Series Editor's Preface

Many will agree that sampling and sample preparation are key parts of the ana-
lytical process. The reliability of analysis is based on the sampling process, stor-
age and preservation of samples, isolation of the analytes, clean-up and final
determination. In all these operations, sampling and sample preparation still
determine the overall analysis time and are the real bottleneck. This book on
Sampling and Sample PreparationTechniques for Fieldand Laboratory, edited
by J. Pawliszyn, addresses the above. The book contains 33 chapters covering
sampling, fundamentals and applications of extraction techniques, and new
materials. Particular attention has been given to the unification of the theory of
extraction, including solid phase micro-extraction, which became very popular
in the late 1990s because of its simplicity. Instrumentation aspects and integra-
tion with other techniques such as chromatography, capillary electrophoresis
and mass spectrometry are covered, followed by a variety of applications in envi-
ronmental, clinical, food and pharmaceutical analysis.
I am glad to introduce this new book in the Comprehensive Analytical Chem-
istry series, since it is the first in the series that comprehensively covers the topic
of sampling and sample preparation. This book is a good example of the "teach-
ing the teachers" approach by introducing sampling and sample preparation
methodologies in a step-by-step way, and at the same time being useful for
undergraduates and graduates as well as practising analytical chemists.
Finally, I am extremely grateful to everyone who has contributed to this
book. In particular, I would especially like to thank Janus Pawliszyn for doing a
great job as Editor. I am sure that this book will help to establish the science of
sampling and sample preparation with all its associated technologies in the field
of analytical chemistry.
D. Barcel

xxxi
Preface

Sampling and sample preparation are integral parts of the analytical process.
They should be part of the analytical chemistry teaching curriculum, but often
are not even mentioned during graduate or undergraduate analytical courses.
The primary reason for this situation is that sample preparation is not consid-
ered a separate part of analytical science with its unique challenges, but rather
the "thing" you do when you develop and perform analytical methods.
The main difficulty in recognizing the scientific principles of sample prepara-
tion is that the fundamentals of extraction, involving natural and frequently
complex samples are much less developed and understood, compared to physico-
chemically simpler systems used in the separation and quantification steps of
the analytical process, such as chromatography and mass spectrometry. This sit-
uation creates an impression that rational design and optimization of extraction
for complex systems is not possible. Therefore, the development of sample prep-
aration procedures is frequently considered to be an "art" but not a "science".
Relative to its significance in the overall success of analysis, there has been
little scientific interest in sample preparation. Little scientific interest in sample
preparation has also been associated with the fact that the technology has been
based, until quite recently, on very simple, "low tech", approaches such as
sample-solvent or sample-headspace partitioning, while underlying more scien-
tifically challenging problems associated with the sample matrix have been
ignored. This situation is currently changing with the introduction of non-
traditional technologies, which address the need for solvent-free alternatives,
automation and miniaturization. These approaches are frequently simpler to
operate, but more difficult to optimize, requiring more fundamental knowledge
by analytical chemists not only about equilibrium conditions, but, more impor-
tantly, about the kinetics of mass transfer in the systems. For some years, I have
been actively involved in teaching the fundamental aspects of modern sample
preparation technology to practitioners of analytical chemistry, mainly indus-
trial chemists. I recognized a need to provide the fundamental background, not
only to assist users, but also to help educators in developing their undergraduate
and graduate programs. Designing teaching programs to address the new devel-
opments in extraction technologies is quite difficult considering that the scien-
tific literature's emphasis is placed on the differences between techniques rather
than on their common features, which would facilitate general understanding.
The present trend in analytical instrumentation is towards miniaturization,
which results in portability. This development will eventually enable the attain-
ment of a major goal of the analytical chemist: to perform analysis at the place a

xxxiii
sample is located, rather than moving the sample to a laboratory, as is currently
the common practice in many cases. The new approach reduces errors and the
time associated with sample transport and storage, resulting in more accurate,
precise and faster analytical data. Simplification of sample preparation technolo-
gies and their integration with sampling and the convenient introduction of
extracted components to analytical instrumentation are significant challenges
to, and opportunities for, the contemporary analytical chemist. The design of
easy-to-use, but powerful, sample preparation tools will facilitate the convenient
implementation of integrated analytical technologies. The results of current
research will have a profound effect on the future of analytical methodology.
The purpose of this book is to address the needs and challenges outlined
above. To facilitate recognition of sample preparation as a separate science with
its unique challenges and research opportunities, the primary focus of the book
is on the fundamental aspects of extraction technology. Leading scientists in this
area have contributed chapters on modern aspects of liquid, solid phase and
membrane extractions with and without derivatization step as well as the chal-
lenges associated with different types of matrices. In the chapter on Unified The-
ory of Extraction, an attempt has been made to outline common features among
extraction technologies. The application chapters are dedicated to different
types of matrices, and focus on the impact of new technologies on the practice of
the science of sample preparation. There are some contributions that are con-
cerned with the characteristics of sample matrices, and their effects on method
development. Also, there are separate chapters focusing on sampling strategies
and equipment; many authors discuss this topic within their own contributions,
emphasizing the fact that extraction technologies should not be considered in
isolation but should be well integrated with the sampling and the introduction to
analytical instrument steps. This is particularly important when implementing
analytical technology directly on-site. Finally, there are a some chapters that
focus on new extraction materials, which could facilitate further improvement in
the technology.
The book is not intended to provide a comprehensive review of the topic of
sample preparation, but rather to be a first step towards a unified treatment of
analytical sample preparation technologies. It is hoped that it will be helpful for
learning more about sample preparation, as well as for encouraging an interest
in it by outlining the present practice of the technology and by indicating
research opportunities.
The book was developed during my scientific visit to Germany sponsored by
the Alexander von Humboldt Foundation award. I would like to thank my Ger-
man hosts, Karsten Levsen from Fraunhofer ITA, Hanover, and Peter Popp,
UFZ, Leipzig, for their hospitality, patience and encouragement during this
project.
Janusz Pawliszyn
Waterloo, Ontario

xxxiv
Chapter 1

Sampling and sample preparation for

indoor air analysis
Jacek A. Koziel

1.1 INDOOR AIR QUALITY

1.1.1 Historical perspective

The recorded history of public air quality awareness goes back to the smoke laws
in London in the Middle Ages. The modern era of air quality research started in
the early 1950s with the tragic London Fog episode which prompted epidemio-
logical studies and risk assessments of polluted outdoor air. Some early measure-
ments of selected indoor air pollutants started in the early 1960s. One of the first
gases measured was nitrogen dioxide (NO2 ) originating in combustion devices.
This was followed by measurements of some components of environmental
tobacco smoke (ETS) and radon in the 1970s. By the early 1980s it was generally
understood that indoor air pollution sources of volatile organic compounds
(VOCs), particulate matter (PM), and NO2 were typically far more significant in
overall personal exposure. Concerns about indoor formaldehyde, bioaerosols,
and allergens peaked in the 1980s and 1990s resulting in the publication of many
monographs, reports, and journal publications. In recent years, the wide use of
heating/air conditioning and energy-efficient building ventilation, the introduc-
tion of engineered building materials, and attached garages have often nega-
tively affected indoor air quality.
Four decades of research carried out by engineers, chemists, toxicologists,
epidemiologists, industrial hygienists, occupational health physicians, and psy-
chologists has resulted in much progress. Still, many unknowns are present and
many challenges remain. These include determination of sources, better under-
standing of release and migration mechanisms, improvements in real-time and
long-term monitoring, improved sampling and analysis methods, lowering of
detection limits, linking exposures with human health risk, improvements in
indoor air modeling and control, improvements in building materials, appli-
ances, consumer products, building codes, and public education.

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
1.1.2 Health effects or air pollutants
Many studies have reported statistical relationships between poor indoor air
quality gauged by concentrations of a specific pollutant, group of pollutants or
specific source or condition, and health effects. Health effects can vary from mild
irritation, headaches, nausea, respiratory infection and responses, cough, short-
ness of breath, allergies, perceived difficulty in concentration, and fevers to dis-
eases, birth defects, cancers, and death. It is often difficult to assess the risk of
exposure because of the lack of toxicity and epidemiological data for humans.
Generalized estimates, such as the total VOCs, cannot serve as risk indicators
because of the various toxicities of their components. Many of the indoor air pol-
lutants discussed in this chapter are considered highly toxic to humans or are
ubiquitous and highly concentrated in indoor environments. In addition, odor
detection threshold or visual assessment offers only limited assistance because
many of the known indoor air pollutants are odorless and colorless.

1.1.3 Sources of indoor air pollutants

Nearly all components of the indoor environment affect indoor air quality (IAQ).
The main components include sources and materials serving as sorption sur-
faces, and their activities and interaction with the outdoor environment. The
main indoor sources include building materials, paints, sealants, cleaning
agents, cosmetics, contaminated drinking water, appliances with combustion
processes, tobacco smoke, fuels, chemical storage, infiltration from outdoor air,
attached garages, cooking and cleaning activities. The presence of many indoor
sources combined with limited air exchange magnifies the resulting concentra-
tions. As a result, indoor air concentrations for many pollutants are much higher
than those in outdoor air. Elevated pollutant concentrations combined with
extended exposures often result in higher risks associated with indoor air.

1.2 MEASUREMENTS OF INDOOR AIR POLLUTANTS

1.2.1 General considerations

The assessment of indoor air quality is a challenging task. It is complicated by

the presence of multiple sources, complex chemistry between individual chemi-
cals, interactions with building materials, ventilation and heating/ cooling
arrangements, and occupant activities. Because of the wide variety of parame-
ters affecting indoor air, it is highly recommended to use a standardized survey
that can provide a comprehensive background to the concentration measure-
ments. The monitoring approach has to be adjusted to the individual case study
that optimizes the choice of target pollutants, appropriate sampling and analysis
method, budget and number of samples, sampling locations, sampling times and
duration, sampling preservation and custody protocols.

2
 Sampling and sample preparation strategy are the major part of the indoor
air quality assessment. The strategy considers (1) type of effect, e.g., acute vs.
chronic, (2) patterns of occupant exposure, (3) pollutant type and potential
health risks, (4) pollutant sources, (5) applicable regulations (for occupational
air), (6) applicable sampling and analysis methods, (7) site selection, (8) number
of samples and their statistical significance, and (8) budgetary, site access, and
logistical limitations. The choice of applicable sampling and analysis methods
continues to be one of the most challenging tasks. As sampling devices and ana-
lytical instruments become more sensitive and economical, air quality profes-
sionals are provided with a greater variety of choices. Ideal methods for indoor
air sampling and analysis could provide a direct (instantaneous/real-time) and
integrated (long-term, time-weighted average) automated reading in real time
for multiple pollutants of concern at multiple locations with no disturbance to
the sampled microenvironments. However, such a sampling is neither possible
nor practical, and the actual solution is the result of a compromise. On-site moni-
tors are often calibrated only to one or a few target pollutants. On-site monitors
for occupational monitoring cannot achieve the low detection limits required for
residential air and can be affected by interferences. Samples based on extraction
to a sorbent media via active or passive sampling or capture to a container
require sample preservation for later analysis in a laboratory. Attempts are
made to speed up and automate the sampling process by on-site analysis with a
portable gas chromatograph and/or mass spectrometer, liquid chromatograph,
and others. In such cases, various sample introduction methods can be used, e.g.,
solid phase extraction, thermal desorption, solid phase microextraction (SPME),
denuders, membrane extraction with solid interface (MESI), membrane inlet,
and other technologies as discussed in this book.
There are several types and categories of indoor air pollutant measuring sys-
tems depending on the sampler mobility, operating characteristics, and output
characteristics [1]. Personal, portable, and stationary samplers can be defined
using the mobility criteria. Personal devices can be worn or carried. These
devices are used to assess workplace exposure to target pollutants and are capa-
ble of a time-weighted average (TWA) measurement. Portable units can be
hand-carried during a survey. Stationary units must operate from a fixed loca-
tion. Depending on the operation mode, air sampling can be active or passive.
Active sampling requires a power source, e.g., pump or vacuum, to draw sample
to a sensor/analyzer or collector. Passive sampling relies on diffusion to acquire
samples. Finally, the measurement can be obtained from analyzers or collectors.
Analyzers provide an instantaneous/real-time (or nearly real-time) signal that
corresponds to the pollutant concentration. Analyzers are very useful in provid-
ing concentration including peak values. The output can be integrated to obtain
TWA concentrations. In contrast, collectors are often limited to TWA measure-
ments. No peak concentrations can be measured unless the averaging period is
very short and the detection methods are sensitive to analyze the lower mass col-
lected. TWA and peak concentrations are very important in indoor air sampling,

3
TABLE 1.1
Summary exposure guidelines for occupational air in the U.S.
Regulating/issuing agency Criteria for peak and averaged concentrations

ACGIH (American TLV: threshold limit value time weighted average (TWA). The
Conference of Governmental TLV is the concentration below which a healthy worker could
Industrial Hygienists) be exposed for 8 h/day, 5 days/week, for 20 years without
developing any disease. TWA means that the TLV can be
exceeded a time if it is counter-weighted by lower exposure.
STEL: shorfor t term exposure limit is a 15 min TWA exposure
that should not be exceeded at any time during a work-day.
OSHA (Occupational Safety PEL TWA: permissible exposure limit TWA. PEL TWAs are
and Health Administration) very similar to ACGIH TLVs but are issued as a regulation by
OSHA to protect workers.
NIOSH (National Institute REL TWA: recommended exposure limits TWA. REL TWAs are
for Occupational Safety very similar to ACGIH TLV TWAs. NIOSH conducts research
and Health) and issues recommendations but not regulations.
IDLH: immediately dangerous to life and health. IDLH values
reflect a concentration that is likely to cause death or
immediate or delayed permanent health effects.

particularly in occupational exposure assessment. A summary of toxic exposure

guidelines is presented in Table 1.1.
Passive samplers are a very popular and convenient means of collecting TWA
samples and for long-term exposure assessment of indoor air. They eliminate the
need for pumps, vacuum, and flow meters. Their small size and simplicity makes
them amenable to personal or fixed location monitoring that could be mailed to
or from a laboratory. Analytes from the vicinity of the sampler diffuse through a
diffusion layer, i.e. stagnant layer of air, and are trapped in a porous sorbent,
chemical reagent, or solvent [2]. TWA concentration (C) is determined based on
the diffusion coefficient (D), amount of analyte sorbed (m), exposure time (t),
length of the diffusion path (L), and cross sectional area of the sampler (A):
mL
CDA (1.1)
DAt
Passive samplers can have different geometry, e.g. "flat" badges and "long"
tubes. These devices include diffusion samplers, permeation samplers (with a
semipermeable, non-porous membrane), and indicator-type samplers (where
analytes react with a suitable reagent and change color or color intensity over a
length proportional to concentration). Determination of analyte in the sample is
based on (1) adsorption, absorption, or chemisorption, followed by desorption
and analysis, and (2) reaction, e.g., analyte concentrations are determined based
on change of mass, color, or electric conductivity [2]. Retracted SPME fibers can
also facilitate passive diffusive sampling as described elsewhere in this book. The

4
main factors affecting the preconcentration process on passive samplers are [2]:
(1) temperature, pressure, humidity, and air (face) velocities; (2) concentration
of analytes, response time and degree of sorbent or reagent saturation; and (3)
back-diffusion, loss, and decomposition of analytes.
This chapter attempts only to summarize sampling and sample preparation
methods for the most important indoor air pollutants with the focus on residen-
tial air. However, the reader is encouraged to seek more detailed information in
several excellent books written specifically on the background, occurrences,
measurements, health effects and controls of indoor air pollutants [1,3-11].
These and other resources contain up-to-date extensive summaries of the com-
mercial air sampling instrumentation that is widely available. Brown and
Monteith compiled gas and vapor sample collectors [12]. Saltzman and Caplan
summarized detector tubes, direct reading passive badges, and dosimeter tubes
[13]. Namiesnik and Gorecki summarized references for applications of passive
samplers to determination of 16 inorganic and 30 organic compounds or group of
compounds in air [2]. Woebkenberg and McCammon discussed direct-reading
gas and vapor instruments [14]. These summaries contain information about
specific compounds, analysis methods, detection limits, instrument weight and
dimensions, and the name of manufacturer and/or supplier. Internet resources
have become very useful in searching for standard methods. These include the
U.S. Environmental Protection Agency (EPA), National Institute for Occupa-
tional Safety and Health (NIOSH), materials and chemicals, instrumentation
and buyers guides (Air &Waste Management Association), and up-to-date litera-
ture searches.

1.2.2 Organic pollutants

Nearly 1000 organic pollutants were known to be present in indoor air by the
end of the 1980s [15,16]. The list is continuously growing as new and more sensi-
tive detection methods are being developed. The concentrations of many of these
organics are often greater indoors than outdoors. Boiling point range is one of
the criteria for classification, e.g., very volatile organic compounds (WOC), vola-
tile organic compounds (VOCs), semivolatile organic compounds (SVOC), and
organic compounds associated with particulate matter (POM). Additional crite-
ria include reactivity, environmental importance, sampling considerations, and
chemical structure. Protocols for collection and analysis of organic pollutants
are being standardized by NIOSH, U.S. EPA, and other agencies [5,17]. This
subsection summarizes sampling and sampling methodologies for major groups
of organic compounds in indoor air.

1.2.2.1 Volatile organic compounds

The definition of VOCs used by IAQ practitioners is somewhat broad and is com-
monly interpreted as all organic vapor-phase compounds [18]. VOCs are ubiqui-
tous in indoor air and originate from almost all building materials, furniture and

5
 Tenax -1.5 grams (60 mm bed depth)
Glass wool plugs 5 mm long

Glass cartridge (13.5 mm OD x 100 mm long)

Tenax -1.5 grams (70 mm bed depth)

Swageiok Glass wool plugs 5 mm long
fitting

Metal cartridge (12.7 mm OD x 100 mm long)

Sampling Siianized glass wool plugs 5 mm long Desorption

direction direction

Carbotrap C Carbotrap Carbosieve S-Ill

graphitized graphitized molecular sieve
carbon black carbon black (traps light
(traps heaviest (traps C5-C8 compounds)
compounds) compounds)

Fig. 1.1. Schematic of typical designs of adsorbent tubes.

clothing, consumer products, fuels and combustion products, pesticides, drink-

ing water and wastewater. VOCs can also penetrate from outdoor air and seep
from contaminated soil or groundwater. Many of the VOCs are considered haz-
ardous air pollutants (HAPs). Typical concentrations measured range from less
than 1 ppbv to more than 1 ppmv [19]. The most common methods include col-
lection on some type of solid sorbent, followed by thermal desorption or solvent
extraction [20-24]. Sorbents used with solvent desorption include silica gel, acti-
vated charcoal, Anasorb 747, carboxens, porous polymers, and carbon molecular
sieves. Sorbents used with thermal desorption include Tenax TA, Chromosorb
106, graphitized carbons, carbon molecular sieves, and multi-bed combinations
[22]. The U.S. EPA Method IP-1B uses solid adsorbent (Tenax) tubes (Fig. 1.1)
sampling followed by thermal desorption and analysis on a gas chromatograph-
mass spectrometer (GC-MS) (Fig. 1.2) [5]. This method is recommended for
many VOCs with a boiling point between approximately 80 and 200°C. In some
cases, this method is applicable for SVOCs. Other detectors such as flame ioniza-
tion detector (FID) and electron capture detector (ECD) may also be used.
The advantages of the Method IP-1B are related to the small size, portability,
and capability of collecting time-weighted average samples for a wide range of
sampling times from approximately 30 min to 12 h. Methods limitation include

6
 sampling sampling I nermal
durtc Trap GC
pump pump undesorption
unit
Fig. 1.2. Schematic of sampling with adsorbent tube and analysis with a thermal
desorption-GC-MS system.

the background contamination of Tenax with toluene and benzene from manu-
facturing, production of false nitrosoamines on Tenax that is in contact with
NOx, low breakthrough volumes for very volatile compounds (e.g., vinyl chlo-
ride), dilution of analytes when solvent desorption is used, accuracy of air flow
rate estimates and pump stability, and often incomplete recovery. The repro-
ducibility is generally between 10 and 30%. Method detection limits depend on
detector responses for each compound, and can be as low as 0.1 ng for MS.
The U.S. EPA Method IP-1A uses pre-evacuated SUMMA canisters for sam-
ple collection (Fig. 1.3) followed by trapping on a cryogenic trap and analysis on a
GC-MS or other GC detector [5]. The advantages of the canister collection
method include the ease of collecting 24 h samples, and ease of deployment,
transportation, and shipping. The disadvantages include incomplete recovery
and incomplete canister clean-up, up-front capital cost and the need to use a spe-
cialized GC interface for sample introduction.
In recent years, solid-phase microextraction has been used as a precon-
centrating device and sample introduction device [25] or as a field sampler capa-
ble of collecting a sample in the field, retaining it, and introducing it into a
conventional GC [26,27]. Sampling times [28] as short as 30 s for spot sampling
to as long as 8 h for time-weighted average can be used for selected VOCs and
SVOCs and quantitative analysis [29-31]. The limits of SPME are associated
with sample losses that can be faster than for many conventional methods and
selected analytes, and the need for determination of several physico-chemical
parameters in addition to typically measured sampling temperature, pressure,
and relative humidity [32].
Passive samplers are capable of collecting samples at a specific rate con-
trolled by the diffusion through a static layer of air or membrane [33,34]. Pas-
sive/diffusive samplers do not require pumps or any other means to move air,
thus they are robust and easily deployable. Samplers use sorption or reactive
sampling to trap analytes. The uptake rate depends on the analyte concentration
in air, the sampler area and length of the static air gap, and the diffusion coeffi-

7
 Fig. 1.3. Sampler configuration for SUMMA canister sampling.

cient of the analyte in air. Air velocity, excessive humidity, sorbent starva-
tion/overload, sample loss, and reverse diffusion may interfere with the accurate
measurement. Typically, passive samplers such as a badge or tube for workplace
sampling can be adapted to indoor air sampling by extending exposure time and
selecting a more sensitive analytical instrumentation. Badges have a higher dif-
fusive uptake than tubes and use stronger absorbents, thus the reverse diffusion
is less likely. However, badges use solvent desorption that is difficult to auto-
mate, and solvent dilution decreases sensitivity [35]. Tube samplers have lower
diffusive uptake rates, use thermal desorption which can be automated, and the
tubes can be reused if the VOC recoveries are satisfactory. Comprehensive guid-
ance on designing a sampling strategy using passive samplers and applications of
diffusive samplers are presented elsewhere [35,36].
Typical badges consist of a charcoal wafer with 1 cm diffusion length, e.g.,
OVM 3500 (3M). Sampling times range from 24 h to 3 weeks. The surrogate
uptake rate per 8 h period as reported by the manufacturer is approximately 30
ml/min. This uptake depends on the analyte. Several research groups used pas-
sive samplers to study VOCs in indoor air [37-40]. Reported mean concentra-
tions of VOCs ranged from less than 1 to several hundred ALg/m 3 for individual
VOCs. Detection limits depend on the sampling time and analyte and typically
start at approximately 1 g/m 3 .
One of the most popular tube diffusive samplers (from Perkin-Elmer) con-
sists of a 90 mm long steel tube (5 mm ID) with a Tenax TA (or other sorbent)
retained by a stainless-steel mesh 15.3 mm from the opening. The opening is also
protected by another stainless-steel mesh to minimize convection inside the
tube. The opposite end of the tube is sealed with a Swagelok fitting. Sampling
time can range from 24 h to 4 weeks. Analysis is typically completed using a ther-

8
mal desorption (TD) CG and FID or MS detector. Several comprehensive indoor
air studies of VOCs involving homes and offices are described in the literature
[33,41,42].
Real-time monitoring devices are very advantageous for screening indoor air
quality because they can detect the presence of intermittent sources, determine
the most current concentration(s), monitor concentration trends, and monitor
effects of variables influencing air quality. Some may be hand-held and can be
used to find a source by following a concentration gradient. However, their
response is often non-specific because the output signal is the sum of instrument
responses to many compounds. The most popular types of detector are based on
photoacoustic spectroscopy (PAS), FID, photoionization detector (PID), non-
dispersive infrared (NDIR), and Fourier transform infrared (FTIR). The PAS
uses the fact that many VOCs absorb infrared light at 3.4 Aum. Absorption of the
pulsed light by the hydrocarbons that pass through a PAS cell generate changes
in pressure that is detectable by microphones and can be related to a surrogate
total VOC concentration [43]. Interferences from water vapor are compensated
by subtracting its concentration measured by another infrared (IR) wavelength.
The PAS detects a wide range of organic compounds from VVOCs to VOCs. The
PAS is very sensitive to n-alkanes, e.g., methane, while responding very poorly to
chlorinated organics [23]. It is possible to adjust the IR wavelength to provide an
enhanced sensitivity for a specific compound, e.g., formaldehyde. However,
interferences from other analytes should be considered [42].
The PID detector ionizes VOCs and VVOCs with UV radiation and detects an
electric current caused by ions and related to the concentrations of analytes
present. Typically, gases with a double bond, e.g., aromatic hydrocarbons,
respond very well. Responses of chlorinated organics are typically very low. Dif-
ferences in response between individual VOCs can be greater than one order of
magnitude. In a flame ionization detector, organic compounds are burned in a
hydrogen flame and produce ions. These ions are collected on an electrode where
they can be amplified. The produced current can be related to the concentration.
The signal is also dependent on the number of carbons and the type of analyte.
For example, response to n-alkanes and aromatic hydrocarbons is very similar,
while it is much lower for oxidized and chlorinated compounds. The NRIR and
FTIR can be calibrated for a specific compound [45,46].

1.2.2.2 Aldehydes
The aldehydes include a large family of compounds that have H-C =O functional
groups. Of these, formaldehyde, acetaldehyde, glutaraldehyde, and acrolein are
the most important aldehydes with respect to indoor air quality because of their
ubiquitous nature and toxicity. Formaldehyde is released from adhesives used in
particle-board and hardwood manufacturing, from textiles that were subjected
to permanent press treatment, from acid-cured coatings, and from selected
paper, insulation, and hard plastic products. It is also produced during combus-
tion or thermal oxidation in combustion appliances, wood burning, and tobacco

9
smoke. Formaldehyde is classified as a probable carcinogen. Several studies have
shown a significant dose-response relationship between formaldehyde concen-
trations in mobile homes and houses with particle-board underlay and health
effects which ranged from eye irritation and sore throat to chest pain, fatigue,
sleeping problems and depression [47,48]. Indoor air acetaldehyde, acrolein, and
glutaraldehyde are produced when organic matter is combusted. Similar to
formaldehyde, these aldehydes are also off-gassed from selected hard plastic
products, and many consumer products such as cleaning agents, cosmetics,
drugs, and textiles.
The most popular sampling methods for airborne aldehydes involve derivati-
zation because of the relatively low concentrations and high reactivity [49]. The
U.S. EPA proposed three methods for sampling and analysis of aldehydes in air:
sorbent tube sampling followed by analysis with HPLC, continuous colorimetric
gas analyzer, and passive sampling followed by analysis with HPLC [5]. The U.S.
EPA Method IP-6A utilizes tubes with 2,4-dinitrophenylhydrazine (DNPH)-
coated sorbent. This method is specific for formaldehyde, but it can be modified
to detect fourteen other aldehydes by optimizing chromatographic conditions,
e.g., using two columns in a series or by changing mobile phase composition
through a gradient program. Derivatizing agent reacts with a carbonyl group to
form more stable derivatives. The sampling time can vary from 5 min to 24 h.
Recommended sampling flow rate is between 500 and 1200 ml/min. It is also rec-
ommended that the derivatizing agent is added in situ on the silica gel adsorbent
because of abnormally high formaldehyde background in some of the commer-
cial pre-coated tubes [5]. Used tubes are desorbed with acetonitrile and stored
until analysis. The DNPH-aldehyde derivative is determined using isocratic
reverse phase HPLC and UV absorption operated at 360 nm. In some cases, an
ozone denuder may be used if high ozone concentrations may interfere with
DNPH and the derivatization product. The method detection limits vary from
approximately 1 ppbv for 10 1 sample volume to 0.01 ppbv for 1000 1 sample
volume.
The U.S. EPA Method IP-6B uses a continuous colorimetric analyzer for
determination of formaldehyde and other aldehydes [5]. Any aldehyde present in
air is scrubbed with sodium pararosaniline (PRA) in the presence of sodium sul-
fite (Fig. 1.4). The formed dye is directed through a flow cell where transmission
of light is measured by photodetectors at 550 nm. The intensity of color is pro-
portional to formaldehyde concentration. The method detection limits are 3
ppbv with the standard range up to 5 ppm. Method IP-6C uses passive diffusive
badges for determination of formaldehyde and other aldehydes (Fig. 1.5) [5].
This method is best suited for long sampling times up to weeks to typical formal-
dehyde concentrations of less than 100 ppbv because of low sampling rates. The
method utilizes a glass filter coated with DNPH that is placed behind a diffusion
gap. After analysis, the filter is desorbed with acetonitrile and analyzed with
HPLC and UV detector at 360 nm. The linear range for this method is approxi-
mately 0.05 to 20 Ag/l.

10
 Reagents Flow meter

Urall

Fig. 1.4. Schematic of continuous formaldehyde analyzer.

=>_____________~_
_~Retaining clip

_.< Perforated plate

200 Mesh diffusion screen
- Retaining clip

Assembled passive sampling device

Fig. 1.5. Passive sampling device for personal sampling.

Other derivatizing agents and methods can be used to extract aldehydes, e.g.,
chromotropic acid method, MBTH method, AHMT method, and acetylacetone
method [50]. The MBTH method (3-methyl-2-benzothiazolinone-hydrazone) is a
non-selective colorimetric method for the determination of aldehydes of low
molecular weight. This method quantifies total aldehydes as equivalents of
formaldehyde. MBTH reacts with aldehydes forming an azide. In parallel, a reac-
tive cation is formed by oxidation of MBTH with Fe(III). The resulting blue ionic
dye can be analyzed for absorbance at 628 nm [50]. This method is less sensitive

11
than the pararosaniline method and strong reducing agents can interfere with
aldehydes. AHMT (4-amino-3-hydrozino-5-mercapto-4H- 1,2,4-triazole) is used
in a strong alkaline media to facilitate reaction with aldehydes to a colorless
product which is then oxidized to form a purple-colored dye [50]. The dye can be
analyzed by colorimetry at 550 nm. This method is most sensitive for formalde-
hyde. The acetylacetone (2,4-pentanedione) method works very well for formal-
dehyde. Air is pumped through a solution, where the Hantzsch synthesis that
involves cyclization of acetylacetone in the presence of formaldehyde and ammo-
nium nitrate to form dihydropiridine-3,5-diacetyl-1,4-dihydrolutidine (DDL)
occurs. The quantification is performed using colorimetry at 412 nm or fluores-
cence at 510 nm [51].
A novel method for sampling and determination of formaldehyde based on
SPME is also available. In this method o-(2,3,4,5,6-) pentafluorobenzyl hydroxy-
amine (PFBHA) is first coated on a PDMS/DVB 65 gm fiber using headspace
extraction for approximately 2 min immediately before sampling. On-fiber
derivatization of formaldehyde occurs when the fiber is exposed to air. Sampling
times can be as low as few minutes for spot sampling with an exposed fiber to as
long as 8 h for time-weighted average sampling with a retracted fiber [30,52].
SPME-based samples are analyzed on a GC-FID. Method detection limits are
approximately 5 ppbv. Thermal desorption tubes have also been used for sam-
pling and determination of aldehydes, albeit with a lesser success than afore-
mentioned methods. Studies with gas standards have shown that the Tenax TA
sorbent is the most effective sorbent for aldehydes from butanal to decanal [53].
However, Tenax TA is not appropriate for acetic aldehyde, propenal, and
propanal.

1.2.2.3 Pesticides
Pesticides are used to control, repel, or kill pests, e.g. insects, weeds, fungus, or
rodents [54]. There are at least 600 pesticides and approximately 50,000 pestici-
dal formulations. Pesticides have bioaccumulating properties and can cause
chronic health effects [5]. On average, the major exposure to pesticides occurs
inside the home [55]. Most pesticides are semi-VOCs and are present in air pri-
marily as gases. These gases can also be volatilized from particulate matter after
sample collection [55]. Pesticides can also be simple inorganic substances,
organometals, VOCs, and nonvolatile organic compounds. Indoor air sources
include treatments for cockroaches, termites, ants, and other insects, disinfec-
tants for kitchen and bathroom, room deodorizers, laundry aids, outdoor lawn
treatments, and others [55].
Because indoor concentrations are typically low, long sampling times, e.g. 24
h, are usually selected to collect enough detectable mass. The U.S. EPA Method
IP-8 for organochlorine pesticides uses active sampling and adsorption on poly-
urethane foam (PUF) [5] (Fig. 1.6). This method is a modified Method EPA 608
3
and is recommended for concentrations greater than 0.01 Atg/m for 4-24 h sam-
pling time and sampling rate of 1-5 1/min. Pesticides are extracted from the PUF

12
 Glass tube Polyurethane foam PTFE filter gasket Screw cap

Sampling pump support screen Particle filter Air sample

Fig. 1.6. Schematic of simple air sampling cartridge with PUF sorbent and particulate filter for
pesticide sampling.

cartridge with 5% diethyl ether in hexane and determined by GC equipped with

an electron capture detector (ECD), nitrogen-phosphorus detector, flame photo-
metric detector, or MS. For some pesticides, HPLC coupled with UV detection
can be also used. Polychlorinated biphenyls (PCBs) can interfere with determi-
nation of some pesticides. Column chromatography can be used to separate
PCBs from pesticides [5]. A modification to this method can include a separating
nozzle for particle size cut-off [52]. Other sorbents such as Tenax GC, Tenax AT,
Amberlite XAD-2, Amberlite XAD-4, Poropak-R, Chromosorb 102, and Florisil
are also suitable for extraction of pesticides from air [54]. Detection limits
depend on the sample volume, type of pesticide, and the type of detector. For typ-
ical GC or HPLC-based methods, a sampling volume of 0.01-1 m3 is sufficient for
typical occupational concentrations from 0.01-10 mg/m3. Sampling volumes
from 1-10 m3 are needed to achieve 0.01-10 .g/m3 detection levels typical in
non-occupational settings [54].
Several NIOSH methods have been developed for organochlorine and
organophosphate pesticides in occupational air, e.g., NIOSH-5602, 5502 [17].
These methods use small traps with particle filter upstream from two Amberlite
XAD-2 sorbent beds. Personal sampling pumps capable of drawing 0.2-1 I/min
are used. Detection limits are typically between 5 and 600 ng/m3 range for sam-
ple volumes from 60-240 1. There are two ASTM methods, i.e., ASTM standard
D4861 for a wide range of pesticides at concentrations in the 0.001-50 g/m3
range, and ASTM standard D4947 for chlordane and heptachlor [56,57]. The
ASTM standard D4861 is based on the U.S. EPA Method TO-10A. The ASTM
and EPA methods are designed for 4 1/min samplers and sampling volumes up to
6 m3 resulting in MDLs from 1-100 ng/m 3 .

1.2.2.4 Polycyclic aromaticcompounds, phthalates, and phenols

Polycyclic aromatic compounds (PAHs) are formed during incomplete combus-
tion in space heaters, environmental tobacco smoke, cooking, grilling, and fry-
ing. Outdoor sources, such as diesel engines can also contribute to indoor PAH
concentration. PAHs often found in indoor air include naphthalene, phen-
anthrene, chrysene, benzo(a)pyrene, and 1,8-dinitropyrene. Many PAHs and
nitro-PAHs are suspected or proven carcinogens or mutagens. An exhaustive

13
summary of common indoor PAHs and nitro-PAHs and their concentrations can
be found elsewhere [58]. Lower molecular weight PAHs, e.g., 2 to 4-ring, are
likely to be found in gaseous phase, while heavier PAHs occur in the particulate
phase. Sampling of PAHs typically requires the use of quartz fiber filters and
sorbents because of the partitioning of PAHs in gaseous and particulate phase.
U.S. EPA Method IP-7 uses a combination of quartz fiber filters and XAD-2
or PUF sorbents [5]. Approximately 30 m3 is drawn through a filter situated
upstream from an adsorbent cartridge at 20 1/min. The filter and cartridge are
extracted by Soxhlet extraction with either methylene chloride (for XAD-2) or
hexane (for PUF), concentrated by a Kuderna-Danish evaporator, and followed
by analysis on GC-MS or HPLC and UV or fluorescence detection. In some cases,
e.g., flame ionization detection, an additional step of silica gel cleanup may be
needed. Method detection limits range from approximately 5-50 g/l for HPLC-
fluorescence, 50-250 Alg/l for HPLC-UV, to as low as 100 ng/l for GC-MS [5].
Phthalates, a.k.a. phthalate esters, are the most commonly used plasticizers
in PVC resin [58]. Phthalates are used to manufacture a wide variety of con-
sumer products and building materials, e.g., food containers and wraps, toys,
tiles, shower curtains, vinyl upholstery, lubricants, sealers, and adhesives. Many
phthalates are classified as endocrine disrupters. Reported indoor concentra-
tions are as high as 1.5 ng/il compared to 0.05 ng/l for outdoor air [58]. Typical
sampling includes particulate collection on a glass fiber filter and gas phase col-
lection on PUF or XAD-2 [59].
Phenols in indoor air typically originate from phenol-formaldehyde bonded
wood products [50]. Alkylphenol polyethoxylates are common surfactants used
in detergents, cleaners, and personal hygiene products [58]. Some of these com-
pounds are likely involved in endocrine disruption. There are several methods of
sampling and analysis of indoor air phenolic compounds. Quartz filters can be
used in combination with PUF or XAD-2 sorbents. Filters and sorbents are then
extracted with methylene chloride, followed by derivatization and analysis on a
GC-MS using single ion monitoring [59]. Another method is based on pumping
air at 2 l/min through 0.1 M NaOH using Muencke-type absorbers [50]. After
neutralization with hydrochloric acid, a p-nitroaniline reagent is added to form
4-nitro-4'-hydroxyazobenzene (dye). The absorbance of total phenolic com-
pounds is measured at 490 nm. Interferences include aromatic amines, hydrogen
sulfide, and sulfur dioxide. The MDLs are approximately 0.8 Ag/m 3 for a total
sampling volume of 100 1 [50]. A variation to this method uses direct analysis
(without derivatization) of phenols after a neutralization step using HPLC and
UV (at 254 nm and 280 nm) and/or fluorescence detection. Fluorescence is pre-
ferred for alkylphenols [60].

1.2.3 Particulate matter

Airborne particulate matter is a complex mixture of organic and inorganic mate-

rial consisting of solid particles and low vapor pressure liquids with sizes ranging

14
from 0.01gm to more than 100 Am. Larger particles are typically solids that were
abraded or ground, whereas smaller size particle often originate in combustion,
condensation, or reaction as a fine nuclei, e.g., 0.05 jAm, which tend to grow to
more stable size below 1 m [5]. Coarse particles are defined as those particles
for which the aerodynamic diameter is less than 10 Am (PM-10). Coarse particles
can be inhaled; however, they are not capable of reaching deep into the lungs
where only fine particulate with an aerodynamic diameter less than 2.5 Am
(PM-2.5) can enter. PM concentrations indoors are affected by combustion
sources (environmental tobacco smoke, wood stoves, gas, and appliances), occu-
pant activities (cooking, vacuuming, and cleaning), and infiltration of outdoor
air. Indoor-to-outdoor ratios often exceed 1, depending on the type of activity,
time of day, ventilation type, and existence of sources [61]. Characterization of
air particles and aerosols is described in a separate chapter in this book.
The U.S. EPA proposed two methods for determination of respirable
particulates (defined as PM-10) [5]. U.S. EPA Method IP-1OA uses a micro-
environmental exposure monitor (MEM) as a fixed site monitor (Fig. 1.7) and a
personal exposure monitor (PEM) for estimation of personal exposure to parti-
cles. Both devices use a specially made nozzle and an impactor plate designed to
provide a predetermined size cut-off at a constant air flow rate of 4 1/min. A
tarred particulate filter is mounted downstream from the separator. For the
MEM device, the total volume sampled is approximately 5.5 m3 per day. The
average concentration is estimated by dividing the weight gain of the filter by
the total volume sampled. A 12 h level of detection (LOD) is typically 8 jg/m3 for
the PEM and 4 jg/m3 for the MEM. These LODs can be affected by the precision
of the collection and size cut-off efficiency, particle resuspension, microbalance
measurements, flow rate estimates, interactions between collected particles, and
interactions with the filter media. In cases where chemical analysis is also con-
duced in addition to mass gravimetric analysis, special filter materials that do
not interfere with sampled gases or analytical procedures should be used.

Air flow Air inlet

over

-10 cut-off

action plate

-2.5 cut-off
)action plate
rticle
lection filter

Pump

Fig. 1.7. Schematic of microenvironmental exposure monitor (MEM) for particulate matter.

15
 Diffusion denuders are used for selective collection of individual gases or
classes of gases upstream from the particulate filter [62,63]. Denuders allow for
separation of gases from particulates and minimizing interferences by preclud-
ing reactions between collected particulate and sampled gases. Sodium carbon-
ate, sodium hydroxide, and potassium hydroxide can be coated on the surface of
a denuder where they can effectively remove acidic gases (formic and acidic
acids), HNO3 , HC1, and SO2. There are several methods for analysis of collected
particulate: these include X-ray fluorescence, proton-induced X-ray emission,
neutron activation analysis, atomic absorption, and ion chromatography [11].
U.S. EPA Method IP-1OB uses a tapered element oscillating microbalance
(TEOM) continuous particulate monitor [5]. This instrument is equipped with
an interchangeable filter cartridge where particulate matter collects. A hollow
glass oscillating tube is attached to the cartridge. As particulates collect on the
filter, the tube oscillation frequency decreases and can be related to the mass
deposited. Drawn air is heated to a designated temperature and the air flow is
measured and adjusted. The sampling rate is between 0.5 to 5 1/min. A pre-
separator for PM-2.5 or PM-10 can be mounted upstream from the heated inlet.
Dissolved semivolatiles, mechanical noise, and rapid changes in air temperature
may interfere with the readings. The LODs are approximately 5 Ag/m 3 .
Stationary or personal nephelometers can be also used [64]. Nephelometers
measure light scattering and relate it to particulate matter concentration. An
averaging time as low as 1 min is possible. Nephelometers illuminate a fixed
sample volume from one side and observe the changes in the amount of light that
is scattered by particles and gas molecules in the direction of a photomultiplier
tube. A particle filter can be inserted periodically into the sample stream to mea-
sure the light scattered by gas molecules and can be subtracted from the total
scattered signal to determine the contribution from the particles alone. The
instruments are calibrated by filling the sample volume with CO2 gas, which has
a known scattering coefficient. Beta attenuation monitors detect changes in beta
radiation scatter inside a fixed volume cell with moving sampled air. Electrical
counters are based on charging passing aerosols and measuring their capability
to cross a controlled electrical field inside a sample cell.

1.2.4 Radon

Radon is a radioactive and inert gas originating from the decay of uranium 238 and
thorium 232 . Radon originates from soil bedrock and often penetrates to the house
envelope through the basement floor, cracks, and deep well water. Radon exists in
several isotopic forms which, shortly after their formation, combine with small
aerosol particles [65]. These particles can be inhaled in the lungs where daughter
radon molecules decay and release alpha, beta, and gamma rays that increase the
probability of cancer, cardiac/gastrointestinal, and nervous system distress.
Indoor air concentrations are often several orders of magnitude higher than those
of ambient air and typically range from 10 to 140 bequerels per cubic meter

16
(bq/m3) [66]. Concentrations as high as 100,000 bq/m3 have been detected. Pre-
ferred measurements last from several months to a year and use passive dosime-
ters because emission rates for radon can vary significantly, hourly, diurnally, and
seasonally. Grab sampling is only used for initial screening and spot check of con-
trol methods.
Most measuring methods are based on detection of the alpha, beta, and
gamma particles produced by radon and radon decay products and include scin-
tillation cells, alpha track detectors, charcoal detectors, electret detectors, and
electronic monitors [65]. Scintillation cells consist of a small metal cylinder
coated on the inside with ZnS(Ag). After the air sample fills the cylinder either
via diffusion or with a pump, alpha particles strike the ZnS(Ag), the signal is
multiplied, and the decay rate is related to the radon concentration [67].
Alpha track detectors (ATDs) are inexpensive and ideal for long-term sam-
pling. ADTs use special plastic or polymeric materials, e.g. LR-115 cellulose
nitrate film, CR-39 thermoset polymer plastic, or Makrofol polycarbonate plas-
tic, that are damaged specifically by the passage of alpha particles. The damage
trail can be made visible by chemical or electrochemical etching and counted
under microscope or by software in a laboratory [68]. Most ADTs consist of few
cm2 of coating inside an almost airtight container to which air enters via diffu-
sion. Charcoal detectors are suitable for short-term (several days to a week) sur-
veys. Radon is adsorbed to charcoal inside a passive, pocket-size sampler. The
gamma radiation emitted by radon and its decay products are later analyzed in
the laboratory with a gamma ray detector such as sodium iodide (NAI(T1)) or
with a thermoluminescent detector [69].
Electret detectors use a charged material such as aluminized Teflon which
will retain a charge for as long as a year [70,71]. Typically, an electret is mounted
inside a small plastic container to which air diffuses. Charged radon decay prod-
ucts decrease the electret charge over time. The charge reduction can be related
to radon concentration on-site. Electrets can be recharged and therefore reused.
The presence of water vapor and other charged particles may interfere with mea-
surements. Electronic monitors are typically active samplers that detect alpha
radiation using solid-state detectors. This type of detector is capable of grab sam-
pling and on-site analysis [72].

1.2.5 Asbestos

Asbestos is the generic name for a wide variety of hydrated silicate minerals such
as chrysolite or amosite with a simplified chemical formula of MenSi 2mOp(OH)2 r,
where Me might be Mg, Na, Fe, or Ca, n = 2,5, or 7, m = 1 or 4,p = 5 or 22, r = 1
or 2. All types of asbestos are considered to be carcinogens due to their capability
of tissue scarring [73,74]. Asbestos has extremely good insulation, fire retarda-
tion, and chemical resistance. Indoor air quality can be adversely affected when
asbestos is released from building materials.
Measurements of asbestos concentrations are very challenging because of

17
possible interferences from other particulate matter. The range of typical asbes-
tos concentrations can vary from as low as 0.1 ng/m3 to 1 m/m 3 , which are often
several orders of magnitude lower that PM concentrations [3]. The "critical"
asbestos fibers are equal or longer than 5 mm, and have a diameter up to 3 gm.
Typical asbestos concentrations expressed in critical fibers per m3 (F/m 3 ) vary
from approximately 100 F/m3 in buildings without specific asbestos sources up to
10,000 F/m 3 in buildings with friable asbestos [3]. Occupational air concentra-
tions can be as high as 100,000,000 F/m 3 .
The most common approach is to sample with a PM sampler using a
polycarbonate cellulose ester filter with 0.4 /m or 0.8 Aum pore size [75]. A seg-
ment of used filter is then analyzed with an electron microscope where asbestos
is identified, counted, and sized. Often an X-ray diffraction analyzer is attached
to the electron microscope to positively identify asbestos fibers. The results are
extrapolated to a number of critical fibers per air volume based on sample vol-
ume and filter area analyzed. Conversion factors for estimation of asbestos mass
per volume of air can also be used [76]. One of the new developments is the use of
PDMS-7 am SPME fibers to collect airborne asbestos followed by Raman micro-
spectroscopy analysis [77]. SPME coating is used as a "glue" to capture airborne
asbestos, which can be positively identified by Raman microspectroscopy.

1.2.6 Ozone

Ozone (03) is a very strong oxidizing compound of great importance in urban

atmospheres because of its central role in regulating the formation of smog by
interacting with nitrogen oxides and volatile organic compounds. Ozone is
faintly bluish in color, has a distinctive odor connotation, and decays quickly
because of its reactivity. The typical half-life for 03 indoors is approximately 30
min compared with approximately 3 days in ambient air at 20°C. Ozone irritates
the respiratory system by inducing coughing, headaches and shortness of breath
[78]. Longer exposures can result in lung tissue damage. The National Ambient
Air Quality Standards (NAAQS) for 1 h and 8 h averages are 120 ppb (-235
Ag/m3 at room temperature and pressure) and 80 ppb (-157 gg/m3 ), respectively.
Infiltrating outdoor air is the main source of indoor ozone if there are no sources
such as electrostatic air cleaners, copiers, printers, and UV-light.
The passive ozone sampler uses a nitrate ion coated filter sandwiched
between diffusion barriers on both sides. The nitrate ion is oxidized to nitrate in
the presence of ozone. The filter is later extracted with deionized and distilled
water and analyzed for nitrate ion by ion chromatography [14].
Ozone concentrations can be measured via active methods [14]. Ultraviolet
absorption photometry is adopted as the ASTM D 5156 method for measuring 03
from approximately 1 ppb to 1 ppm. In this method, absorption of ultraviolet
radiation by 03 is related to absorption in an ozone-free reference cell [79].
In the electrochemical cell method, air is pumped through a cell where an
ozone-specific reaction generates a signal proportional to the 03 concentration.

18
The workable range varies from approximately 30 ppb to 1 ppm. Presence of
humidity and changes in pressure can interfere with the method sensitivity. Dry
colorimetry uses paper tape impregnated with 0O-specific dry reagent which
changes color upon contact with ozone. The color change can be measured with a
photometer and related to 03 concentration. Working range for this method var-
ies from approximately 30 ppb to 300 ppb. The same approach can be used for
other gaseous pollutants by use of other gas-specific dry reagents. The chemo-
luminescence method is based on reaction of O0 with ethylene to produce specific
wavelength in the 300 to 600 nm range. The lower detection limit (LDL) for this
method is approximately 1 ppb. Solid-state ozone sensors are also applicable for
indoor air applications [80].

1.2.7 Nitrogen dioxide

Indoor nitrogen dioxide (NO2 ) is formed at high temperatures by the combina-

tion of oxygen and nitrogen in air during combustion of natural gas, wood, kero-
sene, coal and other fuels in kitchen stoves, fireplaces, clothes dryers, space
heaters, and water heaters. Other sources include tobacco smoking and infiltra-
tions from outdoors or attached garages. It is a water soluble and oxidizing gas
that attacks mucous membranes causing respiratory responses and symptoms
[81]. In the presence of an indoor source, NO2 concentrations can exceed those
found in ambient urban atmospheres, particularly when combustion sources are
unvented. For example, maximum 1 h concentrations in kitchens with natural
gas stoves range from 470 to 1440 Ag/m 3 during cooking [82]. The NAAQS 24 h
average for NO 2 is 100 j/g/m3 (approx. 53 ppb at room temperature).
Nitrogen dioxide can be measured using passive samplers configured as long
acrylic tubes with a cross-sectional area to length ratio of approximately 0.1 cm.
A stainless steel mesh screen coated with triethylamine which acts as an adsor-
bent is placed at the sealed end of the tube (U.S. EPA Method IP-5A) [5]. Used
tubes can be capped and transferred to a laboratory. Formed nitrate can be
extracted within 20 to 30 min using color mixture of sulfanilamide reagent and
N-1-naphthylene-diamine-dichydrochloride (NEDA) reagent and then analyzed
with a spectrophotometer at 540 nm. Concentration of NO2 can be estimated
based on sampling time, amount of nitrate formed, adjusted diffusion coefficient
for NO2 in air, and tube geometry. These tubes, a.k.a. Palmes tubes (Fig. 1.8),
can be reliably used for long sampling times from several hours up to two weeks
[83]. The practical sensitivity of this method is approximately 100 ppb (or 300
ppb-h). Possible interferences include humidity, temperature, wind velocity at
the inlet of the tube, presence of particulate matter and peroxyacetyl nitrate and
some nitroso compounds. A passive sampling device in the form of a badge con-
sisting of triethylamine-coated glass fiber filters behind sets of diffusion barriers
on each side can also be used (EPA Method IP-5C) [5, 84]. After sample collection
the badge is capped and transferred to the laboratory where it is disassembled,
extracted with 10 ml of distilled and deionized water and analyzed by ion chro-

19
 I
A\ ~
o/ Rml~~hre
Air sample /
Up
inlet
/Acrylic tube

ting
on
3 stainless steel
screens coated with TEA

W162
-Fixed cap

Left: Fig. 1.8. Schematic of passive Palmes sampler for NO2 .

Right: Fig. 1.9. Schematic of gas bubbler sampler.

matography. The minimum detectable quantity is 30 ppb h for a typical 8 to 24 h

sampling time.
Continuous samplers for direct measurement of NO 2 concentrations are
based on purging through a solution, absorption and colorimetric analysis, fluo-
rescence, chemoluminescent reaction with ozone, electrochemical sensor, diffu-
sion, and electrochemistry [85]. The modified ASTM Method D 1607 is based on
absorption of NO 2 in a N-1-naphthylene-diamine-dichydrochloride solution
placed in a fritted bubbler (Fig. 1.9) [86]. In this method, the air sample is
pumped at 0.4 1/min through a 60 /um pore size frit submerged in the absorbing
solution. The air flow rate and the pore size affect the absorption efficiency. Typ-
ical sampling time ranges from 10 to 30 min during which the solution changes
color upon contact with NO 2 (Griess-Saltzman reaction) which can be analyzed
on a spectrophotometer or colorimeter at 550 nm. Typical application range var-
ies from approximately 5 ppb to 5 ppm. Lower accuracy is associated with the
alternative sampling that involves drawing air into a 100 ml glass syringe par-
tially filled with the absorbent solution, shaking it for color development, and
spectrophotometric or colorimetric analysis. Another variation of this method is
the use of evacuated sampling bottles that can be used to draw air through a
solution containing the absorbing reagent. Sulfanilamide and 8-anilino-
1-naphthalenesulfonic acid (ANSA) absorbent solution can also be used for 24 h
sampling [86]. The formed nitrate is analyzed at 550 nm. The working range for
this solution is approximately 10 ppb to 350 ppb for 24 h sampling at 200 ml/min.
U.S. EPA Method IP-5A uses a real-time, direct measurement monitor to
detect NO 2 based on fluorescence [5]. This involves detection of fluorescent energy
emitted from the reaction of NO2 with Luminol (5-amino-2,3-dihydro-1,4 phthala-
zine dione) solution. Sampled air is pumped over a fabric wick continuously wet-
ted with Luminol. When NO2 is present, Luminol oxidizes producing fluorescent
energy that is collected by a photomultiplier tube. Nitrogen oxide concentration is
proportional to tube signal. The LDL for this method is very low, i.e., 0.005 ppb.
Some interferences are caused by the presence of NO and humidity [87].

20
 Continuous monitoring for NO2 concentrations as low as 1 ppb to 10 ppb can
be achieved with chemoluminescence [86]. This method is based on a two-stage
process initiated by reaction of NO with internally generated ozone to form NO2 .
At this stage, the NO concentration can be estimated based on the characteristic
luminescence. Next, nitrogen dioxide is converted at 325°C to NOx and the signal
represents the NOx (as NO + NO2 ) concentration. The NO2 concentration is cal-
culated as a difference between the NOx and NO concentration. Some monitors
are capable of measuring ammonia concentrations when the second converter is
used, and the ammonia concentrations are calculated as a difference between
the total nitrogen and the NOx.

1.2.8 Sulfur dioxide

Sulfur dioxide (SO 2) is a colorless gas with a strong pungent odor detectable at
approximately 0.5 ppm. It is classified as a primary air pollutant that is generated
during combustion of sulfur-containing fuels, e.g., natural gas, coal, wood, and
kerosene. SO2 can form acid aerosols when SO2 reacts with water vapor, and sec-
ondary fine particulate matter when SO2 and SO3 react with ammonia. Sulfur
dioxide is highly soluble in water, thus it irritates mucous membranes and causes
respiratory problems. Indoor SO2 concentrations are typically lower than out-
doors and range from approximately 2 Atm/m3 for houses without kerosene heat-
ers to 150 Am/m 3 for households with kerosene heaters and/or gas stoves [88].
Concentrations as high as 2660 tkm/m 3 (approximately 1 ppm) were measured in
houses with unvented kerosene heaters [89].
There are several approaches for SO2 sampling in ambient air that can be
modified for indoor air sampling including U.S. EPA Method 6 for "stationary
sources" by titration [90], U.S. EPA Method 15 for sulfur compounds by GC/ FPD
[91], direct reading methods using pararosaniline reaction and electrochemical
reaction [92]. Method 6 uses a heated probe to collect the sample and filter out
particulates and sulfuric acid mist. Ammonia and water-soluble cations may inter-
fere. Air samples are collected in impingers filled with isopropanol and hydrogen
peroxide. The concentration of SO2 is determined by titration. There are other
variations of this method involving different absorbing solutions, e.g., tetrachloro-
mercurate and pararosaniline, formaldehyde and pararosaniline, and different
analyzers, e.g., spectrophotometer. Typical LDLs are approximately 10 ppb.
Continuous monitoring of SO2 can also be achieved using amperometric
methods [90]. In this method SO2 entering the detector cell (via diffusion or
pump) reacts with free iodine or bromine which serve as a titrant. Any change in
titrant concentration produces a potential on anode/cathode electrodes that is
restored by a current generated by a pair of sensor/reference electrodes. This
current is proportional to the SO concentration. The LDL of this method is
approximately 10 ppb. A flame photometric detector (FPD) can be also used as a
stand-alone detector for continuously flowing air or as a GC detector [85,93]. In
this case, the LDLs are close to 1 ppb.

21
1.2.9 Carbon monoxide and carbon dioxide

Carbon monoxide (CO) is a colorless, odorless, and tasteless gas generated pri-
marily by incomplete combustion of carbon-containing materials. It is one of the
most common air pollutants. Carbon monoxide is typically generated in mal-
functioning or inadequately vented cooking/ heating appliances, fire and tobacco
smoke. In same cases, penetration of vehicular emissions from underground or
attached garages can also be a source of indoor air CO. When inhaled, CO can
quickly combine with blood hemoglobin and drastically reduce the oxygen-
carrying capacity. Health effects vary from minor aggravation to death, depend-
ing on the concentration, exposure time, and the level of physical activity. Typi-
cal concentrations recorded in residential air varied from less than 1 ppm to
nearly 30 ppm, with the average of 2.8 ppm [3]. These concentrations are
approximately one order of magnitude lower than ambient air CO concentra-
tions found in extremely polluted large cities, where automotive emissions are
the major CO source [94]. The NAAQS for ambient air are set to 9 ppm for an 8 h
period, and 35 ppm for 1 h period.
Carbon dioxide (CO 2 ) is a colorless and odorless gas. It is a principal combus-
tion product. The main sources indoors are natural gas, coal, wood-burning
appliances, and exhaled air. Concentrations of CO2 are often used to evaluate
ventilation and air turnover rates indoors. The ASHRAE Standard 62-1981 sug-
gests that enough outdoor air is supplied per occupant so that resulting CO2 lev-
els are not greater than 2,500 ppm.
There are several methods that are applicable to CO and CO 2 sampling and
analysis. Most are based on CO/CO 2 detection via a non-dispersive infrared ana-
lyzer (NDIR) and electrochemical or catalytic oxidation. The NDIR uses infrared
radiation that passes through two parallel cells, the first with an air sample
(enclosed or continuously flowing through), the second filled with CO/CO 2 -free
reference air. Concentrations of CO/CO2 are related to the difference in
absorbance of one or more characteristic CO/CO 2 wavelength IR spectrums
using Beer-Lambert Law. The LDLs of this method are approximately 1 ppm.
Another variation of the IR-based approach is the use of gas filter correlation
(GFC) where an optical filter limits the photometers' sensitivity to the IR wave-
lengths of interest. In the GFC, the IR beam passes through a gas filter alternat-
ing between CO/CO 2 and N2 . The CO/CO 2 gas filter produces a beam that cannot
be attenuated by CO/CO2 in the sample. The N2 is transparent to the IR radia-
tion and can be absorbed by CO/CO 2 . The chopped detector signal is modulated
by the alteration between CO and N2 and is proportional to the CO/CO2 concen-
tration [5]. The MDLs of this method are approximately 0.020 ppm.
A more sensitive and specific detector for CO/CO2 is a photoacoustic-IR [3].
The air sample is irradiated with light of a wavelength coinciding with an
absorption maxima of CO/CO 2 and the gas temperature increases due to absorp-
tion. The gas pressure will increase if the volume is constant and vary if the light
is modulated by a chopper. Pressure variations produce a sound wave which can

22
be detected by a condenser microphone and related to the CO/CO2 concentra-
tions. Electrochemical or catalytic oxidation analyzers are often small in size and
used as personal exposure monitors that can be attached to a person [85]. The
monitor samples either via diffusion or convection aided by a small diaphragm
pump. The CO is converted to CO 2 in a liquid cell producing a small electrical
current that can be related to CO concentration. The typical MDL is approxi-
mately 1 ppm. Carbon dioxide and CO can also be sampled with colorimetric
direct reading tubes where typical MDLs are approximately 5 ppm and 100 ppm
for CO and CO2, respectively. Carbon dioxide can also be sampled with a canister
and analyzed on a GC equipped with a helium ionization detector and/or thermal
conductivity detector, resulting in MDLs of approximately 1 ppm and 100 ppm,
respectively.

1.2.10 Environmental tobacco smoke

Environmental tobacco smoke (ETS) is a mixture of exhaled cigarette smoke,

smoke from the burning tip of a cigarette, and smoke that diffuses through the
cigarette paper. The inhalation of ETS is often referred to as an involuntary
exposure to tobacco smoke, passive or secondary smoking. Tobacco combustion
causes the release of thousands of chemicals, of which many are known or sus-
pected carcinogens or toxic [95-97]. Close to the source the sidestream smoke
has higher concentrations of some toxic and carcinogenic analytes that the
mainstream smoke. However, the room air can dilute these concentrations.
There is growing evidence that ETS can cause a wide range of respiratory and
non-respiratory responses, cancers, heart disease, alterations of menopause age,
birth weight, and sudden infant death syndrome [95,97-100]. Tobacco smoking
increases levels of particulate matter, CO, NO2 , polycyclic aromatic compounds
(PAHs), acrolein, benzene, and other VOCs that can be used as non-specific
markers for ETS. On average, each one-pack-a-day smoker adds approximately
20 Ag/m of respirable particulate matter over a 24 h period [101]. Nicotine,
solanesol, and ultraviolet light absorption of particulate matter can be a more
specific ETS marker [96]. Cotonine, which is formed by oxidation of nicotine can
be used as a biomarker of exposure and measured in body fluids. Nicotine con-
centrations measured over 24 h period ranged from 0.3 gg/m 3 for homes without
smoking to approximately 3.7,g/m 3 for homes with smoking [102]. However,
nicotine concentrations as high as 28.6 gg/m3 were recorded [103]. More than
90% of nicotine found in indoors is in the vapor phase.
Nicotine can be measured using active and passive methods. U.S. EPA
Method IP-2A is a modification of the NIOSH Method S293 anduses XAD-4
sorbent tubes [5]. The schematic of the sorbent tube-based sampling shown in
Fig. 1.10 is also typical for a wide range of VOCs. In this method, air samples are
collected using a small personal sampling pump. Method detection limits are
approximately 0.17 gg/m3 for 1 h sample and 0.02 g/m3 for 8 h sample using 1
l/min sampling rate. Nicotine is extracted using ethyl acetate solution of

23
 C r inlet

;k spring

Glass wool
,ent section

orbent section

, Personal pump

,1 Removable sealing caps

Fig. 1.10. Schematic of sorbent tube for sampling of nicotine and other VOCs.

triethylamine and analyzed on a GC equipped with nitrogen specific detector.

U.S. EPA Method IP-2B uses filter cartridges treated with sodium bisulfate [5].
Filter cassette consists of a particulate filter and a treated filter. Method detec-
tion limits are 0.1 ,ug/m3 for 8 h period and sampling rate of 1.7 /min and 0.03
/sg/m 3 for sampling rate of 3 1/min. Particulate filters are extracted with dichloro-
methane and the nicotine is concentrated in to ammoniated heptane. Treated fil-
ters are extracted with 5% ethanol solution. After sodium hydroxide is added,
the solution is concentrated into ammoniated heptane. The analysis is com-
pleted on a GC equipped with a nitrogen selective detector. A modified filtered
cassette can be used for passive sampling [5]. Method detection limits are 16
F/g/m 3 for 4 to 5 h sampling and 0.2 g/m 3 for 1 week sampling.

1.2.11 Biological agents

Bioaerosols include a wide range of particles and aerosols that affect indoor air.
These include endotoxins, allergens derived from insects, domestic animals, and
natural rubber, pollen, fungi, bacteria and viruses [104,105]. Inhalation of air-
borne endotoxins have been linked to occupational diseases suspected in the
development of sick building syndrome and triggering asthma responses [105].
Endotoxins are composed of lipopolysaccharides (LPS) that are shed (and subse-
quently can become airborne) from the outer cell membrane of gram-negative
bacteria. The potency of endotoxins is measured by comparing the biological activ-
ity of a sample with that of a reference such as Limulus amebocyte lysate (LAL)
assay [105]. The U.S. reference standard is purified from E. coli 0113:H10:K. By
convention, 1 endotoxin unit (EU) is equivalent to 0.1 ng of the reference standard
endotoxin and it does not represent a measure of concentration of a single chemi-
cal [105]. Averaged endotoxin concentrations measured varied from a few tens to
a few hundreds EU/m3 in agricultural and related workplaces. Concentrations

24
exceeding 1000 EU/m 3 were found in machine shops where metal lubricating fluid
was infected and aerosolized [106]. Typical sampling methods use filter media or
glass impingers to collect airborne endotoxins using the LAL assay. Filter selec-
tion depends on the type of aerosol and source sampled. Teflon, cellulose mixed
ester, and PVC can alter, inhibit, or deactivate endotoxin activity [107]. To date,
only polycarbonate capillary pore membrane filters are known to be acceptable for
a wide range of endotoxins [108]. Glass impingers are not sensitive samplers for
submicron particulates, which consist the majority of airborne endotoxins.
Extraction of filters involves sonication and rocking in pyrogen-free water. Sam-
ple analysis usually involves the use of an assay containing an enzyme that is spe-
cifically activated by endotoxin, which in turn, activates other enzymes and
enhances response [105]. The variability of assays can be reduced by the use of
GC-MS, which is based on the analysis of specific markers that are characteristic
of LPS, e.g., 3-hydroxy fatty acids which are derivatized from samples [109]. The
main disadvantage is the lower detection limits compared to the assay method.
Airborne allergens contain antigen proteins that induce immune responses
in humans. The major sources of residential allergens include cats and dogs,
mites, and German cockroaches [110]. Cat and dog allergens are usually
attached to small particles, which cannot be settled easily and can attach them-
selves to walls and clothing, while mites proteins tend to attach themselves to
larger particles of 10 to 20 Am which are less mobile. Electron microscopic analy-
sis identified these particles as mite fecal pellets [111]. The protein identified as
a mite allergen (Der p 1) has MW equal to 24,000 daltons and could be radio-
labeled with 125I and measured using the inhibition radioimmunoassay (RIA)
[112]. This technique allows the measurement of mite allergens in house dust
from approximately 0.4 to 500 Ag of Der p 1 per gram of dust. Special nasal
masks can be used to trap, monitor, and identify particles entering the nose
[113]. Typical sampling methods for airborne allergens use cascade samplers or
liquid impingers, followed by the radioallergosorbent (RAST) or radioimmuno-
assay (RIA) analysis. It is currently possible to measure a single representative
allergen for cats (Mancini technique), and mites, cats, dogs, and cockroaches
using two site monoclonal antibody-based assays [110].
There is growing evidence that immunologically foreign proteins can also be
generated by insects and pests such as cockroaches [114]. These include the Bla g
1 and Bla g 2 associated with German cockroaches which are likely excreted in
their feces. Particles carrying these proteins are likely greater than 5 Am and can
become airborne only when disturbed, e.g., during vacuuming, and likely cannot
get attached to clothing. In contrast, cat allergens are typically attached to
smaller particles [115]. Typical cat allergen concentrations for houses with a cat
range from 2 to as high as 80 Am Fel d 1 per gram of dust. Both cat and dog aller-
gens are sticky, easily transferable to houses without them, and have a potential
to accumulate [116].
Pollen is a male reproductive part of flowering plants that can cause hay
fever, allergenic asthma, and other allergenic symptoms. Pollen can enter indoor

25
spaces and air via vents, cracks, open windows and doors, clothing, and pets. Pol-
len can cause discomfort in persons whose immune system overreacts to pollen
grains that come in contact with mucous membranes in areas such as eyes or the
respiratory system [117]. The typical ratio of outdoor-to-indoor pollen concen-
trations depends on the time of year, time of day, and ventilation scheme, and
ranges from approximately 1 to more than 20 [118,119]. Average measured con-
centrations in homes were 1.2 grains per m3 (gr/m3 ) for airborne pollen and
5.8x 105 gr/m3 for dustborne pollen [120]. The simplest way to sample airborne
pollen consists of a microscope slide coated with a sticky film exposed to settling
pollen for sampling times ranging from hours to days, e.g. Durham sampler. Set-
tling samplers tend to be biased toward particles with larger settling velocity and
the difficulty to relate the number of grains per surface area to the actual air-
borne concentrations. Another variation of the settling sampler is a Taber trap
consisting of a 10 x 10 cm cylinder covered with a specially shaped cover to allow
pollen to settle inside to a glycerol solution. This solution can be periodically
removed and analyzed for pollen settling rate.
Rotating arm impactors use narrow bars wrapped with a clear plastic or
adhesive-coated tape that are rotated around the axis collecting particles [117].
These particles can be stained, e.g. Calberla stain, and counted under a light
microscope. Another popular sampler is the Rotorod sampler capable of collect-
ing particles down to approximately 15 ,Am. This sampler is typically deployed
for periods from 15 to 60 s at 10 min intervals for a 24 h period. Hirst-type sam-
plers use a narrow slit through which air is pulled and impacts a moving
impaction surface coated with an adhesive. Hirst-type samples allow for time
resolution down to 2 h and particle size cut-off at 5 Aum which allows for collection
of the majority of airborne pollen. Filtration samplers use high volume pumps
and a variety of filters to effectively collect airborne pollen. Passive filtration
devices, typically used in outdoor air, e.g. Cour traps, use two 5-layered gauze fil-
ters with an exposed area of approximately 400 cm 2 that can be exposed for days
up to several weeks. Pollen grains can be counted and identified using light
microscopy at magnifications from 400 x to 1000 x. Sample preparation usually
involves dissolving organic matter in a procedure called acetolysis, which does
not affect the resistant pollen outer layer [121]. In the last decade, the use of
labeled antibodies and detection of the actual allergens with fluorescence was
developed. This method is faster than conventional methods and amenable to
automation of counting [122].
Fungi (mold, mildew) are ubiquitous in indoor air environments. Fungi
release water, C0 2 , ethanol, organic acids, enzymes, VOCs, and non-volatile tox-
ins (mycotoxins) [123J. Some of the fungi produce proteases which are likely to
be allergens. VOCs such as aldehydes, ketones, and mycotoxins including afla-
toxins, satratoxins, verrucarins, roridins, griseolvium, and fumonisis are of
interest to indoor air quality. Indoor air sampling often involves the use of an
impactor sampler or liquid impinger. In the impactor sampler, moving air passes
through jets (small nozzles) that impact air against a Petri dish with a nutrient-

26
 Air inlet-*

Separating
nozzles

Fig. 1.11. Schematic of two-stage biological aerosols sampler for collection of non-respirable and
respirable fraction.

coated surface (Fig. 1.11). Staged samplers, where each stage is designed to cap-
ture a predetermined size or range of sizes, can be used to determine the size dis-
tribution of airborne aerosols [124]. Cultures appearing after 3-7 day incubation
can be visually counted. Results are reported as colony forming units per m3 of
air (CFU/m 3 ). Liquid impingers, such as the all-glass impinger (AGI), capture
and retain airborne microorganisms, which can be cultured and counted on a
specific (or non-specific) agar. Mycotoxins are often analyzed with a combination
of analytical methodologies including thin layer chromatography (TLC), HPLC,
SPE, GC-MS, LC-MS, and NMR which are summarized in the handbook of the
Association of Official Analytical Chemists [125]. Airborne bacteria can be
labeled and analyzed using direct count methods based on staining and fluores-
cence [126]. Another method, the polymerase chain reaction (PRC), is based on
detection of selected regions of DNA specific to a particular kind of bacteria of
interest, amplified by 10 4 to 106 times for identification [127]. Mobile and near
real time systems for field sampling of aerosol protein and detection via
absorbance promise automation and on-site analysis [128].

REFERENCES

1 N.L. Nagda, H.E. Rector and M.D. Koontz, Guidelines for Monitoring Indoor Air
Quality. Hemisphere Publishing, New York, NY, 1987.
2 J. Namiesnik and T. Gorecki, LC-GC Europe, 9 (2001) 679.
3 M. Maroni, B. Seifert and T. Lindvall (Eds.), Indoor Air Quality: A Comprehensive
Reference Book. Elsevier, Amsterdam, The Netherlands, 1995.
4 J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.), Indoor Air Quality Handbook.
McGraw-Hill, New York, NY, 2000.
5 W.T. Winberry, L. Forehand, N.T. Murphy, A. Ceroli, B. Phinney and A. Evans
(Eds.), Methods for Determination of Indoor Air Pollutants: EPA Methods. Noyes
Data Corporation, Park Ridge, NJ, 1993.
6 T. Salthammer (Ed.), Organic Indoor Air Pollutants: Occurrence, Measurement,
Evaluation.Wiley-VCH, Weinheim, Germany, 1999.

27
 7 M.J. Boss and D.W. Day, Air Sampling and IndustrialHygiene Engineering.Lewis
Publishers, Boca Raton, FL, 2001.
8 E.W. Miller and R.M. Miller, Indoor Pollution:A Reference Handbook. ABC-CLIO,
Santa Barbara, CA, 1998.
9 L.H. Keith and M.M. Walker, Handbook of Air Toxics: Sampling, Analysis, and
Properties.CRC Press, Boca Raton, FL, 1995.
10 B.S. Cohen and S.V. Hering (Eds.), Air Sampling Instruments for Evaluation of
Atmospheric Contaminants. ACGIH, Cincinnati, OH, 1995.
11 R. Wilson and J. Spengler (Eds.), Particlesin Our Air: Concentrations and Health
Effects. Harvard University Press, Cambridge, MA, 1996.
12 R.H. Brown and L.E. Monteith, in: B.S. Cohen and S.V. Hering (Eds.), Air Sampling
Instruments for Evaluation of Atmospheric Contaminants. ACGIH, Cincinnati, OH,
1995, Ch. 17, p. 369.
13 B.E. Saltzman and P.E. Caplan, in: B.S. Cohen and S.V. Hering (Eds.), Air Sampling
Instrumentsfor Evaluation of Atmospheric Contaminants.ACGIH, Cincinnati, OH,
1995, Ch. 18, p. 401.
14 M.L. Woebkenberg and C.S. McCammon, in: B.S. Cohen and S.V. Hering (Eds.), Air
Sampling Instruments for Evaluation of Atmospheric Contaminants. ACGIH,
Cincinnati, OH, 1995, Ch. 19, p. 439.
15 U.S. Environmental Protection Agency, Report to Congress on Indoor Air Quality,
Vol. II, Assessment and Control of Indoor Air Pollution, Report no EPA
400/1-89-001C, Washington, DC, 1989.
16 L. Molhave, IndoorAir, 4 (1996) 357.
17 National Institute for Occupational Safety and Health, Manual of Analytical
Methods, 4th edn. Centers for Disease Control and Prevention, Cincinnati, OH, 1994.
18 W.G. Tucker, in: J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.), Indoor Air
QualityHandbook. McGraw-Hill, New York, NY, 2000, Ch. 31, p. 31.1.
19 P. Wolfoff, Volatile organic compounds-sources, measurements, emissions and the
impact on indoor air quality. IndoorAir suppl., 3/95, (1995) 1.
20 American Society for Testing and Materials (ASTM), D5466-95: Standard Text
Method for Determination of Volatile Organic Compounds in Atmospheres (Canister
Sampling Methodology), West Conshohocken, PA, American Society for Testing and
Materials, 1995.
21 American Society for Testing and Materials (ASTM), D6196-97: Standard Practice
for the Sampling and Analysis of VOCs in Air by Pumped Sorbent Tube and Thermal
Desorption, West Conshohocken, PA, American Society for Testing and Materials,
1997.
22 M. Harper, J. Chromatogr.A, 885 (2000) 129.
23 A.T. Hodgson, IndoorAir, 5 (1995) 247.
24 E. Ilgen, N. Karfich, K. Levsen, J. Angerer, P. Schneider, J. Heinrich, H-E.
Wichmann, L. Dunemann and J. Begerow, Atmos. Environ., 35 (2001) 1235.
25 K. Elke, E. Jermann, J. Begerow and L. Dunemann, J. Chromatogr. A, 826 (1998)
191.
26 D. Gorlo, B. Zygmunt, M. Dudek, A. Jaszek, M. Pilarczyk and J. Namiesnik,
FreseniusJAnal. Chem., 363 (1999) 696.
27 J.A. Koziel and J. Pawliszyn, J. Air Waste Manage. Assoc., 51 (2001) 173.
28 F. Augusto, J.A. Koziel and J. Pawliszyn, Anal. Chem., 73 (2001) 481.
29 P.A. Martos and J. Pawliszyn, Anal. Chem., 71 (1999) 1513.
30 J. Koziel, J. Noah and J. Pawliszyn, Environ. Sci. Technol., 35 (2001) 1481.
31 M. Jia, J.A. Koziel and J. Pawliszyn, Field Anal. Chem. Technol., 4 (2000) 73.
32 J. Pawliszyn, Solid Phase Microextraction: Theory and Practice. Wiley-VCH, New
York, NY, 1997.

28
33 V.M. Brown, D.R. Crump and D. Gardiner, Environ. Technol., 14 (1992) 367.
34 R.H. Brown and H.J. Saunders (Eds.), Diffusive Sampling. Royal Society for
Chemistry, London, 1987.
35 D. Crump, in T. Salthammer (Ed.), Organic Indoor Air Pollutants: Occurrence,
Measurement, Evaluation.Wiley-VCH, Weinheim, Germany, 1999 Ch. 1.5, p. 57.
36 ECA, Sampling Strategies for volatile organic compounds (VOCs). European
collaborative action-Indoor Air Quality and Its Impact on Man, Report No 14,
European Commission, EUR 16051 EN, Luxemburg, 1994.
37 M.A. Cohen, P.B. Ryan, Y. Yanagisawa and S. Hammond, J. Air Waste Manage.
Assoc., 40 (1990) 993.
38 R. Otson, P. Fellin and Q. Tran, Atmos. Environ., 28 (1994) 3563.
39 K. Levsen, E. Schimming, J. Angerer, H.E. Wickman and J. Heinrich, Proceedingsof
IndoorAir 96, July 1996, Nagoya, Japan, 1996, Vol. 1, p. 1061.
40 C. Krause, R. Nael, C. Schultz, B. Seifert and D. Ulrich, Proceedingsof IndoorAir 87,
August 1987, Berlin, Germany, 1987, Vol. 1, p. 102.
41 D. Cavallo, D. Alcini, P. Carrer, A. Basso, D. Bollini, L. Lovato, F. Vercelli, F. Visigalli
and M. Marconi, Proceedings of the Healthy Buildings/IAQ 97, September/October
1997, Washington, DC, 1997, Vol. 3, p. 141.
42 V.M. Brown, A.H. Cockram, D.R. Crump and H.S. Mann, Proceedings of IndoorAir
96, July 1996, Nagoya, Japan, 1996, vol. 2, p. 115.
43 A. Rosencwaig, in: P.J. Elving, J.D. Winefordner, I.M. Kolthoff (Eds.), Chemical
Analysis. Vol. 57, Chichester, UK, 1980.
44 L.E. Ekberg, in T. Salthammer (Ed.) Organic Indoor Air Pollutants: Occurrence,
Measurement, Evaluation. Wiley-VCH, Weinheim, Germany, 1999, Ch. 1.6, p. 73.
45 J.B. Hicks, E.D. Winegar and C.S. Sims, Proceedingsof IAQ '92: Environments and
People, San Francisco, 1992. American Society of Heating Refrigerating and
Air-Conditioning Engineers, Inc. Atlanta, GA, 1992, p. 251.
46 K.A. Bunding Lee, A.L. Clobes, A.L. Hood, J.A., Schroeder and L. Hawkins, Proc-
eedings of Indoor Air '93, 1993, Helsinki, Finland, 1993, Vol. 2, p. 245.
47 T. Godish, T.W. Zollinger and V. Konopinski, J. Environ. Health., 53 (1990) 34.
48 K-S. Liu, F-Y. Huang and S.B. Hayes, Proceedings of the 4th Int. Conf. Indoor Air
Quality and Climate, 1987, Berlin, Germany, 1987, Vol. 2, p. 610.
49 A. Vairamurthy, J.M. Roberts and L. Newman, Atmos. Environ., 26A (1992) 1965.
50 M. Schultz and T. Salthammer, in: T. Salthammer (Ed.), OrganicIndoorAir Pollut-
ants: Occurrence, Measurement,Evaluation.Wiley-VCH, Weinheim, Germany, 1999,
Ch. 1.2, p. 15.
51 T. Salthammer, JPhotochem. Photobiol. A., 74 (1993) 195.
52 P. Martos and J. Pawliszyn, Anal. Chem., 70 (1998) 2311.
53 E. Uhde and A. Borgschulte, Proceedings of the 8th International Conference on
Indoor Air and Climate. Edinburgh, Scotland, 1999.
54 R.G. Lewis, in J.S. Spengler, J.M. Samet, J.F. McCarthy (Eds.), Indoor Air Quality
Handbook. McGraw-Hill, New York, NY, 2000, Ch. 35, p. 35.1.
55 R.G. Lewis and S.M. Gordon, in: L.H. Keith (Ed.), Principles of Environmental
Sampling. ACS Professional Reference Book. American Chemical Society,
Washington, DC, 2nd ed. 1996, p. 401.
56 American Society for Testing and Materials (ASTM), D4861: Standard practice for
sampling and selection of analytical techniques for pesticides and polychlorinated
biphenyls in air. In: Annual Book of ASTM Standards. West Conshohocken, PA,
American Society for Testing and Materials, Vol. 11.03, 2000.
57 American Society for Testing and Materials (ASTM), D4947: Standard test method
for chlordane and pentachlor residues in indoor air. In: Annual Book of ASTM
Standards. West Conshohocken, PA, American Society for Testing and Materials,

29
 Vol.11.03, 2000.
58 R. Rudel, in: J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.), IndoorAir Quality
Handbook. McGraw-Hill, New York, NY, 2000, Ch. 34, p. 34.1.
59 R.A. Rudel, P.W. Geno, G. Sun, A. Yau, J.D. Spengler, J. Vallarino and J.G. Brody, J.
Air Waste Manage. Assoc., 51 (2001) 499.
60 K. Ogan and E. Katz, Anal. Chem., 53 (1981) 160.
61 L. Wallace, J. Air Waste Manage. Assoc., 46 (1996) 98.
62 P. Koutrakis, J.M. Wolfson, J.L. Slater, M. Brauer and J.D. Spengler, Environ. Sci.
Technol., 22 (1988) 1463.
63 S.K. Poruthoor, P.K. Dasgupta and Z. Genfa, Environ. Sci. Technol., 29 (1995) 1534.
64 C. Howard-Reed, A.W. Rea, M.J. Zufall, J.M. Burke, R.W. Williams, J.C. Suggs, L.S.
Sheldon, D. Walsh and R. Kwok, J. Air Waste Manage. Assoc., 50 (2000) 1125.
65 W.W. Nazaroff and A.V. Nero (Eds.), Measurement Techniques, Radon and Its
Productsin IndoorAir. Wiley Interscience, New York, NY, 1988.
66 A.V. Nero, M.B. Schwehr, W.W. Nazaroff and K.L. Rezan, Science, 234 (1986) 992.
67 H.F. Lukas, Rev. Sci. Instrum., 28 (1957) 680.
68 B.G. Cartwright, E.K. Shirk and P.B. Price, Nuclear Inst. Meth., 153 (1978) 45.
69 E. Stranden, A.K. Kolstad and B. Lind, Radiat. Prot.Dosim., 5 (1983) 241.
70 P. Chittaporn, M. Eisenbud and N.H. Harley, Health Physics, 41 (1981) 405.
71 P. Kotrappa, J. Bigu, Proceedings of 38th Annual Meeting of the Health Physics
Society, Atlanta, GA, July 1993.
72 W.E. Simon and M.B. Schell, Proceedings of the 1990 International Symposium on
Radon and Radon Reduction Technology, U.S. EPA, EPA/600/ 9-90, 1990.
73 P.J. Landrigan, New Engl. J. Med., 338 (1998) 1618.
74 W.N. Rom, in: Rom (Ed.), Environmental and Occupational Medicine. Lippincott-
Raven Publishers, Philadelphia, PA, 1998, p. 349.
75 A.V. Samudra, C.F. Harwood and J.D. Stockham, Electron microscope measurement
of airborne asbestos concentrations. Research Triangle Park, NC. US EPA, EPA
Report No. 60012-77-178-Rev., PB 285945, 1978.
76 B.T. Commings, The Significance of Asbestos and Other Mineral Fibers in Environ-
mental Ambient Air. Commings Assoc. Maidenhead, UK, 1985.
77 M. Odziemkowski, J.A. Koziel, D. Irish and J. Pawliszyn, Anal. Chem., 73 (2001)
3131.
78 U.S. Environmental Protection Agency (EPA), Air quality criteria for ozone and
other photochemical oxidants, Report No EPA-600/8-84-020F, Washington, DC,
1986.
79 American Society for Testing and Materials (ASTM), Atmospheric Analysis; Occupa-
tional Health and Safety; Protective Clothing. Annual Book of ASTM Standards.
West Conshohocken, PA, 1999, Vol. 11.03.
80 P. McGheehin, P.T. Mosely and D.E. Williams, Sensor Review, 1 (1994) 13.
81 World Health Organization (WHO), Air Quality Guidelines for Europe. WHO
Regional Publications, European Series No. 23, WHO Regional Office for Europe,
Copenhagen Denmark, 1987.
82 W.A. Wade, W.A. Cote and J.E. Yocom, J. Air Pollut. Control Assoc., 25 (1975) 933.
83 E.D. Palmes, A.F. Gunnison, J. Di Mattio and C. Tomczyk, Am. Ind. Hyg. Assoc. J.,
37 (1976) 570.
84 Y. Yagisawa and H. Nashimura, Environ. Int., 8 (1982) 235.
85 J.E, Yocom and S.M. McCarthy, Measuring Indoor Air Quality: A Practical Guide.
Wiley-VCH, Chichester, UK, 1991.
86 J.P. Lodge, Methods of Air Sampling and Analysis. Lewis Publishers, Boca Raton,
FL, 1988.
87 E.E Rickman, A.H. Groen, R.S. Wright and J.E. Sickles, Laboratory and Field

30
 Evaluation of Extrasensitive SO2 and NO2 Analyzers for Acid Deposition Monitoring,
U.S. EPA, Quality Assurance Division, Research Triangle Park, NC, RTI/3999/
18-04F, Draft, May 1989.
88 P.B. Leaderer, J.A.J. Stolwijk, R.T. Zagraniski and M. Quin-Shan, Proceedingsof the
77th Ann. Meeting Air Poll. Control Assoc., June 1984, San Francisco, CA, 1984,
Paper No 84-33.3.
89 K.P. Cooper and R.R. Alberti, Am. Rev. Resp. Dis., 129 (1984) 629.
90 U.S. Environmental Protection Agency, Code of Federal Regulations, Title 40, Part
60, Appendix A, Washington, DC, Office of Federal Register, July 1, 1987.
91 U.S. Environmental Protection Agency, Standards of Performance of New Station-
ary Sources. Compilation, EPA-340/1-77-015, 1987.
92 S. Buckestein, K.A. Tucker and P.R. Gifford, Anal. Chem., 52 (1980) 2396.
93 R.K. Stevens, J.D. Mulik, A.E. O'Keefe and K.J. Frost, Anal. Chem., 43 (1971) 827.
94 J.H. Seinfeld, Atmospheric Chemistry and Physics of Air Pollution. John Wiley &
Sons, New York, NY, 1986.
95 U.S. Department of Health and Human Services (USDHHS), A Report of the
Surgeon General: The Health Consequences of Smoking - Chromic and Obstructive
Lung Disease, Washington, DC, U.S. Government Printing Office, 1989.
96 M.R. Guerin, R.A. Jenkins and B.A. Tomkins, in: Center for Indoor Air Research
(Ed.), The Chemistry of Environmental Tobacco Smoke: Composition and Measure-
ment. Lewis Publishers, Chelsea, MI, 1992.
97 C.L. Benner, J.M. Bayona, F.M. Caka, H. Tang, L.D. Lewis and D.J. Eatough,
Environ. Sci. Technol., 23 (1989) 688.
98 California Environmental Protection Agency (Cal EPA), Health Effects of Exposure
to Environmental Tobacco Smoke, Office of Environmental Health Hazard Assess-
ment, 1997.
99 U.S. Environmental Protection Agency (USEPA), Respiratory Health Effects of
Passive Smoking: Lung Cancer and Other Disorders, USEPA-600-006F,
Washington, DC, U.S. Government Printing Office, 1992.
100 National Research Council (NRC), Environmental Tobacco Smoke: Measuring Expo-
sures and Assessing Health Effects, Committee on Passive Smoking, National
Academy Press, Washington, DC, 1986.
101 J.D. Spengler, D.W. Dockery, W.A. Turner, J.M. Wolfson and B.G. Ferris, Atmos.
Environ., 15 (1981) 23.
102 F.W. Henderson, H.F. Reid, R. Morris, O.L. Wang, P.C., Hu, R.W. Helms, L.
Forehand, J. Mumford, J. Lewtas, N.J. Haley and K.S. Hammond, Am. Rev. Resp.
Dis., 140 (1989) 183.
103 K.S. Hammond, Environmental Health Persp., 107 (1999) 329.
104 C.S. Cox and C.M. Wathes (Eds.), Bioaerosols Handbook. Lewis Publishers, Boca
Raton, F1, 1995.
105 T.A. Myatt and D.K. Milton, in: J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.),
IndoorAir Quality Handbook. McGraw-Hill, New York, NY, 2000, Ch. 42, p. 42.1.
106 N.L. Sprince, P.S. Thorne, W. Poppendorf, C. Zwerling, E.R. Miller and J.A.
DeKostner, Am. J. Indust. Med., 31 (1997) 403.
107 D.K. Milton, R.J. Gere, H.A. Feldman and I.A. Greaves, Am. Indust. Hyg. Assoc. J.,
51 (1990) 331.
108 M. Walters, D.K. Milton, L. Larsson and T. Ford,Appl. Environ. Microbiol., 60 (1994)
996.
109 Z. Mielniczuk, E. Mielniczuk and L. Larson, J. Microbiol. Meth., 17 (1993) 91.
110 T.A.E. Platts-Mills, in: J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.), Indoor
Air Quality Handbook. McGraw-Hill, New York, NY, 2000, Ch. 43, p. 43.1.
111 F. De Blay, P.W. Heymann, M.D. Chapman and T.A.E. Platts-Mills, J. Allergy Clin.

31
 Immunol., 88 (1992) 919.
112 M.D. Chapman and T.A.E. Platts-Mills, J. Immunol., 125 (1980) 587.
113 S. DeLucca, R. Sporik, T. O'Meara and E.R. Tovey, J. Allergy Clin. Immunol., 102
(1999) 174.
114 D.L. Rosenstreich, P. Eggleston, M. Kattan, D. Baker, R.G. Slavin, P. Gergen, H.
Mitchel, K. McNiff-Mortimer, H. Lynn, D. Ownby and F. Malveaux, New Engl. J.
Med., 336 (1997) 1356.
115 C.M. Luczynska, Y. Li, M.D. Chapman and T.A.E. Platts-Mills, Am. Rev. Resp. Dis.,
141 (1990) 361.
116 A. Custovic, R. Green, A. Fletcher, A. Smith, C.A. Pickering, M.D. Chapman and A.A.
Woodstock, Am. J. Resp. Crit. CareMed., 155 (1997) 94.
117 M.S. Muilenberg, in: J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.), Indoor Air
Quality Handbook. McGraw-Hill, New York, NY, 2000, Ch. 44, p. 44.1.
118 D.A. Sterling and R.D. Lewis, Ann. Allergy Asthma Immunol., 80 (1998) 279.
119 R. Yankova, Grana, 30 (1991) 171.
120 M.D. Lebowitz, M.K. O'Rourke, R. Dodge, G.C. Holberg, R.W. Hoshaw, J.L. Pinnas,
R.A. Barbee and M.R. Sneller, Environ. Int., 8 (1982) 375.
121 P.D. Moore, J.A. Webb and M.E. Collinson, Pollen Analysis. Blackwell Scientific
Publications, Oxford, Boston, 1991, Ch. 4.
122 Y. Takahashi, T. Nagoya, M. Watanabe, S. Sakaguchi and S.A. Katagiri, Allergy, 48
(1993) 94.
123 H.A. Burge, in: J.S. Spengler, J.M. Samet and J.F. McCarthy (Eds.), Indoor Air
Quality Handbook. McGraw-Hill, New York, NY, 2000, Ch. 45, p. 45.1.
124 M.P. Buttner and L.D. Stetzenbach, Appl. Environ. Microbiol., 59 (1993) 219.
125 P.M. Scott, in AOAC Official Methods of Analysis, 1990, Ch. 49, p. 1185.
126 J.L. Lange, P.S. Thorne and N.L. Lynch, Appl. Environ. Microbiol., 63 (1997) 1557.
127 American Society for Testing and Materials (ASTM), D5952-96: Standard Guide for
Inspecting Water Systems for Legionellae and Investigating Possible Outbreaks of
Legionellosis (Legionnaires' disease or Pontiac fever). In: Annual Book of ASTM
Standards,ASTM Committee on Standards.ASTM, West Conshohocken, PA, 1997,
Vol. 11.03.
128 S.K. Poruthoor and P.K. Dasgupta, Environ. Sci. Technol., 32 (1998) 1147.

32
Chapter2

Sampling water and aqueous solutions

Waldemar Wardencki and Jacek Namiesnik

2.1 INTRODUCTION

In analytical practice the importance of sampling cannot be over-emphasized; if

not carried out with care, errors can arise which cannot be corrected afterwards
and which consequently affect the results of the final analysis. Thus the results
obtained, instead of being the source of information, can lead to misinformation.
Liquid samples must be collected in such a way that they are representative
of the matrix being sampled at as many sampling sites as possible and maintain
compositional integrity prior to analysis. Because of the great diversity of liquid
samples, there is no universal sampling technique or typical sampling equip-
ment. There are only a few applications in which no special sampling system is
required and an appropriate sample container immersed in the analysed liquid
may be adequate. In most cases specific samplers and techniques peculiar to the
source should be used.
Liquid samples are usually taken from the following sources: surface waters
(inland and seawaters); groundwaters; surface microlayers (microfilms) of
water; potable (drinking) water; run-off waters and throughfall waters (from
streets, agriculture areas and roofs); industrial waters (for power plants and
from industrial installation); physiological fluids (urine, sweat, blood); and
commercial fluids (pharmaceuticals, industrial products, bulk solvents, etc.).
The aim of this chapter is to present the basic principles and equipment for
sampling some types of liquid samples, mainly water and wastewater samples.
The sampling approaches focus on sampling for chemical analysis and not for the
determination of physical or other properties. The content of this chapter is to
some extent based on the handbook entitled "Sampling of Environmental Sam-
ples for Analysis" published in Poland in 1995 PWN [1].
One of the basic requirements of sampling is to establish quality assurance
(QA) and quality control (QC) procedures that will ensure the reliability and
validity of the data collected. These procedures should comprise all aspects of the
sampling program, both in the field and the laboratory [2].

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
C 2002 Elsevier Science B.V. All rights reserved
2.2 TYPES OF SAMPLES FROM ENVIRONMENTAL WATERS

The sampling strategy should specify the types of samples to be collected. The
type of sample essentially depends on the character of water or wastewater, flow
variability of water and wastewater streams and required accuracy of the mea-
surements. The most common liquid samples are [3,4]: discrete samples, simple
composite samples, flow proportional composite samples, and sequential com-
posite samples.
A discrete sample is collected separately in its own individual container. It
represents the source at the time the sample was taken and is appropriate if
conditions at the source are essentially constant. The discrete samples are
collected for determination, e.g., pH, total organic carbon or dissolved oxygen.
A simple composite sample, also known as a time proportional composite
sample, comprises a series of identical volume samples (Vc) taken at constant
time intervals (tc) and deposited into the same container. It is defined by the
product of tcVc. In sampling a lake or tank waters a space proportional sample
can be also distinguished.
A flow proportional composite sample is collected depending on the water
flow during the sampling in such a way that it represents the average conditions
of sampling. It can be realized in two ways: (a) the sampler is triggered by a
signal from stream recorder, intervals between sampling of equal volume sub-
samples are changed depending on stream flow; (b) by changing the volume of
each sample depending on stream flow when the intervals between the sequence
operations are constant.
A sequential composite sample is formed from a series of individual samples
(typically from 2 to 8) placed in one container; each container represents a
determined interval (usually 1 h). This way of sampling is useful when the
stream of water or wastewater is not constant (when discharge of wastewater is
expected or self-purification conditions take place). In such conditions, for
example, high or small pH values would be difficult to determine in simple
composite sample.
Representative and appropriate samples of water or wastewater may be
collected in different ways. The general principle of an elementary selection of
sampling technique is given in Table 2.1 [4] and possible types of water sampling
are presented in Fig. 2.1 [5].
The discrete samples are usually taken from pipelines, reservoirs, rivers,
streams and sewers. The discrete sample can be collected either manually or by
using automatic samplers. Each sample should be a source of information on
water quality only at the time when it was sampled.
The composite sample may be obtained by manual mixing of individual
samples or by using an automatic system for sampling. Composite samples
represent the average concentration and constituent loads of the source during
the longer time. However, in most cases, abrupt changes of analyte concentra-
tions may not be noticed in the analysis of such samples. The collection of time

34
TABLE 2.1
Principles of preselection of sampling technique [3]
Rate of fluctuation of Small changes of stream flow Large changes of stream flow
analyte concentration
Small Discrete sample Discrete sample
Large Time proportional composite Volume proportional composite
sample samples

a) v/ , -1 -C)

f t
b) v d) v

111111]1111111111111111. t
iiilll iiiilllili II -
t

Fig. 2.1. Possible types of water sampling techniques: (a) discharge, (b) time dependent
sampling, (c) flow dependent sampling, (d) volume dependent sampling (based on Ref. [5]).

proportional samples consists of sub-samples of equal volume, sampled at equal

time intervals and placed in one container. For quantity proportional samples, in
which volume is proportional to the flow of stream, the samples are taken at
constant intervals of time. When the flow of stream is controlled continuously,
the composite samples are approximately proportional.
Investigations aimed at elaboration of new methods for collecting waste-
waters not completely mixed, with a high degree of inhomogeneity based using a
hydraulic models are known [6].
For representative samples of a stream cross section for constituent load
estimation (mass transport per time), vertical profiles are collected by moving
the sampler at a constant vertical velocity from the surface to the stream-bed at
equally spaced intervals (10 or more) [7].

2.3 GENERAL ASPECTS OF SAMPLING OF LIQUIDS

Due to the great diversity of possible pollutants that can be present in water, a
full analysis of water is not usually performed. Usually water is analysed in order
to evaluate its usefulness for a particular purpose and only specific and harmful
components are determined. The results obtained allow selection of the proper
purification methods or they are used for checking if a process was conducted
properly or if the water fulfils the requirements. Full descriptions of water sam-
pling and analytical methods may be found elsewhere [8-10].
The sample volume depends mainly on the substances to be analyzed and
their expected concentrations but 1 to 5 1 samples are commonly collected [11].

35
When developing a sampling strategy it is important to consider not only the
nature of water sample. The composition of the water in terms of the dissolved
compounds and trace fractions associated with colloids and suspended solids,
and the temporal and spacial variability at the sampling site should be also
considered. Furthermore, the chemical properties of the determinants and the
expected concentrations in the samples also influence the choice of sampling
methodology.
Water samples should be taken to completely sealed bottles, carefully rinsed
with sampling water. Any bubbles of air should be removed under the closing
system of the sampling vessels, otherwise, during transport or sample storage,
volatile components may pass into the headspace leading to their losses. In turn,
some components of gas phase may be dissolved in liquid phase. Both processes
may proceed up to reaching the equilibrium, determined by value of partition
coefficients of each analyte.
Care must be taken in all methods of collecting samples to ensure that
neither the sample storage containers nor any collecting vessels contaminate or
alter the sample. This can occur by:
- leaching of contaminants from the container surfaces
- leaching of chemicals (organic and inorganic) from container materials
- adsorption of trace metals or chemical compounds onto container surfaces
- reaction of the sample with container material
- change in equilibrium between pollutants present in particulate and
solution phases.

2.4 SAMPLING TECHNIQUES AND EQUIPMENT

2.4.1 Sampling of different types of surface water

The type of sampling system depends on the elements or compounds to be ana-

lyzed. Some glasses release sodium, silicon and boron. Therefore, when these el-
ements are to be analyzed plastic (i.e. polyethylene) bottles should be used. In
turn, glass bottles are recommended for determination of organic substances.
During sampling from streams, the bottle is kept at the bottom and the neck
of the bottle is directed against the current. The neck should be immersed in
water sufficiently deep to avoid the objects floating on the surface. It should be
noted that vertical stream cross sections are not commonly well mixed: a result
of sediment settling or proximity to the confluence of streams. Frequently
isokinetic samplers are required to avoid sampling bias by collecting water
through the sampling nozzle at the same velocity that water is flowing past the
sampler. An all-Teflon isokinetic sampler was recently proposed for assessing
sample contamination by mercuric species [7].
For sampling from different but required depths, samplers lowered into the
water using a metered tag line and then opened can be used. Such samplers are
called bathometers. After sampling, the water should be poured into clean

36
 a) b) 6 ,N1l

Left: Fig. 2.2. Bottle for sampling from different depths. 1: Tag line, 2: spring, 3: plastic tube, 4:
rubber stopper, 5: rubber stopper, 6: 1-1 plastic bottle with broad neck, 7: plastic tube, 8: brass
belt bracket (based on Refs. [11-13]).
Right: Fig. 2.3. (a) Collector for sampling from optional depths. 1: Tag line for lowering container
for determined depth, 2: rope for container opening, 3: metal reservoir for sample, 4: lead
basement. (b) Ruttner's sampler. 1: Plexiglass cylinder, 2: metallic lid connected to a rod, 3:
second lid, 4: spring catch on a barrel, 5: thermometer, 6: sinker, 7: tap water (based on Refs.
[11-13]).

bottles and transported to the laboratory. Figures 2.2 and 2.3 give examples of
sampler configuration for sampling from different depths [11-13].
When gases such as 02, CO2 or H2S dissolved in water are determined, a
problem can arise if water flowing into the bottle is mixed with the air inside the
empty bottle. The impossibility of rinsing the bottle with sampling water before
the final filling may generate additional errors. The sampler shown in Fig. 2.4
eliminates such inconveniences [14]. It can be deployed to a depth of up to 40 m.
The system consists of three connected bottles inserted in a lead holder and
closed with polished cork. The first bottle is connected to the atmosphere by a
polyethylene tube. A 40 m nylon tube linked to the third bottle is used for:
- lowering the sampler to the required depth,
- rinsing the system with gas during lowering, e.g., with nitrogen from a
pressurized tank, to avoid filling the bottle at the wrong depth,

Fig. 2.4. Sampler for collecting water from depths of up to 40 m. 1: Hold-fast system to regulate
force of bottle tightening, 2: spring, 3: lead sinker (based on Ref. [141).

37
- connecting the sampler to the atmosphere during filling,
- preventing the losses of sample by tight closing after the sampling but before
its withdrawing to the surface.
Purging with gas stops when the sampler reaches the required level and the end
of the nylon tube above the water surface is open. The first bottle is filled, fol-
lowed by the two others. In this way, the first bottle is rinsed with a water vol-
ume equal to the volume of the other two bottles. The content of the first bottle is
used for determination of gaseous components. Samples from the remaining bot-
tles can be used for different chemical and biological analyses.
The modification of the original Bodman's sampler, called Bodega-
Bodman's sampler [15] (Fig. 2.5), made of high purity aluminium, stainless
steel, Teflon and Viton allows the collection of samples of up 90 litres through
the whole column of water, reducing sample contamination to minimum. An
easily dismantled swagelock type valve at the lid of the sampler for connecting
the purging nitrogen line from the cylinder and stainless steel ball valve placed
at the bottom enables the transfer of the sampler content to an aseptic room for
further treatment without contact with the atmosphere. The outlet valve at the
centre of the bottom lid reduces the possibility of losses of greater, easily sedi-
mentated particulate fraction. A 50 kg sinker, in the form of lead disks, is
attached to the bottom of the sampler. A manual sampler made of transparent
acryl may be also used for the same purpose [16].
Figure 2.6 shows a multichannel (1-4 channels) sampler for collecting sam-
ples for the determination of volatile organic compounds [17]. Each syringe has

.I ra
-U.. 1 I I - I I 1
i2

I1 -
II <2

r___ ------
- 2_ __
c---
- - ---
__
_:_ -
<-3
--- ___ __
I

-~A dK' In -4
-

_ f
/4 p#iajr---1PR
-j.1 _L~-

Left: Fig. 2.5. Bodega-Bodman's sampler: (a) open, (b) closed. 1: Upper construction plate, 2:
upper lid, 3: pressure dropping valve, 4: holding rods, 5: cylinder collar, 6: stainless steel leading
rods, 7: sampler cylinder, 8: 12 kHz pulsator, 9: bottom collar, 10: teflon o-ring, 11: bottom
construction plate, 12: ball valve, 13: ratchet mechanism, 14: magnet of pulsator switcher (based
on Ref. [15]).
Right: Fig. 2.6. Schematic representation of multichannel syringe sampler of water. 1: Syringes,
2: screw for regulation of water volume, 3: circular cam, 4: DC motor (based on Ref. [171).

38
 to shin
·-;
-2

Left: Fig. 2.7. Modified Slocum's sampler for collecting and in situ preconcentration of organic
analytes from water. 1: Teflon stopper (removed under the water surface), 2: steel ropes, 3:
holder of stainless steel filter, 4: reduction connector, 5: supporting frame (steel rods), 6: inlet
tube with blades for directing inlet according to water stream, 7: swagelock-type connection, 8:
connector to sorption tube, 9: teflon column filled with high density polyurethane foam, 10:
nylon collar, 11: column connector to sorbent, 12: one-way valve, 13: connection to tube thread,
14: steel catch, 15: steel catch belts, 16: vacuum rubber tube leading to surface (based on Ref.
[151).
Right: Fig. 2.8. Equipment for water sampling by the "drop-bottle" method. 1: Displacement
buoy, 2: sea level, 3: frame holder of glass bottle, 4: lead sinker (based on Refs. [18,19]).

its own sampling line (Teflon tubes) equipped with a three-way electromagnetic
valve that enables the sampled water to be delivered to a container with a
floating lid. This construction eliminates the headspace above the liquid surface
and avoids the losses connected with partition of organic analytes between
aqueous and gas phases. The use of Teflon and glass as the construction
materials minimizes losses of volatile components by reducing the surface
adsorption. The action of the sampler is controlled by computer using radio-
telemetry.
Frequently, the amount of particulate fraction in water samples is equally
important as compounds dissolved in water. Modified Slocum's sampler [15],
shown in Fig. 2.7, enables independent collection of suspended particulates on
stainless steel filter inserted in the sampler holder and preconcentration of
dissolved organic compounds on high density polyurethane foam, placed in
Teflon adsorption tube. The total weight of the sampler is about 23 kg.

39
Left: Fig. 2.9. DHI sampler (Deutsche Hydrographisches Institut) for sampling deep-water. 1:
Glass tubes, 2: tight closure, 3: movable metallic arm for breaking glass tubes, 4: glass containers
(based on Refs. [18,20]).
Right: Fig. 2.10. Schematic representation of sampler for collecting water from depths of up 3 m.
1: Operational handle, 2: casing of operational rod, 3: holding belts, 4: rod, 5: spring catch, 6:
polyethylene bag, 7: hold-fast element, 8: sampling bottle, 9: grip, 10: deep indicator, 11:
operational rod (based on Ref. [21]).

Water samples from open water bodies from depths of 0.5-1.5 m may be
collected using the "drop-bottle" system [18,19]. The equipment for such samp-
ling is presented in Fig. 2.8. For sampling from deeper layers, bathometers
elaborated by Stadler, commonly known as DHI samplers (Fig. 2.9) are
frequently used [18,20]. The samples are collected to exchangeable glass flasks,
which can be washed and protected against contamination by hermetic closing.
The glass tubes that protect the holes are broken at the required depths by
moving a lever. Extraction and separation of mixtures are performed in the same
flask, which eliminates the need for liquid pouring and a separatory funnel, thus
reducing the possibility of sample contamination. Thick-walled flasks are used
for collecting the samples from depths of up several thousands meters.
Another type of sampler for determination of volatile organic compounds is
shown in Fig. 2.10 [21]. It is essentially made of stainless steel and aluminium
and partly of brass and polypropylene. During the sampling an empty bottle (100
ml capacity) closed by a polypropylene stopper, and provided with an injection

40
 a) b)

r---q

Fig. 2.11. (a) Czapla-1 sampler; 1: Grip, 2: telescoping shaft, 3: striction system, 4: o-ring, 5: body,
6: screw, 7: dipper. (b) Nurnik-1 sampler; 1: regulation of lock force, 2: lock, 3: lock, spring catch,
4: striction system, 5: bearing ring, 6: sinker, 7: dipper (based on Ref. [22]).

chamber with a silicon rubber membrane is firmly attached to the sampler

basement. The bottle with basement is sheltered by a thin polyethylene bag to
avoid contamination of the bottle by the surface hydrocarbon layer. The bag is
tightly rolled around the aluminium casing of the operational rod and guarded
by a spring catch, fastened to the tag line kept by operators. For sample
collection the sampler is carefully dropped into the water, e.g. from a boat, to the
desired depth and kept at the required site using two belts. The polyethylene bag
is removed downwards by releasing a spring catch using the operational rod
attached to the bag. Then, the bottle is opened with the operational rod, filled
completely with water and then closed with a cap. The sampler is carefully
withdrawn to the surface and the bottle with the sample is released.
Relatively simple samplers can be used for sampling water and sewage from
open ducts such as by-pass ducts, overfalls or different modification of scoop
systems.
In hydrochemical investigations two types of sampler are normally used [22]:
the Czapla-1 sampler with self-acting closure (lock) for sampling bottom sedi-
ments from maximum depths of 1.5 m (Fig. 2.11a), and the Nurnik-1 sampler,
closed with a spring mechanism, for sampling bottom sediments from ponds,
rivers and lakes and other water regions from depths of up to 50 m (Fig. 2.1 lb).

41
2.4.2 Sampling of groundwater from small boreholes

Groundwater is sampled to determine its quality and the quality changes with
time [23-25]. Equipment for this purpose was developed for the U.S. Geological
Survey. Samplers for collecting groundwater must fulfil several specific de-
mands [26]:
- the sampler must enable the collection of samples from high depths,
- it must reach the bottom of the drilling hole and collect the sample,
- the material of which the sampler is made must not contaminate the water
sample,
- the sampler must be rinsed with water before collecting the sample,
- the sampler must be capable of being closed immediately after finishing
sampling in order to avoid losses of volatile compounds and protect the
sample from contact with air,
- the sampler must be able to avoid the degassing of water samples by sudden
changes of pressure during the sampling,
- the sampler must be easy to transport to the laboratory.
Therefore, during the designing of such samplers the following requirements
should be considered [26]:
- a small diameter drilling hole (ca. 5 cm),
- the inertness of material for the sampler (stainless steel or Teflon),
- the ability to collect samples from a depth of about 60 meters,
- the container for rinsing water should be three times larger than the
container for the sample,
- the system should be light, portable and easy to operate,
- minimizing the usage of pressure-operated elements (pressurized tank
containers and pumps for gases should be eliminated).
Figures 2.12 and 2.13 show schemes of eight samplers for collecting the ground-
water from drilling holes [26]. In first four samplers, the piston is used to pump
water through the cylinder for samples.
The Geological Survey of Canada developed a sampler for collecting samples
from very small diameter holes drilled by diamond drills [27]. The sampler
shown in Fig. 2.14, made of stainless steel and equipped with a portable travel-
ling block, enables the collection of samples of about 350 ml from small holes at
any depth. A steel rope passes through the central pipe of the sampler and is
fastened to its bottom. Water flows through the sampler during its lowering into
a drill hole thanks to spring-operated jaws installed in holes at the top of the
sampler. After achieving the required depth, a so-called messenger is dropped at
the bottom to a hole along the steel rope. At that moment the jaws are opened
and a sampling tube is liberated which can drop onto a rubber stopper at the
bottom of sampler. Although the top of the sampler is still open, the mixing of
water is minimized during the withdrawal of the sampler. A properly designed
sampler enables the collection of samples from drilling holes that are at angles of
up to 45° .

42
 a) 1 b) c) d)
1 1 1

g2 .2 2 2

3 3
3
3 4
4
5
5

4
~4~ ~6 6

5 1 5 I 7 7

6 8 8

7 9 59
7

Fig. 2.12. Schematic representation of the construction of four samplers for collecting water
from drilling holes. (a) Sampler with coil driven by piston; 1: coiled cable, 2: returning spring, 3:
piston, 4: valve, 5: sample container, 6: valve, 7: controlling valve. (b) Sampler driven by manual
pump; 1: hydraulic pressure, returning spring, 3: piston, 4: valve, 5: sample reservoir, 6: valve, 7:
controlling valve. (c) Sampler with two tubes; 1: sampling tube, 2: tube of pump, 3: double piston,
4: returning spring, 5: controlling valve, 6: valve, 7: sample reservoir, 8: valve, 9: controlling
valve. (d) One-way sampler driven by manual pump; 1: tube for sampling and pump, 2:
controlling valve, 3: double piston, 4: controlling valve, 5: returning spring, 6: valve, 7: sample
container, 8: controlling valve (based on Ref. [26]).

a) b) c) d) 1
2

22

3
~2 ~ ~
3 4

4 5
6
4

Fig. 2.13. Schematic representation of the construction of four samplers for collecting water
from drilling holes. (a) Sampler driven by electric motor; 1: electric motor in sampler head, 2:
thread, 3: piston, 4: sample reservoir, 5: valve, 6: controlling valve. (b) Sampler with electric gear
wheel pump; 1: gear wheel pump, 2: pressure controlling valve, 3: valve, 4: sample container, 5:
valve, 6: controlling valve. (c) Sampler driven by motor; 1: electric motor in sampler head, 2:
thread, 3: upper piston, 4: membrane, 5: cylinder, 6: bottom piston. (d) Manual sampler; 1: cable,
2: opening system, 3: piston sampler, 4: piston, 5: valve, 6: piston, 7: sample reservoir, 8: valve, 9:
controlling valve (based on Ref. [26]).

43
 a) b)
2,5 cm

membrane
=Z.
E
M

nn
C
/1

W/2

r•
3

3
:4

U
I P 5~
5t_ 5

Left: Fig. 2.14. Schematic representation of the construction of a sampler for collecting a sample
from small diameter holes. 1: Casing, 2: sampler tube, 3: rubber closing cap, 4: spring-operated
jaws (based on Ref. [271).
Right: Fig. 2.15. (a) Sampler for in situ collecting of water samples from drilling holes; A: tip
filter, B: disposable double hypodermic needle, C: sterilized, pre-evacuated glass ampoule. (b)
Schematic representation of a sampler for sampling and simultaneous enrichment of analytes.
The flow of water through sorption tubes is controlled by a reversing valve, needle of syringe,
placed at the outflow of each sorption tube. 1: Steel tube (to surface) of 0.45 cm diameter, 2:
swagelock-type connectors, 3: brass sinker, 4: reversing valve with needle, 5: sorption tubes, 6
stream of water from drilling hole (based on Ref. [28]).

Figure 2.15a presents another configuration of the sampler for in situ

collection of water and gas samples from the bottom of small diameter holes
(about 3-5 mm) [28]. It consists of three basic elements: (1) a tip filter firmly
installed at the bottom of the sampler; (2) a double disposable hypodermic
needle; and (3) a sterilized, pre-evacuated glass ampoule.
Depending on the drilling hole diameter, the glass ampoule allows collection
of samples in the range of 35 and 500 ml from small diameter holes (about 3-5
mm). The ampoule, inserted in a heavy holder, after dropping at the bottom is
connected to the tip filter and the needle penetrates through the membrane.
Because the ampoule is evacuated the water and gases from drilled geological
formation are sucked into it.

44
 a)

Fig. 2.16. Schematic representation of simple sampler

for collecting samples of water from small drilling
holes with simultaneous enrichment of organic
analytes: (a) sorption tube, (b) syringe with Teflon B
piston. 1: Tenax-GC sorbent bed, 2: polyethylene tube
for connection of surface pump, 3: holding spring, 4:
teflon, 5: teflon piston, 6: o-rings, 7: holding spring
(hasnl nn R fe [Fl11)

Samplers for continuous preconcentration of volatile organic compounds

using an appropriate sorbent (Tenax-TA) during the sampling from drilling
holes have been also described [29]. Such operations are carried out after
dropping the sampler (up to 50 m) into a drilling hole. The sorption tubes are
disconnected after the sampling/preconcentration step when the sampler is
withdrawn to the surface. After transportation to a laboratory the tubes are
heated to release the adsorbed analytes before chromatographic analysis. A
scheme of such a sampler is shown in Fig. 2.15b [29]. The sampled water
contacts only such materials as stainless steel, glass and sorbent. The beginning
and end of sampling, water flow and the volume of sample are controlled from
the soil surface.
A relatively simple sampler for collecting samples from drilling holes with
simultaneous isolation/preconcentration of organic constituents is shown in Fig.
2.16 [30]. It consists of a small sorption tube attached to a syringe filled with
sorbent. The action of the pistol in the syringe is controlled from the surface by
applying pressure or a vacuum. The total volume of the syringe determines the
volume of the sample collected. The tube is disconnected after withdrawing the
sampler from a hole to the surface and transporting to a laboratory where it is
heated and the released constituents are analysed chromatographically.

2.4.3 Sampling of unsaturated soil water

The quality of groundwater can easily be spoiled by polluted surface water leach-
ing into deeper aquifers often used as sources of high-quality drinking water.
The leaching water may be contaminated by, e.g., pesticides used in agriculture,
migration of pollutants from landfills or from drainage. Such seepage waters
form an unsaturated water zone. A knowledge of the chemistry of waters con-

45
 D vacuum vacuum
Fig. 2.17. Schematic representation of a system for collecting water using a lysimeter for
determination of unsaturated water zones. Numbers in circles from 1 to 10 indicate a particular
route of the sample in a system; 1: 5-cm3 syringe, 2: silicon tube, 3: three-way valve, 4: 5-cm3
ampoule, 5: scrubber with porous partition, 6: sorption tube, 7: burette (based on Ref. [31]).

tained in porous soil is necessary to assess the degree of pollution of groundwa-

ter. The sampling of unsaturated soil water allows us to estimate the flux of
contaminants through the column of water above aquifers [31]. The most popu-
lar method used for sampling of such waters is one using different types of
lysimeter. Regardless of their construction, lysimeters need vacuum to force the
flow of water contained in porous soil into them. Suction sampling should pre-
vent the sample degassing to avoid losses of volatile organic compounds; this re-
quirement cannot be fulfilled when peristaltic pumps are used for this purpose.
Figure 2.17 presents the scheme of a configuration used for the sampling of
30 5 ml water for trichloroethylene determination [31]. The sample is deliver-
ed to a sampler by suction (1-2-5-9 connection). By different connections of the
elements of a system (1-2-5-9, 1-3, 1-3-6-8 and 1-2-4-9) it is possible to analyse
the sample by applying four different analytical methods.

2.4.4 Sampling surface layers (microfilms) of water

The surface microlayer is the part of each water body in which a large number of
contaminants, including insoluble organics and metals are normally concen-
trated. Representative sampling of this layer is rather difficult because the film
is only some 100 m thick. Therefore, a great deal of research is focused on devel-
oping efficient methods of sampling the surface microfilms, frequently thin oily
films, concentrated in organic compounds, both polar and non-polar [32-35].
Organic films on the surface of seawater, biogenic and anthropogenic in
origin, play an important role in seashore waters. Such films affect the physical
and chemical processes taking place in surface seawaters. Special sampling

46
 l

C .' 1

Fig. 2.18. (a) A dip-type sampler with double walls for sampling the surface layers of water. (b) A
section of sampler against A-B line, 1: rectangular plates, 2: cutoff frame, 3: Langmuir's channel
section, 4: glass slide, 5: surface with microfilm, 6: Wilhelm's plate for surface tension measure-
ment, 7: thread, 8: valves, F: connection with torsion balance. The water level in the sampler, C:
during detaching, D: in stand-by position, E: before measurement by using Langmuir's channel
(based on Ref. [36]).

devices are required for collecting of such films. Figure 2.18 presents a scheme of
a submersible rectangular sampler with double walls which cuts off the surface
microlayer from the adjacent subsurface water layer [36]. The sampler consists
of a rectangular plate (40-50 cm) of 6 cm depth made of plastic material with a
shallow Langmuir channel section (30x40x0.7 cm) at the lower part. However,
the so-called Garret's sampler, a metal screen (50x65 cm), is more frequently
used for collecting microfilms of water. It was found that the sampler collected a
surface layer of 0.3 mm thickness [17,37]. The equipment for collecting samples
of the surface microlayer by Garret's method is shown in Fig. 2.19.
Another type of sampler, designed at the Chemical Faculty of the Technical
University of Gdansk, for collecting the surface microlayer is presented in Fig.
2.20. The sampler allows the determination of the oil film thickness on the
surface of water. The water sample is taken from cylinder (1) to glass tube (5)
with a diameter several or more times smaller than the cylinder. The thickness
of the collected film can be measured very simply, depending only on the
diameter of the exchangeable cylinder. Film thickness can be calculated on the
basis of the distance between the pointers (9). The ball of the closing valve (4) is
lifted to about half the height of the glass tube using rope (12). Next, the sampler
is placed in the sampling site perpendicularly to the water surface and immersed
deeply enough that the buoys (7) cause the sampler to float. When the rope slack,
a special spring mechanism prepares to action the rope with the ball attached to
it. After the second tightening of the rope (12) for withdrawing the sampler, the
spring mechanism is liberated, the rope is freed, the ball is lowered and the
cutting-off socket is closed.

47
 1

Left: Fig. 2.19. Sampler for collecting microfilms of water by Garret's method. 1: Garret's
sampler metal gauze, 2: gauze holder, 3: bottle for the sample (based on Refs. [17,37]).
Right: Fig. 2.20. Sampler for collecting a surface microlayer. 1: Exchangeable cylinder, 2: cut-off
sockets, 3: metal o-ring (hung on cut-off rope), 4: ball, 5: glass tube, 6: guarding and mounting
tube, 7: buoys, 8: scale, 9: scale of oil level, 10: lid with special handle, 11: cutoff ball and device
suspension, 12: line.

2.4.5 Sampling of throughfall waters

The sampling of rainwater needs special care and equipment. The simplest sam-
plers for this purpose [38-40] are open containers with known surface, that in
stand-by state are closed with lids. Samples collected in such samplers are used
for typical physicochemical analysis. Samplers of a more complex constructions
are used for the determination of dissolved organic compounds in water; such a
sampler is shown in Fig. 2.21 [29], consisting of the following elements:
- a collecting surface (0.81 m2 gives an 8.1-1 sample for 1 cm rainwater); it is a
specially profiled Teflon plate with a hole in its centre,
- a glass container to which two stainless steel electrodes are installed at
different heights on the side wall, acting as sensors of the water level inside
the container,
- a unit for isolation/preconcentration of organic compounds (adsorption
tubes) from collected water. The flow of rainwater through the adsorption
system is forced by peristaltic pump. The pump is switched on and off by a
signal from the lower electrode sensor. When the rainwater level exceeds the
upper sensor the flow is increased twofold.

48
 20

Fig. 2.21. Sampler for collecting rainwater for determination of volatile organic compounds. 1:
Glass reservoir, 2: glass fibre filter, 3: filter casing, 4: union, 5: glass T-tubes, 6: Teflon unions, 7:
ball joining, 8: top electrode, 9: set of electrodes (front view), 10: stainless steel rod, 11: bottom
electrode, 12: teflon guarding, 13: electric cable, 14: sorbent tube (Tenax-GC) for liquid
extraction, 15: sorbent tube (Tenax-GC) for thermal desorption, 16: connecting glass tube, 17:
stream splitter (capillary), 18: Teflon tube (to pump), 19: Teflon connector, 20: collecting surface
(based on Ref. [29]).

2.4.6 Sampling of water and vapour from closed circuits

A typical configuration for sampling water and vapour from closed circuits con-
sists of a probe, a tube connecting a probe to a condenser, a condenser and a
choke appliance. When the installation is designed for continuous measure-
ments it should be equipped with an additional screen filter with 0.25 mm mesh
and a surface of 25-30 cm 2 . A thermic fuse is also needed to protect the control
device from damage by high temperatures. A system for sampling water and
vapour for periodical analyses is presented schematically in Fig. 2.22 [41], and
for continuous measurements in Fig. 2.23 [41].
Figure 2.24 shows a probe for collecting samples from pipelines [42]. The
internal diameter of the probe depends on the diameter of the pipeline in which
it will be installed. The probe cannot be used for pipelines with diameters
smaller than 50 mm. In such cases, a tube leading to a condenser should be
connected directly to a connector welded to the pipeline wall. The diameter of
the hole in the pipeline wall and the internal diameter of the connector should be
equal to the internal diameter of the tube leading to the condenser. A probe
should be installed on the straight segment of the pipeline in a such way that the
distance between the site of its installation and a valve, a slide or pipeline curve
is at least six times higher than the diameter of the pipeline. A probe can be
installed, depending on the local situation, to perpendicular or horizontal pipe-
lines, with a downward or upward flow of water. If possible, the probe should be
placed either horizontally to a perpendicular pipeline with a downward flow of
water, or perpendicularly or horizontally to a horizontal pipeline.

49
 6

Fig. 2.22. Schematic representation of an installation for

sampling water or vapour for periodical analysis. 1:
Probe, 2: pipeline, 3: blocking valves, 4: tube connecting
probe with cooler, 5: cooler, 6: controlling valve (based on
Ref. [41]). inflow of cooling water

inflow of for control analyses

cooling water

Fig. 2.23. Schematic representation of an installation for sampling water or vapour for conti-
nuous measurements. 1: Probe, 2: pipeline, 3: blocking valves, 4: tube connecting probe with
cooler, 5: by-pass for rinsing of probe and tube 6: filter, 7: controlling valve or throttle, 8: fuse, 9:
sensor of measuring device, 10: cooler (based on Ref. [41]).

A schematic design of a probe and method of sampler installation for the direct
collection of water from containers without a temporary flow of water is shown in
Fig. 2.25. A tube with an internal diameter of 10-12 mm is used as a probe. Its
opening should be placed at the height of the pipeline outlet. For containers with
water flow or to which different waters flow in, a probe located on the outlet pipe-
line, possibly in the vicinity of a container (Fig. 2.24) can be used.

50
 direction of

water flow

Fig. 2.24. Installation of a probe for sampling water from pipelines (based on Ref. [42]).

inflow pipe

010

I
to condenser

Fig. 2.25. Installation of a probe in a reservoir (based on Ref. [42]).

For sampling water from horizontal barrel boilers a probe such as that
shown in Fig. 2.26 is recommended [43]. A probe cannot be installed in the
vicinity of a tube delivering correction chemicals to a boiler. It should also be
noted that the probe must be placed about 50 mm beneath the allowable level of
the water in the boiler. The part of the probe installed on the supports inside the
boiler should be resistant to vibration and forces appearing during the water
flow. The external part of the probe should be strengthened and equipped with
exchangeable connectors of different diameters.

_ S·-~-99~
2

I%
<4
I'~r~~
Fig. 2.26. Schematic representation of a probe for
sampling water from barrel boilers. 1: Orifices, 2:
maximum level of water, 3: common level of water, 4:
minimum level of water (based on Ref. [43]). to condenser

51
2.4.7 Sampling of piped water for determination of volatile
compounds and dissolved gases

Because the concentration of contaminants in water from pipes may change with
distance from the distribution tank and the residence time in the pipe, the condi-
tions for sampling of such water should be defined with respect to sampling site
(along the distribution grid), time of sampling (first open or after stagnation),
number of samples, sampling intervals, pipe material and residence time in the
pipe. The sampling of piped water should be conducted in such a way that the
sample is protected from contact with the atmosphere. After connecting a clean
rubber tube to a tap, a valve is opened and water flows for 10 min. Next, the sec-
ond end of the tube is placed on the bottom of the bottle and water flows into the
bottle until it is full. Before the sampling is finished, the water in the bottle
should be exchanged several times. After that the tube is withdrawn and the bot-
tle is immediately closed without leaving any headspace.

2.4.8 Automatic sampling systems

Because the concentrations of pollutants change over time, manual sampling

can be substituted by automatic sampling systems. Different types of systems
may be used: from simple ones which can be easily constructed by the user to
very sophisticated ones produced by manufacturers [44].
The use of monitors for continuous monitoring of water quality has in-
creased in last years. Samples are collected automatically at 15 minute intervals
for analysis of anthropogenic pollutants [45]. The water can be collected using
different pumps, usually peristaltic ones. Air-pressurized or vacuum-operated
systems can also be used for sampling water or wastewater.
Automatic samplers can be stationary devices, frequently built in closed or
open water ducts or wastewater lines, or portable devices commonly used in the
fields. All automatic systems have the following modules: power supply unit,
control unit, sample splitting assembly, sample storage segment.
Water can be delivered to the sampler by peristaltic pump, syringe pump,
membrane pump (in form of diaphragm), cogged submersible pump and
eccentric submersible pump [46,47].
A scheme of an automatic system based on unforced water discharge with
dosing a constant volume sample is shown in Fig. 2.27 [5]. In this way a constant
sample volume is always obtained. Highly polluted wastewaters may easily clog
the dosage container.
The portable automatic unit for in situ operation is described in Ref. [48]. It
can carry out the following functions: collect the sample from water stream,
preconcentrate trace amount of metals in collected samples, and extract the
organic pollutants.
Automatic operation enables long periods of work without service. An auto-
matic sampler module is shown in Fig. 2.28: it consists of two main elements-an

52
 lifting turning measuring bottling

Fig. 2.27. Schematic representation of automatic collection of water, based on free water outflow
with delivering the constant volume samples. (a) Sample drawing, (b) turning away the excess of
water, (c) measuring sample volume, (d) transferring the sample into the reservoir (based on Ref.
[51).

Fig. 2.28. Portable automatic system for sampling

a time proportional sequential composite sample.
1: Burette, 2: automatic valve, 3: sample
container, 4: stream of water (based on Ref. [48]).

automatic valve and a burette. A sampler is connected to the water source (in the
simplest case, a tap) by Teflon tubing. The stream of water flows continuously
into the sink through a valve and a burette with a total capacity of 32 ml. After a
predetermined time established with the control unit, the motor changes the po-
sition of the valve and the water is directed from the burette to a reservoir. An
electronic counter records the collection of the water sample. The combination of
many discrete samples in defined time intervals allows us to obtain time sequen-
tial composite samples.
Recently, a new device for automatic sampling of runoff during rainfall
events has been described [49]. It is easily made, simple to use, not vulnerable to
operational disturbances and inexpensive. The basic principle is to use the
weight of the rainwater to trigger the sample collection.

2.4.9 Passive water sampling

A new approach to sampling organic pollutants in aqueous environments is pas-

sive sampling, known as passive dosimetry. The latter term underlines the dif-
ference in the method of sampling analytes from water surrounding the sampler,
and common dynamic techniques where analytes are absorbed or adsorbed in/on
the trap. In passive sampling the flow of analytes from the aqueous sample sur-

53
rounding the dosimeter into the inside part of a trap placed in the sampler is
completely free (according to Fick's first law of diffusion). The main driving
force and separation mechanism are based on the differences in concentration.
The equipment used for sampling is therefore not complicated, which is of great
importance since sampling sites are very often situated far from the laboratory.
Another advantage of the passive approach over grab sampling is that only
one device is necessary at a given sampling location for the duration of sampling.
A passive sampler can cover a long sampling period, integrating the pollutant
concentration over time. Moreover, decomposition of the sample by transport
and storage and/or changes during the sample enrichment step (e.g. contamina-
tion, degradation, adsorption) are minimized. Besides, in contrast to dynamic
techniques, passive sampling is simple to perform and after isolation and/or
enrichment, no further sample preparation treatment is usually required. Fur-
thermore, passive sampling is less sensitive to accidental extreme variations of
the organic pollutant concentration in natural waters.
Passive methods may generally be classified as adsorptive or absorptive.
Adsorptive methods take advantage of the physical or chemical retention by sur-
faces and key parameters involve surface binding and/or surface area. Absorp-
tive methods involve not only surface phenomena but also analyte permeation in
the interceding material. This latter approach provides the possibility of com-
pound discrimination due to the membrane's physicochemical characteristics.
The membrane, constituting of a diffusion barrier for transported molecules
of the analyte, is the most important element of permeation passive dosimeters;
hence, the material from which membranes are made should meet specific
conditions such as a large permeability coefficient for the analytes, which is
directly related to the kind of membrane material and to membrane thickness
and homogeneity, so that there are no differences between dosimeters used
during simultaneous exposure. Because permeation is also determined by
dissolution in the membrane, the transport of pollutants through the membrane
is influenced to a large extent by the nature of the membrane material.

2.4.9.1 Designs of passive samplers

Membrane-based passive samplers are based on the process of passive partition-
ing of a compound between water and a lipophilic material enclosed in a
semipermeable polymeric membrane. Several membrane sampler designs have
been proposed, including: solvent-filled devices; semipermeable membrane de-
vices (SPMDs); passive in situ concentration/extraction sampler (PISCES); sup-
ported liquid membrane (SLM) technique; sorbent-filled devices.
Of all the passive systems, semipermeable membrane devices (SPMDs) are
the most frequently used to sample organic contaminants. They can be used to
detect a variety of lipophilic chemicals in water, sediment/soil, and air [50].
SPDMs are designed to sample non-polar, hydrophobic chemicals. The maxi-
mum concentration factor achievable for a particular chemical is proportional to
its octanol-water partition coefficient.

54
TABLE 2.2
General specification of SPMD devices [51]
Parts of SPMD devices Material
Membranes: thin-walled Polyethylene; Silicone or silastic (option of
(25-250 pum) non-porous polymer tubes plasma-treated surface); Polypropylene;
Polyvinylchloride
Sequestration phase: high-molecular- weight Neutral lipids; Silicone fluids; Other lipid-like
(>600 Da) non-polar liquids or fluids organic fluids

SPDMs consist of tubular low-density polyethylene (LDPE) lay-flat mem-

brane manufactured without additives and filled with a high-molecular-weight
lipid: typically high purity synthetic triolen 1,2,3-tri-[cis-9-octacenoyl]glycerol
(95%). The general specification of SPMD devices used for long-term monitoring
analyses of pollutants in aquatic reservoirs is presented in Table 2.2 [51]. The
general dimensions are: 2.5 cm wide (layflat) by 86.4 cm long tubes (wall
thickness -80 gm) containing 1 ml (0.915 g) of triolein as a thin film.
When placed in an aquatic environment, SPMDs passively accumulate hy-
drophobic organic compounds. The LDPE tubing mimics a biological membrane
in its ability to allow selective diffusion of organic compounds. Triolein is a major
non-polar lipid found in aquatic organisms. The passive sampling of the hydro-
phobic chemicals is driven by lipid-water partitioning.
SPMDs have all the advantages over traditional methods of water sampling
that other passive sampling methods exhibit: they are easy to use, can be
standardized and may be representative of the thermodynamically dissolved
organic phase in surface waters. They can be deployed for long periods of time
(days to months) and used to estimate the time-weighted mean concentrations of
the hydrophobic organic compounds in a water body. The real drawback of the
SPMD technique is that SPMD extracts are readily amenable to a toxicity
identification evaluation. A second problem is the time-consuming sample treat-
ment procedure. Although analyte recovery and enrichment procedures for
SPMDs generally require less effort than those for tissue and sediment matrices,
without sample treatment the extracts obtained are not suitable for any analy-
ses. A detailed description and examples of applications of all types of passive
samplers with figures can be found in a review by Kot et al. [52].

2.5 GENERAL ASPECTS OF PREPARING WATER SAMPLES FOR

ANALYSIS

Samples of collected water are usually pretreated in different ways before final
analysis. Some steps of the treatment can be carried out either during or imme-
diately after the sampling process. The main aims of these procedures, which are
frequently labour- and time-consuming, are [53-56]: ensuring the appropriate
physical properties of the samples and removal of any interfering components;

55
TABLE 2.3
Sample pretreatment methods for water [57]
Initial operation of sample Procedure
pretreatment
Removal of suspended Filtration, centrifugation
matter
Preservation pH adjustment to make more generally applicable (e.g. acidif-
ication); biocide addition; derivatization of analytes; application
of UW radiation; storing of sample at a temperature of 4°C.
Isolation and/or Solvent extraction; supercritical fluid extraction; solid phase
preconcentration of extraction; extraction into gaseous phase; membrane extraction;
analytes osmosis processes and ultrafiltration; freezing and lyophilisation.
Enrichment of extracts Evaporation of solvent excess, e.g. in Kuderna-Denish apparatus.
Purification of extracts Liquid chromatography; gel chromatography.
Extracts drying Salt adding, e.g. NaSO,4.

sample preservation; transfer of analytes into another matrix, more convenient

for a given analytical method; and preconcentration of analytes. The basic steps
for preparing water samples for final analysis are given in Table 2.3 [57].

2.5.1 Preservation and water sample storage

Usually, it is difficult to begin the analysis of water immediately after sample col-
lection. However, it should be noted that during storage many changes in the
sample may occur, and ensuring the integrity of the sample is as important as
the analysis itself. The character of the sample, the diversity and differing rates
of processes induced by many factors mean that there is no one universal method
of ensuring the stability of sample composition. Changes in sample composition,
and their reasons, between the sampling and final analysis are different and can
be caused by different reactions; the most important are chemical reactions,
physical processes, biochemical and photochemical reactions.
For chemical reactions, oxidation and reduction processes or hydrolysis of
chemical substances are quite possible. The reaction of chlorine with humic
materials and formation of halomethanes may occur. Depolymerization of con-
densed inorganic phosphates and polymeric silicic acids are also possible.
Changes of pH, as a result of absorption of CO 2 from air, lead to decreasing hard-
ness by precipitation of calcium carbonate. Some components of the sample may
form undissolved salts. Instability of the sample caused by chemical reactions
may be prevented or minimized mainly by the addition of an appropriate chemi-
cal reagent or by pH control.
Among physical changes, the most important are volatilization, adsorption
and diffusion. Volatilization can be minimized by filling the sample collection
vessels, leaving no headspace. Any contact with air is prevented and distribution
of volatile compound between the surface of sample and the headspace is

56
prohibited. Losses by adsorption can be minimized by rapid preservation of the
sample or by eliminating its exposure to the atmosphere.
Diffusion of some organic compounds (e.g. phthlate esters or plasticizers)
may be controlled by collecting samples in glass containers and by using bottles
caps or liners that minimize this process.
Samples can contain organisms (e.g., bacteria or algae) that may consume a
number of substances required for their growth, e.g., nitrogen, phosphorus,
silicon and organic compounds leading to their losses and introducing to the
sample metabolic products of such processes. These changes are generally
minimized by pH and temperature control or by chemical addition. Extreme pH
conditions (low or high) and low temperature are effective for minimizing
degradation. The addition of some chemicals to the sample (e.g., mercuric
chloride and pentaphenol) kills the microorganisms and effectively preserves the
sample, but their toxicity creates environmental hazards.
Generally, in order to prevent significant changes in water and wastewater
samples between sampling and analysis, one of the following preservation
methods is recommended: cooling to a temperature of 2-5°C; acidification to pH
< 2; addition of sodium hydroxide to pH > 11; addition of chloroform or
formaldehyde. The applicability of each of these methods of sample preservation
for determination of inorganic constituents and physicochemical parameters is
given in Table 2.4 [58].
Sampling and storage conditions and preservation methods of water samples
for determination of selected analytes and parameters are tabulated in Table 2.5
[59,60]. It must be noted, however, that complete stability of every sample's
constituents or parameters can never be guaranteed and at best, chemical,
physical and biological processes affecting the sample can only be slowed down.

TABLE 2.4
Applicability of preservation methods for aqueous samples [58]
Preservation method Determined substances (parameters)
Cooling to 2-5°C Organic, total, Kjeldahl and ammonium nitrogen, free and ionized
ammonia, nitrates, nitrites, colour, bromides and bromine
compounds, BOD, phosphates, iodides, acidity and alkalinity,
conductivity, sulphates, cationic surfactants, smell
Acidification to H2SO4 Ammonium and Kjeldahl nitrogen, free and
pH < 2 ionized ammonia, nitrates, COD, permanganate
index, dissolved silicates, silica
HNO3 Total hardness, metals
Addition of NaOH to Cyanides, iodides,
pH >11 sulphides
Addition of chemicals Chloroform Nitrates, nitrites, suspended matter
Formaldehyde Non-ionic surfactants
Other Phenol index (CuSO4 addition), sulphides (addition of zinc or
cadmium acetate)

57
TABLE 2.5
Sample preservation methods for some determinants of water [59,60]
Substance or parameter Sample Preservation Min. Preser- Max.
con- sample vol. vation holding
tainer (ml) temp. (C) time
_

Alkalinity-acidity P,GB 4 24 h
Ammonia P,GB H2SO4 (pH<2) 200 48 h
Arsenic P,G HCI (pH<2) 500 2 months
Nitrogen (Kjeldahl's PG H2SO4 (pH<2) 48 h
method)
Organic carbon G H2SO 4 (pH<2) 24 h
Chlorides P,G 100 15 days
Conductivity P,G In situ measure- 200 48 h
ment recommended
Cyanides GB NaOH (pH<12) 100 48 h
Biochemical oxygen P,G 1000 24 h
demand
Chemical oxygen P,G H2SO 4(pH<2) 100 24 h
demand
Hardness P,G HNO, (pH<2) 100 1 month
Fluorides P 300 7 days
Hydrocarbons G CCI4 (10 ml) 300 7 days
Polynuclear aromatic G C6 H1 4 (10 ml) 100( 46 days
hydrocarbons
Oils and fats G HCl (pH<2) 100( 4 15 days
Lithium G 15 days
Suspended matter Al, P,G HNO, (pH< 1.5) 100( 4 6h
Ag, Cd, Cr, Mn, Pb, Zn 2 month
Mercury GB HNO, (pH<5) 1 month
Nitrates P,G 100 4 48 h
Nitrites P,G 100 4 24 h
Colour, smell, taste G 500 4 24 h
Oxygen, dissolved GB In situ measure- 300 4 24h
ment recommended
Pesticides G 200( 4 7 days
pH GB 4 24 h
Phenols GB CuSO4 (1 g/ml) + H3PO4 500 4 7 days
(pH 4)
Phosphates P,G 100 4 48 h
Radioactivity P 100( species-
dependen
t
Dry residue P,G 500 4 7 days
Ether extract G CHC1I (1 ml/l) or HgC12 4 48 h
(40 mg/l)
Silica P 4 7 days
Sulphates P 4 6 days

58
TABLE 2.5 (continuation)

Substance or parameter Sample Preservation Min. Preser- Max.

con- sample vol. vation holding
tainer (ml) temp. (°C) time
Detergents Gs CHCI 3 (1 ml/l) 200 4 24 h
Turbidity G 100 4 24 h
P = polyethylene, G = glass, GB = borosilicate glass.

REFERENCES

1 J. Namiesnik, J. Lukasiak and Z. Jamrugiewicz, Pobieraniepr6bek srodowiskowych.

PWN, Warsaw, 1995 (in Polish).
2 L.H. Keith (Ed.), Principlesof Environmental Sampling. ACS, Washington, 1996.
3 M.I. Beach, Environ. Protect.Eng., 14 (1988) 117.
4 M. Liess and R. Schulz, Sampling methods in surface waters. In: L.M.L. Nollet (Ed.),
Handbook of Water Analysis. Marcel Dekker, New York, 2000.
5 H.H. Rump and H. Krist, Laboratory Manual for the Examination of Water, Waste
Water and Soil. VCH Verlagsgeselschaft GmbH, Weinheim, 1988.
6 H.G. Stefan, T.R. Johnson, H.L and McCounel, J. Environ. Eng., 118 (1992) 209.
7 J.A. Colman and R.F. Breault, Can. J. Fish. Aquat. Sci., 57 (2000) 1073.
8 D. Rossi, R. Baudo and H. Muntau, Aqua Aria, 4 (1992) 309.
9 A.J. Dobbs and D.T.E. Hunt, Sampling and analysis of water to assess exposure. In:
R.G. Tardiff and B. Goldstein (Eds.), Methods for Assessing Exposure of Human and
Non-Human Biota. Wiley, New York, 1991.
10 M. Ettala and K. Wingrist, Aqua Fennica,21 (1991) 175.
11 W. Hermanowicz, Chemia Sanitarna.Arkady, Warsaw, 1984 (in Polish).
12 W. Hermanowicz, W Dozanska, J. Dojlido and B. Koziorowski, Fizykochemiczne
badaniawody i sciek6w. Arkady, Warsaw, 1976 (in Polish).
13 W.F. Fresenius, K.E. Quentin and W. Schneider (Eds.), Water Analysis. Springer
Verlag, Berlin, 1988.
14 M.I. Abdullah and R.H. Tennant, FreshwaterBiol., 6 (1976) 201.
15 B.W. De Lappe, R.W. Risebourgh, A.M. Springer, T.T. Schmidt, J.C. Shropshire, E.F.
Latterman and J.R.Payne, The sampling and measurement of hydrocarbons in
natural waters. In: B.K. Afgan and D. Mackay (Eds.), HydrocarbonsandHalogenated
Hydrocarbons in the Aquatic Environment. Plenum, New York, 1980.
16 K. Satake, Hydrobiologia, 232 (1992) 149.
17 D.G. Tigwell. D.J. Schaffer and L. Landon, Anal. Chem., 53 (1981) 1199.
18 E. Andrulewicz and D. G16d, Studia i Materialy Oceanologiczne, no 57 (1990) 25 (in
Polish).
19 IOC/WMO, Guide to Operational Procedures for IGOSS Pilot Project on Marine
Pollution (Peroleum) Monitoring, IOC Manuals and Guides, no 7, 1976.
20 IOC/WMO/UNEP Pilot Project on Monitoring Background Levels of Selected
Pollutants in Open Ocean Waters, IOC Technical Series, no 22, 1982.
21 B.J. Harland, F.J. Whitby and M.H.I. Comber, Int. J. Environ. Anal. Chem., 20
(1985) 295
22 J. Siepak, Fizykochemiczna analiza w6d i grunt6w, Uniwersytet im. Adama
Mickiewicza, Poznan, 1992 (in Polish).
23 R.G. Nash and A.R. Leslie, ACS Symp. Ser., 465 (1991) 222.
24 H.B. Pionke and J.B.Urban, Ground Water Monit. Rev., 7, (1987) 79.

59
25 J. Unwim and V. Maltby, ASTM STP, 963 (1988) 240.
26 J.H. Ficken, ASTM STP, 963 (1988) 253.
27 W. Dyck, J.C. Pelchat and G.A. Meilleur, Equipment and Procedures for the
Collection and Determination of Dissolved Gases in Natural Waters, Ecological
Survey of Canada, Paper 75-34, 1976.
28 B.A. Torstensson and A.M. Petsonk, ASTM STP, 963 (1988) 274.
29 M.E. Rosen, J.F. Pankow, J. Gibs and T.E. Imbrigiotta, Ground Water Monit. Rev.
(GWMR), 126, Winter 1992.
30 J.F. Pankow, L.M. Isabelle, J.P. Hewetson and J.A. Cherry, Ground Water, 22 (1984)
330.
31 A. De Schrijver, G.Van Hoydonck, L. Nachtergale, L.De Keesmaeker, S. Mussche and
N. Lust, Water Air Soil Pollut., 122 (2000) 77.
32 J.A. Smith, H.J. Cho, P.R. Jaffe, C.L. MacLeod and S.A. Koehnlein, J. Environ.
Qual., 21 (1992) 264.
33 S.M. Chernyak, C.P. Rice and L.L. McConnell, Marine Pollut. Bull., 32 (1996) 410.
34 K.W. Lin and R.M. Dickhut, Environ. Sci. Technol., 31 (1997) 2777.
35 J.M. Southwood, D.C.G. Muir and D. Mackay, Chemosphere, 38 (1999) 121.
36 D.T.Waite, A.J. Cessna, R. Grover and E.J. Woodsworth, J. Environ. Qual., 29 (2000)
901.
37 S.J. Pogorzelski, Rev. Sci. Instrum., 63 (1992) 3773.
38 IOC/GEMSI Operational Manual for Sampling the Sea Surface Microlayers in the
Marine Pollution Monitoring Progamme for Petroleum (MARPOLMON-P), 1983.
39 W.M.J. Strachan and H. Huneault, Environ. Sci. Technol., 18 (1984) 127.
40 W.M.J. Strachan and H. Huneault, J. Great Lakes Res., 5 (1979) 61.
41 J.F. Pankow, L.M. Isabelle and W.E. Asher, Environ. Sci. Technol., 18 (1984) 310.
42 Polska Norma, PN-74/C-04621. Woda i scieki. Pobieranie pr6bek pary wodnej do
analizy fizycznej i chemicznej (in Polish).
43 E. Sierakowski and J. Mrozek, Kontrola chemiczna obieg6w wodnych i wodno-
parowych w elektrowniach. WNT, Warsaw, 1974 (in Polish).
44 J. Dojlido, Instrumentalne metody badan wody i sciek6w. Arkady, Warsaw, 1980 (in
Polish).
45 D. Hirst, J. Hydrol., 134 (1992) 95.
46 L.V. Parker, Ground WaterMonit. Remed., Spring (1994) 130.
47 T.E. Imbrigiotta, J. Gibs, T.V. Fussillo, G.R. Kish and J.J. Hochreiter, ASTM STP,
963 (1988) 258.
48 J. Turcotte, J.E. Cot, E. Fraser, H. Oullette and S. Langlois, Poll. Atmos., 33 (1991)
192.
49 M. Nieminen, Boreal Env. Res., 5 (2000) 133.
50 J.D. Petty, C.E. Orazio, J.N. Huckins, R.W. Gale, J.A. Lebo, J.C. Meadows, K.R.
Echols and W.L. Cranor, J. Chromatogr.A, 879 (2000) 83.
51 J.N. Huckins, M.W. Tuburgen and G.H. Manuweera, Chemosphere, 33 (1990) 533.
52 A. Kot, B. Zabiegala, J. Namiesnik, Trends Anal. Chem., 19 (2000) 446.
53 D. Barcelo, W.A. House, E.A. Maeir and B. Griepink B., Int. J. Environ. Anal. Chem.,.
57 (1994) 237.
54 K.R. Bull, K.H. Lakhani and A.P. Rowland, Chem. Ecol., 9 (1994) 47.
55 J.T Creed, T.D. Martin and M. Sivaganesan, JAWWA, 87 (1994) 104.
56 M.J. Mogollon, A.J. Raminez and C. Bifano, Chem. Geol., 121 (1995) 263.
57 J. Namiesnik, Z. Jamr6giewicz, M. Pilarczyk, L. Thorez, Przygotowanie pr6bek
srodowiskowych. WNT, Warsaw, 2000 (in Polish).
58 J. Namiesnik and Z. Jamrugiewicz (Eds.), Fizykochemiczne metody kontroli
zanieczyszczen. WNT, Warsaw, 1998 (in Polish).
59 R. Jeannot, Int. J. Environ. Anal. Chem., 57 (1994) 231.
60 A. Kot-Wasik, M. Morawska and J. Namiesnik, Chem. Inz. Ekol, 8 (2001) 177 (in
Polish).

60
Chapter3

Sampling selected solid materials

Bogdan Zygmunt and Jacek Namiesnik

3.1 INTRODUCTION

Solid samples are characterised by even larger variability than liquid samples.
Typical solid materials sampled for analysis include: raw materials for indus-
try/manufacturing, soils, sediments, snow and ice, wastes (industrial, municipal,
hazardous, etc.), biological plant material, electrofilter dusts, road dusts, tissues
and organs of animals, etc.
Analysis can give reliable results only if a sample has the same quantitative
and qualitative composition as the material studied [1-4]. Even the most
accurate analysis of the test sample (Section 3.8 Glossary) is useless if primary
samples are not correctly collected and a gross sample is not representative of
the material studied. The same importance should be given to sampling as is
usually given to subsequent steps of an analytical process. Also, a quality
assurance program is required for this step to ensure that information derived
from analytical data is correct. These sampling problems are discussed in a
number of books and papers [1,3-6].
Generally, analytical protocols contain detailed description of this step of
analysis. Sample collection procedures and systems are also included in techni-
cal standards. If this is not the case general procedures for collecting a primary
sample and producing a representative laboratory sample should be applied.
General discussion of sampling and nomenclature can be found elsewhere
[7-11]. The brief list of definitions of basic terms as used in this chapter is given
in a short glossary (Section 3.8).
Some sampling problems are universal, and general principles for sampling
bulk materials can often be directly applied to numerous matrices. A compre-
hensive description of sample collection at industrial wastes dumps, for example,
has been given by Rasemann and Markert [12].
The aim of sampling is to obtain a small and representative portion of the
population studied. If the samples do not reflect the properties of interest of the
population under study then they are of little or no use. Therefore planning for
informative sampling should be an integral part of any study.
Sampling is a part of the overall analytical process, and before formulating a
sampling procedure it is necessary to define: objectives of sampling, analytes of
ComprehensiveAnalytical Chemistry XXXVIr
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 61
interest, analytical methods to be used, way of reporting the results, and way of
using data. When preparing the sampling plan the following questions should be
answered:
- What kinds of samples are needed?
- When, where and how should samples be collected?
- What containers should be used to collect samples?
- How should the samples be preserved?
- What preparations are necessary for sample analysis?
- What precision of sampling is required?
Each sampling protocol should specify the number and size of samples. They
depend on many factors, including lot size, material granulation, analytical
method to be used, variability within and between particles, concentration of the
component of interest, etc. The related problems should be discussed individu-
ally for each sampling task.
If the product to be sampled is in packages, a specified number should be
randomly selected from the lot, i.e. from the total amount of material which is to
be characterised. One of the factors determining the number of sampled pack-
ages is the number in the lot; generally, if it is less than five, samples should be
taken from each package in the lot.
Statistics were used to answer questions concerning the number and size of
the samples to be collected. Although most work was aimed at solving specific
sampling problems some treatises deal with more unified approaches [13,14]. In
the widely known Ingamells approach describing non-segregated objects the
mass of a sample is calculated from relation (3.1) [15]:
WR 2 = K (3.1)
where W is the mass which should be taken, R is a sampling relative standard
deviation, expressed in percent, found for the material experimentally, and K. is
the sampling constant, corresponding to the mass of sample required to limit the
sampling uncertainty of 1% with 68% confidence.
The number of samples necessary to achieve a given level of confidence can
be estimated from relation (3.2) given by Kratochvil and Taylor [7]:
2 2 2 2
n = t S /iR X (3.2)
2
where t is the Student t-table value for the level of confidence required, s (the
standard deviation of the individual samples) and x (the average of analytical
measurements) are estimated from preliminary measurements or from previous
knowledge of the bulk material, and R is the percent standard deviation in the
average.
For segregated populations Eq. (3.3) was proposed by Wisman [16]:
s 2 = A/wn + B/n (3.3)
where s2 is the variance of the average of n samples using a total mass w of
sample, and A and B are constants for a given bulk material.

62
 As pointed out by Benedetti-Pichler [17] random sampling errors can also
occur in well-mixed particulate mixtures if particles differ considerably in
composition and a sample is relatively small. The relationship between the
allowable error and the number of particles to be counted is as follows [17,18]:
n = (1-p)/Gr
- (3.4)

where p is the fraction of one kind of particles and (1 - p) is the fraction of the
other kind of particles.
The above approach is also extended to describe the bulk material composed
of two kinds of particles differing in content of a component of interest-a higher
percentage is P1 and a lesser content is P2 giving an average percentage of P; the
corresponding densities are d, and d2 giving an average density of d. In this case
Eq. (3.5) can be applied [17-19]:

n = p(l -p)(dldJd2) 2[(P1 -P2)l/P]2 (3.5)

The degree of heterogeneity as measured by P1 - P2 has significant influence on

the number of particles required for a given uncertainty.
In the case of soil sampling, Crepin [20] calculates the number of samples (n)
necessary to obtain a variability accepted for a given task, from Eq. (3.6):
n = t2 s2 /D2 (3.6)
where t is the student table value for an intended probability level and an
assumed number of the degrees of freedom; s2 is the variance which is known
beforehand from other studies or estimated by s2 = (R/4)2 , where R is the
estimated range likely to be encountered in sampling; and D is the variability of
mean estimation we are willing to accept.
If the assumed degrees of freedom differ significantly from the values
obtained a reiteration must be done with the new n value for selecting a t value.
Care should be taken not to change sample composition during sampling,
transport, and storage of samples. The sampled package exterior should be clean
so that product contamination during package opening and sampling product is
avoided. Sampling during rain, strong wind, freezing, etc. should be avoided.
Sampling equipment should allow collecting a sample from all parts of the
material. The sampling equipment as well as containers for samples should be
chemically resistant to the material sampled. Also, there are other factors which
should be taken into account; these should be specified in a sampling protocol.

3.2 PROPERTY VARIABILITY AND ITS EFFECT ON SAMPLING

According to statistical definition, variability means that there are differences in

particular parts of the same population. In the case of sampling bulk materials,
variability of a given property in a lot can be quite significant; in other words,
there may be large differences in a given property among lot parts or segments.

63
 This property can be the content of a selected component, moistness as well as
particle size.
Let us consider lumpy material, e.g. raw material composed of particles of
different size. Lumps can differ in chemical composition and density as well as
size. When such material is unloaded into a pile, segregation occurs; a majority of
the larger lumps gather at the outer bottom part of a cone formed during the
process. Similar segregation takes place during unloading bulk material onto
rail cars; additional segregation occurs during transport. Fine particles tend to
move to the bottom while larger particles remain in the upper layer. If material
piled on the ground or in a rail car is soaked with precipitation, the moisture
content will not be the same throughout but generally will decrease with depth.
The opposite situation can also happen; moist material subjected to sunlight
loses water from outer layers faster, resulting in moisture content increases with
distance from the surface. Such variability can be observed on a conveyor belt
supplying material from a pile or rail car. Transported bulk material undergoes
partial re-mixing during unloading and reloading; however, the effect is
generally not sufficient to obtain the previous state of homogeneity. Often
additional segregation occurs during transport on belt conveyors due to belt
vibrations.
From the above considerations it is clear that variability in a given property
can differ greatly from case to case. This must be strongly stressed since the
nature of the variability can have a significant influence on the sampling
procedure. The difficulty arises mainly from the fact that little is generally
known about the nature and extent of variability before sampling. Usually,
because of time and cost restraints, it is also difficult to conduct corresponding
studies to acquire this knowledge. In such situations experience and knowledge
of the behaviour of materials having similar characteristics can be very helpful.
Furthermore, a thorough examination of a product lot with respect to physical
characteristics, transport history (including extent of loading and reloading),
etc. should be conducted, especially for materials from new suppliers.
The degree of variability or difference in properties among segments of the
same lot shows whether the material homogeneity is fit for the purpose or not. If
differences in a given property among different lot parts are not significant for
the purpose, material can be regarded as homogeneous for that purpose. The
definition does not strictly define the size of the parts. It also allows for
variability within as well as between parts. In other words, if a given part were
divided into a number of smaller portions, the difference in a property among
them can be significant and particular parts could be heterogeneous. If, how-
ever, average values of a property for the segments differed to a statistically
insignificant degree, then the material lot with respect to these segments could
be regarded as homogeneous. The practical importance of this problem for
sampling will be exemplified below.
Let us consider a certain lot of material, composed of a large number of pack-
ages from which several tens were selected for sampling. Measurements on pri-

64
mary samples collected from different points within individual packages showed
significant variations in values of a selected property. The material in these
packages was not homogeneous in the component measured. However, the aver-
age values between packages were not significantly different, so the lot could be
regarded as homogeneous in this component with respect to the packages. A
practical conclusion from this observation is that the number of packages
selected for analysis can be decreased provided that a sufficient number of pri-
mary samples are taken from each package selected for analysis.
A similar situation can happen for a bulk material lot transported in rail
cars. If primary samples from particular cars differ considerably in composition
then the product in them is heterogeneous, and the number of primary samples
collected from individual cars should be increased. If, however, average values of
a given property for different cars differ negligibly, then analytical costs can be
reduced because the number of cars subjected to analysis can be decreased. This
assumes the precision and accuracy of analytical data meet required levels.
A different situation can occur where primary samples from within particu-
lar packages do not differ significantly, but average values between packages are
significantly different. Then the number of samples from a package can be
reduced but the number of packages subjected to analysis should be increased.
If a material is homogeneous throughout the whole lot then collecting a
portion of material from any place in the lot will give a representative sample.
However, assuming that material is fully homogeneous without properly
planned investigations can be risky and expensive. Therefore, even in the case of
liquid materials, well averaged by thorough mixing, more than one primary
sample should be taken.
The effect of variability of the properties studied on sampling is not limited
to the problems described above. Material heterogeneity can manifest itself in
many different ways. Several heterogeneity varieties are recognized [21]: ran-
dom, directional (e.g. axial) and periodic. Heterogeneity is random if a property
value in one portion of the population studied is independent of the values in
neighbouring portions and independent of the position in the whole bulk of the
material. For example, a given lot of material in drums from the same plant but
different production lines and/or different periods may give a final product
which differs in content of a given component. If the products were not properly
mixed before packaging and, due to multiple reloading, packages were mixed up,
it is not possible to tell the characteristics of the product in a given drum even if
the content of a given component is known in all neighbouring drums.
A similar situation can arise in the case of lumpy materials, though the
additional heterogeneity comes from segregation with respect to particle size.
For instance, raw material can come from the same mine but from different
beds, or it can pass through different enrichment equipment; this can lead to
heterogeneity. Although it can be mixed during reloading some heterogeneity
can remain; it will only be random heterogeneity if the material does not
undergo segregation due to difference in particle size.

65
 In the case of random heterogeneity, the location of a sampling point is not
very important; however, the number of primary samples taken is crucial, and
should increase as the difference between the particular samples increases.
Problems with heterogeneity become more complex when property variation
with direction and time needs to be taken into account. If systematic heterogene-
ity occurs, a property value for a given material segment can be obtained from
values in the neighbouring segments. Let us consider again a drum with granu-
lar material of different particle sizes. Once the drum content is thoroughly
mixed, averaging of properties can be sufficient for heterogeneity to be random.
In practice, however, particle-size segregation can occur due to reloading and
transport. Then samples collected from different layers will show axial variabil-
ity, i.e. a property value (e.g. content of small particles) can linearly increase in
bottom direction. Such distribution of a property requires that a given sample is
collected along the whole drum height or samples are taken from several layers
along the height.
On a conveyor belt, vertical and horizontal segregation can occur; this makes
it necessary to take samples from the whole cross-section of the material.
Directional variability of material in a storage place or on a rail car is not as
simple as for the cases described above. Property variation can be multi-
directional, often dependant on how the material was reloaded. Such variability
requires careful selection of sampling points and a larger number of primary
samples. The idea is to collect samples which reflect the distribution of a
property in the whole bulk of the material.
Figure 3.1 shows the distribution of a fine particle fraction (< 10 mesh) in the
cone of coal containing 20% of the above fraction when a chute pipe in a vertical
position is raised slowly to minimise segregation (a) and when a chute pipe is in
an oblique position (b) [21].
Figure 3.2 shows the principle of sampling points location when particle-size
segregation occurs [21]. This is a scheme proposed by the Association of Official
Analytical Chemists for material on, e.g., a tipper trailer or rail car. According to
this scheme, a material lot should be divided into ten parts and samples collected
from each. The total of these samples gives a gross sample which is to be
representative of the material studied. Directional heterogeneity can also be
found in liquids containing suspensions.
Periodic heterogeneity should also be discussed: this occurs in flow manufac-
turing processes and can result from periodicity in feeding raw materials or
a) b)

Fig. \ Distribution
23.1.of fine13,
Fig. 3.1. Distribution of fine 29,6 '27, ' 28,0'
particle fraction (<10 mesh) in / , ,
a cone of coal containing 20% of 8 20 9 35,8/ \14.4\
this fraction (based on Ref.
[21]). vertical shute side shute

66
 7 8

3 1 2 1 r,

6 1_
Fig. 3.2. Principles of sampling points location when 10 9
particle-size segregation occurs (based on Ref. [21]).

intermediate products, or in periodic changes in the process itself. In materials

manufactured in such a way, property values can fluctuate periodically around
average values. Periods and range of fluctuations can be more or less regular. In
regular fluctuations significant systematic errors can happen if samples are reg-
ularly collected and follow fluctuations of the property studied. The likelihood of
coincidence is rather small but feasible. Whenever periodic changes in the mate-
rial produced are suspected, periodicity should be determined and sampling fre-
quency corrected.
A sort of systematic variability can also occur when a given material lot
comes from, e.g., two plants. Products of particular plants can be homogeneous
within a given plant but can show between-plant differences. The material in a
certain number of bags may have the same values of a property but differ from
others in the lot. Depending on whether the origin of the bags can be recognized
or not, the sampling program may treat the products from individual plants
separately or as a whole.

3.3 SAMPLERS FOR GREASY, DOUGHY AND FUSIBLE MATERIALS

3.3.1 Greasy and doughy materials

The sampling equipment for such materials depends on the sample to be

collected. For sampling the upper layer of the material in a container a metal or
ceramic spatula can be used. If a commodity is homogeneous throughout the
container, a spatula can also be used for collection of primary samples to obtain a
representative gross sample.
The upper layer of thickness, depending on size of a container, is removed
and then equal portions of material are collected with a spatula and placed in a
clean container. A stirring rod or a smaller spatula can be helpful in removing
samples from the sampling spatula. Due to possible heterogeneity, primary
samples for preparing a representative sample should be collected from the
whole thickness of the material.
Figure 3.3 [22] presents samplers for greasy and doughy materials. Before
sampling the material an upper layer (e.g., 1-5 cm) is removed from the place the
sampler will be pushed into. The sampler in Fig. 3.3a is made of two halves of a
pipe cut axially and joined with hinges. To the upper ends of both halves handles
are attached which are used to push the sampler into the material and to close it.

67
 a) A-A b)

S2

08

c)

I
1
Fig. 3.3. Samplers for greasy and doughy, and fusible materials (see text for description) (based
on Ref. [22]).

The other ends of the tube halves are tapered. The open sampler is driven down
to a required depth and closed by turning the handles 180° . After withdrawal the
sample is transferred to a container with a rod or spatula.
For more doughy materials the sampler shown in Fig. 3.3b may be used. Its
principle of operation is similar to a gimlet bit: after removal of the upper layer
the sampler is driven into the material by rotating it. Again, the sample collected
is removed from the sampler with a small spatula.

3.3.2 Fusible materials

If a product can be heated to the temperature at which it melts then sampling

equipment for liquids can be applied for sampling this product. Samples of
solidified materials, depending on hardness, can be collected with a knife,
sampler (Fig. 3.3c), axe, and chisel and hammer.
A strong metal knife with a blade ca 3 cm wide should have a handle made of
heat-insulating material since it is often heated before use. Sampling is based on
cutting a piece of material with a hot knife, most often in the form of a cone that
makes sampling convenient from any place in a package. Samples, possibly of the
same or similar size, are placed either in separate containers for individual
analysis or together for preparing a gross sample.
The use of the sampler presented in Fig. 3.3c requires a relatively high
pressure on it when sampling harder materials. The sampler is rotated in the
direction of winding, while uniform pressure is exerted on the sampler. Brittle
material samples will be in the form of chips and smaller lumps which should be
gathered as the sampler is pulled out of the material. Any adhesive or sticky

68
material remaining in the winding should be removed, e.g., with a spatula or
stirring rod and included in the sample.
For sampling brittle and easily breakable materials, tools such as axes or
chisels can also be used; these tools are not convenient for sampling materials
that adhere to them. Before chipping off pieces of material for sampling pur-
poses, the surface layer-which can have changed in composition-is removed.
The pieces of material from different spaces of the package or lot are used as
primary samples or to prepare a gross sample.

3.4 SAMPLERS FOR LOOSE AND LUMPY MATERIAL

The main factors determining selection of sampling equipment for loose and
lumpy solids are granulation, material quantity and whether it is moving or
stationary during sampling. As in the case of other sampled objects, samplers
must be made of a material which does not interact with sample components.
The simplest tools for sampling loose and lumpy material are shovels whose
dimensions depend on particle size and primary sample size. Under ordinary
conditions samples from the outer layer or from small depth can be collected
with these samplers. Therefore, they can be used when material is homogeneous
or of random variability.
Shovels are also used for collecting samples from belt conveyors of small
capacity when the material particle size is not larger than 50 mm. Care should be
taken when collecting samples from the whole cross-section of material to move
the shovel at a constant speed. The shovel should be 2.5-3 times wider than the
largest lumps. It should be noted, however, that shovel sampling directly from a
moving conveyor belt is prohibited.
The most common samplers for fine-grained (<20 mm), granulated and
powdered materials are pipe samplers of different design; they can be either
single-tube or double-tube samplers.
The sampler presented in Fig. 3.4a [22] is made of a metal tube bevelled at
45° at one end; the tube edges are slightly rounded off. At the other end is a
handle closing the tube. Since the sampler has an open end, it should be pulled
out of the material in a horizontal position or at an angle smaller than the
natural slip angle of the material. The sampler length should allow samples to be
taken along the whole diagonal of a container. The tube diameter should be at
least 2.5 times larger than the maximum particle size but never smaller than 20
mm. When the sampler is open at both ends one can collect samples in a
continuous way by allowing the material to flow from the package to a sample
container; this, however, gives unrepresentative samples if the material is
heterogeneous. Moreover, sampling must be done in a horizontal position or at a
rather small angle; an additional disadvantage is that some of the sample is lost
when pulling the sampler from the material. A certain degree of segregation of
the material of different granulation can also occur when sliding the sampler
into the material.

69
 a) b) a)
a)~ b,)~ la).,

I I

A A - ! '

X~~~~~~~~~~~~~~~~~~~~~~~j
__1
o 0
I X I I

il
I' ,,, I
,1 8 I

III I o
1,5 At
iI
I

Left: Fig. 3.4. Samplers for loose materials (see text for description) (based on Ref. [221).
Right: Fig. 3.5. Samplers for loose materials (see text for description) (based on Ref. [22]).

The sampler shown in Fig. 3.4b [22] differs from that shown in Fig. 3.4a in
that one side of the tube is cut out leaving a U shape. The width of the cut-out
opening should not be less than 2.5 times the maximum particle size. The
sampler should be pushed horizontally into the material with the opening
downward, three half-turns should be made, and it should be tapped slightly,
then pulled out with the opening upwards. Particularly in the case of powdered
and very loose materials, the package should be positioned in such a way that the
sampler can be withdrawn horizontally, otherwise some of the sample is lost.
Another type of sampler is presented in Fig. 3.5a [22]. It is T-shaped and
made of tubing terminated at one end in a rounded cone; the other end is open
and has a handle fastened. A rod or tube handle is welded on or fastened in
another way to the outer side of the tube. Along the whole tube length, an
opening is cut out, whose width should not be smaller than half the outer
diameter of the tube. Similarly to the previous one, this sampler is applied in a
horizontal position or at a small angle despite having a closed tip. If used in a
vertical position it may collect more material from the upper layers and part of
the sample may be lost when pulling the sampler out of the material, mainly due
to friction with the material in the package. The disadvantage of this sampler is
that the lowest layer of the material (between the cone tip and the opening) is
not sampled. To minimize this effect, samplers with a cone of very small height
are constructed. Sampling is done in the same way as in the case of the sampler
presented in Fig. 3.4a. Care should be taken during emptying and cleaning the
sampler-in the process the sampler is gently tapped; the sample part which

70
 a) b)

closed open

Fig. 3.6. Samplers for loose materials (see text for description) (based on Ref. [22]).

remains is taken out with a spoon. The sampler can also be applied in such a way
that a fitted metal or wooden rod is slid into the tube before pushing it into the
material. Filling proceeds when the rod is slowly withdrawn while the sampler is
gently rotated.
Other samplers for loose materials are presented in Figs. 3.5b and 3.6 [22].
They consist of two tubes, one inside the other, which are well fitted but can
move with respect to each other. The lower end of the outer tube is terminated
with a slightly rounded cone to protect the package from puncturing. The end of
the inner tube, which extends up to the cone base is closed. Rod or tube handles
are fastened perpendicularly to the upper ends of the tubes. The inner tube
handle does not go across the inner tube bore; this makes emptying and cleaning
easier. Along the length of the outer tube, from below the handle to the cone
base, a longitudinal opening is cut. The width of the opening is generally equal to
half the diameter of the tube; other sizes are allowed, but they should be no less
than 2.5 times maximum particle size. Often, several short longitudinal open-
ings along the same line are made at small distances from each other (Fig. 3.5b).
The inner tube also has one or more openings identical to the outer ones with
respect to dimensions and arrangement. The adjacent edges of openings should
be cut at right angles to protect particles from getting between tubes. The
diameter of the sampler should be at least five times larger than the largest
particle and never smaller than 20 mm. On both tubes marks are made close to
the handles to show sampler position (closed or open).
Before use one should check that both tubes move easily with respect to each
other. The closed sampler is pushed into the material horizontally or at a small
angle with the outer tube openings directed upwards. The sampler is opened by
turning the inner tube to a position such that the openings of both tubes
correspond. To facilitate filling, the sampler is gently tapped or rotated in both
directions. Then the sampler is closed by turning the inner tube 180° , withdraw-
ing it and emptying it through the open upper tube end. Lastly, the outer tube is
pulled out and cleaned up if necessary. For materials with a tendency to enter
between the tubes and make tube rotating difficult, the inner tube can be

71
replaced by a metal or wooden rod of slightly smaller diameter and the sampler
used as a one-tube sampler (Fig. 3.5a). The advantage of this design over
non-closed ones is that the sample is not collected until the sampler is completely
immersed in the material. Also, no sample is lost during removal from the
material. These samplers have some disadvantages, but generally give more
representative samples than the previous ones. The main disadvantages are: if
used in a vertical position the proportion of material from the upper layers is
increased; particles can crumble during sampler closing; the conical end of the
sampler does not allow the lowest layer (from tip to the first opening) to be
sampled.
An improved design of pipe sampler is shown in Fig. 3.6b. The difference
from the previous designs is the division of the inner tube into segments with
lateral barriers. Each segment has a separate opening in both tubes. The
sampler is used as previously described but can be pushed into the material
vertically since segmentation results in more representative sampling along the
sampling axis. With care during emptying, separate samples from individual
layers can be obtained.
The sampler recommended for sampling fertilisers is shown in Fig. 3.6a. It is
constructed of two halves of fitted tubes, the outer one ending with a flattened
cone. The halves are joined by axles around which they can be turned. After
immersion in the material, it is closed by turning the outer half. During sam-
pling the material does not move at all or only negligibly; this is very important
since material movement results in segregation which leads to additional errors.

3.5 SAMPLING SOILS

Different definitions of soil are presented in the literature. Depending on the

aim of the analysis and the kind of soil, different approaches are used to
sampling procedure and design. Special care should be taken and an appropriate
approach applied to sample soil for preparation of soil matrix reference materials
[23]. Again, a proper sampling approach is needed. Substantial differences can
be observed in approaches to profile sampling and sampling different layers and
types of coverage. Sampling depends also on the kind of compounds to be
studied. Special care is required, for example, when volatile organic compounds
are to be determined [24,25]. Quite recently, comparative studies (15 European
participants) on soil sampling for three-dimensional pollution assessment were
conducted [26]. Participants used their own sampling procedures. The results
obtained show that this step of soil analysis needs harmonisation throughout
Europe.

3.5.1 Sampling soil profiles

Selecting the correct sampling place in the soil profile requires the following
initial operations: (1) digging out soil profile, i.e., uncovering the upper part of

72
the soil, at least one layer (usually three), if possible without disturbance, to a
depth of 1.5 m; (2) determination of the genetic classification unit of soil (type,
soil textural group); (3) detailed description of the profile (thickness of strata,
colour, structure, granulation, etc.).
Samples are collected from particular genetic levels or soil horizons,
sometimes also from geological strata. To determine granulation, apparent
density (ratio of mass of sample of undisturbed structure to sample volume),
hygroscopicity, acidity, carbonate content, chemical and mineralogical compo-
sition, samples can be collected without structure preservation. To determine air
and water conditions, i.e. capacity with respect to water (capillary, pore and total
water) and to air (porosity, pore size) as well as soil structure and density,
samples should be taken without natural structure disturbance.
When profile sampling of mineral soils, sampling places are selected after
careful cleaning of the profile wall; sampling starts from the lowest part of the
profile.
Samples without structurepreservation are cut out with a shovel, spatula or
knife from the middle part which is the most typical for a given layer. A sample
from the lowest part of the profile should represent its lower level down to a
depth of 2 m. The soil beyond the profile reach (below the level of 1.5 m) is
sampled with a special drill. The sample from the top organic layer is collected
from an undisturbed area nearby.
The sample mass of mineral soil should be ca 1.5 kg (ca 1 kg air dry soil of
granulation smaller than 1 mm) and of organic soil ca 0.3-0.5 kg. Samples of
gravelly and stony soils should be generally larger than 1.5 kg.
Samples from each genetic level are packed in linen or plastic bags. A plastic
or paper tag is inserted into each bag giving: area description (place, etc.), profile
number, characteristics of genetic level (symbol, thickness, colour, etc.) and
sampling date. The closed bags from a given profile should be tied together.
Samples of undisturbedstructure are obtained with metal or plastic cutting
cylinders of 100 ml capacity, closed at both ends with rubber caps. Before
sampling, numbered cylinders should be cleaned, dried and weighed.
After cleaning a profile wall and identifying particular layers, sampling
points are selected. Starting from the profile bottom, cylinders are driven into
the soil at each level at the point chosen. The cylinders filled with soil are taken
out with a shovel or knife, the soil surface is levelled out at both ends and the
cylinders capped. The cylinders are stood next to each other in a special
container to protect them from falling over during transport. Each sample is
described (as above) in a logbook.
Samples from organic soil layers (humus, peat, etc.) are also collected using
cutting cylinders, mainly of a volume of 250-500 ml.
The collection of undisturbed structure samples of gravelly and stony soils is
particularly difficult. In this case, large cylinders of a volume of up to 1000 ml are
used. If a soil contains large rocky pieces, then their content should be evaluated
and samples collected from spaces of fine earth fraction.

73
3.5.2 Sampling arable layers

Sampling procedures for arable layers assume them to be vertically homo-

geneous. To get a representative sample, horizontal variability resulting from
natural difference and different use should be taken into account. All areas of
arable land which are to be sampled should be drawn on a geological outline or
detailed map and designated with numbers.
In the case of arable land as well as durable grassland, primary samples are
collected from the surface layer of soil. These samples can be collected
throughout the growth stages of plants, from spring to late autumn. Sampling
should not be done directly after application of fertilisers and manure, severe
drought or high soil humidity. Samples can be taken along a zigzag line or
diagonally over the sampled area (Section 3.5.6).
The basic sampling device is a sampling stick as shown in Fig. 3.7 [27]. At
selected points, the sampling stick is pushed down vertically into the soil up to
the crossbar, turned 360 ° , and then pulled out of the soil. Soil from the stick is
removed with a scraper and inserted into a cardboard box. From arable land 15
to 20 primary samples are collected while from durable grassland 30 to 40. The
total sampled material constitutes a composite sample.
The composite sample should be representative of the whole arable land of
similar natural conditions (soil type and texture, topography) and agricultural
conditions (fore-crop, crop, soil fertilisation and manuring). It is assumed that
one composite sample should be prepared for an area of 4 ha, provided that the
area is uniform with respect to soil and close in topography. Composite samples
should be prepared separately for each crop culture. One composite sample is
allowed in the case of an area with different vegetation provided that the plants
have close agricultural and fertilisation requirements and that they are grown
by the same farmer. In field experiments the composite sample can represent
each plot or the whole area under study. The choice depends on the information
sought and the climatic and soil conditions.
Samples of cropland should be placed in one cardboard box, and those of
grassland in two boxes. The boxes with samples should be delivered for
examination within 14 days.
For preparation of the laboratory sample, the composite sample is poured
out on a plastic tray and thoroughly mixed; larger soil pieces are broken up and
visible plant parts and stones removed. The mixed soil is placed in cardboard

1
2

Fig. 3.7. Sampling stick. 1: Handle, 2: bladed arbor, 3: cross-bar (based on

Ref. [27]).

74
boxes and air-dried for approximately two weeks in a dry and well ventilated
space, with the boxes open (slight soil dusting can occur). For determination of
volatile organic compounds in soil, separate standard procedures for collecting
and storing samples are described in Ref. [29]. The samples are generally stored
in glass jars closed with PTFE caps at a temperature of 4°C.
For preparation of a test sample, depending on needs, part of the laboratory
sample is taken and ground in a soil mill with a 2 mm mesh sieve. Particle size
reduction can also be done in an agate or porcelain mortar followed by sieving
through a plastic or iron screen of 2 mm mesh. After sieving and discarding plant
parts, stones and gravel, the soil must be thoroughly mixed. For determination
of general forms of mineral components, part of the laboratory sample is
powdered to such a degree that 80% passes through a sieve of 0.25 mm mesh.
The very fine fraction (<0.25 mm) and that which remains on the screen are
accurately mixed and poured out into a plastic vessel.

3.5.3 Sampling root layer soils

A soil sample which is to characterise a selected area with respect to a given

property should reflect the different kinds of variability that influence the
property. In arable land the soil layer within the reach of plant roots is relatively
uniform. In forest soil and wherever cultivation is carried out not more than
once every several or several tens of years, specific genetic levels differing
considerably in chemical and physical properties are formed in the root layer. In
such soils the vertical variability of the root layer can be more profound than the
horizontal variability.
In the root layer of the forest soil three genetic levels generally occur: the
surface or organic level containing organic matter decayed to a different degree;
accumulation-eluvial or mineral-organic level; and weathering or mineral level.
The thickness and properties of particular levels depend on bed-rock, type of
vegetation and length of vegetation cycle of forest stand. For characterisation of
forest soil and also other soils in which plant roots grow through different
genetic levels, vertical variability (apart from the horizontal) must be taken into
account.
The root layer of forest soil-similar to the arable layer of cultivated soils-is
characterised by the composite sample prepared from primary samples. When
selecting the area for the composite sample, variability must be taken into
consideration resulting from the kind of parent rock (sand, clay, loam), the plant
species predominant in the forest stand (pine, birch, spruce, etc.) and stand age.
The size of the area selected for sampling also depends on the above-mentioned
factors: for a forest nursery it ranges from 0.2 to 0.5 ha, for forest crops it can
approach 10 ha, and for older forest stands on relatively uniform soils one
composite sample can characterise an area of 20-30 ha. Area size characterised
by a representative composite sample depends also on the information one wants
to get from the results of sample analysis. Each composite sample is prepared

75
from individual samples whose number depends on area size, e.g. for areas lower
than 1 ha 5-10 samples are sufficient, for 10-20 ha 20-30 primary samples are
necessary. Sampling sites should reflect the variability of the area studied as
much as possible. Most often samples are collected along a zigzag line (Section
3.5.6).
The most typical technique of sampling soil from the root layer is the
following. In selected sites a soil portion is dug out to a depth of 25-30 cm in a
single operation with a shovel. After removal, the soil is left on the shovel and
samples of 100-150 g are collected with a small shovel or knife from each clearly
distinguished genetic level starting from the lowest. Each successive level is
sampled after removing the previous one. Samples from each level are collected
in separate plastic buckets. Samples from the organic layer should be taken in
such a way that duffs (non-decayed organic matter) are not contained in them.
After collecting the required number of samples from each level, the material in
the buckets should be thoroughly mixed and a portion of ca 1 kg taken from each
bucket. For the area studied three composite samples are generally taken, each
characterising a selected genetic level. If a toxic substance is to be determined in
the root layer, composite samples from all levels are mixed together to give an
composite sample for the whole root layer. In some root layers only two genetic
levels are discerned; this is mainly the case for orchards and parks, where former
arable layers and subsoil are clearly marked. Composite samples of these two
layers are put into linen or plastic bags. Identification tags with additional
description (subsection, section, district, locality) of the area studied are inserted
into each bag. The bags are tied up and kept in dark and cool space.

3.5.4 Sampling soil from surface layers

It often happens that pollution of a soil surface must be determined, e.g.

pollution resulting from breakdown in an industrial plant, a traffic accident
during transport of toxic chemicals, etc. In such situations, generally only a
surface soil layer a few cm thick is analysed. The thickness to be considered
depends on many factors, including time of the pollutants impact on the soil,
nature of the pollutant, nature of soil, vegetation coverage, soil size-grading, etc.
For areas with thick turf, pollutants accumulate mainly on vegetation while they
may move deeply down into the soil for uncovered areas-for sandy soil deeper
than for clay or loam. Pollutant dispersion downwards also depends strongly on
precipitation. Therefore such factors as the area size to be studied, and number
and kind of samples for analysis (primary or composite), and depth and sampling
conditions should be settled on site by a specialist studying the problem.
The degree of area pollution is generally expressed in mass units (mg, g, kg,
etc.) per unit area. Therefore the soil should be sampled, if possible, from a
determined area (e.g. 5x5 cm) to a depth established earlier. If pollutants are
likely to go deeper, two samples should be collected; one from a layer to a depth of
2-5 cm, for instance, and the second 5-10 cm thick. The samples should be

76
handled, labelled and prepared in the ways discussed above.

3.5.5 Soil sample preparation for analysis

Samples should be delivered as quickly as possible to the laboratory (not later

than a few days after sampling) where they are transferred (especially from
plastic bags) into cardboard boxes; identification tags should be checked and
placed in the boxes. If boxes are not available, the bags with samples should be
opened. The open boxes are placed in a dry, well ventilated room. After soil
reaches air dry mass, clusters of mineral layer are ground in a mortar with a
rubber or wooden pestle, if needed, then the soil is sieved through a 1 mm mesh
sieve. From organic (arable) soil, roots, stones etc. are manually removed.
Samples prepared in such a way are stored in cardboard or plastic closed boxes.
Successive steps depend on the analysis conducted and are described in
analytical protocols.
The procedure is different when a fresh sample is to be analysed. In this case,
samples are immediately transported in tightly closed plastic bags to the labora-
tory where they are analysed within a short time. If the analysis must be post-
poned, the samples are stored at a temperature of ca 0-5°C. Samples with
undisturbed structure (in cylinders) should be analysed directly after delivery to
the laboratory.
In the description of sampling procedures the basic principles must be
observed. There are many modifications that are occasionally used when special
analytical tasks must be performed.
Sampling periods depend on the information one wants from the analytical
data. For instance, for evaluating soil properties that are independent of time
(particle size, apparent density, carbonate content, organic matter content, etc.),
samples can be collected at any time although climatic conditions (snow, freeze,
long-term precipitation, etc.) must be taken into account. If properties that
change with time (e.g. easily soluble components) are to be determined,
sampling should be carried out during their periods of stability (i.e., early spring,
late autumn). For determining dynamics of changes, e.g. biological activity,
samples should be collected several times during the vegetation period.

3.5.6 Sampling site selection

The average sample should represent the area of agricultural land of close
natural conditions (soil type, soil textural group, topographic features, etc.). For
similar areas with respect to soil, topographical features, etc., the sampling area
to prepare one composite sample should be ca 0.2-0.4 ha. Sampling sites should
be selected along a zigzag line or along a diagonal (Fig. 3.8) [29].
The approaches presented schematically in Fig. 3.9 are also proposed [30].
They include normal sampling, test sampling, diagonal sampling, and cross
sampling.

77
 a) along zigzag line b) along diagonal

iF\\ / /
\\\ / / / /

Fig. 3.8. Zigzag and diagonal arrangement of sampling points for soils (based on Ref. [29]).

a)

Fig. 3.9. Arrangement of soil sampling sites proposed by Rump and Kirst [30]. (a) Normal
sampling, (b) test sampling, (c) diagonal sampling, (d) cross sampling.

3.6 SAMPLING BIOLOGICAL PLANT MATERIALS [31]

Methods and dates of sampling biological plant materials depend on the aim of
the studies. Most often this material is sampled to determine: (a) crop quality
(pollution level included); nourishment needs; cycling of nutrients in the envi-
ronment; degree of use of soil nutrients on the basis of their accumulation in a
given plant species in a given period; (b) level of supply of nutrients to plants,
distribution of nutrients in plants; effect of different parameters (e.g. fertilisa-
tion, SO2 emission) on the basis of accumulation of these substances in different
parts of a plant at various phases of growth; and (c) direct plant pollution from
given substances accumulation in the whole plant or in some parts.
According to the intended aim of the studies, the sample collected can char-
acterise: a given plant (e.g. tree, bush)-the sample is collected from that
particular plant; a given area occupied by one or more plant species-individual
samples are collected from points representative of the area: the samples can be
analysed separately, however, they are most often combined to give a composite
sample.

3.6.1 Sampling

When sampling from shaded areas, e.g. in parks, orchards, etc., it is important to
take samples that represent the specific area under study (e.g. ecological niches).
When insolation is highly differentiated, separate samples of plants from shaded
and insolated areas should be collected. The sampling principles for plants in
growth phase and for mature plants are the same. The whole plant or only leaves
are gathered; full-grown leaves must be taken from the middle part of a stem;
growth phase should be noted. From leaves collected from single plants repre-

78
sentative of a given area, a composite sample is prepared for analysis. Leaves
should be gathered proportionally to the plant occurrence-each plant should be
properly represented in the composite sample. A similar procedure is applied for
sampling other plant parts (e.g. fruit, flowers). The samples are placed in linen
bags or in paper envelopes (never in plastic bags) and delivered to the laboratory
as soon as possible. The samples should be properly labelled; an identification
tag should be inserted in each bag; sampling site, plant species, plant part
collected and sampling date should be given.
To determine total content of particular components accumulated in a
period of growth by cereals and oil plants, samples of full-grown plants are
taken. If the assimilation rate of a given component, e.g. from a fertiliser or
emission, is to be determined, then samples are collected at different growth
phases. Independently of the aim of the studies, the area sampled should always
be characterised. Over an area of ca 1 ha 30-50 sites (more or less equally
distributed) are selected; at each site at least five plants are cut close to the
ground and placed in a linen bag. The total material collected in the field is
thoroughly mixed and a sample (0.3-0.5 kg) is taken for analysis. For full-grown
plants, seeds and straw are separated and stored for analysis as separate
samples.
Obtaining a representative composite sample of root crops is especially
difficult: plants of different characteristics should be gathered so that the total is
representative of the area studied. As a rule, several tens of individual samples
are collected by digging out individual plants. After separation of bulbs and
above-ground parts, they are packed into separate linen bags. After samples
have been collected from the whole area they are poured out (on paper or sand),
thoroughly mixed, bulbs and above-ground parts separately, and composite
samples are collected.
Primary samples of fodder plants are collected in a similar way to sampling
cereals. Plants from 30-50 sampling sites regularly distributed over an area of 1
ha are cut off at the ground over squares of 10 x 10 cm. After mixing the material
collected a composite sample of ca 0.3-0.5 kg is taken for analysis.
Depending on the aim of the studies, primary samples of wild plants are
taken either as collective samples containing all plant species cut off over a
so-called micro-area or as samples of particular species from given micro-areas.
The sampling procedure is the same as that described above. The number of
samples depends on the area studied, density of coverage, occurrence of
particular species, etc.
In the case of forests, mainly leaves or needles are subjected to analysis,
although sometimes shoots, bark, wood and ground cover are collected. The
principles of sampling ground cover are the same as for wild plants. Plants
higher in evolution are gathered during blooming, and bryophytes (mosses,
lichens) in autumn. Leaves are collected during growth periods, needles after a
period of growth, from the outer part of the middle of the tree head. Leaves from
taller trees are collected with special tools or by shooting twigs.

79
 The time of sampling leaves depends on weather conditions, phases of
growth, growth conditions, and other factors, e.g. the necessity for immediate
determination of pollution in a given region. Therefore, depending on the aim of
the studies and the climatic conditions, the date of sampling leaves should be
decided individually for each case.
The age of needles on a coniferous tree can range from a year to tens of years
depending on the tree's age; which needles are selected for analysis depends on
the aim of the study. Generally, older needles are analysed to determine the
long-term effect of anthropogenic impact on coniferous plants. If the aim of the
study is to determine the quantity of particular components assimilated during a
growing period, needles from the last growing period are collected. Needles are
always gathered during non-growing periods, i.e., November to February in the
northern temperate zone. Which part of the tree head is sampled also depends on
the aim of the analysis: to determine the pollution of needles, samples are
collected from the windward side, i.e. from the prevailing direction of pollutant
influx.
Due to variability in populations of particular tree species, the leafage of each
tree is treated as an individual sampling area; several twigs from each tree are
used to prepare the composite sample. To characterise a selected forest area,
samples are gathered from several trees representative of the area studied. Each
tree is analysed separately. If bark is subject to analysis it must come from the
same tree that the needles or leaves were gathered from. Bark is collected at a
height of ca 1 m above ground level; care should be taken not to damage the
phloem, which is living conducting tissue.
Samples taken from trees are placed in linen or paper bags. Each sample
must be described (sampling date, characteristics and tree number) and the
identification tag placed in the bag.
When sampling leaves and bark of trees and bushes in orchards or parks, the
procedures are as for forests; samples of herbaceous plants are gathered in the
same manner as plants growing wild. The sampling procedures for fruit depend
on the aim of the studies and should be developed separately for each case.

3.6.2 Preparing sampled material for analysis

In the laboratory material should be cleaned and, if needed, dried and reduced in
size. Roots and bulbs should be washed with running water to remove soil
completely and then rinsed with distilled water.
The method of preparation of above-ground parts is strictly related to the
aim of the studies. If a polluting substance taken in by plants is to be determined,
then above-ground parts must be washed before drying. The same amounts of
distilled water should be used for samples being compared. When the effect of
pollutants on plants is to be determined, then samples are not washed because
substances on the plant surface may affect conditions of plant growth, e.g. by
reducing the rate of photosynthesis.

80
 Regardless of whether they are washed or not, plants can be dried, first in air
in well ventilated rooms to constant mass and then in driers or directly in driers
with air circulation and temperature control. In the beginning the temperature
should be kept at 30°C; at the end it can be raised maximally to 60°C. Roots,
bulbs and thick stems should be cut to shorten drying time. Dried material is
milled to dust of uniform particle size. The samples obtained should be stored in
tightly closed glass jars for further study.

3.7 SAMPLING SEDIMENTS

Generally, sampling sediments requires much greater care than sampling other
objects. As always, the sample must resemble the original population as closely
as possible. The sampling device should be lowered slowly to avoid setting up
currents or causing shock waves that might disturb fresh sediments. Samplers
specially designed for the purpose are available and new designs are proposed in
the specialist literature [32,33]. Those commercially available are dredges,
grabs, and corers [34]. Dredges are used only for qualitative studies due to the
impossibility of controlling depth or location of the sampler. Grab samplers are
used to collect surficial samples 1-3 cm thick, which can be used for spatial
variability of a given property. Corers can be applied to obtain both surficial as
well as sediment column samples. With corers mud can be collected with the
least amount of disturbance.
Determination of some properties, e.g. total parameters such as chemical
oxygen demand or total organic content in sediments or solid wastes, requires
not only a proper sampler to collect representative samples but also the use of
correct sample preparation technique before the analysis proper [35].
The problem of sediment sampling has increased in significance because of
growing interest in the relationship between the quality of aqueous environ-
ments and chemical composition of sediment surface layers. The rule is to select
a sampling tool and technique which can give sediment samples with undis-
turbed surface layer structure. Such requirements are satisfied, e.g. by the
Niemisto corer [36] which is commonly used in Baltic studies to collect soft sedi-
ment samples. It consists of a metal tube, inside which a smaller diameter tube
made of plexiglass is fitted. It has stabilising fins near the top which assist verti-
cal penetration into the mud. The auto-valve is open during free fall and sedi-
ment penetration. The plunger is pushed into its machined seat; a vacuum
develops and prevents loss of the sediment on retrieval. This sampler is pre-
sented in Fig. 3.10.
Besides corers, different scoops such as the van Veen grab [37] (Fig. 3.11), or
the McIntyre-Day scoop [37] are used. They give samples whose structure is
partially disturbed during closing and raising the grab. After opening the grab
the sample is collected with a spoon or spatula, scraping the surface layer of
required thickness. Samples from a depth of 40 m or less can be taken by a

81
 a) b) c)

Left: Fig. 3.10. Niemisto corer. : Stabilising fins, 2: catch, 3: metal tube with inner tube made of
plexiglass (based on Ref. [36]).
Right: Fig. 3.11. Van Veen grab used for sampling sediments (based on Ref. [37]). (a) Before
lowering, (b) at the bottom, (c) during withdrawing. 1: Stainless steel jaws, 2: covers, 3: lines
opening covers, 4: central weight, 5: shutting mechanism, 6: weight on longer lever arm.

skin-diver who carefully drives a sampler into the bottom or scrapes the surface
layer of sediment.
For coarse-grained, gravelly and stony sediments, sufficiently heavy grabs
should be used, e.g. the McIntyre-Day grab. Obtaining repeatable results is
difficult in the case of such sediments. Coarse-grained fractions are discarded
before analysis by dry or wet sieving. The most representative results are
obtained when the fraction below 63 Cm is subjected to analysis. This fraction
accumulates organic and inorganic compounds and is very useful in ecological
studies.
Figure 3.12 is a schematic of a cryogenic sampler for sediments [38]. The
sampler is composed of a cold finger (5 cm in diameter) and protecting tube (i.d.
10 cm). The protection tube is used to thermally insulate a frozen sample. The
cold finger must be filled with freezing mixture (dry ice + methanol) before it is
pushed into the sediment layer. Immediately afterwards the protecting tube
must be slid over the finger to thermally insulate the sample. After a freezing
time of 10-40 min, the sampler is pulled out and the sample collected from the
finger.
Sediments are stored wet in glass or metal vessels at a temperature of-20OC.
Drying, if needed, is done by lyophilisation.
Figure 3.13 presents a sampler with a shutter closure [38]. It consists of a
rectangular tube made of plastic, whose bottom end can be closed with a shutter.
The lower part of the tube is divided into sections 5 cm high to collect sample
layers.

82
Fig. 3.12. Cryogenic sampler for sediments. 1: Cold finger, 2: connector pipe, 3: extension rod, 4:
connector pipe supplying cooling medium, 5: rubber flange, 6: ring for fastening transport line, 7:
stiffener, 8: protecting pipe, 9: ring seal, 10: rubber ring (based on Ref. [38]).

a) b)

Fig. 3.13. Shutter sampler for sediments. (a) Rear view, (b) end view. 1: Rod for manipulating
with closure, 2: tube part for fractionation, 3: shutter (based on Ref. [38]).

Samplers for suspended sediments are of different design and may be

classified into three general types: integrating samplers, pumping samplers and
instantaneous samplers [39]. The selection of sampler type depends on the aim
of the studies. Detailed discussion of the effect of sampler design and type on the
results obtained for different suspended matter characteristics and the
information looked for (temporal, directional variations) was given by Horowitz

83
et al. 40]. An appraisal of a simple sampling device for collected time-integrated
fluvial suspended sediment samples was given by Russell et al. [41].
As in the case of soils, special care is needed when sampling sediments for
analysis of volatile compounds. An important measurement for aquatic systems
is determination of bubble gas, whose composition can change during migration
from the sediment through the water column. To obtain gas directly from
profundal sediments of aquatic ecosystems a special sampler was proposed by
Huttunen et al. [42].
To get a sample representative of the whole population studied, and proper
sampler must be selected and sampling must be properly designed. For instance,
the effects of sampling design on estimating the magnitude and distribution of
contaminated sediments in a large reservoir were described by Pearson and Rose
[43].

3.8 GLOSSARY

Composite sample: a mixture of samples collected from more than one

sampling location, or at the same location more than once.
Gross sample: a portion of a material composed of all primary samples collected
from the same lot.
Heterogeneity: a situation when values of a given property or composition of
material in one place of a lot differ from values in other places of the same lot.
Heterogeneity is random if a property value in one portion of the population
studied is independent of the values in neighbouring portions and inde-
pendent of the position in the whole bulk of the material; heterogeneity is
directional when a property value changes along a specified direction in
non-random way; heterogeneity is periodic when a property value fluctuates
periodically around an average value.
Laboratory sample: a sample prepared from primary, gross or composite
samples intended for testing or analysis. The laboratory sample must retain
the composition of an original sample. Often reduction in particle size is
necessary in the course of reducing the quantity; it should be packed and
stored in such a way as to protect its identity.
Lot: a quantity of bulk material of similar composition whose properties are
under study.
Primary sample: an individual portion of material collected by a single
operation of a sampling device from one site of an unpacked material or
from a one site in a single package.
Profile sampling: collecting soil samples from individual layers or horizons of
a soil profile.
Representative sample: a sample that closely resembles, or is subset of, the
population being measured.
Sampling: an attempt to choose and extract a representative portion of a
physical system from its surroundings.

84
Segregation: a process or state in which a substance or particles undergo
distribution throughout a bulk material in a non-random way.
Test sample: a sample prepared from a laboratory sample, e.g., by means of
grinding, reduction, etc. Portions of test sample, termed test portions, are
taken for single measurements
Unit sample: a portion of a material composed of all primary samples collected
from one package.

REFERENCES

1 L.H. Keith, Environmental Sampling and Analysis: A Practical Guide. Lewis

Publishers, Chelsea, Michigan, 1991.
2 L.H. Keith (Ed.), Principlesof EnvironmentalSampling and Analysis, 2nd Edn. ACS
Professional Reference Book, Washington, DC, 1996.
3 L.H. Keith, W. Crummett, J. Deegan jr., R.A. Libby, J.K. Taylor and G. Wentler,
Anal. Chem., 55 (1983) 2210.
4 N.T. Crosby and P. Indu, General Principlesof Good Sampling Practice. The Royal
Society of Chemistry, Cambridge, 1995.
5 Ph. Quevauviller (Ed.), Quality Assurance in Environmental Monitoring:Sampling
and Sample Pre-treatment.VCH, Weinheim, 1995.
6 G.U. Fortunati and M. Pasturenzi, Quim. Anal., 13 (Suppl.) (1994) S5.
7 B. Kratochvil and J.K. Taylor, Anal. Chem., 53 (1981) 924A.
8 R.F. Cross, LC-GC, 18 (5) (2000) 468.
9 S. Nortcliff, Sci. Total Environ., 264 (2001) 163.
10 S.P. Theocharopoulos, G. Wagner, J. Sprengart, M-E. Mohr, A. Desaules, H. Muntau,
M. Christou and P. Quevauviller, Sci. Total Environ., 264 (2001) 51.
11 W. Horwitz, Pure Appl. Chem., 62 (1990) 1193.
12 W. Rasemann and B. Markert, Industrial waste dumps-sampling and analysis. In:
R.A. Mayers (Ed.), Encyclopaedia of Environmental Analysis and Remediation.
Wiley, New York, 1998, pp. 2356-2373.
13 B. Kratochvil, D. Wallace and J.K. Taylor, Anal. Chem., 56 (1984) 113R.
14 W.B. Cochran, Sampling Techniques, 3rd Edn. Wiley, New York, 1977.
15 C.O. Ingamells, Talanta, 23 (1976) 263.
16 J. Wisman, Mat. Res. Stds., 9 (1969) 51, 62.
17 A.A. Benedett-Pichler, In: W.G. Berl (Ed.), PhysicalMethods in ChemicalAnalysis.
Academic Press, New York, 1956, p. 183.
18 G.H. Fricke, P.G. Mischler, F.P. Staffieri and C.L. Housmyer, Anal. Chem., 59 (1987)
1213.
19 D.A. Skoog, D.M. West and F.J. Holler, Fundamentals of Analytical Chemistry, 6th
Edn. Saunders College Publishing, Fort Worth, Texas, 1992.
20 J. Krepin and R.L. Johnson, Soil sampling for environment assessment. In: M.R.
Carter (Ed.), Soil Sampling and Methods of Analysis. Lewis Publishers, Boca Raton,
Florida, 1993.
21 C. Krakowiak, Pobieraniepr6bek.Instrukcja og6lna., P.P.Polcargo, Gdynia, Poland,
1983 (in Polish).
22 Polish Technical Standards: PN/C-60008, Pr6bniki do pobierania pr6bek produkt6w
bezksztaltnych (Samplers for collecting shapeless materials) (in Polish).
23 B.M. Gawlik and H. Muntau (Eds.), Eurosoils II-Laboratory reference materials for
soil-related studies. Environmental Institute, Joint Research Centre, European
Commission, 1999.

85
24 T.E. Lewis, A.B. Crockett and R.L. Siegrist, Environ. Monitor. Assess., 30 (1994) 213.
25 A.D. Hewitt, T.F. Jenkins and C.L. Grant, Am. Environ. Lab., February 1995.
26 S.P. Theocharopoulos, G. Wagner, J. Sprengart, M-E. Mohr, A. Desaules, H. Muntau,
M. Christou and P. Quevauviller, Sci. Total Environ., 264 (2001) 63.
27 Polish Technical Standards: BN-74/9180-02, Analiza chemiczno-rolnicza gleby,
Pobieranie pr6bek (Chemical-agricultural analysis of soil. Sampling) (in Polish).
28 R.L. Siegrist and P.D. Jenssen, Environ. Sci. Techn., 24 (1990) 1387.
29 J. Siepak (Ed.), Fizykochemiczna analiza w6d i grunt6w, University of Adam
Mickiewicz, Poznan, 1992 (in Polish).
30 H.H. Rump and H.H. Kirst, LaboratoryManualfor the Examination of Water, Waste
and Soil. VCH, Weinheim, 1988.
31 J. Namiesnik, J. Lukasiak and Z. Jamr6giewicz, Pobieraniepr6beksrodowiskowych
do analizy. Wydawnictwa Naukowe PWN, Warszawa, 1995 (in Polish).
32 B.E. Markert, M.G. Tesmer and P.E. Parker, Water Res., 17 (1983) 603.
33 A. Murdoch, S.D. MacKnight, CRC Handbook of Techniques for Sediment Sampling.
CRC Press, Boca Raton, Florida, 1991.
34 U.M. Cowgill, Sampling and analysis of lake sediments. In: B. Markert, (Ed.),
Environmental Sampling for Trace Analysis. VCH, Weinheim, 1994.
35 S.D. Wachasunder and S. Satyanarayan, Ind. J. Environ. Protect., 9 (1989) 19.
36 L. Niemistb, A gravity corer for studies of softy sediments, Merentutkimuslait, Julk./
Harseforskiningsinst.Skr. 238, 1974.
37 N.A. Holme and A.D. McIntyre (Eds.), Methods of the Study of Marine Benthos.
Blackwell Scientific, Oxford and Edinburgh, 1971.
38 E. Ristenpart, R. Gitzel and M. Uhl, Water Sci. Technol., 25 (1992) 63.
39 A.J. Horowitz, A Primer on Sediment-Trace Element Chemistry. Lewis Publishers,
Chelsea, 1991.
40 A.J. Horowitz, F. Rinella, P. Lamothe, T. Miller, T. Edwards, R. Roch and D. Rickert,
Environ. Sci. Technol., 24 (1990) 13113.
41 M.A. Russell, D.E. Walling and R.A. Hodgkinson, The Role of Erosion and Sediment
Transport in Nutrient and Contaminant Transfer. Proceedingsof a Symposium held
at Waterloo, Canada, July, 2000. IAHS Publ. No. 263, 2000.
42 J.T. Huttunen, K.M. Lappalainen, E. Saarijarvi, T. Vaisanen and J. Martikainen,
Sci. Total Environ., 266 (2001) 153.
43 S.M. Pearson and K.A. Rose, Environnmetrics, 12 (2001) 81.

86
Chapter4

Analyte losses during storage

Barbara Zielinska and Wendy S. Goliff

4.1 INTRODUCTION

Often too little attention is given to the handling of environmental samples after
their collection and before actual instrumental analysis. This could lead to
errors, despite increasingly sensitive analytical systems. How and for how long
different samples can be stored to preserve their integrity for chemical, physical,
and biological testing is the subject matter of this chapter. This issue is particu-
larly important for the measurements of various organic compounds in ambient
air or in emission sources. Organic compounds are often highly reactive, some-
times unstable, and exhibit a wide range of volatility and are hence distributed
in the atmosphere between the gas and particle phases. Compounds having satu-
rated vapor pressure at 25°C greater than 10' mm Hg are generally classified as
volatile organic compounds (VOC) and are present entirely in the gas phase.
Compounds with saturated vapor pressure at 25°C between 10-1 and 10 - 7 mm Hg
are generally called semi-volatile organic compounds (SVOC) and are distrib-
uted between the gas and particle phases. Compounds having saturated vapor
pressure at 25°C less than 10 - 7 mm Hg are nonvolatile and are predominantly
adsorbed on particles. Different sampling techniques are required for the quan-
titative collection of VOC, SVOC, and nonvolatile organics (see, for example,
Zielinska and Fujita [1] for a review of sampling techniques). The storage and
transport requirements for samples collected by each of these techniques are
briefly reviewed below.

4.2 ELECTROPOLISHED CANISTERS

Electropolished canisters have been used for more than two decades to collect air
samples; they can hold sufficient sample for replicate analyses and are reusable.
Made of stainless steel, canisters passivated by the patented Summa® process
provide a stable environment for a large number of volatile organic compounds
(VOCs). They have been widely used for sampling source and ambient air, and
their use is recommended in the USEPA Method TO-14A [2]. It has been found
repeatedly that humid samples provide improved reproducibility and storage
ComprehensiveAnalytical ChemistryXXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 87
over dry. Water is thought to occupy active sites on the canister walls, increasing
the passivity of the surface.

4.2.1 Hydrocarbons and halocarbons

Past studies of sample integrity in canisters have focused mainly on a select
number of hydrocarbons and halocarbons. Rasmussen and Lovelock [3] investi-
gated the stability of halocarbons over a four-year period, finding no statistical
change during that time. Westberg et al. [4] performed a stability test with a syn-
thetic sample containing a mixture of fourteen compounds including paraffins,
olefins and aromatic hydrocarbons, analyzing the mixture twice per week for a
four-week period. They found poor precision for o-xylene, n-propyl benzene and
1,2,4-trimethyl benzene, and concluded that these aromatic compounds were
lost within the sample containers. Oliver et al. [5] collected ambient air spiked
with fifteen halogenated VOCs in sets of new and used canisters, then deter-
mined concentrations of the contents during 7-day and 30-day storage periods.
They found a large amount of scatter in the data for o-xylene, which they attrib-
uted to its low concentration. Brymer et al. [6] reported the accuracy, precision
and storage stability for 194 VOCs, including alkanes, alkenes, alkynes, alde-
hydes, alcohols, ketones, aromatics halocarbons, and nitrogen- and sulfur-
containing compounds in synthetic mixtures over a 30-day period. They found
167 compounds remained stable for the entire period. Concentrations of methyl
mercaptan, ethyl mercaptan, butyl mercaptan, dimethyl acetal, and bis(chloro-
methyl) ether were below their limit of detection at day 7, which was attributed
to poor compound stability. Three compounds were considered unstable by the
researchers due to their relatively low volatility: n-undecane, 2,3-dirnethyl-
butane and dichlorodifluoro-methane, which is an unlikely explanation. Evans
et al. [7] studied the stability of 31 organic compounds for 32 weeks collected in
seven canisters from the headspaces of high-level radioactive waste tanks. They
found no significant change in concentration in all but two compounds, propanol
and 2-butanone. Goliff and Zielinska [8] reported canister stability for most
VOCs for a period of six months, including olefins, biogenics, aromatics and oxy-
genated compounds (MTBE, ethyl acetate and hexanal), without compromising
analytical integrity, in contrast to Brymer et al. [6].
Zielinska et al. [9] studied canister sample storage stability over a six-month
period. They found that the previous history of the canister, as well as its age,
sample storage temperature and canister cleaning methods, could affect the
recovery of the samples.
The US EPA Method TO-14A recommends storage times of no more than 30
days for samples collected in canisters.

4.2.2 Oxygenated compounds

There have been some studies that have specifically targeted polar volatile
organic compounds for stability testing in canisters. Pate et al. [10] studied a

88
synthetic mixture of ten polar organic compounds (methanol, acetone, isoprene,
acrylonitrile, vinyl acetate, methyl ethyl ketone, t-butyl methyl ether, ethyl
acetate, n-butanol, and ethyl acrylate) and two nonpolar compounds (methyl
chloroform and toluene, as controls) in unpolished and polished canisters under
humid and dry conditions over a 31-day period. They found good recovery (100 ±
10%) with exception of methanol (120% after 7 days and 145% after 31 days) in
humid canisters (passivated and unpassivated). The stability of the polar
compounds in the dry canisters was found to be very poor. Kelly et al. [11]
examined the stability of a humidified synthetic mixture of 16 compounds
(1,3-butadiene, methanol, ethanol, acetonitrile, acetone, 2-propanol, acrylonitrile,
methyl t-butyl ether, vinyl acetate, methyl ethyl ketone, ethyl acetate, ethyl
t-butyl ether, benzene, 1-butanol, ethyl acrylate, and toluene) over a 12-day
period, and observed good precision for most of the compounds (with a coefficient
of variation below 25%), with the exception of vinyl acetate and the alcohols.
Batterman et al. [12] investigated a synthetic mixture of seven aldehydes and four
terpenes in electropolished humidified and dry canisters for 16 days. They found
poor stability in all canisters studied, although humidified samples were reproduc-
ible for a longer period than those in dry environments. In the study of Goliff and
Zielinska [13], the stability of oxygenated compounds in whole air samples in
electropolished canisters was investigated. Three canisters-one each spherical
and cylindrical 6-liter, and one cylindrical 3-liter-were used to collect whole air
samples. Their contents were tested over a four-month period for stability of
selected oxygenated compounds, including acetaldehyde, methanol, acetone,
2-propanol, MTBE, hexanal, octanal, nonanal and benzaldehyde. They found
hexanal and MTBE to be the only oxygenated compounds to show good stability
(relative standard deviation less than 0.10) in all three canisters, while methanol
and acetone displayed the poorest precision.
Gholson et al. [14] studied sample storage of VOCs in aluminum canisters,
passivated and non-passivated, and with and without water. They' found that the
stability of VOCs in aluminum canisters was dependent on the humidity of the
sample and the reactivity of the compounds analyzed. The authors recom-
mended against the usage of aluminum canisters for polar compounds due to
their decreasing concentrations over time. In the USEPA Method TO-14A,
stainless steel canisters are recommended for sampling.

4.3 SOLID ADSORBENTS

The other method of choice for sampling of VOCs in air is collection on solid
adsorbents, where they are removed from possible reaction with other collected
species. A method of preconcentration followed by separation by gas chromatog-
raphy has been standardized by the U.S. EPA (TO-1) [15] for sample collection
and analysis using Tenax, a polymer of 2,6-diphenyl-p-phenylene oxide for
hydrocarbons and halocarbons from C6 to C20 , and by USEPA Method TO-17 [16]
using Carbotrap, Carbosieve, and other adsorbents for more volatile compounds.

89
While there have been many publications regarding artifacts and degradation of
species with a variety of adsorbents, such as Tenax GC, Tenax TA, Carbotrap
and activated carbon [1,17-23], there are relatively few concerning sample
storage. The US EPA Method TO-17 recommends analysis of adsorbent samples
within two weeks of sampling.
Rudling et al. [24] investigated the stability of organic solvents on two
activated carbons, SKC and Merck, during storage times of up to 50 days.
Ketones showed large losses, which the authors attributed to reaction with the
adsorbent material. Compounds which were found stable (less than 5% loss) on
both carbons included toluene, styrene, and 1,1,1-trichloroethane.
Schrader et al. [25] studied the degradation of a-pinene on Tenax-TA under
different storage conditions, including samples placed in sunlight, samples
wrapped in aluminum foil, storage in a refrigerator at 4C, and storage in a
freezer at -18°C. After one month of storage, wrapped samples showed over 99%
recovery. Analysis of samples exposed to daylight showed evidence of degrada-
tion after one day. At one month, 12% of the o-pinene had reacted. The authors
found that storage in a freezer to be most effective in reducing the degradation of
c-pinene.
Zielinska et al. [9] investigated storage stability of hydrocarbons from C to
C1 9 in Tenax samples over a storage period of 7 to 8 months. They found that the
difference in total concentrations was only +5% to -12% (the negative value
indicating a decrease in concentration), but the differences in the concentrations
of the less abundant hydrocarbons were much higher: up to -50%.
There have been some preliminary studies performed by Goliff and Zielinska
[26] looking at sample stability of Tenax-TA stored for more than six months.
Storage conditions were as follows. Sampling tubes were capped with
Swagelok® fittings and then wrapped in aluminum foil. The tubes were then
placed in containers along with activated charcoal. The containers of sampling
tubes were stored in freezers at -20°C. Initial findings suggest good sample
integrity under these storage conditions, including the lightest compound: ethyl
benzene. There was one anomalous sample that showed large losses, which the
authors attribute to sampling error.

4.4 CARBONYL COMPOUNDS

The classic example of a selective preconcentration method for organic gas

sampling is the collection of carbonyl compounds by their derivatization with
2,4-dinitrophenylhydrazine (DNPH). The acid-catalyzed derivatization of
carbonyls proceeds by nucleophilic addition of DNPH to a C=O bond, followed
by 1,2-elimination of water to form 2,4-dinitrophenylhydrazone.
The DNPH-hydrazones, formed during sampling, are non-volatile and
remain on the sampling medium, which is usually either a reagent-impregnated
cartridge or an impinger charged with the reagent solution. The yellow to
deep-orange colored DNPH-hydrazones have UV absorption maxima in the

90
360-375 nm range and can be analyzed by the high performance liquid
chromatography (HPLC) method coupled with UV detection; this method offers
very high selectivity and sensitivity of analysis. The US EPA Method TO-11A
[27] and TO-5 [28] describe the collection and analysis of carbonyl compounds in
ambient air, using either DNPH impregnated cartridges (silica gel or C,,), or
impinger, respectively.
Hydrazones of stable carbonyls, such as formaldehyde, acetaldehyde, pro-
pionaldehyde, acetone, butanal and butanone, maintain their integrity on silica
gel and C18 cartridges, as well as in impinger reagent solutions, for over a month
under refrigerated storage [29,30]. The stability of these hydrazones in the
cartridges at ambient temperature has been also demonstrated [31] by the
collection of duplicate samples in the field. One sample from each duplicate pair
was retrieved and analyzed immediately after collection, while the other
remained in the sampler until it was retrieved during the normal weekly visit by
the field technician. The excellent agreement between the pairs confirmed that
after collection the samples were stable during the seven days of holding time
encountered in the study.
However, it has been reported [30] that the olefinic aldehydes such as
acrolein and crotonaldehyde degrade partially on the cartridges, either during
sampling or storage, and form unknown products. Also, the acrolein-DNPH
derivative is not stable in the strongly acidic DNPH solutions used in the
impinger method. In order to preserve its integrity it has to be extracted from
this solution immediately after sampling [32]. The EPA Method TO-11A [27]
recommends a maximum storage time of 2 weeks at or below 4°C for samples
collected on silica gel or C cartridges. However, the extracts may be stored
refrigerated for a longer period, up to approximately one month.
A study was performed by Fung et al. [33] to assess the effectiveness of vari-
ous types of packing used to protect the DNPH cartridges from contamination
during storage and shipment. In general, the amount of contamination was
reduced significantly by using a protective closure as opposed to an open car-
tridge. Even the PVC caps, the worst performers of the group, reduced formalde-
hyde exposure to approximately 2% of an open cartridge. For formaldehyde, a
polyethylene cap provided ten times more protection than PVC caps and about
twice as much protection as plugs. Screw-capped vials offer additional protec-
tion. Samples should be shipped from the field tightly capped, packed in screw-
capped vials, and chilled with blue ice (or equivalent).
Exposure to sunlight causes significant production of carbonyls, which can
be eliminated by wrapping the cartridges in aluminum foil during sampling and
storage [34].

4.5 SEMI-VOLATILE ORGANIC COMPOUNDS

Semi-volatile organic compounds (SVOC) are usually collected using filters

(either quartz fiber or Teflon-coated glass fiber filters) followed by a solid

91
adsorbent, such as polyurethane foam (PUF) plugs, or XAD-2 or 4 resins, or
"sandwich" type PUF/XAD/PUF cartridges. This method of collection is used for
organic compounds, which are distributed between gas and particle phases, such
as polycyclic aromatic hydrocarbons (PAH), polychlorinated dibenzo-p-dioxin
and furans, pesticides, polychlorinated biphenyls, etc. The US EPA compendium
[35] of methods for the determination of toxic organic compounds in ambient air
describes the standardized procedures for the collection and analysis of some of
these SVOC (Methods TO-4A [36], TO-9A [371, TO-10A [38], TO-13A [39]).
These methods also specify the holding time and the storage requirements for
the specific samples. For example, the U.S. EPA Method TO-13A requires the
extraction of samples collected for PAH analysis within seven days from sample
collection if stored in a refrigerator at or below 4°C. The extracts should be
analyzed within 40 days of extraction. In addition, since PAH are chemically
reactive, and heat, ozone, NO2 , and UV light may cause sample degradation
during transport and analysis, it is recommended that samples are shipped
chilled (in blue ice) overnight from the field to the analytical laboratory and
incandescent or UV-shielded fluorescent lighting should be used in the
laboratory during sample preparation and analysis.
The maximum seven-day requirement for sample holding time is very
stringent for most analytical laboratories. Some studies have suggested that the
holding time can be extended to a month or more if samples are stored in much
lower temperatures in proper containers. Svaderup et al. [40] investigated
several different sample packing techniques and the subsequent stability of the
surrogate samples under different storage conditions. Surrogate samples for
power plant plume fly ash were created by using electrostatic precipitator
hopper ash. Particles smaller than 10 Am were suspended and mixed with three
PAH, one nitro-PAH, salt, and sulfuric acid. The particle mixture was then
collected on quartz filters to create surrogate samples. Samples were stored for
periods of 30 and 120 days at either 20 or -79°C, as well as under a set of variable
conditions of temperature, light, and humidity. Samples were then analyzed by
gas chromatography/mass spectrometry (GC/MS) and scanning electron
microscopy and compared with samples analyzed right after collection (day 0).
The authors concluded that preservation of organic compounds in the samples
was best achieved when the samples were packed in Teflon-wrapped glass dishes
and stored at -79°C in the dark. Under these conditions, loss of the lightest PAH
(fluorene) was minimized and the heaviest PAH was stable for at least 120 days.
However, sulfuric acid had a deleterious effect on two of the spiked PAI. For
nitro-PAH (2-nitrofluorene) the results were inconclusive due to the unequal
loadings on the filters.
Chuang at al. [41] reported the results of storage stability study of PAH on
quartz filters and two different sorbents: PUF and XAD-2. Air samples were col-
lected on quartz filters followed by PUF cartridges or XAD-2 resin. Prior to sam-
pling, the sorbent was spiked with known amount of perdeuterated PAH. After
sampling, the samples were stored in the dark at room temperature (--20"C) for

92
0, 10-, 20-, and 30-day intervals, and at -20°C for 10- and 20-day intervals. The
results showed no significant loss of PAH on XAD-2 after 30 days of storage in
the dark at room temperature, but decreasing concentrations of naphthalene,
anthracene, and benzo(a)pyrene on PUF. Storage losses of PAH collected on
quartz filters were insignificant under these conditions, with the exception of
cyclopenta[c,d]pyrene, which decreased to approximately half of the initial
amount. Levels of its degradation product, pyrene dicarboxylic acid anhydride,
increased more than 1.5 times after 30 days of storage. In addition, levels of 2-
and 3-nitrofluoranthene decreased more than 30% of the initial amount. How-
ever, when samples were stored at -20°C in the dark, all target PAH and nitro-
PAH were stable on quartz filter and XAD-2 cartridges when analyzed after 10
and 20 days of storage.
For analysis of polychlorinated/polybrominated dibenzo-p-dioxins and
furans, Method TO-9A specifies a maximum holding time of 7 days from sample
collection to extraction when stored at or below 4C. The extract should be
analyzed within 40 days of extraction. However, since dioxins and furan are
chemically stable, it is possible that storage in much lower temperatures to
prevent losses due to evaporation would result in much longer safe sample
holding times. EPA Methods TO-4A and 10A, describing the determination of
the broad range of pesticides and polychlorinated biphenyls by high- and low
volume sampling, respectively, requires sample extraction within seven days of
collection when stored at or below 4°C, and analysis of the extract within 40 days
of extraction. The ASTM document D-4861-00, Appendix X3 [42] gives data on
the storage stability of various common pesticides on PUF cartridges, when
stored at room temperature (24°C) for 30 days. With the exception of several
pesticides (for example, Ronnel, Malathion, Chlorothalonil) the losses are in the
range of 0-40%. The ASTM document recommends that the cartridges be stored
at water ice or dry ice temperatures immediately after collection in the field and
at -10°C or lower temperature in the laboratory freezer for no longer than 30
days prior to extraction.
In addition to the US EPA Methods for ambient air monitoring, the National
Institute for Occupational Safety and Health (NIOSH) publishes the Manual of
Analytical Methods [43], suitable for the determination of various toxic
chemicals concentrations in occupational environment. In general, these
methods use solid adsorbents such as PUF, XAD, activated charcoal or Tenax,
and sometimes, filters followed by solid adsorbents. Thus, the same sample
holding time considerations would apply to these samples. It has been reported
[44,45] that organophosphorus and organonitrogen pesticides collected on
OSHA versatile sampler tubes (OVS, consisted of XAD-2 resins) were stable
when stored for up to 30 days under either ambient or refrigerated conditions.
Although the authors did not indicate the necessity for sample refrigeration,
they noted minor improvements in sample recovery at the 21- and 31-day
storage time in some instances. Therefore, refrigeration of samples prior to
analysis was recommended.

93
TABLE 4.1
Recommended storage conditions for sampling media
Sampling media Temperature (EPA) Duration of storage (EPA) Lighting
requirements
Canisters At room temperature 6 months (1 month) Not applicable
Solid adsorbents -20°C (-20°C) 8 months (2 weeks) In the dark
DNPH cartridges (<4°C) (2 weeks) In the dark
Filter-PUF-XAD (<4'C) 7 days In the dark

4.6 SUMMARY

The US EPA recommendations for storage conditions for environmental sam-

pling media are more stringent than those based on experimental data. In the
real world, the EPA requirements are hard for most analytical laboratories to
fulfil. In these cases, the authors recommend alternative storage conditions
listed in Table 4.1 (with US EPA recommendations in parentheses) based on the
available literature.

REFERENCES

1 B. Zielinska and E. Fujita, In: B. Market (Ed.), Environmental Sampling for Trace
Analysis. WCH, Weinheim, 1994, Chapter 7.
2 US EPA, Method TO-14A, In: Compendium of Methods for the Determination of
Toxic Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/010b, January
1999.
3 R.A. Rasmussen and J.E. Lovelock, J. Geophys. Res., 88 (1983) 8369-8378.
4 H. Westberg, W. Lonneman and M. Holdren, In: L.H. Keith (Ed.), Identification and
Analysis of OrganicPollutantsin Air. Butterworth, Woburn, MA, 1984, pp. 323-337.
5 K.D. Oliver, J.D. Pleil and W.A. McClenny, Atmos. Environ., 20 (1986) 1403-1411.
6 D.A. Brymer, L.D. Ogle, C.J. Jones and D.L. Lewis, Environ. Sci. Technol., 30 (1996)
188-195.
7 J.C. Evans, J.L. Huckaby, A.V. Mitroshkov, J.L. Julya, J.C. Hayes and J.A. Edwards,
Environ. Sci. Technol., 32 (1998) 3410-3417.
8 W.S. Goliff and B. Zielinska, Paper 1122, Proceedings of the Air &Waste Manage-
ment Association's 93rd Annual Conference &Exhibition, Salt Lake City, Utah, June,
2000.
9 B. Zielinska, J.C. Sagebiel, G. Harshfield, A.W. Gertler and W.R. Pierson, Atmos.
Environ., 30 (1996) 2269-2286.
10 B. Pate, R.K.M. Jayanty, M.R. Peterson and G.F. Evans, J. Air Waste Manage.
Assoc., 42 (1992) 460-462.
11 T.J. Kelly, P.J. Callahan, J. Plell and G.F. Evans, Environ. Sci. Technol., 27 (1993)
1146-1153.
12 S.A. Batterman, G.-Z. Zhang and M. Baumann, Atmos. Environ., 32 (1998)
1647-1655.
13 W.S. Goliff and B. Zielinska, Paper 403, Proceedings of the Air &Waste Management

94
 Association's 9 4 th Annual Conference &Exhibition, Orlando, Florida, June 2001.
14 A.R. Gholson, J.F. Storm, R.K.M. Jayanty, R.G. Fuerst, T.J. Logan and M.R.
Midgett, JAirPollut. Contr. Assoc., 39 (1989) 1210-1217.
15 US EPA, Method TO-1, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/0
10b, January 1999.
16 US EPA Method TO-17. In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/010b, January 1999.
17 G. MacLeod and J.M. Ames, J. Chromotogr., 355 (1986) 393-398.
18 B. Zielinska, J. Arey, T. Randahl, R. Atkinson and A.M. Winer, J. Chromatogr.,363
(1986) 382-386.
19 H. Rothweiler, P.A. Wager and C. Schlatter, Atmos. Environ., 25B (1991) 231-235.
20 X.-L. Cao and C.N. Hewitt, Chemosphere, 27 (1993) 695-705.
21 X.-L. Cao and C.N. Hewitt, Atmos. Environ., 27A (1993) 1865-1872.
22 A. Calogiroou, B.R. Larsen, C. Brussol, M. Duane and D. Kotzias, Anal. Chem., 68
(1996) 1499-1506.
23 C. Coeur, V. Jacob, I. Denis and P. Foster, J. Chromatogr. A, 786 (1997) 185-187.
24 J. Rudling, E. Bjorkholm and B.-O. Lundmark, Ann. Occup. Hyg., 30 (1986) 319-327.
25 W. Schrader, J. Geiger, D. Klockow and E.-H. Korte, Environ. Sci. Technol., 35
(2001) 2717-2720.
26 W.S. Goliff and B. Zielinska, private communication.
27 US EPA Method TO-11A, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/010Ob, January 1999.
28 US EPA Method TO-5, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/10Ob, January 1999.
29 K. Fung, M. Porter, D. Fitz and E.M. Fujita, PlanningAir QualityMeasurement and
Modeling Studies: A Perspective Through SJVAQS/AUSPEX, P. Solomon (Ed.), Air
&Waste Management Association, Pittsburgh, PA, 1993.
30 A. Vairavamurthy, J.M. Roberts and L. Newman, In: E.D. Winegar and L.H. Keith
(Eds.), Sampling andAnalysis of AirbornePollutants. Lewis Publishers, Boca Raton,
FL, 1993, Chapter 10, pp. 149-174.
31 K. Fung and B. Wright, Aerosol Sci. Technol., 12 (1990) 44.
32 R.R. Freeman, Proceedingsof the Third Annual West Coast Regional Air & Waste
Management Association Conference On Current Issues in Air Toxics, L. Edwards
(Ed.), Air &Waste Management Association, Pittsburgh, PA, 1992, pp. 48-53.
33 K. Fung, M. Porter, D. Fitz and E. Fujita, Planning and Managing Regional Air
Quality Modeling and Measurement Studies-A Perspectivethrough the San Joaquin
Valley Air Quality Study and AUSPEX, P.A. Solomon and T.A. Silver (Eds.), Lewis
Publishers with Pacific Gas and Electric Company, 1995.
34 X. Zhou and K. Mopper, Environ. Sci. Technol., 24 (1990) 1482-1485.
35 US EPA, Compendium of Methods for the Determination of Toxic Organic Com-
pounds in Ambient Air, 2nd Edn. EPA/625/R-96/010Ob, January 1999.
36 US EPA Method TO-4A, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/O10b, January 1999.
37 US EPA Method TO-9A, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/010b, January 1999.
38 US EPA Method TO-10A, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA/625/R-96/01Ob, January 1999.
39 US EPA Method TO-13A, In: Compendium of Methods for the Determinationof Toxic
Organic Compounds in Ambient Air, 2nd Edn. EPA625/R-96/010OlOb, January 1999.
40 G.M. Svaderup, B.E. Buxton and J.C. Chuang, Environ. Sci. Technol., 24 (1990)
1186-1195.
41 J.C. Chuang, N.K. Wilson and R.G. Lewis, FreseniusEnvir. Bull., 8 (1999) 547-556.

95
42 American Society for Testing and Materials (ASTM), Standard practice for sampling
and selection of analytical techniques for pesticides and polychlorinated biphenyls in
air. In: Annual Book of ASTM Standards, D-4861-00, 2000.
43 M.E. Cassinelli and P.F. O'Connor (Eds.), NIOSH Manual of Analytical Methods
(NMAM), 4th Edn. DHHS (NIOSH) Publication 94-113, August 1994.
44 E.R. Kennedy, M.T. Abell, J. Reynolds and D. Wickman, Am. Ind. Hyg. Assoc. J., 55
(1994) 1172-1177.
45 E.R. Kennedy, J. Lin, J.M. Reynolds and J.B. Perkins, Am. Ind. Hyg. Assoc. J., 58
(1997) 720-725.

96
Chapter 5

Automated diffusion-based collection and

measurement of atmospheric trace gases
Purnendu K Dasgupta

5.1 INTRODUCTION

During the last two decades, there has been an increasing realization of the
importance of trace gases and aerosols: they play a significant role in almost all
areas of air quality related concerns, including photochemical smog, strato-
spheric ozone depletion, global climate change, indoor air quality, public and
occupational health, and the genesis of atmospheric acidity and its effects. The
implications of trace gases and aerosol particles on atmospheric visibility span
the gamut from esthetic values of clear sight across the Grand Canyon to the
successful mission of an optically guided armament.
Both gases and particles can be organic or inorganic. In the gas phase, the
dominant components of interest in ambient air are typically simple, small
molecules, with molecular weight (MW) usually <100. Ambient particulate
matter is a more complex mixture than the air it is suspended in. Although
atmospheric fine particulate matter is also dominantly inorganic (urban fine
particles are typically composed of ammonium sulfate and/or nitrate), signifi-
cant quantities of organics, including large and complex macromolecules, are
often present in ambient aerosols in readily detectable concentrations. The
automated measurement of such compounds is a daunting task and not much
progress has yet been made.
The focus of this book is on sample preparation. The best sample preparation
technique is one in which sampling, sample preparation, and analysis are
coupled in an integral and automated manner. This reduces contamination
problems, labor, analysis cost per sample, and improves data quality, system
reliability and throughput.

5.1.1 Gases

Air samples are not spatially and temporally homogeneous. Moreover, the
aerosol may or may not be in equilibrium with the gas phase it is suspended in. If
a sample is collected in a composite manner, often the phase-specific origin of an

Comprehensive Analytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 97
analyte can no longer be discerned. When a fritted bubbler/ impinger filled with
dilute acid is used to collect gaseous ammonia, for example, the collector also
partially collects fine particles that contains ammonium. Once in solution, the
phase specific origin of the NH4+ cannot be identified. Filtration or other means
of removal of aerosols prior to collecting the gas is usually unsatisfactory because
the incoming gas tends to interact with the deposited particles. Ammonia may be
removed by an acid sulfate particle that has already been collected on a filter
through which sampling is taking place. Generally, the use of a filter or other
particle pre-removal means lead to the underestimation of the gaseous and over-
estimation of the particle-phase analyte. However, occasionally the reverse
occurs from particle-particle interactions-gaseous ammonia may, for example,
be liberated when an ammonium sulfate particle impacts on an alkaline crustal
particle.
For these reasons, diffusion discrimination is presently the method of choice
for differentiating between the same analyte present both in the gas and the
particle phase. Even the smallest atmospheric particle of interest has a diffusion
coefficient 103-104 times smaller than that of a gas molecule. As such, when a
gas/particle mixture is passed through a diffusion denuder (in its simplest form,
a tube with its walls coated with a substance that acts as a sink for the gas), the
gas is collected on the walls as gas molecules diffuse to the walls while the
particles are transmitted under laminar flow conditions. The particles can then
be collected and analyzed.
In a previous review, automated approaches to the diffusion based collection
and analysis of atmospheric trace gases were discussed in considerable detail [1].
In the present review, only further developments and the essential principles of
this aspect will be addressed. Most solid phase microextraction (SPME)-based
sampling approaches for gases are also diffusion-based; this is covered
extensively elsewhere in this volume and will not be discussed here further.

5.1.2 Aerosols

The influence of aerosols on global climate is considered to be very important.

Considerable spatial and temporal variations in both aerosol composition and
concentration make it difficult to evaluate the global role of aerosols relative to
that of the gaseous pollutants [2]. There has been a dramatic change in the
attitude towards atmospheric aerosols in recent years, at least in part due to
social, political and judicial pressures [3]. Because of studies that revealed the
detrimental health effects of atmospheric aerosols, regulations now exist on the
highest allowable concentrations of atmospheric particulate matter. In the U.S.,
since the promulgation of National Ambient Air Quality Standards (NAAQS),
many studies have been conducted on the composition and concentration of
atmospheric aerosols. Nevertheless, the existing NAAQS is related to measure-
ments of particle mass concentration within a given particle size fraction, that
with aerodynamic diameters equal to and below 10 and 2.5 Am (PM,, and PM 2 5 ),

98
respectively. This situation largely reflects the capability of extant commercial
instrumentation. While it is obvious that chemical composition of the aerosols
must significantly influence any health effects they may elicit, there are still no
chemically specific regulations. In contrast, the NAAQS has been revised several
times from the original size-indifferent mass concentration parameter (referred
to as total suspended particulate matter or TSP) to a mass concentration
parameter of 50 g/m3 annual average and 150 Ag/m 3 24 h average for PM,, in
1987 and additionally, 15 Ag/m 3 annual average and 65 Ag/m 3 24 h average for
PM2 .5 in 1997 [4]. This has been possible due to the availability of automated
measuring instruments capable of providing size-fractionated mass measure-
ments, some on a near real time basis. A wealth of information on mass
concentration as a function of particle size is available (albeit to date, such data
primarily concerns the PM,, fraction but this is changing). The need for and the
approaches to automated aerosol composition measurement are addressed in a
separate review.

5.2 COLLECTION AND ANALYSIS OF TRACE GASES

5.2.1 Diffusion based collection

Townsend [5] first suggested the use of diffusion denuders to discriminate

between gases and particles and Crider et al. [6] described the first analytical
instrument that utilized such a device. Simple single-tube denuders collect a gas
with a collection efficiency f that is dependent on the length of the tube L, the
diffusion coefficient of the gas D, and the sampling rate Q according to an
equation originally developed by Gormley and Kennedy [7] (assuming that a
collision with the wall is 100% effective in removing the gas):

f = 1 - 0.81905 e 8 - 0.09753 e- 22 30 55 - 0.0325 e-56 '961 - 0.01544 e

656
- ...
(5.1)
where
A = DL/Q (5.2)
A represents a dimensionless quantity as long as units are chosen consistently.
Note that Eq. (5.1) is applicable only for cases where a significant amount of the
gas is collected. (Indeed, even if A is set to a value of zero, an -18% collection is
predicted!). Also, beyond the first exponential term, in most cases further terms
make no significant contribution. Theoretically aerosol transmission through a
simple cylindrical tube is expected to be high and this is indeed experimentally
the case [8].
Assuming that the wall does behave as a perfect sink, a single denuder tube
of 50 cm length will remove 100%, 98.6% and 84% NH3 (D = 0.234 cm2 /s) at flow
rates of 1, 2 and 5 liters per minute (LPM), respectively. The same numbers for a
more slowly diffusing gas, e.g., SO,, are, respectively, 99.1, 91.3 and 66.6%. It

99
should also be borne in mind that the collection efficiency will decrease further
when every collision is not effective in uptake. This aspect has been quantita-
tively examined in many studies and has been adequately reviewed previously
[1]. The salient point is that a simple single tube denuder generally is a
quantitative collector only at relatively low flow rates, this may not be acceptable
in all applications.

5.2.1.1 More efficient denudergeometries

Changing the geometry of a single tube from linear to some other, e.g., a coiled
configuration, does result in an increase in collection efficiency but also
increases the loss of larger particles [9]. Overall, such geometries have not
proven popular. The logical adjunct to a single cylindrical tube is a multiplicity of
such tubes in parallel (often referred to as the Gatling Gun design). When the
same total flow is divided through n parallel channels, the resulting flow through
each tube proportionately decreases and collection efficiency increases accord-
ingly. Krieger and Hites [10] have described a novel twist to the Gatling gun
design of a diffusion denuder: a 30 m x 320 m i.d. capillary gas chromatography
column was cut into 120 segments of 25 cm length and glued together to form an
-1.25 cm diameter bundle, permitting efficient sampling at 1.5 1 min-. Collected
analytes were directly thermally desorbed into an analytical gas chromatogra-
phy column. The results obtained with the denuder for sampling polychlorinated
biphenyls and polynuclear aromatic hydrocarbons were later validated [11].
Other groups have since used gas chromatography capillary column segments as
tubular denuders 12,13].
Glass "honeycomb" denuders with hexagonal cross sections of the individual
tubes, built on the same Gatling gun principle, can be coated with suitable
adsorbents and make for a particularly attractive compact design, especially for
use as a personal sampler [14,15]. In typical deployment, the denuder collects
the desired gases. More than one denuder element can be connected in series,
e.g., one coated with Na2 CO3 for collecting acid gases may be followed by one
coated with citric acid for collecting NH. A filter to collect particles then follows
the denuder assembly. The device has been patented [16] and is available
commercially with different inlet systems in an attractive format [17]. Particle
losses in these denuders can be higher than some other designs, however [18],
and it would not be a simple matter to connect such a denuder to a wet analysis
system in an automated fashion.
Probably the most widely used denuder geometry for the selective collection
of various ambient trace gases is of the annular type. Here, the inner surface of
the outer tube and the outer surface of the inner rod/tube are suitably
sorbent-coated. Introduced originally by Possanzini et al. [191, such denuders
also obey Eq. (5.3) where f becomes a function of the inner and outer diameters
(di and do) defining the annulus:
f = 1 -A exp (-aA(d o + di)/(d -di)) (5.3)

100
where A and a are dependent on the effectiveness of wall collisions and are
generally determined empirically. In the case of perfect sink efficiency, the
values of 0.91 and 3.77 forA and a are respectively expected [20].
Glass annular denuders are commercially available from several vendors
[21]. One interesting addition to the annular denuder family is a carbon-coated
version; this is useful for the collection of NO 2 and peroxyacetyl nitrate [22]. The
gas collection efficiency and entrance flow effects in the annular denuder design
have been theoretically modeled and compared with experimental data [23].
Multiple concentric tubes can be thought of as the same geometric analog to
an annular denuder that a Gatling gun design is to a single tube denuder.
Coutant et al. [24] designed a 12-element concentric denuder for high volume
collection of organics. A program is available from the authors to compute the
collection efficiencies of such denuders (containing any number of elements)
where the wall reaction probability can be input as a separate parameter.
Mathematically, the annular denuder geometry assumes a parallel plate
configuration as the diameter of the tube(s) approaches infinity. Simon and
Dasgupta first reported this parallel plate denuder (PPD) geometry in a
continuously operating configuration [25]. For two parallel plates of active width
w separated by a distance s, the collection efficiency is given by [26]:
f = 1- 0.91 exp (-2.4wA/s)) (5.4)
Such a denuder is much more efficient than its circular counterpart. A denuder
with L = 30 cm, w = 5 cm and s = 0.3 cm will remove SO2 with an efficiency of
99.8% and 95. 2% at flow rates of 5 and 10 LPM, respectively.
To avoid the oddity of a rectangular aperture in a world of circular connect-
ing tubes, Staffelbach et al. [27] described a circular end parallel plate denuder
fabricated by squashing the middle, active, section of a circular tube into a
flattened, rectangular geometry while inlet/outlet connections were made to the
circular termini.
Multiple parallel plates instead of a single pair of active plates provide for very
high flow rates. Eatough et al. [28-33] have pioneered such denuders and incorpo-
rated them in integrated gas-particle differentiating systems and provided them
with memorable acronyms: The BOSS (Brigham Young University Organic
Sampling System) [28], the BIG BOSS (the BOSS capable of sampling at 250
LPM) [32], the PC (Particle Concentrator)-BOSS [33] and the RAMS (Real-Time
Ambient Mass Sampler) [33]. Except for the last device, measurement strategies
are off-line. The multiple PPD in BOSS was composed of seventeen 4.5x58 cm
strips of charcoal impregnated filter (CIF) strips, separated by 2 mm dia. glass
rods at the long edges and contained in a 5 x 5 cm square cross section Al tube. The
denuder is designed to remove vapor phase organic material and is followed by
quartz and CIF filters. Particle losses in the denuder are - <10%. After collection,
CIF and quartz filters are heated in a temperature-programmed manner in an N2
atmosphere. The evolved gases are catalytically oxidized to CO 2 , which is then
measured by a non-dispersive infrared analyzer.

101
 The PC-RAMS consists of a particle concentrator inlet originally developed
by Sioutas et al. [34-36]. The PC is a slotted virtual impactor with a 50% cut
point at a mass median aerodynamic diameter (MMAD) of -0.1 m. Ahead of the
PC is an inlet that rejects large particles, with 50% rejection at an MMAD of -2.5
m. As the aerosol stream (14 LPM) enters the PC, 75% of the flow is drawn from
a direction perpendicular to the entrance. Particles - >0.1 ,tm cannot make the
sharp turn and proceed with the minor flow in the original flow direction
through a slot. Then the stream, four-fold enriched in 0.1- 2.5 ium particles,
flows sequentially through appropriately coated glass annular denuders which
remove NH3 , NO2, and 03, a multiple PPD with CIF strips, a Nafion dryer (that
operates with a countercurrent dry air flow) to dry the aerosol and finally a
sandwich filter (containing a Teflon coated filter followed by a CIF) mounted on
a tapered element oscillating microbalance (TEOM [37,38]) which thus
continuously monitors the dry aerosol mass.
Dasgupta et al. [39] described a multiple PPD housed within a 5 cm i.d. x 30
cm length shell. The plates of the 11-element denuder are composed of twin-ply
polyester fabric. The plates are unequal in width. A computational approach to
the collection efficiency expected from such a design was developed. Based on
this, the spacing between the plates was adjusted according to the widths such
that each individual flow channel was isoefficient. Wetting liquid was pumped
into each element resulting in a wetted area in excess of 0.1 m2. The liquid was
withdrawn from 2 V-grooves at the bottom of the denuder. The device closely
followed theoretical expectations and removed >95% of influent SO2 at a
sampling rate of 50 standard liters per minute (SLPM) at an ambient pressure of
680 mm Hg. Loss of 0.4 Jm MMAD particles was only -0.2%.
It is worth noting here that diffusion denuders are increasingly being used
for purposes other than atmospheric analysis. Lewis et al. used a diffusion
denuder technique to determine that 99% of the nicotine in cigarette smoke is in
the aerosol, rather than the gas phase [40]. Vassilev et al. have used diffusion
denuders to determine reaction isotherms [411.

5.2.2 Automated collection and analysis of denuder samples

5.2.2.1 Thermodenuders
The ability to perform collection and analysis in a coherent integrated auto-
mated manner is the hallmark of a technique that others want to use. The long
history of the use of sampling for organics through a sorbent bed and subsequent
thermal desorption into a gas chromatograph has logically led to thermo-
denuders-sorbent-coated denuders that are thermally desorbed and reused.
Indeed, as discussed elsewhere in this volume, this is the mainstay of SPME as
well. Thermodenuders were originally developed for the thermally reversible
sorption of gases. Subsequently, speciation between aerosol H2 SO4 and various
sulfate salts was made possible by passing the sample through a controlled tem-
perature tube prior to a denuder. The transit temperature was so adjusted as to

102
volatilize sulfuric acid to the vapor phase and thus remove it by the denuder
while the other aerosol sulfur compounds passed through. The previously men-
tioned Krieger-Hites Gatling Gun design [10,11] is configured as a thermo-
denuder.
However, for most small inorganic gases, the thermodenuder approach is not
very attractive because of (a) the difficulty of recovering the collected gas in an
unambiguously identifiable form, (b) interferences from other analytes, (c) large
power requirements in the field, and (d) significant cool down periods before the
denuder can be reused. Since the time of the last review [1], only a handful of rele-
vant articles have appeared: a review [42], a comparative evaluation of different
NH3 monitors [43] (including a thermodenuder based instrument; only a continu-
ous wet effluent denuder based instrument was found suitable for the intended
task), the use of thermodenuders in studying particles present in internal combus-
tion engine exhaust [44,45], a study involving continuous monitoring of sulfate
aerosols and gaseous SO, (largely using established principles) [46], and a tech-
nique for collecting HCHO [47].
The last named technique collects HCHO as a derivative of 2-hydroxy-
methylpiperidine (coated on the walls of a single tube denuder) followed by
off-line thermal desorption and gas chromatography- mass spectrometry (GC-
MS) analysis. The basic problem with most approaches that uses a derivatization
agent followed by thermal desorption is that the derivatization agent does not
generally survive the thermal treatment. Consequently, automated reuse is dif-
ficult. This trait limits the use of derivatization-based thermodenuders and is
not likely to change in the future.

5.2.2.2 Automated denuder wall preparationand analysis.

More than two decades ago, in a terse account, Bos [48] reported coating a glass
denuder in-situ with a methanolic citric acid solution, drying it in situ with clean
air, sampling ambient air for a preset period of time to collect ammonia, washing
it with water and sending the wash solution for analysis to an integrated
air-segmented flow analyzer utilizing indophenol blue chemistry. The whole
sequence was then repeated. This feat has not been repeated as elegantly since.
Gdcs and Ferraroli [49] suggested a most interesting, if somewhat unortho-
dox, way of measuring gases. A stainless steel tube (50 cm, 0.80 mm i.d.) is used
as the denuder. The denuder, located within the loop of a 6-port valve, is initially
filled with 20 mM H202 . The tube is flushed with N2 for 30 s; this drives out most
of the liquid but leaves a thin film of H202 . An air sample containing SO, is next
sampled, typically for 6 min and at flow rates < 300 ml/min whence the analyte
is quantitatively captured as H2SO4 . The H202 reservoir liquid meanwhile
recirculates through a conductivity detector. When the denuder is brought back
to the liquid stream by switching the valve, the collected H2SO4 is swept into the
detector and the signal is registered. One litre of fresh absorbent can be used for
several days until the background increase is considered high enough to warrant
replacement. I presume that if an anion exchanger or a mixed bed exchanger was

103
put in line, it could be used for much longer periods. Conductivity detection is
not specific, and it is to this end that the authors have shown considerable
chemical ingenuity. The air sample is drawn through a 0.2 M sulfamic acid
solution that removes any NH3 or HCl. Nitrogen dioxide is removed by the
sulfamic acid, forming nitrogen. Weaker acid gases like CO2 are not absorbed by
the HO,. I estimate the limit of detection (LOD) to be - 1.5 parts per billion by
volume (ppbv) SO2 . The effective integration time in such a system will be
dictated by the equilibration time of the sulfamic acid solutions to the changing
SO2 concentrations. The choice of this pretreatment solution volume (which is
obviously subject to evaporation from the passage of the sample air) represents a
compromise between the continuous operational period and the response time.
More recently, a wetted single-tube glass denuder arrangement was reported
by Luo et al. [50] which is vaguely similar to the Gacs-Ferraroli approach. A
glass tube, 1 mm i.d., was coated in situ with a film of glycerol/phosphoric acid,
then air was sampled to collect NH3. The collected NH3 was next washed off
directly with reagents suitable for initiating the indophenol blue reaction.
Finally, the blue product was measured; all this was carried out with a single-
pump, sequential injection analysis system. Although the particular exposition
may not have been exemplary in all respects (e.g., collection of NH3 was less than
that would theoretically be expected, probably because the coating solution did
not form an uniform film on the untreated glass walls-less than theoretical
collection efficiency was also the case with the GAcs- Ferraroli approach), the
concept holds promise for a variety of applications and a variety of detection
systems.

5.3 DIFFUSION SCRUBBERS

A denuder in which direct gas-liquid contact takes place (a wet denuder) is

conceptually simple. However, since the liquid phase is necessarily uncontained,
it is more difficult to put this into practice experimentally. Diffusion scrubbers
are membrane-based denuders in which the sample gas flows on one side of a
membrane and the receiver liquid flows in a contained manner on the other side,
usually in a countercurrent fashion. It is obviously possible to build such devices
in planar format, similar to parallel plate denuders. However, tubular mem-
brane devices with a tube-in-shell design, with the membrane tube contained
inside a jacket tube, have been used more commonly. Basically, two geometries
have been used. In the first, the sample gas flows within the membrane tube and
the scrubber liquid flows in the space between the membrane tube and a closely
fitting jacket. This is the geometry used for the very first reported diffusion
scrubber (DS) [51]. In the other geometry the scrubber liquid flows within the
membrane tube and the sample gas flows around it [521. Generally, membrane
tubes of larger i.d. are used with the first geometry and smaller tubes (down to
hollow fiber size) are used with the latter. The interior liquid volume in the
second case is often further reduced by insertion of a filament.

104
 DS devices are easily subdivided into different classes depending on the
nature of the membrane: (a) hydrophilic membranes such as Nafion® present a
wet surface to the sampled gas that readily captures water soluble gases, (b)
porous hydrophobic membranes such as porous poly(tetrafluoroethylene)
(PTFE, Gore-Tex, expanded PTFE (ePTFE), etc.), porous polypropylene
(CelgardT ', Accurel®, Metricel M etc.) in which the analyte gas molecules diffuse
through the pore space into the absorbing liquid on the other side, (c) nonporous
thin-walled membranes such as silicone or Teflon AF in which the analyte gas
molecules permeate through the membrane to be taken up by a suitable
absorber liquid on the other side. Asymmetric membranes, where very thin
nonporous skin is present on a porous support structure, constitute a subclass of
the last group. Although there are many industrial applications of asymmetric
membranes, I know of only one application of such a membrane for trace gas
collection and analysis. This, too, was a custom-made membrane [53].
With appropriate choice of sampling rate, dimensions and scrubber liquid, it
is possible to attain quantitative or near-quantitative collection of desired
analyte gases above for a and b type devices. In general, membrane transfer rate
for type c devices is much lower and quantitative collection is not attained at
practical flow rates. The intrinsic response time of these devices of the different
types also vary (where chemical kinetics to form the product that is detected and
liquid transport time to the detector are not limiting factors). Whereas type a
and c devices involve transport through the membrane matrix, type b devices
involve gas phase diffusion through pore space and this latter process is much
faster. Of course, if it is possible to build membrane devices with very thin
membranes, then type a and c devices can also have very fast response.
Collection efficiencies of DS based devices as a function of sampling rate
show an exponential dependence on the reciprocal of the sampling rate as in the
Gormley-Kennedy equation (Eq. (5.1)). Methods of numerical evaluation are
available for different geometries [1]. In essence, the analyte mass collected
initially increases linearly with sampling rate because collection efficiency is
(nearly) quantitative. Then the slope begins to decrease. At very high flow rates,
the analyte mass collected becomes virtually constant, independent of flow rate.
Obviously, if possible, there are particular advantages to operating in two
different flow regimes: (a) where collection is quantitative and (b) where the
total amount collected is flow-independent. Actual practice is typically dictated
by sensitivity needs.

5.3.1 Ion exchange membrane DS devices

Much of the initial work with diffusion scrubbers were conducted with ion
exchange membrane DS devices [1]. Of ion change membranes, only a cation
exchanger, Nafion®, is commercially available in tubular form in various
diameters. Both anion and cation exchange membranes are available in planar
format from various vendors. Early work on gas collection was carried out with

105
ion exchange membranes for collecting characteristic ionogenic gases (e.g.,
Nafion® was used with dilute H 2SO4 as the DS liquid for collecting gaseous NH 3
as NH4 , or a radiation-grafted custom anion exchange membrane tube with a
K2 SO4 solution as the DS liquid was used for collecting HNO, as NO,-) [51,54].
However, this mode of operation can lead to large integration times because the
analytes occupy ion exchange sites and must be displaced by similarly charged
ions in the DS liquid. As a result, the collected analyte ions move essentially in a
chromatographic fashion along the membrane. Clearly, the greater the affinity
of the analyte ion for the ion exchange sites, greater is the effective integration
time. It is possible to ameliorate this to a degree by using DS liquids that also
contain similarly charged ions with high affinities for the ion exchange sites
(e.g., C104- or Ca 2+) and at high concentrations. Blank contamination, an issue
with any reagent in high concentration, then becomes the limiting factor. Note
that Donnan exclusion prevents like-charged ions from penetrating a membrane
matrix; NO,- is not, for example, transferred through a cation exchange
membrane. For very weakly ionized species, this limitation may not hold; H2S,
for example, appears to be transported through base-treated Nafion®
membranes, but the reproducibility of such behavior has not been clearly
established [55].
As a result, the principal use of ion exchange membrane DS devices has thus
far been in the collection of small polar nonionic molecules, like HCHO or H2 02
as described in the next section.

5.3.1.1. Collection and analysis of formaldehyde with a Nafion membrane DS

(NMDS)
Formaldehyde is the most abundant gaseous carbonyl compound in the ambient
atmosphere, ranging in concentration from 10-100 ppbv in polluted urban air.
Fan and Dasgupta [56] described an NMDS-based method for the continuous
measurement of HCHO. Formaldehyde is present as methylene glycol,
CH 2 (OH)2 , in aqueous solution. However, the hydration of HCHO to CH 2 (OH)2 is
not instantaneous. The rate of the hydration process is both acid- and base-
catalyzed. With a porous membrane based DS, the uptake of HCHO (thus the
collection efficiency) increases significantly when 0.1 M H2 S04 is used as the DS
liquid instead of water [57]. Nafion, however, is itself very strongly acidic, and
the collection efficiency for HCHO appears to be almost the same whether water
or acid is used in the DS [56]. In the Fan-Dasgupta instrument, water is pumped
through the DS and the effluent is merged with a stream containing cyclo-
hexanedione (CHD) and ammonium acetate. The streams are mixed and allowed
to react for -2 min as it passes through a PTFE coil (woven in a Serpentine-II
format [581) immersed in a bath maintained at 95°C. The fluorescence of the
dihydropyridine derivative formed is measured by a commercial filter fluoro-
meter equipped with a 370 nm phosphor coated Hg pen lamp as the excitation
source. The CHD chemistry is not selective for HCHO if the reaction is carried
out for an extended period of time. However, it was shown in this work that if the

106
 6.00-

4.00-

3.00-

U 2.00-

1.00-

0.00-
I 3I 4I
10.00 20.00 30.00 40.00
Hours from 0000, June 1, 1995; Boulder, CO
Fig. 5.1. Formaldehyde data obtained by a Nafion membrane diffusion scrubber collector coupled
to CHD derivatization and fluorometry (NMDS-CHD TTU) and a tunable diode laser absorption
spectrometer operated by the National Center for Atmospheric Research (NCAR).

reaction is carried out for only a short period of time, the solution phase reaction
is > 300 times more selective for HCHO than the nearest congener CH3 CHO. For
a gaseous sample, CH3 CHO will be less efficiently collected than HCHO based
both on diffusion and solubility issues. In the majority of ambient applications,
HCHO concentrations far exceed that of CH 3 CHO (except in countries like
Brazil where ethanol is used as a major fuel for motor vehicles), this approach is
then nearly specific for HCHO. The instrument has a reported (S/N = 3) LOD of
8 parts per trillion by volume (pptv, 10 -12 atm). In a blind intercomparison study
conducted by the National Center for Atmospheric Research (NCAR) for various
HCHO measurement techniques, the instrument produced ambient data
essentially identical to the referee instrument, a tunable diode laser absorption
spectrometer (TDLAS) operated by NCAR [59]. Figure 5.1 shows the correspon-
dence of ambient HCHO measurement results. (We note in passing that more
often than not, this degree of agreement is not observed in intercomparison
studies between two otherwise perfectly good instruments, due to a variety of
reasons.) A similar instrument was built by researchers at Purdue University
and used to measure formaldehyde concentrations during the 1998 polar sunrise
experiment [60]. A more detailed description of this instrument appears
elsewhere [61], this latter paper also describes the use of the instrument to study
HCHO chemistry above a forest canopy. More recently, the instrument has been
used to measure HCHO coming from the snowpack during the ALERT 2000
experiment (Alert, Nunavut, Canada) [62].
A newly designed instrument using the same basic collection principle and
reaction chemistry has been more recently reported [63]. A Nafion tubular
membrane (0.75 mm i.d., 0.90 mm o.d.) is concentrically disposed within a 2.7
mm i.d. PTFE tube in a thermostated enclosure (Fig. 5.2). The complete instru-

107
 vV

Fig. 5.2. Diffusion scrubber and housing. U1, U2, U3: PVDF 1/4-/ tubing union fittings; SI:
Sample inlet; ZA: zero air inlet; AP: to air pump; T1, T2, T3: PVDF tees, E: PVDF elbow; N:
Nation tube; J: PTFE jacket, SS: stainless steel tubes; WI/WO: water inlet/outlet (PTFE
capillaries); RTD: platinum resistance thermometer connected to miniature 3-wire jack M; PF:
Plexiglas frame; C; filler cap; H: heater (top and bottom), HL: heater leads. Reproduced with
permission from Ref. [63] (copyright 2001 John Wiley and Sons).

Peltie,Cooler pL/min

Fig. 5.3. Instrument schematic. LP: peristaltic pump; DS: diffusion scrubber, L: mixing and
reaction coil thermostated at 95°C; AP: air pump; FC: flow control needle valve, V: three-way
solenoid valve controlled by timer; ZAC: cartridge to remove HCHO from pump exhaust, I:
six-way liquid injection valve; FM: flow meter; MTB: moisture trap bottle; DP: debubble port
(normally closed); N: 23 ga. hypodermic needle (supplementary flow); T: HCHO removal trap.
Reproduced with permission from Ref. [63] (copyright 2001 John Wiley and Sons).

ment is schematically shown in Fig. 5.3. A peristaltic pump aspirates water

through the DS. Note that aspiration, rather than pumping, maintains a
constant effluent flow rate in the face of variable evaporation losses as relative
humidity (RH) of the sample air varies. The aqueous DS effluent mixes with the

108
CHD reagent in a Serpentine-II knit reactor (residence time -150 s, potted in a
low-melting bismuth-tin alloy containing a Pt resistance thermometer and
surrounded by two flexible heaters and thus maintained at 95°C) and then flows
through a liquid core waveguide (LCW) based fluorescence detector [64]. The
CHD reagent solution is kept cold (-7°C) by a small Peltier cooler with water as
the heat transfer agent. This reagent "refrigerator" was incorporated within the
instrument enclosure. The debubble port, normally closed, is provided via a
tee-arm to remove recalcitrant bubbles by flushing with methanol or to wash
and clean the detector. Both the water bottle that supplies the DS and the CHD
reagent bottle are protected from ambient HCHO intrusion by traps that
contain sequentially chromic acid-impregnated silica gel and soda lime. The
exhaust of the miniature air pump used for air sampling was connected to a
timer-controlled three-way valve. When the valve is energized, the pump
exhaust is vented. When it is off, the pump exhaust is passed through the HCHO
removal cartridge to feed this HCHO-free gas to the DS. The total aspiration by
AP (controlled by flow control knob FC typically to 1.1 SLPM) constituted the
airflow through the DS, plus a small amount (-2% of the total) through a 23 ga.
hypodermic needle to the AP inlet by a tee. This design ensures that when valve
provides zero gas to the DS inlet, the airflow is sufficient and no ambient air is
drawn through the sample inlet; rather, a small amount of zero air back-flushes
the sample inlet line. In this instrument, the valve is programmed to sample
ambient HCHO for 3 min and zero gas for 7 min. At the stated sampling rate of
1.1 SLPM (p = 680 mm Hg), the DS collects 70% of the HCHO (at 0.76-2.94
SLPM, -81-50% of the HCHO is collected). At the same mass flow rate but at a
higher ambient pressure, sample flow velocity will decrease and collection
efficiency will increase. The liquid core waveguide (LCW) fluorescence detector
design utilizes a GaN based light emitting diode (LED) emitting in the near UV
and provides a very stable, inexpensive, long-life excitation source that produces
no heat. Two disks of blue plastic, placed on top of the photomultiplier tube
(PMT) window, serve as the emission filter.
The entire instrument is housed in a tower-style personal computer (PC)
chassis, and uses the PC power supply. Figure 5.4 shows the front and side views
of the instrument; Fig. 5.5 shows calibration stability and response to sub-ppb
levels of HCHO measured in a 1999 field study [63]. A similar instrument based
on pentanedione chemistry (instead of CHD) has become commercially available
in a rack-mountable version [65].

5.3.1.2 Collection and analysis of gaseousperoxides with a NMDS

In more recent years, H,202 has received a great deal of attention in atmospheric
chemistry studies because of its central role in the conversion of sulfur dioxide to
sulfuric acid in cloudwater, and to a lesser extent, in rainwater [66]. Hydrogen
peroxide and other related organic peroxides, notably methyl hydroperoxide
(MHP) and hydroxymethyl hydroperoxide (HMHP) are of particular interest
and detailed reviews are available [67,68]. An NMDS unit using pure water as

109
Fig. 5.4. Front and side views of HCHO measurement instrument. Thermostated DS unit is
mounted on the side.

.2)
0)

0.00 100.00 200.00 300.00 400.00 500.00

Time, min
Fig. 5.5. Typical instrument stability for an overnight run with the instrument sampling 4.6
ppbv HCHO. The inset shows the lowest concentrations encountered during a field measure-
ment campaign in Atlanta, GA, in 1999; heavy rainfall occurred during the previous evening
resulting in relatively clean air; peaks are 420 and 500 pptv, respectively. Reproduced with
permission from Ref. [63] (copyright 2001 John Wiley and Sons).

DS liquid can collect H20 2. Typically, the collected H202 is determined fluoro-
metrically. A nonfluorescent substrate (typically a p-substituted phenol) is
oxidized by H20 2 in a reaction mediated by a peroxidase enzyme (typically
horseradish peroxidase, HRP) to form a fluorescent product [69]. Genfa and
Dasgupta [70] showed that hematin, an inexpensive product from animal blood,

110
 RC
AAW ?e_: 325
2000x0.3 ?,:m >410
mm

MDS

Fig. 5.6. Single line hydrogen peroxide measurement instrument. H: Ammoniacal hematin
solution; P: pump; R: membrane reactor containing porous polypropylene membrane, M,
suspended over p-cresol, C; NMDS: Nafion membrane diffusion scrubber; RC: reaction coil, D:
fluorescence detector. Reproduced with permission from Ref. [71] (copyright 1992 Elsevier
Science).

can be used as a very effective peroxidase substitute. This led to a simple single
liquid flow channel system for the measurement of gaseous H2 02 [71]. The
instrument is schematically shown in Fig. 5.6. Ammoniacal hematin solution is
pumped into the NMDS via a porous membrane device contained in an enclosure
saturated with p-cresol vapor. The p-cresol is thus dynamically introduced into
the stream (a premixed reagent containing hematin and p-cresol shows a rapid
increase in blank). The collected H2 0 2 reacts with the mixed reagent in a
reaction coil and the fluorescence is then detected with a commercial filter
fluorometer equipped with a Cd-lamp source emitting at 325 nm. Illustrative
performance appears in Fig. 5.7. A similar instrument was flown on an airship
during the summer of 1994 and an extensive account of the results was given
[72]. A fully self contained, self-calibrating H2 02 instrument based on HRP
mediated oxidation of 4-ethylphenol has been extensively used and flown aboard
motor gliders and balloons in exemplary expositions of optimum instrument
design for atmospheric chemistry studies [69,73].
In further work conducted on hydrogen peroxide and HMHP (the adduct of
HCHO and H2 02 , which the authors contend cannot be distinguished from
H2,0), Li and Dasgupta [74] designed a compact field deployable instrument.
The oxidation of phenol derivatives as used in previous work require excitation
in deeper UV than currently available solid-state sources can provide. The new

Fig. 5.7. Diagram for gas phase H20, for the system in Fig. 5.6.
The inset shows S/N performance for 36 pptv H202 . Reproduced
with permission from Ref. [71] (copyright 1992 Elsevier
Science). -

111
Fig. 5.8. (Top) Spectroscopic details relevant to
the HO 2 , measurement system. (a) Excitation 0
spectrum of thiochrome (,, 440 nm); (b) Emis- C
sion spectrum of GaN LED; (c) Emission spect-
rum of thiochrome (ex 370 nm); (d) Absorbance
due to the reaction medium, 5 cm pathlength, E B
right ordinate; (e) Absorbance of 1 disk of m ·f;
colored plastic used as emission filter. (Bottom) L
Transversely illuminated fluorescence detector
schematic. T: Entrance PEEK tee; LI: liquid
inlet; FO: silica fiber optic; OF: colored plastic
optical filter; PMT: miniature photomultiplier
tube; AF: Teflon AF 2400 liquid core wave-
guide; SSJ: stainless steel jacket, LED: UVW Li
emitting light emitting diode; PD: photodiode u
for monitoring source light intensity; U: outlet LED r f
5SJ OF
union; W: waste. Reproduced with permission
from Ref. [74] (copyright 2000 American
PD AF T PO
Chemical Society).

chemistry relies on the hematin-catalyzed oxidation of nonfluorescent thiamine

(vitamin B1) to fluorescent thiochrome by H2 02 ; in the liquid phase, this
reaction is 25-fold more selective for hydrogen peroxide than its nearest alkyl
hydroperoxide congener, MHP. The detector design is the same as that for the
HCHO instrument described in the preceding section. The unique optical
absorption characteristics of the hematin reagent itself provide significant exci-
tation light filtering in addition to the colored plastic emission filters (Fig. 5.8).
The LOD for liquid phase H2 02 is 11 nM. The temperature controlled NMDS is
used to collect gaseous HO22 with near-quantitative efficiency with an S/N = 3
LOD of 15 pptv. The system responds to H2 02 and HMHP but not to MHP.
Remarkably, the use of an NMDS prevents interference from SO2 since anions
do not penetrate through the negatively charged membrane.
In as yet unpublished work, a method has been developed to measure both
H,202 and MHP. MHP is significantly less polar than H2 02 and is collected only
poorly by a NMDS. The H2 02 measurement channel is exactly the same as in the
above instrument. In the MHP measurement channel, an ePTFE membrane)
tube is used to collect both peroxides. The effluent is directed through a small
reactor packed with MnO2 that selectively destroys H,202. The MHP is then
determined using the same thiamine chemistry but using HRP as catalyst
(which unlike hematin, allows the determination of MHP). Two separate
detector cells, with their individual LED excitation sources, are used. The output
optical fibers from both cells lead to the same photomultiplier tube. Because
solid state light sources can be turned on and off very rapidly and reaches stable
output almost immediately, cell 1 and 2 LEDs are alternately turned on (at -5 s

112
intervals) and produces reading for each channel in turn. Software sorts and
processes the signals. The method has been extensively field tested during the
NEO3 PS study in Philadelphia, PA in the summer of 2001 [75].
The advent of LCW based detectors not only makes possible simple fluores-
cence detectors, they allow the ready fabrication of even simpler very sensitive
chemiluminescence (CL) detectors using inexpensive small area photomultiplier
tubes [64,76]. A recent paper [77] makes use of NMDS collection of H2 02 .
Luminol and Co(II) are added directly inside a LCW CL detector. The linear
range is at least up to 100 ppbv H202; an S/N = 3 LOD of 25 pptv H,,202 is
reported. The interference equivalent from ozone is below 0.1% and that from
NO2 is negligible.

5.3.2. Porous membrane DS devices

Until recently, most DS applications involved porous membrane devices. There

are many different types of porous membranes that are applicable for the pur-
pose and many have been used. In the author's laboratory, porous PTFE
(Gore-Tex), ePTFE, porous polypropylene (Accurel® and CelgardTM , Accurel®
membranes are also made with different polymers, e.g., polyvinylidene fluoride,
PVDF) porous membranes have been used. Gore-Tex membrane tubes studied
have relatively large (2-3.5 gm) pores with a thick wall (400 m), considerable
tortuosity and 50% surface porosity. Over a period of time, the collection effi-
ciency decreases, whether due to deposition of concurrently present particulate
membrane in the pores or more likely, membrane deformation, is unknown.
More importantly, while the pressure tolerance is initially reasonable, the mem-
branes slowly lose their hydrophobicity and leak more easily over time. Accurel®
membranes (0.2 m pores, 0.2-6 mm i.d., 70% surface porosity, low tortuosity)
offer a much better alternative in this regard. Celgard ' membrane tubes are in
T

the hollow fiber class (much work have been done earlier with CelgardT' mem-
branes that range from 0.1 to 0.4 mm i.d., 25 Am wall, 0.02 um pores, no
tortuosity, 40% surface porosity; currently these membranes are available only
in 200-240 Am i.d., elongated pores with 0.03-0.04 x 0.10 gm pore size, 30-50
fm wall, 25-40% surface porosity [78]) and offer significant resistance to wall
leakage.
Except as detailed below, none of the above membranes permits the use of
DS liquids containing dissolved solids. As the solution evaporates, the solids are
left in the pores and cause pore blockage. Even with pure water, nanopore
membranes like Celgard"M show evidence of pore blockage over a period of use.
However, weekly cleaning with a wetting solvent, e.g., methanol, can restore the
membrane. By the same token, porous hydrophobic membranes are generally
incompatible with organic or mixed aqueous solvents as DS liquids. The
membrane generally leaks if wetted by the solvent.
The susceptibility to blockage depends on the pore size, pore tortuosity and
wall thickness. Not all porous membranes may be subject to such blockage. An

113
Fig. 5.9. Electron micrograph of an ePTFE membrane. Reproduced with permission from Ref.
[55] (copyright 2002 American Chemical Society).

electron micrograph of an ePTFE membrane is shown in Fig. 5.9. The "pores" in

this membrane are really the space between the fibrils and are very large. The
membrane can be used only in the mode where a nonwetting liquid is fed with a
slight gravity head and aspirated from the other terminus. The very thin wall
(125 Atm) and the very large "pore" structure render it immune to pore blockage.
In this laboratory we failed to find any decrease in collection efficiency of an
ePTFE DS for H202 after one week of continuous sampling of dry air with a 0.2
M NaC1 solution flowing through the tube. Poreflon tubes (type TB 54, 4 mm i.d.,
500 Am wall, 0.45 Atm pores, 79% surface porosity) have been extensively used by
Komazaki et al. [79-82] to construct DS devices in which the air flows through
the membrane tube. In at least one instance [81], a 100+ M solution of
dissolved solids was pumped through the DS. Since the authors do not describe
any specific loss of collection efficiency with time, one imagines that this
membrane is also resistant to fouling.

5.3.2.1 Porous membrane diffusion scrubbers (PMDS) coupled to capillary

scale analyzers
Because some of the porous membranes are available with very small diameters,
they constitute good gas sampling interfaces for capillary-based analysis sys-
T M
tems. A 3-3.5 mm active length, 100 tm i.d. Celgard fiber was interposed near
one end of the capillary in a capillary electrophoresis (CE) system. The test gas
was sampled through a concentric jacket built around the membrane at 100
mi/min for 2 min. Helium or N2 then swept the sample gas off the jacket and elec-
trophoresis was started. The S/N = 3 LOD were 10 ppbv each for HONO and
HNO3 (direct UV detection, borate electrolyte) and 20, 50, 80 and 440 ppbv for
SO2 , HCl, HCOOH and CH3 COOH, respectively (indirect detection, chromate-
borate electrolyte), using an early generation commercial CE system [83]. Very
recently, Rocha et al. [84] described the use of a bundle of 14 porous polypropy-
lene fibers for collection of ammonia over a 2 h period with subsequent analysis
by CE. Oscillometric conductometry was used for detection.

114
 An 8 mm active length, 400 Am i.d. Celgard" fiber, jacketed in a -2 mm i.d.
PTFE tube was used for sampling ammonia in conjunction with a capillary scale,
electroosmotically pumped, sequential injection analysis system using the
indophenol blue chemistry and absorbance detection. Even with the modest 250
Atm optical path length, it was possible to attain an LOD of low ppb levels for a 4
min sample drawn at 120 ml/min [85].

5.3.2.2 Measurement of ammonia and amines with PMDS devices

Gaseous ammonia is of considerable interest in a variety of situations. It is the
principal atmospheric base; it is vital to know gaseous ammonia concentrations
in the ambient environment. In large-scale animal operations, ammonia levels
can be very high. This leads to hygiene, odor and esthetic concerns. Using a
Celgard membrane for collection, in the past we have developed a technique for
measuring low levels of NH3( g) using o-phthaldialdehyde derivatization chemis-
try and fluorescence detection [86]. Others have adapted and refined this further
and showed that sensitive measurements with good time resolution are possible
[87,88]. With careful thermostating, the latter authors report an LOD of 10 pptv
with a time resolution of 10 min and conclude that with some further work, it
would be possible to lower the LOD further. More recently, these types of instru-
ments, along with wetted annular denuders, have been used to measure concen-
trations of gaseous ammonia across the Tenerife islands to assess the effect of
cloud processing on the marine budget of reduced nitrogen compounds [89].
Even very small levels of ammonia are of concern to the semiconductor
manufacturers because of serious quality control issues. Using an unspecified
porous membrane for collection and the same o-pthaldialdehyde fluorogenic
derivatization chemistry, Park et al. [90] report an LOD of 80 pptv in a semicon-
ductor fabrication facility where the ammonia concentrations varied from 0.5 to
-30 ppbv in the vicinity of one tool to another. The authors also showed how
dramatically contact-hole etching patterns deteriorate at "high" NH 3
concentrations.
Chang et al. [91] examined Gore-Tex and Celgard" fiber-based DS devices
for the collection of ammonia. Gore-Tex was found to produce severe memory
effects and Celgard was selected for further work. The effluent was determined
by ion chromatography (IC) following previously described DS-IC coupling
strategies [92]. With a 10 min sample, loading a 150 /il loop at 20 /,l/min (air flow
rate 1.5 LPM), an LOD of 20 pptv could be attained. With highly purified
methanesulfonic acid used as eluent, the background conductance and noise was
sufficiently low that no preconcentration column was needed. The authors
correctly argue that preconcentration columns should be avoided if possible. It
has also been noted that ammonia is by far the dominant (gaseous) base
(catioinogenic species) in atmospheric samples and chromatographic separation
is not really needed for determining ammonia [931. However, it has been more
recently shown that under slightly different conditions (30 min sampling time,
DS liquid flow rate 70 l/min, sample air flow rate 0.5 LPM, injection loop

115
volume 2 ml), amines at very low levels can also be detected [94,95]. Methyl-
amine, dimethylamine and trimethylamine and ethylamine exhibited LODs of
1.0, 0.5, 5.3, 0.7 pptv, respectively. Ambient concentrations of these amines in
Seoul, Korea ranged from less than LOD to tens of pptv: In this latter work, the
authors used a sheet membrane-based planar DS unit with the sample air
flowing on one side of a membrane and liquid on the other.
Researchers at NEC Inc. have been interested for some time in applying
PMDS devices coupled to an IC for the measurement of trace NH3 . Air samples
ranging in concentration from <1 to 10 ppbv could be measured at 10-min
intervals (minimum) with good reproducibility. A multiple-point monitor was
also developed with two sampling arms. Each arm consisted of a DS, an air
pump, and a valve for choosing between one of two sampling ports. When one DS
was measuring the air sample, the other was on standby and was conditioned
with the next air sample for 25 min. Since the air sample measurement and the
conditioning with the next air sample were carried out concurrently, it took 1 h
to measure at four different points [96]. This was later expanded to 16 measure-
ment ports, four DS units, still coupled to one IC and provided the capability of
making nine measurements per day per sampling point. The setup readily
detected an increasing number of people in a cleanroom, and was able to monitor
an area 40 m in radius [971. Ethanolamine is also a critical component for
semiconductor production processes. When an identical approach to that used
for NH3 was tried to measure ethanolamine using multipoint sampling, the
results were poor. This was found to be largely because of strong adsorption of
the analyte on the long sample conduits. Thereafter, individual DS units were
allocated to each measurement point. Satisfactory results were obtained [98].

5.3.2.3 Planardiffusion scrubbers

Planar DS units were likely first described in detail by Frenzel [99,100]. Except
for minor dimensional differences, very similar devices have been used for a very
long time in flow injection analysis (FIA). FIA applications included dialysis and
gas diffusion where liquids (or gas-segmented liquids) flow on both sides of the
membrane [101,102]. The important point that Frenzel [100] makes about
planar DS devices (in his indomitably enthusiastic fashion, in this very readable
review) is that many more membranes are available in sheet form for use in a
planar format than in tubular form. In a study of seven different porous sheet
membranes (two polypropylene, three PTFE, PVDF, and silicone), for the
transport of ammonia, the best (one of the PTFE membranes) transported twice
as much analyte compared to the worst (silicone). Although he makes a case for
surrogate liquid phase calibrations and details various options for the same, in
my opinion liquid phase calibrations should only be used when gas phase
calibration simply cannot be carried out. There is no substitutefor calibrationsin
the gas phase. Liquid phase calibrations do not account for inlet losses. For the
determination of NH 3, the most sensitive method Frenzel found was stopped-
flow trapping of the analyte in dilute H2SO4 . This was followed by gas diffusion

116
FIA analysis of the trapped ammonium into an acid-base indicator stream
receptor. Despite the complexity of the flow scheme, the system was found
reliable with sub-ppbv LODs. For SO2 , the method of choice involved an alka-
line-ethylenediaminetetraacetic acid (EDTA) absorber in a planar DS and
reacting the collected analyte with pararosaniline-formaldehyde with colori-
metric detection. The LOD was -10 ppbv.
Frenzel refers to the above devices as "Permeation Denuders" even though
the paper deals almost exclusively with porous membranes. I do not believe that
this terminology is appropriate. The dominant transfer process in a porous
membrane is diffusion through the pore space, which is porous on an observable,
microscopic scale. In contrast, permeation is a process that is dominant only
with nonporous membranes (more accurately, membranes which are porous
only on a molecular scale) and which occurs by (i) dissolution in the membrane
matrix, (ii) diffusion across the polymer and (iii) evaporation on the other side
accompanied or followed by uptake by a suitable absorber. The terminology is
long established [103,104]. In a subsequent paper dealing with the use of
Griess-Saltzman reagent for determining NO2 by a planar DS utilizing a sheet
Accurel® membrane, to my considerable happiness, Scheppers et al. [105]
dropped the term "permeation". Using a short (20 mm) receiver channel filled
with the chromogenic reagent that contained optical fibers for direct measure-
ment of the reaction product (as gas was sampled in a stopped liquid flow mode),
the Gas-Diffusion Detection Device allowed a working range of -25-125 ppbv
NO 2 . These devices are the forerunners of true permeation based scrubbers
made from Teflon AF which are able to function as long path optical cells
because of their ability to act as LCWs. These are discussed in a subsequent
section.
Toda et al. [106] described a miniaturized detector for SO 2 based on con-
ductometry, using a configuration akin to a planar PMDS. Two pairs of Pt
electrodes were fabricated on a glass-epoxy substrate through conventional
photolithography and sputtering. On this substrate, a cavity for the flow of an
absorbing solution was formed with a Teflon sheet and a porous membrane. The
absorbing solution, H2 SO4 -H 2 02 , was made to flow through this cavity (2 mm
width and 0.5 mm depth) while the sample gas flowed on the other side of the
membrane. Gaseous SO 2 permeated through the membrane and dissolved into
the absorbing solution. The difference in the conductivity, as measured by the
upstream and downstream Pt electrodes, was used to quantitate the SO2 concen-
trations. Levels as low as 10 ppbv SO 2 could be measured, with a dynamic range
up to 1 part per million by volume (ppmv). The 90% response time was 82 s
(liquid flow rate 10 Al/min). Sensitivity to CO 2 was 3 x 10 -5 of that of SO 2. The
effect of RH changes was not addressed.

5.3.2.4 PMDS device applications:H2 S and HCN

An interesting problem of determining trace levels of HCN in the presence of
much larger concentrations of H2S is addressed by Kuban and Dasgupta [107].

117
There are several selective and sensitive methods for the determination of
cyanide. However, few can be made to work in the presence of large amounts of
H2S. In coke oven gas, H2 S is typically present at concentrations an order of
magnitude greater than that of HCN. In the desulfurization of petroleum, the N
content is relatively much lower and the H2 S:HCN ratio can approach 105. It is
nevertheless necessary to keep a close tab on HCN because of its detrimental
effects on some key processes and equipment - it is also believed to be a key agent
responsible for the corrosion of stainless steel making such a measurement of
considerable practical importance. In the Kuban-Dasgupta approach, the test
gas is sampled by a mini-DS composed of a 10 cm active length of Accurel®
PVDF membrane (1.0 mm i.d., 1.5 mm o.d.), disposed within a -1.6 mm i.d.
PTFE jacket, with the gas flowing at 0.2 1/min through the interannular space.
The internal liquid is NaOH (100 gl/min), which captures both the H2 S and
HCN. The pH of the receptor is then adjusted downstream to a pH between 9
and 10. At this pH, a significant amount of the HCN (pKa 9.21) is protonated but
leaves the H2S almost quantitatively ionized (pKa 6.97). This mixture flows
across a silicone rubber membrane device where the HCN selectively permeates
in to an alkaline absorber maintained in stopped flow condition. After a 10 min
preconcentration, it is injected into a colorimetric determination system for
cyanide based on the Kinig reaction. It was possible to achieve an LOD of 35
ppbv HCN in a matrix containing several thousand ppmv H2 S. Just using a
single stage DS device, whether made from porous PVDF or silicone, one could
successfully analyze HCN in presence of H2 S up to about 2-3 orders of magni-
tude greater in concentration, but no higher. Interestingly, when the super-
incumbent pressure was increased on the DS device, the uptake increased for the
silicone rubber DS (permeation rate is a direct function of pressure) while the
uptake decreased for the porous membrane DS. Presumably, application of
external pressure causes the gas liquid interface in a pore to move further to the
bulk liquid side, thus increasing necessary diffusion distance and reducing
collection efficiency.
The same principle of collecting an acid gas using a porous membrane DS
with an alkaline absorber and then acidifying the absorber liquid to release the
collected gas back to the gas phase in more concentrated form was used by
Tarver and Dasgupta [108] to construct a H2 S measurement instrument. The
instrument was capable of measuring sub-ppbv concentrations of H2 S and
deployed off a mobile platform to monitor H2 S fluxes in West Texas oilfields
[109]. The instrument consists of a 20 cm active length Accurel® PVDF mem-
brane (1.8 mm i.d., 2.6 mm o.d.) in a 3 mm i.d. PTFE shell tube. In the annular
space, 0.1 M NaOH is pumped at 110 ,l/min while air is sampled countercurrent
at 1 SLPM. Acidic gases, including H2 S and mercaptans, are captured. The DS
effluent liquid is acidified by merging 110 l/min 0.1 M H3PO4 to re-liberate the
acid gases-the acidic liquid is pumped through a DS operating in reverse (17 cm
active length of Accurel® PVDF membrane 1.2 mm i.d., 1.8 mm o.d., filled with
1.0 mm dia. PTFE filament to reduce internal volume and put in a 2.1 mm i.d.

118
 B
I

Fig. 5.10. Instrument response to ambient air in the vicinity of a natural gas processing plant.
Peaks B and C are due to H2S and CH 3 SH, respectively, the artifact peak A is much smaller
relative to these levels. The highest B and C peaks correspond respectively to 1.5 ppbv H2S and 7
ppbv CH 3SH. Reproduced with permission from Ref. [108] (copyright 1995 Pergamon Press).

jacket, the acidic liquid flows in the membrane tube, N2 gas flows at 3.6 ml/min
on the outside). The liberated gas fills one of the two 2-ml loops of a dual loop
injection valve connected to a packed column gas chromatograph equipped with
a sulfur-selective flame photometric detector. On a Chromasil 310 column, the
dominant compounds, H2 S and CH3 SH, are easily separated and an injection is
made every 2.5 min. The LOD for H2 S is less than 200 pptv. Ambient mercaptan
concentrations are generally lower than this even in the oilfields but are signifi-
cant near natural gas processing plants, as shown in Fig. 5.10.
A more recent DS-based entry to the measurement of H2S is a small portable
instrument (4.5 kg) that is operated in the field from a 12 V battery. The instru-
ment can produce one measurement every 30 s and boasts of an LOD better than
100 pptv [55]. Air is sampled by an ePTFE membrane based diffusion scrubber
and collected into an alkaline fluorescein mercuric acetate (FMA) solution flow-
ing under a controlled and constant pneumatic pressure. The collected sulfide
quenches the fluorescence, which is measured with a miniature blue LED-
photodiode based fluorescence detector. The reaction is nearly selective for H2 S.
All flow control, signal acquisition and interpretation is carried out via a mini-
notebook PC. The instrument provides for self-calibration and zero functions
using an on-board permeation tube enclosed in a thermostated block, at any pre-
programmed desired interval. The system was tested near oilfield operations in
West Texas and a wastewater plant in Louisville, KY. The data showed good cor-
relations with a concurrently operated lead acetate tape based commercial sam-
pler, with a response speed and time resolution much better than that of the
latter instrument. Figure 5.11 shows the schematic layout of the instrument.
The instrument operates in three modes, zero, calibrate and sample. The status
of valves V1 and V2 govern these modes. Only a maintenance flow through the

119
Left: Fig. 5.11. H2S measurement system schematic. P1, P2: air pumps; R: relay; IC: iodized
charcoal H2S removal columns; MFS1, MFS2: mass flow sensors; TE: thermostated enclosure;
TEC: thermal equilibration channel; PT: H2S permeation tube; F: inlet filter; V1, V2: three-way
solenoid valves; WS: water saturator constituting ofjacketed Accurel porous membrane PP fed
on the outside by water bottle WB; DS: ePTFE diffusion scrubber; T: mist trap; H: in-line heater
(prevents condensation); C: compressor; PS: pressure sensor; PC: pressure controller; RB:
reagent bottle; V3: manual on/off valve; LFR: liquid flow restrictor; D: fluorescence detector;
WC: waste container. Reproduced with permission from Ref. [55] (copyright 2002 American
Chemical Society).
Right: Fig. 5.12. Effect of sample gas flow rate on the response. The ordinate is the ratio of the
FMA decomposition in AM per ppb H2S (a) silicone DS (0.41 mm i.d. 0.61 mm o.d., 870 mm long,
2.69 mm i.d. jacket), 82-ppbv H2S, liquid flow 100 Al/min, right hand ordinate scale applies; (b)
NaOH-treated Nafion DS (0.41 mm i.d. 0.58 mm o.d., 700 mm long, 2.69 mm i.d. jacket), 11 ppbv
H2S, liquid flow 120 l/min; (c) ePTFE DS (1.02 mm i.d., 1.04 mm o.d., 600 mm long, filled with
0.53 mm dia. monofilament, 2.69 mm i.d. jacket), 5.5 ppbv H2 S, liquid flow rate 200 L/min; (d)
Teflon AF DS (1.07 mm i.d., 1.27 mm o.d., 16 mm long, 250 ppbv H2S, 10 min stopped flow.

permeation tube chamber is applied during sample and zero periods. The liquid
reagent container has a pressure sensor and this controls the attached compres-
sor to maintain a constant pressure. Several different types of membranes were
tried as DS units and the relative response of three membrane types are shown
in Fig. 5.12. Note that there is no flow rate dependence for a poor collector. For
an active membrane length of 70 cm, the equivalent quantitative sampling rates
for Teflon AF, silicone, Nafion and DS units were 0.5, 2.7, 135 and >1200
ml/min. Figure 5.13 shows how the upper applicable limit varies with the
reagent concentration and typical instrument output.

5.3.2.5 PMDS devices coupled to ion chromatographs

Applications for the specific determination of ammonia and amines with an IC
have already been described. There are more general applications. IC has had a
long history of association with atmospheric analysis, and especially with the
determination of ionogenic atmospheric trace gases [1].

120
 4
a
0

8
8
0s

2.05
F ' I r I '
0 50 100 150 0 2000 4000
Hydrogen Sulfide Concentration, ppbv Time, s

Fig. 5.13. (a) Calibration curves obtained with three different FMA concentrations, 35 cm
filament filled ePTFE DS; (b) multipoint calibration with a 5 AM FMA absorber, 31 cm
filament-filled ePTFE DS; concentrations in ppbv H2S are indicated; liquid flow rate 200 Al/min
in this and (a); (c) low level measurement with a 60 cm filament-filled ePTFE DS, 1 AM FMA;
liquid flow rate 130 l/min. Reproduced with permission from Ref. [55] (copyright 2002
American Chemical Society).

A PMDS-IC was used for the collection and determination of gaseous NH3
and HNO, at ppbv levels using purified water as the DS liquid [110]. For
validation, the gaseous HNO 8 concentration was measured by the impinger
trap-IC method and/or by a reduction (to NO)-CL method. The gaseous NH,
concentration was measured by the impinger trap (boric acid solution)-IC
method. The DS collection efficiencies were more than 97.4% for HNO3 and
>92.3% for NH3 for a relative humidity of 0-70%. Remarkably, the relative
standard deviation (RSD) values were less than 1.1% (n = 5) for successive
collection and analysis. It was confirmed that this system determines gaseous
HNO3 and NH3 correctly in the presence of NH4 NO3 particles.
Recently, Suzuki [111] gave an excellent account of determining low-MW
organic acid gases in air using IC. A particularly long DS (2.2 m porous PTFE, 1
mm i.d., 2 mm o.d., 0.1 Am pores, 60% surface porosity, put in a 3 mm i.d. PTFE
jacket tube and configured as a large diameter coil) was used to attain 92-98%
collection of formic, acetic, propionic, vinylacetic and methacrylic acids at a flow
rate of 1 LPM. The collection efficiencies for isobutyric and n-butyric acids
approached 90%. A clever valve arrangement switched the organic acids from
the much more strongly held inorganic ions which were made to migrate in the
opposite direction in the precolumn after valve switching. With a DS liquid flow
rate of 100 gl/min and a 500 Al sample loop, the LODs of the straight chain
C1-C3 acids ranged from 200-300 pptv. The analytical cycle was 20 min. Some
organic acids were unresolved under the conditions utilized. It is likely that with

121
Fig. 5.14. Automated continuous simultaneous measurement system for monitoring trace acidic
and basic gases in the atmosphere by using a diffusion scrubber coupled to an ion chroma-
tograph. AF, air purifier; F, filter; LT, liquid trap; AD, air drier; MFC, mass flow controller; AP,
air pump; V1, two-port valve; V2, V3 and V4, three-port valves; STD, standard solution; PP,
plunger pump; CC, concentrator column; SC, separator column; CD, conductivity detector.
Reproduced with permission from Ref. [80] (copyright 1999 Royal Society of Chemistry).

a little longer analytical cycle, gradient elution and proper choice of a separation
column, it will be possible to achieve a more complete separation; the method
holds much promise. Evaporative loss from the DS, effects of variations in
sample RH and jacket adsorptive losses were found to be very minor.
Komazaki et al. [80] have also developed a generally applicable ionogenic
trace gas analyzer using a PMDS-IC. This instrument simultaneously collects
the major inorganic trace gases (HCl, HNO3 , SO2 , and NH3 ) using a Poreflon
tube based DS (50 cm active length, vide supra) and an air sampling rate of 1
LPM. The instrument is schematically shown in Fig. 5.14. The liquid flow
through the DS does not occur continuously. The liquid holdup volume of the DS
is about 4 ml and it is gravity-fed from a water reservoir itself fed from an
ultrapure water (UPW) generator. At the end of a 52 min sampling cycle, first
the anion preconcentrator and then the cation preconcentrator are cleaned with
UPW. Then the plunger pump aspirates a total of 12 ml through the DS, taking
in all the collected analyte as well as filling it with fresh UPW, and then loads the
preconcentration columns. The anion concentrator column is followed by the
cation preconcentrator column (both 10x4 mm i.d.) in series; they are loaded at
2.0 ml/min over 6 min. Chromatography is carried out on 2 mm i.d. columns
using a 0.1 ml/min eluent flow rate in each channel, with a total chromato-
graphic cycle time of 1 h. Under these sampling conditions, the collection of the
target gases is essentially quantitative and allows -10 pptv LOD for each gas
even with nonsuppressed IC. No significant NO 2 interference was found in
measuring HNO3 . The instrument was extensively field-tested; the cross corre-

122
lation of the SO 2 data between the described instrument and a pulsed
fluorescence based SO2 analyzer was excellent, with a slope of 1.02 and a linear r2
value >0.98. Collected SO 2 produces both sulfite and sulfate peaks, which are
summed. However, at low SO2 concentrations, virtually all the collected S is
oxidized to sulfate. The instrument provides for on-board calibrations, over a 50
d field measurement effort, the RSD of the individual peaks from the standards
were <5%. A more recent paper reports the placement of the DS outside,
without inlet conduits, to prevent loss of gases such as HNO, or HCl [82]. The
instrument has become commercially available.

5.3.2.6 PMDS measurement of gaseous peroxides and othergases with

peroxide as a surrogate analyte
Using a 40 cm CelgardT ' hollow fiber based DS, Stigbrand et al. [112] make a
very convincing case for peroxyoxalate chemiluminescence (PO-CL) as a highly
sensitive method for measuring H2 02 . The method is demonstrably unaffected
by HMHP or MHP. A 10 mM acetate buffer (pH 5.0) is used as the continuously
flowing DS liquid (1 ml/min or 0.5 ml/min, depending on desired measurement
frequency) and fills a 10 1lsample loop which is injected at frequencies up to
120/h into a carefully dried acetonitrile (ACN) carrier containing dissolved
PO-CL reagents. The solution is pumped through opaque tubing to the detector,
which already contains an immobilized fluorophore. Even at this high a mea-
surement frequency, the LOD was 23 pptv. Ambient measurement data over a
22 h period was presented. Were it not for the vagaries of working with an
organic solvent that must be maintained very dry in the field, this method will be
truly attractive.
The first DS-based application for measuring H2 0 2 also utilized a porous
Celgard M membrane [57]. Ross et al. [113] first used such a Celgard-based DS
along with a glass coil collector (using air segmented liquid flow) for sampling
H2 02 . Later de Serves and Ross [114] compared the two and concluded that the
DS-based approach is superior. Komazaki et al. [81] have also suggested a
selective method for measuring H2 02 . A solution of Ti(IV)-4-(2-pyridylazo)-res-
orcinol (PAR) (100 M Ti(IV), 10,uM PAR, adjusted to pH 2.2 with HCl) is used
as the scrubber liquid in the Poreflon porous PTFE scrubbers favored by this
group. H2 02 reacts to form a Ti(IV)-H 2 0 2 -PAR complex but this must be sepa-
rated from the reagent by chromatography to obtain good quantitation. Air is
sampled at 1 LPM for 55 min and following the general batch mode operation
preferred by this group, 3 ml of the DS liquid is withdrawn, adjusted to a pH of
11.9 (whereupon the peroxo complex is formed) and 80 Al of the sample is
injected into a 2 x 150 mm revered phase high performance liquid chromatogra-
phy (HPLC) column with a 10 mM Na 2 B4,O, eluent flowing at 0.4 ml/min with
detection at 508 nm. The peroxo complex elutes rapidly as a sharp peak; the
unreacted PAR elutes as a very broad peak centered at -25 min.
The above system reports the best LOD for H2 02 thus far reported (9 pptv)
but the sampling time is relatively long. Presumably with an organic modifier in

123
the HPLC eluent, it may be possible to elute the PAR quicker and reduce the
necessary cycle time. Moreover, this will also affect the analyte peak beneficially
(assuming the complex is not destroyed by an organic solvent) in improving its
efficiency. However, the method is not without problems. The scrubber liquid is
held at a low pH supposedly to decrease interference from 03 and SO2 . Regarding
ozone, ozone forms H0 by surface reactions as well as bulk reactions.
Deposited carbonaceous particles in an inlet line can be a large source of artifact
H202 upon reaction with 03 [114]. Nevertheless, barring inlet artifacts, low pH
can understandably help reduce 03 interference by limiting the 03-OH- reaction
to form HO2- and thence HO,. The authors report that they do see a signi-
ficantly lower 03 effect than that observed by de Serves and Ross [114] when
tests are conducted at realistically high levels of 03. However, they also argue
that SO2 interference should decrease with decreasing pH because SO2 solubility
in water decreases with increasing acidity. This is fallacious. The kinetics of the
H202-S(IV) reaction increases with decreasing pH. The problem is compounded
by the fact that the maximum SO2 concentration tested by the authors was only
14 ppbv. At this concentration, an equivalent of 0.4 ppbv of H202 was destroyed;
this would represent a very large negative error in measuring typical ambient
levels of H202. To be fair, the authors use DS liquid pre-doped with H,20 in their
experiments, simultaneous sampling of gaseous H202 and SO, will likely show
less interference.
Some time ago, Matuszewski and Meyerhoff described a simple detection
system suitable for the continuous measurement of atmospheric H202 at the
pptv levels [115]. The method involves the use of a PMDS as a collector, with the
effluent being sensed by a Pt electrode detector polarized at +0.4 V vs. the
saturated calomel electrode (SCE). The anodic oxidation current is directly
proportional to the H202 in the gas phase. In addition, it is well known that many
oxidase enzymes will produce an equivalent amount of HO, when they oxidize
their substrates (e.g., glucose, cholesterol, sulfite, alcohol, etc.). Thus, when
appropriate immobilized enzymes are incorporated into the flow arrangement
prior to the amperometric detector (e.g., sulfite oxidase or alcohol oxidase), the
same measurement scheme can be used to monitor volatile substrates (e.g.; SO2
and low-MW alcohols) of these enzymes, continuously at ppbv levels. The
authors also describe the stopped-flow configuration of the DS for
preconcentration.

5.3.2.7 PMDS measurement of gaseous formaldehyde and acetaldehyde

Formaldehyde has also been measured with a porous membrane DS. As with
H202, the first DS based application for measuring HCHO also utilized a porous
Celgard membrane [57]. de Serves adapted this instrument for use in the high
T

arctic [116]. Trapp and de Serves [117] compared the collection and measure-
ment of HCHO by a PMDS followed by on-line fluorometry, and 2,4-dinitro-
phenylhydrazine (DNPH)-traps followed by off-line high performance liquid
chromatography-UV detection. The time resolution for the DS instrument was

124
5 min while the DNPH-trap samples were collected over 30-60 min. The
measured concentrations ranged from the LODs (0.045 ppbv for the DS, 0.1
ppbv for the DNPH-traps) up to 2 ppbv. The two techniques were well correlated
(r2 = 0.80, n = 48) with a slope indistinguishable from unity (1.02 + 0.03).
Komazaki et al. [118] have also used their DS-Liquid chromatograph (LC)
set up described above for H202 for the measurement of HCHO and CH 3 CHO.
2,4-dinitrophenylhydrazine (60.6 pg/ml) and 3% v/v H3PO4 in ACN constituted
the DS liquid (note the lack of obvious problems with the use of such an organic
solvent) and after collection (typ. at 0.2 LPM, which results in essentially
quantitative collection of both the target gases), an aliquot of the DS liquid is
injected into a conventional format reversed phase column with 40% aqueous
ACN as eluent and detection at 360 nm. All the operations are sequenced by a
programmable controller, and automated continuous measurements are per-
formed with a typical temporal resolution of 1 h. The LOD, based on three
standard deviations of the DS liquid blank were respectively, 0.05 ppbv HCHO
and 0.10 ppbv CH3 CHO. Analytical response from the HPLC during a ten-day
continuous measurement was unchanged (RSD < 1.0%). Interferences from 0O
and NO2 were both insignificant. For both HCHO and CH 3 CHO measurements,
concentrations from this system agreed well with those measured by a
DNPH-impregnated silica cartridge method.
The authors applied this method to diluted exhaust from a methanol-fueled
vehicle (MFV) [79]. Higher sampling frequency compared to the ambient
method above was possible due to higher analyte concentrations. HCHO was the
dominant carbonyl in MFV exhaust and aside from CH 3 CHO, some acetone, at
levels equal to or higher than that of CH 3 CHO was also noted. The emitted levels
were much higher during initial start up (cold engine) than normal running
conditions after warm up. The method showed no response to the relatively high
NOx levels that are typically present in auto exhaust. For HCHO, the results
produced by this instrument were almost identical to those produced by a
method involving impinger-based collection followed by manual HPLC analysis.

5.3.3 Nonporous hydrophobic membrane DS devices

Unlike hydrophilic ion exchange membranes, the surface of most hydrophobic

nonporous membrane surfaces (with an aqueous absorber on the other side), do
not generally represent a good sink for the majority of analyte gases. As a result,
their use in trace gas analysis has been limited. They can, however, have
characteristics that make them attractive in certain applications. Unlike porous
membranes, they can tolerate substantial pressure without liquid leakage. This
property alone accounts for a much larger use of such membranes in gas
diffusion flow injection analysis where liquids flow on both sides of a membrane
and backpressure can also be significantly higher [102]. Nonporous membranes
are also immune to blockage of pores from any particle deposition (this does not
address artifacts or analyte losses that can be caused by particle deposition

125
within the DS or in the inlet lines). Evaporative loss of liquids is much lower
than with other membranes.
Van Nugteren-Osinga et al. [53] were the first to evaluate the behavior of
such a system. They used a bundle of three asymmetric membrane fibers (200
mm length x 0.65 mm o.d. x 0.30 mm i.d.) that were exposed in a test chamber
containing CS2 vapor. The internal liquid was ethanol. After stopped flow
sampling for 15 or 20 s, the internal liquid was pumped out and reacted with a
reagent containing copper acetate and diethylamine in triethanolamine. Yellow
copper diethyldithiocarbamate was formed and was measured at 422 nm. With a
20 s stopped flow, the LOD was -500 ppmv.
As outlined earlier, Toda et al. [55,119] have examined different membranes
for the measurement of H2S, including nonporous hydrophobic membranes such
as silicone rubber and Teflon AF. Although not competitive with ePTFE for high
sensitivity and rapid response applications, in the stopped flow mode, low ppb
LODs are readily attainable with silicone rubber DS devices with the same FMA
chemistry [551]. The FMA chemistry is not completely selective for H2 S; a smaller
response is elicited, e.g., by CH 3 SH. Relative transfer rates of different analytes
through different membranes can vary considerably and by using a multichan-
nel membrane collector with different membranes, it seems possible to obtain
measurements for different reduced sulfur gases by matrix deconvolution of the
multidimensional signal [120].
Olson et al. [121] have described the use of a silicone membrane DS for the
real-world determination of HCN in process streams.

5.3.3.1 Teflon AF LCW-based gas sensors

If a membrane provides relatively low gas collection rates compared to alterna-
tives, it may still be of utility if it offers advantages that offset this. Silicone
rubber ranks very high in permeability among polymer membranes. The perme-
ability of Teflon AF, a relatively recently available polymer, approaches that of
silicone rubber for some gases [122,123]. The most unusual aspect of Teflon AF
is, however, its optical properties. It is nearly transparent from the near-UV to
the near-IR and has a refractive index less than that of water with the result that
water (or aqueous solution, for that matter, essentially any liquid) -filled tubes
of Teflon AF behave as liquid filled fiber optics or LCWs. It is immediately
apparent that long path absorbance measurement is possible. Since the polymer
is gas permeable, when filled with a suitable selective chromogenic reagent and
provided with optical fibers at each end, it can become a versatile optical sensor.
A preliminary exploration was conducted [124] by Hong and Burgess. Dasgupta
et al. [125,126] discussed the principles and applications of this very interesting
sensing technology in greater detail.
The sensor has a simple tube in shell construction as shown in Fig. 5.15 with
double tees at each end that accommodate liquid, gas, and light I/O. The typical
operational mode is with stopped liquid flow. The change of absorbance with
time is monitored. When the absorbance reaches some upper limit determined

126
 Reagent inlet Air in
Optical fiber Glass jacket Optical fiber
to photodiode L LED

" Liquid core waveguide

Air out Reagent out

Fig. 5.15. Basic construction of an absorptiometric liquid core waveguide based gas sensor.
Reproduced with permission from Ref. [125] (copyright 1998 American Chemical Society).

by the detector capabilities, the solution in the DS is dumped. It is filled with

fresh reagent, with no more sophistication needed than a gravity based reagent
reservoir and a timed solenoid valve. Tetramethylbenzidine (TMB) is a reagent
that turns yellow upon exposure to chlorine. Figure 5.16 shows the behavior of
such a sensor filled with TMB reagent and monitored with a blue LED upon
exposure to short periodic pulses of Cl2. Note that the absorbance at any time
reflects the cumulative exposure history of the device since the beginning of the
measurement. The slope at any point reflects the instantaneous concentration.
Barring limitations of light throughput, the underlying mathematics indicates
that a smaller bore fiber provides greater sensitivity, other factors being equal.
Decreasing the fiber diameter also permits a reduction in wall thickness without
loss of structural strength. Further, a thin wall also improves sensitivity, as well
as response speed. The authors were able to dissolve the silica from Teflon AF
coated silica tubes to construct a Teflon AF device with a wall thickness of only
18 Atm. The response time decreases with the square of the membrane thickness.
The TMB-C12 reaction is very fast. This device showed a subsecond response
time upon exposure to chlorine (Fig. 5.17). The behavior of a very inexpensive

i
0.03- r;
i
Fig. 5.16. Output from a Teflon AF 2400
sensor (15 cm x 279 gm i.d., 533gim o.d.) a .2
I-/
periodically exposed to pulses of chlorine 0.01-
(1.4 ppm, 1 min duration of each expo-
sure). Reproduced with permission from
Ref. [125] (copyright 1998 American 0.0 10.0 200 30.0 40.0 500 60.0
Chemical Society). Time (min)

0.02-

Fig. 5.17. The response of a thin film 4o

tubular sensor to a pulse of chlorine. Re-
produced with permission from Ref.
[125] (copyright 1998 American v0.0 I 2.0
I
4.0
I
6.0
I
80
I
100
1
120
Chemical Society). Time (sec)

127
 E

0
B

Time (sec)
Fig. 5.18. System response to NO2 , 14 cm x 78 7 Am i.d., 1041 Am o.d. device. All flow rates were
200 seem except one blank run, as indicated. The linear r2 values of each temporal response are
indicated and the inset shows that these slopes exhibit excellent linearity vs. concentration.
Reproduced with permission from Ref. [125] (copyright 1998 American Chemical Society).

green LED based sensor to monitor a purple product when Griess-Saltzmann

reagent is shown in Fig. 5.18. This sensitivity is more than adequate to monitor
ambient NO 2 levels. The application of these sensors is not limited to gases but to
any material to which the membrane is permeable, e.g., low MW organics
(benzene, toluene, trichloroethylene, etc.) that often contaminate groundwater.
In accordance with established knowledge on permeation, it has also been
observed that application of external pressure can improve the permeation rate
and hence the sensitivity [127].
Nitrous acid is often present in the atmosphere along with NO 2. Nitrous acid
concentration can be significant relative to NO 2 , especially at nighttime when
photodecomposition of HONO is no longer operative. Griess- Saltzman chemis-
try cannot differentiate between HONO and NO2 . Milani and Dasgupta have
explored the possibility of simultaneously determining both analytes by using
more than one Teflon AF collector with dimensional differences. Presented
results show that differences in either length or wall thickness can permit such
differentiation [128].
Gas permeability may be a boon in gas measurement applications, it may be
a nuisance in others. See Inya-Agha et al. [129] for an interesting means to
decrease the permeability of Teflon AF by an ultrathin polymeric coating.

