have been deRned; (iii) whether phobic molecules such as pesticides and herbicides,
standards controlling the operation of a technique hydrocarbons in petroleum products and carotenoids
exist; and (iv) whether the technique has been widely in foods. In situations where GC can determine cer-
accepted in the scientiRc community. Techniques tain compounds more readily with greater selectivity,
used in a forensic laboratory may be applied to the resolution and sensitivity, HPLC offers a secondary,
investigation of a wide variety of crimes. Some exam- conRrmatory method. In other cases, the use of HPLC
ples are illegal drug use, counterfeiting, arson, tam- avoids sample derivatization required by GC, and
pering, fraud, poisoning, terrorism and environ- eliminates steps that could contribute to sample loss
mental crimes. It is this diversity of cases, variety of and increase analysis time. Another advance in HPLC
sample matrices and the staggering number of poten- in recent years involves the introduction of narrow-
tial analytes that necessitate a continuous evaluation bore (2.1 mm inner diameter) and microbore (1 mm
of the testing that is needed to constitute proof in each inner diameter) columns. These smaller columns de-
situation. crease the sample size required for injection and in-
Analytical requests in a forensics laboratory may crease mass detection sensitivity versus the typical
be classiRed into four categories: (i) screen for the 4.6 mm inner diameter analytical columns.
III / FORENSIC SCIENCES / Liquid Chromatography 2871
Thin-layer chromatography (TLC) is also utilized Analytes must be in solution for determination by
in forensic chemistry, particularly for sample screen- HPLC. Sample preparation is necessary to remove
ing. Although TLC permits analysis of many samples compounds such as proteins that might damage an
at one time, facilitating the side-by-side comparison HPLC column, as well as compounds that interfere
of suspect and authentic samples, it can suffer from with an analysis. Liquid}liquid extractions (LLE) are
a lack of resolution and from difRculties in both commonly used, and can be manipulated by choice of
quantiRcation and isolation of an individual compon- solvent, addition of salts (salting-out effect) and con-
ent. Immunoassays are generally more sensitive; how- trol of pH. For biological matrices, extractions using
ever, they may provide class-only determinations, chloroform/2-propanol/n-heptane under alkaline
may be prone to interferences and may not be avail- conditions provide clean extracts with good recove-
able for classes such as neuroleptics and -blockers. ries of basic and neutral compounds. However, acidic
HPLC is useful for sample screening, comparison and compounds such as barbiturates and salicylates are
quantiRcation, and fraction collection for further poorly recovered (20}50%). LLE is not easily auto-
analysis is more straightforward. mated and can require large volumes of solvent.
The UV-visible detector, and more recently the Solid-phase extraction (SPE) cartridges are now wide-
diode array detector (DAD), are the most commonly ly used in toxicology screens, mainly for low viscosity
used detectors for HPLC analyses. As an alternative, samples such as urine or serum. SPE has the advant-
one may utilize more speciRc devices such as Suores- ages of higher efRciencies and selectivity, lower sol-
cence, electrochemical, chiral or mass spectrometric vent volume requirements, absence of emulsions, and
detectors. Because these detectors take advantage automation options. However, the packing materials
of speciRc molecular characteristics of the analyte(s) can be irreproducible, even within batches of the
of interest, they are less susceptible to back- same brand, resulting in variable recoveries and poor
ground interferences from sample matrix, and tend analytical reproducibility. The use of an internal stan-
to be more sensitive. The choice of a particular de- dard is highly recommended for quantitative
tector and method depends upon the requirements of results. SPE cartridges should not be re-used due to
the case in hand } whether the sample is being decreased extraction performance and increased pos-
screened for unknowns, analysed for a particular sibility of the introduction of contaminants.
compound, compared against another sample, or
quantiRed.
General Unknowns
Screening for unknowns is a very challenging task due
Sample Preparation to the vast number of potential contaminants. Screen-
Each manipulation of the forensic sample may irre- ing methods in a forensic laboratory are designed to
versibly alter the evidence and introduces the possibil- detect the most relevant drugs and potentially hazard-
ity of incomplete analyte recovery and inadvertent ous chemicals. Often, screens are performed in re-
contamination. Therefore, sample preparation response to a crisis such as an acute poisoning and as
quires careful consideration and always follows a such require rapid response. While immunoassay
preliminary visual, and perhaps microscopic, exam- techniques and TLC remain invaluable for initial
ination. Sample preparation steps may also affect the screening, these methods must be supplemented by
form of the analyte, which is problematic for speci- HPLC-DAD for those analytes for which the initial
ation work and subsequent toxicological evaluation. screen does not offer sufRcient selectivity or sensitiv-
Additional complications in forensic work are the ity. IdentiRcation of a compound by HPLC-DAD is
variety of sample matrices (drugs, body Suids, food, based on retention time match and spectral match, as
soils, etc.) and limited sample sizes (arson residues, shown in Figure 1.
traces of blood in a syringe, a spot on blotter paper, Additionally, plotting the ratios of absorbance
etc.) frequently encountered. Ideally, a portion of the measurements obtained during a chromatographic
sample should be reserved in case of trial to allow analysis, taken at well-separated and characteristic
independent analysis. If appropriate, the sample can wavelengths, permits evaluation of interferences and
be homogenized. However, portions may need to be a more conRdent identiRcation. For example, vari-
analysed separately to characterize the sample accu- ation in the ratio across a peak indicates co-elution,
rately. Individual samplings may also be advisable as seen in Figure 2.
when there are visual differences within portions of Since forensic screens for unknowns are often per-
a sample. Sampling in the vicinity of a visual conformed in biological matrices, which are inherently
taminant reduces dilution with the matrix and im- variable, the analysis of blanks is an additional safe-
proves detection limits. guard against false positives that could be caused by
Figure 1 Spectrochromatogram showing spectral and chromatographic data for the separation of six benzodiazepines. Solutes: 1,
midazolam; 2, flurazepam; 3, oxazepam; 4, nitrazepam; 5, alprazolam; 6, clonazepam. Column, 25 cm Lichrosorb RP-8; mobile phase,
35% CH3CN in 0.05 mol L\1 KH2PO4}H3PO4 buffer, pH 3; flow rate, 1.5 mL min\1; spectra collected between 190 and 400 nm.
Reproduced from Logan (1994), with permission from Elsevier Science.
the effect of poisons on the matrix or due to putrefac- Statistical toxicological analysis (STA), a general
tion processes. Blanks should include reagents, and screening method for toxins in biological matrices as
matrix that is contaminant-free. Screening methods described by Tracqui et al., often utilizes HPLC-
also need to detect indirect indicators of the com- DAD. Many laboratories have had success in creating
pound of interest. A screen of biological Suids should their own databases and/or using multicomponent
include major drug/poison metabolites such as ben- analysis for the identiRcation of hundreds of substan-
zoylecgonine, the primary indicator of cocaine use, ces from several classes in one run. Widespread use of
which is found in urine. The addition of bleach to HPLC-DAD for STA requires libraries that can be
carbonated beverages leads to the formation of chlor- shared among laboratories. Unfortunately, libraries
ate, chloride and sometimes chloroform. are not as common for HPLC as they are in GC work.
Because the mobile phase in HPLC interacts much
more strongly with analytes than the carrier gas in
GC, small deviations in chromatographic conditions
such as column type and batch, mobile-phase com-
position and pH, temperature and Sow rate may
affect retention time. The use of retention indices is
necessary to minimize interlaboratory differences.
Bogusz et al. proposed the use of a 1-nitroalkane
index scale for toxicological screens since the C1 to
C6 homologues are commercially available and have
high UV absorbance between 200 and 220 nm. The
retention times of 1-nitroalkanes are not affected by
pH changes between 3.2 and 8.5, but are affected by
changes in acetonitrile concentration. Because com-
Figure 2 Use of wavelength ratioing to indicate peak impurity in
the analysis of tricyclic antidepressants and metabolites by LC- pounds are affected differently by changes in
photodiode array detection. (A) Chromatogram at 252 nm; (B) chromatographic conditions, the use of selected drugs
ratiogram 252 nm/230 nm. Asymmetric ratio of peak 3 indicates as retention markers to correct retention indices im-
inhomogeneity and co-elution. Solutes: 1, 10-hydroxynortripty- proves accuracy and precision. One toxicological
line; 2, 10-hydroxyamitriptyline; 3, protryptyline/imipramine (co-
screening library that includes 900 substances is
elution); 4, nortriptyline; 5, amitriptyline. Column, 25 cm Lichros-
pher RP-8 (CH100); mobile phase, 40% CH3CN in 0.05 mol L\1 available commercially. To improve the possibility of
phosphate buffer, pH 3; flow rate, 2.0 mL min\1. Reproduced obtaining a match, it is important to use the same
from Logan (1994), with permission from Elsevier Science. chromatographic conditions as the library.
Figure 3 Chromatograms obtained from extracts of two gastric contents (A and B). Solutes; 1, zopiclone, 2, mesoridazine; 3,
perphenazine; 4, thioridazine; 5 and 6, co-elution of ketoprofen and naproxen. The spectral library was developed over the wavelength
range of 210}367 nm. Solute identification utilized a retention time window of $5% and a peak purity parameter of $1 nm. Columns:
250;4.6 mm i.d. Supelcosil LC-DP (diphenyl) and 250;4.0 mm. i.d. LiChrospher 100 RP-8; mobile phase, isocratic CH3CN}0.025%
(v/v) H3PO4}TEA buffer, pH 3.4; flow rate, 0.6 mL min\1; detection, 229 nm. Reproduced from Koves (1995) with permission from
Elsevier Science.
Figure 3 is an illustration of the process used to purity should be considered. Consequently, a library
identify substances found in gastric contents. The search may yield as many as 10 possible matches.
search window needs to be set wide (20%), and peak Spectra should be compared between a sample and
a standard run in-house for a positive identiRcation. pending on detector selection. For example, mor-
Even in the absence of a positive identiRcation, the phine can be analysed directly in poppy seed extract
diode array spectrum can give class indications. The with electrochemical detection because it is easily
analysis can then be pursued by modifying the HPLC oxidized at low potentials, unlike most opium alka-
conditions or by utilizing another technique. loids from natural products. Figure 4 compares the
sensitivity and selectivity obtained with UV, Suores-
cence and electrochemical detection of various
Analysis for a Known Analyte alkaloids.
HPLC separations can resolve lysergic acid di-
or Analyte Class ethylamide (LSD) from ergot alkaloids. Because LSD is
In contrast to screening for unknowns, sample prep- typically ingested in small amounts, Suorescence de-
aration and separations can be optimized for the tection is commonly used for its sensitivity and selec-
analysis of a known analyte or class of analytes. In tivity. Analysis of opium alkaloids by HPLC must
order to avoid interferences and to maximize analyte also separate caffeine, quinine and strychnine }
recovery from the sample matrix, experimental con- common additives or diluents.
ditions can be tailored for the class of compounds of Although there are more than 2000 known ster-
interest. It is often necessary to determine compo- oids, only a portion are controlled substances. Analy-
nents such as diluents, excipients, metabolites and sis of steroids is required in a variety of matrices,
synthesis or degradation products. Table 1 lists including dosage forms, oils, body Suids and tissues.
examples of compounds that are analysed by HPLC Spot tests are useful for the rapid initial veriRcation of
in forensic laboratories. HPLC may be not be the the presence of steroids. A more speciRc identiRcation
primary method for all of these analytes but may is possible by either GC or HPLC with similar resolv-
instead be the conRrmatory technique. ing power but different co-eluting pairs. HPLC-DAD
Analysis of drugs of abuse constitutes a major por- adds the potential of distinguishing between some
tion of forensic work. While UV may lack the neces- steroids based on differences in UV spectra.
sary sensitivity, electrochemical detection offers sen- In drug abuse cases, creatinine is analysed using ion
sitivity and selectivity for compounds such as mor- pair reversed-phase separation and UV detection at
phine, benzodiazepines, cannabinoids, hallucinogens, 220 nm to determine if urine samples have been
fentanyl and some cyclic antidepressants. Additional diluted. The HPLC method suffers from fewer inter-
compounds can be analysed using post-column ferences than other methods. HPLC reversed-phase
photolytic derivatization followed by electrochemical or ion exchange separations of proteins for blood
detection. Sample preparation may be minimized de- grouping or species identiRcation are rapid and efR-
cient.
Table 1 Types of analytes determined by HPLC in forensic Pesticides, herbicides and rodenticides may be de-
laboratories termined by HPLC in poisoning cases. Warfarin and
its metabolites have been identiRed in matrices such
Analgesics as urine and food. Carbamates are more readily ana-
Anticonvulsants
lysed by HPLC with post-column derivatization than
Antidiabetic drugs
Creatinine by GC because they are thermally labile.
Digitalis glycosides Explosives are difRcult to analyse by GC due to
Drugs of abuse (barbiturates, benzodiazepines, cannabinoids, their thermal instability. HPLC is used for the analy-
cocaine and related compounds, LSD, opium alkaloids) sis of nitroglycerin, propellants, stabilizers, plas-
Dyes and colourings (natural and synthetic)
ticizers and weapon discharge residues. Inorganic ex-
Ergot alkaloids
Explosives, propellants, stabilizers plosives and explosive residues can be determined
Hydrocarbons (petroleum distillates, engine oils, greases) with ion chromatography. Explosives such as am-
Inks monium nitrate and residues such as chloride, chlor-
Inorganic and organic anions (fluoride, chloride, phosphate, ate, sulRde and sulfate can be determined by anion
sulfate, azide, citrate)
exchange with UV or conductivity detection, as illus-
Inorganic cations (sodium, potassium, calcium, ammonium)
Pesticides, herbicides, rodenticides trated in Figure 5.
Pharmaceuticals Separations in ion chromatography (IC) are based
Phenothiazines on ion exchange, ion exclusion and reversed-phase
Plastics, plasticizers, polymers adsorption. The use of a suppressor column to reduce
Proteins
the mobile-phase background chemically and in-
Sugars
Tricyclic antidepressants crease the analyte signal permits conductivity detec-
tion of inorganic ions. There are also methods known
Figure 4 Comparison of detector signals. (A) UV; (B) fluorescence; (C) electrochemical. Chromatograms of orange juice samples
spiked with: A, morphine; B, codeine; C, eserine; D, apomorphine. Column, 250;4.6 mm i.d. Interaction chemicals C18; mobile phase,
55% methanol, 15 mmol L\1 KH2PO4, 3.75 mmol L\1 1-octanesulfonic acid, 7.5 mmol L\1 KCl, adjusted to pH 4.00 with 10% H3PO4;
flow rate, 1.0 mL min\1; 303C. Detection: UV, 254 nm; fluorescence, ex"254 nm, em"408 nm; electrochemical, 1.2 V vs. Ag/AgCl
reference electrode with a glassy carbon working electrode. Reproduced from Lin (1993) with permission from Elsevier Science.
as single column which do not utilize a suppressor, Although conductivity detectors are the most com-
but instead utilize low capacity columns and low monly used, a variety of other detectors are avail-
ionic strength, low conductance mobile phases. able: UV-visible (direct or following post-column
Figure 5 Analysis of residue taken from a black powder pipe bomb using ion chromatography. Solutes: 1, chloride; 2, nitrite;
3, nitrate; 4, sulfate; 5, sulfide; 6, hydrogen carbonate. Column, Vydac 302IC4.6; mobile phase 0.75 g isophthalic acid in 3 L H2O,
adjusted to pH 4.6 with 2 mol L\1 KOH; flow rate, 2.5 mL min\1; detection, 280 nm. Adapted from Hargadon and McCord (1992) with
permission from Elsevier Science.
reactions), amperometry, Suorescence and atomic index detection can be used to analyse the types and
spectroscopy. IC analysis of strong acids and alkalis amounts of sugars found as diluents in illicit street
can be important in poisoning cases. Speciation by IC drugs, possibly linking seized evidence from several
of compounds such as arsenic may be important due cases. Figure 7 demonstrates an excellent separation
to the higher toxicity of inorganic anions compared of cannabinoids and their metabolites. Comparative
to the methylated forms. IC can also be used to analysis of cannabinoids in cannabis samples by
determine counterions of drugs, as well as cleaning HPLC has been shown to connect suppliers with
products and their components. customers.
Through the knowledge of peak area ratios among
carotenoids that occur naturally in orange juice,
Comparison of Samples a suspect orange juice can be analysed to determine
HPLC analysis may be required to compare samples, whether carotenoids have been added to enhance the
either to differentiate one from another or to trace the colour. Dyes extracted from Rbres gathered at a crime
source of a sample. Although retention times can be scene and from Rbres from a suspect could be proRled
used for tentative identiRcation of speciRc com- and compared. The proRle of the subunits of haemo-
pounds, it is not always necessary to identify every globin present in a blood drop permits its source
component in the sample. Often, pattern recognition identiRcation as human adult, human neonatal or
will sufRce when comparing the content of a class of animal.
compounds from one sample to the next. IC has been When comparing chromatograms, the presence of
used in the analysis of sugars present in suspect infant extra peaks in a proRle may suggest the deliberate
formulas, as seen in Figure 6. HPLC with refractive addition of a foreign substance, or simply sample
Figure 6 Comparison of sugar profiles using IC; (A) a known formula and (B,C) suspect infant formulas labelled as the known.
Column, 4;250 mm Dionex Carbopac PA-1; mobile phase, 150 mmol L\1 NaOH in 0}600 mmol L\1 sodium acetate; flow rate,
1.0 mL min\1; detection, pulsed electrochemical detection with a gold working electrode and a pH/Ag/AgCl reference electrode.
Adapted from Kaine and Wolnik (1998) with permission from Elsevier Science.
Figure 7 HPLC of cannabis resin at (A) 254 nm and 263C, (B) 220 nm and 263C. Solutes: 1, cannabidiol and cannabigerol (shoulder);
2, cannabidiolic acid; 3, cannabinol and cannabigerolic acid; 4, tetrahydrocannabinol; 5, cannabichromene; 6, cannabinolic acid; 7,
tetrahydrocannabinolic acid; 8, cannabichromenic acid; 9, di-n-octyl phthalate (internal standard). Chromatographic conditions: 100 mg
resin extracted with 1 mL chloroform}methanol (1:9) containing 8 g L\1 di-n-octyl phthalate; 2 L extract injected. Column,
250;4.9 mm Partisil 5 C18; mobile phase, 80% methanol}20% 0.02 mol L\1 H2SO4; flow rate, 2 mL min\1. Reproduced from Smith
and Vaughan (1976) with permission from Elsevier Science.
degradation. The lack of expected peaks may analysis of blood group and Rh antigens if the distri-
suggest that the sample is not what it claims to be. bution of these factors among the population is
However, because many species absorb at lower known.
wavelengths, the mobile-phase composition may
affect the outcome of an analysis. The molar absor-
ptivity of a compound may change with small cha-
Veri\cation of Analyte Identity
nges in the mobile-phase composition, affecting A high degree of certainty is required for the identi-
quantiRcation. A high background absorbance from Rcation of solutes in forensic cases in order to defend
the mobile phase may obscure species that are present the work in court. An orthogonal technique may be
in the sample at low concentrations. This could used for conRrmation, and the use of selective de-
lead to the erroneous conclusion that there are no tectors aids in the certainty of identiRcation. In
apparent differences between suspect and authentic HPLC, diode array detectors add the ability to match
samples. spectra, while Suorescence and electrochemical de-
Whenever the analysis of samples requires a long tectors are more speciRc. Both the excitation and
period of time, e.g. for a large number of samples, emission wavelengths can be selected for Suorescence
the variation of retention times due to chromato- detection. The oxidation potential can be adjusted to
graphic factors must be considered. An internal stan- reduce interferences of some analytes determined
dard may be added to samples, or a control sample electrochemically.
may be analysed with each set of suspect samples. GC-mass spectrometry (GC-MS) and GC-MS-MS
For comparative analysis, there is added importance are widely used in forensic analysis to verify analyte
to representative sampling and replicate analyses. identiRcation, particularly in cases concerning illegal
It is easier to state that samples are different, rather drugs and drugs of abuse. The retention time identi-
than identical. Results of comparative analyses indic- Res an analyte, and the mass spectrum serves as con-
ating two samples appear the same might be worded Rrmation. Until recently, the argument against the use
as ‘the samples were analytically indistinguishable of HPLC as the primary method in forensic analysis
using techniques x, y, z’. In some cases, the individ- was the difRculty of solute conRrmation. Tremendous
ualization of a sample can be quantiRed. For improvements in instrumentation and interface de-
example, a statistical probability of a blood sample signs have made the coupling of HPLC to a mass
matching a speciRc person can be based upon the spectrometer straightforward, particularly with the
recent introduction of relatively inexpensive bench- consequently is widely used in forensic laboratories
top models. today.
The ionization methods that are available per-
mit the analysis of a wide variety of compounds. See also: III / Carbamate Insecticides in Foodstuff:
Particle beam is effective for moderately polar Chromatography & Immunoassay. Clinical Diagnosis:
compounds (certain steroids and rodenticides), Chromatography. Explosives: Gas Chromatography;
operates more efRciently with narrow-bore columns, Liquid Chromatography; Thin-Layer (Planar) Chromatog-
and causes fragmentation during the ionization raphy. Forensic Toxicology: Thin-Layer (Planar)
process. Both atmospheric pressure chemical ioniz- Chromatography. Heroin: Liquid Chromatography
ation (APCI) and electrospray ionization (ESI) are and Capillary Electrophoresis. Toxicological Analy-
sis: Liquid Chromatography. Steroids: Gas Chrom-
amenable to gradient elution methods, thereby per-
atography; Liquid Chromatography and Thin-Layer
mitting their use in the analysis of a group of com- (Planar) Chromatography.
pounds. APCI is compatible with conventional col-
umns and is used primarily to determine moderately
polar analytes that are not thermally labile (clen- Further Reading
buterol, basic pharmaceuticals). Polar (conjugated
Bogusz M, Franke JP, de Zeeuw RA and Erkens M (1993)
oestrogens, proteins, peptides) and/or thermally frag- An overview on the standardization of chromato-
ile molecules (glucuronide metabolites of morphine graphic methods for screening analysis in toxicology
and codeine) are analysed more effectively by ESI. ESI by means of retention indices and secondary stan-
requires narrow- and microbore columns. In any of dards. Fresenius Journal of Analytical Chemistry 347:
these methods, quantiRcation can be performed in 73}81.
either the full scan mode or in the selected ion Bohan TL and Heels EJ (1995) The case against Daubert:
monitoring (SIM) mode. SIM offers greater sensitiv- the new scientiRc evidence `standarda and the standards
ity (10}100;), comparable to that obtained with of the several states. Journal of Forensic Sciences 40:
GC-MS. 1030}1044.
The utilization of MS-MS provides even greater Busch KL, Glish GK and McLuckey SA (1988) Mass
Spectrometry/Mass Spectrometry: Techniques and
speciRcity, further decreasing the chance of the incor-
Applications of Tandem Mass Spectrometry. New York:
rect identiRcation of an analyte. Even if two com- VCH.
pounds co-elute and have the same [M#H]# ion DeForest P, Gaensslen RE and Lee HC (1983) Forensic
exiting the Rrst quadrupole, it is unlikely that they Science: An Introduction to Criminalistics. New York:
would produce the same fragmentation pattern in the McGraw Hill.
third quadrupole. The technique called selected reac- Hargadon KA and McCord BR (1992) Explosive residue
tion monitoring utilizes the known losses that occur analysis by capillary electrophoresis and ion chromato-
during fragmentation of an analyte or a particular graphy. Journal of Chromatography 602: 241}247.
group of analytes. The method is useful when co- Kaine LA and Wolnik KA (1998) Detection of counterfeit
elution of two or more compounds is suspected. An- and relabeled infant formulas using high pH anion ex-
other beneRt to this method is that its sensitivity is change chromatography-pulsed amperometric detection
for the determination of sugar proRles. Journal of
typically 10}100 times greater than that obtained in
the full scan mode. Koves EM (1995) Use of high-performance liquid
chromatography-diode array detection in forensic
Conclusions toxicology. Journal of Chromatography A 692:
103}119.
Numerous aspects of separation science are appli- Lin LA (1993) Detection of alkaloids in foods with multi-
cable to forensic science. Because HPLC is so versatile detector high-performance liquid chromatographic sys-
and can be used to determine so many different com- tem. Journal of Chromatography 632: 69}78.
pounds, the technique is particularly well suited to Logan BK (1994) Liquid chromatography with photodiode
the demands of a forensic laboratory. Both qualitat- array spectrophotometic detection in the forensic
ive and quantitative information can be obtained, sciences. Analytical Chimica Acta 288: 111}122.
Lurie IS, Sperling AR and Meyers RP (1994) The deter-
often with minimal sample preparation. Because only
mination of anabolic steroids by MECC, gradient
small volumes are needed for analysis, sample con- HPLC, and capillary GC. Journal of Forensic Sciences
sumption can be minimized. Eluting fractions can be 39: 74}85.
collected for further analysis } an important consid- Selevka CM and Krull IS (1987) The forensic determination
eration when dealing with trace evidence. HPLC of- of drugs of abuse using liquid chromatography with
fers a cost-effective technique with the ruggedness electrochemical detection: a review. Journal of Liquid
and reliability necessary for forensic testing and Chromatography 10: 345}375.
III / FORENSIC TOXICOLOGY: THIN-LAYER (PLANAR) CHROMATOGRAPHY 2879
Smith RN (1982) Forensic applications of high- Tracqui A, Kintz P and Mangin P (1995) Systematic toxico-
performance liquid chromatography. In: Saferstein R. logical analysis using HPLC/DAD. Journal of Forensic
(ed.) Forensic Science Handbook. New Jersey: Prentice Sciences 40: 254}262.
Hall. Weiss J (1995) Ion Chromatography, 2nd edn. New York:
Smith RN and Vaughan CG (1976) High-pressure liquid VCH.
chromatogrphy of cannabis. Quantitative analysis of White P (1998) Crime Scene to Court. The Essentials
acidic and neutral cannabinoids. Journal of Chromato- of Forensic Science. Cambridge: Royal Society of
graphy 129: 347}354. Chemistry.
FORENSIC TOXICOLOGY: THIN-LAYER

I. OjanperaK , University of Helsinki, Helsinki, Finland The commonly recognized advantages of classical
manual TLC are high throughput, low cost, easy
sample preparation and versatile visual detection pos-
sibilities. Instrumental TLC extends the scope to re-
The selection of appropriate analytical methodology producible quantitative analysis and allows the utiliz-
for forensic toxicological investigations depends on ation of in situ UV spectral information for identiRca-
the scope of the laboratory. Postmortem forensic tion. The main disadvantage of TLC is low chromato-
toxicology investigates the cause of death, and conse- graphic resolution, which can be partly overcome by
quently a very broad-range screening is needed to instrumental techniques. Another disadvantage is
detect all potential poisons. In trafRc toxicology, that quantitative calibration curves are not reproduc-
only such substances are relevant which may impair ible enough to be stored, making it necessary to
the driver’s ability to control the vehicle. Doping co-analyse several standards along with samples on
control focuses on those substances that have been each TLC plate. Most of the substances frequently
banned by the International Olympic Committee. encountered in forensic toxicology can be readily
Prisoners and rehabilitation clinic patients are tested analysed by TLC. These include therapeutic drugs,
for psychotropic drugs, whereas the US Mandatory drugs of abuse, pesticides and naturally occurring
Guidelines for Federal Workplace Drug Testing alkaloids, which are all relatively small molecular
Programs are limited to the major drugs of abuse, weight organic compounds with functional groups
cannabis, cocaine, amphetamine, opiates and phen- amenable to visualization by colour reactions.
cyclidine. It is practical to divide the discussion of TLC in
Thin-layer chromatography (TLC) has found ex- forensic toxicology into two categories, the broad-
tensive use in forensic toxicology since the early scale screening analysis and target analysis. The for-
1960s when the famous book of Stahl made the mer approach is related to the concepts of systematic
technique well known. The Rrst edition of the toxicological analysis or general unknown, i.e. the
classic laboratory manual by AS Curry, Poison De- search for a rational qualitative analysis strategy for
tection in Human Organs from 1963 (Charles C. hundreds of potential poisons. TLC drug screening is
Thomas, SpringReld, IL), still relies on paper often performed in urine or liver, where the drug
chromatography but the second edition in 1969 util- concentrations are higher than in the blood. In target
izes TLC as a major technique for drugs. Gas analysis, the aim is speciRcally to detect and often
chromatography (GC) and gas chromatography} also to quantify a substance or a limited number of
mass spectrometry (GC-MS) in the 1970s, and espe- substances.
cially high performance liquid chromatography
(HPLC) in the 1980s gradually began to replace
TLC but today the planar technique is having a re- Broad-scale Screening Analysis
naissance due to the progress in instrumentation and
Chromatographic Systems
software. From 1990 to 1996, 26% of published
TLC applications were in the Reld of medical, clinical Evaluation of systems The rational selection of TLC
and biological analysis, which also includes forensic systems for screening analysis differs from the opt-
toxicology. imization of the separation of a few-component
Smith RN (1982) Forensic applications of high- Tracqui A, Kintz P and Mangin P (1995) Systematic toxico-
performance liquid chromatography. In: Saferstein R. logical analysis using HPLC/DAD. Journal of Forensic
(ed.) Forensic Science Handbook. New Jersey: Prentice Sciences 40: 254}262.
Hall. Weiss J (1995) Ion Chromatography, 2nd edn. New York:
Smith RN and Vaughan CG (1976) High-pressure liquid VCH.
chromatogrphy of cannabis. Quantitative analysis of White P (1998) Crime Scene to Court. The Essentials
acidic and neutral cannabinoids. Journal of Chromato- of Forensic Science. Cambridge: Royal Society of
graphy 129: 347}354. Chemistry.
FORENSIC TOXICOLOGY: THIN-LAYER

I. OjanperaK , University of Helsinki, Helsinki, Finland The commonly recognized advantages of classical
manual TLC are high throughput, low cost, easy
sample preparation and versatile visual detection pos-
sibilities. Instrumental TLC extends the scope to re-
The selection of appropriate analytical methodology producible quantitative analysis and allows the utiliz-
for forensic toxicological investigations depends on ation of in situ UV spectral information for identiRca-
the scope of the laboratory. Postmortem forensic tion. The main disadvantage of TLC is low chromato-
toxicology investigates the cause of death, and conse- graphic resolution, which can be partly overcome by
quently a very broad-range screening is needed to instrumental techniques. Another disadvantage is
detect all potential poisons. In trafRc toxicology, that quantitative calibration curves are not reproduc-
only such substances are relevant which may impair ible enough to be stored, making it necessary to
the driver’s ability to control the vehicle. Doping co-analyse several standards along with samples on
control focuses on those substances that have been each TLC plate. Most of the substances frequently
banned by the International Olympic Committee. encountered in forensic toxicology can be readily
Prisoners and rehabilitation clinic patients are tested analysed by TLC. These include therapeutic drugs,
for psychotropic drugs, whereas the US Mandatory drugs of abuse, pesticides and naturally occurring
Guidelines for Federal Workplace Drug Testing alkaloids, which are all relatively small molecular
Programs are limited to the major drugs of abuse, weight organic compounds with functional groups
cannabis, cocaine, amphetamine, opiates and phen- amenable to visualization by colour reactions.
cyclidine. It is practical to divide the discussion of TLC in
Thin-layer chromatography (TLC) has found ex- forensic toxicology into two categories, the broad-
tensive use in forensic toxicology since the early scale screening analysis and target analysis. The for-
1960s when the famous book of Stahl made the mer approach is related to the concepts of systematic
technique well known. The Rrst edition of the toxicological analysis or general unknown, i.e. the
classic laboratory manual by AS Curry, Poison De- search for a rational qualitative analysis strategy for
tection in Human Organs from 1963 (Charles C. hundreds of potential poisons. TLC drug screening is
Thomas, SpringReld, IL), still relies on paper often performed in urine or liver, where the drug
chromatography but the second edition in 1969 util- concentrations are higher than in the blood. In target
izes TLC as a major technique for drugs. Gas analysis, the aim is speciRcally to detect and often
chromatography (GC) and gas chromatography} also to quantify a substance or a limited number of
mass spectrometry (GC-MS) in the 1970s, and espe- substances.
cially high performance liquid chromatography
(HPLC) in the 1980s gradually began to replace
TLC but today the planar technique is having a re- Broad-scale Screening Analysis
naissance due to the progress in instrumentation and
Chromatographic Systems
software. From 1990 to 1996, 26% of published
TLC applications were in the Reld of medical, clinical Evaluation of systems The rational selection of TLC
and biological analysis, which also includes forensic systems for screening analysis differs from the opt-
toxicology. imization of the separation of a few-component
2880 III / FORENSIC TOXICOLOGY: THIN-LAYER (PLANAR) CHROMATOGRAPHY
mixture. In screening systems, the most important formation content, quotient of distribution equality
features are the distribution of RF values across the and principal component analysis. The (MLL)
plate, the reproducibility of the measurement of those method has found widespread use, and it can also be
values, and the correlation of chromatographic prop- used in computerized substance identiRcation. The
erties between systems. The computational methods MLL approach is not related to the separation num-
capable of taking into account these features include ber (SN), i.e. the number of spots that can be separ-
discriminating power, mean list length (MLL), inated by a system with a certain resolution. Table 1
Table 1 TLC systems for broad-scale toxicological screening analysis
Mobile phase Stationary phase Correction hR cF Application

standards
1 Chloroform}acetone Silica gel Paracetamol 15 Acidic and neutral drugs

80#20 Clonazepam 35
Secobarbital 55
Methylphenobarbital 70
2 Ethyl acetate Silica gel Sulfathiazole 20 Acidic and neutral drugs

Phenacetin 38
Salicylamide 55
Secobarbital 68
3 Ethyl acetate}methanol} Silica gel Hydrochlorothiazide 11 Acidic and neutral drugs

conc. ammonia Sulfafurazole 33
85#10#5 Phenacetim 52
Prazepam 72
4 Methanol}water Silica gel Diazepam 16 Acidic and neutral drugs

65#35 RP 18 Secobarbital 35
Phenobarbital 54
Paracetamol 74
5 Methanol}water} Silica gel Hydroxyzine 20 Basic, amphoteric and

conc. hydrochloric acid RP 18 Lignocaine 46 quaternary drugs
50#50#1 Codeine 66
Morphine 81
6 Toluene}acetone}ethanol} Silica gel Codeine 16 Basic and neutral drugs

conc. ammonia Promazine 36
45#45#7#3 Clomipramine 49
Cocaine 66
7 Ethyl acetate}methanol} Silica gel Morphine 20 Basic and neutral drugs

ammonia Codeine 35
85#10#5 Hydroxyzine 53
Trimipramine 80
8 Methanol Silica gel Codeine 20 Basic and neutral drugs

Trimipramine 36
Hydroxyzine 56
Diazepam 82
9 Methanol}ammonia Silica gela Atropine 18 Basic and neutral drugs

100#1.5 Codeine 33
Chlorprothixene 56
Diazepam 75
10 Cyclohexane}toluene} Silica gela Codeine 6 Basic and neutral drugs

diethylamine Desipramine 20
75#15#10 Prazepam 36
Trimipramine 62
a
Impregnated with 0.1 mol L\1 KOH and dried.
shows TLC systems for broad-scale toxicological using aqueous mobile phases, are less dependent on
screening analysis, chosen partly on grounds of the humidity. The RF and hRcF values can be determined
MLL method, while the corresponding RF libraries manually or by using a scanning densitometer. The
can be found from the books of de Zeeuw et al. and correction of RF values makes it possible to obtain
Fried and Sherma. For acidic and neutral drugs, rec- reproducible values in varying conditions and allows
ommended combinations of systems which posses the use of the large hRcF libraries even in interlabora-
low mutual correlation are 2 and 3, and 1 and 3, for tory use.
basic drugs 5 and 6, and 8 and 10. Commercial software are available that utilizes the
concept of hRcF for identiRcation. Chrom TOX
RF correction In screening analysis, where RF libra- (Merck Tox Screening System, Merck, Darmstadt,
ries of hundreds of compounds are utilized, the repro- Germany) is statistical search software that utilizes
ducibility of the values is an essential factor. TLC is the MLL method for identiRcation by RF values and
an open technique, and the RF values, and conse- digitally coded colour reactions, giving a hit list of
quently the separation, are affected by environmental candidates with probability values. The software is
factors, such as humidity, layer activity and temper- also capable of adding information from other ana-
ature. In contrast to column chromatography, the use lytical techniques, such as retention indices from GC,
of a single RF standard for compensating the vari- molecular weights from MS and UV spectra from
ation, corresponding to the relative retention time, HPLC. A drawback is that the TLC data have to be
may produce erroneous results. The method, which fed manually. CATS software (Camag, Muttenz,
uses three to Rve correction standards that are struc- Switzerland) combines instrumental densitometric
turally close to the analytes and linear interpolation evaluation of chromatographic plates with substance
between the standards, is now generally accepted to identiRcation by hRcF values and in situ UV spectra
obtain corrected RF values (hRcF). Table 1 indicates (see below).
the correction standards chosen for the screening
systems listed. Visualization reagents The possibility of using vis-
ualization reagents for the detection and identi-
Identi\cation
Rcation of fractions is a unique feature of TLC.
Migration distance By carefully adjusting the chro- Post-chromatography derivatization by spraying or
matographic and environmental conditions it is pos- dipping has been used more extensively than pre-
sible to obtain reproducible results with precoated chromatography derivatization. The limits of detec-
plates. Reversed-phase (RP) layers show more batch- tion by colour reactions generally range from 0.1
to-batch variation than silica gel but RP separations, to 1 g per fraction and by Suorescence reactions,
Table 2 Visualization reagents for broad-scale screening analysis
Reagent Application
Bratton-Marshall reagent (diazotation and coupling) Benzophenones (from benzodiazepines), sulfonamides

7-Chloro-4-nitrobenzo-2-oxa-1,3-diazole (NBD-Cl) Amphetamines, amino acids
2,6-Dibromoquinone-4-chlorimide (Gibbs reagent) Pesticides
2,6-Dichlorophenol-indophenol (Tillmann’s reagent) Organic acids
p -Dimethylaminobenzaldehyde (Van Urk’s reagent) Drugs, sulfonamides, pesticides
Dragendorff’s reagent Drugs and alkaloids
Fast Black K salt Amphetamines, adrenergic -blocking drugs, nor-metabolites
Fast Blue B salt, Fast Blue BB salt Cannabinoids
Fluorescamine Amphetamines, amino acids, sulfonamides
Forrest reagent Phenothiazines, antidepressants
FPN reagent Phenothiazines, dibenzazepines
Furfuraldehyde Carbamates, phenothiazines
Iodoplatinate, acidic Alkaloids, drugs, quaternary ammonium compunds
Mandelin’s reagent Drugs
Marquis reagent Drugs
Mercuric chloride-diphenylcarbazone Barbiturates
Mercurous nitrate Barbiturates
Ninhydrin Amphetamines, amino acids
4-(4-Nitrobenzyl)pyridine-tetraethylenepentamine Pesticides
Salkowski reagent (FeCl3#H2SO4) Phenothiazines, thioxanthenes
3,3,5,5-Tetramethylbenzidine, o -tolidine, after Cl2 Pesticides, acidic and neutral drugs
especially using pre-chromatography derivatization, Method: C:CAMAGDATA }SC3KRIMSPEC.PAM

20}100 ng per fraction. The visualization reactions Raw data: C:CAMAGDATA }SC3KY140798.DFS
can be divided into class and substance selective. Library: C:CAMAGKRIM1.SCL
In visualization sequences, several reagents can be
oversprayed one after another to amplify the amount Track 15, Analysis n: 2561
Peak C5, Measured hR Cf: 34, Area: 9682.6
of information obtained from a single plate. An
example of such sequence for basic drugs is nin-
hydrin, FPN (FeCl3#HClO4#HNO3 ) reagent,
Dragendorff’s reagent and acidiRed iodoplatinate.
Table 2 lists reagents that are commonly used in
forensic toxicology.
In situ spectra Earlier it was common to scrape off

a separated fraction from the TLC plate and submit it
to a further spectrometric or spot test analysis. Today
it is possible to measure the in situ UV spectrum of No. Substance name Diff Correlation
a fraction and compare this with stored spectrum 1 Clozapine 2 0.984415
libraries for identiRcation. Representative spectra can 2 Thiothixene !9 0.948759
generally be obtained from well-separated fractions 3 Norchlorprothixene 0 0.915627
with substance amounts over 0.5 g on standard TLC 4 Flupenthixol !1 0.902411
plates and over 0.1 g on high performance thin-layer 5 Olanzapine !1 0.891046
6 Norlevomepromazine !3 0.877048
chromatography (HPTLC) plates, although these
limits depend on the shape of the fraction and on the Confirmation: 䊐 necessary 䊐 not necessary
absorption characteristics of the compound in ques- Hit C*** confirmed by A:*** B:***
tion. The upper limit is not a problem as the reSec-
tance saturates after certain level and the spectra
remain practically the same. The CATS software Method: C:CAMAGDATA }SC3RPSPEC.PAM
from Camag allows the complete sequence of instru- Raw data: C:CAMAGDATA }SC3RY140798.DFS
mental screening analysis, including RF correction, Library: C:CAMAGRP1.SCL
spectrum measurement, automated search against Track 15, Analysis n: 2561
hRcF/UV libraries and reporting (Figure 1). Peak C2, Measured hR cf: 46, Area: 13178.0
Other spectrometric techniques than UV have been
tested for the measurement of TLC fractions. In situ
diffuse reSectance Fourier transform infrared (FTIR)
measurements have proved to be feasible for the iden-
tiRcation of drugs using a spectral region where silica
gel has no strong absorption, and a commercial TLC-
FTIR interface is available.
Proprietary Drug Screening Schemes

Particularly in North America, a TLC scheme called No. Substance name Diff Correlation
Toxi-Lab (Ansys, Irvine, CA, USA) has gained popu-
1 Clozapine 3 0.940416
larity in analytical toxicology. This product com- 2 Clothiapine !9 0.937639
prises the extraction, development, visualization and 3 Molindone !9 0.893862
interpretation steps of analysis. The Toxi-Lab A de- 4 Brucine !4 0.868569
tects basic and neutral substances and the Toxi-Lab 5 Apomorphine 2 0.857298
B detects acidic and neutral substances. The plates 6 Metoclopramide !6 0.855726
consist of silica, impregnated with a vanadium salt Confirmation: 䊐 necessary 䊐 not necessary
for detection purposes, in a glass Rbre matrix. The Hit C*** confirmed by A:*** B:***
Toxi-Gram C8 are octylsilica bonded phase plates for
the conRrmation of basic and neutral drugs. All the Figure 1 Identification hit lists produced by CATS software
(Camag) for an analyte fraction on two TLC systems in broad-
plates have holes at the origin for the inoculation of
scale screening analysis for drugs in liver. The reports were
factory-made standard substance discs and discs con- obtained by comparing the hR Fc values (window$9 units) and
taining the evaporated sample residues. Detection is in situ UV spectrum correlation against a library of 325 drug
carried out by using a standardized four-stage substances on each system.
visualization sequence, and the interpretation of the required for the analysis, which may be several hours.
colour patterns is performed with a help of the Toxi- There are no strictly forensic toxicological applica-
Lab Drug Compendium showing indexed colour tions of AMD in the literature but the technique has
photograms for hundreds of compounds. The se- an established position in the broad-scale screening
quence consists of formaldehyde vapour#Man- for pesticides in the environment and has great poten-
delin’s reagent, water, Suorescence under 366 nm UV tial in toxicology.
light and modiRed Dragendorff’s reagent. There
are also procedures and tests available for speciRc Overpressured Layer Chromatography
substances and classes of substances, such as opiates OPLC is based on the forced Sow of the mobile phase
and cannabis. There is ample literature available on against an external pressure, which results in short
the applications of Toxi-Lab to clinical and forensic development times and decreased diffusion of the
toxicology. analyte fractions, making it possible to take advant-
Another TLC screening scheme, Spot Chek (Ana- age of longer developing distances. Silica gel plates
lytical Bio-Chemistries, PA, USA), relies on a single are exclusively used in OPLC as reversed-phase
mobile phase, a set of visualization reactions, and chromatography has become complicated. Method
computerized interpretation of the patterns. Acidic/ development may be laborious due to disturbing ad-
neutral and basic drugs are developed on separate sorption zones, often obtained with multicomponent
plates, and on each plate the sample is divided into mobile phases. A commercial OPLC instrument is
two or three equal portions that are developed in available from OPLC-NIT (Budapest, Hungary).
parallel to facilitate the use of visualization reactions. OPLC has found use in the separation of closely
The migration distance is divided into Rve RF zones related compounds in a particular pharmacological
with reference compounds. The computer program category or compounds originating from a particular
database is based on nine colour reaction responses botanical source. Two complementary OPLC systems
and the plate zone locations for 243 drug substances have been developed for broad-scale screening
but requires entry of only one TLC property to gener- for basic and neutral drugs with SN values close to
ate a matching list. 30, which is more than twice the values obtained
Automated Multiple Development typically with ordinary TLC: trichloroethylene}
methylethylketone}n-butanol}acetic acid}water
There are currently two alternative instrumental 17#8#25#6#4, and butyl acetate}ethanol}
means to improve the Separation number (SN) in tripropylamine}water 85#9.25#5#0.75, with
TLC: automated multiple development (AMD) and layer pre-saturation.
overpressured layer chromatography (OPLC). In
AMD, the plate is developed repeatedly in the same
direction, and each partial run goes over a longer Target Analysis
solvent migration distance than the previous one.
Drugs of Abuse
Each partial run uses a solvent of lower elution
strength than the previous one and in this way a step- Amphetamines and related stimulants, especially
wise gradient is formed. SN values of up to 40}50 can amphetamine, methamphetamine and 3,4-methyl-
be obtained by AMD but a disadvantage is the time enedioxymethamphetamine (MDMA, an Ecstasy
Figure 2 Separation of (1) methamphetamine and (2) amphetamine on the TLC system 6 of Table 1.
component), can be separated by a variety of TLC

systems, such as those in Table 1 (Figure 2), and
sensitively detected as Suorescent derivatives of, for
example, Suorescamine or NBD-Cl (4-chloro-7-
nitro-2,1,3-benzoxadiazole). Ninhydrin is a tradi-
tional reagent for amphetamines. Fast Black K salt
reagent is capable of differentiating aliphatic primary
and secondary amines, giving violet and orange-red
colours, respectively, while tertiary amines do
not react. Thus amphetamine and methamphetamine,
or MDMA and 3,4-methylenedioxyamphetamine,
are readily separated. Amphetamines possess poor
UV characteristics, so they cannot be analysed by
UV densitometric methods at low levels without
derivatization. However, the methylenedioxy-
derivatives can be easily recognized by their UV
spectra (Figure 3). Despite the low limits of detection
obtained with pure amphetamine-like substances, the
limits of detection in urine are of the order of
250}500 ng mL\1 which is close to the standard
immunoassay cutoff value (300 ng mL\1).
The analysis of the main urinary cannabinoid, 11-
nor-delta9-tetrahydrocannabinol-9-carboxylic acid
(THCA), is usually carried out with dedicated TLC
systems, such as ethyl acetate}methanol}water}conc.
ammonia 12#5#0.5#1. The detection of THCA
is performed with various diazonium salts, such as
Fast Blue B salt, Fast Blue BB salt and Fast Blue RR
salt. Detection limits of 2}10 ng mL\1 can be ob-
tained for THCA, and these concentrations compare
favourably with the standard immunoassay cutoff
level of 20 ng mL\1.
Screening for cocaine is usually based on the detec-
tion of its metabolite benzoylecgonine (BE) in urine.
The combination of the following two mobile
phases can be applied to the separation:
methanol}chloroform}ammonia 60#60#1, and
ethyl acetate}methanol}water}ammonia 85#
13.5#1#0.5. The detection of BE is performed with
Dragendorff’s reagent or Ludy Tenger reagent, fol-
lowed by sulfuric acid, with the limit of detection of
Figure 3 The in situ UV spectra of (A) amphetamine,
200 ng mL\1 in urine. The standard immunoassay
(B) methamphetamine and (C) methylenedioxymethamph-
cutoff level is 300 ng mL\1 etamine (MDMA). Amphetamine and methamphetamine have
The opiates of interest in drug abuse testing pro- similar spectra but they can be differentiated, e.g. by using the
grammes include the heroin metabolites, 6-mono- Fast Black K reagent.
acetylmorphine and morphine, and codeine. The
combination of the following two mobile phases can Other Substances
be applied to the separation: ethyl acetate}isopropyl
alcohol}methanol}ammonia 80#15#3#8, and Toxicological Analysis by MuK ller lists 1453 published
1,2-dichloroethane}isopropyl alcohol}methanol} TLC systems for potentially toxic compounds. A TLC
ammonia 20#20#20#7. The limit of detection bibliography is available from Camag on CD-ROM,
for opiates with iodoplatinate ranges from 100 to listing 5500 abstracts of papers from 1982 to 1996.
500 ng mL\1 in urine, depending on the compound, The biennial TLC reviews by Sherma in Analytical
while the standard immunoassay cutoff level is Chemistry provide a wealth of information on sys-
300 ng mL\1. tems for individual substances in forensic toxicology.
III / FRAGRANCES: GAS CHROMATOGRAPHY 2885
Status of TLC in Forensic Toxicology De Zeeuw RA, Franke JP, Degel F et al. (eds) (1992)
Laboratory Thin-layer Chromatographic RF Values of Toxi-
cologically Relevant Substances on Standardized
In forensic toxicology, the unique features of TLC are Systems, 2nd edn. Weinheim: DFG/TIAFT,
best utilized in the broad-scale screening analysis for VCH.
drugs and poisons in urine or liver samples. For this Fried B and Sherma J (eds) (1996) Practical Thin-layer
application, there are equipment, dedicated software Chromatography. Boca Raton: CRC Press.
and reference libraries available from several manu- Gough TA (ed.) (1991) The Analysis of Drugs of Abuse.
facturers. Compared to HPLC or capillary elec- Chichester: John Wiley.
Jork H, Funk W, and Wimmer H (1990) Thin-layer
trophoresis, TLC allows the detection of even poorly
Chromatography, Reagents and Detection Methods,
UV-absorbing compounds using selective visualiz- vol. Ia. Weinheim: VCH.
ation reactions. Compared to GC or GC-MS, TLC Jork H, Funk W, Fischer W and Wimmer H (1994) Thin-
allows the chromatography of polar compounds layer Chromatography, Reagents and Detection
without prior derivatization. Another important ap- Methods, vol. Ib. Weinheim: VCH.
plication of TLC is the screening or conRrmation of Moffat AC (ed.) (1986) Clarke’s Isolation and Identi-
drugs of abuse, although the supremacy of the combi- Tcation of Drugs, 2nd edn. London: Pharmaceutical
nation of immunoassay and GC-MS in this area has Press.
hindered the development of modern dedicated TLC MuK ller RK (ed.) (1995) Toxicological Analysis. Leipzig:
methods. Immunoassay screening, however, is vul- Edition Molina Press.
nerable to sample adulteration and high background OjanperaK I and JaK nchen P (1994) The application of
instrumental qualitative thin-layer chromato-
noise. In larger, broad-service laboratories, the vari-
graphy to drug screening. LC-GC International 7:
ous techniques available today, including TLC, are 164.
considered complementary rather than exclusive. OjanperaK I, Goebel K and Vuori E (1999) Toxicological
drug screening by overpressured layer chromatography.
See also: II/Chromatography: Thin-Layer (Planar): Journal of Liquid Chromatography & Related Tech-
Modes of Development: Conventional; Modes of Develop- nologies 22: 161.
ment: Forced Flow, Over Pressured Layer Chromatogra- Siek TJ, Stradling CW, McCain MW and Mehary TC
phy and Centrifugal; Spray Reagents. III/Alcohol and (1997) Computer-aided identiRcations of thin-layer
Biological Markers of Alcohol Abuse: Gas Chrom- chromatographic patterns in broad spectrum drug
atography. Clinical Chemistry: Thin-Layer (Planar) screening. Clinical Chemistry 43: 619.
Chromatography. Clinical Diagnosis: Chromatogra- Stahl E (1967) Du( nnschicht-Chromatographie, 2nd edn.
phy. Forensic Sciences: Capillary Electrophoresis. Berlin: Springer-Verlag.
Stead AH, Gill R, Wright T, Gibbs JP and Moffat AC
Further Reading (1982) Standardised thin-layer chromatographic sys-
tems for the identiRcation of drugs and poisons. Analyst
Adamovics JA (ed.) (1995) Analysis of Addictive and Mis- 107: 1106.
used Drugs. New York: Marcel Dekker.
FRAGRANCES: GAS CHROMATOGRAPHY
E. R. Adlard, Delryn Burton, Wirral, UK tion) of a foodstuff or beverage. The aroma from
M. Cooke, Royal Holloway University of London, coffee beans on roasting is a prime example but there
Egham, Surrey, UK are many others. The main criterion is that the aroma
Copyright ^ 2000 Academic Press material is essentially all in the vapour phase and the
nose is responsible for sensing the aroma. A Savour is
intimately related to an aroma but may contain in-
Introduction volatile compounds that give rise to the sensation of
It is difRcult to distinguish between aromas, Savours, taste but in practice it is common to have a Savour
taints and perfumes, because to a large extent these with an associated aroma. A tainted foodstuff or
are artiRcial categories that overlap. An aroma may beverage is often unsatisfactory for consumption be-
be deRned as the smell emanating naturally (possibly cause there are compounds present that have an un-
in the process of cooking or other method of prepara- pleasant smell or taste. Taints may arise from natural
2886 III / FRAGRANCES: GAS CHROMATOGRAPHY
chemical reactions such as the oxidation of the acids

in an oil or fat to turn it rancid or the production of
amines in Rsh. Other taints may occur because of
leaching of material from packaging such as phenolic
compounds from paper wrapping and solvents from Figure 1 Structure of capsaicin, the hot agent of peppers.
the ink used in printing labels. Perfumes are natural
or synthetic mixtures that have a pleasant smell to
most people, although some of the constituents may diallyl disulRde, CH2"CH.CH2.S.S.CH2.CH"CH2,
have an unpleasant odour, a different odour or no and lager owes its distinctive aroma to the presence of
odour when present in bulk. For example, coumarin p.p.m. concentrations of lower mercaptans. As can be
in low concentrations has the smell of new-mown hay imagined, the actual amount is a vital part of quality
but this is not apparent at high concentrations. Al- control.
though this article is speciRcally about aromas most, Figure 1 shows the structure of capsaicin, the hot
if not all of the techniques described are equally agent in peppers. It contains a nitrogen-containing
applicable to perfume studies. The only signiRcant amido group, a phenolic group and an ether group as
difference is that perfumes, in their Rnal commercial well as oleRnic double bonds. Aroma compounds
form, normally exist as a solution (usually in ethanol) frequently exist as cis and trans isomers arising from
so that in this form they may be analysed as conven- such double bonds and there is also a possibility of
tional liquid samples. the presence of chiral compounds.
Another group of compounds that have much in IdentiRcation of geometrical and optical isomers is
common with aromas are pheromones, which cause of great importance since the isomers often exhibit
speciRc behavioural effects in animals and insects. a very large variation in physiological properties,
Indeed, it could be claimed that aromas and perfumes including smell. An example of this is carvone
are the equivalent materials in humans (although (Figure 2) where the D isomer has an odour of dill,
there are true human pheromones, they are of small whereas the L isomer has a spearmint smell. In
importance compared to those in the insect world). Figure 2 it can be seen that the centre of asymmetry is
Aromas frequently occur as complex, multicompo- at the carbon atom at which the isopropenyl group is
nent mixtures with many of the important compo- attached to the cyclohexene ring. The situation is also
nents (from an olfactory point of view) in very low complicated by the reverse situation, i.e. it is possible
concentrations } often at the ppm level or less. to have two compounds of quite different chemical
structure that smell the same. The best known exam-
ples of the latter are benzaldehyde and hydrogen
Aroma Compounds cyanide, both of which have a smell of bitter almonds
Since, by deRnition, aromas are volatile mixtures that and both give rise to this smell in natural products.
produce an olfactory response, it follows that they
should be amenable to analysis by gas chromatogra-
phy (GC). This was appreciated quite early in the
Sampling
development of GC and Teranishi et al. did a con- Since the sample is gaseous but is in contact with
siderable amount of work in the 1960s on strawberry a liquid or a solid, all the methods of gas sampling
aroma. The work was hampered by the use of packed may be used as appropriate but the most important
columns and detectors of relatively low sensitivity are static and dynamic headspace sampling. Dynamic
and poor qualitative diagnostic information, but headspace sampling results in greater sensitivity since
these investigations still go on today with modern all the volatile material from a given sample is re-
equipment (see Further Reading). moved from the headspace but it has to be retained in
Aromas contain many different types of com- a trap packed with a sorbent such as Tenax. It is
pounds. Among the commonest are aliphatic, oleRnic difRcult to adjust the purge gas conditions so that all
and aromatic hydrocarbons. A number of essential
oils fall into this category but many aroma com-
pounds have hetero-elements such as oxygen, nitro-
gen and sulfur in the molecule as well as a variety of
functional groups such as alcohol, aldehyde, acid,
phenol, ester and ether moieties. The smell of cheeses
such as Camembert is due to the presence of fatty
acids such as butyric acid. The smell of garlic is due to Figure 2 Structure of carvone showing the position of the chiral
a fairly simple mixture of sulfur compounds including centre.
the compounds of low volatility are removed without be necessary to carry out a large scale GC separation
losing those of high volatility in the trap and vice or some other enrichment procedure before GC anal-
versa. Whilst this may not be a problem if the volatil- ysis. One way of effecting preconcentration is by
ity range of the compounds is small, it may make using supercritical Suid extraction (SFE) with CO2
accurate quantitative analysis in one run impossible if containing small amounts (+5}10%) of a more po-
the range of volatility is large. One way to circumvent lar solvent such as methanol. The great virtues of SFE
this problem is by closed-loop stripping but this is at are that it is conducted at around ambient temperature
the expense of considerable extra experimental com- and that it is very easy to remove the solvent (CO2).
plications. Quantitative analysis is much easier by
static headspace sampling but this technique has
a poorer lower limit of detection, especially for com- GC Separation Conditions
pounds of low volatility. When using the term volatil- Columns
ity, it must be remembered that many samples are in
an essentially aqueous medium and a nonpolar com- All analysis of aroma samples is now carried out on
pound will have a large activity coefRcient in an open tubular columns except if small scale prepara-
aqueous system and a much higher volatility than tive GC is carried out for prior enrichment.
might be expected from the boiling point of the pure Aroma samples consist of compounds ranging
compound. from nonpolar hydrocarbons to polar aldehydes, al-
Transfer of the aroma sample from the sorbent trap cohols and acids but since they are in the gas phase
to the GC is by thermal or solvent desorption. Both under ambient conditions they will be of relatively
have advantages and disadvantages. Thermal desorp- low boiling point. These two conditions point to the
tion is a one-shot method and may cause decomposi- use of polyglycol (CarbowaxTM) phases which can be
tion of some of the components of the sample. It has operated up to about 2303C. Wax columns are parti-
the complication of the need for a secondary trap that cularly favoured for the analysis of fatty acid methyl
can be heated very rapidly so that the sample enters esters because of their ability to separate cis/trans
the column with a plug proRle. The presence of water isomers (Figure 3).
from aqueous samples may also cause problems if Other phases such as phenyl, triSuoromethyl,
steps are not taken to remove most of it. Solvent cyano and hydroxy silicones have also been used. For
desorption has the disadvantage that the solvent may chiral separations silicone phases containing
give a large peak early in the chromatogram that -cyclodextrins dissolved in the silicone have been
masks some of the volatile components of the sample. employed. Figure 4 shows the chiral separation of
The use of CS2 minimizes the problem if a Same #/! 1-octen-3-ol and #/! carvone and Figure 5
ionization detector (FID) is used as the detector since shows the separation of the chiral components of
this compound has a small FID response but it is toxic rosemary oil.
and highly Sammable. The columns should be capable of handling as large
A more recent method of sampling is the use of a sample as possible since some of the important
solid-phase microextraction (SPME) where a quartz aroma constituents may be present at p.p.m. level or
Rbre coated with a Rlm of stationary phase is exposed less. In order to have a large sample capacity, the
to a liquid or gaseous sample followed by thermal or stationary-phase Rlm should be relatively thick and
solvent desorption. Although it is very convenient to Rlms up to 5 m have been used in large bore columns
use, the limit of detection is not likely be as good with (0.53 mm i.d). Thick Rlms cause a signiRcant reduc-
SPME as with dynamic headspace sampling since tion in resolution so a compromise is a Rlm of 1 m
SPME is essentially the same as static headspace samp- thickness. For higher resolution and with mass spec-
ling and the mass of the sorbent Rlm on the quartz Rbre trometric detection, standard 0.25 mm and 0.32 mm
is much smaller than in a conventional headspace trap. i.d columns are used with a Rlm thickness of 0.25 m
Discrimination is also possible if the correct choice of or less. Most applications now standardize on col-
Rbre coating is not made. To Rnd a polymeric station- umns 30 m long, with some samples requiring 60 m
ary-phase material that is equally selective for a broad length; columns longer than this are now seldom
range of compounds of different polarity is difRcult employed in the aroma Reld since they carry the
and there may be memory problems with the Rbre penalty of longer analysis time and a greater possibili-
and low recoveries. In spite of this problem, SPME is ty of decomposition.
becoming more popular and many recent publica-
Carrier Gas
tions use this technique with excellent results.
The concentration of the important compounds in The carrier gas is not of great importance on aroma
an aroma sample may be extremely small and it may analysis; nitrogen gives the highest resolution but the
Figure 3 Separation of cis and trans isomers of fatty acids. 60 m, 0.25 mm i.d., 0.25 m Rtx-Wax; on-column concentration
40}75 ng. Oven temperature 165}2503C at 23C min\1. Injection/detection temperature 220/2503C; carrier gas, helium. Linear velocity
20 cm s\1 set at 1653C; split ratio 50 : 1. Peak identification 1, C14 : 0; 2, C14 : 1 n5cis; 3, C14 : 1 n5trans; 4, C16 : 0; 5, C16 : 1 n7cis; 6,
C16 : 1 n7trans; 7, C18 : 0; 8, C18 : 1 cis isomers (n12, n9, n7); 9, C18 : 1 trans isomers (n12, n9, n7); 10, C18 : 2 n6cis; 11, C18 : 2 n6trans;
12, C20 : 0; 13, C20 : 1 n9cis; 14, C20 : 1 n9trans. Reproduced by permission of Restek Corp.
slowest analysis, as may be seen from van Deemter ionization detector with less than half the gas
curves for hydrogen, helium and nitrogen. Helium is volume (40 L as opposed to 100 L for the FID)
the most commonly used carrier gas. which shows the much bigger response of the latter;
the tailing in the helium detector chromatogram is
due to the response to a water peak. In this instance,
Detection
The FID is the workhorse detector for aroma analy-
sis. It has many advantages but lacks the sensitivity of
some of the selective detectors which may exhibit up
to 103 times better lower limit of detection as well as
giving qualitative information. Figure 6 gives two
headspace chromatograms of coffee aroma, one
obtained with an FID and the other with a helium
Figure 5 Chiral separation of the components of rosemary oil.