128
5.4 DROPS AND FILMS AS COLLECTORS AND REACTION MEDIA
FOR DETERMINING TRACE GASES

Between liquids fully contained within membrane enclosures and liquid flowing
through wetted denuders, drops and static films constitute interesting interme-
diates as gas collection devices. The use of drops for gas collection and analysis
(and other applications) has been reviewed [130,131].
5.4.1 Static drops
Liu and Dasgupta [132] suggested the first application for sampling gases with a
drop. They used an electroosmotically pumped capillary scale sequential injec-
tion analyzer to form a 6-18 1ldrop at the tip of a capillary. In one application,
dilute H202 was used as the drop liquid, SO2 was sampled and determination was
accomplished via conductivity measurement. In another application, NH3 was
sampled into a dilute H2SO4 drop and the drop contents were withdrawn and
analyzed by the Berthelot reaction involving the formation of indophenol blue.
The sampling rate was 100-400 ml/min. Only 1-1l aliquots are necessary for
analysis in this system. It was possible to take 17 successive aliquots from an
18-Al droplet and analyze them individually. This allowed mapping of the radial
analyte distribution within the drop. The results showed reasonable agreement
with the diffusive model of analyte collection. The authors also incorporated a
liquid-air meniscus detector between the drop and the analysis system. In this
way, they probed the volume of the liquid evaporated as a function of the relative
humidity (RH) and found that Frossling's equation below is applicable.
rf = ro2 -k (1-RH)(NRe) 0 5 -t (5.5)
where r and rf are the initial and final droplet radii, k is a constant, NRe is
Reynold's number and t is the time of exposure. With a constant initial droplet
size and the same sampling rate, r2 was linearly related to the sample RH. For a
diffusion-based collector, the collection efficiency f exhibits the general form
ln(1-f) = _b + lna (5.6)
Q
where a and b are constants, Q is the sampling flow rate. This was also found to
be applicable. In a very recent exposition, Felix et al. [133] carried out not only
the collection of ammonia but also the indophenol blue reaction directly in a
suspended drop; the attainable LOD would, however, be insufficient to make
ambient measurements.
Pereira and Dasgupta [134] showed how to carry out sequential reaction in a
drop. A 1/4-28 threaded chromatographic union was configured both as the
optical detector and the "drop head". The reaction of 3-methyl-2-benzothia-
zoline hydrazone (MBTH) with formaldehyde followed by subsequent oxidation
by FeCl3 to form a blue cationic dye is well known. An azine is first formed by the
reaction of MBTH with HCHO and this then oxidatively couples to a second

129
molecule of MBTH with the FeCl3 . Reagents must be added sequentially. In
practice, a 100-1l drop was formed at the drop head through an inlet mounted on
the side (liquid flows down the thread to form a hemispheroid drop). Two 1.5 mm
dia. optical fibers, respectively connected to a 660 nm LED and a photodiode, are
diametrically placed across the bottom of the fitting and can measure light
transmission through the drop. Air is sampled at 0.1 LPM for -5 min around the
drop, then flushed with clean air for 2 min. After allowing some more time to
allow the azine formation to be completed, 40 pl of the FeCl 3 -sulfamic acid
reagent (the latter removes any interference from NO2 ) is added through a
second inlet. The absorbance is noted after some further reaction time. An LOD
of -6.5 ppb was attainable in a 5 min sample. There are many means of
improving this LOD, e.g., by increasing the sample time and sampling rate,
kinetic extrapolation of data, etc. Further work has been subsequently con-
ducted for the determination of HCHO both in indoor and outdoor air [135].
Milani et al. [136] used a drop containing acidic malachite green and probed
it at 625 nm as decolorization occurred by sampled SO 2 . Parts per billion levels of
the gas was detectable. Cardoso et al. [137] have described a fluorornetric drop
sensor for H2 S utilizing previously mentioned FMA chemistry. An optical fiber
brings in blue light directly inside a drop of FMA solution that is created by a
PTFE tube as a drop head. An interference filter equipped photodiode is used to
monitor the fluorescence signal. Even with a small drop (-50 gl), diffusive
mixing within the liquid phase is slow. In a typical protocol, air sample con-
taining H2 S is sampled past the drop for 2 min at 0.16 LPM and following this,
the system is flushed with pure air for 10 s. Mixing brought in by frictional
circulation induced by the air flowing rapidly past the drop is clearly noticeable
and beneficial. This is in agreement with other observations made previously
[132]. All gas flow is then ceased. The current initially increases after the
cessation of all flow and decreases again to a minimum and then increases slowly
to a stable plateau. The analytical signal is taken to be the difference between
the photocurrent at the beginning of the experiment and that at the minimum
occurring between 2.3 and 3.5 min, leading to a linear relationship within the
H2 S concentration range of 0-100 ppb H2S.
More recently, Toda et al. [138] developed a small gas detector in which an
electrode and a gas absorber were combined in a small droplet. A gold film was
coated on the outside of a fused-silica capillary, and a silicone rubber film then
covered the Au film for insulation. When the capillary is cut, a micro-ring
electrode is present on the tip. A small droplet is formed at the tip and gas is
absorbed in it. The absorbed component was measured amperometrically by the
microring electrode in contact with the droplet solution (1 pl). The system was
characterized using SO2 as the test gas.
5.4.2 Falling drops
Liu and Dasgupta [139] carried out several novel analytical experiments
centered around a liquid drop. Animated examples are illustrated in Liu's

130
Fig. 5.19. Schematic diagram of the falling drop
t
chlorine detection unit: h is the vertical distance
between the tip of the droplet forming capillary
and the centerline of the optical fiber. Repro-
duced with permission from Ref. [139] (copyright
1995 American Chemical Society).

website [140]. One of the more intriguing experiments involves continuously

falling drops (liquid is pumped through a capillary at a rate that a drop falls
every -60 s) while gas is sampled around it. Figure 5.19 shows the general
experimental arrangement and Fig. 5.20 shows the absorbance signal during the
lifetime of a drop containing TMB as 400 ppb chlorine is sampled around it. The
yellow oxidation product of TMB is monitored by the absorption of blue light.
Note that until about -10 s, the drop has not grown sufficiently in size to come
between the light source and the detector optical fiber. From this time to -35 s,
the drop not only grows but it intercepts light between the source and the
detector and the absorbance also increases continuously. Past this "mid point"
in the life cycle of the drop, the "light gathering and focusing behavior" (i.e., its
behavior as a lens) of the drop begins to dominate and absorbance decreases. At t
> 45 s it becomes clear that more light is reaching the detector fiber than it
received without any drop being present. This continues until -68 s, when the
drop falls. The reproducibility of drop formation and that of the measurements
as a whole are striking. Figure 5.20 does not actually show the trace from a single
drop but constitutes of four separate drop traces, superimposed on each other.
Figure 5.21 shows the drop traces from three sets of four drops each, from three
separate exposure regimens. The optimum point of measurement (where the
drop traces differ most as a result of Cl2 exposure) is just before the drop falls (at
the minima of the absorbance trace, this is also the time where the detector sees
the maximum light and as a result has the lowest noise. These traces also clearly
show that the minima values between the blank and the 440 ppb trace, and that
between the 440 and the 880 ppb trace are equidistant, i.e., the system exhibits
Beer's law behavior. The reproducibility of the blank minima is tantamount to a
noise equivalent of -1 ppb Cl2 . Obviously, the LOD attainable by this very
simple and robust system consuming very little reagent, is very attractive. A

131
6 N C12
-
D
Q - 440 ppb C2
(I
C 880 ppb Ci2

M
in

0 20 40 60 80 0.0 100.0 200.0 300.0

Time, s Time, seconds

Left: Fig. 5.20. Temporal signal profile during the lifetime of a drop.
Right: Fig. 5.21. Drop traces from three sets of four drops each, from three separate exposure
regimens. The optimum point of measurement (where the drop traces differ most) is just before
the drop falls and a new drop begins.

very interesting aspect of falling drops for gas sampling is the effect of the
sample RH. Different sample RH cause different degrees of drop
evaporation-at any given sample RH, there is a linear relationship between the
minima absorbance and the analyte concentration. However, the slopes of the
individual calibration plots are a function of RH. This may initially seem to be a
major obstacle. However, the drop period is a direct function of sample RH. The
drop dislodges when gravitational force related to the mass of the drop exceeds
the adhesional surface forces that hold the drop to the drophead. At the same
input flow rate, when the sample RH decreases, more liquid evaporates from the
drop and the drop period increases. It is easy to determine the drop period, as
evidenced from Fig. 5.20. The falling drop arrangement thus has a built in RH
sensor and this information can be used to make measurements without the
influence of RH (Fig. 5.22). Dual wavelength absorbance detection [141] should
make measurement with falling drops an even more robust technique and
increase the range of applicability.
Sampling by a drop where the drop falls through a static or moving sample
(this is the way not-so-pure drops of rain fall through the atmosphere and get
even less pure) has not as yet been used for routine analysis. Theoretical
treatment of collection of a gas by a moving drop can be formidably complex.
However, a wealth of information is already available, see, e.g., Amokrane and
Caussade [142].

5.4.3 Films

Compared with ovoid or spheroid drops, suspended films may offer greater
surface/volume ratio and hence greater sensitivity. Cardoso and Dasgupta [143]

132
 a
53

2o
a0
0
.0

D
Chlorine concentration, ppmv Observed response, mAU

Fig. 5.22. Left: Variation in response as a function of RH. Right: When RH is computed from drop
period, a single equation can be used for all data regardless of RH.

investigated a very simple arrangement in which a liquid film, containing ca. 14

gl of Griess-Saltzman reagent, is suspended in a U-shaped wire guide held
between two PTFE tubes (one of which serves as the liquid inlet) and two optical
fibers (located orthogonally to the tubes) for measuring the absorbance, being
respectively coupled to an LED emitting at 555 nm and a photodiode. Air is
sampled at a flow rate of 0.13 LPM for several seconds to several minutes. Then
clean air flushes the system for 10 s. Gas flow is ceased and absorbance is
monitored until there is no further change and the difference between the initial
and the final absorbance is taken to be the analytical signal. The LOD is < 10 ppb
for a 5 min sample.
Very small loops that will support a small-volume liquid film to can be used
to sample gases and constitute an ideal interface to capillary electrophoretic
analyzers. The general area has been reviewed [144]. Dasgupta and Kar [145]
formed a wire loop of 2 mm diameter at the tip of the sampling end of a 360 Am o.
d. fused silica capillary with 100 /,m dia. Pt wire. The volume of an aqueous
liquid film formed by immersion and withdrawal was measured to be 880 70
nl. The capillary tip is at the center of the drop and is in fluid communication
with the film. The standard operating procedure consisted of (a) dipping the
sampling head into a Na 2B4 O7 vial (running electrolyte), pressurizing to flush
the capillary with the running electrolyte, lifting the sampling head and dipping
it into the film making liquid, withdrawing it and introducing it into the gas
sampling chamber. Note that there is no significant hydrostatic difference
between the film contents and the detector end of the capillary during sampling.
Air was sampled immediately after the head sealed itself on the sampling
chamber, typically, for 1 min at 0.1 LPM. Following the sampling period, the
head was lifted to a height of 10 cm and maintained in that position for a fixed
period of time to introduce an aliquot of the film contents into the capillary.
Then the head was returned to a fresh Na2 B4,O, vial and HV (+ 15 kV) applied to
begin the electrophoretic run. With H,202 as the film liquid, the detection limit

133
 3VO.60 01 (20.

oo i .
0.00 0.00 -r-pr-r-r-y-'--r0
0.0 4.0 8.0 12.0 ao0 4.0 8.0 f2.
Time, min Time, min

3 (b 3 01 (d

O 4.0 8.0 12.0 0.0 4.0 8.0 12.0

Time, min Time, min

Fig. 5.23. Electropherograms of (a) outside air of TTU Chemistry building, (b) vapors from a cut
onion, (c) vapors from a concentrated HC104 (15 s sampling), and (d) vapors from a half-filled
diet cola can (half-full and mostly defizzed). Migration-based identifications: 1: acetate, 2:
carbonate, 3: formate, 4: nitrate, : sulfate (probably originally sulfite), 6: perchlorate and
chloride overlapped (can be resolved at lower concentrations), 7: benzoate, 8: phosphate.
Unlabelled peaks could not be identified. Reproduced with permission from Ref. [145] (copyright
1998 American Chemical Society).

for SO 2 with a suppressed conductometric CE system is -I ppbv. Examples of

analysis by using such a system as a "sniffer" are shown in Fig. 5.23.
In a subsequent paper [146], these authors described a direct measurement
of phenolic substances in the gas phase at levels relevant to occupational health
by micellar electrokinetic chromatography (MEKC) using direct ultraviolet
detection using a loop containing 0.5 mM NaOH. A total of twelve chloro- and
nitro-substituted phenols were selected for this study. Under the above samp-
ling conditions, limits of detection (LODs) for various phenols ranged from
high-single-digit to low-double-digit ppbv levels. The effect of several critical
parameters, such as the composition of the liquid film and the MEKC running
electrolyte, sampling period, and film evaporation on the collection efficiency,
separation efficiency, and calibration behavior were studied.
Surowiec and Dasgupta [147] studied exhaustive electromigration injection
from the liquid volume contained in the small wire loop attached to the end of
the separation capillary, using the film forming wire as the electrode. It was
shown that the wire loop could be used both as a gas collection interface and as
an electrode for the exhaustive electromigrative transfer of the collected anions
to the separation capillary. The influences of all relevant experimental parame-
ters were studied. Specific attention was paid to the determination of atmos-
pheric formic and acetic acids; the estimated limits of detection (LODs) were low
single digit ppbv levels for the relatively insensitive but generally applicable
indirect UV absorptiometric detection technique.
Wires can be used not only to hold liquids and as high voltage electrodes, they
are readily used as the active sensing electrodes in an electrochemical sensing

134
Fig. 5.24. Stopped flow, especially at high gas .
sampling rates, remarkably improves sensitivity.
Circles: response under conventional conditions,
gas sampling rate 1 1min'; squares: response with
the liquid flow stopped; triangles: response under
stopped flow conditions with the gas sampling
rate 2 1min'. Reproduced with permission from D
Ref. [148] (copyright 1997 Pergamon Press). H,2 , ppbv

system. Huang and Dasgupta [148] examined electrochemical sensors for hydro-
peroxides based on thin flowing or stationary films. The sensor was composed of
two segments of a Nafion tubing put on a silver wire. A small portion of the silver
wire was exposed and was chloridized to function as the reference electrode. One
Nation segment had a Pt-wire coil wrapped on it to function as the counter elec-
trode and the other had a similar Pt/Rh wire coil, which functions as the working
electrode. A collection solution flows as a thin film on the sensor surface and also
functions as the collection medium. Hydrogen peroxide and cumene hydro-
peroxide were examined as test compounds. The former could be oxidatively
determined with a Pt/Rh electrode over a large range (ppbv-ppmv) without any
significant influence of relative humidity. In the stopped flow mode, the sensitiv-
ity increased further (Fig. 5.24). Cumene hydroperoxide, an industrially import-
ant hydroperoxide, could be determined easily with a relative precision better
than 10% in the vapor phase over simulated process reaction mixtures contain-
ing per cent levels of the analyte by reduction on a Pd electrode. The sensor is
simple and inexpensive to fabricate and requires only a suitably equipped per-
sonal computer for operation. These authors have also reviewed the general area
of electrochemical sensing of gases based on liquid collection interfaces [149].
I conclude this section with an unique continuously flowing, recirculable,
film/drop sampler that exhibits the attributes of direct gas-liquid contact, a
continuous steady state flow of liquid maintaining a stable film and the ability to
recirculate the liquid [150]. This is perhaps the most appropriate transition
device to wetted denuders. This gas-liquid equilibrator or a gas collector is
intended as a low-volume interface between a soluble gaseous sample and a
liquid phase analyzer. It can also be used between a liquid phase sample and a
detector designed for use with gas samples. This paper addresses the application
of the device for the measurement of trace atmospheric ammonia. Gas collection
occurs solely by diffusive sampling such that aerosol particles are not collected.
The device, shown in Fig. 5.25, consists of a tube surrounded externally by a
jacket. Gas flows through the jacket and contacts a liquid film flowing on the

135
 - LO
N1 -
- HS

T -
_ _ N2
I ''

I Vt
Ss
Gas out -

11 (
ili
"
c___-
Left: Fig. 5.25. Continuous film gas-liquid collector/ equilibrator. N1, N2: 1/16-in nuts; N3:
1/8-in nut; T: PEEK tee; LI/ LO: Liquid inlet/outlet lines; S: PEEK sleeve; HS: heat shrink tube;
J: Jacket; C: bottom cap. Inset shows details of the stainless steel SS tube tip. Reproduced with
permission from Ref. [150] (copyright 2000 American Chemical Society).
Right: Fig. 5.26. Calibration behavior of the continuously flowing film collector (fluorometric
analysis in the 0-50 pptv region. Note that the first peak in a series always tends to read higher
due to flow disruption when concentration is changed. Reproduced with permission from Ref.
[150] (copyright 2000 American Chemical Society).

outer surface of the tube. The flowing film forms a drop at the tube terminus and
is aspirated off via the inner bore of the tube. The collected analyte can be (a)
directly sent to an analysis system, or (b) preconcentrated on a suitable station-
ary phase, or (c) the device effluent recycled. With a fluorometric flow injection
analysis system harnessed to measure NH 3 with such a collector, the limit of
detection (LOD, S/N=3) for a sample drawn for 18 min at 200 ml/min was 4.5
pptv, with the linear range extending up to 30 ppbv. Establishing true zero or
removing ammonia from air to prepare a zero level sample for the measurement
of pptv levels of ammonia was shown to be a formidable task. Figure 5.26 shows
system behavior at low pptv levels of ammonia.

5.5 WET DENUDERS

Wetted denuders (wet denuders, wet effluent diffusion denuders, WEDD)

provide for a high efficiency, high flow-rate collector for soluble gases. While
they make attractive front collection ends for highly sensitive analyzers for
gases, they are essential for particle analysis in situations where soluble gases
must be first removed in a continuous manner.

136
5.5.1 Single tube wet denuders

The original development of simple single tube wetted denuders [151,152] was
discussed in the previous review [1]. Using such a wet denuder and dedicated
colorimetric detection, Vecera and Dasgupta [153] explored the production of
indoor HONO from open flame sources. Sources such as gas heaters can produce
alarming levels of HONO.
Taira and Kanda [154] described a similar glass WEDD, that is, however,
much smaller in i.d. (1.6 mm vs. 8 mm). With this much smaller radius of curva-
ture, the authors found that surface treatment with conc. H2SO4 was sufficient
to render the surface wettable. While they demonstrated its successful applica-
tion for the collection and analysis of HNO,, it is noteworthy that at the recom-
mended sampling rate of 4 LPM, the flow is not laminar. The deposition of very
small particles (30-110 nm) was measured in the denuder at a flow rate of 2 LPM
(NRe 1777). The loss was significant (3-11% with NaOH flowing down the walls)
but the authors did not measure the aerosol in the denuder effluent. Rather,
they measured the particle counts before and after the denuder and this can be
misleading because deposition in the nonwetted areas affects the measurement.
Neftel et al. [155,156] used a WEDD as a collector for gaseous N species. The
artifact production of HONO from NO2 and peroxyacetylnitrate was determined
to be 0.1 and 0.4%, respectively. (A cautionary note is necessary regarding the
measurement of outdoor HONO levels with a wet denuder. Although NO2 and/or
NO do not seem to lead to significant levels of artifact HONO in a wet denuder, a
thorough study that includes a more complex mixture (e.g., SO2 ), has never been
rigorously conducted. Ambient HONO levels measured by direct spectroscopic
techniques tend to be lower than those measured by wet denuders.) Neftel et al.
[157] also constructed a miniaturized WEDD (12.5 cm long, 0.5 cm i.d.) for
studying ammonia equilibration in soil [157].
Vecera et al. [158] reviewed the general possibility of using diffusion de-
nuders to sample organic compounds and pointed out the potential of WEDD
devices in particular in such applications. They also followed up on this. Zdrahal
and Vecera [159] used an WEDD for the preconcentration of gaseous 2,4-trichlo-
rophenol (TCP) from air. A mildly alkaline scrubber liquid (water adjusted to pH
8.5) was used as the WEDD liquid. The concentration of TCP was determined
using reversed-phase liquid chromatography with a C-18 stationary phase and
UV detection at 218 nm. The system allowed results to be obtained at 7 min.
intervals with an LOD of 685 ng/m 3. Collection of organic vapors is better
attained with organic solvents as WEDD liquids. Zdrahal et al. [160] studied the
collection of 1,4-dichlorobenzene (DCB) as a model compound using alcohols,
glycols, and their aqueous mixtures as WEDD liquids. 1-Propanol was found to
be the most suitable. With sufficient flow of the WEDD liquid (> 150 tl/min), the
experimental collection efficiencies agreed with theoretical values for air flow
rates from 0.5 to 4.3 LPM at DCB concentrations of 48 and 720 ,Ag/m3. The
denuder operated without difficulties at temperatures from 18-33°C and over an

137
RH range of 10-100%.
Peskova et al. [161] recently studied WEDD preconcentration for the deter-
mination of methanol, ethanol, -propanol, 2-propanol, 1-butanol, 1-pentanol,
acetone, methyl ethyl ketone, diethyl ketone and methyl n-propyl ketone in air,
with gas chromatography- flame ionization detection (GC-FID)or GC-MS
analysis. The compounds are continuously collected into water. The LOD of
alcohols and ketones in the effluent are as low as 1 gg/l by GC-FID and 1 ng/l by
GC-MS. The authors state that the technique should be applicable for the
continuous monitoring of ppbv levels of both alcohols and ketones in the air.
Jaeschke et al. [162] described their static wet denuder system for the collec-
tion of ionogenic gases. Two parallel denuder systems are used, one with an
acidic absorber (pH 2.7 with HC1) for the collection of ammonia, and the other
with an alkaline absorber (pH 10 with NaOH) for the collection of acid gases.
The authors characterize their denuder system as "vertical", a distinction lost
on this author because all the previous denuders described in this section also
operate vertically. Perhaps the intent is to distinguish it from the Netherlands
Energy Research Foundation (ECN) annular WEDDs videe infra) which operate
horizontally. Regardless of the terminology, the denuder system does have char-
acteristics that are unique. The wetted surface is made from fine nylon stock-
ings. (In screening various material as denuder liners [151], this material was
also studied in our laboratory. Similar to the use of polycarbonate liners, we
observed a small but discernible loss of HNQO. However, a thorough study of all
types of available nylon stockings was not conducted!). The denuder is fed from a
constant pressure head using a pressure sensor that controls the feed pump. The
denuder effluent (9 ml recirculated volume) was withdrawn periodically and
analyzed batch mode by IC. It is interesting to note that given the dimensions of
the denuder (25 mm i.d., 350 mm long) the stated collection efficiencies for NH,
(96, 87 and 79% at flow rates of 2, 4, and 6 LPM) are significantly super-
theoretical (at 6 LPM, the computed collection efficiency is 68%; see Eq. (5.1)
and Section 5.2.1). The merit of batch mode operation is not clear. Recently
Costa et al. [1631 have described the use of such denuders with NaF and Na2 CO3
solutions as DS liquids in individual DS units to collect various gases and deter-
mine them by IC analysis.
(Other "fabric denuders", consisting of a piece of porous fabric, with or
without impregnation, stretched across an opening, e.g., a filter holder, have
been reported [164]. These are very different from any other devices described
herein. Similar denuders based on porous metal disks were reported earlier.
These authors modeled the collection with membrane filtration theories [165]. It
has been reported that particle losses can be low. Perhaps this type of denuder
will find increased use, especially in personal monitors [166].)
Buhr et al. [167] have described a single tube wet denuder-fritted filter
combination coupled to an IC for the measurement of gaseous nitric acid and
particulate nitrate and sulfate. Reasonable agreement was obtained for nitric
acid values with collocated filter pack measurements of nitric acid.

138
 Extraction
solution

Fig. 5.27. Photocatalytic single

tube wetted denuder of annular
Sample
design for collecting NO. Repro- solution
duced with permission from Ref.
[168] (copyright 1999 Elsevier
Science).

One very interesting single tube denuder designed by Komazaki et al. [168]
is of annular construction but with only the inside of the outer tube as the active
surface. NOx (both NO and NO2 ) are photocatalytically collected on a
TiO 2 /hydroxyapatite surface. Figure 5.27 shows the device. A tubular fluores-
cent lamp (Hg "black light", 365 nm) surrounded by a quartz tube constitutes
the inner element and the outer tube (glass) has a mixed coating of TiO 2 and
hydroxyapatite. The denuder operates in a batch mode; water washes the walls
after a preset period of illuminated sampling. The effluent is sent to an IC for
analysis and total NOx is measured from the sum of nitrite and nitrate. The cata-
lytic activity is completely recovered on washing and drying. Presumably it does
not work if the liquid is continuously flowing but this is not explicitly addressed.
Figure 5.28 shows the need for both constituents of the mixed coating and UV
illumination to successfully collect NO.
Chang et al. has advanced an equally interesting concept in wet denuders
[169]. The effluent liquid in the cylindrical denuder is not directly added but gen-
1
II ,
-UV illumination-----*
___
Fig. 5.28. Collection of NO by the photocatalytic WEDD
(flow rate 0.2 1/min). Note that NO passes through com- .o
pletely with a coating of only hydroxyapatite (HAP) in the
/ --- no
dark; some is collected upon turning on the light but bulk / .... TO
still passes through. With only a TiO2 coating, nothing is
collected in the dark; with UV illumination some analyte Dor
is initially collected but things rapidly return to the no- S2 I- -···H~
TTi2+HAp
collection condition. Only with both TiO2 and HAP
present and in the presence of UV illumination is NO a-
efficiently collected. Reproduced with permission from 10 20 30
Ref. [168] (copyright 1999 Elsevier Science). Time (min)

139
erated by cooling so that condensation occurs. The temperature and RH of the
sample is controlled by passing the sample through a thermostated Nafion
humidifier (to give 85% RH at 40°C) and the analyte gases are then collected by a
cooled (2°C) cylindrical denuder where condensation occurs on the walls. An IC
analyzes the liquid effluent. The LODs for CH3 COOH, HONO and SO2 are 22, 19
and 9 pptv, respectively. The precision ranges from 0.3 to 3.0% RSD. The method
was successfully applied to urban air analysis in Seoul, Korea. The temporal
results closely followed those produced by a PMDS-IC system operated in paral-
lel. The use of a cation exchange membrane for humidification precludes the
measurement of ammonia or other soluble gases that are not acidic. Although
the method was said to be applicable to weak acid gases such as CHCOOH, no
clear recovery data were presented to indicate whether or not such a weak acid
gas is quantitatively transmitted through the humidifier. The method has been
patented [170] and is discussed in detail in a PhD thesis along with several other
(parallel plate, annular) condensation denuder and DS designs [171]. Particle
deposition in such denuders need to be examined in greater detail since thermo-
phoresis can attract particles to cold surfaces [172].

5.5.2 Rotating wet annular denuders

The wet annular denuder was actually the first wet denuder to be introduced
and predates the less sophisticated single tube wet denuders. The wet annular
denuder was invented by the ECN group who have made many other contribu-
tions to atmospheric analysis [173]. In the originally introduced device, opera-
tion was in the batch mode. It was a wetted annular denuder (do = 45 mm, di =
42 mm, L = 30 cm) held together by PTFE spacers. Twenty-five milliliters of an
aqueous absorber was charged into it and the outer tube was rotated by a
motor-belt-pulley arrangement at 40 revolutions per minute (rpm)-this main-
tained a uniform film on the walls of the device. Near-quantitative collection of
several gases of interest was achieved at flow rates as high as 32 LPM.
Subsequently, Wyers et al. [174] from the same ECN group reported a
rotating wet annular denuder (RWAD) from which solution is continuously
pumped in and out such that analyzers can be continuously coupled. Further
refinements have been reported since [175-177]. The authors strived from early
on to reach excellent analytical precision, to allow these methods to measure
fluxes of different pollutants aided by simultaneous micrometeorological mea-
surements. Several field studies have been conducted with these devices, partic-
ularly with a view to measure the flux of nitrogen compounds [178]. In most
cases, few other alternatives have been available. In one study for example, the
only competition for a continuously operating RWAD was another RWAD oper-
ating in batch mode [179]! The results of these measurements have made a
significant contribution to better descriptions regarding surface exchange of
important pollutants. In a recent review on the advances in micrometeorological
methods to measure gas and particulate N fluxes, Fowler et al. [180] conclude

140
 Fig. 5.29. Schematic diagram of the continuous-flow denuder.

that relative to eddy accumulation techniques, wet denuders can provide the
same value (within 10%) at less than 10% of the cost. In another recent review on
methods for measuring emission rates of ammonia from livestock buildings and
from slurry or manure storage, Phillips et al. [181] also professed their love for
AMANDA (Ammonia Measurement by ANnular Denuder sampling with on-line
Analysis).
The ECN RAWD units operate at a high sampling rate of 25-30 LPM. All
components are packaged into a transportable box, to facilitate field deploy-
ment. Current versions use liquid mass flow meters to monitor peristaltic pump
flows exactly, an IR sensor is used to maintain constant liquid level in the
denuder. Bromide, typically not present in atmospheric samples in discernible
amounts, is added as an internal standard. Anions are measured by carbonate
eluent anion chromatography, ammonia is measured in a portion of the stream
by adding strong NaOH, diffusing the liberated NH3 across a membrane to a
water receptor and measuring conductivity following a procedure originally
outlined by Carlson [182].
Figure 5.29 shows the ECN RAWD. The LOD for the acid gases, as measured
by IC with 20 min time resolution, is easily below 100 ng/m3 for HC1 and HONO
and below 200 ng/m 3 for HNO3 and SO2 [177]. In early work [175], the relative
standard deviation was -4% at gas concentrations of 1-2 g/m 3 , this has been
further improved significantly with the incorporation of internal standards. The
ammonia measurement system has been widely used and cross comparison
against a laser based detection scheme (which has inferior LODs) has provided
favorable results [183].

5.5.3. TTU Gravity flow wet annular and parallel plate denuders

Many contributions from the ECN group and the present author's laboratory at
Texas Tech University (TTU) have proceeded in parallel. Thus, while the ECN
group contributed the first wet denuder, TTU contributed the first continuously
operating wet denuder [151]. The first continuously operating annular denuders

141
 OT- -
LII IT
SN

(a) (b) (c)

Left: Fig. 5.30. TTU Wet Annular Denuder. OT: outer tube, LII: liquid inlet to inner tube, LIO:
liquid inlet to outer tube, IT: inner tube, OR: O-ring, C: cavity, LC: Plexiglas liquid collector cup,
LOI: liquid outlet from inner tube, OF: outer fitting, LOO: liquid outlet from outer tube, SN:
supporting syringe needle, PF: Kynar® porous frit, PG: top Plexiglas insert, SA: silicone
adhesive plug. Reproduced with permission from Ref. [25] (copyright 1993 American Chemical
Society).
Right: Fig. 5.31. Wet parallel plate denuder: (a) overall appearance, (b) view of one plate, (c): top
cross section. GP: Glass plate, LI: hypodermic needle liquid inlets, PG: Plexiglas spacer, TF:
Teflon film, SC: silica coating, LO: liquid outlet, Al: air inlet (the outlet, not shown, is similar.
Reproduced with permission from Ref. [25] (copyright 1993 American Chemical Society).

were described by the ECN and TTU groups in the same year [174,25]. The
design of the TTU wet annular denuder (WAD) is shown in Fig. 5.30. The inner
tube is supported from the outer tube by three symmetrically placed hypodermic
needles, one of which serves as the liquid inlet for the inner member. It fills a cup
shaped cavity and overflows along the outer walls of the inner tube, coated with
easily wettable porous glass [151]. The water flows down into a cup-shaped
collector from which it is withdrawn. Flow in and out along the inner wall of the
outer tube (bearing a similar wettable surface) is very much along the lines of
previous single tube WEDD units. The TTU WAD (7 mm od 300 mm long tube
inserted in 10 mm id, 400 mm length tube) is much smaller than the ECN WAD.
It does not require a mechanical rotating arrangement but quantitative
sampling capability is also far less. The device quantitatively captures SO2 at a
sampling rate of 3 SLPM and efficiency drops to -90% by 5 SLPM (WAD liquid
0.5 mM H, 2,, p = 680 mm Hg).
In the same paper [25], the first parallel plate wet denuder (PPWD) was also
described (Fig. 5.31). Typically each glass plate measured 600 x50 mm (active
coated area width 50 mm) The silica-coated area is represented by the shaded
region. The two plates are separated by a 3 mm thick Plexiglas spacer covered
with PTFE tape. In each plate, ca. 12 cm are left uncoated at the bottom to allow

142
the development of laminar flow before the active collection region begins. At the
top of the coated region, three evenly spaced holes (0.75 mm) are drilled. At the
bottom, one hole is provided just above the point of the vee. Liquid is pumped in
parallel through all three needles, typically for a total flow rate of 200-300
/l/min. It is aspirated at the same rate at the bottom port and the opposing plate
has an identical arrangement. Air inlet or outlet connections to the denuder
were made by thermally deforming and stretching thin-walled PTFE tubes of
the appropriate diameter (ca. 10 mm for the PPWD described) to a rectangular
cross section at one end. The PPWD was substantially more efficient than the
TTU WAD, removing SO 2 quantitatively at 8 SLPM and removing -95% at 10
SLPM. Several other characteristics are noteworthy. At the stated liquid flow
rate, the mean film thickness of the falling water film was estimated to be 25-50
Am. Less water evaporates from the PPWD (which has a greater hydraulic cross
section. At 23°C, no dry spots develop in the PPWD even with a liquid flow rate
as low as 100 bl/min and an airflow rate as high as 16 SLPM (dry air). This is,
however, very dependent on the water affinity of the wetted surface, with
polymer surfaces (vide infra) much more water is seen to evaporate. Significant
amounts of the liquid input are evaporated during transit through the denuder
but the exit air does not reach saturation RH (referring to saturation at ambient
temperature). Due to the significant latent heat of evaporation of water and
near-adiabatic cooling, the temperature of the wetted surface is substantially
below ambient. For example, with the PPWD operating at an air sampling rate
of 15 SLPM and an inlet humidity and temperature of <-30% and 23°C
respectively, within a few cm of the inlet, the temperature of the PPWD, as
measured from the external glass surface, drops to 11.8°C. Nevertheless,
observed collection efficiency seems to be well predicted by extant theoretical
equations for such denuders, using the diffusion coefficient of the gas at ambient
temperature [26].
Figure 5.32 shows the coupling of the PPWD to an IC (the lowest rung IC
then available, a "QIC" from Dionex Corp. was used). The aspirated effluent
from both plates were combined and preconcentrated on a preconcentrator
column. The dual loop injection valve is configured with two such precon-
centrator columns; while one is being loaded with the sample, the other is being
eluted. Typically, the valve was switched every 8 min, a new chromatogram is
thus generated with this frequency.
The lower sampling rate of the TTU PPWD relative to the ECN RAWD is
compensated for by the superior IC LODs afforded by a hydroxide (rather than a
carbonate based) eluent. To date, this difference in eluent use in IC between the
two groups persists. Figure 5.33 shows the triplicate chromatograms resulting
from sampling clean air and 19 pptv SO 2. Reproducibility of the sulfate peak at
this level of SO2 was 0.8% in RSD. The precision level of the blank (clean air)
signal was equal or better. These data lead to a computed limit of detection of 0.5
pptv (1.3 ng/m3), well below any other reported technique. In a similar study to
measure HONO and HNO, by PPD-IC, the method LODs were found to be 0.11

143
Fig. 5.32. Liquid phase analytical system schem-
atic for a PPD coupled to an IC. CC: preconcen-
tration columns, GC: guard column, AC: analytical
column, S: membrane suppressor, D: conductivity
detector. Reproduced with permission from Ref.
[25] (copyright 1993 American Chemical Society).

Fig. 5.33. Triplicate chromatograms resulting from sampling: (left) clean air, (right) 0.019 ppbv
(50 ng/m3 ) SO2 . Peak A is sulfate, B is from CO 2 and C is due to an unknown analyte that is
leached from poly(vinyl chloride) pump tubing. Reproduced with permission from Ref. [25]
(copyright 1993 American Chemical Society).

and 0.23 pptv for these gases [184]. However, gaseous NO2 produces some NO2-
and NO- by hydrolytic reactions (-10 times as much NO2 as NO,-). Although a
very small extent (0.06%, see also Ref. [154]) of the gaseous NO2 is actually con-
verted, this is sufficient to significantly alter practical LODs in ambient air.
Moreover, there is some evidence that N2 0, and other, as yet unidentified nitro-
genous species in ambient air, may also generate such artifacts upon reaction. In
addition, NO2 uptake may be higher in the presence of significant amounts of
SO2 , this has never been directly examined.
Both the WAD and the PPWD exhibited very low particle deposition, as
shown in Fig. 5.34. In this context, it is interesting to note that in all wetted
denuders, particle loss increases when the same denuder is operated dry. When
wet, evaporation of the liquid from the denuder surface creates a flux that
strongly inhibits particle deposition on that surface due to diffusiophoresis

144
 200 ....

0
'o
° 160
I A
C
'C
Mr

0
U)
a
t
a
8-

Air flow rate (1min)

Fig. 5.34. Particle losses in a TTU WAD and a PPWD. The inset shows the size distribution of the
NaNO3 aerosol used. Note that the mean aerosol size grows from the inlet to the outlet due to
deliquescence. Loss was measured by measuring the nitrate content of the denuder liquid
effluent. Reproduced with permission from Ref. [25] (copyright 1993 American Chemical
Society).

[172]. Water evaporating from the wet surfaces set up a flux of molecules that
bombard incoming particles numerous times and make it more difficult for
particles to be deposited. Others have also noted decreased particle deposition in
wetted denuders when the latter are in normal operation compared to when they
are operated dry [154].
Between the WAD and the PPWD, the PPWD is more efficient and is
significantly easier to construct. For this reason, in our laboratory, we have
chosen to work with the PPWD. The present design uses Plexiglas plates with
PTFE spacers, rather than glass plates. This eliminates fragility problems. The
desired area of each Plexiglas plate is microstructured with a grinding tool to
render it more easily wettable (nevertheless, it is not as wettable as their porous
glass counterparts and must initially be filled with water to ensure even
wetting). The denuder stand is conveniently made from an wooden support base
to which threaded pipe flanges are secured by screws and the threaded end of a
3/8 in i.d. galvanized steel piping is used as the support stands. The denuder is
simply bolted to these. If the denuder is used for gas analysis and deployed
without a cyclone with a very short inlet (<5 cm, Pefluoroalkoxy fluorocarbon
(PFA Teflon)), the sample air must enter the denuder at the bottom. As such, the
denuder needs to be on a tall stand that allows the inlet to be -60 cm off the
support base. To minimize interaction of the inlet air sample with the stand
components, especially in still and stagnant air, we also recommend that the
lengths of the iron support stand from the base to the bottom of the denuder be
wrapped with PTFE tape. Each denuder plate is 10.0x55 cm (3/8 in thick) with
the active wettable area of 6.5x42 cm, starting 7.5 cm from the top and 1.75 cm

145
 L

Ai

Fig. 5.35. Left: TTU PPWD shown schematically; AI/AO: air inlet/outlet, LI/LO: liquid
inlet/outlet, LR: Liquid reservoir WA: wetted area, S Spacer between plates, SH Screw holes to
hold plates and spacer together, also affixes denuder to stand. Middle: Complete denuder on
stand. Right: Top and bottom detail of PPWD.

from each edge. The denuder liquid is forced through a fritted PVDF barrier to
allow even flow down the plate and is aspirated from the V-groove at the bottom,
4.5 cm from the bottom. The two plates are spaced by a 3 mm PTFE spacer. The
air inlet/outlet holes, circular at the termini, are machined with a contour that
becomes elliptical as they approach the interior of the denuder to allow a smooth
entrance/exit of the airflow. Pefluoroalkoxy fluorocarbon (PFA Teflon) tubing (1
ga., 8.3 mm o.d., 7.5 mm i.d.) fit tightly into these apertures. The denuder is
shown in detail in Fig. 5.35. Note that serious analyte losses and/or artifact
reactions can occur in a cyclone inlet placed ahead of a denuder [185]; this should
be avoided.
In this configuration, these PPWD units have been used in several USEPA
sponsored field studies, both for trace gas analysis and for removing gases prior
to particle collection, in Atlanta (1999) [186,187], Houston (2000) [188], and
Philadelphia (2001) [75] as well as to remove acid/basic gases before a system
that determines the acidity of the aerosol fraction [189].
A very interesting paper by Rosman et al. [190] appeared during the prepara-
tion of this chapter, in which the authors describe a PPWD constructed of PFA
teflon with an inert screen material as the wetted element. Detailed character-
ization, performance and field evaluation data were presented.

5.5.4 Circular end parallel plate wet denuders (CEPPWD)

Wet denuders based on the circular-end, parallel plate geometry, which facilitate
air inlet/outlet connections, were first used by Staffelbach et al. [27]; few
operational details were given. The same geometry has been subsequently used

146
by Hesterberg et al. [191], Ammann et al. [192] and described in more detail by
Zellweger et al. [193]. Overall, the deployment is similar to that used by the TTU
group, except that -50 M NaHCO 3 was used as the denuder liquid (vide infra)
and a carbonate- bicarbonate eluent was used for IC.
Loflund et al. [194] described the most sophisticated CEPPWD-based analyt-
ical system to date, with very impressive capabilities. The collection device is a
quartz CEPPWD with a silica/silicate coated active area. The air sampling rate is
5 LPM and the liquid flow rate is 167 pl/min. A gradient IC system is used with
an electrodialytic eluent generator for analysis. The system has the capability to
measure the major organic (formic, acetic, propionic, oxalic, malonic and suc-
cinic) as well as inorganic (SO2 , HONO, HNO3 , HCl and HF) acidic gases as well
as ammonia by cation chromatography on a second system. A pair of cation and
anion preconcentrator columns are respectively incorporated in two 10-port
valves. The cation and anion systems are alternately loaded and analyzed (30
min each). The LODs (pptv) stated for the acid gases were as follows: formic (5),
acetic (4), propionic (30) butyric (19), valeric (21), pyruvic (45), oxalic (22),
malonic (6), succinic (42), o-phthalic (10), citric (2), maleic (8), HF (56), HC1 (74),
HONO (10), HNO3 (11), SO 2 (9), H3 PO4 (22). At 400 pptv, the LOD for NH 3 was
significantly higher. The device was successfully used for field measurements at
an alpine station.

5.5.5 Which wet denuder design to choose?

A significant number of designs, all of which offer high collection efficiencies at

modest to high sampling rates, have now been presented. Beyond the obvious
judgement call a user may make as to which design will be simplest for his needs,
another consideration may be important when a continuous analyzer is coupled
to the denuder effluent and fast response is desired. In order to obtain good time
response, one must have a quick washout profile in the denuder. If one is using
similar collecting surfaces (wettable glass or plastic that has the same liquid film
thickness) and in different geometric designs, it follows that the amount of liquid
needed to wash the denuder surfaces to a given degree should be linearly related
to the overall active surface area. Especially if the entire liquid effluent is not
being injected, we should be particularly interested in the liquid effluent concen-
tration from the denuder, assuming that liquid surface velocity is the same in all
cases. Table 5.1 presents such a comparison between four denuders. Among the
parallel plate and annular denuders, I have assumed the same distance of
separation between the collection elements. Curiously, the difference between
the best and the worst is less than a factor of two in terms of the ultimate
effluent concentration. The short and thin WAD is better than the long and fat
WAD, contrary to intuition. The PPWD appears to be marginally superior to the
rest. It should be noted that these considerations do not apply when the primary
intent is to perform particle analysis, a high flow rate must be used in such a case
to attain reasonable LODs for particulate constituents of interest.

147
TABLE 5.1
The tale of the four denuders. Isoefficient (95% efficient in collecting SO2 ) at stated flow rate.
Single tube Annular thin and short Annular fat and long Parallel plate
d = 3 mm di = 1.0 cm di = 3.0 cm w = 5 mm
do = 1.3 cm d = 3.3 cm s= 1.5 mm
L = 50 cm L= 10 cm L = 20 cm L =30 cm
Q = 1 LPM Q= 3.8 LPM Q = 20LPM Q = 20.3 LPM
Relative mass collected
1.00 2.23 11.8 11.9
Active surface area, cm 2 (liq. flow in ul/min to maintain same velocity)
47.1 (283) 72.3 (433.8) 396 (2376) 300 (1800)
Relative analyte concentration in effluent
1.00 1.46 1.41 1.87

5.5.6 Wet denuder coupled IC analysis of acid gases and ammonia

The present TTU configuration used for this analysis is shown in Fig. 5.36.
Typically the air sampling rate is 5 SLPM and 0.5 mM H2 02 is used as the
denuder liquid at 0.5 ml/min on each plate, each stream pumped through a
disposable mixed bed ion exchange resin column MB (0.67 cm i.d. x 15 cm,
PTFE column filled with Dowex MR-3 resin) located immediately before the
PPWD liquid entrance ports. The effluent streams are aspirated at 1.1 ml/min
from each plate (using same peristaltic pump but larger tubing, 1.52 mm vs. 0.89
mm i.d. Pharmed® tubes are used for input vs. aspiration) to ensure all liquid is
aspirated from the bottom of the PPWD. The aspirated flow streams are
combined and sent to the analysis system. An 8-channel peristaltic pump is used
for all low pressure pumping, 7 of 8 channels are used.
A significant difference between the TTU system and other similar systems
such as that of Loflund et al. [194] and the ECN RWAD [174-177] is in feeding
the same sample serially to cation exchange and anion exchange preconcentra-
tors (in that order, which is important because anion preconcentrators used by
us contain accessible cation exchange sites). This increases sensitivity and/or
sample throughput, compared to parallel or temporally sequential arrange-
ments. Anions pass unretained through a cation preconcentrator CC (Na+-form
strongly acidic cation exchanger Dowex 50Wx8, 100 mesh, 4x25 mm) on to an
anion preconcentrator TACLP1, each paired with an identical twin preconcent-
rator in 10-port dual loop injector valves V1 and V2. At the end of a typical 15
min sampling cycle, both of these valves switch, and the other set of precon-
centrators are brought on-line. In the anion analysis system, electrodialytically
generated KOH (15.5 mM) elutes the collected anions at 1.5 ml/min at 29°C
through high capacity AG11HC (4x50 mm) and AS11HC (4x250 mm) guard
and separator columns, a water-regenerated electrodialytic suppressor and a

148
 p MFC )

PPWD
D

LC3S

Fig. 5.36. P: Air aspiration pump, MFC: mass flow controller (5 I/min), PPWD: parallel plate wet
denuder, EG40: electrodialytic eluent generator, IS25: eluent pump (1 ml/min), TAC-LP1: 4 mm
anion preconcentrators, AG11HC: guard column, AS11HC: separator Column, SRS: electro-
dialytic suppressor, CD25: conductivity detector, LC30: Column oven, MB: mixed bed resin
columns, PP: peristaltic pump, PMD: porous membrane device, CC: cation preconcentrators, V1,
V2: 10-port valves, V3: 4-port valve, V4: 3-way valve, R: 0.25x70 mm restrictor tubing to
eliminate formation of air bubbles, W: waste.

conductivity detector. On the cation side, at the time V1 switches, valve V3 is

positioned to recycle the output into its own source container (15 mM NaOH)
while V4 allows water to be pumped through V3 to CC to remove any entrained
air bubbles (PP5 and PP6 both operate at - 1.1 ml/min). After 5 min, V3 and V4
both switch, and water is recycled through PP6 while PP5 pumps NaOH onto
CC. Ammonia elutes from the column and goes across a membrane in the porous
polypropylene membrane device PMD to waste. A portion of the ammonia
transfers across the membrane to the receiver side where pure water flows (PP7,
0.4 ml/min) and carries the transferred ammonia to the conductivity detector
(based on the principle described by Carlson [182] and used by the ECN group as
well). A valving arrangement (as in PP6 and V4) can be put in with PP7 such
that the flow on the receiver side is stopped during the time the liberated
ammonia bolus passes through the donor membrane. This will improve
sensitivity. However, we have not had urban measurement situations where
detection sensitivity has been a problem. The NaOH elution period continues for
5 min. At the end of this time, the system returns to washing CC with water so
that excess NaOH is removed from the column and NH,+ can be captured from
the denuder effluent without hindrance (because of the overwhelming presence
of CO,, the dissolved NH3 in the denuder effluent is always present as NH4+ . Due
to its considerably greater affinity for cation exchange sites, it is easily captured

149
 1.20-
Houston HRM-3
('K 1:38 to 2:08 09/12/00
E
E
'8 0.80
.2
E
a; S
\ Ammonia 1.70 ppb
C

.f 0.40-
E
.2
.5e i
t

0.0.
16.00 18.00 20.00 22.00 24.00
Time (min) Time (min)

Fig. 5.37. Top: Gas phase acids. Bottom: Gas phase ammonia from TTU system.

on a Na+-form column. The output from the detectors during a field campaign in
Houston, TX is shown in Fig. 5.37.
We have also experimented with a simple setup where the same conductivity
detector and the same data processing channel can be used to measure both the
acid gases and ammonia. The arrangement is particularly attractive if one is
primarily interested in the principal acid gases, SO2 and HNO,, corresponding to
the sulfate and nitrate peaks (this is specially true for aerosol phase analysis
where sulfate and nitrate are frequently the focal points), such that only a
rudimentary separation between chloride (plus organic acids), nitrate and
sulfate is needed. This can frequently be accomplished with a short separation
column. The arrangement is shown in Fig. 5.38; the principal difference from
Fig. 5.36 is that the suppressor effluent is routed through a four-port valve V5.
In one position, the suppressed effluent proceeds directly to the detector. Only
when the anion chromatogram has finished (sulfate has eluted), the valve is
switched, directing the suppressor effluent (essentially water) through a mixed
bed resin to the receiver side of the PMD. At this point, V1 is switched to elute
the collected ammonia, the PMD receiver effluent proceeds to the detector,
measuring ammonia. The right panel in Fig. 5.38 shows the output of such a
system, showing peaks due to 10 nmol amounts of chloride, nitrate and sulfate
and 40 nmol NH3 . (Note that these experiments were carried out with bottled,
rather than an electrogenerated hydroxide eluent. The chromatographic back-
ground is high due to the presence of CO,, and the baseline goes down when V5
effluent is switched for ammonia measurement because the background
becomes essentially freshly deionized water.) It should be obvious that ammonia
measurement can be carried out before or after anion elution and the PMD
receptor can be maintained in a stopped flow mode to increase sensitivity. In the
example shown, we have eluted the ammonia where the void volume peak will
normally elute, making efficient use of time.

150
 2-

npnl ,2s,

H4

Time (min)
Fig. 5.38. Left: One detector anions-ammonium measurement system. Legend as Fig. 5.35. V5 is
an additional 4-port valve. Right: First set of peaks in "normal" anion IC, second set of peaks
shows ammonium detection system switched in where void peak normally elutes.

5.5.7 Denuder liquid considerations for IC coupling

With an IC as the analyzer of focus, gases of interest that will be collected by a

denuder are water-soluble ionogenic gases. Acid gases will include most notably
sulfur dioxide, hydrogen chloride, hydrogen fluoride, nitrous acid, nitric acid,
methanesulfonic acid and various organic acids, most notably acetic, formic and
oxalic acid. Ammonia is the sole basic gas that is likely to be collected under most
conditions.
If water is used as a collector, sulfur dioxide is collected as sulfurous acid.
This poses two problems. First, the Henry's law solubility of SO 2 is limited and
quantitative collection may not occur under these conditions. Second, some of
the bisulfite formed undergoes oxidation to sulfate either in the denuder or in
the IC system (or likely, both) leading to both sulfite and sulfate peaks that
unnecessarily complicates quantitation. Addition of a small amount of an oxi-
dant like H2 02 to the denuder liquid eliminates this problem and results in
virtually instantaneous oxidation of the collected SO2 to sulfate. The recom-
mended denuder liquid is thus 0.5 mM H2 02 . All other collected analytes,
including nitrite (originating from HONO) is completely unaffected by the H2 02 .
Dilute HO 2 2 is also easily cleansed of ionic impurities by passing it through a
mixed bed ion exchanger.
Zellweger et al. [193] recently pointed out a potential problem with collection
of the weaker acids, notably HONO in high SO, environments. The point is the
following. It can be easily computed that in an atmosphere containing 100 ppbv
SO,, quantitative collection at an air flow rate of 5 LPM and a total liquid

151
effluent flow rate of i ml/min will lead to 20 ,M H2SO4 in the liquid effluent. This
represents a pH of -4.4. In such a solution, many of the weaker acids will not be
fully ionized and therefore may have solubility limitations. Zellweger et al. were
particularly concerned with nitrous acid (pKa 3.1-3.2), although the problem
may obviously be more of an issue with analytes such as acetic acid (pK 4.75).
Zellweger et al. proposed a dilute solution of the chromatographic eluent,
alkaline in nature (-50 JM NaHCO3), as the denuder scrubbing liquid.
For several reasons, this solution may not be ideal. It is not clear in the
presence of large amounts of SO,, how far along the denuder the influent
NaHCO3 solution will maintain an acceptable pH. Second, weak acids dissolve in
aqueous solution both by their ionization and through their Henry's law parti-
tion (intrinsic solubility). The latter can be high, irrespective of the pH of the
collector-HCN, a very weak acid, has a very high intrinsic solubility, for
example [107]. Also, 100 ppbv or higher levels of SO, are found sporadicallyas a
plume impacts a sampling location but such levels on a sustainedbasis are not
common, at least in the USA. The suggested approach may provide a solution for
an exceptional case but generates problems for other, more common situations.
In a study conducted in Vienna in the winter of 2000 to determine penetration
into a second denuder, Loflund et al. [194] found 97% of the HONO was collected
in the first denuder.
These problems include the incompatibility of large amounts of carbonate in
the sample with hydroxide eluent based anion chromatography, presently the
preferred practice. Use of a carbonate containing denuder liquid generates a
substantial amount of carbonate in the effluent, a broad tailing carbonate peak
can obscure smaller analyte peaks in that region.
An alternative solution to the problem is to buffer the denuder liquid (prefer-
ably at a pH of 5.5-7). This is easily practised if the denuder liquid is not ana-
lyzed for collected gases but the interest is in removing the gases to perform
aerosol analysis. A zwitterionic buffer was used to remove high levels of acidic or
basic gases (as may be present in indoor environments when a kerosene-fueled
heater is operated) or high levels of ammonia (encountered in homes with live-in
pets) before aerosol analysis [189]. However, these approaches have not been
used when the denuder effluent, i.e., the collected gases, are of interest and
moreover, when it is desired to preconcentrate the sample. Zwitterionic buffer-
ing material may still be useful. Glycine, for example, has the correct pKa to be
useful and is known to be suppressible. Morpholinoethanesulfonic acid and
Bis-tris are other potentially useful zwitterionic buffers. It is known that at least
under some conditions it is possible to suppress zwitterionic buffers to provide a
low conductivity background [195]. If an external base must be added, LiOH will
be preferred since Li+ has such weak affinity for other cation exchangers.
(Indeed, it will be better to use LiOH rather than NaOH in ammonia elution and
measurement schemes as well). However, it is a long way from this knowledge to
establish whether any of the available buffer material can be purified to a suffi-
cient degree to make such trace measurements practical. Further, the optimum

152
concentration of such a buffer, as a compromise between maintaining pH and
achieving good preconcentration, will also need to be determined.

5.5.8 Miscellaneous wet denuder applications

A multiple parallel plate wetted denuder consisting of 11 "fusion-stitched" 2-ply

polyester screens has already been described [39] (see Section 5.2.1.1). While the
denuder is compact (5 cm dia., 30 cm tall), with a liquid flow rate of 10 ml/min
and a sampling rate of 50 LPM, it does not necessarily produce an effluent
concentration greater than a conventional PPWD and can be justifiable only
when a large sample flow must be processed, e.g., to perform isotopic analysis.
As a contrasting example, a miniaturized PPWD coupled to a briefcase size
capillary ion chromatography system was demonstrated. The active area (6 x 100
mm) of the Plexiglas denuder was rendered wettable by thermally bonding silica
particles (74-127 ,m) to the active area. At the operating flow rate of 15 Al/min
per plate, the maximum film thickness was calculated to be < 70 Am. The two col-
lection plates were placed 1.5 mm apart and the PPWD interior was provided
with a 4 cm entrance region prior to the wetted area, to develop laminar flow.
The denuder collected SO 2 close to theoretical expectations, with >95% collec-
tion at 0.5 SLPM (p = 680 mm Hg). The denuder effluent was pumped into a vial
and following gravity induced gas-liquid separation, the liquid was precon-
centrated at 17 Al/min on a 0.25x15 mm preconcentration column for 10 min.
The chromatography was conducted isocratically with 25 mM NaOH at 2.2
Al/min on a 0.18 x 350 mm capillary column followed by a hollow fiber membrane
suppressor and a conductivity detector. Under these conditions, backpressure
was only 500 pounds per square inch (psi) and sulfate eluted from the AS-11 HC
resin packed column within 10 min. For the 10 min sample the LOD corre-
sponded to 1.6 pptv SO 2 (or -20 pg/300 fmol). Were capillary scale IC's with
associated hardware and columns commercially available, there is little doubt
that they would replace conventional scale instruments for this kind of applica-
tions.
Globally, the oceans may be the largest source or sink of ammonia. Much
interest exists for accurately measuring sea<-4air flux of NH3 to bring a closure to
the global N budget. Genfa et al. [196] developed an instrument for the measure-
ment of gaseous ammonia concentration, NH3(w, eq), in equilibrium with surface
waters, notably ocean water. The instrument measures the ammonia flux from a
flowing water surface under defined conditions and allows the calculation of
NH(sweq) from the principles of Fickian diffusion. The flux collector is essentially
a PPWD; the seawater runs on one plate of the device (at relatively high flow
rate, to avoid significant donor depletion, the ammonia released from it is
collected by a slow flow of an acidic receptor liquid on the other plate. Coupled
with preconcentration on a silica gel column and a fluorometric FIA system, it is
sufficiently sensitive to measure the ammonia flux from sea-water. The instru-
ment design is such that it is little affected by ambient ammonia.

153
5.6 CONCLUSIONS

Diffusion based collection of trace gases into a liquid not only offers a means of
discrimination from concurrently present particles, the integration of continu-
ous or batch mode analytical systems for the analysis of the liquid effluent
provides unprecedented sensitivity and accuracy. Based on these principles,
many gases can be routinely measured at low to sub-ppb levels. DS devices
provide a contained liquid volume and can be operated in a stopped liquid flow
mode where they can themselves provide further preconcentration. They are
attitude insensitive and can be (and has been) used in airborne applications.
LCW based DS devices constitute a unique new entry and may offer very
inexpensive alternatives for measuring gases at typical urban levels. Selective
luminogenic reactions for gas measurement using LCW DS devices have not
been reported and may be uniquely attractive and sensitive. Wetted denuders
exhibit direct gas liquid contact. Especially in the annular and parallel plate
design, they offer high efficiency removal of gases at high flow rates. They are
however, attitude sensitive and have not been used in airborne applications. In
vertical WADs and PPWDs, the two liquid streams are physically distinct. Is it
possible to use different scrubber liquids with different selectivities (e.g., one
acid, one base) to capture different analytes? DS analogs to the more efficient
wet denuder designs, especially the PPWD, should also be particularly
interesting. Much has transpired in this field since the last review [1] and it is an
exciting time to look towards the future.

Acknowledgements

I thank Kajori Dasgupta for making it possible for me to write this review. Genfa
Zhang is thanked for his assistance to procure many not-so-easy-to-find refer-
ences, often several times a day. I thank USEPA in general and Bill McClenny in
particular, for continued support of our efforts in this area. This review was
made possible in part by the Tropospheric Ozone Research Program of the
National Exposure Research Laboratory, U.S. EPA. In the draft stage, I received
the benefit of meticulous comments from Albrecht Neftel, Don Olson, Kei Toda,
and Zbynek Vecera. I thank them in particular and also many others who read it
and provided positive reinforcement. Finally I thank my Friday, Susie Gonzalez,
for making all the corrections and reformatting and renumbering the citations.

REFERENCES

1 P.K. Dasgupta, ACS Adv. Chem. Ser., 232 (1993) 41-90.

2 0. Preining, Atmos. Environ., 25A (1991) 2443-2444.
3 T. Reichhardt, Enuiron. Sci. Technol., 29 (1995) 360A-364A.
4 U.S. Environmental Protection Agency, Office of Air & Radiation, EPA's Revised
Particulate Matter Standards, http://www.epa.gov/ttn/ oarpg/naaqsfin/pmfact.html

154
 5 J.S. Townsend, Philos. Trans. R. Soc. London, 193 (1900) 129-158.
6 W.L. Crider, N.O. Barkley, M.L. Knott and R. Slater, Anal. Chim. Acta, 47 (1969)
237-241.
7 P.G. Gormley and M. Kennedy, Proc. Roy. Soc. Ir. Acad. Sci., 52A (1949) 163-169.
8 D.B. Ingham, J. Aerosol Sci., 6 (1975) 125.
9 D.Y.H. Pui, C.W. Lewis, C.-J. Tsai and B.Y.H. Liu, Environ. Sci. Technol., 24 (1990)
307-312.
10 M. Krieger and R.A. Hites, Enuiron. Sci. Technol., 26 (1992) 1551-1555.
11 M. Krieger and R.A. Hites, Enuiron. Sci. Technol., 28 (1994) 1129-1133.
12 M. Dudek, L. Wolska, M. Pilarczyk and J. Namiesnik, Chemia Analyticzna, 45 (2000)
781-794.
13 M. Dudek, L. Wolska, M. Pilarczyk, B. Zygmunt and J. Namiesnik, J. High Resolut.
Chromatogr., 23 (2000) 449-454.
14 P. Koutrakis, C. Sioutas, S.T. Ferguson, J.M. Wolfson, J.D. Mulik and R.M. Burton,
Environ. Sci. Technol., 27 (1993) 2497-2501.
15 C. Sioutas, P.Y. Wang, S.T. Ferguson, P. Koutrakis and J.D. Mulik, Atmos. Environ.,
30 (1996) 885-895.
16 P. Koutrakis, P.-Y. Wang, J.M. Wolfson and C. Sioutas, US Patent 5,932,795, August
1999.
17 http://www.rpco.com/products/ambprod/amb3500/index.htm
18 C. Sioutas, P. Koutrakis and J.M. Wolfson, Aerosol Sci. Technol., 21 (1994) 137-148.
19 M. Possanzini, A. Febo and A. Liberti, Atmos. Environ., 17 (1983) 2605-2610.
20 Z. Ali, C.L. Paul Thomas and G.F. Alder, Analyst (London), 114 (1989) 759-769.
21 http: // www.rpco.com / products / ambprod / amb2500 / index.htm http:// snobear.
colorado.edu/ cgi-bin/Kiowa/Kiowa.con.pl?hsdenuder.doc. html
22 F. De Santis, I. Allegrini, P. Di Filippo and D. Pasela, Atmos. Environ., 30 (1996)
2637-2645.
23 B.J. Fan, Y.S. Cheng and H.C. Yeh, Aerosol Sci. Technol., 25 (1996) 113-120.
24 R.W. Coutant, P.J. Callahan, M.R. Kuhlman and R.G. Lewis, Atmos. Environ., 23
(1989) 2205-2211.
25 P.K. Simon and P.K. Dasgupta, Anal. Chem., 65 (1993) 1134-1139.
26 F. De Santis, Anal. Chem., 66 (1994) 3503-3504.
27 T. Staffelbach, A. Neftel, A. Blatter, A. Gut, M. Fahrni, J. Stahelin, A. Prevot, A.
Hering, M. Lehning, B. Neininger, M. Baumle, G.L. Kok, J. Dommen, M. Hutterli
and M. Anklin, J. Geophys. Res., 102 (1997) 23,345-23,362.
28 D.J. Eatough, A. Wadsworth, D.A. Eatough, J.W. Crawford, L.D. Hansen and E.A.
Lewis, Atmos. Environ., 27A (1993) 1213-1218.
29 H. Tang, E.A. Lewis, D.J. Eatough, R.M. Burton and R.J. Farber, Atmos. Environ., 28
(1994) 939-947.
30 D.J. Eatough, H. Tang and W. Cui, J. Machir. Inhal. Toxicol., 7 (1995) 691-710.
31 D.J. Eatough, D.A. Eatough, L. Lewis and E.A. Lewis, J. Geophys. Res., 101 (1996)
19,515-19,532.
32 W. Cui, D.J. Eatough and N. Eatough, J. Air Waste Manage. Assoc., 49 (1998)
1024-1037.
33 D.J. Eatough, F. Obeidi, Y. Pang, Y. Ding, N.L. Eatough and W.E. Wilson, Atmos.
Environ., 33 (1999) 2835-2844.
34 C. Sioutas, P. Koutrakis and B.A. Olson, Aerosol Sci. Technol., 21 (1994) 223-235.
35 C. Sioutas, P. Koutrakis and R.M. Burton, J. Aerosol Sci., 25 (1994) 1321-1330.
36 C. Sioutas, P. Koutrakis, R.M. Burton, Particul.Sci. Technol., 12 (1994) 207-221.
37 H. Patashnick and E.G. Rupprecht, J. Air Waste Manage. Assoc., 41 (1991)
1079-1083.
38 E.G. Rupprecht, M. Meyers and H. Patashnick, J. Aerosol Sci., 23 (1992) S635-S648.

155
39 P.K. Dasgupta, L. Ni, S.K. Poruthoor and D.C. Hindes, Anal. Chem., 69 (1997)
5018-5023.
40 D.A. Lewis, I. Colbeck and D.C. Mariner, Anal. Chem., 66 (1994) 3525-3527.
41 G.P. Vassilev and F. Roubani-Kalantzopolou, J. Chem. Soc. Faraday Trans., 91
(1995) 485-494.
42 J. Slanina and G.P. Wyers, Fres. Z. Anal. Chem., 350 (1994) 467-473.
43 M.G. Mennen, B.G. Van Elzakker, E.M. Van Putten, J.W. Uiterwijk, T.A. Regts, J.
Van Hellemond, G.P. Wyers, R.P. Otjes, A.J.L. Verhage, L.W. Wouters, C.J.G.
Heffels, F.G. Rimer, L. Van Den Beld and J.E.H. Tetteroo, Atmos. Environ., 30
(1996) 3239-3256.
44 G. Skillas and S. Kuenzel, J. Aerosol. Sci., 28 (1997) S43-S44.
45 H. Burtscher and S. Kuenzel, J. Aerosol. Sci., 26 (1996) S129-S130.
46 A.S. Kozlov and A.N. Ankilov, J. Aerosol. Sci., 28 (1997) S131-S132.
47 C.L. Paul Thomas, C.D. McGill and R. Towill, Analyst (London), 122 (1997)
1471-1476.
48 R. Bos, J. Air Pollut. Contr. Assoc., 30 (1980) 1222-1224.
49 I. Gacs and R. Ferraroli, Anal. Chim. Acta, 269 (1992) 177-185.
50 Y. Luo, R. Al-Othman, G.D. Christian and J. Ruzicka, Talanta,42 (1995) 1545-1551.
51 P.K. Dasgupta, Atmos. Environ., 18 (1984) 1593-1599.
52 R.L. Tanner, J. Shen, Aerosol. Sci. Technol., 12 (1990) 86-97.
53 I.C. Van Nugteren-Osinga, M. Bos and W.E. Van der Linden, Anal. Chim. Acta, 226
(1989) 171-175.
54 D.A. Phillips and P.K. Dasgupta, Sep. Sci. Technol., 22 (1987) 1255-1267.
55 K. Toda, P.K. Dasgupta, J. Li, G.A. Tarver and G. Zarus, Anal. Chem. (in press).
56 Q. Fan and P.K. Dasgupta, Anal. Chem., 66 (1994) 551-556
57 P.K. Dasgupta, S. Dong, H. Hwang, H.-C. Yang and Z. Genfa, Atmos. Environ., 22
(1988) 949-963.
58 S. Waiz, B.M. Cedillo, S. Jambunathan, S.G. Hohnholt, P.K. Dasgupta and D.K.
Wolcott, Anal. Chim. Acta, 428 (2001) 163-171.
59 T. Gilpin, E. Apel, A. Fried, B. Wert, J. Calvert, Z. Genfa, P.K. Dasgupta, J.W.
Harder, B.G. Heikes, B. Hopkins, H. Westberg, T. Kleindienst, Y.-N. Lee, X. Zhou, W.
Lonneman and S. Sewell, J. Geophys. Res., 102 (1997) 21,161-21,188.
60 A.L. Sumner and P.B. Shepson, Nature, 398 (1999) 6724-6726.
61 A.L. Sumner, P.B. Shepson, T.L. Couch, T. Thornberry, M.A. Carroll, S. Sillman, M.
Pippin, S. Bertman, D. Tan, I. Faloona, W. Brune, V. Young, O. Cooper, J. Moody and
W. Stockwell, J. Geophys. Res., (2001) in press.
62 A.L. Sumner, P.B. Shepson, A.M. Grannas, J.W. Bottenheim, K.G. Anlauf, D.
Worthy, W.H. Schroeder, A. Steffen, F. Domine, S. Perrier and S. Houdier, Atmos.
Environ. (in press).
63 J. Li, P.K. Dasgupta, Z.Genfa and M.A. Hutterli, Field Anal. Chem. Technol., 5
(2001) 2-11.
64 P.K. Dasgupta, Z. Genfa, J. Li, C.B. Boring, S. Jambunathan and R. Al-Horr, Anal.
Chem., 71 (1999) 1400-1407.
65 http://www.alphaomegapt.com/methanalyzer.htm
66 M.J. Navas, A.M. Jimenez and G. Galan, Atmos. Environ., 33 (1999) 2279-2283.
67 W.G. Gunz and M.R. Hoffman, Atmos. Environ., 24A (1990) 1601-1634.
68 A.V. Jackson, Crit. Rev. Env. Sci. Technol., 29 (1999) 175-228.
69 A. Sigg, A. Neftel and P.K. Dasgupta, in: Proceedings of the Workshop on the
Development of Analytical Techniques for Atmospheric Pollutants, Rome, April
13-15, 1992. Air PollutionResearch Report41 Commission of the European Commu-
nities, I. Allegrini (Ed.), pp. 111-117. Dordrecht, Boston, 1993.
70 Z. Genfa and P.K. Dasgupta, Anal. Chem., 64 (1992) 517-522.

156
 71 G. Zhang, P.K. Dasgupta and A. Sigg, Anal. Chim. Acta, 260 (1992) 57-64.
72 Z. Genfa, P.K. Dasgupta, G.M. Frick and W.A. Hoppel, Microchem. J., 62 (1999)
99-113.
73 J. Dommen, A. Neftel, A. Sigg and D.J. Jacob, J. Geophys. Res., 100 (1995)
8953-8966.
74 J. Li and P.K. Dasgupta, Anal. Chem., 72 (2000) 5338-5347.
75 http://www.cgenv.com/Narsto/
76 www.globalfia.com
77 J. Li and P.K. Dasgupta, Anal. Chim. Acta, 442 (2000) 63-70.
78 http://www.celgard.com/products/hfmproductinfo.cfm
79 Y. Komazaki, Y. Narita and S. Tanaka, Analyst, 123 (1998) 2343-2349.
80 Y. Komazaki, S. Hashimoto, T. Inoue, S. Tanaka, Atmos. Environ. (2002) in press.
81 Y. Komazaki, T. Inouie and S. Tanaka, Analyst, 126 (2001) 587-593.
82 Y. Komazaki, S. Hashimoto, T. Inouie and S. Tanaka, Atmos. Environ., (2002) in
press.
83 L. Bao and P.K. Dasgupta, Anal. Chem., 64 (1992) 991-996.
84 F.R. Rocha, J.A.F, da Silva, C.L. do Lago and I.G.R. Gutz, Abstract SE-79, Proceed-
ings, 11 Encontro Nacional de QuimicaAnalitica, Campinas, Brazil, September 2001.
85 S. Liu and P.K. Dasgupta, Anal. Chim. Acta, 308 (1995) 281-285.
86 Z. Genfa, P.K. Dasgupta and S. Dong, Environ. Sci. Technol., 23 (1989) 1467-1474.
87 R.M. Harrison and I.M. Msibi, Atmos. Environ., 28 (1994) 247-255.
88 L.L. S0rensen, K. Granby, H. Nilesen and W.A.H. Asman, Atmos. Environ., 28 (1994)
3637-3645.
89 C. Milford, M.A. Sutton, A.G. Allen, A. Karlsson, B.M. Davison, J.D. James, K.
Rosman, R.M. Harrison and J.N. Cape, Tellus, 52B (1990) 273-289.
90 J.C. Park, E.Y. Bae, C.G. Park, W.S. Han, Y.B. Koh, M.Y. Lee and J.G. Lee,
Proceedings of the Symposium on Ultrasonic Electronics. Jpn. J. Appl. Phys., 34
(1995) 6770-6773.
91 I.-H. Chang, D.-S. Lee, Y.-K. Lee and K.-J. Whang, Anal. Sci., 13 suppl. (1997)
267-272.
92 P.F. Lindgren and P.K. Dasgupta, Anal. Chem., 61 (1989) 19-24.
93 D.S. Lee, Yonsei University, Seoul, Korea. Personal communication, 2001.
94 D.S. Lee and I.H. Chang, Proceedings of the Insung Chromatech Conference on Ion
Chromatography, Seoul, Korea, August 13, 2001.
95 D.S. Lee, Abstract No. 94, Abstracts, 2001 International Ion Chromatography
Symposium, Oakbrook, IL.
96 U. Hase and Y. Matsuyoshi, NEC Res. Dev., 39 (1998) 95-100.
97 Y. Matsuyoshi, Y. Satoh, T. Shinozaki, E. Suzuki and N. Nagata, NEC Res. Dev., 41
(2000) 93-97.
98 Y. Satoh and Y. Matsuyoshi, NEC Res. Dev., 42 (2001) 314-318.
99 W. Frenzel, Fres. J. Anal. Chem., 342 (1992) 817-821.
100 W. Frenzel, Anal. Chim. Acta, 291 (1994) 305-320.
101 C.J. Patton and A.P. Wade, Continuous flow analyzers. In: G.W. Ewing (Ed.),
Analytical InstrumentationHandbook, 2nd Edn. Marcel Dekker, New York, 1997,
pp. 125-220.
102 J. Ruzicka and E.H. Hansen, Flow Injection Analysis, 2nd Edn. Wiley, 1988.
103 J. Crank and G.S. Park, Diffusion in Polymers. Academic Press, London, 1968.
104 J.E. Lundstrom and R.J. Bearman, J. Polym. Sci., 12 (1974) 97-114.
105 D. Scheppers, G. Schulze and W. Frenzel, Anal. Chim. Acta, 308 (1995) 109-114.
106 K. Toda, H. Inoue and I. Sanemasa, Bunseki Kagaku, 47 (1998) 727-734.
107 V. Kuban and P.K. Dasgupta, Anal. Chem., 64 (1992) 1106-1112.
108 G.A. Tarver and P.K. Dasgupta, Atmos. Environ., 29 (1995) 1291-1298.

157
109 G.A. Tarver and P.K. Dasgupta, Environ. Sci. Technol., 31 (1997) 3669-3676.
110 H. Kobara, K. Takeuchi and T. Ibusuki, Bunseki Kagaku, 46 (1997) 881-886.
111 Y. Suzuki, Anal. Chim. Acta, 353 (1997) 227-237.
112 M. Stigbrand, A. Karlsson and K. Irgum, Anal. Chem., 66 (1996) 3945-3950.
113 H.B. Ross, C. Johansson and C. de Serves, J. Atmos. Chem., 14 (1992) 411-423.
114 C. de Serves and H.B. Ross, Environ. Sci. Technol., 27 (1993) 2712-2718.
115 W. Matuszewski and M,E. Meyerhoff, Anal. Chim. Acta, 248 (1991) 379-389.
116 C. de Serves, J. Geophys. Res., 99 (1994) 25391-25398.
117 D. Trapp and C. de Serves, Atmos. Environ., 29 (1995) 3239-3243.
118 Y. Komazaki, M. Hiratsuka, Y. Narita, S. Tanaka and T. Fujita, Fres.J. Anal. Chem.,
363 (1999) 686-695.
119 K. Toda, P.K. Dasgupta, J. Li, G.A. Tarver, G. Zarus, S. Ohira and E.L. Loree, Anal.
Sci. (in press).
120 K. Toda, Kumamoto University, Japan. Personal communication, August, 2001.
121 D.C. Olson, S.R. Bysouth, P.K. Dasgupta and V. Kuban, Process Control Qual., 5
(2000) 259-265.
122 A.Yu. Alentiev, Yu. P. Yampolskii, V.P. Shantarovich, S.M. Nemser and N.A. Plat,
J. Membr. Sci., 126 (1997) 123-132.
123 I. Pinnau and L.G. Toy, J. Membr. Sci., 109 (1996) 125-133.
124 K.E. Hong and L.W. Burgess, Proc. SPIE, 2293 (1994) 71-74.
125 P.K. Dasgupta, Z. Genfa, S.K. Poruthoor, S. Caldwell, S. Dong and S.-Y. Liu, Anal.
Chem., 70 (1998) 4661-4669.
126 P,K. Dasgupta and S.-Y. Liu, US Patent 6,011,882, January 4, 2000.
127 H. Fein and S.-Y. Liu, US Patent 6,016,372, January 18, 2000.
128 M.R. Milani and P.K. Dasgupta, Anal. Chim. Acta, 431 (2001) 169-180.
129 0. Inya-Agha, S. Stewart, T. Veriotti, M.L. Bruening and M.D. Morris, Appl.
Spectrosc. (2002) in press.
130 H. Liu and P.K. Dasgupta, TrAC, 15 (1996) 468-475.
131 H. Liu and P.K. Dasgupta, Microchem. J., 57 (1997) 127-136.
132 S. Liu and P.K. Dasgupta, Anal. Chem., 67 (1995) 2042-2049.
133 E.P. Felix, C. Ugucione and A.A. Cardoso, Abstract EM-22, Proceedings, 11 Encontro
Nacional de Quimica Analitica, Campinas, Brazil, September 2001.
134 E.A. Pereira and P.K. Dasgupta, Int. J. Environ. Anal. Chem., 66 (1997) 201-213.
135 E.A. Pereira, A.A. Cardoso and P.K. Dasgupta, Quim. Nova, 24 (2001) 443-448.
136 M.R. Milani, J.A.C. Neto and A.A. Cardoso, Microchem. J., 62 (1999) 273-281.
137 A.A. Cardoso, H. Liu and P.K. Dasgupta, Talanta 44 (1997) 1099-1106.
138 K. Toda, M. Kitazaki and I. Sanemasa, Bunseki Kagaku, 49 (2000) 989-995.
139 H. Liu and P.K. Dasgupta, Anal. Chem., 67 (1995) 4221-4228.
140 http://www.geocities.com/ResearchTriangle/Lab/4688/
141 H. Liu and P.K. Dasgupta, Anal. Chim. Acta, 289 (1994) 347-353.
142 H. Amokrane and B. Caussade, J. Atmos. Sci., 56 (1999) 1808-1829.
143 A. Cardoso and P.K. Dasgupta, Anal. Chem., 67 (1995) 2562-2566.
144 S. Kar and P.K. Dasgupta, Am. Lab., 29 (1997) 17C-17M.
145 P.K. Dasgupta and S. Kar, Anal. Chem., 67 (1995) 3853-3860.
146 S. Kar and P.K. Dasgupta, J. Chromatogr., 739 (1996) 379-387.
147 K. Surowiec and P.K. Dasgupta, J. Microcol., 10 (1998) 265-271.
148 H. Huang and P.K. Dasgupta, Talanta, 44 (1997) 605-615.
149 H. Huang and P.K. Dasgupta, Electroanalysis, 9 (1997) 585-591.
150 Z. Genfa and P.K. Dasgupta, Anal. Chem., 72 (2000) 3165-3170.
151 P.K. Simon, P.K. Dasgupta and Z. Vecera, Anal. Chem., 63 (1991) 1237-1242.
152 Z. Vecera and P.K. Dasgupta, Anal. Chem., 63 (1991) 2210-2216.
153 Z. Vecera and P.K. Dasgupta, Int. J. Anal. Chem., 4 (1994) 311-316.

158
154 M. Taira and Y. Kanda, Anal. Chem., 65 (1993) 3171-3173.
155 A. Neftel, A. Blatter, R. Hesterberg and T. Staffelbach, Atmos. Environ., 30 (1996)
3017-3025.
156 A. Blatter, A. Neftel, P.K. Dasgupta and P.K. Simon, in: G. Angletti and G. Restelli
(Eds.), Physico-Chemical Behavior of Atmospheric Pollutants, Proc. 6th European
Symposium, Report EUR 15609/2 EN, Luxembourg, 1994. pp. 767-772.
157 A. Neftel, A. Blatter, A. Gut, D. Hogger, F. Meixner, C. Ammann and F.J. Nathaus,
Atmos. Environ., 32 (1998) 499-505.
158 Z. Zdrahal, P. Mikuska and Z. Vecera, Chem. Listy, 88 (1994) 353-359.
159 Z. Zdrahal and Z. Vecera, J. Chromatogr., 668 (1994) 371-374.
160 Z. Zdrahal, P. Mikuska and Z. Vecera, Anal. Chem., 67 (1995) 2763-2766.
161 J. Peskova, P. Parizek and Z. Vecera, J. Chromatogr., 918 (2001) 153-158.
162 W. Jaeschke, J.P. Dierssen, A. Gunther and M. Schumann, Atmos. Environ., 32
(1998) 365-371.
163 A.M. Costa, A.P.S. Novato, V.P. Campos and T.M. Tavares, Abstract AB-70, Proc., 11
Encontro Nacional de Quimica Analitica, Campinas, Brazil, September 2001.
164 D.R. Fitz and N. Motallebi, J. Air Waste Manage. Assoc., 50 (2000) 981-992.
165 D.S. Poon, D.Y.H. Pui, C.T. Lee and B.Y.H. Liu, J. Aerosol Sci., 25 (1994) 923 -934.
166 C.J. Tsai, C.H. Huang, S.H. Wang and T.S. Shih, Aerosol Sci. Technol., 35 (2001)
611-616.
167 S.M. Buhr, M.P. Buhr, F.C. Fehsenfeld, J.S. Holloway, U. Karst, R.B. Norton, D.P.
Parrish and R.E. Sievers, Atmos. Environ., 26 (1995) 2609-2624.
168 Y. Komazaki, H. Shimizu and S. Tanaka, Atmos. Environ., 33 (1999) 4363-4371.
169 I.H. Chang, N.H. Choi, B.K. Lee and D.S. Lee, Bull. Kor. Chem. Soc., 20 (1999)
329-332.
170 D.S. Lee, N.H. Yu, S.Y. Lee and J.S. Whang, Korean Patent 97-21338.
171 I.H.Chang, Ph.D. Dissertation, Yonsei University, Korea, August 2001.
172 W.C. Hinds, Aerosol Technology. Wiley, New York, 1982. p 161.
173 M. Keuken, C.A.M. Schoonebeek, A. Wensveen-Louter and J. Slanina, Atmos.
Environ., 22 (1988) 2541-2548.
174 G.P. Wyers, R.P. Otjes and J. Slanina, Atmos. Environ., 27A (1993) 2085- 2090.
175 J. Slanina and G.P. Wyers, Fres. J. Anal. Chem., 350 (1994) 467-473.
176 M.T. Oms, P.A.C. Jongejan, A.C. Veltkamp, G.P. Wyers and J. Slanina, Int. J.
Environ. Anal. Chem., 62 (1996) 207-218.
177 P.A.C. Jongejan, Y. Bai, A.C. Veltkamp, G.P. Wyers and J. Slanina, Int. J. Environ.
Anal. Chem., 66 (1997) 241-251.
178 M.A. Sutton, C. Milford, U. Dragosits, C.J. Place, R.J. Singles, R.I. Smith, C.E.R.
Pitcairn, D. Fowler, J. Hill, H.M. ApSimon, C. Ross, R. Hill, S.C. Jarvis, B.F. Pain,
V.C. Phillips, R. Harrison, D. Moss, J. Webb, S.E. Espenhahn, D.S. Lee, M. Hornung,
J. Ullyett, K.R. Bull, B.A. Emmett, J. Lowe and G.P. Wyers, Environ. Pollut., 102S
(1998) 349-361.
179 M.A. Sutton, E. Perthue, D. Fowler, R.L. Storeton-West, J.N. Cape, B.G. Arends and
J.J. Mols, Atmos. Environ., 31 (1997) 2615-2624.
180 D. Fowler, M. Coyle, C. Flechard, K. Hargreaves, E. Nemitz, R. Storeton-West, M.
Sutton and J.W. Erisman, Plant Soil, 228 (2001) 117-129.
181 V.R. Phillips, D.S. Lee, R. Scholtens, J.A. Garland and R.W. Sneath, J. Agric. Eng.
Res., 78 (2001) 1-14.
182 R.M. Carlson, Anal. Chem., 50 (1978) 1528-1531; US Patent 4,209,299, June 24,
1980.
183 H.S.M. Devries, F.J.M. Harren, G.P. Wyers, R.P. Otjes, J. Slanina and J. Reuss,
Atmos. Environ., 29 (1995) 1069-1074.
184 P.K. Simon and P.K. Dasgupta, Environ. Sci. Technol., 29 (1995) 1534-1541.

159
185 X. Li-Jones, D.L. Savoie and J.M. Prospero, Atmos. Environ., 35 (2001) 985-992.
186 http://www-wlc.eas.gatech.edu/supersite/
187 Z. Genfa, B. Hartsell, J. Slanina, K. Baumann, C.B. Boring, P.A.C. Jongejan, E.
Edgerton and P.K. Dasgupta, J. Geophys. Res. (submitted)
188 http://www.utexas.edu/research/ceer/texaqs/
189 K. Ito, C.C. Chasteen, H.-K. Chung, S.K. Poruthoor, Z. Genfa and P.K. Dasgupta,
Anal. Chem., 70 (1998) 2839-2847.
190 K. Rosman, M. Shimmo, A. Karlsson, H.-C. Hansson, P. Keronen, A. Allen and G.
Hoenninger, Atmos. Environ., 35 (2001) 5301-5310.
191 R. Hesterberg, A. Blatter, M. Fahrni, M. Rosset, A. Neftel, W. Eugster and H.
Wanner, Environ. Pollut., 91 (1996) 21-34.
192 M. Ammann, M. Kalberer, DT. Jost, L. Tobler, E. Rossler, D. Piguet, H.W. Gaggeler
and U. Baltensperger, Nature, 395 (1998) 157-160.
193 C. Zellweger, M. Ammann, P. Hofer and U. Baltensperger, Atmos. Enuiron., 33
(1999) 1131-1140.
194 M. Loflund, A. Kasper-Giebl, W. Tscherwenka, M. Schmid, H. Giebl, R.
Hitzenberger, G. Reischl and H. Puxbaum, Atmos. Environ., 35 (2001) 2861-2869.
195 J.P. Ivey, J. Chromatogr., 287 (1984) 128-132.
196 Z. Genfa, T. Uehara, P.K. Dasgupta, T. Clarke and W. Winiwarter, Anal. Chem., 70
(1998) 3656-3666.

160
Chapter 6

Automated measurement of atmospheric

particle composition
Purnendu K. Dasgupta and Simon K. Poruthoor

6.1 INTRODUCTION

In the previous chapter the diffusion-based collection and determination of

gases was addressed. The importance of aerosols and the present lack of com-
mercial instrumentation to make chemically specific measurements were also
mentioned. Measurement methods for physical characteristics of atmospheric
aerosols have been recently reviewed [1]. For recent developments in continuous
aerosol mass monitoring, the reader is referred to Babich et al. [2]. In this
chapter the primary focus is on near real-time analysis of ionic constituents in
atmospheric aerosol. However, other approaches to aerosol composition mea-
surement are also briefly covered. First, a preamble is necessary as to the impor-
tance of aerosol composition measurement.

6.1.1 The need for chemically specific measurement of aerosol

composition (speciation monitors)

Several investigators have concluded that increased mortality is directly

correlated with aerosol concentrations [3-7]. The importance of chemical com-
position of aerosol particles on health, and the inadequacy of the extant practice
of chemically nonspecific particle measurements was recognized very early [8].
After studying mortality data for hundreds of thousands of adults, Pope et al. [3]
concluded that particulate air pollution was associated with cardiopulmonary
and lung cancer related mortality, fine particles in general and sulfate in
particular being of special concern. In contrast, others have suggested that the
primary damage is caused by gaseous pollutants, and aerosols act only as vectors
or modifiers [9,10]. Still other studies have suggested that exposure to acid
aerosols results in decreased pulmonary function, increased hospital admissions
for asthma and other respiratory ailments [5,11-15]. Recent evidence indicates
synergism between ozone damage to the lung and acid aerosols of very small size
[16]. Lack of specific epidemiologic correlations may be due more to the lack of a

Comprehensive Analytical Chemistry XXX,7I

J. Pawliszvn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 161
sufficient database of chemically specific measurements, rather than the health
effects of aerosols being independent of their chemical composition [17]. Most
measurements are carried out for regulatory purposes. Chemical composition
information on the PM2,, fraction (particles with aerodynamic diameters <2.5
im) is still scanty [18]. Epidemiological studies may show relationships but do
not reveal the causality or mechanics of particulate matter induced health
effects. Controlled toxicological studies have shown that specific constituents of
suspended particulate matter are associated with specific health effects, but
such studies have generally been conducted at aerosol concentrations far in
excess of those found in ambient air [19,20]. Organic compounds in automobile
exhaust [21,22], sulfuric acid [11], traces of heavy metals [23,24], etc. have all
been suggested as causing specific ailments. It is difficult to directly relate
laboratory generated toxicologic data to the long-term effects of complex mix-
tures of aerosols to which people are actually exposed in ambient air. The lack of
a sufficient database of chemical composition information of atmospheric aero-
sols can be attributed largely to the lack of instrumentation that can provide this
information in a reasonably cost-effective manner. Many atmospheric species
exhibit short half-lives and concentrations change rapidly. This calls for fast
response measurements, ideally in real time.

6.1.2 Aerosols and gases: an inseparable link

The role atmospheric aerosols play depends on their concentration, chemical

composition, and particle size. These factors may be influenced by the trace gas
concentrations surrounding the particles. Many atmospheric gases exist in
reactive equilibrium with species in the aerosol phase [25-31]. The formation of
fine particles itself involves a number of processes which includes chemical
reaction, nucleation, condensation, coagulation, and evaporation of fog and
cloud droplets in which gases have dissolved and reacted [32]. A clear under-
standing of atmospheric chemistry and air pollution problems requires reason-
able information on the chemical composition of aerosol particles and the
gaseous components. Similarly, other parameters such as pathogens, tempera-
ture, relative humidity, etc. must be evaluated concurrently in any study related
to health damage caused by ambient gases and particles. The lack of such
information leads to the present uncertainty about chemical aspects of the
toxicology of ambient aerosols [33] and lengthy debates on regulatory issues
[18,34].

6.2 SAMPLING AND ANALYSIS CONCEPTS

Atmospheric aerosols exhibit a multimodal size distribution; up to four different

size ranges can be of importance [18]. Particles in each size range are formed
through different dominant routes, thus leading to different chemical composi-
tion in each size fraction [35-411. Therefore, an aerosol analysis system should

162
ideally provide size discriminated collection and analysis in all size ranges of
interest. If only a composite analysis can be performed, it is vital that the aerosol
is quantitatively collected in all the size ranges of interest. Even a small bias
towards collection of larger particles can affect the overall composition result
significantly because larger particles can represent a much larger mass fraction.
In filter collection methods, it is not difficult to select suitable filters to achieve
quantitative collection across the entire size ranges of interest. Not all com-
monly used filter media are actually quantitatively retentive of fine particles
[42,43].
There are two general approaches to aerosol composition measurements.
Traditionally, particles are sampled on some media and later taken to the labora-
tory and analyzed. Many of the instruments available for National Ambient Air
Quality Standard (NAAQS) compliance measurements were compatible with
this approach and the bulk of the extant data on chemical composition informa-
tion of ambient aerosols has been obtained by this technique. Atmospheric aero-
sols can be collected by inertial methods, gravitational settling, centrifugation,
electrostatic and thermal precipitation, and filtration. Of these, filtration, with
or without inertial classification, is the most commonly used approach [27,
44-53]. Collection efficiencies and other characteristics of various filters have
been evaluated and are available in the literature [54]. More often than not,
aerosol samples are collected for an extended period of time, usually many hours.
The collected samples are then stored and later analyzed. It is generally not prac-
tical to complete the analysis step in the field: only the sample collection step is
carried out in the field [18]. The media containing the collected aerosol are
placed in envelopes or sealed containers and transported to a regular laboratory
for analysis. The total elapsed period may range from days to months. Obviously,
such an arrangement can neither provide good temporal resolution nor measure
unstable species. The integrity of the process may be compromised when sam-
ples are stored for long periods of time: collected species may undergo various
chemical and physical changes and therefore the analytical results may not be
meaningful. There are also problems particular to specific analytes and specific
collection/analysis methods (e.g., sampling surfaces absorb or react with gases
and particles leading to positive or negative artifacts) [28,53, 55-62]. As dis-
cussed in a previous chapter, most often it is necessary to remove potentially
interfering gases with a diffusion denuder before the aerosol is collected or oth-
erwise processed. Discrete sampling and analysis is very difficult to automate
and is consequently labor-intensive. In the long run, large-scale measurements
by such approaches are not economically attractive.
In the alternative approach, airborne particles are processed directly
through a real-time or near real-time dynamic measurement system. The collec-
tion (if there is a discrete collection step) and analysis systems are integrated and
the entire system is field deployed, permitting rapid data output and signific-
antly lower cost per analysis. Methods of different types have been reported for
the automated determination of several individual species in aerosols.

163
6.2.1 Thermal decomposition based methods

Thermal decomposition of particles under controlled temperature conditions,

generally to characteristic gases that are measured, was one of the first means of
measuring specific aerosol constituents; early on, sulfur at nanogram levels in
atmospheric aerosol was measured by such techniques. Crider [63] first demon-
strated real-time determination of aerosol sulfur with a flame photometric
detector (FPD) by switching a sulfur dioxide removing filter in and out of line.
Husar et al. [64] reported a flash vaporization-flame photometric detection
(FV-FPD) system for measuring particulate S on filters. Roberts and Fried-
lander [65] measured sulfur directly from collection plates of cascade impactors.
In a typical early method, potentially interfering gases were first removed
and the aerosol stream was then thermally decomposed, under controlled tem-
perature conditions, to characteristic gases that were collected by a diffusion
denuder and then measured periodically. Much of the effort was directed to the
specific measurement of sulfuric acid and the various ammonium sulfates
[66-71]. One of these, originally developed by Harvard researchers, was later
adopted as a standard method [72]. Starting in the mid-1980s, several sophisti-
cated instruments were developed based on diffusion denuders and differing
volatilities of different compounds. Slanina et al. [73] described a computer- con-
trolled thermodenuder that measured sulfuric acid and ammonium sulfates by
first removing all S gases, then using two Cu-CuO coated serial denuders main-
tained respectively at 120 and 240°C. The first one vaporizes and collects H2SO4
and the second decomposes ammonium sulfates and collects the HSO4 . The
denuders are then heated sequentially to 800°C to liberate SO2 measured by a
flame photometric detector. Limits of detection (LODs) were 20 ng/m 3 for either
species for a 1-h sample. (See also comments by Huntzicker [74] on this work.) In
Lindqvist's instrument [75], gases are similarly removed with predenuders and
the H2SO4 is then volatilized and collected on an MnO2-PdO coated denuder
maintained at 138°C. The collected sulfate is decomposed at 800°C and the liber-
ated S is converted to H2S and detected by gas chromatography-photoionization
detection. The LOD was 60 ng/m for a 12 min sample.
Recently, Hering and Stolzenburg [76] described an integrated apparatus
with a collector and vaporizer cell for collecting airborne particles from sample
gas by impacting particles at low pressure on a stainless steel collector strip. The
sample (drawn at 1 l/min) is passed through a pre-impactor (to remove particles
larger than 2.5 gm), a multi-lumen carbon-impregnated ceramic honeycomb
denuder, and then through a Nafion humidifier. A vented housing around the
pre-impactor, denuder, humidifier and cell maintains sample gas temperature
near ambient temperature. Following sample collection for a desired period, a
power source connected to the strip rapidly heats it in a stream of N2 and the
system plumbing now directs the liberated SO2 to a pulsed fluorescence type SO2
analyzer. Ambient particles in the 0.10-0.40 im size range passes through the
humidifier with 97-102% efficiency. Prehumidification of the particles greatly

164
reduces bounce-off from the impaction surface; laboratory generated (NH4 )2 SO4
aerosols in the 100-800 nm size range were collected with 95-99.8% efficiency
[77]. A commercial version of this instrument is now available with a time
resolution of 10 min and a concentration resolution of 200 ng/m3 [78].
Whereas the Hering and Stolzenburg approach is based on actual collection
by impaction, a recent strategy developed by Harvard School of Public Health
researchers [79] is more reminiscent of previous thermodenuder based sulfate
measurement/speciation approaches. Referring to Fig. 6.1, the instrument is
zeroed by allowing it to sample particle-free air that is made to pass by the
sample inlet. A small amount of ammonia is continuously added to the sampled
air stream from an on-board permeation source. The amount is not critical as
long as it is sufficient to neutralize the aerosol sulfate completely to (NH4 )2 SO4
and thus avoid any difference in thermal behavior of (NH4 )2 SO4 from more acidic
sulfates, particularly H2 SO4 . However, large excess should be avoided since NH3
is oxidized to NO and contributes to the baseline noise. In urban areas, where
the aerosol is generally not present as H2 SO4 , it is not necessary to add any NH3
at all. During sampling, the instrument draws ambient air through an inlet
system that removes particles larger than 2.5 gm. Ammonia (few hundred ppb at
10-15 cc/min) is then added to the 0.65 1/min sample stream (the ammonia flow
rate is kept small relative to the main stream so dilution is insignificant). The
mixture proceeds through an alkaline denuder that removes acidic sulfur gases.
The sulfur-gas free aerosol stream then proceeds through a thermal converter (6
ft long stainless steel coil maintained at 800°C), which converts all the sulfate to
SO,. The conversion of sulfate to SO2 is highly efficient. The carbon present in
the steel may play a role in this conversion, but this has not yet been confirmed.
Ambient Air In Systiem Inlet
+ PM,,-lmpacto'r or Cyclone Inlet
I Particle Filter D 7

r' Flow Restriction 2

n Z-Air Solenoid Valve
Approx. 0.8 LPM NH, Permeation Tube

I Approx. 5-10 cc/ nin

Flow Restriction
Approx. 0.65 LPM

Sulfur Gas Scrubber

I

TEl Model 350 Converter

316 Stainless, @ 800°C

Fig. 6.1. Schematic of Harvard

School of Public Health (HSPH) Z-Air Solenoid
PTFE Filter (2 pm) S
Valve Control
continuous sulfate analyzer Signal Modified TEl
(patent applied for). Courtesy Model 43C Trace SO2
Level CO2 Analyzer
George A. Allen, HSPH.

165
Any residual particles are removed from the gas stream prior to measurement of
SO2 concentration with a modified high sensitivity pulsed fluorescence based
analyzer. The system differs from previous thermodenuder-based approaches in
several ways. The most important difference is that the approach is truly
continuous; there are no discrete collection and analysis steps. From a practical
point of view, it also does not rely on the FPD and is therefore intrinsically safe
and does not need compressed gases. Although it is not a factor in urban
measurements, the sensitivity of current SO 2 analyzers is such, however, that
long integration times are needed to measure low sulfate concentrations. The
LOD is 0.5 g/m3 for a 1-h mean, which is equivalent to twice the standard
deviation of clean ambient air. The precision of collocated prototype instruments
for 1-h mean values in urban areas is -7%. It may be better in eventual
commercial versions [80]. A patent has been applied for. The general agreement
with other methods, e.g., ion chromatography (IC) is reportedly excellent.
Similarly, chemiluminescence (CL) detection of liberated NOx has been used
for some time to determine nanogram quantities of nitrate after the thermal
decomposition of nitrate particles [81]. Cox 82] measured nitrite and nitrate in
solution to liberate NO, reducing nitrite with acidic iodide and reducing total
nitrite/nitrate with FeSO4 and ammonium molybdate in conc. H2SO4. The NO
was detected by CL. Klockow et al. [83] described two different approaches for
the measurement of NH4 NO3. In both, all other N species are removed with two
predenuders. NH4NO3 is decomposed and volatilized and collected using either
an MgSO 4 coated denuder at 150°C or a NaF-coated denuder at 140°C. After
collection, the denuder is heated to 7000C and the liberated NOx is detected with
a CL-based monitor. LODs are 60-100 ng/m3 NH4 NO3 , depending on sampling
time and sampling rate.
In coastal areas, significant concentrations of aerosol NaNO3 may be present
from the reaction of gaseous HNO3 with NaCl. It was suggested that NaNO3 and
NH 4NO 3 could be differentiated by their different temperatures of volatilization
[84]. Subsequent work by Sturges and Harrison [85] has shown, however, that
such differentiation is in fact fraught with difficulty; they show that it does not
work for the corresponding chloride case either. More recently, Ten Brink et al.
[86] have reevaluated the thermodenuder based determination of NH4NO 3.
Aside from some problems encountered at high humidities, this instrument is
reported to work well.
Yamamoto and Kosaka [87] collected aerosols on type 304 stainless steel
strips and applied the FV approach with CL-based NO detection. To assure
conversion of NO2 to NO, they added CrO3. This resulted in quantitative
production of NO from NaNO. With NH4 NO,, the NO recovery was only
30-50% and artifact NO production was seen from (NH4)2SO 4 and NH4C1.
Addition of NaOIH prior to heating removed the NH4+ as NH3 and eliminated
this interference.
The recent work of Stolzenburg and Hering [76,77] was mentioned earlier in
connection with a continuous sulfate monitor. Indeed, the same apparatus can

166
 Inlet
Precut
-and
Denuder

Ventilated
Enclosure

p
Fig. 6.2. Stolzenburg-Hering apparatus schematic
for the impaction-flash volatilization measurement
of sulfate and nitrate. Reproduced with permission
from Ref. [77] (copyright 2000 American Chemical
Sointv).

be used to measure nitrate with a CL-based NOx monitor; the system setup is
shown schematically in Fig. 6.2. Laboratory generated NH 4 NO3 particles in the
200-800 nm size range were collected with 97-99.8% collection efficiency. The
FV conditions were optimized to elicit minimum response to ammonium, the
response to (NH4 )2SO4 was only 4% of that of a corresponding amount of nitrate.
The possible vaporization of NH4 NO, during collection was investigated. No
significant losses (2±4%) were found. The LOD is primarily governed by the
field blank, which arises due to the adsorption of nitrogen bearing gaseous
material (that are not removed by the denuder) on the impactor surface. This
could be as high as 0.4-0.7 ptg/m3 in high NOx environments. Nevertheless,
excellent agreement with traditional measurement methods has been generally
observed. In intercomparison studies with more traditional denuder-filter mea-
surement methods in three cities, correlation coefficients were greater than 0.97
and regression slopes ranged from 0.96 to 1.06. Over prolonged field experi-
ments, the system exhibited data recovery rates as high as 97%. A commercial
version of this instrument is now available with a time resolution of 10 min and a
concentration resolution of 200 ng/m 3 [88]. Inter-instrument precision is excel-
lent, the output of two collocated instruments are shown in Fig. 6.3.

6.2.2 Post-sampling aerosol analysis: spectroscopic methods

Spectroscopic methods for aerosol analysis have generally been used only in a
post-sampling off-line mode. The determination of aerosol carbon is an excep-
tion; this is discussed in a later section. Techniques available for analysis of
aerosol particles, once they have been collected, are of course, quite numerous.
An exhaustive review of these approaches is not appropriate here; the present

167
 or n
l JUU ?
A -Nitrate, Instrument 1 ONitrate, instrument 2
O 10.0
E
2
. -
1
20.0- i i i
E0
4~~~~~~~~~~~~~~~~~~~

c:

z
o.1 0.I4Ak,;k&A4*V _1"
0.10 1.00 10.00 8114100 8115/00 8116100 8117/00
Nitrate, pg/m3, Instrument 1 Date

Fig. 6.3. Precision of two collocated nitrate measurement instruments based on impaction-flash
volatilization. Data Courtesy of Susanne V. Hering, Aerosol Dynamics Inc.

focus is on automated methods. Briefly, the general application of optical tech-

niques used for the analysis of collected aerosols was reviewed by Appel [89].
X-ray fluorescence has been commonly used for elemental analysis [90]. Analysis
of single particles has been carried out by various optical methods [91]. Electron
microscopy of single particles or aggregates, coupled to high resolution X-ray
spectrometry, has been clearly established to be one of the most powerful
techniques for the examination and characterization of collected aerosols [92].
Fourier transform infrared spectroscopy also has particular merit, especially in
examining sulfate aerosols and in determining the extent to which acid sulfates
exist in such samples [93-95].

6.2.2.1 Interferences from prolonged sample collection

Air samples are not spatiotemporally homogeneous. Prolonged sample collection
on a filter with intended post analysis can generate positive or negative artifacts,
especially when atmospheric acids, bases, and their salts are the analytes of
interest. The gaseous environment contacting aerosols affect their chemical
composition. This is especially true where the measurement protocol calls for
collection of the particles on a medium that can absorb gaseous species. Even
when the medium is inert, gases can be adsorbed on co-collected particles.
Interference from gaseous nitric acid in determining aerosol nitrate collected
on filters/impactors has been extensively studied [96,97]. This type of HNO3
retention increases with sampling duration, as well as with the alkalinity of the
collection medium/alkalinity of co-collected particles. Similarly the adsorption of
SO2 on the collection media can cause positive interference in aerosol sulfate
measurement. Negative errors in the measurement of particles collected on
filters occur due to volatilization losses during and after collection; reaction with
other collected particles may also lead to artifacts [98,99]. Filters that do not
retain HNO3 well also tend to lose particulate NH4 NO3 by volatilization.
Particulate nitrate measurements that are simply based on collection on Teflon
or quartz fiber filters are thus suspect. Nylon filters, often used to collect both

168
gaseous and particulate nitrate, can cause partial retention of SO 2 and are not
reliable for particulate sulfate measurement. The positive interference from SO2
is variable and can be as high as 70% [28,100-102]. Sulfate measurements with
filters of relatively low alkalinity (Teflon and quartz fiber) tend to be more
reliable. Similar interferences from the adsorption of organic vapors on the
collection media occur for organic and elemental carbon measurement as well.
The use of diffusion denuders for removal of gases prior to collection of
particles, discussed in detail the foregoing chapter, is by far the best way to
prevent artifacts due to reactions with gases. Many aerosol chemical speciation
studies as well as many atmospheric gas measurement efforts have utilized
diffusion denuders [103-107].
The loss of particles by volatilization or artifact generation due to
particle-particle reactions are not, however, eliminated by the use of diffusion
denuders. The only way to ameliorate this is to use a small collection period.

6.2.3 Aerosol mass spectrometry

The application of mass spectrometry (MS) to aerosol analysis has had a long
and illustrious history. Real-time measurement via ionization of collected parti-
cles followed by MS analysis was demonstrated more than two decades ago [108].
Mansoori et al. [109] demonstrated the quantitation of ionic species in single
microdroplets by on-line laser desorption/ionization. For real-time aerosol sam-
pling and analysis, if size information is obtained for the same particle for which
chemical composition data are generated, the resulting size-dependent compo-
sition information will be particularly valuable. As the sample is drawn through
the system across a laser beam path, the size information is deduced from the
magnitude of the scattered light registered on a photodetector placed at right
angles to the beam path. A second high-power pulsed laser is then used for the
photoionization of the constituents of the aerosol particle and this is followed by
quadrupole MS [110]. This arrangement provides a brief ion burst and does not
allow m/e scanning of the quadrupole within the time available for a single
particle. This limitation was removed by using time-of-flight mass spectrometry
(TOFMS) instead of a quadrupole MS and then further developed to permit
simultaneous analysis of both positive and negative ions from a single particle,
by adding a second time-of-flight tube [111,112]. While such an arrangement
provides satisfactory results with monodisperse aerosols, difficulties are still
encountered when particles of different size are present. This is because the
second ionizing laser pulse illuminates the particle at a different spatiotemporal
coordinate, after it has passed through the sizing beam traverse. The travel time
between the two illumination zones is size dependent. Theoretically it is possible
to scan for different size particles, but it has not been reduced to practice. This
difficulty is eliminated by firing the high energy ionizing pulse immediately after
the sizing pulse, so that the particle effectively does not travel any appreciable
distance within this time [113-116]. Actual applications for the measurement of

169
atmospheric aerosol particle composition have been demonstrated 117,118].
Mobile versions of aerosol mass spectrometers have been reported [119-121].
In the last few years, single particle mass spectrometry has blossomed; for a
more detailed account the reader is referred to several recent reviews [122-124].
Maynard [125] has examined the various methods available for the characteriza-
tion of single ultrafine (<100 nm) particles to see how different methods can
determine features in terms of size, morphology, topology, composition, structure
and physicochemical properties. Size, morphology and surface properties are best
characterized by scanning transmission electron microscopy (STEM), while high-
resolution transmission electron microscopy allows structural information on
particles and atomic clusters to sub-0.2 nm resolution. Electron energy loss spec-
troscopy and X-ray emission in the STEM allow the chemical analysis of particles
and particle regions down to nanometer diameters. Scanning probe microscopy
offers the possibility of analyzing nanometer-diameter particles under ambient
conditions. Developments in aerosol mass spectrometry are providing the means
for chemically characterizing size-segregated ultra-fine particles down to 10 nm in
diameter.
Lazar et al. [126] have shown that it is possible to determine just the surface
composition of individual airborne particles by separating desorption and ioniza-
tion steps using a two-laser real-time aerosol mass spectrometry technique. A
weak excimer laser pulse is first used to desorb the semivolatile components
from the particle surface when the particle is in the center of an ion-trap MS.
After a short delay, another excimer laser pulse is used to ionize the semivolatile
surface components in the gas phase and these ions are then subjected to mass
analysis. The conventional one-laser and the two-laser technique above produce
complementary results. The two-laser technique was found to be particularly
suitable for determining polynuclear aromatic hydrocarbons (PAHs) adsorbed
on particle surfaces. PAH compounds adsorbed on contaminated soil particles
have been studied by this group [127]. These authors later reported the first
direct observation of the evolution of the carbonization process in a soot-forming
flame [128,129]. The degree of carbonization of each measured particle, and the
size distribution of both PAH-containing and mature soot particles were deter-
mined. The carbonization process is characterized by rapid hydrogen exchange
between the PAH and pyrolytic addition of small hydrocarbons to form larger
PAH molecules. The hydrogen exchange rate builds until carbon- carbon bond
rearrangement becomes facile. The same authors have shown elsewhere that
even a submonolayer amount of tri(n-)butylphosphate can be monitored in real
time on the surface of micron-sized particles by forming alkali metal ion adducts,
and this can be done with relatively low power lasers because of the ease of pro-
ducing alkali metal cations [130]. Paraquat on the surface of particles has also
been monitored [131].
Real-time aerosol mass spectrometry provides unparalleled power and versa-
tility: the current generation of instruments provide rapid, high throughput
data on both size and chemical composition. Nevertheless, the method is not per-

170
fect: the cost, size, power and operator skill requirements are all substantial.
Quantitative analysis can be matrix and size dependent [132-134]. Kane and
Johnston [133,134] have shown that these dependencies arise from the trans-
mission characteristics of the particle inlet and the intrinsic ability of a particle
to be vaporized and ionized. Small particles are generally more difficult to detect
and analyze than larger particles because of the difficulty of focusing through
the inlet into a tight beam as well as the difficulty of ablation. Particles composed
of PAHs, NH4 NO3 , and alkali metal salts are efficiently ablated by a laser pulse.
Aliphatic organics are less efficiently ablated and (NH 4)2SO4 is very difficult to
detect in a positive ion spectrum. Ultrafine particles fragment so extensively
into small mass fragments that it becomes difficult to distinguish between
aliphatic and aromatic components.
The massive amount of raw data generated by real-time single particle mass
spectrometry is both a blessing and a curse. A significant amount of effort must
be spent to distil the desired information. Despite innovative efforts towards
adapting new and old algorithms to provide artificial intelligence for classifying
the spectra and deriving information from it [135], it is clear that this is not yet a
routine operation. While an aerosol mass spectrometer represents a first rate
research tool, the time for routine ambient monitoring by this technique has not
yet come. It should also be noted that aerosol mass spectrometers differ con-
siderably in their design, performance, and nature of data output. In the Atlanta
Supersite study of 1999 [136], four different aerosol mass spectrometers were
operated at the same site for the first time and the data obtained was both
complementary and overlapping. For an account of this study, see Middlebrook
et al. [137].

6.2.4 Aerosol metal concentrations by real-time emission

spectrometry

Trace metals in aerosols can often be used as unique signatures for a specific
source. This is particularly valuable for source apportionment and other issues
that are otherwise very difficult to address. If such detection/measurement can
be done in real time, there is the exciting possibility of performing eddy
correlations for flux measurement and source apportionment. In favorable
cases, even simple flame emission spectroscopy can be very sensitive. Clark et al.
[138] report that a single 100 nm diameter NaCl particle can be detected with a
modified flame photometer, and this is free from matrix interferences.
A broad multi-element detection scheme in which an air sample is directly
aspirated into a plasma source will seem to be of general use and wide interest.
Unfortunately, most inductively coupled argon plasmas tend to become unstable
when significant amounts of air are introduced. To this end, Bacri and Gomes
initiated studies with air and air-argon plasmas [139]. A preliminary apparatus
for direct analysis of atmospheric aerosols was then introduced [140]; a more
developed version was shown to be adequate for workplace monitoring [141].

171
With an air plasma system, the sample to be analyzed is used directly as the
plasma gas. If an argon plasma is used, the sample to be monitored is injected at
a low flow rate (0.5 l/min) into the plasma. LODs ranged from 1-10 Ag/m 3 with a
response time of less than 1 min [142]. The authors have modeled relevant
relaxation processes in an air plasma [143]. It has also been subsequently shown
that a 50:50 argon:air plasma (total flow 15 l/min) gives a sufficient sample of
airflow for analysis and appreciably lowers the LOD [144].
None of the above instruments are readily transportable. More recently,
Duan et al. [145,146] have described a field-portable low-power instrument
which uses a microwave induced argon plasma that can tolerate up to 20% air.
Although the limits of detection (LODs) are not as yet good enough for use in
ambient air in any of the above instruments, coupled to an inertial particle
concentrators [147-149] such an approach may be practical.

6.2.5 Measurement of aerosol carbon

Having examined 1300 samples from 66 urban and rural sites in the US, Shah et
al. [150] reported that carbonaceous particles constitute an average of 13% of the
total suspended particulate mass. About two-thirds of the carbon may be
organic, derived mostly from combustion. Gray et al. [151] found 40% of the fine
particle mass (2.1 Mm) in Los Angeles to be carbonaceous. Offenberg and Baker
[152] found 49% of summertime fine particle (0.15-0.45 Am) mass to be organic.
On a molar basis, carbon is the most common element in atmospheric fine parti-
cles [153]. Organic and elemental carbon (OC and EC) in such particles are the
primary contributors to visibility degradation. Light extinction by OC occurs
primarily by scattering (with an efficiency of 2-5.5 m2 /g) and EC (this is gener-
ally equivalent to the terms black carbon (BC) or graphitic carbon (GC)) is the
major contributor to light absorption (with an extinction efficiency of -10 m2/g
at 515 nm and varying as 1/) [154]. EC has also been suspected to be a catalytic
agent in the conversion of SO2 to acid sulfates [155] and NO2 to HONO [156,
157]. Inhalation of soot particles from diesel-powered buses can be a threat not
only for those outside, but also those inside a bus [158].
Naturally, the measurement of carbonaceous particles has long been
regarded as an important challenge. The first near real-time measurement was
accomplished by Hansen et al. [159,160]. This instrument, named the aethal-
ometer, measures the reduction in light transmission through a filter as air is
sampled through it. The measurement is generally referred to that of black car-
bon. The original instrument used laser illumination. For a time, commercial
versions of the aethalometer utilized incandescent sources. Currently three fully
automated versions are available with self-advancing quartz-fiber filter tapes,
adjustable sampling rates in the 2-6 1/min range and light emitting Diode (LED)
based light sources at (a) 880 nm, (b) simultaneously at 880 and 370 nm and (c)
simultaneously at 370, 450, 571, 615, 660, 880 and 950 nm [161]. The aethal-
ometer has been used from experiments in the poles [162] to airborne experi-

172
ments [163] including measurements over the Siberian Sea. With high flow
rates, a 10 s temporal resolution is possible with an LOD of 10 ng/m3 [164]. EC is
by far the most important contributor to light absorption by atmospheric aero-
sols; OC, aerosols containing Pb or Fe contribute insignificantly. For these
reasons, the aethalometer is the most popular instrument used for BC measure-
ment [165]. However, to obtain EC concentrations in 'Lg/m 3, it is necessary to use
a value for the specific attenuation cross section. Most investigators simply use
the value supplied by the manufacturer [166]. Although this value is stated to be
10 m2/g, in practice this can vary by a factor of two either way depending on the
aerosol matrix, whether the aerosol is an internal or external mixture and the
aerosol age [167]. It is, of course, also dependent on the nature of the illumina-
tion source and the filter type [168]. Comparison between different aethalo-
meters using different wavelengths and known test aerosols composed of salt,
graphite, and iron oxide has shown that the aethalometer response depends not
only on the absolute amount of the absorbing species on the filter, but also on the
ratio of the absorbing to the nonabsorbing component [169]. Petzold et al. [170]
have also reported on the dependence of the specific attenuation cross section on
the mass fraction and particle size. The fact that the measurement is made on a
scattering matrix (a fiber filter) complicates the measurement. Kopp et al. [171]
carried out an investigation on the specific attenuation cross section of aerosols
deposited on fiber filters with a polar photometer. Interestingly, they found that
at a specific angle (reflected hemisphere 0 = 165°), the specific attenuation cross
section of EC is less dependent of the aerosol matrix than at any other angle.
Ballach et al. [172] have immersed sampled filters in a matching refractive index
oil to eliminate many of the problems from the filter matrix, and report good
agreement with the polar photometer/integrating sphere techniques and Mie
scattering calculations.
Other spectroscopic techniques have been developed that provide a more
direct measurement of BC. Direct photoacoustic spectroscopy using the 514.5
nm line from an Argon ion laser was first validated against a filter collection
technique by Adams et al. [173]. This method permits in situ measurement
without filter collection, but one still must assume a value of the specific
attenuation coefficient for BC. The instrument signal is also affected by humid-
ity and temperature variations and needs to be frequently recalibrated. These
authors subsequently applied the technique to the real-time measurement of BC
in Los Angeles [174]. This was one of the first methods without filter collection
to show a diurnal pattern in BC with low values in night with concentrations
increasing around sunrise and with persistent higher values through mid-
afternoon [175]. They found similar patterns in other urban areas as well as in
rural locations. They noted that visibility degradation due to aerosol light
absorption assumes a relatively greater importance under less polluted condi-
tions. For more recent work on photoacoustic sensing of BC, see Petzold and
Niessner [176] and Arnott et al. [177]. Niessner [178] also developed a photoelec-
tric aerosol sensor which can be used for BC measurements [179].

173
 Other spectroscopic techniques, notably Raman and Fourier Transform
Infra Red (FTIR) spectroscopies, have both been used for the determination of
BC on filer substrates. Keller and Heintzenberg [180] used a diode pumped
Nd3+-YAG laser at 1064 nm in a backscattering mode with a defocused beam
diameter of -1 mm on a polycarbonate filter with an incident power of 20-270
mW to carry out Raman spectroscopy with a liquid nitrogen cooled Ge detector.
Pollard et al. [181] used infrared transmission in the 650-666 cm-' region to
measure BC on a PTFE filter by FTIR spectroscopy. One clear attraction of this
method is that it allows the simultaneous determination of ammonium, nitrate,
sulfate and several organic compounds as general classes (see Section 6.2.2).
Seasonal measurements of BC generally show lower values in the summer
and higher values in the winter [182,183]. In pristine mountain tops where
precipitation removes suspended matter, e.g., atop Mt. Jungfraujoch in
Switzerland, the opposite behavior is observed [184]. Pakkanen et al. [185]
report little seasonal variation in Helsinki, Finland, although they do observe
daily peaks around 8 a.m. except on weekends and holidays, when BC values are
much lower overall. Similar (weekend vs. weekday) behavior has been observed
in Saint-Denis, the largest urban site in la Reunion island in the Southern
Indian Ocean [186]. Here the BC levels are as high as those in any large
European city, and there is no seasonal dependence. It is also interesting to note
that secondary OC in the aerosol, which probably has a photochemical origin, is
produced a lot less in the wintertime [1871. BC and OC particle size distributions
in urban areas and over large water bodies have been determined [152]. In urban
areas, weekday morning peaks are typical and results can be greatly influenced
by local vehicular sources [188].

6.2.5.1 Thermal and thermal/optical methods of elemental, organic, and

total carbon measurement
While BC measurements with an aethalometer are meaningful in gauging
visibility reduction (there is generally excellent correlation between aethalo-
meter measurements and measurements of the "Coefficient of Haze" [189]), it
does not provide an insight on the chemical nature of the carbonaceous fraction.
To this end, there has been a sustained effort over the last two decades to
differentiate and classify the different types of carbon containing compounds in
ambient particulate matter.
The difficulty arises in that the distinction between EC and OC (total carbon
TC = EC + OC) is operational rather than intrinsic. The TC values produced by
different methods generally agree while the subdivision more often than not,
does not. In a recent round robin study involving 17 different laboratories and 9
different methods, this conclusion was reconfirmed [190].
All methods that attempt to measure EC and OC separately rely on initial
filter collection followed by some form of thermal treatment (selective volatiliza-
tion, oxidation, etc.). Broadly, there are two approaches: in one, the filter is first
extracted with a solvent (toluene:DMF 3:1 or toluene:2-propanol 1:1) that is

174
presumed to extract the OC fraction only. Extracted and unextracted filters are
combusted in oxygen to form CO2 (that is typically determined by micro-
coulometry, by a flame ionization detector (FID) after conversion to methane, or
non-dispersive infrared spectroscopy [191]) and thus respectively determine OC
and TC.
In the other, thermal/optical approach, there are many subtle variations. In
general, the sample loaded quartz fiber filter is heated while the transmittance
and/or the reflectance of the filter is monitored. In addition, the evolved gases
are oxidized to CO2, typically passing over an MnO2 bed at a high temperature.
The CO2 is then put through a methanator and measured by an FID. Finally, the
filter is heated at a high temperature in the presence of oxygen to oxidize the
remaining material, taken to be EC, to CO2. The classic paper by Turpin et al.
[192] contains all of these features and is recommended reading. Details of
different methods of this genre differ in: (1) the exact temperatures to which the
sample is subjected in the different steps, (2) the length of treatment at any
temperature, (3) the rate of temperature change during a thermal step, (4) the
composition of the atmosphere surrounding the sample at any step, (5) the
nature of optical monitoring of pyrolysis/volatilization/combustion, and (6) the
nature of calibration standards. Chow et al. [154] provides a most readable
description of the approaches available until that time and also specifically their
own instrument, which subdivides OC into several different subclasses. This
instrumental protocol became the basis of the Interagency Monitoring of
Protected Visual Environments (IMPROVE). More than 100,000 PM,, and PM 2 .5
samples have been analyzed with this protocol. It has been shown that if
allocation of subclasses arising at different temperatures are changed, the
IMPROVE protocol and the National Institute of Occupational Safety and
Health protocol produces basically the same values for EC and OC [193].
Simpler methods of EC/OC speciation have been advocated. Fung [194]
described a procedure in which the sampled filter is heated in intimate contact
with MnO2 in two steps. Carbon dioxide evolving at 525°C is taken to be due to
OC, while that subsequently evolving at 850°C is taken to be due to EC.
However, judging from the trend and given that the line of differentiation is
subjective, it does not appear likely that there will soon be general acceptance of
any EC/OC speciation approach that relies on such a simple step.
Numerous intercomparison studies have been performed [179, 195-197]: a
strong correlation between aethalometric BC measurements and thermal/ opti-
cal EC measurements has almost always been found.

6.3 GENERAL SCHEMES OF COLLECTION AND WET ANALYSIS

Many practitioners of atmospheric analysis shy away from wet chemical analysis
because of the perception that gallons of reagents must be routinely made and
disposed of. The argument is routinely made that wet instruments are somehow
more complex [76] and/or less elegant. Most such feelings are archaic and do not

175
reflect the current practice of wet analysis. At best, the arguments are subjec-
tive: clearly, the beauty is in the eye of the beholder. Current state-of-the-art IC
systems require only water to operate and exhibit reliabilities similar to gas
phase instruments. Present continuous flow analysis systems can operate at low
enough liquid flow rates such that reagent replacement and waste disposal
issues have markedly changed compared to yesteryears. Doubtless, the present
author's view is biased. Nevertheless, it seems oxymoronic for Homo sapiens to
hold the view that a wet instrument is somehow less elegant. One outstanding
advantage that liquid phase analysis has over gas phase techniques is the facility
of preconcentration. This is particularly simple with ionic analytes where ion
exchange preconcentration is applicable. As a result, while the current state-
of-the-art thermal decomposition based sulfate measurement methods are taxed
to attain LODs of the order of a ~jg/m3, orders of magnitude lower LODs are
possible by wet analysis techniques with very comparable integration times
(vide infra).

6.3.1 Methods for continuous particle collection and wet analysis

There are different characteristics that can be utilized to transfer a particle to a
liquid collector. For particles or particle constituents that dissolve readily, it is
desirable that the particles are (eventually) dissolved in as small a solution
volume as possible to obtain the analyte in as concentrated a form as possible.
Second, the collection method should be applicable to small particles, at least
down to 100 nm.
6.3.1.1 Electrostaticcollection
Collection of particles by electrical means is well known [198]. When a charged
particle moves in an electric field, the particle will drift towards the electrode of
opposite charge. The velocity of a charged particle in an electric field can be
much higher than its gravitational or inertial velocity. This is exploited in the
industrial use of electrostatic precipitators for the removal of aerosols. The two
basic steps involved are (a) to charge the particles and (b) to subject them to an
electric field such that their electrostatic field induced migration will cause them
to deposit on a collection surface. One characteristic of electrostatic precipitators
is their high collection efficiencies (99-100%) over a wide range of particle sizes
(-0.05-5 gm) [199]. This principle is used industrially and is increasingly being
used in "air purifier"s intended for residential use. On analytical scale, this has
been utilized in aerosol sizing [200,201]. The present authors carried out an
experiment using a glass parallel plate wet denuder (PPWD, see Chapter 5) in
which a thin film of water flows down glass plates separated by a 3 mm thick
PTFE-coated Plexiglas spacer. The electric field was applied between the liquid
outlets with a variable voltage high voltage (HV) supply. The particles were not
charged in advance; it was assumed that field charging would be adequate. The
falling film was aspirated off (after a suitable grounding joint) and its composi-
tion measured by IC.

176
 In the very moist environment inside a PPWD, the maximum voltage that
could be applied before arcing occurred was 2.5 kV (corresponding to a field
strength of 8.3 kV/cm). Under these conditions, with -1 Avm mass median
aerodynamic diameter (MMAD) (NH4 )2 SO4 as the test aerosol, the highest
collection efficiency obtained was <50% (for a 40 cm x 36 mm active area on each
plate and a sample flow of 5 standard liters/min (SLPM). Subsequent work has
shown that smaller particles are collected with much better efficiency. Also,
having a dry region prior to the wet area where particles are charged by applying
voltage to the outer glass surface dramatically improves collection. There are
some intrinsic safety questions in such systems, however, and this approach was
not pursued further.
Liu and Dasgupta [202] subsequently described an aerosol collection inter-
face that is similar to a wire-in-tube electrostatic precipitator [198]. The
collection device consists of two concentric stainless steel tubes of 30x2.7x3.4
mm and 80x0.2x0.4 mm (length x o.d. x i.d.), functioning as the outer and the
center electrode, respectively. The tubes are fixed in position through polypropy-
lene tees (Fig. 6.4). The top of the center electrode extends through the tee and
connects to the preconcentration column of an IC system via a peristaltic pump.

ssrosal I

Fig. 6.4. Corona discharge based aerosol collection apparatus for coupling to an ion chroma-
tograph. Reproduced with permission from Ref. [202] (copyright 1996, Elsevier Science).

177
The exposed part of this metal tube is connected to the HV supply. The outer
electrode is electrically grounded. The bottom end of the center electrode pro-
trudes ca. 2 mm beyond the outer electrode end. This avoids arcing between the
ends of the electrodes. The aerosol sample is isokinetically aspirated into the
collection chamber and after it is collected for a period of time (typ. 20 min), HV
is turned off. Deionized water is pumped into the annular space and aspirated
back through the center tube. The washings are preconcentrated on the IC
system. To ensure that all of the input liquid wash solution is aspirated back, the
aspiration flow rate is set at nearly twice the input flow rate. Typically, -2 ml is
sufficient for complete washing. Before aerosol sampling is begun again, the
system is cleaned and dried by admitting compressed N2 (8 psi) into the system.
A stable corona discharge could not be generated if the center electrode was
made significantly larger to improve rigidity. On the other hand, too small a cen-
ter electrode made it difficult to be fixed rigidly in position, Generation of corona
discharge inside the collection chamber results in charging of the particles. In
industrial precipitators, negative corona discharge is almost exclusively used. In
the Liu-Dasgupta arrangement, it was easy to obtain a stable positive corona in
the current range of -6 to -70 AMA but a stable negative corona was elusive. A
voltage of -2.7 kV was minimally required to generate a corona. This threshold
voltage varied with experimental parameters such as aerosol particle composi-
tion and concentration, gas phase composition, pressure, temperature, sampling
flow rate, etc. No corona discharge was ever observed below 2.5 kV.
Before the presence of any corona discharge, the collection efficiency was
low. But as the discharge voltage is approached, the collection efficiency inc-
reased linearly with increasing voltage. Aerosol collection occurs in this regime
because even in the absence of charging by a corona, there is equilibrium
Boltzmann charge on the particles [200]. Similar to the situation with an
Electrostatic Aerosol Analyzer [203], more particles deposit on the collection
electrode as a higher voltage is applied. After corona discharge initially starts at
-2.5 kV, the current increased very quickly with the voltage up to 2.8 kV,
reaching a stable current value of -70 MA. The collection efficiency, like the
corona current, increased exponentially with the applied voltage in this range.
The collection efficiency increased essentially linearly with the corona current
up to -20 AMA at which point collection efficiency was quantitative (0.5 SLPM)
for -1.15 and 2.14 d/m MMAD Na 2SO4 particles. Collection efficiency decreased
with increasing sampling rate. Similar results were obtained for 0.5 /xm
particles.
The major problem with the above system was that, much as happens during
an electrical thunderstorm [204], NO, was generated in corona discharge. This
showed up as artifact nitrate (5-10% was present as nitrite). The relationship
between artifact nitrate and corona current was linear, -2 ng//iA of nitrate was
produced. There is little doubt that the nitrate originates from the reaction of
excited N and O atoms in the corona followed by a reaction with water vapor.
Substitution of cylinder N2 for air as the carrier gas virtually eliminated nitrate

178
formation. Nevertheless, the proposed arrangement will only be applicable
where nitrate and nitrite in the aerosol are not of interest.
In a subsequent paper [205], the same authors describe a system that does
not use corona discharge. It uses only field-charging and NOx production is
eliminated or reduced to very low values. The authors also provide a detailed
theoretical description of the field-charging method and show that the experi-
mental results agree closely with theoretical expectations. Basically, an arrange-
ment identical to Fig. 6.4, except for different collector dimensions (150x2.4x
3.0 mm (length x i.d. x o.d.) tube inserted inside a 100 x2.4x3.0 mm tube), was
used. Compared to the previously used dimensions, the annular gap is smaller
but the residence time is nearly five times longer at the same flow rate. At a
voltage of 2 kV applied to the center electrode, the field strength is -28 kV/cm
and corona discharge is not observed. At a sampling rate of 300 ml/min, the
collection efficiency is 85% and is nearly independent of the particle size in the
0.86-2.14 Am range. The collection efficiency decreases with increasing flow rate
and is -60% at 1 /min. The system was applied to ambient sulfate and nitrate
measurement using integrally coupled ion chromatography. The amount of
artifact nitrate formed was well below an equivalent ambient concentration of
0.1 gug/m 3 .

6.3.1.2 Impaction
Impaction on a solid surface is the process most commonly used for particle
collection. The same approach can be utilized for continuous transfer of particles
into a liquid surface. If the liquid constitutes a moving film, then the liquid
containing dissolved and undissolved particles can be directly sent to an analysis
system. Quantitative collection of small particles may necessitate serial,
multiple impaction, which can be achieved by multiple and frequent changes in
flow direction. One way of maintaining a thin liquid film under such a condition
is to introduce the liquid at the inlet along with the aerosol to be sampled. Also, if
the liquid enters the impactor through a narrow orifice, it will itself be aerosol-
ized. For an instrument of such a design, in addition to the impaction of particles
on to the wetted walls, there will be particle growth through particle coagulation
and thus increase the particle collection efficiency of the impaction system. On
the other hand, if the fluid velocity at the impactor exit is reduced by making the
cross-sectional area large, the liquid containing the aerosol particles should
come out as larger droplets, which can be separated from gas phase by an inertial
separator and sent to an analysis system.
This principle was tested with a maze type impactor as represented in Fig.
6.5. As shown, a perpendicular maze was cut into the bottom section; the top
plate is a flat piece of same length and width as the bottom plate. When
assembled, it forms a sealed enclosure. Water (0.5-1 ml/min) was introduced at
the inlet along with the sample aerosol (5 1/min). At the exit, liquid droplets were
collected with an inertial air/ liquid separator represented in Fig. 6.6. Collection
efficiency for 2.2 m MMAD Na2 SO4 aerosol was essentially quantitative but this

179
 ^.
..... o nn - Outlet
- m 38mm
mm

Inlet 3 mm
m Air/Liquid
Inlet

Coverplate 3tn al ~ *.< IMaleNut

Botom plUe Ruber O-Ring

3 flue N 13:' |i .

3 mnmi mm
69 rrn I
The cioerel is 5 amdeep LiquidOutlet

Left: Fig. 6.5. Maze-style impactor, schematically shown.

Right: Fig. 6.6. Inertial gas liquid separator. Reproduced with permission from Ref. [223]
(copyright 1995, American Chemical Society).

dropped to < 70% for 0.7 Avm MMAD aerosol. Further experiments clearly estab-
lished that such an arrangement could not collect submicron particles with
acceptable efficiency. Similar results were obtained with an impactor consisting
of a flattened glass tube (elliptical cross section) with 22 U-turns. Collection of
large particles is efficient but submicron particles are difficult to capture. Of
course, this is known to be the case for impaction in general. The conclusion
drawn from these experiments was that if small particles could somehow be
increased in size prior to impaction, without altering their chemical composi-
tion, then this would be an attractive approach.
Nevertheless, direct impaction into a liquid has its attractions. Karlsson et
al. [206] reported a novel single stage flowing liquid film impactor for continuous
on-line particle composition measurement. A continuous liquid flow is provided
to a support plate made of etched glass that constitutes the impaction area.
Continuous transport of the impaction liquid makes on-line analysis by an
integrally coupled IC possible. With multiple nozzles (74 holes, 0.3 mm i.d.) and
an air sampling rate flow of 10 l/min, the 50% cut-off was 0.41 ± 0.02 Alm. The
design is shown in Fig. 6.7. The actual performance in terms of particle size cut
off is shown in Fig. 6.8. While this is an interesting design, for the collection and
analysis of ambient aerosols, a significant mass of particles exist below the cutoff
size of this impactor, and it is not practical for that purpose.

6.3.1.3 Filtration
Filtration is one of the most common methods of aerosol sampling. Although the
general principles are well known, there are still significant gaps between theory
and experiment. A common misconception is that aerosol filters work like
macroscopic sieves through which only particles smaller than the holes can pass.
This is not how aerosol filtration works [207]. All of the following mechanisms
lead to deposition of aerosol particles on or within a filter: interception, inertial

180
 G

C X
D

Fig. 6.7. Assembly drawing of the wet impactor. The supporting base A is equipped with three
screws B for vertical and horizontal adjustment of the impaction liquid support plate C. Also
shown are the impaction liquid drain spout D, inlet fitting for impaction liquid E, nozzle plate F
and the impaction liquid film G. Liquid I/O tubing H were connected through bulkhead fittings.
Reproduced with permission from Ref. [206] (copyright 1997, Elsevier Science).

e
90-
a'
a'
.2 60 - A
40 - __C]-_
C
--a --
#2
U
e 20
a'
10
0 I I
0 0.2 0.4 0.6 0.8 1
Dp/pm

Fig. 6.8. Impactor performance (Karlsson et al.). Reproduced with permission from Ref. [206]
(copyright 1997, Elsevier Science).

impaction, diffusion, gravitational settling, and electrostatic attraction. Multi-

layer fiber filters such as those of glass and quartz fiber are very efficient in
retaining particles even down to 0.01 gm size.
The problem with using conventional filters for automated aerosol analysis
centers on the difficulty of devising a method that allows adequate automation
for wet analysis while providing enough retention of particles and air sampling
rate. A paper by Buhr et al. [208] probably represents the best compromise. The
authors combined a fritted glass filter (6.5 cm dia., 0.6 cm thick, pore size
170-220 ,m) at the bottom of a single tube wet denuder. The arrangement is
shown in Fig. 6.9. The denuder effluent (see Section 5.5.1 in Chapter 5) is first
taken out through a gas-liquid separator and more water is then introduced.
This water continuously washes the frit on which the aerosol is collected and the
washings are collected through the bottom gas/liquid separator and sent to the
IC system shown in Fig. 6.10. The system was used for the measurement of
gaseous nitric acid and aerosol phase nitrate and sulfate. Because there are
limitations on how efficiently simple single-tube denuders can remove a gas, a
sample flow rate of 1 1/min was used. Nevertheless, on 15 min samples, the LODs
were 10 pptv or equivalent for each species. Using 0.3-1 ,um NH4 NO3 test

181
 Entrance length

Water in
Denuder I
Wet Effluent
Denuder

3as/Liquid
ri
L
i
r
Separator ? Water in
Frit

Pyrex Frit E

Fig. 6.9. Single tube wet denuder-wetted glass frit filter for
simultaneous gas and aerosol sampling. Reproduced with permission
Gas/Liquid
from Ref. [210] (copyright 1995, Elsevier Science). Separator

Fig. 6.10. Analytical system schematic for gas and aerosol analysis according to Buhr et al. [208].
Note provisions for liquid phase calibration and measurement of blank. Reproduced with
permission from Ref. [208] (copyright 1995, Elsevier Science).

particles, an average retention efficiency of 79 ± 8% was found for the glass frit.
Field measurements and comparisons with 1-h filter pack (FP) samples were
consistent with this: the data showed close agreement for sulfate in terms of
temporal variations, with the frit system measuring 12% lower sulfate on the
average. The data for aerosol nitrate as measured by the frit system was not so
straightforward. The measured nitrate value was much higher than that

182
measured by the FP and showed a diurnal variation not exhibited by the FP. The
artifact nitrate generated appears not to be related to NO2 or NOx but to some
photochemically generated intermediate that produces nitrate at the wet frit.
This general issue of intermediates that react with water to produce some key
nitrogen species appears in many places and has not been resolved. There is
accumulating evidence that direct liquid contact based instrumentation (e.g.,
wet denuders) tend to produce significantly higher values for HONO than
collocated direct spectroscopic instruments [209]. It is vital to resolve these
differences and get a better understanding of the responsible species before an
accurate picture of nitrogen deposition in ecosystems can be obtained.
An aspect of wet effluent systems that are coupled to solid phase pre-
concentrators is the necessity to aspirate all of the effluent liquid so that liquid
does not enter the sample inlet region. This is difficult to do without aspirating
bubbles in the process. A very interesting finding of Buhr et al. was that solid
phase preconcentration columns exhibit greater analyte capture efficiency for a
segmented flow stream (i.e., one containing air bubbles), rather than an
unsegmented one.
Samanta et al. [210] described a glass-fiber filter-based instrument for the
automated measurement of aerosol Cr(VI). Hexavalent Cr is a notorious inhala-
tion carcinogen that is produced in the aerosol form in several workplace situa-
tions. They solved the automation problem by using a system that alternately
collects the sample on one of two glass-fiber filters. After 15 min sample
collection on one filter, the sampling switches to the second filter. The freshly
sampled filter is washed for 8.5 min and the washings are preconcentrated on a
minicolumn packed with anion exchange resin. The washed filter is dried with
filtered hot air for the next 6.5 min so that it is ready for sampling at the end of
the 15 min cycle. The preconcentrated Cr(VI) on the column is eluted with 0.1 M
sodium perchlorate and then reacted with sym-1, 5 diphenylcarbazide prior to
absorbance detection with a light emitting diode (LED)-based dedicated flow-
through absorbance detector. The detection limit (S/N = 3) is 5 ng Cr(VI)/m 3,
orders of magnitude lower than current regulatory requirements. The instru-
ment operates unattended over long periods. In continuous round-the-clock
operation, filter replacement frequency is every 24-72 hours depending on dust
loading. The system is portable for facile field deployment and permits fully
automated rapid determinations at a very low analysis cost per sample.

Filter-basedautomatedparticle collection system (PCS)-IC system for general

analysis of ambient aerosols
Boring et al. [211] have developed the filter-based approach further for the
general measurement of ambient aerosols via coupling to an IC. A cyclone
removes large particles and the aerosol stream is then processed by a PPWD (see
Chapter 5) to remove potentially interfering gases. The particles are then
collected by one of two glass fiber filters, which are alternately sampled, washed
and dried. The washings are preconcentrated and analyzed by IC. Detection

183
 Drying Air Drying Air
Sampling
Air

(b)

FA

FA FB
ON OFF

Fig. 6.11. (a) Total system airflow schematic. FA, FB: Glass fiber filters, T: trap bottles,
MFC-A,B,C,D: Mass flow controllers, C; cyclone, FC: 47 mm filter for MS analysis, P1,2: air
sampling pumps, PP: peristaltic pump, F: filter, P: purifer, H: heater. The dotted section
including the denuder is on the roof and the air pumps are either below the instrument shelter or
in a modified doghouse with forced air ventilation. Vi: Aerosol switching valve, shown in detail
in (b).

limits of low to sub ng/m 3 concentrations of most gaseous and particulate

constituents can be readily attained. The instrument has been extensively
field-tested.
A Teflon® coated aluminum cyclone is used as the first element of the inlet
system to remove particles larger than 2.5 am. The cyclone exhibits the desired
size cut point at the design flow rate of 10 1/min. Referring to the overall airflow
arrangement in Fig. 6.11a, the air sample passes through the cyclone at 10
SLPM and is divided by a Y-connector into two flow streams of 5 SLPM each.
One is drawn through a glass fiber filter F1 for archival purposes and for IC-
suppressed conductivity-UV-MS (IC-CD-UV-MS) analysis of the filter extract in
home laboratory via mass flow controller MFC-C. The cyclone and the filter
holder are mounted on a modified camera tripod. The feet of the tripod are
bolted to the roof of the instrument shelter; the air inlet is maintained -2 m
above the roofline. The second flow stream from the cyclone exit proceeds
through a copper conduit or aluminized PFA Teflon tube to a PPWD located
within the instrument shelter. The metal is electrically grounded to minimize
aerosol loss. The PPWD is fed with -1 ml/min streams of 10 mM Na 2 HPO 4

184
(adjusted to pH 7) containing 0.5 mM H2 02 , on each plate. This solution serves to
remove both acidic and basic gases; the denuder effluent (aspirated at -1.5
ml/min per plate) is sent to waste. The gaseous effluent from the denuder
bearing the aerosol proceeds to the PCS.
The first element of the PCS is a specially constructed rotary valve V1 that
directs the ambient air stream to either filter A or Filter B. This valve must
provide a straight passageway for the sample stream to one of the two sample
filters without aerosol loss. The valve is shown in functional detail in Fig. 6.11lb.
The stator plate has three holes, the central port is connected to the sample air
stream (from the PPWD) while the two other ports are connected in common
through a Y-connector to a sequential trap containing a particle filter (F), acid-
washed silica gel (T1, this removes NH3 ) followed by a soda-lime trap (T2, this
removes acid gases) and a heater (H) that thus provides a hot dry clean air
source. The rotor plate has two holes, connected to Filter A (FA) and Filter B
(FB), respectively, and is rotated by a spring-return rotary solenoid. With the
solenoid un-energized, ambient air is sampled on Filter A and with the solenoid
energized, ambient air is sampled on Filter B; flow is thus switched without
aerosol loss. Other air valves (V2-V4) govern airflow in the PCS.
Purpose-machined holders hold 25 mm diameter filters. A glass fiber filter
with a paper filter underneath rest on a stainless screen; ports on opposite sides
of the top half of the filter holder provide entry of wash liquids. The bottom half
of the filter holder is designed as a shallow cone with the air outlet at the center.
The liquid exit port is located equidistant from the inlet apertures such that the
inlet/outlet apertures constitute an equilateral triangle in top view. Air/liquid
separators are incorporated below each filter holder in the air exit path. These
contain air in/out ports, as well as a port to remove accumulated water (period-
ically, e.g., every 24 h) using a syringe. These separators serve to keep any wash
liquid from entering the respective mass flow controllers (MFC-A, B). The
diaphragm pump (P2, same as P1) used for sampling is capable of aspirating at
>8 1/min through each filter holder simultaneously. Perfluoroalkoxy (PFA)
Teflon tubes, used for connecting PCS components upstream of the filter
holders, are externally wrapped with electrically grounded Al tape and then with
bare Cu wire. This served the dual purpose of improving its structural strength
and reducing electrostatically induced aerosol loss.
A six-channel peristaltic pump provides liquid pumping. In Fig. 6.12, valves
V5-V8 are liquid-handling solenoid valves. Valves V5 and V6 direct the flow of
deionized water to the filter holders. Valves V7 and V8 (pinch type valves)
handle filter extract in which stray glass fibers may be present. A low volume
fiber-trap-filter (FTF) placed prior to the injection valve prevents glass fiber
intrusion to the preconcentration columns. Injection valve IV contains two
low-pressure drop anion preconcentration columns PCi and PC2 that alternate
in function.
Table 6.1 shows the air and liquid valves and their respective on/off status.
Figures 6.12a and b illustrate the four states of the instrument cycle. The first

185
 (a) (b) 1FA , Liquid
Phase

PPWD
Na2HPo X X
H202
Ambien
Air In Air In V2 Gas
TI T i lPPWD Phase
F
FA
Ff FA --p Ambient
Air In
MFC-I MFC-A
V3) (VS
vP
P1 1-

Fig. 6.12. PCS set up. (a) State 1, (b) State 2. V5,6: All-PTFE solenoid valves, V7,8: Pinch-type
solenoid valves, FTF: fiber trap filter, IV: 10-port injection valve bearing preconcentration
columns PC1 and PC2, IC2: PCS Ion Chromatograph, V2-4: on/off solenoid valve for air flow, T1:
acid washed silica gel, T2: soda-lime (T1 and T2 together constitute the purifier P in Fig. 6.2),
PPWD: PCS gas denuder, WB: water bottle, W: waste. The top and bottom portions respectively
indicate liquid and gas flow.

state (Fig. 6.12a) is 8.5 min in duration. In the particle collection system, the
soluble gas denuded aerosol flow stream is directed to Filter A by valve V1. Air
passes through Filter A though MFC-A, which regulates the airflow to 5 SLPM,
and finally through valve V4, which is on during state 1. Valves V2 and V3 are
off, and filter holder B (FB) is under airlock.

TABLE 6.1
Four states of the instrument
Status IC using V1 V2 V3 V4 V5 V6 V7 V8
Begin sampling on FA, start washing PC2 OFF OFF OFF ON OFF OFF OFF OFF
FB, water being aspirated through FB
and being loaded on to PC1

Sampling still on FA, washing FB PC2 OFF ON ON ON ON OFF OFF ON

complete dry FB, water purge PC1

FB dry, switch sampling to FB PC1 ON OFF ON OFF OFF ON ON OFF

put PC1 in injection mode begin
washing FA

Sampling still on FB. Begin drying FA PC1 ON ON ON ON ON OFF OFF ON

water purge on PC2

186
 In the liquid extraction portion of the instrument, deionized water is
contained in a bottle WB. The air entrance to the water bottle is equipped with a
soda-lime trap to minimize acid gas intrusion into the bottle. Water from WB is
aspirated and then pumped at 1 ml/min by the peristaltic pump PP through a
mixed bed ion exchange column MB1 to remove any trace impurities present in
the deionized water. Valve V5 directs flow to valve V6, which in turn directs the
water to filter FB. The water enters FB through the two ports in the top of the
holder and is simultaneously aspirated from the bottom of FB through valves V7
and V8 by the peristaltic pump. Since FB is under airlock, water does not enter
the air outlet tubing at the bottom of the filter holder. The extracted material
from the filter is pumped through the fiber trap filter FTF to remove glass fibers
from the flow stream before passing to the appropriate preconcentration
column. Valve IV is configured such that while one preconcentration column is
chromatographed, the other preconcentration column is loaded with sample or
washed with water. In the present case, preconcentration column PC1 is loaded
with sample. After 8.5 min, state 2 begins (Fig. 6.12b).
During state 2 in the PCS, ambient air continues to be sampled on FA, just as
in state 1. Valves V2 and V3 are activated in state 2, allowing clean hot air to pass
through Filter FB for the duration of this state. Clean (ammonia/acid gas- and
particle-free) air, produced by passing ambient air through F, T1 and T2, is
heated to -75°C by passing it over a heater switched in parallel with valve V2.
This clean, hot air is aspirated through the previously extracted filter FB, to dry
it prior to state 3.
Note that at the time the instrument enters state 2 from state 1, although all
the analyte has been extracted from filter FB and preconcentrated, the last
portion of the wash water is still contained in the filter housing. This water is
aspirated into the trap bottle ahead of MFC-B. Water that enters into the trap
bottle is generally of the order of 1 ml/cycle. This volume may be used to monitor
the filter extraction process; excessive water accumulation in the water trap
bottle indicates flow problems through the filter or through the relevant pre-
concentration column.
In the liquid extraction system, valves V5 and V8 are activated. Valve V5 now
directs water used to wash Filter FB in state 1, back into the water bottle. This
recycling procedure helps maintain the purity of the water in WB. As a result of
liquid being aspirated faster from the filter housing than it is pumped in, air
bubbles inevitably enter into the preconcentration column. To remove the air
bubbles before the sample is injected, valve V8 is activated and water is aspirated
by the pump through a mixed bed ion exchange column MB2 through V8 and
pumped through the preconcentration column PC1. The duration of state 2 is
6.5 min.
After state 2 ends, state 3 (8.5 min) and state 4 (6.5 min) follows. States 3 and
4 are identical to states 1 and 2, respectively, except that the roles of filters A and
B are interchanged relative to those in states 1 and 2. States 1-4 constitute an
instrument cycle and the instrument starts anew with state 1 at the end of state

187
 Ir - _;Z . -;

I;> *If An, Nitrate

10-

..t' , ij

- :I: r
Nit rous acid

A Nitrit

10
45-
Sulfur Dioxide
A Sulfate

a 30

15-

0-
8116f99 8118199 8120199

Fig. 6.13. HNO 3 /Nitrate, HONO/Nitrite and SO2 /Sulfate patterns at a Midtown location in
Atlanta, GA. Note nocturnal maxima in the middle panel and opposite behavior in others.

4. The entire process is endlessly repeated until shut down by software

command.
Figure 6.13 shows HNO 3/Nitrate, HONO/Nitrite and SO2/Sulfate patterns at
a Midtown location in Atlanta, Georgia. Note nocturnal maxima in the middle
panel and opposite behavior in others. Figure 6.14 shows HCl/chloride, oxalic
acid/oxalate levels at a heavily industrialized site close to the shipping channel in
Houston, Texas. This site is known to be close to one of the largest chlorine
emission sources in the region. A typical chromatogram from the particle
analysis system is shown in Fig. 6.15.
When carefully examined for minor components, the chromatograms
(especially those for the aerosol samples) reveal a far greater degree of complex-
ity. A gradient chromatogram of a 30 min sample collected in Atlanta is shown in
Fig. 6.16, with overlays representing 10x and 100x magnifications of the base
chromatogram. Considering that the baseline is essentially completely flat for a
blank run even at the 100x magnification, the number of real components
present in such a sample becomes readily apparent. Not surprisingly, a majority
of these peaks are organic acids. While MS is ultimately the only completely
unambiguous means of identification when confirmed by a matching standard,
in many cases the charge on the analyte ion can be estimated by determining

188
 2.5-
Oxalic acid
A Oxalate

2.0-

1.0 - Q.

ii
0.5 -

'"-"'
.
VIANSWUNL IDA
0.0T- I 7- - IWIIE7 -
8119/00 8121100 8123100 8125100 8/31/00 9/2/00 9/4100 916/00

Fig. 6.14. HC1/Chloride, Oxalic acid/Oxalate levels at a heavily industrialized site close to the
shipping channel in Houston, TX. This site is known to be close to one of the largest chlorine
emission sources in the region.

8.00

6.00

.P
4.00

g 2.00

0.00

0.00 4.00 8.00 12.00

Time, min
Fig. 6.15. Representative chromatograms from the PCS-IC, Houston HRM-3 site.

void volume corrected retention times (tR) under isocratic elution conditions at
three or more different eluent concentrations. Under these conditions, it is well
known that the slope of a log t vs. log [eluent] plot is equal to the ratio of the
charge on the analyte ion to that on the eluent ion (unity for hydroxide) [212];

189
 I
,
9

0.0 5.0 10.0 15.0 20.0 25.0 30.0

Elution Time,min

Fig. 6.16. Gradient ion chromatogram of an aerosol collected during the Atlanta experiment.
Readily identifiable peaks: (1) fluoride, formate, acetate, (2) chloride, (3) nitrite, (4) carbonate,
(5) nitrate, (6) sulfate, (7) oxalate.

Unk4 _ t nk7 slope -1.

slope -2.
1.00-
/
Unk3 slope -2.6 - Unk6 slope -1.1
Nitrate Slope-1.17
Sulfate Slope -1.80
arbonate slope -1.62
Nitrite slope -1.10
Chloride slope .90
0.00- Unk2
. slope .1.04 U-kIslI: f 110
Uel
sUpe UnO slope-.0
(3 -

- 0.50- Unk1"slope -1t

-1.00- I ' I I
1.10 1.20 1.30 1.40 1.50
log Hydroxide Eluent Concentration, mM]

Fig. 6.17. Log t vs log [eluent] plots reveal charge on analytes, aiding search for a confirmatory
standard.

this is shown in Fig. 6.17. While MS is ultimately the only completely unambigu-
ous means of identification when confirmed by a matching standard, IC informa-
tion above and the nature of UV response of the analyte often make possible to
determine the identity of the analyte.
Table 6.2 lists average daytime and nighttime aerosol composition for a rela-
tively polluted period during the Atlanta measurement campaign. The analysis
was conducted by IC-CD-UV-MS, with identification confirmed by MS and con-

190
TABLE 6.2
Average anion composition of day and night-time aerosol in Midtown Atlanta, August, 1999a
Retention time (min) Concentration (g/m 3 )
Conductivity detector UV detector Analyte Samples (day) Samples (night)
8.34 Fluoride 1.1 0.58
8.95 Glycolate 0.28 0.19
9.37 Acetate 0.58 0.25
9.56 Lactate 0.81 0.32
9.83 Formate 0.91 0.71
10.96 a-Hydroxyisobutyrate 0.02 0.03
11.23 Unknown [0.015] [0.02]
11.87 Methanesulfonate 0.05 0.04
13.04 Chloride 9.8 5.5
13.27 Pyruvate tr tr
14.93 Unknown [0.004] [0.01]
15.52 Nitrite 0.11 0.15
15.60 Carbonate nd nd
16.23 Malate 0.30 0.24
16.57 Malonate 0.36 0.26
17.23 Sulfate 16 11
18.13 18.34 Oxalate 0.34 0.27
20.46 Unknown [0.01] [0.02]
21.58 Phosphate 0.03 0.03
23.28 23.52 Nitrate 1.9 1.7
24.33 24.66 Unknown [0.02] [0.03]
24.87 Unknown [0.03] [0.03]
25.87 26.06 Unknown [0.004] nd
26.72 Unknown [0.003] [0.007]
28.50 28.83 o-Phthalate tr tr
29.10 Unknown [0.004] [0.072]
aRetention times using a hydroxide gradient chromatographic protocol. Numbers in parentheses
provided for unknown peaks are calculated from the conductivity peak areas based on the
average response. These likely the lower limits. tr: trace.

ductivity providing quantitation. Several peaks remain unidentified; numbers in

parentheses provided for these are calculated from the conductivity peak areas
based on the average response. These should be taken as lower limits because
the average response per unit weight is dominated by strong acid anions and
these unidentified species are almost certainly organic acids for which response
per unit weight is likely to be smaller. It is interesting that the list of the organic
acids identified here overlaps in a major fashion with the list of aliphatic organic
acids that are used as metabolic pathway markers in the human physiological

191
 Air suction

Hydrophobic

i Bonnet

Water out
Fig. 6.18. Particle collector, incorporating integral cyclone, _I Water in
liquid spray and hydrophobic refluxing filter.

system [213]. Perhaps many of the low molecular organic analytes we find in
ambient aerosol are biogenic.

Refluxing with a hydrophobicfilter

Early work conducted by Cofer et al. [214,215] and subsequently by Janak and
Vecera [216,217] had involved generation of liquid aerosols to collect gases.
Sample particles also interact with the generated liquid aerosol and grow in size.
These systems did not use a denuder, as a result, both gases and aerosols were
collected. In the Cofer et al. design, the system exit was through a hydrophobic
polytetrafluoroethylene (PTFE) filter. This prevented escape of the liquid aero-
sol and the system basically acted like a reflux condenser. Recently, Al-Horr et
al. [218] have used this principle that provides for a simple particle collector,
readily interfaced to a solution phase analysis system; the device is shown in Fig.
6.18. The inlet air is processed through a denuder and then enters the initial
cyclone-like inlet of the device, this can be designed for any particular size cut
point. Liquid is aspirated by Venturi action or deliberately pumped through a
capillary and the air is drawn through a small aperture surrounding it. The
liquid generates a spray that attaches to the aerosol particles. The flow is
ultimately drawn out through a 0.5 Am pore size PTFE filter. Liquid droplets
coalesce there and fall below, Electrodes are built into a microliter volume
collection well and when it is filled with solution, the liquid is aspirated by a
peristaltic pump and sent to a preconcentration/analysis system. The system can
collect particles down to 100 nm with high efficiency. It was extensively field
tested in a month-long field study at Philadelphia, PA in the summer of 2001 and
in Tampa, FL in the summer of 2002.

6.3.1.4 Particlecollection by vapor condensation

It has already been stated that several approaches for particle collection will be
applicable for larger, but not smaller, particles. If smaller particles can be grown

192
into larger particles without a change in their composition, then these methods
become applicable. Fortunately, fine particles are excellent nucleation sites for
condensation. As early as 1888, Aitken [219] used vapor condensation to make
very small particles grow so that they are large enough to be optically counted.
Even today, available condensation nuclei counters rely on this principle [220].
Typically, the aerosol is equilibrated with a liquid such as 2-propanol or water.
The vapor saturated aerosol is cooled, often by adiabatic expansion, to induce
supersaturation.
Water is more compatible with most analysis systems than 2-propanol. One
can deliberately introduce steam to the sample aerosol and then cool the
mixture. Cooling causes a large supersaturation, causing the steam to condense
on the available particles to form liquid droplets. These droplets have much
larger size than the original aerosol present and can be collected by various
means, including impaction. A flow-through chamber to provide sufficient
residence time allows the growth to occur. If the impactor itself is cooled,
collection on the cold impactor surface is further aided by thermophoresis
(aerosol particles are attracted to a cold surface) [221].
It is instructive to compute the degree to which such growth occurs. If we
assume an idealized case of a monodisperse aerosol of density 2.0 g/cm3 , a
sampling rate of 10 /min, an aerosol concentration of 100 tg/m 3 (this is much
higher than typical ambient fine particle concentration), a steam injection rate
of 500 mg/min, and homogeneous growth, it is readily computed that a particle
will grow in the radial dimension by 100-fold (final particle density is taken to be
the same as that of liquid water), i.e., a particle with an original diameter of 10
nm will grow to 1 Am. Simple considerations will also indicate that the factor by
which each particle grows increases with higher steam injection rates, lower
sampling rates, and lower aerosol concentrations. The process of particle growth
is vital to impaction-based collection processes and the importance of growth in
size due to vapor condensation cannot be over-emphasized. The mixing chamber
thus serves the function of a cloud chamber.
Blatter et al. [222] and Simon and Dasgupta [223,224] first described a particle
collection and analysis system based on this principle; their arrangement is shown
in Fig. 6.19. The soluble gas-denuded aerosol stream (5 1/min) proceeds to the mix-
ing chamber where steam is introduced at a rate of 800-900 mg/min. Then it
enters a thermoelectrically cooled maze (see Fig. 6.5), which largely functions as a
rain chamber. Agglomeration and impaction occurs, and the process of vapor con-
densation is completed by cooling. Note that the moisture content of the sample
stream exiting the wet denuder is high and cooling to 2°C can add almost another
- 100 mg/min of liquid water to the system effluent. An inertial gas-liquid separa-
tor is added at the exit. The liquid is collected from here and pumped to the analy-
sis system. The overall collection efficiency for the system is shown in Table 6.3,
for all practical purposes, the collection is quantitative.
A number of other researchers have also used steam condensation prior to
aerosol growth. The apparatus of Khlystov et al. [225] developed at the Nether-

193
Fig. 6.19. Vapor condensation particle collection system. Reproduced with permission from Ref.
[223] (copyright 1995, American Chemical Society).

lands Energy Research center (ECN) is shown in Fig. 6.20; the device has been
christened the Steam Jet Aerosol Collector (SJAC) by its developers. In this
arrangement, the soluble gas denuded sample air mixes with steam in the
mixing reservoir. Then two sequential glass cyclones are used to collect the
condensed droplets bearing the aerosol. The liquid effluent from the cyclones are
combined and sent to the analysis system. The gaseous effluent is sent through a
cooler to remove residual moisture prior to passage through flow controllers.

TABLE 6.3
Efficiency of aerosol collection in the Simon-Dasgupta [225] vapor condensation-PCS
Composition of aerosol MMAD (m) Concentration Percent collected by
O(g/m 3) system
Sulfuric acid 0.5475 + 0.0034 1.63 96.97a
Sulfuric acid 0.7139 + 0.0068 9.85 - 2.06 99.95±0.06
Ammonium sulfate 0.7175 + 0.0054 6.56 + 0.17 99.23+1.08
Ammonium sulfate 1.0057 + 0.0065 27.96 ± 0.53 99.21±0.14
Ammonium nitrate 0.6134 - .0065 6.13 ± 0.36 99.88-+0.04
Ammonium nitrate 0.840 ± 0.01 13.27 + 0.51 99.31±0.66
Sodium nitrate 0.6038 ± 0.0097 72.11 + 3.5 100.27 ± 0.52
Sodium nitrate 0.86 b 307.4 + 2.2 99.46 + 0.09
Sodium sulfate 0.4012 ± 0.0021 64.46 + 1.4 99.61 + 0.10
Sodium sulfate 0.67 b 302.4 + 3.6 99.44 ± 0.31
aUncertainty was not determined. At these low levels, the finite filter blank for the post-system
filter is the source of a significant negative bias for the computation of collection efficiency.
Reproduced with permission from Ref. [2251 (copyright 1995, American Chemical Society).

194
 steam
In condenser
airin air out

mixin
reservoir /
Fig. 6.20. ECN Steam Jet Aerosol Collector cyclnes condensed water out
(SJAC). Reproduced with permission from Ref. sampled aerosol solution
[225] (copyright 1995, Elsevier Science). out (for analysis)

The ECN apparatus can sample a much larger volume of air, typically 20-30
1/min, compared to the 5-10 /min flow rates used in the apparatus described in
the foregoing section. When the atmosphere is supersaturated with water vapor,
the growth of the particles is very rapid. Figure 6.21 shows the rapid growth of
100 nm MMAD particles in a supersaturated atmosphere. Figure 6.22 shows the
effect of the extent of steam injection on the particle size distribution.
More recently, the ECN group has elaborated [226] on the refinements made
to their original instrument over the 6 years since its original introduction.
Current practice includes the use of a 10 /M sodium carbonate solution in the
denuder, which is said to effectively scrub out all gases of interest, including SO 2,
HONO, HNO3 , and NH3 . Air sampling is conducted with a pump equipped with a
critical orifice. After the introduction of steam (2.5 g/min), all particles over 10
nm initially in size grow to at least 2 CLm. More than 99.9% of these particles are
collected in the system. Indeed, even at a flow rate of 60 1/min, the collection
efficiency is >99%. The system contains liquid mass flow controllers on all
critical liquid flow lines for quality assurance. The combined sample from the
cyclones is split into two streams. To one stream, NaOH is added and the mixed
stream passes across a porous PTFE membrane based gas diffusion device [227].
Under the test conditions, -30% of the ammonia transfers to the receptor side,
which is pure water. This is measured by conductometry. This principle for
ammonia measurement was originally described by Carlson [228] and adapted
by the ECN group [229,230]; it is commercially available as well [231]. If 50-60
/g/l ammonium is added to the NaOH, the calibration curve is linear. The

Dry

-J
zi I---- 20%

Fig 6.21. Growth of originally

-100 nm particles in a 10% and
20% supersaturated atmosphere.
Reproduced with permission from . .'
Ref. [225] (copyright 1995, Else- o.ol 0.1 1 10 100
vier Science). Diameter (pm)

195
 100000

100001

a 100

10

Fig. 6.22. Growth and removal of 1

(NH 4 )2SO4 test aerosol at different
steam injection rates. Reproduced n1
with permission from Ref. [225] 0.01 0.1 1
(copyright 1995, Elsevier Science). Diameter (pm)

response is dependent on temperature but a thermistor integral to the measure-

ment cell corrects for it.
Bromide is added to the rest of the sample as an internal standard before it is
preconcentrated on a anion preconcentrator column. An eluent of 3 mM Na2CO3,
1.3 mM NaHCO3 is used for suppressed ion chromatography. A correction is
made for nonlinearity inherent in suppressed carbonate based ion chromatogra-
phy. Extensive intercomparison results are presented. The LODs for chloride,
3
nitrate, sulfate and ammonium are typically 20-50 ng/m for a 15 min sample.
Zellweger et al. [232] described a gas and aerosol collection analysis system
that is largely similar to the system described by Blatter et al. [222], and Simon
and Dasgupta [223]. Loflund et al. [233] described what is likely the most
sophisticated vapor condensation based aerosol collection-ion chromatographic
analysis system to date. The device has been named the Gas and Aerosol
Monitoring System (GAMS). Their apparatus uses a quartz circular end parallel
plate wetted denuder (CEPPWD, see Section 5.5.4 in Chapter 5) at a flow rate of
5 1/min. Provisions for steam injection, mixing, impaction in a cooled maze and
an inertial air-liquid separator are similar to the previous Simon- Dasgupta
design. The liquid sample is sent to two 10-port valves, each containing cation
and anion preconcentrators, which alternately acquire cation and anion samples
for 15 min and analyze them on individual dedicated ion chromatographs. The
chromatographs respectively use electrodialytically generated CH 3 SO3 H and
KOH eluents; suppression is also carried out electrodialytically. A differential
mobility particle spectrometer was extensively used too determine system
behavior. These investigators studied for the first time the ability of the vapor
condensation approach to remove particles over a broad input size range of
20-1000 nm. Figure 6.23 shows the efficiency of the GAMS system to collect
(NH4 )2SO4 particles across this entire range. They also looked at bis(2-ethyl-
hexyl)sebacate as an illustrative non-hygroscopic aerosol. These particles do not
change their size distribution in the wet denuder. In this case, as shown in Fig.
6.24, all particles above -100 nm grew in size and were removed by the GAMS.
The GAMS system LODs are shown in Table 6.4. The LODs for cationic species

196
 80000-

60000-
x 0
* I 10
40000-
*·

20000-

·
, ,*,0,t
, t I , , ,l
0-
I . . . I II ' ' ' I' 7 -I
10.00 100.00 1000.00
d,(nm)

Fig. 6.23. Efficiency of the GAMS [233] to remove (NH4 )2 SO4 aerosols. Data courtesy of Maria
Loflund, Vienna University of Technology, 2001. Solid and open circles respectively represent
quantitative size distribution of the aerosol before and after the GAMS.
an
- ---

10000-

1000-,

100-
A A
A

10-
A oA A A
A Primary size distribution
O Size distribution after denuder
A4 Size distribution after GAMS
AAAAAAAAAAAAA

0-

10.00 100.00 1000.00

d,(nm)

Fig. 6.24. Size distribution of a non hygroscopic test aerosol, bis(2-ethyl-hexyl)sebacate before
and after the wet denuder and after the GAMS [233]. Data courtesy of Maria Loflund, Vienna
University of Technology, 2001.

may actually be much better in a comparably constructed instrument since the

authors had problems with a noisy eluent generator.
In a collaborative effort between Georgia Tech (GT) and the Brookhaven
National Laboratory (BNL), Weber et al. [234] described their instrument for

197
TABLE 6.4
Analytical figures of merit for GAMSa
Species. LOD (ng) Limit of determination Repeatability
Aerosol (g/m 3) CV (%)
Formate 3.6 0.02 7
Acetate 2.6 0.09 6
Propionate 1.4 0.04 7
Butyrate 10 0.07 9
Valerate 13 0.09 9
Pyruvate 5.6 0.04 5
Flouride 5.0 0.03 9
Chloride 0.45 0.03 8
Nitrite 3.2 0.05 9
Nitrate 1.1 0.07 5
Sulphate 5.3 0.08 4
Phosphate 1.4 0.01 11
Oxalate 0.75 0.02 1
Malonateb 3.8 0.03 12
Succinate 7.8 0.05 10
Phthalate 10 0.07 2
Citrate 0.75 0.01 13
Maleate 2.0 0.20 3
Sodium 29 0.19 3
Ammonium 26 0.17 12
Potassium 4.8 0.05 5
Calcium 120 0.79 9
Magnesium 28 0.33 4
aLimits of determination are calculated for 150 1 air samples, 30 min samples at 5 I/min.
bIs dependent on carbonate concentrations.
Reproduced with permission from Ref. [233] (copyright 2001, Elsevier Science).

aerosol collection and analysis. This instrument also uses steam introduction for
particle growth followed by impaction and dual IC analysis for rapid measure-
ment of key species of interest, most notably nitrate, sulfate and ammonium.
Conventional dry denuders, replaced periodically, are used with this instrument.
In this instrument, the particle growth device is modeled after a design by
Okuyama [235]. A single jet inertial impactor is used to collect the droplets onto
a vertical glass plate that is continually washed with a constant water flow. The
device is shown in Fig. 6.25. The exit liquid flow is divided and then analyzed by a
dual channel ion chromatograph. In this initial implementation, integrated
samples were measured every 7 min. The instrument provides detection limits of
approximately 100 ng/m3 for chloride, nitrate, sulfate, sodium, ammonium,
calcium and potassium. The authors also experimented with butanol as a solvent
for vapor condensation on the aerosol using a commercial condensation particle

198
 5Umi.n Vacuum
Pump
Mixing Cooling Glass
Coils from Impactor 0.ImLJmin.
Chamber H20
Water Jacket \
Cartridge
Heater
\ from .. S GrowthChamber

Sample
A~ Drai~/ YTo IC
H20 from
Penstaltic Am-Ienr
Pump Aerosol

Fig. 6.25. Vapor condensation-impaction apparatus of Weber et al. Reproduced with permission
from Ref. [234] (copyright 2001, American Association of Aerosol Research).

Atlanta: Aug. 18 to 30, 1999

3a5 a: I-MM
S, Eastern Standard Time

0_

f00:00 2 exO
0:00 22. :c00:00 24:t02 0 2$:0 :000 26o : 2t e
Day: HH: MM: SS, Eastem Standard Time

Fig. 6.26. Intercomparison of the SJAC [226] and the GT/BNL [234] nitrate data from the
Atlanta Supersite 1999 experiment. Reproduced with permission from Ref. [226] (copyright
2001, Elsevier Science).

counter. Particle growth and collection was successfully accomplished but

butanol created problems with IC analysis. The use of such solvents may be more
compatible with other analytical techniques. Figure 6.26 shows the superb
agreement 'of nitrate measurement data between the GT/BNL and ECN
instruments at the 1999 Atlanta supersite.

Interferences from gases

At this time, none of the denuders in common use with wet aerosol collectors
effectively remove NO or NO2 . These gases can potentially generate nitrite and
nitrate ions with water (and steam may actually cause a greater extent of such a

199
 + SJAC ,t Filterpack O Thermoden.
25

E 20
0)

E
a_ 15
E

.g 10

0
.> 5

S
0 5 10 15 20 25
"3)
Average of 3 methods (pglm

Fig. 6.27. Intercomparison of nitrate measurement by three techniques. Caveat: Laboratory

study. Reproduced with permission from Ref. [226] (copyright 2001, Elsevier Science).

reaction), thereby causing errors in the measurement of particulate nitrite and

nitrate. The wet frit collection of nitrate aerosol [208] probably suffers from such
an artifact. The intrinsic solubilities of NO and NO2 are very low; Henry's law
constants are 1.93 x 10 3 and 1.0 x 10 2, respectively. These gases can react with
water to form HONO and HNO3 according to the following equations [236-237].
NO 2 + NO + H2 0, >2 HONO (6.1)
and
2 NO 2 + H2O -- HONO + HNO 3 . (6.2)
However, the reaction kinetics are so slow at room temperature that these
cannot constitute important pathways for the removal of these gases by particle
collection systems. As shown in Fig. 6.27, the nitrate data collected by the ECN
SJAC system and the dry FP and ECN thermodenuder systems are in excellent
agreement. It would appear therefore that artifact nitrate formation in the
vapor condensation systems is low or insignificant. All of the vapor condensation
systems see, however, measurable amount of what appears to be particulate
nitrite. It has been suggested that soot aerosols may provide catalytic sites to
form HONO from NO x and all or part of this particulate nitrite may be an
artifact of these systems.

6.3.2 Applications of aerosol collection and wet analysis

Numerous examples of ambient air measurement in outdoor ambient air for the
measurement of aerosol ionic composition have already been given. Here we look
at various other applications.

200
6.3.2.1 Indoor air quality: kerosene-fueled space heaters
In most societies, people are spending more and more time indoors, especially in
the winter. Assessment of total human exposure to pollutants obviously requires
measurement of indoor pollutants, which can be qualitatively and quantitatively
different from those outdoors. Residential indoor environments are significant
contributors to human exposure to airborne pollutants, including inhalable
particulate matter. Residential wood combustion [239] and environmental
tobacco smoke [240] have long been known to be major contributors to indoor
aerosols; it has more recently been shown that improper use of ultrasonic
humidifiers can lead to significant contributions to inhalable particulate matter
concentrations indoors [241]. Meat cooking has been shown to contribute to
organic aerosols [242].
Unvented kerosene space heaters have become a popular way to provide
localized supplementary domestic heating. The annual sale of such heaters
approaches a million units [243]. These are a potential indoor source of fine
particles, especially sulfate and acid aerosols, in many homes. Sulfate and acidic
aerosol levels in such homes can substantially exceed peak outdoor concentra-
tions [244]. Kerosene heaters can also release various organic products including
some mutagenic compounds [245,246].
The steam condensation-PCS instrument was utilized to measure the partic-
ulate anions and acid gases emitted by a kerosene-fueled space heater in a con-
trolled laboratory study; the results are shown in Fig. 6.28 [224]. To determine
worst-case conditions, air ventilation systems were deliberately shut off. Two
different heaters, rated at 9000 and 89000 BTU/h, were sequentially tested. Very
substantial concentrations of gaseous SO2 and HONO, particulate nitrite and
slightly lower concentrations of particulate sulfate were measured. Note that
the formation of particulate nitrite lags behind that of gaseous HONO and
HONO concentration decreases at a rate much slower than does SO2 after the
heaters are turned off. Further, the ratio of particulate nitrite over gaseous

Fig. 6.28. Concentrations of gaseous SO2 and

HONO, and particulate sulfate and nitrite in
emissions from kerosene fueled space heaters. (A)
Heater 1 (9000 BTU) lighted; (B) Heater I turned
off and Heater II (89000 BTU) lighted; (C) Cham-
ber door opened, (D) Heater II turned off. Inset:
an enlarged (x-expansion -2.8x y-expansion
-4.6 x) view of the boxed portion of the main plot
(ca. 8-10.5 min). Reproduced with permission
from Ref. [224] (copyright 1995, American
Chemical Society).

201
HONO was lower while the heater was lighted than the ratio observed an hour
after the heaters were switched off. It is likely that the primary emission consists
of HONO, NO, NO2 and SO 2 along with other gaseous and particulate species.
Particulate nitrite in these cases does not appear to be an artifact. What is being
measured as particulate nitrite can have two sources: the generation of nitrite
on the particle surface by the reaction of NOx with moisture, with or without
co-adsorbed SO 2 , and the adsorption of HONO already formed on to the parti-
cles. Both of these processes can cause the particulate nitrite concentration to
lag behind that of HONO. Particulate nitrite concentration predictably in-
creases with time. Production of indoor HONO by reaction of NO2 with surface
has been suggested earlier to account for the higher HONO concentrations
observed in homes that do not have unvented combustion appliances [247]. Sim-
ilar processes have been suggested to be responsible for at least part of outdoor
HONO [248,249].
Once the nitrite concentration on particles is sufficiently high, there will be a
net desorption of HONO after the heaters are turned off and primary emission of
HONO ceases. This likely explains the change in HONO to nitrite ratio as well as
the slower fall of HONO concentration after the heaters are turned off. When
significant amounts of NO2 are present, the reaction between sulfite and NO is
more important in the generation of nitrite [250,251]. It is clear that very
significant amounts of SO2 are present in kerosene heater emissions. While no
significant amounts of particulate S(IV) were seen, it would have been rapidly
oxidized to sulfate in the alkaline elution conditions in the IC (the IC eluent was
not deoxygenated). Large quantities of particulate sulfate were indeed observed;
this is not only due to the formation from the oxidation of sulfite, but also due to
primary emission of H2SO4 . Evidence of direct emission of H2 SO4 aerosols has
been acquired by the particle acidity measurement system described in a subse-
quent section, It is clear that the simple hydrolytic reaction of NO2 plays a
relatively minor role in these situations since such a reaction must produce
equal amounts of nitrite and nitrate. Significant amounts of aerosol nitrate were
not observed. It is unlikely that nitrate was not observed due to loss of nitric acid
to the walls as the measurement was continuous and the sampling point was
closer to the heater than to the walls.
Measurement in actual home environments with modest kerosene heaters
showed relatively low concentrations of particulate nitrite, <1 /g/m3. The con-
centration of HNO, and particulate nitrate was even lower. Maximum HONO
concentration was in the range of 10 to 20 g/m 3 (ca. 5-10 ppb) and is similar to
the values reported with open flame combustion sources in other studies
[247,252].

6.3.2.2 Indoor bioaerosolmeasurement

Case histories indicate that acute indoor pollution problems, so-called "sick
building syndrome", are as often caused by "bioaerosols" [253] (viral/fungalbac-
terial matter in airborne particle form), as from nonbiological sources. The

202
market for diagnosing and treating indoor air pollution problems is estimated to
be several billion dollars, of which a tenth is likely devoted to analytical and
consulting services [254]. Sick building syndrome represents the tip of the ice-
berg: ten times as many buildings that have not been designated "sick" are still
operating with an unhealthy indoor air environment, their occupants function-
ing sub-optimally. Public awareness of health effects of indoor air pollution is
presently at a peak. New standards for indoor air quality have been proposed.
The biggest concern lies with the children: the General Accounting Office
reports that severe indoor air quality problems exist in 15,000 US schools,
affecting nearly nine million children [255]. In the wake of today's bioterrorism,
these threats and concerns are likely to remain for a very long time to come.
To enable rapid treatment of indoor air pollution problems in a technically
sound manner, it must first be determined whether the problem is primarily
chemical or biological in nature. If, for example, it is diagnosed to be chemical,
portable gas chromatograph-mass spectrometers can be brought on site to
unambiguously characterize unusual amounts of any species present. However,
it is not completely straightforward-unusual concentrations of certain volatile
organic compounds may originate also from fungal or microbiological sources
[256]. It would be better to first decide if a biological problem exists.
Bioaerosols include airborne bacteria, vira, spores, pollens, plant/insect frag-
ments, etc. Above a threshold concentration, many of these species can elicit
pathogenic and allergenic effects. Indoor bioaerosols are currently being investi-
gated with the goal of discovering a cause for ongoing disease or discomfort [257]
to connect specific health effects with specific biocomponents, and to increase
understanding the ecology and physiology of the source organisms and the fate
of their airborne effluents [258]. Whether it is dead tissue or a living organism,
bioaerosols have a greater protein content than the ambient background of
dominantly inorganic aerosol. The vapor condensation based PCS coupled to an
online protein measurement system was used by Poruthoor et al. [259] to
measure protein concentrations in ambient air, in home environment as well as
a biologically contaminated hotel room.
Several modifications to the original steam condensation-PCS were made.
The new design is shown in Fig. 6.29. One limitation of the original PCS is
Peltier cooling, which requires relatively large current. Instead of the Peltier
cooled maze, a flattened glass coil (Fig. 6.29, inset b) made by coiling a glass tube
(120 cm x 0.5 cm i.d.) was used as the impactor. These authors found that
increasing the residence time after steam introduction facilitated the collection
of the particles. In the final arrangement, the mixing chamber, impactor, the
air/liquid separator and other downstream components are all mounted inside a
circular acrylic jacket and air cooled by a fan placed at the bottom of the acrylic
housing. (In this context, glass bead packed column impactors were also tested
for aerosol collection after steam introduction, with or without further water
addition [260]. It is doubtful that the results are competitive with the other
alternatives already described.)

203
 Aerosol

Air (b)

Fig. 6.29. Simplified vapor condensation particle collection system [259]. Inset: the details of (a)
the air/liquid separator and (b) the flattened glass coil impactor. Reproduced with permission
from Ref. [259] (copyright 1998, American Chemical Society).

The chemistry of the protein determination process relies on the Coomassie

Blue G chromogenic reaction with proteins. The Coomassie Blue-G (CBG) dye
binding method utilizes an adduct formation of CBG with proteins. This results
in a large shift in the absorption maximum. The unbound dye has an absorption
maximum at -470 nm whereas the protein dye complex has an absorption
maximum at -595 nm. The exact wavelengths depend on the pH and the
concentration of the dye. This reaction is not equally sensitive for all proteins
but does occur with almost all proteins [261]. The LOD (SIN = 3) for bovine
serum albumin as the test protein aerosol was calculated to be 100 ng/m 3 for a
300-1 air sample (10 /min, 30 min). With the advent of Teflon AF based long path
detection cells [262], it may be possible to improve on this detection limit.
Protein concentrations in ambient outdoor air were below the detection
capability of the system. On the other hand, proteinaceous aerosol levels above
background were found in a home environment during meat cooking activities.
In the case of a mold-contaminated hotel room, protein aerosol levels were very
significantly above background. Figure 6.30 shows the representative results for
the above measurements. These results indicate the potential capability of such

204
 nnl;. 7 ^
u.ul~
0 4 1 9 I
lo

o 0.010
o0
o

<0.005
1 2 3

n nn 11<k 4I I
u.uu
U (a) (b) (c) (d)

Fig. 6.30. Representative system output. (a) Blank; peaks 1-3 are from loading 8 ml blank each.
(b) 50 ng/ml BSA standard; peaks 4-6 are from loading 8 ml solution each. (c) System output in
an apartment dwelling, during cooking; peaks 7 and 8 are from 700 and 650 1 air sample. (d)
System output in a hotel room known to have fungal contamination; peaks 9 and 10 are from 450
and 420 1 air sampled inside the contaminated room. Reproduced with permission from Ref.
[259] (copyright 1998, American Chemical Society).

systems for identifying bioaerosol contamination. Presumably a detection meth-

od based on fluorescence tagging of the protein (e.g., with fluorescamine) will
greatly increase sensitivity.

6.3.2.3 Aerosol acidity measurement

The health effects of acid aerosols are of specific interest. While the upper
respiratory tract (the moist walls behaving as a 'wet diffusion denuder') effi-
ciently removes soluble gases such as SO2, inhalable particles (aerodynamic
diameter ca. 0.1-2 j-m) can penetrate to the deep lung and be deposited in the
alveolar region with high efficiency. Some studies conducted with "high
ambient" levels of acid aerosols show evidence of bronchoconstriction occurring
in young asthmatics [263]. The reliability of controlled animal/human exposures
at low levels can also be a concern because of neutralization of acid aerosols by
expired ammonia [264]. Laboratory studies strongly suggest that a synergism
exists between ozone damage to the lung and aerosol acidity [265].
In aerosol acidity measurement, it is vital that acidic/basic gases are first
removed without removing particles. This is accomplished with diffusion
denuders, as previously discussed. Because acidity in the ambient aerosol is
essentially confined to the fine particle fraction, it is also important to remove
larger particles (that tend to be crustal and alkaline) before measuring acidity;
this is typically accomplished with a cyclone at the front end.
If the aerosol is continuously dissolved into an aqueous phase to form a
solution, as in a vapor condensation-PCS, there are more opportunities for
designing a continuous monitoring system. If the anions and the cations (other
than H+ ) are determined, e.g., by IC, it becomes possible to determine H+ by
difference. However, with two ion chromatographs at the back end, such a
system would be bulky and expensive. The same objectives can be accomplished
if one measures the total anion equivalents and the total (non-H + ) cation
equivalents in the sample using the principles of suppressed conductometric

205
 H20

I Separator
W

Air '

Anion
0.50
2

mL/min
Peristaltic Pump

Fig. 6.31. Aerosol acidity instrument schematic. See text for details. Reproduced with permission
from Ref. [267] (copyright 1998, American Chemical Society).

detection [266] without chromatographic separation of the individual ions. This

is the operating principle of the instrument developed by Ito et al. [267]. The H+
equivalents are taken to be the measured anion equivalents minus the measured
non-H+ cation equivalents. The system is shown in Fig. 6.31. All processing is
carried out with a multichannel peristaltic pump. The aerosol extract from the
PCS is first passed through a cation preconcentration column (CPC) then
through an anion preconcentration column (APC). Water was pumped as a
carrier in the cation system. Valves were programmed to switch in a temporal
sequence. The loop of V3 is filled with 10 mM HCl. The OH--form packed column
suppressors were packed with an anion exchange resin of intermediate basicity.
In the anion measurement system, the eluent was 2 mM NaOH + 0.5 mM
Na 2 CO3 . An acid-regenerated tubular Nafion suppressor [268] was used. Be-
cause the liquid effluent from the PCS must be completely removed, over-
aspiration is applied and some air bubbles are aspirated. Before analysis,
trapped air bubbles are removed by switching V4 and pumping deionized water
through the columns while the VCACS effluent is sent to waste. The operational
cycle was typically 8-10 min: 5 min for sampling, 1 min for debubbling and 2-4
3
min for elution, washout and return to baseline. The LOD was 7-38 ng/m for a
5-8 min sample at 5 1/min. For a greater sample volume, e.g., 300 1,the LOD was
0.6-3.2 ng/m3, depending on the amount of total neutral salts concurrently
present. This is shown in more detail in Table 6.5.
Figure 6.32 presents unequivocal evidence as to how kerosene heaters
produce acid aerosols. Concurrent measurement of sulfate (not shown) indicate

206
TABLE 6.5
Aerosol acidity limits of detection as a function of background aerosol composition [269]
H2SO4 :(NH4 )2 SO4 Ratio Background No. data H+ LOD (ng/m3 )
(NH 4 )2SO4 points
Nominal in Found in coneh. g/im3) Experimental Projected
nebulizer feed aerosol 5SLPM, 5 min 5SLPM, 60 min
4:1 3.7:1 2.35 16 8.3 0.7
1:1 1:0.98 5.1 19 7.0 0.6
1:5 1:5.4 16.5 28 21.6 1.8
1:7.5 1:8.5 17.4 16 31.8 2.7
1:10 1:11 15.1 26 38.2 3.2
1:15 1:17 8.25 12 9.7 1.6
Reproduced with permission from Ref. [269] (copyright 1998, American Chemical Society).

I He.ater Off

-- r---l
14.00 16.00 18.00 20.00 22.00 24.00
Time, hours, March 2, 1996

Fig. 6.32. Kerosene heaters cause readily measurable acid aerosol emission. Reproduced with
permission from Ref. [267] (copyright 1998, American Chemical Society).

that the acidity is clearly traceable to the production of H2SO4 aerosol. The
application of this instrument during inhalation toxicology studies concerned
with the effect of low levels of H2 SO4 aerosols on humans in test chambers clearly
showed that almost all of the acid aerosol is actually neutralized by expired
ammonia in the chambers (Fig. 6.33).

6.3.2.4 Measurement of the fate of emitted H2 S in an oil field

Large quantities of H2S are emitted in oilfield operations. It is known that this is
eventually oxidized to sulfate but it has never been established if significant
quantities of SO2 are formed as an intermediate. The emitted H2 S can be

207
 3.0 -

1.0 -

0.0 -

I I I I 1 1 I I I I I
14.00 1450 15.00 15.50 15.50 16.50 17.00 17.5
Diumal inme, h

Fig. 6.33. It is difficult to perform human toxicology studies to low levels of acid aerosol because
expired ammonia neutralizes the aerosol. When three people walk into an exposure chamber
3
containing 200 Ag/m H2 SO,, the acid is rapidly neutralized. Reproduced with permission from
Ref. [267] (copyright 1998, American Chemical Society).

chemisorbed on alkaline dust particles and directly oxidized by photochemically

generated oxidants like ozone. Alternatively, it can be oxidized in a homoge-
neous manner to SO2 followed by slower oxidation to particulate sulfate. The
transport pattern of the pollutants will obviously be different in the two
different scenarios. During field deployment in a West Texas Oil field aboard a
motorhome-based mobile atmospheric research laboratory, a sudden accidental
release of H2 S was detected by a custom built H2 S measurement instrument
[269]. A wet denuder based SO2 measurement instrument and a vapor condensa-
tion-PCS-IC based sulfate measurement instrument then followed the concen-
tration of these species. As the results in Figure 6.34 show, a large concentration
of sulfate formed soon after and decayed slowly. The increase in SO2 concentra-
tion was much lower in magnitude. This clearly supports the mechanism
involving oxidation following chemisorption.

6.3.2.5 Vapor condensation based systems as collector front ends for the
measurement of trace metals.
A unique concern of relevance to aerosol composition measurements relates to
the monitoring of hazardous nuclear material that may be released in particu-
late form. Since the end of the cold war, large quantities of used nuclear
materials are being brought to the US for destruction and reprocessing. Acciden-
tal radioactive exposures due to human error can occur in all facilities handling
radioactive material. In the case of a major disaster, the threat of nuclear

208
 20-

i '"J -S02 [
15 - * Sulfate
[a

f :~''…-^

E 10-
a -

0-
11 12 13 14 15
Diurnal time (hours)
Fig. 6.34. Sulfur dioxide and sulfate concentrations after an accidental release of H2S in a West
Texas oilfield, release occurred at ca. 10:45 am. Oxidation proceeds directly to sulfate, rather
than via SO 2 . Reproduced with permission from Ref. [223] (copyright 1995, American Chemical
Society).

contamination to the population at large is mostly from the atmospheric

transport of the dispersed particulate material. This was amply proven in the
Chernobyl incident. Sensitive atmospheric measurement instrumentation is
also expected to be of great help in identifying clandestine nuclear activities
because typically fingerprint radionuclides are released to the atmosphere
[270,271]. In this connection, Plutonium is a particular concern.
With Pu, the chemical toxicity is at least as great a concern as its radioactive
properties, an aspect not always appreciated. In addition, while the detection of
many actinides and trans-actinides based on oa-spectrometry is very sensitive,
with a half-life of 2.44 x 104 y, the detection of Pu 239 is not especially sensitive. A
single aerosol particle of PuO 2 (any Pu exposed to an oxic atmosphere is rapidly
converted to PuO,), 10 s/m in diameter, contains 5.8 ng of the metal but results in
only 14 disintegrations/s. Based on statistical averages, a period in excess of 72 s
will elapse before a single disintegration can occur in a 1 um PuO, particle. As a
comparison, current state of the art elemental detection systems such as
induction coupled plasma-mass spectrometers (ICP-MS) are typically capable of
a low to double digit pg/ml limit of detection (LOD) using an -1 ml sample.
These instruments have not yet been adapted for the use of air in the plasma.
To that end, feasibility of collecting and detecting a metal containing aerosol
at very low concentrations was evaluated, using a vapor condensation-PCS.
Because of its similar chemical behavior, Cerium is often used as an actinide
surrogate. Trivalent Ce is also detected easily by solution phase fluorescence
with LODs comparable to that obtainable by ICP-MS. LODs of the order of 200
fmol/m3 in air were demonstrated for aerosol Ce(III) [272].

209
6.4 CONCLUSIONS

The demand for (near) real-time aerosol composition measurement has never
been greater. While the focus of this review has been primarily on ambient
outdoor air measurement, there is already worldwide desire to find an effective
means of infective bioaerosol detection. Major inroads have been made in mass
spectrometric identification of intact bacteria [273-275] but the same concerns
that exist with single particle MS studies of ambient air chemistry also exist
here-how practical will it be to deploy such instruments widely? While this
author does not claim to have a crystal ball, it seems likely in the near future that
immunosensing approaches coupled to wet effluent aerosol collectors hold much
promise [276]. Fundamental investigations in both stratospheric and tropo-
spheric chemistry will likely be dominated in the future by particle MS tech-
niques. On the other hand, near real-time systems that integrate collection and
can measure a broad suite of substances at one time (by chromatography or
chromatography-MS techniques) hold much promise in regulatory, epidemiolog-
ical and toxicological applications and more routine measurements.

Acknowledgement

This manuscript would have been impossible without the help of Kajori
Dasgupta and Susie Gonzalez. PKD gratefully acknowledges comments by Don
C. Olson and Zbynek Vecera on the draft manuscript. We also thank Janusz
Pawliszyn for his patience. This review was made possible in part by the
Tropospheric Ozone Research Program of the National Exposure Research
Laboratory, U.S. EPA.

REFERENCES

1 P.H. McMurry, Atmos. Environ., 34 (2000) 1959-1999.

2 P. Babich, W. Peng-Yau, G. Allen, C. Sioutas and P. Koutrakis, Aerosol Sci. Technol.
32 (2000) 309-324.
3 C.A. Pope, M.J. Thun, M.M. Namboodiri, D.W. Dockery, J.S. Evans, F.E. Speizer and
C.W. Heath, Am. J. Resp. Crit. Care, 151 (1995) 669-674.
4 C.A. Pope III, D. Dockery and J. Schwartz, Inhal. Toxicol., 7 (1995) 4-18.
5 R.B. Schlesinger and L.C. Chen, Environ. Res., 65 (1994) 69-85.
6 J. Schwartz, Environ. Res., 64 (1994) 26-35.
7 J. Schwartz, Environ. Res., 64 (1994) 36-52.
8 W.C.L. Hemeon, J. Air Pollut. Control Assoc., 23 (1973) 376-387.
9 R.T. Burnett, R.E. Dales, M.E. Raizenne, D. Krewski, P.W. Summers, G.R. Roberts,
M. Raad Young, T. Dan and J. Brook, Environ. Res., 65 (1994) 172-194.
10 P.H.N. Saldiva, A.J.F.C. Lichtenfels, P.S.O. Paiva, I.A. Barone, M.A. Martins, E.
Massad, J.C.R. Pereira, V.P. Xavier, J.M. Singer and G.M. Bohm, Environ. Res., 65
(1994) 218-225.
11 R.B. Schlesinger, Crit. Rev. Toxicol., 20 (1990) 257-286.
12 J. Ayres, D. Fleming, M. Williams and G. McInnes, Environ. Health Persp., 79 (1989)
83-88.

210
13 D.V. Bates, M. Baker-Anderson and R. Sizto, Environ. Res., 51 (1990) 51-70.
14 G.D. Thurston, K. Ito, M. Lippmann and C. Hayes, Envuiron. Health Persp., 79 (1989)
73-82.
15 M.E. Raizenne, R.T. Burnett, B. Stern, C.A. Franklin and J.D. Spengler, Environ.
Health Persp., 79 (1989) 179-185.
16 T.A. Kimmel, L.C. Chen and C. Nadziejkp, Toxicol. Appl. Pharm., 144 (1997)
348-355.
17 R.F. Phalen and D.V. Bates, Inhal. Toxicol., 7 (1995) 1-163.
18 J.C. Chow, J. Air Pollut. Contr. Assoc., 45 (1995) 320-382.
19 R.B. Schlesinger, Inhal. Toxicol., 7 (1995) 99-110.
20 A.L. Hines, T.K. Ghosh, S.K. Loyalka and R.C. Warder Jr. IndoorAir Quality and
Control. PTR Prentice Hall, New Jersey, 1995.
21 H. Klingenberg and H. Winneke, Sci. Total Environ., 93 (1990) 95-105.
22 I.L. Mauderly, in: M. Lippmann (Ed.), Environmental Toxicants: Human Exposures
and Their Health Effects. Van Nostrand Reinhold, New York, 1992, pp. 119-162.
23 R.A. Goyer, in: C.D. Klaassen, M.O. Amdur and J. Doull (Eds.), Toxicology.
Macmillan, New York, 1986, pp 582-635.
24 D.V. Bates, in: D.E. Gardner (Ed.), Proceedings of the Colloquium on Particulate Air
Pollution and Human Mortality and Morbidity. Inhal. Toxicol., 7(1) (1995) ix-xiii.
25 A. Gupta, D. Tang and P.H. McMurry, J. Atmos. Chem., 20 (1995) 117-139.
26 R.W. Talbot, A.S. Vijgen and R.C. Harris, J. Geophys. Res., 95 (1990) 7553-7561.
27 M. Ferm, Atmos. Environ., 20 (1986) 1193-1201.
28 J.L. Durham, L.L. Spiller and T.G. Ellestad, Atmos. Environ., 21 (1987) 589-598.
29 W. John, S.M. Wall and J.L. Ondo, Atmos. Environ., 22 (1988) 1627-1635.
30 B.R. Appel, V. Povard and E.L. Kothny, Atmos. Environ., 22 (1988) 2535-2540.
31 A.M.N. Kitto and R.M. Harrison, Atmos. Environ., 26A (1992) 235-241.
32 Air quality criteria for particulate matter. US. EPA. Research Triangle Park, NC;
EPA/600-AP-95-1001A.
33 J. Samet and S. Jager, as cited in ref. 34.
34 T. Reichhardt, Environ. Sci. Technol., 29 (1995) 360A-364A.
35 G. Oberdorster, R.M. Gelein, J. Ferin and B. Weiss, Inhal. Toxicol., 7 (1995) 111-124.
36 D.A. Lundgren and R.M. Burton, Inhal. Toxicol., 7 (1995) 131-148.
37 D.J. Jacob, J.M. Waldman, J.W. Munger and M. Hoffman, J. Geophys. Res., 91 (1986)
1089-1096.
38 J.G. Watson, J.C. Chow, J.J. Shah and T. Pace, JAPCA, 33 (1983) 114-119.
39 W.A. Hoppel, J.W. Fitzgerald, G.M. Frick and R.E. Larson, J. Geophys. Res., D4
(1990) 3659-3686.
40 W. John, S.M. Wall, J.L. Ondo and W. Winklmayr, Atmos. Environ., 24A (1990)
2349-2359.
41 C.S. Sloane, J.G. Watson, J.C. Chow, L.C. Pritchett and L.W. Richards, Atmos.
Environ., 25A (1991) 1013-1024.
42 M.E. Kitto and D.L. Anderson, Atmos. Environ., 22 (1988) 2629-2630.
43 J.P. Lodge, Jr., Atmos. Environ., 20 (1986) 1657.
44 C.V. Mathai, J.G. Watson, Jr., C.F. Rogers, J.C. Chow, I. Thombach, J.O. Zwicker, T.
Cahill, P. Feeney, R. Eldred, M. Pitchford and P.K. Mueller, Environ. Sci. Technol.,
24 (1990) 1090-1099.
45 D.A. Lane, N.D. Johnson, S.C. Barton, G.H.S. Thomas and W.H. Schroeder, Environ.
Sci. Technol., 22 (1988) 941-947.
46 S. Rapsomanikis, M. Wake, A.M.N. Kitto and R.M. Harrison, Environ. Sci. Technol.,
22 (1988) 948-952.
47 K.T. Knapp, J.L. Durham and T.G. Ellestad, Environ. Sci. Technol., 20 (1986)
633-637.

211
48 W.T. Sturges and R.M. Harrison, Atmos. Environ., 23 (1989) 1987-1996.
49 W. John, S.M. Wall and J.L. Ondo, Atmos. Environ., 22 (1988) 1627-1635.
50 J.F. Pankow, Atmos. Environ., 2 (1987) 2275-2283.
51 P. Koutraldkis, J.M. Wolfson and J.D. Spengler, Atmos. Environ., 22 (1988) 157-162.
52 W.T.Sturges and G.E. Shaw, Atmos. Environ., 27A (1993) 2969-2977.
53 H.V. Andersen and M.F. Hovmand, Atmos. Environ., 28 (1994) 3495-3512.
54 B.Y.H. Liu, D.Y.H. Pui and K.L. Rubo, in: Aerosols in the Mining and Industrial
Work Environments. Ann Arbor Science, Ann Arbor, MI, 1983.
55 W.H. Chan, D.B. Orr and D.H.S. Chung, Atmos. Environ., 20 (1986) 2397-2401.
56 T.R. Fogg, Atmos. Environ., 20 (1986) 1633-1634.
57 V.R. Highsmith, A.E. Bond and J.E. Howes, Jr., Atmos. Environ., 20 (1986)
1413-1417.
58 X. Zhang and P.H. McMurry, Atmos. Environ., 26A (1992) 3305-3312.
59 S.V. Hering, D.R. Lawson, I. Allegrini, A. Febo, C. Perrino, M. Possanzini, J.E.
Sickles, II, K.G. Anlauf, A. Wiebe, B.R. Appel, W. John, J. Ondo, S. Wall, R.S.
Braman, R. Sutton, G.R. Cass, P.A. Solomon, D.J. Eatough, N.L. Eatough, E.C. Ellis,
D. Grosjean, B.B. Hicks, J.D. Womack, J. Horrocks, K.T. Knapp, T.G. Ellestad, R.J.
Paur, W.J. Mitchell, M. Pleasant, E. Peake, A. MacLean, W.R. Pierson, W.
Brachaczek, H.I. Schiff, G.I. Mackay, C.W. Spicer, D.H. Stedman, A.M. Winer, H.W.
Biermann and E.C. Tuazon, Atmos. Environ., 22 (1988) 1519-1539.
60 P. Masia, V. Di Palo and M. Possanzini, Atmos. Environ.,28 (1994) 365-366.
61 F.W. Lipfert, Atmos. Environ., 28 (1994) 3233-3249.
62 B.L. Davis, Y. Deng, D.J. Anderson, L.R. Johnson, A.G. Detwiler, L.L. Hodson and
J.E. Sickles, Atmos. Environ., 28 (1994) 2485-2491.
63 W.L. Crider, Anal. Chem., 37 (1965) 1770-1773.
64 J.D. Husar, R.B. Husar and P.K. Stubits, Anal. Chem., 47 (1975) 2062-2065.
65 P.T. Roberts and S.K. Friedlander, Atmos. Environ., 10 (1976) 403-408.
66 J.J. Huntzicker, R.S. Hoffman and R.A. Cary, Atmos. Environ., 12 (1978) 83-88.
67 J. Coburn, R.B. Husar and J.D. Husar, Atmos. Environ., 12 (1978) 89-98.
68 R.L. Tanner, T. D'Ottavio, R. Garber and L. Newman, Atmos. Environ., 14 (1980)
121-127.
69 T. D'Ottavio, R.L. Garber, R.L. Tanner and L. Newman, Atmos. Environ., 15 (1981)
197-203.
70 J. Slanina, L.V. Lamoen-Dorrnenbal, W.A. Lingera, W. Meilof, D. Klockow and R.
Niessner, Int. J. Environ. Anal. Chem., 9 (1981) 59-70.
71 R.W. Garber, P.H. Daum, R.F. Doering, T. D'Ottavio and R.L. Tanner, Atmos.
Environ., 17 (1983) 1381-1385.
72 B.R. Appel, R.L. Tanner, D.F. Adams, P.K. Dasgupta, K.T. Knapp, G.L. Kok, W.R.
Pierson and K.D. Reiszner, in: J.P. Lodge (Ed.), Methods of Air Sampling and
Analysis, 3rd edn. Lewis, Chelsea, MI, 1988, pp. 523-532.
73 J. Slanina, C.A.M. Schoonebeek, D. Klockow and R. Niessner, Anal. Chem., 57 (1985)
1955-1960.
74 J.J. Huntzicker, Anal. Chem., 58 (1986) 653-654.
75 F. Lindqvist, Atmos. Environ., 19 (1985) 1671-1680.
76 S. Hering and M.R. Stolzenburg, Integrated collection and vaporization particle
chemistry monitoring, US Patent, 5,983,732, November, 1999.
77 M.R. Stolzenburg and S.V. Hering, Environ. Sci. Technol., 34 (2000) 907-914.
78 http://www.rpco.com/products /ambprod/amb400sindex.htm
79 G.A Allen,. Personal communication, October, 2001.
80 http://www.thermoei.com
81 A.H. Moscowitz, Particle size distribution of nitrate aerosols in the Los Angeles air
basin. EPA-600/3-77-053, 1977.

212
 82 R.D. Cox, Anal. Chem., 52 (1980) 332-335.
83 D. Klockow, R. Niessner, M. Malejczyk, H. Kiendl, B. Vomberg, M.P. Keuken, A.
Wayers-Ypelaan and J. Slanina, Atmos. Environ., 23 (1989) 1131-1138.
84 K. Yoshizumi and A. Hoshi, Environ. Sci. Technol., 19 (1985) 258-261.
85 W.T. Sturges and R.M. Harrison, Environ. Sci. Technol., 22 (1988) 1305-1311.
86 H.M. Ten Brink, A. Wayers-Ypellan and S. Slanina, J. Aerosol Sci., 29 (1998)
S633-S634.
87 M. Yamamoto and H. Kosaka, Anal. Chem., 66 (1994) 362-367.
88 http://www.rpco.com/products/ambprod/amb8400n/index.htm
89 B.R. Appel, in: K. Willeke and P.A. Baron (Eds.), Aerosol Measurement-Principles,
Techniques and Applications. Van Nostrand Reinhold, New York, 1993, pp. 233-259.
90 R. Klockenkimper, H. Bayer, A. von Bohlen, M. Schmeling and D. Klockow, Anal.
Sci., 11 (1995) 495-501.
91 D. Pardess, Z. Levin and E. Ganor, Atmos. Environ., 26A (1992) 675-680.
92 P.J. Sheridan, R.C. Schnell, J.D. Kahl, J.F. Boatman and D.M. Garvey, Atmos.
Environ., 27A (1993) 1169-1183.
93 S. Brown, M.C. Dangler, S.R. Burke, S.V. Hering and D.T. Allen, Aerosol Sci.
Technol., 12 (1990) 172-181.
94 W.L. McClenny, K.J. Krost, E.H. Daughtrey, D.D. Williams and G.A. Allen, Appl.
Spectrosc., 48 (1984) 706-712.
95 K.J. Krost and W.L. McClenny, Appl. Spectrosc., 48 (1984) 702-705.
96 B.R. Appel, Y. Tokiwa and M. Haik, Atmos. Environ., 18 (1984) 409-416.
97 E. Bondietty and C. Papastefanou, J. Aerosol Sci., 20 (1989) 667-670.
98 X. Zhang and P.H. McMurry, Atmos. Environ., 26A (1992) 3305-3312.
99 X. Li-Jones, D.L. Savoie and J.M. Prospero, Atmos. Environ., 35 (2001) 985-992.
100 W.H. Chan, D.B. Orr and D.H.S. Chung, Atmos. Environ., 20 (1986) 2397-2401.
101 F.W. Lipfert, Atmos. Environ., 28 (1994) 3233-3249.
102 R.W. Coutant, Environ. Sci. Technol., 11 (1977) 873-878.
103 Z. Ali, C.L. Paul and J.F. Alder, Analyst, 114 (1989) 759-769.
104 G. Lammel, Fres. Z. Anal. Chem., 340 (1991) 684-986.
105 C.L. Benner and D.J. Eatough, Atmos. Environ., 25A (1991) 1537-1545.
106 C. Rosenberg, W. Winiwarter, M. Gregori, G. Pech, V. Casensky and H. Puxbaum,
Fres. Z. Anal. Chem., 331 (1988) 1-7.
107 B.J. Turpin, S.-P. Liu, K.S. Podolske, M.S.P. Gomes, S.J. Eisenreich and P.H.
McMurry, Environ. Sci. Technol., 27 (1993) 2441-2449.
108 R.L. Myers and W.L. Fite, Environ. Sci. Technol., 9 (1975) 334-336.
109 B.A. Mansoori, M.V. Johnston and A.S. Wexler, Anal. Chem., 66 (1994) 3681-3687.
110 M.P. Sinha, C.E. Giffin, D.D. Norris, T.J. Estes, V.L. Vilker and S.K.J. Friedlander,
J. Colloid Interface Sci., 87 (1982) 140-153.
111 K.P. Hinz, R. Kaufmann and B. Spengler, Anal. Chem., 66 (1994) 2071-2076.
112 K.P. Hinz, R. Kaufmann and B. Spengler, Aerosol Sci. Technol., 24 (1996) 233-242.
113 O. Kievit, J.C.M. Marijinissen, P.J.T. Verheijen and B. Scarlett, J. Aerosol Sci., 23
(1992) S301-S304.
114 P.J. McKeown, M.V. Johnson and D.M. Murphy, Anal. Chem., 63 (1991) 2069-2073.
115 D.M. Murphy and D.S. Thomson, Aerosol Sci. Technol., 22 (1995) 237-249.
116 P.G. Carson, K.R. Neubauer, M.V. Johnson and A.S. Wexler, J. Aerosol Sci., 26
(1995) 535-545.
117 K.A. Prather, T. Nordmeyer and K. Salt, Anal. Chem., 66 (1994) 3540-3542.
118 D.-Y. Liu, D. Rutherford, M. Kinsey and K.A. Prather, Anal. Chem., 69 (1997)
1808-1814.
119 E. Gard, J.E. Mayer, B.D. Morrical, T. Dienes, D.P. Fergenson and K.A. Prather,
Anal. Chem., 69 (1997) 4083-4091.

213
120 D.M. Murphy and D.S. Thomson, J. Geophys. Res., 102 (1997) 6341-6352.
121 D.M. Murphy and D.S. Thomson, J. Geophys. Res., 102 (1997) 6353-6368.
122 D.T. Suess and K.A. Prather, Chem. Rev., 99 (1999) 3007-3035.
123 M.V. Johnston, J. Mass Spectrom., 35 (2000) 585-595.
124 C.A. Noble and K.A. Prather, Mass Spectrom. Rev., 19 (2000) 248-274.
125 A.D. Maynard, Philos. Trans. Roy. Soc. A, 358 (2000) 2593-2609.
126 A. Lazar, P.T.A. Reilly, W.B. Whitten and J.M. Ramsey, Environ. Sci. Technol., 33
(1999) 3993-4001.
127 R.P. Rodgers, A.C. Lazar, P.T.A. Reilly, W.B. Whitten and J.M. Ramsey, Anal.
Chem., 72 (2000) 5040-5046.
128 A. Lazar, R.A. Gieray, W.B. Whitten and J.M. Ramsey, Combust. Flame, 122 (2000)
90-104.
129 P.T.A. Reilly, R.A. Gieray, W.B. Whitten and J.M. Ramsey, J. Am. Chem. Soc., 122
(2000) 11596-11601.
130 A. Lazar, P.T.A. Reilly, W.B. Whitten and J.M. Ramsey, Environ. Sci. Technol., 72
(2000) 2142-2147.
131 A. Lazar, P.T.A. Reilly, W.B. Whitten and J.M. Ramsey, Rapid Commun. Mass
Spect., 14 (2000) 1523-1529.
132 P.T.A. Reilly, A.C. Lazar, R.A. Gieray, W.B. Whitten and J.M. Ramsey, Aerosol Sci.
Technol., 33 (2000) 135-152.
133 D.B. Kane and M.V. Johnston, Environ. Sci. Technol., 34 (2000) 4887-4893.
134 D.B. Kane, B. Oktem and M.V. Johnston, Aerosol Sci. Technol., 34 (2000) 520-527.
135 D.J. Phares, K.P. Rhoads, A.S. Wexler, D.B. Kane and M.V. Johnston, Anal. Chem.,
73 (2001) 2338-2344.
136 http://www-wlc.eas.gatech.edu/supersite/
137 A.M. Middlebrook, D.M. Murphy, S.-H. Lee, D.S. Thomson, K.A. Prather, R.J.
Wenzel, D.-Y. Liu, D.J. Phares, K.P. Rhoads, A.S. Wexler, M.V. Johnston, J.L.
Jimenez, J.T. Jayne, D.R. Worsnop, I. Yourshaw, J.H. Seinfeld, R.C. Flagan, S.V.
Hering, R.J. Weber, P. Jongejan, J. Slanina and P.K. Dasgupta, J. Geophys. Res. (in
press).
138 C.D. Clark, P. Campuzano-Jost, D.S. Covert, R.C. Richter, H. Maring, A.J. Hynes and
E.S. Saltzman, J. Aerosol Sci., 32 (2001) 765-778.
139 J. Bacri, A.M. Gomes, J.M. Fieni, F. Thouzeau and J.C. Birolleau, Spectrochim. Acta,
44B (1989) 887-895.
140 D. Nore, A.M. Gomes, J. Bacri and J. Cabe, Spectrochim. Acta, 48B (1993) 1411-1419.
141 A.M. Gomes, J.-P. Sarrette, L. Madon and A. Almi, Spectrochim. Acta, 51B (1996)
1695-1705.
142 A.M. Gomes, C. Trassy, A. Almi, F. Seddiki and S. Hassaine, High Temp. Mater.
Proc., 1 (1997) 461-472.
143 A. Epifanie, J. Bacri, J.P. Sarrete and A.M. Gomes, High Temp. Mater. Proc., 1 (1997)
83-95.
144 A.M. Gomes, A. Almi, P. Teulet and J.P. Sarrette, Spectrochim. Acta, 53B (1998)
1567-1582.
145 Y. Duan, Y. Su, Z. Jin and S. Abeln. Anal. Chem., 72 (2000) 1672-1679.
146 Y. Duan, Y. Su, Z. Jin and S. Abeln, A/P, 71 (2000) 1557-1563.
147 C. Sioutas, P. Koutrakis and B.A Olson, Aerosol Sci. Technol., 21(1994) 223-235;
Particul. Sci. Technol., 12 (1994) 207-221.
148 C. Sioutas, P. Koutrakis and R.M. Burton, J. Aerosol Sci., 25 (1994) 1321-1330.
149 C. Sioutas, P. Koutrakis and R.M. Burton, Environ. Health Persp., 103 (1995)
172-177.
150 J.J. Shah, J.G. Watson, J.A. Cooper and J.J. Huntzicker, J. Air Poll. Contr.Assoc., 36
(1986) 254-257.

214
151 H.A. Gray, G.R. Cass, J.J. Huntzicker, E.K. Heyerdahl and J.A. Rau, Environ. Sci.
Technol., 20 (1986) 580-589.
152 J.H. Offenberg and J.E. Baker, Atmos. Environ., 34 (2000) 1509-1517.
153 A. Molndr, E. M6szdros, H.C. Hansson, H. Karlsson, A. Gelencs6r, G. Kiss and Z.
Krivdcsy, Atmos. Environ., 33 (1999) 2745-2750.
154 J.C. Chow, J.G. Watson, L.C. Pritchett, W.R. Pierson, C.A. Frazier and R.G. Purcell,
Atmos. Environ., 27A (1993) 1185-1201.
155 M. Bizjak, R. Cigler, A.D.A. Hansen and V. Hudnik, Atmos. Environ., 27A (1993)
1347-1350.
156 M. Ammann, M. Kalberer, D.T. Jost, L. Tobler, E. Rossler, D. Piguet, H.W. Gaggeler
and U. Baltensperger, Nature, 395 (1998) 157-160.
157 M. Kalberer, M. Ammann, H.W. Gaggeler and U. Baltensperger, Atmos. Environ., 33
(1999) 2815-2822.
158 M. Bizjak and J. Tursic, J. Aerosol Sci., 29 (1998) S291-S292.
159 A.D.A. Hansen, H. Rosen and T. Novakov, Appl. Opt., 21 (1982) 3060-3062,
160 A.D.A. Hansen, H. Rosen and T. Novakov, Sci. Total. Environ., 36 (1984) 191-196.
161 www.mageesci.com
162 J.F. Hopper, D.E.J. Worthy, L.A. Barrie and N.B.A. Trivett, Atmos. Environ., 28
(1994) 3047-3054.
163 R.L. Gunter, A.D.A. Hansen, J.F. Boatman, B.A. Bodhaine, R.C. Schnell and D.M.
Garvey, Atmos. Environ., 27A (1993) 1363-1368.
164 A.D.A. Hansen, A.V. Pollissar and R.C. Schnell, Atmos. Res., 44 (1997) 153-165.
165 M. Bizjak, J. Tursic, M. Lesnjak and T. Cegnar, Atmos. Environ., 33 (1999) 2783-
2787.
166 S.G. Jennings, F.M. McGovern and W.F. Cooke, Atmos. Environ., 27A (1993)
1229-1239.
167 C. Liousse, H. Cachier and S.G. Jennings, Atmos. Environ., 27A (1993) 1203-1211.
168 L.A. Gundel, R.L. Dod, H. Rosen and T. Novakov, Sci. Total Environ., 36 (1984)
197-202.
169 K. Ruoss, R. Dlugi, C. Weigl and G. Hanel, Atmos. Environ., 27A (1993) 1221-1228.
170 A. Petzold, C. Kopp and R. Niessner, Atmos. Environ., 31 (1997) 661-672.
171 C. Kopp, A. Petzold and R. Niessner, J. Aerosol Sci., 30 (1999) 1153-1163.
172 J. Ballach, R. Hitzenberger, A.E. Schultz and W. Jaeschke, Atmos. Environ., 35
(2001) 2089-2100.
173 K.M. Adams, L.I. Davis, Jr., S.M. Japar and W.R. Pierson, Atmos. Environ., 23 (1989)
639-700.
174 K.M. Adams, L.I. Davis, Jr., S.M. Japar, D.R. Finley and R.A. Cary, Atmos. Environ.,
24 (1990) 597-604.
175 K.M. Adams, L.I. Davis, Jr., S.M. Japar and D.R. Finley, Atmos. Environ., 24 (1990)
605-610.
176 A. Petzold and R. Niessner, Appl. Phys., B63 (1996) 191-197.
177 W.P. Arnott, H. Moosmiiller, C.F. Rogers, T. Jin and R. Bruch, Atmos. Environ., 33
(1999) 2845-2852.
178 A. Petzold and R. Niessner, J. Aerosol Sci., 17 (1986) 705-714.
179 A. Petzold and R. Niessner, J. Aerosol Sci., 24 (1993) S195-S196.
180 S. Keller and J. Heintzenberg, J. Aerosol Sci., 28 (1997) S609-S610.
181 M. Pollard, J. Jaklevic and J. Howes, Aerosol Sci. Technol., 12 (1990) 182-193.
182 E. Schultz, 27A (1993) 1241-1249.
183 M. Bizjak, J. Tursic, V. Hudnik, P. Pavli, R. Cigler, B. Paradiz and B. Paradiz, J.
Aerosol Sci., 28 (1997) S207-S208.
184 V.M.H. Lavanchy, H.W. Gaggeler, S. Nyeki and U. Baltensperger, Atmos. Environ.,
33 (1999) 2759-2769.

215
185 T.A. Pakkanen, C.H. Ojanen, R.E. Hillamo and T. Mkela, J. Aerosol Sci., 29 (1997)
S243-S244.
186 C. Bhugwant, H. Cachier, M. Bessafi and J. Leveau, Atmos. Environ., 34 (2000)
3463-3473.
187 L.M. Castro, C.A. Pio, R.M. Harrison and D.J.T. Smith, Atmos. Environ., 33 (1999)
2771-2781.
188 A.D.A. Hansen and T. Novakov, Aerosol Sci. Technol., 12 (1990) 194-199.
189 G.A. Allen, J. Lawrence and P. Koutrakis, Atmos. Environ., 33 (1999) 817-823.
190 H. Schmid, L. Laskus, H.J. Abraham, U. Baltensparger, V. Lavanchy, M. Bizjak, P.
Burba, H. Cachier, D. Crow, J. Chow, T. Gnauk, A. Even, H.M. Ten Brink, K.-P.
Giesen, R. Hitzenberger, C. Hueglin, W. Maenhaut, C. Pio, A. Carvalho, J.-P. Putaud,
D. Toom-Sauntry and H. Puxbaum, Atmos. Environ., 35 (2001) 2111-2121.
191 A. Even, A. Khlystov and H.M. Ten Brink, J. Aerosol Sci., 29 (1998) S873-S874.
192 B.J. Turpin, R.A. Cary and J.J. Huntzicker, Aerosol Sci. Technol., 12 (1990) 161-171.
193 J.C. Chow, J.G. Watson, D. Crow, G.H. Lowenthal and T. Merrifield, Aerosol Sci.
Technol., 34 (2001) 23-34.
194 K. Fung, Aerosol Sci. Technol., 12 (1990) 122-127.
195 R. Hitzenberger, S.G. Jennings, S.M. Larsson, A. Dillner, H. Cachier, Z. Galambos,
A. Rouc and T.G. Spain, J. Aerosol Sci., 29 (1998) S751-S752.
196 A. Petzold and R. Niessner, Mikrochim. Acta, 117 (1995) 215-237.
197 A.D.A. Hansen and P.H. McMurry, 40 (1990) 894-895.
198 T.T. Mercer, Aerosol Technology in HazardEvaluation. Academic Press, New York,
1973, p 146.
199 B.J. Finlayson-Pitts and J.N. Pitts, Jr., Atmospheric Chemistry: Fundamentalsand
Experimental Techniques. Wiley, New York, 1986, p. 823.
200 K.T. Whitby and W.E. Clark, Tellus, 18 (1996) 573-586.
201 H-C. Yeh, in: K. Willeke and P.A. Baron (Eds.), Aerosol Measurement Principles,
Techniques andApplications. Van Nostrand Reinhold, New York, 1993, pp. 410-425.
202 S. Liu and P.K. Dasgupta, Talanta, 43 (1996) 1681-1688.
203 R. Gunn, J. Colloid Sci., 10 (1955) 107-119.
204 D.K. Brandvold, P. Martinez and R. Hipsch, Atmos. Environ., 30 (1996) 973-976.
205 S. Liu and P.K. Dasgupta, Anal. Chem., 68 (1996) 3638-3644.
206 A. Karlsson, K. Irgum and H.-C. Hansson, J. Aerosol Sci., 28 (1997) 1539-1551.
207 W.C. Hinds. Aerosol Technology. 1982, Chapter 9, pp. 164-186.
208 S.M. Buhr, M.P. Buhr, F.C. Fehsenfeld, J.S. Holloway, U. Karst, R.B. Norton, D.P.
Parrish and R.E. Sievers, Atmos. Environ., 26 (1995) 2609-2624.
209 A. Neftel, Atmospheric Research Laboratory, Swiss Department of Agriculture,
Liebfeld, Switzerland, personal communication, 2001.
210 G. Samanta, C.B. Boring and P.K. Dasgupta, Anal. Chem., 73 (2001) 2034-2040.
211 C.B. Boring, R. Al-Horr, G. Zhang, P.K. Dasgupta, M.W. Martin and W.F. Smith,
Anal. Chem., (submitted).
212 H. Small, Ion Chromatography.Plenum, New York, 1989, pp. 68-69.
213 http://oxmedinfo.jr2.ox.ac.uk/Pathway/Miscell/24028.htm
214 W.R. Cofer, V.G. Collins and R.W. Talbot, Environ. Sci. Technol., 19 (1985) 557.
215 W.R. Cofer, III and R.A. Edahl, Jr. Atmos. Environ., 20 (1986) 979.
216 J. Janak and Z. Vecera, Anal. Chem., 59 (1987) 1494-1498.
217 J. Janak and Z. Vecera, Mikrochim. Acta, 3 (1990) 29-34.
218 R. Al-Horr. G. Samanta and P.K. Dasgupta, to be published, 2002.
219 J. Aitken, Proc. Roy. Soc. Edinburgh(1888) 35-40.
220 Y.S. Cheng, in: K. Willeke and P.A. Baron (Eds.), Aerosol Measurement Principles,
Techniques andApplications.Van Nostrand Reinhold, New York, 1993, pp 427-448.
221 L. Waldmann and K.H. Schmitt, in: C.N. Davies (Ed.), Aerosol Science. Academic

216
 Press, New York, 1966, pp 137-162.
222 A. Blatter, A. Neftel, P.K. Dasgupta and P.K. Simon, in: G. Angletti and G. Restelli
(Eds.), Physico-Chemical Behavior of Atmospheric Pollutants, Proc. 6th European
Symposium, Report EUR 15609/2 EN, Luxembourg, 1994, pp. 767-772.
223 P.K. Simon and P.K. Dasgupta, Anal. Chem., 67 (1995) 71-78.
224 P.K. Simon and P.K. Dasgupta, Environ. Sci. Technol., 29 (1995)1534-1541.
225 A. Khlystov, G.P. Wyers and J. Slanina, Atmos. Environ., 29 (1995) 2229-2234.
226 J. Slanina, H.M. ten Brink, R.P. Otjes, A. Even, P. Jongejan, A. Khlystov, A.
Waijers-Ypellan, M. Hu and Y. Lu, Atmos. Environ., 35 (2001) 2319-2330.
227 J. Ruzicka and E.H. Hansen, Flow Injection Analysis, 2nd edn., 1981.
228 R.M. Carlson, Anal. Chem., 50 (1978) 1528-1531; US Patent 4,209,299, June 24,
1980.
229 G.P. Wyers, R.P. Otjes and J. Slanina, Atmos. Environ., 27A (1993) 2085-2090.
230 J. Slanina and G.P. Wyers, Fres. J. Anal. Chem., 350 (1994) 467-473.
231 http://www.timberlineinstruments.com
232 C. Zellweger, M. Ammann, P. Hofer and U. Baltensperger, Atmos. Environ., 33
(1999) 1131-1140.
233 M. Loflund, A. Kasper-Giebl, W. Tscherwenka, M. Schmid, H. Giebl, R.
Hitzenberger, G. Reischl and H. Puxbaum, Atmos. Environ., 35 (2001) 2861-2869.
234 R.J. Weber, D. Orsini, Y. Daun, Y.N. Lee, P.J. Klotz and F. Brechtel, Aerosol Sci.
Technol., 35 (2001) 718-727.
235 K. Okuyama, Y. Kousaka and T. Motouchi, Aerosol Sci. Technol., 3 (1984) 353-366.
236 J.H. Seinfeld, Atmospheric Chemistry andPhysics ofAir Pollution. Wiley, New York,
1986, pp 195-246.
237 R. Atkinson, W.P.L. Carter, J.N. Pitts, Jr. and A.M. Winer, Atmos. Environ., 20
(1986) 408-410.
238 M. Ferm and A. Sj6din, Atmos. Environ., 19 (1985) 979-983.
239 K. Sexton, J.D. Spengler and R.D. Treitman, Atmos. Environ., 18 (1984) 1371-1383.
240 J.L. Repace, in: I.K. O'Neill, K.D. Brunnemann, B. Dodet and D. Hoffmann (Eds.),
Environmental Carcinogens. Passive Smoking. IARC Scientific publications No. 81.
International Agency for Research on Cancer, Lyon, France, 1987, Vol. 9, pp.
141-162.
241 V.R. Highsmith, R.J. Hardy, D.L. Costa and M.S. Germani, Environ. Sci. Technol.,
26 (1992) 673-680.
242 W.F. Rogge, L.M. Hildemann, M.A. Mazurek and G.R. Cass, Environ. Sci. Technol.,
25 (1991) 1112-1125.
243 J. Barnes, P. Holland and P. Mihlmester, Characterization of population and usage
of unvented kerosene space heaters. EPA-600/7-90-004. US EPA, Office of Research
and Development, Washington, DC, 1990, pp. 1-37
244 B.P. Leaderer and P.M. Boone, Environ. Sci. Technol., 24 (1990) 908-912.
245 G.W. Traynor, M.G. Apte, H.A. Sokol, J.C. Chuang, W.G. Tucker and J.L. Mumford,
Environ. Sci. Technol., 24 (1990) 1265-1270.
246 J.L. Mumford, R.W. Williams, D.B. Walsh, R.M. Burton, D.J. Svendsgaard, J.C.
Chuang, V.S. Houk and J. Lewtas, Environ. Sci. Technol., 25 (1991) 1732-1738.
247 J.D. Spengler, M. Brauer, J.M. Samet and W.E. Lambert, Environ. Sci. Technol., 27
(1993) 841-845.
248 J. Notholt, J. Hjorth and F. Raes, Atmos. Environ., 26A (1992) 211-217.
249 W. Junkermann and T. Ibusuki, Atmos. Environ., 26A (1992) 3099-3103.
250 L.R. Martin, D.E. Damschen and H.S. Judeikis, Atmos. Environ., 15 (1981)
1615-1621.
251 S.E. Schwartz, in: J.G. Calvert (Ed.), SO , , NO and NO 2 Oxidation Mechanisms:
Atmospheric Considerations.Butterworth Publishers, Boston, 1984, pp. 173-208.

217
252 Z. Vecera and P.K. Dasgupta, Int. J. Environ. Anal. Chem., 4 (1994) 311-316.
253 B. Lighthart and A.J. Mohr (Eds.), Atmospheric Microbial Aerosols. Chapman &
Hall, New York, 1994.
254 Indoor Pollution News, July 25 (1994) 199.
255 Federal Register, 29 CFR parts 1910, 1915, 1926 and 1928, April 5, 1994; U.S.
General Accounting Office: School Facilities. Condition of America's Schools,
GAO/HEHS-95-61, February, 1995.
256 R. Kostiainen, Atmos. Environ., 29 (1995) 693-702.
257 H.A. Burge, Bioaerosols. Lewis, Ann Arbor, 1995, pp. 1-23.
258 R.A. Wood, A.N. Laheri and P.A. Eggleston, Clin. Exp. Allergy, 23 (1993) 733-739.
259 S.K. Poruthoor, P.K. Dasgupta and Z. Genfa, Environ. Sci. Technol., 32 (1998)
1147-1152.
260 S. Liu and P.K. Dasgupta, Microchem. J., 62 (1999) 50-57.
261 B.M. Loffler and H. Kunze, Anal. Biochem., 177 (1989) 100-102.
262 P.K. Dasgupta, Z. Genfa, S.K. Poruthoor, S. Caldwell, S. Dong and S.-Y. Liu. Anal.
Chem., 70 (1998) 4661-4669.
263 E.L. Avol, W.S. Linn, D.A. Shamoo, K.R. Anderson, R.C. Peng and J.D. Hackney, Am.
Rev. Respir. Dis., 142 (1990) 343-348.
264 K.W. Clark, K.R. Anderson, W.S. Linn and H. Gong, Jr., J. Air Waste Manage. Assoc.,
45 (1995) 923-926.
265 J.A. Last and K.E, Pinkerton, Toxicology, 116 (1997) 133-146.
266 H.Small, T.S. Stevens and W.C. Bauman, Anal. Chem., 47 (1975) 1801-1809.
267 K. Ito, C.C. Chasteen, H.-K. Chung, S.K. Poruthoor, Z. Genfa and P.K. Dasgupta,
Anal. Chem., 70 (1998) 2839-2847.
268 S. Gupta and P.K. Dasgupta, J. Chromatogr.Sci., 26 (1988) 34-38.
269 G.A. Tarver and P.K. Dasgupta, Atmos. Environ., 29 (1995) 1291-1298.
270 G. Andrew, Prospects for environmental monitoring in international safeguards.
IAEA-SM-333/213. International Nuclear safeguards 1994: Vision for the Future,
Vol. 1. IAEA, Vienna, 1994.
271 NA. Wogman, R.W. Perkins and G.R. Holdren, Environmental sampling and
analysis as safeguards tool. IAEA-SM-333/98. International Nuclear Safeguards
1994: Vision for the Future. Vol. 1. IAEA, Vienna, 1994.
272 S.K. Poruthoor and P.K. Dasgupta, Anal. Chim. Acta, 361 (1998) 151-159.
273 S. Vaidyanathan, J.J. Rowland, D.B. Kell and R. Goodacre, Anal. Chem., 73 (2001)
4134-4144.
274 P.A. Demirev, J.S. Lin, F.J. Pineda and C. Fenselau, Anal. Chem., 73 (2001) 4566-
4573.
275 F.Y. Meng, B.J. Cargile, L.M. Miller, A.J. Forbes, J.R. Johnson and N.L. Kelleher,
Nat. Biotechnol., 19 (2001) 952-957.
276 See, for example, Autonomous Pathogen Detection System (APDS), www.globalfia.
com

218
Chapter 7

Physicochemical properties of aqueous

and solid environmental matrices
Albrecht Paschke

7.1 INTRODUCTION

This chapter gives a short overview of the important physicochemical properties

of aqueous and solid environmental matrices. The term aqueous environmental
matrices encompasses precipitation (rain, snow, ice), surface water, groundwa-
ter, drinking water, wastewater, leachates, as well as sediment pore water and
soil solutions. Solid environmental matrices here means soils and sedimentary
matter, and partly also waste materials which are not of industrial origin (e.g.
sewage sludge, dredged material, and compost) but not minerals, ores, etc. The
chapter does not cover theoretical knowledge or actual measuring procedures
and equipment; instead examples are given on how selected physicochemical
properties of samples influence analytical results and their interpretation. The
reader requiring theoretical information is referred to textbooks on the funda-
mentals of aquatic chemistry [1-2], and soil and sediment chemistry [3-6]. For
actual information on standard measuring procedures and equipment, because
standardization is continuously in progress, the reader should visit the web sites
of International Organization for Standardization: www.iso.ch, Comit6
Europe6n de Normalisation: www.cenorm.be, or their national member organi-
zations, www.din.de, www.ansi.org, etc.).
Why it is important to know the physicochemical properties of aqueous and
solid environmental matrices? There are at least two main reasons. First, sam-
ple preparation is intended to transfer/transform the analytes into a measurable
form. During this process, it is usually inevitable that the interactions (binding
forms, state, etc.) of elements and compounds with their concrete chemical envi-
ronment are changed. These interactions are determined by the physical and
chemical properties of both analytes and matrices and influence the applicability
of different sample preparation techniques and analytical methods as well as
their efficiency and reproducibility. Hence, the characterization of the initial
physicochemical state of the sample is a pre-condition of all further sample
preparation steps. Second, the physicochemical properties of the environmental

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 219
compartments show large variations in time and space. Therefore "exact
values", or rather mean/median values, of the concentration of elements and
compounds, which serve as key parameters in exposure analysis and risk assess-
ment, are meaningful and usable even through determining and considering the
variations of critical matrix properties.
Often the available sample volume restricts the possible measurements of
physicochemical sample properties. In such cases one should select, based on
literature knowledge, the most important ones (sometimes called the master
variables of the system) which at least must be determined. Moreover, some of
the physicochemical properties must be measured immediately during field
sampling because they can change quickly during transportation and/or deter-
mine (partly) the selection of sample storage, preparation and conservation
method.
The important physicochemical properties of environmental matrices under
consideration are summarized in Table 7.1 together with references on their
determination.

TABLE 7.1
Important physicochemical properties of aqueous and solid environmental matrices with
reference to worldwide (ISO), European (EN), or German (DIN) standardized determination
methods. The properties in bold type should be measured during field sampling
Aqueous matrices Solid matricesa
Temperature DIN 38404-C4 Temperature
Density DIN 38404-C9 / Water content ISO 11465: 1993 /
ISO 15212-1: 1988 EN 12880: 2000
Colouring / ISO 7887: 1994 Density ISO 11508:1998/
UV-Vis absorption ISO 11272: 1998
Turbidity ISO 7027:1999 Cation exchange capacity ISO 11260: 1994/
(CEC)b ISO 13536: 1995
pH value ISO 10523:1994 pH valueb ISO 10390: 1994
EN 12178: 1998
Redox potential DIN 38404-C6 Redox potential ISO 11271: 200?c
Conductivity ISO 7888:1985 Conductivityb ISO 11265:1994
Surface tension ISO 4311: 1979 / Particle size distribution ISO 11277: 1 9 98 d
ISO 9101: 1987 (and soil types)
Suspended solids ISO 11923: 1997 Porosity and surface area ISO 9277:1995
Total / Dissolved ISO 8245: 1999 Total Organic Carbon (TOC) ISO 10694: 1995 /
Organic Carbon ISO 14235: 1998
(TOC/DOC)
aAdditional recommendation: ISO 11464: 1994 (soil quality-pretreatment of samples for
physico-chemical analyses).
bThis parameter is measurable only when the solid material is suspended in water.
CFinal draft (2002-03-27) in approval stage.
dSee also the other ISO standards on particle size analysis in ICS field 19.120 (www.iso.ch).

220
7.2 PHYSICOCHEMICAL PROPERTIES OF AQUEOUS MATRICES

7.2.1 Temperature [7-9]

The solubility and stability of analytes in aqueous media as well as other

physicochemical properties (e.g. density, surface tension, pH value, redox poten-
tial) are influenced by the predominant temperature in the sample. Larger tem-
perature fluctuations can also affect the sample integrity. These facts must be
taken into account when choosing special sample storage containers and tech-
niques. Measurements in a temperature range from -5 to 50°C can traditionally
be made with a calibrated mercury thermometer (usually graduated in 0.1 or 0.5
K steps). Nowadays, electronic pocket thermometers with LCD are used. These
devices, mostly equipped with a Pt-resistance sensor, have the advantages of
robustness and easy readability. In all cases it is necessary to give the thermome-
ter/sensor enough time to thermally equilibrate with the sample before reading
the temperature.
To give a simple example of the temperature influence on analyte concentra-
tion, we consider a gas-saturated water sample which is taken at 10°C and
transported to the laboratory. If the sample is not held at this temperature but
warmed up to 25°C, the oxygen concentration decreases from 11.25 to 8.25 mg/l
[7], i.e. by 27%. Other dissolved gases (CO2, NH3, H2S) show similar behaviour.
Also, organic substances can escape from the sample if there is enough head-
space left. An estimation using recently published temperature-dependent
air-water partition coefficients [10] shows that for a 200-ml (groundwater)
sample in a closed 250-ml bottle a rise from 15 to 30°C increases volatilization
loss, e.g., for methyl tert-butyl ether from 1.0 to 1.7%, for benzene from 3.8 to
6.3%, and for tetrachloroethylene from 5.7 to 10%.

7.2.2 Density [7-9]

Although density measurements are comparatively rare in water analysis, they

provide initial information about the content of dissolved and suspended matter,
especially in more mineralized waters. The density of natural waters can vary
considerably from that of pure water depending on the concentration of dis-
solved salts etc. The density should be measured with sufficient accuracy (to 3-4
decimal places). The use of pycnometers is too time-consuming for routine exam-
inations. To get a rough estimate, one can weigh a definite volume taken with a
pipette. More convenient are precision aerometers or hydrostatic balances. A
good alternative for higher sample through-put is an oscillating-tube density
meter (cf. its application in determining the water salinity, Ref. [9]). The density
measurements have to be performed in a temperature-controlled environment
and the exact temperature of determination must be specified or the result has
to be recalculated for 20 or 25°C.

221
7.2.3 Colouring [7-9]

Pure water in thin layers is practically colourless. Dissolved impurities can

change the water colour by absorbing light at special wavelengths in the visible
spectrum. Thus the colouring of a water sample provides first information on
relevant natural impurities; these are mainly iron or manganese compounds,
dissolved humic material and plant residues. Undissolved substances (particu-
late matter) must be removed before determination of colouring by sedimenta-
tion, centrifugation or filtration because it is necessary to differentiate colouring
from turbidity. The water sample can be classified by visual comparison for
example as "colourless", "slightly coloured" or "strongly coloured". Additionally
a colour shade is specified, e.g. "yellowish", "yellowish-brown", "brownish", etc.
More objective criteria provide the spectrophotometry. Methods are established
to measure the spectral absorption coefficient at fixed emission lines of Hg. The
UV absorption at 254 nm is an unspecific overall parameter for the presence of
dissolved high molecular polar organics (fulvic and humic acids) in the sample
because the absorption interference by nitrate (< 100 mg/l) can be neglected at
this wavelength. The 436 nm Hg line can also be utilized for comparative mea-
surements. It is in the complementary range of the usual yellowish-brown shade
of natural-coloured waters.

7.2.4 Turbidity [7-9]

Suspended matter or colloidal dissolved inorganic and/or organic substances

cause the turbidity of water by scattering (and to a minor extent absorbing) the
light on its way through the sample. Apart from silica particles (clay), and
ferric/aluminium hydroxides, organic macromolecules (humus) and microorgan-
isms (bacteria, algae) also cause such effects. The water sample can be classified
by simple visual testing as "clear", "opalescent", weakly turbid", "strongly
turbid", and "opaque". There exists also an instrumental method, nephelo-
metry, which is based on a comparison of the intensity of light scattered by the
sample under defined conditions with the intensity of light scattered by a
standard reference suspension (formazin polymer) under the same conditions.
With more advanced methods [11,12], e.g. the different variants of field-flow
fractionation (FFF), it is possible to further characterize the size distribution of
stable suspensions or colloidal solutions.

7.2.5 pH Value [7-91

The pH, defined as the negative decadic logarithm of the proton (hydronium ion)
activity, is one of the most important parameters in water chemistry. Together
with redox potential it strongly influences the stability, reactivity and mobility
of elements, inorganic and polar organic compounds in aquatic compartments of
the environment and of biological systems. Pure water saturated with carbonic
acid is not pH-neutral but has a pH value of 5.5-5.8. In natural waters the pH

222
 I

Fig. 7.1. Selected trace metal concentration in leachates of a topsoil after 24 h shaking at
different adjusted pH (expect pH = 6.8 which is the unaffected value; see text and Ref. [13] for
more details).

spans a range between 9 and 4. Even much lower pH values have been measured
with samples of acid rain and with soil solutions. But often the carbon dioxide/
hydrogen carbonate buffer system determines the pH of water and thus it varies
only between 6.8 and 7.5. Leachates can have very much lower or higher pH
compared to natural waters. A first pH test (e.g. during sampling in the field) is
possible using so-called universal indicators (as paper strips or slabs). Accurate
pH values require an electrometrical determination of the potential difference
between a glass electrode and a reference electrode (e.g. a saturated calomel
electrode) by the use of a commercial pH meter.
The role of pH in aqueous samples is illustrated in the following examples.
The first is taken from Ref. [13]. Subsamples of a topsoil (Greppin, Germany,
0-10 cm) were shaken with water for 24 h in a batch apparatus with a solid: liq-
uid ratio of 1:10. The pH value of the leachates were adjusted by continuous
titration with dilute nitric acid in order to simulate a successive deposition of
acid rain. Figure 7.1 shows how the pH of a leachate controls the mobilization of
trace metals from this soil material and thus the concentration levels to
quantify. With the exception of Fe, the metals show, in most cases, a drastic
increase in solubility at lower pH values. But iron and aluminium concentration
in leachates also increase at higher pH values. These latter effects further
increase in both directions of the pH scale (outside the shown range) and are
responsible for the mobility of Fe and Al in soil systems [4,6,13].
The second example, according to Ref. [14], demonstrates the pH influence
on the partitioning of three model compounds for the 4-quinolone group of
antimicrobials between SPME fibre coating and water sample. Whereas the

223
TABLE 7.2
pH Dependence of the apparent distribution coefficient for three 4-quinolones between the
SPME fibre (type CW/TPR from Supelco, Bellefonte, PA, USA) and buffer solution (calculated
from the regression models given in Ref. [14])
Substance pH
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0
Flumequine 49 49/ 49, 48 47 45 38 26 14 8 5 4 4
Oxolinic acid 22 22 22 22 22 21 19 16 10 4 2 1 0.2
Sarafloxacin 15 27 49 75 90 94 88 71 45 25 15 12 10

apparent distribution coefficients of flumequine and oxolinic acid decrease only

slightly from pH = 3 to pH = 6 and then drop to very low values at pH = 8, the
coefficient of sarafloxacin passes a pronounced maximum between these pH
values. Table 7.2 gives the pH-dependent distribution coefficients calculated
using the model equations which were fitted to the experimental data. These
findings are relevant not only for explaining the environmental effects of these
substances but also for their analytical determination because a proper adjust-
ment of sample pH can yield, especially for sarafloxacin, a much better
extraction efficiency and thus a lower limit of quantification.

7.2.6 Redox potential [1,7-9]

Reduction and oxidation (red-ox) reactions mediate the behaviour of many

elements and chemical constituents in most of the aquatic media in the environ-
ment (and in living systems) and can be used for their characterization together
with the pH value and the concentration of dissolved chemical species. The redox
potential (Eh), defined by the Nernst equation, represents all corresponding
redox pairs contained in the solution when chemical equilibrium is established.
Aquatic media with a positive Eh values (in mV) are said to be in aerobic (or oxic)
state, such with Eh below -200 mV under strict anaerobic (anoxic) conditions. In
contrast to pH buffering, it is in general not so simple to hold the Eh of a water
sample constant by adding reduction or oxidation agents without seriously
affecting the sample integrity. The Eh value is measured as the difference in
potential between the indicator electrode (Pt electrode) and the reference elec-
trode with a millivoltmeter (or a pH meter with mV-display). A saturated
calomel electrode is mostly used as reference instead of the normal hydrogen
electrode and this additional potential difference has to be considered in report-
ing the final Eh value. Moreover, it is necessary to determine temperature and
pH in the water sample at the same time to account for the considerable
dependence of Eh on these parameters. Because oxygen transfer from ambient
air into the sample can distort the original Eh value, measurements should be
carried out under continuous flow in a closed vessel, especially in nearly oxygen-

224
free samples (groundwater). Many other instrumental details, for example
impurified electrode surface (i.e. coated with PtO/PtO 2 or PtS) and very small
exchange currents, as well as objective facts, such as interfering reactions at the
electrode surface or the presence of inert or multiple redox couples in the
sample, can limit the interpretation of measured Eh values. Therefore, meaning-
ful determination and interpretation ofEh in natural water is a challenging task.
For more information the interested reader should consult special reviews on
redox potential measurements in the laboratory and in the field [1, p. 495] or [9,
pp. 2-60].
An interesting example of the combined effect of pH and redox potential on
the mobilization of the heavy metals Pb, Cd, and Zn from a contaminated soil can
be found in Ref. [15]. Stirring experiments with soil suspensions were conducted
under three different pH values and three redox potentials (325, 0, -100 mV).
The results show that the metals were sparingly soluble under alkaline condi-
tions (pH = 8.0). Metal solubilities were higher then under slightly acidic condi-
tions (pH = 5), and increased drastically when pH was kept at 3.3. When
solubilities were compared under the same pH values, it was observed that they
increased as redox potential decreased. The authors summarized that acidic and
reducing conditions were most favourable for the mobilization of considered
metals, and that the effect of pH was more significant than that of the redox
potential. However, due to the limited number of tests performed and, above all,
when considering the wide variety of possible binding forms and release/fixation
processes with heavy metals in soil and sedimentary matter, this general conclu-
sion must be looked at with reservation. Investigations by others [16] with sedi-
ment suspensions also show the dominating influence of pH, but at identical pH
values the mobilized portions of Cu and Pb from oxic sediment are ten-fold
higher compared to anoxic conditions. For Cd and Zn two-fold higher concentra-
tions are found in oxic states. Altogether, these converse results show that both
factors, pH and redox potential, can strongly affect the mobilization of heavy
metals.

7.2.7 Conductivity [7-9]

The ability of an aqueous solution to conduct a current-its electrical conductiv-

ity (reported as mS cm -1 or gS cm-l)-depends on the presence of ions: on their
total concentration, mobility, valence and on the temperature. In contrast to
many organic molecules, most inorganic compounds dissociate in aqueous solu-
tion and therefore the conductivity can be regarded as a non-specific measure for
the concentration of the latter in water. The self-dissociation of water causes a
conductivity of 0.06 bkS cm - ' only (at 25°C). The main cations and anions which
cause additional conductivity in water are Na + , K+ , Mg2 +, Ca2 + and HCO,-, CO3 -,
Cl-, SO42-, NO,-. Fresh bidistilled water has a value of around 0.5 uJS cm - l.
Natural waters can show very different conductivity values, depending mainly
on their origin. Spring waters from gneiss, granite or red sandstone have

225
conductivity values of only between 10 and 100 AS cm -1 , but springs from
limestone show values of 1000 AS cm-l and above. Compared with that, sea water
has a fifty times higher conductivity. In soil solutions the conductivity can range
from < 1 mS cm -1 to > 15 mS cm 1'. The conductivity meter is a resistance meter
with bridge fed with high-frequency alternating current which has to be applied
to prevent the polarization of electrodes and additional resistances thus arising.
The measuring cell contains electrodes, platinized or of stainless steel. It is
necessary to read off the temperature during the measurement, i.e. after reach-
ing thermal equilibrium in the measuring vessel. The measured conductivity
value has to be converted by a tabulated correction factor to the reference
temperature of 25°C.

7.2.8 Surface tension [12,17]

The intermolecular attractive forces in liquid states cause the macroscopic prop-
erty of surface tension which has an extremely high value in pure water: 72.8
mN/m at 20°C [18], exceeded only by mercury. This water property is very sensi-
tive to impurities. Not only the so-called surface-active agents (detergents), but
most added organic substances reduce the surface tension of water considerably.
(Only dissolution of inorganic salts, mineral acids, and some sugars increase the
surface tension to higher values than that of pure water.) For example, addition
of only 0.1% of n-propanol to water lowers the surface tension to 67 mN/m [18];
the same reduction results if only 30 ng of di-2-ethyl-hexyl phthalate, a ubiqui-
tous plasticizer, are present per litre water [19]. The existing techniques for
measuring surface tension have self-explanatory names such as capillary height,
drop weight, ring detachment method, etc. In all such methods the hydrostatic
pressure drop against the surface and radii of curvature are implicated in differ-
ent ways. Laboratory equipment is available commercially at different levels of
accuracy. Practical problems with reproducibility can arise because the surface
tension depends on the kinetics of diffusion of surface-active impurities between
the bulk phase and the surface region. Surface tension measurements of real
water samples (fog, rain and wastewater) using the drop weight method have
been reported [20,21] and can serve as an overall pollution indicator. But mea-
surements of surface tension of water samples can also, at least in some cases,
replace expensive isolation and chemical analysis of detergents. Moreover, a rel-
atively small surface tension (< 65 Nm/m) indicates the possible presence of col-
loidal structures (micelles) in the water sample (precisely when the critical
micelle concentration is exceeded) and can cause the addition of a micro-
emulsion-breaking agent before continuing with sample preparation.

7.2.9 Total/dissolved organic carbon [7-9]

In addition to microscopic living matter (algae, etc.), a variety of organic com-

pounds is present in waters. These are the natural degradation products of plant

226
and animal tissue (e.g. polysaccharides and proteins), humic substances,
residues from coal and oil processing and fuel combustion (hydrocarbons),
detergents, pesticides, etc. One can subdivide this organically bound carbon
more or less arbitrarily by filtration with, for example, a membrane of 0.45 Am
pore diameter into a particulate and dissolved fraction. The latter fraction could
be further divided into a colloidal and a truly dissolved part. In many cases,
however, only the total of not-settleable organic material in a water sample is of
interest for sample preparation and analysis. In order to determine TOC/DOC,
the organic substances in the sample are completely oxidized by wet-chemical
means or UV rays, or burnt at high temperature. The amount of CO2 released
can then be measured. Commercial apparatus often use non-dispersive infrared
analysis devices. The measurement requires calibration using standard solu-
tions. Since water samples contain also inorganic carbon (CO2, HCO,-, CO3 2 -),
this must be removed by adding HC1 before TOC/DOC determination. Bidistilled
water has a TOC value < 0.1 mg/l (given as elementary C), deionized water due
to traces from ion-exchange resin already up to 1 mg/l. Untouched natural
waters and drinking water normally show a TOC between 1 and 2 mg/l; in
polluted rivers it is ten times higher. Wastewaters and leachates are often far
above that. The presence and quality of TOC/DOC in aqueous media are import-
ant parameters for the distribution and mobility of trace metal ions as well as
organic substances, both nonpolar and partly dissociating. As mentioned above
in connection with turbidity, it is a challenging task to clear up size distribution
and constitution of TOC/DOC determined summarily in a water sample.
Depending on the sorption behaviour of analytes, TOC/DOC acts as an
additional sink or source for them and modifies their available water concentra-
tion as well as their mass-transfer characteristics during extraction steps, etc.
Such processes can strongly influence analytical sample preparation and calibra-
tion and are often summarized under the term matrix effects. The following
examples illustrate this for both organic analytes and metal ions.
The influence of DOC of different origins (and of pH) on the direct SPME of
halogenated phenols is described in [22]. The authors have tested the natural
organic matter of a brown water lake, of a compost extract, of commercial
Aldrich humic acid, and of the protein bovine serum albumine. The distribution
coefficients between SPME fibre coating and sample were calculated according
to a first-order extraction kinetics. In general, the distribution equilibrium was
established faster in the presence of dissolved organic matter. An additional
transport process of DOC-sorbed analytes to the fibre was assumed (carrier
function of DOC molecules) and also a change of the sorption characteristics of
the SPME fibre itself (partial coating with DOC) could not be excluded. A similar
accelerating effect on the SPME kinetics was observed with very hydrophobic
polyaromatic hydrocarbons (PAHs) with four or more rings in the presence of a
Roth humic acid [23]. Interestingly, for the medium hydrophobic three-ring
PAHs the extraction kinetics is not significantly affected and for less hydropho-
bic compounds (naphthalene, biphenyl) even retarded. Leaving all doubtful

227
mechanistic interpretation aside, these examples lead us to the conclusion that
such a strong influence of DOC on this special preconcentration technique can
only be taken into account by applying matrix-related calibration.
The investigation of particle distribution of the DOC (and of pH) and its
effect on measured Cu concentrations is reported in Ref. [24]. Soil solutions were
filtered through different pore size membrane filters (0.45, 0.22, and 0.1 /tm).
There was no difference for DOC concentrations in the three pore size differenti-
ated filtrates at low pH values where fulvic acid is the predominated dissolved
organic matter. At higher pH values, a reduction in membrane filter pore size
decreased the DOC concentration. This indicates an increasing presence of
humic acid with higher molecular weight. With Cu it was further shown that, by
using different sizes of filters, the evaluation of mobile pollutants could be quite
different. For example, the DOC concentration determined in 0.22 and 0.45 ,-m
filtrates differ by 35%, which resulted in 54% soluble Cu concentration differ-
ence due to the strong binding affinity of Cu to the higher molecular weight
organic matter. The effect of DOC on Cu partitioning was further investigated in
river water samples [25].
To end this section, a few more papers can be recommended which further
illustrate the role of selected physicochemical parameters in preparation and
analytical processing of aqueous samples, especially of landfill leachates [26], or
critically review existing knowledge on the characterization of aquatic colloids
and macromolecules [27] and on estimating dissolved organic carbon partition
coefficients for nonionic organic chemicals [28].

7.3 PHYSICOCHEMICAL PROPERTIES OF SOLID MATRICES

7.3.1 Temperature [29]

Solid environmental samples are heterogeneous in composition and structure

and contain considerable amounts of air and water. Temperature measurement
can be made with the same equipment as used for aqueous samples though it is
necessary to be more cautious to prevent breakage of the thermometer/sensor
and to wait longer than in liquid phase for reaching thermal equilibration before
recording temperature. There are also no large differences between aqueous and
solid matrices regarding the effects of temperature changes on sample integrity
and analyte inventory. It is, for example, well known from soil/sediment sample
handling that a rise in temperature can intensify microbiological activities
which in turn can lead to a degradation of organic compounds in the sample. The
increasing volatility of chemicals is another effect of higher temperature which
is taken advantage of, for example, in headspace analysis.

7.3.2 Water content [30,35]

Water is present in almost all solid matrices. It is necessary to know the water

228
content of a sample for weighing definite amounts of solid material required in
further sample processing (extraction, leaching, mixing) and for relating analyti-
cal results to dry weight. Some other analytical methods take advantage of the
fact that the presence of some percentage of water can enhance volatilization of
nonionic or weakly polar organic analytes (BTEX, pesticides, etc.) from solid
samples by displacing them from adsorption sites. Reproducible work also
requires here the determination of initial water content. Moreover, it is partly
necessary to remove water from the sample before subdivision, extraction, or
conservation of the sample. The major problem in water-content determination
is to define at what point the sample is "dry", especially for porous materials
(soils, sediments, etc.) where water can be present in different forms. Colloi-
dal-bound water and portions adsorbed at the mineral surface are more easily
removable than water of crystallization ("structural water") which only comes
off at elevated temperatures. Thermal dehydration curves can be very specific
for the different materials and so exact reporting of the temperature of determi-
nation is important. Additional critical parameters for reproducible and compa-
rable data are accurate temperature measurement and control of the laboratory
oven used. The traditional procedures for most purposes, where the highest
precision in water content determination is not required, involves weighing a
representative part of a moist sample, oven drying at 105°C, reweighing, and cal-
culating the mass of water lost as a percentage of the mass dried material. The
reader should follow the standard procedures closely because, in addition to the
problem of temperature control, drying duration and cooling the sample in a des-
iccator are also important steps. Usually, one continues drying to constant
weight and this can take up to three days depending on the amount of sample.
Special problems can arise if the sample contains solvents or oil/grease in more
then trace quantities. In that case, it is either possible to distil the sample with
toluene in a special apparatus or to use Karl Fischer titration for water-content
determination.

7.3.3 Density [31,36]

Density of solids is a key parameter for applying particle-separation techniques.

Due to the fractal dimensions of solid environmental matrices, the density of
these materials cannot be calculated simply by dividing the mass of a sample
specimen by its volume. Based on Archimedes' Principle, a calibrated pycno-
meter can be used for density determination of fine-grained solid material. For
coarse-grained material (gravel, stones) one can use the submersion method.
The accuracy of the result will be determined by several factors: the dryness and
air-freeness of specimen, the temperature control and the balance sensitivity.
However, we should consider that only an overall density is obtained by such a
method. The concrete density of specific particles can differ considerably from
that average. One should also not be confused by the term "particle density"
used in soil science which refers not to a single particle but to the soil particles

229
collectively and thus means the overall solid density in the sense described
above. Further on, soil scientists have defined a so-called "bulk density" which is
the ratio of the mass of oven-dried solid material to the bulk volume of the solids
plus the pore space at some specified water content, usually that at sampling.
Due to soil shrinking/swelling on desiccation/moisturisation, this "bulk density"
is a dynamic property. When measured, e.g. by the core method, it can be used to
estimate the total porosity of a soil under defined environmental conditions.

7.3.4 Cation exchange capacity [6,37]

Mainly due to their content of clay and amorphous minerals and of organic
matter, solid environmental matrices possess a negatively charged surface that
results from atomic substitution in the lattice of the minerals or from (pH-
dependent) deprotonation of functional groups. This negative charge is balanced
by positively charged cations and as a result the sorbed "exchange complex" is
formed. The cation exchange capacity (CEC) is an important property of the
considered solid materials because it controls, to a large extent, the retention of
inorganic and organic species. The CEC measuring principle is to saturate the
material with some "index" cation (Ba2 + ) that forces other cations, already
present on the charged surfaces (mainly Mg + , Ca2+, K+ , Na + , NH4+ , Al3+), into
solution, followed by quantification of these displaced cations. The measure-
ments can be complicated (erroneous) due to additional dissolution of soluble
salts (calcite and gypsum), specific adsorption of the small cations (K+ , NH4+ ) in
the interlayer position of particular clay minerals, and of the specific adsorption
of trivalent cations (Al3+, Fe 3+ ) at the material surface. The so-called "potential"
CEC is the value measured at the pH of 8.2 where complete deprotonation of the
carboxyl groups of humic matter is assumed. In contrast, the "effective" CEC is
measured at the native pH of the soil/sediment (solution). To give examples of
the occurring CEC values, we take the clay minerals kaolinitite and montmoril-
lonite as two extremes. The former shows a low CEC from 2-15 cmol/kg. For the
latter, one can measure CEC values from 80-150 cmol/kg. The differences are
caused by the different structures of these minerals, especially the accessibility
of interlayer spaces between silica tetrahedral sheets and aluminium octahedral
sheets. Additional large contributions to the CEC (up to 80%) of soils are
provided by the organic matter present. These contributions are variable (pH-
dependent). Knowing the CEC of the material is advantageous for sample
preparation to prevent (or take into account) possible losses of analytes or
standards during operation steps. One should consider that not only inorganic
species but also organic cations (n-alkyl ammonium ions, etc.) are affected by
such exchange processes. The maximum adsorbed amount of the herbicide
Paraquat (1,1'-dimethyl-4,4'-dipyridylium-dichloride), for example, was deter-
mined to 4.0 cmol/kg on kaolinite and to 78 cmol/kg on montmorillonite [41] and
hence reaching nearly the CEC value of these clay minerals.

230
7.3.5 pH Value [4,5,42]

The pH measurement of a soil or sediment sample is usually carried out in

suspension (solid:liquid = 1:2.5) as described already for aqueous matrices. For
muddy surface sediment samples (e.g. dredged material) and water-rich sludges
the determination is also possible after direct insertion of the glass electrode into
the sample. The measurements should be carried out preferentially in the field
because buffer capacity of the sample, and thus the effect of aeration, is usually
unknown prior to sampling and investigating the samples. The pH variation in
bottom sediments of natural water bodies is only in the range of 6-8. Larger
differences are measurable for soil materials. Acid rain can have a dramatic
effect if the acid neutralization capacity of a soil is low. Soil organic matter
(humus, compost) can produce considerable acidity in the surface horizon. When
mine-spoil soils are drained and pyrite (FeS2) oxidation occurs, extreme acidity is
produced which causes pH as low as 2. Alkaline soils contain mostly calcite
(CaCO,) or dolomite (CaMg(CO3 )2 ), either naturally or as result of liming, and
show pH values up to 8.5. For pH measurements of soils it is common to use
neutral salt solutions (0.1 M KCl or 0.01 M CaCl2 ) instead of water for suspend-
ing the solid material. This is intended to ensure a constant and hence
comparable ion activity in the soil solution.
The important influence of pH on the mobility of trace elements has already
been mentioned and exemplified in the previous section (Fig. 7.1). At low soil pH
values, it can be seen that Al, Fe, and Mn become more soluble and can cause
toxic effects. As pH increases, their solubility decreases and precipitation occurs.
But one should also consider that pH has diverse effects in such complex solid
matrices as soils or sediments. For example, apart from solubilization of metals,
pH also controls the dissolution of organic matter which can then act as complex-
ing agent and colloidal transporting medium for inorganic and organic pollut-
ants and can thus cause "unexpected" high concentrations of metals (e.g. Pb)
and organics (PAHs) in the solution phase (drainage water, pore water).

7.3.6 Redox potential [6,42,43]

Just as in the aqueous phase, the measurement of redox potentials in soils and
sediments is usually done with a Pt electrode in combination with a reference
electrode as described above. But, as already mentioned, gases present in large
quantities can make Eh measurements questionable, especially in aerated soils
(02) and sulphur-rich sediments (H2S), because the electrodes could react with
these gases and form coatings of oxides or sulphides on the electrode surface.
Moreover, a measured Eh value mostly represents mixed potentials due to the
mineralogical heterogeneity of a soil/sediment system compared to the dimen-
sion of the electrode. The degree and variability of soil moisture also influence Eh
as well as the predominant pH value (about -59 mV per pH unit). Thus, mea-
surements of redox potential are most reliable, i.e. unambiguous to interpret, for

231
flooded soils and recent sediments, if the layer structure is not disturbed too
much by inserting the electrodes and any "contamination" with air is avoided.
Furthermore, Eh should be recorded not before the signal becomes stabilized. In
soils and sediments this takes ten minutes and more; sometimes it is recom-
mended to wait at least 30 min before recording and to repeat the measurement
in periods of 5-10 min until potential drift stops. For oxidized soils redox poten-
tials of +400 to + 700 mV can be measured. Seasonally water-saturated soils can
change to a highly reduced status (below -250 mV). The overall redox status is
determined by a number of redox reactions mainly under participation of
iron(III) and manganese(III/IV) oxides/hydroxides, which are quite common in
soils and sediments as suspended particles and as coatings on clay mineral sur-
faces, and of Fe(II) minerals and a number of natural organic substances with
(carboxyl, phenolic, etc.) functional groups as reducers. Although according to
the above-mentioned limitations, Eh is no quantitative measure of redox status,
it provides important information on conditions that are favourable for
increased mobility, bioavailability, etc., of nutrients (e.g. nitrate-nitrite) and
trace elements in soils or sediments. Chromium and arsenic are well-known
examples of toxic trace elements which can exist in several oxidation states and
thus are strongly affected by redox potential changes. This has to be taken into
account when choosing sample preparation and analytical methods.
Masscheleyn et al. [44] describe a laboratory study undertaken to show the
influence of redox potential and pH on the arsenic speciation and solubility in
subsamples of a contaminated soil which were incubated in canisters under
flooded conditions for different periods of time (from 1 to 105 days). At higher
soil redox levels (500-200 mV) arsenic solubility was low and the major part of
arsenic in solution was present as As(V). An alkaline pH or the reduction of
As(V) to As(III) released substantial proportions of arsenic into solution. Under
moderately reduced soil conditions (100-0 mV), arsenic solubility was controlled
by the dissolution of iron oxyhydroxides. Arsenic coprecipitated before as As(V)
together with iron oxyhydroxides, was released upon their solubilization. Upon
reduction to -200 mV, the soluble arsenic content increased 13-fold as compared
to 500 mV. Interestingly, these authors also report that an equilibration period
of 4 h was necessary for the microelectrodes to obtain a constant redox reading.

7.3.7 Particle size distribution [11-12,17,32,38,42]

Particle size (and shape) controls sedimentation rate and packing density of
solid material. Usually, we distinguish only between three fractions: first the
coarse particles (2 m and above) which are observable by the naked eye; second
the fine particles (2-0.5 Am) which settle in a limited period of time; and third
the ultrafine particles (<0.5 Abm) which normally stay dispersed for a long time
and cause the turbidity of a suspension. But the particles contained in a given
solid sample can also be sorted into narrower size classes to obtain a distribution
function. Currently, there exist several classification schemes which set slightly

232
different but comparable limits for the particle classes. These classes are gravel
(8-1 cm), sand (<1 cm to 63 (or 50) Am), silt (<63 (or 50) Am to 2 ptm), and clay
(<2 m). Additionally, each class is subdivided in a fine, medium, and coarse
fraction. Particle-size distribution in itself is not a particularly interesting
function: rather it is the influence of size on the macroscopic properties of a
dispersed system that is important. In soil science, for instance, particle-size
analysis is used to evaluate the soil texture which is based on different combi-
nations of sand, silt, and clay and can be systematized by means of a so-called
textural triangle. The texture has a determining influence on soil properties
such as water content and retention, hydraulic conductivity, etc., which in turn
have great significance for soil fertility. Another aspect, more relevant to
analytical-chemical investigations, is that particle size determines the available
outer particle surface area which gives sorption capacity for inorganic coatings,
biofilms, trace elements, and organic compounds. A simple calculation of surface
area per unit mass of uniform cubes or spheres according to the equation
Area/Mass = 6 /(p x d),
where p is the density of particles and d either the edge of the cube or the
diameter of the sphere, illustrates the increase in specific area with decreasing
particle size [17]. Assuming a density of 2.00 g/m 3 , the specific area of a particle
with d of 1 cm, 1 mm, and 1 im is 3, 30, and 30,000 cm2 /g, respectively. A
knowledge of the particle-size distribution of a solid sample and the separation
and investigation of special size classes can bring more insight into distribution
and transport behaviour of contaminants in soils, sediments, and waste
materials. It is also well known from practice that working with (sub)samples
which have a definite size range gives better reproducibility in leaching tests and
also better comparability of results from different samples.
A number of particle-sizing methods have been developed; in general, they
do not cover the same size ranges, require different amounts of sample material
and different methods of sample pretreatment (removal of carbonates, iron
oxide, organic matter, etc.) and dispersion techniques (chemically by adding
phosphates and detergents or physically by stirring or ultrasonication). Not all
of these methods are useful for preparative purposes. They can be classified
according to the measuring principle used or to their applicability for special size
ranges. Sieving is by far the most common procedure; it can cover a wide range of
sizes (down to 5 Atm) and yields also larger, i.e. analytically useful, quantities of
particular size classes. Sieving can be "wet" or "dry" according to the dispersion
medium. However, particle sizing by sieving is affected by several variables such
as particle shape, cohesiveness of particles, brittleness of particles, shape of the
sieve openings, sieve loading, agitation, and duration of sieving. But most of
these possible pitfalls can be prevented by following the standard procedures
and the descriptions given by the manufacturers of sieving equipment.
Ackermann et al. [45] found a simple procedure to separate the < 60-/vm and the
<20-tim size fraction of sediments for heavy metal analysis which is in routine

233
use now (e.g. regulated by the environmental authority of Hamburg, Germany):
a sediment suspension (siftings of the greater sieve) is transferred to a plastic
sieve which is placed in a beaker in an ultrasonic bath together with some
volume of distilled water, followed by 5-15 min of ultrasonic treatment (see
original publication for further details).
Particle-size distribution may also be determined by measuring sediment-
ation velocity of particle settling under gravitational forces. The general method
is to sample the mass of particles in an initially homogeneous dispersion at
different heights at different times (e.g with a special pipette). It is assumed that
settling of particles obeys Stokes' law, which relates the size of free falling
spherical particles (i.e. at high dilution) to the sedimentation velocity in a
laminar flow. Usually some of these assumptions are not exactly fulfilled and
thus the calculated size may differ from the real particle size. Practically,
sedimentation methods are limited to size ranges larger than 5 m and to
dispersions settling within 30 min. These limitations can be overcome by
applying an elutriator, a cyclone, or a centrifuge and thus the measuring range
can be extended to submicron particles. During centrifuging the particles do not
sediment at a constant velocity, but they accelerate as they move farther and
farther from the axis of rotation. The time required for sampling (i.e. the time for
particle settling up to a definite cut-off size) can be adjusted by varying the
angular velocity (calculable for the given geometry of rotor, viscosity of
dispersion fluid, and density difference between particle and fluid). Optical
microscopy is a particle-sizing method with broad applicability and the only
technique that allows direct observation and measurement of fine particles. A
serious drawback of microscopy is that the specimen examined includes only a
minute fraction of the whole sample and therefore representative sample
splitting is a critical point. Additionally, dispersions cannot be examined directly
but the subsample has to be dried and mounted prior to analysis. Analogous
problems arise with the more advanced microscopic techniques-transmission
and scanning electron microscopy. Radiation scattering of particles is the basic
principle of another common type of method and includes static and dynamic
(laser) light scattering, small-angle neutron scattering or X-ray scattering, and
others. The useful range of these methods is limited by the size of the particles
relative to the wavelength of the radiation. One of the chromatographic tech-
niques useful for particle-sizing is the so-called hydrodynamic chromatography.
Particles in the 30-1500 nm size range can be separated by using packed or
capillary columns. Size-exclusion chromatography uses a column packed with
porous particles providing an additional separation effect and is applicable to
particles smaller than 400 nm. Field-flow fractionation employs an external field
(a centrifugal force, a cross flow, an electrical potential, or a temperature
difference) to induce migration of particles in a direction perpendicular to the
laminar carrier liquid flow in a narrow separation channel. The particle size
amenable to separation may range from 10 nm to 100 um, depending on the field
strength and the method used. More details of these chromatographic methods

234
 8
A

Zn

Fig. 7.2. Metal concentration in the size fractions <20 Am and 63-20g/m of a brook sediment (BS)
and a lake sediment (LS) separated by dry sieving (DS) and wet sieving (WS), respectively (see
text for more details).

as well as other separation techniques not described here (e.g. electric zone
sensing) can be found in the literature. However, it should be mentioned that
sequential filtration of a dispersed sediment sample seems to be, against widely
held opinion, unsuitable for particle-sizing and separating of definite fractions,
irrespective of the filter membrane types used [46].
An example should demonstrate the uneven heavy metal burden of different
particle size fractions of two contaminated sediment samples [47]. Bottom
sediment samples were taken from a stream (Boese Sieben) and a lake (Suesser
See) in the Mansfeld region in Saxony-Anhalt, Germany, where 800 years of
black shale mining and copper smelting generated large quantities of mining
debris and processing waste (slag heaps, tailings, etc.). Wet sieving (with water
jet siever WS1 from Retsch, Haan, Germany) and dry sieving (with sonic sifter
Analysette 28 from Fritsch, Idar-Oberstein, Germany) were used to obtain two
different size classes of fine particles in sufficient quantity for extraction with
aqua regia and following element determination using ICP-AES. It can be seen
from Fig. 7.2 that the < 2 0 -,m particle fractions contain twice as much Zn, Pb,
and Cu than the 63-20-utm fractions. Both sieving methods (including the
different prior conservation/preparation techniques) yield comparable results.
Dependencies of pollutant concentration on particle size have also been
identified with hydrophobic chemicals (e.g. PAHs, PCBs, chlorinated hydrocar-
bons) in natural sediments [48], whereby it is not necessarily the finest fraction
(clay) where the major part of these compounds are concentrated [49,50]. The
organic carbon content of soils/sediments seems to be the primary factor control-
ling the sorption of hydrophobic organic substances (see Section 7.3.9).

235
7.3.8 Surface area and porosity [6,12,17,33,34,391

On solid surfaces atoms/molecules can adsorb and subsequently react among

themselves and under participation of the matrix-forming chemical structures.
Thus, the available (reactive) surface area is a determining, often the control-
ling, factor in many chemical reactions and partly also in biological processes.
The specific surface (surface area per unit mass) of nonporous solid materials is
an indirect expression of the particle size distribution as already shown above
with a calculation for particles taken as uniform cubes or spheres of different
size (cf. Section 7.3.7). But apart from their outer surface area, natural solid
matrices often have a considerable inner surface area. This internal surface can
be subdivided into a part resulting from a network of pores extending into the
interior of the material and another part caused by interlayer regions between
the mineral layer sheets which can sorb ions and small molecules (water).
Charcoal is an example of a highly porous material and has a specific surface of
approximately 800 m2 /g. For the latter ease, two extremes are kaolinite with a
specific surface of 7-30 m2 /g and montmorillonite with values between 600 and
800 m2 /g. It is no accident that these minerals also mark some lower and upper
limit, respectively, of cation exchange capacity (cf. Section 7.3.4).
The surface area can be measured by gas adsorption methods based on the
BET theory (named after Brunauer, Emmett and Teller, who published in 1938
a theoretical model for multilayer adsorption of gases) using nitrogen, noble
gases, ethane, water, or ammonia. Although widely applied, the BET method has
the disadvantage, apart from the vacuum apparatus required, that the result
depends on the nature of adsorbate used as well as on that of the adsorbent
(sample substrate). Nonpolar nitrogen molecules, for instance, do not penetrate
the interlayer surfaces of montmorillonite, whereas polar molecules such as
water or ammonia do, giving a much greater apparent surface area for this
material. Therefore the BET method is classed sometimes as belonging to the
techniques for external surface area determination (including that of macro- and
mesopores). By adsorption from solution with dyes, fatty acids, or radioactive-
labelled substances one can determine indeed the external surface area only.
The major experimental consideration is to be able to adjust the concentrations
of adsorbate and particles in aqueous suspension so that detection is sensitive
enough before and after the batch experiment. A third surface-area measuring
principle uses polar organic liquids such as ethylene glycol monoethyl ether as
the adsorbate. This method is also widely accepted due to its simplicity and
because it is believed that in addition to the external surface the applied polar
molecules cover all interlayers, thus providing a measure of the total surface
area. An uncertainty of all these methods is the inherent assumption on the
molecular surface area occupied by an adsorbate which is difficult to prove.
A number of methods are described for the determination of both the total
porosity and the pore size distribution of solid materials. The total soil porosity
may be calculated if the particle density and the dry bulk density are known (see

236
Section 7.3.3). It gives an estimate of the fraction of total sample volume
occupied by pores. More advanced methods (porosimetry) examine porous
structures often indirectly by filling the pores with a gas (helium) or liquid
(mercury) to measure the pore volume. As with the surface area measurements,
assumptions have to be made about the accessibility of pores of different size by
the measuring fluid and thus the porosity values depend on the method of deter-
mination. This limited comparability applies in particular to environmental
solid materials due to their complex nature and has to be taken into account in
evaluating data from literature. The presence of pores in a solid material can
considerably increase the diffusion path length of atoms/molecules, resulting in
a slowing down of adsorption/desorption kinetics. Hence, porosity can largely
influence sample preparation techniques (possible extraction yields, limits of
quantification), particularly when short-term methods are performed.

7.3.9 Total organic carbon [6,40]

Plant debris and degradation products of animals form the organic matter of
soils and sediments. The contents range from 0.5 to 5% (on a weight basis) in the
surface horizon of mineral soils and can be much higher in sediments, peat mate-
rial, and compost. The main elementary constituents of this organic matter are
C (>50%), O (30-40%), H (3-5%), and N (4%). One can distinguish chemically
between persistent humic matter and easily degradable nonhumic substances
(carbohydrates, proteins, amino acids, fats, waxes, etc.). Humic matter can be
subdivided into fulvic acids, humic acids and humin, based on their solubility in
acid or base. This highly condensed organic matter has a large specific surface
(800-900 m2/g) and CEC (150-300 cmol/kg). Due to these properties, it is an
important adsorbent of plant nutrients, heavy metals, and organic compounds
such as pesticides and PAHs. The linear sorption coefficients of a nonpolar
chemical with different soils/sediments were found directly to be proportional to
the organic carbon content. Based on this fact, an organic-carbon normalized
partition coefficient, K,,, can be defined [48]. Numerous studies have shown
that Koc values for a nonpolar organic compound of limited water solubility
(<10-3 M) are relatively constant and, within the experimental errors, independ-
ent of the particular soil or sediment investigated. Similarly to aqueous-sample
processing, the methods for determining TOC are based either on wet oxidation
of a specimen in an acid dichromate solution, followed by determination of the
remaining dichromate or collection and quantification of evolved CO2 , or alter-
natively on dry oxidation of a sample in a furnace followed by CO 2 determina-
tion. It is necessary to differentiate organic carbon from the inorganic part
either by sample pretreatment with a mineral acid to decompose carbonates or
by employing different furnace temperatures. Commercial instruments are
available for the combustion method, most of them work with temperature
ramping and determine the evolved CO2 by infrared absorption measurement.
Knowledge of TOC of material under investigation is important not only for a

237
reasonable comparison of results (OC-normalization) but also for sample prepa-
ration prior to the analysis of hydrophobic organic substances to account for
incomplete and varying recovery of standards (formation of bound residues).
Another aspect is the potential dissolution of organic carbon constituents which
can lead to undesirable effects during sample processing and analysis (masking
of analytes, cf. Section 7.2.9).
Finally, some additional articles can be recommended here which summarize
existing knowledge on solid-solution partitioning of metals in contaminated
soils, especially the dependence on pH, total metal burden, and organic matter
[51], or review the factors affecting sorption of organic compounds in natural
sorbent/water systems [52,531.

REFERENCES
1 W. Stumm and J.J. Morgan, Aquatic Chemistry. 3rd edn. Wiley, New York, 1996.
2 F.J. Millero, The Physical Chemistry of Natural Waters. Wiley, New York, 2001.
3 K.B. Krauskopf and D.K. Bird, Introductionto Geochemistry, 3rd edn. McGraw-Hill,
New York, 1995.
4 P. Schachtschabel, H.P. Blume, G. Briimmer, K.H. Hartge and U. Schwertmann,
Scheffer/Schachtschabel-Lehrbuch der Bodenkunde, 13th edn. F. Enke Verlag,
Stuttgart, 1992.
5 D.L. Rowell, Soil Science: Methods and Applications. Addison-Wesley Longman,
London, 1994.
6 D.L. Sparks, Environmetal Soil Chemistry. Academic Press, San Diego, CA, 1995.
7 L. Hiitter, Wasser und Wasseruntersuchung, 5. Aufl., Salle-Verlag, Frankfurt a.M.,
Germany &Verlag Sauerldnder, Aarau, Switzerland, 1992.
8 W. Fresenius, K.E. Quentin and W. Schneider (eds.) Water Analysis-A Practical
Guide to Physicochemical, Chemical and Microbiological Water Examination and
Quality Assurance. Springer-Verlag, Berlin, 1988.
9 A.E. Greenberg, L.S. Clesceri and A.D. Eaton (eds.), Standard Methods for the
Examination of Water and Wastewater, 18th edn. American Public Health
Association/American Water Works Association/Water Environment Federation,
Washington, DC, 1992.
10 B.G. Bierwagen and A.A. Keller, Environ. Toxicol. Chem., 20 (2001) 1625-1629.
11 I.D. Wilson, E.A. Adlard, M. Cooke and C.F. Poole (eds.), Encyclopedia of Separation
Science. Academic Press, San Diego, CA, 2000.
12 E. Kissa, Dispersions-Characterization,Testing, and Mesurement. Marcel Dekker,
New York, 1999.
13 A. Kruiger, A. Paschke, B. Schneider, E. Biittner and H. Neumeister, Petermanns
Geograph. Mitteil., 142 (1998) 375-392.
14 H.C. Holten Liitzh0ft, W.H.J. Vaes, A.P. Freidig, B. Halling-S0rensen and J.L.M.
Hermens, Environ. Sci. Technol., 34 (2000) 4989-4994.
15 M.C. Chuan, G.Y. Shu and J.C. Liu, Water Air Soil Pollut., 90 (1996) 543-556.
16 W. Calmano, J. Hong and U. F6rstner, Vom Wasser, 78 (1992) 245-257.
17 S. Ross and I.D. Morrison, Colloidal Systems and Interfaces. Wiley, New York, 1988.
18 D.R. Lide (ed.), Handbook of Chemistry and Physics, 73rd edn. CRC Press, Boca
Raton, FL, 1992.
19. M. Thomsen, L. Carlsen and S. Hvidt, Environ. Toxicol. Chem., 20 (2001) 127-132.
20 P.D. Capel, R. Gunde, F. Zfircher and W. Giger, Environ. Sci. Technol., 24 (1990)
722-727.

238
21 R. Gunde, M. Dawes, S. Hartland and M. Koch, Environ. Sci. Technol., 26 (1992)
1036-1040.
22 G. Ohlenbusch, M.U. Kumke and F.H. Frimmel, Sci. Total Environ., 253 (2000)
63-74.
23 F.D. Kopinke, J. Prschmann, and A. Georgi, Chapter 8 in: J. Pawliszyn (ed.),
Applications of Solid Phase Microextraction. Royal Society of Chemistry, Cambridge,
1999, pp. 111-128.
24 S.J. You, Y. Yin. and H.E. Allen, Sci. Total Environ., 227 (1999) 155-160.
25 Y. Lu and H.E. Allen, Sci. Total Environ., 277 (2001) 119-132.
26 V. Goundaris, P.R. Anderson and T.M. Holsten, Environ. Sci. Technol., 27 (1993)
1381-1387.
27 J. Buffle and G.G. Leppard, Environ. Sci. Technol., 29 (1995) 2169-2184.
28 L.P. Burkhard, Enuiron. Sci. Technol., 34 (2000) 4663-4668.
29 S.A. Taylor and R.D. Jackson, Chapter 37 in: A. Klute (ed.), Methods of SoilAnalysis.
Part 1-Physical and MineralogicalMethods, 2nd edn. American Society of Agron-
omy & Soil Science., Madison, WI, 1986, pp. 927-940.
30 W.H. Gardner, Chapter 21, ibid., pp. 491-544.
31 G.R. Blake and K.H. Hartge, Chapter 13 and 14, ibid., pp. 363-382.
32 G.W. Gee and J.W. Bauder, Chapter 15, ibid., pp. 383-412.
33 D.L. Carter, M.M. Mortland, and W.D. Kemper, Chapter 16, ibid., pp. 413-424.
34 R.E. Danielson and P.L. Sutherland, Chapter 18, ibid., pp. 443-462.
35 G.C. Topp, Chapter 51 in: M.R. Carter (ed.), Soil Sampling and Methods ofAnalysis.
Lewis Publishers, Boca Raton, FL, 1993, pp. 541-557.
36 J.L.B. Culley, Chapter 50, ibid., pp. 529-540.
37 W.H. Hendershot, H. Lalande and M. Duquette, Chapter 19, ibid., pp. 167-176.
38 B.H. Sheldrick and C. Wang, Chapter 47, ibid., pp. 499-511.
39 M.R. Carter and B.C. Ball, Chapter 54, ibid., pp. 581-588.
40 H. Tiessen and J.O. Moir, Chapter 21, ibid., pp.187-200.
41 B.A.G. Knight and T.E. Tomlinson, J. Soil Sci., 18 (1967) 233-243.
42 A. Murdoch and J.M. Azcue, Manual of Aquatic Sediment Sampling. Lewis
Publishers, Boca Raton, FL, 1995.
43 V.P. Evangelou, Environmental Soil and Water Chemistry. Wiley, New York, 1998.
44 P. Masscheleyn, R.D. Delaune and W.H. Patrick, Jr., Environ. Sci. Technol., 25
(1991) 1414-1419.
45 F. Ackermann, H. Bergmann and U. Schleichert, Environ. Technol. Lett., 4 (1983)
317.
46 I.G. Droppo, B.G. Krishnappan, S.S. Rao and E.D. Ongley, Environ. Sci. Technol., 29
(1995) 546-550.
47 A. Paschke and R. Wennrich, unpublished results, UFZ Centre for Environmental
Research Leipzig, Germany, 2000.
48 S.W. Karickhoff, D.S. Brown and T.A. Scott, Water Res., 13 (1979) 241-248.
49 G. Umlauf and R. Bierl, Z. Wasser-Abwasser-Forsch., 20 (1987) 203-209.
50 I.A.Razak, A. Li and E.R. Christensen, Water Sci. Tech., 34 (1996) 29-35.
51 S. Sauv, W. Hendershot and H.E. Allen, Environ. Sci. Technol., 34 (2000)
1125-1131.
52 R.G. Luthy, G.R. Aiken, M.L. Brusseau, S.D. Cunningham, P.M. Gschwend, J.J.
Pignatello, M. Reinhard, S.J. Traina and J.C. Westall, Environ. Sci. Technol., 31
(1997) 3341-3347.
53 A. Delle Site, J. Phys. Chem. Ref. Data, 30 (2001) 187-439.

239
Chapter8

Characterization of analyte binding and

freely dissolved concentrations in
environmental and biological materials
Joop L.M. Hermens and Wouter H.J. Vaes

8.1 INTRODUCTION

8.1.1 Analyte binding and freely dissolved concentration

Freely dissolved concentration is a key entity in environmental chemistry, phar-

macology and toxicology. Within the environment for example, the transport
and distribution of chemicals depend on the chemical in its free form [1]. The
freely dissolved amount of a chemical can substantially deviate from the total
amount by binding to dissolved organic carbon such as humic acids and this may
affect the uptake and accumulation in biota [2-8], or in other words, the
bioavailability. Reductions in the fraction freely dissolved occur in particular for
more hydrophobic compounds such as higher chlorinated benzenes, polychlori-
nated biphenyls (PCBs) and pesticides such as DDT. Servos et al. [9], for
example, observed a decrease in bioconcentration of polychlorinated dibenzo-p-
dioxins in rainbow trout in the presence of Aldrich humic acid. Bioavailability
and freely dissolved concentrations are also important issues in sediment and
soil toxicology [10-12].
Within pharmacology and drug design, the issue of the freely dissolved
fraction has a long tradition. Protein binding of drugs in blood is a key issue in
the in vivo bioavailability. As mentioned by Seydel [13]: "in general only the
unbound drug is active and capable of diffusing across membranes and only the
free drug is glomerularly filtered". Information on freely dissolved concentra-
tions is important in in vitro studies as well. Quantitative data from in vitro
studies are often used as indications of the potency of chemicals to interact with
a certain receptor or target, or in comparisons of degradation rates in a certain
transformation reaction. In most cases, these data are expressed via a nominal
or total concentration. It has been shown several times that the actual active
(freely dissolved) concentration may deviate strongly from the nominal

Comprehensive Analytical Chemistry XXXV7I

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 241
concentrations [14--16]. Actual exposure concentrations can be much lower than
nominal ones, due to binding to serum proteins, loss processes or simply due to
uptake into cells and tissue or (non) specific binding to other components in the
test or culture medium. As a consequence, in vitro effect concentrations may
vary widely between tests and laboratories. Even for the same test and chemical,
effect concentrations may show a large variation because of different test
conditions and the outcome of in vitro tests can depend on the specific test
conditions [14-19]. Nagel et al. [20] for example concluded that "the impact of
xenoestrogens is substantially affected by their specific access to cells from
serum, and that in vitro assays that evaluate the effective free concentrations
and other factors substantially improve the prediction of the bioactivity of
endocrine disrupters". A better characterisation of the freely dissolved concen-
tration in vitro tests may lead to more sound data. This uncertainty in external
exposure data has also led to the development of alternative approaches and
estimation methods to express effect data based on internal or target site
concentrations [16,19, 21-27].
These examples all show the need for methodologies to measure the freely
dissolved concentration and binding affinities of analytes, being pharmacologi-
cally active compounds or environmental pollutants.

8.1.2 Theory of analyte binding

Analyte binding can be due to adsorption as well as absorption. Absorption is

related to partitioning and can be written as an equilibrium process with Kph2,ph
as the partition coefficient and C the concentration.

Xphl Xh2

withKph2.phl = Cph2/Cphl (8.1)

If the two phases are difficult to separate or if phase 2 is "dissolved" in phase 1,

the concentration in phase 1 is usually referred to as the freely dissolved or
unbound concentration (Cu pr Cfree). C. can be calculated from the total concen-
tration Ct and the "concentration of phase 2" in phase 1 which is equal to the
ratio of the mass of phase 2 (Mphs,,2) and the total volume (V) via Eq. (8.2).

Cu (Cfre) = Ct/(1 + Kph2,phl Mph2/V) (8.2)

The binding of drugs to proteins is usually described as an adsorption process in

which the bound concentration can be described as [28]:
Cbd = Ka. C. P (8.3)

The unbound concentration (C.) can be calculated from

C~ = f -Ct (8.4)

242
with fu (fraction unbound):
fu = 1/(1 +K fup Pt) (8.5)
in which: Cbd is the concentration of bound analyte; C, is the concentration of
unbound analyte; Ct is the total concentration of analyte; P is the concentration
of unoccupied protein; Pt is the total protein concentration; fu is the fraction of
the analyte that is unbound; f is the fraction of the total number of binding
sites unoccupied; and K a is the association constant.
These equations and abbreviations, taken from Ref. [28], assume that there
is only one binding site, while in practice a protein molecule can have more
binding sites with different affinities for the analyte [29]. Information about the
unbound analyte is particularly important in understanding and modelling
kinetic behaviour, while information on the bound fraction is relevant in quanti-
fying drug-receptor interactions. The analysis of binding experiments often
assume that the free concentration of analyte is equal to the amount added [30]
and the affinity constant K. is then derived from a dose response curve with the
total concentration on the x-axis. However, in the analysis of drug-receptor
binding, information on the freely dissolved concentration is needed in those
cases where the unbound concentration of the agonist decreases due to binding
and the discrepancy between the total and free concentration is not the same in
each concentration.
So, binding affinities can be calculated by measuring the bound fraction and
in some cases, information on the free fraction is needed. As an alternative,
binding affinities or partition coefficients can also be derived from measuring
only the unbound concentration as a function of the increasing concentration of
the second phase (ph 2) or protein (Pt). Assuming a 100% mass balance, Kph2,phl
from Eq. (8.1) and Ka from Eq. (8.5) can be calculated from:
Cu (Cfie0) C=
t (8.6)
with fu = 1/(1 + Kph2,,h' Mph2 /V)

Cu = C/(1 + Ka' Pt) (8.7)

with f = 1/(1 + K a P).

It should be realized that Eq. (8.7) is only valid in the initial phase of the
binding curve where fup - 1. In principle, if the response of the analytical method
is linear, there is no need for absolute concentration measurements and the
response of the analytical detection method can be used directly to calculate
Kph2,phl and Ka.
Binding affinities are mostly measured in saturation binding experiments
using various concentrations of the analyte or via competitive binding experi-
ments with the single concentration of radioligand at various concentrations of
an unlabeled competitor and, in many cases, a correction is needed for non-
specific binding [30].

243
8.2 TECHNIQUES FOR MEASURING FREELY DISSOLVED
CONCENTRATION AND ANALYTE BINDING

8.2.1 Techniques used in bioanalyses and environmental analyses

for measuring freely dissolved concentrations

In order to be able to measure freely dissolved concentration, the sample needs

to be treated to enable the analytical separation of the bound and free chemical.
Basically, two approaches are used in analytical laboratories to determine freely
dissolved concentrations or partition coefficients: one approach focuses solely on
the measurement of the free analyte; the other focuses on physical separation of
two phases, i.e., the aqueous phase and the phase to which the analyte is bound.
Clearly, the former is experimentally rather new, and characterised by
partition techniques like vial equilibration headspace analysis, and negligible
depletion extraction. The latter, e.g., equilibrium dialysis and ultra-filtration, is
used on a more routine basis.
The focus of this chapter is measurement of the freely dissolved concentra-
tion, with emphasis on solid phase microextraction as a relatively new tech-
nique. A brief overview of available methods for measuring freely dissolved
concentrations and binding affinities is given in Table 8.1. A distinction is made
between methods which are mostly used in the pharmaceutical sciences and
those used in environmental research. Experimental methods used nowadays

TABLE 8.1
Methods used to measure freely dissolved concentrations and or partition/binding phenomena
Method General reference
Methods used in drug research
Equilibrium dialysis [29]
Ultrafiltration [29]
Ultracentrifugation [29]
Affinity chromatography (protein stationary phases) [29]
High performance size exclusion chromatography [29]
Capillary electrophoresis [29]
Fluorescence spectroscopy [29]
Methods used in environmentalresearch
Dynamic headspace analysis [43-45]
Static headspace analysis [46]
Reversed phase cartridge [54]
Equilibrium dialysis [9]
Semipermeable membrane devices (SPMD) [55,56]
Solid phase partitioning (empore disk) [57]
Solid phase micro-extraction (SPME) [39,41,49,50]
Thin-film solid-phase extraction [58]

244
for the determination of freely dissolved concentrations of drugs and serum-
protein binding include equilibrium dialysis, ultra-filtration and ultra-centri-
fugation and gel filtration [29]. Other techniques include chromatographic
methods, solid phase extractions and headspace analyses (see Table 8.1). Each of
these methods has its own advantages, disadvantages and limitations. It is not
the aim of this chapter to give a detailed analysis and evaluation of all available
methods, so the overview is kept short. More detailed information can be found
in review papers, e.g. [29].
Many methods are available to make a physical separation between the two
phases and thereby are very suitable to determine binding affinities. Neverthe-
less, only a few can also be used to determine the free concentration in plasma,
for example. The major problem of a technique like equilibrium dialysis is that
the plasma should equilibrate with the aqueous phase on the other side of the
membrane. Besides the kinetic drawbacks, there still remains the issue of
increased volume. The addition of a volume of competing phase on the other side
of the membrane can influence the equilibrium between the aqueous phase and
the protein bound fraction in the serum sample. Thus, the measured free
concentration is not representative of the free concentration of the non-dialysed
serum sample. Similar drawbacks exist for most other phase-separation tech-
niques, although it must be mentioned that the extent to which bias is
introduced by the physical separation greatly depends on the affinity constants
and volumes used. It should be emphasised that this experimental bias is only of
influence for the free concentration measurements, but not for binding
coefficients for which the analyte is determined in both phases.
For any method of measuring the true freely dissolved concentration, the key
is not to perturb in any way the equilibrium that exists between the aqueous
dissolved fraction and the bound fraction. In principal, this is an impossible
assignment, since almost every measurement either extracts, concentrates or
dilutes the sample during its treatment. Nevertheless, minimising this effect of
the sample treatment by, e.g., extracting a negligible amount, diluting with a
negligible volume (dialysis), or concentrating only a negligible volume (ultra-
filtration), will have only a negligible effect on the measurement of the free
concentration.
In this chapter we focus on the application of negligible depletion solid phase
microextraction (nd-SPME), which is an example of an extraction technique
based on earlier described negligible influence measurements. The nd-SPME
technology, which we introduced in 1996, is rather new and therefore the
mechanisms, advantages and drawbacks will be discussed.

8.2.2 Solid phase microextraction (SPME) for measuring freely

dissolved concentrations

Solid phase microextraction (SPME) is a potentially promising new technique

for measuring analyte binding and freely dissolved concentrations. It was devel-

245
oped by Pawliszyn and co-workers [31] as a simple extraction method with many
advantages over classical solid phase extractions. The principle, theoretical
aspects and experimental details of the method are described in several review
papers [32-35] and others discuss applications within different fields such as in
environmental analyses [32] and analyses of drugs in biological samples [36-38].
In the early stages, the development and application of SPME was mainly
focused on its capacity as an extraction technique. Much attention was given to
optimising recoveries or extraction efficiencies. Later on, it was realized that the
technique also represented a simple method for measuring freely dissolved
concentrations in complex matrices [16,27,39-41]. Based on the assumption that
only the freely dissolved fraction of a chemical partitions to the fiber and under
conditions where the extracted amount is negligibly small in comparison to the
total freely dissolved amount in a sample, the concentration on the fibre is in
principle directly proportional to the freely dissolved concentration. This specific
application of SPME is also referred to by Vaes et al. [39] as "negligible
depletion" SPME; the principle of nd-SPME is illustrated in Fig. 8.1.
The basic condition that needs to be met for nd-SPME is that the sampling
does not (or hardly) affect the amount present in the aqueous phase and,
therefore, the equilibrium between the bound and unbound fraction remains
unperturbed. The concentration on the fiber (Cf) is then directly proportional to
Cu. Additional conditions that need to be met are [39]:
- there is equilibrium between C. and Cb;
- the uptake kinetics is not influenced by the matrix;
- the matrix does not bind so substantially to the fiber that the concentration
becomes higher than expected based on Cu or that the characteristics of the
fiber are affected;
- the depletion of the amount present in the aqueous phase should be neglig-
ible; Vaes at al. [39] used a 5% limit for the depletion, which is represented by
Eq. (8.8). These conditions are discussed in more detail in Section 8.3.
Cf Vf < 0.05 C V (8.8)
with
Cf /C = Kfw
where Cf is the concentration in the fiber; Vf is the volume of the fiber; Cu is the
unbound concentration at equilibrium; V is the volume of the aqueous phase;
and KfW is the fiber-water partition coefficient.
The advantage of nd-SPME is that the sampling is relatively simple and that
no equilibrium is needed, assuming that the kinetics are independent of the test
conditions. The approach became possible due to the very small volume of the
SPME fiber as sampling phase.
SPME is also applied in more complex systems including other phases (see
Fig. 8.2). Headspace SPME is well known and applied in several studies [32] with
the advantage of clean samples. In relation to free concentration measurements,

246
 ., .
IeorlSPME fiber

Negligible Depletion ¢aqusau phase

With dissolved matrix
i p|soladpisePMEiber

Left: Fig. 8.1. Principle of negligible solid phase micro-extraction (nd-SPME) for measuring
freely dissolved concentrations of analytes [39].
Right: Fig. 8.2. Solid phase micro-extraction in a multi compartment system.

it should be realized that the condition of negligible depletion should be fulfilled.

In addition, unless measured at equilibrium, an analysis of rate limiting steps
and its consequences for the analysis is needed. The same holds for systems with
a solid matrix such as soil or sediment. Mayer et al. [42] used disposable SPME
fibers and equilibrium partitioning in a sediment-water system to measure
freely dissolved pore water concentrations. Depletion of the aqueous phase is not
critical in this situation because the aqueous concentration is continuously
refilled from the solid phase (the sediment), which acts as a buffer and which
should be much "larger" than the fiber volume. In equilibrium, the concentra-
tion on the fiber is then directly proportional to the pore water concentration
and the fiber-water partition coefficient [42]. If the flux of refilling is higher than
the flux to the fiber, kinetic measurements are feasible for measuring the freely
dissolved concentration. However, equilibrium measurements as performed by
Mayer et al. [42], better guarantee precise measurements, because they depend
less on actual circumstances and fluctuations. Actually, the same concept of
negligible influence of the measurement was used in this study. The sampling
did not influence the concentration in the aqueous phase because the sample was
allowed to re-equilibrate. The organic phase was used as a dominating reservoir,
from which only a negligible amount was extracted.
These equilibrium SPME measurements are also a tool for measuring
fugacities [42]. Measuring fugacities is a concept often used in analysing the fate
of compounds at large as well as small scales and not many techniques are
available [43-46].

8.3 SOLID PHASE MICROEXTRACTION FOR MEASURING FREELY

DISSOLVED CONCENTRATIONS AND PARTITION BEHAVIOUR:
EXAMPLES IN BIO- AND ENVIRONMENTAL ANALYSES

Although there are many studies and review publications [36,37] on the applica-
tion of SPME in biomedical analyses, only a few studies focus on the measure-
ment of freely dissolved concentrations in biological samples. The negligible

247
depletion SPME method (nd-SPME) was developed specifically to measure
freely dissolved concentrations [16,39]. The first study presented data for serum
protein binding of aniline, nitrobenzene, 4-chloro-3-methylphenol and 4-n-
pentylphenol [39]. Fractions freely dissolved were calculated from plots of total
versus free concentrations of the test chemical at one fixed protein concentra-
tion. A dialysis experiment was performed to check for a disturbance of the
analysis by direct binding of protein to the fiber. Peak areas of the SPME fibers
exposed on both sides from the dialysis membrane that separated the protein
solution from the buffer solution were the same, which shows that under these
test conditions no influence of binding of protein on the fiber is observed. The
same approach was applied to measure freely dissolved concentrations in
suspensions that contain microsomes, cells and membrane vesicles [16]. Koster
et al. [47] applied direct immersion SPME to extract the local anaesthetic
lidocaine from human plasma. The protein binding affinity was calculated from
a plot of extraction yield against the dilution factor of the protein solution using
equilibrium partitioning according to Eq. (8.9).
nf = KfVfno/(KfVf + KpNp + Vpl) (8.9)
where nf is the amount of drug in the fiber coating, Kf is the fiber-matrix
partition constant, n0 is the total amount of drug in the sample, Kp is the protein
association equilibrium constant, N is the number of available sites for protein-
drug binding and Vf and Vp; are the volume of the fiber coating and the plasma
sample, respectively. In this case, depletion of the bound and free fraction in the
sample is allowed because the analysis is performed at equilibrium. If Kf is
known, Kp N can be derived from Eq. (8.9) by varying Vp1.
The mathematical treatment of the analysis results by Koster et al. [47]
immediately points out that total concentration measurements with SPME in
samples where protein binding can play an important role, are doomed to fail.
High binding affinities will decrease nf and thereby negatively influence extrac-
tion recovery (ratio of nf and no). Moreover, interindividual variation in concen-
trations of plasma proteins can give interindividual variations in measured
plasma concentrations of the analyte, that are only a consequence of differences
in protein concentration instead of analyte concentrations. One elegant way to
circumvent this problem by GC-MS and isotopically labelled chemicals has been
described by Poerschmann et al. [40]. They used a deuterated analyte as internal
standard, which behaves according to the same affinity constant or partition
coefficient as the analyte of interest. From the ratio of the abundance of the ions
of both the analyte as the deuterated internal standard, the total concentration
can be determined.
For free concentrations, the advantage of the negligible depletion method is
that one calibration curve can be used to directly derive the freely dissolved
concentration and that, in principle, measurements can be performed under
non-equilibrium conditions. This gets even more important in headspace SPME
in which four phases are involved (fiber, headspace, protein and the aqueous

248
phase with the freely dissolved concentration). Yuan et al. [48] discussed the
measurement of protein binding via headspace SPME and pointed to the calibra-
tion problem and the effects of depletion on the calibration curve of concentra-
tion on the fiber versus the freely dissolved concentration. They demonstrated
that "for the three-phase system the amount of the analyte partitioned in the
headspace could be ignored only in certain circumstances, where the Henry's
law constant and the ratio between headspace volume and sample volume are
sufficiently small" [48].
Kopinke et al. [41] pointed at the possibility of applying SPME in the
measurement of sorption coefficients to humic acids because only the free solute
can interact with the fiber coating. This same group measured sorption coeffi-
cients to dissolved organic matter (DOM) of a series of organotin compounds [49]
using headspace as well as direct immersion SPME. Although some fouling of
the fibers occurred when applying them in a solution, these phenomena did not
affect the extraction. Sorption coefficients measured in headspace and solution
SPME did not show significant differences, although the headspace SPME was
preferred because fiber fouling can be completely excluded. In this same study,
the influence of DOM on the extraction kinetics to the fiber was investigated
with the idea that kinetic measurements would be less time-consuming than
equilibrium measurements. The addition of DOM did not affect the kinetics of
the partition process.
Using the negligible depletion SPME method, Urrestarazu-Ramos et al. [50]
measured sorption coefficients to Aldrich humic acids of penta and hexachloro-
benzene, PCB 77 and DDT. The sorption coefficients were derived from mea-
surements of the freely dissolved concentration at different humic acid
concentration using Eq. (8.6). PDMS fibers were used and the measurements
were performed in the aqueous solution and in the kinetic uptake phase. Figure
8.3 gives plots of the decrease in free concentration of PCB77 with increasing

0
05
Z

_u
1B_

0 5 10 15 20 25 30
concentration of humic acids (mg/l) CDOC

Fig. 8.3. Decrease in free concentration (plotted as freely dissolved fraction) as a function of the
humic acid concentration (mg/1) for PCB#77 (from [50]).

249
humic acid concentration. The advantage of this procedure is that no absolute
calibration of the concentration is needed. In principle, the partition coefficient
can be calculated based on the response of the analytical detection method. The
sorption coefficients were very close to reported literature values determined
with other techniques [50]. In a study of freely dissolved concentrations of
hydrophobic chemicals in soil and in artificial human digestive mixtures, Oomen
et al. [51] observed a slight increase in free concentrations of PCB52 with
increasing protein concentrations. Although the conditions in this study were
rather extreme (very low free fractions of less than 1%), these observed
phenomena may point, as suggested by the authors, to an increased flux in the
diffusion layer due to desorption from the matrix.
In a recent review Ulrich et al. [38] discussed the possibility of using SPME
as a method for measuring freely dissolved concentrations in biomedical analy-
sis. He addressed the topics of fouling of the fiber with proteins and the effects of
the matrix on the kinetics of uptake in the fiber. The problem of fouling of the
fiber has also been discussed in a recent review from Lord and Pawliszyn [52].
Ulrich [38] concluded that headspace SPME should be applied whenever
possible in SPME of body fluids. He mentioned that "the burden of the fiber with
proteins, for instance is considerable decreased. The lifetime of the fiber is
increased because irreversible damage is delayed. A reversible change of
extraction properties of the coating is also avoided and, therefore, the precision
of the method is improved". The introduction of disposable fiber material by
Mayer et al. [42,53] may partly reduce potential fouling problems, because the
fiber is used only once.

8.4 CONCLUDING REMARKS

As mentioned above, the basic assumption behind nd-SPME for measuring

freely dissolved concentration is that the sampling does not affect the amount
present in the aqueous phase and, therefore, the equilibrium between the bound
and unbound fraction remains undisturbed. The concentration on the fiber (Cf)
is then directly proportional to Cu. The conditions that should be fulfilled are
given by Vaes et al. [39] and include the following: there is equilibrium between
C, and Cb; the uptake kinetics is not influenced by the matrix; the matrix does
not bind so substantially to the fiber that the concentration becomes higher than
expected based on Cu or that the characteristics of the fiber are affected; and the
depletion of the amount present in the aqueous phase should be negligible.
In the meantime, a few studies have addressed these topics and have led to
recommendations such as equilibrium measurements and/or the application of
only headspace SPME instead of direct immersion SPME. In some cases, these
recommendation are difficult to fulfill, in particular for very hydrophobic
compounds (very long equilibration times) or chemicals with a low volatility
from the aqueous phase.

250
 For the majority of applications, direct immersion nd-SPME continues to be
a good option for measuring freely dissolved concentrations. The conditions of
nd-SPME should, however, be carefully examined for each application to ensure
that the necessary requirements are met. Information about the freely dissolved
concentrations of chemicals is still of utmost importance in understanding the
fate of chemicals in the environment and exposure in toxicology. At present, the
availability of measurement techniques for the determination of freely dissolved
concentration limits its use. The methods that are now in use each have their
advantages, disadvantages [29] and limitations. As with all new methods,
further development to enable the routine measurement of free concentrations
will take time.

REFERENCES

1 D. Mackay and S. Paterson, Environ. Sci. Technol., 25 (1991) 427.

2 K.E. Day, Environ. Toxicol. Chem., 10 (1991) 91.
3 C.T. Chiou, T.I. Brinton, R.L. Malcom and D.E. Kile, Environ. Sci. Technol., 20
(1986) 502.
4 B.J. Eadie, N.R. Morehead and P.F. Landrum, Chemosphere, 20 (1990) 161.
5 J. Kukkonen, WaterRes., 26 (1992) 1523.
6 J. Kukkonen and J. Pellin, Sci. Total Environ., 152 (1994) 19.
7 A. Kullberg, K.H. Bishop, A. Hargeby, M. Jansson and R.C.J. Petersen, AMBIO, 22
(1993) 331.
8 I.H. Suffet, C.T. Javfert, J. Kukkonen, M.R. Servos, A. Spacie, L.L. Williams and J.A.
Noblet, in: Bioavailability. CRC Press, Lewis Publisher, 1994.
9 M.R. Servos, D.C.G. Muir and G.R. Barrie Webster, Aquat. Toxicol., 14 (1988) 169.
10 D.C.G. Muir, B.E. Townsend and W.L. Lockhart, Environ. Toxicol. Chem., 2 (1983)
269.
11 P.F. Landrum, W.S. Dupuis and J. Kukkonen, Environ. Toxicol. Chem., 13 (1994)
1769.
12 M. Alexandar, Environ. Sci. Technol., 29 (1995) 2713.
13 J.K. Seydel and K.-J. Schaper, Pharmac. Ther., 15 (1982) 131.
14 R.S. Obach, Drug Metabol. Disp., 25 (1997) 1359.
15 J.B. Houston, Biochem. Pharmacol., 47 (1994) 1469.
16 W.H.J. Vaes, E. Urrestarazu Ramos, C. Hamwijk, I. van Holsteijn, B.J. Blaauboer, W.
Seinen, H.J.M. Verhaar and J.L.M. Hermens, Chem. Res. Toxicol., 10 (1997) 1067.
17 M. Giilden, H. Seibert and J.U. Voss, ATLA, 22 (1994) 185.
18 M. Giilden and H. Seibert, Toxicol. in Vitro, 11 (1997).
19 B.I. Escher, M. Snozzi and R.P. Schwarzenbach, Environ. Sci. Technol., 30 (1996)
3071.
20 S.C. Nagel, F.S. vom Saal and W.V. Welshons, Proc. Soc. Exp. Biol. Med., 217 (1998)
300.
21 L.S. McCarty and D. Mackay, Environ. Sci. Technol., 27 (1993) 1719.
22 A.P. Van Wezel and A. Opperhuizen, Crit. Rev. Toxicol., 25 (1995) 255.
23 B. Escher and R.P. Schwarzenbach, Environ. Sci. Technol., 30 (1996) 260.
24 W.H.J. Vaes, E. Urrestarazu Ramos, H.J.M. Verhaar and J.L.M. Hermens, Environ.
Toxicol. Chem., 17 (1998) 1380.
25 M.T. Miller, A.J.B. Zehnder and B.I. Escher, Environ. Toxicol. Chem., 18 (1999)
2767.

251
26 E.M.J. Verbruggen, W.H.J. Vaes, T.F. Parkerton and J.L.M. Hermens, Environ. Sci.
Technol., 34 (2000) 324.
27 J.L.M. Hermens, A.P. Freidig, E. Urrestarazu-Ramos, W.H.J. Vaes, W.M.G.M. van
Loon, E.M.J. Verbruggen and H.J.M. Verhaar, in: PersistentBioaccumulative Toxic
Chemicals. ACS Symposium Series, Vol. 773, American Chemical Society,
Washington, DC, 2000.
28 M. Rowland and T.N. Tozer, Clinical Pharmacokinetics:Concepts and Applications.
Williams and Wilkens, Baltimore, MD, 1998.
29 J. Oravcova, B. Bohs and W. Lindner, J. Chromatogr. B, 677 (1996) 1.
30 H.J. Motulsky, Analyzing Data with GraphPad Prism, GraphPad Software Inc., San
Diego, CA, 1999.
31 J. Pawliszyn, Trends Anal. Chem., 14 (1995) 113.
32 J. Pawliszyn, Solid Phase Microextraction: Theory and Practice. Wiley-VCH, New
York, 1997.
33 D. Louch, S. Motlagh and J. Pawliszyn, Anal. Chem., 64 (1992) 1187.
34 H. Lord and J. Pawliszyn, J. Chromatogr.A, 885 (2000) 153.
35 J. Pawliszyn, J. Chromatogr. Sci., 38 (2000) 270.
36 N.H. Snow, J. Chromatogr.A, 885 (2000) 445.
37 G. Theodoridis, E.H. Koster and G.J. de Jong, J. Chromatogr.B, 745 (2000) 49.
38 S. Ulrich, J. Chromatogr.A, 902 (2000) 167.
39 W.H.J. Vaes, E. Urrestarazu Ramos, H.J.M. Verhaar, W. Seinen and J.L.M.
Hermens, Anal. Chem., 68 (1996) 4463.
40 J. Poerschmann, Z. Zhangh, F.-D. Kopinke and J. Pawliszyn, Anal. Chem., 69 (1997)
579.
41 F.-D. Kopinke, J. Porschmann and M. Remmler, Naturwissenschaften, 82 (1995) 28.
42 P. Mayer, W.H.J. Vaes, F. Wijnker, K.C.H.M. Legierse, R. Kraaij, J. Tolls and J.L.M.
Hermens, Environ. Sci. Technol., 34 (2000) 5177.
43 C. Yin and J.P. Hassett, Environ. Sci. Technol., 20 (1986) 1213.
44 J.W. Sproule, W.Y. Shiu, D. Mackay, W.H. Schroeder, R.W. Russell and F.A.P.C.
Gobas, Environ. Toxicol. Chem., 10 (1991) 9.
45 M. Horstmann and M.S. McLachlan, Environ. Sci. Technol., 26 (1992) 1643.
46 J. Resendes, W.Y. Shiu and D. Mackay, Environ. Sci. Technol., 26 (1992) 2381.
47 E.H. Koster, C. Wemes, J.B. Morsink and G.J. de Jong, J. Chromatogr.B, 739 (2000)
175.
48 H. Yuan, R. Ranatunga, P.W. Carr and J. Pawliszyn, Analyst, 124 (1999) 1443.
49 J. Poerschmann, F.-D. Kopinke and J. Pawliszyn, Environ. Sci. Technol., 31 (1997)
3629.
50 E. Urrestarazu-Ramos, S.N. Meijer, W.H.J. Vaes, H.J.M. Verhaar and J.L.M.
Hermens, Environ. Sci. Technol., 32 (1998) 3430.
51 A. Oomen, J.P. Groten, D.T.H.M. Sijm, J. Tolls and A.J.A.M. Sips, Environ. Sci.
Technol., 34 (2000) 297.
52 H. Lord and J. Pawliszyn, J. Chromatogr.A, 902 (2000) 17.
53 P. Mayer, W.H.J. Vaes and J.L.M. Hermens, Anal. Chem., 72 (2000) 459.
54 P.F. Landrum, S.R. Nihart, B.J. Eadis and W.S. Gardner, Environ. Sci. Technol., 18
(1984) 187.
55 J.N. Huckins, M.W. Tubergen and G.K. Manuweera, Chemosphere, 20 (1990) 533.
56 G.S. Ellis, J.N. Huckins, C.E. Rostad, C.J. Smitt, D.D. Petty and P. MacCarthy,
Environ. Toxicol. Chem., 14 (1995) 875.
57 A.P. Freidig, E. Artola Garicano, F.J.M. Busser and J.L.M. Hermens, Environ.
Toxicol. Chem., 17 (1998) 998.
58 J.B. Wilcockson and F.A.P.C. Gobas, Environ. Sci. Technol., 35 (2001) 1425.

252
Chapter 9

Unified theory of extraction

Janusz Pawliszyn

9.1 INTRODUCTION

The sample preparation step in an analytical process typically consists of the

extraction of components of interest from a sample matrix. This procedure can
vary in the degree of selectivity, speed and convenience, depending on the
approach and conditions used, as well as on the geometric configurations of the
extraction phase and conditions. Optimisation of this process aids enhancement
in the performance of the overall analysis. Proper design of the extraction
devices and procedures facilitates rapid and convenient on-site implementation,
coupled with separation/quantification and/or automation. The key to rational
choice, optimisation and design is an understanding of the fundamental princi-
ples governing mass transfer of analytes in multiphase systems. There is a
tendency to categorise extraction techniques according to random criteria. The
objective of this chapter is to emphasise common principles among different
extraction techniques and describe a unified theoretical treatment.

9.1.1 Steps in the analytical process

The analytical procedure for complex samples consists of several steps typically
including sampling, sample preparation, separation, quantification, statistical
evaluation, and decision making (see Fig. 9.1). Each step is critical for obtaining
correct and informative results. The sampling step includes deciding where to
get samples that properly define the object or problem being characterized, and
choosing a method to obtain samples in the right amounts. The objective of the
sample preparation step is to isolate the components of interest from a sample
matrix, because most analytical instruments cannot handle the matrix directly.
Sample preparation involves extraction procedures and can also include "clean
up" procedures for very complex "dirty" samples. This step must also bring the
analytes to a suitable concentration level for detection and therefore sample
preparation methods typically include enrichment. During the separation step
of the analytical process, the isolated complex mixture containing target
analytes is divided into its constituents, typically by means of a chromatographic
ComprehensiveAnalytical Chemisty XXXVII
J. Pawliszyn (Ed.) 253
© 2002 Elsevier Science B.V. All rights reserved
 -

Fig. 9.1. Steps in the analytical process.

or electrophoretic techniques. Quantification is the determination of amounts of

the identified compounds. The identification can be based on a retention time
combined with selective detection. More frequently, however, more specific
instruments (viz., mass spectrometers) are used to eliminate possible errors in
quantification due to interferences by confirming identity of analytes. Statistical
evaluation of the results provides an estimate of the target compound concentra-
tion in the sample being analyzed. The data will then support appropriate
decisions, which might include a move to take another sample for further
investigation of the object or problem.
It is important to note, as emphasized by Fig. 9.1, that the analytical steps
follow one after another, and a subsequent step cannot begin until the preceding
one has been completed. Therefore, the slowest step determines the overall
speed of the analytical process, and improving the speed of a single step may not
result in a throughput increase. To increase the throughput of analysis, all steps
need to be considered. Also, errors performed in any preceding step, including
sampling, will result in the overall poor performance of the procedure.

9.1.2 Sample preparation as part of the analytical process

There have been major breakthroughs in the development of improved instru-

mentation, which involve miniaturisation of analytical devices and integration
of different steps into one system. It is recognized that an ideal instrument
would perform all the analytical steps without human intervention, preferably
directly on the site where an investigated system is located rather than moving
the sample to the laboratory, as is common practice at the present time. This
approach would eliminate errors and time associated with sample transport and
storage and therefore result in more accurate and faster analytical data.
Although such a device has not yet been built, today's sophisticated instruments,
such as the gas chromatograph/mass spectrometer (GC/MS) or liquid chroma-
tograph/mass spectrometer, can separate and quantify complex mixtures and
automatically apply chemometric methods to statistically evaluate results. It is

254
much more difficult to copuple sampling and sample preparation steps,
primarily because the current state of the art in sample preparation techniques
employs multi-step procedures involving organic solvents. These characteristics
make it difficult to develop a method that integrates sampling and sample
preparation with separation methods, for the purposes of automation. The
result is that over 80% of analysis time is currently spent on sampling and
sample preparation steps for complex samples.
One of the reasons for the slow progress in the area of sample preparation is
that the fundamentals of extraction involving natural, frequently complex
samples are much less well developed and understood compared to more
physicochemically simpler instrumental systems used in separation and
quantification steps such as chromatography and mass spectrometry. This
situation creates a feeling that rational design and optimisation of extraction
systems is not possible. Therefore, the development of sample preparation
procedures is frequently considered to be an "art" not a "science".

9.1.3 Classification of extraction techniques

This situation creates a tendency by practitioners and regulatory agencies to

prefer exhaustive vs. non-exhaustive techniques (see Fig. 9.2) even though this
choice frequently results in labour-intensive and costly procedures. The
objective of the exhaustive techniques is to completely remove analytes from a
sample matrix and transfer them to the extraction phase. The fundamental
advantage of exhaustive methods is that, in principle, they do not require
calibration since the vast majority of analytes are transferred to the extraction
phase. This objective is accomplished by application of overwhelming volumes of
extraction phase. To reduce the amounts of solvents and time required to
accomplish exhaustive removal, batch equilibrium techniques (for example,
liquid-liquid extractions) are frequently replaced by flow-through techniques.
For example, a sorbent bed can be packed with extraction phase dispersed on a
Extraction Techniques

BatchEquilibrium
and Pre-equilibrium

Fig. 9.2. General classification of extraction techniques.

255
supporting material and when the sample is passed through, analytes are
retained on the bed. Large volumes of sample can be passed through a small
cartridge and the flow through a well-packed bed facilitates rapid mass transfer.
The extraction procedure is followed by re-extraction of analytes into a small
volume of solvent resulting in high enrichment. This strategy is used in sorbent
trap techniques and Solid Phase Extraction (SPE) [1]. Alternatively, a sample
(typically a solid sample) can be packed in the bed and the extraction phase can
be used to remove and transport the analytes to the collection point. In super-
critical fluid extraction (SFE) compressed gas is used to wash analytes from the
sample matrix, while in purge and trap atmospheric pressure inert gas performs
the same function. In dynamic solvent extraction, such as Soxhlet, the solvent
continuously removes the analytes from the matrix at the boiling point of the
solvent. In more recent hot solvent extraction techniques smaller volumes of
organic solvent are used to accomplish higher enrichment at the same time,
because of increased solvent capacity and eluting capability at high tempera-
tures and pressures [2].
Alternatively, non-exhaustive approaches can be designed based on equilib-
rium, pre-equilibrium and permeation principles [3]. Equilibrium non-exhaust-
ive techniques are fundamentally analogues to equilibrium exhaustive tech-
niques; however the capacity of the extraction phase is smaller, and in most
cases is not sufficient to remove the majority of analytes from the sample matrix.
This is caused by use of a small volume of the extracting phase relative to sample
volume (microextraction, such as solvent microextraction or solid phase micro-
extraction (SPME) [41 or a low sample matrix-extraction phase distribution con-
stant, such as is typically encountered in gaseous headspace techniques [5].
Pre-equilibrium conditions are accomplished by breaking the contact
between the extraction phase and the sample matrix before the point of equilib-
rium with the extracting phase has been reached. The devices frequently used
are identical to the microextraction systems, but extraction times are shorter. In
permeation techniques, such as membrane extraction [6], continuous steady-
state transport of analytes through the extraction phase is accomplished by
simultaneous re-extraction of analytes. The membrane extraction can be
exhaustive by designing appropriate membrane modules and optimising the
sample and striping flow conditions [7] or it can be optimised for throughput and
sensitivity in a non-quantitative open bed approach [8].
In addition to classification of methods based on more fundamental
principles as discussed above, it is also instructive to divide techniques according
to particular characteristics. For example, there has recently been a trend
towards solvent-free techniques (see Fig. 9.3) [9]. This is an important direction,
not only because it addresses health and pollution prevention issues, but also
because in most cases such approaches are easier to implement for on-site
monitoring in field conditions. This direction has been creating a lot of interest
and research opportunities recently and it is expected to continue to be a very
active area in the near future, The most promising solventless techniques are

256
 ISolvent-free Sample Preparation Methodsl

headspace,
sorbent and approaches. Supercritical Fluid Extraction
membrane
traon-siteimExtraction Etractions.

among
exStatic Staticon process. In all tech-
Cartridge|H Direct
Dynamic | H Dynamic Dick Headepaco

In-tube

Fig. 93. Clssificaxtraction of solvent-free extraction techniques.

9.2 FUNDA EALSSPE SPME

headspace, membrane and sorbent approaches. Supercritical Fluid Extraction
(SFE) is able to remove selectively semivolatile and non-volatile trace compo-
nents from solid matrices; however field implementation of this technology is
very difficult since it wouses heavy and inconvenient components. New develop-
ments in the technology, such as a miniaturised fluid delivery system [10] would
aid on-site implementations.

9.2 FUNDAMENTALS

As the above discussion and Fig. 9.2 indicate, there is a fundamental similarity
among extraction techniques used in the sample preparation process. In all tech-
niques the extraction phase is in contact with the sample matrix and analytes
are transported between the phases. For exhaustive techniques the phase ratio
is higher and geometries are more restrictive compared with non-exhaustive
approaches required to ensure the quantitative transfer of analytes. The ther-
modynamics of the process is defined by the extraction phase/sample matrix dis-
tribution constant. It would be instructive to consider in more detail the kinetics
of processlveoccurring at the extraction phase/sample matrix interface since this
defines the time of the analytical procedure. In many cases the analytes are
re-extracted from the extraction phase, but this step is not discussed here since
this process is analogues and much simpler in principle compared to removing
analytes from a more complex sample matrix. The main objective of this chapter
is to outline the common fundamental principles among various extraction tech-
niques to facilitate a better understanding of selection criteria for appropriate
techniques, device geometries and operational conditions.

9.2.1 Thermodynamics

The fundamental thermodynamic principle common to all chemical extraction

techniques involves distribution of analyte between the sample matrix and the
extraction phase. When a liquid is used as the extraction medium then the
distribution constant, K00:

257
 Fig. 9.4. Partitioning between aqueous sample matrix and organic extraction phase.

=
Kes aa = CJCs (9.1)
defines the equilibrium conditions and ultimate enrichment factors achievable
in the technique, where ae and as are the activities of analytes in the extraction
phase and matrix correspondingly, which can be approximated by the appropri-
ate concentrations. Figure 9.4 shows a schematic example of the system for
liquid-liquid extraction. For solid extraction phase adsorption, equilibria can be
explained by the following equation:

Ke = SCs (9.2)
where Se is the solid extraction phase surface concentration of adsorbed
analytes. The above relationship is similar to Eq. (9.1) with the exception that
extraction phase concentration is replaced with surface concentration. The Se
term in the numerator indicates that the sorbent surface area available for
adsorption must also be considered. This complicates the calibration at equilib-
rium conditions because of displacement effects and the non-linear adsorption
isotherm [11]. The above equations can be used to calculate the amount of
analyte in the extraction phase at equilibrium conditions [4]. For example, for
equilibrium liquid microextraction techniques and large samples, including
direct extraction from the investigated system, the appropriate expression is
very simple:
n = KesVC (9.3)
where K,, is the extraction phase/sample matrix distribution constant, Ve is the
volume of the extraction phase and C. is the concentration of the sample. In het-
erogeneous samples (headspace, immiscible liquids and solids) the components
of the sample partition in the multiphase system are less available for extraction.
This effect depends on analyte affinity and volume of the competing phases and
can be calculated if appropriate volumes and distribution constants are known
using equations discussed in Ref. [4]. The distribution constants depend on vari-
ous parameters including temperature, pressure and sample matrix conditions
such as pH, salt and organic component concentration. All these parameters
need to be optimised for maximum transfer of analytes to the extraction phase
during the method development process. In practice, however, kinetic factors
defined by the dissociation constants, diffusion coefficient and agitation condi-
tion frequently determine the amount of extracted analytes from complex sam-

258
pies since the overall rates are slow and therefore extraction amounts for limited
time experiments do not reach equilibrium values.

9.2.2 Kinetics

9.2.2.1 Liquid-liquidextraction
Let us take the simple case of static extraction of water with organic solvent as
illustrated in Fig. 9.4 to consider the effect of different parameters on extraction
kinetics. The appropriate equation showing concentration profiles in each of the
phases can be obtained by solving Fick's Second Law differential equation for
appropriate boundary conditions:
C, = D a2C(x' t) (9.4)
at ax2
If no convection is present in the system, the distribution constant is defined by
Eq. (9.1) and the two phases are placed in contact with each other at t = 0 then
the solution can be found using Laplace Transform approach to be for aqueous
sample phase (x < 0):
z +erf(zVt-/)
C (x,t)=CO Kes (9.5a)
1+
Kes

and for organic extraction phase (x > 0):

C,(x, t) = C z{l - erf (x / zt) (9.5b)

1+Z
Ke,

where CO is the initial concentration of the analyte in the aqueous phase, D is

the diffusion of analyte in the extraction phase and D in the sample, z = DD,
and K,, is an appropriate distribution constant defined by Eq. (9.1). This solution
to the above equation is shown graphically in Fig. 9.5 for several extraction times
and diffusion of analytes in aqueous and organic phases to be 10-5 cm 2/s. Figure
9.5 illustrates that the concentration gradient is decreasing and extending
deeper into both phases as a function of time. The flux of analytes is decreasing
proportionally with a decrease in the gradient. The concentration effect of
analytes at the boundary on the organic side compared to the bulk aqueous
concentration is not observed at the beginning of extraction due to drop of the
concentration on aqueous side. Therefore decrease in boundary layer thickness
and the diffusion length by agitation of one and particularly of both phases
increases the rate of extraction dramatically. Effects of agitation can be
calculated using the boundary layer model described later. The other way to
improve mass transfer is to use thin films of sample matrix and/or extraction

259
 Boundary

Aqueous organic
>t D\

_ \ _

-2.0 -1.0 0.0 1.0 2.0

x [mm]

Fig. 9.5. Concentration profiles at the interface between infinite volume sample and extraction
phases for analyte characterized by identical diffusion coefficients in aqueous and organic phase
of 10-5 cm2/s. The profiles correspond to 1 s (A), 10 s (B), 100 s (C) and 1000 s (D) after merging
both phases.

phase to decrease the diffusion length. In addition, a combination of agitation of

the sample and thin extraction phase and vice versa also facilitate shorter
extraction times. If the extraction media and matrix phases are of a different
state of matter, then it is more critical to overall extraction kinetics to agitate or
use the thin film of the phase which is characterised by smaller diffusion
coefficient. For example, when extracting a gas or liquid sample with poly(di-
methylsiloxane) it is critical to disperse the extraction phase as a thin film to
rapidly reach equilibria between the phases.

9.2.2.1 Extraction of solids

The most challenging extractions occur when a solid is present as a part of the
sample matrix. This case will be considered as the most general example of
extraction since it involves a number of fundamental processes occurring during
the extraction. If we assume that a matrix particle consists of an organic layer on
an impermeable but porous core and the analyte is adsorbed onto the pore
surface, the extraction process can be modelled by considering several basic
steps as shown in Fig. 9.6. To remove the analyte from the extraction vessel, the
compound must first be desorbed from the surface (A(M,S)) (see Fig. 9.6), then it
must diffuse through the organic part of the matrix (A(M,L)) to reach the
matrix/fluid interface (A(M,I)). At this point the analyte must be solvated by the
extraction phase (A(EP,P)) and then it must diffuse through the static phase
present inside the pore to reach the portion of the extraction phase influenced by
convection, to be transported through the interstitial pores of the matrix and
eventually reach the bulk of the extraction phase (A(EP,B)). The simplest way to
design a kinetic model for this problem is to adopt equations developed by
engineers to investigate mass transport through porous media [12,13].
For the purpose of this discussion, we consider the efficient and frequently
applied experimental arrangement for removing solid bound semivolatile
analytes, involving the use of a piece of stainless steel tubing as the extraction
vessel. The sample is typically placed inside the tubing and a linear-flow

260
 ---

i
Flow
Convection

A(EP,B)

O Particle Core
O Organic Material

Fig. 9.6. Processes involved in extraction of heterogeneous samples containing porous solid
particles. The terms in the figure are discussed in the text.

restrictor is attached to maintain the pressure at the end of the vessel. During
the process, the extraction phase continuously removes analytes from the
matrix, which are then transferred to the collection vessel after the expansion of
the fluid. This leaching process is very similar to chromatographic elution with
packed columns, particularly to the frontal method. The main difference is that
in sample preparation analytes are dispersed in the matrix at the beginning of
the experiment, while in chromatographic frontal analysis a long plug is intro-
duced into the column at the initial stage of the separation process. The principal
objective of the extraction is to remove analytes from the vessel in the least
amount of time, requiring elution conditions under which the analytes are
unretained. In chromatography, on the other hand, the ultimate goal is to sepa-
rate components of the sample, which requires retention of analytes in the
column. Another major difference is that the packing matrix is usually well char-
acterized in chromatography, but in sample preparation it is often unknown.
One way to develop a mathematical model for this extraction approach is to
establish the mass balance equation for the system after careful consideration of
the individual mass transfer steps occurring during the extraction process (see
Fig. 9.3) and specific boundary conditions [14]. Extensive investigations on
similar topics have already been conducted by engineers who have studied the
mass transfer in porous media [12] and chromatographers [13]. In these studies
the relationship between various matrix parameters and flow conditions on the
elution profile were described mathematically, and verified experimentally. In
chromatography this relationship is usually described as contributions from
each of the mass transfer steps to the height equivalent to a theoretical plate
(HETP). The overall performance of the system can be defined as the sum of the
relevant individual components judiciously selected to reflect the most
significant individual steps present in the elution process. For the purpose of
this discussion, this approach is adopted to develop a model for extraction
kinetics in flow-through techniques.

261
 The effect of slow desorption kinetics of analytes from the matrix on the
elution profile, can be described as the contribution to the HETP [8], hRK:

h.K = 2ku, (9.6)

(1 + k) 2 (1+ k )kd (9.6)

where k is the partition ratio, kd is the dissociation rate constant of the analyte-
matrix complex of reversible process, kh is the ratio of the intraparticulate void
volume to the interstitial void space and is expressed as:

ko i1-c e) (9.7)

where ei is intraparticulate porosity and e is interstitial porosity; and ue is the

interstitial linear extraction phase velocity expressed as:
Ue = u(1 + ko) (9.8)
where u = L/t o is chromatographic linear velocity, L is the length of the
extraction vessel and to is the time required to remove one void volume of the
extraction phase from the vessel. Chromatographic and interstitial linear
velocities are identical if matrix particles have low porosity. This analysis can be
extended to elution through a matrix having multiple adsorption sites
characterised by different dissociation rate constants by using the approach
described by Giddings [15].
The diffusion of the analyte in the liquid or swollen solid part of the matrix is
important when polymeric materials are extracted, or the matrix has substantial
organic content. Its contribution can be expressed as hDc:

h c =2 kI d2 ue (9.9)
3 ( +1)2 D,

where dc is the thickness of the matrix component permeable to analyte and D, is

the diffusion coefficient of the analyte in the sample matrix.
The analytes migrate in and out of a pore structure of the matrix during the
elution. This can be described as resistance to mass transfer in the fluid
associated with the porous nature of the environmental matrices which gives
rise to the following HETP component, h0 p:

hP 0(ko +k+kho) 2dp2e (9.10)

30ko(1+ ko) 2 ( l +k) 2 D p

where is the tortuosity factor for the porous particle and D. is the diffusion
coefficient of the analyte in the material filling the pores, which in most practical
cases will be an extraction phase and therefore Dp = De, where De is the diffusion
coefficient of the analyte in the extraction phase. This contribution can be quite
important considering the relatively large particle size (about 1 mm) of

262
environmental matrices, and it becomes particularly important when the pores
are filled with dense organic material, such as humic matter rather than the
extraction phase.
In the flowing bulk of the fluid an analyte experiences resistance to mass
transfer associated with eddy diffusion (random paths of the analytes through
the vessel filled with the particles) which is given by hED:

h,,ED = 2d/ (9.11)

where X is a structural parameter and is close to 1 for spherical matrix particles.
This contribution to band broadening is the most important factor in HPLC
separations and it is expected to remain significant in extractions because of
matrices typically having large particle sizes.
In addition, we should also consider analyte diffusion along the axis of the
vessel (longitudinal diffusion) which can be defined as hLD:

hLD = ~YMDe (9.12)

Ue

where yM is the obstruction factor which characterise the structure of the matrix.
This component is expected to be small. The analyte concentration profile
generated during the experiment as a function of time C(x,t) can be represented
using the equation which describes the dispersion of a plug of finite width [9]:

(x,t) erf 2 xl+k +erf 2 I+k (9.13)

CO 52Y
2

where L is the length of the vessel, COis the initial concentration of analyte in the
extraction vessel and a is the mean square root dispersion of the band expressed
as:

= Ht< (9.14)

where H is equivalent to the HETP in chromatographic systems and is a sum of

the contributions discussed above, H = hR, + h,, + hD + h,, + hL. The mass of
analyte eluted from the vessel during a given extraction time t can be calculated
from the following equation:
1L
2
C(x, t)dx
wmer m~t) is the extracted mass of analyte an is the total amount ofayt(9.15)
mo CsL

where m(t) is the extracted mass of analyte and m0 is the total amount of analyte

263
in the vessel at the beginning of the experiment. We will refer to this function as
the "time elution profile", emphasizing the similarity of the extraction process
in this simple case to chromatographic elution.

9.2.2.3 Convolution model of extraction

The above discussion applies only to the situation when the analytes are initially
present in a fluid phase, which in flow-through techniques corresponds to
elution of uniform spikes from the extraction vessel or when weakly adsorbed
native analytes are removed from an organic-poor matrix such as sand. In other
words, the above relationships are suitable for systems in which the partitioning
equilibrium between the matrix and extraction fluid is reached quickly
compared to the fluid flow. They are also suitable to model static/dynamic
extractions, under good solubility conditions (k = 0), in which the sample is
initially exposed to the static extraction phase (vessel is capped) for a time
required to achieve equilibrium condition prior to elution by fluid flow. If
dynamic extraction is performed from the beginning of extraction, then in the
majority of practical cases, the system is not expected to achieve initial
equilibrium conditions. This is because of slow mass transport between the
matrix and the fluid (for example slow desorption kinetics or slow diffusion in
the matrix). The expected relationship between the amount of analyte removed
from the vessel versus time can be obtained in this case by convoluting the
function describing the rate of mass transfer between the phases F(t) with the
elution time profile m/mo0 (t) derived above (Eq. (9.15)) [16]:

m(t F()d (9.16)

-=0 m

The resulting function describes a process where elution and mass transfer
between the phases occur simultaneously. In this discussion we will refer to this
function as the "extraction time profile" to emphasize the point that in a
majority of extraction cases these two processes are expected to be combined.
F(t) describes the kinetics of the process, which defines the release rate of
analyte from the sample matrix and can include for example: the matrix-analyte
complex dissociation rate constant, the diffusion coefficient, the time constant
that describes swelling of the matrix that will facilitate removal of analyte, or a
combination of the above. Detailed discussion, graphical representations and
applications of this model to describe and/or investigate processes in supercriti-
cal fluid extraction have been described in detail elsewhere [17,18].
The above conclusion can be stated in a more general way. Convolution
among functions describing individual processes occurring during the extrac-
tion, describes the overall extraction process and represents a unified way to
describe the kinetics of these complex processes. The exact mathematical
solution to the convolution integral is frequently difficult to obtain, but graphi-
cal representation of the solution can be calculated using Fourier Transform or

264
numerical approaches. Frequently, it is possible to incorporate mathematical
functions that describe a combination of the unit processes. In the example of
the flow-through system discussed above, the elution function describes the
effect of porosity and analyte affinity towards the extraction matrix on the
extraction rate. It should be emphasised that the convolution approach consid-
ers all processes equivalently. In practice, however, a small number-and
frequently just one unit process-controls the overall rate of extraction so the
equations can be simplified by considering this fact.
Determination of the limiting step is not possible exclusively by qualitative
agreement with the mathematical model since the effect on recovery of most of
the unit processes has an exponential decay nature. To properly recognise them,
quantitative agreement and/or the effect of extraction parameters needs to be
examined. Identification of the limiting process provides a valuable insight into
the most effective approach to optimise the extraction.

9.3 OPTIMISATION OF THE EXTRACTION PROCESS

A fundamental understanding of the process leads to better strategies for opti-

misation of performance. In heterogeneous samples, for example, the release of
solid bound analytes from sample matrix, through a reversal of chemisorption or
inclusion, frequently controls the extraction rate. By recognising this fact,
extraction parameters can be changed to increase the extraction rates. For
example, dissociation of the chemisorbed analytes can be accomplished either by
using high temperature or the application of catalysts. Recognition of this fact
led to the development of high temperature supercritical fluid extraction [19],
followed by the evolution of both the hot solvent extraction approach [20] and
microwave extraction, with more selective energy focusing at the sample matrix/
extraction phase interface [21]. There is also an indication that milder condi-
tions can be applied by taking advantage of the catalytic properties of the extrac-
tion phase or additives [22]. However, to realize this opportunity more research
needs to be performed to gain more insight about the nature of interactions
between analytes and matrices. Benefits are not only improved speed, but also
selectivity resulting from application of appropriate conditions. This strategy of
simultaneous extraction and clean up has been applied successfully to a very dif-
ficult case of extraction of polychlorinated dibenzo-p-dioxins from fly ash [23].
If the extraction rate is controlled by mass transport of analytes in the pores
of the matrix, then the process can be successfully enhanced by application of
sonic and microwave energy, which induce convection even in the small dimen-
sions of the pore. Frequently, diffusion through the whole or a portion of the
sample matrix containing natural or synthetic polymeric material controls the
extraction rate [24]. In this case swelling the matrix and increasing temperature
results in increased diffusion coefficients and therefore increased extraction
rates.

265
9.3.1 Flow-through techniques

For homogeneous samples and flowing fluid extraction phase, the description of
the extraction process is much simpler and can be based directly on the chro-
matographic theory for liquid stationary phases. Let us consider another case of
the flow-through system where the extraction phase is dispersed as a thin layer
inside the extraction bed and the sample flows through the cartridge. The bed
can be constructed of a piece of fused silica capillary, internally coated with a
thin film of extracting phase [25] (a piece of open tubular capillary GC column;
in-tube SPME) [26], or the bed may be packed with extracting phase dispersed
on an inert supporting material (SPE cartridge). In these geometric arrange-
ments, the concentration profile along the axis x, of the tubing containing the
extracting phase as a function of time t, can be described by adopting the expres-
sion for dispersion of a concentration front:

x-
C(x,t) = 0.5C erf (9.17)

where u, is linear velocity of the sample through the tube, k is the partition ratio
defined as:

k =K VI (9.18)

where Kes is a extraction phase/sample matrix distribution constant, V. is the vol-

ume of the extracting phase and Vv is a void volume of the tubing containing the
extracting phase. a is the mean square root dispersion of the front defined as:

= Ht (9.19)
1+k

where H is equivalent to the HETP (height equivalent to theoretical plate) in

chromatographic systems. This can be calculated as a sum of individual contri-
butions to the front dispersion. These contributions are dependent on the
particular geometry of the extracting system as discussed previously in the
theory section of this chapter.
Figure 9.7 illustrates the normalized concentration profiles produced in the
bed during extraction [25]. The full breakthrough is obtained for the last curve
which corresponds to appropriate volume of the sample matrix. The time
required to pass this required volume through this extraction system
corresponds to the equilibration time of the compound with the bed.
Equation (9.17) and Fig. (9.7) indicate that the front of the analyte migrates
through the capillary/bed with a speed proportional to the linear velocity of the

266
sample, and inversely related to the partition ratio. For the in-tube SPME and
short capillaries with a small dispersion, the minimum extraction time for
in-tube SPME at equilibrium conditions can be assumed to be similar to the time
required for the centre of the band to reach the end of the capillary:

L 1+Ks
te -= \ v (9.20)
U

where L is the length of the capillary holding the extraction phase. For packed
bed extractors typical for SPE techniques, analogous equations can be devel-
oped. In that case the calculated time corresponds to the maximum extraction
time before breakthrough occurs. As expected, the extraction time is propor-
tional to the length of the capillary and inversely proportional to the linear flow
rate of the sample. Extraction time also increases with an increase in the extrac-
tion phase/sample distribution constant and with the volume of the extracting
phase but decreases with an increase of the void volume of the capillary.
9.3.2 Batch techniques
Coupling equations for systems involving convection caused by flow through a
tube as discussed above, are frequently not available for other means of agitation
and other geometric configurations. In these cases the most successful approach
is to consider the boundary layer formed at the interface between the sample
matrix and the extraction phase. Independent of the agitation level, fluid
contacting the fibre's surface is always stationary, and as the distance from the
fibre surface increases, the fluid movement gradually increases until it
corresponds to bulk flow in the sample. To model mass transport, the gradation
in fluid motion and convection of molecules in the space surrounding the fibre
surface can be simplified as a zone of a defined thickness in which no convection
occurs, and perfect agitation occurs in the bulk of the fluid everywhere else. This
static layer zone is called Prandtlboundary layer (see Fig. 9.8) [27].
,= nha
-rait-n

/ Dounuay Sample
i Layer

,
0,
a
0

A X
Pn-iti-n

Left: Fig. 9.7. Normalized concentration profiles for in-tube SPME calculated using the equation
discussed in the text.
Right: Fig. 9.8. Boundary layer model.

267
9.3.3 Boundary layer model

A precise understanding of the definition and thickness of the boundary layer in

this sense is useful. The thickness of the boundary layer (6) is determined by
both the rate of convection (agitation) in the sample and an analyte's diffusion
coefficient. Thus, in the same extraction process, the boundary layer thickness
will be different for different analytes. Strictly speaking, the boundary layer is a
region where analyte flux is progressively more dependent on analyte diffusion
and less on convection, as the extraction phase is approached. For convenience
however, analyte flux in the bulk of the sample (outside the boundary layer) is
assumed to be controlled by convection, whereas analyte flux within the bound-
ary layer is assumed to be controlled by diffusion. 6 is defined as the position
where this transition occurs, or the point at which convection towards the
extraction phase is equal to diffusion away. At this point, analyte flux from
towards the extraction phase (diffusion controlled) is equal to the analyte flux
from the bulk of the sample towards 6, controlled by convection.
In many cases when the extraction phase is well dispersed well forming a
thin coating, the diffusion of analytes through the boundary layer controls the
extraction rate. The equilibration time, t can be estimated as time required to
extract 95% of the equilibrium amount and calculated for these cases from the
equation below 4]:
t, =B (9.21)
D,
where b is the extraction phase thickness, D. is the analyte's diffusion coefficient
in the sample matrix, K is the analyte's distribution constant between the
extraction phase and the sample matrix. B is a geometric factor referring to the
geometry of the supporting material upon which the extraction phase is dis-
persed on. The boundary layer thickness can be calculated for given convection
conditions using engineering principles and it is discussed in more detail later.
Equation (9.21) can be used to predict equilibration times when the extraction
rate is controlled by diffusion in the boundary layer, which is valid for thin
extraction phase coatings (b < 200 microns) and high distribution constants (Ke
> 100).

9.3.4 Solid vs. liquid sorbents

The properties of the extraction phase should be carefully optimised. These

include both bulk physicochemical properties such as polarity as well as physical
properties. The discussion to this point has been limited to extraction by liquid
phases. These exhibit wide linear dynamic ranges associated with linear absorp-
tion isotherms. They also facilitate "gentle" sample preparation since chemi-
sorbtion and catalytic properties, frequently associated with solid surfaces, are
absent. No loss or modification of the analyte occurs during extraction and/or

268
 t =°
Absorption Adsorption

a b
Fig. 9.9. Extraction using absorptive (a) and adsorptive (b) extraction phases immediately after
exposure of the phase to the sample (t = 0) and after completion of the extraction (t = t,).

desorption. Despite these attractive properties of liquid extraction media, solid

phases are frequently used because of their superior selectivity and sensitivity
for some groups of compounds. There is a substantial difference in performance
between liquid and solid coatings (Fig. 9.9). In the case of liquid coatings the
analytes partition into the extraction phase, where the molecules are solvated by
the coating molecules. The diffusion coefficient in the liquid coating allows the
molecules to penetrate the whole volume of the coating within a reasonable
extraction time if the coating is thin (see Fig. 9.9a). In the case of solid sorbents
(Fig. 9.9b) the coating has a well-defined crystalline structure, which if dense,
substantially reduces the diffusion coefficients within the structure. Therefore
within the experimental time sorption occurs only on the porous surface of the
coating (see Fig. 9.9b). During extraction by solid phase, compounds with poor
affinity towards the phase are frequently displaced at longer extraction times by
analytes characterised by stronger binding, or those present in the sample at
high concentrations. This effect is associated with the fact that there is only a
limited surface area available for adsorption. If this area is substantially occu-
pied then a competition effect occurs [6] and the equilibrium amount extracted
can vary with concentrations of both the target and other analytes. On the other
hand, in the case of extraction with liquid phases, partitioning between the sam-
ple matrix and extraction phase occurs. In this case, equilibrium extraction
amounts vary only if the bulk coating properties are modified by the extracted
components, which only occurs when the amount extracted is a substantial por-
tion (a few percent) of the extraction phase. This is rarely observed, since extrac-
tion/enrichment techniques are typically used to determine trace contamination
samples however cannot be neglected as a possible course of non-linearity, when
quantifying very complex matrices.

9.3.5 Diffusion-based calibration

The only way to overcome this fundamental limitation of porous coatings is, as
Fig. 9.9 suggests, to use an extraction time much less than the equilibrium time,
so that the total amount of analytes accumulated onto the porous coating is

269
 - Fiji m-

II"I i - pores
i bulk air
Ii movement
i

i
i
iI solid coating
surface (A)
I'- .-~ boundary layer
I

/, concentration profile

Fig. 9.10. Schematic of the diffusion based calibration model. The terms are defined in the text.

substantially below the saturation value. At saturation, all surface available for
adsorption is occupied. When performing such experiments, not only is it critical
to precisely control extraction times, but also convection conditions since they
determine the thickness of the diffusion layer. One way of eliminating the need
for compensation of differences in convection is to normalize (use consistent)
agitation conditions. For example, the use of a stirring means at a well defined
rotation rate in the laboratory, or fans for field air monitoring, will ensure
consistent convection [28,29]. The short time exposure measurement described
above has an advantage associated with the fact that the rate of extraction is
defined by diffusivity of analytes through the boundary layer of the sample
matrix, and corresponding diffusion coefficients, rather than by distribution
constants. This situation is illustrated in Fig. 9.10 for cylindrical geometry of the
extraction phase dispersed on the supporting rod.
The analyte concentration in the bulk of the matrix can be considered con-
stant when a short sampling time is used, and there is a constant supply of an
analyte via convection. These assumptions are true for most cases of sampling,
where the volume of sample is much greater than the volume of the interface,
and the extraction process does not affect the bulk sample concentration. In
addition, the solid coating can be treated as a perfect sink. The adsorption bind-
ing is frequently instantaneous and essentially irreversible. The analyte concen-
tration on the coating surface is far from saturation and can be assumed to be
negligible for short sampling times and relatively low analyte concentrations in a
typical sample. The analyte concentration profile can be assumed to be linear
from C, to C. In addition, the initial analyte concentration on the coating sur-
face (Ce) can be assumed to be equal to zero when extraction begins. Diffusion of
analytes inside the pores of a solid coating controls mass transfer from outer to
inner surface of the coating.

270
 The function describing the mass of extracted analyte with sampling time
can be derived [30], which results in the following equation:

n(t) BADs f Cs (tdt (9.22)

60

where n is the mass of extracted analyte over sampling time (t) in ng; Ds is the
gas-phase molecular diffusion coefficient; A is the surface area of the sorbent; 6
is the thickness of the boundary layer surrounding the extraction phase; B1 is a
geometric factor and C. is the analyte concentration in the bulk of the sample. It
can be assumed that the analyte concentration is constant for very short
sampling times and therefore Eq. (9.20) can be further reduced to:

n(t) = BDsA Ct (9.23)

6
where t is the sampling time [31].
It can be seen from Eq. (9.23) that the amount of extracted mass is
proportional to the sampling time, the D for each analyte, bulk sample concen-
tration, and inversely proportional to 6. This is consistent with the fact that an
analyte with a greater D will cross the interface and reach the surface of the
fibre coating faster. Values of D, for each analyte can be found in the literature or
estimated from physicochemical properties [25]. This relationship allows for
quantitative analysis. Equation (9.23) can be modified to estimate the analyte
concentration in the sample for rapid sampling with solid sorbents:
n6
Cs= -(9.24)
B,D,At
The amount of extracted analyte (n) can be estimated from the detector
response.
The thickness of the boundary layer (6) is a function of sampling conditions.
The most important factors affecting 6 are geometric configuration of the
extraction phase, sample velocity, temperature and D s for each analyte. The
effective thickness of the boundary layer can be estimated for the coated fibre
geometry (see Fig. 9.10) using Eq. (9.25), adapted from heat transfer theory:
d
6 = 9.52 0 62 (9.25)
Re SC0 .38
where Re is the Reynolds number = 2u5d/yv; us is the linear sample velocity; v is
the kinematic viscosity of matrix; Sc is the Schmidt number = v/D, and d is the
fibre diameter. The effective thickness of the boundary layer in Eq. (9.25) is a
surrogate (or average) estimate and does not take into account changes of the
thickness that may occur when the flow separates and/or a wake is formed.
Equation (9.25) indicates that the thickness of the boundary layer will decrease
with an increase of the linear sample velocity (Fig. 9.10). Similarly, when the

271
sample temperature (Tg) increases, the kinematic viscosity also increases. Since
the kinematic viscosity term is present in the numerator of Re and in the
denominator of Sc, the overall effect on 6 is small. A reduction of the boundary
layer and an increase of the mass transfer rate for an analyte can be achieved in
at least two ways, i.e., by increasing the sample velocity and by increasing the
sample temperature. However, the temperature increase will reduce the solid
sorbent efficiency. As a result, the sorbent coating may not be able to adsorb all
molecules reaching the surface and therefore stop behaving as a zero sink for all
analytes.

9.3.6 Headspace extraction

Equations (9.21) and (9.23) indicate that the use of the headspace above the sam-
ple as an intermediate phase might be an interesting approach to accelerate
extraction for analytes characterised by high Henry constants. When a thin
extraction phase is used, the initial extraction rate and hence the extraction time
is controlled by the diffusion of analytes through the boundary layer present in
the sample matrix. The presence of a gaseous headspace facilitates rapid trans-
port into the extraction phase because of the high diffusion coefficients. To
increase the transport from the sample matrix into the headspace, the system
can be designed to produce a large sample/headspace interface. This can be
accomplished by using large diameter vials with good agitation, purge or even
spray systems. At room temperature only volatile analytes are transported
through the headspace. For low volatility compounds, heating of the sample is a
good approach, if loss in magnitude of the distribution constant can be accepted.
The ultimate approach is to heat the sample and cool the extraction phase at the
same time. Heating of the sample does not only increase the Henry constant, but
also induces convection of the headspace due to density gradients associated
with temperature gradients present in the system, resulting in higher mass
transport rates. The cooling of the sorbent increases its capacity. Collection of
analytes can be performed in the same vial [32] or can be separated in space simi-
larly as in the purge and trap technique. In the heating-cooling experiments,
both kinetic and thermodynamic factors are addressed simultaneously. This
technique is discussed in more detail in Chapter 13 on Solid Phase Micro-
extraction. Headspace approaches are also interesting since adverse affects asso-
ciated with the presence of solids, oily or high molecular weight interferences,
which can cause fouling of the extraction phase are eliminated.

9.3.7 Passive time weighted average sampling

Consideration of different arrangements of the extraction phase is always bene-

ficial. For example, extension of the boundary layer by a protective shield that
restricts convection would result in time weighted average (TWA) measurement
of analyte concentration (see Eq. (9.21)). Various diffusive samplers have been

272
 z -

0 Z

Fig. 9.11. SPME/TWA approaches based on the in-needle fibre and in-needle coating.

developed based on this principle. For example, when the extracting phase in the
SPME device is not exposed directly to the sample, but is contained in a protec-
tive tubing (needle) without any flow of the sample through it (see Fig. 9.11), the
diffusive transfer of analytes occurs through the static sample (gas phase or
other matrix) trapped in the needle. The system consists of an externally coated
fibre with extraction phase withdrawn into the needle (Fig. 9.11). This geomet-
ric arrangement represent a very powerful method, capable of generating a
response proportional to the integral of the analyte concentration over time and
space (when the needle is moved through space) [33]. In this case, the only mech-
anism of analyte transport to the extracting phase is diffusion through the
matrix contained in the needle. During this process, a linear concentration pro-
file (shown in Fig. 9.11) is established in the tubing between the small needle
opening, characterized by surface area A and the distance Z between the needle
opening and the position of the extracting phase. The amount of analyte
extracted, dn, during time interval, dt, can be calculated by considering Fick's
first law of diffusion [4]

dn = ADmc dt = ADm AC(t) dt (9.26)

dz Z
where AC(t)/Z is an expression of the gradient established in the needle between
the needle opening and the position of the extracting phase, Z; AC(t) = C(t) - C,
where C(t) is a time dependent concentration of analyte in the sample in the
vicinity of the needle opening, and Cz is the concentration of the analyte in the
gas phase in the vicinity of the coating. Cz is close to zero for a high extraction
phase/matrix distribution constant capacity, then: AC(t) = C(t). The concentra-
tion of analyte at the coating position in the needle, Cz, will increase with
integration time, but it will be kept low compared to the sample concentration
because of the presence of the extraction phase. Therefore the accumulated
amount over time can be calculated as:

n AC
f (t)d (27)t)dx

As expected, the extracted amount of analyte is proportional to the integral of

the sample concentration over time, the diffusion coefficient of analyte in the
matrix filling the needle, D,, in the area of the needle opening, A, and inversely
proportional to the distance of the coating position in respect of the needle

273
opening, Z. It should be emphasized that Eqs. (9.24) and (9.25) are valid only in a
situation where the amount of analyte extracted onto the sorbent is a small
fraction (below RSD of the measurement, typically 5%) of the equilibrium
amount in respect to the lowest concentration in the sample. To extend integra-
tion times, the coating can be placed further into the needle (larger Z), the
opening of the needle can be reduced by placing an additional orifice (smaller A),
or a higher capacity sorbent can be used. The first two solutions will result in low
measurement sensitivity. An increase of sorbent capacity presents a more
attractive opportunity. It can be achieved by either increasing the volume of the
coating, or its affinity towards the analyte. An increase of the coating volume
will require an increase of the device size. Therefore the optimum approach to
increased integration time is to use sorbents characterized by large coating/gas
distribution constants. If the matrix filling the needle is other than sample ma-
trix then an appropriate diffusion coefficient and distribution constant should be
used in Eq. (9.27) as discussed below in the case of membrane extraction.

9.3.8 Extraction combined with derivatisation

The capacity of the extraction phase for the difficult to extract analytes such as
polar or ionic species is frequently enhanced by introducing a derivatisation
step. The objective of this approach is frequently not only to convert the native
analytes into less polar derivatives that are extracted more efficiently, but also to
label them for better detection and/or chromatography. The most interesting
implementation of this approach is simultaneous extraction/derivatisation. In
this technique the derivatisation reagent is present in the extraction phase
during the extraction. The major advantage of this approach is that two steps are
integrated into one. There are two limiting cases describing the combination
between extraction and derivatisation. The first occurs when mass transfer to
the fibre is slow compared to the reaction rate. In this case Eq. (9.22), as
discussed above, describes the accumulation rate of analytes, assuming that the
derivative is trapped in the extraction phase.
In the second limiting case the situation is reversed in that the reaction rate
is slow compared to the transport of analytes to the extraction phase. In other
words, at any time during the extraction, the extraction phase is at equilibrium
with the analyte remaining in a well-agitated sample, resulting in uniform
reaction rate throughout the coating. This is a typical case for thinly dispersed
extraction phase, since the equilibration time for well-agitated conditions is very
short compared to a typical reaction rate constant. The accumulation rate of the
product in the extraction phase n/t can then be defined by:
n = VkrKe X C(t)dt (9.28)
where Cs is the initial concentration of analyte in the sample and kr is chemical
reaction rate constant. In the other words, when the sample is of large volume,
such as direct sampling in the field, the reaction and accumulation of analyte in

274
the extraction phase proceeds with the same rate as long as reagent is present in
an excess amount. It is worth noting that the rate is also proportional to the
extraction phase/sample matrix distribution constant. If the concentration
varies during the accumulation, the collected amount corresponds to the
integral over concentration and time, similarly as discussed above in the case of
time weighted average sampling. For limited sample volume, however, the
concentration of analyte in the sample phase decreases with time as it is
partitioned into the coating and converted to trapped product, resulting in a
gradual decrease of the rate. The time required to exhaustively extract analytes
from a limited volume can be estimated using experimental parameters [4].

9.3.9 Membrane extraction techniques

For continuous monitoring applications, membrane extraction is an attractive

approach. Permeation through membrane is a specific extraction process where
the sorption into and desorption out of the extraction phase occurs simulta-
neously. The sample (donor phase) is in contact with one side of the membrane
where extraction into the membrane material occurs, while permeated analytes
are removed by the stripping phase (acceptor). For membrane extraction with
good flow (agitation) conditions at both acceptor and donor sites and efficient
stripping, the rate of mass transport through the membrane is controlled by the
diffusion of analytes through the membrane material. The concentration
gradient which facilitates transport across the membrane is formed by the
difference in analyte concentration between the sample side (KesCs) and the
stripping phase is close to zero for high flow rates of stripping phase (see Fig.
9.12). The mass transfer rate through the membrane, nt, can be estimated at
steady-state conditions using the following equation:
nit = BzADK,,CsC/b (9.29)
where A is the surface area of the membrane, De is the diffusion coefficient in the
membrane material, Kes is the membrane material/sample matrix distribution
constant, b is a thickness of the membrane and B2 is a geometric factor defined
by the shape of the membrane. The permeation rate through the membrane is

Strip

Fig. 9.12. Membrane extraction at good sample agitation and stripping conditions. The terms are
defined in the text.

275
proportional to both the diffusion coefficient (D.) and the distribution constant
(Kes) and inversely proportional to b. De determines the rate of analyte migration
through the membrane and Kes the magnitude of the concentration gradient
generated in the membrane [341. This information can be used to calibrate the
extraction process a priori if these parameters are obtained from tables or
experimental data [35].
The concentration of unknown can be calculated by converting Eq. (9.29):
bn
C, - (9.30)
B2 ADeKe,,t (9

The membrane material/sample matrix distribution constant Kes determines the

sensitivity of membrane extraction (see Eq. (9.29)) indicating that the mem-
brane, although a physical barrier, is also a concentrating medium, analogous to
the extraction phase in other configurations. However, the concentration in the
stripping gas phase are lower compared to the sample since the gradient needs to
exist in the membrane to generate diffusive mass transfer through the mem-
brane material (see Fig. 9.12). Therefore incorporating sorbent after the mem-
brane would allow concentration of the extract and therefore sensitive analysis.
When the membrane is in direct contact with the aqueous phase then the mass
transfer through the boundary layer surrounding the membrane could contrib-
ute to the overall mass transfer in the system. Therefore for analytes character-
ised by high Henry constant it is important to consider headspace membrane
extraction arrangement. The topic of membrane extraction is discussed in detail
in Chapter 14.

9.4 SUMMARY: SIGNIFICANCE OF FUNDAMENTAL

DEVELOPMENTS

Whenever a new type of complex sample is considered, a small research project

must be conducted to find optimum extraction conditions that give the fastest
and most complete release of native analytes from the matrix and their parti-
tioning into extraction phase. Typically, an empirical approach is taken and
several parameters are varied, such as change of chemical properties of the
extraction phase or possibly type of additives used. Better understanding of
analyte-matrix interaction would facilitate a more rational choice of extraction
based on models, when the characteristics of the analyte and the matrix are
known. However, in most cases involving solid matrices, there is still more
research effort required to reach the level of fundamental knowledge to practi-
cally implement such an approach.
Reliance on physicochemical constants in calibration might appear uncon-
ventional or even uncomfortable to some researchers. However, as the theory
indicates these constants define the extraction process, and there is an opportu-
nity to take advantage of this fact. Physicochemical constants can be frequently
estimated from simple experiments or calculated by considering the molecular

276
structures of analytes, extraction phase and matrix, which adds to the attrac-
tiveness of this approach. For equilibrium microextraction techniques the
extraction phase/sample matrix distribution constant is used to quantify the
concentration of analytes in the sample matrix (see Eq. (9.3)). For extraction
approaches controlled by the mass transfer in the boundary layer, calibration
can be based on the diffusion coefficient in the sample matrix at constant extrac-
tion time under well-defined agitation conditions. When the derivatisation reac-
tion kinetics controls the extraction rate, the rate constant can provide a means
of calibration. In some cases, as in membrane extraction, a combination of con-
stants defines the extraction rate and can be used for calibration as well. The
major argument against using this approach is that physicochemical constants
are affected by many experimental parameters, such as temperature and matrix
conditions. However the impact of temperature change can be compensated for.
This can be accomplished by monitoring the temperature and using correction
factors and therefore direct calibration for simple matrices is possible [36]. For
more complex matrices, internal standard or standard addition calibration,
applied routinely in exhaustive techniques to monitor recoveries, can be used to
compensate for matrix variations [37]. In future research, correlations between
distribution constants and simple measurements such as turbidity and pH may
be found to account for small matrix variations and therefore eliminate the need
for internal calibration for samples of known origin.
The advantages of non-exhaustive extraction are its fundamental simplicity
and fewer geometric restrictions, which facilitate a number of interesting on-site
implementations by integrating the sampling and sample preparation steps.
Additionally, more information can be obtained about the investigated system.
For example, it is possible to speciate and determine distribution of analyte in
multiphase systems since the extraction process does not disturb the equilibria
naturally present in the system. Therefore different forms of an analyte are
extracted and quantified according to their corresponding distribution constants
and/or diffusion coefficients.
A better understanding of the fundamentals of the extraction process
facilitates exploration of new opportunities, which makes sample preparation a
more vital and scientifically exciting part of the analytical process. The focus in
this discussion has been on chemical extraction, purposely neglecting extraction
approaches involving mechanical principles, which would introduce an addi-
tional level of complexity, although these would be interesting to explore in
future developments of the technology.

REFERENCES

1 E.M. Thurman and M.S. Mills, Solid Phase Extraction. Wiley, New York, 1998.
2 J. Dean, Extraction Methods for EnvironmentalAnalysis. Wiley, New York, 1998.
3 A. Handley, Extraction Methods in Organic Analysis. Sheffield Academic Press,
Sheffield, 1999.
4 J. Pawliszyn, Solid Phase Microextraction.Wiley-VCH, New York, 1997.

277
 5 B.V. Ioffe and A.G. Vitenberg, Headspace Analysis and Related Methods in Gas
Chromatography.Wiley, New York, 1984.
6 J. Pawliszyn, Chapter 14 in: J. Pawliszyn (ed.), Sampling and Sample Preparation
for Fieldand Laboratory. Fundamentalsand New Directionsin Sample Preparation.
Comprehensive Analytical Chemistry XXXVII. Elsevier, Amsterdam, 2002, pp.
479-502.
7 K. Pratt and J. Pawliszyn, Anal. Chem., 64 (1992) 2101.
8 M. Yang, M. Adams and J. Pawliszyn, Anal. Chem., 68 (1996) 2782.
9 J. Pawliszyn, Trends Anal. Chem., 14 (1995) 113.
10 M. Adams, E. Otu, M. Kozliner, J. Szubra and J. Pawliszyn, Anal. Chem., 67 (1995)
212.
11 T. Gorecki, X. Yu and J. Pawliszyn, Analyst, 124 (1999) 643.
12 F.A.L. Dullien, PorousMedia. Academic Press, San Diego, 1992.
13 C. Horvath and H.J. Lin, J. Chromatogr., 149 (1978) 43.
14 J. Crank, Mathematics of Diffusion. Clarendon Press, Oxford, 1989.
15 J.C. Giddings, Anal. Chem., 35 (1963) 1999.
16 J.A. Cadzow and H.F. van Landingham, Signals, Systems, and Transforms. Prentice
Hall, Englewood Cliffs, NJ, 1985.
17 J. Pawliszyn, J. Chromatogr. Sci., 31 (1993) 31.
18 J. Langenfeld, S. Hawthorne, D. Miller and J. Pawliszyn, Anal. Chem., 67 (1995)
1727.
19 J. Langenfeld, S. Hawthorne, D. Miller and J. Pawliszyn, Anal. Chem., 65 (1993) 338.
20 B.E. Richter, B.A. Jones, J.L. Ezzell, N.L. Porter, N. Avdalovic and C. Pohl, Anal.
Chem., 68 (1996) 1033.
21 J.R.J. Pare, J.M.R. Belanger, K. Li and S.S. Stafford. J. Microcolumn Sep., 7 (1995)
37.
22 N. Alexandrgu and J. Pawliszyn, Anal. Chem., 61 (1989) 2770.
23 Z. Miao, Z. Zhang and J. Pawliszyn, J. Microcolumn Sep., 6 (1994) 459.
24 K.D. Bartle, T. Boddington, A.A. Clifford and N. Cotton, Anal. Chem., 63 (1991)
2371.
25 R. Eisert and J. Pawliszyn, Anal. Chem., 69, (1997) 3140.
26 R. Eisert and J. Pawliszyn, Crit. Rev. Anal. Chem., 27 (1997) 103.
27 A.D. Young, Boundary Layers. BSP Professional Books, Oxford, 1989.
28 F. Augusto, J. Koziel and J. Pawliszyn, Anal. Chem., 73 (2001) 481.
29 K. Sukola, J. Koziel, F. Augusto and J. Pawliszyn, Anal. Chem., 73 (2001) 13.
30 H.S.Carslaw and J.C. Jaeger, Conductionof Heat in Solids. Clarendon Press, Oxford,
1986.
31 J. Koziel, M. Jia and J. Pawliszyn, Anal. Chem., 72 (2000) 5178.
32 Z. Zhang and J. Pawliszyn, Anal. Chem., 67 (1995) 34.
33 M. Chai and J. Pawliszyn, Environ. Sci. Technol., 29 (1995) 693.
34 Y. Luo, M. Adams and J. Pawliszyn, Anal. Chem., 70 (1998) 19.
35 Y. Luo and J. Pawliszyn, Anal. Chem., 72 (2000) 1064.
36 P. Martos and J. Pawliszyn, Anal. Chem., 69 (1997) 206.
37 C. Grote and K. Levsen, in: J. Pawliszyn (ed.), Applications of Solid Phase Micro-
extraction. RSC, Cambridge, 1999.

278
Chapter10

Headspace gas chromatography

Zelda E. Penton

10.1 OVERVIEW

10.1.1 Introduction

Headspace gas chromatography (GC) is a sample preparation method for

determining volatile compounds in solid and liquid samples. The technique has
existed since the late 1950s [1] and is still actively used. The popularity of
headspace analysis is due to its simplicity and the fact that it is a very clean
method of introducing volatile analytes into a GC-the injector system and
column should require virtually no maintenance.
A typical headspace analysis involves a liquid or solid sample in a sealed vial,
containing a gas phase. A portion of the gas phase is removed and introduced
into a GC. The analytes are volatile compounds that are dissolved in a water-
based matrix or in an organic solvent with a high boiling point (>150°C). The
matrix can also be a solid or semi-solid material. The analytes are normally
present at concentrations ranging from parts per billion to low percentage levels.
Some common applications of the headspace technique are determination of
blood alcohol, residual solvents in pharmaceuticals, and monomers in polymers.
However, there are many other applications of headspace GC. Some of these will
be discussed in Section 10.6.

10.1.2 Static versus dynamic headspace

The term "headspace GC" is commonly used to describe static headspace but is
sometimes used to refer to dynamic headspace (purge and trap).
In the static headspace technique (Fig. 10.1), the sample is placed in a vial,
which is sealed and usually heated. The volatile analytes begin to partition
between the sample and the gas phase. Eventually an equilibration state is
reached. At this point, an aliquot of the gas phase is transferred to the GC.
With dynamic headspace (Fig. 10.2), an inert gas is bubbled through the
sample and the volatiles are transferred to an absorbent trap. The trap is heated
and the volatiles are released or desorbed and transferred to the GC. Finally the

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
( 2002 Elsevier Science B.V. All rights reserved
 to vent TRAP Purge gas

carrier gas TRAP to GC column

A B C D

Left: Fig. 10.1. Static headspace sampling. The sample is collected in a container with no space
over the liquid so that volatiles are not lost (A). An aliquot of the sample is placed in a headspace
vial that is sealed with a septum and a crimp-top vial (B). The volatiles in the liquid phase enter
the gas phase until a state of equilibrium is reached (C). An aliquot of the gas phase is removed
with a gas-tight syringe (D) and injected into the GC.
Right: Fig. 10.2. Dynamic headspace sampling or purge and trap. The top diagram is a schematic
of the purge mode. The purge gas (helium or nitrogen) passes through the sample, extracting the
volatiles, which pass onto an adsorbent trap. The lower diagram shows the desorb mode where
the gas flow is reversed and the trap is heated. The desorb gas sweeps the analytes out of the trap
to the GC column for separation. Finally, the trap is heated to a higher temperature, to clean out
residual analytes and water.

trap is heated to a higher temperature than was used during the desorb step, to
remove residual analytes and moisture, and the system is ready for another
sampling.
Purge and trap is generally more sensitive than static headspace, since the
former technique should theoretically remove all of the analytes from the
matrix, while the latter is limited to removal of a single aliquot of the gas phase.
However, a purge and trap system requires more maintenance and is subject to
problems such as foaming of the sample. The two techniques are compared in
Table 10.1. This chapter will focus on static headspace sampling.

10.2 THEORY OF CONVENTIONAL STATIC HEADSPACE

10.2.1 Partition coefficient and phase ratio (P)

In order to develop and optimize headspace methods, the analyst should

understand the effect of two parameters on the sensitivity that will be achieved
in a headspace analysis. These are the partition coefficient and the phase ratio.
These parameters are discussed below.
A sample that is to be analyzed with the headspace technique is usually
collected from the source and refrigerated in a vial with no gas phase over the
sample. The sample is transferred to a headspace vial just prior to analysis.
These vials normally hold a total volume of 5 to 22 ml and are sealed with a
septum and crimped metal cap. The sample is placed in the vial, leaving a

280
TABLE 10.1
Comparison of static and dynamic headspace
Static headspace Dynamic headspace (purge and
trap)
Concentration Parts per billion to low percent levels Parts per trillion to parts per
range million
Hardware Ranges from a simple gas-tight syringe to Complex dedicated instrument
required a sophisticated dedicated instrument.
Automation Can be manual with a gas-tight syringe Purge and trap autosamplers can
but several automated systems can handle handle a maximum of -70
over 100 samples samples
Quantitation Highly matrix-dependent since there is Less matrix-dependent since most
only partial recovery of the analyte of the analyte are removed
Ease of Little maintenance required Labor intensive; traps must be
maintenance cleaned; foaming a problem
Common 1. Blood alcohol 1. Monomers in polymers
applications 2. Residual solvents in pharmaceuticals 2. Flavors in foods
3. Monomers in polymers 3. Volatiles in drinking and
4. Flavors in foods wastewater and soil (United
5. Volatiles in drinking and wastewater States)
(Europe), soil (United States)

significant amount of headspace over the sample; then the vial is sealed and
heated. The analytes (volatile compounds) begin to move into the gas phase
above the sample until a state of equilibrium is reached. At this point, the ratio of
the analyte concentrations in the gas phase and in the liquid or solid phase is a
constant. The constant is defined as the partition coefficient.

Partition coefficient: K = CL (10.1)

Co
The ratio () of the gas phase to the liquid phase is:

= V0 (10.2)
VL
The equilibration process is illustrated in Fig. 10.3 where: CL = concentration in
the liquid phase after equilibration; C, = concentration in the gas phase after
equilibration; Co = original concentration of analyte in the sample; VL = volume

Fig. 10.3. Showing the equilibrium process in static

headspace. In the vial on the left, the analytes are at the
concentration that they were in the original sample. On LiW VLI
the right, the analytes have reached equilibrium
between the gas and liquid phases.

281
-- - A 1 I . I: I I_=I M r =l.C1flnnh
rig. 10.4. Hypotnetlcal curves calculated trom .
Eq. (10.3) for three different analytes with parti- :
tion coefficients of 4, 40 and 400. For all of the 8
compounds, the original concentration of the =-
analyte in the liquid sample was 100 ppb. The
x-axis represents volumes of liquid sample in a '
10-mil vial ranging from 0.5 to 9.5 ml. The y-axis .-
is the concentration of the analyte in the gas o
phase for each volume of liquid sample. Note that
volume of the sample has virtually no effect on E
t1_ L, -1 ---- ,^-
- ---- + - IL, Ri
tile: ll~flubpiCe CVI$UfhtSLldlllUll Wtll tle pitt ta~Lut: E 4 K = 40
coefficient is 400, but sample volume has a sig- 0 ..
nificant effect when the partition coefficient is 4. 0 - -
0.5 2.5 4.5 6.5 8.5
Volume of liquid sample (mL) in a 10-mL vial

of the liquid sample; V, = volume of the gas phase; V = VL + VG (total volume of

the headspace vial).
Equation (10.3), the relationship of the partition coefficient (K) and the
phase ratio () to headspace sensitivity (CG) can be calculated from Eqs. (10.1)
and (10.2):
C
CG O (10.3)
K+P
Equation (10.3) is based on the assumption that the solution is dilute. It is
possible to draw several conclusions from Eq. (10.3):
- The volume of the headspace vial has no effect on sensitivity.
- As K increases, the sensitivity is reduced and D becomes less significant.
- As K decreases, becomes more significant; a small value of A will tend to
increase sensitivity.
Figure 10.4 illustrates these conclusions, showing plots of three compounds with
partition coefficients of 4, 40 and 400. These values roughly correspond to the
partition coefficients of benzene, ethyl acetate and isopropanol in water at 50°C.
Note that the headspace sensitivity of a compound comparable to benzene with a
low partition coefficient shows a large increase as the relative volume of the gas
phase decreases (smaller A). The compound with a partition coefficient of 400
(comparable to isopropanol) shows almost no change in sensitivity as P
decreases.

10.2.2 Parameters affecting the partition coefficient

Some parameters affecting the partition coefficient of a compound in a given

matrix are solubility and temperature. As the solubility of an analyte increases
in a given matrix, the partition coefficient increases and the headspace sensi-
tivity is reduced. In this situation, altering the matrix will have a significant
effect on the partition coefficient. Thus, addition of salt to water will reduce the
solubility of polar compounds in the water and increase the sensitivity of

282
 * No salt
:::Sodium chloride
1 Sodium sulfate
Fig. 10.5. Effect of saturation with salt
on increasing the headspace sensitivity Co
of various compounds in water. Note s
that "salting out" has a significant ,,
effect only on dioxane, the polar
compound where the response was
enhanrrl q-fnld (nff-sarle) when
saturating with potassium carbonate. MeCI2 Benzene TCE Chloroform Toluene Dioxane

headspace analysis. Non-polar compounds with a low partition coefficient in

water will not be affected significantly by the addition of salt (Fig. 10.5).
The partition coefficient is also affected by temperature:
A
In K = -B (10.4)
RT
where K = the partition coefficient, R is the ideal gas constant, A and B are
thermodynamic constants, and T is the absolute temperature (K).
Table 10.2 lists some partition coefficients of organic compounds in water at
different temperatures. Compounds that are more soluble in the matrix
generally show more variation in partition coefficient with temperature.
Compare the changes in the partition coefficients of 1,1,1-trichloromethane and
dioxane in water as the temperature increases.
The temperature of headspace samples normally must be kept constant to
ensure excellent precision. There are a few exceptions, however, when the parti-
tion coefficient varies only slightly with temperature or an internal standard is
present that shows the same variation in partition coefficient as the analyte [2].

TABLE 10.2
Air-water partition coefficients [12] of selected compounds versus temperature
Compound Partition coefficient K
40°C 60°C 80°C
Dioxane 1618 642 288
Ethanol 1355 511 216
Ethyl acetate 62.4 29.3 17.5
Benzene not available 2.27 1.66
Toluene 2.82 1.77 1.27
o-Xylene 2.44 1.31 0.99
Dichloromethane 5.65 3.31 2.07
1,1,1-Trichloromethane 1.65 1.47 1.18
Tetrachloroethylene 1.48 1.27 0.87
Cyclohexane 0.08 0.05 0.02

283
10.3 OPTIMIZATION OF HEADSPACE RESPONSE

10.3.1 Decrease the partition coefficient of analytes in the matrix

As discussed above, a typical approach to decreasing the partition coefficient is to

increase the temperature of the sample. However, problems can arise when a
headspace vial is heated to a temperature approaching the boiling point of the
matrix. These include leaks due to increasing pressure in the vial, decomposition
of the sample and interference from the matrix solvent.
Altering the matrix to reduce the solubility of the analytes can also decrease
the partition coefficient. "Salting out", as mentioned above, is an effective
method for driving polar analytes from biological fluids and soil. It is important
to saturate the sample with salt to maximize the effect of the salt and also to
avoid variations in salt concentrations from sample to sample, which will affect
the relative responses of the analytes.
It is sometimes useful to change the pH of the matrix to maximize
response-i.e. lower the pH to convert volatile acids to the molecular state or
raise the pH to determine amines in the sample.

10.3.2 Mixing

Most commercially available headspace systems have the ability to mix the
sample during the equilibration period. Mixing tends to reduce the time required
to achieve equilibration [3]. This brings up the question of sampling the head-
space before reaching equilibrium. The main advantage of sampling the head-
space after the analytes have reached full equilibrium between the matrix and
the gas phase, is to achieve maximum sensitivity and precision. In many
laboratories, speed of analysis is more important than maximizing sensitivity
and precision. An example of an application where the GC run time is much
shorter than the time necessary to reach equilibrium is the determination of a
residual monomer in a polymer. In the case of volatiles in a solid, equilibrium
may take hours or even days to achieve. The only practical solution may be to
sample the headspace before equilibrium is attained, and to heat all of the
samples for the same length of time prior to analysis, to assure comparability.
When headspace GC is used in thermodynamic studies such as the
determination of activity coefficients [4-6], it is important to sample after
equilibrium has been reached.

10.3.3 Injection volume

In most headspace GC analyses, fused silica columns with inner diameters

ranging from 0.25 to 0.53 mm are utilized. A film thickness of 0.5 micron or
greater is often selected to increase the capacity of the GC column. Nevertheless,
these columns have a limited ability to focus very volatile compounds such as
freons at temperatures of 35°C and above. Therefore, a smaller injection volume

284
 200 pL J

Fig. 10.6. GC/MS chromatograms of volatile halogenated hydro- 500,L

carbons in the headspace over water. Note that the chromato-
K
graphy improves as the injection volume is reduced. Although, a
high capacity fused-silica column was used to help focus the volatile 1000 pL _ ,J
compounds, a cryogenic trap would have been useful for this 4
sample. Time (min)

is desirable to achieve narrow early-eluting peaks. The default injection volume

in many commercial headspace systems is usually 1 ml. This volume is often too
large for optimum chromatographic results. When developing a headspace meth-
od, it is a good idea to use as small an injection volume as is consistent with the
sensitivity requirements (Fig. 10.6). If it is necessary to inject volumes of 1 ml or
greater, a cryogenic oven or a cold trap should be used to focus the analytes on
the column.

10.3.4 Vial size

As discussed in Section 10.2.1, the sensitivity is determined by the partition

coefficient and the phase ratio and is independent of the vial size. However,
there are advantages to large vials. It is easier to add solid samples to a 20-ml vial
than to a 2-ml vial and if a relatively large sample is added to a vial, it is more
likely to be representative of the material to be analyzed.

10.3.5 Multiple sampling from a headspace vial

In a typical headspace analysis, the vial is sampled only once and then discarded.
After one sampling, some of the analyte is removed from the headspace and after
a new equilibrium occurs, there is less analyte in the gas phase. Therefore,
unlike the situation with liquid injection, only the first injection represents the
original sample.
There is a technique that utilizes multiple sampling from a single vial.
Multiple headspace extraction (MHE) was proposed in 1970 by Suzuki et al. [7]
then modified by Kolb and Ettre [8]. With MHE, a liquid or solid sample is sealed
in a headspace vial and sampled repeatedly at equal time intervals. It is assumed
that the concentration of volatiles under these conditions will decay exponen-
tially (Fig. 10.7). If an infinite number of extractions is carried out, the volatiles
will be completely removed from the vial. The total area count of the peaks
corresponding to the analyte should be equal to the sum of the areas from each
individual extraction. MHE is normally used with solid samples but it can also be
applied to liquids. The technique is useful only if the partition coefficient
between the matrix and the analyte is small, so that a substantial quantity is
removed at each extraction.

285
Fig. 10.7. FID chromatograms of vinyl chloride injected four times from the same vial in a
multiple headspace extraction analysis. The vial contained a polyvinylchloride polymer.

W
0
To
0
0.
o

a
U.

0 4 8 12 16
Gasoline (ppm)
Fig. 10.8. An illustration of the matrix effect on recovery of gasoline from sand, soil and water
by headspace GC.

10.4 QUANTITATION IN HEADSPACE

10.4.1 Quantitation techniques

Headspace GC differs from most other analytical chemistry techniques in that

recovery is almost never near 100%-sometimes it is well under 1%! In spite of
this, it is possible to obtain extremely reproducible and accurate results with
headspace. The main difficulty in achieving good quantitation with headspace is
in matching the matrix of the samples and standards. I described the signifi-
cance of the partition coefficient in determining headspace response. Slight
differences in the matrix, such as content of organic material in soil samples
(Fig. 10.8) or protein or fat in biological fluids, differences in ionic strength and
pH will affect headspace response. Some approaches to overcoming the matrix
effect are discussed below.

286
10.4.1.1 Using a blank that exactly matches the sample matrix
This method could be used for the analysis of organic volatiles in drinking water.
It is not difficult to prepare a solution of water that does not contain measurable
volatile organics. The water is spiked with volatile organics of known concentra-
tion and a calibration curve is prepared.

10.4.1.2 'Dilute the samples and standardswith a matrix modifier

This method is used in determining blood alcohol (Section 10.6.1). It is also the
basis of the United States Environmental Protection Agency (USEPA) method
5021 [9] for the analysis of volatile organics in soil (Section 10.6.3).

10.4.1.3 Spike the sample with known increments of analyte

This technique, known as "standard additions", is useful when it is not possible
to find a blank matrix and the sensitivity requirements of the analysis do not
permit dilution with a matrix modifier. An example would be the determination
of a trace flavor compound in wine such as linalool. Three to five aliquots of a
wine samples are spiked with increasing quantities of linalool. The spiked
samples and an unspiked sample are analyzed. The quantity spiked is plotted
versus the response and the resulting calibration curve is extrapolated to the x
intercept to determine the original quantity of the analyte in the sample. Figure
10.9 illustrates the use of standard additions for the determination of ethyl
acetate in an industrial solution. When using this calibration method, the
analyst should verify that the analyte is not bonded to components in the matrix.

10.4.1.4 Use the multiple headspace extraction technique

MHE minimizes the matrix effect by essentially imitating dynamic headspace or
purge and trap. Kolb and Ettre [8] derived a simplified method of quantitation
for MHE where the standards and samples can be in quite different matrices.
Ethylene oxide was determined in surgical silk sutures. The silk sample was
placed in a headspace vial and sealed. Ethylene oxide was injected into empty
headspace vials to form the calibrators. Kolb showed that when an analyte is
sampled several times from a headspace vial, the total area from all of the

Quantity in original m
sample 0
X = 0.5 mg/mL

-0.750 -0.375 0.375 0.750 1.125

Ethyl acetate spiked (mg/mL)

Fig. 10.9. Standard additions quantitation in the headspace analysis of ethyl acetate in an
industrial mixture. The detector responses were determined for the unspiked sample and for the
sample spiked with increasing quantities of ethyl acetate. The resulting curve was back-
extrapolated to the x-axis to determine the quantity of ethyl acetate in the original sample.

287
 -- _ sample
o
a
0o ar
-
z

1 2 3 4
Run number
Fig. 10.10. Calibration curves for determining vinyl chloride in polyvinylchloride with multiple
headspace extraction.

samplings can be represented by Eq. (10.5):

E, Ai
A A k
(10.5)
1-e
YCA is the total area count, A, is the area from the first sampling, and k is the
slope of the plot obtained by plotting the natural logarithm of the area counts
versus the number of the sampling. Calibration curves for MHE samplings are
shown in Fig. 10.10.
After validating the method by demonstrating a linear response, a simplified
form of Eq. (10.5) can be used, which requires only two samplings:

CAi (A'A) - (10.6)

YlAi is the total area count, A1 is the area from the first sampling, and A2 is the
area from the second sampling. While results would be expected to be more
accurate with more than two samplings, Eq. (10.6) is practical for routine
analysis. The area counts in the calibration sample would also be determined
using Eq. (10.6). Finally, the mass of the analyte would be calculated with Eq.
(10.7).

W-mp = (Amp )(1W4t ) (10.7)

Astd

Wsp is the weight of the analyte in the sample, Am,,p is the total area of the
analyte in the sample as calculated in Eq. (10.6), WStd is the weight of the analyte
in the standards vial, and AId is the total area of the analyte in the standard vial
as calculated in Eq. (10.6).

288
10.4.1.5 Use the full evaporative technique
The full evaporative technique (FET) [10] is an approach for eliminating the
matrix effect in headspace GC, but it is not really an equilibrium headspace
method. With FET, a very small liquid sample (a few microliters) is injected into
an empty headspace vial and the vial is heated until the entire sample is in the
vapor state. Then a sample of the gas phase in the vial is injected into the GC.
Since there is no equilibrium between two phases, there is no matrix effect. The
sensitivity that one can obtain with this method depends on the partition coeffi-
cient of the analyte in the original sample. For example with dioxane in water at
40°C (K = 1618), there is almost no change in sensitivity with FET if one com-
pares the sensitivity from a 10-g sample in a 20-ml headspace vial to the sensi-
tivity when a 10-ml sample is in the same 20-ml vial, However with o-xylene in
water at 40°C (K = 2.44), the sensitivity with FET is lower by a factor of over
100.

10.4.2 Internal standards in headspace quantitation

The use of internal standards in headspace is not as common as in liquid

injection. As is the case in liquid injection, if there is a complex mixture of
analytes, the internal standards may co-elute with some of the compounds of
interest. It can also be difficult to find a compound that has a partition coeffi-
cient which closely matches several analytes. In a simple analysis such as blood
alcohol determination, an internal standard is normally used.

10.4.3 Problems with quantitation

The most frequent cause of problems with quantitation in headspace is due to

loss of volatiles during sample handling. This includes losses during preparation
of standards. It is important to fill the container to the top when collecting
samples, store samples in the refrigerator and analyze them as soon as possible.
Standards should be prepared and stored with equal care. Occasionally, analysts
spike solvents with levels of standards that exceed the solubility of these com-
pounds in the matrix. This is another error that must be avoided to obtain valid
results.
Other problems such as low response or no response are usually due to
improper sealing of the vials. Some vials may leak only at higher temperatures.
With automated systems, where the sample is removed from the vial by
puncturing the septum with a needle, the needle position may be set too high, so
that the septum is not fully punctured.

10.5 HEADSPACE HARDWARE

This section gives a brief description of the various types of apparatus that are
commercially available.

289
10.5.1 Manual headspace

The simplest method of headspace analysis is to sample the vials with a manual
gas-tight syringe. The vials may be unheated or heated in a temperature bath.
Manual headspace is practical if the lab does not have many samples and the
samples contain volatile compounds with low partition coefficients. This type of
sample does not require precise temperature control and there is no danger of
condensation in an unheated syringe. Manual headspace is an inexpensive way
for a laboratory to evaluate the headspace technique for a particular application.

10.5.2 Automated headspace

Dedicated headspace systems should have the following features: constant

heating time for each sample; heating of the next sample in a sequence during
the GC run of the current sample, so that there is no wasted time between runs;
an inert sample path; variable injection volume; mixing; automated method
optimization (the ability to download different methods to study the effect of
varying parameters such as sample temperature or equilibration time); ability to
handle a large number of samples, if required.
10.5.2.1 Syringe-based systems
A syringe-based automated system can be as simple as a modified liquid auto-
sampler that has been equipped with a gas-tight syringe [2,11]. These systems
perform well for some applications, but the sample vials and the syringe are
unheated; they are therefore inadequate for analyzing solid samples or relatively
high-boiling analytes such as diesel fuels. There are also dedicated headspace
systems that use a heated syringe and have all of the desirable features listed
above.
10.5.2.2 Valve and loop-based systems
Figure 10.11 is a schematic of a valve and loop-based headspace autosampler.
These systems are equipped with a 6-port valve, usually a 1-ml sample loop and a
transfer line that connects to the GC column. (With some of these systems, the
sample loop is accessible and can be easily changed to a different size.) This can
be a direct connection where the GC injector is bypassed. The GC column can
also be installed in an injector in the usual way, and the transfer line can replace
the carrier gas inlet to the injector. A third possibility involves a needle at the
end of the transfer line, which is inserted through the injector septum. The steps
in introducing the sample are: (1) The vial is heated for the length of time
specified by the user, then the vial is pierced with a needle and pressurized with
an inert gas. (2) The valve is turned so that the flow of gas changes direction, and
a portion of the headspace flows into the sample loop. (3) The valve is turned
again so that the gas in the sample loop is flushed through the transfer line and
into the GC. (4) The system returns to the standby position where clean inert gas
flushes the sample loop and the transfer line.

290
 Transfer line to GC
.;'Sp Heated zone

Gasin Sarnpe v A

IA (19) Vial in heated

carrousel

Restrictor
Gas
Gas
w inb Pressurize
alve ^ Vent valve

to vent

Fig. 10.11. Schematic of a valve-loop automated headspace system. The valve, sample loop,
tubing that leads to the needle that penetrates the sample vial, and the transfer line that leads to
the GC, are heated. Note that there are two sources of carrier gas. The sequence of events is
described in the text. The "inject step" is shown here.

10.5.2.3 Pressure-balancedheadspace sampling

Pressure-balanced headspace sampling was invented and patented by
Perkin-Elmer Instruments. Systems using this type of headspace sampling are
available only from this company. The technique is quite different from other
headspace sampling methods, in that the quantity of headspace gas that is
introduced into the GC column is controlled by the pressure in the headspace
vial and the sampling time, rather than by volume.
Pressure-balanced headspace sampling can best be understood by referring
to the schematic in Fig. 10.12A where the system is in standby while the sample
in the vial reaches equilibrium. After reaching equilibrium, a rotating needle
penetrates the headspace vial (Fig. 10.12B) and pressurizes it with carrier gas
until the pressure is equal to the carrier gas inlet pressure of the column. In the
sampling stage (Fig. 10.12C), the solenoid valve closes, interrupting the flow of
carrier gas and the pressurized headspace in the vial, flows through the column
for the specified injection time. This technique has been modified over the years
[12] so that the GC injector can maintain its original source of carrier gas flow
and the headspace system has an independent source of carrier gas.
All three types of dedicated headspace apparatus described here, should
perform adequately for most applications. Table 10.3 summarizes some of the
advantages and disadvantages of these systems.

10.6 HEADSPACE APPLICATIONS

There are literally hundreds of headspace GC applications in the literature. This

section will briefly cover the most common applications and the scope of
applications where headspace is used. Some recent interesting applications will
also be described.

291
 SAMUNG

pj

r
Fig. 10.12. Schematic of the Perkin-Elmer pressure-balanced headspace system. The sequence of
events is described in the text.

TABLE 10.3
Summary of advantage and potential disadvantages with dedicated automated headspace
systems
Automated Advantages Disadvantages
headspace
technique
Syringe- Inert sample path. GC injector Gas-tight syringes have Teflon-tipped
based accessible for trouble-shooting in plungers which can leak (once they are used
headspace and for liquid injection. at a given temperature, the syringe should
Simple to vary injector volume not be used at a lower temperature).
Occasional needle bending.

Valve and GC injector usually accessible for Activity in the sample path has been a
loop trouble-shooting in headspace and problem. Ability to change sample loop (and
for liquid injection. Usually easy to sample volume) is limited in some systems.
move from one GC to another.

Pressure- Inert sample path. Simple system GC injector usually not accessible.
balanced with minimum of moving parts. Trouble-shooting difficult. Works best with a
column that requires a high back pressure.

10.6.1 Volatiles in biological fluids

Determination of ethanol in the blood of suspected drunk drivers in one of the

most important applications of headspace GC. An internal standard is almost
always used-most frequently n-propanol but other compounds including
tert-butanol, acetonitrile, methyl ethyl ketone and isoamyl alcohol have also

292
been reported [13]. There is a significant matrix effect when comparing the
response of ethanol in blood to the response in water or urine. To minimize
differences from sample to sample and from samples to standards, Penton [2]
diluted the samples and standards four-fold with a solution that was saturated
with sodium chloride and also contained n-propanol internal standard.
In addition to ethanol, there are headspace methods for numerous other
solvents in biological fluids. Sharp [14] developed a method for screening for
more than 40 organic compounds in biological matrices, using a two-column
system with FID detection and MS confirmation. Matrix effects were eliminated
with a 15-fold dilution. Frequent findings in samples were ethanol, toluene and
ethyl ether. Russo and Campanella [15] used headspace to determine benzene at
a level of 10 ng/g in small tissue samples (50-70 mg).

10.6.2 Pharmaceutical and other industrial applications

10.6.2.1 Residual solvents in pharmaceuticals

Apart from ethanol in blood, the other major application for headspace is the
determination of residual solvents in pharmaceuticals. Analytical procedures
are published by the European Pharmacopoeia and the United States
Pharmacopeia (USP). USP Method 467 [16] lists five compounds at specified
levels (Table 10.4); in practice, laboratories screen for many additional solvents,
including methanol, ethanol, acetone, toluene and pyridine [17,18].
In Method 467, the pharmaceutical compound is dissolved in water and there
is an external standard comparison of the headspace response of the various sol-
vents with the response of the same solvent in an aqueous standard. USP
method 467 refers to water-soluble pharmaceuticals but many pharmaceutical
compounds are insoluble in water. The practice in many laboratories is to dis-
solve water-insoluble pharmaceuticals in high-boiling solvents such as N,N-di-
methyl formamide (DMF) and N,N-dimethyl acetamide (DMA). Hong and
Altorfer [18] compared water as a dissolution medium with DMF, DMA and DMI
(1,3-dimethyl-2-imidazolidinone). They compared the headspace responses and
precision of 14 compounds in the four dissolution media and concluded that
while most of the compounds exhibited the best headspace response in water,

TABLE 10.4
Limits* on residual solvents in pharmaceuticals (United States Pharmacopeia Method 467)
Solvent Limit (g/g)
Dichloromethane 600
Benzene 2
Trichloroethylene 80
Chloroform 60
1,4-Dioxane 380
*As of 2000, benzene removed later.

293
DMI was a better dissolution medium for hydrophobic drugs than DMF or DMA.
Later, the same authors [19] developed a micro method for this analysis. The
sample size was reduced from 100-200 mg to 5-30 mg. The equilibration time
was reduced from 45-60 to 5-10 min.

10.6.2.2 Food andpackaging

The food and packaging industries use headspace GC to characterize flavors
[20-22], to look for decomposition products [23] and to detect compounds that
contribute to an "off" taste from packaging materials [24,25]. Kaipainen et al.
[261 looked at the decomposition of sugars with headspace and compared GC/MS
detection with an electronic nose. Pattern-recognition techniques are important
in the food industry; with this technique, individual compounds are not
identified; the analyst looks for changes in a headspace chromatogram as a
quality assurance technique.

10.6.2.3 Volatiles in high-boilingoils

Headspace GC is used not only for determining volatile organic solvents but also
for permanent gases. Jalbert et al. [27] measured dissolved gases (H2, N2, 02, CO,
CO 2, methane, ethane, ethylene, acetylene, propane) and traces of water in
mineral insulating oils. The authors separated the analytes with a two-channel
system. The first channel consisted of two columns for separating the perma-
nent gases; the second channel separated the water with an ethylene glycol
phase. A helium photoionization pulsed discharge detector was used in this
work.
Levermore et al. [28] published an interesting study in which they identified
certain "signature" compounds in engine oil that indicated when it was time to
change the oil. Some of these compounds were identified by GC/MS as acet-
aldehyde, acetone, butanal, benzaldehyde, and acetic and benzoic acids.

10.6.3 Environmental methods

Volatile organic compounds in drinking water and wastewater are determined

with static headspace or purge and trap, depending on the country. In Europe,
static headspace methods tend to dominate, while most of the methods in the
United States use purge and trap. USEPA Method 5021 [9] is a static headspace
method for determining volatile organic compounds in soils and other solid
matrices. This method covers a wide range of aromatics and halogenated
hydrocarbons. A matrix modifying solution is used to minimize the differences
between samples and decrease the partition coefficients of the analytes. This
consists of water, saturated with sodium chloride, and adjusted to a pH of 2.0
with phosphoric acid. The matrix modifier is spiked with volatiles to prepare
calibration standards. Unspiked matrix modifier is mixed with the soil samples
as a diluent to minimize matrix differences (10 ml matrix modifier to 2 g of soil).

294
 One would not normally think of headspace as a technique for determining
pesticides, but Royer et al. [29] proposed an automated headspace method for
determining dithiocarbamates in plant matrixes. The method was suggested as a
replacement for the spectrophotometric method, European Norm EN 12396-1
(1996), and the manual headspace method, EN 12396-2 (1999). The new method
offers several advantages including greater sample throughput and the need for
fewer reagents. The authors mixed 2 g of plant matrix with 6 ml of a stannous
chloride/HC1 solution and 4 ml of a diethanolamine solution in a headspace vial;
the vial was then heated to 90°C. The dithiocarbamates released carbon
disulfide, which was determined in the headspace. Recoveries from potato, onion
and currants varied from 85 to 103%. The detection limit was 0.05 mg/kg.

10.7 HEADSPACE GC AS A TOOL FOR DETERMINING

THERMODYNAMIC CONSTANTS

Headspace GC is more than an analytical tool; it has been used for determining
thermodynamic constants including partition coefficients [30-32], activity
coefficients [4-6] and Henry's constant [33,34]. An example follows: Kolb et al.
[31] determined the partition coefficients of 12 compounds in water at tempera-
tures ranging from 40 to 80°C with the vapor phase calibration technique. For
calibration, a known quantity of analyte was injected into an empty vial and
allowed to completely vaporize. An aliquot of the vapor phase was injected into
the GC and the area count of the resulting peak was measured. Then a known
quantity of the analyte was injected into a vial containing the matrix, and the
vial was kept at a given temperature until equilibration was reached between the
matrix and the headspace. After equilibration occurred, a portion of the head-
space, equal to the volume injected in the calibration step, was injected into the
GC. The area count of the resulting peak was compared to the area count that
was obtained in the calibration step and the partition coefficient was calculated.
Some of the data from this work is shown in Table 10.2. There was excellent
agreement with partition coefficients derived from other sources.

10.8 CONCLUSION

In this era of rapid technological changes, headspace GC has survived for several
decades as a useful analytical tool for determining volatiles in liquids and solids.
The potential user can determine if the technique is feasible for a given
analytical problem by examining a vast body of literature, and by a few simple
experiments using a GC, some vials and a gas-tight syringe.

Acknowledgements

Figures 10.1-10.11 are printed with the permission of Varian Inc. Figure 10.12 is
printed with the permission of Perkin-Elmer Instruments.

295
REFERENCES

1 B.V. Ioffe and A.G. Vitenberg, Headspace Analysis and Related Methods in Gas
Chromatography.Wiley-Interscience, New York, 1984.
2 Z. Penton, Clin. Chem., 31 (1985) 439.
3 Z. Penton, J. High Resolut. Chromatogr., 15 (1992) 834.
4 N. Asprion, H. Hasse and G. Maurer, J. Chem. Eng. Data, 43 (1998) 74.
5 C.B. Castells, D.I. Eikens and P.W. Carr, J. Chem. Eng. Data, 45 (2000) 369.
6 C.B. Castells, D.I. Eikens and P.W. Carr, J. Chem. Eng. Data, 45 (2000) 376.
7 M. Suzuki, S. Tsuge and T. Takeuchi, Anal. Chem., 42 (1970) 1705.
8 B. Kolb and L. Ettre, Chromatographia,32 (1991) 505.
9 Test Methods for Evaluating Solid Waste Physical/Chemical Methods, SW-846, Vol.
IB, Chap. 4, Sec 4.2.1, U.S. Environmental Protection Agency, Cincinnati, OH,
December 1996.
10 M. Markelov and J. Guzowski, Anal. Chim. Acta, 276 (1993) 235.
11 Z. Penton, J. High Res. Chromatogr., 17 (1994) 647.
12 B. Kolb and L. Ettre, Static Head-space Gas Chromatography: Theory and Practice.
Wiley-VCH, New York, 1997.
13 F. Tagliaro, G. Lubli, S. Ghielmi, D. Franchi and M. Marigo, J. Chromatogr. B, 580
(1992) 161.
14 M.E. Sharp, J. Anal. Toxicol., 25 (2001) 631.
15 M.V. Russo and L. Campanella, Anal. Lett., 34 (2001) 883.
16 USP 24/NF 19, United States Pharmacopeial Convention, Rockville, MD, 1999, 1877.
17 R.B. George and P.D. Wright, Anal. Chem., 69 (1997) 2221.
18 L. Hong and H.R. Altorfer, Pharm. Acta Helv., 72 (1997) 95.
19 L. Hong and H. Altorfer, Chromatographia,53 (2001) 76.
20 L. Alonso and M.J. Fraga, J. Chromatogr.Sci., 39 (2001) 297.
21 E. Fernandez-Garcia, J. Agric. Food Chem., 44 (1996) 1833.
22 F. Mathieu, C. Malosse, A. Cain and B. Frerot, J. High Resolut. Chromatogr., 19
(1996) 298.
23 I. Medina, M.T. Satue-Gracia and E.N. Frankel, J. Am. Oil Chem. Soc., 76 (1999)
231.
24 L. Du, Q. Xu and A. Lin, Fenxi Ceshi Xuebao, 18 (1999) 59.
25 G. Forsgren, H. Frisell and B. Ericsson, Nord. Pulp Pap. Res. J., 14 (1999) 5.
26 A. Kaipainen, S. Ylisuutari, Q. Lucas and L. Moy, Int. Sugar J., 99 (1997) 403.
27 J. Jalbert, R. Gilbert and P. T6treault, Anal. Chem., 73 (2001) 3382.
28 D.M. Levermore, M. Josowicz, W.S. Rees and J. Janata, Anal. Chem., 73 (2001) 1361.
29 A. Royer, M. Menand, A. Grimault and P.Y. Communal, J. Agric. Food Chem., 49
(2001) 2152.
30 T.M, Cummins, G.A. Robbins, B.J. Henebry, R.C. Goad, E.J. Gilbert, M.E. Miller and
J.D. Stuart, Environ. Sci. Technol., 35 (2001) 1202.
31 B. Kolb, C. Welter and C. Bichler, Chromatographia,34 (1992) 235.
32 A. Chaintreau, A. Grade and R. Munoz-Box, Anal. Chem., 67 (1995) 3300.
33 X. S. Chai and J.Y. Zhu, Anal. Chem., 70 (1998) 3481.
34 J. Peng and A. Wan, Environ. Sci. Technol., 31 (1997) 2998.

296
Chapter 11

Liquid-liquid extraction
Frederick F. Cantwell and Manon Losier

11.1 INTRODUCTION TO LIQUID-LIQUID EXTRACTION FOR

SAMPLE PREPARATION

Liquid-liquid extraction (LLE), or solvent extraction, is one of the oldest and

most widely used techniques in the preparation of samples for qualitative and
quantitative analysis. It involves the distribution of sample components
between two immiscible liquid phases. In its most classic form it is performed in
a separatory funnel by agitating the mixture to disperse drops of one liquid in
the other, discontinuing agitation to allow drop coalescence, and then separating
the bulk liquid phases from one another [1,2].
In recent decades, because of practical problems associated with the
separatory funnel mode of operation, LLE has been significantly replaced in
sample preparation by other types of batch-phase distribution techniques, such
as solid phase extraction (SPE) and solid phase microextraction (SPME) [3].
However, more recently, a variety of nonclassical modes of performing LLE have
been described which circumvent the practical problems and beg a critical
re-evaluation of the role of LLE in sample preparation. This chapter is intended
to summarize the principles underlying the modern use of LLE in sample
preparation. It is not intended as a literature review.

11.1.1 Goals of performing LLE

As for any sample preparation technique, sample clean-up and/or analyte-

component preconcentration are the most common goals of LLE. The first
requires a high selectivity for partitioning of the analyte component over that of
potential interferents; while the second is favored by a high distribution ratio, so
that the analyte component can be extracted from a large volume of sample
solution into a small volume of extractant.
Because the extractant phase is a liquid, it is often possible to perform the
quantitative measurement step directly on the extract, which may or may not
contain a reagent that reacts with the analyte component to produce a detectable
species. Thus, as examples, "extraction photometry" of inorganic and organic
ions as complexes [4-6] was one of the most frequent goals of LLE in years past,
ComprehensiveAnalytical Chemistry XXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 297
and extraction of drugs followed by direct injection of the organic extract into a
gas chromatograph is still a popular analytical method.

11.1.2 Terminology of LLE in sample preparation [1,2]

There are at least two bulk-liquid phases: The aqueous phase, which initially
contains the dissolved sample, and the organic extractant phase. When the
mixture is agitated, either may become the dispersed phase and the other the
continuous phase, depending on a variety of conditions. Agitation produces not
only a dispersion of one phase as drops, but also convection or stirring in both
phases. Up to a point, more vigorous agitation produces both smaller drops of the
dispersed phase, with a concomitant increase in interfacial area between the two
phases, and more rapid convective mixing within both bulk phases.
The movement of a chemical species from one bulk phase to another is the
result of a thermodynamic driving force in the form of a difference in chemical
potential for a neutral species, or in electrochemical potential for an ionic
species. The movement continues until phase distribution equilibrium is
reached, where the ratio of the concentration of a species i in the organic phase
(C,,) to that in the aqueous phase (Ci,,) is given by the equilibrium distribution
coefficient or partition coefficient (i):

Co
Ki =- (11.1)

The mechanism by which species i moves from one phase to the other is
convective-diffusive mass transfer, which is generally faster for more rapid
convection.
The equilibrium distribution isotherm for i is a plot of Ci,o vs. Ci,. In the
range of concentrations, starting at zero, over which the isotherm is linear, l is
constant. The maximum concentration in this range is sometimes called the
linear sample capacity.
Surface-active species can become adsorbed at the liquid-liquid (LL)
interface in addition to being dissolved in the bulk liquid phase. Depending upon
how tightly packed the adsorbed species can become and on whether multi-
layers can be formed, the adsorbed film can have the properties of a two-
dimensional gas, or liquid, or solid. However, adsorption of sample species at the
LL interface is not very common, and most LLE systems involve only
partitioning between bulk liquid phases.
A sample species may be involved in secondary chemical equilibria with
other dissolved components that are either present in the sample or added as
reagents, in either or both the aqueous or organic bulk phases. In such cases the
sample component (I) may be present as more than one species in these phases so
that the ratio of the sum of its species concentrations in the organic phase to the
sum of its species concentrations in the aqueous phase, or the distribution ratio
D,, is a useful quantity to define.

298
 C '
D = - C(i-containing)'o
L (11.2)

C, C(i-containing),a

For systems with secondary chemical equilibria present, the isotherm is a plot of
C,. vs. C, , for which D1 is constant in its linear region. D1 can be expressed in
terms of the c's for the various species. A very common and simple example is
that of a basic sample species, B, which is present as B in the organic phase and
as both B and its conjugate acid BH+ in the aqueous phase, for which:

D CB O KBKEH (11.3)
CB, +CBHa [H30] + Ka,BH

where K,,BH is the acid ionization constant of the species BH + in the aqueous
phase. If Ka,H were equal to zero, D = cB. Quantitatively, the fraction of I
extracted from a volume Va of aqueous phase into a volume V of organic phase,
f,,, at equilibrium is given by the expression:

ID-.Vo
moles I in o V, ..
V at equilibrium. (11.4)
totalmolesI i+Dl.V°
vs
The extent of preconcentration of I can be expressed by the preconcentration
enrichment factor, EF. At equilibrium [7]:
C D
EF I = C, 1- D ,at equilibrium. (11.5)
Clainitia 1+ D· °
Va

Equation (11.5) shows that the maximum value of EF, which is achieved when
D is very large, is equal to the inverse phase ratio, VJVo.
The extent of sample clean-up with respect to a particular component J in
the sample, which interferes with the quantification of analyte component I, is
related to the selectivity of the extraction. This can be expressed as the extent of
separation of I from J in the extract, [8]:

5 = f,o -J,O (11.6)

From Eqs. (11.4) and (11.6) it is evident that a large value of D, and a small value
of Dj favors separation of I from J.
Operationally, if it is desired, the forward extraction of a sample component
from aqueous to organic phase can be followed by a back extraction from the
organic phase into a different aqueous phase. Continuing the above example of
base B, the forward extraction of this component into the organic phase is
favored by high pH (i.e. low [H,0+]) and the back extraction is favored by low pH
(i.e. high [H30+]).

299
11.1.3 Advantages of LLE

Compared to other batch-type phase-distribution techniques, LLE offers large

linear sample capacities. This is because its full bulk-phase volume is available so
that competition for space in the extract phase, which would cause a decrease in
D, is minimal. Thus, DI is constant at high concentrations of analyte component
I in the extract, but even more importantly, it is constant in the presence of high
concentrations of non-analyte sample components in the extract.
A second advantage is the fact that the organic extract itself can be directly
subjected to the quantitative analytical measurement step such as gas or liquid
chromatography. It is not necessary to de-sorb the analyte from the extractphase.
Additionally, the fact that a fresh portion of extracting solvent is used for each
sample means that "carryover" and other "memory" effects are negligible.
On the other hand, back extraction, in which the analyte component is
removed from the organic extract phase, can intentionally be used to add extra
selectivity or an extra degree of preconcentration of the analyte component and,
furthermore, back extraction can be performed simultaneously with forward
extraction [9,10].
A final advantage of LLE derives from the vast body of solvent extraction
literature, accumulated over many decades, which provides information on
choice of organic solvent, of pH and of type and concentration of reagents (e.g.
metal-chelation agents) and how these choices affect both the selectivity, which
is necessary for sample clean-up, and the quantitativeness of extraction, which is
necessary for preconcentration of the analyte component.

11.1.4 Problems with the classical approach

When LLE is performed in a separatory funnel it is necessary agitatively to

disperse drops of one phase in the other phase in order to facilitate mass
transport, and then to allow coalescence [11,12] and phase separation. This
mode of operation introduces some practical problems. For one thing, it is not
possible to work with phase ratios (Vo/Va) smaller than about 0.01, which
militates against a high preconcentration ratio, as seen from Eq. (11.5). In
addition, coalescence and phase separation can be slow, especially if emulsions
are formed [11,13-15]. This is a particular problem for samples of natural origin,
such as urine and vegetable matter extracts, which contain surface-active agents
that adsorb at the LL interface to produce emulsions. Another disadvantage of
the separatory funnel technique is the relatively large amount of waste organic
solvent that is generated.
Despite these problems, about 45% of respondents to a 1996 survey indicated
that they were still using LLE for the preparation of aqueous samples [3]. This
percentage is close to the 47% who said that they were using SPE and/or SPME.
The continued attractiveness of LLE is due in part to the several advantages that
it possesses, as noted in Section 11.1.3, and in part to the development of

300
alternative simple and accurate methods of performing LLE, some of which are
noted in Section 11.1.5 and are described in later sections.

11.1.5 Modern approaches to LLE

Since the problems associated with the separatory funnel mode of operation
arise largely from the agitative drop dispersion/coalescence processes, it is no
surprise that recent LLE techniques, which appear to be growing in popularity
in the literature on chemical analysis, do not employ these processes. Such is the
case for the four techniques identified below.
Single liquid drop techniques with two phases employ a microlitre or smaller
size drop of organic solvent phase suspended in a large volume of aqueous
sample phase. When extraction is terminated, the drop can be injected directly
into an instrument such as a chromatograph or a graphite furnace atomic
absorption spectrophotometer, or it can be interrogated in situ by absorption or
fluorescence spectroscopy.
Unsupported liquid membrane techniques with three phases involve an
aqueous sample phase separated from an aqueous receiver phase by a layer of
organic solvent. The conditions in the aqueous receiver phase are made different
from those in the aqueous sample phase so that the analyte component is
forward-extracted out of the sample phase into the organic phase and back-
extracted out of the organic phase into the receiver phase, all in one simulta-
neous operation.
Supported liquid membrane techniques employ either two or three phases,
with simultaneous forward- and back-extraction in the latter configuration. The
aqueous sample phase is separated from the bulk organic or from an aqueous
receiver phase by a porous, polymer membrane, in the form of either a flat sheet
or a hollow fiber, that has been impregnated with the organic solvent phase. The
sample phase is continuously pumped, the receiver phase may be stagnant or
pumped, and the organic phase in the membrane pores is stagnant and not
replaced between samples. Since this technique of LLE is the subject of a chapter
in this book by Professor Jbnsson, it will not be further described here, except to
note that, in contrast to this supported membrane design, the unsupported
organic membrane phase, discussed above, is not stagnant but rather is con-
vected by momentum transfer from the convected aqueous sample phase.
Segmented flow techniques in small bore tubes performed in the flow
injection analysis mode, were initially developed in the late 1970s and early
1980s as a logical alternative to the Technicon AutoAnalyzer® continuous
extraction system. Continuous, cocurrent flow of alternating aqueous and
organic solvent segments occurs, either with or without air bubble segmenta-
tion. Mechanical phase separators sort the organic from the aqueous segments
at the outlet end of the extraction tube, or on-tube detection is used.
The first three techniques mentioned above, taken as a group, are sometimes
referred to as liquidphasemicroextraction (LPME) techniques.

301
11.1.6 Kinetics and mechanism

Equations (11.1-11.6) apply to LLE only after phase distribution equilibrium

has been attained. This requires time because extraction proceeds at a finite
rate. In practice, just as for SPME and for other techniques requiring time, it is
not necessary to wait for the attainment of equilibrium in LLE, provided that
the rate equation is the same among samples and standards and that the same
amount of time is allowed to elapse for all of them. For this reason, it is
important to have some understanding of the various processes which influence
the rate of LLE.
Generally speaking, there are two categories of rate (kinetic) processes:
chemical (CHEM) and mass-transfer (MT). CHEM kinetics, with which we are
all familiar, involves reactions between species once they find themselves in
close proximity to one another on a molecular scale; whereas MT kinetics relates
both to the movement of species from a considerable distance apart into close
proximity and the reverse [16,17]. Examples of CHEM kinetic processes that
may be important in LLE include [16,18-21] solvation/desolvation and
adsorption/desorption, both of which can occur at the LL interface; and protona-
tion/deprotonation and complexation/decomplexation, both of which can occur
at either the LL interface or in bulk solution. MT kinetic processes are generally
diffusion and convection toward or away from the LL interface, and diffusion
across the interface itself [16,19]. When convection is present in a liquid phase
the term "convective-diffusive mass transfer" is used, since diffusion must also
be present. If convection is absent, then we speak of purely "diffusive mass
transfer". Since the MT and CHEM processes all occur in series with one
another, any one or all of them could, in principle, be slow enough to contribute
to the kinetic rate equation of solvent extraction. Often, however, just one or two
of these steps will be much slower than the others so that it becomes the rate
determining step(s).
Extending the example started in Section 11.1.2, the extraction of a basic
analyte component from a well-buffered, low pH, stirred bulk aqueous phase, in
which it is present mostly as the species BH + , into a stirred bulk organic solvent
phase, in which it will be present mostly as the species B, might involve
convective-diffusive MT of BH+ to the vicinity of the LL interface, CHEM
deprotonation, MT diffusion of B across the interface, CHEM resolvation of B as
it crosses, and convective-diffusive MT of B away from the interface. Since
deprotonation, resolvation and diffusion across the interface are usually very
fast, it is likely that one or both of the remaining MT steps would be rate
determining.
Although the overall rate equation can sometimes have a complicated form,
there are many practical situations in which it is quite simple [7,16,19,22-26]. In
particular if the CHEM step in the aqueous phase can be represented by an
irreversible first-order (or pseudo-first-order) rate equation, and if we consider
only times after the convective-diffusive MT steps have achieved steady-state

302
(this usually occurs quickly if both liquid phases experience convection), then all
of the steps are described by equations that are linear in concentration, and the
overall rate equation is relatively simple. If the CHEM reaction in the aqueous
phase is:
ia -->pa (11.7)
and the extraction is:
Pa 4P 0 (11.8)
then both the original analyte-containing species, i, and the analyte-containing
product of the CHEM reaction, p, constitute the total analyte component I. The
overall extraction rate law is [7,16,19,23,24]:

dCt =koverall (Cl ,o,equil -C,o) = D overall (C,,,,i -C,) (11.9)

dt
which is a first-order rate law with a rate constant [21,27]:

k0 v (D=A= 6f 1 + AD· )D . +1 (11.10)

Vo P kINT DIa V kCHEM V

The first three terms within the parentheses relate to the MT steps (Eq. 11.8)
and include the reciprocal of the mass transfer coefficients (cm/s) for analyte
component movement toward (kMT,a = D, 0/a), across (kINT) and away from (kMT,,
= DJoo), the interface, respectively. The fourth term in the parentheses inc-
ludes the first-order CHEM rate constant kHEM (s-') for Eq. (11.7). In Eq.
(11.10), A is the area of the LL interface (cm 2 ), 6band 60 are the thicknesses (cm)
of the Nernst diffusion layers in the "a" and "o" phases, D1, and D, o0 are the dif-
fusion coefficients (cm2 /s) of the analyte component in the "a" and "o" phases,
and kINT is the interfacial mass transfer coefficient (cm/s) of the analyte compo-
nent.
In Eq. (11.10), the so-called Whitman two-film model has been employed to
describe the convective-diffusive MT in the "a" and "o" phases [10,16,25,27]. In
this view, all of the stirred (bulk) liquid phase is completely and continuously
mixed except for a thin Nernst diffusion layer of thickness 6 which is located
immediately adjacent to the interface (Fig. 11.1). This layer is considered to be
totally stagnant so that analyte crosses it exclusively by diffusion. In this view,
faster convection produces a smaller value of 6.
From Eq. (11.5) it can be seen that the equilibrium concentration of I in the
organic phase which should be used in Eq. (11.9) is:

C,o,equil Cl,a,initial (11.11)

303
Fig. 11.1. Schematic view of the Whitman two-film model for Bulk Aq. 8. Bulk Org.
convective-diffusive mass transfer at a LL interface. The Phase a o Phase
dark vertical line is the interface. Completely stagnant I
Nernst diffusion layers exist on either side. Their thickness 6 -I - - - N-
is smaller for greater convection. Bulk solutions are com- cton -
pletely convectively homogenized at all times. Solid arrowsonvection Diffusion onvecon
represent pure convective MT and dashed arrows represent Films
pure diffusive MT.

Integration of Eq. (11.9), gives the concentration extracted at time t:

Clo = Cl,o,equil (1 - e k.... ) (11.12)

For LLE which has not yet reached equilibrium, the fraction extracted at time t
is:

C______ - D V' (1 -e-k".,1lt)

I=
, = V,-_ V. before equilibrium (11.13)
V,

the preconcentration enrichment factor at time t is:

EF, = D
DI1(1-
I - e-o......)
Vo- ) before equilibrium (11.14)
(11.14)
1+D .°
Va
and the extent of separation of I from J at time t is:

= f,o -J,o, before equilibrium (11.15)

Equations (11.13) and (11.14) differ from Eqs. (11.4) and (11.5) by the presence
of the rate term (1 - e - k.. t.)
It is often the case in LLE that the rate of crossing the liquid-liquid interface
and the rates of all chemical reactions are much faster than the rates of diffusion
across the Nernst diffusion films. When this is so, k,,, and kH,,,, are both so large
that the terms containing them can be ignored and Eq. (11.10) simplifies to:

k.-alail A( D D d LDI
V+1] (11.16)

Furthermore, it may happen that only one of the liquid phase mass transfers
becomes the rate-determining step. This is true for example if D. a > 8>
6, in
which case Eq. (11.16) may be approximated as:

k =A . D DI
.[D V. +1i (11.17)

304
If, in addition, it is also true that DI · VJV a >> 1, then Eq. (11.16) may be
approximated as:

koveral A D (11.18)

Conversely, when the opposite is the case, i.e. 6o> > D- 6a, Eq. (11.16) becomes:

kvr = D,.[DV +1] (11.19)

and, if it is also true that 1 > > DI. Vl/Va, then:

kove = A D.o (11.20)

11.1.7 Convection, momentum transfer and interfacial adsorption

Since two liquid phases are in direct contact with one another, in the absence of
any porous solid support at the interface, then when one of the phases is
mechanically stirred (i.e. directly convected), the other will also experience
convective mixing. This is because the first phase transfers momentum to the
second phase as a result of frictional drag at the LL interface [16,19,25,28]. This
is true regardless of whether the second phase is layered over (or under) the first
or whether it has the form of a small drop suspended in or moving through the
first phase.
The significance of this indirectly induced convection in unsupported liquid
phases is two-fold. First, the rate of the resulting convective-diffusive MT in
each phase is faster than if MT were occurring by diffusion alone and second, it
can be treated quantitatively in terms of the Whitman two-film model, as
embodied in Eq. (11.10).
When discussing MT rates it is necessary to mention the possible effect of
adsorption of a species from solution [29]. Surface-active substances coming
from a sample or from reagents used in an analysis can adsorb at the LL
interface and thereby may reduce the interfacial MT coefficient (h,,,N) and/or
increase the diffusion film thickness (6). The latter would reduce the liquid
phase MT coefficient (D6/8), in any liquid phase which is convected via moment-
um transfer from an adjacent liquid phase [13,19,24-26,30-36]. If, in an extreme
case, convection is completely stopped in a liquid phase then diffusion becomes
the only liquid phase mass transfer process in that stagnant liquid phase and the
MT coefficient can no longer by approximated by D1 /6 but must be represented
by a different expression. The form of the expression for purely diffusive MT
depends on the geometry of the phase (e.g. planar, spherical). The subject is
treated in Section 11.3.2.

305
11.1.8 Quantitative analysis by LLE
Determining the amount of an analyte component I in a sample involves
subjecting both the aqueous sample solution and an aqueous standard solution
to the same LLE process, followed by instrumentally obtaining signals on both
the sample and standard extracts that are proportional to the concentration of I.
Instrumental techniques commonly employed include molecular absorption
spectroscopy, fluorescence, atomic spectroscopy, chromatography, etc. If only a
forward extraction is involved, then the instrument signals would be proportion-
al to C,, (sample) and C,o,(standard). The value C,o O (sample) can be obtained by
multiplying the known value of C,o (standard) by the ratio of the sample and
standard signals.
An essential precondition for a quantitative determination is that the frac-
tion of I extracted (f,) either must be quantitative (fi, = 1) or it must be the
same for both sample and standard. Equation (11.4) shows that when the extrac-
tion is allowed to proceed to equilibrium, assuming that the same ratio V/Va is
used, then conditions such as pH, ligand concentration, etc., used with both sam-
ple and standard, must be such that either D, is very large or it has the same
value. On the other hand, Eq. (11.13) shows that when the extraction is termi-
nated before equilibrium is reached the requirements are more stringent: not
only must D, be the same, but also both koer,,,, and t must be the same for sample
and standard. Another way of stating this is that unless extraction is unambigu-
ously quantitative, then for equilibrium extraction only those "matrix effects"
and operating conditions that influence Dr need to be made the same in standard
and sample; while for non-equilibrium extraction "matrix effects" and operating
conditions that influence both Dr and kove 1 1 must be made the same. For many
types of samples there are no additional matrix effects on kverall than there are on
D, so that the only operating conditions that require more careful control are
the interfacial area (A), the details of stirring (which effects 6a and 6) and tem-
perature. Usually a specially designed extraction vessel and a constant speed
stirring device are used for non-equilibrium extraction, and room temperature is
often adequate.
When surface-active substances are present in a sample the risk exists of a
smaller value of kover.ll for the sample than for the standard, as discussed in
Section 11.1.7. This particular type of matrix effect on non-equilibrium extrac-
tion can be addressed by "matrix-matching" or by employing the standard
addition technique in which the standard substance is added to the sample.
Since the presence of a surface-active component generally has a negligible effect
on DI, it does not exert a matrix effect on equilibrium extraction.

11.2 SEGMENTED TWO-PHASE FLOW TECHNIQUES

11.2.1 Introduction
The automation of analytical methods received a major boost in the mid-1950s

306
with the introduction of the Technicon AutoAnalyzer®, an air-bubble seg-
mented, continuous flow system [37-39]. In this system, in which aqueous
samples were injected periodically into a flowing aqueous stream, the presence
of air bubbles served both to reduce the longitudinal zone broadening of the
sample zone and to facilitate mixing of reagents with the sample. The subse-
quent implementation of LLE in the Technicon system in the mid-1960s
involved cocurrent flow of aqueous and organic solvent segments through coiled
glass tubes, sometimes packed with glass beads, either with or without simulta-
neous air-bubble segmentation [39-42].
Two decades after the development of the Technicon Autoanalyzer®, the
concept of continuous flow analysis in the "flow injection analysis" (FIA) mode
of operation was introduced [39,43,44]. In contrast to the AutoAnalyzer, FIA
involves the use of smaller diameter flow tubing and does not employ air-bubble
segmentation. The implementation of LLE in the FIA mode took place in the late
1970s [45-48]. Shown in Fig. 11.2 is a simple FIA instrument such as might be
used for segmented flow LLE [49]. The pump(s), sample injector, segmentor,
extraction tube, phase separator, detector and readout device are shown. Pumps
were initially of either the peristaltic type [45,46] or constant pressure type [50],
but now are as likely to be of the constant displacement syringe or reciprocation
piston type [51,52]. Sample injectors are valves, not unlike those used in HPLC
[39].
Segmentors appear in a wide variety of designs, the simplest being a
T-shaped tube [50,53]. Many of these designs have been reviewed [51,54]. The
function of the segmentor is to create a uniform sequence of alternating aqueous
and organic segments which then flow through the extraction tube. The pres-
ence of alternating segments facilitates mass transfer of the extractable analyte
component from the aqueous into the organic phase (see Section 11.2.2) and,
coincidentally, favors both reduced sample zone broadening and enhanced
reagent mixing. The extraction tube is typically smaller than 2 mm internal
diameter. Its material of construction, length and tightness of coiling all affect
the rate of solvent extraction and the sample zone broadening, as discussed in
Section 11.2.2. Phase separators, just like segmentors, appear in a wide variety

Fig. 11.2. Simple segmented flow, flow injection analysis (FIA) instrument for LLE showing
cocurrent alternating segments of organic (black) and aqueous (white) phases in the Teflon
extraction tube. I is the injector, S the segmentor and PS the phase separator. From Ref. [49].

307
 I l{

2.0

W
a 10
C
C

0
.Q I

Fig. 11.3. Peaks for eight replicate injections of a

sample in a segmented flow FIA instrument for I I I
LLE. Each injection gives one peak. From Ref. [50], 0 1.0 2.0
with permission. Time min)

of designs, with their principles of operation based on gravity (i.e. density differ-
ence), wetting (i.e. contact angle) and interfacial tension [13,29,51]. After sepa-
ration of the organic extractant phase from the aqueous phase, the former flows
to a detector. Each sample injection produces a peak, the height or area of which
can be used for quantification [50,55] based on similarly extracted standards.
Figure 11.3 [50,56] presents the absorbance signals obtained from replicate
injections of a caffeine solution into a segmented flow LLE instrument which
employs chloroform as the extractant.
The types of instruments used for detection of the analyte component in the
extract at the end of a segmented flow LLE extraction include molecular absorp-
tion and fluorescence spectrometers, atomic absorption and emission spectrom-
eters, mass spectrometers and electrochemical instruments. In addition, it is
common for the extract to be introduced into a high efficiency separation instru-
ment such as a gas or liquid chromatograph, or a capillary electrophoresis col-
umn, or related instrument [51]. The way in which the segmented flow extractor
is "interfaced" with the detection instrument can be "on-line", with the detector
located in the flow stream after the extraction column and phase separator;
"off-line", by collecting the extract and transferring it to a separate detection
instrument; or "on-tube", with the detector located at some point along the
extraction tube. One of the currently most challenging instruments with which
to interface a continuous flow extractor is a capillary electrophoresis instru-
ment. Common ways of doing so have been reviewed [57]. A requirement of any
on-line or on-tube detector is that it must have a relatively small hold-up volume
so as to minimize additional zone broadening and the resulting reduction of
detector signal [51].
As is true for LLE in general, a great variety of secondary chemical equilibria
are routinely employed in segmented flow LLE to control the values of the
distribution ratios of various sample components for purposes such as: (i)
increasing the fraction of analyte component extracted (f,o) when using a small
phase ratio (VoVa), for greater preconcentration; (ii) increasing the extent of
separation () of the analyte component over other sample components, for
greater selectivity in sample clean up; and (iii) converting the analyte compo-

308
Fig. 11.4. Somewhat more complicated segmented flow, FIA instrument for LLE based on
chloroform extraction from an alkaline aqueous phase. The injector is V4 . The segmentor is T,.
The phase separator M, which passes some of the chloroform phase containing extracted
pheniramine, is located between two sections of the Teflon extraction coil. The phase separator
Ma which passes some of the aqueous phase containing phenylephrine is located after the second
section of the Teflon extraction coil. See Fig. 11.5 for detector output. From Ref. [58], with
permission.

nent into a more detectable species. Reagents commonly employed for these
purposes include buffers, ligands, ion-pair reagents and crown ethers [51].
Because of this, a segmented flow LLE instrument is often more complicated
than that which is shown in Fig. 11.2. Pumps and tubes for aqueous reagents are
commonly present.
By way of illustration, such a system is pictured in Fig. 11.4 [49,58]. In this
example the reagent is 0.2 M NaOH which raises the pH of the aqueous phase
prior to segmentation with chloroform. Two phase separators are used in series,
the first of which isolates some of the organic phase and sends it to the detector
and the second of which isolates some of the aqueous phase and sends it to the
detector. Shown in Fig. 11.5 [49,58] are the peaks obtained from one injection of
a commercial nasal spray containing the antihistamine pheniramine maleate

Pheniramine 1
1,o

8.05

Fig. 11.5. Peaks obtained from the two-channel detector for

one sample injection in the instrument shown in Fig. 11.4.
The pheneramine peak is for chloroform phase from phase
.pnsrtr M n l thA nh.n-Al n+rin ne- i fr nmlne l °
-C'-L-aU' -. g -LLU lit: H--ll'-lJl-l-lllr rCIL 1t - ut YUCVUUm I I
phase from phase separator Ma. From Ref. [58], with o 1 2
permission. Time (min)

309
and the decongestant phenylephrine hydrochloride. At high pH pheneramine is
present as the neutral free-base form which extracts into chloroform, while
phenylephrine is present as its anionic conjugate base form which stays in the
aqueous phase. Also included in the instrument are a cation exchange resin trap
and an anion exchange resin trap which remove potential interferents from the
aqueous phase prior to the segmentor.
Analyte components that are routinely determined with the aid of secondary
chemical equilibria include metal ions, inorganic anions, ionic and nonionic
surfactants, pharmaceuticals and, in general, any analyte components for which
secondary chemical equilibria can be employed in conventional "separatory
funnel" LLE.

11.2.2 Equilibrium, kinetics and zone broadening in segmented flow

LLE

When the segmented flow of two immiscible phases has proceeded through a
sufficient length of extraction tube, phase distribution equilibrium is achieved
and the concepts that were introduced in Section 11.1.2 such as distribution
coefficient of a species (), secondary chemical equilibria and distribution ratio
of a component (D,), describe the system.
The mechanistic details of mass transfer kinetics in segmented flow LLE
have been studied [49,51,59-62] and a nodding acquaintance with them is
essential for understanding any of the several variations of this technique. The
first important characteristic of segmented flow is the fact that one of the liquid
phase solvents (i.e. the one with the smaller contact angle with the wall of the
extraction tube) forms a wetting film on the wall. As examples, a Teflon tube is
wetted by the organic phase and a glass tube is wetted by the aqueous phase.
This is illustrated in Fig. 11.6. The thickness of the wetting film is greater for
larger tube radius, higher viscosity of the wetting phase, higher flow velocity and
lower interfacial tension [49,51,60,62,63]. For typical segmented flow LLE
conditions the wetting film thickness has an order of magnitude value of 10 Am.
As a result of the presence of the wetting film, the wetting phase is a "continuous
phase" while the non-wetting phase is a "dispersed" phase whose segments are
completely surrounded by the continuous phase.
The second important characteristic of segmented flow, which is also illus-
trated in Fig. 11.6, is that convection is present in both the wetting and the
non-wetting phase segments. For the wetting phase this is the result of frictional
drag on the wall, while for the non-wetting phase it results from frictional drag
on the wetting film.
In segmented flow LLE, the injected aqueous sample solution occupies
several, to many, aqueous segments. The transfer of analyte component from
the aqueous to the organic phase segments occurs by the following steps:
liquid-phase MT through the aqueous phase to the liquid-liquid interface;
crossing of the interface; and liquid-phase MT away from the interface in the

310
 FLOW -

I :,"

Fig. 11.6. Longitudinal section of a Teflon extraction tube showing alternating segments of
organic (o) and aqueous (a) phases. The convective toroidal circulation patterns are shown in
each segment. The wetting film of organic phase can be seen between the aqueous segments and
the wall. Its relative thickness is exaggerated. From Ref. [491.

organic phase. Since each non-wetting phase segment is surrounded by the

continuous phase, this process occurs at all of its surfaces-front, back and side.
The presence of convection in both wetting and non-wetting segments causes
liquid-phase MT to be of the convective-diffusive type, explaining why the rate
constant in this technique is given by Eq. (11.10), or by one of its simpler
approximations, Eqs. (11.16-11.20). The pattern of the flow streamlines shown
in Fig. 11.6 is toroidal. The Nernst diffusion layers of thickness 6, though not
drawn in Fig. 11.6, are present on both sides of the interfaces. More vigorous
convection leads to smaller 6, larger koveri (Eq. 11.10) and a faster rate of
extraction (Eq. 11.9). If the extraction tube is coiled with a small radius of
curvature, then a radial "secondary flow" pattern becomes superimposed on the
toroidal flow pattern [49,61] and the secondary flow further decreases and
increases the rate of extraction.
Empirically, the rate of extraction for a given analyte component has been
shown to depend on extraction tube diameter, tightness of coiling, phase ratio,
segment lengths and flow velocity. For most of these, it has been possible to
interpret the effect quantitatively in terms of the constants A, V, Va, a and 6 in
Eq. (11.10). How close an extraction is to equilibrium when the sample zone
reaches the phase separator depends not only on the extraction rate, but also on
the length of time taken for the zone to flow from the segmentor to the phase
separator (i.e. t in Eq. (11.12)). This is equal to the extraction tube length divided
by the flow linear velocity.
When surface-active substances are present in segmented flow LLE, they
become adsorbed at the LL interface and may reduce the rate of extraction, as
discussed in Section 11.1.7 [51,64]. In such a case, the toroidal convective
circulation within the segments of the non-wetting phase are reduced.
Another aspect of segmented flow LLE that can be understood in terms of
the presence of a wetting film, is analyte zone broadening, which is also called
sample dispersion [49,51,52,59,60,62,65,66]. The number of segments spanned
by the analyte component zone increases as the zone moves along the extraction
tube. This is primarily because segments of the non-wetting phase do not fully
extract the analyte component from the wetting film during the short contact
time as they move past it, so that some of the analyte component gets left behind.
The extent of zone broadening is greater for thicker wetting films. Because it

311
arises primarily from the wetting film, zone broadening is minimal when the
analyte component is present nearly exclusively in the segments of the non-
wetting phase. In some, rather rare, cases analyte zone broadening occurs
because the analyte component adsorbs on the wall of the extraction tube [67].

11.2.3 Modes of segmented flow LLE

The discussions of segmented, two-phase flow LLE in Sections 11.2.1 and 11.2.2
have described mainly instruments having the traditional design which employs
continuous, segmented flow of both phases and includes a phase separator prior
to the detector. However, over time, designs have evolved toward the employ-
ment of more sophisticated electronic/computer control of instrument compo-
nents such as pumps, valves and detectors, and toward more elegant exploitation
of those fundamental properties of two-phase flow which were identified in
Section 11.2.2. Two trends, in particular, are noted. One trend is the greater use
of "on-tube" spectroscopic detectors employing fluorescence and absorbance
measurements, which can continuously measure analyte concentrations in the
segments as they flow past a point on the extraction tube [62,68,69]. This
eliminates the need for a phase separator. The second trend is toward new ways
of exploiting the presence of a wetting film on the walls of the extraction tube.
The terms "monosegmented flow extraction" [68,70,71] and "sequential injec-
tion extraction" [72] have been used for several related techniques. In these
techniques a single, short segment of an aqueous sample solution is injected into
a capillary extraction tube. Introduced just behind it is a short segment of the
organic extractant. Surrounding these two segments may be an aqueous carrier
fluid or air bubbles. The extraction tube often goes to an on-tube detector where
the concentration of the analyte component is measured photometrically in the
organic segments as they pass through it. Extraction tubes may be made of glass
or fused silica, in which case the wetting film is aqueous. For these, mass
transfer from the aqueous into the organic phase occurs largely via the wetting
film as described in Section 11.2.2. Alternatively, extraction tubes may be made
of Teflon or polyethylene, in which case the wetting film is organic. Mass
transfer from the aqueous sample segment into the organic segment is slow in
this case because there is no organic wetting film present on the tube wall
adjacent to the aqueous segment because it is ahead of the organic segment.
Something must be done to speed up the rate of transfer. Two quite different
approaches to circumventing this slow MT problem may be mentioned for their
clever exploitation of the fundamental principles.
In one approach [68], a homogeneous solution containing an aqueous
sample, a nominally immiscible organic solvent and the co-solvent ethanol, all in
one phase, is present as a monosegment between two air bubbles in a polyethyl-
ene extraction tube and is pumped toward an on-tube detector. Before it reaches
the detector, a concentrated aqueous NaNO3 solution is added to this single
segment causing it to separate, via the phenomenon of salting-out, into two

312
adjacent segments: one being a water-rich phase and the other being an organic-
rich phase. Since the analyte component becomes part of the organic-rich
segment during the process of two-phase formation in the tube, there is no need
for MT between the two segments after they have formed.
Another approach to speeding up the rate of extraction in a sequential
injection system with a hydrophobic extraction tube [72] involves drawing a
short organic solvent segment into a Teflon extraction tube followed by a
segment of aqueous sample solution whose composition is such that D for the
analyte component is large. Long air-bubble segments are present before and
after the two adjacent liquid segments, but not between them. The leading
organic phase segment lays down a wetting film on the Teflon tubing wall and
analyte component is extracted into this wetting film as the following aqueous
phase flows past it. In fact, the volume of the organic segment used is small
enough that, after flowing some distance, all of it becomes coated on the wall and
the organic phase is no longer present as a segment. Now the direction of flow is
reversed so that the aqueous phase flows back past the organic wetting film and
out of the tube to waste. Next, a different aqueous solution, for which D is very
small, is drawn into the extraction tube and through the organic wetting film on
the wall where it back-extracts the analyte component out of the film. Reversing
the flow direction now causes this aqueous back-extractant phase to flow back
past the film and out of the tube, where it is collected and subjected to HPLC
analysis.
For many "monosegmented" and "sequential injection" extractions, phase
transfer equilibrium is not achieved [69], so that both the concepts of segmented
flow that were discussed in Section 11.2.2 and the kinetic and mechanistic
concepts discussed in Section 11.1.6, are essential to a proper understanding of
these techniques.

11.2.4 Advantages, shortcomings and future directions

The advantages of using LLE for sample preparation, as opposed to using

solid-phase extraction, was pointed out in Section 11.1.3. Compared to tradi-
tional "separatory funnel" techniques, segmented continuous flow LLE in the
flow injection extraction (FIE) mode, described in Section 11.2.1, allows easy
automation of multi-reagent and multi-step procedures. A relatively simple
example of this was shown in Fig. 11.4. The use of computer-controlled auto
samplers, pumps, valves and on-line or on-tube detectors is routine. The waste of
solvents and reagents-a shortcoming of continuous flow extractors-has been
greatly reduced in the newer monosegmented and sequential injection systems
described in Section 11.2.3.
Another shortcoming of the continuous flow FIE mode is the fact that the
extent of preconcentration (i.e. EF1) that can be achieved is limited by an
inability to achieve stable segmented flow with phase ratios smaller than about
0.05 [51,52,73]. For this reason, EFI is usually smaller than 20, even in systems

313
that perform back-extractions [74], and is as high as 50 only in systems that
exploit the wetting film in special ways to enhance back-extraction [52].
In the 1996 survey mentioned above [151 it was noted that laboratory
workers do not often use segmented flow LLE in sample preparation. This is
probably due to the considerable instrumental complexity required, compared to
the relatively simple solid phase extraction systems that are commercially
available in profusion.

11.3 DROP TECHNIQUES WITH TWO PHASES

11.3.1 Introduction

The practical disadvantages of performing LLE in the separatory funnel mode,

in which a dispersion of drops is created by agitation and then destroyed by
coalescence, were identified in Section 11.1.4. Other modes of introducing drops,
which circumvent these disadvantages, are also employed in LLE.
A well-established one is the ascending/descending drop mode, in which a
single stream of evenly spaced drops of a liquid phase is introduced at the
bottom/top of a relatively wide cylinder filled with a non-flowing immiscible
liquid phase. The drops rise/fall under the influence of gravity, to be collected at
the top/bottom of the cylinder for analysis. This technique, which has long been
popular for performing mechanistic studies of LLE [18,30,75,76], was modified
for use in chemical analysis in the form of so-called Droplet Countercurrent
Chromatography (DCCC) [77].
In the DCCC technique the cylindrical tube has a 2 mm i.d. and the moving
drops are only slightly smaller than 2 mm diameter [78]. Commercial equipment
for DCCC finds its major use in the isolation of natural products [79]. The DCCC
technique differs from segmented flow LLE (Section 11.2) in that the continuous
phase is not flowing in DCCC. However, the convective-diffusive mechanism of
mass transfer between the phases is the same, since the ascending/descending
drops both induce convection in the continuous phase through which they move,
and acquire their own convection in the process. Because the continuous phase is
not flowing and because longitudinal mixing in the continuous phase is reduced
by using drops having diameters only slightly smaller than the tube width,
DCCC, as its name implies, can exhibit many "theoretical plates" and is capable
of fractionation of multicomponent mixtures. This also distinguishes DCCC
from segmented flow LLE which typically exhibits only one "theoretical plate",
except for versions of the latter in which very thick wetting films are employed,
such as "co-current chromatography" [66].
Since gravity-induced ascent or descent of drops in DCCC yields a slow
means of performing LLE, centrifugal versions of the technique have been
devised [80]. A disadvantage of DCCC is its instrumental complexity. Even in the
gravity-based design, it is necessary to pump the drop phase through a long
system of connected tubes.

314
 5-H GC
|Syringe Z

Fig. 11.7. Typical device for Microdrop-LPME showing a 1-1 drop of

organic phase suspended from the tip of a microsyringe needle into a
stirred aqueous sample solution. The panel in the upper right shows the tll 1 pi Octane
needle tip with the pendent drop. The dashed arrow outside the drop I -Drop
indicates local convection in the stirred aqueous phase and the dashed -1 ml Aq. Phase
arrow inside the drop indicates circulation in the microdrop which is o Magnetic
induced by momentum transfer from the aqueous phase. v Stirring Bar

In contrast to the relatively complex instrument designs that are required to

flow the drop phase through the continuous phase, a simpler approach to the use
of drops has recently come to prominence. This employs a single, stationary,
organic-phase drop of small volume, past which a relatively large volume of the
aqueous phase is made to flow. The flow can be caused by simple devices. In the
example shown in Fig. 11.7, a magnetic stirring bar is used [7,81]. In other
versions of the technique the drop-phase may be directly agitated, both phases
may be directly agitated, or neither phase may be directly agitated. These
variations will be described in Section 11.3.3. Since drop volumes are in the
microliter to picoliter range the phase ratio, V/Va, is always very small. Based on
the drop volume these Single Drop Liquid Phase Microextraction techniques
may be categorized as either: Microdrop-LPME in which a microliter drop is
employed; Nanodrop-LPME in which a nanoliter drop is employed; or
Picodrop-LPME in which a picoliter drop is employed.

11.3.2 Equilibrium and kinetics of Single Drop-LPME

As is true for all LLE techniques, when the extraction process is allowed to occur
for a sufficiently long time that equilibrium is achieved, then the equations
presented in Section 11.1.2 apply. The equations which describe the time-course
(rate) of extraction and therefore indicate how long it takes to reach equilibrium,
depend on which, if either, of the two phases experience convection, regardless of
whether convection in a given phase is caused directly or by momentum transfer
from the other phase, as described in Section 11.1.7. For Single Drop-LPME,
three MT situations are commonly encountered:
Case I: Directly convected aqueous phase, and indirectly convected organic drop
Microdrops of >0.1 l experience convection via momentum transfer from the
aqueous phase [19,25,30,31]. The convective circulation pattern for such a
system is shown in Fig. 11.7. The overall extraction rate equation is given by Eq.
(11.12) and kveral is given by Eq. (11.10) or by one of its approximate forms, Eqs.
(11.16-11.20).

315
 Case II: Directly convected aqueous phase, in the absence of convection in the
organic drop
Drops smaller than about 0.1 L1 (i.e. nanodrops and picodrops) do not experience
convection via momentum transfer from the aqueous phase. The exact cut-off
volume below which a drop becomes stagnant depends on such factors as the
vigorousness with which the aqueous phase is stirred (i.e. its Reynolds number),
the interfacial tension between the organic and aqueous phases and their
density difference, and the presence of minute traces of surface-active impurities
which exist even in highly purified solvents [19,25,30,31].
However, even microliter-size drops can have their convection reduced or
even completely suppressed by the presence of an adsorbed interfacial film.
When the drop is not convected, MT through the aqueous phase toward the
interface is convective-diffusive and is represented by the term D. ba/Di. a in Eq.
(11.10). On the other hand, MT through the organic drop away from the
interface is purely diffusive and can be approximated by an equation for
diffusion into a sphere [82]. This complex equation, which must be evaluated
numerically, does not lead to a first order rate equation and is not correctly
represented by the term SJD,,o in Eq. (11.10).
If k,,, and h,,,HEM are large and if the rate determining MT step occurs in the
aqueous phase, then the first order Eqs. (11.9) and (11.12) describe the kinetics,
and koveral is given by Eq. (11.17) or (11.18); but if the rate determining MT step
occurs in the non-convected organic drop, then Eqs. (11.9) and (11.12) do not
describe the kinetics and Eqs. (11.19) and (11.20) do not apply. To make an
order-of-magnitude estimate of the conditions under which the control of Case II
MT kinetics will reside in the drop phase rather than in the aqueous phase, the
fact that drop phase MT is not first order can be ignored and the inverse mass
transfer coefficient term o/D,,o in Eq. (11.10) can be replaced by tadius/rdrop, where
rdrop is the drop radius and t,.adus is the time required for an average molecule of I
to diffuse the distance rd,,,. Rearrangement of the Einstein diffusion equation
gives [25,83,84]

tradius- rdrp s/cm (11.21)

rdrop 2D,o

Since the radius of a spherical drop is equal to 3 . , Eq. (11.21) can be

written as: 4

tradis - 0.31 Vb (11.22)

drop Dlo

Values of tadi/rdrop are presented in Table 11.1 for various drop volumes V,
assuming a typical value of 3x10 ° cm2 /s for D, 0o [17]. Also shown in Table 11.1
are values of Di. 6aD,, for various values of the distribution ratio D1, assuming
typical values of 20x 10-4 cm for 6a [7,81] and 1 x 10 -5 cm 2/s for Dia [17].
If tradius/rdrop > > D b8/dD,, then the extraction rate is controlled by diffusive

316
TABLE 11.1
Drop size, distribution ratio and the rate-determining step for Case II mass transfer kinetics

V. (t1) rdrp trf us (s/cm)

rdrop

Microdrop 100 2.88 mm 4800

10 1.34 mm 2233
1 620/zm 1033
Nanodrop 0.1 288 m 480
0.01 134 m 223
0.001 62 m 103
Picodrop 0.0001 28.8 m 48

D, D-,6a (s/cm)

1 200
10 2,000
100 20,000
1000 200,000

MT in the drop, and the rate equation will be the one for spherical drop diffusion,
rather than Eq. (11.12). From Table 11.1 it can be seen that drop diffusion con-
trols the extraction rate for microdrops (Vo = 1 to 100 Al) when D is relatively
small.
Iftradius/rdrOP< < D aI/Di,, then convective-diffusive MT in the aqueous phase
controls extraction rate. Equation (11.12) is the rate equation and kovera is given
by either Eq. (11.17) or (11.18) depending on whether DrVVV is greater than 1.
From Table 11.1, the aqueous phase MT is seen to be rate determining for small
microdrops and large nanodrops (V, = 0.1 to 10 gl) when D1 is large, and it is
always rate-determining for small nanodrops and picodrops (V = 0.01 l).
If traiuJr, 0rp and D· 6Ja/D, are not very different from one another then MT in
both phases affect the extraction rate and the "mixed" rate equation will be even
more complicated than both Eq. (11.12) and the spherical drop diffusion
equation.
In the above discussion it has been assumed that 1/,,NT is negligibly small,
even though an adsorbed film may be present at the LL interface. This is a
reasonable approximation since the suppression of drop convection is usually
found to be a more important effect of adsorbed surface-active substances than is
their diffusional barrier effect [19,85].

Case III: No convection in eitherphase

Here the aqueous phase is not stirred. From the discussion of Case II above, it
will be obvious that the rate determining step for LLE will always be purely

317
diffusive MT in the aqueous phase. Because the phase ratio VoVa is small (i.e.
10- 5 to 10 - 6) and because the extraction process is terminated short of equilib-
rium, the fraction extracted, f, 0, will generally be very small. This means that the
concentration of analyte in the aqueous phase will not be significantly decreased
during the extraction (i.e. C,a,equi C,ainitial). The differential rate law for revers-
ible semi-infinite spherical diffusion, which applies well before equilibrium is
achieved, has the form [84,86-89]:

dC,' 4dC2 =4r

D( CD I0 [VD 1( o' +t) 1 (11.23)
dt Dir/
1 . [(7DJ t) 2drop ]

For Picodrop-LPME (rdrop < 6 Am), (Di,t)" / 2 < < rp, even at times as short
as 1 second, so that the term in square brackets becomes [1/rdrop], and Eq. (11.23)
can be integrated to give the following first-order rate equation:

Cl,o = DI Claiitial (1- exp(-kovral. t)) (11.24)

where the overall rate constant is:

koveral 3 DI, (11.25)

DI rdrop

The rate constant is larger for smaller drops; however the fraction of the total
analyte component which is extracted (f,o) in a given time is greater for large
drops.
For Nanodrop-LPME and especially for Microdrop-LPME, drop is smaller
so that Eq. (11.23) cannot be simplified and must be integrated as is. The
integrated rate equation obtained from Eq. (11.23) (not shown) gives a generally
convex shape rate curve, but because of the presence of t /2, it is not a first-order
equation.

11.3.3 Drop dissolution in the aqueous phase

In Single Drop-LPME it is desirable that the volume of the organic drop, V,

remain constant during the LLE process. In fact, some decrease in V will occur
because no solvent is completely immiscible with water so that there will be
some dissolution of the organic solvent in the aqueous phase (and the reverse).
However, a slight decrease in Vo is acceptable because it will occur to the same
extent for the standard as for the sample and its effects will cancel out.
The decrease in Vo can be estimated. If the dissolution process reaches
equilibrium then the fractional decrease of the mass (and volume) of the drop
will be:

Anorg - Corg,asat Va (11.26)

-org, initial Vo,initial P org

318
where Corg,a,sat is the solubility of the organic solvent in the aqueous phase (g/ml)
and Porg is its density, at the temperature of interest.
Often in LPME equilibrium is not reached, so that a knowledge of the
dissolution rate equation is useful. The rate of dissolution of a spherical drop of
organic solvent in water at constant temperature depends upon the solubility of
the solvent in water, the vigorousness with which the aqueous phase is con-
vected, the radius of the drop and presence of an interfacially adsorbed film
[17,19].
For an organic microdrop in a convected aqueous phase, a very thin layer (i.e.
< < 6 a) of aqueous phase immediately adjacent to the interface is and remains
saturated with organic solvent and serves as a "source-layer" away from which
dissolved organic solvent experiences convective-diffusive MT in the aqueous
phase [19]. This involves diffusion across a Nernst film of thickness a . The
resulting first-order dissolution rate law is:
1 dnorg Dorg. C
Ad -r
= (Dorga,sat- org,a) (11.27)
A dt 6 (,

Equation (11.27) can be transformed to take into account the fact that A
decreases with drop dissolution, and the resulting expression can be integrated
to give a rather complicated equation. However, for the purpose of assessing the
suitability of an organic solvent for use as an extraction drop, a "worst case" can
be estimated from the initial rate of dissolution. Substituting A = constant and
Corg,a = 0 into Eq. (11.27); dividing both sides by the initial grams of organic
solvent norg,initial, rearranging and integrating, gives an expression for the fraction
of the drop dissolved in time t:

Anlorg A Dorg a 'Crga,at t (11.28)

norg,initial a 'norg,initial

As an example consider a 1-Al drop of n-octane suspended in 1 ml of stirred

aqueous phase. Here, the water solubility of n-octane is Corg,a,, = 6.6 10- 7 g/ml,
Dorg a 5x10- 6 cm 2/s (nominal), 6a = 1x10 3 [90], A = 0.048 cm 2 and norg,initial =
1x10 ml · 0.70 g/ml = 7 x10- 4 g. Using these values in Eq. (11.28) shows that
-3

the fraction of a 1-l drop which will have dissolved after 30 min (i.e. 1800 s) is
4x 104, or 0.04%. At equilibrium, after a longtime, Eq. (11.25) gives the fraction
dissolved as 6.6 x 104, or 0.066%, which is still negligible.
For an organic picodrop in a non-convected aqueous phase the dissolution
rate is controlled by pure spherical diffusion into the aqueous phase from the
saturated source-layer surrounding the drop. Assuming that the experiment is
terminated well before drop dissolution-equilibrium is achieved, which is usually
the case, the integrated fractional rate equation is for semi-infinite spherical
diffusion, which has the form [84,86]:
Anorg, a
4
t drop Dorg,a Corg,a,sat t(11.29)
og gitioongitil(11.29)
norg, initial norg, initial

319
If the drop radius is assumed to stay constant, this equation pertains to the
initial dissolution rate. The fact that the rate decreases with time because rdrop
decreases, is ignored.
As an example, consider a 40 pl (i.e. 40 x 10 - 9 ml) drop of n-octane in contact
with any volume of non-stirred aqueous phase that is > >40 pl. Here rdrop = 21
Am and the other parameters are the same as those given in the example above.
After 30 min the fraction of the drop that will have dissolved is 0.004, or 0.4%,
which is negligible. The fraction dissolved when dissolution equilibrium has
been reached depends on Va, as shown in Eq. (11.25). For instance, at equilib-
rium, 24% of a 40 pl drop will be dissolved in a 10 Al aqueous phase.
A point which has been ignored in the above discussions of drop dissolution is
the fact that Crga,sat is larger for smaller drops, as given by the Kelvin equation
[91]. However, the effect is negligible for drops with rdrop > 0.1 Avm and can
justifiably be ignored, even in picodrops.
Loss of an organic solvent from a small drop can occur not only by dissolution
in the aqueous phase but also by vaporization into the headspace of the vessel in
which the LLE is being performed. The number of moles of organic solvent vapor
in the headspace at equilibrium, n'orgvapor, can be estimated from the vapor
pressure of pure organic solvent, Porg, assuming ideal gas behavior:
Porg Vheadspace
norg,vapor = (11.30)
R.T
Of course, before dissolution and vaporization equilibria are reached n'org,vapor
will be smaller than the value given by Eq. (11.30).
Experimentally, drop dissolution and vaporization can be completely pre-
vented by pre-equilibratingthe aqueous phase and headspace with vapor of the
organic solvent, from an independent source, prior to introducing the organic
drop into the system. In this way, even organic solvents having considerable
solubility in water and considerable vapor pressure can be used for Single
Drop-LPME.

11.3.4 Modes of single drop-LPME

A variety of experimental designs has been employed to contact the single drop
of organic phase with the aqueous sample solution. Also, a variety of analytical
measurement techniques has been used to quantify the amount of analyte
extracted into the drop. These include molecular spectroscopic and electrochem-
ical techniques that are used directly on the drop, in situ, and chromatographic,
electrophoretic, atomic spectroscopic and mass spectrometric techniques that
require injection of the organic extract into the instrument in either an off-line
or on-line manner. In this section, the discussion will include modes of contact-
ing the phases, modes of quantifying the analyte and experimental results.
Microdrop-LPME, Nanodrop-LPME and Picodrop-LPME will be discussed in
Sections 11.3.4.1, 11.3.4.2 and 11.3.4.3, respectively.

320
 It will be recalled that, no matter what the experimental design, LLE is being
performed for sample preparation, usually to accomplish sample clean-up and/or
analyte-component preconcentration, and that the quantitative measurement
step is being made directly on the organic drop, without the need to remove the
extracted analyte from it. A recent review gives an overview of the formation and
properties of aqueous and organic microdrops and of drop-based analytical mea-
surements [92].

11.3.4.1 Microdrop-LPME
One of the simplest ways of contacting an organic microdrop with a stirred
aqueous sample solution is to suspend the drop in the sample solution from the
tip of a microsyringe needle, as is shown in Fig. 11.7. Experimental characteriza-
tion of the equilibrium and kinetic features of this system was performed on both
an 8-jl drop [7] and a 1 -ul drop [81], of n-octane in 1 ml of a stirred aqueous
sample solution. The presence of indirectly-induced convection in the organic
drop was made visible by suspending charcoal powder in the drop. The kinetics
are therefore expected to be Case I mass transfer (Section 11.3.2). Shown as the
points in Fig. 11.8 are plots of measured concentration of the analyte component
4-methylacetophenone (4-MAP) in the 1 Al drop of n-octane versus extraction
time, for experiments performed at four different stirring rates. The concentra-
tion of 4-MAP was measured by injecting the 1 l n-octane drop directly into a
gas chromatograph. The solid lines in Fig. 11.8 are from nonlinear least squares
fits of Eq. (11.12) to the data points. From the measured limiting concentration,
C4M,o,equ, which all four curves are approaching at long times, the value D4 MA,
= 41 ml/ml is obtained by rearranging Eq. (11.4). The value of k,,,over,, increases
with stirring rate as shown in Table 11.2. Since D4 P is large, kovera, is expressed
by Eq. (11.17), from which the value of the mass transfer coefficient D4M a/6a,
shown in Table 11.2 is calculated. The diffusion coefficient D4 MAP was measured
to be 7.68 x 10- 6 cm 2 /s in a different experiment and is used to obtain the Nernst
diffusion film thicknesses that are shown in the last column of Table 11.2. As
expected, a decreases at faster stirring rates.
A Microdrop-LPME device, based on the design shown in Fig. 11.7, with Case
I MT, employed a 2 -/tl drop of organic solvent in a stirred aqueous sample
solution of between 1 and 9 ml volume in order to preconcentrate and measure

E 4-

Fig. 11.8. Plots of concentration of the extracted 1500 rp

component 4-methylacetophenone in a microdrop, ± 2- 10 rpm
measured by GC, versus time of extraction, at four 1
different stirring speeds. Solid lines are for non-
linear least squares fits of Eq. (11.12) to the points. 2 4 6 1 12
Adapted from Ref. [81]. Stining time(min)

321
TABLE 11.2
Equilibrium and kinetic parameters for the extraction of 2.06 x 104 mo/i of 4-MAP from water
into 1.00 l1of n-octane
6
Stirring rate C4 -MAPoe uil koral D-MAPaa/
4 * a
-1
(rpm) (mol/cm ) X 106 (S ) X 103 (cm/s) x 103 (cm) x 104
900 7.97 ± 0.08 4.68 ± 0.11 4.53 ± 0.15 16.9 + 0.7
1200 7.92 + 0.07 6.42 0.17 6.18 ± 0.22 12.4 ± 0.5
1500 7.87 ± 0.03 7.98 ± 0.09 7.64 ± 0.12 10.1 ± 0.3
1800 7.98 ± 0.03 9.05 ± 0.11 8.78 ± 0.14 8.8 + 0.2
2
*D4_MAP, a = 7.68 x 104 cm /s from a separate experiment.

the concentrations of pesticides in river water [93,94] and cocaine and its
metabolites in urine [95]. In these studies a variety of different organic solvents
were used for the microdrop including hexane, chloroform, toluene and mixtures
of these. Extracted analyte was quantified in the drop by gas chromatography
employing an internal standard. Typically, stirring rates were rather slow and 5
min extraction times were used so that the fraction of analyte extracted was
small (f,0 < 1%). Also, no special precautions were taken to keep the stirring rate
constant. These conditions led to rather high relative standard deviations (RSD
= 10%) which, however, were quite adequate to screen for the drugs and
pesticides.
The microsyringe set-up shown in Fig. 11.7 has also been used with a
different mode of contacting the organic microdrop with the aqueous sample
solution [96,97]. In this mode, which the authors call "Dynamic-LPME", the
aqueous phase in the sample vial is not stirred and the organic phase does not
have the shape of a drop during the mass transfer process but, rather, is a
segment inside the microsyringe needle. Convection is created in both the
aqueous phase and the organic phase by alternately moving the microsyringe
plunger in and out 15 or 20 times in succession. To begin with, at the time the
microsyringe needle is immersed in the 4-ml of aqueous sample solution the 1-gl
of organic solvent is present just inside the needle tip. As the plunger is then
retracted the organic solvent is pulled farther back in the needle and is followed
by a specified, few 1lof aqueous sample solution (Va,,amp plug) After a few seconds
pause (the dwell time) the plunger is advanced to expel all of the aqueous
solution, leaving the 1 Aldof organic solvent just inside the tip again. Both the
organic and aqueous liquids in the needle experience convection due to frictional
drag during the times that the plunger is moving out and in. The majority of the
extraction of analyte component occurs from the aqueous sample plug that is in
the needle, into a wetting film of the organic solvent that coats the inside of the
needle, which is made of stainless steel.
Because extraction is occurring into a thin wetting film of organic phase
rather than a drop, the kinetic conditions for Case I in Section 11.3.2 do not
apply. Furthermore, the extraction occurs as a series of steps, one for each cycle

322
of the plunger, rather than continuously. If it can be approximated that the mass
transfer between the thin wetting film and the 3 Al aqueous sample plug is fast so
that they reach distribution equilibrium, then after the first plunger cycle the
concentration in the organic phase is:
2. DI C, a,initial 'o, film Va,samp plug (11.31)
(rneedle +2 D, o, flm ) Vo

where 6 ,fim is the wetting film thickness, r,eedle is the inner radius of the needle
and V0 is the total 1 l volume of the organic solvent [96]. If the following
additional approximations are made: (i) Ci,a CI,,initiM, because Va >> V0 and
because the extraction is always stopped far short of extraction equilibrium; (ii)
the organic wetting film is completely renewed for each plunger cycle; and (iii) n,
the number of plunger cycles, is not very large so that nV,,,lm < < Vo; then the
relationship between the fraction of analyte component extracted and n is:
6
Cl,. Vo 2 n D,- o, Vasampplg
film (11.32)
Cl,a,initi Va needlee +2' D, '5o,fi ) V

Equation (11.32) predicts that the fraction extracted should increase approxi-
mately linearly with the number of plunger cycles and with the volume of the
aqueous sample plug that is drawn into the needle. This has been experimentally
demonstrated to be true [96]. The wetting film thickness and the closeness of
approach to equilibration between the wetting film and the aqueous sample plug
depend on how fast the plunger is moved [97]. It should be evident from the
details of the mode of contacting the phases that "Dynamic-LPME" more closely
resembles the "sequential injection extraction" mode of segmented flow LLE
(Section 11.2.3) than it does Microdrop-LPME.
Analytically, after the extraction operation the 1 l of organic solvent is
directly injected into a gas chromatograph with either electron capture or mass
spectrometric detection. Since the detector signal is proportional to C,o. V, it is
evident from Eq. (11.32) that if n, V,, pl,ug and the details of plunger movement
are kept constant then a linear calibration curve will be obtained for detector
signal vs C,a,initiaVa in the similarly-extracted aqueous standard which can be
used to determine I in the sample. In the studies cited the solvents toluene,
chloroform, butyl acetate and isooctane were all employed as the organic phase
and a variety of chlorobenzenes were determined at 2 Ag/l concentration in
industrial wastewater with RSDs < 10% in most cases.
Another mode of contacting the phases in Microdrop-LPME, rather than
stirring the aqueous sample solution as in Fig. 11.7, is to cause it to flow past the
microdrop which is positioned in a flow cell of small dimensions. The flow
induces MT in the drop via momentum transfer, producing Case I kinetics. The
flow also continuously brings fresh aqueous sample solution into contact with
the drop, making C,,equil = CI,a,initia in Eq. (11.12).

323
 Chloroform
Aqueous
_I,

CHl
P-

Left: Fig. 11.9. Diagram of a flow cell containing a hydrophobic PEEK capillary tube with a
microdrop of organic solvent standing on top of it. Aqueous sample solution flows out the tip end
of the PEEK tube and past the drop. It fills the flow cell and leaves via the top of the flow cell. The
syringe is used to withdraw some of the organic drop for analysis. Adapted from Ref. [98].
Right: Fig. 11.10. Diagram of a flow cell containing an organic microdrop (CHC1) suspended
from the tip of a capillary into a flowing aqueous sampling solution. In the design, the flowing
aqueous phase oscillates in size from filling the cell to the size shown by the dashed curved line.
Detection is performed photometrically in the chloroform drop, in situ. Adapted from Ref. [99].

In one flow design shown in Fig. 11.9, the organic microdrop (1-5 l) is
positioned at the tip of a hydrophobic capillary tube (polyetheretherketone,
PEEK, 0.254 mm i.d.) and the aqueous sample solution is pumped through the
PEEK tube and past the microdrop [98]. The concentration of analyte compo-
nent in the drop is determined by withdrawing 0.5 Al of the drop into the syringe
and injecting it into a gas chromatograph fitted with an electron capture
detector. A series of eleven nitro- and chloro-aromatics were used as analyte
components. The rate of extraction is reported to increase with increasing flow
rate of the aqueous solution, consistent with a decrease in thickness of the
Nernst diffusion films in Eq. (11.16). The fact that the analyte components all
have large D, values suggests that MT rate control may reside in the aqueous
phase, rather than in both phases, but this has not been experimentally verified.
In fact, there is no need to know this in order to perform extractions for
analytical purposes. For a 10-min extraction time with a 1- 1I drop of toluene and
a flow rate of 0.1 ml/min, preconcentration enrichment factors (EF) ranged
from 260 to 1600 depending on D, for the compound, and the fractions extracted
(f,,) were greater than 0.7. RSDs were < 10%. Seawater samples were analyzed.
A second flow design involves an approximately 1-1 drop of chloroform
suspended in an approximately 50-L1 glass flow cell through which the aqueous
sample is pumped (Fig. 11.10) [99]. The analyte dodecylsulfate anion (DS-) was
extracted as a colored ion pair with methylene blue (MB + ) cationic reagent which
was initially added to the aqueous sample solution. After the extraction, water
was pumped instead of sample in order to backwash the organic drop and then
the concentration of MBDS in the chloroform drop was determined by measur-
ing the absorbance of the drop in situ by means of an LED/optical fiber-based
detector (not shown in Fig. 11.10). At the end of the experiment the organic drop

324
Left: Fig. 11.11. Diagram of a drop-in-a-drop which is
used for Microdrop-LPME in an acoustic levitation field.

was pumped out and a fresh one was formed by pumping in chloroform. The
entire instrument is computer controlled. For a 60-s extraction time the extrac-
ted concentration increased with an increase in sample flow rate which is
consistent with Case I or Case II kinetics. It is likely that in this determination,
the MT kinetics are Case II with rate control in the aqueous phase, because DS-
is surface-active, MB-DS is probably surface active and D, is large.
Yet another mode of contacting the phases in Microdrop-LPME employs
acoustic levitation of a drop-in-a-drop (Fig. 11.11). The levitation force is pro-
vided by the acoustic radiation pressure of an ultrasound source [100,101]. In
the experiment described, 0.74 r1 of an aqueous sample solution and about 4 pAl of
toluene were levitated at a pressure node in an ultrasonic field. The drop-in-
drop, with the water drop inside the toluene drop, forms spontaneously. After 2
min, 0.2 Al of the toluene phase is withdrawn from the outer drop with a
microsyringe and 0.1 l of it is injected into a gas chromatograph. The analyte
component was n-hexanol. It was necessary to include an internal standard
(n-octanol) because evaporation of toluene during the 2 min levitation is signifi-
cant. Since the ultrasonic field induces convection in the drops, this is probably
Case I kinetics. RSD was -7% and was attributed to the precision of the GC step
rather than to that of the extraction step.
In some Microdrop-LPME studies the microsyringe set-up shown in Fig. 11.7
has been used with no convection in either phase [96,102]. This gives Case III
kinetics, in which the integral of Eq. (11.23) is the rate equation. In these studies
a 1 or 2 -ul drop of toluene or hexane is suspended in the non-stirred aqueous
sample solution (e.g. 4 ml) for a specified period of time after which the micro-
drop is injected into a gas chromatograph. For a given extraction time the
fraction of I extracted is greater for a large drop volume, but it is significantly
less than when the aqueous phase is stirred. Linear calibration curves are
obtained when the extraction conditions are kept constant. For a 15-min extrac-
tion, typical preconcentration enrichment factors (EFI) of about 10 are obtained.

11.3.4.2 Nanodrop-LPME
LLE from a volume of 1-3 Al of a convected aqueous sample solution into a 1-30
nl drop of chloroform at the tip of a fused silica capillary was used to pre-
concentrate and clean up analyte components in preparation for microspot
matrix-assisted laser desorption-ionization mass spectrometry (MALDI-MS)
[103]. The extraction device is shown in Fig. (11.12). Drop sizes up to 30 nl are

325
 Pipette Tip
Capillary Aqueous
Filled with a Sample ( 1-5 pL)
Water-lmmiscible
Organic Solvent

Organic,
Solvent
Drop L
(1-30 nL) A
Rotation by
Small Electric
Motor

Microscopic
Observation [40x]

Right: Fig. 11.12. Diagram of a Nanodrop-LPME device for extracting analyte components from
a rotationally-stirred aqueous phase into a nanodrop formed at the tip of a fused silica capillary.
The microscope is used to help position and monitor the drop. From Ref. [103], with permission.

very stable even at a 3000-rpm rotation speed of the plastic pipet tip containing
the sample solution. Extraction times of 15 min are typical. After the extraction
the organic drop was withdrawn into the capillary, the capillary was removed
from the pipet tip and all the chloroform in the capillary (e.g. 200 nl) was pushed
out and deposited onto the MALDI matrix layer where the chloroform rapidly
evaporates leaving the analyte component as a microspot. Hydrophobic peptides
were separated from large concentrations of interfering hydrophilic peptides
and from interfering metal ions. Also, it is claimed that, since the organic drop
does not touch the wall of the container holding the aqueous sample solution, the
extraction of contaminants that might have been present on the wall is
minimized.
Because the organic drop is in the nanoliter range it will not experience
momentum-induced convection and Case II kinetics prevails, with MT in the
aqueous phase being rate determining. MALDI is not a quantitative technique,
so fraction extracted (fi, ) and preconcentration enrichment factor (EFI ) were not
quantified, but both preconcentration and extent of separation from inter-
ferents () were inferred to be very high.

11.3.4.3 Picodrop-LPME
LLE into picoliter-size drops has been performed with both convected and
non-convected aqueous phases. Shown in Fig. 11.13 is an apparatus for "micro-
injection-microspectroscopy" in which a drop containing between 1 and 80 pl of
tributylphosphate organic phase is formed at the tip of a microcapillary in a glass
microtube (400 m deep x 1 mm side x 10 mm long [104]) The aqueous sample
solution is pumped through the glass microtube at flow rates from 5 to 20 l/min.
The sample solution contains the extractable metal-ligand complex Al-2,2'-di-
hydroxyazobenzene. The concentration of complex in the picodrop is determined
by in situ measurement of its absorbance at 480 nm using the associated optical
microscope system. The kinetics here are Case II, with film-diffusion rate

326
 Objective
Lens

organic phase, flow OFF

*--- ~ aqueous sample, flow ON

Drop - -Pinhole

| Detector |

Fig. 11.13. Diagram of a flow cell containing an organic picodrop past which aqueous sample
solution is continuously pumped. Detection of analyte component in the picodrop is performed
photometrically on the drop in situ using microspectroscopy. Adapted from Ref. [104].

control in the convected aqueous phase. Because fresh aqueous sample solution
is continuously brought to the drop, Cl,, equi = C a,initial.
Since extraction rate studies of C,,o vs. t were performed for only short times,
Ci,1 << Clequi in the first-order rate equation (11.9), and the integrated rate
equation for initial rate is:

Ct,o overall ·DI . C, ainitia (11.33)

where, from Eq. (11.18):

k r 4TrdrOP Da 3 Dia (11.34)

1 6
4
overall irop a
37rdrop drp

As predicted from these equations, experimental plots of C, o vs. t were linear and
their slopes increased with increasing flow rate of aqueous sample, which gives a
smaller 6, and with decreasing drop radius. The preconcentration enrichment
factor of the aluminum complex, for which D, = 162, was EFt - 103 in a 17-pl drop
after a 10-min extraction.
Picoliter-LPME with laser trapping of the drop in an unconvected aqueous
phase has been the subject of several reports [84,87,88,105,106]. More recently,
a microcapillary injector has been used to position a 31-pl drop of nitrobenzene
on top of a small gold disk electrode in a cell containing 2 ml of aqueous sample
solution [89]. Laser trapping was not needed. The concentration of the electro-
active analyte compound 1-ferrocenyl-2-butanol in the nitrobenzene drop was

327
 1' . . . I I

i ' ' ' I

Fig. 11.14. Plot of quantity of charge Q in nano-

coulombs versus time for Picodrop-LPME performed
in an unstirred aqueous solution. Since Q is prop-
ortonal to q. 1.4) predicts a nrst-oraer plot. 0 5 10 15
The line is for a first-order fit to the data points. 0 5 10 15 20 25 30 35 40
Adapted from Ref. [89].

determined in situ by a coulometric technique. The fact that the mass transfer
kinetics for this system are Case III was demonstrated experimentally by the
good fit of Eq. (11.24) to the C, vs. t data (Fig. 11.14) and by the linearity of plots
of observed rate constant vs. raT20p [84], as predicted by Eq. (11.25).

11.3.5 Advantages, shortcomings and future directions

For sample preparation prior to an analytical measurement, Single Drop-LPME

is more suitable than is Droplet Countercurrent Chromatography because of its
relative simplicity and much smaller organic solvent consumption. Techniques
in which the single drop is in contact with a convected (i.e. stirred or flowing)
aqueous sample solution appear usually to be preferable to those in which the
aqueous solution is not convected, because the large values of D, mean that
kinetic rate control is usually in the aqueous phase, for which Case I and Case II
kinetics almost always give faster extraction than Case III. For reproducibility,
the conditions of convection and temperature should be fairly well controlled.
Single Drop-LPME techniques in which the analyte components in the organic
drop are determined by injection into an instrument with the capability of
resolving and simultaneously determining multiple components are more suit-
able for sample preparation than are those techniques in which in situ analyses
are performed. This is because most samples contain several to many compo-
nents and the in situ measurements, such as absorption and fluorescence
spectroscopies and electrochemistry, are not capable of simultaneous analyses
and also are subject to interference from non-analyte components that are
present in the sample.
In this context, ways and means of accomplishing the injection of part or all
of the small organic drop into a chromatographic column, an electrophoresis
capillary or an atomic or mass spectrometer are an important consideration.
Syringe injection was identified in a variety of contexts in Section 11.3.3; it can
be automated. Transfer of the drop from a capillary was also identified [103].
The transfer of small volumes of solution into capillary liquid chromatographic
columns and capillary electrophoresis columns is an area of active research and
development.

328
 Since the drop to be analyzed after Single Drop-LPME is composed of a
water-immiscible organic solvent it is readily compatible with direct injection in
gas chromatography, normal phase liquid chromatography, mass spectrometry
and atomic spectrometry. However, it may not be compatible with direct injec-
tion in capillary electrophoresis or reversed-phase liquid chromatography where
the carrier fluid is mainly aqueous. One way of dealing with this potential
problem is to evaporate the drop, perhaps on a small wire loop [107,108] and
redissolve it in a suitable aqueous solvent. Another way is via back-extraction of
the analyte component into an aqueous solution. This latter technique is the
subject of Section 11.4.
It has generally been considered impractical to perform headspaceanalysis
of volatile analyte components above an aqueous sample solution using
Microdrop-LPME in which the microdrop would be suspended in the air above
the aqueous phase. However, very recently it has been demonstrated to be quite
feasible [109]. A 1-1l drop of 1-octanol suspended from the tip of a microliter
syringe in the headspace was shown to be an excellent preconcentration medium
for the BTEX components (benzene, toluene, ethylbenzene and o-xylene) in a
stirred aqueous sample. Quantification of analyte concentration in the micro-
drop was achieved by split-mode injection into a gas chromatograph using a
quadrupole mass spectrometer for detection, with the aid of n-decane as internal
standard. Detailed extraction rate studies of C,, vs. time showed that the
extraction rate is limited by mass transfer in both the aqueous sample solution
and the organic drop. In headspace analysis the drop is not in contact with the
stirred aqueous sample solution so that the drop cannot be convected by momen-
tum transfer. When desired, the drop was convected by gently vibrating the
needle from which it was suspended. This study opens the realm of headspace
analysis for volatile organic compounds (VOCs) by Microdrop-LPME.
In this chapter the emphasis is on the theoretical aspects of LLE, an
understanding of which is useful, and sometimes essential, in designing, opti-
mizing and troubleshooting an LLE technique. However, once a procedure has
been established it is not at all necessary to deal with equations such as Eq.
(11.10) and it is not necessary to know the values of parameters such as D, D,, 6
and A, in order to perform routine LLE operations, such as Microdrop-LPME,
for sample preparation in quantitative analysis. In routine use standards are
always run through the same extraction steps as the sample.

11.4 UNSUPPORTED LIQUID MEMBRANE TECHNIQUES WITH

THREE PHASES

11.4.1 Introduction

When a layer of a solvent is located between two layers of another, immiscible

solvent, it is often referred to as a liquid membrane. If the liquid membrane is
free of any supporting solid matrix such as a porous plastic, then the membrane

329
 I I
II ... nr LI
NaH 2PO4 (pH2.1)
.........a2 ) -
-4 Teflon Ring
Fig. 11.15. Unsupported Membrane-LPME in which the Y N
n-octane membrane separates the stirred aqueous sample al N 80 or 40 pL
solution from a layer of aqueous receiving phase. The 14Q n-Octane
wetting properties of the liquid membrane keep it
touching the Teflon ring. The shape of the membrane . \n .= -
results from frictional drag of the toroidally circulating
aqueous sample that can be seen in Fig. 11.16b. From Ref.
[9], with permission. V Aqueous Sample
(pH13)

is said to be unsupported. It can be that the membrane phase is aqueous while

the phases on either side of it are organic [110] or it can be the reverse [111].
Analyte components starting initially in the sample solution on one side of the
membrane extract into the liquid membrane and simultaneously back-extract
out of the membrane into the receiving phase on the other side. If the volume of
the membrane phase is relatively large (e.g. ml) then a Y-shaped tube or an
inverted Y-shaped tube may be used to allow gravity to keep the three phases
intact, with a more dense membrane located at the bottom of the Y tube, or a less
dense membrane located at the top of an inverted Y tube. Direct stirring may be
performed in none, one, two or all three of the liquid phases.
However, if the volume of the membrane phase is small (e.g. tpl), it is not
necessary to employ bent tubes to prevent one of the phases from falling through
the liquid membrane. Rather, it is possible to use the wetting properties of the
membrane phase, which are the opposite of the wetting properties of the sample
and receiving phases. Since the use of LLE in sample preparation implies that
the sample solution is aqueous, we will continue the description with the
assumption that the liquid membrane is an organic solvent. Shown in Fig. 11.15
is an unsupported liquid membrane (40 or 80 txl of n-octane) system in a 1-ml
microreaction vial. A snug-fitting ring of Teflon has been slid down the inside
wall and positioned at the level where the n-octane membrane will be located.
Since the n-octane wets the Teflon, the membrane layer resists dislodgment and
the upper aqueous phase does not fall through it. A feature of this Unsupported
Membrane-Liquid Phase Microextraction (Unsupported Membrane-LPME)
technique is the fact that direct convection (stirring) in one phase results in
indirectly induced convection in the other two phases as a result of momentum
transfer across both LL interfaces. The convected flow patterns are illustrated in
Fig. 11.16B.
In order to use the Unsupported Membrane-LPME technique for simulta-
neous forward and back extraction to transport an analyte component from the
aqueous sample phase (al) into the aqueous receiving phase (a2), it is necessary
that a2 and al have different compositions, such as to produce a large value of
D, 1, between the membrane and al and a small value of D, 2 between the mem-

330
 a k

Fig. 11.16. Diagrammatic representation of: (b) The convective-flow circulation patterns in the
device shown in Fig. 11.15; and (a) the resulting Nernst diffusion films of thickness 6. The
horizontal components of circulation are omitted for clarity. From Ref. [9], with permission.

brane and a2. This condition is typically achieved by a pH difference when the
analyte components are acids or bases, as was discussed in Section 11.1.2.

11.4.2 Equilibrium, kinetics and modes of operation for

unsupported membrane-LPME [9,10]

A series of two reversible extractions is involved. For analyte component I they

can be represented as:
1 ___Io k j
k\2 k 2 (11.35)

where k-k 4 are first-order extraction rate constants. The preconcentration

enrichment factor for I extracted into the receiving phase at equilibrium is given
by:

EF = C1a2,equil Dll at equilibrium (11.36)

Va 1 V,
in which

DI,,= C'' (11.37)

CI,al,equil

and

D 2 = CIo3equil (11.38)
The
fraction
euil of extracted into a2,

The fraction of I extracted into a2 is:

331
 D' V.
f EF at equilibrium (11.39)
al D 2 + DI,, D 2 V + Dll Va2

From Eq. (11.36) it can be seen that a high preconcentrationis favored by very
small D, 2 and Vo < V02 < V,,al, so that EFrtends toward a high limiting value equal
to Vai/Va,. However, if quantitative extraction of I from al into a2 is the goal, Eq.
(11.39) shows that a very small D,2, is desired along with V < V, and V < V,,, 2 so
that fl, 2 tends toward the value 1. Thus, completeness of extraction is favored by
larger V,, while preconcentration is favored by smaller V,,2. This consideration
has led to two variations in design for Unsupported Membrane-LPME which will
be discussed in connection with Figs. 11.15 and 11.18.
The kinetics of extraction can be treated in terms of the Whitman two-film
model at each LL interface, assuming that both hkINT and kCHEM are much larger
than the mass-transfer coefficients for convective-diffusive MT in the three
liquid phases. Shown in Fig. 11.16A are the locations of the four Nernst diffusion
films. The two steps in Eq. (11.35) can be treated as two reversible, consecutive
first-order processes [9]. The four rate constants in Eq. (11.35) can be expressed
as functions of the four mass transfer coefficients, D 0l/60 l, D,o/6, ,, D,6,,o2 and
D,,/6a2. A set of three differential equations can be written and explicitly solved
to obtain theoretical expressions for the concentrations of I in each of the three
phases as a function of time (see Appendix in Ref. [9]).
Shown as points in Fig. 11.17 are experimentally measured rate curves of
C,2, vs. time for the extraction of two drug compounds, mephentermine and
2-phenylethylamine. The solid lines through the points are nonlinear least
squares fit lines of the theoretical equation to the data. Also shown on the figure
are the theoretically-predicted curves for the concentrations in the organic
membrane phase. The extractions represented in Fig. 11.17 involved 1 ml of
stirred (2050 RPM) aqueous sample (al) at pH 13, 80 ul of n-octane liquid
membrane (o) and 200 ul of pH 2.1 aqueous receiving phase (a2). Quantification

Fig. 11.17. Plots of experimentally meas- 6 x 10'

ured concentrations of mephentermine ()
and 2-phenylethylamine () in 200 Al of pH ' 5 x 104
2.1 aqueous receiving phase from 1.0 ml of 0
pH 13 stirred aqueous sample solution in an E 4 x 10 4
apparatus like that in Fig. 11.15. The n-
octane membrane was 80 gLL. The solid lines ' 3 x 10 4
are for nonlinear least squares fits of the
theoretical equation to the data points. The 4 2 x 104
curved dashed and dotted lines are for the e
compounds in the membrane phase and the 0 1 x 10 4
dotted horizontal line corresponds to extrac-
tion equilibrium, +
at +which
AAA..A0 extraction is
XA1A~dA 0
quantltatlve or ootn compounds. r rom ei.
[9], with permission. Stirring Time (min)

332
was performed by injecting 10 Culof the a2 extract into an HPLC. In al, both
drugs exist nearly exclusively as their neutral free-base species, B, while in a2
they exist nearly exclusively as their cationic conjugate acid species, BH + . At pH
13 D, 1 = 45 for mephentermine and D,1 = 1.1 for 2-phenylethylamine; while at
pH 2.1 D1,2 = 3.2x10 7 for mephentermine and D, 2 = 1.8x10 -8 for 2-phenyl-
ethylamine. Note that the extraction is slower for the 2-phenylethylamine, for
which D, is smaller; but that at equilibrium, which is shown by the horizontal
dotted line, even this compound is quantitatively extracted from 1 ml of al into
0.2 ml of a2. This of course is due to the very small value of D1 ,2 at pH = 2.1,
which provides a large thermodynamic driving force for the overall process.
A much simpler rate equation for C,, vs. t than the one given in the Appen-
dix of Ref. [9] and used in Fig. 11.17 can be obtained by employing the steady-
state approximation for C,, in the organic membrane phase [10], at which moles
of I leaving al and entering a2 are approximately the same. This is justified by
the small fraction of I that is in the membrane at any time. The rate equation is:

C a2 Claliniia Val' {1-exp[-k1over(t-tMIg)]} (11.40)

Va2

The first-order rate constant k is given by:

overall - k3 (11.41)
kg +k3
and tag is the "lag time" constant that is related to the time required to reach
diffusional steady-state across the Nernst film [82,112]. (Note that there are two
different steady-state concepts here.) Equation (11.40) can be rearranged to give
the preconcentration enrichment factor as a function of time:

EF- {1 -exp[kvl (t - tg)]} (11.42)

It has been shown that a larger value of kov, 0rl is favored by faster convection
(smaller 6's) and larger Va2 (larger A2).
When high preconcentration is desired, as opposed to quantitative
extraction, then V= should be minimized because the maximum EF I is Va1/V2.
The apparatus shown in Fig. 11.18 employs a microdrop of receiving phase
suspended in the organic liquid membrane. Shown in Fig. 11.19 are extraction
rate curves for mephentermine and 2-phenylethylamine plotted as EF vs. t [10].
Here V., is 1.6 ml (pH 13), V is 30 gl and V, is 0.5 Al (pH 2.5). The solid lines are
for nonlinear least squares fitting of Eq. (11.42) to the data points. The lag time
at which the line intersects the time axis is 1.2 min for mephentermine and 0.7
min for 2-phenylethylamine. The fact that, after their lag times, these two
first-order rate curves deviate only slightly from linearity is an indication that
after 25 min they are still far from extraction equilibrium. This is because the
small V, has resulted in small rate constants (ko-el = 0.024 min- and 0.0056
min' for mephentermine and 2-phenylethylamine, respectively). Nevertheless,

333
Fig. 11.18. Apparatus for Unsupported Membrane-LPME with a microliter drop of receiving
phase suspended in the membrane from the tip of a microsyringe. The shape of the liquid
membrane results from frictional drag by the toroidally circulating aqueous sample phase. From
Ref. [10], with permission.

151

Fig. 11.19. Plots of experimentally measured con- 121

centrations of mephentermine () and 2-phenyl-
ethylalmine () into a 0.5-L1 drop of pH 2.5 t 91
aqueous receiving phase suspended in a 30-41 IL

liquid membrane that is in contact with 1.6 ml of U-

aE 64
C
stirred, pH 13 aqueous sample solution in an 0
4,
apparatus like that in Fig. 11.18. The lines are non- 3C
linear regression fit lines of Eq. (11.42) to the data.
Lag times are seen as the greater than zero inter-
cepts on the time axis. From Ref. [10], with D
Q 5 10 15 20 25
permission. Extraction Time (min)

even after only 10 min, 2-phenylethylamine for which D,1 = 1.1 has already been
preconcentrated over 100 fold. By contrast if only forward extraction from
aqueous pH 13 into n-octane were employed, this compound would have a
maximum possible preconcentration of only 1.1-fold, at equilibrium (Eq. 11.5).

11.4.3 Advantages, problems and future directions

Extraction with simultaneous back extraction yields the analyte components in

an aqueous phase ready for direct injection into a reversed-phase liquid chroma-
tograph or electrophoresis capillary. The back extraction step represents a
means of achieving very high preconcentration even for compounds with low D,,1.
This is because it is often possible to make D,2 very small. The fact that a

334
preconcentrated analyte component is present in a microdrop which can be
injected in toto avoids the need to evaporate the extract prior to injection. The
technique also represents a second clean-up step for separating potentially
interfering sample components from the analyte component. The liquid
membrane itself is replaced for every sample and standard, eliminating concerns
of carry-over and memory. The membrane is very easy to prepare.
On the negative side, back extraction for sample clean up and precon-
centration is practical only when it is possible to make D, 2 < < D,,l by means of
differences in composition of the aqueous sample and receiving phases such as
different pH for acids and bases, the presence of ligands for metal ions and the
employment of other types of secondary chemical equilibria.
The possibility of increasing preconcentration (EF,) dramatically by
replacing microliter drops of Va2 with nanoliter or picoliter drops is very real.
Such drop volumes are compatible with the injection volumes required by
capillary separation techniques. One can easily imagine an electrophoresis
capillary replacing the microsyringe that protrudes into the liquid membrane
phase in Fig. 11.18; with a nanoliter drop of its running buffer replacing the
microliter drop on the syringe. After extraction the nanodrop could be brought
back into the capillary, the capillary removed from the vial and CZE commenced.

11.5 SPECIATION: A WARNING AND AN OPPORTUNITY

11.5.1 Introduction

The fact that an analyte component (I) may be present in an aqueous sample
solution as more than one kinetically labile (i.e. rapidly interchanging) species
was discussed briefly in Section 11.1.2. It led to using the component distribu-
tion ratio D, in place of the species distribution coefficient Ki. Other aspects of the
present discussion were foreshadowed in Section 11.1.8. Here we consider this
situation from another perspective, asking the question: "When I perform a
sample preparation step that involves distribution of the analyte between two or
more phases, have I recovered an amount of analyte that is proportional to the
concentration of the analyte component or to a species of that component?"
These considerations apply equally to LLE, SPE, SPME, purge-and-trap
methods, etc.
Consider for simplicity the example that was started in Section 11.1.2: that
of a basic analyte compound which is present in the sample solution as two
species B and BH+. The fraction of each is determined by the pH. Let us say that
the pH happens to be equal to PK,,BH. Now imagine that we also prepare a
standard of this compound by dissolving the pure free base, B, in water. We
perform LLE on both the sample and standard solutions, completely oblivious of
the fact that in the sample only 50% of the analyte component is present as the
species B, while in the standard nearly 100% is present as B. The extraction
equation is:

335
Ba B (11.43)

for which the species distribution coefficient is iB. Since the species BH+ does not
distribute between the two phases, ,,H+ = 0. We now perform GC on the extracts
from both sample and standard and calculate the concentration of the compound
in the sample as the product of the ratio of sample GC signal to standard GC
signal times the concentration in our original aqueous standard solution. Will we
be determining the component concentration (B + BH+) in the sample or will we
merely have determined the concentration of the species B?
Combining the equilibria

BH + + H20 - =>B a + H0 +
,T' KB (11.44)
Bo,
If the extraction conditions are exhaustive, such as a very large KB, so that D in
Eq. (11.3) is large then f, in Eq. (11.4) could well be close to 1 meaning that the
extraction of B has shifted the acid ionization equilibrium in Eq. (11.44) far to
the right and we have quantitatively removed from the aqueous sample solution
virtually all of both BH + B. That is, we have determined component concen-
tration. We say that the acid-base equilibrium has been completelyperturbedby
the extraction.
On the other hand, if the extraction conditions that we are using yield a very
small B or if we are operating very far from extraction equilibrium then the
extraction of B will be slight so that [B]a,,,eqi = [B]a,iitia, and [B] o, equil = 1KB [B]a,initi.a
and we have determined species B concentration. Here we say that the acid-base
equilibrium has been left completely unperturbed. Obviously, intermediate
extraction conditions yield intermediate degrees of perturbation.
Often the conditions that are used for solvent extraction remove only some
fraction of the analyte component and do not quantitatively (exhaustively)
extract it; while a different fraction is extracted from the standard (see Section
11.1.8). The obvious and most notable condition leading to non-perturbation of
the sample occurs in those techniques in which the aqueous sample solution is
caused continuously to flow past a small drop of organic phase, maintaining the
aqueous concentration at C,ainitia. If the analyte component is present as more
than one species in the sample but not the standard then we will not have
determined the total concentration of the analyte component.
Looked at from a different perspective, an awareness of perturbation and its
causes tells us how we can use LLE (or SPE, SPME, etc.) to measure the
concentration of a particular species in a sample solution. This is called a
speciationmeasurement. One example is the determination of the concentration
of free (i.e. unbound) progesterone species (P) in the presence of the protein
bovine serum albumin (BSA) which binds P, using Microdrop-LPME [113].

336
 The equilibria are:
Pa + BSAa i P. BSAa
I? (11.45)
Po

By using Microdrop-LPME with Vo = 1 l of n-octane suspended from a micro-

syringe needle, and Va = 500 ml of stirred aqueous sample solution, the fraction
of the Pa,initia extracted is small enough that C,a,initia = Cp,a,equil, so that CP,,eqil - Kp
Cp,a,,,ta. Quantification in the drop was by gas chromatography. The measure-
ment is non-perturbing.
In a second Microdrop-LPME speciation measurement, the concentration of
free (hydrated) Cu2+ ion was measured in the presence of the ligand EDTA by
extracting Cu 2+ from a 500-ml stirred aqueous sample solution into a 1-1 l drop of
n-octane containing the ligand LIX63 [114]. Quantification of copper in the drop
was done by direct sample insertion ICP-MS. The method was non-perturbing.
Here the equilibria were:
Cua+ + EDTAa 4-Cu. EDTAa
+
Lix63a (11.46)

Cu(Lix63)o

REFERENCES

1 L.C. Craig and D. Craig, in: A. Weissberger (Ed.), Techniques of Organic Chemistry,
Vol. III, 2nd Edn. Interscience, New York, 1956.
2 H. Irving and R.J.P. Williams, in: I.M. Kolthoff and P.J. Elving (Eds.), Treatise on
Analytical Chemistry, Vol. 3, Part I. Interscience, New York, 1961, Ch. 31.
3 R.F. Majors, LC-GC, 14 (1996) 754.
4 J. Stary, The Solvent Extraction of Metal Chelates. MacMillan, New York, 1964.
5 A. Ringbom, Complexation in Analytical Chemistry. Wiley, New York, 1973.
6 G. Schill, in: J. Marinsky and Y. Marcus (Eds.), Ion Exchange and Solvent Extraction,
Vol. 6. Marcel Dekker, New York, 1974, Ch. 1.
7 M.A. Jeannot and F.F. Cantwell, Anal. Chem., 68 (1996) 2236.
8 P.R. Rony, Separat.Sci., 3 (1968) 239.
9 M. Ma and F.F. Cantwell, Anal. Chem., 70 (1998) 3912.
10 M. Ma and F.F. Cantwell, Anal. Chem., 71 (1999) 388.
11 D.J. Shaw, Introduction to Colloid and Surface Chemistry. Butterworths, London,
1980, Ch. 4 and 10.
12 T.E. Belk, Chem. Eng. Progr., 61 (1965) 72.
13 J.C. Lee and T.D. Hodgson, Chem. Eng. Sci., 23 (1968) 1375.
14 G.S. Laddha and T.E. Degaleesan, in: T.C. Lo, H.M. Baird and C. Hanson (Eds.),
Handbook of Solvent Extraction. Wiley, New York, 1983.
15 R.E. Majors, LC-GC, 14 (1996) 936.
16 P. Danesi and R. Chiarizia, Crit.Revs. Anal. Chem., 10 (1980) 1.

337
17 E.L. Cussler, Diffusion: Mass Transfer in Fluid Systems. Cambridge University
Press, London, 1984.
18 I. Nahringbauer and B. Larsson, Anal. Chim. Acta, 151 (1983) 153.
19 J.T. Davies and E.K. Rideal, Interfacial Phenomena. Academic Press, New York,
1961, Ch. 7.
20 M. Harada, T. Imamura, K. Fujiyoshi and W. Eguchi, J. Chem. Eng. Jpn., 8 (1975)
233.
21 W.J. Albery, J.F. Burke, E.B. Leffler and J. Hadgraft, J. Chem. Soc. FaradayTrans.,
72 (1976) 1618.
22 E.L. Cussler, Diffusion: Mass Transfer in Fluid Systems. Cambridge University
Press, London, 1984, Ch. 9, 11 and 13.
23 0. Levenspiel, Chemical Reaction Engineering. Wiley, New York, 1972, Ch. 11 and
13.
24 G.J. Hanna and R.D. Noble, Chem. Rev., 85 (1985) 583.
25 G.S. Laddha and T.E. Degaleesan, TransportPhenomenain Liquid Extraction. Tata
McGraw-Hill, New Delhi, 1976, Ch. 3, 6, 7 and 15.
26 W. Nitsch and A. Hoffmann, Chem. Ing. Tech., 53 (1981) 367.
27 R. Fleming, R.H. Guy and J. Hadgraft, J. Pharm. Sci., 72 (1983) 142.
28 V.G. Levich, Physicochemical Hydrodynamics, Prentice-Hall, Englewood Cliffs, NJ,
1962, Ch. 8.
29 G. Persaud and F.F. Cantwell, Anal. Chem., 59 (1987) 2.
30 M. Linton and K.L. Sutherland, Proc. 2nd Int. Conf. Surface Activity, 1 (1957) 494.
31 M.J. Slater, in: J.C. Godfrey and M.J. Slater (Eds.), Liquid-Liquid Extraction
Equipment. Wiley, New York, 1994, Ch. 4.
32 R.G. Borwanker and D.T. Wassan, Chem. Eng. Sci., 43 (1988) 1323.
33 D.C. England and J.C. Berg, Am. Inst. Chem. Eng., 17 (1971) 313.
34 G.C. Quintana, in: R.P. Chhabra and D. De Kee (Eds.), Transport Processes in
Bubbles, Drops and Particles. Hemisphere Publishing, New York, 1992, Ch. 4.
35 W. Nitsch and K. Roth, Colloid Polym. Sci., 256 (1978) 1182.
36 W. Nitsch and G. Weber, Chem. Ing. Tech., 48 (1976) 715.
37 L.T. Skeggs, Am. J. Clin. Pathol., 28 (1957) 311.
38 W.B. Furman, Continuous Flow Analysis: Theory and Practice.Marcel Dekker, New
York, 1976.
39 C.J. Patton and A. Wade, in: G.W. Ewing (Ed.), Analytical Instrumentation
Handbook. Dekker, New York, 1990, Ch. 27.
40 V. Wallace, Anal. Biochem., 20 (1967) 411.
41 D.A. Burns, Anal. Chem., 53 (1981) 1403A.
42 L. Valentine, Advances in Automated Analysis, Technicon International Congress,
Vol. II. Mediad, Inc., White Plains, New York, 1970, 87.
43 K.K. Stewart, G.R. Beecher and R.E. Hare, Fed. Proc., 33 (1974) 1439.
44 J. Ruzicka and E.H. Hansen, Anal. Chim. Acta, 78 (1975) 145.
45 B. Karlberg and S. Thelander, Anal. Chim. Acta, 98 (1978) 1.
46 H. Bergamin, J.X. Madeiros, B.F. Reis and E.A.G. Zagatto, Anal. Chim. Acta, 101
(1978) 9.
47 B. Karlberg, P-A. Johansson and S. Thalander, Anal. Chim. Acta, 104 (1979) 21.
48 J.F.M. Kinkel and E. Tomlinson, Int. J. Pharmaceutics, 6 (1980) 261.
49 C.A. Lucy, Ph.D. Thesis, University of Alberta, 1988.
50 L. Fossey and F.F. Cantwell, Anal. Chem., 54 (1982) 1693.
51 V. Kuban, Crit. Rev. Anal. Chem., 22 (1991) 477.
52 C.A. Lucy and S. Varkey, Anal. Chem., 67 (1995) 3036.
53 F.F. Cantwell and J.A. Sweileh, Anal. Chem., 57 (1985) 329.
54 V. Kuban and F. Ingman, Crit. Revs. Anal. Chem., 22 (1991) 491.

338
55 J.A. Sweileh and F.F. Cantwell, Can. J. Chem. 63 (1985) 2559.
56 L. Fossey, Ph.D. Thesis, University of Alberta, 1985.
57 M. Valcarcel, A. Rios and L. Arce, Crit. Revs. Anal. Chem., 28 (1998) 63.
58 C.A. Lucy and F.F. Cantwell, Anal. Chem., 58 (1986) 2727.
59 L. Nord, K. Backstrom, L.G. Danielsson, F. Ingman and B. Karlberg, Anal. Chim.
Acta, 194 (1987) 221.
60 L. Nord, Ph.D. Thesis, Royal Institute of Technology, Stockholm, 1984.
61 C.A. Lucy and F.F. Cantwell, Anal. Chem., 61 (1989) 101.
62 C.A. Lucy and F.F. Cantwell, Anal. Chem., 61 (1989) 107.
63 H.T. Evensen, D.R. Meldrum and D.L. Cunningham, Rev. Sci. Instr., 69 (1998) 519.
64 E. Castaneda, M.Sc. Thesis, University of Alberta, 1986.
65 C.A. Lucy and K.K.C. Young, Anal. Chem., 66 (1994) 2220.
66 C.A. Lucy and B.P. Housermann, Anal. Chim. Acta, 307 (1995) 173.
67 C.C. Lindgren and P.K. Dasgupta, Talanta, 39 (1992) 101.
68 I. Facchin, J.W. Martins, P.G.P. Zamora and C. Pasquini, Anal. Chim. Acta, 285
(1994) 287.
69 K. Carlsson and B. Karlberg, Anal. Chim. Acta, 415 (2000) 1.
70 I. Facchin and C. Pasquini, Anal. Chim. Acta, 308 (1995) 231.
71 L. Nord and B. Karlberg, Anal. Chim. Acta, 164 (1984) 233.
72 K.L. Peterson, B.K. Logan, G.D. Christian and J. Ruzicka, Anal. Chim. Acta, 337
(1997) 99.
73 L. Nord and B. Karlberg, Anal. Chim. Acta, 145 (1983) 151.
74 T.M. Rossi, D.C. Shelby and I.M. Warner, Anal. Chim., 54 (1982) 2056.
75 W. Nitsch, Dechema Monogr., 55 (1965) 143.
76 A. Brodin and A. Agren, Acta Pharm. Suecica, 8 (1971) 609.
77 T. Tanimura, J.J. Pisano, Y.Ito and R.L. Bowman, Science, 169 (1970) 54.
78 Y. Ito, in: N.B. Mandava and Y. Ito (Eds.), Counter-CurrentChromatography:Theory
and Practice. Dekker, New York, 1988, Ch. 3.
79 K. Hostettmann and A. Marston, in N.B. Mandava and Y. Ito (Eds.), Counter-Current
Chromatography:Theory and Practice, Dekker, New York, 1988, Ch. 5.
80 W. Conway, in: N.B. Mandava and Y. Ito (Eds.), Counter-CurrentChromatography:
Theory and Practice. Dekker, New York, 1988, Ch. 3.
81 M.A. Jeannot and F.F. Cantwell, Anal. Chem., 69 (1993) 235.
82 J. Crank, The Mathematicsof Diffusion. Clarendon Press, Oxford, 1975, Ch. 4 and 5.
83 W.J. Moore, Physical Chemistry. Prentice-Hall, New York, 1962, Ch. 9.
84 K. Nakatani, T. Uchida, N. Kitamura and H. Masuhara, J. Electroanal.Chem., 375
(1994) 383.
85 P.C. Blokker, Proc. 2nd Int. Conf. Surface Activity, 1 (1957) 503.
86 A.J. Bard and L.R. Faulkner, Electrochemical Methods: Fundamentals and
Applications. Wiley, New York, 1980, Ch. 5.
87 K. Nakatani, T. Uchida, H. Misswa, N. Kitamura and H. Mashuhara, J. Phys. Chem.,
97 (1993) 5197.
88 K. Nakatani, M. Wakabayshi, K. Chikama and N. Kitamura, J. Phys. Chem., 100
(1996) 6749.
89 K. Nakatani, M. Sudo and N. Kitamura, Anal. Chem., 72 (2000) 339.
90 M.A. Jeannot, Ph.D. Thesis, University of Alberta, 1997.
91 P.C. Hiementz, Principles of Colloid and Surface Chemistry. Dekker New York,
1986, Ch. 6.
92 K.E. Miller and R.E. Synovec, Talanta, 51 (2000) 921.
93 L.S. de Jager and A.R.J. Andrews, Chromatographia,50 (1999) 733.
94 L.S. de Jager and A.R.J. Andrews, Analyst (London), 125 (2000) 1943.
95 L.S. de Jager and A.R.J. Andrews, J. Chromatogr.A, 911 (2001) 97.

339
 96 Y. He and H.K. Lee, Anal. Chem., 69 (1997) 4634.
97 Y. Wang, Y.C. Kwok, Y. He and H.K. Lee, Anal. Chem., 70 (1998) 4610.
98 W. Liu and H.K. Lee, Anal. Chem., 72 (2000) 4462.
99 H. Liu and P.K. Dasgupta, Anal. Chem., 68 (1996) 1817.
100 E. Welter and B. Neidhart, Fres. J. Anal. Chem., 357 (1997) 345.
101 R. Eberhardt and B. Neidhart, Fres. J. Anal. Chem., 365 (1999) 475.
102 T. Ligor and B. Buszewski, ChromatographiaSupplement, 51 (2000) S279.
103 B.O. Keller and L. Li, Anal. Chem., 73 (2001) 2929.
104 0. Kogi, H.B. Kim and N. Kitamura, Anal. Chim. Acta, 418 (2000) 129.
105 K. Nakatani, T. Uchida, H. Misawa, N. Kitamura and H. Masuhara, J. Electroanal.
Chem., 367 (1994) 109.
106 K. Nakatani, M. Sudo and N. Kitamura, J. Phys. Chem. B, 102 (1998) 2908.
107 P.K. Dasgupta and K. Surowiec, Anal. Chem., 68 (1996) 4291.
108 P.K. Dasgupta, K. Surowiec, Anal. Chem., 68 (1996) 1164.
109 A.L. Theis, A.J. Waldack, S.M. Hansen and M.A. Jeannot, Anal. Chem. ASAP Article
AC015569c, Web Release Date October 30, 2001.
110 S. Watanabe, T. Tani, S. Watanabe and M. Seno, Biochim. Biophys. Acta, 1073 (1991)
275.
111 M. Ma, D. He, Q. Wang and Q. Xie, Talanta, 55 (2001) 1109.
112 R.H. Guy and R. Fleming, J. Colloid Interface Sci., 83 (1981) 130.
113 M.A. Jeannot and F.F. Cantwell, Anal. Chem., 69 (1997) 2935.
114 Y. Wang, Ph.D. Thesis, University of Alberta, 2000.

340
Chapter 12

Principles and practice of solid-phase

extraction
Colin F. Poole

12.1 INTRODUCTION

Solid-phase extraction (SPE) is a method used for the isolation and concentra-
tion of selected analytes from a gas, fluid or liquid (usually) flowing sample
stream by their transfer to and retention (sorption) on a solid phase. The solid
phase is then isolated from the sample and the analytes recovered by elution
using a liquid or fluid, or by thermal desorption into the gas phase. Occasionally
the sampling step is performed by suspension of a suitable SPE device in a liquid,
fluid or gas phase long enough to reach equilibrium with the sample, or some
other reproducible end-point, followed by mechanical separation of the solid
phase from the sample. The analytes are then recovered by elution or thermal
desorption, as before. The principal goals of SPE are trace enrichment (concen-
tration), matrix simplification (sample clean-up) and medium exchange
(transfer from the sample matrix to a different solvent or to the gas phase). SPE
was initially developed as a complement or replacement for liquid-liquid extrac-
tion (LLE). It is now the most common sampling technique in many areas of
chemistry, including environmental, pharmaceutical, clinical, food and indus-
trial chemistry. Over time various sampling formats and sorbents have been
developed to facilitate the convenient processing of different sample types and to
extend the scope of the method. A high level of automation is also possible using
robotics or on-line interfaces to separation and spectroscopic instruments. Full
system integration for unattended extraction, separation and detection is
possible although flexible, inexpensive manual sample processing remains the
more common practice in most laboratories.
In this chapter I will describe the principles and practice of SPE for the
analysis of liquid samples using cartridge and disc devices. SPE using cartridges
for atmospheric sampling and solid-phase microextraction, in-tube solid-phase
microextraction and stir bar extraction are discussed elsewhere in this book,
while specific applications of SPE can be found in the applications section and in
Refs. [1-6].

Comprehensive Analytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 341
12.2 EVOLUTION OF SOLID-PHASE EXTRACTION

12.2.1 Historical development

Fortuitous application of the SPE principle predates written history, but its gen-
eral use in laboratory and field sciences is more recent. Inorganic oxide columns
were used from the 1930s for the clean-up of samples in organic solvents, but
this was rarely considered an extraction process. Probably the first significant
analytical application of SPE was the use of metal pipes filled with activated car-
bon in the early 1950s for the isolation of organic contaminants from various
water sources [7]. These studies were performed to isolate sufficient material for
structural identification with the relatively insensitive instruments available at

TABLE 12.1
Milestones in the development of solid-phase extraction
Early 1950s Pipes filled with activated charcoal used to isolate microcontaminants from
surface waters.
Late 1960s Macroreticular porous polymers introduced as a replacement for active charcoal.
Standardized protocols developed using small columns. Applications to biological
fluids and air sampling developed.
Mid 1970s Silica-based chemically bonded phases used in solid-phase extraction and
eventually dominate the practice of solid-phase extraction. Disposable cartridges
introduced in 1978 and the open syringe barrel format in 1979.
Early 1980s Short precolumns introduced for automated solid-phase extraction and liquid
chromatography. This did not become a routine technique until the early 1990s.
Mid 1980s Start of the development of sorbents with improved molecular recognition
properties. Mixed mode (reversed phase/ion exchange) sorbents introduced for
improved isolation of drugs from biological fluids (1985). Restricted access
materials introduced for the selective isolation of drugs in biological fluids
without prior removal of proteins (1985). Immunosorbents developed to take
advantage of the specificity of antibodies for the isolation of target analytes from
complex mixtures (Biomedical applications developed in the early 1980s followed
by specific procedures for food contaminants and then environmental
applications by the mid-1990s). Silica or other supports with tethered mixed
oxygen-nitrogen donor cryptands introduced for the selective isolation of metals
by a molecular recognition mechanism (1988).
1989 Matrix dispersion introduced as a sample processing technique.
1989 Particle-loaded membranes for disc extraction introduced. Soon followed by
particle-embedded glass fiber discs and laminar discs. Starts the development of
disc technology as a complement to cartridge sampling devices.
1992 Solid-phase microextraction is brought to commercial fruition.
1994 Molecularly imprinted polymers introduced as synthetic analogs of
immunosorbants.
Late 1990s Multiwell plates for high throughput sample processing introduced. Stir bar
extraction (1999) and in-tube solid-phase microextraction (early work in the
1980s but no general acceptance until late 1990s as an interface for microcolumn
liquid and gas chromatography) continue the development of new formats for
convenient sampling using a solid-phase sorption mechanism.

342
that time and for toxicity assessment. The large volume of water generally sam-
pled (>1000 liters over several days) meant that LLE procedures were inappro-
priate or difficult to implement. Since then a number of different sorbents and
formats for SPE have been introduced, as summarized in Table 12.1.
The heterogeneous nature of activated carbons available at that time ham-
pered any extensive development of SPE. These materials had poorly defined
properties characterized by variable and weak adsorption of some analytes, irre-
versible adsorption and low recovery of others, and occasional chemical modifi-
cation of some analytes [8]. The introduction of poly(styrene-divinylbenzene)
and poly(methacrylate) macroreticular porous polymer beads, prepared by sus-
pension polymerization, in the late 1960s was responsible for rekindling interest
in the SPE of organic contaminants from water and for extending the scope of
SPE to air sampling and the analysis of biological fluids [6,7,9-13]. Polymer
sorbents played a key role in the development of purge-and-trap techniques for
the routine determination of volatile organic compounds in water [14,15]. These
chemically inert sorbents had reasonable mechanical strength (at least com-
pared to gels), with a large surface area, high sample capacity, low water reten-
tion and provided high sample recoveries by either solvent elution or thermal
desorption. Compared with carbon analyte recovery was generally better and
irreversible adsorption and catalytic activity greatly diminished. Standardized
sampling procedures using small columns, early versions of modern cartridge
devices, were introduced but the problem of extract contamination by sorbent
impurities remained a persistent problem.
SPE for liquid samples became a widely used laboratory technique in the
early 1980s with the introduction of disposable cartridges containing silica-
based chemically bonded sorbents of a suitable size for sample processing by
gentle suction. Typical cartridge devices consist of short columns (generally an
open syringe barrel) containing a sorbent with a nominal particle size of 50-60
Am, packed between porous plastic or metal frits (Fig. 12.1) [1-4,6,16-19]. A
large number of non-selective and selective sorbents (including restricted access
media, mixed-mode sorbents, immunosorbents, molecularly imprinted poly-
mers, chemically bonded cryptands, etc.) are available today providing for the
diverse application base of modern SPE. Low volume cartridge or precolumn
devices are the basis of on-line hyphenated systems (SPE-LC, SPE-GC) for

Sample
(polyl

Fritted
(20pmI

Sorben

Fritted
Luer tip

Fig. 12.1. Schematic diagram showing the typical construction of a solid-phase extraction
cartridge and a vacuum manifold for parallel sample processing.

343
 Reservoir

Clamn
" ' r
disk.

Collection
( tube

Fig. 12.2. Typical cartridge and vacuum filtration formats for solid-phase extraction using discs.

integrated sample processing and separation that matured into robust practical
systems in the mid-1990s [6,19-23].
SPE discs were developed initially to provide higher sample processing rates
for large sample volumes and to minimize plugging by suspended particles and
matrix components [24-29]. Since small-diameter discs are easily prepared,
discs are also favored for handling small-volume samples as well. At least three
different disc formats are offered today. Particle-loaded membranes contain
sorbent particles of 8-12-utm diameter immobilized in a web of short poly(tetra-
fluoroethylene) (PTFE) fibrils [24-27]. These are formed into 0.5-mm thick discs
of various sizes containing about 90% by weight of sorbent. The discs are flexible
and superficially resemble filter paper discs. They are used with some support-
ing structure such as a porous glass or plastic support (Fig. 12.2). Particle-loaded
membranes are also available in a conventional cartridge format. In this case,
the sorbent bed contains particles of a larger size, about 50 Arm, in a thicker disc
of about 1.0 mm, sealed into the base of an open syringe barrel. These discs have
an integral poly(propylene) prefilter attached to their top surface. Parti-
cle-embedded glass fiber discs contain particles of about 10-30 Am in diameter
woven into a glass fiber-supporting matrix [28,29]. Small diameter discs are
self-supporting but larger discs require a supporting structure similar to the par-
ticle-loaded membranes. Particle-embedded glass fiber discs are also available
with an integral prefilter attached to their top surface. Laminar discs contain
10-,tm sorbent particles in a consolidated 0.5- or 1.0-mm thick bed (usually)
retained by two glass fiber filters held in place by screens and a retaining ring in
a preassembled cartridge [301]. Disc technology has contributed directly to the
automation of SPE through development of the multiwell extraction plate (Fig.
12.3), which is used for isolation and sample clean-up in high-throughput screen-
ing techniques, such as combinatorial chemistry and drug development [31-36].

12.2.2 As a replacement for liquid-liquid extraction

SPE was initially considered as a replacement for liquid-liquid extraction (LLE).

Conventional LLE is labor intensive, difficult to automate, and is frequently

344
 Fig. 12.3. Multiwell plate for automated SPE. (From Ref. [31]; copyright Elsevier).

plagued by practical problems, such as emulsion formation. In addition, it

consumes relatively large volumes of high purity solvents with expensive
disposal requirements. In contrast SPE benefits from low intrinsic costs, shorter
processing times, low solvent consumption and simpler processing procedures.
SPE methods are easier to automate using robotics, centrifugal or special
purpose sequential or parallel flow processing units that simultaneously extract
and prepare samples for separation. In addition, SPE has favorable properties
for field sampling, eliminating the need to transport and store bulk samples for
processing by the receiving laboratory.
SPE techniques have their own, although different, problems to those of
LLE. The surface chemistry, and therefore sorption properties, of solid phases
are not as reproducible as solvent properties. Silica-based, siloxane-bonded
sorbents retain basic compounds by a mixed retention mechanism involving
low-selectivity sorption by the bonded phase and ion-exchange interactions with
accessible dissociated silanol groups of the silica substrate. The mixed retention
mechanism can interfere in analyte recovery, since elution with neutral solvents
is ineffective for displacing ionically bound analytes, and the contribution of
ionic binding to retention can vary for different sorbents. Solid phases tend to
have a higher level of contamination by manufacturing and packaging materials
than is the case for solvents. The chemical background from impurities may
interfere in the subsequent determination of the analytes. Sample processing
problems in SPE related to the limited sorption capacity of sorbents and analyte
displacement or plugging of sorbent pores by matrix components, easily pass
unnoticed, resulting in changes in analyte recovery. Sample overload and
displacement are more important for SPE than LLE because of the limited
sorption capacity of sorbents compared with the solvation properties of liquids.
A large number of factors have to be considered when selecting SPE or LLE
for a particular problem. These include economic as well as technical features,
whether manual or automated operation is to be pursued and the matrix
chemistry and quantity. It is quite likely that a LLE or SPE method can be
developed for the same problem and a final selection is based on a feasibility

345
study where one technique emerges as a preferred choice from consideration of
the positive and negative features of each method. Increasingly as external
factors dictate a reduction in the use of popular solvents this has tended to favor
SPE. Also, as laboratories look at approaches to increase sample throughput and
reduce labor costs, this has also tended to favor SPE methods. For field sampling
SPE provides a simpler approach than solvent extraction methods.

12.2.3 Choice of cartridge or disc formats

Cartridge and disc devices use the same sorbent technology and are
distinguished only by differences in format. Cartridges are easily prepared in the
laboratory while discs, so far, have only been produced in a manufacturing
setting. Discs, therefore, are only available in a limited range of sorbent chemis-
tries, largely dictated by market forces. Also, it is easier to scale-up cartridge
devices to handle large sample loads and to perform sample clean-up than it is
for discs. Discs are significantly more expensive than cartridges and for
relatively simple, routine applications economic consideration often dictate the
use of cartridges. Discs, on the other hand, offer specific format advantages that
favor their use for some applications.
For large sample volumes containing suspended particles, discs are likely to
function better than cartridges. Discs provide shorter sample processing times
on account of their larger cross-sectional area and decreased pressure drop,
allowing higher sample flow rates. This is important in environmental surveil-
lance programs, such as the analysis of surface waters for trace contaminants,
where large sample volumes are commonly employed to achieve adequate detec-
tion limits.
Discs have a large surface area per unit bed mass. This facilitates their use
for passive sampling by immersing the disc in the sample as opposed to the
conventional approach of passing the sample through the disc in a manner
similar to filtration [37-391. Passive sampling is convenient for both field and
laboratory applications, but has been little explored so far. One reason is the
slow equilibrium of the extraction process even for stirred solutions. The sorp-
tion of organic contaminants from water by an octadecylsiloxane-bonded silica
disc using passive sampling under conditions that avoid sample depletion was
used as an indirect measure of bioconcentration by biota [40,41]. The disc
concentrates the more hydrophobic compounds from water in preference to
hydrophilic compounds and is theorized to be a more realistic approach to
assessing potential environmental consequences of organic contaminants in
complex environmental systems. This is a relatively new approach to toxicity
assessment and requires further work to establish a satisfactory relationship
between the sorption properties of the disc and those of target aquatic species.
Because of the low packing density of typical cartridge devices, longer
sorbent beds than are needed for extraction are used to compensate for reduced
retention resulting from channeling. Increased bed mass results in increased

346
non-specific matrix adsorption and dirtier extracts. The use of smaller particles
and the greater mechanical stability of discs minimize channeling, and the opti-
mized use of bed mass results in a cleaner chemical background due to reduced
matrix adsorption. For small sample sizes it is easier to miniaturize discs than
cartridges, and several disc devices (e.g. microdiscs, pipette tips, etc.) that con-
tain only a few milligrams of sorbent are available. Immobilized analytes on
microdiscs facilitate integrated sample processing techniques such as in-vial
desorption [42-44] and on-disc derivatization [42,44,45]. Solid-phase deriva-
tization techniques for conventional sorbent cartridges are also described [46].
In-vial desorption reduces the amount of organic solvent used for recovery and
eliminates tedious sample preparation steps compared to conventional sample
processing methods. The in-vial desorption technique is an equilibrium-based
approach to sample recovery. After extraction the dried disc is placed in an
autosampler vial and covered with a few ml of solvent. Recovery depends on the
selection of the solvent, temperature and equilibration time. An equilibration
time of 4 h, or overnight, at room temperature is sufficient in many cases to pro-
vide acceptable recovery (> 90%) and method precision as well as being compatible
with automated sample analysis in gas and liquid chromatography. Addition of a
derivatizing reagent to the solvent used for desorption allows derivatization and
analyte recovery to be performed simultaneously in the same vial without removal
of the disc. The vial can be sealed and heated to enhance the rate of reaction with
the derivatizing reagent. Common reactions are trialkylsilylation and alkylation
for gas chromatographic analysis. Ion-exchange discs catalyze the rate of alkyla-
tion of acid herbicides, surfactants and pesticides using a solution of an alkyl
iodide as the derivatizing reagent. An example is shown in Fig. 12.4 [42].
The disc format supports other, if minor, applications at present, such as in
situ detection using radioactivity counting [47], phosphorescence [48] and
MALDI mass spectrometry [49,50]. The particle-embedded glass fiber discs can
be inserted directly into a hole in a glass fiber sheet and developed in a conven-
tional manner for thin-layer chromatography combining recovery and separa-
tion into a single step [51,52]. This approach is the basis of the Toxi-Lab®
system for toxicological screening of biological fluids.

12.3 SORBENT TYPES AND THEIR APPLICATIONS

12.3.1 Inorganic oxide adsorbents

The most important inorganic oxide adsorbents for SPE are silica gel, alumina,
Florisil and diatomaceous earth. Silica gels used for SPE have surface areas of
about 300-800 m2/g, pore sizes from 4-10 nm, and an apparent pH of 5.5-7.5
(measured indirectly as the pH of a 5% (m/m) sorbent suspension in water).
Alumina used for SPE has a surface area of about 150 m2/g and a pore size of 6
nm. Depending on processing conditions it is available as neutral (pH 7.5 ± 0.5),
weakly acidic (pH 6.0 ± 0.5), acidic (pH 4.5 ± 0.5) and basic (pH 9.5 ± 0.5) forms.

347
 C
C
0
o
Q.
2
a
W
O
LUI

Time (min)

Fig. 12.4. Use of in-vial desorption and derivatization for the determination of chlorinated acid
herbicides in an extract from a spiked (< 10 aug/l each) river water sample by gas chroma-
tography. A 500-ml sample was processed through a 25-mm strong anion-exchange disc, the disc
dried by suction, sealed in a autosampler vial with 1 ml of acetonitrile and 0.2 ml of methyl iodide
and heated at 80°C for 1 h, and excess methyl iodide removed by evaporation. (From Ref. [42];
copyright Elsevier).

Florisil is a magnesium silicate with a surface area of about 250-300 m2 /g and an

apparent pH of about 8.5. Diatomaceous earth is a flux-calcined silica with a low
surface area used primarily as a filter aid and dispersant for solvent extraction
by the matrix dispersion technique [53].
The general extraction mechanism and applications of the inorganic oxide
adsorbents are summarized in Table 12.2 [1,4,5,54-57]. Adsorbent properties
that increase retention are a larger surface area and a high activity. Adsorbent
activity is usually controlled by the intentional addition of water to the dried
adsorbent prior to use and by drying samples with anhydrous sodium sulfate, or
a similar drying agent, prior to applying the sample to the adsorbent. A sodium
sulfate cartridge connected in series to the adsorbent cartridge is a convenient
approach for maintaining a constant adsorbent activity during sampling. Anal-
yte properties that increase retention depend on the number and type of func-
tional groups present. Hydrogen-bonding functional groups are strongly
retained (e.g. sulfonic acid, carboxylic acid, phenol, hydroxyl, etc.), those with a
significant dipole-character are retained to a lesser extent (nitro, ester, ketone,
etc.) and polarizable functional groups (e.g. aromatic, alkene, etc.) are the least
retained. Irreversible adsorption and catalytic breakdown of sensitive analytes
can occur on all inorganic oxide adsorbents and is a source of low recovery for
some analytes. Alumina and silica can function as selective ion-exchangers with
buffered aqueous samples (see Table 12.2). Coating inorganic oxides with a
complexing agent (e.g. silver nitrate, caffeine) or an acid or base (an acid can be
used for the selective isolation of bases and vice ersa) can be used to modify
selectivity for the isolation of target compounds.

348
TABLE 12.2
General application of inorganic oxide adsorbents
Uses:
· Isolation of low and medium polarity analytes from non-aqueous solutions
· Isolation of cations (alumina and silica) and anions (alumina) from buffered aqueous
solutions
* Matrix simplification by fractionation into groups containing a similar number and type of
functional group
Examples:
· Isolation of organochlorine pesticides and polychlorinated biphenyls from transformer oil,
animal fats and oils, etc., using Florisil.
* Isolation of lipids by chromatography over silica gel using chloroform to elute simple lipids,
acetone to elute glycolipids and methanol to elute phospholipids [54].
* Group fractionation of polycyclic aromatic compounds (hydrocarbons, N-containing, and
OH-containing) in synthetic fuels over alumina using a step solvent gradient.
* Isolation of paraquat and diquat from high moisture crops in a pH 9 aqueous extract using
silica gel [55,56]
* Mycotoxins in feeds using silica gel
* Pesticides in foods, feeds, and soil extracts [57]; alkaloids, pigments and flavor compounds
from plants; sugars and caffeine in cola beverages, inorganic anions, and organic acids in
aqueous solution using alumina; steroids and vitamins from creams and oil-based
suspensions

12.3.2 Low-specificity sorbents

Low-specificity sorbents are widely used for the isolation of contaminants from
aqueous solution. Because of the prevalence of water as a general solvent in
environmental, biological and industrial chemistry these methods represent the
most common applications of SPE. Typical sorbents include silica-based,
chemically bonded materials, porous polymers [18,58-60] and graphitic carbon
[8,61]. Silica-based, chemically bonded sorbents can be prepared with a wide
range of bonding density, pore sizes and functional group types (Table 12.3).
They are generally synthesized by reaction of monofunctional or trifunctional
silanes with silica gel followed by endcapping in some cases. Trifunctional
reagents result in sorbents with a polymeric bonded layer of higher carbon
loading and greater pH stability. Silica gels of high surface area, 500-600 m2 /g,
are generally used to prepare sorbents for the isolation of small molecules.
Chemically bonded sorbents with high surface areas, long alkyl chains and high
phase loading maximize retention of small molecules from aqueous solution
while wide pore materials with a low phase loading and short alkyl chains are
used to isolate macromolecules (Table 12.4) [1,4,5,62-69]. Chemically bonded
sorbents with polar functional groups are used mainly to isolate analytes from
organic solvents based on their selective interactions with analyte polar
functional groups. They are less retentive than alkylsiloxane-bonded phases for
aqueous samples. When the sampling process is not dominated by breakthrough

349
TABLE 12.3
Structures of silica-based chemically bonded sorbents
Type Functional group Structure
C18 Octadecyl Si-C18H27

C8 Octyl -SiCH,7

C2 Ethyl Si-C2H5

CH Cyclohexyl S

PH Phenyl -i-

CN Cyanopropyl -SiCH2 CH 2CH 2CN

NH2 Aminopropyl Si-CH 2CH 2CH 2NH 2

DIOL 2,3-Dihydroxypropoxypropyl Si-CH2 CH2 CHOCH2 CH-CH 2

OH OH
SAX Trimethylaminopropyl (Quaternary SiHCH CH
2 2N-(CH3)3Cl-
amine)
CBA Carboxypropyl Si-CH2 CH 2CH 2 COOH

SCX Benzenesulfonic acid -Si- SO3t

PRS Propylsulfonic acid Si-CH2CHCHSO,-"H

volume considerations they can be used to minimize interference from matrix

adsorption and to aid analyte recovery in a small solvent volume.
Silica-based, chemically bonded sorbents are the most widely used sorbents
for SPE but are unsuitable for some applications. Breakthrough volumes for
small, highly polar molecules in water are often too small for adequate quantifi-
cation. Silica-based sorbents contain a low concentration of ionized silanol
groups capable of retaining basic solutes by an ion-exchange mechanism [70].
Low recovery of these compounds can result from the inability of the eluting sol-
vents to displace the analytes from the ion-exchange sites. Sometimes the addi-
tion of a competing base to the eluting solvent can solve this problem [4,70,71].
Silica-based sorbents are unstable at pH extremes (2 > pH > 8). Porous polymer
and graphitic carbon sorbents offer a potential solution to these problems. They
are stable throughout the full pH range and do not possess ionized silanol
groups.
Modern porous polymer sorbents are generally copolymers of styrene and
divinylbenzene processed to enhance their properties for SPE [19,58-60]. They

350
TABLE 12.4
Samples in aqueous solution
Samples in Aqueous Solution
* Isolation of neutral and ionic analytes from aqueous solution. Weak acids and bases by ion
suppression. Strong acids and bases using ion-pair extraction (alternative to ion exchange)
* Retention increases with solute size and is reduced by polar interactions (particularly
hydrogen-bonding) and ionization
* Polar bonded phases provide only weak retention and are not particularly useful unless
elution of the analyte is a problem from non-polar sorbents
Examples
* Isolation of agricultural and industrial chemicals from surface waters using C18, Carbon or
PS-DVB
* Isolation of polycyclic aromatic compounds from water by C18 and PS-DVB [62,63]
* Isolation of phenols from water by C18, Carbon and PS-DVB [64]
* Isolation of drugs from biofluids using C18, C8, PS-DVB, or CN [65]
* Isolation of macromolecules from biofluids and fermentation broth using C4
* Isolation of pesticides from biofluids using C18 and C8 [66,67]
· Isolation of antibiotics from biomatrices using C18 and C8 [68]
* Isolation of pigments and coloring materials from beverages and food extracts using C18
* Isolation of carbohydrates and nucleosides from biofluids using AMINO
* Isolation of proteins, peptides and surfactants using DIOL
Samples in Organic Solvents
* Retention depends on the type and number of functional groups. Solute size is less
important.
CN Strong dipole-type interactions and weak hydrogen-bond acidity
AMINO Strong hydrogen-bond base, weak hydrogen-bond acid and weak dipole-type
interactions.
DIOL Strong hydrogen-bond acid, weak hydrogen-bond base and significant
dipole-type interactions
Examples
· Isolation of polar pesticides from fats and oils
· Isolation of polycyclic aromatic compounds from fuel oils
Active ingredients from ointments and suppositories

are either similar in chemistry to porous polymers developed for HPLC and have
moderate surface areas (< 600 m2 /g), or are biporous and highly crosslinked with
surface areas from 700-1200 m2/g. The larger surface areas of the highly
crosslinked polymers results in higher retention. The highly strained porous
structure is swollen to different extents by aqueous and organic solvents and is
partially responsible for the favorable retention and elution properties of these
materials [58]. Porous polymer particle-loaded membranes are selectively
solvated by organic solvents, which are added to the sample as a processing aid
(1% v/v). The solvated polymers exhibit significant selectivity differences result-
ing in significant compound-specific changes in breakthrough volumes (Table
12.5) [72]. The retention properties of poly(styrene-divinylbenzene) polymers
can also be increased by light surface modification with polar functional groups,
such as acetyl or sulfonate [59,73]. The polar substituents reduce the interfacial

351
TABLE 12.5
Breakthrough volumes (cm 3) of some organic compounds on porous polymer particle-loaded
membrane discs with 1% (v/v) organic solvent in water
Compound Processing solvent
Acetonitrile Tetrahydrofuran 2-Propanol Methanol
Anisole 575 300 1750 1300
Acetanilide 95 25 100 75
2-Phenylethanol 150 25 100 200
Heptan-l-ol 2700 300 2700 1000
Propyl propanoate 1400 200 1250 750

tension between the polymer surface and aqueous sample increasing contact
between the analyte and polymeric sorbent. For sulfonated polymers increased
retention of neutral polar compounds was strongly dependent on the concentra-
tion of sulfonate groups with an optimum concentration of 0.6 mmol sulfonate/g
providing the largest increase in retention. The same goals are achieved by a
macroporous poly(divinylbenzene-co-N-vinylpyrrolidone) polymer (Oasis TI
HLB), which has a surface area of about 800 m2/g [74]. The sulfonated polymers
and vinylpyrrolidone copolymer have the added advantage of being fully water
wettable. Consequently, a sorbent conditioning step prior to sample application
is no longer required and samples can be fully recovered even if the sampling
device runs dry during sample processing.
The modern forms of carbon used for SPE are graphitized carbon blacks and
porous graphitic carbon [8,61,75-78]. Graphitized carbon blacks are (largely)
non-porous with moderate surface areas of 100-210 m2/g. Their surfaces are con-
taminated by oxygen complexes (hydroquinone, quinones, chromene) and benz-
pyrylium salts. Analyte retention occurs by a combination of adsorption and
anion exchange. Porous graphitic carbon is a more refined material developed
for HPLC and produced in a larger particle size for SPE. The retention mecha-
nism is different to chemically bonded phases with high retention obtained for
planar molecules containing several polar functional groups and for systems rich
in polarizable electrons [8,77,78]. Recovery of adsorbed analytes from carbon
sorbents using conventional solvents, such as methanol and acetonitrile, is often
more difficult than for chemically bonded sorbents. Stronger (less polar) binary
solvent mixtures are often more efficient and desorption by backflushing is rec-
ommended for well-retained compounds.
Low-specificity sorbents are commonly used for multiresidue analysis of
pesticides, herbicides, etc., and their decomposition products in environmental
and food samples [30,79-82]. These applications are relatively recent but
represent a natural extension of SPE methods for use in surveillance programs.
No single sorbent is considered ideal for these applications at present, sustaining
interest in the development of new sorbents with optimized retention properties

352
to meet regulatory requirements. Ion-pair extraction is a useful approach for
isolating ionized compounds by low-specificity sorbents but is not widely used at
present [83].

12.3.3 Compound and class-specific sorbents

Various selective sorbents based on ion exchange, bioaffinity, molecular

recognition, and restricted access materials are used to supplement the sorbents
discussed above. Ion exchange is used to isolate inorganic and organic ions
(usually) from aqueous solution with sorbents containing fixed ionic sites of
opposite charge to the analytes of interest (Table 12.6) [1,4,5,84]. Ion-exchange
sorbents are usually classified as weak or strong depending on the identity of the
ionic group and whether its charge is independent of the sample pH (strong ion
exchanger) or can be manipulated by changing pH (weak ion exchanger). Some
examples of typical silica-based, ion-exchange sorbents are indicated in Table
12.3. Porous polymer, ion-exchange resins are also commonly used in SPE and
have a higher exchange capacity and a wider pH operating range than silica-
based materials. For many applications silica-based and porous polymer ion-
exchange materials with the same immobilized ionic groups can be used inter-
changeably, although because of non-specific adsorption of matrix components
the chemical background of the extracts might be different. Ion-exchange
materials are particularly attractive for the isolation of ions since the neutral
molecules, which may interfere in the final chromatographic analysis, are easily
rinsed from the ion exchanger without affecting the recovery of the ions.
In recent years disc-based SPE has become a popular technique for sample
clean-up in ion chromatography and capillary electrophoresis [85,86]. Inter-
fering ions, often at a relatively high concentration, can mask, broaden or
change the migration time of the ions of interest. Removal of interfering ions is

TABLE 12.6
General applications of ion-exchange sorbents
· In general strong ion exchanges are used to isolate weak acid/bases of opposite charge and
weak ion exchanges strong acid/bases
· Retention selectivity can be adjusted by manipulating the sample pH and ionic strength
· Choice of competing ion, its concentration and eluent pH controls selectivity for matrix
simplification and elution
· Isolation of macromolecules in an active form may require special non-denaturing sorbents
based on cellulose, agarose or dextran
Examples
· Isolation of carboxylic, sulfonic and phosphoric acids, phenols, amines and inorganic ions
from water
· Isolation of amino acids, organic acids, nucleosides and nucleotides from biofluids
· Isolation of organic acids and bases from coal-derived and synthetic fuels
Isolation of organic acids, phenols and amines from wine, fruit juices and food extracts
* Isolation of acid herbicides from water [84]

353
performed by disc extraction using a sulfonic acid-functionalized resin in the
hydrogen, silver or barium forms. The hydrogen form is used to remove hydrox-
ide and carbonate ions from samples by neutralization. The silver form is useful
for the removal of excess halide ions from samples by formation of insoluble sil-
ver halide salts that precipitate in the disk matrix. The barium form is used to
remove sulfate from samples through the formation of insoluble barium sulfate.

12.3.3.1 Mixed-mode and multimode extractions

Mixed-mode sorbents containing co-bonded ion-exchange and alkyl groups in
either cartridge or disc format are popular in clinical and pharmaceutical
laboratories, where they are used for the isolation of drugs with ionizable
functional groups and their metabolites from biological fluids [4,5]. Standard
protocols using mixed-mode sorbents have been developed for the isolation of
most drugs of abuse (e.g., amphetamines, barbiturates, cocaine, opiates, etc.)
and for classifying extracts into acid/neutral and basic drugs for systematic
toxicological analysis [86-90]. Drugs are initially retained by a reversed-phase
mechanism and excess salts and polar neutral molecules (e.g. urea) are washed
from the sorbent. An acid rinse is used to protonate bases increasing their
retention by an ion-exchange mechanism. Elution with an organic solvent
selectively isolates the neutral and acid fraction. Subsequent elution with an
aqueous-organic eluent containing a base elutes the basic drugs. The strong
retention and the use of efficient rinse solvents result in cleaner extracts
compared with single mode sorbents, suitable for screening by thin-layer chro-
matography and confirmation by gas chromatography-mass spectrometry or
liquid chromatography-mass spectrometry.
Discs or cartridges with different sorbent chemistries can be staked on top of
each other and analytes recovered with a single elution or as fractions of
different chemistries after physically separating the discs or cartridges [3,4,
43,91]. This approach is useful for complex samples containing analytes that
differ significantly in polarity or ionization, for optimizing the breakthrough
volume of analytes, and for reducing interference in the chromatogram by the
selective sorption of the matrix by the first SPE device and the analytes by the
other. Because of ease of stacking discs with minimal dead volume this approach
is more common using discs than cartridges, but in both cases general use is less
than might be anticipated given the simplicity of the procedure. Stacked discs of
a reversed phase and cation exchange type are a suitable alternative to mixed-
mode sorbents for isolating drugs and their metabolites from biological fluids.
Cation-exchange discs have been used to remove humic and fulvic acids from
surface waters and soil extracts with retention of phenylurea and triazine
herbicides on a octadecylsiloxane-bonded silica disc [92]. A combination of a
octadecylsiloxane-bonded silica disc and a carbon disc was used to isolate
N-nitrosodimethylamine from water in which the octadecylsiloxane-bonded
silica disc was used to sorb matrix components and set aside before elution of the
analyte from the carbon disc [931].

354
12.3.3.2 Restricted access materials
Restricted access sorbents were initially developed for the isolation of low-
molecular-mass drugs from biological fluids with minimum sample pretreat-
ment [94-100]. Recent applications include the isolation of acidic herbicides
from surface waters burdened by humic and fulvic acids [101]. Restricted access
materials work by preventing access of macromolecules (proteins) to those
regions of the sorbent where analyte retention occurs (Fig. 12.5). Restricted
access to the retentive part of the sorbent is provided by either a physical
diffusion barrier, such as a pore diameter in internal surface reversed-phase
sorbents, or by a chemical diffusion barrier, such as a polymer network at the
outer surface of the particle in semi-permeable surface materials. In addition,
the outer surface of the particles must be non-adsorptive and matrix-compatible.
Restricted access sorbents are commonly used for automated on-line sample
processing in liquid chromatography. In this case a short precolumn (0.5 to 3.0
cm) packed with the restricted access material is interfaced to a separation
column by a multiport switching valve (see Section 12.6.1). The biological fluid is
injected directly onto the precolumn, which retains the analytes of interest.
Interfering macromolecules (proteins) pass through the precolumn unretained
and are flushed to waste. The analytes retained on the precolumn are eluted
on-line to the separation column and detected. Simultaneously the precolumn is
reconditioned (or exchanged) before processing the next sample. An important
consideration for automated sample processing is the ability of the restricted

InternalSurface Reversed Phase (ISRP) concept

I l

...
;.k kr---;rkir
""" - -~

workof
rophilic
r phase

Hydrophobic
inner phase

Semi PermeableSurface (SPS) concept

Fig. 12.5. Schematic representation of the separation mechanism of restricted access materials.
(From Ref. [101]; copyright Elsevier).

355
access material to repeatedly extract the analyte without change in properties or
accumulation of sample matrix components.

12.3.3.3 Surface-bound macrocyclic ligands

Silica and porous polymer sorbents with surface-bound macrocyclic ligands can
be used for the selective isolation and concentration of alkali-metal, alkaline-
earth-metal, post-transition metal ions, and some anions from aqueous solution
and complex sample matrices of high ionic strength (e.g. seawater) or extreme
pH [47,102-105]. It is particularly useful for those conditions where ion
exchange is likely to be ineffective because of a lack of retention or selectivity.
Metal complexation by macrocyclic ligands is generally tolerant of extreme pH
and ionic strength. The stability of the host-guest complex, and therefore the
selectivity of the isolation process, depends on matching the metal ion radius to
the size of the macrocycle cavity and the selection of the donor atom types. The
metal ion is sorbed in the cavity of the macrocycle during the sample application
phase and later released by elution with a solution of a complexing agent with a
higher binding constant for the metal. Both processes provide a high level of
specificity allowing sorbents to be developed for specific applications. One signif-
icant application is the use of RAD discs in the nuclear industry for analyzing
groundwater and nuclear fuel storage-pool water for trace quantities of radio-
nuclides such as strontium, radium, technetium, etc. [47,105].

12.3.3.4 Surface-boundphenylboronic acids

Phenylboronic acid groups covalently attached to silica or porous polymer
supports are used for the selective isolation of compounds containing vicinal diol
groups (e.g., nucleosides, catecholamines, steroids, drugs, etc.), a-hydroxy acids,
1,2-aminoalcohols, 1,2-diketones, 1,3-diols, etc., by formation of 5- or 6-mem-
bered cyclic boronate complexes [4,5,106,107]. The selectivity of the retention
mechanism depends on the stability of the covalent complex, which is largely
governed by ring size, functional group type and pH. Complex formation
requires a basic pH. This is achieved by activating the sorbent with an alkaline
buffer at pH 10-12 followed by a rinse and application of the sample at (usually)
a pH of 8.0-8.5. Zwitterionic buffers (Good's buffers) are typically used for pH
control. After formation of the covalent complexes, the cartridge is washed with
alkaline aqueous-organic solvent mixtures to remove non-specifically adsorbed
matrix components. The analytes are recovered by elution with an acidic
aqueous-organic solvent mixture (pH < 5), which decomposes the covalent
complexes releasing the bound analytes.

12.3.3.5 Immunosorbents
Immunosorbents (or immunoaffinity sorbents) have been used for a long time
for sample pretreatment in medicine, biology and food science but more general
applications, such as to environmental samples, are relatively recent [108-113].
In part this is due to the difficulty of making antibodies selective to small

356
molecules as well as a lack of familiarity among analytical chemists with the
procedures used to make specific antibodies. The interest in immunosorbent
extraction results from its high selectivity, which allows extraction, concentra-
tion and clean-up from complex matrices in a single step and from large sample
volumes when required. The high degree of molecular selectivity is due to the
specificity of the antibody-antigen (analyte) spatial fitting and interactions.
Cross-reactivity, considered a disadvantage in the development of single
compound schemes, is advantageous for multiresidue methods. Class-specific
immunosorbents for mycotoxins, phenylurea herbicides, and polycyclic aromatic
hydrocarbons, for example, are commercially available. Manufactured immuno-
sorbents have been available for a relatively short time and the range of products
remains narrow. Most applications continue to emanate from a few research
groups. It is clear, however, that immunosorbents have a good potential for use
with difficult sample matrices. Products based on immunochemistry have been
well received in other areas of analytical chemistry [114].
Immunosorbents are prepared by covalently bonding a suitable antibody to
an appropriate sorbent. These sorbents are used in disposable cartridges or as
short precolumns in coupled-column liquid chromatography, which has received
the most attention in research laboratories. For the latter application it is
important that the sorbent is pressure stable and the precolumn reusable for
many samples. Coupled-column SPE-GC is not as advanced [115] as SPE-LC but
has good potential for future growth.
The breakthrough volume of an immunosorbent cartridge or precolumn
depends on the total number of accessible binding sites and on the analyte-anti-
body binding constant. It is desirable to minimize non-specific analyte and
matrix interactions with the support. As a consequence there is no universal
antibody support suitable for all applications. The analyte capacity of immuno-
sorbents is generally low, but since these materials are usually used for trace
analysis and their selectivity for the target analytes is high, this is not normally a
problem. The strong analyte-antibody binding facilitates rinsing the sorbent
with solvents to eliminate matrix interference. Finding conditions that release
the analytes in a small elution volume can be problematic, particularly if the
immunosorbent is to be reused. Aqueous organic solvent mixtures are commonly
used for elution but may denature the antibodies; alternatives include the use of
competitive binding (displacing) agents, change in pH or temperature variation.
It is clear that the selection of sample processing conditions with immuno-
sorbents requires some adaptation of conventional SPE procedures, but when
fully optimized they are no more difficult to use than conventional sorbents.

12.3.3.6 Molecularly imprinted polymers

Molecularly imprinted polymers are sometimes referred to as plastic antibodies
and are used in SPE as synthetic analogs of immunosorbents [116-123].
Molecular imprinting is a technique used for preparing polymers with synthetic
recognition sites having a predetermined selectivity for a specified analyte (or

357
 group of similar analytes). The imprint is obtained by the polymerization of
functional and cross-linking monomers in the presence of a template molecule
(the analyte). The template-monomer system is chosen such that in solution the
imprint molecule complexes one or several of the functional monomers, which
then become spatially fixed in a solid polymer by the polymerization reaction.
The resultant imprints possess a steric (size and shape) and chemical (spatial
arrangement of complementary functional groups) memory for the template
molecule. Removal of the template from the polymer matrix creates vacant
recognition sites that enable the polymer to selectively rebind the imprint
molecule from a mixture of closely related compounds. In some cases the binding
affinity and selectivity for template-polymer binding approach those
demonstrated by antibody-antigen complexes.
For the time being, it is difficult to predetermine the experimental condi-
tions for successful imprinting of target analytes. In the case of non-covalent
binding, the most common synthetic method (other approaches include revers-
ible covalent interactions and metal ion mediated interactions), the key to a suc-
cessful imprint polymer is the formation of a stable complex between monomer
and template molecules in the pre-polymerization mixture. Maximal efficiency
of imprint formation occurs when the solvent (porogen) has sufficient solvating
power to adequately solubilize the reaction mixture without interfering with the
formation of template-monomer interactions. In current practice apolar organic
solvents for synthesis of the imprint polymer are generally used, which are usu-
ally different to solvents used for sample application. Shrinking or swelling of
the polymers in different solvents during sample processing and recovery can
have a profound effect on the affinity of the polymer for the template molecule
I117,122-124]. To date, block polymerization using UV radiation or thermal ini-
tiation to create a macroporous polymer, followed by grinding and particle siz-
ing, has been the technique most often used for the preparation of polymer
particles for cartridge SPE. The synthesis of the imprint occurs in the presence
of a large amount of the template molecule, which can be difficult to quantita-
tively extract from the polymer reducing its binding capacity, but more seri-
ously, residual template molecules may leak into sample extracts disqualifying
analyte quantification, particularly at trace concentrations. One solution to this
problem is to use a template molecule similar to the analyte for the imprint pro-
cess together with an analytical method that is able to separate the template and
analyte for the determination step. Only a few practical applications of molecu-
larly imprinted polymers for SPE have been demonstrated so far [117,119], most
of which are for the isolation of drugs from biological fluids, but there is consid-
erable promise for this technology. Methods for controlling non-specific adsorp-
tion and polymer morphology and stability are required, but seam well within
the bounds of polymer technology used to prepare sorbents for HPLC. Molecu-
larly imprinted polymers should be easier and cheaper to produce in chemical
laboratories than antibodies while, at least in theory, they should be capable of
similar specificity.

358
12.4 THEORY OF SOLID-PHASE EXTRACTION

Most of the parameters that describe the sequence of processing steps in SPE are
amenable to measurement by liquid chromatography or estimation using theo-
retical principles derived from the theory of liquid chromatography
[4,6,125-127]. Analyte concentrations are generally low and the volume of sam-
ple that can be processed, and therefore the amount of analyte isolated, is deter-
mined by the breakthrough volume of the sampling device. The volume and type
of rinse solvent for matrix simplification is determined by the type and amount
of sorbent required for the isolation step. Here, the need is to identify the volume
of the strongest solvent that eliminates matrix components from the sorbent
without analyte loss. Optimum conditions can be established by selecting eluting
conditions that preserve a certain minimum value for the retention factor of the
least retained analyte. To accomplish a significant concentration of the analytes
with minimal further sample manipulation it is desirable to recover the analytes
in a small solvent volume. Generally the minimum elution volume that can be
safely employed is about 2-3 times the hold-up volume for the sampling device.
This corresponds to a retention factor less than 2 for the most retained analyte.
In addition, the sorption mechanisms for sample application occur according to
the characteristics of frontal analysis and the rinse and desorption step by elu-
tion. Typical SPE devices contain short sorbent beds and cannot be expected to
provide a high plate number. The possibility that retention may be influenced by
kinetic properties has also to be taken into consideration.

12.4.1 Breakthrough volumes

The breakthrough volume for a particular analyte and sampling device is

determined by its breakthrough curve (Fig. 12.6). In the initial sampling phase
the analytes are quantitatively retained by the sorbent up to the point that the
sample volume exceeds the retention capacity of the sorbent. Further sample
passing through the sorbent bed is not quantitatively retained by the sorbent
and eventually the analyte concentration entering and exiting the sampling
device become identical. The position on the curve at which some arbitrary
amount of sample is detected at the outlet of the sampling device, typically 1%,

Co
z 0.841 -

Fig. 12.6. Typical representation of a breakthrough curve. VB is 0.oO-----

the breakthrough volume, VR the chromatographic elution vol- .-
ume, Vc the sample volume corresponding to the isolation of the E 01599
maximum amount ofanalyte, C0 the concentration of analyte in 1
the sample, and v the standard deviation of the derivative V, VR Vc
curve for the plot. (From Ref. [127]; copyright Elsevier). Sample volume

359
5% or 10% is arbitrary defined as the breakthrough volume (VB). The chosen
level reflects the difficulty of experimentally determining small changes in the
concentration of the analytes at the outlet of the sampling device. For the
purpose of modeling the breakthrough process a value of 1% is generally chosen
in keeping with the desire to define a maximum sample volume that can be
processed with minimal (acceptable) analyte loss. A second point on the break-
through curve, V, corresponds to the sample volume at which the retention
capacity of the sorbent is saturated and the concentration of analyte exiting the
sampling device is the same (actually 100% used to define the breakthrough
volume) as that entering the sampling device. It corresponds to the volume of
sample that will result in the isolation of the maximum amount of analyte but
with a lower overall recovery, because a fraction of the sample is lost during the
sorption process. The point of inflection for the breakthrough curve corresponds
to the chromatographic retention volume, VR, provided that the plate number for
the sampling device is not too small [128].
In general, there are two common causes of premature breakthrough in fron-
tal chromatography. The retention capacity of the sorbent bed is overloaded due
to a high concentration of either analyte or sorbed matrix components. Or, the
sorbent bed may fail to adequately retain the analytes due to the provision of an
insufficient plate number for retention volumes to be independent of the plate
number for the sampling device.

12.4.1.1 Determination of breakthroughvolumes

The most straightforward method for determining breakthrough volumes is the
direct method using either on-line or off-line detection [16,129]. This method is
particularly convenient for precolumn devices used in on-line SPE-LC. A
solution containing a low but detectable concentration of analyte is pumped at a
constant flow rate through the precolumn, which is connected directly to the
detector. The detector response for the analyte at the exit of the precolumn is
similar to Fig. 12.6. Since the breakthrough volume may be flow rate dependent
the flow rate used to record the breakthrough curve should be similar to the flow
rate used for sample application.
For standard cartridge and disc devices off-line sample processing is
commonly used. Samples are processed in aliquots in the same way as regular
samples with an off-line detection method used to determine the analyte concen-
tration in the extracts recovered from each sample [72,130-132]. Each aliquot
contains the same amount of analyte but in a different sample volume. Initially,
an approximate value of the breakthrough volume is established by using decade
changes in aliquot volumes, followed by a more systematic experimental design.
For compounds with an estimated breakthrough volume between 0 and 50 ml,
measurements are made at 2.5 ml volume increments, between 50 and 100 ml at
5 ml increments, 100 and 1000 ml at 10 ml increments, and greater 1000 ml at
100 ml increments. Plotting the observed recovery for the complete sampling
process against the corresponding aliquot volumes generates a breakthrough

360
curve from which the breakthrough volume is estimated from the best-fit line
through the experimental data. A similar approach has been used to determine
the breakthrough volume of precolumn traps in on-line SPE-LC [133-135].

12.4.1.2 Methods for estimating breakthroughvolumes

The determination of breakthrough volumes is time consuming and identifying
the breakthrough point on the breakthrough curve somewhat subjective.
Consequently, methods that enable breakthrough volumes to be estimated from
solute properties or calculated from models that require a minimal number of
experimental measurements are particularly attractive. The octanol-water dis-
tribution constant (log Kow) is a widely used parameter for estimating compound
hydrophobicity and as a general compound descriptor for modeling numerous
environmental and biological properties [136]. Hennion et al. [125] investigated
the use of log Kow to estimate log kw (sorbent-water retention factor) for the
purpose of modeling SPE breakthrough volumes but concluded that it was of
limited value. These results were confirmed by Poole et al. [136] and explained in
a theoretical context.

12.4.1.3 Models for predictingbreakthroughvolumes

From the general theory of frontal chromatography it is possible to derive a
relationship between the breakthrough volume and the sorption properties of
SPE devices [16,76,125-127,137-141]. At the 1% breakthrough level the break-
through volume, VB, is related to the retention volume, VR, through Eq. (12.1)
VB = VR- 2.3Cv (12.1)

where av is the standard deviation depending on the axial dispersion of the

analyte along the sorbent bed (see Fig. 12.6) and is evaluated through Eq. (12.2)

oG= VM(1 + k)/N (12.2)

where VM is the interparticle volume of the sorbent bed (hold-up volume), k the
retention factor, and N the plate number for the sorbent bed calculated by Eq.
(12.3)
2
N = VR(VR - C)/v (12.3)
In principle it should be possible to calculate VB from Eqs. (12.1) to (12.3) by
determining VM and N for the sampling device and measuring VR (or k) for the
analytes of interest. N and VM are easily determined for precolumn cartridges in
coupled SPE-LC since the on-line arrangement allows direct recording of the
breakthrough curves. There is no convenient way to make measurements of
these parameters for off-line sampling conditions, and therefore estimates must
be made from results obtained for the SPE device by HPLC [127,128,142].
Equations (12.1) to (12.3) are derived assuming that the conditions of linear
chromatography apply and the plate number for the sorbent trap is reasonably

361
TABLE 12.7
Coefficients for the Lovkvist and Jonsson model, Eq. (12.4), for sorbent beds with a low plate
number
Breakthrough level (%) Coefficients
a. al a,

0.1 0.998 29.12 57.54

0.5 0.990 17.92 26.74
1.0 0.980 13.59 17.60
5.0 0.903 5.36 4.60
10.0 0.810 2.88 1.94

large. For sorbent beds with low plate numbers the above equations can result in
a poor estimate of breakthrough volumes.
Lovkist and Jonsson [128] proposed a model described by Eq. (12.4) to
characterize the sampling properties of sorbent beds with a small plate number
that has been adopted by other groups [127-131,142-144] to calculate break-
through volumes under SPE conditions. The coefficients ao, a, a2 are character-
istic of the breakthrough level and are summarized in Table 12.7 [128]
VB = (ao + a1 /N + a2 /N2)-1(1 + k)VM (12.4)
The calculation of the breakthrough volume requires the determination of N and
VM for the sampling device and either a measurement or estimation of the reten-
tion factor. The influence of these parameters with respect to the performance of
solid-phase extraction devices can now be discussed. The main requirements are
the optimization of size (VM and N), kinetic properties (N, particle size, flow rate)
and retention (N, k and VM).

12.4.2 Kinetic optimization of cartridge and disc devices

For off-line or on-line sampling the sorbent bed cannot be too large because it is
desirable to recover the analytes in a small solvent volume. In addition, in
off-line sampling the pressure drop available to transport the sample through
the sampling device at a practical velocity is limited. The pressure drop per unit
length of sorbent bed is given by Eq. (12.5) [127]

AP/L = uq/d 2 (12.5)

where AP is the pressure drop across a sorbent bed of length L, u the linear
velocity of the sample solution through the sorbent bed, Tjthe viscosity of the
sample solution, the flow resistance parameter for the sorbent bed (typically
103 for cartridges and particle-loaded membranes [126]) and d the average

362
 10

0 25 50 75 100 125
Particle Size
Fig. 12.7. Plot of (AP/L) against the average particle size (dr) for different sample processing
rates (u). A = 0.11 mm/s, B = 0.22 mm/s, C = 0.43 mm/s and D = 1.08 mm/s. The horizontal line
represents the cut off for a pressure drop of 0.9 atm for a 1 cm bed length. (From Ref. [127];
copyright Elsevier).

particle diameter for the sorbent. For off-line sampling with cartridge devices
the available pressure drop is limited to about 0.9 atm. In practice this means
that sorbent beds are restricted to particle diameters greater than about 40 Am
with larger particles providing a wider range of sample flow rates (Fig. 12.7)
[127]. For coupled-column systems high-pressure pumps are used for sample
application and pressure is no longer the factor that dictates the sorbent particle
diameter employed. These columns are also generally short resulting in
conditions favorable for fast sample processing. For off-line sampling the easiest
solution to increase sample-processing rates is to increase the diameter of the
sampling device at a constant bed height, since the flow rate is proportional to
the diameter squared. This is conveniently achieved using particle-loaded mem-
branes and particle-embedded glass fiber discs.
Sorbent cartridges with short beds packed with coarse particles, or for that
matter, very short beds packed with fine particles, cannot be expected to provide
large plate numbers. A plot of the plate height against the interparticle velocity
for some typical SPE sorbents is shown in Fig. 12.8 [126]. There is no observed
minimum in the plots over the interparticle velocity range of 0.5-5.0 mm/s
(equivalent to a flow rate of about 3-30 ml/min through a 1-cm diameter car-
tridge). The main contribution to the plate height arises from flow anisotropy in
the packed bed and resistance to mass transfer. The sorbents in Fig. 12.8 provide
from 5-40 plates per centimeter of bed length with considerable variation for the
different sorbent types. A cartridge with a 1-cm diameter operated at a sample
flow rate of 1 ml/min cannot be expected to provide more than about 20 plates
per centimeter of bed length. At higher sample flow rates only lower plate num-
bers are expected. For cartridges with a lower packing density than shown in
Fig. 12.8, a realistic scenario based on measurements of average cartridge pack-
ing density [126] even smaller plate numbers are to be expected.

363
 11n _ 1n
2 A

:ZF
;~100 I
aH 9
8S~~~~~W
a
9~~~m
S~~~~~~~~
1.5-
(D
Li
0
U80 7n

LiI
H W
H
I
1- l54f
H f 41
Alt -
4

0 0.5 i 1.5
INTERPARTICLE VELOCITY
0 0 50 .
100
1 2 3 4 5 o
INTERPARTICLE VELOCITY uJbr rLk. 1MH

Left: Fig. 12.8. Plot of the plate height (mm) against mobile phase interparticle velocity (mm/s)
for silica-based cartridge sorbents. The test compound was anthracene and the mobile phase
methanol-water (80:20 v/v). 1 = Octadecylsiloxane-bonded sorbent (light loading); 2 = spacer
bonded propanediol sorbent; 3 = cyanopropylsiloxane-bonded sorbent; 4 = butylsiloxane-
bonded sorbent. (From Ref. [126]; copyright Elsevier).
Fig. 12.9. Variation of the efficiency of a particle-loaded membrane as a function of the mobile
phase velocity. (A) Plot of the observed plate height (am) as a function of the interparticle mobile
phase velocity, and (B) the transformation of(A) to indicate the plate number (N) as a function of
the sample flow-rate through a 0.5 mm disk with a 38 mm-diameter active sampling area (From
Ref. [126]; copyright Elsevier).

The kinetic properties of particle-loaded membranes are different to those of

cartridges (Fig. 12.9) [126,145]. A minimum value for the plate height, corre-
sponding to about seven particle diameters, is observed at an optimum inter-
particle velocity of about 0.19 mm/s. Over the typical flow rate range for a
47-mm-diameter disc with a 38-mm active sampling area, 10-100 ml/min, the
particle-loaded membrane will provide about 4-9 theoretical plates, with the
highest value in the region of the optimum flow rate (about 13 ml/min).
The influence of low plate numbers on breakthrough curves for SPE devices
can be calculated using Eq. (12.4). Breakthrough curves are given in Fig. 12.10
for a cartridge with a hold-up volume of 0.42 ml, a solute retention factor of 100,
providing 5, 20 and 100 plates per cm bed length. Increasing plate numbers
results in a sharper front boundary and a larger breakthrough volume (23 ml for
N = 5, 34 ml for N = 20 and 40 ml for N = 100). Larger plate numbers are desir-
able for optimal sample retention but small values of N are capable of providing
useful breakthrough volumes.
For recovery of analytes by elution it is necessary to consider the shape of the
front as well as the retention capacity of the sorbent [76]. The required elution
volume for 99% analyte recovery, V5, on a cartridge with a low plate number is
given by Eq. (12.6)

V = V[1 + k][l + (2.3/)] (12.6)

The only practical way to minimize the volume of eluting solvent is to use a small
sorbent bed (minimize VM) and a strong solvent (k < 3 and ideally 0). Cartridges

364
 0.7
.u

O0.
C
0.2

0.2

0
Sample Volume

Fig. 12.10. Plot of the breakthrough curves for a sampling device with a hold-up volume of 0.42
ml, retention factor 100 and (A) 5, (B) 20, and (C) 100 plates per cm of bed length. (From Ref.
[127]; copyright Elsevier).

with relatively large values of N provide sharper desorption front profiles and
require a smaller elution volume to quantitatively recover the analyte from the
cartridge.

12.4.3 Retention

It is convenient to rearrange Eq. (12.4) into the general expression

log V = log QVM + log (1 + ks) (12.7)

where V is either the breakthrough volume, volume of rinse solvent or elution

volume, Q the contribution of kinetic factors to retention resulting from the
small plate number, VM the hold-up volume for the sorbent bed, and k s the
retention factor for the analyte with the sample solvent, rinse solvent or elution
solvent as mobile phase. For a limited range of flow rates and a specified
sampling device, the product QV, is approximately constant. Given the typical
numerical values for QVM it is easy for log (1 + ks) to become the dominant term
in Eq. (12.7) and the most important factor determining the breakthrough
volume. To optimize the breakthrough volume requires the selection of sample
processing conditions that provide large retention factors. Selection of a rinse
solvent requires identification of experimental conditions that preserve a suffi-
ciently large retention factor to avoid loss of analyte during the rinse step. The
selection of the elution conditions requires identification of experimental condi-
tions that minimize the retention factor. Methods that allow the measurement
or estimation of retention factors for relevant sampling conditions are then of
prime importance for method development in SPE.

365
12.4.3.1 Experimental determinationof retention factors
There are two general approaches for determining retention factors on sorbents
used for SPE. The equilibrium method uses a fixed volume of sample solution
with a known concentration of analyte that is continuously circulated by a
mechanical pump through the sampling device and returned to the sample
solution reservoir until a steady state is reached [36,140,141]. On-line monitor-
ing of the analyte solution exiting the sampling device is a convenient method to
establish the time required to reach a steady state. The amount of analyte
retained by the sorbent is then determined by elution with a small volume of
strong solvent using any suitable method for quantification. Provided that the
sorption isotherm is linear the retention factor can be calculated from the
volume and concentration of the analyte in the sample solution, the hold-up
volume for the sorbent bed, and the amount of analyte taken up by the sorbent
under steady-state conditions.
The alternative method of determining the retention factor is by direct
measurement of retention in a typical chromatographic experiment using col-
umns of different lengths packed with the SPE sorbent [139,142,146-150] or
forced-flow planar chromatography [28,145,151] for discs. Compounds with
retention factors up to about ten thousand can be determined in this way.
Compounds with a retention factor greater than ten thousand are more than
adequately retained for most likely sampling conditions and an accurate deter-
mination of their retention factors is rarely required.

12.4.3.2 Methods for estimating retention factors

The general approach for estimating retention factors with sample solutions
containing predominantly water is by extrapolation from retention factors
determined at more convenient mobile phase compositions providing shorter
separation times. Extrapolations are usually based on either linear, Eq. (12.8),
or quadratic, Eq. (12.9), models

logk = logkw-Sos (12.8)

logk = logkw + S,¢s + S 2 ¢s2 (12.9)

where kw is the equation intercept (assumed to be equivalent to the retention

factor for the analyte with water as the mobile phase), Os the volume fraction of
organic solvent in a binary mixture of water and organic solvent, and the S
coefficients are regression constants obtained by fitting the experimental data to
the models. The validity of Eqs. (12.8) and (12.9) is debated elsewhere [125-127,
136,152,153]. Here we simply note that either equation may function adequately
for a limited range of aqueous organic solvent mixtures of intermediate composi-
tion but both equations are unreliable when data for the region of low organic
solvent composition of interest for estimating breakthrough volumes in SPE are
included [125-127,136]. Some typical results are summarized in Table 12.8

366
TABLE 12.8
Comparison of experimental and extrapolated values for log kw for an octadecylsiloxane-bonded
silica sorbent
Compound Log kw
Linear Eq. (12.8)' Quadratic Eq. (12.9)2 Experimental
Methanol-Water
2-Phenylethanol 2.00 2.36 2.45
4-Chlorophenol 2.52 2.50 2.73
4-Nitrobenzyl alcohol 1.58 2.15 2.40
Acetanilide 1.60 2.19 2.52
Acetophenone 2.26 2.81 3.01
Benzaldehyde 1.87 2.43 2.56
Hexan-2-one 1.91 2.49 2.62
Acetonitrile-Water
2-Phenylethanol 1.27 2.05 2.45
4-Chlorophenol 1.81 2.35 2.73
4-Nitrobenzyl alcohol 1.21 1.89 2.40
Acetanilide 1.03 1.92 2.52
Acetophenone 1.71 2.33 3.01
Benzaldehyde 1.66 1.94 2.56
Hexan-2-one 1.74 2.31 2.62
1From 40-70% (v/v) methanol or acetonitrile.
2
For the full data range from 1-100 %(v/v) methanol or acetonitrile.

[136]. Because extrapolation methods provide unreliable estimates of log kw with

large errors likely in some cases, extrapolation methods cannot be used to
estimate log kw with any reasonable level of confidence for estimating break-
through volumes by Eq. (12.7).

12.4.3.3 Solvation parametermodel

The appropriate form of the solvation parameter model for SPE with liquid
samples is given by Eq. (12.10) [72,126,127,132,151,153].

logSP= c + mVx + rR2 + ' +aya' +b E (12.10)

The model equation is made up of product terms representing solute properties

(descriptors) and system properties characteristic of the sampling system. Each
product term represents the contribution of a defined intermolecular interaction
to the correlated solute property (SP), in this case either the retention factor (log
k) or breakthrough, rinse and elution volumes (log V) when log QVM in Eq. (12.7)
is approximately constant.
The solute descriptors in Eq. (12.10) are McGowan's characteristic volume
Vx, the excess molar refraction R2, the solute's dipolarity/polarizability 04, and

367
the solute's effective hydrogen-bond acidity and hydrogen-bond basicity, S ah
and A S, respectively. The solute's characteristic volume and excess molar
refraction are additive properties that can be calculated for any analyte whose
structure is known [154,155]. The solute dipolarity/polarizability and hydro-
gen-bond acidity and basicity parameters can be obtained experimentally from
gas-liquid chromatographic data or water-solvent distribution constants [156].
For some solutes such as anilines, pyridines, and sulfoxides, which exhibit vari-
able hydrogen-bond basicity in aqueous and non-aqueous solutions, the CH
descriptor is replaced by I O for the extraction of samples from aqueous solu-
tion. The P descriptor is retained for extraction of analytes from organic sol-
vents. Solute descriptors are available for about 4000 compounds with others
available through parameter estimates for compounds of known structure [157].
The system constants in Eq. (12.10) are defined by their complementary
interactions with the solute descriptors. The r constant determines the differ-
ence in capacity of the solvated sorbent and sample solution to interact with
solute n- or -electrons; the s constant to the difference in capacity of the
solvated sorbent and sample solution to take part in dipole-dipole and dipole-
induced dipole interactions; the a constant is a measure of the difference in
hydrogen-bond basicity of the solvated sorbent and the sample solution; the b
constant is a measure of the difference in hydrogen-bond acidity of the solvated
sorbent and sample solution; and the m constant is a measure of the relative ease
of cavity formation for the solute in the solvated sorbent and sample solution
together with contributions from dispersion interactions that fail to cancel when
the solute is transferred between phases. For any sampling system, the system
constants can be obtained by multiple linear regression analysis of experimental
retention properties acquired for a group of varied solutes with known
descriptors. The salvation parameter model is strictly applicable to neutral
compounds and ionizable compounds in their neutral form. Ionization tends to
reduce retention compared to the neutral form of the solute in SPE from
aqueous solution.
A plot of the system constants as a function of solvent composition provides a
system map such as the one shown in Fig. 12.11 [127]. System maps are the most
6

4 m

o r C
= '

b
Fig. 12.11. System map for a heavily loaded octadecyl- -4
siloxane-bonded silica sorbent with methanol-water as the 0 25 50 75 100
sample solvent. (From Ref. [127]; copyright Elsevier). Volume %of Methanol

368
TABLE 12.9
System maps for selecting SPE conditions for aqueous samples using the solvation parameter
model
Sorbent* Type Solvent Ref.
IST C-18 (HL) Octadecylsiloxane-bonded silica (high Methanol 142
carbon loading)
JTB C-18 (LL) Octadecylsiloxane-bonded silica (light Methanol 132,136
carbon loading) Acetonitrile
Tetrahydrofuran
Empore C-18 Octadecylsiloxane-bonded silica Methanol 151
particle-loaded membrane
SPEC-C18AR Octadecylsiloxane-bonded silica Methanol 28
particle-embedded membrane
JTB C-4 Butylsiloxane-bonded silica Methanol 150
2-Propanol
Acetonitrile
Tetrahydrofuran
JTB CN Cyanopropylsiloxane-bonded silica Methanol 147,149
2-Propanol
Acetonitrile
Tetrahydrofuran
JTB DIOL Silica-based spacer bonded propanediol Methanol 148
2-Propanol
Acetonitrile
Tetrahydrofuran
PLRP-S Poly(styrene-divinylbenzene) Methanol 146
2-Propanol
Acetonitrile
PGC Porous graphitic carbon Methanol 77
*IST = International Sorbent Technology; JTB = J.T. Baker; PLRP-S = Polymer Laboratories;
PGC = Hypercarb; Empore = 3M; and SPEC = Ansys.

useful approach for method development. The individual system constants

change smoothly with solvent composition and can be fit to simple linear or
polynomial functions for computer-aided calculation of sampling properties.
Once generated the system maps are permanent and can be used to estimate
sample-processing conditions based on Eqs. (12.7) and (12.10) for any analyte
whose solute descriptors are known or can be reasonably estimated.
In general terms there are three regions of the system map of interest for
method development in SPE. The left-hand side of the map, corresponding to
low organic solvent, is the region of interest for establishing a safe sampling vol-
ume. From system maps for different sorbents the preferred sorbent for the iso-
lation step can be identified. The intermediate region of the system map is of
interest for selection of the rinse solvent. The right-hand side of the system map
is the region of interest for selecting solvent compositions for the elution of
analytes in a small solvent volume. A number of system maps are available for
different sorbents and solvent compositions as indicated in Table 12.9.

369
TABLE 12.10
System constants for the estimation of solute retention in solid-phase extraction for aqueous
samples (water or 1% v/v methanol in water)
Sorbent* System constants
m r s a b c
IST C-18 (HL) 4.39 0 0 -0.79 -1.90 -0.27
JTB C-18 (LL) 3.92 0 -0.11 -0.54 -1.53 -0.90
IST C-8 4.91 0 -0.84 -1.43 -2.27 -0.54
IST CHX 3.94 0 -0.32 -0.83 -1.98 -0.31
IST CH 3.22 0.35 0 -0.92 -1.61 -0.37
JTB C-4 3.36 0 0 -0.46 -1.53 -1.38
JTB CN 2.06 0.53 0 -0.51 -1.45 -0.88
JTB DIOL 1.57 0.61 0 -0.45 -0.80 -1.05
PLRP-S 5.22 0.84 -0.49 -1.39 -4.01 -0.18
Oasis HLB 3.48 1.16 0 0 -2.94 -0.32
PGC 5.14 0.82 0.36 0 -2.76 -2.32
*IST = International Sorbent Technology; JTB = J. T. Baker; PLRP-S = Polymer Laboratories
(styrene-divinylbenzene porous polymer); Oasis HLB = poly(divinylbenzene-co-N-vinylpyrro-
lidone); PGC = Hypersil porous graphitic carbon (Hypercarb); HL = high loading; LL = light
loading; CHX = cyclohexanesiloxane-bonded; CH = phenylsiloxane-bonded; CN = cyanopropyl-
siloxane-bonded; and DIOL = spacer bonded propanediol.

Sorbent selection is based on identifying an appropriate sorbent that pro-

vides an adequate breakthrough volume for the analytes of interest. The break-
through volume can be calculated from Eq. (12.7) when the parameters for log
QVM are known or confidently estimated, as is generally the case, and only the
retention factor need be calculated using Eq. (12.10) [132]. The system con-
stants for a wide variety of sorbents used for SPE with water or water containing
1% (v/v) methanol as the sample solvent are summarized in Table 12.10. Given
that only system constants with a positive sign contribute to sorbent retention
we can obtain some general insight into the factors responsible for sorbent
extraction from aqueous solutions. For all sorbents the cavity/dispersion term
(m system constant) is most important for promoting retention. All sorbents are
at least as competitive as water for electron lone pair interactions (r = 0) and in
some cases these interactions contribute favorably to retention (r is positive), if
only in a minor way compared to the cavity/dispersion term. Oasis HLB is the
exception, having a large positive r system constant, which together with the
favorable cavity/dispersion term is the source of its unique selectivity and high
retention capacity. In general, polar interactions favor transport in the sample
solution and result in smaller breakthrough volumes. The exception is the car-
bon-based materials, which are the most competitive with water for dipole-type
interactions. This is the distinguishing feature of carbon-based sorbents com-
pared to porous polymer and chemically bonded silica sorbents and combined
with favorable cavity/dispersion and electron lone pair interaction terms is the

370
TABLE 12.11
System constants for retention in liquid-solid chromatography with organic mobile phases
Stationary Mobile phase System constants
phase*
m r s a b
DIOL Hexane -1.05 0 1.63 2.10 3.86
AMINO -0.85 0 1.40 1.65 3.81
CYANO -0.37 0 1.88 2.47 0.99
Silica Hexane-methanol -0.83 0 1.06 2.23 1.56
DIOL (99:1) -0.85 0 1.07 2.37 1.47
AMINO -0.72 0 0.94 2.94 1.20
CYANO -0.61 0 0.95 1.86 1.15
Silica Hexane-methyl 0 -0.86 1.67 1.84 3.00
CYANO t-butyl ether (95:5) -1.20 0 1.43 1.10 2.80
Silica Hexane-methyl -0.46 -0.21 1.10 1.16 3.02
CYANO t-butyl ether (80:20) -1.08 0 1.01 0.53 2.26
*AMINO = 3-aminopropylsiloxane-bonded silica; CYANO = 3-cyanopropylsiloxane-bonded
silica; and DIOL = spacer bonded propanediol siloxane-bonded silica.

root of their high retention of polar compounds and complementary selectivity to

the other sorbents. For the chemically bonded sorbents dipole-type interactions
are of limited importance (s is either zero or small and negative). The cyano-
propylsiloxane-bonded and spacer bonded propanediol sorbents have small val-
ues for the cavity/dispersion term compared to other chemically bonded sorbents
indicating greater cohesion and a lower retention capacity. The greater cohesion
of these sorbents easily eclipses their capacity for selective polar interactions
with the result that they are less effective for the isolation of polar analytes from
water than the alkylsiloxane-bonded sorbents. None of the sorbents are competi-
tive with water for hydrogen-bond interactions except for carbon-based sorb-
ents. The general difficulty in isolating hydrogen-bond bases arises because the b
constant for chemically bonded and porous polymer sorbents is quite large and
negative. The system properties most important for controlling retention are the
high cohesive energy and hydrogen-bond acidity of water for which the solvated
sorbents compete to varying extents but fail to dominate. Ironically, these prop-
erties are in opposition with respect to sorbent retention. The high cohesive
energy of water promotes retention while the hydrogen-bond acidity of water is
the principle reason for low retention of hydrogen-bond bases. The equation con-
stant (c term) is not related to fundamental properties of the analytes but is
important for estimating retention. It is a complex combination of factors, one of
which is the phase ratio for the sampling system when the retention factor or
breakthrough volume is used as the dependent variable in Eq. (12.10).
There are only a limited number of models that characterize extraction from
organic solvents, all of which have been obtained by liquid-solid chromatogra-
phy (Table 12.11) [148,158-161]. The driving force for retention under normal

371
phase conditions is the capacity of the solvated sorbent for polar interactions.
Increasing solute size generally reduces retention, in contrast to reversed-phase
extraction where the m system constant is invariably positive and the most
important term controlling solute extraction. Electron lone pair interactions are
unimportant for many applications. The main contribution to retention by the
3-aminopropylsiloxane-bonded sorbent is the hydrogen-bond basicity of the
solvated sorbent (a constant). For the 3-cyanopropylsiloxane-bonded sorbent it
is the capacity of the solvated sorbent for dipole-type interactions (s constant).
For the spacer bonded propanediol siloxane-bonded sorbent it is the capacity for
hydrogen-bond interactions, particularly as a hydrogen-bond acid (b constant),
combined with a significant capacity for dipole-type interactions (s constant)
that are important. Silica is a strong hydrogen-bond acid and significantly
hydrogen-bond base and dipolar. The range of sorbent selectivity variation is
much greater than observed for extraction of aqueous solutions and is very
strongly dependent on the composition of the sample solvent. This is seen in the
large changes in the system constants observed for hexane and hexane contain-
ing 1% (v/v) methanol. For this reason it is unwise to attempt to reduce retention
properties to a sorbent property alone. The limited data available indicates that
it will be difficult to obtain reasonable breakthrough volumes for analytes that
lack polar functional groups and that breakthrough volumes will change signifi-
cantly with small changes in the composition of the sample solvent. Although the
solvation parameter model can model retention on chemically bonded sorbents
using organic solvents without apparent difficulty it probably does not correctly
account for solute size effects and site-specific interactions for inorganic oxide
sorbents [158].

12.5 METHOD DEVELOPMENT

12.5.1 Method and mode selection

Sorbent selection is based on considerations summarized in Fig. 12.12. The

sample solvent (aqueous or organic), the analyte type (non-polar, polar or
ionized), and if ionized (strong or weak acid or base) provides a logical guide for
method selection. Organic compounds soluble in polar organic solvents but
difficult to dissolve in solvents of intermediate polarity, if they can be reconsti-
tuted in aqueous solution can be extracted in the reversed-phase mode.
The same sorbents used for cartridge SPE are generally used for disc extrac-
tion, but since discs are used for a narrower range of applications at present, the
number of frequently used sorbent types is fewer, Table 12.12. The majority of
disc applications proposed so far are for aqueous samples.
The sample volume is generally selected to conform to the needs of the
instrumental detection step, and as these instruments have improved in sensi-
tivity sample volumes have decreased in size. Regulatory authorities often indi-
cate action levels in concentration units, which can also be used to define an

372
Fig. 12.12. Method selection guide for the isolation of organic compounds from solution. SAX =
strong anion exchanger; SCX = strong cation exchanger; WCX = weak cation exchanger; RP =
reversed-phase sampling conditions; NP = normal-phase sampling conditions; IE = ion-
exchange sampling conditions. (From Ref. [127]; copyright Elsevier).

adequate sample volume for analysis. The sample volume that can be processed
by a particular SPE device depends primarily on the breakthrough volume of the
analyte, the concentration of the analyte matrix, sample flow rate, and the
sorbent mass. SPE cartridges are available in a range of sizes containing from
about 35 mg to 10 g of sorbent with the 100 mg and 500 mg sorbent cartridges
being the most widely used for extraction and the larger cartridge sizes for sam-
ple clean-up. As a rough guide the sample capacity of a SPE cartridge is about
1-5% of the sorbent mass. The parameters of interest for selecting a disc for a
particular application are summarized in Table 12.13. The large diameter discs
are used for processing large sample volumes in an acceptable time. Small discs
are used for processing small sample volumes and to recover analytes in a small
solvent volume to eliminate the need for solvent evaporation prior to analysis.
Small discs are frequently used in clinical, forensic and pharmaceutical analysis.

12.5.2 Sample processing considerations

Sample processing involves four distinct steps. Initially, the sorbent is condi-
tioned with solvent to remove impurities and to facilitate sorption of the
analytes. The conditioning step is critically important for processing aqueous
samples using particle-loaded membranes. The high surface tension of water
combined with the microporosity of the discs results in slow and uneven flow
through the discs and low analyte recovery if the discs are not first conditioned
with an organic solvent. The conditioning solvent is then replaced with the same
solvent as the sample solvent and the sample passed through the sampling
device at a controlled flow rate. For large sample volumes a small amount of
organic solvent (1% v/v) is added to aqueous samples to maintain a constant sam-
pling velocity. This usually has little influence on the breakthrough volumes

373
TABLE 12.12
Sorbent selection for disc applications
Sorbent General applications
Octadecylsiloxane- and Largely used for extracting aqueous solutions. Applications
octylsiloxane-bonded dominated by octadecylsiloxane-bonded sorbents. Many
silica applications to the extraction of non-polar and moderately polar
pesticides (all classes), herbicides, phthalate and adipate esters,
polycyclic aromatic compounds, food additives, pharmaceutical
compounds, hydrocarbons and grease, etc. A large number of
approved regulatory methods for water analysis.
Poly(styrene-divinyl- Used for compounds poorly extracted by octadecylsiloxane-bonded
benzene) silica sorbents because of high water solubility (polar pesticides,
herbicides, phenols and pharmaceutical compounds). Higher
recoveries of neutral polar compounds claimed for polar group
functionalized sorbents because of better surface contact with
aqueous samples.
Activated carbon Applications similar to poly(styrene-divinylbenzene) but less
frequently used. Methods proposed for triazine herbicides,
some polar pesticides and N-nitrosodialkylamines.
Mixed mode Usually co-bonded alkylsiloxane and benzenesulfonic acid groups
(or sorbent mixtures) used predominantly in toxicological and
pharmaceutical analysis for the simultaneous isolation of drugs
and their metabolites. A number of established methods for drugs
of abuse (marijuana and cocaine metabolites, amphetamines,
phencyclidine and opiates) in biological fluids.
Sulphonic acid and Isolation of acidic and basic compounds as well as some metals.
quaternary amine ion Many methods for acidic herbicides and pesticides in water and
exchangers basic drugs in biological fluids. Hydrogen-ion and metal-loaded
cation exchangers available for clean-up of samples analyzed by ion
chromatography and capillary electrophoresis.
Crown ethers Commonly used for the isolation of precious metals (Pd, Pt, Rh)
and radionuclides (Cs, Sr, Pb) from high concentrations of other
metals to determine environmental burden and for dating
geochemistry samples. Other applications include the removal of
base metal impurities (Bi, Sb, Fe, Pb, Bi, Cu, Hg) from refinery
streams, plating baths, etc.

TABLE 12.13
Guide to the physical properties of particle-loaded extraction discs
Property Disc diameter (mm)
4 7 10 25 47 90
2
Surface area (cm ) 0.13 0.38 0.80 4.9 17 64
Bed mass (mg) (silica-based sorbents) 4 10 25 140 500 1850
Flow rate (ml/min) 0.5 1.5 3 20 60 250
Elution volume (ml) 0.15 0.25 0.5 3 10 35
Typical sample volume (ml) <1 <5 <25 <250 <1000 <5000

374
except for some porous polymer sorbents, which selectively absorb the organic
solvent changing sorbent selectivity and the system phase ratio [72]. Porous
polymer sorbents, such as poly(divinylbenzene-co-N-vinylpyrrolidone), that are
solvated by water alone, allow samples to be processed without a conditioning
step. A sample processing solvent is not usually required for small sample vol-
umes.
Optionally, after the sample is processed, the sorbent can be rinsed with a
weak solvent to displace undesired matrix components from the sorbent without
displacing the analytes. Finally the analytes of interest are eluted from the
sorbent in a small volume of strong solvent for subsequent determination. The
drying step between processing aqueous samples and eluting the retained
analytes with a water-miscible organic solvent is also important. Water retained
by the sorbent (and the support structure when discs are used) contaminates the
elution solvent and may cause difficulty if the solvent is to be reduced in volume
or the analytes analyzed by gas chromatography. Water may also reduce the
efficiency of the elution step, particularly for solvents only partially miscible
with water. Drying, by suction or vacuum desiccation, is usually sufficient to
remove water trapped in the pores, but excessive drying can result in low analyte
recovery from vaporization or inefficient elution. General advice for establishing
optimum sample processing conditions for cartridge and disc sampling devices is
summarized in Table 12.14. A generic outline for processing samples using disc
technology is outlined in Table 12.15. This can be re-scaled for different disc and
sample sizes using the data in Table 12.13.

12.5.3 Computer-aided methods

In spite of its relative maturity computer-aided strategies for SPE method

development are still uncommon. A significant contributing factor to this
situation has been the poor understanding of the theoretical principles of SPE
until recently and the low status accorded to sample preparation compared with
separations by many professional scientists best able to develop such theories.
Expert systems have been described for the optimization of sample processing
conditions for selected drugs in biological fluids [162,163], but there is little
evidence from the literature that expert systems are widely used at present. It
was shown recently that all sample-processing steps in SPE are amenable to
computer simulation [132]. Seibert and Poole used Eq. (12.7) and the solvation
parameter model, Eq. (12.10), to predict the sample processing conditions for the
isolation of estrogens from urine. System maps were used to determine break-
through volumes for sample processing, the composition and volume of rinse
solvents for matrix simplification, and the composition and volume of elution
solvents. A sample volume of 45 ml was required to provide sufficient estrogens
for convenient analysis by gas chromatography with flame-ionization detection.
Estriol, the least retained of the estrogens, had a predicted breakthrough volume
greater than 45 ml on an octadecylsiloxane-bonded silica sorbent cartridge with

375
TABLE 12.14
Sample processing parameters and their influence on recovery by solid-phase extraction
Conditioning solvent (typically 3-5 bed volumes)
· Ensures reproducible retention and flow. Critical step for particle-loaded membranes
· Helps to minimize contamination of extracts by sorbent impurities
· Replace by sample solvent before processing sample
Flow rates (typical range 0.2-1.5 mm/s)
* More critical for cartridges than discs due to their variable and heterogeneous packing
density (channeling)
* More critical when the sample volume exceeds the breakthrough volume as typical
sampling devices provide too few theoretical plates for flow independent retention
Sample properties
· Dilute viscous samples with a weak low viscosity solvent to reduce sample processing time
* Remove excessive particle matter by filtration or centrifugation to maintain a constant
sample-processing rate.
* Add small volume of organic solvent (1-3% v/v) to large volume water samples to ensure
sorbent remains solvated and to maintain a constant (fast) sample-processing rate.
Important for particle-loaded membranes
* Adjust pH to reduce ionization of weak acids and bases for reversed-phase sampling
* Maintain approximately constant ionic strength for samples and standards when using
reversed-phase sampling conditions. Ionic strength is a critical parameter for ion-exchange
extraction
* Deproteination of biofluids may be required for acceptable recovery of low molecular mass
analytes for reversed-phase sampling
* Precipitation of inorganic acids (sulphate, phosphate, etc.) by barium hydroxide is
sometimes required for acceptable recovery of organic acids from biofluids using
ion-exchange extraction
Drying time (typically 1-5 min, but sometimes considerably longer)
· Sufficient to remove all sample solvent trapped in the sorbent pores
* Excessive drying may result in low recovery of analytes from evaporation or retention in
poorly solvated regions of the sorbent
Rinse solvent (optional)
* Small volume of intermediate strength solvent to elute matrix components. Analytes
remain immobilized on the sorbent
* Biological fluids, plant extracts and soil extracts often require a rinse step but surface
waters may not
Eluting solvent (ideally 2-3 bed volumes but often larger)
· Should be a strong solvent able to displace all analyte from the sorbent in a small volume
· Normally should be volatile and miscible with the sample solvent

a sample solvent containing less than 25% (v/v) methanol. A methanol-water

mixture was selected to minimize matrix sorption during sample application. A
volume of 6 ml of 40% (v/v) aqueous methanol was identified as an appropriate
rinse solvent for matrix simplification. The solvent composition required to
quantitatively elute the estrogens, predicted for estrone as the most retained
estrogen, were three cartridge hold-up volumes of methanol. The good agree-
ment between model predictions and experiment (recovery 95-101% with an
RSD = 3.5%, n = 5) was taken as a validation of the approach. From the system

376
TABLE 12.15
Generic guide for sample processing using solid-phase extraction discs. In this example a 47-mm
disc and a 1-1 water sample are used for illustration
Decontamination With disc installed in filtration apparatus (or cartridge) rinse with 10 ml
of solvent (acetonitrile) by allowing the solvent to soak into the
membrane for a few minutes and then remove by suction.
Condition As described above using 10 ml of methanol. Before the last drop of
methanol has been sucked through the disc add 10 ml of deionized water.
The disc should not be allowed to suck dry until after the sample has been
processed.
Sample The sample containing 1% (v/v) methanol is passed through the disc with
a suitable reservoir in place. Filter aid or a prefilter may be required for
samples with a heavy burden of particulate matter. The sample is
processed at a flow rate between about 50 and 100 ml/min using a vacuum
of about 10-20 mm Hg.
Drying Bulk water is removed from the disc and sampling apparatus by sucking
air through the disc under full vacuum for about 3 minutes. Volatile
compounds may be lost.
Recovery The analytes are recovered by passing two 5 ml volumes of acetonitrile
through the disc. The first volume of acetonitrile is allowed to soak into
the disc for a few minutes and then gently sucked through the disc.
Without letting the disc run dry the second volume of solvent is added to
the disc and sucked through the disc. The disc is then allowed to suck dry.

maps given in Table 12.8 thousands of methods could be simulated without

resort to initial trial-and error experiments for compounds with known or
estimated solute descriptors.

12.6 AUTOMATION AND COUPLED-COLUMN SYSTEMS

SPE lends itself to automation to an extent that has been difficult to achieve
using LLE procedures [1,4,166]. A major impetus for the development of auto-
mated laboratory procedures has come from increasing demand for high
throughput methods in clinical, pharmaceutical and environmental laborato-
ries. Automation is viewed as an essential feature of achieving this goal by reduc-
ing time and cost of labor while improving method accuracy and precision,
reducing assay tedium and enhancing safety by limiting exposure to chemicals
and pathogens. In the most favorable case automated procedures provide for
out-of-hours operation and require a minimum of operator intervention. This is
not always achieved, however.
Automated methods can be considered in three categories. The most com-
mon are semi-automated instruments where operator intervention is required
at some stages during the extraction. These instruments are designed to reduce
the tedium of repetitive manual operations. They may increase sample through-
put and improve accuracy and precision, but are really method assistants.

377
Workstations, on the other hand, are dedicated instruments that usually carry
out all stages of the SPE method without intervention, and are usually easily
coupled to separation instruments. Most early versions, however, operated in a
serial sampling mode or semi-batch mode and offered limited improvement in
sample throughput. The best of these systems can extract between 25-50
samples per hour, but the majority are nearly an order of magnitude slower than
this. For coupled-column systems this was not a problem when the separation
was slow, but with increasing use of fast separation methods and mass spectro-
metric detection, which is less demanding of extraction selectivity, this was no
longer adequate. Really high sample throughput was achieved by parallel sample
processing using workstations designed around the 96-well plate format. The
best of these systems can extract up to about 400 samples per hour, but are
usually somewhat limited in the sample volume that can be handled, and are
mainly used in clinical and pharmaceutical laboratories, where large sample
volumes are rarely required for quantification of target analytes. A higher level
of automation can be achieved using robotic systems that are capable of many
activities besides SPE (e.g. weighing, centrifugation, filtration, capping, evapo-
ration, bar code reading, etc). These can often process the sample from an earlier
stage in the analysis through to injection of extracts into the separation system.
Automation is probably not a high priority for all laboratories. In the absence
of large sample numbers requiring a short turn around time fully automated
systems are difficult to justify. The best of the modern systems use convenient
graphical interfaces that are easy to use but earlier and some existing systems
require specialized knowledge of computer programming and electronics for
operation and maintenance. Carryover from the sharing of common modules by
the samples can limit performance and reduce accuracy through contamination
and limit throughput by the need for repetitive programmed rinse procedures.
Systematic errors arise when the instrument fails to take account of sample
properties, such as incomplete withdrawal of sample volumes due to changes in
viscosity, precipitation or clotting. Sample stability can be a problem for some
samples and conditions, for example, when samples are delayed for various times
prior to being processed. Differences in liquid management, for example, the use
of positive pressure versus vacuum, often means that manually developed
procedures do not function well on automated systems without extensive re-
optimization. On the other hand, automated workstations provide a more
controlled environment for method development with the possibility of running
larger sets of trial experiments in a shorter time to identify rugged extraction
conditions.

12.6.1 Coupled-column techniques

On-line coupling of SPE-LC is quite straightforward and is in common use

[4,6,21,86,94,98,165-167]; SPE-GC is somewhat more complicated but has been
achieved in a robust fashion in the last few years [20-22,168]; and SPE-CE is

378
feasible but largely restricted to use in research laboratories for the present
[169-171]. The majority of applications are for aqueous samples reflecting the
main use of the technique in clinical, pharmaceutical and environmental labora-
tories for the analysis of biological matrices and surface waters. Typical sample
volumes for SPE-LC are 1-10 ml while larger volumes, 10-1000 ml, are required
for environmental applications. For SPE-GC sample sizes are often an order of
magnitude smaller because of the greater sensitivity provided by typical gas
chromatographic detectors. The design of precolumn sampling cartridges is
similar for SPE-LC and SPE-GC as are the sorbents used. Low-specificity
sorbents are used for general applications, mainly octadecylsiloxane-bonded
silica gel and porous polymers [167,169,172], and class-specific sorbents, partic-
ularly restricted access materials and immunosorbents, for selective extraction
of target analytes [96-101,173-177]. Restricted access materials have gained
acceptance for the direct analysis of drugs in biological fluids and the analysis of
polar pesticides in the presence of humic acid in surface waters. Breakthrough
volume considerations are generally unimportant for sampling biological fluids,
since only small sample volumes are required for adequate detection of target
analytes but for environmental applications this is a critical parameter and
dominates sorbent selection. High surface area porous polymers and to a lesser
extent graphitized carbon sorbents, are generally used for environmental appli-
cations [6,43,178-181], but no single sorbent is suitable for all applications.
The choice of precolumn dimensions and sorbent properties is a compromise
between the need for sufficient retention to accommodate the desired sample
volume and efficient analyte desorption for transfer to the separation column
without unacceptable loss in its separation power. Typical precolumns are 10-30
mm long with internal diameters of 1-3 mm when used with standard 4-4.6 mm
internal diameter separation columns in liquid chromatography or 2-20 mm
long and 1-4.6 mm internal diameter for coupling to gas chromatography. These
columns contain about 10-50 mg of sorbent with an average particle diameter in
the range of 10-40 gm. Less frequently precolumns are prepared from several
small particle-loaded membrane discs assembled to form a continuous bed. For
automated operation the precolumn must be either reusable or automatically
exchangeable.
The on-line coupling of precolumn and separation column in liquid chroma-
tography is achieved using a single 6-port automated switching valve with flow
paths indicated in Fig. 12.13, or by using two automated valves for additional
versatility in automating a wider range of sample processing steps. The control
system has to provide for regeneration and conditioning of the precolumn,
loading of the sample, rinsing the precolumn to minimize matrix constituents
and transfer of the analytes to the analytical column for separation. Usually the
sample processing steps occur concurrently with the separation of the previous
sample. All events are programmable and optimized by time and flow control.
For strongly retained analytes, backflushing the precolumn is often the best
approach for solvent desorption. The solvent used for desorption is usually the

379
 (I) (I)

Fig. 12.13. Arrangement for automated coupled-column switching in liquid chromatography.

Position (I) is for sample application and fractionation and position (II) for transfer of extracted
analytes and subsequent separation. P = pump, AS = autosampler, ASV = automated switching
valve, PC = precolumn, AC = analytical column, D = detector and W = waste. (From Ref. [94];
copyright Elsevier).

same or a weaker solvent than the mobile phase used for the start of the
separation. The analytes are also distributed over a significant length of the
precolumn and their transfer usually occurs in a solvent volume that is too large
and perhaps too strong to avoid additional band broadening compared with
conventional injection techniques. The precolumn dimensions (short column of
smaller internal diameter than the separation column), sorbent particle size
(small particle diameter) and the retention capacity of the sorbent for the
analytes (weak retention capacity) are optimized to assist in minimizing band
broadening due to the transfer process. On the other hand for successful
sampling the sorbent should have strong retention properties, the precolumn
should be as large as practical and packed with coarse particles to assist in
sample application when large volumes are used. For environmental samples,
sample volumes usually exceed 100 ml and sample application rates of 5-15
ml/min are common. The flow resistance of the precolumn is an important
consideration for these sampling conditions. The criteria for extraction and
desorption are generally opposite and practical operating conditions are always a
compromise.
In most cases preconcentration of analytes occurs with simultaneous matrix
accumulation. Matrix simplification can be achieved through the sampling pro-
cess, the use of a rinse solvent prior to analyte transfer, and selective transfer of
front-, heart- or end-cuts of the desorbed sample extract. Dual precolumn sys-
tems allow solvent exchange and separate trapping of analytes with a wide range
of breakthrough volumes. The first column of a dual precolumn system can be
used to retain strongly sorbed matrix components that would contaminate the
second precolumn or interfere in the separation of the transferred extract.

380
Fig. 12.14. Typical valve switching arrangement for automated SPE-GC. (From Ref. [168];
copyright Elsevier).

Automated SPE-GC coupling is achieved in a similar manner to SPE-LC.

The precolumn sample processing and clean-up steps are the same except that
sample volumes are often smaller and solvent selection for desorption is limited
to solvents compatible with GC injection techniques, which are generally differ-
ent to the selection criteria for LC [20-23]. A suitable apparatus for automated
SPE-GC is shown in Fig. 12.14 [168,182-184]. The desorption solvent is trans-
ferred to the gas chromatograph via a narrow bore deactivated fused silica
capillary to a standard GC injector connected to a retention gap and possibly
retaining precolumn and an early solvent vapor exit.
For on-column injection transfer occurs under conditions of partial concur-
rent solvent evaporation with the solvent vapor exit initially open during trans-
fer and closed just before completion of the evaporation process. An efficient
precolumn drying process is important, since even a few percent of water
absorbed by the desorption solvent will cause destruction of the deactivated
retention gap. Apart from purging with nitrogen at room temperature, which is
a slow process, a drying cartridge containing anhydrous sodium sulfate or silica
gel can be inserted between the SPE cartridge and the inlet of the gas chroma-
tograph [184]. The loss of partially trapped analytes is more critical in on-line
SPE-GC than for conventional injection techniques because of the varying
analyte distribution in the transfer solvent. The introduction of a small volume
of organic solvent, presolvent, immediately in front of the extract to ensure that

381
a solvent film is already present in the retention gap when extract transfer
starts, can be used to minimize this loss.
Loop-type interfaces with partial concurrent solvent evaporation in a long
retention gap are more robust and easier to automate than the on-column
interface but are restricted to the transfer of analytes of low volatility [20,
185-187]. The interface is equipped with a 6-port and a 14-port switching valve.
The 6-port valve is used to divert the carrier gas either to the gas chromatograph
or to the 14-port valve, on which the SPE cartridge and two loops for storing
organic solvent are mounted. The first loop contains the presolvent and is
usually of smaller volume (e.g. 50 Al). The second loop contains the solvent used
to desorb the analytes from the SPE cartridge and to transfer them to the gas
chromatograph. After the sample processing steps are completed and the
cartridge dried, both valves are switched simultaneously and the solvent vapor
exit opened. The carrier gas pushes the contents of both loops, the second loop
via the SPE cartridge, into the retention gap. Both valves are switched to the
load position when a preselected pressure drop corresponding to near comple-
tion of the evaporation process is recorded and the solvent vapor exit closed after
a selected delay time. The transfer is carried out at a temperature just above the
solvent boiling point.

12.7 CONCLUSIONS

Solid-phase extraction is now a well-established sample preparation technique

with an ever-widening applications base. The methodology continues to evolve
through changes in format more than principle. The driving forces being the
desire to simplify the sampling process or enhanced automation possibilities.
Advances are expected in sorbent chemistry, particularly class-specific sorbents
for the isolation and clean-up of target analytes in complex matrices. Advances
are also expected in the use of computer-aided method development strategies as
the logical replacement for tedious trial-and-error procedures commonly used
today. Further development of automated SPE systems can be expected in those
laboratories where high sample throughput or round-the-clock process monitor-
ing is important. In addition, SPE technology has defined new opportunities for
use in passive sampling, storage and preservation of extracts, field sampling, and
as a biometric indicator. Fundamental research into the development of SPE
techniques and their potential applications is likely to continue at an increased
pace in the next decade as analysts attempt to capitalize on the above
opportunities.

REFERENCES

1 E.M. Thurman and M.S. Mills, Solid-Phase Extraction. Principles and Practice.
Wiley, New York, 1998.
2 J.R. Dean, ExtractionMethods for EnvironmentalAnalysis. Wiley, Chichester, 1998.

382
3 J.S. Fritz, Analytical Solid-Phase Extraction. Wiley, New York, 1999.
4 N.J.K. Simpson (Ed.), Solid-Phase Extraction: Principles, Strategies and
Applications. Marcel Dekker, New York, 2000.
5 I.D. Wilson, E.R. Adlard, M. Cooke and C.F. Poole, Encyclopedia of Separation
Science. Academic Press, London, Vol. 1-10, 2000.
6 M.-C. Hennion, J. Chromatogr.A, 856 (1999) 3.
7 I. Liska, J. Chromatogr. A, 885 (2000) 3.
8 E. Matisova and S. Skrabakova, J. Chromatogr. A, 707 (1995) 145.
9 G.A. Junk, J.J. Richard, M.D. Grieser, D. Witiak, J.L. Witiak, M.D. Angello, R. Wick,
H.J. Svec, J.S. Fritz and G.V. Calder, J. Chromatogr., 99 (1974) 745.
10 M. Dressier, J. Chromatogr., 165 (1979) 167.
11 A. Zlatkis, H.A. Lichtenstein and A. Tishbee, Chromatographia,6 (1973) 67.
12 M. Harper, J. Chromatogr. A, 885 (2000) 129.
13 J.S. Fritz and M. Macka, J. Chromatogr.A, 902 (2000) 137.
14 J.W. Eichelberger, T.A. Bellar, J.P. Donnelly and W.L. Budde, J. Chromatogr. Sci.,
28 (1990) 460.
15 B. Kolb, J. Chromatogr. A, 842 (1999) 163.
16 M.-C. Hennion and V.C. Pinchon, Environ. Sci. Technol., 28 (1994) 576A.
17 L.A. Berrueta, B. Gallo and F. Vicente, Chromatographia,40 (1995) 474.
18 M. Gilar, E.S.P. Bouvier and B.J. Compton, J. Chromatogr. A, 909 (2001) 111.
19 N. Masque, R.M. Marce and F. Borrull, Trends Anal. Chem., 17 (1998) 384.
20 J.J. Vreuls, A.J.H. Louter and U.A.Th. Brinkman, J. Chromatogr.A, 856 (1999) 279.
21 E.R. Brouwer, S. Kofman, U.A.Th. Brinkman, J. Chromatogr.A, 703 (1995) 167.
22 A.J.H. Louter, J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr.A, 842 (1999) 391.
23 E.C. Goosens, D. de Jong, G.J. de Jong and U.A.Th. Brinkman, Chromatographia,47
(1998) 313.
24 D.R. Hagen, C.G. Markell, G. Schmitt and D.D. Blevins, Anal. Chim. Acta, 236 (1990)
157.
25 H. Lingeman and S.J.F. Hoekstra-Oussoren, J. Chromatogr.B, 689 (1997) 221.
26 E.M. Thurman and K. Snavely, Trends Anal. Chem., 19 (2000) 18.
27 J.S. Fritz and J. Masso, J. Chromatogr. A, 909 (2001) 79.
28 M.L. Meyer, C.F. Poole and M.P. Henry, J. Chromatogr.A, 695 (1995) 267.
29 I. Urbe and J. Ruana, J. Chromatogr. A, 778 (1997) 337.
30 V. Pinchon, M. Charpak and M.-C. Hennion, J. Chromatogr.A, 795 (1998) 83.
31 R.S. Plumb, R.D.M. Gray and C.M. Jones, J. Chromatogr., B 694 (1997) 123.
32 J. Hempenius, J. Wieling, J.P.G. Brakenhoff, F.A. Maris and J.H.G. Jonkman, J.
Chromatogr. B, 714 (1998) 361.
33 U.J. Nilsson, J. Chromatogr.A, 885 (2000) 305.
34 M.S. Lee and E.H. Kerns, Mass Spectrom. Rev., 18 (1999) 187.
35 M. Jemal, Biomed. Chromatogr., 14 (2000) 422.
36 M.C. Rouan, C. Buffet, L. Masson, F. Marfil, H. Humbert and G. Maurer, J.
Chromatogr., B 754 (2001) 45.
37 C.E. Green and M.H. Abraham, J. Chromatogr.A, 885 (2000) 41.
38 W.C. Koskinen and B.L. Barber, J. Environ. Qual., 26 (1997) 558.
39 J.D. Mattice, S.K. Park and T.L. Lavy, Bull. Environ. Contam. Toxicol., 60 (1998)
202.
40 E.M.J. Verbruggen, W.M.G.M. Van Loon, M. Tonkes, P. Van Duijn, W. Seinen and
J.L.M. Hermens, Environ. Sci. Technol., 33 (1999) 801.
41 A.P. Freidig, E.A. Garicano, F.J.M. Busser and J.L.M. Hermens, Environ. Toxicol.
Chem., 17 (1998) 998.
42 J.A. Field and K. Monohan, J. Chromatogr. A, 741 (1996) 85.
43 L.E. Sojo and J. Djauhari, J. Chromatogr.A, 840 (1999) 21.

383
44 L.K. Ng, P. Lafontaine and J. Harnois, J. Chromatogr.A, 873 (2000) 29.
45 J.W. King and L.J. King, J. Anal. Toxicol., 20 (1996) 262.
46 J.M. Rosenfeld, J. Chromatogr.A, 843 (1999) 19.
47 N.A. Marley, J.S. Gaffney, K.A. Orlandini, P.J. Drayton and M.M. Cunningham,
Radiochim. Acta, 85 (1999) 71.
48 E.D. Hagestuen, A.F. Arruda and A.D. Campiglia, Talanta, 52 (2000) 727.
49 R.L. Vermillion-Salsbury and D.M. Hercules, Int. J. Envuiron. Anal. Chem., 73 (1999)
297.
50 A.W.T. Bristow, C.S. Creaser, S. Nelieu and J. Einhorn, Analyst, 121 (1996) 1425.
51 D.M. Steinberg, L.J. Sokoll, K.C. Bowles, J.H. Nichols, R. Roberts, S.K. Schultheis
and C.M. O'Donnell, Clin. Chem., 43 (1997) 2099.
52 P.D. Whitter and P.L. Cary, J. Anal. Toxicol., 10 (1986) 68.
53 S.A. Barker, J. Chromatogr.A, 885 (2000) 115.
54 V. Ruiz-Gutierrez and M.C. Perez-Camino, J. Chromatogr.A, 885 (2000) 321.
55 T.M.P. Chichila and D.M. Gilvydis, J. AOAC Int., 76 (1993) 1323.
56 Y. Pico, G. Font, J.C. Molto and J. Manes, J. Chromatogr.A, 885 (2000) 251.
57 G.R. van der Hoff and P. van Zoonen, J. Chromatogr.A, 843 (1999) 301.
58 C.W. Huck and G.K. Bonn, J. Chromatogr.A, 885 (2000) 51.
59 M.E. Leon-Gonzalez and L.V. Perez-Arribas, J. Chromatogr.A, 902 (2000) 3.
60 N.A. Penner, P.N. Nesterenko, M.M. Ilyin, M.P. Tsyurupa and V.A. Davankov,
Chromatographia,50 (1999) 611.
61 M.-C. Hennion, J. Chromatogr. A, 885 (2000) 73.
62 R.M. Marce and F. Borrull, J. Chromatogr.A, 885 (2000) 273.
63 G. Michor, J. Carron, S. Bruce and D.A. Cancilla, J. Chromatogr. A, 732 (1996) 85.
64 I. Rodriguez, M.P. Llompart and R. Cela, J. Chromatogr.A, 885 (2000) 291.
65 M.A. Moeller, S. Steinmeyer and T. Kraener, J. Chromatogr., B 713 (1998) 91.
66 A. Pauwels, D.A. Wells, A. Covaci and P.J.C. Schepens, J. Chromatogr., B 723 (1999)
117.
67 T. Kumazawa and 0. Suzuki, J. Chromatogr., B 747 (2000) 241.
68 R.W. Fedeniuk and P.J. Shand, J. Chromatogr.A, 812 (1998) 3.
69 J.M. Soriano, B. Jimenez, G. Font and J.C. Molto, Crit. Revs. Anal. Chem., 31 (2001)
19.
70 K. Albert, R. Brindle, P. Martin and I.D. Wilson, J. Chromatogr. A, 665 (1994) 253.
71 P. Martin, A. Taberner, A. Fairbrother and I.D. Wilson, J. Pharm. Biomed. Anal. 11
(1993) 671.
72 S.K. Poole and C.F. Poole, Analyst, 120 (1995) 1733.
73 L. Schmidt and J.S. Fritz, J. Chromatogr., 640 (1993) 145.
74 Y.-F. Cheng, D.J. Phillips and U. Neue, Chromatographia,44 (1997) 187.
75 M.-C. Hennion, V. Coquart, S. Guenu and C. Sella, J. Chromatogr.A, 712 (1995) 287.
76 S. Guenu and M.-C. Hennion, J. Chromatogr. A, 725 (1996) 57.
77 C. Lepont, A.D. Gunatilleka and C.F. Poole, Analyst, 126 (2001) 1318.
78 S.K. Poole and C.F. Poole, Anal. Commun., 34 (1997) 247.
79 V. Pinchon, J. Chromatogr. A, 885 (2000) 195.
80 H. Sabik, R. Jeannot and B. Rondeau, J. Chromatogr.A, 885 (2000) 217.
81 V. Pignon, R. Jeannot and E. Sauvard, Int. J. Environ. Chem., 75 (1999) 345.
82 S.R. Rubern, W.M. Draper and S.K. Perera, J. Agri. Food Chem., 48 (2000) 4109.
83 M.C. Carson, J. Chromatogr. A, 885 (2000) 343.
84 M.J.M. Wells and L.Z. Yu, J. Chromatogr. A, 885 (2000) 237.
85 R. Saan-Nordhaus and J.M. Anderson, J. Chromatogr. A, 706 (1995) 563.
86 P.R. Haddad, F, Doble and M. Macka, J. Chromatogr. A, 856 (1999) 145.
87 X.-H. Chen, J.-P. Franke, J. Wijsbeek and R.A. de Zeeuw, J. Chromatogr., 617 (1993)
147.

384
 88 P. Degel, Clin. Biochem., 29 (1996) 529.
89 O.H. Drummer, J. Chromatogr.B, 733 (1999) 27.
90 S. Paterson, R. Cordero, S. McCulloch and P. Houldsworth, Ann. Clin. Biochem., 37
(2000) 690.
91 E.H.R. Vanderwal, E.R. Brouwer, H. Lingeman and U.A.Th. Brinkman,
Chromatographia,39 (1994) 239.
92 I. Ferrer, D. Barcelo and E.M. Thurman, Anal. Chem., 71 (1999) 1009.
93 B.A. Tomkins and W.H. Griest, Anal. Chem., 68 (1996) 2533.
94 K.-S. Boos and C.-H. Grimm, Trends Anal. Chem., 18 (1999) 175.
95 K.K. Unger, Chromatographia,31 (1991) 507.
96 J. Haginaka, Trends Anal. Chem., 10 (1991) 17.
97 K.-S. Boos, A. Rudolphi, S. Vielhauer, A. Walfort, D. Lubda and F. Eisenbeiss,
Fresenius'J.Anal. Chem., 352 (1995) 684.
98 Z. Yu and D. Westerlund, Chromatographia,44 (1997) 589.
99 C.-H. Grimm, K.-S. Boos, Ch. Apel, K. K. Unger, P. Onnerfjord, L. Heintz, L.-E.
Edholm and G. Marko-Vargra, Chromatographia,52 (2000) 703.
100 R. Koeber, C. Fleischer, F. Lanza, K.S. Boos, B. Sellergren and D. Barcelo, Anal.
Chem., 73 (2001) 2437.
101 E. Hogendoorn and P. van Zoonen, J. Chromatogr.A, 892 (2000) 435.
102 R.M. Izatt, J.S. Bradshaw and R.I. Bruening, Pure Appl. Chem., 68 (1996) 1237.
103 R.M. Izatt, J. Inclusion Phenom. Mol. Recogn. Chem., 29 (1997) 197.
104 N.E. Izatt, R.L. Bruening, K.E. Krakowiak and S.R. Izatt, Ind. Engin. Chem. Res., 39
(2000) 3405
105 D.M. Beals, W.G. Britt, J.P. Bibler and D.A. Brooks, J. Radioanal. Nucl. Chem., 236
(1998) 187.
106 P. Martin, B. Leadbetter and I.D. Wilson, J. Pharm.Biomed. Anal., 11 (1993) 307.
107 H. Tsuchiya, J. Chromatogr., B 720 (1998) 225.
108 J.M. Van Emon, C.L. Gerlath and K. Bowman, J. Chromatogr.B, 715 (1998) 211.
109 V. Pichon, M. Bouzige, C. Miege and M.-C. Hennion, Trends Anal. Chem., 18 (1999)
219.
110 N. Delaunay, V. Pichon and M.-C. Hennion, J. Chromatogr., B 745 (2000) 15.
111 D. Stevenson, J. Chromatogr., B 745 (2000) 39.
112 P.B. Carrasco, R. Escola, M.P. Marco and J.M. Bayona, J. Chromatogr. A, 909 (2001)
61.
113 N. Delaunay-Bertoncini, V. Pichon and M.-C. Hennion, Chromatographia,53 (2001)
S224.
114 M.-C. Hennion and D. Barcelo, Anal. Chim. Acta, 362 (1998) 3.
115 J. Dalluge, T. Hankemeier, R.J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr.A,
830 (1999) 377.
116 D. Stevenson, Trends Anal. Chem., 18 (1999) 154.
117 B. Sellergren, Trends Anal. Chem., 18 (1999) 164.
118 L. I. Anderson, J. Chromatogr.B, 739 (2000) 163.
119 L. I. Anderson, J. Chromatogr. B, 745 (2000) 3.
120 F. Lanza, B. Sellergren, Chromatographia,53 (2001) 599.
121 A. Martin-Esteban, Fresenius J. Anal. Chem., 370 (2001) 795.
122 J.G. Karlsson, L.I. Andersson and I.A. Nicholls, Anal. Chim. Acta, 435 (2001) 57.
123 A. Martin-Esteban, F. Turiel, D. Stevenson, Chromatographia, 53 (2001) S434.
124 P. Martin, I.D. Wilson and G.R. Jones, Chromatographia,52 (2000) S19.
125 M.-C. Hennion, C. Cau-Dit-Coumes and V. Pichon, J. Chromatogr.A, 823 (1998) 147.
126 C.F. Poole, S.K. Poole, D.S. Seibert and C.M. Chapman, J. Chromatogr.B, 689 (1997)
245.
127 C.F. Poole, A.D. Gunatilleka and R. Sethuraman, J. Chromatogr.A, 885 (2000) 17.

385
128 P. Lovkvist and J.A. Jonsson, Anal. Chem., 59 (1987) 818.
129 I. Liska, J. Chromatogr. A, 655 (1993) 163.
130 M.L. Mayer and C.F. Poole, Anal. Chim. Acta, 294 (1994) 113.
131 M.L. Larrivee and C.F. Poole, Anal. Chem., 66 (1994) 139.
132 D.S. Seibert and C.F. Poole, J. High Resolut. Chromatogr., 21 (1998) 481.
133 N. Masque, M. Galio, R.M. Marce and F. Borrull, J. Chromatogr.A, 771 (1997) 55.
134 V. Pichon and M.-C. Hennion, J. Chromatogr., 665 (1994) 269.
135 V. Pichon, L. Chen, S. Guenu and M.-C. Hennion, J. Chromatogr.A, 711 (1995) 257.
136 C.F. Poole, S.K. Poole and A.D. Gunatilleka, Adv. Chromatogr., 40 (2000) 159.
137 HG.J. Mol, J. Staniewski, H.-G. Jansson, C.A. Cramers, R.T. Ghijsen and U.A.Th.
Brinkman, J. Chromatogr., 630 (1993) 201.
138 H.G.J. Mol, H.-G. Janssen and C.A. Cramers, J. High Resolut. Chromatogr., 16
(1993) 413.
139 M.-C. Hennion and V. Coquart, J. Chromatogr., 642 (1993) 211.
140 A. Gelencser, G. Kiss, Z. Krivacsy, Z. Varga-Puchony and J. Hlavay, J. Chromatogr.
A, 693 (1995) 217.
141 A. Gelencser, G. Kiss, Z. Krivacsy, Z. Varga-Puchony and J. Hlavay, J. Chromatogr.
A, 693 (1995) 227.
142 K.G. Miller and C.F. Poole, J. High Resolut. Chromatogr., 17 (1994) 125.
143 E. Baltussen, F. David, P. Sandra, H.-G. Janssen and C.A. Cramers, J. Chromatogr.
A, 805 (1998) 237.
144 E. Baltussen, H. Snijders, H.-G. Janssen, P. Sandra and C.A. Cramers, J.
Chromatogr. A, 802 (1998) 285.
145 W.P.N. Fernando, M.L. Larrivee and C.F. Poole, Anal. Chem., 65 (1993) 588.
146 D. Bolliet and C.F. Poole, Chromatographia,46 (1997) 381.
147 D.S. Seibert and C.F. Poole, J. High Resolut. Chromatogr., 18 (1995) 226.
148 D.S. Seibert, C.F. Poole and M.H. Abraham, Analyst, 121 (1996) 511.
149 D.S. Seibert and CF. Poole, Chromatographia,41 (1995) 51.
150 D.S. Seibert and C.F. Poole, Anal. Commun., 35 (1998) 147.
151 M.L. Mayer, S.K. Poole and C.F. Poole, J. Chromatogr. A, 697 (1995) 89.
152 T. Baczek, M. Markuszewski, R. Kaliszan, M.A. van Straten and H.A. Claessens, J.
High Resolut. Chromatogr., 23 (2000) 667.
153 C.F. Poole and S.K. Poole, J. Chromatogr.A (2002), in press.
154 M.H. Abraham and J.C. McGowan, Chromatographia,23 (1987) 243.
155 M.H, Abraham, G.S. Whiting, R.M. Doherty and W.J. Shuely, J. Chem. Soc. Perkin
Trans., 2 (1990) 1451.
156 M.H. Abraham, C.F. Poole and S.K. Poole, J. Chromatogr. A, 842 (1999) 79.
157 J.A. Platts, D. Butina, M.H. Abraham and A. Hersey, J. Chem. Inf. Comput. Sci., 39
(1999) 835.
158 W. Kiridena and C.F. Poole, Analyst, 123 (1998) 1265.
159 J.H. Park, M.H. Yoon, Y.K. Ryu, B.E. Kim, J.W. Ryu and M.D. Jang, J. Chromatogr.
A, 796 (1998) 249.
160 F.Z. Oumada, M. Roses, E. Bosch and M.H. Abraham, Anal. Chim. Acta, 382 (1999)
301.
161 J. Li and D.A. Whitman, Anal. Chim. Acta, 368 (1998) 141.
162 J.C. Pearce, A. Churchill and A.C. Terry, Chemom. Intell. Lab. Sys., 17 (1992) 213.
163 P. Hubert, P. Chiap, M. Moors, B. Bourguignon, D.L. Massart and J. Crommen, J.
Chromatogr. A, 665 (1994) 87.
164 D.T. Rossi and N. Zhang, J. Chromatogr.A, 885 (2000) 97.
165 I. Liska, J. Chromatogr., 655 (1993) 163.
166 G. Font, J. Manes, J.C. Malto and Y. Pico, J. Chromatogr., 642 (1993) 135.
167 L. Bovanova and E. Brandstetenova, J. Chromatogr. A, 880 (2000) 149.

386
168 R. Sasano, T. Hamada, M. Kurano and M. Furuno, J. Chromatogr. A, 896 (2000) 41.
169 J.R. Veraart, H. Lingeman and U.A.Th. Brinkman, J. Chromatogr.A, 856 (1999)
483.
170 D. Martinez, M.J. Cugat, F. Borrull and M. Calull, J. Chromatogr.A, 902 (2000) 65.
171 M. Valcarcel, L. Arce and A. Rios, J. Chromatogr.A, 924 (2001) 3.
172 P. Simon, M. Lafontaine, P. Delsaut, Y. Morele and T. Nicot, J. Chromatogr.B, 748
(2000) 337.
173 S. Salado and L.E. Vera-Avila, J. Chromatogr.B, 690 (1997) 195.
174 E. Schoenzetter, V. Pichon, D. Thiebaut, A. Fernandez-Alba and M.-C. Hennion, J.
Microcol. Sep., 12 (2000) 316.
175 E.A. Hogendoorn, E. Dijkman, B. Baumann, C. Hidalgo, J.V. Sancho and F.
Hernandez, Anal. Chem., 71 (1999) 1111.
176 Z. Yu and D. Westerlund, Chromatographia,47 (1998) 299.
177 R.A.M. vander Hoeven, A.J.P. Hofte, M. Frenay, H. Irth, U.R. Tjaden, J. vander
Greef, A Rudolfi, K.-S. Boos, G.M. Varga and L.E. Edholm, J. Chromatogr.A, 762
(1997) 193.
178 R. Wissiack, E. Rosenberg and M. Grasserbauer, J. Chromatogr. A, 896 (2000) 159.
179 J. Slobodnik, H. Lingeman and U.A.Th. Brinkman, Chromatographia,50 (1999) 141.
180 N. Masque, R.M. Marce and F. Borrull, J. Chromatogr.A, 793 (1998) 257.
181 A.C. Hogenboom, M.P. Hofman, D.A. Jolly, W.M.A. Niessen and U.A.Th. Brinkman,
J. Chromatogr.A, 885 (2000) 377.
182 A.J.H. Louter, U.A.Th Brinkman and R.T. Ghijsen, J. Microcol. Sep., 5 (1993) 303.
183 J. Hankemeier, S.P.J. van Leeuwen, R.J.J. Vreuls and U.A.Th. Brinkman, J.
Chromatogr.A, 811 (1998) 117.
184 T. Hankemeier, A.J.H. Louter, J. Dalluge, R.J.J. Vreuls and U.A.Th. Brinkman, J.
High Resolut. Chromatogr., 21 (1998) 450.
185 T.H.M. Noij and M.M.E. van der Kooi, J. High Resolut. Chromatogr., 18 (1995) 535.
186 A.J.H. Louter, S. Ramalho, D. Jani, J.J. Vreuls and U.A.Th. Brinkman, J. Microcol.
Sep., 8 (1996) 469.
187 K.K. Verma, A.J.H. Louter, A. Jain, E. Pocurull, J.J. Vreuls and U.A.Th. Brinkman,
Chromatographia,44 (1997) 372.