30 m, 0.32 mm i.d., 0.25 m Rt-DEXsm. Oven temperature 403C
(hold 1 min) to 2003C at 23C min\1 (hold 3 min). Carrier gas:
hydrogen 80 cm s\1. Detector: FID set at 2203C. Peak identifica-
Figure 4 Chiral separation of #/! octen-3-ol and #/! tion: 1, (!/#) -pinene; 2, (#/!) camphene; 3, (#/!)
carvone. 30 mm, 0.32 mm i.d., 0.25 m, Rt-DEXsa. Oven tem- -pinene; 4, (!/#) limonene; 5, eucalyptol (1,8-cineole); 6,
perature: 403C (hold 1 min) to 2303C at 23C min\1 (hold 3 min). (!/#) linalool; 7, (#/!) camphor; 8, (!/#) teripinen-4-ol;
Carrier gas: hydrogen 80 cm s\1 set at 403C. Detector: FID set at 9, (#/!) isoborneol; 10, (#/!) borneol; 11, (#/!) -ter-
2203C. Reproduced by permission of Restek Corp. pineol. Reproduced by permission of Restek Corp.
Figure 6 Helium ionization detector and FID chromatograms of coffee aroma. Column: 100 m;0.5 mm i.d. stainless steel coated
with Witconol LA-23. Column temperature 603C isothermal. Sample introduced via a gas sampling valve. (Reproduced from Andrawes
FF and Gibson EK Journal of High Resolution Chromatography (1982), with permisson from Wiley}VCH.)
the helium detector clearly has great advantages over

the FID.
As pointed out earlier, many aroma compounds
contain hetero-elements and Figure 7 shows head-
space chromatograms of coffee aroma obtained with
an FID and the pulsed Same photometric detector in
the sulfur mode, which shows the presence of a large
number of sulfur compounds. This detector can also be
used in a nitrogen-selective mode to reveal the presence
of pyrroles, pyrazines and caffeine in the coffee aroma.
The Fourier transform infrared detector (FTIR)
should be extremely useful in aroma analysis because
of its ability to give a response to speciRc functional
groups in a molecule. This detector is gradually
coming into use for this type of work (see Further
Reading) but the sensitivity to different functional
groups varies considerably and the capital cost is
quite large. The mass spectrometer in the selective ion
monitoring mode probably gives the best all-round
sensitivity. Full scan mass spectra can give sensitivi-
ties in the pg range under favourable circumstances.
Figure 8 shows a mass chromatogram and a Gram-
Schmidt FTIR chromatogram of a dynamic head-
space sample of the vapour above a particular variety
of strawberries. The identity of the peaks is given
Figure 7 Pulsed flame photometric detector (in S mode) and in Table 1. Although there is a general similarity be-
FID chromatograms of coffee aroma. (From Varian publication tween the two chromatograms, there is a considerable
03-914625-00, 1998. By courtesy of Varian Associates.) difference in detailed quantitative response, with
Table 1 Compounds in strawberry aroma identified by GC-MS and GC-FTIR
Peak no.a Compound RRT m/z max (cm\1)

(%SD, n"3)
1 Acetic acid 0.141 43 (100%), 60 (M#, 83%) No data

(5.4%)
2 Methyl acetate 0.156 43 (100%), 74 (M#, 15%) 2966, 1777, 1757, 1448
(5.2%) 1372, 1240, 1048
3 Ethyl acetate 0.255 43 (100%), 61 (5%), 2992, 1769, 1755, 1373,
(3.6%) 88 (M#, 19%) 1238, 1093, 1052
4 Isopropyl acetate 0.310 43 (100%), 59 (6%), 2985, 2904, 1755, 1383,
(2.6%) 61 (26%), 87 (2%), 1239, 1138, 1021
102 (M#, 9%)
5 Ethyl propionate 0.393 57 (100%), 74 (12%), 2992, 2958, 1753, 1185
(1.7%) 102 (M#, 33%)
6 Methyl butyrate 0.415 43 (100%), 55 (11%), 2968, 1760, 1444,
(3.4%) 59 (13%), 71 (28%), 1359, 1297, 1256,
74 (33%), 87 (8%), 1184, 1103
102 (M#, 12%)
7 4-Methyl- 0.440 43 (100%), 58 (13%), 2960, 2881, 1729, 1373,
2-pentanone (1.8%) 85 (40%), 100 (M#, 26%) 1285, 1240, 1176
8 Ethyl 0.499 43 (100%), 71 (29%), 2973, 2895, 1755, 1461, 1442,
isobutryate (1.5%) 88 (14%), 116 (M#, 22%) 1390, 1365, 1249,
1234, 1188, 1154, 1096
1083, 1021
9 Methyl 0.518 41 (55%), 43 (100%), 2966, 2889, 1759, 1440,
2-methylbutyrate (1.7%) 55 (20%), 57 (58%), 1363, 1292, 1242, 1186,
59 (25%), 69 (12%), 1112, 1017
74 (12%), 85 (20%)
88 (52%), 101 (13%),
116 (M#, 4%)
10 Methyl 0.518 41 (40%), 43 (100%), 2969, 2896, 1757, 1466,
isobutyrate (1.5%) 57 (17%), 59 (17%), 1445, 1378, 1363, 1299,
74 (22%), 85 (9%), 1257, 1187, 1161, 1112,
101 (5%), 116 (M#, 6%) 1099, 1018
11 n-Hexanal 0.567 41 (100%), 44 (45%), 2940, 2885, 2811, 2714,
(1.1%) 56 (30%), 72 (7%), 1744
82 (16%), 99 (7%),
100 (M#, 4%)
12 Ethyl butyrate 0.567 43 (100%), 60 (10%), 2983, 1754, 1255, 1181
(1.4%) 70 (4%), 71 (45%),
88 (19%), 89 (15%),
101 (4%), 116 (M#, 14%)
13 Isobutyl acetate 0.587 41 (18%), 43 (100%), 2969, 2886, 1764, 1485,
(1.1%) 56 (14%), 61 (21%), 1372, 1234, 1064, 1032
69 (8%), 71 (8%),
116 (M#, 10%)
14 Isopropyl 0.637 41 (43%), 43 (100%), 2981, 2944, 2890, 1750,
isobutyrate (1.2%) 71 (40%), 89 (35%), 1468, 1376, 1238, 1184,
130 (M#, 2%) 1152, 1091, 1030
15 Ethyl 0.674 41 (100%), 43 (50%), 2979, 2948, 2891, 1750,
2-methylbutyrate (0.7%) 57 (67%), 69 (21%), 1466, 1377, 1248, 1182,
74 (12%), 85 (11%), 1149, 1088, 1033
115 (5%), 130 (M#, 12%)
Table 1 Continued

(%SD, n"3)
16 Ethyl isovalerate 0.678 43 (100%), 57, (59%) 2971, 2884, 1753, 1468,
(0.8%) 69 (10%), 87 (20%), 1374, 1295, 1250, 1184,
130 (M#, 10%) 1115, 1039
17 Hex-2(Z )-enal 0.681 41 (100%), 55 (41%), 2972, 2949, 2885, 2814,
(0.9%) 69 (18%), 83 (11%), 2727, 1715, 1634, 1151,
98 (M#, 14%) 1091, 1037, 981
18 Isoamyl acetate 0.712 43 (100%), 55 (24%), 2969, 2887, 1761, 1468
(1.5%) 61 (7%), 70 (17%), 1371, 1234, 1038
87 (4%), 130 (M#, 4%)
19 2-Methylbutyl 0.72 43 (100%), 55 (13%), 2972, 2899, 1762, 1468
acetate (1.6%) 61 (6%), 70 (12%), 1373, 1233, 1039
87 (1%)
20 3-Methyl- 0.747 45 (100%), 55 (26%), No data
2-heptanolb (0.6%) 57 (23%), 69 (8%),
83 (7%), 92 (7%),
112 (1%)
21 Amyl acetate 0.798 43 (100%), 55 (13%), 2963, 2944, 1767, 1361,
(0.3%) 61 (25%), 70 (11%), 1234, 1143, 1048
130 (M#, 2%)
22 Methyl 0.816 43 (100%), 55 (24%), 2963, 2879, 1760, 1440,
caproate (0.5%) 59 (21%), 69 (13%), 1241, 1215, 1173, 1110
74 (40%), 87 (13%),
99 (11%), 130 (M#, 13%)
23 Ethyl 3-methyl- 0.849 43 (20%), 55 (100%), 2987, 2948, 2936, 2904,
2-butenoate (1.5%) 83 (40%), 100 (13%), 1731, 1652, 1268, 1138,
113 (27%), 128 (M#, 5%) 1081, 1053
24 2,5-Dimethyl- 0.933 43 (100%), 55 (30%), No data
4-methoxy- (0.8%) 69 (10%), 85 (13%),
3(2H )-furanone 101 (5%), 127 (4%),
142 (M#, 2%)
25 Ethyl caproate 0.966 43 (100%), 55 (27%), 2969, 2943, 2882, 1754,
(1.6%) 60 (72%), 73 (40%), 1464, 1375, 1241, 1172,
88 (42%), 99 (37%), 1109, 1041
101 (18%), 115 (10%),
144 (M#, 8%)
26 2,5-Dimethyl- 0.990 43 (70%), 67 (45%),
3-hydroxy- (0.1%) 83 (100%), 112 (6%), No data
4-methoxy- 129 (2%), 128 (3%),
2,3-dihydrofurana 144 (M#, 6%)
27 Hexyl acetate 0.994 43 (100%), 55 (13%), 2966, 2942, 2871, 1762,
(0.1%) 56 (22%), 61 (14%), 1369, 1234, 1060, 1030
69 (7%), 84 (5%),
144 (4%)
28 Hex-2-(E )-enyl 0.997 43 (100%), 55 (17%), 2969, 2942, 2883, 1762
acetate (0.1%) 67 (48%), 82 (31%), 1675, 1455, 1358, 1230
142 (M#,3%) 1081 1024, 968
29 Cyclohexyl 1 43 (100%), 55 (13%), 3018, 2970, 2947, 2908,
acetate 67 (19%), 82 (35%), 2886, 1752, 1465, 1375,
83 (44%), 142 (M#, 4%) 1233, 1042
Table 1 Continued

(%SD, n"3)
30 2-Ethyl hexenoate 1.028 41 (55), 55 (100%), 2975, 2939, 1743, 1650,

(isomer) (1.1%) 68 (18%), 69 (18%), 1528, 1312, 1252, 1176,
73 (22%), 97 (31%), 1047, 991
142 (M#, 16%)
31 Amyl 1.054 43 (100%), 55 (29%), 2969, 2908, 2887, 1752,
butyrate (0.5%) 60 (6%), 70 (27%), 1460, 1353, 1238, 1177,
71 (52%), 89 (10%), 1096
158 (M#, 3%)
32 Unidentified 1.073 41 (100%), 55 (78%), 2966, 2934, 2882, 2817,
unsaturated (0.6%) 69 (42%), 83 (32%), 2738, 1787, 1716, 1623
aldehyde 93 (7%), 109 (57%),
128 (48%), 144 (50%),
33 Nona-2,4-dienal 1.155 43 (39%), 81 (100%), 2749, 1745, 1673
(isomer) (0.9%) 95 (19%), 138 (M#, 9%)
34 Non-2-en-1-ol 1.175 57 (100%), 67 (36%), No data
(isomer) (0.8%) 68 (18%), 69 (67%)
70 (34%), 81 (39%),
83 (36%), 95 (13%)
96 (10%), 124 (7%),
142 (M#, 2%)
35 Methyl caprylate 1.194 43 (100%), 55 (40%), 2937, 2867, 1758, 1443,
(0.9%) 69 (11%), 74 (65%), 1353, 1238, 1191, 1113,
87 (25%), 101 (9%), 1045
115 (8%), 127 (11%),
158 (M#, 7%)
36 Benzyl acetate 1.244 43 (100%), 51 (9%), No data
(1.3%) 69 (9%), 77 (14%),
79 (29%), 91 (87%),
108 (54%), 150 (M#, 4%)
37 Ethyl benzoate 1.267 43 (33%), 51 (15%), No data
(1.2%) 69 (10%), 77 (53%),
105 (100%), 122 (19%)
150 (M#, 10%)
38 n-Hexyl butyrate 1.316 43 (100%), 56 (30%), 2943, 2895, 2877, 1754,
(1.2%) 71 (54%), 84 (7%), 1263, 1176, 1097
89 (52%), 117 (16%),
172 (M#, 15%)
39 Hexyl isobutyrate 1.319 43 (97%), 55 (48%), 2971, 2943, 1754, 1265,
(1.1%) 71 (89%), 84 (100%), 1173, 1095, 1053, 976
89 (16%), 101 (9%),
172 (M#, 26%)
40 Ethyl caprylate 1.322 43 (100%), 57 (35%), 2967, 2938, 2678, 1753,
(1.2%) 60 (29%), 61 (20%), 1465, 1366, 1342, 1263,
69 (28%), 81 (17%), 1188, 1167, 1107, 1042
88 (26%), 101 (13%),
115 (5%), 127 (8%),
172 (M#, 10%)
41 Decanal 1.339 43 (100%), 57 (78%), 2934, 2865, 2780, 2765,
(1.2%) 69 (37%), 70 (28%), 1746
83 (75%), 95 (28%),
109 (9%), 156 (M#, 2%)
Table 1 Continued

(%SD, n"3)
42 Octyl acetate 1.354 43 (100%), 55 (30%), 83 (12%), 2937, 2866, 1760, 1462,
(0.1%) 56 (17%), 57 (30%), 1369, 1233, 1038, 1018
61 (37%), 69 (28%),
70 (15%), 71 (42%),
112 (23%), 172 (M#, 12%)
43 Amyl caproate 1.427 43 (100%), 55 (32%), 2967, 2907, 2880, 1753,
(1.5%) 60 (8%), 70 (30%), 1466, 1369, 1239, 1194
71 (31%), 99 (8%), 1163, 1113
117 (16%), 186 (M#, 1%)
44 Nonyl acetate 1.511 43 (100%), 55 (31%), No data

(1.6%) 61 (13%), 69 (22%),
83 (11%), 97 (9%),
186 (M#, 1%)
45 n-Decyl acetate 1.657 43 (100%), 55 (31%), 2936, 2982, 2865, 2846,
(4.4%) 69 (28%), 83 (40%), 1755, 1456, 1272, 1175
97 (16%), 200 (M#, 2%) 1094
c
46 Hex-3(Z )-en-1-ol 41 (100%), 55 (19%), No data
67 (38%), 69 (8%),
81 (11%), 82 (15%),
100 (M#, 3%)
a
Numbering 1}45 from Figure 1.
b
Tentative identification.
c
Reproduced with permission from Marco et al., Journal of High Resolution Chromatography (1997), 20: 276}278.
some minor peaks in the mass chromatogram giving
a large FTIR response.
One detector speciRc to aroma/perfumery studies
is the human nose. The efSuent from the GC is
split between a conventional detector such as the
FID and a snifRng port which is purged with humidi-
Red nitrogen. Because the ability to recognize the
presence of an odour varies considerably from one
individual to another it is necessary to select a panel
from people who have been shown to possess a keen
sense of smell and to train them to recognize the
odour of particular compounds. Although it is always
stated how insensitive the human nose is compared to
those of animals, nevertheless it is still a highly sensi-
tive organ. It is possible, apparently, for trained
panellists to indicate the emergence of an odoriferous
compound from a GC column in parts of the
chromatogram where no signal is obtained from con-
ventional detectors. Under these circumstances the
procedure is to use small scale preparative GC and to
collect fractions at the points indicated by the panel;
these fractions are then re-run under analytical GC
conditions.
Conclusion
Figure 8 (A) Mass chromatogram of a headspace sample
above strawberries; (B) FTIR chromatogram of the same sample. The study of aromas is intimately connected to the
For peak identities see Table 1. (By courtesy of the Journal of study of Savours, taints and perfumes in that they all
High Resolution Chromatography (1997), 20: 279.) make extensive use of GC with a variety of detectors,
2894 III / FUELS AND LUBRICANTS: SUPERCRITICAL FLUID CHROMATOGRAPHY
of which the mass spectrometer is the most impor- De la Calle-Garcia, Reichenbacher D, Danzer M et al.
tant. Advances in this type of work will depend on (1998) Analysis of wine bouquet components using
advances in the instrumentation, particularly in the headspace solid phase micro-extraction-capillary gas
sensitivity of the mass spectrometer and on general chromatography. Journal of High Resolution
advances in knowledge of food components under Chromatography 21(7): 373}377.
Gomes da Silva MDR and Chaves das Neves HJ (1997)
various circumstances.
Differentiation of strawberry varieties through purge-
and-trap HRGC-MS, HRGC-FTIR and principal com-
See also: II/Chromatography: Gas: Detectors: Mass ponents analysis. Journal of High Resolution
Spectrometry; Detectors: Selective. III/Natural Pro- Chromatography 20(5): 275}283.
ducts: Liquid Chromatography. Solid Phase Micro- Guadayol JM, Caixach J, Ribe J et al. (1997) Extraction
Extraction: Environmental Applications; Food Tech- and identiRcation of volatile organic compounds from
nology Applications. Tobacco Volatiles: Gas paprika oleoresin (Spanish type). Journal of Agriculture
Chromatography. and Food Chemistry 45(5): 1868}1872.
Kralj Cigic I and Zupancic-Krajl L (1999) Changes in odour
Further Reading of Bartlett pear brandy inSuenced by sunlight irradia-
tion. Chemosphere 38(6): 1299}1303.
There do not seem to be any modern texts which deal Morales, MT Berry AJ, McIntyre PS and Aparicio R (1998)
speciRcally with aroma analysis but the literature contains Tentative analysis of virgin olive oil aroma by supercriti-
numerous references. The following is a partial list of pa- cal Suid extraction}high resolution gas chromatogra-
pers published from 1997 to 1999 showing the wide variety phy}mass spectrometry. Journal of Chromatography
of foodstuffs and drinks covered, ranging from wine, yo- A 819(1}2): 267}275.
gurt and tomato juice to strawberries and alligator meat. Ott A, Fay LB and Chaintreau A (1997) Determination and
Although mass spectrometry is the main method of detec- origin of the aroma impact compounds of yogurt Sa-
tion, other techniques are covered, together with a variety vour. Journal of Agriculture and Food Chemistry 45(3):
of methods for extraction prior to analysis. 850}858.
Pinnel V and Vandegans J (1997) Study of the aroma proRle
Ahn DU, Jo C and Olson DG (1999) Volatility proRles of of gherkin by purge-and-trap followed by GC-MS. Jour-
raw and cooked turkey thigh as affected by purge tem- nal of High Resolution Chromatography 20(6): 343}346.
perature and holding time before purge. Journal of Food Rapior S, Breheret S, Talou T and Bessiere J-M (1997)
Science 64(2): 230}233. Volatile Savour constituents of fresh Marasmius al-
Baek HH and Cadwallader KR (1997) Aroma volatiles in leaceus (garlic Marasmius). Journal of Agricultural and
cooked alligator meat. Journal of Food Science 62(2): Food Chemistry 45(3): 820}825.
321}325. Sucan MK and Russell GF (1997) A novel system for purge
Clarke RJ and Macrae R (1985) Coffee, Vol. 1 Chemistry. and trap with thermal desorption: optimization using
Amsterdam: Elsevier. tomato juice volatile compounds. Journal of High Res-
De la Calle-Garcia, Reichenbacher D, Danzer M et al. olution Chromatography 20(6): 310}314.
(1997) Investigations on wine bouquet components by Tarantilis PA and Polissiou MG (1997) Isolation and iden-
solid-phase micro-extraction-capillary gas chromatogra- tiRcation of the aroma components from saffron (Cro-
phy (SPME-CGC) using different Rbres. Journal of High cus sativus). Journal of Agricultural and Food Chemistry
Resolution Chromatography 20(12): 665}668. 45(2): 459}462.
FUELS AND LUBRICANTS:

SUPERCRITICAL FLUID
CHROMATOGRAPHY
M. M. Robson, University of Leeds, Leeds, UK (PAH). SFC operates with diffusivities that are more
gas-like, viscosities that are lower than liquids, and
densities that are more liquid-like. The resulting mass
Supercritical Suid chromatography (SFC) has a transfer coefRcients lead to more rapid analysis in
number of advantages over gas chromatography (GC) SFC than in HPLC. The diffusion and viscosity range
and high performance liquid chromatography (HPLC) available in SFC allows GC-like separations on capil-
for mixtures such as polyaromatic hydrocarbons lary columns but at much lower temperatures.
III / FUELS AND LUBRICANTS: SUPERCRITICAL FLUID CHROMATOGRAPHY 2895
SFC on Packed Columns a modiRer to the CO2 mobile phase substantially

reduces the retention times of the PAH. This effect is
During the initial development of SFC, commercial
due to the intermolecular attraction between the
HPLC columns were used. The length and internal
modiRer and the solute molecules and the subsequent
diameter of SFC packed columns are constrained by
increased solvating power. The successful separation
the large pressure drop as compared to open tabular
of the 16 Environmental Protection Agency (EPA)
columns and high mass Sow rates, making interfacing
target list of PAH using a single (15 cm;4.6 mm)
to GC detectors more difRcult. Therefore, narrow-
column packed with specially bonded C18 silica has
bore packed columns with diameters of 1}2 mm were
been achieved in 6 min (Figure 1).
frequently used because they can be installed in a cap-
The analysis of PAH and their derivatives from
illary SFC instrument and are compatible with many
particulate matter has been of recent interest due to
GC detectors. Packed fused silica columns have sig-
possible human exposure of the highly toxic nitro-
niRcant advantages, allowing the use of a large var-
PAH compounds. Sandra et al. provides an interest-
iety of liquid chromatography (LC) stationary phases
ing approach to the analysis of PAHs using semip-
with GC-based detectors.
reparative SFC to separate the PAHs initially into the
The separation of PAH standards has been used
required types to analyse the peaks of interest, includ-
throughout the development of SFC to determine
ing nitro-PAH (Figure 2). A large amount of work
chromatographic efRciency and performance of the
has been carried out using supercritical Suid extrac-
system. Various-size stationary-phase particles have
tion (SFE) of particulates to extract the organic ma-
been used (10, 5 and 3 m particle diameter) and it
terial followed by GC to characterize the extract.
has been shown that the particle size inSuences the
efRciency of the columns and that SFC reduced analy-
sis times considerably in comparison with HPLC. Open Tubular Columns
The elution order of PAH in SFC can be varied by
The use of open tubular capillary columns in SFC for
changing the operating temperature and/or the pres-
petroleum-derived mixtures was initiated by Lee et al.
sure. Also, the mobile-phase modiRer used in SFC can
Capillary column SFC is mainly preferred because it
signiRcantly affect the retention behaviour of PAH;
provides the highest resolution. SFC can be carried
dramatic changes in retention together with different
out on capillary columns to achieve unique separ-
selectivities have been demonstrated. The addition of
ations, especially for complex hydrocarbon mixtures
Figure 1 SFC of the EPA 16 priority PAH using 5 m Waters Spherisorb PAH. Peak identification: 1, naphthalene; 2, acenaphthene;
3, acenaphthylene; 4, fluorene; 5, phenanthrene; 6, anthracene; 7, fluoranthene; 8, pyrene; 9, benzo[a ]anthracene; 10, chrysene; 11,
benzo[b ]fluoranthene; 12, benzo[k ]fluoranthene; 13, benzo[a ]pyrene; 14, dibenzo[a,h ]anthracene; 15, benzo[ghi ]perylene; 16, 16
indeno [1,2,3-cd ]pyrene. (Courtesy of the University of Leeds.)
Figure 2 Semipreparative SFC separation of PAHs from particulate matter. Using a 25 cm;4.6 mm i.d. silica column at 503C and
200 bar, supercritical fluid flow rate 2 mL min\1, a 1 : 1 methanol : acetonitrile modifier was used and programmed from 1% (up to 5 min)
to 2.5% at 0.15% min\1, and then 27.5% and 5% min\1. With ultraviolet detection: 1.5}3.7 min 2}3-ring PAH; 4.4}5.8 min 4-ring PAH;
7.0}8.3 5-ring PAH; 9.0}10.9 6 ring PAH. (Reprinted with permission of Sandra et al. (1998) J. Microcol. Sep., 10: 89.)
such as PAH. Capillary columns provide selectivity simultaneous separation of a mixture of PAH
and high efRciency. Essentially, the same range of isomers. The method not only reduces the develop-
stationary phases that have been used for analysis in ment time by 50% but also provides two sets of
capillary GC have also been used in capillary SFC. retention data for identiRcation of unknowns. The
These stationary phases include 100% methyl, 100% simultaneous separation of a mixture of three-, four-,
phenyl, 5% phenyl, 5% biphenyl and cyanopropyl Rve-ring, PAH isomers on biphenyl and smectic col-
siloxanes. An n-octylpolysiloxane coated stationary umns is shown in Figure 4.
phase has been used to perform SFC on petroleum- However, unless column diameters are greatly re-
derived vacuum residues. The increased alkyl content duced, open tubular column SFC cannot compete
of the n-octyl phase over methyl phases provides an with conventional packed column SFC in terms of
increase in van der Waals interactions of the station- analysis time. Comparison of Figures 1 and 3 shows
ary phase with solutes. that separation of high-molecular-weight PAH can be
SFC separations on liquid crystal phases are based achieved in less than 6 min with a conventional
on the length-to-breadth (L/B) ratio and the planarity packed column, while more than an hour is required
of the PAH, and they interact with the ordered rod- for an open tubular column.
like structure of the liquid crystal phase. In addition,
interactions such as dispersion, dipole and induced
dipole interactions contribute to a good separation.
Hydrocarbon Group-type Separations
Kithinji et al. optimized chromatographic conditions Hydrocarbon group-type separations refer to the sep-
(pressure and temperature) to achieve better SFC separation of alkanes, alkenes and aromatic compounds
arations of high-molecular-weight PAH on a capillary in petroleum feedstocks and products. LC is com-
column coated with a smectic mesomorphic crystal- monly used for this purpose but suffers from a lack of
line phase with simultaneous temperature and density resolution, lack of a universal detector and long anal-
programming. The separation of PAH isomers on ysis time. GC has also been used but is limited to the
a liquid crystalline polysiloxane stationary phase is analysis of light distillates due to the column temper-
shown in Figure 3. Many of these isomers cannot be atures required. Packed column SFC with CO2 as the
resolved on any other stationary phase. mobile phase has been used for the determination of
Raynor et al. used dual capillary column SFC with saturates, oleRns and aromatics in petroleum prod-
phases of different polarity and selectivity to perform ucts boiling below 3503C. Separation is achieved
Figure 3 Chromatogram of total coal tar PAH obtained with linear density programming at a constant temperature of 1103C,
10 m;50 m, SB-smectic column. (Courtesy of the University of Leeds.)
using two different columns connected in series hydrocarbons used as solvents in LC. Another ap-
} a silica and a silver nitrate-impregnated silica col- proach uses supercritical sulfur hexaSuoride (SF6) as
umn } the analysis takes only 4 min per sample. The the mobile phase. SF6 provided less peak tailing and
effect of temperature, pressure and column stationary a shorter analysis time. Although SF6 is reasonably
phase on the separation has also been studied. At compatible with the Same ionization detector (FID),
a low temperature (353C), the saturates are better the detector has to be protected against the decompo-
separated from the monoaromatics. sition product hydrogen Suoride, either by being gold-
Another advantage of using a low column temper- plated or by having a platinum jet and upper electrode.
ature is the shorter retention times and hence shorter A more accurate, reproducible and rapid SFC
analysis time per sample. An increase in pressure also method to separate hydrocarbon mixtures by chem-
results in improved separation between saturates and ical class for samples ranging from C4 to C40 uses
aromatics. Separations using a combination of silica a column-switching technique which allows inter-
and cyano or amino columns or the combination of change between a microbore (1 mm i.d.) silica gel
silica and 20% silver nitrate column are not as good column and a silver-ion loaded strong cation ex-
as the separation obtained with a single 5 m silica change silica gel microbore column, with 10% CO2 in
column at the same temperature. SF6 as mobile phase. The silver-loaded cation ex-
The separation of aromatic types in middle distil- change column gives the saturate and oleRn separ-
lates according to ring number has been studied. Two ation, while temperature programming is used to
packed columns, a silica and an amino-modiRed sil- elute the aromatic peaks. The only problem encoun-
ica, were used with a switching valve. The saturates tered with the use of silver-modiRed columns is that
are separated as a group from the aromatics on the certain types of compounds can react with the silver
silica column. The aromatics are then switched to the ions, causing column instability.
amino column where further separation into mono-, The three-column system with column switching
di- and polyaromatic types can be performed. and backSushing can be used to separate saturates,
It has generally been found that the separation of aromatics and polar compounds in high-boiling resi-
saturates from oleRns is incomplete when CO2 is used dues. A multicolumn system for quantitative deter-
as the mobile phase. This observation was thought to mination of crude oil and high-boiling fraction class
be due to the polarizability of CO2 compared to the separation has been developed. Three different col-
Figure 4 Simultaneous separation of three-, four- and five- ring PAH isomers on biphenyl and liquid crystalline columns. Conditions:
CO2 at 1203C; programmed from 150 bar (10 min) to 450 bar at 4 bar min\1. (Courtesy of the University of Leeds.)
umns are used } cyano, silica and silver-loaded silica by the addition of air as make-up gas (approximately
} in order to separate saturate, aromatics and resins. 15 mL min\1).
The replacement of the silver-loaded silica column The use of a specially treated silica-based packed
with a silver-loaded cation exchange column results column to separate saturate and aromatic compounds
in a more stable system. Similar effects have been in diesel fuels by SFC has been reported. Separations
observed when the silica column is replaced with are achieved in less than 8 min, with good resolution
a cyano column. The silica column prevents the aro- (greater than 9) being achieved in the separation of
matic components with strong -interactions from one-, two- and three-ring aromatics.
entering the silver-loaded silica column and thus In 1991, SFC was approved for the separation of
allows the aromatics to be backSushed from the silica saturate and aromatic hydrocarbons in diesel fuels.
column as a narrow band. The cyano column is used The American Society of Testing Materials (ASTM)
to trap the resins. Both carbon dioxide (CO2) and method D5186 requires that temperatures in the
nitrous oxide (N2O) can be used as mobile phases; the range of 30}403C be used for this separation. At these
latter provides better solubility of the resins and less low temperatures, the separation of saturates and
tailing of the peaks. monoaromatics is easily achieved. However, low
A quantitative study of the determination of aro- temperatures are not adequate for separation be-
matics in jet and diesel fuels by SFC with FID used tween the mono- and polyaromatics.
a single column (25 cm;2.0 mm i.d. column packed A packed capillary column has also been used with
with 5 m Chromegasphere SI-60 silica), a CO2 mo- the same methodology, and showed that good separ-
bile phase under constant density and no temperature ation can still be achieved at temperatures as high as
programming. It was found that the nonlinear re- 853C. A further increase in pressure and temperature
sponse of the detector can be signiRcantly improved also results in apparently higher efRciencies for the
by using true distillation techniques or analytical

techniques which simulate the distillation process.
Simulated distillation (SD) is a technique which is
inexpensive and rapid compared to distillation tech-
niques but does not yield fractions for further charac-
terization. GC simulated distillation techniques re-
quire the use of special high-temperature columns,
and the high temperatures (up to 4003C) may com-
promise the integrity of the sample. SFC uses much
milder conditions (typically below 1503C) than those
necessary for simulated distillation by GC and can be
used for the characterization of heavy crude compo-
nents, with boiling points up to 7603C. The effect of
temperature and pressure on resolution and retention
have been studied, and generally analysis is conduc-
ted by pressure programming while keeping the tem-
perature constant, although simultaneous temper-
ature and pressure programming has been success-
fully used. A relatively nonpolar, 5% phenyl}
95% dimethylpolysiloxane (DB5) phase is used. Such
columns are very stable at SFC temperatures and,
with the use of an integral restrictor installed at the
end of the column, good results are obtained.
Figure 5 Chromatogram of base oil mixture containing 57.8% Linear density programming of the supercritical
aromatic content. Operating conditions: column 1.3;250 m i.d. mobile phase provides a more nearly linear relation-
packed with Waters Spherisorb S5W. (Courtesy of the University ship between the elution pressure and homologous
of Leeds.)
series boiling point. The use of open tubular columns
eliminates the pressure drop effects which are com-
heavier aromatic peaks with reduced analysis time. mon in packed columns. If an n-octylpolysiloxane
However, the ASTM method requires a minimum stationary phase is used, the discrimination between
resolution of 4 between the saturates and aromatics aliphatic versus aromatic boiling points is minimized.
peaks. As the temperature is raised, the separation While SFC with open tubular columns seem to be
between the saturates and aromatics decreases. For well suited to SD, low sample capacity and loadabil-
highly complex mixtures, such as coal liquefaction ity pose difRculties for complex mixtures. Recently,
recycle solvents, an optimized group-type separation the analysis of hydrocarbon mixtures up to n-C130 has
procedure involves SFC on a 250 m i.d.;1.3 m long been achieved using a 300;0.3 mm i.d. column
silica column operated at 803C and 300 bar (Fig- packed with 5 m C18 bonded silica. The true boiling
ure 5). points and retention times of n-alkanes, alkylben-
The separation of petroleum distillate into zenes, PAH and thiophenes have been correlated and
aliphatic and aromatic fractions has been achieved it has been found that the retention time differences
using a two-dimensional SFC-SFC system with a Sow- do not exceed 1 min for chemically different solutes
switching interface. The columns used were a liquid with similar boiling points.
crystal polysiloxane capillary column and a SB- The effect of C1 to C18 alkyl groups bonded to silica
Biphenyl-30 capillary. The use of a liquid crystal has been investigated and the oligomer peak resolu-
column in the second dimension to provide shape tion obtained with packed capillary columns ap-
selectivity allows separation of various isomers, in- proaches that obtained with open tubular columns.
cluding chrysene, triphenylene, benz[a]anthracene Figure 6 shows the SFC chromatogram of the SD
and the benzoSuoranthenes. calibration standard on a packed capillary hexylsilyl
(C6) column. The SD data from SFC correlated well
Simulated Distillation of Petroleum with those obtained by GC. Higher-molecular-weight
hydrocarbons can easily be eluted at operating pres-
Compounds and Crude Oils sures below 415 bar (density of CO2 mobile phase
Distillation is the primary separation process in the approximately 0.71 g L\1). At this maximum pres-
petroleum used to characterize petroleum products sure a column packed with hexyl (C6) bonded silica
before processing. Distillation data can be obtained elutes hydrocarbons boiling at more than 7563C,
between the retention times of aromatics and straight

chain alkanes of apparently similar boiling points
which occurs when SD is performed by GC may be
reduced using open tubular SFC and further mini-
mized when packed capillary columns are used, espe-
cially for aromatic compounds with three or more
rings. However, comparisons may not be valid in
view of the discrepancies between published values of
PAH boiling points.
Analysis of Zinc Dialkyldithio-

phosphate in Lubricating Oil
Zinc dialkyldithiophosphates (ZDDPs) are used in
lubricating oils as extreme-pressure anti-wear addi-
tives. It is possible to analyse ZDDPs using SFC; if the
sample is in a lubricating oil matrix then it is essential
that a phosphorus-speciRc detector is used, i.e. nitro-
gen phosphorous detector (NPD) (Figure 8). This re-
moves the interference from the base oil which is
obtained if FID detection is used. Figure 8 also shows
that it is possible to determine other lubricating oil
additives e.g. Irgafos 168.
Figure 6 Simulated distillation chromatogram of Polywax 655
on a 0.25 m;250 m i.d. column packed with Waters Spherisorb
S5W C6. Conditions: linear pressure programming from 100 to Conclusions
415 bar (hold for 60 min) at 3.5 bar min\1, FID detection. (Cour-
tesy of the University of Leeds.)
The analysis of high-boiling hydrocarbon mixtures
has historically been difRcult due to their complex
nature. Because of the favourable properties of super-
whereas a column packed with octadecyl silanol critical Suids } low viscosity, low density and high
(ODS-2) (C18; Figure 7) is more retentive and only diffusivity } SFC has found many applications in this
elutes hydrocarbons boiling up to 6863C. Hydrocar- area. The technique is becoming increasingly popular
bons of even higher boiling points can be eluted if the in the petroleum industry, especially for group-type
column length is changed. The signiRcant deviation separation and for simulated distillation. The main
Figure 7 Graph of elution pressure versus the boiling point for packed columns. Diamonds, methyl (C1); squares, hexyl (C6);
triangles, octyl (C8); circles, ODS2 (C18). (Courtesy of the University of Leeds.)
III / FULLERENES: LIQUID CHROMATOGRAPHY 2901
Further Reading
ASTM Method D5186 (1992) In: Annual Book of ASTM
Standards, vol. 05.03, p. 855. Philadelphia, PA: Ameri-
can Society for Testing and Materials.
Kithinji JP, Raynor MW, Egia B et al. (1990) Analysis of
coal-tar polycyclic aromatic hydrocarbon LC-fractions
by capillary SFC on a liquid-crystalline stationary
phase. Journal of High Resolution Chromatography 13:
27}33.
Lee ML and Markides KE (eds) (1990) Analytical Super-
critical Fluid Chromatography and Extraction. Provo,
UT: Chromatography Conferences Inc.
Medvedovici A, David F, Desmet G and Sandra P (1998)
Fractionation of nitro and hydroxy polynuclear aro-
matic hydrocarbons from extracts of air particulates by
supercritical Suid chromatography. Journal of Micro-
column Separations 10: 89}97.
Raynor MW, Moulder R, Davies IL et al. (1990)
Dual capillary column supercritical Suid chromat-
ography. Journal of High Resolution Chromatography
13: 22}26.
Roberts I (1995) Supercritical Suid chromatography. In:
Figure 8 Chromatogram of used lubricating oil. Conditions:
Diol packed column, temperature 503C; ramp rate of 3 bar min\1, Adlard ER (ed.) Chromatography in the Petroleum In-
NPD detection. (Courtesy of the University of Leeds.) dustry. Journal of Chromatography Series, vol. 56, pp.
305}346. Oxford: Elsevier Science.
Shariff SM, Robson MM, Myers P et al. (1998) Hydrocar-
advantages of SFC are its speed of analysis and im- bon group-type separations for high aromatic fuels by
proved column efRciency when compared to liquid supercritical Suid chromatography on packed capillary
chromatography. The lower column temperature columns. Fuel 77: 927}931.
than needed for GC allows the analysis of higher- Westwood SA (ed.) (1993) Supercritical Fluid Extraction
molecular-weight mixtures with carbon numbers up and its Use in Chromatographic Sample Preparation.
to C120 and beyond. London: Blackie Academic & Professional.
FULLERENES: LIQUID CHROMATOGRAPHY
V. L. Cebolla, L. Membrado and J. Vela, a decade based mainly on C60, and to a lesser extent
Instituto de CarboquOH mica, CSIC, Zaragoza, Spain on higher fullerenes. The starting reactives for this
Copyright ^ 2000 Academic Press chemistry have been the compounds previously iso-
lated by LC. It is now possible to bind covalently
many types of compounds to the fullerene molecule.
Introduction LC is currently the method of choice for the separ-
Since the Rrst separation in 1990 of C60 and C70 using ation, isolation and puriRcation of fullerenes. Pro-
column liquid chromatography (LC), this technique gress in fullerene chemistry therefore depends on the
has played a very important role in fullerene chem- development of improved chromatographic methods,
istry. LC has allowed macroscopic quantities of i.e. those with the highest efRcacy and best resolution
fullerenes (particularly C60) to be isolated and puriRed between the different components of fullerene mix-
from the processing products. Obtaining sufRcient tures, at both analytical and preparative scales, and at
amounts of pure fullerenes has been crucial both for the lowest cost.
determining physical and chemical properties in order
Contribution of LC in the Field of Fullerene
to investigate practical applications of this new var-
Production
iety of carbon, and for developing a chemistry of
these spherical and polyfunctional carbon molecules. In early 1990, the design of LC methods was mostly
A very rich chemistry has been developed in less than geared towards isolating the most abundant
2902 III / FULLERENES: LIQUID CHROMATOGRAPHY
fullerenes obtained from the different production percentages of compounds to be separated are differ-
methods (e.g. laser vaporization of graphite, electric ent. Finally, some addents on the fullerene core have
arc discharge, hydrocarbon Same combustion pro- a dramatic inSuence on the solubility properties and
cess, pyrolysis of carbon material). Each of these the retention behaviour.
methods has its own speciRcity, with its advantages
and drawbacks. In general, the objective of these
methods was to obtain C60 and, to a lesser extent, C70. Separations Using Normal and
On occasions, endohedral metallofullerenes (i.e. Reversed Stationary Phases
M@C60) were also isolated. Therefore, chromato-
In general, the poor solubility of fullerenes in most
graphic methods developed for this purpose should be
organic solvents limits the choice of mobile and sta-
(semi-) preparative. Likewise, the presence ‘as impu-
tionary phases that can be used. Conventional sta-
rities’ of higher fullerenes and metallofullerenes (in
tionary phases used for normal- and reversed-phase
low or very low concentration), polyaromatic com-
elution may exhibit reasonable fullerene selectivity,
pounds (PACs, created during production), residual
but only when using mobile phases in which the
solvents, solvates (due to strong interactions between
fullerenes are poorly soluble (e.g. n-hexane, dich-
fullerenes and some solvents), fullerene adducts
loromethane, acetonitrile). These phases usually pro-
(mostly with oxygen), and other artefacts produced
vide weak interactions with fullerenes when using
during the production must be taken into account
eluents in which fullerenes are most soluble (e.g.
when designing an analytical procedure.
o-dichlorobenzene, CS2), which produce elution
In this case, an extraction step for pre-puriRcation
without adequate separation. Detection of fullerenes
is required. The origin of the soots, which is deter-
is not a problem: conventional or diode-array UV are
mined by the starting carbon material and the pro-
used in the range 230 to 600 nm, depending on the
duction method, inSuences the extraction yields and
cutoff of the solvent used.
the nature of the fullerenes extracted.
Normal-Phase Elution
LC in the Field of Organic and Organometallic
Open-column and low-pressure liquid chromato-
Chemistry of Fullerenes
graphic techniques, rather than HPLC, have been
At the same time that C60 and C70 became available in used for separations based on adsorption chromatog-
signiRcant amounts, the organic chemistry of raphy and performed with conventional normal-
fullerenes started to be developed. New reactions phase stationary phases (e.g. silica gel or alumina). In
were performed and plentiful data about C60 reactiv- this subsection, we summarize the application of
ity were obtained. C60 was initially considered to be these methods, including the use of charcoal columns,
a polycyclic aromatic hydrocarbon (PAH), and the to fullerene separations.
chromatographic methods that were developed were C60 was Rrst separated from C70 using open-column
based on previous knowledge accumulated on PAHs. LC on alumina as stationary phase, and n-hexane or
Further research, to which LC contributed signiR- n-hexane/toluene (95 : 5) as mobile phase. This
cantly, demonstrated that the reactivity of C60 is more method presents several disadvantages: as n-hexane is
like that of a localized polyoleRn. The LC contribu- not a good solvent for fullerenes, a high consumption
tion to this chemistry includes separation, identi- of both mobile and stationary phases is required; it is
Rcation and isolation of products. LC is also time-consuming, because several sequential separ-
used to monitor the reactions typical of fullerenes: ations are necessary in order to obtain sufRcient
nucleophilic additions, cycloadditions, hydrogenation amounts of C70 and other higher fullerenes, owing to
reactions, and oxidation and reactions with elec- peak tailing; and on-column degradation of C70 and
trophiles. In addition, different isomers, dia- higher fullerenes occurs (this phenomenon had been
stereomers and enantiomers have been separated already described for PAHs). The preadsorption of
(Figure 1). Different examples of these separations fullerenes on alumina did not solve these problems.
are shown in Table 1. A modiRcation was introduced to overcome these
The analytical designs required to separate deRciencies: the combination of Soxhlet extraction
fullerene derivatives from organic reactions differ and LC on alumina, in which hot solvent was con-
considerably from those used to isolate fullerenes tinuously recycled, allowed solvent consumption and
from their production methods. Several points must labour to be reduced. However, the most inexpensive
be considered. First, the structural similarities of these and efRcient method for rapid separation of C60 and
fullerene derivatives are often an obstacle in fullerene C70 involves an activated charcoal (Norit A) and silica
separations. Furthermore, the matrices and relative gel mixture (1 : 2) as the stationary phase. As carbon
Figure 1 Positions of the ligand-carrying bonds in the eight possible regioisomers of C60 with symmetrical additions to 6-6 bonds and
symmetry of the corresponding isomers. (Reprinted from Hirsch A (1994) The Chemistry of the Fullerenes, pp. 68 and 69, with
permission from Thieme.)
Table 1 Some examples of LC contribution to organic and organometallic chemistry of fullerenes
Reaction mechanism category Reaction LC contribution
Nucleophilic additions Hydroalkylation and hydroarylation of C60 Monitoring of products formation; consumption of
and C70 C60 followed from HPLC peak areas
C60#RLiPC60R\PC60HR (protonation) Purification of 1,2-organodihydrofullerenes
(preparative C18, chloroform/acetonitrile, 60/40)
Synthesis of [6,6]-methanofullerenes from Monitoring of the reactions by HPLC
stabilized sulfonium ylides with C60 or C70 Pure monoadducts obtained after column
chromatography on silica gel
Resolution of a racemic derived-amide on a chiral
(S,S)-Whelk-O HPLC column
Cyclopropanation of C60 and C70 and higher Isolation of seven stable regioisomeric bisadducts
adducts (stabilization by intramolecular SNi) of C62 (COOEt)4 of the eight possible ones (see
Figure 1). Identified by NMR and HPLC elution
order
Isolation of chiral trisadducts of C60 and
di-(ethoxycarbonyl) methylene
[4#2] Cycloadditions Cycloadducts from C60 with: Purification using GPC columns and toluene
}cyclopentadiene
}anthracene
Cycloadducts from C60-4 (4-fluoro-3-nitro Separation of products by GPC columns and
benzoyl) benzo-cyclobutene with: chloroform, or toluene
}4,13-diaza[18]-crown-6
}1,6-bis (aminomethyl) hexane
}4-amino-azobenzene
Cycloadducts from C60 with: Synthesis through flash chromatography
}2-trimethyl-silyloxy-1,3-butadiene
[3#2] Cycloadditions Synthesis of fullerene-bound dendrimers Isolation by GPC
Methano-bridged cycloadducts from C60 with Isolation on silicagel using toluene
diazoamides (biological importance)
[2#2] Cycloadditions Photochemical cycloaddition of enones to C60 Analysis of products using HPLC-Buckyclutcher威
The corresponding enantiomers of each
diastereomer for a methyl substituent of the
enone have been resolved and isolated by
HPLC using chiral stationary phases
Hydrogenation Synthesis of C60H2 via hydroboration and The only regioisomer formed can be isolated from
hydrozirconation, and also by Zn/acid the reaction mixture by HPLC
reduction
Polyhydrofullerenes from Birch}HuK ckel C60H36, C60H18 obtained from the mixture using LC
reduction on silica gel in CH2Cl2/hexane
Oxidation and reactions Oxygenation of C60 and C70 C60O, C70O and others isolated from fullerene
with electrophiles extracts by preparative HPLC (neutral alumina)
Thermal treatment of fullerene systems Isolation of C120O, C120O2, C180O2 using Cosmosil
Buckyprep威 and PBB威 in several steps, with
o-dichlorobenzene, and toluene
Bisosmylation of C60 Isolation of five regioisomers of C60 [OsO4(t-
bupy)2]2
Asymmetrical osmylation of chiral C76 using HPLC analysis of C76[OsO4LH] shows that two
chiral ligands (LH) regioisomers are predominantly formed upon
osmylation
compounds can adsorb fullerenes in a similar manner raphy on silica gel has occasionally been able to
to alumina, silica gel is added to mitigate this effect. isolate pure C60 and C70 derived adducts from organic
This system readily elutes C60 (using toluene, at low reactions (see Table 1). Likewise, silica gel has been
pressure or even slight vacuum). By adding o-dich- used to separate PAHs cosynthesized during fullerene
lorobenzene to the solvent system, gram quantities of production.
both C60 and C70 can be puriRed and a fraction en-
Normal Elution Using Nonconventional Phases
riched in higher fullerenes collected, all in a single
column pass. -Cyclodextrin bound to silica gel, in HPLC mode, is
Silica gel, in contrast, is not able to separate parent mostly used for enantiomer separations. However, in
fullerenes on its own. However, column chromatog- the Reld of fullerenes, it has been used for separating
C60 from C70. -Cyclodextrin is a cyclic oligosacchar- fullerene molecules. These phases afford greater in-
ide that forms complexes with fullerenes. The interac- teraction with fullerenes, thus allowing the use of
tion of C70 with cyclodextrin is stronger than that toluene-based mobile phases.
with C60, and both molecules can be separated, at In the second-half of the 1990s, several of these
analytical scale, even using n-hexane or n- stationary phases became commercially available and
hexane/toluene (70 : 30). However, the problem of are used now in many different types of fullerenes
fullerene solubility in these solvents has hampered separations, at both analytical, semipreparative or
further research on this stationary phase. preparative scale. Figure 2 shows, among other
phases described in this work, the chemical structures
Bonded-Phase Sorbents
of Buckyclutcher威 (tri-(dinitrophenil)-silica), Cos-
Octadecylsiloxane-bonded silica phases (C18) have mosil PBB威 (pentabromobenzyl-silica) and Cosmosil
been used in HPLC mode for analytical-scale separ- Buckyprep威 (2-[(1-pyrenyl)ethyl]-silica). The Buckyc-
ations of fullerenes. Although n-hexane or n-heptane lutcher column, developed by Pirkle, can be used in
have been used as mobile phases, toluene has been both normal and reversed phase mode. The capacity
mostly used, together with a counter solvent (e.g. of this phase to separate fullerenes seems to be due, in
acetonitrile, methanol) in variable proportions. Al- addition to electron donor}acceptor interaction, to
though the solubility of parent fullerenes is low in a steric effect created by a cone-shaped arrangement
these solvent compositions in general, the use of tol- of dinitrophenyl groups.
uene/acetonitrile provides adequate separation be- High-capacity stationary phases containing con-
tween C60 and C70 with selectivities (expressed as densated aromatic systems (e.g. Buckyprep威, 5-cor-
C70/C60) near 2 for a 50 : 50 mixture. The use of onenylpentyl-silica, COP) or heavy heteroatoms such
toluene/methanol, in turn, shortens the analysis time as sulfur, chlorine or bromine (e.g. Cosmosil PBB威)
in comparison with toluene/acetonitrile mixtures. showed high retentivity with excellent efRciency for
Other proRle phases, such as CHCl3/acetonitrile on separation of fullerenes, particularly Cosmosil PBB威.
CH2Cl2/acetonitrile, have also proven suitable for This phase allows the use of solvents providing high
separation of C60 and C70. fullerene solubilities, such as CS2 or 1,2,4-trich-
Monomeric and polymeric C18 phases have also lorobenzene, for gram-scale separations with ordi-
been used. They provide differences in retention for nary HPLC equipment. These solvents exceed toluene
fullerenes. A ‘topological’ recognition ability of the in solubility of fullerenes.
polymeric phase has been invoked to explain these Although large aromatic systems were expected to
differences. Therefore, fullerenes would be separated show high fullerene retentivity according to electron
as a function of their spherical diameters or their donating}accepting the positive effect of heavy atoms
geometries in general. is not correlated with the electron-withdrawing or
donating properties and more research is needed on
Separations Using Charge-Transfer this point.
The use of the above-mentioned commercially
Stationary Phases available columns usually involves two or three steps
Considerable research has been carried out in devel- in the process of separation, isolation and puriRcation
oping supports that can provide strong interactions of fullerene derivatives. As an example, a typical
with fullerenes when using eluents that solubilize scheme of proceeding may be a Rrst step using an
fullerenes. Charge-transfer chromatography is based analytical Cosmosil PBB column威 (with CS2, o-di-
on the donation or acceptance of electrons between chlorobenzene or 1,2,4-trichlorobenzene as eluents)
the stationary phase and, in this case, the fullerene and a second step using a preparative Buckyprep威
derivatives. The migration of fullerenes through the or Buckyclutcher威 column (e.g. with toluene). HPLC
column is selectively retarded to a degree depending with these columns has allowed the identiRcation,
on the strength of the charge-transfer complex formed. separation and/or isolation of endohedral metalloful-
Based on research on separation of PAHs, phases such lerenes (e.g. M@C74, M@C60, M@C70), fullerenes
as dinitroanilinopropyl silica (DNAP) or tetrach- higher than C60 (e.g. C86, C92, C94, C120), odd-
lorophthalimidopropyl silica (TCPP) were studied in numbered clusters (e.g. C119), products derived from
the early 1990s. The latter presented adequate selec- fullerene oxidation (e.g. C120O, C120O2, C180O2), and
tivity for C60 and C70 ("2.75 in the best case). the separation of isomers (e.g. C60H4).
Many of these phases, called Pirkle-type, have been Tetraphenylporphyrin (TPP)-silica based sta-
synthesized and tested. In general, they are based on tionary phase is another promising family for
electron-deRcient aromatic rings capable of simulta- fullerene separation at both analytical and pre-
neous interaction with the spherical, electron-rich, parative scale. Figure 2 shows chemical structures of
Figure 2 Some charge-transfer stationary phase used for fullerene separation (for abbreviations, see text).
[5-(p-hydroxyphenyl)-10,15,20-triphenyl]porphyrin- These phases have demonstrated strong reten-