387
Chapter 13

Solid phase microextraction

Janusz Pawliszyn

13.1 INTRODUCTION

Solid phase microextraction was developed to address the need for rapid sample
preparation both in the laboratory and on-site. With this the technique a small
amount of extracting phase dispersed on a solid support is exposed to the sample
for a well defined period of time. In one approach a partitioning equilibrium
between the sample matrix and extraction phase is reached. In this case
convection conditions do not affect the amount extracted. In a second approach
utilizing short time pre-equilibrium extraction, if convection/agitation are
constant, then the amount of analyte extracted is related to time. Quantification
can then be performed based on timed accumulation of analytes in the coating.
Figure 13.1 illustrates several implementations of SPME which have been
considered. They mainly include open bed extraction concepts such as coated
fibres, vessels, agitation mechanism disks, but in-tube approaches are also
considered. Some better address issues associated with agitation while others
address ease of implementing sample introduction to the analytical instrument.
It should be noted that solid phase microextraction was originally named after
the first experiment using an SPME device which involved extraction on solid
fused silica fibres, and later as such, as a reference to the appearance of the
extracting phase, relative to a liquid or gaseous donor phase, even though it is
recognized that the extraction phase is not always technically a solid.
Solid phase microextraction is often mistakenly considered as another form
of solid phase extraction or micro-SPE. However, there are significant differ-
ences between the methods. Solid phase extraction is essentially a three-step
process (see Chapter 12 "Principles and Practice of Solid-Phase Extraction" ). A
sample is initially passed through the sorbent bed, and analytes present in the
sample are exhaustively extracted from the sample matrix to the solid sorbent.
In a second step, unwanted analytes are selectively desorbed from the solid
sorbent by washing with a solution capable of desorbing unwanted analytes, but
leaving desired analytes retained on the sorbent. In the final step the wash solu-
tion is changed for one able to desorb analytes of interest. The resulting eluent
may then be concentrated by evaporation, to the desired volume. Solid phase
microextraction however takes advantage of equilibrium extraction and selec-
ComprehensiveAnalytical Chemistry XXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 389
 WIN Extraction Phase

Tube
I

F---I X1 particle, -----1- ' - ---

ti I I

Suspended Particles Stirrer Disk i Membrane

Fig. 13.1. Configurations of solid phase microextraction.

tive sorption from the matrix onto the coating. In the first step, the coating is
exposed to the sample and analytes with a high affinity for the sorbent are selec-
tively extracted. In the second step, everything extracted by the fibre is desorbed
into the analytical instrument. No intermediate clean-up step is normally imple-
mented. Micro-SPE is more related to SPE as it is a total extraction method, but
utilizes a reduced sample and sorbent volume. A comparison with SPME is
therefore inappropriate.
A degree of selectivity is required for any sample preparation method. It is
impractical to introduce all compounds present in a sample to an analytical
instrument. The method developed must eliminate compounds incompatible
with the instrument including matrix components. It is also desirable to remove
as many of the unwanted compounds as possible, to make the resulting data
interpretation as clean and simple as possible. Thus, with selective extraction,
sample preparation is simplified and typically results in significant savings in
time and precision.
Selectivity is therefore quite important when choosing an SPME coating.
High capacity, even for a range of analytes, is more important for solid phase
extraction, where prevention of break-through is a significant concern. Because
break-through is not an issue to be addressed in an equilibrium extraction
method such as SPME, more emphasis may be placed on sorbent selectivity.
SPME differs from SPE in another significant way: because of the large
volume of sorbent required relative to SPME, SPE sorbent has the potential to
retain non-adsorbed components in the void volume. It is difficult to design a
wash regime that removes unwanted compounds completely, without impacting
retention of the analyte(s) of interest. In this way there is the potential that
unwanted compounds may remain, either adsorbed, or present as non-adsorbed
analytes in the bulk of the sorbent. Because of the geometry of the SPME device,
and the modes of extraction used, unwanted analytes are not normally present
in the sorbent at the time of desorption.
SPME devices have an open-bed structure, relative to SPE devices where the
extraction medium is packed into a cartridge-like device. In SPME the surface of

390
the extraction phase is itself accessible for analysis. This is less true with in-tube
SPME, although limited surface characterization has been performed with the
capillaries as well. Therefore, with SPME it is possible to perform convenient
spectroscopic analysis of surface adsorbed components and not only extracted
chemical species, but also composition of collected aerosols or particulates. This
can have an important advantage in speciation and characterization of natural
system.
In this chapter the focus is placed on two primary implementations of the
technique to date: fibre and in-tube SPME. These have been explored theoretic-
ally and experimentally to a large extent. The first is based on externally coated
fibres mounted in a syringe-like needle for protection, and the second on
internally coated tubes or capillaries. General conclusions obtained in the
studies involving these systems can be extended to other geometries.

13.2 HISTORICAL PERSPECTIVE ON THE EVOLUTION OF SPME

TECHNOLOGY

The early work on laser desorption/fast gas chromatography conducted in our

laboratory resulted in rapid separation times, even for very high molecular mass
species [1]. However, the preparation of samples for this experiment took hours,
which was over an order of magnitude longer than the separation times. In this
experiment, optical fibres were used to transmit laser light energy to the GC
instrument. The sample preparation process was analogous to standard solvent
extraction procedures. The fibre tip was coated with the sample by dipping one
end of the optical fibre in the solvent extract, coating the fibre and then
removing volatile solvents through evaporation. The fibre tip, prepared in such a
way, was inserted into the injector of a gas chromatograph, and analytes were
volatilized onto the front of the GC column by means of a laser pulse. During
that work, a need was recognized for rapid sample preparation techniques to
retain the time efficiency advantages made possible by the use of the laser pulse
and a high speed separation instrument. The challenge was addressed using
fibres, since optical fibres could be purchased coated with several types of
polymeric films. The original purpose of these coatings was simply to protect the
fibres from breakage. Because of the thin films used (10-100 tm), the expected
extraction times for these systems were very short. In addition, novel films could
be prepared, since chromatographers have a good knowledge base on fused silica
coating methods gained from capillary columns manufacturing experience.
In the initial work on SPME, sections of fused silica optical fibres, both
uncoated and coated with liquid and solid polymeric phases, were dipped into an
aqueous sample containing test analytes and then placed in a GC injector [2].
The process of introducing and removing the fibres required the opening of the
injector which resulted in loss of head pressure at the column. Despite their basic
nature, those early experiments provided very important preliminary data that
confirmed the usefulness of this simple approach, since both polar and nonpolar

391
 syringe needle syringe barrel plunger cap

/o n \ic f stainless steel

coating silica fiber microtubing epoxy glue

Fig. 13.2. The custom-made SPME device based on Hamilton 7000 series syringe.

chemical species were extracted rapidly and reproducibly, from aqueous

samples.
The development of the technique accelerated rapidly with the implementa-
tion of coated fibres incorporated into a microsyringe, resulting in the first
SPME device [3]. Figure 13.2 shows an example of an SPME device based on the
Hamilton TM 7000 series microsyringe. The metal rod, which serves as the piston
in a microsyringe, is replaced with stainless steel microtubing having an inside
diameter (i.d.) slightly larger than the outside diameter (o.d.) of the fused silica
rod. Typically, the first 5 mm of the coating is removed from a 1.5 cm long fibre,
which is then inserted into the microtubing. High temperature epoxy glue is
used to permanently mount the fibre. Sample injection is then very much like
standard syringe injection. Movement of the plunger allowed exposure of the
fibre during extraction and desorption and its protection in the needle during
storage and penetration of the septum. SPME devices do not need expensive
syringes like the HamiltonM syringes. As Fig. 13.3a illustrates, a useful device
can be built from a short piece of stainless steel microtubing (to hold the fibre),
another piece of larger tubing (to work as a "needle"), and a septum (to seal the
connection between the microtubing and the "needle"). The design from Fig.
13.3a is the basic building block of a commercial SPME device described later

attach enbhub
a-attachment hub
" \\

sealing septum _
cooling
Coils

- septum piercing T
needle

microtubing -coating
coated fused silica fiber

Fig. 13.3. Simple versions of the SPME device using (a) coated fibre and (b) internally coated
tubing.

392
 Plunger

Barrel
-Plunger Retaining Screw
Z-slot

Hub-Viewing Window

Adjustable Needle
GuidelDepth Gauge

TensioningSprin
g , - SeptumPiercingNeedle
Sealing Septum
4 - FiberAttachment
Tubing
F u s ed
Coated SilicaFiber

Fig. 13.4. Design of the first commercial SPME device made by Supelco.

and illustrated in Fig. 13.4. Another simple SPME construction is based on a

piece of internally coated tubing [4]. This tubing can be mounted inside a needle
or it can constitute the "needle" of a syringe itself [5]. Elimination of mechanical
movement of a plunger of a syringe can be accomplished by sealing the tubing at
one end and installing a microheater as illustrated in Fig. 13.3b. Expansion of air
caused by temperature increase allows removal of desorbed analytes from the
extracting phase located inside the tubing. A coated tubing approach is useful in
the design of passive sampling devices discussed later, since in this case, the
extraction rate is limited by the diffusion of analytes into the needle [6]. In
addition, active sampling is possible by heating and cooling of air contained in
the upper part of the tubing, which causes movement of liquid or gaseous
samples into and out of the tubing, facilitating mass transport of analytes from
the sample to the coating.
This in-tube concept has also been expanded to facilitate automation of sam-
ple preparation for HPLC. In that approach the sample components are
extracted by the coating located on the inner surface of the hollow tubing and
after the extraction is completed the analytes are washed into the HPLC column
using the mobile phase or solvent. Everything is easily automated using a con-
ventional autosampler (see Section 13.3.6). This concept is very similar to SPE
(some researchers used packed tubes) [7]; however the difference is associated
with the principle of the process-total extraction vs. equilibrium-different
selectivities, less flow restrictions and taking full advantage of phase capacity.
The extraction coating has also been coated on other elements of the analyti-
cal system. Recently there have been reports of coating the interior of vessels [8]
the exterior of magnetic stirring bars [9] or even pieces of poly(dimethyl)siloxane
(PDMS) tubes and thin membranes [10]. Several of these implementations are
shown in Fig. 13.1. The main reason for developing these alternative approaches
is to enhance sensitivity by using a larger volume of the extraction phase
(PDMS) and improving kinetics of the mass transfer between sample and PDMS

393
by increasing the surface to volume ratio of the extraction phase. The main dis-
advantage of these approaches, however, is loss of convenience associated with
syringe configuration, in particular in introduction to the analytical instrument.
This step necessitates the use of high volume desorption devices and creates dif-
ficulties in automation of the extraction process as well as handling volatile com-
pounds which are lost during transfer of the extraction phase from sample to the
injection system. Since high sensitivities are obtained for hydrophobic high
molecular weight compounds with fibre SPME the advantages of using larger
volume phases are limited, especially for small sample volumes [11].
Although to date SPME devices have been used principally in laboratory
applications, more current research has been directed toward remote
monitoring, particularly for clinical, field environmental, and industrial hygiene
applications. In their operating principles, such devices are analogous to the
devices described above, but modifications are made for greater convenience in
given applications. For example, as shown in Fig. 13.5a, adding a tube with a
small opening to cover the needle of the SPME syringe results in a useful device
for breath analysis in a non-invasive clinical application [12]. This design can be
further improved by adding a one-way valve mounted on the aperture, but the
concept of operation remains the same.
An important feature of a field device is the ability to preserve extracted
analytes in the coating. The simplest practical way to accomplish this goal is to
seal the end of the needle with a piece of septum. Additionally, cooling extends
the storage time. Polymeric septum material, however, may cause losses of
analytes from the fibre. Therefore, a more appropriate approach is to use metal
to metal (or solid polymer) seals. Figure 13.5b illustrates an example of a device
construction based on a "leaf"' closure. It is anticipated that future devices
designed for field applications will be more rugged than the current laboratory
versions and will look more like "sticks" or "pens" than syringes.
Recently, different configurations of field samplers were evaluated for the
analysis of volatile compounds [13]. A percentage of compound retained in the
coating was evaluated for the four devices, at different storage times and

exposed fiber
aperture

inert tubing

(a)

Fig. 13.5. SPME device modified for (a) breath analysis and (b) field application.

394
temperatures, and for different fibre coatings. The PDMS fibre demonstrated
the lowest ability to retain these compounds. Carboxen-PDMS had the highest
and PDMS-DVB was intermediate. The devices employed various sealing
methods, to preserve the samples on the fibre, and to protect from external
contamination. The Supelco field sampler seals by retracting the outer needle
behind a silicone septum. Alternatively, a conventional manual sampler may be
sealed by pressing the end of the needle into a piece of septum. Two prototype
devices constructed were sealed by either a leaf system (Fig. 13.5b), which opens
automatically when the outer needle is exposed, or by capping the outer needle
after sampling, with a teflon cap.
Another promising technique uses a valve syringe, where the fibre is
withdrawn into the barrel of the syringe along with a sample of the air being
analyzed, by means of retracting the syringe plunger. The valve is then sealed
until analysis. Initial data describing the new approaches clearly demonstrate
advantages of the new designs to seal the fibre in the needle, compared to the
commercial devices, by eliminating the losses and contamination to the septa
material of volatile components.

13.3 PRINCIPLES OF SOLID PHASE MICROEXTRACTION

In SPME a small amount of extracting phase associated with a solid support is

placed in contact with the sample matrix for a predetermined amount of time. If
the time is long enough, a concentration equilibrium is established between the
sample matrix and the extraction phase. When equilibrium conditions are
reached, exposing the fibre for a longer time does not accumulate more analytes.
Two different implementations of the SPME technique have been explored
extensively to date (see Fig. 13.6): one associated with a tube design and the
other with fibre design. The tube design can use very similar arrangements as
SPE; however the primary difference, in addition to volume of the extracting
phase, is that the objective of SPME is never an exhaustive extraction. This
substantially simplifies the design of systems. For example, in-tube SPME for
analysis of liquids uses 0.25 mm i.d. tubes and about 0.1 41 of extraction phase,
because concern about breakthrough is not relevant since exhaustive extraction
is not an objective. In fact the objective of the experiment is to produce full
breakthrough as soon as possible, since this indicates equilibrium extraction has
been reached.
A more traditional approach to SPME involves coated fibres. The transport
of analytes from the matrix into the coating begins as soon as the coated fibre has
been placed in contact with the sample (Fig. 13.7). Typically, SPME extraction is
considered to be complete when the analyte concentration has reached distribu-
tion equilibrium between the sample matrix and the fibre coating. In practice,
this means that once equilibrium is reached, the extracted amount is constant
within the limits of experimental error and it is independent of further increase
of extraction time. The equilibrium conditions can be described as [14]:

395
 (a)

- fused silica fiber

extracting phase solid support
- Coating
Vf Sp
Sampte
(b)
Vi Co

Left: Fig. 13.6. Two different implementations of the SPME technique: (a) polymer coated on
outer surface of fibre; (b) polymer coated on internal surface of capillary tube.
Right: Fig. 13.7. Microextraction with SPME. Vt, volume of fibre coating; Kr,, fibre/sample
partition coefficient; V., volume of sample; CO, initial concentration of analyte in the sample.

n K= ,Vf Co (13.1)
KrfVf +V
where n is the number of moles extracted by the coating, Kf, is a fibre coating/
sample matrix distribution constant, Vf is the fibre coating volume, Vs is the
sample volume, and CO is the initial concentration of a given analyte in the
sample.
Strictly speaking, this discussion is limited to partitioning equilibrium
involving liquid polymeric phases such as poly(dimethylsiloxane). The method of
analysis for solid sorbent coatings is analogous for low analyte concentration,
since the total surface area available for adsorption is proportional to the coating
volume if we assume constant porosity of the sorbent. For high analyte concen-
trations, saturation of the surface can occur, resulting in nonlinear isotherms as
discussed later. Similarly high concentration of a competitive interference com-
pound can displace the target analyte from the surface of the sorbent.
Equation (13.1), which assumes that the sample matrix can be represented
as a single homogeneous phase and that no headspace is present in the system,
can be modified to account for the existence of other components in the matrix
by considering the volumes of the individual phases and the appropriate
distribution constants. The extraction can be interrupted and the fibre analyzed
prior to equilibrium. To obtain reproducible data, however, constant convection
conditions and careful timing of the extraction are necessary.
Simplicity and convenience of operation make SPME a superior alternative
to more established techniques in a number of applications. In some cases, the
technique facilitates unique investigations. Equation (13.1) indicates that, after
equilibrium has been reached, there is a direct proportional relationship
between sample concentration and the amount of analyte extracted. This is the
basis for analyte quantification. The most visible advantages of SPME exist at
the extremes of sample volumes. Because the set-up is small and convenient,
coated fibres can be used to extract analytes from very small samples. For
example, SPME devices are used to probe for substances emitted by a single
flower bulb during its life span; the use of sub-micrometer diameter fibres

396
permits the investigation of single cells. Since SPME does not extract target
analytes exhaustively, its presence in a living system should not result in
significant disturbance. In addition, the technique facilitates speciation in
natural systems, since the presence of a minute fibre, which removes small
amounts of analyte, is not likely to disturb chemical equilibria in the system. It
should be noted however, that the fraction of analyte extracted increases as the
ratio of coating to sample volume increases. Complete extraction can be achieved
for small sample volumes when distribution constants are reasonably high. This
observation can be used to advantage if exhaustive extraction is required. It is
very difficult to work with small sample volumes using conventional sample
preparation techniques. Also, SPME allows rapid extraction and transfer to an
analytical instrument. These features result in an additional advantage when
investigating intermediates in the system. For example, SPME was used to
study biodegradation pathways of industrial contaminants [15]. The other
advantage is that this technique can be used for studies of the distribution of
analytes in a complex multiphase system [16] and speciate different forms of
analytes in a sample [17].
In addition, when sample volume is very large, Eq. (13.1) can be simplified to:
n = KfVfCo (13.2)
which points to the usefulness of the technique for field applications. In this
equation, the amount of extracted analyte is independent of the volume of the
sample. In practice, there is no need to collect a defined sample prior to analysis
as the fibre can be exposed directly to the ambient air, water, production stream,
etc. The amount of extracted analyte will correspond directly to its concentra-
tion in the matrix, without being dependent on the sample volume. When the
sampling step is eliminated, the whole analytical process can be accelerated, and
errors associated with analyte losses through decomposition or adsorption on
the sampling container walls will be prevented. This advantage of SPME could
be enhanced practically, by developing portable field devices on a commercial
scale.

13.3.1 Extraction modes with coated fibre

Three basic types of extraction can be performed using SPME: direct extraction,
headspace configuration, and a membrane protection approach. Figure 13.8
illustrates the differences among these modes. In the direct extraction mode
(Fig. 13.8a), the coated fibre is inserted into the sample and the analytes are
transported directly from the sample matrix to the extracting phase. To
facilitate rapid extraction, some level of agitation is required to transport
analytes from the bulk of the solution to the vicinity of the fibre. For gaseous
samples, natural convection of air is sufficient to facilitate rapid equilibration.
For aqueous matrices, more efficient agitation techniques, such as fast sample
flow, rapid fibre or vial movement, stirring or sonication are required [6,18].

397
These conditions are necessary to reduce the effect caused by the "depletion
zone" produced close to the fibre as a result of fluid shielding and slow diffusion
coefficients of analytes in liquid matrices.
In the headspace mode, the analytes need to be transported through the bar-
rier of air before they can reach the coating. This modification serves primarily
to protect the fibre coating from damage by high molecular mass and other
non-volatile interferences present in the sample matrix, such as humic materials
or proteins. This headspace mode also allows modification of the matrix, such as
a change of the pH, without damaging the fibre. Amounts of analyte extracted
into the coating from the same vial at equilibrium using direct and headspace
sampling are identical as long as sample and gaseous headspace volumes are the
same. This is caused by the fact that the equilibrium concentration is independ-
ent of fibre location in the sample/headspace system. If the above condition is not
satisfied, a significant sensitivity difference between the direct and headspace
approaches exists only for very volatile analytes.
The choice of sampling mode has a very significant impact on extraction
kinetics, however. When the fibre coating is in the headspace, the analytes are
removed from the headspace first, followed by indirect extraction from the
matrix, as shown in Fig. 13.8b. Overall mass transfer to the fibre is typically
limited by mass transfer rates from the sample to the headspace. Therefore,
volatile analytes are extracted faster than semivolatiles since they are at a
higher concentration in the headspace which contributes to faster mass
transport rates through the headspace. Temperature has a significant effect on
the kinetics of the process by determining the vapour pressure of analytes. In
fact, the equilibration times for volatiles are shorter in the headspace SPME
mode than for direct extraction under similar agitation conditions. This outcome
is produced by two factors: that a substantial portion of the analyte is in the
headspace prior to extraction, and that diffusion coefficients in the gaseous
phase are typically four orders of magnitude larger than in liquid media. Since
concentrations of semivolatiles in the gaseous phase at room temperature are
typically small, however, overall mass transfer rates are substantially lower and
result in longer extraction times. They can be improved by using very efficient
agitation or by increasing the extraction temperature [19].

SampleHeadspace Fiber Membrane

Coaig Sample Coating Sample

a b c

Fig. 13.8. Modes of SPME operation: (a) direct extraction; (b) headspace SPME; (c) membrane-
protected SPME.

398
 a
' 15
o tO
I 10
xc
A

5
0 . 0

0 10 20 30 40 50 60 70 80

Extraction Time (minutes) Extraction Time (minutes)

Fig. 13.9. Time extraction profile obtained for headspace solid phase microextraction of several
PAHs from aqueous samples at (a) 75% and (b) 100% stirring rates. A, naphthalene; B,
acenaphthene; C, phenanthrene; D, chrysene.

Figure 13.9 illustrates the effect of agitation on the extraction time profile
obtained for polynuclear aromatic hydrocarbons (PAHs). As the rotational speed
of the magnetic stirrer increases, the equilibration time of naphthalene and
acenaphthene decreases from 8 min to 3 min and from 25 min to 10 min,
respectively. For less volatile analytes, phenanthrene and chrysene, the equili-
bration is not reached during the experimental period in either case, but the
amount of analyte extracted after 70 min extraction at low agitation (Fig. 13.9a)
is about the same as after 45 min of more efficient stirring (Fig. 13.9b). The use
of even more efficient agitation techniques, such as direct sonication, further
cuts the extraction time [20].
The effect of elevated sampling temperature is a lowering of fibre sample
partition coefficients. This decreases the amount extracted at equilibrium but it
may be acceptable if target limits of detection still can be reached. This effect is
demonstrated dramatically with the analysis of amphetamines. As shown in Fig.
13.10, room temperature extraction produces a very long equilibrium extraction
time, but ultimately the highest amount extracted. Conversely, the highest
temperature tested (73°C) produces a very short equilibration time (ca. 5 min)
but a significantly lower equilibrium amount extracted. An increase in the
Henry constant and diffusion coefficient improves kinetics at higher
temperature combined with a decrease in fibre capacity determined by the
distribution constant contributes to a substantial decrease in extraction time.
To enable the simultaneous analysis of very volatile substances (gases) and
less volatile analytes, the headspace SPME technique can be combined with
static headspace sampling by means of the gastight/SPME device illustrated in
Fig. 13.11 [21]. A fused silica fibre, coated with poly(dimethylsiloxane) (PDMS),
is connected with 30 gauge stainless steel (SS) tubing. The empty end of this SS
tubing/fibre assembly is then mounted to the plunger of a Hamilton 500 lI
gas-tight syringe. When the fibre is withdrawn into the syringe needle, a certain
volume of gas is also withdrawn into the gas-tight syringe through the needle
opening. For sampling, the fibre is first withdrawn into the syringe needle,

399
 plunger-
-fWs
syringe
barrel
f

!
Teflon
tip-

SS
tubing

needle -

fiberand
0 20 40 60 80 100 coating -
Time (min)

Left: Fig. 13.10. Effect of extraction temperature on equilibrium time and amount extracted, for
headspace methamphetamine analysis. Extraction conditions: PDMS 100 Am fibre, 0.5 M KOH,
saturated NaCi, 2 ml sample in a 4 ml vial, extraction temperature as shown, extraction time 15
min., analysis by GC-FID, desorption time 15 min. (): 22C; (): 40°C; (): 60°C; (): 73°C.
Right: Fig. 13.11. Gastight SPME device.

which is used to punch through the sample vial septum. The fibre is then
exposed into the headspace by lowering the plunger for a predetermined period
of time to establish analyte equilibrium among the coating, the headspace, and
the sample matrix. Movement of the plunger up and down can be used to
increase mass transport from the headspace to the coating and shorten the
equilibration times. Then the plunger is raised to a pre-marked position, which
allows a predetermined volume of headspace gas to be withdrawn into the
gastight/SPME device. The raised plunger also withdraws the fibre into the
needle for protection. Upon injection, the volume of headspace is injected at the
same time as analytes are desorbed.
Figure 13.12 illustrates the distribution of 28 typical volatile analytes,
between a PDMS coating of about 1 L1lvolume and the 110 /1 headspace con-
tained in the sampling device from Fig. 13.11. The presence of the gaseous
headspace in the gastight SPME device improves sensitivity only for very vola-
tile analytes with boiling temperatures below room temperature. The resulting
limits of detections (LODs) are mid to low parts per trillion (ppt) for the full
range of volatiles [21].
Figure 13.8c shows the principle of indirect SPME extraction through a
membrane. The initial purpose of the membrane barrier was to protect the fibre
against damage, similar to the use of headspace SPME when very dirty samples
are analyzed. Membrane protection is advantageous for determination of
analytes having volatilities too low for the headspace approach. In addition, a
membrane made from appropriate material can add a certain degree of
selectivity to the extraction process. The kinetics of membrane extraction are
substantially slower than for direct extraction though, because the analytes

400
 Combination
1.2

U) 0.8

)- 0.6
)

aZ 0.4

0.2

0
0 5 10 15 20 25 30
Elution Order

Fig. 13.12. Normalized peak area of 28 VOCs, which were sampled from the headspace of a
salt-saturated water sample at 100 ppb using three extraction methods; the peak area of VOCs
sampled by the gastight SPME was normalized to 1 for all analytes. The x axis gives the elution
order numbers of VOCs.

must diffuse through the membrane before they can reach the coating. The use
of thin membranes and increased extraction temperatures will result in faster
extraction times [22]. The thicker membranes can be used to slow down the
mass transfer through the membrane resulting in the time weighted average
(TWA) measurement for aqueous and gaseous samples discussed latter.

13.3.2 Extraction modes with in-tube SPME

Figure 13.13 illustrates that there are two fundamental approaches to in-tube
SPME: active or dynamic where the analytes are passed through the tube, and
passive or static where the analytes are transferred into the sorbent using
diffusion. In either approach, the coating may be supported on a fused silica rod,
or coated on the inside of a tube or capillary. Below we briefly discuss the
theoretical aspects of the extraction processes which use these geometric
arrangements.

Dynamic in-tube SPME

In this system we assume the use of a piece of fused silica capillary, internally
coated with a thin film of extracting phase (a piece of open tubular capillary GC
column), or that the capillary is packed with extracting phase dispersed on an
inert supporting material (a piece of micro-LC capillary column [23,24].
The front of analyte migrates through the capillary with a speed
proportional to the linear velocity of the sample, and inversely related to the
partition ratio. For short capillaries with a small dispersion, the extraction time
can be assumed to be similar to the time required for the centre of the band to
reach the end of the capillary. The extraction time is proportional to the length

401
 a b
- ~~~~~~~~~--·---
~_ --f-
II

P I
I

:6 "j i

5
mc:
9

static sample t sample flow :

Fig. 13.13. Comparison of (a) passive versus (b) dynamic modes of in-tube extraction.

of the capillary and inversely proportional to the linear flow rate of the fluid.
Extraction time also increases with an increase in the coating/sample distribu-
tion constant and with the thickness of the extracting phase but decreases with
an increase in the void volume of the capillary. An increase in the coating/sample
distribution constant produces an increase in the absolute amount extracted. It
has been observed that increases in amounts extracted can be achieved in many
cases by pre-conditioning the capillary with methanol or some other appropriate
solvent, prior to extraction. Enhancement has even been observed when a plug
of methanol is aspirated into the capillary before the sample is drawn in, and
follows the sample plug in the capillary during the extraction aspirate/dispense
steps. This is analogous to the solvent pre-conditioning used in SPE to enhance
extraction.
In practice, in-tube SPME is implemented by replacing a section of the
tubing in a commercially available autosampler, and then programming the
autosampler to pass sample in and out of the extraction capillary until equilib-
rium or a suitable extraction level has been reached. Several options for
implementing in-tube SPME are summarized in Fig. 13.14.
It should be emphasized that the above discussion is valid only for direct
extraction when the sample matrix passes through the capillary. This approach
is limited to particulate-free gas and clean water samples. The headspace SPME
approach can broaden the application of in-tube SPME. In that case, careful con-
sideration to the mass transfer between sample and headspace should be given
in order to describe the process properly. Also, if the flow rate is very rapid pro-
ducing turbulent behaviour and the coating/sample distribution constant is not
very high, then perfect agitation conditions are met and Eq. (13.3) can be used to

402
 HPLC Pump HPLC Pump

Buffer tubing Column

Column -Column

Six-port valve alve

t a) Transfer Line In-tube SPME

Sample vial
In-tube SPME capillary
rdInjection
Loop In-tube SPME

P
c) Flush Loop In-tibe SPME To column

Fig. 13.14. Schematics of different implementations of in-tube SPME.

estimate equilibration times. In this case equilibration time, t,, is assumed to be

achieved when 95% of the equilibrium amount of analyte is extracted from the
sample:

te = t95% b (13.3)
2Dr
In this equation, bt refers to the thickness of the sorbent material, and Df refers
to analyte diffusion coefficient in the sorbent.
The removal of analytes from a tube is an elution problem analogous to
frontal chromatography and has been discussed in detail [24]. In general, if the
desorption temperature of a GC is high and thin coatings are used, then all the
analytes are in the gas phase as soon as the coating is placed in the injector, and
the desorption time corresponds to the elution of two void volumes of the
capillary. For liquid desorption (into a liquid chromatography system, for exam-
ple) (Fig. 13.15), the desorption volume can be even smaller since the analytes
can be focused at the front of the desorption solvent [25].

Static in-tube SPME time weighted average sampling

In addition to the analyte concentration measurement at a well defined place in
space and time, obtained by using the approaches discussed above, an integrated
sampling is possible with a simple SPME system. This is particularly important

403
 Dispenser

*_Sf~'l~ Y~ ' Six-port valve

A
'ME capillary INJECT position

LOAD position

Adjustable capillary guideldepth gauge

Adjustable needle guideldepth gauge

Vial retainer arm
- Septum piercing needle

ample vial

Fig. 13.15. SPME with modified needle to allow in-needle extraction and automated flow-
through analysis of small samples and automation of a coated tubing solid phase microextractor
using Spark Holland Autosampler.

in field measurements when changes of analyte concentration over time and

place variations, must often be taken into account.
When the extracting phase is not exposed directly to the sample, but is con-
tained in a protective tubing (needle) without any flow of the sample through it
(Fig. 13.16a), the extraction occurs through the static gas phase present in the
needle. The integrated system can consist of extraction phase coating the inte-
rior of the tubing, or it can be an externally coated fibre withdrawn into the nee-
dle. These geometric arrangements represent a very powerful method which is
able to generate a response proportional to the integral of the analyte concentra-
tion over time and space (when the needle is moved through the space) [26]. In
these cases, the only mechanism of analyte transport to the extracting phase is
diffusion through the gaseous phase contained in the tubing. During this pro-
cess, a linear concentration profile (shown in Fig. 13.16) is established in the tub-
ing between the small needle opening, characterized by surface area A and the
distance Z between the needle opening, and the position of the extracting phase.
The amount of analyte extracted, dn, during time interval, dt, can be calculated
by considering Fick's first law of diffusion [27]:

dn = AD dt = AD Ctdt (13.4)
dz Z
where AC(t)/Z is a value of the gradient established in the needle between needle
opening and the position of the extracting phase, Z; AC(t) = C(t)-Cz, where C(t) is
a time dependent concentration of analyte in the sample in the vicinity of the
needle opening, and C. is the concentration of the analyte in the gas phase in the

404
 A

a i) I

-C=-
s M RX, concentration
gradient inside tube

C=O0

Addionai groves (5 mm spaced)

L Oto30mm

SPME fiber inside the needle

Fig. 13.16. Use of SPME for in-tube time weighted average sampling: (a) schematic, (b)
adaptation of commercial SPME manual extraction holder.

vicinity of the coating. Cz is close to zero for a high coating/gas distribution

constant capacity, then: AC(t) = C(t). The concentration of analyte at the coating
position in the needle, Cz, will increase with integration time, but it will be kept
low compared to the sample concentration because of the presence of the sorbing
coating. Therefore the accumulated amount over time can be calculated as:

n =D - C(t)dx (13.5)

As expected, the extracted amount of analyte is proportional to the integral of

the sample concentration over time, the diffusion coefficient of analytes in gas-
eous phase, Dg, in the area of the needle opening, A, and inversely proportional to
the distance of the coating position in respect of the needle opening, Z. It should
be emphasized that Eq. (13.5) is valid only in a situation where the amount of
analyte extracted onto the sorbent is a small fraction (below RSD of the mea-
surement, typically 5%) of the equilibrium amount in respect to the lowest con-
centration in the sample. To extend integration times, the coating can be placed
further into the needle (larger Z), the opening of the needle can be reduced by
placing an additional orifice (smaller A), or a higher capacity sorbent can be
used. The first two solutions will result in a low measurement sensitivity. An
increase of sorbent capacity presents a more attractive opportunity. It can be
achieved by either increasing the volume of the coating, or its affinity towards
the analyte. An increase of the coating volume will require an increase of the
device size. The optimum approach to increased integration time, is to use
sorbents characterized by large coating/gas distribution constants.

405
 The exploitation of restricted access to the absorbing medium allows the
implementation of SPME for time-weighted average (TWA) sampling. Where
diffusion to the sorbent surface is limited, the sorbent can act as a sort of "zero
sink" such that extraction is very far from equilibrium under normal sampling
conditions. In practice then, any analytes reaching the sorbent surface are
absorbed, essentially exhaustively. The rate of diffusion however, is still depend-
ent on the sample concentration, so the total amount absorbed by the coating is
proportional to the average of analyte concentrations over time, hence time-
weighted average sampling is achieved. This has been implemented to date with
the conventional fibre assembly, by retracting the fibre a known distance inside
the needle (Fig. 13.16b). The small size of the needle orifice limits diffusion to
the sorbent surface, and the ultimate diffusion rate is also a function of the
distance between the fibre tip and the end of the needle. Depending on the
volatility and concentration of the analyte of interest, the fibre may be
positioned either closer to or further from the end of the needle, to achieve the
desired degree of non-equilibrium extraction and sensitivity. It would also be
possible to implement this type of sampling with the sorbent coated on the
interior wall of a capillary. To date however, the retractable needle implementa-
tion has gained the most attention, due to its ease of use and adjustability for the
analyte and sample at hand.

Time-weighted averagesamplers
The theory of TWA sampling using an SPME device, was evaluated both in the
laboratory using a standard gas mixture of airborne hydrocarbons, and by
comparison to standard methods for field sampling [28,29]. One hundred ,um
PDMS fibre was used in the device shown in Fig. 13.16b to determine mass
uptake rates or sampling rates (SR). These values were determined for Z values
of 0.3 cm, 1 cm, and 3 cm. The intercept value for the linear uptake rate equation
represents the amount adsorbed on the stainless steel tubing. The higher the
molecular mass of the compound, the higher is the intercept value. However, for
a specific analyte, the SR value is proportional to the concentration and
diffusion coefficient of the analyte in the gas phase and inversely proportional to
the pathlength Z. Therefore if these SR values are used to determine an
unknown concentration without accounting for analyte adsorbed to the needle, a
positive error will be introduced, which is the intercept value. This error
becomes more significant as the volatility of the compound decreases.
Experimental values of uptake rates for the most volatile compounds like
pentane and hexane and PDMS as a coating were less than expected from the
theory. These compounds have high vapour pressure and diffusion coefficients,
which allows a fraction of them to leave the fibre coating before desorption in the
GC. In addition, due to their low partition coefficients for PDMS, these com-
pounds reach equilibrium in a short time which makes it difficult to load less
than 5% of the equilibrium amount. These problems were solved by using an
adsorptive coating like PDMS-DVB.

406
 The problem of adsorption of the least volatile compounds on the stainless
steel tubing is aggravated by retracting the fibre further into its housing needle
(3 cm). The area of the adsorptive surface increases as we retract the fibre
further into the needle. However, retracting the fibre further slows the
equilibrium process and minimizes the loss of the most volatile compounds. It
may be possible to use the PDMS-DVB fibre to minimize the loss of volatile
compounds from the fibre coating.

Field TWA sampling using the modified SPME device

Indoor air samples were collected. Active TWA sampling through charcoal tubes
was carried out following National Institute for Occupational Safety and Health
(NIOSH) method 1550 for determination of hydrocarbons in air. A mass flow
controlled air sampling pump was used to draw air through small charcoal tubes
at a certain flow rate, which was calibrated with a bubble flow meter. The results
obtained from the field study for the determination of TWA concentration of
airborne styrene using the SPME device with PDMS coating, and charcoal tubes
are summarized in Table 13.1.
In summary, the commercially available SPME fibre can be successfully used
as a diffusive TWA sampler for volatile and semi-volatile compounds. The com-
mercial SPME fibre should be modified for TWA sampling according to Fig.
13.16b. Deactivated stainless steel needles should be used when sampling com-
pounds with very low vapour pressure such as pentadecane. There is a need for a
new coating which has high capacity for volatile compounds. This coating should
have a strong binding phase that can resist high temperatures. The main advan-
tage of using an SPME device as a TWA sampler over other available techniques
is the ability to change the distance between the coating and needle opening.
This makes the technique very flexible for lower and higher concentrations and
for different sampling times. TWA sampling by SPME requires no pumps and no
polluting organic solvents, which significantly reduces sampling costs. The
above approach is not limited to sampling of gaseous phases. It was also applied
to passive sampling in aqueous media [30]. It was observed however that
adsorbtion of highly hydrophobic compounds on the needle surface might limit
performance of the technique and therefore careful choice of materials and sur-
face modification should be used for construction of the needles. Alternative way
in reducing access of the analytes to the extraction phase can include use of
membranes as indicated in Fig. 13.8c. This approach would be particularly use-

TABLE 13.1
Summary of data for analysis of styrene, obtained from field study using (A) TWA charcoal
tubes, (B) TWA by SPME with 100 gam PDMS fibre (retracted 0.3 cm)
Sample type SPME 100 Am PDMS Charcoal tubes Passive badge PID
5 min sample (g/l) 130 97 90 50-250
30 min TWA (g/1l) 56 54 72 N/A

407
ful for designing TWA devices for aqueous sample monitoring using hydrophilic
membranes since this approach reduces effect of needle surface adsorbtion.

13.3.3 Fibre coatings

Equations (13.1) and (12.2) indicate that the efficiency of the extraction process
is dependent on the distribution constant Kfs. This is a characteristic parameter
that describes properties of a coating and its selectivity toward the analyte
versus other matrix components. Specific coatings can be developed for a range
of applications. Coating volume determines method sensitivity as well (see Eq.
(13.1)), but thicker coatings result in longer extraction times (Eq. (13.3)).
Therefore, it is important to use the appropriate coating for a given application.
This is clearly demonstrated in Fig. 13.17, which compares the performance of
two different coatings for analysis of polar and nonpolar compounds from an
aqueous matrix. The distribution constant and the sensitivity of the method
drop over two orders of magnitude for o-xylene, and increase by an order of
magnitude for 2,4-dichlorophenol when the film is changed from nonpolar
PDMS (Fig. 13.17a) to polar poly(acrylate) polymer (PA) (Fig. 13.17b) [31].
Coating selection and design can be based on chromatographic experience.
To date, several experimental coatings have been prepared and investigated
for a range of applications. In addition to liquid polymeric coatings such as
PDMS for general applications, other more specialized materials have been
developed. For example, ion exchange coatings were used to remove metal ions
and proteins from aqueous solutions [32,33], liquid crystalline films to extract
planar molecules and carbowax for polar analytes [2], metal rods to electro-
deposit analytes [19,34], pencil "leads" to extract pesticides [35], and Nafion
coatings to extract polar compounds from nonpolar matrices [36].
Polypyrole coatings have been recently developed to extract polar or even
ionic analytes and possibly to explore the conductive polymer properties of the
polymer [37]. This could involve applying a charge to the polymer during
extraction in order to selectively extract analytes of interest, and then reversing
the charge to facilitate desorption. One of the main difficulties limiting the wide

oxylene
a E b
E)
:
/ K=794 2,4-dichlorophenol
0C .

0 o-xylene K\\
K =4
a
-FU
-~ i , \ --
`
A~~~~~~~~1
~ ~15 - -
-7-~~~~~~~~~~~~~~-
a6

-~~~~~~~~~~~~~~~~~~~
" -- --
10 10 15
Retention time (min) Retention time (min)

Fig. 13.17. Total ion current GC-MS chromatogram of benzene, toluene, ethylbenzene, and
o,m,p-xylenes (BTEX) and phenol in water extracted with: (a) a poly(dimethylsiloxane) coating,
and (b) a poly(acrylate) coating.

408
application of SPME/LC is the absence of a suitable stationary phase that not
only has high extraction ability for the polar analytes but is also stable in
solutions of various matrices. Polypyrole coatings have been investigated to fill
this gap.
Conducting polymers are versatile materials in which molecular/analyte
recognition can be achieved in different ways, including: (1) the incorporation of
counter ions that introduce selective interactions, (2) utilising the inherent and
unusual multifunctionality (hydrophobic, base-acid and TI-it interactions, polar
functional groups, ion-exchange, hydrogen bonding, electroactivity, and etc.) of
the polymers, (3) the introduction of functional groups to the monomers, (4) the
co-deposition of metals or other monomers within the polymer, and (5) the
application of appropriate electrochemical potentials. Based on these properties,
conducting polymers have found wide application in many fields including
separation science [38], chemical sensors [39] and electrochemical analysis [40].
So far, the widely used conducting polymers are based on polypyrrole, poly-
thiophene and polyaniline. Of these three classes of materials, polypyrrole (PPY)
and its derivatives have seen the most use and have been intensively studied in
recent years, due to additional advantages: (1) they can be easily polymerized
from organic or aqueous media at neutral pH by electrochemical or chemical
methods, (2) they are relatively stable in air and solution, and (3) pyrrole
monomer and some of its derivatives are available commercially.
Polypyrrole has been coated on the inner surface of a fused silica GC
capillary by the chemical polymerisation method. This PPY coated capillary has
been used successfully for automated in-tube SPME and determination of
B-blockers in urine and serum samples when coupled with liquid chromatogra-
phy/electrospray ionisation mass spectrometry (LC/ESI-MS) [41]. Compared
with the previous study using Omegawax 250 capillary for extraction of
P-blockers [42], PPY coated capillary has shown better extraction ability for
most of the compounds studied and therefore lower detection limits have been
achieved. This method can be extended to other compounds such as amine
containing drugs and anionic species due to the acid-base interaction and ion
exchange properties of polypyrrole coating.
A very pronounced difference in selectivity toward target analytes and
interferences can be achieved by using surfaces common to affinity chromatog-
raphy. Using the method of polymer imprinting [43], antibody mimics can be
generated with specificities to an analyte of choice. Briefly, the desired affinity
can be introduced by adding an amount of the compound of interest to the
polymerization reaction. This "pattern" chemical may be removed after poly-
merization, leaving vacant sites of a specific size and shape, suitable for binding
the same chemical again from an unknown sample. In this technique, we have
observed that non-specific binding should be controlled for, but that enhance-
ments in sensitivity are seen, particularly at low analyte concentrations.
Antibodies shows high affinity and extraordinary specific recognition ability
towards their complementary antigen in biological systems. Immunoassays,

409
 Absorption Adsorption Adsorption
(large pores) (small pores)

Fig. 13.18. Schematic representation of absorptive vs. adsorptive extraction, and adsorption in
small vs. large pores.

which is an analytical method based on the antibody-antigen specific reactions,

has become an essential routine and research tool throughout the biological
sciences. It is particularly useful in clinical analysis because of its unique
selectivity, extremely low limits of detection and applicability to a variety of
compounds. The selectivity of antibodies allows for the quantification of
compounds at trace levels in the presence of other structurally or chemically
similar compounds. They enable a selective preconcentration that is essential for
the analysis of complex samples such as biological matrices. Therefore, affinity
supports, such as those based on immobilized antibodies, play an ever-increasing
role in the development of new biological separation and analysis methods [44].
Recently, the extraordinary selectivity property of antibodies were combined
with SPME by immobilization of the antibodies on the fused silica fibre. The
specificity of immobilized antibodies, the maximum antigen binding capacity
and the competitive binding of different antigens to the immobilized antibodies
were studied and applied practically to determine concentration of drug
(antigen), was determined in biological samples [45].
There is a substantial difference in performance between the liquid and solid
coatings (Fig. 13.18). In the case of liquid coatings the analytes partition onto the
extraction phase, where the molecules are solvated by the coating molecules.
The high diffusion coefficient in the liquid coating allows the molecules to
penetrate the whole volume of the coating within a reasonable extraction time, if
the coating is thin. In the case of solid sorbents the coating has a well defined
crystalline structure, which if dense, substantially reduces the diffusion
coefficients within the structure. Therefore within the experimental time the
extraction occurs only on the surface of the coating. This can be demonstrated by
considering extraction of proteins illustrated in Fig. 13.19. The original mixture
contains three compounds: myoglobin, cytochrome and lysozyme. During fibre
extraction with polyacrylic acid, compounds with weaker affinity are only ob-
served at short extraction times. When extraction time is longer the displace-
ment of analytes with lower affinities occurs. In this case, lysozyme with a

410
 Originalsampleconsistingof myoglobin Proteinsadsorbed onto and
ontothen
(M),cytochrore C (C), and ysozyme (L), releasedfrom a po yacrylc acid
coatedfiber L
C
L

L C

Min t,t = 5 s te, = 240 s

0 5 10

Fig. 13.19. Protein extraction using poly(acrylic) acid coated fibre. Demonstration of myoglobin
and cytochrome C displacement by lysozyme with time.

stronger affinity for the coating, replaces the other two compounds during
extraction. This effect is associated with the fact that there is only limited
surface area available for adsorption. If this area is substantially occupied then
the displacement effects occur [46,47] and the equilibrium amount extracted can
vary with concentrations of both the target and other analytes. In extraction of
analytes with liquid coatings on the other hand, partitioning between the sample
matrix and extraction phase occurs. In this case, equilibrium extraction
amounts vary only if the coating property is modified by the extracted compo-
nents, which only occurs when the amount extracted is a substantial portion (a
few percent) of the extraction phase. This is very rarely observed since SPME is
typically used to determine trace contamination samples.
One way to overcome this fundamental limitation of the porous coatings is,
as the above figure suggests, to use an extraction time much less than the equi-
librium time, so that the total amount of analytes accumulated onto the fibre is
substantially below the saturation value. When performing such experiments,
not only is it critical to precisely control extraction times, but also convection
conditions must be monitored to ensure that they are constant or can be com-
pensated for. One way of eliminating the need for compensation of convection is
to normalize (use the same) agitation conditions. For example, the use of stirring
means at well defined rotation rates in the laboratory or fans for field air moni-
toring will ensure consistent convection.
The short time exposure SPME measurement described has an advantage
associated with the fact that the rate of extraction is defined by diffusivity of
analytes through the boundary layer of the sample matrix, and their
corresponding diffusion coefficients, rather than distribution constants. This is
demonstrated in Fig. 13.20.
The amouNt of analytes accumulated on the fibre as a function of time for
fibre geometry can be estimated as:

n= 2DLCt (13.6)
ln((b +6)/ b)

411
 benzene y= 546.7x -1188, 10 = 0.9964
2=
,.l, ,, v = 514.7x
+ 1. R 09986
3 E+04-
Diffusion Coefficients
E
benzene: 0.088
2.E+04
-
toluene: 0.084

IL E-04 p- xylene 0:071

0E+00
0 10 20 30 40 50 60 70
Sampling time (s)

Fig. 13.20. Correlation of uptake rate with diffusion coefficient, for short sampling time
(non-equilibrium) extraction of VOCs by PDMS-DVB fibre.

where n is the mass of extracted analyte over sampling time (t) in ng; D, is the
analyte molecular diffusion coefficient in sample matrix (cm2/s); b is the outside
radius of the fibre coating (cm); L is the length of the coated rod (cm); 6 is the
thickness of the boundary layer surrounding the fibre coating (cm); and C is
analyte concentration in sample matrix (ng/ml).
The differences in diffusion coefficients between compounds are small
compared to the differences in distribution constants. This makes it easier to
calibrate the system. Because of the large differences in distribution constants
between analytes, the resulting chromatograms are characterized by small peak
areas for compounds with small distribution constants, and large areas for those
with large constants. With uptake dependent on diffusion coefficients, all
compounds in a chromatogram with similar molecular masses will have similar
peak areas, given similar detector responses. Also, it is relatively simple to
calculate the diffusion coefficients for a given analyte and therefore correct for
the small differences in it. It must be understood that this system is only suitable
for trace analysis. When sample concentrations become too high, saturation of
the active sites occurs, and uptake rates are no longer linear. Shorter exposure
times, where smaller amounts are extracted, can solve this problem. Also, at
these higher concentrations, samples are easily extracted and analyzed with
PDMS fibre, using conventional SPME extraction methods. Results of extrac-
tion by the diffusion type approach are shown in Fig. 13.21. Accumulation of
volatile components on the solid coating in 10 s is much larger compared to the
10 min equilibrium extraction on PDMS. This approach to extraction is not
limited to devices using the fibre geometry, but is generally applicable.
At this point it might be instructive to draw parallels between developments
and applications of SPME and electrochemical methods. The coulometric tech-
nique corresponds to the total extraction method. Although the most precise,
this technique is not often used because of time required to complete it. SPME is
capable of producing exhaustive extraction when the volume of the extraction
phase is large enough combined together with high distribution constants. In
fibre geometry, the larger volume translates into thicker coatings which results

412
 A -PDMS - 10min, B - PDMS/DVB -10 s
0.8
A

1 2 netn
. .. I ----
...
2. hexane
3 heptane
4 octane
5 nonane
6. decade
7 undecane
8. dodecane
7 8
-11n
Fig. 13.21. Comparison of VOC mass loading I I
between PDMS and PDMS/DVB. 1 2 3 4 5 6 (min)

in long extraction times. The alternative approach is to disperse the whole vol-
ume of the extraction phase onto a larger surface area, resulting in a thinner
coating and faster equilibration times. For example, solid support material may
include particulate matter, a stirring mechanism or the vessels walls (see Fig.
13.1). In this case, however, there would be more handling required to conve-
niently introduce the extraction phase into the sample introduction system (GC
or HPLC). It might necessitate the use of organic solvent to desorb the analytes
from the extraction phase. Equilibrium potentiometric techniques are more fre-
quently used (pH electrode), particularly in cases where the sample is a simple
mixture and/or selectivity of the membrane is sufficient to quantify target
analyte in complex matrices. The equilibrium SPME method has some advan-
tages in this regard, since the technique is typically coupled with separation
and/or mass spectrometry detection methods, which allows identification and
quantification of many components simultaneously. The advantage of electro-
chemical methods is response time because of low capacities of electrodes.
Some electrochemical methods, like amperometry, are based on mass trans-
port through the boundary layers as in pre-equilibrium SPME. Analogously in
SPME, calibration based on diffusion coefficients can be accomplished as long as
the agitation conditions are constant and the extraction times are short and
coating has high affinity towards the analytes. Figure 13.21 illustrates the
results obtained for the 10 s extraction times using solid coating. In some
implementations of the technology the rate of mass transfer to the extraction
phase can be purposely restricted by placing it in the needle and therefore
achieving the time-weighted average measurements of concentration in a
specific time period (Fig. 13.16).
The compounds adsorbed onto the Carboxen coating are strongly bound.
This fact could be used to prepare solvent-free standards. Table 13.2 shows that

413
TABLE 13.2
Effect of fibre exposure time on the amount of BTEX remaining on the 75-gAm thick carboxen
coated fibre
Compounds Amount of BTEX (%) remaining on the fibre";
2 min 3 min 4 min 5 min 6 min
extraction extraction extraction extraction extraction
2 min 3 min 4 min 5 min 6 min
exposure exposure exposure exposure exposure
Benzene 98.6 99.0 97.7 96.1 89.9
Toluene 99.5 99.2 100.3 101.2 93.6
Ethylbenzene 100.7 99.1 100.9 97.7 94.5
p-Xylene 99.8 100.3 103.1 95.7 95.2
o-Xylene 100.6 99.9 99.2 95.5 93.7
1,3-Dichlorobenzene 99.3 100.8 100.1 98.6 97.2
1,1,2-Trichloroethane 98.8 99.7 98.1 96.7 93.5
Tetrachlloro-ethylene 100.1 101.3 98.5 97.2 94.1
*'Compared with the corresponding data obtained by 2-min, 3-min, 4-min, 5-min and 6-min
extraction without exposure to the air, respectively.

75-/um carboxen coating could preserve adsorbed components even through a

6-4 min exposure to zero air under the same conditions. Additional needle pro-
tection approaches, as in the field samplers discussed earlier, would result in
improved stability of the compounds on the coating. Therefore, the fibre can be
used to deliver solvent-free standards into analytical instruments for easy cali-
bration. To accomplish this, a SPME coated fibre can initially be introduced for a
well defined time to a standard gas or liquid containing the calibration compo-
nent to load a well defined mass of standard onto the fibre. Then the fibre can be
introduced to the analytical instruments to perform the calibration. This proce-
dure can be conveniently performed in the field. In addition, this approach can
be used in field sampling to check for losses and potential contamination of the
coating during storage. Leaks could be detected by the loss of standard from the
coating. Also, it is possible to perform multiple integrating sampling with one
fibre. This same coating can be exposed to samples from different locations or
times and/or standards to facilitate integrated form of determination together
with calibration [48].

Fibrepreparation
There are several different methods of depositing coatings onto fibres. The
dipping technique typically consists of placing a fibre in a concentrated organic
solvent solution of the material to be deposited for a short time. After removal of
the fibre from the solution, the solvent is evaporated by drying and the deposited
material can be crosslinked [2]. An extension of this method is electrodeposition,
which can be used to deposit selective coatings on the surface of metallic rods
[31]. The limitation of this approach is coating thickness variance from fibre to

414
fibre. Therefore, the preparation of films for commercial devices is carried out
simultaneously during the drawing of the fused silica rod. While this process
requires dedicated and very expensive equipment, reproducibility of the coating
thickness is excellent. This process is identical to the preparation of optical
fibres [49], and so the equipment needed is commercially available. Similar
results can be obtained using a piece of hollow fibre membrane (small i.d.
tubing), made from the desired extraction material. Preparation consists of
swelling the membrane by means of an appropriate volatile solvent, placing the
enlarged membrane onto the tip of the fibre, and evaporating the solvent. The
thickness of the coating is determined by the membrane thickness. Therefore
the volume of the coating can be large, reaching to 3 gl for a 300 m thick PDMS
hollow fibre membrane. A porous hollow fibre membrane can also be used for
adsorption of target analytes, or its pores can be filled with organic solvent to
allow for solvent microextraction [50].

13.3.4 Experimental parameters affecting extraction efficiency

A better understanding of SPME theory has allowed us to more rationally

optimize extraction conditions [51]. This is particularly important for equilib-
rium of a dynamic method. When performing extraction from gaseous samples
at short sampling times (pre-equilibrium) it has been observed that increasing
wind speed, up to about 5 cm/s (10 ft/min), enhances extraction. Figure 13.22
shows the effect of wind speed on the adsorption of benzene, p-xylene and
toluene from air samples on a 65 Am PDMS-divinylbenzene (DVB) fibre. The
curves indicate that the wind speed or air bulk movement significantly affects
the volatile organic contaminant (VOC) mass transfer process from the bulk air
to the fibre. The VOC mass loading on the fibre increases as the wind velocity

5 s sampling with CarboxenTM/PDMS

Ai
4.L-U
4.E-U

x 3.E-03

I :A benzene
I1 a toluene
'
E 2.E-03 + p-xylene

o ethylbenzene i
N

E 1.E-03
0 P f
I I
* r U
P
0.E+00
Fig. 13.22. Effect of wind speed on 0 30 60 90
non-equilibrium extraction of
Air velocity (cm/s)
VOCs (exposed fibre).

415
 Fig. 13.23. Constant agitation device for field air sampling by SPME.

increases from 0 to 5 cm/s. No further change was observed as the wind speed
further was increased from 5 to 20 cm/s. This indicates that the thickness of the
boundary layer between the fibre and air is diminished as the wind speed
increases, which results in an increase of the mass transfer rate and the mass
loading on the fibre. After the thickness of the boundary layer has been
decreased to certain degree, uptake becomes limited by diffusion within the
pores of the polymer, and no further increase in extraction is seen with
increasing wind speed. In practice, when using SPME to sample in the field, it
would be helpful to use a fan to move air samples across the fibre during
sampling, to eliminate possible imprecision in extraction due to variations in
wind speed. Figure 13.23 shows an example of an agitation device for field air
sampling, consisting of a modified hair-dryer fan with a mounting for the SPME
device. Appropriate devices for water sampling can be designed as well.
Control of salt concentration and sample pH can be used to enhance extrac-
tion; the principle is similar to that for solvent extraction procedures. An appro-
priate salt or buffer is added directly to the sample without the need for a specific
instrument. The other parameter that is very important to optimize is extrac-
tion temperature. At elevated temperatures native analytes can effectively dis-
sociate from the matrix and move into the headspace for rapid extraction by the
fibre coatings. However, the coating/sample distribution coefficient also
decreases with an increase of temperature, resulting in a diminution in the equi-
librium amount of analyte extracted, as described previously. To prevent loss of
sensitivity, the coating can be cooled simultaneously with sample heating. This
idea was implemented in the design of the device shown in Fig. 13.24. In this
device, a fused silica tubing is sealed and coated at one end (outer surface of the
capillary). Liquid carbon dioxide is delivered via the inner capillary to the coated
end of the outer capillary resulting in a coating temperature lower than that of
the sample. This "cold finger" effect results in accumulation of the volatilized

416
 inner capillary
C02in X C0o2 out
-outer capillary

- syringe barrel

ferrule
.<~needle

headspaceoating
sample vial
Fig. 13.24. Design of internally cooled SPME device. heater

analytes at the tip of the fibre. There is additional enhancement in the sample
matrix-fibre coating distribution constant associated with the temperature gap
present in the system which can be given by the following equation:

KT = K° T exp[Cp AT +ln T] (13.7)

T,. LR T T
where K T = Cf(Tf)/Cs(T s ) = n,(Tj)Vls/n(Tf)Vf is the distribution constant of the
analyte between cold fibre and hot headspace; and Cp is the constant pressure
heat capacity of the analyte AT = T, - Tf and K0 is the coating/headspace distri-
bution constant of the analyte when both coating and headspace are at tempera-
ture Tf. Quantitative extraction of many analytes, including volatiles, is possible
with this method [52].
Table 13.3 illustrates that complete recoveries are possible with the internal
cooling approach, even for volatile organic compounds such as substituted
benzene, in a range of matrices. This approach can be used effectively to increase
the sensitivity and extraction speed of the SPME methods without changing the
chemical nature of the fibre. One significant drawback of increased fibre
capacity is loss of selectivity, since in this case not only the analytes but also most
of the interferences are extracted exhaustively into the coating.
Water has proven to be a very effective additive to facilitate the release of
analytes from the matrix and is often used to accelerate extraction [52]. It can
also be used in combination with high temperature extractions to remove and
dissolve even very nonpolar analytes such as PAHs (see Chapter 18). This is pos-
sible because the dielectric constant of water decreases rapidly with temperature
increase [53]. This property can also be used in the solid phase microextraction
of solids. Figure 13.25 shows the dynamic hot water extraction system. The solid
sample is placed into the extraction vessel which is in the form of a piece of
thick-walled tubing. This is important to facilitate efficient removal of the
extracted components from the extraction cell [24]. The vessel is then fitted into
a high pressure system consisting of a pump delivering water, a heater to control

417
TABLE 13.3
Quantitative extraction of BTEX from different matrices
Analytes %Extracted
Water (80°C) Soil (15% water) Sand (110°C) Clay (5% water)
(120°C) (170°C)
_

Benzene 50 48 84 64
Toluene 65 101 105 84
Ethylbenzene 85 96 106 91
m,p-Xylene 90 99 102 95
o-Xylene 100 104 100 100
_ , -

extraction temperature, and a restrictor to maintain the extraction pressure.

Water with extracted analytes is collected in a cooled vial supplied with agitation
and an SPME device (see Fig. 13.25). The analytes are extracted into the fibre
coating from the water immediately upon delivery to the vial. Good quantitation
has been obtained for extraction of PAHs from contaminated soil when the col-
lection vial is cooled and the SPME fibre is immersed in water continuously dur-
ing the extraction [54].
Figure 13.26 shows how the extraction system can be made simpler and less
expensive: when a static extraction technique is performed in a high pressure
vessel, the need for a high pressure pump is eliminated. The procedure consists
of adding the sample and water to the vial, inserting the fibre, and sealing it. The
extraction can be performed in either direct, headspace, or membrane protection
mode (see Fig. 13.8). The vial is heated to facilitate the release of analytes to
water. After a cooling period to ensure good partitioning onto the fibre, the

ExtractIon cell r

I i

Fig. 13.25. Hot water dynamic extraction system combined with SPME for analysis of solid
samples.
Fig. 13.26. Static hot water-SPME system.

418
extracted analytes are desorbed into the analytical instrument and typically
quantified, based on an isotopically labelled standard spike (introduced to the
sample prior to extraction).

13.3.5 Derivatization

The main challenge in organic analysis is polar compounds: they are difficult to
extract from environmental and biological matrices and difficult to separate on
the chromatographic column. Derivatization approaches are frequently used to
address this challenge. Figure 13.27 summarizes various derivatization tech-
niques that can be implemented in combination with SPME [55]. Some of the
techniques, such as direct derivatization in the sample matrix, are analogous to
well-established approaches used in solvent extraction. In the direct technique,
the derivatizing agent is first added to the vial containing the sample. The
derivatives are then extracted by SPME and introduced into the analytical
instrument (Fig. 13.28). For example, this approach has been applied to extract
and separate phenols from aqueous samples by first converting the target
analytes to their acetate derivatives [56].
Because of the availability of polar coatings, extraction efficiency for polar
underivatized compounds is frequently sufficient to reach the sensitivity
required. Occasionally, however, there are problems associated with the separa-

< DERIVATIZATION I SPME ;>

Direct Derivatizatin inn Derivatzation in GC

Sample Matrix Injector Port

Derivatization in
SPME Fiber Coating

Simultaneous Derivatization
derivatization following
and extraction extraction

Fig. 13.27. SPME derivatization techniques.

IHbmt> f000[i I; gnu>

Place Add Fiber desorption,

sample derivatizing Extract separation, and
in vial reagent derivatives quantitation

Fig. 13.28. In-sample matrix direct derivatization SPME technique.

419
 (in coating post-extraction derivatization)
4

3
a
2
c

:D 1 palmitic acid stearic acid

,- 0
S 8 palmitic acid ataric acid
F 6 b methyl ester methyl ester
.2)

2 0 n . a~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i
0
10 11 12 13 14 15 16
Time (min.)
Fig. 13.29. Separation of derivatized (b) and underivatized (a) long chain carboxylic acids.

tion of these analytes. Good chromatographic performance and detection can be

facilitated by in-coating derivatization following extraction. This has been
shown with high molecular mass carboxylic acids. After exposing an SPME coat-
ing containing extracted analytes to diazomethane, the resulting ester deriva-
tives can be separated as narrow bands on a GC column (see Fig. 13.29). In
addition, selective derivatization to analogues containing high detector response
groups will result in enhancement in sensitivity and selectivity of detection.
Derivatization in the GC injector is an analogous approach, but it is performed at
high injection port temperatures. For example, long chain carboxylic acids can
be extracted onto the coating as ion pairs when tetramethylammonium hydro-
gen sulfite is added to the sample. During volatilization, analytes are converted
to methyl esters [57].
The most interesting and potentially very useful technique is simultaneous
derivatization and extraction, performed directly in the coating. This approach
allows high efficiencies and can be used in remote field applications. The sim-
plest way to execute the process is to dope the fibre with a derivatization reagent
and subsequently expose it to the sample. Then the analytes are extracted and
simultaneously converted to analogues having high affinity for the coating. This
is no longer an equilibrium process as derivatized analytes are collected in the
coating as long as extraction continues (see Fig. 13.30). This approach, which is
used for low molecular mass carboxylic acids, results in exhaustive extraction of
gaseous samples [58]. When 1-pyrenyldiazomethane is used as the deriva-
tization reagent, it is introduced into the coating by first dissolving the reagent
in a volatile solvent. The fibre is then immersed in the solution. The fibre coating
swells and is doped with the reagent. After evaporation of the solvent, the fibre is
ready to perform extraction. The reagent, having low vapour pressure and high
affinity toward the coating, remains on the fibre during its exposure to the sam-

420
 a 0

IAa r- A l
W DDDM1DPt w
DDmjD

it a
Fg__J
Doping the SPME Placing the doped fiber into Fiber desorption,
fiber with the gaseous phase or headspace above separation, and
derivatizing reagent aqueous phase in reaction vial for quantitation
in-fiber derivatization/SPME

Fig. 13.30. In-coating derivatization technique with fibre doping method.

pie. Volatilities of the pyrenylmethyl esters formed during the reaction also are
low, resulting in the accumulation of the product onto the fibre until analyte or
reagent is exhausted or decomposed. At a high injector temperature the deriva-
tized analytes are removed from the coating and the fibre can be reused.
A similar approach is used for the analysis of formaldehyde from gaseous
samples [59]. This approach is not an equilibrium extraction, but it is based on
the fast reaction. Overall extraction rate is therefore controlled by mass trans-
port, analogous to the non-equilibrium extraction of benzene, toluene and
p-xylene using porous coatings, described earlier (Fig. 13.20). The derivatizing
agent, o-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine is first doped onto the fibre
by room temperature headspace extraction from an aqueous solution. Formalde-
hyde is subsequently extracted from an unknown sample, and converted to the
oxime derivative in the fibre. In this case the kinetics of the derivatizing reaction
are fast, and uptake is controlled by mass transport. If reaction kinetics were
slow, uptake may be controlled by both the distribution constant and the reac-
tion kinetics [51]. With fast derivatization reaction kinetics, the process is simi-
lar to the mass transport controlled extraction described above for VOCs
analysis from air, where uptake rates for short sampling times are proportional
to diffusion coefficients. The significant difference between the two approaches
is that the formaldehyde analysis approach has a much higher capacity.
This simple, but powerful procedure as described above for this analysis, is
limited to low volatility reagents. The approach can be made more general by
chemically attaching the reagent directly to the coating. The chemically bound
product can then be released from the coating either by high temperature in the
injector, light illumination or change of applied potential etc. The feasibility of
this approach was recently demonstrated by synthetizing standards bonded to
silica gel, which were released during heating. This approach allowed solvent-
free calibration of the instrument [60]. The other interesting option is to use an
SPME device similar to the one shown in Fig. 13.24, but with the derivatization
reagent, instead of carbon dioxide, delivered to the coating.
In addition to using a chemical reagent, electrons can be supplied to produce
redox processes in the coating and convert analytes to more favourable deriva-
tives. In this application, the rod as well as the polymeric film must have good

421
 C

4
Fig. 13.31. SPME-electrochemistry minicell. 1, Reference electrode;
2, 100-ml plastic syringe; 3, Teflon capillary; 4, 8-ml Teflon vial; 5,
SPME fibre working electrode (WE); 6, platinum wire counter
electrode (CE).

electrical conductivity. Figure 13.31 shows the schematic of the three-electrode

cell used to deposit mercury species onto a gold-coated metallic fibre [31,34]. A
similar principle has been used to extract amines onto a pencil "lead" electrode
[61]. In addition the SPME approach can be used to study the properties of
electrochemical processes occurring on the electrode surface. The use of conduc-
tive polymers, such as polypyrole will introduce additional selectivity of the
electrochemical processes associated with coating properties.

13.3.6 Interfaces to analytical instrumentation

Because of its solvent-free nature and the small size of the fibre or capillary,
SPME can be interfaced conveniently to analytical instruments of various types.
Only extracted analytes are introduced into the instrument, since the extracting
phase is non-volatile. Thus there is no need for complex injectors designed to
deal with large amounts of solvent vapour, and these components can be
simplified for use with SPME. The sensitivity of determinations using the SPME
technique is very high, facilitating trace analysis. Although in most cases the
entire complement of analytes is not extracted from the sample, all extracted
material is transferred to the analytical instrument, resulting in good perform-
ance. Also, the solvent-free process results in narrow bands reaching the
instrument, giving taller, narrower peaks and better quantification.

SPME-GC interface
The analytical instrument used most frequently with SPME has been the gas
chromatograph. Standard GC injectors, such as split/splitless can be applied to
SPME as long as a narrow insert with an inside diameter close to the outside
diameter of the needle is used. The narrow inserts are required to increase the
linear flow around the fibre, resulting in efficient removal of desorbed analytes.
The split should be turned off during SPME injection. Under these conditions,
the desorption of analytes from the fibre is very rapid, not only because the
coatings are thin but because the high injector temperatures produce a dramatic
decrease in the coating/gas distribution constant and an increase in the diffusion
coefficients. The speed of desorption in many cases is limited by the time

422
carrier -

supply

upply

0 6 12 18
Time (s)

Left: Fig. 13.32. Schematic diagram of the flash SPME injector. 1, Injector body; 2, washer; 3,
septum; 4, nut; 5, needle guide; 6, 0.53 mm i.d. fused silica capillary; 7, nut; 8, ferrule; 9, heater;
10, butt connector; 11, relay; 12, capacitor; 13, switch.
Right: Fig. 13.33. Rapid analysis of BTEX in water by SPME-flash injector-GC. 1, Benzene; 2,
toluene; 3, ethylbenzene; 4, m,p-xylenes; 5, o-xylenes.

required to introduce the fibre into the heated zone.

One way to facilitate sharper injection zones and faster separation times is to
use rapid injection autosampling devices. An alternative solution is to use a
dedicated injector, which should be cold during needle introduction, but which
heats up very rapidly after exposure of the fibre to the carrier gas stream. A
schematic diagram of such an injector is presented in Fig. 13.32. During desorp-
tion, the fibre is located inside the heated part of the fused silica capillary, its end
being close to the bottom of the heated zone. The distance between the fibre and
the capillary wall is approximately 0.15 mm. A close match between inner
diameter of the capillary and the outer diameter of the fibre assures effective
heat transfer from the heater to the fibre, and a high linear flow rate of the
carrier gas along the fibre. The injector is rapidly heated via a capacitive
discharge. Heating rates of 1000C/s have been determined experimentally [62].
The injector has achieved separation of BTEX (benzene, toluene, ethylbenzene,
xylenes) in 9 s (see Fig. 13.33). Separation of the 28 volatile organic compounds
listed in U.S. EPA method 624 has been accomplished in 150 s with repro-
ducibility better than 5% RSD for most analytes [63]. The fibre can be also be
designed to contain the heating element, as shown in Fig. 13.34. In this case, no
injector is necessary. The modified fibre can be introduced directly into the front
of the column, and analytes can be desorbed rapidly by heating with a capacitive
discharge current after the fibre has been exposed from the needle.

423
 Leads connected to capacitive
discharge supply

holder body plunger with grooves

oblong openingmm
in the plunger x,.32 mm I.D
fused silica 0.057 mm O.D.
capillary / heating wire
silicone rubber
seal ,

nut capillary
needle .

Fig. 13.34. Internally heated SPME device.

GC Oven
5

To lon-Trap MS

Fig. 13.35. Direct capacitive discharge desorption system. 1, SPME syringe; 2, electric
connection I; 3, injector body; 4, steel wire; 5, gold coating; 6, electric connection II; 7, transfer
line; 8, capacitor; 9, relay; 10, butt connector.

Flash desorption injectors can alternatively be designed by passing a current

directly through the fibre. This is possible if the rod is made of conductive mate-
rial, as in the case of the electrochemical SPME devices already mentioned; Fig.
13.35 illustrates such an interface [31]. When the electrical connection is made
at the bottom of the interface, the fibre is rapidly heated by the discharging cur-
rent. The other option is to use laser energy to desorb analytes from the surface
of fused silica optical fibre, as discussed previously.

424
Fig. 13.36. High sensitivity of toluene detection by
directly hyphenating SPME device to an ion trap mass o
spectrometer. Time (min)

Flash desorption injectors can be applied to directly interface SPME to a

range of detection devices such as mass spectrometers and atomic emission
devices. The sharp bands obtained during the desorption process result in very
sensitive detection. For example, Fig. 13.36 shows a sharp peak corresponding to
the toluene band desorbed from the SPME fibre and directly detected by the
mass spectrometer. The limit of detection is about two orders of magnitude
lower than is obtainable with conventional GC-MS techniques, because of a
much sharper band. To facilitate proper quantification, the extract needs to be
very clean, which puts an additional demand on the coating selectivity. Some
help in proper quantification can be obtained if the apparatus for tandem mass
spectrometry (MS-MS) is available.
An important improvement of the SPME flash interface would be a cooling
option, which would eliminate the broadening of the injection bands of volatile
analytes because of desorption prior to heating pulse. This would result in even
faster desorption and separation.

SPME-HPLCinterface
Research has also been focused on designing interfaces for liquid phase separa-
tion techniques to address the need for analysis of nonvolatile and thermally
labile analytes. The interface to high performance liquid chromatography
(HPLC) can be a straightforward analogue of the traditional loop injection sys-
tem. A typical SPME-HPLC interface consists of a custom-made desorption
chamber and a six-port injection valve (Fig. 13.37) [64]. The upper part of the
PEEK tubing (c), fitted into a tee-union, is enlarged to fit the needle of the
syringe. The internal tubing of the SPME device, which holds the fibre, can be
sealed by the PEEK tubing and the tee-union tightly enough to withstand sol-
vent pressures as high as 4500 psi. The desorption chamber is placed in the posi-
tion at which the injection loop normally resides on the injection valve. When the
injection valve is in the "load" position, it allows the fibre to be introduced into
the desorption chamber under ambient pressure. It also allows for the introduc-
tion of a desorption solvent if different from the mobile phase. The valve is then
switched to "inject" to transfer the desorbed analytes to the column. A heater

425
 Desorpt
chamber

¢tPcim

cbU-

Frame
Enlarged

Interface
SPME
Device

bore HPLC pmp

UV-VIS detector

Fig. 13.37. SPME-HPLC interface: (a) stainless steel (SS) 1/16 in tee; (b) 1/16 in SS tubing; (c)
1/16 in PEEK tubing (0.02 in i.d.); (d) two-piece, finger-tight PEEK union; (e) PEEK tubing
(0.005 in i.d.) with a one-piece PEEK union.

can be installed in the device to facilitate the desorption process. The interface
performs well, and its desorption volume is similar to the volume of the typical
injection loop. The use of small-volume desorption chambers results in very effi-
cient supercritical fluid chromatography separation (see Fig. 13.38) with very
narrow bore capillary columns [65].

In-tube SPME-HPLC
There is a significant need to automate SPME-HPLC analyses, and this is
addressed conveniently through the development of in-tube SPME using
internally coated tubing [66]. In this method, a section of coated fused silica GC
capillary is placed between the needle of an HPLC autosampler and the injection
valve. The capillary sections selected have coatings similar to common commer-
cially available SPME fibres. Figure 13.15 illustrates the system based on a
modified Spark Holland micro LC autosampler [67]. In this system the analytes
are extracted first into the coating by passing sample through the tubing,
followed by desorption of the compounds using a small volume of solvent or
mobile phase. These approaches, which are suitable for analysis of very small
samples, also offer convenient interfacing to micro HPLC instrumentation.
The theory of extraction and desorption by this method was investigated and
validated with the automated analysis of six phenylurea pesticides. For extrac-
tion, sample is aspirated and dispensed from the capillary a number of times,
preferably until an equilibrium is nearly achieved. It was expected that the

426
 3

Fig. 13.38. Separation of PAHs on a 50 gm i.d. capil-

lary using the SPME device as a sample introduction
technique. 1, Naphthalene; 2, acenaphthene; 3, phen- o 10 20 30
anthrene; 4, fluoranthene; 5, benz[a]anthracene. time (min)

number of aspirate/dispense steps required to reach equilibrium would increases

with the K-values of the analytes. Experimentally, equilibrium extraction was
not obtained for any of the analytes. It was reasoned that during each dispense
step, analyte would at least partially desorb into the mobile phase that follows
the sample in the capillary, thus complicating the extraction. One significant
difference between in-tube SPME and manual fibre-based SPME-HPLC is the
decoupling of desorption and injection possible with the in-tube method. In the
fibre-based method, analytes are desorbed during injection, as the mobile phase
passes over the fibre. With in-tube SPME, analyte are desorbed either by mobile
phase flow or by aspirating a desorption solvent of choice from a second vial, and
then later transferring the solvent with desorbed analytes to the injection loop
for injection onto the column. While fibre-based HPLC has the advantage of
eliminating the need to filter cloudy samples, the method does suffer peak
broadening in applications where analytes are slow to desorb from the fibre to
the mobile phase due to the coating thickness. With in-tube SPME , analytes are
completely desorbed prior to injection, so peak broadening is less of a factor. The
coating thickness is only a fraction of a micron, so desorption is fast. If analytes
are sufficiently solvated by the mobile phase, there is no need to use additional
solvent for desorption. For the analysis of phenylurea pesticides, a 60 cm section
of carbowax GC capillary (0.25 mm i.d., 0.25 gm film thickness) was used for
extraction, with a total of 10 sample aspirate/dispense steps, with subsequent
desorption into 38 pl methanol. An aspirate/dispense rate of 50 to 100 ul/min was
found optimal for extraction and desorption. Below this level, aspiration and
desorption require an inconveniently long time, and above this level, bubbles
formed on the inside of the capillary, reducing extraction/desorption efficiency.
Method precision was found to vary between 1.6 and 5.6% RSD, and linearity of
the method was observed over the range of 10-10,000 gg/l. The limit of detection
achieved was below 5 g/1l.
This technique was applied to the analysis of the thermally labile carbamate
pesticides, using both the FAMOS autosampler from LC Packings and the
Hewlett Packard (HP) 1100 autosampler, in both cases without modification of
the autosampler itself [68]. Extraction, desorption, separation and MS detection
parameters were optimized, including capillary selection, desorption conditions,

427
 gradient composition, and MS fragmentor voltage. Both standard and capillary
LC formats were employed. For capillary LC detection limits were reduced by a
factor of 10. The effect of sample alcohol content was determined, and the
method was applied to the analysis of spiked water (12% ethanol) and wine
samples.
Automated in-tube solid-phase microextraction (SPME) coupled with liquid
chromatography/electrosprayionisation mass spectrometry (LC-ESI-MS) using
the HP1100 system, was evaluated for the analysis of 5-blockers [42], ranitidine
[69] and amphetamines [70] in biological and pharmaceutical samples. Extrac-
tions required 10-15 min per sample. For quantification, LC-MS analyses were
initially performed by liquid injection onto the LC column with analysis in
selected ion mode. Several in-tube SPME parameters were optimized. These
included capillary column stationary phase selection, sample pH, extraction flow
rate and number and volume of aspirate/dispense steps. The compounds
extracted by the capillary column were easily desorbed by either mobile phase
flow or a small quantity of methanol. Carryover was not observed.
A serum sample from a patient administrated propranolol was analyzed
using this method and both propranolol and its metabolites were detected. This
was a significant finding as metabolites are typically more challenging to
analyze, due to their higher polarity. For ranitidine analysis both tablet and
urine samples were analyzed with a detection limit at S/N = 3 of 1.4 ng/ml. and
within-day and between-day variations in ranitidine analysis were 2.5 and 6.2%
(n = 5), respectively. For the analysis of amphetamine, methamphetamine, and
their 3,4-methylenedioxy derivatives in urine samples detection limits (S/N = 3)
of 0.38-0.82 ng/ml were observed.
Mutagenic heterocyclic amines were also analyzed by this technique [71].
The amines extracted by the capillary column were desorbed by aspiration of 30
/b1 of methanol prior to injection, and carryover of heterocyclic amine was not
observed. The detection limits (S/N = 3) were 0.2-3.1 ng ml- l and IQ, MeIQx and
PhIP concentrations in grilled beefsteak were determined.

Other interfaces
Smaller injection volumes for applications of SPME to micro-HPLC and capil-
lary electrophoresis can be accomplished by modifying microinjector designs;
the sliding injector developed for capillary isotachophoresis is one example [72].
The other approach is to design an appropriate sample introduction system
based on guides, to introduce the fibre with the extracted components directly
into the capillary (see Fig. 13.39). Using this method very efficient separation
was achieved (Fig. 13.40) [73].
SPME can be directly combined to optical detection, based on reflectometric
interference spectrometry [74-76]. A light beam passing through an optically
transparent fibre coated with transparent sorbing material interacts with ab-
sorbed substances through internal reflection. Therefore, if any of the extracted
analytes strongly absorb the transmitted light, there is a loss in intensity that

428
 a Capillary electropherograph 3

ler 7

0.001 AU 4
By ~~~~~~~1
l

b leflon rod , Polymer-coated silica fiber 9

(-40 m diameter)
Conical guide tubc A

To detector - f B

Separation capillary 0 5 10 15 20
(75 nmId. x 360pm o.)/ S fiberassembly Time (min)
Buffer reservoir Pt electrode

Left: Fig. 13.39. Schematic of the SPME-CE system (a) and interface (b).

Right: Fig. 13.40. Electropherogram of ten phenolic compounds obtained by SPME-CE using (A)
a 40 gLm o.d. PA-coated silica fibre and (B) a 40 gAm o.d. bare silica fibre. Peak identities: 1:
2,4-dimethylphenol, 2: phenol, 3: 4-chloro-3-methylphenol, 4: pentachlorophenol, 5:
2,4,6-trichlorophenol, 6: 2-methyl-4,6-dinitrophenol, 7: 2,4-dichlorophenol, 8: 2-chlorophenol, 9:
4-nitrophenol, 10: 2-nitrophenol.

can be detected with a simple optical sensor. These devices demonstrate poor
sensitivity primarily because it is difficult to find light wavelengths that are spe-
cifically adsorbed by the analytes and not by the coating or interferences. In an
alternative design, the light can be passed directly through the absorbing poly-
mer which is then cooled to facilitate high sensitivity of determination [77]. Flu-
orescence can be used to detect analytes in the coating. The selectivity of the
extraction process and spectroscopy can be combined with selectivity of the elec-
trochemical process resulting in a spectroelectrochemical sensor [78].

13.4 COMMERCIAL DEVICES

Fibre assemblies: The first commercial version of the laboratory SPME device
was introduced by Supelco in 1993 (see Fig. 13.4). The device is similar in
operation principle to the custom-made device shown in Figs. 13.2 and 13.3a.
Some additional improvements include the adjustment for depth of the fibre
with respect to the end of the needle, which allows control of the exposure depth
in the injector and extraction vessel. The device incorporates such useful
features as colour marking of the fibre assemblies to distinguish among various
coating types. In addition to standard PDMS coatings of various thickness and
polyacrylate (PA), Supelco developed new mixed phases based on solid/liquid
sorption, such as Carbowax-DVB and PDMS-DVB. Supelco also introduced an
HPLC interface (Fig. 13.41) that integrates the original concept with the injec-
tion valve (Fig. 13.37).

429
 SPMEfiberholder

Septumpiercingneedle
le I
Needleguide
Doubletapered
ferrule ression union
Comp

SPMEfiber_

Solvent
desorption
chamber From To
-r interface interface
r-' Solvent - Solvent
from from
To syringe From synnge Vdte
I~Slvn interface
VWaste

Static desorption Vole Sample

(no flow) njection

HPLC HPLcl Mobil phase

column Mobilephase colum frompump
frompump

Fig. 13.41. Schematic diagram of the Supelco SPME-HPLC interface.

Autosamplers for GC: Varian has developed an SPME autosampler based on

their 8000 GC autosampler system, taking advantage of the fact of the SPME
device is analogous to a syringe in its operation and that after desorption the
coating is cleaned and ready for re-use [79]. The major challenge is to incorpo-
rate agitation and temperature control as well as other enhancements, such as
fibre internal cooling or dedicated injectors. One improvement is an SPME sys-
tem that incorporates an agitation mechanism consisting of a small motor and a
cam to vibrate the needle. The fibre in this design works as a stirrer. The vibra-
tion causes the vial to shake and the fibre to move with respect to the solution;
the result is a substantial decrease of equilibration times compared to a static
system. This mode of agitation simplifies fibre handling because it does not
require the introduction of foreign objects into the sample prior to extraction.
CTC Analytics (Switzerland) incorporated SPME sampling on their
CombiPal autosampler. This is a robotic system with a great deal of flexibility
T M

for programming SPME analyses. Samples are loaded onto trays accommodat-
ing five vial sizes, and samples are heated and agitated during extraction, using a
separate sample preparation chamber. To facilitate agitation, the extraction
chamber is rotated at a programmable rotation speed during extraction. Addi-
tional vials/stations are present to accommodate wash solutions, derivatizing
agents, temperature control, derivatization and fibre conditioning, to facilitate
operation of SPME at optimal conditions. While the built-in software can be
used to perform basic SPME analyses, extra programming flexibility is provided
by the Cycle Composern software, available as an accessory. We expect this new
instrument with its greatly enhanced flexibility, will significantly expand the

430
 43 1 6 7 10

Fig. 13.42. Design of dedicated injector. 1, Modified Swagelok

fitting; 2, stainless steel tubing; 3, moulded septum; 4, nut; 5,
needle guide; 6, nut; 7, blind ferrule; 8, stainless steel tubing; 9,
0.53 mm i.d. fused silica capillary, 10, contact.

range of SPME methods amenable to automation. New coatings and devices are
expected to follow, as interest in SPME grows along with the unprecedented
numbers of new applications appearing in the literature.
Field portable SPME-FastGC: Solid phase microextraction coupled to high
speed gas-chromatography is a good combination to perform rapid and cost-
effective investigations in the field, even of complex organic samples. As dis-
cussed above, SPME is particularly suited for fast GC, as it is solvent-free, and
the thin coatings can provide very fast desorption of analytes at high tempera-
tures. Some instrumental modifications were performed recently in order to
achieve successful fast separations [62].
A portable system was optimized for SPME-fast GC field investigations and
was commercialized by SRI Instruments (model 8610C, SRI Instruments,
Torrance, CA). The instrument was tested in combination with a flame ioniza-
tion detector (FID), a photoionization detector (PID), and a dry electrolytic
conductivity detector (DELCD). A dedicated injector, presented in Fig. 13.42,
was mounted on the portable system in order to use SPME for high-speed
separation. The injector guarantees very fast thermal desorption of the analytes
from the SPME fibre [80].
The injector for high-speed GC should produce as narrow an injection band
as possible. Internal volumes of regular injector ports are too large (e.g.,
split/splitless injector), since they have been designed to accommodate large
volumes of gaseous samples or vapours produced by solvent injection. Thermal
focusing for separation improvement is not convenient for fast separations, since
temperature programming is impractical for high-speed GC. Hence, an injector
port with a small internal volume was required for this application. Also, very
fast thermal desorption from the SPME fibre was required to produce a narrow
injection band and achieve effective separation. In the dedicated injector for
SPME-fast GC, the injector port was maintained cold during needle introduction
and was rapidly heated only when the fibre was exposed to the carrier gas
stream. The desorption area of the injector was heated by capacitive discharge
that allowed heating rates as fast as 4000°C/s, and very narrow injection bands
were observed, as required by fast GC.
Fibre conditioners: SRI has also has a dedicated fibre injector available for
their systems, and CTC Analytics has introduced a fibre cleaning station for
their SPME autosampler. New SPME fibres require initial conditioning at
manufacturer-recommended temperatures ranging from 210-320°C, for time
periods ranging from 0.5-4 h. Conditioning is also recommended at the begin-
ning of the work day, and between runs in the case where analyte carry-over is a
possibility. In addition, many SPME-fast GC applications, such as field

431
 iA) C'··li--·· , nh"· .......

LCcolimnll
S slt,
co.,iolr
aid daa anolysis

(B) .. .......... .....

ad a1tllyl
OtsisfO

Pig. 13.43. Agilent LC 1100 with in-tube SPME.

sampling, require additional fibre cleaning in order to reduce desorption time in

the GC injector and eliminate carry-over of non-target analytes [81].
Typically, fibre conditioning is performed by desorption in a temperature-
controlled GC injector for the recommended conditioning time. However, this
procedure reduces available GC time by occupying an injector and in many cases
prevents analytical work from being performed on the instrument. It may also
load a GC column with unwanted products of desorption, which in turn may
require additional column conditioning and column "blank" determinations.
To address the aforementioned concerns, we have studied the process of fibre
conditioning, and designed, built and tested a stand-alone SPME fibre condi-
tioner [82]. This new device allows for fast conditioning of SPME fibres, using
high temperature and gas flow for the desorption and subsequent purging of
fibre contaminants. This device, intended for laboratory and field sampling
applications, was based on a modified commercial syringe cleaner. The perform-
ance of the new fibre conditioner was tested for several types of commercially
available SPME fibres and compared with the traditional, GC-injector fibre
conditioning method. The new device performed equal to, or better than GC
injectors for both new fibre conditioning and the desorption of n-alkanes repre-
senting a wide range of boiling points, in addition to being significantly less
expensive.

432
 Small custom modifications of commercial devices can lead to new possibili-
ties. For example, the replacement of part of the loop in an Agilent 1100 instru-
ment results in automation of the sample preparation and introduction into an
LC instrument (Fig. 13.43). This in-tube SPME system was applied to the analy-
sis of drugs and pesticides as well as their degradation products in real samples
such as waste water, blood plasma and cell culture extracts with minimum sam-
ple processing prior to placement into the autosampler. These ideas are explored
in Chapter 20-29.
The addition of input and output connections to the GC autosampler vial
allows the system to be used to continuously monitor flowing streams, as shown
in Fig. 13.44. The flow-through design facilitates agitation of the sample [83].
Alternatively, when a connection is added directly to the needle of the auto-
sampler syringe, as in Fig. 13.13b, the system can analyze samples present in the
vial without the need to expose the fibre [4]. This modified approach, which
would facilitate simplified automated sampling, can be designed to rely on air
pressure to push the sample through the needle. Then the fibre containing the
extracted analytes in its coating is introduced to the instrument for desorption.
The in-needle concept can be expanded to the trap in-needle system [84].
Example of such design is illustrated in Fig. 13.45a. In this embodiment the side
hole domed tip needle was used for convenience of packing the needles. This
design is able to perform conveniently sampling, sample preparation and sample
introduction to analytical instrument. The device can be used as active sampler,
by drawing the gas or liquid sample through the needle using a gas tight syringe
attached to the needle. This can be performed in exhaustive mode, by careful
packing of the trap and drawing limited amount of the sample to prevent break-
through or it can be more convenient equilibrium or diffusion based as discussed
above. This device can also perform well as passive sampler when the needle is
exposed to the investigated system directly allowing components of the sample

b Holding rod PDMS

membrane

a
Trap Gas tight syringe
Z a
;-Yr~~~~~
T~~1
autosampler carousel
Left: Fig. 13.44. On line monitoring of flowing streams by Varian autosampler using the
modified vial design.
Right: Fig. 13.45. Alternative embodiments of SPME: (a) In-needle trap and (b) membrane
SPME.

433
 ~~
a.~~ _·

Fig. 13.46. Schematic diagram of

SPME/ MALDI-QqTOF system.
1, Laser source; 2, focusing lens;
3, photodiode; 4, fibre holder;
5, SPME/MALDI fibre; 6, QqTOF;
7, computer.
1
to diffuse into the needle. The components of the sample which are captured can
be both chemical compounds as well as particulate matter present in the sample
since the trap can be constructed from either quartz wool to collect particulate
matter and aerosols or can be a sorbent material to extract chemicals from the
sample matrix.
Extraction rate after exposure of the SPME device to the sample is propor-
tional to the contact surface area between the sorbent and the sample (see Eqs.
(9.22) and (9.23) in Chapter 9). Therefore to increase the mass uptake rates and
therefore sensitivities, large surface area sorbent geometries can be used. For
example, the PDMS extraction phase can be in the form of the thin membrane
(see Fig. 13.45b). In that case the high surface area to volume ratio is obtained
resulting in very high accumulation rates [10]. This approach is particularly
beneficial for hydrophobic semivolatile components characterised by very high
distribution constants [85]. To facilitate the convenient introduction to analyti-
cal instrument the membrane can be attached to the holding rod, as illustrated
in Fig. 13.45b, and after extraction the membrane can be rolled around the rod
and introduced to the injection system for introduction of exacted components.
Solid phase microextraction (SPME) was recently coupled to a matrix
assisted laser desorption/ionization (MALDI) for the detection of large bio-
molecules [86]. The tip of an optical fibre was silanized for extraction of analytes
of interest from the sample. The optical fibre thus served as the sample extrac-
tion surface, the support for the sample plus matrix, and the optical pipe to
transfer the laser energy from the laser to the sample. Both an ion mobility
spectrometer and a quadrupole/time-of-flight (QqTOF) mass spectrometer were
used for the detection of the SPME/MALDI signal.
The QqTOF mass spectrum obtained for a solution of 5 pol//Al of each
peptide using system from Fig. 13.46 is shown in Fig. 13.47. Peaks at 726.6 and
1347.7 represent the protonated MH+ ions of the enkephalin and substance P
respectively. For these experiments, the sample tested was a mixture of enk-
ephalin and substance P solution with alpha-cyano-4-hydroxy cinnaminic acid
used as the matrix. Increased selectivity and performance of SPME-MALDI can
be expected by employing the antibody immobilized SPME technique for the
extraction of the complementary antigen, as discussed above. The application of
this technique holds promise especially in biochemical analysis, pharmaceutical
research, clinical diagnostics and screening. The combination of SPME/MALDI
with a QqTOF system offers the simplicity of sample handling with the
specificity and sensitivity of high performance mass spectrometry; it allows con-

434
 726.4
100 .

5 pmol/uL

so

1347.7

" j4
200 400 600 800 1000
. . .
1200 1400 1600 1800 2000 2200 2400
m/z, amu
2600
.I- . .
2800 3000 3200

Fig. 13.47. Mass spectrum of enkephalin (m/e = 726.4) and substance P (m/e = 1347.7) formed by
SPME/MALDI.

venient implementation of the same technology in parallel format. Laser

desorption from the fibre is also a very powerful sample introduction approach
in fast GC [1].

13.5 METHOD DEVELOPMENT AND OPTIMIZATION OF SPME

Solid phase microextraction has been introduced as a modern alternative to

traditional sample preparation technology, and it is able to address many of the
requirements put forward by analytical researchers. This technique eliminates
the use of organic solvents, substantially shortens the time of analysis and
allows convenient automation of the sample preparation step. It can integrate
sampling with sample preparation which makes it suitable for on-site analysis
and process monitoring. The configurations and operation of the SPME devices
are very simple. For example, for the coated fibre implementation of the technol-
ogy, anyone who knows how to use a syringe can operate the SPME device. In the
case of automated in-tube extraction for HPLC, fitting a piece of the GC capillary
to the system and then turning on the autosampler is all that is required to start
its operation. The technology is designed to greatly simplify sample preparation.
This feature, however, creates a false impression that the extraction is a simple,
almost trivial process. This misunderstanding frequently results in disappoint-
ment. It should be emphasised that the fundamental processes involved in solid
phase microextraction are similar to more traditional techniques and therefore

435
challenges to develop successful methods are analogous. The nature of target
analytes and complexity of sample matrix determine the level of difficulties in
accomplishing a successful extraction. The simplicity, speed and convenience of
the extraction devices primarily impacts the costs of practical implementation
and automation of the developed methods.
The potential savings in analysis time, reduced solvent use and apparent
simplicity of SPME techniques will continue to attract interest among analytical
chemists searching for improved analysis methods. So long as analysts have a
sound understanding of the theory and principles behind this technique, good
accuracy and precision will follow. This part of the chapter describes method
development and optimization.

13.5.1 Theoretical aspects of SPME optimization and calibration

13.5.1.1 Thermodynamics
Solid phase microextraction is a multiphase equilibration process. Frequently,
the extraction system is complex, as in a sample consisting of an aqueous phase
with suspended solid particles having various adsorption interactions with
analytes, plus a gaseous headspace. In some cases specific factors have to be
considered, such as analyte losses by biodegradation or adsorption on the walls
of the sampling vessel. In the discussion below we will only consider three
phases: the fibre coating, the gas phase or headspace, and a homogeneous matrix
such as pure water or air. During extraction, analytes migrate between all three
phases until equilibrium is reached.
The mass of an analyte extracted by the liquid polymeric coating is related to
the overall equilibrium of the analyte in the three-phase system. Since the total
mass of an analyte should remain constant during the extraction, we have:

COVs =C Vf +CTvh +Cs

V (13.8)
where C0 is the initial concentration of the analyte in the matrix; Cf, CT, and CT
are the equilibrium concentrations of the analyte in the coating, the headspace,
and the matrix, respectively; Vf, Vh, and V. are the volumes of the coating, the
headspace, and the matrix, respectively. If we define the coating/gas distribution
constant as K, = CT / Ch, and the gas/sample matrix distribution constant as Khs
= CT /C, the mass of the analyte absorbed by the coating, n = Cf Vf, can be
expressed as:

n= KfhKhVCOV (13.9)
K, Khs Vf +Kh, V + Ve

Also, Kf = KKh, = KfgKg (13.10)

since the fibre/headspace distribution constant, K, can be approximated by the

fibre/gas distribution constant Kfg, and the headspace/sample distribution con-

436
stant, Khs, by the gas/sample distribution constant, Kgs, if the effect of moisture
in the gaseous headspace can be neglected. Thus, Eq. (13.2) can be rewritten as:

n Kf VfCOVs (13.11)
Kf V + Ks Vh +Vs

The equation states, as expected from the equilibrium conditions, that the
amount of analyte extracted is independent of the location of the fibre in the
system. It may be placed in the headspace or directly in the sample as long as the
volumes of the fibre coating, headspace, and sample are kept constant. There are
three terms in the denominator of Eq. (13.11) which give measures of the analyte
capacity of each of the three phases: fibre (KfsVf), headspace (KhSVh), and the
sample itself (Vs). Therefore, it is always important to optimize volumes of all the
phases present in the system. If we assume that the vial containing the sample is
completely filled (no headspace), the term KhSVh in the denominator, which is
related to the capacity (CVh) of the headspace, can be eliminated, resulting in
Eq. (13.1).
Equation (13.1) describes the mass absorbed by the polymeric coating after
equilibrium has been reached in the system. In most of determinations, Kfs is
relatively small compared to the phase ratio of sample matrix to coating volume
(Vf < < V). In that situation the capacity of the sample is much larger compared
to capacity of the fibre resulting in a very simple relationship defined by Eq.
(13.2), which states that the amount extracted at equilibrium is proportional
only to the distribution constant, volume of the fibre coating and the concentra-
tion.
Equation (13.2) emphasizes the field sampling capability of the SPME
technique. It is not necessary to perform a separate sampling procedure where a
well defined volume of the matrix is obtained since the amount of analyte
extracted is independent of V, as long as KSVf < Vs. The SPME device can be
placed directly in contact with the investigated system to allow quantitation.
Strictly speaking, the above discussion is limited to partitioning equilibrium
involving liquid polymeric phases such as poly(dimethylsiloxane). The method of
analysis for solid sorbent coatings is analogous for low analyte concentration,
since the total surface area available for adsorption is proportional to the coating
volume if we assume constant porosity of the sorbent. For high analyte concen-
tration the saturation of the surface can occur resulting in nonlinear isotherms.
Similarly high concentration of competitive interference compound can displace
the target analyte from the surface of the sorbent. The simplest way to consider
these high concentration effects is to replace the volume of the fibre coating, Vf in
the above equations as a measure of the total fibre surface area by a fraction of
the original coating volume corresponding to a free surface area available for
adsorption.
The way to overcome these effects is to use diffusion based sampling by
exposing fibre to the sample for a short period of time. The amount of analytes

437
collected on the coating is given by Eq. (13.7), which also does not contain
volume of sample. The extraction mass is dependent only on the diffusion
coefficient of given analyte in the sample matrix and agitation conditions, which
needs to be constant during sampling. This approach is therefore also suitable
for field sampling.

Predictionof distributionconstants
In many cases, the distribution constants present in equations above which
determine the sensitivity of SPME extraction can be estimated from physico-
chemical data and chromatographic parameters. This approach eliminates need
for calibration. For example, distribution constants between a fibre coating and
gaseous matrix (e.g., air) can be estimated using isothermal GC retention times
on a column with stationary phase identical to the fibre coating material. This is
possible because the partitioning process in gas chromatography is analogous to
the partitioning process in solid phase microextraction, and there is a well-
defined relationship between the distribution constant and the retention time.
The nature of the gaseous phase does not affect the distribution constant, unless
the components of the gas, such as moisture, swell the polymer, thus changing
its properties. A most useful method for determining coating-to-gas distribution
constants uses the linear temperature programmed retention index (LTPRI)
system, which indexes compounds' retention times relative to the retention
times of n-alkanes. This system is applicable to retention times for tempera-
ture-programmed gas-liquid chromatography. The logarithm of the coating-to-
air distribution constants of n-alkanes can be expressed as a linear function of
their LTPRI values [87]. The LTPRI values for many compounds are available in
the literature, hence this method allows estimation of Kfg values without experi-
mentation. If the LTPRI value for a compound is not available from published
sources, it can be determined from a GC run. Note that the GC column used to
determine LTPRI should be coated with the same material as the fibre coating.
More discussion on this topic is conducted in Section 13.6 on applications.
Estimation of the coating/water distribution constant can be performed
using Eq. (13.10) [88]. The appropriate coating/gas distribution constant can be
found by applying techniques discussed above, and the gas/water distribution
constant (Henry's constant) can be obtained from physicochemical tables or can
be estimated by the structural unit contribution method [89].
Some correlations can be used to anticipate trends in SPME coating/water
distribution constants for analytes. For example, a number of investigators have
reported the correlation between octanol/water distribution constant Kow and
Kf,. This is expected, since Ko, is a very general measure of the affinity of
compounds to the organic phase. It should be remembered, however, that the
trends are valid only for compounds within homologous series, such as aliphatic
hydrocarbons, aromatic hydrocarbons or phenols; they should not be used to
make comparisons between different classes of compounds, because of different
analyte activity coefficients in the polymer.

438
Effect of extractionparameters
Thermodynamics theory predicts the effects of modifying certain extraction
conditions on partitioning and indicates parameters to control for reprodu-
cibility. The theory can be used to optimize the extraction conditions with a
minimum number of experiments and to correct for variations in extraction
conditions, without the need to repeat calibration tests under the new condi-
tions. For example, SPME analysis of outdoor air may be done at ambient
temperatures that can vary significantly. A relationship that predicts the effect
of temperature on the amount of analyte extracted allows calibration without
the need for extensive experimentation. Extraction conditions that affect K,
include temperature, salting, pH, and organic solvent content in water. A brief
discussion about the use of extraction parameters in SPME method optimization
can be found the next section.

13.5.1.2 Kinetics
The kinetic theory is very useful in optimization of the extraction conditions by
identifying "bottlenecks" of solid phase microextraction and indicates strategies
to increase extraction speed. In the discussion below we will limit our consider-
ation to direct extraction (Fig. 13.48).
Perfect agitation:Let us first consider the case where the liquid or gaseous
sample is perfectly agitated. In other words, the sample phase moves very rap-
idly with respect to the fibre, so that all the analytes present in the sample have
access to the fibre coating. In this case, the equilibration time, defined as the
time required to extract 95% of the equilibrium amount of an analyte from the
sample, corresponds to:

te = t9 % = 2(b -a) (13.12)

Df

Fig. 13.48. Graphic representation of

the SPME/sample system configura-
tion, with dimensions and parame-
ters labelled as follows: a, fibre
coating inner radius; b, fibre coating
outer radius; L, fibre coating length; Liquid
Polymer
d, vial inner radius; Cf, analyte con- C, KK
centration in the fibre coating; Df,
analyte diffusion coefficient in the Aqueous
fibre coating; Cs, analyte concentra- Solution
tion in the sample; D., analyte diffu- C,
sion coefficient in the sample; Kf,,
analyte distribution coefficient be- Vial
tween fibre coating and sample; Kf
= C/C.

439
Fig. 13.49. Mass absorbed versus time from 0 1 2 3 4 5
agitated solution of infinite volume. D t / SKf(b-a)

Using this equation, one can estimate the shortest equilibration time possible
for the practical system by substituting appropriate data for the diffusion
coefficient of an analyte in the coating (Df) and the fibre coating thickness (b - a).
For example, the equilibration time for the extraction of benzene from a
perfectly stirred aqueous solution with a 100 Am PDMS film is expected to be
about 20 s. Equilibration times close to those predicted for perfectly agitated
samples have been obtained experimentally for extraction of analytes from air
samples (because of high diffusion coefficients in gas) or when very high sonica-
tion power was used to facilitate mass transfer in aqueous samples. However, in
practice, there is always a layer of unstirred water around the fibre. A higher
stirring rate will result in a thinner water layer around the fibre.
Practical agitation: Independent of the agitation level, fluid contacting a
fibre's surface is always stationary, and as the distance from the fibre surface
increases, the fluid movement gradually increases until it corresponds to bulk
flow in the sample. To model mass transport, the gradation in fluid motion and
convection of molecules in the space surrounding the fibre surface can be
simplified by a zone of a defined thickness in which no convection occurs, and
perfect agitation in the bulk of the fluid everywhere else. This static layer zone is
called the Prandtl boundary layer and is discussed in detail in Chapter 9. Its
thickness is determined by the agitation conditions and the viscosity of the fluid.
Figure 13.49 shows theoretical equilibration time for this case. The equili-
bration time can be estimated for practical cases from the equation below:

8K
t e = t95 = 3 (b -a) (13.13)
D,

where (b - a) is the fibre coating's thickness, Ds is the analyte's diffusion

coefficient in the sample fluid, Kf is the analyte's distribution constant between
the fibre and the sample. This equation can be used to predict equilibration
times when the extraction rate is controlled by the diffusion in the boundary
layer. In other words, the extraction time calculated using Eq. (13.13) must be
longer than the corresponding time predicted by Eq. (12.12).

440
13.5.2 Method development

Developing a new SPME method in most cases requires the following steps: (1)
Selection of fibre coating. (2) Selection of the derivatization reagent, if required.
(3) Selection of the extraction mode. (4) Selection of the agitation method. (5)
Selection of separation and/or detection technique. (6) Optimization of desorp-
tion conditions. (7) Optimization of sample volume. (8) Determination of the
extraction time profile in a pure matrix. (9) Determination of extraction time.
(10) Calculation of the distribution constant. (11) Optimization of extraction
conditions (pH, salt, temperature). (12) Determination of the linear range for
pure matrix at optimum extraction conditions. (13) Selection of the calibration
method. 14) Optimization of the extraction conditions for heterogeneous sam-
ples. (15) Verification of the equilibration time, sensitivity and linear dynamic
range for complex samples. (16) Determination of method precision. (17) Deter-
mination of method detection limit. (18) Validation of the method. (19) Automa-
tion of the method
In most cases, not all the steps need to be performed. Knowledge gained from
previous experiments, as well as from literature, can often be applied to the
problem at hand. Most SPME methods developed to date are used in combina-
tion with gas chromatographic separation and an appropriate detection method.
Hyphenation with HPLC and other techniques should be also considered. A dis-
cussion of the particular steps in method development follows, with the main
focus on aqueous samples.

13.5.2.1 Selection of fibre coating

The fibre coatings that are commercially available (2001) from Supelco, Inc.
(Bellefonte, PA): poly(dimethylsiloxane) (PDMS) of three thicknesses: 7, 30 and
100 Am; 85 ,.m poly(acrylate) (PA); 65 yAm poly(dimethylsiloxane)/divinylbenz-
ene (PDMS/DVB); 60 Am PDMS/DVB optimized for HPLC; 65 [cm. poly(ethylene
glycol)/divinylbenzene (Carbowax/DVB); 65 m poly(ethylene glycol)/divinyl-
benzene template resin (Carbowax/TR); and 80 m poly(dimethylsiloxane)/
Carboxen as well Carboxen/DVB. The PDMS and the PA coatings, still the most
widely used, extract analytes by absorption. PDMS is a very viscous liquid
(rubber), while PA is a low density solid polymer at room temperature, which
allows analytes to diffuse into the coating, but the diffusion coefficients are
lower than for PDMS. The remaining coatings are porous polymers, in which
DVB or Carboxen porous microspheres are immobilized onto the fibre by using
either Carbowax or PDMS as the glue to hold them together. These coatings
extract analytes by adsorption rather than absorption. The pores in the template
DVB polymer are uniform, resulting in less adsorption discrimination as a
function of analyte molecular weight [90].
Figure 13.50 presents general guidelines for choosing a coating for a given
application. Typically, the chemical nature of the target analyte determines the
type of coating used. A simple general rule, "like dissolves like", applies very well

441
 t 100 prn
jM
30 pm PDMS
lM 7pm J
MEN Poly(acrylate)
* PDMS\DVB
,*Carbowax\DVB (also TR)
= Carboxen

Fig. 13.50. Coating selection guide.

to liquid coatings. Porous polymer coatings are mostly non-selective. Selection of

the coating is based primarily on polarity and volatility characteristics of the
analyte. For example, according to Fig. 13.46, a 100 Aum PDMS coating should be
used for compounds of high to medium volatility, and low to medium polarity. In
general, PDMS is the most useful coating and should be used whenever possible.
It is very rugged and able to withstand high injector temperatures, up to about
300°C. PDMS is a non-polar phase, thus it extracts non-polar analytes very well.
However, it can also be applied successfully to more polar compounds, particu-
larly after optimizing extraction conditions. An additional advantage of this
phase is the possibility of estimating the distribution constants for organic
compounds from retention parameters on PDMS-coated GC columns (see Figs.
13.61 and 13.63).
Both the coating thickness and the distribution constant determine the sen-
sitivity of the method and the extraction time. Thick coatings offer increased
sensitivity, but require much longer equilibration times. As a general rule, to
speed up the sampling process, the thinnest coating offering the sensitivity
required should be used. For 100 /m PDMS fibres, the equilibration time in
direct aqueous extraction with magnetic stirring is less than 1 h for compounds
characterized by distribution constants K lower than - 10,000. Thinner PDMS
coatings should be considered for compounds of larger K. For more polar com-
pounds, such as phenols [91], the poly(acrylate) coating is more suitable.
Porous polymer coatings have complementary properties compared to
PDMS and PA. They are more suitable for volatile species, the respective
distribution ratios typically being larger than for PDMS. Changing the "glue"
from PDMS to Carbowax results in different selectivity towards polar com-
pounds, such as ketones and alcohols. The adsorption times are typically shorter
for gaseous samples compared to 100 lam PDMS, since the analytes do not need
to diffuse through the liquid polymeric phase.
The template resin coating is designed to reduce molecular weight discrimi-
nation for homologues differing in chain length. With regular DVB coatings
characterized by non-uniform pore size distribution, the amount of analyte
extracted from a solution of a given concentration may vary as a function of the

442
size of a molecule. The template resin has uniform pore diameters, therefore it
exhibits uniform extraction efficiency.
When a group of analytes of different characteristics (e.g., pesticides) is to be
determined by SPME, primary consideration should be given to the analytes
that are the most difficult to extract. A coating should be chosen that enables the
determination of all the analytes with enough sensitivity. When none of the
commercially available coatings meets the requirements, custom coatings can
sometimes be prepared. For example, Fig. 13.47 shows the separation of alcohols
extracted from unleaded gasoline using a fibre coated with commercially avail-
TM
able Nafion perfluorinated resin [92]. SPME device also facilitates micro-
solvent extraction. For example, porous coatings can be used to support organic
solvent by placing the fibre into the solvent prior to extraction [93]. Solvent
located in the pores extracts analytes from the matrix, when the fibre is placed in
contact with a sample.

13.5.2.2 Selection of the derivatizationreagent

Derivatization performed before and/or during extraction can enhance sensitiv-
ity and selectivity of both extraction and detection, as well as enable SPME
determination of analytes normally not amenable to analysis by this method,
while post extraction methods can only improve chromatographic behaviour and
detection. Incorporation of the derivatization step complicates the SPME proce-
dure, so should only be considered when necessary. Selective reactions produc-
ing specific analogues result in less interference during quantitation. This
approach can be used for analyte determination in complex matrices. Sensitivity
enhancement can be achieved when the derivatizing reagent contains moieties
that enhance detection. For example, the sensitivity of SPME determination of
carboxylic acids in water can be greatly improved by using 1-(pentafluoro-
phenyl)diazoethane as the derivatizing reagent [94]. This compound converts
carboxylic acids directly in the aqueous matrix to pentafluorophenyl ester deriv-
atives which can be detected by ECD with very good sensitivity. Method selectiv-
ity is also greatly improved in this way.
In many cases, derivatization reagents designed specifically for liquid extrac-
tion procedures can be used in SPME. For example, sodium tetraethylborate was
used to enable the analysis of metal ions, including mercury and lead, in water,
as well as to perform partial speciation of organometallic compounds by SPME/
GC [95,96].
The most interesting implementation of the derivatization procedure
involves the use of doped fibres. Figure 13.51 presents a chromatogram obtained
by exposing a poly(acrylate)-coated fibre doped with 1-pyrenyldiazomethane to
the headspace of a sewage sample [97]. All volatile carboxylic acids are detected.
Without derivatization, SPME is not capable of extracting sufficient amounts of
these acids because no appropriate selective coating exists as of yet. Such an
approach results in high sensitivity and is compatible with field analysis require-
ments, since the derivatizing reagent and the derivatization products are non

443
 3 5
25°ZDy

CH 2 OCOC 2H5

TOT 7In 8 INT8

.t~j X I ------r
I UU I 0U0 150UU
- 0u0 40 120 200 280 360
Retention Time (s) Mass / Charge
1. Acetic; 2. Propionic; 3. Isobutyric; 4. Butyric; 5. Pivalic 6. Isovaleric; 7. Valeric; 8. Hexanoic Acids

Fig. 13.51. (a) Reconstructed GC/MS chromatogram illustrating the presence of short-chain
fatty acids in a real sewage sample (SPME extraction with PA-coated fibre). Peak assignment: 1,
acidic; 2, propionic; 3, isobutyric; 4, butyric; 5, pivalic; 6, isovaleric; 7, valeric; and 8, hexanoic
acids. The peaks shown correspond to pyrenylmethyl esters of these acids. (b) An example of the
mass fragmentogram and the structure of propionic acid/PDAM ester.

A.High Capacity B. Low Capadcity

KhsV >> Kfn Kh.Vh<' K,

Fig. 13.52. SPME extraction form headspace: (A) high

Henry constant compounds (high headspace .
capacity); (B) low Henry constant compounds (low
headspace capacity).

volatile. The doped fibres not only facilitate spot sampling, but can also measure
long-term exposure to a given contaminant, since analytes are continuously
accumulated onto the fibre until all the reagent is consumed.

3.5.2.3 Selection of the extraction mode

Extraction mode selection is based on the sample matrix composition, analyte
volatility, and its affinity to the matrix. For very dirty samples, the headspace or
fibre protection mode should be selected. For clean matrices, both direct and
headspace sampling can be used. The latter is applicable to analytes of medium
to high volatility. Headspace extraction is always preferential for volatile anal-
ytes because the equilibration times are shorter in this mode compared to direct
extraction. Extraction conditions for many compounds, including polar and ionic
ones, can be improved by matrix modifications, as described later. Application of
headspace SPME can be extended to semi-volatile compounds and analytes
strongly bound to the matrix by increasing the extraction temperature, also
discussed below. The need for careful optimisation of headspace procedures is
illustrated in Fig. 13.52. Rapid extraction is obtained when the headspace
contains a high concentration of analytes (Fig. 13.52A). When the concentration
of analytes in headspace is low (Fig. 13.52B), the extraction rate is defined by
mass transfer of analytes from the aqueous matrix. An increase of temperature
(resulting in an increase in Henry constant) and/or agitation rate (decrease of

444
TABLE 13.4
Sampling mode selection criteria
Sampling mode Analyte properties Matrices
Direct medium to low volatility Gaseous
Headspace high to medium volatility liquid (including complex); solid
Membrane protection low volatility complex samples

TABLE 13.5
Agitation methods in SPME
Method Advantages
Static (no agitation) simple, performs well for gaseous
samples
Magnetic stirring common equipment, good
performance
Intrusive stirring very good performance
Vortex/moving vial good performance, no need for a
stirring bar in the vial
Fibre movement good performance, no need for a
stirring bar in the vial
Flow through good agitation at rapid flows
Sonication very short extraction times
Disadvantages
limited to volatile analytes and
headspace SPME
requires stirring bar in the vial
sealing the sample vial difficult
stress on needle and fibre
stress on needle and fibre, limited
to small sample volumes
potential for cross contamination,
requires constant flow
noise, heating of the sample

the boundary layer thickness) results in the shortening of equilibration time.

Fibre protection should be used only for very dirty samples, in cases where
neither of the first two modes can be applied. General sampling mode guidelines
are presented in Table 13.4.

13.5.2.4 Selection of the agitation technique

Equilibration times in gaseous samples are short and frequently limited only by
the rate of diffusion of the analytes in the coating. A similar situation occurs
when analytes characterized by large air/water distribution constants are
determined in water by the headspace technique. When the aqueous and gaseous
phases are at equilibrium prior to the beginning of the sampling process, most of
the analytes are in the headspace (see Fig. 13.52A). As a result, the extraction
times are short even when no agitation is used. However, for aqueous samples,
agitation is required in most cases to facilitate mass transport between the bulk
of the aqueous sample and the fibre. Table 13.5 summarizes the properties of
several agitation methods which have been tested with SPME.

445
 -S

Fig. 13.53. Extraction time profiles

obtained for direct extraction with mag-
netic stirring of 1 ppm benzene solution 0 200 400 600
using 100 gum PDMS coating. Time (s)

Care must be taken when using magnetic stirring to ensure that the rota-
tional speed of the stirring bar is constant and that the base plate does not
change its temperature during stirring. This usually implies the use of high
quality digital stirrers. Alternately, with cheaper stirrers, the base plate should
be thermally insulated from the vial containing the sample to eliminate varia-
tions in sample temperature during extraction. Magnetic stirring is efficient
when fast rotational speeds are applied. Figure 13.53 illustrates the dependence
of the equilibration time for benzene on the rotational speed. The equilibration
time progressively decreases with stirring speed increase, and at 2500 rpm is
only about 200 s. Intrusive stirring can improve the agitation further, but it
requires a direct connection between the stirrer and the motor, which is difficult
to seal.
Needle vibration technique uses an external motor and a cam to generate a
vibrating motion of the fibre and the vial. This technique has been implemented
by Varian in the SPME autosampler. In the Vortex technique the vial is moved
rapidly in a circular motion. Both techniques generally provide good agitation,
resulting in equilibration times similar to those obtained for magnetic stirring.
However, for the needle vibration technique, good performance is limited to
small vials and direct extraction mode. Both techniques can be conveniently
applied to process a large number of samples since the sample vials do not
require any manipulations such as introduction of stirring bars.
Flow through techniques are very useful in continuous monitoring applica-
tions and can also be automated. However, some additional flow metering
devices may be required to ensure reproducible mass transfer conditions. The
most efficient agitation method evaluated to date for SPME applications is the
direct probe sonication, which can provide very short extraction times, fre-
quently approaching the theoretical limits calculated for perfectly agitated sam-
ples. However, this technique has substantial drawbacks associated with large
amounts of energy introduced into the system, which cause the sample to heat
up and in some cases can lead to decomposition of the analyte. Most of these dis-
advantages are eliminated when sonication is applied in the flow-through mode.

446
13.5.2.5 Selection of separationand/or detection technique
So far, most SPME applications have been developed for gas chromatography,
but other separation techniques, including HPLC, capillary electrophoresis (CE)
and supercritical fluid chromatography (SFC), can also be used in conjunction
with this technique. The complexity of the extraction mixture determines the
proper quantitation device. Regular chromatographic and CE detectors can
normally be used for all but the most complex samples, for which mass spectrom-
etry should be applied. As selective coatings become available, the direct
coupling to MS/MS and ICP/MS becomes practical.

13.5.2.6 Optimizationof desorption conditions

Standard gas chromatographic injectors, such as popular split/splitless types,
are equipped with large volume inserts to accommodate the vapours of the
solvent introduced during liquid injections. As a result, the linear flow rates of
the carrier gas in those injectors are very low in splitless mode, and the transfer
of the volatilized analytes onto the front of the GC column is very slow. No
solvent is introduced during SPME injection, therefore the large insert volume is
unnecessary. Opening the split line during SPME injection is not practical, since
it results in reduced sensitivity. Efficient desorption and rapid transfer of the
analytes from the injector to the column require high linear flow rates of the
carrier gas around the coating. This can be accomplished by reducing the
internal diameter of the injector insert matching it as closely as possible to the
outside diameter of the coated fibre. Narrow bore inserts for SPME are commer-
cially available from Supelco for a range of GC instruments.
Other GC injectors, including PTV (programmed temperature vaporization)
and SPI (septum-equipped programmable injector, Varian), can be used with
good results for SPME analysis. The use of on-column injectors is limited to
those equipped with independent heating and capable of accommodating large
diameter columns. Table 13.6 summarizes the characteristics of GC injectors
suitable for SPME analysis.
Desorption time is determined by the temperature of the injector and the
linear flow rate of the carrier gas around the fibre. Theoretical desorption times
are very short since the diffusion coefficients of analytes in the coating increase
and the gas/coating distribution constants rapidly decrease with temperature
increase. Table 13.7 illustrates the decrease in the gas/coating distribution
constant for three analytes when the temperature is increased from ambient to
250°C, which is a typical injector temperature.
In practice, desorption temperature is determined by thermal stability of the
coating. It is advisable to use high desorption temperatures in order to achieve
fast desorption. However, application of excessive temperatures adversely
affects the coating's longevity and results in bleeding of the polymer, rendering
the separation and quantitation difficult.
To achieve good separation efficiency, it is crucial that the injection band
width is as small as possible. The few seconds needed to transfer the analytes

447
TABLE 13.6
GC injectors and their compatibility with SPME
Split/splitless PTV SPI
Can be used for SPME in splitless Can be used for SPME; low Low internal volume; best
mode; low volume insert required. volume insert required. suitable for SPME.

TABLE 13.7
Effect of temperature on gas/coating distribution constant, Kf5
Compound Kfg
25°C 250°C
Benzene 300 0.4
Xylene 3,000 1.4
Undecane 26,000 13

onto the front of the column when narrow bore inserts are used might still be too
long to produce narrow bands for very volatile analytes. It is necessary, there-
fore, to refocus the band after the desorption is finished. A thick film column or
cryofocusing can be used for this purpose. It is always advisable to consider the
application of a short length of deactivated narrow bore fused silica tubing as a
retention gap. This will eliminate the broadening effect created by the tempera-
ture gradient existing in the column close to the injector. Except for the most vol-
atile analytes (gases), the combination of a thick film column and the retention
gap is usually sufficient to refocus the injection band, and cryofocusing does not
have to be applied.
Parameters which control the desorption process in the HPLC interface are
analogous to those in GC applications. In addition to temperature and flow rate,
the composition of the mobile phase also affects the process. In many cases, it is
possible to use the mobile phase without any modifications, as in PAH analysis
[98]. In some instances, addition of an appropriate solvent to the interface will
assist desorption. The linear flow rate of the mobile phase should be maximized
by choosing a small i.d. tubing for the desorption chamber. Temperature also
plays an important role in accelerating the desorption process as illustrated in
Table 13.8. At ambient desorption temperatures, a significant carryover of
Triton X-100 was observed, which was corrected when the temperature was
raised to 135°C [99].

13.5.2.7 Optimization of sample volume

Significance of volume optimisation is illustrated graphically in Fig. 13.54 for K
= 4. When the volume sample is large (Fig. 13.54A) then the amount of analyte
extracted is an insignificant portion of the total amount of analyte in the system.
Therefore the concentration of analyte in the sample remains constant during

448
TABLE 13.8
Effect of desorption chamber temperature on carryover of Triton X-100 using Carbowax/DVB
fibre
No. of units in the ethoxylate chain (n =) %Carryover
21°C 135°C
4 15 0
5 20 2
6 30 2
7 25 3
8 26 1
9 26 2
10 27 2
11 27 3
12 26 2
13 28 3
14 26 0

extraction. This results in optimum sensitivity and better precision since the
variation in sample volume does not affect the amount of analyte extracted. In
the case when the sample volume is low (Fig. 13.54B) then substantial depletion
of the sample occurs during extraction, resulting in loss of sensitivity and
precision. For example, in Fig. 13.54B, only half the amount of analytes is
extracted compared to Fig. 13.54A. Therefore the optimum sample volume of the
sample should be selected based on the estimated distribution constant Kfs. The
distribution constant can be estimated using literature values for the target
analyte or a related compound with the coating selected. For the PDMS coating,
it can also be directly estimated from the chromatographic parameters. In all
other cases, Kfs has to be determined experimentally by equilibrating the sample
with the fibre and determining the amount of analyte extracted by the coating.
Care must be taken to avoid analyte losses via adsorption, evaporation,
microbial degradation, etc., when very long extraction times are required to
reach the equilibrium.

For Example: Kfs = 4

A. V, >> KV, B. s >/KVf

f"= 1.00; 111-=25 =4 =f

f K V, ' = 0 .5
C;, eC o << o
= 0.96 c, C c, C' =0.5 oc,,

Fig. 13.54. Sample volume effect on extraction amount.

449
 The sensitivity of the SPME method is proportional to the number of moles
of the analyte n extracted from the sample and, for direct extraction method, is
given by Eq. (13.1).
As the sample volume Vs increases, so does the amount of analyte extracted
until the volume of the sample becomes significantly larger than the product of
the distribution constant and volume of the coating (fibre capacity K 6 < < Vs). At
this point, the sensitivity of the method does not increase with further increase
in volume. In practice, the limiting volume can be calculated from the following
dependence:

v = 100KV
E
(13.14)

where E is the magnitude of the relative measurement error (%). For example,
for a 5% error, it can be estimated that Vs = 20KfsVf. This means that for Kf,
values of up to about 200 and a 100 m coating, a 2 ml vial is sufficient to give
maximum sensitivity, while a 40 ml vial can be used for Kfs values smaller than
4000, etc. Using sample volumes larger than the limiting value not only maxim-
izes the sensitivity, but also results in better precision since variations of the
sample volume do not affect the results. Figure 13.55 illustrates this relationship

a 20

100 100 pm coating

XO

K I0
Q000

40

0 10 20 30 40
Simple volme [tL
20
0 coal&K
ting0

80
m K=I 5000

o ltuame
Salpie [r ml

0 10 20 30 40 50

Fig. 13.55. Dependence of n/n o on V. for (a) 100 Am fibre (V, = 0.65 l), (b) 30 xm fibre (Vg 0.14
/l)
,u1) for direct SPME from small volume samples (two-phase system).

450
 4 mL vial, 100 pm coating 40 mL vial, 100 pm coating
101
1060 ' _ --- 160 _ ...-

20 20
0 2 C 3
0 1 2 3 4 0 10 20 30 40
Vhs Vhs
15 mL vial, 100 pm coating
100 ) . _

---
--·e-e
80
·-,
60
-x
`·- --- chloroform
40
- 1,1,1-trichloroethane
20
- carbon tetrachloride
fl
0 2 4 6 8 10 12 14
Vhs

Fig. 13.56. The effect of headspace volume on the amount of analyte extracted in a system of a
constant volume of (a) 4, (b) 15, and (c) 40 ml by a 100 um PDMS fibre for chloroform,
1,1,1-trichloroethane and carbon tetrachloride.

graphically; no corresponds to the amount of analyte extracted from an infinite

sample volume. Even for a moderate distribution constant of 1000 characteristic
for many volatiles, for the 100 m PDMS coating (see Fig. 13.55a) there is a
substantial difference in the amount extracted when varying the sample volume
within a few ml range. Increasing the volume from 1-10 ml results in sensitivity
improvement by over 50%. The change is more dramatic for semivolatile
analytes. The variation is less significant for thinner coatings (Fig. 13.55b), but
the sample volume should still be considered when analytes characterized by
large distribution constants, such as PAHs, are analyzed. In practice, the sample
size is determined by the available vials. The discussion above should assist in
the choice of the most appropriate vial size.
In headspace SPME, the analytes partition to the gaseous phase as well as to
the coating. Very volatile compounds preferentially accumulate in the headspace
resulting in a very substantial loss in sensitivity when the headspace volume is
large. Figure 13.56 shows the dependence of the amount of analyte extracted by
a 100 urm fibre on the headspace-to-sample volume ratio for three vial sizes: 4, 15
and 40 ml. The following values of distribution constants have been used: K0s =
400, 2500 and 1000, and Khs = 0.15, 0.7 and 1.24, corresponding to chloroform,
1,1,1-trichloroethane and carbon tetrachloride, respectively. In this figure, n is
the amount of analyte extracted from an infinite sample volume. The n/no ratio is
always the largest for chloroform, which has the lowest Khs value, even though
its K f0 value is also the lowest. This is due to the fact that only a small fraction of
the analyte is present in sample headspace, therefore equilibrium concentration

451
of the analyte in the sample remains relatively high. For the two other curves
and low headspace volumes, n/no is higher for carbon tetrachloride than for
1,1,1-trichloroethane, which is consistent with the values of their distribution
constants Kf,. However, as carbon tetrachloride has the largest Kh,, n/no drops
faster for this compound than for 1,1,1-trichloroethane, as a result of which the
respective two lines cross each other. The biggest relative increase in n/n, when
moving from small to large vials (-44%) is observed for 1,1,1-trichloroethane,
characterized by the highest Kf, value. For compounds with lower Kf values the
relative increase in the amount extracted is much lower (-16% for chloroform
and - 12% carbon tetrachloride). It is therefore the combination of Kf, and Kh, for
a given compound that determines the magnitude of the effect of sample volume
on the amount of analyte extracted by the fibre in three-phase system involving
headspace [100].
Headspace volume can have a significant effect on equilibration times
(extraction kinetics). If the amount of the analyte extracted by the fibre at equi-
librium is negligibly small compared to the amount present in the headspace
equilibrated with the sample prior to the extraction, the analyte is extracted
almost exclusively from the gaseous phase, and the process is much faster com-
pared to the case when significant amounts of the analyte have to transported
from the liquid sample to the fibre via the headspace. Assuming that this situa-
tion occurs when 95% of the analyte is extracted exclusively from the headspace,
the criterion that must be fulfilled can be described by Eq. (13.15). The assump-
tion is reasonable, as a 5% difference usually falls within the limits of experimen-
tal error for trace SPME/GC analysis:

Kffi Vf (13.15)
KhVh

The above criterion means that the capacity of the headspace (KhVh,,) needs to be
at least 20 times larger than the capacity of the fibre (KfV) to achieve rapid
extraction. For a given sample volume V,, this can be achieved by using a large
enough headspace volume, or by increasing Khs. The latter can be accomplished
by increasing the temperature or by salting the analyte out of the liquid phase.
When the criterion (13.15) is fulfilled, equilibration can take as little as a few
minutes, and is almost independent of the agitation conditions (provided the
analyte is equilibrated between the liquid phase and its headspace before the
extraction begins). It should be emphasized, however, that the increase in
headspace capacity causes a significant loss in sensitivity. In any case, it should
be remembered that changing the vial size during the optimization process can
not only affect the sensitivity, but also the equilibration time if the headspace/
sample volume ratio is different in the new vial.

13.5.2.8 Determination of the extraction time profile in a pure matrix

Extraction time profiles for the analyte should be determined first in a pure
matrix, such as dry air, pure water or sand, to obtain basic understanding of the

452
 A p,p'- DDT 13 Dichlorvos
GC Response(area counts) GC IResponse (areacounts)
100,000 2,5C
80,000 2,0c0 ii

60000 15Eo O %
rel.error
40000 ; : 7% rel. error 1,0

20,000 5 10 5/i m ro
20% rel error
o 0
0 50 100 150 200 250 300 350 i
0 50 100 150 200 250 300 350
Timeminutes) Time(minutes)
PDMS fiber, 100 pm
Fig. 13.57. Selection of the extraction time based on extraction time profiles.

kinetics of analyte transfer from the matrix to the fibre under given mass trans-
fer conditions. The profile is determined by extracting samples of identical com-
position for progressively longer periods of time and plotting the resultant area
counts vs. time. The mass of the analyte extracted can be quantified by a proper
calibration procedure, e.g. syringe injection of a standard solution of the analyte
in an appropriate solvent. In gas chromatography, this type of calibration is reli-
able only when non-vaporizing injectors are used (PTV, SPI, heated on-column).

13.5.2.9 Determinationof the extraction time

An optimal approach to SPME analysis is to allow the analyte to reach
equilibrium between the sample and the fibre coating. The equilibration time is
defined as the time after which the amount of analyte extracted remains
constant and corresponds within the limits of experimental error to the amount
extracted after infinite time. Care should be taken when determining the
equilibration times, since in some cases a substantial reduction of the slope of
the curve might be wrongly taken as the point at which equilibrium is reached.
Such a phenomenon often occurs in headspace SPME determinations of aqueous
samples, where a rapid rise of the equilibration curve corresponding to extrac-
tion from the gaseous phase only is followed by a very slow increase related to
analyte transfer from water through the headspace to the fibre.
When equilibration times are excessively long, shorter extraction times can
be used. However, in such cases the extraction time has to be strictly controlled
to assure good precision. Figure 13.57 presents equilibration time profiles for
p,p'-DDT (a) and Dichlorvos (b). The equilibration time for DDT is about two
hours. At equilibrium, small variations in the extraction time do not affect the
amount of the analyte extracted by the fibre. On the other hand, at the steep part
of the curve, even small variations in the extraction time may result in
significant variations of the amount extracted. The relative error is larger, the
shorter is the extraction time. For a 10 min extraction (Fig. 13.57a), the relative
error caused by varying the extraction time by 1 min is as high as 20%. It drops
to 5% after 50 min extraction since more analyte accumulates onto the fibre. The
timing is even more critical for short exposure times when the equilibration

453
TABLE 13.9
Ranges of K values obtained when the amount of the analyte extracted by the fibre n determined
experimentally falls within 5% (relative) of the true value, for two different fibre coating
thickness and two sample volumes
2 ml sample 35 ml sample
K Range of K K Range of K
100 um
100 105-95 100 105-95
1,000 1067-935 1,000 1051-949
10,000 12537-8172 10,000 10598-9413
100,000 -36190 100,000 115748-86928
1,000,000 -55072 1,000,000 4700000-492593
10,000,000 -58104 10,000,000 -923611
100,000,000 -584 26 100,000,000 o-1012177

7 gm
100 105-95 100 105-95
1,000 1051-949 1,000 1050-950
10,000 10574-9434 10,000 10504-9496
100,000 112903-88785 100,000 105422-94622
1,000,000 3500000-558824 1,000,000 1093750-913462
10,000,000 o-1187500 10,000,000 17500000-6785714
100,000,000 -1338028 100,000,000 -19000000

curve rises rapidly. Figure 13.57b shows an example when a 1 min deviation
from 10 min sampling time results in 50% relative error of determination.
Autosamplers can measure the time very precisely and the precision of
analyte determination can be very good, even when equilibrium is not reached in
the system. However, this requires the mass transfer conditions and the temper-
ature to remain constant during all experiments.

13.5.2.10 Calculationof the distributionconstant

The target analyte's distribution constant defines the sensitivity of the method.
It is not necessary to calculate the fibre coating/sample matrix distribution
constant, Kf, when the calibration is based on isotopically labelled standards or
standard addition, or when identical matrix and headspace volumes are used for
the standard and for the sample with external calibration. However, it is always
advisable to determine Kfs since this value gives more information about the
experiment and aids optimization. K, can be used for calculation of the head-
space volume, sample volume, and coating thickness required to reach the
desired sensitivity.
The distribution constant for direct extraction mode can be calculated from
the following dependence obtained from Eq. (13.1):

454
Kh nV (13.16)
Vf(CoV -n)

It is important to consider the sample volume when calculating the value of Kf.
Table 13.9 presents the ranges of K values that are obtained when the amount of
the analyte extracted by the fibre n determined experimentally falls within + 5%
(relative) of the true value, for two sample volumes (2 and 35 ml) and two fibre
coating thicknesses (100 and 7 m). A +5% error can be assumed typical for
trace analysis by SPME/GC. To correctly estimate the distribution constants
over 1000 with 100 gm thick PDMS fibre, volumes of the sample larger than 2 ml
need to be used. For very large Kfs values (>1,000,000), typical for high
molecular weight hydrocarbons in water or air, no reliable estimates ofKf, can be
obtained even with 35 ml sample when the 100 gm fibre is used, therefore only
the 7 m fibre should be used for this purpose.
If the distribution constant is low and the total volume of the sample is high,
a good estimate of Kfs can be obtained from the following relationship:
n
Kf = n (13.17)
C0Vf
However, extreme caution is advised when using this equation, since it assumes
that after the extraction, the analyte concentration in the sample does not
change significantly. If this equation is used by mistake when the Kfs value is
large or the sample volume is too small, the result of the calculation does not cor-
respond to Kf,, but rather to the sample/coating phase volume ratio (V /Vf).
To determine the coating/matrix distribution constant when headspace is
present in the vial, knowledge of the matrix/headspace distribution constant is
required:

K n(KhSVh +V) (13.18)

Vf (C V - n)
Kfs can be obtained for aqueous samples from tables of Henry constants (KH):

Ks = KHRT (13.19)
RT
For analytes with an unknown Henry constant, the following equation can be
used:
Kfs = nKf (13.20)
Kf VfVsCo -nKfh V -nVh
Equation (13.20) is obtained by substituting Kh, = KfS /K into Eq. (13.18) and
rearranging it. The fibre/gas distribution constant can be found by extracting
target compounds from air mixtures by using a simple bulb experiment [101]. In
addition, Kfg can be obtained directly from a chromatographic run when the sta-
tionary phase is made of the same material as the fibre coating. For compounds

455
 I I

a)
* pH 1035
* p 11
C0 2 H 13|

Amphetamine Methamphetamine

Fig. 13.58. Effect of pH on amphetamine and methamphetamine extraction.

which have low vapour pressures, the amount of analyte present in the head-
space is very small, especially if the volume of the headspace is kept to a mini-
mum. Therefore, in this situation, the headspace volume can be neglected
altogether and Eq. (13.1), which corresponds to direct extraction, can be applied.

13.5.2.11 Optimization of extraction conditions

Extraction temperature increase causes an increase in the extraction rate, but
simultaneously a decrease in the distribution constant. In general, if the extrac-
tion rate is of major concern, the highest temperature which still provides
satisfactory sensitivity should be used. Figure 13.10 demonstrates the effect of
temperature on extraction of methamphetamine from water. At room tempera-
ture equilibration takes several hours. When the temperature is raised to 60°C,
the equilibration time is less than 20 min for both compounds. When the
temperature is further increased to 73°C, the equilibration time is reduced to a
few minutes (Fig. 13.10). Very short equilibration times indicate that the
majority of the analytes extracted by the fibre originate from the headspace. At
this temperature, the amount of analytes extracted is almost independent of the
agitation conditions. The amount extracted is lower by over an order of magni-
tude compared to room temperature, but still provides sufficient sensitivity to
quantify these drugs at the ppb level, which is adequate for screening
applications.
Adjustment of the pH of the sample can improve the sensitivity of the
method for basic and acidic analytes. This is related to the fact that unless
special coatings are used, SPME can extract only neutral (non-ionic) species
from water. By properly adjusting the pH, weak acids and bases can be converted
to their neutral forms, in which they can be extracted by the SPME fibre. To
make sure that at least 99% of the acidic compound is in the neutral form, the pH
should be at least two units lower than the pKa of the analyte. For the basic
analytes, the pH must be larger than pKb by two units. Figure 13.58 illustrates
the increase in the amount of amphetamine and methamphetamine extracted as

456
 5000000

4000000

0)
U 3t)0000
IE 15o salt
O iti0 5 ~1
al 2000000

1000000

0
Amphetamine Methamphetanile

Fig. 13.59. Effect of salt concentration on amphetamine and methamphetamine extraction, pH =

11.7.

the pH is increased [102]. In practice, it may be difficult to use pH adjustment

with direct extraction, since the coating might be damaged by solutions of very
high or very low pH values. Headspace SPME is therefore a natural choice for
pH-adjusted matrices. The sample should be buffered to ensure good repro-
ducibility when basic or acidic compounds are present in it.
Addition of salt to aqueous samples generally increases the fibre/matrix dis-
tribution constants of neutral organic molecules. However, when the analytes
are in the dissociated form, a decrease in the amount extracted is observed. This
is related to the fact that the activity coefficients of ionic species in water
decrease with the increase in the ionic strength of the solution. It is important,
therefore, to first convert the analytes to neutral forms. Figure 13.59 illustrates
the increase in the amount of amphetamines extracted from a solution when salt
is added after the pH is lowered by KOH addition.

13.5.2.12 Determinationof the linear dynamic range of the method for a pure
matrix at optimum extraction conditions
Modification of the extraction conditions affects both the sensitivity and the
equilibration time; therefore, it is advisable to check the previously determined
extraction time before proceeding to the determination of the linear dynamic
range. This step is required if substantial changes of the sensitivity occur during
the optimization process.
SPME coatings include polymeric liquids such as PDMS, which by definition
have a very broad linear range. For solid sorbents, such as Carbowax/DVB or
PDMS/DVB, the linear range is narrower because of a limited number of
sorption sites on the surface, but it still can span over several orders of magni-
tude for typical analytes in pure matrices. In some rare cases when the analyte
has extremely high affinity towards the surface (e.g., basic proteins adsorption
on poly(acrylic acid) [103] or bioaffinity coatings), saturation can occur at low
analyte concentrations. In such cases, the linear range can be expanded by
shortening the extraction time. In the majority of practical applications

457
developed to date, it is the dynamic range of the detector that limits the linear
response of SPME methods.

13.5.2.13 Selection of the calibrationmethod

Standard calibration procedures can be used with SPME. Fibre blank should be
first checked to ensure that neither the fibre nor the instrument cause interfer-
ences with the determination. The fibre should be conditioned prior to the first
use by desorption in a GC injector or a specially designed conditioning device.
This process ensures that the fibre coating itself does not introduce interfer-
ences. Fibre conditioning may have to be repeated after analysis of samples
containing significant amounts of high-molecular weight compounds, since such
compounds may require longer desorption times than the analytes of interest.
When the matrix is simple (e.g., air or groundwater), the distribution con-
stants are very similar to those for pure matrix. It has been shown, for example,
that typical moisture levels in ambient air, as well as the presence of salt and/or
alcohol in water in concentrations lower than 1%, usually do not change the K
values beyond the 5% RSD typical for SPME determinations. In many such
instances, calibration might not be necessary since the appropriate distribution
constants which define the external calibration curve are available in the
literature or can be calculated from chromatographic retention parameters.
External calibration can also be used successfully when the matrix is more
complex, but well defined (e.g., process streams of relatively constant composi-
tion). Of course calibration standards have to be prepared in such cases in the
matrix of the same composition rather than in the pure medium (e.g., water).
A special calibration procedure, such as isotopic dilution or standard addi-
tion, should be used for more complex samples. In these methods, it is assumed
that the target analytes behave similarly to spikes during the extraction. This is
usually a valid assumption when analyzing homogeneous samples. However, it
might not be true when heterogeneous samples are analyzed, unless the native
analytes are completely released from the matrix under the conditions applied.
Moreover, whenever any of these methods is used, an inherent assumption is
made that the response is linear in the concentration range between the original
analyte concentration and the spiked concentration. While this is usually true
for fibres extracting the analytes by absorption (PDMS, PA) and detectors with
wide linear range (FID), problems may arise when porous polymer fibres
(PDMS/DVB, Carbowax/DVB) are used, or when the detector applied has a
narrow linear dynamic range. It is important, therefore, to check the linearity of
the response using standard solutions before applying standard addition or
isotopic dilution for calibration. To improve the accuracy and precision, multi-
point standard addition should be used whenever practical.

13.5.2.14 Optimization of extraction conditions for heterogeneous samples

Optimizing the process that releases native analytes from the matrix compo-
nents (e.g., solid particles) is more difficult since the analyte/matrix association

458
TABLE 13.10
Quantitation of PAHs from urban air particulate (NIST 1649) by static high temperature water
extraction
Compound Cert conc. Lg/g Estimated concentration as %of certified concentration
(%RSD) (%RSD)
100 mg/270°C 50 mg/270°C 50 mg/300°C
Fluoranthene 7.1 (7) 137 (6) 149 (8) 128 (2)
Pyrene 7.2 (7) 113 (9) 121 (7) 105 (7)
Benzo(a)pyrene 2.9 (17) 49 (14) 72 (10) 70 (4)

is usually poorly understood. Whenever a new type of complex matrix is consid-

ered, a study must be conducted to find the optimum extraction conditions
which give the fastest and most complete release and extraction of native
analytes. Typically, an empirical approach is taken and several parameters are
varied, such as temperature and type of additives. Change of extracting phase
type, used in optimization of many solvent extraction methods, is of limited
value in SPME since the fibre coating does not interact directly with solids
present in the matrix. Heating can release analytes from a solid matrix, but
sometimes heating is not adequate because analytes are bound too strongly or
the matrix or analytes are unstable at higher temperatures. Additives (matrix
modifiers) which enhance the extraction of analytes are an alternative to heat-
ing. The additive selected should enhance the release of analytes from the matrix
but not interfere with the partitioning of analytes into the fibre coating or with
the analysis of the extracted compounds. Successful modifiers for SPME include
nonvolatile acids, salts, and water [104]. Better understanding of the analyte-
matrix interaction would allow a more rational choice of appropriate additives.
In-depth investigation of sample matrices should be considered.
Frequently, conditions optimal for releasing the analytes from the matrix
are different from conditions that are optimal for partitioning them into the
coating. In most cases, high temperature is used to remove native analytes from
the matrix. Higher temperatures, however, reduce the distribution constant and
thus the sensitivity of the method. To eliminate this disadvantage, the cooled
fibre approach can be applied to condense analytes onto the fibre as was dis-
cussed above (see Fig. 13.24).
Optimization of extraction parameters is considered complete when the
recoveries of analytes present in native samples match those for spikes. This
verification is performed based on extraction results obtained for certified
standard reference materials, or by a comparison with standard extraction
methods. Once the method is verified, quantitation is performed by comparing
extracted amounts of native and spiked analytes. Table 13.10 illustrates the
comparison of results of PAH analysis by SPME with certified reference values
obtained for Urban Air Particulate (NIST 1649) [105]. The method consists of

459
static hot water extraction at 250°C, followed by a cool-down period to increase
the amount of analytes absorbed by the PDMS coating.

13.5.2.15 Verification of the equilibrationtime, sensitivity and linear dynamic

range for real sample matrices
Sample matrix can affect not only the distribution constant, but also the
equilibration time. When the matrix is heterogeneous (e.g., an aqueous sample
with solid particulate matter), the kinetics of analyte release might determine
the overall extraction rate. To operate at optimum sensitivity, extraction time
needs to be adjusted accordingly.
The amount of analyte extracted by the fibre changes substantially when the
coating is swollen due to absorption of a large amount of an interfering major
matrix component. To produce swelling, the amount of interference which needs
to be absorbed must be substantial compared to the mass of the coating (more
than 1%), which translates to about 10 micrograms for a 100 1Am PDMS fibre. In
trace analysis of homogeneous aqueous samples this would rarely be the case,
since it requires that the interfering compound has good solubility in both the
aqueous matrix and the coating. However, the phenomenon might occur if the
interference exists as a separate phase, such as an oily suspension or dispersed
hydrophobic humic material [106]. In such cases, SPME is of limited value as a
quantitation tool, since both the distribution constant and the volume of the
coating can change during extraction. The effect of high molecular weight
compounds on the results of SPME analysis of aqueous samples can be
eliminated by using microporous hollow fibre membranes to isolate the coating
from direct contact with the sample.
Apart from the extraction time, the linear dynamic range and method
sensitivity should also be examined for real matrices. Typically, no differences
between model solutions and real samples occur when fibres extracting the
analytes by absorption (PDMS, PA) are used. However, when porous polymer
fibres are used, competitive adsorption of interfering matrix components with
large distribution ratios can reduce the uptake of analytes of interest and cause
non-linearity of the calibration curves. The adverse effect of competitive
sorption can be minimized by extracting the analytes under non-equilibrium
conditions. For example, short extraction can be performed with a well agitated
sample, or static extraction can be used. In both cases, the uptake of volatile
analytes with lower distribution ratios is only slightly affected, while that of less
volatile compounds with large distribution ratios is significantly reduced.
Sample matrix can affect not only the extraction process, but also the
separation. It should be verified that no matrix components co-elute with the
analytes of interest, interfering with proper quantitation.

13.5.2.16 Determinationof method precision

The most important factors affecting precision in SPME are: (1) agitation condi-
tions; (2) sampling time (if non-equilibrium conditions are used); (3) tempera-

460
ture; (4) sample volume; (5) headspace volume; (6) vial shape; (7) condition of
the fibre coating (cracks, adsorption of high M.W. species); (8) geometry of the
fibre (thickness and length of the coating); (9) sample matrix components (salt,
organic material, humidity, etc.); (10) time between extraction and analysis; (11)
analyte losses (adsorption on the walls, permeation through Teflon, absorption
by septa); (12) geometry of the injector; (13) fibre positioning during injection;
(14) condition of the injector (pieces of septa); (15) stability of the detector
response; (16) moisture in the needle. The majority of these factors and their
effect on method precision have already been discussed. A brief explanation of
the remaining parameters follows.
The reproducibility of the sample and headspace volumes can impact the
precision of SPME methods. The volume of the sample has an impact on the
amount extracted when only a few ml sample is used (see the discussion above).
Therefore, small samples such as 1 ml need to be measured very carefully. For
large samples, small variations in the sample volume do not have a substantial
effect on the precision of the data. Obviously, the volume of all the samples and
standards must be the same to obtain valid results.
Headspace volume can be an important factor determining precision of the
results in three-phase systems. It is relatively easy to measure sample volume
accurately. However, vials are not manufactured to have exactly the same total
volume. Wall thicknesses and bottom shapes may differ from vial to vial. Also,
the shape of the septum in a closed vial can vary from concave to convex. All
these factors will affect the total volume of the system, and therefore the
headspace volume, which is the difference between the total volume and the
sample volume. Such differences (especially related to septum shape) are usually
more pronounced in small vials, therefore worse precision can usually be expect-
ed when they are used. In addition to keeping the sample and standard volumes
constant throughout all the experiments, vials of the same volume should be
used for all samples and standards, so that the headspace volume is constant
from vial to vial. This becomes a bigger challenge for small vials, when a small
absolute variation in headspace volume could translate to a large relative
difference.
The condition of the fibre is very important: the presence of high molecular
weight species, such as proteins and humic matter, adsorbed on the surface of
the fibre, causes a change in sorption characteristics. In some cases, the sub-
stances adsorbed can be removed by washing the fibre in appropriate solvents.
When the coating is badly cracked, a small amount of the matrix itself kept in the
cracks by capillary forces might be transferred into the analytical instrument
and adversely affect its operation.
The volume of the coating determines the amount of analytes extracted.
Fibres which begin to lose the coating phase should be replaced immediately.
The length of the fibre should be monitored to ensure equivalency when
changing the fibre. If the length of the fibre (thus the volume of the coating) is
different, an appropriate correction factor should be incorporated.

461
 Time between preparation of standards or sample collection and extraction
should be minimized. After placing aqueous samples or standards in a vial,
analytes begin to interact with the walls of the container. Very hydrophobic sub-
stances are frequently strongly adsorbed on the surfaces and are not available
for extraction by the coating. This results in lower than expected extracted
amounts. One way to prevent this problem is to use vials with more inert walls,
e.g., silanized glass or Teflon. For volatile analytes, it is crucial that new Tef-
lon-lined septa are always used. Once the Teflon lining is pierced, the analytes
can come into direct contact with the septum material (typically polydimethyl-
siloxane), which usually results in losses through absorption. Finally, it should
be remembered that many volatile compounds can permeate through Teflon at a
relatively high rate. This phenomenon can also contribute to analyte losses
when the samples or standards are stored for prolonged periods of time.
The time between extraction and analysis should also be minimized. When
this time is longer that a few seconds, the needle of the fibre assembly should be
sealed with a septum and preferentially cooled down to reduce losses and
interferences. After the fibre is withdrawn from the sample, it begins to equili-
brate with ambient air. This process is slow when the fibre is withdrawn into the
needle. Therefore, the few seconds that are normally required to transfer the
fibre to the injection port of a gas chromatograph do not result in any substantial
losses of the analytes. However, when the fibre is to be stored for some time
before the analysis, sealing of the needle is necessary. In addition, this protection
prevents contamination of the fibre with interferences or even target analytes
which might be present in ambient air. Cooling of the SPME device might be
necessary to extend the stability of the volatile analytes in the coating [11].
13.5.2.17 Determinationof method detection limit
There are several methods described in the literature to determine the method
detection limit. The most widely accepted definition is based on estimating the
detection limit using low concentration spikes and calculating the standard
deviation of the determination. The detection limit is then defined as three times
the standard deviation obtained for the measurement value not higher than 10
times the method detection limit.

TABLE 13.11
Regression line parameters for SPME vs. purge and trap for clean water analysis
Compound Regression parameters-purge and trap on x-axis
Slope y-Intercept Correlation
Benzene 1.07 + 0.087 -5.9 + 6.6 0.9951
Toluene 1.06 + 0.060 -4.6 + 4.6 0.9976
Ethylbenzene 1.06 + 0.046 -3.8 - 3.7 0.9986
m/p-Xylene 1.05 + 0.073 -8.4 + 11 0.9964
o-Xylene 1.07 + 0.060 -4.0 + 4.6 0.9977

462
TABLE 13.12
Statistical characteristics of the results obtained in the round robin test on pesticide analysis by
SPME (sr = repeatability standard deviation, sL = inter-laboratory standard deviation, s, =
reproducibility standard deviation, r = repeatability, R = reproducibility, G.A. = gross average,
C.I. = confidence interval of the gross average, T.V. = confidence interval of the "true" value. All
values are expressed in g/l)
Compound sr sL sR r R G.A. C.I. T.V.
Dichlorvos 2.06 5.04 5.44 5.83 15.4 227.3 27±5.8 25±1.35
EPTC 0.56 1.56 1.66 1.57 4.7 9.9 10+1.6 10+0.54
Ethoprofos 0.82 4.79 4.86 2.32 13.74 15.5 16±2.3 17+0.92
Trifluralin 0.27 0.57 0.63 0.76 1.79 1.6 1.6±0.76 2±0.11
Simazine 2.34 3.45 4.17 6.61 11.79 23.6 24+6.6 25±1.35
Propazine 1.21 2.04 2.37 3.42 6.71 9.5 10-3.4 10±0.54
Diazinon 0.63 2.13 2.22 1.79 6.29 8.2 8+1.8 10-0.54
M. chlorpyriphos 0.12 0.32 0.34 0.35 0.97 1.6 1.6±0.35 2-0.11
Heptachlor 2.03 2.89 3.53 5.75 10 8.9 9+5.8 10+0.54
Aldrin 0.54 0.73 0.91 1.53 2.58 2.0 2+1.5 2+0.11
Metolachlor 0.73 2.83 2.92 2.07 8.28 15.7 16±2.1 17±0.92
Endrin 0.87 3 3.13 2.47 8.85 8.8 9±2.5 10±0.54

13.5.2.18 Validation of the method

Validation of the method might include comparison of quantitation results with
certified values obtained for standard reference materials which have similar
matrix and target analytes. Another approach is to validate the method against
officially accepted techniques for the analysis of target samples and analytes.
Table 13.11 summarizes the results of multilevel validation of SPME against the
standard Purge and Trap technique for the analysis of BTEX compounds in
water. The regression line has a slope close to one, with a very small intercept
value and linear correlation coefficients better than 0.99 for all species. The very
good agreement between both methods indicates suitability of headspace SPME
for the analysis of volatile organic compounds in aqueous samples [107].
Finally, inter-laboratory studies are frequently performed to validate a
method. Table 13.12 summarizes the results obtained in a round robin test on
pesticide analysis by SPME, which involved several laboratories in Europe and
North America [108]. In general, the results are characterized by good repeat-
ability, which proves that SPME is a valid method for the determination of pesti-
cides at trace levels. As expected, the inter-laboratory and reproducibility
standard deviations are higher since they include differences between laborato-
ries. However, they are still satisfactory. The results indicate also that SPME is
an accurate method. In all cases, the confidence intervals of the gross average
and the "true" value overlap, which indicates that any differences between the
two respective values are due to random factors. Interestingly, for 10 out of 12
compounds, the values of the gross average are slightly lower than the "true"

463
values. This might be due in part to losses of analytes through adsorption (as
described above) in cases where the aqueous pesticide solutions were not pre-
pared directly before the analysis, as required by the test protocol.

13.5.2.19 Automation of the method

SPME is a very powerful investigative tool, but it can also be a technique of
choice in many applications for processing a large number of samples. To
accomplish this task would require automation of the methods developed. As
automated SPME devices with more advanced features and capabilities become
available, automation of the methods developed becomes easier. The SPME
autosampler currently available from Varian enables direct sampling with
agitation of the sample by fibre vibration, and static headspace sampling. In
some cases, custom modifications to the commercially available systems can
facilitate operation of the method closer to optimum conditions.

13.6 APPLICATIONS OF SPME

This section discusses the more fundamental concepts related to the application
of SPME for analysis of real matrices and physicochemical applications. The
applications of the technique are covered in more detail in the application chap-
ters, as well as in the bookApplications of Solid PhaseMicroextraction[109]. An
updated list of SPME publications is posted in a database at the University of
Waterloo website (www.spme.uwaterloo.ca). A basic understanding is emerging
regarding approaches which can be taken to the basic matrix types: gaseous, liq-
uid and solid. Progress in this direction is summarized in the first three parts of
this section. General approaches to facilitate successful extraction and quantifi-
cation in these systems are emphasized. The objective of this section is not to
demonstrate a comprehensive review, but rather to discuss fundamental princi-
ples facilitating successful quantification. Physicochemical applications empha-
size the capability of SPME to characterise distribution of analytes in the
natural system; the section on investigation of living objects briefly summarizes
this exciting new direction.

13.6.1 Gaseous matrices

All SPME methods developed to date for gas analysis are directed towards air
analysis. However, the approaches described are also suitable for the analysis of
other gas mixtures.
Model studies on extraction from air have been performed by two methods:
static and (the more realistic) dynamic. In the static method, the target com-
pounds are introduced to a glass bulb equipped with a septum. After the analytes
vaporize completely, SPME fibre is introduced to the bulb through a septum. In
practical ambient air measurements, the system is not static, since convection is
always present. It is more appropriate, therefore, to use dynamic gas chambers

464
 logK = a-)+b

4.50

4.00 - 2' '...'1.i.mesitylene

*
~ ~ ~~
' '~.. ....~~~.....1:.
....... _".. . ~ .; ... -l
3.50 .~~~ p-xylene
~~~~~~-
- .... -- - - ...........----- -.-- ---.I. .. n
S 3.00 ! ~~~~~~~~ ;.:........
...
E- *.~~~~- . 4.E - "'!benzene
1
2.50
r P.- ~ I I I i I ------------
_5a---!a
--
. -------
I -
1-r. - I I
2.00

1.50

1/Temp (I/K) (x 103)

Fig. 13.60. Representative plots for pentane, benzene, p-xylene, mesitylene and n-undecane
demonstrating the linear relationship between logKf and 1/T.

for the modelling studies [110]. Equilibration times for extraction of trace
contaminants from moving air are short, as predicted from the theory. Increased
air flow decreases the equilibration times for less volatile analytes.
PDMS coating is generally not affected by major air components. A small
decrease in the response for organics was observed only for samples of relative
humidity close to 100%. The primary experimental parameter which controls
the response is temperature. However, the effect of temperature on the distribu-
tion constant can be easily predicted, since logKfg is linearly related to 1/T and
the heat of vaporization of the pure solute. In other words, Kfg can be calculated
for given extraction conditions using the temperature measured and the heat of
vaporization for the target compound. Figure 13.60 illustrates the above rela-
tionship graphically for a range of compounds varying in volatility. The linear
relationship is clearly illustrated. In addition, as has been described in the
section on theory, the heat of vaporization and activity coefficient are related to
the retention time of a compound on the GC column using the same coating
material as the SPME fibre. Thus, the appropriate distribution constants can be
determined directly from the retention indices of the target analytes. For PDMS
coating, the following relationship between the gas/coating distribution con-
stant and the linear temperature programmed retention index of a compound
(LTPRI) has been found (see Fig. 13.61):
logoKfg = 0.00415- LTPRI- 0.188 (13.16)
Thus, the LTPRI system permits interpolation of the Kfg values from the plot of
logKfg versus retention index. The differences between the values of distribution
constants calculated from the retention indices and determined directly were

465
 nn

5.00

*E 4.00

0 I
m_ 3.00
2.00

1.00

0.00
400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

Linear Temperature Programmed Retention Index (LTPRI)

Fig. 13.61. Correlation between analyte logKfg and LTPRI for n-alkanes.

generally found to be within experimental error. This approach can be extended

to other coatings as long as columns with appropriate coatings are available. In
addition, using this approach it is possible to determine the distribution con-
stants of unidentified compounds, therefore it is possible to use SPME for the
determination of parameters like total petroleum hydrocarbons (TPH) in air. A
comparison of SPME/PDMS with the standard charcoal tube technique obtained
for an air sample drawn near a gas station, resulted in 262 and 247 ug/l,
correspondingly [5]. The values show a very good agreement indicating that the
approach described above can be used to quantify analytes in gaseous samples
without the need for identification. It should be pointed out, however, that this
approach can be used only with detectors whose response is independent of the
nature of the analyte (e.g., FID for the analysis of hydrocarbons).
SPME has superior sensitivity for short-term monitoring compared to tradi-
tional devices, which are limited by the gas throughput. The SPME fibres can
also be used at high temperatures. For example, Fig. 13.62 illustrates the identi-
fication of several polycyclic aromatic hydrocarbons (PAHs) directly in diesel
exhaust. Particulate matter can be collected at the same time and characterised
using spectroscopic means [111]. Alternatively, the particulate collection device
can be also constructed as a packed needle (see Fig. 13.45). The needle can be
then introduced to the analytical instrument for desorption and analysis.
Typically, the fibre reaches equilibrium with the air components in the first
few minutes and then does not accumulate any more analytes, independently of
the exposure time. However, the sensitivity of the SPME technique can be
improved by using thicker, more selective coatings or by cooling the fibre to
increase the PDMS/gas distribution constant. Another approach is to incorpo-
rate the derivatization reagent in the coating to allow "trapping" of the analyte
in the coating. All these modifications enhance the sensitivity of SPME in air
monitoring.

466
 n~ l

100%
(a)
TIC

i I I I I
0.46% Naphthalene
128 (b)
(a) TIC
wI I I 1 * (b) m/z = 128
3.14% Phenanthrene (c) m/z = 178
(C) (d) m/z = 202
178 Anthracene

Fluoranthene
(d)
202

10:00 11:40 13:20 15:00 16:40 18:20

Time (min)

Fig. 13.62. GC/MS traces corresponding to polycyclic aromatic hydrocarbons present in diesel
exhaust sampled by SPME/PDMS. (a) Total ion chromatogram; selected ion traces at: (b) m/z =
128, (c) 178, (d) 202.

SPME can also be used as an integrated sampling device. Theoretical princi-

ples of this approach have been described in the section on theory. As mentioned
previously, the simplest way to accomplish this task is to retract the fibre into
the needle. An alternative approach is to use a chemical reaction in the coating.
This approach is particularly useful for integrated sampling of very low analyte
concentrations in air. An important application of SPME coupled to an on-fibre
derivatization step was demonstrated for the analysis of aldehydes in air. In this
case o-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine was used as derivatization
reagent and the products, oximes were analyzed by GC [112].
After the extraction, the fibres with analytes in the coating often cannot be
analyzed immediately. This is particularly true in field applications, since an
appropriate portable instrument might not be available. In such situations, the
fibres have to be stored in a way which ensures that analyte losses are minimal.
This can be simply accomplished by retracting the fibre into the needle and
sealing its open end with a piece of a septum. Cooling the needle provides
additional protection against losses. For PDMS fibres, even with such precau-
tions, losses of the most volatile analytes might occur when the fibre is stored for
more than a few hours. When properly protected (retracted fibre, needle sealed
with a septum or other sealing approach [113], low temperature), fibres coated
with PDMS/DVB can be safely stored for much longer periods of time.

13.6.2 Liquid matrices

The majority of methods for liquid matrices reported in the literature were
developed for aqueous samples. The fibre coating/water distribution constant
can be calculated from the following equation:

467
 1
iogK ,= 0.0042*LTPRI - 0.188 + logK,,aw I
5.50

5.00 Branched alkanes l

4.50 logK>,=0.0041*LTPRtI1 33 * - w,

*
4.00 -

350
3 00
3; -

Cycopentane and
250 ,' BenzeneSubstituted Aromatics
*Cylohexe
2.50 Derivatives ' IogK,
ogK, = 0 0043LTPRI -0
00043*LTPR1 88
osS
~- ~IgK,
0.0040
ULTPRI 0.69
2.00 C C C =K
.i50 K C cf h K1
.oo
1.00 ,i,
10 200 400 600 800 1000 1200 1400
LTPRI

Fig. 13.63. LogK, (PDMS) as a function of LTPRI for isoparaffins, substituted benzenes and
cycloalkanes at 25C.

Kfw = KfgKaw (13.17)

where Kfg can be determined from chromatographic data as discussed above, and
Kaw is the air/water distribution constant for a given analyte which can be found
in the tables of Henry's constant values. For example, the equation for a PDMS
coating and aqueous matrix can be calculated from the equation below:

Kfw = 0.00415 LTPRI-0.188 + logK 0 (13.18)

This equation is obtained after substituting the expression for Krg given by Eq.
(13.17) into Eq. (13.16). The Kfw values found using this equation agree very well
with experimental data considering that typically the errors in Henry's constant
determination are larger than 10%. In addition, Henry's constants are similar
for closely related compounds, as illustrated in Fig. 13.63. As could be predicted
from Eq. (13.18), the slopes of the lines are similar, but their intercepts vary
because of the differences between the average values of Henry's constant for
the different classes of compounds. This relationship enables quantitation of
unknown analytes in water provided that the class to which a compound belongs
is known.
Several reports indicated the existence of a linear relationship between the
coating/water distribution constant and octanol-water distribution constant,
Kow,, [114]. Considering the discussion above, such relationships are expected to
exist only for individual classes of compounds, such as aromatic hydrocarbons.
Because the activity coefficients of various classes of analytes in octanol are
expected to be different than in PDMS or other fibre coatings, the relationship
between the two extracting phases will vary with the change of chemical proper-

468
ties of the analytes. It is possible, however, to predict the general trends in Kf,
within a group of related compounds by using the corresponding values of Kow.
The above discussion pertains to pure water as the sample matrix. The
presence of other components in water might affect the distribution constants of
the analytes. Also, liquid chromatographic experience gives some clues about the
trends in the distribution constant change with modification of the matrix. For
example, addition of salt would generally result in an increase in the distribution
constant for neutral organics, but the change is expected to be noticeable only if
the concentration of the salt exceeds 1%. Also, the presence of water miscible
polar organic solvents may change the properties of the matrix by reducing its
polarity. In addition, swelling of the polymer with the solvent might occur for
polar coatings, resulting in a change of the coating volume and possibly also of
Kf,. However, the change is not expected to be substantial when the concentra-
tion of the solvent is below 1% [115]. When samples contain higher amounts of
salt and/or dissolved organics, but they are well defined so that pure matrix can
be prepared, external calibration using standards in the matrix may still be
appropriate. Otherwise, standard addition (preferentially of isotopically labelled
analytes) should be used to compensate for variations in the matrix composition.
A major analytical challenge is always associated with the analysis of sam-
ples containing solids, such as sludge. Several approaches can be implemented
with SPME. Sometimes modification of the extraction conditions, such as tem-
perature, pH, salt and other additives, facilitates the displacement of analytes
into the aqueous phase or the headspace, resulting in distribution constants sim-
ilar to those for pure water. In certain cases, direct extraction is not possible
because of a very dirty matrix or pH conditions which damage the fibre (above
pH 12 for the PDMS coating). In such situations, the headspace mode is more
suitable for many applications. Even semivolatiles can be analyzed by this
method as long as the extraction temperature is sufficiently high, and good agi-
tation conditions are provided.

13.6.3 Solid matrices

The accurate quantitation of target analytes in solids is a very significant

challenge to the analytical community. Although SPME cannot be used directly
to extract analytes from solids, several approaches can be taken to facilitate
simple sample preparation. For volatiles, the typical approach is to perform
headspace analysis. To quantitatively release the analytes from the matrix, the
temperature needs to be increased. This facilitates extraction and improves the
kinetics of the process, but it also decreases the distribution constant. Typically,
a maximum is observed for the relationship between the amount of analyte
extracted by the fibre and temperature. Initially, the loss of sensitivity due to
decreased distribution ratio is more than compensated for by the increased
concentration of the analyte in the headspace. At even higher temperatures, the
loss of sensitivity becomes predominant and the amount of analyte extracted by

469
the fibre decreases. The inherent loss of sensitivity associated with the decrease
of the distribution constant at high temperatures can be compensated for by
cooling the fibre, as discussed earlier. However, this approach has not yet been
commercialized. Alternatively, matrix modifier can be added to the solid sample.
Water has been found to be very effective in displacing the analytes from the
solid surfaces for many types of samples, especially at elevated temperatures.
Another successful approach to SPME analysis of solids involves the use of
water or a polar organic solvent, such as methanol, for the extraction of the
analytes from the solid matrix prior to SPME analysis. When a polar solvent is
used, the extract is spiked into pure water. Low temperature water extraction
followed by SPME was found to be a very useful approach for polar compounds,
such as herbicides [116]. Application of methanol with water spiking, on the
other hand, has been found to be useful for the analysis of volatile hydrocarbons
[117]. An interesting modification to the above procedure involves volatizing the
extract followed by fibre extraction from the gaseous phase [118]. Quantitation
of analytes in the gas phase is easier, as discussed above.
For less volatile analytes, the methanol approach can give good results.
High-pressure hot water extraction is also a suitable solvent-free alternative
(see Chapter 18). Both static and dynamic hot water extraction have their advan-
tages: the static method is very simple and inexpensive as it does not use high
pressure pumps; the dynamic approach, on the other hand, provides an extract
which is much cleaner than the original matrix. However, it might be possible to
extract many semivolatile target analytes from very dirty matrices with the high
temperature static system, when using the headspace mode of SPME.

13.6.4 Physicochemical applications

Solid phase microextraction can be applied not only for extraction purposes, but
also to perform measurements which better characterize the extraction system.
The measurements can include studying the properties of the fibre coating and
investigation of multiphase equilibria in the matrix. These investigations are
convenient because SPME is and equilibrium technique and the fibres are small
and therefore to not disturb chemical equilibria in the system as it is illustrated
on Fig. 13.64. In addition, since detailed mathematical treatment of the solid

Fig. 13.64. Distribution of analyte in heterogeneous large sample

--A ---
illlll

470
illlall II CVaLU
---I-A RUMS
TiYlllr riht
IIVlt·
l
phase microextraction process is available, SPME can be applied to study the
parameters involved in the extraction, including diffusion coefficients in both
the coating and the extracted phase, as well as various distribution constants.
For understanding solute-solvent interactions at the molecular level and the
thermodynamic processes involved in forming the solution, the study of infi-
nite-dilution activity coefficients of probe solutes in a polymer phase is a funda-
mental approach. SPME can be a useful tool for determining these coefficients.
The stationary phases of interest can be coated on fibres made of suitable mate-
rials (fused silica, stainless steel, etc.). The SPME device with a selected fibre
coating can be used to extract a group of probe compounds, which are then sepa-
rated on a standard commercially available column and quantified by a GC/MS.
The SPME measurements of infinite-dilution activity coefficients of probe
solutes in a selected stationary phase consist of two steps. The first step is to
measure the infinite-dilution partition coefficients of the probe solutes between
the stationary phase and the gas phase. Then, from these partition coefficients,
the corresponding activity coefficients are determined [119].
Investigation of analyte distribution in natural systems is critical for many
disciplines of science including environmental chemistry and biochemistry.
Information gained during these studies is also very important to analytical
chemists, since it facilitates the development of optimum sampling protocols and
can lead to more rational optimization of extraction parameters. One approach
to study multiphase equilibria involves separation of the individual phases after
equilibrium has been reached and analysis of the individual phases to determine
the concentrations of target compounds. For example, SPME was used to study
the concentration of organic substances from octanol saturated water [12].
However, the separation of phases is not always necessary. SPME has been
applied to measure the distribution of chemicals between the components of a
system, for example, consisting of water and macromolecules such as dissolved
humic organic matter (HOM) [121]. There are two methods for performing this
experiment [122]. Typically, the calibration curve is first constructed by employ-
ing the amount of the analytes partitioned on the fibre vs. the free analyte
concentration for pure water. In method I designed for large volume systems
(see Fig. 13.65), the extraction is then performed in the binding macromolecule

SPME
fibre

vial

sample solution

dissolved protein

Fig. 13.65. Distribution of drug between solution, fibre

and protein.

471
 l lI .Protein Soiition

Polethylee Inseralt

Fig. 13.66. Fibre delivery approach to s

qtiidv nrntpin-drn hbinling for mill1
volume of protein solution. Loading
i!0;
Desgrin

solution (e.g., protein, humic material) with a known total amount of the analyte
(e.g., drug, pesticide). In method II designed for small samples (see Fig. 13.66), a
known amount of the analyte is first loaded onto the fibre and then this fibre is
put into the protein solution for delivery of the analyte. The amount of analyte
remaining on the fibre is analyzed after the system reaches equilibrium. The
free analyte concentration is then calculated from the calibration curve obtained
during the first step in both of the methods. Since only a very small amount (150
Al for each extraction) of the macromolecule solution is required, method II is
very, useful for situations where the macromolecule is very limited, for example,
in the case of some proteins. Method I does not require determination of absolute
masses of analyte since the only information needed is the area count change
(ratio) of the fibre injections before and after the macromolecule was introduced
into the system, to calculate the distribution in this two phase system. These
methods have been applied to drug binding studies [123]. The procedures
described above can be extended to natural systems containing native analytes
by using isotopically labelled analogues [124]. This topic is discussed in more
detail in Chapter 8.

13.6.5 Investigation of living systems

Recently, interest has been shown in monitoring levels of biologically active

compounds in living systems in natural environments. These efforts are a
significant departure from conventional sampling techniques, where a portion of
the system under study is removed from its natural environment, and the
compounds of interest extracted and analyzed in a laboratory environment.
There are two main motivations for exploring these types of investigations: the
first is the desire to study chemical processes in association with the normal
biochemical milieu of a living system, and the second is the lack of availability or
impracticality of removing suitable samples for study from the living system,
frequently associated with size. SPME, using an externally coated extraction
phase on a microfibre mounted in a syringe-like device, seems a logical target for
the development of such tools. As with any microextraction technique, the
compounds of interest are not exhaustively removed from the investigated
system. On the contrary, conditions can be devised where only a small propor-
tion of total compound is removed, thus avoiding disturbance of the normal

472
TABLE 13.13
Spacial distribution of pesticides in tomato plant stem
Height from base (cm) Mass extracted (pg)
Atriazine Simazine Prometryn
113 2 2 0
99 26 52 9
92 0 0 8
68 0 26 0
43 0 0 0
26 0 7 0
10 9 35 0
6 43 101 148

balance of chemical components. Secondly, because it is a syringe-like device

that can be physically removed from the laboratory environment for sampling, it
is amenable to the monitoring of a living system in its natural environment,
rather than trying to move the living system to an unnatural laboratory environ-
ment. Early in vivo investigations focused on fragrances emitted by living
plants, fungi and bacteria; these are summarized in Chapter 21. More recently,
these investigations were extended to volatiles emitted by animals [125]. It is
also possible to effectively sample components or parts of the whole plant or
animal since the puncture made by a micro fibre does not cause much damage to
a living system. Table 13.13 shows data obtained for SPME direct sampling of a
tomato plant along its stem, which had been exposed to pesticides [126].
The data clearly shows a non-uniform distribution of pesticide in the stem.
Higher concentrations of pesticides are located at the base of the plant. This
result has been confirmed by the results obtained using an independent, more
traditional analytical protocol involving homogenising sections of the stem. The
agreement clearly indicates that SPME is suitable for direct semiquantitative
investigations of living systems.
The direct sampling approach requires the application of micro-fibres, much
smaller diameter compared to commercial devices. The amount of analyte col-
lected on these fibres is very small. Therefore, to successfully detect extracted
components, micro-separation technologies, such as capillary electrophoresis
(see Fig. 13.39) combined with fluorescence detection, might be necessary. For-
tunately, a coated fibre is well suited to perform on-fibre labelling since, after
extraction, the isolated analytes can be derivatized by exposing the fibre to the
reagent. Alternatively the labelling procedure can be performed during samp-
ling by doping the fibre with the appropriate reagent prior to exposure to the liv-
ing system (see Fig. 13.30). Figure 13.67 illustrates the results of such an initial
study involving SPME sampling of small grapes with a carbowax coated micro
fibre (coating thickness: 10 nm, o.d.: 70 gm), then derivatization with 4-fluoro-

473
 8

E .5
D
6

,0 . 4
@
c

o
0 2 4 6 8 0 2 4 6 8
Migration Time / minutes Migration Time / minutes

Fig. 13.67. Electropherograms corresponding to separation of labelled SPME extracts from a

green grape (G) and a red seedless grape (R). E is the glutamic acid derivative. SPME/CE
interface: off-column desorption. Desorption solution: methanol, CE running buffer, 0.1 M
NaOH (3:3:4). CE condition: separation capillary (75,um i.d. and 365gm o.d.) was 55 cm long.
The CE running buffer was 20 mM Borate (Na+, pH 9.2), 10 mM Brij 35 and 2.5% methanol. The
running voltage was 12 kV.

7-nitro-2,1,3-benzoxadiazole (NBD-F) in the headspace of triethylamine (TEA)

to label aminoacids extracted into the coating, and followed by CE separation
with fluorescence detection [127]. Glutamic acid is identified based on standards
as one of the extracted components from both red and green grapes. To further
improve the capability of SPME for in vivo sampling, new specific coatings, such
as affinity phases discussed briefly in this chapter, should be developed for a
range of important target analytes. Coupling of micro fibres with micro instru-
mentation would result in on-site monitoring capabilities, which will make
micro technology more accessible in the future to medical research community.

REFERENCES

1 J. Pawliszyn and S. Liu, Anal. Chem., 59 (1987) 1475.

2 R.G. Belardi and J. Pawliszyn, Water Pollut. Res. J. Can., 24 (1989) 179.
3 C.L. Arthur and J. Pawliszyn, Anal. Chem., 62 (1990) 2145.
4 J. Pawliszyn, Method and Device for Solid Phase Microextraction and Desorption.
PCT, International Patent Publication Number WO 91/15745 and national
counterparts.
5 M. McComb, E. Giller and H.D. Gesser, in: 78th Canadian Society for Chemistry
Conference and Exhibition. University of Guelph, Guelph, ON, 1995, Abs. 528.
6 M. Chai and J. Pawliszyn, Environ. Sci. Technol., 29 (1995) 693.
7 E. Baltussen, F. David, P. Sandra, H.-G. Janssen and C. Cramers, J. High Resolut.
Chromatogr., 21 (1998) 333.

474
 8 M.A. Nickerson, Sample screening and preparation within a collection vessel. U.S.
patent number 5,827,944.
9 E. Baltussen, P. Sandra, F. David, H-G. Janssen and C. Cramers, Anal. Chem., 71
(1999) 5213-5216.
10 X. Liu, I. Bruheim, J. Wu and J. Pawliszyn, Anal. Chem., in preparation.
11 P. Popp and M. Moeder, in: ExTech'2001. Barcelona, Spain, September, 2001.
12 C. Grote and J. Pawliszyn, Anal. Chem., 69 (1997) 402.
13 L. Miller, in: J. Pawliszyn (Ed.), Applications of Solid Phase Microextraction. RSC
Chromatography Monographs, The Royal Society of Chemistry, Cambridge, UK,
1999.
14 D. Louch, S. Motlagh and J. Pawliszyn, Anal. Chem., 64 (1992) 1187.
15 J. Al-Hawari, in: 78th Canadian Society for Chemistry Conference and Exhibition.
Memorial University of Newfoundland, St. Johns, NF, 1996.
16 J. Poerschmann, F-D Kopinke and J. Pawliszyn, Environ. Sci. Technol., 31 (1997)
3629.
17 Z. Mester and J. Pawliszyn, Rapid Comun. Mass Spectrom., 13 (1999) 1999.
18 S. Motlagh and J. Pawliszyn, Anal. Chim. Acta, 284 (1993) 265.
19 Z. Zhang and J. Pawliszyn, Anal. Chem., 65 (1993) 1843.
20 S. Motlagh and J. Pawliszyn, Anal. Chim. Acta, 284 (1993) 265.
21 Z. Zhang and J. Pawliszyn, J. High Resolut. Chromatogr., 19 (1996) 155.
22 Z. Zhang, J. Poerschmann and J. Pawliszyn, Anal. Con., 33 (1996) 219.
23 J. Crank, Mathematics of Diffusion. Clarendon Press, Oxford, 1989, p. 14.
24 J. Pawliszyn, J. Chromatogr.Sci., 31 (1993) 31.
25 R. Eisert and J. Pawliszyn, Anal. Chem., 69 (1997) 3140.
26 M. Chai and J. Pawliszyn, Environ. Sci. Technol., 29 (1995) 693.
27 J. Crank, Mathematics of Diffusion. Clarendon Press, Oxford, 1989, p. 2.
28 A. Khaled and J. Pawliszyn, J. Chromatogr., 892 (2000) 455.
29 P. Martos and J. Pawliszyn, Anal. Chem., 71(8) (1999) 1513.
30 B. Shurmer, Investigation of the Bioavailability of Organic Contaminats by Solid
Phase Microextraction. University of Waterloo, Waterloo, Canada, 1999.
31 Z. Zhang, M.J. Yang and J. Pawliszyn, Anal. Chem., 66 (1994) 844A.
32 E. Otu and J. Pawliszyn, J. Mikrochim. Acta, 112 (1993) 41.
33 J.L. Liao, C.M. Zeng, S. Hjerten and J. Pawliszyn, J. Microcol. Sep., 8 (1996) 1.
34 F. Guo, T. Gorecki, D. Irish and J. Pawliszyn, Anal. Con., 33 (1996) 361.
35 H.B. Wan, H. Chi and M.K. Wong, C.Y. Mok, Anal. Chim. Acta, 298 (1994) 219.
36 T. Gorecki, P. Martos and J. Pawliszyn, Anal. Chem., 70 (1998) 19.
37 J. Wu, X. Yu, H. Lord and J. Pawliszyn, Analyst, 125 (2200) 391.
38 H. Ge, K. Gilmore, S. Ashraf, C. Too and G. Wallace, J. Liq. Chromatogr., 16 (1993)
1023.
39 G.G. Wallace, M. Smyth and H. Zhao, Trends Anal. Chem., 18 (1999) 245.
40 A. Michalska, A. Ivaska and A. Lewenstam, Anal. Chem., 69 (1997) 4060.
41 J. Wu, H. Kataoka, H. Lord and J. Pawliszyn, J. Microcol. Sep., 12 (2000) 255.
42 H. Kataoka, S. Narimatsu, H. Lord and J. Pawliszyn, Anal. Chem., 71 (1999) 4237.
43 G. Vlatakis, L.I. Andersson, R. Miller and K. Mosbach, Nature, 361 (1993) 645.
44 D.S. Hage, Anal. Chem., 65 (1993) 420R.
45 H. Yuan, W. Mullett and J. Pawliszyn, Analyst, 126 (2001) 1456.
46 J-L. Liao, C-M. Zeng, S. Hjerten and J. Pawliszyn, J. Microcolumn Sep., 8 (1996) 1-4.
47 T. Gorecki, X. Yu and J. Pawliszyn, Analyst, 124 (1999) 643-649.
48 G. Xiong, Y. Chen and J. Pawliszyn, Anal. Chem., in preparation.
49 P. Cheo, Fiber Optics. Prentice-Hall, Englewood Cliffs, NJ, 1985, p. 88.
50 J. Chongrong and J. Pawliszyn, J. Microcol. Sep., 10 (1998) 167-173.
51 J. Pawliszyn, Solid Phase Microextraction:Theory and Practice.Wiley, NY, 1997.

475
52 Z. Zhang and J. Pawliszyn, Anal. Chem., 67 (1995) 34.
53 S. Hawthorne, Y. Yang and D. Miller, Anal. Chem., 66 (1994) 2912.
54 H. Diamon and J. Pawliszyn, Anal. Corn., 33 (1996) 421.
55 L. Pan and J. Pawliszyn, Anal. Chem., 69 (1997)196.
56 K. Buchholz and J. Pawliszyn, Anal. Chem., 66 (1994) 160.
57 L. Pan and J. Pawliszyn, Anal. Chem., 69 (1997) 196-205.
58 L. Pan and J. Pawliszyn, Anal. Chem., 67 (1995) 4396.
59 P. Martos and J. Pawliszyn. Anal. Chem., 70 (1998) 2311.
60 P. Konieczka, L. Wolska, Ei Luboch, J. Namiesnik, A. Przyjazny and J. Biernat, J.
Chromatogr.A, 742 (1996) 175.
61 E.D. Conte and D.W. Miller, J. High Resolut. Chromatogr., 19 (1996) 294.
62 T. G6recki and J. Pawliszyn, J. High Resolut. Chromatogr., 18 (1995) 161.
63 T. Gorecki and J. Pawliszyn, Anal. Chem., 67 (1995) 3265.
64 J. Chen and J. Pawliszyn, Anal. Chem., 67 (1995) 2530.
65 Y. Hirata and J. Pawliszyn, J. Microcolumn Sep., 6 (1994) 443.
66 R. Eisert and J. Pawliszyn, Anal. Chem., 69 (1997) 3140.
67 R. Eisert and J. Pawliszyn, Crit. Rev. Anal. Chem., 27 (1997) 103.
68 Y. Gou, C. Tragas, H. Lord and J. Pawliszyn, J. Microcol. Sep., 12(3) (2000) 125.
69 H. Kataoka, H. Lord and J. Pawliszyn, J. Chromatogr. B, 731 (1999) 353.
70 H. Kataoka, H. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 257.
71 H. Kataoka and J. Pawliszyn, Chromatographia,50 (9/10) (1999) 532.
72 T. McDonnell and J. Pawliszyn, Anal. Chem., 63 (1991) 1884.
73 C-W. Whang and J. Pawliszyn, Anal. Commun., 35 (1998) 353.
74 B.L. Wittkamp and D.C. Tilotta, Anal. Chem., 67 (1995) 600.
75 G.L. Klunder and R.E. Russo, Anal. Chem., 49 (1995) 379.
76 H.M. Yan, G. Kraus and G. Gauglitz, Anal. Chim. Acta, 312 (1995) 1.
77 J. Pawliszyn, Device and Process for Increasing Analyte Concentration in a Sorbent,
U.S. Patent 5,496,741.
78 William R. Heineman, University of Cincinnati, USA, personal communication.
79 C. Arthur, L. Killam, K. Buchholz, J. Berg and J. Pawliszyn, Anal. Chem., 64 (1992)
1960.
80 T. Gorecki and J. Pawliszyn, Field Anal. Chem. Technol., 1(5) (1997) 277-284.
81 J.A. Koziel, M. Jia, A. Khaled, J. Noah, and J. Pawliszyn, Anal. Chim. Acta, 400
(1999) 153.
82 J. Koziel, B. Shurmer and J. Pawliszyn J. High Resol. Chromatogr., 23 (2000) 343.
83 R. Eisert and K. Levsen, J. Chromatogr.A, 737 (1996) 59.
84 J. Koziel, M. Odziemkowski and J. Pawliszyn, Anal. Chem., 73 (2001) 47.
85 X. Liu, J. Wu and J. Pawliszyn, Analyst, in preparation.
86 H. Tong, N. Sze, B. Thomson, S. Nacson and J. Pawliszyn, Analyst, in preparation.
87 P. Martos, A. Saraullo and J. Pawliszyn, Anal. Chem., 69 (1997) 402.
88 P. Martos, A. Saraullo and J. Pawliszyn, Anal. Chem., 69 (1997) 1992.
89 R. Schwarzenbach, P. Gschwend and D. Imboden, Environmental Organic
Chemistry. Wiley, New York, 1993, pp. 109-123.
90 V. Mani (Supelco), private communication.
91 K. Buchholz and J. Pawliszyn, Environ. Sci. Technol., 27 (1993) 2844.
92 T. G6recki, P. Martos and J. Pawliszyn, Anal. Chem., 70 (1998) 19.
93 C. Jia, Y. Luo and J. Pawliszyn, J. Microcolumn. Sep., 10 (1998)167.
94 L. Pan and J. Pawliszyn, Anal. Chem., 69 (1997) 196.
95 Y. Cai and J. Bayona, J. Chromatogr., 696 (1997) 113.
96 J. Poerschmann, F-D Kopinke and J. Pawliszyn, Environ. Sci. Technol., 31 (1997)
3629.
97 L. Pan, M. Adams and J. Pawliszyn, Anal. Chem., 67(1995) 4396.

476
 98 J. Chen and J. Pawliszyn, Anal. Chem., 67(1995) 2530.
99 H. Daimon and J. Pawliszyn, Anal. Comm., 33 (1996) 365.
100 T. G6recki and J. Pawliszyn, Analyst, 122 (1997) 1079.
101 M. Chai and J. Pawliszyn, Environ. Sci. Technol., 29 (1995) 693.
102 L. Wang, Determination of Amphetamine and Methamphetamine in Urine and
Blood Plasma by Headspace Solid Phase Microextraction, M.Sc. Thesis, University of
Waterloo, Waterloo, Canada, 1996.
103 J-L. Liao, C-M Zeng, S. Hjerten and J. Pawliszyn, J. Microcol. Sep., 8 (1996) 1.
104 K. Buchholz and J. Pawliszyn, Anal. Chem., 66 (1994) 160.
105 H. Daimon and J. Pawliszyn, Anal. Corn., 33 (1996) 421.
106 Z. Zhang, J. Poerschmann and J. Pawliszyn, Anal. Corn., 33 (1996) 129.
107 B. MacGillivray, P. Fowlie, C. Sagara and J. Pawliszyn, J. Chromatogr. Sci., 32
(1994) 317.
108 T. G6recki and J. Pawliszyn, Analyst, 121(1996)1381.
109 J. Pawliszyn (Ed.), Applications of Solid Phase Microextraction. Royal Society of
Chemistry, Cambridge, UK, 1999.
110 P. Martos and J. Pawliszyn, Anal. Chem., 69(1997) 206.
111 M. Odziemkowski, J. Koziel, D. Irish and J. Pawliszyn, Anal. Chem., 73 (2001) 3131.
112 P. Martos and J. PawliszynAnal. Chem., 70(1998) 2311.
113 L. Muller, T. Gorecki and J. Pawliszyn, Fresenius J. Anal. Chem., 364 (1999) 610.
114 J. Dean, W. Tomlinson, V. Makovskaya, R. Cumming, M. Hetheridge and M. Comber,
Anal. Chem., 68 (1996)130.
115. C. Arthur, L. Killam, K. Buchholz and J. Pawliszyn, Anal. Chem., 64 (1992) 1960.
116 A. Boyd-Boland and J. Pawliszyn, J. Chromatogr., 704 (1995)163.
117 B. MacGillivray, Analysis of Substituted Benzenes in Environmental Samples by
Headspace Solid Phase Microextraction, M.Sc. Thesis, University of Waterloo,
Waterloo, 1996.
118 A. Saraullo, Determination of Petroleum Hydrocarbons in the Environment by
SPME, M.Sc. Thesis, University of Waterloo, 1996.
119 Z. Zhang and J. Pawliszyn, J. Phys. Chem., 100 (1996) 17648.
120 P. Popp, A. Paschke, U. Schroter and G. Oppermann, ChemiaAnalityczna (Warsaw),
40 (1995) 897.
121 J. Poerschmann, Z. Zhang, F. Kopinke and J. Pawliszyn, Anal. Chem., 69 (1997) 597.
122 H. Yuan and J. Pawliszyn, Anal. Chem., 73 (2001) 4410.
123 H. Yuan, Bioapplications of Solid Phase Microextraction, Ph.D. Thesis, University of
Waterloo, Waterloo, 2000.
124 J. Pawliszyn, Solid PhaseMicroextraction: Theory and Practice.Wiley, NY, 1997, pp.
185-189.
125 B. Smith, C. Zinni, E. Caramdo, J. Pawliszyn, M. Tyler andY. Hayasaka, Chem. Ecol.,
17 (2000) 215.
126 H. Lord, M. Moeder and J. Pawliszyn, Anal. Chem., in preparation.
127 A. So, N. Sze, A. Namera and J. Pawliszyn, J. Chromatogr.(2002) in press.

477
Chapter 14

Membrane extraction
Janusz Pawliszyn

14.1 INTRODUCTION TO MEMBRANE SEPARATIONS

The topic of membrane separation is very broad since it covers both basic mate-
rial science as well as numerous engineering implementations of the technology.
However, analytical chemists familiar with the theory of separation methods
know fundamental physicochemical principles of membrane separations. Ana-
lytical scale separations, such as solvent extraction or size-exclusion chromatog-
raphy, are typically performed as batch processes. Membrane implementation of
the same fundamental principles facilitates continuous separation. The optimi-
sation of mass transfer in membrane separations often differs compared to cor-
responding batch processes because of different geometric arrangements
implemented to facilitate continuous operation. It is not surprising that most
developments and commercial applications have been conducted by engineers
focused on large-scale continuous separation of products in process streams.
Analytical chemists frequently adopted new membrane materials to develop
their applications. Numerous books and reviews have been published on the
topic of membrane separations, principally by engineers [1-3] but more recently
by analytical chemists as well [4-6]. The objective of this chapter is briefly to
introduce different membrane separation technologies, and then give a funda-
mental and practical discussion of membrane extraction since they are in princi-
ple more selective and therefore more interesting for analytical applications. In
this chapter, the focus is a discussion on polymeric extraction membranes and in
particular their applications in gas chromatography.
A membrane is a selective barrier between two phases. Analytes are trans-
ferred from the sample (donor phase) through the membrane into the acceptor
(or stripping) phase. An essential characteristic of membrane-based separation
processes is the ease with which they can be adopted for continuous operation.
This feature makes membranes very attractive sample preparation tools since
their application allows conversion of the analytical separation and detection
instruments into sensor-like devices suitable for monitoring operations. It is
expected that chemical monitors manufactured using micro-machining technol-
ogies and incorporating membranes will be used directly on-site, similarly to pH
and other selective membrane electrodes presently used.
ComprehensiveAnalyticalChemistry XXXVII
J. Pawliszyn (Ed.)
( 2002 Elsevier Science B.V. All rights reserved
 A SapeEffluent containing concentrated
Sample C0 * . 0 species above cut-off

Stripping phase Effluent containing dissolved

species below cut-off

B
Sample [Aj1 Purified effluent

[Aj Stripping p ase with

Stripping phase target analytes

Fig. 14.1. Schematic representation of separations in (A) size-exclusion and (B) extraction
membranes.

There are two basic types of membrane separations: one involves the mech-
anical separation of sample components based on size exclusion (Fig. 14.1A). A
wide variety of membrane materials can be used, each with its own advantages
and disadvantages [7]. In many cases, a membrane is a porous network of a syn-
thetic polymer, such as polypropylene, polysulfone or a cellulose derivative. Sep-
aration with these membranes is based on size-exclusion principles: sufficiently
small molecules can permeate through the pores, whereas larger ones cannot. A
membrane is usually characterised by its molecular weight cut-off (MWCO)
value. This value is defined as the molecular mass of the smallest compound that
is retained by the membrane to a larger extent than 90%. A usual way of classify-
ing size-exclusion membrane separation processes is by means of the driving
force applied. The filtration process is achieved by applying a pressure differ-
ence; dialysis is driven by a concentration difference and electrodialysis by an
electric potential difference coupled with a concentration gradient.
The other type of membrane separation (see Fig. 14.1B) involves a chemical
interaction between the membrane material and the analytes resulting in parti-
tioning of target components between the membrane material and the sample
matrix. These extraction-type membranes can be homogeneous liquids or solids
with microporous structures of different selective characteristics (e.g., affinity
membranes [8]). As the pore size of the porous membrane decreases, the selec-
tivity of transport becomes defined by the sorption characteristics of the mem-
brane material rather than their porosity. In fact, liquid membranes can be
viewed as having molecular size channels. These nonporous membranes can be
composed of a solvent or polymeric liquid film (for example silicone), into which
a molecule must actually dissolve in order to be able to pass through. In this case,
the efficiency of membrane transport for a particular compound is largely
dependent on its partition coefficient between the different parts of a membrane
separation system. Only compounds that are easily extracted from the donor
phase into the membrane and, in addition, easily extracted from the membrane
into the acceptor phase will be transported. The separation of different com-
pounds is based on the same principle as liquid extraction followed by a back-
extraction. Therefore, molecules with different physicochemical properties can

480
 I Membrane Separation Methods i

Based on S ze Exclusion Based on Extraction

Filtration Selective Microporous MembrE3e.]

anes
Dialysis ' Homogeneous Extaction Membr anes

Electrodialysis Liquid Membranes I

Po ymeric Membranes

Fig. 14.2. Classification of membrane separation methods discussed in this chapter.

be separated even if they are of equal size. In selecting a material, the objective
for a size-exclusion membrane is to minimise chemical interaction between the
membrane and the sample matrix as well as the analytes, while in extraction
type membranes, the basis of separation is defined by the way different sample
components chemically interact with the material. The classification of mem-
brane separation techniques discussed in this chapter is shown in Fig. 14.2.
There are two basic ways membrane extraction can be implemented practi-
cally. One involves the use of flat sheet membranes. This configuration requires
the extraction module to support the membrane and to define the access of the
donor phase (typically a sample) and the acceptor phase to the membrane. Fig-
ure 14.3a gives an example of such a module used in Membrane Extraction with
Sorbent Interface (MESI), which is discussed in detail later. It has been designed
to expose one side of the membrane to the sample and the other side to the gas
that strips analytes that permeate through the membrane. The other implemen-
tation is the use of a hollow-fibre membrane which is, in principle, a piece of
microtubing made of suitable material with the desired permeation characteris-
tics. This structure is naturally self-supporting. Figure 14.3b illustrates the
implementation of this geometry as an MESI sampler. It is a very simple struc-
ture consisting of a hollow-fibre membrane and two connecting tubings, where
the stripping gas flows through the centre of the membrane and the whole out-
side surface area of the membrane is exposed to the sample. In addition to the
simplicity of making hollow-fibre modules, they are characterised by a high sur-
face area to volume ratio, resulting in faster mass transfer compared to a flat
sheet format for the same thickness. The major drawback is the need to use
thicker membranes to facilitate a self-supporting structure. They can only with-
stand a small pressure difference across the membranes. Most common mem-
brane materials such as silicone are available commercially in both flat sheet and
hollow-fibre formats.
It is possible to consider three basic types of mass transfer through mem-
branes: passive, facilitated and active [9]. In the first, most commonly used in
analytical applications, the membrane acts as a barrier through which the com-
ponents are transported under the influence of a gradient in their chemical
potential. In facilitated transport, together with a gradient in the chemical

481
 (a) (b)
Bolt Stripping fluid in , ' Stripping fluid out

Supporting _ - ~
Connection
Wire rmesh tubing

G Z Z , J sheet
Fiat
membrane

Membrane
Teflon
__ __ =~washer

I I L Bottom
rig-~~~ ~plate
Threaded hole
Carrier gas in Carrier gas out

Fig. 14.3. Membrane module designs for (a) flat sheet membrane and (b) hollow-fibre membrane.

potential, a carrier such as a complexing ligand is present in the membrane,

which results in increased permeability for the chosen analytes. Finally, active
transport, which is mainly seen in cellular membranes, is produced against the
gradient of chemical potential of the compounds via a reaction inside the mem-
brane. This last approach for introducing high specificity of mass transport
through the membrane is very interesting, but has not yet been explored by ana-
lytical chemists.

14.2 MEMBRANE SEPARATION INVOLVING SIZE EXCLUSION

14.2.1 Filtration

The filtration process requires a means of applying pressure across the

membrane. The number of molecules passing through the membrane per unit
time is defined by the Hagen-Poiseuille equation:

Jv = -HpA(dP/dx) (14.1)

which defines volume flux Jv associated with pressure gradient dP/dx across the
membrane, where A is surface area of the membrane and Hp is hydrodynamic
permeability defined by the sample viscosity and the resistance of the mem-
brane, which depends on pore size. The influence of pore size on volume flow is
very substantial; for example, the flux obtained for pure water increases about
25 times by an increase in the MWCO value from 10 to 54 kDa, at a constant
membrane area and thickness [10].

482
 Filtration offers some advantages over other size-exclusion approaches such
as a higher speed of separation and analyte recovery. However, it has one major
disadvantage associated with concentration polarisation, which arises due to sol-
ute concentration next to the membrane. The resistance to mass transfer is not
only caused by the membrane itself, but also by a layer which is formed on the
membrane surface by the accumulation of compounds which cannot pass
through the membrane pores. The contribution of this so-called concentration
polarisation layer is dependent on the applied pressure: at low pressures the con-
centration polarisation resistance is negligible (which indicates that no build-up
of retained compounds occurs), whereas at high pressures it accounts for a major
part of the total resistance. For an efficient and constant flux through the mem-
brane it is essential that the concentration polarisation layer be removed from
the membrane as much as possible. This can be achieved by applying a stirring
action during filtration on the sample side of the membrane. More frequently,
however, cross-flow filtration is used. In this case, the sample is pumped through
a flow channel or a hollow fibre and the sample flow itself removes accumulated
compounds from the membrane. Turbulent rather than laminar sample flow
along the membrane is more efficient in removing the polarisation effect. Lami-
nar flows result in zero velocity at the membrane surface, whereas turbulent
flows have nonzero velocities, which facilitate the removal of the accumulated
layer of sample components. In practice, increasing the sample flow-rate and
increasing the diameter of the channel facilitate the generation of turbulent
flow.
Practical filtration is achieved by placing a sample on one side of the
membrane and applying a pressure difference to drive all molecules of appropri-
ate size including the solvent, through the membrane pores to the other side. In
traditional off-line applications with disposable membranes, the driving force is
often applied by a vacuum or a centrifugal force. In on-line filtration the sample
is pumped on the feed side (the name for the donor side in the filtration
processes) of the membrane and a pressure is created by restricting the outlet
tubing of the filtration unit. In analytical applications, the filtrate flow thus
obtained is typically directed to an injection loop, from where it is introduced
into an analytical system. Higher filtration speeds can be obtained by increasing
the membrane surface area, which is most conveniently achieved by using
hollow-fibre membranes instead of planar ones.

14.2.2 Dialysis

In dialysis a sample is placed on the donor side of a porous membrane and analytes
diffuse through the membrane pores to the acceptor side as the result of a concen-
tration gradient dC/dx. The mass flux, Jm in this case is defined by Fick's law:

Jm= -D(dC/dx) (14.2)

where Dm is the diffusion coefficient in sample matrix that is determined by the

483
sample viscosity, the temperature and the molecular dimensions of the analyte
in comparison to the membrane pore size. Equation (14.2) indicates that flux is
dependent upon the surface area and the thickness of the membrane as well as
the diffusion coefficient. In order to maximise the flux and thus obtain a high
analyte recovery, these parameters as well as sample viscosity (because it affects
the diffusion coefficient) should be optimised.
To increase the membrane area, hollow-fibre membranes can be used
instead of planar ones. Typically, both ends of a bundle of hollow fibres are glued
into a fitting and the bundle is immersed into the sample, while the acceptor
phase is present inside the fibres. Since the area available for diffusion is much
larger for hollow-fibre than for planar membranes, the number of analyte mole-
cules recovered per unit time is much higher, which leads to distinctly improved
detection limits. Another important parameter with a distinct influence on the
flux is the membrane pore size or, since pores of varying size exist within a mem-
brane, rather the pore-size distribution. For each application, a proper MWCO
value should be selected so that the interfering material is sufficiently retained,
but at the same time rapid transport of analyte molecules is ensured. In many
cases on-line dialysis is applied to samples of biological origin and the major aim
is to retain proteins and other large biopolymers in order to protect the separa-
tion system designed to separate small molecules. For these applications, mem-
branes with MWCO values of 10-15 kDa are almost exclusively used, since they
have been proven to efficiently remove interfering bio-macromolecules. These
membranes are very rugged and can be used many times before a degradation in
analytical performance. If smaller compounds are to be removed, such as humic
substances from environmental samples, smaller pores are more appropriate
It is very important to use a properly constructed dialysis block and to select
a suitable membrane. It has been shown that analyte recovery can be improved
not only by decreasing the membrane thickness, but also by decreasing the
depth of the donor channel 11]. This indicates that mass transfer resistance not
only takes place in the membrane but also in the boundary layer on the donor
phase side in typical flow conditions. It is crucial to keep the concentration
gradient across the membrane as high as possible. If both the donor and acceptor
phases are kept stagnant, as it is in equilibrium dialysis, then after a while the
analyte concentration gradients extend into these phases resulting in reduced
flux and longer separation times. Maximum recovery of 50% in equilibrium
dialysis can be obtained (for equal donor and acceptor volumes). To improve
recovery, the concentration gradient must be kept higher and this can be
achieved by removing the analytes from the acceptor channel using an acceptor
phase that moves either continuously or in pulses. This continuous dialysis
technique, in principle, allows the quantitative transfer of an analyte from the
sample to the analytical system. However, the moving acceptor phase gives rise
to dilution, which must be overcome by reconcentration of the analytes for trace
analysis. The efficiency of dialysis is decreased when analytes bind to the
membrane material. Both electrostatic and hydrophobic interactions are

484
possible. Low surfactant concentrations, present in the sample below the critical
micelle concentration, decreases the effect.
The application of a dialysis membrane is very attractive in combination
with micro separation technologies. This approach has been used for on-line
extraction of living systems [12]. The coupling of dialysis membranes with CE
was proposed recently [13]. The sample is pumped through the outside of a
hollow-fibre dialysis membrane that connects two sections of a capillary and,
after diffusion of the analytes to the interior, an appropriate voltage is applied to
introduce them into the electrophoresis system. The dialysis membrane allows
not only the coupling of the micro-separation instrument to the investigated
system [14], but can also constitute an interface between two orthogonal
separation techniques resulting in two-dimensional separation [15].

14.2.3 Electrodialysis

The principles of electrodialytic sample preparation have been described in

detail [16]. The transport of charged compounds through the membrane is
caused by two driving forces: a concentration gradient (as in dialysis) and an
electric potential difference. Upon application of an electric potential across the
membrane, charged analytes are actively transported to the electrode in the
acceptor channel, while oppositely charged compounds move towards the other
electrode (and are therefore retained). Neutral compounds only undergo passive
diffusion. In this way, selectivity based on the charge of a molecule is introduced
in addition to selectivity based on its size. The membranes can be neutral or
charged. With neutral membranes, flows higher than conventional dialysis are
achieved, while in the charged ones, selectivity is increased because the
membrane retains the species that possess the same charge.

14.3 MEMBRANE EXTRACTION

The use of microporous or homogeneous extraction membranes is an alternative

to traditional extraction methodology. When hydrophobic microporous mem-
branes are used, the membrane acts as a support that allows contact between the
phases without merging them. With homogeneous membranes, the separation of
analytes is directly related to their transport rate within the membrane phase,
which is determined by their diffusivity and solubility in the membrane matrix.
The main advantages over conventional liquid-liquid extraction are the avoid-
ance of emulsion formation, the lack of the phase separation step and the use of
modules with a high surface-area-to-volume ratio.
Membrane-based extraction techniques offer real chemical selectivity. The
most analytically important and interesting groups of homogeneous membranes
used for extraction processes are those made of polymer and liquid membranes
[17], especially supported liquid membranes (SLMs). Either silicone rubber or a

485
supported liquid can be used as the membrane material and although both
options have the same general characteristics, SLM offers more flexibility
because selectivity of the extraction can be adjusted by selecting an optimal
solvent. In both cases, separation is accomplished in two consecutive extraction
processes between the membrane and the donor phase and between the acceptor
phase and the membrane.
The main advantage of liquid homogeneous membranes, as compared with
polymeric membranes, is the greater transport velocity through them, which is
due to the greater diffusivity of species in a liquid medium. Additionally, it is
easier to incorporate carriers in these membranes with a view to selectively
increasing the permeability of certain species, giving rise to facilitated or coupled
transport processes. However, the membrane lifetime is usually longer for
polymeric membranes. Liquid membranes tend to have solvent leakage, which
becomes more pronounced as the polarity of the solvent increases.
Frequently, the liquid membrane is used to separate two aqueous phases and
the pH of the sample in the donor channels are adjusted to such a value that the
analytes of interest are uncharged and easily extracted into the membrane liquid
or the silicone polymer film. The acceptor phase has the proper pH to effect
ionisation of the analytes immediately after passing the membrane, which
implies that they cannot be back-extracted into the membrane and are trapped
in the acceptor channel. The fundamentals and applications of these and other
similar approaches are presented in Chapter 15.

14.3.1 Fundamentals

The permeation through a membrane is a specific extraction process where the

sorption into and desorption out of the extraction phase occur simultaneously
(see Chapter 9). The sample is in contact with the donor side of the membrane
where extraction into membrane material occurs, while permeated analytes are
removed by the stripping phase from the acceptor side. For membrane extrac-
tion with good flow (agitation) conditions at acceptor and donor sides and effi-
cient stripping, the rate of mass transport through the membrane is controlled
by the diffusion of analytes through the membrane material. Figure 14.4 illus-
trates the concentration profile expected in a system under good convection con-
ditions. The extraction process consists of several steps: (1) mass flux of the
analyte from the bulk sample to the boundary layer outside the membrane sur-
face; (2) diffusion of the analyte through the boundary layer to the membrane
outer surface, a diffusion process; (3) partitioning of the analyte between the
sample matrix and the membrane at the membrane outer surface, a partitioning
process; (4) random movement of the analyte in and through the membrane, a
diffusion process; (5) release and stripping of the analyte by the stripping phase
at the inner surface of the membrane, a partitioning process; (6) diffusion of the
analyte through the stripping boundary layer which is close to the stripping side
of the membrane surface, a diffusion process; (7) mass transfer of the analyte

486
 Membrane Stripping Membrane Sample
z C-4foS---..
----

Be
La
'E
a
o

1
-_ - ~
I ---- I

r g. 1±.4. uiiciuviraliiuli gratleuiLa I

present during membrane extraction. Position

away from the membrane surface by the stripping phase. There are two basic
approaches to membrane extraction. The first is based on quantitative recovery
of analytes and the other involves steady state mass transfer. The exhaustive
approach results in easier quantification since all analytes are removed from a
well defined volume of the sample, while steady state conditions result in higher
sensitivities since the mass transport through the membrane is maximised. In
the next two sections, the impact of different parameters on the extraction per-
formance for each approach is discussed theoretically. For the purpose of this
discussion, we will assume that hollow-fibre geometry is used and the stripping
phase is a gas, but the general conclusion from this discussion can be extended to
other geometries and stripping phases.

14.3.2 Exhaustive extraction

Figure 14.5 shows an example of a module which can be used for quantitative
removal of analytes from a sample. It is constructed of a hollow fibre positioned
in the centre of a glass tube and extending beyond the tube. Glue is used to fix
the membrane in a position and to connect the two tubings. Typically, a sample
is introduced through either end of the hollow-fibre membrane. The stripping
fluid is delivered to the module using the tubings and flows around the fibre.
When the flows are appropriate, the exhaustive transfer of analytes from sample
to gas phase occurs. The value of suitable flows can be found by solving the
appropriate differential equation which describes the mass transfer in the
system illustrated schematically in Fig. 14.6. The aqueous (or gas) sample passes
through the centre of the fibre and the stripping gas phase flows in a counter-
current direction. The appropriate mathematical solution to the equation is
, .......
~. inlet
!Gas
Sample inlet

Gas outlet ' s Eepoxy

Glass tube

Fig. 14.5. Design of the membrane extraction module for exhaustive extraction.

487
 Gas flow

Sampe flow (v) i .. ......

Membrane wall b
Fig. 14.6. Schematic of the hollow-fibre membrane
Gas flow
extraction system. Gas flow

quite complex [18], but can be presented graphically in a dimensionless space of

r/R (reduced radial position in the fibre, where R is inner radius of the fibre),
2
DmL/VR (parameter proportional to the length, L, of the fibre) versus reduced
concentration (see Fig. 14.7). Dm is the diffusion coefficient of the analytes in the
sample matrix and v is a linear flow rate of the sample. The graph shows
concentration profiles generated in the flowing aqueous sample along the hollow
fibre during extraction. The gradients created in the system reflect the presence
of a parabolic flow profile in the system. Velocity of the fluid in the centre of the
tube is highest and at the same time, this is located the largest distance from the
hollow fibre walls where extraction occurs. From this plot, it can be seen that by
increasing the DmL/vR2 parameter, the extraction efficiency improves. There-
fore, as Dm and L increase and v and R decrease, the concentration in the centre
of the fibre decreases, resulting in improved recoveries. The amount of analyte
remaining in the sample can be calculated by integrating the concentration
profiles along the cross-sectional area at the end of the hollow fibre and
considering a parabolic flow profile.
Figure 14.8 compares the extraction efficiency as a function of the volumet-
ric flow obtained using the above model and experimental data for several chlori-
nated hydrocarbons. The agreement is excellent, indicating that indeed the
rate-limiting step in the mass transfer of analytes in the system occurs in the
sample matrix. Figure 14.7 also shows that exhaustive extraction is obtained
only for very low volumetric flows: a few microlitres per minute. The decrease in
the diameter of the hollow-fibre membrane does not help since the linear flow
needs to be substantially increased to compensate for the decrease in the

Fig. 14.7. Dimensionless plot showing analyte concentration distribution inside the hollow-fibre
membrane.

488
 100
90
7 80
t 70
.) 60
60
= so
. 40 - Analycal aoltion
,30 U trichloroethene
V 20 0 tetrachloroethene
Fig. 14.8. Comparison of theoretical W A 1,1,l1trichloroethan
modelling and experimental 10
extraction efficiency data for flow ° I 40 80 120 160
through module system. Volumetric flow [pUminl

cross-sectional area. The only way to improve mass transfer is to generate a tur-
bulent component in the sample flow through the hollow fibre, for example, by
coiling the membrane. However, it is very difficult to generate substantial tur-
bulence in such a small diameter tube. The above discussion clearly demon-
strates that, although it is feasible to reach exhaustive extraction, the sensitivity
obtained in the process is low since the amount of sample extracted under the
exhaustive conditions is very low. It is possible to increase the time of extraction
and/or use multi-fibre modules to obtain higher sensitivities, but both
approaches lead to the greater use of the stripping phase and difficulties in refo-
cusing analytes. This exhaustive approach is more effective when extracting gas
samples, since the diffusion coefficient in gas is higher compared to liquid, how-
ever, good metering pumps will always be required since the extraction recovery
is dependent on the flow conditions.

14.3.3 Maximum rate extraction

The hollow-fibre membrane module used in steady-state membrane extraction

that provides the maximum extraction rate has a very simple open structure and
is illustrated in Fig. 14.3B. The stripping gas (acceptor phase) flows through the
inner membrane and the outer surface of the membrane is in direct contact with
the well-agitated sample matrix. In this case, the mass transfer is controlled by
analyte diffusion through the membrane material. The appropriate equations
describing the mass transport rates as functions of extraction parameters have
been derived by solving the diffusion equation for the membrane geometry and
boundary conditions. Details of the derivation of the equations have been
described [19]. Figure 14.9 shows a comparison between theoretical and experi-
mental results. Curve A shows the theoretical prediction of the extraction time
profile for benzene. This profile indicates that the extraction is composed of two
permeation processes: non-steady-state permeation and steady-state perme-
ation. The term non-steady-state permeation refers to the process of forming the
concentration gradient of analyte in the membrane soon after the system is
exposed to the sample. The term steady-state permeation refers to the process

489
 a: model prediction

b: expermental result
0.2

Fig. 14.9. Comparison of theoretical and experi- ction time profi (benzene)
mental transients and steady state extraction rate X
for direct exposed membrane configuration 0 50 100 150 200 250 300
system. ime (sec)

after the concentration gradient is formed, when a constant permeation through

the membrane wall is reached. In curve A, the positive slope corresponds to the
non-steady-state permeation process and the flat line corresponds to the steady-
state permeation process. An experimental extraction time profile was obtained
and is shown as curve B. When comparing curves A and B, it can be seen that the
theoretical prediction and the experimental results are close. Steady-state time
was considered to be when the signal intensity reached 90% of the steady-state
permeation signal. The experimental result shows the steady-state time as 66 s
and the theoretical prediction as 62 s.
At steady-state extraction of well-agitated samples, the amount extracted, Z,
can be expressed as [1]:

Z ACsDKes,,t (14.3)
ADe/e5 +2Rln R+b
Q R
where R and b are the inner radius and the thickness of the hollow-fibre
membrane, respectively; A is the membrane inner surface area; Cs is the analyte
concentration in the sample; De is the diffusion coefficient in the membrane; fis
the ratio of the average concentration of the stripping gas and the analyte
concentration in the stripping gas at the exit of the membrane (constant at
steady-state conditions); Ke, is the membrane/air partition coefficient; Q is the
stripping gas flow rate; and t is the trapping time.
This equation includes analyte and membrane physico-chemical properties
reflected in the distribution constant between the membrane and gas phase and
the diffusion coefficient of analyte molecules in the membrane. Equation (14.3)
illustrates that steady-state transport can be calculated for given geometric
configurations and experimental conditions as long as K and D are known. These
values can be stored in the computer or calculated based on a theoretical formula
or even determined experimentally as illustrated for the MESI system discussed
latter. Equation (14.3) also indicates that several experimental parameters
affect the extraction efficiency. A longer membrane probe leads to a higher
extraction rate if the stripping gas flow rate is high enough. Membrane thickness
is another factor affecting extraction rate and the membrane response time for a
given analyte. A thinner walled membrane would be beneficial for this applica-
tion. Temperature is an important factor in the extraction, affecting both the

490
distribution constant and the diffusion coefficient of an analyte. The effects are
opposite for these two parameters. A relatively low extraction temperature
results in a higher extraction rate.
In summary, the mass transfer in the open configuration illustrated in Fig.
14.3b is much faster than in the closed module design from Fig. 14.5 since
efficient agitation of the sample can be easily generated. The calibration of the
exhaustive extraction method (Fig. 14.5) is simpler since it requires only infor-
mation about the volume of sample passing through the module; however, the
use of Kes and De as a basis for calibration of the open configuration (Fig. 14.3b)
eliminates the need for external calibration of this system, making it as attrac-
tive. The temperature should be monitored during the experiment and results
corrected for its change.
The calibration approach described above for well agitated clean homoge-
neous samples can also be extended to quantification of volatile analytes in more
complex aqueous and even solid matrixes by the headspace membrane arrange-
ment, as long as the gas/matrix distribution constant (Henry constant for aque-
ous and solid samples) is known. This approach to calibration is analogous to one
recently described for the SPME technique (see Eqs. 13.7 and 13.18) [20]. Practi-
cally, the headspace membrane extraction arrangement can be implemented by
using the "cap" sampling system (see Fig. 14.10) [21]. To increase mass transfer
through the headspace created by the cap, helium can be injected into the cap.
Also, a micro-fan can be installed in the cap to enhance the mass transfer of
analytes through the headspace. The extraction cap can be placed at the water
surface or at a required depth to perform on-site or on-line monitoring. In this
arrangement headspace extraction is used and the membrane does not come in
contact with the sample matrix directly. This reduces the possibility of mem-
brane contamination and mechanical damage, which prevents a decrease of
membrane performance with time. This ensures that the extraction probe can be
placed in the sampling environment for long-term monitoring without the need
for replacement. This technique can also be extended to solid sample analysis by
covering for example soil, by the "cap". Figure 14.11 shows the range of linear
hydrocarbons that are able to permeate through a 0.15-mm thick PDMS hol-
low-fibre membrane. As expected, the relative permeation rate initially
increases with the increase in the molecular weight of the compound due to
increase in the sample matrix-PDMS distribution constant. The diffusion coeffi-
cient is at the same time decreasing, but this effect is much smaller. However,
above C-12 there is a dramatic decrease in the amount of extracted analytes
since the stripping of semivolatile compounds is not very efficient at room tem-
perature. The headspace approach to membrane extraction can be extended to
less-volatile analytes by heating the sample, the membrane, and the transfer
lines. By increasing the membrane temperature from room temperature to 80°C
hydrocarbons up to C-16 can be transmitted through the same membrane. An
alternative approach is to periodically heat the membranes since this process
cleans the system and produces concentration injection pulses of the semiv-

491
 I-

mbrane
ule .>

8
6 7 8 9 10 11 12 13 14 15 16
arhn nllmhr

Left: Fig. 14.10. Cup membrane extraction system.

Right: Fig. 14.11. Relative steady state membrane extraction rate for range of n-alkanes.

olatile compounds accumulated in the membrane material into the analytical

instrument [22]. This, however, can be better accomplished by using a micro-
sorbent trap that follows the membrane because of better refocusing and
sharper injections. This approach is discussed in the following section.
Figure 14.10 illustrates the cap extraction system used in combination with
the flat sheet membrane module. It is frequently useful to consider these struc-
tures since the selection of commercially available flat sheet membranes is con-
siderably larger compared to that of hollow-fibre membranes. In addition, it is
much more convenient to make very thin flat sheet membranes, for example by
spin coating, compared to tubular structures. Externally supported flat sheet
membranes can withstand a much higher-pressure differential between donor
and acceptor phases than the self-supporting hollow-fibre membranes. These
facts, as well as the faster response time and shorter memory effect attainable
for thin membranes make them very attractive. A membrane module can be
designed very simply, as illustrated in Fig. 14.3A. Carrier (stripping) gas is sup-
plied to the module through a small diameter PTFE tube, fitted tightly into the
PTFE washer. The pressure of the stripping gas lifts the membrane, allowing
free passage of the gas to the outlet tube. A fine stainless steel mesh, mounted in
a stainless steel ring, supports the membrane. The ring is screwed to the bottom
plate of the module with three bolts. The body of the module is made out of stain-
less steel and has a sandwich-like design. A Teflon washer is inserted between
the two stainless steel pieces,. Teflon is relatively inert and because it is soft, it
also helps to seal the membrane eliminating the need for an o-ring. There is no
need to use glue or other sealant.
Membrane extraction can be used in combination with supercritical fluid
extraction [23]. This approach uses hollow-fibre membranes contained in a
similar module to that showed in Fig. 14.5. The casing of the unit is made of
metal rather than glass tubes to enable it to withstand high pressures. The key
to operating such a system is to ensure that no pressure differential exists
between sample inside the hollow fibre and the stripping fluid outside the fibre,
eliminating the possibility of damage to the membrane. This is accomplished by
using appropriate flow restrictions.

492
14.4 MEMBRANE EXTRACTION WITH SORBENT INTERFACE (MESI)

The MESI approach to membrane extraction was developed to address the need
for enhanced sensitivity in membrane extraction and simple continuous opera-
tion. To efficiently transfer analytes through the membrane, the stripping gas
flows should be high to prevent the accumulation of analytes in the membrane.
This process results in the dilution of analytes. As Fig. 9.14 (in Chapter 9) indi-
cates, the concentration of analytes in the stripping gas is low at the optimum
permeation rates. The use of high sensitivity detectors, such as mass spectrome-
ters, allows effective on-line determination of the diluted sample components.
However, to achieve high sensitivities, refocusing of analytes prior to detection
would be beneficial. In addition, some analytical instruments, for example gas
chromatographs, require pulse sample introduction, which can be generated by
a sorbent interface present during desorption.
Figure 14.12 illustrates that the MESI process consists of two steps: select-
ive permeation and trapping of extracted analytes by the sorbent, and then
desorption of accumulated components from the trap into an analytical instru-
ment. In the trapping mode, the sorbent is kept at room temperature, or if a
cooler is used the temperature can be considerably lowered to increase the trap
capacity. The power supply is off, and the carrier gas is free of analytes after
passing over the sorbent. Therefore, only pure gas enters the instrument.
During the desorption step, the sorbent is rapidly heated and the accumulated
analytes are released to enter the instrument as a high concentration narrow
pulse, which facilitates high sensitivity determinations. In addition, use of a
sorbent on line with membrane extraction eliminates the need for valves to
produce pulse injection into the analytical instrument.
The sensitivity of the MESI system in the trapping mode is directly related to
the trapping time-the longer the trapping time, the more analytes are accumu-
lated at the sorbent interface. Trapping time is limited by the breakthrough time

11

carrier

Fig. 14.12. Two steps in

the MESI process. Desorpi signal

493
 Sorbentrap

Fig. 14.13. PDMS sorbent trap with solid-state

cooling. Heat sin

of the analytes through the sorbent interface. To increase the breakthrough

time and enhance the sensitivity of the MESI system, a large sorbent volume can
be used and/or the trapping temperature can be reduced. An increase of the
sorbent volume often leads to an increase of the thermal capacity of the sorbent
trap which determines the width of the desorption pulse. For rapid desorption a
small sorbent trap is preferable. Alternatively, the analyte breakthrough time at
the sorbent trap can be increased by decreasing the trapping temperature. The
cooling can be accomplished by using a coolant such as liquid nitrogen or solid
carbon dioxide or by using semiconductor devices, such as Peltier coolers. Figure
14.13 illustrates the design of such a trap. PDMS sorbent is located in the fused
silica capillary surrounded by the heating wire and sandwiched between two
Peltier devices. The desorption is accomplished by passing the current through
the wire for a short time and heating the trap. PDMS is a very good sorbing
material which produces quantitative desorption of accumulated analytes dur-
ing the desorption pulse. However, its capacity is low and it requires cooling for
efficient sorption of more volatile components.
Alternatively, solid sorbent can be used, which does not require cooling and
is therefore suitable for the design of small trapping units, such as the one illus-
trated in Fig. 14.14. It is designed using a 6 cm long deactivated stainless steel
tubing (having an outer diameter of 0.75 mm) packed with about 0.0019 g of
XAD-2 resin [24]. The packing is about 1 cm long. Using stainless steel tubing
instead of deactivated fused-silica tubing, the need for a heating coil is elimi-
nated since the heating current can pass directly through the tubing. The pack-
ing material is immobilised in the middle of the stainless steel tube, by placing
glass wool and sometimes squeezing the tube at the ends of the packing. When
- Power supply-

Fig. 14.14. Design of solid sorbent trap with direct heating. 1, Sorbent material; 2, quartz wool; 3,
gold-plated ferrule; 4, Valco connector; 5, stainless steel tubing connecting the membrane
module to the sorbent trap; 6, stainless steel tubing connecting the sorbent trap to the GC
column; 7, electrical connection to power supply.

494
an electrical pulse is applied to the trap, the gas inside expands very quickly and
the analytes can be pushed back partially towards the carrier gas lines instead of
going directly into the analytical instrument, which is placed at the opposite end
of the sorbent trap. The injection band would be broadened in this case, and the
separation would be worsened. To eliminate this effect, the stainless steel tubing
of very low i.d. is used to connect to the carrier gas lines at the entrance to the
trap producing flow restriction and, therefore, preventing back flush. With this
configuration, when a heating pulse is applied, the carrier gas, following the less
restricted path, flows mainly towards the instrument, moving the analytes in
this direction.

14.4.1 MESI-GC

An interesting implementation of the MESI sample preparation system is in

combination with gas chromatography. There are many variations of this
concept, but they all have similar characteristics [25-29]. The MESI-GC system
uses a membrane module, a sorbent interface, a capillary gas chromatograph
and a data acquisition system as illustrated in Fig. 14.15. This instrumental
concept is capable of performing all steps of the analytical process directly in the
field [30]. The membrane module represents the sampling part of the system.
The volatile compounds from the sample permeate the membrane, which can be
in direct contact with the sample or exposed to its headspace. The membrane
represents a barrier between the sample and the carrier gas that flows through
the MESI components and usually has a nonpolar character, thus keeping water
from entering the system. It also acts as a selective element, since permeation
rates of different molecules vary with membrane material. During sampling, the
analytes diffuse in the membrane along the concentration gradient created
between sample and the stripping phase. Once they reach the inner side they are
stripped by the carrier gas and directed to the sorbent interface. The membrane
is typically nonporous silicone material. Hollow-fibre membranes are often used,
(as shown in Fig. 15.15) presenting the advantages of being self-supported and
easy to connect to the carrier gas line. However, their relatively thick walls (over
100 gm) give long response times and long-lasting memory effects. On the other
hand, very thin flat sheet membranes are available, but since they are not

Injection signal Detector signal

MUemrane -Colurmn
t
probe

Sorbent
interface

TOC profile TOC profile

Fig. 14.15. Schematic representation of the TOC profile TOC profile
MESI-GC system. C

495
Fig. 14.16. Chromatogram obtained
from the MESI system with prog-
ressively longer trapping times (30 s
increments). The bottom trace
corresponds to a trapping time of 1
min, and the top trace corresponds
to 3 min. Peaks: 1.benzene: 2 tolu- t
-U.U3m Pek 1 e . 1.0 L.U .3 . 3.0
ene; 3, m, p-xylenes 4, o-xylene. Minutes

self-supported they cannot be connected to the gas lines without the use of
special holders (see Fig. 14.3A).
Once the analytes cross the membrane, the carrier gas stream carries them
away to the sorbent interface, where concentration occurs. The sorbent interface
includes a sorbent trap, a heating means and a power supply as discussed above.
Figure 14.16 presents a set of five chromatograms obtained with the previ-
ously described MESI system, composed of the membrane module illustrated in
Fig. 14.3a and a trap as in Fig. 14.14, using progressively longer trapping times
(30, 60, 90, 120 and 180 s) for a dynamically generated standard gas mixture.
The bottom trace corresponds to a trapping time of 1 min, and the top trace to 3
min. The first peak in each group corresponds to traces of water and air, which to
some extend also permeate through the membrane and are trapped by the
sorbent. The remaining four peaks in each group correspond to benzene,
toluene, m,p-xylene and o-xylene (from left to right). It is clear from Fig. 14.15
that within the time frame examined, the response of the system was propor-
tional to the trapping time. For example, the benzene peak height for 1 min
trapping time was -0.5 V, while for 3 min trapping time it was 1.5 V. This proves
that no breakthrough of analytes occurred in the trap. Each individual separa-
tion was completed in -25 s, and the peak shapes were satisfactory. The width of
the benzene peak at half height was -250 ms, which was very good taking into
account that the desorption of the analytes was carried out without reversing
the direction of the carrier gas flow through the sorbent trap. This proves that
the MESI system in the configuration examined can be successfully used for
semi-continuous monitoring of trace levels of volatile analytes.
The most important aspect of MESI is its ability to selectively concentrate
volatile analytes on-line, before fast separation. This can result in dramatically
improved sensitivity, as illustrated in Fig. 14.17. The lower trace in this figure is
a MESI chromatogram obtained for a standard BTEX mixture and micro GC
equipped with a low sensitivity universal thermal conductivity detector (TCD
[31]. The trapping time was 1 min. The scale for this chromatogram is presented

496
 A r
4.1

4.
5.0

3.

.5
2.5

2.0

1.5 0.0
Minutes

Fig. 14.17. Comparison of results obtained using the MESI system with those obtained using
regular gas-phase injection through a built-in injector. The lower chromatogram (lefty axis) has
been obtained using the MESI system; the upper trace (righty axis) is a chromatogram for direct
gas injection. BTEX concentration -4 ppm (v/v) of each compound. Peaks: 1, benzene; 2,
toluene; 3, ethylbenzene; 4, m,p-xylenes; 5, o-xylene.

on the lefty axis. The upper trace is a chromatogram obtained for the same mix-
ture with a regular micro-sampling loop injection, using the second GC module
in the same instrument. This chromatogram is presented in a 100 times larger
scale (see right y axis). It is clear from Fig. 14.17 that even with a precon-
centration time of only 1 min, the sensitivity of MESI was higher by more than
two orders of magnitude compared to direct injection of a gas sample. The use of
a selective membrane results in the elimination of a sloping baseline noticeable
in the upper trace associated with the presence of large amounts of oxygen and
moisture in the loop injection system. It has been demonstrated that use of
MESI coupled to a micro-GC allows trace analysis with a low sensitivity TCD
detector. For example, chloroform in drinking water present at low ppb level has
been detected by this system (see Ref. [31]).
Calibration is an important issue in MESI applications since it uses a
non-exhaustive extraction method. External calibration shows good precision
and wide linear range [32] in MESI similar as has been demonstrated before for
MIMS [33] systems. However, for field application, because of the demands of
fast, simple and accurate monitoring, traditional calibration methods such as
internal and external calibration are not appropriate choices, and in some cases
they are not applicable. Alternatively, an on-line calibration method can be
based on the mathematical model that was introduced in a previous section of
this chapter. Equation (14.3) can be rearranged to express the relationship
between the concentration in air, Cs, and other parameters:

C
c, Zt ( Qf 2Rln((R +b)IR)
ADK
(14.4)
(14.4)

497
 In Eq. (14.4), R, b, A and Q are constant. The trapping time (t) can be
experimentally fixed. De and Kes are constant if the extraction temperature does
not change and assuming low sample concentration. Kes and De values are
available in the literature or can be measured on site. Parameter f is a constant
when extraction conditions such as stripping gas flow rate and extraction
temperature, are constant. Therefore, Eq. (14.4) can be simplified to:

Cs = BZ (14.5)

where the constant

i( f 2Rln((R+b)/R
t Q ADKe
5

and can be calculated using experimental dimensions and parameters. Equation

(14.5) shows that at steady-state extraction, the concentration in a sample is
proportional to the amount extracted. In other words, if the amount extracted is
known, the concentration in sample can be calculated. The amount extracted Z,
can be obtained from the analyte peak area and the response factor. Thus no
external calibration is need.
As discussed above, the model indicates that calibration under good convec-
tion conditions can be performed based on the diffusion coefficient and distribu-
tion constant related to the membrane material and a given analyte. These
values can be stored in the data analysis unit or can be determined experimen-
tally on-line. The advantages of performing calibration on-line include compen-
sation for changes in membrane properties with time, quantification of
unidentified compounds, and even possible identification of unknown analytes
based on their distribution constants and diffusion coefficients.
The above discussion is true if extraction conditions are stable. In some cir-
cumstances, however, conditions are not stable, mainly because of temperature
variation. Temperature variation results in changes in the stripping gas flow
rate, the size of the membrane probe, but most importantly in changes of Kes and
De values. Figure 14.18 shows the chromatogram obtained during continuous
monitoring of BTEX during a temperature change from 97 to 25°C. As the tem-
perature decreases the amount of analytes extracted increases. This effect might
be a little surprising, but can be explained by the fact that the increase in the
membrane-sample distribution constant for a given analyte is much larger than
the decrease in the diffusion coefficient in the membrane material. Obviously in
this case, Ke and D. values calculated at room temperature should not be used,
as this would cause a large calibration error. Therefore, in on-site applications,
the temperature should be monitored and Ke and D. values corresponding to
appropriate temperatures should be used to ensure the calibration is correct.
The appropriate Kes and De values can be stored in the computer for on-site use.
There are various strategies, which can be used to determine Kes and De val-
ues on-line during a MESI-GC measurement [34]. Probably the most interesting

498
 97

I-
25

___ _
32151.0

, 25720.8

= 19290.6
1~~~~~~~~
i
.6 12860.4

L
6430.2 I

n00 .I- Ul, j

0.00 12.00 24.00 36.00 48.00 60.00
Retention Time (min)

Fig. 14.18. (A) Plot of the temperature profile during the experiment. (B) Monitoring chromato-
gram of BTEX extraction during the temperature change. 1, Benzene; 2, toluene; 3, ethyl-
benzene; 4, o-xylene.

involves pulse heating of the membrane and observing the transients generated
by the system.
Figure 14.19 shows a chromatogram obtained for benzene by membrane
pulse heating. The two highest peaks, 1 and 2, correspond to two heating pulses,
hence thermal desorption from the membrane. The two peaks are the same
height, which means that a constant amount was desorbed. The amount of
analyte desorbed from the membrane material into the stripping gas is repre-
sented by the difference between the intensity of peak 1 (or peak 2) and the
average intensity of the peak obtained at steady state. Curve a is a smooth plot of
each peak height from non-steady state to steady state. The non-steady state
transient obtained after a heating pulse, as shown in curve b, is identical to the
transient associated with curve a corresponding to the exposure of the
membrane to the sample. This transient can be used to calculate the diffusion

Fig. 14.19. Monitoring chromatogram of

benzene during probe pulse heating ex-
periment. Concentration of benzene: a
12.7 g l-l; membrane probe length: 4
cm; flow rate of stripping gas: 2.2 ml
min'; extraction temperature: 25°C;
pulse voltage on the membrane probe
and sorbent interface: 38.6 V; pulse
width: 1 s; trapping temperature at 000 _ : -
sorbent interface: -40°C; trapping time: 0.00 2.00 4.00 6.00 8.00 10.00
40 s. Retention Time (min)

499
 5:50 PM - 'JL 3:50 P '-God1 methanol
2 acetone
3:30 PM __J L L ' 3 propanol
4. butan-2-ol
5. 1I 1-trichloroethane
6 benzene
1:30 PM ul 7 toluene
11:50 AM
9:30 AM -
Fig. 14.20. Laboratory air monitoring with 0.5 1 1.5 2 2.5 3 3.5 4 4.5
MESI-GC. Time (min)

coefficient in the membrane. In addition, the height of the desorption peak is

related to the amount of the analytes in the membrane material, and therefore it
is proportional to the membrane-sample matrix distribution constant and can be
used to calculate its value. This approach to calibration has been tested experi-
mentally and good agreement has been found as reported in Ref. [34].

14.4.3 Application of MESI-GC

MESI-GC has been used for analysing volatile organic compounds in various
samples. Environmental monitoring has been demonstrated for tap water and
various types of air. Figure 14.20 illustrates the results of air quality monitoring
in a research laboratory. It is possible to notice an increase in air contamination
as a result of increased research activity during the day. In particular, the
presence of toluene is detected at 1:30 p.m. associated with the use of the Soxhlet
extraction system.
Monitoring of human breath by MESI could become a very powerful tech-
nique for clinical chemistry and toxicology since the last part of the breath (end
tidal volume) represents the headspace of blood [35]. Figure 14.21 illustrates the
acetone measured in breath over a 4.5 h period. During this time, the subject ate
various amounts of food. Acetone is used to monitor diets and diabetic patients.
---
OUu,

a 400
0
to
300

a 200

100
Fig. 14.21. Breath acetone
profile recorded over a 4.5 h
period with MESI-GC-Ion 0 20 40 90 180 210 240 270
Mobility Spectrometer. Time (min)

500
 0
0)
C

C.
a

To

q
C
C)
a
Fig. 14.22. Breath ethanol
profile with MESI-GC after
application on the skin of 1 ml 5
of vodka. Time (min)
700

650
C-~----*
600

C 550
I~~~`-----~"-

2 500

t 450

o 400
Fig. 14.23. Monitoring of 35o
breath for alcohol during
application on the skin of 1 ml 3 20 40 60 80 100 120 140
of vodka every 5 min. Time (min)

The correlation between the level of the acetone in the breath and the amount of
food ingested is showed in Fig. 14.20. After lunch, the concentration of acetone
in the breath drops, and it begins increasing again after 4 h followed by another
drop after a snack. It can be seen that the concentration of acetone in breath is
directly related to the timing of food consumption.
Exposures monitoring with MESI system are illustrated in Figs. 14.22 and
14.23. In this experiment alcohol was used as a chemical marker to visualise the
capabilities of the MESI-GC approach for this application. 1 ml of vodka was
applied on the skin. The collection of the breath started 3 min after the applica-
tion. One breath sample was analysed each time, using a trapping time of 4 min.
It can be seen from the figure that the ethanol travelled rapidly through the body
to reach the breath. The profile is very sharp, with the concentration dropping
rapidly after the first sampling. Different vodka application time intervals were
used, and the concentration of ethanol was measured in the breath. It has been
found that by applying 1 ml of vodka every 5 min, the concentration of ethanol
maintains constant over a long period of time (see Fig. 14.23). Samples were
recorded every 15 min, using 4-min trapping times and one breath per sample.
The above examples illustrate the flexibility of the membrane-sorbent sys-
tem in combination with GC. This instrumental approach results in a single
device being able to perform sampling, sample preparation, separation and
quantification and therefore it enables GC to act as a sensor for environmental

501
and clinical monitoring. Such applications are expected to be used widely in the
future, when small size separation micro-machining devices become available.

REFERENCES

1 M. Mulder, Basic Principlesof Membrane Technology. Kluwer, Dordrecht, 1991.

2 P.M. Bungay, H.K. Lonsdale and M.N. de Pinho (Eds.), Synthetic Membranes-
Science, Engineeringand Applications. Reidel, Dordrecht, 1986.
3 S.A. Stern, Membrane Separation Technology. Elsevier, Amsterdam, 1995.
4 N.C. van de Merbel, J. Chromatogr.A, 856 (1999) 55.
5 J-A. Jonsson and L. Mathiasson, J. Chromatogr.A, 902 (2000) 205.
6 B. Montero Cordeno, J.L. P6rez Pav6n, J.C. Garcia Pinto, M.E. Fernandez Laespada,
R. Carabias Martinez and E. Rodriguez Gonzalo, J. Chromatogr.A, 856 (2000) 195.
7 I. Pinnau and B.D. Freeman (Eds.), Membrane Formation and Modification,, ACS
Syposium Series 744, ACS, Washington DC, 2000.
8 E. Plein, Affinity Membranes: Their Chemistry and Performance in Adsorptive
Separation Processes.Wiley, New York, 1991.
9 P. Meares, in: P.M. Bungay, H.K. Lonsdale and M.N. de Pinho (Eds.), Synthetic
Membranes-Science, Engineeringand Applications. Reidel, Dordrecht, 1986.
10 N.C. van de Merbel, I.M. Kool, H. Lingeman, U.A.Th. Brinkman, A. Kolhorn and L.C.
de Rijke, Chromatographia,33 (1992), 525.
11 M.M.L. Aerts, W.M.J. Beek and U.A.Th. Brinkman. J. Chromatogr.,500 (1990) 453.
12 Y. Song and C.E. Lunte, Anal. Chim. Acta, 400 (1999) 143.
13 L. Bao and P.K. Dasgupta, Anal. Chem., 64 (1992) 991.
14 J. Wu, J. Pawliszyn, Anal. Chem., 67 (1995) 2010.
15 C. Tragas and J. Pawliszyn, Electrophoresis,21 (2000) 227.
16 J.P. Brun, Procdds de SeparationparMembranes. Hasson, Paris, 1989.
17 J-A. Jdnsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 318.
18 K.F. Pratt and J. Pawliszyn, Anal. Chem., 64 (1992) 2101.
19 Y.Z. Luo, M. Adams and J. Pawliszyn, Anal. Chem., 70 (1998) 248.
20 A. $araullo, P. Martos and J. Pawliszyn, Anal. Chem., 69 (1997) 1992.
21 Y. Luo and J. Pawliszyn, Anal. Chem., 72 (2000) 1058.
22 G. Matz, G. Kibelka, J. Dahl and F. Lennemann, J. Chromatogr. A, 830 (1999) 365.
23 M. Yang and J. Pawliszyn, Anal. Chem., 65 (1993) 2538.
24 T. Gorecki and J. Pawliszyn, LC-GC International, 12 (1999) 123.
25 K.F. Pratt and J. Pawliszyn, Anal. Chem., 64 (1992), 2107.
26 M.J. Yang and J. Pawliszyn, Anal. Chem., 65 (1993) 2538.
27 S. Mlitra Mitra, L. Zhang, N. Zhu and X. Guo, J. Chromatogr. A, 736 (1996) 165.
28 M.J. Yang, S. Harms, Y.Z. Luo and J. Pawliszyn, Anal. Chem., 66 (1994) 1339.
29 B. Hauser, P. Popp and A. Paschke, Int. J. Environ. Anal. Chem., 74 (1999) 107.
30 J. Pawliszyn, Trends Anal. Chem., 14 (1995) 113.
31 A. Segal, T. Gorecki, P. Mussche, J. Lips and J. Pawliszyn, J. Chromatogr., 873
(2000) 13.
32 Min Yang, Membrane Extraction with Sorbent Interface, Ph.D. Thesis, University of
Waterloo, 1995.
33 T. Kotiaho, F. Lauritsen, T.K. Choudhury and R.G. Cooks, Anal. Chem., 63 (1991)
875A.
34 Y. Luo and J. Pawliszyn, Anal. Chem., 72 (2000) 1064.
35 M. Phillips, Sci. Am., 267 (1992) 74.

502
Chapter 15

Liquid membrane techniques

Jan Ake J6nsson

15.1 INTRODUCTION

In recent years a number of membrane techniques for sample preparation have

been suggested [1-8]. Membrane techniques can provide several clear advant-
ages over other extraction techniques such as LLE or SPE, especially regarding
selectivity, enrichment power and automation potential.
In all types of membrane extraction, the membrane separates the sample
phase (often called donor or feed solution) from the acceptor or strip phase and
the analyte molecules pass through the membrane from the donor to the
acceptor, a process sometimes called pertraction (permeation-extraction).
Membrane extraction techniques can be divided into two main categories:
porous and non-porous membrane techniques. Another distinction is between
one-, two- or three-phase membrane extraction techniques.
A typical example of a one-phase technique is dialysis. The membrane in
one-phase membrane systems is porous, so there is a liquid (or gas) contact
through the pores between the donor and acceptor phases, which are of similar
chemical composition (i.e., both are either aqueous, organic or gaseous). There is
no phase boundary and therefore no partition between phases. Physical rather
than chemical properties govern the process. This chapter will not deal further
with one-phase systems; for information on dialysis, especially its analytically
interesting version microdialysis, the reader is referred to the literature [9,10].
Non-porous membrane techniques utilise two or three different phases
separated by distinct phase boundaries. In three-phase membrane systems, a
separate membrane phase is surrounded by two different phases (donor and
acceptor) forming a system with two phase boundaries. This provides possibili-
ties for two different extraction (partition) steps that can be tailored to different
types of chemical reactions, leading to a high degree of selectivity. The mem-
brane phase can be a liquid, a polymer or a gas, and the donor and acceptor
phases can be gaseous or liquid (aqueous or organic).
In two-phase membrane systems one of the surrounding phases is the same
as the membrane phase, so there could be for example an organic solvent in the
membrane and the acceptor, while the donor is aqueous or gaseous. There is only

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 503
one phase boundary and consequently only one partition reaction (often
analogous to liquid-liquid extraction in separation funnels).
By leaving an extraction system for a long enough time, equilibrium between
the phases is reached. Usually the aim in sample preparation in analytical
chemistry is to transfer as much of the analyte as possible from the donor to the
acceptor. To improve this recovery, there are in principle two methods: the
acceptor can be continuously changed, usually flowing past the membrane, so
extracted molecules are removed from the membrane surface by convection.
This typically leads to a dilution of the analyte. Alternatively, and more
efficiently, the analyte molecules can in many cases be trapped in the acceptor by
a chemical reaction or simply because of a high partition coefficient. With
efficient trapping there is the potential to obtain high enrichment factors. To
improve the overall amount of analyte extracted, a flowing donor is often used:
i.e., the sample is pumped past the donor side of the membrane.
This chapter will mainly cover extraction systems with liquid membrane
phases and trapping in the acceptor. Both two- and three-phase systems are
considered. These types of system have been successfully applied to a number of
analyte and sample types and can be important alternatives to more orthodox
approaches to sample preparation.

15.2 NON-POROUS MEMBRANE EXTRACTION TECHNIQUES

15.2.1 Liquid membrane extraction techniques (SLM and MMLLE)

A common type of a liquid membrane format is the supported liquid membrane

(SLM). Here, the pores of a porous hydrophobic polymer "membrane" are filled
with an organic liquid, which is held by capillary forces. This liquid in the pores
then provides a separate phase (although dispersed) between the donor and
acceptor. Typical solvents in this context are long-chain hydrocarbons like
n-undecane or kerosene and more polar compounds like di-hexyl ether, tri-octyl
phosphate and others. With such liquids, a liquid membrane can be stable from a
few days to a few months. Different additives to the organic phase can be used.
This can increase the efficiency and selectivity of extraction considerably,
possibly with some decrease in membrane lifetime. SLM extraction is chemically
analogous to LLE, extracting from an aqueous sample into an organic solvent,
followed by a "back extraction" of the analytes in the organic phase to a second,
different, aqueous phase.
A version of the SLM uses gas (air) as the membrane phase, i.e. the pores in
the hydrophobic support are not filled with any liquid. This can be used for the
extraction of volatile compounds like amines [11,12], and will not be much
discussed here.
The use of SLM in sample preparation in analytical chemistry was suggested
by Audunsson [13] and the field has been reviewed several times [2,3,6,8,14].
There are a number of examples where SLM extraction has been used for

504
industrial separations, for example extraction of metal ions [15,16] and organic
acids [17,18] from wastewater. Also, extraction of large polyelectrolytes as
lignosulfonate [19] and proteins [20] was described.
There are a number of other ways for arranging a three-phase liquid
membrane extraction; e.g., as a bulk liquid between semipermeable membranes
or in completely unsupported forms. These types are commonly called bulk
liquid membranes (BLM). This principle leads to slow mass transfer and long
extraction times. Membrane extraction can also be arranged in the form of an
emulsion-an emulsion liquid membrane (ELM) [16,21]. The ELM is useful for
removal of compounds from the feed solution, but the quantitative recovery of
the extract is difficult. The techniques mentioned are therefore not used in
analytical chemistry, but extensively in industrial processes and in fundamental
membrane extraction studies.
A two-phase membrane extraction system with a porous membrane
(support) that separates one aqueous and one organic phase is called micro-
porous membrane liquid liquid extraction (MMLLE) [22]. In principle there are
two approaches: with a hydrophobic or a hydrophilic porous membrane. Using a
hydrophobic membrane, the organic liquid fills the pores and a direct contact
between the phases is obtained close to the surface of the membrane where the
interfacial mass transfer takes place. This is a liquid membrane system where
the liquid membrane is the organic liquid in the pores of the porous membrane
support, in analogy with SLM. The membrane could also be hydrophilic, which
would lead to an aqueous phase in the membrane pores, but this does not yet
seem to have been tried for analytical purposes.
In MMLLE, virtually the same extraction chemistry can be obtained as with
LLE and it can be seen as a way of performing LLE in a continuous and
instrumental way. There are other, more instrumentally complicated ways of
performing continuous LLE, as described in the literature [23]. These typically
involve mixing the organic and aqueous phases as segments in a flow system,
and then sequentially separating these phases again, the latter operation being
quite critical.

15.2.2 Polymeric membrane extraction techniques

Using a polymeric membrane, such as a silicone rubber membrane, instead of a

supported liquid, the life of the membrane can be considerably increased. One of
the potential drawbacks of SLM extraction, i.e., the relative instability of the
liquid membrane, is thereby bypassed. However, this involves a fixed composi-
tion of the membrane, so the possibilities for chemical tuning (e.g. the applica-
tion of carriers) the extraction process are reduced. This especially limits the
possibilities of extraction of relatively polar analytes, where the addition of
various ion-pair or complex formers to the membrane is imperative. Numerous
examples of this are listed below. Also, polymeric membranes lead to slower
extraction as diffusion coefficients are larger in polymers than in liquids. On the

505
other hand, the membrane is virtually insoluble in most common solvents, so
any combination of aqueous and organic liquids can be used as the donor and
acceptor phases. Application of polymeric membranes has been described both
with an aqueous, trapping acceptor and with an organic solvent in the acceptor
channel [24-26]. The latter version is sometimes termed membrane-assisted
LLE [27], and is somewhat similar to MMLLE, with the additional feature that
the dissolution of the analytes into the membrane polymer will influence the
mass transfer, leading to slower extraction but a more stable system.

15.3 THEORY AND PRINCIPLES OF SUPPORTED LIQUID

MEMBRANE (SLM) EXTRACTION

15.3.1 Mass transfer

There are a large number of chemical principles that have been used for SLM
extraction of various classes of compounds. A number of such principles are
summarised in Table 15.1 and discussed in some detail below.

15.3.1.1 Simple permeation-acidsand bases

Figure 15.1 shows the principle of simple permeation in SLM (and, in fact, also
PME) extraction, set up for extraction of basic compounds, e.g. amines. First, the
pH of the sample is adjusted to a sufficiently high value so the amines are
uncharged. When the sample is pumped through the donor channel, the
uncharged amines (B) are partitioned into the organic membrane phase.
To obtain an efficient transport of the amines by trapping, the acceptor chan-
nel on the other side of the membrane is filled with a stagnant acidic buffer.
Then an amine molecule, which has diffused through the organic membrane, is
immediately protonated at the membrane-acceptor interface and thereby pre-
vented from re-entering the membrane. The result is a transport of amine mole-
cules from the donor to the acceptor phase.
Referring again to Fig. 15.1, it is clear that acidic compounds (HA) will be
charged already in the alkaline donor phase and will therefore be completely
excluded from the membrane. The same is true for permanently charged com-
pounds. Neutral compounds (N) may be extracted, but will not be trapped in the
acceptor, so the concentration in the acceptor phase will never exceed that in the
donor phase and no enrichment is obtained. Further, hydrophilic neutral
compounds will be very reluctant to partition into the membrane, while hydro-
phobic neutral compounds might accumulate in the membrane but not continue

Fig. 15.1. Schematic description of the SLM

principle. For details see the text.

506
TABLE 15.1
Schematic overview of different chemical principles for extraction and trapping used for SLM
extraction. TOPO = trioctylphosphine oxide; DTPA = diethylenetriaminepentaacetic acid;
DEHPA = diethylhexylphosphoric acid.
Analytes Donor Membrane Acceptor Transported Trapping Ref.
species
Simple permeation
Acids acidic org (+TOPO) basic neutral anions 6,28
Bases basic org acidic neutral cations 6

Carrier-mediated transport
Metal ions 8-hydroxy- org DTPA complexes charged 29
quinoline complexes
Metal ions, acidic, DEHPA acidic, complexes counter- 30
Amino acids pH = 3 pH = 0 transport
of H+
Amino acids, basic trioctylmethyl- acidic, ion pairs cations 31-34
amino ammonium chloride
phosphonates
Amino acids basic Pd-complex acid complexes cations 35,36
Sugars, diol- neutral boronic acid acid covalently protonation 37,38
containing carrier bound of carrier
compounds complexes
Anionic acidic + org basic ion pairs deprotonation 39
surfactants amine of amine
carrier
Immunological trapping
Triazine neutral org atrazine permeation immunological 40
herbicides antibodies as antigen-
antibody
complex

to the acceptor. Charged macromolecules, as proteins, will be rejected and the

extraction rate of uncharged macromolecules will be very low due to their low
diffusion coefficients. In summary, with the conditions mentioned, the SLM
extraction will be highly selective for small, basic molecules.
Obviously, acidic compounds may be extracted in a similar way to amines by
reversing the pH conditions.

15.3.1.2 Carrier-mediatedtransport-metals,organic ions

By adding, e.g., ion-pairing reagents or chelating reagents to the donor phase,
SLM extraction systems for various permanently charged compounds and metal
ions can be devised. Various carrier molecules or ions can be incorporated in the
membrane phase to enhance selectivity and mass transfer, as well as trapping
reagents in the acceptor phase preventing analytes from being extracted back
into the membrane.

507
 MeL+- C (HD) 2 - FMe +
+
C
2 H' e MeD 2 - - 2H

Fig. 15.2. Membrane extraction schemes for of metal ions (Me). HQ = 8-Hydroxyquinoline,
DTPA 5- = diethylenetriaminepentaacetic acid anion, R4N+ = methyltrioctylammonium cation,
HD = diethylhexyl phosphoric acid. For a,b,c, see the text.

As an analytical example of adding a reagent to the donor phase, metals can

be extracted from solutions containing a complex former as 8-hydroxyquinoline,
which forms extractable complexes with many metals [29] (see Fig. 15.2a). This
complex is transported through the membrane in a similar way to that described
above. The extracted analyte can be trapped in the acceptor by another ligand, in
the cited work DTPA (diethylenetriaminepentaacetic acid), forming a stronger
and charged complex.
There are many examples of the addition of a carrier to the membrane phase.
A common carrier that has been used both for extraction of metals and organic
acids is Aliquat-336 (methyltrioctylammonium chloride). It is a tertiary ammo-
nium ion, i.e., permanently positively charged in ion pair with chloride and it can
be added to a suitable membrane solvent. For metals, it has been used for
extraction of Cu, Cd, Co, Zn [29]. The addition of thiocyanate ions to the donor
permits a negatively charged metal-thiocyanate complex to be formed in the
donor. This is extracted as ion pair with the Aliquat-336 cation into the mem-
brane and eventually trapped in the acceptor using DTPA as described above
and presented in Fig. 15.2b. A similar extraction system can be used for extrac-
tion of organic acids, especially amino acids. Amino acids cannot be extracted by
simple permeation, as there is no pH where the amino acids are uncharged. They
are efficiently extracted with Aliquat-336 from a basic solution, where the amino
acid is negatively charged [31]. This is shown in Fig. 15.3a. In this case, a
gradient of chloride ions from the acceptor to the donor phase provides a driving
force for the extraction.

508
 Donor phase Membrane Acceptor phase
-

A- R4NCI - HA

CI
C[ew RNA - IT H'
a

H2A- (HD)2, - -H2A

H+ H:AD(HD) 3 - EH
b
pH- 3 pH<

S-
RNHS- -RNH2

RNH;
OH

Fig. 15.3. Extraction schemes for membrane extraction of amino acids (HA) and anionic
surfactants (S-). RNH 2 = hexylamine. Other abbreviations as for Fig. 15.2.

Another much used extractant is diethylhexyl phosphoric acid (DEHPA). It

can be used for extraction of a number of metal ions [41,42] (Fig. 15.2c). In this
case a pH gradient is created over the membrane, so the acceptor is kept more
acidic than the donor (typically pH -1 and pH -3, respectively). Speciation of
different chromium species (chromate and chromium ion) was performed by the
combination of two extraction systems, one working with DEHPA for extraction
of Cr 3+ and the other with Aliquat-336 as described above for the chromate
anions [43].
Extraction of amino acids with DEHPA can be made in a similar way as for
metals [30] (see Fig. 15.3b). In an interesting work, a number of chiral analogues
to DEHPA have been used to obtain an enantioselective extraction process,
although with moderate success [44]. To improve the extraction of very hydro-
philic compounds, DEHPA was also employed [45], where biogenic amines in
wine were successfully extracted from a donor phase with pH = 5 and the accep-
tor with pH = 1, using DEHPA in the membrane. The analytes in question were
badly extracted with simple permeation (Section 15.3.1.1) as their high hydro-
philicity led to very low partition coefficients. The corresponding scheme is anal-
ogous to that in Fig. 15.3b.
There are a number of other examples of carrier-mediated transport. By the
addition of certain palladium-containing complexes to the membrane, amino
acids were extracted as described by Calzado et al. [35,36] and in a similar way
nitrite [46].
An ion-pair membrane extraction strategy was also used for extraction of
anionic surfactants [39]. Such surfactants (sulfonic acid anions) cannot be
protonated as conventional organic acids, and can therefore not be extracted

509
 100
90
k
80
70
60
E(%) 50
40
30
20
Fig. 15.4. Influence of TOPO 10
content in di-n-hexyl ether on 0
extraction efficiency E. (After
Shen et al. [28], with permission
from Elsevier Science B.V.) TOPO (%)in di-n-hexyl ether

with simple permeation. The addition of an aliphatic amine (e.g., hexyl amine) in
acidic donor solution permitted the formation of an ion pair that could readily be
extracted. As shown in Fig. 15.3c, a higher pH in the acceptor caused the amine
to be non-charged, leading to the destruction of the ion pair, so the surfactants
were efficiently trapped in an indirect way.
There is an example of covalent binding between the analyte and a mem-
brane additive, namely the extraction of sugar and other compounds with vicinal
hydroxyl groups by borate reagents. This principle has been studied by Smith
[37,38], but it seems not yet to have been applied to analytical problems.
A final example of a membrane-phase additive is TOPO (tri-octyl phosphine
oxide). An example of this is shown in Fig. 15.4, where extraction of carboxylic
acids of different polarities is strongly influenced by the contents of TOPO in the
membrane [28]. The most polar acid, lactic acid, was not extracted at all without
TOPO, but the extraction was significantly improved by the additive. Butanoic
acid, being the most hydrophobic acid in the study, was well extracted without
TOPO and essentially unaffected by its concentration. TOPO is known to form
hydrogen bonds, thereby improving the extraction of relatively polar
compounds.

15.3.1.3 Mathematical modeling

As discussed above, SLM extraction can be seen as a combination of a liquid- liq-
uid extraction into an organic solvent followed by a back-extraction into a second
aqueous phase. The kinetics of these operations are very different to SLM where
the two extraction steps occur simultaneously. The mass transfer kinetics in
SLM will be generally more efficient. The general mass transfer theory for SLM
extraction was described in detail [47], with some additional aspects described
more recently [48,49]. In short, the mass transfer from donor to acceptor is
proportional to the concentration difference, AC, over the membrane. With some
simplifications (especially related to activity effects at different ionic strengths)
the following holds:

510
.AC = YD CD- aA C (15.1)
where CD and CA are the concentrations in the donor and acceptor phase, respect-
ively; a, and aA are the fractions of the analytes that are in extractable
(uncharged) form in the actual phase. Typically, extraction conditions are such
that aD is close to 1 and aA is a very small value (corresponding to, e.g., an
alkaline donor and an acidic acceptor intended for extraction of bases, c.f. Fig.
15.1). CA is zero from the beginning of the extraction and increases during the
operation, usually to values well over CD. The maximum concentration enrich-
ment factor possible is reached when AC has eventually reached zero:
Ee(m) = (CA/CD)max = QD/(A (15.2)
The rate at which the conditions in Eq. (15.2) are approached depends on many
parameters, as detailed earlier [47] and briefly discussed in Section 15.3.4. There
are in principle two different cases: Membrane-controlledextraction and donor-
controlled extraction. If the extraction is membrane-controlled, the rate-limiting
step is the diffusion of the analyte compound through the membrane, which
usually leads to a slow extraction. On the other hand, with a donor-controlled
process the mass transfer rate is typically considerably larger. Here, it is limited
by diffusion in the donor phase, so the rate of mass transfer depends mainly on
the diffusion coefficient in the donor phase, D, and on the donor flow condi-
tions. If the partition coefficient K between the donor and membrane phases is
less than about 1, the extraction is typically membrane-controlled, while the
mass transfer is mainly donor-controlled when K > 10. Other than this, the
value of the partition coefficient has no large influence on the efficiency of
extraction although the rate of mass transfer increases slowly with K. As K is not
involved in Eq. (15.2), it does not influence the maximum possible enrichment
factor, but only the rate of mass transfer. This is in contrast to the conditions for
classical LLE.
Further, it seems that too large partition coefficients are not favourable [49],
probably as the transfer of analyte out of the membrane into the acceptor phase
may become less efficient. This influence of K is further discussed in Section
15.3.3.
The extraction efficiency E is usually expressed as the fraction of analyte
amount input to the system that is recovered in the acceptor:
E = nA/n, (15.3)

where n1 and nA are the number of moles input during the extraction time and
those collected in the acceptor, respectively. This parameter is analytically
important and it is not identical to the concept of recovery. An alternative to Eq.
(15.3) is the formula:
E' = (n,-n )/n, (15.4)
where nw is the number of moles leaving the donor channel. Equation (15.3) thus
tells how much of the input material is recovered in the acceptor, while Eq.

511
(15.4) measures how much is removed from the donor phase. We can define a
recovery R as:
R = E/E' (15.5)
If E = E', the recovery is 100% and no analyte is lost in the process. If E < E'
some analyte is either adsorbed in the apparatus or left in the membrane. This is
the memory effect, which is discussed in a number of cases [28,39,50], and
constitutes a limitation of SLM extraction. In practice, the problem can be
overcome by careful design of the experimental conditions. This is also related to
the adverse effects of too high partition coefficients, as mentioned above.

15.3.1.4 Temperature dependence

The temperature dependence of SLM extraction has not been very thoroughly
investigated. In the only systematic study [51], the temperature dependence of
the permeation through an SLM was measured with some phenols in the range
0-50°C, observing a roughly three times change in permeation rate over that
temperature range. For several reasons, it is not straightforward to translate
this into extraction efficiencies. For example, the experiments measure only
membrane-controlled transport, while the donor-controlled mechanism usually
dominates in practical use. If the conditions in a practical experiment are such
that the extraction efficiency is large (i.e. large K and low flow rate) a threefold
increase in K hardly influences the result, while a threefold increase in DD has a
considerable effect and these effects are smaller when E is close to 100%. There
are some observations regarding the temperature dependence of extraction of
phenoxyacids from natural water [52]. In this work, extractions were performed
at temperatures between 5 and 20 ° , referring both to the extracted water and the
ambient temperature where the apparatus was set up. No significant differences
in extraction efficiency were seen in that work. It is obvious that more work is
needed in this field.

15.3.2 Analyte trapping

From Eqs. (15.2-15.3) it can be seen that for efficient SLM extraction, a neutral,
extractable species should be formed in the donor phase (or at the donor-
membrane interface) so aD is practically unity. Then, this species should be
moved though the membrane and into the acceptor phase, where it shall become
transformed to a nonextractable species, i.e. a, should be close to zero. This
transformation is termed trapping and it can be arranged in several chemical
ways.

15.3.2.1 Direct trapping

Looking again at simple permeation of basic compounds (e.g., amines), almost all
of the analyte molecules will be in the form of ammonium ions if there is a
sufficiently low pH in the acceptor phase. As the ions are not extracted, this

512
constitutes an efficient trapping. The a,-values that are obtained are easily
calculated. For a monoprotonic basic analyte, the fraction aA of non-ionised
molecules in the acceptor is given by:
aA = Ka/([H +] + Ka) (15.6)
Ka is the dissociation constant of the corresponding ammonium ion. It was found
[47] that a sufficiently low value of cA in practice is 0.0005. With conditions
giving such a value, the extraction efficiency will be constant over a reasonable
period of time. This, together with Eq. (15.6), leads to the rule of thumb that the
acceptor pH should be 3.3 units below the pKa to obtain sufficient trapping of
amines.
Obviously, acidic compounds are trapped in an analogous way by a high
acceptor pH, and analogous equations can be set up for other acidic compounds.
This leads to the rule that the acceptor pH should be 3.3 units above the pKa of
acidic analytes, for the trapping to be satisfactory.
With acceptor conditions fulfilling these rules and with a donor pH leading to
a, = 1, Eq. (15.2) leads to a maximum enrichment factor of 2000 times.

15.3.2.2 Indirecttrapping
There are a number of SLM-systems where the transport is driven by principles
other than the simple direct trapping by forming a non-extractable charged spe-
cies as a result of a pH gradient or the addition of a reagent for the purpose
described above. In the application of DEHPA for extraction of various cationic
compounds (metal ions, amino acids, amines) (Figs 15.2c and 15.3b, the trans-
port is driven by a pH gradient, by providing an excess of protons in the acceptor
phase compared to the donor phase. The DEHPA anion is then protonated, so
the metal ion is released and the return of the carrier from the acceptor to the
donor is facilitated. This is an example of a counter-transport mechanism, and in
this case, the transported analyte is in the same form (in itself non-extractable)
on both sides of the membrane. A similar situation is found when anions are
extracted by means of Aliquat-336 (Figs. 13.2b and 13.3a). Here a gradient of
counter ions (chloride in the figures, but other ions could be more efficient) from
the acceptor to the donor phases, also in a counter-transport mechanism. In
those cases a direct trapping also helps to drive the mass transfer (by the employ-
ment of DTPA in Fig. 15.2b and protonation in Fig. 15.3a).
In the case of ion-pair transport as exemplified above for anionic surfactants
(Fig. 15.3c) [39], the ion-pair former (aliphatic) amine is neutralised so the
ion-pair is "killed" in the acceptor. In this case, the amine can in principle be
extracted back into the membrane and become "trapped" in the donor, but if this
happens in the actual system has not been investigated.

15.3.2.3 Immunological trapping

Utilising soluble antibodies in the acceptor phase can selectively trap analytes as
antigen-antibody complexes (ImmunoSLM). When the immunocomplex forms,

513
 0.9

>, 0.8

Fig. 15.5. Extraction efficiency against 07 A

log K,ow plots for phenols. Extraction effi- 2 0.6
ciencies calculated according to Eq. A
(15.3) for an SLM impregnated with () 0s
di-n-hexylether and (A) n-undecane; ()
extraction efficiency calculated accord- 04
ing to Eq. (15.4) for an SLM impregnated A
with di-n-hexylether. For details see text. 03
(After Chimuka et al. [49] with I 2 3 4 5 6
permission from Elsevier Science B.V.). g P_

the antigen can no longer re-dissolve in the organic membrane, so the analyte is
trapped in the acceptor. With a surplus of antibody, the concentration of analyte
in the acceptor phase will easily exceed the initial sample concentration. In a
preliminary work [40], the immunocomplex was quantified on-line, using a
fluorescein flow immunoassay in a sequential injection analysis (SIA) setup. The
outlined scheme was successfully applied for the extraction of 4-nitrophenol
(4-NP) from spiked water solutions as well as from a spiked wastewater sample,
indicating that the immunoextraction can be suitable when dealing with
difficult matrices.

15.3.3 Influence of the hydrophobicity of analytes

According to the modelling 47] of the SLM extraction process discussed in

Section 15.3.1.1, the main sources of resistance to mass transfer are the diffusion
through the donor phase and the diffusion through the membrane. With a
"donor-controlled" extraction process, the extraction mass transfer increases
steadily with increasing K. This provides an easy explanation why the extraction
efficiency typically increases with K, up to a certain level. However, with very
large values of K, extraction efficiencies are frequently observed to decrease, and
this effect cannot be explained with the existing theory of mass transfer kinetics.
In Fig. 15.5, the measured extraction efficiency of some phenols is plotted as a
function of the logarithm of the octanol-water partition coefficient, Kow of these
compounds. The results marked with squares and triangles indeed show such a
decrease for hydrophobic compounds, for which log Ko,, is > 3.
If the extraction efficiency is calculated according to the alternative defini-
tion in Eq. (15.4) instead of the "normal" definition in Eq. (15.3), the results
increase with Kow as expected. This means that the extraction of the analytes
into the membrane occurs as expected, but for compounds with high hydro-
phobicity only a fraction is subsequently transported to the acceptor, where it is
recovered. This is an illustration of the so-called "memory effect", a low recovery

514
according to Eq. (15.5). It is clear that the SLM technique is best applicable to
compounds with moderate hydrophobicity, i.e., log Kow less than 3-4. On the
other hand, it is possible to extract compounds with very low values for Ko. Most
hydrophilic compounds hitherto extracted with SLM seem to be some biogenic
amines that were successfully extracted by Romero et al. [45], using DEHPA as a
carrier (see Section 15.3.1.2). For these compounds like spermine, spermidine,
putrescine, etc. log K,,ow is in the range of -0.5 to -1.0. Also lactic acid (c.f. Fig.
15.4) is very hydrophilic with log Kow -- 0.6.

15.3.4 Sample flow rate influence

The extraction efficiency depends on the donor flow rate and other physical
parameters as the following equation (assuming complete trapping (A = 0) in a
stagnant acceptor) [47,53]:

E=l-·expch
E=l-expLh D
3D
3D K kl+
lnK1+ Kl+aDKkM
3
23DD
.DD+
6D 1
j3.
(15.7)

The extraction efficiency is a function of the three independent parameters: 0,

DD/hD and an' K kM, where = FD/(L-w). FD is the volume flow rate of the donor
phase and L and w are the length and width of the donor channel, respectively.
Note that is not the linear velocity, but the volumetric flow divided by the free
membrane surface area. The parameter D,/hD characterises the mass transfer in
the donor phase (DD is the diffusion coefficient in the donor phase and h is the
height of the donor channel) and aD - K. kM characterises the mass transfer
processes in the membrane. The membrane mass transfer coefficient kM is
proportional to DM/hM, where DM is the diffusion coefficient in the membrane and
hM is the thickness of the membrane.
From Eq. (15.7) and Fig. 15.6 it can be seen that the extraction efficiency
approaches unity as the flow-rate approaches zero. Therefore, the most efficient
extractions are obtained at low donor flow rates, which is not unexpected, as a
low flow rate increases the residence time of an analyte molecule in the donor
channel. However, in practical work time is an important issue, and it is there-
fore often more relevant to maximise the enrichment factor i.e. the accumulated
amount of analyte in the acceptor phase during a given time rather than to maxi-
mise the extraction efficiency. This will lead to larger instrumental signals (e.g.,
peak areas) in the analysis of the extract and thus to a more time-efficient analy-
sis. The enrichment factor E, is related to the extraction efficiency as:

E =E EVA = E FD t/VA
JVA (15.8)

Vs and VA are the volumes of the extracted sample and the acceptor channel,
respectively, and t is the time of enrichment. When the donor flow rate is
increased, E decreases as shown above, but this is compensated for by the
increasing amount of analyte being input to the system, so the enrichment factor
typically increases with donor flow rate for a given time, as seen in Fig. 15.6. On

515
 120

100
0.8

80
0.6
uJ 60
0.4
40

0.2
20
Fig. 15.6. Extraction efficiency, E and
anvi-1rnon+ Cantor Vx it ---r- a 0
e-lmlllmllL au-culrVne ttarllu-.ry unitl) a 0 0.05 0.10 0.15
functions of the reduced flow parameter
(see text). )

the other hand, a high flow rate will consume a large volume of sample, which is
usually no problem for extraction of environmental samples (e.g., river water) or
similar. If the available sample volume is limited (e.g., biological samples),
extraction could be better performed at low flow rates to maximise the
extraction efficiency.

15.3.5 Concentration enrichment

With SLM extraction, very high enrichment factors can be potentially obtained,
considerably higher than with LLE. Generally, Eq. (15.2) describes the maxi-
mum possible enrichment factor. With a knowledge of the relevant pKa values,
Eqs. (15.2) and (15.6) easily give estimations of the theoretically obtainable
enrichment factors for acidic and basic compounds. For a number of weakly
basic compounds, such calculations have been performed and compared with
experimental data [48]. In Fig. 15.6 some results are seen. For aniline withpKa =
4.6, a value of aA = 2.3 10 4 is calculated at the experimental conditions (acceptor
pH = 1), leading to a maximum enrichment factor of about 4300 times (assuming
a, = 1). This value is not reached even after long extraction times (the experi-
ment was ended after 25 h of extraction and 6 1 of sample, giving a final enrich-
ment factor of about 2000 times and still increasing). The weaker base
3,5-dichloroaniline, pKa = 2.5, gives acA 3.10- 2 so the maximum enrichment fac-
tor is only 33, and a similar value is quickly reached after a short time of extrac-
tion. This situation can be improved by increasing the acid concentration in the
donor, as shown in the cited work. High acid concentrations in the acceptor
might lead to differences in the ionic strength between the donor and acceptor
phases. In that case, activity effects will have to be considered in connection with
Eq. (15.2), as described in the cited work.
For ultra trace analysis, very high enrichment factors can in principle be
obtained by realising very low values of aA by an efficient trapping reaction. It is
also convenient in analytical sample preparation that the extraction is linear, i.e.
that the extraction efficiency is independent of extraction time or volume, so the

516
amount of analyte collected in the acceptor is directly proportional to the
concentration in the extracted sample. This condition will be true if the enrich-
ment factor obtained is considerably less than the theoretical maximum enrich-
ment factor, which is the same as aAcA << ac,. This is the basis for the
recommendation [47] that the acceptor pH shall be 3.3 units below the pKa for
basic analytes and correspondingly for other types of analytes.

15.3.6 Selectivity

The term selectivity can mean two slightly different things in the present
context of sample preparation of small molecules in biological and environmen-
tal matrices. The first consideration is the discrimination between different
small molecules and the second is the discrimination between small and large
molecules, usually with the objective of removing the large molecules and
recovering the small ones. From the discussions above, it is obvious that there
are a large number of possibilities to tune the membrane transport processes so
a certain class of compounds is efficiently extracted while other classes are not
extracted to appreciable extents. It is also clear that for several reasons macro-
molecules are not usually extracted at all by SLM. Many biological macro-
molecules (proteins, peptides etc.) are charged at most conditions and thus not
extractable. Even if it is possible to design extraction systems that can extract
such compounds, they are totally excluded from systems designed to extract
small molecules. A similar situation exists with humic compounds in environ-
mental samples (e.g., water). Completely uncharged macromolecules exist, but
even so, these compounds have, by virtue of their size, transport properties that
make their transport through an SLM exceedingly slow. There are many
examples of analytical applications to biological and environmental systems,
which show nearly complete selectivity of both the above types. For example,
triazine herbicides were extracted from spiked river water both with SLM and
with SPE [54]. The results in Fig. 15.7 show that in the SLM extract virtually
only the expected compounds are found, while a typical "humic hump" from
humic acids together with disturbing signals from unknown compounds are seen
in the SPE extract.
250

200

150

Fig. 15.7. Enrichment factors of E 100

aniline (1), 3-chloro-4-methylaniline
(2), 3,5-dichloroaniline (3) and 3- L 50
methyl-5-nitroaniline (4), all 0.1
mg/l. Acceptor: 0.1 M sulfuric acid
(psi-
/ -
I).
- I -Rae1 .
rom reI. L41 wlUn u luu zuuu .iuuu 4Uo UUU r
0oUU 1UUU
permission. Extracted Sample Volume [mL]

517
 4 50
a
400
7 350.
'
.6
300
2501 LI
e
- 200
150
°-I l
AV lj
5 _ a. Blood plasma
V
U ll I l l
0 2 4 6 8 o
200 b
80
1
160
140
l 20.
100 I Il
30
30
40

00 ____I__I
J
0 2 4 6 8 10 0 5 10 150 5 10 15
Time (min) Minutes Minutes

Left: Fig. 15.8. Chromatograms (LC-UV) of methoxy-s-triazine herbicides. (a) SPE-extraction of

spiked river water (1.0 ;ugl of each analyte); (b) SLM-extraction of spiked river water (0.5 jug/l of
each analyte). Peak designation: 1, simetone; 2, atratone; 3, secbumetone; 4, terbumetone.
(Reprinted from Megersa et al. [54] with permission from Elsevier Science.)

Right: Fig. 15.9. (a) Chromatograms of Amperozide (I), its metabolite (II) and homologue (III)
with the subsequent blank after enrichment from blood plasma. (b) Corresponding
chromatograms after enrichment from an aqueous buffer solution. Concentrations 4g/ml of I
and II, 8 ug/ml of III. (Reprinted from Lindeghrd et al. [55]. Copyright 1994 American Chemical
Society.)

Another example is the SLM extraction of a basic drug from blood plasma, as
shown in Fig. 15.8 [55]. Here, the chromatogram from spiked blood plasma is vir-
tually identical to that from a water solution of the actual drugs. A difference is
that the extraction efficiency from the blood plasma is lower than from water.
This is traced to the influence of drug-protein binding, and the phenomenon can
be used to study such binding constants, possibly including the corresponding
kinetics.

15.4 THEORY AND PRINCIPLES OF MICROPOROUS MEMBRANE

LIQUID-LIQUID EXTRACTION (MMLLE)

15.4.1 General

The MMLLE technique is a complement to the SLM extraction, permitting

membrane-based extraction to be extended to further classes of compounds and
it can be performed in similar membrane units as used for SLM. Here also,

518
hydrophobic membranes are used: typically PTFE or polypropene membranes.
The acceptor phase is an organic solvent, also filling the pores of the hydrophobic
membrane. This forms an aqueous-organic two-phase system, with the organic
phase partly in the membrane pores and partly in the acceptor channel. MMLLE
extends the concept of membrane extraction, as it is best applicable to hydropho-
bic, preferably uncharged compounds. i.e. those that cannot easily be extracted
with SLM. The extract is typically organic, not aqueous as with SLM. Thus
MMLLE is more easily interfaced to gas chromatography and normal-phase
HPLC than SLM, which is better compatible with reversed-phase HPLC.

15.4.2 Differences from SLM

With SLM, the concept of trapping in a stagnant acceptor is central, but with
MMLLE and an organic acceptor, no trapping reactions are used. The maximum
extraction efficiency is directly given by the partition coefficient as in LLE, the
only driving force for the mass transfer being the attainment of distribution
equilibrium between the aqueous and organic phases. If the partition coefficient
is very high, it is possible to work with a stagnant acceptor and still get a
considerable enrichment into a very small extract volume. The extraction effi-
ciency will be higher if the hydrophobicity of the analyte is large; i.e. if the
partition coefficient is large. With smaller partition coefficients, the mass
transfer can be improved if the acceptor phase is continuously or intermittently
flowing, removing the extracted molecules successively from the acceptor and
maintaining the diffusion through the membrane.
There is a recent example of MMLLE with the phases reversed, i.e. extrac-
tion of phenolic compounds from an organic phase into an aqueous phase [56]. In
this case the phenols were trapped in a highly basic aqueous acceptor, which was
kept stagnant during the extraction.

15.4.3 Acceptor flow rate influence

To describe the influence of different experimental parameters, especially the

donor and acceptor flow rates on the extraction efficiency E, a simplified theory
related to the SLM equations above has been derived [22]. Assuming that the
partition is in equilibrium and that the flows in the channels are parallel, the
following equation for the extraction efficiency results:
E = 1 -FD/(FD+ FAK) (15.9)
As seen from this equation, compounds with large distribution coefficients K are
more efficiently extracted and the extraction efficiencies increase with acceptor
flow rate. A high acceptor flow rate maintains low concentrations of the analytes
in the acceptor phase and maximises the concentration gradient leading to high
mass transfer rates. However, as the acceptor is flowing, the analyte is diluted,
which leads to a smaller concentration enrichment factor, Ee:

519
E = 1/(1/K + FA/FD) (15.10)

With a stagnant acceptor, the maximum value of the enrichment factors is:

Ee(max) = K (15.11)
This equation for MMLLE should be compared to the corresponding equation
for SLM, Eq. (15.2), which does not contain the value of the partition efficiency
and the maximum value of the enrichment factor is determined by the trapping
conditions.
MMLLE is conceptually similar to continuous dialysis. Assuming equilib-
rium between the phases and with the value of K equal to unity, Eqs. (15.10) and
(15.11) are also valid for that technique. In these equations it is assumed that the
acceptor flow is parallel to the donor flow. By reversing the acceptor flow
(counter-current flow) the concentration gradient between the two phases will
be larger leading to more efficient mass transfer. The matter was discussed by
Valcdrcel [23, p. 66]. This situation is mathematically far more complex than
when parallel flow is assumed. For the corresponding problem in dialysis,
numerical solutions to the appropriate partial differential equation system have
been derived [57].

15.5 APPARATUS FOR LIQUID MEMBRANE EXTRACTION

15.5.1 Basic membrane extraction apparatus

Some typical membrane extraction units or contactors are shown in Fig. 15.10.
They are made of two blocks of inert material with a machined groove in each.
After clamping the blocks together with a membrane between, flow-through
channels (donor and acceptor) are formed on each side of the membrane.
Channel volumes are typically in the range 10-1000 Al. Another type of mem-
brane unit is based on a hollow fibre membrane. The inside (lumen) of the fibre
contains the acceptor and the annular volume between the outside of the fibre
and the inside of a surrounding tube or cylindrical hole contains the donor. Such

(a) (b)

A
....... Membrane ~ _

Acceptor
Fig. 15.10. Membrane units for liq- channel | \
uid membrane extraction. (a) Flat
membrane module with 1 ml chan-
nel volume (A = blocks of inert
(c) Donor
material, B = membrane). (b) Flat inlet ~
membrane module with 10 Al chan- Acceptor
outlet
nel volume. (c) Hollow- fibre module Acceptor inlet -- IAcceor o
with 1.3 l acceptor channel t - Porous hollow
(lumen) volume. Donor outlet j fibre <2pl

520
 (a) Peristaltic
Acid pump Mixing
coil
Waste

Sample Membrane
unit
(b)

Vast

Dot

pH 1

Mix

Wash

(C

Acceptor buffer LC-pump LC-column

Fig. 15.11. Apparatus for liquid membrane extraction. (a) Manual off-line instrument based on a
peristaltic pump. (b) Instrument with on-line connection to HPLC for environmental studies. (c)
Experimental set-up for SLM-HPLC determination of biomolecules in blood plasma or urine.

units can be made with channel volumes as small as 1 bl. [58]. These types of
membrane units are in principle applicable to all versions of membrane
extraction for analytical sample preparation or sampling. Membrane units can
be purchased from Global FIA (Gig Harbor, WA, USA) or other suppliers of FIA
equipment. They can also be prepared in-house or obtained from the Lund
University Workshop via the author of this chapter.
Simple and cheap membrane extraction flow systems for reasonably large
sample volumes can be built up around a peristaltic pump. An example of such a
system is seen in Fig. 15.11a. With this set-up, the acceptor phase is manually
removed by the use of a syringe after each extraction. Such systems have been
used both for laboratory work [50] and for sampling in natural waters [52].

521
15.5.2 On-line connection to chromatography

With large membrane units (channel volumes around 1 ml) direct transfer of the
acceptor phase to a HPLC can be arranged by means of a precolumn [59,60] in
order to inject as much as possible of the extracted analyte (see Fig. 15.11b).
Such systems can be completely or partially automated with pneumatically or
electrically actuated valves controlled by timers or by computer systems.
Smaller membrane channels can also be used. Then a heart-cut which contains a
major part of the extract can be accommodated in the injection loop for direct
injection into the HPLC column without a precolumn [25,61,62].
For smaller samples (<1 ml) the liquid delivery precision of peristaltic
pumps is not adequate. For such applications, membrane extraction equipment
based on syringe pumps connected to so called robotic liquid handlers have been
designed [55,63]. A typical example is shown in Fig. 15.11c. Here, a robotic nee-
dle connected to a syringe pump submits reagent to adjust the pH of samples in
vials, picks up an aliquot and passes it through the donor channel of a small
membrane unit (channel volume around 10 l). The entire extract collected in
the acceptor channel is transferred to an injection loop injector connected to the
HPLC system in the usual manner. Thus, the entire extract from, say, 1 ml sam-
ple ends up in one chromatographic injection. It is possible to arrange this appa-
ratus so that while one sample is chromatographed, the next sample is extracted,
so the cycle time of the system is determined by the chromatogram time.
For gas chromatography, the most suitable membrane extraction technique
is MMLLE. The organic acceptor is better compatible with GC than with LC, as
are the analytes that are best extracted in such a system, i.e., relatively hydro-
phobic compounds. One approach is that the organic acceptor can be injected
directly into a GC by means of a large-volume injector [22]. A new development
is the ESy instrument (Personal Chemistry AB, Uppsala) [64] where a MMLLE
extraction in microscale (extract volume ca 1 Il) is automatically performed and
the organic extract is directly injected into the GC by means of an injection
needle, directly connected to the extraction cell (see Fig. 15.12).

Fig. 15.12. Schematic picture of the extraction

syringe. (After Norberg and Thordarson [64].
Reproduced by permission from the Royal
Society of Chemistry.). A: Extraction unit; B:
needle; C: pneumatically actuated piston. 1:
Hollow fiber; 2: needle; 3: Kel-F block; 4: nut;
5: solvent inlet; 6: sample inlet; 7: sample ,
outlet.

522
15.5.3 Off-line equipment

In contrast to the membrane extraction instruments described above, there are

some examples of completely off-line devices utilising membrane extraction
technology. In this way a number of individual extraction units can be handled
either manually or with autosamplers in a parallel, off-line way.
A PME device for "membrane-assisted solvent extraction" was developed by
Hauser and Popp and the device was applied to the determination of chlorinated
and other compounds in environmentally contaminated water [27]. It utilises
vials with a non-porous polyethylene film separating two compartments, one
containing the aqueous sample and the other an organic phase (see Fig. 15.13a).
After equilibrium has been reached, the organic phase is injected into a GC.
Another type of a PME-system was described by Mullins [65], who utilised a
home-made thin silicone rubber membrane mounted in the end of a plastic tube
placed in an autosampler vial as shown in Fig. 15.13b. Analytes are extracted
into the tube from blood plasma and are directly injected into LC-MS by the
autosampler. The device was applied to the determination of sulfonylurea drugs.
Liquid membrane extraction in a porous hollow fibre was presented by
Rasmussen and Pedersen-Bjergaard [66-71] under the name "Liquid Phase
Microextraction". The fibre is impregnated with an organic phase to create both
SLM and MMLLE systems, depending on the nature of the acceptor phase,
inside the fibres. Physically, the fibre is arranged under the cover of an auto-
sampler vial using two hypodermic needles, thus creating a disposable extrac-
tion unit (see Fig. 15.13c). It has been successfully applied to various drugs in
biological samples.
(a) PI-FEgStopger Organic Solvent (b)

.DPE-Membrane

GinsAZ appeo. gm s/iaseeaWasher

Glass
(approx. 5pmthickness)

Analyte

(a) Ne le for inteCtion

of acceptorsolution

Scre- top with

siliconeptum
s t Needletr collection
of acceptor
solution

Samplesolution I Porous ollow fiber

(donmsolution) . impregnated with
organicsotlent)
Vial LAcceptor solution

Fig. 15.13. Off-line membrane extraction devices according to (a) Hauser and Popp [27], (b)
Mullins [65], and (c) Gr6nhaug et al. [71]. (With permission from the authors and Wiley VCH.)

523
15.6 APPLICATIONS

15.6.1 Biomedical analysis

Membrane extraction techniques have been applied to the determination of

various compounds, mainly drugs but also other compounds, in biological fluids
(blood plasma, urine etc.). In these applications selectivity is crucial as well as
the possibility of automation. The area was recently reviewed [5,14], and is here
summarised in Table 15.2.

TABLE 15.2
Applications of membrane extraction to biological samples. Abbreviations: LPME2 = Liquid
phase microextraction, two-phase system, and LPME3 = Liquid phase microextraction, three-
phase system. Other abbreviations are found in the text or are generally used.
Analytes Matrices Membrane Analytical Ref.
technique technique
Aliphatic amines Urine SLM GC 72
Aliphatic acids Manure SLM GC 73
Aliphatic amines Blood plasma SLM GC 74
Amperozide Blood plasma SLM HPLC 55
Bambuterol Blood plasma SLM CE 58,75,
76
Bambuterol Blood plasma SLM LC-CE 77
Diprivan (Propofol) Urine SLM HPLC 78
Phenols Blood plasma SLM HPLC-biosensor 79
Lead Urine SLM AAS 42
Lead Urine SLM PSA 80
Local anaesthetics Blood plasma SLM GC 22
Local anaesthetics Blood plasma MMLLE GC 81
Amphetamine Blood plasma, urine LPME3 (LLLME) CE 66
Amhetamines, benzodia- Blood plasma, urine LPME2, LPME3 HPLC, GC, CE 68
zepines, Naproxen,
Citalopram
Benzodiazepines Blood plasma, urine LPME2 GC 69
Ibuprofen, naproxen, Urine LPME3 CE 67
ketoprofen
Metabolites of Ropivacaine Urine SLM HPLC 63
Citalopram and metabolites Blood plasma LPME3 CE 70
Amphetamines Blood, urine LPME3 FIA-MS/MS 71
Organo-phosphate esters Blood plasma MMLLE GC-MS 82
Sulfonylurea drugs Blood plasma PME (off-line) LC-MS/MS 65
Amphetamine Urine SLM HPLC 83
Benzimidazole anthelmintics Urine, tissue, milk SLM HPLC, LC-MS 84

524
15.6.2 Environmental analysis

Determination of pollutants and natural compounds in natural waters and other

environmental matrices, demand both high enrichment factors for compounds
in low concentrations, and selectivity to discriminate against, for example,
humic compounds. The different membrane extraction techniques can provide
these characteristics and there are numerous applications of membrane
extraction in this important field of practical analysis, as listed in Table 15.3.

TABLE 15.3
Applications of membrane extraction in environmental analysis. For abbreviations of membrane
techniques, see the text
Analytes Matrices Membrane Analytical Ref
technique technique
Phenoxy acids Natural water SLM HPLC 85
Chlorinated organics Water PME GC 86
Phenoxy acids Natural water SLM HPLC 87
Sulfonylurea herbicides Natural water SLM HPLC 88
Phenoxy acids Natural water SLM HPLC 52
Aliphatic amines Air (Impinger samples) SLM (gas) GC 11
Aliphatic amines Air (Impinger samples) SLM (gas) GC 89
Uncharged organics Water PME GC 26
Carboxylic acids Air (Impinger samples) SLM IC 90
Sulfonylurea herbicides Natural water SLM HPLC 59
Cu Natural water SLM AAS 91
Cu, Cd, Co, Ni, Zn Natural water SLM AAS 29
Phenolics Natural water SLM HPLC 60
Phenolics Nutrient solutions SLM HPLC 50
Carboxylic acids Soil liquid SLM IC 61
Carboxylic acids Soil liquid SLM IC 92
Triazine herbicides Natural water SLM HPLC 93
Triazine herbicides Natural water MMLLE FIA 94
Anionic surfactants Natural water SLM HPLC 39
Triazine herbicides Natural water SLM HPLC 95
Anilines Natural water SLM HPLC 96
Cu, Pb, Cd Natural water SLM AAS 97
Cu, Pb, Cd Natural water SLM AAS 41
Triazine herbicides Natural water SLM HPLC 98
Zn, Ni, Co, Mn Natural water SLM HPLC 99
Carboxylic acids Air SLM IC 100

continued

525
TABLE 16.3 (continuation)

Analytes Matrices Membrane Analytical Ref

technique technique

Triazine herbicides Natural water SLM HPLC 54

Cr (speciation) Natural water SLM AAS 43
Triazine herbicides Natural water SLM HPLC 101
Organotin compounds Natural water MMLLE GC-MS 102
Cationic surfactants Natural water MMLLE HPLC (NP) 103
Thiophanate methyl and Natural water SLM, MMLLE HPLC 104
metabolites
Vinclozolin Natural water MMLLE HPLC 105
Semi-volatile organics Water PME HPLC 106
Triazine herbicides Natural water SLM HPLC 107

TABLE 15.4
Various applications of membrane extraction in chemical analysis of foods and industrial
materials
Analytes Matrices Membrane Analytical Ref.
techniques techniques
Phenols Kerosene, naphtha PME FIA 108
Triazine herbicides Cooking oil MMLLE FIA, HPLC 109
Phenols Crude oil PME HPLC 110
Phenols Gasoline, Kerosene PME HPLC 111
Phenols Crude oil, fuels PME HPLC 112
Vitamin E Butter PME HPLC 62
Nicotine Snuff SLM (solid sample) UV 113
Caffeine Coffee, Tea SLM (solid sample) UV 114
Pesticides Eggs PME HPLC 115
Vanillin Various food SLM (solid sample) Amperometric 116
Anionic surfactants Detergents MMLLE FIA 117
Phenols Pyrolysis oil MMLLE LC-LC 56
Biogenic amines Wine SLM HPLC 45

15.6.3 Food and industrial analysis

In addition to the bioanalyticai and environmental applications of membrane

extraction, as summarised in Tables 15.2 and 15.3, respectively, there are a
number of applications in other field of chemical analysis, mainly concerning
food stuffs and various oils and fuels. These are summarised in Table 15.4.

526
Acknowledgements

The work of the Lund membrane group was supported by grants from the
Swedish Natural Science Research Council (NFR), the Swedish Environmental
Protection Agency (SNV), the Swedish Council for Forestry and Agricultural
Research (SJFR), the Swedish Institute, the Swedish International Develop-
ment Co-operation Agency (SIDA), the Crafoord Foundation and the European
Community (DG XII). The companies Pharmacia AB, Astra Draco AB and Astra
Pain Control AB have contributed with funds and interesting applications. The
work would not have been possible without a number of colleagues, especially
Prof. Lennart Mathiasson, graduate and undergraduate students as well as
guest researchers who have made important contributions.

REFERENCES

1 J.A. Jsnsson and L. Mathiasson, Trends Anal. Chem., 11 (1992) 106.

2 J.A. Jbnsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 318.
3 J.A. JBnsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 325.
4 C. Garcia Pinto, E. Fernandez Laespada, J.L. Perbz Pav6n and B. Moreno Cordero,
Lab. Autom. Inform. Managem., 34 (1999) 115.
5 J.A. J6nsson and L. Mathiasson, Chromatographia,Suppl. 52 (2000) S8 .
6 J.A. JBnsson and L. Mathiasson, J. Chromatogr.A, 902 (2000) 205.
7 B. Moreno Cordero, J.L. Perbz Pav6n, C. Garcia Pinto, E. Fernandez Laespada, R.
Carabias Martinez and E. Rodriguez Gonzalo, J. Chromatogr.A, 902 (2000) 195.
8 J.A. Jnsson and L. Mathiasson, in: P. Brown and E. Grushka (Eds.), Advances in
Chromatography, Vol. 41. Marcel Dekker, New York, 2001, p. 53.
9 S.M. Lunte and C.E. Lunte, in: P. Brown and E. Grushka (Eds.), Advances in
Chromatography. Marcel Dekker, New York, 1995, p. 383.
10 M.I. Davies, Anal. Chim. Acta, 379 (1999) 227.
11 L. Grbnberg, P. Lbvkvist and J.A. J6nsson, Chromatographia,33 (1992) 77.
12 E. Fernandez Laespada, L. Calvo Seronero, J.L. PBrez Pav6n, C. Garcia Pinto and B.
Moreno Cordero, J. Sep. Sci., 24 (2001) 526.
13 G.A. Audunsson, Anal. Chem., 58 (1986) 2714.
14 J.A. Jinsson and L. Mathiasson, J. Sep. Sci., 24 (2001) 495.
15 A.M. Sastre, A. Kumar, J.P. Shukla and R.K. Singh, Sep. Purific. Meth., 27 (1998)
213.
16 R.A. Bartsch and J.D. Way, Chemical Separations with Liquid Membranes. ACS
Symposium Series, Vol. 642. ACS, Washington, 1996.
17 S.V. Ho, Environ. Progr., 18 (1999) 273.
18 A. Schafer and M.M. Hossain, Bioprocess Eng., 16 (1996) 25.
19 A.-K. Kontturi, K. Kontturi, P. Niinikoski and G. Sundholm, Progr. Colloid Polym.
Sci., 88 (1992) 90.
20 S.-W. Tsai, C.-L. Wen, L.-J. Chen and C.-S. Wu, J. Membr. Sci., 100 (1995) 87.
21 T. Araki and H. Tsukube, T. Araki and H. Tsukube, Liquid Membranes, Chemical
Applications. CRC Press, Boca Raton, FL, 1990.
22 Y. Shen, J.A. Jonsson and L. Mathiasson, Anal. Chem., 70 (1998) 946.
23 M. Valcarcel and M.D. Luque de Castro, Non-chromatographic Continuous
Separation Techniques. The Royal Society of Chemistry, Cambridge, 1991.

527
24 R.G. Melcher, Anal. Chim. Acta, 214 (1988) 299.
25 R.G. Melcher and S.A. Bouyoucos, Process Contr. Qual., 1 (1990) 63.
26 P.L. Morabito and R.G. Melcher, Process Contr. Qual., 3 (1992) 35.
27 B. Hauser and P. Popp, J. Sep. Sci., 24 (2001) 551.
28 Y. Shen, L. Gr6nberg and J.A. Jonsson, Anal. Chim. Acta, 292 (1994) 31.
29 M. Papantoni, N.-K. Djane, K. Ndung'u, J.A. J6nsson and L. Mathiasson, Analyst,
120 (1995) 1471.
30 P. Wieczorek, J.A. J6nsson and L. Mathiasson, Anal. Chim. Acta, 346 (1997) 191.
31 P. Dzygiel, P. Wieczorek, L. Mathiasson and J.A. J5nsson, Anal. Lett., 31 (1998) 1261.
32 P. Dzygiel and P. Wieczorek, J. Chromatogr.A, 889 (2000) 93.
33 P. Dzygiel and P. Wieczorek, J. Sep. Sci., 24 (2001) 561.
34 M. Rak, P. Dzygiel and P. Wieczorek, Anal. Chim. Acta, 433 (2001) 227.
35 J.A. Calzado, C. Palet, J.A. Jonsson and M. Valiente, Anal. Chim. Acta, 417 (2000)
159.
36 J.A. Calzado, C. Palet and M. Valiente, J. Sep. Sci., 24 (2001) 533.
37 B.D. Smith, Supramol. Chem., 7 (1996) 55.
38 M. Di Luccio, B.D. Smith, T. Kida, C.P. Borges and T.L.M. Alves, J. Membr. Sci., 174
(2000) 217.
39 T. Miliotis, M. Knutsson, J.A. J6nsson and L. Mathiasson, Int. J. Environ. Anal.
Chem., 64 (1996) 35.
40 E. Thordarson, J. Emneus and J.A. J6nsson, Anal. Chem., 72 (2000) 5280.
41 N.-K. Djane, K. Ndung'u, F. Malcus, G. Johansson and L. Mathiasson, Fresenius J.
Anal. Chem., 358 (1997) 822.
42 N.-K. Djane, I.A. Bergdahl, K. Ndung'u, A. Schiitz, G. Johansson and L. Mathiasson,
Analyst, 122 (1997) 1073.
43 N.-K. Djane, K. Ndung'u, C. Johnson, H. Sartz, T. Tornstrom and L. Mathiasson,
Talanta, 48 (1999) 1121.
44 P. Dzygiel, P. Wieczorek, J.A. Jnsson, M. Milewska and P. Kafarski, Tetrahedron,
55 (1999) 9923.
45 R. Romero, J.A. J6nsson, D. Gdzquez, M.G. Bagur and M. Sdnchez-Vifias, J. Sep. Sci.
(2002), in press.
46 J.A. Calzado, C. Palet and M. Valiente, Anal. Chim. Acta, 403 (2000) 101.
47 J.A. J6nsson, P. L6vkvist, G. Audunsson and G. Nilve, Anal. Chim. Acta, 227 (1993)
9.
48 L. Chimuka, N. Megersa, J. Norberg, L. Mathiasson and J.A. Jbnsson, Anal. Chem.,
70 (1998) 3906.
49 L. Chimuka, L. Mathiasson and J.A. Jonsson, Anal. Chim. Acta, 416 (2000) 77.
50 M. Knutsson, J. Lundh, L. Mathiasson, J.A. Jnsson and P. Sundin, Anal. Lett., 29
(1996) 1619.
51 L. Chimuka, M.M. Nindi, M.E.M. ElNour, H. Frank and C. Velasco, J. High Resol.
Chromatogr., 22 (1999) 417.
52 M. Knutsson, G. Nilv6, L. Mathiasson and J.A. J6nsson, J. Agric. Food Chem., 40
(1992) 2413.
53 P. Wieczorek, J.A. J6nsson and L. Mathiasson, Anal. Chim. Acta, 337 (1997) 183.
54 N. Megersa, T. Solomon and J.A. Jonsson, J. Chromatogr.A, 830 (1999) 203.
55 B. Lindegdrd, H. Bj6rk, J.A. J6nsson, L. Mathiasson and A.-M. Olsson, Anal. Chem.,
66 (1994) 4490.
56 T. Hy6tylainen, T. Andersson, M. Jussila, S.K. Wiedmer, M. Rautiainen and M.-L.
Riekkola, J. Sep. Sci., 24 (2001) 544.
57 B. Bernhardsson, E. Martins and G. Johansson, Anal. Chim. Acta, 167 (1985).
58 S. Palmarsdottir, E. Thordarson, L.-E. Edholm, J.A. Jnsson and L. Mathiasson,
Anal. Chem., 69 (1997) 1732.

528
59 G. Nilv6, M. Knutsson and J.A. Jonsson, J. Chromatogr.A, 668 (1994) 75.
60 M. Knutsson, L. Mathiasson and J.A. Jonsson, Chromatographia,42 (1996) 165.
61 Y. Shen, V. Obuseng, L. Gr6nberg and J.A. Jonsson, J. Chromatogr. A, 725 (1996)
189.
62 M.M. Delgado Zamarreno, A. Snchez Prez, M. Bustamante Rangel and J.
Hernandez M6ndez, Anal. Chim. Acta, 386 (1999) 99.
63 J.A. Jnsson, M. Andersson, C. Melander, J. Norberg, E. Thordarson and L.
Mathiasson, J. Chromatogr. A, 870 (2000) 151.
64 J. Norberg and E. Thordarson, Analyst, 125 (2000) 673.
65 F. Mullins, J. Sep. Sci., 24 (2001) 593.
66 S. Pedersen-Bjergaard and K.E. Rasmussen, Anal. Chem., 71 (1999) 2650.
67 S. Pedersen-Bjergaard and K.E. Rasmussen, Electrophoresis,21 (2000) 579.
68 K.E. Rasmussen, S. Pedersen-Bjergaard, M. Krogh, H.G. Ugland and T. Gr0nhaug,
J. Chromatogr.A, 873 (2000) 3.
69 H.G. Ugland, M. Krogh and K.E. Rasmussen, J. Chromatogr.B, 749 (2000) 85.
70 T. Gr0nhaug Halvorsen, S. Pedersen-Bjergaard and K.E. Rasmussen, J.
Chromatogr. A, 909 (2001) 87.
71 T. Gr0nhaug Halvorsen, S. Pedersen-Bjergaard, J.L.E. Reubsaet and K.E.
Rasmussen, J. Sep. Sci., 24 (2001) 615.
72 G.A. Audunsson, Anal. Chem., 60 (1988) 1340.
73 L. Mathiasson, M. Knutsson, G. Bremle and L. Martensson, Swedish J. Agri. Res., 21
(1991) 147.
74 B. Lindeghrd, J.A. Jonsson and L. Mathiasson, J. Chromatogr., 573 (1992) 191.
75 S. Palmarsdottir, B. Lindegard, P. Deininger, L.-E. Edholm, L. Mathiasson and J.A.
Jinsson, J. CapillaryElectroph., 2 (1995) 185.
76 S. Palmarsdottir, L. Mathiasson, J.A. Jonsson and L.-E. Edholm, J. Chromatogr.B,
688 (1997) 127.
77 S. Palmarsdottir, L. Mathiasson, J.A. Jnsson and L.-E. Edholm, J. Capillary
Electroph., 3 (1996) 255.
78 J. Trocewicz, Z. Suprynowicz and J. Markowicz, J. Chromatogr.B., 685 (1996) 129.
79 J. Norberg, J. Emneus, J.A. J6nsson, L. Mathiasson, E. Burestedt, M. Knutsson and
G. Marko-Varga, J. Chromatogr. B., 701 (1997) 39.
80 N.-K. Djane, S. Armalis, K. Ndung'u, G. Johansson and L. Mathiasson, Analyst, 123
(1998) 393.
81 Y. Shen, L. Mathiasson and J.A. Jonsson, J. Microcol. Sep., 10 (1998) 107.
82 O.B. Jonsson, E. Dyremark and U.L. Nilsson, J. Chromatogr.B, 755 (2001) 157.
83 J. Trocewicz, J. Sep. Sci., 24 (2001) 587.
84 T. Msagati and M.N. Nindi, J. Sep. Sci., 24 (2001) 606.
85 G. Nilv6, G. Audunsson and J.A. J6nsson, J. Chromatogr.,471 (1989) 151.
86 R.G. Melcher and P.L. Morabito, Anal. Chem., 62 (1990) 2183.
87 L. Mathiasson, G. Nilve and B. U16n, Int. J. Environ. Anal. Chem., 45 (1991) 117.
88 G. Nilve and R. Stebbins, Chromatographia,32 (1991) 269.
89 L. Gronberg, P. Lvkvist and J.A. Jonsson, Chemosphere, 24 (1992) 1533.
90 L. Gronberg, Y. Shen and J.A. J6nsson, J. Chromatogr., 655 (1993) 207.
91 N. Parthasarathy and J. Buffle, Anal. Chim. Acta, 284 (1994) 649.
92 Y. Shen, L. Str6m, J.A. J6nsson and G. Tyler, Soil Biol. Biochem., 28 (1996) 1163.
93 J. Trocewicz, J. Chromatogr.A, 725 (1996) 121.
94 R. Carabias Martinez, E. Rodriguez Gonzalo, M.P. Santiago Toribio and J.
Hernandez M6ndez, Anal. Chim. Acta, 321 (1996) 147.
95 L. Chimuka, M.M. Nindi and J.A. Jonsson, Int. J. Environ. Anal. Chem., 68 (1997)
429.
96 J. Norberg, A. Zander and J.A. Jonsson, Chromatographia,46 (1997) 483.

529
 97 N. Parthasarathy, M. Pelletier and J. Buffle, Anal. Chim. Acta, 350 (1997) 183.
98 N. Megersa and J.A. J6nsson, Analyst, 123 (1998) 225.
99 K. Ndung'u, N.-K. Djane and L. Mathiasson, J. Chromatogr.A, 826 (1998) 103.
100 L. Martensson, M. Magnusson, Y. Shen and J.A. Jonsson, Agri. Ecosys. Environ., 75
(1999) 101.
101 N. Megersa, T. Solomon, B.S. Chandravanshi and J.A. Jnsson, Bull. Chem. Soc.
Ethiopia, 14 (2000) 9.
102 K. Ndung'u and L. Mathiasson, Anal. Chim. Acta, 404 (2000) 319.
103 J. Norberg, E. Thordarson, L. Mathiasson and J.A. J6nsson, J. Chromatogr.A, 869
(2000) 523.
104 M. Sandahl, L. Mathiasson and J.A. J6nsson, J. Chromatogr.A, 893 (2000) 123.
105 M. Sandahl, E. DUlfsson and L. Mathiasson, Anal. Chim. Acta, 424 (2000) 1.
106 X. Guo and S. Mitra, J. Chromatogr.A, 904 (2000) 189.
107 N. Megersa, L. Chimuka, T. Solomon and J.A. Jonsson, J. Sep. Sci., 24 (2001) 567.
108 E. Rodrigues Gonzalo, J.L. Perez Pav6n, J. Ruzicka, G.D. Christian and D.C. Olson,
Anal. Chim. Acta, 259 (1992) 37.
109 R. Carabias Martinez, E. Rodriguez Gonzalo, E. Herndndez Fernandez and J.
Herndndez M6ndez, Anal. Chim. Acta, 304 (1995) 323.
110 T. Garcia Sanchez, J.L. PBrez Pav6n and B. Moreno Cordero, J. Chromatogr.A, 766
(1997) 61.
111 E. Fernandez Laespada, J.L. Perez Pav6n and B. Moreno Cordero, J. Chromatogr.A,
823 (1998) 537.
112 E. Ferndndez Laespada, J.L. P6rez Pav6n and B. Moreno Cordero, J. Chromatogr.A,
852 (1999) 395.
113 E. Luque-Perez, A. Rios, M. Valcdrcel, L.-G. Danielsson and F. Ingman, Anal. Chim.
Acta, 387 (1999) 155.
114 E. Luque-Prez, A. Rios, M. Valcarcel, L.-G. Danielsson and F. Ingman, Lab. Autom.
Inform. Managem., 34 (1999) 131.
115 R. Carabias Martinez, E. Rodriguez Gonzalo, P.H. Paniagua Marcos and J.
Herndndez MBndez, J. Chromatogr.A, 869 (2000) 427.
116 M. Luque, E. Luque-Prez, A. Rios and M. Valcdrcel, Anal. Chim. Acta, 410 (2000)
127.
117 L. Jing-fu and J. Gui-bin, Microchem. J., 68 (2001) 29.

530
Chapter 16

Membrane inlet mass spectrometry

Tapio Kotiaho and Frants R. Lauritsen

16.1 INTRODUCTION

Volatile organic compounds can be sampled and analysed directly from soils,
liquids and gases by membrane inlet mass spectrometry (MIMS). This unusual
flexibility in matrixes that can be analysed is achieved through an inert polymer
membrane, which is used as the only interface between the sample and a mass
spectrometer. The membrane is simply brought into contact with the sample
where, after the volatile organic compounds diffuse through it and evaporate
into the mass spectrometer, they are ionized and analysed. In this fashion most
volatile organic compounds can be analyzed directly and in most cases a chang-
ing chemical environment in the sample can be monitored on-line.
Membrane inlet mass spectrometry is a relatively old technique; it was
introduced by Hoch and Kok in 1963 [1] as a method for the continuous
monitoring of gases dissolved in water. Soon after it was tested as an interface
between a gas chromatograph and a mass spectrometer [2]. However, a special
characteristic of membrane inlet systems-a very high selectivity in the trans-
port through the membrane-has limited this application. Normally the MIMS
membranes are very hydrophobic and hydrophobic compounds dissolve very
well in the membrane. They have low detection limits (ng/l in water and -1
Ag/m 3 in air), whereas hydrophilic compounds have detection limits in the high
jg/il range in water and mg/m 3 in air.
For many years the main applications of the MIMS technique were in vivo
monitoring of blood gases [3] and monitoring of dissolved gases and small
organic compounds (MW < 100 Dalton) in fermentors [4]. This changed in the
late eighties when the direct insertion membrane probe was developed [5]. The
direct insertion membrane probe turned the strategy of membrane inlet design
upside down. Instead of having the membrane mounted at the end of a probe
which was inserted into the sample, the membrane probe was inserted into the
mass spectrometer and the sample had to be transported to the membrane inlet.
With this design volatile organic compounds are delivered directly to the ion
source of the mass spectrometer and do not have to be transported through an
evacuated tube, where chromatographic effects at the surfaces can take place
[6]. The direct insertion membrane probe initiated a very productive decade
Comprehensive Analytical Chemistry XXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
when the MIMS technique evolved from a specialist technique into a mature
method for the detection of volatile organic compounds in environmental
samples [7] and in biological matrixes [8].
During the nineties, the application of the MIMS technique was expanded
beyond the analysis of gases and small organic compounds in aqueous matrixes.
Methods that made it possible to analyse complex organic solutions like gasoline
were developed [9] and an interest in the analysis of gaseous samples appeared
[10]. For example, a fast and sensitive method for the detection of sulfur
compounds in air was recently presented [11]. Solid analysis is a new application
area for the MIMS technique and optimized methods for the analysis of residual
volatiles in soil samples are emerging [12,13]. A review of MIMS applications can
be found in Srinivasan et al. [14] and Johnson et al. [15].
At the moment the detection of volatile organic compounds (boiling point <
200°C) have become a routine matter. However, effort is put into the develop-
ment of methods that will allow the direct detection of less and non-volatile
organic compounds [16-19]. Among these methods the desorption chemical
ionization method [20] is currently the most versatile allowing the detection of
large and complex biomolecules like steroid hormones [21], atrazine, PCB and
phthalates [22].
As well as the main areas of MIMS development and application described
above, many specialised MIMS methods have been presented in the past, and
this chapter gives an overview of the many possibilities of the technique. Empha-
sis is on the flexibility of the MIMS technique, when it comes to the analysis of
samples having completely different matrixes. First, it describes methods for the
characterization of gaseous samples, next characterization of aqueous and org-
anic liquids, followed by analysis of sediments, and finally, solid samples.

16.2 MIMS THEORY

The sampling of compounds into a mass spectrometer via a polymer membrane

is a complex process, which involves solvation in the membrane, transport
through the membrane and finally evaporation from the membrane surface into
the mass spectrometer. Each of the processes is highly selective towards various
organic compounds, the selectivity being dependent on the chosen type of mem-
brane. In most cases a hydrophobic and inert silicone membrane is used. This
hydrophobic membrane favours non-polar organic compounds which dissolve
very well in it. It has a nonporous structure and the transport process is a simple
diffusion process, with the size of the sample playing a major role. The last step,
evaporation of the sample from the membrane surface, limits the applicability of
standard MIMS to the detection of compounds with a relatively low boiling point
(<200°C).
Generally, the transport parameters are independent of the sample concen-
tration and a very simple set of equations describe the system. The steady state
flow (I,,) through a flat sheet membrane is given by

532
I, =AxDxKxx (16.1)
1
where A is the surface area of the membrane, D is the diffusion constant, K is the
distribution ratio of the analyte between the concentration in the membrane and
in the sample, C is the analyte concentration in the sample, and I the thickness of
the membrane. In other words, the flow through the membrane is proportional
to the surface area of the membrane, the product between the diffusion constant
and the distribution ratio (a parameter called the permeability) and the concen-
tration gradient across the membrane.
The response time (t1 -90 %) defined as the time it takes the signal to increase
from 10% to 90% following a step change in concentration is given by
12
t0-90 = 0.237 x - (16.2)
D
It increases with the square of the membrane thickness and decreases
proportionally to the diffusion constant. In contrast to the steady-state flow the
response time does not depend upon the distribution ratio.
The distribution ratio is properly the most important parameter, when it
comes to the detection limits of a MIMS system. As mentioned above, most
MIMS systems use a very hydrophobic silicone membrane and this membrane
discriminates strongly against compounds containing polar groups. This is best
demonstrated by a comparison of detection limits obtained from structurally
similar compounds. For example, the detection limit for toluene is normally in
the ng/l range, whereas that of the anisol (methoxybenzene) is 10 times higher,
benzaldehyde is 10-100 times higher, benzyl alcohol more than 100 times higher
and benzoic acid as the most polar compound in the series is more than a
thousand times higher. The extreme variation in solubility of relatively similar
compounds in the membrane makes MIMS a special technique in that it gives a
biased picture of the hydrophobic components in a sample. Often, constituents
in a sample that hitherto has been unnoticed using other analytical methods,
suddenly give a signal that dominates the analytical output and suppresses the
signal from well-known constituents. The identification of 3-Cl-4-methoxy-
benzaldehyde in growth media of the white-rot fungus Bjerkandera adusta is a
good example of this [23]. Unfortunately this feature occasionally makes it
impossible to determine the concentration of polar constituents present at high
concentration.
The hydrophobic nature of some compounds can cause analytical difficulties
when present at high concentrations. They might act as plasticizers and cause
softening (swelling) of the membrane. The signal becomes nonlinear with con-
centration. 3-Cl-4-methoxy-benzaldehyde is an example of a compound that
softens silicone membranes. It gives linear signals from the detection limit (< 10
gg/l) and up to about 10 mg/Il, where after the signal increases too fast up to about
50 mg/l before it becomes linear again (Fig. 16.1). This behaviour can be
explained by the simple assumption, that the sample molecules interact with the

533
 1 _
0. -
/
0.6-

0.4-

Fig. 16.1. Calibration curve for 3-Cl-4-methoxybenz- 0.2 02O 0 405v

aldehyde showing the effect of a sample that causes a Q
softening (swelling) of the membrane at high 0.o0 o 20 30 SIo so
concentration. Concentration (mgIl)

polymer chains and cause a local softening of the membrane, where diffusion is
faster and perhaps the distribution ratio is also higher. At low concentration the
distance between individual sample molecules in the membrane is so big that
one molecule will not experience the soft areas created by other molecules. At a
certain point the soft areas begin to overlap and a molecule diffusing through the
membrane will experience both soft and dense places in the membrane. The
higher the sample concentration the more soft areas are experienced and the
result is a signal that increases faster than linearity. At a certain concentration
the membrane is fully softened (swelled), and the signal becomes linear again,
but the permeability has increased.
The effect of membrane swelling can be difficult to observe due to saturation
effects [24] in the mass spectrometric system. When it is observed, it does not
just cause nonlinearities for the hydrophobic constituent itself, it will also influ-
ence the detection of other compounds in the sample. A related effect can be
observed in connection with the analysis of liquid samples that have a high salt
concentration (-40 gl). Such a matrix can influence the distribution ratio
between the sample and the membrane with nonlinear effects as the result.
The diffusion constant influences both the steady-state flow through the
membrane and the response time. In general, the larger the compound the
slower it diffuses through the membrane, since more free space between the
polymer chains is needed before the molecule can pass. As a result large com-
pounds have long response times, whereas small compounds have short response
times. However, the small diffusion constant for large compounds does not nec-
essarily mean that the steady-state flow through the membrane becomes small.
As already mentioned, the steady-state flow through the membrane depends on
the product between the diffusion constant and the distribution ratio. Often the
increase in distribution ratios by far exceeds the decrease in diffusion constant
and the result is a higher flow through the membrane for the large compound.
This is exemplified by octanol which has a steady-state flow through the mem-
brane which exceeds that of ethanol by approximately a factor of 50 [6].
The evaporationof volatile organic compounds from the membrane surface
into the mass spectrometer is normally not a problem. The vapour pressure is
high and the flow through the membrane so low that heat conduction can
compensate for the energy used in the evaporation process. It is after the

534
evaporation from the membrane has taken place that problems may arise.
Interactions between the vaporized analyte and surfaces in the vacuum [6,25]
may cause prolonged response times or even a complete suppression of the
signal. Most membrane inlets are therefore designed to have as short a distance
as possible from the membrane to the ion source of the mass spectrometer.
Compounds of low volatility (boiling point > 250°C) cannot be detected by
standard MIMS. They stick to the membrane and have to be released by some
sort of stimulated evaporation. Thus far, three methods have been tested: in
trap-and-release MIMS [16,17] a sudden thermal heating of the membrane up to
300°C is used; in laser desorption MIMS [19] the vacuum side of the membrane is
irradiated by laser; and in desorption chemical ionization MIMS [20,21] a
plasma-assisted evaporation is used.
The membrane temperature is very important for the performance of a
membrane inlet. Both the diffusion constant, the distribution ratio and the
permeability (equal to the product of the diffusion constant and the distribution
ratio) depend on the temperature, according to Arrhenius equations [26]:

D =Do xexp -Ed X(R RTo )) (16.3)

K = K xexpAHRT )) (16.4)

P =P0 xexp -Ep Xl - 1 )) (16.5)

=DO xKo x-(Ed +AH,)X(RT RT ))

where Do, Ko and P0 represent the diffusion constant, the distribution ratio and
the permeability at some initial temperature, To. Ed is the energy of activation
for diffusion, AH. is the difference in heats of solution between the membrane
and the sample matrix and Ep the activation energy for permeation.
In general the activation energy for diffusion, Ed, is larger than zero, which
means that the diffusion constant increases with temperature, whereas for most
organic compounds the difference in heats of solution between the membrane
and the matrix, AHs, is less than zero, which means that the analyte concentra-
tion inside the membrane decreases with temperature. In all circumstances, the
response time, which is inversely proportional to the diffusion constant, becomes
shorter the higher the temperature. The steady-state flow through the mem-
brane can both increase or decrease, when the temperature is increased. This is a
question of the sign of the sum of Ed and AH. For gas analysis AHs is often larger
than Ed, whereas for water analysis AH is smaller than Ed. The result is that for
gas analysis the steady-state flow through the membrane decreases with an

535
increase in temperature, whereas it increases for water analysis. Gas analysis is
therefore carried out at a temperature that is chosen as a compromise between a
desire for maximum signals (low temperature) and short response time (high
temperature). Water analysis is carried out at the practically maximal tempera-
ture, which is normally around 70°C. At higher temperatures bubble formation
in front of the membrane can cause serious instability in the signals [27].

16.3 MEMBRANE INLET DESIGNS

Membrane inlets exist in many different designs, each created to fit a specific
application. However, most of the designs belong to one of six categories (Fig.
16.2): (a) membrane probes, where the membrane is mounted at the end of a
probe to be inserted directly into the sample matrix; (b) the direct insertion
membrane probe, where the membrane is mounted inside the mass spectrome-
ter and the sample (gas or liquid) has to be transported to it; (c) measuring cells,
where a sample (liquid or solid) is transferred to a small measuring cell that has
a membrane built into the wall; (d) the helium purge inlet, where permeate
molecules at the inside of the membrane is purged by helium to the mass
spectrometric ion source; (e) two stage inlet systems, where two different
systems (one being the membrane) for separation and/or preconcentration of the
sample are used in series; and (f) membrane inlets using a stimulated desorption
for the analysis of semi- and/or non-volatile organic compounds.

a) _____ _ d)

He

Membrane

b) e)

bII He

Member
_:Membrane Memban
Membrane I Jet separator

C) b11 ' Me nl~jjm

Membrane 1 Sample in

Fig. 16.2. Different types of membrane inlet design. (a) Membrane probe mounted directly in a
reactor. (b) Flow-through membrane probe. (c) Sample cell mounted directly at the mass spectro-
metric ion source. (d) Helium purge membrane inlet. (e) A two-stage membrane inlet system using
a helium purge membrane inlet followed by ajet separator that reduces the relative amount of the
helium in the carrier gas. (f) A membrane inlet using stimulated desorption to release less volatile
organic compounds from the membrane surface. In the system presented in (f) (T&R-MIMS) the
membrane is continuously irradiated by light and electrons from the filament.

536
 The membraneprobe is probably the simplest membrane inlet to use. It con-
sists of a capillary (steel) perforated, and covered by a polymer membrane in one
end, while the other end is connected via an evacuated tube to the mass spec-
trometer. The probe can be inserted directly into almost any sample. For exam-
ple they have been inserted into arteria of humans [3], into plants [28], into
sediments [29] and into chemical and biological reactors of all kinds. The draw-
back of the membrane probe is that it can only be used to analyse gases and
highly volatile organic compounds (b.p. < 100°C). Compounds of a lower volatil-
ity interact with the surfaces in the evacuated tube which connects the inlet to
the mass spectrometer and very long response times are often the result.
In order to circumvent the problem with condensation of sample molecules
in the tube that connects a membrane probe to the mass spectrometer, the direct
insertion membrane probe was invented [5]. In this inlet the membrane is
mounted at the end of a probe which is inserted directly into the mass spectro-
metric ion source. The liquid or gas sample is then flushed through the probe.
The advantage of the inlet is the elimination of the connection tube between the
membrane and the ionizing region and volatile organic compounds with a boiling
point up to around 200°C can be measured. Further, the inlet has the advantages
that the sample can be chemically modified on-line prior to the detection and
calibration using an external standard is straight forward. The inlet has found
extensive use in connection with detection of volatile contaminants in water [7]
and with monitoring of growing microbial cultures [30].
The measuringcell is a small vessel (1-5 ml) mounted as close as possible to
the ionizing region of the mass spectrometer. It is interfaced directly to the mass
spectrometer via a sheet membrane that constitutes a part of one of the walls.
The sample, liquid or solid, is simply transferred to the vessel and after a few
minutes a MIMS spectrum of the volatile constituents can be recorded. When
the measuring cell is attached directly to the ion source [31] it can be used to
measure the same compounds as the direct insertion membrane probe.
An alternative to the close coupling of the membrane inlet to the ionizing
region of the mass spectrometer is found with the helium-purge membrane inlet
[32]. In this inlet a liquid or gas sample is flushed across one side of the
membrane, whereas the other side is continuously purged by a helium stream
that carries the permeated molecules to the ionizing region of the mass spectro-
meter. The purging of molecules from the membrane to the ion source reduces
the problem with condensation effects in connection tubes and the inlet has
almost the same applications as the direct insertion membrane probe.
In order to improve either selectivity or sensitivity, many devices have been
developed which combine the membrane inlet with another analytical chemical
technique for separation or concentrating. The two-stage inlets can be split into
systems using: two steps of membrane separation, a pre-separation or pre-
concentrating device before the membrane inlet; and systems using a post-
concentrating device after the membrane inlet. Most of the two-step devices
have only been introduced or briefly discussed in the literature and will not be

537
 described here. One combination that is frequently used is the combination of a
helium-purge membrane inlet and a jet separator [33]. In this combination the
jet separator is used for selective removal of helium from the purge stream
before it reaches the mass spectrometer. Examples have recently been published
where either a sorbent (Tenax) [34] or a cold trap [35] has been used after the
membrane inlet to concentrate the membrane permeate prior to the detection.
With these systems a considerable improvement in detection limits is to be
expected. The sorbent trap can also be used prior to the membrane inlet in
connection with gas analysis. Here, a pre-separation of related compounds can
even be achieved in addition to the preconcentration if the adsorbed molecules
are released from the trap using a controlled temperature gradient [36].
The five inlet types presented are designed for the measurement of gases
and/or volatile organic compounds and are only of limited use for the measure-
ment of less volatile (b.p. >200°C) organic compounds. In order to measure such
compounds with a membrane inlet, it is necessary to stimulate their evaporation
from the membrane surface. Currently, three techniques are in use: the trap-
and-release technique [16,17], laser desorption [19], and desorption chemical
ionization [20]. In the trap-and-release technique the sample (gas or liquid) is
passed through the membrane inlet and the less volatile organic compounds dif-
fuse into the membrane, but do not evaporate from it. After a sampling period
the membrane is rapidly heated and the less volatile organic compounds evapo-
rate from the membrane into the mass spectrometer. The desorption chemical
ionization method is an improved version of the trap-and-release method, where
the membrane is positioned in the centre of a chemical ionization plasma. The
plasma assists in the liberation of nonvolatile compounds from the membrane
during the heating process by ionizing the analyte molecules directly at/on the
membrane surface. With this method it is possible to measure compounds like
polyaromatic hydrocarbons and estrogenic compounds like phthalates at ng/l
concentrations in water, whereas many steroid hormones and pesticides can be
detected at low pg/l concentrations [22]. In laser desorption MIMS the evapora-
tion of less volatile organic compounds from the membrane is stimulated by
irradiation of the membrane surface with laser light. This technique has success-
fully been applied to measure polyaromatic hydrocarbons in water at low ng/l
concentrations.

16.4 SAMPLING FROM GASES

16.4.1 Characterization

Measurement of environmentally significant compounds directly from air has

become an important application of MIMS; a clear increase of publications in
this area has been seen since 1990 [11,18,26,37-42]. Table 16.1 summarizes the
typical performance characteristics of a simple membrane inlet mass spectrome-
ter in air monitoring. The detection limits shown are not necessarily the lowest

538
TABLE 16.1
Detection limits, response times and linear ranges measured by MIMS for a selected set of
environmentally significant compounds from air samples (adapted from Refs. [37,38])
Compound Response time Detection limit Linear range
(s)a
3
(Lg/m )b (Ag/m3 )
Benzene 1 1 1-10 000
Toluene 1 0.5 1-10 000
Xylenes 2 2 2-10 000
Chlorobenzene 2 0.5 2-10 000
Chloroform 1 5 5-30 000
Carbon tetrachloride 1 3 10-20 000
1,2-Dichloroethene 1 1 5-10 000
1,1-Dichloroethane 1 1 1-5 000
1,1,1-Trichloroethane 2 4 10-30 000
1,1,2,2-Tetrachloroethane 3 1 1-5 000
Tetrachloroethene 1 0.5 1-10 000
Dimethyl sulfide 1 1 1-40 000
Dimethyl disulfide 2 2 2-40 000
Ethanethiol 2 5 5-50 000
aMembrane (polydimethylsiloxane) thickness 25 tm. Response time is defined as the time it
takes for the signal to rise from 10 to 90%. bS/N = 3.

ones published, but they are levels that should be relatively easy to obtain with a
flow-through membrane inlet and a new mass spectrometer. Many of the VOCs
can be measured at low /tg/m 3 levels directly from air. As an example of the
results, single ion monitoring of tetrachloroethene close to the detection limit is
shown in Fig. 16.3. The response times that can be obtained using a MIMS
system (seconds) are very good when measurements are done from air using a
thin membrane (25 /tm). Table 16.1 also shows that wide linear dynamical
ranges are obtained, typically 3-5 orders of magnitude. The reliability of the
quantitative results has also been confirmed by comparing the total VOC-
contents measured by MIMS to those obtained with an on-line FID-analyzer in
analysis of paintshop exhaust air [39], the agreement of the results was very
good.
16
14
12
10

Fig. 16.3. The response of tetra- E

chloroethene (m/z 166) in the < 6
analysis of air in which two suc- 4
cessive samplings were made at 2
two different concentration levels -

(R.A. Ketola, private communica- 0

tion). Time, minutes

539
 In addition to the typical VOCs listed in Table 16. 1, it has been demonstrated
that MIMS can be used to measure organometallic (molybdenum hexacarbonyl
and ferrocene) [40], semivolatile [18] and polar [41] organic compounds directly
from air. However, because the polydimethylsiloxane membrane discriminates
against polar compounds it was proposed that direct sampling glow discharge
mass spectrometry is better suited for polar compounds [41]. Semivolatiles, such
as malathion, nitrobenzene, 2-chlorophenol, and dimethyl methylphosphonate
could be measured at ppb levels or lower [18].
A very innovative application of MIMS in air monitoring was a study by Colo-
rado et al. [42]. They built an automatic MIMS apparatus utilizing a membrane
and cryogenic concentration unit for monitoring isoprene from greenhouse air
below ppb levels. The necessary selectivity for the measurement was obtained
using chemical ionization with vinyl methyl ether (VME) reagent gas and tan-
dem mass spectrometry. VME reacts with isoprene to form an adduct ion at m/z
94 corresponding to [isoprene+VME-MeOH] +.Under collision induced dissocia-
tion the ion m/z 94 fragments by methyl radical loss to ion m/z 79, on-line moni-
toring this single reaction gave the necessary selectivity and sensitivity for the
on-line greenhouse air monitoring experiments performed.
Figure 16.4 shows an example of the results obtained in the study where a
method called temperature-programmed desorption membrane inlet mass
spectrometry (TPD-MIMS) was introduced [36]. In this technique VOCs are con-
centrated, typically from air, into an adsorbent material, and subsequently de-
sorbed using a rapid temperature program and detected with a MIMS apparatus.
The method allows separation of mixtures within a few minutes and detection at
sub-nanogram levels.
Monitoring of fermentation products/off-gases (e.g., N2, O,, CO, and ethanol)
from outlet gas of fermentors is a very important application of MIMS [30,
0.5

0.4

U 0.3

0.2

0.1

0
0 1 2 3
Time, min
Fig. 16.4. Separation of a four component mixture using the TPD-MIMS technique. Compounds
in the mixture are: 1 dichloromethane (6 ng, m/z 84), 2 trans-1,2-dichloroethene (7.8 ng, m/z 61),
3 chloroform (5.5 ng, m/z 83) and 4 tetrachloroethene (3.6 ng, m/z 166). The volume of the air
sample was 25 ml and the adsorbent material used was HayeSepD. (Reproduced with permission
of John Wiley & Sons Limited from R.A. Ketola, G. Gron, F.R. Lauritsen, Temperature-
programmed desorption for membrane inlet mass spectrometry, Rapid Commun. Mass
Spectrom., Vol 12, p. 773-778, Copyright 1998 by John Wiley & Sons Limited.)

540
43-47]. The use of adsorbents has also been adapted with MIMS methods to
increase selectivity and detection limits [36,48]. Another interesting demonstra-
tion of the wide applicability of MIMS is on-line monitoring of industrial solvents
(e.g., styrene, 1,1,1-trichloroethane and toluene) in breath [49].

16.4.2 Gas-phase reactions

To our knowledge, MIMS has not hitherto been used to follow gas phase
reactions-a potentially very good application of MIMS. The paper which comes
closest to this is a study in which MIMS was used to introduce monochloramine,
a reactive initiator of chlorination reactions, into the ion source of a mass
spectrometer [50].

16.5 SAMPLING FROM LIQUIDS

16.5.1 Water

16.5.1.1 Characterization
Measurement of volatile organic compounds directly from aqueous phase is the
standard application of MIMS [8,14,15,26,30,32,37,44-47,51-60] and recent
work has shown that many semivolatile organic compounds can also be mea-
sured reliably [20-22,61-65]. Table 16.2 summarizes the typical performance
characteristics of MIMS in direct measurement of VOCs and SVOCs from water.
The detection limits given for VOCs are not necessarily the lowest published val-
ues, but are levels that are relatively easy to obtain with most MIMS systems
using a flow-through membrane inlet and a new mass spectrometer. Many of the
VOCs can be measured at ng/l (pptr, parts-per-trillion) levels directly from aque-
ous phase and in an exceptional case detection of certain compounds at high pg/l
(ppq, parts-per-quadrillion) have been demonstrated, as shown in Fig. 16.5.
Detection limits of SVOCs are much higher with standard MIMS methods, but
with the newly developed trap-and-release methods (T&R-MIMS) SVOCs can be
measured at ,ugfl levels (ppb, parts-per-billion) [21,22,61]. Figure 16.6 shows
that steroid hormones can be reliably measured using desorption chemical ion-
ization membrane inlet mass spectrometry (DCI-MIMS) [21], a recently intro-
duced technique combining trap-and-release MIMS and chemical ionization.
Response times in the MIMS method are typically in the range of 1-3 min when
VOCs are measured from water using a 100 m thick silicone membrane. In the
trap-and-release method a pre-set concentration time is used before the analytes
are desorbed from the membrane. The complete analysis time (5-30 min) is
therefore more important than membrane response times.
Table 16.2 also shows that with standard MIMS a wide linear dynamical
range is obtained for VOC analysis, typically 3-5 orders of magnitude, whereas
the trap-and-release method only has a linear dynamical range of 2-3 orders of
magnitude. The reliability of the quantitative results obtained by MIMS has

541
TABLE 16.2
Detection limits, response times and linear ranges measured by MIMS for a selected set of
environmentally significant compounds from aqueous samples (adapted from Refs. [21,22,
37,611)
Compound Response Detection limit Linear range
time (min)' (ig/l)d (jig/l)
VOCsa
Benzene 2.0 0.1 0.1-10 00
Toluene 1.5 0.1 0.3-2 000
Xylenes 2.0 0.1 0.1-5 000
Chlorobenzene 1.5 0.2 ND'
Chloroform 1.5 0.1 0.1-1 000
Carbon tetrachloride 2.0 0.1 0.1-1 000
1,2-Dichloroethene 1.0 0.1 ND'
1,1-Dichloroethane 1.0 0.4 NDe
1,1,1-Trichloroethane 1.5 0.1 0.4-1 0OO
e
Dimethyl sulfide ND 0.5 ND'
Dimethyl disulfide NDe 0.5 NDe
Ethanethiol ND' 0.1 ND'

SVOCsb Standard T&R-MIMS

DDT [61] NA f 1000 25 ND'
Phenoxyacetic acid [61] NAf 10 000 100 ND'
Fluoranthene [61] NA' 25 5 ND'
Phenanthrene [61] NA f 4 0.5 ND e
Caffeine [61] NA f Not detectable 600 1 000-1 000 000
DCI-MIMS
Testosterone [21] NA' 30 ND'
Testosterone acetate [21] NA f 3 ND'
Progesterone [21] NA f 16 30-6 000
Levonorgestrel [21] NA' 30 NDe
Ethynylestradiol [21] NA f 150 ND'
Fluoranthene [22] NA 0.2 ND'
Phenanthrene [22] NAf 0.02 ND'
Atrazine [22] NA 5 ND'
Aldrine [22] NA 5 ND'
aAdapted from Ref. [37]. bAdapted from Refs. [21,22,61]. CFor VOCs membrane thickness 100
im. Response time is defined as the time it takes for the signal to rise from 10 to 90%. dS/N = 3.
eND = Not determined. fNA = Not applicable.

been confirmed by comparing MIMS, purge-and-trap gas chromatography-mass

spectrometry and static headspace gas chromatography for analysis of VOCs
from water samples [54] and for T&R-MIMS and liquid chromatography in the
analysis of caffeine from tea and coffee [61]. Agreement of the results obtained
by the various methods was very good in both of the studies.

542
 ?n

+
0

E
a

Scan number (5 sec/scan)

Fig. 16.5. The response of trans-dichloroethylene (m/z 96 and 98) in the analysis of water in
which samplings were made at two different concentration levels, 1 pptr and 500 ppq. (Reprinted
with permission from M. Soni, S. Bauer, J.W. Amy, P. Wong, R.G. Cooks, Direct determination of
organic compounds in water at parts-per-quadrillion levels by membrane introduction mass
spectrometry, Analytical Chemistry, Vol 67, pp. 1409-1412, Copyright 1995 American Chemical

A _ are~'"
1.0 n
{D)

2.5 - extract extract 3

o 0.8 extract2 m/z 313
2.0 mih 279
M 0.6

O 0.4 .0
< 1.0 -
a 0.2 0.5 -
blank
0.0 0.0
270 280 290 300 310 320 0 100 200 300 400
m/z Time (cycles)

Fig. 16.6. (a) DCI-MIMS mass spectrum of a birth control pill and (b) multiple ion monitoring
data for extracts of three birth control pills containing the synthetic sex hormones ethynyl-
estradiol (30 gLg,m/z 279 water loss from protonated molecule) and levonorgestreol (75 ag, m/z
313 protonated molecule) dissolved in 20 ml of distilled water and analyzed as is. (Reproduced by
permission of The Royal Society of Chemistry from F.R. Lauritsen, J. Rose, Determination of
steroid hormones by membrane inlet mass spectrometry and desorption chemical ionization,
Analyst, Vol 125, p.1577-1581, Copyright 2000 by The Royal Society of Chemistry.)

An important aspect of the MIMS measurement is identification of the

compounds to be measured, especially since a standard MIMS measurement
does not include a separation step. Tandem mass spectrometry (MS/MS) has, in
many cases, provided the necessary identification capability to the MIMS meth-
od developed. For example, the MS/MS technique has been used in identification
of fermentation products of Bacillus polymyxa organism [66], metabolites of
white rot fungus Bjerkanderaadusta [67] and metabolites of parasitic flagellates
Trichomonos vaginalis [68,69]. Interestingly, the differences in the permeation

543
 snn as_

180

160

140

120

0 100

80

60

40

20

Fig. 16.7. On-line MIMS measurement of the waste water stream of an oil refinery. (Reproduced
with permission of VSP International Science Publishers from T. Kotiaho, R. Kostiainen, R.A.
Ketola, T. Mansikka, I. Mattila, V. Komppa, T. Honkanen, K. Wickstrim, J. Waldvogel, O. Pilvi6,
Development of a fully automatic membrane inlet mass spectrometric system for on-line
industrial waste water monitoring, Proc. Control and Qual., Vol 11, pp. 71-78, Copyright 1998 by
VSP International Science Publishers.)

rates were also used to aid the identification of the metabolites of Trichomonos
vaginalis [68]. Recently, it has also been shown that mathematical methods can
be used for identification of the individual compounds based on the electron
ionization multicomponent mass spectra measured by MIMS [70].
One of the most interesting recent applications of MIMS is on-site and
on-line monitoring of environmentally significant compounds from water. A cus-
tom-made MIMS apparatus to be used in a mobile laboratory has been built for
on-site measurements [52], a specialized MIMS system has been designed for
on-line measurement of VOCs of river water in a boat [51] and for on-line
wastewater monitoring of chemical plants (Fig. 16.7) [55], a comprehensive kit
for analytical field work utilizing a GC/MS-instrument equipped with a
membrane inlet has been constructed [71] and a submersible membrane inlet
GC/MS-instrument has been built for marine pollution measurements [72].
Other interesting instrumental developments have been, e.g., development of a
cryotrap MIMS for measurement of VOCs at low pptr levels [35] and demonstra-
tion that polar organic compounds can be measured at ppb levels using a
microporous membrane combined with glow discharge ionization [73]. Some
nonconventional MIMS applications published are measurement of active prin-
ciples in tisanes of vegetable drugs [74], studies of aroma fraction of wines [75],
and classification of cola beverages [76]. The latter study clearly demonstrated
that MIMS (Fig. 16.8) is an excellent alternative or complementary method to
the electronic nose sensors used for classification of foodstuffs.

544
 2 . 4

1.5 ; c- Classes

x
0 2
- I x
x x x 3
0 *0 +4
05

0 0 7
7 8

.. *~~~~~~~~ A 9
;> < 10
D >11
** * 12
-1 It* *13*
Zip a 13

-1.5 0

-6 -4 -2 0 2 4 6 8 10 12
1. principal component

Fig. 16.8. Separation of thirteen cola beverages according to their first two principal components.
(Reproduced with permission of John Wiley & Sons Limited from R.A. Ketola, J. Heikkonen, S.
Piepponen, F.R. Lauritsen, T. Kotiaho, Classification of Cola beverages on the basis of mass
spectra measured by membrane inlet mass spectrometry, Rapid Commun. Mass Spectrom. Vol
12, pp. 1011-1017, Copyright 1998 by John Wiley & Sons Limited.)

16.5.1.2 Monitoring of reactions

On-line measurement of products of fermentations and other biological reac-
tions is the main application area of this subsection [31,66,77-81]; many reviews
have summarized this topic [8,15,30,44-47,59]. For example, an automated
MIMS system has been built for measurement of phenoxyacetic acid in penicillin
fermentations (Fig. 16.9) [77] and for measurement of ethanol, acetic acid,
2,3-butanediol and acetoin in Bacilluspolymyxa and Klebsiella oxytoca fermen-
tations [66], for control of a bioreactor for the anaerobic pre-treatment of indus-
trial waste water [79] and monitoring of a 9000 litre pilot plant fermentor [81].
MIMS has also been used to study biodegradation of trans- and cis-1,2-di-
chloroethylene [78,80] and to study production of halogenated aromatic com-
pounds formed by the white-rot fungus Bjerkandera adusta [31].
Another interesting area of reaction monitoring is the formation of halo-
genated inorganic amines [82-87]. As one of the few techniques MIMS can be
used to detect the unstable chloramines [82]. It has been shown that formation
of monochloramine and its conversion to dichloroamine and trichloroamine can
be followed on-line by MIMS (Fig. 16.10) [83], formation of organic chloramines
and brominated chloramines can be studied [84-85]. Recently, MIMS was used
in detailed studies of reactions of monochloramine with formaldehyde [86] and
for measurement of volatile byproducts formed from chlorination of organic-N
compounds [87].

545
 7
, EI
-6 0
2.0- *POAA Penicillin-V ,
-5 (
a. 1.5-
-4 o
to

Production phase / 04 g/l POAA 0.51 g/ilPOAA

medium added added added -3
, 1.0-
/JrJ -2 <
'
0.5
-1

0.0- 0_
50 100 150 200
Time (hours)

Fig. 16.9. On-line monitoring ofphenoxyacetic acid during a 200-h penicillin-V fermentation (A).
Penicillin-V was measured off-line using the hydroxylamine method (D). (Reprinted by permission
of Wiley-Liss, Inc., a subsidiary of John Wiley &Sons, Inc., from K.F. Hansen, F.R. Lauritsen, H.
Degn, An on-line sampling system for fermentation monitoring using membrane inlet mass
spectrometry (MIMS): Application to phenoxyacetic acid monitoring in penicillin fermentation,
Biotechnol. Bioeng. Vol 44, pp. 347-353, Copyright 1994 by John Wiley &Sons, Inc.)

Other examples of monitoring of aqueous phase reactions are measurement

of gas-exchange rates in the Belousov-Zhabotinskii reaction [88], on-line moni-
toring of reactions of epichlorohydrin [89], studies of photocatalytic reactions of
phenol and trichloroethylene [90] and benzyl acetate and 3,5-dimethoxybenzyl
acetate [91]. In addition, kinetics and mechanism of degradation of benzene
derivatives with Fenton's reagent have been studied by MIMS [92].

16.5.2 Organic matrixes

16.5.2.1 Characterization
A generic name, reversed-phase membrane inlet mass spectrometry, has been
adapted for MIMS methods in which water or organic compounds are measured
from organic matrix [9]. The name was borrowed from terminology used in
liquid chromatography, since typically MIMS is used to measure VOCs from
aqueous samples. The reversed phase MIMS papers published are dealing with
two different application areas, namely measurement of water activity in org-
anic liquids of low dielectric constant [93-95] and measurement contaminants of
organic solvents [9]. However, measurement of styrene and tetrachloroethylene
directly from olive oil [96] and measurement of Kathon CG (an antimicrobial
agent) from cosmetic emulsions [97] can also be considered reversed-phase
MIMS studies.
The method for determination of water activity developed by Degn et al.
utilizes a hydrophilic polyethylene terephthalate membrane instead of the nor-

546
 35 + 37
NH 2 CI / N C1+' 51
90"] (11-L,
30.6-
NH237CI+ 53

- 27.4- NH35CI2+ 85
35 37
NH CI CI+. 87

+ .
N35CI3 119
in
00.0
4.2- 35 37 +.
.. ~ N CI2 CI 121

35C2
+' 70

639.
35C137CI+. 72

looo 1500 2000 SCAMN

9820 1640 25100 33,20 TIM1E

Fig. 16.10. On-line monitoring of the reactions of monochloramine with HCI. Additions of HCl
were made at the times indicated by the arrows. Data were collected by scanning the full El mass
spectra (m/z 33-300). (Reprinted from T. Kotiaho, A.K. Lister, M.J. Hayward, R.G. Cooks,
On-line monitoring of chloramine reactions by membrane introduction mass spectrometry,
Talanta, Vol 38, pp. 195-200, Copyright (1991), with permission from Elsevier Science.)

mal hydrophobic silicone membrane [93-95]. The terephthalate membrane

allows easy passage of water and has very low permeability to nonpolar organic
solvents and atmospheric gases. The detection limits of water from various
organic solvents were at low ppm levels. Later, the method was simplified by
using a Penning vacuum gauge instead of a mass spectrometer to measure the
amount of water permeating through the terephthalate membrane [95].
Detection of contaminants in organic solvents was realized through the use
of a microporous polypropylene membrane [9]. With such a membrane the flow
of solvent into the ion source of the mass spectrometer is very high and the
vaporized solvent can be used to create conditions for chemical ionization with
the vaporized solvent acting as reagent gas. No sample enrichment is obtained in
the transport through the membrane, but this is compensated for by the much
higher flux. Low or sub ppm levels were obtained for several possible contami-
nants in hexane and toluene. As an example of the possible applications of the
method, it was tested in the analysis of additives and contaminants of gasoline.
the method showed great promise for these applications, especially together
with tandem mass spectrometry. Identification of tert-butyl methyl ether, a
gasoline additive, is presented as an example in Fig. 16.11.
16.5.2.2 Monitoringof reactions
The reversed-phase-MIMS methods described above have been used for on-line
monitoring of water activity in condensation reaction of 2,4-dimethyl-3-penta-

547
 5 i A N
A
LUo d 1.0 - b
0
o 0.8 o 0.8 -
0
0 6
= . c 0.6-

e 0.4-
.01
0 0.2-4
0.2 c 0.2 I

10
U0
I 1..
4
I 50 6b
I
f0
T
8U 0 100
V. V
'^
IU
.
^^
4U
I
.
^^ '^..
4
^I.
50
_I
^^
o
-^
o
^^
0
^^ '^^I
9 100
mlz m/z
Fig. 16.11. Collision induced dissociation spectra of m/z 89 obtained from (a) gasoline and (b) 100
ppm of tert-butyl methyl ether in hexane. (Reprinted with permission from F.R. Lauritsen, T.
Kotiaho, T.K. Choudhury, R.G. Cooks, Direct detection and identification of volatile organic
compounds dissolved in organic solvents by reverse phase membrane introduction tandem mass
spectrometry, Analytical Chemistry Vol 64, pp. 1205-1211, Copyright 1992 American Chemical
Society.)

none with phenylhydrazine in hexanol solution and hydrolysis of acetylchloride

in octane solution [93], and determination of the rate of hydrolysis of diphenyl
carbonate by porcine liver esterase as a function of water activity in diisopropyl
ether [94].

16.5.2.3 Combined HPLC-MIMS

The first demonstration of a membrane inlet being used as an LC/MS interface
was published in 1975 [98]. In the apparatus the solvent from the liquid
chromatograph was first evaporated using a flash evaporator and the resulting
gas stream was conducted into a three-stage Llewellyn-Littlejohn GC/MS, which
rejected most of the polar solvent used and allowed passage of the volatile
polyaromatic hydrocarbons analyzed into the ion source of the mass spectrome-
ter. In more recent studies it was demonstrated that an LC/MS apparatus
equipped with a membrane LC/MS interface can be used for analysis of small
halogenated solvents, BTEX compounds and other volatile aromatics [99,100].
The method developed was simpler than that of Jones and Yang [98], since an
eluent evaporation step was not used; instead the aqueous eluent stream was
directly conducted into a membrane inlet for mass spectrometric analysis.

16.5.3 Biological matrixes

16.5.3.1 Blood, bone and tissue

One of the earliest applications of membrane inlet mass spectrometry was in
vivo analysis of blood gases [3]. Woldring et al. capped the tip of a polyethylene
cannula (i.d. 0.4 mm, o.d. 1.1 mm) with a natural rubber membrane and
measured gases directly from circulating blood of animals. Subsequently other
researchers have utilized small cannula type of membrane inlets for blood gas
measurements [101,102], for measurement of halothane [103] or other volatile

548
organic compounds in blood [104] and for measurement of partial gas pressures
in bone tissue [102]. Most of these studies have been done using animals
[3,101-104], but interestingly subcutaneous tissue gas tensions have also been
measured for humans [105]. A recent application of MIMS is the demonstration
that NO can be measured from very small human plasma sample volumes
(10-300 Al) at gM levels [106]. In addition to the animal tissue gas tension
measurements, MIMS was shown to give valuable information about the gas
concentrations in living plants [28].
16.5.3.2 Urine
A MIMS method for determining the toluene metabolite p-cresol from urine has
been developed [107]. Urine was not directly analyzed, even though that is also
possible; instead the metabolite was measured from 1-propanol solutions after a
three-step sample preparation (hydrolysis, extraction, dissolution). Average
p-cresol concentration of unexposed workers was measured to be 11 Ag/ml.
16.5.3.3 Saliva and stomach fluid
Membrane inlet mass spectrometry has played a very import role in the studies
of changes in gases dissolved in the rumen fluid in situ and in vitro [108]. MIMS
has been used for measurement of dissolved gases (e.g. CH4, CO2, H2 and O,2)
directly in the rumen liquor of ten fistulated sheep on variety of diets. The in situ
MIMS experiments also allowed one to design laboratory experiments that
mimic the natural conditions much better than before.
MIMS has also been used to measure the time persistence of monochlor-
amine in human saliva and stomach fluid [109]. This information is very
important since monochloramine is widely used for disinfection of drinking
water and can be toxic. The study was done by monitoring on-line the decay of
monochloramine signal when an aliquot of a saliva or stomach fluid sample was
added to the monochloramine solution continuously circulated through a direct
insertion membrane probe. The conclusion of the study was that any toxicity
associated with the ingestion of monochloramine is most probably due to the for-
mation of disinfection byproducts rather than absorption of intact inorganic
chloramine(s).

16.6 SAMPLING FROM SEDIMENT AND PEAT

Measurement of dissolved gases (e.g., CH4, N2 , N2 0O,02 and CO,) in peat and
sediment samples with MIMS was shown to give valuable information about the
microbial physiology and processes (e.g., denitrification and methanogenesis) in
sediments and peat cores [29,110-113]. The advantage of the MIMS technique
compared to other methods is the possibility for simultaneous measurement of
multiple species [111]. Gas concentrations can be measured at 1 .M levels and
with spatial resolution of the order of 1 mm [112]. As an example vertical profiles
of dissolved oxygen, methane and carbon dioxide in a sub-core of a peat monolith
is shown in Fig. 16.12.

549
 CH4 (pm)
I

Fig. 16.12. Dissolved oxygen (O), methane (A), and carbon dioxide (0) in a a
sub-core of a peat monolith. Water table corresponds to the surface of the
peat. (Reprinted from D. Lloyd, K. Thomas, D. Price, B. O'Neil, K. Oliver,
T.N. Williams, A membrane-inlet mass spectrometer miniprobe for the 10
direct simultaneous measurement of multiple gas species with spatial
- -,l of
-inn T T,/rrnl__l MtA_ I, A T-1 nn 1 A 1 lnanvrht
-tU -
UIU -
u.tJ -
lle ---
IJ~IUVULI. -l- ll-U V Vl ,]l.'-m3 V,3, 0 CO,(Jm) so
(1996), with permission from Elsevier Science.) CO2 (pm)

16.7 SAMPLING FROM SOLIDS

16.7.1 Soil

Recent studies have shown that the use of a static [13] or a dynamic [12,114,115]
headspace method combined with MIMS is very suitable for measurement of
VOCs from soil samples, slurry samples or other types of solid samples, widening
the applicability of MIMS considerably. It has also been shown that MIMS can be
used to measure the VOC content of subsurface soil samples using a penetrom-
eter membrane interface probe [116].
The performance characteristics of the dynamic headspace membrane inlet
method, i.e. the purge-and-membrane mass spectrometry (PAM-MS) method
have been studied thoroughly [12,114,115]. One of the most important results of
these studies is the observation that for most types of real soil samples quanti-
tation can be done using only one standard soil sample. This is due to the fact
that in principle in the PAM-MS analysis the entire amount of a particular VOC
is measured, since it is completely purged out of the sample, i.e. quantitation is
matrix independent. The PAM-MS method is a sensitive method with detection
limits of chlorinated and aromatic solvents in the range of a few /tg/kg, the
linearity is 3-4 orders of magnitude and relative standard deviations in determi-
nations of VOCs from soil samples is around 5%. The potential of the PAM-MS
method for fast screening of soil samples was proven by analysis of many real
environmental samples and comparing the results with data obtained by head-
space gas chromatographic methods in two different laboratories (Fig. 16.13)
[115]. Good correlation between the various methods was obtained.
Another very interesting MIMS application is the simultaneous in vivo mon-
itoring of gases in plants and soil around the plants [28]. The experimental
set-up of a multi-capillary membrane inlet system and an example of the results
obtained are shown in Fig. 16.14. The capillary membrane inlets are mounted
either into the soil or into the plant cavities. The gas concentrations in the soil

550
 1000
0n -
800
o 700
pi 600
E 500
a 400
=) 3000
200
Fig. 16.13. Correlation of the 100
quantitative results obtain- 0
ed by HSGC and PAM-MS 200 400 600 800 1000
method. PAM, mg/kg

a) b) 4 Soil:brownforestsoil Plant:cornI
N,
3

1 0CO,

0-
lobo 2obo
S Soil: brownforestsoil Plant:tomato
CO,
.2
.0 0.8; u, N
-0
S
1.4 0o,
oo
0.
00 2doo 3doo00
O6- Soil: slightlyacidicsandysoil Plant:corn
W
a CO,
2.
8:
a 41 lo A, Nz ____ .k___
0-
10UU 2000 3000
Time (h)

Fig. 16.14. (a) The experimental set-up for in vivo monitoring of gases in plants and soils; (b)
measured gas concentrations in three different types of soils. The plant grown was corn or
tomato. (Reproduced with permission of John Wiley &Sons Limited from S. Bohdtka, Process
monitoring in fermentors and living plants by membrane inlet mass spectrometry, Rapid
Commun. Mass Spectrom. Vol 11, p. 656-661, Copyright 1997 by John Wiley &Sons Limited.)

have been observed to be characteristics of life in the soil and of the quality of the
soil.

16.7.2 Adsorbents

Adsorbents together with MIMS are used, as with other techniques, for sample
concentration before the concentrated sample is directed into the mass spectro-
meter [34,36,48]. This set-up can also be used to determine breakthrough curves
for various trapping materials and a thorough study of the stability of common
adsorbents such as Chromosorb, Carboxen, Tenax and Carbosieve has been
conducted [117].

551
16.7.3 Characterization of polymers

During the various development steps of MIMS, many different polymers have
been tested for their potential use as membranes in MIMS [118-121]. Polymers
tested include materials such as polyethylene [118,119], regenerated cellulose
[118], polypropylene [118], silicone rubber [118], polyethylene vinyl acetate
copolymer [118], poly(dimethylsiloxane) [119,120], latex [119], polyurethane
[119], nitrile [119], poly(vinyl) chloride [119], polyimide [119], Teflon [119],
poly(dimethylsiloxane) with zeolite [120], polyvinyl alcohol [120], and poly-
etherimide (polyester)/silicone composite membrane [121]. As an outcome of
these studies transport parameters such as the distribution ratio, the diffusion
constant and the permeability constant for many analytes diffusing through
various polymers have been established. In practice this possibility of determin-
ing transport parameters through polymers has been used to characterise the
transport of flavour compounds through polymer packing materials [122].
From a MIMS point of view, the study of many polymer materials has not
resulted in major progress. The poly(dimethylsiloxane) is still the best all-round
membrane for gas and VOC analysis, although Teflon offers a very good alterna-
tive for gas analysis. However, recent studies with low vapour pressure liquids as
membranes [123] and zeolite membranes for isomer separation [124], have
demonstrated interesting scientific advancement in this field.

16.7.4 Other solid materials

A demonstration involving the combination of a pyrolyzer unit and a membrane

inlet mass spectrometer to measure formation of isoprene, styrene and two
isomers of limonene during pyrolysis of Kraton 1107 (an isoprene-styrene
co-polymer) [125] showed that many other types of solid materials can also be
studied using this combination. Subsequently, pyrolysis-MIMS was used to
study oxidative pyrolysis products of peptides [126], to characterize four patho-
genic bacteria [127] and for direct measurement of free radicals formed during
thermolysis of aromatic diazoamino compounds (ArNH-N2 Ar) [128]. The
PAM-MS method has been used for the analysis of residual solvents from solid
pharmaceutical products [129] and VOCs from building materials.

16.8 FUTURE PERSPECTIVES

In its present form, standard MIMS has reached a mature stage where it is hard
to imagine fundamental changes in the future. It is currently used routinely for
the detection of VOCs in liquid matrixes and every year new applications within
the analysis of gas and solid samples occur. There are still many unexplored
possibilities for the practical use of MIMS and here the simple sample handling,
the possibility for fast analysis and on-line monitoring are tempting advantages
for many potential users. From an industrial and environmental point of view

552
the technique might become very interesting along with the development of
compact and/or miniaturized MIMS systems for on-site monitoring purposes.
For example, at the 2001 American Society for Mass Spectrometry Conference
an underwater MIMS system that can operate autonomously or under user
control via a wireless rf-link was presented [130]. Such on-site MIMS systems
[131] are unique compared to other on-site sensors, because they allow analysis
of a relatively broad spectrum of compounds and the interference from matrix
components is relatively limited.
Recently, new MIMS methodologies have been developed that expand the
analytical capabilities of the MIMS technique with respect to the analysis of
larger and polar compounds. Steroid hormones, estrogenic compounds, pesti-
cides and polyaromatic hydrocarbons are currently detectable at relatively low
levels using stimulated desorption from the membrane surface. When fully
developed, these new systems might set new standards for on-site equipment.

Acknowledgements

TK thanks Raimo Ketola and Marja Ojala for the help in preparation of the
figures. The Danish Center for Water Quality Sensors (VAKS) and the National
Technology Agency (Tekes) and acknowledged for financial support.

REFERENCES

i G. Hoch and B. Kok, Arch. Biochem. Biophys., 101 (1963) 160-170.

2 P.M. Llewellyn and D.P. LittleJohn, US Patent 1969, 3429105.
3 S. Woldring, G. Owens and D.C. Woolford, Science, 153 (1966) 885-887.
4 M. Reuss, H. Piehl and F. Wagner, Eur. J. Appl. Microbiol., 1 (1975) 323-325.
5 M. Bier and R.G. Cooks, Anal. Chem., 59 (1987) 597-601.
6 F.R. Lauritsen, Int. J. Mass Spectrom. Ion Process., 95 (1990) 259-268.
7 T. Kotiaho, J. Mass Spectrom., 31 (1996) 1-15
8 F.R. Lauritsen and D. Lloyd, in: C. Fenselau (Ed.), Mass Spectrometry for the
Characterization of Microorganisms. ACS Symposium Series 541, American
Chemical Society, 1994, pp. 91-106.
9 F.R. Lauritsen, T. Kotiaho, T.K. Choudhury and R.G. Cooks, Anal. Chem., 64 (1992)
1205-1211.
10 M.E. Cisper, C.G. Gill, L.E. Townsend and P.H. Hemberger, Anal. Chem., 67 (1995)
1413-1417.
11 R.A. Ketola, T. Mansikka, M. Ojala, T. Kotiaho and R. Kostiainen, Anal. Chem., 69
(1997) 4536-4539.
12 R. Kostiainen, T. Kotiaho, I. Mattila, T. Mansikka, M. Ojala and R.A. Ketola, Anal.
Chem., 70 (1998) 3028-3032.
13 M.A. Mendes, R. Sparrapan and M.N. Eberlin, Anal. Chem., 72 (2000) 2166-2170.
14 N. Srinivasan, R.C. Johnson, N. Kasthurikrishnan, P.W. Wong and R.G. Cooks,
Anal. Chim. Acta, 350 (1997) 257-271.
15 R.C. Johnson, R.G. Cooks, T.M. Allen, M.E. Cisper and P.H. Hemberger, Mass Spec.
Rev., 19 (2000) 1-37.
16 M. Leth, F.R. Lauritsen, Rapid Commun. Mass Spectrom., 9 (1995) 591-596.
17 G. Matz and P. Kesners, Analusis Magazine, 23 (1995) M12-16.

553
18 M.E. Cisper and P.H. Hemberger, Rapid Commun. Mass Spectrom., 11 (1997)
1449-1453.
19 M.H. Soni, J.H. Callahan and S.W. McElvany, Anal. Chem., 70 (1998) 3103.
20 R.A. Ketola and F.R. Lauritsen, Rapid Commun. Mass Spectrom., 13 (1999) 749-751.
21 F.R. Lauritsen and J. Rose, Analyst, 125 (2000) 1577-1581.
22 T. Aggerholm and F.R. Lauritsen, Rapid Commun. Mass Spectrom., 15 (2001)
1826-1831.
23 F.R. Lauritsen, T. Kotiaho and D. Lloyd, Biol. Mass Spectrom., 22 (1993) 585-589.
24 S. Chauvatcharin, K.B. Konstantinov, K. Fujiyama, T. Seki and T. Yoshide, Microb.
Util. Renewable Resour., 9 (1996) 493-506.
25 K.F. Hansen, S. Gylling and F.R. Lauritsen, Int. J. Mass Spectrom. Ion Process., 152
(1996) 143-155.
26 M.A. LaPack, J.C. Tou and C.G. Enke, Anal. Chem., 62 (1990) 1265-1271.
27 M. Bier, T. Kotiaho and R.G. Cooks, Anal. Chim. Acta, 231 (1990) 175-190.
28 S. Bohdtka, Rapid Commun. Mass Spectrom., 11 (1997) 656-661.
29 J. Benstead and D. Lloyd, FEMS Microbiol.Ecol., 13 (1994) 233-240.
30 F.R. Lauritsen, in: J.F.M. Van Impe et al. (Eds.), Advanced Instrumentation, Data
Interpretation,andControl of BiotechnologicalProcesses.Kluwer, 1998, pp. 67-103.
31 F.R. Lauritsen and A. Lunding, Enzyme Microbiol. Technol., 22 (1998) 459-465.
32 L.E. Slivon, M.R. Bauer, J.S. Ho and W.L. Budde, Anal. Chem., 63 (1991) 1335-1340.
33 L.E. Dejarme, S.J. Bauer, R.G. Cooks, F.R. Lauritsen, T. Kotiaho and T. Graf, Rapid
Commun. Mass Spectrom., 7 (1993) 935-942.
34 A.A. Rivlin, Rapid Commun. Mass Spectrom., 9 (1995) 397-399.
35 M.A. Mendes, R.S. Pimpirn, T. Kotiaho and M.N. Eberlin, Anal. Chem., 68 (1996)
3502-3506.
36 R.A. Ketola, C. Gr0n and F.R. Lauritsen, Rapid Commun. Mass Spectrom., 12 (1998)
773-778.
37 T. Kotiaho, R.A. Ketola, M. Ojala, T. Mansikka and R. Kostiainen, Am. Env. Lab.,
3/97 (1997) 19-21.
38 R.A. Ketola, Method Development in Membrane Inlet Mass Spectrometry, Air
Analysis and Desorption Techniques. VTT Publications, 364, 1998, p. 90.
39 R.A. Ketola, M. Ojala, H. Sorsa, T. Kotiaho and R.K. Kostiainen, Anal. Chim. Acta,
349 (1997) 359-365.
40 M.E. Cisper and P.H. Hemberger, Rapid Commun. Mass Spectrom., 11 (1997)
1454-1456.
41 S.M. Gordon, P.J. Callahan, D.V. Kenny and J.D. Pleil, Rapid Commun. Mass
Spectrom., 10 (1996) 1038-1046.
42 A. Colorado, D.J. Barket Jr., J.M. Hurst and P.B. Shepson, Anal. Chem., 70 (1998)
5129-5135.
43 E. Pungor Jr., C.R. Perley, C.L. Cooney and J.C. Weaver, Biotechnol. Lett., 2 (1980)
409-414.
44 H. Degn, J. Microbiol.Meth., 15 (1992) 185-197.
45 E. Heinzle, J. Biotechnol., 25 (1992) 81-114.
46 E. Heinzle and M. Reuss (Eds), Mass Spectrometry in Biotechnological Process
Analysis and Control. Plenum Press, New York, 1987, 241 pp.
47 T. Kotiaho, F.R. Lauritsen, T.K. Choudhury, R.G. Cooks and G.T. Tsao, Anal. Chem.,
63 (1991) 875A-883A.
48 G. Matz, A. Walte, W. Miinchmeyer and H.-E. Rikeit, Soc. Automot. Eng. [Spec.
Publ.] SP-1161 (1996) 175-179.
49 H.K. Wilson and T.W. Ottley, Biomed. Mass Spectrom., 8 (1981) 606-610.
50 T. Kotiaho, B.S. Shay, R.G. Cooks and M.N. Eberlin, J. Am. Chem. Soc., 115 (1993)
1004-1014.

554
51 B.J. Harland and P.J. Nicholson, Sci. Total Environ., 135 (1993) 37-54.
52 V.T. Virkki, R.A. Ketola, M. Ojala, T. Kotiaho, V. Komppa, A. Grove and S. Facchetti,
Anal. Chem., 67 (1995) 1421-1425.
53 M. Soni, S. Bauer, J.W. Amy, P. Wong and R.G. Cooks, Anal. Chem., 67 (1995)
1409-1412.
54 R.A. Ketola, V.T. Virkki, M. Ojala, V. Komppa and T. Kotiaho, Talanta, 44 (1997)
373-382.
55 T. Kotiaho, R. Kostiainen, R.A. Ketola, T. Mansikka, I. Mattila, V. Komppa, T.
Honkanen, K. Wickstr6m, J. Waldvogel and O. Pilvio, Process Contr. Qual., 11 (1998)
71-78.
56 M. Ojala, R.A. Ketola and T. Kotiaho, LC-GC, 16 (1998) 1026-1032.
57 R.M. Alberici, R. Sparrapan, W.F. Jardim and M.N. Eberlin, Environ. Sci. Technol.,
35 (2001) 2084-2088.
58 T. Kotiaho, R. Kostiainen, R.A. Ketola, M. Ojala, I. Mattila and T. Mansikka, in: E.J.
Karjalainen, A.E. Hesso, J.E. Jalonen and U.P. Karjalainen (Eds.), Advances in Mass
Spectrometry. Elsevier, Vol. 14, 1998, pp. 501-527.
59 F.R. Lauritsen and T. Kotiaho, Rev. Anal. Chem., 15 (1996) 237-264.
60 P.S.H. Wong, R.G. Cooks, M.E. Cisper and P.H. Hemberger, Environ. Sci. Technol.,
29 (1995) 215A-218A.
61 F.R. Lauritsen and R.A. Ketola, Anal. Chem., 69 (1997) 4917-4922.
62 G. Matz, M. Loogk and F. Lennemann, J. Chromatogr. A, 819 (1998) 51-60.
63 G. Matz, G. Kibelka, J. Dahl and F. Lennemann, J. Chromatogr. A, 830 (1999)
365-376.
64 F.R. Lauritsen, M.A. Mendes and T. Aggerholm, Analyst, 125 (2000) 211-215.
65 M.A. Mendes and M.N. Eberlin, Analyst, 125 (2000) 21-24.
66 M.J. Hayward, T. Kotiaho, A.K. Lister, R.G. Cooks, G.D. Austin, R. Narayan and
G.T. Tsao, Anal. Chem., 62 (1990) 1798-1804.
67 H.C. Beck, F.R. Lauritsen, J.S. Patrick and R.G. Cooks, Biotechnol. Bioeng., 51
(1996) 23-32.
68 F.R. Lauritsen, L.T. Nielsen, H. Degn, D. Lloyd and S. Bohdtka, Biol. Mass
Spectrom., 20 (1991) 253-258.
69 D. Lloyd, F.R. Lauritsen and H. Degn, J. Gen. Microbiol., 137 (1991) 1743-1747.
70 R.A. Ketola, M. Ojala, V. Komppa, T. Kotiaho, J. Juujarvi and J. Heikkonen, Rapid
Commun. Mass Spectrom., 13 (1999) 654-662.
71 G. Matz, W. Schr6der, A. Harder, A. Schillings and P. Rechenbach, Field Anal. Chem.
Technol., 1 (1997) 181-194.
72 G. Matz and G. Kibelka, in Proceedingsof the PITTCON'98,New Orleans, Lousiana,
March 1-5, 1998, 1864 pp.
73 F.R. Lauritsen, T.K. Choudhury, L.E. Dejarme and R.G. Cooks, Anal. Chim. Acta,
266 (1992) 1-12.
74 F.F. Vincieri, N. Mulinacci, G. Mazzi, D. Favretto, M. D'Alpaos and P. Traldi, Eur.
Mass Spectrom., 1 (1995) 503-505.
75 D. Favretto, G. Grandis, G. Allegri and P. Traldi, Rapid Commun. Mass Spectrom.,
12 (1998) 1595-1600.
76 R.A. Ketola, J. Heikkonen, S. Piepponen, F.R. Lauritsen and T. Kotiaho, Rapid
Commun. Mass Spectrom., 12 (1998) 1011-1017.
77 K.F. Hansen, F.R. Lauritsen and H. Degn, Biotechnol. Bioeng., 44 (1994) 347-353.
78 F.R. Lauritsen and S. Gylling, Anal. Chem., 67 (1995) 1418-1420.
79 G. Matz and F. Lennemann, J. Chromatogr. A, 750 (1996) 141-149.
80 J.-P. Arcangeli, E. Arvin, M. Mejlhede and F.R. Lauritsen, Water Res., 30 (1996)
1885-1893.
81 R.C. Johnson, N. Srinivasan, R.G. Cooks and D. Schell, Rapid Commun. Mass

555
 Spectrom., 11 (1997) 363-367.
82 P.J. Savickas, M.A. LaPack and J.C.Tou, Anal. Chem., 61 (1989) 2332-2336.
83 T. Kotiaho, A.K. Lister, M.J. Hayward and R.G. Cooks, Talanta, 38 (1991) 195-200.
84 T. Kotiaho, M.J. Hayward and R.G. Cooks, Anal. Chem., 63 (1991) 1794-1801.
85 M. Gazda, L.E. Dejarme, T.K. Choudhury, R.G. Cooks and D.W. Margerum, Environ.
Sci. Technol., 27 (1993) 557-561.
86 E.J. Pedersen, E.T.Urbansky, B.J. Marinas and D.W. Margerum, Environ. Sci.
Technol., 33 (1999) 4239-4249.
87 C. Shang, W.-L. Gong and E.R. Blatchley III, Environ. Sci. Technol., 34 (2000)
1721-1728.
88 H. Degn and F.R. Lauritsen, J. Phys. Chem., 93 (1989) 2781-2783.
89 R.C, Johnson, K. Koch and R.G. Cooks, Ind. Eng. Chem. Res., 38 (1999) 343-351.
90 R.F.P. Nogueira, R.M. Alberici, M.A. Mendes, W.F. Jardim and M.N. Eberlin, Ind.
Eng. Chem. Res., 38 (1999) 1754-1758.
91 P.SH. Wong, N. Srinivasan, N. Kasthurikrishnan, R.G. Cooks, J.A. Pincock and J.S.
Grossert, J. Org. Chem., 61 (1996) 6627-6632.
92 R. Augusti, A.O. Dias, L.L. Rocha and R.M. Lago, J. Phys. Chem. A, 102 (1998)
10723-10727.
93 S. Bohatka and H. Degn, Rapid Commun. Mass Spectrom., 5 (1991) 433-436.
94 H. Degn, S. Bohdtka and D. Lloyd, Biotechnol. Techniques, 6 (1992) 161-164.
95 H. Degn and N.-U. Frigaard, Anal. Chim. Acta, 274 (1993) 355-359.
96 T. Kotiaho, S. Gylling, A. Lunding and F.R. Lauritsen, J. Agric. Food Chem., 43
(1995) 928-930.
97 D. Favretto, P. Traldi, C.A. Benassi and A. Bettero, Biol. Mass Spectrom., 20 (1991)
669-676.
98 P.R. Jones and S.K. Yang, Anal. Chem., 47 (1975) 1000-1003.
99 S. Ouyang, Y.H. Chen and Y. Xu, Anal. Chim. Acta, 337 (1997) 165-172.
100 S. Ouyang, Y. Xu and Y.H. Chen, Anal. Chem., 70 (1998) 931-935.
101 J.S. Lundsgaard, B. Jensen and J. Gronlund, J. Appl. Physiol.: Respirat. Environ.
Exercise Physiol., 48(2) (1980) 376-381.
102 T. Kiaer, B. Dahl and G. Lausten, J. Orthop. Res., 10 (1992) 807-812.
103 M. Roberts, E.T. Colton III, G. Owens, D.D. Thomas and G.M. Watkins, Med. Biol.
Eng., 13 (1975) 535-538.
104 J.S. Brodbelt, R.G. Cooks, J.C. Ton, G.J. Kallos and M.D. Dryzga, Anal. Chem., 59
(1987) 454-458.
105 W.E. Donavan and B. Myers, in: H.I. Bicher and D.F. Bruley (Eds), Oxygen Transport
to Tissue, Instrumentation,Methods and Physiology. Plenum Press, New York, 1973,
pp. 67-72.
106 E.V. Trushina, N.J. Clarke, L.M. Benson, A.J. Tomlinson, C.T. McMurray and S.
Naylor, Rapid Commun. Mass Spectrom., 12 (1998) 985-987.
107 A. Sturaro, L. Doretti, G. Parvoli and P. Traldi, Anal. Chim. Acta, 224 (1989)
119-122.
108 D. Lloyd, J.E. Ellis, K. Hillman and A.G. Williams, J. Appl. Bacteriol. Symp. Suppl.,
73 (1992) 155S-163S.
109 T. Kotiaho, J.M. Wood, P.L. Wick Jr., L.E. Dejarme, A. Ranasinghe, R.G. Cooks and
H.P. Ringhand, Environ. Sci. Technol., 26 (1992) 302-306.
110 K.L. Thomas and D. Lloyd, FEMS Microbiol. Ecol., 16 (1995) 103-114.
111 K.L. Thomas, D. Price and D. Lloyd, J. Microbiol. Meth., 24 (1995) 191-198.
112 D. Lloyd, K. Thomas, D. Price, B. O'Neil, K. Oliver and T.N. Williams, J. Microbiol.
Meth., 25 (1996) 145-151.
113 G. Cowie and D. Lloyd, J. Microbiol. Meth., 35 (1999) 1-12.
114 M. Ojala, I. Mattila, T. Srme, R.A. Ketola and T. Kotiaho, Analyst, 124 (1999)

556
 1421-1424.
115 M. Ojala, I. Mattila, V. Tarkiainen, T. Srme, R.A. Ketola, A. Miittanen, R.
Kostiainen and T. Kotiaho, Anal. Chem., 73 (2001) 3624-3631.
116 J. Costanza and W.M. Davis, Field Anal. Chem. Technol., 4 (2000) 246-254.
117 R.J.B. Peters and H.A. Bakkeren, Analyst, 119 (1994) 71-74.
118 L.B. Westover, J.C. Ton and J.H. Mark, Anal. Chem., 46 (1974) 568-571.
119 A.J. Maden and M.J. Hayward, Anal. Chem., 68 (1996) 1805-1811.
120 N. Kasthurikrishnan, R.G. Cooks and S. Bauer, Rapid Comm. Mass Spectrom., 10
(1996) 751-756.
121 R. Alberici, R. Sparrapan, M.E. Eberlin, D. Windmoller and R. Augusti, Anal.
Commun., 36 (1999) 221-223.
122 J.C. Tou, D.C. Rulf and P.T. DeLassus, Anal. Chem., 62 (1990) 592-597.
123 R.C. Jonhson, K. Koch, N. Kashuriksrishnan, W. Plass, J.S. Patrick and R.G. Cooks,
J. Mass Spectrom., 32 (1997) 1299-1304.
124 K.H. Bennett, K.D. Cook, J.L. Falconer and R.D. Noble, Anal. Chem., 71 (1999)
1016-1020.
125 R.J. Vreeken and R. Houriet, Anal. Chim. Acta, 313 (1995) 237-241.
126 A.D. Hendricker, F. Basile and K.J. Voorhees, J. Anal. Appl. Pyrol., 46 (1998) 65-82.
127 A.D. Hendricker, C. Abbas-Hawks, F. Basile, K.J. Voorhees and T.L. Hadfield, Int. J.
Mass Spectrom., 190/191 (1999) 331-342.
128. N. Kasthurikrishnan, R.G. Cooks and M.J. Thompson, J. Am. Soc. Mass Spectrom., 9
(1998) 234-241.
129 M. Ojala, M.Poutanen, I. Mattila, R.A. Ketola, T. Kotiaho and R. Kostiainen, Rapid
Commun. Mass Spectrom., 14 (2000) 994-998.
130 D.P. Fries, R.T. Short, G. Kibelka and M.L. Kerr, 49th ASMS Conference on Mass
Spectrometry andAllied Topics, Chicago, 2001.
131 G. Matz, W. Schr6der and T. Kotiaho, in: R.M. Meyers (Ed.), Encyclopedia of
Analytical Chemistry: Applications, Theory, and Instrumentation, Vol. 5. Wiley,
2000, pp. 3783-3804.

557
Chapter 17

Extraction techniques for solid samples

John R. Dean and Sarah L. Cresswell

17.1 INTRODUCTION

Extraction is normally a laboratory procedure prior to analysis. It is expected to

be performed in a reproducible manner, in the fastest possible time and at low
cost. Subsequent analysis is then supposed to capitalise on the effectiveness of
this process to generate good quality data. However, the weakest link in this
process is often neglected as unimportant and time-consuming. It might involve
a journey to a site of interest some distance away. The weather might be
unfavourable-wet, windy and cold. However, the sampling strategy performed
and the nature of the sample type probably have the greatest influence on the
result obtained than the rest of the process. Sampling inherently requires a
strategy of how best to obtain a representative sample from the location. Then,
how best to remove the sample; it is not necessarily a surface sample, but may be
sub-surface. Also, how to effectively store the sample, i.e., type of container and
method of preservation required. And, finally, how to safely transport it back to
the laboratory. It may be that some preliminary sample pre-treatment is
necessary, or that the sample requires drying at room temperature, or not for
fear of losing volatile compounds. All these stages are necessary to obtain a
homogenous and representative sample of appropriate size for the selected
extraction techniques.
The choice of extraction technique is often made on the basis of what is
available and what has been used before. The skill and experience of the analyst
is often key to the operation of extraction and analysis procedures. But why
should this be so? This rhetorical question is perhaps best explained by the
diversity of methods available in the scientific literature for the extraction of one
type of analyte from a soil matrix. If the entire process was straightforward and
uncomplicated then a standard method could be produced based on a definitive
method. However, anyone familiar with standard methods, e.g., U.S. EPA
methods for extraction, will realise that even the simplest of approaches, e.g.,
Soxhlet extraction, offers a choice of solvent(s) and extraction times to allow a
range of analytes to be effectively extracted from a range of matrices. Often this
is complicated further by the use of so-called instrumental approaches, e.g.,
ComprehensiveAnalytical Chemistry XXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 559
 Solid sample matrix, e.g. soil

Addition of solvent

[Heat pressuree

EXTRACTION TECHNIQUE |

Fig. 17.aMAE solid matrices.

L PFE(ASE)
of analyses from

Fig. 17.1. Extraction of analyses from solid matrices.

MAE/SFE/PFE, which might not only be vague in terms of solvent choice but
might have situations where the operation of the instrument requires some
intervention. These method variabilities are not the fault of the regulators who
produce the methods, but are necessary because of the complex interactions of
the analytes with it.
Analyte-matrix interactions obviously need to be overcome in order to
release the analyte and make it available for subsequent analysis. The nature of
these interactions is very dependent upon the type of molecule to be removed.
Organic molecules occur in different forms ranging from the non-polar (e.g.,
PAHs) through to more polar molecules (e.g., phenols). While the former may be
only weakly bound (Van der Waals forces) to the matrix, polar molecules may be
strongly bound and have the possibility of electrostatic attraction, depending
upon the pH of the matrix. In this way it is difficult to generalise on the exact
nature of extraction technique and solvent choice. This is where the skill and
experience of the analyst becomes important. Often the choice of solvent to be
used to remove an analyte from a matrix is based on past experience. That
assumes that the analyte(s) to be extracted are known. For unknown samples,
several extractions may be required with at least two different sample types to
assess the effectiveness of a particular solvent system.
Extraction of organic pollutants from solid environmental samples has
frequently been done using organic solvents with or without the addition of heat
(Fig. 17.1). This process is typified by the technique of Soxhlet extraction. A flow
diagram detailing a procedure for the extraction of analytes from soils using
Soxhlet extraction is shown in Fig. 17.2. Soxhlet extraction normally requires
large volumes (up to 150 ml) of often chlorinated solvent to be refluxed through
the solid sample for an extended period of time (6-24 h). While it is not
uncommon for the analyst to assemble several experimental set-ups at the same
time it can be labour-intensive at the start and end of the process. As the process
is time-consuming, alternative extraction strategies have been developed, based
on instrumentation that allows the process to be controlled more effectively.
This chapter mainly focuses on these newer instrumental approaches to
compare and contrast both their application base and robustness for extraction.

560
 Soxhlet apparatus assembled. Sample e.g. soil, accurately weighed
Solvent placed in round-bottomed I and mixed with similar weight of
flask on isomantle anhydrous sodium sulphate.

Sample +dispersion agent placed

in porous extraction thimble.

Heat applied via isomantle. Solvent circulation

rate is 4 cycles per hour (typically).
4
Sample refluxed for an appropriate amount
of time e.g. 6, 12, 18, 24 hours.

After completion of extraction solvent

containing analytes retained.

Concentration of analyte by
solvent evaporation (e.g.
rotary evaporator, solvent
blow-down) or SPE.
I~~~~~ \\\

=Analysis

Fig. 17.2. Flow diagram for Soxhlet extraction.

17.2 SUPERCRITICAL FLUID EXTRACTION

17.2.1 Theoretical aspects

A supercritical fluid can be defined by looking at a phase diagram for a pure

substance (Fig. 17.3). The regions where the substance occurs, as a consequence
of temperature or pressure, as a single phase, i.e. solid, liquid or gas, can be
clearly seen. The divisions between these distinct regions are bounded by curves
indicating the co-existence of two phases, e.g., solid-gas corresponding to subli-
mation, solid-liquid corresponding to melting, and liquid-gas corresponding to
vaporization. These three curves intersect at the triple point where the three

cP
gas

Fig. 17.3. Phase diagram for carbon dioxide. Temperature

561
TABLE 17.1
Critical temperature and pressures for selected solvents
Solvent Critical temp. (C) Critical pressure
Atmospheres psi
Carbon dioxide 31.1 74.8 1070.4
Nitrous oxide 36.6 73.4 1050.1
Water 374.4 224.1 3208.2
Xenon 16.7 59.2 847.0

phases co-exist in equilibrium. At the critical point, designated by both a critical

temperature (T.) and a critical pressure (P.), no liquefaction will take place on
raising the pressure and no gas will be formed on increasing the temperature. It
is this region which is, by definition, the supercritical region. Table 17.1 shows
the T and P values for a range of substances that are suitable for use in
analytical supercritical fluid extraction (SFE).
After consideration of Table 17.1 and the temperature and pressures
required, it can clearly be seen that the choice of supercritical fluid is often deter-
mined on the basis of practical issues, e.g., equipment required to establish
supercritical fluid conditions and/or the availability with high purity and low
cost of the material. However, the low critical temperature and moderate critical
pressure of carbon dioxide have made it the substance of choice for most analyti-
cal applications of SFE. The non-polar nature of CO2 (it has no permanent dipole
moment) means that for some applications the solvent strength of the supercriti-
cal fluid is inadequate for the task (see later).

17.2.2 Instrumentation

All SFE systems contain six basic components, namely a supply of high purity
carbon dioxide (CO2), a supply of high purity organic modifier, the pumps, the
oven for the extraction cell, the pressure outlet or restrictor, and the collection
vessel. A schematic diagram of a typical SFE arrangement is shown in Fig. 17.4.
The CO2 is supplied in a cylinder fitted with a dip tube that allows liquefied CO 2
only to be obtained. This is in contrast to the common cylinder type which

Fig. 17.4. Schematic diagram of apparatus

for SFE. or

562
comprises liquid CO2 at the bottom and vapour at the top (when positioned
vertically). Modified CO2 can be produced via the use of two pumps or via a single
cylinder containing CO2 and an organic modifier; the latter is the preferred
approach.
Two pumps are required to pump the pressurized CO2 and organic modifier
through the SFE system. After pumping out from the cylinder, CO2 and organic
modifier are mixed with a T-piece before introducing into the extraction cell.
The combination of the two pumps allows a degree of control in terms of modifier
composition (1-20% v/v) and flexibility of solvent choice, e.g., methanol, toluene.
The ideal pump should deliver a constant flow rate (ml min -') at a suitable pres-
sure (3500-10000 psi). Two kinds of pump-the reciprocating pump (piston
pump) and syringe pump-are usually used in SFE systems, but the former is
the most common due to its lower cost. However, when the reciprocating pump
is employed, external cooling of the pump head is required to prevent cavitation,
i.e., gas entrapment. No such modification of the pump head is required for the
modifier.
An oven is used to generate the critical temperature of the solvent. Located
within the oven is the sample cell or vessel. The temperature range of the oven is
ideally located between T. and 250°C. The sample vessel, which is typically made
of stainless steel, must be capable of safely withstanding high pressures (up to
10000 psi). It is essential to remember that some equilibration time is required
to achieve the pre-set temperature for the newly prepared and inserted sample-
containing vessel prior to starting the extraction process. Almost all commercial
extraction vessels are of the flow-through design. This allows fresh, unadulter-
ated supercritical fluid to pass over the sample.
The first commercial instruments relied on an inadequate method of pres-
sure regulation based on the use of narrow-bore silica or stainless steel tubing
which acted as the restrictor. These were largely superseded in the 1990s by
more robust methods of pressure regulation based on the use of 'moving ori-
fices', typified by the use of such terms as 'nozzles' or 'back-pressure regulators'.
For the extraction of real sample it is more appropriate to use a variable
restrictor.
Supercritical fluid extraction relies on the diversity of properties exhibited
by the supercritical fluid to (selectively) extract analytes from solid, semi-solid or
liquid matrices. The important properties offered by a supercritical fluid for
extraction are: (a) good solvating power; (b) high diffusivity and low viscosity,
and (c) minimal surface tension.
SFE can be operated in two modes: off-line and on-line. While a significant
part of the early work was done on utilising the compatibility of SFE to
chromatographic separation, the lack of robust instruments and the inflexibility
of such systems has probably precluded their continued usage. The situation is
reversed for off-line SFE. In this situation, the flexibility of off-line operation
allows the analyst to focus on the sample preparation only. This allows SFE to be
optimised to maximise analyte recovery, to process larger samples without fear

563
of overloading the chromatographic column, a certain amount of freedom of
choice pertaining to the analysis (e.g. GC, HPLC, IR etc.), and finally, that the
analytical measurement instrument is available to analyse other samples.
It would appear however that the initial growth in the popularity of SFE,
based on its environmentally friendly solvent (CO,), has largely been ignored in
recent years, at least in the scientific literature. However, SFE has found a niche
market. Three U.S. EPA methods exist for the SFE of PAHs (method 3561), total
petroleum hydrocarbons, TPHs (method 3560) and PCBs and organochlorine
pesticides (method 3562) from soils.

17.2.3 Methods of analysis

Many reviews on SFE have been published with an environmental bias [1-7]. It
is not the intention of this chapter to review the extensive application literature
that exists for SFE. Accordingly, a short table of applications has been provided
(Table 17.2) to highlight suitable analytes and conditions. Based on this and
other information a set of recommendations for the use of SFE are proposed.

TABLE 17.2

A summary of supercritical fluid extraction (SFE) using for extracting PAHs, PCBs and
pesticides from soils

Analyte(s) Sample types/ Operating conditions Recovery (%) Ref.

matrices

PAHs Sample: urban Sample size: 20-50 mg. Phenanthrene-d0,: 89-102 8

dust (SRM Fluid: CO2, NO2 , ethane, CO 2 pyrene-dl0: 67-74
1649); spiked with 5% methanol; NO 2 with perylene-d 0 : 28-44
river sedi- 5% methanol; Pressure: 300
ment. atm; Temperature: 45°C for
the pure fluid, 65°C for the
modified fluid. Extraction
time: 30-60 min.
PAHs, pesti- Sample 1: Sample size: 400 mg-1.5 g. SFE at 200 'C 9
cides, etc. agricultural Fluid: pure CO2; Pressure: 400 Pesticides from sample 1:
soil atm; Temperature: 50°C; 200°C; 110-119.
Sample 2: 350°C; Extraction time: PAHs from sample 2: 84-100
railroad bed 30 min. (11 out of 19 compounds),
soil 51-72 (8 out of 19 compounds).
Sample 3: PAHs from sample 3: 88-115 (6
diesel soot out of 19 compounds), 18-75
(13 out of 19 compounds)
PAHs Sample: active Sample size: 1 g. Fluid: CO2; (variable with the change of 10
contaminated Modifier: methanol, 0-20%. operating conditions)
2
soil; Pressure: 100-300 kg cm- ;
CONTEST Temperature: 40-100°C;
soil. Extraction time: 10-60 min.

continued

564
Analyte(s) Sample types/ Operating conditions Recovery (%) Ref.
matrices
PCB conge- Samnle: Fluid: pure CO. or CO. with quantitative 11
ners industrial soil methanol modifier (2-5%).
(CRM 481) Different conditions employed in
21 different laboratories.

OCPs Sample: Fluid: CO2 . Modifier: toluene; Typically, 87.5-100.6 12

spiked and acetonitrile; methanol.
real soil Pressure: 20 MPa; Tempera-
ture: 50°C; Extraction time:
10-min static plus 20-min dy-
namic, or 30 dynamic.
Pesticides Sample: Fluid: CO 2. Modifier: methanol, OCPs from Celite: 87.6-95.4 13
2
(herbicides, spiked Celite 0-10%. Pressure: 250 kg cm ; Triazines and urea herbicides
OCPs, and soil. Temperature: 50°C; from Celite: 77.0-91.7
OPPs) Extraction time: 15-min static, OPPs from Celite: 87.6-97.8
30-min dynamic. OPPs from soil: 93.6-104.4
(org. content 2.0%)
76.3-96.7 (org. content 15.0%)
42.9-47.7 (org. content 35.0%)
Hexa- Sample: Sample size: 4 g. Fluid: CO2 ; Typically, 50-90% (compared to 14
conazole weathered Modifier: methanol, 20%. Soxhlet)
2
soils. Pressure: 250 kg cm- ;
Temperature: 55 C;
Extraction time: 20 min.
Herbicides Sample: Sample size: 2 g. >90 15
spiked soil Fluid: CO 2 ;
Modifier: methanol
Temperature: 85 C;
Extraction time: 20 min.

Some abbreviations in this table: OCPs-organochlorine pesticides; OPPs-organophosphorus

pesticides; DCM- dichloromethane.

Recommendations for SFE: Equipment issues [7]

- Use an SFE system with two pumps.
- Choose the most appropriate modifier, combined modifier mixture or
reactive modifier.
- Use a variable restrictor, not a fixed linear restrictor.
- Choose an appropriate collection system, i.e., liquid-solid trap and/or
collection solvent.
- Maintain collection solvent at 5°C.

Recommendations for SFE: sample issues [7]

- Smaller particle size is important for increasing recovery.
- Use certified reference materials for method development, if possible. Alter-
natively extract from native contaminated samples.
- Pack the extraction cell appropriately, i.e., use drying agents and copper (for
soils with high sulphur content).

565
 SFE system operational and
connected to electric and gas
supply. A two-pump system is
required for extraction of polar
analytes.
Sample e.g. soil, accurately weighed
and mixed with similar weight of
drying agent and copper (for soils
with high sulphur content).

Sample +drying agent (+copper)

placed in extraction cell.

SFE conditions applied.

'
Pressure, 20-35 MPa; Temperature, 50-250 C; flow rate, 1-2 ml min CO2;
10% methanol required for polar analytes.
Static conditions maintained for 5-15 mins, then dynamic extraction for
appropriate time

Sample extracted for an appropriate amount of time e.g. 5 -

60 mins. Collection of analyses in solvent (at 5 °C) or a
liquid-solid trap.

i-1
Fig. 17.5. Flow diagram for supercritical fluid extraction.

Recommendations for SFE: method development [16]

- General rule: supercritical CO2 only will generally solvate GC-able analytes
at normal extraction conditions, such as, 400 atm and 50°C. For analytes
that are fairly polar or have high molecular masses, then the addition of an
organic modifier (10% v/v) may be necessary with an increase in temperature
(70°C). In the case of ionic analytes the addition of an ion-pairing reagent
may be beneficial.
- Optimization of SFE conditions: the main SFE operating variables are
temperature, pressure, modifier addition and flow rate of CO 2 . Optimization
can be achieved using a univariate (one at a time) approach or as preferred in
our own laboratory a multivariate approach using chemometrics [10].
A flow diagram detailing a procedure for the extraction of analytes from soils is
shown in Fig. 17.5.

17.3 MICROWAVE-ASSISTED EXTRACTION

17.3.1 Theoretical aspects
Microwave-assisted extraction is a technique which grew out of the late 1980s
microwave digestion work carried out by Gedye et al. [17]. Gedye and co-workers
realized that using microwaves to heat solvents or acids in closed vessels

566
resulted in temperatures far greater than the atmospheric boiling points of the
liquids. This elevated boiling phenomenon is due to the special way in which
microwaves induce heat.
On a standard hot-plate or heating mantle heating system, heat is trans-
ferred from the hot-plate, through the walls of the vessel and then into the liquid
contained within it. This is a relatively slow process, which results in a tempera-
ture gradient whereby the liquid is hottest at the walls of the vessel and coolest
in the centre.
In contrast, microwave heating works by exciting the molecules in the liquid
causing heating by either rotation of dipoles or charges within the molecules of
the liquid. Microwaves have a specific wavelength of 12.2 cm or a frequency of
2.45 GHz set so as not to interfere with radar and telecommunication networks.
The presence of an electric field generated by the microwaves would cause polar
molecules to align themselves with the field. However, since microwave fields
oscillate, the molecules align and relax with the field of electromagnetic radia-
tion. It is the energy dissipated during such movements, which leads to the
heating effect (Fig. 17.6).
The physical parameter, which describes the ability of a material to heat
when placed in the microwave field, is termed the dielectric loss coefficient. The
larger this value, the more polar the solvent and the greater the ability of the sol-
vent to transform electromagnetic energy into heat via mechanical motion. In
order for microwave heating to occur, one of the following three conditions
should be used [18]: (1) a single solvent or mixture of solvents that have high
dielectric loss coefficients; (2) a mixture of solvents with high and low dielectric
loss; (3) a microwave susceptible sample that has a high dielectric loss in a low
dielectric loss solvent.
The increase in temperature, which can be reached by microwave heating,
accelerates the rate at which extraction takes place by increasing the solubility
of the analytes in the extraction solvent. This means that the sample throughput
can be increased. In addition, since far smaller volumes of solvent are needed for
MAE as compared to Soxhlet methods, this reduces costs. Parr and co-workers,
who first showed how microwaves could be used in extractions from plant and
animal tissues, described the fundamental physical phenomena involved [19].
The disadvantage of this technology is the fact that the solvent needs to be physi-

Fig. 17.6. Conventional versus microwave- assisted

heating. Conventional heating: heat from the hot Conventional Heating Microwave-assisted
plate or heating mantle passes through the glass
base of the beaker and into the solution. The solu-
tion heats slowly causing a heat gradient where the
liquid is hotter at the bottom than the top. Micro-
wave-assisted heating: Microwaves pass through
the vessel walls directly into the polar liquid. A
Microwave interaction with dipolar molecules H TPLATE
causes rotation and vibration of the molecular
bonds causing heating.

567
cally removed from the sample matrix upon completion of the extraction prior to
analysis.

17.3.2 Instrumentation

The first use of MAE for the extraction of analytes with organic solvents
appeared in 1986 [17]. In MAE, both the organic solvent and the sample are
subjected to radiation from a magnetron and since microwaves are able to cause
rapid heating of samples, safety is an important issue.
Although domestic microwave ovens can be used for extraction, safety
considerations have led to a number of manufacturers producing microwave
ovens specially designed for laboratory use. The advantages of these dedicated
instruments over domestic ovens especially in terms of safety are great,
although they are considerably more expensive.
Initially, MAE was always performed using closed-vessels and involved
heating the solvents under pressure beyond their boiling point and consequently
these experiments carried a risk of explosion. For example, reactions that cause
decomposition using acids often produce gaseous by-products, which can result
in a rapid increase in pressure in a closed vessel. If a sample of a polymer
containing a large number of the same type of bonds is heated, when the bonds
break there will be a rapid release of carbon dioxide. It is because of this fact that
sample size is limited to less than 2 g in closed vessels.
This problem led researchers to develop microwave systems that could be
open to the atmosphere which, although limiting the temperature to which the
solvent could be heated, would handle larger sample sizes since gaseous release
would no longer be a problem. Such systems are referred to as "open-vessel
systems" and can be used for sample sizes greater than 2 g.
Laboratory microwave ovens are available in both closed-vessel style and
open-vessel style. The former is typified by the MES-1000 Microwave Solvent
Extraction System supplied by CEM Corp., USA (recently superseded by the
MARS5 system), and the latter by the Soxwave system from Prolabo Ltd.,
France (now, CEM Corp.). In addition, some laboratory microwave ovens such as
QLAB 6000 developed by Questron Technologies Corp., Canada, can be set for
closed or open vessel operation or flow-through. Figures 17.7 and 17.8 are sche-
matic diagrams of pressurised (closed) and atmospheric (open) MAE systems. As
an example of the current technology, the MARS5 system from CEM Corp.
allows up to 14 extraction vessels (XP-1500 plusTM ) to be irradiated simulta-
neously. Safety features include a function for monitoring both pressure and
temperature and the instrument is equipped with a solvent alarm to call atten-
tion to an unexpected release of flammable and toxic organic solvent. Pressure
(up to 800 psi) is continuously measured (measurements taken at the rate of 200
s-]), while the temperature (up to 300°C) is monitored for all vessels every 7 s.
The sample and solvent are placed into the inert vessel liner, which has a
volume of 100 ml and is made from a TFM fluoropolymer. The vessels are

568
Fig. 17.7. Pressurised-vessel
microwave system.

Fig. 17.8. Open-vessel micro-

waverrsvstm

equipped with an AutoVent Plus "' system, which allows venting of excess
pressure within each extraction vessel. This works by lifting of the vessel cap
slightly to release any excess pressure and then immediately resealing to prevent
loss of sample. If any solvent leaks from the extraction vessel(s), a solvent
monitoring system will automatically shut off the magnetron but leaves the
exhaust fan to continue working.
The use of microwaves for environmental sample preparation has increased
at a very fast rate over the past 15 years. Many traditional Soxhlet extraction
procedures were adapted to be carried out much faster and with much less
solvent in a microwave oven. The biggest hindrance to this process was the
reduced in sample size, which could be accommodated safely in a pressurised
microwave system.
However, with the advent of atmospheric microwave systems, much larger
sample sizes can be used and this, coupled with the lower limits of detection
offered by modern chromatographic equipment, has meant that most traditional
extractions can be performed in a microwave oven. To illustrate this, there are
now a number of EPA methods, based on microwave extraction, which would
previously have employed a different extraction technique such as Soxhlet.
Although microwaves have found a place in many routine environmental
analysis laboratories, there is a problem with the through-put capabilities of
these instruments. In comparison with Soxhlet they are certainly no worse, but
there is a growing need to further increase the number of sample preparations
which can be performed per day. The systems described above are batch pro-

569
cesses, i.e. each microwave vessel holds a single discrete sample. A maximum of
14 samples can be irradiated simultaneously and all vessels must be allowed to
cool before opening to prevent loss of volatile materials. If it were possible to
perform these extractions in a continuous system, the increase in through-put
would be vast.
CEM and Milestone (now part of CEM) have manufactured continuous flow
systems but the problem with these systems lies in the way in which the
magnetron produces the microwaves. The magnetrons on these systems work on
a duty cycle. This means that the time the microwaves are produced is varied
rather than the power of the microwaves. For example, to obtain 50% power, the
magnetron would be on full-power for 10 s and then off for 10 s so on average,
50% power. This is a problem for continuous flow samples, because if the sample
is passing through the microwave cavity during the power-off section of the duty
cycle, there will be no irradiation of that sample.
The problem caused by duty cycles in closed-vessel systems has been elimi-
nated by the production of the MARS5 instrument, which works on a variable
power not variable time magnetron. If this technology could be introduced into
continuous-flow systems, the impact on MAE would be great. Routine analyses
could be performed much quicker and would require much lower volumes of
solvents than used with traditional Soxhlet extraction methods causing consid-
erable cost reductions.

17.3.3 Methods of analysis

Many reviews on MAE have been published with an environmental bias [18,
20-21]. It is not the intention of this chapter to review the extensive application
literature that exists for MAE. Accordingly, a short table of applications has
been provided (Table 17.3) to highlight suitable analytes and conditions. Based
on this and other information a set of recommendations for the use of MAE are
proposed.

Recommendations for MAE: equipment issues

- Magnetron should produce variable power rather than work on a duty cycle
to ensure complete extraction of samples.
- Introduction of a continuous-flow microwave extraction system would
dramatically reduce the volumes of solvents required and dramatically
increase the throughput of samples.
- Microwave vessels must be transparent to microwaves and of a substance
inert to solvents.
- Microwave cavity should be fitted with a solvent leak detector for safety.

Recommendations for MAE: sample issues

- Closed system for small sample sizes (<2 g).
- Atmospheric system for large sample (>2 g).

570
TABLE 17.3
A summary of microwave-assisted extraction (MAE) using for extracting PAHs, PCBs and
pesticides from soils
Analyte(s) Sample types / Operating conditions Recovery (%) Ref.
matrices
PAHs, OCPs, etc. Samples: standard Closed-vessel system; Sample 75 (average) 22
reference sediments size: 5 g. Extractants: ace-
(HS-3, HS-4, HS-5, tone-hexane (1:1, v/v), 30 ml;
and SRM 1941) and A temperature of 115°C and
soils (SRS 103-100, an extraction time of 5 min
and ERA lot no. were selected.
321)

187 Compounds Sample 1: fresh Closed-vessel system; For the compounds 23

(PAHs, pesticides, topsoils; Extractants: acetone-hexane from sample 1: >120
etc.) and 4 Sample 2: aged soils; (1:1, v/v), 30 ml; (1); 80-120 (79); <80
Aroclors (or PCBs) Sample 3: reference Temperature: 115°C; (14). For OCPs from
soils. Extraction time: 10 min. sample 2: 80-120 (38
out of 45). For OPPs
from sample 2: 80-
120 (34 out of 47). For
the compounds from
sample 3: at least as
good as Soxhlet
results (14 out of 15).
PAHs Sample: contami- Closed-vessel system; A sol- (Variable with the 24
nated soil. vent of 40-ml acetone, a tem- change of operating
perature of 120'C and an ex- conditions)
traction time of 20 min were
selected. A central composite
design was used to elucidate
the optimum operating condi-
tions.

PAHs, PCBs, pes- Samples: marine Domestic oven; Sample size: 97-102 25
ticides, etc. sediment. 2-10 g. Solvents: 10-ml toluene
+ 1-ml H2O; Extraction at 660
w for 6 min.
PAHs Sample: sewage Closed-vessel system; 97.7-117.3 26
sludge. Extractants: acetone-hexane
(1:3), 30 ml;
Temperature: 130°C.
PCBs Sample: soil, sewage Idem, but temperature 80°C. 123.4-149 26
sludge.

PCBs Sample: soil, Closed-vessel system; Sample >83 27

sediment (both size: 5-15 g. Extractant:
real and spiked). methanolic 1M KOH, 30 ml;
Pressure: 0.5 MPa;
Extraction time: 6 min.

(continued)

571
TABLE 17.3 continuation

Analyte(s) Sample types / Operating conditions Recovery (%) Ref.

matrices
Pesticides Sample: spiked Domestic oven; Sample size: 100 28
soil. 0.5-1.0 g.
Solvent: 2-3 ml methanol;
Extraction: irradiation for 30 s
followed by an interval of a
few minutes to cool the vial to
room temperature, repeated
up to 7 times.
OCPs Sample: sediment Domestic oven; 74-995.3 29
(spiked and then Sample size: 1 g.
aged for >1 month). Solvents: 2 ml
Before extraction, iso-octane-acetonitrile(1:1);
sample was Extraction: irradiation for 30 s
saturated with followed by a cooling with
water. ice-bath for 2-5 min, 5 times.
Imidazolinone Sample: a variety of Closed-vessel system; 92±13 30
herbicides agricultural soils. Extractants: 0.1 M
NH 4OAc/NH 4 0H at pH 10;
Temperature: 125°C;
Extraction time: 3 min.
Atrazine and its Sample: soil. Domestic oven; For DIA and DEA: 31
degradation prod- Extractans: organic-free 85-115;
ucts water and 0.35 N HC1; For atrazine and
Extraction with organic-free TBA: 50-65
water at 95-98°C, twice;
followed by extraction with
0.35 N HCI, 3 times.
Atrazine Sample: soils Closed-vessel system; 91.5 32
(spiked and real Sample size: 1 g.
sandy loam). Extractant: water 25 ml;
Pressure: 0.5 MPa;
Extraction time: 4 min.

Atrazine, simazine Sample: spiked Closed-vessel system; Sample For atrazine: 33

and prometryne soils. size: 1-4 g. 87.7-99.8
Extractants: different solvents For simazine:
[water, methanol, DCM, 88.5-100.4
acetone-hexane (1:1, v/v)]; For prometryne:
Pressure: 0.5 MPa; 38.5-97.6
Extraction time: 4 min.

OCPs: organochlororine pesticides; OPPs: organophosphorus pesticides; DCM: dichlorometh-

ane; DIA: deisopropylatrazine; DEA: de-ethylatrazine; TBA: terbuthylazine.

Recommendation for MAE: method development

- Use a high dielectric loss solvent such as acetone.
- Use a sample which contains some moisture (water has a very high dielectric
loss coefficient) as this will heat well.

572
 MAE system: atmospheric or
pressurised. Check that system is
operational and connected to
electric supply.
Sample e.g. soil, accurately weighed,
and placed in extraction vessel along
with solvent (hexane:acetone, 1:1, v/v,
acetone or DCM only). 30-45 ml of
solvent per 2-5 g of sample.

I-
MAE conditions: pressurised system.
Pressure, <200 psi; Temperature,
110-145 C; extraction time, 5-20
mins; microwave power, 100% at 900
W. Cooling time of extraction vessels,
20-30 mins, if unaided.
MAE conditions: atmospheric system.
Temperature, bp of solvent; extraction
time, 5-20 mins; microwave power,
100% at 300 W. Cooling time of
extraction vessel, negligible.

Solvent containing analyses filtered.

Concentration of analyte by solvent

evaporation (e.g. rotary evaporator,
solvent blow-down) or SPE.

Analysis

Fig. 17.9. Flow diagram for microwave assisted extraction.

- When extracting non-polar substances such as PAHs, use a 1:1 v/v mixture of
acetone and hexane to ensure high temperatures and good solubility of
analytes in solvent.
A flow diagram detailing a procedure for the extraction of analytes from soils is
shown in Fig. 17.9.

17.4 PRESSURISED FLUID EXTRACTION

Pressurised fluid extraction (PFE) has become the generic name for the extrac-
tion technique originally launched by Dionex Inc., in Spring 1995, called acceler-
ated solvent extraction (ASE ). The technique relies on the use of temperature
and pressure to extract organic compounds from different environmental,
industrial and food matrices.

17.4.1 Theoretical considerations

Liquid solvents at elevated temperatures and pressures should provide

enhanced extraction capabilities compared to their use at or near room tempera-
ture and atmospheric pressure for two main reasons [34]: (1) solubility and mass
transfer effects, and (2) disruption of surface equilibria.

573
Solubility and mass transfereffects
Three factors are considered important:
- Higher temperature increases the capacity of solvents to solubilise analytes.
- Faster diffusion rates occur as a result of increased temperature.
- Improved mass transfer and hence increased extraction rates occur when
fresh solvent is introduced, i.e., the concentration gradient is greater
between the fresh solvent and the surface of the sample matrix.
Disruptionof surface equilibria
As both temperature and pressure are important both are discussed separately.
Temperature effects
- Increased temperatures can disrupt the strong solute-matrix interactions
caused by van der Waal's forces, hydrogen bonding, and dipole attractions of
the solute molecules and active sites on the matrix.
- A decrease in viscosity and surface tension of solvents occurs at higher
temperature thus allowing improved penetration of the matrix, and hence
improved extraction.
Pressure effects
- The utilisation of elevated pressures allows solvents to remain liquified
above their boiling points.
- Extraction from within the matrix is possible, as the pressure allows the
solvent to penetrate the sample matrix.

17.4.2 Instrumentation

A commercial PFE system (ASE 200) consists of a fully automated sequential

extractor. It consists of a solvent supply system, extraction cell, oven, collection
system and purge system all of which are under computer control. A schematic
diagram of a PFE system is shown in Fig. 17.10. Employing a carousel, the

action

Fig. 17.10. Schematic diagram of apparatus for PFE.

574
extractor can operate with up to 24 sample-containing extraction vessels and up
to 26 collection vials plus an additional four vial positions for rinse/waste
collection. All the sample vessels have the same internal diameter of 19 mm, but
six volume sizes are available: 0.5, 1, 5, 11, 22 and 33 ml. With two removable end
caps, the sample vessel is easy to clean and fill with sample. The step-by-step
sample-filling procedure is as follows: screw one of the sample vessel's end caps
on to finger-tightness; then, introduce a filter paper into the vessel followed by
the sample itself; lastly, screw the other end cap on to finger-tightness and then
place the vessel in the carousel. Before performing extraction, an auto-seal
actuator places the selected extraction vessel into the oven.
For doing extraction, the sample cell vessel is positioned vertically in the
oven, and filled with the selected solvent (or solvent mixture) by the solvent sup-
ply system. Then, the cell is heated to a pre-set temperature (up to 200°C) and
pressure (up to 20 MPa) and held for a few minutes (typically 5 min). After com-
pletion, the static valves are released. Then, a few ml of clean solvent passes
through the extraction cell to exclude the existing solvent(s) and extracted
analytes. Finally, a gas-purging process follows in which N2 gas evacuates both
the sample vessel and the stainless steel transfer line. All extracted analytes pass
through the stainless steel tubing that punctures a septum (solvent resistant)
located on the top of the collection vial (40 or 60 ml capacity). The vessel will be
automatically returned to the carousel after extraction. Moreover, the inherent
safety features adopted by the PFE system include an IR sensor to monitor the
arrival and level of solvent in the collection vial, as well as an automatic shut-off
procedure that initiates in the case of system failure.

17.4.3 Methods of analysis

Several reviews on PFE have been published with an environmental bias. It is

not the intention of this chapter to review the application literature that exists
for PFE. Accordingly, a short table of applications has been provided (Table
17.4) to highlight suitable analytes and conditions. A suggestion for a procedure
for method development for PFE is given below based on the scientific literature,
as discussed above, and our own experience in the use of a commercial PFE
instrument since 1996 [41]:

Recommendations for PFE: equipment issues

- An automated system allows unattended operation.

Recommendations for PFE:sample issues

- Smaller particle size is important for increasing recovery.
- Use certified reference materials for method development, if possible. Alter-
natively extract from native contaminated samples.
- Pack the extraction cell appropriately i.e. use drying agents and copper (for
soils with high sulphur content).

575
TABLE 17.4
A summary of pressurised fluid extraction (PFE) using for extracting PAHs, PCBs and pesticides
from soils
Analyte(s) Sample types/ Operating conditions Recovery Ref.
matrices (%)
PAHs Sample: native Sample size: 7-9 g. Solvents: acetone- DCM quantitative 35
contaminated (1:1, v/v); Temperature: 100 C; Pressure:
soil. 2000 psi; Time: 5-min heat-up plus 5-min
static extraction.
PAHs Sample: real soil. Sample size: 20 g. Solvents: hexane- quantitative 36
acetone-toluene (10:5:1); Temperature:
100°C; Pressure: 13.8 MPa; Time: 5-min
heat-up plus 10-min static extraction.
PAHs, Sample 1: urban Sample size: 1-5 g. quantitative 37
OCPs, dust (SRM 1649a) Solvents: DCM; Temperature: 100°C;
PCB Sample 2: sediment Pressure: 2000 psi; Time: 5-min heat-up
congeners (SRM 1944). Sample plus 5-min static extraction.
3: marine sediment
(SRM 1941a)
PAHs Sample: contam- Solvents: acetone-hexane (1:1, v/v) quantitative 38
inated soils or acetone-DCM (1:1, v/v); Temperature:
100°C; Pressure: 10 MPa/14 MPa; Time:
5-min heat-up plus 2x5-min static extrac-
tion. Solvents: toluene; Temperature:
175/200C; Pressure: 14 MPa; Time: 8- min
heat-up plus 2x5-min static extraction.
OCPs Sample: contam-i idem quantitative 38
nated soils
PCDDs, Sample: fly ash Solvents: toluene; Temperature: quantitative 38
PCDFs 175/200°C; Pressure: 14 MPa; Time: 8-min
heat-up plus 2x10-min static extraction.
PCB Sample 1: marine Sample size: 0.5-1.5 g. Solvents: For sample 1: 39
congeners sediment (SRM 1944) acetone-DCM (1:1, v/v); Temperature: 83-112; For
Sample 2: sewage 100°C; Pressure: 2000 psi; Time: 5-min sample 2: 91-
sludge (BCR 392) heat-up plus 5-min static extraction. 109; For
Sample 3: harbor Others idem, but 2x5 min static sample 3: 93-
sediment (CRM 536) extraction. 123; For sam-
ple 3: 99-131.
Herbicides Sample: agricultural Sample size: 20 g. Solvents: methanol; 47-99 40
soil and spiked soil. Temperature: 125°C; Pressure: 10 MPa;
Time: combined heat-up and static
extraction, 10 min.

OCPs: organochlororine pesticides; PCDDs: polychlorinated dibenzo-p-dioxins; PCDFs: poly-

chlorinated dibenzofurans.

Recommendations for PFE:method development

- Optimise solvent choice. Changing the solvent and flushing the system
extends the time for method development. However, knowledge of the most
effective solvent system is required for achieving maximum recovery of

576
 analytes. A mathematical approach for the selection of the optimum solvent
has been developed [42,43] (for further details see later).
- Optimise static/flush cycles. The success of Soxhlet extraction is based on its
ability to recycle fresh, clean solvent through the sample at the rate of
approximately 4 cycles per hour. While this means that a Soxhlet extraction
may take up to 24 hours to achieve quantitative recovery of analytes, the
process of successive flushing with fresh solvent is an important concept. It is
recommended in PFE that the cycling of solvent through the sample is
mimicked. In order to achieve this PFE can perform up to three static-flush
cycles in any single extraction, allowing the PFE to mimic the action of
Soxhlet extraction. It has been reported [44] that the number of static flush
cycles required to quantitatively extract chlorinated pesticides from soil is
two. It was concluded [44] that two successive 5 min cycles was more
effective at extraction than longer (10 and 15 min) static extraction times.
- Optimization of PFE conditions: the main PFE operating variables are:
temperature, pressure, time. In order to evaluate these three variables it is
necessary to select operational (safe working) limits prior to commencement.
Typically, the following limits can be used: pressure, 1000 and 2400 psi;
temperature, 40 and 200°C; and time, 2 and 16 min.
- The preferred approach in our own laboratory is to use a multivariate
approach using chemometrics [45].
A flow diagram detailing a procedure for the extraction of analytes from soils is
shown in Fig. 17.11.

17.5 MISCELLANEOUS

17.5.1 Solid phase microextraction of soil slurries

One approach is to use solid phase microextraction (SPME) for the extraction of
analytes from soil slurries. This is achieved by stirring a known quantity of soil
with a solvent (e.g., water) and then exposing the SPME fibre directly to the
resultant slurry. An initial attempt to demonstrate the applicability of the
approach was done by Cisper et al. [46]. The group were able to determine
pyrene and anthracene at the 10 ppm level by exposure of an SPME fibre to a
soil/methanol slurry spiked sample for 3 min. Analysis was via laser desorption
ion trap mass spectrometry. It was suggested that the addition of a suitable
internal standard to the slurry sample would allow quantitation. A polar
poly(acrylate) fibre was utilised by Boyd-Boland and Pawliszyn [47 ] for the
qualitative extraction of nitrogen-containing herbicides from an aqueous slurry.
The soil sample (garden lawn) had been previously treated with benfluralin; a
GC-MS chromatogram indicated the presence of the herbicide. The same group
[48] were able to obtain quantitative data for the assay of metachlor, a pesticide,
on soil. The soil sample was prepared by stirring 0.5 g in 4 ml of water and then
exposing the SPME fibre directly to the resultant slurry for 50 min. It was found

577
 PFE system operational and Sample e.g. soil, accurately weighed
connected to electric and gas and mixed with drying agent and
supply. copper (for soils with high sulphur
content).

Sample +drying agent (+copper)

placed in extraction cell.

PFE conditions applied.

Pressure, 2000 psi; Temperature, 100 C; solvent, DCM-acetone, 1;1, v/v.
Operating conditions achieved in approx 5 mins, then static extraction for 5-
10 mins followed by N2 purging (1-2 mins).

and ak1 Collection of analyte-containg solvent.

Analyte concentration by solvent s

evaporation (e.g. rotary evaporator,
solvent blow-down or Kuderna-
Danish sample concentrator) or SPE.

Analysis

Fig. 17.11. Flow diagram for pressurised fluid extraction.

that the level of metachlor in soil was 1.84 mg kg- and 1.85 mg kg by SPME and
Soxhlet extraction, respectively. This approach demonstrated the applicability
of SPME for the analysis of analytes from solid matrices. However, some concern
was expressed by the authors in terms of the limited applicability of the
approach as metachlor is a relatively water-soluble compound (solubility = 530
mg 1) and this is not likely to be the case for other compounds.

17.5.2 Combined hot water extraction-SPME

An alternative to the slurry method is to extract the solid sample with hot water
and then isolate the analytes from the water using SPME prior to chromato-
graphic separation and detection. This approach has been applied to a range of
semivolatile compounds of environmental interest including polycyclic aromatic
hydrocarbons (PAHs) [49]. Sub-critical water (250°C) is used to extract soil sam-
ples for up to 60 min. Then, the sample cell is cooled (by placing under tap water)
and a known volume (1.8 ml) of the water removed and the solubilised organic
analytes are isolated by SPME prior to GC-MS analysis. A 100 Lm film thickness
poly(dimethylsiloxane)-coated fibre was used. For the determination of PAHs,
internal standards of the perdeuterated forms of the PAHs under investigation
were added directly to the water in the extraction cell prior to the heating step.

578
Quantitation was based on a comparison of the GC/MS peak areas found for the
sample PAHs versus the peak areas (and spike quantities) of the same deuter-
ated PAH internal standards (e.g. phenanthrene concentrations were based on
the response of the phenanthrene-d,, spikes). The effect of recoveries, relative to
24 h Soxhlet extractions, for 11 PAHs from railroad bed soil were assessed in
terms of water extraction temperature (200 and 250°C), extraction time (15 and
60 min) and operator. The highest recoveries were obtained at 250°C where
recoveries for the PAHs ranged from 45 to 197% (15 min extraction) and from 61
to 278% (60 min extraction). Similar recoveries, using a 250°C extraction tem-
perature, were obtained from an urban dust sample (SRM 1649) for 8 PAHs (or
grouped isomers): 33-149% for a 15 min extraction time and 58-141% for a 60
min extraction time. In addition to PAHs a range of volatile and polar organics
were extracted from an industrial soil using hot water (at 250°C) and a 15 min
SPME. For the determination of the aromatic amines and other contaminants
an internal standard, 2,4-diethylamine, which was added directly to the extrac-
tion cell after the heating step but prior to removal of the water. Quantitation
was based on the use of external standard solutions of the target organics pre-
pared in water. The results were compared with an 18 h sonication method and a
24 h Soxhlet extraction (both using a 1:1 mixture of dichloromethane:acetone).
The authors concluded that the three extraction procedures produced reason-
ably similar concentrations for the less volatile organics, including aniline and
the chlorinated anilines. However, a significantly higher concentration (at least
>33 times) was obtained for N-methylaniline by SPME. It was postulated that
analyte degradation had lead to this spuriously high concentration being
recorded for N-methylanaline. The authors stressed the importance of being vig-
ilant for possible analyte degradation at high temperature.
An extension to this work was provided by Daimon and Pawliszyn [50] who
compared two different approaches for the combined high temperature water
extraction: SPME of PAHs in solid matrices. A 30 Am poly(dimethylsiloxane)
coated fibre was used for SPME. The two approaches evaluated were: dynamic
and static extraction. In the dynamic extraction mode, analytes leached from the
solid matrix with hot water are collected in a vial and simultaneously isolated
from the water using SPME. In the static extraction mode a high pressure cell is
used in to which is inserted the SPME fibre. In this approach analytes released
from the solid matrix are partitioned into the fibre coating as the cell is cooling.
After SPME sorption the fibre is removed from either the collection water or the
high pressure cell and inserted in to the injector of the gas chromatograph for
separation and detection without the need for any further clean-up or pre-
concentration. Both approaches were evaluated for the analysis of PAHs from
urban air particulate (NIST SRM 1649). In the dynamic approach, extraction
was done for 15 min at 250°C and 50 atm while for the static extraction the
sample was heated to either 270 or 300°C for 120 min in the presence of water
and SPME was done for 120 min while the cell was cooling. Recoveries for the
three PAHs investigated were 134.0, 87.5 and 72.0% for fluoranthene, pyrene

579
and benzo[a]pyrene, respectively, by the dynamic extraction approach. Recov-
eries by the static approach ranged from 137, 149 and 128% for fluoranthene at
270°C (100 mg sample), 270°C (50 mg sample) and 300°C (50 mg sample),
respectively; 113, 121 and 105% for pyrene at 270°C (100 mg sample), 270°C (50
mg sample) and 300°C (50 mg sample), respectively; and 49, 72 and 70% for
benzo[a]pyrene at 270°C (100 mg sample), 270°C (50 mg sample) and 300°C (50
mg sample), respectively. It was suggested that the advantage of the dynamic
approach was the simple external calibration (samples and spikes, 40 ml volume,
were isolated by the SPME fibre for 70 min with rapid stirring) and fibre
protection, since the fibre coating is exposed to relatively pure water under low
temperature conditions. On the other hand, static high temperature water
extraction with simultaneous SPME uses only simple apparatus and procedure,
and extends the approach for analytes tightly bound to soils. Quantitation for
2
this approach was achieved by spiking isotopically labelled standards i.e. 10 [ H]
2 2
fluoranthene, 10 [ H] pyrene and 12 [ HI benzo[a]pyrene, as internal standards
in to the sample prior to extraction.

17.5.3 Headspace SPME of solids

Instead of using SPME to extract from the aqueous slurry an alternative

strategy has been utilised. This uses SPME to extract volatile or semi-volatile
analytes from the headspace above a sample. Preliminary work by Zhang and
Pawliszyn [51] investigated the use of headspace SPME for the analysis of
volatile and semi-volatile compounds from sand, clay and sludge. Extraction
time profiles (signal versus extraction time curves) were produced for benzene,
toluene, ethylbenzene and xylene isomers (BTEX) extracted from sand, using a
poly(dimethylsiloxane) fibre, at 50°C. It was established that equilibrium was
rapidly reached (< 1 min). The effects of increasing the temperature and modify-
ing the matrix were investigated by extracting the BTEX compounds from clay.
It was found that increasing the temperature (room temperature to 50°C) and
the addition of water (10-30%) increased the sensitivity of the measurement.
This approach was then applied to a sewage sample. Using a sampling time of 2
min at room temperature and GC-MS detection it was possible to identify six
native volatile organic compounds (toluene, tetrachloroethene, ethylbenzene,
m-andp-xylene, o-xylene and naphthalene) in the sample. The paper also reports
initial attempts to extract PAHs from the headspace above sand.
A simple one-dimensional kinetic model has been reported to study the
diffusion process involved in headspace SPME [52]. The results of the theoreti-
cal model were consistent with the data obtained for the headspace SPME, using
a 56 Am coated poly(dimethylsiloxane) fibre, of BTEX compounds from aqueous
samples. The sampling time for BTEX compounds was reduced to 1 min as
compared to 5 min for direct analysis in aqueous samples. It was suggested that
at ambient temperature the headspace SPME approach can be effective at
isolating compounds with Henry's constants >90 atm cm3/mol i.e. three-ring

580
PAHs compounds or more volatile compounds. This approach was then applied,
in addition to aqueous samples, to a sewage sludge and a soil sample known to be
contaminated with PAHs. For the sewage sludge sample, 0.2 Al of 10 gg ml- l
standard solution of each deuterated PAH (naphthalene-d8 and phenanth-
rene-d,,), in acetone, were used as internal standards. Using a sampling time of
30 min and maximum stirring of the sample it was possible to quantify the levels
of the two PAHs (2.0 and 0.3 ng ml- l for naphthalene and anthracene, respect-
ively) in the sewage sample using single-ion mode GC-MS. Using the same
approach the levels of four PAHs were determined in a soil sample. In this
situation, phenanthrene-d,, was used as an internal standard for anthracene
and phenanthrene while chrysene-d,2 was used as the internal standard for
benz[a]anthracene and chrysene. The deuterated PAHs (0.1 ml of a 10 g ml- l
standard solution) were spiked into a 1.0 g sample of the soil. The concentrations
determined were as follows: anthracene (17.0 pg gl), phenanthrene (4.0,g gl),
benz[a]anthracene (0.23 ,g gl) and chrysene (0.23 g g).
James and Stack [53] exposed a 100 Am poly(dimethylsiloxane) fibre to the
headspace above a soil sample (1.0 g) heated up to a temperature of 60°C (22, 30,
40, 50 and 60°C). Both spiked soils (control site) and soil samples from a landfill
site were used in the study. After 180 min the fibre was inserted into the injector
of a gas chromatograph and analysed for a selected range of volatile organic
compounds. Calibrations, using soil samples spiked with selected solvents
(1,1,1-trichloroethane, benzene, tetrachloromethane, trichloroethene, methyl-
benzene, tetrachloroethene, ethylbenzene, p-xylene and styrene) in the range
0.5-75 gg g' were linear (r2 > 0.9889). This approach (50°C for 30 min) was
applied to eight sites within a municipal landfill, at three depths, and resulted in
the detection of xylene isomers in 18 out of the 24 samples. The maximum
concentration of xylene determined was 2.8 pg g-l. In addition, a number of other
non-target organic contaminants were identified e.g. 4-methylphenol (found in
four samples) and alkylated benzene compounds and terpenes. 4-Methylphenol
is used as a wood preservative and so it is not uncommon to find it in municipal
landfill sites. The authors concluded that the application of headspace SPME
provided a rapid protocol for the quantification of residual solvents in landfill
soil samples.

17.6 COMPARISON OF EXTRACTION TECHNIQUES

Any comparison of extraction techniques needs to consider a range of factors in

addition to their effectiveness [54]. Table 17.5 identifies and then compares
some selected analytical figures of merit for the main extraction techniques con-
sidered in this chapter (except SPME). By simply viewing this table it is obvious
that the choice of the optimum extraction technique is not easy. It is obvious,
however, that any laboratory working to an accreditation system will only use
methods that are approved. However, all the techniques considered have 'ap-
proved methods' for extraction of a range of analytes.

581
TABLE 17.5

Analytical figures of merit of extraction techniques

Soxhlet SFE MAE PFE

Sample mass (g) 10 1-10 2-5 Up to 30 g
Extraction time 6, 12 or 24 h 30 min-1 h 20 min (plus 30 min 12 min
cooling and pressure
reduction)
Solvent type Acetone:hexane CO 2 (plus organic Typically, acetone: Acetone:hexane (1:
(1:1, v/v); ace- modifier). Tetra- hexane (1:1, v/v). 1, v/v) or acetone:
tone: DCM (1:1, chloroethene used The solvent(s) is/are DCM (1:1, v/v) for
v/v); DCM only; as the collection required to be able OCPs, semivolatile
or, toluene:meth- solvent for TPHs to absorb microwave organics, PCBs or
anol (10:1, v/v) for determination energy. OPPs; acetone:
by FTIR, otherwise DCM:phosphoric
DCM. acid (250: 125:15,
v/v) for chlorinated
herbicides
Solvent consump- 150-300 10-20 25-45 25
tion (ml)
Extraction method heat Heat + pressure Heat + pressure Heat + pressure
Sequential or Sequent ial (but Sequential Simultaneous (up to Sequential
simultaneous multiple assem- 14 vessels can be ex-
blies can operate tracted simulta-
simultaneously) neously)
Method develop- Low High High High
ment time
Operator skill Low High Moderate Moderate
Equipment cost Low High Moderate High
Level of automation Minimal Minimal to high Minimal Fully automated
up to 24 samples
can be extracted.
EPA method 3540 3560 for TPHs, 3546 3545
3561 for PAHs and
3562 for PCBs and
OCPs

TPHs: total petroleum hydrocarbons; PAHs: polycyclic aromatic hydrocarbons; OCPs: organo-
chlorine pesticides

While Soxhlet extraction is applicable to a wide range of analytes from solid

matrices the technique itself is slow (solvent is passed through the sample at the
rate of 3/4 cycles per hour for up to 24 h). However, most users of this technique
would use multiple Soxhlet assemblies. The cost of the Soxhlet assemblies and
related heating devices (iso-mantles) is relatively low cost in terms of capital
expenditure, but high in terms of solvent consumption and cost and disposal of
solvent. SFE is used to extract samples sequentially using an environmentally
friendly solvent (CO2). While some organic solvent is required for the removal of
polar analytes its popularity as a method of extraction has waned in recent years
(not assisted by the lack of commercial instruments available in the market

582
place). MAE is capable of extracting multiple samples simultaneously in a short
time; however, additional cleanup is required to remove the sample matrix from
the analyte-containing solvent, after cooling of the sample vessels. PFE allows
multiple samples to be extracted sequentially in an automated system. The
major disadvantage of PFE is the highest capital cost (of any of the techniques
described) and the need for sample treatment for high sulphur-containing soils,
or risk the blockage of tubing between the hot sample cell and the cooler collec-
tion vial.
Five main factors are considered important when making a decision as to
which technique to use for extracting environmental samples: capital cost,
operating costs, requirements for method development, environmental impact
(risk) and level of automation. It is possible that after taking these factors into
account that the 'easy option' is to retain the traditional, tried and tested
techniques currently available in your laboratory (Soxhlet extraction). However,
it is likely that a cost-benefit analysis of the instrumental techniques might well
enable the user to improve sample throughput, reduce solvent consumption (and
hence, disposal costs) and re-deploy staff in other activities.
So, while the authors have a tendency to favour the newer instrumental
extraction techniques because of their speed of extraction and smaller consump-
tion of organic solvent, they also know that current approaches are by no means
the end commodity. The current research focus towards miniaturisation should
lead to next generation systems that incorporate complete automated extrac-
tion-analysis systems.

17.6.1 Solvent selection: prediction of the optimum solvent

An alternative approach to solvent optimization is to predict the solvent

required that would give quantitative extraction of the analyte. Recently, a
model based on the Hildebrand solubility parameter has been developed [42,43].
The solubility parameter, 6, is defined as the square root of the cohesive energy
density or
2
= (AEV)' (17.1)
where AE v is the energy of vaporisation at a given temperature, and V is the
molar volume of the molecule [55].
The solubility parameter is used as a measure of the solubility of compounds
in various solvents.
However, calculations of this sort require knowledge of the heat of vaporisa-
tion at various temperatures as well as the molar volume of the substance, and
these values are not widely available for a wide range of molecules of environ-
mental interest. Hansen [56] has taken this work further and assumes the total
cohesive energy is a linear addition of three components: h, hydrogen bonding
ability contribution; d, dispersion coefficient contribution; and 6,, polarity
contribution. They are linked by the following equation.

583
TABLE 17.6
Calculation of individual group contributions for methanol
Group Group contribution to Group contribution to Group contribution to
dispersion
3
(Fd) polarity (F ) hydrogen bonding (Uh)
J1 2 cm /2 mo m mol' J mol-'

CH 3 420 0 0
OH 210 500 20000
Total 630 500 20000

2 6 2 2 6 2
5t = h + 5p + d (17.2)

Calculation of each component is achieved via group contribution data in the

literature. Thermodynamic data to develop equations for the calculation of the
individual parameters of the total solubility parameter are available [57,58]. The
method used was addition of group contributions. Tables of data are set up
containing each group's contribution to polarity, dispersion and hydrogen bond-
ing (Fp, Fd, and Uh, respectively). Using Eqs. (17.3) to (17.6) 6,, 6 h, and 6,d can be
calculated.

d (Z'Fd)/V (17.3)
* (17.4)
6p = (zZZFp)/V
2 1 /2
6p = ( Fp ) /V (17.5)
2
6h = ((zZ Uh)/V)' (17.6)

":For molecules with more than one polar group present, Eq. (17.5) must be used
instead of Eq. (17.4), to take into account the interactions between the polar
groups [58].
An example calculation of the individual components of the solubility
parameter for a solvent (methanol) is shown in Table 17.6. The values for the
group contributions were taken from Ref. [58].
Fractional parameters of the Hildebrand solubility parameter can be calcu-
lated using Eqs. (17.3)-(17.6) and plotted on a triangular graph in order to give a
visual representation of the extent of contribution from the three components
(polarity, dispersion and hydrogen bonding). A visual representation between
the three components is via a triangular graph using fractional parameters (Fig.
17.12). The triangle can be used to predict the best solvent for dissolving a com-
pound. The optimum solvent should be in the same position as the target
analyte. This thinking can be applied to extraction techniques. To quantitatively
extract an analyte from a matrix, the solvent should have similar properties as
the compound, i.e. non-polar solvents are better at extracting non-polar
analytes. Using this approach, an informed selection of the most suitable solvent
for a particular analyte can then be made.

584
 100

olvent value
onalyte values
Methanol
Acetone
fichloromethane
,cetonitrile
so-hexane
Foluene
,cetonitrile:dichloromethane (:1, v/v)

0o V / t 100

Fig. 17.12. Prediction of the optimum solvent for extraction: modelling the process.

REFERENCES

1 V. Camel, A. Tambute and M. Claude, J. Chromatogr., 642 (1993) 263.

2 V. Janda, K.D. Bartle and A.A. Clifford, J. Chromatogr., 642 (1993) 283.
3 I.J. Barnabas, J.R. Dean and S.P. Owen, Analyst, 119 (1994) 2381.
4 T. Greibrokk, J. Chromatogr., 703 (1995) 523.
5 S. Bowadt and S.B. Hawthorne, J. Chromatogr., 703 (1995) 549.
6 I.A. Stuart, J. MacLachlan and A. McNaughton, Analyst, 121 (1996) 11R.
7 J.R. Dean, Analyst, 121 (1996) 85R.
8 S.B. Hawthorne and D.J. Miller, Anal. Chem., 59 (1987) 1705.
9 S.B. Hawthorne and D.J. Miller, Anal. Chem., 66 (1994) 4005.
10 I.J. Barnabas, J.R. Dean, W.R. Tomlinson and S.P. Owen, Anal. Chem., 67 (1995)
2064.
11 S. B0wadt, B. Johansson, Anal. Chem., 67 (1995) 2424.
12 E.G. van der Velde, M. Dietvorst, C.P. Swart, M.R. Ramlal, P.R. Kootstra, J.
Chromatogr.A, 683 (1994) 167.
13 J.R. Dean, I.J. Barnabas and S.P. Owen, Analyst, 121 (1996) 465.
14 S.P. Frost, J.R.Dean, K.P. Evans, K. Harradine, C. Cary and M.H.I. Comber, Analyst,
122 (1997) 895.
15 J. Atienza, J.J. Jim6nez, A. Herguedas and J.L. Bernal, J. Chromatogr. A, 721 (1996)
113.
16 S.B. Hawthorne, D.J. Miller, M.D. Burford, J.J. Langenfeld, S. Eckert-Tilotta and
P.K. Louie, J. Chromatogr., 642 (1993) 301.
17 R. Gedye, F. Smith, K. Westaway, H. Ali, L. Baldisera and L. Laberge, Tetrahedron
Lett., 27, 279 (1986).
18 L. Jassie, R. Revesz, T. Kierstead, E. Hasty and S. Matz, in: H.M. Kingston and S.J.
Haswell (Eds.) Microwave-Enhanced Chemistry: Fundamentals, Sample Prepara-
tion and Applications. ACS, 1997, Chapter 12, p. 569.
19 J.R.J. Parr, J.M.R. Belanger and S.S. Stafford, Trends Anal. Chem., 13 (4), 176
(1994).
20 H.M. Kingston, P.J. Walter, S. Chalk, E. Lorentzen and D. Link, in: H.M. Kingston
and S.J. Haswell (Eds.), Microwave-Enhanced Chemistry: Fundamentals, Sample
Preparationand Applications. ACS, 1997, Chapter 3, p. 223.

585
21 S.J. Chalk, H.M. Kingston, P.J. Walter, K. McQuillin and J. Brown, in: H.M.
Kingston and S.J. Haswell (Eds.), Microwave-Enhanced Chemistry: Fundamentals,
Sample Preparationand Applications. ACS, 1997, Chapter 15, p. 667.
22 V. Lopez-Avila, R. Young and W.F. Beckert, Anal. Chem., 66 (1994) 1097.
23 V. Lopez-Avila, R. Young, J. Benedico, P. Ho, R. Kim and W.F. Beckert, Anal. Chem.,
67 (1995) 2096.
24 I.J. Barnabas, J.R. Dean, I.A. Fowlis and S.P. Owen, Analyst, 120 (1995) 1897.
25 A. Pastor, E. Vazquez, R. Ciscar and M. de la Guardia, Anal. Chim. Acta, 344 (1997)
241.
26 B. Enders and G. Schwedt, GITFachz. Lab. (German), 40 (1996) 172.
27 G.-H. Xiong, X.-Q. He and Z.-X. Zhang, Anal. Chim. Acta, 413 (2000) 49.
28 K. Ganzler, A. Salgo and K. Valko, J. Chromatogr., 371 (1986) 299.
29 F.I. Onuska and K.A. Terry, Chromatogr., 36 (1993) 191.
30 S.T. Stout, A.R. dacunha and D.G. Allardice, Anal. Chem., 68 (1996) 653.
31 T.R. Steinheimer, J. Agric. Food Chem., 41 (1993) 588.
32 G.-H. Xiong, J.-M. Liang, S.-C. Zou and Z.-X. Zhang, Anal. Chim. Acta, 371 (1998) 97.
33 G.-H. Xiong, B.-Y. Tang, X.-Q. He, M.-Q. Zhao, Z.-P. Zhang and Z.-X. Zhang, Talanta,
48 (1999) 333.
34 B.E. Richter, B.A. Jones, J.L. Ezell and N.L. Porter, Anal. Chem., 68, 1033 (1996).
35 N. Saim, J.R. Dean, Md. P. Abdullah and Z. Zakaria, J. Chromatogr., 791 (1997) 361.
36 J.D. Berset, M. Ejem, R. Holzer and P. Lischer, Anal. Chim. Acta, 383 (1999) 263.
37 M.M. Schantz, J.J. Nichols and S.A. Wise, Anal. Chem., 69 (1997) 4210.
38 P. Popp, P. Keil, M. M6der, A. Paschke and U. Thuss, J. Chromatogr. A, 774 (1997)
203.
39 E. Bjbrklund, S. Bowadt, T. Nilsson and L. Mathiasson, J. Chromatogr.A, 836 (1999)
285.
40 E. Conte, R. Milani, G. Morali and F. Abballe, J. Chromatogr.A, 765 (1997) 121.
41 J.R. Dean, Anal. Commun., 33, 191 (1996).
42 L.J. Fitzpatrick and J.R. Dean, 21st InternationalSymposium on Capillary Chroma-
tography and Electrophoresis, Park City, Utah, USA, 20-24 June 1999.
43 L.J. Fitzpatrick and J.R. Dean, Anal. Chem., 74 (2002) 74.
44 P. Popp, P. Keil, M. Moder, A. Paschke and W. Thuss, J. Chromatogr., 774A, 203
(1997).
45 N. Saim, J.R. Dean, Md. P. Abdullah and Z. Zakaria, Anal. Chem., 70, 420 (1998).
46 M.E. Cisper, W.L. Earl, N.S. Nogar and P.H. Hemberger, Anal. Chem., 66 (1994)
1897-1901.
47 A.A. Bowland and J.Pawliszyn, J. Chromatogr., 704 (1995) 163-172.
48 A.A. Boyd-Bowland, S. Magdic and J.B. Pawliszyn, Analyst, 121 (1996) 929-938.
49 K.J. Hageman, L. Mazeas, C.B. Grabanski, D.J. Miller and S.B. Hawthorne, Anal.
Chem., 68 (1996) 3892-3898.
50 H. Daimon and J. Pawliszyn Anal. Comm., 33 (1996) 421-424.
51 Z. Zhang and J. Pawliszyn, J. High Res. Chromatogr., 16 (1993) 689-692.
52 Z. Zhang and J. Pawliszyn, Anal. Chem., 65 (1993) 1843-1852.
53 K.J. James and M.A. Stack, J. High Resolut. Chromatogr., 19 (1996) 515-519.
54 J.R. Dean, Extraction Methods for Environmental Analysis. John Wiley and Sons,
Chichester, 1998.
55 R.F. Fedors, Polymer Eng. Sci., 14, 147 (1974).
56 C.M. Hansen, J. PaintTechnol., 39, 105 (1967).
57 A.F.M. Barton, The Handbook of Solubility Parametersand other Cohesion Para-
meters. CRC Press, Florida, 1983.
58 D.W. van Krevelen and P.J. Hoftzyer, Propertiesof Polymers; Their Estimation and
Correlation with Chemical Structure. Elsevier, Amsterdam, 1976.

586
Chapter 18

Hot (subcritical) water extraction

Steven B. Hawthorne and Alena Kubdtovd

18.1 INTRODUCTION

The desire to replace organic solvents used for extracting organic compounds
from solids and semi-solids has resulted in a large volume of literature describ-
ing the use of supercritical carbon dioxide. One other fluid, which certainly fits
"green" solvent criteria, is water. However, water at ambient conditions is much
too polar to be generally useful for extracting non-polar and moderately polar
organics from solid and semi-solid matrices, ranging from soils and sludges to
biological tissues. The polarity of water can be reduced dramatically (and the sol-
ubility of organics increased) by heating and pressurizing water to its supercriti-
cal state. However, the very high pressure (> 220 bar) and temperature (> 374°C)
required complicates its use for analytical extractions. In addition, supercritical
water is quite reactive (stainless steel is generally not sufficiently resistant to
corrosion). Thus, the major uses of supercritical water for organic compounds
have focused on their destruction, rather than extraction and recovery of
organics for analytical or process purposes. Similarly, the characteristics and
uses of superheated steam are well known, but generally are not considered
broadly applicable to organic extractions.
Interestingly, the characteristics of one area of water's phase diagram has
largely been ignored by the scientific community. As shown by the shaded area
in Fig. 18.1, the subject of this chapter is the use of water as an extraction fluid,
where the water is heated anywhere up to its critical temperature, and where
enough pressure is applied that the water remains liquid rather than turning to
steam. This technique has variously been named "subcritical water", "hot
water", or "superheated water" extraction, but regardless of the terminology,
the approach is defined by hot water kept under sufficient pressure to maintain
the liquid state.
When we performed our initial work with "subcritical" water [1], we had
approximately 400 references related to supercritical water extraction in our
files, but could only locate a few citations which were even vaguely related to hot
water extractions. None of these exploited the properties of hot water for analyt-
ical extractions. The lack of scientific literature on the properties of water in this
ComprehensiveAnalytical Chemistry XXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 587
 I
I

200

150-
Ice

. 100-
Liquid
Melting point curve --

50-

10000

0 . I ,

-270 -170 -70 30 130 230 330 430

Temperature, °C

Fig. 18.1. Phase diagram of water. Note the shaded area representing hot subcriticali" water.
Generated using NIST/ASME Steam Formulation Database 10, version 2.2, A.H. Harvey, A.P.
Peskin, S.A. Klein [43].

region is particularly surprising since those who make cappuccino coffee, or who
use a home pressure cooker have long been exploiting these beneficial proper-
ties. While analytical chemists have largely ignored hot water, the venerable
Professor Franck had published one data set in 1983 on the solubility of
anthracene under "subcritical" water conditions which was very enlightening
[2]. Franck's data showed that the solubility of anthracene increased by a factor
of 1000 over its ambient solubility by simply heating the water to 150°C (at 60
bar). This surprising data set (although ignored for many years) clearly demon-
strates the potential uses of hot water for organic extractions.
In subsequent years since our first work with hot water extraction [1], the
amount of published work in this field has grown steadily. We hope that this
chapter will give the reader a realistic view of the potential of water as an
extraction fluid (both positive and negative aspects), as well as provide a survey
of work in the field. Since the majority of readers will be more familiar with
supercritical carbon dioxide, some comparisons of the two fluids will be made
throughout the text based on the qualitative comparisons shown in Table 18.1.

18.2 BACKGROUND

Three major characteristics of water change when it is heated and kept in the
liquid state, i.e., the polarity (as measured by the dielectric constant, ), the
surface tension, and viscosity all drop dramatically as shown in Figs. 18.2, 18.3,
and 18.4 [3]. In fact, when water is heated from ambient to 250°C, the changes in

588
TABLE 18.1
Properties of supercritical carbon dioxide versus hot "subcritical" water
CO2 HO
Solubility control 10-100x 50-100000Ox
Major parameter T&P T
"Easiest analyte" non-polar polar
"Hardest analyte" polar non-polar
Reactivity low low to moderate
Concentrating analyte extract usually simple simple to difficult
Selectivity for solute polarity fair good
Selectivity against matrix (e.g. soils) good poor
Polarity range (e) 1-2 10-80

these three parameters are nearly identical to changing the fluid to pure
methanol or acetonitrile (Figs. 18.2 to 18.4). While the drop in viscosity and
surface tension would be expected to enhance extraction rates based on mass
transfer considerations, the major advantage gained is the drop in water's
polarity from a very polar fluid at room temperature ( = 80) to a fairly low
polarity fluid at higher temperatures (e.g., E = ca. 20 at 300°C).
The practical effect of being able to control water's polarity with tempera-
ture is the ability to control the solubility of various organic compounds. The
surprisingly large increase in anthracene's solubility has subsequently been con-

%Methanol or Acetonitrile in Water at 20 °C

0

Fig. 18.2. Effect of temperature and concentra-

tion of organic solvent in water on the dielectric 50 100 150 200 250
0
constant [3]. Pure Water Temperature, C

589
 %Methanol or Acetonitrile in Water at 20 C %Methanol or Acetonitrile in Water at 20 °C
on An n A
... . ..0 20 40 60 80 100
u AU 4U OU OU IUU

E
0
C
0

.2 a.
C
U)
a
0
C
V)
0
2A
I
a,

50 100 150 200 250 50 100 150 200 250

Pure Water Temperature,?C Pure Water Temperature,?C

Left: Fig. 18.3. Effect of temperature and concentration of organic solvent in water on the
surface tension [3].
Right: Fig. 18.4. Effect of temperature and concentration of organic solvent in water on the
viscosity [3].

firmed by measuring the solubility of several solids (mostly PAHs and pesticides)
[4-7]. In fact, the solubility of such compounds tends to increase by an order of
magnitude for every 50°C increase in water's temperature, as shown for
benzo[a]pyrene and the pesticides atrazine and chlorothalonil in Fig. 18.5. Thus,
solubility increases of four to five orders of magnitude are typical for heating
water to 250°C and continue to increase at hotter temperatures. In fact, the solu-
bility of the PAH benzo[e]pyrene increases by an amazing factor of 25 million
fold when the water is heated to 350°C [7]. For comparison purposes, the solubil-
ity changes which occur for similar compounds over the useful temperature and
pressure range for supercritical carbon dioxide are typically in the one- to
two-orders of magnitude range, rather than the four- to six-orders of magnitude
increases found for hot water [8].
Although the solubility of liquid solutes such as alkyl benzenes and flavor
and fragrance compounds do not increase quite as dramatically as solid solutes,
the increase in solubility with temperature is still quite dramatic [9-11] with sol-
ubility changes of ca. 50 to 100-fold for nonpolars (like monoterpenes) occurring
in 200°C water. As might be expected, solubility increases are much more limited
if the solute already has significant solubility at room temperature [10]. For
example, the solubility of nerol only increases from ca. 2 x 104 mole fraction at

590
 80
Benzo[a]pyrene
62
60 _-- -
J

40

-5
co
20

6
0.004 0.04
23 100 150 200
0D
Temperature, C

2000
Atrazine 1780
1600
CI

E 1200
CH3 N N

2- 800 _
O H3C NH N NH \CH 3
500
400 210
70

50 75 100 125
Temperature, °C

N C I

ci C
CI

Temperature, °C

Fig. 18.5. Solubilities of different organic pollutants at different water temperatures. Adapted
from [4-6].

room temperature to 50 x 10 4 mole fraction at 200°C [10]. Similarly, the solubil-

ity of atrazine increases only 15-fold when water is heated from 50 to 125°C [6].
Finally, the use of pressure with hot water extraction is a minor variable (as
opposed to being a major variable with supercritical carbon dioxide, for example)

591
as long as the pressure is sufficient to maintain the liquid state. While increasing
pressure most often increases the solubility of organics in carbon dioxide,
increasing pressure actually increases water's dielectric constant, and therefore,
decreases organic solubilities. For example, the solubility of the PAH, pyrene, in
100°C water drops from 9 x 10 - 7 mole fraction at 40 bar to 1 x 10- 7 mole fraction at
400 bar [5].

18.3 APPARATUS AND TECHNIQUES

18.3.1 Dynamic (flowing) extractions and analyte collection

Much of the early work in hot water extraction was performed using equipment
analogous to that used for supercritical CO2 extraction, as shown in Fig. 18.6. In
essence, a pump is used to provide the water to the sample extraction cell via a
pre-heating coil. The pump can be run in a constant flow mode by utilizing a
small HPLC back-pressure regulator [5], or by using a flow control valve at the
outlet of the unit. Care must be taken to control the internal pressure so that the
phase of the water (liquid versus steam) in the extraction unit is controlled.
Analyte collection methods depend on the solubility of extracted compounds
in ambient water, and the final method used to analyze the extracted com-
pounds. For solutes fairly soluble in water, simply cooling and collecting the
water can suffice. For non-polar solutes, cooling the water prior to collection can
cause loss of the extracted compounds in the transfer line between the oven and
the collection device. (Remember, if the solubility of a compound increases sev-
eral orders of magnitude in the heated water, it decreases when the water is
cooled for collection.) For such compounds, we suggest introducing a small flow
of organic collection solvent in the heated zone as shown in Fig. 18.6. This allows
the collection solvent to mix with the hot water, and extract the organic solutes
from the water as cooling occurs. The result is a very nicely controlled flow of
cooled (and phase separated) water and organic solvent at the collection device.
Typical solvents good for non-polar organics (e.g., hexane, toluene, methylene
chloride) which are not miscible in water perform well. However, at higher tem-
peratures (ca. >250°C), chlorinated solvents can dehalogenate forming hydro-
chloric acid, and must not be used. Since nonpolar organics are generally
I~~~~~c~ollectio Miniature Back
Solvent >4 * Pressure Regulator
Pump Inlet Safety
Ives _ffRelief
Water - Valve

Collection
Vial
Preheating- Extraction
Coil Cell

Fig. 18.6. Hot water GC Oven

extraction apparatus.

592
analyzed by GC, their final collection in a solvent which is compatible with GC is
convenient.
An alternative to collecting analytes in the extract water, or in an organic
solvent is to pass the extract water through a solid-phase trap, such as C18
bonded silica [12-16]. This approach eliminates the use of organic solvents and
has been used to couple hot water extraction with HPLC. However, extreme care
must be taken to ensure that any non-polar organics are not deposited in the
transfer line as the water is cooled. Alternate approaches to solid phase trapping
have utilized membrane discs [17-19], and solid-phase microextraction (SPME)
[20-23].
Although the concept of off-line hot water extractors is similar to supercriti-
cal CO 2 extractors, the required characteristics of the various hardware compo-
nents are somewhat different, especially the pump and the extraction cell
heater. Overall, water is much easier to pump at a consistent flow rate than CO2,
since water is a non-compressible fluid in the pump. While many commercial
CO2 pumps are effective for pumping water, simpler and cheaper HPLC pumps
also suffice. Also, while the pressure required for supercritical CO2 extractions is
often 400 bar (or higher), the pressure required for subcritical water extractions
is only a few bar, as shown in Fig. 18.1. Thus, at the highest temperatures
typically used to extract non-polar organics with hot water (e.g., 300"C), the
pump need only be capable of working at ca. 100 bar (just above the pressure of
86 bar which is required to maintain the liquid state of water at 300°C).
Naturally, the cell heater must also be capable of 300°C, which is normally
accomplished by placing the sample cell (and a water pre-heating coil) inside a
GC oven.
Much lower temperatures and pressures will suffice if the operator only
desires to extract moderately polar compounds (e.g., most pesticides, oxygenated
flavor and fragrance compounds). For many such compounds, a temperature of
150°C or lower will suffice, which only requires a pump capable of delivering
water at ca. 5 bar. This keeps the water liquid, although for the sake of
simplicity, extractions are often performed at 50 bar. In fact, some commercial
SFE and PSE (pressurized liquid extraction) units can perform hot water
extractions at these conditions.
An approximate idea of the temperature required to extract a particular
class of organic compounds is shown in Fig. 18.7. Again, subcritical water is
essentially opposite to supercritical CO,, i.e., the most extreme conditions (high-
est temperature) are required for hot water extraction of non-polar solutes such
as PAHs and PCBs. In fact, the "easiest" compounds for extraction by supercriti-
cal carbon dioxide, alkanes (which are very non-polar) are virtually impossible to
extract with hot water (except by decompressing the water to steam, as
described later in the text). Similarly, the "easiest" organics to extract with hot
water, polar organics, are generally the most difficult to extract with supercriti-
cal CO 2 (and require the addition of an organic modifier to be extracted at all in
supercritical CO,).

593
 Hard Nonpolar 280 °C

PCBs
i J PAHs I
organochlorine pesticides
monoterpenes
triazine and nitro-pesticides
explosives (HMX,RDX,TNT)
flavor oxygenates
Fig. 18.7. Water temperature
ran r-irrl to extrrt m- phenols, amines
poundswith differentpolarities. Easy Polar 100 °C

Similar to SFE, the time required for a particular extraction can be deter-
mined by collecting and analyzing fractions. As shown in Fig. 18.8, the time
required to extract chloransulam-methyl from soil can be dramatically reduced
by increasing the extraction temperature to 150°C, and reasonably good recover-
ies are achieved in ca. 100 min [24].
Special mention should be made of the plugging problems associated with
hot water extraction since this problem was so important during the develop-
ment of SFE. Unfortunately, plugging problems also exist with hot water extrac-
tion. However, at least in our laboratory (which mostly extracts soils and waste
sludges, as well as aromatic herbs), plugging at the outlet restrictor (e.g., the
back pressure regulator shown in Fig. 18.6), is not a significant problem.
Plugging of the outlet line (where the water is cooled) can be major if very high
concentrations of non-polars (e.g., a soil highly contaminated with PAHs) are
extracted at high temperatures (e.g., 250°C), since the solutes deposit when the
water temperature is dropped. However, adding the organic collection solvent to
the heated zone (as shown in Fig. 18.6), has completely solved this problem in

Ann

91
81
7(
S 6'
8 5
' 4

3
2
1

0
Time (min)

Fig. 18.8. The effect of time and temperature on the extraction efficiency of cloransulam-methyl
from soil. Used by permission from Krieger et al. [24].

594
our lab. Plugging of the sample matrix itself remains a significant problem. For
example, many soils appear to swell by 20% or more during extraction. Thus, a
void must be left in the sample cell upon filling to allow for this expansion. Fre-
quently, the sample swells enough at the higher water temperatures that mixing
it with a dispersant (or reducing the sample size) is required.

18.3.2 "Static" (non-flowing) extractions and analyte collection

For SFE, "static" extractions are often performed to allow a modifier or chemical
reagent to be in contact with a sample for a certain time, after which a "dyna-
mic" extraction was performed in a conventional manner to collect the analyte.
While such experiments are certainly reasonable with hot water extractions, the
steam/water equilibrium can also be exploited to allow the construction of
extremely simple and inexpensive hot water extraction units [17]. The unit
consists simply of a threaded stainless steel pipe with threaded end caps sealed
with Teflon tape. At the time of this writing, the total cost per extractor was ca.
$10 (USD), and if tightened properly, the extractors typically last for more than
100 uses.
To perform an extraction with these simple static units, a sample is placed in
the cell, water is added, and the cell is capped. A headspace of air (or an inert gas
like nitrogen) is left above the water. When the water is heated (typically in an
old GC oven), the internal pressure is controlled by the steam/liquid equilibrium
pressure, i.e., the bottom of the cell contains the sample and the liquid phase of
water, while the top contains steam. After 15-60 min of heating, the cells are
cooled under tap water, and opened to collect the water extract.
It is critically importantnot to overfill these cells. To avoid excessive pressure
buildup, sufficient headspace must remain so that the pressure is controlled by
the steam/water equilibrium. (A convenient reference for the steam/liquid water
equilibrium pressures at various temperatures can be found in Refs. [25] and
[43], along with several other useful tables of water characteristics.)
The initial use of these cells was to provide a simple way to extract organic
pollutants from contaminated soils, and then determine their concentrations in
the extract water using SPME [20]. (In essence, we were attempting to turn a
soil sample into a water sample suitable for SPME analysis.) For more polar
analytes (e.g., chloroanilines), the solubility in ambient water was sufficient
enough that the target organics remained in the water phase after the static hot
water extraction of the soil. Thus, the water could be removed and the concen-
trations of the pollutants determined using normal SPME calibration
procedures.
However, non-polar solutes like PAHs showed substantial partitioning back
to the soil samples when the cell was cooled. While this initially made determin-
ing the concentration of the PAHs on the soil impossible, the addition of
perdeuterated PAHs as internal standards to the soil/water mixture prior to
extraction allowed good quantitative results from this simple static cell extrac-

595
Fig. 18.9. Comparison ofPAH concentrations obtained by 1 h 250°C hot water/SPME extraction
and by 48 h Soxhlet extraction (certified values) for urban dust (SRM 1649) [201.

tion followed by SPME and GC/MS analysis, as shown in Fig. 18.9 by the analysis
of urban dust [20]. However, the internal standard must behave the same as the
target analyte both in partitioning between the soil and water (upon cooling the
extraction cell) and in the SPME sorption step from the extract water. For
compounds where isotopically labeled internal standards are available, this
approach yields good quantitative data. However, if isotopically labeled internal
standards are not available, quantitation is less reliable. For example, semi-
quantitative determinations of soil PCB concentrations were possible using two
congeners (PCB 103 and 169), which are not present in environmental samples,
as internal standards [23]. SPME has more recently been used to determine
chlorophenols and methyl mercury in subcritical water extracts [21,22].
The static extraction cells have also been used with solid-phase trapping by
including an "Empore" sorbent disc in the cell during the static extraction step
[17,18]. In this technique, the cell is slowly rotated during the heating step so
that the water, soil, and sorbent disc are well-mixed. The mixing continues dur-
ing cooling, and the organics collect on the disc as their solubility in water is
reduced. The disc is then removed and extracted in a few ml of organic solvent to
recover the extracted organics.
Surprisingly, this very simple (and inexpensive) procedure is capable of
yielding quantitative extraction and collection of organic pollutants from soil.
Collection efficiencies of non-polars is quite good, with > 90% of the mass of each
PAH from a coal tar contaminated soil being collected on the sorbent disc, rather
than repartitioning to the soil during the cooling step [18]. This "low-tech"
approach yields quite good quantitative agreement with the concentrations of
PAHs determined in a coal tar contaminated soil, as compared to the concentra-
tions determined using 18 h of Soxhlet extraction (Fig. 18.10) [18].

596
 500

400

E 300

E 200
w

100

Fig. 18.10. Comparison of PAH concentrations obtained by 1 h 250°C hot water/SPE (Empore
disc) extraction and by 18 h Soxhlet extraction for manufactured plant soil [18].

18.3.3 Addition of Modifiers

As described above, ambient water is too polar to be an effective extraction

solvent for moderate- and low-polarity organics, just like supercritical CO2 is too
non-polar for all but the lower-polarity organics. The majority of investigators
who are developing hot water extraction have a background in SFE with CO,
and are familiar with adding organic modifiers to increase the polarity of CO2.
Similarly, it may be useful to add organic modifiers (e.g., methanol, or other
water/miscible solvents) to decrease the polarity of water, and thus lower the
temperature required to solvate a particular analyte.
Our initial investigations with several organic modifiers (e.g., methanol,
ethanol, acetonitrile, acetone) had little effect on the extraction efficiencies of
non-polar analytes such as PAHs, until the modifier concentrations were quite
high (e.g., >20%). Therefore, we preferred to raise the temperature of pure
water as the simpler (and "greener") alternative.
However, other investigators have improved the recoveries of a variety of
compounds with the aid of additives to the hot water. For example, 30% ethanol
in water was used to improve the recovery of nonylphenol polyethoxy carboxy-
lates from waste sludges [26]. The recovery of PAHs from soil was improved by
adding sodium dodecyl sulfate (SDS) as a micelle former Fig. 18.11 [27, 28].
Analogous to SFE with CO2 , modifiers have been added which affect the matrix
rather than simply increasing the analyte solubility. For example, phosphate-
buffered water was found to recover herbicides which were sequestered in soil
humic material [12]. Finally, acidified water has been used to remove metals
from waste oils (Fig. 18.12) [29], and from coal [30].

597
 120 SDS
100
-

-80 1
0 i / water
o 40

0
0 10 20 30 40 50
Dynamic extraction time Imin)

Left: Fig. 18.11. Hot water extraction of benzo[ajpyrene from soil with and without the addition
of sodium dodecyl sulfate (SDS). Adapted from Ref. [271.
Right: Fig. 18.12. Extraction of metals from industrial oils with 150°C hot water modified with
4% (v/v) HNO3 + 0.1 M KCI. Adapted from [29].

The effect of solvent polarity, changed either by temperature or by modifier,

on the solubility of atrazine is shown in Fig. 18.13. Although the solubility trends
are similar, the slope of ethanol/water curve is higher, which can only be
explained by additional analyte cosolvent interactions with the ethanol [6]. This
was confirmed by additional tests where the use of water modified with 12% eth-
anol at 100°C ( = 49) resulted in a 2-fold increase in atrazine solubility in com-
parison to solubility in pure water at 125°C ( = 48) [6].

18.4 EXTRACTION SELECTIVITY

Supercritical CO2 has often been praised as being a "selective" extraction fluid,
since pressure and temperature can be used to change its solvent strength. As
auuu

water modified with ethanol

2500

2000

E
-8 1500oo

1000

500

Fig. 18.13. Dependence of atrazine solubility on

v-a- r
--
pultai- y
1---y
y
.la,
lllliru
ie 1-
y
t c f-n c a -It,, Jr
v-
u
Lilrrau~, I
changed by adding ethanol. Used by permission 75 70 65 60 55 50 45
from M. Curren and J. King [6]. dielectric constant,

598
 1UU

80 374.9 °C

6.
0 Liquid water
'E
0
25 C
251C S water,
8
S0

Q 40
Z -methano
o-nitrololuene
5
20
-pnethyamiP _

CCI4 f PAHs _
00 200 300 400 500
Temperature,°C

Fig. 18.14. Dielectric constants of various solvents and solutes at room temperature compared to
the dielectric constant of water at different temperatures (220 bar).

shown in Table 18.1, the polarity of CO 2 (as measured by its dielectric constant)
can be controlled from ca. £ = 1 to 2 (or roughly in the range of alkane solvents).
In contrast, the polarity of hot water can be controlled from ca. 13 (at 350°C) to
80 (at ambient temperature). Thus, while hot water cannot achieve the low
polarity of CO,, its polarity can easily be controlled from E = 80 to that of
solvents like acetone ( = 18) and ethanol ( = 24), simply by raising the
temperature (Fig. 18.14). In our lab, this ability to control water's polarity
enables us to selectively extract different classes of compounds better than with
supercritical CO,.
An example of water's selectivity is shown in Fig. 18.15 by a comparison of
extracts from urban air particulate matter [31]. While both techniques yield
quantitative recoveries of PAHs compared to Soxhlet extraction, supercritical
CO 2 extracts large quantities of n- and branched alkanes, which completely
obscure the PAHs in the GC chromatogram (Fig. 18.15). In contrast, the
chromatogram from the hot water extract shows primarily PAHs, with essen-
tially no interfering alkanes. In fact, hot water is not capable of extracting
alkanes larger than ca. C14, unless the water is depressurized to steam [32,33].
While we have found the selectivity of hot water for extracting different
compound classes to be much better than that of supercritical CO2, it must be
noted that hot water is much less selective than supercritical CO 2 in regards to
extracting matrix components, especially for soils and sediments [31]. Hot water
extracts of soils and sediments are often dark brown, especially at higher
extraction temperatures, while supercritical CO2 extracts of the same soils will
typically be clear or yellow. A recent comparison of various extraction methods
for extracting PAHs from soil showed that water at 250°C and supercritical CO2
both extracted ca. 8 mg/g soil of matrix material (based on the residue weight
after evaporation of the collection solvent). However, when 300°C water

599
 0

41 0
2 .0 0

I
IS
0 0B
At Subcritical Waer
K

0 00
U
0
0.1 s~~~~~~~~~~ 0
21

AL
0 e

N 0h

0 0

- 0
9;
0
3. i5 Retention Time (min) 35

la

25 Retention Time (min) 35

40
V)

3.
Ai

25 Retention Time (min) 35

Fig. 18.15. Comparison of GC/MS chromatograms of urban air particulate extracts obtained with
subcritical water, SFE with pure CO2 , and Soxhlet extraction [31].

600
 100

80

0 60

-40

20
Fig. 18.16. Selective extrac-
tion of oxygenates versus
-

terpenes from savory using 0

100°C subcritical water 0 0 10 15 20 25 30 35 40
[34]. Time, min

extractions were used, the extracted residue increased to 13 mg/g soil, which is
similar to the 15 mg/g extracted by pressurized liquid extraction. However, each
technique removed less matrix material than Soxhlet extraction, which removed
ca. 100 mg/g soil. SFE with pure CO2 also removed the least organic matrix
material from the soil of all of the techniques, and its extracts showed less matrix
material in the early parts of the GC/MS chromatograms [31].
The relative ability to selectively and quantitatively extract flavor and
fragrance compounds has also been demonstrated [34]. For example, hot water
extraction of savory yields much faster recovery of desirable (and more valuable)
oxygenates such as thymol, carvacrol, borneol, and linalool than of the less
desirable (and less valuable) monoterpenes (Fig. 18.16). With 100°C water, ca.
80% of the oxygenated compounds can be recovered in 20 min, while the extract
contains only a few percent of the mono- and sesquiterpenes. In contrast,
supercritical CO 2 shows little or no selectivity between oxygenates and terpenes.
Supercritical CO2 extracts also contain substantial concentrations of the
undesirable plant wax alkanes, while hot water extracts contained no detectable
concentrations of the alkanes (Fig. 18.17) [34].
A very elegant combination of supercritical CO 2 extraction with hot water
extraction utilizes the strengths of each fluid [19]. The non-polar herbicide
"Dacthal" (the dimethyl ester of 2,3,5,6-tetrachloro-1,4-benzenedicarboxylate)
was first extracted from the soil using SFE with pure CO2 . Next, the same soil
sample was extracted with hot water to extract the mono- and diacid metabo-
lites. This combination of SFE and hot water's strengths can be performed on a
single instrument, and deserves further attention.

18.5 HOT WATER REVERSE PHASE LC

Although the focus of this chapter is on hot water extractions, its use for
reverse-phase liquid chromatographic separations has received some recent
attention [35-42,44]. The concept for reversed-phase LC using water is easily

601
 a) subcritical water extract

a
/:
0
o
o

CD

i0I 2I I -
10 20 30
Retention time, min

a)
c
0
0~
5a
P
C:
Fig. 18.17. Comparison gCD
of GC chromatograms
of savory extracts CD
obtained with selective
subcritical water and 1- I I I I I I
SFE with pure CO 2 0 10 ... . 20 30
[34]. Retenton time, min

seen in Fig. 18.2. Note that heating the water to ca. 200°C achieves the same drop
in solvent polarity as solvent programming from 0 to 100% methanol or aceto-
nitrile. Also, water's surface tension and viscosity also drop with temperature
(Figs. 18.3 and 18.4), both of which enhance LC separations. Several examples
utilizing this idea have been reported for polar (e.g., underivatized amino acids)
to moderately-polar (e.g., phenols and aliphatic alcohols).
In addition to the desire to reduce the use of organic solvents, one of the more
intriguing characteristics of water is that it has no response in a flame ionization
detector (FID), which has led to successful use of a GC/FID detector for aliphatic
and aromatic alcohols and underivatized amino acids [38]. However, subsequent
investigators have had less success with this approach, likely because the effect
of various manufacturer's FID designs (and related flame gas flow rates) on the
ability of a particular FID detector to accommodate high water flow rates has not
been well-understood [44].

18.6 USEFUL REACTIONS

A major concern with the use of hot water is the potential for degrading the
organics one wishes to extract. While this certainly occurs with some com-
pounds, the fact that the samples are not exposed to oxygen (water purged with

602
 0
U 100 C 40 min [ 150C 20 min ' 175°C 12 min

0
'o

V
55

Fig. 18.18. Degradation of some savory oxygenates at water temperatures higher than 100°C [34].

nitrogen to remove dissolved oxygen is used in our laboratory), may actually

reduce the amount of degradation for easily-oxidized compounds. For example,
hydrodistillation is normally used to extract essential oils from herbs such as
savory. With hydrodistillation, the exposure to atmospheric oxygen causes
almost complete loss of thymoquinone. However, when savory is extracted with
subcritical water (or, by SFE with CO2), no oxidation of the thymoquinone
occurs, and it is efficiently recovered [34]. Even with 175°C water extraction, the
recovery of thymoquinone is much higher using hot water extraction than
hydrodistillation. However, some compounds may degrade in hot water, as noted
by the loss of linalool, borneol and thymoquinone from savory if the extraction
was performed with 175°C water (Fig. 18.18) [34].
While the possibility of oxidative reactions is less likely in hot water, water
certainly becomes more reactive at higher temperatures. This is particularly
true for base- and acid-catalyzed reactions such as hydrolysis. In addition to the
increased reactivity based on the higher temperatures used, the dissociation con-
stant of water increases with temperature so that the actual concentrations of
hydroxide and hydronium ions at 250°C is ca. 25-fold higher than at ambient
temperature [45].
If the organic compound of interest is easily hydrolyzed, hot water extraction
is likely not an appropriate method. However, the hydrolysis of some organics
during hot water extraction has actually been advantageous to their
quantitative determinations. Natural pyrethrins (cinerin I, jasmolin I, and
pyrethrin I) have been hydrolyzed to their parent acids as a extraction/ prepara-
tion step for their subsequent GC determination using SPME [46], and soybean
and vegetable oils have been hydrolyzed to their free fatty acid components for
processing uses, and for their subsequent GC determination as fatty acid methyl
esters (FAME) [47,48].
Hydrolysis reactions during extraction with hot water have also been applied
to the determination of acid herbicides on soil [17]. Acid herbicides are applied as

603
the mixture of many different esters, but are normally analyzed after extraction
and derivatization as a single compound. For acid herbicides, this is often
performed by extracting the herbicides from soil using a Soxhlet apparatus,
hydrolyzing to convert the various esters to the acid form of each herbicide, and
then derivatizing (e.g., with diazomethane) the acid form to a single ester form
for analysis. Thus, for example, even though several different esters of 2,4-di-
chlorophenoxyacetic acid (2,4-D) may be used on a field, they are all converted to
their methyl esters for analysis.
Hydrolysis of acid herbicides on soil with water at 100-150°C has been
performed using the simple static cell described above, with a small piece of an
"Empore" anion exchange disc being placed in the cell during the heating step
[17]. Thus, the herbicides are extracted from the soil, hydrolyzed to their acid
form, then collected on the anion exchange disc as the static cell is cooled. The
acid forms of the herbicides are then derivatized with 1 ml of BSTFA (N,O-
bis(trimethylsilyl)-trifluoroacetamide), and analyzed by GC/MS. Recoveries for
a variety of acid herbicides were > 80% with a 30-min extraction, and the entire
procedure only required 3 ml of water and 1 ml of BSTFA reagent.
Other reactions occur which may or may not be useful, depending on the
investigator's goal. For example, attempts to develop hot water extraction meth-
ods for polychlorinated dioxins (PCDDs) from incinerator fly ash were complete
failures, and even spiked PCDDs could not be recovered with hot water. Subse-
quent studies showed that hot water may be an excellent method for destroying
PCDDs, but is not likely to be a useful analytical extraction method [49].
Similarly, even though PCBs can be quantitatively extracted from soils and sedi-
ments at 250-300°C [23,50,51], higher temperatures can result in dechlorin-
ation of the PCBs, particularly with added reactants such as iron [52]. Some
organics react at even lower temperatures. Attempts to use subcritical water
extraction to recover high explosives (TNT, RDX, and HMX) from contaminated
soils showed substantial degradation even at temperatures as low as 100-125°C.
This observation led to the development of a method to decontaminate soil by
destroying the explosives with subcritical water. Soils contaminated with as
much as 12 wt.% TNT have been cleaned to low ppm levels with just one hour of
exposure to hot water in a static system [53].

18.7 APPLICATIONS AND FUTURE OF HOT WATER EXTRACTION

Although subcritical water extraction is a fairly new approach to analytical

extractions, there has been an impressive range of applications since the tech-
nique was first described in 1994 [1]. Table 18.2 contains a list of representative
applications of hot water extraction from environmental samples, as well as for
recovering desirable compounds from biological samples.
Although the reader is referred to the original papers for additional details, it
is interesting to note that the applications include a very broad range of applica-
tions from a number of sample types which have been developed in a relatively

604
TABLE 18.2
Applications of hot water extraction
Matrix Analytes Comments Ref.
Soil and sediments alkanes SPE' 33
selective extraction 31
alkyl benzenes SPE online HPLC 15
SPME 2 20
aromatic amines SPME 20
chlorinated hydrocarbons SPME 20
chlorophenols modifiers, SPME 21
fungicide (tricyclazole) 54
herbicides SPE 13
consecutive SF CO 2 19
and subcritical water
24
chlorinated acid herbicides SPE, hydrolysis 17
metals (methylmercury) SPME 22
pesticides modified water 55
C18, modified water 12
56
PAHs SPME 20
SPE 33, 18
selective extraction 31
SPE-online LC-GC 14
modifier, SPE 28, 27
SPE online HPLC 15
PCBs 51, 57
SPE 50
SPME 23
PCDFs 58
soil organic carbon, phenanthrene matrix interactions 59
Air particulate/ fly alkanes selective extraction 31
ash PAHs SPME 20
selective extraction 31
SPE 18
Waste/sludges alkanes, alkyl benzenes, phenols, PAHs selective extraction 32
surfactant-nonylphenol polyethoxy modified (30% ethanol) 26
carboxylate
metal (Se) derivatization 60
organic waste (epoxy resin) hydrolysis 61
Food fungicides 62
Industrial oil metals acidification 29
Coal metals (Se, As, Hg) modifier, derivatization 30
Chrysanthemum insecticide hydrolysis 46
Clove buds eugenol, eugenyl acetate, caryophyllene - 63
Eucalyptus terpenes, oxygenates - 64
Fennel monoterpenes, oxygenates - 65
continued

605
Matrix Analytes Comments Ref.
Ground tea waste lignocellulosic materials - 66
Kava lactones - 67
Laurel terpenes, oxygenates - 68
Marjoram terpenes, pinenes, alcohols - 69
Oleander glycosides - 70
Peppermint oxygenates, caryophyllene - 71
selective extraction 34
Rosemary terpenes, oxygenates 72
Savory terpenes, oxygenates selective extraction 34
Coconut oil fatty acids from triglycerides hydrolysis 47
Linseed oil fatty acids from triglycerides hydrolysis 47
Soybean oil fatty acids from triglycerides hydrolysis 47, 48
Veronica longifolia glycosides, terpenes 73
2
1SPE = solid phase extraction; SPME = solid phase microextraction.

short time. In our laboratory, hot water extractions have been easier to perform
than SFE (especially in the early days when all SFE instrumentation had to be
fabricated in the lab), and the development of quantitative extraction methods
has gone more quickly. For samples such as soil, hot water extracts do tend to be
"dirtier" than SFE (at least those generated with pure CO,) in terms of contain-
ing more soil organic matter. However, hot water extracts of herbs can be
"cleaner" than SFE extracts, since plant waxes are not recovered with hot water.
The ability of hot water extraction to selectively extract different chemical
classes of analyte compounds far exceeds any selectivity we have been able to
achieve with SFE. In our laboratory, we have attempted to selectively extract
different polarities of compounds from samples ranging from waste sludges to
herbs using both SFE and hot water. In every case, the hot water extractions
yielded better selectivity among the various compound classes.
Quantitative hot water extraction methods for organics ranging from quite
low polarities (e.g., PAHs, PCBs, monoterpenes) to very high polarities (e.g.,
pesticide metabolites, amino acids) have been reported by a variety of laborato-
ries. In contrast, SFE methods for many of the more polar analytes were never
successfully developed, especially without the need for high concentrations of
organic modifiers. Given the characteristics of the two fluids, perhaps the
combination of SFE for non-polar organics, and hot water extraction for moder-
ate- and high-polarity compounds will be the optimal approach in the future.

REFERENCES

1 S.B. Hawthorne, Y. Yang and D.J. Miller, Anal. Chem., 66 (1994) 2912.
2 G.L. Rossling and E.U. Franck, Ber. Bunsen-Ges. Phys. Chem., 87 (1983) 882.
3 Y. Yang, M. Belghazi, A. Lagadec, D.J. Miller and S.B. Hawthorne, J. Chromatogr.A,
810 (1998) 149.
4 D.J. Miller and S.B. Hawthorne, Anal. Chem., 70 (1998) 1618.
5 D.J. Miller and S.B. Hawthorne, A.M. Gizir, A.A. Clifford, J. Chem. Eng. Data, 43

606
 (1998) 1043.
6 M.S.S. Curren and J.W. King, Anal. Chem., 73 (2001) 740.
7 N.D. Sanders, Ind. Eng. Chem. Fundam,. 25 (1986) 169.
8 D.J. Miller, S.B. Hawthorne, A.A. Clifford and S. Zhu, J. Chem. Eng. Data, 41 (1996)
779.
9 Y. Yang, D.J. Miller and S.B. Hawthorne, J. Chem. Eng. Data, 42 (1997) 908.
10 D.M. Miller and S.B. Hawthorne, J. Chem. Eng. Data, 45 (2000) 315.
11 D.M. Miller and S.B. Hawthorne, J. Chem. Eng. Data, 45 (2000) 78.
12 C. Crescenzi, A. Di Corcia, M. Nazzari and R. Samperi, Anal. Chem., 72 (2000) 3050.
13 C. Crescenzi, G. D'Ascenzo, A. Di Corcia, M. Nazzari, S. Marchese and R. Samperi,
Anal. Chem., 71 (1999) 2157.
14 T. Hyotyl'ainen, T. Andersson, K. Hartonen, K. Kuosmanen and M.-L. Riekkola,
Anal. Chem., 72 (2000) 3070.
15 Y. Yang and B. Li, Anal. Chem., 71 (1999) 1491.
16 B. Li, Y. Yang, Y. Gan, C.D. Eaton, P. He and A.D. Jones, J. Chromatogr. A, 873
(2000) 175.
17 X. Lou, D.J. Miller and S.B. Hawthorne, Anal. Chem., 72 (2000) 481.
18 S.B. Hawthorne, S. Trembley, C.L. Monoit, C.B. Grabanski and D.J. Miller, J.
Chromatogr.A, 886 (2000) 237.
19 J.A. Field, K. Monohan and R. Reed, Anal. Chem., 70 (1998) 1956.
20 K.J. Hageman, L. Mazeas, C.B. Grabanski, D.J. Miller and S.B. Hawthorne, Anal.
Chem., 68 (1996) 3892.
21 L. Wennrich, P. Popp and M. Moder, Anal. Chem., 72 (2000) 546.
22 A. Beichert, S. Padberg and B.W. Wenclawiak, Appl. Organomet. Chem., 14 (2000)
493.
23 S.B. Hawthorne, C.B. Grabanski, K.J. Hageman and D.J. Miller, J. Chromatogr.A,
814 (1998) 151.
24 M.S. Krieger, J.L. Wynn and R.N. Yoder, J. Chromatogr., 897 (2000) 405.
25 L. Haar, J.S. Gallagher and G.S. Kell, National Bureau of Standards/National
Research Council Steam Tables. Hemisphere Publishing, Bristol, PA, 1984.
26 J.A. Field and R.L. Reed, Environ. Sci. Technol., 33 (1999) 2782.
27 V. Fernandez-Perez and M.D.L. de Castro, J. Chromatogr., 902 (2000) 357.
28 B. Vallejo-Pecharroman, L.E.G. Ayuso and M.D.L. de Castro, Chromatographia,53
(2001) 5.
29 V. Fernbndez-Prez, M.M. Jimbnez-Carmona and M.D.L. de Castro, Anal. Chim.
Acta, 433 (2001) 47.
30 V. Ferndndez-PBrez, M.M. Jim6nez-Carmona and M.D.L. de Castro, J. Anal. Atom.
Spectrom., 14 (1999) 1761.
31 S.B. Hawthorne, C.B. Grabanski, E. Martin and D.J. Miller, J. Chromatogr.A, 892
(2000) 421.
32 Y. Yang, S.B. Hawthorne and D.J. Miller, Environ. Sci. Technol., 31 (1997) 430.
33 K. Hartonen, G. Meissner, T. KesalB and M.-L. Riekkola, J. Microcolumn Sep., 12
(2000) 412.
34 A. Kubatovd, A.J.M. Lagadec, D.J. Miller and S.B. Hawthorne, Flavour Fragr.J., 16
(2001) 64.
35 R.M. Smith, O. Chienthavorn, S. Saha, I.D. Wilson, B. Wright and S.D. Taylor, J.
Chromatogr.A, 886 (2000) 289.
36 R.M. Smith and R.B. Burgess, Anal. Commun., 33 (1996) 327.
37 R.M. Smith and R.B. Burgess, J. Chromatogr.A, 785 (1997) 49.
38 D.J. Miller and S.B. Hawthorne, Anal. Chem., 69 (1997) 623.
39 Y. Yang, A.D. Jones and C.D. Eaton, Anal. Chem., 71 (1999) 3808.
40 T.M. Pawlowski and C.F. Poole, Anal. Commun., 36 (1999) 71.

607
41 S.M. Fields, C.Q. Ye, D.D. Zhang, B.R. Branch, X.J. Zhang and N. Okafo, J.
Chromatogr.A, 913 (2001) 197.
42 T.E. Young, S.T. Ecker, R.E. Synovec, N.T. Hawley, J.P. Lomber and C.M. Wai,
Talanta, 45 (1998) 1189.
43 Generated using NIST/ASME Steam Formulation Database 10, version 2.2, A.H.
Harvey, A.P. Peskin and S.A. Klein.
44 B.A. Ingelse, H.-G. Janssen and C.A. Cramers, J. High Resol. Chromatogr., 21 (1998)
613.
45 W.L. Marshall and E.U. Franck, J. Phys. Chem. Ref. Data, 10 (1981) 295.
46 M. Krappe, S.B. Hawthorne and B.W. Wenclawiak, Fresenius J. Anal. Chem., 364
(1999) 625.
47 R.L. Holliday, J.W. King and G.R. List, Ind. Eng. Chem. Res., 16 (1997) 932.
48 J.W. King, R.L. Holliday and G.R. List, Green Chemistry (1999) 261.
49 I. Windal, S.B. Hawthorne and E. De Pauw, Organohalogen. Compounds, 40 (1999)
40 591.
50 K. Hartonen, K. Inkala, M. Kangas and M.-L. Riekkola, J. Chromatogr.A, 785 (1997)
219.
51 Y.Yang, S. Bowadt, S.B. Hawthorne and D.J. Miller, Anal. Chem., 67 (1995) 4571.
52 H.K. Yak, B.W. Wenclawiak, I.F. Cheng, J.G. Doyle and C.M. Wai, Environ. Sci.
Technol., 33 (1999) 1307.
53 S.B. Hawthorne, A.J.M. Lagadec, D. Kalderis, A.V. Lilke and D.J. Miller, Environ.
Sci. Technol., 34 (2000) 3224.
54 M.S. Krieger, W.L. Cook, L.M. Kennard, J. Agric. Food Chem., 48 (2000) 2178.
55 A. di Corcia, A.B. Caracciolo, C. Crescenzi, G. Giuliano, S. Murtas and R. Samperi,
Environ. Sci. Technol., 33 (1999) 3271.
56 M.M. Jim6nez-Carmona, J.J. Manclus, A. Montoya and M.D. Luque de Castro, J.
Chromatogr., 785 (1997) 329.
57 S. Pross and B.W. Wenclawiak, Fres. J. Anal. Chem., 367 (2000) 89.
58 B. van Bavel, K. Hartonen, C. Rappe and M.L. Riekkola, Analyst, 124 (1999) 1351.
59 M.D. Johnson, W.L. Huang, Z. Dang and W.J. Weber, Environ. Sci. Technol., 33
(1999) 1657.
60 C.M.R. Varade and M.D.L. de Castro, J. Anal. Atom. Spectrom., 13 (1998) 787.
61 C. Fromonteil, Ph. Bardelle and F. Cansell, Ind. Eng. Chem. Res., 39 (2000) 922.
62 T.M. Pawlowski and C.F. Poole, J. Agric. Food Chem., 46 (1998) 3124.
63 A.A. Clifford, A. Basile and S.H.R. Saidi, Fres. J. Anal. Chem., 364 (1999) 635.
64 M.M. Jimenez-Carmona and M.D.L. de Castro, Chromatographia,50 (1999) 578.
65 L. Gdmiz-Gracia and M.D.L. de Castro, Talanta, 51 (2000) 1179.
66 A. Demirbas, A. Caglar, A. Ayas and S. Karslioglu, Fuel Sci. Technol. Int., 14 (1996)
395.
67 A. Kubatova, D.J. Miller and S.B. Hawthorne, J. Chromatogr., 923 (2001) 187.
68 V. Ferndndez-P6rez, M.M. Jim6nez-Carmona and M.D.L. de Castro, Analyst, 125
(2000) 481.
69 M.M. Jim6nez-Carmona, J.L. Ubera and M.D.L. de Castro, J. Chromatogr., 855
(1999) 625.
70 X. M. Wang, J.B. Plomley, R.A. Newman and A. Cisneros, Anal. Chem., 72 (2000)
3547.
71 A. Ammann, D.C. Hinz, R.S. Addleman, C.M. Wai and B.W. Wenclawiak, Fres. J.
Anal. Chem., 364 (1999) 650.
72 A. Basile, M.M. Jim6nez-Carmona and A.A. Clifford, J. Agric. Food Chem., 46 (1998)
5205.
73 J. Suomi, H. Siren, K. Hartonen and M.-L. Riekkola, J. Chromatogr., 868 (2000) 73.

608
Chapter 19

Recent developments in the chemistry and

application of analytical derivatizations
Jack M. Rosenfeld

19.1 INTRODUCTION

19.1.1 Overview

Analytical derivatizations are considered by many separation scientists as the

means of last resort. The additional steps, interferences arising from excess
reagents and matrix effects all mitigate against use of this technique. Despite
such apparent disadvantages, numerous applications continue to be reported, as
are new developments in the basic chemistry and innovations in instrumenta-
tion. The utility of, and interest in, this sub-discipline of the separation sciences
is perhaps best exemplified by the number of reviews over the last 18 months
that cited analytical derivatizations as an important aspect of analytical prob-
lems being considered [1-3]. Other current reviews are cited below in the
context of specific functional groups or detection techniques. The work of
Dovichi's group is telling on the importance of analytical derivatizations. Krylov
et al. [4] established that the science has now reached the point where it is
feasible and necessary to consider how determination of analytes from a single
cell eliminates biases resulting from the analysis of numerous (e.g., 106) cells [4].
The bias is generated by enzymes which continue non-metabolic reactions
during standard lysis procedures. These reactions are reduced in capillary zone
electrophoresis (CZE) by the fast lysis of the cell and the rapid separation of
enzymes from substrate. Such developments demonstrate the intellectual
vibrancy of the field and have immense impact on studies in biology, medicine
and in environmental sciences.
Recent reviews of the field have focused on determination of specific classes
of compounds and analytical problems and/or techniques, including the determi-
nation of phosphorus based pesticides [1], electrophoresis [2] and food analysis
[3]. Accordingly, this chapter takes the approach of functional group analysis.
For instance, analytical derivatization of amino acids is considered under the

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 609
reactions of amines and thiols. Saccharide analyses are, in general, dependent on
reactions at the reducing terminal and so are considered as aldehyde reactions.
Given the extent of the recently reviewed literature on the subject, the focus
is on the most recent publications with earlier work cited where appropriate.
The source of the analytical problems is identified since issues in biology,
environmental toxicology or drug regulation set the analyte(s), the matrix and
the sensitivity required.
A major purpose of analytical derivatizations is to enhance detection by a
selected technique. The chapter is therefore organized on the basis of the
detection techniques. Within the sections on detection, specific classes of
analytes are examined (e.g., amines). Subsections on the analyte class contain
discussions of the different reagents.

19.1.2 Reactions and mechanisms

The chemistry of analytical derivatizations is of necessity fairly simple. These

most common reagents are alkyl bromides, acid chlorides, alkylchloroformates,
N-hydroxysuccinimidyl esters and chlorides or ethers of trimethylsilyl groups. A
series of one step reactions that derivatize a hydroxyl, amine or a thiol are shown
in Fig. 19.1. An important feature here is the concomitant release of acid or some
other product of the reaction. This can either alter the reaction conditions by
changing pH or in the case of residual N-trimethylsilyltrifluoroacetamide follow-
ing trimethylsilylation can interfere with the chromatography. Derivatization of
carboxylic acids shown in Fig. 19.2 are also one step reactions. Again, acid or
some other product of the esterification or amidation remains in the reaction
mixture. The acid in this case is often buffered by the excess of base used to
derivatize the carboxyl group. If strong acids are used then these must be
removed from the derivatives to preserve the injection devices, columns and
detectors.
The reactions pathways shown in Fig. 19.3A indicate that derivatization of
carbonyls is somewhat more complicated. The hydroxylamine intermediate can
decompose to the initial or can dehydrate to the hydrazone which is stable. Con-
ditions have to be such as to favour the latter reaction. Formation of the
hydroxylamine moves the carbon from a planar SP2 to a tetrahedral SP3. From
this it is evident that steric factors on the analyte will affect the equilibrium
between the hydroxylamine and the starting materials.
In certain instances the formation of the hydrazone is not the final step. If
the analyte is a di-aldehyde, or has a moiety that is in conjugation with an alde-
hyde (or possible a ketone), then further reaction is possible. The secondary
amine on the hydrazone can further react with the second carbonyl to form the
cyclic product (Fig. 19.3B).
Detection methods that provide the best sensitivity are established by the Rd
which is defined as the moiety of the derivatizing agent that imparts the
detectability. Numerous options exist for selecting an appropriate Rd. Choices

610
 A RdXH B R.XH

II o I I0
R CI RR CO

C 'n + RXH D RdXnH

"o (CH,) 3S
0~CR

Io
+

RXH,
70 1
R.XSi(CH 3 )3
OH

Fig. 19.1. Prototypical derivatizations of amines, hydroxyls and thiols. X = nucleophile

(NH,OH,SH). Rd = Detection moiety on the reagent.

/CHBr (cH ROH

)

(Et)3N R O

R R /C\ =N\OR

RdNH ECF

NHR

Fig. 19.2. Prototypical derivatizations of carboxylic acids.

are particularly varied for the chromophoric or fluorophoric techniques

associated with high performance liquid chromatography (HPLC) or CZE. These
have recently been summarized by Oguri 2].
The few reactions summarized here can also be combined with other re-
agents to produce more complex derivatives, providing always that the reaction
conditions themselves are simple in execution. Several such examples will be
throughout this
discussed throughout this chapter.
chapter.

611
 A 0 B H H
C=N-N-R d
,C\R

Hi H2NNRd

I H'
4H

7 RI OHNNR
H

R. HNNRJ

AC Rd
H H
H.

R, NHRd
R, NHRd
C=N "' Rd
C=N HH~ ~R
/¥ R2

Fig. 19.3. Derivatization of a carbonyl with a hydrazine.

19.2 FLUORESCENCE

High sensitivity available from fluorescence combined with automation and

miniaturization has greatly extended the scope of applications and has led to
development of new analytical tools. Innovations in this field involve standard
derivatizations that are used in new and complex biomedical or environmental
problems, preparation and characterization of new reagents as well as a
continuing evolution of advanced instrumentation.

19.2.1 Amines

19.2.1.1 1,2 Aromatic dialdehydes

Reaction of amines and amino acids with the fluorogenic reagent o-pthaldi-
aldehyde (OPA) preparatory to HPLC with detection by fluorescence (FLU) is a
mature and extensively used technique [3,5-15]. The field was recently reviewed
by Stalikas and Konidari [1], Molnar-Perl [3,5] Boucherau [6], and Wan and
Blomberg [7]. Molnar-Perl [5] also recommended reaction conditions for pre-
column derivatization and provided useful guidance for the application of
existing technologies and for further investigation.
Reaction mechanisms for formation of the fluorescent 1-alkylthiol 2-alkyl
substituted isoindoles (AAI) have been extensively studied. Alvarez-Coque et al.
[8] contributed to the fundamental understanding of the chemistry of formation
and decomposition of derivatives as well as their fluorescent properties.

612
 o 0
II II

c H + H2NR k'=± /
OPA II / O NR
HO H
O
O
11 JZZ/" II
SC-H
CC'H
.1,
11
NR S
R,

H
/H

H SR2
I-
7
C
cI NR

SR2
A -alkylthiol-2-alkyl substituted isoindole

Fig. 19.4. Reactions of OPA/THIOL with amines.

Synthetic reaction mechanisms are summarized in Fig. 19.4. Alvarez-Coque et

al. [8] demonstrated that while amines react rapidly with OPA to form iso-
indoles, the derivatives decompose by two distinct pathways: spontaneously and
under catalysis by excess OPA. Inclusion of thiols such as 2-mercaptoethanol
(2ME), 3-mercaptopropionic acid (MPA) and N-acetyl cysteine (NAC) in the
reaction mixture produced 1-alkylthiol 2-alkyl substituted isoindoles which are
more stable. Such knowledge is essential in applying this useful but complex
reaction.
As a general rule, derivatives with more bulky and lipophilic thiol groups
were more stable than those with smaller lipophilic moieties but they also had
reduced fluorescence. This former class of derivatives, however, proved useful
for HPLC or CZE with electrochemical detection (discussed below). A combina-
tion of OPA and one of the thiols 2ME, MPA or NAC produces derivatives with a
useful combination of chemical stability and high fluorescence as well as provid-
ing water soluble reagents that had less objectionable odour. Despite these com-
plicating issues, OPA derivatization, when aided by understandings of the basic
chemistry, is a reaction with numerous applications. The development of highly
automated methods for determination of amino acids subsequent to reaction

613
with OPA/Thiols [10,11] demonstrated the utility and practicality of this
approach.
New ideas on, and new applications of, derivatizations with OPA continue to
be reported, particularly the use of linked multiple detectors. Vasanits et al. [12]
described an extensive and detailed study of the determination of 31 amino acids
present in apples via derivatization with OPA using either MPA or NAC. A
major difference in the use of either of these two thiols is that formation of
OPA/MPA derivatives required a higher temperature than those obtained from
OPA/NAC derivatization. Optimization of chromatographic conditions and the
linking of fluorescent and photodiode array detectors in series maximized the
information derived from these analyses. Sensitivity remained unaffected by the
type of column used, allowing wide choice in selection of columns for developing
complex separations without undue concern about sacrificing sensitivity for
resolution.
Van Eijk et al. [14] simultaneously determined both plasma concentrations
of amino acids and their isotopic enrichment in a study on relative rates of syn-
thesis and degradation of proteins. The amino acids were derivatized with
OPA/MPA. Fluorescence detection provided concentrations and the effluent
from the fluorimeter was fed directly to an interfaced mass spectrometer for
determination of the isotopic composition. Differentially lower sensitivities for
mass spectrometric detection of the more polar analytes were attributed to
excess hydration interfering with the ionization process and not a matter inher-
ent in the formation or stability of the derivative. Sensitivity to the more polar
analytes was improved to some extent by increased temperature of the transfer
line rather than modification of the derivatizing reagent or reaction conditions.
Sampling by microdialysis is one of the more advanced techniques for in uivo
studies in physiology [15]. Yang et al. [16] combined this technique with analyti-
cal derivatization to determine extracellular glutamate and other amino acids in
cerebrospinal fluid (CSF). That biofluid was sampled via microdialysis and
mixed on-line with OPA to form fluorescent derivatives preparatory to
HPLC/FLU. The authors pointed out the importance of developing the relatively
standard HPLC/FLU as well as the more advanced CZE with laser induced
fluorescence (LIF) as the former was more widely available. Glutamate is an
excitatory amino acid involved in a variety of neurodegenerative disorders such
as Parlinson's and Alzheimer's diseases. Many investigators will wish to under-
take studies that involve the in vivo determination of glutamate and other amino
acids and these efforts would benefit from a wider range of instrumental options
such as those described by Yang and co-workers.
Wu and co-workers [17] as well as Ruberu et al. [18] described innovative
techniques using post-column derivatization by OPA/2ME and detection by
fluorescence. The latter group determined phenylurea pesticides. After separa-
tion these analytes by HPLC were decomposed to monoalkylamines by post-
column photolysis. The monoalkylamines were then derivatized with OPA/2ME.
Wu and co-workers [17] used a double function of quaternary ammonium

614
surfactants to determine N-methylcarbamate (NMC) pesticides. The surfactant
served as a hydrophobic pseudophase in micellular electrokinetic capillary chro-
matography (MECC) as well as (particularly cetyltrimethylammonium bromide,
CTAB) the catalyst for the thermal decomposition of NMCs to methylamine.
Again the alkylamine was derivatized with OPA/2ME for high sensitivity
detection by fluorescence. Separation, thermal decomposition and analytical
derivatization of the methylamine were performed in the capillary without any
interfacing. Despite multiple uses of the single capillary, the authors reported
column efficiencies of 50,000 and limits of detection below 0.5 ppm. The
processes and chemistry involved are complex and the methods required careful
optimization, yet they are simple in actual implementation and require minimal
sample preparation and interfacing.
Herraez-Hernandez and Campins-Falco [19] extended the OPA/THIOL
derivatization to include secondary amines. In developing the method for deter-
mination of ephedrine, a secondary amine, NaOC1 or chloramine-T provided a
Cl- that catalyzed the formation of an imine with subsequent hydrolysis to the
primary amine which was derivatized with OPA/NAC. The authors detailed the
optimization of this technique but did not clearly identify the structure of the
compounds.
Naphthalene-2,3-dicarboxaldehyde (NDA) also found considerable applica-
tion to the determination of amines. Again, a nucleophile is required to produce
a stable and highly fluorescent derivative. In this case, however, cyanide (CN)
rather than a thiol proved superior [20-27]. The structure of these 1-cyano-2-
substituted-benz[flisoindoles has been characterized by matrix assisted laser
desorption ionization (MALDI) and time-of-flight (TOF) mass spectrometry
[20]. Shah et al. [21] tested the stability of the cyano derivatives at 10°C over a 15
h period. Under these conditions there was no evidence of decomposition. In
contrast, OPA derivatives were stable for 5 h at 0°C.
Complex analytical problems are readily addressed through the use of
NDA/CN. Rapid derivatization of peptides [22], including those containing
highly basic lysine residues [23], expanded the biological applications. The
derivatives were readily separated by MECC and derivatives of these analytes
[22] were detected by LIF. Applications and analytical techniques were not
limited to biogenic molecules. Chiang et al. [24,25] separated the enantiomers of
bacolfen (4-amino-3-p-chlorophenylbutyric acid-an anti-inflammatory drug)
by derivatizing with NDA/CN and using CZE with a cyclodextrin in the buffer as
a chiral selector.
Advanced automated techniques also exploited sensitivity available for
derivatization with this reagent. A microchip device [26] combined enzymatic
reactions, electrophoretic separation of the reactants from the products and
post-column derivatization of proteins and peptides with NDA. Gottschlich et al.
[26] constructed this device and used it to measure peptides and amino acids
derived from insulin. They digested an oxidized insulin B-chain with trypsin
under stopped flow conditions in a heated channel of the device. Subsequent

615
separation was completed in 1 min and the separated peptides and proteins etc.
were derivatized with NDA in a post-column reactor and detected by laser-
induced fluorescence. It was also possible to reduce the disulphide bonds of
insulin on this device (reduction of disulphide bonds is discussed below). The
work demonstrated that microchip devices can perform all reactions, separa-
tions, derivatizations and detections involved in studies of proteins, peptides and
amino acids.
Parmentier et al. [27] described elegant chemistry and instrumentation used
for simultaneous determination of reactive oxygen species (ROS) and gluta-
thione (GSH). The bio-protective peptides, GSH and y-glutamylcysteine were
derivatized at the amine NDA/CN to form the corresponding fluorescent
products although the structure of these were not identified (see below re
derivatization of GSH with OPA). The toxic ROS were reduced by dihydro-
rhodamine-123 (DHR-123) which was itself oxidized to the fluorescent dye
rhodamine-123 (Rh-123). All fluorescent compounds were then separated by
CZE and detected LIF. The technique was sophisticated in all its components
but it provided a rapid and simultaneous measure of the cellular concentrations
of the toxic ROS and their endogenous scavenger.

19.2.1.2 5-Furoylquinoline-3-carboxaldehyde
Despite a structure that is quite dissimilar to either OPA or NDA, 5-furoyl-
quinoline-3-carboxaldehyde (FQ) also reacts with the amino acids in the
presence of CN to produce fluorophoric isoindoles [28]. As in the case of both
OPA and NDA, this compound has the desirable property of being a fluorogenic
rather than a fluorophoric reagent [29]. Sensitivity of analysis engendered by
the derivatives is exceptionally high-although this can vary with analyte
structure. In addition, since FQ is neutral its mobility is approximately equal to
the electro-osmotic flow and further reduces likelihood of interference.
The reaction has been very thoroughly studied. Activation energies and
reaction rates constant for derivatization of several amino acids with FQ were
recently determined for a number of amino acids [30]. While the activation
energy was relatively low (11-34 kJ), so was the rate constant. For most amino
acids the low rate constant indicated that complete reaction required about an
hour at 60°C. There was also high variation in sensitivity amongst the different
amino acids. This was attributed to differences in spectral properties (molar
absorbtivity, emission spectrum, fluorescence quantum yield) rather than less
efficient yield. Dovichi's group has provided considerable data on mechanisms
for derivatization of amino acids with FQ but there is yet more to learn regarding
this very important reaction. For instance, it is not clear why the derivatization
of proteins with FQ at 65°C [31] and was complete within minutes rather than
the hour required for amino acids [32]. Although, in some respects, detailed
knowledge of the reaction may be incomplete, the sensitivities achieved were
sufficiently high to detect amino acids and biogenic amines in 4 ul of brain

616
dialysate [28]. The slower reaction rates for these analytes necessitated analysis
of the samples in the off-line mode.

19.2.1.3 Chemical cytometry by profiling proteins

The high sensitivities available from the combination of CZE/LIF and deriva-
tization with FQ led to the technique of chemical cytometry. This technique is
defined as the qualitative and quantitative determination of chemicals in a
single cell [33,34]. Classical flow cytometry exploits diverse ligands such as
calcium chelators that fluoresce when bound to this cation or fluorescent
labelled antibody that can tag a cell with a specific antigen. Cells are passed
through a laser beam and both forward scatter and fluorescence measurements
are obtained. Different blood cells such as lymphocytes, neutrophils, and mono-
cytes produce a distinctive forward light scatter pattern that identifies the
individual type. The fluorescent spectrum identifies those cells that are labelled
with the fluorophore. For an individual cell it is thus possible to identify cell
types, immunogenicity and specific biochemical events such as calcium release.
Chemical cytometry is based on labelling of proteins or oligosaccharides with
fluorophores and determining profiles of these analytes from a single cell by
CZE/LIF.
Rapid reaction between FQ and proteins facilitated on-column derivati-
zation of the analytes followed by CZE/LIF. The reaction chemistry was similar
to that from the "test tube" reaction requiring NaCN and high concentrations
(10 mM) of FQ. A reaction time of 30 s was sufficient to lyse the cell and
derivatize the proteins. Subsequently, the derivatized proteins were separated
at 16 Kv. Protein profiles were obtained from HT29 human colon adeno-
carcinoma cells (HT29 cells) [35] and from a single-cell embryo of the intestinal
parasite caenorhabditis elegans [36]. In general, the profiles from single cells
and the cell extract from 106 cells were similar but not identical. The protein
profile from individual HT29 cells were qualitatively and quantitatively differ-
ent from each other. Profiles from single cells could also be distinguished from
the one obtained from an extract. Studies on single embryo of caenorhabditis
elegans embryo focused on the developmental biology and gene expression. It
may lead to a deeper understanding of how a founder cell differentiates into
progeny.

19.2.1.4 9-Fluorenylmethylchloroformate
Although 9-fluorenylmethylchloroformate is a long established and useful
reagent for determination of amines [6], it remains the subject of on-going study.
The simple question of its reactivity, surprisingly, is still being discussed.
Aymard et al. [37] determined the secondary amines ephedrine (Eph) and
norephedrine (NEph) from human plasma using derivatization with FMOC and
subsequent determination by HPLC with fluorometric detection. They argued
that formation of the FMOC derivatives provided specificity by two mechanisms.
The first was detection at an emission wavelength that has no corresponding

617
signal from endogenous compounds. The second was selective reaction with
primary and secondary amines.
Sparidans et al. [38] demonstrated that FMOC has a more extensive
reactivity towards amines than that proposed by Aymard and co-workers [37].
The former group developed a method for determination of olpadronate, a drug
used to treat osteoporosis. This analyte is a bis-phosphonate that contains a
tertiary amine. Sparidans and co-workers found that FMOC derivatized the
tertiary amine to give the secondary amine-FMOC [38]. This was unexpected
and the authors reported difficulty in obtaining confirmation of the structure by
mass spectrometry. Nevertheless, there was corroborating and recent data in
the literature that tertiary dimethylamines do react with FMOC and that there
is congruent demethylation [39].
As with many other techniques, sample preparation prior to or after derivati-
zation was a factor in optimizing yield of FMOC-derivatives. Sample handling
after derivatization of Eph and Neph requires considerable care [38]. FMOC
derivatives contain an ester functionality and are labile to hydrolysis. Conse-
quently, injection solutions were made in anhydrous acetonitrile, a solvent that
is quite hygroscopic. The requirement for anhydrous acetonitrile precluded
dissolution in the mobile phase and resulted in peak distortion when 150 Al were
injected, thus limiting sensitivity. These observations demonstrated the struc-
ture and stability of a derivative can affect various aspects of the analytical
technique.
Derivatization of olpandronate was sensitive to sample handling at almost
all stages of sample preparation [38]. Extraction of bisphosphonates typically
involves co-precipitation with Ca and subsequent removal of the metal ion
exchange chromatography. In this case cation exchange, but not anion exchange,
resin was compatible with derivatization using FMOC. Moreover, oplandronate
bound irreversibly to the resins. Addition of an analogous bisphosphonate to the
sample inhibited this binding. This may have had direct consequences to the
derivatization. At higher concentration of analyte (or other amines such as the
added carrier) the pH of the reaction and, consequently, the reaction rate
decreased as the reaction progressed and buffer was required to maintain the
pH. Even dissolution of reagent required optimization as did selection of the
buffer for the reaction. Acetone, but not other solvents, dissolved sufficient
reagent to allow a fast reaction.
In their study on determination of amino acids and peptides with FMOC,
Shangguan and co-workers [40] addressed problems resulting from excess
fluorophoric reagent, from additional reagent derived peaks and from sample
preparation. These workers performed the derivatization on a solid support of
alkaline silica gel as the sorbent (herein termed solid phase analytical deriva-
tization or SPAD) and compared the results with those from derivatization in
solution. Solid phase analytical derivatizations substantially reduced formation
of byproducts of FMOC. In addition, excess FMOC, a neutral molecule, was
selectively eluted from the phase with EtOAc. Since FMOC reacts with the

618
amine functionality, the derivatized amino acids and acids retained a free
carboxyl group. As a result during elution of excess reagent and side reaction
products with organic solvent, the FMOC derivatives adhered to the column and
were eluted with more polar solvents that contained methanol and water.

19.2.1.5 Isothiocyanates
Derivatization of amines and amino acids with fluoresceneisothiocyanate isomer
1 (FITC) produces highly fluorescent derivatives. Molina and Silva [41] deter-
mined the phosphorus-containing amino acid herbicides (glufosinate and
glyophosate) as well as aminomethyl phosphonic acid (the major metabolite of
glufosinate) as their FITC derivatives. They identified the optimal pH, FITC
concentrations, reaction temperature and reaction time to provide optimal yield.
The derivatives were separated by MECC and detected by LIF. Excess reagent
was an initial problem. With the first running buffer, derivatized analytes eluted
between 5 and 6 min but the excess FITC eluted as an intense broad peak at 3-5
min. Inclusion of Triton X-100, a non-ionic surfactant, in the running buffer
shifted the retention time of the FITC to less than 3 min but did not affect the
retention times of the analytes. With effective separation of FITC from the
reagent, the linear dynamic range was 2-3000 gg/ml and this was sufficient to
obviate the need for pre-concentration.
Asymmetric dimethyl-L-arginine (ADMA) is an endogenous inhibitor of
nitric oxide synthase (NOS). High concentrations of ADMA would indicate a
reduced formation of nitric oxide, which is a potent vasodilator. Reduced capac-
ity for vasodilation is thrombogenic since blood vessels cannot expand in
response to an occlusion. In principal, elevated ADMA concentrations may be
markers of vascular disease. This is borne out in practice where individuals with
renal failure, with peripheral arterial occlusive disease, or with clinically asymp-
tomatic hypercholesterolemia all exhibit elevated plasma concentrations of
ADMA. Determination of this analogue of arginine is therefore an important
analytical problem. Asymmetric dimethyl-L-arginine and symmetric dimethyl-
L-arginine (SDMA) as well as other methylated arginines were derivatized with
FITC preparatory to CZE with detection by LIF. Causse et al. [42] demonstrated
an effective separation of these methylated analogues of arginine that was com-
plete within 14 min. These basic amino acids also separated from neutral and
acidic amino acids that have more negative charges at the basic pH of the run-
ning buffer. The authors, however, report a slow decomposition of the FITC
derivatives of both ADMA and SDMA to products that elute prior to the parent
compounds but can be accounted for by proper sample storage.
Toyo'oka et al. [43] exploited the reactivity of isothiocyanates and combined
this moiety with one containing a chiral centre and another containing a
fluorophore. The technique was used for the resolution of racemic mixtures of
amines and amino acids with subsequent detection by HPLC/FLU). The chiral
tag R(-)-4-(3-isothiocyanatopyrrolidin-1-yl)-7-(N,N-dimethylaminosulfonyl)-2,
1,3-benzoxadiazole [R(-)-DBD-PyNCS] reacted readily with amines although the

619
yield for the more hindered amines such as those in the P blockers was reduced
by steric hindrance. Despite these difficulties, the method provided useful
information of chiral amines and amino acids. Separation of the enantiomers
was accomplished by simple isocratic elution which decreased analysis time. In
addition, high excitation (x = 460) and emission (em = 550) could be expected
to add to the specificity of the detection.
A reagent consisting of a near infrared dye linked to an isothiocyanate
functionality (NN382, LI-COR Inc) gave ultrasensitive determination for
peptides [44]. The isothiocyanate functionality provided facile derivatization of
peptides. Absorption and emission wavelengths of this dye are 790 and 800 nm,
respectively. At the high wavelengths in the near infrared (670-1000 nm) there
are few endogenous interferences. Moreover, scatter (both Raman and Rayleigh)
is substantially reduced at higher wavelengths. The study by Baars et al. [44]
found that concentrations of reagents, buffers and temperatures were important
determinants of the derivatization yield. Reaction did not occur at room temper-
ature but required 35-45°C. The traces from derivatization at 35°C, however,
were cleaner than those at higher temperature. A surprising finding was that
the ionic strength as defined by [NaCl] was also critical in maximizing yield and
had to be maintained at 2 M.

19.2.1.6 4-Fluoro-7-nitro-2,1,3-benzoxadiazole
Hu and Li 45] discussed 4-fluoro-7-nitro-2,1,3-benzoxadiazole (NBD-F) as a
reagent for derivatization of amino acids. They noted that the reaction was
complete within 1 min at 60 ° and this was considerably faster than the rate noted
for reagents such as FITC. In addition, since these derivatives can be detected
using a less expensive Argon laser there are cost efficiencies. Finally, there were
only two extraneous peaks and these are attributed to hydrolysis products of
NBD-F (i.e., NBD-OH and NBD-NH2).
Workers investigating D and L leucine required and exploited the high
sensitivity obtained with NBD. Determination of naturally occurring D-leucine
is an essential component of elucidating its possible physiological role. This
analyte must be separated from and detected in the presence of L-leucine which
is present in much higher concentrations. Derivatization with NBD-F provided
the requisite sensitivity and Inoue et al. [46] also reported the added benefit of a
very rapid reaction. Linked ODS-(C2 chiral) columns provided the separations.
Sensitivity of 1 fmol (S/N = 5) for leucine was reported and this was sufficient to
detect native D enantiomer in the brain. Kato et al. [47] tested enantiomeric
separations on a monolithic column in which the silica gel had been modified
with chiral phases. Amino acids derivatization with NBD-F were detectable at
very high sensitivity facilitating the use of capillary electro-chromatography.
Reductive amination (discussed below) of the aldehyde on a reducing sugar
with methylamine converts the aldehyde into and N-methylglyeamine [48]. The
secondary amine reacts with NDB-F to produce N-methylglycamines tagged
with NBD. Reaction between the secondary amine and NDB-F was more rapid

620
than for primary amine. This reactivity was attributed to a higher electron
density at the nitrogen. The NBD-tagged N-methylglycamines were separated
by CZE and detected by LIF. Although preparation of the NBD tagged
N-methylglycamine is a two-step reaction, Honda and co-workers [48] stressed
that it is a "one-pot" synthesis that is quantitative and complete within 50 min.
Moreover the derivatization is the last step and occurs in borate buffer which is
suited for the subsequent CZE. Detection by an argon laser-induced fluorescence
yielded an on-capillary sensitivity in the attomole range. This was sufficient to
detect N-glycans in a microgram quantity of a glycoprotein.

19.2.1.7 Other reagents

Biogenic polyamines such as putrescine, cadaverine, spermidine and spermine
are involved in cell-cell signalling and are believed to be markers for the
development of cancers and their regression during drug therapy. Knowledge of
their concentrations and fate has great significance in advancing the under-
standing of this and other neoplastic diseases, in developing methods for early
identification of cancer, and assessing the therapeutic efficacy of treatment.
Molins-Legua and co-workers [49] prepared the DANSYL derivatives of these
amines by SPAD. The analytes were sorbed onto the cartridges and while still on
the solid phase were contacted with a solution of DANSYL chloride that was
passed through the cartridge. The authors tested several solid phases and
reaction conditions. Optimal derivatization on C18 cartridges proceeded at
alkaline pH, ambient temperature and a 30 min reaction time. In comparison,
solution chemistry required 70'C and this put at risk the recovery of thermally
labile derivatives. Use of alkaline pH did not seem to affect the base labile silica
gel that is the basis of the C18 cartridges suggesting that this solid phase may be
more rugged than expected.
The importance of biogenic polyamine analysis attracted the attention of
other investigators. Nohta et al. [50] derivatized the analytes with the fluoro-
phore 4-(1-pyrene)butyric acid N-hydroxysuccinimide ester (PSE). Rather than
measuring the standard fluorescence they measured the excimer fluorescence as
a means to enhance selectivity. These analytes were derivatized with PSE at
primary amines (putrescine and cadaverine) or secondary amines (spermidine
and spermine). Since these polyamines are flexible, the pyrene groups were
sufficiently close to allow excimer fluorescence between 450 and 500 nm. Normal
fluorescence of the pyrene derivatives of monoamines is between 350 and 400
nm. As the biological matrix contains considerably more monoamines than
polyamines, the higher emission wavelength from excimer fluorescence substan-
tially improved selectivity of determination.
Asthana [51] compared methods based on amperometry with those using
derivatization plus CZE. The target analytes were aniline derivatives and the
matrix was environmental water. With the first method the anilines were
separated as cations by free zone electrophoresis at low pH, and detected by
amperometry at a detection limit in the low ,ug/l. The alternative method

621
involved derivatization of the anilines with fluorescamine, the separation of the
derivatives formed by MECC, and fluorescence detection. Detection limits were
comparable (in the order of 1 zg/l). The analytical derivatization/fluorimetric
detection technique, however, gave higher separation efficiency and shorter
analysis times (approximately 4 min) and was reported as being more reliable as
well as having greater ease of handling.
Several groups [52-54] reported on derivatization with 4-(4,5-diphenyl-
1H-imidazol-2-yl)-benzoyl chloride (DIBC) preparatory to HPLC analysis with
fluorescence detection. Al-Dirbashi et al. [52] determined concentration of meth-
amphetamine and amphetamine in human urine. The highly fluorescent deriva-
tives could be detected when the analytes were analyzed from as little as 10 Al of
urine. Such a small volume did not require any work-up other than evaporation.
Moreover, the small sample size facilitated use of miniaturized equipment
suggesting even more extensive applications where small samples sizes are
important. Other reports describe application of this reagent to determination of
phenols (discussed below) [53,54].
Wang et al. reported [55] synthesized N-hydroxysuccinimidyl fluorescein-
O-acetate (SIFA), which was developed for determination of catecholamines by
HPLC with fluorescence detection. SIFA proved to be an amine selective reagent
that reacted with NE, E and DA under mild conditions. The detection limits of
the pure compounds were 3.2, 12, and 56 fmol, respectively, and a determination
of these catecholamines in urine was demonstrated. The issues with this tech-
nique were the high dilution factor and both reagent and matrix interferences.
Only 20 / 1 of 25 ml derivatized solution was injected onto the HPLC column.
Hydrolyzed reagent, derivatized amino acids and aliphatic amines could be
separated from the derivatized catecholamines but under such conditions, the
retention time for derivatized catecholamines was longer than for other meth-
ods. There appears to be considerable scope for enhancing the sensitivity and
speed of analysis for catecholamine determination using this useful reagent.

19.2.2 Thiols

Measurement of thiols, particularly those of biological origins continues to

generate considerable interest and is a worthy challenge to the analytical
chemist. Sulphur containing amino acids and peptides are natural anti-oxidants
and also act as nucleophiles to scavenge reactive alkylating biotransformation
products. For instance, GSH is the nucleophile for detoxification of N-acetyl-
immidoquinone the toxic biotransformation product of acetaminophen or
tylenol [56]. This thiol is also a natural scavenger for ROS and other free radicals
[571. The concentrations of GSH are considered a measure of the protection
available to the cell. Another example of important biogenic thiols is homo-
cysteine. Elevated concentrations of this amino acid play a significant role in
development of heart disease. Elucidating the basic physiology of this amino acid
and clinical interpretation of its concentrations requires sensitive and rapid

622
analytical methods, as reviewed by Ubbink [58]. Analytical derivatizations are
fundamental in almost all methods for measuring the biogenic thiols.
There are several problems present in the determination of thiols. There
must, of course, be the usual derivatization to incorporate a group to enhance
detection. It is also often necessary to block oxidation of thiols (RSH) to the
disulphides (RS-SR). A third and frequent requirement is the reduction of the
RS-SR dimer to the RSH monomer. Although measurement of the disulphides
such as oxidized glutathione (GS-SG) may be important, it is the reduced thiols
that are biologically active. Reduction of the disulphide bond and maintaining
the reduced state during sample preparation is arguably, according to MacCoss
[59], the "greatest problem associated with measuring homocysteine" and the
same can be claimed for the measurement of all biogenic or exogenous thiols
found in plasma.
DeGraff et al. [60] reported a prototypical example of the chemistry involved
in the analyses of thiols. These authors studied diagnosis and management of
cystinosis, a rare metabolic disorder of cystine transport. At issue was the
determination of intracellular cystine concentrations. Inter and intra-acellular
cysteine, however, can be oxidized during sample handling and preparation.
Blocking the thiol of cysteine with excess N-ethylmaleimide (NEM) [60,61]
prevented oxidation of cysteine to cystine during sonication of cells. It also
blocked the disulfide exchange reactions of cystine with available sulfhydryl
moieties such as glutathione and proteins. Once all the possibilities of thiol
oxidation and/or exchange were blocked, cystine was reduced with sodium
borohydride. The resulting cysteine was derivatized with monobromobimane
(discussed below) to produce a highly fluorescent product suitable for HPLC
separation with detection by fluorescence.

19.2.2.1 Reduction of the RS-SR bond

Although borohydride is an effective reducing agent it also reacts with water and
the resulting foaming makes this a less desirable reagent. Several other classes
of compounds have been developed to reduce the RS-SR groups. A number of
authors have relied on phosphines [62-68]. Salazar et al. [62] and Bald et al. [63]
reduced the disulphides with tri-n-butylphosphine. This reagent, however, is
relatively volatile and unstable as well as having a very strong odour. Two other
reagents are potential alternatives. Ivanov and co-workers [64-66] used tri-
phenylphosphine which is less volatile but is relatively water insoluble. In
contrast, tris-2-carboxyphosphine is stable, water soluble and less odoriferous
[67] and may be a useful alternative to the other phosphines.
MacCoss et al. [59] discussed reduction with thiols pointing out that
mercaptoethanol and dithiothreitol hadpKa > 9 and as a result they were ionized
to a very small extent at physiological pH, thus the reduction of the disulphides
was slow and stability in the reduced state tenuous. They found that
N,N'-dimethyl-N,N'-bis(mercaptoacetyl)hydrazine (DMH) with a pKa -7 was
able to hold homocysteine in the thiol form during work-up from plasma.

623
19.2.2.2 Blocking/protectingthe thiol group
Just as there are varied approaches to the problem of disulphide reduction, there
are also a variety of blocking/protecting reagents for the reduced thiols.
Hammermeister [68] took a very classical approach and blocked reduced thiols
from cell extracts by acetylation to prevent oxidation to the RS-SR dimers.
Haj-Yehia et al. [69] took a similar approach and derivatized the thiols of
dihydrolipoc acid (DHLA) with ethylchloroformate. A more standard reagent is
N-ethylmaleimide which has been successfully used by many investigators
[60,61,70-72] offers substantial specificity for reaction at the mercapto group.
Another class of thiol protecting/blocking reagents are based on iodo groups
[9,73]. Iodoacetic blocked the thiol of homocysteine and cysteine leaving the
amine to react with OPA [9]. As a result, the method could be used to determine
the reduced forms of the two thiol amino acids as well as five other, non-sulphur
containing, amino acids such as glutamic acid. Boyd et al. [73] utilized iodo-
acetamide, not to derivatize a thiol amino acid for but to scavenge excess
t-butylthiol subsequent to an OPA derivatization. It is likely that iodoacetamide
would react equally well with the reduced sulphur containing acids. MaCoss et
al. [59] drew on the literature on protein sequencing and selected 4-vinyl-
pyridine (4-VP) as a thiol protecting/blocking reagent and this produced a very
stable product.

19.2.2.3 Fluorophorictagging of thiols

If the sample preparation technique does not involve blockade of the thiol on the
analyte then it is possible to derivatize at that position. This would enhance the
specificity of analysis, particularly of amino acids, as other than cysteine or
homocysteine (and related compounds) amino acids do not contain an SH group.
These analytes would remain invisible to the detectors following derivatization
with reagents specific for thiols. There are several reagents that react with thiols
to provide derivatives that can be detected at very high sensitivity with
specificity being determined by a variety of means.
Halobimanes: Derivatization with monochloro [74] or monobromobimane
[60,64-66] coupled to HPLC/FLU produces a sensitivity of 70 attomole [59].
With narrow-bore HPLC and fluorometric detection it is possible to determine
biogenic thiols with good precision and a wide dynamic range using I l of
physiological fluids and this makes it a valuable technique in biomedical sci-
ences. Monobromobimane (mBrBi) is reactive towards all thiols and can be used
to determine homocysteine, cysteine and reduced glutathione. The general
reactivity produced a facile analytical derivatization of cysteine [60] and homo-
cysteine [58] and resulted in high sensitivity analysis by HPLC/FLU. This
derivatization, however, places a premium on separations. All thiols in the
sample are derivatized and must be separated from each other and from interfer-
ences. Monobromobimane is water labile but can be stored for short periods of
time in a solvent of acetonitrile/water and in the dark [66]. Despite the best
precautions, however, the final reaction mixture contains some fluorescent

624
hydrolysis product of mBrBi and these are potential interferences. If these
caveats are recognized and accounted for, then excellent sensitivities can be
achieved by derivatization with mBrBi.
In contrast, and as would be expected, monochlorobimane is less reactive and
derivatization with this reagent requires enzymatic catalysis [74] with GSH
transferase. While this increases specificity of the derivatization it limits the
application to analytes that are enzymatic substrates. For instance, reaction
between GSH and monochlorobimane can only occur when catalyzed by GSH
transferase but the enzyme will not catalyze derivatization of homocysteine.
This affords considerable specificity and permits determination of intracellular
GSH simply by adding monochlorobimane to cultured cells. The reagent crosses
the membrane and intracellular GSH transferase catalyses the derivatization to
produce a highly fluorescent derivative that allows measurement of intracellular
GSH. Kamencic et al. [74] reviewed this technique and extended the scope of
derivatizations with monochlorobimane. They fortified tissue homogenate with
this GSH transferase and found that it allows a direct measure of GSH in
homogenate as well as in intact cells. It is likely that the enzyme catalyzed
derivatization would also be applicable to the determination of GSH in such
matrices as plasma of cerebrospinal fluid that do not contain this enzyme.
4-Aminosulfonyl-7-fluoro-2,1,3-benzoxadiazole (ABD-F): A highly fluores-
cent derivative is obtained from reaction between thiols and the fluorogenic
reagent [62,75]. Salazar and co-workers [62] reacted ABD-F with homocysteine
and determined this analyte from plasma by HPLC/FLU. They also confirmed
the structure of the derivative with electrospray ionization-mass spectrometry.
In this work, concentrations of homocysteine in plasma were determined by the
HPLC/FLU, by fluorescence polarization immunoassay and by capillary gas
chromatography-mass spectrometry. There was good correlation between the
three techniques providing investigators with options for determining valid
concentrations of this important amino acid.
In keeping with the current requirements of analysis, Kang et al. [75] com-
bined on-column derivatization with ABD-F and separation/detection by
CZE/LIF to develop a fully automated technique for determination of homo-
cysteine. The analyte was directly injected into the capillary in a "sandwich"
between two plugs of reagent solution and derivatized at 50°C prior to separa-
tion. A theoretical examination of reaction rates and mobility under the CZE
conditions permitted calculation of injection rates. The result was a reproducible
on-column derivatization technique that improved the cost-effectiveness of the
determination but increased somewhat the detection limit. The limit of detec-
tion for on-column derivatization with LIF detection was 5.0 nM, as compared to
2.5 nM when a pre-column derivatization was used. The method is a very simple,
fast, and practical approach to fully automated determination of homocysteine
and other thiols contained in low-volume and low-concentration samples.
6-Iodoacetamidofluoresceine: Causse et al. [67,76,77] and Ducros et al. [78]
built upon the known reaction of iodoacetic acid and iodoacetamide with thiols

625
[9,68]. They investigated 6-iodoacetamidofluoresceine (IAF) as a thiol specific
reagent and concluded that, compared to other reagents, it had superior
properties of speed of reaction, selectivity, sensitivity, stability of derivatives and
absence of hydrolysis products. The sensitivity and specificity for determination
of homocysteine following derivatization with IAF enabled comparison between
high performance liquid chromatography/conventional fluorescence detection
CZE/LIF and fluorescence polarization immunoassay. Although all three
technique produced equivalent results CZE/LIF was the most cost-effective [76].
The reports by Ducros and co-workers [78] also highlighted the importance
of matrix effects that must always be considered in analytical derivatization.
Derivatization of GSH, homocysteine, etc. with IAF was dependent on the
method of precipitating the protein in the sample of plasma or serum. A detailed
study demonstrated that protein precipitation with sulfosalicylic acid and
acetonitrile provided high yields of derivative whereas the use of other solvents
or reagents was considerably less efficacious.
Maleimides: There are several reagents that combine maleimide functional-
ity with a fluorescent moiety and these are thiol specific. Ercal and co-workers
[79] studied 9- acetoxy-2-(4-(2,5-dihydro-2,5-dioxo-lH-pyrrol-1-yl)phenyl)-3-
oxo-3H-naphtho [2,1-b]pyran maleimide (ThioGloTM3) as a derivatizing agent.
Derivatization with this reagent was rapid at room temperature and the fluores-
cent ThioGlo-SR adducts were detected fluorimetrically (e = 365 nm, m =
445 nm). This reagent also produced less hydrolysis products than found for
derivatization with a more standard maleimide reagent, N-(1-pyrenyl)-male-
imide. Oxidized glutathione (GSSG) was determined by difference. All free thiol
groups were blocked with 2-vinylpyridine, GSSG was reduced enzymatically
with GSH reductase and the resulting GSH was derivatized with ThioGlo.
Reactions at sulphur in determinationof GSSG and GSH: Cereser et al. [80]
and Lenton et al. [81] recently addressed the difficult problem of determining
both GSH and GSSG. Concentrations of these two compounds reflect the oxida-
tive demand within a cell. Their quantitative determination has considerable
import for physiology, pharmacology and toxicology. Cereser and co-workers
took a more classical approach and used pre-column derivatization of GSH with
OPA [80] but in the absence of thiol reagent. Formally, under these conditions
(the exact mechanism is unknown) the reaction of the nitrogen with OPA can be
followed by a intramolecular reaction between the thiol terminus of GSH with
the isoindole [61] (Fig. 19.5). The resulting tricyclic derivative with an unusual
10 member ring is highly fluorescent. The relationship between yield vs pH was
a curve with a relatively narrow maximum at pH = 10 and indicated the need for

Fig. 19.5. Reaction product for derivatization of GSH with OPA in 1

the absence of thiol.

626
close attention to buffering [80]. The optimization produced a detection limit for
GSH at 50 fmol. It was not possible to directly determine GSSG under these con-
ditions but the disulphide could be measured indirectly by difference. The first
step was the determination of GSH and this was followed by reduction of GSSG
with dithiothreitol. Subtraction of the GSH concentrations before reduction
from those following reduction produced the value of the disulphide.
Post-column derivatization, however, provided a simultaneous determina-
tion of GSH and GSSG via derivatization with OPA. The reaction conditions
were pH = 11.6 and 60°C [81]. Again thiol was omitted to allow derivatization at
both the NH2 and SH functionalities. Lenton et al. [81] demonstrated that GSH
was not oxidized to GSSG during sample preparation and that their direct
measure of GSSG reflected the true concentration of this analyte. It is not clear
whether GSSG forms a OPA derivative or whether the elevated temperature and
pH in the post column reaction produces some other compound which then
reacts with OPA. The hypothesis of reaction at both the NH2 and SH of the GSH
remains unproven. Elucidation of derivative structure may be an important
problem. That intramolecular reaction product exhibits a substantially
increased fluorescence relative to that obtained for the simpler thiols; for
instance, these investigators reported 1000 times greater sensitivity for GSH
and GSSG than for cysteine.

19.2.3 Determination of saccharides via derivatization of the

carbonyl

Determination of sugar monomers, oligomers and polymers by HPLC/FLU is a

fundamental tool in the burgeoning field of glycobiology. As monomers, saccha-
rides are sources of energy. Oligo and polysacharrides (polymers of sugars) play a
yet more extensive role as structural and recognition molecules as well as
markers of disease [82].
Whether as monomers, oligomers or polymers, these compounds possess no
functionality that can be readily recognized by detectors used in chromatogra-
phy. These characteristics are guarantors of a requirement for analytical
derivatizations. Even mass spectrometric analysis of saccharides with a variety
of ionization techniques including MALDI-TOF (discussed below) are improved
by derivatization. Characterization of the complex structures of polysaccharides
requires identification of branching, number of sialic acids etc. as well as the
need for high sensitivity analysis of small amounts of sample.
Formation of fluorophoric derivatives [48,83-88],90-98] has application to
wide ranging investigations in glycobiology. Derivatizations of a carbonyl by an
amine are the predominant technique in these studies.

19.2.3.1 DANSYL hydrazine

Reaction of hydrazines with carbonyls is one of the most frequently used
derivatizations for determination of compounds with this functional group.

627
Volpi et al. [83] derivatized unsaturated disaccharides with DANSYL hydrazine.
The disaccharides were prepared by enzymatic digestion of glycoaminoglycans
and then converted to their DANSYL hydrazones. Following derivatization a
series of seven unsaturated disaccharides with various degrees of sulphation
(0-2 sulphate groups) was separated by single column and an isocratic system.
Detection of less substituted and early eluting analytes was achieved but it
appeared that peaks from reagent or other sources could interfere at lower
concentrations.

19.2.3.2 1,2-Diamino-4,5-methylene-dioxybenzene
Inoue's group investigated derivatized sialic acids with 1,2-diamino-4,5-
methylene-dioxybenzene (DMDB) [84,85]. Unlike most derivatizations of sac-
charides which occur on the carbonyl, the sialic acids were determined as
ketoacids (Fig. 19.6). Derivatization with DMDB produced a sensitivity of 13
fmols of a silaic acid dimer and Sato et al. [84] found that this compared to the
sensitivity obtained by electrochemical detection. Both techniques, electrochem-
ical detection of native analyte and derivatization with DMDB, proved useful in
analysis of oligomers and polymers of sialic acids (Sia chains). Lin and
co-workers argued that derivatization methodology, however, provided higher
sensitivity and, more importantly, was effective in the presence of proteins [85].
This could be expected. The reagent, DMDB, will not react with proteins and
these biopolymers will remain invisible to the detector. At alkaline conditions,

HO

~Rs~~i
j H

N
.
XAOJ

/ .
2
C.
CHCH

OH
HO

R2 =R r

Fig. 19.6. Derivatization of a sialic acid = otier siaic acds

=
as an a ketoacid. er sia acids

628
however, the electrochemical detector would respond quite sensitively to the
sulphydryl and hydroxyl groups on the proteins produce a high background.
Although there are advantages to this derivatization, the acidic reaction
conditions hydrolyzed the Sia chains.

19.2.3.3 Reductive amination

Reductive amination is an essential analytical derivatization and impacts on sep-
arations, detections and structural analysis of the saccharides that are discussed
below. These reactions convert an aldehyde into an amine that either is itself a
fluorophore [86,87] or can be converted to a fluorophore by further reaction [47]
(see Section 19.2.1.5). The derivatives are formed by a sequential reaction
between an aldehyde on a reducing sugar and an amine to formation of a Schiff
Base followed by reduction of that Schiff Base to an aminoglycan (Fig. 19.7). The
reaction mixture for reductive amination consists of the saccharide, the deriva-
tizing amine and reducing agent. Reduction occurs in situ without the need for
isolation of the intermediate Schiff Base. There is debate regarding preferred
techniques for these reductive aminations because of potential unwanted side
reactions.
One potential side reaction arises from the reducing conditions. During
reduction it is possible that the saccharides can be converted to the glucitols
rather than derivatized to fluorophoric products. The consequences of this are
self evident. Reduction to the alcohol removes the functionality that is important
both for biological effect and derivatization to fluorescent compounds.

O
II

JI

H R,

Fig.
19.7.
Reductive
examination.

Fig. 19.7. Reductive amination.

629
 A second side reaction can be generated by the -7.5% acetic acid used in
many of these derivatizations as typified by the conditions used by Charlwood et
al. [86,87]. Acidic conditions can hydrolyze sialic acids residues from the poly-
saccharide [47,84,88,89], and this is a significant matter [82]. Sialic acids are
nine carbon carboxylated sugars located at the terminal end of glycoproteins and
glycolipids [89]. This location enables the biological functions of sialic acids for
cell-cell signalling and adhesion which are fundamental to processes of multi-
cellular life. Their association with polysaccharides is an important measure in
elucidating many biological processes and their loss during derivatization limits
the quality and interpretation of the data produced.
Honda et al. [48] argued that a combination of a monoamine and a dimethyl-
amine-borane complex as reactant and reductant, respectively, is a preferred
condition for reductive amination. Unwanted side reactions are minimized or
eliminated by a reaction temperature at 40°C and maintaining pH = 4.5. In a
technique described in Section 2.1.5 reductive amination with methyl amine and
the dimethylamine-borane complex produced a N-methylglycamine, with
minimal loss of sialic acids and with little or no reduction to glucitols.
A widely used reaction mixture for reductive amination [86,87], includes the
saccharide, amine, sodium cyanoborohydride in a solution of dimethylsulph-
oxide containing a 7.5% acetic acid. The reaction temperature varies from
70-80°C and the reaction time from 1-2 h depending on the temperature. These
conditions appear harsh, but for derivatization of saccharides with 2-amino-
acridone (2-AMAC) or 3-acetylamino-6-aminoacridine (AA-Ac) there are consis-
tent reports giving mass spectral evidence of sialic acid residues retained on the
oligosaccharides prepared for analysis by this technique [90-92].
It should also be recognized that there are examples of saccharide analysis
where such issues are not a concern. The specifics of the analytical problem need
to identified. For instance, in a study on inflammatory bowel disease, Araki et al.
[93] required the determination of dextran sulphate from rat faeces to demon-
strate retention of this polysaccharide in the GI tract. They described reductive
amination of dextran sulphate sodium (DSS) with 2-aminopyridine and sodium
cyanoborohydride. Reaction conditions were stringent requiring 20 h at 90°C but
were adequate for this particularanalyticalproblem.The fluorescent derivatives
were subsequently separated by size exclusion chromatography.

19.2.3.4 Reagents derived from benzoic acids

The fluorescent aminobenzoic acid derivatives, prepared by reductive
amination, also proved useful in studies of mono and oligosaccharides. Anumula
et al. [94] fractionated glycoproteins and then hydrolyzed the membrane blots.
The saccharides were derivatized by reductive amination with 2 aminobenzoic
acids. In this technique hydrolysis with 20% TFA provided the monosaccharides
whereas digestion with peptide N-glycosidase F (PNGase F) released the
N-linked oligosaccharides. The latter technique provided an oligosaccharide

630
map. Methyl-4-aminomethyl benzoate (M4-AB) was used to derivatize a dextran
ladder for a specific and useful application which is discussed in more detail
below [87,92].

19.2.3.5 Aminoacridines
Charlwood and co-workers [86,91,92,95-97], Birrell et al. [87], Hanrahan et al.
[90] and Huterrer et al. [98] of Camilleri's group conducted extensive studies on
quantitative determination of mono, oligo and polysaccharides and also
determined the complex branched structure of the oligomers and polymers.
Derivatives of the aldehydic groups were prepared by reductive amination with
either 2-AMAC or more recently AA-Ac and with sodium cyanobohydride. The
authors reported basic developments including preparation and validation of a
new aminoacridine reagent as well as numerous applications.
In the course of these studies [86,87] the authors addressed the problem of
excess reagent and other impurities. They demonstrated that semi-preparative
column chromatography of the reaction mixture on an OASIS column allowed a
selective elution of derivatives and used this method extensively [91,92,95-97.
Introduction of the 2-AMAC group provided the fluorescence necessary for
high sensitivity detection and also increased lipophilicity as well as including the
aminoacridine group that ionized at acid pH. The latter two properties facili-
tated separations by HPLC or CZE [86,87]. When used in combination with the
appropriate enzymatic hydrolyses, analytical derivatization with 2-AMAC and
chromatographic separation allowed assignment of structures for glycans in
complex mixtures. These derivatives were compatible with CZE/ LIF,
HPLC/FLU, HPLC-MS/MS and MALDI-TOF [86-88,90-93,95-97]. Hutterer
and co-workers [98] further extended the utility of the 2-AMAC derivatives.
They reported that oligosaccharides tagged with 2-AMAC were well suited to the
MECC at 100 kV which considerably improved the resolution over that obtained
at 20 kV - an important finding. Increase in resolution adds to the quality of
data available from the electrophoretic analysis.
More recently Charlwood and co-workers [86] prepared AA-Ac and compared
it to 2-AMAC. They reported a two-fold increase in the detection sensitivity. For-
mation of the AA-Ac as the products were apparently more lipophilic, were
better retained and resolved MECC relative to that found for the 2-AMAC deriv-
atives. As a result, approximately twice as many oligosaccharides were detected
with the AA-Ac derivatives. In addition, the AA-Ac derivatives carry more posi-
tive charge and fluoresce under acidic conditions. With these properties it is fea-
sible to use CZE and MECC. Accordingly the analyst can select the order of
elution based on molecular size of the AA-Ac derivatives. Capillary zone electro-
phoresis elutes the derivatives in order of size (for instance AA-Ac glucose is the
first peak because of its high charge density) and the order is reversed for MECC.
A double derivatization with M4-AB and 2AMAC [87,92] provided an elegant
technique for determination of molecular weights of mono, oligo and polysaccha-
rides. A dextran ladder was derivatized with M4-AB whereas the saccharide

631
were derivatized with 2-AMAC. The derivatized dextran ladder was added then
to the solution containing the 2-AMAC derivatized saccharides. The 4MAB
derivatives were monitored by absorption at 305 nm and 2-AMAC derivatives
were monitored by fluorescence. Fluorescence properties of 4MAB did not
interfere with that of the 2-AMAC and the two classes of compounds could be
detected independently. The derivatized dextran ladder provided an internal
calibration of molecular weight for derivatized oligosaccharides and the
independent detection ensured that there was no cross-interference.

19.2.4 Derivatization of other classes of carbonyls

Visser et al. [99] prepared fluorescent derivatives of carbonyl containing steroids

by derivatization with dansyl hydrazine. Overall recoveries for the resulting
dansylhydrazones were in excess of 90% and the sensitivity was sufficient to
determine 50 ng in 50 Al of plasma. The reaction times, however, were 4-20 h
depending on the steroidal analyte. This possibly reflected the hindered nature
of the carbonyls at C-11 and C-20. Moreover, despite a liquid/liquid extraction
there were interferences from both matrix and excess reagent. Shangguan et al.
[100] also used a hydrazine reagent in a study of ginsonsides. This derivatization
followed ozonolysis of the C24-25 double bond and reduction of the ozonide to
the aldehyde. Although this was an aldehydic position derivatization with
FMOC hydrazine was also slow, requiring 4 h at 40°C.
Acrolein and chloroacetaldehyde are two low molecular weight carbonyls
that are produced during the biotransformation of cyclophosphamide and ifo-
phosphamide. These are two of the premier anticancer alkylating drugs. The
toxicity and side effects of these drugs can be linked to the biotransformation
reactions that generate reactive and toxic intermediates. Determination of
acrolein and chloroacetaldehyde is fundamentally important in elucidating the
toxicity of the therapeutic agents and improving therapy.
Huang and Waxman [101] reacted chloracetaldehyde with adenosine to give
1-N(6)-ethenoadenosine-a highly fluorescent compound (Fig. 19.8). Develop-
ment of the optimal reaction conditions required some effort to balance reaction
time, temperature and decomposition of the product. Derivatization at pH 4.5
and 80°C for 2 h produced the derivative. Since this reaction required a
condensation at the 1 position via displacement chlorine of chloroacetaldehyde,
it could not be used for determination of acrolein.
Paci et al. [102] developed a similar technique for measuring the latter alde-
hyde by derivatization of acrolein with Luminarin 3. This hydrazine based
derivatizing agent produced a fluorescent hydrazone. A possible interference
resulting from the reaction of chloracetaldehyde was not found. The chlorinated
derivative was well separated from the acrolein derivative. Moreover, the fluo-
rescence of condensation product between luminarin 3 and chloroacetaldehyde
was quenched due to the presence of the chlorine on the molecule.

632
 R

NH1
+ CI-CH 2CHO

iR

cl-c~.

CH
2

Xcr
R=sugay residue

Fig. 19.8. Reaction of an adenine derived drug or nucleoside with chloroacetaldehyde. R = Ribose
or deoxyribose.

19.2.5 Fluorogenic reactions between nucleosides and aldehydes

Recognizing the fluorophoric reaction between chloroacetaldehyde and adeno-

sine, several investigators [99-102] "reversed" the derivatization for determin-
ing adenosine derived anti-AIDS medication. Sparidans et al. [103] reported the
determination of 9-(2-phosphomethoxyethyl)adenine, Zhang and co-workers
[104] developed methods for determining 2'-f-fluoro-2',3'-dideoxyadenosine
(F-ddA), and Dai et al. [105] described the measurement of 5'-triphosphate of
2'-j-fluoro-2',3'-dideoxyadenosine. All drugs reacted with chloroacetaldehyde at
the adenosine moiety to form the 1-N(6)-ethenoadenosine. Analysis of human
lymphocytes for determination of 5'-triphosphate of 2'-P-fluoro-2',3'-dideoxy-
adenosine (F-ddA, lodenosine), required degradation of natural nucleic acid
ribosides, such as ATP which caused large interferences in chromatography
[105]. Reaction of ribosides with HIO4 cleaves the vicinal hydroxyls and altered
the structure sufficiently that the products did not interfere with the
determinations.
Guanine nucleotides were converted by reaction with 3,4-dimethoxy-
phenyl)glyoxal to fluorescent derivatives [106]. High performance liquid chro-

633
matography with fluorescence detection provided limit of detection for cGMP at
10 fmol injected on column. Such sensitivity was sufficient to assay guanylate
cyclase (GCase) activity by measuring GTP and cGMP. The two nucleotides are
the substrate and the product of GCase, respectively. Although this method was
less sensitive than generally used radioisotopic methods, it was sufficient to
determine GCase activity in human platelets. Moreover, the HPLC method is
simpler and does not require the use of radio-isotopes.

19.2.6 Derivatization of carboxylic acids

Frey et al. [107] derivatized oxidatively modified phospholipids with chloro-

methylanthracene. They focused on the fragmented acyl chain phosphatidyl-
choline (PC) and found that it contained an oxidized residue, which most likely
had four carbons. Although a classical derivatization procedure, the results were
important for investigators studying the toxic reactions of oxygen. Compared to
age-matched controls, concentrations of fragmented PC were higher in elderly
individuals with coronary heart disease, elevated in smokers and also increased
following reperfusion after treatment with cardiopulmonary bypass. This data
suggests that the oxidatively modified phospholipids may be excellent markers
for oxidative stress.
Finding possible contamination of a marine environment by okadaic acid or
dinophysistoxin, particularly in algae, is indicative that the shellfish from that
source are toxic to humans and to local marine organisms. Confirming the
presence of these diarrheal toxins has deleterious economic consequences for the
economy of a maritime region and so reporting of such a finding must be certain.
The existence of these compounds in a sample can be inferred from bio-assay.
Confirmation, requires, in part, chromatographic analysis. Derivatization of the
acid by anthranyldiazomethane (ADAM) derivatization [108,109] provides the
high sensitivity available from fluorometric detection. In addition, this reagent
reacts almost exclusively with carboxylic acids and so the specificity of analysis is
also high. These characteristics are particularly useful when dealing with toxins.
The amount of material that is to be handled is reduced by the high sensitivity
and selectivity. At the same time the method has the high validity necessary for
confirmation
Hattori et al. [110,111] selected 4-N,N-dimethyl-aminosulfonyl-7-piper
-azino-2,1,3-benzoxadiazole (DBD-PZ) to derivatize the carboxyl group of a
newly discovered natriuretic hormone (2,7,8-trimethyl-2-(j-carboxyethyl)-
6-hydroxy chroman). The reaction required triphenylphosphine and 2,2'-pyrid-
inyl disulphide as condensing agents. Sample preparation involved liquid/liquid
extraction to separate the excess basic reagent from the more neutral product.
The first derivative, however, retained a phenolic group that was susceptible to
oxidation. This was stabilized by O-acetylation. Despite the preliminary
separation, the possible interferences from reagent, the analysis required HPLC
using coupled-phenyl and octadecylsilica columns connected by a six-port valve

634
to allow for switching and washing. While this is an effective methodology, the
run time is long-just under 50 min.
Haj-Yehia et al. [69] determined a-lipoic acid (LA) and DHLA in biological
fluids. After extraction and protection of the thiols in DHLA, the carboxyl
groups were derivatized with 2-(4-aminophenyl)-6- methylbenzothiazole using a
diimide under basic conditions to couple the analyte and reagent. In addition to
providing a determination of LA and DHLA the reagent may prove useful in the
analytical derivatization of other carboxylic acids.

19.2.7 Derivatization of alcohols

Several groups combined derivatization with a fluorophore and detection by

fluorescence. Sun et al. [53] determined bisphenols in extracts from baby bottles
by derivatization with DIBC but used peroxylate chemiluminesce (PO-CL) for
detection. The very large reagent interference was substantially reduced but not
eliminated with a micro-column "clean-up". Nevertheless a sensitivity of 2.8
fmol injected on-column was reported. Wada and co-workers [54] optimized the
derivatization and chromatographic conditions for the determination of phen-
ols. Following derivatization with DIBC, the sensitivity was an order of magni-
tude better than that obtained by HPLC with native fluorescence of the analytes
and was also superior to GC. Finally, Yamada et al. [112] combined DANSYL-
ation of estradiol with detection by chemiluminescene. They tested four oxalates
as reagents for fluorescence and found that bis(2,4-dinitrophenyl)oxalate
produced the desired combination of high and stable chemiluminescence.
Derivatization of an alcohol was the basis for determination of arachadonyl
ethanolamide (anandamide), which is the premiere endogenous cannabinoid.
Yagen and Burstein [113] prepared the dansyl derivative reported detection of as
little as 15 fmol of the derivatized anandamide on a silica gel thin-layer plate
irradiated by long wave ultraviolet light. This enabled a pre-purification for
quantitation of anandamide by HPLC. These authors used detection at 255 nm
and obtained a detection limit in hepatocyte media of 4.3 nmol. It is apparent,
however, that a lower detection limit could be achieved with fluorimetric
detection. Arai et al. [114] derivatized the primary hydroxyl of anandamide with
DBD-COCl to provide a fluorophoric derivative. Pre-chromatographic separa-
tion of excess reagent from derivatized analyte was affected by a combination of
post-derivatization solid-phase extraction (SPE). Although pre-chromato-
graphic separation was used, the analytical separation required linked column
(phenyl-ODS) methodology to reduce interferences. Sensitivity for the entire
technique was 37 pmol/G of tissue.
Assaf et al. [115] described highly fluorescent esters of alcohols obtained by
derivatization with (4-carboxyphenyl)-6-N,N-diethylaminobenzofuran. The
esterification at 60°C required a coupling technique with a diimide and using
4-dimethylaminopyridine as a basic catalyst. With hX 0 = 387 and kem = 537 nm)
the detection limit ranged from 0.2-0.5 pg of the derivatized alcohol at a S/N = 3.

635
19.2.8 Photolytic activation
Photochemical reactions produce highly fluorescent compounds and this can be
used to advantage most often with post-column derivatization [116-118].
Sharma [116] studied the formation of adducts between tamoxifen, an anti-
cancer drug, and DNA. This work demonstrated that during biotransformation,
tamoxifen (or its 4-hydroxy biotransformation product), is activated and binds
covalently to DNA. Post-column photochemical derivatization cyclized the
4-hydroxytamoxifen-DNA adduct to a phenanthrene and production of this fluo-
rescent compound increased sensitivity by two orders of magnitude. Adduct for-
mation between an oxidatively activated compound and DNA is often mutagenic
in its own right. It is therefore possible that treatment of cancer with tamoxifen
may also induce cancer at a later stage. A similar photochemical induced
cyclization was used to determine the cycloxyenase II inhibitor 5-chloro-3-(4-
methanesulfonylphenyl)-6'-methyl-[2,3']bipyridinyl. This analyte rapidly
formed fluorescent products when exposed to UV light at 254 nm in a post col-
umn reactor [117]. These compounds were detected by excitation at 260 nm and
emission 375 nm. At 5 ng/ml the response was well above the base-line noise and
the reproducibility was excellent.
On-line post-column irradiation at = 254 nm produced photo-induced
modifications of equilin and equilenin to fluorescent derivatives [118]. Limits of
detection were 10 fmol for equilenin ( = 280 nm; Kem = 410 nm) and 50 fmol
for equilin (x = 280 nm; Xk, = 312 nm). Differences in kex and hXm indicated that
irradiation at 7 = 254 did not simply oxidize the dihydronaphthalene of the A-B
ring in equilin to the naphthalene of the A-B ring in equilenin.
Determination of melatonin in rat and mouse pineal gland was by HPLC
with fluorescence detection [119]. Melatonin was derivatized with hydrogen
peroxide in carbonate and produced a fluorophore detected at the emission
wavelength of 380 nm with the excitation at 245 nm. The structure of the
derivative was not defined but may be N-acetyl-N-formyl-5-methoxy-ky-
nuramine or some analogue of it.

19.2.9 Metabolic ytometry by labelling of saccharides with

rhodamine

Krylov and co-workers [1201 developed metabolic cytometry as a sub-class of

chemical cytometry. The former is defined as "analysis of metabolic conversions
in individual cells". Analytical derivatizations again were crucial in evolution of
this field. Both chemical [121,122] and metabolic derivatizations were reported.
For chemical derivatization reducing sugars were converted to amino alditols
and reacted with 5-carboxytetramethylrhodamine succinimidyl ester
(5-CTRSE) [121]. Reductive amination with benzylamine/sodium borohydride
produced the benzylamines of simple sugars such as glucose or galactose. Subse-
quent hydrogenolysis provided a 1 amino-l-deoxy alditol. Aminosugars were
simply reduced with sodium borohydride to provide the 2 amino 2-deoxyalditol.

636
 pGal(l -- 4)pGIcNAC-O-TMR
a fuc(l -- 3)
z(1-03)FTase t [LeX]

pGal(l - 4)pGIcNAC-O-TMR
[LacNac--TMR]
tr(-2)FTase

pGal(l - 4)PGlcNAC-O-TMR
a fuc(1-02) Il +3)Frase

Gal(1 - 4)PGIcNAC-O-TMR
a fiic(-2) a lc(1 -3)
[LeY]

Fig. 19.9. Biosynthetic pathways for LacNac-O-TMR to for LeX and LeY.

The amine functionality then provided the reactive moiety for derivatization by
5-CTRSE. Detection limits for the tetramethylrhodamine derivatives were on
the order of 60 molecules injected on-capillary. This derivatization was not
applicable to the study of biological problems because of the lengthy deriva-
tization procedure and high concentrations of analyte required to compensate
for hydrolysis or reagent. Nevertheless it was a powerful proof-of-principal.
Following demonstration of this very low detection limit, a biochemical
derivatization was developed [122]. A methoxycarbonyloctyl group was attached
to the oligosaccharide gal (1->4)PGlcNac (N-acetyl-lactosamine abbreviated
LacNac) and this product was reacted with ethylenediamine to form Pgal(1-4)
PGlcNac-O(CH 2 ),CONH(CH 2)2 NH2 . The amine group was the site of reaction
with 5-CTRSE to form the derivative gal(1-*4)jGlcNacO (CH2 )sCONH(CH
2) 2NHTMR (LacNac-O-TMR). Upon incubation, cells took up this reagent which
then became a substrate enzymatic process within the cell. Both degradation
and biosynthetic pathways were observed [33] and are summarized in Fig. 19.9.
In the degradation pathways galactosidase cleaved LacNacTMR to GlcNacTMR
and subsequent reaction with hexosaminidase cleaved the sugar and released
the HO(CH2)8CON(HCH2)2NHTMR moiety. In the presence of GDP-fucose
(GDPF), fucosyl transferases (FTase) incorporated the LacNac-TMR into more
complex oligosaccharides labelled LeX and LeY as shown in Fig. 19.9. Three
types of activities were defined. In the G1 phase of the cell cycle neither LeX nor
LeY was produced as would be expected in the resting phase. In the G2/M phase
some cells remained inactive (i.e., no degradation or biosynthesis) or produced
either LeX or LeY. No cell was found to produce both LeX and LeY. Thus the cell
cycle of individual cells can be probed by chemical cytometry.

637
19.3 ABSORPTION OF ULTRAVIOLET AND VISIBLE LIGHT

Detection in the ultraviolet and visible (UV/VIS) wavelengths does not, as a rule,
provide the high sensitivities typical of many modern methods. Nevertheless the
techniques are well understood, the equipment is rugged and there are many
applications that can benefit from use of such reliable methods. This was
exemplified by continuing reports [123-141] describing methods based on
derivatization followed by detection by absorption in the UV/VIS region.

19.3.1 Amines

19.3.1.1 Isothiocyanate
Miller et al. [123] recognized that study of complex biochemical problems
requires rapid and simple methods as well as rugged, relatively inexpensive
equipment. They derivatized isopeptides with phenylisothiocyanate (PITC) and
applied this technique in studies on cross-linking of proteins during clotting.
Their's was a simplified method for determining the frequency of cross-linking
as it eliminated complex multicolumn systems or the more expensive fluori-
metric detectors used in other methodologies.
New information continues to refine derivatizations with PITC; for instance,
Albin et al. [124] recently reported that standard hydrolysis conditions (6 M HCl
at 110°C for 24 h) were not optimal. They suggested a sequential hydrolysis
technique to optimize yield for derivatization of amino acids by PITC. Zahou and
co-workers [125] used derivatization with PITC but relied on the high extinction
coefficient to detect at 200 nm to provide the high sensitivity. CZE separated the
18 amino acids produced by the hydrolysis of peptides with 10-600 residues.
Sensitivities were in the 50-60 fmol range or three orders of magnitude better
than the ninhydrin technique used in routine analysis of amino acids.

19.3.1.2 Nitroaromatics
Nitroaromatic reagents continue to find extensive application in the determina-
tion of carbonyls, amines, alcohols and acids. The nitro groups impart high
molar extinction coefficients thus increasing the sensitivity. Nitroaromatic
moieties are attached to hydrazines or acyl chlorides to provide reactivity. In the
case of 1-fluoro-2,4-dinitrobenzene (FDNB), the nitro groups also exert a power-
ful electron withdrawing effect and in conjunction with the electron withdraw-
ing effect of F produce an electrophilic centre on the aromatic ring. Nucleophilic
attack and displacement of the fluorine by amines tags the analyte with
2,4-dinitrobenzene (DNB).
The neuro-excitatory non-proteogenic amino acid 3-N-oxalyl-2,3-diamino-
propionic acid (-ODAP) and other such non-proteogenic amino acids were
determined as their DNB derivatives. These amino acids are potentially toxic to
humans and animals. Under certain environmental conditions these compounds
can be present in some food crops such as grasspea (Lathyrussativus). Measure-

638
ment of ODAP is useful in assessing potential toxicity. Extracts of grasspea were
treated with FDNB and ODAP as well as other amino acids were derivatized at
the amine [126]. The derivatives could be detected in the pmol range. The
resulting method is an important addition to the options available to the food
chemist.
Determination of gentamicin, an aminoglycoside antibiotic, presents an
analytical problem that is somewhat more complex than usual. First, the drug
itself is a mixture of gentamicins C(1), C(la), and C(2) with other gentamicins
also possibly being present in the pharmaceutical preparations. Second, pharma-
cokinetic and other pharmacological or toxicological studies require that all
three compounds be measured. Third, the drug has no functionalities that can be
used for detection. Isoherranen and Soback [127] determined the gentamicins in
plasma or urine via a SPAD technique. They extracted the gentamicins from the
bio-fluids by SPE and the sorbed gentamicins were contacted with a solution of
1-fluoro-2, 4-dinitrobenzene directly in the extraction cartridge itself. This
considerably simplified the manipulations required in sample preparation.
Studies on diffusion of gentamicin through agar matrix [128] presented less
severe requirements than pharmacokinetics but the selection of the preferred
method is not completely resolved. In developing this technique, the authors
compared derivatization with FDNB and OPA as well as stability of the resulting
derivatives. The former reagent provided a more facile reaction and more stable
reaction products than those achieved by the OPA reaction. On the other hand
Kaale et al. [129], using a central composite experimental design, optimized an
NDA/CN derivatization and followed detection by absorption at Xm = 330 nm.
The fully optimized capillary zone electrophoresis provided a baseline separa-
tion of gentamicin C1, Cla, C2, C2a, C2b, sisomicin and several minor
components.
Biogenic amines are readily derivatized with 3,5-dinitrobenzoyl chloride
(DNBC) and this reaction was developed to determine such analytes in food
[130]. This reaction has a distinct advantage of rapidity, being complete within 3
min in a solution of 1 M NaOH, acetonitrile and 2-propanol. Kirschbaum et al.
[130] confirmed structures of the derivatives by HPLC-MS with electrospray
ionization. Rapid derivatization of amines with this reagent was also reported by
Herraez-Hernandez et al. [131]. These authors compared sensitivities for
derivatization by DNBC with those obtained from FMOC and recommended the
latter reagent for determination of amines.

19.3.2 Carbonyls

Derivatization of carbonyls with 2,4-dinitrophenylhydrazine (2,4-DNPH) is a

classical reaction that remains the basis of recent work for monitoring environ-
mental toxicity as well as in pharmacological studies. Grosjean et al. [132]
derivatized carbonyls with DNPH and provided a detailed description of
structure and properties of the 2,4-dinitrophenylhydrozone (2,4-DNPHone)

639
derivatives of numerous carbonyls found in car exhaust. They reported the
concentrations and speciation of carbonyls collected in Tuscarora Mountain
tunnel as an example of the diversity of these environmental contaminants [132]
produced by heavy diesel trucks. Despite its extensive use over time, application
of 2,4-DNPH to a variety of problems requires careful study and attention to the
needs of the specific problem and the chemistry of the analytes.
Innovation in this field remains essential-even when the analytical
problem consists of a frequently measured analyte such as formaldehyde and a
long used reagent such as 2,4-DNPH. Determination of formaldehyde, a potent
toxin, is required for quality control in the manufacture of water soluble poly-
mers used in waste water treatment. In this case, however, the standard pre-
column derivatization with 2,4-DNPH produces unreliable measures of
formaldehyde as the polymers themselves react with this reagent. Michels [133]
circumvented the problem by using post-column derivatization. This allowed a
separation of the polymer from the analyte prior to formation of the chromo-
phoric derivative.
Despite the long experience with this derivatization there are still potential
problems with specific analytes. Dusi and Gamba [134] studied determination of
polyether antibiotics (PEs) lasalocid, monensin, salinomycin and narasin that
are used in poultry feeds. Under the acidic conditions of the reaction, these
polyethers are hydrolyzed to carbonyls that react with this reagent. While the
yield of the derivatization of the other antibiotics was acceptable, that for
lasalocid was low. This was attributed to steric hindrance. Lasalocid has one
carbonyl with two highly substituted a carbons. Monensin, in contrast, has a
ketal and hemiketal which, at acidicity of the reaction, are converted to ketones
but the a positions are either methylenes or hydroxy methyl groups.
Reaction of the hydrazone group via intramolecular condensation [135] was
recently described. Such reactions occur when there are additional carbonyls or
conjugated unsaturation in an analyte. Following derivatization of 3-butyl-2-one
with the 2,4-DNPH the expected product butynone DNP-hydrazone was formed.
A subsequent rearrangement (as in Fig. 19.3B) led to 3-methyl-1-(2',4'-dinitro-
phenyl)pyrazol (DNPP). This did not interfere with determination of the
analytes but co-eluted near the 2,4-DNPone of formaldehyde. Similar reaction
products were observed for analytical reactions between other hydrazines and
polyfunctional carbonyls.

19.3.3 Carboxylic acids

The usual analytes selected for derivatization with hydrazines contain a car-
bonyl functionality. Nevertheless, Miwa [136] successfully used 2-nitrophenyl-
hydrazine hydrochloride to form the 2-nitrophenylhydrazides of low molecular
weight mono and di-carboxylic acids. The derivatization was catalyzed by
diimide to condense the reagent with the carboxylic acid and required heating to
60°C. A subsequent treatment with strong alkali, also carried out at 60°C, was

640
required to remove excess diimide and other interferences. Although the
reaction was somewhat involved and lengthy, the resulting derivatives have
excellent chromatographic and detection properties.

19.3.4 Reactions at sulphur

Glutathione was determined by micellar electrokinetic chromatography and

using in-capillary derivatization with 2,2'-dipyridyldisulfide [137]. One molecule
of GSH reacts with one molecule of 2,2'-dipyridyldisulfide to give the mixed
disulphide and 2-thiopyridone. Determination of the 2-thiopyridone provides an
indirect measure of GSH. The in-capillary reaction exploited the differences in
the mobilities of the analyte and reagent which provided the mixing required for
the reaction. The 2-thiopyridone readily separated from the other analytes and
spectrophotometric determination at 343 nm further enhanced the specificity of
the analysis.
2-Chloro-l-methylquinolinium tetrafluoroborate (CMQTFB) is a thiol
specific reagent that reacts with GSH in aqueous medium at pH > 8.2. Bald et al.
[63] optimized reaction conditions and showed the derivatives to be stable for
more than 24 h. Glowaki and co-workers [138] investigated determination of
2-mercaptoethanesulphonic acid (MESNA) via derivatization with this reagent.
MESNA is an important adjunct to cancer chemotherapy with alkylating drugs
as it ameliorates the toxicity of these agents. Its measurement is essential in the
studies on the pharmacology of this drug. Glowaki et al. [138] demonstrated that
CMQTFB absorbs at 320 nm whereas the maximal absorption for the derivative
is at 350 nm, thus reducing reagent interference. This provides pmol detection
and quantitation limits. Mesna is a protective agent against reactive chemicals
that is widely used to ameliorate the toxic effects of alkylating anti-cancer drugs.

19.3.5 Alkylating anticancer drugs

Busulfan [139] is an antineoplastic drug that acts by alkylating DNA via the two
sulphate leaving groups. The drug itself is a neutral molecule and so was readily
isolated from plasma by liquid/liquid extraction. In alkaline medium one
molecule of busulfan reacted with 2 molecules of 2,3,5,6-tetrafluorophenol
(TFTP) to form di-TFTP-butane. This derivative was readily analyzed by HPLC
and at 275 nm the detection limit was 50 ng/ml. Results from analysis of
busulfan by this HPLC/UV technique correlated well with those obtained by
mass spectrometry. These data indicate a high specificity with the chromophoric
detection method. This analytical derivatization reagent and technique can
likely be applied to other alkylating drugs used in cancer chemotherapy. In
addition a thiophenol with a higher molar absorptivity would undoubtedly
increase the sensitivity of the technique.
Diethyldithiocarbamate (DDTC) also proved a useful nucleophile for deter-
mining busulfan [140]. A facilitating feature of derivatization with this reagent

641
is that it is compatible with plasma and can be carried out directly in that
biofluid. The subsequent sample preparation, however, was somewhat extens-
ive. It required extraction with ethyl acetate followed by further separation of
the analyte by chromatography on C8 columns. Despite the complexity of the
method, recoveries were high (99%) and the overall method had a sensitivity of
25 ng/ml. More importantly, it only required 200 Al of plasma. This is a particu-
larly important characteristic when studying pediatric patients. Their low body
weight combined with the illness and cancer chemotherapy, make children a
very difficult population to study: they can be very anemic and unable to give
large volumes of blood. This sensitivity was sufficient to determine plasma
concentrations after oral administration and to perform both therapeutic drug
monitoring and studies of the pharmacokinetics of this drug in children.
Another pairing of a thiol and an antineoplastic drug produced a successful
method. Cisplatin has no functional group that can be used to detect low levels of
cisplatin by HPLC. This is a standard feature of analytes that require derivatiza-
tion. The required product formed when DDTC reacted with the platinum in
cisplatin to yield a platinum-DDTC (Pt-DDTC) complex [141]. This derivative
has a high molar-extinction coefficient and can thus be detected at high sensitiv-
ity after chromatographic separation. Raghavan et al. [141] indicated that
DDTC has a use beyond analysis. They suggested that the reagent can be used to
destroy cisplatin by converting it to platinum-DDTC. This can be used to
inactivate pharmacy counter-tops or any other area where the drug may be on
surfaces or in solution.

19.4 ELECTROCHEMICAL DETECTION

Boyd et al. [73] combined the hindered t-butylthiol with OPA to derivatize and
determine amino acids in microdialysate. The t-butylthio derivatives were more
stable and more lipophilic, the latter property enhancing preconcentration on a
capillary column. Electrochemical detection of the 2-t-butylthio isoindole deriva-
tives provided the requisite sensitivity. A critical aspect of this work, however,
was removal of the excess thiol by scavenging with iodoacetamide and excess
OPA with Na2SO3 . The latter reaction produced OPA-based N-alkyl-1-isoindole-
sulfonate which eluted in the solvent front. The double scavenging reduced the
background by 90% and substantially increased the sensitivity. High sensitivity
and automated throughput allowed sampling with a 10 s temporal resolution. As
many neural processes occur on time scales of a second or less, the temporal
resolution is a crucial feature of the methodology.
Wang et al. [142] developed an integrated miniaturized analytical device
which also exploited the OPA/2ME reaction as well as the electroactivity of the
final derivative. Micromachining technology and a miniaturized electrochemical
detector provided the instrumentation that allowed derivatization, separation
and detection on a glass micro chip. The authors reported on the influence of a
number of variables on the analytical process. Relationships between yield and

642
reagent concentrations showed a fairly sharp maximum indicating that small
error in these variables can have a large effect on recoveries of the derivatives.
The sample/reagent (S/R) ratio had a more "forgiving" profile in that the
maximum yield was obtained over a wide range of S/R values. As with all current
analytical methods speed was an important consideration. A step of the driving
voltage reduced migration times for late eluting components. The entire process
is easily automated and integrated into high throughput techniques.
Tcherkas and co-workers reported on the use of sulphite as a nucleophilic
reagent for OPA derivatization preparatory to HPLC with electrochemical
detection [143]. This is an interesting innovation and is analogous to the
scavenging reaction described by Boyd et al. [73]. At present, however, it is
limited to the more lipophilic amino acids. The more polar analytes in this group
are eluted in the solvent front when derivatized with OPA/sulphite.

19.5 DERIVATIZATION HYDROXYLS, AMINES, CARBOXYLIC ACIDS

AND CARBONYLS FOR GAS CHROMATOGRAPHIC ANALYSIS

Many of the following reagents affect several functionality groups. Silylation can
occur on carboxyl hydroxyl and amine functionalities. Similarly, alkylation can
form esters, ethers and enol ethers from carboxyls, hydroxyls and carbonyls.
Indeed, there are instances where these different groups are on the same mole-
cule. This section is therefore organized according to derivatization reaction.

19.5.1 Silylation

Silylation was one of the first analytical derivatizations [144]. The enhanced
volatilization resulting from silylation of compounds with amino, hydroxyl and
carboxyl functionalities was fundamentally important in both qualitative and
quantitative gas chromatographic analysis of numerous and structurally diverse
organic compounds. Little [145] reviewed artifact formation in the course of
silylation but this authoritative report also summarized reagents, functional
groups and reactions involved in this derivatization technique. Despite the
decades-long use of this derivatization, it remains fundamental to many current
techniques.
Trimethyliodosilane (TMIS) and other strong silyating methods have seen
considerable use. Thevis and co-workers [146] demonstrated selective trimethyl-
silylations of steroidal glucuronides. Derivatization with perdeutero N, O-bis (tri-
methylsilyl) acetamide (BSA) gave the perdeuterotrimethylsilyl ethers of
hydroxyl groups but left the keto groups free. The former were predominantly of
the glucuronyl moiety, whereas the carbonyls were found only on the steroid.
Subsequent reaction with the highly reactive TMIS enolized and trimethyl-
silylated the keto groups. This sequential derivatization tagged hydroxyl and
ketone groups with different isotopes yet finally produced a pertrimethysilylated
derivative which provided valuable structural information.

643
 Other applications of TMIS have also appeared recently. Shinohara et al.
[147] addressed derivatization of a sterically hindered steroidal alcohol. Regula-
tory requirements for the International Olympic Committee (IOC) were the
driving force for the study and so the focus was detection in urine. These
investigators developed a method for determining the two urinary methyl-
testosterone metabolites: 17a-methyl 5a-androstan-3a, 17p-diol and 17a-methyl
5a-androstan-3(,17-diol. Again, both hydroxyl groups in these molecules are
hindered: the 3a hydroxyl is an axial configuration and the 17P hydroxyl is a
tertiary position and is also a to the 18 methyl group. Reaction with TMIS,
trimethylsilyated these highly hindered hydroxyls and provided the necessary
gas chromatographic methodology.
The IOC also generated a study on determination of plasma expanders such
as hydroxyethyl starch (HES). Plasma expanders are usually administered in
cases of hypovolaemic shock. In normal individuals, however, they increase
plasma volumes and dilute concentrations of drugs in blood. The IOC banned the
use of HES from January 2000 and required methods for the determination of
this compound in biological fluids. Thevis and co-workers [148] hydrolyzed HES
which produced ethylated monosaccharides. Reaction with TMIS produced the
pertrimethylsilyl derivatives of these monosaccharides and they were
determined by GC.
These reports demonstrated the utility of TMIS. This reagent, however, is
both highly reactive and unstable. It must be prepared with care and the
reaction is usually performed in situ by a combination of N-methyl-N-tri-
methylsilyltrifluoroacetamide, ammonium iodide and ethanethiol accompanied
by heating to 60°C.
The problem of silylating a hindered hydroxyl groups was also encountered
in the determination of gentrisone (13P ethyl-17a-ethynyl-17-hydroxygona-
4,9,11-trien-3-one), an oral contraceptive possessing androgenic as well as anti-
estrogenic and anti-progestagenic properties. These properties make it a
possible doping agent for the high performance athlete and this again generated
an analytical problem. The hydroxyl group is very hindered and the triene
structure coupled with the 3 keto makes this molecule difficult to derivatize for
confirmation of doping by GC-MS. Kim et al. [149] showed that trimethyl-
silylimidazole saturated with sodium acetate coupled with heat will give both
17p-trimethylsilyl (TMS) and 3 TMS-enol. The latter reaction produces three
tautomeric forms with extended conjugation. These can be separated chromato-
graphically and have quite distinct mass spectra. Although formation of the
three tautomers reduces sensitivity, it can add substantially to the specificity of
analysis by providing distinct patterns of peaks and three unique spectra from
one compound.
Ohie [150] et al. compared the efficacy of t-butyldimethylsilyl (TBDMS) and
TMS derivatives for gas chromatographic analysis amino acids, their metabo-
lites and hydroxy fatty acids. Trimethylsilyl derivatives fragment via the loss of a
TMS group (m/z = 73) accompanied by extensive fragmentation. t-Butyl-

644
dimethylsilyl ethers and esters fragment via the loss of the t-butyl group (m/z =
57). The resulting intense ion at M-57 retains most of the structure of the mole-
cule. Accordingly, the signal is highly specific as it is attributable to an ion that
truly represents the structure of the analyte. Selected ion monitoring at higher
masses provided greater sensitivity for most of the compounds studied and in
some instances detection limits were five-fold lower for the TBDMS compounds.
The coefficient of variation for the TBDMS derivatives was approximately half
that of analyses using the TMS ethers/esters. The superior analytical perform-
ance of TBDMS products, however, translated into only a moderate improve-
ment in identifying genetic diseases. Analyses based on the TBDMS derivatives
gave the correct diagnosis in 52 of 53 cases, whereas those based on MS
derivatives gave a correct diagnosis in 49 of the 53 cases.
One frequent problem with all silylations is that reagents and derivatives are
variously labile to water. Mol et al. [151] address this issue in an extensive study
of analysis of phenol endocrine disruptors. They optimized derivatization of
these organic acids with several silylating reagents including N-methyl-N-
(t-butyldimethylsilyl) trifluoroacetamide (MSTFA). A variety of temperatures,
reaction times and solvents were tested to determine optimal conditions. In the
course of this study they found derivatization with MSTFA but not other
silylating reagents can tolerate up to 0.5% water in the silylating solution.
Compatibility of this reagent with a small amount of water was a particularly
important consideration for one of the most complex analytical problems under
investigation. This is the determination of amino acids in the Martian terrain
using GC-MS aboard the spacecraft [152]. That planetary atmosphere contains
some water but is below 0.5%. Rodier et al. [152] found derivatization with
MSTFA was simple, compatible with some water contamination and readily
automated. The investigators, however, pointed out that the TBDMS adds to the
molecular weight of the reaction products and puts the molecular ion of the
larger amino acid derivatives above the mass range of the instrument on the
spacecraft. The structure of these will have to be inferred from the gas chromato-
graphic properties alone.
Techniques for reducing sample handling and reducing solvent use during
sample preparation derivatization attract interest. Staerk and Kulpmann [153]
combined trimethylsilylation with head-space sampling at elevated tempera-
tures using solid-phase microextraction (SPME). In this technique, drugs (e.g.
amphetamines, cocaine, benzodiazepines, barbiturate) were isolated from aque-
ous biological sample by liquid/liquid extraction using a small volume. The
extract was evaporated to dryness in a head-space vial. The vial was sealed and
depending on the analytes two alternative procedures were used. For determina-
tion of opiates or cocaine the analytes were silylated in the vial using MSFTA. An
SPME syringe was inserted and the temperature increased to volatilize the
silylated derivative which were then sorbed onto the fibre for solventless injec-
tion. Alternatively, the fibre was doped with acetic anhydride/pyridine and
inserted into the vial. Upon heating, the analytes were volatalized, sorbed onto

645
the fibre and derivatized in situ (the on-fibre derivatization technique is dis-
cussed further below). Following derivatization, the fibre was inserted into the
injector port and the derivatives desorbed for gas chromatographic separation.
Miniaturization of classical derivatization techniques facilitate analysis of
small samples. Harrison and co-workers [154] developed and applied such
techniques to the determination of oxidized lipids. The analytes were deposited
in capillary tubes which were placed in an all-glass apparatus containing a
receptacle for reagents: either N,O-bis(trimethylsilyl)trifluoroacetamide
(BSTFA) or methoxylamine hydrochloride. Upon heating, reagent gas (BSTFA
or methoxylamine) was released and the vapours derivatized the analytes.
Excess reagent was then removed under vacuum. Silylated analytes were
prepared for electrospray ionization (ESI) analysis by dissolution in aqueous
ammonium acetate-methanol. This treatment hydrolyzed the TMS esters but
left TMS ethers of alcohols intact. Increases in masses of 72 for trimethyl-
silylation and 29 for oximation were used to identify hydroxyl and carbonyl
groups. Ozonolysis and subsequent derivatization readily identified the number
and position of double bonds. Ozonolysis was performed simply by exposing the
sample in a capillary to ozone. Further oxidation of the ozonides with HO 2 2 and
formic acid, however, was likewise performed by heating the reagents in the
sealed container containing the capillaries with the analytes.
Anari et al. [155] combined t-butylmethylsilylation with pentafluorobenzyl-
ation and NICI to determine valproic acid and its products from enzymatic
oxidation. As is found for many PFB esters, NICI of the mixed TBDMS/PFB
derivatives exhibited an uninformative base peak at m/z = 181 corresponding to
the PFB group. The fragmentation of the TBDMS/PFB derivatives, however,
was not as extensive as that found for the TMS/PFB products. The mixed
TBDMS/PFB derivatives of the hydroxylated products exhibited reasonably
intense peaks corresponding to loss of the tbutyl and TBDMS groups and
provided the necessary structural information. The resulting method produced
sensitivities of 2 ng/ml with excellent precision. This was sufficiently sensitive
for studies with the small amounts of cDNA-expressed human cytochrome that
are available to investigators. In brief, a judicious selection and combination of
derivatization pathways and ionization techniques provided a highly sensitive
and selective method.
Yang et al. [156] proposed pentafluorophenyldimethylsilyl (flophemesyl)
chloride as a potential alternative to PFBBr which is quite noxious. As a
silylation, the reaction does not require catalysis and derivatization of saturated
fatty acids is complete at room temperature within 15 min. The flophemesyl
derivatives have excellent chromatographic properties. Electron impact spectra
show reasonably intense M-15 peaks that provide specificity and high sensitiv-
ity. The base peak for all derivatized analytes, however, was at m/z = 121
resulting from a loss of 2 fluorines from the fluorohydrocarbon tropylium ion at
m/z = 159. Flophemsyl was also used to derivatize and determine unsaturated
fatty acids [157]. Although the chromatographic properties were again excellent,

646
in the case of unsaturated fatty acids fragmentation was controlled by the double
bond position and there were relatively fewer ions of higher mass.

19.5.2 Alkylation

Two groups determined polar, hydrophilic acids by derivatization in the aqueous

matrix by using high concentrations of sulphuric acids. Formic acid is the
terminal biotransformation product of methanol and its presence in biological
fluid is evidence for ingestion of this toxic alcohol. A combination of concen-
trated sulphuric acid and methanol added to whole blood or urine produced
methyl formate [158]. This volatile ester was then isolated by sorption onto an
SPME carboxen/polydimethylsiloxane fibre placed in the head-space above the
reaction mixture. Despite the low recoveries, precision was high and the tech-
nique proved sufficient for determination of formic acid in biological fluids. A
similar combination but with ethanol as the esterifying agent [159] yielded the
ethyl esters of haloacetic acids found in natural waters. Following deriva-
tization, ethyl esters were sorbed onto a polydimethyl-siloxane SPME fibre in
the head-space by saturating the aqueous sample with sodium sulphate and
heating. Reaction conditions described in these two reports are severe. The
work, however, demonstrates that if a compound is stable a simple deriva-
tization procedure can be carried out directly in natural waters, urine or plasma
and sampling by head-space SPME protects the instruments.
More recently, milder conditions were developed to derivatize the haloacetic
acids. Sarrion et al. reported [160] that dimethyl sulphate or diethyl sulphate
under catalysis with tetrabutylammonium sulphate esterify these analytes at
55°C. Again, the use of head-space SPME protected the instrumentation from
overloading with the large excess of reagent and/or catalyst as well as providing a
facile isolation and a solvent-less gas chromatographic injection.
Catalina et al. [161] determined chlorophenoxy acetic acid herbicides in
water by aqueous based derivatization, extraction and high volume injection.
Dimethyl sulphate and the carboxylic acids reacted in the water sample itself.
Most analytes were esterified by catalysis with tetrabutylammonium salts but
high yield methylation of 4-(4-chloro-2-methylphenoxy) and 4-(2,4-dichloro-
phenoxy)butyric acids required NaOH. The methyl esters were extracted with a
small volume (800 l) of n-hexane. Fully 200 pl or 25% of the extract was injected
into the GC-MS system. A 3 min sample preparation time coupled to injection of
a substantial percentage of the final extract resulted in a high quality method
with a detection limit of 0.5 ig/1.
Stalikas and Pildis [162] optimized the derivatization of phosphoric and
amino acid groups containing pesticides for determination by GC-MS. The
analytes were derivatized in a solution of acetic acid and trimethyl orthoacetate.
The latter compound methylated the phosphoric or carboxylic acids and the
acetic acid acetylated the amines to provide neutral reaction products. These
authors investigated the relationship between yield and amount of reagents,

647
reaction temperature, and reaction time. Optimization was based on the central
composite design. The method achieved virtually quantitative recoveries of
analytes from spiked water samples (96-103%) with excellent precision (RSD <
3.5%). Detection limits ranged from 0.05 to 14 gug/l.
Aging of old masters' paintings can be evaluated by studying oxidation of
abietic acids. Van den Berg et al. [163] investigated several derivatizations for
determining these oxidation products in varnish from the old works of art. They
pointed out that methylation with either TMS-diazomethane or diazomethane
(the former is a less toxic methylating agent) itself is insufficient as many
hydroxylated compounds would remain undetected. Trimethylsilylation of hydr-
oxyl groups-including the tertiary hydroxyl group at the tertiary position-aids
in chromatography of the hydroxylated materials but the mass spectra are domi-
nated by fragmentation via loss of a methyl group. Methylation of the carboxylic
acids followed by silylation was viewed with concern as more extended sample
handling could contaminate the samples which are exceedingly limited in both
size and supply. The authors opted for an on-line derivatization with tetra-
methylammonium hydroxide (TMAH). At elevated temperatures TMAH affects
the following reactions: (1) it esterifies the C-19 carboxyl, (2) it forms the methyl
ether of the tertiary hydroxyl at C-15, and (3) it enolizes and methylates the
ketone at C-7 but also produces the side products via a number of pathways. The
permethylated abietic acids and various reaction products can be separated by
GC. The TMAH was preferred to the "softer" methylating reagents such as
trimethyltrifluorotoluylammonium hydroxide (TMTFH) because the latter does
not methylate tertiary hydroxyls. Van den Berg and co-workers [163] argued
that although there are more degradation products with TMAH, these are
known. It is therefore possible to compensate for the degradation in the analysis.
The advantages of minimal sample handling predominate for the analysis of
these very rare and unique samples.

19.5.3 Alkyl chloroformates

Oxidation of arginine and proline to y-glutamyl semialdehyde may be a new and

useful measure of the degree of protein oxidation. Such oxidation is implicated in
the pathogenesis of disease and toxicity of drugs. Reduction of y-glutamyl
semi-aldehyde to 5-hydroxy-2-aminovaleric acid (HAVA) provides a chemically
more stable marker for this deleterious oxidative process. Derivatization of
HAVA with ethylchloroformate in water/ethanol produced the N(O)-ethoxy-
carbonyl ethyl esters that were determined at the femtomole levels by GC-mass
spectrometry [164]. Measurements of protein oxidation were comparable to
those obtained by the standard, but non-specific, technique for determination of
protein oxidation by measuring the formation of aldehydes such as malonyl-
dialdehyde (MDA). The sensitivity and specificity obtained by this technique,
however, were far greater.
Several groups described an analogy to extractive alkylation. Water immisci-

648
ble organic solvents with lipophilic chloroformates simultaneously extract and
derivatize analytes from water. Kim and Kim [165] described extractive deriva-
tization of phenols from natural waters. The analytes were phenol itself, which
is very hydrophilic, and a substantial number of increasingly substituted (and
more lipophilic) phenols in water. The aqueous phase was acidified and
contacted with methylene chloride containing isobutylchloroformate as the
reagent and trimethylamine as a basic catalyst. The phenols were extracted and
derivatives were retained in the organic phase which was evaporated and the
isolate was further purified prior to analysis.
Extractive derivatization provided a method for determination of estrone
(El) and estradiol (E2). Agitation of mixture consisting of the aqueous sample at
pH = 7, isobutyl-chloroformate and hexane, simultaneously derivatized and
extracted the analytes [166]. Under these conditions, only the phenolic group at
C-3 was derivatized. Prior to gas chromatographic analysis the secondary
hydroxyl at C-17 was derivatized with chlorodifluoroacetic anhydride to produce
the dichlorofluoroacetyl product. The resulting derivatives fragmented to give
high intensity fragments resulting from a loss of the isobutyryl group.
Blair et al. [167] derivatized y-hydroxybutyrate (GHB) with hexylchloro-
formate directly in aqueous matrices, including urine. The derivatives were
isolated from the sample by SPME and analyzed with GC-MS. Results obtained
by the SPME-GC-MS method were comparable to that obtained by standard
methods. The SPME is simpler, faster and is a solvent-less injection. In addition,
the overall technique can differentiate GHB from y-butyrolactone (GBL).
Reaction between amphetamines and 2,2,2-trichloroethylchloroformate
produces an electrophoric reagent suitable for separation by GC and detection
by the classical electron capture detector or mass spectrometry. March and
co-workers developed high quality methods based on this reagent [168].

19.5.4 Perfluorinated reagents

Halogenated acid anhydrides are a classical derivatization for amines and this
derivatization was combined with SPME to yield techniques that meet the more
current requirements of speed, simplicity and automation. Jurado and co-
workers [169] and Lee et al. [170] used SPAD for determination of amphet-
amines. In both instances, the analytes were first sorbed from alkalinized matrix
onto a polydimethylsiloxane fibre. The syringe was then transferred to another
vial and exposed to vapours of trifluoroacetic anhydride [169] or heptafluoro-
butyric anhydride [170]. Finally, the fluorinated derivatives were desorbed by
heating in the injector for 5 min. Reaction time and temperature were critical
both for efficient sorption and derivatization. The range of conditions was quite
varied. Jurado et al. [169] carried out their procedures at temperatures between
60 and 100°C and 20 min extraction time to isolate the analytes and prepare the
derivatives. Lee's group [170] performed the derivatization at 270°C to achieve a

649
reaction with heptafluorobutyric anhydride in 10 s, but perhaps the need for ele-
vated higher also reflected the reduced volatility of the reagent.
Namera et al. [171] isolated basic drugs (fenfluramine, amphetamine and
methamphetamine) from whole blood via head-space SPME and using heat to
volatilize the analytes. The approach differed from the SPME derivatizations
described above [169,170]. In this technique heptafluorobutyric anhydride was
first injected into the injector of the GC-MS and upon insertion of the SPME
fibre the compounds were simultaneously desorbed and derivatized. Sensitiv-
ities were as low as 5 ng/G of blood. This unusual measure for concentration was
used because the samples were of forensic origin and the blood had clotted.
There were two recent reports on pentafluorobenzoylation for determina-
tion of alcohols by GC-MS [172,173]. The work was undertaken because in both
instances other reagents did not produce the necessary chromatographic proper-
ties, informative mass spectra and high sensitivities. Xiao and McCalley [172]
compared fragmentation patterns and sensitivity obtained for pentafluoro-
benzoyl (PFBO) derivatives with those obtained from trifluoroacetyl, hepta-
fluorobutyl, pentadecafluorooctanoyl and perfluorotolyl derivatives. PFBO
derivatization in quantitative yield occurred at each hydroxyl in the three
natural estrogens. The phenol of El was pentafluorobenzoylated as were the
phenol and 17a hydroxyl on E2 and the phenol and the 16,17 vicinal diol on E3.
There was, however, no derivatization of the synthetic and highly hindered 173
hydroxyl of 17a ethinylestradiol. Base peak of mass spectra for PFBO deriva-
tives was the molecular ion with diagnostic fragments arising from the loss of
one, two or three PBO groups.
Weintraub and co-workers [173] reviewed the literature and discussed their
own work on analytical derivatization of platelet activating factor (PAF). They
also found that pentafluorobenzoylation of PAF coupled with GC-MS/negative
ion chemical ionization (NICI) provided the necessary separations and sensitiv-
ity. Initially they reacted PAF with pentafluorobenzoyl chloride (PFBOC1) but
found this problematic because of the reactivity of the Cl group. They recently
prepared and used pentafluorobenzoic anhydride (PFBanh). The highest yields
were obtained with pure molten PFBanh at 110°C. The resulting transacylation
produced mixtures of 1-Oacyl-2-acetyl-3-PFB and 1-O-acyl-2-PFB-3-acetyl
glycerols.

19.5.5 Carbonyls as analyses and reagents

Malondialdehyde is one of the most frequently used markers of oxidative stress.

Derivatization of this dialdehyde with 2,4,6-trichlorophenylhydrazine produced
reaction at only one aldehydic position. The hydrazone then cyclized (Fig. 19.3B)
so that the final product was the cyclic propylenehydrazine [174]. Due to the
three, chlorines the derivatives were readily determined by GC with electron
capture detection and the presence of the molecular ion as the base peak gave
excellent mass spectrometric sensitivity.

650
 As previously noted, formaldehyde is a frequently observed environmental
toxin. Koziel et al. [175] determined formaldehyde in air using SPAD with
0-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride (PFBONH2 ) as
the derivatizing reagent that was deposited on an SPME fibre. A theoretical
treatment of the reaction mechanism demonstrated that sorption-reaction times
should be relatively short and that the loading of reagent on the fibre should be
high. This combination ensures that the amount of reagent used up in the
reaction would be negligible and the reaction rate would be high. Results
obtained by this method were equivalent to those obtained by NIOSH methods,
and, as in the case of most SPME methods, it was found to be simpler in
implementation.
Pentafluorobenzaldehyde (PFBAld) derivatized the biogenic primary alkyl-
amines in wines by GC-MS to form the Schiff base [176]. This is the converse of
an analogous reaction reported by Cancilla et al. [177] who determined
carbonyls in natural waters by reaction with PFBONH2 under acid conditions.
In the former work, however, because the analytes were amines the optimum
yield was achieved at pH = 12.
Alpha ketoisocaproic acid (KIC) and other keto acids present interesting
analytical problems. The need for methods for determining KIC derives from its
function as a dietary replacement for L-leucine and its role in the conversion of
D-leucine into L-leucine. It is relatively difficult to extract and isolate this and
other low molecular weight analytes of intermediary metabolism. Matsukawa et
al. [178] derivatized KIC with N-phenyl-1,2-phenylenediamine to form a stable
heterocyclic derivative of the analyte. This product exhibited a sharp peak on GC
and provided spectra with strong molecular ions as well as informative frag-
ments. The limit of quantitation using 50 tkl of plasma was 50 nM. Since the
reaction was based on the a ketoacid moiety, this derivatization could be used to
determine other a ketoacids. Reaction conditions used in this work are harsh
(100°C) and may have to be modified for other analytes which may not be as
stable.

19.5.6 Other reagents and analytes

19.5.6.1 Determination of D and L-Leucine

The biological role of D-leucine and its conversion to the L-leucine remains an
important biochemical problem that has produced interesting developments in
analytical methodology. The natural enantiomer is L-leucine, but D-leucine is
found at low concentrations in mammalian systems and, more importantly, it is
also a side product of food processing. The D enantiomer can be converted into
the L form via KIC (the reverse reaction does not occur in mammalian biochem-
istry). Rate of conversion is one determinant of the nutritional value of
processed food. For these reasons, sensitive and specific methods are required
for the chiral separation and determination of the two enantiomers of leucine.

651
 Despite the high quality of information available from mass spectrometry,
determination of D and L-leucine by this instrumental technique remains
dependent on proper selection of derivatizing reagent. Hasegawa and co-
workers [179,180] selected a-methoxy-a-trifluoro-methylphenylacetyl chloride
(MTPA-Cl) as a chiral reagent because the fully substituted chiral centre is
impervious to racemization. The derivatization and chromatographic
procedures were relatively straightforward with good separation between D and
L-leucine with both the + and - MTPA but the presence of L-isoleucine, an
isomer of leucine, provided the analytical challenge. It was necessary not just to
separate the two enantiomers bot to all three amino acids prior to mass
spectrometry this was achieved only with +MTPA.

19.5.6.2 Determinationof the 3C/12 C isotopic ratios in sugars

Two different analytical derivatizations were evaluated for determination of the
13
C/' 2 C ratio in sugars. Knowledge of this ratio is necessary for studies in such
diverse fields as environmental microbiology and human genetic diseases. Dur-
ing investigations of galactosemia, Fen et al. [181] prepared the pentaacetyl-
aldonitrile as a derivative for GC-MS with positive ion chemical ionization
(PICI) and determined the 13C/' 2 C ratio in the galactose pool. Ning [182] com-
pared this derivatization to n-butylboronylation of galactose. Formation of the
pentaacetylaldonitrile gave systematically higher readings of the atom percent
excess and this overestimated the apparent galactose appearance rate. A prefer-
ence for the alkylboronylation reaction was supported by the results of van
Dongen and co-workers [183] who determined the natural 13C content of mono-
saccharides in freshwater algae. They provided evidence that alkylboronylation
does not result in fractionation of 13C during derivatization thus ensuring that
values of 13C/1 2C ratios determined by this procedures are true measures of natu-
ral 3 C content.

19.5.6.3 Bromoacetonitrile
Investigators are beginning to use bromoacetonitrile (BrACN) for analytical
derivatizations. Very long chain fatty acids [184] and phenols [185] were
derivatized with BrACN to the cyanomethyl esters and ethers. Separation of
these products by GC and detection with a nitrogen phosphorus selective (NPD)
detector sensitivity and specificity. Detection limits of 100 pg/ml [185] were
achieved and this was sufficient to detect the analyte in tap water as well as
environmental waters.

19.6 LIQUID CHROMATOGRAPHY-MASS SPECTROMETRY AND

RELATED TECHNIQUES

Liquid chromatography coupled with mass spectrometric analysis is one of the

most technologically complex and scientifically advanced types of instrumenta-
tion available to the separation sciences. Indeed, with modern instrumentation

652
the mass spectrometer itself can provide substantial separation and quanti-
tation. Despite the technological and scientific sophistication of the technique,
there are numerous instances where derivatization can enhance both the
sensitivity of detection and the quality of information on fragmentation and
structure of the analyte.

19.6.1 Derivatization of amino acids and peptides on the nitrogen

Under carefully controlled conditions 2,4-dinitrofluorobenzene will react with

the alpha amino group but not the epsilon amine of lysine or the imidazole of
histidine [186]. A combination of this selectivity plus the use of an equimolar
mixture of 2,4-dinitrofluorobenzene and [2H3]2,4-dinitrofluorobenzene tagged
the N-terminus of tryptic peptides such as Val-His-Leu-Thr-Pro-Val-Glu-Lys. In
the electrospray ionization-tandem mass spectrometry (ESI-MS/MS) and liquid
chromatography (LC)-ESI-MS the 2,4-dinitrobenzene modified peptides
produced very characteristic patterns. The tag produced a doublet of equal
intensity at [M+H]+ but with a delta m/z of 3 and at [M+2H]2+ a delta m/z.
Since the tag was specific for the N terminus this pattern distinguished the N
and C termini of the peptide.
Derivatization with FMOC was also used in the determination of amino acids
and peptides. At the usual pH of the reaction (i.e. pH = 10), FMOC labelled the
epsilon N of lysine, the imidazole N of hisidine and the OH of tyrosine [187].
Detection limits for the determination of free amino acids and peptides
decreased from 10-50 nmol for the native analytes to several hundred picomoles
following derivatization. Moreover, collision-induced dissociation of the
derivatized peptides produced mostly b-type ions. It is relatively less complicated
to assign peptide sequence based on these b-type ions.

19.6.2 Derivatization at carboxylic acids and phenols

Pentafluorobenzyl bromide (PFBBr) has been used extensively for gas chroma-
tographic determination of organic acids [137]. Derivatization is very straight
forward and the five fluorine atoms give very high sensitivity via the electron
capture detector and by NICI. Singh et al. [188] reasoned that in conventional
atmospheric pressure chemical ionization the displacements of electrons from
the nitrogen sheath gas in the corona discharge creates a source of gas phase
electrons. Under these conditions pentafluorobenzyl (PFB) derivatives should
be detected at very high sensitivity because these derivatives will efficiently
capture electrons and then fragment. They prepared and determined PFB esters
and ethers of amino prostaglandins, phenolic estrogens and adducts of 2'-deoxy-
guanosine produced by the reaction between 4-oxo-2-nonenal and DNA. The
PFB derivatives fragmented by dissociative electron capture and produced very
characteristic spectra. The technique was termed electron capture atmospheric
pressure chemical ionization (ECAPCI). HPLC-MS/ECAPCI coupled with

653
selected reaction monitoring produced detection limits as low as 140 attomole
for prostaglandins and thromboxanes. The results for the prostaglandins are
particularly interesting. Pentafluorobenzylation is only one step in the prepara-
tion of these eicosanoids prior to determination by GC-MS. A highly sensitive
HPLC-MS/MS technique would eliminate the additional silylations and
oximations that are required for analysis by GC-MS.

19.6.3 Saccharides

Derivatization of oligosaccharides with 1-phenyl-3-methyl-5-pyrazolone (PMP)

[84,182-184] is under active study as an important technique for HPLC-MS/MS
of these bio-polymers. Reaction typically occurs in heated water/methanol
solution and is catalyzed by NaOH. As with other reagents, the derivatization is
at the aldehydic position but in this case two PMP molecules are incorporated at
that position (Fig. 19.10). The PMP moiety incorporates a chromophore that can
be detected by absorption in the ultraviolet and also imparts lipophilicity that
improves separations.
Suzuki et al. [189] reported a study of sialo-N-glycans. A combination of
derivatization and enzymatic hydrolysis provided PMP derivatives of oligo-
saccharide that gave the attaching positions of the sialic acids. Micellular
electrokinetic capillary chromatography in sodium dodecylsulphate separated
the PMP derivatives. The technique demonstrated that sialic acids are retained
on the polysaccharides under the alkaline conditions of the reaction between
polysaccharides and PMP.
Relative to the native analytes, these derivatives had superior chromato-
graphic properties as well as more informative mass spectra [88,190,191]. In the
case of sialo and asialo N linked oligosaccharides the PMP derivatives were
better retained than the native analytes on the standard reverse phase C-18
columns. As a result, derivatization provided separations that were not feasible
with the natural compounds. There was little difference in the sensitivity for

HO
2

RMP CH,
HO
HO

OH
Fig. 19.10. Reaction between PMP and a O o o
sugar. H H

654
derivatized and underivatized asialo oligosaccharides, but substantial improve-
ments were found for the derivatized mono and disialo compounds. Compared to
the native N-linked oligosaccharides, derivatization of the monosialo compounds
increases the sensitivity by 100-fold and detection of the disialo analyte is
increased by 100% following derivatization [88]. Moreover, the mass spectra of
the native compounds are uninformative as to the branching. In contrast, the
PMP derivatives fragment to produce ions that contained the reducing end of
the molecule and this facilitates interpretation. Derivatization of neutral and
N-acetylated proved equally beneficial. In this case sensitivity of detection by
mass spectrometry was seen for all analytes. Again, PMP derivatives underwent
directed cleavage towards the reducing end of oligosaccharide.
The sulphonated fluorophore, 2-aminonaphthalene trisulphone (ANTS) is a
useful reagent for the analysis of oligo and polysaccharides. The derivatizing
group imparts a charge onto neutral oligo and polysaccharides. As such it
produces derivatives suitable for electrophoresis as well as providing fluores-
cence for high sensitivity detection [190]. Poly and oligosaccharides derivatized
with ANTS were separated by polyacrylamide gel electrophoresis and by CZE.
This utility of this reagent for HPLC-ESI-MS was compared to that of PMP to
determine whether a single reagent can be used for all these separation/detec-
tion techniques. Due to its greater lipophilicity, PMP provided superior
chromatographic properties in both normal and reversed phase HPLC. In
addition, while ANTS was difficult to separate from the highly water derivatives,
a simple liquid/liquid extraction separated excess lipophilic PMP from the
derivatives which were hydrophilic by virtue of their saccharide moiety.
As noted above, sialic acids are critical recognition molecules in biological
processes. Their importance in physiology is matched by the difficulty in produc-
ing fragments in MALDI-TOF that retain the sialic acids. Leavell and Leary [89]
addressed the important problem of stabilizing sialic acid on oligosaccharides
during mass spectral analysis. They prepared the diethylenetriamine (DETA)
derivative of sialylated oligosaccharides and then complexed these compounds
with transition metals. They demonstrated that whereas the underivatized
oligosaccharide lost the sialic acids the cobalt complexed DETA derivative
produced an intense M-H + ion as well as several other fragments, including the
base peak, that retained the sialyl group.

19.6.4 Vitamin D3 and dienophiles

Determination of VitD3 and its therapeutic analogues by HPLC-MS/MS offers a

viable alternative to the well studied GC-MS. The latter instrumentation puts at
risk the thermally labile unsaturated and silylated derivatives. The former tech-
nique, however, requires tagging with reagents that provide a site for protona-
tion to maximize the sensitivity obtained by APCI or ESI. An alternative class of
derivatizing reagents is dienophiles. Because of the conjugated dienes in VitD3,
derivatization with dienophiles is very specific [192,193]. The triazoline 4-

655
phenyl-1,2,4-triazoline-3,5-dione (PTAD) [192] is one dienophile that reacts
with Vit D3 and its pharmaceutical analogue EWB-1089 [193]. PTAD deriva-
tives of the latter analyte in particular require more detailed study. The 22,
23-24, 24a diene in the side chain of EWB 1089 provides an alternative site for
derivatization. Weiskopf and co-workers [185] characterized the structure of the
diagnostic ion at 314. This information will enhance development of other meth-
ods for Vit D3, its analogues and other diene containing analytes. Higashi et al.
[193] derivatized VitD3 with 4-[4-(6-methoxy-benzoxazoyl)phenyl]-phenyl-
1,2,4-triazoline-3,5-dione (MBOTAD) and used chromatography on OASIS and
silica gel columns to purify the resulting derivative. Although sample prepara-
tion is lengthy the method is highly sensitive with a detection limit of 18 fmol
injected on column but in this work the mass spectrometry of the derivatives was
not studied.

19.6.5 Alcohols

Quirke and Van Berkel [194] continued their work on the preparation of
ferrocene derivatives for the derivatization of alcohols. Derivatives are simply
prepared by heating the alcohols with ferrocenoyl azide in anhydrous toluene.
The neutral derivatives are oxidized to the corresponding ferrocenium products
(F+ ) during the processes involved in electrospray ionization. The combination
of a variety of mass spectrometric techniques gives a ready identification of
ferrocenoyl derivatives of the primary, secondary and tertiary alcohols as well as
providing information on other structural features.

19.6.6 Preparation of proteins for MALDI-TOF.

Khorana's group provided a tour-de-force on the use of analytical derivatizations

preparatory to MALDI-TOF [70-72]. This group studied the mutant forms of
rhodopsin (RHO) and the binding between rhodopsin and transducin (T). These
are critical studies in the biochemistry of vision. Analytical derivatizations were
used to block mercapto groups, impart chromatographic properties, cross-link
proteins and tag active sites and disulphide binding in misfolded rhodopsin
found in retinitispigmentosa.
A sequence of blocking and labelling reactions identified the cysteines that
formed the disulphide link in normal and the misfolded RHO associated with
retinitispigmentosa [70]. The first step blocked free Cysteinyl (Cys) SH groups
with NEM. The NEM-RHO was then bound to anti-RHO ld4-Sepharose. While
bound to Sepharose, the disulphide residue (CysS-SCys) was reduced with DTT
and produced free CysSH. At this stage, the only free CysSH groups were those
that were originallyin a disulphide linkage. Without eluting from the Sepharose,
the CysSH groups were derivatized with maleimido-butyryl-biocytin (MBB) to
give NEM-RHO-MBB. Thus all the original CysSH groups of RHO were tagged
with NEM and all the original CysS-SCys were tagged with MBB. The MBB was

656
linked via the maleimido group and this left the biocytin (or biotin) group free for
subsequent sorption onto a column loaded with covalently bound avidin. Avidin
readily binds biotin and provided a selective separation technique. The NEM-
RHO-MBB protein was eluted from the Sepharose column by displacement with
nonapeptide and was hydrolyzed by proteinase K. The peptide fragments were
then added to an avidin-agarose column and those peptides containing the biotin
moiety were sorbed. Washing with water removed the peptides without MBB
tags. Finally the MBB labelled peptides were eluted by displacement with biotin
and analyzed by MALDI-TOF. In the normal RHO, H tagged the cysteine at the
185 position. The MBB label was attached to cysteines at the 110 and 187
position, demonstrating that in the protein these two cysteines had originally
formed an CysS-SCys bond. In misfolded RHO the cysteine at the 185 and the
187 positions were tagged with MBB whereas the cysteine at the 110 position
carried the NEM tag. It follows that the disulphide bond in misfolded RHO was
between 185 and 187.
A similar approach was used to identify the contact sites between RHO and T
[71,72]. Sensitization and desensitization are controlling features in light activa-
tion of visual signal transduction. A critical feature of this process is the binding
of T and rhododopsin kinase to light activated RHO. Competition between T and
the kinase for binding interaction controls activation and quenching of the
biochemical cascades involved in vision. Elucidation and understanding of these
phenomena requires a more detailed knowledge of the binding sites. Analytical
derivatizations at the thiol group of cysteine and other moieties were key in
identifying these sites.
The first step was binding RHO to Sepharose beads through the RHO-
1D4-antibody. Subsequent reactions were then performed on RHO immobilised
on the solid surface (S-RHO). The first reaction was with one of two pyridyl-
dithio reagents that were designed to ultimately cross-link the RHO and T using
either light activation or chemical activation. The light activated reagent was
N-((2-pyridyldithio)ethyl)-4-azidosalicylamide (PEAS) and N-succinimidyl 3-
(2-pyridyldithio) propionate (SPDP) [71] was the chemically activated [72]
reagent. These reagents were linked to the S-RHO via a disulphide group that
could later be reduced.
Addition of a GDP/T complex to the S-RHO-linking group complex followed
by activation with light at >495 nm bound the GDP/T complex with the
S-RHO-linking group [71]. Irradiation at 310 nm activated the PEAS labelled
S-RHO while adjusting the pH activated the SPDP labelled S-RHO [72]. The
activation cross-linked the S-RHO and T. Washing the cross-linked complex
with GTP removed the GDP required for the binding of T to RHO. Deriva-
tization with NEM blocked the remaining CysSH groups and reduction with
DTT broke the link between the S-RHO and the now labelled T. The freed and
tagged T was then derivatized with MBB, as described above. The purification
and hydrolysis were also the same as used for identification of the disulphide link
in the misfolded RHO.

657
19.7 CHIRAL SEPARATIONS

An interesting isothiocyanate was described and used by two groups [195,196].

The isothiocyanate group was linked to an acetylated sugar to yield 2,3,4,6-
tetra-o-acetyl-3-D-glucopyranosyl isothiocyanate (GITC). The chiral amine of A
receptor active drugs (terbutaline, atenolol), reacted with GITC to form the
diastereomers which were separated by a relatively simple chromatographic
system (an ODS column with acetonitrile-acetate buffer as a mobile phase). Due
to the rather bulky groups, GITC appeared to be less reactive than either PITC
or FITC [196] and quantitative derivatization required 1 h at room temperature
but once recognized the longer reaction time was not a problem.
Marfey's reagent, or 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide forms
reaction products suitable for chiral separation of amino acids [197,198]. The
2,4-dinitrophenol moiety of this reagent engenders high molar absorptivity. At
340 nm, = 30,000 M-1 cm-l and this produces detection in the sub-nmol range.
Moreover, the 340 nm region is relatively free of endogenous interferences.
Incorporation of an amino acid moiety provides a chiral centre and so the
enantiomeric derivatives readily separate on standard HPLC columns. Torok
and co-workers [198] extended these studies and compared chiral separations
based on derivatization with Marfey's reagent to those obtained by separations
on various cyclodextrin phases. Peter et al. [199], however, found that deriva-
tization with Marfey's reagent was more susceptible to steric hindrance and
suggested that GITC was preferable for chiral separation.
L-alanine-j-naphthylamide (L-Ala-p-NA) is another reagent that coupled an
enantiomerically pure amino acid with a reactive group. In this case the amino
acid provided both chiral centre but since the carboxylic acid was amidated the
amino group was also reactive towards acetyl groups. Kagawa et al. [200]
determined the enantiomeric purity of acetyl-L-carnitine by derivatization with
this reagent. They tested a number of chiral amino compounds and obtained the
best results with L-Ala-P-NA. The chromophore was detected with good sensitiv-
ity at 254 nm and less than 0.05% D acetyl carnitine could be detected in the
presence of L-acetylcarnitine.
An intermediate reagent from the production of chloramphenicol and its
diastereoisomer provided Peter et al. [201] with a new chiral derivatizing
reagent, (S,2S)- or (1R,2R)..1,3-diacetoxy-1-(4-nitrophenyl)-2-propylisothiocya-
nate (DANI) that was easily accessible in both enantiomeric forms after a simple
two-step synthesis. Derivatization conditions were optimized with respect to pH,
reaction temperature, excess of reagent etc. Its utility was demonstrated on the
example of the resolution of a series of a-amino acids. The diastereomeric
thiourea derivatives were separated by HPLC.
A combination of chromatography on cyclodextrin and derivatization of L
and D enantiomer of homnocysteine with ABD-F gave chiral separation [202].
The lipophilic ABD group readily fits into the pore of cyclodextrin. In this
inclusion complex, the amino acid moiety was oriented towards the opening

658
where hydrogen bonding interactions enabled the chiral separations. Under
these conditions, a useful and interesting result is that the reagent elutes after
the derivatized analyte as the latter is more hydrophilic.

19.8 INORGANIC ANALYSES

Analytical derivatizations are usually associated with organic compounds. There

is, however, an accumulating body of literature on the derivatization of inor-
ganic anions and cations. These reports impact a broad range of applications.
Nitric oxide is an important mediator and its role in inflammation, vaso-
dilation etc. is long known and well documented [203]. This compound is
difficult to determine because it is both volatile and unstable. Several groups
have measured the production of NO by determining nitrite [204-206]. Reaction
of nitrite with 2,3-diaminonaphthalene (DAN) under acid conditions produces a
highly fluorescent derivative that is readily analyzed by HPLC [204]. Nitrate
was determined by difference after conversion to nitrite by nitrate reductase.
The authors report a detection limit in the picomole range.
Nitrate and nitrite in biological fluids have also been determined by GC with
mass spectrometric detection [205,206]. In both cases, PFBBr derivatized nitrite
to the PFB derivatives. Extractive alkylation with tetradecyldimethylbenzyl-
ammonium chloride as the phase-transfer catalyst extracted the nitrite from the
aqueous into the organic phase (ethyl acetate) where it was converted into PFB
derivative [205]. Nitrate was again determined by difference following reduction
to nitrite by hydrazine sulphate in the presence of Cu 2+ and Zn2+. It is also
feasible to directly derivatize nitrite with PFBBr directly in the aqueous phase
[206]. In this case, the lipophilic PFBBr is solubilized by adding equal parts of
acetone to the aqueous sample. Upon completion of the reaction, the acetone was
evaporated prior to extraction of the derivative with toluene. In this last report
cadmium was used to reduce nitrate to nitrite.
Mercury is an environmental toxin that is particularly hazardous in its
monomethyl and dimethyl forms: it is an occupational hazard, particularly for
chemists. Ingestion of dimethyl mercury in the course of a nuclear magnetic
resonance experiment was responsible for the death of K.E. Wetterhahn, a
chemist with considerable expertise in mercury chemistry [207]. Deitz et al.
[208] reported a semiautomated method for the speciation of mercury via
derivatization with sodium tetraethylborate (NaBEt4 ). The instrumentation
performs cryogenic trapping and chromatographic separation within a capillary.
Chromatograms are obtained within 3 min. Derivatization, gas phase extraction
and preconcentration are all instrumentally controlled. Reported detection
limits were 33 ng/l for dimethyl mercury, 39 ng/l for monomethyl mercury and
71 ng/ml for mercury.
Although derivatization of mercury with NaBEt4 is reasonably established,
Cai et al. [209] suggest that sodium tetraphenylborate (NaBPh 4 ) may be a
preferred reagent. Diethyl mercury present in soil would not be distinguished

659
from the mercury derivatization product by reaction with NaBEt4 . Sodium
phenyltetraborate produces unique derivatives for all species of mercury.
Sodium tetraethylborate is also effective at producing volatile derivatives of
tin and lead. These derivatizations occur in the aqueous sample. Subsequent
isolation and injection into the chromatographic system require the develop-
ment of rapid and facile techniques. Milan and Pawliszyn [210] prepared the
derivatives of butyltin species using NaBEt4 . Alkyl lead and inorganic lead were
the subjects of an investigation by Yu and Pawliszyn [211] who derivatized these
analytes with deuterium-labelled sodium tetraethylborate NaB(C2 Ds)4
(DSTEB). Detection by mass spectrometry readily distinguished between Pb2 + ,
Pb(CH 3 )3+ , Pb(C2 H5 )3 +, and Pb(C2 H5 )4 since any derivative had the deuterated
tag which is not found in nature. This was analogous to the use of NaBPh4 to
produce an unambiguous determination of metal species in the environment. In
both instances, for study of tin or lead, the isolation was affected by SPME with
collection in the head-space. The techniques provided results that matched the
requirements of regulatory agencies and because of the SPME technique are
well suited to field analysis.
Methylarsonic acids were converted to lipophilic species with thioglycol
methylate [212]. Derivatization was carried out directly in acidified urine and
the derivatives were extracted into hexane for subsequent analysis by GC-MS.
Although there is a substantial dilution factor in the extraction, the overall
technique is simple, rapid and provides good sensitivity and reproducibility.

19.9 SUMMARY

It is evident that there are advantages to exploiting analytical derivatizations;

there are also cautions and difficulties. The developments in CZE/LIF, SPME,
SPAD, miniature and automated devices provide some resolution to the
problems of the extra steps, interferences from excess reagents as well as from
the matrix of the sample and the matrix of the derivatization reaction. Such
studies demonstrate that analytical derivatizations can be made more "user
friendly". The improvements in sensitivity and ease of sample preparation are
worth the effort that goes into these endeavours.

19.10 GLOSSARY

Alkylating anticancer drugs: Bifunctional drugs containing an alkylating

group. These drugs bind covalently to DNA or RNA and cross-link the
strands of the nucleic acid. Because of their potent chemical reactivity these
drugs have a low therapeutic index defined as the ratio of therapeutic to
toxic dose or plasma concentrations.
Analytical derivatizations: Reactions carried out on the analytical scale to
incorporate a sensitizing group into a molecule preparatory to instrumental
analysis.

660
Biogenic amines: Any amine, other than amino acids or nucleic acids that are
part of normal physiology. This includes the catecholamine, indoleamines,
y-hydroxybutyric acid. The catecholamines include epinephrine (adren-
aline) nor epinephrine (nor-adrenaline) and dopamine. Indoleamines
include serotonin, N-acetylserotonin and melatonin. These compounds are
involved in the functioning of the central and autonomic nervous system
and are responsible for the fight or flight response.
Biogenic polyamines: Diamines such as putrescine, cadaverine, spermidine
and spermine. These are compounds involved in signalling between cells
and are thus important markers of normal tissue development and develop-
ment of tumours.
Bio-protective peptides and amino acids: Peptides or amino acids that
contain a free thiol (-SH) that can react with ROS. The peptides are
glutathione (GSH) and y-glutamylcysteine and the amino acid is cysteine.
The major bio-protective peptide is GSH. In addition to reacting with ROS,
the peptide GSH can also react with alkylating species. Alkylating species
are usually drugs or reactive compounds formed from the biotrans-
formation of drugs.
Chemical cytometry: The qualitative and quantitative determination of
compounds produced by a single cell and use of this data to interpret the
biology of the cell.
Dextran ladder: A group of oligosaccharides derived from dextran that provide
a series of saccharides with increasing molecular weight. Used to calibrate
columns for determination of molecular weight of unknown mono, oligo
and polysaccharides.
Liquid chromatography: Any chromatographic technique in which the
mobile phase is liquid and the stationary phase is a solid. This includes high
performance liquid chromatography, thin layer chromatography and
capillary zone electrophoresis.
Metabolic cytometry: The analysis of metabolic interconversions in a single
cell.
Rhododopsin: One of two proteins that control the activation of the retina in
response to light.
Solid phase extraction (SPE): Isolation of an analyte from aqueous matrix
by sorption onto a solid phase and subsequent elution with a volatile
organic solvent.
Solid phase micro-extraction (SPME): Isolation of an analyte from aqueous
matrix by sorption onto a solid phase consisting of a glass fibre coated with
sorbent. The glass fibre is on a device that permits direct insertion into an
injector port of a gas chromatograph for thermal desorption or a port of an
liquid chromatograph for on-line elution.
Solid phase analytical derivatizations (SPAD): A combination of sorption
of analyte onto a solid phase and derivatization on the phase.
Transducin: One of two proteins that controls the activation of the retina in

661
 response to light.
Vasodilator: Any endogenous or exogenous compound that increases the
lumen of the blood vessels. Endogenous vasodilators such as nitric oxide
serve to maintain blood pressure and respond to normal or pathophysi-
ological events that may occlude the blood vessel. Exogenous vasodilators
are drugs that are used to overcome deficits in the functioning of the
endogenous vasodilators.

REFERENCES

1 C.D. Stalikas and C.N. Konidari, J. Chromatogr. A, 907 (2001) 1-19.

2 S. Oguri, J. Chromatogr. B, Biomned. Sci. Appl., 747 (2000) 1-19.
3 I. Molnar-Perl, J. Chromatogr. A, 891 (2000) 1-32.
4 S.N. Krylov, E. Arriaga, Z. Zhang, N.W. Chan, M.M. Palcic and N.J. Dovichi, J.
Chromatogr. B, Biomed. Sci. Appl., 741 (2000) 31-35.
5 I. Molndr-Perl, J. Chromatogr. A, 913 (2001) 283-302.
6 A. Bouchereau, P. Guenot and F. Larher. J. Chromatogr. B, Biomed. Sci. Appl., 747
(2000) 49-67.
7 H. Wan and L.G. Blomberg, J. Chromatogr. A, 875 (2000) 43-88.
8 M.C. Garcia Alvarez-Coque, M.J. Medina Hernandez, R.M. Villanueva Camanas and
F.C. Mongay, Anal. Biochem., 178 (1989) 1-7.
9 Y.V. Tcherkas and A.D. Denisenko, J. Chromatogr. A, 913 (2001) 309-313.
10 G.P.M.A., Hardy, F.J. Van Hemert, A.J. Meijer and J. Goudsmit, Anal. Biochem., 291
(2001) 297-299.
11 H. Liu, Meth. Mol. Biol., 159 (2000) 123-140.
12 A. Vasanits, D. Kutlan, P. Sass and I. Molnar-Perl, J. Chromatogr. A, 870 (2000)
271-287.
13 F.R. Antoine, C.I. Wei, R.C. Littell and M.R. Marshall, J. Agric. Food Chem., 47
(1999) 5100-5107.
14 H.M. van Eijk, D.R. Rooyakkers, P.B. Soeters and N.E. Deutz, Anal. Biochem., 271
(1999) 8-17.
15 B.H. Westerink, J. Chromatogr.B, Biomed. Sci. Appl., 747 (2000) 21-32.
16 C.S. Yang, P.J. Tsai, W.Y. Chen, W.J. Tsai and J.S. Kuo, J. Chromatogr. B, Biomed.
Sci. Appl., 734 (1999) 1-6.
17 Y.S. Wu, H.K. Lee and S.F. Li, Anal. Chem. 72 (2000) 1441-1447.
18 S.R. Ruberu, W.M. Draper and S.K. Perera, J. Agric. Food Chem., 48 (2000)
4109-4115.
19 R. Herraez-Herandez and P. Campins-Falco, J. Chromatogr. A, 893 (2000) 69-80.
20 K. Linnemayr, A. Bruckner, R. Korner, R. Hahn, A. Jungbauer, D. Josic, P.
Roepstorff, A. Rizzi and G. Allmaier, J. Mass Spectrom., 34 (1999) 427-434.
21 A.J. Shah, S.G. de Biasi, V. Taylor, C. Roberts, P. Hemmati, R. Munton, A. West, C.
Routledge and P. Camilleri, J. Chromatogr. B, Biomed. Sci. Appl., 735 (1999)
133-140.
22 S. Zhao, C. Prasad, H.J. Robertson and Y.M. Liu, Fresenius J. Anal. Chem., 369
(2001) 220-224.
23 A.L. Freed and S.M. Lunte, Electrophoresis,21 (2000) 1992-1996.
24 M.T. Chiang, S.Y. Chang and C.W. Whang, Electrophoresis 22 (2001) 123-127.
25 M.T. Chiang, S.Y. Chang and C.W. Whang, J. Chromatogr.A, 877 (2000) 233-237.
26 N. Gottschlich, C.T. Culbertson, T.E. McKnight, S.C. Jacobson and J.M. Ramsey, J.
Chromatogr. B, Biomed. Sci. Appl., 745 (2000) 243-249.