silica (HPTPP) and [5-(p-carboxyphenyl)-10,15,20- tion and unmatched selectivity in the separation
triphenyl]porphyrin-silica (CPTPP). of C60 and C70 using strong mobile phases such as
Table 2 Fullerene selectivity, expressed as C70/C60 , of different (2,4,5,7-tetranitro-9-Suorenylideneaminooxy)prop-

stationary phases, using toluene as the mobile phase, under ionic acid] (TAPA) bound to silica gel (see Figure 2),
similar conditions
have been used for the separation of enantiomeric
Stationary phase C70/C60 fullerene derivatives.
Taking advantage of the fact that (R)-N-(3,5-dinit-
Buckyclutcher 1.5 robenzoyl)phenylglicine (DNBPhG) was able to sep-
PBB 2.5 arate fullerenes, other chiral stationary phases were
Buckyprep 1.8
designed. A stationary phase derived from S-nap-
CPTPP 5.7
HPTPP 7.0 roxen (Whelk-O columns; see Figure 2) has provided
successful separation of chiral fullerene derivatives
(Table 1).
1,2-dichlorobenzene or CS2. Table 2 gives a rough A strong charge-transfer interaction is necessary
picture of the inSuence under similar conditions of but on its own it is not a sufRcient requirement for
different stationary phases on fullerene selectivity, chiral discrimination. Both -donor}acceptor interac-
expressed as C70/C60 , using toluene as the mobile tions and one stereochemically signiRcant interaction
phase. Selectivity using toluene is the highest for all via hydrogen bonds are required.
phases with regard to stronger eluents: tol'CS2' TAPA-bounded silica gel has also been used for
dichlorobenzene. Another important point for TPP-silica nonchiral separations (e.g. separation between
stationary phases is their potential for preparative- C60 and C70).
scale separations. This includes higher fullerenes.
Owing to the usually low percentage of these kinds of
fullerenes in the parent mixtures, their efRcient separ-
Size Exclusion Chromatography (SEC)
ation using conventional or commercial phases be- SEC is a molecular sieving technique that separates
comes difRcult and usually involves several reinjec- molecules according to their selective permeation into
tions, which are both solvent- and time-consuming. the gel pores on the basis of differences in their size in
Currently, TPP phases present a good resolution in solution. Solute permeation into the gel increases
the separation of higher fullerene isomers in a single with decreasing molecular size, resulting in later elu-
column pass, using strong eluents such as 1,2-dich- tion. By virtue of these particular characteristics of
lorobenzene, as shown in Figure 3. SEC, the total time of the chromatographic runs is
previously known. This technique is also named GPC
Enantiomer Separation
(gel permeation chromatography) when carried our
Some charge-transfer stationary phases, such as using organic solvents.
(R)-N-(3,5-dinitrobenzoyl)phenylglicine or R-(!)[2- SEC stationary phases consist of a gel, typically
a three-dimensional network of cross-linked poly-
meric chains of controlled porosity. Compatibility
between solvent and stationary phase is of prime
importance. If the mobile phase is not suitable, gels
can swell or shrink, often damaging the packing bed;
this changes the pore distribution and hence the ex-
clusion volume. SEC is therefore conRned to isocratic
elution using a limited range of pure solvents.
The differences in size between the fullerenes are
sufRcient to allow SEC separation. Because pure tol-
uene can be used as the mobile phase (allowing fewer
constraints for solvent recycling) and these columns
have a greater lifetime than other LC columns (owing
to the theoretical absence of interaction between the
Figure 3 Separation of higher molecular mass fullerenes on fullerenes and SEC stationary phase) it is possible to
a 250 mm;4.6 mm column packed with HPTPP(4)-silica. Injec- use SEC at preparative scale.
tion of 12 mg of fullerene soot; mobile phase"100% 1,2-dich- Polystyrene-divinylbenzene-based columns, from
lorobenzene; flow rate"0.85 mL min\1; detection wavelength" 10 to 100 nm pore size, have been used for fullerene
410 nm; ambient column temperature. (Reprinted from Coutant
separations. In spite of their low selectivity
DE, Clarke SA, Francis AH and Meyerhoff ME (1998) Selective
separation of fullerenes on hydroxyphenyl-triphenylporphyrin- (C70/C60"1.1), the possibility of using automated
silica stationary phases. Journal of Chromatography A 824: systems with reinjection/sample collection and sol-
147}157, with permission from Elsevier Science.) vent recycling has allowed the separation of 10 g of
2908 III / FUNGICIDES / Gas Chromatography
an extract (from production) a day, yielding C60 with Coutant DE, Clarke SA, Francis AH and Meyerhoff ME
high purity. The relatively short time of SEC runs (1998) Selective separation of fullerenes on hydro-
allows the frequency of injections to be increased xyphenyl-triphenylporphyrin-silica stationary phases.
with respect to other LC techniques. Journal of Chromatography A 824: 147}157.
This kind of automated system has also been used Gross B, Schurig V, Lamparth I and Hirsch A (1997)
Enantiomer separation of [60]fullerene derivatives by
to separate and isolate metallofullerenes from empty
micro-column high-performance liquid chromatography
cage fullerenes in sufRcient amounts, despite their using (R)-(!)-2-(2,4,5,7-tetranitro-9-Suorenylidene-
low concentration in the production mixtures (e.g. aminooxy) propionic acid as chiral stationary phase.
200 mg of metallofullerenes isolated in 16 h). Journal of Chromatography A 791: 65}69.
However, the fact the C60 is eluted before C70 and Hirsch A (1994) The Chemistry of the Fullerenes. Stuttgart:
other higher fullerenes, regardless of the mobile phase Georg Thieme-Verlag.
used (e.g. toluene, CHCl3), means that solutes are not Kimata K, Hirose T, Moriuchi K et al. (1995) High-capa-
separated according to their size. Non-size effects city stationary phases containing heavy atoms for HPLC
(due to adsorptions and other types of interactions) separation of fullerenes. Analytical Chemistry 67:
have been described in the case of other relatively 2556}2561.
small molecules (e.g. PACs) with relative molecular Pirkle WH and Welch CJ (1991) An unusual effect of
temperature on the chromatographic behavior of buck-
masses lower than 1000.
minsterfullerene. Journal of Organic Chemistry 56:
6973}6974.
Further Trends Scrivens WA, Cassell AM, North BL and Tour JM (1994)
Research is now focused on Rnding more selective Single column puriRcation of gram quantities of C70.
Journal of the American Chemical Society 116:
stationary phases (mainly based on charge-transfer
6939}6940.
chromatography) to improve fullerene separation.
Taylor R, Hare JP, Abdul-Sada AK and Kroto HW (1990)
An efRcient method for separating individual Isolation, separation and characterisation of the
fullerenes on a large (preparative) scale is still re- fullerenes C60 and C70: the third form of carbon. Journal
quired. Most of the separation methods reported here of Chemical Society, Chemical Communications 20:
are limited to gram scale. This has hampered the 1423}1425.
study of higher molecular mass fullerenes. Theobald J, Perrut M, Weber JV, Millon E and Muller JF
(1995) Extraction and puriRcation of fullerenes: a com-
Further Reading prehensive review. Separation Science and Technology
30: 2783}2819.
Ajie H, Alvarez MM, Anz SJ et al. (1990) Characterization Welch CJ and Pirkle WH (1992) Progress in the design
of the soluble all-carbon molecules C60 and C70. Journal of selector for buckminsterfullerene. Journal of
of Physical Chemistry 94: 8630}8633. Chromatography 609: 89}101.
FUNGICIDES
diseases. The Rrst fungicide of proven efRciency was

Gas Chromatography the Bordeaux mixture, developed in 1882. This is
a mixture of lime and copper(II) sulfate that has been
JoseH L. Bernal, Department of Analytical Chemistry,
used for a long time; nowadays a wide variety of
Faculty of Sciences, University of Valladolid, Valladolid, compounds are used in a more selective way to Rght
Spain speciRc fungi in speciRc plants. It is necessary to pay
attention to their environmental impact (on water,
Copyright ^ 2000 Academic Press the atmosphere, soil and food) and also to their pres-
ence in vegetables which are intended for direct hu-
Fungicides, a class of pesticides, are toxic substances man consumption. Among the important fungicides
that are used to prevent or kill the growth of fungi are those which are applied in greenhouses and in
which are hazardous for plants, animals and human wine production.
beings. Most fungicides for agricultural use are fumi- There are two options for fungicide analysis: the
gated or sprayed over seeds, leaves or fruits to control determination of the composition of formulations,
and avoid a variety of economically important fungal where the concentrations are relatively high, and the
III / FUNGICIDES / Gas Chromatography 2909
evaluation of residues which appear after their use. Technique Selection

Theoretically, it is assumed that the correct use of
Chromatographic techniques are the common choice
fungicides does not imply problems with residue be-
for determining pesticide residues, and among them
cause they should always be applied at nonhazardous
GC is useful for analysing organochlorine and or-
levels. New pesticides are being developed with the
ganophosphorus residues because they have good
aim of greater efRciency and lower environmental
thermal stability, they are volatile and they have
risk.
a strong hydrophobic character. That this is valid for
In formulation analysis there are not many analyti-
a great number of fungicides can be appreciated by
cal problems. In most cases high performance liquid
the entries in the last column in Table 1. Nevertheless
chromatography (HPLC) is successfully applied. In
the thermal instability of some groups } EBDCs and
some cases UV-Vis spectrophotometry or gas
benzimidazoles particularly } make the use of HPLC
chromatography (GC) is used. In contrast, residue
more appropriate, although it is also possible to pro-
determination presents serious problems not only
duce derivatives that allow their determination by
because it is focused on trace levels but also because
GC.
the compound is usually found with other com-
pounds or degradation products, and in a matrix that Column
can present problems. This implies that the analyst
must pay attention to all the steps in the analytical Due to the large range of polarities of fungicides, it is
method from sampling to the interpretation of re- difRcult to select just one speciRc column, but many
sults. According to the problem the analyst will compounds can be separated using the mid-polarity
choose the best technique. This choice will affect columns. In several ofRcial methods for fungicide,
other steps, especially which sample treatment is glass columns (185 cm;4 mm i.d.) packed with OV-
selected to achieve the lowest quantity of possible 101 or similar, on Chromosorb WHP (80}100 mesh)
interference, good reproducibility and high re- are still recommended, although nowadays fused sil-
coveries. Sensitivity and selectivity always appear as ica open tubular capillary columns (FSOT) are prefer-
opposed criteria, so it is necessary to balance the red with lengths up to 60 m, inner diameter
chance of a better signal-to-noise ratio against 0.2}0.7 mm and Rlm thickness between 0.15 and
the risk of incompletely determining all the residues 0.5 m (DB-35, DB-1301, DB-1701 or similar). In
present. commercial catalogues or application notes, many
There are three types of required analyses: the examples of fungicide separations are shown, to-
global determination of a group of compounds, e.g. gether with the main chromatographic conditions
ethylenbis(dithiocarbamate) (EBDCs); individual de- and the equivalence between the different manufac-
terminations; and, separation and quantiRcation of turers.
chiral compounds. The method of analysis will vary
Detectors
according to the objective.
Table 1 summarizes common fungicides. The From the molecular formulae shown in Table 1, it is
trend is to use several compounds of the same or clear that the use of a nitrogen}phosphorus detector
a different chemical family, with the aim of achieving (NPD) is advantageous, although it is not as stable as
several modes of action over the development of the the Same ionization detector (FID) and because of
fungi. Because of this, a number of methods are that it must be calibrated frequently. The electron-
devoted to the determination of several compounds in capture detector (ECD) is useful for many com-
the same sample. Not all fungicides are shown in the pounds, but in several cases, the co-extracted com-
table: inorganic materials such as sulfur, borax, mer- pounds can interfere with its response. As there are
cury salts, arsenic and copper salts have been omitted. fungicides that have S and/or P atoms, the Same
Recently developed compounds whose efRcacy has photometric detector (FPD) is also useful. ConRrma-
not yet been proved or which are not registered tion of identity can be obtained using two columns,
in many countries, or whose mode of action is one a 5% phenyl 95% methylsilicone bonded-phase
still unknown, are also omitted. As can be ob- column coupled to ECD and NPD and the other
served, there are a number of chemical families, a 50% phenyl 50% methylsilicone column coupled to
the most important being phthalimides, benzimi- ECD and FPD. Such a system is very versatile and
dazoles, dicarboximides, carbamates, sulfamides, sensitive, allowing easy identiRcation of the eluted
anilinopyrimidines and phenylpyrrols. The nitrogen compounds. Nevertheless, the best option to conRrm
atom is common to all of them. In several cases there compound identity is the use of coupled techniques
is a halogen atom and in fewer cases, sulfur or phos- such as gas chromatography}mass spectrometry
phorus atoms are included. (GC}MS); the speciRcity of GC}MS provides low
Table 1 Chemical group, molecular formula and recommended method of residue analysis of the most common fungicides
Fungicide Chemical group Molecular formula Method for residues
Anilazine Triazine C9H5Cl3N4 GC, HPLC

Benalaxyl Acylalanine C20H23NO3 HPLC, GC
Benomyl Benzimidazole C14H18N4O3 HPLC
Bitertanol Azole C20H23N3O2 GC
Bromuconazole Azole C13H12BrCl2N3O GC
Bupyrimate Pyrimidine C13H24N4O3S GC, HPLC
Captafol Phthalimide C10H9Cl4NO2S GC
Captan Phthalimide C9H8Cl3NO2S GC
Carbendazim Benzimidazole C9H9N3O2 HPLC
Carboxin Phenylamide C12H13NO2S GC
Chlorothalonil Methoxybenzene C8Cl4N2 GC
Chlozolinate Phthalimide C13H11Cl2NO5 GC
Cymoxanyl Acetamide C7H10N4O3 GC
Cyproconazole Azole C15H18ClN3O GC
Cyprodinil Pyrimidine C14H15N3 HPLC
Dichlofuanid Sulfamide C9H11Cl2FN2O2S2 GC
Diclomezine Pyridazinone C11H8Cl2N2O GC
Dicloran Nitrobenzamine C6H4Cl2N2O2 GC
Diethofencarb Carbamate C14H21NO4 GC
Difenoconazole Azole C19H17Cl2N3O3 GC
Dimetomorph Morpholine C21H22ClNO4 GC, HPLC
Diniconazole Azole C15H17Cl2N3O GC
Dinocap Dinitrophenol C18H24N2O GC
Diphenylamine Amine C12H11N GC
Dodemorph Morpholine C18H35NO GC
Edifenphos Organophosphorus C14H15O2PS2 GC
Ethirimol Pyrimidine C11H19N3O HPLC/GC
Etridiazole Azole C5H5Cl3N2OS GC
Fenarimol Pyrimidine C17H12Cl2N2O GC
Fenfuran Carboxamide C12H11NO2 GC
Fenpliconil Pyrrole C11H6Cl2N2 HPLC
Fenpropidin Morpholine C19H31N GC, HPLC
Fenpropimorph Morpholine C20H33NO GC, HPLC
Ferbam Dithiocarbamate C9H18FeN3S6 HPLC
Fludioxonil Phenylpyrrole C12H6F2N2O2 GC
Fluoroimide Phenylpyrrole C10H4Cl2FNO2 GC
Flusilazole Azole C16H15F2N3Si GC, HPLC
Flutolanil Phenylamide C17H16F3NO2 GC
Flutriafol Azole C16H13F2N3O GC
Folpet Phthalimide C9H4Cl3NO2S GC
Fosetyl Organophosphorus C6H18AlO9P3 GC
Hexaconazole Azole C14H17Cl2N3O GC
Hymexazol Azole C4H5NO2 GC
Iprodione Phthalimide C13H13Cl2N3O3 HPLC,GC
Imazalil Azole C14H14Cl2N2O GC
Mancozeb Dithiocarbamate (MnZnSCNH)x HPLC
Maneb Dithiocarbamate C4H6MnN2S4 HPLC
Mepanypirim Pyrimidine C14H13N3 GC
Mepronyl Carboxamide C17H19NO2 GC
Methalaxyl Phenylamide C15H21NO4 GC
Metiram Dithiocarbamate (C16H33N11S16Zn3)x HPLC
Myclobutanyl Azole C15H17ClN4 GC
Nabam Dithiocarbamate C4H6N2Na2S4 HPLC
Nuarimol Pyrimidine C17H12ClFN2O GC
Ofurace Phenylamide C14H16ClNO3 GC
Oxadixyl Phenylamide C14H18N2O4 GC
Oxycarboxin Carboxamide C12H13NO4S GC
Penconazole Azole C13H15Cl2N3 GC
Phthalide Benzofuranone C8H2Cl4O2 GC
Prochloraz Azole C15H16Cl3N3O2 GC
Procymidone Carboximide C13H11Cl2NO2 GC
Propamocarb.HCI Carbamate C9H21Cl N2O2 GC
Table 1 Continued
Fungicide Chemical group Molecular formula Method for residues
Propiconazole Azole C15H17Cl2N3O2 GC

Propineb Dithiocarbamate (C5H8N2S4Zn)x HPLC
Pyrazophos Organophosphorus C14H20N3O5PS GC
Pyrifenox Oxime C14H12Cl2N2O GC, HPLC
Pyrimethanyl Pyrimidine C12H13N3 HPLC
Tebuconazole Azole C16H22ClN3O GC
Tetraconazole Azole C13H11Cl2F4N3O GC, HPLC
Thiabendazole Benzimidazole C10H7N3S HPLC
Thiophanate methyl Benzimidazole C12H14N4O4S2 GC, HPLC
Thiram Thiocarbamate C6H12N4S4 HPLC
Triadimefon Azole C14H16ClN3O2 GC
Triadimenol Azole C14H18ClN3O2 GC
Tricyclazole Azole C9H7N3S GC
Tridemorph Morpholine C19H39NO GC
Triflumizole Azole C15H15ClF3N3O HPLC
Vinclozolin Phthalimide C12H9Cl2NO3 GC
Zineb Dithiocarbamate C4H6N2S4Zn HPLC
Ziram Thiocarbamate C6H12N2S4Zn HPLC
Fungicides which appear in bold are those which are used most widely. (Reproduced from JimeP nez JJ, Bernal JL, del Nozal MJ,
Toribio L and MartOP n MT (1998) Journal of Chromatogrphy A 823: 381}387, with permission from Elsevier Science.)
detection limits and unambiguous spectral conRrma- poor resolution between compounds, as does GC}
tion in complex matrixes. MS. However, many laboratories cannot afford GC}
Taking into account that the presence of hetero- AED or GC}MS because they are more expensive
atoms is common, it is also possible to use the atomic than other options.
emission detector (AED) to monitor characteristic
wavelengths. Monitoring the emission lines for Practical Considerations
elements such as nitrogen, chlorine, phosphorus and
sulfur ensures speciRc chromatograms for those ele- According to the aim, a technique will be selected, as
ments, increasing the selectivity, which is especially mentioned before and this will determine the prior
desirable when dealing with environmental and food steps in the method. There are always some general
samples. recommendations, such as the need to employ stan-
dards and surrogates, whose addition (spiking) gives
Multiresidue Methods recoveries (which should be higher than 80%). The
Multiresidue methods are desirable for the deter- use of solvents of adequate purity is necessary; each
mination of speciRc components in samples of un- batch must be tested for a potential source of interfer-
known origin or those which have been subjected to ence; at the same time all glassware must be ad-
unknown pretreatments. Unfortunately, nitrogen- equately cleaned. Apart from these general pre-
containing pesticides have been poorly investigated in cautions, it is necessary to be aware of the importance
comparison to the halogen- or phosphorus-contain- of other aspects that have a notable inSuence in the
ing pesticides as regards their possible combination in analysis of fungicide residues. Some are summarized
multiresidue methods. Nevertheless, there are several here.
methods in which the behaviour of some fungicides is
Standards
considered; some include up to 20 different fungi-
cides. This type of research commonly relies on the One of the Rrst and most important steps in fungicide
use of more than one type of capillary column for the residue evaluation in food and environmental sam-
separation of broad groups of pesticides; usually the ples is the correct preparation of standard solutions,
main column has a low polarity stationary phase, preferably from solid reagents of certiRed purity, be-
employing another one of mid or high polarity as a cause of their low stability in solution. For example,
conRrmatory column. The detection can be made dir- Imazalil solutions are sensitive to light; Fosetyl resi-
ectly or after derivatization, and using either a single or dues decompose during storage at !183C. It is very
several detectors (ECD, NPD, FPD, AED, MS). common for derivatives to not last more than 24 h in
The use of the AED for multiresidue analysis par- a refrigerator; the stability should be checked for
tially overcomes some of the problems derived from longer storage times. It has also been demonstrated
that the amount of fungicide residue in food is in- placed by fast SPE clean-up, with the additional
Suenced by storage, handling and processing. advantage of a high enrichment factor.
The most recommended phase is octadecylsilane,
Sample Treatment although for some groups, the diol or cationic ex-
The sample may be simple or very complex; this will change phases may be better; graphitized carbon
clearly have a great inSuence on the sample treat- black is also a possibility nowadays. Florisil is fre-
ment. Isolation of the compounds using an extraction quently used to remove co-extractive interferences.
technique frequently needs a further clean-up step When this is not sufRcient, further clean-up can be
before determination. There are a great variety of achieved by gel permeation chromatography.
possible approaches, from classic liquid}liquid ex- Solvent selection for recovery of fungicides from
traction (LLE) to the use of supercritical Suids (SFE), cartridges is very important. The results vary for
and ofSine or online procedures. individual compounds. Figure 1 provides an example
of the recovery of different fungicides and acri-
Liquid}Liquid Extraction cides from the analysis of must samples.
In all cases it is necessary to optimize the type of
There are many methods based on the use of a separ- sorbent, sorbent mass, Sow rate, sample volume, pH,
ating funnel, drying over anhydrous sodium sulfate ionic strength, drying time and soaking time. Some-
and clean-up; Soxhlet extraction is also employed. times the complete elution of the compounds from the
The commonly used solvents are ethyl acetate, disposable extraction column requires several portions
acetonitrile, methanol, dichloromethane, acetone and of eluting mixture instead of only one. It is well known
n-hexane. Frequently, phase separation is hindered by that, for conazole fungicides and Captan, low recove-
emulsion formation in the separatory funnel, in ries are obtained because the sample volume and Sow
which case Rltration through a loose glass wool plug rate of extraction seriously affect the recoveries.
may be appropriate or another extraction procedure In the analysis of modern fungicides, extraction
may be more suitable. with methanol, partitioning with chloroform, puriR-
It is usual to Rnd anomalous results for Vinclozolin, cation of the extract by column chromatography on
Captan, Folpet and Iprodione when LLE is used. sodium sulfate/Florisil/celite/charcoal is often recom-
mended.
Solid-phase Extraction (SPE)
In the multiresidue methods acetonitrile is usually
The laborious liquid}liquid partitioning clean-up preferred, with SPE ofSine using C18 or polymeric
procedures described in the literature have been re- cartridges, followed by GC}MS. This gives better
Figure 1 Recovery of pesticides from 500 mL of synthetic sample spiked with 4 g L\1 by octadecylsilica cartridges eluted with 2 mL
of solvent. (Reproduced from Bernal JL, del Nozal MJ, JimeP nez JJ, and Rivera JM (1997) Journal of Chromatogrphy A 778: 111}117,
with permission from Elsevier Science.)
results than online HPLC-diode array detection Figure 2; it can be observed that SPE gives the sim-
(DAD) which has drawbacks for trace level deter- plest chromatograms.
mination as a result of many interferences.
In water analysis, SPE on disc (C18 Empore) gives Matrix Effects
good results and it has been successfully applied to GC analysis for fungicides frequently presents con-
the determination of fungicide residues, but in Vin- siderable errors due to the so-called matrix effect
clozoline determination errors are obtained, with the which has been described in the analysis of diverse
major losses occurring when the fungicide was col- compounds in wine, grape juice, honey, milk, butter,
lected from the surface of the disc. fruits and vegetables. This effect is explained by
Solid-phase microextraction is also useful for the a higher transference of analytes from the injection
analysis of fungicide residues in water samples, al- port to the chromatographic column either as a result
though in complex matrices it gives low reproducibil- of the presence in the extract of associated carrier
ity, which suggests that it is only useful for semiquan- substances from the matrix or of a protective effect in
titative purposes. In addition, the duration of the the injection port performed by these substances.
process in relation to other extraction procedures can The matrix effect is usually greater for lower
seriously limit its application to large numbers of quantities of analyte, as can be seen in Table 2, where
samples. The extraction conditions } stationary some data for common fungicides are shown. The
phase, time, temperature, type and concentration of recovery can also be inSuenced by the total amount of
compound and matrix } must be taken into account. sample (Table 3). From both tables it can be deduced
In Vinclozolin and Captan residue analysis on semi- that serious errors can arise when the effect is not
solid spiked samples, lower recoveries are obtained considered.
when the amount added increases. Once the sample preparation has been checked,
To prevent degradation or hydrolysis of certain attention must be paid to calibration. To reduce
fungicides (e.g. benzimidazoles), sometimes other ex- quantitative errors from the matrix effects, a standard
traction techniques such as those based on the use of addition method or an external standard calibration
pressurized hot water or supercritical CO2 are recom- with standards dissolved in an unspiked sample ex-
mended; even a cloud point preconcentration has tract can be used. The use of two or more certiRed
been used, nevertheless, these procedures are not reference materials to establish a calibration curve
common as yet. The importance of sample pre- can help recognize matrix effects. If the measure-
paration on the Rnal chromatogram is seen from ment shows the same slope with the regression, it
can be concluded that the matrix has no dominant
inSuence.
In general, an analyte addition method (AAM),
a sample variation AAM (varying the test sample
mass and keeping the added analyte amount con-
stant) or, better, a sample and analyte variation AAM
must be used for calibration.
The Analysis of Various Fungicide

Groups
Phthalimides
Captan, Captafol, Folpet There are several methods
devoted to the analysis of residues of these com-
pounds. Usually an extraction with n-hexane}acetone
mixtures is carried out; after drying with sodium
sulfate and evaporating the solvent, n-hexane is ad-
ded. This is followed by clean-up on a Florisil car-
tridge, eluting it with an n-hexane}acetone mixture,
evaporating and dissolving the residue in toluene and
Figure 2 Chromatograms by EI-MSS (scan mode) for Vin-
injecting an aliquot of the solution into the GC.
clozolin residues in a larvae extract. (A) Hexane}acetone (70 : 30,
v/v); (B) SPE. (Reproduced from Bernal JL, del Nozal MJ, Rivera To prevent hydrolysis, cloud-point preconcentra-
JM, JimeP nez JJ, and Atienza J (1996) Journal of Chromatogrphy tion employing the nonionic surfactant Triton X-114
A 754: 507}513 with permission from Elsevier Science.) has been proposed.
Table 2 Recovery (%) of some fungicides from honey samples, after conventional solvent extraction, spiked at different levels
(average seven determinations)
Fungicide 0.025 mg kg\1 0.125 mg kg\1 0.25 mg kg\1 1.0 mg kg\1 2.5 mg kg\1
Captan 1028 329 165 99 99

Folpet 2380 321 164 108 107
Iprodione 948 405 350 259 174
Vinclozolin 647 335 250 209 171
Captan and Captafol tend to decompose on col- mode, with detection limits in the range of
umns that have been in use for some time. To avoid p.p.b.}p.p.t.
decomposition removal of the glass wool from the These fungicides are frequently investigated in
inlet of the GC column is recommended. wine analysis where it is known that the recovery not
only depends on the concentration level but also on
Benzimidazoles the variety of wine; attention must be paid to their
metabolites, mainly those belonging to the 3,5-di-
Benomyl, Carbendazim, Thiabendazole, Thiophan-
chloroaniline group.
ate methyl This group is usually analysed by HPLC;
nevertheless there are GC-MS methods using acid Triazole
hydrolysis, re-extracting the amine and forming the
tert-butyldimethylsilyil derivatives. Bitertanol, Triadimefon, Triadimenol, Tryciclazole
Carbendazim is a fungicide and the main metab- Usually they are analysed by GC-NPD and GC-MS in
olite for Benomyl, both of which are widely used on the selected ion monitoring mode. Triadimefon is eas-
vegetables intended for direct human consumption. ily reduced to Triadimenol, so both appear together.
Methods usually provide residual levels in terms of Bitertanol and Triadimenol have diastereoisomers
Carbendazime because Benomyl degrades rapidly. To that cannot easily be separated; depending on their
analyse Carbendazime by GC, the compound is usu- relative proportion they frequently produce peaks
ally extracted with ethyl acetate and derivatized with with shoulders.
pentaSuorobenzylbromide. After clean-up on a silica Dithiocarbamate fungicides
column, the product is determined by GC-ECD or
GC-NPD. These are habitually classiRed into three families of
compounds depending upon their structure:
Dicarboximides
1. Dimethyldithiocarbamates (Ferbam, Ziram,
Chlozolinate, Iprodione, Procymidone, Vinclozolin Thiram)
These compounds are frequently included in multi- 2. Ethylenebisdithiocarbamates (Mancozeb, Maneb,
residue methods and in many applications a signiR- Zineb, Nabam)
cant inSuence of the spiking levels on recovery is 3. Propylenedithiocarbamates (Propineb)
observed. The most used techniques are GC-FID, GC-
ECD and ion trap GC-MS in multiple ion-monitoring It is very difRcult to isolate and determine speciRcally
the fungicides which belong to the same family due to
the fact they possess the same organic moiety. Thus,
Table 3 Recovery of Vinclozolin from honey and larvae
samples, spiked with 25 mg kg\1, by solvent extraction with the typical determination of these compounds is car-
hexane}acetone at different proportions and octadecylsilica clean- ried out as a group, performing an acid hydrolysis to
up (n"5) (Reproduced from Bernal JL, del Nozal MJ, Rivera JM, carbon disulRde which is then quantiRed by tech-
JimeP nez JJ, and Atienza J (1996) Journal of Chromatogrphy niques such as headspace GC-ECD.
A 754: 507}513, with permission from Elsevier Science.) When an FPD is used to detect CS2, n-hexane must
Honey Larvae be avoided because it may co-elute, resulting in the
quenching of the S emission in the detector.
Sample amount 1g 5g 1g
EBDCs differ chemically from dithiocarba-
Extractant solvent (%) Recovery Recovery
Hexane (100) 45 42 48 mates because they have reactive hydrogen on the
Hexane}acetone (90 : 10, v/v) 64 63 67 nitrogen atom, which reduces their stability and re-
Hexane}acetone (80 : 20, v/v) 81 82 78 sults in different biological behaviour. One of the
Hexane}acetone (70 : 30, v/v) 99 98 95 most characteristic decomposition products is
Hexane}acetone (60 : 40, v/v) 99 98 96
ethylenethiourea (ETU). GC determination of ETU
Acetone (100) 99 98 95
can only be achieved after derivatization, forming
III / FUNGICIDES / Liquid Chromatography 2915
triSuoroacetylated S-benzyl or butyl ETU derivatives Kidd H and James DR (eds) (1993) The Agrochemicals
that can be analysed by GC-NPD, GC-ECD or GC- Handbook, 3rd edn. London: Royal Society of Chem-
MS. In real samples EBDCs and ETU content de- istry.
crease with storage time. To prevent this, the addition Middleditch BS (1989) Analytical Artifacts. Amsterdam:
of cysteine hydrochloride has been recommended. Elsevier.
Milne GWA (1995) CRC Handbook of Pesticides. Boca
Raton, FL: CRC Press.
See also: II/Chromatography: Gas: Detectors: Selective;
Nielsen SS (1998) Food Analysis, 2nd edn. Gaithersburg,
Detectors: Mass Spectrometry. Extraction: Solid-Phase
MA: Chapman and Hall.
Extraction; Supercritical Fluid Extraction. III/Pesticides:
Pleger K, Manner HH and Weber A (1992) Mass Spectral
Gas Chromatography. Herbicides: Gas Chromatography;
and GC Data of Drugs, Poisons, Pesticides, Pollutants
Solid-Phase Extraction.
and their Metabolites. Parts I, II and III. Weinheim:
VCH.
Further Reading Robinson J (1982) Analysis of Pesticides in Water.
Vol. III. Nitrogen-containing Pesticides. Boca RatoH n:
BarceloH D (1993) Environmental Analysis. Techinques, Ap- CRC Press.
plications and Quality Assurance. Amsterdam: Elsevier. Thier HP and Kirchoff J (eds) (1992) Manual of Pesticide
BarceloH D and Hennion MC (1997) Trace Determination of Residue Analysis, vols I and II. Weinheim: VCH.
Pesticides and their Degradation Products in Water. Tomlin CDS (ed.) (1997) The Pesticide Manual, 11th edn.
Amsterdam: Elsevier. Farnhan, Surrey: British Crop Protection Council.
Inspectorate for Health Protection (1996) The Dutch Man- US Environmental Protection Agency (1990) Methods for
ual of Analytical Methods for Pesticide Residues in Determination of Organic Compounds in Drinking
Foodstuffs, 6th edn. Alkmaar, The Netherlands: Minis- Water. SpringReld, VA: National Technical Information
try of Public Health, Welfare and Sport. Service.
M. Jesuf s del Nozal Nalda, University of Valladolid, tending to displace it for many applications. Some
Valladolid, Spain considerations related to the use of HPLC are sum-
Copyright ^ 2000 Academic Press marized below, with more attention being paid to the
groups of fungicides most frequently determined by
this technique.
Introduction
There are some groups of fungicides of wide use
Technique Selection
(benzimidazoles, ethylenebisdithiocarbamates) whose Most applications are based on the use of reversed-
thermal instability, high polarity and low vola- phase HPLC, nevertheless for some fungicides ion-
tility make them difRcult to determine by gas pair HPLC (ethylenebisdithiocarbanates) (EBDC),
chromatography (GC) unless derivatization methods micellar HPLC (Thiram) or chiral HPLC (Metalaxyl)
are employed. This usually makes the process longer are used. Normal phase HPLC, with amino-bonded
and introduces new errors. These compounds are stationary phases, is sometimes recommended, main-
easily measured by high performance liquid ly for the benzimidazole group.
chromatography (HPLC) as are many pesticides that Chiral HPLC is very important for the determina-
were typically analysed by GC in the past. Integrated tion of enantiomeric purity, mainly for large-scale
systems of solid-phase extraction sample cleanup and synthesis. Resolution of C}chiral enantiomers seems
on line HPLC allows multiple options, not only by to be easier than that of axial}chiral enantiomers
including fungicides of very different polarity in the (atropoisomers).
same analysis but also by achieving very high concen-
tration factors and, at the same time, analysing
a large number of samples. The use of pre- or post-
Columns
column derivatization reactions allows the analysis of The most widely used stationary phases for fungicide
compounds that are very difRcult to determine or residue analysis are the n-octyl and n-octadecylsilica
have a low sensitivity. because they allow the separation of compounds with
Given these advantages HPLC not only comp- a wide range of polarity. Some fungicides, mainly
lements GC in fungicide residue analysis but is EBDCs, are easily ionized and because of this some
2916 III / FUNGICIDES / Liquid Chromatography
methods propose the use of ion-exchange phases. The Suorescence detector gives higher sensitivity
It may, however, be better to use C18 in the ion- and selectivity, so it is preferred for residue analysis
pairing mode, adding a counter ion to the mobile (e.g. benzimidazoles, bitertanol). It is also possible to
phase. When it is necessary to separate enantiomers, programme the excitation and emission wavelengths
then chiral columns are preferred although there is to optimize the signal for all eluted compounds.
also the possibility of using the C18 with a chiral Occasionally the use of an electrochemical detector
mobile phase. is recommended (Phthalimides, Thiram, DisulRram,
Usually columns with a diameter of 4.6 mm, etc.); although it gives great sensitivity, it is more
packed with 5 m material are employed, but now- difRcult to operate, and frequently the electrodes are
adays it is possible to use shorter columns or even contaminated; sometimes, for example for EBDC de-
narrow bore, microbore or packed capillary columns. termination, it is coupled on-line after the UV detector.
These later make the coupling to an MS detector Nowadays there is an increasing trend to MS detec-
easier. In both cases the lower mobile phase Sow rate tion but some difRculties are still encountered when
provides a big reduction in reagent consumption. coupling it to HPLC. It is advisable to use micro-
Several manufacturer’s offer equivalent columns. HPLC and avoid, if possible, the presence of salts in
Attention must always be paid to batch-to-batch re- the mobile phase. This, in addition to its high cost,
producibility. The use of a pre-column helps to pre- means that only a few applications of its use to
serve the life of the column, and, if it is possible to fungicides have been published.
work at room temperature, the column will last lon-
ger than when used with higher temperatures.
Derivatization
A very useful option, in HPLC, is derivatization made
Mobile Phase pre- or post-column, which facilitates compound
The selection of the stationary phase and the mode of detection. A great number of derivatizing reagents
detection is determined by the characteristics of the lead the formation of products with a high absorb-
analytes to be separated; both, also, control the selec- ance or Suorescence, and in fungicide analysis they
tion of the mobile phase. As C18 is usually employed, are often used in post-column reactions but care is
the mobile phase is frequently composed of a mixture needed to minimize band broadening, particularly for
of water with an organic solvent, mainly methanol or slow reactions. The present use of solid phase reac-
acetonitrile. To improve the peak shape or to separate tors has several advantages such as the simplicity in
compounds with acidic or basic character the addi- the instrumentation and compatibility with most mo-
tion of acid or buffer to vary the pH can be very bile phases. A clear example is the monitoring of the
useful. Also the temperature at which the separation carbamate pesticides.
is made, must be established when looking for the As pre-column derivatization can be carried out
best resolution. The reagents used to prepare the with an automatic injector and post-column derivat-
mobile phase must be compatible with the detection ization can be automated with modern devices, this
mode. It is very important when a UV detector is facilitates improved precision.
used, because not all commercial methanol or
acetonitrile are transparent enough in the low UV
region. Attention must be paid to the transmission
Sample Treatment
spectrum of the solvents and to the changes in batch Many applications of fungicide determination require
or manufacturer. When there is a great difference a preliminary sample extraction using an organic
between the polarity of the (aqueous) mobile phase, solvents such as ethyl acetate followed by clean up by
and the organic solvent used to inject the sample, it is liquid}liquid partitioning. Obviously, the matrix has
possible that the Rrst peaks will be distorted and in a great inSuence on the method. The heavy pigment
this case it is better to reconstitute or dilute the content in many crops and vegetables has made the
sample in the mobile phase. popular UV detector almost unusable; even when
analysing fungicides in, for example, citrus, celery
heart, mint and coriander. A large amount of Suor-
Detection escent coextractives can appear, causing inference in
Most of the fungicides that are analysed by HPLC can detection. In these cases the sample treatment must be
be detected in the UV region. In multiresidue methods optimized, including for example a clean up with
it is more convenient to employ a diode array detector Florisil or changing the mobile phase polarity, so the
(DAD) which allows multiple wavelengths to be em- coextracted interference elutes together with the sol-
ployed and peak purity to be checked. vent front. If the sample preparation step can be
carried out using solid-phase extraction, this favours cause of the intense colour of the extracts. As is
direct coupling to HPLC and overall automation, shown, pollen or beeswax are better extracted with
facilitating routine multiresidue analysis. methanol because ethyl acetate extraction gives an
emulsion that makes the separation difRcult. Another
problem that must be taken into account is the inSu-
Analysis of some speci\c fungicide ence of the fortiRcation level on the carbendazime}
groups benomyl recoveries. Thus when analysing vegetable
Phthalimides (Captan, Captafol, Folpet) samples, higher fungicide concentrations added to the
samples results in lower recoveries, even for smaller
For formulation analysis extracting the compound samples. With samples bigger than 2 g, problems also
with diethylphthalate in methylene chloride and appear because the pigments are extracted, giving
chromatography on silica gel using degassed CH2Cl2 a greenish-yellow colour and therefore the determina-
as mobile phase is recommended while for residue tion of carbendazime is hindered. Some relevant data
analysis GC is usually preferred. Nevertheless, re- are shown in Table 1.
cently an isocratic HPLC method using electrochemi- Thiabendazole has also received special attention.
cal detection with single and dual glassy-carbon elec- Several HPLC methods have been proposed for its
trodes has been evaluated, showing good recoveries determination, usually employing ethyl acetate as ex-
and precision and with detection limits of about tractant and C18 columns and acetonitrile}water or
4 g L\. methanol/aqueous buffer mixtures as mobile phases. A
mixture of n-hexane/ethanol/0.2 N HCl, with cation
Benzimidazoles (Benomyl, Carbendazime,
exchange columns or ion pairing HPLC has also been
Thiabendazole, Methyl Thiophanate)
used. Detection can be made by UV (298 nm) although
The high use of these post harvest fungicides means usually Suorescent detection is preferred. Changes in
that many methods have been proposed for the deter- the extractant or of the chromatographic parameters
mination of their residues. require selection of the wavelength to achieve the best
It is possible to evaluate the total content (benomyl, sensitivity. The use of several wave-lengths, mainly
carbendazime, methyl thiophanate) by transforming the couples 285/350 nm and 305/345 nm have
them into carbendazime, by reSuxing at pH"6.8. been proposed. In some cases, the metabolite 5-hy-
Multiresidue methods have been proposed, extracting droxythiabendazole has also been determined.
the sample with HCl and analysing on LiChrosorb Si
60. Recently a clean up on strong cation exchange Dicarboximides (Iprodione, Vinclozolin)
cartridges and analysis on C18 with UV and water-
There are not many HPLC methods to determine
acetonitrile as mobile phase has been proposed, al-
residues of these fungicides. Vinclozolin is perhaps
though ion-pairing HPLC coupled to UV or Suores-
the most frequently used compound and so, some
cence detection can be used. Normal phase HPLC for
methods have been proposed for its analysis, and of
carbendazime can be employed after extraction with
its metabolite (3,5-dichloroaniline) using reversed-
methanol, partitioning in n-hexane-dichloromethane
phase HPLC with UV detection. In complex mixtures
and Suorescence detection at 285/315 nm. Neverthe-
analysis it has also been proved that SPE cartridges
less, the majority of the proposed methods are de-
are more selective than extraction with organic sol-
voted to the study of the pair benomyl}carbendazime,
vents and they provide simpler chromatograms.
using reversed-phase HPLC.
This pair of compounds is normally analysed by
Triazoles (Bitertanol, Triadimefon, Triadimenol,
monitoring carbendazime, although using light petro-
Tryciclazole)
leum ether and a special drying step it is possible to
analyse benomyl without conversion to carben- Residues of Triadimefon and its metabolite Triadimenol
dazime. The analysis involves an extraction with an are seldom determined by HPLC, but they can be
organic solvent (methanol, ethyl acetate or acetone) analysed by reversed-phase HPLC on C18 columns
followed by partitioning with n-hexane or an alkaline with UV detection. The same stationary phase is recom-
solution, using C18 columns and UV detection at mended for Bitertanol, with acetonitrile}water as
224 nm or better by Suorescence at 285/317 nm. The mobile phase and Suorescent detection at 254/322 nm.
type of matrix, even quite similar ones, strongly con- The selection of the extracting solvent (acetone/water,
ditions the sample preparation. In Figure 1 some acetone, methanol) is very important in order to
schemes for the determination of carbendazime in achieve high recoveries. If a cleanup is required, SPE
apiarian samples are shown. In the case of pollen an on C18 cartridges eluted with cyclohexane}ethyl acet-
additional partition with n-hexane is required be- ate seems to be the most adequate.
Figure 1 Flow charts showing the sample preparation procedures used in the analysis of carbendazime in apiarian products.
(Reproduced from Bernal JL, del Nozal MJ, Toribio L, JimeP nez JJ and Atienza J (1997) Journal of Chromatogrphy A 787: 129}136, with
permission from Elsevier Science.)
Dithiocarbamate fungicides Rnal product is the fungicide Nabam. This is water

soluble so some methods are based on this trans-
The US Food and Drug Administration (FDA) has formation. On the other hand, ethylene thiourea
recommended the evaluation of new uniresidue (ETU), ethylene urea and 2-imidazoline are decompo-
methods for the analysis of dithiocarbamates in veg- sition products of the EBDCs. The parent compounds
etables, because of problems which have arisen from have a relatively low toxicity but ETU has been dem-
the application of the carbon disulRde method. This onstrated to be goiterogenic, carcinogenic and
lacks speciRcity because naturally occurring carbon teratogenic, so there is a great interest in determining
disulRde and degradation products of the EBDCs, this compound.
such as dialkyldithiocarbamate and thiuram disulRde, To analyse ETU by HPLC an extraction with meth-
can give serious interference. anol, a clean up on a mixture of sorbents and a mo-
EBDCs have little or no solubility in water and, bile phase of ethanol}isooctane has been frequently
because of this, in many methods the compounds are used. The detection can be electrochemical or by
converted into their soluble sodium salt by means of HPLC/MS, with similar detection limits. Extraction
EDTA and subsequently hydrolysed to form carbon of ETU from vegetables is preferred with methanol
disulRde, but if the hydrolysis is not carried out the and analysis on a CN column, with a mobile phase of
Table 1 Recovery of carbendazime obtained by using an SFE- mon the ethylenebisdithiocarbamate group and there-
hplc procedure on spiked lettuce samples (n"5). (Reproduced fore it is very difRcult to separate them from their
from JimeP nez JJ, Atienza J, Bernal JL and Toribio L (1994)
mixtures. So in some situations it is easier to deter-
Chromatographia 38: 395}405, with permission from Vieweg-
Publishing.) mine the residue of only one fungicide. Another
approach is to try to separate mixtures of three
Sample amount Fortification level Recovery n 1 fungicides belonging to different chemical groups and
\
(g) (mg kg\1 ) (%)
the third and most difRcult one is to try to separate all
0.20 1.0 98.4 3.0 of them.
0.20 6.0 98.3 2.9 Some methods attempt to distinguish between
0.20 12.0 96.4 3.3 compounds using both HPLC and atomic absorption
0.50 0.3 98.3 3.2 methods. This can cause problems because in EBDC
0.50 6.0 98.0 3.4 manufacture there is always an excess of the metallic
0.50 12.0 83.3 3.5 ion that has not been incorporated into the com-
1.00 0.3 98.2 3.3 pound, so if the extraction of the compound is not
1.00 0.6 98.0 3.3 speciRc, the extract will contain not only the metal
1.00 12.0 72.4 3.9
belonging to the fungicide but also the remaining
2.00 0.3 88.3 3.7 metal coextracted. As a consequence, atomic absorp-
2.00 0.6 68.4 5.5 tion data are usually very much higher compared
2.00 12.0 53.7 7.0
with those from HPLC.
To analyse individual fungicides, transition metal
methanol in chloroform/cyclohexane and detection salts are frequently employed as ion-pairing reagents
at 240 nm. Another possibility is to extract ETU and for reversed-phase HPLC with detection in the UV
react it with dihaloquinones to produce a yellow region. According to the complex used the wave-
derivative that can easily be detected at 385 nm. length selected is obviously different, so for Ziram,
There is always a problem with fungicide deter- Maneb and Zineb forming as 1 : 1 Cu(II)-
mination because they are very similar in chemical dithioligand the wavelengths are in the 260}287 nm
structure and behaviour. Ferbam and Ziram have the range, with detection limits at the nM level.
same organic moiety and the difference is in the Sometimes the problem arises of the presence of
metallic ion. Nabam, Maneb, Zineb, Mancozeb and Thiram and DisulRram which could interfere with the
Propineb, frequently used in agriculture, have in com- dithiocarbamate determination. This situation is usu-
Figure 2 Post-column derivatization reaction for EBDCs.

Figure 3 Chromatogram of a mixture of EBDCs. 1 g L\1 each. Detection of 254 nm. Mobile phase: EDTA/MeOH/AcCN
(60 : 13 : 27, v/v).
ally solved by using HPLC on a C18 column with detection at 272 nm. However, the lifetime of the
electrochemical detection or by determining Thiram column is only about 15 analyses.
using micelles of CTAB in the mobile phase, so A good separation between compounds of the three
Nabam, Ziram and Ferbam do not interfere. families (EBDC, PBDC and DMDC) can be obtained
Nabam determination is of interest because EBDC using reverse-phase HPLC on a C18 column, with
fungicides can be converted into this fungicide and so a mobile phase of EDTA 0.05 M, pH"7.7, and
they can be indirectly determined. In this case there is detection in series (UV at 280 nm and amperometric
a method, very similar to one devoted to carbamate at 400 mV), but it is not possible to distinguish be-
residues determination, based on a post-column reac- tween compounds of the same group.
tion through an acid hydrolysis to form ethylene- A method that allows the separation of Rve com-
diamine that is afterwards Suorogenically labelled pounds (see Figure 3) uses ion-pair HPLC with tetra-
with o-phthalaldehyde-mercaptoethanol and detec- butylammonium bromide as counterion on C18 col-
ted at 356/450 nm. The scheme of the post-column umns, with a mobile phase of EDTA/methanol/
reaction is shown in Figure 2. acetonitrile and detection at 254 nm. However, there
Attention must be paid to carrying out the separ- are still some problems because the separation is strong-
ation at the lowest possible temperature. The hy- ly dependent on the analyte concentration, achieving
drolysis temperature must be considered because only the overall separation for very low concentrations.
when the temperature is near 1003C the possibility of As a conclusion, it can be said that the analysis of
more by-products and background noise increases, these fungicides is very difRcult when there are several
and OPA degrades easily at higher temperatures. Sep- of them in the sample and that further work is neces-
aration of Nabam is carried out by micellar HPLC sary.
with a mobile phase of cetylpyridinium (CPC) phos-
phate buffer/acetonitrile. See also: II/Chromatography: Liquid: Derivatization.
III/Fungicides: gas chromatography.
The real problem is the difRculty to convert quant-
itatively EBDCs into Nabam and recoveries lower
than 30% are usually obtained although if EDTA is Further Reading
used, the conversion is favoured. Thus a simpler Aizawa H (1982) Metabolic Map of Pesticides. Orlando:
method has been proposed based on the transforma- Academic Press.
tion into Nabam by means of an aqueous EDTA Frei RW and Lawrence JF (1982) Chemical Derivatization
solution, followed by reverse-phase chromatography in Analytical Chemistry, Volumes 1 and 2. New York:
on an NH2 column with acetonitrile}methanol and Plenum Press.
III / FUSED SALTS: ELECTROPHORESIS 2921
Helrich K (1995) OfTcial Methods of Analysis, 16th edn, Milne GWA (1995) CRC Handbook of Pesticides. Boca
Vol. I. Arlington, VA: Association of OfRcial Analytical Raton: CRC Press.
Chemists. Moye HA (1980) Analysis of Pesticide Residues. New
Krull IS (1986) Reaction Detection in Liquid Chromatogra- York: John Wiley & Sons.
phy. New York: Marcel Dekker. Pawliszyn J (1997) Solid-phase Microextraction. Theory
Lawrence JF (1982) High Performance Liquid Chromatog- and Practice. New York: Wiley-VCH.
raphy of Pesticides. New York: Academic Press. Thurman EM and Mills MS (1998) Solid Phase Extrac-
Lingeman H and Underberg WJM (1990) Detection- tion. Principles and Practice. New York: John Wiley &
Oriented Derivatization Techniques in Liquid Sons.
Chromatography. New York: Marcel Dekker.
FUSED SALTS: ELECTROPHORESIS
M. Lederer, UniversiteH de Lausanne, Lausanne, Techniques

Switzerland
Moving Boundary Method
This article is reproduced from Encyclopedia of Analyti- Migrating boundaries can be observed using a cell
cal Science, Copyright ^ 1995 Academic Press like that shown in Figure 1.
Flat Bed Methods
The interest in this technique is mainly centred
Electromigration in a support to eliminate convection
around the solution chemistry of molten salts, which
is carried out much as in normal electrophoresis,
had its renaissance in the nuclear Reld and in the
study of nonhydrated ions for the purpose of separat-
ing isotopes.
Figure 2 Apparatus for zone electrophoresis in molten salts.