662
27 C. Parmentier, M. Wellman, A. Nicolas, G. Siest and P. Leroy, Electrophoresis,20
(1999) 2938-2944.
28 Z. Chen, J. Wu, G.B. Baker, M. Parent and N.J. Dovichi, J. Chromatogr.A, 914 (2001)
293-298.
29 Z. Zhang, S. Krylov, E.A. Arriaga, R. Polakowski and N.J. Dovichi, Anal. Chem., 72
(2000) 318-322.
30 J. Wu, Z. Chen and N.J. Dovichi, J. Chromatogr.B, Biomed. Sci. Appl., 741 (2000)
85-88.
31 S. Hu, Z. Zhang, L.M. Cook, E.J. Carpenter and N.J. Dovichi, J. Chromatogr. A, 894
(2000) 291-296.
32 I.H. Lee, D. Pinto, E.A. Arriaga, Z. Zhang and N.J. Dovichi, Anal. Chem., 70 (1998)
4546-4548.
33 S.N. Krylov, Z. Zhang, N.W. Chan; E. Arriaga, M.M. Palcic and N.J. Dovichi,
Cytometry, 37 (1999) 14-20.
34 S.N. Krylov, D.A. Starke, E.A. Arriaga, Z. Zhang, N.W. Chan, M.M. Palcic and N.J.
Dovichi, Anal. Chem., 72 (2000) 872-877.
35 Z. Zhang, S. Krylov, E.A. Arriaga, R. Polakowski and N.J. Dovichi, Anal. Chem., 72
(2000) 318-322.
36 S. Hu, R. Lee, Z. Zhang, S.N. Krylov and N.J. Dovichi, J. Chromatogr.B, Biomed. Sci.
Appl., 752 (2001) 307-310.
37 G. Aymard, B. Labarthe, D. Warot, I. Berlin and B. Diquet, J. Chromatogr. B,
Biomed. Sci. Appl., 744 (2000) 25-31.
38 R.W. Sparidans, J. den Hartig, S. Cremers, J.H. Beijnen and P. Vermeij, J.
Chromatogr. B, Biomed. Sci. Appl., 738 (2000) 331-341.
39 R. Herraez-Hernandez and P. Campins-Falco, Analyst, 125 (2000) 1071-1076.
40 D.H. Shangguan, Y.X. Zhao, H.W. Han, R. Zhao and G.Q. Liu, Anal. Chem., 73 (2001)
2054-2057.
41 M. Molina and M. Silva, Electrophoresis,22 (2001) 1175-1181.
42 E. Causse, N. Siri, J.F. Arnal, C. Bayle, P. Malatray, P. Valdiguie, R. Salvayre and F.
Couderc, J. Chromatogr.B, Biomed. Sci. Appl., 741 (2000) 77-83.
43 T. Toyo'oka, D. Jin, N. Tomoi, T. Oe and H. Hiranuma, Biomed. Chromatogr., 15
(2001) 56-67, 2001.
44 M.J. Baars and G. Patonay, Anal. Chem., 71 (1999) 667-671.
45 S. Hu and P.C. Li, J. Chromatogr.A, 876 (2000) 183-191.
46 T. Inoue, K. Hamase, A. Morikawa and K. Zaitsu, J. Chromatogr.B, Biomed. Sci.
Appl., 744 (2000) 213-219.
47 M. Kato, M.T. Dulay, B. Bennett, J. Chen and R.N. Zare, Electrophoresis, 21 (2000)
3145-3151.
48 S. Honda, J. Okeda, H. Iwanaga, S. Kawakami, A. Taga, S. Suzuki and K. Imai, Anal.
Biochem., 286 (2000) 99-111.
49 C. Molins-Legua, P. Campins-Falco, A. Sevillano-Cabeza and M. Pedron-Pons,
Analyst, 124 (1999) 477-482.
50 H. Nohta, H. Satozono, K. Koiso, H. Yoshida, J. Ishida and M. Yamaguchi, Anal.
Chem., 72 (2000) 4199-4204.
51 A. Asthana, D. Bose, A. Durgbanshi, S.K. Sanghi and W.T. Kok, J. Chromatogr. A,
895 (2000) 197-203.
52 O.Y. Al-Dirbashi, M. Wada, N. Kuroda, M. Takahashi and K. Nakashima, J. Forensic
Sci., 45 (2000) 708-714.
53 Y. Sun, M. Wada, O. Al Dirbashi, N. Kuroda, H. Nakazawa and K. Nakashima, J.
Chromatogr. B, Bioamed. Sci. Appl., 749 (2000) 49-56.
54 M. Wada, S. Kinoshita, Y. Itayama, N. Kuroda and K. Nakashima, J. Chromatogr.B,
Biomed. Sci. Appl., 721 (1999) 179-186.

663
55 H. Wang, J. Li, X. Liu, T.X. Yang and H.S. Zhang, Anal. Biochem., 281 (2000) 15-20.
56 Z. Yan, J.G. Nikelly, L. Killmer, Jr. and J.B. Tarloff, Drug Metab Dispos., 28 (2000)
880-886.
57 F.Q. Schafer and G.R. Buettner, Free Radic. Biol. Med., 30 (2001) 1191-1212.
58 J.B. Ubbink, Semin. Thromb. Hemost., 26 (2000) 233-241.
59 M.J. MacCoss, N.K. Fukagawa and D.E. Matthews, Anal. Chem., 71 (1999)
4527-4533.
60 A. de Graaf-Hess, F. Trijbels and H. Blom, Clin. Chem., 45 (1999) 2224-2228.
61 D. Tsikas, J. Sandmann, D. Holzberg, P. Pantazis, M. Raida and J.C. Frolich, Anal.
Biochem., 273 (1999) 32-40.
62 J.F. Salazar, H. Schorr, W. Herrmann, B. Herbeth, G. Siest and P. Leroy, J.
Chromatogr. Sci., 37 (1999) 469-476.
63 E.D. Bald, R. Glowacki and J. Drzewoski, J. Chromatogr. A, 913 (2001) 319-329.
64 A.R. Ivanov, I.V. Nazimov, L. Baratova, A.P. Lobazov and G.B. Popovich, J.
Chromatogr. A, 913 (2001) 315-318.
65 A.R. Ivanov, I.V. Nazimov and L. Baratova, J. Chromatogr.A, 895 (2000) 157-166.
66 A.R. Ivanov, I.V. Nazimov and L.A. Baratova, J. Chromatogr.A, 870 (2000) 433-442.
67 E. Causse, N. Siri, H. Bellet, S. Champagne, C. Bayle, P. Valdiguie, R. Salvayre and F.
Couderc, Clin. Chem. 45 (1999) 412-414.
68 D.E. Hammermeister, J. Serrano, P. Schmieder and D.W. Kuehl, Rapid Commun.
Mass Spectrom. 14 (2000) 503-508.
69 A.I. Haj-Yehia, P. Assaf, T. Nassar and J. Katzhendler, J. Chromatogr.A, 870 (2000)
381-388.
70 J. Hwa, J. Klein-Seetharaman and H.G. Khorana, Proc. Natl. Acad. Sci. USA, 98
(2001) 4872-4876.
71 K. Cai, Y. Itoh and H.G. Khorana, Proc. Natl. Acad. Sci. USA, 98 (2001) 4877-4882.
72 Y. Itoh, K. Cai and H.G. Khorana, Proc. Natl. Acad. Sci. USA, 98 (2001) 4883-4887.
73 B.W. Boyd, S.R. Witowski and R.T. Kennedy, Anal. Chem., 72 (2000) 865-871.
74 H. Kamencic, A. Lyon, P.G. Paterson and B.H. Juurlink, Anal. Biochem., 286 (2000)
35-37.
75 S.H. Kang, W. Wei and E.S. Yeung, J. Chromatogr.B, Biomed. Sci. Appl., 744 (2000)
149-156.
76 E. Causse, C. Issac, P. Malatray, C. Bayle, P. Valdiguie, R. Salvayre and F. Couderc,
J. Chromatogr.A, 895 (2000) 173-178.
77 E. Causse, P. Malatray, R. Calaf, P. Charpiot, M. Candito, C. Bayle, P. Valdiguie, R.
Salvayre and F. Couderc, Electrophoresis,21 (2000) 2074-2079.
78 V. Ducros, M. Candito, E. Causse, R. Couderc, K. Demuth, M.E. Diop, J. Drai, A.M.
Gachon, I. Garcia, M.F. Gerhardt, C. Philippe-Bourgeois, M.H. Read and M.P.
Sauvant, Ann. Biol. Clin. (Paris), 59 (2001) 33-39.
79 N. Ercal, P. Yang and N. Aykin, J. Chromatogr.B, Biomed. Sci. Appl., 753 (2001)
287-292.
80 C. Cereser, J. Guichard, J. Drai, E. Bannier, I. Garcia, S. Boget, P. Parvaz and A.
Revol, J. Chromatogr. B, Biomed. Sci. Appl., 752 (2001) 123-132.
81 K.J. Lenton, H. Therriault and J.R. Wagner, Anal. Biochem., 274 (1999) 125-130.
82 G. Durand and N. Seta, Clin. Chem., 46 (2000) 795-805.
83 N. Volpi, Anal. Biochem., 277 (2000) 19-24.
84 C. Sato, S. Inoue, T. Matsuda and K. Kitajima, Anal. Biochem., 266 (1999) 102-109.
85 S.L. Lin, Y. Inoue and S. Inoue, Glycobiology, 9 (1999) 807-814.
86 J. Charlwood, H. Birrell, A. Gribble, V. Burdes, D. Tolson and P. Camilleri, Anal.
Chem., 72 (2000) 1453-1461.
87 H. Birrell, J. Charlwood, I. Lynch, S. North and P. Camilleri, Anal. Chem., 71 (1999)
102-108.

664
 88 J.A. Saba, X. Shen, J.C. Jamieson and H. Perreault, Rapid Commun. Mass
Spectrom., 13 (1999) 704-711.
89 M.D. Leavell and J.A. Leary, J. Am. Soc. Mass Spectrom., 12 (2001) 528-536.
90 S. Hanrahan, J. Charlwood, R. Tyldesley, J. Langridge, R. Bordoli, R. Bateman and
P. Camilleri, Rapid Commun. Mass Spectrom., 15 (2001) 1141-115.
91 J. Charlwood, J.M. Skehel and P. Camilleri, Anal. Biochem., 284 (2000) 49-59.
92 J. Charlwood, H. Birrell, D. Tolson, J. Connelly and P. Camilleri, Anal. Biochem., 283
(2000) 250-257.
93 Y. Araki, A. Andoh, Y. Fujiyama, K. Hata, J. Makino, T. Okuno, F. Nakanura and T.
Bamba, J. Chromatogr. B, Biomed. Sci. Appl., 753 (2001) 209-215.
94 K.R. Anumula and P. Du, Anal. Biochem., 275 (1999) 236-242.
95 J. Charlwood, J. Langridge, D. Tolson, H. Birrell and P. Camilleri, Rapid Commun.
Mass Spectrom., 13 (1999) 107-112.
96 J. Charlwood, J. Langridge and P. Camilleri, Rapid Commun. Mass Spectrom., 13
(1999) 1522-1530.
97 J. Charlwood, H. Birrell, A. Organ and P. Camilleri, Rapid Commun. Mass
Spectrom., 13 (1999) 716-723.
98 K.M. Hutterer, H. Birrell, P. Camilleri and J.W. Jorgenson, J. Chromatogr. B,
Biomed. Sci. Appl., 745 (2000) 365-372.
99 S.A. Visser, C.J. Smulders, W.W. Gladdines, H. Irth, P.H. van der Graaf and M.
Danhof, J. Chromatogr.B, Biomed. Sci. Appl., 745 (2000) 357-363.
100 D. Shangguan, H. Han, R. Zhao, Y. Zhao, S. Xiong and G. Liu, J. Chromatogr.A, 910
(2001) 367-372.
101 Z. Huang and D.J. Waxman, Anal. Biochem., 273 (1999) 117-125.
102 A. Paci, A. Rieutord, D. Guillaume, F. Traore, J. Ropenga, H.P. Husson and F. Brion,
J. Chromatogr.B, Biomed. Sci. Appl., 739 (2000) 239-246.
103 R.W. Sparidans, A. Veldkamp, B.M. Hoetelmans and J.H. Beijnen, J. Chromatogr.B,
Biomed. Sci. Appl., 736 (1999) 115-121.
104 H. Zhang, H. Ford, J.S. Roth and J.A. Kelley, J. Pharm. Biomed. Anal., 25 (2001)
285-297.
105 F. Dai, J.A. Kelley, H. Zhang, N. Malinowski, M.F. Kavlick, J. Lietzau, L. Welles, R.
Yarchoan and H. Ford Jr., Anal. Biochem., 288 (2001) 52-61.
106 K. Soda, Y. Ohba and K. Zaitsu, J. Chromatogr.B, Biomed. Sci. Appl., 752 (2001)
55-60.
107 B. Frey, R. Haupt, S. Alms, G. Holzmann, T. Konig, H. Kern, W. Kox, B. Rustow and
M. Schlame, J. Lipid Res., 41 (2000) 1145-1153.
108 M. Comesana Losada, J.M. Leao, A. Gago-Martinez, J.A. Rodriguez Vazquez and
M.A. Quilliam, J. Agric. Food Chem., 47 (1999) 618-621.
109 L. Ten-Hage, N. Delaunay, V. Pichon, A. Coute, S. Puiseux-Dao and J. Turquet,
Toxicon, 38 (2000) 1043-1054.
110 A. Hattori, T. Fukushima, K. Hamamura, M. Kato and K. Imai, Biomed.
Chromatogr., 15 (2001) 95-99.
111 A. Hattori, T. Fukushima and K. Imai, Anal. Biochem., 281 (2000) 209-215.
112 H. Yamada, Y. Kuwahara, Y. Takamatsu and T. Hayase, Biomed. Chromatogr., 14
(2000) 333-337.
113 B. Yagen and S. Burstein. J. Chromatogr.B, Biomed. Sci. Appl., 740 (2000) 93-99.
114 Y. Arai, T. Fukushima, M. Shirao, X. Yang and K. Imai, Biomed. Chromatogr., 14
(2000) 118-124.
115 P. Assaf, J. Katzhendler and A.I. Haj-Yehia, J. Chromatogr.A, 869 (2000) 243-250.
116 M. Sharma, Biochem. Biophys. Res. Commun., 273 (2000) 40-44.
117 C.Z. Matthews, E.J. Woolf, L. Lin, W. Fang, J. Hsieh, S. Ha, R. Simpson and B.K.
Matuszewski, J. Chromatogr. B, Biomed. Sci. Appl., 751 (2001) 237-246.

665
118 R. Gatti, M. Franchina, M.G. Gioia and V. Cavrini, Biomed. Chromatogr., 14 (2000)
82-88.
119 K. Hamase, T. Tomita, A. Kiyomizu and K. Zaitsu, Anal. Biochem., 279 (2000)
106-110.
120 S.N. Krylov, E.A. Arriaga, N.W. Chan, N.J. Dovichi and M.M. Palcic, Anal. Biochem.,
283 (2000) 133-135.
121 J.Y. Zhao, P. Diedrich, Y. Zhang, O. Hindsgaul and N.J. Dovichi, J. Chromatogr.B,
Biomed.Appl. 657 (1994) 307-313.
122 Y. Zhang, X. Le, N.J. Dovichi, C.A. Compston, M.M. Palcic, P. Diedrich and 0.
Hindsgaul, Anal. Biochem., 227 (1995) 368-376.
123 M.L. Miller and G.V. Johnson, J. Chromatogr. B, Biomed. Sci. Appl., 732 (1999)
65-72.
124 D.M. Albin, J.E. Wubben and V.M. Gabert, J. Agric. Food Chem., 48 (2000) 1684-
1691.
125 E. Zahou, H. Jornvall and T. Bergman, Anal. Biochem., 281 (2000) 115-122.
126 F. Wang, X. Chen, Q. Chen, X. Qin and Z. Li, J. Chromatogr.A, 883 (2000) 113-118.
127 N. Isoherranen and S. Soback, Clin. Chem., 46 (2000) 837-842.
128 C. Arcelloni, B. Comuzzi, R. Vaiani and R. Paroni, J. Chromatogr. B, Biomed. Sci.
Appl., 753 (2001) 151-156.
129 E. Kaale, S. Leonard, A. Van Schepdael, E. Roets and J. Hoogmartens, J.
Chromatogr.A, 895 (2000) 67-79.
130 J. Kirschbaum, K. Rebscher and H. Bruckner, J. Chromatogr.A, 881 (2000) 517-530.
131 R. Herraez-Hernandez, P. Campins-Falco and J. Verdu-Andres, Analyst, 126 (2001)
581-586.
132 D. Grosjean, E. Grosjean and A.W. Gertler, Environ. Sci. Technol., 35 (2001) 45-53.
133 J.J. Michels, J. Chromatogr. A, 914 (2001) 123-129.
134 G. Dusi and V. Gamba, J. Chromatogr.A, 835 (1999) 243-246.
135 M. Vogel, W. Potter and U. Karst, J. Chromatogr.A, 886 (2000) 303-307.
136 H. Miwa, J. Chromatogr. A, 881 (2000) 365-385.
137 Z. Glatz and H. Maslanova, J. Chromatogr. A, 895 (2000) 179-187.
138 R. Glowacki, K. Wojcik and E. Bald, J. Chromatogr.A, 914 (2001) 29-35.
139 M.H. Quernin, B. Poonkuzhali, Y. Medard, D. Dennison, A. Srivastava, R. Krishna-
moorthy, M. Chandy and E. Jacqz-Aigrain, J. Chromatogr. B, Biomed. Sci. Appl., 721
(1999) 147-152.
140 N. Bleyzac, P. Barou and G. Aulagner, J. Chromatogr.B, Biomed. Sci. Appl., 742
(2000) 427-432.
141 R. Raghavan, M. Burchett, D. Loffredo and J.A. Mulligan, Drug Dev.Ind.Pharm. 26
(2000) 429-440.
142 J. Wang, M.P. Chatrathi and B. Tian, Anal. Chem., 72 (2000) 5774-5778.
143 Y.V. Tcherkas, L.A. Kartsova and I.N. Krasnova, J. Chromatogr. A, 913 (2001)
303-308.
144 D.R. Knapp, Handbook of Analytical Derivatization Reactions. WileyInterscience,
John Wiley &Sons, Inc. 1979.
145 J.L. Little, J. Chromatogr.A, 844 (1999) 1-22.
146 M. Thevis, G. Opfermann, H. Schmickler and W. Schanzer, J. Mass Spectrom., 36
(2001) 159-168.
147 Y. $hinohara, K. Isurugi and T. Hashimoto, J. Chromatogr.B, Biomed. Sci. Appl.,
741 (2000) 271-278.
148 M. Thevis, G. Opfermann and W. Schanzer, J. Chromatogr. B, Biomed. Sci. Appl.,
744 (2000) 345-350.
149 Y. Kim, Y. Lee, M. Kim, Y.H. Yim and W. Lee, Rapid Commun.Mass Spectrom. 14
(2000) 1293-1300.

666
150 T. Ohie, X.W. Fu, M. Iga, M. Kimura and S. Yamaguchi, J. Chromatogr.B, Biomed.
Sci. Appl., 746 (2000) 63-73.
151 H.G. Mol, S. Sunarto and O.M. Steijger, J. Chromatogr. A, 879 (2000) 97-112.
152 C. Rodier, R. Sternberg, F. Raulin and C. Vidal-Madjar, J. Chromatogr. A, 915 (2001)
199-207.
153 U. Staerk and W.R. Kulpmann, J. Chromatogr. B, Biomed. Sci. Appl., 745 (2000)
399-411.
154 K.A. Harrison, S.S. Davies, G.K. Marathe, T. McIntyre, S. Prescott, K.M. Reddy, J.R.
Falck and R.C. Murphy, J. Mass Spectrom., 35 (2000) 224-236.
155 M.R. Anari, R.W. Burton, S. Gopaul and F.S. Abbott, J. Chromatogr. B, Biomed. Sci.
Appl., 742 (2000) 217-227.
156 Y.J. Yang, M.H. Choi, M.J. Paik, H.R. Yoon and B.C. Chung, J. Chromatogr. B,
Biomed. Sci. Appl., 742 (2000) 37-46.
157 M.H. Choi and B.C. Chung, Anal. Biochem., 277 (2000) 271-273.
158 X.P. Lee, T. Kumazawa, K. Kondo, K. Sato and 0. Suzuki, J. Chromatogr.B, Biomed.
Sci. Appl., 734 (1999) 155-162.
159 M.N. Sarrion, F.J. Santos and M.T. Galceran, J. Chromatogr. A, 859 (1999) 159-171.
160 M.N. Sarrion, F.J. Santos and M.T. Galceran, Anal. Chem., 72 (2000) 4865-4873.
161 M.I. Catalina, J. Dalluge, R.J. Vreuls and U.A. Brinkman, J. Chromatogr.A, 877
(2000) 153-166.
162 C.D. Stalikas and G.A. Pilidis, J. Chromatogr.A, 872 (2000) 215-225.
163 K.J. Berg, J.J. Boon, I. Pastorova and L.F. Spetter, J. Mass Spectrom., 35 (2000)
512-533.
164 J. Pietzsch, Biochem. Biophys. Res. Commun., 270 (2000) 852-857.
165 K.R. Kim and H. Kim, J. Chromatogr.A, 866 (2000) 87-96.
166 M.H. Choi, K.R. Kim and B.C. Chung, Analyst, 125 (2000) 711-714.
167 S. Blair, M. Song, B. Hall and J. Brodbelt, J. ForensicSci., 46 (2001) 688-693.
168 C. March, H.T. Karnes, A. McLean and P.S. Mukherjee, Biomed. Chromatogr., 15
(2001) 100-107.
169 C. Jurado, M.P. Gimenez, T. Soriano, M. Menendez and M. Repetto, J. Anal. Toxicol.,
24 (2000) 11-16.
170 M.R. Lee, Y.S. Song, B.H. Hwang and C.C. Chou, J. Chromatogr. A, 896 (2000)
265-273.
171 A. Namera, M. Yashiki, J. Liu, K. Okajima, K. Hara, T. Imamura and T. Kojima,
Forensic Sci. Int., 109 (2000) 215-223.
172 X. Xiao and D. McCalley, Rapid Commun. Mass Spectrom., 14 (2000) 1991-2001.
173 S.T. Weintraub, R.K. Satsangi, E.A. Sprague, T.J. Prihoda and R.N. Pinckard, J. Am.
Soc. Mass Spectrom., 11 (2000) 176-181.
174 C.D. Stalikas and C.N. Konidari, Anal. Biochem., 290 (2001) 108-115.
175 J.A. Koziel, J. Noah and J. Pawliszyn, Environ. Sci. Technol., 35 (2001) 1481-1486.
176 K.K. Ngim, S.E. Ebeler, M.E. Lew, D.G. Crosby and J.W. Wong, J. Agric. Food
Chem., 48 (2000) 3311-3316.
177 D.A. Cancilla and S.S. Que Hee, J. Chromatogr., 627 (1992) 1-16.
178 T. Matsukawa, H. Hasegawa, Y. Shinohara and T. Hashimoto, J. Chromatogr. B,
Biomed. Sci. Appl., 751 (2001) 213-220.
179 H. Hasegawa, T. Matsukawa, Y. Shinohara and T. Hashimoto, Drug Metab. Dispos.,
28 (2000) 920-924.
180 H. Hasegawa, T. Matsukawa, Y. Shinohara and T. Hashimoto, J. Chromatogr. B,
Biomed. Sci. Appl., 735 (1999) 141-149.
181 P.T. Fenn, C. Ning, S. Segal and I.A. Blair, J. Mass Spectrom., 35 (2000) 218-223.
182 C. Ning, P.T. Fenn, I.A. Blair, G.T. Berry and S. Segal, Mol. Genet. Metab., 70 (2000)
261-271.

667
183 B.E. van Dongen, S. Schouten and J.S. Damste, Rapid Commun. Mass Spectrom., 15
(2001) 496-500.
184 M.J. Paik, K.O. Lee and H.S. Shin, J. Chromatogr.B, Biomed. Sci. Appl., 721 (1999)
3-11.
185 H.S. Shin, C.H. Park, S.J. Park and H. Pyo, J. Chromatogr.A, 912 (2001) 119-125.
186 X. Chen, V.E. Anderson and Y.H. Chen, J. Am. Soc. Mass Spectrom., 10 (1999)
448-452.
187 K. Gartenmann and S. Kochhar, J. Agric. Food Chem., 47 (1999) 5068-5071.
188 G. Singh, A. Gutierrez, K. Xu and I.A. Blair, Anal. Chem., 72 (2000) 3007-3013.
189 S. Suzuki, R. Tanaka, K. Takada, N. Inoue, Y. Yashima, A. Honda and S. Honda, J.
Chromatogr.A, 910 (2001) 319-329.
190 J.A. Saba, X. Shen, J.C. Jamieson and H. Perreault, J. Mass Spectrom., 36 (2001)
563-574.
191 X. Shen and H. Perreault, J. Mass Spectrom., 34 (1999) 502-510.
192 A.S. Weiskopf, P. Vouros, J. Cunniff, E. Binderup, F. Bjorkling, L. Binderup, M.C.
White and G.H. Posner, J. Mass Spectrom., 36 (2001) 71-78.
193 T. Higashi, D. Awada and K. Shimada, Biomed. Chromatogr., 15 (2001) 133-140.
194 J.M. Quirke and G.J. Van Berkel, J. Mass Spectrom., 36 (2001) 179-187.
195 K.H. Kim, D.S. Kim, S.P. Hong and O.S. Keon, Arch. Pharm. Res., 23 (2000) 26-30.
196 X. Li, T.W. Yao and S. Zeng, J. Chromatogr. B, Biomed. Sci. Appl., 742 (2000)
433-439.
197 S. Kochhar, B. Mouratou and P. Christen, Meth. Mol. Biol., 159 (2000) 49-54.
198 G. Torok, A. Peter, A. Gaucher, M. Wakselman, J.P. Mazaleyrat and D.W.
Armstrong, J. Chromatogr. A, 846 (1999) 83-91.
199 A. Peter, E. Olajos, R. Casimir, D. Tourwe, Q.B. Broxterman, B. Kaptein and D.W.
Armstrong, J. Chromatogr. A, 871 (2000) 105-113
200 M. Kagawa, Y. Machida and H. Nishi, J. Chromatogr.A, 857 (1999) 127-135.
201 M. Peter, A. Peter and F. Fulop, J. Chromatogr.A, 871 (2000) 115-126.
202 I.J. Kim, S.J. Park and H.J. Kim, J. Chromatogr.A, 877 (2000) 217-223.
203 C.R. Craig and R.E. Stitzel, Modern Pharmacology. Little-Brown, 1994.
204 H. Li, C.J. Meininger and G. Wu, J. Chromatogr. B, Biomed. Sci. Appl., 746 (2000)
199-207.
205 S. Kage, K. Kudo and N. Ikeda, J. Chromatogr. B, Biomed. Sci. Appl., 438 (2000)
363-368.
206 D. Tsikas, F.M. Gutzki, J. Sandmann, E. Schwedhelm and J.C. Frolich, J. Chroma-
togr. B, Biomed. Sci. Appl., 731 (1999) 285-291.
207 D.W. Nierenberg, R.E. Nordgren, M.B. Chang, B. Morris, R.W. Siegler, M.B. Blayney,
F. Hochberg, T.Y. Toribara, E. Cernichiari and T. Clarkson, N. Eng. J. Med., 338 (23)
(1.998) 1672-1676.
208 C. Dietz, Y. Madrid, C. Camara and P. Quevauviller, Anal. Chem., 72 (2000) 4178-
4184.
209 Y. Cai, S. Monsalud, R. Jaffe and R.D. Jones, J. Chromatogr. A, 876 (2000) 147-155.
210 E. Millan and J. Pawliszyn, J. Chromatogr.A, 873 (2000) 63-71.
211 X. Yu and J. Pawliszyn, Anal. Chem., 72 (2000) 1788-1792.
212 Z. Mester, G. Horvath, G. Vitany, L. Lelik and P. Fodor, Rapid Commun. Mass
Spectrom., 13 (1999) 350-353.

668
Chapter20

Sampling and sample preparation for

pheromone analysis
Jocelyn G. Millar

20.1 INTRODUCTION

The remarkable biological activity of insect pheromones has been known for
over a century, since pioneering experiments in the late 19th century by J.H.
Fabr6 [1] and others demonstrated that male moths were attracted to
pheromone-producing females over distances of a kilometer or more. However,
identification of pheromones from insects and other organisms was hindered by
the small quantities that are produced, often in the nanogram or less range per
individual. Research on the first identification of an insect pheromone began in
earnest in the mid-1930s, and culminated 25 years later in the identification of a
few milligrams of (10E,12Z)-hexadecadienol as the sex attractant pheromone of
the silk moth Bombyx mori, from an extract of 500,000 virgin female moths [2].
The pheromone had remarkable biological activity, stimulating a behavioral
response in male moths at femtogram levels. Since that extraordinary effort, the
development of optimized techniques for collection, analysis, and bioassay of
pheromones, coupled with the quantum leaps in chromatographic and spectro-
scopic methods, has greatly simplified the process of pheromone identification.
Several thousand insect pheromones of various types are now known [3-7],
along with a number of pheromones from other organisms. Identifications of
new pheromones have also been aided by biosynthetic parsimony, in that related
organisms tend to produce pheromone components with similar or identical
structures. In optimal cases, it has been possible to identify a complete phero-
mone blend from less than ten individuals. However, this is usually the excep-
tion rather than the rule. In general, the identification of pheromones remains
challenging, particularly if the bioactive compound(s) are new to science,
because of the very small quantities that can be collected.
It is also instructive to consider the differences between extraction of
analytes from samples such as water or tissue, and collection of pheromone sam-
ples. Although pheromones have been collected numerous times by whole-body
extraction or by extraction of specific tissues, this may be less useful than
dynamic collection of pheromones from living organisms, for several reasons.
ComprehensiveAnalytical ChemistryXXXVII
J. Pawliszyn (Ed.)
C 2002 Elsevier Science B.V. All rights reserved 669
First, although some pheromones are stored in glands or reservoirs, others are
biosynthesized and released as they are needed. Consequently, far more phero-
mone may be collected by continuous dynamic headspace sampling for periods of
minutes to days or longer than would be collected by one-time destructive
extraction of the same number of organisms. Second, headspace collections pro-
vide much cleaner and simpler extracts containing only the volatiles produced by
the test organisms, and are easier to analyze than whole-body extracts. Third,
because some pheromones are biosynthesized at the moment of release, extrac-
tion of tissues that contain both pheromones and pheromone precursors may
result in a chemical profile that is markedly different from that of the correct
pheromone blend.
Furthermore, the difference between developing effective sampling methods
for compounds whose structures are unknown, versus sampling compounds
whose structures, physical properties, and chemical properties are known, must
be borne in mind. That is, although some generalizations can be made with
regard to the chemical and physical properties of different types of pheromones
or related attractants, most sampling methods are biased. For example, acti-
vated charcoal or synthetic polymers effectively trap hydrophobic, semivolatile
compounds, but they are ineffective or even useless at trapping small polar
molecules such as ammonia, formic acid, acetic acid, and ethanol, all of which are
known insect semiochemicals.
Problems with identification of trace amounts of compounds can be further
compounded if the structures are unstable. Identifications of several insect
pheromones were hindered by the instability of the bioactive compounds to gas
chromatography and other typical analytical chemistry methods. For example,
the sex attractant pheromones of the tobacco hornworm moth Manduca sexta
((10E,12E,14Z)- and (0E,12E,14E)-hexadeca-10,12,14-trienals; [8]), the tuss-
ock moth Orgyia pseudotsugata ((6Z,8E)-heneicosadien-11-one [9]), and the
stink bug Thyantapallidovirens(methyl (2E,4Z,6Z)-decatrienoate) [101 were all
partially or totally destroyed during attempted analysis by gas chromatography.
This chapter will review the methods that have been developed for the
collection and isolation of insect pheromones and related compounds. Where
appropriate, examples from other taxa will be included. In practice, the appara-
tus and techniques for collection of trace quantities of pheromones, with adjust-
ments for scale, are dictated more by the chemical and physical properties of the
pheromones, and the physical circumstances required for their production or
extraction, than by the particular organism. For example, the same general
sampling techniques have been used to sample pheromones from organisms as
small as single-celled gametes to as large as Asian elephants [11,12]!

20.2 TYPES OF PHEROMONES

Pheromones represent a subset of the naturally occurring chemical signals

termed "semiochemicals". The term "pheromone" is defined as a chemical signal

670
produced by one organism for communication with another organism of the
same species. Within this broad definition, pheromones are further classified
into different categories, depending on their function. Several recent reviews
and databases provide compilations of known insect pheromones [3-7].
Some of the more common types of pheromones are briefly described below,
along with their key characteristics and selected examples.

20.2.1 Sex pheromones

The volatile sex attractant pheromones that many insects produce to attract
mates are probably the most widely known type of pheromones [7]. These
pheromones are produced by one sex, and responded to by the opposite sex.
Males rather than females are frequently the attractive sex, for various
biological and ecological reasons that are beyond the scope of this chapter to
explore. Sex attractant pheromones also are used by many other animals, and
even by single-celled organisms such as yeasts. The pheromone properties vary
with the producing organism, and with the medium in which the signal is
transmitted. Thus, terrestrial organisms whose signals are airborne produce sex
attractants that are volatile or semivolatile, which restricts the molecular
weights to about 350 daltons or less, and also limits the number of functional
groups. In contrast, aquatic and marine organisms may produce polar or nonvol-
atile sex pheromones whose properties may be limited more by constraints on
solubility than on volatility and molecular size. However, this is not always the
case; for example, sex pheromones of brown algae are small hydrophobic polyun-
saturated hydrocarbons [13-17].

20.2.2 Recognition pheromones

Many insects have sex and species recognition pheromones that act only at close
range or on contact [4,5]. These pheromones are used to confirm that an
individual is of the correct species and sex before mating attempts progress.
These types of pheromones are also used by social insects such as ants, termites,
and honeybees to distinguish nestmates from non-nestmates, and to distinguish
caste status. The pheromones tend to be simple straight- or branched-chain
hydrocarbons of limited volatility, with specificity usually being conferred by a
particular blend of compounds rather than the presence or absence of a unique
compound.

20.2.3 Marking pheromones

Marking pheromones are used, for example, by parasitic insects to mark hosts in
which they have laid an egg [18]. The marks deter oviposition by conspecifics,
limiting competition for the resource. The marks also may prevent the same
female from superparasitizing a host. Scent marks are also ubiquitous in higher

671
animals, being used to define territory and to display the dominance status of
individuals.

20.2.4 Alarm pheromones

Alarm pheromones alert conspecifics to danger, and are widely used by insects
that live in groups or colonies, such as ants, bees, and aphids [4,5]. The alarm
pheromone may cause receivers to move away from the signal or drop off the
plant (e.g. aphids), or the pheromone may recruit nestmates to defend the nest,
such as occurs with Africanized honeybees. In some cases, these pheromones
may also have a secondary role as defensive compounds. For example, many true
bugs (Hemiptera) produce copious quantities of volatile defensive secretions,
and in addition to deterring predators, the compounds also cause conspecifics to
move away [19]. These pheromones tend to be small volatile compounds,
allowing rapid dispersion of the signal.

20.2.5 Trail pheromones

Trail pheromones are used by organisms such as ants to mark trails to food
sources [20]. Because the food sources are usually ephemeral rather than
permanent, ant trail pheromones tend to be semivolatile, and must be continu-
ally reinforced in order for the trail to persist. Trail pheromones are also used by
processionary insects such as tent caterpillars and bagworms, to establish trails
between the protective tent and food sources [21].

20.2.6 Aggregation pheromones

Aggregation pheromones, as distinct from sex pheromones that attract only one
sex, are produced by one (or both) sex(es), and attract both sexes. They are used
by insects such as bark beetles to assemble the critical mass of individuals
needed to mass attack and overwhelm a host tree's defenses [22]. Aggregation
pheromones are also used by some beetle species that infest grain and related
stored products [23], and by first instar stink bugs, causing the nymphs to
cluster together [24].

20.2.7 Other pheromones

There is a wide variety of other types of pheromones associated with all aspects
of reproduction. For example, primer pheromones induce physiological changes
such as egg maturation and ovulation, or the inhibition of reproduction by
worker castes in bees [25]. Aphrodisiac pheromones cause sexual stimulation
and excitation in males, or render females compliant to mating attempts
[26-28]. Releaser pheromones trigger the release of specific behaviors, such as
gregariousness in locusts [29], or the release of sperm or eggs into the water by

672
aquatic organisms [30]. Other pheromones transferred during mating reduce
female receptivity to further mating attempts, block further sex attractant
pheromone production by females, and stimulate oviposition [31].
Predators and parasites frequently exploit the pheromones of their hosts.
This exploitation may be eavesdropping, using the pheromones as chemical
beacons to locate hosts [32,33]. Alternatively, exploiting organisms may actively
mimic the pheromones of their hosts. For example, bolas spiders lure male
moths towards their sticky threads by producing the female sex attractant
pheromone for that moth species [34]. Furthermore, the spiders appear to be
able to change the blend that they produce to match the flight times and periods
of specific prey species.

20.3 PHEROMONES IN BIOLOGICAL CONTEXTS

An example may provide insight into the variety and importance of pheromones
in the biology and ecology of organisms. Consider an ant nest or a honeybee hive,
consisting of a queen, males, and numerous female workers of different castes,
frequently totalling thousands or even millions of individuals. The structure,
organization, and functioning of these immense and complicated societies is
largely controlled by the exchange of chemical signals between individuals, with
an intricate vocabulary of chemical "words" providing the "language" of
communication that enables the colony to function.
This also illustrates another important point: when planning to sample and
identify a pheromone, a detailed knowledge of the biological function of the
pheromone is a vital prerequisite. Many behaviors, such as mating or host
location, actually consist of a sequential series of behavioral steps. In some cases,
one pheromone may trigger the whole behavioral cascade that culminates in
copulation, that is, activation of a resting individual, long range orientation and
directed movement towards a potential mate, arrestment, assessment and rec-
ognition, sexual stimulation, and finally copulation. For example, males of some
beetle and moth species, when stimulated with a single female pheromone, will
fly upwind, land, and, in a sexual frenzy, attempt to mate with the pheromone
source, even with such unnatural sources as the end of a glass rod or a pencil
treated with pheromone [5,6]. With other organisms, different steps in the
mating behaviors are mediated by different chemical signals, and if the full
repertoire of chemicals is not present, copulation will not occur. For example,
males of many species of ticks are attracted by 2,6-dichlorophenol [35], but in the
absence of other chemical signals, copulation will not occur. Thus, it is crucial
both to know the biology of the study organism thoroughly, and to develop a
bioassay that tests for the specific behavior mediated by a particular pheromone.
An example may be illustrative. When first perceiving one or more components
of the female sex pheromone blend, male moths begin wing-fanning, and this
behavior has been exploited in simple bioassays for identification of moth sex
pheromones. However, this is only the first step in the behavioral cascade that

673
ends in a male moth copulating with a female, and wing-fanning can be triggered
by incomplete or incorrect pheromone blends that will not result in a male moth
flying upwind, or even taking flight. More sophisticated bioassays that test the
whole series of behavioral steps, such as flight tests in wind tunnels, are required
in order to identify the complete pheromone blend [36].
Control and knowledge of the physiological state of the test organisms is also
crucial because many pheromones are produced only in specific contexts. For
example, sex pheromones are produced only by sexually mature individuals that
are in a reproductively receptive state. It is pointless to try to collect pheromones
from immature individuals, or from organisms that are otherwise in non-
reproductive states, such as insects that are in seasonally induced reproductive
diapause. Production of sex (and other) pheromones also may occur only during
certain specific periods of the day, with production usually being linked to daily
light-dark cycles. Consequently, when collecting pheromones from live organ-
isms, light, temperature, and other physical conditions must be correct in order
for the desired pheromone to be produced. Furthermore, female insects usually
cease production of sex attractant pheromones after mating, either perma-
nently, or fora refractory period while some eggs are laid. Thus, for collection of
sex pheromones, it is prudent to use insects whose reproductive history is
known, and if possible, to use virgin insects. Analogous considerations apply to
other types of pheromones. In short, insects must be provided with a facsimile of
the conditions under which they would normally perform a specific behavior
associated with a pheromone in order for that pheromone to be produced.
Although the volatile sex pheromones are most well known, it must be
remembered that the physical and chemical properties of a pheromone are
suited to its function, the medium through which it is transmitted, and the
target receptor organs. Thus, sex attractant pheromones for terrestrial organ-
isms, whose function is attraction of mates from a distance, tend to be relatively
small molecules. Conversely, pheromones that act on contact, such as sex and
species recognition pheromones, tend to be larger and of limited volatility.
Consequently, sampling methods must take into account the likely properties of
the particular pheromones being sampled.

20.4 METHODS FOR SAMPLING AND COLLECTING PHEROMONES

Techniques used in the study of pheromones, including methods of collection,

isolation, and identification of the bioactive compounds, have been the subject of
several books and reviews [,37-41]. A useful compilation of methods used to
sample and collect insect pheromones up to 1984 can be found in Golub and
Weatherston [42]. Consequently, I will encapsulate the general methods used in
the majority of the studies discussed in the books and chapters mentioned above,
and then review more recently developed methods, such as solid phase micro-
extraction (SPME), in greater detail. The main advantages and limitations of
each method are summarized in Table 20.1, for easy comparison.

674
TABLE 20.1
Advantages and limitations of various methods of sampling and collecting pheromones
Method Advantages Limitations
Whole body Extracts most material present; can Active compounds are buried in a complex
extraction provide large samples for background of extraneous material; extensive
fractionation and bioassay cleanup and fractionation may be required
Extract may not be representative of profile of
compounds actually released by living organ-
ism
Organisms are destroyed; can only sample
them once
Enzyme action during tissue maceration may
alter profile of extractables
Amounts and types of compounds extracted bi-
ased by choice of solvent
Need to remove relatively large amounts of sol-
vent without losing or degrading bioactive com-
pound(s)
Solvent purity is critical

Steam Extracts most material present; can
distillation provide large samples for fraction-
ation and bioassay
Separates volatiles from nonvolatile
materials; extracts can go straight
on GC
Sensitive compounds destroyed during extrac-
tion
Extract may not be representative of profile of
compounds actually released by living organ-
ism
Need to extract sample from aqueous distillate
Organisms are destroyed; can only sample
them once

Headspace Not usually appropriate for use in chemical

sampling ecology due to lack of sensitivity

Thermal Solventless method Sample may not be representative of blend pro-

desorption duced by living organism
and purge Sampling method only; not suitable for bulk
and trap collections
Thermal degradation of sensitive compounds
Purge and trap adsorbents biased in their re-
tention of different classes of compounds

Trapping Simple Small-scale short-term sampling method only

on glass Samples recovered in small volume Suitable only for semi-volatile hydrophobic
of solvent compounds
Closed loop Similar to dynamic headspace Similar to dynamic headspace sampling (below)
stripping sampling (below) Memory" effects and difficulty of completely
Recirculation of air minimizes cleaning apparatus between runs
buildup of contaminants within Limited in size by necessity for closed system
one run

continued

675
TABLE 20.1 (continuation)

Method Advantages Limitations

Dynamic Collect continuously or repeatedly Biased in favor of hydrophobic semivolatile
headspace from same group of living organisms compounds. Not good for very volatile or small
sampling/ for periods of minutes-weeks polar compounds
collection on Sample is representative of the Volatile components may be masked by solvent
Porapak or blend of compounds actually Losses during extraction and concentration
Tenax released by the test organism
Extracts contain only volatile
materials; can be injected directly
onto GC with no cleanup steps
required
Extracts much simpler than whole
body or steam distillation extracts;
higher ratio of desired compound(s)
to extraneous materials
Extracts recovered in relatively
small volume of solvent; minimal
concentration required
Simple and straightforward; adapt-
able for microscale sampling to bulk
collections including field collections
Can be used with thermally sensitive
compounds

Solid phase Collect continuously or repeatedly
micro- from same group of living organisms
extraction for periods of minutes-weeks
(SPME) Sample is representative of the
blend of compounds actually released
by the test organism
Different fibers optimized for reten-
tion of different classes of analytes
Simple and inexpensive; fibers can
be reused many times
Solvent-free: volatile components
not masked by solvent
Essentially quantitative desorbtion
of sample
Loaded fibers can be shipped between
labs
Sampling method only; cannot be used for bulk
collections for fractionation or bioassays.
Each collection provides only a single sample;
sample cannot be split easily, eg., for analysis
and bioassays
Fibers are biased in their retention of different
chemical classes and also retain only a small
fraction of very volatile compounds
Cannot be used with thermally sensitive com-
pounds

Cryogenic Traps all components including Traps water; need to isolate active compounds
trapping CO2 and other permanent gases from water
Equipment may be cumbersome Safety concerns due to plugging of apparatus
with ice; pressure buildup on warming

676
20.4.1 Whole body and pheromone gland extracts

Historically, pheromones have been extracted using methods similar to those

used for extraction of other volatile and semivolatile analytes, such as solvent
extraction or steam distillation. These methods provide relatively large quanti-
ties of material, and extraction is relatively complete. However, there are
significant disadvantages. First, the large amounts of extraneous materials that
are coextracted necessitate laborious bioassay-driven fractionation to isolate the
biologically active materials from the complex background. The problem may be
compounded by the similarities in properties between the pheromones and other
components of the extracts. For example, in the isolation of the first insect
pheromone, bombykol, 500,000 pheromone glands were extracted, producing
280 g of crude extract, from which a few milligrams of pure pheromone were
eventually isolated [2].
Second, the profile of chemicals in the extract may bear little relationship to
the profile of chemicals that are actually released by the intact living organism
[43-47]. In particular, gland extracts may be rich in pheromone precursors,
which may actually antagonize behavioral responses. Analogous problems have
been noted when comparing the profile of chemicals obtained by solvent or
steam distillation of plant materials versus the profile obtained from collections
from the headspace above the intact plants, with the steam distillates and
solvent extracts producing qualitatively and quantitatively different profiles
than headspace collections. Furthermore, enzymes released by tissue macera-
tion can rapidly change the chemical profile. For example, the "green leaf
volatiles" Z3-hexenal, Z3-hexenol, and Z3-hexenyl acetate are produced by oxi-
dation of fatty acids in disrupted plant tissues [48].
Finally, solvents may extract compounds that are sequestered by the intact
organism, such as defensive compounds. These compounds can mask the activity
of the compounds under study, and make bioassays difficult to interpret.
Because the defensive compounds are present in much larger amounts than the
pheromone, defensive compounds also may obscure the pheromone components
in chromatograms.
Whole body extraction is still used occasionally, particularly in cases where
the site of pheromone production is unclear, or the pheromone is produced from
diffuse tissues rather than a discrete macroscopic gland. Extraction of dissected
pheromone glands is still widely used, particularly in combination with tech-
niques such as coupled gas chromatography-electroantennogram detection that
can be used to pinpoint trace amounts of pheromone compounds in a complex
background of extracted lipids [37,38]. In addition, pheromones or related
semiochemicals may be extracted from insect excrement (frass) [23,49-52] or
insect exuvia (the cast skins) [24].
Whole insects, dissected glands, or tissues are normally soaked for periods of
a few seconds to days in solvents chosen for their volatility, for ease of removal
without losing the volatile pheromones. Other considerations for choosing a

677
solvent include the ease of purification and the selective solvating power for the
compounds of interest, which are normally of limited solubility or insoluble in
water. Thus, solvents such as pentane, hexane, methylene chloride, and diethyl
ether have been used extensively in extraction of pheromone-containing tissues.
It is critically important to use appropriate sized apparatus and glassware for
working with the small quantities of pheromone extracts that are normally
available. The fabrication and use of microscale glassware and other apparatus
is described in Millar and Haynes [38,53]. It is pointless to extract nanogram
quantities of volatile pheromones using tens or hundreds of milliliters of solvent
in the glassware normally used in organic chemistry laboratories because of
problems with contamination, loss of the bioactive material during concentra-
tion, or adsorbtion on the surfaces of glassware. The number of handling steps
also must be minimized to reduce losses and contamination. To minimize losses
during concentration, the bulk of the solvent can be removed with a fractional
distillation column [50]. For smaller quantities, solvent is removed by passive
evaporation from an open vial, or is blown off with a gentle stream of nitrogen.
Often a single cleanup step, or no cleanup step at all, is used before going directly
to high resolution chromatographic methods.

20.4.2 Continuous collection of pheromones

Because of drawbacks with solvent extraction, headspace sampling methods

have been used widely with volatile pheromones. The test organisms are usually
held in an "aeration chamber" to minimize contamination and strictly control
conditions, but field collections are also possible (see below). Collection of
entrained volatiles from the headspace above live, healthy organisms perform-
ing natural behaviors has numerous advantages. First and foremost, it provides
an accurate representation of the profile of volatiles actually released by an
organism. Second, aeration extracts are less complex than solvent extracts
because they contain only volatile components, and so they can be analyzed
directly by GC or GC-MS without clean-up steps. The pheromone components
also represent a much larger percentage of the total extract than with solvent
extracts, making it easier to locate and identify the pheromone components. Aer-
ation extracts also are recovered in small volumes of solvent, simplifying concen-
tration. Third, as long as organisms are provided with food and water, aerations
can be continued indefinitely, with periodic changes of the collectors. This allows
for collection of the microgram to milligram quantities that are usually required
for the spectroscopic identification of an unknown pheromone. Furthermore,
the dynamics of production of semiochemicals, either in relation to the age of the
organisms or in relation to the diurnal rhythms of production, are easily deter-
mined by changing the collectors at specified intervals. Finally, because head-
space collection methods are nondestructive, the same individuals can be
sampled repeatedly. This is critically important when the numbers of individu-

678
als are limited, or when following changes in chemical profiles during develop-
ment, for example, during sexual maturation, or before and after mating. Even
when the number of individuals is not limiting, it is usually more economical to
sample the same individuals several times than to use them once in a single sol-
vent extraction.
Several points should be considered when choosing methods for collection of
volatile pheromones. First, the amounts of material obtained are small, being a
function of their rate of release and volatility, and the flow rate in the aeration
setup. Normally, nanograms to a milligram or so of material are obtained per
aeration cycle. If only small quantities of pheromone are required, for example
for quantification of a known chemical, then straightforward sampling methods
using simple aeration chambers are usually satisfactory. However, if larger
amounts of material are required for fractionations and bioassays, or NMR
analysis, then repeated collections may be required.
The characteristics of the bioactive pheromone blend must also be consid-
ered. If the blend consists of nonpolar to intermediate polarity components with
approximately similar volatilities, sampling or preparative-scale collections can
be done using well established methods (see below). However, if the blend
consists of components with widely differing properties, then the collection of
samples that accurately reflect the blend being released becomes more difficult.
For example, adsorbents such as Porapak Q, Tenax, or activated charcoal are
satisfactory for collecting low to medium polarity chemicals of intermediate size
and volatility, but they are poor to useless for trapping low molecular weight
and/or polar compounds such as ethanol, ammonia, and short-chain amines.
This problem is discussed in more detail below. The fact that the blend of
compounds trapped by most sampling and collection methods is quantitatively
and qualitatively biased has been largely ignored. In cases where pheromone
blends have proven to consist of hydrophobic molecules with molecular weights
in the 150-350 dalton range, this bias is not important. However, in other cases,
such as the attraction of mosquitoes to hosts, in which the attractant blend may
consist of components as diverse as carbon dioxide, lactic acid, and more hydro-
phobic alcohols and phenols, this sampling bias may be critically important. It is
undoubtedly responsible in part for slow progress in other areas of research,
such as the attraction of insects to host plants, in which complex blends of
chemicals with widely differing properties are involved.
Furthermore, volatile semiochemicals may be hidden under solvent peaks
during GC analyses of extracts prepared by elution of trapping media. It may be
possible to circumvent this problem by solventless thermal desorbtion of the
trapped volatiles, but purge-and-trap type thermal desorbtion is also not with-
out problems, including degradation of fragile compounds and production of
artifacts from the trapping media, as discussed below.
In the discussion that follows, where appropriate, I will draw distinctions
between methods that are appropriate only for sampling, and methods that are
appropriate for both sampling and preparative scale collection of pheromones.

679
20.4.2.1 Apparatusfor continuous collections
Apparatus for collection of volatiles in the headspace above live insects or other
organisms should be made from glass, metal, Teflon, polycarbonate, or Plexiglas
[38,50]. Other plastics, rubber, glues and adhesives, wood, and other building
materials should not be used because they bleed and/or retain volatiles. Parts
must be thoroughly cleaned and where possible, oven-dried at -125°C before
use. Connections are made with ball and socket ground-glass joints that allow
some motion. Tubes are connected with brass Swagelock unions with Teflon
ferrules. Connections through solid surfaces such as screw-cap lids are made by
drilling holes to accommodate Swagelock bulk-head unions. For small systems,
Teflon tubing connected to syringe needles with Teflon heat-shrink tubing is
used to connect the apparatus to air or vacuum lines. Rubber or plastic tubing
should only be used downstream from the collector. Apparatus of appropriate
sizes usually can be built from commonly available jars or containers. For large
numbers of aerations on a continuing basis, custom-built apparatus may be
worthwhile.
During aerations, all air entering the aeration system is precleaned by
passage through a charcoal filter, readily constructed using the granulated
charcoal used in aquariums. The airstream is usually humidified to prevent
desiccation of the organisms by bubbling the air through a water-filled gas-
washing bottle or passing it through a tube loosely packed with water-soaked
glass wool, placed in front of the charcoal scrubber. Nitrogen can be used if
anaerobic conditions are required, bearing in mind that volatiles produced
under aerobic and anaerobic conditions may differ [54]. Appropriate air sources
include purified compressed air (e.g., so-called "medical air"), or ambient air
supplied by an oil-free diaphragm pump. Alternatively, air can be pulled through
the system by vacuum; in "push-pull" systems, both pressure and vacuum are
used [47,55]. The aeration system must be neither overpressurized nor held
under partial vacuum, particularly when collecting volatiles from live
organisms. Either case may occur due to flow constrictions from the charcoal air
scrubber on the inlet side, or the volatiles collection trap on the outlet end.
More recently, "open" systems for the collection of volatiles from larger
organisms such as intact plants have found widespread use [47,56]. With this
method, a large volume of clean air is passed over a plant in an open-ended sleeve
or tube, with a small portion of the airstream being sampled.
Aeration chambers may be very small (e.g., multiple simultaneous aerations
in small vials [57]) to quite large. Small chambers are usually glass, whereas
intermediate to large scale aeration chambers are fabricated from glass, poly-
carbonate, metal, or Plexiglass. Apparatus such as wide-mouthed containers
with ground-glass joints (e.g., vacuum desiccators, wide-mouthed Erlenmeyer
flasks, etc.) or Teflon-lined screw-on lids is readily adapted for aerations. Cus-
tom-made cylindrical flow-through chambers which open in the middle or at one
end, with a securely clamped, O-ring sealed middle joint, and ground-glass ball
and socket joints at either end for connections are also excellent. Insects often

680
like to have something to climb or perch on, so clean metal screen or analogous
substrates can be inserted.
Larger chambers can be constructed from steam-cleaned, solvent-washed 55
gall drums [58,59], modifying the pouring hole to accept a charcoal filter. The
chamber is loaded through a large hole cut in the bottom, which is then closed
with a bolted-on metal plate and a Teflon gasket, with an outlet for the volatiles
collector.
More recently, sophisticated setups for the aeration of intact plants have
been developed, incorporating techniques for purification of large volumes of
inlet air, and computer controlled sequential sampling and monitoring of envi-
ronmental conditions [47,55,60]. These systems are commercially available in
custom designs (Analytical Research Systems, Inc., Gainesville FL; website
www.ars-fla.com). Related systems also have been described [61,62].
It is also possible to sample from living organisms in the field. For smaller
scale sampling, polyvinyl fluoride [63] or polyacetate cooking bags [64,65] were
used to bag flower heads for volatiles collection, polythene terephthalate bags
were used to bag fungal fruiting bodies [66], and Tedlar® bags were used in
sampling pine needle volatiles [67]. Larger apparatus can be constructed from
aluminum [58] or polyvinyl fluoride sheets. For example, we sampled volatiles
from trunks of living Eucalyptus trees by "bagging" sections with Tedlar®
fabric (Millar et al., unpublished data). Because these large bags cannot be
completely sealed, a push-pull system is advisable, with clean air pumped in, and
only a portion of the output being sampled so that the system remains under a
slight positive pressure.
Field sampling of odor sources can also be carried out in the open air, for
example, with flower odors [68]. Battery-operated personal air sampling pumps
with flows of several 1/min, such as those used for air sampling in industrial
settings, are ideal for these applications. Field sampling can be important
because odors released by organisms in their natural habitats may differ from
those produced under less natural conditions. However, background volatiles
may be problematic.
During the identification of novel semiochemicals, significant quantities are
needed for bioassay, fractionation, and identification of the active principles,
necessitating collection of relatively large samples from groups of insects. For
these bulk collections, headspace volatiles are usually collected on a bed of
granulated adsorbent (-40-80 mesh, to compromise between high surface area
and resistance to air flow), held in place with glass wool plugs in a glass or metal
tube. Porapak Q (and its refined version, Super Q; Waters, Supelco), Tenax GC
(refined version, Tenax TA; Chrompack, Alltech), and activated charcoal are the
most commonly used adsorbents. Porapak and Tenax are hydrophobic polymeric
resins with a high affinity for lipophyllic organic compounds. Tenax is more
thermally stable than Porapak, but this may not be an issue if traps are not
thermally desorbed. For most applications, either adsorbent can be used inter-
changeably [69]. Porapak and Tenax adsorbents need conditioning before use,

681
typically by heating under a slow flow of inert gas overnight (Porapak, 200°C,
Tenax 270-340°C for 2-24 h; [70-72]), followed by Soxhlet extraction with
pentane, methylene chloride, or ether [50,73,74]. Once conditioned, both adsorb-
ents require storage in sealed containers protected from light. Artefacts from
both types of adsorbents have been tabulated (Porapak [72,75,76]; Tenax [71,77,
78]). To control for artefacts, a system blank (i.e., aeration of an empty setup)
should always be run.
Volatiles are recovered from either adsorbent by elution (5-10 ml/g
adsorbent) or Soxhlet extraction with low-boiling solvents such as pentane or
methylene chloride, or for polar compounds, carbon disulfide. Small extracts are
concentrated by passive evaporation from open vials, or blown down under a
stream of clean N2 , whereas larger volumes of solvent are removed by fractional
distillation. Eluted sample tubes can be reconditioned by further elution with
clean solvent, while Soxhlet-extracted adsorbents are usually ready for reuse
after a second extraction with solvent. Residual solvent vapor is removed from
the cleaned resin under a stream of N2 .
The volume of adsorbent required for a particular application depends on the
polarity and volatility of analytes, flow rate, and volume of air to be sampled.
Because collections usually are made under ambient conditions, temperature is
not an issue. Highly volatile and/or more hydrophyllic compounds are not
retained well and will break through traps rapidly. Breakthrough volumes per
gram of adsorbent for analytes of different classes have been tabulated [79,80]
and are available from adsorbent suppliers (e.g. Scientific Instrument Services).
Practically, the easiest way to check for breakthrough is by putting two traps in
series, checking both for the analytes of interest. For most applications with
semivolatile pheromones, volatiles from tens to hundreds of liters of air can be
trapped on a relatively small bed of adsorbent (a few milligrams to a gram or so,
-5-50 mm in length), remembering that longer beds offer greater resistance to
air flow. As a general rule, the quantity of adsorbent is kept small to minimize
artifacts and the quantity of solvent required to elute the analytes.
Activated coconut charcoal is also an excellent adsorbent, with several
desirable characteristics. It is highly retentive with a high capacity, which
minimizes the quantity of adsorbent needed. It is easily cleaned in bulk by
heating under a clean N2 flow overnight at 250C, after which further solvent
cleanup is rarely necessary. Other researchers have recommended conditioning
with exhaustive solvent stripping with, e.g., hot EtOH, CH 2C12 , and CS 2 [81], or
CH 2 Cl2 and pentane [82]. Charcoal traps can be eluted with pentane, methylene
chloride, or CS2 (-5-20 volumes). Unlike the polymeric resins, charcoal is cheap,
so it can be discarded after use to eliminate the chance of cross-contamination.
Alternatively, small reusable traps have been fabricated by permanently sealing
a few milligrams of charcoal between stainless steel screens in glass tubes
[82-85]. These are commercially available (Brechbfihler AG, Urdorf Switzer-
land; http://www.brechbuehler.ch/).

682
 Until the advent of SPME, trapping of insect pheromones on Porapak,
Tenax, or charcoal was the method of choice for both sampling and bulk
collection. SPME is now replacing these methods to some extent in sampling
applications, but SPME is not suitable for bulk collections over extended
periods, which will continue to rely on methods based on these adsorbents.

20.4.3 Solid phase micro-extraction

This recently developed sampling technique [86] usually involves trapping

volatiles on adsorbent-coated fibers, followed by thermal desorbtion by insertion
of the fiber directly into a GC injector. The method has been rapidly adopted for
semichemical research. The apparatus and techniques are simple, consisting of a
fused silica fiber coated with various adsorbents, mounted on a modified GC
syringe. In its simplest form, the materials to be sampled (live insects, plants,
aqueous solutions, etc.) are held in a closed container. The needle of the device is
inserted into the container through a septum, and the SPME fiber is extended to
adsorb volatiles in the headspace. After an adsorbtion period of minutes to half
an hour or more, the fiber is withdrawn into the needle, the needle is inserted
through the GC septum, and the loaded fiber is extended into the hot injector for
thermal desorbtion of the adsorbed volatiles. Alternatively, volatiles from flow-
through systems can be sampled by inserting the SPME fiber into the airstream
exit. For aqueous samples, such as mosquito oviposition waters, the needle can
be inserted directly into the solution. However, both volatiles and nonvolatiles
will be adsorbed from solutions, so aqueous samples containing a lot of organic
matter may foul the fiber with nondesorbable contaminants fairly rapidly.
Each fiber is reusable for many adsorbtion cycles, and SPME fibers of
different polarities (e.g. polydimethylsiloxane, polyacrylate, Carbowax/divinyl-
benzene, CarboxenT'-polydimethylsiloxane) and thicknesses (7-100 um) are
available (Supelco, Bellefonte PA). These different fibers have been developed to
allow efficient sampling of compounds of differing polarities and volatilities.
New 100 Am polydimethylsiloxane fibers are efficiently cleaned by desorbtion at
200°C for an hour or more in a clean inert gas stream, for example by inserting
the fiber into a GC split-splitless injector with the outlet blocked and the purge
valve left open [87]. Cleaning of other fiber types is described in the product
literature.
Several useful studies have delineated the characteristics of the SPME
method, and defined the operational parameters for use in semiochemical
research [87-91]. First, to obtain reproducible quantitative results in closed
systems, the fiber must be exposed to headspace volatiles long enough to reach
equilibrium, with equilibration time inversely proportional to volatility. Second,
retention of analytes depends on molecular size and polarity. For nonpolar
polydimethylsiloxane coatings, larger molecules are adsorbed several orders of
magnitude more effectively than very volatile compounds, and somewhat sur-
prisingly with PDMS fibers, analytes with polar functional groups are adsorbed

683
more efficiently than hydrocarbons with similar GC retention indices. Con-
versely, coatings such as CarboxenT"-polydimethylsiloxane have been developed
specifically for sampling of very volatile compounds. Thus, a prudent choice of
fiber coating, and calibration with appropriate standards covering the entire
range of interest, is crucial for quantitative work.
Third, adsorbtion is inversely dependent on temperature. However, when
sampling less volatile substrates, it may be necessary to heat the sample to
increase the headspace concentrations of analytes to measurable levels. In
extreme cases, such as sampling long-chain hydrocarbons from insect cuticles,
samples have been heated to as high as 170°C during sampling (see below).
Finally, adsorbtion appears to depend slightly on concentration, with lower
concentrations of analytes being adsorbed better than higher concentrations
[87].
SPME provides only enough material for GC, GC-MS, possibly GC-FTIR
[92], and HPLC with certain types of compounds and detectors. It is an excep-
tional method for analysis of known compounds, but when attempting to
identify novel compounds in trace amounts, the sensitivity may not be good
enough even for some GC-based methods. SPME's primary value is as a sampl-
ing tool for compounds of known structure, for compounds with structures
similar to known compounds, or for compounds with structures simple enough
to be determined solely by GC-linked methods.
Within these limitations, SPME has a number of advantages. First, it is
rapid and simple, and virtually any odor source can be sampled under laboratory
conditions, or in the field with devices modified for field collections [93]. Loaded
devices can even be sealed up and shipped between laboratories. Second, the
dynamics of release of compounds from an odor source can be followed readily by
sequential sampling of the odor source continuously or at any desired interval
with a pair of devices, for any desired duration, from minutes to months.
However, for sampling in a closed container, compounds that are not produced
continuously may become depleted [87]. Third, one of the key advantages of
SPME is that samples are solvent-free, so that volatile analytes are not hidden
under solvent peaks in GC analyses. The problem of loss of volatile samples
during concentration of extracts is also eliminated. Fourth, SPME, with the
appropriate choice of fiber coating and coating thickness, is amenable to samp-
ling analytes spanning the entire range of volatilities and polarities, including
gases such as ammonia and volatile amines [94,95], low-boiling compounds such
as short-chain alcohols and hydrocarbons, all the way through to very high-
boiling compounds such as hydrocarbons with chain lengths of C30 and higher.
The former group of compounds in particular may be difficult to sample
accurately by other methods.
Because the desorbtion of analytes from the SPME fiber takes an appreciable
period of time in the GC injector, injection techniques must be modified to retain
good chromatographic resolution and symmetrical peaks, particularly as there is
no solvent effect to refocus the analytes at the front of the GC column.

684
Refocusing of the desorbed compounds is accomplished by appropriate choice of
temperature (low initial temperature, or even cryogenic focusing at the column
head) and stationary phase (thicker film phases for volatile compounds) [89,96],
and by use of narrow bore, small-volume injector liners (0.75 mm i.d.; Supelco)
developed especially for use with SPME.
Application of SPME for qualitative analysis of volatiles is straightforward.
However, the determination of relative and absolute amounts of each compound
in blends is not trivial, for several reasons. First, the adsorbtion of compounds on
SPME fibers is proportional to the analytes' molecular weights, and to a lesser
extent, their polarities. Thus, for quantitative studies, a proportionality
constant K that relates the mass adsorbed on the fiber to the concentration in
the headspace must be calculated for each analyte [87]. Second, when sampling
from closed systems, at equilibrium, the amount of a given analyte on the SPME
fiber is proportional to the analyte concentration in the headspace. However,
higher molecular weight compounds take longer to equilibrate on the fiber than
smaller molecules. Consequently, sampling times should be long enough to allow
equilibration of all components of interest. Third, GC detector response varies
with molecular structure and molecular formula, and relative response factors
must be calculated for each compound of interest.
To address these problems with quantitation, Bartelt [87] developed a
regression model for closed systems that found a strong correlation between log
K and the GC Kovats retention index, over a number of different chemical
classes. Using this model, the K value can be accurately predicted for compounds
in 13 chemical classes, using only the Kovats retention index (which is trivial to
determine), the temperature at which SPME sampling was conducted, and the
functional group of the analyte. The K values, in combination with the GC
detector response factors, can then be used to calculate the absolute headspace
concentrations of all compounds in a complex blend from the SPME-GC peak
areas. K values were found to vary slightly with analyte concentrations, but this
effect was minor in comparison to the major changes in K values with increasing
molecular size. For example, the SPME detection limits using a PDMS fiber
were calculated to be about 5.6 ng/ml of air for acetaldehyde, versus 0.00036
ng/ml of air for ethyl decanoate.
This landmark work was extended to dynamic systems in two further
studies. In the first [97], a general model was developed that allowed the
determination of the absolute concentrations of analytes in a uniform airstream.
A key feature of the model is that it does not require the analytes to have reached
equilibrium with the fiber, which is of crucial practical importance for sampling
relatively large molecules that take a long time to equilibrate. A further conse-
quence that emerged from the study was the fact that an SPME fiber inserted
into a 1.5 mm to 4.5 mm outlet port, with a slow air flow rate (2 ml/min) provides
essentially quantitative collection of analytes with GC retention indices > 1500.
The second study examined the effects of airflow and SPME sampling port
diameter in more detail [98]. For a given release rate of an analyte (mass/unit

685
time), trapping efficiency was inversely proportional to air flow rate, with low
flow rates providing the highest trapping efficiencies. Surprisingly, outlet port
diameters of 1.5 and 4.5 mm provided approximately equal trapping efficiencies.
These three studies have provided a foundation for using SPME to make
quantitative measurements of complex mixtures of analytes of widely differing
polarities and volatilities, from both closed and dynamic biological systems. The
importance of these studies in the continuing development of SPME methodol-
ogy cannot be underestimated. To further appreciate the differences between
SPME and trapping on polymers such as Porapak, or purge and trap analysis,
several studies have compared the techniques for collecting volatiles from
various biological systems [99-103]. Each has its advantages and limitations,
and no single method can be used in all situations.

20.4.3.1 Applications of SPME in chemical ecology

Because of its advantages, the use of SPME in chemical ecology is increasing
rapidly. It has proven to be particularly useful for sampling and analysis of small
and/or polar molecules, such as fermentation volatiles, that are highly attractive
to many insect species. For example, SPME with poly(dimethylsiloxane)
(PDMS) or Carboxen® fibers used with thick film GC columns allowed the
analysis of low levels of ammonia and other volatile amines, pyrazines, and
acetic acid in tryptic soy broth cultures of the bacterium Staphylococcus aureus
that are attractive to Mexican fruit flies [94,95]. SPME analyses of tryptic soy
cultures of other bacterial species found other types of small, polar compounds,
including dimethyl disulfide, and short-chain alcohols, ketones, and carboxylic
acids [104]. A further study used SPME to analyze the profile of volatiles from
bird faeces, which serves as a natural protein source for female fruit flies [105].
Ethanol, propanol, ammonia and low-molecular weight amines, phenol, and
pyrazines were identified. In situ derivatization of small primary amines with
4-nitrophenyltrifluoroacetate during SPME headspace sampling, or derivatiza-
tion of amines in aqueous samples with pentafluorobenzaldehyde [106] followed
by SPME sampling can significantly increase the selectivity and sensitivity of
these analyses [107].
In another application, SPME with PDMS fibers was used to sample volatiles
produced by the fungus Fusarium verticillioides growing on corn [108]. The
fungal volatiles, in combination with the insects' pheromone, attract several sap
beetle species that vector the fungus. Sequential SPME sampling of the fungal
cultures determined that the profile of volatiles changed with time, initially
being dominated by ethanol and other short-chain alcohols, ethyl acetate, and
acetaldehyde. Eight days after inoculation, these compounds were essentially
absent, and the volatiles profile was dominated by substituted phenols. SPME
played a crucial role in an analogous study that examined the volatiles produced
by fermenting bread dough, a source of attractants for nitidulid beetles [109].
SPME was used to sample the volatiles from the bread dough, which consisted
primarily of ethyl acetate, short chain alcohols, and acetaldehyde, and to accu-

686
rately reconstruct release rates from synthetic blends of the identified
compounds.
SPME was used to determine that the rhinoceros beetles Scapanes australis
and Strategus aloeus produce very volatile pheromones with remarkably simple
structures. The pheromone of the former species consisted of a 2:1 blend of the R
and S enantiomers of 2-butanol in combination with 3-hydroxy-2-butanone and
2,3-butanediol, whereas the latter species produced a blend of 2-butanone,
3-pentanone, and sec-butyl acetate [110]. These pheromone components vary
widely in volatility and polarity, and might have been difficult to analyze by
other methods.
Palm weevils, such as the American palm weevil Rhynchophoruspalmarum,
are attracted by a synergistic combination of pheromones and volatiles produced
from decaying host plant material. SPME with PDMS fibers was used in combi-
nation with collection of volatiles on an adsorbent polymer (Supelpak-2) to
identify more than 100 compounds from the headspace above decaying samples
of sugar cane, coconut palm, oil palm, and the tree Jacaratiadigitata [111]. The
combination of SPME with adsorbent trapping was critically important because
the most volatile compounds, such as acetaldehyde and short-chain alcohols and
esters, were minimally retained on the adsorbent traps; in fact, all compounds
with retention times less than 12 min broke through the adsorbent traps to a
greater or lesser extent.
In all of the studies above, the critical role of SPME cannot be overempha-
sized. Traditional methods of trapping the analytes on adsorbents such as
charcoal, Porapak, or Tenax would not have worked because the highly volatile
semiochemicals would not have been retained, and the concentrations of most of
the volatiles would have been too low for direct analysis of headspace samples.
Furthermore, the more volatile constituents would have been hidden under the
solvent peaks during GC analysis.
SPME is also being used in sampling and identification of insect pheromones
and related attractants of moderate to relatively high molecular weight. In one
of the first reported studies of the application of SPME to insect chemical
ecology, SPME was used to quantify the diurnal cycleof production of the
pheromone components 2-methyl-4-heptanol, 2-methyl-4-octanol, 5-nonanol,
3-hydroxy-4-methyl-5-nonanone, and 4-methyl-5-nonanol from sugarcane weev-
ils [112]. With continuous sampling at 30 or 60 min intervals, SPME analysis
provided an accurate picture of the time course of pheromone production [112].
As an example of the critical importance of determining the correct ratio of
compounds in attractant blends, Zhang et al. [113] reexamined volatiles released
from apples, using SPME coupled with electroantennographic detection. The
analyses resulted in the development of an improved blend of attractants for
apple maggot fly consisting of five short-chain esters of 8-10 carbons.
SPME in combination with coupled GC-electroantennographic detection
also played a critical role in the development of improved attractants for house-
flies [103]. Volatiles collected from fresh pig manure that elicited strong

687
responses from housefly antennae included butanoic and 3-methylbutanoic
acids, dimethyldisulfide, dimethyltrisulfide, dimethyltetrasulfide, phenol,
phenylethanol, indole, and skatole. An optimized blend of skatole, butanoic acid,
and dimethyltrisulfide was approximately as attractive as pig manure. The same
research group compared SPME to more traditional methods of headspace
analysis (collection on Tenax resin, purge and trap analysis) when analyzing
buffalo gourd root powder for attractants for corn rootworms [114]. Results from
the three sampling methods were similar, and a number of volatiles were
identified, of which (E)-3-octen-2-one and (E,E)-3,5-octadien-2-one were most
attractive in field tests.
A number of insect species utilize the fruiting bodies of fungi as breeding or
feeding substrates, and the insects locate their hosts by olfaction. SPME in
combination with collection of volatiles on Porapak was used to analyze the
volatiles from intact and chopped fruiting bodies [66]. Compounds identified
included ethanol, acetic acid, mono-and sesquiterpenes, and intermediate chain
length aldehydes, alcohols, and ketones. As in the studies mentioned above,
SPME was crucial to the detection of ethanol and acetic acid. In related studies,
SPME is finding increased use in botany, for example in sampling volatile
chemicals from intact and damaged plants [89,115,116]. SPME was also used to
sample the patterns of emission of odors from whole plants, buds, and flowers of
the plant Boronia megastigma while manipulating the light regimes under
which the plants were held. The identified volatiles included mono- and ses-
quiterpenes, esters, alcohols, and straight-chain hydrocarbons [117].
In an example of SPME used in the extraction of pheromones from aqueous
media, SPME was used to sample the attractant pheromones produced by the
sessile female gametes of the brown algae Laminaria digitata to attract the
motile male gametes [15]. These compounds, consisting of C11 hydrocarbons and
epoxides, were extracted by immersion of the PDMS fiber in 20 ml of medium
from a culture that had released the female gametes. Previous studies of this
type had used closed-loop-stripping analyses (see below), but the SPME method
was much simpler and faster.
A series of papers has reported the use of SPME in the identification of
pheromones from leafminer moths [118-121]. These moths are very small (-2-3
mm long), and SPME had distinct advantages over dissection and extraction of
pheromone glands. In particular, SPME was used to collect pheromone for
extended periods from undisturbed calling female moths, and samples could be
taken from individual females multiple times. The SPME-collected volatiles
from a single calling female equaled the amount obtained from an extract of fifty
pheromone glands [119]. This finding is important because lepidopteran phero-
mones are frequently produced in such small amounts that even identification of
the major components can be difficult. Thus, any technique that significantly
increases the amount of pheromone collected is of tremendous value.
SPME was recently used to study the dynamics of pheromone production by
male southern green stinkbugs [122]. Sexually mature males produced

688
nerolidol, bisabolene, and two bisabolene epoxides, but the amounts produced by
individuals were highly variable. Repeated sampling of the same individuals
revealed that the ratio of compounds produced by individuals from the same
population was variable, but that the ratio produced by a given individual was
stable over time. This type of study would have been more difficult using
traditional methods of pheromone collection on adsorbents.
SPME has also been used to analyze larger molecules with limited volatility
by heating the relevant tissues in a sealed vial to temperatures as high as 170°C,
and sampling the headspace by SPME. This method was used to sample long-
chain fatty acids in sternal glands from paper wasps [123], and hydrocarbons
from cuticular fragments from the wasps Vespa crabro, V. orientalis, and Polistes
dominulus [124].
Finally, SPME is finding increasing use in the sampling of larger molecules
of limited volatility, using an adaptation termed "contact SPME", in which the
SPME fiber is wiped over an insect's cuticle or over an extruded gland. As with
headspace sampling, the method is nondestructive so that the same organism(s)
can be sampled repeatedly, and specific areas of the cuticle or specific glands can
be sampled without fear of contamination from other structures or tissues, as
frequently happens with dissected glands. In one of the first reports of contact
SPME, Frerot et al. [125] reanalyzed the pheromone of the moth Sesamia
nonagriodes,collecting pheromone by wiping the SPME fiber over the extruded
gland. More pheromone was obtained by SPME than by gland washes, the SPME
extracts were cleaner, and the ratio of pheromone components was slightly
different than that obtained by gland washes. In another example, contact
SPME was used to confirm the sternal gland as the site of production of a trail
pheromone, (Z)-3-dodecen-1-ol, in the termite Macrotermesannandalei,by rub-
bing the SPME fiber on the exposed gland [126]. However, because each termite
produced only about 1 ng of the pheromone, the compound was initially identi-
fied from extraction and fractionation of thousands of termites and excised
glands.
Contact SPME has also been used in the analysis of the large cuticular
hydrocarbons of insects that are used for recognition of species, sex, and nest-
mate status. For example, contact SPME was used to determine the hydrocar-
bon profiles of several Reticulitermes and Coptotermes species [127]. In a related
example, SPME proved crucial in studying cuticular pheromones associated
with dominance status [128,129] and fertility [130] in the ants Dinoponera
quadriceps and Harpegnathossaltatorrespectively. In both cases, the ability to
sample the same individuals repeatedly for up to four months was crucial in
following the increase in pheromone with the development of dominant repro-
ductive status within the colony.
Sledge et al. [131] compared solvent extraction with two methods of SPME
extraction in the analysis of Dufours gland secretions of a stenogastrine wasp,
Parischnogastersp. The first SPME method consisted of headspace extraction of
glands held in a sealed vial at 170°C for 10 min, whereas the second consisted of

689
simply puncturing the gland and adsorbing the contents on the SPME fiber. The
results from all three methods were reported to be similar. The same group also
used contact SPME to study mechanisms used by a social parasite, the wasp
Polistes sucifer, to usurp host nests [132,133]. This wasp has no worker caste,
and must take over host nests in order to produce offspring. Repeated sampling
of the cuticular hydrocarbons of the parasite and its hosts by contact SPME
revealed that the parasite rapidly acquires the hydrocarbon profile of its hosts,
and is thus accepted as a legitimate member of the colony.
From the examples above, it is apparent that the SPME technique has
proven enormously useful in chemical ecology, and it is rapidly becoming the
method of choice for nondestructive and repetitive sampling of insect phero-
mones and related compounds. However, the reader is again cautioned that, as
with most other methods of sampling volatiles, the profile of adsorbed volatiles
does not represent the true proportions of the compounds in the headspace, and
calibration is crucial for accurately determining both the relative and absolute
amounts of compounds present.

20.4.4 Miscellaneous methods

20.4.4.1 Headspace analysis

Traditional headspace analysis methods, in which an aliquot of the headspace
gases above an insect or group of insects is removed with a gas-tight syringe and
analyzed directly by GC, are not appropriate for use with insect pheromones
because the quantities of pheromones produced are too small. In all cases, some
type of concentration is required in order to collect detectable quantities of
material.

20.4.4.2 Thermal desorbtion and purge and trap

Several methods have been devised for direct vaporization of volatiles from
insect secretions or tissue samples into a GC. In one of the simplest methods,
dissected glands were introduced into the GC injector using a grooved syringe
needle to hold the glands [134,135]. Alternatively, analytes can be coated on a
solids injection syringe [136]. In a related method, secretions from swallowtail
butterfly larvae were collected in glass capillaries, one end of which was then
sealed, followed by the open end being inserted into the injector to desorb the
volatiles off the glass [137].
Another solventless method involves modifying a GC injector port to permit
the insertion of sealed soft glass capillaries containing small glands and other
tissue samples. After heating in the injection port for several minutes, the
capillary is broken in situ, and the volatiles from the sample are swept onto the
GC column [53,138]. These methods suffer from the disadvantages that the
profile of volatiles represents the contents of the entire sample, rather than the
profile of compounds actually released by the organism, and all introduce
significant amounts of water into the GC column which may degrade chromato-

690
graphic performance. Furthermore, in the latter method, glass fragments from
the capillaries can adsorb analytes. Nevertheless, the latter method has been
used to analyze dissected glands, fragments of insect cuticle, or other body parts
for pheromones of a number of ant species [138].
Standard purge and trap techniques, in which volatiles are collected or ther-
mally desorbed onto an adsorbent cartridge, followed by purging the cartridge
with dry inert gas to remove water and then thermal desorbtion directly into the
GC (short-path thermal desorbtion), have found only limited use in chemical
ecology, despite turn-key systems being available from several suppliers. Purge
and trap analysis has been used, for example, to analyze insect attractants from
ripe or fermenting fruit [139-142]. Purge and trap methods suffer some of the
same limitations as the aeration methods described above, such as the fact that
the profile of chemicals obtained may be biased by the lack of retention of some
classes of compounds. Furthermore, some of the trapping materials, such as sil-
ica gel, may catalyze reactions of the analytes during the rapid heating of the
desorbtion cycle. Artefacts also may be produced by thermal degradation of the
adsorbents. Several reports have addressed these problems [43,70,143,144].

20.4.4.3 Trappingon glass

Small amounts of pheromones can be collected from individual insects over
limited time periods by trapping the volatiles on glass beads [145] or plugs of
glass wool [146-148]. Similarly, pheromones have been collected by drawing air
through glass capillaries positioned close to the exposed pheromone gland of a
calling female insect [149-152]. The pheromone is adsorbed with surprising
efficiency onto the glass walls of the capillaries. For all of the above methods,
trapped pheromones are recovered in a small amount of solvent, and the cleaned
glass materials contribute minimal background. However, short sampling inter-
vals and low airflow rates must be used to minimize the possibility of break-
through.

20.4.4.4 Closed loop stripping

Closed loop stripping systems, adapted from water sampling methods [82,153],
has found sporadic use in semiochemical research. Air is circulated with a
diaphragm pump through a closed system which includes an aeration chamber
and a volatiles trap. Closed loop stripping was used to collect and sample algal
pheromones from aqueous matrices [13-16,154-156]. It has also found use in
studies of plant metabolism [153], of the diel periodicity of pheromone release by
female moths [157], and in identification of pheromones produced by desert
locusts [158]. In another interesting example, closed loop stripping was used in
the collection of plant volatiles induced by oviposition by elm leaf beetles into
leaf surfaces [159]. Closed loop stripping has the advantage that because the air
is recirculated, chances of contamination during a single run are decreased.
However, care must be taken to eliminate "memory" effects due to adsorbtion of
compounds to system components.

691
20.4.4.5 Cryogenic trapping
Insect semiochemicals have been collected occasionally by cryogenic trapping,
for example, using liquid nitrogen cooled traps to condense air containing
entrained volatiles [58]. To recover the volatiles, the trap is warmed slowly to
distill off the condensed air, leaving a concentrate. Unlike most other methods,
cryogenic trapping can collect compounds with widely differing volatilities and
polarities, including gases like ethylene and CO 2 [160,161]. However, it has a
number of drawbacks which preclude its general use, including the large volume
of water that is also trapped, from which analytes of interest have to be extracted
or otherwise separated. Safety may also be a concern because formation of ice
plugs can create explosion hazards as the condensed air vaporizes.

20.5 SUMMARY AND CONCLUSIONS

Since the identification of the first insect pheromone more than 40 years ago, a
series of specialized techniques have been developed to streamline the identifica-
tion process. Whole body extractions and gland extractions in which the test
organisms were sacrificed have largely given way to dynamic collection methods
in which the volatiles released by living organisms are collected for extended
periods of time. Such dynamic collection methods provide a more accurate
representation of the pheromone blends produced by organisms than tissue
extractions, and allow repeated sampling from the same individuals. For small-
scale collections, analytes may be trapped on short beds of Porapak Q, Tenax, or
activated charcoal. Alternatively, analytes may be sampled by SPME, which has
increased flexibility and ease of use in comparison to trapping on adsorbents.
Furthermore, because SPME requires no solvent for desorbtion, volatile
analytes are not obscured by solvent peaks in GC analyses. For large scale
collection of analytes for bioassay or for spectroscopic analysis, large numbers of
organisms can be placed in appropriate sized aeration chambers supplied with
food and water, with collections continued indefinitely, changing the collectors
periodically. Overall, it must be remembered that each of the methods described
above has limitations with regard to the types of analytes that can be collected,
the efficiency of collection of different classes of analytes, and the quantities that
can be collected. No single method is appropriate for all sampling and collection
scenarios.

REFERENCES

1 J.H. Fabr6, Bilder aus der Insektenwelt. Verlag Kosmos, Stuttgart, 1914.
2 E. Hecker and A. Butenandt, in: H.E. Hummel and T.A. Miller (Eds.), Techniques in
Pheromone Research, Springer-Verlag, New York, 1984.
3 M.S. Mayer and J.R. McLaughlin, Handbook of Insect Pheromones and Sex
Attractants. CRC Press, Boca Raton, FL, 1991.
4 W. Francke and S. Schulz, in: K. Mori (Ed.), Comprehensive Natural Products
Chemistry, Vol. 8. Elsevier, Amsterdam and New York, 1999.

692
5 E.D. Morgan and N. Bhushan Mandava (Eds.), Handbook of Natural Pesticides. Vol.
4. Pheromones, PartsA and B. CRC Press, Boca Raton, FL, 1988.
6 J. Hardie and A.K. Minks, Pheromones of Non-lepidopteran Insects Associated with
AgriculturalPlants. CABI Publishing, Wallingford, UK, 1999.
7 The Pherolist, http://www.nysaes.cornell.edu/pheronet/
8 J.H. Tumlinson, M.M. Brennan, R.E. Doolittle, E.R. Mitchell, A. Brabham, B.E.
Mazomenos, A.H. Baumhover and D.M. Jackson, Arch. Insect Biochem. Physiol., 10
(1989) 255.
9 G. Gries, K.N. Slessor, R. Gries, G. Khaskin, P.D.C. Wimalaratne, T.G. Gray, G.G.
Grant, A.S. Tracey and M.J. Hulme, J. Chem. Ecol., 23 (1997) 19.
10 J.G. Millar, Tetrahedron Lett., 38 (1997) 7971.
11 L.E.L. Rasmussen, T.D. Lee, W.L. Roelofs, A.J. Zhang and G.D. Davies, Nature, 379
(1996) 684.
12 L.E.L. Rasmussen, T.D. Lee, A.J. Zhang, W.L. Roelofs and G.D. Davies, Chem.
Senses, 22 (1997) 417.
13 I. Maier, D. G. Muller, G. Gassmann, W. Boland, F.-J. Marner and L. Jaenicke,
Naturwissenchaften, 71 (1984) 48.
14 I. Maier and D.G. Muller, Biol. Bull., 170 (1986) 145.
15 I. Maier, G. Pohnert, S. Pantke-Bocker and W. Boland, Naturwissenschaften, 83
(1996) 378.
16 D.G. Muller, G. Gassmann and K. Luning, Nature, 279 (1979) 430.
17 L. Muller, T. Gorecki and J. Pawliszyn, Fresenius'J. Anal. Chem., 364 (1999) 610.
18 J. Hurter, E.F. Boiler, E. Stadler, B. Blattmann, H.R. Buser, N.U. Bosshard, L.
Damm, M.W. Kozlowski and R. Schoni, Experientia, 43 (1987) 157.
19 W.S. Leal and T. Kadosawa, Biosci. Biotech. Biochem., 56 (1992) 1004.
20 E.O. Wilson and B. Holldobler, The Ants. Belknap Press of Harvard University Press,
Cambridge, MA, 1990.
21 T.D. Fitzgerald and F.X. Webster, Can. J. Zool., 71 (1993) 1511.
22 M.C. Birch, in: W.J. Bell and R.T. Card6 (Eds.), Chemical Ecology of Insects. Sinauer,
Sunderland, MA, 1984.
23 R. Plarre and D.C. Vandervel, in: J. Hardie and A.K. Minks (Eds.), Pheromones of
Non-lepidopteran Insects Associated with Agricultural Plants. CABI Publishing,
Wallingford, UK, 1999.
24 M. Borges and J.R. Aldrich, Experientia, 48 (1992) 893.
25 J. Pettis, T. Pankiw and E. Plettner, in: J. Hardie and A.K. Minks (Eds.), Pheromones
of Non-lepidopteran Insects Associated with AgriculturalPlants. CABI Publishing,
Wallingford, UK, 1999.
26 L.M. Roth and G.P. Dateo, J. Insect Physiol., 12 (1966) 255.
27 L. Sreng, J. Chem. Ecol., 16 (1990) 2899.
28 Q. Wang and J.G. Millar, Ann. Entomol. Soc. Am., 93 (2000) 972.
29 A. Hassanali and B. Torto, in: J. Hardie and A.K. Minks (Eds.), Pheromones of
Non-lepidopteran Insects Associated with Agricultural Plants. CABI Publishing,
Wallingford, UK, 1999.
30 E. Zeeck, T. Harder and M. Beckman, J. Chem. Ecol., 24 (1998) 13.
31 P.S. Chen, Experientia, 52 (1996) 503.
32 K.F. Haynes and K.V. Yeargan, Ann. Entomol. Soc. Am., 92 (1999) 960.
33 W. Powell, in: J. Hardie and A.K. Minks (Eds.), Pheromones of Non-lepidopteran
Insects Associated with Agricultural Plants. CABI Publishing, Wallingford, UK,
1999.
34 C. Gemeno, K.V. Yeargan and K.F. Haynes, J. Chem. Ecol., 26 (2000) 1235.
35 W.F. Wood, M.G. Leahy, R. Galun, G.D. Prestwich, J. Meinwald, R.E. Parnell and
R.C. Payne, J. Chem. Ecol., 1 (1975) 501.

693
36 C.E. Linn and L.K. Gaston, Environ. Entomol., 10 (1981) 751.
37 H.E. Hummel and T.A. Miller (Eds.), Techniques in PheromoneResearch. Springer
Verlag, New York, 1984.
38 J.G. Millar and K.F. Haynes, Methods in Chemical Ecology, Vols. 1 and 2. Chapman
and Hall, New York, 1998.
39 P.E. Howse, I. Stevens and 0. Jones (Eds.), Insect Pheromones and Their Use in Pest
Management. Chapman and Hall, London, 1998.
40 A.R. McCaffery and I.D. Wilson (Eds.), Chromatography and Isolation of Insect
Hormones and Pheromones. Plenum, New York, 1989.
41 G.R. Jones and N.J. Oldham, J. Chromatog.A, 843 (1999) 199.
42 M.A. Golub and I. Weatherston, in: H.E. Hummel and T.A. Miller (Eds.), Techniques
in Pheromone Research. Springer-Verlag, New York, 1984.
43 R. Teranishi, R.G. Buttery and H. Sugisawa, Bioactive Volatile Compounds from
Plants, ACS Symposium Series 525. American Chemical Society, Washington DC,
1993.
44 G.R. Takeoka, R.A. Flath, M. Guntert and W. Jennings, J. Agric. Food Chem., 36
(1988) 553.
45 L. Tollsten and G. Bergstrom, Phytochemistry, 27 (1988) 4013.
46 E. Lorbeer, M. Mayr, B. Hausmann and K. Kratzl, Monat. Fur. Chem., 115 (1984)
1107.
47 R.R. Heath and A. Manukian, J. Chem. Ecol., 18 (1992) 1209.
48 C.S. Charron, D.J. Cantliffe, R.M. Wheeler, A. Manukian and R.R. Heath, J. Am. Soc.
Hort. Sci., 121 (1996) 488.
49 J.G. McConnell, J.H. Borden, R.M. Silverstein and E. Stokkink, J. Chem. Ecol., 3
(1977) 549.
50 H.D. Pierce, Jr., A.M. Pierce, J.G. Millar, J.H. Borden and A.C. Oehlschlager, Proc.
3rd Int. Conf. Stored ProductEntomology, Kansas State University, Manhattan, KS,
1984.
51 J.H. Tumlinson, D.D. Hardee, J.P. Minyard, A.C. Thompson, R.T. Gast and P.A.
Hedin, J. Ecol. Entomol., 61 (1968) 470.
52 T.D. Paine, J.G. Millar, C.C. Hanlon and J.S. Hwang, J. Chem. Ecol., 15 (1999) 433.
53 A.B. Attygalle and E.D. Morgan, Angew. Chem. Int. Ed., 27 (1988) 460.
54 T.R. Hamilton-Kemp, J.G. Rodriquez, D.D. Archbold, R.A. Andersen, J.H. Loughrin,
C.G. Patterson and S.R. Lowry, J. Chem. Ecol., 15 (1989) 1465.
55 R.R. Heath and A. Manukian, J. Chem. Ecol., 20 (1994) 593.
56 J.H. Loughrin, A. Manukian, R.R. Heath, T. Turlings and J.H. Tumlinson, Proc.
Natl. Acad. Sci. USA, 91 (1995) 11836.
57 J.F. Chang, J.H. Benedict, T.L. Payne, B.J. Camp and S.B. Vinson, J. Chem. Ecol., 15
(1989) 767.
58 L.E. Browne, D.L. Wood, W.D. Bedard, R.M. Silverstein and J.R. West, J. Chem.
Ecol., 5 (1979) 397.
59 J.G. Millar, C.-H. Zhao, G.N. Lanier, D.P. O'Callaghan, M. Griggs, J.R. West and
R.M. Silverstein, J. Chem. Ecol., 12 (1986) 583.
60 R.R. Heath, P.J. Landolt, B. Dueben and B. Lenczewski, Environ. Entomol., 21
(1992) 854.
61 B. Weissbecker, J.J.A. van Loon, M.A. Posthumus, H.J. Bouwmeester and M. Dicke,
J. Chem. Ecol., 26 (2000) 1433.
62 N.K. Agelopoulos, K. Chamberlain and J.A. Pickett, J. Chem. Ecol., 26 (2000) 497.
63 M.H. Pham-Delegue, P. Etievant, E. Guichard and C. Masson, J. Chem. Ecol., 15
(1989) 329.
64 H.E.M. Dobson, J. Bergstrom, G. Bergstrom and I. Groth, Phytochemistry, 26 (1987)
3171.

694
 65 H.E.M. Dobson, in: H.F. Linskens and J.F. Jackson (Eds.), Essential Oils and Waxes.
Springer-Verlag, New York, 1991.
66 J. Faldt, J.M. Jonsell, G. Nordlander and A.K. Borg-Karlson, J. Chem. Ecol., 25
(1999) 567.
67 M. Kleinheitz, H. Jactel and P. Menassieu, J. Chem. Ecol., 25 (1999) 2741.
68 B.V. Burger, Z.M. Munro and J.H. Visser, J. High Res. Chromatogr., 11 (1988) 496.
69 J. Schaefer, in: Peter Schreier (Ed.), Flavour '81. Walter de Gruyter, New York,
1981.
70 H. Rothweiler, P.A. Wager and C. Schlatter, Atmos. Environ., 25B (1991) 231.
71 G. MacLeod and J.M. Ames, J. Chromatogr., 355 (1986) 393.
72 M.J. Lewis and A.A. Williams, J. Sci. Food Agric., 31 (1980) 1017.
73 K.J. Byrne, W.E. Gore, G.T. Pearce and R.M. Silverstein, J. Chem. Ecol., 1 (1975) 1.
74 L.B. Bjostad, L.K. Gaston and H.H. Shorey, J. Insect Physiol., 26 (1980) 493.
75 A. Sturaro, G. Parvoli and L. Doretti, Chromatographia,33 (1992) 53.
76 P.H. Krumperman, J. Agric. Food Chem., 20 (1972) 909.
77 E.D. Pelizzari and K.J. Krost, Anal. Chem., 56 (1984) 1813.
78 R.J.B. Peters, J.A.D.V. Renesse, V. Duivenbode, J.H. Duyzer and H.L.M. Verhagen,
Atmos. Environ., 28 (1994) 2413.
79 J.J. Manura, Adsorbent Resins, PartI: Calculationand Use of Breakthrough Volume
Data. The Mass Spec. Source 7: 3-11. Scientific Instrument Services, Ringoes, NJ,
1994.
80 I. Maier and M. Fieber, J. High Res. Chromatogr., 11 (1988) 566.
81 P. Matile and R. Altenburger, Planta, 174 (1988) 242.
82 K. Grob and F. Zurcher, J. Chromatogr.117 (1976) 285.
83 R.R. Heath, J.R. McLaughlin, F. Proshold and P.E.A. Teal, Ann. Entomol. Soc. Am,.
84 (1991) 182.
84 J.H. Tumlinson, R.R. Heath and P.E.A. Teal, in: B.A. Leonhardt and M. Beroza
(Eds.), Insect Pheromone Technology: Chemistry and Applications, ACS Symposium
Series 190. American Chemical Society, Washington DC, 1982.
85 P.A. Hollingdale-Smith, Chem. Ind., (1975) 226.
86 J. Pawliszyn, Solid Phase Microextraction Theory and Practice.Wiley-VCH, New
York, 1997.
87 R.J. Bartelt, Anal. Chem., 69 (1997) 364.
88 A.J. Matich, D.D. Rowan and N.H. Banks, Anal. Chem., 68 (1996) 4114.
89 B. Schafer, P. Hennig and W. Engewald, J. High Res. Chromatogr.18 (1995) 587.
90 X. Yang and T. Peppard, LC-GC, 13 (1995) 882.
91 H. Verhoeven, T. Beuerle and W. Schwab, Chromatographia,46 (1997) 63.
92 J. Auger, S. Rousset, E. Thibout and B. Jaillais, J. Chromatogr.A, 819 (1998) 45.
93 L. Muller, T. Gorecki and J. Pawliszyn, Fresenius'J.Anal. Chem., 364 (1999) 610.
94 D.C. Robacker and R.A. Flath, J. Chem. Ecol., 21 (1995) 1861.
95 D.C. Robacker and R.J. Bartelt, J. Agric. Food Chem., 44 (1996) 3554.
96 J.J. Langenfeld, S.B. Hawthorne and D.J. Miller, J. Chromatogr.A, 740 (1996) 139.
97 R.J. Bartelt and B.W. Zilkowski, Anal. Chem., 71 (1999) 92.
98 R.J. Bartelt and B.W. Zilkowski, Anal. Chem., 72 (2000) 3949.
99 N.K. Agelopoulos and J.A. Pickett, J. Chem. Ecol., 24 (1998) 1161.
100 J. Faldt, M. Eriksson, I. Valterova and A.K. Borg-Karlson. Zeit. Naturforsch., C 55
(2000) 180.
101 J.S. Elmore, D.S. Mottram and M.A. Erbahadir, J. Agric. Food Chem., 45 (1997)
2638.
102 J. Vercammen, P. Sandra, E. Baltussen, T. Sandra and F. David, J. High Res.
Chromatogr. 23 (2000) 547.
103 A.A. Cossb and T.C. Baker, J. Agric. Entomol., 13 (1996) 301.

695
104 D.C. Robacker and R.J. Bartelt, J. Chem. Ecol., 23 (1997) 2897.
105 D.C. Robacker, J.A. Garcia and R.J. Bartelt, J. Chem. Ecol., 26 (2000) 1849.
106 M.J. Avery and G.A. Junk, Anal. Chem., 57 (1985) 790.
107 L. Pan, J.M. Chong and J. Pawliszyn, J. Chromatog.A, 773 (1997) 249.
108 R.J. Bartelt and D.T. Wicklow, J. Agric. Food Chem., 47 (1999) 2447.
109 R.J. Bartelt and B.W. Zilkowski, J. Chem. Ecol., 24 (1998) 535.
110 D. Rochat, P. Ramirez-Lucas, C. Malosse, R. Aldana, T. Kakul and J.P. Morin, J.
Chromatogr.A, 885 (2000) 433.
111 D. Rochat, P.N. LeMeillour, J.R. Esteban-Duran, C. Malosse, B. Perthuis, J.P. Morin
and C. Descoins, J. Chem. Ecol., 26 (2000) 155.
112 C.P. Malosse, P. Ramirez-Lucas, D. Rochat and J.P. Morin, J. High Res.
Chromatogr., 18 (1995) 669.
113 A. Zhang, C. Linn, S. Wright, R. Prokopy, W. Reissig and W.L. Roelofs, J. Chem.
Ecol., 25 (1999) 1221.
114 A.A. Coss6 and T.C. Baker, J. Chem. Ecol., 25 (1999) 51.
115 S.F. Vaughn and R.A. Boydston, J. Chem. Ecol., 23 (1997) 2107.
116 C.S. Charron and C.E. Sams, J. Am. Soc. Hort. Sci., 124 (1999) 462.
117 H.S. MacTavish, N.W. Davies and R.C. Menary,Annals Bot. (London), 86 (2000) 347.
118 A.K. Borg-Karlson and R. Mozuraitis, Zeit. Naturforsch., 51c (1996) 599.
119 R. Mozuraitis, A.-K. Borg-Karlson, A. Eiras, P. Witzgall, A. Kovaleski, E.F. Vilela and
C.L. Unelius, Abstracts of the Int. Soc. Chem. Ecol. 13th Annual Meeting, Prague,
Aug. 18-22, 1996, pp. 193.
120 R. Mozuraitis, A.K. Borg-Karlson, V. Buda and P. Ivinskis, J. Appl. Entomol., 123
(1999) 603.
121 R. Mozuraitis, V. Buda, V. Jonusaite, A.K. Borg-Karlson and R. Noreika, Entomol.
Exp. Appl., 94 (2000) 15.
122 N. Miklas, M. Renou, I. Malosse and C. Malosse, J. Chem. Ecol., 26 (2000) 2473.
123 R. Maile, F.R. Dani, G.R. Jones, E.D. Morgan and D. Ortius, J. Chromatogr. A, 816
(1998) 169.
124 G. Monetti, F.R. Dani, G. Pieraccini and S. Turillazzi, Rapid Comm. Mass Spec., 11
(1997) 857.
125 B. Frbrot, C. Malosse and A.H. Cain, J. High Res. Chromatogr.20 (1997) 340.
126 A. Peppuy, A. Robert, E. Semon, C. Ginies, M. Lettere, O. Bonnard and C. Bordereau,
J. Insect Physiol., 47 (2001) 445.
127 J.M. Bland, W. Osbrink, D. Woodson and C.B. Vigo, Abstracts of Papers, 218th
NationalMeeting Am. Chem. Soc., 218 (1999) AGRO 130. New Orleans, LA, August
22-26 1999.
128 T. Monnin, C. Malosse and C. Peeters, J. Chem. Ecol., 24 (1998) 473.
129 C. Peeters, T. Monnin and C. Malosse, Proc. R. Soc. Biol. Sci. B, 266 (1999) 1323.
130 J. Liebig, C. Peeters, N.J. Oldham, C. Markstaedter and B. Hoelldobler, Proc. Nat.
Acad. Sci. USA, 97 (2000) 4124.
131 M.F. Sledge, G. Moneti, G. Pieraccini and S. Turillazzi, J. Chromatogr. A, 873 (2000)
73.
132 S. Turillazzi, M.F. Sledge and G. Moneti, Ethol. Ecol. Evol., 10 (1998) 293.
133 S. Turillazzi, M.F. Sledge, F.R. Dani, R. Cervo, A. Massolo and L. Fondelli,
Naturwissenschaften, 87 (2000) 172.
134 K. Dettner, G. Schwinger and P. Wunderle, J. Chem. Ecol., 11 (1985) 859.
135 C. D6scoins and M. Gallois, Ann. Zool. Ecol. Anim., 11 (1979) 521.
136 R. Deml and K. Dettner, J. Chem. Ecol., 20 (1994) 2127.
137 B.V. Burger, Z. Munro, M. Roth, H.S.C. Spies, V. Truter, H. Geertsma and A. Habich,
J. Chem. Ecol., 11 (1985) 1093.
138 E.D. Morgan, Anal. Chim. Acta, 236 (1990) 227.

696
139 A.A. Cosse, J.J. Endris, J.G. Millar and T.C. Baker, Entomol. Exp. Appl., 72 (1994)
233.
140 A.A. Coss6, J.L. Todd, J.G. Millar, L.A. Martinez and T.C. Baker, J. Chem. Ecol., 21
(1995) 1823.
141 P.L. Phelan and H. Lin, J. Chem. Ecol., 17 (1991) 1253.
142 T.C. Leskey, R.K. Prokopy, S.E. Wright, P.L. Phelan and L.W. Haynes, J. Chem.
Ecol., 27 (2001) 1.
143 J.M. Patt, D.F. Rhoades and J.A. Corkill, Phytochemistry, 27 (1988) 91.
144 A. Calogirou, B.R. Larsen, C. Brussol, M. Duane and D. Kotzias, Anal. Chem., 68
(1996) 1499.
145 R.E. Charlton and R.T. Carde, J. Insect Physiol., 28 (1982) 423.
146 K.F. Haynes and R.E. Hunt, J. Chem. Ecol., 16 (1990) 509.
147 K.F. Haynes, L.K. Gaston, M. M. Pope and T.C. Baker, Environ. Entomol., 12 (1983)
1597.
148 T.C. Baker, L.K. Gaston, M.M. Pope, L.P.S. Kuenen and R.S. Vetter, J. Chem. Ecol.,
7 (1981) 961.
149 P. Witzgall and B. Frerot, J. Chem. Ecol., 15 (1989) 707.
150 P. Witzgall, M. Bengtsson, G. Karg, A.-C. Backman, L. Streinz, P.A. Kirsch, Z. Blum
and J. Lofqvist, J. Chem. Ecol., 22 (1996) 191.
151 A. Shani, J. Chem. Ecol., 16 (1990) 959.
152 M.J. Lacey and C.J. Sanders, J. Chem. Ecol., 18 (1992) 1421.
153 W. Boland, P. Ney, L. Jaenicke and G. Gassmann, in: P. Schreier (ed.), Analysis of
Volatiles. Walter de Gruyter, Berlin, New York, 1984.
154 K. Stratmann, W. Boland and D.G. Muller, Angew. Chem. Int. Ed., 31 (1992) 1246.
155 D. Wirth, I. Fischer-Liu, W. Boland, D. Icheln, T. Runge, W.A. Konig, J. Phillips and
M. Clayton, Helv. Chim. Acta, 75 (1992) 734.
156 D. Wirth and W. Boland, Helv. Chim Acta, 73 (1990) 916.
157 H.J. Bestmann, J. Erler and 0. Vostrowsky, Experientia, 44 (1988) 797.
158 K. Seidelmann, K. Luber and H.-J. Ferenz, J. Chem Ecol., 26 (2000) 1897.
159 R. Wegener, S. Schulz, T. Meiners, K. Hadwich and M. Hilker, J. Chem. Ecol., 27
(2001) 499.
160 B.E. Hibbard and L.B. Bjostad, J. Chem. Ecol., 14 (1988) 1523.
161 B.E. Hibbard and L.B. Bjostad, J. Econ. Entomol., 82 (1989) 773.

697
Chapter21

Sampling and sample preparation for

fragrance analysis
Fabio Augusto and Claudia Alcaraz Zini

21.1 INTRODUCTION

The term "odor" refers to the biological, physical and psychological effects
caused by the interaction between chemical stimulants-aromas and fra-
grances-and the olfactory systems of living creatures [1]. The study of the
chemical composition of fragrances and aromas is relevant to several fields. In
food science, chemicals or their blends associated with odors discerned as
desirable or agreeable are related to the multivariate set of sensorial responses
known as "flavor" [2]. They are the predominant elements in the perception of
food and beverage quality [3] and, therefore, its acceptance by consumers. Also,
"off-flavors"-connected to volatile substances with unpleasant odors-may be
caused by microbial contamination of foodstuffs [4]. The investigation of these
substances can be an important tool for food safety studies.
Odor can be a significant environmental parameter in some situations, since
it can be related to the human perception of comfort and to the sanitary
conditions of an indoor atmosphere [5], to the contamination of air resulting
from agricultural [6] and industrial [7] activities, as well as to the quality of
natural and treated water [8,9].
Chemical characterization of fragrances released by vegetables and animals
is important to several branches of industry and science. Natural essential oils
from plants are the traditional base ingredients for perfumes and related toiletry
products. However, the rarity of several plants and possible dermatotoxic effects
caused by some of the natural products cause a demand for synthetic alternative
mixtures. The designing of these "synthetic" essential ingredients depends
fundamentally on reliable analytical data on the composition of their natural
correspondents [10]. Moreover, knowledge about the composition and emission
dynamics of floral scents and similar biogenic mixtures of volatile organic
compounds is fundamental in several biological studies due to their multiple
roles in plant reproductive processes, defense against predators and in intra-
species communication [11].

ComprehensiveAnalytical Chemistry XXVII

J. Pawliszyn (Ed.)
C 2002 Elsevier Science B.V. All rights reserved 699
TABLE 21.1
Low odor threshold concentrations (OT) for selected odor-producing substances
Odorant OT Ref.
Ethyl 4-methyl-4-pentenoate 0.06 (nll') 12
Ethyl 3-methyl-2-butenoate 0.22 (nl-l') 12
Methyl (E)-2-methyl-2-butenoate 0.22 (nl.lt') 12
2-Iodophenol 1 g. - ' 13
4-Iodophenol 500 Ag l-' 13
2-Methyl-3-furanthiol 0.4 ng[ l' 14
3-Methylbutanal 0.2 nl1' 15
Butyric acid 240 nl-ll' 15
Phenylacetaldehyde 4 nl.l' 15
2-Ethyl-3,5-dimethylpyrazine 0.04 nl.l-' 15
Methional 0.2 nl.lt' 15
Dichloroiodomethane 5.8 tg- ' 16
Iodoform 0.03 nll'1 16
2-Ethyl-3,5-dimethylpyrazine 0.007 ng[l' 17

The development and application of methodologies for the determination of

the chemical composition of aromas and similar mixtures is a challenging task.
This is a consequence of some inherent general properties of such samples:
(a) The concentration of relevant analytes in fragrance samples can be
extremely low. Table 21.1 shows reported literature values for low odor
threshold concentrations(which can be defined as the minimum concentration
for an odor-producing substance that has a 50% probability of being perceived as
a faint odor by a human subject [1]) for selected substances important in food,
biological and environmental aromas.
The typical perceptible concentration for some odorants, as exemplified in
Table 21.1, can be less than 1 ng 1l- of the substance. In consequence, the corre-
sponding analytical procedures must provide extremely high sensitivities, ade-
quate to the detection and quantification of these species at such low levels.
(b) Besides the typical low concentration of the odor-active analytes, most
aromas are extremely complex blends of substances. Kotserides and Baumes
[18] reported 48 different organic volatile substances as impact odorants in
Bordeaux Cabernet Sauvignon and Merlot wines. In a similar study, Reiners and
Grosch [19] found 41 compounds as odorants responsible by the aroma of
Italian, Spanish and Moroccan virgin olive oils. The same complexity can be
found in floral scents: Bayrak and Akgiil [20] identified more than 60 volatile
organic compounds in samples of essential oils from Turkish rose (Rosa
damascena Miller).
(c) For aromas generated by biological sources such as plants and animals,
further analytical difficulties arise from the dynamic nature of such systems. For

700
example, the composition of samples obtained from detached parts of plants may
not correspond to the actual mixture released by undisturbed live organisms
[21-23], which causes a demand for methods allowing in vivo sampling. Also, the
fact that the production and emission of plant volatiles can be affected or
triggered by factors such as light, environmental temperature, stress and the
presence of trace atmospheric pollutants [24,25], can introduce further
complications.
(d) Some odorants have limited chemical stability, due to photolysis, oxida-
tion and other reactions; e.g., the atmospheric chemical lifetime of mono-
terpenes under daylight conditions was estimated as ranging from less than 5
min (a-terpinene) to 3 h (a- and P-pinene, sabinene) [24].
In consequence, chemical characterization of fragrances and aromas (F&A)
normally demands state-of-the-art techniques for sampling and sample prepara-
tion, analyte separation, detection and quantitation. Usually, the application of
an analytical separation technique in the final steps [2] is involved. Gas chroma-
tography coupled to mass spectrometry (GC-MS) and other similar detection
schemes, such as infrared absorption spectrometry [26], are the conventional
devices employed for F&A chemical analyses. The coupling of olfactometric
detection to gas chromatography (GC-O) is also extremely relevant for qualita-
tive and quantitative chemical analysis of fragrances [27]; several instrumental
and methodological variants of GC-O are described in the literature, such as
CHARM-Analysis and aroma extraction dilution analysis (AEDA) [28]. Along
with the chromatographic techniques, use of "electronic noses" (arrays of elec-
trochemical sensors that generate an electric signal which emulates the expected
response from the human olfactory system) has been growing in recent years.
The development of these fascinating devices involves the arrival of adequate
new materials, such as piezoelectric crystals and synthetic conductive polymers,
as well as sophisticated data processing and chemometric techniques [29]. Their
more remarkable features, from the analytical standpoint, are the possibility of
fast direct qualitative and quantitative evaluation of fragrances and aromas
with limited or no preliminary sample preparation procedures, although the low
sensitivities provided by the devices presently available still prevent their use in
solving most F&A analytical problems.
Except for applications involving measurement with the above-mentioned
electronic noses and similar techniques, adequate isolation and pre-concentra-
tion of the odor-active analytes are mandatory and critical steps in the methodol-
ogies for chemical characterization of fragrances and aromas [2]. Several
sorbent-extraction approaches-classical liquid-liquid extraction (LLE), solid
phase extraction (SPE), solid phase microextraction (SPME) and supercritical
fluid extraction (SFE)-have been employed for sample preparation in F&A
analyses, as well as dynamic and static headspace methods and procedures based
on analytical distillation. In the following sections, the application of some of
these techniques to procedures for chemical characterization of aromas, flavors
and fragrances will be addressed.

701
21.2 ANALYTICAL DISTILLATION PROCEDURES

Distillation is usually carried out in two slightly different ways: in the first, the
matrix to be extracted is mixed or suspended with water in a suitable vessel
fitted with a condenser and, while the mixture is boiled, a condensate phase is
collected (hydrodistillation). After the process, an organic, water-insoluble
fraction (for vegetable materials, the essential oil) can be separated from the
water. In the second procedure, steam is passed through a vessel containing the
matrix-water mixture (steam distillation) to yield a similar condensate. In
either case, an important parameter affecting the composition of the condensate
is the pH of the water [30].

21.2.1 Vapor distillation

Steam-assisted distillation and hydrodistillation are traditional procedures for

isolation of volatile aroma-related compounds from odoriferous samples such as
food and detached parts of plants. Being simple and straightforward procedures,
they are still extensively applied for F&A characterization, either alone or com-
bined with other sample preparation procedures. For example, Saritas et al. [31]
isolated the essential oils from aromatic lichens of various genera (Mnium,
Plagiomnium and others) by hydrodistillation of fresh and dried plant parts.
Using GC-MS combined with 3C-NMR and other techniques, they were able to
identify several volatile terpenoid and aliphatic compounds in these hydro-
distillates, including two unreported volatile sesquiterpenes ((+)-10-epi-
muurola-4,11-diene and 10,11-dihydro-a-cuparenone). Other samples studied
recently after isolation of their essential oils by hydrodistillation include Cuban
rose geranium (Pelargonium sp.) [32], Brazilian camphor trees (Cinnamomum
camphora Nees & Eberm) [33], Moroccan rosemary (Rosmarinus officinalis L.)
[34] and several others.

21.2.2 Simultaneous distillation-extraction (SDE)

A widespread distillation-based sample preparation method for chemical analy-

sis of F&A is Simultaneous Distillation-Extraction (SDE), also known as the
Lickens-Nickerson method. Figure 21.1 shows one of the proposed designs for
the Lickens-Nickerson SDE apparatus. An amount of the fragrance-generating
sample (food, beverage, sliced plant tissue, etc.), along with distilled water for
dry samples, is contained in flask B, and flask C receives a convenient volume of
an extracting solvent denser than water (dichloromethane, chloroform, etc.).
Flasks B and C are heated; the water and solvent vapors are conducted to the
extractor body, where they are allowed to condense over the surface of the cold
finger D. In this operation, aroma analytes are removed from the matrix by the
water vapor and transferred to the organic phase when the liquids condense
together on the cold finger. Both water and solvent are collected in the extractor

702
Fig. 21.1. Likens-Nickerson simultaneous distillation-extraction
apparatus (modified from Perpbte et al. [38]): 1, body; 2, sample flask; 3, 3
extracting solvent flask; 4, cold finger; 5, inlet for purge gas; 6, cold water
inlet and outlet.

body after their condensation and return to the corresponding flasks, allowing
continuous reflux. A modified apparatus allows use of solvents lighter than
water, such as pentane or ethyl acetate, as extractors. Among others, this device
was employed by Schlotzhauer et al. [35] to study insect attractors in the
fragrance of flowers from Japanese honeysuckle (Lonicera japonica). Large
amounts of linalool and a-farnesene were found in samples collected during
daylight periods. Perpete et al. [36] combined liquid-liquid extraction and SDE
to find the key compound responsible for the worty off-flavor in alcohol-free
beers (1-methylpropionaldehyde), which was not previously found in reports
using other sample preparation techniques such as dynamic headspace analysis.
Fragrances from fresh and processed plums [37], hops [38] and y-ray irradiated
mushrooms [39] were also studied using SDE followed by GC analysis of the
extracts.

21.2.3 Other distillation methods

Atmospheric pressure distillation-based sample preparation methods may have

some serious disadvantages when applied to F&A characterization, since the
temperatures usually necessary for the operation can lead to degradation of
some analytes. For example, after studying the aroma from saffron (Crocus
sativus L.), Tarantilis and Polissou [40] found that the most important odorant
in these samples-safranal--is oxidized during conventional steam distillation,
forming artifacts such as 2,6,6-trimethyl-1,3-cyclohexadien-l-carboxylic acid.
Other compounds found in these distillates were also determined as artifacts
resulting from the thermal degradation and oxidation of carotenoids from the
saffron matrix. The same problems can be observed in SDE: Siegmund et al. [41]
discovered that 5,6-dihydro-2,4,6-trimethyl-4H-1,3,5-dithiazine, usually pointed
out as an important aroma-active compound in cooked and cured meat aromas,
is actually an artifact formed during the distillations performed according the
Lickens-Nickerson SDE method. Vacuum distillation, which can be performed
under milder conditions, can overcome these problems and is an alternative to
regular distillation methods. Moio et al. [42] combined vacuum distillation and
liquid-liquid extraction to identify the key odorants from Gorgonzola cheese.

703
After GC-MS and GC-O analysis of the distillates/extracts, 2-nonanone,
1-octen-3-ol, 2-heptanol, ethyl hexanoate, methylanisole and 2-heptanone were
determined to be the most relevant odorants on the aromas of natural and
creamy Gorgonzola cheeses. Other interesting application of low-pressure distil-
lation was described by Bouchilloux et al. [43]: using high-vacuum distillation
(temperatures between 35 and 45°C and pressures between 0.5 and 1.5 Pa) and
simultaneous derivatization with p-hydroxymercuribenzoic acid, the authors
were able to determine trace levels of some powerful odorous thiols in Bordeaux
red wines. The analytes were identified and quantified in the samples as their
volatile organomercury derivatives, in concentrations ranging from 0.25-10
-
ggl-l (3-mercapto-2-methylpropanol), 10 ngl-l to 5 tgl1l (mercaptohexanol) and
1-200 ngl- (3-mercaptohexyl acetate). Another alternative is microwave-
assisted distillation, which has been described recently for analysis of the
off-flavors geosmin and methylisoborneol in catfish tissue, in combination with
either adsorbent trapping [44] or SPME [45].

21.3 HEADSPACE METHODS

Procedures based on the manipulation of the headspace in contact with odorous

materials are popular and extremely convenient for chemical analysis of F&A
[46,47]. Different approaches have been employed, either using direct HS
analysis or by collection of the odorants in the HS using sorbent devices or cold
traps. These techniques and their applications for the chemical characterization
of F&A are discussed in the following sections.

21.3.1 Static headspace analysis (SHS)

The simplest way to assess the chemical composition of an aroma is direct analy-
sis (by GC or another convenient technique) of a portion of the air in contact
with the odor source, without any other sample treatment step. However, since
little or no pre-concentration of the analytes is involved and, considering the typ-
ical trace or ultratrace levels of the components, usually it is not feasible to apply
SHS to chemical fragrance analysis [48]. For example, out of 118 references
indexed in an extensive literature review on headspace analysis of floral fra-
grances [49], only nine employed SHS as the sampling or sample preparation
procedure. Nonetheless, when techniques and devices with adequate detection
limits are available, and depending on the concentration of the analytes, SHS
can be particularly suitable due to its inherent simplicity. For example, Clarkson
and Cooke [50] employed GC-MS and GC-AED (Atomic Emission Detection) sys-
tems equipped with high-volume injection ports to analyze aroma compounds in
the headspace of commercial cigarettes. Injecting 1 ml of the HS air in contact
with the samples, the authors pointed out that, for these samples, both analysis
time and detection limits were better than those obtained using a commercial
dynamic headspace analyzer.

704
21.3.2 Dynamic headspace analysis (DHS)

A large number of reports describing the use of DHS methods can be found in the
literature regarding the chemical characterization of F&A. Due to the constant
depletion of the analytes from the sample or from the adjacent atmosphere, the
general approach on DHS potentially provides improved analytical sensitivity
when compared to SHS and other equilibrium extraction procedures [2].
Desorption of trapped analytes for subsequent analysis can be performed either
with small volumes of adequate solvents [51] or using on-line automated therm-
al desorption devices [52], the later being eminently suitable for routine
procedures.
Experimental set-ups such as that depicted in Fig. 21.2 [53] can be used,
either in the laboratory or under field conditions, to collect fragrance compounds
emitted by live plants in dynamic conditions. A vase containing a live plant or
some of its parts can be isolated from the environment by a glass bell or similar
device. A controlled flow of purified air is passed through the chamber, carrying
the aroma compounds to a tube containing a sorbent or to a cold finger, where
the volatiles are accumulated for further desorption and analysis. Using a
similar apparatus with polyvinylacetate bags as the isolating device instead of
the glass bell, to emulate the usual procedure employed in field sampling of floral
scents, Raguso and Pellmyr [54] performed an extensive study on the method-
ological aspects of sampling and pre-concentration of floral fragrances by DHS
with solvent desorption and chromatographic analysis. Live flowers of Clarkia
breweri were used as fragrance source, as well as synthetic mixtures of typical
flower odorants. It was found that Porapak Q was more efficient for than Tenax
TA or Tenax TA/charcoal mixtures as sorbent for the evaluated fragrances.
However, the relative abundance of the detected compounds-a useful
parameter in biochemical and ecochemical studies-were not dependent with
the composition of the sorbent. The trapped compounds were desorbed using
either diethyl ether, dichloromethane or hexane; the last solvent provided better
recoveries. In the range studied (0.5-1.0 min-'), the effect of the stripping air
flow rate on the recovery of the trapped analytes generally was not significant

At
Fig. 21.2. Chamber for DHS isolation and collec-
tion of volatiles emitted by living plants (modified
from Baltussen [53]).

705
when compared to other sources of variation on the results between experi-
ments, such as variation of the composition of the volatile emission between
different specimens of the same plant. Sample collection and pre-concentration
by DHS with solvent desorption and GC-MS analysis was also the approach
adopted by Helsper et al. [25] to assess the circadian emission profiles of volatile
compounds by live "Honesty" roses (Rosa hybrida). A similar approach was
adopted by Christensen et alii [55] in a study of the fragrance ofjasmine flowers
(Jasminumpolyanthum) irradiated with y-rays in an attempt to produce floral
fragrances with reduced dermatotoxic or allergic properties. Live flowers were
enclosed in glass bulbs and the volatiles released in 2 h periods carried to
Porapak Q cartridges by a 200 ml-min 1 air stream. The entrapped compounds
were desorbed with 2 ml of purified pentane and the extracts analysed by
GC-MS, GC-FID and GC-O. It was found that the major odorants in the head-
space of non-irradiated flowers were linalool, benzyl acetate, p-cresol, iso-
eugenol, eugenol, 2-methoxy-p-cresol, phenethyl acetate; benzyl alcohol,
(Z)-3-hexenol, (Z)-3-hexenyl acetate and butyrate also have some impact on the
fragrance. No qualitative difference was observed between the fragrance
composition of irradiated and non-irradiated specimens; however, the relative
concentrations of benzenoid and phenylpropanoid odorants such as iso-eugenol
and 2-methoxy-p-cresol were reduced, as well as the overall production of
volatile compounds per plant after irradiation.
Although it can demand more complex and expensive hardware (prices up to
US$15,000) and cannot be applied to fragrances containing temperature-sensi-
tive compounds, DHS with thermal desorption (DHS-TD) has several advan-
tages over solvent desorption: simpler procedure, improved detection limits and
no interference of solvent peaks in the chromatograms. Ageloupoulos et al. [56]
monitored the temporal emission pattern of the volatile compounds (unsatu-
rated C 6 -alyphatic alcohols and aldehydes) released by single intact and damaged
leaves of potato (Solanum tuberosum) and broad bean (Vicia faba) by DHS
(using a specially designed sampling chamber) coupled to GC-MS. The volatiles
were trapped in Tenax TA cartridges and desorbed either thermally (inserting
the cartridges inside the programmed-temperature injector of the GC-MS sys-
tem) or by a solvent (diethyl ether). It was determined that TD, which provided
better detectability and therefore demanded reduced sampling times, allowed
assessing the time emission profiles for the monitored compounds. Vercammen
et al. [57] also employed a commercial automated on-line thermal desorption
apparatus (ATD) coupled to a GC-MS to compare sorption tubes packed with
Tenax TA and with pure grinded polydimethylsiloxane (PDMS), as well as static
headspace SPME, for DHS collection of fragrance compounds released by live
roses and jasmine. PDMS was found to be more efficient for the isolation of vola-
tile odorants released by these plants, resulting also in cleaner chromatograms.
Use of DHS with ATD devices is also predominant for analysis of the fragrances
of detached parts of plants. Mazza and Cottrell [58] determined that the preva-
lent HS components in the aerial parts of three species of Echinacea tradition-

706
ally employed as medicinal plants by North-American native peoples (E.
angustifolia, E. pallida, and E. purpurea) were -myrcene, a- and 3-pinene,
limonene, camphene, trans-ocimene, 3-hexen-l-ol and 2-methyl-4-pentenal. For
the roots, the HS contained mainly a-phellandrene, dimethyl sulfide, 2-methyl-
butanal, 3-methylbutanal, 2-methylpropanal, acetaldehyde, camphene,
2-propenal, and limonene. The volatiles were trapped in Tenax TA cartridges
and analyzed by GC-MS after thermal desorption. Tenax TA was also deter-
mined as adequate for collection of organic volatiles in milk by DHS-TD [59].
Analytical methodologies involving DHS-TD for characterization of the
aroma of foods and beverages are also common. To study the release of volatile
flavor compounds by leaves of Japanese pepper (Xanthoxylum piperitum), a
spice employed in traditional oriental cuisine, Jiang and Kubota [60] collected
the volatiles in the headspace of crushed, mechanically disturbed or intact leaves
in Tenax TA cartridges. The analytes were thermally desorbed in the injector
port of a GC-MS, where they were cryofocused, separated, detected and quanti-
fied (Fig. 21.3). It was determined that the number of detected volatile
compounds increased from 12 (from intact leaves) to 22 (mechanically disturbed
5

s4 intact leaves
c

10 20 30 40 50 60

4 disturbed leaves
a
3
C

2
C
0C
IC

U
10
I . 1.
26
I).i
I

30
. J 40 56 60

I . crushed leaves
C I
3

Fig. 21.3. GC-MS chromatographic Fig.GC-MS

21.3.
chromatographic
a'
profiles after DHS-ATD collection of
volatiles released from Japanese
pepper (Xanthoxylum piperitum)
leaves (modified from Jiang and
Kubota [60]).
1

oJ
1t
I
;

20
w

30
. 11
-t
40
Time / min
-----I-L-
Ali...
50 60

707
leaves) and then to 36 (crushed leaves). For the disturbed leaves, it was found
that the main compounds responsible for the characteristic aroma were limo-
nene (0.56 g.ggl) and (Z)-3-hexen-l-ol (0.63 bug.gl). For crushed leaves, higher
concentrations of C6 -aldehydes, especially (Z)-3-hexenal (17.88 tggl), and
hexanal (6.77 ug.g-), imparted an undesirable grassy odor to the samples.
The combination of DHS, thermal desorption and other sample preparation
methods is occasionally described. Bylait6 et al. [61] studied the composition of
the volatile fraction from different parts of lovage (Levisticum officinale, Koch),
a plant widely employed in the perfume industry. Essential oils from different
parts of the plant (leaves, stems, flowers, seeds, and roots) were prepared by
hydrodistillation and the volatile fractions of these oils were isolated by
DHS-TD, using Tenax TA as sorbent. They found 98 compounds in the volatile
part of the oils, including 41 unreported substances for this plant (e.g.,
2-methyl-2-propenal and carvacrol); the most concentrated compound in all
samples was a-phellandrone.
Introduction of artifacts produced by degradation of the sorbent can be a
major difficulty in DHS-TD; e.g., Canac-Arteaga et al. [62] found out that the
interaction of water vapor, volatile compounds and the trapping material (Tenax
TA) can produce artifacts during the analysis of aromas of dehydrated cheese
and Parmesan cheese. Retention of water by the trapping material can be also a
potential problem, especially when it is necessary to cryofocus the analytes in
the chromatographic column before separation (moisture can freeze and clog the
column). Carbon molecular sieves (Carbosieve, Carboxen) can retain large
amounts of water, as opposed to non-polar polymeric sorbents such as Tenax and
graphitized carbon blacks. Polar polymers as Porapak T and N can also hold
appreciable amounts of water which, however, can be easily removed prior to
analyte desorption by a current of dry air [63].
Introduction of products from the thermal degradation of adsorptive
materials can be avoided in DHS procedures by using cryotrapping. StrAnsky
and Valterovd [64] determined the emission profile of compounds related to the
putrid fragrance from Hydrosme rivieriflowers during their flowering period. A
single live specimen was enclosed in a glass cylinder, which was purged by 15
ml-min - ' of pure air. The compounds released were trapped in a U-shaped tube
containing methanol at -78°C. It was found that emission of dimethyl di- and
trisulphides started after 70 h of flower enclosement, peaking at 118 h and
ending after 142 h (Fig. 21.4). Before 70 h of enclosure, only n-alkanes were
detected. To eliminate water co-volatilized with the analytes, Kolb et al. [65]
proposed a modification in a commercial on-line DHS-cryotrapping device by
incorporation of a water trap (glass-lined steel tube packed with 65% LiCl on
Chromosorb W/AW). The authors reported good results in the application of this
technique to several processes, including the characterization of aroma from
ground coffee. An interesting variation of DHS with cryotrapping is Short-Path
Thermal Desorption (SPTD) [66]: a small amount of solid sample (food or plant
part) is placed in a glass-lined stainless-steel tube, which is directly connected to

708
 2-

1-

0-

Fig. 21.4. Profile of emission of a -1-

U
volatiles from a single Hydr- E
osme rivieri flower (modified a'
o -2-
from Strdnsk& and ValterovA
[641) after DHS-cryotrapping
sampling and GC-FID analysis. -3-
Compounds: * = dimethyl
disulphide; · = dimethyl
trisulphide; 0 = n-decane; O =
n-undecane; f = n-dodecane o 1 2 3 4 5 6
Time / Days
and A = n-tridecane.

the cooled tip of the chromatographic column. The sample inside the tube is
carefully heated and the volatile F&A compounds released are directly trapped
and cryofocused at the entrance of the column.

21.3.3 Headspace solid phase microextraction (HS-SPME)

SPME, a fast, simple and convenient sample preparation method introduced in

1990 [67], has gained increasing popularity for F&A analysis, especially as an
alternative to DHS methods. It is specially suitable for qualitative and quantita-
tive analysis of fragrance compounds released by odorous samples, only demand-
ing exposure of a fiber to the HS above the sample for a convenient period of
time, followed by direct thermal desorption on the heated injection port of a GC
[68]. For these reasons, HS-SPME is being presently considered as the best
available choice for sample preparation in fragrance and F&A analysis [2].
Due to the small dimensions of the sampling device and to the simplicity and
speed of the extraction procedure, HS-SPME is able to collect fragrances from
live plants with minimum disturbance of the specimen, under both laboratory
and field conditions. Vereen et al. [69] employed HS-SPME and GC-MS to study
volatile compounds released by intact and mechanically damaged leaves of
Fraser firs (Abies fraseri) in the field. Branches of fir were enclosed inside 100 ml
Tedlar bags and the air inside was extracted with 100 Am PDMS fibers over
periods of up to 4 h. After 5 min extractions, monoterpernes such as 3-carene are
predominant; for 3 h extractions, the major component in the chromatograms is
bornyl acetate, with minor amounts of heavier compounds (e.g., camphor and
borneol). Two species (-phellandrene and y-terpinene), detected in the
fragrance exhaled from damaged leaves, were assigned as wound response
compounds. However, inadequate precision (RSD > 20%) and slow equilibration
times for heavier compounds prevented quantitative application of HS-SPME to

709
 23
8

Fig. 21.5. Glass chamber for SPME

sampling of volatiles emitted from single
live leaves (based on Zini et al. [70]). 1,
Silanized glass cylindrical body; 2, silan-
ized glass lid; 3, SPME sampling holes
topped with silicone septa; 4, SPME holder
+ fiber; 5, DC power supply for microfan;
6, microfan; 7, Teflon support for micro-
fan: 8. Teflon tane seal.

these samples. Zini et al. [70] employed HS-SPME and GC-ITMS (ion trap mass
spectrometry) to assess the emission profiles from intact and mechanically
damaged leaves of living Eucalyptus citriodora trees, using the sampling
chamber shown in Fig. 21.5 and extractions with PDMS/DVB fibers performed
every 30 min for continuous periods between 8 and 10 h. The main compounds
identified were isoprene, citronellal, citronellol and P-caryophyllene. Different
patterns of dependence between extracted amounts and leaf enclosure times
were observed; e.g., for rose oxide (cis-4-methyl-2-(2-methyl-l-propenyl)-tetra-
hydropyran) a maximum in the area versus time curves appears after 300-400
min of leaf enclosure while for citronellal the peak areas decay exponentially
from the start of the experiments. A similar study was conducted on flowering
Boroniamegastigma plants by MacTavish et al. [71]. Volatiles from specimens
in vases placed inside glass or steel vessels were extracted for 30 min with 100
/Am PDMS fibers and analyzed by GC-MS and GC-FID. The principal substances
identified were a-pinene, 5-acetoxylinalool, dodecyl acetate, Z-n-heptadec-8-ene
and f-ionone isomers. From the examination of emission profiles under varied
illumination conditions during 26 h periods of plant enclosure (Fig. 21.6), it was
determined that the pattern is both endogenously and environmentally
controlled.
HS-SPME also has been extensively employed in recent applications related
to food flavors. Roberts et alii [72] studied methodological and operational
aspects of application of HS-SPME to analysis of volatile flavor compounds. For
coffee aroma, best overall sensitivity was achieved using fibers coated with 65
/gm PDMS-DVB; this fiber was also found as the most adequate for heavier polar
odorants such as vanillin. Carbowax-DVB fibers were found to be suitable for
-
organic acids. Non-polar odorants were detectable at the /g.1 levels, were polar
compounds could generally be detected only at mg.l-l levels; e.g., in the extrac-
tion of coffee aroma with PDMS-DVB fiber for 5 min followed by GC-MS analy-
-
sis, 7.7 mgl 1 l vanillin, 116 mgl-1 furaneol, 2.3 mgl l ethylguaiacol, 5.7 mgl-l
-
guaiacol, 40 mgl-l 4-vinylguaiacol, 0.2 mgl-l P-damascenone, 0.4 mg-l ' 2,3-di-
ethyl-5-methylpyrazine and 1 mg l-' 2-ethyl-3,5-dimethylpyrazine were found.
In a similar study conducted by Augusto et al. [73], different SPME fibers were
compared for the characterization of the aromas of industrialized pulps of Bra-

710
 A
U

m
a.
U
U
U

Time / min
Fig. 21.6. Dependence between corrected peak areas (peak area per plant) and plant enclosement
time for Boronia megastigma flowering plants (modified from MacTavish et alii [71]). A,
5-acetoxy-linalool; B, a-pinene; 0, continuous darkness; *, continuous illumination; A, A,
alternating darkness/illumination.

zilian tropical fruits: cupuassu (Theobroma grandiflorum, Spreng.), cajd

(Spondias lutea, L.), siriguela (Spondias purpurea, L.) and graviola (Anona
reticulata, L). Best overall efficiencies for these aromas were obtained with
Carboxen-PDMS fibers, especially for low-molar mass compounds. Other fruit
aromas studied by HS-SPME include kiwi [74], raspberry [75], banana [76],
dried plum [77] and cantaloupe [78].
Another area where HS-SPME is quickly becoming favored is in studies of
the aroma of alcoholic beverages, essential in the assessment of the quality of
such products and in the optimization of the production routines. Sala et al. [79]
developed an HS-SPME methodology for determination of the aroma-active
heterocycles 3-alkyl-2-methoxypyrazines in musts from Cabernet Sauvignon
and Merlot grapes. Levels as low as 0.1 ng l-' of the analytes could be detected
after extraction of the sample HS with PDMS-DVB fibers and GC-NPD (nitro-
gen-phosphorous detection) analysis of the extracts. Weber et al. [80] showed
that chromatographic profiles of aroma obtained by HS-SPME and GC-MS can
be useful to classify wine samples according to production area, grape variety
and vintage. Nonato et al. [81] compared HS-SPME and conventional liquid-liq-
uid extraction (LLE) in the analysis of secondary aroma compounds (e.g., esters
and heavy alcohols) in Brazilian sugar-cane spirits (cachaqa). The precision in
the quantification of the target analytes with SPME was better when compared
to the standard LLE procedure. HS-SPME was also successfully applied to
quantify dimethyl sulfide (a powerful off-odorant) in beer aroma [82].
Along with these reports, a new HS-SPME approach for food flavor analysis
was recently proposed by P6res et al. [83]. To characterize the ripening of
Camembert cheese, the HS in contact with a 2 g sample contained in 10 ml flasks
was exposed for 10 min to a 75,um Carboxen-PDMS SPME fiber. In preliminary
assays, the extracted materials were analyzed by GC-MS for identification of

711
odorant compounds. For subsequent experiments, the extracted compounds
were thermally desorbed from the fiber and directly introduced into the ioniza-
tion chamber of a mass spectrometer without chromatographic separation. More
than 60 compounds were identified in the HS-SPME extract by GC-MS;
Carboxen-PDMS fiber was found to be especially efficient for collecting sulfur
compounds (e.g., dimethyl sulfide, methyl thioacetate), which are important
impact odorants for this cheese variety. The HS-SPME-MS data obtained
without chromatographic separation was statistically processed and found to be
valuable for qualitative chemometric classification of the samples.
Some drawbacks of HS-SPME should be considered. Since the extracted
amounts are limited by the low mass of sorbent present in the commercial fibers
(less than 1 mg coating on a 100 gm PDMS SPME fiber), the quantification
limits may be higher when compared to DHS methods [57]. However, significant
increases in analyte recovery can be achieved using schemes such as that
proposed by Zhang and Pawliszyn [84]. Using SPME fibers modified to be cooled
by an internal flow of liquid CO2 (which improves the efficiency of the transfer-
ence of analyte from the headspace to the fiber), these authors report near
quantitative extraction of aromatic hydrocarbons by HS-SPME and results
similar to those achieved with a DHS-ATD device. Another problem, related to
the use of fibers with mixed adsorptive coatings (Carboxen-PDMS and
PDMS-DVB) is the lack of analytical accuracy caused by competition between
the analytes for the active sites available on the fiber [85]. When these materials
are employed, extractions with very short exposures of the fiber to the sample
should be adopted to circumvent these problems [86].

21.4 DIRECT EXTRACTION PROCEDURES

Methods involving application of LLE, SPE or SFE directly to the samples to

isolate odorants are sometimes found in the recent literature. Although HS-
based methods are usually more convenient for fragrance analysis, in some cases
direct extraction of the target analytes from the matrixes is necessary; e.g.,
species with high odor impact and in extremely reduced concentrations can be
important to some F&A, but may not be detected using HS methodologies if
their volatility is not high enough to provide a convenient concentration in the
sample HS [1].

21.4.1 Liquid-liquid extraction (LLE)

Despite its simplicity, the modern tendency is to replace LLE by other tech-
niques, due to the necessity for high purity solvents for trace analysis and to
reduce the environmental and health risks associated with their manipulation.
Also, in the case of biological F&A, they cannot be applied to live samples.
However, there are several contemporary reports on the use of this technique for
F&A analysis, especially for collection of preliminary data [87]. L6pez and

712
G6mez [88] addressed some operational parameters on the application of LLE to
extract aroma compounds from wines. Several solvents-diethyl ether, n-pent-
ane, Freon-11, n-hexane, 1:1 ether/pentane, 1:1 ether/hexane and dichloro-
methane-were compared for extractions of odoriferous terpenic substances
from "artificial wine" (12% v/v ethanol in water). It was concluded that dichloro-
methane and 1:1 ether/pentane are the best solvents for extracting these
analytes from wine. In a related study, continuous LLE was employed by Rocha
et al. [89] to extract free and releasable odor volatiles from Portuguese Bairrada
white grapes employed in the local wine industry. A volume of 250 ml of raw or
enzymatically treated must was continuously extracted for 25 h with 75 ml of
dichloromethane at 50°C. The organic phase was dried and concentrated by
vacuum distillation prior to the GC-MS analysis. Chemometric analysis of the
data allowed differentiation of several grape varieties and of the enzyme treat-
ment studied. Conventional batch LLE is still also occasionally employed in F&A
analysis; recent examples include the identification of odorants in sugar beet
juice [90] and in Grenache Noir fortified wines [91].

21.4.2 Solid phase extraction (SPE)

SPE can be directly applied to isolate odorants from liquid or liquefiable odorif-
erous samples such as beverages, fruit pulps and tissues. A typical contemporary
application of SPE to aroma analysis was presented by Wada and Shibamoto
[92], who studied the direct extraction of odorants from Chablis red wine using
Porapak Q columns. Samples spiked with selected aroma analytes (2-methyl--
1-propanol, 3-methyl-l-butanol, 2-phenylethanol, ethyl hexanoate, ethyl octano-
ate, diethyl butanedioate and hexanoic, octanoic and decanoic acids) were passed
through 30 cm glass columns packed with the sorbent. Diethyl ether, dichloro-
methane and pentane were evaluated to desorb the extracted materials, which
were analyzed by GC-FID and GC-MS. Dichloromethane was found to be the best
desorbing solvent, recovering up to (103.0 + 3.7) % of the spiked analytes. Analy-
sis of unspiked samples revealed 67 volatile odor compounds in this variety of red
wine. SPE has also been applied to characterize butter aroma: Adahchour et al.
[93] assessed the use of SPE cartridges packed with different sorbents- C,,, C,
NH2 and CN-bonded silica, as well as SDB-1 (PS-DVB copolymer)-to extract
aroma compounds from this material. Butter samples were melted and the aque-
ous fraction, after separation of the fat by centrifugation, was passed through the
SPE cartridges. The cartridges were eluted by methyl acetate, which was dried
and analyzed by GC-MS. Best overall recoveries were obtained with SDB-1 car-
tridges (average recovery ca. 80%); detection limits for impact odorants such as
vanillin and diacetyl were in the low-pg range. For samples such as fruit pulps,
SPE can be combined with isolation techniques such as distillation, as exempli-
fied by Boulanger et al. [94]. After distillation of cupuassu pulp and extraction
with XAD-2 resin, the main odorants were identified as linalool, a-terpineol,
2-phenylethanol, myrcene, limonene, ethyl 2-methylbutanoate, ethyl hexanoate,

713
butyl butanoate, 2,6-dimethyl-oct-7-en-2,6-diol, (E)- and (Z)-2,6-dimethyl-octa-
2,7-dien-1,6-diol and methoxy-2,5-dimethyi-3(2H)-furanone.
Stanfill and Ashley [95] presented an interesting application of SPE to
analysis of flavor-related alkylbenzenes in cigarette smoke. Tobacco from
disassembled cigarettes was "smoked" in a smoking machine and the particulate
material produced collected with filter pads. The collected particulate was then
suspended in hexane and the target analytes extracted from these solutions with
CN-silica SPE cartridges. After elution with a hexane/toluene/tetrahydrofuran
mixture and concentration of the extract by vacuum evaporation, alkylbenzene
odorants were detected and quantified by GC-MS. Detection limits per cigarette
ranged from 1.1 ng (methyleugenol) to 27.8 ng (eugenol). This method was
tested to quantify aroma alkylbenzenes in commercial brands sold in the USA,
and levels of target analytes up to several micrograms per cigarette were found.

21.4.3 Supercritical fluid extraction (SFE)

Compared with the direct extraction methods discussed above, SFE presents
some advantages for fragrance analysis. For example, since the critical point for
CO 2 (the most popular and most convenient fluid for SFE) is 31.1° C at 7.38 MPa,
extractions can be carried out under milder temperature conditions and without
need of further aggressive procedures such as distillation of excess solvent, as is
common in LLE and SPE [96].
Due to its widespread application on industrial scale to process vegetable
materials, most of the analytical applications of SFE are related to fragrance
analysis of plant materials, using leaves, barks or flowers as samples [97].
Volatile compounds from flowers, leaves and stems of guaca (Spilanthes
americana, Mutis-a South American plant with medicinal and insecticide
properties) were isolated by Stashenko et al. [98] using both SFE and SDE, and
analyzed by GC-MS, GC-FID and GC-NPD. SFE was performed with CO 2 at
40-45°C and 7.24-7.53 MPa, using up to 30 g of vegetable sample. Extraction
time was 2 h and, after depressurization, the collected analytes were dissolved in
2 ml CH 2C12 . Compared to SDE, a larger number of compounds could be detected
at levels above 100 ppb in the SFE extracts (e.g., 67 analytes with SFE and 43
analytes with SDE, for floral extractions); the overall extraction yield for SFE
was up to twice as large of that for SDE. SFE was found to be especially effective
to isolate sesquiterpenes and nitrogenated compounds from these samples.
There are also several SFE applications reported for food flavor analysis.
Leunissen et al. [99] developed a SFE-GC-MS method for fast analysis of roasted
peanut aroma. For aroma isolation, 2.00 g to 2.50 g of ground roasted peanuts
were extracted for 10 min with CO2 at 50°C and 9.6 MPa. The conditions were
adjusted to minimize the extraction of greasy materials. The CO2 was passed
through a silica trap at -5°C to collect the extracted materials; the trapped
analytes were desorbed by dichloromethane and analysed by GC-MS. The char-
acteristic aroma compounds were found to be methylpyrrole, hexanol, hexanal,

714
2-furanmethanol, 2-furancarboxyaldehyde, benzeneacetaldehyde and several
alkylpyrazines. Quantification limits for these analytes were estimated to be in
the range between 80 ng.g (methylpyrrole) and 0.4 ng g (ethyl- and 2,3,5-tri-
methylpyrazine). Supercritical CO 2 was also employed by Calvey et al. [100] to
study organosulfur lacrimatory odorants in ramp/wild leek (Allium tricoccum,
Ait.) a native North American variety of garlic. Ramp bulb homogenates were
mixed with Hydromatrix (diatomaceous earth plus crystalline silica) and
extracted with 2 ml/min CO2 at 35°C and 24.3 MPa for 30 min. The extracted
materials were cold-trapped on glass beads, dissolved in methanol and analyzed
by LC-MS. A total of 24 sulfur compounds (thiosulfinates and cepaenes) were
detected in the samples.
Supercritical CO 2 is a poor extractor for polar substances; therefore, for such
analytes addition of modifiers or use of other fluids, such as supercritical water,
is advisable [2]. Jimdnez-Carmona et al. [101] isolated the aromatic essential oils
from marjoram leaves using supercritical water (2 mlmin-l of water at 150°C
and -5 MPa, for 0.4 g of ground sample), with further analysis by GC-FID.
Compared to hydrodistillation, the essential oils produced after supercritical
water extraction had a higher content of oxygenated monoterpenes, which are
the most significative odorants for this spice (Fig. 21.7). Kubatova et al. [102]
also compared water-SFE and CO2-SFE with hydrodistilation to separate
aromatic essential oils from savory (Satureja hortensis) and peppermint
(Mentha piperita). Extraction of savory with supercritical water for 40 min
removed almost 100% more thymol and carvacrol and 150% more borneol and
linalool than hydrodistillation. For CO2 extractions, yields of borneol and lina-
lool were similar than those for hydrodistillation; however, for thymol and
carvacrol the efficiency of CO,-SFE was only half of that for hydrodistillation.
Water-SFE was determined as highly selective for polar oxygenated flavor
compounds, when compared to CO,-SFE or hydrodistillation.

21.5 FINAL REMARKS

As in any analytical procedure, an adequate choice of sample preparation

technique is an essential step for accurate and reliable characterization of the
chemical composition of F&A. However, due to the peculiarities of these
samples-especially the common presence of thermally or chemically labile
analytes in trace or ultratrace concentrations-the selection of the analyte
isolation and pre-concentration technique, as well as the careful optimization of
the corresponding operational parameters, is of paramount importance.
Some of the general techniques discussed in the previous sections-- namely
distillation procedures, LLE and SPE-are still occasionally employed for F&A
analysis. However, for more critical analysis, the present tendency is their
replacement by methodologies which are simultaneously less aggressive to the
analytes and capable of dealing with them when present in ultra-low concentra-
tions in the samples. Most of the state-of-the-art, contemporary applications in

715
 (A)

75

e 50
8

,>
2 9 10
25 IS
6
3 !1
7

~~.. ...-... ~-..

< -= Mmsss
. - ---I - - 11

10 Time (min)

8 (RI
S-1
Fig. 21.7. GC-FID chromato-
grams of marjoram essential oil 75
extracted by hydrodistillation - 9
(A) and water/SFE (B) (from
Jim6nez-Carmona et al. [101]). .
Peak identification: 1, a-pinene; e 50o
2, -pinene; 3, ]-myrcene; e- ! 11
4, eucalyptol; 5, linalool; 6, 10
2-methyl-6-methylen-7-octen-2-
ol; 7, terpinen-4-ol; 8, a-terp- 25, 6
IS
ineol; 9, geraniol; 10, geranyl i
acetate; 11, 4-ethenyl-a,a,4-
trimethyl-3-(1-methylethenyl)-
cyclohexanemethanol; IS,
2 ;!? i ; i~~~-'
-
'-------'- L'-'.-'-...--' In 7n

cnromaLgapULnlc 1nirlal u Time (min)

standard.

F&A Analytical Chemistry are focused on the several variations of DHS

sampling or, more recently, on HS-SPME. Despite the fact that these techniques
have drawbacks (e.g., expensive hardware for DHS-ATD, limitations on the
extracted masses for HS-SPME), for nearly all cases in F&A chemistry, either
DHS or HS-SPME can be applied to obtain dependable results.
Perhaps the next frontier to be broken in this area is related to field sampling
and analysis. The absolute majority of the analytical work on plant fragrances in
general is still performed, totally or in part, in the laboratory and/or under in
vitro conditions (i.e., using detached parts of the plants as the source of
fragrance). Nonetheless, the adequate and realistic characterization of the
composition of the organic volatile emissions from these organisms can only be
made in their natural habitats, with minimal biological, chemical and physical

716
disturbance. Therefore, both adequate portable or semi-portable instruments
(such as GC-MS) and sampling procedures based on small and rugged hardware
are needed. On the sample preparation side, SPME seems to be the technique
that most closely fits these demands.

REFERENCES

1 N. Nuener-Jehle and F. Etzweiler, in: P.M. Miller and D. Lamparsky (Eds.),

Perfumes: Art, Science and Technology. Blackie Academic & Professional, London,
1991, p. 153.
2 A. Sides, K. Robards and S. Helliwell, Trends Anal. Chem., 19 (2000) 322.
3 G.V. Civille, J. Food Qual., 14 (1991) 1.
4 J.W. Arnold and S.D. Senter, J. Sci. Food Agric., 78 (1998) 343.
5 J. Pejtersen, H. Brohus, C.E. Hyldgaard, J.B. Nielsen, O. Valbjorn, P. Hauschildt,
S.K. Kjaergaard and P. Wolkoff, IndoorAir, 11 (2001) 10.
6 J.R. Miner, J. Animal Sci., 77 (1999) 440.
7 L.Y. Chen, P.T. Jeng, M.W. Chang and S.H. Yen, Environ. Sci. Technol., 34 (2000)
1166.
8 J. Noblet, L. Schweitzer, E. Ibrahim, K.D. Stolzenbach, L. Zhou and I.H. Suffet,
Water Sci. Tech., 40 (1999) 185.
9 I.H. Suffet, D. Khiari and A. Bruchet, Water Sci. Tech., 40 (1999) 1.
10 R. Kaiser, in: P.M. Miller and D. Lamparsky (Eds.), Perfumes: Art, Science and
Technology. Blackie Academic & Professional, London, 1991, p. 213.
11 N. Dudareva and E. Pichersky, Plant Physiol., 122 (2000) 627.
12 G.R. Takeoka, R.G. Buttery, L.C. Ling, R.Y. Wong, L.T. Dao, R.H. Edwards and J.J.
Berries, Lebensm. -Wiss. -Technol., 31 (1998) 443.
13 A.M. Dietrich, S. Mirlohi, W.F. Costa, J.P. Dodd, R. Sauer, M. Homan and J. Schultz,
Water Sci. Tech., 40 (1999) 45.
14 P. Bouchilloux, P. Darriet and P. Dubourdieu, Vitis, 37 (1998) 177.
15 R.G. Buttery and L.C. Ling, J. Agric. Food Chem., 46 (1998) 2764.
16 B. Cancho, C. Fabrellas, A. Diaz and F. Ventura, J. Agric. Food Chem., 49 (2001)
1881.
17 M. Czerny, R. Wagner and W. Grosch, J. Agric. Food Chem., 44 (1996) 3268.
18 Y. Kotseridis and R. Baumes, J. Agric. Food Chem., 48 (2000) 400.
19 J. Reiners and W. Grosch, J. Agric. Food Chem., 46 (1998) 2754.
20 A. Bayrak and A. Akgil, J. Sci. Food Agric., 64 (1994) 441.
21 L. Tollsten and G. Bergstrom, Phytochemistry, 27 (1988) 4013.
22 J. Takabayashi, S. Takahashi, M. Dicke and M.A. Posthumus, J. Chem. Ecol., 21
(1995) 273.
23 J.B.F. Gervlieet, M.A. Posthumus, L.E.M. Vet and M. Dicke, J. Chem. Ecol., 23
(1997) 2395.
24 J. Kesselmeier and M. Staudt, J. Atmos. Chem., 33 (1999) 23.
25 J.P.F.G. Helsper, J.A. Davies, H.J Boumeester, A.F. Krol and M.H. van Kampen,
Planta, 207 (1998) 88.
26 J. Auger, S. Rousset, E. Thibout and B. Jaillais, J. Chromatogr.A, 819 (1998) 45.
27 P. Pollien, L.B. Fay, M. Baumgartner and A. Chaintreau, Anal. Chem., 71 (1999)
5391.
28 W. Grosch, Trends Food Sci. Technol., 4 (1993) 68.
29 R.I. Stefan, J.F. van Staden and H.Y. Aboul-Eneim, Crit. Rev. Anal. Chem., 29 (1999)
133.
30 C.P. Milner, R.D. Trengrove, C.M. Bignell and P.J. Dunlop, in: H.F. Linskens and

717
 J.F. Jackson (Eds.), Modern Methods of Plant Analysis. Springer-Verlag, Berlin,
1997, Vol. 19, p. 141.
31 Y. Saritas, M.M Sonwa, H. Iznaguen, W. A. Knig, H. Muhle and R. Mues,
Phytochemistry, 57 (2001) 443.
32 J.A. Pino, A. Rosado and V. Fuentes, J. Essent. Oil Res., 13 (2001) 21.
33 C.D. Frizzo, A.C. Santos, N. Paroul, L.A. Serafini, E. Dellacassa, D. Lorenzo and P.
Moyna, Braz. Arch. Biol. Technol., 43 (2000) 313.
34 A. Elamrani, S. Zrira, B. Benjilali and M. Berrada, J. Essent. Oil. Res., 12 (2000) 487.
35 W.S. Schlotzhauer, S.D. Pair and R. J. Horvat, J. Agric. Food Chem., 44 (1996) 206.
36 P. Perpete and S. Collin, J. Agric. Food Chem., 47 (1999) 2374.
37 L.F. di Cesare, G. Cortellino and M. Proietti, Ind. Aliment. (Italy), 37 (1998) 14.
38 P. Perp6te, L. Ml1otte, S. Dupire and S. Collin, J. Am. Soc. Brew. Chem., 56 (1998)
104.
39 J.L. Mau and S.J. Hwang, J. Agric. Food Chem., 45 (1997) 1849.
40 P.A. Tarantilis and M.G. Polissiou, J. Agric. Food Chem., 45 (1997) 459.
41 B. Siegmund, E. Leitner, I. Mayer, W. Pfannhauser, P. Farkas, J. Sadecka and M.
Kovac, Food Res. Technol., 205 (1997) 73.
42 L. Moio, P. Piombino and F. Addeo, J. DairyRes., 67 (2000) 273.
43 P. Bouchilloux, P. Darriet, R. Henry, V. Lavigne-Cruege and D. Dubourdieu, J.
Agric. Food Chem., 46 (1998) 3095.
44 E.D. Conte, C.Y. Shen, D.W Miller and P.W. Perschbacher, Anal. Chem., 68 (1996)
2713.
45 S.W. Lloyd and C.C. Grimm, J. Agric. Food Chem., 47 (1999) 164.
46 H. Dobson, in: H. Linskens and J.F. Jackson (Eds.), Modern Methods of Plant
Analysis. Springer-Verlag, Berlin, 1997, Vol. 19, p. 231.
47 H. Steinhart, A. Stephan and M. Bficking, J. High Resolut. Chromatogr., 23 (2000)
489.
48 R.J. Stevenson, X.D. Chen and O.E. Millls, Food Res. Int., 29 (1996) 265
49 J.T. Knudsen, L. Tollsten and L.G. Bergstrom, Phytochemistry, 33 (1993) 253.
50 P. Clarkson and M. Cooke, Anal. Chim. Acta, 335 (1996) 253.
51 S.A. Rankin and F.W. Bodyfelt, J. Food Sci., 60 (1995) 1205.
52 E. Valero, E. Miranda, J. Sanz and I. Martinez-Castro, Chromatographia,44 (1997) 59.
53 H.A. Baltussen, New concepts in sorption based sample preparation for chromato-
graphy (Ph.D. Thesis), Technical University of Eindhoven, Eindhoven, 2000, p. 97.
54 R.A. Raguso and 0. Pellmyr, Oikos, 81 (1998) 238.
55 L.P. Christensen, H.B. Jakobsen, K. Kristiansen and J. Miller, J. Agric. Food Chem.,
45 (1997) 2199.
56 N.G. Agelopoulos, A.M. Hooper, S.P. Maniar, J.A. Pickett and L.J. Wadhams, J.
Chem. Ecol., 25 (1999) 1411.
57 J. Vercammen, P. Sandra, E. Baltussen, T. Sandra and F. David, J. High Resolut.
Chromatogr., 23 (2000) 547.
58 G. Mazza and T. Cottrell, J. Agric. Food Chem., 47 (1999) 3081.
59 B. Vallejo-Cordoba and S. Nakai, J. Agric. Food Chem., 41 (1993) 2378.
60 L. Jiang and K. Kubota, J. Agric. Food Chem., 49 (2001) 1353.
61 E. Bylaite, J.P. Roozen, A. Legger, R.P. Venskutonis and M.A. Posthumus, J. Agric.
Food Chem., 48 (2000) 6183.
62 D. Canac-Arteaga, C. Viallon and J.L Berdague, Analusis, 28 (2000) 550.
63 J. Gawlowski, T. Gierczak, A. Jezo and J. Niedzielski, Analyst, 124 (1999) 1553.
64 K. Stransky and I. Valterova, Phytochemistry, 52 (1999) 1387.
65 B. Kolbe, G. Zwick and M. Auer, J. High Resolut. Chromatogr., 19 (1996) 37.
66 J.G. Wilkes, E.D. Conte, Y. Kim, M. Holcomb, J.B. Sutherland and D.W. Miller, J.
Chromatogr. A, 880 (2000) 3.

718
 67 J. Pawliszyn (Ed.), Applications of Solid Phase Microextraction. RSC, Cambridge,
1999.
68 A.J. Matich, in J. Pawliszyn (Ed.), Applications of Solid PhaseMicroextraction. RSC,
Cambridge, 1999. p.3 4 9 .
69 D.A. Vereen, J.P. McCall and D.J. Butcher, Microchem. J., 65 (2000) 269.
70 C.A. Zini, F. Augusto, E. Christensen, B.P. Smith, E.B. Caramao and J. Pawliszyn,
Anal. Chem., 73 (2001) 4729.
71 H. S. MacTavish, N.W. Davies and R.C. Menary, Ann. Bot., 82 (2000) 347.
72 D. D. Roberts, P. Pollien and C. Milo, J. Agric. Food Chem., 48 (2000) 2430.
73 F. Augusto, A.L.P. Valente, E.S. Tada and S.R. Rivellino, J. Chromatogr. A, 873
(2000) 117.
74 X.M. Wan, R.J. Stevenson, X.D. Chen and L.D. Melton, FoodRes. Int., 32 (1999) 175.
75 B. Ancos, E. Ibanez, G. Reglero and M.P. Cano, J. Agric. Food Chem., 48 (2000) 873.
76 S.G. Wyllie and J.K. Fellman, J. Agric. Food Chem., 48 (2000) 3493.
77 H.T. Sabarez, W.E. Price and J. Korth, J. Agric. Food Chem., 48 (2000) 1838.
78 J.C. Beaulieu and C.C. Grimm, J. Agric. Food Chem., 49 (2001) 1345.
79 C. Sala, M. Mestres, M.P. Marti, O. Busto and J. Guasch, J. Chromatogr.A, 880
(2000) 93.
80 J. Weber, M. Beeg, C. Bartzsch, K.H. Feller, D.D. Garcia, M. Reichenbacher and K.
Danzer, J. High Resolut. Chromatogr., 22 (1999) 322.
81 E.A. Nonato, F. Carazza, F.C. Silva, C.R. Carvalho and Z.L. Cardeal, J. Agric. Food
Chem., 49 (2001) 3533.
82 C.J. Scarlata and S.E. Ebeler, J. Agric. Food Chem., 47 (1999) 2505.
83 C. P6res, C. Viallon, J.L. Berdagu6, Anal. Chem., 73 (2001) 1030.
84 Z. Zhang and J. Pawliszyn, Anal. Chem., 67 (1995) 34.
85 R.A. Murray, Anal. Chem., 73 (2001) 1646.
86 J. Koziel, M.Y. Jia and J. Pawliszyn, Anal. Chem., 72 (2000) 5178.
87 M. Mestres, O. Busto and J. Guasch, J. Chromatogr. A, 881 (2000) 569.
88 E.F. L6pez and E.F. G6mez, Chromatographia,52 (2000) 798.
89 S. Rocha, P. Coutinho, A. Barros, M.A. Coimbra, I. Delgadillo and A. D. Cardoso, J.
Agric. Food Chem., 48 (2000) 4802.
90 P. Pihlsgard, M. Larsson, A. Leufv6n and H. Lingnert, J. Agric. Food Chem., 48
(2000) 4844.
91 R. Schneider, R. Baumes, C. Bayonove and A. Razungles, J. Agric. Food Chem., 46
(1998) 3230.
92 K. Wada and T. Shibamoto, J. Agric. Food Chem., 45 (1997) 4362.
93 M. Adahchour, R.J.J. Vreuls, A. van der Heijden and U.A.Th. Brinkman, J.
Chromatogr.A, 844 (1999) 295.
94 R. Boulanger and J. Crouzet, FlavourFrag.J., 15 (2000) 251.
95 S.B. Stanfill and D.L. Ashley, J. Agric. Food Chem., 48-(2000) 1298.
96 Q. Lang and C.M. Wai, Talanta, 53 (2001) 771.
97 R.M. Smith, J. Chromatogr.A, 856 (1999) 83.
98 E.E. Stashenko, M.A. Puertas and M.Y. Combariza, J. Chromatogr.A, 752 (1996)
223.
99 M. Leunissen, V.J. Davidson and Y. Kakuda, J. Agric. Food Chem., 44 (1996) 2694.
100 E.M Calvey, K.D White, J.E. Matusik, D. Sha and E. Block, Phytochemistry, 49 (1998)
359.
101 M.M. Jim6nez-Carmona, J.L. Ubera and M.D. Luque de Castro, J. Chromatogr.A,
855 (1999) 625.
102 A. Kubatova, A.J.M. Lagadec, D.J. Miller and S.B. Hawthorne, FlavourFragr.J., 16
(2001) 64.

719
Chapter22

Sample preparation for water analysis

Karsten Levsen

22.1 INTRODUCTION

22.1.1 General considerations

Water is the most frequent chemical compound covering 71% of the earth's
surface and more vital to the survival of humankind than food. Thus, to preserve
what is indispensable to life and to protect human health and the environment,
one of man's most important goals is to provide high quality water in sufficient
quantity at affordable costs, while maintaining the various functional roles of
the ecosystems. To achieve this objective, critical limits for various pollutants
have been established in many countries. The surveillance of these critical limits
and of water pollution in general calls for precise and accurate methods for the
analysis of contaminants in water.
Water analysis deals with very different types of aqueous samples, such as
ground and surface water used directly or indirectly for drinking water produc-
tion, rain water, municipal and industrial waste water and process water. These
various kinds of water may differ not only with respect to the types of pollutants
encountered, but in particular with respect to the pollution level, which has to be
considered during sampling, sample preparation and instrumental analysis.
Sample preparation of ground and surface water with a low pollutant level is
usually less cumbersome than, e.g., that of soil and biological samples (including
most food samples). Thus, in organic analysis sample preparation is often
restricted to the extraction of pollutants from the aqueous sample, while a
further cleanup is only required with highly polluted samples or in ultra-trace
analysis. Except for the basic matrix, the water itself, other matrix components
are less dominant in aqueous samples than in soil and biological samples. Humic
compounds and inorganic salts are the major matrix components which have to
be considered in the analysis of aqueous samples. Humic substances interfere in
particular with the analysis of other organic pollutants if analysis is done by LC
with UV detection, as they lead to a pronounced "hump" at the beginning of the
LC chromatogram. The use of a more specific instrument such as a mass
spectrometer will reduce matrix interferences and thus the need for sample
cleanup. However, even if matrix components such as humic compounds and
Comprehensive Analytical Chemistry XXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
inorganic salts do not directly interfere with the instrumental analysis, they may
deteriorate the performance of a chromatographic column and thus the robust-
ness of an analytical method or influence the instrumental detection (e.g.
leading to an ion suppression in LC/MS with electrospray ionization); however,
these effects are again less pronounced than in soil or biological samples. Thus,
in water analysis the analytical chemist will have to balance the extent of sample
preparation (in particular sample cleanup) against the resulting labor costs.
Water samples do not only consist of the aqueous, but also of a solid phase in
form of suspended particles. Hydrophobic compounds may preferentially be
adsorbed onto these particles. Many standardized methods of water analysis
include a filtration step using, e.g., a 0.45 pim membrane filter, which removes
the larger fraction of these suspended particles. Further sample preparation of
compounds adsorbed on these suspended particles is not considered in this
chapter. Moreover, the discussion is restricted to organic pollutants. (Note that
many inorganic compounds.in aqueous samples can be subjected to an instru-
mental analysis without any further sample preparation).

22.1.2 Target versus non-target analysis, off-site versus on-site

analysis

As with the analysis of other matrices, in water analysis one differentiates

between "target" and "non-target" analysis. In target analysis, one or a limited
number of pre-selected compounds, for which e.g. critical limits are defined by
regulations, are determined. Sample preparation can be optimized for target
analysis which usually provides quantitative results. Long-term monitoring of
environmental waters and, in particular, drinking water is usually restricted to
target analysis, which often uses very specific detection methods (e.g. selected
ion monitoring in GC/MS and LC/MS). Thus, important pollutants may be
overlooked. In routine monitoring, therefore, target analysis should be comple-
mented by less frequent non-target analyses. Such non-target analysis often
only provides qualitative data. Sample preparation in non-target analysis is
more difficult, as the physico-chemical properties of the potential pollutants may
be unknown. In this case, sample extraction should be as universal as possible
and is usually done in several steps to extract non-polar hydrophobic as well as
very polar water-soluble compounds including organic ions.
Today, sample preparation as part of an entire analysis of aqueous samples is
still predominantly carried out "off-site" in central laboratories. This may be the
most cost efficient solution, but obviously it is not the optimum one, in particular
for water analysis where 1 1 glass bottles have to be transported from the field to
the laboratory. Sample degradation (microbial or chemical degradation, loss of
very volatile pollutants) may occur and analytical results are often obtained with
delay. Thus, considerable efforts are currently being made to develop field
analytical methods ("on-site" methods). Even if field analysis is carried out
discontinuously, the above-mentioned disadvantages can be largely overcome.

722
In particular, analytical data are provided rapidly allowing fast decision making,
which is particularly important if an accidental discharge of (toxic) chemicals
into the aquatic environment has occurred.
In this chapter, various sample preparation methods used for water analysis
are also assessed with respect to their potential for field analysis.

22.1.3 Sample extraction and cleanup

Extraction and cleanup methods of organic and, in particular, of environmental

analysis have recently been described in several books [1-5]. As with other
matrices, sample preparation for water analysis comprises (a) extraction of the
analytes and (b) sample cleanup. Sample cleanup is less important for water
analysis and will thus be dealt with briefly in this chapter, while emphasis is laid
on extraction methods. It will be demonstrated that these extraction methods
usually simultaneously provide some selectivity and thus sample cleanup. It is
apparent that a high selectivity (and sample cleanup) can be achieved if a single
target compound or a class of very similar compounds are to be extracted in one
step from an aqueous sample. The simultaneous extraction of compounds or
compound classes with a wide range of polarities is a much more challenging
task. Thus, multi-residue analysis of pesticides usually includes an extraction
step which allows the simultaneous recovery of compounds of very different
physico-chemical properties from aqueous samples.
The ideal extraction method for aqueous samples should provide a high
analyte enrichment, a near quantitative recovery, good accuracy and precision
and low detection limits. Some of these factors are discussed below.
Enrichmentfactor. For many extraction methods described in this chapter,
the enrichment factor is given by the ratio of the initial volume of the aqueous
sample (usually 0.2-1 1) to the final volume of the eluate (assuming a 100% recov-
ery). Note that very high enrichment factors are achieved in specific cases by
using large water sample volumes (Ž1001). If enrichment is followed by GC (MS)
analysis, usually only 1 A l of the eluate is injected into the chromatograph. Thus,
assuming that the final volume of the eluate is 1 ml, only 0.1% of the sample is
analyzed, 99.9% discarded. (As discussed below, a much higher fraction of the
eluate (-10%) is used for the GC analysis if a large volume injector is employed).
The fraction of the eluate used in the instrumental analysis is larger in LC. Nev-
ertheless, sample preparation methods in which the entire "eluate" is trans-
ferred to the chromatographic column should be preferred, as is the case in
on-line solid-phase extraction (SPE)-GC or in SPE-LC, in solid-phase micro-
extraction (SPME) and in some headspace methods employing thermal
desorption (vide infra).
Recovery. A near quantitative analyte recovery is desirable during the ex-
traction of aqueous samples, but not a prerequisite. Thus, the US-EPA accepts
extraction methods with recoveries ranging from 70 to 130%. Even lower recov-
eries are acceptable if isotopically labeled internal standards are used. Moreover,

723
higher recoveries are only achieved with exhaustive extraction methods such as
SPE, liquid-liquid extraction (LLE) and purge-and-trap (dynamic headspace),
while with other sample preparation methods, such as static headspace or
SPME, an equilibration (or pre-equilibration) between the aqueous sample and
the extraction phase is attempted, which leads to an only partial recovery of the
analytes.
Limit of detection (LOD). With respect to sample preparation, only the limit
of detection of the entire analytical method (sample preparation and instrumen-
tal analysis) is of relevance; the instrumental detection limit is not. The LOD
should be sufficiently low to verify critical limits. In the European Union, the
critical limit for pesticides in drinking water was set at 0.1+0.05 ug/l. For a
reliable verification of this limit, a LOD of approx. 0.02 /ug/l is necessary. Even
lower method detection limits are desirable, not primarily to detect even lower
concentrations of analytes, but particularly to use smaller sample volumes, as
realized, for instance, in on-line SPE-GC or on-line-SPE-LC and in SPME.
Further criteria used for assessing sample extraction methods for water
analysis are (a) solvent consumption, (b) speed, (c) commercial availability of
equipment and material, (d) ease of automation, (e) suitability for field analysis.
These criteria are discussed below while describing the various sample prepara-
tion methods of water analysis.

22.1.4 Parameters influencing the extraction

Breakthrough. Extraction methods which are based on the adsorption of the

analyte on a solid (or quasi solid) extraction phase, such as SPE, are limited by
analyte breakthrough, as discussed in more detail in the section on solid-phase
extraction. This breakthrough may have two reasons: first, as SPE is a chro-
matographic process (frontal chromatography), the water in the aqueous sample
will finally displace the analyte from the sorbent; second, a breakthrough is also
observed at high concentrations if all adsorption sites are occupied.
Inorganic ions. In many extraction methods, a high content of inorganic ions
leads to a higher extraction efficiency (salting-out effect) and may thus be used
intentionally to improve the extraction. This effect is discussed in more detail in
the section on SPME.
pH. As a result of the chemical nature of the most common extraction
phases, the partitioning of neutral compounds into these extraction phases is
usually favored over that of ionic organic compounds. Thus, the extraction yield
of ionizable compounds depends significantly on the pH value. Weakly acidic
compounds are transformed into their neutral form at low pH and best extracted
in this pH range, basic compounds at high pH. This approach is used for most
extraction techniques (such as LLE, SPE, SPME, membrane techniques), but
fails if a compound exists in its ionic form over the entire pH range.
Temperature. Depending on the extraction method, the temperature may
influence the extraction yield of a given analyte in various ways: it may influence

724
the kinetics of the extraction (in particular the diffusion of the analytes in the
aqueous phase, through the boundary layer water/extraction phase and within
the extraction phase) and the thermodynamics, i.e. partitioning of the analytes
between the aqueous and the extraction phase, as discussed below.
In this chapter, the methods of extraction and cleanup of aqueous samples
are discussed. Some of these methods are described in separate chapters where
the fundamental aspects are discussed in more detail. To provide a general
overview of sample preparation for water analysis, a certain overlap with previ-
ous chapters is accepted. Emphasis is placed on new methods, while established
methods such as LLE are only briefly discussed. Applications of established
methods such as headspace, LLE and SPE in water analysis are numerous and
will not be summarized here. Instead, applications of new extraction methods
for water analysis are presented in a tabular form. The potential of the various
extraction methods for field analysis is also discussed.

22.2 LIQUID-LIQUID EXTRACTION (LLE)

22.2.1 General considerations

Liquid-liquid extraction is described in several recent books [1-5]. In LLE, the

analytes (solutes) are partitioned between two immiscible solvents in which they
have different solubilities. When applied to aqueous samples, one of the solvents
is water, the other is an organic solvent. The partitioning is described by
Nernst's distribution law, also known as partition law. For a given temperature,
the ratio of the solute activity in each phase is constant [1-3]. At low concentra-
tions, the activities may be equated by the equilibrium concentrations:
Kd = CJCw (22.1)
where C is the solute concentration in the organic solvent, C that in the
aqueous phase and Kd the distribution coefficient. The fraction of analyte
extracted into the organic solvent is
E = C,Vs/(CsV + C, Vw) (22.2)
where Vs or Vw are the volumes of the organic and aqueous phase, respectively, or
E = Kd VVw/(1 +Kd Vs/V,) (22.3)
For n successive multiple extractions it follows
E = 1- [1/(1 + Kd' VsVw)] ' (22.4)
For a one-step extraction and practical phase ratios, V, /Vw, ranging from 0.1 to
10, Kd must be large, i.e. > 10, for a near quantitative recovery (> 99%). Equation
(22.4) demonstrates that to extract a sample it is always better to use several
small portions of solvents (e.g. 3 x 20 ml) for 1 1aqueous sample than one large
portion (e.g. 100 ml) [3].

725
 If weak acids or bases are to be extracted, the pH has a significant influence
on the actual chemical form of the analyte in the aqueous and the organic phase
and thus the extraction. In this case, the distribution ratio D is used to describe
the partitioning, where D is the concentration ratio of the analytes in all
chemical forms in the organic and the aqueous phase, i.e., for a weak acid HA:

D = C(HA),/(C(HA)w + C(A-),) (22.5)

D = Kd/[l + (Ka/C(H+)w)] (22.6)

where Ka is the acidity constant (see Ref. [3]).

22.2.2 Procedures

For LLE the selection of the optimum solvent is of particular importance, where
the general principle "like dissolves like" applies. Strategies for solvent selection
have been reviewed recently [6]. In this review, a classification of a large number
of solvents according to the polarity and the solvatochromic parameters is given.
Further solvent properties to be considered are summarized in Ref. [3], e.g.
distribution coefficient, selectivity, solubility of the solvent in the aqueous
phase, density, viscosity and vapor pressure. As mentioned above, weak acids or
bases are transformed into their neutral form by a proper choice of the pH value,
i.e., low pH values for acids and high pH values for bases. However, for water
analysis it should be kept in mind that at low pH values more interfering humic
substances are coextracted. If the analyte is ionic over the entire pH range, the
method of ion-pair extraction may be used, i.e. a counter ion is added to the
aqueous sample and the newly formed neutral complex extracted into a non-
polar solvent. Addition of inorganic salts may improve the recovery in LLE
(salting-out effect).
LLE is usually performed as discontinuous extraction either manually or in a
shaking apparatus using a separating funnel, as described by Holden [3]. After
repeated extraction, the organic phases are combined, usually dried over non-
hydrous sodium sulfate, if analyzed by GC (MS), and reduced in volume, as
described below.
If Kd is very small or if the sample is very large, continuous LLE should be
favored [2], where fresh solvent is boiled, condensed and left to percolate
repetitively through the aqueous sample. These continuous liquid-liquid extrac-
tors are commercially available and use organic solvents that are either lighter
than or heavier than water. Several extractors can be operated in parallel
unattended, e.g., overnight, and may provide very high enrichment factors of up
to 10 5 [2].
The formation of emulsions often renders phase separation difficult. Among
other methods [2], filtration through a glass wool plug or centrifugation is used
to break up the emulsion.

726
 Liquid-liquid extraction procedures are very simple and straightforward,
and continue to play an important role in water analysis, especially if a small
number of samples are to be analyzed. Thus, many EPA methods are still based
on LLE. However, these methods are time-consuming and difficult to automate
and thus hardly suitable for routine analysis of a large number of water samples.
The consumption of large quantities of solvents and the environmental and
health hazards of these solvents (in particular the frequently used dichloro-
methane) are further severe disadvantages of LLE. LLE is hardly suited for field
analysis of aqueous samples.

22.2.3 Automated liquid-liquid extraction and field analysis

Although LLE is more difficult to automate than the other extraction methods
presented below, there have been several approaches to an automatic LLE.
Here, only the LLE coupled on-line to GC is discussed.
In on-line LLE-GC [7-9], the aqueous sample is periodically injected via a
T-piece (segmentor) into the organic phase. The combined streams flow through
an extraction coil out of glass, fused silica or PTFE in which a segmented flow is
formed and extraction occurs. Phase separation is achieved using a semiperme-
able membrane or a sandwich-type phase separator [10] before the organic
phase is injected via a valve into the GC. As the volume ratio of the organic phase
(Vs) to the aqueous phase (Vw) is near unity, the distribution coefficient Kd
between both phases should be > 10 to achieve near quantitative extraction (see
Eq. (22.4)), a prerequisite not easily met in real-life water samples where
pollutants cover a wide range of polarities. Moreover, for such real-life samples,
the formation of emulsions poses a problem. The range of analytes amenable to
this technique may be extended if the on-line LLE-GC system involves simulta-
neous extraction and derivatization [11].
On-line LLE-GC coupling can also be used for automated monitoring of
pollutants in water samples in the field, as shown for the monitoring of volatile
organic compounds [12].

22.2.4 Applications of on-line liquid-liquid extraction coupled with

GC (LLE-GC)

Applications of LLE in water analysis are so numerous that they will not be
presented here. Rather, the applications of on-line LLE-GC to the analysis of
aqueous samples are summarized. The on-line LLE-GC system described above
was used for monitoring volatile organic pollutants in aqueous samples [7] and
for determining halocarbons in seawater [8], organochlorine pesticides in
ground water [9] and pesticide intermediates in water [13]. The approach of
simultaneous extraction and derivatization using on-line LLE-GC was applied to
the analysis of organic acids after ion-pair extraction [11], phenols [14, 15], and
N-methylcarbamates [16].

727
22.2.5 Solvent evaporation

LLE usually leads to a relatively large volume of an organic extract and thus low
analyte concentration, so that a reduction of the extract volume by evaporation
becomes necessary. Thus, solvent evaporation techniques will be briefly
described here. The same techniques are also applicable to other extraction
methods, such as solid-phase extraction, but they will only be discussed here.
Solvent evaporation is a critical part of sample preparation of aqueous
samples as volatile analytes may be readily lost. Thus, special skill and attention
of the operator are necessary. By adding a keeper, i.e., 0.5-1 ml of a high boiling
solvent in which the analytes readily dissolve, volatile losses may be reduced.
In a rotary evaporator, the solvent is removed under reduced pressure by
mechanical rotation of a flask containing the organic extract (or eluate) in a
water bath at both controlled pressure and controlled temperature.
In a Kuderna-Danish(K-D)evaporator,the solvent is distilled and condensed
in a Snyder column. The returning condensate assists in the process of
recondensing volatile analytes [2].
In the automated evaporative concentration system (EVACS), a distillation
column is also used [2]. Distillation appears to be better controlled, e.g., by
sensing the sample level. The final concentration of 10 to 1 ml is achieved by
nitrogen blow-down.
Nitrogen blow-down is often performed manually, e.g., for solid-phase
extraction eluates. With this technique thermal stress is avoided, but special
attention has to be paid. Thus, automatic devices that allow the blow-down of
large solvent volumes and incorporate a sensing device for the final solvent
volume (e.g., 1 ml) should be preferred.

22.3 HEADSPACE SAMPLING

22.3.1 General considerations

A headspace sample is a gas sample which has been previously in contact with a
liquid or solid sample from which volatile compounds were released into the gas
phase. The headspace sample is subsequently analyzed by gas chromatography
(HS-GC). In this chapter, only the analysis of water samples by HS-GC is
discussed. The water sample is placed in a closed container in contact with the
gas phase (the extracting gas). The gas extraction can be carried out as a
one-step extraction (static or equilibrium headspace), similar to solvent extrac-
tion, or as a continuous extraction by stripping off the volatile compounds in a
continuous flow of an inert gas (dynamic headspace). As with most sample
preparation methods, headspace sampling is used for the extraction of the
analytes from the water sample, and simultaneous sample cleanup as interfer-
ences from non-volatile matrix components (humic substances, inorganic salts)
during the subsequent analysis are minimized. If headspace sampling replaces

728
LLE, the numerous problems with the use of solvents in the subsequent chro-
matographic separation and analyte detection are avoided.
Static headspace sampling is usually a one-step procedure, where an aliquot
of the vapor phase is directly transferred to the gas chromatograph. Dynamic
headspace is usually a two-step procedure, where volatile analytes stripped off
the aqueous phase by an inert purge gas are separated from the matrix or
headspace gas in a trap. From there, they are released into a stream of the GC
carrier gas by thermal desorption and transferred to the column. Thus, dynamic
headspace sampling leads to an enrichment of the analytes and is the method of
choice for low analyte concentrations. Headspace solid-phase microextraction
(SPME) represents an alternative to the classical dynamic headspace and will be
discussed in a separate section.
While headspace sampling is straightforward and appears to be simple,
problems may arise during the subsequent transfer to the GC and the chromato-
graphic separation. Headspace sampling is also described in Chapter 10 and has
recently been reported in another book [17] and a review [18], where the
problems arising from the coupling of headspace sampling and subsequent GC
analysis are discussed in detail. For routine analysis in water laboratories, a
deciding criterion is the automation of HS-GC.

22.3.2 Static (equilibrium) headspace

In static headspace, the liquid phase of the water sample (in a vial closed tightly
by a septum) is in equilibrium with the gas phase. The partial pressure of the
analyte in the gas phase is described by Henry's law which for ideal dilute
solutions is given by
Pi = Hxi (22.7)
wherepi is the partial pressure of the component i in the vapor phase, xi its mole
fraction in the aqueous phase (i.e., if n represents the number of moles present,
x i = ni/n,,t,) and H is Henry's constant, which is the basis for a quantification by
HS-GC. Henry's constant and thus the analyte response vary significantly from
analyte to analyte. In addition, Henry's constant depends on the temperature
and increases significantly with temperature. Thus, for reproducible headspace
analyses the water sample has to be thermostatted where enhanced tempera-
tures (e.g., 80°C) significantly increase the partial pressure of the analytes and
thus the response.
In water analysis, the partial pressure of the water in the headspace gas
dominates by far over the partial pressures of the analytes, which, as discussed
below, may cause problems in the subsequent GC analysis.
In static headspace, sampling of the gas phase can be done using a gas-tight
syringe or a sample loop as used in gas analysis by GC. The latter technique is
easier to automate. Filling the loop with headspace is achieved by first pressuriz-
ing the vial with an inert gas (e.g. the GC carrier gas). Alternatively, instead of

729
 carrier gas
Fig. 22.1. Schematic set-up of a static
headspace gas chromatograph with bal-
ance pressure headspace system, a cryo-
genic refocusing unit (cryogenic trap)
and a water trap. The system is shown in
the standby mode with backflushing of
the water trap for regeneration. V1-V4 =
solenoid valves. The whole sampling
cycle comprises the additional steps:
pressurizing (needle down, V1, V2, V3
open, V4 open to V1), sampling (needle
down, V1,V2 closed, V3 open, V4 closed
to cryogenic trap) [17] (courtesy of
Elsevier, Amsterdam, Netherlands).

first filling the loop, the pressurized gas can be expanded directly into the GC col-
umn. This method is called "balanced pressure sampling" [191. A commercially
available instrumental set-up which is completely automated is shown in Fig.
22.1. In a first step, the carrier gas pressurizes the sample vial by a proper actua-
tion of the four valves (usually up to the column head-pressure). In a second step,
sample transfer is performed for a short time, while disconnecting the carrier
gas flow. The pressurized headspace in the vial expands directly into the column
and no headspace gas is wasted due to an unnecessary expansion to atmosphere.

22.3.3 Dynamic headspace

The dynamic headspace technique separates the volatile sample constituents

from the sample by a continuous flow of an inert gas either through or above the
water sample. The method of bubbling the extraction gas through the water
sample is also termed "purge and trap". For quantitative analysis, an exhaustive
and time-consuming gas extraction is required. Thus, a direct introduction of
the headspace into the GC column is not possible. Rather, the analytes are
focused in a trap, in general a cartridge packed with adsorbent (e.g., Tenax),
from which they are released by thermal desorption and transferred via the
carrier gas into the column.
As discussed in detail by Kolb [17], the analyte trapping on a small sorbent
with subsequent thermal desorption in dynamic headspace leads (if coupled with
capillary GC) to (a) a flow problem (the high flow rate during desorption is not
acceptable for instantaneous sample introduction into a capillary GC) and (b) a
time problem (the desorption from the trap may take longer than acceptable for
high resolution in capillary GC leading to band broadening [20,211). The flow
problem can be resolved by using a splitter at the GC inlet [22,23], which,
however, reduces the sensitivity of the method. The flow and time problems are
better solved by additionally refocusing the analyte in the GC column using a
cryogenic trap [24,25]. This cryogenic trapping can be either achieved by cryo-
genic condensation, i.e., by condensation of the analytes in traps which do not

730
contain a stationary phase (see Ref. [17]), or better by cryogenic focusing, where
the analytes are trapped in the liquid phase of a GC column at low temperatures.
Cryogenic focusing is readily achieved using the technique of whole-column
cryotrapping (WCC) [26-29], in which the entire GC including the column is
cooled (e.g., to -100°C). Most gas chromatographs are equipped with a so-called
subambient accessory, where, e.g., liquid nitrogen is introduced into the oven
through a valve and cryofocusing can be performed automatically [30,31].
However, in order to generate a narrow starting band of the analytes, it is
sufficient to cool only the beginning of the coated capillary column. A very
effective band concentration is achieved if cryogenic focusing is combined with
an additional focusing by a negative temperature gradient (i.e. the front of the
moving zone is cooler than its rear) [32-34]. This can be achieved with the
instrumental set-up shown in Fig. 22.1. During the sample introduction, cold
nitrogen gas (prepared outside the oven by passing nitrogen through a liquid
nitrogen bath) flows through a PTFE tube outside the fused silica pre-column,
but counter to the warm headspace gas. This set-up is commercially available
and leads to a double focusing of the analyte and to a very narrow band of the
analytes leaving the trap.
Both static and dynamic headspace sampling have to solve the "water
problem", in particular in water analysis. The water in the headspace vapor may
block the GC capillary and cause a peak distortion. The problem is more severe
in dynamic headspace (e.g., with purge and trap) than in static headspace. Water
in dynamic headspace can be removed with a condenser cooled to +5 to -10°C
and placed in front of the Tenax trap [35,36]. Alternatively, the water can be
removed using a Nafion dryer [37-42]. In static headspace, the water is mostly
removed using a water trap with a sorbent coated with LiCl [43]. The water trap
is regenerated in the backflush mode at the end of the GC run by increasing the
oven temperature above + 120°C. Such a water trap is also shown in Fig. 22.1.
Static cryogenic headspace has the main advantage of running fully auto-
mated under computer control. The commercially available set-up shown in Fig.
22.1 can automatically handle 40 vials each containing 2.0 ml of an aqueous sam-
ple. Such automated systems have a good long-term reproducibility. With the
analysis of BTX in water, for instance, a precision of <3% is achieved [17].
Headspace sampling (both static and dynamic, i.e. purge and trap) have been
established methods of water analysis for many years and are well suited to the
routine analysis of volatile compounds in aqueous samples (see e.g., EPA method
624 [76]). Thus, applications of this method are not reported here.

22.4 SOLID-PHASE EXTRACTION (SPE)

22.4.1 General considerations

A historical overview of the development of SPE has recently been presented by

Liska [47]. In conventional solid-liquid extraction (SLE) or solid-phase extrac-

731
tion (SPE), a liquid is passed over a sorbent packed in a cartridge or embedded in
a disk. As a result of the strong attraction, the analytes are retained on the
sorbent. In a second step, the analytes are eluted by passing a small volume of
organic solvent through the cartridge or disk which disrupts the bonds between
analytes and sorbents.
SLE was introduced in the mid 1970s [44]. In these early applications,
enrichment was achieved by adsorption on polystyrene-divinylbenzene
(PS-DVB) polymeric resins (e.g., XAD-2) packed in glass cartridges followed by
elution with a small quantity of suitable organic solvents. The term "solid-phase
extraction (SPE)" was introduced for this sample preparation technique and is
still predominantly used today. Although this term is neither suited to a system-
atic classification of the various sample enrichment methods nor does it describe
the physical process underlying this sample preparation, it will also be used here.
In these early applications, good recoveries and an acceptable precision were
achieved for several analytes; however, the procedure was still time-consuming
as the cartridge had to be packed by the analyst and the sorbents had to be
precleaned thoroughly.
The breakthrough of the SPE method was achieved at the advent of commer-
cially available, prepacked, disposable cartridges with chemically modified
reversed phase silica (such as C8 and C,,). However, in these early days the
recoveries using these modified silica often varied from batch to batch. Today,
different formats such as cartridges or disks and a large variety of sorbents with
different chemical matrices and reproducible recoveries from many suppliers are
on the market (vide infra). They permit an enrichment of organics not only from
environmental but, in particular, from biological samples. In water analysis,
SPE is probably the most widely used sample preparation technique which is
applicable not only to non-polar (hydrophobic) analytes but, more recently, also
to polar (more hydrophilic) compounds. The method is described in several
books [4,5,45-46] and reviews [47-51]. The reader is particularly referred to
Chapter 12 and a recent review by Hennion [48]. The major advantages of
solid-phase extraction are (a) decreased use of (toxic) organic solvents, (b) ease of
automation (allowing a high sample throughput with minimum labor involve-
ment) and (c)its suitability for field analysis (which has not been fully explored).
Method development may be more tedious and the reproducibility poorer with
SPE when compared to LLE.
Solid-phase extraction as a sample preparation step in water analysis is
mainly used for the enrichment of analytes. As with other sample preparation
techniques, a partial sample cleanup is achieved simultaneously so that in
general (at least for the analysis of drinking, ground and surface water), no
additional cleanup is required, in particular if sample preparation is followed by
a selective and specific instrumental analysis as GC/MS and LC/MS. Sample
cleanup in water analysis by SPE is discussed in Section 22.7, while specific
sorbents with enhanced selectivity are described in Section 22.4.6.
The term solid-phase microextraction (SPME), as introduced by Pawliszyn

732
in 1990 [52], suggests that this new method is a miniaturized solid-phase
extraction. Although with some fibre coatings analytes are extracted from the
aqueous phase by adsorption as with SPE, the extraction principle differs
(non-exhaustive extraction with batch equilibrium or preequilibrium with
SPME when compared to exhaustive extraction with flow-through equilibrium
or preequilibrium with SPE) and will thus be presented in Section 22.5.
As with most analytical techniques, the commercial availability of the equip-
ment or material and the incorporation into official standardised methods are
necessary prerequisites for a wide-spread use in routine analysis. Thus, for the
promotion of SPE it has been of advantage that many supplier (more than 50
companies [48,53]) can make products for SPE because of lacking severe patent
restrictions. Moreover, in Germany as early as 1994 several official standard
methods of water analysis (DIN methods) were exclusively based on SPE. More
recently, the technique has also been included in the European CEN methods
and the US EPA methods.

22.4.2 SPE formats and procedures

Formats [48]: Cartridges are still the most popular format. They are typically
filled with 40-60 Aum dp packing material and in general made out of polypropy-
lene, polyethylene or glass. (The larger particle size is, e.g., necessary to achieve
sufficiently high flow rates.) Originally, sample contamination by extraction of
polymer additives or impurities represented a problem, in particular for water
analysis, but the use of high purity medical grade polymers for the production of
the cartridge bodies and the frits reduced such contamination. For water analy-
sis, additional reservoirs can be adapted to the cartridge to increase the sample
volume. Vacuum manifolds which allow a parallel processing of several (e.g., 12)
water samples are supplied by several manufacturers. They may include devices
for the drying of the cartridge with vacuum or nitrogen and for the volume
reduction of the eluate by nitrogen blowdown. If suspended particles are not
removed from water samples by filtration, plugging of the cartridge or exces-
sively long percolation times may represent a problem. Thus, nowadays filters
are often integrated into the SPE cartridge. For on-line SPE-GC/MS and
SPE-LC/MS special cartridges are manufactured (vide infra).
With the second most popular format, the disk, much higher flow rates can
be realized due to its large cross-sectional area and thin adsorber bed. The
sorbent (on a polymer or silica substrate) is embedded in a web of PTFE or glass
fibre. Glass fibres are thicker and more rigid and therefore provide higher flow
rates. The sorbent particles are smaller than those used in cartridges (approx. 8
Am instead of 40 Am dp). As mentioned above, the shorter sampling rate and the
smaller particle size allow a more efficient trapping of analytes at higher flow
rates when compared to cartridges. Moreover, disks are more suited to the
extraction of large water sample volumes containing suspended particles. They
are available in several different diameters, the most frequently used size being

733
 47 mm. Small (4.6 mm) disks were used for on-line coupling of SPE with LC [54].
While disks are particularly suited to water analysis, it should be mentioned that
other formats have been introduced for a (high throughput) analysis of biological
samples such as the 96-well-plate format [55] or the SPE pipette tip.
Procedures for water analysis: When conventional SPE is used for water
analysis, samples of up to 1 1 are typically handled (in contrast to biological
samples where the sample volume is limited to a few ml). In water analysis, the
SPE procedure (e.g., with C1, sorbents) consists of five steps: (1) conditioning of
the sorbent with one or several organic solvents to reduce background
contaminants and to open the hydrophobic chains to increase the effective
surface area; (2) application of the water sample; (3) washing of the adsorbed
sample by passing a small volume of pure water through the cartridge or disk; (4)
drying of the sorbent bed by passing nitrogen through the bed or by applying a
vacuum; and (5) desorption and recovery of the sorbent with suitable solvents.
Many official standardised SPE methods of water analysis include pre-
filtration through a 0.45 /m PTFE filter. This avoids a plugging of the sorbent
(vide supra) and ensures that only the water-soluble fraction of the analyte is
enriched and analyzed. In particular, hydrophobic analytes may be adsorbed on
suspended particles. Although such suspended particles may be trapped in the
adsorber beds, it is not the primary aim of SPE to extract and enrich the analytes
in both the aqueous and the particulate phase (suspended particles).
The procedure described above is simple and straightforward. Nevertheless,
depending e.g., on the analyte, a thorough method development and validation
are necessary. It has been stated [48] that most SPE methods published to date
have largely been developed empirically in a labor-intensive and tirne-consum-
ing trial and error process with little consideration of the chemistry and physics
involved.
Various approaches to SPE method development that have been described
are based on the analogy between SPE and LC. The following parameters are to
be determined during the method development, as indicated in the SPE
sequence outlined above: (a) selection of the type and amount of sorbent; (b)
determination of the sample volume which can be applied without loss in
recovery (the so-called safe sample volume); (c) composition and volume of the
washing solution which can be applied without loss of analyte and (d) composi-
tion and volume of the elution solution. Some knowledge of the breakthrough
volume is of particular importance.

22.4.3 Breakthrough volume

The theory of solid-phase extraction and analyte breakthrough is only summa-

rised, because it is discussed in detail in Chapter 12 and has also been reviewed
recently [65].
SPE uses the principles of LC. The processes involved are frontal chromatog-
raphy during the extraction step and displacement chromatography during the

734
desorption (elution) step [48]. Thus, in a first approximation, SPE can be
described as a simple chromatographic process where the sorbent is the station-
ary phase and (in water analysis) water the mobile phase during extraction and
the appropriate solvent during the desorption (elution) step.
For sample enrichment by SPE in water analysis, reversed phase sorbents
(and sorbents with comparable properties) are frequently selected.
When an aqueous sample is spiked with a solute and percolated through a
SPE cartridge, a breakthrough curve can be observed (e.g., by measuring the UV
absorption), as shown in Fig. 22.2 [48]. The volumes Vb and Vm are defined at 1%
of the initial absorbance A, and at 99% of the initial absorbance A, respectively.
Under ideal conditions, this curve has a bilogarithmic shape, where the inflec-
tion point corresponds to the retention volume of the analyte Vr. The break-
through volume represents the maximum sample volume which can be applied
to a SPE cartridge or disk with a 100% recovery. Thus, knowledge of the
breakthrough volume for a given analyte and sorbent is important for the
development of a SPE method.
The breakthrough volume can be determined by using a spiked water sample
and by monitoring the UV signal continuously and discretely at the outlet of a
pre-column or cartridge [5,56-61] (provided the spike level is lower than the
sorbent capacity). However, such measurements are time-consuming and not
very accurate, in particular for analytes with strong retention (leading to a sig-
nificant spread of the breakthrough curve). Hence, various methods have been
developed to estimate or predict the values of breakthrough volumes, as
described in detail in Chapter 12 and by Hennion [48]. Thus, as a result of the
analogy between SPE and LC, breakthrough volumes may be predicted using
retention factor values [5,48,58]. Vb (at the 1% level, Fig. 22.2) is related to the
retention volume by

Vb = VR- 2.3 (Vm N½-(1 + ks)) (22.8)

where ks is the retention factor of the solute and N the number of theoretical
plates [48]. Vm can be calculated using the porosity of the sorbent and the
geometric volume of the pre-column, cartridge or disk. For C1s silica in
cartridges, Vm is estimated as 0.12 ± 0.01 ml/100 mg sorbent [48]. N is difficult to
determine for cartridges or disks and has to be estimated. For a SPE cartridge
packed with 500 mg of C18 silica N has been determined as approx. 20 for flow
precum- Detector
Water spiked
UV response with re

Fig. 22.2. Typical breakthrough curve

obtained by recording the UV signal of
the effluent of the cartridge (see text for
symbols). The shaded area represents the
maximum amount than can be precon-
centrated [48] (courtesy of Elsevier,
Amsterdam, Netherlands). 2

735
rates of 5 ml/min [62]. For more detailed information, the reader is referred to
Chapter 12 and to Refs. [48,61-65,92].
Moreover, for SPE cartridges with C,8 silica, k, can be estimated based on
octanol/water partitioning coefficients [66-68]. Since the retention mechanism
is primarily governed by hydrophobic interactions between analyte and the alkyl
chains bond to silica, a relation has been observed between the retention factors
of the analytes and their octanol-water partitioning coefficient (Kw). In particu-
lar, a linear relationship was found between log k, using methanol/water as the
mobile phase and log K,,. (Such a linear relationship was even found for
polystyrene-divinylbenzene polymers, but not for porous graphitic carbon [68],
vide infra). However, it has been shown for very polar analytes (log Ko < 1.5)
that these log Ko values are of limited help for predicting SPE breakthrough
volumes [48]. Since the breakthrough volume is small with such a compound, it
is more reliable to determine k s with measurements of a short C1, column using
methanol/water as the mobile phase in different ratios and by an extrapolation
to zero methanol content.
It should be kept in mind that a 100% recovery of the analyte (as assumed in
the discussion above) is not a necessary prerequisite for water analysis. (The US
EPA accepts methods with recoveries ranging from 70-130%). Thus, as dis-
cussed by Hennion [48], recovery curves (i.e., the recovery as a function of
sample volume) are of more relevance to SPE method development than break-
through curves. Such recovery curves demonstrate that sample volumes higher
than the breakthrough volume can be used in SPE of compounds with high log ks
with only small losses in recovery.

22.4.4 Binding energies

Apart from an efficient retention of the analyte on the sorbent (as discussed
above), a quantitative desorption (elution) of the analyte from the sorbent is the
second important step in SPE method development. To optimise the step, a good
knowledge of the interactions between analyte and sorbent is required. These
interactions are the same as in LC and include mainly (a) hydrophobic interac-
tions (dipole-dipole, dipole-induced dipole and dispersive interactions with
binding energies ranging from 1-10 kcal/mol (approx. 10-40 kJ/mol)), (b)
hydrogen bonding with binding energies ranging from 5-10 kcal/mol and (c)
ionic interactions with binding energies ranging from 50-200 kcal/mol [48,69].

22.4.5 Conventional sorbents

In the past, a large variety of sorbents have been developed and marketed which
allow the extraction of both non-polar and/or polar organic compounds from
water. In addition, sorbents with enhanced selectivity have been developed
which allow sample extraction and cleanup to be conducted in one step. These
sorbents are based on size-exclusion (restricted access materials, RAMs) or

736
molecular recognition (immunosorbents, ISs; molecular imprinted polymers,
MIPs). They are described in Section 22.4.6.
Conventional sorbents include chemically bonded reversed phase silica,
normal phase sorbents, styrene-divinylbenzene polymers, carbon-based sorb-
ents, ion pair and ion-exchange sorbents and mixed mode sorbents. The various
types of sorbents are described in detail in Chapter 12 and Ref. [48].
Chemically bonded reversedphase silica. Initially, chemically bonded revers-
ed phase silica (C8, Cl 8) were mainly used for SPE of water samples. They were
made out of the same stationary phase as those in LC columns, but with a larger
particle diameter (vide supra). These chemically modified silica are well suited to
the extraction of non-polar to moderately polar analytes such as many classes of
pesticides including organophosphorous, triazines, phenylurea and carbamate
pesticides. Poor recoveries were observed with more polar compounds (log Kow <
1.5), in particular polar pesticide metabolites. Phenol (log Kow = 1.5) and
deisopropylatrazine (DIA, log Kw = 1.1) are often considered as representatives
of polar water pollutants and are used by analysts and manufacturers to charac-
terize SPE sorbents [48]. Using conventional SPE cartridges with 500 mg Cl8
packing and a 1 1 water sample, the recovery for phenol is low [70]. It can be
increased by using higher amounts of sorbents or lower sample volumes and by
adding salt (salting-out effect).
The trend in LC is to produce totally non-polar chemically modified silica by
reducing the number of residual silanol groups of the original silica. To achieve
this, trifunctional silane is used for bonding the n-alkyl chains and end-capping
is carried out with trimethylsilane after bonding (although some residual silanol
groups will always remain) [71-73]. In contrast, it was realized that the presence
of residual silanol groups is of advantage for SPE to enhance the interaction with
more polar analytes. Thus, basic analytes are retained by an ion exchange mech-
anism. A second requirement for a good analyte retention is a high coverage of
the surface by alkyl chains, which is often expressed as carbon content percent-
age [48]. These requirements (high percentage of alkyl chains and simultaneous
presence of residual silanol groups) led to the design and development of specific
C,8 silica for SPE. A high carbon content percentage is achieved using silica with
a high specific surface area (500-600 m2/g) as a starting material, while the sur-
face modification by monofunctional silane and no end-capping was used to
enhance the number of residual OH groups and to also provide, in addition to
hydrophobic interactions, hydrogen bonding and ionic interactions. This allows
more polar compounds to be extracted even if large sample volumes are used.
Both properties (high surface coverage by alkyl chains and a high number of
residual silanol groups) are difficult to realize simultaneously. An increase in the
number of residual OH-groups increases the hydrogen bonding interaction,
while the use of monofunctional silanes for surface modification reduces the
hydrophobic interactions. The choice of the optimum sorbent depends on the
individual analyte (in particular on the range of analytes) to be trapped and the
sample volume necessary to reach the desired detection limit.

737
 Manufacturers often provide information on the silane function (mono-
functional, trifunctional, end-capping and percent carbon content). A list of
commercially available Cl 8 sorbents and their characteristic properties is avail-
able in Ref. [48].
There is another difference between LC and SPE with C18 silica in water
analysis: the mobile phase in LC contains an organic solvent which, when
adsorbed to the stationary phase, provides a good contact between the solute and
the hydrophobic sorbent surface, while this organic solvent is absent in the
aqueous sample to be enriched. Therefore, the addition of a small amount of
organic solvent (0.5%) to the water sample is recommended.
In addition to C, and C18 silica, other modified silica are available on the
market, namely C2 silica with carbon content in the range 3-6% (most of them
end-capped) and cyclohexyl and phenyl silica. With the latter an increase in
retention of aromatic analytes was reported [105]. Cyanopropyl and amino-
propyl silica represent polar phases which exhibit both polar and non-polar
interactions. Hydrophobic interactions are achieved by using a high carbon
loading of typically 8-9% for cyanopropyl silica.
Finally, normal phase SPE sorbents, e.g., bare silica, alumina, Florisil (a
synthetic magnesium silicate), and silica modified by amino, cyano and diol
groups are available. For these sorbents, SPE is described as adsorption chroma-
tography where only organic solvents are used as mobile phase. Therefore, in
water analysis they are not used for sample enrichment, but rather for sample
cleanup, as described in Section 22.7.3.
Poly (styrene-divinyl benzene) copolymers. The principles and new develop-
ments in SPE based on polymer sorbents have recently been reviewed [77]. As
mentioned above, polystyrene-divinyl benzene (PS-DVB) copolymer resins as
Amberlite XAD resins were among the first sorbents used for SPE [44] and have
been used in many analytical laboratories in spite of the tedious cleaning and
packing of the cartridge. Later, disposable pre-columns designed for on-line
extraction became commercially available. It was only with the advent of PS-
DVB resins with high specific surface areas (700-1200 m2/g), prepacked and
precleaned in disposable cartridges, that the use of these SPE sorbents became
wide-spread in routine water analysis. They are available in cartridge or disk
format [75]1. The higher degree of cross-linking in these polymeric resins leads to
a higher porosity material. The resulting increased specific surface area allows
enhanced F-e interactions and leads to a significant increase in retention of the
polar compounds. For such polar compounds, the retention power is increased
by a factor of 20-60 when compared to the original PS-DVB resins with lower
specific surface area and by a factor of approx. 100 when compared to C18 silica
[5,74,77-89]. For deisopropylatrazine, DIA, in aqueous samples, for instance, a
retention factor logk, = 2.3 + 0.1 was observed for C, 8 material. Log h, increased
to 3.2 + 0.2 for PS-DVB resins with specific surface areas of 350 m2/g and even to
4.4 + 0.3 for a similar resin with a specific surface area of 1060 m2 /g [90]. (Log k,
for porous graphitized carbon, discussed below, is >3.5 for this compound.)

738
 In other words, for analytes with a polarity similar to that of DIA and for
large sample volumes, high surface PS-DVB copolymers appear to be the
sorbents of choice. Another advantage over C18 silica is their stability over a pH
range of 1-14, while silica-based sorbents are only stable over a pH range of 2-8.
Unfortunately, there are only few chromatographic data available for these
resins which are not yet routinely used as LC packings [77,91,93,94].
More recently, chemically modified resins have been developed [95-101] and
have become commercially available. The PS-DVB polymer was, e.g., modified
by introduction of acetyl groups or by sulfonation. Even more attractive are
polystyrene-divinylbenzene-N-vinylpyrrolidone (PS-DVB-NVP) copolymers
which have simultaneously hydrophilic and lipophilic properties and are advert-
ised as "universal" sorbents. They are able to enrich acidic, basic and neutral
compounds varying substantially in polarity. For these new sorbents, a
significant increase in retention was reported for several very polar analytes
(such as catechol), which is explained by both the high specific surface area (800
m2/g) and the presence of a pyrrolidone group which acts as hydrogen acceptor
[101]. These hydrophilic/lipophilic sorbents appear to be particularly attractive
for the extraction of drugs and their metabolites from various body fluids. For
water analysis they have been used for the extraction of polar pesticides [291]
and multiresidue analysis [292]. A list of commercially available SPE sorbents
based on PS-DVB or PS-DVB-NVP resins is given in Ref. [48].
Carbon-basedsorbents [102]. The most popular carbon-based sorbent used in
SPE is graphitized carbon black (GCB), a non-porous sorbent with a relatively
low specific surface area (100-200 m2 /g). It is produced by heating carbon black
to high temperatures (2700-3000°C). With this material, chemisorption of oxy-
gen leads to various functional groups which are important for the interaction
with the analyte. GCB sorbents are commercially available on SPE cartridges
and disks [103] and are also produced as porous material (porous graphitic
carbon). The retention mechanism differs from those of C18 silica and PS-DVB
copolymers [104]: compounds are retained by both hydrophobic and electronic
interactions (ion exchange mechanisms). Thus, non-polar, hydrophobic and very
polar analytes in water are well retained which makes the sorbent particularly
attractive for a multi-residue analysis of, e.g., pesticides (including the polar
metabolites), as extensively shown by Di Corcia and others [103-113]. A particu-
larly high retention is achieved for planar molecules containing several
functional groups and for systems which are rich in polarizable electrons [137].
Ion exchange and ion pair sorbents. Ionic or ionizable analytes can be
extracted and enriched from water samples using ion-exchange sorbents which
are mainly produced by chemical modification of PS-DVB copolymers. Weak
cation exchangers normally have carboxylic groups and strong cation
exchangers sulfonic acid groups, while weak anion exchangers have primary or
secondary amino groups and strong anion exchangers quaternary amino groups.
SPE of ionizable analytes is straightforward: retention occurs at a sample pH
where the analyte is present in its ionic form and desorption at a pH where the

739
analyte is present in its neutral form. If the analyte is ionic over the entire pH
range, desorption is achieved with a mobile phase of higher ionic strength. With
water samples (environmental samples), the main problem is caused by the
inorganic ions present in high amounts, as they readily adsorb to the ion
exchange sorbent and cause an overload of the sorbent. This problem can be
partly overcome by a pretreatment of the water sample to remove the major
inorganic ions, e.g., by precipitation or complexation [114]. When the organic
ions are more hydrophobic, an additional hydrophobic interaction occurs with
the matrix of the ion exchanger leading to a better retention.
Another important approach to the extraction of ionic organic analytes from
water represents the method of ion-pair extraction. With organic anions such as
aromatic sulfonic acids, for instance, ion pairs are formed with quaternary
ammonium salts (tetramethyl-, tetrabutyl- or cetyltrimethyl ammonium) which
are then enriched on a conventional C or Cs sorbent. Typical applications are
aromatic sulfonates such as naphthalene sulfonates and linear alkylbenzene-
sulfonates (LAS) where the latter are used as anionic surfactants [115-119].
Mixed mode sorbents. Mixed mode sorbents contain reversed phase alkyl
chains and cation exchangers bonded to the same matrix (e.g., octyl chain and
cation exchange groups). They are particularly used for a simultaneous extrac-
tion and cleanup of drugs and their metabolites in body fluids [120]. Drugs
containing nitrogen are transformed into their protonated forms at low pH and
bond to ion exchange sorbents, while all other analytes and interferences that
first bound to the reversed phase mode of the sorbents by hydrophilic interac-
tions are eluted with methanol. Finally, the analytes are desorbed with a basic
solution to break the ionic interactions. Thus, these mixed mode sorbents allow a
simultaneous extraction and cleanup. Apart from drug analysis, they have been
applied to the analysis of aqueous samples, e.g., the extraction of triazines from
water [120].

22.4.6 Sorbents with enhanced selectivity

Restricted access materials (RAMs). Restricted access sorbents combine size

exclusion of high molecular mass matrix components with simultaneous enrich-
ment of low molecular mass analytes at the inner pore surface [121-129].
Usually, the inner surface has a reversed phase coating (C8 , C,,), but ion
exchange groups may also be incorporated. These sorbents are particularly
suited to the analysis of drugs and their metabolites in plasma and serum
samples where the proteins are excluded from access into the small pores due to
their size. In addition, diol groups may be bound to the outer surface of the
sorbent particles to avoid precipitation of the proteins and thus plugging of the
pores. To date, the production of disposable RAM cartridges remains too expens-
ive. Instead, these cartridges are coupled on-line with LC (MS) as discussed
below. They can be regenerated and used for up to 1000 times (see Section
22.4.7). RAM sorbents have been used for water analysis to remove humic

740
substances prior to the subsequent chromatographic analysis, e.g., for the
determination of acidic herbicides [130] or triazines [131].
Immunoaffinity sorbents (ISs). In trace and ultra-trace analysis of organic
pollutants in water, an unequivocal identification and quantification of target
analytes may be difficult because of many interferences present at higher
concentrations which may coelute with the target analyte. This is particularly
the case if the chromatographic unit (GC or LC) is not coupled with a specific
detector such as a MS. In water analysis of polar compounds by LC with UV
detection, there is the additional problem of coelution with humic substances
which are observed as a large unresolved hump in the first part of the chromato-
gram. Such interferences can be avoided if highly selective SPE sorbents, such as
immunoaffinity sorbents (ISs) or molecular imprinted polymers (MIPs), are
used for sample enrichment. These techniques are based on molecular recogni-
tion. Immunoaffinity sorbents are described in detail in Chapter 33. Here, only
the basic principles and the application of these techniques to water analysis are
briefly repeated.
In the case of immunoaffinity sorbents, the molecular recognition is
achieved by antigen-antibody interaction. The antibodies are covalently bound
to an appropriate sorbent to form the immunosorbent, which is packed into a
SPE cartridge or pre-column. The binding of the analyte is the result of good
spatial complementation and specific intermolecular interactions between
analyte and antibody. Binding can also occur to analytes with structures similar
to those of the target molecule, leading to a so-called cross reactivity of antibod-
ies. This cross reactivity, which is a negative property of immunoassays, can be
used to extract structurally related analytes.
Immunosorbents have mainly been applied to the analysis of biological
samples, as discussed in Chapter 33. In this field, commercial ISs have been
introduced in the last decade. In environmental analysis, they have been pre-
dominantly used for a selective extraction of target pesticides and their major
metabolites [132-136] and steroid estrogens [137]. In addition, they have been
developed for the extraction of a whole compound class by exploiting the above-
mentioned cross reactivity or by mixing two antibodies in the same cartridge.
Therefore, ISs were again developed for the extraction of classes of pesticides
[138,139] as well as for BTEX, polycyclic aromatic hydrocarbons (PAH) and
azodyes [140-150].
Molecularly imprinted polymers (MIPs). The preparation of antibodies is
still cumbersome and involves animals. Therefore, it has been attempted to
synthesise polymers with molecular recognition sites analogous to those in
antibodies. In recent years, this has led to the development of molecularly
imprinted polymers (MIPs). The synthesis is achieved using suitable monomers
assembled around the target molecule (the "template molecule") and a cross-
linker for subsequent polymerisation. In a second step, template molecules are
washed from the porous polymer with a suitable solvent resulting in polymers
with cavities which are "imprints" of the target molecule. These cavities act as

741
recognition sites if the MIP is used for the selective extraction of target
molecules from the sample. The selectivity is based on both the shape of the
cavity and the specific interactions between the target molecule and the polymer
(usually hydrophobic, hydrogen or electronic interactions). When compared to
antibodies, such MIPs can be prepared more rapidly and easily. In addition, they
are stable at high temperatures and over a large pH range. The use of MIPs as
SPE sorbent was first described by Sellergren in 1994 [151]. This technique is
also termed "molecular imprinted SPE" (MISPE) and has so far mainly been
applied to the selective extraction of drugs from body fluids and other biological
samples [152,153]. Synthesis of MIPs and their applications have been reviewed
repeatedly [154-158].
A specific problem of the use of MIPs as selective SPE sorbents in trace
analysis results from residual template molecules which were not completely
removed during the washing step and which may elute together with the target
molecule during the subsequent sample extraction/elution. Such interferences
are avoided if closely structural analogues are used as template molecules during
MIP synthesis.
MIPs are usually prepared in an apolar (non-aqueous) solvent. There are
some difficulties using MIPs for water analysis since the hydrogen bonding and
other interactions are weakened in a highly polar aqueous environment which
can decrease the interaction between the analyte and the MIP and thus the
selectivity. This is one reason why applications to water analysis are still scarce.
So far, MISPE has mainly been applied to the extraction of pesticides [159-164,
294,295] as well as 4-nitrophenol [165] from different aqueous samples. The
selectivity of the sample cleanup can be further enhanced by combing RAM and
MIP sorbents as shown for the analysis of triazines in river water [296].
The commercial availability of MIPs as SPE sorbents will be a necessary
prerequisite for their wide-spread use for water analysis.

22.4.7 Automated solid-phase extraction

As with other sample preparation methods described in this chapter, automated

solid-phase extraction in water analysis is of interest for two reasons: (a)
automation, i.e. unattended operation at minimal operator interventions, allows
a higher sample throughput, which reduces the involved labor and thus the
costs; and (b) automation is particularly important for field analysis, i.e. for a
continuous on-site monitoring of pollutants in water.
Automation in solid-phase extraction has been reviewed recently [166]. Time
(and cost) saving and higher sample throughput are the most important advant-
ages of automation. In general, with automated systems individual samples are
processed in series. The next sample in a series is not started before the extrac-
tion of the preceding sample is on its way or completed. It has been pointed out
[166] that with serial sample processing automated SPE systems are slower than
manual systems. However, as such automated systems operate continuously

742
during the day, night or weekend, time savings are still achieved. The currently
fastest serial processing systems extract 25-50 samples per hour [166,167]. A
significant increase in sample throughput, however, is only achieved if solid-
phase extraction is done in parallel [167-169], i.e., if large numbers of samples
are extracted simultaneously videe infra). Parallel processing systems can han-
dle up to 400 samples per hour [167]. Although with automated SPE a higher
precision (and accuracy) is generally achieved, systematic errors (resulting, for
instance, from blocked cartridges) may occur undetected. Another problem with
automated systems is analyte carryover which may lead to erroneously elevated
analyte responses at low levels. Moreover, sample stability may become a prob-
lem, in particular if a large number of samples is processed over a long period,
such as the weekend.
For automated solid-phase extraction to be worthwhile, a minimum number
of samples is required. It has been stated that this break-even number is about
10 [166]. Automated solid-phase extraction can be performed on-line or off-line
as described below.

22.4.7.1 On-line solid-phase extraction

The conventional SPE with cartridges can be coupled on-line mainly with GC
(MS) or HPLC (MS). With both methods, the analytes in aqueous samples are
first trapped on a small cartridge or so-called pre-column and then, in a second
step, eluted. The entire eluate is thereby transferred onto the analytical GC or
LC column, which causes a significant decrease in the limit of detection (LOD)
when compared to the off-line method, provided that identical sample volumes
are used.
SPE-LC. In on-line SPE-LC, the extraction is carried out in analogy to the
off-line approach employing pre-columns/cartridges for analyte enrichment and
switching techniques. Thus, SPE-LC represents a special type of the many
column-switching techniques. After conditioning the pre-column with a suitable
solvent or solvent mixture (usually by flushing with methanol and, next, with
water), the aqueous sample is applied to the pre-column. A drying step is not
required. The retained analytes are eluted directly from the pre-column onto the
analytical column using the mobile phase of the LC system. The schematic
set-up for the on-line SPE-LC system is shown in Chapter 12.
Devices for on-line LC/MS are commercially available. They use small
pressure resistant pre-columns for sample enrichment packed with 5-40 )m
particles [48]. With such commercial instruments, more than 100 pre-columns
can be loaded automatically into the instrument prior to the initiation of a run.
As most components are retained at the beginning of the pre-column, back-flush
elution leads to a better peak shape in the successive LC analysis, although in
practice forward elution is usually used.
There are two limitations of on-line SPE-LC systems:
(a) As a result of the small pre-column (10 x 2-4 mm i.d.) [170] packed with
20-100 mg sorbent, breakthrough volumes are significantly lower when

743
compared to conventional SPE cartridges. Breakthrough is avoided by using
smaller sample volumes (e.g., 10-100 ml). As discussed above, breakthrough of
more polar compounds is reduced if polystyrene-divinyl benzene (PS-DVB) is
used instead of the conventional C8 or C,,. As the entire analytes trapped on the
sorbent are transferred onto the analytical column, comparable or even better
detection limits are achieved with on-line SPE-LC using 10-100 ml sample
volumes when compared with conventional SPE extraction of a 1 1water sample.
Thus, the use of on-line SPE-LC is particularly attractive if only small sample
volumes are available.
(b) The sorbents of the pre-column and the analytical column have to be
compatible. Ideally, the sorbents of the pre-column and the analytical column
should be identical. This is easily realised if C18 sorbents are used for SPE.
However, an on-line coupling of SPE with LC using PS-DVB pre-columns and
C, 8 analytical columns [99,171-183,270] or graphitized carbon [184-186] is also
readily achieved.
While the accuracies are comparable [189-191], a better precision is often
achieved with on-line SPE-LC than with off-line extraction [187,188].
Apart from commercial equipment specially designed for on-line SPE-LC,
conventional LC instruments with autosamplers may be modified and prog-
rammed for on-line extraction [192]. In this case, the aqueous sample is injected
onto a pre-column in several 1 ml fractions for sample enrichment before elution
onto the analytical column follows.
Sample enrichment using a restricted access material (RAM), as discussed in
Section 22.4.6, is usually only done on-line, as disposable RAM cartridges are too
expensive and automation is desirable for routine analysis of a large number of
similar samples. Sample volumes are usually restricted to < 50 Al. After each
injection, the RAM pre-column is regenerated by washing and the regeneration
is done in parallel to the chromatographic separation of the next sample.
In water analysis, the use of RAM sorbents is particularly attractive for the
removal of humic substances, as mentioned above [133,134]. In addition,
immunosorbents (ISs) [140,141,193-197] and molecular imprinted polymers
[MIPs] [163-165,198-201] have also been used successfully in on-line SPE-LC
applications for water analysis.
SPE-GC. For the analysis of volatile compounds amenable to GC, in water
analysis on-line coupling of SPE with GC is more attractive than on-line coup-
ling with LC, as the higher separation efficiency of GC columns and the ease of
GC/MS couplings lead to a better selectivity and specificity of the overall analyti-
cal method. On-line SPE-GC coupling has been reported and reviewed in detail
[170,202]. Sample enrichment on the SPE cartridge is similar to that for SPE-LC
(pre-column conditioning, sample application, washing with a small amount of
water). However, for a better compatibility with the GC, typical sample volumes
for water analysis are 1-10 ml with SPE-GC compared to 10-100 ml with
SPE-LC. In contrast to SPE-LC, in SPE-GC residual water on the pre-column is
not compatible with the GC and has to be removed in a drying step (vide infra).

744
Desorption (elution) is achieved with 50-100 pL of a suitable organic solvent. In
order to transfer the entire desorption solution to the GC, the GC system has to
be equipped with a "large-volume-injection-system". A schematic set-up of an
on-line SPE-GC system is shown in Fig. 22.3. The system has at least three
switching valves: one to switch between aqueous (conditioning and sampling)
and organic (elution) solvents, a second to supply drying gas, and a third to direct
the desorption solution to the GC system. As shown in the figure, for on-column
large volume injection (e.g., 100 l) a system of two pre-columns and an analyti-
cal column is used. The first uncoated pre-column (e.g., 5 m long) acts as a
retention gap using partially concurrent solvent evaporation (PCSE) [203-205],
while the analytes are focused on a second coated pre-column (e.g., 2 m long), the
so-called retaining column. After this retaining column the solvent is split off via
a T-splitter (solvent vapour exit, SVE) [202,206,207].
For analyte retention, PS-DVB sorbents are best suited because of their
higher breakthrough volume (when compared to C or C,,). Non-polar to
medium-polar compounds are quantitatively retained on a 10 x 2-4 mm i.d.
pre-column and with sample volumes of < 10 ml (or even up to 100 ml). For
desorption (elution) of typical water pollutants, methyl- or ethylacetate are used
[202,208,209]. Originally, the cartridge was dried by purging with nitrogen. This
procedure is rather time-consuming. Thus, a short drying cartridge (e.g., packed
with silica) inserted between the SPE pre-column and the GC is recommended
[170,210-212]. The drying agent is reconditioned by heating after the GC run
and can be re-used for up to 100 times.
Devices for on-line SPE-GC are commercially available and similar to those
used for on-line SPE-LC (see Fig. 22.3). With these devices all steps are carried
out automatically under computer control. One of these systems allows > 100
cartridges to be automatically loaded into the instrument. However, for the
analysis of water with a low pollutant level (e.g., surface or tap water), cartridges
can be re-used at least 100 times [202], while it is recommended to replace the
retention gap after 50-100 GC runs [202].
It has been stated that this approach to on-line SPE-GC is applicable to all
analytes which can be trapped on the (polymeric) pre-column and dissolved by

Fig. 22.3. Schematic set-up of

a system for automated on-
line SPE-GC-MS: 1, solvent
channels; 2, purge leak
restriction; 3, waste; 4, single
piston LC pump; 5, SPE pre-
column; V1-V3, valves; SDU,
solvent delivery unit; SVE,
solvent vapor exit; OCI, on-
column injector [202] (court-
esy of Elsevier, Amsterdam,
Netherlands).

745
methyl- or ethylacetate and which are amenable to GC. The potential of on-line
SPE for on-site field analysis is evident and will be reported in Section 22.4.8.
A potential problem of on-line SPE-GC is the loss of volatile analytes due to
the approach of partially concurrent solvent evaporation (PCSE) and the instal-
lation of solvent vapor exit [170,213,214]. With an optimized system, analytes as
volatile as monochlorobenzene and xylene are recovered.

22.4.7.2 Off-line automation of SPE

Off-line automation of SPE was reviewed by Rossi and Zhang [166]. Originally,
these robots used only cartridges (discrete columns) for sample enrichment
[215]. With some of these systems, serial and parallel processing are possible. In
water analysis, these automated SPE systems are widely used for routine
analysis and are the technique of choice in many German water laboratories for
pesticide analysis. Such systems can also be used for an automated method
development.
More recently, SPE material has also become available in a 96-well-format
(microtiter plates), and workstations which handle this format are on the
market. They use parallel sample processing and are thus ideally suited to high
throughput analysis in drug development, while discrete extraction columns are
still preferred for water analysis [1661.

22.4.8 Solid-phase extraction in field analysis

In water analysis, sampling is predominantly still done in the field, while sample
preparation and instrumental analysis are carried out in a central laboratory.
Rapid screening tests and various sensors have been developed and are, in part,
commercially available for many pollutants. They are mainly used to monitor,
e.g., ground and surface water near hazardous waste sites and surface water
after an accidental release of pollutants into the aqueous environment. How-
ever, these tests are usually at best semi-quantitative and give information on a
very limited number of target pollutants. For a detailed qualitative and
quantitative analysis of water samples in the field, chromatographic instru-
ments such as GC or LC coupled with selective detectors (in particular with a
mass spectrometer) are still needed and here SPE plays an important role.
Water extraction and instrumental analysis in the field can be done off-line
or on-line. Off-line methods are important. After solid-phase extraction of water
samples using, e.g., a cartridge, it is much easier and safer to transport the
cartridge to the central laboratory than a 1 L water sample in a glass bottle. This
holds in particular for samples collected at remote sites [216-222]. Moreover,
using this approach sample stability is usually improved. The stability of various
analytes sorbed on disposable cartridges has been investigated as a function of
the storage time and temperature, type of sorbent and cartridge, pH and matrix.
While in general an acceptable stability was observed with all analytes, this was
not the case for some non-stabile organophosphorous pesticides [173]. Polymeric

746
sorbents appear to be better suited to long-term storage. Polar phenolic com-
pounds, for instance, could be stored at -20°C for two months [81] while a good
stability over a seven-week-period was reported for several groups of pesticides
[224].
For an on-site monitoring of individual pollutants, the fully automated
on-line SPE-GC or SPE-LC systems are particularly suitable. Thus, under the
"Rhine Basin programme", a System for Automated Monitoring of Organic
Compounds in Surface Waters (SAMOS) based on on-line SPE-LC was devel-
oped and tested [188]. This system was considered as an early warning system.
At analyte concentrations of 1 lg/l, a good precision ranging from 1-15% was
observed, with the exception of polar compounds (due to matrix interference and
partial breakthrough) [225]. The robustness of the system was studied in two
laboratories during five and seven months long periods. No major problem was
encountered in over 1000 analyses. An even better performance is achieved
when the water sample is filtered [226].
For more volatile compounds, the on-line SPE-GC coupled with a mass
spectrometer is probably better suited (although most likely less robust than
on-line SPE-LC systems). If combined with a sample delivery unit (as, for
instance, available on commercial on-line SPE-GC/MS systems), an unattended
run is possible [202]. Non-target analysis is done with the MS running in the
scanning mode, while selected ion monitoring (SIM) is used for target analysis.

22.4.9 Applications of solid-phase extraction for water analysis

During the last two decades numerous applications of solid-phase extraction for
water analysis have been reported. At present more than 200 papers appear
every year in which the use of SPE for water analysis is reported. Thus, an
exhaustive overview of the application of SPE in this area is not possible here.
Various applications of special SPE techniques such as the use of polymeric and
carbon based sorbents, restricted access materials, immunosorbents and molec-
ular imprinted polymers were reported in Sections 22.4.5 and 22.4.6. Many
further applications are found in a review by Hennion [48].
SPE is particularly suited to the extraction of moderately polar to polar and
semivolatile compounds from aqueous matrices. Pesticides are typical analytes
which belong to this polarity range. Thus method developments for pesticides
dominate the SPE applications as summarised in a recent review [49] in which
the monitoring of triazines and their degradation products in natural water is
discussed in detail. Moreover solid-phase extraction of carbamate pesticides
[227], acidic herbicides [228] and quaternary ammonium herbicides [229] has
been reviewed. An overview of multiresidue analysis based on SPE was present-
ed [230]. In further reviews the application of SPE to the extraction and analysis
of polycyclic aromatic hydrocarbons [231], phenols [232], antifouling compounds
[224] and aromatic sulfonates [117, 231] has been summarised. In addition SPE
was extensively applied to the extraction of surfactants and estrogens. More

747
TABLE 22.1
Recent applications of on-line solid-phase extraction for water analysis (coupled to liquid
chromatography (LC), capillary electrophoresis (CE), supercritical fluid chromatography (SFC)
or gas chromatography (GC))
Compound class (type of agent) Instrumental Special methods Ref.
analysis
Pesticides (metabolites) LC 49,227,271-280
(quaternary ammonium) LC Ion pair extr. 274,277
(carbamates) LC Post-column deriv. 278
(acidic herbicides) LC 275
CE 287
SFC 288
Polar pollutants LC 279,280
LC Dual pre-column 293
Endocrine disruptors (estrogenic LC 246,253
hormones)
(alkylphenols, bisphenol A) LC 259
Antifouling compounds LC 281,282
GC 290
Naphthalene sulfonates LC Ion-pair extr. 283
Phenols LC 279,284,285
4-nitrophenol LC Mol. imprinted polymer 165
Anilines LC 285
Cyanobacterial toxins LC 286
Organic pollutants GC 289

recent reports include non-ionic [233-238], anionic [239-242] and cationic

[243-245] surfactants, estrogens [233,246-259] and aromatic sulfonates, in
particular naphthalene sulfonates [191,260-269].
Most extractions by SPE are still performed off-line, although SPE coupled
not only to GC (MS), CE (MS), but in particular LC (MS) gain in importance.
Earlier applications of on-line SPE have been cited in Sections 22.4.5-22.4.8
[133, 134,141,163-165,174-201]. Recent applications of on-line SPE for water
analysis coupled to LC, CE, SFC and GC are summarised in Table 22.1.

22.5 SOLID-PHASE MICROEXTRACTION (SPME)

22.5.1 General considerations

In solid-phase microextraction, the sample (gaseous or aqueous) is exposed to a

polymer-coated surface as stationary phase. The analytes partition into the
stationary phase until an equilibrium (or preequilibrium) is achieved, followed
by thermal elution (for GC) or solvent elution (for LC). In contrast to solid-phase
extraction (SPE), where the sample is passed through the sorbent (stationary
phase), a flow-through equilibrium (or preequilibrium) is realised and exhaust-
ive extraction attempted, SPME uses a non-exhaustive batch extraction.

748
 This relatively new sample preparation method was invented in 1990 by
Pawliszyn [52] and rapidly spread into various areas of analytical chemistry with
more than 700 papers published today. The method is described in several books
and reviews [297-314]. Aqueous samples were among the first matrices studied
and water analysis is still one of the main application fields of SPME.
SPME is a very simple solventless sample preparation method which
combines sampling, sample enrichment and sample cleanup in one step. With
SPME, the extracting polymer phase is either exposed directly to the aqueous
sample or to the headspace above the sample. The theory of SPME is presented
in detail in Chapter 13. Here, only the basic equations describing the extraction
process from an aqueous sample and its headspace are repeated. If the extracting
polymer phase is exposed to the headspace of an aqueous sample (and the
humidity in the headspace is neglected), the equilibrium between the three
phases is described by two distribution coefficients, the distribution coefficient
between the gas and the aqueous phase, Kgw, and the one between the polymer
and the gas phase, Kpg, where the overall distribution coefficient between the
polymer phase and the aqueous phase, Kpw, is
KpW = Kpg. Kgw (22.9)
If extraction occurs by absorption (rather than adsorption) by the coating, the
extracted mass of the analyte is
Kpw Vp CoVW (22.10)
n- (22.10)
Kpw .V. +Kgw .Vg +Vw
Vp, Vg, V, are the volumes of the polymer, gas and aqueous phases, respectively,
and C0 the initial concentration of the analyte in the aqueous sample. Equation
(22.10) states that the amount of analyte extracted is independentof the location
of the extractingpolymerphase, which may be placed in the headspaceor directly
in the aqueous sample. If there is no headspace, Eq. (22.10) simplifies to

n= (22.11)
Kw Vp +Vw
which applies to direct SPME.
Usually, the volume of the polymer coating is much smaller than the volume
of the aqueous sample (Vp < < Vw) so that Eq. (22.11) further simplifies to
N = Kpw Vp. Co (22.12)
i.e., the extracted analyte mass is independent of the volume of the aqueous
sample which is important for field sampling.

22.5.2 Procedures

SPME devices. The original and still most popular procedure for SPME of
aqueous samples uses a small fused silica fibre coated with a thin polymer phase

749
attached to a syringe-like holder (see Chapter 14). (As mentioned above, the
volume of the extracting phase is very small, usually less than 1 L, when
compared to the sample volume, therefore "microextraction"). The syringe-like
holder provides protection of the fibre during transport and allows a piercing of
the rubber septum of the sample vial and the GC or LC injector. For GC
applications in water analysis (which by far prevail), the fibre is first exposed
either directly to the aqueous sample (direct SPME) or to the headspace (head-
space SPME) until an equilibrium (or pre-equilibrium) is reached and then
introduced (manually or by autosampler) into the hot injector, where thermal
desorption occurs. A polymer-coated fibre can also be used for LC application.
After analyte extraction, the fibre is exposed directly to the mobile phase in a
special interface [315], where solvent elution of the analytes occur. Other SPME
devices have also been reported for water analysis: for GC analysis a stir bar can
be coated with a polymer phase [316,474]) leading to a significant increase in the
extracting phase volume and thus to lower detection limits for water analysis.
For LC applications, the inner surface of a capillary can be coated with the
extracting phase (in-tube SPME) or the coating of a GC open tubular column can
be used for extraction [317-323]. The extraction capillary is placed between the
sample injector loop and the injection needle of a commercial LC instrument. To
reach an equilibrium or preequilibrium in in-tube SPME-LC, the sample or an
aliquot of the sample is repeatedly drawn through the extraction capillary by the
autosampler.

Direct SPME versus headspaceSPME. For water analysis, both direct SPME
and/or headspace SPME [300,325-329] can be used. Headspace SPME is the
method of choice for dirty water samples, such as waste water, if the analytes are
sufficiently volatile. The extraction phase (fibre) is protected from adverse
effects by the matrix, e.g., precipitation of non-volatile and high molecular
weight compounds such as humic substances. It appears that most compounds
amenable to GC can also be extracted by headspace SPME [327]. As mentioned
above, the amount of extracted compounds does not depend on the location of
the fibre. However, the choice of the sampling mode has a significant impact on
the extraction kinetics. When the fibre is exposed to the headspace, the analytes
are first removed from the headspace, followed by an indirect extraction from
the aqueous phase. Thus, volatile analytes are faster extracted than semi-
volatiles. In general, the equilibration times for volatile compounds are shorter
for headspace SPME than for direct SPME. This is i.a. due to the higher
diffusion coefficient in the gas phase when compared to the aqueous phase.
Headspace analysis of semi-volatiles can be accelerated by increasing the
temperature and by agitating the sample (e.g., by stirring). With increasing
temperature the vapor pressure of the analytes increases, but the distribution
coefficient Kw decreases. Thus, under equilibrium conditions, the extraction
yield as a function of the temperature passes through a maximum when the fibre
is not cooled [330].

750
 Sample volume. The sample volume can be neglected if the distribution
coefficient between polymer and aqueous phase is not very high [331,332]. It is a
significant advantage for water analysis that small sample volumes (e.g., 2 ml)
are sufficient for quantitative analysis. Thus, the aqueous samples can be placed
directly into an autosampler, provided the sample is sufficiently agitated
[333-338]. Thus, SPME is particularly attractive if only small volumes of water
samples are available. In spite of the small sample volume and although in
general only a small fraction of the analytes is extracted by SPME, low method
detection limits are still achieved as the entire extracted analytes are transferred
to the chromatographic system.

22.5.3 Extraction phases

For fibre SPME in particular a variety of extraction phases with different polari-
ties and thus different selectivities are commercially available [339]. The selec-
tion of a suitable extraction phase is often based on the principle "like solves
like". The extraction phases can be differentiated according to the underlying
extraction principle: Liquid polymeric phases such as polydimethylsiloxane
(PDMS) and polyacrylate (PA) extract by absorption and have been widely used
for water analysis. Solid sorbents such as carboxene or polystyrene-divinyl-
benzene (PS-DVB) extract by adsorption [341]. If extraction occurs by absorp-
tion into the liquid polymeric film (until a constant concentration in the bulk of
the polymer is reached), the extraction equilibrium of a given compound is only
described by its distribution coefficient and not influenced by other analytes.
Thus, even if an analyte is present in large excess it will not displace minor com-
ponents [52,327,343-346]. In contrast, SPME fibres coated with porous solid
sorbents extract by adsorption. They are sometimes preferred because of their
high sensitivity (and selectivity). As only a limited surface area is available for
adsorption, analytes will compete at higher concentrations for the adsorption
sites. This causes a displacement of compounds with poor affinities towards the
sorbent at longer extraction times. In addition, a non-linear extraction yield will
be observed at higher concentrations. Porous particles such as carboxene and
PS-DVB are often embedded in other partially cross-linked phases, such as
PDMS or carbowax (CW). A carbowax-templated resin is used for SPME-LC
applications [339].

22.5.4 Calibration

Unlike exhaustive sample preparation methods where calibration of the analyti-

cal instrument can be done with standard solutions in organic solvents, for
SPME calibration over the entire method using spiked water samples is
required. For liquid extraction coatings, four calibration procedures may be used
[297,346]: external standard calibration, internal standard calibration (and
isotope dilution), standard addition, and use of distribution constants. For the

751
last method the distribution coefficient Kpg between the polymeric and the gas
phase can be determined based on GC retention times [347,348], while the
coefficients for the polymer-water phase (Henry's constant) can be obtained
from physicochemical tables [349]. For solid extraction phases, these calibration
modes are only applicable at low concentrations, i.e., within the linear range of
the adsorption isotherm. At higher concentrations, a calibration based on the
diffusion constant is possible [324].

22.5.5 Derivatization
A coupling of SPME with GC is easier than with LC, but restricted to compounds
amenable to GC (i.e. compounds of sufficient thermal stability and volatility).
The range of compounds which can be analyzed by SPME-GC can be extended if
non-volatile compounds are derivatized. Derivatization can be done in the
aqueous sample (i.e. prior to extraction), in the injector, and directly in the fibre
coating [350-360].

22.5.6 Parameters determining the extraction

Polymer coatings [297,298,330,339]. As for any given analyte very different

distribution constants between the polymer and the aqueous phase are observed
for different polymer coatings some selectivity can be achieved during extraction
by proper choice of the polymer coating. For water samples containing analytes
in concentrations covering several orders of magnitude, liquid polymer coatings
are the extraction phases of choice, since with these coatings displacements of
trace analytes by other analytes with higher affinity to the extraction phase are
of minor concern videe supra). If compounds of very different polarities are to be
analyzed simultaneously, the polyacrylate phase is better suited than the polydi-
methylsiloxane phase. For trace analysis of, for instance, aromatic hydrocarbons
or halogenated hydrocarbons in water samples, substantially better sensitivities
are achieved with the carboxene fibre [361].
Temperature. As discussed above, the temperature influences the kinetics
and the thermodynamics of the extraction process in opposite directions [297,
345]. Thus, the extraction yield as a function of the temperature usually passes
through a maximum as discussed with headspace SPME [327]. To achieve good
precision and accuracy, water samples should be thermostatted prior to analysis,
which is particularly important for field analysis (ide infra).

22.5.7 Matrix effects

pH. For ionizable compounds, i.e. weak acidic or basic compounds, SPME shows
the expected strong dependence on pH both for direct and headspace SPME
[353,362-365], i.e. the extraction yield increases if, by proper choice of the pH,
the ionic organic compound is transformed into its neutral form as already
reported for other extraction methods.

752
 Inorganic salts. It was observed repeatedly that salt addition to aqueous
samples enhances the amount of analyte extracted by SPME [327,343-346,
362,364,366,367] as it is also observed in other extraction methods (LLE, SPE).
In general, the effect of salt addition increases with the polarity of the compound
[343]. The effect is observed both for direct and for headspace SPME [327].
Organic solvents, lower alcohols. For method validation, aqueous samples
were often spiked with reference compounds dissolved in polar volatile solvents.
It was demonstrated for methanol both for direct SPME and for headspace
SPME that the extraction yield only decreases significantly at methanol concen-
trations of >5% [327,343,345,368-370].
Natural organic matter. Humic compounds (natural organic matter, NOM)
are often present in groundwater of concentrations of > 1 mg/l, i.e., in substan-
tial excess to anthropogenic pollutants. Thus, displacement reactions during the
absorption of analytes may be anticipated. However, studies with pesticides and
other compounds in aqueous samples did not reveal a pronounced effect of
humic compounds on the extraction of other pollutants, provided the concentra-
tion did not exceed 10 mg/l [343,345,371,372]. The impact of natural organic
matter in aqueous samples on the extraction can be completely eliminated by
using headspace SPME.

22.5.8 Field analysis

SPME is particularly well suited to field analysis of environmental water. As

discussed above, the amount of analytes extracted by SPME from an aqueous
sample hardly depends on the sample volume. Thus, a SPME fibre can be placed
directly into a lake or water reservoir. Analysis can be done on-site with a
portable GC or GC/MS. Alternatively, the fibre (well protected against any
contamination) may be shipped to a central laboratory for analysis allowing the
monitoring of environmental waters at very remote sites.
A fully automated instrumental set-up for a quasi-continuous on-site moni-
toring of organic pollutants in surface, waste and process water or during ground
water remediation is described [373] which can be operated unattended for
several weeks. The schematic design is shown in Fig. 22.4. This set-up allows an
automatic pH adjustment (which is of importance for waste water analysis).
Quantification is achieved by adding a small amount of internal standard during
each sampling cycle, which at the same time allows a monitoring of the proper
performance of the instrument by remote control. For difficult aqueous matrices
such as industrial waste water, headspace analysis is preferred as it enhances
the robustness of the instrumental set-up [327].

22.5.9 Applications

Various applications of the SPME method to the analysis of (environmental)

water are summarized in Table 22.2. By far the largest number of applications

753
Fig. 22.4. Schematic set-up of automatic analyzer based on solid-phase microextraction: 1,
sample preparation vessel; 2, peristaltic pump; 3, magnetic stirrer; 4, pH electrode; 5, level
switch; 6, programmable burette (titration); 7, progammable burette (internal standard); 8 and
9, peristaltic pumps; 10, extraction vessel) [2571.

were reported for pesticides that are amenable to GC. Phenoxy acid herbicides
are derivatized prior to GC analysis. Moreover, polar pesticides were analyzed by
LC after SPME. Many studies have been published on the SPME-GC analysis of
volatile organic compounds (VOCs), including halogenated hydrocarbons and
aromatic hydrocarbons (fuel related compounds), where headspace SPME is
often preferred to direct SPME.

22.6 MEMBRANE-BASED TECHNIQUES

22.6.1 General considerations

Membranes are particularly well suited to sample preparation of aqueous

samples. However, they have found their way into the analytical laboratory only
in the last 10 years. Again, analyte enrichment and sample cleanup are achieved
simultaneously.
Generally, a membrane can be defined as a "selective barrier between two
phases" [479]. When a driving force is applied across a membrane, transport of
matter from one phase (the donor phase), i.e., in water analysis the aqueous
sample, to the other (the acceptorphase)occurs, referred to as flux. The driving
forces are (a) pressure difference, (b) electric potential difference, and (c) con-
centration difference. Concentration gradients are mainly applied as driving
force for membranes used for sample preparation techniques in water analysis.
Either porous or non-porous membranes are used to separate the analyte
from the aqueous phase and from matrix interferences. With porous membrane
techniques (polypropylene, polysulfone or cellulose derivative polymers), the

754
TABLE 22.2
Applications of solid-phase microextraction (SPME) to the analysis of aqueous samples
Compound class (type of Instrumental Special methods Ref.
agent) analysis
Pesticides Review 374
GC 343, 366, 370,
375-400, 473, 478
GC Derivatization 401-403
LC 404,405
Phenols GC 354, 362, 406, 408
GC Headspace 353, 365, 407
Polycyclic aromatic GC 300, 328, 337,
hydrocarbons 409-411,413, 415,474
LC 412
Aromatic hydrocarbons, fuel GC 414-421
related compounds GC Headspace 300, 328, 329, 422-425
IR 426
Volatile organic compounds, GC 427-431, 456, 477
chlorinated hydrocarbons GC Headspace 361, 432-435
IR 436
Volatile polar compounds GC (headspace) 437, 438
(solvents)
Amines GC 439, 440
Derivatization 355
Nitroaromatic compounds, GC 441
explosives IR 442
Off-odour compounds GC 443, 444
Organometallic compounds GC (AED) 358, 359, 445-447
(Hg, Mn, Pb, Sn) GC (AED, ICP) Derivatization, 448-450
headspace
LC 451
Inorganic ions GC 452
Complexation 453
Polychlorinated biphenyls GC 410,411,454, 455
Headspace 338
Surfactants LC (Derivatization) 457, 458
Carbonyl compounds GC Derivatization, 459, 460
headspace
Hydroxyaromatics LC 461
GC 462
Miscellaneous GC 463-467, 469-471
LC 468, 472
Biologically active substances GC 475
Phthalates GC 476

two (liquid) phases are in direct physical contact through the pores. Separation
with these membranes is only based on size exclusion, where sufficiently small
molecules can permeate through the pores and larger ones cannot. Non-porous
membranes consist of a liquid or polymer film. Analyte molecules must dissolve

755
into this film to be able to pass through. Thus, separation with non-porous
membranes is based on the same principles as in LLE [480].
With porous membranes all small molecules may pass the membranes. With
non-porous membranes, however, one exploits the different physico-chemical
properties of the varying (small) analytes to achieve a separation between the
different compounds in much the similar way as in LLE (even if the analyte
molecules are equal in size).
Thus, while porous membranes are mainly used for sample cleanup to, e.g.,
remove proteins from biological samples, non-porous membranes allow the
extraction and enrichment of a sample and, in addition, a cleanup.
Depending on the driving force, three sample preparation techniques use
porous membrane: (a) dialysis (a concentration-driven process), (b) electro-
dialysis (an electric potential-driven process) and (c) filtration (a pressure-
driven process) [480]. Filtration is an important step in water analysis, as many
standard methods require that water samples be filtered before extraction to
exclude particles (and, unfortunately, also hydrophobic organics bound to these
particles). The other two techniques have been mainly applied to biological and
food samples. They are not further discussed here; the reader is referred to a
review 480] where these methods (including filtration) are described and
discussed in detail. Dialysis may in principle be used for cleanup of water
samples, i.e., to remove humic substances from environmental samples. For this
purpose, membranes with much smaller pores than those used for biological
samples have to be applied [481]. This approach may be promising in particular
if analyte enrichment is achieved in a second step, i.e., by solid-phase extraction,
but does not seem to have been used so far.

22.6.2 Extraction with non-porous membranes

Sample preparation methods using non-porous membranes have mainly been

developed by Jinsson and Mathiasson. For details, the reader is referred to
several reviews by these authors [482-455] and Chapter 15. In this section the
principles and the application of this technique to water analysis and to field
analysis of aqueous samples are summarised.
For sample extraction and enrichment, five different approaches have been
developed which are well suited to the extraction of water samples:
Supported liquid membrane (SLM) extraction. The application of supported
liquid membranes to sample preparation has been reviewed several times [480,
482-487] and is also discussed in Chapter 15. A SLM device consists of a
hydrophobic porous membrane as support impregnated with a water-immiscible
organic solvent which is held by capillary forces in the pores of the support
membrane. The liquid in these pores represents the actual "liquid membrane",
which is in contact with two aqueous phases. Typical solvents used as liquid
membranes are long-chain hydrocarbons such as n-undecane or, for more polar
compounds, dihexylether or trioctylphosphate [482]. SLM extraction is

756
particularly suited to the extraction of ionizable organic compounds and uses the
following principle: in the donor phase (the sample), the analytes are first
transformed into their neutral extractable form using a pH adjustment. They
are then transported through the liquid membrane into the acceptor phase,
where they are transformed into their ionic form by further proper pH adjust-
ment and thus prevented from re-entering the membrane, i.e., the analytes are
trapped in the acceptor phase. For a complete trapping, the pH of the acceptor
phase should be at least three units lower than the pKa of the analyte. Neutral
compounds may be extracted by SLM, but will never be trapped in the acceptor
phase and no enrichment is achieved.
If organic compounds are present in their ionic form over a large range of pH
values, an addition of an ion pairing agent allows the extraction in the neutral
form, as shown for anionic surfactants [488]. As the extract is aqueous, the SLM
technique is best compatible with reversed phase LC or ion chromatography
(IC).
Microporous membrane liquid-liquid extraction (MMLLE). With the
MMLLE technique, the acceptor phase is an organic solvent and the same
solvent forms the liquid membrane by filling the pores of a porous hydrophobic
membrane [482,483]. In the case of water analysis the aqueous sample repre-
sents the donor phase. MMLLE is particularly suited to the extraction of highly
hydrophobic compounds which are readily extracted from water into an organic
solvent. In contrast to the SLM approach, they are not back-extracted into a
second aqueous phase. Thus, MMLLE is based on the same principles as LLE
but usually performed in a flowing system. In contrast to SLM, the extract ends
up in an organic phase and is thus best compatible with GC analysis. The
extraction efficiency in MMLLE is determined by the distribution coefficient
between water and the organic solvent.
Polymeric membrane extraction (PME). Moreover, a non-porous polymeric
membrane (such as a silicon rubber) may also be used for extraction. In water
analysis, the donor phase is the aqueous sample and the acceptor phase in
general an organic solvent (although a system, i.e., aqueous phase/membrane/
aqueous phase, is also conceivable). As the organic solvent penetrates into the
polymer causing it to swell considerably, the situation is similar to MMLLE. The
extraction efficiency is determined by the analyte-membrane distribution coeffi-
cient. As diffusion coefficients in polymers are lower than in liquids the mass
transfer is lower than in MMLLE [482].
Two further special membrane extraction methods are described in Section
22.6.4.

22.6.3 Analyte enrichment and selectivity with membrane

techniques

Analyte enrichment. In membrane extraction techniques, the enrichment factor

is defined as [482]

757
E = CA/Cw (22.13)

where CA and Cw are the concentrations in the acceptor phase and the aqueous
sample, respectively [482]. In MMLLE and PME, the maximum enrichment
factor for an aqueous sample, Ee(,,, is equal to the distribution constant, K5, of a
given analyte between water and the organic solvent used as acceptor phase, i.e.,
the equilibrium concentration C(eq) of the analyte in both phases is given as

Ee(m-) = CA(eq)/Cw(eq) (22.14)

As with conventional LLE, high distribution constants are required to achieve a

high sample enrichment.
In contrast, the analyte enrichment in SLM is not limited by the distribution
constant, as analytes are trapped immediately in the acceptor phase. Here, the
maximum enrichment factor is determined by the pH of the acceptor phase and
the dissociation constant of the analyte [482]:
logEe(m), = PKa-pH (22.15)
Sample enrichment is described in more detail in Chapter 15.
With a flowing acceptor phase which is continuously replenished, the enrich-
ment factor will be significantly lower than Ee(mx). In this instance, analyte
enrichment is possible using an intermediate trap placed between the extraction
cell and the analytical instrument. Such trapping with a sorbent is difficult with
an organic solvent as acceptor phase, but readily achieved with a gaseous
acceptor phase as realised in the MESI technique (vide infra).
Cleanup and selectivity. Non-porous membranes not only provide an analyte
enrichment, but also simultaneously allow for a sample cleanup. For analyte
extraction, analytes have to dissolve in the membrane, diffuse through the
membrane and redissolve in the'acceptor phase. Diffusion coefficients of ionic
and of high molecular mass compounds in hydrophobic membranes are very low
so that diffusion is negligible. This is used in SLM for a selective extraction of
ionizable organic compounds as discussed above, but in all other techniques it is
used to remove interfering high molecular weight matrix compounds such as
proteins in biological samples and humic substances in environmental (water)
samples. A very illustrative example was given by Megersa and Jbnsson [489].
They compared solid-phase and supported liquid membrane extraction with
river water spiked with low levels of methoxy-s-triazine herbicides. LC (UV)
analysis of this water sample after SPE is significantly disturbed by a hump of
humic substances at the beginning of the chromatogram, while these interfer-
ences are removed after SLM extraction.
It has been claimed [482] that membrane extraction provides the highest
degree of selectivity and cleanup of all known sample preparation techniques.
Furthermore, it should be stressed that membrane extraction techniques use (if
at all) very small amounts of solvents.

758
22.6.4 Special membrane techniques

Membrane extraction with sorbent interface (MESI). While the membrane tech-
niques discussed so far use both a liquid donor and a liquid acceptor phase, the
MESI technique uses a gaseous acceptor phase. This technique was developed by
Pawliszyn and coworkers [311,490-495] and is described in detail in Chapter 14.
MESI systems consist of a membrane (usually a silicon hollow fibre) in which the
analytes are extracted from the surrounding liquid or gaseous sample. A flow of
inert gas transports the analytes through a cool sorbent trap where they are
enriched. In a subsequent step, they are eluted by thermal desorption and trans-
ferred to the GC. This technique is particularly well suited to the enrichment of
trace organics in air, but also for the analysis of volatiles in water. As discussed
below, for water analysis sampling of the headspace above the water is preferred
[494, 495] (see Section 22.6.6).
Membrane inlet mass spectrometry (MIMS). In membrane inlet mass spectro-
metry, the organic compounds diffuse from the aqueous sample through a non-
porous hydrophobic polymer membrane (usually a silicon membrane) into the
vacuum of the ion source of a mass spectrometer. This hydrophobic membrane
favors non-polar organic compounds which dissolve well in such a membrane.
The MIMS technique was pioneered by Cooks [496] who introduced the
concept of a direct insertion membrane probe. With this technique the mem-
brane is inserted directly into the mass spectrometer and the aqueous sample
transported to the membrane inlet. The technique was reviewed [497-506] and
is described in detail in Chapter 16 in which the theory of membrane extraction
is also presented. As discussed in Chapter 16, the entire extraction process
consists of three steps: (a) solvation of the analyte in the membrane, (b) trans-
port through the membrane and, finally, (c) evaporation from the membrane
surface into the mass spectrometer. Further parameters determining the
analyte transport through the membrane are the diffusion constant in the
membrane and the distribution coefficient between the aqueous and the
membrane phase, whereby the distribution coefficient is the more important
parameter. This distribution coefficient varies significantly from compound to
compound depending on the solubility in the membrane phase. The MIMS
technique has a high selectivity for hydrophobic compounds.
Applications of the MIMS technique were recently reviewed [500,501]. The
MIMS technique has mainly been applied to water analysis where measure-
ments of volatile organic compounds are standard applications. More recently is
has been shown that also semi-volatile organic compounds (SVOC) can be
determined (see Table 22.3).
Many of these VOCs can be measured at the ng/1 level directly from the aque-
ous phase while detection limits for SVOCs with standard MIMS techniques are
much higher.
It should be emphasised that MIMS does not include any chromatographic
separation steps which is a limiting factor if complex mixtures of pollutants,

759
TABLE 22.3
Applications of membrane techniques to the analysis of organic compounds in water
Compound class (type of agent) Extraction method Instrumental Ref.
analysis
Volatile organic compounds (VOCs) MIMS MS 496-506, 517,
518, 522-526
Phenols MIMS MS 527
Dicarboxylic acids MIMS DCI-MS 528
Steroid hormones MIMS DCI-MS 529
Hormones, biomolecules MIMS DCI-MS 530
Polycyclic aromatic hydrocarbons, MIMS DCI-MS 531
estrogenic comp., pesticides
Semivolatile compounds MIMS (trap,release) MS 532
Volatile, semivol. compounds MIMS (trap,release) MS 533
Volatile, semivol. compounds MIMS GC-MS 534
Volatile organic compounds MESI GC-MS 535
Pesticides (triazines) SLM LC 536-541
(phenoxy acids) SLM LC 542-544
(sulfonylurea) SLM LC 545-546
Phenols SLM LC 547
Anionic surfactants SLM LC 548
Anilines SLM LC 549
Pesticides (triazines) MMLLE FIA 550
Organotin MMLLE GC-MS 551
Cationic surfactants MMLLE LC 552
Other pesticides MMLLE LC 553, 554
Chlorinated organics PME GC 555
Neutral organics PME GC 556
Semivolatile organics PME LC 557
SLM = supported liquid membrane; MMLLE = microporous membrane liquid-liquid extraction;
PME = polymeric membrane extraction; MIMS = membrane inlet mass spectrometry; DCI =
desorption chemical ionization; MESI = membrane extraction with sorbent interface.

including many isomeric or isobaric compounds, are to be analysed. Thus, for

compound identification at least MS/MS techniques are required.

22.6.5 Automation

If in membrane extraction techniques the acceptor phase is not a stagnant but a

flow system, this flow can be readily coupled with GC or LC. Such coupling forms
the basis of automated devices used for on-site monitoring in the field (see
below).

22.6.6 Field analysis

Membrane extraction methods have been shown to be particularly suited to field

analysis. As with solid-phase extraction, there are two approaches to field analy-

760
sis: (a) sampling and analyte enrichment are done in the field (i.e. at a remote
site) while further cleanup (if necessary) and instrumental analysis are carried
out in a central laboratory; (b) sampling, enrichment and instrumental analysis
are performed on-site in the field.
The first approach is realised with passive membrane-based sampling
devices which have been developed for ultra-trace enrichment of hydrophobic
pollutants, i.e., for air, soil and water analysis. Here, only aqueous samples are
considered. They use the approach of polymeric membrane extraction (PME).
The samplers have been termed semi-permeable membrane devices (SPMD) and
were developed by Petty and coworkers [508-514] and have been reviewed [507].
They mimic the bioaccumulation of hydrophobic micropollutants by organisms
containing lipid compartments. Standard SPMD devices are commercially avail-
able and consist of a lay flat non-porous low density polyethylene membrane
(standard size tube: 106 cm long, 2.4 cm wide, 75-90 m thick), which contains 1
ml triolein (glyceroltrioleate) as acceptor phase. Only neutral compounds with a
molecular weight of < 600 Da diffuse through the membrane, thus precluding, in
the case of water samples, the sampling of humic substances and compounds
bound to suspended particles. The amount of accumulated analytes is propor-
tional to the concentration of the dissolved (and thus bioavailable) fraction in
water. The capacity is related to the compound's octanol-water partition coeffi-
cients, K,: the higher the K,,, the greater the capacity of the SPMD for that
chemical, where the device is only attractive for compounds with log Ko > 3, i.e.
for compounds with high bioaccumulation potential. They are typically employ-
ed in environmental water (such as river or lake water) for a period of 14-30
days, i.e. they sample water-bound contaminants integratively. The subsequent
ultra-trace analysis in a central laboratory involves a substantial cleanup, i.e.
organic solvent dialysis, size exclusion chromatography, fractionation on
Florisil, silica gel or alumina and, if necessary, additional reversed-phase chro-
matography before instrumental analysis by, e.g., GC/MS. The concentration of
water-borne pollutants is given by
Cw = CL VL/R t (22.16)
where C, is the concentration of the analyte in water, CL that in the lipid triolein,
VL is the volume of triolein, R is the sampling rate and t the sampling time.
Further details for the estimation of the ambient water concentrations have
been presented [508-514]. Moreover, pitfalls (such as biofouling) and interfer-
ences (coextraction of polyethylene oligomers, oleic acid and methyloleate) have
been discussed [507].
In a second approach, membrane extraction techniques have also been
employed for direct on-site monitoring in the field. Such on-site monitoring is
readily achieved if these techniques are on-line coupled with LC (MS) or GC
(MS) or direct with a MS. Mobile or even portable GC and GC/MS systems are
commercially available and used for such on-site monitoring. Polymeric mem-
brane extraction (PME) with an inert gas as acceptor phase is particularly suited

761
to air analysis because of the higher diffusion coefficients of organic compounds
in the gas phase when compared to the liquid one. However, it is also well suited
to on-site monitoring of organic pollutants in water (in particular volatile
compounds). In this case, it is advantageous to sample the headspace of water
samples, as proposed for the MESI technique using a cup immersed in the water
[494,495]. The MIMS technique is also particularly well suited to on-site and
on-line monitoring of organic pollutants in water [515]. For these on-site mea-
surements, a MIMS was placed in a mobile laboratory [516] or a boat [517] for
on-line measurements of VOCs in river water. The method is also suited to
on-site waste water monitoring in chemical plants [518]. In addition, membrane
inlet systems coupled with GC/MS are employed [519,520], e.g., for below
surface measurements of marine pollution. For a direct field analysis of a water
stream a membrane extraction flow cell (hollow fibre) was connected to a mobile
gas chromatograph after enrichment onto a sorption tube [521] in a similar way
to that realised with the MESI technique.

22.6.7 Applications

Applications of the various membrane techniques to the analysis of (environ-

mental) water are summarised in Table 22.2.

22.7 SAMPLE CLEANUP IN WATER ANALYSIS

As mentioned in the introduction (Section 22.1), sample cleanup is less import-

ant for water analysis than, e.g., for the analysis of soil or biological samples, in
particular as many extraction methods provide some sample cleanup at the same
time. Thus, many routine methods used for the analysis of organic pollutants in
aqueous samples do not consider any additional cleanup. There are many
reasons why further sample cleanup of an aqueous sample may be necessary: (a)
a high content of humic substances may lead to interferences with the subse-
quent instrumental analysis, and (b) in particular in samples with high pollution
levels, co-elution may occur. The latter problem is particularly severe in ultra-
trace-analysis.

22.7.1 Natural organic matter

Natural organic matter (NOM), such as humic and fulvic acids, may influence
the extraction process, as discussed in the preceding sections, and may also
interfere with the LC analysis with UV detection, leading to a broad unresolved
hump at the beginning of the chromatogram. In particular, with LLE and SPE
using conventional sorbents (e.g., C sorbents), a substantial fraction of the
NOM is coextracted, especially if low pH values are used for extraction, while
with solid-phase microextraction or headspace extraction, only a small part or
none of this NOM is extracted. With non-porous polymer membrane techniques,

762
high molecular weight NOM cannot diffuse through the membrane, while with
microporous membrane extraction methods, a size exclusion of NOM is achieved
along with a proper selection of the pore size. MMLLE methods combine size
exclusion with a (selective) extraction of smaller analytes. This approach is also
realized in solid-phase extraction using restricted access material as discussed
above.
If a size exclusion is not part of the extraction process itself, high molecular
weight NOM can be removed from the aqueous sample prior to extraction by
conventional size exclusion chromatography (SEC).

22.7.2 Column switching methods

Column switching methods using the heart cut principle are used for automated
sample cleanup of many different types of samples including water samples.
Thus, in coupled column LC (LC-LC), after sample extraction, e.g., by LLE a
coarse separation of the analytes is achieved on a first column. Using switching
valves the fraction containing the analytes of interest is separated and automati-
cally transferred onto a second column (e.g., of different polarity) for final
analysis. A similar approach is used in LC-GC coupling [558]. With normal phase
LC, this approach is straightforward and robust [559-561], as an organic solvent
is finally injected into the GC. The direct coupling of reversed phase LC with GC
is more difficult as the LC heart cut contains a high fraction of water which is
problematic for the subsequent GC analysis, as normal retention gaps cannot
handle this water. The method of normal phase LC-GC, as applied to aqueous
samples, has been critically reviewed by Louter et al. [170], who concluded that
serious limitations exist with this approach.
Although column switching modes are well established in, e.g., pesticide
residue analysis of food, they are less well established in routine analysis of
water samples. Coupled column LC has recently been used for pesticide analysis
in natural water [562-567].

22.7.3 Polluted aqueous samples

At higher pollution levels, an additional sample cleanup may be required to avoid

a coelution of analytes. The classical approach to such non-target sample clean-
up uses a separation of the analytes into an acidic, a neutral and a basic fraction,
followed by normal phase chromatography using silica, alumina or Florisil. To
achieve this, the aqueous sample is, e.g., first adjusted to a high pH (>12) and
then extracted with an organic solvent. Acidic components are transformed into
anions and remain in the aqueous phase. After acidification of this aqueous
phase, they are back-extracted with an additional portion of solvent (in the past
traditionally dichloromethane was used). If the original organic phase is acidi-
fied, basic compounds can be re-extracted into the aqueous phase and separated
at a low pH (<2). The final organic phase is subjected to normal phase column

763
chromatography. Fractionation of these neutral compounds is achieved by elu-
tion with solvents or solvent mixtures of increasing polarity. In environmental
analysis, traditional column chromatography, well known from natural
compound isolation, can be replaced by using prepacked normal phase SPE
cartridges.
This sample cleanup is particularly suited to non-target analysis of moder-
ately polluted aqueous samples. However, it is rather time-consuming (even if
the normal phase fractionation is done automatically, e.g., with automated
fraction collection, or by applying preparative LC). Highly selective sorbents for
SPE such as immunoaffinity sorbents and molecular imprinted polymers will
facilitate this sample cleanup in target analysis (see Section 22.4.6).

REFERENCES

1 A.J. Handley (Ed.), Extraction Methods in Organic Analysis. Sheffield Academic

Press, Sheffield, 1999.
2 J.R. Dean, Extraction Methods in EnvironmentalAnalysis. Wiley, Chichester, 1998.
3 A.J. Holden in: A.J. Handley (Ed.), Extraction Methods in Organic Analysis.
Sheffield Academic Press, Sheffield, 1998, Ch. 2, pp. 5-25.
4 M.C. Hennion, P. Scribe, in: D. Barcelo (Ed.), EnvironmentalAnalysis: Techniques,
Applications and Quality Assurance. Elsevier, Amsterdam, 1993, pp. 24-77.
5 D. Barcelo and M.-C. Hennion, in: Determination of Pesticides and Their
DegradationProducts in Water. Elsevier, Amsterdam, 1997, pp. 249-356.
6 V.J. Barwick, TrendsAnal. Chem., 16 (1997) 293.
7 J. Roeraade, J. Chromatogr., 330 (1985) 263.
8 E. Fogelqvist, M. Krysell and L.-G. Danielsson, Anal. Chem., 58 (1986) 1516.
9 E.C. Goosens, R.G. Bunschooten, V. Engelen, D. de Jong and J.H.M. van den Berg, J.
High Resolut. Chromatogr., 13 (1990) 438.
10 C. de Ruiter, J.H. Wolf, U.A.Th. Brinkman and R.W. Frei, Anal. Chim. Acta, 192
(1987) 267.
11 E.C. Goosens, M.H. Broekman, M.H. Wolters, R.E. Strijker, D. de Jong, G.J. de Jong
and U.A.Th. Brinkman, J. High Resolut. Chromatogr., 15 (1992) 242.
12 E.C. Goosens, D. de Jong. G.J. de Jong, F.D. Rinkema and U.A.Th. Brinkman, J.
High Resolut. Chromatogr., 18 (1995) 38.
13 E.C. Goosens, D. de Jong, G.J. de Jong and U.A.Th. Brinkman, J. High Resolut.
Chromatogr., 20 (1997) 325.
14 E. Ballesteros, M. Gallego and M. Valcarcel, Anal. Chem., 62 (1990) 1597.
15 E. Ballesteros, M. Gallego and M. Valcarcel, J. Chromatogr., 518 (1990) 59.
16 E. Ballesteros, M. Gallego and M. Valcarcel, J. Chromatogr., 633 (1993) 169.
17 B. Kolb, J. Chromatogr.A, 842 (1999) 163.
18 B. Kolb and L.S. Ettre (Eds.), Static Headspace-GasChromatography: Theory and
Practice.Wiley-VCH, New York, 1997, p. 298.
19 D. Jentzsch, H. Kruger, G. Lebrecht, G. Denks and J. Gut, Z. Anal. Chem., 236 (1968)
96.
20 K.J. Krost, E.D. Pillizzari, S.G. Walburn and S.A. Hubbard, Anal. Chem., 54 (1982)
810.
21 J.-G. Lo, T.-Y. Chen and T.-L. Tso, Chromatographia,38 (1994) 151.
22 E.R. Adlard and J.N. Davenport, Chromatographia,17 (1983) 421.
23 F.A. Dreisch and T.O. Munson, J. Chromatogr. Sci., 21 (1983) 111.

764
24 Th.A. Brettell and R.L. Grob, Part One, Int. Lab., Nov./Dec. (1985) 14.
25 Th.A. Brettell and R.L. Grob, Part Two, Int. Lab., April (1986) 30.
26 J.F. Pankow, J. High Resolut. Chromatogr., Chromatogr. Commun., 6 (1983) 292.
27 J.F. Pankow and M.E. Rosen, J. High Resolut. Chromatogr., Chromatogr. Commun.,
7 (1984) 504.
28 J.F. Pankow, J. High Resolut. Chromatogr., Chromatogr. Commun., 9 (1986) 18.
29 J.F. Pankow, J. High Resolut. Chromatogr., Chromatogr. Commun., 10 (1987) 409.
30 X.-P. Lee, T. Kumazawa, K. Sato, K. Watanabe, H. Seno and 0. Suzuki, Analyst, 123
(1998) 147.
31 K. Watanabe, H. Seno, A. Ishii and 0. Suzuki, Anal. Chem., 69 (1997) 5178.
32 R. E. Kaiser, Anal. Chem., 45 (1973) 965.
33 J.A. Rijks, J. Drozd and J. Novak, J. Chromatogr., 186 (1979) 167.
34 B. Kolb, B. Liebhard and L.S. Ettre, Chromatographia,21 (1986) 305.
35 P. Werkhoff and W. Bretschneider, J. Chromatogr., 405 (1987) 87.
36 J.F. Pankow, Environ. Sci. Technol., 25 (1991) 123.
37 J.W. Cochran and J.M. Henson, J. High Resolut. Chromatogr., Chromatogr.
Commun., 11 (1988) 869.
38 P.G. Simmonds, J. Chromatogr., 289 (1984) 117.
39 M. Shimoda and T. Shibamoto, J. High Resolut. Chromatogr., Chromatogr.
Commun., 13 (1990) 518.
40 Th. Noij, A. van Es, C. Cramers and J. Rijks, J. High Resolut. Chromatogr., Chroma-
togr. Commun., 10 (1987) 60.
41 B.B. Baker, Jr., Am. Ind. Hyg. Assoc. J., (1974) 735.
42 J.W. Cochran, J. High Resolut. Chromatogr., Chromatogr.Commun., 11 (1988) 663.
43 B. Kolb, G. Zwick and M. Auer, J. High Resolut. Chromatogr., 19 (1996) 37.
44 G.A. Junk, J.J. Richard, M.D. Grieser, D. Witiak, J.L. Witiak, M.D. Angello, R. Wick,
H.J. Svec, J.S. Fritz and G.V. Calder, J. Chromatogr., 99 (1974) 715.
45 E.M. Thurman and M.S. Mills, Solid-Phase Extraction-Principlesand Practice.
Wiley, New York, 1998.
46 N.J.K. Simpson, Solid Phase Extraction-Principles, Strategies and Applications.
Marcel Dekker, New York, 1998.
47 I. Liska, J Chromatogr. A, 885 (2000), 3.
48 M.C. Hennion, J. Chromatogr.A, 856 (1999)3.
49 H. Sabik, R. Jeannot and B. Rondeau, J. Chromatogr.A, 885 (2000) 217.
50 A. Balinova, J. Chromatogr.A, 754 (1996) 125.
51 M.E. Leon-Gonzalez and L.V. Peres-Arribas, J. Chromatogr.A, 90 2621.
52 C. Arthur and J. Pawliszyn, Anal. Chem., 62 (1990), 2145.
53 M.C. Hennion, C. Cau-Dit-Coumes and V. Pichon, J. Chromatogr.A, 823 (1998) 147.
54 D. Barcelo, G. Durand, 40 v. Bouvot and M. Neilen, Environ. Sci. Technol., 27 (1993)
271.
55 D.D. Blevins and D.O. Hall, LC-GC Int., September (1998) 17-30.
56 R.E. Majors, LC-GC Int., September (1988) 8.
57 P. Subra, M.-C. Hennion, R. Rosset and R.W. Frei, J. Chromatogr., 456 (1988) 121.
58 C.E. Werkhoven-Goewie, U.A.Th. Brinkman and R.W. Frei, Anal. Chem., 53 (1981)
2072.
59 W. Golkiewicz, C.E. Werkhoven-Goewie, U.A.Th. Brinkman, R.W. Frei, H. Colin and
G. Guiochon, J. Chromatogr. Sci., 21 (1983) 27.
60 R. Ferrer, J.L. Beltran and J. Guiteras, Anal. Chim. Acta, 346 (1997) 253.
61 M.L. Larrivee and C.F. Poole, Anal. Chem., 66 (1994) 139.
62 K.G. Miller and C.F. Poole, J. High Resolut. Chromatogr., 176 (1994) 125.
63 C.F. Poole, S.K. Poole, D.S. Seibert and C.M. Chapman, J. Chromatogr.B, 689 (1997)
245.

765
64 D. Seibert and C.F. Poole, Chromatographia,41 (1995) 245.
65 C.F. Poole, A.D. Gunatilleka and R. Sethuraman, J. Chromatogr.A, 885 (2000), 17.
66 T. Braumann, G. Weber and L.H. Grimme, J. Chromatogr., 261 (1983) 329.
67 T. Braumann, J. Chromatogr., 373 (1986) 191.
68 G. Machtalere, V. Pichon and M.C. Hennion, J. High Res. Chromatogr., 23 (2000),
437.
69 R.D. McDowall, LC-GC Int., 7 (1994) 638.
70 P. Mussmann, K. Levsen and W. Radeck, FreseniusJ. Anal. Chem., 348 (1994) 654.
71 A.B. Scholten, H.A. Claessens, J.W. de Haan and C.A. Cramers, J. Chromatogr. A,
759 (1997) 37.
72 J. Nawrocki, J. Chromatogr.A, 779 (1997) 29.
73 R.J.M. Vervoort, M.W.J. Derksen and A.J.J. Debets, J. Chromatogr. A, 765 (1997)
157.
74 D. Puig and D. Barcelo, Chromatographia,40 (1995) 435.
75 D. Barcelo, S. Chiron, S. Lacorte, E. Martinez, J.S. Salau and M.-C. Hennion, Trends
Anal. Chem., 13 (1994) 352.
76 L.H. Keith (Ed.), Compilation of EPA's Sampling and Analysis Methods, 2nd edn.
Lewis Publishers, Boca Raton, FL, 1996.
77 C.W. Huck and G.K. Bonn, J. Chromatogr.A, 885 (2000), 51.
78 S. Guenu and M.-C. Hennion, J. Chromatogr.A, 737 (1996) 15.
79 V. Pichon, Analusis, 26 (1998) 91.
80 C. Molina, P. Grasso, E. Benfenati and D. Barcelo, J. Chromatogr.A, 737 (1996) 47.
81 M. Castillo, D. Puig and D. Barcelo, J. Chromatogr.A, 778 (1997) 301.
82 A. Junker-Buchheit and M. Witzenbacher, J. Chromatogr. A, 737 (1996) 67.
83 C. Crespo, R.M. Mare6 and F. Burrull, J. Chromatogr. A, 670 (1994) 135.
84 F. Hernandez, C. Hidalgo, JV. Sancho and F.J. Lopez, Anal. Chem., 70 (1998) 3322.
85 J, Hadgeson, J. Collins and W. Bashe, J. Chromatogr.A, 659 (1994) 395.
86 A.M. Kvistad, E. Lundanes and T. Greibrokk, Chromatographia,48 (1998) 707.
87 C. Aguilar, F. Borrull and R.M. Marc6, J. Chromatogr.A, 771 (1997) 221.
88 T.A. Albanis, D.G. Hela, T.M. Sakellarides and I.K. Konstantinou, J. Chromatogr.A,
823 (1998) 59.
89 I. Tolosa, J.W. Readman and L.D. Mee, J. Chromatogr.A, 725 (1996) 93.
90 M.C. Hennion, C. Cau-Dit-Coumes and V. Pichon, J. Chromatogr. A, 823 (1998) 93.
91 M.-C. Hennion and V. Pichon, Environ. Sci. Technol., 28 (1994) 576 A.
92 M.L. Mayer, S.K. Poole and C.F. Poole, J. Chromatogr.A, 697 (1995) 89.
93 M.-C. Hennion and V. Coquart, J. Chromatogr., 642 (1993) 211.
94 D. Bolliet and C.F. Poole, Chromatographia,46 (1997) 381
95 N.Masque, R.M, Marce and F. Borrull, Trends Anal. Chem., 17 (1998) 384.
96 L. Schmidt, J.J. Sun, J.S. Fritz, D.F. Hagen, C.G. Markell and E.E. Wisted, J.
Chromatogr., 641 (1993) 57
97 T.K. Chambers and J.S. Fritz, J. Chromatogr.A, 797 (1998) 139.
98 N. Masqu6, M. Galia, R.M. Marc6 and F. Borrull, J. Chromatogr.A, 803 (1998) 147.
99 N. Masqu6, R.M. Marc6 and F. Borrull, J. Chromatogr.A, 793 (1998) 257.
100 N. Masqu6, M. Galia, R.M. Marc6 and F. Borrull, J. Chromatogr.A, 771 (1997) 55.
101 E.S.P. Bouvier, P.C. Iraneta, U.D. Neue, P.D. McDonald, D.J. Phillips, M. Capparella
and Y.F. Cheng, LC-GC int., September (1998) 35.
102 M.C. Hennion, J. Chromatogr.A, 885 (2000), 73
103 A. Di Corcia, C. Crescenzi, A. Marcomini and R. Samperi, Environ. Sci. Technol., 32
(1998) 711.
104 M.-C. Hennion, V. Coquart, S. Guenu and C. Sella, J. Chromatogr.A, 712 (1995) 287.
105 A. Di Corcia and M. Marchetti, Environ. Sci. Technol., 26 (1992) 66.
106 A. Di Corcia, S. Marchese and R. Samperi, J. Chromatogr., 642 (1993) 175.

766
107 A. Di Corcia, S. Marchese and R. Samperi, J. Chromatogr., 642 (1993) 163.
108 H. Sabik and R. Jeannot, J. Chromatogr.A, 818 (1998) 197.
109 A. Lagana, G. Fago and A. Marino, J. Chromatogr. A, 796 (1998) 309.
110 C. Crescenzi, A. Di Corcia, R. Samperi and M.A. Marcomini, Anal. Chem., 67 (1995)
1797.
111 A. Di Corcia and M. Marchetti, Anal. Chem., 63 (1991) 580.
112 G. D'Ascenzo, A. Gentili, S. Marchese, A. Marino and D. Perret, Chromatographia,
48 (1998) 497.
113 C. Crescenzi, A. Di Corcia, E. Guerriero and R. Samperi, Environ. Sci. Technol., 31
(1997) 479.
114 M.W.F. Nielen, U.A.Th. Brinkman and R.W. Frei, Anal. Chem., 57 (1985) 806.
115 E.R. Brouwer, J. Slobodnik, H. Lingeman and U.A.Th. Brinkman, Analusis, 20
(1992) 121.
116 F.T. Lange, M. Wenz and H.J. Brauch, J. High Resolut. Chromatogr., 18 (1995) 243.
117 T. Reemtsma, J. Chromatogr.A, 733 (1996) 473.
118 C. Sarzanini, M.C. Bruzzoniti, G. Sacchero and E. Mentasti, J. Chromatogr.A, 739
(1996) 63.
119 R.A. Gimeno, J.L. Beltran, R.M. Marce and F. Borrull, J. Chromatogr. A,., 890
(2000), 289.
120 M.S. Mills, E.M. Thurman and M.J. Pedersen, J. Chromatogr., 629 (1993) 11.
121 I.H. Hagestam and T.C. Pinkerton, Anal. Chem., 57 (1985) 1757.
122 K.K. Unger, Chromatographia,31 (1991) 507.
123 S. Vielhauer, A. Rudolphi, K.S. Boos, D. Seidel, J. Chromatogr.B, 666 (1995) 315.
124 A. Rudolphi, K.S. Boos and D. Seidel, Chromatographia,41 (1995) 645.
125 B.J. Gurley, M. Marx and K. Olsen, J. Chromatogr.B, 670 (1995) 358.
126 Z. Yu and D. Westerlund, J. Chromatogr. A, 725 (1996) 137.
127 Z. Yu and D. Westerlund, J. Chromatogr. A, 725 (1996) 149.
128 Z. Yu, D. Westerlund and K.S. Boos, J. Chromatogr. B, 698 (1997) 379.
129 K.S. Boos and A. Rudolphi, LC-GC Int., 15 (1997) 602.
130 P. Onnerfjord, D. Barcelo, J. Emnens, L. Gorton and G. Marko-Varga, J.
Chromatogr.A, 737 (1996) 35.
131 E. Hogendoorn and P. van Zoonen, J. Chromatogr. A, 892 (2000) 435.
132 A. Marx, T. Giersch and B. Hock, Anal. Lett., 28 (1995) 267.
133 D.H. Thomas, M. Beck-Westermeyer and D.S. Haga, Anal. Chem., 66 (1994) 3823.
134 S.J. Shahtaheri, M.F. Katmeh, P. Kwasowski and D. Stevenson, J. Chromatogr. A,
697 (1995) 131.
135 S.J. Shataheri, P.W. Kwasowski and D. Stevenson, Chromatographia,47 (1998) 453.
136 K.A. Bean and J.D. Henion, J. Chromatogr. A, 791 (1997) 119.
137 E. Matisova and S. Skrabakova, J. Chromatogr.A, 707 (1995) 145.
138 E. Schoenzetter, V. Pichon, D. Thiebaut, A. Fernandez-Alba and M.C. Hennion, J.
Microcolumn Sep., 12 (2000) 316.
139 S. Ben Rejeb, N.F. Durand, A. Martel, B. Le Poullennec, J.F. Lawrence, M.C.
Hennion and F. Le Goffic, Anal. Chim. Acta, 379 (1998) 41.
140 V. Pichon, L. Chen, M.-C. Hennion, R. Daniel, A. Martel, F. Le Goffic, J. Abian and D.
Barcelo, Anal. Chem., 67 (1995) 2451.
141 V. Pichon, L. Chen and M.-C. Hennion, Anal. Chim. Acta, 311 (1995) 429.
142 V. Pichon, L. Chen, N. Durand, F. le Goffic and M.-C. Hennion, J. Chromatogr.A, 725
(1996) 107.
143 V. Pichon, H. Rogniaux, N. Fischer-Durand, S. Ben Rejeb, F. Le Goffic and M.-C.
Hennion, Chromatographia,45 (1997) 289.
144 S. Ouyang, Y. Xu and Y.H. Chen, Anal. Chem., 70 (1998) 931.
145 M. Bouzige, V. Pichon and M.-C. Hennion, J. Chromatogr.A, 823 (1998) 197.

767
146 M. Bouzige, V. Pichon and M.-C. Hennion, Environ. Sci. Technol., 33 (1999) 1916.
147 M. Cichna, D. Knopp, R. Niessner, Anal. Chim. Acta, 339 (1997) 241.
148 A. Martin-Esteban, P. Kwasowski and D. Stevenson, Chromatographia,45 (1997)
364.
149 V. Pichon, M. Bouzige and M.-C. Hennion, Anal. Chim. Acta, 376 (1998) 21.
150 V. Pichon, M. Bouzige, C. Midge and M.C. Hennion, Trends Anal. Chem., 18 (1999)
219.
151 B. Sellergren, Anal. Chem., 66 (1994) 1578.
152 P. Martin, I.D. Wilson, D.E. Morgan and G.R. Jonas and K. Jones, Anal. Commun.,
34 (1997) 45.
153 M.T. Muldoon and L.H. Stanker, Anal. Chem., 69 (1997) 803.
154 J. Steinke, D.C. Sherrington and I.R. Dunkin, Adv. Polym. Sci., 123 (1995) 81.
155 B. Sellergren, Trends Anal. Chem., 18 (1999) 164.
156 D. Stevenson, Trends Anal. Chem., 18 (1999) 154
157 B. Sellergren, Trends Anal. Chem., 16 (1997) 310.
158 N. Masqu6, R.M. Marc and F. Borrull Trends Anal. Chem., 20 (2001) 477.
159 C. Baggiani, F. Trotta, G. Giraudi, C. Giovannoll and A. Vanni, Anal. Commun., 36
(1999) 263.
160 J. Matsui, M. Okada, M. Tsuruoka and T. Takeuchi, Anal. Commun., 34 (1997) 85.
161 J. Matsui, K. Fujiwara, S. Ugata and T. Takeuchi, J. Chromatogr.A, 889 (2000) 25.
162 I. Ferrer, P. Lanza, A. Tolokan, V. Horvath, B. Sellergren, O. Horval and D. Barcelo,
Anal. Chem., 72 (2000) 3934.
163 B. Bjarnason, L. Chimuka and 0. Ramstrom, Anal. Chem., 71 (1999) 2152.
164 I. Ferrer and D. Barcel6, Trends Anal. Chem., 18 (1999) 180.
165 N. Masque, R.M. Marc6, F. Borrull, P.A.G. Cormack and D.C. Sherrington, Anal.
Chem., 72 (2000) 4122.
166 D.T. Rossi and N. Zhang, J. Chromatogr. A, 885 (2000) 97.
167 G.A. Smith and T.L. Lloyd, LC-GC, 5 (1998) S22.
168 F.X. Diamond, W.E. Vickery and J. de Kanel, J. Anal. Toxicol., 20 (1997) 587.
169 E.L. Johnson, K.L. Hoffman and L.A. Pachla, in: J.R. Strimaitis and G. Hawk (Eds.),
Advances in LaboratoryAutomation Robotics. Zymark, Hopington, MA, 1988, p. 111.
170 A.J.H. Louter, J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr.A, 842 (1999) 391.
171 D. Barcelo and M.C. Hennion, Anal. Chim. Acta, 318 (1995) 1.
172 D. Barcelo and M.-C. Hennion, in: DeterminationofPesticides and theirDegradation
Productsin Water. Elsevier, Amsterdam, 1997, pp. 357-428.
173 I. Ferrer and D. Barcelo, J. Chromatogr. A, 778 (1997) 161.
174 J. Slobodnik, A.J.H. Louter, J.J. Vreuls, I. Liska and U.A.Th Brinkman, J. Chroma-
togr. A, 768 (1997) 239.
175 G.R. Mills, J. Chromatogr. A, 813 (1998) 63.
176 C. Hidalgo, J.V.C. Sancho, F.J. Lopez and F. Hernandez, J. Chromatogr. A, 823
(1998) 121.
177 R. El Harrak, M. Calull, R.M. Marce and F. Borrull, J. High Resolut. Chromatogr.,21
(1998) 667.
178 R.J. Vreeken, P. Speksnijder, I. Bobeldijk-Pastorova and Th.H.M. Noij, J. Chroma-
togr. A, 794 (1998) 187.
179 S. Guenu and M.C. Hennion, Anal. Methods Instrum., 2 (1995) 247.
180 D. Puig and D. Barcelo, J. Chromatogr.A, 778 (1997) 313.
181 R.W. Fedeniuk and P.J. Shand, J. Chromatogr.A, 812 (1998) 3.
182 N. Masqu6, E. Pocurull, R.M. Marc6 and F. Borrull, Chromatographia,47 (1998) 176.
183 E. Pocurull, R.M. Marce and F. Borrull, Chromatographia,41 (1995) 521.
184 J.M. Soriano, B. Jimenez, M.J. Redondo and J.C. Molto, J. Chromatogr.A, 822 (1998)
67.

768
185 M. Ibanez, Y. Pico and J. Manes, Chromatographia,45 (1997) 402.
186 S. Dupas, S. Guenu, V. Pichon, A. Montiel, B. Welte and M.C. Hennion, Int. J.
Environ. Anal. Chem., 65 (1996) 53.
187 V. Pichon and M.-C. Hennion, J. Chromatogr.A, 665 (1994) 269.
188 U.A.Th. Brinkman, J. Slobodnik and J.J. Vreuls, Trends Anal. Chem., 13 (1994) 373.
189 S. Lacorte and D. Barcelo, Anal. Chim. Acta, 296 (1994) 223.
190 S. Lacorte and D. Barcelo, J. Chromatogr.A, 725 (1996) 85.
191 S. Lacorte, N. Ehresmann and D. Barcelo, Environ. Sci. Technol., 29 (1995) 2834.
192 R. Reupert, G. Brausen and R. Schuster, Acta Hydrochem. Hydrobiol. 26 (1998) 318.
193 A. Farjam, G.J. de Jong, R.W. Frei, U.A.Th. Brinkman, W. Haasnoot, A.R.M.
Hamers, R. Schilt and F.A. Huf, J. Chromatogr., 452 (1988) 419.
194 A. Farjam, N.C. van de Merbel, H. Lingeman, R.W. Frei and U.A.Th. Brinkman, Int.
J. Environ. Anal. Chem., 45 (1991) 73.
195 I. Ferrer, V. Pichon, M.-C. Hennion and D. Barcelo, J. Chromatogr.A, 777 (1997) 91.
196 I. Ferrer, M.-C. Hennion and D. Barcelo, Anal. Chem., 69 (1997) 4508.
197 J. Henion, E. Brewer and G. Rule, Anal. Chem., 70 (1998) 650A.
198 K.-S. Boos, J. High Resolut. Chromatogr., 23 (2000) 272.
199 C.T. Fleischer, K.-S. Boos, F. Lanza and B. Sellergren, J. High Resolut. Chromatogr.
23 (2000) 284.
200 C.T. Fleischer, P. Chiap and K.-S. Boos, 1s t International Workshop on Molecularly
Imprinted Polymers, Cardiff, 2000.
201 R. Koeber, C.T. Fleischer, K.-S. Boos, D. Barcel6, F. Lanza, B. Sellergren and Mica
network, 1 t International Workshop on Molecularly Imprinted Polymers, Cardiff,
2000.
202 A.J.H. Louter, C.A. van Beckvelt, P. Cid Montanes, J. Seobodnik, J.J. Vreues and
U.A. Th. Brinkman, J. Chromatogr. A, 725 (1996) 67.
203 H.J. Cortes, C.D. Pfeiffer, G.L. Jewett and B.E. Richter, J. Microcol. Sep., 1 (1989)
28.
204 K. Grob Jr. and Z. Li, J. Chromatogr., 473 (1989) 423.
205 K. Grob, J. Chromatogr.A, 473 (1989) 73.
206 E. Noroozian, F.A. Maris, M.W.F. Nielen, R.W. Frei, G.J. de Jong and U.A.Th.
Brinkman, J. High Resolut. Chromatogr. Commun., 11 (1987) 17.
207 Th.H.M. Noij, E. Weiss, T. Herps, H. Van Cruchten and J. Rijks, J. High Resolut.
Chromatogr. Chromatogr. Commun., 11 (1988) 181.
208 J.J. Vreuls, W.J.G.M. Cuppen, E. Dolecka, F.A. Maris, G.J. de Jong and U.A.Th.
Brinkman, J. High Resolut. Chromatogr., 12 (1989) 807.
209 J.J. Vreuls, W.J.G.M. Cupen, G.J. de Jong and U.A.Th. Brinkman, J. High Resolut.
Chromatogr., 13 (1990) 157.
210 J.J. Vreuls, R.T. Ghijsen, G.J. de Jong and U.A.Th. Brinkman, J. Chromatogr., 625
(1992) 237.
211 Y. Pic6, J.J. Vreuls, R.T. Ghijsen and U.A.Th. Brinkman, Chromatographia, 38
(1994)461.
212 Y. Pic6, A.J.H. Louter, J.J. Vreuls and U.A.Th. Brinkman, Analyst 119 (1994) 2025.
213 E. Boselli, B. Grolimund, K. Grob, G. Lercker and R. Amadb, J. High Resolut.
Chromatogr., 21 (1998) 355.
214 B. Grolimund, E. Boselli, K. Grob, R. Amadb and G. Lercker, J. High Resolut.
Chromatogr., 21 (1998) 378.
215 H. Huang, J.R. Kagel and D.T. Rossi, J. Pharm.Biomed. Anal., 19 (1999) 613.
216 D. Barcelo, S. Chiron, S. Lacorte, E. Martinez, J.S. and M.C. Hennion, Trends Anal.
Chem., 13 (1994) 352.
217 S.A. Senseman, T.L. Lavy and J.D. Mattice, Anal. Chem., 67 (1995) 3064.
218 E. Martinez and D. Barcelo, Chromatographia,42 (1996) 72.

769
219 C. Crescenzi, A. Di Corcia, M.D. Madbouly and R. Samperi, Environ. Sci. Technol., 29
(1995) 2185.
220 G.A. Penuela and D. Barcelo, J. Chromatogr. A, 823 (1998) 81.
221 G. Carrera, P. Fernandez, R. Vilanova and J.O. Grimalt, J. Chromatogr. A, 823
(1998) 189.
222 G.A. Penuela and D. Barcelo, J. Chromatogr. A, 795 (1998) 93.
223 S. Lacorte, N. Ehresmann and D. Barcelo, Environ. Sci. Technol., 29 (1995) 2834.
224 E.Pocurull, L. Brossa, F. Borrull and R.M. Marce, J. Chromatogr.A, 885 (2000) 361.
225 J. Slobodnik, M.G.M. Groenewegen, E.R. Brouwer, H. Lingeman and U.A.Th.
Brinkman, J. Chromatogr., 642 (1993) 359.
226 S. Lacorte, J.J. Vreuls, J.S. Salau, F. Ventura and D. Barcelo, J. Chromatogr. A, 795
(1998) 71.
227 J.M. Soriano, B. Jimenez, G. Font and J.C. Molto, Crit. Rev. Anal. Chem., 31 (2001)
19.
228 M.J.M. Wells and L.Z. Yu, J. Chromatogr.A, 885 (2000) 237.
229 Y. Pico, G. Font, J.C. Molto and J. Manes., J. Chromatogr.A, 885 (2000) 251.
230 V. Pichon, J. Chromatogr.A, 885 (2000) 195.
231 R.M. Marce and F. Borrull, J. Chromatogr.A, 885 (2000) 273.
232 I.Rodriguez, M.P. Llompart and R. Cela, J. Chromatogr.A, 885 (2000) 291.
233 P.L. Ferguson, C.R. Iden and B.J. Brownawell, J. Chromatogr.A, 938 (2001), 79.
234 M. Petrovic and D. Barcelo, J. Mass Spectrom., 36 (2001) 1173.
235 S.H. Benomar, M.R. Clench and D.W. Allen, Anal. Chim. Acta, 445 (2001) 255.
236 M. Castillo, G. Penuela and D. Barcelo, Fresenius J. Anal. Chem., 369 (2001) 620.
237 P. de Voogt, O. Kwast, R. Hendriks and N. Jonkers, Analusis, 28 (2000) 776.
238 M. Petrovic and D. Barcelo, FreseniusJ. Anal. Chem., 368 (2000) 676.
239 P. Eichhorn, M.E. Flavier, M.L. Paje and T.P. Knepper, Sci. Total. Environ., 269
(2001) 75.
240 J. Riu and D. Barcelo, Analyst, 126 (2001) 825.
241 J. Riu, E. Martinez, D. Barcelo, A. Ginebreda and L. Tirapu, Fresenius J. Anal.
Chem., 371 (2001) 448.
242 C.A. Moody, W.C. Kwan, J.W. Martin, D.C.G. Muir and S.A. Mabury, Anal. Chem., 73
(2001) 2200.
243 W.H. Ding and Y.H. Liao, Anal. Chem., 73 (2001) 36.
244 I. Ferrer and E.T. Furlong, Environ. Sci. Technol., 35 (2001) 2583.
245 S. Hui-Feng, T. Hase, N. Hata, I. Kasahara and S. Taguchi, Anal. Sci., 17 (2001)
1291.
246 M.J.L. de Alda and D. Barcelo, J. Chromatogr.A, 938 (2001) 145.
247 M. Katayama, T. Sasaki, Y. Matsuda, S. Kaneko, T. Iwamoto and M. Tanaka,
Biomed. Chromatogr., 15 (2001) 403.
248 S.A. Snyder, D.L. Villeneuve, E.M. Snyder and J.P. Giesy, Environ. Sci. Technol., 35
(2001) 3620.
249 H.M. Kuch and K. Ballschmiter, Environ. Sci. Technol., 35 (2001) 3201.
250 X.Y. Xiao, D.V. McCalley and J. McEvoy. J. Chromatogr.A, 923, (2001) 195.
251 A. Lagana, G. Fabo, A. Marino and D. Santarelli, Anal. Lett., 34 (2001) 913.
252 S. Nakamura, T.H. Sian and S. Daishima, J. Chromatogr.A, 919 (2001) 275.
253 M.J.L. de Alda and D. Barcelo, J. Chromatogr. A, 911 (2001) 203.
254 Y. Ishii, S. Okita, M. Torigai and S.J. Yun, Bunseki Kagaku, 49 (2000) 753.
255 M.J.L. de Ala and D. Barcelo, J. Chromatogr.A, 892 (2000) 391.
256 H.M. Kuch and K. Ballschmiter, Fresenius J. Anal. Chem., 366 (2000) 391.
257 A. Lagana, A. Bacaloni, G. Fago and A. Marino, Rapid Commun Mass Spectrom., 14
(2000) 401.
258 S.A. Snyder, T.L. Keith, D.A. Verbrugge, E.M. Snyder, T.S. Gross, K. Kannan and

770
 J.P. Giesy, Environ. Sci. Technol., 33 (1999) 2814.
259 M. Careri, L. Elviri and A. Mangia, J. AOAC Int., 84 (2001) 1383.
260 M.C. Alonso and D. Barcelo, J. Chromatogr.A, 889, (2000) 231.
261 M.C. Alonso and D. Barcelo, Anal. Chim. Acta, 400 (1999) 211.
262 M.C. Alonso, M. Castillo and D. Barcelo, Anal. Chem., 71 (1999) 2586.
263 R. Loos, J. Riu, M.C. Alonso and D. Barcelo, J. Mass Spectrom., 35 (2000) 1197.
264 E. Marengo, M.C. Gennaro and V. Gianotti, Chemometr. Intell. Lab. Syst., 53 (2000)
57.
265 C. Wolf, T. Storm, F.T. Lange, T. Reemtsma, H.J. Brauch, S.H. Eberle and M. Jekel,
Anal. Chem., 72 (2000) 5466.
266 R. Loos, M.C. Alonso and D. Barcelo, J. Chromatogr.A, 890 (2000) 225.
267 R.A. Gimeno, R.M. Marce and F. Borrull, Chromatographia,53 (2001) 22.
268 C.H. Liu and W.H. Ding, J. Chromatogr.A, 926 (2001) 341.
269 C. Creszenzi, A. Di Corcia, A. Marcomini, G. Pojana and R. Samperi, J. Chromatogr.
A, (2001) 97.
270 J. Kruppa, A. Preiss, K. Levsn and H.P. Kalbus, Acta Hydrochim. Hydrobiol., 24
(1996) 226.
271 C. Aguilar, I. Ferrer, F. Borrull, R.M. Marce and D. Barcelo, Anal. Chim. Acta, 386
(1999) 237.
272 A.C. Hogenboom, W.M.A. Niessen and U.A.T. Brinkman, J. Chromatogr. A, 841
(1999) 33.
273 H. Bagheri, J. Slobodnik and U.A.T. Brinkman, Anal. Lett., 33 (2000) 249.
274 R. Castro, E. Moyano and M.T. Garceran, J. Chromatogr. A, 869 (2000) 441.
275 T.C.R. Santos, J.C. Rocha and D. Barcelo, J. Chromatogr.A, 879 (2000), 3.
276 A.C. Hogenboom, W.M.A. Niessen and U.A.T. Brinkman, Rapid Commun Mass.
Spectrom., 14 (2000) 1914.
277 R. Castro, E. Moyano and M.T. Galceran, J. Chromatogr.A, 914 (2001), 111.
278 M.P.G. De Llasera and M. Bernal-Gonzalez, Water Res., 35 (2001) 1933.
279 N. Masque, M. Galia, R.M. Marce and F. Borrull, J. High Res. Chromatogr., 22 (1999)
547.
280 J. Patsias and E. Papadopoulou-Mourkidou, J. AOAC Int., 82 (1999) 968.
281 I. Ferrer and D. Barcelo, J. Chromatogr. A, 854 (1999) 197.
282 R.A. Gimeno, C. Aguilar, R.M. Marce and F. Borrull, J. Chromatogr.A, 915 (2001),
139.
283 E. Pocurull, C. Aguilar, M.C. Alonso, D. Barcelo, F. Borrull and R.M. Marce, J.
Chromatogr. A, 854 (1999) 187.
284 R. Wissiack, E. Rosenberg and M. Grasserbauer, J. Chromatogr.A, 896 (2000) 159.
285 J. Patsias and E. Papadopoulou-Mourkidou, J. Chromatogr.A, 904 (2000), 171.
286 C. Rivasseau, G. Vanhoenacker, P. Sandra and M.C. Hennion, J. Microcolumn Sep.,
12 (2000) 323.
287 P. Hinsmann, L. Arce, A. Rios and M. Valcarcel, J. Chromatogr.A, 866 (2000) 137.
288 A. Medvedovici, V. David and P. Sandra, Rev. Roum Chim., 45 (2000) 827.
289 J. Slobodnik, S. Ramalho, b.M. van Baar, A.J.H. Louter and U.A.T. Brinkman,
Chemosphere, 41 (2000) 1469.
290 E. Pocurull, L. Brossa, F. Borrull and R.M. Marce, J. Chromatogr.A, 885 (2000) 361.
291 M. Peruzzi, G. Bartolucci and F. Cioni, J. Chromatogr. A, 867 (2000) 169.
292 S. Lacorte, I. Guiffard, D. Fraisse and D. Barcelo, Anal. Chem., 72 (2000) 1430.
293 A.C. Hogenboom, M.P. Hofman, D.A. Jolly, W.M.A. Niessen and U.A.T. Brinkman, J.
Chromatogr. A, 885 (2000) 377.
294 E. Turiel, A Martin-Esteban, P. Fernandez, C. Perez-Conde and C. Camara, Anal.
Chem., 73 (2001) 5133.
295 A Martin-Esteban E. Turiel and D. Stevenson, Chromatographia,53 (2001) S 434.

771
296 R. Koeber, C. Fleischer, F. Lanza, K.S. Boos, B. Sellergren and D. Barcelo, Anal.
Chem., 73 (2001) 2437.
297 J. Pawliszyn, Solid-PhaseMicroextraction:Theory and PracticeVCH, New York, NY,
1997.
298 J. Pawliszyn (Ed.), Applications of Solid Phase Microextraction. Royal Society of
Chemistry, Cambridge, 1999.
299 A. Boyd-Boland, M. Chai, L. Luo, Z. Zhang, M. Yang, J. Pawliszyn and T. Gorecki,
Environ. Sci. Technol., 28 (1994) 569A.
300 Z. Zhang and J. Pawliszyn, Anal. Chem., 65 (1993), 1843.
301 C.L. Arthur, L. Killam, K.D. Buchholz, D. Potter, M. Chai, Z. Zhang and J. Pawliszyn,
Environ. Lab., 11 (1992) 10.
302 C.L. Arthur, D. Potter, K. Buchholz, S. Motlagh and J. Pawliszyn, LC-GC, 10 (1992)
656.
303 H. Lord and J. Pawliszyn, J. Chromatogr.A, 885 (2000) 153.
304 H. Prosen and L. Zupancic-Kralj, Trends Anal. Chem., 18 (199) 272.
305 J. Pawliszyn, Trends Anal. Chem., 14 (1995) 113.
306 Z. Zhang, M. Yang and J. Pawliszyn, Anal. Chem., 66 (1994) 844A.
307 T. Gorecki, A. Boyd-Boland, Z. Zhang and J. Pawliszyn, Can. J. Chem., 74 (1996)
1297.
308 B. Zygmunt, A. Jastrzebska and J. Namiesnik, Crit. Rev. Anal. Chem., 31 (2001) 1.
309 R. Eisert and K. Levsen, J. Chromatogr. A, 733 (1996) 143.
310 R. Eisert and J. Pawliszyn, Crit. Rev. Anal. Chem., 27, (1997) 103.
311 Y. Luo and J. Pawliszyn in: A.J. Handley (Ed.), Extraction Methods in Organic
Analysis, Sheffield Academic Press, Sheffield, 1998, Ch. 4, 75.
312 M.D. Alpendurada, J. Chromatogr.A, 889 (2000), 3.
313 N. Prosen and L. Zupancic-Kralig, Trends Anal. Chem., 18 (1999) 272.
314 A. Penalver, E. Pocurull, F. Borrull and R.M. Marce, Trends Anal. Chem., 18 (1999)
557.
315 J. Chen and J. Pawliszyn, Anal. Chem., 67 (1995) 2530.
316 P. Sandra, B. Tienpont, J. Vercammen, A. Tredoux, T. Sandra and F. David, J.
Chromatogr. A, 928 (2001) 117.
317 R. Eisert and J. Pawliszyn, Anal. Chem., 69 (1997) 69.
318 H. Kataoka, S. Narimatsu, H. Lord and J. Pawliszyn, Anal. Chem., 71 (1999) 4237.
319 J. Wu, H.J. Lord, J. Pawliszyn and H. Kataoka, J. Microcolumn Sep., 12 (2000), 255.
320 J. Wu, X.M. Yu, H.L. Lord and J. Pawliszyn, Analyst, 125 (2000) 391.
321 J.C. Wu, H. Lord and J. Pawliszyn, Talanta, 54 (2001) 655.
322 W.M. Mullett, P. Martin and J. Pawliszyn, Anal. Chem., 73 (2001) 2383.
323 Y. Saito, Y. Nakao, M. Imaizumi, T. Takeichi, Y. Kiso and K. Jinno, Fresenius J.
Anal. Chem., 368 (2000) 641.
324 S.N. Semenov, J.A. Koziel and J. Pawliszyn, J. Chromatogr.A, 873 (2000) 39.
325 J. Ai, Anal. Chem., 69 (1997) 3260.
326 J. Ai, Anal. Chem., 70 (1998) 4822.
327 E. Belau, C. Grote, M. Spiekermann and K. Levsen, Field Anal. Chem. Technol., 5
(2001) 37.
328 Z. Zhang and J. Pawliszyn, J. High Resolut. Chromatogr.16 (1993) 689.
329 B. MacGillivray, J. Pawliszyn, P. Fowlie and C. Sagara, J. Chromatogr. Sci., 32
(1994) 317.
330 Z. Zhang and J. Pawliszyn, Anal. Chem., 67 (1995) 34.
331 T. Gorecki and J. Pawliszyn, Analyst, 122 (1997) 1079.
332 T. Gorecki, A. Khaled and J. Pawliszyn, Analyst, 123 (1999) 2819.
333 D. Louch, S. Motlagh and J. Pawliszyn, Anal. Chem., 64 (1992) 1187.
334 S. Motlagh and J. Pawliszyn, Anal. Chim. Acta, 284 (1993) 265.