A and A, platinum wires for measurement of potential difference;
B and B, electrodes; C and C, reservoirs; D and D, electrode
Figure 1 Cell for the determination of mobilities by observation compartments provided with sintered discs at the bottom; E,
of migrating boundaries. Reproduced with permission from Her- supporting glass plate; F, electrophoretic strip. Reproduced with
zog and Kelmm (1961). permission from Alberti et al. (1964).
2922 III / FUSED SALTS: ELECTROPHORESIS
Figure 3 Cross-sectional view of an electrophoretic apparatus. A, furnace; B, electrophoresis chamber; C, capillary; D, tube
supporting the capillary; E, screw to raise or lower tube D; F, glass fibre paper; G, heat-resistant glass plate; H and H, graphite
electrodes; I and I, heat-resistant glass vessels; L and L6, thermocouples; M, Ni}Cr heating wire; N, insulating jacket. Reproduced with
permission from Albert et al. (1964).
except that here evaporation is negligible and high Supports

temperatures are used to keep the molten salt liquid.
Asbestos paper and asbestos sheets were initially
Figure 2 shows a typical apparatus. It can be made of
employed as supports; then glass Rbre papers
glass or fused silica. Glass Rbre paper is mainly used
became available (e.g. from Whatman). These papers
as the support material with this apparatus, but it
must be washed to removed impurities and are rather
lends itself equally well to electrophoresis on asbestos
sheets. This kind of apparatus is best housed in an
oven, as shown in Figure 3. Here provision is also
made to circulate a gas to remove gaseous electrolysis
products such as chlorine.
Figure 5 Apparatus for column electrophoresis. 1, separation

Figure 4 Apparatus for column electrophoresis. (A) Front view; column; 2, anode compartment; 3, foam at platinum anode; 4,
(B) side view. 1, electrode space; 2, ground glass joint; 3 tube for cathode compartment; 5, molten zinc cathode; 6, molten salt
gas evacuation; 4, separation column; 5, tube for admitting the bath; 7, stirrer; 8, thermocouple in fixed position; 9, movable
salts to be separated; 6, sintered glass plate. Reproduced with thermocouple for measuring vertical temperature distribution. Re-
permission from KuK hnl and Khan (1966). produced with permission from Ljubimov and LundeH n (1996).
III / FUSED SALTS: ELECTROPHORESIS 2923
Table 2 Effect of mass in electromigration of halogens
Isotopes Fused medium Effect of mass
Cl ZnCl2 0.043
Cl TlCl 0.086
Cl PbCl2 0.052
Br PbBr2 0.042
From Chemla (1959).
as the cathode. Platinum may be attacked by the

alkali metals formed on the cathode during elec-
trophoresis.
Column Electrophoresis
Figures 4 and 5 show two arrangements that can be
used for column electrophoresis.
Fused Salts Used as Electrolytes

Numerous electrophoretic mobilities of cations and
Figure 6 Apparatus used by Klemm for the separation of lith- anions have been published. Most work has been
ium isotopes. Chlorine is admitted at the cathode to prevent the done at relatively low temperatures, i.e. between 150
deposit of metallic lithium. Reproduced from Klemm et al. (1947). and 3003C. A lithium nitrate}potassium nitrate eu-
tectic (43 : 57) can be used at 1603C, a sodium ni-
trate}potassium nitrate eutectic (50 : 50) or lithium
fragile, requiring careful handling. Some authors rec- chlorate}potassium chlorate eutectic (76 : 24) at
ommend converting the surface silanol groups to salt 2503C and 3003C, respectively, and potassium nitrate
forms, e.g. by dipping into 4}6 mol L\1 KNO3 at pH or sodium nitrate at 3503C.
8}9, otherwise the melt does not wet the glass Rbre
paper.
Powdered thin layers for electrophoresis are made
Isotope Separations
by spraying aqueous suspensions of ceramic oxides The separation of isotopes performed in aqueous
onto sintered ceramic strips. solutions is generally poorer than expected from mass
Column electrophoresis is carried out in columns
of glass powder, Al2O3 splinters, or quartz powder.
Electrodes
Platinum is generally used for the anode, and other
materials such as tungsten, nickel, copper or graphite
Table 1 Effect of mass in electromigration of metals
Isotopes of Fused medium Effect of mass
Li LiCl 0.14
Li LiBr 0.26
Li LiNO3 0.05
Zn ZnCl2 0.078
Zn ZnBr2 0.11
K KNO3 0.037
Cu CuCl 0.080
Ag AgCl 0.064
Cd CdCl2 0.067
Tl TlCl 0.040 Figure 7 Some representative separations of inorganic ions by
Pb PbCl2 0.024 zone electrophoresis in molten LiNO3}KNO3 eutectic}10%
NH4NO3 at 1603C. A, application point. Reproduced from Alberti
From Chemla (1959). et al. (1962).
2924 III / FUSED SALTS: ELECTROPHORESIS
Table 3 Movement of metal ions in fused salts
Distance moved KClILiCl eutectic KNO3}LiNO3 eutectic

in 4 h (cm) (T"4503C; 2 V cm\1) (T"1603C; 5 V cm\1)
Anionic 0.5}3 Zn(II), Co(II)

Isoelectric Th(IV)a Th(IV)a
0.5}3 Ce(III) Cd(II)
3}5.5 Pb(II), Cd(II) Pb(II)
Cationic 5.5}8 Cu(II) Sr(II), Ba(II), Cs(I), Rb(I)
8}10.5 Cs(I), Rb(I)
10.5}13 Na(I), Ag(I)
a
Insoluble precipitate formed. From Alberti et al. (1962).
differences, since smaller ions are more hydrated than 16.1% 6Li in 4 days). This work served as the basis
larger ones and thus mass differences are diminished for a commercial separation of lithium isotopes.
with fully hydrated ions. This is not the case in molten Mass effects depend on the temperature as well as
salts and hence the ionic mobility differences of iso- on the anion(s) in the melts. Typical data are shown
topes are nearer to those expected. in Tables 1 and 2.
Countercurrent electrophoresis has been used for
isotope separation of molten salts. In the system
shown in Figure 6 molten lithium chloride is sub- Analytical Separations of
jected to electrophoresis and the lithium metal Inorganic Ions
formed on the cathode is reoxidized with a stream of Table 3 gives some data on the movement of metal
chlorine. At 6503C and a current of 0.5 A, using ions in fused salts and some representative separ-
granular quartz medium to decrease convection, ations by electrophoresis on glass Rbre paper are
rather high enrichments were reported (from 7.3 to shown in Figure 7.
A number of binary metal mixtures have been
separated, as shown in Figure 8.
The main interest in molten salt separations seems,
however, to reside in isotope separations and in the
study of ionic mobilities in molten salts. It is worthy
of note (see Table 3) that Ag# travels as a cation in
a KCl-LiCl eutectic, while it readily forms an anionic
complex AgCl\ 2 in aqueous concentrated HCl
solution.
See also: II/Electrophoresis: Theory of Electrophoresis.

III/Isotope Separations: Gas Centrifugation.
Further Reading
Alberti G and Allulli S (1968) Chromatography and elec-
trophoresis of inorganic ions in fused salts. Chromato-
graphic Reviews 10: 99.
Alberti G, Grassini G and Trucco R (1962) Separation of
inorganic ions in fused salts by electrophoresis on glass
Rber paper. Journal of Electroanalytical Chemistry 3:
283.
Alberti G, Allulli A and Modugno G (1964) Separation of
inorganic ions in fused salts by means of chromatogra-
phy and electrophoresis on glass Rber paper. III. Effect
Figure 8 Separations of inorganic ions by column electrophor- of water, oxygen and support on the migration of inor-
esis in molten KHSO4}K2S2O7 eutectic. Reproduced with per- ganic ions dissolved in the LiCl-KCl eutectic at 4503.
mission from KuK hnl and Khan (1966). Journal of Chromatography 15: 420.
III / GAS ANALYSIS: GAS CHROMATOGRAPHY 2925
Chemla M (1959) SeH paration d’isotopes par chromatog- Klemm A, Hintenberger H and Hoernes P (1947) An-
raphie et par eH lectrophorèse. Chromatography Reviews reichung der schweren Isotope von Li and K durch
1: 246. elektrolytische Wanderung in geschmolzenen Chloriden.
Forcheri S and Berlin A (1967) The determination of trans- Zeitschrift fu( r Naturforschung 2a: 245.
port quantities in molten salts with thin layer elec- KuK hnl H and Khan MA (1966) Journal of Chromatography
trophoresis and diffusion on Rtted ceramic oxides. Jour- 23: 149.
nal of Chromatography 15: 420. Ljubimov V and LundeH n A (1966) Electromigration in
Herzog W and Klemm A (1961) Electronenleitung in ges- molten and solid binary sulfate mixtures: Relative cation
chmolzenem PbCl2}Pb. Zeitschrift fu( r Naturforschung mobilities and transport numbers. Zeitschrift fu( r Natur-
16a: 523. forschung 21a: 1592.
GAS ANALYSIS: GAS CHROMATOGRAPHY
C. J. Cowper, GQ Tech, Walton on Thames, Surrey, UK materials as liquid or solid samples, they can be
modiRed to handle those components in a gas
mixture.
This article is not intended to give details of how to
Introduction analyse all the chloroSuorocarbons or the hydrides of
germanium, but aims to show the characteristic dif-
If gases are deRned so as to be distinguished from ferences in equipment and procedures used for gas
vapours, which is to say only those gases whose analysis.
critical temperatures are below ambient, and hence
cannot be liqueRed by pressure at ambient temper-
ature, then the task of gas analysis appears to be Equipment and Procedures
a simple one. Applying the criterion of a critical
A chromatograph which is conRgured for gas analysis
temperature below, say 153C, produces a very small
will differ in a number of respects from one designed
list of elements and compounds, ranging from helium
for liquids’ analysis. Sample injection will almost
to ethene. It is, of course, more appropriate to deRne
invariably be by valve, and other valves may be used
gases as those which are handled in the gas phase at
to alter the relative positions of different columns
ambient conditions. Widening the criterion to allow
during the analysis. Some columns are speciRcally
components with boiling points below 153C pro-
used for gas analysis, and others may be used in
duces a rather larger list, ranging from xenon to
a different way from that for other applications. Car-
cyclobutane.
rier gas must be chosen with some care, as it may be
Any gas mixture will be based on one or more of
a component of the sample, or have properties which
these components, but can in addition contain higher
do not favour the measurement of sample compo-
boiling compounds whose low concentration allows
nents. The thermal conductivity detector (TCD) is
them to be present without condensing out from the
likely to play a major role; the Same ionization de-
mixture. As an example, natural gas is treated before
tector (FID) may be regarded as a selective detector in
it is distributed so that it remains stable in the gas
this context.
phase over a wide range of temperatures and pres-
sures. It consists predominantly of methane, but con-
Sample Handling and Injection
tains a large number of other hydrocarbons, up to
and including decane, at concentrations which are A packed column will handle sample sizes typically in
amenable to direct analysis by gas chromatography. the region of 0.1 to 10 mg. For samples which are
Analysis for decane and similar components in liquid, or solids dissolved in a solvent, this means
liquid hydrocarbon mixtures is well established, and volumes of 0.1 to 10 L, which are conveniently
similar analytical procedures can be applied to its measured and injected using microsyringes. For gas
measurement in natural gas. The main difference is in samples, this mass range approximates to volumes of
sample handling and introduction. This will generally 0.1 to 10 mL at ambient conditions.
be true for other low concentration components of Gas-tight syringes will easily cope with such
gas mixtures which would normally be liquids or sample sizes, but the main drawback with using
solids. Where techniques exist for analysis of such them is poor repeatability, due to injection of a
2926 III / GAS ANALYSIS: GAS CHROMATOGRAPHY
compressible sample into an already compressed car- sample gas in the loop are constant just before injec-
rier gas. Most chromatographs equipped for gas anal- tion, the technique is capable of excellent sample
ysis will be Rtted with a gas sampling valve, some- size precision, and hence very good quantitative
times referred to as a bypass injector. Figure 1 shows behaviour.
a typical design of a six port valve. It consists of a Gas sampling valves can use other conRgurations
base through which six holes or ports are drilled, } the motion can be linear rather than rotary, but the
equally spaced around the circumference of a circle, principle of isolating the sample in a deRned volume
and a rotor, which rotates around the axis of the same and then purging it on to the column with carrier gas
circle. The rotor has three grooves machined into it, remains unchanged. Valves with more ports can also
which connect adjacent pairs of ports. Rotation of the be used, to combine gas sampling with other switch-
rotor through 603 alters the internal plumbing by con- ing operations which may be required.
necting different pairs of ports. The ports are connec- Capillary columns need much smaller sample sizes
ted into the chromatograph by small bore tubing. for efRcient operation, and hence the proliferation of
In Figure 1, the carrier gas inlet is connected to techniques for liquid samples, ranging from sample
port 1, and port 2 goes to the column. Sample gas splitting to solvent effects and retention gaps. It is
enters and exits through ports 5 and 4 respectively. possible to minimize the dead volume in a gas samp-
A sample loop is connected between ports 3 and 6. ling valve to allow direct injection, but this becomes
The sample loop, usually a length of 2 mm i.d. tubing, more difRcult as the column i.d. is reduced. If a gas
deRnes the size of sample injected. Figure 1A shows sample is stable when introduced into the chromato-
the sample loading position, with the carrier gas graph, it is most likely that splitting the carrier gas
going directly to the column, and Figure 1B shows Sow downstream of the valve will give a representa-
the inject position. Here, the carrier gas sweeps the tive sample on to the column. In most instances this
entire contents of the sample loop on to the column. would be the preferred option.
Provided that the temperature and pressure of the
Columns
Gas chromatographic separations are mainly in-
Suenced by the volatility of the components of the
mixture. By using selective stationary phases, groups
of components of higher polarity can be retained in
the column for longer than components of lower
polarity. Within similar groups, however, the order of
elution will be dictated by boiling point. It is also the
case that an isothermal analysis would use a column
temperature of somewhere around the middle of the
boiling range of the sample, and a temperature pro-
gramme would very approximately mimic the distil-
lation characteristics.
Permanent gases in particular do not have boiling
points which would suggest a convenient choice of
column temperature, and polarity differences do not
appear to be strong enough to be helpful. Although
columns have been operated at liquid nitrogen tem-
perature for the separation of hydrogen isotopes, in
general subambient operation is unattractive. Gas
analysis, therefore, requires different separation
mechanisms to allow use of more or less standard
equipment at normal temperatures. The principle
difference is the use of adsorption onto stationary
phases with active surfaces as the means of separ-
ation, rather than partition into a dispersed liquid
phase. Such adsorption phases separate, at least
in part, by molecular size or shape. As a consequence,
some have relatively limited applicability, and are
Figure 1 Gas sampling valve. (A) Sample loading. (B) Sample used as part of a range of columns required for a com-
injection. plete analysis.
Molecular sieve The term is general, but in the con-

text of gas analysis it refers to aluminosilicates of the
alkali and alkaline earth metals. The most commonly
used are type 5A, based on calcium, and 13X, based
on sodium. They have average pore diameters of 0.5
and 1 nm respectively. Molecular sieves are the only
materials available which can separate oxygen and
nitrogen at normal chromatographic temperatures
(35}1003C). On the other hand, they retain carbon
dioxide under the same conditions for so long that
it is sometimes regarded as being permanently
absorbed. They also strongly adsorb water, and are
widely used as drying agents. Figure 2 Molecular sieve 5A.
Molecular sieves must be activated before use, to
drive off water and other strongly adsorbed materials
and to make the pores available to sample compo- a porous-layer open-tubular or PLOT column. The
nents. This can be done in situ, by heating the column combination of high efRciency and high carrier gas
to around 3003C for several hours with dry carrier linear velocity means that argon and oxygen can be
gas Sowing. During use, a column will slowly lose separated within an overall analysis time comparable
separating power due to adsorption of moisture from to that of a 2 m packed column.
samples or carrier gas, and will eventually need to be 5A molecular sieve is also capable of exclusion
conditioned again. With reasonable precautions, chromatography, based on molecular shape. Ethane,
a column should continue to give good separations propane and n-butane have increasingly longer reten-
for a year or longer. tion times, so as to be unmeasurable, under the condi-
5A and 13X sieves have broadly similar behaviour, tions of Figure 2. However, isobutane (2-methylpro-
with some detail differences which may cause one to pane) elutes as a tailing peak just before methane;
be preferred over the other for particular applica- Figure 3 shows this effect. Similar behaviour is
tions. When each is packed into a typical 2 m column, found for neopentane (2,2-dimethylpropane), al-
5A sieve will give longer retention times, most evi- though it is less of a problem, since neopentane is
dently for carbon monoxide. Figure 2 shows likely to be at much lower concentration than iso-
a chromatogram, from a 5A molecular sieve column, butane. A further example is sulfur hexaSuoride,
of helium, hydrogen, oxygen, nitrogen, methane and which elutes before oxygen. This can be used where
carbon monoxide, using argon carrier gas at 503C. SF6 is measured as an atmospheric tracer by electron-
Under these conditions, rare gases can also be ana- capture detector; oxygen is mildly electron-capturing,
lysed, with neon eluting just after hydrogen, argon but does represent 21% of the atmosphere, and so
co-eluting with oxygen, krypton just before nitrogen, having the trace of SF6 eluting Rrst makes detection
and xenon after carbon monoxide. Component rela- easier.
tive retention times are inSuenced by the temperature Under the same conditions, 13X molecular sieve
and time of activation, and so in the unlikely event of gives more uniform separation of components, with
rare gases being signiRcantly present in a mixture rather shorter retention times, as shown in Figure 4.
such as that in Figure 2, it should be possible to Rnd 13X sieve does not display the exclusion mode of
an activation procedure which will allow all to be
separated.
Argon is the most abundant rare gas (0.93% v/v in
air), and so its co-elution with oxygen can create
a problem. It can be resolved before oxygen by using,
for example, a 2 m column at !503C, or a 5 m
column at 03C. The obvious drawback is the excess-
ive retention times for other components. An alterna-
tive way of measuring oxygen without interference
from argon is to use argon as carrier gas, as in
Figure 2.
Capillary columns containing 5A sieve are avail-
able. The Rnely divided material is dispersed as
a layer on the wall of the capillary. This is known as Figure 3 Molecular sieve 5A.
Figure 4 Molecular sieve 13X.

Figure 6 Porous polymer beads as packing material.
behaviour shown by 5A. Another difference is that

the order of elution of methane and carbon monoxide overnight near the maximum operating temperature
can be reversed. If the temperature of acti- will remove residual low relative molecular mass ma-
vation is limited to 1503C, then a chromatogram terial and give more stable baselines. With operating
similar to that in Figure 5 is produced. With this temperatures from subambient to around 2503C, the
reduced level of activation, the long-term stability of range of samples to which they can be applied is
the column is excellent. With the higher activation large. Since there is no change in activity due to
temperature required for Figure 4, the column perfor- adsorbed moisture, porous polymers are frequently
mance, as with 5A sieve, slowly deteriorates. 13X used in temperate-programmed applications, which
sieve must have at least two types of pores, from one considerably increases their Sexibility.
of which water is removed at relatively low temper- A 2 m column packed with nonpolar material at
ature, giving the chromatogram in Figure 5; the other 503C does not separate oxygen, nitrogen and carbon
requiring higher temperatures to give the perfor- monoxide, which elute at the beginning of the
mance in Figure 4. Obviously, at some intermediate chromatogram, closely followed by methane. It does
activation temperature, methane and carbon monox- separate carbon dioxide, ethene and ethane in that
ide will co-elute. order, which makes it a natural complementary col-
umn to molecular sieve for light gas analysis
Porous polymer beads Porous polymer beads are (Figure 6). Propene elutes just before propane, but
based on polyaromatic cross-linked resins. They have C4 saturated and unsaturated hydrocarbons are
a regular pore size and form beads of uniform dia- mixed together. At higher temperatures, porous poly-
meter, making a good packing material. They are mers can analyse hydrocarbons to C8. They are also
available in a range of polarities, according to the good for sulfur-containing gases, separating H2S,
method of preparation, which allow differences in COS and SO2 in that order. With temperature
elution order. They are not hygroscopic and hence programming, organic thiols and sulRdes can be
need no activation before use, although treatment included.
Alumina Activated alumina has a high polarity

which is suitable for mixtures of saturated and un-
saturated hydrocarbons. To avoid tailing peaks for
unsaturated components, some controlled surface de-
activation is necessary. Originally this was done with
water, or a mixture of water and silicone oil to obtain
the desired polarity. The water would slowly be strip-
ped off by the dry carrier gas, increasing polarity and
the tendency to tailing peaks. Alumina PLOT col-
umns are now available, deactivated with inorganic
salts, and these offer the optimum solution for this
type of analysis. Figure 7 shows a chromatogram of
C1 to C4 saturated and unsaturated hydrocarbons.
Figure 5 Molecular sieve 13X } partially activated. This was produced using an FID, so the lack of
Figure 7 Porous-layer open-tubular (PLOT) alumina column.
separation between methane and inorganic gases is materials in the sample path, not just the packing
not a problem. material, must be considered.
Column Switching
Carbon Active carbon has been used as a packing
since the introduction of gas analysis by chroma- Most gas analyses require the use of more than one
tography. Modern packings are based on graphitized column, given the restricted applicability of each.
carbon black or on carbon molecular sieve. Both have Rather than use the columns individually in separate
similar retention characteristics to porous polymers, chromatographs, they can be combined in a single
but carbon molecular sieve, with a high surface area, unit by means of switching valves. Such valves are
requires higher temperatures. Another characteristic similar to the gas sampling valve described earlier,
of carbon molecular sieve is its very low retention for but conRgured for different uses.
water, typically eluting before CO2. Either type of Figure 8 shows a conRguration suitable for the
packing is used for samples containing adsorptive common combination of molecular sieve and porous
components, in which case the inert nature of all the polymer columns. V1 is the gas sampling valve and
Figure 8 Column isolation.

V2 the column switching valve. Column 1 is porous

polymer and column 2 molecular sieve. The restrictor
on V2 is adjusted so that the carrier Sow to the
detector is the same for both positions of V2. Those
components for which molecular sieve is appropriate
(O2, N2, CH4 and CO) are rapidly eluted from the
porous polymer column with little or no separation.
With V2 in the position shown, they pass into the
molecular sieve column. Before CO2 and other com-
ponents have reached the end of the porous polymer
column, valve V2 is switched, allowing them to by-
pass the molecular sieve and pass directly to the
detector. Switching V2 also isolates the light compo-
nents in the molecular sieve column, with no carrier
gas Sow. After the components directly eluted from Figure 10 Combined injection and switching.
the porous polymer column have been detected, V2 is
switched back, and the light gases are measured.
Figure 9 shows a chromatogram. Multiple systems Natural gas and reRnery gas are
Another procedure is possible. If the gap on the two examples of complex mixtures which require
porous polymer column between the initial un- more elaborate systems than those described above.
resolved components and CO2 is sufRciently large, Although neither mixture should contain oxygen,
then the Sow can continue through the molecular sieve a molecular sieve column is recommended because air
to allow the light components to be measured before contamination of the sample is always possible;
CO2 reaches V2. V2 is then switched, allowing CO2 measurement of the oxygen concentration allows the
and the other components to be measured. air content to be calculated and allowed for. A porous
polymer column allows optimum separation of car-
Combined use For certain applications, one 10-port bon dioxide, ethane and, if present, ethene. C3,
valve can be used in place of two 6-port valves. C4 and C5 hydrocarbons can be dealt with by an
Figure 10 shows the conRguration which would appropriate liquid-phase column, and in a common
allow the second procedure described above to be implementation all heavier hydrocarbons are mea-
achieved with one valve. With the valve in the Rrst sured as a backSushed C6# group.
position, sample is being purged through the loop, Figure 11 shows the conRguration. Column 1 is
and the carrier gas is following the sequence column divided into two sections, the longer 1B, on which C3,
2Pcolumn 1Pdetector. Switching the valve injects C4 and C5 hydrocarbons are separated, and the
the sample on to column 1 (porous polymer). After shorter 1A, from which C6# is backSushed.
the light gases from column 2 have been measured, Valve V1 injects the sample and also alters the se-
the valve is returned to the sample load position for quence of columns 1A and 1B. Valve V2 isolates and
measurement of CO2 and other components. reconnects the porous polymer (column 2) and mo-
lecular sieve (column 3), and V3 does the same for
column 3 alone.
After sample injection, V1 is left in that position
until the C5 components have passed on to column
1B, while the C6 and heavier are still on column 1A.
This time is found by trial and error. At this point, V1
is returned to the sample load position, which also
reverses the Sow through column 1A and places it
directly before the detector. All heavier components
are rapidly measured as a recombined C6# peak.
After the light gases have been eluted from column
1B, rotation of V2 isolates them in columns 2 and 3.
The timings and column lengths are chosen so that
each group of light gases is isolated on the appropri-
ate column. C3, C4 and C5 hydrocarbons elute from
column 1B, pass through column 1A for the second
Figure 9 Combined column chromatogram. time, and are detected. After that, valves V2 and V3
Figure 11 Multiple column analysers.
are rotated to reconnect column 2, and CO2 and best choice for measurement of other trace compo-
C2 hydrocarbons are measured. Finally, V3 recon- nents in air.
nects column 3, and the components from the mol- When other detectors, such as the FID, are used,
ecular sieve column are measured. Figure 12 shows the choice of carrier gas is less critical. Maximum
a chromatogram of natural gas. sensitivity from the FID is obtained with nitrogen or
argon, but this is rarely the most important need.
Carrier Gas
Helium allows higher carrier gas linear velocities
Since the TCD is normally used for the major compo- without loss of efRciency, and is also preferred in
nents of gas mixtures, the thermal conductivity of the cases where both TCD and FID are used in series.
carrier gas is one of its most important properties.
Detectors
Helium and hydrogen have similar high thermal con-
ductivities, and allow other components to be detec- For speciRc applications, most types of detector are
ted with good sensitivity. All other things being equal, used or have been used. Electron capture, helium
helium would be preferred for its inertness. Unfortu- ionization, ultrasonic and Same photometric de-
nately, the thermal conductivities of helium/hydrogen tectors are applicable where the properties or very
mixtures have a signiRcantly nonlinear relationship to low concentrations of sample components require
concentration, which makes quantitative measure- them. For the majority of applications, however, the
ment of hydrogen in helium carrier gas difRcult. Ar- TCD is the most popular, followed by the FID. TCDs
gon or nitrogen is suitable for the measurement of use either Rlaments or thermistors as sensing ele-
hydrogen and/or helium. As described above under ments. Thermistors give greater sensitivity at lower
‘Molecular sieve’, argon can be used to measure oxy- detector temperatures, but are less good if the ap-
gen without interference, and nitrogen may be the plication demands a high detector temperature. When
Figure 12 Natural gas chromatogram.

using adsorption columns, it is not always necessary through a membrane into a measured diluent gas Sow
to maintain a detector temperature higher than that under controlled conditions, are widely used for trace
of the column oven, and thermistors may still be analysis.
preferred. The higher temperatures of Rlaments can If only a few speciRc components are to be mea-
cause sample components to react with them. As an sured, then the individual response factors calculated
example, a sample containing both oxygen and hy- from the calibration gas are critical. It is necessary to
drogen sulRde can cause step baseline changes in calibrate with sufRcient frequency that uncontrol-
opposite directions as each component in turn passes lable effects, such as that of barometric pressure on
through the detector. TCD response, do not degrade the result. Consistency
of sample size between calibration and analysis is also
Quantitative Analysis important. A gas sampling valve (see above under
‘Sample handling and injection’) allows this, but the
The importance of calibration is as great for analysis operator must ensure that the temperature and pres-
of gases as it is for liquid samples. In some respects it sure of the calibration gas and sample are uniform.
is even greater, since the TCD, which is used for the If the analysis is comprehensive, as in the natural
major components of gas samples, does not share the gas example shown in Figure 12, then the resulting
‘carbon counter’ characteristic of the FID. If, for composition will be normalized to 100%. This
example C3 to C5 hydrocarbons in a gas are analysed procedure effectively converts the individual response
using an FID, it would be possible to calibrate using factors to relative ones. Since relative response factors
a standard gas containing C3 alone, and then quantify for a single detector remain stable over long periods,
the other components by means of relative response consistency of sample size is less critical. Regular
factors. This is not the case when using TCD; while calibration is still important, but is used as much as
the relative response for a particular detector will be a quality control test as for calibration in the tradi-
stable, they are not predictable as for the FID, and tional sense.
there is not a sufRcient basis of knowledge to allow
them to be transferred to other models of TCD from
other suppliers. Further Reading
Calibration gas mixtures of the highest accuracy Cowper CJ (1995) The analysis of hydrocarbon gases. In:
can be prepared in cylinders gravimetrically. Al- Adlard ER (ed.) Chromatography in the Petroleum In-
though the mass of the cylinder is very much greater dustry. Amsterdam: Elsevier.
than the masses of the added components, the dis- Cowper CJ and DeRose AJ (1983) The Analysis of Gases by
crimination and accuracy of weighing are such that Chromatography. Oxford: Pergamon Press.
the uncertainties of the composition are very small. Jeffery PG and Kipping PJ (1972) Gas Analysis by Gas
Chromatography. Oxford: Pergamon Press.
Such mixtures are, of course, expensive, and for regu-
Leibrand RJ (1967) Atlas of gas analysis by gas chromatog-
lar use certiRed calibration mixtures, which have been raphy, Journal of Gas Chromatography 5: 518}524.
analysed against a gravimetric mixture, are normal. If Mindrup R (1978) The analysis of gases and light hydrocar-
components are likely to react with or adsorb on to bons by gas chromatography. Journal of Chromato-
cylinder walls, then calibration mixtures can be pre- graphic Science 16: 380}389.
pared dynamically, at the time and place of use. Thompson, B. (1977) Fundamentals of Gas Analysis by
Permeation tubes, where a component diffuses Gas Chromatography. California: Varian Associates.
GAS CENTRIFUGE: ISOTOPES SEPARATION

See III / ISOTOPE SEPARATIONS: Gas Centrifugation
GAS CHROMATOGRAPHY-MASS SPECTROMETRY

IN MEDICINE
See III / BIOMEDICAL APPLICATIONS: Gas Chromatography^Mass Spectrometry
III / GAS SEPARATION BY METAL COMPLEXES: MEMBRANE SEPARATIONS 2933
GAS SEPARATION BY METAL

COMPLEXES: MEMBRANE SEPARATIONS
N. Toshima and S. Hara, Science University of Tokyo temperature or pressure:
in Yamaguchi, Yamaguchi, Japan
Copyright ^ 2000 Academic Press S}M#G & S}M2G [1]
Figure 1 shows a schematic illustration of carbon

Introduction monoxide separation from a carbon monoxide}nitro-
gen gas mixture by a solid copper complex. Only
Gases are separated on a very large scale by cryogenic carbon monoxide can coordinate to copper to form
distillation, membrane or sorption methods. For a complex when the adsorbent is exposed to the gas
instance, the production of pure oxygen from air mixture. The adsorbed gas is released when the ad-
by cryogenic distillation is one of the most important sorbent is subjected to high temperature or reduced
separation processes. However, the membrane pressure.
method is preferable when O2-enriched air is The characteristics of metal complex adsorbents
required for medical use or effective combustion, are as follows:
and the sorption method is best for removal of
O2 from packaging. The fundamental mechanism 1. Various combination of metal}ligand systems are
of cryogenic distillation is due to a difference in available for the development of appropriate ad-
boiling points of the various gases. In the case of sorbents.
the membrane method, a difference in solubility 2. Even a very dilute gas component in a mixture can
and diffusion of gases is essential for nonporous be removed due to strong coordinative binding
membranes, while molecular mass or size of gases between the central metal and gas molecules.
is a decisive factor for porous membranes where 3. Selective adsorption can be achieved using suitable
Knudsen diffusion or molecular sieving occurs. functional groups of ligands.
The sorption method involves physical sorption 4. Such a selective adsorption can possibly be used as
and chemisorption, which are characterized by a sensor or an indicator by showing colour change
binding energies of about 5}50 kJ mol\1 and about of adsorbents.
150}500 kJ mol\1. 5. The amount of adsorbed gas molecules per unit
Unlike most physical sorbents, gas adsorbents con- mass of the adsorbents is limited because only one
taining metal complexes adsorb gas molecules by or two molecules can coordinate to a central metal
chemical coordination of the molecules to a central atom.
metal atom. Some metal complex adsorbents, be- 6. The metal complex adsorption system cannot sep-
cause of their strong chemical adsorption, can com- arate gases by their molecular size.
pletely adsorb gas molecules, even at low gas concen- 7. In general, the metal complex adsorbent is less
trations. The equilibrium between the metal complex stable than other types of adsorbents, such as
(S}M) and the gas (G) is shown in eqn [1]. When this active carbon and zeolite.
equilibrium shifts to the right, the adsorbent adsorbs
gas molecules. When the equilibrium shifts to the Metal complexes can be used as free-standing
left, the adsorbent releases the adsorbed molecules. adsorbents, but preferably should be spread on to
The equilibrium can be controlled by changing an inert support, such as activated carbon, porous
Figure 1 A schematic illustration of the separation of carbon monoxide from a carbon monoxide}nitrogen mixture by a supported
copper complex.
2934 III / GAS SEPARATION BY METAL COMPLEXES: MEMBRANE SEPARATIONS
carbon Rbres, zeolite or porous polymer beads. These The following are industrially important applica-
supports should have a large surface area to increase tions of metal complexes for gas separation. How-
the adsorption capacity. Also the adsorbent support ever, most have been investigated on just a small
should be mechanically strong so as not to break up laboratory scale. Only the separation of oxygen from
in use. air has been the subject of practical research for
Polymeric supports have several advantages. First- industrial application on a large scale.
ly, polymers can be processed into many shapes, such
as thin membranes, hollow Rbres or porous beads.
Secondly, the wide variety of available polymers
Separation of Carbon Monoxide
means that the structure and surface chemistry of the Carbon monoxide, one of the most important starting
polymer are easily changed. Thirdly, the metal com- materials for synthesizing organic chemicals, is usu-
plex can be easily inserted into a hydrophobic ally obtained as a gas mixture containing methane,
domain, which cannot be achieved with inorganic ethane, nitrogen, carbon dioxide and water vapour,
materials. This offers some protection to water-sensi- by steam reforming of hydrocarbons, partial
tive metal complexes. Finally, composite materials oxidation processes or gasiRcation of coal. Thus,
can be easily formed. separation of carbon monoxide from gas mixtures is a
Polymer-supported metal complexes or poly- signiRcant target for investigation. Although the
mer}metal complexes are quite often more stable, cryogenic separation process is widely used for car-
more effective, and more easily handled than metal bon dioxide separation, it is difRcult to separate car-
complexes without support for the following reasons: bon monoxide from gas mixtures containing N2 be-
cause their boiling points are close (CO"!191.53C,
1. Diluting effect: polymers can immobilize the metal N2"!195.83C). For this reason adsorption of car-
complex separately so that aggregation of the bon monoxide onto metal complexes is still of use.
metal complex is difRcult. One process is adsorption of carbon monoxide by
2. Concentrating effect: metal complexes Rxed in a copper liquor process using copper(I) ammonium
a polymeric support give a reaction Reld with solution, but dry systems have also been proposed.
a higher concentration of active centres than in Some adsorbents investigated are listed in Table 1.
a solution of the corresponding metal complex. Introduction of copper(II) ion into Y-type zeolite
3. Field effect: polymeric supports form a speciRc by cation exchange, followed by reduction under
reaction Reld to promote the reaction or to inhibit 150 mmHg (1 mmHg133.3 Pa) of carbon monox-
undesirable side reactions. ide at 4003C, gives a Y-type zeolite-supported Cu(I)
4. Steric effect: polymeric supports sterically control adsorbent which adsorbs 1.65 mmol of carbon mon-
the approach of molecules to the metal complex. oxide per g-adsorbent at 253C, 99 mmHg. The
amount of CO adsorbed is almost equal to the moles
The polymer}metal complexes can be used for gas of attached Cu(I) ion, suggesting that a 1 : 1
separation in the form of adsorbents and membranes. Cu(I)}carbon monoxide complex is formed. A similar
In both cases, the coordination of gas molecules to the adsorbent can be made by replacing Y-type zeolite by
metal complex is much stronger than physical ad- ZSM-5. The disadvantages of the Cu(I) zeolite-type
sorption or solubilization into polymers, resulting in adsorbents are as follows:
higher selectivity for gas separation than metal com-
plexes supported on inorganic materials where the 1. Cu(I) is oxidized to Cu(II) by oxygen in the
inorganic support itself may adsorb a considerable presence of water vapour or ammonia, losing the
amount of gas molecules. ability to adsorb carbon monoxide.
Table 1 Adsorption and release of CO by various adsorbents containing Cu
Adsorbent Capacity of adsorption Adsorption conditions Release conditions

(cm3 g adsorbent I1)
Cu/Y-type zeolite 30}40 100 Torr, (1003C '3003C, vacuum

Cu/ZSM-5 16 200 Torr, (503C '3003C, vacuum
Cu(I)/active carbon 24 0.9 atm, 203C 1203C, 1 atm or 203C, vacuum
Cu(I)/PS-NH2 20 0.9 atm, 203C 803C, vacuum
AlCuCl4/active carbon 28 0.9 atm, 203C 1803C, vacuum
AlCuCl4/PS 70 1.0 atm, 203C 903C, vacuum
2. Release of carbon monoxide requires a temper- Table 2 Capacity of CO adsorption for various ion exchange
ature of 300}4003C under vacuum because of the resin-supported CuCla
strong carbon monoxide coordination of Cu(I).
Ion exchange resin Functional group Adsorbed
These severe conditions are unsuitable for good CO b
carbon monoxide recovery.
Anion exchange resin INH2, INHI 15.9
Although active carbon is capable of adsorbing Weak cation exchange resin ICOOH 1.2
Strong cation exchange resin ISO3H 0.8
gases, it cannot be used as a selective carbon monox-
Polystyrene resin None 2.2
ide adsorbent because of the general physical adsorp-
tion involved. However, activated carbon, having a
The adsorbents were prepared from 10.0 g of resin and 9.9 g
a very large surface area, can be used as a support. An (100 mmol) of CuCl in 80 cm3 acetonitrile.
b
adsorbent can be made holding 14.1 mmol of CuCl Adsorbed CO under 1 atm of CO}N2 (9 : 1) mixture at 203C.
(94%), with a surface area of 744 m2 g\1, about 70%
of that of the activated carbon (1044 m2 g\1) used.
The activated carbon-supported CuCl adsorbs 88% salt with aluminium chloride, and the double
carbon monoxide per Cu(I) (24 cm3 STP per g-adsor- salt forms a -complex with aromatic hydrocarbons
bent) under a CO}N2 (9 : 1) mixture at 203C. such as toluene, giving a stable carbon monoxide
The adsorbent totally releases the adsorbed carbon absorbent even in the presence of oxygen or carbon
monoxide at 1203C under 1 atm or 203C under dioxide:
0.4 mmHg, and can be used repeatedly. The high
capacity of carbon monoxide adsorption results from
AlCuCl4(toluene)#CO & AlCuCl4(CO)#toluene
the highly dispersed CuCl\ 2 on active carbon pre-
pared from CuCl in hydrochloric acid solution. Car-
bon-supported CuCl prepared in ethanol or water This absorbent (AlCuCl4 toluene solution, so-called
does not show high carbon monoxide adsorption. COSORB solution) is very unstable in water, how-
Carbon-supported CuCl prepared from CuCl in 20% ever, decreasing irreversibly the capacity of carbon
ammonia solution exhibits 83% carbon monoxide monoxide absorption. Therefore, the water content
adsorption to the Cu(I) content under the same condi- of gas mixtures must be reduced to less than 1 p.p.m.
tions mentioned above. A similar carbon monoxide prior to using this solution. Addition of linear PS to
adsorbent, prepared from CuCl2 instead of CuCl, an AlCuCl4}toluene solution dramatically increases
adsorbs 67% carbon monoxide to the CuCl2 added. the stability to water, presumably because PS coordi-
The active site is however found to be Cu(I) by X-ray nates AlCuCl4, forming a hydrophobic domain
photoelectron spectroscopy (XPS), presumably be- around the water-sensitive AlCuCl4 molecule, there-
cause the Cu(II) is reduced by the activated carbon by protecting it from water.
during the preparation. MR cross-linked PS resin (Bio-Beads SM-2, divinyl-
Macroreticular (MR) polystyrene (PS) resin with benzene content: 20%, surface area: 300 m2 g\1) is
primary and secondary amino groups can be used as an excellent support for AlCuCl4, providing a water-
a support for CuCl. The polymeric adsorbent adsorbs resistant solid carbon monoxide adsorbent. The ad-
15.9% carbon monoxide per Cu(I) (21 cm3 STP per sorbent adsorbs an equimolar amount of carbon
g-adsorbent) under the same conditions described monoxide in 10 min under 1 atm of carbon monoxide
above. The adsorbent partially releases carbon mon- at 203C, and partially releases the carbon monoxide
oxide in 10 min at 803C under 1 atm, and repeatedly under 7 mmHg at 203C. After 10 min desorption of
adsorbs 9.1% carbon monoxide per Cu(I). The other CO, the adsorbent still adsorbs 54% carbon monox-
ion exchange resins show poor capacity of carbon ide per Cu(I) the second time, and this is a reversible
monoxide adsorption (Table 2), showing that the co- capacity for carbon monoxide adsorption under the
ordination of the amino groups to Cu(I) is essential. conditions mentioned above. The capacity of revers-
The adsorbent does not adsorb methane, hydrogen or ible carbon monoxide adsorption increases by ap-
nitrogen, but does adsorb some carbon dioxide plying higher vacuum or higher temperatures during
(3.7 cm3 per g-adsorbent). The capacity is however the desorption part of the cycle. The water-resistivity
much smaller than that of carbon monoxide (21 cm3) of the adsorbent strongly depends on the solvent used
because most of the amino groups of the resin coordi- for preparation. Carbon disulRde gives a water-resis-
nate to CuCl, and few free amino groups remain. tant adsorbent, while toluene gives a water-sensitive
Polystyrene-supported CuCl shows very poor material. The water-resistant adsorbent has a uni-
capacity for carbon monoxide adsorption (Table 2). form distribution of AlCuCl4 in PS resin, whereas the
Copper(I) chloride is known to make a double water-sensitive adsorbent possesses a lot of crystalline
deposits, consisting mainly of CuCl, in the beads and Table 3 Effects of the washing solvent on the ability for NO
much salt, mainly AlCl3, on the surface. adsorption and the surface area for chelate resin-supported Fe(II)
Gel-type PS (divinylbenzene content: 1%)-sup- Solvent Adsorption rate a Adsorbed NO b Surface areac
ported AlCuCl4 adsorbs carbon monoxide much (10\2 mmol min\1) (m2 g\1)
more slowly than the MR-type adsorbent, and it
Chloroform 0.36 0.789 2.3
takes about a day to attain equilibrium. In conse-
Water 0.30 '0.425 3.2
quence, the MR-type resin, which possesses macro- Acetone 1.04 0.933 17.2
pores even in a dry state, is essential for the prepara- 2-Propanol 31.7
tion of solid gas adsorbents. Ethanol 1.54 0.997 31.1
Methanol 1.54 0.997 43.1
a
Separation of Nitrogen Monoxide (NO) b
Amount of adsorbed NO in 15 min.
Equilibrium NO adsorbed against NO introduced.
Nitrogen oxides and sulfur oxides are major air pollu- c
Measured by a Brunauer}Emmett}Teller (BET) method.
tants. Unlike sulfur oxides which have been decreas-
ing in the atmosphere, nitrogen oxides, particularly
nitrogen monoxide (NO) in combustion and exhaust 1203C. The apparent equilibrium constant of the
gases, still cause difRculties. For removal of nitrogen adsorbent is around 1500 atm\1.
monoxide, in general, catalytic processes using am- Iron(II) ethylenediaminetetraacetic acid (EDTA)
monia, carbon monoxide or hydrocarbons as a reduc- solution is a well-known nitrogen monoxide absorb-
ing agent are applied. ent, adsorbing 0.5 mol of nitrogen monoxide per
Another possibility of removing nitrogen monox- mole of Fe from NO}N2 mixtures (NO: 1000 p.p.m.)
ide from gas mixtures might be the use of metal at 553C. However, the absorbent is easily oxidized
complex adsorbents at lower temperatures. The ni- by oxygen due to the unstable Fe(II) ion. To overcome
trogen monoxide adsorbents are expected to adsorb this problem, a chelate resin, cross-linked PS with
nitrogen monoxide completely, even at very low con- iminodiacetic acid moieties, has been used as a poly-
centrations of nitrogen monoxide, and to be stable to meric support. The Fe(II) supported on chelate
water vapour, SO2, dust, etc., which are often con- resin adsorbs almost all the nitrogen monoxide in
tained in combustion and exhaust gases. 15}20 min from a 6 dm3 of NO}N2 mixture
Natural jarosite, KFe3(SO4)2(OH)6, possesses a (NO: 1000 p.p.m.) when the gas mixture is circulated
layer structure in the crystal. Jarosite synthesized from at 1.6 dm3 min\1. The adsorbent releases all the ad-
Fe(SO4)2 and K2SO4 (3 : 1) adsorbs 5.3;10\4 mmol sorbed nitrogen monoxide under 3 mmHg at 1003C,
of nitrogen monoxide per g-adsorbent (surface and can be reused repeatedly. The capacity of nitro-
area: 7.3 m2 g\1). The other synthetic jarosites, gen monoxide adsorption for the chelate resin-sup-
MFe3(SO4)2(OH)6 (M"Na, Rb), show no difference ported Fe(II) is dependent on the washing solvent.
in nitrogen monoxide adsorption. This is because the porosity or the surface area
-FeOOH crystals can be prepared from Fe2(SO4)3 increases (Table 3) when the water-swollen resin-
and Na2CO3 by heat treatment. Powdery crystal - supported Fe(II) is dried after washing with water-
FeOOH, whose surface area is 60 m2 g\1, adsorbs miscible organic solvents. Coexistence of a high
0.4 mmol of nitrogen monoxide per g-adsorbent at valent cation such as Fe(III) further increases the
303C under 600 mmHg of nitrogen monoxide partial surface area to achieve more effective nitrogen mon-
pressure, whereas -FeOOH supported on activated oxide adsorption.
carbon Rbre (surface area: 870 m2 g\1) adsorbs
4.67 mmol of nitrogen monoxide under the same
condition (a 12 times higher capacity than that with-
Separation of Ethylene (C2H4)
out activated carbon Rbre as a support). This adsor- Ethylene is one of the most important raw materials
bent adsorbs 0.043 mmol per g-adsorbent even under in the petrochemical industry and is also an acceler-
300 p.p.m. of nitrogen monoxide at 1003C, but little ator for ripening fruit. Ethylene is produced on a large
more than 10% of the nitrogen monoxide adsorbed scale by the thermal cracking of naphtha and natural
can be released under vacuum due to its strong gas, and obtained in mixtures with methane, ethane,
adsorption. propane, propylene, hydrogen, carbon dioxide, nitro-
Active carbon-supported FeCl2 completely adsorbs gen and water. In this case, a large quantity of gas
nitrogen monoxide in 25 min from a 6 dm3 NO}N2 mixture has to be treated to obtain pure ethylene. On
mixture (NO: 1000 p.p.m.) when the gas mixture is the other hand, a small amount of ethylene must be
circulated at 1.6 dm3 min\1. The adsorbent releases thoroughly removed from fruit packaging so that
28% adsorbed nitrogen monoxide in 15 min at fruit does not ripen too quickly.
In industry ethylene is puriRed mostly by large scale Separation of Oxygen

distillation, and sometimes by using membrane separ-
ation. For removal of ethylene from fruit packaging, Oxygen, which can be prepared by separation from
no good adsorbent has yet been developed. The solid air, is one of the most important industrial chemical
metal complex adsorbent mentioned here may pro- products and is produced on the largest scale by
vide a new material for both purposes. weight in the world. Thus, separation of oxygen from
Y-type zeolite-supported Cu(I) adsorbs 3.10 mmol air is an important process in industry, and is mainly
of ethylene per g-adsorbent (1.9 mmol of ethylene per performed by low temperature distillation. It is also
mole of Cu(I)) under 250 mmHg of ethylene at 253C important to increase (or decrease) oxygen concentra-
and releases 73% of the adsorbed gas in 60 min. tion for medical use and effective combustion (or
Y-type zeolite-supported Ag(I) adsorbs 3.28 mmol of inhibition of nitrogen monoxide production).
ethylene per g-adsorbent and releases 54% of the gas Haemoglobin and nyoglobin are typical examples
under the same conditions. of metal complexes which can reversibly bind an
MR-type PS resin with primary and secondary oxygen molecule in nature. So, the separation of
amino groups, which is used as a support for CuCl as oxygen from air with solid metal complexes has
described above for carbon monoxide adsorption, received considerable attention in the search for
also works as an ethylene adsorbent. An adsorbent alternative procedures to low temperature distillation.
prepared from 10 g of resin and 15.0 g (152 mmol) of Cobalt(II) Schiff’s base complexes, such as a com-
CuCl in water}acetonitrile (1 : 1) reversibly adsorbs plex of cobalt(II) with bis(salicylidene)diaminoethane
15 mmol of ethylene and 1.9 mmol of ethane, while (Co(salen)), were objects of initial research in the
the resin itself (10 g) adsorbs 5.0 mmol of ethylene 1940s on reversible oxygen absorbents and were later
and 6.5 mmol of ethane at 203C under 1 atm. The developed for on-board oxygen support systems for
coverage of CuCl over the surface of the resin is the US Air Force. Oxygen binds preferentially to
supposed to restrict physical adsorption and increases cobalt by exposing the complex to air at room tem-
chemisorption or coordination. perature to form cobalt(III) superoxide. Heating
MR-type PS-supported AlCuCl4 adsorbs ethylene (thermal swing adsorption: TSA) or reducing pressure
very rapidly, and the equilibrium ratio of the adsor- (pressure swing adsorption: PSA) facilitates release of
bed ethylene to the CuCl added is 1.40 mol- oxygen from the complex. Up to 3000 oxygena-
C H /mol-Cu (89 cm3 STP of ethylene per g-adsor- tion/deoxygenation cycles have been carried out with

bent). The adsorbent releases a part of the adsorbed the same sample. After 3000 cycles the remaining
gas under 8 mmHg at 203C for 10 min, and adsorbs activity was still 50%. Chemical engineering research
0.29 of mol-ethylene per mol-Cu in the second was carried out using the system involving Co(salen)-
and later passes. When an ethylene-adsorbed ad- type adsorbents. The US Air Force studied the
sorbent is cycled at 903C under 1 atm and at cobalt(II) complex of bis(3-Suorosalicylidene)dia-
1423C under 8 mmHg to release ethylene, the adsor- minoethane or Suomine for potential use in providing
bent removes 0.47 and 0.87 of mol-ethylene per breathing oxygen for crews of military aircraft. This
mol-Cu, respectively, in the second adsorption cycle. compound gives the best performance among all oxy-
Although the PS-supported AlCuCl4 adsorbs both gen adsorbents so far with its fast and reversible
ethylene and carbon monoxide, the coordination of binding of oxygen, as well as good stability, but could
ethylene to Cu(I) is stronger than that of carbon not be applied in practice due to the high cost.
monoxide. Thus, the equilibrium molar ratio of ad- Several other transition metals, such as manganese,
sorbed ethylene}carbon monoxide is 5 : 1 when the iron, chromium, nickel, copper, titanium, and so on,
adsorbent is exposed to a 1 : 1 ethylene}carbon have been used to synthesize various kinds of oxygen
monoxide mixture. adsorbents by complexing with various types of
A similar adsorbent, MR-type PS-supported ligands. They show reversible oxygen binding and
AgAlCl4, adsorbs an equimolar amount of ethylene release, but none has achieved commercial success for
per Ag under 1 atm at 203C and releases almost all separation of oxygen from air.
the ethylene under 8 mmHg at 203C. The adsorbent Cobalt(II) Schiff’s base complex has been tried,
does not adsorb carbon monoxide at all under 1 atm Rxed in a polymer matrix. A poly(4-vinylpyridine)-
at 203C, and therefore can be used as a selective attached cobalt(II) Schiff’s base complex can be used
ethylene adsorbent. The adsorbent is water-resistant, for gas chromatography because oxygen comes
although AgAlCl4 is water-sensitive, as is AlCuCl4. out later than the other gases, such as nitrogen.
The stability of the adsorbent is attributable to the Cobalt bis(salicylideneamino)propylamine-attached
location of AgAlCl4 at the hydrophobic sites of the PS polystyrene and polyoctylmethacrylatecobalt dis-
resin surrounded by several aromatic rings. alicylidenethylenediamine Rlms have been found to
be useful for concentrating oxygen. Membranes of However, from a practical viewpoint, there are many
this latter material concentrate 8.3% oxygen in 5 h limitations, for example, lifetime of the materials and
and 13.1% in 42 h from air. The permeability coefRc- the cost of the separation. Since the greatest advant-
ient and the separation factor (O2/N2) under age of systems using solid metal complexes is com-
10 mmHg of feed gas are 10\9 and 15, respectively, plete separation, the adsorbents of solid metal com-
for this membrane containing 12 wt% of the complex. plexes may be used for the purpose of the complete
The experimental results were analysed using the dual- removal of small amount of gaseous molecules from
mode sorption model. Although several interesting closed systems in the future.
studies have been reported on facilitated oxygen trans-
port membranes employing the transition metal}
oxygen complexes, they have a number of inherent Further Reading
limitations from a practical viewpoint, for example, Endo T, Toshima N and Yamamoto T (1998) Chemistry of
the lifetime of membranes, the relationship between Functional Polymeric Materials. Tokyo: Asakura.
quality and quantity of the separation, and so on. Giddings JC (1991) UniTed Separation Science. New York:
John Wiley.
Li GE and Govind R (1994) Separation of oxygen from air
Conclusion using coordination complexes: A review. Ind. Eng.
Metal complex adsorbents used for carbon monox- Chem. Res. 33: 755}783.
ide, nitrogen monoxide, ethylene and oxygen separ- Senoo M, Takagi M, Takeda K et al. (eds) (1993) Hand-
book of Separation Science. Tokyo: Kyoritsu.
ation have been described in this article. Few exam-
Toshima N (ed.) (1992) Polymers for Gas Separation. New
ples of adsorbents for other gases are known, but York: VCH.
theoretically any gas capable of coordinating to metal Toshima N, Kaneko M and Sekine M (1990) Macro-
complexes can be separated in this way. A variety of molecular Complexes. Tokyo: Kyoritsu.
metal complexes can be synthesized by changing Tsuchide E (ed.) (1991) Macromolecular Complexes. Dy-
metal and ligand, and therefore supported metal com- namic Interactions and Electronic Processes. New York:
plexes are promising materials for gas separation. VCH.
III / GENE TYPING : TWO-DIMENSIONAL ELECTROPHORESIS 2939
GENE TYPING: TWO-DIMENSIONAL