772
335 Z. Penton, H. Geppert and V. Betz, GIT Spez. Chromatogr., 2 (1996) 112.
336 R. Eisert and J. Pawliszyn, J. Chromatogr.A, 776 (1997) 293.
337 Z.Y. Zhang and J. Pawliszyn, Anal. Chem., 65 (1993), 1843.
338 B.D. Page and G. Lacroix, J. Chromatogr.A, 757 (1997), 173.
339 V. Mani in: J. Pawliszyn (Ed.), Applicationof Solid Phase Extraction. Royal Society of
Chemistry, Cambridge, Ch. 5, 57.
340 S.L. Chong, D. Wang, J.D. Hayes, B.W. Wilhite and A. Malik, Anal. Chem., 69 (1997)
3889.
341 F. Mangani and R. Cenciarini. Chromatographia,41 (1995) 678.
342 Z.Y. Wang, C.H. Xiao, C.Y. Wu and H.M. Han, J. Chromatogr. A, 893 (2000) 157.
343 R. Eisert and K. Levsen, J. Am. Soc. Mass Spectrom., 6 (1995), 1119.
344 P. Popp, S. Mothes and L. Briiggemann, Vomn Wasser, 85 (1995) 229.
345 C. Grote, E. Belau, K. Levsen and G. Wiinsch, Acta Hydrochim. Hydrobiol., 27 (1999)
193.
346 C. Grote and K. Levsen in: J. Pawliszyn (Ed.), Application of Solid Phase Extraction.
Royal Society of Chemistry, Cambridge, 1999, Ch. 12, 1.
347 B. Schafer, P. Hennig and W. Engewald, J. High Resolut. Chromatogr., 20 (1997),
217.
348 P.A. Martos, A. Saraullo and J. Pawliszyn, Anal. Chem., 69 (1997) 402.
349 B. Shurmer and J. Pawliszyn, Anal. Chem., 72 (2000) 3660.
350 L. Pan and J. Pawliszyn, Anal. Chem., 69 (1997) 196.
351 P.A. Martos and J. Pawliszyn, ACS Symp. Ser., 705 (1998), 92.
352 L. Pan, M. Adams and J. Pawliszyn, Anal. Chem., 67 (1995) 4396.
353 P. Bartak and L. Cap, J. Chromatogr.A, 767 (1997) 171.
354 K.D. Buchholz and J. Pawliszyn, Enuiron. Sci. Technol., 27 (1993) 2844.
355 L. Pan, J.M. Chong and J. Pawliszyn, J. Chromatogr.A, 773 (1997) 249.
356 R. Eisert and K. Levsen, GIT Fachz. Lab., 40 (1996) 581.
357 Y. Cai and J.M. Bayona, J. Chromatogr.A, 696 (1995) 113.
358 T. Gorecki and J. Pawliszyn, Anal. Chem., 68 (1996), 3008.
359 Y. Morcillo, Y. Cai and J.M. Bayona, J. High Resolut. Chromatogr., 18 (1995), 767.
360 M. Guidotti and M. Vitali, Ann. Chim., 87 (1997) 497.
361 P. Popp and A. Paschke, Chromatographia,46 (1997) 419.
362 M. Mder, S. Schrader, U. Franck and P. Popp, Fresenius J. Anal. Chem., 357 (1997)
326.
363 B. Schafer and W. Engewald, Fresenius'J. Anal. Chem., 352 (1995) 535.
364 L. Miller, E. Fattore and E. Benfenati, J. Chromatogr.A, 791 (1997) 221.
365 K. D. Buchholz and J. Pawliszyn, Anal. Chem., 66 (1994) 160.
366 A.A. Boyd-Boland, S. Magdic and J. Pawliszyn, Analyst, 121 (1996) 929.
367 J. Dewulf, H. van Langenhove and M. Everaert, J. Chromatogr.A, 761 (1997) 205.
368 C.L. Arthur, L.M. Killam, K.D. Buchholz, J. Pawliszyn and J.R. Berg, Anal. Chem.,
64 (1992) 1960.
369 R. Eisert and K. Levsen, G. Wiinsch, Vom Wasser, 86 (1996) 1.
370 L. Urruty and M. Montury, J. Agric. Food Chem., 44 (1996) 3871.
371 K.K. Chee, M.K. Wong and H.K. Lee, J. Microcolumn Sep., 8 (1996) 131.
372 S.D. Huang, C.P. Cheng and Y.H. Sung, Anal. Chim. Acta, 343 (1997) 101.
373 C. Grote, K. Levsen and G. Winsch, Anal. Chem., 71 (1999) 4513.
374 J. Beltran, FJ Lopez and F. Hernandez, J. Chromatogr.A, 885 (2000) 389.
375 R. Hu, D. Elia, JM Berthion and S. Poliak, Chromatographia,53 (2001) 306.
376 D. Lambropoulou, T. Sakellarides and T. Albanis, Fresenius J. Anal. Chem., 368
(2000) 616.
377 M.C Sampedro, O. Martin, C.Lopez de Armentia, M.A. Goicolea, E. Rodriguez, Z.
Gomez de Balugera, J. Costa-Moreira and R.J. Barrio, J. Chromatogr. A, 893 (2000)

773
 347.
378 J. Lipinski, FreseniusJ. Anal. Chem., 367 (2000) 445.
379 D.A. Lambropoulou, I.K Konstantinou and T.A. Albanis, J. Chromatogr. A, 893
(2000) 143.
380 A.Penalver, E. Pocurull and R.M. Marce, J. Chromatogr.A, 839 (1999) 253.
381 R. Eisert and K. Levsen, Fresenius J. Anal. Chem., 351 (1995) 555.
382 I. J. Barnabas, J.R. Dean, I.A. Fowlis and S.P. Owen, J. Chromatogr. A, 705 (1995)
305.
383 A.A. Boyd-Boland and J. Pawliszyn, J. Chromatogr.A, 704 (1995) 163.
384 S. Magdic, A.A. Boyd-Boland, K. Jinno and J. Pawliszyn, J. Chromatogr.A, 736
(1996) 219.
385 R. Eisert, K. Levsen and G. Wiinsch, J. Chromatogr. A, 683 (1994) 175.
386 T. Gorecki, R. Mindrup and J. Pawliszyn, Analyst, 121 (1996) 1381.
387 T.K. Choudhury, K.O. Gerhardt and T.P. Mawhinney, Environ. Sci. Technol., 30
(1996) 3259.
388 R. Young, V. Lopez-Avila and W.F. Beckert, J. High Resolut. Chromatogr., 19 (1996)
247.
389 K.N. Graham, L.P. Sarna, G.R.B. Webster, J.D. Gaynor, and H.Y.F. Ng, J. Chroma-
togr. A, 725 (1996) 129.
390 M.T. Sng, F.K. Lee and H.A. Lakso, J. Chromatogr. A, 759 (1997) 225.
391 I. Valor, J.C. Molto, D. Apraiz and G. Font, J. Chromatogr.A, 767 (1997) 195.
392 V. Lopez-Avila, R. Young and W.F. Beckert, J. High Resolut. Chromatogr., 20 (1997)
487.
393 R. Ferrari, T. Nilsson, R. Arena, P. Arlati, G. Bartolucci, R. Basla, F. Cioni, G. Del
Carlo, P. Dellavedova, E. Fattore, M. Funi, C. Grote, M. Guidotti, S. Morgillo, L.
Miller and M. Volante, J. Chromatogr. A, 795 (1998) 371.
394 J. Dugay, C. Miege and M.C. Hennion, J. Chromatogr.A, 795 (1998) 27.
395 C. Aguilar, S. Penalver, E. Pocurull, F. Borrull and R.M. Marce, J. Chromatogr.A,
795 (1998) 105.
396 J. Beltran, F.J. Lopez, O. Cepria and F. Hernandez, J. Chromatogr.A, 808 (1998)
257.
397 G.P. Jackson and A.R.J. Andrews, Analyst, 123 (1998) 1085.
398 C. Miege and J. Dugay, Analusis, 26 (1998) M137.
399 A. Wenner and M. Wortberg, J. High Resolut. Chromatogr., 21 (1998) 661.
400 B.V Stolyarov and V.V Boiko, Russian J. Appl. Chem., 73 (2000) 1963.
401 T. Nilsson, D. Baglio, I. Galdo-Miguez, J.O. Madsen and S. Facchetti. J. Chromatogr.
A, 826 (1998) 211.
402 M.R. Lee, R.-J. Lee, Y.W. Lin, C.M. Chen and B.H. Hwang, Anal. Chem., 70 (1998)
1963.
403 T.Nilsson, D. Baglio, I. Galdo-Miguez, J.O. Madsen and S. Facchetti, J. Chromatogr.
A, 826 (1998) 211.
372 D.C Stahl and D.C Tilotta, Environ. Sci. Technol., 35 (2001) 3507.
404 S.H. Salleh, Y. Saito, Y. Kiso and K. Jinno, Anal. Chim. Acta, 433 (2001) 207.
405 K. Jinno, T. Muramatsu, Y. Saito, Y. Kiso, S. Magdic and J. Pawliszyn, J. Chroma-
togr. A, 754 (1996) 137.
406 M.R. Lee, Y.C. Yeh, W.S. Hsiang and B.H. Hwang, J. Chromatogr.A, 806 (1998) 317.
407 J. Porschmann, F.D. Kopinke and J. Pawliszyn, J. Chromatogr.A, 816 (1998) 159.
408 M. Guidotti and G. Ravaioli, Ann. Chim. (Rome), 88 (1998) 629.
409 R.A. Doong, S.M. Chang and Y.C. Sun, J. Chromatogr. Sci., 38 (2000) 528.
410 D.W. Potter and J. Pawiiszyn, Environ. Sci. Technol., 28 (1994) 298.
411 Y. Liu, M.L. Lee, K.J. Hageman, Y. Yang and S.B. Hawthorne, Anal. Chem., 69 (1997)
5001.

774
412 M.R. Negrao and M.F. Alpendurada, J. Chromatogr. A, 823 (1998) 211
413 A. Paschke, P. Popp and G. Schiiiirmann, FreseniusJ. Anal. Chem., 363 (1999) 426.
414 D.W. Potter and J. Pawliszyn, J. Chromatogr., 625 (1992) 247.
415 J.J. Langenfeld, S.B. Hawthorne and D.J. Miller, Anal. Chem., 68 (1996) 144.
416 C.L. Arthur, L.M. Killam, S. Motlagh, M. Lim, D. Potter and J. Pawliszyn, J.
Environ. Sci. Technol., 26 (1992) 979.
417 I. Valor, C. Cortada and J.C. Molto, J. High Resolut. Chromatogr., 19 (1996) 472.
418 S.P. Thomas, R. Sri Ranjan, G.R.B. Webster and L.P. Sarna, Environ. Sci. Technol.,
30 (1996) 1521.
419 M. Eriksson, A. Swartling and G. Dalhammar, Appl. Microbiol. Biotechnol., 50 (1998)
129.
420 A.A.M. Hassan, E. Benfenati, G. Facchini and R. Fanelli, Toxicol. Environ. Chem., 55
(1996) 73.
421 B. Christall, UFZ-Ber (1997) 12. Optimierung Umweltvertriglicher Analysen-
verfahren fr Mineralolkohlenwasserftoffe im Boden, 14.
422 E. Javorszky, E. Molnar and K. Torkos and J. Borossay, Chromatographia,51 (2000)
328.
423 J. Ritter, V.K. Stromquist, H.T. Mayfield, M.V. Henley and B.K. Lavine, Microchem.
J., 54 (1996) 59.
424 D. Djozan and Y. Assadi, Chromatographia,45 (1997) 183.
425 M. Llompart, K. Li and M. Fingas, J. Chromatogr.A, 824 (1998) 53.
426 D.C. Stahl and D.C. Tilotta, Environ. Sci. Technol., 33 (1999) 814.
427 F.J. Santos, M.T. Galceran and D. Fraisse, J. Chromatogr. A, 742 (1996) 181.
428 J. Czerwinski, B. Zygmunt and J. Namiesnik, FreseniusEnviron. Bull., 5 (1996) 55.
429 M.L. Chen, I.F. Mao and C.C. Hsu, Toxicol. Environ. Chem., 60 (1997) 39.
430 R. Mindrup, R. Shirey and V. Mani, Proc. Water Qual. Technol. Conf. 1995 (PT. 1)
(1996) 57.
431 T. Nilsson, L. Montanarella, D. Baglio, R. Tilio, G. Bidoglio and S. Facchetti, Int. J.
Environ. Anal. Chem., 69 (1998) 217.
432 M. Chai, C.L. Arthur, J. Pawliszyn, R.P. Belardi and K.F. Pratt, Analyst, 118 (1993)
1501
433 T. Nilsson, F. Pelusio, L. Montanarella, B. Larsen, S. Facchetti and J.O. Madsen, J.
High Resolut. Chromatogr., 18 (1995) 617.
434 K.J. James and M.A. Stack, FreseniusJ. Anal. Chem., 358 (1997) 833.
435 T. Nilsson, R. Ferrari and S. Facchetti, Anal. Chim. Acta, 356 (1997) 113.
436 D.L. Heglund and D.C. Tilotta, Environ. Sci. Technol., 30 (1996) 1212.
437 D.A. Cassada, Y. Zhang, D.D. Snow and R.F Spalding, Anal. Chem., 72 (2000) 4654.
438 E. Matisova, J. Sedlakova, M. Slezackova and T. Welsch. J. High Resolut.
Chromatogr., (1999) 109.
439 H. van Doorn, C.B. Grabanski, D.J. Miller and S.B. Hawthorne. J. Chromatogr.A,
829 (1999) 223.
440 L. Miller, E. Fattore and E. Benfenati, J. Chromatogr.A, 791 (1997) 221.
441 S-A. Barshick and W.H. Gries, Anal. Chem., 70 (1998) 3015.
442 D.C Stahl and D.C Tilotta, Environ. Sci. Technol., 35 (2001) 3507.
443 S.W. Lloyd, J.M. Lea, P.V. Zinba and C.C. Grimm, Water Res., 32 (1998) 2140.
444 R. McCallum, P. Pendleton, R. Schumann and M.-U. Trinh, Analyst, 123 (1998)
2155.
445 S. Tutschku, S. Mothes and R. Wennrich, FreseniusJ. Anal. Chem., 354 (1996) 587.
446 Y. Cai, S. Monsalud, K.G. Furton, R. Jaffe and R.D. Jones, Appl. Organomet. Chem.,
12 (1998) 565.
447 F. Yang and Y.K. Chau, Analyst, 124 (1999) 71.
448 T. De Smaele, L. Moens, P. Sandra and R. Dams, Microchim. Acta, 30 (1999) 241.

775
449 Y. Cai and J.M. Bayona, J. Chromatogr.A, 696 (1995) 113.
450 S. Mothes and R. Wennrich, J. High Resolut. Chromatogr., 22 (1999) 181
451 J. Wu, Z. Mester and J. Pawliszyn, J. Anal. Atomic Spectrom., 16 (2001) 159.
452 M. Guidotti, J. AOAC Int., 83 (2000) 1082.
453 C. Jia, Y. Luo and J. Pawliszyn, J. Microcolumn Sep., 10 (1998) 167.
454 M. Llompart, K. Li and M. Fingas, Anal. Chem., 70 (1998) 2510.
455 A. Paschke, P. Popp and G. Schueuermann, FreseniusJ. Anal. Chem., 360 (1998) 52.
456 Z. Takats and K. Torkos., Chromatographia,48 (1999) 74.
457 R. Aranda and R.C. Burk, J. Chromatogr.A, 829 (1998) 401.
458 A.A. Boyd-Boland, J.B. Pawliszyn, Anal. Chem., 68 (1996) 1521.
459 M-L. Bao, F. Pantani, O. Griffini, D. Burrini, D. Santianni and K. Barbieri, J.
Chromatogr.A, 809 (1998) 75.
460 P.A. Martos and J. Pawliszyn, Anal. Chem., 70 (1998) 2311.
461 Y.C. Wu and S.D. Huang, J. Chromatogr.A, 835 (1999) 127.
462 S-D. Huang, C-P. Cheng and Y.H. Sung, Anal. Chim. Acta, 343 (1997) 101.
463 M. Winkler, J.V. Headley and K.M. Peru, J. Chromatogr., (2000) 203.
464 K.K. Chee, M.K. Wong and H.K. Lee, J. Microcolumn Sep., 8 (1996) 131.
465 E. Benfenati, P. Pierucci, R. Fanelli, A. Preiss, M. Godejohann, M. Astratov, K.
Levsen and D. Barcelo, J. Chromatogr. A, 831 (1999) 243.
466 B. Aikawa and R.C. Burk, Int. J. Environ. Anal. Chem., 66 (1997) 215.
467 S-D. Huang, C. Yi ting and C-S. Lin, J. Chromatogr.A, 769 (1997) 239
468 R. Battle, C. Sanchez and C. Nerin, J. AOAC Int., 84 (2001) 431.
469 H.R. Rogers and S.D.W. Comber, Chemosphere, 37 (1998) 1413.
470 N.Huppert, M. Wuertele and H.H. Hahn, FreseniusJ. Anal. Chem., 362 (1998) 529.
471 H. Lakso, and WF. Ng. Anal. Chem., 69 (1997) 1866.
472 M. Moder, P. Popp and J. Pawliszyn, J. Microcolumn Sep., 10 (1998) 225.
473 P. Popp, C. Bauer, M. Moder and A. Paschke, J. Chromatrogr.A, 897 (2000) 153.
474 P. Popp, C. Bauer and L. Wennrich, Anal. Chim. Acta, 436 (2001) 1.
475 M. Moder, S. Schrader, M. Winkler and P. Popp, J. Chromatogr.A, 873 (2000) 95.
476 K. Luks-Betlej, P. Popp, B. Janoszka and H. Paschke, J. Chromatogr.A, 938 (2001) 93.
477 L. Wennrich, W. Engewald and P. Popp, Acta Hydrochim. Hydrobiol., 25 (1997) 329.
478 D.A. Volmer and J.P.M. Hui, Arch. Environ. Contam. Toxicol., 35 (1998) 1.
479 M.H.V. Mulder, BasisPrinciplesof Membrane Technology. Kluwer, Dordrecht, 1991.
480 N.C. van de Merbel, J. Chromatogr.A, 856 (1999) 55.
481 N.C. van de Merbel, F.M. Lagerwerf, H. Lingeman and U.A.Th. Brinkman, Int. J.
Environ. Anal. Chem., 54 (1994) 105.
482 J.A. J6nsson and L. Mathiasson, J. Chromatogr. A, 902 (2000) 205.
483 J.A. Jonsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 318.
484 J.A. Jonsson and L. Mathiasson, Trends Anal. Chem., 11 (1992) 106.
485 S. Palmarsdottir, E. Thordarson, L.-E. Edholm, J.A. Jonsson and L. Mathiasson,
Anal. Chemr., 69 (1997) 1732.
487 J.A. Jonsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 325.
488 P. Dzygiel, P. Wieczarek, L. Mathiasson and J.A. Jonsson, Anal. Lett., 31 (1998) 1261.
489 NT.Megersa, T. Solomon and J.A. Jdnsson, J. Chromatogr.A, 830 (1999) 203.
490 K.F. Pratt and J. Parliszyn, Anal. Chem., 64 (1992) 2101.
491 M. Yang, S. Harms, Y. Lou and J. Pawliszyn, Anal. Chem., 66, (1994) 1339.
492 M. Yang, M. Adams and J. Pawliszyn, Anal. Chem., 68 (1996) 2782.
493 Y. Luo, M. Adams and J. Pawliszyn, Anal. Chem., 70 (1998) 19.
494 A. Segal, T. Gorecki, P. Mussche, J. Lips and J. Pawliszyn, J. Chromatogr., 873A
(2000) 13.
495 Y. Luo and J. Pawliszyn, Anal. Chem., 72 (2000) 1058.
496 M.E. Bier and R.G. Cooks, Anal. Chem., 59 (1987) 597.

776
497 T. Kotiaho, J. Mass Spectrom., 31 (1996) 1.
498 F.R. Lauritsen, T. Kotiaho, T.K. Choudhury and R.G. Cooks, Anal. Chem., 64 (1992)
1205.
499 S. Bauer, Trends Anal. Chem., 14 (1995) 202.
500 N. Srinivasan, R.C. Johnson, N. Kasthurikrishnan, P.W. Wong and R.G. Cooks,
Anal. Chim. Acta, 350 (1997) 257..
501 R.C. Johnson, R.G. Cooks, T.M. Allen, M.E. Cisper and P.H. Hemberger, Mass Spec.
Rev., 19 (2000) 1.
502 T. Kotiaho, R.A. Ketola, M. Ojala, T. Mansikka and R. Kostiainen, Am. Env. Lab.,
3/97 (1997) 19.
503 T. Kotiaho, F.R. Lauritsen, T.K. Choudhury, R.G. Cooks and G.T. Tsao,Anal. Chem.,
63 (1991) 875A.
504 M. Ojala, R.A. Ketola and T. Kotiaho, LC-GC, 16 (1998) 1026.
505 T. Kotiaho, R. Kostiainen, R.A. Ketola, M. Ojala, I. Mattila and T. Mansikka, in: E.J.
Karjalainen, A.E. Hesso, J.E. Jalonen and U.P. Karjalainen (Eds.), Advances in Mass
Spectrometry, Vol. 14. Elsevier, 1998, pp. 501.
506 F.R. Lauritsen and T. Kotiaho, Rev. Anal. Chem., 15 (1996) 237.
507 J.D. Petty, C.E. Orazio, J.N. Huckins,R.W. Gale, J.A. Lebo, K.R. Echols and W.L.
Cranor, J. Chromatogr. A, 879 (2000) 83.
508 T. Colborn, F.S. vom Sall and A.M. Soto, Environ. Health Perspect., 101 (1993) 378
509 J.N. Huckins, G.k. Manuweera, J.D. Petty, D. MacKay and J.A. Lebo, Environ. Sci.
Technol., 27 (1993) 2489.
510 J.D. Petty, J.N. Huckins, C.E. Orazio, J.A. Lebo, B.C. Poulton and R.W. Gale,
Environ. Sci. Technol., 29 (1995) 2561.
511 J.A. Lebo, R.W. Gale, J.D. Petty, D.E. Tillitt, J.N. Huckins and J.C. Meadows,
Enuiron. Sci. Technol., 29 (1995) 2886.
512 J.D. Petty, J.N. Huckins and J.L. Zajicek, Chemosphere, 27 (1993) 1609.
513 J.A. Lebo, J.L. Zajicek, C.E. Orazio, J.D. Petty and J.N. Huckins, Polycyclic Arom.
Comp., 8 (1996) 53.
514 J.D. Petty, B.C. Poulton, C.S. Charbonneau, J.N. Huckins, S.B. Jones and J.T.
Cameron, Environ. Sci. Technol., 32 (1998) 837.
515 P.S.H. Wong, R.G. Cooks, M.E. Cisper and P.H. Hemberger, Environ. Sci. Technol.,
29 (1995) 215A.
516 V.T. Virkki, R.A. Ketola, M. Ojala, T. Kotiaho, V. Komppa, A. Grove and S. Facchetti,
Anal. Chem., 67 (1995) 1421.
517 B.J. Harland and P.J. Nicholson, Sci. Total Environ., 135 (1993) 37.
518 T. Kotiaho, R. Kostiainen, R.A. Ketola, T. Mansikka, I. Mattila, V. Komppa, T.
Hankanen, K. Wickstrom, J. Waldvogel and 0. PilviS, Proc. Contr. Qual. 11 (1998) 71.
519 G. Matz, W. SchrBder, A. Harder, A. Schillings and P. Rechenbach, Field. Anal.
Chem. Technol., 1 (1997) 181.
520 G. Matz and G. Kibelka, in: Proceedingsof PITTCON '98, New Orleans, Lousiana,
March 1-5, 1998, 1864P.
521 B. Hauser and P. Popp, J. Chromatogr.A, 909 (2001) 3.
522 M. Soni, S. Bauer, J.W. Amy, P. Wong and R.G. Cooks, Anal. Chem., 67 (1995) 1409.
523 R.A. Ketola, V.T. Virkki, M. Ojala, V. Komppa and T. Kotiaho, Talanta, 44 (1997)
373.
524 M.A. LaPack, J.C. Tou and C.G. Enke, Anal. Chem., 62 (1990) 1265.
525 L.E. Slivon, M.R. Bauer, J.S. Ho and W.L. Budde, Anal. Chem., 63 (1991) 1335.
526 R. Kostiainen, T Kotiaho, I. Mattila, T. Mansikka and R.A. Ketola, Anal. Chem., 59
(1987) 597.
527 R.M. Alberici, R. Sparrapan, W.F. Jardim and M.N. Eberlin, Environ. Sci. Technol.,
35 (2001) 2084.

777
528 R.A. Ketola and F.R. Lauritsen, Rapid Commun. Mass Spectrom., 13 (1999) 749.
529 F.R. Lauritsen and J. Rose, Analyst, 125 (2000) 1577.
530 F.R. Lauritsen, M.A. Mendes and T. Aggerholm, Analyst, 125 (2000) 211.
531 T. Aggerholm and F.R. Lauritsen, Rapid Commun. Mass Spectrom., 15 (2001) 1826.
532 F.R. Lauritsen and R.A. Ketola, Anal. Chem., 69 (1997) 4917.
533 M.A. Mendes and M.N. Eberlin, Analyst, 125 (2000) 21.
534 G. Matz, M. Loogk and F. Lennemann, J. Chromatogr.A, 819 (1998) 51.
535 G. Matz, G. Kibelka, J. Dahl and F. Lennemann, J. Chromatogr. A, 830 (1999)
365-376.
536 J. Trocewicz, J. Chromatogr.A, 725 (1996) 121.
537 L. Chimuka, M.M. Nindi and J.A. J6nsson, Int. J. Environ. Anal. Chem., 68 (1997), 429.
538 N. Megersa and J.A. J6nsson, Analyst, 123 (1998) 225.
539 N. Megersa, T. Solomon and J.A. Jonsson, J. Chromatogr.A,., 830 (1999) 203.
540 N. Megersa, T. Solomon, B.S. Chandravanshi and J.A. Jnsson, Bull. Chem. Soc.
Ethiopia, 14 (2000) 9.
541 N. Megersa, L. Chimuka, T. Solomon and J.A. J6nsson, J. Sep. Sci., 24 (2001)567.
542 G. Nilv6, G. Audunsson and J.A. Jonsson, J. Chromatogr., 471 (1989) 151.
543 L Mathiasson, G. Nilv6 and B. U16n, Int. J. Environ. Anal. Chem., 45 (1991) 117.
544 M. Knutsson, G. Nilv6, L. Mathiasson and J.A. J5nsson, J. Agric. Food Chem., 40
(1992) 2413.
545 G. Nilve and R. Stebbins, Chromatographia,32 (1991) 269.
546 G. Nilve, M. Knutsson and J.A. Jonsson, J. Chromatogr.A, 668 (1994) 75.
547 M. Knutsson, L. Mathiasson and J.A. Jonsson, Chromatographia,42 (1996) 165.
548 T. Miliotis, M. Knutsson, J.A.Jonsson and L.Mathiasson, Int. J. Environ. Anal.
Chem., 64 (1996) 35.
549 J. Norberg, A. Zander and J.A. Jonsson, Chromatographia,46 (1997) 483
550 R. Carabias Martinez, E. Rodriguez Gonzalo, M.P. Santiago Toribio and J.
Hernandez Mendez, Anal. Chim. Acta, 321 (1996) 147.
551 K. Ndung'u and L. Mathiasson, Anal. Chim. Acta, 404 (2000) 319.
552 J.Norberg, E. Thordarson, L. Mathiasson and J.A. Jonsson, J. Chromatogr.A, 869
(2000) 523.
553 M. Sandahl, L. Mathiasson and J.A. Jonsson, J. Chromatogr. A, 893 (2000) 123.
554 M. Sandahl, E. Ulfsson and L. Mathiasson, Anal. Chim. Acta, 424 (2000) 1.
555 R.G. Melcher and P.L. Morabito, Anal. Chem., 62 (1990) 2183
556 P.L. Morabito and R.G. Melcher, Proc. Contr. Qual., 3 (1992) 35.
557 X. Guo and S. Mitra, J. Chromatogr.A, 904 (2000) 189.
558 K. Grob, On-line coupled LC-GC. Hithig, Heidelberg, 1991.
559 K. Grob, J. Chromatogr. A, 703 (1995) 265.
560 M.-L. Riekkola, J. Chromatogr. A, 473 (1989) 315.
561 I.L. Davies, K.E. Markides, M.L. Lee, M.W. Raynor and K.D. Bartle, J. High Resolut.
Chromatogr., 12 (1989) 193.
562 E. Dijkman, D. Mooibroek, R. Hoogerbrugge, E. Hogendoorn, J.V. Sancho, O. Pozo
and F. Hernandez, J. Chromatogr. A, 926 (2001) 113.
563 F. Hernandez, C. Hidalgo, S. Grimalt and J.V. Sancho, Quim. Anal., 20 (2001) 81.
564 E.A. Hogendoorn, K. Westhuis, E. Dijkman, E.A.G. Heusinkveld, P. Chamraskul, P.
Biadul, R.A. Baumann, A.A. Cornelese and M.A. van der Linden, Int. J. Environ.
Anal. Chem., 78 (2000) 67.
565 J.L.M. Vidal, P.P. Vazquez and J.M. Fernandez, Chromatographia,51 (2000) 187.
566 A. Motoyama, A. Suzuki, O. Shirota and R. Namba, Rapid Commun. Mass Spectrom.,
13 (1999) 2204.
567 L.N. Konda, M.B. Barroso, G. Morovjan and P. Csokan, J. Chromatogr. Sci., 37
(1999), 71.

778
Chapter23

Sampling and sample preparation for

clinical and pharmaceutical analysis
Hiroyuki Kataoka and Heather L. Lord

23.1 INTRODUCTION

23.1.1 Objectives and strategies for sampling and sample

preparation

Analysis of drugs in biological samples and pharmaceutical products is growing

in importance owing to the need to understand therapeutic and toxic effects of
drugs and the continuing development of more selective and effective drugs
[1,2]. A knowledge of drug levels in body fluids such as serum and urine, allows
the optimization of pharmacotherapy and provides the basis for studies on
patient compliance, bioavailability, pharmacokinetics and genetics, organ func-
tion and the influences of co-medication. The quantitative and qualitative
analysis of drugs and metabolites is extensively applied in pharmacokinetic
studies. Variables such as time to maximal concentration in plasma, clearance
and bioavailability have to be known for the approval of a new drug [3,4]. For
example, pharmacokinetic interactions, the pharmacokinetics in special popula-
tions and relationships between the concentration of drug and pharmacological
effect, are investigated in post-marketing surveillance. In addition, therapeutic
drug monitoring (TDM) is used as a tool for the improvement of drug therapy
[5-8]. Drugs of abuse, illicit drugs and intoxications by drugs and poisons are
analyzed in clinical and forensic toxicology [1,9-16]. The screening and confir-
mation of abused drugs in body fluids is also important for the detection and
treatment of users of illicit drugs and the control of drug addicts following
withdrawal therapy.
Drug analyses have been carried out using various analytical instruments in
many circumstances such as clinical control for diagnosis and treatment of dis-
eases, doping control, forensic analysis and toxicology, depending on sample ori-
gins and analytical objectives. However, despite the advances in the development
of highly efficient analytical instrumentation for the end point determination of
analytes in biological samples and pharmaceutical products, a sample pretreat-
ment is usually necessary in order to extract and isolate the analytes of interest

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
from complex matrixes. In general, the analytical process is divided into five
steps-sampling, sample preparation, separation, detection and data analysis
-and over 80% of analysis time is spent on the sampling and sample preparation
steps [17]. Furthermore, the quality of these steps is a key factor in determining
the success of analysis from complex matrices such as biological samples.
The sampling step includes a decision on where to get samples that properly
define the object being analyzed, and choosing a method to obtain samples in the
right amounts. Sampling for subsequent trace analysis is doubtless by far the
most crucial step in an analytical procedure. If not properly planned and
practically performed by using appropriate sampling tools with the utmost care
and expertise [18,19], the total errors for sampling can range from a small
percentage to several orders of magnitude. Depending on the analytical task and
the material to be collected, possible errors stem from many sources. The main
objective in any sampling strategy is to obtain a representative portion of the
sample. This requires a detailed plan of how to carry out the sampling. Therefore
the planning of the sampling strategy is an important part of the overall
analytical procedure as the consequences of a poorly defined sampling strategy,
as well as costing both time and money, could well lead to the wrong answer. A
checklist of the necessary criteria for carrying out an effective sampling strategy
for drug analysis is summarized in Table 23.1. In addition, it is necessary to
collect appropriate blank samples. The blank samples are matrices that have no
measurable amount of the analyte of interest. The ideal blank will be collected
from the same source as the samples but will be free of the analyte. All conditions
relating to collection of the blank sample, storage, pretreatment, extraction,
concentration and analysis will be carried out as for the actual samples. Only
once these questions can be answered, is it appropriate to collect the samples.

TABLE 23.1
Checklist for carrying out an effective sampling strategy in drug analysis
1. What are your data quality objectives?
2. Is specialized sampling equipment needed and/or available?
3. Are the samplers experienced in the type of sampling required/available?
4. Have all analytes been listed? Have detection level and methods been specified for each
analyte?
5. Are all analytes stable in the sample? Is a special preservation method needed after the
sampling?
6. Are there specific types of quality control samples? Does the instrument require
optimization of its operating parameters?
7. What type of sampling approach will be used? Random, systematic, judgemental or a
combination of these?
8. Will the type of sampling meet your data quality objectives? Is the sampling approach
compatible with data analysis method?
9. How many samples are needed? How many methods are specified? How many test
samples are needed for each method?
10. What types of quality control samples are needed? How many exploratory samples are
needed? How many supplementary samples will be taken?

780
 The sample preparation step is necessary to isolate the components of inter-
est from a sample matrix because most analytical instruments cannot handle the
matrix directly. Sample preparation can include clean-up procedures for very
complex (dirty) samples. This step must also bring the analytes to a suitable con-
centration level. During a separation the isolated mixture of analytes is divided
into its constituents. Separation techniques such as chromatography and elec-
trophoresis are well suited for the analysis of complex multi-component samples.
The direct injection of complex matrixes such as plasma, serum, whole blood, tis-
sue homogenates, saliva or urine into these separation systems was studied in
the past with moderate success [20,21]. Typically, however, injection of samples
with large amounts of protein results in a rapid deterioration of the column's
separation performance, an increase in background interferences, and a dra-
matic increase in column backpressure. In addition, the chromatographic sorb-
ent selectivity may be altered by irreversible adsorption of matrix compounds
such as proteins. Therefore, the analysis of drugs present in these biological
samples requires well-designed sample preparation procedures. The isolation
and measurement of organic compounds present in a biological matrix,
especially at low concentration, presents a significant analytical challenge.
Objectives of sample preparation for drug analysis in biological samples and
pharmaceutical products are summarized in Table 23.2. The objectives of the
analytical method will indicate how much effort will be necessary to put into a
sample preparation step. For example, TDM usually requires specificity to dis-
tinguish the drug to be monitored from similar compounds, metabolites or
co-administered drugs. In contrast, a pharmacokinetic study of a potential drug
candidate requires a specific and sensitive analytical method. Therefore, it is
important to obtain the required analyte at a stable and high recovery rate for
each individual application.
TABLE 23.2
Objectives of sample preparation for drug analysis
1. Removal of unwanted macromolecular contaminants (mainly proteins) that would
interfere with analyte determination.
2. Removal of selected analyte components if the resolving power of the chromatographic
and electrophoretic column is insufficient to separate all the components in the sample
completely or in a reasonably practical time.
3. Removal of material that would affect chromatographic and electrophoretic resolution or
reproducibility.
4. Removal of material that could block the chromatograph and electrophoresis tubing,
valve, column or frits.
5. Determination of total analyte present in a complex.
6. Solubilization of analytes to enable injection under the initial chromatographic and
electrophoretic conditions.
7. Dilution to reduce solvent strength or avoid solvent incompatibility.
8. Concentration of the analyte within the detection limits of the analytical instrument.
9. Stabilization of the analyte to avoid hydrolytic or enzymatic degradation.
10. Derivatization of analytes for enhancement of sensitivity and selectivity.

781
23.1.2 Classification of sampling and sample preparation for drug
analysis

As shown in Table 23.3, the sampling and sample preparation for drug analysis
can be mainly classified into five separate operations: sample collection, release
of drugs from matrix, liquid handling, removal of endogenous compounds and
enhancement of sensitivity and selectivity [26]. For the sample collection, proce-
dures are adopted to ensure the integrity of the determinant during collection,
transport and storage. Various sampling techniques have been used for different
samples such as blood, urine, milk, hair and tissues. These samples must be col-
lected without errors by contamination from sampling equipment (e.g., syringe,
pipette, cannulas, tube and vessel) or by use of the wrong sampling technique.
Biological matrices are complex mixtures of compounds and usually have
high protein content. High proportions of the drugs may be bound to protein,
glucuronide and sulphate, and be present as free drug only at low concentra-
tions. Therefore, for analysis of total drug content, the analyst must first release
drugs from the matrix by hydrolysis. The hydrolysis can be performed either by
enzymes or by acid or base. Sonication, dilution or heating are also possible
means of improving the recovery of drugs from biological samples. In the
opposite case, where the goal is to assess the free and therefore therapeutically
relevant concentration, the analyst must ensure that the equilibrium between
free and bound drug is not perturbed during sampling and sample preparation.
Liquid handling procedures mainly provide the links between each opera-
tional step. Procedures for liquid handling involve the addition, mixing, separa-
tion or removal of liquids. They can often be the rate-limiting steps in a sample
preparation process because they are typically labour-intensive and tedious
procedures.
A biological matrix may be solid, particulate or a mixed composition of
organic compounds in an aqueous solution, and consists of many components
such as macromolecules and small molecules. Ultrafiltration and protein precip-
itation can be used to remove protein from biological samples. The membrane
filters used for ultrafiltration are efficient, although care must be taken to avoid
binding of the drug to the membrane. Precipitation using organic solvents and
inorganic acids or salts is effective for removal of proteins. Dialysis can be used to
separate an analyte from the matrix by diffusion through a semi-permeable
membrane rather than centrifugal force with ultrafiltration. On the other hand,
liquid-liquid extraction LLE), solid-phase extraction (SPE) and solid-phase
microextraction (SPME) are useful sample preparation techniques, and can
produce clean extracts for analysis very efficiently. Supercritical fluid extraction
(SFE), immunoaffinity extraction (IAE), molecularly imprinted polymer (MIP)-
based extraction and membrane-based extraction are also used as specific and
efficient sample preparation techniques. These techniques, which are based on
the adsorption or partitioning of analyte, are responsible for removing the
majority of the biological material from the sample matrix prior to analysis.

782
TABLE 23.3
Classification of sampling and sample preparation
Sample collection Suction
Drawing
Cutting
Pipetting
Head-space sampling
Release of drugs from matrix Hydrolysis Acid
Base
Enzyme Proteases
Lipase
[-Glucosidase
Arylsulphatase
Sonication
Liquid handling Aspiration
Centrifugation
Dilution
Evaporation
Filtering
Mixing
Salting-out
Separation
Removal of endogenous Precipitation Organic solvents
compounds Inorganic acids and salts
Ammonium sulphate
Ultrafiltration
Dialysis
Membrane extraction
Liquid-liquid extraction
Supercritical fluid extraction
Solid-phase extraction
Solid-phase microextraction
Immunoaffinity extraction
Liquid chromatography
Procedures for enhancement of Pre-column derivatization
sensitivity and selectivity Post-column derivatization
Use of selective detectors

Derivatization of analyte can also be used to enhance the sensitivity and

selectivity by pre- or post-column reaction. The use of selective detectors for gas
chromatography (GC), high performance liquid chromatography (HPLC) and
capillary electrophoresis (CE) is also effective for this purpose.
As mentioned in Table 23.3, the final aim of the sample preparation must be
to isolate and purify the analyte and to introduce it to the GC, HPLC and CE in a
manner that is compatible with each instrument. Thus, sample preparation
from biological samples for drug analysis is usually carried out in combination
with each technique mentioned above. Furthermore, automated sample prepa-
ration systems are developed in conjunction with a single or a combination of
these techniques.

783
23.1.3 Objectives and scope of this chapter

In this chapter, we review advances in sampling and sample preparation tech-

niques for clinical and pharmaceutical analysis in the past decade. The review
consists of three main parts: in Sections 23.2 and 23.3, general aspects of
biological matrices and pharmaceutical products for sampling and sample prepa-
ration are described; in Section 23.4, sample preparation techniques for clinical
and pharmaceutical analysis are considered according to the matrix type and
extraction type. This review focuses on the main techniques of sample prepara-
tion such as LLE, SFE, SPE, SPME, IAE as well as other new extraction
techniques. The release of drugs from matrix, liquid handling and derivatization
techniques are not covered. The details of sampling and sample preparation for
clinical and pharmaceutical analysis are also described in books [1,2,16-19,
22-24] and well-documented reviews [25-48].

23.2 CONSIDERATIONS FOR BIOLOGICAL MATRICES

The process of interaction between drugs and human organs is divided into three
phases: the pharmaceutical (exposure) phase, the pharmacokinetic (toxicokin-
etic) phase and the pharmacodynamic (toxicodynamic) phase. Pharmacodyma-
nics is concerned with the study of interactions between the drug (or its active
metabolites) and the receptors in target tissues. While this subject is of great
importance in understanding the mechanism of toxicological effects of drugs, it
is normally not a major concern to drug analysts. On the other hand, many
processes that take place during the pharmaceutical and pharmacokinetics
phases are highly relevant to drug analysis. The manner in which a drug is
introduced into a biological system, that is, the route of administration, dictates
the set of digestive and enzymatic conditions to which the drug is subjected and,
therefore, the metabolic fate of the drug and its distribution in various tissues.
Common routes of introduction include oral ingestion, intravenous and intra-
muscular injections, inhalation, and dermal application. Once a drug is intro-
duced into a biological system, it is subjected to the processes of absorption,
distribution, metabolism, and excretion, which are subjects of pharmacokin-
etics. Some basic principles governing the interactions between drugs and the
human biological system with respect to drug introduction, absorption, distribu-
tion, and excretion will be addressed. Such knowledge facilitates the proper
selection of specimens for analysis, sample pretreatment, and analytical data
interpretation. Various biological matrices used for drug analysis in clinical and
forensic applications are summarized in Table 23.4. In this section, characteris-
tics, collection and handling of plasma, serum, blood, urine, saliva, milk, sweat,
feces, tissues and breath are described. Other matrices in the table are not
commonly used in conventional clinical and toxicological analysis, but they may
provide valuable information otherwise not available in clinical and forensic
applications. The details of these matrices for sampling and sample preparation

784
TABLE 23.4
Biological matrices used for drug analysis in clinical and forensic applications
Liquids and gases Blood: Whole blood, plasma, serum
Urine
Secretion and effusion: Saliva, tears, sweat, milk, cerebrospinal fluid,
gastric juice, bile, semen, liquor amnii
Breath

Solids and semi-solids Feces

Keratinaceous tissues: Hair, skin, nails
Organs and other tissues: Brain, liver, lung, kidney, muscle, fat, bone
Stomach contents

are also described in the books [1,19,50] and well-documented reviews [39,42,
49,51-56].

23.2.1 Serum, plasma and whole blood

Blood is perhaps the most useful sample for the identification and quantification
of drugs and for the interpretation of data of toxicological significance. Blood is a
complex fluid containing solubilized proteins, dissolved fats and salts, and
suspended cells. The major constituents-the red blood cells (erythrocytes)-
can be separated from the clear fluid (plasma) by centrifugation, if the blood is
mixed with a suitable anticoagulant during the draw. If blood is allowed to stand
without the addition of anticoagulating agents, the red cells will eventually clot
and the resulting straw-coloured serum can be decanted. In plastic tubes,
clotting may be delayed for more than one hour and some hemolysis can take
place. Hemolysis may also occur during the blood draw if the needle is positioned
too close to the wall of the vessel, or if it is drawn too quickly and the red cells
impact the wall of the sampling tube too hard. Hemolysis may or may not impact
the analytical results, depending on the specific test used. As matrices for drug
analysis, the significant difference between serum and plasma is that serum does
not contain fibrinogen and some clotting factors (2.5-5% of proteins). In general,
both plasma and serum are routinely used for drug analysis. A method developed
for plasma can normally be applied without modification to serum. On the other
hand, whole blood may be the sample of choice when trying to detect the
presence of a drug, or when quantitative information is needed on a deteriorated
blood sample where complete separation of plasma or serum from red cells is not
possible. When anticoagulation is used there is a potential for analytical interfer-
ence. Whole blood is often extracted with no pre-treatment. The blood sample is
added to a sample vial, and extraction is initiated. It may however be preferable
to extract from blood collected with heparin or another anti-coagulation agent.
As heparin is a mucopolysaccharide and may interact with some analytes,

785
ethylenediamine tetraacetic acid (EDTA) or one of the other anticoagulants may
be preferable depending on circumstances. EDTA should not be used where
metallics or organometallics are to be analyzed. The containers or their stoppers
may also cause contamination but these problems are well recognized. A com-
mon error in blood collection is sample haemolysis which causes analyte dilution
and interference in a serum assay. Because serum and plasma consist of a large
amount of protein, drugs will bind to proteins to varying degrees depending on
their individual physicochemical properties. In general, acid and neutral drugs
bind primarily to albumin, and basic drugs bind primarily to acid glycoprotein.
Only free drug is available for extravascular distribution and elimination, and is
able to cross cellular membranes and interact with drug receptors. Where the
goal is analysis of total drug concentration, direct analysis of drugs in protein-
free serum or plasma may not reflect the total drug concentration. Therefore,
protein is often denatured prior to extraction with an organic solvent. Alterna-
tively, drugs strongly bound to protein may be freed with appropriate adjust-
ment of pH.

23.2.2 Urine

Urine is one of the most commonly studied biological matrices for drug analysis,
particularly because of its relative ease of collection. However, samples should
not be taken from women during menstruation, and men should be warned that
there is a risk of contamination by sperm remains. For pharmacokinetic studies,
urine is usually sampled according to detailed instructions over 24 hours into
pre-cleaned bottles. Preservatives may be required to stabilize the analyte
particularly if it is susceptible to bacterial degradation or the urine may be
chilled during the collection period. To guarantee the stability of the urine, the
time that elapses between sampling and analysis should be kept as short as
possible. To avoid the risk of contamination the urine is not acidified. After
through shaking, the urine is divided and the subsamples are frozen immedi-
ately. As a matrix, urine has moderate complexity, and typically contains both
organic and inorganic constituents, as well as a relatively high salt content. In
addition, urine suffers from high variability. Results are typically normalized by
expressing drug concentration per gram creatinine, in order to circumvent the
highly variable dilution that results from the body's attempts to maintain water
balance [57,58]. Compared to blood, plasma, or serum, urine is relatively free of
protein, thus making it possible for direct extraction with an organic solvent.
However, the interpretation of results obtained from urine samples is compli-
cated by several factors, such as the amount of excretion, the variation of pH
values, the variation of ionic strength, and the time lapse after the intake of the
drug. Variations in urine pH effects both elimination of many drugs, as well as
extraction efficiency when effects on the drug's acid/base balance are significant.
Variations in ionic strength can also have a significant effect on drug extraction.
The additional components of the sample may also provide for competitive

786
sorption, further reducing the amount extracted. While this provides an oppor-
tunity for studying distribution of a drug between the various components of a
urine sample, it is often total urine concentration that must be reported. This is
accomplished by using an appropriate internal standard, or standard addition
for quantification. The internal standard selected should be structurally related
to the other compounds. If urine samples must be significantly diluted prior to
extraction, because of very high drug concentrations, variability in the matrix
becomes of less significance for extractions.

23.2.3 Saliva

Saliva is a colorless fluid excreted into the oral cavity from three principle
glands: (1) the parotid gland, exiting at the top of the mouth, secretes saliva
derived mainly from blood plasma (serous fluid); (2) the sublingual glands, exit-
ing at the sides of the mouth, excrete both serous fluid and mucin; and (3) the
submandibular glands, exiting at the base of the tongue, also excrete both serous
fluid and mucin. Saliva is approximately 99% water, 0.3% protein (mostly
enzymes) and 0.3% mucin with the balance salts. The mucin gives saliva its
sticky character. The low protein concentration in saliva makes drug binding
minimal compared to that observed in plasma. Unstimulated saliva pH is in the
range of 5.6-7.0 and increases with stimulation (to more approximate the pH of
blood, i.e., 7.4) to a maximum of 8.0. Therefore, as discussed above, drug concen-
trations in saliva partially depend on the pH of the saliva and the degree of stim-
ulation. Compared to urine collection, saliva sample-collection procedures are
less invasive and cause less concern about violation of privacy and adulteration
of samples. Because saliva can be used to estimate the actual, non-protein
bound, circulating concentration of some drugs and their metabolites at the time
of collection, test results may have potential for correlation with performance
impairment [59]. Compared to blood, saliva samples can be directly analyzed
without a prior extraction step. These advantages have promoted numerous
investigations and several review articles on the suitability of saliva in TDM and
in forensic applications [59-61]. Limited amounts of mixed saliva may be col-
lected by spitting. Larger amounts of saliva may be collected by stimulating
saliva flow by mastication of rubber bands, wax, Teflon tape or gum; sucking on
pebbles, marbles or candy; by placing citric acid on the tongue; adsorption on
cotton rolls or administration of pilocarpine. Selection of material for saliva sim-
ulation must be carefully chosen because lipophilic drugs may be adsorbed into
the material. For identification purposes, adsorption (similar to SPE or SPME)
could be employed to directly extract the drug from the saliva while inside the
mouth. Little research has been done in this area, as the sample size would be
limited and the adsorption process variable (depending on the cooperation of the
individual, i.e. the more sucking/chewing, the greater the adsorption). Some
devices (OraSure®) both stimulate saliva flow and collect the saliva on an absor-
bent pad [62-64]. Other devices have been developed to acquire saliva from

787
selected glands [65]. In general, these devices acquire saliva by placing the end of
the device over the gland and applying suction [66]. Although selected gland
secretions have an advantage in reducing saliva/plasma ratios and minimizing
oral contamination, most studies collect mixed saliva specimens because it is less
invasive. An ultrafiltration device (SalivaSac®) has been developed to reduce the
viscosity of saliva for easier analysis, and is similar to that collected by suction
from the parotidal gland. It consists of a dialysis membrane enclosing sucrose
crystals and is approximately 3.5 cm in diameter and a few millimeters thick
[67,68]. The device is placed in the mouth and massaged with the tongue for a
few minutes until all of the crystals are dissolved, collecting 1-2 ml of saliva
ultrafiltrate. Both the sucking on the device and the sweet taste stimulate saliva
production. The SalivaSac® may also be externally coated with citric acid to
stimulate saliva production. The dialysis membrane is chosen to exclude most
higher-molecular-mass substances, therefore, the mucopolysaccharides, food
particles and bacteria are not collected. The SalivaSac® may also have a handle
for easier insertion and removal. However, two problems concerning the correla-
tion of drug levels of the ultrafiltrate with saliva levels are apparent when quan-
titative information is needed. The sucrose in the SalivaSac® takes up an
appreciable amount of the molar volume of water inside the device. Therefore,
measurement of the density of the fluid is necessary by careful weighing and a
correction factor must be calculated [69]. Furthermore, diffusion through the
dialysis membrane is related to molecular mass and, therefore, water is prefer-
entially collected relative to drugs [69]. Thus, the concentration of drugs in the
SalivaSac® is lower than in the external saliva and this ratio may vary depend-
ing on how the user moves the device during the saliva collection. If one only
wishes to have qualitative information on drug use, then these concerns are not
applicable.

23.2.4 Milk

The main constituents of human milk are water (88%), proteins (3%), lipids (ca.
3%), carbohydrate in the form of lactose (6.8%) and other solids (minerals,
vitamins etc.). The lipids are in the form of fat droplets suspended in the watery
matrix. The colostrum that appears in early lactation differs significantly from
true milk in that it contains less lactose and virtually no fat. The non-water
constituents are present in different physical forms; dissolved (lactose), colloid-
ally dispersed (protein) and emulsified in water (lipids or fats) [70]. Human milk
can serve both as a means of exposure of a newborn to compounds the mother
has been previously exposed to, and provide a relatively convenient means of
biomonitoring for toxicant exposure. Human milk is quite constant in composi-
tion from the third week on, but the colostrum from early lactation is signifi-
cantly different as noted above. Such variation could impact analyses if not
taken into account. The sampling of human milk is always done with the
assistance of a trained person. The breast, in particular the nipple, is cleaned

788
with sterile compress and bidistilled water, in order to remove possible contami-
nants (e.g., breast cosmetics). The sampling equipment, which includes an
electric milk pump is unpacked and installed at the place where the sampling is
done. The breast adaptor is pressed on to one breast, so that no air is drawn in
and a light vacuum is created. The pump is started and controlled by the test
person herself. As soon as the desired volume has been reached, the pump is
stopped and the filled milk bottle is removed from the breast. To minimize
contact with the air, the sample is rapidly subdivided into prepared tubes. If
possible this procedure should be done under clean-room conditions in the
laboratory. If the let-down of the milk dose not take place, even though the
breast is full, the test person may use a nasal spray of oxytocin in order to
stimulate the milk flow. Milk samples are commonly used for the trace analysis
of pesticides, heavy metals, antibiotics and some drugs.

23.2.5 Hair

Hair is an attractive target for drug analysis as it is relatively non-invasive to

collect, and provides a historical record of exposure, even though it does suffer
from variability in drug concentrations by hair type, due to drug affinity varia-
tions. Hair is now being recognized as a third fundamental biological specimen
for drug testing after urine and blood. An additional advantage of hair as a
matrix for analysis is that it is fairly nonpolar and so tends to absorb parent drug
molecules, as they are typically less polar than metabolites. Because of this it is
an ideal matrix for extraction of analytes to non-polar extraction phases, partic-
ularly where the parent drug is extensively metabolized and often non-detect-
able in other tissues. Throughout most of its length the hair shaft consists of the
keratinized remains of cells in three distinct layers: an outer cuticle, an inner
medulla and a central cortex. Generally, the cuticle is less intact toward the
distal end of the hair shaft than the proximal end. Hair grows in predictable
patterns, lengths and textures in different body areas. The type of hair deter-
mines the lengths and textures, which in turn is dependent on sex and age. The
patterns, colors, textures, hair diameter and hair growth rates are all dependant
on anatomical location, race, gender and age. The average rate of hair growth is
0.44 mm per day (range, 0.38-0.48) for men and 0.45 mm per day (range,
0.4-0.55) for women in the vertex region of the scalp. Although the mechanisms
of drug incorporation into hair have still not been sufficiently clarified, drugs are
thought to be distributed into hair by two processes: incorporation into the
growing hair shaft from blood and/or adsorption from other media such as sweat
from the environment. The most important considerations in hair sampling are:
collecting hair from a preferable anatomical location (posterior vertex) where
hairs are relatively uniform, collecting the hair a uniform distance from scalp
(especially if sectional analysis is to be performed), collecting a sufficient sample
for the number of tests to be performed, preventing contamination and accu-
rately identifying the sample.

789
 Hair is mainly collected from the area at the back of the head, called the
posterior vertex. This area has less variability in hair growth rate than other
areas. Moreover, the hair is less subject to age and sex-related influences [71].
The sample size collected should be 200-250 mg at a minimum because of the
relatively large interindividual variability. When head hair is not available or is
too short, pubic or axilla hairs may be collected. Other than for drugs of abuse
testing (violation of laws on narcotics, probation, drug abstinence in detoxifica-
tion treatments, child custody, divorce), testing hair for pharmaceuticals in
forensic science is most frequently performed on deceased persons. Thus, the
quantity of the collected hair could be significantly higher than for living people,
except indeed for newborns, the very young or the very old. The problem of
external contamination is very important for drugs of abuse as people can be
exposed to a smokey or dusty environment. Papers dealing with the problem,
which also impacts the establishment of cut-off values, are numerous and
various studies have been published [72-76]. The approach is very different for
pharmaceutical analyses as the drugs are normally to be taken by mouth or by
another therapeutic route under normal pharmaceutical presentation. Another
important serious concern is the change in the drug concentration induced by
cosmetic treatments of hair. The strong bases used for such treatments may
cause hair damage resulting in loss from the hair matrix or alternatively under
favorable environmental contamination conditions, a higher incorporating ratio
[76]. In fact, the history of hair analysis is the history of extraction procedures,
because after the drug has been extracted it can be handled as if it had been
extracted from urine or blood samples. Various extraction procedures have been
proposed including chronologically [77]: methanolic sonication, 0.1 M sodium
hydroxide, 0.1 M hydrochloric acid, water, buffers, 1 M sodium hydroxide, 0.1%
sodium dodecylsulfate, acetone, pronase, performic acid-pronase, methanol, 5 M
hydrochloric acid, proteinase K, dithiothreitol, phosphate buffer pH 7.4, -glucu-
ronidase-arylsulfatase and SFE [78,79]. With the exception of SFE and methan-
olic sonication, the resulting mixture from the pretreatment step must be
further purified prior to analytical examination. It could be as simple as LLE
with common solvents or as complex as a multi-step trace enrichment procedure
as that devoted to the analysis of 17-keto steroids and their esters [80]. The fields
in which hair analysis has so far been applied are mainly forensic toxicology and
drug abuse studies, followed by clinical toxicology and clinical chemistry. The
largest number of papers on hair analysis has dealt with cocaine, opiates and
amphetamines. These drugs are the subject of almost 50% of all papers on hair
analysis. However, recent hair analysis studies have addressed other kinds of
drugs, for example doping agents like clenbuterol, therapeutic drugs like benzo-
diazepines, methadone and carbamazepine.
23.2.6 Sweat
Sweat is an alternative target for the determination of drug-excretion patterns,
and drug compositions in sweat may partially contribute to the observation of

790
drugs in hair [81,82]. Sweat is excreted from eccrine and apocrine glands and
skin is bathed with aqueous and sebaceous secretions, especially on the face and
scalp. The sebaceous secretions are primarily lipids that may transport and
absorb many drugs. Sebum is excreted more on the scalp and forehead than on
other areas of the body. Therefore, different concentrations may be expected,
depending upon the area of the body in which the sample is taken, because
fat-soluble drugs may be sequestered or secreted in sebum. Almost all studies
obtain mixed secretions of sweat and sebum, which is incorrectly referred to as
sweat. Sweating may be induced by exercise and several milliliters of sweat may
be collected in conjunction with an occlusive bandages, consisting of one to three
layers of filter paper or pieces of cotton, gauze or towel. Significant advances
have been made during the past years with the development a "sweat patch"
technology, which was proposed by PharmChem Labs. (Menlo Park, USA).
"Sweat patches" used for collecting perspiration samples from human skin have
improved from an earlier occlusive design to the current dressing technology
using very thin polyurethane and acrylate adhesives. This new nonocclusive
approach allows the moisture content of perspiration to evaporate and makes it
possible to apply a single dressing to skin and maintain sterile attachment for
several days with rare dermatological reactions. The final content of the collec-
tion pad is an accumulation or integration of that contributed to it by skin [83]. A
patch was developed that included a chemical binding layer in the absorbent pad
to prevent back-diffusion of the drug through the skin. This design has been
used to monitor theophylline in monkeys [84] and caffeine in infants [85]. Both
of these early patches used aqueous media and an occlusive covering to stimulate
sweat and prevent evaporation of the drug. A device has also been developed that
has a covering that allows the passage of sweat from the skin through the device
and prevents external water and other molecules from back-diffusing into the
absorptive pad [86]. This device is being marketed as the PharmChek® sweat
patch [87]. A potential problem with the PharmChek® patch is the absence of a
layer between the skin and the absorptive pad, to prevent bacterial transfer into
the pad and, therefore, the possibility of bacterial growth and drug degradation.
Careful preparation of the skin prior to application of the patch should kill or
remove bacteria and prevent these problems, and the absence of substantial
moisture in the pad decreases the possibility of bacterial growth. Even with
these many devices, sweat is more difficult to collect non-invasively than is
saliva, due to the lower amounts/unit area secreted in a given time. As an
alternative collection of sweat, wiping the skin with a cotton pad moistened with
alcohol was developed [88]. This procedure allows rapid collection of drugs that
may arise both from sweat evaporating on the surface of the skin and from
external contamination. The exact composition of the fluid analyzed is not
known because both the aqueous secretions (true sweat) as well as sebum are
collected. Whether or not the presence of drugs on skin wipes indicates use only
or use and exposure is unknown. In a survey of a university population, skin
wipes detected more cocaine use/exposure than did hair analysis [89].

791
23.2.7 Other matrices

Tissues: While tissue may be obtained at biopsy and analyzed it is more

frequently examined as a post operative sample or as a forensic sample. In the
former case it is often important to avoid histological fixatives and in the latter
putrefaction can cause significant interference in assays.
Faeces: Faecal analysis is disagreeable but sometimes necessary. Collection
of timed samples is difficult as the mixing of gut contents makes it almost impos-
sible to identify which stools were formed during the collection period or those
that might contain the determinant (if exogenous). Two approaches are used to
produce a fluid sample for analysis: ashing to reduce the bulk followed by dis-
solving of the temperature stable remnants or more commonly, homogenization
with an appropriate volume of fluid to produce a faecal slurry, removal of
remaining solids and subsequent processing. The unpleasant nature of the sam-
ple dictates that the early stages of sample preparation are conducted in a fume
hood.
Breath: Analysis of compounds in breath is a newer area of clinical analysis,
with reports appearing for analysis of ethanol, acetone and for isoprene as an
indicator of metabolic state [90,91]. A custom designed mouthpiece and Tedlar
bags were used for collection of expired breath. In these cases the samplers were
fitted to a manual SPME fibre device for extraction. Adsorbent tubes may also be
used for pre-concentration prior to analysis.

23.3 CONSIDERATIONS FOR ILLICIT DRUG PREPARATIONS AND

PHARMACEUTICAL PRODUCTS

Illicit drug samples encountered in crime laboratories are normally exhibited in

dosage units and presented as their own entities or as counterfeits of brand-
name pharmaceutical products. Their characteristics are summarized in Table
23.5. In addition to identifying the presence of a specific drug in an exhibit,
laboratories are often requested to provide additional information that may be
helpful to the investigation process. On the basis of the characteristics listed in
Table 23.5, exhibits may be profiled and linked to common sources or routes of
distribution.

TABLE 23.5
Characteristics of illicit pharmaceuticals
1. Various forms of carriers or containers are used.
2. Various unintentional impurities, such as degradation products, synthesis by-products,
and residual solvents are present.
3. Intentional multiple-component drug combinations and the addition of adulterants and
excipients are common.
4. The contents are often unexpected.
5. Many of these drugs may be sold as counterfeits of pharmaceutical products.

792
 The nature of the active ingredient is only one of several important consider-
ations in the design of pharmaceutical products. It is widely recognized that
pharmaceutical products from different manufacturers, containing the same
active ingredient and dosage, can produce markedly different circulating con-
centrations of the active ingredient. Optimal drug formulation is a crucial step in
the design of products that produce optimal therapeutic effect. In addition to
meeting regulatory standards for active ingredient quality, identity, purity and
stability, products are typically optimized for utility and bioavailability to assure
the best competitive advantage. The nature of the formulation procedures
and/or auxiliary (non-medical) components of a product can for example, make a
product easier to handle, limit degradation, maintain homogeneity, enhance
bioavailability and targeting of the active ingredient to receptor sites, provide for
additional routes of administration, and target the product for additional thera-
peutic indications for use. These auxiliary inactive ingredients, known collec-
tively as excipients, can also solubilize the active ingredient, suppress the growth
of microorganisms, provide bulk or a coating for a tablet or capsule, provide color
or mask an unpleasant taste or odor. A good working definition of excipients is
"agents in a medicinal preparation regardless of the nature, purpose or quanti-
ties employed, other than the components intended as the active ingredients".
With so many tasks to perform, it is obvious therefore that the nature and num-
bers of excipients in various pharmaceutical products can vary widely. It is
essential therefore, that for analysis of pharmaceutical products, the nature of
these excipients is taken into account, for their potential impact on the process
of analysis of the active ingredient. Excipients can be divided based on either for-
mulation type or excipient action. Table 23.6 provides an overview of common
excipients for various drug formulations. In this section, characteristics, collec-
tion and handling of illicit preparations and pharmaceutical products for foren-
sic and pharmaceutical analysis are described.

23.3.1 Illicit pharmaceuticals

Illicit preparations of drugs of abuse may be prepared by various synthetic

routes, and for each of these, incomplete and side reactions may also occur.
Extensive purification of the active ingredients is not normally performed fol-
lowing synthesis. As a result, several impurities are typically present in these
preparations, and impurity profiling may be used to track the source of these
products when they are confiscated. LLE or occasionally SPE have been used to
prepare samples of confiscated materials for profiling by either GC or HPLC.
Recently, the use of SPME has been evaluated for impurity profiling in confis-
cated amphetamine and ecstasy tablets [92]. Headspace extraction was prefer-
red for the ecstasy tablets to avoid contamination of the SPME fiber by tablet
components. For amphetamine powders, direct immersion extraction was feasi-
ble, and PDMS/DVB and 100 im PDMS fibers provided equivalent profiles. The
repeatability of profiles was consistent with those attained from LLE, indicating

793
TABLE 23.6
Common excipients for pharmaceutical preparations

Bulk materials Controlled release Flavorings

Oral-solids binding agents Sugars
Starch Stearate esters Artificial sweeteners
Lactose Cellulose derivatives Sodium / potassium chloride
Calcium salts Talc Fruit flavors
Citric acid Aliphatic alcohols Essential oils
Bicarbonate salts Carboxylate polymers Imitation flavors
Dextrose Gel polymers
Cellulose Copolymeric derivatives Preservatives
Oral-liquids Resins Antimicrobials
Water Phospholipids Antioxidants
Alcohol Waxes and wax esters Chelating agents
Syrups Silicon dioxide
Glycerin Coloring
Glycols Tartrazine
Unabsorbed Carmine
monosaccharides Amaranth
Oils Capsules Eosin
Topical-solids Gelatin Erythrosine
Mineralhydrocarbons +sucrose (hardening)
Glycols +glycerin (softening) Suspendings, stiffening,
Lanolin +sorbitol (softening) emulsifying
Silicates Microencapsulation Ionic agents
Beeswax Plastics / proteins Non-ionic agents
Topical-liquids Animal and vegetable Tragacanth and acacia
Water origin Mucilages and gums
Alcohol Albumin Methyl cellulose
Oils Phospholipids Paraffin
Glycerin Cholesterol Lubricants - talc
Dimethylsulfoxide Poly(methyl methacrylate) Lubricants - edible oils
Injectables Fatty alcohols
Water (5% glucose, 0.9%
saline)
Glycols Coatings Propellants
Glycerin Shellac Hydrocarbons
Nutrient fats Silicones Fluorocarbons
Fatty acid esters Mucilages / gums Other halocarbons
Oils Gluten Compressed ambient gases
Suppositories Acidic polymers
Fats e.g. carbomers
Mono-, di-, tri-glycerides Basic polymers Buffers
Hydrogenated, ethoxylated e.g. poly-L-lysine Phosphate
derivatives Cellacephate Borate
Beeswax Paraffins Acetate
Polyethylenes Waxes Citrate
Glycerin Beeswax Amino acid
Gelatin Vegetable waxes Tris

794
similar abilities in differentiating between closely related but different drug sei-
zures. Interestingly, for ecstasy, isosafrole was identified as a precursor for syn-
thesis. With LLE, this compound was difficult to analyze, due to its high
volatility. There has also been a report on the use of SPME for profiling of manu-
facturing by-products and impurities from an illicit drug seizure of 4-methoxy-
amphetamine [93]. The results indicated the synthetic route likely used in the
preparation of the compound, and the method was found to be rapid, non-
destructive and give results complementary to those from conventional LLE.

23.3.2 Pharmaceutical excipients by formulation type

23.3.2.1 Oral solidformulations:e.g. tablets

In addition to fillers, excipients are commonly used to control disintegration,
dissolution, release kinetics, pH, and compressibility and provide effervescence,
encapsulation, coating, flavour and sweetness. Excipients may also be used to
improve product flow during production although it is more common to carefully
choose all excipients such that the flow of the end product is satisfactory. It is
difficult to 'fix' a blend with poor flow characteristics by the addition of one or
more excipients, and this also makes the product more expensive and difficult to
manufacture.
23.3.2.2 Oral liquid formulations:e.g. elixirs
Elixirs are sweetened alcoholic solutions with an ethanol content that is either
low (8-10%) or high (73-78%), with glycerol and amounts of water varying from
none to over 90%. Common excipients include fillers, flavorings, preservatives
and those to address solubility and stability problems. Controlled release is often
achieved by entrapping or coupling the active ingredient in an unabsorbable
complex. Encapsulating may also be used. Ethyl alcohol is often used to enhance
solubility of the active ingredient.
23.3.2.3 Topicals
Excipients in topicals may be present to assist the active ingredient in penetrat-
ing the skin, preventing it from washing off, or providing for an occlusive
dressing at one extreme, to a vanishing effect at the other. In terms of penetra-
tion, the formulation may be required to allow penetration to the skin only, to
the underlying tissues, or to the circulatory system. Topicals are further subdi-
vided into solid and liquid formulations.
Solid formulations:e.g. creams/ointments/gels. It is often difficult to distin-
guish between solid and liquid topicals. While dusting powders are obviously
solid, other formulations considered solid for this treatment include anything
that does not run when poured. In addition to fillers and vehicles to control
tissue penetration, excipients are used to solubilize the active ingredient, pro-
vide antibacterial action, enhance stability, emulsify and act as suspending
agents. Mineral hydrocarbons are commonly used as the base for ointments and
gels and creams typically have a more aqueous base.

795
 Liquid formulations: e.g. lotionsemulsionslsuspensions. Water as a bulk
filler is probably the most common excipient for liquid topicals. Excipients to
control pH and suppress microbial growth are also common. Isopropyl and
larger alcohols are commonly used to improve consistency, feel and solubiliza-
tion. Oils are used to produce emulsions and suspensions, and other compounds
provide humectant (water-retaining) and emollient (softening or soothing) prop-
erties. Topicals for ophthalmologic applications require excipients to control
tonicity, pH and sterility as well.

23.3.2.4 Injectables
Injectables are prepared for intramuscular, intravenous and subcutaneous
administration. Critical factors for these preparations are their sterility and
tolerance of the tissue at the site of injection. Water is the predominant solvent,
although non-aqueous solvents can also be used to improve the solubility of the
active ingredient. Glucose (5%) and saline (0.9%) are generally used, although
some preparations may be hypertonic. Some preparations may also incorporate
fixed oils, and nonglyceryl esters of fatty acids.

23.3.2.5 Suppositories
Excipients for these are generally fatty preparations with melting points below
body temperature and solidification points above room temperature.

23.3.2.6 Inhalants
The most obvious excipient in these preparations is the propellant. These may be
hydrocarbons, fluorocarbons, other halocarbons or compressed ambient gases
such as nitrogen and carbon dioxide. Other major excipients in inhalants can
also include solid or liquid components, for example, a filler such as lactose is
often used in dry powder inhalers.

23.3.3 Pharmaceutical excipients by action

23.3.3.1 Bulk materials

These are generally considered as carriers of the active ingredient. Where the
amount of active ingredient is too small for convenient handling, an additive is
required to provide suitable bulk. In many cases, the amount of bulk material far
surpasses the amount of active ingredient in a product. Additional uses for bulk
materials are as controlled release agents, or agents to improve flow characteris-
tics, stability or compressibility. Matrix variation in a particular product, due to
variation in a bulk material should be minimal. Due to increasingly stringent
standards for pharmaceutical products, particularly in the United States, suppli-
ers tend to have only one manufacturer for an excipient of a particular grade.
Drug manufacturers will tend to use one supplier, but often have a back-up
supplier to ensure continuity of supply in case of emergency. It is important to
note though that "fillers" are often mixtures of several ingredients, often of

796
biological origin, the nature of which can vary. Some matrix variation may
therefore result between lots for either of these reasons. The list of compounds
that can potentially be used as fillers in pharmaceutical preparations is substan-
tial (see Table 23.6). In many cases fillers will not dissolve in a sample and will
therefore be present as a discontinuous phase that may compete with the
extraction phase for the drug.

23.3.3.2 Capsules
These small single-dose containers are pre-manufactured and then filled with
either a solid or liquid medication. They are manufactured primarily from
gelatin. Common additives are sucrose, to increase hardness, and glycerin or
sorbitol to increase softness. As a protein, it would be expected that gelatin could
bind to some drug compounds. Gelatin content should not be permitted to vary
significantly in samples for analysis, without the use of a suitable internal
standard.

23.3.3.3 Coatings
Of the wide variety of tablet coatings used, the proteins and polymers would be
expected to compete with the extraction phase for the analyte of interest, while
the waxes and paraffins have the potential to change the sample polarity, and
hence impact partition coefficients. If the total amount of coating present is low,
the effect may not be significant.

23.3.3.4 Flavorings
Pharmaceuticals have for the most part, left the era where nasty-tasting medi-
cine was considered strong, and therefore a positive characteristic. The public
now expects oral pharmaceuticals to be palatable, particularly in pharma-
ceuticals for children. Flavor preference and consistency for pharmaceuticals are
areas of significant concern. Sucrose and glucose are commonly used sweeteners,
and glycerin may be used as both a sweetener and a solvent. Other sweeteners
may be present because of the use of natural flavorings. Sorbitol, xylitol and arti-
ficial sweeteners may be used where there is concern over disturbing the carbo-
hydrate control of diabetics. In addition to its use to provide desired tonicity or
solubility, sodium chloride may also be used to improve palatability. Many fruit
flavors prepared as syrups from natural sources are used in flavorings. However,
with the significant variability that exists in these preparations it is difficult to
precisely predict their impact on extraction, where their concentrations are
high. Synthetic flavorings by contrast, are more consistent and cost effective.
There is however a common consensus, whether founded or not, that natural fla-
vors taste better and are safer than synthetics, a fact that may limit the use of
synthetics. Where significant variation in flavoring composition exists between
samples extraction efficiency may be variable and hence external calibration
impractical. Standard addition or the use of an internal standard may be
required. Essential oils are also commonly used in flavoring, and like natural

797
fruit flavors, can be a rather ill-defined mixture of many compounds. Because
they consist mainly of compounds with low boiling points, they could interfere
with headspace extraction of pharmaceuticals by solid sorbents.

23.3.3.5 Colorings
While color in pharmaceuticals may be looked upon as a matter of aesthetics
rather than of necessity, the adding of color for psychological or aesthetic rea-
sons has been shown to influence therapeutic response. Coloring also enhances
the effect of flavoring. Most importantly, color (and tablet shape) can also give a
unique identity to a product, reducing confusion, errors and noncompliance. Col-
orings used in pharmaceuticals can be divided into two classes. Dyes are soluble
colored materials that go into true solution, generally in water. Pigments on the
other hand are colored particulate material. A subclass of the pigments is the
'lakes', which are prepared by the tight adsorption of dyes to otherwise un-
colored insoluble particulate materials such as alumina. Particulate coloring
agents, if present in significant amounts, may compete with the extraction phase
for affinity for the active ingredient. Where amounts vary from one sample to
another, some consideration will have to be given to this in analytical method
development.

23.3.3.6 Controlled-releasebinding agents

By their very nature, most of these compounds are designed to compete strongly
with the extraction phase, for the active ingredient. A very specific internal
standard, or standard addition will be required to adequately quantify total
concentration of the active ingredient. By contrast, SPME techniques are ideal
for studying or monitoring the therapeutically relevant dose delivered by these
preparations.

23.3.3.7. Buffers
The pH of a pharmaceutical product typically must be controlled, either to
optimize solubility, bioavailability or stability, or for compatibility with the site
of administration. A salt of a drug compound is more water-soluble than the free
acid or free base form, and upon dissolution is less likely to result in a strongly
acidic or basic solution. Non-ionic species however are more lipid-soluble and
better absorbed. It is therefore critical in many drug formulations that buffering
components are included to achieve an appropriate balance between water and
lipid solubility so that a compound reaches the required absorptive surface, with
an adequate concentration of the lipid-soluble moiety to permit absorption. In
the case of injectable medications, solutions for intramuscular and subcutane-
ous injection must be close to the pH of the tissue into which they are injected.
Modest deviation from physiologic pH is better tolerated for solutions for slow
intravenous infusion, particularly where the molarity (buffering capacity) is low.
Topicals for skin application are slightly acidic to avoid skin irritations, intra-
vaginal preparations typically have a pH of 4-5 to maintain normal vaginal flora,

798
and ophthalmologic topicals require physiologic pH. The pH appropriate for
optimum formulation and/or bioavailability may not be appropriate for optimal
extraction, and significant pH adjustment may be required to counter the
buffering capacity of the excipients, and provide for adequate extraction
efficiency.
23.3.3.8 Suspending, emulsifying and stiffening agents
These excipients are used in both liquid and solid dosage forms, typically in a
small quantity. Many of these compounds are surfactants that act as suspending
and emulsifying agents to aid in the development of solutions and topical
formulations of materials that may otherwise be difficult to prepare in an
acceptable form. They may also be used in aqueous solutions to be inhaled, to
hydrate and liquify bronchiopulmonary secretions or to help clear debris. They
also have cleansing and antimicrobial action and may be used as preservatives.
The surfactants are either ionic or non-ionic. Other compounds that may be used
include mucilages and gums, methylcellulose, paraffin, fatty alcohols and lubri-
cants such as talc and edible oils. All of these compounds could potentially
change the polarity of a sample, and the partitioning of an analyte between a
sample and an extraction phase. If only a very low concentration of one of these
excipients is employed, the impact on extraction efficiency/precision would be
low. Where the concentration is higher, the primary concern would be in
ensuring consistency between samples, and between samples and standards.
Poor precision would result if there were variation in the nature and/or concen-
tration of these excipients.
23.3.3.9 Preservatives
There are three primary classes of preservatives, antimicrobials, antioxidants
and chelating agents. The surfactants discussed previously may also be employ-
ed as preservatives. Antimicrobials include such things as phenol and thymol,
thimerosal, parabens, chlorobutanol and benzyl alcohol. Concentrations range
from about 0.1-10%. Compounds present at the high end of this range may
interfere with extraction. Antioxidants include butylated hydroxy anisol and
butylated hydroxy toluene at about 0.02% for non-aqueous formulations, and
sulfites at the ppm level for aqueous formulations. At these levels, interference
with extraction is unlikely. The primary chelating agent used is ethylenediamine
tetraacetic acid (EDTA). It is typically the disodium salt that is employed. Unless
the compound being extracted is a metal species, interference with extraction is
unlikely.

23.4 SAMPLE PREPARATION TECHNIQUES IN CLINICAL AND

PHARMACEUTICAL ANALYSIS

Biological materials and pharmaceutical products are complex as mentioned

above. They often contain proteins, salts, acids, bases and numerous organic
compounds with similar chemistry to the analytes of interest. In addition, the

799
analytes often exist at low concentration in these samples. Thus, sample prepa-
ration is usually necessary in order to extract and isolate the analytes of interest
from these complex matrices. Many sample preparation approaches such as LLE
[25-30,94], SFE [31], SPE [26-37,94-100], SPME [36-46,94,101-106], affinity
sorbent extraction [97,107,108], membrane-based extraction [34,35,98,
109-111] and others are used for drug analysis. Such steps may be the only
sample preparation or quite often they are used in combination with other
methods. In this section, the characteristics of these techniques and their
applications in clinical and pharmaceutical analysis are discussed. Applications
of these sample preparation techniques reported during the past five or six years
are summarized in Tables 23.7-10.

23.4.1 Protein removal

Protein in biological fluids or tissues must often be removed from the sample
matrix before a suitable extraction process is performed. If the analytes have
high protein-binding capacities, the adopted protein-removal protocol should
provide for release of these analytes from protein so that they may not be lost.
Deproteinization approaches include the use of (1) extreme temperatures,
osmotic or ionic strength, or pH conditions; (2) organic precipitation agents; (3)
proteolytic enzymes. Protein precipitation, ultrafiltration and dialysis can be
used to remove protein from biological samples. Common deproteinization
procedures involve protein denaturation by acids such as trichloroacetic acid
and sulfosalicylic acid, the use of extreme temperature of 90-100°C or freezing-
thawing cycles, and co-precipitation with bases such as zinc hydroxide. However,
methods using such precipitants rarely exceed 80% recovery, frequently less.
On the other hand, the use of water-miscible organic solvents has several
advantages. The mild conditions used under the precipitation process minimize
the possibility of decomposition of labile drugs. Furthermore, appropriate
solvents may be selected so that they are compatible with the mobile phases used
in HPLC analysis. Commonly used solvents include methanol, ethanol, acetone
and acetonitrile. Methanol is often preferred because it produces a flocculent
precipitate to give a clearer supernatant suitable for direct injection. Aceto-
nitrile is the most popular, but there can be problems of poor recovery and also
late eluting peaks. There are other problems that should be considered. Protein
precipitation may not be complete. Some globulins can remain in solution, and
react in the analytical system. Furthermore, such procedures by their nature are
dilutional and are best utilized for high concentration determinants. The pre-
cipitants can react in the analytical system, eluting as a peak in HPLC or
interfering in a spectrophotometric assay. The membrane filters used for ultra-
filtration are efficient, although the analyst should ensure that binding of the
drug to the membrane is not significant. The technique of ultrafiltration may be
used to measure total drug in serum if it is only marginally protein bound.
Ultrafiltration works best for acidic and neutral drugs; basic drugs tend to bind

800
to the membrane. In addition, care must be taken to not compromise the
membrane integrity during centrifugation.
Dialysis can be used to separate an analyte from the matrix by diffusion
through a semi-permeable membrane rather than centrifugal force with ultra-
filtration. Dialysis is the classic procedure for protein removal from small
molecular weight analytes; the automation of this technique in continuous flow
analyzers revolutionized clinical chemistry enabling an explosion of throughput
to occur in that area. The combination of this technique with a trace enrichment
column is available commercially for linkage to an HPLC (e.g., Gibson ASTED®
system). These membrane-based techniques are also described in Section 23.4.7.

23.4.2 Liquid-liquid extraction

LLE is a classic sample preparation technique and depends on the partitioning of

the determinant between two immiscible liquids. Many methods have been
developed for the extraction of drugs from biological fluids involving extraction
under acidic (pH 3) or basic (pH 8) conditions, followed by a sequential back-
extraction with bicarbonate and sodium hydroxide to remove the strong acids
and weak acids, respectively. Neutral drugs will remain in the organic solvent.
Important parameters that have to be considered when selecting an appropriate
solvent for the intended extraction system include density, volatility, polarity,
selectivity, and solubility of drugs. Obviously, solvents that are harmful to
health or the environment should be avoided. The selected solvent also has to be
compatible with the subsequent analytical step. A solvent heavier than water is
preferred if a conventional separating funnel is used. On the other hand, it
would be beneficial to use a lighter solvent if extraction is conducted in centri-
fuge tubes or if the phase separation is to be achieved by freezing the lower
aqueous phase. The concentration process will be facilitated by using a solvent of
high volatility. Solvent polarity determines the solubility of the drug. On the
other hand, the solvent selectivity parameters reflect the solvent's ability to
function as a proton acceptor, a proton donor, and a strong dipole interactor.
LLE utility is susceptible to many factors such as safety (the risk of explosion,
flammability, toxicity and carcinogenicity) and the technique's influence on
precision and recovery. LLE is also labor intensive, difficult to automate and
connect in-line with analytical instruments, and there are problems of analyte
adsorption on glassware or solvent mediated decomposition. Furthermore,
impurities in the solvent can cause interferences, emulsion formation increases
the complexity and decreases recovery and there are the delays engendered in
the often lengthy evaporation stage. Details of the LLE technique are also
described by F. Cantwell in Chapter 11.
In spite of these several drawbacks, LLE is widely used. With LLE, it is
possible to efficiently achieve both clean-up and enrichment for many applica-
tions. Selected applications of LLE techniques reported recently are summa-
rized in Table 23.7. Hexane, chloroform, dichloromethane, diethyl ether, ethyl

801
TABLE 23.7
Selected applications for drug analysis using liquid-liquid extraction
Analyte Matrix Extraction solvent Analytical Ref.
method*
Thiopentane Plasma Pentane HPLC-UV 112
Sameridine Plasma Hexane GC-NPD 113
Eliprodil Plasma, urine Hexane HPLC-FL 114
Methylenedioxy Serum, tablet Hexane HPLC-FL 115
amphetamine
Tricyclic antidepressants Serum, plasma Hexane HPLC-UV 116
Candidates &metabolite Plasma Hexane LC/MS/MS 117
Propranolol, furosemide Plasma Hexane/diethyl ether HPLC-UV 118
HIV protease inhibitors Plasma Hexane/ethyl acetate HPLC-UV 119
Deramciclane &metabolite Plasma Hexane/diethyl ether LC/MS 120
Atovaquone Plasma Hexane/isoamyl alcohol HPLC-UV 121
Amitriptyline, Serum Hexane/isoamyl alcohol GC-NPD 122
noramitriptyline
Citalopram Plasma Hexane/isoamyl alcohol HPLC-FL 123
Clozapine &metabolite Plasma Hexane/isoamyl alcohol HPLC-UV 124
Clemastine Plasma Toluene GC-NPD 125
Disopyramide &metabolite Plasma Toluene CE-UV 126
Dextrophane, guaifenesin Plasma Chloroform HPLC-FL 127
Lamotrigine Serum Chloroform HPLC-UV 128
ThioTEPA, Plasma Chloroform GC-NPD 129
cyclophosphamide
Flunitrazepam Hair Chloroform/ diethyl ether GC/MS 130
Nalbuphine Plasma Chloroform/isopropanol HPLC-ECD 131
Bezitramide &metabolite Urine Chloroform/isopropanol GC-NPD 132
Ranitidine Serum Dichloromethane HPLC-UV 133
Cetirizine Plasma Dichloromethane HPLC-UV 134
Albendazole &metabolite Plasma Dichloromethane CE-UV 135
Disopyramide &metabolite Plasma, urine Dichloromethane HPLC-UV 136
Anabolic steroids Urine Dichloromethane HPLC-DAD 137
Phenmetrazine Urine 1-Chlorobutane GC/MS 138
Benzodiazepines Whole blood 1-Chlorobutane HPLC-DAD 139
Arbidol Plasma tert.-Butyl methyl ether HPLC-UV 140
Losigamone Plasma tert.-Butyl methyl ether HPLC-DAD 141
HIV protease inhibitors Plasma tert.-Butyl methyl ether HPLC-UV 142
Staurosporine Plasma Diisopropyl ether HPLC-FL 143
Triazolam Muscle Diethyl ether GC/MS 144
Efavirenz Plasma Diethyl ether HPLC-UV 145
Docetaxel Plasma Diethyl ether HPLC-UV 146
Vinorelbine Plasma, blood Diethyl ether HPLC-FL 147
Rapamycin Whole blood Diethyl ether/butyl chloride HPLC-UV 148
Cimetidine Plasma Ethyl acetate HPLC-UV 149
Nitrofurantoin Plasma, urine Ethyl acetate HPLC-UV 150
Penem antibiotics Plasma, urine Ethyl acetate HPLC-UV 151
Captopril Plasma Ethyl acetate HPLC-UV 152
continued

802
TABLE 23.7 (continuation)

Analyte Matrix Extraction solvent Analytical Ref.

method*
Ifosfamide &metabolite Plasma, urine Ethyl acetate HPLC-UV 153
Ifosfamide &metabolite Plasma Ethyl acetate GC-NPD 154
Clozapine &metabolite Plasma Ethyl acetate HPLC-UV 155
Seratodas &metabolite Serum, urine Ethyl acetate HPLC-UV 156
Cyclophosphamide Urine Ethyl acetate LC/MS/MS 157
Minocycline Plasma Ethyl acetate HPLC-UV 158
Theophillin &metabolite Plasma, urine Ethyl acetate/isopropanol HPLC-UV 159
Paclitaxel Plasma Acetonitrile/n-butyl chloride HPLC-UV 160
"See list of abbreviations.

acetate, and their solvent mixes are used for the LLE of various drugs. There is
little to distinguish these solvents in terms of their extraction power, although
the polar solvents will often also give a higher background. A serious drawback is
the large consumption of pure solvents. There is usually a considerable cost for
the acquisition and disposal of these solvents and many classical methods
demand chlorinated and/or fluorinated solvents, the use of which is at odds with
current environmental awareness and legislation. Macek et al. [123] used hex-
ane-isoamyl alcohol (98:2 v/v) for the LLE of citalopram from human plasma,
with analysis by HPLC with FL after back-extraction to 0.02 M hydrochloric
acid. Bouley et al. [142] developed a rapid, sensitive, and specific HPLC method
for the simultaneous TDM of HIV-protease inhibitors (indinavir, neldinavir,
ritonavir and saquinavir) in human plasma, after LLE with tert-butyl methyl
ether and a sequential washing of the reconstituted sample with hexane. A
number of flow-system LLE approaches have been presented based on mixing
the aqueous and organic phases in a tube coil and their subsequent separation.
These approaches are not widely used in sample preparation for chromato-
graphic analysis. On the other hand, semi-automated LLE methods using
96-deep-well plates have been developed for drug analysis [161,162]. All liquid
transfers during the sample preparation were automated, and the extraction
time was relatively short. The details of other LLE techniques are also described
in well-documented reviews [13,25-30,94].

23.4.3 Supercritical fluid extraction

Supercritical fluid extraction (SFE) is a rapidly expanding sample preparation

technique that can be automated and interfaced to both chromatography and
mass spectrometry. SFE considerably reduces sample preparation time and
provides efficient analyte recovery from liquid, solid and semisolid samples. The
technique offers unique advantages since it combines liquid-like solvating capa-
bilities with almost gas-like transport properties. Near-zero surface tension

803
contributes to efficient penetration of porous materials. An attractive property
is that the solvent power may be tuned by varying the temperature or pressure of
the media. Most commonly the pressure is varied, allowing the temperature to
be kept low (<40°C), ensuring that no analyte degradation occurs during extrac-
tion. Carbon dioxide or CO2-rich mixtures have been used almost exclusively as
the extraction media, because of ease of manipulation, good solvent strength and
compatibility with solutes. Carbon dioxide (with critical temperature and
pressure of 31°C and 73.8 bar, respectively) is inert, non-toxic, non-flammable,
non-corrosive, odorless and inexpensive [163]. The diffusion properties of super-
critical CO 2 allow the rapid penetration and extraction of samples, and the low
critical temperature allows good stability of thermally labile materials [163].
Hence, the use of hazardous organic solvents can be kept to a minimum in the
SFE procedure [164]. The use of ammonia or hydrocarbons as supercritical
fluids is less convenient and safe than CO2 . The solvating power of supercritical
CO2, particularly for the extraction of polar compounds, may be increased by the
addition of modifier substances, which are usually organic solvents that are
added to the source of compressed fluid before the pump or to the extraction gas
after it is compressed (e.g., CO2 mixed with methanol) [164-166. Gradient
elution, as opposed to the use of pre-mixed solvents gives shorter analysis times
[166,167]. The use of modifiers does limit the choice of detectors although there
are a large number to choose from. For example UV detectors, flame ionization
detectors, nitrogen-phosphorous detectors, electron capture detectors and mass
spectrometers are all commonly employed for these analyses. Adjusting parame-
ters such as pressure, temperature and the amount/composition of modifier can
readily alter the selectivity of the extraction [163] as can the humidity of the
sample. Generally a partial dehydration allows for a faster extraction [167]. The
density of the extracting solvent can also be altered by pressure changes, giving
the advantage of being able to alter the solubility parameter (selectivity). Also
the nature of the matrix material affects the extraction. Generally a rapid and
complete extraction depends upon the matrix particles being small. An increase
in the porosity of the matrix will also lead to improved extraction [167].
To date, most studies involving the SFE of illicit drugs and pharmaceuticals
have been applied to the extraction of biological materials, solid and semisolid
dosage forms and bulk drug substances. Opiates [168], cocaine and its metabo-
lite benzoylecgonine [169] and amphetamines [170] were extracted from hair
samples using CO 2 as the supercritical fluid modified with the polar mixture of
tetraethylamine, methanol and water. Simmons and Stewart [1711 extracted
benzodiazepine, an anabolic agent and a non-steroidal anti-inflammatory from
water and serum using a supercritical CO2 mobile phase. Scott and Oliver [172]
used SFE for the extraction of temazepam from whole blood samples. SFE has
also been used for the extraction of various pharmaceuticals from tablets,
capsules, creams, ointments, aqueous matrices and infusions [173]. Lawrence et
al. [174] employed SFE in extracting benzodiazepines from the matrices of
standard dosage forms. Scalia et al. [175] reported on the potential of SFE for the

804
isolation of vitamins A and E and their esters from tablet matrices. Tena et al.
[176] demonstrated an improved SFE of sulphonamides from solid supports
using supercritical CO 2 and methanol modified CO2 as mobile phases. Howard et
al. [177] used SFE for the sample preparation of sustained release felodipine
tablets. Masuda et al. [178] designed an SFE-supercritical chromatography-UV
system for quantitative analysis of retinal palmitate and tocopherol acetate in a
hydrophobic ointment. Recently, Wang et al. [179] reported an in situ SFE and
chemical derivatization procedure followed by GC/MS for the determination of
amphetamines in urine. Brewer et al. [180] also demonstrated SFE using CO 2
modified with methanol for the extraction of cocaine benzoylecgonine, codeine
and morphine from human hair sample in combination with GC/MS. This
method was faster and gave higher recoveries than the conventional acid
hydrolysis method. The details of other SFE techniques are also described in
well-documented reviews [31,164,173,181].

23.4.4 Solid-phase extraction

Solid-phase extraction (SPE) is a method involving a series of operations to

extract, concentrate and clean up the component of interest from a sample using
an extraction phase chemically bonded silica gel or polymer gel (polystyrene,
polymethacrylate, etc.) in a cartridge. The concepts were devised during the
development process of the HPLC technology, which was accomplished in a
rapid advance in the 1970s. SPE offers the following advantages over conven-
tional liquid-liquid procedures: (1) higher recovery; (2) effective concentration;
(3) less organic solvent usage; (4) no foaming or emulsion problems; (5) shorter
sample-preparation time; (6) easy operation; (7) ease of incorporation into an
automated process. SPE has been widely adopted for the analysis of pharma-
ceuticals and drugs of abuse in biological matrices, and its position is now
ensured as a main-stream means of sample preparation. SPE is based on the
partitioning of compound between a liquid (sample) phase and solid (extraction)
phase, and the intermolecular forces between the phases impact retention and
elution. The fundamental concept of SPE is equivalent to that of column
chromatography. SPE consists of four basic operational steps: conditioning,
loading, rinsing, and eluting. The first solvent for conditioning should be as
strong as or stronger than the elution solvent. The second conditioning solvent
should be the same as or as close to the strength of the loading solvent as
possible. The solvent used for loading should be as weak as possible to result in
the tightest or narrowest band of adsorbed analyte on the sorbent. A solvent that
is slightly stronger or the same strength as the loading solvent is used as the
rinse solvent. The rinse will eluate unwanted sample components that are not as
strongly retained as the analytes and also wash down small droplets of loading
solvent adhering to the walls of the tube to ensure that all sample comes in
contact with the sorbent. An ideal eluting solvent should elute the analytes
within 5-10 bed volumes. The optimal amount of solvent to elute the analytes

805
from a 500-mg cartridge is about 0.6-1.2 ml. Using a solvent that is too strong
will result in the elution of unnecessary sample components that are more
strongly retained than the analytes, whereas a solvent that is too weak will
result in excessive elution solvent volumes, which negate the advantage of
reducing solvent consumption with SPE cartridges. Sometimes, a desired
solvent strength may have to be obtained by blending appropriate amounts of
miscible solvents. If a water-immiscible eluant is selected and the final analysis
is to be performed with GC, a "drying" procedure should be applied between the
rinsing and the eluting steps. It has been reported that the combination of
vacuum and a small amount of methanol can produce a "dry" eluate without
causing a substantial loss of drugs.
The fundamental mechanisms of the retention are based on: (1) nonpolar
interaction, (2) polar interaction and (3) ionic interaction, and a wide selection of
SPE products is possible in the combination of these interactions using commer-
cial SPE sorbents. The primary decision for analysts in SPE method develop-
ment is the selection of the type of sorbent to obtain optimal extraction. Various
SPE sorbents, diatomaceous earth Extrelut®, Chem Elut®, Bond Elut Certify®
and Chromabond® mixed-mode columns, from different manufactures are now
widely used. Mixed-mode sorbents and restricted access matrix sorbents have
also been introduced as well as the emerging selective sorbents such as immuno-
sorbents or molecularly imprinted polymers (see Section 23.4.6). The mixed-
phase extraction columns (Bond-Elut Certify®, Chromabond®, Isolute HCX®
and TSC®) show good recoveries and allow retention of all functional groups
and differing polarities. Bond-Elut Certify® is a mixed phase of C8 and SCX,
with the functions of both hydrophobicity and ion exchange, and it is suitable for
the extraction of basic drugs that have been dissociated, in blood and urine
samples. SPE discs are also a useful and rapid way to extract drugs from liquid
specimens. In the conventional packed type solid phase, it is difficult to elute the
analyte of interest using minial solvent, if the organic solvent composition is not
raised to around 100%. Therefore, it is necessary to evaporate to dryness and
redissolve in the HPLC mobile phase. The Empore® disk cartridge has a
membrane structure, and the elution can be achieved with HPLC mobile phase.
It is then possible to directly inject into the HPLC system. New formats have also
been introduced, e.g., the 96-well SPE plates [182-184] and the microfibers for
SPME (see Section 23.4.5) for drug analysis. Further details of the SPE tech-
nique are also described by Poole in Chapter 12.
SPE can be performed off-line, with the sample preparation being separated
from the subsequent chromatographic analysis, or on-line by direct connection
to the chromatographic system. On-line techniques do not require further
handling of the samples between the trace-enrichment and the separation step
and, therefore are highly suitable for fully automated techniques. With on-line
fully-automatic solid phase extraction equipment, a column switching system is
incorporated, a precolumn is fixed in the sample loop of the injector, and the
pre-concentration is carried out through the load of a comparatively large

806
sample and the retention of the analyte of interest in the solid phase. Next, the
solute is eluted and introduced into the separation column, by switching the
injection valve to the injection position. Special attention for hyphenation has
been given to on-line SPE-HPLC followed with various detection modes, which
represents a fast, modern and reliable approach of trace analysis. Important
advantages are a decrease in the risk of sample contamination, the removal of
analyte losses by evaporation and finally the transfer and analysis of the entirety
of the extracted species. In contrast to off-line SPE where only an aliquot of the
extract is injected into the chromatograph, the analysis of the complete sample
allows the sample volume to be dramatically reduced. It has been surprising to
see that gas chromatography (GC) has been the preferred method for analytical
chemists for a long time but that the on-line coupling of SPE with LC became the
first robust on-line technique. This is explained by the good compatibility of the
LC aqueous mobile phases with the SPE of biological or environmental samples
that are mainly aqueous. On-line coupling of SPE with GC is more difficult
because of the inherent incompatibility between the aqueous nature of the SPE
step and the requirement for dryness in the GC system. Much work has been
done in this area and automatic devices have recently become available. D. Wells
and T.L. Lloyd also describe the details of an automated technique for clinical
and pharmaceutical analysis in Chapter 24.
SPE is applied for isolation, concentration, and clean up of a target substance
from a biological specimen, which may be a complicated matrix like serum,
urine, saliva, tissues, etc. Selected applications of SPE techniques reported
recently are summarized in Table 23.8. C18 is the most popular sorbent for drug
analysis. Prieto et al. [193] used C8 and C18 SPE cartridges for the extraction of
the antihypertensive drug cilazapril and its active metabolite cilazaprilat from
pharmaceuticals and urine samples, with analysis by HPLC-DAD or ED. Fluni-
trazepam (Rohypnol), and its metabolites were extracted with RP-18 [196] and
Bond-Elut Certify® [242] from plasma and urine samples, respectively. Bevalot
et al. [221] reported a LC/MS method for the analysis of doping agents and
corticosteroids in hairs, by methanolic extraction followed by SPE on a C18
cartridge. Cocaine and other opiates were also extracted from hair [239], plasma
and urine samples [238] using Bond-Elute Certify®. In addition, SPE was
applied for the clean-up of milk samples [206,220,240]. The details of other SPE
techniques are also described in well-documented reviews [26-37,94-106].

23.4.5 Solid-phase microextraction

Solid-phase microextraction (SPME), developed by Pawliszyn and co-workers in

1990, is a new sample preparation technique using a fused-silica fiber that is
coated on the outside with an appropriate stationary phase. Analyte in a sample
is directly extracted to the fiber coating. In contrast to conventional SPE with
packed-bed columns, micro or non-micro columns, this arrangement allows the
combination of all steps of sample preparation into one step. The method saves

807
TABLE 23.8
Selected applications for drug analysis using solid-phase extraction
Analyte Matrix SPE sorbent* Hyphenated Ref.
analysis*
Almokalant Plasma C2 HPLC-FL 185
Haloperidol, perphenazine Serum C2 HPLC-UV 186
Citalopram Plasma C4 HPLC-FL 187
Pyrimethamine, Blood, urine C8 HPLC-UV 188
sulphadoxine
Modafinil Plasma C8 HPLC-UV 189
Cilagapril &metabolite Pharmaceuti, cal, urine C8 HPLC-ED 190
Atovaquone Plasma, who]le blood C8 HPLC-UV 191
Olanzapine &metabolite Plasma C8 HPLC-ED 192
Cilazapril &metabolite Pharmaceuti cals, C8 / C18 HPLC-ED, 193
urine DAD
Verapamil, norverapamil Urine C8 / C18 HPLC-FL 136
Gilbendamide Plasma RP-8 / RP-18 HPLC-UV 194
Flunitrazepam & Plasma RP-18 HPLC-UV 195
metabolite
Anabolic steroids Urine C18 HPLC-UV 196
Benzodiazepines Plasma, serum C18 HPLC-DAD 197
Illicit drugs Urine C18 LCMS 198
Fluconazole Plasma C18 MECC-UV 199
P-Blockers Blood, urine C18 GC/MS 200
Drugs Hair C18 HPLC-DAD 201
Methotrexate &metabolite Urine C18 HPLC-UV 202
Doxorubicin, Plasma C18 HPLC-FL 203
prochlorperazine
Fluticasone propionate Plasma C18 LC/MS 204
Iornoxicam Plasma C18 HPLC-UV 205
Phenytoin Plasma, milk C18 HPLC-UV 206
HIV-protease inhibitors Serum C18 HPLC-UV 207
Budesonide Plasma C18 LC/MS/MS 208
Cefotaxime Pharmaceutica]Is, C18 HPLC-UV 209
urine
Sirolimus Blood C18 LC/MS/MS 210
Opiate agonists Serum, blood, urine C18 HPLC-UV 211
HIV-protease inhibitors Plasma C18 HPLC-UV 212,
213
Mefloquine &metabolite Plasma. Serum, blood C18 HPLC-UV 214
Chlorambucil Plasma, serum C18 LCMS 215
Olanzapine Serum C18 LC/MS 216
Omeprazole Serum C18 HPLC-UV 217
Caffeine Medicinal prescription C18 HPLC-UV 218
Vitamins Pharmaceuticals C18 HPLC-UV 219
Carbamazepine & Plasma, milk C18 HPLC-UV 220
metabolite
Corticosteroids Hair C18 LC/MS 221
Retinoids Pharmaceuticals C18 HPLC-FL 222
continued

808
TABLE 23.8 (continuation)

Analyte Matrix SPE sorbent* Hyphenated Ref.

analysis*
Tacrolimus, sirolimus Blood C18 LC/MS 223
Tramadol &metabolite Plasma C18 HPLC-UV 224
Vincristine Plasma C18 HPLC-ED 225
Ketorolac Plasma PH HPLC-UV 226
Anti-inflammatory drugs Urine Chromosorb P GC/MS 227
Illicit drugs Serum, urine Chromabond mixed FIA-MS/MS 228
mode
Pranlukast &metabolite Plasma PROSPEKT LC/MS/MS 229
Quinfamide Plasma, urine, faces Sep-Pack HPLC-UV 230
Aconitum alkaloids Blood, urine Sep-Pack PHPS-1 HPLC-UV 231
THC, THC-COOH Plasma, urine CBA/CH HPLC-ED 232
Abused drugs Urine Bakerbond NARC-2 GC/MS 233
Domperidone Plasma Nitrile SPE HPLC-FL 234
Hydroxytyrosol Plasma Oasis HLB HPLC-DAD 235
copolymer
Mefenamic acid, flufenamic Pharmaceuticals, SAX HPLC-UV 236
acid, telfenamic acid urine
Ribavirin Serum PBA HPLC-UV 237
Cocaine, benzoylecgonine Plasma, urine Bond-Elute Certify HPLC-UV 238
Cocaine, opiates & Hair Bond-Elute Certify GC/MS 239
metabolites
Nefazodone &metabolite Plasma, milk Bond-Elute Certify HPLC-UV 240
LSD Urine Bond-Elute Certify GC/MS/MS 241
Flunitrazepam & Urine Bond-Elute Certify GC/MS 242
metabolite
Raclopride Plasma SPE disk LC/MS 243
Tolterodine &metabolite Plasma Kromosil C18 SPE-MS 244
capillary
*See list of abbreviations.

preparation time, solvent purchase and disposal cost, and can improve the
detection limits. It has been used routinely in combination with gas chromatog-
raphy (GC) and GC/mass spectrometry (GC/MS), and successfully applied to a
wide variety of compounds, especially for the extraction of volatile and semi-
volatile organic pollutants from water samples. SPME was also introduced for
direct coupling with high performance liquid chromatography (HPLC) and
LC/MS in order to analyze weakly volatile or thermally labile compounds not
amenable GC or GC/MS. An SPME/HPLC interface equipped with a special
desorption chamber is utilized for solvent desorption prior to HPLC analysis in
place of thermal desorption in the injection port of a GC. Moreover, a new
SPME/HPLC system known as in-tube SPME, was recently developed using an
open tubular fused silica capillary as the SPME device instead of the SPME fiber.
The main advantages of SPME are simplicity, rapidity, solvent elimination,
high sensitivity, small sample volume, lower cost and simple automation. Since

809
1995, a number of SPME methods have been developed to extract drugs from
various biological samples such as urine, serum, plasma, whole blood, saliva and
hair (Table 23.9). The fiber SPME device consists of a fiber assembly with a
built-in extraction fiber inside the needle and an assembly holder. In fiber
SPME, analytes are extracted directly from the sample onto a polymeric station-
ary phase coated on the fiber. When the fiber is inserted into the sample, the
target analytes partition from the sample matrix into the stationary phase until
equilibrium is reached. Two types of fiber SPME techniques can be used to
extract analytes: headspace (HS) SPME and direct immersion (DI) SPME. In
HS-SPME, the fiber is exposed in the headspace of gaseous, liquid or solid
samples. In DI-SPME, the fiber is directly immersed in liquid samples. Minor
variants in the method depend on whether or not derivatization is applied and in
which phase, the type of sample agitation and whether or not additives are
required to optimize extraction. The fiber with concentrated analytes is trans-
ferred to an instrument for desorption, followed by separation and quantitation.
HS- and DI-SPME techniques can be used in combination with any GC, GC/MS,
HPLC and LC/MS systems. The affinity of the fiber coating for an analyte is the
most important factor in SPME. As shown in Table 23.9, fiber coatings of
different polarities and thicknesses were selected for different drugs. Most drugs
in biological samples were extracted with 100 gm PDMS for nonpolar drugs and
either 85 gm PA or a porous polymer DVB fibre for polar drugs.
In-tube SPME using an open tubular capillary as the SPME device has been
used for coupling with HPLC or LC/MS. It is suitable for automation, and extrac-
tion, desorption and injection are performed continuously using a standard
autosampler. Automated sample handling procedures not only shorten the total
analysis time but also usually provide better accuracy and precision relative to
manual techniques. With the in-tube SPME technique, organic compounds in
aqueous samples are directly extracted from the sample into the internally
coated stationary phase of a capillary and then desorbed by introducing a moving
stream of mobile phase or static desorption solvent when the analytes are more
strongly absorbed to the capillary coating. Desorbed compounds are finally
injected onto the column for analysis.
Although the theories of fiber and in-tube SPME methods are similar, the
significant difference between these methods is that the extraction of analytes is
performed on the outer surface of the fiber for fiber SPME and in the inner sur-
face of the capillary for in-tube SPME. Commercially available SPME fibers for
drug analysis are limited, but GC capillary columns with a vast array of station-
ary phases are commercially available for in-tube SPME. Headspace fiber SPME
is suitable for the extraction of drugs in gaseous, liquid and solid samples, and
has the advantage of avoiding contact with an aggressive matrix incompatible
with the fiber. Direct immersion fiber SPME can be used to extract drugs from
both clear and cloudy liquid samples, whereas in-tube SPME is limited to the
extraction of clear liquid samples. Therefore, the headspace SPME technique is
suitable for direct extraction from whole blood samples, but direct immersion

810
TABLE 23.9
Selected applications for drug analysis using solid-phase microextraction
Analyte Matrix SPME device Extractn. Hyphenated Ref.
modea analysis*
Anorectics Urine 30 m PDMS DI GC/MS 246
Cannabinoids Hair 30 gm PDMS DI GC/MS 247
Cannabinoids Saliva 30 m PDMS DI GC/MS 248
Amphetamines Blood 100 m PDMS HS GC/MS 249
Amphetamines Urine 100 gm PDMS HS GC/MS 250
Amphetamines Urine 100 m PDMS HS GC-FID 251
Amphetamines Hair 100 gm PDMS HS GC-NPD 252
Amphetamines Hair 100/Am PDMS HS GC/MS 253
Amphetamines Hair 100 m PDMS D + HS GC/MS 254
Amphetamines, MDA, Urine 100 m PDMS HS GC/MS 255-
MDMA, MDEA 257
Amphetamines Urine 100 m PDMS D + DI GC-NPD, 258
GC/MS
Amphetamines, Urine 100 m PDMS DI GC/MS 259
MDMA
Amphetamines, MDA, Urine 100 m PDMS D + DI GC-NPD, 260
MDMA, MDEA GC/MS
Amphetamines, Whole blood 100 m PDMS D + HS GC/MS 261
fenfluramine
Methamphetamine, Urine 100,um PDMS DI GC-NPD, 262
cocaine, diazepam GC/MS
Anaesthetics Blood 100 m PDMS HS GC-FID 263
Local anaesthetics Blood 100 m PDMS HS GC/MS 264
Anaesthetics Blood 100 m PDMS DI GC-FID 265
Phencyclidine Blood, urine 100 m PDMS HS GC-SID 266
Halothane Blood 100,um PDMS HS GC/MS 267
Lidocaine Plasma 100 m PDMS DI GC-FID 268
Lidocaine Plasma 100 m PDMS DI GC-FID 269
Lidocaine Urine 100 m PDMS DI HPLC-UV 270
Tricyclic antidepress. Blood 100 am PDMS HS GC-FID 271
Tricyclic antidepress. Blood 100 am PDMS HS GC/MS 272
Tricyclic antidepress. Plasma 100 m PDMS DI GC-NPD 273
Tricyclic antidepress. Urine 100 m PDMS HS GC-FID 274
Amitriptyline Urine 100 am PDMS DI HPLC-UV 275
Valproic acid Plasma 100 m PDMS DI GC-FID 276
Diphenylmethane Blood, urine 100 m PDMS HS GC-FID 277
antihistaminics
Phenothiazines Blood, urine 100 m PDMS HS GC-FID 278
Clozapine Plasma 100 m PDMS DI GC-NPD 279
Cocaine Urine 100 gm PDMS DI GC-NPD 280
Meperidine Blood, urine 100 m PDMS HS GC-FID 281
Benzoylecgonine Urine 100 am PDMS D + HS GC/MS 282
Cannabinoids Saliva 100 am PDMS DI GC/MS 283
Methadone Hair 100zm PDMS DI GC/MS 284
Nicotine, cotinine Urine 100gAm PDMS HS GC/MS 285
continued

811
TABLE 23.9 (continuation)

Analyte Matrix SPME device Extractn. Hyphenated Ref.

modea analysis*
Benzodiazepine & Urine 100/am PDMS, 65 m DI GC-ECD 286
metabolites PDMS/DVB
Amphetamines Urine 65 Am PDMS/DVB DI GC-NPD 287
Amphetamine, Tablet 65 Am PDMS/DVB DI GC-FID, GC- 288
MDMA NPD, GC/MS
Benzodiazepines Urine 65 m PDMS/DVB DI GC-FID 289
Methadone & Hair 65 m PDMS/DVB HS GC/IMS 290
metabolite
Nitrous oxide, Urine 50-30 mm HS GC/MS 291
isoflurane, halothane DVB-carboxen/PDMS
Illicit drugs Serum, urine 85 m PA HS GC/MS 292
Benzodiazepines Plasma, urine 85,um PA DI GC-NPD 293
Benzodiazepines Urine 85 m PA DI LC/MS 294
Diazepam Plasma 85 m PA DI GC-NPD 295
Estrogens Serum 85am PA DI + D GC/MS 296
8
Phenothiazines Blood, urine 5 Am PA DI LC/MS 297
Anabolic steroids, Urine 85 m PA, 65am DI + D GC/MS 298
estrogens CW/DVB
Benzophenone & Urine 65 Am CW/DVB DI GC/MS 299
metabolite
Lidocaine Hair 65am CW/DVB HS GC/MS 300
Barbiturates Urine 65 m CW/DVB DI GC/MS 301
Corticosteroids Urine 65 m CW/DVB DI LC/MS 302
Benzodiazepines Serum, urine 65am CW/DVB DI GC-FID, 303
GCfMS
Tetracyclines Milk 50,um CW/TPR DI LC/MS 304
Benzodiazepines Urine 50 m CW/TPR DI HPLC-UV 305
Barbiturates Urine PVC-coated DI CE-UV 306
Theophylline Serum Antibody-coatec Id DI HPLC-UV 307
Amphetamines, MDA, Urine Omegawax 250 IT LC/MS 308
MDMA, MDEA
Ranitidine Urine, tablet Omegawax 250 LC/MS 309
D-Blockers Serum, urine Omegawax 250 LC/MS 310
Drugs Serum, urine Omegawax 250 LC/MS 311
3-Blockers Serum, urine PPY-coated LC/MS 312
Amphetamines, MDA, Urine, hair PPY-coated LC/MS 313
MDMA, MDEA
Benzodiazepines Serum Supel-Q plot IT LC/MS 314
Tricyclic antidepress. Urine Wire-packed DI 3-1 IT HPLC-UV 315
Propranolol Serum MIP-packed IT HPLC-UV 316
aHS: headspace; DI: Direct immersion; IT: in-tube; D: derivatization.
*see List of abbreviations.

fiber SPME or in-tube SPME methods generally require deproteinization or

ultrafiltration of samples prior to extraction. The extraction efficiency of fiber
SPME depends on extraction time, agitation, heating, sample pH and salt con-

812
centration. For in-tube SPME, number, volume and speed of draw/eject cycles,
and sample pH are important factors for efficient extraction. On the other hand,
the desorption of analyte from a fiber or capillary coating depends on the tem-
perature of injection port and exposure time in combination with GC or GC/MS,
or component and volume of solvent when used in combination with HPLC or
LC/MS. Therefore, these SPME parameters should be optimized when develop-
ing a new SPME method for drug analysis. SPME principles and optimization
strategies are also described by J. Pawliszyn in Chapter 13.
There are numerous references to the SPME analysis of drugs from biologi-
cal fluids and matrices. A summary of references is given in Table 23.9. Most of
the applications have been related to forensic and toxicological analysis; how-
ever, they demonstrate the potential of SPME for other drug analyses, such as
clinical, metabolic and pharmaceutical. Most of the methods shown to date have
involved HS techniques for volatile drugs. Some methods have employed DI and
derivatization for less volatile analytes. It has provided low detection limits and
excellent quantitation. Especially in the HS mode, SPME extractions offer the
potential for very clean analyses, with little to no interference from non-volatile
compounds. Because of the relatively low partition coefficients between polar
drugs and the commercially available HPLC/SPME fibres, the application of
SPME for the assay of low volatility drugs and metabolites in plasma may be
limited to those drugs with high therapeutic concentrations, in the range of 1 to
100 ,ug/ml. The situation can be improved with the use of current tandem
quadrupole LC/MS instrumentation, where concentrations approaching 1 ng/ml
in plasma can be analysed [245]. The analysis of drugs in plasma for pharmaco-
kinetic studies and TDM is an important area of biomedical analysis. Currently
however there are few practical SPME/HPLC methods available. The bulk of the
SPME applications in drug analysis to date have involved GC analysis. Amphet-
amines were extracted from urine [250,251,255-257], hair [252-254] and blood
[249] by HS-SPME with PDMS fiber under alkaline conditions. Lidocaine was
extracted by DI-SPME with PDMS fiber from plasma and urine, and analyzed by
GC-FID [268,269] and HPLC-UV [270]. PDMS/DVB [289], PA [293,294] and
CW/DVB [303] fibers were used for the extraction of benzodiazepines in plasma,
serum and urine samples by DI-SPME. Benzodiazepines were also analyzed in
combination with DI-SPME with CW/TPR and HPLC-UV [305]. Recently Yuan
et al. [307] developed an immunoaffinity (IE)-SPME technique for the specific
extraction of theophylline in serum sample. The IE-SPME fiber, covalently
immobilized a theophylline antiserum on the surface of a fused silica fiber, was
used as a selective and sensitive extraction medium. For automated SPME/
HPLC analysis, Kataoka et al. [308-311] reported automated on-line methods
for the analysis of amphetamines, ranitidine and 3-blockers by in-tube SPME
coupled with LC/MS. In these methods, Omegawax 250 capillary (Supelco,
Bellefonte, PA) was used as the extraction device. Wu et al. [312] later demon-
strated improvements obtained for the extraction of a series of P-blocker drugs
in urine and serum by the use of novel polypyrrole (PPY) polymers as compared

813
to the conventional Omegawax GC phase with in-tube. Saito et al. [315] used a
DB-1 capillary for the in-tube SPME of tricyclic antidepressants in urine, and
extraction efficiency was raised by insertion of stainless steel wire into the
capillary. Mullet et al. [316] recently developed a new in-tube SPME technique
using a polyether ether ketone (PEEK) capillary packed with propranolol MIP
particles. The details of other SPME techniques are also described in books [17]
and well-documented reviews [36-46,94,101-106].

23.4.6 Affinity sorbent extraction

23.4.6.1 Immunoaffinity extraction

Immunoaffinity extraction (IAE) is a method for selectively extracting the
chemical compound of interest using an antibody immobilized to an insoluble
carrier, based on molecular recognition. When the biological specimen is added
to the immunosorbent, the target compound is specifically retained to the
immobilized antibody based on the principle of the antigen-antibody binding.
After washing and removal of contaminants, the purified and concentrated
compound is obtained by the dissociation of the complex. Extraction and clean-
up are achieved in the same step from complex biological aqueous samples, and
from large volumes when required. Single analytes can be targeted, but since a
polyclonal antibody can also bind one or more analytes having structures similar
to the original immunogen, immunosorbents have been developed for targeting
both a single analyte and its metabolites. The cross-reactivity of monoclonal
antibodies has also been exploited for developing immunosorbents that select-
ively extract a class of structurally related compounds, with the advantage that
cross-reactivities are well characterized and reproducible, as discussed below.
The IAE of antibodies, hormones, peptides, enzymes, recombinant proteins,
viruses and subcellular components have been widely described because of the
ease of their preparation and the availability of antibodies which can be very
selective for large molecules [317]. Small molecules are usually poorly antigenic.
Obtaining selective antibodies for small molecules has been more difficult and
the development of immunochemical methods in the SPE field targeting low
molecular weight analytes is rather recent [317-320].
The binding of analyte to antibody is the result of a good spatial comple-
mentarity and is a function of the sum of several intermolecular interactions.
Therefore, an antibody can also bind one or more analytes with a structure simi-
lar to the target analyte that was used to induce the immune response. This
cross-reactivity of antibodies was initially considered as a negative feature for an
immunoassay, but it has been more recently exploited in extraction, so that
immunosorbents have been made for single analytes, single analytes and associ-
ated metabolites, or a class of structurally related analytes [318]. Commercial
immunosorbents have been introduced during the last decade for the clean-up of
samples, for the analysis of natural food contaminants such as aflatoxins,
ochratoxins and fumonisins or veterinary drugs such as clenbuterol. In addition

814
to the selectivity of IAE, an important parameter is the extraction recovery,
which is linked to the breakthrough volume and to the capacity. In order to
achieve low detection limits the extraction sorbent must allow the precon-
centration of analytes from large sample volumes without breakthrough of the
analyte. Since the breakthrough volume depends on the total number of accessi-
ble binding sites and on the strength of the interaction between the analytes and
the antibodies, the selection of the materials used for the covalent bonding has
become of prime importance. These should allow a high capacity and avoid any
kind of non-specific interactions with the analytes. Immunosorbents can be used
in off-line procedures in disposable cartridges. A minimum desorption volume
should be used in order to obtain a high enrichment factor. This cannot be
achieved using the conventional aqueous elution solutions of affinity chromatog-
raphy. Organic solvents are required for the elution of small molecules. There is
also interest in coupling extraction on-line with LC [321,322]. Pichon et al. [318]
developed several silica-based immunosorbents and demonstrated that a single
immunoprecolumn could be directly desorbed on-line using a water-organic sol-
vent mixture, and was easily regenerated for re-use [318,320,321,324]. It has
been demonstrated that the coupling of immunoaffinity SPE in off-line or
on-line procedures is a powerful technique for the determination and easy quan-
tification at low levels of many organic compounds in various complex samples.
The sample handling step is greatly simplified. The selectivity in the extraction
step also allows easier identification of analytes of interest and their confirma-
tion is reinforced since trapping occurs on the basis of structure recognition.
Moreover, a clear baseline enhances the overall sensitivity of the analysis and
allows the handling of a smaller sample in comparison to non-selective extrac-
tion procedures. Hennion and Pichon describe the details of IAE in Chapter 33.
IAE coupled on-line with LC/MS has been described [325-327]. Cai and
Henion [326] have reported an automated on-line clean-up and enrichment
procedure using a commercially available workstation to perform the complete
process. Three different columns were used in the column switching system: an
IAE column packed with the bound antibodies, a trapping column and an
analytical column. The work-up process consisted of the following steps: precon-
ditioning of the IAE column, loading of the urine sample, elution of the unbound
matrix to waste, equilibration of the trapping and analytical columns with
mobile phase, elution of the bound analytes and trapping in the trapping
column, back-flushing of the trapped analytes onto the analytical column,
separation and introduction into the MS. The advantage of such a procedure is
not only the full automation but also the highly selective and sensitive detection.
For example, the LOD for the IAE-LC-LC-MS-MS detection of lysergic acid
diethylamide (LSD) in urine was 20-fold below that using off-line SPE and
LC-MS-MS [325]. IAE was also applied to LSD analysis in whole blood [328] and
hair [329] samples. IAE units (LSD ImmunoElute®) are commercially available
for the analysis [329]. Rashid et al. [330] developed morphine-specific immuno-
sorbents to concentrate morphine from urine samples prior to HPLC analysis

815
with electrochemical detection. IAE columns demonstrated remarkably high
specificity towards morphine, showing minimal binding with other opiates such
as codeine, normorphine, norcodeine, morphine-glucuronide. Recently, Clarke
et al. [331] reported ultrafast IAE of warfarin by a 2.1-mm-i.d. sandwich micro-
column that contained a 1.1-mm layer of an anti-warfarin antibody support. In
addition, the IAE technique was applied to specific clean-up of anabolic steroids
[332], P-2 agonists [333] and fluoroquinolines [334] from biological samples.
More recently, IA-SPME for theophylline analysis [307] was also developed as
unique IAE application. The details of other IAE techniques are also summa-
rized in well-documented reviews [97,107,108,335,336].

23.4.6.2 Molecularly imprinted polymer-based extraction

The technique of molecular imprinting (MI) is a template polymerization meth-
od for the synthesis of a polymer in the presence of a target molecule. It is
possible to construct specific binding sites for the target molecule in the growing
polymer. The MI method is fundamentally based on the following three proce-
dures. In the first step, the imprint molecule is combined with the functional
monomer (a bifunctional monomer having functional groups that interact with
both the target molecule and the polymerizable vinyl group) by covalent or
noncovalent bonding. In the process an imprint molecule-functional monomer
complex is formed. In the second step, crosslinking agent and initiator are added
to the solution including the imprint molecule-functional monomer complex,
and the polymerization reaction is completed. As a result, an insoluble polymer
molecule is produced. Crosslinking agents are necessary to maintain the comple-
mentation with the imprint molecule and polymer monomer during the poly-
merization reaction. In the third step, the imprint molecules are removed from
the polymer, and the specific binding sites are exposed. These specific binding
sites exist both internally and externally to the polymer network with a comple-
mentarity for the shape and functional groups of the imprint molecule. Thus the
binding sites are formed at a distance and angle that are optimum for the
re-combination of the imprint molecule and functional groups of the monomer.
The imprint enables the polymer to selectively rebind the imprint molecule from
a mixture of closely related compounds. This organic polymer with artificial
functionality prepared by the MI, method is called a molecularly imprinted
polymer (MIP). In theory MIP for a target molecule can be easily synthesized by
the three steps described above, and this synthetic molecular recognition should
be possible for nearly every small molecule. MIPs provide a combination of
mechanical and chemical robustness with highly selective molecular recognition
comparable to biological systems. The simple and rapid MIP synthesis is particu-
larly suited for low-molecular weight compounds, whereas antibody preparation
for these so-called haptens requires conjugation of the hapten to a carrier
protein before immunization by injection into an animal.
The selectivity of the MIP can be pre-determined by the choice of template
employed for its preparation. This tunable selectivity is a major benefit of

816
MIP-SPE sorbents, leading to efficient sample clean-up. The selectivity of the
extraction leads to distinctly cleaner chromatographic traces and the ability to
improve sensitivity by extraction from larger sample volumes. The analytical
performance of the MIP-SPE based method was found to be equivalent to or
better than that of a standard method based on the use of LLE for sample
clean-up [337].
As a disadvantage, MIPs are made in the presence of large amounts of
template molecules and small amounts of the imprint molecule remaining in the
resultant polymer may later leak during extraction. This has been observed in
several cases. Hence, method development must include a confirmation that
leakage of remaining traces of the imprint species does not interfere with the
assay, and increase uncertainty in the concentration determination. This is
particularly important when dealing with trace analysis. One approach to com-
pletely avoid this risk is the use of a close structural analogue of the analyte(s) of
interest for the preparation of the MIP. Provided the imprint species and the
analyte(s) can be separated by the subsequent LC or GC, which in most
instances is a valid assumption, the leakage appears as a separate peak and
presents no problem. A limitation of MIP based SPE is somewhat of a lack of
knowledge in the use of MIPs for biological samples. MIP preparation entails the
use of organic solvents and, in consequence, most studies on rebinding to
imprints have been conducted using organic solvents as the incubation medium.
A key success factor for aqueous rebinding is the ability to balance specific
binding to the imprints and non-specific binding to the polymer by hydrophobic
interactions. For each compound, (both analyte and all other components of the
sample), the observed retention is due to the sum of specific and nonspecific
binding. Hence, if the non-specific element dominates, any selectivity shown by
the imprints will be obscured. Problems with non-specific adsorption to the
polymer can be reduced by the use of small amounts of MIP, thereby reducing
the polymer surface area available for non-specific adsorption. Another means of
reducing problems with non-specific adsorption is the use of proper washing
schemes prior to elution. The rationale is that the selective imprint analyte
binding increases in strength and non-specific adsorption of hydrophobic nature
is weakened. This leads to redistribution of non-specifically bound analyte to
imprint sites and washing off of non-related structures.
MIPs are capable of molecular recognition and are stable for long-term
storage, easy to prepare and inexpensive, thus they may be considered as new
artificial affinity media. Different modes of MIP based SPE have been demon-
strated, including various modes of off-line and on-line SPE for pre-concentra-
tion or pre-treatment of analytes, conventional SPE where the MIP is packed
into columns or cartridges, and batch mode SPE where the MIP is incubated
with the sample. Several applications for the use of MIP for biological samples
have now appeared. The first MIP-SPE was reported for pentamidine, which is
used for the treatment of AIDS-related pneumonia [338]. The high selectivity of
the polymer was demonstrated with on-line sample enrichment of a urine

817
sample spiked with pentamidine. In this system, 20% (v/v) phosphate buffer in
acetonitrile was used for both loading and elution, and the adsorption (pH 5) and
desorption (pH 3) were accomplished by changing the pH of the buffer. Thus
electrostatic interaction is the main factor for the specific MIP extraction of
pentamidine. MIP-SPE of sameridine [339] and bupivacaine [340], which are
provided for anesthesia during surgery and prolonged post-operative analgesia,
was reported for batch-wise pre-concentration from plasma samples prior to GC
analysis. A non-steroidal estrogen antagonist, tamoxifen, was extracted from
biological samples by MIP-SPE [341]. As another example, this technique has
been extended to the extraction of theophylline in serum [342,343], phenytoin in
plasma [344], clenbuterol in urine [345-347] and propranolol in biological
samples [348,349]. More recently, magnetic MIP beads for drug radioligand
binding assay [350], MIP-membrane SPE for terbumeton extraction [351], and
MIP-SPME for propranolol analyses [316] were also developed as unique MIP
applications. The details of other MI techniques are also summarized in well-
documented reviews [337,352-354].

23.4.7 Membrane-based sample preparation

Membrane-based sample preparation is widely used in drug analysis, primarily

because it allows continuous operation. There are a number of different mem-
brane techniques, which have been suggested as alternatives to the SPE and
LLE techniques [109]. A wide variety of membrane materials can be used, each
with its own advantages and disadvantages [355]. In many cases, a membrane is
a porous network of a synthetic polymer, such as polypropylene, polysulfone or a
cellulose derivative. Separation with these membranes is based on size-
exclusion: sufficiently small molecules can permeate through the pores, whereas
larger ones cannot. More selectivity can be obtained with other membrane types
such as ion-exchange membranes, in which positively or negatively charged
groups are covalently attached to the polymeric membrane material. Separation
with these membranes is not only based on a difference in size, but also on the
fact that ionic compounds with the same charge as the membrane ions are
excluded. Nonporous membranes consist of a liquid or polymer film, in which a
molecule must actually dissolve in order to be able to pass through. In this case,
the efficiency of membrane transport for a particular compound is largely
dependent on its partition coefficients between the different parts of a mem-
brane separation system. Only compounds which are easily extracted from the
donor phase into the membrane and from the membrane into the acceptor phase
will be transported. Therefore, the separation of different compounds is based
on the same principle as a liquid extraction followed by a back-extraction.
Molecules with different physicochemical properties can be separated, even if
they are of equal size. For sample preparation purposes, four membrane extrac-
tion techniques are of importance. Three techniques make use of porous mem-
branes: dialysis, a concentration-driven process, electrodialysis, an electrically

818
driven process and filtration, a pressure-driven process. The fourth technique
uses nonporous membranes and will be referred to as membrane extraction.
In porous membrane techniques the liquids on each side of the membrane
are physically connected through the pores. These membranes are used in
Donnan dialysis to separate low-molecular-mass analytes from high-molecular-
mass matrix components, leading to an efficient clean-up but no discrimination
between different small molecules. No enrichment of the small molecules is
possible, instead the analytes are diluted as the driving force of the mass transfer
process is a simple concentration difference across the membrane. Dialysis is
widely used for protein concentration, etc., in biochemistry, but has not seen
much use in sample preparation for chromatography. As one example, in the
ASTED® (automated sequential trace enrichment of dialysates) process, both
clean-up and enrichment can be performed by a combination of dialysis and SPE
and is applied to drug analysis in blood plasma. Other variations of porous
membrane techniques are microdialysis [356], which is extensively used in
neuroscience research for in vivo sampling, and electrodialysis, where an electric
field over a dialysis membrane promotes selective transport of charged com-
pounds. Lange et al. [357] reported the intracerebral microdialysis technique for
monitoring of free drug concentrations in brain extracellular fluid, in which the
knowledge of associated free drug concentrations provide information on blood-
brain barrier drug transport. Furthermore, the dialysis sampling technique was
applied to the pharmacokinetic studies of unbound cefepime in rat bile [358] and
unbound cephaloridine in rat blood [359]. In addition, a number of micro- and
nanofiltration techniques belong to the field of porous membrane techniques. In
contrast, a nonporous membrane is a liquid or solid (e.g., polymeric) phase that
is placed between two other phases, usually liquid but sometimes gaseous. One
of these phases is the sample to be processed, the donor phase. On the other side
of the membrane is the acceptor phase, where the extracted analytes are
collected and transferred to the analytical instrument. This arrangement
permits the versatile chemistry of LLE to be used and extended, which can
provide a highly effective clean-up as well as high enrichment factors. The
operations can normally be automated and in most cases there is no, or an
insignificant, use of organic solvents. The details of membrane techniques are
also described by J. Pawliszyn (Chapter 14), J.A. Jonsson (Chapter 15), T.
Kotiaho and F.R. Lauritsen (Chapter 16), and T. Matsuura (Chapter 30).
Membrane techniques have been applied to the determination of various
drugs in biological fluids. Selected applications of membrane-based sample
preparation techniques reported recently are summarized in Table 23.10. On-
line dialysis has been applied in a fully automated HPLC method for analysis of
the non-ionic X-ray contrast agent iodixanol in plasma [360]. Continuous moni-
toring of free (i.e., non protein bound) drug concentrations in different body
compartments by microdialysis represents an important step towards correct
interpretation of data in pharmacokinetic or toxicokinetic studies. This samp-
ling methodology is undoubtedly an in vivo experimental alternative to the

819
TABLE 23.10
Selected applications for drug analysis using membrane-based separation techniques
Analyte Matrix Membrane Hyphenated analysis* Ref.
Antiepileptics Plasma DM HPLC-UV 360
Benzodiazepines Plasma DM GC-FID, GC-NPD, GC/MS 361
Clozapine Plasma DM HPLC-UV 362
Sildenofil &metabolite Plasma DM HPLC-UV 363
Levosimendan Plasma DM HPLC-UV 364
Velapamil Plasma DM HPLC-UV 365
Antidepressants Plasma DM HPLC-UV 366
Anticonvulsants Plasma DM HPLC-UV 367
Dopamine, cocaine Brain DM HPLC-UV 368
Amphetamines Plasma, serum DM HPLC-FL 369
Sulphonamides Serum, urine DM CE-UV 370
5-Fluorourasil Plasma DM CE-UV 371
Iodixanol Plasma DM HPLC-UV 372
Albendazole &metabolite Plasma DM HPLC-UV 373
Cefazolin Blood, brain DM HPLC-UV 374
Propranolol Blood DM HPLC-FL 375
Drugs Blood DM LC/MS/MS 376
Cephaloridine Blood DM HPLC-UV 377
Cefsulodin Blood DM HPLC-UV 378
Cefepime Bile DM HPLC-UV 379
Tricyclic antidepressants Serum, urine DM CE-UV 370
Bambuterol Plasma SLM CE-UV 381
Basic drugs Plasma SLM CE-UV 382
Ziprasidone Plasma SLM LC.MS 383
Lamivudine, fluticasone Plasma, serum, SLM LCMS 384
urine
Local anesthetics Plasma MMLLE GC-FID 385
Ropivacaine &metabolite Urine SLM HPLC-UV 386
Methotrexate &metabolite Serum, urine SLM LC/MS/MS 387
aDM: dialysis membrane; SLM: supported liquid membrane; MMLLE: microporous membrane
LLE. *See list of abbreviations.

commonly used indirect in vitro or ex vivo methods (e.g., equilibrium dialysis or

ultrafiltration), since it enables monitoring of biotransformation directly in the
tissues, the determination of free drug concentration at the site of drug action, or
eventually in tissues acting as drug (metabolite) depot compartments. In these
cases, lower enrichment is usually obtained, as the sample volumes are generally
smaller, but here it is the selectivity that is crucial. Also the potential for
automation is of significance.
Supported liquid membrane extraction (SLM) permits studies of drug-
protein binding properties. Palmarsdottir et al. [381,382] reported the SLM
technique for selective on-line enrichment of basic drugs in plasma combined
with CE. The analyte is extracted from the outside of the hollow fiber (donor)
through the liquid membrane into the acceptor solution in the fiber lumen, in

820
which the process is driven by difference in pH between the donor and acceptor
solutions. Then the whole volume of the acceptor solution is injected into the CE
capillary by using the double stacking procedure for large volume injection. In
SLM-CE and other drug analysis applications using microporous membrane liq-
uid-liquid extraction (MMLLE)-GC [384], all pertaining to blood plasma sam-
ples, similar high degrees of selectivity have been demonstrated. In these
applications, enrichment factors of 30-70 times were obtained. The available
volume of a blood plasma sample is limited, so the systems were optimized so
that the entire extract from each sample (<1 ml) was injected, resulting in one
chromatographic run. Recently, Rule et al. [387] reported a new membrane-
based extraction technique for LC/MS/MS determination of methotrexate and
its metabolite in human plasma and urine samples using 384-well plates con-
taining C18 particles incorporated into a glass-fiber membrane. The details of
other membrane-based extraction techniques are also summarized in well-
documented reviews [34,35,109-111, 388-395].

23.4.8 Other sample preparation techniques

23.4.8.1 Liquid-phase microextraction

Liquid-phase microextraction (LPME) is a newly developed solvent-minimized
sample preparation technique, which is quick, inexpensive and since very little
solvents is used, there is minimal exposure to toxic organic solvents. This
technique is compatible with capillary GC, CE and HPLC [396]. In this example
the target analytes were extracted from the sample matrix and into a small
volume of acceptor solution with a preconcentration factor of 30-125. The
acceptor solution was directly injected into the analytical instrument. For LPME
combined with GC a porous fiber was filled with a suitable organic solvent. For
LPME combined with CE and HPLC an organic solvent was immobilized in the
pores of a hollow fiber and the internal volume of the fiber was filled with an
aqueous solution in which the analyte was highly soluble. The solvent consump-
tion was reduced by 99% compared to the traditional methods used for sample
preparation. LPME provided extracts with highly enriched analytes and excell-
ent clean-up of endogenous compounds. Solvent consumption was greatly re-
duced compared to traditional LLE. Furthermore in the LPME-GC method
there was no need for evaporation of solvent and reconstitution of analytes prior
to injection into the GC. Disposable extraction units eliminated the possibility of
carry-over and the need for regeneration of the fiber. This was feasible as the
costs of the extraction unit were low. Only the porous polypropylene hollow fiber
had to be replaced for subsequent extractions. The technique was applied as a
sample preparation technique for biological samples, for the analysis of various
drugs such as benzodiazepines, amphetamines and non-steroidal anti-inflam-
matory drugs [395,396], and antidepressants [398]. The LPME technique for
sample extraction has been also described in more detail in several papers
[399-403].

821
23.4.8.2 Microwave-assisted extraction
Microwave-assisted extraction (MAE) is a process for using microwave energy to
heat solvents in contact with a sample in order to partition analytes from the
sample matrix into the solvent. The ability to rapidly heat the sample solvent
mixture is inherent to MAE and is the main advantage of this technique. By
using closed vessels the extraction can be performed at elevated temperatures,
accelerating the mass transfer of target compounds from the sample matrix. A
typical extraction procedure takes 15-30 min and uses solvent volumes in the
range of 10-30 ml. In addition, sample throughput is increased as several
samples can be extracted simultaneously. In most cases analyte recovery and
reproducibility are improved compared to conventional techniques. Although
sample preparation for chemical analysis is a major activity for the pharmaceuti-
cal industry, relatively few papers have been published in this area using MAE.
MAE has been used for extraction of antibiotics such as chloramphenicol in eggs
[404], sulphamethazine in swine tissue [405], vitamins from foodstuffs [406] and
puerarin in Chinese herbal medicine [407] using a household microwave oven.
Felodipine and one of its degradation products have been extracted from tablets
[408]. By optimizing the extracting solvent (5% methanol in acetonitrile) whole
tablets could be extracted with full recovery without grinding the tablet prior to
extraction. The details of other MAE techniques are also summarized in a
well-documented review [47].

23.5 CONCLUSIONS

Drug analysis in biological samples and pharmaceutical products is very import-

ant for TDM, pharmacokinetic studies, screening of illicit drugs and develop-
ment of new medicines. Various biological specimens, particularly plasma,
serum and urine, have been used as samples for analysis. Blood (plasma, serum)
is the most useful choice for simultaneous screening and quantification. Blood
samples usually require deproteinization and, if necessary, cleavage of conju-
gates of the drugs and their metabolites, prior to sample pretreatment and
chromatography or electrophoresis. Urine is the sample of choice for a compre-
hensive screening and identification of unknown drugs and their metabolites
[10], because of the relatively high concentrations of drugs typically present in
urine. Urine collection is easy, but it is necessary to adequately address the risk
of contamination and degradation of drugs during storage. Saliva, sweat and
hair analyses are today to be considered as useful adjuncts to conventional drug
testing. In particular, toxicological analysis from hair samples allows the detec-
tion of past, chronic drug use [409]. However, there is still controversy on how to
interpret the results, particularly concerning external contamination, cosmetic
treatment and ethnic bias. The use of milk, tissues, faces and breath samples for
drug analysis are unusual in clinical applications. On the other hand, pharma-
ceutical products provide various formulation types such as solids, liquids,
topicals, injectables and inhalants, and various excipient actions for tablets, cap-

822
sules, coatings, flavorings, colorings, buffers, suspendings and preservatives.
Clean-up of these samples for the analysis of target drugs are commonly
performed by LLE, SPE, SPME and their combination after precipitation,
centrifugation and dialysis.
Because most analytical instruments cannot handle the matrix directly sam-
ple preparation is necessary to isolate the components of interest from the sam-
ple matrix. This however has always been a rather neglected part of analytical
chemistry, even though it is also one of the most significant sources of inaccura-
cies in the entire analytical procedure. Suitable sample preparation is an impor-
tant prerequisite for analysis of biological samples. Many of the techniques
currently used for the preparation of complex samples prior to chromatographic
or electrophoretic analysis, such as filtration, precipitation and extraction with
organic solvents, have been in use for several decades with essentially no modifi-
cation over the years. Universal LLE procedures are preferable for general
screening procedures, because substances with very different physico-chemical
properties may be isolated from heterogeneous matrices. The acceptance of new
technologies has been slow, both by the users themselves and by regulatory
agencies. In general, sample preparation is still regarded as "low tech" and in
many laboratories the associated tasks are assigned to the least trained staff,
who may often be reluctant to accept new technologies. This is one of the main
reasons why the implementation of new sample preparation techniques has been
slow and thereby sample preparation in many cases is still the most time-
consuming and often rate-limiting step of the total analytical procedure.
As described in this chapter, SFE, SPE, SPME, IAE, MD, ME, LPME and
MAE have been developed as newer sample preparation techniques. Among
these techniques, SPE is the most popular sample preparation technique for
drug analysis and has come to play a truly crucial role in laboratories all over the
world. It has also replaced older techniques to a relatively large extent. The
increased development of SPE has occurred during the past decade, with many
improvements in formats, automation and the introduction of new phases. One
reason for the general acceptance of SPE has been the pressure to decrease
organic solvent usage in laboratories, which has encouraged the requirement for
solvent-free procedures and has greatly contributed to the growth of SPE at the
expense of LLE procedures. Sample pretreatment for SPE depends on the
sample type: whole blood and tissue need deproteinization and filtration/centri-
fugation steps before application to the SPE columns, whereas for urine usually
a simple dilution step and/or centrifugation is satisfactory. However, clogging,
channeling and percolation are typical problems of SPE encountered in every-
day laboratory work. Some promising approaches in SPE are based on special
packings such as restricted access materials (RAMs) and MIPs [410]. Other
methods of sample preparation are SFE, on-column sample preparation with
column-switching techniques and on-line SPE in LC using RAMs [410,411],
LC-GC coupling [412] and membrane-based sample preparation such as
dialysis, electrodialysis, ultrafiltration. Although these methods have their own

823
merits, most of them are only found in isolated applications, and often they do
not achieve the sensitivity and selectivity of LLE and SPE. Finally, some
methods require the use of expensive equipment. Other problems are fouling of
membranes in membrane-based sample preparation and irreversible binding of
some high-molecular-mass material in on-line SPE.
SPME is becoming an attractive alternative to SPE and LLE for some
applications. SPME is a solvent-free and concentrating extraction technique and
applied for the analysis of drugs and metabolites in biological fluids. The most
striking attribute of SPME is its low total recovery as reported for many
methods. This is not unexpected because SPME is an equilibrium extraction
technique rather than an exhaustive extraction. However, several SPME meth-
ods have also been presented with high total recovery and the performance of the
majority of methods was good with regard to sensitivity, linearity, precision and
accuracy. As evidenced with SPE, immunoaffinity sorbent [307] and MIP [316]
are also expected to be applied as new sorbent materials for highly efficient
extraction of drugs from various biological samples by SPME. With the develop-
ment of more sensitive phases it may be possible to further miniaturize the
technique. As the market for SPME increases in the future this could lead to the
introduction of disposable low-cost extraction fibers (e.g. in the form of a
carousel) or tubes such as in other areas of sample preparation, e.g., SPE
multiwell plates. A more approach to the design and concept of SPME has been
recently proposed in the form of Twister®, available from Gerstel [413--415]. In
this technique, known as stir bar sorptive extraction, a coated magnetic stirring
bar is used, with the same PDMS phase as for SPME but in a thicker layer
(0.3-1.0 mm) and it is also compatible with both GC and HPLC desorption
procedures. It can theoretically improve sensitivity compared to a SPME fiber
for certain applications, but it requires the special desorption unit and is difficult
to automate. To date the Twister has only been applied to environmental
samples.
The trends are clearly to simplify the labor of sample preparation, increasing
its reliability and precision, and eliminating the clean-up step by using more
selective extraction procedures. The development of more selective sorbents will
remain an active area of research, as it is today for IAE sorbents and MIPs. IAE
has been shown as another feasible procedure for trace analysis of biological
samples. As more antibodies become available and as procedures to develop
them become more sophisticated it is likely that more methods will be developed.
MIP antibody mimics are much easier to synthesize and consequently much less
expensive. MIPs have provided a new tool in affinity separation-based SPE,
because conventional affinity sorbents are expensive. The development of SPE
based on inexpensive MIPs allows a disposable type of affinity-SPE, potentially
making the availability of affinity extraction with excellent selectivity wide-
spread. Many applications can be expected in clinical, pharmaceutical and
biochemical analyses for molecular recognition techniques such as affinity
extraction media using immobilized antibodies, substitution of the immuno-

824
assay for antibody extraction, sensor material which may replace the biochip,
novel carriers for capillary electrophoresis, selective membranes, and other new
intelligent materials for molecular recognition.
Dialysis is less used as a sample preparation technique in this field and is
being succeeded by batch-like techniques such as 96-well plate SPE, which
allows several hundreds of samples to be processed per day. It can therefore be
expected that the importance of on-line dialysis for biomedical samples will
decrease in the next few years. Non-porous membrane extraction has been
shown to be compatible with HPLC, GC and CE and applicable to many different
sample types. Its robustness appears to be comparable to that of on-line dialysis,
although the leakage of membrane-incorporated solvent could be problematic
for routine use. Method development should not be more difficult than for liquid
extraction (followed by a back-extraction). However, despite its favorable chara-
cteristics, membrane extraction has seen little use outside the university
laboratory. An important reason could be that the technique has not been com-
mercialized so far, while most users do not want to manufacture analytical sys-
tems themselves, but rely upon commercially available equipment. As long as
this does not change, there is little chance that membrane extraction will have a
similar breakthrough as dialysis when it was commercialized some 10 years ago.
Users have had an increasing interest in the automation of sample prepara-
tion for faster sample-preparation procedures, improved precision and more
cost-effective analyses. The key attractive features of automated sample prepa-
ration techniques include miniaturization, throughput, reproducibility and
traceability. In the last decade, new concepts have been developed, which allow
the on-line coupling of sample preparation devices to separation and detection
systems, and which are especially designed for automation. In the future,
advances in better integration of sampling/sample preparation and instrumen-
tal analysis will allow wider use of automated on-line analysis in clinical and
pharmaceutical analysis.

REFERENCES

1 R.H. Liu and D.E. Gadzala (Eds.), Handbook of Drug Analysis: Applications in
Forensic and Clinical Laboratory. American Chemical Society, Washington, DC,
1997.
2 K. Valko (Ed.), SeparationMethods in Drug Synthesis and Purification.Handbook of
Analytical Separation, Vol. 1. Elsevier, Amsterdam, 2000.
3 J.H. Lin and A.Y. Lu, Pharmacol.Rev., 49 (1997) 403.
4 A. Marzo, Pharmacol.Res., 36 (1997) 425.
5 K.A. Jackson and S.E. Rosenbaum, Drug Dev. Ind. Pharm., 24 (1998) 1155.
6 S. Ulrich, C. Wurthmann, M. Brosz and F.P. Meyer, Clin. Pharmacokinet., 34 (1998)
227.
7 M.H.H. Ensom, G.A. Davis, C.D. Cropp and R.J. Ensom, Clin. Pharmacokinet., 34
(1998) 265.
8 S. Ulrich, I. Schrodter, G. Partscht and P. Baumann, Psychopharmakotherapie,7
(2000) 2.

825
 9 V. Iyengar, J. Kumpulainen, K. Okamoto, M. Morita, S. Hirai and S. Nomoto, Prog.
Clin. Biol. Res., 380 (1993) 329.
10 H.H. Maurer, J. Chromatogr.B, 580 (1992) 3.
11 H.H. Maurer, J. Chromatogr. B, 713 (1998) 3.
12 H.H. Maurer, J. Chromatogr.B, 733 (1999) 3.
13 O.H. Drummer, J. Chromatogr. B, 733 (1999) 27.
14 A. Polettini, J. Chromatogr. B, 733 (1999) 47.
15 P. Marquet and G. Lachiatre, J. Chromatogr.B, 733 (1999) 93.
16 M.J. Bogusz, Forensic Science. Handbook of Analytical Separation, Vol. 2. Elsevier,
Amsterdam, 2000.
17 J. Pawliszyn, Solid Phase Microextraction: Theory and Practice. Wiley-VCH, New
York, 1997.
18 N.T. Crosby and I. Patel, General Principles of Good Sampling Practice. The Royal
Society of Chemistry, Cambridge, UK, 1995.
19 M. Stoeppler (Ed.), Sampling and Sample Preparation. Springer-Verlag, Berlin,
Germany, 1997.
20 E. Riva, R. Merati and L. Cavenaghi, J. Chromatogr., 553 (1991) 35.
21 A. Haque and J.T. Stewart, Biomed. Chromatogr., 13 (1999) 51.
22 D. Stevenson and I.D. Wilson (Eds.), Sample Preparation for Biomedical and
EnvironmentalAnalysis. Plenum Press, New York, 1994.
23 E. Thurman and M. Mills, Solid Phase Extraction: Principles and Practice.
Wiley-VCH, New York, 1998.
24 J. Fritz, Analytical Solid Phase Extraction. Wiley-VCH, New York, 1999.
25 S.P. Bjergaard, K.E. Rasmussen and T.G. Halvorsen, J. Chromatogr.A, 902 (2000)
91.
26 R.D. McDowall, J. Chromatogr.B, 492 (1989) 3.
27 W. Thormann, S. Lienhard and P. Wernly, J. Chromatogr., 636 (1993) 137.
28 D.K. Lloyd, J. Chromatogr.A, 735 (1996) 29.
29 O.H. Drummer, J. Chromatogr. B 713 (1998) 201.
30 T. Kumazawa and 0. Suzuki, J. Chromatogr.B, 747 (2000) 241.
31 R.E. Majors, LC-GC, 13 (1995) 547.
32 R.A. de Zeeuw, J. Chromatogr.B, 689 (1997) 71.
33 J.P. Franke and R.A. de Zeeuw, J. Chromatogr.B, 713 (1998) 51.
34 J.R. Veraart, H. Lingeman and U.A.Th. Brinkman, J. Chromatogr.A, 856 (1999)
483.
35 M. Gilar, E.S.P. Bouvier and B.J. Compton, J. Chromatogr.A, 909 (2001) 111.
36 F. Degel, Clin. Biochem., 29 (1996) 529
37 Z.E. Penton, Adv. Chromatogr., 37 (1997) 205.
38 L. Junting, C. Peng and 0. Suzuki, Forensic Sci. Int., 97 (1998) 93.
39 F. Sporkert and F. Pragst, ForensicSci. Int., 107 (2000) 129.
40 G. Theodoridis, E.H.M. Koster and G.J. de Jong, J. Chromatogr.B Biomed. Sci. 745
(2000) 49.
41 N.H. Snow, J. Chromatogr.A, 885 (2000) 445.
42 H.L. Lord and J. Pawliszyn, J. Chromatogr. A, 902 (2000) 17.
43 S. Ulrich, J. Chromatogr. A, 902 (2000) 167.
44 G.A. Mills and V. Walker, J. Chromatogr. A, 902 (2000) 267.
45 K.G. Furton, J. Wang, Y-L. Hsu, J. Walton and J.R. Almirall, J. Chromatogr.Sci., 38
(2000) 297.
46 H. Kataoka, H.L. Lord and J. Pawliszyn, in: I.D. Wilson, T.D. Adlard and C.F. Poole
and M. Cook (Eds.), Encyclopedia of Separation Science. Academic Press, London,
2000, p. 4153.
47 C.S. Eskilsson and E. Bjorklund, J. Chromatogr.A, 902 (2000) 227.

826
48 J. Segura, R. Ventura and C. Jurado, J. Chromatogr. B, 713 (1998) 61.
49 T. Inoue and S. Seta, ForensicSci. Rev., 4 (1992) 89.
50 I.D. Watson, in: I.D. Wilson (Ed.), Sample Preparation for Biomedical and
Environmental Analysis. Plenum Press, New York, p.71, 1994.
51 D.A. Kidwell, J.C. Holland and S. Athanaselis, J. Chromatogr. B, 713 (1998) 111.
52 C. Staub, J. Chromatogr. B 733 (1999) 119.
53 P. Kintz and N. Samyn, J. Chromatogr. B, 733 (1999) 137.
54 Y. Nakayama, J. Chromatogr.B, 733 (1999) 161.
55 Y. Gaillard and G. Pepin, J. Chromatogr.B, 733 (1999) 231.
56 M.A. Huestis, J.M. Oyler, E.J. Cone, A.T. Wstadik, D. Schoendorfer and R.E. Joseph,
J. Chromatogr. B, 733 (1999) 247.
57 M. Yashiki, N. Nagasawa, T. Kojima, T. Miyazaki and Y. Iwasaki, Jpn. J. Forensic
Toxicol., 13 (1995) 17.
58 B. Dehon, L. Humbert, L. Devisme, M. Stievenart, D. Mathieu and N. Houdret, J.
Anal. Toxicol., 24 (2000) 22.
59 D.B. Menkes, R.C. Howard, G.F.S. Spears and E.R. Cairns, Psychopharmacology,103
(1991) 277.
60 E.J. Cone, J. Anal. Toxicol., 14 (1990) 1.
61 W. Schramm, R.H. Smith, P.A. Craig and D.A. Kidwell, J. Anal. Toxicol., 16 (1992) 1.
62 L.M. North, N.D. Gaudette, M.L. Cordeiro, J.H. Fitchen, S.L. Davidson and M.S.
Hindahl, Ann. N.Y. Acad. Sci., 694 (1993) 332.
63 M.L. Cordeiro, C.S. Turpin and S.A. McAdams, Ann. N.Y. Acad. Sci., 694 (1993) 330.
64 T. Thieme, J. Fitchen, F. Bartos, M. Salinsky, T. Green and W. Holden, Ann. N.Y.
Acad. Sci., 694 (1993) 337.
65 M. Navazesh, Ann. N.Y. Acad. Sci., 694 (1993) 72.
66 M. Slavik, J. Wu, N. Brown, A. Williams, S. Wagner and J. Slavik, Ann. N.Y. Acad.
Sci., 694 (1993) 317.
67 W. Schramm, R.H. Smith and P.A. Craig, Ann. N.Y. Acad. Sci., 694 (1993) 311.
68 W. Schramm and R.H. Smith, Clin. Chem., 37 (1991) 114.
69 W. Schramm, O.F. Pomerleau, C.S. Pomerleau and H.E. Grates, Prevent. Med., 21
(1992) 63.
70 F. Harding (Ed.), Milk Quality. Blackie, Glasgow, 1995, 75.
71 M.R. Harkey, Forensic Sci. Int., 63 (1993) 9.
72 T. Mieczkowski, Microgram, 28 (1995) 193.
73 D.L. Blank and D.A. Kidwell, ForensicSci. Int., 70 (1995) 13.
74 P. Kintz and P. Mangin, ForensicSci. Int., 70 (1995) 3.
75 D.A. Kidwell and D.L. Blank, Forensic Sci. Int., 63 (1993) 137.
76 C. Jurado, P. Kintz, M. Menendez and M. Repetto, Int. J. Leg. Med., 110 (1997) 159.
77 H. Sachs, ForensicSci. Int., 84 (1997) 7.
78 C. Staub, ForensicSci. Int., 84 (1997) 295.
79 M. Chiarotti, ForensicSci. Int., 63 (1993) 161.
80 Y. Gaillard, F. Vaysette, A. Balland, G. PBpin, J. Chromatogr.B, 735 (1999) 189.
81 W.L. Wang and E.J. Cone, ForensicSci. Int., 70 (1995) 39.
82 D.L. Blank and D.A. Kidwell, Forensic Sci. Int., 63 (1993) 145.
83 M. Burns and R.C. Baselt, J. Anal. Toxicol., 19 (1995) 41.
84 D.P. Conner, R.G. Almirez, P. Rhyne, K. Zamani, B.J. Bolden and C.C. Peck, Skin
Pharmacol., 2 (1989) 155.
85 D. Conner, E. Millora, K. Zamani, D. Nix, R. Almirez, P. Rhyne-Kirsch and C. Peck,
J. Invest. Dermatol., 2 (1990) 186.
86 G. Skopp, L. Potsch, H.-P. Eser and M.R. Miller, J. Forensic Sci. 41 6 (1996) 933.
87 I. Sunshine and J.P. Sutliff, in: S.H.Y. Wong and I. Sunshine (Eds.), Handbook of
Analytical Therapeutic Drug Monitoring and Toxicology. CRC Press, Boca Raton,

827
 FL, 1997, p. 253.
88 F.P. Smith and D.A. Kidwell, Forensic Sci. Int., 83 (1996) 179.
89 D.A. Kidwell, M.A. Blanco and F.P. Smith, Forensic Sci. Int., 84 (1997) 75.
90 C. Grote and J. Pawliszyn, Anal. Chem., 69 (1997) 587.
91 R. Hypler, Crhova, J. Gasparic, Z. Zaddk, M. Cikova and V. Balasova, J. Chromatogr.
B, 739 (2000) 183.
92 K.E. Kongshaug, S. Pedersen-Bjergaard, K.E. Rasmussen and M. Krogh, Chroma-
tographia, 50 (1999) 247.
93 J.C. Coumbaros, K.P. Kirkbride and G. Klass, J. Forensic Sci., 44 (1999) 1237.
94 J.J. Vreuls, A.J.H. Louter and U.A.Th. Brinkman, J. Chromatogr.A, 856 (1999) 279.
95 D. Barcelo and M.-C. Hennion, Anal. Chim. Acta, 318 (1995) 1.
96 R.W. Fedeniuk and P.J. Shand, J. Chromatogr.A, 812 (1998) 3.
97 M.-C. Hennion, J. Chromatogr.A, 856 (1999) 3.
98 P.R. Haddad, P. Doble and M. Macka, J. Chromatogr.A, 856 (1999) 145.
99 M.E.L. Gonzalez and L.V.P. Arribas, J. Chromatogr. A, 902 (2000) 3.
100 D. Martinez, M.J. Cugat, F. Borrull and M. Calull, J. Chromatogr.A, 902 (2000) 65.
101 J. Pawliszyn, J. Chromatogr.Sci., 38 (2000) 270.
102 H. Kataoka, H. Lord and J. Pawliszyn, J. Chromatogr. A, 880 (2000) 35.
103 H. Lord and J. Pawliszyn, J. Chromatogr. A, 885 (2000) 153.
104 J. Beltran, F.J. Lopez and F. Hernandez, J. Chromatogr.A, 885 (2000) 389.
105 M. de F. Alpendurada, J. Chromatogr. A, 889 (2000) 3.
106 H. Kataoka, Anal. Bioanal. Chem., 373 (2002) 31.
107 N. Delaunay, V. Pichon and M.-C. Hennion, J. Chromatogr. B, 745 (2000) 15.
108 D. Stevenson, J. Chromatogr.B, 745 (2000) 39.
109 N.C. van de Merbel, J. Chromatogr. A, 856 (1999) 55.
110 B.M. Cordero, J.L. Perez, C.G. Pinto, M.E.F. Laespada, R.C. Martinez and E.R.
Gonzalo, J. Chronatogr.A, 902 (2000) 195.
111 J.A. Jonsson and L. Mathiasson, J. Chromatogr.A, 902 (2000) 205.
112 D.J. Jones, K.T. Nguyen, M.J. McLeish, D.P. Crankshaw and D.J. Morgan, J.
Chromatogr.B, 675 (1996) 174.
113 C. N.-Hoog, P. Neidenstrom and T. Arvidsson, J. Chromatogr.B, 760 (2001) 123.
114 B. Malavasi, M. Ripamonti, A. Rouchouse and V. Ascalone, J. Chromatogr.A, 729
(1996) 323.
115 F. Sadeghipour and J.L. Veuthey, J. Chromatogr. A, 787 (1997) 137.
116 R. Theurillat and W. Thormann, J. Pharm.Biomed. Anal., 18 (1998) 751.
117 Y.Q. Xia, D.B. Whigan and M. Jemal, Rapid Commun. Mass Spectrom., 13 (1999)
1611.
118 M. Walshe, M.T. Kelly and M.R. Smyth, J. Pharm. Biomed. Anal., 14 (1996) 475.
119 K.C. Marsh, E. Eiden and E. McDonald, J. Chromatogr.B, 704 (1997) 307.
120 A. Tolokan, V. Horvath, G. Horvai, A. Egresi, K.B. Nemes and I. Kiebovich, J.
Chromatogr. A, 896 (2000) 279.
121 S.L. Hannan, G.A. Ridout and A.E. Jones, J. Chromatogr. B, 678 (1996) 297.
122 S. Ulrich, T. Isensee and U. Pester, J. Chromatogr. B, 685 (1996) 81.
123 J. Macek, P. Ptacek and J. Klima, J. Chromatogr. B, 755 (2001) 279.
124 A. Llerena, R. Berecz, M.J. Norberto and A. dela Rubia, J. Chromatogr. B, 755 (2001)
349.
125 N.N. Davydova, S.U. Yasuda, R.L. Woosley and I.W. Wainer, J. Chromatogr. B, 744
(2000) 177.
126 V.A. Jabor, V.L. Lanchote and P.L. Bonato, Electrophoresis,22 (2001) 1406.
127 S. Stavchansky, S. Demirbas, L. Reyderman and C.K. Chai, J. Pharm. Biomed. Anal.,
13 (1995) 919.
128 A.P. Hart, S. M.-Proo, W. Blackwell andA. Dasgupta, Ther. DrugMonit., 19 (1997) 431.

828
129 A.D. Huitma, M.M. Tibben, T. Kerbusch, J.W. Zwikker, S. Rodenhuis and J.H.
Beijnen, J. Chromatogr.B, 716 (1998) 177.
130 V. Cirimeie, P. Kintz, C. Staub and P. Mangin, Forensic Sci. Int., 84 (1997) 189.
131 F. de Cazanove, J.M. Kinowski, M. Audran, A. Rochette and F. Bressolle, J. Chroma-
togr.B, 690 (1997) 203.
132 S.M. De Baere, W.E. Lambert and A.P. De Leenheer, J. Anal. Toxicol., 22 (1998) 18.
133 C.L.-Calull, L.G.-Capdevila, C. Arroyo and J. Bonal, J. Chromatogr.B, 693 (1997)
228.
134 J. Macek, P. Ptacek and J. Klima, J. Chromatogr. B, 736 (1999) 231.
135 A. Prochazkova, M. Chouki, R. Theurillat and W. Thormann, Electrophoresis, 21
(2000) 729.
136 R. Bortocan, V.L. Lanchote, E.J. Cesarino and P.S. Bonato, J. Chromatogr.B, 744
(2000) 299.
137 R.G. Lumbreras, D.P. Trapero and R.I. Hornillos, J. Chromatogr. B, 754 (2001) 419.
138 A. Dasgupta and C.E. Mahle, J. Forensic. Sci., 42 (1997) 937.
139 A. El Mahjoub, C. Staub, J. Pharm.Biomed. Anal., 23 (2000) 1057.
140 R. Metz, P. Muth, M. Ferger, V.G. Kukes and H. Vergin, J. Chromatogr.A, 810 (1998)
63.
141 C.D. Torchin, W.D. Yonekawa, I.M. Kapetanovic and H.J. Kupferberg, J. Chroma-
togr. B, 724 (1999) 101.
142 M. Bouley, C. Briere, C. Padoin and 0. Petitjean and M. Tod, Ther. Drug Monit., 23
(2001) 56.
143 R. van Gijn, 0. van Tellingen, J.J. de Clippeleir, M.J. Hillebrand, E. Boven, J.B.
Vermorken, W.W. ten Bokkel Huinink, S. Schwertz, P. Graf and J.H. Beijnen, J.
Chromatogr. B, 667 (1995) 269.
144 H. Inoue, Y. Maeno, M. Iwasa, J. Monma and R. Matoba, J. Chromatogr. B, 701
(1997) 47.
145 P. Villani, M. Pregnolato, S. Banfo, M. Rettani, D. Burroni, E. Seminari, R. Maserati
and M.B. Regazzi, Ther. Drug Monit., 21 (1999) 346.
146 J. Ciccolini, J. Catalin, M.F. Blachon and A. Durand, J. Chromatogr.B, 759 (2001)
299.
147 A. Gauvin, F. Pinguet, S. Poujol. C. Astre and F. Bressolle, J. Chromatogr.B, 748
(2000) 389.
148 J.O. Svensson, C. Brattstrom and J. Sawe, Ther. Drug Monit., 19 (1997) 112.
149 M.T. Kelly, D. McGuirk and F.J. Bloomfield, J. Chromatogr.B, 668 (1995) 117.
150 P. Muth, R. Metz, B. Siems, W.W. Bolten and H. Vergin, J. Chromatogr.A, 679 (1996)
169.
151 M. Matsuoka, T. Maki, K. Banno and T. Sato, J. Chromatogr.B, 685 (1996) 237.
152 C. Arroyo, C. L.-Calull, L. G.-Capodevila, I. Gich, M. Barbanoj and J. Bonal, J.
Chromatogr. B, 688 (1997) 339.
153 G.P. Kaijser, J.H. Beijnen, A. Bult, H.J. Keizer and W.J. Underberg, J. Chromatogr.
B, 690 (1997) 131.
154 T. Kerbusch, M.J. Jeuken, J. Derraz, J.W. van Putten, A.D. Huitema and J.H.
Beijnen, Ther. Drug Monit., 22 (2000) 613.
155 C. Guitton, J.M. Kinowski, R. Aznar and F. Bressolle, J. Chromatogr.B, 690 (1997)
211.
156 R. Ohta, T. Amano, K. Yamashita and M. Motohashi, J. Chromatogr.B, 704 (1997)
325.
157 C. Sottani, R. Turci, L. Perbellini and C. Minoia, Rapid Commun. Mass Spectrom., 12
(1998) 1063.
158 H.J. Mascher, J. Chromatogr.A, 812 (1998) 339.
159 B.B. Rasmussen and K. Brosen, J. Chromatogr.B, 676 (1996) 169.

829
160 A. Sparreboom, P. de Bruijn, K. Nooter, W.J. Loos, G. Stoter and J. Verweij, J.
Chromatogr. B, 705 (1998) 159.
161 S. Steinbormer and J. Henion, Anal. Chem., 71 (1999) 2340.
162 N. Zhang, K.L. Hoffman, W. Li and D.T. Rossi, J. Pharm. Biomed. Anal., 22 (2000)
131.
163 C. Staub, Forensic Sci. Int., 84 (1997) 295.
164 J.W. King and J.E. France, in: B. Wenclawiak (Ed.), Analysis with Supercritical
Fluids: Extraction and Chromatography.Springer-Verlag, Bellin, 1992, p. 32.
165 B.J. Perrigo and B.P. Joynt, Can. Soc. Forens. Sci. J., 25 (1992) 70.
166 R.F. Cross, J.L. Ezzell and B.E. Richter, J. Chromatogr. Sci., 31 (1993) 162.
167 C. Staub, P. Edder and J.L. Veuthey, in: P. Kintz (Ed.) Drug Testing in Hair. CRC
Press, Boca Raton, FL, 1996 p. 1 2 2 .
168 V. Cirimele, V. Kintz, R. Majdalani and P. Mangin, J. Chromatogr.B, 673 (1995) 173.
169 J.F. Morrison, S.N. Chesler, W.J. Yoo and C.M. Selavka, Anal. Chem., 70 (1998) 163.
170 D.L. Allen and J.S. Oliver, Forensic Sci. Int., 107 (2000) 191.
171 B.R. Simmons and J.T. Stewart, J. Chromatogr. B, 688 (1997) 291.
172 K.S. Scott and J.S. Oliver, J. Anal. Toxicol., 70 (1997) 297.
173 L. Karlsson, A. Torstensson and L.T. Taylor, J. Pharm. Biomed. Anal., 15 (1997) 601.
174 J.K. Lawrence, A.K. Larson and I.R. Lebbett, Anal. Chim. Acta, 288 (1994) 123.
175 S. Scalia, G. Ruberto and F. Bonina, J. Pharm. Sci., 84 (1995) 433.
176 M.T. Tena, M.D. Luque de Castro and M. Valcarcel, Chromatographia,40 (1995) 197.
177 A.L. Howard, M.C. Shah, M.A. Brooks, J.T. Strode and L.T. Taylor, J. Pharm. Sci.,
83 (1994) 1537.
178 M. Masuda, S. Koike, M. Handa, K. Sagara and T. Mizutani, Anal. Sci., 9 (1993) 29.
179 S.M. Wang, Y.S. Giang and Y.C. Ling, J. Chromatogr. B, 759 (2001) 17.
180 W.E. Brewer, R.C. Galipo, K.W. Sellers and S.L. Morgan, Anal. Chem., 73 (2001)
2371.
181 C. Radcliffe, K. Maguire and B. Lockwood, J. Biochem. Biophys. Meth., 43 (2000) 261.
182 M. Jemal, M. Huang, Y. Mao, D. Whigan and A. Schuster, Rapid Commun. Mass
Spectrom., 14 (2000) 1023.
183 G. Rule, M. Chapple and J. Henion, Anal. Chem., 73 (2001) 439.
184 M.C. Rouan, C. Buffet, L. Masson, F. Marfil, H. Humbert, G. Maurer, J. Chromatogr.
B, 754 (2001) 45.
185 H. Svennberg and P.O. Lagerstrom, J. Chromatogr. B, 689 (1997) 371.
186 H.R. Angelo and A. Petersen, Ther. Drug Monit., 23 (2001) 157.
187 D. Ohman, B. Carlsson and B. Norlander, J. Chromatogr. B, 753 (2001) 365.
188 H. Astier, C. Renard, V. Cheminel, O. Soares, C. Mounier, F. Peyron and J.F.
Chaulet, J. Chromatogr. B, 698 (1997) 217.
189 P. Burmat and F. Do B. Robles, J. Chromatogr. B, 706 (1998) 295.
190 J.A. Prietro, R.M. Jimenez and R.M. Alonso, J. Chromatogr. B, 714 (1998) 285.
191 N. Lindegardh and Y. Bergqvist, J. Chromatogr. B, 744 (2000) 9.
192 M.A. Raggi, G. Casamenti, R. Mandrioli and V. Volterra, J. Chromatogr. B, 750
(2001) 137.
193 J.A. Prietro, R.M. Jimenez, R.M. Alonso and E. Oritiz, J. Chromatogr. B, 754 (2001)
23.
194 E.B.A. Adjaye and G.K. Shiu, J. Chromatogr. B, 707 (1998) 161.
195 J.R. Valdes Santurio and E. Gonzalez Porto, J. Chromatogr. B, 682 (1996) 364.
196 A.E.I. Mahjoub and C. Staub, J. Chromatogr. B, 754 (2001) 271.
197 K.K. Akerman, J. Jolkkonen, M. Parvainen and I. Penttila, Clin. Chem., 42 (1996)
1412.
198 M. Tatsuno, M. Nishikawa, M. Katagi and H. Tsuchihashi, J. Anal. Toxicol., 20
(1996) 281.

830
199 F. von Heeren, R. Tanner, R. Theurillat and W. Thormann, J. Chromatogr.A, 745
(1996) 165.
200 S.B. Black, A.M. Stenhouse and R.C. Hansson, J. Chromatogr.B, 685 (1996) 67.
201 Y. Gaillard and G. Pepin, J. Chromatogr.A, 762 (1997) 251.
202 T. Hirai, S. Matsumoto and I. Kishi, J. Chromatogr. B, 690 (1997) 267.
203 C. Mou, N. Ganju, K.S. Sridhar and A. Krishan, J. Chromatogr. B, 703 (1997) 217.
204 Y.N. Li, B.N. Tattam, K.F. Brown and J.P. Seale, J. Pharm.Biomed. Anal., 16 (1997)
447.
205 S. R.-Welte and P. Dittrich, J. Chromatogr.B, 707 (1998) 151.
206 R. Shimoyama, T. Ohkubo, K. Sugawara, T. Ogasawara, T. Ozaki, A. Kagiya and Y.
Saito, J. Pharm. Biomed. Anal., 17 (1998) 863.
207 S. Frappier, D. Breilh, E. Diarte, B. Ba, D. Ducint, J.L. Pellegrin and M.C. Saux, J.
Chromatogr. B, 714 (1998) 384.
208 K. Kronkvist, M. Gustavsson, A.K. Wendel and H. Jaegfeldt, J. Chromatogr.A, 823
(1998) 401.
209 L.G. Martinez, P. C.-Falco, A. S.-Cabeza and F. B.-Reig, J. Chromatogr.B, 718 (1998)
143.
210 P.J. Taylor and A.G. Johnson, J. Chromatogr.B, 718 (1998) 251.
211 M.J. Bogusz, R.D. Maier, K.D. Kruger and U. Kohls, J. Anal. Toxicol., 22 (1998) 549.
212 M.L. Foisy and J.P. Sommadossi, J. Chromatogr.B, 721 (1999) 239.
213 C. Marzolini, A. Telenti, T. Buclin, J. Biollaz and L.A. Decosterd, J. Chromatogr.B,
740 (2000) 43.
214 M.D. Green, Y. Bergqvist, D.L. Mount, S. Corbett and M.J. D'Souza, J. Chromatogr.
B, 727 (1999) 159.
215 I.D. Davies, J.P. Allanson and R.C. Causon, J. Chromatogr.B, 732 (1999) 173.
216 M.J. Bogusz, K.D. Kruger, R.D. Maier, R. Erkwoh and F. Tuchtenhagen, J. Chroma-
togr. B, 732 (1999) 257.
217 G.G. Encina, R. Farran and L. Martinez, J. Pharm.Biomed. Anal., 21 (1999) 371.
218 Y.Y. Ku, K.C. Wen, L.K. Ho and Y.S. Chang, J. Pharm.Biomed. Anal., 20 (1999) 351.
219 P. Moreno and V. Salvado, J. Chromatogr.A, 870 (2000) 207.
220 R. Shimoyama, T. Ohkubo and K. Sugawara, Ann. Clin. Biochem., 37 (2000) 210.
221 F. Bevalot, Y. Gaillard and M.A. Lhermitte and G. Pepin, J. Chromatogr.B, 740
(2000) 227.
222 R. Gatti, M.G. Gioia and V. Cavrini, J. Pharm. Biomed. Anal., 23 (2000) 147.
223 P.J. Taylor, P. Salm, S.V. Lynch and P.I. Pillans, Ther. Drug Monit., 22 (2000) 608.
224 S.H. Gan and R. Ismail, J. Chromatogr. B, 759 (2001) 325.
225 P. Koopmans, C.E. Gidding, S.S. de Graaf and D.R. Uges, Ther. Drug Monit., 23
(2001) 406.
226 Z. Wang, R.M. Dsida and M.J. Avram, J. Chromatogr. B, 755 (2001) 383.
227 K.R. Kim and H.R. Yoon, J. Chromatogr. B, 682 (1996) 55.
228 W. Weinmann and M. Svoboda, J. Anal. Toxicol., 22 (1998) 319.
229 A. Marchese, C. McHugh, J. Bi and H. Kehler, J. Mass Spectrom., 33 (1998) 1071.
230 J.M. Morales, C.H. Jung, A. Alarcon and A. Barreda, J. Chromatogr. B, 746 (2000)
133.
231 H. Ohta, Y. Seto and N. Tsunoda, J. Chromatogr.B, 691 (1997) 351.
232 E. Kramer and K.A. Kovar, J. Chromatogr. B, 731 (1999) 167.
233 S. Paterson, R. Cordero, S. McCulloch and P. Houldsworth, Ann. Clin. Biochem., 37
(2000) 690.
234 M. Kobylinska and K. Kobylinska, J. Chromatogr. B, 744 (2000) 207.
235 V. R.-Gutierrez, M.E. Juan, A. Cert and J.M. Planas, Anal. Chem., 72 (2000) 4458.
236 E. Mikami, T. Goto, T. Ohno, H. Matsumoto, K. Inagaki, H. Ishihara and M. Nishida,
J. Chromatogr. B, 744 (2000) 81.

831
237 J.O. Svensson, A. Bruchfeld, R. Schvarcz and L. Stahle, Ther. DrugMonit., 22 (2000)
215.
238 P. Femandez, N. Lafuente, A.M. Bermejo, M. L.-Rivadulla and A. Cruz, J. Anal.
Toxicol., 20 (1996) 224.
239 K.M. Hold, D.G. Wilkins, D.E. Rollins, R.E. Joseph and E.J. Cone, J. Chromatogr.
Sci., 36 (1998) 125.
240 S. Dodd, A. Buist, G.D. Burrows, K.P. Maguire and T.R. Norman, J. Chromatogr.B,
730 (1999) 249.
241 S.A. Reuschel, S.E. Percey, S. Liu, D.M. Eades and R.L. Foltz, J. Anal. Toxicol., 23
(1999) 306.
242 H. Nguyen and D.R. Nau, J. Anal. Toxicol., 24 (2000) 37.
243 B.H. Forngren, N. Tyrefors, K.E. Markides and B. Langstrom, J. Chromatogr.B, 748
(2000) 189.
244 R. Swart, P. Koivisto and K.E. Markides, J. Chromatogr.B, 736 (1999) 247.
245 H. Lord, R. Grant and J. Pawliszyn, unpublished data.
246 M. Chiarotti, S. Strano-Rossi and R. Marsili, J. Microcolumn Sep., 9 (1997) 249.
247 S. Strano-Rossi and M. Chiarotti, J. Anal. Toxicol., 23 (1999) 7.
248 N. Dfucci, N. De Giovanni, M. Chiarotti and S. Scarlata, ForensicSci. Int., 119 (2001)
318.
249 N. Nagasawa, M. Yashiki, Y. Iwasaki, K. Hara and T. Kojima, Forensic Sci. Int., 78
(1996) 95.
250 M. Yashiki, T. Kojima, T. Miyazaki, N. Nagasawa, Y. Iwasaki and K. Hara, Forensic
Sci. Int., 76 (1995) 169.
251 H.L. Lord and J. Pawliszyn, Anal. Chem., 69 (1997) 3899.
252 I. Koide, O. Noguchi, K. Okada, A. Yokoyama, H. Oda, S. Yamamoto and H. Kataoka,
J. Chromatogr.B, 707 (1998) 99.
253 J. Liu, K. Hara, S. Kashimura, M. Kashiwagi and M. Kageura, J. Chromatogr.B, 758
(2001) 95.
254 J. Liu, K. Hara, S. Kashimura, M. Kashiwagi and M. Kageura, J. Chromatogr.B, 758
(2001) 95.
255 F. Centini, A. Masti and I.B. Comparini, ForensicSci. Int., 83 (1996) 161.
256 C. Battu, P. Marquet, A.L. Fauconnet, E. Lacassie and G. Lachatre, J. Chromatogr.
Sci., 36 (1998) 1.
257 C. Jurado, M.P. Gimenez, T. Soriano, M. Menedez and M. Repetto, J. Anal. Toxicol.,
24 (2000) 11.
258 H.G. Ugland, M. Krogh and K.E. Rasmussen, J. Chromatogr.B, 701 (1997) 29.
259 S.-W. Myung, H.-K. Min, S. Kim, M. Kim, J.-B. Cho, K. Taek-Jae, J. Chromatogr.B,
716 (1998) 359.
260 H.G. Ugland, M. Krogh and K.E. Rasmussen, J. Pharm. Biomed. Anal., 19 (1999)
463.
261 A. Namera, M. Yashiki, J. Liu, K. Okajima, K. Hara, T. Imamura and T. Kojima,
Forensic Sci. Int., 109 (2000) 215.
262 Y. Makino, T. Takagi, S. Ohta and M. Hirobe, Chromatography, 18 (1997) 195.
263 T. Kumazawa, X.-P. Lee, K. Sato, H. Seno, A. Ishii and 0. Suzuki, Jpn. J. Forensic
Toxiol., 13 (1995) 182.
264 T. Watanabe, A. Namera, M. Yashiki, Y. Iwasaki, T. Kojima, J. Chromatogr.B, 709
(1998) 225.
265 T. Kumazawa, K. Sato, H. Seno, A. Ishii and 0. Suzuki, Chromatographia,43 (1996)
59.
266 A. Ishii, H. Seno, T. Kumazawa, K. Watanabe, H. Hattori and 0. Suzuki, Chroma-
tographia, 43 (1996) 331.
267 F. Musshoff, H. Junker and B. Madea, J. Anal. Toxicol., 24 (2000) 372.

832
268 E.H.M. Koster, C. Wemes, J.B. Morsink and G.J. de Jong, J. Chromatogr. B, 739
(2000) 175.
269 E.H. Koster, C. Wemes, J.B. Morsink and G.J. de Jong, J. Chromatogr. B, 739 (2000)
175.
270 E.H.M. Koster, N.S.K. Hofman and G.J. de Jong, Chromatographia,47 (1999) 678.
271 X.-P. Lee, T. Kumazawa, K. Sato and 0. Suzuki, J. Chromatogr. Sci., 35 (1997) 302.
272 A. Namera, T. Watanabe, M. Yashiki, Y. Iwasaki, T. Kojima and J. Anal. Toxicol., 22
(1998) 396.
273 S. Ulrich, J. Martens, J. Chromatogr. B, 696 (1997) 217.
274 T. Kumazawa, X.-P. Lee, M.-C. Tsai, H. Seno, A. Ishii and K. Sato, Jpn. J. Forensic
Toxiol., 13 (1995) 25.
275 K. Jinno, M. Kawazoe and M. Hayashida, Chromatographia,52 (2000) 309.
276 M. Krogh, K. Johansen, F. Tonnesen and K.E. Rasmussen, J. Chromatogr. B, 673
(1995) 299.
277 M. Nishikawa, H. Seno, A. Ishii, O. Suzuki, T. Kumazawa, K. Watanabe and H.
Hattori, J. Chromatogr. Sci., 35 (1997) 275.
278 H. Seno, T. Kumazawa, A. Ishii, M. Nishikawa, K. Watanabe, H. Hattori and 0.
Suzuki, Jpn. J. ForensicToxicol., 14 (1996) 30.
279 S. Ulrich, S. Kruggel, H. Weigmann and C. Hiemke, J. Chromatogr.B, 731 (1999)
231.
280 T. Kumazawa, K. Watanabe, K. Sato, H. Seno, A. Ishii and 0. Suzuki, Jpn. J.
ForensicToxicol., 13 (1995) 207.
281 H. Seno, T. Kumazawa, A. Ishii, M. Nishikawa, H. Hattori, O. Suzuki, Jpn. J.
ForensicToxicol., 13 (1995) 211.
282 B.J. Hall, A.R. Parikh and J.S. Brodbelt, J. ForensicSci., 44 (1999) 527.
283 B.J. Hall, M. Satterfield-Doerr, A.R. Parikh and J.S. Brodbelt, Anal. Chem., 70
(1998) 1788.
284 A.C.S. Lucas, A.M. Bermejo, M.J. Tabernero, P. Fernandez and S. S.-Rossi, Forensic
Sci. Int., 107 (2000) 225.
285 M. Yashiki, N. Nagasawa, T. Kojima, T. Miyazaki and Y. Iwasaki, Jpn. J. Forensic
Toxicol., 13 (1995) 17.
286 F.Y. Guan, H. Seno, A. Ishii, K. Watanabe, T. Kumazawa, H. Hattori and 0. Suzuki,
J. Anal. Toxicol., 23 (1999) 54.
287 A. Ishii, H. Seno, T. Kumazawa, M. Nishikawa, K. Watanabe, H. Hattori and 0.
Suzuki, Jpn. J. Forensic Toxicol., 14 (1996) 228.
288 K.E. Kongshaug, S.P. Bjergaard, K.E. Rasmussen and M. Krogh, Chromatographia,
50 (1999) 247.
289 P. Okeyo, S.M. Rentz and N.H. Snow, J. High Resolut. Chromatogr., 20 (1997) 171.
290 F. Sporkert and F. Pragst, J Chromatogr.B, 746 (2000) 255.
291 D. Poli, E. Bergamaschi, P. Manini, R. Andreoli and A. Mutti, J. Chromatogr.B, 732
(1999) 115.
292 U. Staerk and W.R. Kulpmann, J. Chromatogr.B, 745 (2000) 399.
293 K.J. Reubsaet, H.R. Norli, P. Hemmersbach and K.E. Rasmussen, J. Pharm.Biomed.
Anal., 18 (1998) 667.
294 K. Jinno, T. Taniguchi and M. Hayashida, J. Pharm.Biomed. Anal., 17 (1998) 1081.
295 M. Krogh, H. Grefslie and K.E. Rasmussen, J. Chromatogr.B, 689 (1997) 357.
296 P. Okeyo, S.M. Rentz and N.H. Snow, J. High Resol. Chromatogr., 20 (1997) 171.
297 T. Kumazawa, H. Seno, K. Watanabe, H. Hattori, A. Ishii, K. Sato and 0. Suzuki, J.
Mass Spectrom., 35 (2000) 1091.
298 P.D. Okeyo and N.H. Snow, J. Microcol. Sep., 10 (1998) 551.
299 T. Felix, B.J. Hall and J.S. Brodbelt, Anal. Chim. Acta, 371 (1998) 195.
300 F. Sporkert and F. Pragst, J. Anal. Toxicol., 24 (2000) 316.

833
301 B.J. Hall and J.S. Brodbelt, J. Chromatogr. A, 772 (1997) 275.
302 D.A. Volmer and J.P.M. Hui, Rapid Commun. Mass Spectrom., 11 (1997) 1926.
303 Y. Luo, L. Pan and J. Pawliszyn, J. Microcolumn Sep., 10 (1998) 193.
304 C.M. Lock, L. Chen and D.A. Volmer, Rapid Commun. Mass Spectrom., 13 (1999)
1744.
305 K. Jinno, M. Taniguchi, H. Sawada and M. Hayashida, Analusis, 26 (1998) M27.
306 S. Li and S.G. Weber, Anal. Chem., 69 (1997) 1217.
307 H. Yuan, W.M. Mullett and J. Pawliszyn, Analyst, 126 (2001) 1456.
308 H. Kataoka, H.L. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 257.
309 H. Kataoka, H.L. Lord and J. Pawliszyn, J. Chromatogr. B, 731 (1999) 353.
310 H. Kataoka, H.L. Lord and J. Pawliszyn, Anal. Chem., 71 (1999) 4237.
311 H. Kataoka, H.L. Lord, S. Yamamoto, S. Narimatsu and J. Pawliszyn, J. Microcol.
Sep., 12 (2000) 493.
312 J. Wu, H.L. Lord, J. Pawliszyn and H. Kataoka, J. Microcol. Sep., 12 (2000) 255.
313 J. Wu, H.L. Lord and J. Pawliszyn, Talanta, 54 (2001) 655.
314 H. Yuan, Z. Mester, H.L. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 718.
315 Y. Saito, M. Kawazoe, M. Hayashida and K. Jinno, Analyst, 125 (2000) 807.
316 W. Mullett, P. Martin and J. Pawliszyn, Anal. Chem., 73 (2001) 2383.
317 D.S. Hage, J. Chromatogr. B, 715 (1998) 3.
318 V. Pichon, M. Bouzige and M.-C. Hennion, Anal. Chim. Acta, 376 (1998) 21.
319 J.M. Van Emon, C.L. Gerlack and K. Bowman, J. Chromatogr.B, 715 (1998) 211.
320 V. Pichon, M. Bouzige, C. Miege and M.-C. Hennion, Trends Anal. Chem., 18 (1999)
219.
321 M.-C. Hennion, J. Chromatogr.A, 856 (1999) 3.
322 M.-C. Hennion, C. C.-D.-Coumes and V. Pichon, J. Chromatogr.A, 823 (1998) 147.
323 V. Pichon, L. Chen and M.-C. Hennion, Anal. Chim. Acta, 311 (1995) 429.
324 V. Pichon, L. Chen, N. Durand, F. Le Goffic and M.-C. Hennion, J. Chromatogr.A,
725 (1996) 107.
325 J. Cai and J. Henion, Anal. Chem., 68 (1996) 72.
326 J. Cai and J. Henion, J. Chromatogr.B, 691 (1997) 357.
327 M.L. Nedved, S. H.-Goudarzi, B. Ganem and J.D. Henion, Anal. Chem., 68 (1996)
4228.
328 S. Kerrigan and D.E. Brooks, J. Immunol. Meth., 224 (1999) 11.
329 J. Rohrich, S. Zorntlein and J. Becker, Forensic Sci. Int., 107 (2000) 181.
330 B.A. Rashid, G.W. Aheme, M.F. Katmeh, P. Kwasowski and D. Stevenson, J.
Chromatogr. A, 797 (1998) 245.
331 W. Clarke, A.R. Chowdhuri and D.S. Hage, Anal. Chem., 73 (2001) 2157.
332 M. Dubois, X. Taillieu, Y. Colemonts, B. Lansival, J. De Graeve and P. Delahaut,
Analyst, 123 (1998) 2611.
333 A. Koole, J. Bosman, J.P. Franke and R.A. de Zeeuw, J. Chromatogr.B, 726 (1999) 149.
334 C.K. Holtzapple, S.A. Buckley and L.H. Stanker, J. Chromatogr.B, 754 (2001) 1.
335 S.A. Reuschel, D. Eades and R.L. Foltz, J. Chromatogr.B, 733 (1999) 145.
336 D.S. Hage, Clin. Chem., 45 (1999) 593.
337 L.I. Andersson, J. Chromatogr.B, 739 (2000) 163.
338 B. Sellergren, Anal. Chem., 66 (1994) 1578.
339 L.I. Andersson, A. Paprica and T. Arvidsson, Chromatographia,46 (1997) 57.
340 L.I. Andersson, Analyst, 125 (2000) 1515.
341 B.A. Rashid, R.J. Briggs, J.N. Hay and D. Stevenson, Anal. Commun., 34 (1997) 303.
342 W.M. Mullett and E.P.C. Lai, Anal. Chem., 70 (1998) 3636.
343 W.M. Mullett and E.P. Lai, J. Pharm.Biomed. Anal., 21 (1999) 835.
344 A. Bereczki, A. Tolokan, G. Horvail, V. Horvath, F. Lanza, A.J. Hall and B.
Sellergren, J. Chromatogr. A, 930 (2001) 31.

834
345 C. Berggren, S. Bayoudh, D. Sherrington and K. Ensing, J. Chromatogr. A, 889
(2000) 105.
346 G. Brambilla, M. Fiori, B. Rizzo, V. Crescenzi and G. Masci, J. Chromatogr. B, 759
(2001) 27.
347 C. Crescenzi, S. Bayoudh, P.A.G. Cormack, T. Klein and K. Ensing, Anal. Chem., 73
(2001) 2171.
348 P. Martin, I.D. Wilson, D.E. Morgan, G.R. Jones and K. Jones, Anal. Commun. 34
(1997) 45.
349 P. Martin, I.D. Wilson and G.R. Jones, J. Chromatogr.A, 889 (2000) 143.
350 R.J. Ansell and K. Mosbach, Analyst, 123 (1998) 1611.
351 T.A. Sergeyev, H. Matuschewski, S.A. Piletsky, J. Bendig, U. Schedler and M.
Ulbricht, J. Chromatogr.A, 907 (2001) 89.
352 T. Takeuchi and J. Haginaka, J. Chromatogr.B, 728 (1999) 1.
353 B. Sellergren and L.I. Andersson, Methods: A Companion to Methods in Enzymology,
22 (2000) 92.
354 L.I. Andersson, J. Chromatogr.B, 745 (2000) 3.
355 R.E. Majors, LC-GC Int., 8 (1995) 375.
356 S.M. Lunte and C.E. Lunte, Adv. Chromatogr., (1995) 383.
357 E.C. de Lange, B.A. de Boer and D.D. Breimer, Adv. DrugDeliv.Rev., 36 (1999) 211.
358 Y.L. Chang, M.H. Chou, M.F. Lin, C.F. Chen and T.H. Tsai, J. Chromatogr.A, 914
(2001) 77.
359 T.H. Tsai, H.Y. Kao and C.F. Chen, Biomed. Chromatogr., 15 (2001) 79.
360 K. Johansen, M. Krogh, A.T. Andresen, A.S. Christophersen, G. Lehne and K.E.
Rasmussen, J. Chromatogr. B 669 (1995) 281.
361 R.H. Hernandez, A.J.H. Louter, N.C. van de Merbel, U.A.Th. Brinkman, J. Pharm.
Biomed. Anal., 14 (1996) 1077.
362 K. Johansen, M. Krogh and K.E. Rasmussen, J. Chromatogr.B, 690 (1997) 223.
363 J.D.H. Cooper, D.C. Muirhead, J.E. Taylor and P.R. Baker, J. Chromatogr. B, 701
(1997) 281.
364 M. Karlsson, T. Korkolainen and T. Wikberg, Biomed. Chromatogr., 11 (1997) 54.
365 A. Ceccato, P. Chiap, Ph. Hubert, B. Toussaint and J. Crommen, J. Chromatogr. A,
750 (1996) 351.
366 K. Johanssen and K.E. Rasmussen, J. Pharm.Biomed. Anal., 16 (1998) 1159.
367 G. Higgins, A. Hughes, J. Dutton and N.B. Roberts, Ann. Clin. Biochem., 35 (1998)
534.
368 L.H. Parsons, T.M. Kerr and F. Weiss, J. Chromatogr.B, 709 (1998) 35.
369 F. Sadeghipour and J.L. Veuthey, J. Pharm.Biomed. Anal., 17 (1998) 801.
370 J.R. Veraart, M.C.E. Groot, C. Gooijer, H. Lingeman, N.H. Velthorst and U.A.Th.
Brinkman, Electrophoresis, 19 (1998) 2944.
371 R.M. Mader, M. Brunner, B. Rizovski, C. Mensik, G.G. Steger, H.G. Eichler and M.
Muller, Electrophoresis, 19 (1998) 2981.
372 P.B. Jacobsen, J. Chromatogr.B, 749 (2000) 135.
373 P. Chiap, B. Evrard, M.A. Bimazubute, P. de Tullio, Ph. Hubert, L. Delattre and J.
Crommen, J. Chromatogr. A, 870 (2000) 121.
374 T.H. Tsai and Y.F. Chen, Biomed. Chromatogr., 14 (2000) 274.
375 C. Mislanova, A. Stefancova, J. Oraveova, J. Horecky, T. Tmovec and W. Lindner, J.
Chromatogr. B, 739 (2000) 151.
376 N. Kobayashi, M. Kazui and T. Ikeda, J. Pharm.Biomed. Anal., 21 (2000) 1233.
377 T.H. Tsai, H.Y. Kao and C.F. Chen, Biomed. Chromatogr., 15 (2001) 79.
378 T.H. Tsai, F.C. Cheng, Y.F. Chen and C.F. Chen, J. Chromatogr.A, 914 (2001) 83.
379 Y.L. Chang, M.H. Chou, M.F. Lin, C.F. Chen and T.H. Tsai, J. Chromatogr.A, 914
(2001) 77.

835
380 J.R. Veraart and U.A.Th. Brinkman, J. Chromatogr. A, 922 (2001) 339.
381 S. Palmarsdottir, L. Mathiasson, J.A. Jonsson and L.-E. Edholm, J. Chromatogr. B,
688 (1997) 127.
382 S. Palmarsdottir, E. Thrdarson, L.-E. Edholm and J.A. Jonsson, L. Mathiasson,
Anal. Chem., 69 (1997) 1732.
383 J. Janiszewski, P. Schneider, K. Hoffmaster, M. Swyden, D. Wells and H. Fouda,
Rapid Commun. Mass Spectrom., 11 (1997) 1033.
384 R.S. Plumb, R.D. Gray and C.M. Jones, J. Chromatogr.B, 694 (1997) 123.
385 Y. Shen, J.A. Jonnson and L. Mathiasson, Anal. Chem., 70 (1998) 946.
386 J.A. Jonsson, M. Andersson, C. Melander, J. Norberg, E. Thordarson and L.
Mathiasson, J. Chromatogr.A, 870 (2000) 151.
387 G. Rule, M. Chapple and J. Henion, Anal. Chem., 73 (2001) 439.
388 H. Lingeman and S.J. H.-Oussoren, J. Chromatogr.B, 689 (1997) 221.
389 L.A. Dawson, J. Chromatogr.B, 697 (1997) 89.
390 L. Denoroy, L. Bert, S. Parrot, F. Robert and B. Renaud, Electrophoresis, 19 (1998)
2841.
391 D.K. Hansen, M.L. Davies, S.M. Lunte and C.E. Lunte, J. Pharm. Sci., 88 (1999) 14.
392 D.J. Weiss, C.E. Lunte and S.M. Lunte, Trends Anal. Chem., 19 (2000) 606.
393 L. Stable, Adv. Drug Deliv. Rev., 45 (2000) 149.
394 M.I. Davies, J.D. Cooper, S.S. Desmond, C.E. Lunte and S.M. Lunte, Adv. DrugDeliv.
Rev., 45 (2000) 169.
395 J.A. Jonsson and L. Mathiasson, J. Chromatogr.A, 902 (2000) 205.
396 K.E. Rasmussen, S.P. Bjergaard, M. Krogh, H.G. Ugland, T. Gronhaug, J.
Chromatogr. A, 873 (2000) 11.
397 H.G. Ugland, M. Krogh and K.E. Rasmussen, J. Chromatogr.B, 749 (2000) 85.
398 T.G. Halvorsen, S.P. Bjergaard and K.E. Rasmussen, J. Chromatogr.A, 909 (2001)
87.
399 M.A. Jeannot and F.F. Cantwell, Anal. Chem., 68 (1996) 2236.
400 M.A. Jeannot and F.F. Cantwell, Anal. Chem., 69 (1997) 235.
401 M.A. Jeannot and F.F. Cantwell, Anal. Chem., 69 (1997) 2935.
402 Y. He and H.K. Lee, Anal. Chem., 69 (1997) 4634.
403 L. Zhao and H.K. Lee, J. Chromatogr.A, 919 (2001) 381.
404 M.H. Akhtar, L.G. Croteau, C. Dani and K.A. Elsooud, Spectroscopy, 13 (1997) 33.
405 M.H. Akhtar, M. Wong, S.R.H. Crooks and A. Sauve, Food Addit. Contam., 15 (1998)
542.
406 G.M. Greenway and N. Kometa, Analyst, 119 (1994) 929.
407 Z. Guo, Q. Jin, G. Fan, Y. Duan, C. Qin and M. Wen, Anal. Chem., 436 (2001) 41.
408 C.S. Eskilsson, E. Bjorklund, L. Mathiasson, L. Karlsson and A. Torstensson, J.
Chromatogr. A, 840 (1999) 59.
409 H. Sachs and P. Kintz, J. Chromatogr.B, 713 (1998) 147.
410 K.S. Boos and C.H. Grimm, Trends Anal. Chem., 18 (1999) 175.
411 H. Weigmann, J. Bierbrauer, S. Hartter and C. Hiemke, Ther. DrugMonit., 19 (1997)
480.
412 J.J. Vreuls, A.J.H. Louter and U.A.T. Brinkman, J. Chromatogr.A, 856 (1999) 279.
413 E. Baltussen, P. Sandra, F. David and C. Cramers, J. Microcol. Sep., 11 (1999) 737.
414 T. Benijts, J. Vercammen, R. Dams, H.P. Tuan, W. Lambert and P. Sandra, J.
Chromatogr. B, 755 (2001) 137.
415 J. Vercammen, C. Peres, C. Devos, P. Sandra, F. Vanhaecke and L. Moens, Anal.
Chem., 73 (2001) 1509.

836
Chapter24

Automation of sample preparation for

pharmaceutical and clinical analysis
David A. Wells and Thomas L. Lloyd

24.1 INTRODUCTION

The use of automation has been shown to be very successful for the many varied
sample preparation techniques performed in pharmaceutical and clinical
analysis. The choices for automation range in complexity according to the
particular function to be performed. While traditionally the term "sample prep"
refers to concentration of analyte, exchange of solvent, and/or removal of inter-
fering substances prior to analysis, automation processes exist for a multitude of
supporting functions as well, such as solvent delivery, sample dissolution,
sample aspiration and dispensing, sample reformatting from tubes to plates,
homogenization, capping/uncapping, sealing, and delivery of sample to the
detection system. These individual processes can be combined to form a semi-
automated or fully automated method. Also, the manner in which tasks can be
combined varies from a single benchtop workstation approach to multiple
modules linked with communications software and a robotic arm to shuttle
components, as will be described in this chapter.
There are many motivating factors for automating a specific sample prepara-
tion procedure. Most commonly, processes are automated to achieve higher
throughput. Although throughput considerations often meet or exceed goals,
sometimes they fall short of expectations. Automation is not always faster than
manual operation, but it can reduce hands-on time significantly, thereby freeing
a scientist to perform other tasks. Unattended operation may also allow over-
night runs, in some cases. Freeing an individual from hazardous (e.g., radioactiv-
ity or toxic chemical usage) and/or mundane tasks is another valid reason for
implementing automation. The automation of most processes has been shown to
bring a degree of reproducibility and quality to the results that cannot be real-
ized among different workers each performing the method manually. This com-
monality in the approach lends itself to easier assay duplication or transfer.
Automation can often facilitate troubleshooting by removing any manual vari-
ability from processing. By then varying the process parameters in a controlled

ComprehensiveAnalytical Chemisty XXIXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
manner as documented in the software, an experimental method can be devel-
oped or optimized. In addition, sample tracking can be ensured via bar code read-
ers or various instrument feedback mechanisms (e.g., gravimetric, optical,
resistance) and process documentation is retained. Sometimes the task required
for sample prep is too miniaturized or complex for manual processing and having
an instrument with computer software calculate and perform the desired func-
tions allows an operator with a lower skill level to perform the work. Addi-
tionally, automation can be viewed as a strategic investment allowing rapid
implementation of new technology (e.g., test tube to 96-well microplates to
denser well formats with more flexible systems) and is often best utilized with
automated processes. An important overall goal for implementing automation
into a laboratory workflow is greater employee job satisfaction.
Recent focus on the importance and prevalence of automation has resulted in
an increase in the number of focused conferences, journals and news sources.
Several scientific conferences are now held yearly that prominently feature
laboratory robotics and automation, e.g., ISLAR (International Symposium on
Laboratory Automation and Robotics), Lab Automation, EuroLab Automation,
MipTec-ICAR (International Conference on Microplate Technology, Laboratory
Automation and Robotics) and the Society for Biomolecular Screening (SBS)
Conference. Automation-specific news and literature can be found in journals
(e.g., Journal of the Association for Laboratory Automation and Laboratory
Automation News, both sponsored by the Association for Laboratory Automa-
tion) and via LRIG, a special interest group (Laboratory Robotics Interest
Group) that maintains regional chapters and hosts a discussion group on the
Internet.
There are many disciplines within pharmaceutical analysis that utilize
automated processes for sample preparation. Bioanalysis (the quantitative
determination of drugs and their metabolites in biological fluids) is used very
early in the drug development process to provide support to drug discovery
programs on the metabolic fate of chemicals in animals as well as in living cells in
an in vitro setting, and its use continues throughout the preclinical and clinical
drug development phases. The automation of bioanalysis is perhaps most varied
among all pharmaceutical disciplines as many different sample preparation
techniques are employed. These techniques will thus be used as representative
models for describing applications and techniques for automation of sample
preparation in this chapter.
The discipline of clinical analysis has invested heavily in automated
processes for sample preparation. After a drug is approved and marketed,
therapeutic drug monitoring laboratories may utilize many of the same auto-
mated bioanalysis techniques as used in pharmaceutical analysis when it is
important to relate the plasma concentrations of a drug to a patient's therapeu-
tic response and/or adjust drug dosage to minimize adverse effects or toxicity.
The clinical laboratory utilizes various testing workstations and analyzers that
perform bioanalysis as well as many other sample preparation procedures

838
automatically, offering walk-away automation solutions. In addition, clinical
laboratories are moving toward modular automation in an effort to improve cost
efficiency. Clearly, the role of automation for sample preparation is wide-
ranging and encompasses many different functions within both pharmaceutical
and clinical analysis.
The objectives for this chapter are to: (a) discuss strategies and attributes to
consider when implementing automated processes into analytical laboratories;
(b) overview specific applications demonstrating the extensive use of automation
in pharmaceutical bioanalysis and clinical analysis; and (c) assess and compare
automated systems, as well as offer a glimpse into future automation trends.

24.2 STRATEGIES AND ATTRIBUTES OF AUTOMATION

In today's modern laboratories, automation has become a familiar sight. The

type and extent of automation varies greatly between organizations as well as
between different labs within a single institution. For example, there are instru-
ments that perform only a specific task, instruments that are flexible in their
tasks, multiple tasks integrated into an application-specific workstation, semi-
automated solutions, and fully automated walk-away configurations which often
utilize a robotic arm or gripper unit. This section describes the many attributes
of automated sample preparation and the important role they play in finding the
'right' equipment for a specific laboratory and its personnel [1].
When reviewing laboratory automation options for sample preparation, it is
important to carefully evaluate many attributes in order to achieve the best
value and outcome for the task to be performed. The following list of considera-
tions may be used as a framework for appraising the utility of existing systems in
a laboratory, guiding new purchase decisions and formulating an overall lab
automation strategy. Clearly there are explicit tradeoffs that should be weighed
with respect to the constraints and needs of each unique situation. In consider-
ation of both the high cost and long term impact (good or bad) of lab automation
choices, these attributes should be evaluated very carefully prior to choosing the
automation type. Purchasers often develop a checklist of performance bench-
marks and selection criteria to help guide their decision, combined with a
hands-on evaluation or demonstration of the tools.

(1) Degree of automation

A fairly broad scope must be applied to offer a thorough approach in defining and
evaluating all the different types of automation that can be applied for sample
preparation. The following classification is provided:
(a) function-specific workstations or modules;
(b) on-line instruments including autosamplers with preparative and/or
liquid addition and transfer capabilities, switching valves, and on-line extraction
instruments;
(c) XYZ liquid handlers with various numbers of tips (probes);

839
 (d) robotic workstations having a gripper arm that can pick and place
labware;
(e) full robotic systems linking multiple workstations and/or modules with
other peripheral devices.
Although discrete classifications are offered for the purpose of noting closely
related instrument equivalents, one should view this attribute more as a
continuum.

(2) Cost and budget

Implementation of laboratory automation requires a significant investment of
both time and capital. In today's research environment, budgets are one of the
constraining factors in what system(s) to purchase. Nonetheless, it is still wise to
consider which equipment has the best cost/benefit ratio for a laboratory. The
expected duration of the project will impact return on investment decisions,
whereas the opportunity cost of not having certain capabilities sooner will
influence the amount to spend. The graph of 'diminishing returns' has become
well known throughout the automation industry (Fig. 24.1). Quite simply, it
depicts how automating part of a process, usually the rate-limiting or labor
intensive step(s), gives the best return on an investment. As a process continues
to be automated, the costs usually increase whereas the incremental benefits
decrease. This observation should not deter people from evaluating the benefits
of fully automated systems, as many applications are worthy, but it suggests
perhaps a more conservative stepwise approach in which continuous cost and
overall process analyses might be done. Although a gradual implementation
involves less risk, it can also result in greater overall investments of time and
money for those processes that do end up fully automated.

(3) Type and Quantity of Work

Certainly a very important consideration is the suitability of an automated
system to the type and quantity of work being performed. How many samples

100%

. 75 %

E
o 0%

25 %

50 100 150 200

Cost (US$ x 1000)
Fig. 24.1. Graph of the extent of automation versus cost.

840
need to be analyzed? How many different assays need to be accommodated and
how frequently will changeovers occur? How similar are the different assays?
How challenging are the analytical requirements of the different assays? Will
the system be used for method development? What level of process documenta-
tion is required and how much of that information should be captured by the
automated system? These are all fundamental questions to be posed when
considering specific needs with regard to an automated system.

(4) Functionality
Consider the extent to which a system automates a process (e.g., semi-
automated versus fully automated). Determine the exact elements or steps
which are to be performed on a particular instrument. For example, the process-
ing steps that can be used to evaluate a system's functionality might include:
capping/uncapping samples, weighing, loading or exchanging consumables,
conditioning, sampling, washing, eluting, vortexing or mixing, evaporating,
filtering, derivatizing, incubating, reagent additions, vacuum manifold control,
centrifugation, bar code labeling and reading, sealing and injecting.

(5) Flexibility and Adaptability

Both the hardware and software architecture determine the versatility of a
system. In addition to assessing the number of process steps a system can
automate, also consider whether it can handle other processes to (a) encompass a
broader application, (b) extend the useful life of the instrument by allowing
adaptation to new technologies, or (c) postpone obsolescence by allowing for
upgrades. A system may offer added value by supporting different types of
sample extraction techniques (e.g., liquid-liquid extraction, protein precipita-
tion, dialysis, solid-phase extraction) or smaller related processes (e.g., prepara-
tion of standards and quality controls, sample dilutions, or sample transfer from
one format to another) that will increase the usage and value of an instrument.

(6) Space and Utilities

Laboratory automation can vary in size from small portable benchtop units to
very large 'on the floor' systems. Evaluate space and specialized renovation
requirements for each option. Good planning involves more than just having a
place to put the equipment. The location of a system can influence whether or
not, or at least how often, staff makes use of it. If a system is being targeted for
shared use, try to place it in a common, accessible area. Space planning should
also allow for placement near other key instruments (e.g., an LC/MS/MS near an
on-line analysis option or space for HPLC systems on mobile carts) depending on
overall process plans. Different equipment will require different utilities such as
sufficient number and type of electrical outlets, an uninterruptible power
supply, computer network and data acquisition ports, compressed air and gases,
water, drains and sinks, ventilation, and phone jacks. Space usage and
renovation costs should enter into a cost assessment.

841
(7) Degree of Difficulty
The learning curve associated with operating an automated sample preparation
system will vary depending on both the skill level of staff and the complexity of
the equipment and/or its software. Simpler systems usually begin to pay small
dividends almost immediately, whereas complex robotic systems may take
months or even years to fully implement with the realization of greater returns.
Equipment installation, operation, maintenance and continuing system devel-
opment roles and timeframes should be clearly defined. An evaluation of the
in-house skills and resources (chemistry, automation, programming and engi-
neering) required for each automation option should be made. This assessment
should then be matched with existing talent available, training options, and
contract support options. The more complicated workstations and robotics
systems usually require the identification of a champion or system administra-
tor early in the process.

(8) Vendor Support

The quality of vendor support networks can vary considerably. Their relative
importance depends on how complex a system is purchased and how much
reliance is made on the vendor's support to install, operate, maintain and
continually develop the system. The real cost of each automated system should
consider all associated warranties, training, documentation, technical and
applications support, continuing service and the upgrade policy.

(9) Ruggedness
It is difficult to determine a system's reliability prior to purchasing and using the
equipment in support of daily lab activities. Ask the manufacturer to supply
references of users who may give a candid review of their experiences. Some
companies are willing to offer the use of a demonstration unit for a hands-on
evaluation prior to making a purchase. The process of acquiring and installing
the demonstration unit will often provide insight into the quality and availabil-
ity of the vendor's technical and service support personnel for this system. As
each individual step of the automated process is evaluated, keep in mind that the
system will only be as reliable as the weakest link in the process. For more
complicated or customized systems, it is wise to agree on performance specifica-
tions as part of the purchase agreement. This performance expectation shared
up-front serves as a good, and fair, source of protection for both the customer
and the manufacturer.

(10) Speed, Sample Capacity and Sample Throughput

Depending on the quantity of work supported, the speed and sample capacity of a
system may be important. A system's sample capacity can vary from tens to
thousands of samples or more. The processing speed is highly dependent on the
application and the processing mode. For a typical extraction application,
sample preparation times can vary from four samples per hour to 400 or more

842
samples per hour, depending on the system. Sometimes, too much attention is
placed on the speed of the instrument, rather than its functionality. In such
cases, it is not uncommon for laboratories to increase the speed of a single link in
a process only to realize that they have merely shifted the bottleneck and had
little impact on the overall productivity of the application.

(11) ProcessingMode
Automated systems also vary in the manner in which samples are processed,
which can influence overall system speed and throughput. Some systems process
each sample individually in a serial or staggered-serial fashion. This mode of
operation may afford reproducibility advantages such as gravimetric
confirmation of liquid transfers, unique control and monitoring of each sample
processed, and uniform sample history or exposure. Other systems process
samples in parallel using multiple tip liquid handling transfers. The processing
mode frequently involves a tradeoff of giving up some control of sample
processing in return for greater speed. A good example of increasing parallel
processing speed involves moving from a single sample 1-tip serial mode to a
96-sample parallel mode with the implementation of a 96-well microplate format
and 96-tip liquid handler for sample preparation. Other combinations offering
improved throughput also exist where a system concurrently extracts samples
serially on multiple, distinct processing modules.

(12) On-line vs. Off-line Analysis

Many automated systems operate as off-line sample preparation stations. The
term "off-line" refers to performing the sample prep procedure independent of
the chromatographic analysis. Conversely, "on-line" systems refer to the inte-
gration of sample preparation and analysis functions in a single, linked system.
Off-line operation provides the most flexibility, by minimizing coordination
issues with analytical devices and allowing open access by multiple users. One
system can prepare samples for multiple analytical systems or vice versa. Such
an approach also avoids the potentially difficult issue of interfacing equipment.
Some automated systems afford the opportunity to operate either off-line or
on-line. Advantages of operating in an on-line manner include a reduced cycle
time for the overall process (with analysis taking place concurrently with sample
preparation) and a seamless automated process (perhaps free of manual inter-
vention steps such as transfers to an evaporator or analytical system). Still other
systems offer direct elution (chromatography) or introduction (e.g., counter or
plate reader) on-line with the detection system. In the case of chromatographic
analyses, these systems offer an additional benefit over other on-line systems,
introducing the entire sample mass onto the analytical system instead of leaving
some fraction of an eluate behind. However, such systems can be restrictive in
that they do not allow for repeated injections of a given extracted sample and
they require that the elution reagent be compatible with the mobile phase of the
analytical system.

843
(13) ProcessingFlow Control
Another important variable to consider is the means by which solvents are
passed through an extraction or filtration device. Processing flow control can
involve using either vacuum, positive pressure or centrifugation, controlled
either manually or by the automated system. Many automated systems use
vacuum manifolds since they are relatively simple, inexpensive and their small
size allows easy integration with the automation deck or its hardware. Other
automated systems complement or substitute for vacuum processing by using
positive pressure flow control or centrifugation. There is ample evidence of
improved analytical precision using positive pressure as opposed to vacuum flow
control for solid-phase extraction applications [2,3]. Centrifugation also offers
varying control and has been applied for a number of techniques. Different
automated systems offer varying degrees of flow control ranging from requiring
manual intervention to simple on/off control and to finer, varying control with
pulsing and multiple stepped levels of applied force. Some systems also offer a
means for attempting to identify sorbent bed blockages by liquid level detection.

(14) ContainerFormat Standardization

Decisions about container standardization are associated with any choice of an
automated system. There are benefits in supply process when standardizing on
one or several container formats. However, it is important to consider the
flexibility in handling other consumable formats (e.g., cartridges, cassettes,
microplates, chips) that currently exist or may evolve and how they would be
used with the system. Additionally, consider what restrictions may be associated
with a particular system in terms of compatible reagents, vendor material
selection, sample working volumes and price of consumables.

24.3 APPLICATIONS FOR AUTOMATION

24.3.1 Background

The earliest efforts to automate sample preparation concentrated on the

analytical product testing and clinical bioanalysis application areas, targeting
repetitive functions with the largest numbers of samples. An early example of
automation being applied for such tasks was the proliferation of autosamplers
for chromatography systems, which replaced manual syringe injections with
switching valves. In the early 1980s, computerized data acquisition, processing,
and reporting tools were introduced. Shortly thereafter, the first commercial
laboratory robots and automated liquid handlers were introduced. By the end of
the 1980s, such robotic systems were versatile enough to perform all types of
sample preparation techniques including protein precipitation, liquid-liquid or
solid-phase extraction, filtration, dissolution and homogenization. They often
shifted the bottleneck over to the chromatographic analysis, with typical 5-15
min run times. As a result, these systems tended to process samples serially or in

844
a staggered serial manner, especially for on-line applications. At this stage, there
were liquid handlers with multiple tips (e.g., 4 or 8) for performing at least
certain steps in parallel.
In the early 1990s, the attention of pharmaceutical automation efforts was
redirected to early drug discovery applications by the introduction of systematic
research approaches such as high throughput screening and combinatorial
synthesis. These approaches sought to harness iterative, automated procedures
executed in parallel, microplate array format. By the mid-1990s, the concept of
parallel sample preparation and the automated tools created for microplate
formats were beginning to be adapted for applications further along in the drug
development cycle. The first solid-phase extraction microplates became
commercially available in 1996 and scientists continued to explore performing
other types of sample prep in parallel microplate format as well. Such principles
even spread to dosing animals in parallel with multiple test compounds "n-in-1"
[4] to reduce animal handling or, alternatively, pooling of analytical samples [5].
Automation was further driven by advances in detection systems, namely
liquid chromatography interfaced with tandem mass spectrometry (LC/MS/MS),
and their more widespread affordability within the last five years. These systems
have allowed for the processing of a greater number of samples than ever before
[6]. Although mass spectrometry has not obviated the need for matrix removal
and chromatographic separation, it has loosened the requirements for resolving
multiple analytes prior to detection. These LC/MSMS systems have become
prevalent in the pharmaceutical industry and have recently begun to appear in
the larger clinical reference laboratories for some esoteric analyses. As a result,
laboratories are now using higher throughput sample preparation schemes
combined with automation to better realize the potential of their LC/MS/MS
systems. Because of the cost and size of such analytical instruments, more recent
automated systems have tended to be off-line. Serial processing systems in many
cases are no longer able to keep pace with the faster analysis times. However, in
many instances, sample prep platforms based on parallel processing have again
reached a point where they are capable of supplying samples for analysis on
multiple LC/MS/MS systems.

24.3.2 Automation for bioanalysis

The accurate and precise determination of drugs and their metabolites in

biological fluids is an important task performed to support therapeutic drug
monitoring within clinical drug analysis and the entire pharmaceutical drug
development process. Laboratories typically process from tens to hundreds of
thousands of samples per year in the course of their work. The preparation of a
biological sample before LC/MS/MS analysis is an important task which has
often been the rate-limiting step in the overall bioanalytical process. Automation
of sample preparation is thus a necessary goal which has been approached by
many different routes. Since there are many types of sample preparation

845
techniques used for both clinical and pharmaceutical analysis, as described in
the chapter by Kataoka and Lord, there are a variety of automation tools
available. The bioanalytical sample preparation techniques in common use
today include:
- protein precipitation;
- liquid-liquid extraction;
- off-line solid-phase extraction using cartridges and microplates;
- on-line solid-phase extraction techniques using microplates, cartridges and
small diameter LC columns;
- on-line sample preparation techniques such as turbulent-flow chromato-
graphy, restricted access media and immunoaffinity extraction.
Automated liquid handling workstations using 4 or 8 tips are commonly used for
bioanalytical sample preparation tasks. Examples of these workstations include
the MultiPROBE IIM (Packard Bioscience Company, Meriden, Conn., USA)
M
(Fig. 24.2), Genesis" (Tecan Group Ltd., Mannedorf, Switzerland), Biomek"'
2000 (Beckman Coulter, Fullerton, Calif., USA), Speedy" (Zinsser Analytic,
Frankfurt, Germany), and the Model 215"' (Gilson, Middleton, Wisc., USA).
Note that the MicrolabM Series of liquid handling instruments (Hamilton
Instruments, Reno, Nev., USA) is available in a 1-tip version as well as in
multiple tip versions for liquid handling tasks. All of these instruments are
interfaced to a computer and are software-controlled for precise adjustment of
parameters. These workstations are also used for addition of reagents to adjust
pH, addition of internal standard, dilutions and liquid transfers (e.g., from tube
to tube, plate to plate, tube to plate, and tube or plate to autosampler vials).
Higher throughput liquid aspiration and dispensing is achieved by 96-tip
pipettors, such as the Quadra96" (Tomtec, Hamden, Conn., USA), Multimek"'
(Beckman Coulter) (Fig. 24.3), Personal PipettorTM 96 (Apricot Designs,
Monrovia, Calif., USA), Sciclone'" (Zymark, Hopkinton, Mass., USA) or

Fig. 24.2. Example of a 4-tip liquid handling workstation, the MultiPROBE II (Packard
Bioscience Company). Photo reprinted with permission from Packard Bioscience.

846
Fig. 24.3. Example of a 96-tip liquid handling workstation, the Multimek96'" (Beckman
Coulter). Photo reprinted with permission from Beckman Coulter, Inc. Copyright 2001.

Hydra96TM (Robbins Scientific, Sunnyvale, Calif., USA). Systems capable of

automating more steps of the sample prep process then involve linking such
systems with physical movement capability by adding on plate stackers and
shuttles, or a robotic arm.

Proteinprecipitation
A fast and simple method of sample preparation is protein precipitation. This
non-selective technique involves adding a water-miscible organic solvent (e.g.,
acetonitrile) to the biological matrix (usually in a 3:1 ratio, v/v) to denature and
precipitate the proteins. The proteins are removed by centrifuging or filtering
and an aliquot of the diluted supernatant is injected for analysis. Protein
precipitation is often used in drug discovery support when an analytical method
needs to be developed quickly and concentrations of analyte are relatively high.
In this situation, the generic extraction approach, with its associated disadvan-
tage of matrix carryover causing ion suppression, is a more important criterion
than achieving greater sensitivity by using a more selective technique. Typical
sample matrices seen in drug discovery support that are used with protein
precipitation are plasma, serum and tissue homogenates.
Bioanalytical sample prep using protein precipitation is usually a semi-
automated procedure, accomplished in part by having a liquid handling worksta-
tion aspirate and dispense volumes of organic solvent to samples contained in
test tubes or microplates. The liquid handler, if desired, can also aspirate fixed or
varying volumes of sample from source tubes into the tube or well used for the
precipitation; disposable tips are frequently used to avoid carryover. Additional

847
tasks that can be automated include preparation of calibration and quality
control standards, and internal standard additions. The tubes or plates contain-
ing organic solvent and sample then are capped/sealed, vortexed and then
centrifuged to pellet the protein. An aliquot of the supernatant can be aspirated
using the same liquid handling instrument. The volume aspirated should not
disrupt the protein pellet upon removal, which can be facilitated by choosing an
instrument with liquid level detection and tracking. This aliquot is transferred
to a clean vial or microplate, often containing a small volume of buffer that is
capped, vortexed and ready for injection into the analytical system for separation
and detection. Note that another option instead of dilution of supernatant
(depending on volume used) is evaporation, followed by reconstitution. Although
this procedure is simple and can be done manually, automation is introduced
when the number of samples to be extracted becomes large and the procedure is
to be performed routinely. An application reported by Watt details a protein
precipitation procedure using the Biomek 2000 in which throughput increased
to 400 samples per day in conjunction with LC/MS/MS analysis [7].
Filtration is another approach to protein precipitation and it can provide a
more easily automated sample prep procedure from start to finish. In this
application, a filtration 96-well plate is assembled on top of a vacuum manifold,
and a clean collection plate is placed inside the manifold to capture the filtrate. A
liquid handling workstation delivers acetonitrile and plasma to the wells of the
filtration plate. Precipitation occurs instantly and vacuum is applied to collect
the filtrate. Proteins remain on top of the filter and are discarded with the plate.
The filtrate contained in the underlying microplate can be evaporated before
reconstitution and analysis, or an aliquot of the diluted supernatant can be
injected directly.
It has been demonstrated that the order of addition of reagent and sample
can influence the efficiency of protein precipitation in filter plates, as well as the
degree of mixing or vortexing occurring when these two items are added
together. Also of note is that not all filter varieties available in plates possess the
necessary porosity to capture all proteins and the necessary resistance to organic
solvents to retain the organic solvent until vacuum or centrifugation is applied.
Thus, two common problems seen using filter plates are the breakthrough of
proteins into the filtrate and organic solvent that drips through prematurely.
A resourceful approach to obtain adequate efficiency of mixing is to first aspirate
acetonitrile, followed by an air gap, and then aspirate plasma in the same
disposable tip, as described by Biddlecombe [8]. This mixture is then forcibly
dispensed into wells of the filtration plate, thus avoiding a manual intervention
to vortex the plate. A condition of using the liquid handler for this filtration
approach is that it be able to work with a vacuum manifold on its deck and,
optionally, be able to sequentially aspirate solvents separated by an air gap. An
alternative to vacuum is to use centrifugation to filter the precipitated mixture.
This approach can be realized by placing a filter plate, mated with a collection
plate underneath, onto the deck of the liquid handler. Liquids are delivered

848
directly into the wells as described above, but then the plate combination is
manually inserted into a centrifuge for the filtration step.

Liquid-liquid extraction
Liquid-liquid extraction (LLE) involves mixing an immiscible organic solvent
with an aqueous biological sample (e.g., plasma) to selectively extract the
analyte(s) into the organic phase. The organic layer is transferred, evaporated to
dryness, and reconstituted prior to analysis. While this methodology provides
clean extracts, it has traditionally been difficult to automate, until recently.
There are three common approaches to automating LLE in a fully or semi-
automated setting: LLE performed in tubes, LLE in microplates, and LLE using
solid-supported media in flow-through columns or wells.
Typical 4- and 8-tip workstations can semi-automate LLE in test tubes but
manual intervention is often required for capping/vortexing. The detection of
the interface between the aqueous and organic layers is important in order to
aspirate the organic phase while leaving the aqueous phase behind in the
extraction tube. Removal of the organic layer from on top of the aqueous layer in
a tube can be performed by aspirating from a fixed depth in the tube. Freezing of
the aqueous layer at the tube bottom can aid in this transfer. The depth required
is determined by trial and error or, upon freezing, the upper organic layer can be
manually poured off into a clean tube. Such an application utilizing a liquid
handling instrument (MultiPROBE II) with liquid dispensing into test tubes is
reported by Jemal [9].
The technology now exists to automatically detect this phase boundary, and
an example of a single probe workstation performing automated LLE in this
manner is the Myriad ALLEX (Mettler Toledo Bohdan, Mundelein, Ill., USA).
The ALLEX is able to use different sizes of test tubes and automatically detect
the liquid-liquid phase boundaries, performing up to 60 separations per hour. It
also displays multiple and back extraction capability. A special note is that if the
organic layer is denser than the aqueous layer and remains at the well bottom,
the aqueous layer must be aspirated from the sample at a fixed depth, leaving the
organic layer in the well bottom for evaporation, reconstitution and analysis. In
this situation, the Myriad ALLEX workstation provides an efficient and superior
solution.
A higher throughput approach to performing LLE is to use microplates
instead of test tubes. Since the volume capacity of microplates (2 ml or less) is
smaller than tubes, sample and solvent volumes are reduced (e.g., 100 Al plasma
to 800 1I organic solvent). Multiple probe liquid handlers can be utilized for
sample and solvent aspiration/dispensing, and are especially convenient to
reformat samples from source tubes into a microplate. The microplates are
sealed and then manually vortexed to perform the mixing step. Liquid handlers
with 4-, 8- and 96-tips can all perform this action using microplates. The Myriad
ALLEX cannot use microplates, so aspiration of the organic layer is performed
using fixed depth techniques as described above, followed by evaporation and

849
reconstitution of the organic layer, all in the microplate format. The 96-tip liquid
handling instruments, e.g., Quadra96 (Tomtec), provide the highest throughput
for LLE in microplates and there are published applications of their use [10,11].
An important safety consideration for LLE applications is to vent the volatile
organic solvent vapors by either placing the instrument in a fume hood or
locating trunks from the ceiling near the work area to remove the vapors.
A more completely automated approach to LLE uses flow-through 96-well
microplates filled with inert diatomaceous earth particles of high surface area
(hydromatrix), available from numerous vendors (e.g., Varian, Harbor City,
Calif., USA; Orochem Technologies, Westmont, Ill., USA; and International
Sorbent Technology Ltd., South Wales, UK). When using the solid-supported
particles to perform LLE, plasma with buffer solution is added to the plate wells
and allowed to partition for a few minutes on the particle surface. A hydrophobic
filter on the bottom of each well prevents the aqueous phase from breaking
through into the collection plate. Organic solvent is then added to the wells, the
analyte partitions into the organic solvent as it flows through, then analyte is
eluted through the device, collecting in a container underneath. Multiple probe
and 96-tip liquid handlers can automate this entire extraction procedure for
limited volumes using microplates, as detailed in the report by Peng [121.
Solutions do flow through the particle bed via gravity, which maximizes the time
for interaction and partitioning of analyte from the aqueous into the organic
phase, so little or no vacuum is necessary. Advantages of using solid-supported
LLE include the avoidance of capping/mixing steps and the potential formation
of emulsions. A special note with the use of these particle beds is that the
aqueous phase is not retrievable.
Solid-supported LLE products are also available as individual cartridges,
essentially syringe barrels with a frit on the bottom and containing the diatom-
aceous earth particles. The cartridge format of diatomaceous earth is usually
used with gravity flow. One example of an application that efficiently uses these
solid-supported LLE cartridges in a fully automated mode is a custom configured
Zymate XP robotics system (Zymark). Zymark provides the Zymate"' robot
system, in which a central robotic arm is the focal point for a range of activities
custom configured about its perimeter (Fig. 24.4). The XP robot is now the third
generation laboratory robot for such an automation system. The robot
incorporates technology for interchangeable hands to carry various tubes and
containers, and to perform functions such as gripping, weighing, pipetting,
vortexing, centrifuging and evaporating. Tactile sensing capability is included in
the robot hand and all robot axes have optical encoders to verify successful
completion of functions. An application described by Alianti [13] uses a unique
tapered cartridge format ("LRC" configuration) of diatomaceous earth. The
robotics system is equipped with an analytical balance, an extraction station, a
1 ml pipet hand, a general purpose hand, a test tube dispenser, a vortex station,
an evaporating station, and an LC sipping section interfaced to an autosampler.
The robot transfers serum or plasma to an unconditioned cartridge, internal

850
Fig. 24.4. Example of a Zymate XP robotic system (Zymark) configured for performing LLE
using solid-supported cartridges. Photo courtesy of DuPont Pharmaceuticals.

standard is added, ethyl acetate is added and elution occurs using two 3 ml
aliquots, eluant is evaporated to dryness and the sample is reconstituted and
transferred to an autosampler vial for injection.

Solid-phase extraction: off-line/cartridges

Solid-phase extraction (SPE) is a specific type of sample preparation in which an
analyte, contained in a liquid phase, comes in contact with a solid phase (sorbent
particles contained within a column or a disk) and is selectively adsorbed onto
the surface of that solid phase. All other materials not adsorbed remain in the
liquid phase and pass through the sorbent particle bed. Generally, a wash
solution is then passed through the sorbent bed to remove possible adsorbed
contaminants from the sample matrix, while the analyte of interest is still
retained on the solid phase. A selective elution step is then performed, in which
the analytes partition away from the solid support into an organic solvent.
Details of this technology have been described by Kataoka and Lord in Chapter
23.
The traditional format for SPE has been single disposable columns and
cartridges (commonly 1 and 3 ml reservoirs for bioanalysis; 6 ml size less often)
filled with a range of masses of solid sorbent silica or polymeric particles (from
25-500 mg) held between two polyethylene frits. Teflon or glass disks embedded
with bonded phase material in cartridges are an alternative to packed particles,
as described by Kataoka and Lord. There are many automation choices that
process liquids through this cartridge format, usually in a serial approach. A
single-probe liquid handler accommodates SPE tubes, e.g., the Microlab SPE
(Hamilton) which applies positive pressure. Other serial processing benchtop

851
instruments produced specifically for SPE sorbent-filled tubes include the
single-tip ASPEC XL (Gilson) and its higher throughput version ASPEC XL4
that processes cartridges using positive pressure with four tips in parallel. The
ASPEC XL can be configured with one or two Rheodyne injection valves
(Rheodyne, Cotati, Calif., USA) to inject the prepared eluate on-line. An applica-
tion using the ASPEC XL that illustrates fully automated solid-phase extraction
coupled on-line with injection is reported by Streel [14].
A more recent and higher throughput Gilson product for SPE is the "SPE
215" system, which eliminates the need for caps on top of the cartridges via use
of an integrated silicone sealing foot. This design seals the tops of cartridges and
applies positive displacement for liquid processing of up to 8 samples at a time
via its 8-tips. An injection module can be added to perform injections of up to 8
samples in parallel (when interfaced with 8 LC systems). In terms of product for-
mat capabilities, the SPE 215 (like the ASPEC XL4) accepts the 1 and 3 ml car-
tridge sizes and is compatible with SPE microplates. When "tab-less" cartridges
are used, a greater density of cartridges can be placed in the available footprint
to retain a 9 mm center to center spacing as in microplates; e.g., 96 "tab-less"
1 ml cartridges can be placed on the rack and elution can be accomplished in a
microplate instead of test tubes.
Another example of automation providing high throughput in the syringe
barrel format for SPE is available via serial processing among multiple stations
(e.g., 10 modules placed side by side) using an instrument called the Rapid-
Trace" (Zymark), shown in Fig. 24.5. Each module processes liquids through
SPE cartridges via positive displacement. This modular design allows for
capacity to be added as demand requires. In terms of throughput, about 50

Fig. 24.5. Example of modular automation for SPE using cartridges, the RapidTrace T'(Zymark).
Photo reprinted with permission from Zymark.

852
samples per hour have been reported by Huang for a quantitative bioanalytical
LC/MS/MS application [15]. An additional application utilizing the
RapidTraceTm approach to automating SPE is illustrated by a forensic urine drug
testing method [16].
With today's rapid pace of drug development, the time required for method
development has been reduced from several weeks to just a few days, so the pri-
ority to automate this necessary task has become much greater. The develop-
ment of solid-phase extraction methods using cartridges was traditionally
performed manually and, once optimized, the methods were transferred to auto-
mation. However, with the advent of the RapidTrace format, method develop-
ment can now be routinely automated. The modular approach of the RapidTrace
makes it ideal for performing method development since the extraction condi-
tions can be manipulated one at a time among different modules, and even
among different samples within the same module. A discussion of its versatility
for accelerated bioanalytical methods development is provided by Xen [17].
A second workstation design using SPE cartridges and configured for meth-
od development purposes is the Spe-ed Wiz (Applied Separations, Allentown,
Penn., USA). This instrument uses positive pressure to process up to 60 samples
per hour. It uses standard cartridges and the software controls the drying times,
flow rates and solvents delivered to each cartridge. By using different sorbent
chemistries in the particle beds, many variables of solid-phase extraction can be
examined at once in an effort to develop and optimize an extraction method. A
good example of method development and optimization using SPE is provided by
Bakklai [18]. A more comprehensive automation solution for bioanalysis, in
terms of both capability and flexibility, is exemplified by a full robotics system.
Such systems frequently offer additional process controls such as a balance to
confirm liquid transfers and afford the opportunity to correct processes on the
fly. An application illustrating the range of activities of a custom configured
Zymate robotics system for performing solid-phase extraction method develop-
ment in bioanalysis is described by Parker [19].
A particular benchtop workstation called duo-PREP' (Pharmaceutical
Technology Ltd., Cambridgeshire, UK) processes liquids through SPE
cartridges in a unique manner. In contrast with conventional liquid processing
through SPE cartridges, in which solutions are passed through a stationary
solid-phase bed from top to bottom and out, duo-PREP operates by the move-
ment of the sorbent bed through each solution. The process is performed as
follows. An array of up to 48 cartridges (specifically duo-PREP SPE cartridges,
as regular SPE cartridges are not compatible) is lowered into an array of
multiwells containing the process solutions. As the cartridge descends into each
well below it, an outer flexible seal engages with the inner wall of the well,
generating a pocket of air above the solution. Once the cartridge inlet enters into
the solution, the liquid is forced upward, through the solid sorbent bed and into
the cartridge. On withdrawal of the cartridge, the arrangement of the seal with
the well creates a partial vacuum on the underside that pulls the solution back

853
out of the cartridge and into the well underneath. Thus, controlled reversible
flow is achieved. This procedure is performed for the condition, load, wash and
elution steps. For elution, clean multiwells are used and eluate remains in the
wells for injection by an XYZ autosampler [20].

Solid-phase extraction: off-line/microplates

The pharmaceutical industry has responded to the challenge of higher
throughput sample preparation by utilizing solid-phase extraction in a 96-well
microtiter plate format. This technique utilizes single blocks or plates having 96
wells that contain disks or packed beds of sorbent particles arranged in an 8-row
by 12-column rectangular matrix [21]. Although the plates can be processed
manually, the use of automated instruments for processing is widely preferred.
The advent of this high throughput sample preparation format, in combination
with automation and the proliferation of LC/MS/MS analytical techniques, have
all occurred to create a superior working environment for the analytical chemist
to be able to meet the demands for faster sample processing and data generation
to support the drug development process.
In order to efficiently use microplates for solid-phase extraction and other
applications, there is first a need for sample reformatting from individual tubes
to a microplate. Various brands of XYZ liquid handlers have proven ideal for this
application, e.g., MultiPROBE II (Packard), Genesis (Tecan), Microlab
(Hamilton), Biomek (Beckman Coulter), Model 215 (Gilson) and Sciclone
(Zymark), among others. The flexible deck design of a liquid handling
instrument allows placement of components (test tube racks and microplates)
and labware (solvent troughs, disposable tips, wash bowl, etc.) on its deck. The
variable tip spacing feature of multiple probe liquid handlers allows them to
expand their tip-to-tip width to aspirate from various test tube sizes, and reduce
the tip spacing width when dispensing into wells of a microplate (9 mm well-to-
well spacing). This sample reformatting step is necessary in order to automate
bioanalysis using a 96-tip liquid handling workstation; e.g.. a MultiPROBE II
can first be used to reformat samples from tubes into a microplate; it also adds
buffer and internal standard to the wells, as required, then the assembled
microplate is taken to a 96-tip liquid handling workstation to perform 96-well
SPE, as detailed by Matthews [22].
The best combination of workstations for throughput and efficiency per-
forming off-line SPE applications is a 96-tip liquid handling workstation used
with a 96-well SPE microplate (Fig. 24.6). One example of such a system is the
Quadra96 (Tomtec) or the newer Quadra 3. The Quadra workstation automates
all of the liquid handling steps for SPE in a semi-automated mode. Its 96
separate precision pipettes operate simultaneously and deliver from 2-450 gl
volumes. Its deck is a six-position automated shuttle which accommodates a
vacuum manifold, samples in microplate format, a clean set of pipette tips, and
three solvent troughs. The instrument is attended by a user who manually
controls vacuum conditions (on/off), swaps pipette tips for a new set, and swaps

854
 i%

'-i
Fig. 24.6. Example of a 96-tip liquid handling workstation configured for microplate solid-phase
extraction using a vacuum manifold, the Quadra96TM (Tomtec). Photo reprinted with permission
from Tomtec.

solvent troughs (when necessary) during plate processing. Time required for an
extraction varies depending on the solvent volumes used and number of steps,
but can be as fast as 10 min; in practice, as many as four plates (384 samples) can
conveniently be processed per hour. When considering the total time require-
ment, however, the time to reformat samples from tubes to plate before use on
the Quadra96 must be considered. Several applications demonstrate the
versatility and speed of this 96-tip liquid handling approach to SPE [23-25].
There are also 96-tip pipettors available that can speed up common pipetting
operations, such as diluting samples, adding internal standard solution, and
sample reconstitution in microplates, as required. Such instruments are exem-
plified by the Personal Pipettor M 96 (Apricot Designs), Hydra96 (Robbins Sci-
TM
entific), Multimek 96 (Beckman Coulter), GenMateTM (Tecan) and
T M
RapidPlate (Zymark). The volume limitations vary among tip selected and
instrument, e.g., Multimek uses either 50 or 200 g1 disposable tips; thus, for dis-
pensing 500 gl volumes, multiple aspirations and dispenses are required. Many
of these instruments have an interchangeable pipetting head system so that they
can utilize 384-well labware on their deck.
The common XYZ liquid handlers described previously also can perform
solid-phase extractions; although not 96-tip based, they commonly offer 4-tip or
8-tip processing capabilities. The first few semi-automated 96-well SPE applica-
tions published used the MultiPROBE workstation [26,27]. The software offered
total control of the vacuum for processing via a manifold, and useful features of
the innovator Packard model included liquid level sensing, variable span tips
TM
and detection of blocked SPE wells. The Packard WinPREP software also

855
 TM
Fig. 24.7. Example ofa gripper arm on a Biomek workstation used to assemble and disassemble
components and move items around the deck (Beckman Coulter). Photo reprinted with
permission from Beckman Coulter, Inc. Copyright 2001.

allows for automating method development procedures in microplates. Other

brands of liquid handlers in time demonstrated SPE applications, e.g.,
Beckman's Biomek, Gilson's SPE 215 and their recently acquired Cyberlab
series, Tecan's Genesis (28) and Zymark's Sciclone.
Many of these XYZ liquid handling systems were at first semi-automated, in
that a manual step was required to remove the waste tray and insert a clean
collection plate for the elution step. As model lines were expanded, the function-
ality and usefulness of an integrated gripper arm became clearly evident. Lab-
ware movement around the deck and into external devices is now enabled in
such workstations (Fig. 24.7). The Speedy (Zinsser Analytics) was introduced
with this functionality built-in, as was the Biomek (Beckman), Sciclone ALH
(Zymark) and Cyberlab (Gilson). Packard offers its gripper arm with an option
as part of a "gripper integration platform" for both the MultiPROBE II EX and
MultiPROBE II HT EX (where EX = Extended deck configuration). Tecan's
Genesis robotic manipulator arm is called RoMA and now has built in access to a
centrifuge in one model. Most all of these vendors have designed a special
vacuum manifold to be assembled and disassembled by the robotic gripper
system (after sample clean up and before elution) and they support most
varieties of extraction plates on the market [291.
Note that while versatile liquid handling workstations have been adapted for
use in sample preparation by the means discussed, another choice for automa-
tion is to custom configure a workstation to meet the needs for a specific method.
Mettler Toledo Bohdan is one example of a vendor that adds the desired

856
capabilities to a deck platform to meet the goals of the method, e.g., weighing,
pipetting, mixing, homogenizing/grinding, LLE, SPE, centrifuging, evaporation
and filtration. Cyberlab is another example of a vendor that has built a variety of
custom workstations for automating sample preparation. Capping and
uncapping is another function that is automatable; workstations or modules for
capping/uncapping are available, e.g., from Mettler Toledo Bohdan and Zymark.
The previously mentioned ASPEC XL4 and SPE 215 systems from Gilson,
which accommodate discrete SPE cartridges, also process most types of 96-well
extraction plates. The XL4 is a 4-tip system which requires special caps be placed
on top of the microplate wells prior to use; a capper unit is available to quickly
add these caps in the form of strips to an extraction plate. In contrast, the SPE
215 uses positive pressure via a silicone sealing mat and does not require caps; it
is also faster using its 8 tips. Both these systems are fully automated and can
elute directly into deep well microplates for subsequent injection by a microplate
compatible autosampler.
Full automation for SPE microplates using a custom configured Zymate
robotics system is exemplified by the work of Biddlecombe, Smith and Lloyd
[30-33]. The systems described are built around a Zymate XP robot designed to
handle microplates in conjunction with various modules, e.g., a liquid handler
(Tecan RSP 8051) for sample transfers, a reagent addition station (RAS)
equipped with dual 12-port solvent selection valves for all liquid additions, a
microplate vortex station, a 96-well evaporation station, a plate sealer, a temper-
ature controlled carousel for the storage of all microplates and extraction
devices, and a custom microplate centrifuge (Fig. 24.8). In addition to SPE,
lI

25 PORT
TOLV
4T
SELECT

DRY - t~~flp

ix I A T

`'Gv

VORTEX BALANCE

MAND

CENTRIFUGE .>/

SEALER fu

Fig. 24.8. Schematic diagram of a robotics system configured to perform protein precipitation,
LLE and SPE in microplates. Photo courtesy of Glaxo SmithKline.

857
systems such as this one can also perform protein precipitation and LLE in
microplate format [34,35]. Note that full automation for sample prep in micro-
plates using a robotics system is not limited to the Zymark XP series of robots.
CRS Robotics Corporation (Burlington, Ontario, Canada) has its own line of
robotic arms upon which full workstations can be custom configured.

Solid-phase extraction: on-line

Solid-phase extraction microplates, when used with multiple tip liquid handling
workstations, have dramatically increased the number of samples that can be
prepared off-line, independent of the detection system. Eluates collected in
96-well microplates are injected by autosamplers that accept this same micro-
plate format. An example of such an autosampler is the CTC PAL from Leap
Technologies (Carrboro,sNorth Carolina, USA). There is another approach to
automation of sample preparation in bioanalysis, one in which a sample is
prepared and the eluent is subsequently injected "on-line" with the chromato-
graphic detection system. The throughput of such a system often does not match
that possible by parallel processing, but it can be maximized with fast chromato-
graphic run times coupled with fast extraction times. The advantage of this
on-line approach is that the system is fully automated from start (the sample
delivered in its container to the instrument) to finish (LC/MS/MS analysis).
Examples of these systems that utilize solid-phase extraction include the SPE
Twin PAL (Leap Technologies) using microplates, the Prospekt' (Spark
Holland, Emmens, Netherlands) and the OSP-2 (Merck, Darmstadt, Germany)
using special cartridges filled with sorbent media. On-line SPE can also be
performed using a narrow diameter LC column (e.g., 1x50 mm; Oasis HLB®,
Waters Corporation, Milford, Mass., USA) with large particle size sorbent media
(30-50 um) placed before the analytical column.
The SPE Twin PAL, processes a 96-well SPE plate one well at a time; the
eluate from well #1 is injected as the extraction begins on well #2. One syringe
(typically 1 ml capacity) is used for SPE plate conditioning, loading and washing
and the second syringe (typically 100 il capacity) is used for injecting small
volumes of eluate. The system uses positive pressure for liquid processing
through the wells. The SPE Twin PAL is also capable of adding internal
standards and performing dilutions of the sample, if desired; samples are
presented to the system in a microplate format. The eluate is collected in a clean
microplate that is moved into place below the extraction plate processing area at
the designated time.
Solid-phase extraction can be coupled on-line with LC/MS/MS using a versa-
tile automated system such as the Prospekt-M (PRogrammable On-line Solid
Phase ExtraKtion [sic] Technique) from Spark Holland. The Prospekt is an
integrated sample clean-up and injection system. The total system includes
three modules: a solvent delivery unit, the Prospekt module containing the
cartridge transport and sealing mechanisms, and an autosampler (Fig. 24.9).
Individual disposable cartridges containing various sorbent media are used, as a

858
Fig. 24.9. On-line SPE can be accomplished with the Prospekt which includes a solvent delivery
unit, cartridge transport and sealing module, and an autosampler; (a) extraction mode (b)
elution mode. Photo reprinted with permission from Spark Holland.

cassette of 10 cartridges; 120 cartridge capacity. Samples are introduced by the

autosampler and loaded onto disposable cartridges (2x10 mm, containing
10-50 mg sorbent). Cartridge exchange and valve switching is performed by the
Prospekt module. A weak solvent elutes unretained salts and polar matrix
components. An optimized sequence of solvents, each with increasing solvent
strength, is used to wash out weakly retained components. The valve is switched
and mobile phase elutes the analytes of interest from the SPE cartridge onto an
analytical column for chromatographic separation followed by detection. The
unit operates in staggered serial mode; one sample is being analyzed while the
next sample is undergoing clean-up. Each sample is processed by a fresh extrac-
tion cartridge so that there is no carryover of analyte by extraction media [36].
The Prospekt system performs method development quite readily, using differ-
ent sorbent chemistries and solvent wash/elution schemes. It is possible to
improve throughput even further by using a dual Prospekt system for parallel
on-line SPE [37].

Sample preparation:on-line turbulent flow

Turbulent flow chromatography is an on-line sample preparation technique that
allows for the direct injection of sample without previous extraction (usually a
single filtration or centrifugation of plasma is employed). This approach utilizes
large particle size stationary phases (50 /Am) and extremely high flow rates
(4-8 ml/min). When the flow rate (and thus the linear velocity) is high enough, a
point is reached where the mobile phase flow ceases to be laminar and becomes
turbulent. At this point, improved mass transfer and flow equilibration
increases the analyte diffusion within the pores of the packing material. The
larger matrix components are thus separated from the smaller analyte

859
molecules and diverted to waste. Column switching with the introduction of a
strong organic solvent flow then elutes the analyte from the turbulent flow
column onto the mass spectrometer. The Turboflow T column can be back-
flushed and re-equilibrated for repeated use, up to a maximum number of
injections which varies based on matrix material and volumes injected. An entire
system for turbulent flow chromatography is configured by Cohesive
Technologies (Franklin, Mass., USA) to minimize carry-over and provide for
optimum integration of components with the mass spectrometer. Numerous
applications have been demonstrated using this technique [38-401.
Another on-line sample preparation approach is reported that uses a
1x50 mm LC column packed with 30-50 [km stationary phases. The large
particle size allows a higher flow rate than normal to be used (e.g., 4 ml/min),
leading to a short analysis cycle time. These conditions do not represent a
condition of true turbulent flow, thus the terms "ultra-high flow-rate liquid
chromatography" and "direct on-line bioanalysis" have been used to describe
this technique [41]. A splitter is placed after the extraction column, permitting
about 0.2 ml/min to enter the mass spectrometer. Although only minimal
separation resolution can be achieved, it affords a rapid extraction to keep pace
with fast LC/MS/MS run times. While this example illustrates a generic
procedure, it has been adopted to provide for rapid method development for
SPE. An optimized SPE method can be achieved within one working day by
manipulating pH and the percentage of organic in the mobile phase to affect the
retention of analyte and matrix interferences 42]. This method can then be
transferred to a SPE microplate by matching sorbent chemistries.

Sample preparation:on-line RAM columns

Another technique allowing direct injection of plasma or serum on-line with the
chromatographic system, using normal laminar flow liquid chromatography,
involves the use of analytical columns containing "restricted access media"
(RAM). Commercially available RAM columns, all silica-based, include internal
surface reversed phase (ISRP), semi-permeable surface (SPS) and hydrophobic
shielded phase (Hisep). The silica surfaces of the stationary phases are modified
with a hydrophobic bonded phase, which in turn is coated with a hydrophilic
polymer. This outer hydrophilic polymer layer, due to its pore size, excludes pro-
teins and other large molecules and they are not retained. Drugs and other small
molecules enter the pores of the hydrophobic phase to partition and retain. The
stationary phases (and external hydrophilic phases) for ISRP, SPS and Hisep
RAM columns are commonly tripeptide (diol-glycine), octadecylsilane (polyeth-
ylene glycol), and disubstituted benzene (polyethylene glycol), respectively.
Essentially, a combination of size exclusion chromatography and partition chro-
matography is observed. An application comparing these three RAM columns for
non-steroidal anti-inflammatory drugs has been reported by Haque [43]. Some
disadvantages for high throughput considerations for RAM column usage is that
retention times can be long (>10 min), washing of the column is required

860
between injections, and required mobile phases are not always compatible with
some ionization techniques used in LC/MS/MS. The overall approach is auto-
mated, however, with low capital outlay for automation equipment since it uses
common HPLC hardware configurations, and is another choice in bioanalysis.

Sample preparation:on-line immunoaffinity extraction

Immunoaffinity extraction using antibody-antigen interactions can provide a
very high specificity for the molecules of interest in an analysis. Antibodies are
immobilized onto solid supports such as silica to provide highly selective extrac-
tions. In some cases, immunoaffinity extraction may be preferable to traditional
solid-phase extraction when SPE carries over interfering matrix components
and more specificity cannot be achieved, given time limitations or choices of
sorbent chemistry. Automated procedures with column switching valves have
been demonstrated using immunoaffinity extraction on-line prior to liquid
chromatography separation and tandem mass spectrometric detection [44,45].

Sample preparation:on-line ASTED

The ASTED M (Automated Sequential Trace Enrichment of Dialysates) extracts
biological fluids by dialysis followed by trace enrichment of dialysate on a short
column to concentrate the extract before analysis. The sample is diluted and
passed through the donor channel of the dialysis unit. Buffer is passed through a
recipient channel on the other side of the dialysis membrane. Analyte diffuses
across the dialysis membrane while plasma proteins are retained. The recipient
fluid is passed through a trace enrichment cartridge which concentrates the
analyte. The cartridge is then switched in-line with the mobile phase and analyte
elutes and reaches the analytical column for separation and analysis. This
approach is not a universal one, in that highly protein-bound drugs will not
dialyze to a great extent. Representative applications using the ASTED are dem-
onstrated by Turnell [46] and Lloyd [47].

24.3.3 Automation for clinical chemistry

The focus of laboratory automation for clinical chemistry, found in hospital

laboratories and commercial laboratories around the world, has undergone a
change in recent years. The movement now is toward greater productivity and
efficiency with the aim of controlling costs in the face of difficult market
competition. Laboratory automation technology consists of integrated hardware
and software designed to perform complete specimen processing and analysis.
This hardware is installed in the form of complete automation systems "total
laboratory automation" or as discreet hardware devices that perform specific
tasks "modular automation". Automation vendors have focused on the less
extensive hardware configurations called modular automation which is more
affordable, can be integrated into larger systems when needed, and appeals to
about 90% of the laboratories [48].

861
Fig. 24.10. A pre-analytical automated sample preparation system for a clinical laboratory, the
Power Processor M (Beckman Coulter). Photo reprinted with permission from Beckman Coulter,
Inc. Copyright 2001.

An aspect of automation unique to clinical chemistry is the integration of

pre- and post-analytical automation with the analytical test itself. Pre-analytical
automation aspects include the test request, the patient accurately identified,
the appropriate sample collected and stored in the proper tube (e.g., with or
without anticoagulant), complete sample labeling using barcodes, sample ins-
pection for errors, transport of sample to the laboratory and specimen
verification upon receipt, aliquoting and re-labeling. Since 60% of a laboratory's
labor costs can be attributed to specimen processing, automation of these
pre-analytical tasks is an important goal. The Power Processor T" (Beckman
Coulter) is an example of a pre-analytical automated sample preparation system
for a clinical laboratory (Fig. 24.10). Analytical automation may thus include any
or several of the following tasks:
1. Capping/uncapping.
2. Addition of reagents or buffers to sample, followed by mixing.
3. Centrifugation to isolate plasma or serum (directed by barcode labeling).
4. Sampling direct from tube or separation of plasma or serum.
5. Sealing specimen for transport and/or integrity.
6. Sorting samples by test(s) to be performed.
7. Delivering sample to appropriate workstation.
8. Sample preparation (yes/no).
9. If yes, use alternate designated pathway (extraction) to isolate or prepare
sample for the specific test; if no, proceed to direct sampling and measure-
ment with liquid level sensors and clot detection devices.
Post-analytical automation may include sorting and mapping samples to appro-
priate storage racks, review of out-of-range results, reporting test results, billing
for cost of the test performed, disposal of used reagents and sample wastes, and
archiving of sample tubes for further disposition.

862
 Larger clinical laboratories have Laboratory Information Systems (LIS) that
are interfaced with separate information systems for billing, and to patient
information systems for sample test requests and display of results. In the
laboratory, the testing requested for a specific sample is accessed by the worksta-
tion through "Host Query" functions and the test request processed in real time
by the analyzer. Once the bar-coded sample identification is read, the instru-
ment receives information on the tests to be performed directly from the LIS.
The instrument's "review by exception" software evaluates results and the
acceptable results are forwarded directly to the LIS. After a further review
process, these results can be released to the physician for review via the patient
information system.
The clinical laboratory uses test tubes as preferred sample containers
(not microplates as in pharmaceutical analysis). Safety concerns from
potentially hazardous or infected blood from patients can be satisfied using
"through the stopper" aspirations for many sample types. Tests performed are
most often immunoassay-based and detection endpoints include fluorescence,
colorimetric, kinetic rate, and/or chemiluminescence (not LC/MS/MS as in
pharmaceutical). This format approach permits the use of dedicated benchtop
workstations and analyzers that perform clinical chemistry tests automatically,
offering walk-away automation solutions (via large on-board capacity for
reagents and supplies) and integration with LIS. Examples of benchtop work-
stations offering immunoassay detection include the Access®2 (Beckman
Coulter; Fig. 24.11), the ADx System (Abbott Diagnostics, Chicago, Ill., USA)
and the ACS 180 (Bayer Diagnostics, Tarrytown, New York, USA).
A large, floor model modular automation approach is the AccelNet' labora-
tory automation network from Beckman Coulter. The AccelNet seamlessly com-
bines two SYNCHRON CX®9 ALX clinical chemistry systems, a specimen
manipulator, an automated sample processor, a centrifuge, and a data manager

Fig. 24.11. A benchtop automated immunoassay workstation for clinical analyses, the Access'
(Beckman Coulter). Photo reprinted with permission from Beckman Coulter, Inc. Copyright
2001.

863
to create a versatile workstation. The SYNCHRON series of clinical chemistry
analyzers comprise immunoassay systems in which up to 150 tests are available.
These tests include critical care and general chemistries, urine chemistries, eso-
teric chemistries, as well as tests for drugs of abuse, proteins, serologies, and
therapeutic drug monitoring. In terms of throughput, the CX500 Pro offers 53
on-board assays and throughput up to 700 tests per hour, the CX1000 Pro offers
57 assays at 1000 tests per hour and LX2000 Pro offers 65 assays at 1540 tests
per hour. Two additional examples of a scalable, modular automation system are
the Advia® LabCell Bayer Diagnostics [49] and the Lab-Frame® SELECT
(Ortho Clinical Diagnostics, Raritan, New Jersey, USA). Laboratories can adapt
an upgrade path with this equipment through the addition of conveyer belt mod-
ules that allow the addition of other work cells (e.g., chemistry, hematology).
Therapeutic drug monitoring (TDM) laboratories perform bioanalysis to
relate the plasma concentrations of a drug to a patient's therapeutic response
and/or adjust drug dosage to minimize adverse effects or toxicity. Typical
analytes for which TDM is performed using immunoassay detection include
digoxin, acetaminophen, phenobarbital, gentamicin, phenytoin, theophylline
and valproic acid. Lower volume, highly specific drug testing requires special
handling and sample preparation techniques. Immunoassay detection cannot be
used in cases where there is metabolite cross-reactivity (e.g., cyclosporine)
and/or when multiple analytes must be resolved and measured simultaneously
(e.g., antidepressant drugs), thus the more specific procedure LC/UV is used.
Sample preparation for LC/UV analysis most commonly involves solid-phase
extraction using individual columns or cartridges, and to a lesser extent involves
on-line SPE and LLE. Some common choices for automating the use of solid-
phase extraction cartridges in the clinical laboratory are the ASPEC XL/XL4 and
SPE Model 215 (Gilson) and the RapidTrace (Zymark), as discussed previously
for pharmaceutical analysis. The Prospekt (Spark Holland) has also been
demonstrated to perform reliably for automated on-line SPE in the clinical
laboratory setting. Note that although LC/UV is commonly used for detection,
there are growing instances where LC with mass spectrometry detection (LC/MS
and LC/MS/MS) is warranted (e.g., tacrolimus and sirolimus, two anti-rejection
drugs used in transplant cases; some of the newer anti-AIDS drugs; for newborn
screening in detecting inborn errors of metabolism).
Although dedicated workstations and modular automation for clinical labo-
ratory testing are widely used and continue to proliferate, there is the occasional
need for a more flexible and affordable automation solution, particularly for
molecular diagnostic applications performed in clinical research laboratories. An
example of this approach is the automation of the polymerase chain reaction
(PCR) amplification response to detect specific RNA and DNA sequences. The
PCR test is best performed by highly skilled technologists because of the
complexity of the assay and the potential for laboratory contamination. Thus,
automating this labor-intensive procedure is a desired goal and applications
have been demonstrated using an XYZ liquid handling workstation, e.g., the

864
MultiPROBE II (Packard Bioscience Company) and the Biomek 2000 (Beckman
Coulter). Integral components for this PCR assay are mounted on the worksta-
tion deck, namely a DNA engine with a remote alpha dock system and an
automated thermal cycling device. PCR protocols require the mixture of
multiple reagents in specific ratios for the amplification process. The exact
volumes of the reagents and buffers vary based on the total number of amplifica-
tions to be run and the desired final sample concentration and volume. An XYZ
liquid handling workstation has been found to provide a cost efficient alternative
to molecular diagnostic assays while demonstrating minimal inter-sample
contamination [50].
Automation in the clinical laboratory continues to advance in sophistication
and the full treatment of the subject is outside the scope of this text. Cost-
effective modular systems are common, and evolve into integrated laboratory
automation systems. Automation of the sample preparation step is but one small
component of the focus of hospital and clinical laboratories. Full robotics appli-
cations using robotic arms on tracks and the interfacing of multiple analyzer
components and processing systems have been demonstrated for high through-
put to accomplish "total laboratory automation", practised by some of the
largest laboratories in the world. There exist already even mobile robots that
travel across hospital floors and ride in elevators to pick up and deliver samples,
part of the pre-analytical automation component. Many disciplines of engineer-
ing and robotics are thus involved in this rapidly advancing approach to
automate all laboratory processes in the clinical laboratory and hospital setting,
and the reader is referred to other texts for their further description [51,52].

24.4 FUTURE TRENDS IN AUTOMATED PROCESSES

A variety of automated solutions have been demonstrated for sample

preparation. The rationale for this diversity is driven by many factors that vary
according to the application and laboratory setting, as previously discussed in
this chapter. As interrelated process technologies evolve, in search of increased
productivity, the bottleneck continues to shift from one part of the process to
another. For example, just as the emergence of widespread mass spectrometry
(MS) detection made sample preparation the rate-limiting step in many
laboratories, a subsequent proliferation of microplate format automation tools
then shifted the bottleneck back onto the analytical system. Meanwhile, as mass
spectrometers continue to achieve lower limits of detection and acquire more
data, aspects of the process that have been quite satisfactory for some time, such
as autosamplers and computerized data processing, became challenged as well.
Autosamplers have recently been adapted to handle microplates, input more
microplates via stackers or robotic arms, and now must further decrease the
cycle time between injections while at the same time reduce carryover.
Currently, we see new developments with the analytical systems for parallel
introduction of samples that may match or even outpace automated sample prep

865
cycle times again. Examples include parallel LC systems switching onto a mass
spectrometry source [53] or the combination of a reduced separation time by
implementing a monolithic column coupled with a parallel MS (MUX®) inter-
face [54]. Meanwhile, direct injection techniques (discussed previously) per-
formed in parallel continue to improve and challenge the need for sample prep at
all in certain cases [55].
Miniaturization is a continuing theme as detection limits improve, through-
put increases, and therefore sample sizes decrease. The density of microplate
formats continues to increase from 96-well to now 384-well and applications
using 384-well solid-phase extraction have been reported [56,57]. Also, liquid
handler working volumes continue to decrease and approach the point in many
instances where automation is no longer optional, but necessary to process sam-
ples. For example, the development and production of DNA/RNA biochip arrays
for diagnostic applications [58] presents unique liquid handling requirements
for nanoliter volumes. Tecan has developed a novel nano-pipetting system as an
option for the Genesis workstation to dispense volumes as small as 0.5 nl with a
spatial resolution of 100 Am.
In the meantime, many product lines from automation vendors have
diversified, through either expansion or acquisition, to the point of overlap and
broader competition. Companies that once concentrated on fully automated,
robotic systems now offer workstations and liquid handlers, while those that
began producing liquid handlers now have added plate stacking capability,
robotic arms, and a variety of other peripheral devices (e.g., carousels, incuba-
tors, centrifuges). A critical component to the integration and functional success
of linking peripherals and modular devices is software, which continues to
improve in utilizing different standards from multiple vendors.

REFERENCES

1 G.A. Smith and T.L. Lloyd, LC/GC Supplement (1988) S22.

2 T.L. Lloyd and J.R. Lang, Adv. Lab. Autom. Rob., 6 (1989) 73.
3 J. Wieling, J. Hempenius, R.J.J.M. Steenvoorden and J.H.G. Jonkman, in: Proc-
eedings of the International Symposium on Laboratory Automation and Robotics,
1998.
4 J. Berman, K. Halm, K. Adkison and J. Shaffer, J. Med. Chem., 40 (1997) 827.
5 C. Hop, Z. Wang, Q. Chen and G. Kwei, J. Pharm. Sci., 87 (1998) 901.
6 J. Henion, S. Prosser, T. Corso and G. Schultz, Am. Pharm. Rev., 3(4) (2000) 19.
7 A.P. Watt, D. Morrison, K. Locker and D. Evans, Anal. Chem., 72 (2000) 979.
8 R.A. Biddlecombe and S. Pleasance, J. Chromatogr.B, 734 (1999) 257.
9 M. Jemal, D. Teitz, Z. Ouyang and S. Khan, J. Chromatogr.B, 732 (1999) 501.
10 N. Zhang, K.L. Hoffman, W. Li and D.T. Rossi, J. Pharm. Biomed. Anal., 22 (2000)
131.
11 S. Steinborner and J. Henion, Anal. Chem., 71 (1999) 2340.
12 S. Peng, Todd M. Branch and S.L. King, Anal. Chem., 73 (2001) 708.
13 John R. Alianti, Christine M. Grosse and Glenn A. Smith, in: Proceedings of the
International Symposium on Laboratory Automation and Robotics, 1998.
14 B. Streel, Ph. Hubert and A. Ceccato, J. Chromatogr. B, 742 (2000) 391.

866
15 N. Helen Huang, John R. Kagel and David T. Rossi, J. Pharm. Biomed. Anal., 19
((1999) 613.
16 F.X. Diamond, W.E. Vickery and J. de Kanel, J. Anal. Toxicol., 20 (1996) 587.
17 X. Ren and A. Witkowski, in: Proceedings of the International Symposium on
Laboratory Automation and Robotics, 1996.
18 A. Bakkali, L.A. Berrueta, B. Gallo and F. Vicente, J. Chromatogr.B, 729 (1999) 139.
19 D. Parker, D.T. Rossi and D.S. Wright, Anal. Chem., 68 (1996) 2437.
20 J. Bates, UKLaboratory, February (1998) 12.
21 D.A. Wells, LC/GC, 17(7) (1999) 600.
22 C.Z. Matthews, E.J. Woolf, L. Lin, W. Fang, J. Hsieh, S. Ha, R. Simpson and B.K.
Matuszewski, J. Chromatogr.B, 751 (2001) 237.
23 J. Janiszewski, R.P. Schneider, K. Hoffmaster, M. Swyden, D. Wells and H. Fouda,
Rapid Comm. Mass Spectrom., 11 (1997) 1033.
24 G. Rule and J. Henion, J. Am. Soc. Mass Spectrom., 10 (1999) 1322.
25 S.X. Peng, S.L. King, D.M. Bornes, D.J. Foltz, T.R. Baker and M.G. Natchus, Anal.
Chem., 72 (2000) 1913.
26 J.P. Allanson, R. A. Biddlecombe, A.E. Jones and S. Pleasance, Rapid Comm. Mass
Spectrom., 10 (1996) 811.
27 H. Simpson, A. Berthemy, D. Buhrman, R. Burton, J. Newton, M. Kealy, D. Wells and
D. Wu, Rapid Comm. Mass Spectrom., 12 (1998) 75.
28 D. Schutze, B. Boss and J. Schmid, J. Chromatogr.B, 748 (2000) 55.
29 P. Ahrweiler, T. Sitko, S. Glass, K. Knotts and K. Ruterbories, in: Proceedings of Lab
Automation Conference, 2001.
30 S.L. Callejas, R.A. Biddlecombe, A.E. Jones, K.B. Joyce, A.I. Pereira and S.
Pleasance, J. Chromatogr.B, 718 (1998) 243.
31 R.A. Biddlecombe and S. Pleasance, in: Proceedings of the International Symposium
on Laboratory Automation and Robotics, 1997.
32 T.L. Lloyd, E. Lynch, J. Short and A. Wernicki, in: Proceedings of the International
Symposium on Laboratory Automation and Robotics, 2001.
33 G. Smith, J. Bruner and J. Alianti, in: Proceedings of the International Symposium
on Laboratory Automation and Robotics, 1997.
34 R.A. Biddlecombe and S. Pleasance, in: Proceedings of the International Symposium
on Laboratory Automation and Robotics, 1998.
35 R.A. Biddlecombe and S. Pleasance, in: Proceedings of the International Symposium
on Laboratory Automation and Robotics, 1999.
36 M.W.J. van Hout, C.M. Hofland, H.A.G. Niederlander and G.J. de Jong, Rapid
Commun. Mass Spectrom., 14 (2000) 2103.
37 A. Schellen, B. Ooms, M. van Gils, O. Halmingh, E. van der Vlis, D. van de Lagemaat
and E. Verheij, Rapid Commun. Mass Spectrom., 14 (2000) 230.
38 C.J. Oberhauser et al., LC-GC, 18(7) (2000) 716.
39 D. Zimmer, V. Pickard, W. Czembor and C. Muller, J. Chromatogr.A, 854 (1999) 23.
40 N. Brignol, R. Bakhtiar, L. Dou, T. Majumdar and F.L.S. Tse, Rapid Commun. Mass
Spectrom., 14 (2000) 141.
41 J. Ayrton, G.J. Dear, W.J. Leavens, D.N. Mallett and R.S. Plumb, J. Chromatogr.A,
828 (1998) 199.
42 J. Ding and U. Neue, Rapid Commun. Mass Spectrom., 13 (1999) 2151.
43 A. Haque and J.T. Stewart, Biomed. Chromatogr., 13 (1999) 51.
44 M.L. Nedved, S. Habibi-Goudarzi, B. Ganem and J.D. Henion, Anal. Chem., 68
(1996) 4228.
45 C.S. Creaser, S.J. Feely, E. Houghton and M. Seymour, J. Chromatogr.A, 794 (1998)
37.
46 D.C. Turnell and J.D.H. Cooper, J. Chromatogr., 395 (1987) 613.

867
47 P.-H. Hsyu and T.L. Lloyd, J. Chromatogr.B, 655 (1994) 253.
48 R.A. Felder, Clin. Chim. Acta, 278 (1998) 257.
49 M. Campanelli, J. Autom. Lab. Robotics, 3(3) (1998) 46.
50 T.E. Mifflin, C.A. Estey and R.A. Felder, Clin. Chim. Acta, 290 (2000) 199.
51 R.A. Felder, J. Clin. Lig. Assay, 22(1) (1999) 13.
52 S. Bauer and C. Teplitz, Med. Lab Observ., 7(9) (1995) 44.
53 L. Yang, T.D. Mann, D. Little, N. Wu, RP. Clement and P.J. Rudewicz, Anal. Chem.,
73 (2001) 1740.
54 Y. Deng, J.-T. Wu, T.L. Lloyd, C.L. Chi, T.V. Olah and S.E. Unger, Rapid Commun.
Mass Spectrom., 16 (2002) 1116.
55 C. Cameron, R.P. Grant, S. Mackenzie and M. Young, in: American Society for Mass
Spectrometry Annual Meeting, Chicago, 2001.
56 G. Rule, M. Chapple and J. Henion, Anal. Chem., 73 (2001) 439.
57 R.A. Biddlecombe, C. Benevides and S. Pleasance, Rapid Commun. Mass Spectrom.,
15 (2001) 33.
58 M.R. Knapp et al., Am. Lab., November (1998) 22.

868
Chapter25

Sampling and sample preparation for

food analysis
Meredith S.S. Curren and Jerry W. King

25.1 FOOD SAMPLING

25.1.1 Considerations

The term "food" refers to the broad range of edible materials that comprise the
essential body nutrients required for life and growth, such as proteins, carbohy-
drates, fats, vitamins, or minerals. Foodstuffs are described variously as "liquid"
or "solid", and "wet" or "dry", depending on the amounts of water and fat they
contain. Samples of plant origin are classified for analytical purposes as having a
high or medium water content and a lower content of saccharides (from 5% to
15%), very low water content (dry), or a high content of oils [1]. Similarly, food
samples can be divided into four main groups based on water and fat content [2].
Food samples of biological origin (liquid or solid) have been divided generally
into the five categories described in Table 25.1. This coarse division is important
when considering the choice of isolation technique, extraction solvent, and
sample clean-up method during an analytical procedure [3].
Moisture content is an important consideration during sampling procedures,
in part because it affects the extent of sample heterogeneity. Virtually all foods
are heterogeneous, and the analyst should be familiar with their variability in
composition and structure. In general, fresh foods of plant origin are more vari-
able in composition than fresh foods of animal origin. The analyst should be also
aware of the postmortem or postharvest physiological changes that can occur
after a fresh food is sampled and which can affect sample heterogeneity. A com-
bination of cold storage and chemical preservation may be required to maintain
sample integrity in the event of prolonged storage.
Although the chemical and physical properties of foods are inherently
variable, even between samples that originate from the same breed or strain, the
variability in composition of a single food sample can be minimized with proper
sampling and sample pretreatment techniques. Two approaches can be used for
sampling a food mass that is larger than the amount required for analysis in the

ComprehensiveAnalytical ChemistryXXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 869
TABLE 25.1
General classification of food samples according to their content (with permission from Ref. [3])
Sample Character Typical analytes
Milk Aqueous, proteins, lipids Veterinary drugs, toxic elements,
pesticides, industrial contaminants
Eggs High lipids and albumin content Veterinary drugs, industrial
contaminants, pesticides
Other samples of Various fat, proteins, or water Drugs, industrial contaminants,
animal origin (e.g. pesticides
muscle, liver, fat)
Plant material (e.g. Various water, plant pigments, Pesticides, toxic elements,
fruits, vegetables, lipids, proteins, essential oils or industrial contaminants
seeds) waxes
Food (e.g. meat, fish, Various fat, oils, lipids, proteins, Pesticides, industrial
milk, cereals, wine, sugar, starch, water, or pigments contaminants, synthetic colorants,
juices, plant oils, additives, synthetic sweeteners,
sugar) antioxidants

laboratory. Many minute increments of a solid material can be collected and

blended to represent the entire foodstuff, or a quantity of material that is large
enough to be compositionally representative of the whole can be collected and
then reduced to a fine mixture before being subsampled [4]. The first approach is
usually avoided, since it is difficult to obtain a statistically representative sample
and the sampling time can also be very long. The latter approach is more
practical, accurate, and reproducible.
Since virtually no food material can be analyzed in its entirety, careful
sampling techniques are required to obtain representative, laboratory-sized
primary samples, in addition to subsequent subsamples, or secondary samples
[5]. The amount of subsample required for an analytical procedure usually
varies from a fraction of a gram to several grams. The sampling techniques
discussed in the sections that follow are used to produce small, discrete primary
and secondary samples that are representative of the entire food material, with
minimal error.
The required sample size is defined in part by the nature of the target
compound, that is, to what extent the analyte is retained in the matrix. Xeno-
biotics are generally present at trace levels, i.e., in tgxg or ng.g - l concentrations,
or even lower. A sufficiently large amount of sample must be collected and
analyzed in order to be able to measure minute quantities of the compound of
interest and to satisfy the method's limit of detection. Conversely, relatively
small samples may be collected for the macro analysis of gross food components,
i.e., to measure crude fat, crude protein, crude fiber, or ash. Although proximate
analysis of these food components is sometimes sufficient, more exact analyses
are usually required.

870
 The sample size is also dependent on the relationship that exists between the
mass required to adequately represent a sample and the characteristics of that
sample [6]. If a foodstuff consists of some mixture of different-sized particles,
enough sample mass needs to be collected in order to adequately represent all of
the particles. Because large particles are more difficult to represent than smaller
ones, a mass that is large enough to represent the larger particles will also be
representative of the smaller ones. The segregation of finer, denser particles to
the bottom of the sample container must be recognized during the sampling
process to ensure that all particles are represented and to avoid large sampling
errors. The theory of sampling along with solutions for correct sampling are
well-described in other texts [7-9].

25.1.2 Techniques

Food lots are sampled in either a manual or continuous manner in order to

obtain a representative specimen. Containers holding loose foodstuffs can be
sampled manually with devices that trap the material in a compartment such as
a probe or tube. Slots or openings placed at intervals in the tube allow for
simultaneous sampling at different depths of the product. When employing this
technique, however, the analyst must consider the segregation effect and ensure
that all particle sizes are accessible. The foodstuff may ultimately need to be
removed from the sample container and poured onto a flat surface. The amount
of material may then be reduced with a coning-and-quartering method [10], and
a subsample collected in multiple random increments. No particle size should be
excluded during the sampling process since food components or contaminants
that collect in certain-sized particles might be omitted from the final analysis,
thereby resulting in an increase in sampling error.
Large mixtures may also be reduced with a riffle cutter, which is a box-like
device that has equally spaced dividers to divide the sample stream. The sample
may be further cut or quartered by passing it through successive riffles. Other
proportional dividers are available for reducing a sample, such as the straight-
line sampler and the spinning riffle sample divider [10].
Uniformly solid or liquid products are perhaps the most straightforward to
sample. Drill-type devices are used to obtain a core from solid products such as
cheese or frozen foods. Liquid samples are thoroughly mixed before a subsample
is removed with a syringe-type sampler or by submerging a container under the
liquid's surface (a so-called "grab" sample) [11]. For obvious reasons, many
complex foods such as vegetables, fruit, or animal tissues may require blending
prior to being sampled. These blending methods are discussed in the section that
follows.
Throughout the sample preparation procedure, it is essential for the analyst
to recognize the necessity of utilizing methods that satisfy statistical sampling
and analysis requirements. The inherent variability in the composition of raw
materials, basic ingredients, and processed foods requires the use of statistical

871
methods for obtaining representative and replicate samples, and for estimating
the error involved in sampling. Measures of the precision of the mean results are
also required, in addition to statistical analysis and interpretation of the data
obtained [12]. The reader is referred to standard text and reference sources in
the field for these purposes [7-9,12,13].

25.2 FOOD PRETREATMENT

25.2.1 Removal of extraneous matter

Before sample blending is done, it is often necessary to wash, remove, or drain

irrelevant extraneous matter. Soil or sand that adheres to fresh fruit or vegeta-
bles can be removed by washing or wiping the surface of the produce; however,
excessive washing should be avoided to prevent the leaching of soluble solids.
Depending on the objective of the analysis, fresh produce may be separated into
the core and the outer and inner tissues. Shells are usually separated from nut
kernels and pits from stone fruits. Large fish are cleaned, scaled, and eviscer-
ated, while small fish can be blended whole. Shellfish are shucked, eggs are bro-
ken to isolate the liquid interior, and meat is removed as completely as possible
from bone. Suspended matter or sediment present in liquids such as beer, wine,
juice, or cooking oil is removed by filtration or separated by centrifugation.
Canned fruit and vegetable products may be drained through screens if it is not
necessary to analyze the composite sample [14].

25.2.2 Sample reduction

Once a food sample has been collected using the sampling techniques discussed
in Section 25.1.2, a suitable method is required to make the material less
heterogeneous. Various approaches may be utilized for reducing the particle
weight and size in a primary sample, so that smaller subsamples can be taken for
a representative analysis of the whole [4]. Finely divided materials also dissolve
faster and are easier to extract because of their greater surface area.
Methods for reducing solid or semi-solid foods include mechanical grinding,
mixing, rolling, agitating, stirring, chopping, crushing, macerating, mincing,
pressing, pulverizing, or any other reasonable means of comminuting the sam-
ple. Sample reduction can also be achieved with a Wiley or ball mill, mortar and
pestle, mechanical high-speed beaters or blenders (for soft or wet foods), and
meat grinders. Liquid samples can be mixed using magnetic stirrers or sonic
oscillators. Figure 25.1 demonstrates the importance of selecting the appropri-
ate hardware for sample mixing, and of blending the sample for a sufficient
period of time [15].
There are several other factors to consider when reducing a food sample.
Food choppers, blenders, and mixers should be constructed of metal alloys that
resist corrosion or erosion, and that are inert enough to prevent contamination

872
 BUichi B-400 MixerA Mixer B Mixer C

U
0
47

n
W

U
e

Fig. 25.1. Effect of choice of mixing equipment and blending time on sample heterogeneity for
sunflower seeds. Reprinted with permission from Ref. [15].

of the product. Aeration of the product during the blending process should be
avoided since this can result in appreciable changes in oxidizable components. It
is also important to avoid heating the material during the grinding step since
this can accelerate chemical changes in the foodstuff. The surfaces of all mixing
equipment should be clean and dry, since changes in sample moisture content
can change the chemical and physical nature of the foodstuff. Care should also be
taken to prevent the release of volatile constituents during grinding, if this is of
concern [14].
The analyst should be aware of the enzymatic changes that can rapidly occur
in crushed plant and animal tissues. In animal tissue, rapid enzymatic changes
may result in appreciable changes of certain food components, particularly in
the case of carbohydrate and nitrogenous compounds [14]. It may also be
necessary to inactivate food enzymes, for example by denaturation in boiling
methanol-water or ethanol-water mixtures [16].
In conclusion, it is imperative for the analyst to be familiar with the food
matrix that is being analyzed. Since it is not feasible to discuss all possible cases
here, it is important that the analyst to consult the appropriate sources for
information before beginning a new sampling procedure.

25.2.3 Moisture

Recognition of the level of moisture in food samples is important for several

reasons. As previously discussed, moisture can contribute to the extent of
sample heterogeneity. A sample may also need to be dried prior to being blended

873
or stored, since the material may lose moisture during blending, or deteriorate
during storage. Determining the moisture content through sample drying may
be necessary in order to calculate the nutritive value of a food product or to
express analytical results on a uniform scale, for example in the determination of
the dry matter in flour [17]. Moisture is also important in terms of food quality,
since it affects food freshness, preservation, and resistance to deterioration.
Water is present in food samples in three forms [11]: as a solvent or
dispersing media; adsorbed on the internal or external surfaces, or as fine
capillaries by capillary condensation; and as water of hydration.
Solvent or free water is most easily removed. The rate at which moisture is
removed from foods is affected by drying temperature, particle size, vacuum,
crust formations on the surface, and surface area of the sample [11]. Bound
water is quite difficult to remove and normally requires a vacuum process.
Vacuum drying is preferred nonetheless, since this technique significantly
reduces the deterioration of samples during heating. For example, plant tissues,
which are often dried prior to the analysis step, might undergo extensive
enzymatic changes during the drying process, especially when they are exposed
to air. Utilizing vacuum drying also accelerates the drying time, which can take
up to 16 h under ideal conditions.
Other precautions need to be considered when drying foods at elevated
temperatures, since chemical reactions such as hydrolysis can occur and chemi-
cal reactions can be accelerated. Moisture determinations can be erroneous if
hydrolysis has occurred, since the water of hydrolysis has not been released from
the sample. On the other hand, very dry samples may absorb water from the air
before the moisture determination has been completed.
A general rule of thumb for sample drying is that it should be as rapid and at
as low a temperature as possible. Vacuum methods that can used to dry a sample
include vacuum ovens and lyophilization, or freeze-drying. Other methods that
can be employed are distillation, microwave drying, and the Fischer titration
method. The titration method is particularly applicable to low-moisture foods
that give erratic results when heated or under vacuum [11].
Finally, when drying a sample, the analyst should be aware that a certain
level of moisture might be required for prolonged food storage, since chemical
reactions such as oxidative deterioration can occur when moisture levels are too
low, for example in vegetables such as carrots and potatoes, which will develop
oxidized flavours or become rancid in two to three weeks at a 2 or 3% moisture
content. Oxidative deterioration of these foods is inhibited for several months
when they have a 8-10% moisture content [14].

25.2.4 Removal of co-extractives

An inherent difficulty in the extraction of food samples is the co-extraction of

matrix components that are also soluble in the extraction solvent. A common
example of this is the co-extraction of lipids during supercritical carbon dioxide

874
extraction (or any other type of extraction) of non-polar compounds from animal
and vegetable matrices [18-22]. The presence of matrix interferences in sample
extracts can result in a multitude of problems, including the generation of
emulsions, sample turbidity, contamination or plugging of equipment, and,
perhaps most importantly, the masking of the analytical signal for the target
analyte and the consequent increase in the method limit of detection.
Co-extractives are frequently removed during a post-extraction clean-up
step that requires passing the liquid extract through a clean-up column for sorp-
tion or filtration of the interferences. Commonly used clean-up materials include
Florisil, alumina, silica gel, in addition to gel permeation chromatography,
solid-phase extraction materials, etc. Solid-phase materials can also be used to
exclude co-extractives from the analyte concentration step, that is, the material
may only retain the target analyte and not the interferences. This step is also
referred to as analyte enrichment, since the analyte concentration is increased
over that of the matrix background signal, if indeed any occurs at all. The factors
that affect the choice of clean-up material are similar to those considered when
choosing a solid-phase for the extraction of liquid food samples. Overviews of
both types of applications will therefore be presented together in Section 25.3.
Of particular interest are analytical methods that incorporate an in situ or
on-line clean-up technique. Sample clean-up in this case can be achieved in situ,
for example, with a simple and elegant extraction technique called matrix
solid-phase dispersion (MSPD) [23,24]. The advantage to MSPD is that it
combines sample blending, clean-up, and extraction into one technique. During
an MSPD procedure, the sample matrix is mixed with an appropriate polymer
resin, such as the reverse-phase chromatographic sorbent, C,,. The solid or
semi-solid sample is prepared for extraction by grinding it in the presence of the
sorbent using a mortar and pestle, which facilitates disruption of the sample
matrix. Total disruption is achieved once the cell components are disrupted and
the sample is evenly dispersed over the polymer material [23]. The end result is
that the entire dispersed sample becomes a unique chromatographic phase from
which either the analyte or matrix components can be selectively eluted using an
organic solvent or solvent mixture with the appropriate eluent strength. The
solvent mixture is usually water-immiscible.
The MSPD technique was originally applied to the isolation of drug residues
from animal tissues [25]. It has since been successfully applied to the wide
variety of food matrices shown in Table 25.2, including dairy or medical
products, animal tissues, vegetables, fruits, and aquatic species. It should be
noted that adsorbent consumption can be high for samples with high lipid
content, and that an additional clean-up step may be required for an extract
obtained from a complex biological matrix.
A particularly interesting application uses a miniaturized and automated
MSPD extraction method for the isolation of pesticides from fruit samples [40].
This method was optimized for a variety of organophosphorous pesticides and a
pyrethroid from oranges, but satisfactory recoveries were also obtained from

875
TABLE 25.2
MSPD clean-up and extraction of food matrices
Sample Analyte MSPD material Ref.
Citrus fruit Various pesticides C8 26
Oranges, grape, onions, Carbamate pesticides Cl, C8, cyano, amine and 27
tomatoes phenyl solid phases
Milk Organochlorine and ClS 28
organophosphorus pesticides
Milk Veterinary drugs C18 29
Medical foods Vitamin K, C18 30
Meat, milk, cheese Tetracyclines ClS 31
Fish Surfactants C18 32-34
Fish Triazine pesticides C8s 35
Beef fat Chlorinated pesticides C18 36
Liver P-agonists C8 , C18 37
Liver Clenbuterol ClS 38
Chicken muscle Sulfonamides C8S 39

pears and grapes. The method requires only 25 mg of sample and 100 A of
organic solvent. Solid-phase C8 was determined to be the optimum dispersion
material for this application.
The technique of matrix-solid phase dispersion can be adjusted to retain
particular compounds by choosing an appropriate dispersion material in addi-
tion to using a specific eluent. Most applications have utilized the reverse-phase
material C,,, in part because the solid silica support facilitates sample disruption
while silanol groups on the silica surface may associate with polar components in
the sample matrix [23]. However, a recent application has demonstrated that
clean-up from kidney tissue can be achieved with a cross-linked acrylic polymer
[41]. In this case, the acrylic polymer XAD-7 HP was able to retain lipid
components, such as fatty acids, sterols, and triglycerides, in addition to protein
matter in the presence of an ethanol-modified water eluent at 100°C. Figure 25.2
demonstrates how the kidney sample clean-up was achieved with the XAD-7 HP
resin.
A slightly different in situ sample clean-up technique has been employed for
the selective extraction of polychlorinated biphenyls (PCBs) from lard, fish, fish
meal, and cod-liver oil [42,43]. In these cases, a fat retainer was placed in
sequence inside an extraction thimble, rather than being dispersed through the
sample. The packing of the extraction thimble shown in Fig. 25.3 demonstrates
how matrix interferences are initially co-extracted from the sample, but are then
trapped by the fat-retainer inside the thimble. Several fat retainers have been
investigated, including sulfuric acid, Florisil, and basic, neutral, and acidic alu-
mina, for static extractions performed at 100°C, followed by elution with n-hex-
ane. Figure 25.4 demonstrates that the magnitude of fat retention is similar for

876
 1 filter paper
flow SFE support
direction
Sand +Na2 SO4

Matrix +
Na 2SO 4 +sand

1 filter paper

Fat retainer

r
Silica gel
2 filter papers

Left: Fig. 25.2. Extract from kidney samples pretreated with or without the acrylic polymer
XAD-7 HP prior to pressurized liquid extraction with 30% ethanol in water at 100°C and 50 atm.
Samples: 0.5 g beef kidney + 2 g diatomaceous earth.
Right: Fig. 25.3. Packing of an extraction thimble with fat retainer. Reproduced with permission
from Ref. [42].

iB,
05"
I-

0.05 0.10 015 0.20

Fat / fat retainer ratio

Fig. 25.4. Amount of fat retained for five fat retainers using different fat/fat retainer ratios.
Reproduced with permission from Ref. [42].

877
each retainer, but is dependent on the fat/fat retainer ratio. Fat retainers have
been utilized in a similar mode when conducting supercritical fluid extraction
(SFE) with CO 2 [44-48].

25.3 LIQUID FOOD SAMPLES AND EXTRACTS

25.3.1 Choice of extraction methods

The analyses of liquid food samples have an advantage over those associated
with solid samples in that they usually require one less pretreatment step, due to
their liquid form. In some cases, very little sample preparation may be required
if the liquid is sufficiently free of matrix interferences. Straightforward tech-
niques that may used to prepare "clean" liquid samples prior to the analysis step
include sample dilution, evaporation, distillation, microdialysis, lyophilization,
or liquid-liquid extraction (LLE) [49]. Sample drying by lyophilization was
discussed in the previous section, and is particularly useful for the analysis of
nonvolatile organics. The technique of microdialysis is further discussed in
Section 25.3.4.
The technique of LLE is included in this list of "conventional" or straightfor-
ward methods since it is well-described in standard texts and references, and has
also been described in some recent reviews [50,51]. Further information on LLE
methods is also found in Chapter 11. The LLE technique is frequently utilized in
the analysis of toxicants, but can also be applied to food components, for example
in the extraction of low relative molecular mass compounds from food samples,
such as milk, soft drinks, wine, or beer. The extraction procedure generally
results in the separation of hydrophilic and lipophilic compounds, such as fat and
proteins, following a protein denaturation step with an acid or organic solvent,
or following solvent extraction under gentler conditions [16].
Other major techniques for the isolation or purification of liquid food
samples are solid-phase extraction (SPE), including immunoaffinity extraction
(IAE) and molecularly-imprinted polymers (MIPs); microextraction techniques,
including solid-phase microextraction (SPME) and spin bar sorptive extraction
(SBSE); and membrane extraction techniques, including dialysis. These meth-
ods are characterized by a reduced use of organic solvents, and the associated
toxic effects to the laboratory worker and the environment.

25.3.2 Solid-phase extraction

The clean-up and concentration of target analytes in liquid samples or solvent

extracts is frequently achieved through sorption onto a solid-phase extraction
material that is loaded in a separate cartridge or disk, or placed in-line down-
stream from the extraction vessel. Table 25.3 presents an overview of select SPE
applications for liquid food samples, in addition to examples of the SPE clean-up
of solvent extracts from solid foods. Most of the examples cited refer to the

878
TABLE 25.3
Solid-phase extraction and clean-up of liquid foods and solvent extracts
Sample Anlalyte Solid-phase Ref.
SPE clean-up of solvent extracts
Fruits and vegetables Pe sticides C,,, CN 54
Apples, pears Be nzoylurea insecticides Silica 55
Meat extract Heeterocyclic amines Cs8 56
Liver, kidney, muscle Pe nicillin antibiotics Ion-exchange 57
SPE of liquid food
Red wine Pii gments Silica, C, CN, alumina, NH2, 58
Florisil, carbon black
Wine Various pesticides C8s, PS-DVB, Oasis cartridge 59
Alcoholic beverages Sy nthetic colours NH2 60
Orange juice Ca .rotenoids C1
, 61
Soft drinks Caffeine Cl 8 62
Fruit juices Phlenolic acids Cl 8 63
Butter (liquid) FiE avour compounds C,,, Cs, NH, CN, PS-DVB 64
Immunoaffinity SPE and clean-lup
Orange juice s-t Lriazine pesticides Immunoaffinity 65
Fruit and vegetables PhLenylurea herbidices Immunoaffinity 66
Fruit and vegetables Tr iazine herbicides Immunoaffinity 67
Fish Mi icrocystins Immunoaffinity 68
Grains M)icotoxin Immunoaffinity 69
Herbs Ni trated polycyclic Immunoaffinity 70
ariomatic hydrocarbon
Peanut butter, paprika, Af latoxins Immunoaffinity 71
pistachios
MIP SPE and clean-up
Liver Cl1enbuterol MIP 72
Chewing Gum Ni cotine and analogs MIP 73
Liver Tr iazine pesticides MIP 74
PS-DVB = Polystyrenedivinyl-benzene.

analysis of xenobiotics or trace components. However, SPE is also amenable to

the analysis of lipid classes and related compounds, as described in a recent
review [52]. Automated SPE is easily achieved with a dedicated SPE worksta-
tion, for example in the determination of resveratrol derivatives in wine [53].
Normal- and reversed-phase chromatographic materials continue to find
widespread use in the food industry. However, the use of analyte-specific materi-
als, such immunoaffinity-based solid-phase extraction and molecularly imprint-
ed polymers, is becoming increasingly advantageous. In the case of IAE, the
appropriate antibodies are developed against the compound of interest. This
technique may also be utilized on-line, for example in the determination of

879
s-triazines in orange juice, where the cartridge containing the immobilized anti-
bodies is coupled on-line to a gas chromatograph via a reversed-phase cartridge
[65].
Table 25.3 also cites examples in which molecularly imprinted polymers
were used as solid-phase sorbents for the enrichment of analytes from liquid
foods and solvent extracts. MIPs are highly stable polymers that possess recogni-
tion sites within the polymer matrix that are specific for the three-dimensional
shape and functionalities of the analyte of interest [75]. For example, an MIP
material was utilized as part of a two-tier, on-line sample clean-up method
performed concurrently with sample extraction for the determination of clen-
buterol in liver. The liver samples were first blended with C,, in a MSPD
clean-up procedure, then a molecularly imprinted solid-phase extraction
cartridge was placed in-line after the MSPD cartridge to selectively trap the
analyte during elution with acetonitrile [72].
SPE methods may also use ion-exchange materials. For example, anion
exchange membranes have been utilized for the determination of glucosinolates
in canola and mustard seeds. The analytes in this case were isolated by
immersing the membranes in an aqueous suspension of the ground seeds. The
membrane was then removed from the suspension, washed, and submerged in
an appropriate solvent for elution inside of a shaken vial [76,77].

25.3.3 Microextraction techniques

Two equilibrium-based microextraction techniques serve as alternatives to

classical solid-phase extraction: solid-phase microextraction (SPME) and
stir-bar sorptive extraction (SBSE). The advantages of utilizing SPME have
been well-discussed in previous sections. Table 25.4 lists a few of the many liquid
food applications that have been developed utilizing SPME fibers, in addition to
the SPME sampling of solvent extracts from solid foods (headspace sampling of
solid foods will be discussed in Section 25.4). Each of the examples cited in Table
25.4 utilize a "classical" sampling method consistent with SPME, that is, either
by immersing the fiber directly in the sample, by sampling the headspace, or by
sampling the effluent from a gas stream (the latter two are classified together as
"headspace" in Table 25.4). Alternative SPME sampling methods have been
investigated, for example in the determination of catechins and caffeine in tea by
utilizing automated in-tube solid-phase microextraction [78].
Stir bar sorptive extraction is a similar equilibrium technique that requires
submersion of a stir bar (that is encapsulated in a glass jacket and coated with a
solid-phase) into the liquid sample. In this case, the solid-phase is usually a rela-
tively high amount (25-125 1 ) of polydimethylsiloxane (PDMS) polymer. The
stir bar is then thermally desorbed on-line in the heated injector of a gas
chromatograph. The advantage to utilizing SBSE for sampling liquid samples or
extracts that are amenable to the PDMS solid-phase technique is that a 500-fold
increase in enrichment, and therefore sensitivity, can be achieved compared

880
TABLE 25.4
SPME sampling of liquid foods and solvent extracts
Sample Analyte Fiber Ref.
Liquid food, headspace
Beer Alcohols and esters PA 79
Wine Flavours PDMS 80
Milk Fatty acids PA 81
Alcoholic beverages Flavours PA 82
Vanilla extracts Volatiles PA 83
Orange juice Flavours PDMS 84
Vegetable oils Volatiles DVB/Carboxene/PDMS 85
Wine Sulphides Carboxen/PDMS 86
Wine Diacetyl PDMS, CW-DVB 87

Liquid food and extracts, immersion

Fruit juices Organophosphates PDMS, CW-DVB, PDMS-DVB, PA 88
Honey Pesticides PA, PDMS 89
Beverages Caffeine Silica 90
Strawberries Pesticides PDMS-DVB 91
Cheese Mycotoxin CW-DVB 92
Kidney Triazine pesticide CW-DVB 41
PA = polyacrylate; PDMS = polydimethylsiloxane; CW-DVB = Carbowax divinylbenzene.

with a 100 tm PDMS SPME fiber [93]. However, such a SBSE technique does
not have the same selectivity as SPME.
Although the SBSE technique has only been recently developed, it has
already seen modest use in the food industry. Stir bar sorptive extraction has
been applied to the determination of dicarboximide fungicides in wine [94],
organochlorine pesticides and chlorobenzenes in fruit and vegetables [95,96],
benzoic acid in lemon-flavoured beverages [97], and flavour compounds in
strawberries [98].

25.3.4 Membrane techniques

Membrane extraction methodologies encompass both the non-porous tech-

niques of supported liquid membrane extraction (SLM), microporous membrane
liquid-liquid extraction (MMLLE), polymeric membrane extraction (PME), and
membrane-extraction with a sorbent interface (MESI), in addition to the porous
membrane technique of dialysis [99,100]. Variations of the latter are micro-
dialysis and electrodialysis. Unlike the non-porous membrane methodologies,
the porosity-based techniques are not characterized by analyte enrichment.
There is no discrimination between small-sized molecules that are similar in size

881
to the analyte, and only partial sample clean-up is achieved by membrane
separation of lower molecular weight species from higher molecular weight
matrix components. A dialysis clean-up step is therefore often combined with a
subsequent enrichment technique, for example on an automated trace
enrichment of dialysates system, also known as ASTED.
Dialysis techniques are strictly not extraction techniques, unlike the non-
porous membrane extraction methods. However, they will be discussed here
nonetheless, since they are highly effective for the clean-up of liquid foods and
solvent extracts. Several on-line microdialysis methodologies have been devel-
oped for this purpose. For example, on-line microdialysis clean-up has been
coupled with liquid chromatography and programmable fluorescence detection
for the analysis of chicken liver fortified with fluoroquinolone antibacterials
[101]. Fluoroquinolones in eggs [102] and chicken liver and muscle [103], in
addition to sarafloxacin residues in fortified and incurred eggs [104], can be
determined in a similar manner.
On-line microdialysis has also been utilized to improve the sensitivity of a
disposable lactate biosensor used in the flow-injection mode for the analysis of
L-lactate in milk and yoghurt [105]. The measurement of lactulose in milk did
not require pre-treatment when a microdialysis probe was used as the sampling
system [106]. Glucose determination in milk and juice samples can be achieved
with little or no sample pretreatment using a microsystem that integrates a
microdialysis probe with a glucose oxidase bioreactor [107].
During a nonporous membrane extraction technique, a liquid or solid (e.g.
polymeric) phase is placed between two other phases, which are usually liquid
but sometimes gaseous [100]. The sample to be processed may be viewed as part
of the donor phase, while an acceptor phase on the other side of the membrane
collects the analyte for transfer to the analytical instrument. In this fashion,
unparalleled sample clean-up and analyte enrichment can be achieved when
compared with classical liquid-liquid extractions.
Nonporous membrane techniques have tremendous potential for the food
industry, although they as yet have seen limited use in this field. For example,
Vitamin E has been determined in butter samples after dissolution of the butter
in a micellar medium. Following on-line saponification, the nutrient was
enriched across a silicone membrane and taken up in acetonitrile prior to liquid
chromatographic analysis using an electrochemical detector [108]. Continuous
extraction of Vitamin E isomers from vegetable oils has been achieved in a
similar manner [109]. Another membrane separation device has been coupled to
a liquid chromatograph for the enrichment of pesticide multiresidues from egg
extracts generated via Soxhlet extraction [1101.
The supported liquid membrane (SLM) extraction principle can also be
extended to solid or semi-solid samples by incorporating a donor channel unit
that permits close contact between the sample and the membrane. Using such a
configuration, it has been possible to extract and quantify vanillin in food
samples (e.g. chocolate) [111] and caffeine in coffee and tea [112].

882
25.4 EXTRACTION OF SOLID SAMPLES

25.4.1 Headspace solid-phase microextraction

The application of solid-phase microextraction to the analysis of solvent extracts

from solid food samples has already been discussed in Section 25.3.3. Several
examples of the direct immersion of a SPME fiber into food extracts were
provided in Table 25.4. This section provides a short review of a second approach
that may be utilized to sample volatile species from solid food samples, that is, by
sampling the headspace above the food with a SPME fiber. Sampling solid foods
in such a manner allows the solvent extraction step to be omitted from the
analytical procedure. Table 25.5 lists several examples of this approach for a
variety of foodstuffs.

25.4.2 Microwave-assisted extraction

The traditional method for the determination of compounds in many foodstuffs
is Soxhlet extraction, whereby the solid sample is placed in a porous thimble and
is continuously extracted in a glass apparatus with a sub-boiling solvent. The
thimble in this case also serves as a filtration medium. Soxhlet methods are
fairly simple, standard, and continue to have widespread use in the food
industry, for example in the determination of pesticide residues in eggs [110].
However, these methods can also be inefficient and slow, and they can consume
large quantities of organic solvents.
Microwave-assisted extraction (MAE) is one of several techniques that have
been developed in response to the increased demand for techniques that have a
shortened extraction time and reduced solvent consumption, as discussed in a
recent review [123]. One of the primary benefits of MAE is the ability to directly
TABLE 25.5
Headspace SPME sampling of solid and semi-solid foodstuffs
Sample Analyte Fiber Ref.
Mustard paste Flavour compounds PDMS-DVB 113
Cucumber Flavour compounds PDMS 114
Butter Reduced sulphur PA 115
compounds
Apple Flavour compounds PDMS 116
Onion Volatiles PDMS 117
Cheese Volatiles PDMS, PA 118
Smoked ham Nitrosoamines PA 119
Catfish Off-flavours PDMS 120
Tomato and strawberry Flavour compounds PDMS, PDMS-DVB, CW-DVB 121
Processed poultry Volatiles produced by PDMS 122
bacteria

883
heat the sample with the application of microwaves. This type of heating is fast
and temperature gradients are kept to a minimum. A drawback to the technique
is the requirement for an extraction solvent that is able to absorb microwaves. In
addition, a subsequent clean-up step is usually required once the microwave
vessel has cooled sufficiently for handling.
Microwave techniques have been applied to biological and food samples quite
extensively. The first use of the microwave domestic oven in the laboratory was
for the determination of trace metals in biological samples [124]. This was
followed by the extraction of crude fat and nutrients from food [125], and such
solutes as pyrimidine-glucoside from seeds and fava beans [126]. A patented
variation of MAE is the microwave-assisted process [127], or MAP, which was
first applied to the extraction of essential oils from plant products [128]. MAP
methods mainly concern biological applications ranging from analytical to
processing scale.
All the applications cited thus far have utilized closed-vessel systems.
Recently, closed-vessel MAE has been used for a number of marine tissue
applications. For example, organic contaminants such as polychlorinated bi-
phenyl (PCB) congeners, chlorinated pesticides, and polycyclic aromatic hydro-
carbons (PAHs) have been extracted from standard reference materials [129]. In
this case, it was determined that the moisture content in the samples greatly
influenced analyte recovery, and it was necessary to standardize the moisture
content in all samples in a batch prior to extraction. Closed vessel MAE has also
been utilized for the determination of PCBs in freeze-dried mussels [130],
organochlorine compounds in cod liver and fish fillets [131], xenoestrogens in
liver and muscle tissue from rainbow trout [132], and ionic arsenic species in
oyster tissue [133], as well as sulphamethazine in swine tissue [134] and fat in
chocolate [135].
Several authors have investigated novel microwave extraction applications.
For example, a simultaneous extraction-derivatization procedure was developed
for the analysis of methylmercury in biological samples [136]. In addition, the
determination of the release of dimethyl sulfide from cereals and canola was
facilitate by MAE, producing a gaseous sample in the headspace above the
microwave-extracted food sample which could be injected to a gas chromato-
graph [137].
A microwave extraction procedure may also make use of an open vessel, in
what is called focused open-vessel microwave-assisted extraction (FOV-MAE).
In a closed vessel procedure, an additional sample treatment step may be
required to remove co-extracted water from a wet sample. However, in an open
vessel procedure, the water may be removed from the sample via azeotropic
distillation. This was shown to be the case during the determination of organo-
chlorine compounds in cod liver and fish fillets [131]. Open vessel procedures
have also been applied to the determination of polychlorinated biphenyls and
chlorinated pesticides in standard reference materials of cod liver oil and freeze-
dried mussel tissue [138], and mercury in fish [139].

884
 Most of the MAE applications have concerned the determination of exoge-
nous species. Microwave extraction procedures are also useful for the character-
ization of food components. The MAE technique has been applied to the
determination of the fatty acid profile of mackerel and cod [140], trace element
analysis in plant materials [141], and the extraction of free amino acids from
foods [142]. During the extraction of lipids from milk samples, it was determined
that triglycerides are more stable using an MAE procedure compared with the
conventional Weibull-Berntrop extraction procedure, since there is less chemic-
al transformation of the triglycerides via hydrolysis [143].

25.4.3 Pressurized liquid extraction

A second technique that has been employed for the rapid extraction of food
samples at elevated temperatures is pressurized liquid extraction (PLE). PLE
methods frequently utilize the Accelerated Solvent Extraction (ASE) system
developed by Dionex, or any other system that performs static or dynamic
solvent extractions at elevated temperatures and pressures. The advantage to
performing extractions under pressurized conditions is that the upper extrac-
tion temperature is not limited by the boiling point of the solvent, as is the case
with the traditional Soxhlet system. A flow-through system such as the ASE is
also particularly beneficial in food analysis. Static extractions are performed
inside steel extraction vessels that have ample capacity for food samples, from
11-100 ml. The static extraction period is followed by elution of the extraction
solvent into a collection vial. PLE extracts are usually cleaner than microwave
extracts, since matrix components that do not dissolve in the extraction solvent
may be retained inside of the vessel. In addition, in situ clean-up methods can be
employed during PLE methods. While most food samples are blended with only
diatomaceous earth or sand prior to the PLE extraction step, further in situ
clean-up can be achieved by utilizing the matrix solid-phase dispersion tech-
nique (MSPD) [24-41] or by placing fat retainers in series inside of the vessel
[42,43]. Further details on these applications have been discussed in Section
25.2.4.
PLE methods have been utilized for both the extraction of food components
and the isolation of food contaminants. With regards to food composition, PLE
has been particularly useful in the determination of fats and lipids in food
samples such as meats [144] and egg-containing food [145]. The PLE technique
has also been applied to the determination of the fatty acid composition in cereal,
egg yolk, and chicken breast muscle [146]. PLE techniques have been employed
for the extraction and speciation of arsenic in freeze-dried carrots [147] and to
assess the levels of vitamin K-1 in medical foods [30].
Pressurized liquid extraction has also seen widespread use in the isolation of
contaminants from foodstuffs. There has been particular interest in applying
the technology to the analysis of lipid-containing foods. For example, organo-
chlorine compounds have been isolated from cod liver and fish fillets [131], and

885
various pesticides have been removed from baby foods [148]. PLE has also been
utilized for the determination of polycyclic aromatic hydrocarbons in smoked
fish and pork [149], and polychlorinated biphenyls in cod-liver oil and milk
powder [150], as well as fish tissue [151]. Corticosteroid residues in bovine liver
have been quantified using a two-step extraction method. Fat components were
first removed from the sample with hexane in a defatting step. This was followed
by the actual extraction of the analytes with a 1:1 hexane-ethyl acetate mixture
[152].
Pressurized liquid extraction has also been useful for the rapid analysis of
toxicants in fruits and vegetables. The technique has been applied to the
determination of fungicides in oranges and bananas [153], organophosphorous
pesticides in apples and carrots [154], and various pesticide residues in fresh
pear, cantaloupe, white potato, and cabbage [155].
A discussion on the utilization of hot, aqueous extraction solvents during the
analysis of food samples is included in this section. Hot, pressurized water, also
called subcritical water, is a novel extraction solvent that has been discussed at
length in an earlier chapter. The benefit of utilizing subcritical water for
analytical extractions is that the solvent strength can be tuned by varying the
extraction temperature and/or through the addition of a cosolvent. Water as a
solvent is easily obtained and disposed of, being benign to the laboratory worker
and the environment. Aqueous extractions of food samples are also convenient,
since the sample matrix does not need to be dried prior to the extraction step.
The application of this relatively new technology to the analysis of foods has
been limited thus far. The technique has been largely restricted to the isolation
of food components or contaminants from foods of plant origin. This is likely due
to the fact that hot water extracts from animal tissues can be quite turbid and
highly coloured [41]. Applications to foods of plant origin have included the
selective extraction of oxygenates from savory and peppermint [156], essential
oils from oregano [157], fennel [158], and marjoram [159], fungicides from
various vegetables [160], and organochlorine compounds from strawberries
[95,96], kohlrabi, lettuce, and tomatoes [96].
In our laboratory, the problem of coextractives from foods of animal origin
was overcome by incorporating an in situ MSPD clean-up technique into the
extraction procedure. In this case, the pesticide atrazine was isolated from beef
kidney that was dispersed with an acrylic polymer [41]. It was necessary to
modify the water with another benign solvent (ethanol at 30% v/v) in order to
attain the solvent strength necessary to achieve complete recovery of the target
analyte from the dispersed matrix.

25.4.4 Supercritical fluid extraction

The use of supercritical fluids in food analysis has grown tremendously in the
past decade. Two applications that have had significant development in recent
years have been the determination of fat and associated nutrients, as well as the

886
isolation of pesticides from food matrices. Supercritical fluid extraction (SFE)
technology has also been applied to the determination of PAHs and PCBs, drugs,
and other food toxicants. Many of these applications are summarized in Table
25.6, all of which utilize supercritical carbon dioxide (SC CO2). SC CO2 continues
to be the fluid of choice, since its critical parameters (31.1°C, 72.8 bar) are easily
achieved with high pressure instrumentation. Further, it is non-toxic and easy
to obtain. Some of the SF-based methodologies utilize suitable modifiers to
enhance analyte recovery. Protocols have been developed that encompass
sample clean-up, derivatization, and automation in the methodology. The
clean-up of lipids has been of critical importance during the extraction of
non-polar compounds such as pesticides, PAHs and PCBs, due to the propensity
of these toxicants to accumulate in the lipid phase.
It is probably fair to say that analytical SFE will be a method of choice for the
fat determination of foods in the future. Toward that end, several collaborative
and peer-verified methods have been published utilizing SFE for the determina-
tion of lipid levels in oilseeds, meats, and food products.
There has been limited use of alternative fluids for the analysis of food
samples. For example, three fluids were examined for the removal of ethoxyquin
from lean beef and beef fat: carbon dioxide, trifluoromethane, and 1,1,1,2-tetra-
fluoroethane. While CO2 appeared to react with the analyte during the
extraction, methanol-modified hydrofluorocarbons provided more complete ex-
tractions than the pure fluids. Quantitative extraction was achieved at the 0.5
ppm level [197]. Binary mixtures of CO 2 and nitrogen have also been utilized for
the selective extraction of organochlorine and organophosphate pesticides from
poultry. The binary mixtures provided quantitative recoveries of the analytes
while significantly reducing lipid solubility and coextraction [198].

25.5 FINAL REMARKS

In this concise chapter, we have attempted to provide an overview of modern

methods that can be used in preparing samples for food analysis. Food analysis is
a complex area that has led to a plethora of approaches due to the complexity of
the sample matrices. An emphasis has been placed here on recently developed
techniques and methods that are rapid and minimize the generation of chemical
waste (e.g. organic solvents). The authors hope that the near 200 references that
have been cited may aid the analyst in further selecting the sample preparation
method that is most appropriate for solving the problem being investigated.

REFERENCES

1 A. Ambrus, H.-P. Their, Pure Appl. Chem., 58 (1986) 1035.

2 V. Leoni, A.M. Caricchia and S. Chiavarini, J. AOAC Int., 78 (1992) 511.
3 J. Tekel, T. HudecovA and K. Pecnfkovd, Eur. Food Res. Technol., 213 (2001) 250.
4 R. Majors, LC/GC, 16 (1998) 436.
5 P.M. Gy, LC/GC, 12 (1994) 808.