ELECTROPHORESIS
N. J. van Orsouw, S. B. McGrath and J. Vijg, principle, in combination with PCR ampliRcation to
Institute for Drug Development, Cancer Therapy and prepare the target sequences. In DGGE, DNA frag-
Research Center, San Antonio, TX, USA ments are subjected to electrophoresis in a polyac-
R. K. Dhanda, Mosaic Technologies, Boston, MA, USA rylamide gel against a gradient of ever higher temper-
C. B. Scott, CBS Scientific Company, Del Mar, CA, USA ature or chemical denaturants (i.e. a mixture of urea
Copyright ^ 2000 Academic Press and formamide). Unlike nucleotide sequencing,
DGGE detects mutations, including base pair substi-
Introduction tutions and small insertions and deletions, on the
basis of differences in the melting temperature of
With the human genome program drawing to a close, the target fragments. A given DNA fragment com-
attention is now rapidly shifting from obtaining con- prises one or more domains, each representing
sensus sequences of all human genes to the detection a stretch of between 50 and 300 base pairs with equal
of individual DNA sequence variations. Based on melting temperature (the temperature at which each
complete sequence information for all human genes, base pair has a 50% probability of being in either the
it is theoretically possible to generate a catalogue of helical or the denatured state). Since the stability of
all gene mutations and polymorphisms in the human each domain depends on its sequence, mutational
genome and test them directly for association to differences among different fragments are
relevant phenotypes, e.g. of health and disease. revealed as migrational differences in the gel
Unfortunately, current methods for detecting DNA (Figure 1A).
sequence variants are not optimized for generating In order to obtain virtually 100% accuracy in
data on multiple genes in large numbers of individuals, mutation detection, fragments to be subjected to
e.g. in population-based studies or in the clinical set- DGGE can be clamped to a GC-rich sequence (a
ting. The most reliable system for comprehensive gene stretch of 30}50 G and C bases). A convenient way of
sequence analysis is still nucleotide sequencing itself, attaching a GC-clamp to the target fragment is by
which is not compatible with cost-effective large making it part of one of the primers in a PCR. With-
scale population-based genetic screening. out GC-clamping, a DNA fragment consisting of one
Recently, various systems have been proposed to melting domain will become completely single-
analyse gene-coding and regulatory sequences more stranded upon denaturation and run off the gel.
effectively for all possible variations. Here we By adding a GC-clamp, a single high-melting domain
review the development and application of one such is artiRcially created at one end of the target frag-
system, two-dimensional gene scanning (TDGS). This ment. As the GC-clamped target fragment migrates
method is based on the two-dimensional separation through the gradient of denaturants, melting of the
of polymerase chain reaction (PCR)-ampliRed gene target domain causes partial branching and halting of
fragments on the basis of both size and base pair the fragment in the gel (Figure 1B). Thus, one func-
sequence in polyacrylamide gels. Attention will be tion of the GC-clamp is to ensure branch formation
focused on most recent developments in automation after melting of the target fragment. However, when
and miniaturization of the two-dimensional elec- the target DNA fragment consists of multiple melting
trophoresis procedure. Future developments towards domains (Figure 1C), only mutations in the lowest
a dedicated fully automated high-throughput system melting domain are readily detected. To facilitate
for gene analysis will be discussed. detection of all possible mutations, it is imperative
that the target fragment represents only one melting
domain. Fortunately, since the addition of a GC-
Two-Dimensional Gene Scanning: clamp allows for stacking interactions with neigh-
Background and Principles bouring bases, the entire fragment will often behave
as one melting domain (Figure 1C). However, this is
Denaturing Gradient Gel Electrophoresis
not always the case, and in practice, the target frag-
TDGS is based on denaturing gradient gel elec- ment needs to be designed, e.g. through the strategic
trophoresis (DGGE) as the mutation detection positioning of PCR primers to achieve the ideal single
2940 III / GENE TYPING : TWO-DIMENSIONAL ELECTROPHORESIS
Figure 1 Principles of denaturing gradient gel electrophoresis. (A) Single base changes affect the melting temperature of a fragment,
which results in a gel shift. (B) After complete denaturation, single-stranded fragments will run off the gel; the addition of a GC-clamp to
the target fragment prevents complete denaturation and therefore fragments will be retarded in the gel. (C) The addition of a GC-clamp
to a multiple-domain fragment can make the fragment behave as a single-domain fragment. Continuous line: target fragment without
a GC-clamp. Dashed line: target fragment, including a 40 bp GC-clamp. (D) The introduction of a heteroduplex cycle at the end of PCR
amplification of target fragments facilitates detection of heterozygous mutations as four molecules: two homoduplexes and two
(early-melting) heteroduplexes.
melting domain. In general, target fragments in mutation is revealed as four different double-
DGGE have an average size of 275 bp, including stranded fragments: two homoduplex molecules (one
a GC-clamp and PCR primer sequences. wild-type homoduplex and one mutant homoduplex)
The sensitivity of DGGE for detecting variants is and two heteroduplex molecules (each comprising
further enhanced by the introduction of a heterodup- one wild-type and one mutant strand). Since the stab-
lexing step using one round of denaturation/renatura- ility of heteroduplexes is so much lower, they always
tion, usually at the end of PCR ampliRcation of the melt earlier than the homoduplex molecules (Fig-
target fragment. In this manner, a heterozygous ure 1D).
Figure 1 Continued
Although DGGE has the crucial advantage of hav- analysis system offers only a limited advantage.
ing virtually 100% sensitivity in detecting mutations, Multiplex PCR systems for genes and genetic markers
it has typically been applied in a serial fashion, e.g. on are now becoming available and it has been demon-
a fragment-by-fragment basis. For analysing large strated that as many as 26 fragments can be co-
genes or multiple genes this is not practical. A solu- ampliRed in one single tube under the same reaction
tion for this problem, which we adopted, is to apply conditions.
the DGGE principle in the format as it was originally
described, i.e. a two-dimensional system of separ-
Two-Dimensional DNA Electrophoresis
ation by size followed by DGGE. Successful imple-
mentation of such a two-dimensional DNA elec- The major advantage of two-dimensional elec-
trophoresis system in mutation scanning of large trophoresis is that it provides a high resolution system
genes requires an efRcient multiplex PCR proto- to screen multiple fragments under the same condi-
col. Indeed, without the possibility to PCR-amplify tions. It has been demonstrated that DGGE provides
multiple target fragments (i.e. typically 10 or more) virtually 100% mutation detection sensitivity even
in one single reaction, the application of a parallel when applied with a broad range gradient of
denaturants. This opens up the possibility to analyse an entire gene for a particular DNA sample and
multiple fragments for all possible mutations under immediately recognize each exon and variants there-
the same set of experimental conditions. The total in. This has been demonstrated for several large hu-
number of target fragments that can be analysed man disease genes, including CFTR, RB1, MLH1,
simultaneously depends on the resolution of the gel TP53, TSC1, BRCA1, as well as for a part of the
system used. Although high resolution can be ob- mitochondrial genome.
tained by using one-dimensional denaturing gradient
gels, two-dimensional separation allows characteriza-
tion of each fragment on the basis of two independent Design Software and Instrumentation
criteria, size and melting temperature. In practice, for TDGS Tests
a fragment mixture corresponding to all exons of
Computer-Automated Design of Target Fragments
a gene is electrophoresed in a nondenaturing size gel.
for PCR and DGGE
Fragments are further sorted out in a denaturing
gradient gel as the second dimension (Figure 2). By A potential hindrance to the widespread application
using the two-dimensional system, it is possible to of TDGS to multiple novel genes involves the dif-
visualize completely all fragments corresponding to Rculties in the design of PCR primers generating
Figure 2 Schematic depiction of a TDGS test. All exons are amplified in an extensive multiplex reaction, and the fragments are
resolved by size separation, followed by separation in a gradient of denaturants. Heterozygous mutations would show up as four spots
instead of one.
single-domain fragments which can be resolved under sion (separation according to size) and the subsequent
one set of electrophoretic conditions. To design com- second-dimension separation of these fragments by
plete gene tests for mutational analysis by TDGS, an DGGE on the basis of their melting temperature. The
automated generally applicable computer program Rrst-dimension separation was carried out in poly-
was developed, which was based on a commercially acrylamide slab gels, which required staining of the
available primer design program (Primer Designer 3; gel to visualize the one-dimension separation pattern
ScientiRc and Educational Software, State Line, PA), before this could be excised and transferred to
the melting routine MELT87 and a newly generated the second-dimension denaturing gradient gel.
spot distribution routine. After entering a gene’s cod- Alternatively, tube gels have been employed for size
ing sequence as exons with their Sanking intronic separation, which obviated the need for gel staining
sequences, a rank of suitable PCR primers for each and lane excision. However, routine application of
exon is designed by the PCR design subroutine. Next, TDGS requires standardization and automation,
the best primer pair is used in the melt subroutine to which is incompatible with the labour-intensive step
check for a one-domain target fragment. The program of manual interference between the Rrst- and second-
uses different GC-clamps at either the 5 or 3 end dimension separation.
of the target fragment and, if necessary, additional Recently, we developed a simple automated two-
small GC- or AT-clamps at either side of the target dimension instrument, which is based on an existing
fragment. If it is impossible to design a one-domain vertical electrophoresis system with an isolated hori-
fragment, the next optimal primer pair is tested, and zontal unit on top (Figure 3). This top unit consists of
so on. If a primer pair suitable to create a one-domain two outer chambers and one middle chamber. The
fragment cannot be found, the exon is split. necessary contacts between the outer buffer
As soon as primers fulRl PCR and melting criteria, chambers and the gel are provided by two strategi-
the fragment is positioned according to its size (x) and cally located openings in the inner glass plate.
melting (y) coordinates. The spot distribution routine In this system only one gel (a denaturing gradient
then checks for possible overlap. The output Rle of gel) is used with the top part nondenaturing. This
the program is a complete list of primers to be used in nondenaturing part functions as the lane for the Rrst-
TDGS. (Questions regarding the use of the TDGS dimension size separation. A slot former is placed in
software should be directed to Accelerated Genomics, the top left part (Figure 4). In the current conRgura-
Concord, NH, http://www.accelerated genomics.com tion, a gel is attached to each side of a gel holder,
Tel.: (210) 616-5910; fax: (210) 692-7502.) which can be placed in a buffer tank. Buffer
tanks can hold multiple units so that multiple gels can
Electrophoresis
be run simultaneously (Figure 5). The sample is elec-
For two-dimensional DNA electrophoresis, originally trophoresed on the basis of size horizontally in the
two different gels were used for the Rrst-dimen- nondenaturing top gel, and the second-dimension
Figure 3 Two automatic dual-gel TDGS systems.

Figure 4 Schematic depiction of the automated two-dimensional electrophoresis unit. Buffer chambers for the first-dimension
separation (A) are connected with the gel through openings in the (inner) glass plate (B). During the first-dimension electrophoresis, the
middle buffer chamber (C) is isolated from the outer chambers. For the second-dimension run, buffer chamber C is flooded with buffer
and the upper electrode is turned on in conjunction with a positive electrode in the lower reservoir (not shown in this figure). The gel
cassette is sealed to the top unit with a serpentine silicone gasket (D), and sample is loaded in the single slot (E). The dashed line
indicates the beginning of the gradient of urea/formamide.
electrophoresis is carried out vertically in Miniaturization

the denaturing gradient gel. All components of the
Miniaturization of gene analysis systems, such as the
automatic TDGS electrophoresis system are depicted
TDGS system described here, offers two major
in Figure 6.
advantages: increased speed and lower cost.
Gradient gels can be poured, up to nine at a time,
using a simple linear gradient maker in combination
with a multiple gel caster. The exact gradient that is Speed The duration of electrophoresis depends on
to be applied is dependent on the GC-content of the the voltage applied. For example, the optimal elec-
gene(s) of interest and is determined by the TDGS trophoresis conditions for the retinoblastoma suscep-
primer designer software. tibility gene RB1 using standard 1.0 mm thick gels,
Figure 5 The entire automatic TDGS system. In this version of the system, four gels can be run simultaneously submerged in a buffer
tank, which is equipped with a heater/stirrer to provide for a constant temperature. For more information, see http://www.cbssci.com
are 100 V, 5 h for the size separation and 100 V, 16 h for the occurrence of variations (in the form of four
for the second-dimension separation. Increasing the spot patterns; see Figure 1D). An example of a TDGS
voltage increases the heat production, which nega- pattern is shown in Figure 7, depicting the RB1 gene,
tively affects gel resolution. An obvious strategy containing a mutation in exon 2.
is the use of thinner gels, which facilitate rapid heat However, for large scale application of TDGS, dye
dissipation into the surrounding buffer and primer technology for the in-gel detection of two-
thereby allow increasing the voltage while maintain- dimensional spot patterns is an obvious strategy. Test
ing a good resolution. Currently, gels as thin as results indicate similar two-dimensional patterns and
0.35 mm are now run at 500 V, 0.8 h for the size sensitivity for Suorescein-labelled primers compared
separation and 500 V, 3.5 h for the second-dimension to Sybr-green-stained gels.
separation. Introduction of Suorescent detection offers
two advantages over gel staining. First, the reduction
Cost The cost factor is of major importance for the in labour is considerable and loss of gels due to
large scale implementation of genetic testing. Since breakage is prevented. Second, since there is no need
this is determined to a major extent by reagent and to release the gel from between the glass plates it has
material cost, as well as space, miniaturization of become possible to use thinner gels, which will allow
analytical systems is of crucial importance. Miniatur- shorter electrophoresis times (see above). To increase
ization of TDGS results in thinner and smaller gels, the efRciency even further it is possible to label
which require less sample (smaller PCR volumes can different samples with different Suorophores.
now be applied) and lower gel and buffer Current Suorescence imagers have the option to ana-
amounts. Moreover, they take up less space. Instead lyse multiples Suorophores in the same gel.
of the current 17;22 cm format, two-dimensional
patterns have already been produced on 10;10 cm
mini-gels, and it is not unreasonable to expect that Future Developments
ultimately electrophoresis will be carried out on glass Routine application of TDGS requires standardiz-
slides. ation and further streamlining of the procedure. Ulti-
mately, one could envisage a fully automated system
Detection of TDGS Patterns
of PCR ampliRcation, sample loading, electrophor-
After electrophoresis, the two-dimensional DNA esis, scanning of gels by a Suorescent imager, fol-
fragment patterns can be visualized by incubating the lowed by online interpretation of gels by image analy-
gels with DNA staining dyes. Examples are ethidium sis systems.
bromide or the more sensitive dye Sybr-green. Pat- Much of the labour that is involved in PCR ampliR-
terns are photographed under UV light and evaluated cation, as well as the error rate, can be greatly
Figure 6 Line drawing of all the components of a 4-gel automatic TDGS system.
diminished by PCR robotics. Such instruments have tual interpretation of spot patterns is currently most
now become widely available and, in combination conveniently done by eye, automated image analysis
with an ongoing effort to increase multiplex software is commercially available. The use of such
groups, are expected to increase greatly the front-end software may, for example, facilitate the detection of
throughput of genetic testing. Multiple two-dimen- subtle positional changes in the context of other spot
sional gels can be stacked for simultaneous elec- variations. In this respect, one could envisage a pro-
trophoresis of manifold samples. A simple robot arm gramme with information on all possible spot
could load the gel sandwiches into the Suorescent positional variants to identify quickly recurrent muta-
imager for quick gel scanning. Finally, while the actions and polymorphisms on the basis of their unique
Figure 7 Empirical TDGS pattern of the retinoblastoma susceptibility gene RB1, containing a mutation in exon 2.
conRguration. Such software should also be capable Eng C and Vijg J (1997) Genetic testing: the problems and
of storing two-dimensional patterns and link subsets the promise. Nature Biotechnology 15: 422}426.
of them in particular experiments requiring compari- Fischer SG and Lerman LS (1979) Length-independent sep-
sons of large numbers of individuals. It could also aration of DNA restriction fragments in two-dimen-
provide for a sample tracking system. sional gel electrophoresis. Cell 16: 191}200.
Lerman LS and Silverstein K (1987) Computational simula-
tion of DNA melting and its application to denaturing
Further Reading gradient gel electrophoresis. Methods in Enzymology
155: 501}527.
Cotton RGH (1997) Mutation Detection. Oxford: Oxford ShefReld VC, Cox DR, Lerman LS and Myers RM
University Press. (1989) Attachment of a 40-base pair G#C rich se-
Dhanda RK, Smith WM, Scott CB et al. (1998) A simple quence (GC-clamp) to genomic DNA fragments by the
system for automated two-dimensional electrophoresis: polymerase chain reaction results in improved detection
applications to genetic testing. Genetic Testing 2: of single-base changes. Proceedings of the National
67}70. Academy of Science of the USA 86: 232}236.
2948 III / GEOCHEMICAL ANALYSIS: GAS CHROMATOGRAPHY AND GC-MS
van Orsouw NJ, Li D, van der Vlies P et al. (1996) Muta- electrophoretic gene mutational scanning tests. Nucleic
tional scanning of large genes by extensive PCR multi- Acids Research 10: 2398}2406.
plexing and two-dimensional electrophoresis: applica- Vijg J and van Orsouw NJ (1999) Two-dimensional gene
tion to the RB1 gene. Human Molecular Genetics 5: scanning: exploring human genetic variability. Electro-
755}761. phoresis 20: 1239}1249.
van Orsouw NJ, Dhanda RK, Rines DR et al. (1998) Rapid
design of denaturing gradient-based two-dimensional
GEOCHEMICAL ANALYSIS:
GAS CHROMATOGRAPHY AND GC-MS
R. P. Philp, University of Oklahoma, Norman, OK, onmental studies one is more concerned with cor-
USA relating a spilled product with its original source
Copyright ^ 2000 Academic Press material or trying to evaluate the extent of removal
during clean-up procedures.
Geochemical samples are extremely complex mix-
Introduction tures of a wide variety of compound classes. The
analytical techniques commonly used to characterize
Geochemical analysis, and more speciRcally such mixtures involve some form of chromatography,
chromatography, is concerned with samples derived such as gas chromatography (GC), gas chromatogra-
from two different sources: those of relatively phy}mass spectrometry (GC-MS), gas chromatogra-
recent origin, related to environmental problems; and phy}mass spectrometry/mass spectrometry (GC-
those of a much greater geological age, related to MS/MS), and more recently gas chromatography}
fossil fuel exploration and exploitation. The isotope ratio mass spectrometry (GC-IRMS). Liquid
chromatographic techniques utilized to analyse and chromatography (LC) and combined liquid
characterize such samples are virtually identical re- chromatography}mass spectrometry (LC-MS) are
gardless of the age and origin of the sample. The also used in certain applications, but not to the same
extracts from geochemical samples, whether they are extent as GC and GC-MS. In addition to the analyti-
rocks, soils, crude oil spills, contaminated wildlife or cal chromatographic separations, most geochemical
spills of reRned products, are very complex mixtures analyses require some sort of fractionation into com-
of a wide variety of organic compounds. Compounds pounds classes prior to the actual analysis. There are
derived from fossil fuels typically will be complex certain cases where total sediment extracts or whole
mixtures of hydrocarbons, and the environmental crude oils are analysed directly but generally the mix-
samples from more recent sediments probably will tures are so complex that an initial fractionation(s) is
contain a variety of other compounds such as chlorin- required to simplify the extracts for subsequent ana-
ated compounds, pesticides or herbicides. In view of lyses. For example gas chromatograms of many crude
the similarities of the techniques used for analysing oils (Figure 1) are dominated by n-alkanes but, for
the samples from these different sources, the ma- the most part, compounds that are of much greater
jority of examples used in this article to illustrate the geochemical importance are not readily observable in
techniques will be based on the characterization of these chromatograms but are hidden in the baseline
fossil fuel samples. of the chromatogram. It should be noted that there
The major goal of any geochemical analysis is to are also many naphthenic crudes not dominated by
take a sample and, through a variety of fractionations n-alkanes, e.g. Venezuelan and Russian crudes. Most
and analytical techniques, reach a point where either of these naphthenic crudes are either severely biode-
the presence or absence of speciRc target compounds graded or have been generated at relatively low levels
can be determined, or Rngerprints for speciRc classes of maturity from sulfur-rich kerogens. A fractiona-
of compounds can be obtained and used for correlation step involving thin-layer chromatography, col-
tion purposes. Applications related to petroleum umn chromatography or liquid chromatography, all
exploration might use such Rngerprints for oil}source of which involve partitioning of components between
rock or oil}oil correlation studies, whereas in envir- a liquid and solid phase, leads to the separation of
III / GEOCHEMICAL ANALYSIS: GAS CHROMATOGRAPHY AND GC-MS 2949
Figure 1 Gas chromatograms of crude oils, rock extracts, or

refined petroleum products are typically dominated by n-alkanes
and isoprenoids. While GC alone does not permit their identifica-
tion, the fact that the isoprenoids pristane and phytane have very
similar elution times to the C17 and C18 n-alkanes, respectively,
generally make it relatively easy to identify the other members of Figure 2 The whole oil chromatogram shown in Figure 1 does
the homologous series with a reasonably high degree of confi- not give a true impression of the complexity of the mixture of
dence. compounds in a crude oil. While the n-alkanes are the dominant
components in the chromatogram, a vast array of branched,
cyclic, aromatic and polar compounds are also present. This
compounds on the basis of factors such as polarity, figure shows the chromatograms for a saturate and aromatic
shape and size. For hydrocarbon-containing samples, fraction separated from a crude oil by thin-layer chromatography.
the fractionation step typically involves separation
into three fractions } saturate and aromatic hydrocar-
bons and a polar fraction containing nitrogen, sulfur compounds can be further concentrated by such pro-
and oxygen compounds. In general it is the saturate cesses as molecular sieving or urea adduction, both of
and aromatic fractions that receive most attention in which will separate the n-alkanes from the branched
terms of additional analyses. Although the nitrogen, and cyclic alkanes as shown in Figure 3. At this point
sulfur and oxygen fractions contain many com- fractionation and sieving of the original extract or
pounds that possess useful information, the complex- crude oil will have produced a fraction that is readily
ity of these fractions has precluded their detailed
analyses. It is not proposed to go into the experi-
mental details of such chromatographic fractiona-
tions since these are very basic techniques and de-
scriptions of speciRc methods for particular classes of
compounds are readily available in the literature.
Fractionation of the crude oil used for Figure 1 into
various fractions produces saturate and aromatic
fractions as shown in Figure 2. It should be noted
when comparing Figures 1 and 2 that the result of the
fractionation and evaporation of the solvents used in
the fractionation process will lead to the loss of some
of the more volatile compounds in the C1}C15 range
of the saturate and aromatic fractions. GC analyses
of the saturate fraction produces a chromatogram
dominated by n-alkanes, typically in the carbon num-
ber range from C15 to around C40 when the analyses
Figure 3 The saturate fraction shown in Figure 2 is again dom-
are performed using conventional GC. The iso- inated by n-alkanes, which tend to mask the presence of a very
prenoids pristane (Pr) and phytane (Ph) can also be complex mixture of branched and cyclic compounds also present
clearly discerned on these chromatograms. Once in this fraction. The n-alkanes can be separated from these
again, it should be emphasized that although the branched and cyclic compounds by processes such as molecular
sieving or urea adduction to produce the branched and cyclic
n-alkanes are the predominant compounds in the
fraction shown in the bottom chromatogram of this figure. The top
chromatograms, there are many more minor com- chromatogram (A), shown for comparison purposes, is the total
pounds in the fraction that generally provide a great saturate fraction from which the branched and cyclic compounds
deal more information than the n-alkanes. These were isolated.
amenable to analysis by the techniques mentioned injectors have improved and temperature control of
above, such as GC, GC-MS, GC-MS/MS or GC- the ovens has improved, but probably the greatest
IRMS. In the following sections a brief description of advances have occurred in the Reld of column tech-
each technique and typical applications will be given. nology. Early columns were short, large-diameter
Again it should be reiterated that for the most part packed columns made of stainless steel or copper.
this article uses hydrocarbons for illustration pur- Over the years narrow-bore capillary columns were
poses, but the majority of the techniques are equally developed, initially made of stainless steel, then glass
applicable to the analysis of other samples of geo- and more recently fused silica. At the same time as the
chemical interest such as environmental samples, pos- evolution of capillary columns, the variety of liquid
sibly with some slight modiRcations in the operating phases for different applications has also greatly
conditions. improved. The most recent advance, and one that is
quite signiRcant for petroleum exploration and pro-
duction, has been the development of high temper-
Gas Chromatography ature GC phases. Traditionally most crude oil ana-
As can be seen from the chromatograms used in the lyses have characterized hydrocarbons in the C1}C40
Introduction, GC provides a great deal of informa- range but the advent of high temperature GC
tion on the composition of geochemical samples. (HTGC) signiRcantly changed the way in which we
With the inclusion of an internal standard, this in- look at hydrocarbon distributions of crude oils. The
formation can be both quantitative and qualitative in use of HTGC has demonstrated that many crude oils
nature. However, it is important to remember that, contain a wide range of hydrocarbons signiRcantly
for the most part, GC does not provide any informa- above C40, extending to as far as C100 and possibly
tion on the identiRcation of individual components. higher. This in turn has also led to changes in one of
In the saturate hydrocarbon fractions, individual the very basic premises of geochemistry } that oils
compounds such as n-alkanes in the chromatograms with a high wax content were thought to be derived
can be readily recognized and their identiRcation con- only from higher plant sources. It is now clear that
Rrmed either by the use of co-injected standards or such oils may also be derived from lacustrine and
analysis of the sample by GC-MS as described below. marine source rocks as a result of analysing a number
In other cases where it may be necessary to detect of samples from such source rocks using HTGC.
certain classes of chlorinated compounds, or or- Figure 4 provides an excellent illustration of the addi-
ganosulfur compounds, additional information on tional information obtained from the use of HTGC.
the presence or absence of these compounds can be The upper chromatogram shows the analysis of an
obtained by using detectors that are speciRc for these ozocerite extract by conventional GC and the bottom
classes of compounds, such as electron capture, Same chromatogram shows the same sample analysed by
photometric or Hall detectors. It should also be noted HTGC. Clearly the distribution of hydrocarbons is
that a number of recent studies have shown that quite different in the lower chromatogram. The
atomic emission detection (AED) is a particularly signiRcance of this is related to the fact that the
sensitive and speciRc method of detection for sulfur- greater the concentration of the higher molecular
containing compounds. Although the Same photo- weight alkanes, the greater the production problems
metric detector (FPD) is probably the most widely associated with oils that contain such compounds. In
used detector for sulfur-containing compounds, it has other words, if the oils were only characterized by
some drawbacks including nonlinear compound-de- conventional GC, high molecular weight hydrocar-
pendent response, and the quenching of sulfur signals bons would remain undetected. Once a production
by co-eluting hydrocarbons. The AED has a linear programme was initiated it would not be long before
response and is compound-independent, permitting the wellhead facilities and pipelines would become
easy calibration using any sulfur-containing com- blocked with parafRn deposits, which require
pounds. This feature is unique because other de- costly measures to remove. While HTGC analyses do
tectors require the construction of curves using target not eliminate the problem, production engineers
analytes as standards, which becomes a very time- would be aware of the potential for such a problem
consuming exercise. and steps could be introduced to minimize its occur-
GC has been utilized widely in geochemistry since rence.
the 1950s. As column technology and instrumenta- With the availability of HTGC an increasing num-
tion have improved, so has the quality of the analyti- ber of samples have been analysed using this ap-
cal data. There are many similarities between the gas proach and steps have been taken to develop methods
chromatographs of today and the systems that were that will quantitatively separate high molecular
developed in the 1950s and 1960s. Detectors and weight alkanes from the asphaltene fraction. Analyses
could readily convert these compounds to the corre-

sponding hydrocarbon and could easily represent
a viable source for such hydrocarbons.
Another area of geochemistry that has become par-
ticularly important in the past few years is reservoir
geochemistry, and GC has played an extremely im-
portant role in its development. Oil and gas reservoirs
are very complex geological features with many com-
partments. A knowledge of the relative position of
these compartments is extremely important for reser-
voir management, determination of where additional
production wells should be drilled, and evaluating
how a speciRc reservoir may have been Rlled. There
are a number of ways in which the reservoir compart-
ments may be delineated but one particularly interest-
ing, innovative and relatively cheap method involves
the utilization of high resolution GC. As indicated
above, crude oils are very complex mixtures of hydro-
carbons, but when the chromatograms are expanded
the complexity of the mixtures becomes far more
apparent and the presence of a large number of minor
Figure 4 One of the most significant recent developments with-
in GC has been the development of high temperature phases for
components is clearly visible. Reservoir geochemistry
the columns. Before this development it was generally only pos- utilizes these minor components to assist in the delin-
sible to analyse compounds with up to approximately 40 carbon eation of the reservoir compartments. In brief it is Rrst
atoms. The newer HTGC columns permit samples containing up necessary to determine whether all of the oils in the
to approximately 120 carbon atoms to be analysed. This figure reservoir are derived from the same source materials.
illustrates the comparison between the analyses of the hydrocar-
bons in the fossil bitumen ozocerite by (A) conventional GC and
Once this has been established, all of the oils need to
(B) HTGC. The difference in distributions is very clear and also be analysed by high resolution GC, the early eluting
demonstrates how the composition of a fraction obtained by region of the chromatogram expanded and a number
conventional GC may not necessarily reflect the true composition of pairs of minor peaks selected as shown in Figure 6.
of the sample being investigated. Ratios based on pairs of selected peaks are measured
and subsequently plotted on a star or polar diagram.
of a wide range of such samples have shown that, in This process is then repeated for all the oils to be
addition to the n-alkanes, there is also a wide range of examined from the reservoir. It is important to ensure
additional compounds in the higher molecular weight that the same pairs of peaks are selected for each oil,
fraction including branched alkanes and alkylcy- even if the identity of these peaks in unknown. Since
clohexanes. The distribution patterns for these com- the differences between the pairs of peaks in
pounds has provided an additional powerful tool for individual samples are often quite small it is extreme-
determination of the type of environment from which ly important to ensure that the GC analyses are highly
a sample has been generated. For example Figure 5A reproducible for this particular application. How-
and B show the distributions of high molecular ever, once all of the oils have been analysed and the
weight fractions from oils whose source materials data plotted on the star diagram, it will be found that
were known to be deposited in lacustrine and marine oils that are in the same compartment or in commun-
depositional environments, respectively. Note the dif- ication will appear virtually on top of each other,
ference in the distributions of these monocyclic whereas those oils in different compartments will
hydrocarbons, which are characteristic of the dif- be slightly separated (Figure 7). These small dif-
ferent environments. At present relatively little inference may result from slight differences in
formation is available concerning the origin of these oil}rock interactions, slight maturity differences
compounds, although their widespread distribution or generation from slightly different sources.
suggests that they are probably related to an al-
gal/phytoplanktonic source, possibly with additional Gas Chromatography^
contributions from higher plant waxes. It is known
that many marine organisms contain abundant
Mass Spectrometry
quantities of higher molecular weight esters, alcohols While GC can provide a great deal of information
and fatty acids. Relatively simple transformations that is of interest and useful from a geochemical
Figure 5 The availability of HTGC has led to the discovery of numerous new series of compounds present in oils and source rocks
above C40. Several of these series, and in particular a series of alkylcyclohexanes, have been shown to be useful in discriminating
between oils derived from source materials deposited in lacustrine (A) versus marine settings (B). More subtle variations in these
distributions allow the salinity levels of the depositional environments to be distinguished.
perspective, it should also be noted that in most cases amounts of useful information for both an environ-
GC only provides information on the distribution of mental and exploration context.
the major components in the sample, and for the most To illustrate the utility of the biomaker Rnger-
part these are generally dominated by the n-alkanes. prints, the gas chromatograms of three oils are shown
The more useful compounds are the more complex in Figure 8. From the gas chromatograms alone it is
molecules, or biomarkers, which are typically present virtually impossible to determine what relationship, if
in relatively low concentrations and which require the any, exists between these samples. In other words, are
use of GC-MS and more speciRcally single ion or they from the same source rock or can they be corre-
multiple ion detection (MID) in order to determine lated with each other? The effects of biodegrada-
their distributions. While there are many classes of tion are clearly evident in sample B since all of the
biomarkers that are commonly used for correlation n-alkanes have been removed, making it appear even
and other purposes, compounds such as the steranes more signiRcantly different from the other two
and terpanes will typically provide the greatest samples. Detailed analyses of the same samples by
Figure 6 Reservoir geochemistry has provided an important means of determining continuity and communication within reservoir
compartments. Once it has been established that the oil in a reservoir is from a common source, high resolution gas chromatograms are
obtained for individual samples and ratios of various pairs of peaks are determined, as shown in the figure. The identity of the
components does not have to be known; the important point is that the same pairs of peaks are used for all the samples examined in
any particular study. Although the early studies typically used peaks in the early part of the chromatogram, it has been shown that the
minor components in the higher regions of the chromatogram can also be used for the same purpose.
GC-MS and MID using the characteristic ions for the speciRc for a variety of applications, in addition to
sterane and terpane biomarkers at m/z 217 and 191, this type of correlation. The presence of individual
respectively, produces the additional data shown in compounds, for example oleanane and gam-
Figure 9. On the basis of the chromatograms shown macerane, can provide information on the presence of
in Figure 9, samples B and C are in all probability speciRc types of source materials or the nature of the
related to each other. It is not necessary to identify depositional environment.
each component, rather one should think of the mass To illustrate this type of application, Figure 10
chromatograms as Rngerprints. If two samples are shows the m/z 191 and 217 mass chromatograms of
derived from the same source, then their Rngerprints an oil that is derived from source material dominated
should be the same, or at least very similar; samples by higher plant or terrestrial source material. This
from different sources will be different from evidence is contained in the fact that the predominant
each other. Hence when the Rngerprint for sample component in the m/z 191 mass chromatogram is the
A is compared with those for B and C in Figure 9, terpane called 18(H)-oleanane. It has been estab-
there are a number of signiRcant differences be- lished that this compound has its precursor in higher
tween these samples that permit one to conclude that plant material and hence the presence of this com-
A is from a different source than B or C. The pound in an oil will indicate that the sample is derived
biomarker Rngerprints obtained in this way are very from such material. In support of such evidence is the
manner, pieces of evidence can be put together that in

many cases will provide a very clear indication as to
the origin of the material being analysed.
In the second example, shown in Figure 11, the
presence of another very speciRc compound, gam-
macerane, can also be very clearly seen in the m/z 191
chromatogram. This compound is a very speciRc indi-
cator of depositional environments of enhanced salin-
ity. Recent attempts have been made to relate the
presence of certain compounds, for example, dino-
sterane to the age of the source rock from which the
sample was generated. SpeciRc ratios of different
sterane isomers or terpane isomers are also used ex-
tensively for determining the relative maturity of oils
or source rocks.
The sterane and methylsterane distributions in
crude oils are far more complex than the terpanes and
no matter how good the GC resolution, it is imposs-
ible to obtain complete separation of all co-eluting
Figure 7 The ratios of the pairs of peaks measured in Figure 6 isomers, epimers and homologues (Figure 12). In or-
are plotted on a star or polar diagram. If two oils are in the same
der to optimize this separation it is necessary to utilize
compartment, or in communication with each other, then on such
a star plot the two oils will have identical plots (i.e. B and C). If they GC combined with tandem mass spectrometry, or
are not in communication with each other, then their plots will MS/MS, which provides an additional degree of
show some subtle differences (i.e. oil A). separation based on the utilization of the MS/MS
capability.
fact that the sterane chromatogram is dominated by
the C29 steranes. For oils of this nature it has been
clearly established that the C29 steranes are also asso- Gas Chromatography^Mass
ciated with higher plant source materials. In this
Spectrometry/Mass Spectrometry
To demonstrate the utility of the GC-MS/MS ap-
proach to the characterization and determination of
biomarkers in geochemical samples, the resolution of
a complex mixture of sterane isomers and homo-
logues will be described. While this example utilizes
the steranes, it should be borne in mind that the same
approach can be used to resolve any very complex
mixture of organic compounds from geochemical
samples.
The mixture of steranes commonly analysed by
MS/MS is in the C27}C30 carbon number range and
each homologue has a molecular mass at m/z 372,
386, 400 and 414, respectively. For each of the
steranes the parent ions will undergo a collision-
activated decomposition to produce a daughter ion at
m/z 217. Hence a series of MS/MS parent}daugh-
ter experiments are performed utilizing these par-
ent}daughter transitions in combination with the GC
Figure 8 Gas chromatograms of three different oils provides separation. The GC-MS analysis and single ion
limited information concerning the relationship between the sam- monitoring of m/z 217 produces the mass chromato-
ples. In this figure oil B clearly appears to be quite different from gram shown in Figure 12 but with the GC-MS/MS
oils A and C, but all that can really be said is that oil B has been
analyses, the results shown in Figure 13 are obtained.
heavily biodegraded and the n-alkanes have been removed. It is
impossible to say what, if any, similarities there may have been It can be seen in Figure 13 that by using the C27 par-
between the alkane distribution for this sample and the other two ent}daughter ion pair at m/z 372/217, respectively,
samples. the result of analysing the sample by MS/MS is to
Figure 9 The terpane and sterane distributions for the same three oils as used in Figure 8 provide a more specific indication of the
relationship between different samples. It is not necessary to know the identification of all the individual peaks but rather to think of the
total chromatogram as a fingerprint. In this case it can be seen that oil A has quite a different set of fingerprints than oils B and C,
suggesting it is not related to these other two samples. While the fingerprints of oils B and C may show some differences, these are
actually due to small maturity differences in the samples.
totally resolve the C27 components from the rest of some type of degradation step prior to analysis. At
the complex mixture. present for geochemical purposes this degradation
Similar results would be obtained if the par- step typically consists of some type of pyrolysis
ent}daughter pairs for the other members of the series reaction with the pyrolyser interfaced to the gas
were also illustrated. A similar approach could be chromatograph or GC-MS system. There are
applied to the methylsterane mixture using the parent also reports of the use of various NMR techniques
ions and the daughter ion at mass 231 and a similar to characterize this insoluble fraction, although
simpliRcation of the mixture would be obtained. In this is a little less speciRc than the pyrolysis
this particular application the MS/MS serves to intro- approach.
duce an additional element of separation following An example of the pyrolysis of the insoluble frac-
the initial separation by GC. tion of an organic-rich source rock in shown in Fig-
ure 14. This was produced by pyrolysing the sample
at a temperature of 6003C for a short period of time
Pyrolysis^Gas Chromatography^ and allowing the pyrolysis products to be transferred
directly to the GC column. (There are of course a
Mass Spectrometry wide variety of pyrolysis conditions that could be
While a large proportion of the geochemical samples used, but those cited here give a general idea of the
analysed are soluble in organic solvents and readily typical conditions used.) The products of a sample
amenable to direct analysis by GC or GC-MS, there is pyrolysed in this manner produce a chromatogram
another aspect to geochemical samples that is often dominated by alkane/alkene doublets plus a wide
overlooked. Samples of geochemical interest such as variety of minor components. From these distribu-
soils, source rocks or coals also have a signiRcant tions it is often possible to gain information about the
insoluble organic component such as the humic frac- nature of the source materials originally responsible
tion of soils or the kerogen fraction of a source rock. for the formation of the kerogen plus the type of
Characterization of these insoluble fractions requries products it will subsequently produce if buried to
Figure 12 Sterane distributions in crude oils as typically deter-

Figure 10 GC-MS analysis of crude oils reveals complex fin-
mined by single ion monitoring of m/z 217 reflect the complex
gerprints of biomarkers. In many cases these compounds may be
nature of the mixture of these components in a crude oil. It is
very specific indicators of particular types of source materials
virtually impossible to separate all of the co-eluting isomers,
responsible for sourcing the oil. In this example the predominant
epimers and homologues by GC-MS, no matter how good the
component in the terpane chromatogram is 18(H)-oleanane. Not
chromatography.
only is this compound very specific in terms of being derived from
higher plant source materials, but it is also more specifically
related to the flowering plants or angiosperms that have only
evolved since the Late Cretaceous}Early Tertiary periods. The
presence of this compound can therefore be used to constrain the by pentane precipitation and pyrolysed. In this way it
age of the source rock from which the oil was generated. will be observed that the n-alkane/alkene Rngerprints
generated from the degraded and nondegraded
samples will be virtually identical if the samples
are derived from the same source, but quite dif-
appropriate depths and subjected to thermal degrada- ferent if the samples are unrelated.
tion.
Another useful application is the pyrolysis of as-
phaltenes, particularly those isolated from bio- Gas Chromatography^High Resolution
degraded crude oil samples. It is often difRcult to
determine the origin of a biodegraded oil sample.
Mass Spectrometry
However, if there are a number of possible nonde- The majority of geochemical analyses reported in
graded samples with which it can be compared, then the literature are concerned with the detection and
the asphaltenes can be isolated from all the samples identiRcation of hydrocarbons. However, many
Figure 11 Gammacerane is another very specific biomarker Figure 13 To simplify the sterane fingerprint to some degree
that can be readily observed in the m/z 191 chromatogram. The GC-MS/MS becomes an important tool. In this diagram the same
relative concentration of this compound varies with the salinity of sterane mixture used for Figure 12 has been analysed in the
the original depositional environment. Hence samples deposited GC-MS/MS mode. By filtering out the parent}daughter ions for
in a very saline lacustrine setting will generally have high gam- the C27 steranes, the top chromatogram shows only the C27 com-
macerane contents whereas those from freshwater lacustrine pounds. Distributions for the C28}C30 steranes could also be
environments are generally depleted in gammacerane. obtained using the appropriate parent}daughter ions.
Used in conjunction with the conventional detection

of biomarkers, this method provides a very powerful
and additional tool for geochemical analyses.
Gas ChromatographyÎsotope
Ratio Mass Spectrometry
Carbon naturally exists as a mixture of its two stable
isotopes, 12C and 13C, in an approximate 12C/13C ratio
of 99 : 1. The carbon isotopic composition of living
organic matter in part depends on the species but is
also determined by a number of environmental prop-
erties. For example, atmospheric carbon dioxide is
assimilated by living plants during photosynthesis
Figure 14 One of the most useful ways for characterizing the
insoluble organic matter in a source rock, coal, shale or soil and the nature of the plants and whether they assimi-
sample is by pyrolysis-GC. This figure illustrates the results ob- late CO2 via a C3 or C4 photosynthetic cycle will
tained from pyrolysis-GC of Messel shale and shows that the determine the extent of preferential assimilation of
major components obtained in the approach are a series of the lighter 12C isotope. C3 plants are typically asso-
alkane/alkene doublets. Variations in these distributions can be
ciated with warmer and more arid climates and in
used to characterize the organic matter in terms of whether it is
algal or terrestrial as well as provide information on the relative general have isotopic values in the !10 to !18
maturity of the samples. range. C4 plants are more typically associated with
colder climates and have lighter isotopic values in the
!22 to !30 range. To determine the 13C com-
position, the sample is combusted to convert all of the
geochemical mixtures contain complex mixtures of
carbon to CO2, which is analysed in a stable isotope
compounds in which heteroatoms are mixed with the
ratio mass spectrometer and compared with the iso-
hydrocarbons in varying amounts. In certain applica-
topic composition of a standard material (Pee Dee
tions knowledge of these components, particularly
Belimnite, PDB), whose isotopic composition has
sulfur-containing compounds, may be extremely im-
been assigned a value of 0.
portant. There are two approaches by which such
distributions may be obtained. The Rrst is by the use
of element-selective GC detectors, such as the FPD, GC-IRMS and Isotopic Composition
Hall detector, or one of the more recent types of AED
that are selective for sulfur-containing compounds at
of Individual Components
the exclusion of non-sulfur-containing compounds. GC-IRMS permits acquisition of 13C values for indi-
The second method combines GC with high resolu- vidual components in complex mixtures. The impor-
tion MS. Although this approach is not used routine- tant part of the system is the interface between the
ly, it is a very powerful and speciRc technique for this GC and the isotope ratio mass spectrometer. This
type of application, as discussed by Tibbetts and consists of a reactor tube, generally a narrow-bore
Large in 1988. While the use of low resolution GC- ceramic tube, containing a bundle of wires, typically
MS and ancillary techniques such as single ion copper, nickel or platinum where complete combus-
monitoring and multiple ion detection have been dis- tion to CO2 must occur. After combustion the water
cussed elsewhere, it needs to be recognized that utiliz- and CO2 pass through a membrane separator to re-
ation of nominal masses in MID may lead to ambigu- move the water before the CO2 enters the mass spec-
ous results. As demonstrated by Tibbets and Large, trometer, where the isotopic composition of the gas is
while the ion at nominal mass 184 may be used for the determined relative to the standard. The isotopic
determination of dibenzothiophenes (DBT), it is also values of individual components can be interpreted to
the nominal mass for the C4-substituted naphthalenes, obtain information on the diagenetic history of an
leading to possible misinterpretation of the resulting individual component and the nature of the microbial
chromatograms. However, use of the accurate mass at community during deposition.
m/z 184.0347 permits detection of only the DBT and The isotopic composition of individual compounds
no substituted naphthalenes. Several examples have is also of importance from an environmental view-
been given by Tibbetts and Large on the use of this point. For example, analysis of a whole oil, or the
approach for the correlation of crude oils or to distin- saturate fraction of a whole oil, allows the ready
guish oils from different sources or reservoirs. determination of 13C values of the n-alkanes and the
major isoprenoids, pristane and phytane. These that, despite the loss of some of the light ends through
values can be used for correlation purposes, to distin- evaporation, a good relationship between the two
guish oils from different sources, to correlate oil samples can be established. These data could also be
spills with their suspected sources, or to determine the used in support of the biomarker and other properties
source of hydrocarbons that have contaminated normally used to establish relationships between sam-
wildlife. Examples of this approach are shown in ples thought to be related. In an additional example,
Figure 15. Gas chromatograms of oil extracted from Figure 16 shows the results from the GC-IRMS ana-
the feathers of birds that had been exposed to a crude lyses of 20 oils from a region in SE Asia. It can be seen
oil spill and a suspected source are compared with that on the basis of these analyses two distinct fami-
each other and show certain differences, parti- lies of oils are present in the region. One family has
cularly at the lighter end of the chromatograms. Such isotopic values of around !20 for each compound
differences could lead to a dispute as to whether whereas the other family has values around !28.
or not this oil was actually the one that was respon- This information can be used in conjunction with the
sible for contaminating the birds. The lower part of biomarker data to determine the signiRcance of these
the Rgure shows the carbon isotope data for indi- differences.
vidual compounds in the two samples. It can be seen While these are just two examples, we have also
shown that GC-IRMS can be used in an environ-
mental context to correlate weathered and un-
weathered oils and reRned products and their
weathered counterparts. If there are small amounts of
the n-alkanes remaining it should be possible to ob-
tain their isotopic composition and subsequently use
these values to make the correlation. Alternatively if
the samples have been so extensively biodegraded
that all of the n-alkanes have been removed, it is
possible to isolate the asphaltenes, pyrolyse them and
analyse the pyrolysates by GC-IRMS. Correlations
can then be made using these data. This is a parti-
cularly valuable approach for the correlation of re-
Rned products that only contain lower carbon num-
ber compounds and none of the more reliable bio-
marker compounds that are typically used for
correlation purposes.
Summary
This article has attempted to illustrate the importance
of GC to geochemical analyses. Geochemical samples
from all sources, whether recent or ancient, oils or
synthetic chemicals, reRned or crude, are incredibly
complex mixtures of organic compounds in most
cases. To try and analyse such samples, whether sim-
ply for correlation purposes or to detect and identify
unknown compounds, almost inevitably requires
some level of chromatography to facilitate the
analytical process. The most common forms of
chromatography generally involve some form of
Figure 15 (A) Chromatogram of an extract from bird feathers
contaminated by a crude oil; (B) shows the chromatogram for the liquid chromatography in the initial steps to simplify
suspected source of the contamination. The two chromatograms the mixture into compound classes, followed by GC
show some significant differences, particularly in the early part of to separate and resolve as many compounds as pos-
the chromatogram, mainly as a result of weathering. However, in sible in the resulting fractions. Chromatography
(C) we see that there is a strong relationship between the isotopic
alone simply separates the components, hence it is
compositions for each individual n-alkane in the two samples.
This isotopic information can be used in conjunction with the very common in most geochemical analyses for the
biomarker data and other parameters to establish a relationship chromatographic step to be combined with an identi-
between the two samples. Rcation technique such as MS. The results of such
Figure 16 Analysis of 20 oils by GC-IRMS provides isotopic data for each major compound present in the oils. The oils plotted on this
chart can be divided into isotopically heavy and isotopically light groups. Supporting evidence for these groupings can be obtained from
the GC-MS data.
analyses generally provide the information necessary tion of the Indonesian Petroleum Association, Jakarta,
to determine the origin of a particular sample and, in pp. 483d507.
the context or crude oil exploration, relate it to pos- Connan J (1984) Biodegradation of crude oils in reservoirs.
sible source rocks and such information as age, ma- In: Brooks J and Welte DH (eds) Advances in Petroleum
turity, and migration pathways. For environmental Geochemistry, vol. 1, pp. 299}335. London: Academic
Press.
samples the information obtained is generally for the
England WA (1989) The organic geochemistry of petro-
purpose of determining the source of a spill and hence leum reservoirs. Organic Geochemistry 16(1}3):
a great deal of use is made of these distributions in 415}425.
terms of their Rngerprinting capability. As chromato- Hayes JM, Freeman KH, Popp BN and Hoham CH (1990)
graphic and spectroscopic techniques continue to im- Compound-speciRc isotopic analyses: a novel tool for
prove, clearly the degree of separation achievable for reconstruction of ancient biogeochemical processes. Or-
these complex mixtures will also greatly improve, but ganic Geochemistry 16 (1d3): 1115}1128.
mixtures of even greater complexities will always be Johnson D, Quimby BD and Sullivan JJ (1995) An atomic
available to provide that next level of challenge. emission detector for gas chromatography. American
Laboratory, pp. 13}20.
See also: II/Chromatography: Gas: Column Techno- Killops SD and Readman JW (1985) HPLC fractionation
logy; Detectors: General (Flame Ionization Detectors and and GC-MS determination of aromatic hydrocarbons
Thermal Conductivity Detectors); Detectors: Mass Spec- from oils and sediments. Organic Geochemistry 8:
247}257.
trometry; Detectors: Selective; Historical Development;
Levy JM (1994) Fossil fuel applications of SFC and SFE:
Pyrolysis Gas Chromatography; Theory of Gas a review. Journal of High Resolution Chromatography
Chromatography. 17: 212}216.
Peters KE and Moldowan JM (1992) The Biomarker
Further Reading Guide: Interpreting Molecular Fossils in Petroleum and
Ancient Sediments. Englewood Cliffs, NJ: Prentice
Baskin DK, Hwang RJ and Purdy RK (1995) Prediction of Hall.
gas, oil, and water intervals in Niger Delta reservoirs Philp RP and Oung J-N (1992) Biomarkers, occurrence,
using gas chromatography. American Association of Pe- utility and detection. In: Voress L (ed.) Instrumentation
troleum Geologists Bulletin 79: 337}350. in Analytical Chemistry, 1988}1991, pp. 368}376.
Brooks J and Fleet AJ (eds) (1987) Marine Petroleum Washington, DC: American Chemical Society.
Source Rocks, Geological Society Special Publication, Del Rio JC and Philp RP (1992) High molecular weight
vol. 26, 444 pp. Oxford: Blackwell ScientiRc Publica- hydrocarbons: a new frontier in organic geochemistry.
tions. Trends in Analytical Chemistry 11 (15): 187}193.
Carlson RMK, Teerman SC, Moldowan JM, Jacobson SR, Tibbetts PJC and Large R (1988) Improvements in oil
Chan EI, Dorrough KS, Seetoo WC and Mertani Rngerprinting: GC/HRMS of sulphur heterocycles. In:
B (1993) High temperature gas chromatography of high Crump GB (ed.) Petroanalysis 87, pp. 45}57. London:
wax oils. In: Proceedings of the 20th Annual Conven- John Wiley.
2960 III / GLYCOPROTEINS: LIQUID CHROMATOGRAPHY
GLYCOPROTEINS:
K. Miyazaki, Hokkaido University Hospital, number of antennae. The basic branch structures are
Hokkaido, Japan composed in most instances of one GlcNAc and
Copyright ^ 2000 Academic Press one Gal residue (Figure 1).
O-Glycosyl glycoproteins contain at their reducing
end a GalNAc residue that is linked to the hydroxyl
Introduction group of either serine or threonine residues of a poly-
Many proteins in cells and biological Suids are peptide. This linkage is called an O-linked or mucin-
glycosylated, and these glycoproteins are present in type sugar chain. In general, O-linked structures
animals, plants, microorganisms and viruses. The appear to be less complex than N-linkages in terms of
most commonly occurring monosaccharides found in the number of antennae and monosaccharides. How-
oligosaccharide attachments to mammalian proteins ever, they can be fucosylated and sialylated. Some
are D-mannose (Man), D-galactose (Gal), D-glu- glycoproteins have both the N-linked and O-linked
cose (Glu), L-fucose (Fuc), N-acetylglucosamine forms in their molecules (N, O-glycoproteins).
(GlcNAc), N-acetylgalactosamine (GalNAc), and N- The addition of carbohydrate to a peptide chain
acetylneuraminic acid (sialic acid or NeuAc). changes the shape and size of the protein structure.
The primary structure of glycoprotein glycans and Several important discoveries have revealed the fol-
their biological functions have been gradually un- lowing biological roles of glycans: (i) protection of
ravelled by the improvements of methods for isola- polypeptide chains against proteolytic enzymes; (ii)
tion and structure determination. High performance inSuence on heat stability, solubility, and many
liquid chromatography (HPLC) is one of the most physicochemical properties; and (iii) interaction with
commonly used methods for the isolation and analy- other proteins or nonprotein components of the
sis of both glycoproteins and their derived carbohy- cell, including control of the lifetime of circulating
drates, mainly due to the excellent resolution, ease of glycoproteins and cells.
use, the generally high recoveries, the excellent repro-
ducibility of repetitive separations, and the high pro- Microheterogeneity of Glycans
ductivity in terms of cost parameters.
In addition to genetically determined variants ex-
pressed as variations in their polypeptide chains,
Chemistry and Importance almost all glycoproteins exhibit polymorphism
of the Glycan Chain of Glycoproteins associated with their glycan moieties. This type of
diversity is termed microheterogeneity, and these
Basic Structure of Glycoprotein
different forms have recently been called glyco-
In glycoproteins, glycans are conjugated to peptide forms. These variants were Rrst characterized in the
chains by two types of primary covalent linkage, 1-acid glycoprotein (AAG) from human serum using
N-glycosyl and O-glycosyl. The former is called an electrophoresis. As shown in the structure of major
N-linked sugar chain and contains a GlcNAc residue oligosaccharides of AAG (Figure 1), micro-
that is linked to the amide group of asparagine resi- heterogeneity was found to be due to the occurrence
dues of a polypeptide. As shown in Figure 1, almost of di-, tri-, and tetraantennary glycans of the N-
all N-linked glycoproteins have a common core of acetyllactosamine type at the Rve glycosylation sites.
two GlcNAc and three Man residues. N-linked glyco- This feature is widespread and has been observed
proteins have three types of carbohydrate moieties: in natural as well as in recombinant DNA glycopro-
complex type (Figure 1), high mannose type and hy- teins. The existence of microheterogeneity gives rise
brid type. The hybrid type is a mixture of the complex to many interesting questions regarding the origin of
and high mannose types. Complex type structures this phenomenon and its relevance to the biological
usually have from two to four branches attached to functioning of the glycoproteins that can be distin-
the two outer core Man residues. The branches are guished.
distributed over the two terminating core Man resi- Recent interest has been shown in glycoproteins in
dues. The complex structures are termed diantennary, the industrial Reld of genetic engineering of human
triantennary and tetraantennary, according to the glycoproteins of therapeutic interest. This gives rise to
III / GLYCOPROTEINS: LIQUID CHROMATOGRAPHY 2961
Figure 1 Structure of the major oligosaccharides of 1-acid glycoprotein (a complex type of the N-linked form). Several NeuAc are
linked to Gal residues.
an enormous problem, because the production of glycoprotein molecules as well as for the development
recombinant human glycoproteins in nonhuman eu- of a recombinant DNA-derived glycoprotein as
karyotic cells or in prokaryotic cells devoid of glycan a pharmaceutical agent. However, the composition,
biosynthesis machinery leads to the production type and branching pattern of carbohydrates are
of incorrectly glycosylated proteins. Incorrectly complex due to the diversity of monosaccharides and
glycosylated glycoproteins may have an undesirable the variety of possible linkages. Unlike amino acids,
effect on therapeutic effectiveness and safety which are linked through an amido bond, monosac-
due to changes in the properties of the products, charides are joined through a variety of hydroxyl
including a decrease in the stability against heat or groups present on the sugar to form glycosidic link-
protease, shortening of the in vivo life span of the ages. For example, two different amino acids can
molecules by an increase in clearance, a decrease in form only two dipeptides, while two different
the afRnity for speciRc receptors, and an increase monosaccharides can lead to more than 60 disacchar-
in antigenicity. ides. However, the availability of improved and
sophisticated methods for the isolation and character-
ization of glycoproteins and their derived glycans has
Isolation and Quantitation paved the way for the analysis and characterization of
of Glycoprotein Molecules and the carbohydrate chains of glycoprotein. These ana-
Analysis of Glycan Chains lytical methods include mass spectrometry, enzymatic
Determination of the primary structure of glycopro- microsequencing, nuclear magnetic resonance
teins necessitates analysis of the protein sequence, (NMR), capillary electrophoresis, reversed-phase
identiRcation of the glycosylation sites, unravelling of HPLC (RP-HPLC), and high pH anion exchange
the glycan structures and determination of the micro- chromatography with pulsed amperometric detection
heterogeneity of the glycans at each glycosylation. (HPAEC-PAD), and generally a combination of sev-
For these studies, it is essential that adequate starting eral complementary analytical methods is needed to
materials are available. determine the carbohydrate structure.
For the isolation or puriRcation of glycoproteins, In this section, we give an outline of the recently
a combination of several complementary separation developed HPLC procedures for the puriRcation, sep-
methods such as gel permeation chromatography, aration, and determination of glycoproteins and their
afRnity chromatography (lectin or others), anion- glycoforms.
or cation-column chromatography, high performance
Examples of Isolation or Puri\cation
capillary electrophoresis, and HPLC using several
of Glycoproteins by Using HPLC
sorbents is generally used.
Determination of the carbohydrate composition, 1-Acid glycoprotein (AAG) AAG is a characteristic
type and branching pattern is an important step for and dominant fraction of human serum sialoglyco-
understanding the biological function of the native proteins with a molecular mass approximately 44 000
Da, an unusually high carbohydrate content (45%) procedure involves two membrane Rltration steps,
and a large number of sialyl residues. Although its Sephadex G-25, two DEAE-agarose steps, Sephadex
exact biological function is still unknown, AAG is an G-75, wheat germ agglutinin (WGA)-agarose, and
acute-phase reactant that increases following cancer, RP-HPLC. The Rnal HPLC step is essential for the
myocardial infarction, and congestive heart failure removal of nucleic acids.
and has also been reported to play an important role
in immunoregulation. Chromatographic Determination of AAG
Ion exchange chromatography and gel permeation
There have been very few reports on chromato-
chromatography previously used for puriRcation are
graphic methods for determining concentrations of
time-consuming and require a large volume of plasma
glycoproteins other than AAG in biological samples.
or serum because of the low quantities recovered.
Radial immunodiffusion (RID) utilizing the anti-
These methods also lead to a strong possibility of
body against AAG has been widely used to determine
denaturation and desialylation. Moreover, separation
AAG in serum because of its strong speciRcity. This
of AAG and 1-antitrypsin has been difRcult, be-
method, however, is time-consuming and is not easily
cause the chromatographic behaviour of these com-
applicable to experimental animals. To overcome
pounds during anion exchange chromatography is
these problems, some simple and rapid HPLC
similar. Recently, these problems have been over-
methods have been developed.
come by the introduction of an HPLC system equip-
After pretreatment of human serum with a chloro-
ped with a hydroxyapatite column as the last step
form/methanol mixture (2 : 1, v/v), 500 L of the
after the clean-up procedures with commercially
upper phase was applied to the anion exchange FPLC
available cartridge ion exchange columns.
system (Mono Q HR 5/5 column), and AAG was
eluted with a pH/NaCl gradient elution programme.
Ceruloplasmin (CP) CP is a serum 2-glycoprotein
To measure the serum AAG content, the Mono Q HR
that carries more than 95% of the copper present in
column was calibrated with commercial AAG in the
plasma and is believed to have an active role in the
range 100}200 g/500 L of sample volume.
regulation of copper and iron homeostasis. It has
A rapid and sensitive determination method start-
been pointed out that fragmentation of CP during
ing from the diluted human serum itself has also been
puriRcation and storage has hampered the study of its
reported. This procedure involves the anion exchange
structure. The rapid degradation of puriRed CP re-
step for cleaning up serum (commercially available
ported by many laboratories may be largely due to the
cartridge column, DEAE-M) and a hydroxyapatite
presence of one or more copurifying or contamina-
HPLC system. A linear relationship between standard
ting proteases, at least one of which is a metallo-
AAG concentration and peak height was observed
proteinase.
over the concentration range 0.5}2.5 mg mL\1
Recently, a highly puriRed and nonlabile CP has
serum. The coefRcient of variation at
been obtained from human plasma by combining the
0.5 mg mL\1 AAG was 3.7% (n"8). A good cor-
previously reported chromatographic steps with addi-
relation was observed between this HPLC method (y)
tional gel permeation and fast protein liquid
and the conventional RID (x) (y"1.009x #0.004,
chromatography (FPLC) steps. In the latter steps,
r"0.996).
further puriRcation of CP by Sephadex G-50
chromatography and Mono Q FPLC were essential
Determining the Carbohydrate Composition, Type
for the removal of plasma metalloproteinase, and this
and Branching Pattern
puriRcation procedure yielded a protein that was
completely stable even after incubation at 373C for RP-HPLC has become a commonly used method for
4 weeks. the analysis and puriRcation of peptides, proteins and
glycoproteins. The RP-HPLC experimental system
Erythropoietin (EPO) EPO, an acidic glycoprotein usually comprises an n-alkylsilica-based sorbent. By
hormone, is synthesized in the kidney and circulates using modern instrumentation and columns, complex
in the blood to stimulate red cell proliferation and mixtures of peptides and proteins can be separated
differentiation in bone marrow. Native human and low picomolar amounts of resolved components
EPO was Rrst puriRed from the urine of patients can be collected. Separation can be easily performed
suffering from severe aplastic anaemia. Since by changing the gradient slope of solvents such as
then, several methods for the puriRcation of urinary acetonitrile containing an ionic modiRer (e.g. tri-
human EPO (uHuEPO) have been developed. RP- Suoroacetic acid (TFA)); column temperature; or the
HPLC has recently been used for the puriRcation of organic solvent composition. The technique is equally
uHuEPO with high in vivo activity. This puriRcation applicable to the analysis of enzymatically derived
III / GLYCOPROTEINS: LIQUID CHROMATOGRAPHY 2963
mixtures of peptides from proteins and glycoproteins.

Separated fractions can be subsequently subjected to
further analysis of carbohydrates and amino acids.
A new HPLC method, HPAEC-PAD, which by-
passes the derivatization steps by using pulsed elec-
trochemical detection on gold electrodes, has been
developed. Monosaccharides and oligosaccharides
can be directly resolved by anion exchange
chromatography, because the hydroxyl groups of
carbohydrates are weakly acidic and reveal anionic
forms at pH values greater than pH 12. In addition to
high sensitivity in the low picomole range of PAD,
a major advantage of HPAEC-PAD is its usefulness in
analysing both monosaccharides and all classes of
oligosaccharides without derivatization. HPAEC-
PAD has therefore been used successfully for resolv-
ing and quantitating the constituent monosaccharides
released by acidic hydrolysis (e.g. TFA) of glycan
chains and for resolving N-linked oligosaccharides
separated by enzyme digestion (e.g. PNGase F).
Figure 2 shows the HPAEC-PAD chromatograms
of fractions 23}27 from the RP-HPLC separation of
a tryptic digest of recombinant tissue plasminogen
activator (tPA). The peaks from RP-HPLC separation Figure 3 Representative chromatograms of monosaccharides
after treatment of standard and plasma samples. (A) Standard
were collected manually, and aliquots of all 62 frac-
sample containing 5.0 g mL\1 of each monosaccharide; (B)
tions were analysed for neutral and amino monosac- healthy subject; (C) patient with renal insufficiency. Peaks:
charides after acid hydrolysis. The chromatograms in 1"mannitol (internal standard); 2"Fuc; 3"GlcNAc; 4"Gal;
5"Man. (Reproduced with permission from Kishino et al., 1995.)
Figure 2 indicate that fractions 24}26 contain

glycopeptides, and the ratio of constituent monosac-
charides suggests that their oligosaccharide structures
are those of fucosylated N-acetyllactosamine-type
oligosaccharides. Another N-acetyllactosamine-type
chain and an oligomannose-type chain were identiRed
similarly by the same analytical procedure.
The HPAEC-PAD method is also applicable to the
quantitation of concentrations of monosaccharides
after release by acid hydrolysis and following clean-
up procedures with commercially available cartridge
columns. Figure 3 shows the chromatograms of four
monosaccharides in puriRed serum AAG from
healthy subjects and from patients with renal insuf-
Rciency. The concentration of NeuAc can also be
determined under different solvent conditions.
This method enables composition analysis of the
carbohydrate moiety of AAG with only 1.0 mL of
plasma. Linear relations between the amount of
NeuAc or monosaccharides and the peak-height ratio
of NeuAc or monosaccharides to the internal stan-
Figure 2 Monosaccharide analysis of fractions 23}27 from
dards are observed over the concentration range of
RP-HPLC separation of a tryptic digest of recombinant tissue
plasminogen activator. Elution positions of monosaccharide stan- 5.0 to 100 g mL\1. N-Glycolylneuraminic acid and
dards are indicated on the upper trace. (Reproduced with per- mannitol are used as the internal standard for NeuAc
mission from Townsend et al., 1996.) and the four monosaccharides, respectively.
Table 1 Analysis of NeuAc and monosaccharide levels in purified 1-acid glycoprotein (AAG) from plasma of healthy subjects,
patients with renal insufficiency and patients with myocardial infarction
Healthy subjects Renal insufficiency Myocardial infarction

(n"8) (n"6) (n"4)
Age (years) 65.8$12.5 60.2$7.2 69.7$11.7

AAG conc. (mg mL\1) 0.79$0.14 1.01$0.19a 1.66$0.34b
NeuAc (mg g\1 ) AAG) 82.77$12.55 89.29$15.55 132.70$6.89b

Fuc (mg g\1 ) AAG) 9.84$3.08 12.79$3.37 10.40$3.42
GluNAc (mg g\1 ) AAG) 113.01$10.07 135.44$10.51b 142.68$12.96b
Gal (mg g\1 ) AAG) 76.97$4.23 88.78$5.61b 86.92$11.73a
Man (mg g\1 ) AAG) 49.14$2.57 57.26$3.80b 50.16$1.67
GluNAc/Man 2.18$0.37 2.50$0.35 2.84$0.21a
Source: Kishino et al. (1995).

AAG concentration was determined by the HPLC method with a hydroxyapatite column.
NeuAc and each monosaccharide concentration were determined by HPAEC-PAD.
Values in the table are means$SD.
a
Significantly different (p(0.05) from healthy subjects.
b
Significantly different (p(0.01) from healthy subjects.
The resultant quantitation data (Table 1) for charides (NeuAc, Fuc, GlcNAc, Gal, Man) in each
healthy subjects, patients with renal insufRciency fraction by HPAEC-PAD, it was found that glyco-
and patients with myocardial infarction show that forms rich in carbohydrates were eluted later (frac-
not only AAG levels but also the concentrations of tions 4, 5, 6) and that NeuAc was relatively abundant
several monosaccharides in patients increased signiR- in these highly adsorbed glycoforms, especially in
cantly compared to those of healthy subjects, sugges- fraction 6. Furthermore, the ratio of GlcNAc/Man in
ting a change in the carbohydrate branching pattern fraction 2 was signiRcantly higher than those in the
in such pathologic conditions. other fractions, suggesting the presence of a highly
It is well known that the microheterogeneity of branched glycan chain. Interestingly, it has been also
AAG is due to the occurrence of di-, tri-, and tetra- shown that fractions 1 and 2, both relatively rich in
antennary glycans of the N-acetyllactosamine type at highly branched glycan chains, showed a signiRcantly
the Rve glycosylation sites. Moreover, the Man con- lower binding capacity to disopyramide, a drug for
tent is constant among the antennary glycans, and the the treatment of arrhythmia, than did the other frac-
number of branches increases with the addition of tions. This result suggests that the binding sites of
GlcNAc to Man residues. A highly branched glycan AAG to disopyramide are hindered by relatively large
chain of AAG is constructed by the linkage of Gal to carbohydrate moieties, such as a tetraantennary
GlcNAc (Figure 1), which results in the formation of structure. These results are consistent with the Rnd-
an antennary structure (N-acetyllactosamine). There- ings that the binding capacity of puriRed AAG iso-
fore, in the case of AAG, determination of the con- lated from patients (with renal insufRciency or
centration ratio of GlcNAc to Man (GlcNAc/Man) is myocardial infarction) to disopyramide is signiR-
important for estimating whether the carbohydrate cantly lower than that of healthy subjects and that the
moiety of glycoforms has a highly or less-branched AAGs of these patients revealed a higher concentra-
glycan chain. The signiRcantly higher GlcNAc/Man tion ratio of GlcNAc/Man, an index of the abund-
ratio in the patients with myocardial infarction sug- ance of highly branched glycan chains.
gests that a highly branched glycan chain was syn- In conclusion, in order to gain further insight into
thesized. Changes in the carbohydrate moiety in the the structure}function relations of the carbohydrate
glycoproteins have been reported in patients with moiety, it is essential that sufRcient quantities of
various types of disease. glycoprotein variants are available. An effective
As shown in Figure 4, at least six fractions, which combination of the sophisticated separation methods
are possibly based on carbohydrate-mediated micro- for glycoforms, such as the HPLC system using a hy-
heterogeneity, have been obtained from healthy droxyapatite column, and the qualitative and quantit-
human (Japanese) serum AAG by HPLC using a ative analytical methods for monosaccharides/
hydroxyapatite column under a gradient elution pro- oligosaccharides, such as HPAEC-PAD, must be
gramme. From the determination of Rve monosac- established.
III / GOLD RECOVERY: FLOTATION 2965
Figure 4 Typical chromatograms of the glycoforms of 1-acid glycoprotein (AAG) from the serum of healthy subjects by HPLC. Inlet is
the gradient programme for the fractionation of the glycoforms of AAG. (Sampling time of each fraction: fraction 1, 17}22 min; 2, 27}36
min; 3, 43}50 min; 4, 53}57 min; 5, 58}62 min and 6, 65}72 min.) (Reproduced with permission from Kishino et al., 1997.)
See also: III/Carbohydrates: Liquid Chromatography. Kishino S and Miyazaki K (1997) Journal of Chromatogra-
Peptides and Proteins: Liquid Chromatography. Poly- phy 699: 371}381.
saccharides: Liquid Chromatography. Kishino S, Nomura A, Saitoh M, Sugawara M, Iseki K,
Kitabatake A and Miyazaki K (1997) Journal of
Further Reading Montreuil J (1995) In: Montreuil J, Schachter H and
Vliegenthart JFG (eds) Glycoproteins, pp. 1}12. Amster-
Clemetson KJ (1997) In: Montreuil J, Vliegenthart JFG and
dam: Elsevier.
Schachter H (eds) Glycoproteins II, pp. 173}201.
Schmid K (1989) In: Bauman P, Eap CB, Muler WE and
Amsterdam: Elsevier.
Tillement J-P (eds) Alpha1-Acid Glycoprotein, pp. 7}22.
Hancock WS, Chakel AAJ, Souders C, M’Timkulu T, Pun-
New York: Alan R. Liss.
gor E Jr and Guzzetta AW (1996) In: Karger BL and
Townsend RR, Basa LJ and Spellman MW (1996) In:
Hancock WS (eds) Methods in Enzymology, vol. 271,
Karger BL and Hancock WS (eds) Methods in Enzymol-
pp. 403}427. San Diego: Academic Press.
ogy, vol. 271, pp. 135}147. San Diego: Academic
Hardy MR and Townsend RR (1994) In: Lennarz WJ and
Press.
Hart GW (eds) Methods in Enzymology, vol. 230,
Vliegenthart JFG and Montreuil J (1995) In: Montreuil J,
Schachter H and Vliegenthart JFG (eds) Glycoproteins,
Kishino S, Nomura A, Sugawara M, Iseki K, Kakinoki S,
pp. 13}28. Amsterdam: Elsevier.
GOLD RECOVERY: FLOTATION
S. Bulatovic and D. M. Wyslouzil, Lakefield Research, mineralogy of the ore and the distribution of
Lakefield, Ontario, Canada gold in the ore. The methods used for the re-
Copyright ^ 2000 Academic Press covery of gold consist of the following unit opera-
tions:
Introduction
E The gravity preconcentration method, which is
The recovery of gold from gold-bearing ores mainly used for recovery of gold from placer de-
depends largely on the nature of the deposit, the posits that contain coarse native gold. Gravity is
Figure 4 Typical chromatograms of the glycoforms of 1-acid glycoprotein (AAG) from the serum of healthy subjects by HPLC. Inlet is
the gradient programme for the fractionation of the glycoforms of AAG. (Sampling time of each fraction: fraction 1, 17}22 min; 2, 27}36
min; 3, 43}50 min; 4, 53}57 min; 5, 58}62 min and 6, 65}72 min.) (Reproduced with permission from Kishino et al., 1997.)
See also: III/Carbohydrates: Liquid Chromatography. Kishino S and Miyazaki K (1997) Journal of Chromatogra-
Peptides and Proteins: Liquid Chromatography. Poly- phy 699: 371}381.
saccharides: Liquid Chromatography. Kishino S, Nomura A, Saitoh M, Sugawara M, Iseki K,
Further Reading Montreuil J (1995) In: Montreuil J, Schachter H and
Vliegenthart JFG (eds) Glycoproteins, pp. 1}12. Amster-
Clemetson KJ (1997) In: Montreuil J, Vliegenthart JFG and
dam: Elsevier.
Schachter H (eds) Glycoproteins II, pp. 173}201.
Schmid K (1989) In: Bauman P, Eap CB, Muler WE and
Amsterdam: Elsevier.
Tillement J-P (eds) Alpha1-Acid Glycoprotein, pp. 7}22.
Hancock WS, Chakel AAJ, Souders C, M’Timkulu T, Pun-
New York: Alan R. Liss.
gor E Jr and Guzzetta AW (1996) In: Karger BL and
Townsend RR, Basa LJ and Spellman MW (1996) In:
Hancock WS (eds) Methods in Enzymology, vol. 271,
Karger BL and Hancock WS (eds) Methods in Enzymol-
ogy, vol. 271, pp. 135}147. San Diego: Academic
Hardy MR and Townsend RR (1994) In: Lennarz WJ and
Press.
Hart GW (eds) Methods in Enzymology, vol. 230,
Vliegenthart JFG and Montreuil J (1995) In: Montreuil J,
Schachter H and Vliegenthart JFG (eds) Glycoproteins,
Kishino S, Nomura A, Sugawara M, Iseki K, Kakinoki S,
pp. 13}28. Amsterdam: Elsevier.
GOLD RECOVERY: FLOTATION
S. Bulatovic and D. M. Wyslouzil, Lakefield Research, mineralogy of the ore and the distribution of
Lakefield, Ontario, Canada gold in the ore. The methods used for the re-
Copyright ^ 2000 Academic Press covery of gold consist of the following unit opera-
tions:
Introduction
E The gravity preconcentration method, which is
The recovery of gold from gold-bearing ores mainly used for recovery of gold from placer de-
depends largely on the nature of the deposit, the posits that contain coarse native gold. Gravity is
2966 III / GOLD RECOVERY: FLOTATION
often used in combination with Sotation and/or gold may occur in several of the above forms, which
cyanidation. affects Sotation recovery.
E Hydrometallurgical methods are normally em- During Sotation of gold-bearing massive sulRde
ployed for recovery of gold from oxidized deposits ores, the emphasis is generally placed on the produc-
(heap leach), low grade sulRde ores (cyanidation, tion of base metal concentrates and gold recovery
carbon-in-pulp (CIP), carbon-in-leach (CIL)) and becomes a secondary consideration. In some cases,
refractory gold ores (autoclave, biological de- where signiRcant quantities of gold are contained in
composition followed by cyanidation). base metal ores, the gold is Soated from the base
E A combination of pyrometallurgical (roasting) and metal tailings.
hydrometallurgical route is used for highly refrac- The Sotation of gold-bearing ores is classiRed ac-
tory gold ores (carbonaceous sulRdes, arsenical cording to ore type (i.e. gold ore, gold}copper ore,
gold ores) and the ores that contain impurities that gold}antimony ores), because the Sotation methods
result in a high consumption of cyanide, which used for the recovery of gold from different ores is
have to be removed before cyanidation. vastly different.
E The Sotation method is a widely used technique for
the recovery of gold from gold-containing copper Geology and General Mineralogy
ores, base metal ores, copper}nickel ores, platinum of Gold-bearing Ores
group ores and many other ores where other pro-
cesses are not applicable. Flotation is also used for The geology of the deposit and the mineralogy of the
the removal of interfering impurities before hydro- ore play a decisive role in the selection of the best
metallurgical treatment (i.e. carbon preSoat), for treatment method for a particular gold ore. Geology
upgrading of low sulRde and refractory ores for of the gold deposits varies considerably, not only
further treatment. Flotation is considered to be the from deposit to deposit, but also within the deposit.
most cost-effective method for concentrating gold. Table 1 shows major genetic types of gold ores and
their mineral composition. More than 50% of the
SigniRcant progress has been made over the past total world gold production comes from clastic sedi-
several decades in the recovery of gold using hydro- mentary deposits.
metallurgical methods, including cyanidation (CIL, In many geological ore types, several subtypes can
resin-in-pulp) and bio-oxidation. All of these probe found, including primary ores, secondary ores and
cesses are well documented in the literature and
abundantly described. However, very little is known
Table 1 Common genetic types of gold deposits
about the Sotation properties of gold contained in
various ores and the sulRdes that carry gold. The Ore type Description
sparse distribution of discrete gold minerals, as
Magmatic Gold occurs as an alloy with copper,
well as their exceedingly low concentrations in the nickel and platinum group metals
ore, is one of the principal reasons for the lack of Typically contains low amount of gold
fundamental work on the Sotation of gold-bearing Ores in clastic Placer deposits, in general conglomer-
ores. sedimentary rock ates, which contain quartz, sericite, chlor-
ite, tourmaline and sometimes rutile and
In spite of the lack of basic research on Sotation of
graphite. Gold can be coarse. Some de-
gold-bearing ores, the Sotation technique is used, not posits contain up to 3% pyrite. Size of the
only for upgrading of low grade gold ore for further gold contained in pyrite ranges from 0.01
treatment, but also for beneRciation and separation to 0.07 m
of difRcult-to-treat (refractory) gold ores. Flotation is Hydrothermal This type contains a variety of ores, in-
cluding:
also the best method for recovery of gold from base
Gold}pyrite ores
metal ores and gold-containing platinum group meta- Gold}copper ores
ls (PGM) ores. Excluding gravity preconcentration, Gold}polymetallic ores
Sotation remains the most cost-effective beneRciation Gold}oxide ore, usually upper zone of
method. sulfide zones
The pyrite content of the ore varies from
Gold itself is a rare metal and the average grades
3% to 90%. Other common waste min-
for low grade deposits vary between 3 and 6 p.p.m. erals are quartz, aluminosilicates, dolomite
Gold occurs predominantly in its native form in sili- Metasomatic or Sometimes very complex and refractory
cate veins, alluvial and placer deposits or encap- scarn ores gold ores. Normally the ores are com-
sulated in sulRdes. Other common occurrences of posed of quartz, sericite, chlorites, cal-
cite, magnetite. Sometimes the ore con-
gold are alloys with copper, tellurium, antimony,
tains wolframite and sheelite
selenium, PGMs and silver. In massive sulRde ores,
Table 2 Major gold minerals
Group Mineral Chemical formula Impurity content
Native gold and its alloys Native gold Au 0}15% Ag

Electrum Au/Ag 15}50% Ag
Cuproauride Au/Cu 5}10% Cu
Amalgam Hg/Au 10}34% Au
Bismuthauride Au/Bi 2}4% Bi
Tellurides Calaverite AuTe2
Sylvanite (Au, Ag)Te2
Petzite (Au, Ag)Te
Magyazite Au(Pb, Sb, Fe)(S, TeII) Unstable
Gold associated with Krennerite AuTe2(Pt, Pl)
platinum group metals Platinum gold AuPt Up to 10% Pt
Rhodite AuRh 30}40% Rh
Rhodian gold AuRh 5}11% Rh
Aurosmiride Au, Ir, Os 5% Os#5}7% Ir
oxide ores. Some of the secondary ores belong to also recovers gold. Normally, for the Sotation of
a group of highly refractory ores, such as those from PGMs and associated gold, a combination of xan-
Nevada (USA), and El Indio (Chile). The number of thate and dithiophosphate is used, along with gangue
gold minerals and their associations are relatively depressants guar gum, dextrin or modiRed cellulose.
small and can be divided into the following three In the South African PGM operations, gold recovery
groups: native gold and its alloys, tellurides and gold into the PGM concentrate ranges from 75% to 80%.
associated with PGMs. Perhaps the most difRcult problem in Sotation of
Table 2 lists major gold minerals and their associ- native gold and its alloys is the tendency of gold to
ations. plate, vein, Sake and assume many shapes during
grinding. Particles with sharp edges tend to detach
Flotation Properties of Gold Minerals from the air bubbles, resulting in gold losses. This
shape factor also affects gold recovery using a gravity
and Factors Affecting Floatability method.
Native gold and its alloys, which are free from surface In Sotation of gold-containing base metal ores,
contaminants, are readily Soatable with xanthate col- a number of modiRers normally used for selective
lectors. Very often, however, gold surfaces are con- Sotation of copper}lead, lead}zinc and copper}lead}
taminated or covered with varieties of impurities. The zinc have a negative effect on the Soatability of gold.
impurities present on gold surfaces may be argentite, Such modiRers include ZnSO4;7H2O, SO2, Na2S2O5,
iron oxides, galena, arsenopyrite or copper oxides. and cyanide when added in excessive amounts.
The thickness of the layer may be in the order of The adsorption of collector on gold and its Soata-
1}5 m. Because of this, the Sotation properties of bility are considerably improved by the presence of
native gold and its alloys vary widely. Gold covered oxygen. Figure 1 shows the relationship between col-
with iron oxides or oxide copper is very difRcult to lector adsorption, oxygen concentration in the pulp
Soat and requires special treatment to remove the and conditioning time. The type of modiRer and the
contaminants. pH are also important parameters in Sotation of gold.
Tellurides on the other hand are readily Soatable in
Flotation of Low Sul\de-containing Gold Ores
the presence of small quantities of collector, and it is
believed that tellurides are naturally hydrophobic. The beneRciation of this ore type usually involves
Tellurides from Minnesota (USA) were Soated using a combination of gravity concentration, cyanidation
dithiophosphate collectors, with over 95% gold and Sotation. For an ore with coarse gold, gold is
recovery. often recovered by gravity and Sotation, followed by
Flotation behaviour of gold associated in the plati- cyanidation of the reground Sotation concentrate. In
num group metals is apparently the same as that for some cases, Sotation is also conducted on the cyani-
the PGMs or other minerals associated with the dation tailing. The reagent combination used in
PGMs (i.e. nickel, pyrrhotite, copper and pyrite). Sotation depends on the nature of gangue present
Therefore, the reagent scheme developed for PGMs in the ore. The usual collectors are xanthates,
developed using the Sow sheet shown in Figure 2,

where the best metallurgical results were obtained
using a three-stage grinding and Sotation approach.
The metallurgical results obtained with different grin-
ding conRgurations are shown in Table 3.
Flotation was carried out at an alkaline pH, con-
trolled by lime. A xanthate collector with cyclic alco-
hol frother (pine oil, cresylic acid) was shown to be
the most effective. The use of small quantities of
a dithiophosphate-type collector, together with xan-
thate, was beneRcial.
Flotation of Carbonaceous Clay-containing

Gold Ores
These ores belong to a group of refractory gold ores,
where Sotation techniques can be used to remove
interfering impurities before the hydrometallurgical
treatment process of the ore for gold recovery and to
Figure 1 Relationship between adsorption of xanthate on gold preconcentrate the ore for further pyrometallurgical
and conditioning time in the presence of various concentrations of or hydrometallurgical treatment. There are several
xanthate. Triangles, O2 2 mg L\1, circles; O2 9 mg L\1; squares, Sotation methods used for beneRciation of this ore
O2 45 mg L\1. type. Some of the most important methods are de-
scribed as follows:
dithiophosphates and mercaptans. In the scavenging E PreSotation of carbonaceous gangue and carbon.
section of the Sotation circuit, two types of collector In this case, only carbonaceous gangue and carbon
are used as secondary collectors. In the case of a par- are recovered by Sotation, in preparation for fur-
tially oxidized ore, auxiliary collectors, such as hy- ther hydrometallurgical treatment of the Soat tails
drocarbon oils with sulRdizer, often yield improved for gold recovery. Carbonaceous gangue and car-
results. The preferred pH regulator is soda ash, which bon are naturally Soatable using only a frother, or
acts as a dispersant and also as a complexing reagent a combination of a frother and a light hydrocarbon
for some heavy metal cations that have a negative oil (fuel oil, kerosene). When the ore contains clay,
effect on gold Sotation. Use of lime often results in regulators for clay dispersion are used. Some of the
the depression of native gold and gold-bearing sul- more effective regulating reagents include sodium
Rdes. The optimum Sotation pH ranges between 8.5 silicates and oxidized starch.
and 10.0. The type of frother also plays an important E Two-stage Sotation method. In this case, carbon-
role in the Sotation of native gold and gold-bearing aceous gangue is preSoated using the method
sulRdes. Glycol esters and cyclic alcohols (pine oil) described above, followed by Sotation of gold-con-
can improve gold recovery signiRcantly. taining sulRdes using activator}collector combina-
Amongst the modifying reagents (depressant), so- tions. In extensive studies conducted on carbon-
dium silicate starch dextrins and low molecular aceous gold-containing ores, it was established that
weight polyacrylamides are often selected as gangue primary amine-treated copper sulfate improved
depressants. Fluorosilicic acid and its salts can also gold recovery considerably. Ammonium salts and
have a positive effect on the Soatability of gold. The sodium sulRde (Na2S;9H2O) also have a positive
presence of soluble iron in a pulp is highly detrimen- effect on gold-bearing sulRde Sotation, at a pH
tal to gold Sotation. The use of small quantities of between 7.5 and 9.0. The metallurgical results ob-
iron-complexing agents, such as polyphosphates and tained with and without modiRed copper sulfate
organic acids, can eliminate the harmful effect of iron. are shown in Table 4.
E Nitrogen atmosphere Sotation method. This tech-
Flotation of Gold-containing Mercury/Antimony Ores
nique uses a nitrogen atmosphere in grinding and
In general, these ores belong to a group of difRcult-to- Sotation to retard oxidation of reactive sulRdes,
treat ores, where cyanidation usually produces poor and has been successfully applied on carbonaceous
extraction. Mercury is partially soluble in cyanide, ores from Nevada (USA). The effectiveness of the
which increases cyanide consumption and reduces method depends on the amount of carbonaceous
extraction. A successful Sotation method has been gangue present in the ore, and the amount and type
Figure 2 Flotation flow sheet developed for the treatment of gold-containing mercury}antimony ore.
of clay. Ores that are high in carbon or contain Gold in the porphyry copper ore may appear as
high clay content (or both) are not amenable for native gold, electrum, cuproaurid and sulfosalts
nitrogen atmosphere Sotation. associated with silver. During the Sotation of
porphyry copper}gold ores, emphasis is usually
Flotation of Gold-containing Copper Ores
placed on the production of a marketable cop-
The Soatability of gold from gold-containing copper} per}gold concentrate and optimization of gold recov-
gold ores depends on the nature and occurrence of gold ery is usually constrained by the marketability of
in these ores, and its association with iron sulRdes. its concentrate.
Table 3 Gold recovery obtained using different flow sheets

Flow sheet Recovery in concentrate (%) Tailing Assays (%, g t\1)
Au Ag Sb As S Au Ag Sb As S
Single-stage grind flotation 88.1 89.2 72.9 68.4 70.1 1.7 5.0 0.04 0.035 0.38
Two-stage grind flotation 92.2 91.8 93.4 78.7 81.2 1.0 4.1 0.015 0.022 0.27
Three-stage grind flotation 95.3 95.2 95.7 81.2 85.7 0.7 2.2 0.005 0.015 0.19
Reproduced from Sristinov (1964) with permission.
The minerals that inSuence gold recovery in these In the presence of soluble cations (e.g. Fe, Cu), addi-
ores are iron sulRdes (i.e. pyrite, marcasite), in which tions of small quantities of organic acid (e.g. oxalic, tar-
gold is usually associated as minute inclusions. Thus, taric) improve gold recovery in the copper concentrate.
the iron sulRde content of the ore determines gold Some porphyry copper ores contain naturally Soa-
recovery in the Rnal concentrate. Figure 3 shows the table gangue minerals, such as chlorites and alumino-
relationship between pyrite content of the ore and silicates, as well as preactivated quartz. Sodium
gold recovery in the copper concentrate for two differ- silicate, carboxymethylcellulose and dextrins are com-
ent ore types. Most of the gold losses occur in the pyrite. mon depressants used to control gangue Sotation.
The reagent schemes used in commercial opera- Gold recovery from massive sulRde copper}gold
tions treating porphyry copper}gold ores vary con- ores is usually much lower than that of porphyry
siderably. Some operations, where pyrite rejection is copper}gold ores, because very often a large portion
a problem, use a dithiophosphate collector at an of the gold is associated with pyrite. Normally, gold
alkaline pH between 9.0 and 11.8 (e.g. OK Tedi, recovery from these ores does not exceed 60%.
PNG Grasberg, Indonesia). When the pyrite content During the treatment of copper}gold ores containing
in the ore is low, xanthate and dithiophosphates are pyrrhotite and marcasite, the use of Na2H2PO4 at
used in a lime or soda ash environment. alkaline pHs depresses pyrrhotite and marcasite, and
In more recent years, in the development of com- also improves copper and gold metallurgy.
mercial processes for the recovery of gold from por-
Flotation of Oxide Copper+Gold Ores
phyry copper}gold ores, bulk Sotation of all the sul-
Rdes has been emphasized, followed by regrinding of Oxide copper}gold ores are usually accompanied by
the bulk concentrate and sequential Sotation of cop- iron hydroxide slimes and various clay minerals.
per}gold from pyrite. Such a Sow sheet (Figure 4) can There are several deposits of this ore type around the
also incorporate high intensity conditioning in the world, some of which are located in Australia (Red
cleaner}scavenger stage. Comparison of metallurgi- Dome), Brazil (Igarape Bahia) and the Soviet Union
cal results using the standard sequential Sotation Sow (Kalima). Treatment of these ores is difRcult,
sheet and the bulk Sotation Sow sheet is shown in and even more complicated in the presence of clay
Table 5. A considerable improvement in gold recov- minerals.
ery was achieved using the bulk Sotation Sow sheet. Recently, a new class of collectors, based on ester-
During beneRciation of clay-containing copper} modiRed xanthates, has been successfully used to
gold ores, the use of small quantities of Na2S (at treat gold-containing oxide copper ores, using a sul-
natural pH) improves both copper and gold metal- Rdization method. Table 6 compares the metallurgi-
lurgy considerably. cal results obtained on the Igarape Bahia ore using
Table 4 Effect of amine-modified CuSO4 on gold-bearing sulfide flotation from carbonaceous refractory ore
Reagent used Product Weight (%) Assays (%, g t\1) Distribution (%)
Au S Au S
CuSO4#xanthate Gold sulfide conc. 30.11 9.63 4.50 69.1 79.7

Gold sulfide tail 69.89 1.86 0.49 30.9 20.3
Head 100.00 4.20 1.70 100.0 100.0
Amine modified Gold sulfide conc. 26.30 13.2 5.80 84.7 90.8
CuSO4#xanthate Gold sulfide tail 73.70 0.85 0.21 15.3 9.2
Head 100.00 4.10 1.68 100.0 100.0
Stibnite Soats well in neutral and weak acid pH,

while in an alkaline pH (i.e. '8), it is reduced.
Utilizing this phenomenon, gold-bearing sulRdes
are Soated with xanthate and alcohol frother in
alkaline medium (pH'9.3) followed by stibnite
Sotation at about pH 6.0, after activation with lead
nitrate. Typical metallurgical results using this
method are shown in Table 7.
E Bulk Sotation followed by sequential Sotation of
gold-bearing sulRdes, and depression of stibnite.
This method was practised at the Bradly concentra-
tor (USA) and consisted of the following steps: Rrst,
bulk Sotation of stibnite and gold bearing sulRdes
at pH 6.5 using lead nitrate (Sb activator) and
xanthate; second, the bulk concentrate is reground
in the presence of NaOH (pH 10.5) and CuSO4,
and the gold-bearing sulRdes are reSoated with
additions of small quantities of xanthate; third,
Figure 3 Effect of pyrite content of the ore on gold recovery in cleaning of the gold concentrate in the presence of
the copper}gold concentrate at 30% Cu concentrate grade. 1, NaOH and NaHS. The plant metallurgical results
Ore from Peru; 2, ore from Indonesia. employing this method are shown in Table 8.
Recent studies conducted on ore from Kazakhstan
xanthate and a new collector (PM230, supplied by have shown that sequential Sotation using thionocar-
Senmin in South Africa). bamate collector gave better metallurgical results
The modiRer used in the Sotation of these ores than those obtained with xanthate.
included a mixture of sodium silicate and Calgon.
Good selectivity was also achieved using boiled starch. Flotation of Arsenical Gold Ores
There are two major groups of arsenical gold ores of
Flotation of Gold+Antimony Ores economic value. These are the massive base metal
sulRdes with arsenical gold (e.g. the lead}zinc Olym-
Gold}antimony ores usually contain stibnite (1.5}
pias deposit, Greece) and arsenical gold ores without
4.0% Sb), pyrite, arsenopyrite gold (1.5}3.0 g t\)
the presence of base metals. Massive, base metal
and silver (40}150 g t\). Several plants in the USA
arsenical gold ores are rare, and there are only a few
(Stibnite, Minnesota and Bradly) and Russia have
deposits in the world. A typical arsenical gold ore
been in operation for some time. There are two com-
contains arsenopyrite as the major arsenic mineral.
mercial processes available for treatment of these ores.
However, some arsenical gold ores, such as those
E Selective Sotation of gold-containing sulRdes fol- from Nevada in the USA (Getchel deposit), contain
lowed by Sotation of stibnite with pH change. realgar and orpiment as the major arsenic-bearing
Figure 4 Bulk flow sheet used in the treatment of pyritic copperIgold ores. Reproduced with permission from Bulatovic (1997).
Table 5 Comparison of metallurgical results using conventional and bulk flotation flow sheets on ore from Peru
Flow sheet used Product Weight (%) Assays (%, g t\1) Distribution (%)
Cu Au Cu Au
Conventional Cu/Au conc. 2.28 27.6 32.97 95.4 76.7

(sequential Cu/Au) Cu/Au tail 97.72 0.031 0.23 4.6 23.3
Head 100.00 0.66 0.98 100.0 100.0
Bulk Cu/Au conc. 2.32 27.1 36.94 95.2 85.8
(Figure 4) Cu/Au tail 97.68 0.032 0.14 4.8 14.2
Head 100.00 0.66 0.96 100.0 100.0
minerals. Pyrite, if present in an arsenical gold ore, arsenopyrite separation. In this particular case, most
may contain some gold as minute inclusions. of the gold was associated with pyrite. Successful
Flotation of arsenical gold ores associated with pyrite}arsenopyrite separation can also be achieved
base metals is accomplished using a sequential Sota- with the use of potassium peroxydisulRde as the
tion technique, with Sotation of base metals followed arsenopyrite depressant.
by Sotation of gold-containing pyrite}arsenopyrite. The second method involves depression of pyrite
The pyrite}arsenopyrite is Soated at a weakly acid and Sotation of arsenopyrite. In this method, the bulk
pH with a xanthate collector. concentrate is treated with high dosages of lime
Arsenical gold ores that do not contain signiRcant (pH'12), followed by a conditioning step with
base metals are treated using a bulk Sotation method, CuSO4 to activate arsenopyrite. The arsenopyrite is
where all the sulRdes are Rrst recovered into a bulk then Soated using a thionocarbamate collector.
concentrate. In case the gold is contained either in Separation of arsenopyrite and pyrite is important
pyrite or arsenopyrite, separation of pyrite and ar- from the point of view of reducing downstream
senopyrite is practised. There are two commercial processing costs. Normally, roasting or pressure oxi-
methods available. The Rrst method utilizes ar- dation followed by cyanidation is used to recover
senopyrite depression and pyrite Sotation, and con- gold.
sists of the following steps:
Flotation of Gold from Base Metal Sul\de Ores
E Heat the bulk concentrate to 753C at a pH of 4.5
Very often lead}zinc, copper}zinc, copper}lead}zinc
(controlled by H2SO4) in the presence of small
and copper}nickel ores contain signiRcant quantities
quantities of potassium permanganate or disodium
of gold (i.e. between 1 and 9 g/t). The gold in these
phosphate. The temperature is maintained for
ore types is usually found as elemental gold. A large
about 10 min.
portion of the gold in these ores is Rnely disseminated
E Flotation of pyrite using either ethylxanthate or
in pyrite, which is considered nonrecoverable. Be-
potassium butylxanthate as collector. Glycol
cause of the importance of producing commercial-
frother is also usually employed in this separation.
grade copper, lead and zinc concentrates, little or no
This method is highly sensitive to temperature. consideration is given to improvement in gold recov-
Figure 5 shows the effect of temperature on pyrite} ery, although the possibility exists to optimize gold
Table 6 Effect of collector PM230 on copper}gold recovery from Igarape Bahia oxide copper}gold ore
Reagent used Product Weight (%) Assays (%, g t\1) Distribution (%)
Cu Au Cu Au
Na2S"2500 g t\ Copper Cl conc. 9.36 33.3 14.15 67.0 50.0

PAX"200 g t\ Copper tail 90.64 1.61 1.46 33.0 50.0
Feed 100.00 4.65 2.65 100.0 100.0
Na2S"2500 g t\ Copper Cl conc. 10.20 39.5 21.79 88.0 85.5

PAX/PM230 Copper tail 89.80 0.61 0.42 12.0 14.5
(1 : 1)"200 g t\ Feed 100.00 0.61 0.42 12.0 14.5
PAX, potassium amylxanthate.

Reproduced from Bulatovic (1997) with permission.
Table 7 Metallurgical results obtained using a sequential can be used. Depressant combinations such as
flotation method Na2S#NaCN, or Na2SO3#NaCN, may be used.
The type of collector also plays an important role in
Product Weight Assays (%, g t\1) Distribution (%)
(%) gold Sotation of lead}zinc ores. A phosphine-based
Au Ag Sb Au Ag Sb collector, in combination with xanthate, gave better
gold recovery than dithiophosphates.
Gold 2.34 42.3 269.3 20.0 53 13 15
concentrate
Copper}zinc gold-containing ores Gold recovery
Stibnite 4.04 6.2 559.8 51.0 13 51 64
concentrate from copper}zinc ores is usually higher than that
Tailing 93.62 0.65 18.7 0.7 34 36 21 obtained from either a lead}zinc or copper}lead}zinc
ore. This is attributed to two main factors. When
Feed 100.00 1.86 46.4 3.2 100 100 100 selecting a reagent scheme for treatment of cop-
per}zinc ores, there are more choices than for the
other ore types, which can lead to the selection of
recovery in many cases. Normally, gold recovery a reagent scheme which is more favourable for gold
from base metal ores ranges from 30 to 75%. Sotation. In addition, a noncyanide depressant sys-
In the case of a copper}zinc and copper}lead}zinc tem can be used for the treatment of these ores, which
ore, gold collects in the copper concentrate. During in turn results in improved gold recovery. This option
the treatment of lead}zinc ores, the gold tends to is not available during treatment of lead}zinc ores.
report to the lead concentrate. Information regarding Table 9 shows the effect of different depressant com-
gold recovery from base metal ores is sparse. binations on gold recovery from a copper}zinc ore.
The most recent studies conducted on various base The use of a noncyanide depressant system resulted
metal ores revealed some important features of Sota- in a substantial improvement in gold recovery in the
tion behaviour of gold from these ores. It has been copper concentrate.
demonstrated that gold recovery to the base metal
concentrate can be substantially improved with the Gold-containing copper}lead}zinc ores Because of
proper selection of reagent schemes. Some of these the complex nature of these ores, and the requirement
studies are discussed below. for a relatively complex reagent scheme for treatment
of this ore, the gold recovery is generally lower than
Gold-containing lead}zinc ores Some of these ores that achieved from a lead}zinc or copper}zinc ore.
contain signiRcant quantities of gold, ranging from One of the major problems associated with the Sota-
0.9 to 6.0 g t\ (e.g. Grum, Yukon, Canada; Greens tion of gold from these ores is related to gold min-
Creek, Alaska and Milpo, Peru). The gold recovery eralogy. A large portion of the gold is usually con-
from these ores ranges from 35 to 75%. Laboratory tained in pyrite, at sub-micron size. If coarser elemen-
studies have shown that the use of high dosages of tal gold and electrum are present, the gold surfaces
zinc sulfate, which is a common zinc depressant used are often coated with iron or lead, which can result in
in lead Sotation, reduces gold Soatability signiR- a substantial reduction in Soatability.
cantly. The effect of ZnSO4;7H2O addition on gold The type of collector and Sow sheet conRguration
recovery in the lead concentrate is illustrated in play an important role in gold recovery from these
Figure 6. ores. With a Sow sheet that uses bulk copper}lead
In order to improve gold recovery in the lead con- Sotation followed by copper}lead separation, the
centrate, an alternative depressant to ZnSO4;7H2O gold recovery is higher than that achieved with a
Table 8 Plant metallurgical results obtained using a bulk flotation method
Product Weight (%) Assays (%, g t\1) Distribution (%)
Au Ag Sb Au Ag Sb
Gold concentrate 1.80 91.1 248.8 1.5 61.0 31.3 2.0

Antimony concentrate 1.80 13.0 684.2 51.3 9.0 58.6 75.0
Middlings 0.50 46.6 248.8 20.0 8.6 6.0 8.0
Bulk concentrate 4.10 51.7 440.0 29.0 78.6 85.9 85.0
Tailing 95.90 0.6 3.1 0.2 21.4 14.1 15.0
Feed 100.00 2.7 21.0 1.3 100.0 100.0 100.0

Figure 5 Effect of temperature on separation of pyrite and

arsenopyrite from a bulk pyrite}arsenopyrite concentrate. Figure 6 Effect of ZnSO4 additions on gold recovery from
lead}zinc ores. Circles, Greens Creek ore (Alaska); triangles,
Grum ore Yukon (Canada).
sequential copper}lead Sotation Sow sheet. In labora-
tory tests, an aerophine collector type, in combina- recovery processes (i.e. pyrometallurgy, hydro-
tion with xanthate, had a positive effect on gold metallurgy) or for recovery of gold to base metal
recovery as compared to either dithiophosphate or concentrates, is a very important method for con-
thionocarbamate collectors. Table 10 compares the centrating the gold and reducing downstream
metallurgical results obtained with an aerophine costs.
collector to those obtained with a dithiophosphate E The Sotation of elemental gold, electrum and tellu-
collector. rides is usually very efRcient, except when these
Because of the complex nature of gold-containing minerals are Soated from base metal massive sul-
copper}lead}zinc ores, the reagent schemes used are Rdes.
also complex. Reagent modiRers such as ZnSO4, E Flotation of gold-bearing sulRdes from ores con-
NaCN and lime have to be used, all of which have taining base metal sulRdes presents many chal-
a negative effect on gold Sotation. lenges and should be viewed as Sotation of the
particular mineral that contains gold (i.e. pyrite,
arsenopyrite, copper), because gold is usually asso-
Conclusions ciated with these minerals at micron size.
E The Sotation of gold-bearing ores, whether for E Selection of a Sotation technique for gold precon-
production of bulk concentrates for further gold centration depends very much on the ore
Table 9 Effect of different depressant combinations on gold recovery to the copper concentrate from lower zone Kutcho Creek Ore
Depressant system Product Weight (%) Assays (%, g t\1) Distribution (%)
Au Cu Zn Au Cu Zn
ZnSO4, NaCN, CaO Cu concentrate 3.10 20.4 26.2 3.30 45.1 85.6 2.8
pH 8.5 Cu, 10.5 Zn Zn concentrate 5.34 1.20 0.61 55.4 4.6 3.4 82.2
Tailings 91.56 0.77 0.11 0.58 50.3 11.0 15.0
Feed 100.00 1.4 0.95 3.60 100.0 100.0 100.0
Na2SO3, NaHS, CaO Cu concentrate 3.05 32.5 28.1 2.80 68.3 87.4 2.3
pH 8.5 Cu, 10.5 Zn Zn concentrate 5.65 1.20 0.55 54.8 4.7 3.2 84.6
Tailings 91.30 0.43 0.10 0.52 27.0 9.4 13.1
Feed 100.00 1.45 0.98 3.66 100.0 100.0 100.0
III / GRADIENT POLYMER CHROMATOGRAPHY: LIQUID CHROMATOGRAPHY 2975
Table 10 Effect of collector type on Cu-Pb-Zn-Au metallurgical results from a high lead ore
Collector used Product Weight (%) Assays (%, g t\1) Distribution (%)
Au Cu Pb Zn Au Cu Pb Zn
Xanthate"30 g/t Cu concentrate 2.47 22.4 25.5 1.20 4.50 41.6 78.6 2.3 1.3
Dithiophosphate Pb concentrate 1.80 2.50 0.80 51.5 8.30 3.4 1.8 71.3 1.7
3477"20 g/t Zn concentrate 13.94 1.10 0.60 0.80 58.2 11.5 10.4 8.6 92.2
Tailing 81.79 0.71 0.089 0.28 0.52 43.5 9.2 17.8 4.8
Feed 100.00 1.33 0.80 1.30 8.80 100.0 100.0 100.0 100.0
Aerophine 3418A"20 g/t Pb concentrate 1.92 2.80 0.90 51.1 9.20 4.1 2.1 72.5 2.0
Zn concentrate 13.91 0.90 0.50 0.72 58.5 9.6 8.5 7.4 92.5
Tailing 81.65 0.41 0.093 0.30 0.44 25.7 9.3 18.0 4.1
Feed 100.00 1.30 0.82 1.35 8.80 100.0 100.0 100.0 100.0
mineralogy, gangue composition and gold particle Bulatovic SM (1993) Evaluation of new HD collectors in
size. There is no universal method for Sotation of Sotation of pyritic copper}gold ores from BC Canada.
the gold-bearing minerals, and the process is LR-029. Interim R&D report.
tailored to the ore characteristics. A speciRc reagent Bulatovic SM (1996) An investigation of gold Sotation
scheme and Sow sheet are required for each ore. from base metal lead}zinc and copper}zinc ores. Interim
Report LR-049.
E There are opportunities on most operating plants
Bulatovic SM (1997) An investigation of the recovery of
for improving gold metallurgy. Most of these im- copper and gold from Igarape Bahia oxide copper}gold
provements come from selection of more effective ores. Report of Investigation LR-4533.
reagent schemes, including collectors and modi- Bulatovic SM and Wyslouzil DM (1996) Flotation behav-
Rers. iour of gold during processing of porphyry copper}gold
E Perhaps the most difRcult ores to treat are the clay- ores and refractory gold-bearing sulphides. Second In-
containing carbonaceous sulRdes. SigniRcant pro- ternational Gold Symposium. Lima, Peru.
gress has been made in treatment options for these Fishman MA and Zelenov BI (1967) Practice in treatment
ores. New sulRde activators (e.g. amine-treated of sulphides and precious metal ores. Izdatelstro Nedra
CuSO4, ammonium salts) and nitrogen gas Sota- (in Russian) 5: 22}101.
tion are amongst the new methods available. Kudryk V, Carigan DA and Liang WW (1982) Precious
Metals. Mining Extraction and Processing, AIME.
Martins V, Dunne RC and GelR P (1991) Treatment of
Further Reading Partially Refractory Gold Ores. Perth, Australia:
Randol Gold Forum.
Baum W (1990) Mineralogy as a Metallurgical Tool in Sristinov NB (1964) The effect of the use of stage grinding
Refractory Ore, Progress Selection and Optimization. in processing of refractory clay-containing gold ore.
Squaw Valley, Salt Lake City: Randol Gold Forum. Kolima 1: 34}40.
GRADIENT POLYMER CHROMATOGRAPHY:

G. GloK ckner, Dresden University of Technology, With low molecular weight substances this condition
Dresden, Germany is easily fulRlled, because the entropy contribution is
Copyright ^ 2000 Academic Press large owing to the large number of particles involved.
But with polymer compounds, the entropy of dissolu-
tion is comparatively small and the enthalpy contri-
Classical Precipitation Chromatography bution gains in importance. The precept that ‘similia
similibus solventur’ becomes a stringent requirement;
Polymer Solubility and Precipitation
in terms of Hildebrand’s solubility concept, this
Solubility is governed by the general requirement that means that a polymer can dissolve only in Suids
the change in Gibbs’ free energy must be negative. whose solubility parameters are very closely related
Table 10 Effect of collector type on Cu-Pb-Zn-Au metallurgical results from a high lead ore
Collector used Product Weight (%) Assays (%, g t\1) Distribution (%)
Au Cu Pb Zn Au Cu Pb Zn
Dithiophosphate Pb concentrate 1.80 2.50 0.80 51.5 8.30 3.4 1.8 71.3 1.7
3477"20 g/t Zn concentrate 13.94 1.10 0.60 0.80 58.2 11.5 10.4 8.6 92.2
Tailing 81.79 0.71 0.089 0.28 0.52 43.5 9.2 17.8 4.8
Feed 100.00 1.33 0.80 1.30 8.80 100.0 100.0 100.0 100.0
Aerophine 3418A"20 g/t Pb concentrate 1.92 2.80 0.90 51.1 9.20 4.1 2.1 72.5 2.0
Zn concentrate 13.91 0.90 0.50 0.72 58.5 9.6 8.5 7.4 92.5
Tailing 81.65 0.41 0.093 0.30 0.44 25.7 9.3 18.0 4.1
Feed 100.00 1.30 0.82 1.35 8.80 100.0 100.0 100.0 100.0
mineralogy, gangue composition and gold particle Bulatovic SM (1993) Evaluation of new HD collectors in
size. There is no universal method for Sotation of Sotation of pyritic copper}gold ores from BC Canada.
the gold-bearing minerals, and the process is LR-029. Interim R&D report.
tailored to the ore characteristics. A speciRc reagent Bulatovic SM (1996) An investigation of gold Sotation
scheme and Sow sheet are required for each ore. from base metal lead}zinc and copper}zinc ores. Interim
Report LR-049.
E There are opportunities on most operating plants
Bulatovic SM (1997) An investigation of the recovery of
for improving gold metallurgy. Most of these im- copper and gold from Igarape Bahia oxide copper}gold
provements come from selection of more effective ores. Report of Investigation LR-4533.
reagent schemes, including collectors and modi- Bulatovic SM and Wyslouzil DM (1996) Flotation behav-
Rers. iour of gold during processing of porphyry copper}gold
E Perhaps the most difRcult ores to treat are the clay- ores and refractory gold-bearing sulphides. Second In-
containing carbonaceous sulRdes. SigniRcant pro- ternational Gold Symposium. Lima, Peru.
gress has been made in treatment options for these Fishman MA and Zelenov BI (1967) Practice in treatment
ores. New sulRde activators (e.g. amine-treated of sulphides and precious metal ores. Izdatelstro Nedra
CuSO4, ammonium salts) and nitrogen gas Sota- (in Russian) 5: 22}101.
tion are amongst the new methods available. Kudryk V, Carigan DA and Liang WW (1982) Precious
Metals. Mining Extraction and Processing, AIME.
Martins V, Dunne RC and GelR P (1991) Treatment of
Further Reading Partially Refractory Gold Ores. Perth, Australia:
Randol Gold Forum.
Baum W (1990) Mineralogy as a Metallurgical Tool in Sristinov NB (1964) The effect of the use of stage grinding
Refractory Ore, Progress Selection and Optimization. in processing of refractory clay-containing gold ore.
Squaw Valley, Salt Lake City: Randol Gold Forum. Kolima 1: 34}40.
GRADIENT POLYMER CHROMATOGRAPHY:

G. GloK ckner, Dresden University of Technology, With low molecular weight substances this condition
Dresden, Germany is easily fulRlled, because the entropy contribution is
Copyright ^ 2000 Academic Press large owing to the large number of particles involved.
But with polymer compounds, the entropy of dissolu-
tion is comparatively small and the enthalpy contri-
Classical Precipitation Chromatography bution gains in importance. The precept that ‘similia
similibus solventur’ becomes a stringent requirement;
Polymer Solubility and Precipitation
in terms of Hildebrand’s solubility concept, this
Solubility is governed by the general requirement that means that a polymer can dissolve only in Suids
the change in Gibbs’ free energy must be negative. whose solubility parameters are very closely related
2976 III / GRADIENT POLYMER CHROMATOGRAPHY: LIQUID CHROMATOGRAPHY
to those of the polymer. Therefore, most liquids are size 0.1 mm diameter. The sample size was 300 mg,
non-solvents and the number of solvents available for the gradient ran from ethanol (non-solvent) to methyl
a given polymer is far fewer than the number of ethyl ketone, and the temperature gradient spanned
solvents available for a low molecular weight sub- 10}603C. The multistage mechanism ensured a high
stance of comparable structure. separation power, which was conRrmed by both the-
The solubility of polymers decreases with increas- ory and experiment. The method became popular in
ing molecular weight (MW) and can be measured polymer characterization; the second citation in the
easily by the controlled addition of a non-solvent to Bibliography provides a survey of application of the
the solution of a polymer. The volume fraction NS of method to about 30 different polymers.
non-solvent at the cloud point is related to the square The development of size exclusion chromatogra-
root, M0.5, of molecular weight by phy (SEC) made separation according to MW feasible
and convenient. SEC allows the investigation of dif-
100 "C1#C2/M0.5 ferent polymers in a common eluent with very little
NS
preliminary work. Dissolved samples can be injected
into a running eluent, e.g. tetrahydrofuran, which has
where C1 and C2 are constants for the particular
sufRcient solvent strength for a great many poly-
system.
mers. The elution curve can be monitored with a suit-
This dependence can be used to separate polymers
able detector and provides at least a Rrst guess at the
by either fractional precipitation or dissolution. The
MW distribution (MWD). Using MW-sensitive de-
latter method can also be performed in packed col-
tectors and sophisticated software, reliable MWD
umns by gradients whose solvent power increases in
curves can be measured within minutes. Interest in
the course of the elution.
the demanding Baker}Williams technique therefore
The solubility of polymers also depends on temper-
faded away. Although this technique is no longer
ature. Usually, the temperature coefRcient is pos-
a competitor in analytical separations according to
itive, i.e. fractional dissolution can be carried out
MW, it should still be considered a powerful tool for
with a given solvent (or a non-solvent/solvent mix-
separations according to chemical differences
ture at constant composition) by raising the temper-
and for preparative fractionation. The chemical
ature. This procedure can also be performed in
composition distribution of copolymers, blends
columns.
or modiRed polymers can be measured by SEC only
in rare cases (if coupled with MWD in a known
Baker^Williams Fractionation
ratio). This was realized some years ago (see Biblio-
In 1956, Baker and Williams described ‘a new graphy).
chromatographic procedure and its application to
high polymers’. This was column elution combining High Performance Precipitation Liquid
the effects of solvent strength and temperature. Chromatography (HPPLC)
The important innovation was a temperature gradi-
ent along the column. The top of the column was
Principle and Instrumentation
heated to a temperature about 50 K higher than that
of the cooled bottom. An aluminium jacket ensured The renaissance of precipitation chromatography
a linear temperature proRle. The polymer to be inves- requires modern equipment, e.g. detectors and pro-
tigated was coated onto the part of the inert packing grammable gradient devices. The samples to be inves-
that subsequently was put into the uniformly heated tigated should be applied in solution and injected into
uppermost section of the column. The temperature the eluent stream ahead of the column.
gradient enabled multistage separation to be per- About 80% of all high performance liquid
formed. Any component dissolved from the sample chromatography (HPLC) investigations are per-
bed was reprecipitated in a cooler zone of the column. formed in the reversed-phase mode. Reversed-phase
Here it was redissolved later by a non-solvent/solvent packing materials have a nonpolar surface. They usu-
mixture of higher solvent strength and transported to ally consist of particles with a silica core and
the next cooler zone for another reprecipitation. a bonded layer of alkane chains. Reversed-phase
Thus, Baker}Williams fractionation was described as gradients run from a highly polar initial eluent to
‘a chromatographic method based upon the equilibra- a Rnal eluent of low polarity. The polar eluent forces
tion of substances between a stationary precipitated nonpolar solutes to be retained by the stationary
phase and a moving solution’. Baker and Williams phase. Retention increases with decreasing polarity of
investigated polystyrene in a glass tube, 350 mm long the sample components. The mechanism is under-
and 24 mm wide, packed with glass beads of average stood to be a solvophobic interaction that requires the
mobile phase to be an unfavourable environment for fering in polarity, which is of basic interest in the
the solute. framework of chromatography. Styrene}acrylonitrile
The measures taken to force the polymer towards copolymers therefore seemed to be well suited to
the stationary phase may easily reach or even trans- early studies of high performance precipitation
gress the limits of solubility. The latter effect has chromatography.
been observed occasionally in reversed-phase Preliminary studies published in 1982 showed that
chromatography of low molecular weight com- tetrahydrofuran (THF) has the capacity to separate
pounds, but is almost the rule with polymers whose a mixed styrene}acrylonitrile sample into its constitu-
solubility is more restricted. ents, provided that the starting gradient component
In normal-phase chromatography, the column is A enables proper retention of the injected samples.
polar and gradient elution is performed with a non- This was achieved by using at least 80% n-hexane in
polar starting component A and a polar component THF, i.e. with a non-solvent. The injected polymer
B is added during the run. Retention increases with was therefore precipitated at the top of the column.
increasing polarity of the sample constituents. The elution characteristic (percentage THF in the
In order to achieve proper retention of a polymer, eluent versus acrylonitrile content of the sample) was
the starting eluent A must usually be a non-solvent. similar to the solubility borderline determined by
This means that sample solutions cannot be prepared turbidimetric titration. It was found that equivalent
in a portion of the starting eluent and that the poly- separations could be achieved on a silica column as
mer is precipitated at the top of the column. Since well as on a nonpolar C8 column. This surprising
proper retention is required, the separation is by this result was conRrmed in systematic studies performed
step classiRed as precipitation chromatography. The by GloK ckner and van den Berg in 1987 using other
precipitation at the top of the column yields precon- polar and nonpolar columns including silica CN
centration of the sample. Thus, HPPLC can cope with bonded phase, small-pore C18, wide-pore C18 and
samples differing widely in concentration, e.g. -Bondagel E1000-10. Thus, the surface of the pack-
SEC fractions. The column permeability is not af- ings did not actively participate in the separation. It
fected. If the sample solvent is a portion of eluent B, was found that the peak shapes obtained could not be
the amount of solvent injected will not cause dif- improved further by a temperature gradient along
Rculties. The use of another solvent is not recommen- the column (which had been so essential in Baker}
ded because it could overload the column with an Williams fractionation).
additional substance. Multistage separation without the use of a tempera-
The mechanism of separation is, in general, a com- ture gradient or interaction with the surface can be
bination of precipitation and adsorption. The de- achieved on porous packings where the polymer sol-
tector must be capable of measuring the eluting ute is excluded from the pores. The polymer solute
sample components without being affected by then has a higher linear velocity than the eluent,
the solvent gradient. Suitable equipment became which Rlls the interstitial volume as well as the pore
available in the late 1970s. volume of the column. The polymer bypasses the
pores and thus overtakes the eluent which has suf-
High Performance Precipitation Liquid Rcient solvent strength. The polymer is precipitated
Chromatography of Styrene}Acrylonitrile
and retained until a more powerful eluent reaches its
Copolymers
position.
Styrene is a polymerizable substance of formula In chromatographic terms, the gradient
CH2"CH(C6H5), whose homopolymerization yields hexanePTHF is a normal-phase gradient, i.e. in-
polystyrene (PS). It can be polymerized with numer- creasing in polarity. In combination with a polar
ous other monomers to yield copolymers. Styrene column, e.g. silica or a CN bonded phase, it forms
units have a strong UV absorption, which means that a standard normal-phase system, which should elute
polystyrene and styrene-containing copolymers can more polar sample constituents after less-polar ones.
be monitored by UV detectors. Copolymers of styrene Thus, the observed efRciency of irregular combi-
and acrylonitrile are of commercial interest. Well- nations with nonpolar C8 or C18 columns shows that
characterized samples graded in composition are the separation was not governed by the common
available together with a considerable knowledge of polarity rules of chromatography. The separation of
styrene}acrylonitrile dissolution/precipitation behav- styrene}acrylonitrile copolymers was, under the con-
iour. The polarity of acrylonitrile is higher than that ditions of these studies, dominated by a precipitation
of styrene units. Therefore a separation of mechanism. Another example of precipitation mech-
styrene}acrylonitrile copolymers according to com- anism in styrene}acrylonitrile gradient chromatogra-
position is also a separation into constituents dif- phy is given in the next section. However, it should be
Rrmly stated that styrene}acrylonitrile is an exception

rather than the rule. In general, gradient chromato-
graphy of synthetic polymers is governed by the com-
bination of precipitation and adsorption. Irregular
phase combination will not often work, but they do
with styrene}acrylonitrile.
Normal- and reversed-phase chromatography are
like mirror images. It was a challenge to Rnd out
whether or not a given synthetic copolymer could be
separated by both mechanisms. The Rrst positive re-
port appeared in 1987 when copolymers of styrene
and ethyl methacrylate were measured by both modes
of chromatography. All previous related work was by Figure 1 Merged plot of elution curves of seven styrene}acrylo-
normal-phase separation. As expected, the elution nitrile copolymers on a column (250 mm;7.1 mm i.d.) packed
order achieved by reversed-phase chromatography with polystyrene gel. Gradient: methanolPtetrahydrofuran,
0}100% B in 25 min: UV signal detected at 254 nm. The acrylonit-
was the opposite of that in normal-phase chromatog- rile (AN) content of the samples is indicated on the curves; the
raphy. Since then, several polymer systems have been amount of each injected was 30 g. (Reproduced from GloK ckner
separated by normal-phase and reversed-phase et al., 1991, by courtesy of Vieweg-Verlag.)
chromatography with inversion of elution order, e.g.
styrene}methyl methacrylate copolymers, styrene}
methyl acrylate copolymers or methacrylate
homopolymers graded in polarity of the ester group. labelled 26.1% acrylonitrile, but is was eluted be-
tween the samples 17.6% and 12.5% acrylonitrile.
High Performance Precipitation Liquid This puzzling observation can be understood with the
Chromatography of StyreneÂcrylonitrile
help of the solubility diagram for styrene}acrylonit-
Copolymers with Inversion of Elution Order
rile in THF/MeOH, which is shown in Figure 2. The
The separation of styrene}acrylonitrile copolymers solubility boundary has a maximum at 20}25%
with an elution order as in reversed-phase acrylonitrile content, where samples require only
chromatography was achieved on a column packed about 45 vol% THF in MeOH for dissolution,
with polystyrene gel. Eluent A was methanol whereas copolymers with more or less acrylonitrile
(MeOH), a polar non-solvent for styrene}acrylonit- need up to 12% more THF.
rile samples; the less polar eluent, B, was either THF Along the left-hand branch of the solubility bound-
or dichloromethane. In both cases, the gradient rate ary (0}20% acrylonitrile), both polarity and solubil-
was 0}100% B in 25 min. Copolymers with acrylo- ity decrease with decreasing acrylonitrile content.
nitrile content between 2.3% and 27.3% were re- The sequence of the Rve late-eluting peaks in Figure 1
tained longer the less acrylonitrile they contained is supported by polarity and solubility. Beyond the
(see Figure 1). Although the phase system and the point of inSection, polarity increases but solubility
elution order conformed to the rules of reversed- decreases with increasing acrylonitrile content.
phase chromatography, the solubility mechanism A sample with 36.2% acrylonitrile requires about
prevailed. 50% THF for dissolution but should, according to
The samples were prepared by copolymerization to polarity, already be released from the column in
only about 5% conversion, but they still consisted of a mixture of 35% THF in MeOH. The measured
macromolecules differing in composition. The peak position between the peaks for 17.6% and
chemical composition distributions of the samples are 12.5% acrylonitrile is determined by solubility. This
essentially responsible for the shape of the elution is another indication of precipitation prevailing
curves. The chemical composition distributions of in the chromatography of styrene}acrylonitrile
samples with, say, 8.6% or 17.6% average acrylonit- copolymers.
rile content are obviously narrower than that of Along the left-hand branch of the solubility bound-
a sample with 12.5% acrylonitrile. ary, polarity supports the effect of solubility but
The shape of the elution curve for 36.2% acrylonit- beyond the turning point the two effects counter-
rile in Figure 1 looks rather odd. In addition, the act each other. This is the reason why the elution
position of its maximum is not where it might be curve for 36.2% acrylonitrile is much broader than
expected. According to its high acrylonitrile content, the others. With normal-phase gradients, the 36.2%
the sample is the most polar of the series investigated. sample yielded an elution curve of the usual narrow
It should therefore be eluted before the copolymer shape, even in irregular phase systems (see Figure 3).
system. If solubility prevails (and both temperature

and concentration are suitable), both curves coincide.
Noticeable adsorption shifts the elution characteristic
above the solubility boundary, i.e. a higher concen-
tration of solvent B is necessary for eluting a given
sample than for dissolving it. The least adsorption
was observed in reversed-phase systems with polysty-
rene samples. The predominance of an adsorption
mechanism causes a retention behaviour differ-
ent from (or even opposite to) that observed with
a precipitation mechanism (see Table 1).
Baker and Williams reported that classical precipi-
tation chromatography can be performed with ‘a col-
Figure 2 Solubility boundary for styrene-acrylonitrile umn of inert material2 providing that the polymer
copolymers in THF/MeOH as measured by turbidimetric titration gel does not Sow through the column’. This type of
at 203C using THF as a sample solvent and methanol as the
precipitating non-solvent. Phase separation (precipitation) occurs
Sow can also occur in high performance precipitation
on crossing the curve from the upper part of the diagram (homo- chromatography in which case an optical detector
geneous solutions) to the lower part. (Reproduced from GloK ckner may register the strong signal characteristic of a tur-
et al., 1991, by courtesy of Vieweg-Verlag.) bid liquid. Such a signal is affected by many
parameters including time and is therefore poorly
reproducible. Gel breakthrough can be avoided if
there is some contribution of adsorption to retention.
Sudden-Transition Gradient Separation according to composition with the least
Chromatography of Synthetic superimposition of molecular weight effects re-
Polymers quires adsorption to dominate, i.e. elution by chang-
ing polarity of the eluent rather than by solvent
Interaction of Precipitation and Adsorption strength. On the other hand, the unfavourable ef-
in Polymer Gradient Chromatography
fect of sample size owing to strong adsorption can be
Chromatographic retention and elution of synthetic compensated by increasing the contribution of pre-
polymers is generally governed by precipitation/dis- cipitation to retention.
solution and adsorption/desorption. The contribu-
tion of adsorption can be judged by comparing the
Independent Control of Adsorption
solubility and elution characteristics of the sample
and Precipitation
In common binary gradients, the solvent power and
polarity of the mobile phase change simultaneously in
the course of the run. An optimum can be sought by
using a variety of different eluents A and B and
their combinations. However, this is a cumbersome
procedure requiring an adequate supply of chemicals
and a prolonged time. In addition, it may not be
successful because thermodynamic reasons restrict
the number of possible solvents for a given polymer,
and several of these may be further ruled out by
physical, physiological, or Rnancial reasons. More
promising and efRcient is the use of ternary sys-
tems consisting of two non-solvents (A and B) and
Figure 3 Merged plot of elution curves of six styrene}acrylonit- one solvent (C) for the polymers under investigation.
rile copolymers on a reversed-phase column (250 mm;4.3 mm A and B must be opposite in polarity, i.e. if A is
i.d.) packed with C18 bonded phase; in irregular combination with a polar non-solvent, B is a nonpolar one. The polarity
a normal-phase gradient n-heptaneP(THF#20% methanol), of solvent C is in between those of A and B. Solvent
0}100% B in 25 min. Detection by signal from an evaporative
C must be miscible with A and B and must have
light-scattering detector. The acrylonitrile (AN) content of the
samples is indicated on the curves; the amount of each injected sufRcient strength to dissolve samples in the
was 30 g. (Reproduced from GloK ckner et al., 1991, by courtesy whole range of molecular weight and composition
of Vieweg-Verlag.) under investigation.
Table 1 Features of polymer gradient chromatography with predominance of either precipitation/dissolution or adsorption/desorption
Dominating Elution characteristic Irregular phase Increasing temperature Increasing sample Increasing
mechanism combinations size (overload) molecular weight a
Percipitation Coincident with the Separating like Decreases retention Increases retention Increases,
solubility boundary standard ones retention,
C2"2}4;103
Adsorption Above the Ineffective in Increases the retention Decreases retention Increases
solubility boundary separation of polymers retention slightly,
C2"2}5;102
a
For C2 factors see the equation in the text. Values of C2 are compiled in GloK ckner G (1991) Gradient HPLC of Copolymers and
Chromatographic Cross-Fractionation, p. 107. New York: Springer-Verlag.
Since with gradients of this kind the chromato- for baseline separation. The best result was obtained
graphically signiRcant process is the result of inter- after addition of 30% THF.
actions between non-solvents there is, owing to the The advantage of this technique in comparison
large variety of the latter, more freedom in adjusting with binary gradient elution is obvious (see Figure 5).
optimum conditions than with binary non-sol- The chromatogram in Figure 5 was obtained in the
vent/solvent gradients. same laboratory as those of Figure 4 with the same
The samples to be investigated are dissolved in instrument and identical solvents. The baseline shift
solvent C and injected into a starting eluent (e.g. A), in Figure 5 is due to the UV absorption of THF
whose polarity and precipitating power ensure proper which, at 259 nm, is slightly higher than that
retention at the top of the column. Solvent C is then of iso-octane. This causes the baseline rise with
added to the eluent at a concentration that in itself
does not sufRce for elution. In order to achieve
short chromatograms, the concentration of C is
changed as rapidly as the apparatus allows. No un-
favourable side effects of the shock caused by the
sudden transition from injection to elution conditions
have ever been observed. The disturbance is visible
with the help of optical detectors. With cyanopropyl
or C18 packings, it is swept through by the approxim-
ately three-fold volume of mobile phase in the col-
umn. The elution of the sample is then triggered by
a gradient APB at a constant level of solvent C.
The Rrst results of gradient elution with sudden
transition of solvent concentration were achieved in
the normal-phase mode of chromatography. The col-
umn packing was polar (CN-modiRed silica), A was
iso-octane (with addition of 2% MeOH to the start-
ing eluent), B was MeOH and C was THF. The
gradient APB was performed at 5% min\1 and ap-
plied to copolymers of styrene and ethyl methacrylate
(EMA), methyl methacrylate (MMA), or 2-
methoxyethyl methacrylate.
Figure 4 is the merged plot of UV signals measured
Figure 4 Separation of the mixture of five styrene}EMA
on the elution of a mixture of Rve styrene}EMA copolymers at 503C on a column (60 mm;4 mm i.d.) packed with
copolymers through a gradient iso-octanePMeOH cyanopropyl bonded phase. Gradient: iso-octanePmethanol
after sudden transition to 20, 25, 30 or 35% THF (5% min\1) after increase of THF concentration from zero to the
solvent. Both iso-octane and MeOH are non-solvents percentage indicated at the curves; flow rate 0.5 mL min\1.
Sample: 1.8 g copolymer A (4.7% EMA)#1.2 g C (32.2%
for styrene}EMA. The addition of 35% THF yielded
EMA)#2.0 g E (54.6% EMA)#1.2 g G (68.0%
too high a solubility: the sample with 4.7% EMA was EMA)#2.0 g I (92.5% EMA); UV signal detected at 230 nm.
swept through the column by the sample solvent. (Reproduced from GloK ckner, 1991, by courtesy of Springer-
A proportion of 20 or 25% THF did not sufRce Verlag.)
increasing THF content of the eluent. The effect

would be still more dramatic at a shorter wavelength,
e.g. at 230 nm. Figure 4 presents horizontal baselines
although the chromatograms were monitored at
230 nm. This is due to the constant concentration of
THF throughout the elution, which disturbs the
traces much less than a changing amount of THF
does. The higher the THF addition, the higher the
level of the baseline at the end of the chromatogram
in comparison to the starting position in Figure 4.
The poor separation in Figure 5 is explained by the
comparatively low molecular weight of these samples
(50}80;103) and the superimposition of separation
Figure 5 Separation of the mixture of four styrene}EMA
by molecular weight and by composition. The peaks
copolymers at 503C on a column (60 mm;4 mm i.d.) packed with are indeed quite well separated when SEC fractions of
silica. Gradient: iso-octanePmethanol (5% min\1); flow rate the copolymer mixture are injected. Figure 4 indi-
0.5 mL min\1. Samples A to G as in Figure 4, 2.5 g each; UV cates that the molecular weight effect in the
signal detected at 259 nm. (Reproduced from GloK ckner, 1987a, investigation of the raw copolymers can be sup-
by courtesy of Elsevier Science Publishers.)
pressed by the sudden-transition technique. Table 2
Table 2 Characteristics of polymer separation with separate control of solubility and adsorptiona
Factor Details
Sample 20}100 g polymer per injection, dissolved in about 50 L solvent

Solvent C, capable of dissolving samples of the system under investigation in the whole
range of composition and molecular weight, used also for sample solutions
(recommended: tetrahydrofuran, dichloromethane)
Non-solvents A and B, opposite in polarity, both miscible with solvent C, e.g. A, acetonitrile,
methanol; B, n-heptane.
In general, the variety of non-solvents for a given polymer system is much
broader than the list of suitable solvents
Interactions of eluents and detector Eluents must not impede the monitoring of the eluting sample components, they
must be transparent if optical detection is employed. This demand is more
stringent for the gradient components A and B than for solvent C, whose
concentration is not changed during the elution of sample components. For
instance, separations at constant concentration of THF can be monitored at
230 nm or at constant DCM concentration with an evaporative light scattering
detector without disturbance
Reversed-phase separation Non-polar column, e.g. reversed-phase C18 bonded phase, injection into polar
non-solvent A, gradient APB after adjusting the solvent concentration to
a suitable constant value
Normal-phase separation Polar column, e.g. cyanopropyl bonded phase, injection into non-polar non-
solvent B, gradient BPA after adjusting the solvent concentration to a suitable
constant value
Reversed-phase and normal-phase separations Can be performed with a common set of three eluents
Automated search for optimum separation method Possible with programmable apparatus equipped with three storage bottles and
a device for column switching
Balance between solubility and adsorption Can be adjusted by the solvent concentration, which remains constant during
the elution
Length of chromatograms Can be optimized by sudden transitiona of solvent concentration from zero to
the selected level
a
Information on how to perform sudden-transition gradients is available in GloK ckner G, Wolf D and Engelhardt H (1994) Chromato-
graphia 39: 557}563.
summarizes the characteristics of sudden-transition in normal-phase mode with 25}40% dichloro-

gradient elution. methane are indicated. The characteristics of rever-
sed-phase elutions form a group in the left-hand area
Chromatography in Normal-Phase and of the triangle. The proportion of acetonitrile present
Reversed-Phase Modes Using a Solvent and
means that eluent systems in this region have a higher
Two Non-Solvents
polarity than those in the right-hand region. Rever-
Independent control of adsorption and solubility en- sed-phase chromatography starts with retention in
ables normal-phase and reversed-phase separations to a strongly polar medium. Sample components are
be performed with a common set of three liquids. released when the polarity of the eluent is no longer
This was Rrst demonstrated with styrene}MMA sufRcient for retention. Thus, the characteristics
copolymers in the system A (acetonitrile), B (n-hep- of reversed-phase elution are to be expected on the
tane) and C (dichloromethane, DCM) on either CN polar side of the composition diagram. On the other
or C18 bonded phases. hand, normal-phase elution characteristics are
Figure 6 shows chromatograms measured under located in the right-hand part of the triangle. This can
reversed-phase and normal-phase conditions. Both be understood by complementary reasoning because
modes yielded good separations. The elution order is normal-phase chromatography starts with retention
inverted in the reversed-phase mode, as expected. in a nonpolar medium.
The elution of styrene}MMA copolymers by the The characteristics in Figure 7 are due to samples
strong precipitant heptane (Figure 6A) is rather containing methyl methacrylate in the proportions
surprising. (from left to right) 83.7%, 62.2%, 48.1%, 34.1%,
Figure 7 shows the composition triangle of the elu- 14.1% or 0% (polystyrene homopolymer). This se-
ent system used in Figure 6 with dichloromethane at quence holds true with reversed-phase as well as with
the top, the polar non-solvent acetonitrile at the bot- normal-phase elutions. In both modes, the copolymer
tom left and the non-polar precipitant heptane at the with the highest content in polar methyl methacrylate
bottom right. Acetonitrile and heptane have a misci- units yields characteristics nearer to the polar (left)
bility gap that diminishes as dichloromethane is side of the diagram than the other samples. As ex-
added. Eluent mixtures containing 25% or more pected, the least polar sample (polystyrene) marks the
dichloromethane are homogeneous. The elution char- right border of the elution area in each mode.
acteristics of the styrene}MMA copolymers investi- All polymers considered here are soluble in the
gated in reversed-phase mode with 25}50% DCM or region beneath the solvent apex. The addition of
Figure 6 Separation of the mixture of five styrene}MMA copolymers at 353C and flow rate 1 mL min\1 by gradient elution in
reversed-phase (A) or normal-phase mode (B) after a sudden increase of dichloromethane concentration from zero to 30%, monitored
by an evaporative light-scattering detector. Sample in each mode: 6.76 g copolymer A (14.1% MMA)#5.54 g C (34.1%
MMA)#5.28 g E (48.1% MMA)#5.48 g G (62.2% MMA)#5.02 g I (83.7% MMA), dissolved in 10 L DCM. (A) Column
(250 mm;4.1 mm i.d.) packed with reversed-phase C18 bonded phase. Gradient: acetonitrilePn-heptane (4.99% min\1). (B) Column
(250 mm;4.1 mm i.d.) packed with cyanopropyl bonded phase. Gradient: n-heptanePacetonitrile (4.99% min\1). (Reproduced from
GloK ckner et al., 1994, by courtesy of Vieweg-Verlag.)
vents and control of solubility by a sudden increase of

solvent concentration to a constant level will not only
offer the opportunity to improve separations
with small additional effort, but will also con-
tribute to a better understanding of the mechanisms
of polymer chromatography.
See also: II/Chromatography: Liquid: Mechanisms:

Normal Phase; Mechanisms: Reversed Phases; Mecha-
nisms: Size Exclusion Chromatography. III/Polyethers:
Liquid Chromatography. Synthetic Polymers: Liquid
Chromatography.
Figure 7 Composition triangle for acetonitrile/n-heptane/ Further Reading

dichloromethane with elution characteristics of styrene}MMA
Baker CA and Williams RJP (1956) A new chromato-
copolymers in normal-phase and reversed-phase sudden-
graphic procedure and its application to high polymers.
transition gradients. Samples: 䢇, 83.7% MMA; *, 62.2% MMA;
#, 48,1% MMA, 䉫, 34.1% MMA; *, 14.1% MMA; 䊐, polysty- Journal of Chemistry Society (London) 1956:
rene. (Reproduced from GloK ckner, 1996, by courtesy of Gordon 2352}2362.
& Breach.) GloK ckner G (1987a) Normal- and reversed-phase separ-
ation of copolymers prepared from styrene and ethyl
methacrylate. Journal of Chromatography 403:
small quantities of heptane or acetonitrile to dich- 280}284.
loromethane will impair solubility but will not im- GloK ckner G (1987b) Polymer Characterization by Liquid
mediately cause precipitation. The polymers are still Chromatography. Amsterdam: Elsevier Science Pub-
soluble in mixtures of dichloromethane with about lishers.
GloK ckner G (1991) Gradient HPLC of Copolymers and
40% acetonitrile or 40% heptane. Thus, the upper
Chromatographic Cross-Fractionation. New York:
sections of the solubility boundary follow the left and Heidelberg, Tokyo: Springer-Verlag.
right sides of the eluent triangle. With increasing GloK ckner G (1996) Solubility and chromatographic separ-
concentration of nonsolvent, a precipitation thre- ation of styrene/methacrylate copolymers in ternary elu-
shold is reached on each side. From these points, both ent systems. International Journal of Polymer Analysis
branches of the solubility boundary bend towards and Characterization 2: 237}251.
each other. These sections may be determined experi- GloK ckner G and van den Berg JHM (1987) Copolymer
mentally by turbidimetric titration. For example, the fractionation by gradient high-performance liquid
elution characteristics of the copolymer containing chromatography. Journal of Chromatography 384:
48.1% MMA run almost parallel to the correspond- 135}144.
ing sections of the solubility boundary. In both rever- GloK ckner G, Wolf D and Engelhardt H (1991) Separation
of copoly(styrene/acrylonitrile) samples according to
sed-phase and normal-phase modes, the elution char-
composition under reversed phase conditions.
acteristics are shifted from the solubility boundary Chromatographia 32: 107}112.
towards the centre of the solubility window. This GloK ckner G, Wolf D and Engelhardt H (1994) Control of
shift indicates the contribution of adsorption to reten- adsorption and solubility in gradient high performance
tion, which is well known in gradient HPLC of liquid chromatography 5: separation of styrene/methyl
styrene}methyl methacrylate copolymers. Finally, methacrylate copolymers by sudden-transition gradients
both branches of the boundary will merge inside the in normal-phase and reversed phase mode. Chromatog-
triangle (above the miscibility gap). For details, see raphia 39: 557}563.
GloK ckner G (1996). Mourey TH and Schunk TC (1992) In E. Heftmann (ed.)
Solubility windows of similar shape can be ex- Chromatography } Fundamentals and Application of
pected with many polymers in mixtures of a solvent Chromatography and Related Differential Migra-
tion Methods. Chapter 22. Amsterdam: Elsevier Science
with two non-solvents differing in polarity.
Publishers.
Hence, HPLC separation generally should be possible Pasch H and Trathnigg B (1997) HPLC of Polymers. New
in normal-phase as well as in reversed-phase mode York: Heidelberg, Tokyo: Springer-Verlag.
with a suitable ternary eluent system. These separ- Quarry MA, Stadalius MA, Mourey TH and Snyder LR
ations should be achievable near the respective side of (1986) General model for the separation of large mol-
the solubility boundary. Thus, the use of ternary ecules by gradient elution: sorption versus precipitation.
gradients consisting of a solvent and two non-sol- Journal of Chromatography 358: 1}16.
2984 III / HERBICIDES / Gas Chromatography
Schultz R and Engelhardt H (1990) HPLC of synthetic Schunk TC (1993) Chemical composition separation of
polymers: characterization of polystyrenes by high per- synthetic polymers by reversed-phase liquid chromato-
formance precipitation liquid chromatography graphy (review). Journal of Chromatography A 656:
(HPPLC). Chromatographia 29: 205}213. 591}615.
HERBICIDES
chromatography (GC) allowed the analysis of these

Gas Chromatography compounds in commercial formulations with high
selectivity and sensitivity. The analytical procedure is
J. L. Tadeo and C. Sanchez-Brunete, Department of
commonly based on the dissolution of a known
Sustainable Use of Natural Resources, INIA, amount of the formulation in an organic solvent,
Madrid, Spain which often contains an internal standard to improve
the precision and accuracy of the determination. An
Copyright ^ 2000 Academic Press aliquot of this solution is analysed by GC. Packed
columns were used initially, but have now been re-
placed by capillary columns of low or medium polar-
Herbicide Formulations ity and Same ionization is the detection technique
Weeds have been controlled by humans since the more widely used. When herbicides are not volatile or
beginning of agriculture by means of mechanical thermally stable, high performance liquid chromatog-
tools or by hand. It was early in the 20th century that raphy (HPLC) is the preferred technique for their
some inorganic compounds were Rrst used with this determination in commercial formulations. Figure 1
aim. The discovery of the herbicidal properties of shows the gas chromatographic separation of a mix-
2,4-D (2,4-dichlorophenoxyacetic acid) in 1945 can ture of phenoxy esters.
be considered the initiation of use of organic herbi-
cides in agriculture. Since then, more than 130 dif-
ferent active compounds have been synthesized for
Herbicide Residue Analysis
their application as herbicides. These compounds can Residues of herbicides will persist in the plant or in
be grouped, according to their chemical structures, the soil for a variable time, depending on their
into different herbicide classes (Table 1). physicochemical properties and on the environmental
Compounds belonging to the principal herbicide conditions. Analysis of herbicide residues in these
groups will be considered in this study. These com- matrices is important, not only from the point of view
pounds control weeds in a variety of ways, showing of the efRcacy of application, but also to know
different modes of action, selectivity and ap- the distribution and persistence of these compounds
plication characteristics. Soil-applied herbicides are in food and in the environment. Therefore, herbicides
absorbed by roots or emerging shoots and foliage- of a wide range of polarities have to be determined in
applied herbicides are absorbed into the leaves, where complex environmental matrices at very low levels.
they may be translocated to other parts of the plant. Initially, herbicide residues were analysed by
The active ingredient of a herbicide is a compound, colorimetric methods. These procedures were gener-
usually obtained by synthesis, which is formulated by ally based on acidic or basic hydrolysis followed by
a manufacturer in soil particles or liquid concen- formation of derivatives. These methods are time-
trates. These commercial formulations of herbicides consuming and do not usually distinguish between
are diluted with water before application in agricul- the parent herbicide and metabolites.
ture at the recommended doses. Herbicide formula- Since the development of GC, this technique has
tions generally contain other materials to improve the been widely used in the analysis of these compounds.
efRciency of application. Table 2 summarizes the preparation of different
Analysis of herbicide formulations was initially types of samples for residue determination. These
carried out by wet chemical procedures, such as de- samples are generally analysed by a procedure with
termination of total chlorine, nitrogen or phos- the following main steps: sample extraction, clean-up
phorus, or by spectrometric procedures like ultra- of extracts, then GC determination and identiRcation.
violet absorption. The development of gas Some compounds are not volatile or thermally stable
III / HERBICIDES / Gas Chromatography 2985
Table 1 Chemical structures of herbicides
Benzonitriles
Bromoxynil; ioxynil
Phenoxyacids
2,4-D; MCPA; MCPP; dichlorprop; diclofop; fenoxaprop
Carbamates
EPTC; triallate
Chloroacetamides
Alachlor; metolachlor
Dinitroanilines
Butralin; ethalfluralin; pendimethalin; trifluralin
Triazines
Ametryn; atrazine; cyanazine; simazine; terbutryn
Uracils
Bromacil; lenacil; terbacil
Ureas
Chlorotoluron; isoproturon; linuron; chlorsulfuron; metsulfuron; triasulfuron
coveries is 70}120%, with a standard deviation of

20%.
Several injection techniques are used in herbicide
residue analysis. The techniques most often employed
are splitless, on-column and programmed-temper-
ature vaporizer injection and the volume normally
injected is 1}2 L.
Various selective detectors are used in trace analy-
sis of herbicides. The electron-capture detector (ECD)
was Rrst used for the determination of halogen-con-
taining compounds or halogenated derivatives, due to
its high sensitivity for these compounds. The nitro-
gen}phosphorus detector (NPD), also known as ther-
mionic or alkali Same detector, is commonly used in
the analysis of nitrogen-containing herbicides. Both
detectors are highly sensitive and can detect herbicide
concentrations lower than 1 pg. The Same photomet-
ric detector (FPD) has sometimes been used in the
determination of sulfur-containing herbicides. The
limit of detection (LOD) reSects the sensitivity of
Figure 1 Analysis of a mixture of phenoxy esters by gas a detector for a given compound and it is deRned as
chromatography on a BP-5 fused silica column, 12 m;0.53 mm the amount producing a signal-to-noise ratio equal to
i.d., with helium as carrier gas at 10 mL min\1 and flame ioniz- 3. When this ratio is determined with extracts of real
ation detection. Oven temperature was held at 1803C for 5 min,
samples, processed through the whole analytical pro-
increased at 253C min\1 to 2503C and held for 10 min. 1, 2,4-D
isobutyl ester; 2, MCPA 2-butoxyethyl ester, 3, 2,4-DP 2- cedure, this parameter is known as the limit of quan-
butoxyethyl ester; 4, 2,4-D 2-butoxyethyl ester. Adapted from tiRcation (LOQ), which depends on the efRcacy
SaH nchez-Brunete C, PeH rez S and Tadeo JL (1991) Determination of the extraction and clean-up procedures and on the
of phenoxy ester herbicides by gas and high-performance liquid selectivity and sensitivity of the detector. The coup-
chromatography. Journal of Chromatography 552: 235 with per-
ling of mass spectrometry (MS) with GC allows the
mission from Elsevier Science.
determination of herbicide residues with high selec-
tivity and sensitivity and, in addition, the identiRca-
tion of residues by means of their mass spectra ob-
and need to be derivatized before being analysed by tained at trace levels. The atomic emission detector
GC. allows monitoring characteristic wavelengths for car-
Currently capillary columns are most commonly bon and hydrogen atoms, as well as the more speciRc
used for residue analysis. Table 3 shows some repre- emission lines for phosphorus, nitrogen and sulfur.
sentative examples of different packed and capil- Table 4 summarizes the detection techniques for her-
lary columns used in the determination of herbicides. bicide residue determination by GC.
In trace analysis, blank samples are commonly pro- The analysis of herbicides, grouped into various
cessed through the analytical method to identify the chemical classes, is considered in more detail
possibility of interferences in the herbicide determina- below.
tion from other sample or reagent components. In
Benzonitriles and Phenoxy Acids
addition, recoveries of the analysed compound are
also carried out through the extraction and clean-up Benzonitriles and phenoxy acids are widely applied as
procedures. The accepted normal range of these re- salts or esters, but they are hydrolysed to their
Table 2 Procedures used in sample preparation for herbicide residue determination
Matrix Amount sampled Sample preparation Amount extracted Extraction procedure
Soil 1 kg Sieving ((2 mm) 10}20 g Shaking, Soxhlet, SFE

Water 1L Filtration 0.1}1 L SPE, LLE
Plants 1 kg Blending 20}50 g Homogenization
Air 25}250 L Adsorption or trapping Thermal or solvent desorption
SFE, Supercritical fluid extraction; SPE, solid-phase extraction; LLE, liquid}liquid extraction.
Table 3 Chromatographic columns used in herbicide residue analysis
Column (length/diameter) Stationary phase Applications
Packed (1}3 m/2}4 mm) Dimethylpolysiloxane (SE-30, DC-200) Carbamates, ureas, dinitroanilines, triazines
Phenylmethylpolysiloxane (OV-17, OV-25) Benzonitriles, phenoxy acids
Trifluoropropylpolysiloxane (QF-1, OV-210) Carbamates, chloroacetamides, phenoxy acids
Polyethyleneglycol (Carbowax) Triazines
Cyanoethylpolysiloxane (XE-60) Triazines
Cyanopropylpolysiloxane (OV-225) Ureas
Capillary (10}30 m/0.2}0.5 mm) Dimethylpolysiloxane Nitrogen-containing herbicides

Phenylmethylpolysiloxane Triazines
Polyethylene glycol Triazines, phenoxy acids, benzonitriles
Cyanopropylphenylmethylpolysiloxane Multiresidue
respective phenols or acids in the matrix. Extraction nitriles can be determined directly by GC, but the
of residues from soil and water is commonly per- sensitivity and reproducibility achieved are poor.
formed at acidic pH with organic solvents of medium Various derivatives overcome these problems and
polarity. The extraction of these herbicides from veg- diazomethane and heptaSuorobutyric anhydride are
etable matter is often done with aqueous solutions at the reagents most often used.
basic pH, followed by extraction with organic sol- The determination of herbicides is widely carried
vents. out by GC with ECD, if the compound has halogen
PuriRcation of extracts is required in most cases substituents or halogenated derivatives are obtained.
and this step is accomplished by liquid}liquid parti- MS detection has the advantage of being more selec-
tion at basic pH or by chromatography on silica tive and requiring less clean-up of extracts.
columns.
Analysis of these compounds in air is carried out by
Carbamates
trapping herbicides in ethylene glycol or in various
adsorbents, like polyurethane or amberlite resins. Carbamates are a wide group of pesticides and some
Derivatization of phenoxy acids, before GC deter- of them have herbicide properties, like the thiocarba-
mination, is necessary to make them volatile. Various mates S-ethyl dipropylthiocarbamate (EPTC) and
alkyl, silyl or pentaSuorobenzyl derivatives are ob- triallate. These compounds are extracted from soil
tained with this aim. Methyl esters have been com- with methanol or acetone and from water by means
monly prepared for the determination of phenoxy of hexane or dichloromethane. The extraction from
acids and the reagents most often used are diazo- plants is commonly accomplished with acetonitrile or
methane and boron triSuoride}methanol. Benzo- by steam distillation. Clean-up of extracts is often
Table 4 Detection of herbicides in environmental samples
Herbicides Detectors LOD ( g g\1) Derivatives
Benzonitriles ECD, MS 0.05}0.0003 Methyl ethers

MS 0.001 Heptafluorobutyryl
Phenoxyacids ECD, MS 0.05}0.001 Methyl esters
Carbamates ECD, NPD, FPD, MS 0.1}0.001
Chloroacetamides NPD, MS 0.05}0.001
Dinitroanilines ECD, NPD, MS 0.05}0.0001
Triazines NPD, ECD, MS 0.01}0.0001
Uracils NPD, ECD, MS 0.04}0.001
Ureas
Phenylureas NPD, MS 0.1}0.01 Methyl or ethyl
ECD 0.01}0.001 Heptafluorobutyryl
Sulfonylureas ECD 0.1}0.002 Methyl or PFB
Multiresidue NPD, MS 0.02}0.001
ECD, Electron capture detector; MS, mass spectrometry; NPD, nitrogen}phosphorus detector; FPD, flame photometric detector; PFB,
pentafluorobenzyl.
done by chromatography on silica columns. Most Triazines

carbamates are thermally unstable and are usually
These compounds form a wide group of herbicides
analysed by HPLC. Some carbamates, such as the
often employed in fruit trees and cereals; in particu-
thiocarbamates considered above, can be determined
lar, simazine and atrazine are the most widely used
by GC and their residues are determined by this
triazines.
technique using different detectors, like ECD,
Triazine herbicides are extracted from soil by polar
NPD, FPD and MS.
or medium polarity organic solvents, followed by
liquid}liquid partition or column chromatography
Chloroacetamides clean-up if necessary. Supercritical Suid extraction is
also used in the analysis of triazines in soil.
These compounds, also known as anilides, are very These herbicides are extracted from water by rever-
often used for weed control in maize, in combination sed-phase SPE, by anion exchange columns or by
with triazines. Chloroacetamides are extracted from liquid}liquid extraction with low polarity solvents,
soil by polar and medium polarity solvents and their followed in some cases by Florisil column clean-up.
determination is generally carried out without further Analysis of these compounds in plants is performed
puriRcation. Extraction from water is performed by by sample homogenization with polar organic sol-
reversed-phase solid-phase extraction (SPE) or by vents and column chromatography or liquid}liquid
liquid}liquid extraction with low polarity solvents partition clean-up. Triazines are sometimes analysed
and clean-up of extracts on silica columns is neces- in air and plugs of polyurethane foam have been used
sary in some cases. to extract these compounds from air.
Analysis of these herbicides in plants is done The gas chromatographic determination of
by extraction with polar solvents, followed by triazines is commonly performed with NPD, due to
puriRcation by liquid}liquid partition or column the high response obtained with this detector because
chromatography on silica or alumina adsorbents. of the number of nitrogen atoms in their molecules.
Analytical methods for the determination of these ECD is also employed but its sensitivity and selectiv-
herbicides in air have been reported, using several ity for these compounds are lower. GC-MS is used for
adsorbents or an ethylene glycol phase for their ex- conRrmation of residues as well as for routine deter-
traction from air. NPD and MS are the detectors most mination, due to the good sensitivity obtained with
widely used in the determination of chloroacet- selected ion monitoring. Figure 2 shows some
amides; the ECD is also sometimes used. chromatograms of the determination of various
triazines, together with other herbicides, by GC with
Dinitroanilines NPD and MS detection.
Dinitroaniline herbicides are highly lipophilic with Uracils
a low solubility in water and some compounds have
a remarkable volatility. Extraction of these com- These compounds, also named pyrimidines, are used
pounds from soil is carried out by polar as well as by to control weeds in some fruit trees, vegetables and
low polarity solvents. Dinitroanilines are extracted sugar beet. Extraction of these herbicides from soil is
from water by SPE or by liquid}liquid extraction with carried out with polar organic solvents or with basic
low polarity solvents, such as dichloromethane. aqueous solutions. Extraction from water is per-
Clean-up of water and soil extracts on silica columns formed by SPE or by liquid}liquid extraction with
is sometimes needed. low polarity solvents. Uracils are extracted from
Extraction with polar solvents, like methanol, is plants by homogenization with basic aqueous solu-
the method often used for the determination of ditions or with mixtures of polar solvents with water.
nitroanilines in plants, followed by liquid}liquid par- Clean-up of extracts is commonly carried out by
tition or Florisil column clean-up of extracts. liquid}liquid partition at different pHs. The gas
Analysis of these herbicides in air is performed by chromatographic determination of these herbicides is
means of several adsorbents or organic solvents as performed using ECD and NPD detectors, and also
trapping phases to remove these compounds from by MS and atomic emission detection.
air.
Ureas
ECD is often used in the determination of these
compounds due to the large response obtained, parti- Substituted ureas constitute one important group of
cularly with some halogen-containing dinitroanilines. herbicides formed by two different classes of
NPD and MS are also used in the gas chromato- compounds: phenylureas, one of the Rrst used herbi-
graphic analysis of these herbicides. cides, and sulfonylureas, introduced later and applied
Figure 2 Gas chromatograms of herbicide residues in soil, fortified at 0.01 g g\1, separated on an HP-1 capillary column,
12.5 m;0.20 mm i.d., with helium as carrier gas at 1 mL min\1 and detected (A) by GC-MS with selected ion monitoring or (B) by GC-
NPD. Oven temperature was maintained at 1003C for 1 min, programmed at 153C min\1 to 2503C and held for 1 min. Sim, simazine;
Th, thiazopyr; Pen, pendimethalin; Hex, hexazinone. Adapted with permission from PeH rez RA, Sanchez-Brunet C, Miguel E and Tadeo
JL (1998) Analytical methods for the determination in soil of herbicides used in forestry by GC-NPD and GC-MS. Journal of Agriculture
and Food Chemistry 46: 1864.
at lower doses due to their high herbicidal activity. from plants has also been reported. Clean-up of
Both types of compounds suffer from thermal extracts through column chromatography or
instability and their decomposition products, rather liquid}liquid partition is necessary before GC deter-
than the parent compounds, have been determined in mination.
some cases. To overcome this problem, various deriv- Detection of these herbicides is performed by NPD
atives amenable to GC have been obtained, mainly or by ECD, mainly when halogenated derivatives are
their methyl, heptaSuorobutyryl or perSuorobenzyl obtained. MS is also used in the determination of
derivatives. substituted ureas in environmental matrices.
Substituted ureas are extracted from soil by shak-
Multiresidue Analysis
ing with organic solvents of medium or high polarity,
like methanol or acetone, sometimes followed by The wide range of polarities and physico-chemical
silica column or liquid}liquid partitioning clean-up. properties of herbicides does not allow the determina-
These herbicides are extracted from water by re- tion of the more than 130 available herbicides in one
versed-phase SPE or by liquid}liquid extraction analytical method. Nevertheless, it is advisable to use
with dichloromethane. analytical procedures that allow the determination of
Urea herbicides are generally extracted from plants as many herbicides as possible in one method. These
by homogenization with polar organic solvents. multiresidue methods permit reduction in time and
Supercritical Suid extraction of these compounds cost of analysis as well as detection of the possible
presence of herbicide residues in samples with un- methanol, ethyl acetate and acetone. Vegetable sam-
known origin or contamination. A multiresidue ples are generally homogenized with organic solvent
method is able to determine all the herbicides that can and soil samples are normally shaken and Rltered.
be extracted, cleaned up, separated and detected in Matrix solid-phase dispersion (MSPD) is a technique
the conditions used in the analytical procedure. which has recently been employed for pesticide resi-
In some cases, two detectors } generally ECD due determination in food samples: it performs sample
and NPD } are connected at the end of the extraction and puriRcation at the same time. A simple
same chromatographic column to allow the detection multiresidue method, based on soil extraction in small
of a wider type of compounds. MS is used as a univer- columns, has been reported. Figure 3 shows represen-
sal detector and also for residue identiRcation tative chromatograms of the analysis of various nitro-
purposes. Atomic emission detection is a more recent gen-containing herbicides by this method.
detection technique which is increasingly used in Analysis of herbicides in water is generally based
trace analysis. on liquid}liquid extraction with dichloromethane
Herbicide residues are extracted from soil and or on SPE using reversed phase columns, mainly
plants with medium or high polarity solvents, like C18. Determination of herbicide residues in air is
Figure 3 Multiresidue analysis of soil extracts separated on an HP-1 column, 30 m;0.25 mm i.d., with helium as carrier gas at
1 mL min\1 and detected by GC-NPD. Oven temperature was kept at 803C for 1 min, programmed at 53C min\1 to 1403C, held for
10 min and programmed at 53C min\1 to 2503C, held 15 min. (A) Soil fortified with nitrogen-containing herbicides at 0.5 g g\1. (B)
Blank soil. 1, EPTC; 2, molinate; 3, propachlor; 4, ethalfluralin; 5, trifluralin; 6, atrazine; 7, terbumeton; 8, terbuthylazin; 9, dinitramine,
10, triallate; 11, prometryn; 12, alachlor; 13, metribuzin; 14, bromacil; 15, terbutryn; 16, cyanazine; 17, thiobencarb; 18, metolachlor; 19,
butralin; 20, oxadiazon; 21, lenacil. Reproduced from SaH nchez-Brunete C, PeH rez RA, Miguel E and Tadeo JL (1998) Multiresidue
herbicide analysis in soil samples by means of extraction in small columns and GC with NPD and MS detection. Journal of
Chromatography A 823: 17, with permission from Elsevier Science.
III / HERBICIDES / Solid-Phase Extraction 2991
accomplished by trapping these compounds on adsor- Further Reading

bents, followed by extraction with organic solvents.
BarceloH D and Henion MC (1997) Trace Determination of
Pesticides and their Degradation Products in Water.
Future Developments Amsterdam: Elsevier.
GC will continue to be the main chromatographic Blau K and King GS (eds) (1978) Handbook of Derivatives
technique used in herbicide residue analysis in the for Chromatography. London: Heyden.
Dobrat W and Martijn A (eds) (1998) CIPAC Handbook:
near future, due to the high sensitivity and selectivity
Vol. H Analysis of Technical and Formulated Pesticides.
given by the detectors that can be coupled with this Cambridge: Black Bear Press.
technique. In particular, the use of less expensive and Hutson DH and Roberts TR (eds) (1987) Progress in Pesti-
more robust and sensitive GC-MS equipment will cide Biochemistry and Toxicology, vol. 6 Herbicides.
keep growing in the routine determination and con- New York: Wiley.
Rrmation of herbicide residues. Milne GWA (1995) CRC Handbook of Pesticides. Boca
The time needed for sample processing is expected Raton: CRC Press.
to be reduced as a consequence of the continuation in Nollet LML (ed.) (1996) Handbook of Food Analysis. New
the development of automatic processes for sample York: Marcel Dekker.
preparation, extraction and clean-up. These processes Sherma J (ed.) (1989) Analytical Methods for Pesticides and
will use less sample and lower volumes of organic Plant Growth Regulators: vol. XVII. Advanced Analyti-
cal Techniques. San Diego: Academic Press.
solvents in the analytical procedure.
Tadeo JL, SaH nchez-Brunete C, GarcmH a-Valcarcel AI et al.
New improvements in the gas chromatographic (1996) Review: determination of cereal herbicide resi-
equipment to allow higher injection volumes of less dues in environmental samples by gas chromatography.
puriRed extracts can also be expected. Journal of Chromatography A 754: 347.
See Colour Plate 85. Tekel’ J and Kovac\ ic\ ovaH J (1993) Review: chromatographic
methods in the determination of herbicide residues in
See also: II/Chromatography: Gas: Detectors: Mass crops, food and environmental samples. Journal of
Spectrometry; Detectors: Selective. Insecticides: Gas Chromatography 643: 291
Chromatography. Pesticides: Supercritical Fluid Chroma- Zweig G (ed.) (1972) Analytical Methods for Pesticides and
tography; Gas Chromatography. Solid-Phase Matrix Plant Growth Regulators: vol. VI. Gas Chromato-
Dispersion: Extraction. III/Sorbent Selection for Solid- graphic Analysis. New York: Academic Press.
Phase Extraction.
Solid-Phase Extraction
Y. PicoH , Universitat de Vale% ncia, Vale% ncia, liquid chromatography (HPLC) or gas chromatogra-
Spain phy (GC); accuracy and precision are improved; and
Copyright ^ 2000 Academic Press it is rapid and easily automated.
Another impressive feature of SPE is the commer-
Solid-phase extraction (SPE) methods, using bonded- cial availability of sorbents in small and inexpensive
silicas, were Rrst introduced in 1971 as an alternative cartridges. C18-bonded silica cartridges, styrene-
to liquid partitioning. The method combines extrac- divinylbenzene Empore威 extraction discs and Car-
tion and preconcentration of organic compounds in bopack威 cartridges have been extensively used for the
water by adsorption on proper solid material fol- extraction of organic molecules from water samples.
lowed by desorption with a small quantity of an Automated column switching systems and on-line
organic solvent. In comparison with liquid}liquid SPE coupled to determination devices have also been
extraction, the following advantages are offered: often reported for determination of pollutants in
the amount of solvent required for the clean up is drinking and surface water.
greatly reduced, thus saving time for the evaporative Because of the reasons given above, in recent years
concentration step and minimizing exposure of the much analysis of herbicides in fruit, vegetable and
analyst to the toxic solvent; the Rnal eluate has less water has been conducted using SPE. Phenoxy
interfering material, and it could be analysed using acids, phenylureas, aryloxyphenoxypropionic acids,
any of a variety of detection, separation and identi- triazines, sulfonylureas, imidazolinones, glyphosate,
Rcation techniques, including high performance phenoxyacetic acids, bipyridynium compounds,
accomplished by trapping these compounds on adsor- Further Reading

bents, followed by extraction with organic solvents.
BarceloH D and Henion MC (1997) Trace Determination of
Pesticides and their Degradation Products in Water.
Future Developments Amsterdam: Elsevier.
GC will continue to be the main chromatographic Blau K and King GS (eds) (1978) Handbook of Derivatives
technique used in herbicide residue analysis in the for Chromatography. London: Heyden.
Dobrat W and Martijn A (eds) (1998) CIPAC Handbook:
near future, due to the high sensitivity and selectivity
Vol. H Analysis of Technical and Formulated Pesticides.
given by the detectors that can be coupled with this Cambridge: Black Bear Press.
technique. In particular, the use of less expensive and Hutson DH and Roberts TR (eds) (1987) Progress in Pesti-
more robust and sensitive GC-MS equipment will cide Biochemistry and Toxicology, vol. 6 Herbicides.
keep growing in the routine determination and con- New York: Wiley.
Rrmation of herbicide residues. Milne GWA (1995) CRC Handbook of Pesticides. Boca
The time needed for sample processing is expected Raton: CRC Press.
to be reduced as a consequence of the continuation in Nollet LML (ed.) (1996) Handbook of Food Analysis. New
the development of automatic processes for sample York: Marcel Dekker.
preparation, extraction and clean-up. These processes Sherma J (ed.) (1989) Analytical Methods for Pesticides and
will use less sample and lower volumes of organic Plant Growth Regulators: vol. XVII. Advanced Analyti-
cal Techniques. San Diego: Academic Press.
solvents in the analytical procedure.
Tadeo JL, SaH nchez-Brunete C, GarcmH a-Valcarcel AI et al.
New improvements in the gas chromatographic (1996) Review: determination of cereal herbicide resi-
equipment to allow higher injection volumes of less dues in environmental samples by gas chromatography.
puriRed extracts can also be expected. Journal of Chromatography A 754: 347.
See Colour Plate 85. Tekel’ J and Kovac\ ic\ ovaH J (1993) Review: chromatographic
methods in the determination of herbicide residues in
See also: II/Chromatography: Gas: Detectors: Mass crops, food and environmental samples. Journal of
Spectrometry; Detectors: Selective. Insecticides: Gas Chromatography 643: 291
Chromatography. Pesticides: Supercritical Fluid Chroma- Zweig G (ed.) (1972) Analytical Methods for Pesticides and
tography; Gas Chromatography. Solid-Phase Matrix Plant Growth Regulators: vol. VI. Gas Chromato-
Dispersion: Extraction. III/Sorbent Selection for Solid- graphic Analysis. New York: Academic Press.
Phase Extraction.
Solid-Phase Extraction
Y. PicoH , Universitat de Vale% ncia, Vale% ncia, liquid chromatography (HPLC) or gas chromatogra-
Spain phy (GC); accuracy and precision are improved; and
Copyright ^ 2000 Academic Press it is rapid and easily automated.
Another impressive feature of SPE is the commer-
Solid-phase extraction (SPE) methods, using bonded- cial availability of sorbents in small and inexpensive
silicas, were Rrst introduced in 1971 as an alternative cartridges. C18-bonded silica cartridges, styrene-
to liquid partitioning. The method combines extrac- divinylbenzene Empore威 extraction discs and Car-
tion and preconcentration of organic compounds in bopack威 cartridges have been extensively used for the
water by adsorption on proper solid material fol- extraction of organic molecules from water samples.
lowed by desorption with a small quantity of an Automated column switching systems and on-line
organic solvent. In comparison with liquid}liquid SPE coupled to determination devices have also been
extraction, the following advantages are offered: often reported for determination of pollutants in
the amount of solvent required for the clean up is drinking and surface water.
greatly reduced, thus saving time for the evaporative Because of the reasons given above, in recent years
concentration step and minimizing exposure of the much analysis of herbicides in fruit, vegetable and
analyst to the toxic solvent; the Rnal eluate has less water has been conducted using SPE. Phenoxy
interfering material, and it could be analysed using acids, phenylureas, aryloxyphenoxypropionic acids,
any of a variety of detection, separation and identi- triazines, sulfonylureas, imidazolinones, glyphosate,
Rcation techniques, including high performance phenoxyacetic acids, bipyridynium compounds,
2992 III / HERBICIDES / Solid-Phase Extraction
chloroacetamides, dinitroanilines and substituted The sample is extracted under neutral or slightly
phenols are examples of herbicides usually extracted alkaline conditions, and the pH is adjusted before the
and isolated by this technique. extraction to between 6 and 8. Under these condi-
It is undeniable that SPE is gaining in importance tions, salts of humic acids, which generally cause
and, today, is a well-established and validated considerable interference in herbicide determination,
method, since the Environmental Protection Agency are unlikely to be adsorbed during enrichment. As
(EPA) in the United States currently offers one acid herbicides are highly polar, they are soluble in
SPE procedure for the analysis of organic compounds water and in aqueous solutions and are less soluble
(including neutral herbicides) and two for the analysis (in their dissociated form) in apolar sorbents. To
of acid herbicides. overcome this difRculty, the aqueous phase has to
be acidiRed before extraction to suppress the dissocia-
tion of this class of herbicides and to facilitate the
Solid-Phase Extraction transfer of the undissociated molecular species to the
Analyte Characteristics solid phase.
The recoveries and the relative standard deviation
The determination of herbicide residues is an intricate of the performance of different devices and solid
problem because of the large number of chemicals phases are compared in Tables 2 and 3 for
involved. As a general rule, to classify them into basic/neutral and acid/phenolic herbicides, respec-
a wide variety of classes depending on their chemical tively. The C18 cartridges showed good recoveries
structure results in a lot of groups that barely provide with most of the basic/neutral and acidic/phenolic
enough information in order to select the best SPE herbicides. Compounds having a small (deiso-
procedure. propylatrazine, tribensulfuron-methyl) or a very high
In this way, the most practical approach is to or- (beta-cySuthrine) afRnity to the C18 material gave
ganize the herbicides according to their acid/base the worst recoveries. In comparison with Empore威
character or other properties that condition the pro- discs a lower breakthrough of polar metabolites of
tocol following by SPE. Table 1 shows these charac- atrazine was reported, possibly due to the fact that
teristics for the major classes of herbicides and Empore威 discs contain only half the quantity of
some examples of the structures included in each C18 material. However, lower recoveries were
group. achieved for medium polar and non-polar pesticides
except triSuralin and trialate.
Disposable Solid Phases
As reported in the literature on the subject, the
The modern SPE technique began in 1978 with the bonded-silicas, in cartridge or in disc conRguration,
introduction of Sep-Pack cartridges, the Rrst compact are the most commonly used supports, but they also
silica-based solid-phase extraction device for sample have some limitations:
preparation on the market. Present-day, disposable
prepacked columns or cartridges are available from E For polar analytes, the retention is weak and often
more than 30 manufacturers, who offer phases results in breakthrough during the loading step.
such as C18, C8, cyano and amino. The containers are E Basic analytes interact strongly with the residual
generally made of polypropylene. The sorbent bed silanols, which in turn cause low recovery.
varies from 100 to 1000 mg and is retained between E The sorbent must remain wet prior to sample load-
two porous frits. ing. (If one accidentally lets the cartridges run dry,
The use of Empore威 discs are described in more the recovery is low and variable.)
recent studies. These devices include Sat discs with E Poor stability in very acidic and basic media,
large cross-sectional areas that provide advantages which limits their use to the pH range of between
for preconcentration and clean-up methods with re- 2 and 8.
spect to the sorption, capacity, back pressure and
stability after repeated use. These limitations have led to a search for new
Reversed-phase silica-based sorbents, especially materials with improved characteristics. For
C8 and C18 bonded-silicas, are the most widely used example, the styrene-divinylbenzene resins have been
packings for SPE. A typical SPE requires a previous extensively checked for their use in the extraction of
sorbent activation step (wetting), usually with meth- pesticides. These polymers show higher retention of
anol, and removal of activation solvent excess (condi- analytes and a wider pH range than C18 silicas. The
tioning), usually with water. LC-grade polymers used as stationary phases have
Neutral herbicides can be extracted from 1 L sam- more commonly been used in precolumns (mainly
ples with an average amount of sorbent (500 mg). PRP-1 and PLRP-S) for on-line purposes, because
Table 1 Chemical structure of major classes of herbicides according to the character that determines the SPE procedure used
Character Class Typical herbicide Chemical structure
Basic/neutral Triazine Atrazine
Chloroacetamide Metolachlor
Urea Monuron
Carbamate Desmedifam
Dinitroaniline Pendimethalin
Acid phenolic Phenoxy acid 2,4,5-T
Substituted phenol Bromoxynil
Aryloxyphenoxy Fluazifop
propanoic acid
Cationics Bipyridilium compound Diquat
Very soluble Organophosphate Glyphosate

Table 2 Relative standard deviation (RSD), recoveries (Rec.) and determination limits (DL, P 95%) for the preconcentration of basic/neutral pesticides from 1000-mL Milli-Q-water
2994
Concentration Preconcentration on
range (ng L\1)
Bond Elut C18 cartridges Empore威 C18 discs Empore威 SDB discs ENVI-Carb cartridges
RSD Rec. DL RSD Rec. DL RSD Rec. DL RSD Rec. DL

(%) (%) (ng L\1) (%) (%) (ng L\1) (%) (%) (ng L\1) (%) (%) (ng L\1)
Desisopropylatrazine 54}270 9.9 33 133 10.6 21 143 15.9 16 221 6.9 99 96

Desethylatrazine 50}248 8.7 82 113 8.8 49 110 3.6 72 46 7.3 101 94
Metoxuron 108}538 1.9 95 58 3.4 106 95 4.8 94 134 5.7 106 159
Hexazinon 107}535 8.1 96 224 6.9 102 186 6.7 104 182 7.4 108 200
Simazine 49}245 1.5 102 20 4.5 102 57 5.2 100 66 9.3 95 117
Metribuzin 94}470 4.7 94 120 4.0 104 99 4.1 109 99 10.2 97 249
Cyanazine 57}285 7.9 81 119 5.0 110 73 6.3 93 91 5.6 82 81
Carbofuran 179}895 6.4 92 309 4.4 107 211 6.8 95 311 23.1 80 1057
Methabenzthiazuron 100}498 5.2 89 138 5.0 98 127 3.9 103 101 5.2 86 135
Chlortoluron 101}503 4.9 96 133 4.5 102 118 2.3 98 61 5.9 97 158
Atrazine 50}250 5.6 96 75 5.6 100 72 3.6 93 47 7.1 90 92
III / HERBICIDES / Solid-Phase Extraction
Monolinuron 149}745 4.0 96 166 13.7 105 473 5.7 114 221 7.7 89 296
Diuron 102}510 7.3 89 194 5.1 101 134 4.9 94 130 6.5 92 171
Isoproturon 109}543 6.7 90 191 5.7 102 158 7.3 95 199 6.0 100 164
Metobromuron 102}508 13.4 88 334 9.2 93 230 4.5 96 119 9.4 96 248
Metazachlor 123}165 7.3 85 235 6.5 106 204 2.2 100 72 5.3 89 177
Sebutylatrazine 53}265 4.9 97 69 4.8 103 66 4.3 102 58 4.1 97 56
Terbutylatrazine 53}263 6.3 104 87 6.6 104 89 5.7 98 77 6.2 92 83
Linuron 102}508 2.3 101 66 3.6 96 97 3.2 106 87 3.5 98 94
Napropamide 45}225 5.8 98 69 2.1 100 24 4.9 100 57 3.1 98 36
Terbuconazol 164}820 6.5 90 282 2.1 101 98 5.2 101 224 3.9 98 168
Metolachlor 127}633 3.3 103 115 6.3 100 204 7.1 104 226 5.2 89 166
Propiconazol 225}1125 4.9 96 316 2.3 99 156 4.7 107 290 6.3 97 391
Dinosebacetate 138}688 6.6 41 239 16.2 52 509 7.6 70 262 32.2 33 1108
Parathion-ethyl 148}740 4.6 90 189 8.0 101 294 7.1 97 264 6.0 88 224
Pyrazophos 125}623 7.1 82 230 3.6 93 120 5.1 107 165 32.8 38 1058
Bifenox 67}333 3.9 98 70 3.5 93 61 8.8 92 147 4.2 89 70
Prosulfocarb 140}698 4.8 91 182 11.6 94 385 5.5 72 198 6.2 66 224
Pendimethalin 121}603 8.0 84 248 9.9 93 290 4.5 107 142 10.4 62 327
Trifluoralin 125}625 8.3 63 268 14.2 21 415 0.0 0 0 10.9 60 458
Triallate 128}638 4.1 96 144 14.7 35 434 0.0 0 0 8.5 52 513
Fluoroxypyrester 81}403 3.2 86 71 5.1 89 105 6.0 98 123 8.3 52 170
Beta-cyfluthrine 190}950 3.7 61 202 8.1 82 383 2.2 89 122 15.0 42 818
(Reproduced with permission from SchuK lein J, Martens D, Spizauer P and Kertrup A (1995) Comparison of different solid phase extraction materials and techniques by application of
multiresidue methods for the determination of pesticides in water. Fresenius Journal of Analytical Chemistry 352: 565}571.)
they are too expensive for use in disposable SPE
(Reproduced with permission from SchuK lein J, Martens D, Spizauer P and Kertrup A (1995) Comparison of different solid phase extraction materials and techniques by application of
Table 3 Relative standard deviation (RSD), recoveries (Rec.) and determination limits (DL, P 95%) for the preconcentration of acidic phenolic pesticides from 1000-mL Milli-Q-water
cartridges. Empore威 extraction discs have recently
(ng L\1)
become available, with styrene divinylbenzene (SDB)
126
153
252
206
156
113
69
67
80
69
copolymer sorbents enmeshed in the matrix.
DL
ENVI-Carb cartridges
Recoveries of acid herbicides from water samples
Rec. have been compared by using C18 and SDB discs; the
(%)
124
104
76
77
97
89
91
88
96
91
results of this comparison are controversial. Some
authors documented an improvement in the recove-
ries of phenoxycarboxylic acid and phenols on SDB
RSD
10.7
5.3
9.5
6.5
7.1
2.4
7.0
3.6
3.8
4.2
(%)
discs for the enrichment of samples down to 500 mL

(see Table 4). The addition of salt considerably en-
hances the recovery and decreases the differences
(ng L\1)
between the extraction efRciency of C18 and resin

116
300
111
140
220
127
67
66
84
80
discs. However, other authors reported that SDB
DL
discs showed worse recovery rates under acidic condi-

Empore-SDB discs
tions, in comparison with C18 when preconcentra-

Rec.
(%)
115
106
104
68
71
59
67
83
76
97
tions were carried out with 1-L samples (see Table 3).
In any case, salting out the water sample enhances the
Preconcentration on
retention of substances on both materials; this in-

multiresidue methods for the determination of pesticides in water. Fresenius Journal of Analytical Chemistry 352: 565}571.)
RSD
12.8
5.2
5.0
4.9
3.8
3.0
6.3
9.9
4.4
4.7
(%)
creases recovery rates for hydrophilic substances.

However, salting out is avoided as it may introduce
impurities into the samples. The addition of a small
(ng L\1)
quantity of methanol or other organic solvent also

164
177
119
110
121
121
99
66
93
85
enhances the recovery by the so-called ‘dynamic sol-

DL
vation’. However, it is not recommended, as it pro-

Empore威 C18 discs
duces a relatively early breakthrough of hydrophilic

Rec.
(%)
103
93
85
46
99
65
94
95
94
99
substances.
Graphitized carbon black (GCB) has been con-
Rrmed to be a valuable adsorbing material for SPE of
RSD
(%)
pesticides in aqueous environmental samples. GBC

7.4
4.8
3.8
6.5
6.1
4.2
4.7
5.0
4.5
4.3
cartridges proved to be more efRcient than the

more commonly used C18 bonded-silica cartridges for
(ng L\1)
the SPE of polar herbicides, whereas the extraction of

Bond Elut C18 cartridges
208
119
118
151
202
104
166
71
92
96
non-polar compounds showed inferior results (see

DL
Table 3). Although GCB is known to behave as

a natural reversed phase, it contains chemical hetero-
Rec.
(%)
100
100
38
43
63
98
76
99
88
88
geneities on its surface, which are able to bind anions

via electrostatic forces. GBC can behave as both a re-
versed-phase sorbent and an anion exchanger, retain-
RSD
(%)
ing the acidic pesticides in their ionic form under

5.3
6.7
3.9
8.5
3.9
4.1
6.6
8.7
5.6
6.1
acidic conditions. In this situation, the base}

neutral/acid fractionation can be achieved by using
range (ng L\1)
Concentration
solvent mixtures at different pHs.

Silica-based ion exchangers are found in disposable
110}561
105}533
104}528
52}258
53}263
91}464
96}487
88}446
91}464
70}357
SPE cartridges. They are not widely used for the

preconcentration of environmental samples owing to
their low capacity. Strong anion exchanger discs have
been used for the analysis of chlorinated acid and
phenoxy acid herbicides. The main problem with
Trifensulfuron-methyl
these comes from the fact that environmental waters

Metsulfuron-methyl
contain high amounts of inorganic ions, which over-

load the capacity of the sorbent.
Bifenox acid
Dichlorprop
Bromoxynil
Selective SPE from environmental waters has

Haloxyfop
Dicamba
been accomplished by using different sorbents

MCPA
MCPB
Ioxynil
coupled in the same or in different cartridges.

Table 4 C18 and resin recoveries and effect of salting water
Analyte Recoveries$RSD (%; n"3)
C181 Resin1 C182 Resin 2
Acifluorfen 77$20 82$5 104$5 121$1

Bentazon 0 ND 90$13 71$5
Chloramben 8$11 3$15 72$14 77$7
2,4-D 86$12 83$6 81$8 94$15
Dalapon 0 42$25 12$75 31$30
2,4-DB 81$13 80$14 118$10 130$8
Dacthal 53$17 99$8 67$16 97$5
Dicamba 73$13 71$14 83$3 94$15
3,5-Dichlorobenzoic acid 70$17 76$2 86$25 107$20
Dichloroprop 77$11 78$3 85$9 94$10
Dinoseb 72$16 75$5 92$26 85$6
Pentachlorophenol 69$14 70$2 65$15 73$8
Pichloram 49$19 74$7 96$24 99$21
2,4,5-T 76$11 75$14 93$10 89$5
Silvex 73$14 74$14 82$9 80$5
1
Fortified, unsalted reagent water.
2
Fortified reagent water with 20% (w/w) Na2SO4.
ND, no data.
(Reproduced with permission from Hodgeson J, Collins J and Bashe W (1994) Determination of acid herbicides in aqueous samples by
liquidIsolid disk extraction and capillary gas chromatography. Journal of Chromatography A 659: 395}401.)
One of the sorbents is non-speciRc, such as GCB, Glyphosate, due to its ionic form, can be precon-
which traps the analytes of interest and many other centrated using anionic and cationic resins. Derivatiz-
compounds, while the other is more speciRc, such as ation of the analyte, prior to SPE of the water
a cation or anion exchanger, which retains and recon- sample, seems to help the concentration from water
centrates the analytes of interest. Table 5 shows the samples.
recoveries of nine phenoxy acid herbicides extracted
by one miniaturized cartridge containing 50 mg of
GCB at the top and 70 mg of a silica-based strong Table 5 Recovery of herbicides from 200-mL groundwater
anion exchanger (SAX) at the bottom, compared with samples by using one cartridge containing GCB and anion
C18 and anion exchanger extraction. A large loss of exchanger compared with that from two other extraction
dicamba and incomplete recovery of phenoxyacids methods
were obtained by using the resin-based exchanger
% Recovery1
material.
Ammonium quaternary compounds and gly- C18 Anion GCB#anion
phosate constitute a special and complicated case. exchanger exchanger
Their determination is very important because they
are among the top herbicides used in the world. An pH 2 pH 7.9
important drawback in the preconcentration of such Dicamba 94.2 (10 23.0 97.3
compounds from water is their high polarity. Several 2,4-D 96.1 (10 78.3 98.4
efforts have been made to analyse them in envir- MCPA 96.4 (10 77.7 97.6
onmental samples. 2,4-DP 97.7 (10 76.8 96.4
MCPP 94.4 (10 82.0 97.3
SPE of ammonium quaternary herbicides has been
2,4,5-T 93.5 (10 81.4 95.4
mainly performed with silica, which is a well-known 2,4-DB 96.0 (10 82.3 98.9
example of the solid phase using adsorption and ionic MCPB 95.8 (10 81.5 96.5
interaction mechanisms with the silanol groups. 2,4,5-T 93.1 (10 77.3 95.2
Recoveries are rather acceptable in the pH range 1
Mean values were calculated for two determinations.
7.5}9. Taking into account that at pH values higher
(Reproduced with permision from Di Corcia A, Marchetti M and
than 7 the silanol groups of the stationary phase are Sampieri R (1989) Extraction and isolation of phenoxyacid herbi-
ionized, at these pH values the cation-exchange capa- cides in environmental waters using two adsorbents in one mini-
city of the solid-phase will be increased. cartridge. Analytical Chemistry 61: 1363}1367.)
Table 6 Desorption efficiencies from the solid phase and overall recoveries of herbicides
Herbicide Desorption efficiency (%) Recovery from filtered

river water (%)5
Methanol Ethyl acetate Acetone (3 mL) Acetone Acetone (6 mL)
3 mL 6 mL 3 mL 6 mL Mean1 RSD2 3 mL#HX3 3 mL#DCM4 Mean1 RSD 2 Mean1 RSD 2
ACN 68.9 68.9 78.3 78.3 74.6 8.4 74.6 74.6 81.9 12 128 14
Alachlor 67.7 67.7 80.5 80.5 80.1 7.4 80.1 80.1 101 6.1 105 3.4
Benfluralin 48.1 48.1 73.4 73.4 55.0 13 69.6 70.3 67.9 1.8 83.6 2.2
Bifenox (5 (5 78.3 78.3 69.2 9.1 69.2 79.0 92.9 8.2 98.1 5.7
Bromobutide 77.1 77.1 78.2 78.2 80.1 8.5 80.1 80.1 103 2.7 108 3.9
Bromobutide-debromo 76.4 76.4 76.6 76.6 79.5 7.8 79.5 79.5 92.7 7.8 101 7.1
Butachlor 50.6 64.0 79.6 79.6 73.4 9.3 83.5 83.1 100 5.2 112 1.1
Butamifos 73.9 73.9 81.4 81.4 81.2 9.2 81.2 81.2 96.5 3.9 104 2.6
Chlomethoxyfen (5 (5 74.9 74.9 70.9 5.9 70.9 83.2 91.3 8.7 97.9 6.7
Chlornitrofen (5 (5 77.8 77.8 63.2 13 70.6 80.5 71.9 3.2 81.6 4.4
Chlorprofam 57.5 57.5 81.1 81.1 68.2 5.1 68.2 79.1 95.7 5.4 103 4.6
Dimepiperate (5 42.2 80.1 80.1 71.7 7.8 84.0 85.8 104 8.9 110 4.8
Dimethametryn 64.5 64.5 83.1 83.1 79.4 6.9 79.4 79.4 96.4 3.2 105 5.8
Dithiopyr 73.5 73.5 79.5 79.5 74.4 6.6 74.4 82.4 84.1 8.6 85.9 5.0
Esprocarb (5 35.9 76.2 76.2 63.4 13 77.5 78.3 87.3 3.6 94.1 0.23
MCPA-ethyl 22.5 47.3 72.6 72.6 57.9 12 70.3 74.8 89.7 5.8 89.3 5.3
MCPA-thioethyl (5 (5 76.6 76.6 57.9 14 74.8 82.6 98.1 4.0 98.9 6.5
Mefenacet (5 55.4 78.2 78.2 82.5 2.5 82.5 82.5 90.7 2.9 103 12
Molinate 25.8 45.9 76.8 76.8 64.1 9.7 75.9 80.5 90.8 7.0 98.7 6.2
Naproanilide (5 60.0 76.8 76.8 79.8 2.4 79.8 79.8 96.0 3.1 103 9.5
Oxadiazon 38.0 59.4 76.1 76.1 70.8 6.7 83.2 84.0 93.2 1.8 96.8 1.6
Pendimethalin (5 42.9 92.6 92.6 71.1 8.9 86.4 86.5 73.2 7.3 86.4 2.8
Piperophos 54.1 71.8 79.9 79.9 85.9 5.7 85.9 85.9 93.5 5.3 100 9.1
Pretilachlor 64.1 64.1 78.5 78.5 79.8 8.6 79.8 79.8 108 6.5 112 2.7
Prometryn 70.9 70.9 79.4 79.4 82.8 8.2 82.8 82.8 93.6 4.5 101 4.1
Simazine 84.6 84.6 78.1 78.1 78.9 11 78.9 78.9 102 9.6 112 7.3
Simetryn 78.8 78.8 77.9 77.9 80.4 9.2 80.4 80.4 93.5 9.4 120 11
Thiobencarb (5 30.7 80.3 80.3 64.4 13 77.6 80.6 97.6 5.1 110 0.63
Trifluralin 50.3 50.3 76.8 76.8 57.3 13 71.1 72.3 72.1 1.2 87.1 5.8
1
Percentage mean recovery (n"3).
2
Percentage relative standard deviation.
3
A 3-mL volume of hexane.
4
A 3-mL volume of dichloromethane.
5
III / HERBICIDES / Solid-Phase Extraction
The herbicides collected on the cartridges were eluted with 6 mL of acetone.

(Reproduced with permission from Tanabe A, Mitobe H, Kawata K and Sakai M (1996) Monitoring of herbicides in river water by gas chromatographyImass spectrometry and solid phase
extraction. Journal of Chromatography A 754: 159}168.)
2997
It can be concluded that C18 material is inappropri- results obtained when herbicides were eluted from
ate for some herbicides, especially more polar and a cartridge using different solvents, such as meth-
very non-polar herbicides. In these cases, the SDB anol, ethyl acetate, acetone, hexane and dichloro-
polymers and GCB offer a valuable alternative. methane following acetone.
The appropriate choice of solid phase for application Desorption of acid herbicides from the sorbents
to a separation problem will vary from case to case can also be performed using a solution adjusted to
and must be adapted accordingly. a pH where the analytes are in their ionic form (two
units below or above the pKa ). The uniqueness of
GCB is that acid compounds are retained in their
Elution of the Target Analytes
ionic forms and neutral compounds are adsorbed by
Desorption of the compounds from the concentration unspeciRc mechanisms. In this situation, base}neu-
columns is mainly performed with a small volume of tral/acid fractionation can be easily achieved by Rrst
liquid. The partition coefRcient in a given solid- eluting base}neutral species with a neutral organic
phase eluent system should favour the shift of the solvent mixture and then passing a basiRed or acidi-
studied herbicides. On the other hand, SPE is not Red solvent system to desorb acidic compounds.
a separate step, but it is part of a process that includes Table 7 reports the results obtained using base}neu-
subsequent determination and so it should be taken tral/acid fractionation in three kinds of GCB. In all
into account that some determination systems, such cases, there was some carryover of 2,4-DB, which is
as GC, are incompatible with the presence of water. the weakest compound included in this table.
In this way, the selection of the eluting solvent de- With ion exchange sorbents, the analytes can be
pends on the selected sorbent, the analytes and the eluted from the SPE column by either adjusting the
detection method. Air-drying is often applied before pH in order to neutralize the charge on the analyte or
analyte elution in order to remove residual water. by using a buffer of high ionic strength.
Methanol and acetonitrile are recommended sol-
Sample Requirements
vents for the elution of herbicides adsorbed to C8 or
C18 silicas. Dichloromethane and ethyl acetate have Samples undergoing SPE need to be Rltered to separ-
also been extensively used, especially when the pres- ate suspended matter. Filtration is especially neces-
ence of water is undesirable. Table 6 presents the sary before extraction of surface water, but is also
Table 7 Base}neutral/acid fractionation by differential elution of selected compounds with cartridges containing three different types
of GCB at two eluents
Compound Sorbent material
Carbograph 1 Carbograph 4 Carbograph 5
Eluent A1 Eluent B 3 Eluent A Eluent B Eluent A Eluent B
Base/neutral
Atrazine 97 * 95 * 94 *
Linuron 99 * 98 * 95 *
Aldicarb 92 * 92 * 92 *
Acidic
Dichlorprop (3.5)3 * 95 * 97 30 73
2,4,5-T (2.2) * 97 * 102 * 99
Ioxynil (3.9) * 101 * 102 * 93
2,4-D (2.6) * 99 * 100 * 93
2,4-DB (4.8) 40 63 18 81 50 49
Mecoprop (3.7) * 99 * 99 * 96
Extraction from 1 L of Aldrich humic acid-spiked drinking water (spiked level, 10 g L\1). Mean recovery values obtained from three
measurements.
1
Eluent phase: CH2Cl2}CH3OH (80 : 20).
2
Eluent phase: CH2Cl2}CH3OH (80 : 20)#10 mmol L\1 tetrabutylammonium chloride (TBACI).
3
Reported pKa values of the acidic compounds are given in parentheses.
(Reproduced with permission from Crescenzi C, Di Corcia A, Passariello G, Samperi R and Turnes MI (1996) Evaluation of two new
examples of graphitized carbon blacks for use in solid-phase extraction cartridges. Journal of Chromatography A
733: 41}55.)
often advisable for extraction of ground water to humic acids, surfactants, inorganic salts, phenols,
avoid blocking up the cartridge material. However, polycyclic aromatic hydrocarbons (PAH), other
waters from different sources are very dif- pesticides and related compounds, can negatively
ferent in chemical composition. Matrix effects affect the analysis, signiRcantly diminishing the
from the water itself can cause errors in quantitation recovery efRcacy or interfering with the posterior
and determination. The presence in waters of com- determination. Figure 1 shows that when natural
mon contaminants (natural or xenobiotics), such as samples are acidiRed, humic and fulvic acids are
Figure 1 Effect of the pH of the sample on the preconcentration of 500 mL of drinking water spiked at 0.1 g L\1. Sample (A)
adjusted to pH 3 with perchloric acid and (B) not adjusted (pH 7). Analytical conditions: flow-rate, 1 mL min\1, loop, 50 L; mobile
phase, acetonitrile gradient with 0.005 M phosphate buffer acidified to pH 3 with HClO4, gradient from 10 to 30% acetonitrile from 0 to
10 min, and from 30 to 77% from 10 to 80 min; UV detection at 220 nm. Peaks: 1, chloridazon; 2, dicamba; 3, aldicarb; 4, methoxuron;
5, simazine; 6, cyanazine; 7, bentazone; 8, atrazine; 9, carbaryl; 10, isoproturon; 11, ioxynil; 12, MCPP; 13, difenoxuron; 14, 2,4-DB;
15, 2,4,5-T; 16, metolaclor; 17, dinoterb. (Reproduced with permission from Pichon V, Cau Dit Coumes C, Chen L, Guenu S and Henion
MC (1996) Simple removal of humic and fulvic acid interferences using polymeric sorbents for the simultaneous solid-phase extraction
of polar acidic, neutral and basic pesticides. Journal of Chromatography A 737: 25I33.)
Table 8 Recoveries of cationic herbicides (4 g L\1) from 0.25 L of water samples containing various concentrations of different
surfactants
Surfactants Concentration (g L\1) Recovery (%)1
Diquat Paraquat Difenzoquat
Cetrimide 5 98 99 90
50 93 92 92
300 95 91 93
3000 87 89 92
Benzalkonium chloride 5 114 114 94
50 107 107 93
300 102 102 95
3000 105 105 103
Sodium tetradecyl sulfate 5 99 102 87
50 98 95 93
300 41 47 41
3000 34 30 37
Lauryl sulfate 5 84 85 92
50 109 100 93
300 42 35 41
3000 34 30 39
Laurylsulfobetaine 5 89 83 79
50 91 94 84
300 47 57 42
3000 54 55 48
Brij-35 5 85 89 92
50 83 101 91
300 45 33 37
3000 36 30 42
Triton X-100 5 86 89 92
50 102 101 91
300 45 33 37
3000 36 30 42
1
Average recovery calculated from four determinations.
(Reproduced with permission from IbaH n ez M, PicoH Y and Man es J (1996) Influence of organic matter and surfactants on solid-phase
extraction of diqua, paraquat and difenzoquat from waters. Journal of Chromatography A 727: 245}252.)
co-extracted and co-eluted, which generates a large, Although some common contaminants of natural
unresolved peak in the chromatogram when HPLC waters have a negative effect on the recoveries,
with UV detection is used (chromatogram A). At pH SPE is useful for analysing herbicides in drinking and
7, humic and fulvic acids are not co-extracted, as can surface water because only in very extreme conditions
be seen by the Sat baseline from the beginning to the does the concentration of these contaminants reach
end of chromatogram (B). levels at which recoveries are signiRcantly decreased.
Organic matter and anionic or non-ionic surfac- The application of SPE to the isolation of herbicide
tants have demonstrated negative effect on the residues from other matrices presents difRculties
recovery of any class of herbicides. Although these that must be overcome, which have, up to now,
undesirable effects are well known, only a few discouraged investigation into the use of other ma-
analytical studies have focused on ways in which to trices. For liquid matrices (plasma, urine, blood or
avoid them. The proposed methods for removing milk), acceptable recoveries have been obtained using
interferences are based on the use of chemical re- protein precipitation prior to SPE but the impurities
agents, such as sulRte or cationic surfactants. In these present can accumulate in the analytical columns and
cases, the recovery values after chemical treatment affect the chromatogram. The recoveries ob-
were similar to those when a Milli-Q-quality water tained by SPE for determining triazines from milk are
standard was analysed. The recoveries reported in compared with those obtained by liquid}liquid
Table 8 show that the quantitative SPE of diquat, extraction (Hajs\ lova et al.) in Table 9. SPE was
paraquat and difenzoquat is affected by the pres- performed using a double trap: Rrst, a non-speciRc
ence of anionic, zwitterionic and non-ionic surfac- adsorbent (GCB), and then a cation exchanger. The
tants when they are present in water at a level of up to liquid}liquid extraction method, after an initial
50 g L\1. double protein precipitation using methanol in
Table 9 Recovery (n"6) of triazines from fortified On-line and Off-line Procedures
(50 ng mL\1) skimmed milk using the proposed method and that
of Hajs\ lovà et al. Nowadays, SPE methods using off-line proced-
ures can be converted into on-line SPE methods by
Compounds Recovery % (mean$RSD) direct connection of the precolumn to the analytical
SPE method Hajs\ lova% et al. column via switching valves. The concentrated
analytes are then directly desorbed and transferred to
Simazine 89.7$4.1 86.1$5.2
the analytical system. Such systems often involve
Atrazine 89.3$3.9 84.3$4.7
Prometon 90.4$4.0 92.4$4.3 microprocessor control of the stages for sample
Ametryn 89.5$3.5 90.0$4.7 switching and Sushing of solvents and eluents through
Propazine 93.4$3.6 88.6$4.2 the concentration and chromatographic columns.
Terbutylazine 91.6$3.4 87.3$4.2 On-line procedures have gained popularity since
Prometryn 87.2$3.8 85.5$4.0
European Union (EU) guidelines were introduced
Terbutryn 77.8$3.2 80.9$3.9
which limited the maximum amount allowed for
(Reproduced with permission from Lagana A, Marino A and Fago a single pesticide in drinking water to 0.1 g L\1 and
C (1995) Evaluation of double solid-phase extraction system for for several pesticides to 0.5 g L\1, including toxic
determining triazine herbicides in milk. Chromatographia 41:
transformation products. Very sensitive methods are
178}182.)
required for monitoring herbicide residues in drink-
ing water at such low concentrations. Furthermore
basic and acid environments, used a partition
the recent commercialization of automatic devices
with chloroform followed by a sample clean up using
has certainly helped in the development of on-line
a silica cartridge. There were no signiRcant dif-
trace enrichment methods in environmental analysis,
ferences in the triazine recovery using the two
because the sequence can be totally automated using
methods.
systems such as the Prospect module (Spark Holland)
Solid matrices can also be extracted by SPE with
or the OSP-2 system (Merck).
cartridge or disc devices but require a separate hom-
On-line SPE-LC is the most common procedure
ogenization step and other laborious processes. The
used because it is easily performed in any laboratory.
reported recoveries are lower than those obtained
The extracted compounds are eluted directly from the
with water, and the addition of methanol or acetonit-
precolumn to the analytical column by a suitable
rile as organic modiRer is necessary. However, these
mobile phase, which permits the separation of the
recoveries are comparable to those obtained by other
trapped compounds. It is well established that on-line
well known extraction methods for solid matrices.
procedures enable lower concentrations of pesticides
Table 10 gives a comparison of the features of three
to be determined, and most compounds can be kept
extraction procedures for tribenuron methyl analysis
within EU limits. Table 11 illustrates the improve-
in soil. Solid-phase and supercritical Suid extractions
ment in detection limits obtained for triazine and
are the most adequate in terms of recovery percentage
phenylurea herbicides using on-line procedures when
and precision, with acceptable detection limits; never-
compared with off-line ones.
theless, the recovery is affected by the amount of
Breakthrough is the key parameter in on-line SPE
herbicide present in soil. SPE can also be performed
because it indicates the sample volume and the
by blending directly a homogenized sample with
amount of analyte that can be preconcentrated. Two
C18 sorbent, transferring the mixture to a glass
factors can be responsible for breakthrough: insuf-
chromatography column and eluting the analytes
Rcient retention of the analytes by the sorbent and
with appropriate solvent.
overloading of the sorbent. One important factor of
The SPE of matrices other than water requires
the concentration procedure is the selection of the
further investigation.
Table 10 Comparison of the extraction procedures for tribenuron methyl analysis
Extraction Efficacy Precision Selectivity Operation time Affecting factors Detection limit
Solvent # ## ### ## No data No data

Solid-phase ## ### ### ## Concentration ###
Supercritical fluid ## ### ### ### Concentration ###
#, Bad; ##, regular; ###, good.

(Reproduced with permission from Berna JL, JimeH nez JJ, Herguedas A and Atienza J (1997) Determination of chlorsulfuron and
tribenuron-methyl residues in agricultural soils. Journal of Chromatography A 778: 119}125.)
Table 11 Range of linearity, r 2 and detection limit (LOD) for the on-line method
Pesticide Off-line method On-line method
Range of linearity r2 LOD (g L\1) Range of linearity r2 LOD (g L\1)
(g L\1) (g L\1)
Simazine 0.5}50 0.9985 0.1 0.1}8 0.9990 0.03

Cyanazine 0.5}50 0.9973 0.1 0.1}8 0.9987 0.03
Chlortoluron 0.5}50 0.9960 0.1 0.2}8 0.9956 0.05
Atrazine 0.5}50 0.9980 0.1 0.1}8 0.9999 0.03
Isoproturon 1.0}50 0.9990 0.1 0.2}8 0.9993 0.05
Ametryn 0.5}50 0.9985 0.05 0.1}8 0.9993 0.03
Prometryn 0.5}50 0.9989 0.05 0.1}8 0.9995 0.03
Terbutryn 0.5}50 0.9980 0.05 0.1}8 0.9985 0.03
Chlorpyriphos-methyl 2.0}50 0.9962 0.5 0.5}8 0.9980 0.20
Fenitrothion 2.0}50 0.9983 0.5 0.5}8 0.9993 0.20
Fenchlorphos 5.0}50 0.9950 1.0 1.0}8 0.9927 0.30
Parathion-ethyl 5.0}50 0.9944 1.0 1.0}8 0.9995 0.30
(Reproduced with permission from Aguilar C, Borrull F and MarceH RM (1996) On-line and off-line solid-phase extraction with
styrene-divenylbenzene-membrane extraction disks for determining pesticides in water by reversed-phase liquid-chromatography-
diode array detection. Journal of Chromatography A 754: 77}84)
sorbent, which must allow a convenient break- limits and quantiRcation. Figure 2 shows some
through of the analytes. Table 12 shows a compari- chromatograms obtained with different waters.
son of SPE sorbents for analysis of phenyl carbamate In spite of the presence of interference peaks, it can be
herbicides. The results were unsatisfactory with some seen that making a good choice of preconcentration
herbicides. GCB is not used much in online SPE parameter and analytical conditions, allows low
because it is not sufRciently pressureresistant. levels of many pesticides to be determined, even in
Another factor in the procedure is to evaluate highly contaminated surface waters.
the maximum sample volume that can be precon- In this way, the EPA in the United States currently
centrated without breakthrough of analytes, thus offers an on-line SPE procedure followed by
avoiding peak broadening. Generally, 50 mL was HPLC for the analysis of acidic herbicides in drinking
considered as optimum, but it could be increased for water. The sample is Rrst adjusted to pH 12 to hydro-
a particular kind of herbicide. lyse esteriRed analytes, then it is acidiRed to a pH of
It should be taken into account that sorbents used 1 and a 20-mL aliquot is pumped through a reversed-
in on-line SPE are not selective and numerous com- phase concentration column. By use of a switching
pounds from the matrix of natural samples are pre- valve, the concentration column is then pumped in
concentrated and can be eluted with the analytes of line with the analytical column and the sample con-
interest. Interferences depend on the nature of the stituents are then passed to the analytical column for
water. They have an effect on both detection separation and detection.
Table 12 Average recoveries and RSDs (%) of the analytes by the proposed on-line SPE-LC-DAD procedures in environmental
water samples spiked at different levels
Compound C18 pre-column PRP-1 pre-column
Drinking water Surface water Drinking water Surface water
0.5 g L\ 1
4 g L\ 1
0.5 g L\ 1
4 g L\ 1
0.2 g L\ 1
1 g L\ 1
0.2 g L\1 1 g L\1
Carbetamide * 105 (3) * 105 (4) 84 (12) 101 (3) 102 (8) 101 (3)
Propham 101 (2) 98 (3) 99 (3) 97 (3) 90 (5) 102 (5) 97 (6) 102 (3)
Desmedipham 84 (9) 86 (8) 94 (7) 98 (7) * * * *
Phenmedipham 87 (2) 97 (6) 98 (10) 108 (7) 87 (3) 101 (2) 93 (6) 104 (3)
Chlorbufam 105 (5) 99 (2) 102 (4) 97 (1) 106 (7) 101 (3) 99 (5) 105 (2)
Chlorpropham 103 (2) 99 (1) 108 (3) 106 (2) 105 (5) 101 (4) 99 (5) 108 (2)
(Reproduced with permission from Hidalgo C, Sancho JV, LoH pez FJ, and HernaH ndez F (1998) Automated determination of phenylcar-
bamate herbicides in environmental water by on-line trace enrichment and reversed-phase liquid chromatography-diode array
detection. Journal of Chromatography A 823: 121}128.)
Figure 2 On-line analysis of 150 mL of different water samples spiked with 0.3 g L\1 of (1) simazine, (2) methabenzthiazuron, (3)
atrazine, (4) carbaryl, (5) isoproturon, (6) propanil, (7) linuron, (8) fenamiphos, (9) fenitrothion and (10) parathion. Precolumn, PRLP-S.
(A) Blank gradient; (B) Milli-Q-purified water; (C) drinking water; (D) surface water from the Seine (28 June 1993). (Reproduced with
permission from Pichon V and Henion MC (1994) Determination of pesticides in environmental water by automated on-line trace-
enrichment and liquid chromatography. Journal of Chromatography A 665: 269I281.)
On-line SPE-GC is another interesting approach uncoated, deactivated capillary precolumn, also
that has gained popularity over the last few years. known as a retention gap, which accommodates the
The SPE-GC coupled techniques generally use an liquid SPE eluent while it vaporizes, thereby providing
Figure 3 SPE-GC-NPD chromatograms obtained after preconcentration of 10 mL of (A) HPLC grade water, (B) Amsterdam drinking
water, and drinking water spiked with (C) triazines (0.1 g L\1) and (D) OPPs (0.03 g L\1). Peak assignment for the herbicides: S,
simazine; A, atrazine; P, propazine; SB, secbumeton; T, trietazine; and TB, terbutylazine. GC programme: 753C during sample
introduction, then to 3003C at 153C min\1; held at 3003C for 5 min. (Reproduced with permission from PicoH Y, Louter AJH, Vreuls JJ
and Brinkman UATh (1994) On-line trace-level enrichment gas chromatography of triazine herbicides, organophosphorus pesticides,
and organosulfur compounds from drinking and surface waters. Analyst 119: 2025I2031.)
solute preconcentration. Figure 3 shows typical re- E savings in amount of solvent used
sults for the SPE-GC-nitrogen phosphorus detector E an improved chromatographic separation
(NPD) analysis of triazines. The most striking obser- E a reduction of sample volume needed to achieve
vation is the good baseline stability, because NPD is good results (up to 200 mL)
a very selective detector. The drawback of this tech- E a ten-fold improvement detection limit over that
nique is the high cost involved, which makes it unaf- required by EPA and EU regulations (limit values)
fordable by most of the laboratories involved in
herbicide analysis. The advantages cited for on-line procedures are
The analysis of herbicides, using an automated convenient for some analysts, but many prefer the
on-line solid-phase extraction device results in: off-line approach, which gives a convenient ex-
tract in an organic solvent suitable for multiple ana-
E a reduction in error lyses. Moreover, such an extract is generally much
E a more efRcient use of time more stable than the aqueous sample from which it
was derived, and is therefore more suitable for long- Capillary electrophoresis (CE) is very much suited
term storage. Also, the off-line approach allows for those analytes that are not amenable to GC or
the processing of many samples at one time, an ap- when existing LC methods do not offer sufR-
proach which is generally more productive in lab- cient separation power. Many impressive CE separ-
oratories that are not fully automated. ations, including the separation of triazines and
Both off-line and on-line techniques are not sulfonylureas, have been demonstrated in the last few
mutually exclusive. The possibility of employing both years. The main disadvantages of these techniques are
methods gives the analyst more tools at his/her dis- its inadequate detection limits and lack of selective
posal for performing the analysis adequately. detectors for the determination of residues in environ-
mental matrices. As a result of coupling with SPE, use
of CE has become competitive in trace analysis, and
Future Developments the door has been opened to environmental applica-
Today, SPE has become generally accepted as the tions in real matrices. Thus, the potential of CE is
analytical method of choice for determination of all very good indeed.
major herbicide groups in water. It is suitable for
detecting approximately 300 pesticides and pesticide- See also: II/Extraction: Solid-Phase Extraction. III/Im-
related compounds and has undergone rigorous munoaffinity Extraction. Porous Graphitic Carbon:
multi-laboratory calibration studies. SPE is also the Liquid Chromatography. Solid-Phase Extraction with
Cartridges. Sorbent Selection for Solid-Phase Extrac-
backbone of residue analysis protocols for govern-
tion.
ment agencies such as the EPA in the US. However,
there is still much to be done. The development of Further Reading
new, more selective supports for SPE, its coupling
with high separation power techniques, such as capil- BarceloH D (ed) (1993) Environmental Analysis. Techniques,
lary electrophoresis (CE), and its application to Applications and Quality Assurance. Amsterdam:
extract herbicide residues from solid samples, may Elsevier.
further reduce the detection limit and will represent BarceloH D and Hennion MC (eds) (1997) Trace Determina-
tion of Pesticides and Their Degradation Products in
an exciting challenge for researchers working in the
Water. Amsterdam: Elsevier.
area of herbicide residue analysis. Berrueta LA, Gallo B and Vicente F (1995) A review of
Looking to the future, it is interesting to note that solid-phase extraction: basic principles and new devel-
new SPE sorbents involving antigen}antibody in- opments. Chromatographia 40: 474}483.
teraction, so-called immunosorbents, have been CsarhaH rti T and ForgaH cs E (1998) Phenoxyacetic
described. Due to their high afRnity and high acids: separation and quantitative determination. Jour-
selectivity for these interactions, extraction and clean nal of Chromatography A 717: 157}178.
up of complex aqueous environmental samples is Dean JR, Wade G and Barnabas IJ (1996) Determination
achieved in the same step. Their application to ex- of triazine herbicides in environmental samples. Journal
tracts from solid samples is solvent-free and simpler of Chromatography A 733: 295}335.
than any other clean-up procedure. Two class-selec- Font G, Man es J, MoltoH JC and PicoH Y (1993) Solid phase
extraction in multi-residue pesticide analysis of water.
tive immunosorbents have been optimized up to now
Journal of Chromatography A 642: 135}161.
that enable the trapping of two groups of widely used International Union of Pure and Applied Chemistry (1994)
herbicides, phenyl urea and triazines. Analyte isolation by solid-phase extraction (SPE) on
Experiments have been recently designed to ex- silica-bonded phases. ClassiRcation and recommended
plore the possibility of recovering herbicide residues practices. Pure and Applied Chemistry 62: 277}304.
from food, soil, biological liquid and tissue samples Liska I, Kupcik J and Leclercq PA (1989) The use of solid
by SPE. For liquid matrices, such as plasma, urine, sorbents for direct accumulation of organic compounds
fruit juice or milk, acceptable residue recovery may be from water matrices. A review of solid-phase extraction
obtained almost without clean up. Before SPE can be techniques. Journal of High Resolution Chromatogra-
used with solid matrices (e.g. muscle, vegetables or phy 12: 577}590.
soil) a separate homogenization step and often mul- Nollet LML (ed.) (1996) Handbook of Food Analysis. New
York: Marcel Dekker.
tiple Rltration, sonication and centrifugation are
Pichon V (1998) Multiresidues solid-phase extraction for
required. Despite these drawbacks, SPE has been used trace analysis of pesticides and their metabolites in en-
a few times to extract residues of triazines, carba- vironmental waters. Analusis 26: M91}M98.
mates, ureas and other herbicides. More work is PicoH Y, MoltoH JC, Man es J and Font G (1994) Solid phase
needed to further develop SPE for use with the many techniques in the extraction of pesticides and related
different types of matrices that may contain her- compounds from food and soils. Journal of Micro-
bicide residues. column Separations 6: 331}359.

Chromatography Encyclopedia of Separation Science Elsevi

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chromatography Encyclopedia of Separation Science Elsevi

Uploaded by

Copyright:

Available Formats

2012 III / AMINO ACIDS / Thin-Layer (Planar) Chromatography

acetyl-cysteinyl and with OPA/N-L(D)-isobutyrylcys- See also: II/Chromatography: Liquid: Derivatization;

Thin-Layer (Planar) Chromatography

Table 1 Solvent systems for TLC of amino acids on silica gel

Solvent system Ratio v/v

Table 2 Solvent systems for two-dimensional TLC

Table 3 hRF (RF;100) values for amino acids on different layers

FXA FXB FXC

Ala 41.9 29.0 32.4 28.8 50.9 51.2 53.6

A, Baker Flex cellulose sheets; B, Baker Flex microcrystalline cellulose sheets; C,

Table 4 Group separation of amino acids

System as in Group Amino acids resolved

A I Leu, Phe, Try, Ala, Glu, Ser, Lys, Cys, Tyr

Table 6 Characteristic colours of PTH amino acids on polyamide plates containing

PTH amino acid Colour after

Second treatment Alkaline treatment

Valine Red Red

Table 8 Characteristic colours of PTH amino acids following ninhydrin application

PTH derivative Colour properties NH4OH colour change

Proline UV, colourless Light blue after heating

Silica gel plates, without fluorescent indicator, developed in heptane}CH2Cl2}propionic

Dns amino acid RF in solvent systems

Solvent systems: A, benzene}acetic acid (9 : 1, v/v) : B, toluene}acetic acid (9 : 1, v/v); C,

Table 10 Various solvent systems for TLC of dansyl amino acids

Solvent systems Ratio

Solvents at serial no. 1}8 : two-dimensional TLC on polyamide layers.

Table 12 hRF Values of 10 dansylamino acids on silica gel thin

Table 14 hRF of amino acids in presence of halides

Solvent system: n -butanol}acetic acid}chloroform (3 : 1 : 1, v/v); temperature 25$23C.

Solvent, butyl acetate}methanol}acetic acid}pyridine (20 : 20 : 5 : 5, v/v). Developing

Aminoacids Unimpregnated Plate impregnated with

Table 18 TLC of amino acids on impregnated silica gel layers

Solvent system Ratio (v /v) Impregnation

iso-Amyl alcohol}H2O}HOAc 6:5:3 Pyridinium tungstoarsenate

Sl. PTH-amino Alone Fe 2# Co 2# Ni 2# Zn 2#

PS1 M1 M2 M3 M4 PS2 M1 M2 M3 M4 PS3 M4

Table 22 TLC of PTH amino acids on impregnated silica gel layers

Solvent system Ratio (v/v) Impregnation a

CHCl3}H2O}EtOAc 28 : 1 : 1 Zn2#, Cd2#, Hg2#

1. CH2Cl2 3.5 5 Pro

Layers: 1, 2, Whatman C-18; 3, 5, Merck RP-18; 4, 6, Merck

propionaldehyde (43 mL) is added. The mixture is

Table 25 hRF Values of amino acid standards on normal-phase

Amino acid TLC system

Table 27 Enantiomeric resolution of -dialkyl amino acids by TLC

Parent amino acid R1 R2 RF value Configuration Mobile phase

Asp CH2CO2H CH3 0.52 (D) 0.56 (L) A

Mobile phase: A, methanol}water}acetonitrile (1 : 1 : 4, v/v); B, methanol}water}acetonitrile (5 : 5 : 3, v/v). Development distance

Table 28 Enantiomeric resolution of racemates by TLC

Racemate RF value Configuration Mobile phase

Valine 0.54(D) 0.62(L) A

A, Methanol}water}acetonitrile (1 : 1 : 4, v/v); B, methanol}water}acetonitrile (5 : 5 : 3,

Table 31 Separation data for enantiomeric compounds on -CD-bonded-phase plates

Compound, D,L mixture RF Mobile Detection

Dns-leucine 0.49 0.66 40/66 Fluorescence

Table 32 RP-TLC enantiomeric separation using vancomycin

Compound hRF Vancomycin

Dansyl DL-amino acid Pure L hRF from DL mixture Solvent system

DNP Amino Acids evaluated by measuring the optical density at 360 nm

Table 34 Results from resolution of dansyl DL-amino acids

Dansyl DL-amino acid Pure L hRF from DL mixture Solvent system

AMINO ACIDS AND DERIVATIVES:

phase mode, consequently, retention and selectivity

(mol L\1 cm\1) abs/ex (nm) em (nm) Remark