887
TABLE 25.6
Supercritical fluid extraction of food samples
Sample Analyte Ref.
Fatand Nutrients
Ground beef Total fat 161-163
Ground cumin Volatile oils 164
Egg-containing foods Cholesterol 165
Chocolate Triacylglycerols 135
Infant formula Fat 166
Milk products Total fat 167
Milk products Total fat 168
Deep-fried chicken and potato Lipids 169
Oilseeds Total fat 162,163
Bakery samples Total fat 163
Milk powder, infant formula Fat-soluble vitamins 170,171
Tomato skin Lycopene 172
Raisins 5-Hydroxymethyl-2-furaldehyde 173
Pesticides
Butter fat, corn oil Organochlorine and organophosphate 174
Wheat, maize Organophosphate 175
Eggs Triazine 176
Eggs Organochlorine 177
Sugar cane, oranges Diuron 178
Garlic Organochlorine 179
Cereals Various 180
Beef and chicken tissue Carbamate 181
Apples Fenpyroximate 182
Apples Various 183
Chicken fat, ground beef, lard Organochlorine and organophosphate 184
Grains Organochlorine and organophosphate 185
Poultry tissue Organophosphate 186
Drugs
Bovine liver 3-agonists 187
Poultry eggs and muscle Nicarbazin 188
Animal tissues Steroids 189
Eggs Sulfamethazine 190
Pig fat Androstenone 191
PAHs and PCBs
Smoked meat PAHs 18
Toasted bread PAHs 19
Liver PAHs 20
Fish PCBs 21
Fat PCBs 22
Miscellaneous
Irradiated foods Hydrocarbons and 2-alkylcyclobutanones 192,193
Cured meats Nitrosamines 194
Beef Liver Aflatoxin Ml 195
Corn Aftatoxin B1 196

888
6 C.A. Ramsey and J. Suggs, Environ. Test. Anal., 10(2) (2001) 13.
7 F.F. Pitard, Pierre Gy's Sampling Theory and Sampling Practice, 2 d Edn. CRC
Press, Boca Raton, 1993.
8 P.M. Gy, Sampling of Heterogeneous and Dynamic Material Systems. Elsevier,
Amsterdam, 1992.
9 P.M. Gy, Sampling of Particulate Materials, Theory and Practice. Elsevier,
Amsterdam, 1982.
10 R.F. Cross, LC/GC, 18 (2000) 468.
11 L.W. Aurand, A.E. Woods and M.R. Wells, Food Composition and Analysis. Van
Nostrand Reinhold, New York, 1987, Chapter 2.
12 M.A. Joslyn, Methods in Food Analysis, 2nd Edn. Academic Press, New York, 1970,
Chapter 2.
13 W.G. Cochran, Sampling Techniques. Wiley, New York, 1959.
14 M.A. Joslyn, Methods in Food Analysis, 2nd Edn. Academic Press, New York, 1970,
Chapter 3.
15 B. Berilter, X. Giard, C.J. Zizek and J. Fssler, Am. Lab., 33(21) (2001) 10.
16 H. S0rensen, S. Sorensen, C. Bjergegaard and S. Michaelsen, Chromatographyand
Capillary Electrophoresisin Food Analysis. Royal Society of Chemistry, Cambridge,
UK, 1999, p. 73.
17 Ibid., p. 71
18 M.Y. Ali and R.B. Cole, J. Agric. Food Chem., 49 (2001) 4192.
19 M.N. Kayali-Sayadi, S. Rubio-Barraso, R. Garcia-Iranzo and L.M. Polo-Diez, J. Liq.
Chrom. Rel. Technol., 23 (2000) 1913.
20 S.G. Amigo, M.S.G. Falcon, M.A.L. Yusty and J.S. Lozano, Fres. J. Anal. Chem., 367
(2000) 572.
21 K. Yasumura, E. Kitamura, M. Uno and M. Tamaki, J. FoodHyg. Soc. Jpn., 42 (2001)
1.
22 E. Bjtrklund, M. JBremo and L. Mathiasson, J. Liq. Chrom. Rel. Technol., 23 (2000)
2337.
23 S.A. Barker, J. Chromatogr.A, 885 (2000) 115.
24 S.A. Barker, J. Chromatogr.A, 880 (2000) 63.
25 S.A. Barker, A.R. Long and C.R. Short, J. Chromatogr., 475 (1989) 353.
26 A.I. Valenzuela, R. Lorenzini, M.J. Redondo and G. Font, J. Chromatogr.A, 839
(1999) 101.
27 M. Fernandez, Y. Pic6 and J. Mafies, J. Chromatogr.A, 871 (2000) 43.
28 C. Yagie, S. Bayarri, R. Ldzaro, P. Conchello, A. Arifo and A. Herrera, J. AOAC Int.,
84 (2001) 1561.
29 S.A. Barker and A.R. Long, J. AOAC Int., 77 (1994) 848.
30 G.W. Chase, Jr. and B. Thompson, J. AOAC Int., 83 (2000) 407.
31 E. Brandsteterova, P. Kubalec, L. BovanovA, P. Simko, A. Bednarikovd and L.
Machackovd, Z. Lebensm. Unters. Forsch. A, 205 (1997) 311.
32 J. Tolls, M. Haller and D.T.H.M. Sijm, Anal. Chem., 71 (1999) 5242.
33 J. Tolls, M. Haller and D.T.H.M. Sijm, J. Chromatogr.A, 839 (1999) 109.
34 M. Zhao, F. van der Wielen and P. de Voogt, J. Chromatogr. A, 837 (1999) 129.
35 P. Guant and S.A. Barker, Int. J. Environ. Pollut., 13 (2000) 284.
36 A.R. Long, M.M. Soliman and S.A. Barker, J. AOAC Int., 74 (1991) 493.
37 S. Collins, M. O'Keefe, R. Calverley and M.R. Smyth, Proceedingsof EuroresidueIII
(1996) 340.
38 E. Horne, M. O'Keefe, C. Desbrow and A. Howells, Analyst, 123 (1998) 2517.
39 K. Kishida and N. Furusawa, J. Chromatogr.A, 937 (2001) 49.
40 E.M. Kristenson, E.G.J. Haverkate, C.J. Slooten, L. Ramos, R.J.J. Vreuls and
U.A.Th. Brinkman, J. Chromatogr. A, 917 (2001) 277.

889
41 M.S.S. Curren and J.W. King, J. Agric. Food Chem., 49 (2001) 2175.
42 E. Bjorklund, A. Miller and C. von Holst, Anal. Chem., 73 (2001) 4050.
43 Dionex, Application Note ASE 322, Dionex Corporation, Sunnyvale, CA, 1996.
44 M. Jaremo, E. BjSrklund, M. Milsson, L. Karlsson and L. Mathiasson, J. Chromatogr.
A, 877 (2000) 167.
45 J.E. France, J.W. King and J.M. Snyder, J. Agric. Food Chem., 39 (1991) 1871.
46 R.C. Hale and M.O. Gaylor, Environ. Sci. Technol., 29 (1995) 1043.
47 E.G. Alley and G. Lu, J. AOAC Int., 78 (1995) 1051.
48 E. Bjorklund, M. Jiaremo and L. Mathiasson, J. Liq. Chrom. Rel. Technol., 23 (2000)
2337.
49 R.E. Majors, LC/GC, 14 (1996) 936.
50 J.R. Dean, Extraction Methods for Environmental Analysis, .iley, Chichester, UK,
1998.
51 A.J. Holden, in: A.J. Handley (Ed.), Extraction Methods in Organic Analysis.
Sheffield Academic Press, Sheffield, UK, 1999, p. 5.
52 V. Ruiz-Gutierrez and M.C. Perez-Camino, J. Chromatogr.A, 885 (2000) 321.
53 C. Dominguez, D.A. Guillen and C.G. Barroso, J. Chromatogr. A, 918 (2001) 303.
54 K. Nordmeyer and H.-P. Thier, Z. Lebensm. Unters. Forsch. A, 208 (1999) 259.
55 N.G. Tsiropoulos, P.G. Aplada-Sarlis and G.E. Miliadis, J. AOAC Int., 82 (1999) 213.
56 F. Toribio, E. Moyano, L. Puignou and M.T. Galceran, J. Chromatogr.A, 880 (2000)
101.
57 Y. Ito, Y. Ikai, H. Oka, H. Matsumoto, Y. Miyazaki, K. Takeba and H. Nagase, J.
Chromatogr. A, 911 (2001) 217.
58 G.A. Csiktusnddi Kiss, E. Forgacs, T. Cserhdti, M. Candeias, L. Vilas-Boas, R. Bronze
and I. Spranger, J. Chromatogr. A, 889 (2000) 51.
59 J.J. Jimbnez, J.L. Bernal, M.J. del Nozal, L. Toribio and E. Arias, J. Chromatogr.A,
919 (2001) 147.
60 S.M. Dugar, J.N. Leibowitz and R.H. Dyer, J. AOAC Int., 77 (1994) 1335.
61 J.F. Fisher and R.L Rouseff, J. Agric. Food Chem., 34 (1986) 985.
62 Y. Daghbouche, S. Garrigues, M.T. Vidal and M. de la Guardia, Anal. Chem., 69
(1997) 1086.
63 Y. Amakura, M. Okada, S. Tusji and Y. Tonogai, J. Chromatogr.A, 891 (2000) 183.
64 M. Adahchour, R.J.J. Vreuls, A. van der Heijden and U.A.Th. Brinkman, J.
Chromatogr. A, 844 (1999) 295.
65 J. Dallfige, T. Hankemeier, R.J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr.A,
830 (1999) 377.
66 J.F. Lawrence, C. M6nard, M.-C. Hennion, V. Pichon, F. Le Goffic and N. Durand, J.
Chromatogr. A, 732 (1996) 277.
67 J.F. Lawrence, C. M6nard, M.-C. Hennion, V. Pichon, F. Le Goffic and N. Durand, J.
Chromatogr. A, 752 (1996) 147.
68 J.F. Lawrence and C. M6nard, J. Chromatogr.A, 922 (2001) 111.
69 P. Zillner, J. Jodlbauer and W. Lindner, J. Chromatogr. A, 858 (1999) 167.
70 B. Spitzer, M. Cichna, P. Markl, G. Sontag, D. Knopp and R. Niessner, J.
Chromatogr.A, 880 (2000) 113.
71 J. Stroka, R. van Otterdijk and E. Anklam, J. Chromatogr.A, 904 (2000) 251.
72 C. Crescenzi, S. Bayoudh, P.A.G. Cormack, T. Klein and K. Ensing, Anal. Chem., 73
(2001) 2171.
73 A. Zander, P. Findlay, T. Renner, B. Sellergren and A. Swietlow, Anal. Chem., 70
(1998) 3304.
74 M.T. Muldoon and L.H. Stanker, Anal. Chem., 69 (1997) 803.
75 0. Ramstrom, K. Skudar, J. Haines, P. Patel and 0. Briiggeman, J. Agric. Food
Chem., 49 (2001) 2105.

890
76 A.M. Szmigielska, J.J. Schoenau and V. Levers, J. Agric. Food Chem., 48 (2000) 4487.
77 A.M. Szmigielska and J.J. Schoenau, J. Agric. Food Chem., 48 (2000) 5190.
78 J.C. Wu, W. Xie and J. Pawliszyn, Analyst, 125 (2000) 2216.
79 H.H. Jelen, K. Wlazly, E. Wasowicz and E. Kaminski, J. Agric. Food Chem., 46 (1998)
1469.
80 C. Sala, M. Mestres, M.P. Marti, O. Busto and J. Guasch, J. Chromatogr. A, 880
(2000) 93.
81 A.F. Gonzdlez-C6rdova and B. Vallejo-Cordoba, J. Agric. Food Chem., 49 (2001)
4603.
82 S.E. Ebeler, G.M. Sun, M. Datta, P. Stremple and A.K. Vickers, J. AOAC Int., 84
(2001) 479.
83 T. Sostaric, M.C. Boyce and E.E. Spickett, J. Agric. Food Chem., 48 (2000) 5802.
84 M. Jia, Q.H. Zhang and D.B. Min, J. Agric. Food Chem., 46 (1998) 2744.
85 H.H. Jelen, M. Obuchowska, R. Zawirska-Wojtasiak and E. Wasowicz, J. Agric. Food
Chem., 48 (2000) 2360.
86 M. Mestres, C. Sala, M.P. Marti, O. Busto and J. Guasch, J. Chromatogr. A, 835
(1999) 137.
87 Y. Hayasaka and E.J. Bartowsky, J. Agric. Food Chem., 47 (1999) 612.
88 A.L. Simplicio and L.V. Boas, J. Chromatogr.A, 833 (1999) 35.
89 J.J. Jimenez, J.L. Bernal, M.J. del Nozal, M.T. Martin and A.L. Mayorga, J. Chroma-
togr. A, 829 (1999) 269.
90 S.B. Hawthorne, D.J. Miller, J. Pawliszyn and C.L. Arthur, J. Chromatogr., 603
(1992) 185.
91 Z. Wang, B. Hennion, L. Urruty and M. Montury, FoodAdditives Contam., 17 (2000)
915.
92 C.G. Zambonin, L. Monaci and A. Aresta, Food Chem., 75 (2001) 249.
93 E. Baltussen, P. Sandra, F. David and C. Cramers, J. Microcolumn Sep., 11 (1999)
737.
94 P. Sandra, B. Tienpont, J. Vercammen, A. Tredoux, T. Sandra and F. David, J.
Chromatogr.A, 928 (2001) 117.
95 L. Wennrich, P. Popp, G. Koller and J. Breuste, J. AOAC Int., 84 (2001) 1194.
96 L. Wennrich, P. Popp and J. Breuste, Chromatographia,Suppl. 53 (2001) S-380.
97 A.G.J. Tredoux, H.H. Lauer, T. Heideman and P. Sandra, J. High Resolut. Chroma-
togr., 23 (2000) 644.
98 M. Kreck, A. Scharrer, S. Bilke and A. Mosandl, Eur.FoodRes. Tech., 213 (2001) 389.
99 J.A. Jonsson and L. Mathiasson, in: P.R. Brown and E. Grushka (Ed.), Advances in
Chromatography. Marcel Dekker, New York, 2001, Chapter 2.
100 J.A. J6nsson and L. Mathiasson, J. Chromatogr. A, 902 (2000) 205.
101 R.J. Maxwell and E. Cohen, J. High Resolut. Chromatogr., 21 (1998) 241.
102 M.J. Schneider and D.J. Donoghue, J. AOAC Int., 83 (2000) 1306.
103 M.J. Schneider, J. Chromatogr. Sci., 39 (2001) 351.
104 R.J. Maxwell, E. Cohen and D.J. Donoghue, J. Agric. Food Chem., 47 (1999) 1563.
105 F. Palmisano, M. Quinto, R. Rizzi and P.G. Zambonin, Analyst, 126 (2001) 866.
106 D. Moscone, R.A. Bernardo, E. Marconi, A. Amine and G. Palleschi, Analyst, 124
(1999) 325.
107 S. Mannino, O. Brenna, S. Buratti and M.S. Cosio, Electroanalysis9 (1997) 1337.
108 M.M.D. Zamarreno, A.S. Perez, M.B.Rangel and J.H. Mendez, Anal. Chim. Acta, 386
(1999) 99.
109 A. Sanchez-Perez, M.M. Delgado-Zamarreno, M. Bustamante-Rangel, and J.
Hernandez-Mendez, J. Chromatogr. A, 881 (2000) 229.
110 R. Carabias-Martinez, E. Rodriguez-Gonzalo, P.H. Paniagua-Marcos and J.
Hernandez-Mendez, J. Chromatogr. A, 869 (2000) 427.

891
111 M. Luque, E. Luque-Perez, A. Rios and M. Valcdrcel, Anal. Chim. Acta, 410 (2000)
127.
112 E. Luque-PBrez, A. Rios, M. Valcarcel, L.-G. Danielsson and F. Ingman, Lab. Autom.
Inform. Manage., 34 (1999) 131.
113 J.B. Cai, B.H. Liu and Q.D. Su, J. Chromatogr.A, 930 (2001) 1.
114 C. Palma-Harris, R.F. McFeeters and H.P. Fleming, J. Agric. Food. Chem., 49 (2001)
4203.
115 D. Shooter, N. Jayatissa and N. Renner, J. Dairy Res., 66 (1999) 115.
116 J. Sung, B.D Gardner, J.F. Holland and R.M. Beaudry, J. Agric. Food Chem., 45
(1997) 1801.
117 E.P. Jarvenpaa, Z. Zhang, R. Huopalahti and J.W. King, Z. Lebensm. Unters. Forsch
A, 207 (1998) 39.
118 H.W. Chin, R.A. Bernhard and M. Rosenberg, J. Food Sci., 61 (1996) 1118.
119 N.P. Sen, S.W. Seaman and B.D. Page, J. Chromatogr.A, 788 (1997) 131.
120 M. Zhu, F.J. Aviles, E.D. Conte, D.W. Miller and P.W. Perschbacher, J. Chromatogr.
A, 833 (1999) 223.
121 J. Song, L. Fan and R.M. Beaudry, J. Agric. Food Chem., 46 (1998) 3721.
122 J.W. Arnold and S.D. Senter, J. Sci. Food Agric., 78 (1998) 343.
123 C.S. Eskilsson and E. Bjorklund, J. Chromatogr.A, 902 (2000) 227.
124 A. Abu-Samra, J.S. Morris and S.R. Koirtyohann, Anal. Chem., 47 (1975) 1475.
125 K. Ganzler, A. Salg6 and K. Valk6, J. Chromatogr., 371 (1986) 299.
126 K. Ganzler, A. Salg6, Z. Lebensm. Unters. Forsch., 184 (1987) 274.
127 J.R.J. Par6, J.M.R. BBlanger and S.S. Stafford, Trends Anal. Chem., 13 (1994) 176.
128 J.R.J. Pare, US Patent 5 002 784 (1991).
129 S. Jayaraman, R.J. Pruell and R. McKinney, Chemosphere, 44 (2001) 181.
130 N. Carro, I. Garcia and M. Llompart, Analusis, 28 (2000) 720.
131 M. Weichbrodt, W. Vetter and B. Luckas, J. AOACInt., 83 (2000) 1334.
132 S.N. Pedersen and C. Lindholst, J. Chromatogr. A, 864 (1999) 17.
133 A. Chatterjee, Talanta, 51 (2000) 303.
134 M.H. Akhtar, M.L. Wong, S.R.H. Crooks and A. Sauve, Food Additives Contam., 15
(1998) 542.
135 C. Simoneau, C. Naudin, P. Hannaert and E. Anklam, Food Res. Int., 33 (2000) 733.
136 M. Abuin, A.M. Carro and R.A. Lorenzo, J. Chromatogr. A, 889 (2000) 185.
137 Y.L. Ren, J. Agric. Food Chem., 49 (2001) 1737.
138 S. Thompson and H. Budzinski, Int. J. Env. Anal. Chem., 76 (2000) 49.
139 C.S. Chiou, S.J. Jiang and K.S.K. Danadurai, Spectrochim. Acta B, 56 (2001) 1133.
140 A. Batista, W. Vetter and B. Luckas, Eur. Food Res. Technol., 212 (2001) 377.
141 J. Borkowska-Burnecka, Fres. J. Anal. Chem., 368 (2000) 633.
142 A. Kovacs, K. Ganzler and L. Simon-Sarkadi, Z. Lebensm. Unters. Forsch., 207 (1998)
26.
143 L.E. Garcia-Ayuso, J. Velasco, M.C. Dobarganes and M.D.L. de Castro, Int. Dairy J.,
9 (1999) 667.
144 Dionex, Application Note ASE 334, Dionex Corporation, Sunnyvale, CA, 1999.
145 E. Boselli, V. Velazco, M.F. Caboni and G. Lercker, J. Chromatogr. A, 917 (2001) 239.
146 K. Schafer, Anal. Chim. Acta, 358 (1998) 69.
147 N.P. Vela, D.T. Heitkemper and K.R. Stewart, Analyst, 126 (2001) 1011.
148 J.C. Chuang, K. Hart, J.S. Chang, L.E. Boman, J.M. Van Emon and A.W. Reed, Anal.
Chim. Acta, 444 (2001) 87.
149 G.D. Wang, A.S. Lee, M. Lewis, B. Kamath and R.K. Archer, J. Agric. Food Chem., 47
(1999) 1062.
150 A. Miiller, E. Bjbrklund and C. von Holst, J. Chromatogr. A, 925 (2001) 197.
151 Dionex, Application Note ASE 322, Dionex Corporation, Sunnyvale, CA, 1996.

892
152 R. Draisci, C. Marchiafava, L. Palleschi, P. Cammarata and S. Cavalli, J. Chroma-
togr. B, 753 (2001) 217.
153 S. Kakimoto, H. Obana, M. Okihashi and S. Hori, J. Food Hyg. Soc. Japan,38 (1997)
358.
154 B.E. Richter, F. Hoefler and M. Linkerhaegner, LC/GC, 19 (2001) 408.
155 K. Adou, W.R. Bontoyan and P.J. Sweeney, J. Agric. Food Chem., 49 (2001) 4153.
156 A. Kubdtovb, A.J.M. Lagadec, D.J. Miller and S.B. Hawthorne, Flav. Frag. J., 19
(2001) 64.
157 R.S. Ayala and M.D.L. de Castro, Food Chem., 75 (2001) 109.
158 L. Gamiz-Garcia and M.D.L de Castro, Talanta, 51 (2000) 1179.
159 M.M. Jimenez-Carmona, J.L. Ubera and M.D.L. de Castro, J. Chromatogr.A, 855
(1999) 625.
160 T.M. Pawlowski and C.F. Poole, J. Agric. Food Chem., 46 (1998) 3124.
161 F.J. Eller and J.W. King, J. Agric. Food Chem., 49 (2001) 4609.
162 S.L. Taylor, F.J. Eller and J.W. King, Food. Res. Int., 30 (1997) 365.
163 F.J. Eller and J.W. King, J. Agric. Food Chem., 46 (1998) 3657.
164 D.L. Heikes, B. Scott and N.A. Gorzovalitis, J. AOAC Int., 84 (2001) 1130.
165 E. Boselli, M.F. Caboni and G. Lercker, Eur. Food Res. Technol., 212 (2001) 244.
166 M. Ashraf-Khorassani, R. Hellmer, L.T. Taylor and D.C. Messer, Am. Lab., 32 (2000)
40.
167 L. Manganiello, A. Rios and M. Valcarcel, J. Chromatogr.A, 874 (2000) 265.
168 F. Dionisi, B. Hug, J.M. Aeschlimann and A. Houllemar, J. FoodSci., 64 (1999) 612.
169 N. Devineni, P. Mallikarjunan, M.S. Chinnan and R.D. Phillips, J. Am. Oil. Chem.
Soc., 74 (1997) 1517.
170 C. Turner, M. Persson, L. Mathiasson, P. Adlercreutz and J.W. King, Enzyme Microb.
Technol., 29 (2001) 111.
171 C. Turner and L. Mathiasson, J. Chromatogr.A, 874 (2000) 275.
172 M. Ollanketo, K. Hartonen, M.L. Riekkola, Y. Holm and R. Hiltunen, Eur. Food Res.
Technol., 212 (2001) 561.
173 M. Palma and L.T. Taylor, J. Agric. Food Chem., 49 (2001) 628.
174 M.L. Hopper, J. Chromatogr. A, 840 (1999) 93.
175 K.N.T. Norman and S.H.W. Panton, J. Chromatogr.A, 907 (2001) 247.
176 J.W. Pensabene, W. Fiddler and D.J. Donoghue, J. Agric. Food Chem., 48 (2000)
1668.
177 W. Fiddler, J.W. Pensabene, R.A. Gates and D.J. Donoghue, J. Agric. Food Chem., 47
(1999) 206.
178 F.M. Lancas and S.R. Rissato, J. Microcolumn. Sep., 10 (1998) 473.
179 J.H. Wang, Q.A. Xu and K. Jiao, J. Chromatogr. A, 818 (1998) 138.
180 K. Yoshii, Y. Tonogai, Y. Tsumura, Y. Nakamura and T. Shibata, J. Food Hyg. Soc.
Japan,39 (1998) 184.
181 K.J. Voorhees, A.A. Gharaibeh and B. Murugaverl, J. Agric. Food Chem., 46 (1998)
2353.
182 B.L. Halvorsen, C. Thomsen, T. Greibrokk and E. Lundanes, J. Chromatogr. A, 880
(2000) 121.
183 R. Stefani, M. Buzzi and R. Grazzi, J. Chromatogr. A, 782 (1997) 123.
184 K.-S. Nam and J.W. King, J. High Resolut. Chromatogr., 17 (1994) 577.
185 J.W. King, M.L. Hopper, R.G. Luchtefeld, S.L. Taylor and W.L. Orton, J. AOAC Int.,
76 (1993) 857.
186 J.M. Snyder, J.W. King, L.D. Rowe and J.A. Woerner, J. AOAC Int., 76 (1993) 888.
187 M.J. O'Keefe, M. O'Keefe and J.D. Glennon, Analyst, 124 (1999) 1355.
188 D.K. Matabudul, N.T. Crosby and S. Sumar, Analyst, 124 (1999) 499.
189 A.A.M. Stolker, P.W. Zoontjes and L.A. van Ginkel, Analyst, 123 (1998) 2671.

893
190 J.W. Pensabene, W. Fiddler and D.J. Donoghue, J. Food Sci., 63 (1998) 25.
191 M.A. Magard, H.E.B. Berg, V. Tagesson, M.L.G. Jremo, L.L.H. Karlsson, L.J.E.
Mathiasson, M. Bonneau and J. Hansen-Moller, J. Agric. Food Chem., 43 (1995) 114.
192 P. Horvatovich, M. Miesch, C. Hasselmann and E. Marchioni, J. Chromatogr.A, 897
(2000) 259.
193 E.M. Stewart, W.C. McRoberts, J.T.G. Hamilton and W.D. Graham, J. AOAC Int., 84
(2001) 976.
194 J.W. Pensabene and W. Fiddler, Supercrit. Fluid Meth. Protoc., 13 (2000) 23.
195 S.L. Taylor, J.W. King, J.I. Greer and J.L. Richard, J. Food Protect., 60 (1997) 698.
196 S.L Taylor, J.W. King, J.L. Richard and J.I Greer, J. Agric. Food Chem., 41 (1993)
910.
197 D.R. Brannegan, M. Ashraf-Khorassani and L.T. Taylor, Chromatographia, 54
(2001) 399.
198 J.W. King and Z. Zhang, Anal. Chem., 70 (1998) 1431.

894
Chapter26

Sample preparation techniques for soil

analysis
Annamaria Halasz, Josep M. Bayona and Jalal Hawari

26.1 INTRODUCTION

Products and byproducts from various industrial activities including petroleum,

mining, pulp and paper, printing, polymer, electrochemical, agricultural, food
and pharmaceutical industry have all contributed to the contamination of vast
volumes of soil. The most dangerous pollutants currently found in soil include
petroleum products such as PAHs, BTEX, chlorinated solvents such as PCBs,
PCE, TCE and VC, explosives such as RDX, HMX and TNT, pesticides and heavy
metals. Once in soil most pollutants undergo several transport and transforma-
tion mechanisms and migrate through subsurface soil to cause groundwater con-
tamination. To understand the environmental fate and the impact of pollutants
we must first determine the type, level and distribution of these contaminants in
soil. However, soil is an extremely heterogeneous matrix and the presence of
chemicals is governed by temporal and spatial variations at the site. Further-
more, most common environmental pollutants are labile, undergoing several
transformation pathways, causing severe variation in pollutant distribution at
the site. These transformation pathways can be greatly affected by factors such
as zero valent metals (e.g. Fe(0)), sunlight irradiation, soil indigenous microflora
and other plant species which can all lead to a partial or complete (bio)transfor-
mation of the chemical in soil. Consequently, the concentration variation of cer-
tain pollutants in soil due to spatial sampling heterogeneity can sometimes
exceed the analytical error by a factor of 10 [1,2].
For analyses of contaminated soil, a sampling strategy that minimizes soil
heterogeneity and increases sample representativity must be developed. Table
26.1 summarizes some of the most practised sampling techniques for soil
analysis. Techniques of sampling and field analysis may be found in Chapter 11
and in Refs. [6] and [7]. This chapter describes post-sampling methods for soil
preparation before analyses. In order to identify and quantify targeted
pollutants in soil(s), particularly field samples, rigorous sample preparation and
separation techniques are necessary to eliminate analytical bias caused by the

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 895
TABLE 26.1
Soil sampling plans
Sampling plans Process Characteristics, Examples

Exploratory sampling Few samples are selected to Qualitative assessment of soil

[3,4] confirm contamination. An
in-depth knowledge of the
sampler is required.
Simple random Probability sampling of five to
sampling [3] ten samples (max 25) for small
sites (<0.5 ha).
Stratified random The total sampling area is broken
sampling [3,4] into a number of strata (sampling
units) and random samples are
taken from each stratum.
Systematic or grid Sampling units are selected to
sampling [3,4] completely cover the targeted
soil volume.
Composite sampling The collected samples are mixed
[3,4] to form a composite using
different compositing scheme.
Geostatistics [3,5] Two-step process using
three-dimensional data analyses
such as kriging, a linear-weighted very short distances due to
gridding method.
Sampling for linear Requires the division of study
disturbance [3] area into strata and the
implementation of a stratified
random sampling plan.
where an impact or damage is
visible or anticipated
(Single-industry waste site; Small
waste site; Chemical spills).
Postreclamation assessment,
land-farming operation, soil
storage piles.
More precise than simple random
sampling, the stratification must
eliminate some variation
attributable to sampling.
For spill sites where there is a
likely concentration gradient.
For a given analytical cost, a more
representative sample is obtained,
when only the average value is
needed. Only for soils without
physical disturbance.
Used when the concentrations of
pollutants vary remarkably over
heterogeneity of the soil and the
accidental nature of
contaminating processes.
Used for impact assessment to
assess linear disturbance or
environmental damage caused by
roads, power lines, or pipelines
accidents.

presence of other interference. Table 26.2 summarizes currently known and

practised sample preparation techniques for soil analysis. Following sample
preparation and isolation of the analyte, a number of analytical techniques
comprised of separation and detection including LC-MS, SPME-GC/MS, SPME-
ICP-MS, IR, NMR and micellar electrokinetic chromatography (MEKC-UV
detection) can then be used for the determination of targeted chemicals. This
chapter will first highlight the desired results of environmental analytical
chemistry and then describe some of the most widely applied sample preparation
techniques for soil analysis.

896
26.2 DESIRED DELIVERABLES OF ENVIRONMENTAL ANALYTICAL
CHEMISTRY

Environmental analytical chemistry should provide rigorous data on the type,

concentration, breakdown products, and distribution of the pollutant in various
compartment of real soil samples. Such data can then be used by site managers
and engineers to understand and predict the fate and impact of the pollutant at
contaminated sites. Pollutants have characteristic physicochemical properties
and thus are expected to experience transformation and transport mechanisms
differently. Often water solubility and soil-water partition coefficients, Kd,
provide quantitative evidence on the migration of chemicals from soil through
subsurface soil to cause groundwater contamination. For example TNT has
water solubility equals to 145 mg/l and a soil-water partition coefficient (Kd) > 4
whereas RDX and HMX have lower water solubilities (45 mg/l and 5 mg/l,
respectively) and lower Kd values < 1 [19-23], implying that TNT will migrate
faster through subsurface soil to groundwater. Also the octanol/water partition
coefficient, Kw, can provide an understanding of the distribution of the chemical
between soil organic matter and water or between soil and plant tissue. Thus Kow
values can provide insight about important transport and transformation pro-
cesses such as natural attenuation and phytoremediation. It has been shown
that compounds with 1 < Kow < 3 are most easily transported into the plants. For
example, TNT with a reported K.o around 2 has been successfully phyto-
remediated [24,25].
Therefore, environmental analytical chemistry should be able to provide the
physicochemical data that can provide understanding of the factors leading to
the immobilization of chemicals and the factors that determine their binding
stability in soil. Immobilization that leads to irreversible binding with long term
stability in soil can be considered as a harmless in situ soil remediation technol-
ogy. Analytical chemistry should thus broaden our understanding of the mecha-
nisms and kinetics of binding of chemical pollutants and their intermediate
products onto soil or other biological matrices such as plant tissue. This requires
accurate measurements of the pollutant sequestration and distribution between
available compartments within the soil matrix.
Environmental analytical chemistry must also focus more on on-site moni-
toring because field monitoring is an important first step in the characterization
of contaminated sites, which can provide rapid initial estimates of the concentra-
tion of the pollutant at the site and will thus help decide whether an extensive
laboratory analysis of the soil is needed. Contaminated sites including brown-
fields are a worldwide problem, necessitating the development of unified, inter-
national on-site characterization protocols. Such protocols should be conceived
to provide an analytical data base that can be used to identify pollutants, assess
environmental risks and make recommendations of whether site remediation is
needed. At present, on-site characterization is not widely used but is emerging as
a fast-growing technology for site monitoring and assessment, particularly in

897
studies related to natural attenuation [7,26]. Current on-site characterization
methods based on immunoassay [27-29] and colorimetry [30] suffer from severe
interference from other soil organic constituents. The immunoassay is based on
a reaction between a targeted chemical and a specific antibody to produce a
response that can be quantitated using radioactivity or fluorescence detection.
Colorimetric detection is based on reacting the pollutants with a chemical
reagent to produce a color that can be monitored by UV/VIS detection. In the
case of cyclic nitramine explosives such as RDX and HMX the compounds are
allowed to react with Zn/HCl to produce a pink to yellow color which can then be
quantified using a portable spectrophotometer at X 607 nm [29,30].
Analytical chemistry must also be able to provide an accurate and precise
assessment of the performance of soil (bio)remediation technologies. At present,
most field-scale remediation technologies are assessed based on the removal of
the pollutant while the degradation products and their fate generally remain
unknown [26,31]. The need for fast analytical techniques capable of performing
on-line analysis of soil during remediation technologies is a challenge that
should attract exploration by analytical chemists.
Last but not least, analysis of environmental samples, particularly soil,
should focus more on solvent-free sample preparation techniques. Solventless
sample preparation should be a legitimate strategy for soil analysis since the
ultimate objective of identifying the type and concentration of pollutants is to
safely remove them from the contaminated soil. Many solventless sample prepa-
ration techniques have already been developed and are available commercially.
Some of these techniques are based on SFE, SPME and headspace techniques as
summarized in Fig. 26.1. Mester et al. [32] have reviewed most currently
practised solventless sample preparation techniques for the analysis of several
environmental matrices including biological, sediment and plants. The following
discussion will describe some of these solvent-free sample preparation tech-
niques and compare their performance with other solvent-based traditional
methods such as the century-old Soxhlet [33].

Fig. 26.1. Solvent-free sample preparation techniques for soil analysis.

898
26.3 SAMPLE PREPARATION PROTOCOLS FOR SOIL ANALYSIS

The most difficult step during soil analysis is sample preparation, which requires
that the targeted analyte(s) be effectively extracted from the soil matrix suffi-
ciently pure for subsequent detection and quantification. Several types of sam-
ple extraction techniques are currently available for soil analysis (Table 26.2).
Selection among these techniques depends on the type of soil and the type(s) of
analyte(s) to be extracted. To achieve optimal analyte recovery from soil, a
proper solvent with a particular polarity is routinely used in either a Soxhlet or a
sonicator. For instance, hydrophobic aliphatic hydrocarbons (oil and grease) can
easily be extracted from soil with a nonpolar solvent such as hexane (EPA 9071
[4]), whereas the more polar PAHs require a more polar solvent such as methyl-
ene dichloride. Also, polynitroorganic compounds such as explosives (EPA 8330
[4]) and polychlorinated compounds such as PCBs [34] are best extracted with
polar solvents such as acetonitrile or a mixture of hexane and a polar solvent
such as acetone, respectively. Where the soil is contaminated with VOCs such as
BTEX, gasoline, TCE, PCE, VC and metal hydrides, a headspace technique is
used for direct injection into a GC. Following analyte extraction from soil,
extracts are subject to several other manipulations including concentration and
chromatographic separation prior to detection. As mentioned earlier, for envi-
ronmental consideration the trend in soil analysis is now focusing increasingly
on solventless sample preparation techniques as shown in Fig. 26.1.
In the following sections we will describe specific examples of analyte extrac-
tion from soil and other related biological matrices followed by proper clean-up
procedures prior to analyte detection with a suitable detector for identification
and quantification.

26.3.1 Analysis of petroleum hydrocarbons (PHCs)

Generally, more attention has been given to PHCs than any other family of
compounds due to their vast abundance and wide spread use. Over 2 billion
metric tons of petroleum products are used annually in the chemical, petrochem-
icals and transportation industries, resulting in severe accidental spills and
contamination in the terrestrial and marine environments [35]. Due to the
toxicity and carcinogenicity of some of these products, particularly PAHs, consi-
derable effort and capital are allocated to monitor their fate in the environment
[36]. For years the Environmental Protection Agency (EPA) Methods 418.1,
413.2 [37], 3560/8440 and 9071 [4] have been widely used to analyze PHCs in
soils, but recently these methods have been found to be inappropriate for certain
types of soil [38]. For example, discrepancies as high as 300% are obtained when
PHCs are analyzed in sediments [39]. Douglas et al. [35] recently reviewed the
limitations associated with some of the most prominent Standard Methods in
this field. The standardized EPA methods for the analysis of petroleum hydro-
carbons are derived from the American Public Health Association Method

899
TABLE 26.2
Soil sample preparation methods
Extraction type Analytes Extraction solvents Characteristics/Limitation
Soxhlet extraction Nonvolatile and Methylene chloride, EPA certified # 3540/
[4]; automated semivolatile organic
hexane, cyclohexane, 3541. Large volumes of
Soxhlet extraction compounds (TPHs, acetone/hexane (1:1, organic solvents (500 ml)
[4] PCBs, PAHs) v/v), toluene/methanol and long extraction times
(10:1, v/v) are needed
Pressurized fluid Water insoluble Acetone/hexane (1:1) EPA certified # 3545. Uses
extraction (PFE) semivolatile organics for organochlorine less solvent and less time
[4,8,9] (organophosphorus pesticides or PCBs. than Soxhlet.
and organochlorine Methylene chloride for
pesticides, chlorin- organophosphorus
ated herbicides, and pesticides. Acetone/
PCBs). methylene chloride/
phosphoric acid for
chlorinated herbicides.
Ultrasonic Nonvolatile and Similar to PFE method EPA certified # 3550/
extraction [4] semivolatile organic without organophos- 8330. Unsuitable for low
compounds (not phorus compounds. concentrations. Normally
validated for Acetonitrile for rapid but takes 16 h for
organophosphorus explosives. explosives.
compounds [10]).
Supercritical fluid TPHs, PAHs. Supercritical carbon EPA certified # 3560/
extraction (SFE) Organochlorine and dioxide (SC-CO2) for 3561. Can extract a wide
[4] organophosphate THPs [11] and PAHs. range of nonvolatile and
pesticides. Metals. SC-CO 2 with methanol semi-volatile organics and
Explosives [14]. co-solvent for pesticides metals. Not suitable for
[12] and metals [13]. volatiles such as gasoline.
SC-CO 2 with acetonitrile Co-solvent enhances
co-solvent for explosives. recoveries.
Equilibrium Volatile organic Solvent-free. EPA certified # 5021.
headspace [4,15] compounds (VOCs; Suitable for the extraction
BTEX, VC, TCE, of a wide range of volatile
PCE) organics with bp lower
than 180°C.
Vacuum Volatile organic Solvent-free. EPA certified # 5032.
distillation [4,16] compounds (VOCs)
Headspace SPME VOCs, HAPs, Solvent-free. Suitable for automatic
[17] chlorinated on-line SPME/GC and
benzenes, heavy SPME-ICP-MS analysis.
metals [18]
Purge-and-trap Volatile organic Solvent-free. EPA certified # 5035.
[4] compounds (VOCs) Suitable for on-line
injection and analysis.
Thermal Semi-volatile Solvent free. EPA certified # 8275.
extraction (TE) organics. Suitable for on-line
[4] injection and analysis.
Acid digestion [4]; Total recoverable Strong acids. EPA certified # 3050/
microwave metals. 3051. Corrosive.
assisted acid
digestion

900
(APHA [40]), originally developed to analyze mineral oil and grease in water and
sludge. According to the APHA method, any extractable chemical with a C-H
bond that absorbs in the IR region 2700-3200 cm 1 is measured as mineral oil
and grease.
In general, current understanding of the environmental fate of PHCs is
limited due to the lack of uniform analytical methods for the identification and
quantitation of complex mixtures of PHCs, particularly in soil. Sorption of PHCs
onto soil particles and the co-extraction of other organic material from soil are
some of the major problems encountered during soil analysis for PHCs. In
addition petroleum consists of a complex mixture of products ranging from
straight chain and branched aliphatics (paraffins) to simple and complicated
PAHs, necessitating a compound class separation prior to analysis. A successful
analytical protocol for the determination of PHCs in soil must thus take the
above limitations into consideration. Recently, there has been a genuine interest
from several researchers and agencies to improve and optimize standard
methods that have been developed earlier for the analysis of PHCs in soil [35,41].
At present, more specific techniques based on fingerprinting are considered for
the determination of TPHs in several soil matrices for the identification of the
source of contamination. Source determination is important for companies
facing liability claims [42]. Fingerprinting tools are based on extensive and
elaborate chromatographic and mass spectrometric techniques.
Extraction of semi-volatile petroleum hydrocarbons with an organic solvent
In a typical example, a soil sample from a refinery site with at least 10-15 years'
history of contamination with bunker crude oil is sieved (10 mesh), mixed with
MgSO4 and homogenized. Hexadecane and p-terphenyl are used as recovery
standards for aliphatic hydrocarbons and PAHs, respectively. The sample is
extracted as described in EPA Method 3450 [4] with an organic solvent such as
methylene chloride (Freon was used earlier but banned following Montreal
protocol in 1995 [43]) in Soxhlet for a total of 80 cycles. In some cases the Soxhlet
method was replaced with sonication using a suitable solvent such as hexane (30
min). Final extracts are dried over sodium sulfate, concentrated and eluted
through a silica column using solvents with different polarities, starting with the
least polar (hexane) and ending with the most polar one (methanol). Hydropho-
bic aliphatic compounds are first eluted with hexane whereas PAHs are eluted
with carbon tetrachloride or hexane mixed with methylene chloride (4:1, v/v).
Purified extracts are then analyzed for PHCs with IR spectrometry at 2700-
3200 cm- ' (vH, 2932 cm'), GC/MS or GC/FID. Figure 26.2 represents a typical
GC/MS chromatogram of various fractions collected during the analysis of TPH
in soil. Alternatively PHCs extracts can be quantified gravimetrically after
completely removing the solvent. A mass balance can then be drawn between
crude soil extracts and the collected fractions.
Solventless extraction of PHCs from soil
To reduce the amount of solvents and the time required for analysis, researchers

901
 Aliphatic

L
Polar

Time (min)

Fig. 26.2. Typical GC/MS chromatogram of various fractions collected during silica clean-up of
Soxhlet extracts of petroleum contaminated soil.

have recently focused their attention on solventless sample preparation methods

such as supercritical fluid extraction (SFE) [4,12]. The theory and application of
SFE has been extensively reviewed by Chester et al. [44]. SC-CO2 normally
requires less than one-tenth the time required by traditional techniques such as
Soxhlet and sonication. Carbon dioxide is nontoxic, the supercritical fluid has
low viscosity and negligible surface tension; CO 2 can thus diffuse through the
matrix as a gas and extract the analyte as a liquid. However, in contrast to the
Soxhlet method, extractability of analytes by SC-CO2 depends on the type of SFE
extractor and on the type of matrix used in the extraction cell. SFE often
requires extensive optimization of several instrumental (extractor type,
pressure, flow rate, restrictor) and experimental (soil type, compaction and
modifiers) [45-49] parameters. Consequently, estimates on the actual SFE
concentrations of chemicals in soil are often established through interlaboratory
studies and by comparing SFE recoveries with those obtained by other standard
techniques such as Soxhlet. In an interlaboratory study using several commer-
cial SFE extractors, EPA released an SFE/IR method for the determination of
PHCs in soil (EPA 3560 [4]). However, when SC-CO2 (EPA 3560 [4]) was
compared with Soxhlet (APHA 5520 [40]), large variations in the concentration
of PHCs between the two techniques were found. It has been reported that the
century old laborious Soxhlet seems to produce more realistic and reproducible
data than the SFE on the actual concentration of PHCs in aged soil [45].
In a typical SFE soil extraction experiment, a contaminated soil sample (3 g)
is mixed with MgSO4 (1:1 w/w) and extracted with SC-CO 2 (90°C, 340 atm and
350-400 ml gas/min) for 60 min and analytes are collected in vials designed
specifically to contain a solvent such as methylene dichloride. The CO2 gas is
allowed to pass through a restrictor during collection. The collected extracts are

902
 4+5 A
1 31

.1
r"@+1-
i
',Q'S1N
I6 ,,,, ,-

I lr-n~~n nn M~~~~lirI1

4+5
36

___4'J ,I ..; 1 - ,

4+5 C

6
3
I

Time (min.)

Fig. 26.3. A headspace GC/FID chromatogram of BTEX in a contaminated soil. (A) BTEX
standard: 1, benzene; 2, toluene; 3, ethylbenzene; 4+5, m,p-xylenes; 6, o-xylene). (B) Gasoline
standard. (C) contaminated soil sample.

either cleaned through a silica column or analyzed directly with a suitable

detector. For a more comprehensive overview of the SFE extraction and its
application, the reader is advised to consult Refs. [50,51].

Direct analysis of BTEX and gasoline in soil

For the analysis of volatile organic compounds (VOCs) such as BTEX and
gasoline, the soil can be analyzed directly using either headspace (soil) or
purge-and-trap technique (aqueous phase) for direct sample injection into a GC
fitted with a suitable detector normally FID or MSD (Fig. 26.3). The extensive
demand has led to the development of standard techniques for rapid screening of
VOCs in soil [4]. Voice and Kolb [52] recently reported that the static headspace
technique for the analysis of VOCs in soil is superior to the dynamic purge-
and-trap technique recommended by EPA [4].

26.3.2 Analysis of chlorinated compounds PCBs, TCE, PCE

Polychlorobiphenyls (PCBs) and their mixtures are commercially known as

Aroclors. They are produced industrially by chlorinating biphenyl using iron as a

903
 29.6 min.
A

Time (min.) 292 B

710

74 110
92 150

50 2 1 I 1 128 14 _ 4
' i! ' i I j / 1194 '148
2Fs 28 .13 327340
20' 40 60 80 io0 ii20 40 0 ibo0 0o
i 22o 240 20 20O 360 320 340
m/z
Fig. 26.4. A GCJMS Selective Ion Monitoring (SIM) chromatogram of: (A) soil extract contam-
inated with Aroclor 1248; and (B) the mass spectrum of 2,3',5,5'-tetrachloro-1,1'-biphenyl
congener at 29.6 min.

catalyst. This produces a mixture of Aroclors with Cl substitution ranging from

1-10 Cl/biphenyl molecule giving a total of 209 PCB congeners [34]. For
example, in Aroclor 1248 (the first two digits signify the 12 Cs in biphenyl and
the last two digits represent the wt% of Cl in the Aroclor) there are roughly 70
chlorinated biphenyls with an average C1 content of about 4 Cl/biphenyl. This
complexity is indicated in the gas chromatographic pattern shown in Fig. 26.4.
The unique thermal and chemical stability that makes PCBs industrially useful
has also made them a threat to the environment, because they are toxic and
suspected carcinogens. They are also resistant toward solar photodegradation
and microbial biodegradation and therefore tend to persist indefinitely. For
understanding the fate and impact of PCBs in soil we must first identify and
quantify their presence in the affected soil. The complex distribution of various
PCBs congeners and their interactions with soil necessitates an elaborate
sample preparation protocol for their analysis in soil. Briefly, a sample of the
contaminated soil is cleaned from rough debris and small rocks and homoge-
nized. The soil is spiked with 4-bromobiphenyl as a recovery standard and then
extracted by hexane/acetone (1:1, v/v) using ultrasonic extraction (60 Hz) for 30

904
 2 3

Time (min.)

Fig. 26.5. Headspace GC/FID chromatograms of: (A) VOCs standard; 1, chloroethane; 2,
1,1-dichloroethene; 3, 1,1-dichloroethane; 4, 1,2-dichloroethane; 5, 1,1,1-trichloroethane); and
(B) VOCs obtained from the headspace of a contaminated soil.

min. The extracts are treated with acid and copper and fractionated on an
Alumina column using methylene chloride as eluent. The collected methylene
chloride fractions are concentrated and analyzed with a GC/ECD or GC/MSD.
In cases of volatile chlorinated solvents such as PCE, TCE and VC, the soil
sample can be analyzed directly with a suitable GC using a headspace technique
as described earlier with BTEX (Fig. 26.5).

26.3.3 Analysis of explosives in soil

Of the nearly 20 different families of used highly energetic chemicals possessed

by the military today RDX, HMX and TNT represent the most powerful and
commonly used. General contamination of soil and water by RDX, HMX and
TNT is widespread and often caused by various military activities (manufactur-
ing, testing and training, demilitarization and open burning/open detonation-
(OB/OD)) [53-55]. It has been estimated that during the manufacture of RDX,
up to 12 mg/l may be discharged to the environment in process waste waters [56]
while a single TNT manufacturing plant can generate over 1.8 megalitres of
wastewater per day [57]. In the past, energetic chemicals received much less
attention in the scientific community than petroleum and other industrial
chemical pollutants although more than 12,000 sites are known to be contami-
nated with several explosives since military activities started in World War I
[58]. In general, data on the extent of contamination by energetic chemicals is
scarce. In the following section we will describe some of the recent methods
developed for the determination of these chemicals in soil environments.

905
 At present, explosives are extracted from soil by sonication with acetonitrile
by EPA Method # 8330 [4]. Briefly, soil samples are air-dried in a fume hood in
the dark to a constant weight at room temperature then saturated with acetone
to achieve homogeneous distribution of the contaminant in the soil. The soil is
ground with a mortar, then thoroughly sonicated with acetonitrile for 18 h for
subsequent analysis by HPLC/UV (X54,m) Detection limits for TNT, RDX and
HMX are < 1 mg/kg. Alternatively, the soil can be directly extracted by shaking
the soil sample in acetone for 30 min. However, acetone is a powerful polar
solvent that can extract several other polar organic compounds from soil in
addition to the explosive thus causing interference with analysis. Also, acetone
absorbs at 2 54 nm, which is the wavelength used in EPA Method # 8330 [4] to
detect and quantify explosives. As discussed below, it is possible to minimize
interference caused by co-extracted chemicals from soil using supercritical fluid
extraction with carbon dioxide (SC-CO,). Furthermore, techniques that are
more specific than IR such as GC-MS and LC/MS can be used for explosives
detection and quantification.
The recent discovery of solid phase microextraction (SPME) [17,59] has
reduced the analysis time required for analysis and lowered the detection limit of
analytes. SPME is a solventless and rapid sample preparation technique that
uses a polymer-coated fiber for the adsorption of organic compounds from an
aqueous solution or the headspace of a soil sample followed by direct thermal
desorption into a GC or HPLC for subsequent detection and quantification. The
technique is known for its sensitivity, which enables detection in the ug 1 ' range.
Further details on the theory, applications and limitations of the SPME tech-
nique are available elsewhere [17,59]. Table 26.3 shows the concentration of
TNT and its derivatives as determined by SPME/GC-MS in a wet soil sample
obtained from a ditch near a TNT manufacturing plant. The analysis was done
by inserting a polydimethylsiloxane coated fiber in the water phase followed by
desorption inside a GC injector connected to a MS detector. Table 26.3 shows
TNT and several of its derivatives including 2-NT (2-nitrotoluene), 3-NT (3-
nitrotoluene), 4-NT (4-nitrotoluene), 2,6-DNT (2,6-dinitrotoluene), 2,5-DNT
(2,5-dinitrotoluene), 2,3-DNT (2,3-dinitrotoluene), 2,4-DNT (2,4-dinitrotolu-
ene), 3,5-DNT (3,5-dinitrotoluene), 3,4-DNT (3,4-dinitrotoluene), TNT (2,4,6-
trinitrotoluene), 4-ADNT (4-aminodinitrotoluene) and 2-ADNT (2-aminodi-
nitrotoluene). Table 26.3 shows the concentration of these products as found by
the standard HPLC based EPA method 8330 and by SPME/GC-MS [4]. In gen-
eral, a correlation factor of 90-100% was obtained between the two methods,
although the limit of detection (DL) by both techniques was found to be 10 ppb
and 0.5 ppm, respectively.

26.3.4 Agrochemicals

Pesticides including insecticides and herbicides are used worldwide in agricul-

ture and other domestic and industrial activities; the most widely used include

906
TABLE 26.3
Determination of TNT and its derivatives (mg/kg) in water taken from a contaminated area
nearby a ditch at a TNT manufacturing plant: SPME/GC-MS versus EPA Method 8330
Compounds SPME/GC-MS (RSD) HPLC/UV (RSD)*
2-Nitrotoluene 3.78 (2.26) 2.0 (1.50)
3-Nitrotoluene 1.11 (2.74) nd
4-Nitrotoluene 0.29 (2.10) nd
2,6-Dinitrotoluene 97.27 (5.40) 100.0 (0.21)
2,5-Dinitrotoluene 8.60 (5.85) nd
2,3-Dinitrotoluene a 12.6 (0.17)
2,4-Dinitrotoluene 100.49 (4.93) 80.4 (0.03)
3,5-Dinitrotoluene 2.39 (4.87) nd
3,4-Dinitrotoluene 35.1 (4.56) b
2,4,6-Trinitrotoluene 50.16 (3.25) 60.5 (0.09)
TNT-isomer 1.06 (7.47) nd
TNT-isomer 0.10 (6.18) nd
TNT-isomer 0.44 (18.9) nd
4-Amino 2,6-dinitrotoluene 0.34 (6.85) nd
2-Amino 4,6-dinitrotoluene 0.58 (10.1) nd
*Detected at 254 nm, RSD was calculated based on triplicate analysis. nd: Not detected.
a: 2,3-DNT overlapped with 2,4-DNT; b: 3,4-DNT overlapped with 2,6-DNT.

triazines, chloroacetamides, bipyridilium salts, chlorophenoxy acids, nitro-

anilines and organophosphorus and organochlorine compounds. Their wide-
spread use has led to the contamination of vast volumes of soil. The long-term
persistence of pesticides in the terrestrial ecosystem and their long-range trans-
port have led to their classification as priority pollutants in both Europe and the
USA and they are therefore placed under strict monitoring program.
Most of the currently used pesticides are polar and contain very reactive
functional groups such as hydroxyl and carboxyl groups, thus leading to strong
interactions with soil humic and fulvic acids. Such strong interactions can lead
to irreversible binding and the formation of covalently bound complexes with
soil (bound residues) whose recovery from soil by conventional methods is
extremely difficult. Therefore, the release of this fraction from soil may require
the application of high temperature processes such as pyrolysis or high-temper-
ature distillation, but these chemicals are thermally labile and often decompose
on heating. Alternatively, mild SFE extraction with either supercritical MeOH
(150 bar, 250°C), water (120°C, 200 bars) or CO 2 is used [60-64]. In the latter
case, a polar modifier such as MeOH, water [11], ethyl amine [65] and trifluoro-
acetic acid [66] are added to disrupt the strong interactions between polar
pesticides and soil organic materials. However, certain organophosphorous
pesticides (OPPs) such as dimethon-S and dimethon-O are best extracted with
neat CO 2 without the assistance of modifiers [67]. On the other hand, ionic
herbicides such as phenoxy acids are first converted (e.g., in situ derivatization

907
with tetrabutylammonium hydroxide or methyl iodide [671) to less polar and
more soluble forms in CO2 .
Alternatively bound pesticide residues may be determined by immunochemi-
cal methods such ELISA that can recognize the pesticide directly bound to humic
acid materials without bond cleavage [68]. Other extraction methods such as
MAE and PFE have been developed to minimize the use of solvents or to reduce
the extraction time [69].

26.3.5 Analysis of metals in soils and sediments.

Soil contamination with metals, particularly heavy metals is widespread and

caused by various anthropogenic activities including mining, printing, food
packaging and electrochemical industry. Most metals are toxic and carcinogenic
and furthermore can easily undergo several biogeochemical transport and
(bio)transformation processes. As an example, metallic tin is almost non-toxic
but its methylated or butylated organic derivatives are very toxic at low
concentrations (ppt). Furthermore, its tetraalkylated derivative are volatile and
easily released to the atmosphere. Likewise, metallic mercury is almost
non-toxic and does not bioaccumulate, but its methyl derivative is extremely
toxic and can be easily transported through the food chain.
Another important aspect of soil contamination by heavy metals is their
mobility and transport through surface and subsurface soil leading to vast vol-
umes of contaminated soil and water. Therefore, in addition to the determina-
tion of the total soil content of heavy metals we must determine the distribution
of the chemical species (speciation) in soil in order to evaluate their risk.
Traditionally, acid digestion is either carried out in a heated oven bomb or in
a microwave oven to dissolve the metals from soil, according to standardized
EPA Methods [4]. Acid digestion is not suitable for metals that bind strongly
onto the silicate part of the soil matrix. Such cases require a two-step acid
digestion method comprising, firstly, digesting with a mixture of HNO3 and HF
acids in a closed microwave oven and, secondly, digestion with a mixture of
HCiO4 and H2 SO4 [70].
SFE has been introduced as a "green" extraction technique because it does
2+
not use harmful solvents or strong acids. Transition metal ions such as Cu ,
2+ 2+ 2+
Co , Cd and Zn have been extracted from sand with SFE in the presence of
suitable complexing agents that are mixed with the matrix prior to extraction
[13]. The complexion or chelation of metal ions neutralizes the charge of the ions
and enhances their solubility in the supercritical fluid. This in situ chelation-
3+
SFE technique can also be used to extract trivalent lanthanide ions (La3+, Eu ,
3+
and Lu ) from sand with a fluorinated P-diketone as chelating agent [71]. As
discussed above, the trend in environmental analysis is to focus more on sol-
vent-free sample preparation techniques and to identify the different chemical
species occurring in the environment. Bayona and Cai [72] reported strategies
for the determination of organometallic compounds, mainly alkylated tin and

908
their degradation products in aqueous and solid matrices using in situ deriva-
tization or acid-modified SC-CO2 . For a wider application of the SFE technique
for the extraction and determination of heavy metals, the reader can refer to
another review by Chester et al. [44]. With the ability to extract a wide range of
heavy metals, this technique is important for the study of nuclear solid waste
remediation and other soils heavily contaminated with metals.
Extraction of metals from soil can be accomplished using the previously
described SPME. Mester et al. [74] reported a coupled SPME-ICP/MS technique
for the determination of volatile metal hydrides such as Se, Sn, As and Sb
(generated in situ) or methylated species in case of tin and mercury [75]. For a
comprehensive review of some of the most recent sample preparation techniques
used for the analysis of metals in soils, sediments and plants including SPME,
the reader is referred to two reviews by Mester et al. [74] and Sturgeon [76].
Finally, some workers have adopted a sequential extraction technique to
enhance the recovery of metals. F6rstner [77] reported sequential application of
several steps in the determination of metals distribution in soil according to
their matrix interaction. These steps include a cation exchange with 1 M
ammonium acetate followed by extraction with a reducible phase with 0.1 M
hydroxyl amine-HCl at pH = 2, and extraction of oxidizable phases with 30%
hydroxy minerals.

26.3.6 Analysis of pollutants in plant tissues grown in contaminated

soil: phytoextraction

Phytoremediation is receiving considerable attention as an alternative in situ

remediation technology for contaminated soil environments and shallow aqui-
fers [78]. Methods are described for the phytoextraction of radioactive waste
[79], heavy metals [80], petroleum hydrocarbons (PHCs) [81], PAHs [82], TCE
[83], TNT [84-86] and RDX [87,88], with an on-going development of analytical
protocols for the application and assessment of this relatively new technology
[88,89].
Several plant species including wheat (Triticum aestivum) and perennial rye
grass (Lolium perenne) have been effectively used on HMX contaminated soils
[90]. Extracts from plant shoots grown on contaminated soil are prepared and
analyzed using the methods outlined by Larson et al. [91]. Briefly, finely cut
samples of the rye grass (approximately 4 g) are suspended in 10-20 ml of
ice-cold deionized water and homogenized for subsequent lyophilization prior to
sonication with acetonitrile. The sonicated mixture is then centrifuged and the
supernatant is decanted, mixed with deionized water and filtered using Millex
HV 0.45 jum. Separations to detect explosives or their metabolites in plant
extracts are performed using HPLC/UV or a capillary electrophoresis (CE)
coupled with a suitable detector [90]. The sufficiently irrigated wheat and rye
grass cultivars could accumulate HMX in their senescent leaf tissue to a level of
more than 500 mg/kg (plant dry weight basis). The average HMX concentration

909
is 30 mg/kg (dry weight basis) [90]. LC/MS analyses confirm the absence of any
HMX degradation in the plant tissue, emphasizing that the explosive HMX is
bioextracted rather than biodegraded.
The current sonication method, which involves sonication of plant tissue
with an organic solvent in a refrigerated bath for 16 h, may not be suitable for
the determination of short-lived intermediate products. As described above, to
reduce the amount of solvents and time required for analysis, researchers have
recently focused their attention on solventless sample preparation methods such
as SFE [92]. Furthermore, the low supercritical temperature of CO2 also makes
it possible to solubilize thermally labile non-volatile chemicals such as the
explosive HMX, allowing their extraction without thermal decomposition. How-
ever, the non-polar character of CO2 limits its affinity for the extraction of polar
molecules such as polynitroorganic explosives, but this difficulty can be over-
come by the addition of a suitable co-solvent or by choosing an alternate
supercritical fluid for extraction.
Plant tissue samples from either the leaves, stem or roots are freeze-dried
and mixed with sodium sulfate or inert sand (particle diameters less than 0.5
mm) to sufficiently fill the extraction cell (5 ml) for the SFE extractor. After
passing through the extraction cell, SC-CO2 is allowed to pass through a
restrictor wherein SC-CO2 is depressurized to allow CO 2 gas to escape leaving
behind the analytes in the collection vial. The extract is then analyzed for
explosives content with either HPLC/UV (254 nm) or GC connected with ECD or
MSD. Both sonication with acetonitrile and SC-CO 2 were found to be compatible
in their extraction efficiency of HMX from the plant tissue, although at higher
concentration (> 300 mg/kg), sonication is found to be slightly better [93].
Previously we demonstrated that SC-CO2 recovery of TNT from soil [14] is
compatible with that of EPA Method # 8330 [4] which requires sonication with
acetonitrile (Table 26.3). The extractability of explosives from soil by SC-CO 2
can be improved by adding a suitable co-solvent either directly to the sample
(static solvent addition) or to CO2 (dynamic addition) prior to extraction [14].
For instance, the relatively poor SC-CO2 recoveries of RDX, HMX and TNT (26,
0 and 70%, respectively) obtained from a contaminated dry sandy soil (100
mg/kg) are improved drastically (90, 70 and 87%, respectively) when the soil is
first treated with 5-15% v/v of acetonitrile [14].
Interestingly, the SC-CO2 recoveries of TNT amine derivatives 2-ADNT and
4-ADNT are found to be lower than that of TNT itself (Table 26.4). The
extracted amounts correlate with their Freundlich soil sorption isotherms [14].
Amino-TNT derivatives, frequently found with TNT as degradation products,
with higher water-soil partition coefficients, Kd, are poorly extracted [14].
Amines are basic compounds and are thus expected to interact strongly with soil
humic acids to produce irreversibly bound complexes. The efficiency of SC-CO2
extractability of energetic chemicals from soil is also found to depend on several
other instrumental parameters such as pressure, temperature and time of
extraction. In general, any increase in pressure or temperature that does not

910
TABLE 26.4
Extraction of explosives and related compounds from several soil matrices using either SC-CO 2
extraction or sonication with acetonitrile (EPA # 8330)
Soil source Analyte Method of extraction (mg/kg) %SC-CO,/EPA
SC-CO2* (RSD) EPA** (RSD)
Sand RDX 75.0 (5.4) 76.6 (4.0) 97.9
HMX 57.4 (6.5) 81.3 (3.3) 70.6
Burning pit RDX 3310 (7.5) 3620 (3.8) 91.4
HMX 1150 (6.9) 1710 (4.0) 67.3
Top soil TNT 205.3 (5.1) 220.5 (13.4) 93.1
Disposal site TNT 5841 (4.5) 5265 (5.2) 110
Fungi treated TNT 0.75 0.78 96.1
Contaminated soil 2-ADNT 3.31 6.33 52.3
4-ADNT 2.92 5.04 57.9
*Samples (2 g) were each mixed with acetonitrile (AcN) (10% v/w) before extraction with SC-CO 2
(300 atm, 280 ml/min gas, oven and restrictor temperatures were 60°C and 150°C, respectively).
MeOH (5% v/v) was used as a modifier in the extraction of the sample from the burning pit.
Extraction time was 30 min. Analytes were collected in acetonitrile (10 ml).
**Samples (2 g) were each extracted with acetonitrile (5 ml) using sonication for 16 h. Extracts
from both methods were analyzed by HPLC (254 nm).

cause the explosive to decompose, leads to an increase in the amounts of

explosive extracted from soil [14].

26.3.7 Monitoring the fate of chemicals in related biological systems

Biodegradationof RDX in soil

Soil (bio)remediation is a multi-million dollar growth industry, but the contro-
versy surrounding the use of these technologies is also growing partly due the
absence of a valid analytical protocol capable of assessing the fate of the
chemical. Contaminant removal (or disappearance) during remediation does not
necessarily mean safe removal: the chemical might have been converted into
another more toxic chemical [55]. One major problem in providing a reasonable
assessment on the extent of soil remediation is the compositional differences
amongst the samples obtained at different stages of remediation. Monitoring the
disappearance of a targeted chemical and the appearance and subsequent trans-
formation of its degradation products can provide insight into the fate of the
chemical and the effectiveness of remediation.
In a typical example, a contaminated soil sample is first analyzed for explo-
sives using sonication with acetonitrile as described by EPA method # 8330 [4].
The contaminated soil is then treated with a domestic anaerobic sludge used as
an exogenous source of microorganisms. Some serum bottles (microcosms) are
supplemented with a uniformly labeled [UL-14 C] RDX (100,000 dpm) and then

911
fitted with a small test tube containing KOH (0.5 M) to trap liberated carbon
dioxide ( 4 CO,). Uniformly ring labeled [UL- 1 N] RDX is used to confirm the
identity of RDX breakdown products such as N2 , N2 0O,and other ring cleavage
products. Full details on biodegradation experiments can be found in Refs.
[94,95]. Degradation products in the soil-slurry mixture are either directly
analyzed with an SPME/GC-MS system or with a LC-MS. Figure 26.6 is a typical
LC/MS chromatogram of products detected during RDX biodegradation with
sludge. The chromatogram shows the three nitroso derivatives of RDX as
determined by their deprotonated molecular mass ions [M-H]. These ions are
detected at m/z 206, 190 and 174 Da, matching molecular mass formulas of
C,HNO,,
3 CH 6N6,04 and CH 6N0,,, respectively. The peaks are identified as the
mononitroso hexahydro-l-nitroso-3,5-dinitro-1,3,5-triazine (MNX), hexa-
hydro-1,3-dinitroso-5-nitro-1,3,5-triazine (DNX) and hexahydro-1,3,5-tri-
nitroso-1,3,5-triazine (TNX). Figure 26.6 also shows a second set of LC-MS
peaks at the beginning of the chromatogram with lower retention times (<4
min) with [M-H] appearing at m/z 135 Da. Using reference standards and
uniformly ring labeled [N]-RDX, the peak is identified as the ring cleavage
products methylenedinitramine (O2NNHCH2NHNO2) [94]. The observation of
the latter metabolites confirm the occurrence of ring cleavage which is a prereq-
uisite step prior to mineralization (liberation of CO,). A combined LC/MS and
SPME/GC-MS time course shows that none of the above RDX degradation
intermediate accumulates indefinitely. Instead, they all (bio)transform further
to eventually produce HCHO, CO 2 and nitrous oxide, N2 0. The formation of N2 0O
from RDX biodegradation is seemingly universal since both N2 0O(detected by
GC/ECD) and HCHO (detected by SPME/GC-MS) also can be generated during
biodegradation of both RDX under aerobic conditions using the fungus
Phanaerocheate chrysosporium [96]. Furthermore chemical reduction and
hydrolysis have been found to lead to N2 0 and HCHO. The universal transfor-
mation of the cyclic nitramine explosive to eventually produce nitrous oxide,

100

TNX

[M-H]:135

2.5
1~~ V
5.0 75
I
10.0
,MNX
~~~~I.
12.5
RDX
15.0 17.5 20.0
Retention time (min)

Fig. 26.6. Typical LC/MS chromatogram of products detected during RDX biodegradation with
sludge in a soil. The figure shows the initial metabolites MNX, DNX and TNX and the ring
cleavage product methylenedinitramine, which ultimately decomposes in water to produce
HCHO and N2 O.

912
HCHO and CO 2 under abiotic and biotic conditions could be exploited for the
development of chemical or biochemical sensor for on-site monitoring of cyclic
nitramine explosives in the environment.

Biodesulfurization of dibenzothiophene (DBT)

Perhaps one of the most useful applications of analytical chemistry is the
provision of insight into biotransformation pathways in environmental and
biological matrices. Until recently, lengthy sample preparation and separation
techniques (e.g. liquid-liquid extraction followed by chromatographic clean-up
procedures) were necessary to isolate and identify intermediates produced dur-
ing biotransformation processes [97,98]. When such intermediates are formed in
trace amounts, traditional techniques are not practical or fast enough for their
detection, thus leading to the loss of valuable information regarding the trans-
formation pathways. Recently, SPME in combination with GC/MS has been
successfully used to identify metabolites formed during desulfurization by
Rhodococcus sp. strain ECRD-1, of model thiophenic compounds commonly
found in fossil fuel, i.e., DBT [99,100]. These model bitumen compounds are
selected because reserves of this fossil fuel are extremely large, but its fuel value
is low due in part to the high organic sulfur content, which upon combustion can
release sulfur dioxide into the atmosphere causing acid rain. To increase the fuel
value without causing harm to the environment, the crude oil must be desulfur-
ized without an excessive reduction of its calorific value. Several studies have
described products that are generated from model bitumen compounds using
different microorganisms under both aerobic and anaerobic conditions. For
example, through extensive GC/FTIR/MS analysis, Olson et al. [101] reported
the formation of key metabolites including dibenz[c,e][1,2]oxathiin-6-oxide
(sultine) and dibenz[c,e][1,2]oxathiin-6,6-dioxide (sultone) during the desulfur-
ization of DBT by Rhodococcus sp. strain IGTS8. Recently the degradation
pathway of the biodesulfurization of DBT by Rhodococcus sp. ATCC 55309
(strain 309) has been determined using an SPME/GC-MS [100, 102]. A time
course for the formation and disappearance of DBT transient metabolites show-
ed that biodesulfurization occurred through a metabolic pathway that involved
the sequential formation of DBT-5-oxide (sulfoxide), DBT-3,5-dioxide (sulfone),
dibenz[c,e] [1,2]oxathiin-6,6-dioxide (sultone), dibenz[c,e] [1,2]oxathiin-6-oxide
(sultine), and finally the S-free end product 2-hydroxybiphenyl [100].

Biodegradation of biphenyl generatedfrom photoreductive dechlorinationof

PCBs
A combined solid phase microextraction/GC-MS analytical technique has been
used to monitor the formation of metabolites in the biodegradation of biphenyl
(BP), which was originally obtained from the solar photodechlorination of
Aroclor 1254 in the presence of a photosensitizer, by Pseudomonas pseudo-
alcaligenes [103]. The metabolites that were detected include 2-hydroxy-
biphenyl (2-OH-BP), 2,3-dihydroxybiphenyl (2,3-di-OH-BP), and benzoic acid

913
(detected as its benzoate derivative 1-methylethylbenzoate). A time course study
for the formation and disappearance of these metabolites is used to construct a
degradation pathway which involves the formation of 2-OH-BP and 2,3-di-
OH-BP that later undergo ring cleavage, ultimately resulting in mineralization
(CO, formation) [104].

Acknowledgements

The author would like to thank the Department of National Defence, Canada for
their partial support on conducting the characterization part on energetic
chemicals. We thank Mme. C. Beaulieu and Mr. S. Deschamps for providing the
figures on TPHs and VOCs. We would like to thank Dr. An-Lac Nguyen for
revising the manuscript. Finally, we would like to thank the Strategic Environ-
mental Research and Development Program (SERDP # CU1213), USA, for
partial funding of the present work on RDX.

REFERENCES

1 T.F. Jenkins, M.E. Walsh, P.G. Thorne, P.H. Miyares, T.A. Ranny, C.L. Grant and
J.R. Esparza, Site characterization for explosives contamination at a military firing
range impact area, Special Report 98-9. US Army Corps of Engineers, CRREL,
Hanover, NH, 1998.
2 S. Thiboutot, G. Ampleman, T.F. Jenkins, C.L. Grant, P.G. Thee, MF. Walsh, T.A.
Ranney, J. Esparza and M.H. Stutz, Proceedings of the Third Tri-Service Environ-
mental Technology Workshop, San-Diego, CA, 18-20 August, 1998.
3 J. Crepin, and R.L. Johnson, in: M.R. Carter (Ed.), Soil Sampling and Methods of
Analysis. Lewis Publishers, Boca Raton, FL, 1993, pp. 5.
4 EPA SW-846 update 3rd edn., Methods 3051, 3540, 3541, 3545, 3550, 3560, 3561,
5021, 5032, 5035, 8330, 8275, Office of Solid Waste, Washington, DC, 1997.
5 C. Carlon, A. Critto, A. Marcomini and P. Nathanail, Environ. Pollut., 111 (2001)
417.
6 J.C. Pennington, J.M. Brannon, D. Gunnison, D.W. Harrelson, M. Zakikhani, P.
Miyares, T.F. Jenkins, J. Clarke, C. Hayes, D. Ringleberg, E. Perkins and H.
Fredrickson, Soil Sediment Contain., 10 (2001) 45.
7 T.F. Jenkins, C.L. Grant, G.S. Brar, P.G. Thorne, P.W Schumacher,. and T.A
Ranney, FieldAnal. Chem. Technol., 1 (1997) 151.
8 B. Richter, J. Ezzell and D. Felix, Single Laboratory Method Validation Report.
Extraction of TCL/PPL (Target Compound List/Priory Pollutant List) BNAs and
Pesticides using Accelerated Solvent Extraction (ASE) with Analytical Validation by
GC/MS and GC/ECD. Document 116064.A, Dionex Corporation, 1994.
9 B. Richter, J. Ezzell and D. Felix, Single Laboratory Method Validation Report.
Extraction of TCL/PPL (Target Compound List/Priory Pollutant List) Chlorinated
herbicides and PCBs using Accelerated Solvent Extraction (ASE). Document
101124.A, Dionex Corporation, 1994.
10 C.S. Hein, P.J. Marsden and A.S. Shurtleff, Evaluation of Methods 3540 (Soxhlet)
and 3550 (Sonication) for Evaluation of Appendix IX from Solid Samples. S-CUBEC,
Report for EPA, Work Assignment No 03, Doc. No. SSS-R-88-9436, 1988.
11 J.L. Snyder, R.L. Grob, M.E. McNally and T.S. Oostdyk, Anal. Chem., 64 (1992)
1940.

914
12 S.B. Hawthorne and D.J. Miller, Anal. Chem., 66 (1994) 4005.
13 Y. Liu, V. Lopez-Avila, M. Alcazar, W.F. Beckert and E.M. Heithmar. J. Chromatogr.
Sci., 31 (1993) 310.
14 J. Hawari, A. Halasz, L. Dusseault, J. Kumita, E. Zhou, L. Paquet, G. Ampleman and
S. Thiboutot, Proceedings of the 5th Meeting On Supercritical Fluids, Nice, France,
1998. pp. 161.
15 P. Flores and T. Bellar, Determination of Volatile Organic Compounds in Soils using
Equilibrium Headspace Analysis and Capillary Column Gas Chromatography/Mass
Spectrometry. U.S. Environmental Protection Agency, Office of Research and Devel-
opment, Cincinnati, OH, 1992.
16 M.H. Hiatt, Anal. Chem., 53 (1981) 1541.
17 J. Pawliszyn, Solid Phase Microextraction: Theory and Practice. Wiley-VCH, New
York, NY. 1997.
18 J.R. Dean and P. Hancock, in: J. Pawliszyn (Ed.), Application of Solid Phase
Microextraction.The Royal Society of Chemistry, Cambrige, UK, 1999. pp. 248.
19 J.M. Brannon, P. Deliman, C. Ruiz, C. Price, M. Qasim, J.A. Gerald, C. Hayes and S.
Yost, Conceptual model and process descriptor formulations for fate and transport of
UXO, Technical Report IRRP-99-1, U.S. Army Corps of Engineers, Waterways
experimental station, Vicksburg, MS, 1999.
20 S.D. Comfort, P.J. Shea, L.S. Hundal, Z. Li, B.L. Woodbury, J.L. Martin and W.L.
Powers, J. Eviron. Qual., 24 (1995) 1174.
21 H.M. Selim, S.K. Xue and I.K. Iskandar, Soil Sci., 160 (1995) 328.
22 J. Singh, S.D. Comfort and P.J. Shea, J. Environ. Qual., 27 (1998) 1240.
23 T.W. Sheremata and J. Hawari, Environ. Sci. Technol., 34 (2000) 3462.
24 HSDB, TNT. Hazardous substances data bank, National library of medicine,
National toxicology information program, Bethesda, MD, 1994.
25 R.J. Spanggord, W.R. Mabey, T.W. Chou and J.H. Smidth, in: D.E. Rickert (Ed.),
Chemical Industry Institute of Toxicology Series: Toxicity of Nitroaromatic Com-
pounds. Hemisphere, Washington, DC, 1985, pp. 15.
26 D.L. Kaplan, in: S.K. Sikdar and R.L. Irvine (Eds.), Bioremediation:Principlesand
Practice, Vol. II, Biodegradation Technology Developments. Technomic, Lancaster-
Basel, 1998, pp. 549.
27 G.B. Teany and R.T. Hudak, Proceedings of the 87th Annual Air and Waste Manage-
ment Association, paper # 94-RP 143.05, Cincinnati, Ohio, 1994.
28 G.B. Teany, R.T. Hudak and J.M. Melby, Proceedings of Field Screening Methods for
Hazardous Wastes and Toxic Chemicals, Las-Vegas, Nevada, 1995, p. 965.
29 P.R. Cauger, D.B. Holt, C.H. Patterson Jr., P.T. Charles, L. Shriver-Lake and A.W.
Kusterbeck, J. Hazard. Materials, 83 (2001) 51.
30 K.F. Myers, E.F. McCormick, A.B. Strong, P.G. Thorne and T.F. Jenkins, Com-
parison of Commercial Colorimetric and Enzyme Immunoassay Field Screening
Methods for TNT in Soil, Technical Report IRRP-94-4, Waterways Experiment
Station and U.S. Army CREEL, 1994.
31 J.A. Sundquist and S.S. Sisodia, Proceedings of 28 th Mid-Atlantic Industrial and
Hazardous Waste Conference, Vol. 28, 1996, p. 93.
32 Z. Mester, R. Sturgeon and J. Pawlyszin, Spectrochim. Acta B, 56 (2001) 233.
33 F. Soxhlet, Dingers Polytech. J., 232 (1879) 461
34 M.D. Erikson, Analytical Chemistry of PCBs. Butterworth, Stoneham, MA, 1986.
35 G.S. Douglas, K.J. McCarty, D.T. Dahlen, J.A. Seavey, W.G. Steihauer, R.C Prince
and D.L. Elmendorf, in: P.T. Kostecki and E.J. Calabrese (Eds.), ContaminatedSoils:
Diesel Fuel Contamination. Lewis Publishers, Chelsa, MI, 1992.
36 M.L. Lee, M.V. Novotny and K.D. Bartle, Analytical Chemistry of Polycyclic Aromatic
Compounds. Academic Press, New York, 1981, pp. 17.

915
37 Test Methods for Chemical Analysis of Water and Wastes Environmental Monitoring
and Support Lab., Cincinnati, Ohio, PB-297 686, Method 418.1, 413.2, 1978.
38 J.H. Martin, Jr., A.J. Siebert and R.C. Loehr, J. Environ. Eng., 117 (1991) 291.
39 L.H. Hilpert, W.E. May, S.A. Wise, S.N. Chesler and H.S. Hertz, Anal Chem., 50
(1987) 458.
40 APHA, 19th edn. Standard Methods 5520, 1995.
41 G.S. Douglas and A.D. Uhler, Environ. Test. Anal., 2 (1993) 46.
42 G.S. Douglas and J.S. Brown, Advanced Chemical Fingerprinting of Petroleum
Contaminated Soils and Water. 10 th Annual Conference on Contaminated Soils,
October 23-26, 1995.
43 Montreal protocol on Substances that Deplete the Ozone Layer, Montreal Protocol
Unit (MPU), ESDG/UNDP, New York, NY., Sep., 1987.
44 T.L. Chester, J.D. Pinkston and D.E. Raynie, Anal. Chem., 68 (1996) 487R.
45 J.A. Hawari, C. Beaulieu, D. Ouellette, A. Halasz and H. Van Tra, Int. J. Environ.
Anal. Chem., 60 (1995) 123.
46 J.L. Snyder, R.L. Grob, M.E. McNally and T.S. Oostdyk, J. Chromatogr. Sci., 31
(1993) 183.
47 M.T. Fahmy, M.E. Paulaitis, D.M. Johnson and M.E. McNally, Anal. Chem., 65
(1993) 1462.
48 J.W. Hills and H.H. Hill, J. Chromatogr. Sci., 31 (1993) 6.
49 J.J. Langenfeld, S.B. Hawthorne, D.J. Miller and J. Pawliszyn, Anal. Chem., 66
(1994) 909.
50 S.B. Hawthorne, in: J. Pawliszyn (Ed.), Sample Preparation in Field and Laboratory,
Elsevier, Amsterdam, 2002 (this book and references therein).
51 G. King, in: J. Pawliszyn (Editor), Sample Preparation in Field and Laboratory,
Elsevier, Amsterdam, 2002 (this book and references therein).
52 T.C. Voice and B. Kolb, Environ. Sci. Technol., 27 (1993) 709.
53 T. Urbanski, in: M. Jurecki (Trans.) and S. Laverton (Ed.), Chemistry and Tech-
nology of Explosives. Pergamon, Oxford, 1967, Vol. III, pp. 17.
54 R. Haas, I. Schreiber, E. v. Low and G. Stork, FreseniusJ. Anal. Chem., 338 (1990)
41.
55 A. Myler and W. Sisk, in: G.S. Sayler, R. Fox and J.W. Blackburn (Eds.), Bio-
remediation of Explosives Contaminated Soils (Scientific Questions/Engineering
Realities), Environmental Biotechnology for Waste Treatment. Plenum, New York,
1991, pp. 137.
56 M. Jackson, J.M. Green, R.L. Hash, D.C. Lindsten and A.F. Tatyrek, Nitramine
(RDX and HMX) wastewater treatment at the Holston Army Ammunition Plant,
Report ARLCD-77013, US Army Armament Research and Development Command,
Dover, NJ, 1978.
57 J. Yinon, Forensic and Environmental Detection of Explosives. John Wiley, Chich-
ester, UK, 1999.
58 J. Akhavan, The Chemistry of Explosives. The Royal Society of Chemistry,
Cambridge, UK, 1998.
59 J. Pawliszyn, in: R.M. Smith (Ed.), RSC Chromatography Monographs Series.
Loughborough University of Technology, The Royal Society of Chemistry, UK, 1999.
60 S.S.Yang, A.I. Goldsmith and I. Smetena, J. Chromatogr. A, 754 (1996) 3.
61 A. Di Corcia, A.B. Caracciolo, C. Crescenzi, G. Giuliano, S. Murtas and R. Samperi,
Environ. Sci. Technol., 33 (1999) 3271.
62 J.R. Dean, I.J. Barnabas and S.P. Owen, Analyst, 121 (1996) 465.
63 A.M. Robertson and J.N. Lester, Environ. Sci. Technol., 28 (1994) 346.
64 S. Paillound and W. Haerdi, Chromatographia,38 (1994) 514.
65 R. Alzaga, J.M. Bayona and D. Barcel6, J. Agric. Food Chem., 43 (1995) 395.

916
66 R. Alzaga, J.M. Bayona and D. Barcel6, J. High Resol. Chromatogr., 19 (1996) 23.
67 V. Lopez-Avila, N.S. Dodhiwala and W.F. Beckert, J. Agric. Food Chem., 41 (1993)
2038.
68 M.G. Weller, A. Zeck, P. Pfortner, E. Simon and R. Niessner, Int. J. Environ. Anal.
Chem., 75 (1999) 201.
69 V. Camel, Analyst, 126 (2001) 1182.
70 L. Yang, J.W.H. Lam, R.E. Sturgeon and J.W. McLaren, J. Anal. Atom. Spectrom., 13
(1998) 1245.
71 Y. Lin, R.D. Brauer, K.E. Laintz and C.M. Wai, Anal. Chem., 65 (1993) 2549.
72 J.M. Bayona and Y. Cai. TrAc, Trends Anal. Chem., 13 (1994) 327.
73 J.M. Bayona, Tech. Instrum. Anal. Chem., 17 (1995) 465
74 Z. Mester, R. Stergeon and J.W. Lam, J. Anal. At. Spectrom., 15 (2000) 1461.
75 Y. Morcillo, Y. Cai and J.M. Bayona, J. High Resol. Chromatogr., 18 (1995) 767.
76 R. Sturgeon, Commun. Soil Sci. Plant Anal., 31 (2000) 1479.
77 U. Forstner, Contaminated Sediments. Springer, Berlin, 1989. pp. 157.
78 R.B Meagher, Curr. Opin. PlantBiol., 3 (2000) 153.
79 A. Guivarch, P. Hinsinger and S. Staunton, PlantSoil, 211 (1999) 131.
80 W.H.O. Ernst, New Phytol., 146 (2000) 357.
81 K.V. Nedunuri, R.S. Govindaraju, M.K. Banks, A.P. Schwab and Z. Chen, J. Environ.
Eng., 126 (2000) 335.
82 S.P. Pradhan, J.R. Conrad, J.R. Paterek and V.J. Srivastava, J. Soil Contamin., 7
(1998) 467.
83 R.L. Brigmon, N.C. Bell, D.L. Freedman and C.J. Berry, J. Soil Contam., 7 (1998) 433.
84 J.G. Burken, J.V. Shanks and P.L. Thompson, in: J.C. Spain, J.B. Hughes and H.-J.
Knackmuss (Eds.), Biodegradation of Nitroaromatic Compounds and Explosives.
Lewis Publishers, Boca Raton, FL, 2000, pp. 239.
85 P.L. Thompson, L.A. Ramer and J.L Schnoor, Environ. Sci. Technol., 32 (1998) 975.
86 S.D. Harvey, R.J. Fellows, D.A. Cataldo and R.M. Bean, Environ. Toxicol. Chem., 10
(1991) 845.
87 P.L. Thompson, L.A. Ramer and J.L. Schnoor, Environ. Toxicol. Chem., 18 (1999)
279.
88 S.D. Harvey, R.J. Fellows, D.A. Cataldo and R.M.J. Bean, J. Chromatogr. A, 518
(1990) 361.
89 S.L. Larson, Ann. N.Y. Acad. Sci., 829 (1997) 195.
90 C. Groom, S. Beaudet, A. Halasz, L. Paquet and J.A. Hawari, J. Chromatogr.A, 909
(2001) 53.
91 S.L. Larson, A.B. Strong, S.L. Yost, B.L. Escalon and D. Parker, Tech. Report
IRRP-98-5 US Army Corps of Engineers, Washington, DC, 1998.
92 S.B. Hawthorne, Anal. Chem., 62 (1990) 633A.
93 A. Halasz, C. Groom, L. Paquet, C. Beaulieu, S. Deschamps, A. Corriveau, S.
Thiboutot, G. Ampleman and J. Hawari, J. Chromatography(2002), in press.
94 J. Hawari, A. Halasz, T. Sheremata, S. Beaudet, C. Groom, L. Paquet, C. Rhofir, G.
Ampleman and S. Thiboutot, Appl. Environ. Microbiol., 66 (2000) 2652.
95 C.F. Shen, S.R. Guiot, S. Thiboutot, G. Ampleman and J. Hawari, Biodegradation,8
(1998) 339.
96 T. Sheremata, A. Halasz, L. Paquet, S. Thiboutot, G. Ampleman and J. Hawari,
Environ. Sci. Technol., 35 (2001) 1037.
97 J. Poerschmann, Z. Zhang, F.-D. Kopinke and J. Pawliszyn, Anal. Chem., 69 (1997)
597.
98 F.-D. Kopinke, J. Porschmann and M. Remmler, Naturwissenschaften,82 (1995) 28.
99 C. Denis-Larose, D. Labbe, H. Bergeron, A.M. Jones, C.W. Greer, J. Hawari, M.J.
Grossman, B.M. Sankey and P.C.K Lau, Appl. Environ. Microbiol., 63 (1997) 2915.

917
100 T. MacPherson, C. Greer, E. Zhou, A.M. Jones, G.Wisse, P.C.K. Lau, B. Sankey, M.J
Grossman and J.A. Hawari, Environ. Sci. Technol., 32 (1998) 421.
101 E.S. Olson, D.C. Stanley and J.R. Gallagher, Energy Fuels, 7 (1993) 159.
102 M.K. Lee, J.D. Senius and M.J. Grossman, Appl. Environ. Microbiol., 61 (1995) 4362.
103 J.A. Hawari, A. Demeter and R. Samson, Environ. Sci. Tech., 26 (1992) 2022.
104 C. Rhofir and J. Hawari, J. Chromatogr.A, 873 (2000) 53.

918
Chapter27

New developments in sampling and

sample preparation for forensic analysis
Jose R. Almirall and Kenneth G. Furton

27.1 INTRODUCTION

Forensic science is defined as the application of scientific principles and method-

ology to legal disputes. Forensic chemists work to interpret chemical evidence in
civil and/or criminal proceedings involving a variety of possible analytes of
interest. Advances in sampling and sample preparation in forensic analysis are
of particular interest to forensic scientists because of the implications that
improved sampling and sample preparation could have on the value of the
evidence in question. The opportunity to collect good evidence from a crime
scene is usually limited in time by the potential for contamination, evidence
degradation and the possible loss of security of the sample.
In recent years, a variety of new sample preparation techniques has been
applied to forensic evidence. Figure 27.1 illustrates some of the recent advances
and the evolution of sampling and sample preparation for six main types of
analytes of interests to forensic scientists, with solid phase microextraction
(SPME) as the most prominent advance in recent years. Considerable research
activity has been recently reported for the application of SPME as a sampling or
sample preparation strategy in four out of the six areas within the forensic
sciences. For example, sample preparation techniques developed for the
recovery of drugs and poisons have included liquid-liquid extraction (LLE),
solid-phase extraction (SPE), supercritical fluid extraction (SFE), accelerated
solvent extraction (ASE) and, most recently, SPME.
Figure 27.2 illustrates an application of SFE to the recovery of Phenobarbi-
tal from human liver tissues with comparable recovery and precision to
LLE/SPE with reduced sample handling, reduced analysis times and the elimi-
nation of halogenated solvents by employing non-toxic carbon dioxide. Unfortu-
nately, SFE has often proven to be too selective and the instrumentation
relatively expensive. SPME on the other hand is inexpensive, rapid and sensitive
and has other advantages as a sample preparation technique in forensic science.
While other recent sample preparation methods are important for specific sam-
ples ranging from Laser Ablation of materials such as glass prior to elemental

ComprehensiveAnalytical Chemistry XXXYVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 919
 =Il SPME I SPME On-Chip

1 ACS I I SFE/ASE I Microarrays

I
SolvcntExtraction SPE CR

team Distillation I LLE

Arson/Aeelerants Drugs/Poisons DNA/Biologicals

FORENSIC EVIDENCE SAMPLE PREPARATION METHODS I

Explosives/Firearm Glass/Fibers | Soils/Other Materials

E x t ra io n
Solven c Acid Digestion Soxhlet extraction

o |gTLC
TLC ~ 1~Utrasonication
SFEASE

SPE Microwave Digest.

=i E I i
!~l SPME I Laser AblationI S'ME

Fig. 27.1. Recent advances in sample preparation methods for common forensic evidence.

80

T ,'

60 I'
1~~~~~~~~.
. 40

L 20
.·-./". ,'"
u)

0
0 20 40 60 80
LLEISPEIGC (mg/l phenobarbital)

Fig. 27.2. Comparison of recoveries by SFE/EIA to LLE/SPE/GC [117].

analysis by ICPMS [1] to On-Chip technologies for DNA analysis, the focus of
this chapter is on recently developed methods employing SPME including ignit-
able liquid residues analyses, explosive traces analyses, and the extraction and
identification of drugs and poisons from biological specimens, amongst other
forensic applications. A recent review of sample preparation and analysis meth-
ods has shown that SPME has proven to be an important sample preparation
technique for the analysis of forensic specimens due to the many advantages the

920
 Odors I Water I Debris
]
Fig. 27.3. Some forensic applications employing SPME under various different modes.

technique offers when it is applied to these types of samples [2]. SPME allows for
multiple sampling, preservation of the sample and minimizes the risk of sample
contamination. It is often faster than traditional techniques and can be readily
automated. The lower detection limits generally afforded by SPME allow for
confirmation of positive samples that previously went undetected. An additional
benefit of SPME is the reduction of solvents in order to minimize the risk of ana-
lysts being exposed to toxic substances and also to save forensic science laborato-
ries money. This chapter describes the application of SPME for the analysis of
forensic specimens including ignitable liquid residues, often referred to as
accelerants, explosive traces, drugs and poisons from biological specimens and
other forensic applications. Recently developed SPME methods are also pre-
sented, including novel methods for the analysis of ignitable liquids on human
skin, odor signatures, and several drug applications such as the analysis of
free-fraction antipsychotic drugs levels and GHB identification without the need
for derivitization.
Figure 27.3 illustrates some applications of the SPME sampling and sample
preparation strategy within the forensic sciences. The major areas discussed in
this chapter include: drug analysis for toxicology; bulk drug analysis; trace anal-
ysis including high explosives and ignitable liquids; and sampling from various
matrices and a number of SPME sampling modes. A recent review paper [2] on
the application of SPME to forensic samples has also addressed this topic while
adding some new results by the authors.

27.2 IGNITABLE LIQUID RESIDUES FROM SUSPECTED ARSON

One very important aspect of an arson investigation is the detection and

identification of an ignitable liquid residue at the scene of the fire. Arson

921
investigators are trained to determine the possible point (location) of origin of
the fire and often collect the debris in that area in the hope that a small amount
of ignitable liquid residue has remained. The most widely used method for
sampling and pre-concentration is the multi-step process involving pre-
concentration of the volatile compounds in the debris onto an activated charcoal
strip. The investigators at the scene collect and package the debris into a new
metal paint can in such a way that the volatile species are contained within the
can. The can is transported to the laboratory for analysis some time later (the
time frame between collection and analysis may be several weeks). A small strip
of activated charcoal is placed in the heated headspace of the sealed can (at 80°C)
for 14 h (overnight) and the strip is then eluted of the analytes of interest with a
solvent such as carbon disulfide. This process is time- and labor-intensive and is
reported to detect approximately 0.1 i1 of an ignitable liquid residue from a
sample [3] of a gasoline spike into a 1 gal can when the typical extraction
volumes (50 A1)and GC-MS conditions are used.
The same year Pawliszyn et al. introduced SPME as a sampling and sample
preparation alternative [4], a coated wire adsorption technique was being
applied to the recovery of ignitable liquid residues by Tranthim-Fryer [5]. This
early technique involved the heated headspace (70-80°C) adsorption of ignitable
liquids onto carbon-coated aluminum or copper wire followed by n-pentane
elution with ultrasonic vibration [5]. The first report of SPME applied to the
recovery of ignitable liquid residues in 1994 [6] demonstrated improved sensitiv-
ity for the recovery of ignitable liquid residues with reduced analysis times and
the elimination of toxic solvents compared to the established activated charcoal
strip/solvent elution method [7]. The SPME analyses of gasoline and kerosene
has been compared to headspace, cold trap and solvent extraction methods and
shown to provide accurate information with less interference peaks [8].
Headspace SPME has also been applied to the analysis of flammable and
combustible substances in human body fluids [9,10]. A detailed study of the
recovery of gasoline residues confirmed the utility of the SPME technique
including lack of interference problems in the presence of wood or plastic
pyrolysis products and the ability of SPME to provide reproducible multiple
analyses from a single sample [11]. The recovery of ignitable liquids directly
from aqueous solvents has also been demonstrated, with SPME proving to be
more than an order of magnitude more sensitive than the conventional solvent
extraction method on 500 ppb preparations and allowed for positive identifica-
tion of diesel fuel in aqueous samples [121. SPME has also been used to identify
the presence of gasoline in a real arson-suspected fire debris sample, while
conventional methods lacked adequate sensitivity for identification (13).
SPME methods have been optimized for a variety of ignitable liquids and
conditions [14,15] and applied to the recovery and identification of ignitable
liquids from human skin [16]. The SPME method developed for recovery of
ignitable liquid residues from human skin used 100 m polydimethylsiloxane
(PDMS) fibers with gentle heating for 5 min followed by 10 min sampling from a

922
plastic bag shrouding the suspected hand. The recovery was found to be depend-
ent on the initial amount, environmental conditions, ignitable liquid type and
time since application [16]. Figure 27.4 shows chromatograms of a blank hand
(top) and hands contaminated with 10 Al of gasoline (middle) and diesel fuel

14000 Blank hand sampled by SPME

12000

10000

', 8000

c 6000

4000

2000

0
ji
Innn
0 5 10 15 20 25 30 min.
14000
Gasoline residues on hand after 1.5 hours
12000

.10000
,Z
8000

6000

4000 C4 Alkyl Benzenes

2000

-LUUU ······
_/ mSJ-<L.._YI!
0 10
bI 4Villll~\l
15 20 25
i i
30 min.

100000 Diesel fuel residues on hand after 3.5 hours

pentadecane hexadecane

80000

pristane
_ 60000

40000 tetradecane

20000

0 10 20 30 40 50

Fig. 27.4. Headspace SPME/GC/FID of blank and hands contaminated with accelerants and
sampled for 15 min using headspace SPME.

923
(bottom) several hours after exposure. A proposed comprehensive SPME scheme
for detecting ignitable liquid residues in suspected arson cases used a low tem-
perature carboxen/PDMS fiber followed by an elevated temperature PDMS fiber
extraction demonstrating simplified sample preparation and high recovery effi-
ciency compared to headspace charcoal adsorption and solvent extraction
methods [17]. A review of contemporary sample preparation methods, including
SPME, for the detection of ignitable liquids in suspected arson cases was
recently published [18]. The E30 committee on Forensic Sciences of the Ameri-
can Society for Testing and Materials (ASTM) approved a standard method
entitled E 2154-Standard Practice for the Separation and Concentration of
Ignitable Liquid Residues from Fire Debris Samples by Passive Headspace
Concentration with Solid Phase Microextraction (SPME) in 2001 for publication
in the Annual Book of ASTM Standards, Vol 14.01 of 2002.

27.3 EXPLOSIVES ANALYSIS

The detection and identification of very low levels of the organic explosives
commonly referred to as the high explosives can be a very important aspect of a
bombing investigation. Explosives detection before the detonation (pre-blast)
can provide information that could prevent a bombing and lead to the arrest of a
perpetrator and the identification of an explosives reside following a bombing
(post-blast) can provide investigative leads to solving the crime. The analysis of
organic explosives from soil found around storage facilities is also of importance
to environmental scientists as these compounds are monitored due to their
carcinogenic properties. The application of SPME to the analysis of these
compounds offers definite advantages over solvent extraction protocols.
Analysis of the semi-volatile explosive compounds, including nitrobenzene and
dinitrotoluenes in water has been reported using a PDMS fiber and GC/FID
analysis with detection limits reported at 9-15 ng/ml [19]. Headspace and direct
aqueous immersion SPME using PDMS/DVB fibers followed by GC/MS and
LC/UV were used to recover 14 explosives with varying results depending on the
extracted explosive [201. An optimized headspace SPME/GC/MS method has
been published using polyacrylate resin with 30-min adsorptions at 100°C and
desorptions at 200°C with reported detection limits of 0.5-10 ng/750 ml head-
space for EGDN, NG, PETN, TNT and RDX [21]. Trace explosive signatures
have been determined from World War II unexploded undersea ordinance using
direct immersion SPME/GC/MS and SPME/GC/Reversal Electron Attachment
Detection (READ) yielding improved extractions over solid phase extraction
with sensitivities of 10 parts per trillion for DNT and TNT for 15 min extractions
using PDMS/DVB fibers [22]. Direct immersion SPME using a Carbowax/DVB
fiber with 10 min sampling followed by a GC/ITMS analysis yielded limits of
detection of 10-325 ppt for TNT, RDX, and amino-DNTs in seawater [23]. Direct
immersion SPME/GC, SPME/LC and SPME/CZE methods have been optimized,
compared and successfully applied to actual post-explosion debris samples

924
 Analytical Pump(2)
len-port Valve at "Sample Loading" positionn

tion
I)

I
Analytical Pump(2)
Ten-Port Valve at "Sampie lajectiol" Positio

Fig. 27.5. SPME/HPLC interface employing a refocusing column.

[24,25]. Headspace SPME has also been applied to the characterization of odor
signatures, including explosive odor signatures [26]. The relative effects of major
controllable variables for the SPME recovery of explosives and ignitable liquid
residues have been reviewed including analyte chemistry, sampling mode
(direct, headspace, and partial headspace), fiber chemistry, adsorption times,
adsorption temperatures, desorption temperature, desorption time, and matrix
effects including water content relative to sample container size [27].
An improved HPLC interface (see Fig. 27.5) utilizing a 10 port valve and a
pre-focusing column was reported [27] to improve the resolution of the high
explosives analysis to the point where all fourteen [14] compounds in the EPA
8330 method standard mixture are readily resolved using a single solvent system
as the mobile phase. A comparison of the detection limits calculated for direct
injection vs SPME extraction for MECC/PDA, GC/ECD and LC/UV methods is
summarized in Table 27.1 [27].

925
TABLE 27.1
Detection limits (5 replicates, S/N > 3) of explosives by SPME/GC, SPME/HPLC and SPME/
MECC (measured at 254 nm, *220 nm; *'200 nm)
Explosives LOD (ug/ml) LOD (ng/ml)
MECC/PDA GC/ECD LC/UV (254 nm)
MECC SPME/ GC SPME/ GC HPLC SPME/
MECC HPLC
NB 0.74 (6.8) 0.69 (5.3) 53 (3.3) 0.24 (3.5) 3.9 (2.7) 1.2 (2.7)
1,3-DNB 0.41 (7.6) 0.38 (5.8) 44 (2.7) 0.22 (2.7) 2.3 (1.2) 0.8 (1.2)
1,3,5-TNB 0.62 (7.6) 0.62 (11.8) 39 (3.7) 0.18 (2.1) 2.6 (1.1) 0.6 (1.2)
2-NT 0.92 (4.3) 0.53 (11.7) 55 (3.2) ' 0.24 (3.4) 2 (3.7) 1.8 (3.8)
4-NT 1.12 (5.8) 0.51 (10.2) 47 (2.7) 0.24 (2.8) 7.4 (4.1) 1.9 (4.2)
3-NT 0.87 (4.3) 0.42 (11.8) 64 (2.7) 0.24 (2.9) 7.3 (4.0) 1.7 (4.0)
4-A-2,6-DNT 0.6 (7.4) 0.21 (4.0) 30 (1.7) 0.1 (1.6) 2.9 (1.6) 1.2 (1.6)
2-A-4,6-DNT 0.49 (9.2) 0.18 (3.8) 29 (1.4) 0.09 (1.4) 3.4 (2.1) 1.2 (19)
2,6-DNT 0.63 (6.3) 0.28 (5.3) 27 (1.9) 0.09 (1.5) 4.4 (3.0) 1.3 (2.9)
2,4-DNT 0.41 (7.5) 0.19 (3.4) 29 (1.5) 0.1 (1.7) 3.4 (1.9) 1.2 (1.7)
TNT 0.54 (7.8) 0.25 (4.2) 22 (1.7) 0.09 (1.7) 2.9 (1.2) 11 (1.3)
HMX 1.61 (6.6) 0.73 (9.8) - - 3.8 (2.6) 1.2 (2.7)
RDX 1.24 (7.4) 1.01 (17.4) 130 0.61 (5.2) 2.9 (2.3) 1.1 (2.5)
Tetryl 0.67 (6.9) 0.24 (4.1) 43 (4.0) 0.25 (4.2) 3.2 (2.8) 1.3 (2.9)
EGDN 2.96 (8.5)* 2.35 (18.2) 51 (3.7) 0.22 (3.8) 550 120 (4.6)*
NG 1.71 (4.5)* 0.37 (7.8)* 89 (2.8) 0.58 (2.9) 500 110 (4.8)*
PETN 1.89 (6.7)* 0.20 (6.4)* 94 (3.7) 0.61 (3.8) 380 80 (4.3)*

27.4 TOXICOLOGICAL ANALYSIS

The extraction of poisons from biological specimens, including urine, serum and
blood, continues to be a difficult and time-consuming task. Increasingly, tradi-
tional liquid-liquid extraction techniques have been improved on and replaced
by newer methods including solid-phase extraction (SPE) and supercritical fluid
extraction (SFE) [28]. Most recently, SPME has emerged as a promising alterna-
tive for the rapid recovery of drugs from biological fluids such as urine or blood
and of poisons from biological matrices [14,29] with recent applications detailed
below.

27.4.1 Toxicological analysis from urine

A number of recent review papers have described the use of SPME within
toxicology applications including a review [30] that provides a good overview of
the theoretical and practical understanding of the use of microextraction tech-
nologies for drug analysis and three reviews [31-33] focusing on the extraction of

926
drugs from biological matrices. A fourth review [34] in the same year would also
cover the theory and application of SPME in biomedical analysis. Recent reports
of SPME-GC-MS for drug analysis [35], the use of SPME with HPLC and CE
[36] and specifically dealing with the analysis of drugs from hair using a
headspace sampling method [37] are further evidence of the wider acceptance of
SPME as a sampling tool in toxicology laboratories.
A large number of SPME applications to toxicology have involved the
recovery of drugs from urine or the recovery of drugs from blood and serum.
SPME/GC has been successfully applied to the recovery of methadone, benzo-
diazepines, cannabinoids, phencyclidine, methaqualone, amphetamines and
their metabolites from urine, but was not as successful for the recovery of
cocaine or opiates and their metabolites at pH 12 using 20 min direct immersion
SPME at 40°C with PDMS and polyacrylate fibers [38]. A direct immersion
SPME method for the detection of cocaine down to 6 ng/0.5 ml human urine has
been reported using a PDMS fiber with 30 min sampling from urine with added
NaF followed by GC/nitrogen-phosphorous detection (NPD) [39]. A SPME
method has been developed for the determination of benzoylecgonine in urine
with hexyl chloroformate derivitization using a 100 Aum PDMS fiber for 10 min at
55°C and ion trap MS detection [40].
The determination of amphetamine, methamphetamine (MA) and dimeth-
amphetamine in urine has been performed by 100 Atm PDMS SPME/GC/MS
sampled for 30 min at pH 12.4 with 30% salt added [41]. Methamphetamine and
amphetamine were detected down to ca. 10 ng/ml using PDMS/DVB fibers to
extract the drugs from urine treated with 1 g/ml sodium carbonate (Na 2CO3 ) for
30 min at 65°C followed by GC/NPD analysis [42]. Another method for the
recovery of methamphetamine and amphetamine used 100 Am PDMS fibers
from urine treated with NaCl (0.5 g/ml) adjusted to pH 12 and extracted for 20
min followed by a NaOH-H3 BO4 buffer wash prior to GC analysis with reported
sensitivity 10-100 times greater than headspace methods including SPME [43].
Amphetamines and inflammable compounds in urine and blood have been
analyzed using a 100 Am PDMS fiber immersed in basic solutions for 5 min at
80°C with analysis by GC/MS [44]. The automated determination of amphet-
amines and ecstasy in urine has been reported using 100 m PDMS fibers
immersed at pH 10 for 16 min followed by analysis by GC/NPD or GC/MS [45].
Although sometimes less sensitive, depending on the volatility of the drug,
headspace SPME has the advantage of yielding cleaner extracts from biological
samples. A 5 min headspace SPME/GC/CI-MS procedure with 100 m PDMS
fibers for amphetamines from urine containing potassium carbonate at 80°C was
shown to be 20 times more sensitive than the conventional headspace method
[46]. Another headspace SPME/GC/MS procedure using a 100 Am PDMS fibers
extraction for 15 min over a solution containing NaCl at 75°C was shown to be
sensitive enough for routine confirmation of positive EMIT and RIA results for
amphetamine and its analogs (MA, MDMA, MDEA) [47]. A headspace SPME/
GC/FID method has also been reported for the analysis of amphetamines in clini-

927
cal urine samples using 100 um PDMS fibers for 15 min at 60°C with further
method development and derivatization suggested for conclusive differentiation
between the various types of amphetamines [48]. A screening procedure for 21
amphetamine related compounds has been developed using 100 m PDMS
headspace SPME/GC/MS using a 10 min extraction at 80°C with salt added [49].
Finally, the rapid analysis of amphetamine, MA, methylenedioxyethylamphet-
amine (MDA) and methylenedioxymethamphetamine (MDMA) has been report-
ed using 100 Am PDMS headspace SPME/GC/MS using a 10 min extraction at
100°C with 2 N NaOH followed by direct on-fiber derivatization for 20 min at
60°C using trifluoroacetic anhydride [50].
Methadone has been quantified in urine using 100 Am PDMS fiber for 15 min
at pH 7.7 followed by GC/MS quantitation [51]. Nicotine and cotinine in urine
have been analyzed down to 5 and 300 ng/ml, respectively, using 100 Am PDMS
headspace SPME/GC/MS using a 5 min extraction at 80°C with the addition of
potassium carbonate [52]. Tricyclic antidepressants in urine have been analyzed
down to 24-38 ng/ml using 100 Am PDMS headspace SPME/GC-FID using a 15
min extraction at 100°C with the addition of sodium hydroxide [53]. Meperidine
in urine and blood have been analyzed down to 20 and 100 ng/ml, respectively,
using 100 pm PDMS headspace SPME/GC-FID using a 30 min extraction at
100°C with the addition of sodium hydroxide and sodium chloride [54]. There
have recently been several reports of the determination of benzodiazepines and
metabolites using direct immersion methods. A method using 65 am PDMS/DVB
sampling for 30 min followed by GC-FID analysis was applied to the detection of
benzodiazepines in urine [55]. Another method for determining benzodiazepines
in urine used 65 gm Carbowax/DVB sampling for 60 min at 45°C with added salt
followed by GC-FID and GC-MS analysis [56]. The analysis of benzophenones
for the detection of benzodiazepines in urine has been reported using 100 Am
PDMS SPME/GC/ECD using a 30 min extraction at pH 9.4 with the addition of
potassium hydroxide [57]. The analysis of benzophenone and metabolites in
water and human urine has been reported using 65 Am Carbowax/DVB SPME/
GC/M$ with a 45 min immersion time [58]. Barbiturates have been determined
in urine using 65 pm Carbowax/DVB SPME/GC/MS using a 20 min immersion
time [59].
Non-routine volatiles including methylene chloride and petroleum products
were confirmed in urine and in a gastric sample using 100 Am PDMS headspace
SPME/GC/MS using a 10 min extraction at 60°C with the addition of sodium
chloride [60]. The absence of an air peak in the GC/MS afforded by the SPME
procedure offered a tremendous advantage in the identification of the unknown
volatiles, which proved to be crucial evidence in the investigation of traffic
fatalities [61]. Headspace SPME using 100 am PDMS fibers for 15 min at 60°C
has been used to determine benzene, toluene, ethylbenzene and xylenes (BTEX)
in urine with GC/MS [62]. An immersion method using 100 /m PDMS fibers for
5 min with added NaCl has been used to analyze urinary toluene and xylene from
workers using organic solvents with GC/FID [63]. The analysis of chlorophenols

928
in urine has been accomplished using 85 m polyacrylate SPME/GC/MS using a
50 min extraction at pH 1 [63]. Another method for the analysis of chlorophenols
in urine of subjects exposed to chlorobenzene used 85 gm polyacrylate SPME/
GC/MS using a 30 min extraction with added salt [64].

27.4.2 Toxicological analysis from blood and plasma

The free concentration of valproic acid in human plasma has been determined at
1000 ng/ml using 100 m PDMS direct immersion SPME/GC-FID using a 3 min
extraction from a sample previously dialyzed for 25 min and adjusted to pH 2.5
[65]. A calibration curve for valproic acid recovered from buffered standards and
plasma samples centrifugally purified to remove proteins followed by direct
immersion SPME/GC-MS with a 65 gm Carbowax/DVB fiber for 10 min is shown
in Fig. 27.6. A 7.5 gg/ml caprylic acid internal standard was added to plasma
solutions containing valproic acid (1.25-12.5 g/ml) at pH 6.86. Two milliliters of
each plasma sample was pipetted into a 2 ml Centricon YM-30 cell, and then
centrifuged 30 min to get 1 ml filtrate followed by SPME and desorption into the
GC inlet for 5 min at 250°C. Figure 27.6 illustrates the excellent linearity of this
SPME method for quantifying the free fraction concentration of this drug.
Antidepressant drugs and metabolites in human plasma have been analyzed
down to 90-200 ng/ml using 100 m PDMS headspace SPME/GC-NPD using a
10 min extraction at 22°C with high protein binding appearing to be the limiting
mechanism for better extractions [66]. Four tricyclic antidepressants in blood
have been analyzed down to 61-2000 ng/ml using 100 gm PDMS headspace
SPME/GC-FID using a 60 min extraction at 100°C with the addition of sodium
hydroxide [67]. The analysis of antidepressants in blood and urine have been
analyzed, including an analysis in a fatal intoxication case using 100 gm PDMS
headspace SPME/GC-MS using a 45 min extraction at 120°C with the addition of
sodium hydroxide [68,69]. Methamphetamine and amphetamine in blood have
been analyzed down to 10 ng/ml using 100 gm PDMS headspace SPME/GC/MS
using a 5 min extraction at 80°C with the addition of sodium hydroxide [70].
Phencyclidine has been recovered from whole blood and urine analyzed down to
1.0 ng/ml and 0.25 ng/ml, respectively, using 100 gm PDMS headspace SPME/
GC/surface ionization detection (SID) using a 30 min extraction at 90°C with the
addition of sodium hydroxide and K2 C0 [71]. Methadone and metabolites have
been recovered from plasma via 100 gm PDMS direct immersion SPME/GC-MS
using a 30 min extraction at a pH 7.7 [72]. Diphenylmethane in blood and urine
has been analyzed by 100 gm PDMS headspace SPME/GC-FID using a 10 min
extraction at 98°C with 10 N NaOH added [73]. Diazepam in human plasma has
been determined using 85 gm polyacrylate direct immersion SPME/GC-FPD
using a 4 min extraction at pH 5.5 with 1-octanol added [74]. The authors
applied a method using 100 gm PDMS SPME/GC/MS with 5 min immersion at
25°C to confirm a suspected drink tampering case with the equivalent of ten
tablets containing 2 mg of diazepam each added to a whisky bottle.

929
 1
q
0.9
0.8
0.7
0.6
'. 0.5
v 0.4
o 0.3
X 0.2
0.1
0
-0.1 0 2000 4000 6000 8000 10000 12000 14000

Valproic Acid Concentration (ng/ml)

Fig. 27.6. Aqueous calibration curve for valproic acid in a pH 6.8 buffer using caprylic acid as an
internal standard analyzed by SPME/GC-MS.

Steroids have been analyzed from human serum using 85 tLm polyacrylate
direct immersion SPME/GC/MS using a 30 min extraction followed by headspace
derivitization with BSTFA [75,76]. Caffeine metabolites have been recovered
from blood and urine via 65 Am carbowax/DVB SPME/GC-NPD using a 60 min
extraction at 40°C with added acid [77]. SPME has been successfully applied to
the detection of ethanol in human body fluids, including urine and blood, down
to 10 and 20 jLg/ml, respectively, using 65 A8m Carbowax/DVB headspace SPME/
GC-FID using a 15 min extraction at 70°C with the addition of (NH4 )2SO4 [78].
An automated headspace SPME method for the analysis of blood alcohol concen-
tration using 65 Am Carbowax/divinylbenzene headspace SPME/GC-FID using 3
min exposures showed excellent precision and linearity with results in close
agreement with the conventional static headspace method [79]. An improved
method for the recovery of ethanol from blood and urine employed a 75 Atm
carboxen/PDMS headspace SPME/GC/FID using a 15 min extraction at 60°C
with the addition of (NH4 )2SO4 and sodium dithionite, improving sensitivity by
one to three orders of magnitude compared to previous methods [80]. The analy-
sis of methanol in whole blood has also been reported using headspace SPME/
GC-FID with a carboxen/PDMS fiber using 10 min exposures at 60°C [81]. An
example of SPME analysis of a blood alcohol casework sample analyzed by 65 m
Carbowax/DVB headspace automated SPME/GC-MS using 3 min headspace
adsorption at 25°C.

27.4.3 Unconventional poisons

Although the most common toxicological analysis involve drug extraction and
recovery, drugs actually represent only ca. 60% of the more than 1500 possible
poisons [41]. The second largest class of poisons are the pesticides representing
ca. 30% of possible compounds followed by anions, metals, gases and volatiles,

930
which comprise a total of less than 10% of possible poisons [41]. Organo-
phosphate pesticides in blood and urine have been analyzed down to 5-80 ng/ml
and 2-24 ng/ml, respectively, using 100 Am PDMS headspace SPME/GC/NPD
using a 20 min extraction at 100°C with the addition of HC1 only for blood and
HC1, NaCl and (NH 4 )2 SO4 for urine [82]. The common organophosphorous
pesticide, malathion, in blood has been detected down to 1000 ng/ml using 100
tm PDMS headspace SPME/GC/MS using a 5 min extraction at 90°C with the
addition of (NH 4 )2 SO4 and H2 SO4 [83]. Six carbamate pesticides in blood and
urine have been analyzed down to 100-500 ng/ml and 10-50 ng/ml, respectively,
using 100 tm PDMS headspace SPME/GC/NPD using a 30 min extraction at
70°C with the addition of NaC1 [84]. Phenothiazines have been detected using
100 um PDMS headspace SPME/GC-FID with a 40 min extraction at 140°C with
added NaC1 [85]. Dinitroaniline herbicides have been analyzed using 100 tm
PDMS SPME/GC-ECD with 30 min extractions at 70°C for water and 90°C for
blood [86]. The analysis of parathion in blood has also been reported using
headspace SPME/GC-MS [87]. The natural insecticide, nereistoxin, has been
analyzed using 65 gm PDMS/DVB headspace SPME/GC-MS with a 30 min
extraction at 70°C with added NaCI [88].
A headspace SPME/GC/MS method has been developed for the analysis of
toluene, xylenes and hydrocarbons at 100-1000 ng/ml in human blood and
applied to the medico-legal autopsy of a fire victim with kerosene substances
detected [89]. The analysis of thinner components in whole blood and urine
down to 1 ng/ml has been performed using 100 Am PDMS headspace SPME/
GC-FID using a 5 min extraction at 80°C [90]. Cresol and phenols have been
detected using 85 Am polyacrylate headspace SPME/GC-FID with a 30 min
extraction at 100°C with added NaC1 [91]. Benzene and toluene in human blood
have also been monitored using 65 m carboxen/PDMS headspace SPME/
GC-MS with a 30 min extraction at 20°C [92]. Headspace sampling of BTEX in
blood and urine of cyclists exposed to air pollutants employed 75 Am carboxen/
PDMS SPME/GC-MS with a 30 min extraction at 50°C [93]. Tetrachloroethylene
and trichloroethylene have been analyzed in serum tissue and urine in a fatality
using 100 m PDMS headspace SPME/GC-ECD with a 1 min extraction at 60°C
[94].
Local anesthetics have been analyzed down to 58-830 ng/ml using 100 gm
PDMS headspace SPME/GC-FID using a 40 min extraction at 100°C from
human blood deproteinized with perchloric acid and sodium hydroxide and
(NH4 )2SO4 added [95]. The recovery of local anesthetics from human blood has
also been accomplished with a direct immersion using 100 Mm PDMS SPME/
GC-FID with a 40 min extraction [96]. Another recent method for the recovery of
local anesthetics employed 100 Mm PDMS headspace SPME/GC-MS using sel-
ected ion monitoring (SIM) with a 45 min extraction at 120°C with added NaOH
[97]. The analysis of the highly toxic methylmercury in aqueous and tissue
samples has been reported using 100 Am PDMS headspace SPME/GC-atomic
fluorescence spectrometry (AFS) using simultaneous extraction/derivitization

931
with sodium tetraethylborate at 25°C and pH 4.5 with an acetate buffer [98]. The
simultaneous determination of mercury, alkylmercury, lead and tin in human
body fluids was accomplished using 100 Am PDMS headspace SPME/GC-MS-MS
also using derivitization with sodium tetraethylborate for 10 min at pH 5.3 [99].
Methylmercury in biological samples has also been reported using headspace
SPME/GC-atomic absorption spectrometry (AAS) using hydride derivitization
[100]. Cyanide in blood has been analyzed using 65 tkm carbowax/DVB head-
space SPME/GC-NPD using a 45 min extraction at 50°C with Na2 SO4 added
[101]. Lead in blood and urine has been analyzed below 10 ppb using 65 ~m
PDMS/DVB headspace SPME/GC-FID using a 15 min extraction/derivitization
time with sodium tetraethylborate at 25°C and pH 4.0 with an acetate buffer
[102].

27.5 MISCELLANEOUS APPLICATIONS

A modified SPME device has been developed which is capable of detecting down
to 5.8 nmol/l ethanol, 1.8 nmol/l acetone and 0.3 nmol/l isoprene in human
breath using a 65 m PDMS/DVB fiber and 1 min sampling followed by GC/MS
analysis [103]. SPME has been shown to be a useful tool to determine phospho-
lipid/water partition coefficients and free (bioavailable) concentrations in vitro
systems, which may make experimental data more meaningful for quantitative
in vivo extrapolations [104]. SPME has been used to determine residual organic
solvents in pharmaceutical samples with an optimized method using 65 Jum
PDMS/DVB with 30 min headspace extraction with NaCl added [105]. A method
for profiling confiscated ecstasy and amphetamine uses 65 tm PDMS/DVB
headspace SPME/GC-NPD with a 30 min extraction at 90°C with a pH of 5.0
using an acetate buffer [106]. SPME has also been used to characterize street
narcotic odors. The room-temperature recoveries of cocaine odor chemicals,
including methyl benzoate, required optimized headspace extraction times of up
to 12 h with the optimum fiber found to be 65/ m carbowax/DVB [107]. Although
it has been established that most US currency in circulation is contaminated
with gg quantities of cocaine, the corresponding odor chemical (methyl benzo-
ate) levels are shown to be well below the threshold levels of drug detector dogs
[107,108]. Recently, odor signature chemicals for other controlled substances
have been identified using SPME/GC/MS including street MDMA (Ecstasy)
samples as shown in Fig. 27.7.
A method for determining cannabinoids in water and saliva used 75 Aum
carboxen/PDMS direct immersion SPME/GC-MS with a 10 min extraction [109].
SPME has also been applied to the analysis of cannabinoids and amphetamines
in hair [110,111] and has been used to characterize items including cinnamon,
vodka, coffee and tobacco [112-115] which could prove useful investigative infor-
mation. Recently, a fast and simple SPME method has been developed for the
identification of the y-hydroxybutyric acid (GHB) from aqueous samples [116].
The direct injection of a GHB extraction into a GC port under normal tempera-

932
 I
6500000 1. Methanphetamie

6000000 2. Piperonal
2 4
5500000 3. 3,4 methyedioxyphenyl
5000000 acetone (MD-P2P) I
4500000 4. 1-(3,4 methylenedioxyphenyl-2-
. 4000000 propanol)
2 3500000 5. Bylated Hydroxytoluene
3000000
(B3Hl)
2500000
5
2000000

1500000
1000000
500000
1 - II .I ,
2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00 30.00
Time

Fig. 27.7. Odor compounds commonly identified in MDMA tablets with PDMS fiber sampled for
3 h.

ture (280°C) conditions followed by electron ionization in the mass spectrometer

results in the thermally induced intra-molecular esterification forming the
butyrolactone (see Fig. 27.8, top, for the EI spectra) and preventing the positive
identification of the drug. A 15 min 100 /tm PDME SPME headspace extraction
at 60°C followed by a low temperature (100°C) desorption into the injector of a
GC outfitted with a short (2 m) DB-1 column followed by positive chemical ion-
ization using methanol as the reagent gas yields good fragmentation and a con-
firmatory molecular ion + 1 peak (see Fig. 27.8, bottom, for the CI spectra).

27.6 REMOTE SAMPLING METHODS FOR ANALYTES OF FORENSIC

INTEREST

One obvious advantage of the SPME sampling and sample concentration tech-
nique is the adaptability of the technique for field sampling. The apparatus is
portable and amenable to remote sampling schemes. Possible future applica-
tions include the collection of explosives at the scene of a post-blast bombing
scene and the scene of a suspected arson. One of the major drawbacks of current
sampling schemes for these types of evidence is the time lag between the
occurrence of the event and the analysis of the analytes of interest. The advent of
portable analytical instrumentation such as GC/MS systems could resolve some
of the problems associated with this time lag including the deterioration of
sample, possible contamination and evidence integrity. A remote sampling
scheme involving SPME with subsequent analysis would improve the quality of
these types of evidence.

933
 1(00% 42

75%
87

0%-

25%
56
0%
73
00%

75%

50%

25% 45 89
59 . 105

0%
50 75 160o 125 150
m/z

Fig. 27.8. GC/EIMS of y-hydroxybutyric acid (GHB) (top) yielding an inconclusive identification
of this drug compared to the SPME/GC/CIMS (bottom) providing improved fragmentation and a
molecular ion + 1 peak.

27.7 CONCLUSIONS

Overall, most reported SPME methods have proven to be superior to existing

extraction methods with overall improved recoveries translating to lower
detection limits with shorter analysis times, minimal sample handling and the
significant reduction or elimination of organic solvents. Future developments of
SPME for forensic applications will likely include further improvements in
automation and field sampling and analysis including Fast GC methods. Field
portable SPME/GC methods could dramatically improve law enforcement
response time by providing rapid on-site confirmation of drugs, accelerants, and
explosives. Hurdles to more widespread acceptance in the forensic science
community include the need for additional optimization and the establishment
of more standard methods (i.e., ASTM) for analysis using SPME.

REFERENCES

1 J.R. Almirall, in: B. Caddy (Ed.), ForensicExaminations of Glass and Paint;Analysis

and Interpretation. Taylor and Francis, London and New York, 2001, Ch. 4, pp.
65-84.
2 K.G. Furton, J. Wang, Y. Hsu, J. Walton and J.R. Almirall, J. Chromatogr. Sci., 38
(2000) 297.

934
 3 ASTM E 1412-00, 2001 Annual Book of ASTM Standards, 14.02, (2001) pp. 431-433.
4 C.L. Arthur and J. Pawliszyn, Anal. Chem., 62 (1990) 2145.
5 D.J. Tranthim-Fryer, J. Forensic Sci., 35 (1990) 271.
6 J.R. Almirall, K.G. Furton and J.C. Bruna, Southern Association of Forensic
Scientists Fall 1994 Meeting, Orlando, Florida, September 7-10, 1994.
7 K.G. Furton, J.R. Almirall and J. Bruna, J. High. Resolut. Chromatogr., 18 (1995) 625.
8 T. Kaneko and M. Nakada, Reports of the National Research Institute of Police
Science: Research on ForensicScience, 48 (1995) 107.
9 Y. Iwasaki, M. Yashiki, N. Nagasawa, T. Miyazaki and T. Kojima, Jpn. J. Forensic
Toxicol., 13 (1995) 189.
10 X.P. Lee, T. Kumazawa and K. Sato, Int. J. Legal. Med., 107 (1995) 310.
11 K.G. Furton, J.R. Almirall and J. Bruna, J. ForensicSci., 41 (1996) 12.
12 J.R. Almirall, J. Bruna and K.G. Furton, Sci. Justice, 36 (1996) 283.
13 Steffen and J. Pawliszyn, Anal. Communications, 33 (1996) 129.
14 J.R. Almirall and K.G. Furton, in: S.A. Wercinski (Ed.), Applications in Solid Phase
Microextraction: A Practical Guide. Marcel Dekker, New York, 1999, Ch. 7, pp.
203-216.
15 K. Higgins (Ed.), Proceedings of the SPIE Conference on Investigation and Forensic
Science Technologies, Boston, 1998, 3576, pp. 136-141.
16 J.R. Almirall, J. Wang, K. Lothridge and K.G. Furton, J. ForensicSci., 45 (2000) 461.
17 Q. Ren and W. Bertsch, J. ForensicSci., 44 (1999) 504.
18 W. Bertsch and Q. Ren, ForensicSci. Rev., 11 (1999) 141.
19 J.Y. Horng and S.D. Huang, J. Chromatogr.,678 (1994) 313.
20 M. Bi, M.S. thesis, Florida International University, Miami, FL, 1998.
21 K.P. Kirkbride, G. Klass and P.E. Pigou, J. ForensicSci., 43 (1998) 76.
22 M.R. Darrach, A. Chutjian and G.A. Plett, Environ. Sci. Technol., 32 (1998) 1354.
23 S.A. Barshick and W.H. Griest, Anal. Chem., 70 (1998) 3015.
24 K. Higgins (Ed.), Proceedings of the SPIE Conference on Investigation and Forensic
Science Technologies, Boston, 1998, 3576, pp. 18-23.
25 K.G. Furton, L. Wu and J.R. Almirall, J. Forensic Sci., 45 (2000) 845.
26 K.G. Furton and L.J. Myers, Talanta, 54 (2001) 487.
27 K.G. Furton, J.R. Almirall, M. Bi, J. Wang and L. Wu, J. Chromatogr.A, 885 (2000)
419.
28 K.G. Furton and J. Rein, Anal. Chim. Acta, 236 (1990) 99.
29 L. Junting, C. Peng and 0. Suzuki., ForensicSci. Int., 97 (1998) 93.
30 H. Lord and J. Pawliszyn, J. Chromatogr.A, 902 (2000) 17.
31 N.H. Snow, J. Chromatogr.A, 885 (2000) 445.
32 G.A. Mills and V. Walker, J. Chromatogr.A, 902 (2000) 267.
33 G. Theodoridis, E.H.M. Koster and G.J. de Jong, J. Chromatogr. B, 745 (2000) 49.
34 S. Ulrich, J. Chromatogr.A, 902 (2000) 167.
35 U. Staerk and W.R. Kiilpmann, J. Chromatogr. B, 745 (2000) 399.
36 K.E. Rasmussen, S. Pederson-Bjergaard, M. Krogh, H.G. Ugland and T. Gronhaug,
J. Chromatogr.A, 873 (2000) 3.
37 F. Sporkert and F. Pragst, ForensicSci. Int., 107 (2000) 129.
38 K. Singer, B. Wenz, V. Seefeld and U. Speer, Labor-Med. (German), 18 (1995) 112.
39 T. Kumazawa, K. Watanabe, K. Sato, H. Seno, A. Ishii and 0. Suzuki, Jpn. J.
Forensic Toxicol., 13(3) (1995) 207.
40 B.J. Hall, A.R. Parikh and J.S. Brodbelt, J. Forensic Sci., 44 (1999) 527.
41 S.W. Myung, H.K. Min, S. Kim, M. Kim, J.B. Cho and T.J. Kim, J. Chromatogr., 716
(1998) 359.
42 A. Ishii, H. Seno, T. Kumazawa, M. Nishikawa, K. Watanabe, H. Hattori and 0.
Suzuki, Jpn. J. Forensic Toxicol., 14 (1996) 2228.

935
43 K. Ameno, C. Fuke, S. Ameno, H. Kinoshita and I. Ijiri, Can. Soc. Forens. Sci. J., 29
(1996) 43.
44 M.Yashiki, T. Kojima and T. Miyazaki, Jpn. J. Forensic Toxicol., 12 (1994) 120.
45 H. Ugland, M. Kogh and K. Rasmussen, J. Pharm. Bomed. Anal., 19 (1999) 463.
46 M. Yashiki, T. Kojima, T. Miyazaki, N. Nagasawa, Y. Iwasaki and K. Hara, Forensic
Sci. Int., 76 (1995) 169.
47 F. Centini, A. Masti and I.B. Comparini, Forensic Sci. Int., 83 (1996) 161.
48 H. Lord and J. Pawliszyn, Anal. Chem., 69 (1997) 3899.
49 C. Battu, P. Marquet, A. Fauconnet, E. Lacassie and G. Lachatre, J. Chromatogr.
Sci., 36 (1998) 1.
50 C. Jurado, M. Gimenez, T. Soriano, M. Menedez and M. Repetto, J. Anal. Toxicol., 24
(2000) 11.
51 M. Chiarotti and R. Marsili, J. Microcolumn, 6 (1994) 577.
52 M. Yashiki, N. Nagasawa, T. Kojima, T. Miyazaki and Y. Iwasaki, Jpn. J. Forensic
Toxicol., 13 (1995) 1724.
53 T. Kumazawa, X.P. Lee, M.C. Tsai, H. Seno, A. Ishi and K. Sato, Jpn. J. Forensic
Toxicol., 13 (1995) 25.
54 H. Seno, T. Kumazawa, A. Ishii, M. Nishikawa, H. Hattori and 0. Suzuki, Jpn. J.
Forensic Toxicol., 13 (1995) 211.
55 H. Seno, T. Kumazawa, A. Ishii, K. Watanbe, H. Hattori and 0. Suzuki, Jpn. J.
ForensicToxicol., 15 (1997) 16.
56 Y. Luo, L. Pan and J. Pawliszyn, J. Microcolumn, 10 (1998) 193.
57 F. Prosen, H. Seno, A. Ishii, K. Watanabe, T. Kumazawa, H. Hattori and 0. Suzuki,
J. Anal. Toxicol., 23 (1999) 54.
58 T. Felix, B. Hall and J. Brodbelt, Anal. Chim. Acta, 371 (1998) 195.
59 B. Hall and J. Brodbelt, J. Chromatogr.A, 777 (1997) 275.
60 W.E. Brewer, R.C. Galipo, S.L. Morgan and K.H. Habben, J. Anal. Toxicol., 21 (1997)
286.
61 S. Fustinoni, R. Giampiccolo, S. Pulvirenti, M. Buratti and A. Colombi, J.
Chromatogr. B, 723 (1999) 105.
62 F. Asakawa, F. Jitsunari, J. Choi, S. Suna, N. Takeda and T. Kitamado, Bull.
Environ. Contamin. Toxicol., 62 (1999) 109.
63 M. Lee, Y. Yeh, W. Hsiang and C. Chen, J. Chromatogr. B, 707 (1998) 91.
64 M. Guidotti, G. Ravaioli and M. Vitali, J. High Resolut. Chromatogr., 22 (1999) 427.
65 M. Krogh, K. Johansen, F. Tonnesen and K.E. Rasmussen, J. Chromatogr.B, 673
(1995) 299.
66 S. Ulrich and J. Martens, J. Chromatogr.B, 696 (1997) 217.
67 X.P. Lee, T. Kumazawa, K. Sato and 0. Suzuki, J. Chromatogr.Sci., 35 (1997) 302.
68 T. Watanabe, A. Namera, M. Yashiki, Y. Iwasaki and T. Kojima, Jpn. J. Legal. Med.,
52 (1998) 69.
69 A. Namera, T. Watanabe, M. Yahiki, Y. Iwasaki and T. Kojima, J. Anal. Toxicol., 22
(1998) 396.
70 N. Nagasawa, M. Yashiki, Y. Iwasaki, K. Hara and T. Kojima, Forensic Sci. Int., 78
(1996) 95.
71 A. Ishii, T, Kumazawa, K. Watanabe, H. Hattori, O. Suzuki, Chromatographia,43
(1996) 331.
72 A. Bermejo, R. Seara, A. dos Santos Lucas, M. Tabernero, P. Fernandez and R.
Marsili, J. Anal. Toxicol., 24 (2000) 66.
73 M. Nishikawa, H. Seno, A. Ishii, O. Suzuki, T. Kumazawa, K. Watanabe and H.
Hattori, J. Chromatogr. Sci., 35 (1997) 275.
74 M. Krough, H. Grefslie and K. Rasmussen, J. Chromatogr.B 689 (1997) 357.
75 P. Okeyo, S. Rentz and N. Snow, J. High Res. Chromatogr., 20 (1997) 171.

936
76 P. Okeyo and N. Snow, J. Microcolumn Sep., 10 (1998) 551.
77 T. Kumazawa, H. Seno, X. Lee and A. Ishii, Anal. Chim. Acta, 387 (1999) 53.
78 T. Kumazawa, H.Seno, X.P Lee, A. Ishii, O. Suzuki and K. Sato, Chromatographia,
43 (1996) 393.
79 Z. Penton, Can. Soc. Forens. Sci. J., 30 (1997) 7.
80 Z. Lee, T. Kumazawa, K. Sato, H. Seno, A. Ishii and 0. Suzuki, Chromatographia,47
(1998) 593.
81 X. Lee, T. Kumazawa, T. Jurosawa, K. Akiya, Y. Akiya, S. Fruta and K. Sato, Jpn. J.
Forensic Toxicol., 16 (1998) 64.
82 X.P Lee, T. Kumazawa, K. Sato and 0. Suzuki, Chromatographia,42 (1996) 135.
83 A. Namera, M. Yashiki, N. Nagasawa, Y. Iwasaki and T. Kojima, ForensicSci. Int., 88
(1997) 125.
84 H. Seno, T. Kumazawa, A. Ishii, M. Nishika, K. Watanabe, H. Hattori and 0. Suzuki,
Jpn. J. ForensicToxicol., 14 (1996) 199.
85 H. Seno, T. Kumazawa, A. Ishii, H. Hattori, M. Nishikawa, K. Watanabe and 0.
Suzuki, Jpn. J. ForensicToxicol., 14 (1996) 30.
86 F. Prosen, K. Watanabe, A. Ishii, H. Seno and 0. Suzuki, Jpn. J. ForensicToxicol., 15
(1997) 151.
87 F. Musshoff, H. Junker and B. Madea, Clin. Chem. Lab. Med., 37 (1999) 639.
88 A. Namera, T. Watabe, Y. Yashiki, T. Kojima and T. Urabe, J. Chromatogr. Sci., 37
(1999) 77.
89 Y. Iwasaki, M. Yashiki, N. Nagasawa, T. Miyazaki and T. Kojima, Jpn. J. Forensic
Toxicol., 13 (1995) 189.
90 X.P Lee, T. Kumazawa and K. Sato, Int. J. Legal. Med., 107 (1995) 310.
91 X. Lee, T. Kumazawa, S. Furuta, T. Kurosawa, K. Akiya, I. Skiya, K. Sato, Jpn. J.
ForensicToxicol., 15 (1996) 21.
92 E. Schimmiing, K. Levsen, C. Koehme and W. Schuermann, Fresenius' J. Anal.
Chem., 363 (1999) 88.
93 R. Andreoli, P. Manini, E. Bergamaschi, A. Brustolin and A. Mutti, Chroma-
tographia,50 (1999) 167.
94 B. Dehon, L. Humbert, L. Devisme, M. Stievenart, D. Mathieu, N. Houdret and M.
L'hermitte, J. Anal. Toxicol., 24 (2000) 22.
95 T. Kumazawa, X. P Lee, K. Sato, H. Seno, A.Ishii and 0. Suzuki, Jpn. J. Forensic
Toxicol., 13 (1995) 182.
96 T. Kumazawa, K. Sato, H. Seno, A. Ishii and 0. Suzuki, Chromatographia,43 (1996)
59.
97 T. Watanabe, A. Namera, M. Yaskiki, Y. Iwasaki and T. Kojima, J. Chromatogr.B,
709 (1998) 225.
98 Y. Cai, S. Monsalud, K.G.Furton, R. Jaffe and R.D. Jones, Appl. Organomet. Chem.,
12 (1998) 565.
99 L. Dunemann, H. Hajimiragha and J. Begerow, Fresenius'J.Anal. Chem., 363 (1999)
466.
100 B. He, G. Jiang and Z. Ni, J. Anal. Atom. Spectrom. 13 (1998) 1141.
101 K. Takekawa, M. Oya, A. Kido and 0. Suzuki, Chromatographia,47 (1998) 209.
102 X. Yu, H. Yuan, T. Gorecki and J. Pawliszyn, Anal. Chem., 71 (1999) 2998.
103 C. Grote and J. Pawliszyn, Anal. Chem., 69 (1997) 587.
104 W.H.J. Vaes, E.U. Ramos, C. Hamwijk, I.V. Holsteijn, B.J. Blaauboer, W. Seinen,
H.J.M. Verhaar and J.L.M. Hermens, Chem. Res. Toxicol., 10 (1997) 1067.
105 C. Camarasu, M. Mezei and A. Szabo, Acta Pharm.Hung., 69 (1999) 77.
106 K. Kongshaug, S. Pedersen-Bjergard, M. Krogh and K. Rasmussen, Chroma-
tographia,50 (1999) 247.
107 K.G. Furton, Y.-C. Hong, Y.-L. Hsu, T. Luo, S. Rose and J. Walton, J. Chromatogr.

937
 Sci., 40 (2002) 147.
108 K. Higgins (Ed.), Proceedings of the SPIE Conference on Investigation and Forensic
Science Technologies, Boston, 1998, 3576, p. 41-46.
109 B. Hall, M. Satterfield-Doeerr, A. Parikh and J. Brodbelt, Anal. Chem., 70 (1998)
1788.
110 S. Yamamoto and H. Kataoka, J. Chromatogr.B, 707 (1998) 99.
111 S. Strano-Rossi and M. Chiarotti, J. Anal. Toxicol., 23 (1999) 7.
112 K. Miller, C. Poole, T. Pawlowski, Chromatographia,42 (1996) 639.
113 L. Ng, M. Hupe, J. Harnois and D. Moccia, J. Sci. Food Agric., 70 (1996) 380.
114 C. Bicchi, O., Panero, G. Pellegrino and A. Vanni, J. Agric. Food Chem., 45 (1997)
4680.
115 J. Clark and J. Bunch, J. Agric. Food Chem., 45 (1997) 844.
116 J.R. Almirall, A.D. Garcia, J. High Resolut. Chromatogr., in preparation
117 K.G. Furton, A.J. Sabucedo, J. Rein and W.L Hearn, J. Chromatogr., in press.

938
Chapter28

Sampling and sample preparation for

trace element speciation
Zoltan Mester and Ralph E. Sturgeon

28.1 INTRODUCTION

28.1.1 Definition of speciation

It is currently recognized that the determination of the total trace metal burden
in most biological and environmental samples provides insufficient information
about lability, i.e., bioavailability and toxicity, or for use in risk assessment
scenarios. The separation and quantification of identifiable forms of an element
constitute the field of speciation and atomic spectrometry detectors, when
coupled to an appropriate separation technique, currently play a major role in
the development of this discipline, with potential impact in areas relating to
nutrition, toxicology, geochemistry and environmental chemistry.
Speciation of trace elements provides more specific information about the
real status and impact of a given element in an environmental or biological
system. The term "chemical species" was originally used to refer to a specific
form (monoatomic or molecular) or configuration in which an element can occur,
or to a distinct group of atoms consistently present in different compounds or
matrices [1]. More recently, the International Union of Pure and Applied
Chemistry (IUPAC) has proposed definitions of terms relating to speciation [2]:
"the analytical activity of identifying and measuring the quantities of one or
more individual chemical species, where the latter are defined as specific forms
of an element defined as to molecular, complex, or nuclear structure or oxidation
state."
To understand the impact and availability of target elements in any system
requires different approaches, which comprise the extended directions of
speciation analysis:
(i) Determination of the exact chemical form of a given element includes not
only its oxidation state, but also the structure of the molecule, if the metal is
covalently bonded. It is well known that the different chemical forms of an
element may exhibit significant differences in toxicity, mobility or reaction
kinetics.

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 939
 (ii) Determination of the distribution of metallic compounds in a real or
pseudo-multiphase system. For example, the toxic trace metals in aqueous
environmental systems containing natural complexing agents such humic and
fulvic acids are partially associated with these complexing agents, with the
consequence that the free metal content, which determines the biological effect,
can be seriously altered. There is little doubt that studies devoted to this will
become one of the fastest growing areas of speciation analysis because this type
of equilibrium/distribution problem arises not only in the fields of the environ-
ment, but also the realm of bio-inorganic chemistry [3-9].
(iii) Determination of the bioavailability of metals arising from solid
materials (such as sediments, soils, fly ash, bottom ash) through solubilization
with various "soft" aqueous leaching agents in an effort to model metal release
in different environmental situations. For example, extracting fly ash samples
with weakly acidic solutions may simulate the effect of acid rain; extraction of
soils with different complexing agents, such as EDTA, may simulate the effect of
the presence of naturally occurring complexing substances in soil systems.
The need for speciation, particularly with respect to point (i), has had a
significant impact in two areas of classical elemental analysis. The first is sample
preparation, wherein preservation of the sample's integrity requires use of
"soft" extraction techniques and, when gas chromatography (GC) analysis is to
be done, transfer of the organometallic compounds from the aqueous phase to a
nonpolar organic phase. The second aspect encompasses the necessity to apply
separation techniques, such as liquid chromatography (LC) [10,11], gas chroma-
tography [12] and capillary electrophoresis (CE) [13] prior to element selective
detection based on optical and mass spectroscopic sources such as inductively
coupled plasmas, microwave induced plasmas, glow discharge and flame photo-
metric and ionization sources. Hundreds of research articles have been
published on the subject of speciation and three recent books, edited by Caroli
[14], Ure [15] and Caruso [16], have been entirely devoted to the subject. The
coupling of atomic spectroscopic detectors with separation systems has been
reviewed in a recent special issue of the Journal of Analytical Atomic Spectrome-
try [17]. It is the purpose of this contribution to focus on the various aspects
relating to sample preparation for subsequent speciation analysis.

28.1.2 Reference materials for speciation analysis

A reference material (RM) by definition, is a material or substance, one or more

of whose property values are sufficiently homogeneous and well established to be
used for the calibration of an apparatus, the assessment of a measurement
method, or for assigning values to materials. A Certified Reference Material
(CRM), is a RM accompanied by a certificate, one or more of whose property
values are certified by a procedure which establishes its traceability to an
accurate realization of the unit in which the property values are expressed, and
for which each certified value is accompanied by an uncertainty at a stated level

940
of confidence [18]. Chemical measurement traceability is the property of the
result of a measurement, either physical or chemical, or the value of a standard
whereby it can be related to stated references, usually national or international
standards, through an unbroken chain of comparisons [18]. The use of CRMs is
necessary for all analytical laboratories undertaking methods development to
ensure the validity of a proposed new approach. During the past decade, a
number of CRMs have become available for methylmercury, trimethyllead [19],
organoarsenic [20-22], alkyltin [23,24] and chromium oxidation states [25].
Table 28.1 summarizes currently available CRMs and their sources for organo-
metallic species. Most reference materials are based on matrices of environmen-
tal interest, mirroring the increase in environmental security issues which arose
during the 1980s and 1990s. Only very few biological/clinical CRMs are available
for trace element speciation; the biotech revolution will likely induce production
of a number of health related speciation standards in the future.
An important issue for the speciation field is the provision of fully character-
ized CRMs, wherein not only the amount of a given species has been defined, but
the total content of the element along with information values for macro (N, C,
0) and micro constituents (trace elements) is also available. This is important
from the viewpoint of methods development, to know precisely the nature of the
material in question, and what type of interferences may be encountered. This
approach has been adopted by the National Institute of Standards and Technol-
ogy (NIST) and the National Research Council of Canada (NRCC).
Several articles [26-29], and recently an entire book [29], have been devoted
to questions associated with the production of reference materials for speciation
purposes.

28.1.3 Instrumentation

Typical instrumentation employed for trace element speciation comprises the

coupling of a (chromatographic) separation technique with a molecular or
element specific detector. The principal technical difficulties arising from the
combination of chromatographic and atomic spectroscopic detectors include the
interfacing of the two systems and the long data collection times required for the
characterization of (numerous) transient signals comprising the chromato-
grams. Most atomic spectroscopic detectors have been designed primarily for
steady-state sample introduction [30]. Many of the above problems have been
resolved by the detector manufacturers by integrating chromatographic soft-
ware packages into ICP-MS operating systems; one ICP-MS producer also offers
a GC-ICP interface.

28.1.3.1 Gas chromatography

A significant portion of speciation directed measurements has been achieved
using gas chromatographic separation. The attractive features of GC include the
extremely high resolution, high sample introduction efficiency into an atomic

941
TABLE 28.1
Commercially available CRMs for organometallic speciation
Arsenic Arsenobetaine: Solution BCR CRM 626 1031 tmol/kg
Tuna fish tissue BCR CRM 627 52 mol/kg
Dogfish muscle NRC CRM DORM-2 16.4 mg/kg
Dimethylarsenic acid: Tuna fish tissue BCR CRM 627 2 gmol/kg
Tetramethylarsonium: Dogfish Muscle NRC CRM DORM-2 0.248 mg/kg

Tin Dibutyl: Mussel tissue BCR CRM 477 1.54 mg/kg

Marine sediment NRCC CRM PACS-2 1.09 mg/kg
Coastal sediment BCR CRM 462 0.068 mg/kg
Monobutyl: Mussel tissue BCR CRM 477 1.5 mg/kg
Marine sediment NRCC CRM 0.45 mg/kg
Tributyl: Mussel tissue BCR CRM 477 2.2 mg/kg
Fish tissue NIES CRM 11 1.3 mg/kg
Marine sediment NRCC CRM PACS-2 0.98 mg/kg
Coastal sediment BCR CRM 462 0.054 mg/kg
Triphenyl: Fish tissue NIES CRM 11 6.3 mg/kg

Lead Trimethyl: Urban dust BCR CRM 605 0.0079 mg/kg

Mercury Methyl: Human hair IAEA 085 22.9 mg/kg

Dogfish mussel NRCC CRM DORM-2 4.47 mg/kg
Tuna fish BCR CRM 464 5.5 mg/kg
Human hair NIES CRM 13 3.8 mg/kg
Tuna fish BCR CRM 463 3.04 mg/kg
Dogfish liver NRCC DOLT-2 0.693 mg/kg
Cod mussel BCR CRM 422 0.43 mg/kg
Human hair IAEA 086 0.258 mg/kg
Lobster hepatopancreas NRCC CRM 0.152 mg/kg
TORT-2
Mussel tissue NIST SRM 1974a 0.0772 mg/kg
Mussel tissue NIST SRM 2974 0.0772 mg/kg
Non-defatted lobster hepatopancreas 0.063 mg/kg
NRCC CRM LUTS-1
Mussel tissue NIST SRM 2976 0.0278 mg/kg
Estuarine sediment BCR CRM 580 0.0075 mg/kg
Fucus (sea plant) IAEA-140/TM 0.000626
mg/kg

Chromium Cr(III): Lyophilised solution BCR 544 26.8 g/l

Cr(VI): Lyophilised solution BCR 544 22.8 ,g/l
Welding dust loaded on a filter BCR 545 40.2 g/kg
BCR, Community Bureau of Reference; IAEA, International Atomic Energy Agency; NIES,
National Institute for Environmental Studies; NIST, National Institute of Standards and
Technology; NRCC, National Research Council of Canada.

942
spectroscopic detector, and very low background due to the high purity inert gas
used as the mobile phase. Typical drawbacks of such approaches are that most of
the target organometallic compounds are usually not directly suited to GC
separation because of their thermal instability, high reactivity or ionic nature in
the environment. Therefore, it is necessary to apply some form of derivatization
prior to analysis. Several reviews have been published on the application of GC
for speciation analysis [12,31,32].
28.1.3.2 HPLC and CE
As most of the species of interest are ionic, polar compounds and are generally
found in the aqueous environment, liquid chromatography and capillary electro-
phoresis are a natural choice for separation. The matrix load on the atomic or
ionic source used for detection can present a significant limiting factor. Typical
LC flow-rates (1-2 ml/min) can be introduced only in conjunction with low
efficiency nebulization to split matrix load about 100-fold. The situation is only
worsened if the eluent systems contain organic solvents, such as methanol or
acetonitrile, because the tolerance of plasma- or flame-based detection systems
is much lower for these solvents than for water. As a result, detection limits for
HPLC- or CE-based methods are several orders of magnitude worse than GC-
based methods. Miniaturization of the aqueous phase separation system will
likely overcome many of the limitations presented by the mobile phase. Use of
aqueous phase separation is very popular for arsenic [33], selenium [34-36],
chromium [37,38] and platinum [39,40] speciation, partly because there are no
readily available gas chromatographic derivatization methods for these species.
Condensed phase separation is gaining momentum in the speciation field, also
due to the increasing interest in metal containing biopolymers such as proteins
and DNA [9,41-46].
28.1.3.3 Detection
The very low detection limits required to quantitate trace element species relies,
in most cases, on powerful atomic spectroscopic detectors. Atomic absorption
[47], atomic fluorescence [48-51], atomic emission with inductively coupled
plasma (ICP) sources [52-54], glow-discharges [55,56] or microwave induced
plasma (MIP) sources and, recently, electrospray ionization mass spectrometry
[57], are most commonly used. Additionally, several of the conventional gas
chromatographic detectors are finding direct application, principally the flame,
pulsed flame photometric and ionization detectors [58]. However, the detection
system of choice in most cases is currently ICP-MS [59-65]. These systems
utilize an argon plasma, which serves as the elemental ion source for the mass
spectrometer. The identification of the analyte is based entirely on chromato-
graphic retention times. Since no molecular information is available from
conventional plasma sources, structural information is lost and identification
becomes impossible. Recent work with multidimensional instrumentation has
helped to rectify this problem [17]. A combination of liquid chromatography with
electrospray/ionspray sources can be used for measurement of fragment ion

943
patterns, generating (potentially near simultaneous) structural information
which can be used to deduce species identity in the absence of a standard.
Additionally, "dual mode" analysis, achieved by judicious selection of operating
conditions in the atmospheric pressure ionization source interface, permits
elemental atomic and molecular information to be obtained on the same sample.
"Soft" ionization sources can also provide for a dual mode of operation to permit
atomic and molecular information to be achieved with glow discharges [66,67],
furnace atomization plasma ionization mass spectrometry (FAPIMS) [68], low
pressure ICPs [69] and low power microwave induced plasmas (MIPs) [70].

28.1.4 Sampling for speciation analysis

Knowledge accumulated in the speciation field relating to sampling and storage

conditions used to ensure sample integrity focus primarily on standard solutions
and highly processed reference materials [71]. Only limited experience is avail-
able on the sampling and storage of real environmental or biological samples.
Ariza et al. [71] have reviewed current knowledge on storage and stability of
organometallic compounds in standards and in various environmental matrices.
Storage conditions and stability of organotin species have been studied in
freshwater [72], organotin in environmental samples [73], arsenic in urine [74],
selenium in urine [75] and in environmental samples [76], volatile species from
ambient air [77], arsenic III and V in water [78], selenium standard solutions
[79], organolead in environmental matrices [80] and mercury [81,82]. Generally,
adequate sampling and storage conditions for organometallic species are
common to those applied in the organic analytical field. The general approach is
to prevent or limit any chemical and biological degradation processes from
occurring in the sample. Storage of samples at low temperatures (+4 to -80°C)
and in the dark to prevent thermal and photocatalytic decomposition is most
frequently adopted. Storage containers should carefully be selected to avoid any
loses of analyte due to wall adsorption. Numerous species, mainly due their
hydrophobic character, require special inert containers.

28.2 EXTRACTION/LEACHING TECHNIQUES

The ideal analytical instrument would, without human intervention, perform all
analytical steps, from sampling to data analysis, and even undertake decision
making based on the results obtained. Although today's sophisticated instru-
ments can analyze complex samples and evaluate results, most sampling and
sample preparation practices are based on 19th century technologies, such as the
common Soxhlet extraction method. Traditional sample preparation methods
are typically time consuming, employ multi-step procedures having high risk for
loss of analytes, and use extensive amounts of organic solvents. These character-
istics make such methods very difficult to automate and integrate into modern
sampling/separation systems. Consequently, most of the analysis time is

944
consumed with sampling and sample preparation, which contribute significantly
to the cost of analysis.
Extensive use of organic solvents in analytical laboratories is no longer
tolerated because of the associated health risks and disposal concerns. As a
result, several different extraction approaches have been studied in the past
decade aimed at enhancing the extraction efficiency. The major goals for
improvement include reducing solvent usage, decreasing the extraction time,
decreasing the amount of sample required for analysis, and automation of the
extraction procedure, preferably online with separation and detection. Historic-
ally, the first extraction methods used were conventional solvent extraction
approaches. In the last ten years, however, a significant scientific effort has been
made to enhance the efficiency of classical solvent leaching procedures by using
microwave energy, ultrasonic waves, supercritical fluids and pressurized
solvents. Table 28.2 summarizes some typical working conditions for these new
extraction methods compared to the conventional Soxhlet process. A brief
review of these techniques is presented below.

28.2.1 Accelerated solvent extraction

Accelerated solvent extraction (ASE) is a static solvent extraction method. It

uses a combination of high pressure (1.0x 10 4-1.4x 10 4 kPa) and temperature
(50-200°C) to increase the efficiency of the extraction. The increasing popularity
of this approach is due to the low solvent volume usage and the automation capa-
bilities of the system which provide good control over the entire procedure [83].
In practice, the sample is enclosed in a stainless steel vessel filled with
solvent, which is pressurized and heated. The sample is typically allowed to
statically extract for 5-10 min, with the expanding solvent vented to a collection
vial. Next, compressed nitrogen purges the remaining solvent into the same vial.
The entire procedure typically requires less than 15 min and consumes approxi-
mately 10-15 ml of solvent for a 5-30 g sample. Accelerated solvent extraction
has, as its basis, the fact that the solubility of the analyte increases at tempera-
tures above the boiling points of commonly used solvents. Naturally, at high
temperature, the desorption kinetics are also improved. Solvent usage is also
reduced as a consequence of the higher analyte solubility in the heated solvent.
Elevated pressure serves to keep solvents in their condensed state, and facilitate
the mass transfer of analyte from the micro-structure of the solid sample [83].
To date, only a handful of applications has been reported on the speciation of
arsenic [84-86] and tin [87] using ASE.

28.2.2 Microwave assisted extraction

Several books [88-90] and review articles [91-94] have been published on the
theoretical and practical approaches to the use of microwave chemistry and
microwave assisted extraction. The microwave source is a non-ionizing radiation

945
 -els~~ 5" ~~~~~~~10.
.5'
o ~ gt
=0 ao 0
C) -
S
O 0£

"o ~
.".B ~ IE $5 0C)00~ed
0
vF a: 0"1 m
ae P - i=W.= 0 cl . z
4C ) ..

5 ®
o 0
Cd
0 o

0"0 11~~.
aO) C-
0 0"~~~~ o 4°
C)
'o C)

w
F
VA
0 M:
L Si .

o0 d
001 m oz
'ji.Q e E e iL i , 2-
a000
aC) ,bX

, 0
a4) i
0 O ~~ ~ ~ ~ t .
a)4
'd a: .)
o 0
.00 0" At
co 0 9u
C)

:t
01)
kF - w E r0BC- )OC)0 e

R la
C -- 1<
.-
,0
t

Oa
0 .
CID 0
00
a
U, O5
. o

.,

*.-
0. I
0
10
-a
¢ 01cv

a0

946
TABLE 28.3
Selected physical parameters for some of the most commonly used solvents. (All parameters
obtained from Refs. [89] and [94].)
Solvent Dielectric Dipole Dissipation Boiling point' Closed vessel
constant (e') momentb factor (tan 8) (°C) temperatured (°C)
-
(x 10 4)

Acetone 20.7 56 164

Acetonitrile 37.5 82 194
Ethanol 24.5 1.96 2500 78 164
Hexane 1.89 69 no microwave heating
Water 78.3 2.3 1570 100
aDetermined at 20°C; bdetermined at 25°C; Cdetermined at 101.4 kPa; ddetermined at 1207 kPa.

field in the frequency range 300-300,000 MHz that induces motion of molecules
by ion migration and rotation of dipoles. Dipole rotation can occur not only in the
selected solvent, but also in the sample itself. The alignment and relaxation of
the molecular dipoles in the microwave field occurs 109 times per second
resulting in extremely quick heating. Commercial microwave extraction devices
operate at 2.45 GHz.
The ability of a solvent to absorb microwave energy and pass it on to other
molecules in the form of heat will partly depend on the dissipation factor (tan 6).
The dissipation factor is given by the following equation [88]:
tan 6 = E"/c' (28.1)
where £" is the dielectric loss (a measure of the efficiency of converting micro-
wave energy into heat) and ' is the dielectric constant (a measure of the
polarizibility of a molecule in an electric field). Polar molecules and ionic
solutions will strongly absorb microwave energy because they have a permanent
dipole moment that will be affected by the microwaves. However, non-polar
solvents, such as hexane, will not heat up when exposed to microwaves. Table
28.3 presents selected physical parameters, including dielectric constants and
dissipation factors, for solvents that are used in most applications [94].
Because of the high dielectric constant of water, the free water content of the
sample induces a very rapid and concentrated heating during the microwave
extraction process. The resulting pressure increase may result in the rupture of
membranes in biological samples, enabling various constituents to be released. A
similar mechanism is suspected to occur in sediments and soils. Microwave heat-
ing of the water in the matrix leads to boiling, bubble formation and increased
pressure, resulting in disintegration of the microstructure of the matrix, thereby
increasing the available surface for extraction by the solvent [91].
Microwave assisted extraction can be performed either with closed vessels
(under controlled pressure and temperature), or with open vessels (at atmos-
pheric pressure). Both systems are illustrated schematically in Fig. 28.1.

947
 Diffused microwave system

I '/- -A
-I
I hnrntablc,

Focused microwave system

MLtgnetrol ter

I vessel

Solvent
Iocsed | t2··

I QAlm uc .... Ca'es WMOi-- Sample

Fig. 28.1. Closed and open vessel microwave units. Reprinted from Trends in Analytical
Chemistry, Vol. 19, V. Camel: Microwave-assisted solvent extraction of environmental samples,
pp. 229, Copyright (2000), with permission from Elsevier Science.

In closed vessels, the solvent can be heated above its standard boiling point,
thus enhancing both extraction speed and efficiency. Such systems permit tem-
perature control of the extraction process. In addition, they usually lead to an
increase in sample throughput because several vessels can used simultaneously.
Because of the non-homogeneous electric field in the microwave cavity, sample
vessels are usually placed on a turntable. In so-called open systems, extractions
proceed under atmospheric pressure. As a consequence, the maximum possible
temperature is determined by the boiling point of the solvent at ambient pres-
sure. Such systems use focused microwaves, so that the heating of the sample is
homogeneous and very efficient. Vapour loss is prevented by the presence of a
reflux system fitted to the top of the extraction vessel. Organometallic com-
pounds are typically extracted using such open vessel systems, as the close con-
trol of the energy delivered to the sample prevents destruction of carbon-metal
bonds [95].
Microwave assisted extraction has been popular for leaching of organotin
from biological and environmental samples such as fish-tissue and sediment
[95-101]. The polar nature of the butyltin species permits it to easily absorb
microwave energy, thereby facilitating solubilization of the species and
minimizing the problem of re-adsorption. Microwave assisted extraction has also
been used for efficient recovery of methylmercury [102-107].

948
28.2.3 Supercritical fluid extraction

The properties of fluids under supercritical conditions, i.e., between those of liq-
uids and gases, render salvation power (similar to liquids), while the low viscos-
ity and high diffusivity (similar to gases) facilitate transport phenomena, thus
making extraction faster and more effective. The most common SF extractant
used at present is CO2 because of its low toxicity and inflammability, reasonable
critical conditions and chemical inertness. The most significant aspect of SC-
CO 2 is its weak interaction with both analytes and matrices which, when pure,
provide poor efficiencies in the extraction of environmentally persistent pollut-
ants. This problem can be overcome either by the addition of co-solvents (metha-
nol, pentane, toluene) to increase the polarity of the solvent system, or by
reducing the polarity of the target analytes by complexation or ion pair forma-
tion, esterification and reverse-micelle formation. The advantages of SFE vs
both Soxhlet and other conventional leaching techniques include: reduced
extraction time, a compatibility for on-line coupling to either detectors or
chromatographs, reduced need for cleanup due to the high selectivity achieved
through manipulation of both pressure and temperature and, most significantly,
elimination of solvent removal steps as the extractant is released from the
leached species after depressurization. Trapping techniques employed for collec-
tion of analyte after extraction include: liquid collection, cryogenic trapping and
solid-phase trapping, the latter being the most effective as it allows for simulta-
neous collection, cleanup and concentration prior to either individual chromato-
graphic separation or direct detection. Despite the advantages of SFE and the
wide variety of commercially available apparatus and instruments, which have
stimulated the development of new applications, this technique has so far not
fulfilled the expectations of suppliers and users owing to the following: (a) the
large discrepancies in extraction efficiency between spiked and endogenous
analytes in samples (b) the number of methods reported in the literature for
which the efficiency is lower than that provided by classical methods; (c) the lack
of information about both the means of overcoming analyte-matrix interactions
and the selection of the most appropriate modifier.
Recently, an excellent review article by Bayona [108] has appeared outlining
SFE applications in the speciation field. Most compounds of interest exhibit high
polarity and/or are of an ionic nature wherein direct extraction by non-polar
SF-CO2 is not, in most cases, a suitable approach. As noted earlier, to overcome
the issue of high polarity of the analyte, complexation/derivatization may be
undertaken or the polarity of the extraction phase may be increased using
matrix modifiers. Both approaches have been examined in a number of studies.
The typical approach used for butyltin leaching involves application of metha-
nol/acetic acid/formic acid/HCl modifiers. The use of complexing agents, such as
ammonium pyrrolidine dithiocarbamate (APDC) in the presence of the above-
mentioned polar modifiers, does not improve the extraction efficiency. Most
likely, the resulting complexes are not stable under the supercritical conditions

949
[109]. Grignard derivatization has also been used in an attempt to convert ionic
compounds into non polar species which may be more soluble in supercritical
carbon dioxide [110]. Several combinations of modifiers and different matrices
have been described for the extraction of organotin species [108,109,111-116].
Supercritical fluid CO 2 has also been used for the extraction of di- and trialkyl-
lead species. Determination of alkyllead following SFE was successfully used in a
CRM certification campaign, although the material, an urban dust sample, was
spiked, which may have conferred a simplified leaching process [19]. SFE has
also been applied for methylmercury extraction [105,117,118-1211.

28.2.4 Sonication

Ultrasonic extraction is based on the enhancement of mass exchange in the

pores of a solid phase when exposed to ultrasound, wherein repeated deforma-
tion of the material particle skeleton occurs along with generation of convection
cells. Ultrasonic radiation is created within a liquid by means of transducers,
which convert electrical energy into acoustic waves. The transducers, consisting
of vibrating elements tuned to a specific frequency, are bonded to the side or
underside of a tank containing the liquid or are encased in stainless steel for
immersion within a liquid. Ultrasonic cleaning equipment is available in three
primary operating frequencies: 25, 40 and 80 kHz.
The mechanism of extraction by sonication is an effect created by the action of
sound waves at high frequency introduced into a liquid cleaning medium. This
action, known as cavitation, consists of the formation and instantaneous collapse of
millions of tiny cavities, or bubbles in the liquid. These bubbles occur throughout
the liquid, even in hidden recesses and crevices. Cavitation is produced by the alter-
nating patterns of compression and rarefaction generated by the rapidly expanding
and contracting transducers during sound wave transmission. As the liquid is
stretched beyond its tensile strength during rarefaction, these bubbles grow from
microscopic nuclei and then, upon compression, they implode violently. This phe-
nomenon occurs at a rate proportional to the ultrasonic frequency used. Individ-
ually, these minute bubbles release only an extremely small amount of energy;
however their cumulative effect is intense [122]. Sonication is widely used as a sim-
ple and cost-effective means of intensifying various leaching procedures applied to
elemental speciation and it has been used for the extraction of organotin species
[123-125], organoarsenic [86,126-128] and methylmercury [129].

28.3 SAMPLING AND TRACE ENRICHMENT METHODS FOR

VOLATILE SPECIES

Enrichment techniques become necessary if the analyte concentration in the

headspace gas above the sample is below detectability. For this purpose, the
target analytes must be separated from the headspace gas by absorption into a
liquid, by adsorption onto a solid support or by condensation in a cold trap.

950
Solvent-free techniques are particularly desirable for trace analyses so as to
avoid the cumbersome problems associated with solvent impurities. However, a
significant portion of the work related to speciation employs solvent trapping/
preconcentration. This normally involves trapping the analyte in an organic
solvent. If the volatility of the analyte is significantly lower than that of the
trapping solvent, further preconcentration can be achieved by evaporation of the
solvent. Organic solvent extraction/trapping is typically used in combination
with Grignard [130] derivatization or ethylation with sodium tetraethylborate
(NaBEt 4) [131].

28.3.1 Solid phase microextraction

The theory and practice of SPME has been presented elsewhere in this book and
can also be found in a recent text [132]. Therefore, the focus here is on some
typical elemental speciation applications of this technique. Mester et al. [133]
recently reviewed the determination of organometallic species using SPME
sampling with gas chromatographic separation/detection. Rapid advances in this
field have served to expand the scope of applications, interface development,
instrument technique and fiber preparation.
Most elemental speciation applications of SPME are based on their coupling
with classical gas chromatographic instrumentation. Methods have been
described for mercury, tin, lead, arsenic and selenium. Both headspace and
direct extraction methods have been used for the sampling of organometallic
compounds and, generally, some type of derivatization process is necessary for
their GC separation. The typical derivatization method used for tin, lead and
mercury species is ethylation using NaBEt 4 reagent. Moens et al. [134] reported
a comparison of sensitivity between 'conventional' liquid/liquid- and headspace
SPME-extraction of butyltin and organolead compounds. They found that head-
space SPME provided an approximately 300-fold better sensitivity for butyltin
compounds and about a 35-fold enhancement for trimethyllead. As a result,
analysis of butyltin species based on SPME can thus be accomplished using a
conventional Flame Ionization Detector (FID) instead of the more expensive
mass spectrometric systems.
Several researchers have reported results on the headspace sampling/deter-
mination of metal hydrides using solid phase microextraction techniques. Jiang
and coworkers [135,136] developed a sampling method for organomercury
species based on hydride generation using KBH 4 reagent. Mester et al. [137]
tested the compatibility of two different fibers (and two different extraction
phenomena) for sampling volatile metal hydrides coupled with ICP-MS detec-
tion. An adsorption-based carboxen coating provided better sensitivity than an
absorption-based extraction with a liquid-type polydimethylsiloxane polymeric
coating (PDMS). The success of absorptive sampling confirmed the relatively
high stability of the studied metal hydrides (As, Se, Sn and Sb) because they
survive diffusion into and release from the polymeric liquid.

951
 Guidotti et al. [138,139] described the determination of Se(IV) by headspace
SPME-GC/MS following derivatization based on either piaselenol formation or
ethylation. The method can also be applied to Se(IV) and Se(VI) speciation by
determining the Se(IV) content first, followed by the total selenium concentra-
tion after converting all selenium species to the Se(IV) oxidation state.
Mester and Pawliszyn [140] reported a method for the speciation of di-
methylarsinic acid (DMA) and monomethylarsonic acid (MMA) by SPME-
GC/MS. Thioglycolmethylate (TGM) was used for derivatization, having as its
basis the known "affinity" of the arsenic compound for the thiol-group [30,141].
Organometallic species which are normally saturated (non-ionic) and suffi-
ciently volatile can be sampled by SPME and determined by GC without deriva-
tization. Gorecki and Pawliszyn [142] described a simple sampling procedure for
tetraethyllead. A GC-MS was used for quantitation. Similar studies were per-
formed by Snell et al. [143] for determination of dimethylmercury in the head-
space of natural gas condensates. This method was characterized by a very short
sampling time (30 s). The extremely complex matrix, comprising volatile organic
material (which is also extracted with the dimethylmercury), presented a serious
problem even for the extremely selective microwave induced plasma atomic
emission detection system used.
Mester et al. [144] recently described a SPME method for methylmercury
determination based on the relatively high vapour pressure of methylmercury
chloride. Headspace SPME sampling was performed above a methylmercury
solution which was previously saturated with sodium chloride. A slightly polar
solid coating (PDMS/DVB) was used for extraction. Sample introduction into an
ICP-MS was achieved with a unique thermal desorption interface consisting of a
heated glass-lined splitless type GC injector placed directly at the base of the
torch to minimize the length of transfer-line. This arrangement provided for fast
desorption and high sample introduction efficiency. A schematic diagram of the
thermal desorption ICP-MS interface is shown in Fig. 28.2. This unit provides a
very versatile interface, which can be connected to virtually any atomic spectro-
scopic source.
SPME can also be used for sampling of non-volatile/ionic metal species from
aqueous media. Jia et al. [145] studied the extraction of mercury species from
INDUCTIVELY COUPLED
PLASMA MASS SPECTROMETER

Heated and insulated 2 mm ID

desorption unit
Swagelok "T"

Fig. 28.2. Schematic of the

SPME-TD-ICP-MS system. gas- Glassinsert Ausitiarygasow

952
aqueous solutions. The SPME fiber was modified for sampling in that different
crown ethers were adsorbed onto the non-polar SPME coating by simply dipping
the fiber into a solution of the crown ether. An HPLC system was subsequently
employed for the separation of free complexing agent from the metal complex.
Because there are no commercially available ion-exchange or specific metal
selective coatings available, the study of new coating materials and extraction
principles is of interest. Caruso and coworkers [146] recently reported on
development of SPME sol-gel coatings for organometal determination. Sol-gel
coatings are very mechanically, thermally and chemically stable in organic
solvents as well as acidic and basic solutions where commercial SPME coatings
cannot be used. The fibers were evaluated for SPME-HPLC applications using
diphenylmercury, trimethylphenyltin and triphenylarsine. The effect of organic
solvents and acids on the fibers was also studied.
In a recent paper, Wu et al. [147] described the application of a polypyrrole
polymer coating for SPME extraction of small anionic analytes such as C1-, F-,
Br, NO,- P04 3 , S042-, SeO4 2- , SeO32- and AsO43- . Extraction was based on the ion
exchange characteristics of the polypyrrole coating. The same coating has been
used for sampling arsenobetaine from aqueous samples [126].
A different approach to the use of liquid phase microextraction is to substi-
tute an open tubular capillary for the sample introduction loop in an HPLC
system. Such an in-tube SPME system has been described for the extraction of
trimethyllead and triethyllead [148,149] arsenobetaine [126], and tributyltin
[125] from biological samples.

28.3.2 Cryogenic trapping

Cold traps are used for two main reasons: enrichment purposes and solute band
concentration. Most cold traps are home-made and are operated manually,
requiring skill and experience of the operator. In keeping with the general trend
in analytical instrumentation, however, any technique will be successful only if
it can be operated automatically and unattended.
A distinction has to be made between 'cryogenic condensation' and 'cryogenic
focusing', although the common term 'cryogenic trapping' is used for both. The
term cryogenic condensation is used here for techniques in which the volatile com-
pounds are trapped simply by condensation in traps, which usually contain no sta-
tionary phase. Condensation is also the prevailing mechanism if a coated capillary
is cooled down to such a low temperature that the liquid phase will solidify and
lose its properties as a chromatographic phase. The term cryogenic focusing is
used if the volatile compounds are trapped in the liquid phase of a column at a low
temperature which, however, is still above its glass transition temperature. Dur-
ing sample introduction, the compounds dissolve in the liquid phase and immedi-
ately but slowly migrate downstream on the cold column.
In his recent article, Kolb [150] reviewed headspace sampling methods,
including cryotrapping: among many others, it is an excellent starting point for

953
headspace sampling techniques. Donard's research group has pioneered the
adaptation of cryotrapping technology for speciation studies [102,151-163].
Cryotrapping is widely used in the speciation field, especially for precon-
centration of very volatile metal(oid) species following hydride generation/
ethylation derivatization. The typical trapping temperature is in the -150 to
-196°C range obtained with liquid nitrogen cooling. Cryotrapping can also be
used as a means of sampling volatile metal(oid) species in the environment. The
trapped volatile species can be introduced into a separation system (typically
GC) after a fast heat-up, or the cryotrap can be applied as a 'crude' separation
device where, with gradual heating, the analytes leave the trap according to their
boiling points.
Cryogenic sample collection has been used for the determination of organotin
species after hydride generation [164-167], ethylation with sodium tetraethyl-
borate [168] and for the sampling of saturated alkyltin species from sea water and
air [161]. It has also been extensively applied in speciation studies of arsenic
[169-174] and mercury [151,154,155,158,163,175-178]. Volatile metal(oid) spe-
cies have been detected in air [77,156], water [153,157,161,162] and in landfill gas
samples [179,180] using cryogenic sampling. Most of these studies employed
detectors having multielement capabilities (manly ICP-MS) for simultaneous
monitoring of the volatile species of different elements, such as arsenic, antimony,
tin, selenium and mercury. Volatile metal carbonyl species (Ni(CO) 4 Mo(CO)6 and
W(CO) 6) have been detected in fermentation gases from a municipal sewage treat-
ment plant using cryogenic gas sampling/trapping [181,182].
One of the main advantages of cryogenic trapping compared to all other
preconcentration methods is that unstable/reactive volatile organometallic
species do not come into contact with any liquid or solid sorbent material,
significantly reducing the possibility of alteration/decomposition of analyte and
also reducing the possibility of introducing contaminants via the solvent used in
other techniques. At same time, cryotrapping technology, especially for field
application-despite its clear advantages-remains not very appealing because
of its bulkiness and the requirement for liquid nitrogen.

28.4 DERIVATIZATION METHODS

Following the sample leaching procedure described in Section 28.2, organo-

metallic compounds can, in some cases, be directly determined by HPLC. How-
ever, for GC analysis, it is necessary to transform non- or semi-volatile
organometallic compounds into thermally stable volatile compounds. This
transformation or 'derivatization' process will be briefly reviewed in the next
section. Table 28.4 summarizes the different derivatization approaches used for
elemental speciation and Fig. 28.3 shows the periodic table, wherein individual
elements are marked according their capability of forming volatile chloride,
ethyl, or hydride species. Generation of carbonyl species requires extreme condi-
tions and is thus not considered further.

954
 ,o SS
o

o, 3
0

= ;-8
' a a) --4 .-
aa)
o
S
C- .a_
0
0 Cr,
- 0Cr5
' 6 S
-S 8¼
Cr~

S~*S o 6
G1E
~ .2 ,2
o.2
Cr m
o. .B . -° at
*5r
M
-C
O .

~
o ~, O
S.®

C
.2
w
o,
0 a)2 o O~r t
,o a .s .eCoa0._ O
~
at C)

-I 2 2
C
0
oC a)
9s
C)

; ~ _O ..oO
ata

r -6 ~ ~ O X
0 0
S o
o '
E
>4
~'~ ~o
R r
I
52 a~-l
Ca 7
a)
a) Op'at C-
.2 52aCr-g
-0 baa-' e Mf:,Qi
a) oa
9 S
ag Cr a) Ca C.C
a)
aC a_a)
C 2Cr 5 a))

O-S
C) o Od. ,.o 3
o. - . 255
o, o)C.
S
a m a C
Cr- Sm o, a>~ a 3 )

7H
m s
0 D
0 a
*2 C,
Cr
C
c0 a P r
r a) 2
D
6 .0 o
a Cr
Sd >4 i S tE- 5
Cr
0 ~o
Cr
Ca)
D

955
 [A 1IA VIllA
-

H He

IiA IVA VA
"' VIA
-" v1

Li Be B C N 0 F Ne

A
Na M, Al osi P S Cl Ar

HIB IVB VB VIB VIIB Vill lB lie I, A

K Ca Sc Ti V Cr Mn F Co Ni Cu Zn Ga OGe K0
OA
A. Se Br Kr

Ai A A IA A A
iO
Rb Sr .Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In . A
Sn Sb Xe

] A A A- A A A A
Cs Ba La Hf Ta W Re Os Ir Pt Au Hg
A DA
Pb F'
Pi,
At Rn

* A A A ** A
Fr Ra Ac

Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No L.

Fig. 28.3. Periodic Table of the elements indicating elements which can be volatilized at
atmospheric pressure and room temperature as: O, ethylated species (with sodium tetra-
ethylborate); A, hydride with conventional hydride generation (sodium tetrahydroborate); A,
newly discovered hydride forming elements; *, carbonyl formation (using carbon monoxide); 0,
halide generation. *Osmium tetraoxide. **Elemental vapour (Hg° and Cd°). Molybdenum and
tungsten carbonyl species have been detected in the environment [182].

28.4.1 Hydride generation

Hydride generation (HG) is one of the most commonly used techniques in

inorganic analytical chemistry for the volatilization of trace elements. In this
process, tetrahydroborate is typically used because of its reducing and hydride
transfer properties. The reaction of borohydride in aqueous media can be
described as:
BH + 3H 2 0 + H+ - H3 BO3 + 8H (28.2)
+
M' + (m+n)H -- MH n +mH (28.3)
where M is the analyte and m and n are the oxidation state of the analyte [183].
Decomposition of the tetrahydroborate requires only milliseconds (Eq. (28.2))
and the nascent hydrogen produced reacts with metal or metalloid ions,
reducing them to metal hydrides. The resulting hydride species are generally
very volatile and thus suitable for gas chromatographic analysis.

956
 During hydride generation, alkyl group(s) bonded to the metall(oid) center
remain unchanged, providing an excellent means of differentiating between
alkyl and inorganic species. Hydride generation combined with GC has been
used for the determination of inorganic, mono and dimethylarsenic [184,185].
Because of the specific requirements regarding the acidity of the sample matrix
during hydride generation of different species, the process can be fine-tuned to
selectively derivatize certain species [186-188]. Generally, the hydride genera-
tion process is rapid, shows high efficiency, and is inexpensive. On the down side,
the gaseous products are difficult to transfer to a GC and the efficiency of
hydride generation can be heavily influenced by the presence of transition
metals.
Hydride generation can also be used as a gaseous sample introduction/inter-
face unit between aqueous phase separation (HPLC, CE) and atomic spectroscopic
detectors. The clear advantage of this approach is the high sample introduction
efficiency compared with conventional nebulization processes. However, the
hydride interface tends to further complicate the already complex hyphenated
systems, introducing additional uncertainty factors. It remains a popular inter-
face/sample introduction technique between chromatography and atomic spectro-
scopic detectors, especially when the detectors are less sensitive and/or robust and
the trade-off for better detection limits is worthwhile [189-193].

28.4.2 Ethylation

Currently, sodium tetraalkylborate-based alkylation is the most commonly used

derivatization reaction. An advantage of NaBEt4 is that derivatization can be
accomplished in an aqueous environment, the natural medium for most environ-
mental and biological samples and there is no need to change phases, as in the
case of Grignard reagents. It has been used for the derivatization of alkyllead
[194,195], organotin [124,131,161,168,196-210] and selenium [138,211] species.
Fernandez et al. [212] recently reviewed the various derivatization approaches
used for methylmercury determination. Morabito et al. [213] compared various
derivatization methods, including ethylation by sodium tetraethylborate for
organotin compounds. Zufiaurre et al. [214] compared different alkylation meth-
ods for trimethyllead determination. Derivatization with NaBEt4 opens the
possibility for multielement speciation of tin, mercury and lead species [134,
215]. Although the derivatization process is suitable for multielement studies,
the limiting factor in achieving this is usually that the leaching/extraction
procedures required for the individual species of various elements are quite
different. One of the main limitations of the ethylation technique is that it
cannot be applied to the speciation of ethyl ligand containing species. For
example, reaction of triethyllead and inorganic lead with NaBEt4 produces the
same compound: tetraethyllead. Yu and Pawliszyn [195] overcame this limita-
tion by utilizing deuterated NaBEt 4 for the derivatization of organolead

957
compounds. The application of the propyl-[216,217] and phenylborate [218,219]
reagent also appears to be a promising approach.

28.4.3 Grignard reaction

The transformation of analyte into a more volatile compound can also be achieved
by reaction with a Grignard reagent (R-MgX) in a suitable solvent. The R alkyl
group can be methyl, ethyl, propyl, butyl, pentyl, or hexyl. Alkylation via the Grig-
nard reaction is a widely used derivatization technique for organotin [220-222],
alkyllead [223], antimony [224] and germanium [225] prior to their determina-
tion. The Grignard reaction is generally performed after an organic solvent
extraction because the Grignard reagent is not compatible with water. Grignard
derivatization permits the determination of different organometallic species in
various environmental matrices (water, sediment, biota) with high derivatiza-
tion yields and excellent reproducibility. However, the derivatization procedure is
quite lengthy, requires several sample manipulation steps which increase the risk
of contamination, decomposition and losses, and significantly increases the analy-
sis cost. In a recent study, Morabito et al. [213] compared Grignard pentylation
and ethylation with ethylation using sodium tetraethylborate for determination
of alkyltin in mussel tissue (Fig. 28.4). The two Grignard derivatization methods
provided very similar results, however, sodium tetrethylborate-based ethylation
was always biased somewhat lower for the studied species.

28.4.4 Halide generation

Halide generation aims at the formation of volatile metal(oid)- and organo-

metal-halide species [226]. The chemistry involved is generally quite straightfor-
ward: the halide forming species are simply exposed to a significant excess of a
halide in the form of a salt (such as sodium chloride), but in some cases high
acidity is also required. The generated metal halide species can be directly

Fig. 28.4. Comparison of Grignard penty- .

lation and ethylation with ethylation using 3
sodium tetraethylborate for alkyltin deter- n
mination in mussel tissue. Reprinted from
Trends in Analytical Chemistry, Vol. 19, R. t
Morabito, P. Massanisso, P. Quevauviller: -
Derivatization methods for the determi-
nation of organotin compounds in environ-
mental samples, pp. 119, Copyright (2000),
with permission from Elsevier Science
[213].

958
introduced into a detector. Alternatively, they can be pre-concentrated on
sorbent material. Inorganic arsenic, mono- and dimethylarsenic [227],
germanium [228], tributyltin [229], methylmercury [144], silicon [230,231] and
gallium have been reported. The high boiling point and the thermal instability of
these species makes them unsuitable for gas chromatographic analysis. At the
same time, the high boiling points facilitate room temperature trapping as
opposed to cryotrapping necessary for hydrides. This process is also free from
classical transition metal interferences typical of hydride generation methods.

28.5 CONCLUDING REMARKS

The need to preserve the integrity of various chemical forms of an element

during their sampling and extraction process and subsequently attempting to
separate them is an entirely new approach for the traditional inorganic analyti-
cal chemist. However, this is not a novel problem for organic separation science,
which is accustomed to dealing with such issues. Speciation analysis is thus
evolving in the same manner that organic analytical chemistry did years ago.
There are several specific issues, however, which characterize the speciation
field.
Because of the extremely diverse nature of organometallic species, extrac-
tion/ leaching procedures typically focus on the extraction of a single species of
an element, as in many instances they are ineffective for others. For example,
organotin species can be extracted from sediment samples using acetic acid/
methanol or an organic solvent system. This approach is specifically tailored for
organotins and largely ineffective for inorganic tin. The same picture can be
drawn for arsenic; different extraction protocols have been developed for arseno-
sugars/arsenobetaine and arsenocholine/and anionic arsenic species (mono and
dimethylarsenic and As +3 and As+5).
In the case of butyltin determination, extraction protocols essentially aim to
co-extract the hydrophobic tributyltin and strongly cationic monobutyltin. If for
any reason it is important to ascertain not only the tributyltin content, but also
its degradation products, different extraction procedures may be required.
Because of the complexity of the leaching, derivatization and separation
processes, it is crucial to ask the right question at the beginning of any exercise:
which are the compounds of real interest? If information concerning only
tributyltin instead of all three butyltins is required, this simplifies the extraction
protocol. Most importantly, because of the significantly different extraction
characteristics between tributyltin and its degradation products, specific leach-
ing/ extraction procedures can be designed to extract not only the tributyltin
from the matrix, but also to separate it from inorganic and other butyltin
species. Following selective extraction, the remaining tin content can be
measured directly with an atomic spectroscopic detector without chromato-
graphic separation. A well-designed extraction procedure can provide a suffi-
cient degree of separation to permit bypassing the chromatographic step and

959
complete the measurement with an extremely selective atomic spectroscopic
detector. This approach can be entertained for methylmercury/inorganic
mercury and alkyllead/inorganic lead.
Because serious matrix interferences may arise in many cases during speci-
ation analysis, this has to be seriously considered when deciding on the calibra-
tion approach to be taken. Analytical methods are sufficiently robust and the
sample matrix is sufficiently well-characterized to employ external calibration in
only a limited number of cases. The method of standard additions or isotope
dilution is thus strongly encouraged. Isotopic spiking and isotope dilution mea-
surements provide the most advanced calibration/quantification approaches,
based on changes in the known isotopic distribution of an element before and
after spiking with a single isotope. When used for speciation analysis, isotope
dilution requires advanced mass spectrometric facilities to enable accurate and
precise isotope ratio measurements to be achieved characterizing a single chro-
matographic peak. Many laboratories are equipped with such instruments but,
unfortunately, isotopically enriched species-specific standards are not commer-
cially available.

REFERENCES

1 E. Nieboer, Analyst, 117 (1992) 550.

2 D.M. Templeton, F. Ariese, R. Cornelis, L.G. Danielsson, H. Muntau, H.P. Van
Leeuwen and R. Lobinski, Pure Appl. Chem., 72 (2000) 1453.
3 J. Szpunar, Analyst, 125 (2000) 963.
4 J. Szpunar, Trends Anal. Chem., 19 (2000) 127.
5 J. Szpunar, H. Chassaigne, A. Makarov and R. Lobinski, Chemia Analityczna, 44 (3A)
Special Iss. SI (1999) 351.
6 J. Szpunar and R. Lobinski, Pure Appl. Chem., 71 (1999) 899.
7 J. Szpunar, P. Pellerin, A. Makarov, T. Doco, P. Williams, B. Medina and R. Lobinski,
J. Anal. Atom. Spectrom., 13 (1998) 749.
8 J. Szpunar, P. Pellerin, A. Makarov, T. Doco, P. Williams and R. Lobinski, J. Anal.
Atom. Spectrom., 14 (1999) 639.
9 J. Szpunar, A. Makarov, T. Pieper, B.K. Keppler and R. Lobinski, Anal. Chim. Acta,
387 (1999) 135.
10 C. Sarzanini and E. Mentasti, J. Chromatogr.A, 789 (1997) 301.
11 C. Sarzanini, J. Chromatogr. A, 850 (1999) 213.
12 R. Lobinski and F.C. Adams, Spectrochim. Acta B, 52 (1997) 1865.
13 J.W. Olesik, J.A. Kinzer, E.J. Grunwald, K.K. Thaxton and S.V. Olesik, Spectrochim.
Acta B, 53 (1998) 239.
14 S. Caroli, Element Speciation in Bioinorganic Chemistry. Wiley, New York, 1996.
15 A.M. Ure and C.M. Davidson, Chemical Speciation in the Environment. Blackie,
London, 1995.
16 J.A. Caruso, K.L. Sutton and K.L. Ackley, Elemental Speciation:New Approaches for
Trace Element Analysis. Elsevier, Amsterdam, 2000.
17 J. Anal. Atom. Spectrom. Special issue, 2000.
18 International Vocabulary of Basic and General Terms in Metrology, ISO, 1993.
19 P. Quevauviller, L. Ebdon, R.M. Harrison and Y. Wang, Appl. Organomet. Chem., 13
(1999) 1.

960
20 A. Chatterjee, Y. Shibata, J. Yoshinaga and M. Morita, Appl. Organomet. Chem., 15
(2001) 306.
21 J. Yoshinaga, A. Chatterjee, Y. Shibata, M. Morita and J.S. Edmonds, Clin. Chem., 46
(2000) 1781.
22 F. Lagarde, M.B. Amran, M.J.F. Leroy, C. Demesmay, M. Olle, A. Lamotte, H.
Muntau, P. Michel, P. Thomas, S. Caroli, E. Larsen, P. Bonner, G. Rauret, M.
Foulkes, A. Howard, B. Griepink and E.A. Maier, Fresen. J. Anal. Chem., 363 (1999)
18.
23 R. Morabito, H. Muntau, W. Cofino and P. Quevauviller, J. Environ. Monit., 1 (1999)
75.
24 J. Yoshinaga, H. Kon, T. Horiguchi, M. Morita and K. Okamoto, Anal. Sci., 14 (1998)
1121.
25 K. Vercoutere, R. Cornelis, L. Mees and P. Quevauviller, Analyst, 123 (1998) 965.
26 P. Quevauviller, Fresen. J. Anal. Chem., 354 (1996) 515.
27 B. Michalke, Fresen. J. Anal. Chem., 363 (1999) 439.
28 P. Quevauviller, Spectrochim. Acta B, 53 (1998) 1261.
29 P. Quevauviller, Method Performance Studies for Speciation Analysis. RSC,
Cambridge, UK, 1998.
30 Z. Mester, G. Vitanyi, R. Morabito and P. Fodor, J. Chromatogr.A, 832 (1999) 183.
31 Y.S. Drugov, J. Anal. Chem., 53 (1998) 606.
32 H. Tao, Bunseki Kagaku, 46 (1997) 239.
33 T. Guerin, A. Astruc and M. Astruc, Talanta, 50 (1999) 1.
34 K. Pyrzynska, Chemia Analityczna, 40 (1995) 677.
35 K. Pyrzynska, Analyst, 121 (1996) R77.
36 K. Pyrzynska, Anal. Sci., 14 (1998) 479.
37 H. Ding, L.K. Olson and J.A. Caruso, Spectrochim. Acta B, 51 (1996) 1801.
38 K. Vercoutere and R. Cornelis, Qual. Assurance Environ. Anal., 17 (1995) 195.
39 W.R.L. Cairns, L. Ebdon and S.J. Hill, Fresen. J. Anal. Chem., 355 (1996) 202.
40 S. Lustig, J. De Kimpe, R. Cornelis, P. Schramel and B. Michalke, Electrophoresis,20
(1999) 1627.
41 M.M. Bayon, A.B.S. Cabezuelo, E.B. Gonzalez, J.I.G. Alonso and A. Sanz Medel, J.
Anal. Atom. Spectrom., 14 (1999) 947.
42 H. Koyama, K. Omura, A. Ejima, Y. Kasanuma, C. Watanabe and H. Satoh, Anal.
Biochem., 267 (1999) 84.
43 K.T. Suzuki, Analusis, 26 (1998) M57.
44 A.T. Lombardi and A.A.H. Vieira, Phycologia, 37 (1998) 34.
45 C.N. Ferrarello, M.M. Bayon, R.F. de la Campa and A. Sanz Medel, J. Anal. Atom.
Spectrom., 15 (2000) 1558.
46 K. Polec, O. Garcia Arribas, M. Perez Calvo, J. Szpunar, B. Ribas Ozonas and R.
Lobinski, J. Anal. Atom. Spectrom., 15 (2000) 1363.
47 I. Havezov, Fresen. J. Anal. Chem., 355 (1996) 452.
48 M. Vilano, A. Padro and R. Rubio, Anal. Chim. Acta, 411 (2000) 71.
49 J.L. Gomez Ariza, D. Sanchez Rodas, R. Beltran, W. Corns and P. Stockwel, Appl.
Organomet. Chem., 12 (1998) 439.
50 A. D'Ulivo and S. Rapsomanikis, Anal. Lett., 30 (1997) 2109.
51 Z. Mester and P. Fodor, Anal. Chim. Acta, 386 (1999) 89.
52 Y.L. Feng, H. Narasaki, H.Y. Chen and L.C. Tian, Anal. Chim. Acta, 386 (1999) 297.
53 I.R. Pereiro, A. Wasik and R. Lobinski, Fresen. J. Anal. Chem., 363 (1999) 460.
54 P.C. Uden, S.M. Bird, M. Kotrebai, P. Nolibos, J.F. Tyson, E. Block and E. Denoyer,
Fresen. J. Anal. Chem., 362 (1998) 447.
55 M.A. Dempster and R.K. Marcus, J. Anal. Atom. Spectrom., 15 (1999) 43.
56 N.G.O. Velado, R. Pereiro and A. Sanz Medel, J. Anal. Atom. Spectrom., 15 (1999) 49.

961
57 Stewart, II, Spectrochim. Acta B, 54 (1999) 1649.
58 M.B. DelaCalleGuntinas, C. Brunori, R. Scerbo, S. Chiavarini, P. Quevauviller, F.
Adams and R. Morabito, J. Anal. Atom. Spectrom., 12 (1997) 1041.
59 K. Sutton, R.M.C. Sutton and J.A. Caruso, J. Chromatogr.A, 789 (1997) 85.
60 J.S. Becker and H.J. Dietze, Spectrochim. Acta B, 53 (1998) 1475.
61 A.E. Croslyn, B.W. Smith and J.D. Winefordner, Crit. Rev. Anal. Chem., 27 (1997)
199.
62 J.A.C. Broekaert, Mikrochim. Acta, 120 (1995) 21.
63 S.J. Hill, L.J. Pitts and A.S. Fisher, Trends Anal. Chem., 19 (2000) 120.
64 K.L. Sutton and J.A. Caruso, J. Chromatogr.A, 856 (1999) 243.
65 G.K. Zoorob, J.W. McKiernan and J.A. Caruso, Mikrochim. Acta, 128 (1998) 145.
66 J.P. Guzowski and G.M. Hieftje, Anal. Chem., 72 (2000) 3812.
67 C. Lewis, S.K. Doom, D.M. Wayne, F.L. King and V. Majidi, Appl. Spectrosc., 54
(2000) 1236.
68 R. Guevremont and R.E. Sturgeon, J. Anal. Atom. Spectrom., 15 (2000) 37.
69 G. O'Connor, L. Ebdon, E.H. Evans, H. Ding, L.K. Olson and J.A. Caruso, J. Anal.
Atom. Spectrom., 11 (1996) 1151.
70 N.P. Vela, J.A. Caruso and R.D. Satzger, Appl. Spectrosc., 51 (1997) 1500.
71 J.L.G. Ariza, E. Morales, D. Sanchez Rodas and I. Giraldez, Trends Anal. Chem., 19
(2000) 200.
72 C. Bancon Montigny, G. Lespes and M. Potin Gautier, Water Res., 35 (2001) 224.
73 J.L. Gomez Ariza, I. Giraldez, E. Morales, F. Ariese, W. Cofino and P. Quevauviller,
J. Environ. Monit., 1 (1999) 197.
74 J. Feldmann, V.W.M. Lai, W.R. Cullen, M.S. Ma, X.F. Lu and X.C. Le, Clin. Chem., 45
(1999) 1988.
75 M.M. Gomez, T. Gasparic, M.A. Palacios and C. Camara, Anal. Chim. Acta, 374
(1998) 241.
76 J.L. Gomez Ariza, J.A. Pozas, I. Giraldez and E. Morales, Int. J. Environ. Anal.
Chem., 74 (1999) 215.
77 K. Haas and J. Feldmann, Anal. Chem., 72 (2000) 4205.
78 G.E.M. Hall, J.C. Pelchat and G. Gauthier, J. Anal. Atom. Spectrom., 14 (1999) 205.
79 R.M. Olivas, P. Quevauviller and O.F.X. Donard, Fresen. J. Anal. Chem., 360 (1998)
512.
80 Y. Wang, R.M. Harrison and P. Quevauviller, Appl. Organomet. Chem., 10 (1996) 69.
81 P. Quevauviller and O.F.X. Donard, J. Environ. Monit., 1 (1999) 503.
82 M.T. Perez Corona, Y. Madrid Albarran and C. Camara, Fresen. J. Anal. Chem., 368
(2000) 471.
83 R. Draisci, C. Marchiafava, L. Palleschi, P. Cammarata and S. Cavalli, J.
Chromatogr. B, 753 (2001) 217.
84 P.A. Gallagher, J.A. Shoemaker, X.Y. Wei, C.A. Brockhoff Schwegel and J.T. Creed,
Fresen. J. Anal. Chem., 369 (2001) 71.
85 P.A. Gallagher, X.Y. Wei, J.A. Shoemaker, C.A. Brockhoff and J.T. Creed, J. Anal.
Atom. Spectrom., 14 (1999) 1829.
86 J.W. McKiernan, J.T. Creed, C.A. Brockhoff, J.A. Caruso and R.M. Lorenzana, J.
Anal. Atom. Spectrom., 14 (1999) 607.
87 Dionex Application Note 339, 1999.
88 H.M. Kingston and L.B. Jassie, Introduction to Microwave Sample Preparation.
American Chemical Society, Washington, DC, 1988.
89 H.M. Kingston and S.J. Haswell, Microwave-Enhanced Chemistry. American Chem-
ical Society, Washington, DC, 1997.
90 H.M.S. Kingston and P.J. Walter, in: A. Montaser (Ed.), Inductively Coupled Plasma
Mass Spectrometry. Wiley, New York, NY, 1998, p. 33.

962
 91 V. Camel, Trends Anal. Chem., 19 (2000) 229.
92 Q.H. Jin, F. Liang, H.Q. Zhang, L.W. Zhao, Y.F. Huan and D.Q. Song, Trends Anal.
Chem., 18 (1999) 479.
93 K.J. Lamble and S.J. Hill, Analyst, 123 (1998) R103.
94 C.S. Eskilsson and E. Bjorklund, J. Chromatogr.A, 902 (2000) 227.
95 J. Szpunar, V.O. Schmitt, O.F.X. Donard and R. Lobinski, Trends Anal. Chem. 15
(1996) 181.
96 B.S. Tselentis, M. Maroulakou, J.F. Lascourreges, J. Szpunar, V. Smith and O.F.X.
Donard, Mar. Pollut. Bull., 38 (1999) 146.
97 V.O. Schmitt, J. Szpunar, O.F.X. Donard and R. Lobinski, Can. J. Anal. Sci. Spect.,
42 (1997) 41.
98 I.R. Pereiro, V.O. Schmitt, J. Szpunar, O.F.X. Donard and R. Lobinski, Anal. Chem.,
68 (1996) 4135.
99 J. Szpunar, J. Bettmer, M. Robert, H. Chassaigne, K. Cammann, R. Lobinski and
O.F.X. Donard, Talanta, 44 (1997) 1389.
100 J.M. Ruiz, J. Szpunar and O.F.X. Donard, Sci. Total Environ., 198 (1997) 225.
101 H. Garraud, M. Robert, C.R. Quetel, J. Szpunar and O.F.X. Donard, Atom. Spectrosc.,
17 (1996) 183.
102 C.M. Tseng, A. deDiego, F.M. Martin and O.F.X. Donard, J. Anal. Atom. Spectrom.,
12 (1997) 629.
103 M.J. Vazquez, A.M. Carro, R.A. Lorenzo and R. Cela, Anal. Chem., 69 (1997) 221.
104 I.R. Pereiro, A. Wasik and R. Lobinski, J. Anal. Atom. Spectrom., 13 (1998) 743.
105 R.A. Lorenzo, M.J. Vazquez, A.M. Carro and R. Cela, Trends Anal. Chem., 18 (1999)
410.
106 M.J. Vazquez, M. Abuin, A.M. Carro, R.A. Lorenzo and R. Cela, Chemosphere, 39
(1999) 1211.
107 M. Abuin, A.M. Carro and R.A. Lorenzo, J. Chromatogr. A, 889 (2000) 185.
108 J.M. Bayona, Trends Anal. Chem., 19 (2000) 107.
109 Y. Cai, M. Abalos and J.M. Bayona, Appl. Organomet. Chem., 12 (1998) 577.
110 Y. Cai, R. Alzaga and J.M. Bayona, Anal. Chem., 66 (1994) 1161.
111 I. Fernandez Escobar and J.M. Bayona, Anal. Chim. Acta, 355 (1997) 269.
112 V. Lopez Avila, Y. Liu and W.F. Beckert, J. Chromatogr.A, 785 (1997) 279.
113 N.P. Vela and J.A. Caruso, J. Anal. Atom. Spectrom., 11 (1996) 1129.
114 J.M. Bayona, Qual. Assurance Environ. Anal., 17 (1995) 465.
115 Y. Liu, V. Lopez Avila, M. Alcaraz and W.F. Beckert, J. AOACInt., 78 (1995) 1275.
116 K.M. Attar, Appl. Organomet. Chem., 10 (1996) 317.
117 H. Emteborg, E. Bjorklund, F. Odman, L. Karlsson, L. Mathiasson, W. Frech and
D.C. Baxter, Analyst, 121 (1996) 19.
118 W. Holak, J. AOAC Int., 78 (1995) 1124.
119 R. Cela Torrijos, M. Miguens Rodriguez, A.M. Carro Diaz and R.A. Lorenzo Ferreira,
J. Chromatogr.A, 750 (1996) 191.
120 Y.C. Sun, J. Mierzwa, Y.T. Chung and M.H. Yang, Anal. Commun., 34 (1997) 333.
121 N.S. Simon, Int. J. Environ. Anal. Chem., 68 (1997) 313.
122 http://www.tmasc.com, 2001.
123 M.Abalos, J.M. Bayona and P. Quevauviller, Appl. Organomet. Chem., 12 (1998) 541.
124 S. Shawky, H. Emons and H.W. Durbeck, Anal. Commun., 33 (1996) 107.
125 J.C. Wu, Z. Mester and J. Pawliszyn, J. Anal. Atom. Spectrom., 16 (2001) 159.
126 J.C. Wu, Z. Mester and J. Pawliszyn, Anal. Chim. Acta, 424 (2000) 211.
127 J.A. Caruso, D.T. Heitkemper and C. B'Hymer, Analyst, 126 (2001) 136.
128 J.L. Gomez Ariza, D. Sanchez Rodas, I. Giraldez and E. Morales, Analyst, 125 (2000)
401.
129 G.L. Hu and X.R. Wang, Anal. Lett., 31 (1998) 1445.

963
130 Y.K. Chau, F. Yang and M. Brown, Anal. Chim. Acta, 338 (1997) 51.
131 M.B. delaCalleGuntinas, R. Scerbo, S. Chiavarini, P. Quevauviller and R. Morabito,
Appl. Organomet. Chem., 11 (1997) 693.
132 J. Pawliszyn, Solid Phase Microextraction: Theory and Practice. Wiley, New York
(NY), 1997.
133 Z. Mester, R. Sturgeon and J. Pawliszyn, Spectrochim. Acta B, 56 (2001) 233.
134 L. Moens, T. De Smaele, R. Dams, P. Van den Broeck and P. Sandra, Anal. Chem., 69
(1997) 1604.
135 B. He, G.B. Jiang and Z.M. Ni, J. Anal. Atom. Spectrom., 13 (1998) 1141.
136 B. He and G.B. Jiang, Fresen. J. Anal. Chem., 365 (1999) 615.
137 Z. Mester, R.E. Sturgeon and J.W. Lam, J. Anal. Atom. Spectrom., 15 (2000) 1461.
138 M. Guidotti, J. AOAC Int., 83 (2000) 1082.
139 M. Guidotti, G. Ravaioli and M. Vitali, HRCJ. High Resolut. Chromatogr.,22 (1999)
414.
140 Z. Mester and J. Pawliszyn, J. Chromatogr.A, 873 (2000) 129.
141 Z. Mester, G. Horvath, G. Vitanyi, L. Lelik and P. Fodor, Rapid Commun. Mass.
Spectrom., 13 (1999) 350.
142 T. Gorecki and J. Pawliszyn, Anal. Chem., 68 (1996) 3008.
143 J.P. Snell, W. Frech and Y. Thomassen, Analyst, 121 (1996) 1055.
144 Z. Mester, J. Lam, R. Sturgeon and J. Pawliszyn, J. Anal. Atom. Spectrom., 15 (2000)
837.
145 C.R. Jia, Y.Z. Luo and J. Pawliszyn, J. Microcolumn Sep., 10 (1998) 167.
146 T.P. Gbatu, K.L. Sutton and J.A. Caruso, Anal. Chim. Acta, 402 (1999) 67.
147 J.C. Wu, X.M. Yu, H. Lord and J. Pawliszyn, Analyst, 125 (2000) 391.
148 Z. Mester and J. Pawliszyn, Rapid Commun. Mass. Spectrom. 13 (1999) 1999.
149 Z. Mester, H. Lord and J. Pawliszyn, J. Anal. Atom. Spectrom., 15 (2000) 595.
150 B. Kolb, J. Chromatogr. A, 842 (1999) 163.
151 C.M. Tseng, A. DeDiego, F.M. Martin, D. Amouroux and O.F.X. Donard, J. Anal.
Atom. Spectrom., 12 (1997) 743.
152 F. Martin, F.M. Corrigan, O.F.X. Donard, J. Kelly, J.A.O. Besson and D.F. Horrobin,
Human. Exp. Toxicol., 16 (1997) 512.
153 D. Amouroux, E. Tessier, C. Pecheyran and O.F.X. Donard, Anal. Chim. Acta, 377
(1998) 241.
154 C.M. Tseng, A. deDiego, H. Pinaly, D. Amouraoux and O.F.X. Donard, J. Anal. Atom.
Spectrom., 13 (1998) 755.
155 A. deDiego, C.M. Tseng, T. Stoichev, D. Amouroux and O.F.X. Donard, J. Anal. Atom.
Spectrom., 13 (1998) 623.
156 C. Pecheyran, C.R. Quetel, F.M.M. Lecuyer and O.F.X. Donard, Anal. Chem., 70
(1998) 2639.
157 C. Pecheyran, D. Amouroux and O.F.X. Donard, J. Anal. Atom. Spectrom., 13 (1998)
615.
158 C.M. Tseng, A. De Diego, J.C. Wasserman, D. Amouroux and O.F.X. Donard,
Chemosphere, 39 (1999) 1119.
159 A. de Diego, C. Pecheyran, C.M. Tseng and O.F.X. Donard, in: A. Sanz-Medel (Ed.),
Flow Analysis with Atomic Spectrometric Detectors. Elsevier, Amsterdam, 1999, p.
375.
160 C.M. Tseng, D. Amouroux, I.D. Brindle and O.F.X. Donard, J. Envuiron. Monit., 2
(2000) 603.
161 D. Amouroux, E. Tessier and O.F.X. Donard, Environ. Sci. Technol., 34 (2000) 988.
162 C. Pecheyran, B. Lalere and O.F.X. Donard, Environ. Sci. Technol., 34 (2000) 27.
163 A. de Diego, C.M. Tseng, N. Dimov, D. Amouroux and O.F.X. Donard, Appl.
Organomet. Chem., 15 (2001) 490.

964
 164 J. Sanz, M. Perez, M.T. Martinez and M. Plaza, Talanta, 50 (1999) 149.
165 J. Sanz, M. Perez, M.T. Martinez and M. Plaza, Talanta, 51 (2000) 849.
166 J. Sanz Asensio, M.T. Martinez Soria, M. Plaza Medina and M. Perez Clavijo, Anal.
Chim. Acta, 409 (2000) 171.
167 S. Cabredo, J. Galban and J. Sanz, Talanta, 46 (1998) 631.
168 R. Eiden, H.F. Scholer and M. Gastner, J. Chromatogr. A, 809 (1998) 151.
169 T. Kaise, M. Ogura, T. Nozaki, K. Saitoh, T. Sakurai, C. Matsubara, C. Watanabe and
K. Hanaoka, Appl. Organomet. Chem., 11 (1997) 297.
170 A.G. Howard and C. Salou, Anal. Chim. Acta, 333 (1996) 89.
171 J.L. Burguera, M. Burguera, C. Rivas and P. Carrero, Talanta, 45 (1998) 531.
172 A.G. Howard and C. Salou, J. Anal. Atom. Spectrom., 13 (1998) 683.
173 U. Pyell, A. Dworschak, E. Nitschke and B. Neidhart, Fresen. J. Anal. Chem., 363
(1999) 495.
174 J.Y. Cabon and N. Cabon, Fresen. J. Anal. Chem., 368 (2000) 484.
175 I.R. Pereiro, A. Wasik and R. Lobinski, Anal. Chem., 70 (1998) 4063.
176 M. Ceulemans and F.C. Adams, J. Anal. Atom. Spectrom., 11 (1996) 201.
177 A. Wasik, I.R. Pereiro, C. Dietz, J. Szpunar and R. Lobinski, Anal. Commun., 35
(1998) 331.
178 C. Dietz, Y. Madrid, C. Camara and P. Quevauviller, Anal. Chem., 72 (2000) 4178.
179 J. Feldmann, I. Koch and W.R. Cullen, Analyst, 123 (1998) 815.
180 J. Feldmann, E.M. Krupp, D. Glindemann, A.V. Hirner and W.R. Cullen, Appl.
Organomet. Chem., 13 (1999) 739.
181 J. Feldmann, J. Environ. Monit., 1 (1999) 33.
182 J. Feldmann and W.R. Cullen, Environ. Sci. Technol., 31 (1997) 2125.
183 J. Dedina, in: A. Sanz-Medel (Ed.), Flow Analysis with Atomic Spectrometric
Detectors. Elsevier, Amsterdam, 1999.
184 Z. Mester and P. Fodor, Spectrochim. Acta B, 52 (1997) 1763.
185 S.A. Pergantis, W. Winnik, E.M. Heithmar and W.R. Cullen, Talanta,44 (1997) 1941.
186 A.G. Howard, J. Anal. Atom. Spectrom., 12 (1997) 267.
187 J.H. Weber, Trends Anal. Chem., 16 (1997) 73.
188 S.N. Willie, Spectrochim. Acta B, 51 (1996) 1781.
189 Z. Mester and P. Fodor, J. Chromatogr.A, 756 (1996) 292.
190 Z. Mester, A. Woller and P. Fodor, Microchem. J., 54 (1996) 184.
191 S. Yalcin and X.C. Le, Talanta, 47 (1998) 787.
192 Z. Slejkovec, J.T. van Elteren and A.R. Byrne, Talanta, 49 (1999) 619.
193 J.L. Gomez Ariza, D. Sanchez Rodas, E. Morales, O. Herrgott and I.L. Marr, Appl.
Organomet. Chem., 13 (1999) 783.
194 N. Mikac, M. Branica, Y. Wang and R.M. Harrison, Environ. Sci. Technol., 30 (1996)
499.
195 X.M. Yu and J. Pawliszyn, Anal. Chem., 72 (2000) 1788.
196 P.W. Looser, M. Berg, K. Fent, J. Muhlemann and R.P. Schwarzenbach, Anal. Chem.,
72 (2000) 5136.
197 J.R. Encinar, J. Alonso and A. Sanz Medel, J. Anal. Atom. Spectrom., 15 (2000) 1233.
198 B.S. Tselentis and E.S. Tzannatos, Fresen. Environ. Bull., 9 (2000) 499.
199 S. Aguerre, C. Bancon Montigny, G. Lespes and M. Potin Gautier, Analyst, 125
(2000) 263.
200 H. Tao, R.B. Rajendran, C.R. Quetel, T. Nakazeto, M. Tominaga and A. Miyazaki,
Anal. Chem., 71 (1999) 4208.
201 A.A. Ansari, I.B. Singh and H.J. Tobschall, Sci. Total Environ., 223 (1998) 157.
202 C. Montigny, G. Lespes and M. PotinGautier, J. Chromatogr.A, 819 (1998) 221.
203 C.G. Arnold, M. Berg, S.R. Muller, U. Dommann and R.P. Schwarzenbach, Anal.
Chem., 70 (1998) 3094.

965
204 R. Ritsema, T. deSmaele, L. Moens, A.S. deJong and O.F.X. Donard, Environ. Pollut.,
99 (1998) 271.
205 C. CarlierPinasseau, G. Lespes and M. Astruc, Environ. Technol., 18 (1997) 1179.
206 D. Milde, Z. Plzak and M. Suchanek, Collection Czech. Chem. Commun., 62 (1997)
1403.
207 C. Gerbersmann, M. Heisterkamp, F.C. Adams and J.A.C. Broekaert, Anal. Chim.
Acta, 350 (1997) 273.
208 S. Girousi, E. Rosenberg, A. Voulgaropoulos and M. Grasserbauer, Fresen. J. Anal.
Chem., 358 (1997) 828.
209 C. CarlierPinasseau, G. Lespes and M. Astruc, Appl. Organomet. Chem., 10 (1996)
505.
210 S. Tutschku, S. Mothes and R. Wennrich, Fresen.J. Anal. Chem., 354 (1996) 587.
211 R. Allabashi, J. Rendl and M. Grasserbauer, Fresen. J. Anal. Chem., 360 (1998) 723.
212 R.G. Fernandez, M.M. Bayon, J.I.G. Alonso and A. Sanz Medel, J. Mass Spectrom., 35
(2000) 639.
213 R. Morabito, P. Massanisso and P. Quevauviller, TrendsAnal. Chem., 19 (2000) 113.
214 R. Zufiaurre, B. Pons and C. Nerin, J. Chromatogr. A, 779 (1997) 299.
215 U.M. Gruter, J. Kresimon and A.V. Hirner, Fresen. J. Anal. Chem., 368 (2000) 67.
216 K. Bergmann and B. Neidhart, J. Sep. Sci., 24 (2001) 221.
217 M. Heisterkamp and F.C. Adams, J. Anal. Atom. Spectrom., 14 (1999) 1307.
218 Y. Cai, S. Monsalud and K.G. Furton, Chromatographia,52 (2000) 82.
219 Y. Cai, S. Monsalud, R. Jaffe and R.D. Jones, J. Chromatogr.A, 876 (2000) 147.
220 S. Diez, L. Ortiz and J.M. Bayona, Chromatographia,52 (2000) 657.
221 G. Binato, G. Biancotto, R. Piro and R. Angeletti, Fresen. J. Anal. Chem., 361 (1998)
333.
222 1. FernandezEscobar, M. Gibert, A. Messeguer and J.M. Bayona, Anal. Chem., 70
(1998) 3703.
223 J.R. Baena, S. Cardenas, M. Gallego and M. Valcarcel,Anal. Chem., 72 (2000) 1510.
224 M.B. delaCalleGuntinas and F.C. Adams, J. Chromatogr.A, 764 (1997) 169.
225 G.B. Jiang and F.C. Adams, Anal. Chim. Acta, 337 (1997) 83.
226 P. Smichowski and S.V. Farias, Microchem. J., 67 (2000) 147.
227 Z. Mester and R.E. Sturgeon, J. Anal. Atom. Spectrom., 16 (2001) 470.
228 X.W. Guo and X.M. Guo, Anal. Chim. Acta, 330 (1996) 237.
229 Z. Mester, R.E. Sturgeon, J.W. Lam, P.S. Maxwell and L. Peter, J. Anal. Atom.
Spectrom., 16 (2001) 1313.
230 A.L. Molinero, L. Martinez, A. Villareal and J.R. Castillo, Talanta, 45 (1998) 1211.
231 A.L. Molinero, A. Morales, A. Villareal and J.R. Castillo, Fresen. J. Anal. Chem., 358
(1997) 599.

966
Chapter29

Sample cleanup and concentration

techniques for capillary electrophoresis
Xing-Zheng Wu

29.1 INTRODUCTION

Capillary electrophoresis (CE) is becoming one of the most powerful micro-

separation techniques. In the last two decades, various CE modes such as
capillary zone electrophoresis (CZE), capillary gel electrophoresis (CGE),
micellar electrokinetic capillary chromatography (MEKC), capillary electro-
chromatography (CEC), capillary isoelectric focusing (CIEF), and capillary iso-
tachophoresis (CITP) have been developed [1]. Furthermore, these CE modes
have been successfully adapted to a microchip format in the last decade [2]. Also,
various samples ranging from simple inorganic ions to biomolecular ions such as
proteins and peptides can be easily analyzed with CE. Compared with other
separation methods such as HPLC, the features of CE are high separation
efficiency, short analysis time, and low sample consumption.
In various detection methods developed for CE, although mass sensitivity is
extremely high because of the small detection volumes in CE, the concentration
sensitivity is usually not sufficient. Particularly for the most widely used UV
absorbance detection of CE, the concentration detection limit is generally on the
order of 10-100-fold higher than that of HPLC due to the short optical pass-
length. Improvement of the capillary geometry and optical design, such as the
use of a bent capillary (Z-cell) [3,4], a silver-coated capillary (multireflection flow
cell) [5], end column [6] and off-column [7] detections, is one approach to lower-
ing the detection limit of CE. Another important and more effective approach is
sample preconcentration [3].
Other problems in applying CE to some real samples (particularly biological
samples) are that inorganic constituents of high concentration in the samples
(e.g., serum and plasma) would reduce the efficiency of CE separation, and some
components (e.g., proteins in biological samples and surfactants in environmen-
tal samples) may cause binding to the capillary wall and thus result in a signifi-
cant change of migration times and the presence of interfering peaks in the
electropherograms [8,9]. Therefore, sample preparation including cleanup and

ComprehensiveAnalytical Chemistry XXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
preconcentration, plays a key role in widespread applications of CE in various
samples.
So far, there have been many attempts at simple sample preparation which
can not only improve separation efficiency but also lower the detection limit [9].
This chapter will introduce basic considerations and approaches in sample prep-
aration for CE, emphasizing on-line sample cleanup and concentration tech-
niques.

29.2 OFF-LINE SAMPLE CLEAN-UP AND PRECONCENTRATION

METHODS FOR CE

As shown in Fig. 29.1, a sample can be treated with conventional sample cleanup
and preconcentration methods such as solid phase extraction (SPE) and liquid-
liquid extraction (LLC). Then, the prepared sample (usually greatly concen-
trated in comparison with the original sample) is injected into a capillary by
either hydrodynamic or electrokinetic injection methods, and CE separation and
determination are followed. This off-line sample cleanup and preconcentra-
tion-CE has been used for many practical samples. For example, different struc-
tural types of industrial dye samples can be preconcentrated on a SAX Bond Elut
cartridge [10]. Then the dyes are eluted by mixture of HCl and MeOH. After
evaporating the solvent, the dyes are re-dissolved in 0.5 ml running buffer of CE
and are analyzed by CE. This off-line SPE concentration could lower the detec-
tion limit by 50-fold [10]. Various phenols in wastewater included in the 8041
U.S. Environmental Protection Agency (EPA) method can be preconcentrated in
a styrene-divinylbenzene SPM cartridge, then separated and determined by
non-aqueous CE using a buffer of ammonium acetate dissolved in N-methyl-
formamide-acetonitrile [11]. Quinolone antibiotics in plasma samples can also
be separated and quantitatively determined by CE with SPE pre-treatment [12].
The plasma matrix is removed, and the quinolone antibiotics are preconcen-
trated in a SPE cartridge [12]. In addition to the SPE cartridges, a SPE disk has
also been used for extraction ethylenediaminetetraacetic acid (EDTA) in envi-
ronmental water as nickel EDTA chelate. Then, the preconcentrated nickel
EDTA chelate is quantitatively determined by CE-MS [13].
1 Clean up and 2 Injection 3 CE
Preconcentration
(Hydrodynamic injection,
Electroldnetic injection)

+Uroj

.......... _ sm-pl. -l tpl)

Fig. 29.1. Schematic of off-line sample cleanup/preconcentration-CE process.

968
 Liquid-liquid extraction has also been used for the off-line cleanup and
preconcentration of CE. Fluorescently labelled amino acids and peptides are
extracted and concentrated into an organic phase such as acetophenone, then
they are separated and determined by CE with fluorescence detection [14]. A
membrane, gel, or hollow fiber also can be used for the off-line preconcentration
of CE [15-17]. Water in a hollow fiber filled with sample solution can be removed
by evaporation or Donnan effect; thus, sample in the fiber is concentrated into a
zone with a length of only a few mm [17]. Then, the concentrated sample zone
can be injected into a capillary for following CE determination. The method can
provide a high concentration factor up to 1000-fold for protein samples [17].
Isotachophoretic (ITP) has also been used for the off-line sample pre-treatment
of CE [18].
Although the above off-line preconcentration methods can lower the concen-
tration detection limit of CE -100-fold or even -1000-fold, the main disadvan-
tage is inefficient use of the concentrated sample because usually less than 1% of
the concentrated sample is injected into the capillary for further CE determina-
tion. Also, sample loss and contamination may occur during the preconcentra-
tion process, especially for preparation of a minute volume of sample. Therefore,
on-line sample cleanup and preconcentration methods are desirable for CE.

29.3 ON-LINE SAMPLE CLEAN-UP AND CONCENTRATION METHODS

FOR CE

29.3.1 On-line chemical or chromatographic concentration

Simple on-line coupling of an SPE microcolumn to a CE capillary can be

accomplished by use of a valve [19] or T-connector [20]. In sample enrichment,
the SPE microcolumn is connected with a drain tube or drain capillary.
Relatively large volumes of sample solution can be flowed through the SPE
microcolumn. Then the SPE microcolumn is switched to connect with the CE
capillary. A small amount of appropriate eluting buffer followed with CE
running buffer is injected into the SPE microcolumn, and CE separation of the
concentrated sample is carried out by applying a separation voltage. This on-line
SPE-CE has been used for determinations of the drugs papaverine, quinidine
and hydroquinidine, and propranolol in urine and serum [19,20]. The detection
limit can be lowered by over 100-fold.
Recently, a more compact on-line coupling of SPE to CE without the use of
any valve or T-connector has been proposed. A short capillary filled with a chro-
matographic sorbent [21-23] or membrane [24,25] is directly connected with the
CE capillary (Fig. 29.2A). This type of on-line coupling is also called in-line
SPE-CE coupling [9]. The sample enrichment procedure includes washing, wet-
ting, conditioning, sorption, washing, filling, and desorption. Finally, CE separa-
tion follows. Figure 29.2B shows one example of terbutaline determination by
the SPE-CE with C alkyl-diol silica as sorbent. It permits detection of 0.6 nM

969
 A Frit or filter
X - -o -L - - ww 9 Ha w
-,vaa
v ww "

Sorbent or membrane CE capillary

B
A).]1nM terbutaline with enrich'rhi
.
<

B) 10 nM terbutaline with enrichment 1mAUi

C) 10 pM terbutaline without enrichment

0 1 2 3
Migration time (min)

Fig. 29.2. Schematic of the compact on-line coupling of SPE-CE (A) and CE electropherograms of
terbutaline (B). In (B), A and B are electropherograms of 1 nM and 10 nM terbutaline with
on-line SPE concentration, and C is an electropherogram of 10 /M terbutaline without on-line
SPE concentration, respectively. For details see Ref. [21]. (B) is reprinted with permission from
Ref. [21].

terbutaline while detection limit is only 4.4 AM without the SPE enrichment
[21]. A similar SPE-CE has also been used for peptide separation [23]. Peptides
from a sample matrix are adsorbed on a reversed phase resin (C-8 or C-18)
sorbent and subsequently released for CE separation by injection of organic
elutant. This technique allows the peptide separation on the level of -10 ng/ml
[23]. Membrane materials with cationic-exchange or hydrophobic characteris-
tics also have been used for on-line preconcentration and cleanup of CE for pro-
tein and peptide samples [24,25]. This compact SPE-CE lowers the detection
limit of CE by a factor of several hundreds or even thousands.
Solid phase microextraction (SPME) has also been coupled to CE [26]. Figure
29.3A shows a coupling attachment for SPME to CE. A 1.5 cm capillary segment
(180 m i.d., 380 m o.d.) is closely attached to a CE capillary in a heat shrink-
able tube. A glass fiber of 150 /Am coated with poly(dimethylsiloxane) is used for
concentrating polycyclic aromatic hydrocarbons (PHAs). The glass fiber with
absorbed analytes is immersed into the capillary segment, and the absorbed
analytes are directly released into the CE buffer stream, and then enter into the
capillary. Neutral PHAs are separated by CE with borate CE running buffer con-
taining cyclodextrin and its derivatives. Figure 29.3B shows one example of the
CE separation. It can be seen that SPME-CE greatly lowers the detection limits
of CE for PHAs. The detection limit of pyrene is 8 ppb [26].
Furthermore, a SPME-CE interface that realizes a zero-dead volume con-
nection between SPME and CE has also been proposed recently (Fig. 29.3C) [27].
A 40 Am fiber coated with polyacrylate film is used for extraction and concentra-

970
 A E
Separation capillary
50 pm id, 350 pm od

c
Heat shrinkable
tubing 2 cm a
a
a

Capillary segment 1.6 cm

180 pm id, 380 pm od

Glassfiber150pm
diameter with
absorbed analytes,
A__
--
Nylon coating ' 6'' (b
on glass fiber

4
2 7 1 t

12 14 16 I 2 22
Time (min)

C -HV

(i3) Interface SPME device

Fig. 29.3. Schematics of on-column SPME-CE (A, C) and CE separation results of PAHs (B). In
(B), (a) and (b) are electropherograms obtained without and with on-column SPME concentra-
tion (by system shown as A), respectively. Standard sample solution of EPA610 is 40-fold (a) and
4000-fold (b) diluted. Peaks 1-15 are dibenz[a,h]anthracene, acenaphthylene, acenaphthene,
naphthalene, fluorene, anthracene, phenanthrene, chrysene, benz[a]anthracene, benzo(k)fluor-
anthene, fluoranthene, benzo[a]pyrene, pyrene, benzo[b]fluoranthene, indeno[1,2,3-cd]pyrene,
respectively. For details see Ref. [26]. (A) and (B) are reprinted with permission from Ref. [26].

tion of phenol samples. Then, the fiber is directly immersed into the inlet end of
a 75 C/m CE capillary. It has been used for the determination of various phenol
compounds.
Recently, SPME-CE has been further applied to investigate the chemical
composition of small living objects such as portions of plants and organs of
insects by combining with on-fiber fluorescence derivatization techniques and
laser-induced fluorescence detection [28]. Firstly, an SPME fiber was dipped
into a vial containing fluorescent derivatization reagent 4-fluoro-7-nitro-2,1,3-
benzoxadiazole (NBD-F, 2-3 mg/ml in ethanol) solution for 10 min with mag-
netic stirring at 1000 rpm. The fluorescent derivatization reagent was adsorbed
on the fiber. Secondly, the fiber was immersed into a small living object such as a
grape for 20 s. Amino acids such as glutamic acid in the grape was extracted and
concentrated on the fiber. Thirdly, the fiber was inserted into the headspace of a

971
4-ml amber vial containing 1 ml of triethylamine (TEA) at 60°C water bath, so
that the fluorescent derivatization reaction of amino acids occurred easily.
Finally, the fiber was immersed into the zero-dead volume SPME-CE interface,
followed by desorption with methanol and micellar electrokinetic capillary chro-
matography (MECC) separation with laser-induced fluorescence detection.

29.3.2 On-line electrophoretic concentration methods for CE

29.3.2.1 Sample stacking

The sample stacking technique [29-31] might be the most frequently used and
studied on-line concentration method in CE since it does not need any pre-
concentration component connected to the capillary. In sample stacking, sample
is usually prepared in a low concentration (low conductivity) buffer, while CE
running buffer is with relatively high concentration (high conductivity). It is
well known that the electrophoretic mobility of an ion in low conductivity matrix
is larger than that in high conductivity matrix. Therefore, when the ion electro-
migrates across a boundary between the low and high conductivity matrix, the
sudden slowing down of the electromigration velocity results in stacking (con-
centration) of the sample zone (Fig. 29.4). The length of the stacked sample zone
(L') is expressed as follows:
L' = L(Veof + Vepb)/(Vcof + Vep) (29.1)

where L, Vof, Veps, and Ver b are length of sample solution plug, electroosmotic
flow, electrophoretic velocities in sample plug and background buffer, respect-
ively. Then, the concentration factor F is expressed as follows:
F = (Veo f + Veps)/(Veof + Vepb) (29.2)

Equation (29.2) shows that a high concentration factor F can be obtained with a
small Vepb and large Vep,. In order to achieve large Veps and small Vepb, sample is
usually dissolved in a much lower conductivity buffer than the running buffer or
even in pure water. In some cases, organic solvent is added to the sample to
decrease the conductivity of the sample plug.

A Salllple plug Capillary

Fiu 9C2Q r'nurl I
sample stacking model eVp Q
for anions. (A) A o Q o o a +
sample plug (length L)
bounded with CE il >1 Backolaad buffer
running buffer in a
capillary before Sacked sample zone
stacking. (B) The B
sample zone (length
L') after crossing the +
boundary of sample
peauala/a- lg ouIce
plUg/rsauking Du,l Ie
after stacking. 3

972
 One merit of sample stacking is that it can inject a long sample plug into the
capillary. In some experiments, even more than half of the capillary is filled with
sample plug [32]. Either electroosmotic flow is adjusted by adding a modifier [33]
into the buffer or polarity switching [34] is employed after finishing the sample
stacking so that the long water plug will not affect the separation efficiency.
Sample stacking can also be accomplished during the electrokinetic injection
process when the inlet capillary is immersed into a sample vial with a low con-
ductivity matrix. This stacking mode is also called field amplification sample
injection [35]. Injection of a short plug of water before the sample is reported to
be able to enhance the electric field at the injection point and thus increase the
stacking effect [36].
The stacking technique has been further extended to concentration of neu-
tral analytes in MEKC separation [36,37]. A concentration factor up to 106 has
been reported [38]. This technique has been well reviewed [39]. Similar to sam-
ple stacking, other concentration methods such as sample self stacking [40],
acetonitrile stacking [41], sweeping [42], and the use of pH junction [43] are also
based on the moving velocity difference of ions across the boundary of the sample
plug and the running buffer.

29.3.2.2 CITP-CE
Capillary isotachophoresis is a unique electrophoresis mode where a sample plug
is sandwiched between a leading electrolyte (with a faster mobility than any ion
in the sample) and a terminating electrolyte (with a lower mobility than any ion
in the sample) in a capillary. When an electric field is applied across the capillary,
the resulting field is not homogeneous throughout the capillary. Separation
occurs between the boundaries based on the individual ion mobilities. As a
result, sample ions are separated in consecutive zones according to their mobili-
ties, and all zones migrate at the same velocity. An ion diffusing out of its zone
speeds up or slows down, depending on the velocity of the neighbouring zone it
encounters, thereby rejoining its focused zone. This CITP can also be on-line
coupled to CE.
Basically, there are two approaches to coupling CITP to CE. A simple
approach is to carry out CITP and CE in a single capillary [44]; this approach is
also called transient CITP-CE [9]. After filling the running buffer into the capil-
lary, a leading buffer (the running buffer can be used as the leading buffer in
some case), sample solution, and terminating buffer are subsequently injected
into the capillary from the inlet end. After applying a voltage across the two ends
of the capillary, CITP occurs in the first part including the inlet end of the capil-
lary. Then, concentrated sample zones by the CITP are separated by CE in the
remaining part of the capillary. This CITP-CE has been used for separation of
sugars [45], proteins [46] and other biochemical samples; concentration factor
can be up to 1000-fold [47]. The drawback of this technique is that CE resolution
might be decreased because of overloading in some cases.
Another approach is to carry out CITP and CE in two different capillaries

973
 HighVoiage Power Supply

I ,I

LE
TE EIzE ILE 3 TE =t
CZED
'tD G
LE
2 E iLE 4 TE LE

Fig. 29.5. Schematics of instrumentation for comprehensive CITP-CE (A) and its operation
principle (B). In (A), BR1, BR2, and BR3 represent buffer reservoirs 1, 2, and 3, respectively. V1
and V2 represent valves 1 and 2, respectively. In (B), CZED is CE detector; TE, LE, s represent
terminating electrolyte, leading electrolyte, and sample, respectively. For details see Ref. [51].
Reprinted with permission from Ref. [51].

that are connected either by a T-junction [48,49] or by immersing the inlet end of
the CE capillary (with small diameter) into the CITP capillary (with large inter-
nal diameter) [50]. Firstly, a voltage is applied across the CITP capillary for car-
rying out ITP concentration of sample ions. When the concentrated analyte
zones move to the inlet end of the CE capillary, a voltage is applied across the
CITP and CE capillaries for injecting the concentrated zones into the CE capil-
lary. Then CE separation is carried out. This CITP-CE has been used for separa-
tion of small anions [49], amino acids [51], and so on, a concentration factor as
high as 10 5 -10 5 has been reported [49]. However, this CITP-CE can only analyze
a few concentrated zones at a time to prevent CE overloading. Moreover, two
detectors are required (one for CITP and the other for CE). The CITP detector is
used to determine when the concentrated bands reach the injection end of the
capillary. The timing for injection of the concentrated zone into the CE capillary
is important [52].
Recently, a comprehensive CITP-CE coupling configuration (Fig. 29.5) has
been reported [52]. Firstly, the sample is focused in the CITP capillary. Second-
ly, the focused sample zone is injected into the CE capillary by electrokinetic
injection. Thirdly, the remaining sample zones are forced to move back away
from the injection end of the CE capillary by injecting leading electrolyte from
the syringe pump. Fourthly, CE separation is carried out. Comprehensive

974
CITP-CE involves repetitive runs of the above four steps. Since only small por-
tions of the concentrated zones are sequentially injected for CE separation, the
problem of overloading does not exist. The system has been evaluated using a
mixture of angiotensins. Under optimized conditions, a detection limit of ap-
proximately 5 nM can be achieved by injecting 10/ l of angiotensin solution [52].

29.3.2.3 On-column electrophoreticconcentration with a hollow fiber or

porous membrane
Figure 29.6 shows an electrophoretic concentration method using a hollow fiber
or porous membrane [53]. The separation capillary can be directly immersed
into a hollow fiber (or porous membrane) by etching the inlet end of the capillary
in HF solution [54]. During sample injection, a sampling voltage is applied across
the sample vial and the hollow-fiber holding vial as shown in Fig. 29.6A. Since
the hollow fiber does not permit macromolecular ions such as proteins pass
through, the macromolecular ions are concentrated in the hollow fiber. After a
certain period of the concentration, the hollow-fiber holding vial is immersed
into a buffer vial, and normal CE is followed. Figure 29.6B shows one example of
the concentration effect of the method. A concentration factor of 1000-fold is eas-
ily obtained for model proteins such as cytochrome C. The concentration factor
is proportional to sampling (concentration) time and sampling electric field.
This simple concentration principle has also been applied in microchip for-
mat CE [55]. A microfabricated injection valve incorporating a porous mem-
brane structure is schematically shown in Fig. 29.7. The injection tee and side

~B ~ ~~ 2 32
Sample concentration :0.05 mg/ml I Abs
Without concentration

Sample concentration: 0.0005 ml

' Conceintration time: min. \ I

Fig. 29.6. Schematic of the on-column electrophoretic concentration with a hollow fiber (A) and
CE results of proteins (B). Peaks 1, 2, and 3 are cytochrome c, lysozyme, and ribonuclease A,
respectively.

975
 buffers
A buffer C aay side
porous membrane 3
0
analyte side channel 100 P I

injection tee
of
: IaIwaste
separation
channel

Waste r

B buffer porous membrane

i
l.. /
/D/ SE

....
analyte main channel
(I f/:1X
rI
.side.channel

separation channel

Top view

PM
" sodium silicate
bonding interface
cover plat

substrate

50
Cross setionview time [Is]
Fig. 29.7. Schematics of a porous membrane microchip for DNA concentration and electro-
phoretic analysis (A) and porous membrane structure (B). (C) CCD images of analyte
concentrated for different times and injection after concentration. (D) Successive electro-
pherograms of a DNA PCR marker (50, 150, 300, 500, 750, and 1000 bp) after 150 s, 200 s, and
250 s of preconcentration. For details see Ref. [54]. Reprinted with permission from Ref. [54].

channel are separated from each other by a distance of 3-12 mm and connected
by a thin porous silicate layer. The porous silicate layer allows the passage of cur-
rent to establish an electrical connection between the separated channels but
prevents large molecules (e.g., DNA) from traversing the porous silicate layer.
Therefore, when a voltage is applied between the analyte and side reservoirs,
DNA samples will be concentrated in front of the porous membrane (Fig. 29.7C).
Then, the concentrated DNA samples are injected in the separation channel by
switching the voltage. Figure 29.7D shows that the concentration factor depends
on the concentration time.
Comparing with stacking and CITP, one merit of the on-column electropho-
retic concentration method is that it does not need to use discontinuous buffers,
i.e., the sample can be directly prepared in the CE running buffer.

976
29.3.3 On-line membrane cleanup for CE

A porous membrane or a hollow fiber can also be used for cleanup of analytes
from the sample matrix to prevent particulate material or high molecular mass
matter from entering the CE capillary. Figure 29.8 shows one configuration of
the coupling between capillary and tubular membrane. The tubular membrane
can be used as sampling interface. Sample solution flowed into the jacket is
introduced by diffusion and/or permeation through the membrane. The jacket
Sample Waste

Wash solution 'K._J

-' \Membe
Capillary Capillay

F . l out
Dram Capilln
a r

Fig. 29.8. Schematic of a membrane sampling interface for CE.

high positive voltage

blank 1.2 h 2,3 h 3.2 h 3.9 h 18.0 h

dialysate

I nA F 5 5 5 0 5

0 5 0 s 0 5 0o 5 6 D 5
Tite (anin)

Fig. 29.9. Schematic of the on-line microdialysis-CE-electrochemical detection system (A) and
electropherograms of transdermally delivered nicotine in microdialysate after administration of
nicotine patch (B). Details see Ref. [60]. Reprinted with permission from Ref. [60].

977
 can be connected to a valve that allows easy replacement of the solution in the
jacket. This membrane sampling allows CE direct determination of gaseous
samples, low molecular weight constituents in blood plasma, ionisable/non-ionic
solutes in wastewater [561.
A dialysis system can also be on-line connected to CE by a flow-gating
approach [57,58] or via a micro-injection valve-based interface [59,601. Figure
29.9 shows one example of the on-line coupling of in ivo microdialysis with
CE/electrochemicaldetection [61]. As shown in Fig. 29.9A, a microdialysis probe
is immersed into a rabbit body. Outlet of the microdialysis probe is connected to
a micro-injection valve (60 nl) using a capillary. Injection of analytes into the CE
capillary is accomplished electrokinetically by applying the separation voltage at
the gap junction interface. Figure 29.9B shows monitoring results of nicotine
concentration in the animal after administration of the nicotine patch.
Figure 29.10 shows one example of on-line dialysis-CE coupling with a
FIA-CE interface for determination of small anions [62]. The sample is pumped
into the dialysis unit, and small anions are dialysed into an acceptor stream of
water. The acceptor steam is led into an injector loop. When injection takes place
the sample is transferred to the electrolyte. The injected sample then passes by
the end of the capillary situated in the FIA interface and a small fraction is
electrokinetically introduced into the capillary where separation takes place.
Experimental results show that this system can be used for determination of
small anions in various drinks such as milk with good reproducibility.
Capillary isoelectric focusing (CIEF) is known to be an effective tool with
high separation resolution in biomolecular analysis [63]. In particular, for the
recently developed whole column imaged CIEF [64,65], it greatly decreases anal-
ysis time, and enables direct observation of the isoelectric focusing and separa-
tion process. However, the protein sample is usually required to desalt before the
CIEF experiment. Figure 29,11 shows an on-line desalting for CIEF with a hol-
low fiber [66]. The protein sample is injected into the hollow fiber, which is
immersed into an ampholyte solution for desalting. After desalting, the protein
sample is flowed into the capillary and whole column imaging CIEF is carried
out [661.

29.4 SAMPLE PREPARATION FOR PROTEOME ANALYSIS

A combination of high-resolution two-dimensional (2-D) polyacrylamide gel

electrophoresis, high sensitive biological mass spectrometry, and the rapidly
growing protein and DNA databases has paved the way for high-throughput
proteomics [67]. Sample preparation also plays an important role in the success
of proteome analysis. Adsorbing columns [68,69], reverse phase chromato-
graphic beads [70] and silica-based "restricted access" support ADS C column
[71] have been used for sample purification and concentration for proteome
analysis. Recently, high throughput sample preparation technique using matrix
a-cyano-4-hydroxycinnamic acid (CHCA) and prestructured sample supports

978
 a) (
PTFE tubing

- ·
b)

3 4 5 6 7 8 9 rmin

Fig. 29.10. Schematics of the on-line dialysis-FIA-CE system (a) and CE results (b). In (a) A is a
cross-sectional view of the FIA-CE interface. S, A, E, M, D, V1, V2, W, C, Pt, and HV represent
sample, acceptor stream, electrolyte, dialysis membrane, UV detector, injection valve in filling
position, injection valve in injection position, waste, capillary, platinum electrodes, and high-
voltage supply, respectively. In (b), A and B are electropherograms of 24 anion standard mixture
without and with dialysis, respectively. Peaks 1-20 are S2032 Cl-, SO,', NO 3 , oxalate, HS,
SO3, citrate, maleinate, fumarate, F-, tartrate/succinate/formate/malate, HPO4,% HCO3 ,
acetate, OH, propionate, lactate, butyrate, and benzoate, respectively. For details see Ref. [61].
Reprinted with permission from Ref. [61].

was reported for matrix-assisted laser desorption/ionization mass spectrometry

(MALDI-MS) peptide analysis in proteomics [72]. The technique integrates
sample purification based on the affinity of microcrystalline CHCA for peptides,

979
 ailaries

Ampholytes '
solution i

Fig. 29.11. Schematic of on-line desalting with a dialysis hollow fiber for CIEF. For details see
Ref. [65]. Reprinted with permission from Ref. [65].

thereby simplifying the analysis of crude peptide mixtures. It also allows

preparation of large numbers (up to several hundreds) of samples with little
effort and without the need for automation.
Ultramicroanalyses such as single-cell analysis have also been reported
[73-75]. In these reports, miniaturized micropipettes are used to transfer the
intracellular content of a single cell directly into a separation capillary for follow-
ing CE determination of cellular analytes. Very recently, direct injection of
whole cell suspensions of bacteria without prior analytes separation for
electrospray ionization mass spectrometry has also been reported [76]. The
results show that the direct injection of whole cell suspensions without prior
analyte separation is highly informative and can be useful in microbial identifi-
cation with discrimination capability at the subspecies level [76].

29.5 SUMMARY

Sample preparation including cleanup and preconcentration is very important

to the widespread application of CE in various fields and various samples. All the
above on-line sample cleanup and preconcentration methods can greatly lower
the detection limit of CE, and some of them have been successfully applied to
practical samples. However, direct injection of a sample with complex matrix
(e.g., biological samples) is still limited. More advanced and practical sample
preparation techniques for CE are required. Since more and more researchers
are using CE, and more and more samples are analyzed with CE, it is expected
that new sample preparation techniques will appear in the near future.

REFERENCES

1 S.F.Y. Li, Capillary Electrophoresis,Principles,Practice,and Applications. Elsevier,

Tokyo, 1992.
2 F.J. Harrison and A. van den Berg (Eds.), Proceedings of the ATAS'98 Workshop.
Kluwer, Dordrecht, 1998.
3 M. Albin, P.D. Grossman and S.E. Moring, Anal. Chem., 65 (1993) 489A.
4 J.P. Chervet, R.E.J. Van Soest and M. Ursem, J. Chromatogr.543 (1991) 439.
5 T. Wang, J.H. Aiken, C.W. Huie and R.A. Hartwick, Anal. Chem., 63 (1991) 1372.

980
 6 X. Xi and E.S. Yeung, Appl. Spectrosc., 45 (1991) 1199.
7 X.-Z. Wu, A. Hosaka, E. Kobayashi and T. Hobo, J. Chromatogr.A, 726 (1996) 205.
8 Dankova, D. Kaniansky, S. Fanali and F. Ivanyi, J. Chromatogr.A, 838 (1999) 31.
9 J.R. Veraart, H. Lingeman and U.A.Th. Brinkman, J. Chromatogr. A, 856 (1999)
483.
10 L. Farry, D.A. Oxspring, W.F. Smyth and R. Marchant, Anal. Chim. Acta, 349 (1997)
221.
11 S. Morales and R. Cela, J. Chromatogr.A, 896 (2000) 95.
12 M. Hernandez, F. Borrull and M. Calull, J. Chromatogr., B: Biomed. Sci. Appl., 742
(2000) 255.
13 R.L. Sheppard and J. Henion, Electrophoresis, 18 (1997) 287.
14 W. Zhan, T. Wang and S.F.Y. Li, Electrophoresis,21 (2000) 3593.
15 R. Zhang and S. Hjerten, Anal. Chem., 69 (1997) 1585.
16 S. Hjerten, J.-L. Liao and R. Zhang, J. Chromatogr.A, 676 (1994) 409.
17 J.-L. Liao, R. Zhang and S. Hjerten, J. Chromatogr.A, 676 (1994) 421.
18 K. Dusan, K. Eva, M. Vlasta and M. Marian, Electrophoresis,18 (1997) 260.
19 A.J.J. Debets, M. Mazereeuw, W.H. Voogt, D.J. van Iperen, H. Lingeman, K.-P. Hupe
and U.A.Th. Brinkman, J. Chromatogr.A, 608 (1992) 151.
20 I. Morita and J.I Sawada, J. Chromatogr.A, 641 (1993) 375.
21 M. Petersson, K. Wahlund and S. Nilsson, J. Chromatogr.A, 841 (1999) 249.
22 A.J.J. Debets, M. Mazereeuw, W.H. Voogt, D.J. van Iperen, H. Lingeman, K.-P. Hupe
and U.A.Th. Brinkman, J. Chromatogr.A, 608 (1992) 151.
23 M.A. Strausbauch, J.P. Landers and P.J.Wettstein, Anal. Chem., 68 (1996) 306.
24 S. Naylor and A.J. Tomlinson, Biomed. Chromatogr., 10 (1996) 325.
25 E. Rohde, A.J. Tomlinson, D.H. Johnson and S. Naylor, J. Chromatogr.B, 713 (1998)
301.
26 A.L. Nguyen and J.H.T. Luong, Anal. Chem., 69 (1997) 1726.
27 C-W. Whang and J. Pawliszyn, Anal. Commun., 35 (1998) 353.
28 A. So, Sampling Small Object with Solid Phase Microextraction. Master Thesis, 2000,
University of Waterloo, Ontario, Canada.
29 R.L. Chien and D.S. Burgi, Anal. Chem., 64 (1992) 489A.
30 D.S. Burgi and R.L. Chien, Anal. Chem., 63 (1991) 2042.
31 R.L. Chien, Anal. Chem., 63 (1991) 2866.
32 Y. He and H. K. Lee, Anal. Chem., 71 (1999) 995.
33 J.P. Quirino and S. Terabe, Electrophoresis,21 (2000) 355.
34 M. Albert, L. Debusschere, C. Demesmay and J.L. Rocca, J. Chromatogr. A, 757
(1997) 281.
35 R.L. Chien and D.S. Burgi, J. Chromatogr., 559 (1991) 141.
36 Z. Liu, P. Sam, S. Sirimanne, P. McClure, J. Grainger and D. Patterson, J. Chroma-
togr. A, 673 (1994) 126.
37 J.P. Quirino and S. Terabe, J. Chromatogr.A, 781 (1997) 119.
38 J.P. Quirino and S. Terabe, Anal. Chem., 72 (2000) 1023.
39 J.P. Quirino and S. Terabe, J. Cap. Electrophor., 4 (1997) 233.
40 P. Gebauer, W. Thormann and P. Bocek, J. Chromatogr., 608 (1992) 47.
41 Z.K. Shihabi, J. Cap. Electrophor., 2 (1995) 267.
42 J.P. Quirino and S. Terabe, Science, 282 (1998) 465.
43 P. BritzMckibbin, J. Wong and D.D.Y. Chen, J. Chromatogr.A, 853 (1999) 535.
44 F. Foret, E. Szoko and B.L. Karger, J. Chromatogr., 608,(1992) 3.
45 C.-C. Wang, W.P. McCann and S.C. Beale, J. Chromatogr.B, 676 (1996) 19.
46 J. Bergmann, U. Jaehde, M. Mazereeuw, U.R. Tjaden and W. Schunack, J. Chroma-
togr. A, 734 (1996) 381.
47 J.-L. Liao, R. Zhaang and S. Hjerten, J. Chromatogr.A, 676 (1994) 421.

981
48 A.W. Moore Jr. and J.W. Jorgenson, Anal. Chem., 67 (1995) 3456.
49 D. Kaniansky, I. Zelensky, A. Hybenova and F.I. Onuska, Anal. Chem., 66 (1994)
4258.
50 N.J. Reinhoud, A.P. Tinke, U.R. Tjaden, W.M.A. Niessen and J. Van der Greef, J.
Chromatogr., 627 (1992) 263.
51 D. Kaniansky and J. Marak, J. Chromatogr., 498 (1990) 191.
52 S. Chen and M.L. Lee, Anal. Chem., 72(2000) 816.
53 X.-Z. Wu, A. Hosaka and T. Toshiyuki, Anal. Chem., 70 (1998) 2081.
54 X.-Z. Wu and 'T. Kasashima, Anal. Sci., 16 (2000) 329.
55 J. Khadurina, S.C. Jacobson, L.C. Waters, R.S. Foote and J.M. Ramsey, Anal. Chem.,
71 (1999) 1815.
56 L. Bao and P.K. Dasgupta, Anal. Chem., 64 (1992) 991.
57 M.W. Lada and R.T. Kennedy, Anal. Chem., 68 (1996) 2790.
58 M.W. Lada, G. Schaller, M.H. Carriger, T.W. Vickroy and R.T. Kennedy, Anal. Chim.
Acta, 307 (1995) 217.
59 B.L. Hogan, S.M. Lunte, J.F. Stobaugh and C.E.Lunte, Anal. Chem., 66 (1994) 596.
60 C.E. Lunte, D.O. Scott and P.T. Kissinger, Anal. Chem., 63 (1991) 773A.
61 J. Zhou, D.M. Heckert, H. Zuo, C.E. Lunte and S.M. Lunte, Anal. Chim. Acta, 379
(1999) 307.
62 P. Kuban and B. Krlberg, Anal. Chem., 69 (1997) 1169.
63 S. Hjerten and M. Zhu, J. Chromatogr., 346 (1985) 265.
64 J. Wu and J. Pawliszyn, Am. Lab., 26 (1994) 48.
65 J. Wu and J. Pawliszyn, Anal. Chem., 64 (1992) 224.
66 J. Wu and J. Pawliszyn, Anal. Chem., 67 (1995) 2010.
67 K. Gevaert and J. Vandekerckhove, Electrophoresis,21 (2000) 1145.
68 H. Zhang, P. Andren and R. Caprioli, J. Mass Spectrom., 30 (1995) 1768.
69 P. Courchesne and S. Patterson, BioTechniques, 22 (1997) 244.
70 K. Gevaert, H. Demol, M. Puype, D. Broekaert, S. Deboeck, T. Houthaeve and J.
Vandekerckhove, Electrophoresis, 18 (1997) 2950.
71 C. Bratt, C. Lindberg and G. Marko-Varga, J. Chromatogr., 909 (2001) 279.
72 J. Gobom, M. Schuerenberg, M. Mueller, D. Theiss, H. Lehrach and E. Nordhoff,
Anal. Chem., 73 (2001) 434.
73 F. Han and S.J. Lillard, Anal. Chem., 72 (2000) 4073.
74 S.N. Krylov, E. Arriaga, Z. Zhang, N.W.C. Chan, M.M. Palcic and N.J. Dovichi, J.
Chromatogr. B, 741 (2000) 31.
75 E.S. Yeung, J. Chromatogr.A, 830 (1999) 243.
76 S. Vaidyanathan, J. Rowland, D. Kell and R. Goodancre, Anal. Chem., 73 (2001) 4134

982
Chapter30

Recent progresses in membrane separation

science and technologies: a review
Takeshi Matsuura

30.1 INTRODUCTION
It is not possible to cover the entire scope of progress in membrane technologies
in a single chapter since, nowadays, they touch almost all aspects of human life:
e.g., membrane technologies are penetrating rapidly into biotechnology and bio-
medical applications. Serious attempts are currently being made to develop
membranes that allow enhancement in efficiencies of fuel cells and batteries.
The author's experience is limited to research in the area of conventional mem-
branes and membrane separation processes, hence this article is focused on
progress made in reverse osmosis, ultrafiltration, nanofiltration, microfiltra-
tion, gas and vapor separation and pervaporation. However, even in this rather
limited area of membrane applications, remarkable progress has been achieved;
i.e. recovery of drinkable water from seawater has been increased by developing
high pressure resistant membranes; membranes operable at ultra-low operating
pressures have been developed; performance of gas separation membranes has
been significantly improved; inorganic membrane module for pervaporation has
been commercialised; a number of new techniques and instruments are being
used daily to characterize the membrane surface, which will have impact not
only on separation membrane performance but also on biotechnology and bio-
medical applications of membrane surface in the future; new methods for the
measurement of nano-pores have been developed and used; transport equations
of charged membranes have enabled the prediction of nanofiltration mem-
branes; and a new hybrid system is presently used for the treatment of heavily
polluted wastewater. This chapter also includes a brief outline of a new extrac-
tion method using reverse micelles that seems to have potential in applications
in biotechnology.

30.2 PROGRESS IN MEMBRANE DEVELOPMENT

30.2.1 Progress in reverse osmosis

Since the announcement of the first cellulose acetate membrane for seawater
desalination by Loeb and Sourirajan in 1960, reverse osmosis, nanofiltration,
ComprehensiveAnalytical Chemistry XX/XVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved
ultrafiltration and microfiltration membranes have been used increasingly in
many different aspects of water treatment. Membrane research and develop-
ment have allowed the evolution of the use of membranes in many aspects of
human life. Seawater desalination remains, however, the central subject for the
development of membrane technologies, therefore a brief review is given on the
progress in membrane development for seawater and brackish water
desalination.
Two opposite trends are observed in the development of novel reverse
osmosis membranes: one is to develop reverse osmosis membranes durable at
high operating pressures and the other is to develop membranes that are
operable at as low pressures as possible. The former type of membrane is to be
used in the treatment of solutions with high salt concentrations (e.g., sea water
desalination) and the latter type of membrane is to be used in the treatment of
solutions with low salt concentrations (e.g., desalination of brackish water and
production of ultra-pure water) [1]. In seawater desalination, for example,
attempts are being made to increase water recovery from the conventional 40%
to 60%, which inevitably requires high pressure reverse osmosis operation to
overcome very high osmotic pressure of concentrated brine. Hence, the first type
of membrane is necessary. On the other hand, low-pressure reverse osmosis
operation is desired to reduce the running cost of ultra-pure water production.
When the water recovery by a reverse osmosis (RO) module is 40%, the
osmotic pressure of the concentrate is 4.5 MPa, while it would become 7.0 MPa
when the water recovery is 60%. This means that while an operating pressure of
5.5-6.5 MPa is required for 40% water recovery, 8.0-9.0 MPa would be required
for 60% water recovery. To enable the operation of an RO module at a pressure
as high as 9.0 MPa is an engineering challenge in many respects, but the
development of RO membranes durable at 9.0 MPa is of course the most
important.
After a thorough analysis of the performance data of a conventional TFC
membrane with a thin selective layer of aromatic polyamide, Kawada concluded
that the flux decline of the membrane at 9.0 MPa was caused not by the
compaction of the thin selective layer but by the compaction of the surface layer
of the porous support; i.e., the flux of the porous support membrane decreased by
95% as a result of compaction. It was also found from a SEM picture that the
pores on the surface layer of the porous support membrane had been crushed by
compaction Based on the above findings, Kawada attempted to make a porous
support membrane with a surface layer possessing a large number of small
pores, which would enable the compaction of the porous support membrane to be
minimized. It was also attempted to make the pore sizes as uniform as possible.
Then, RO membranes were prepared by coating the top surface of the newly
made porous support membrane with aromatic polyamide by interfacial poly-
condensation. The results of the compaction tests are summarized in Fig. 30.1
which shows that the flux decline of the conventional NTR-70SWC TFC mem-
brane was not severe when the membrane was used at 5.6 MPa, but very severe

984
 A

0
E
E

a
.

.
no

Operating time (h)

Fig. 30.1. Flux versus operating time for conventional and newly developed high pressure RO
membrane (from Ref. [1]).

COCI COCI
TMC A TPC

Controlof coc Cntrol of

Membrane ISurface Charge
Thickness NH2 COCI O
COl CNH2 COOH End Acid

0 0 0, 0\ Organic Layer
Highly rosslinked Aromatic olyamide Membrane

Diffusion of Aqueous Layer

Amines
NH2 NH2
TAB m-PDA
H2N JNH2 'NH2

Fig. 30.2. Formation of composite polyamide membrane by interfacial polycondensation (from

Ref. [21).

at the operating pressure of 9.0 MPa. On the other hand, the compaction of the
novel NTR-70SWH membrane was moderate even at an operating pressure of
9.0 MPa.
Reverse osmosis membranes that exhibit high fluxes at low operating pres-
sures are also prepared by interfacial poly-condensation [2]. As shown in Fig.
30.2, a mixture of aromatic dicarboxylic acid chloride and tricarboxylic acid chlo-
ride dissolved in an organic solvent, which is immiscible to water, is brought into
contact with an aromatic diamine dissolved in water. An interfacial poly-
condensation will take place in situ, forming a very thin layer of aromatic poly-
amide at the top of the support layer. This is a well established method of prepar-
ing a thin film composite membrane. In a new approach, some additives were

985
 Low Pressure RO-I Low Pressure RO-i Ultra Low Pressure RO Super UftraLow Pressure RO

Fig. 30.3. Cross-sectional pictures of newly developed ultra-low pressure RO membranes (from
Ref. [2]).

99.9

99.8

1 99.5

t 99.0

95.0

90.0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Normalized water flux, m3/m2 day MPa

Fig. 30.4. Progress in the performance of RO membrane for seawater desalination (from Ref.
[2]).

brought into the monomer solution to increase the interfacial area. Figure 30.3
shows cross-sectional pictures taken near the surfaces of advanced low-pressure
reverse osmosis membranes. The increase in the effective surface area can be
observed in the figure, which resulted in an increased membrane flux. Figure
30.4 illustrates the progress made in the performance of reverse osmosis mem-
brane during the past three decades. Remarkable progress has been seen both in
selectivity and water flux since the 1970s. In particular, an order of magnitude
increase has been achieved in water flux of reverse osmosis membranes, which
enabled RO operation under ultra-low pressures. Currently, no limit is seen in
the horizon for the improvement of the RO membrane flux.
There is another interesting report in which the performance of a thin film
composite membrane was enhanced by surface modification [3]. Mukherjee et al.
applied surface chemical reaction on top of an aromatic polyamide thin film com-
posite membrane (FT-30). Membranes were soaked at controlled temperatures

986
in solutions of various concentrations of hydrofluoric acid (HF), fluorosililic acid
(FSA) and their mixtures. The membranes were taken out after various inter-
vals of time, rinsed with deionized water and then tested for their RO perfor-
mance. A dramatic effect of the surface modification on the membrane
performance was observed when the membrane was soaked in 15 wt.% HF solu-
tion for 7 days. The flux increased from 3.5 1/m 2 h to 18 1/m 2 h while NaCl separa-
tion increased from 94.5 to 95.3% under the same operating pressure of 250 psig
(1724 kPag). According to X-ray photoelectron spectroscopy (XPS), the fluorine
ratio (F/C) at the membrane surface increased from zero of fresh FT-30 mem-
brane to 0.012 of a membrane immersed in 15 wt.% HF solution for 4 days. It fur-
ther increased to 0.044 when the membrane was immersed in 15 wt.% HF
solution for 75 days. The SEM picture revealed the thinning down of the polymer
network of the selective skin layer due to the etching of the surface by HF. Thus,
both the chemical modification of the surface and the thinning of the selective
skin layer contributed to the dramatic increase of the flux, while NaCl separa-
tion was practically unchanged.

30.2.2 Progress in gas separation membranes

Progress in the development of gas separation membranes is occurring in two

areas: one is in the search of membrane materials and the other in membrane
preparation using conventional polymeric materials. New polymeric materials
that exhibit both high permeability and high permselectivty are constantly
being searched for. However, because of higher mobility associated with macro-
molecular chains, a limit in the selectivity by polymeric membrane materials is
expected, hence the major activity in the search for materials in the last decade
has been among inorganic materials. For the preparation of high performance
gas separation membranes, on the other hand, the phase inversion-technique
seems still to be in the main stream, especially for the preparation of high flux
gas separation membranes.
Molecular sieve membranes enable the separation of gaseous mixtures by
difference in sizes of component gases. There are currently two types of
molecular sieve membranes, the one is zeolite membranes and the other carbon
molecular sieve (CMS) membranes. While commercial zeolite membranes are
not yet available for gas separation, commercial CMS membranes are available
as a product of Carbon Membrane Inc., Israel. Hence, the development of
molecular sieve membranes based on carbon materials is briefly outlined.
A review article written recently by Suda and Haraya gives a summary of
CMS membranes [4]. A CMS membrane can be prepared by coating a thin layer
of polymer precursor on a porous support, followed by pyrolysis. Another
method is to prepare an asymmetric membrane of polymer precursor by phase
inversion technique, which is also followed by pyrolysis. A dense homogeneous
membrane can also be subjected to pyrolysis to prepare a dense homogeneous
CMS membrane. Precursors so far used are polyfurfuryl alcohol, thermosetting

987
 100
: :
":.']'T;' tradeoff f
T'"1997 "'// , ...
...
for OMS membl no

.. ....
: ::..... o si

0u
0

1
104 102 10' 10° 10' 10' 10' 10o
Permeability of 02 [Barrer]

Fig. 30.5. Oxygen/nitrogen selectivity versus oxygen permeability (from Ref. [41).

polymers and phenol formamide. More recently, polyimide precursors were

used. The pyrolysis temperature and the heating rate affect the performance of
the resulting membrane. The pyrolysis environment is another important
factor. Figure 30.5 compares the gas permeability and selectivity characteristics
of CMS membranes with those of polymeric membranes. As can be seen in the
figure, CMC membrane materials shifted the points to the direction of higher
permeability and higher selectivity (towards upward right); therefore, the
intrinsic properties of CMS materials are better than those of polymeric
materials. However, the major drawback of CMS materials (and that of zeolite
materials as well) is the poor processability. Construction of a large separation
unit is a serious engineering problem.
Asymmetric gas separation membranes were prepared by Kawasaki et al. by
dry-wet phase inversion technique based on 6FDA-APPS polymer (2,2'-bis(3,4-
dicarboxyphenyl)hexafluoropropane dianhydride (6FDA) and bis[4-(4-amino-
phenoxy)phenyl]sulfone (APPS)) [5]. A polymer solution with a composition of
polyimide (12 wt%), methylene chloride (55 wt%), 1,1,2-trichloroethane (23
wt%) and butanol (10 wt%) was cast on a glass plate to a thickness of 250 /m.
After solvent evaporation of 15-600 s, the cast film was immersed in a methanol
bath for gelation. The membrane was washed by methanol for 12 h, air-dried for
24 h and dried in a vacuum oven for 15 h at 150'C. The gas permeation
experiments were performed with membranes the effective area of which was
1.0 cm 2 at a feed pressure of 76 cmHg (100 kPa). Carbon dioxide permeance of

988
TABLE 30.1
Permeance data for fluorinated polyimide membranes (from Ref. [5])
Membrane Permeance (GPU a ) Permeance ratio

CO2 02 CO,/CH,4 O
2 N2
6FDA-6FAP Asymmetric 1490 345 43 5.2
6FDA-6FAP Denseb 0.75 0.16 36 4.5
6FDA-DDS Asymmetric 0.57 0.15 144 12
6FDA-DDS Dense b 0.0032 0.0048 116 8.5
al1GPU = 1 x 10 - 6 cm 3 STP/cm 2 s cmHg = 3.349 x 10-10 mol/m2 s Pa.
bMembrane thickness = 50 xm.

350 GPU (1172 x 10 -1° mol/m 2 s Pa) was obtained with CO2/CH 4 selectivity of 40.
This high selectivity indicates that the thin skin layer, the thickness of which
was calculated to be 25 nm, was defect-free. Moreover, the selectivity of the
asymmetric membrane was higher than that of the dense homogeneous
membrane with a thickness of 50 gm. Higher selectivity of the thinner film is
probably due to the formation of a more highly ordered polymer structure as the
film thickness decreases. Table 30.1 shows the results of gas permeation experi-
ments for asymmetric membranes as well as homogeneous membranes prepared
from other polyimides. The table also shows that the permeability ratio in-
creased from 36 (which corresponds to the dense homogeneous membrane) to
43, as the thickness of the top skin layer decreased. They have also reported an
asymmetric membrane with CO, permeance of 1490 GPU (4990 x 10 -1° mol/m2 s
Pa) with CO)CH 4 permeance ratio of 43. It may be difficult to make a membrane
of a large area with sufficiently high pressure resistance according to their
method, but Kawakami's work indicates that there is practically no limit for the
permeation flux of gas separation membranes.

30.2.3 Progress in pervaporation membranes

Zeolite membranes have been developed on a small laboratory scale for gas
separation, separation of organic solvents and for pervaporation. Commercial
zeolite membranes are now available for pervaporation. Kita et al. reported
recently on the preparation of X- and Y-type zeolite membranes and their
pervaporation performance [6]. Zeolite membranes were grown hydrothermally
on the surface of a porous cylindrical mullite support. The aluminosilicate gel
was prepared by mixing an aqueous sol (NaOH 14% and SiO, 27%) and an
alkaline aluminate solution prepared by dissolving sodium aluminate (Al/NaOH
= 0.81) and sodium hydroxide in distilled water. After gel formation and aging
(for NaY only), the gel was poured into a reaction vessel fitted with a condenser
and a heater, and then a porous support tube, the surface of which was treated
with seed crystals of zeolite, was immersed in the gel. After hydrothermal
treatment at 100°C for a specific period the coated tube was washed with water

989
TABLE 30.2
Some pervaporation data pertinent to NaX and NaY zeolite membranes (from Ref. [6])
Zeolite Feed mixture (A/B) 10 wt.% of A T Separation factor Flux
type (IC) (A/B) (kg/m2 h)
NaX Water/ethanol 75 360 0.89
Methanol/benzene 50 24 1.25
Methanol/MTBE 50 320 0.26
NaY Water/ethanol 75 130 1.59
Methanol/benzene 50 3800 0.93
Methanol/MTBE 50 3300 1.11
Methanol/MTBE 60 5300 1.70
Ethanol/ETBE 65 1300 0.48
Ethanol/benzene 60 930 0.22
Ethanol/cyclohexane 60 1000 0.27

and dried. The pervaporation data of NaX and NaY zeolite based membranes are
given in Table 30.2. It is shown that membranes are water selective for the
separation of water/alcohol mixtures and both flux and selectivity of these
membranes are very high. It is thought that mass transfer occurs through
microporous cavity in zeolite material and inter-crystalline pores. Table 30.3
summarizes the pervaporation data of NaY zeolite membrane together with
those of polymeric membranes. Methanol and MTBE were separated in the
above experiments. It is obvious that the flux of the zeolite membrane is
extremely high compared with the polymeric membranes. Commercial zeolite
membrane is now available as a product of The Smart Chemical Company. The
membrane combines both a well-defined pore size (0.42 nm) and a high porosity
(47%) and is used for dehydration of organic solvents at a temperature as high as
100°C. The volume of organic solvent treated is from 500 ml to 3 1.

30.2.4 Surface modification by surface modifying macromolecules

(SMM)

Membrane surface plays an important role in the membrane performance, as

manifested by the structure of composite membranes that generally comprise
two distinct layers made of two different polymeric materials; i.e. the thin
surface layer that governs the selectivity and flux of composite membranes and
the thick substrate layer that is porous and provides the mechanical strength.
Hence many attempts have been made to modify the membrane surface aimed at
improving membrane performance. The surface modification methods include:
in situ polymerisation, coating, grafting, plasma deposition, CVD, chemical
modification, UV treatment, etc. Many of the surface modification methods are
complicated and require at least an additional step in the membrane preparation
process.

990
TABLE 30.3
Pervaporation performance of various membranes (from Ref. [61)
Membrane MeOH in feed T Separation factor Flux
(wt.%) (°C) (MeOH/MTBE) (kg/m2 h)
Cellulose acetate 0.83-6.9 23-49 14-454 0.05-1.41
Poly(vinyl alcohol) 5-30 45 3-4 0.12-0.16
BPDA polyimide 4.1 60 1400 0.6
Poly(phenylene oxide) 1-20 22 8-24 0.18-0.65
Nafion 417 3.2-5.3 50 35 0.64
Poly(styrene-co- styrene-sulfonic 5.0-14.3 25 25,000-35,000 0.001-0.02
acid) in Mg form
Silicalite/ stainless steel 5-50 (vol.%) 30-50 4-9 0.08-0.12
NaY/alumina 1-50 35-60 2,000-24,000 1.0-2.3

Recognizing the drawbacks of conventional surface modification methods,

the University of Ottawa and Toronto University team has proposed an alterna-
tive approach for membrane surface modification [7]. It is known that surface
active additives in a bulk phase migrate to the surface and change the surface
properties of the material of the bulk. A species is surface active if it migrates to
the surface of the phase in which it is dispersed. It usually has an amphipathic
structure, consisting of low and high polar components. The driving force for
spontaneous surface migration is the tendency to minimize interfacial energy,
which is universal to all condensed phase of matter. However, the kinetics of this
phenomenon is slower in polymeric systems than in low viscosity liquid systems
and has not been studied very well. In a polymer blend, thermodynamic incomp-
atibility between polymers usually causes demixing of polymers to occur. If the
polymer system is equilibrated in air, the polymer with the lowest surface energy
(hydrophobic polymer) will concentrate at the air interface and reduce the
system's interfacial tension as a consequence.
Based on the above concept, surface modifying macromolecules (SMM) were
synthesized and blended into polymer solutions. During the process of casting
the polymer solution into a film and solvent evaporation that follows, SMM
migrates to the membrane surface, rendering the surface of the membrane
ultimately obtained more hydrophobic than the bulk membrane phase (Fig.
30.6). The type of surface modifying macromolecules developed by the Univer-
sity of Ottawa/University of Toronto team has a co-polymeric nature. It has an
amphipathic structure consisting of a main polyurethane chain terminated with
two low polarity polymer chains. Since the surface characteristics are largely
determined by the low polarity components, they can be chosen to give a specific
property. It is preferable to use a fluorine based component due to additional
features such as surface lubrication, reduced fouling and increased chemical
resistivity associated with the carbon-fluorine (C-F) bond. Figure 30.7 shows the

991
 OO00000000 SURFACE O0- 00~* 0**
O000O00000 00-0000000
00000000 BULK Q.QQ*QQQQ*
0000000000000
0000000 000
000000004 0000000000
*0-000-00 '000000000
(A) (C)

*0000 Q 00 SURFACE

(B) j0 fcl BULK

0000
0000000
Fig. 30.6. A schematic diagram illustrating SMM migration. Closed circles, SMM molecules; open
circles, base polymer molecules; A, time zero; B, time in-between; C, time infinity (from Ref.
[101).

molecular structure of a typical SMM. The solubility in the base polymer can
depend on the additive's average molecular weight, its molecular structure and
its low polar component, which can be controlled by varying factors such as the
main chain length, the number of C-F bond and the type of chemicals used to
synthesize the main chain [7].
SMM was blended in polyethersulfone (PES) membrane. It was found that
the contact angle of the surface increased from 76 ° of PES to 116 ° by blending
SMM, which value is nearly equal to that of Teflon. XPS analysis also confirmed
that SMM was concentrated at the surface of the membrane. These SMM
blended PES membranes were found to remove chloroform effectively from
water by pervaporation [7]. They were found to be less fouled in the treatment of
cutting machine oil/water emulsion by ultrafiltration [8]. It was also found that
the mechanical strength of PES membranes is generally enhanced by blending
SMM into PES [9]. The time required for the SMM to migrate to the membrane
surface was measured and the kinetics of the surface migration was established
[10]. The change in fluorine content at the surface is given in Fig. 30.8 as the
functions of evaporation time and cast membrane thickness.
An attempt is currently being made to develop microfiltration membranes
with hydrophobic surface. These membranes are expected to be useful for the
purpose of membrane distillation. A similar technique was used to render the

HO CH 10H
F-(CFz ) W-(CH,)- - CH 2 0H-o-)
N-C-O-(CH2 3 -I CH
o z C:0(Ch2)z (CF 2 ) F

Fig. 30.7. Structure of a surface modifying macromolecule (from Ref. [10]).

992
 40 ¢(
A
35

30
A
S
C 25

20
o

105 A
.r
To 10 A
A,
A 0.24mm PES/SMM film
5 A A 0.12mm PES/SMM film

0 -

I T r I I I n . Ir i.i \I
0 1 2 3 4 5 6 7 8 9 10 2000
Evaporation time in oven at 1300 [min]

Fig. 30.8. Florine content versus evaporation time (from Ref. [10]).

membrane surface hydrophilic [11]. Recently, Uragami et al. and Miyata et al.
blended surface modifying macromolecules they synthesized into poly(1-tri-
methylsilyl-l-propyne) and polydimethylsiloxane membranes and tested the
performance of the membranes in pervaporation [12,13].

30.3 PROGRESS IN MEMBRANE CHARACTERIZATION

30.3.1 Membrane surface characterization by the measurement of

pore sizes

Pore sizes down to 1 nm were measured by a modified bubble point method. The
bubble point method to measure the pore size and the pore size distribution is
based on Cantor's equation:
Ap = 4( cos0/d
The pressure that allows a gas bubble to penetrate through a water-filled pore of
diameter, d, is given as Ap. Using the surface tension, , of water and zero
contact angle (0), the diameter of the pore measurable by the above method is
calculated to be 64 nm. A lower surface tension, or interfacial tension, is
necessary, when a pore size as small as 1 nm is measured. The interfacial tension
of two immiscible liquids A and B can be used for this purpose.
Suppose two liquids A and B are in contact with each other in the membrane
pore. Suppose also A does not wet the membrane surface, whereas B does. When
pressure is applied on liquid B that can wet the membrane surface, the contact

993
 16

14

12

10

8
0 1 2 3 4 5 6 7 8
Diameter (nm)

Fig. 30.9. Pore size distributions of some nanofiltration membranes obtained by modified bubble
point method (from Ref. [14]).

angle tends to be zero and Ap tends to be the maximum pressure. When a

pressure less than Ap is applied the interface is stable. When the pressureis
more than Ap, liquid B will penetrate through pores, the diameters of which are
larger than d. When pressure is applied on liquid A whose wettability is lower,
the contact angle tends to be 90 ° , and Ap tends to be zero. Liquid A penetrates
through all pores and the interface is unstable.
Courtois was successful in measuring the pore size of nanofiltration
membranes, as shown in Fig. 30.9 [14]. Bowen also claimed that pores as small as
1 nm were observable by AFM method [15]. It is expected that pores of reverse
osmosis membranes will be seen in the near future as progress will be made in
the technique of measuring small pore sizes.

30.3.2 Membrane surface characterization by atomic force

microscopy (AFM)

Scanning electron microscopy (SEM) has been a powerful tool for investigating
the morphology of membranes from the onset of reverse osmosis. Atomic force
microscopy (AFM) became popular in the 1990s and many AFM pictures of
reverse osmosis membrane surfaces have been taken. There are two examples in
which has AFM contributed to the development of high flux reverse osmosis
membranes. Hirose et al. found a relationship between the flux of reverse
osmosis membranes and their roughness parameters measured by AFM. Accord-
ing to their experiments, an increase in surface roughness resulted in a higher
water permeation flux. As an extension of this observation, they prepared an
aromatic polyamide composite membrane with an extremely rough surface. The
roughness of the surface was confirmed by transmission electron microscope.
The membrane so prepared, called ES 20 membrane, maintained a high salt

994
rejection while the flux was doubled from the conventional low pressure RO
membrane [16]. Fusaoka also found a similar rough structure on the surface of
their super ultra low pressure composite reverse osmosis membrane [1].
Bessieres et al. observed that the actual surface area of an ultrafiltration
membrane (molecular weight cut off = 200 kDa) is almost 40% more than the
geometric surface area (based on smooth surface) of the membrane [17].
The surface roughness of a microscopic scale also has an effect on the fouling
of reverse osmosis membranes by colloidal particles. Elimelech et al., by
comparing the fouling of cellulose acetate and aromatic polyamide composite
membranes, concluded that the effects of surface roughness on the interaction
force between particles and membrane surface resulted in enhanced attachment
of colloids onto the membrane surface, and hence more severe fouling [18]. The
important role of surface roughness in enhancing the attachment rate between
particles or between a particle and a surface has been pointed out by several
investigators. It has also been shown that an increase in the macroscopic surface
roughness in hollow fibers decreased the mass transfer coefficient and thus
increased concentration polarization [19].
The effect of surface roughness on the membrane transport and fouling will
be known more in detail as progresses will be made in the instrument to observe
the membrane surface. The development of hydrodynamics in a microscopic
scale is also called for.
The AFM method also played an important role in studying the morphology
of gas separation membranes. Integrally skinned asymmetric membranes were
prepared from poly(2,6-dimethyl-1,4-phenylene) oxide (PPO) using different
nonsolvents added to the casting solution. These nonsolvent additives consisted
of branched and linear alcohols ranging from C3 to C,,. Permeation data for pure
gases of CO2, CH 4, 02 and N2 were obtained for the membranes so prepared. The
membranes were further characterized by atomic force microscopy [20]. Figure
30.10 shows the permeance ratio of CO2/CH 4 versus mean diameter of nodules
observed under AFM. It is found that the membrane prepared from a solution
that contained 2-ethyl-l-hexanol as its additive (m in Fig. 30.10) had the
highest permeance ratio of 24.2 and the lowest mean diameter, 27.8 nm. There-
fore, the smallest nodule resulted in the highest permeance ratio. A possible
explanation for the tendency of decreasing permeance ratio with an increase in
nodule size is that the disentanglement of polymer molecules between nodules
occurs faster than the polymer vitrification. The smaller capillary force that
works between larger nodules also decreases the tendency for nodules to be
closed. An important observation in this study was the physical appearance of
nodules. There were two types of nodules identified as merged and discrete
(given in Fig. 30.10 as m and d). The discrete nodules are defined by the regularly
occurring and distinct depressions of the AFM image. The merged nodules, on
the other hand, show a flat surface. Another criteria for the discrete and merged
surface are the roughness parameter, Ra. If the roughness parameter is above
0.62 nm, the surface is defined as discrete while it is defined as merged if the

995
 28

26 -

Im_m
, 24-
X 4m

c 20-

c 18- 3di
c.) 6d
d 16

' 14- 0 t1 ~~~~7d

1)
9m -

10 1 i r . r C- I
20 40 60 80 100 120
Nodule diameter (m)

Fig. 30.10. Pure CO 2/CH, permeance ratio as a function of nodule diameter (from Ref. [20]): (1)
2-ethyl-l-hexanol; (2) 1-octanol; (3) 2-propanol; (4) 2-decanol; (5) 3,5,5-trimethyl-l-hexanol; (6)
2,4-dimethyl-3-pentanol; (7) 2,4,4-trimethyl-1-pentanol; (8) 2,2-dimethyl-3-pentanol; (9)
3-ethyl-2,2-dimethyl-3-pentanol; m, merged; d, discrete.

roughness is below 0.62 nm. Figure 30.10 shows that membranes with merged
nodules form the upper boundary of the data, while those with discrete nodules
form the lower boundary, with exceptions of 5m and 9m. Therefore, membranes
with smooth surfaces formed by merged nodules potentially exhibit high
selectivity. This information is important for the choice of nonsolvent additives
to be used in the preparation of casting solutions. A similar conclusion was
obtained also for the choice of solvent.

30.3.3 Membrane surface characterization by electron spin

resonance (ESR)

A new characterization method is now being developed at the University of

Ottawa using the electron spin resonance (ESR) technique. In one of their
papers, the selective skin layer of asymmetric reverse osmosis membranes was
studied by the ESR method; in particular, it was attempted to know if the skin
layer is indeed the same as the dense film prepared by complete evaporation of
solvent from a cast film of polymer solution [21]. By the ESR method, one is able

996
to know the degree in freedom of the motion of spin probes in a given
environment by analysing ESR signals obtained when the spin probes are
subjected to a magnetic field; i.e., the signal will consist of symmetric peaks
when the spin probes are freely mobile, while the asymmetricity of the peak
shape will increase as the motion of the spin probe becomes more restricted. In
this work, spin probes were incorporated into membranes in two ways. In one,
spin probes were brought into the skin layer of a reverse osmosis membrane
under a reverse osmosis condition, where an aqueous solution of the spin probe
is used as feed. In the other, spin probes were incorporated into a homogeneous
dense film by dissolving the spin probes into a polymer solution from which the
film was cast. Then, both the asymmetric membrane and the homogeneous film
were subjected to ESR analysis.
Figure 30.11 shows the ESR spectra of an aqueous solution of the radical
probe. The spectra consist of three symmetric peaks, since the NO radical of the
probe is freely mobile in the solution. The ratio of the peak heights b/a is unity. It
is expected that the b/a ratio will increase as the mobility of the radical probe
decreases. Figure 30.12 shows the ESR spectra of a cellulose acetate membrane
when radicals were brought into the membrane by immersing the membrane
into a radical solution. Figure 30.12 shows two overlapping spectra correspond-
ing to weakly restricted and highly restricted radical probes: one, which is
weakly restricted, is located inside the pores and the movement is restricted due
to the narrowness of the space in the pore; the other, whose motion is highly
restricted, is located in the polymer matrix and interacting with polymer
molecules. Even though the spectra are the overlap of signals from different
sources, the ratio of the peak heights, b/a, was considered as a measure of the
overall restriction on the motion of the radical probe. It was found that the b/a
-

20 G
H

Left: Fig. 30.11. Typical ESR spectra of TEMPO in an aqueous solution containing 0.1 wt.%
NaCI and 0.05 wt% TEMPO (from Ref. [21]).
Right: Fig. 30.12. ESR spectra of an asymmetric cellulose acetate membrane after immersing the
membrane in TEMO solution for 24 h and rinsing with water to remove the residual solution
(from Ref. [21]).

997
ratio increased from 1.3 to 1.9 as the temperature of heat treatment of the
cellulose acetate membrane was increased from 60 to 90 0C. The corresponding
change in the separation of sodium chloride was 0-95%. Interestingly, the b/a
ratio of 1.9 was also obtained for the dense homogeneous membrane in which the
radical probe was incorporated from the casting solution.

30.4 PROGRESS IN MEMBRANE TRANSPORT

Seawater is a complicated mixture of many components. Hence, it is difficult to

predict the membrane performance under different operating conditions based
on transport theories. Most of the theories treating the separation of multi-
component electrolytic systems are based on Debye-Hfickel theory, Donnan
effect and Nernst-Planck equation [22]. Although these theories are known to
be applicable to the mixture of any number of ions involved in the feed, there are
only few works in which agreement of experimental results and theoretical
predictions was tested for the mixtures of more than four ions. One of noted
exceptions is the work of Rangarajan et al. [23]. They have expanded the
Kimura-Souriajan theory to treat a multicomponent system involving nine ions
of primary concern in seawater. The separations of individual ions were
predicted theoretically and the results were examined by experiments.
Recent progress in membrane transport has been made mostly for
membranes carrying ionic charges, reflecting the development of a number of
charged nanofiltration membranes in the 1990s. In one such attempt, Chevalier
developed a model to predict membrane performance based on a pore-model
[22]. Since the measurement of pore sizes of nanofiltration membranes is now
possible by the modified bubble point method, a new development in the trans-
port theory is expected when the experimental technique to measure pore sizes
and her model is combined. She started from a general Nernst-Planck equation

Ji(r, z) = AkKaC(r,z)u -D 1 da (r;z) + F Cr; z) d (z)

[·rr·I
d y(r;z) dz RT 'dz)dz)

Note that the above equation includes friction factors, ai and i, which are
related to the ratio of the size of an ion and the size of the pore. After simplifying
the above equation by applying Boltzmann's distribution equation and the
Debye-Hiickel equation, a mono-dimensional equation that includes only the
axial co-ordinate in the pore was established. The general form is a complicated
matrix soluble for any number of ions by supplying numerical values specified
below:
- membrane; membrane porosity, pore size, pore length, charge density,
dissociation constant of the charge
- electrolytes; ionic valence, ionic diffusivity, ionic radius
- operating conditions; ionic concentrations in feed, pH of feed, operating
pressure

998
The model was tested by the experimentally obtained separation data for NaC1,
NaNO8, Na2SO4 , CdCl2 and CdSO 4 since the treatment of wastewater was her
primary concern. Calculation was also made for a mixture of NaCl and CdC 2.

30.5 HYBRID SYSTEMS FOR WATER TREATMENT

Immersed membrane, known as ZeeWeed®, has been developed by Zenon

Environmental Inc., Burlington [24]. The U-shaped hollow fiber bundles shown
in Fig. 30.13 are immersed into wastewater in a open container and a slight
vacuum (-2 to -8 psi) is applied on the bore side of hollow fibers to draw water
from the shell side. The company claims that the suction of water under vacuum
is advantageous over application of high pressure on the shell side since (1) the
energy requirement is reduced; (2) membrane fouling is reduced, and (3) the
membrane life is increased. Furthermore, high-speed air bubbles are introduced
from the bottom of the wastewater vessel along the hollow fiber bundles. This
aeration minimizes the settling of solid particles and also scrubs the hollow fiber
surface, resulting in the reduction of membrane pore blocking and fouling. The
membrane is chlorine resistant and can be cleaned by a high dose of chlorine.
When the wastewater vessel is used as a bioreactor, the aeration also supplies
oxygen that is required for the aerobic digestion of the sludge. This device
enables the immersed hollow fibers to work at very high solid contents of
1-1.5%. Another advantage of immersed membrane is its modularity: the system
capacity can be easily expanded by adding the modules.
The conventional method of wastewater treatment requires many steps of
pre-treatment before the water is finally treated by RO membranes. The pre-
treatment system requires a large area of land and capital investment. The sys-
tem is also complicated to operate. In order to reduce the number of steps
involved in the pre-treatment the immersed membranes can be used as micro-
filtration (MF) and ultrafiltration (UF) between the secondary clarifier and
reverse osmosis. Thus, MF and UF membranes filter the particles leaking from
the clarifier.

Fig. 30.13. Prototype of ZeeWeed Module (from Technical

Bulletin of Zenon Environmental Inc).

999
 In a membrane bioreactor, called ZenoGem®, the immersed membrane and
a bioreactor are integrated, thus combining clarification and filtration of an
activated sludge process in a single step. The main advantage of this system is
that no suspended solids will enter the effluent since particulate matters are
completely filtered by the immersed membrane. Since the settling of the sludge
can be minimized, a very high biomass concentration (1.0-1.5%) and high sludge
retention time (SRT 20-50 days) are allowed. This means that the membrane
bioreactor can be 2-5 times smaller than the conventional bioreactor. The
membrane bioreactor is simple in construction and operation.

30.6 PROGRESS IN MEMBRANE EXTRACTION

Reversed micelles are molecular assemblies that are formed when a compound
with an amphipathic structure is added to an organic solvent such as hydrocar-
bon. A micelle can contain an aqueous phase, the size of which is as small as a few
nanometers. Amino acids, peptides and proteins that are usually immiscible in
organic solvents can be extracted into the micelles in the organic phase. Reverse
micellar extraction uses this unique property of micelles [25].
Protein extraction by reversed micelles is schematically shown in Fig. 30.14.
In the first step, proteins in aqueous phase are extracted to the interior of
micelles, when the aqueous phase is brought into contact with an organic phase
in which the micelles are retained. This is called the forward extraction step.
After separating the aqueous and organic phases, the organic phase is brought
into contact with another aqueous phase into which proteins are extracted from
the organic phase. This is the backward extraction step. Forward and backward
extraction are enabled by the proper adjustment of pH or salt concentration of
the aqueous phases. Because of the simplicity of the process, reverse micellar
extraction is expected to become a bio-separation technology by which a large
amount of bio-products can be processed.
Forward Extraction Backward Extraction
Aqueous Phase i l

Reversed Micella
Reversed Micellar MiA
Mixing R 0 eancPhase Aqueous Phase IOrganie Phase
Organic Phase Extraction

Salt oeR

Aq hese
Pha A Aqueous Phase

Amphiphilic Molecule Recycle

Hdrophobic
Hydrophyic

Fig. 30.14. Protein extraction by reverse micelles (from Ref. [25]).

1000
 Ichikawa reported his results of reverse micellar extraction using di-2-
ethylhexyl sodium sulfosuccinate, called AOT, as a compound with an amphi-
pathic structure and isooctane as an organic solvent. Extractions of cytochrome
c, ribonuclease A and lysozyme were attempted.
Figure 30.15 shows how many percent of cytochrome c was extracted into the
organic phase from the aqueous phase in the forward extraction step. It depends
on the concentration of AOT in the organic phase and a certain concentration of
AOT was necessary to achieve 100% extraction of cytochrome c. This concentra-
tion was called the minimal AOT concentration.
Forward extraction experiments were carried out while maintaining the
AOT concentration at its minimal AOT value. As the concentration of cyto-
chrome c in the aqueous phase increased, cytochrome c concentration in the
organic phase increased and so did water concentration in the organic phase. In
Fig. 30.16 the concentration of water and cytochrome c, both in the organic
phase, are correlated. Interestingly, those concentrations were proportional to
each other and the slope was equal to 3500. This means that 3500 water
molecules were necessary to extract one protein molecule in the micelle. This
number was called hydrophobic surroundings (H.S.). H.S. was found to be
linearly correlated to the polarity ratio of the protein molecule.
Figure 30.17 shows the circular dichroism spectra of cytochrome c. The spec-
tra in the aqueous phase before the forward extraction step and after the Back-
ward Extraction step are the same. However, the spectra from cytochrome c in
the organic phase after the forward extraction has shifted remarkably from the
spectra from cytochrome c in aqueous phase. This means that the configuration
of cytochrome c molecule in the micelle is quite different from that in the aque-
ous phase. The reason is probably that cytochrome c molecule is not floating in
the water pool inside the micelle but it is interacting strongly with AOT mole-
cules that constitute the micelle wall. It was further confirmed that the activity
of cytochrome c molecule was almost completely retained after transferring the
cytochrome c molecule from one aqueous phase to another aqueous phase by the
forward and backward extraction steps.
The applications of this technique for the extraction of antibiotics and DNA
are being investigated [26,27]. In order to make reverse micellar extraction use-
ful in practice, a reverse micellar system with high bio-compatibility has to be
developed. Improvement in selectivity in extraction of a targeted molecule by
introducing bio-affinity in the micelle is also necessary.

30.7 CONCLUSIONS

The development of separation membranes, both in gas and liquid phases, is

continuing, achieving remarkable improvement in performance. Currently,
there is no foreseeable limit in either flux or selectivity.
Membrane surface modification has potential in the enhancement of
membrane performance. Surface modification by blending surface modifying

1001
 rI,. . .....
n
...
100
o
oh 80 ,K -
x : 60
60 E
a) Minimal AOT T
tm -40 Concentration

iL 20
... .. I
10 100 0 0.2 04
AOT [mM] cytochrome c [mM]

Left: Fig. 30.15. Degree of protein extraction versus AOT concentration (from Ref. [25]).
5
Cytochrome c concentration, 7.8 x 10 moll1; KCI concentration, 0.1 mol/l; pH 7-8.

Right: Fig. 30.16. Water concentration versus protein concentration at minimal AOT (from Ref.
[25]).

macromolecules may lead to the development of novel membranes with reduced

fouling. For surface modification to be applied more effectively, the relationship
between hydrophobicity/hydrophilicity and membrane fouling has to be firmly
established.
Membrane characterization by surface roughness has already led to the
development of ultra-low pressure membranes. It is expected that the effect of
roughness in a microscopic scale on the particle-surface interaction will be more
clearly known, which will enable the reduction of deposition of the particles on
the membrane surface.
Membrane surface characterization by novel techniques is progressing very
rapidly. This will have a tremendous impact not only on separation membranes
but also on various biomedical applications based on polymeric surfaces that
include membrane surface.
The technique to measure the sizes of nanoscale pores together with appro-
priate transport equations will enable more precise prediction of nanofiltration
membrane performance.
0
E
U
1'
E
Ma

E
x

Fig. 30.17. Circular dichroism spectra of cytochrome c in "

aqueous phase and organic phase containing reverse micelles s
(from Ref 1251).
)

1002
 Attempts have been made, some of them very successfully, to develop hybrid
systems where membrane separation is combined with other separation
processes. We will hear more of the success stories of hybrid system in the future.
Extraction with reversed miscelles has potential in biotechnology applica-
tions. Attention should also be paid to recent development of stereo-specific
membranes [28]. Even though it is still in its infancy, its potential in the
separation of optical isomers is very high.

REFERENCES

1 I. Kawada, Maku (Membrane), 24 (1999) 336.

2 Y. Fusaoka, Maku (Membrane), 24 (1999) 319.
3 D. Mukherjee, A. Kulkarni and W.N. Gill, J. Membr. Sci., 97 (1994) 231.
4 H. Suda and K. Haraya, in: I. Pinnau and B.D. Freeman (Eds.), Membrane Formation
and Modification, ACS Symposium Series 744. ACS, Washington D.C., 2000, Ch. 20,
p. 295.
5 H. Kawakami, Maku (Membrane), 24 (1999) 200.
6 H. Kita, H. Asamura, K. Tanaka and K. Okamoto, in: I. Pinnau and B.D. Freeman
(Eds.), Membrane Formation and Modification, ACS Symposium Series 744. ACS,
Washington D.C., 2000, Ch. 22, p. 330.
7 T. Matsuura, P. Santerre, R. Narbaitz, Y. Fan, V. Pham, H. Mahmud and F. Baig,
Membrane Composition and Method of Preparation, US Patent, 5,954,966 (1999).
8 A.V. Hamza, V.A. Pham, T. Matsuura and J.P. Santerre, J. Membr. Sci., 131(1997)
217.
9 D. Suk, Comprehensive Exam Report, Department of Chemical Engineering, Univ-
ersity of Ottawa, 2000.
10 D. Suk, G. Chowdhury, T. Matsuura, R.M. Narbaitz, P. Santerre, G. Pleizier and Y.
Deslandes, Study on the kinetics of surface migration of surface modifying macro-
molecules in surface modification. Macromolecules, 35 (2002) 3017.
11 Membrane Separation Technology News, June 2000, p.8.
12 T. Uragami, T. Doi and T. Miyata, in: I. Pinnau and B.D. Freeman (Eds.), Membrane
Formation and Modification, ACS Symposium Series 744. ACS, Washington D.C.,
2000, Ch. 18, p. 263.
13 T. Miyata, Y. Nakanishi and T. Uragami, in: I. Pinnau and B.D. Freeman (Eds.),
Membrane Formation andModification, ACS Symposium Series 744. ACS, Washing-
ton D.C., 2000, Ch. 19, p. 280.
14 T. Courtois, in: Proceedings of the First Nanofiltration and Application Workshop,
Trois Rivibres, June 1-4, 1997.
15 W.R. Bowen, N. Hilal, R.W. Lovitt and C.J Wright, in: Proceedings of ICOM'99,
Toronto, June 12-18, 1999.
16 M. Hirose, H. Itoh, Y. Minamizaki and Y. Kamiyama, in: Proceedings of ICOM'96,
Yokohama, Aug. 18-23, 1996, pp. 178-179.
17 A. Bessibres, M. Meireles, R. Coratger, J. Beauvillain and V. Sanchez, J. Membr. Sci.,
109 (1996) 271.
18 M. Elimelech, X. Zhu, A.E. Childress and S. Hong, J. Membr. Sci., 127 (1997) 101.
19 S. Deng, Roughness of Membrane Surface and Its Effect on Membrane Performance,
Seminar on the Ecological Applications of Innovative Membrane Technology in the
Chemical Industry, held by UN Economic Commission for Europe, Cetrato, Italy,
May 1-4, 1996.
20 J.M.A. Tan and T. Matsuura, J. Membr. Sci., 160 (1999) 7.

1003
21 K.C. Khulbe, T. Matsuura, G. Lamarche, A.-M. Lamarche, C. Choi and S.H. Noh,
Polymer, 42 (2001), 6479-6484.
22 S. Chevalier, Mod6lisation math6matique des m6canismes de separation en nano-
filtration, Ph.D thesis, University of Bordeaux, 1999.
23 R. Rangarajan, M.A. Majid, T. Matsuura and S. Sourirajan, Ind. Eng. Chem. Proc.
Des. Dev., 24 (1985) 977.
24 D. Mourato, Desalinationand Water Reuse, 9/4 (2000) 27.
25 S. Ichikawa, Maku (Membrane), 22 (1997) 331.
26 N.W. Fadnavis, B. Satyavathi, and A.A. Deshpande, Biotechnol. Progr., 13 (1997)
503-505.
27 M. Goto, T. Ono, A. Horiuchi and S. Furusaki, J. Chem. Eng. Japan, 32 (1998)
123-125.
28 M. Yoshikawa, in: Molecularly and Ionic Recognition with ImprintedPolymers, ACS
Symposium Series, 703. ACS, Washington D.C., 1998, Chap. 12, pp. 170-187.

1004
Chapter 31

Novel fibrous systems for contaminant

removal
James Economy and Christian Mangun

31.1 INTRODUCTION

During the past several decades, there has been extensive work done to monitor
and quantify the harmful effect of various pollutants to the environment. At the
same time, considerable effort has been dedicated to develop alternative
chemicals and processes which might not be as damaging to the environment.
However, very little emphasis has been placed on designing improved materials
that could selectively remove a wide range of impurities and permit for recovery
and reuse. In particular, new fibrous systems based on coated glass fibers appear
to offer many advantages over the currently available technology of granular
activated carbon (GAC) and ion exchange beads (IEB). These advantages include
the design flexibility to be fabricated in the form of felts or fabrics, which display
greatly improved contact efficiency leading to faster rates of reaction/ regenera-
tion. In addition, the wear resistance and dimensional stability are greatly
improved through the reinforcement of a glass fiber substrate.
In this chapter, the technological background of both activated carbons and
ion exchange materials is described. This is followed by a more detailed consider-
ation of recent work on the activated carbon fibers (ACF) and ion exchange
fibers (IEF).

31.2 REVIEW OF ENVIRONMENTAL MATERIALS

31.2.1 Activated carbons

Activated carbons have been used for many years for both air and water
purification and are described extensively in the literature [1-6]. The primary
raw material for activated carbons can be any organic material with an accept-
able char yield (i.e., coal, wood, coconut shells, or certain polymers). Activation
occurs when the carbon layers in the char are etched away through an oxidation
reaction resulting in the formation of a porous carbon network with high surface

ComprehensiveAnalytical Chemistry )YXXVII

J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 1005
area. Amorphous, activated carbons have a structure consisting of layers of
twisted, ribbon-like turbostratic planes. The narrow space between these planes
can extend down to molecular dimensions to produce materials that behave like
molecular sieves. Physical adsorption in these pores is the primary means by
which activated carbons work to remove contaminants. Micropores (less than 20
A), rather than the larger mesopores (between 20 and 500 A) and macropores
(greater than 500 A), are preferentially filled at low relative pressures due to the
higher overlap in potential from the van der Waal forces of opposite pore walls
[6].
Commercially available GAC has been the industry standard for many years.
They are used as adsorbents to purify polluted waste streams because of their
low cost and because they have been determined to be the best available
technology for removal of many contaminants. However, GAC suffers from a
number of drawbacks including poor selectivity, slow kinetics, the need for
expensive containment systems, less than 100% working capacity, and difficult
reactivation. The production of activated carbon fibers (ACF) has been limited
as a result of the extreme weight losses realized during activation leading to
increased material costs. The technique of carbonizing the synthetic precursors
to such fibers also produces materials with poor mechanical properties that are
brittle or friable, limiting their utility to systems in which some type of struc-
tural support or containment is required (again, increasing the cost). Perhaps
the most serious problem is the absence of a comprehensive understanding as to
the effect of pore size and chemistry on the adsorption properties. Such
knowledge would allow one to tailor activated carbons for removal of specific
contaminants.

31.2.2 Ion exchange beads

Conventional IEB consist of three-dimensional covalent networks to which ion

exchange groups are bound. The network preserves structural integrity of the
material while the bound ions provide either cationic or anionic exchange sites.
For example, the typical cationic styrene-based matrix is prepared by suspension
polymerization of styrene and varying ratios of divinylbenzene followed by
swelling with organic solvents and sulfonation with sulfuric acid [7]. IEB are
used extensively for purification and demineralization of water. While ion
exchange beads can be very effective, it should be noted that they have a number
of drawbacks. During the activation stage, solvents must be used to pre-swell the
cross-linked beads to facilitate sulfonation. These toxic solvents also impart
strong odors or taste that remain in the finished product. The beads are very
susceptible to fracture and breakage due to osmotic shock and must be kept wet
at all times (Fig. 31.1a). In addition, the beads frequently require costly service
containment systems and the added expense of EPA required disposal once
spent. More importantly, there are a number of major needs with respect to
environmental pollution where greatly improved ion exchange materials are

1006
Fig. 31.1. SEM micrographs of (a) Purolite cationic exchange bead displaying effects of osmotic
shock and (b) ion exchange fibers.

required. These needs include materials with better contact efficiency, a

capability for rapid regeneration, much longer service life, and the ability to
selectivity remove specific ions (such as heavy metals to acceptable levels of less
than 1 part per billion).

31.3 RECENT DEVELOPMENTS

31.31 Activated carbon fibers

Undoubtedly the first major innovation in this field occurred with the discovery
that ACF could be made by coating glass fibers (either woven or non-woven) with
a phenolic resin. The coating is then carbonized and activated from 600-800°C in
a single step process to yield the adsorbent material (Fig. 31.2). Given the low
costs of the constituents and the relatively simple processing, the potential exists
for producing a new adsorbent material that is cost competitive to GAC and with

Fig. 31.2 (a) Broken end of ACF showing the glass core surrounded by carbon coating. (b) Broken
end of an etched ACF showing a hollow carbon layer.

1007
greatly improved wear resistance over current ACF. In addition, the contact
efficiency is significantly improved compared to current ACF and almost an
order of magnitude better than GAC. Faster adsorption/desorption rates (due to
a decrease in the diffusion path length) maximizes the carbon usage resulting in
reduced critical bed depths and pressure drops. Better wear resistance means
lower attrition losses during handling and the potential for greatly increased
utility as protective garments. This ACF can be manufactured in a wide variety
of product forms (textiles, felts, papers), which allows for vast design flexibility
thus meeting the demands of a variety of applications.
Typical activated carbons have limited selectivity since their adsorption char-
acteristics are controlled to a considerable extent by the acidic nature of their sur-
face. In addition, the pore size distributions tend to be very broad with little
consideration as to what would be the most advantageous size range for a particu-
lar contaminant. Thus recent research describing the relationship between pore
size/surface chemistry and adsorption capacity becomes important. This led to
specific techniques for tailoring these variables through careful control of the acti-
vation conditions. The time/temperature of activation dictates the pore size while
the atmosphere can greatly alter the pore surface chemistry. Thus a wide range of
tailored materials with micropores ranging from 5-30 A and chemical structures
including nitrided, sulfonated, unsaturated, or a more highly oxidized surface
could be made to enhance adsorption of specific contaminants.

31.3.1.1 Chemical activation

The latest discovery was the development of chemical activation techniques to
lower the required heat treatment temperature. In this case, the adsorbent
fibers were produced by coating a glass, mineral or organic fiber substrate with a
polymer solution or melt containing a dehydration catalyst. Coatings that have
been examined include phenolic resin, polyvinyl alcohol, polyacrylonitrile or
cellulose in combination with reagents such as ZnCl2 , phosphoric acid, or
potassium hydroxide [8]. This is followed by drying the coated fiber and then
heating to a temperature of 170-350°C in nitrogen or air to introduce micro-
porosity. The final step consists of washing the resultant fiber with base or acid
to remove any remaining reagent followed by rinsing with DI water. Further
heating of the fiber to above 350 up to 800°C may be advantageous in certain
cases where a carbonaceous structure is desired. Through control of the process-
ing steps, the surface area, weight percent carbon, yield, surface chemistry, etc.
can all be modified to match a specific requirement. Figure 31.3 gives an example
of the effect of temperature on the surface area and the coating content on glass
for phenolic resin with ZnCl2 activation. It is noted from Fig. 31.3 that washing
with HC1 was very important since lower surface areas result from the presence
of zinc salts, zinc hydroxide and even zinc oxide generated during activation. In
contrast with thermally-treated ACF from phenolic fibers where a temperature
above 600°C is needed for activation [9], ZnCl 2 -treated phenolics begin to develop
porosity at a much lower treatment temperature. Surface area development

1008
 - ,-- -,Samples were washed with 0.5N of HCIand DI. water.
-,
- -,Samples were only washed with D.I. water.
1000 60

...
--- 50

< 600 - -

o/, Based only on the coating -30

at 400 C.
o , o
Z~ - -20 .;
0-,, . 0
& 200 Based on total materials
t, 10

100 200 300 400 500 600 700 800 900

Activation Temperature (C)

Fig. 31.3. Effect of activation temperature on the surface area of phenolic resin-based ACF
produced by ZnC2, activation.

starts at about 300°C and increases rapidly up to 350, after which it gradually
rises to a broad maximum of -852 m2 /g of coating. At temperatures above 550°C
the surface area begins to decline due to significant widening of the pores.
The chemically activated ACF exhibit notable advantages over current acti-
vated carbon materials. For example, a wider range of fiber substrate materials
may be used offering greater versatility. This includes low melting point glass
microfibers, as found in current HEPA filters, allowing for potential adsorption/
filtration in a single step (of considerable importance in reducing the size of gas
masks, for example). Use of polymeric fiber substrates would result in even
better wear resistance over the glass substrate, further extending their utility
for protective clothing. This new approach makes it far easier and less expensive
to manufacture due to the low temperature activation and higher yields. By
utilizing different starting polymers, a wide variety of chemically modified
surfaces can be created which are capable of adsorbing/chelating many different
contaminants from both air and water.

31.3.1.2 Effect of pore structure on adsorptionbehavior

The characterization of the nature of the pore size and shape of ACF was initially
examined with a variety of molecular probes ranging from large dye molecules
down to small organics such as butane, benzene, iso-butane and cyclohexane.
Considerable insights were obtained from these indirect studies; however, a
more direct approach was needed to quantify the porous structure. The prevail-
ing view up to several years ago was that the pores in activated carbons were slit
shaped. Thus the successful imaging of pores in the ACF using scanning tunnel-
ing microscopy (STM) techniques [9,10] permitted an irrefutable depiction of
the pore size/shape and the pore size distributions. It was discovered that the

1009
 Fig. 31.4. STM images of ACF (a) surface and (b) cross-section.

fiber surface was composed of dispersed mesopores while the bulk of fiber
contained a homogeneous arrangement of micropores that were ellipsoidal in
shape (Fig. 31.4). This research also elucidated the mechanism of pore formation
showing that the first step of heating and charring the phenolic fiber creates a
microporosity of -600 m2 /g just from the evolution of volatiles. In the second
step, etchants such as steam, CO2 , air or ammonia act to produce a mesoporous
region to a depth of about 10 nm. The microporous structure in the core remains
relatively uniform in diameter but increases in size with increasing time of
activation. This type of process is in contrast to GAC where the activation must
initiate from the outside surface thus resulting in a very broad pore size
distribution.
In addition, experiments were carried out on a series of ACF with a range of
surface areas to examine the effect of pore size on adsorption of gases [11]. The
adsorption isotherms of n-butane for this series of ACF showed a counter-
intuitive relationship such that adsorption capacities at low concentrations
increased as surface areas decreased. This was explained by assuming that the
lower surface area material had a smaller pore size; this hypothesis was later
confirmed through STM. The smaller pores led to an increase in the overlapping
potentials from neighboring pore walls thus binding the molecules more tightly
at low concentrations. This resulted in the condensation of butane at much
lower concentrations, allowing for immediate use of its total pore volume. How-
ever, as the concentration was raised, the butane would eventually condense in
the larger pores. Since the higher surface area ACF also had a larger total pore
volume, they would eventually win out over the smaller pore materials resulting
in a crossover regime. Recently, this work was extended to examine the effect of
boiling point on the crossover phenomenon [12]. When the adsorption tempera-
ture is much greater than the boiling point of the adsorbate, the thermal energy
possessed by these molecules is sufficient to overcome the forces leading to lique-
faction. Therefore the molecule would prefer to remain in the gas phase. This
means that for gases with low boiling points such as methane, only the high-

1010
energy sites will be occupied because pore size is the critical variable. Since
methane does not condense in the pores, no crossover was observed (the small
pore ACF adsorbed better over the entire concentration range). It was shown
that as the boiling point of the adsorbate was increased, the crossover regime
shifted to higher concentrations. For low boiling point gases or at lower concen-
trations, the smaller pore ACF adsorbs to the highest capacity. But for higher
boiling point gases and at higher concentrations, pore volume becomes the con-
trolling feature. Therefore, this becomes a competition between the pore size
and the pore volume of the ACFs to determine which will be the most important
factor.
This type of information is critical in tailoring an activated carbon for
increased adsorption capacity. Thus, in most cases, it is the low surface area
(small micropore) material that is the most advantageous since one is typically
concerned about removing very low concentrations of contaminants. This is
counter-intuitive to the notion that higher surface area materials will always
perform better. As discussed previously, GAC is at an extreme disadvantage due
to its large pore size distribution whereas in ACF the distribution is much
narrower. In GAC, the same type of micropores only exist in the very deep
portions of the structure while the entrances on the surface are extremely wide
by comparison. Therefore, the pore structure of GAC has a weak attraction at
the surface and it takes a much longer time for molecules to diffuse and
penetrate to the point where they can be captured and held. Thus, the process of
filling the pores in GAC is more time consuming and much less efficient than
that of ACF.

31.3.1.3 Effect of pore surface chemistry on adsorptionbehavior

The carbon edges at the surfaces of the micropores can be combined with a
number of different atoms depending on the activation conditions and post-
treatment. Conversely, these heteroatoms can be removed upon heat treatment.
Currently GAC is prepared by a high temperature oxidative etching process that
results in a pore surface containing a few percent oxygen in the form of phenolic
hydroxyls, carboxylic acids, and quinones. Hence the pore surface is slightly
acidic which of course strongly influences the adsorption characteristics toward
acidic or basic materials. Therefore techniques have been developed to alter the
pore surface chemistry to make it basic, more acidic, neutral or polar [13].
By this relatively simple approach, one can dramatically increase adsorption
efficiency and capacity. For example, an ACF further oxidized from 3-5% oxygen
to 25-30% oxygen content will have 30x the adsorption capacity for ammonia
gas as demonstrated by breakthrough experiments (Fig. 31.5) [14]. Note that the
untreated ACF does not adsorb an appreciable amount of ammonia and thus the
breakthrough time is fairly short. Following the oxidation treatment, the break-
through time increases tremendously from 25 up to 800 min per gram. This
improvement is due to the enhanced interaction of the ammonia molecule with
the acidic functional groups that have been introduced onto the pore surface.

1011
 1

0.8

ACF-Untreated
--- ACF-Oxidized !
0.6 i ........... .... _
0
U
U
0.4

I..........
0.2 J............. .. . ..

., , ,,}1 i X
0
0 200 400 600 800 1000 1200 1400 1600
Time (min/g)

Fig. 31.5. Normalized breakthrough curves for ammonia (Co = 970 ppmv) on ACF.

-'---- Basic (10% N)

- Basic (5% N)
15 --_- Basic (2% N) ............... .
Acidic (5% i)

-n 10
-e
0

0 ' ' ' ' ' ' ' ' ' ' ' ' ' '
0 1000 2000 3000 40 00 5000 6000
Concentration (ppmv)
Fig. 31.6. Adsorption isotherms of HCI gas on ammonia treated ACF.

Conversely, an ACF fiber activated in an atmosphere of ammonia gas will

contain basic nitrogen groups (typically in the form of pyridine/aniline type
units), which are effective for adsorption of acidic contaminants (Fig. 31.6) [15].
The amount of HCl gas adsorbed scales with the quantity of basic nitrogen (as
determined by elemental analysis). For example, a basic ACF containing 10%
elemental nitrogen will adsorb up to 7 x by weight of hydrogen chloride gas as
compared to a slightly acidic ACF containing 3-5% oxygen. Other dramatic

1012
increases in adsorption capacity over currently used systems have been demon-
strated for SO2 , acetone, and trichloroethlyene [16-18]. Through these exam-
ples, one can note the effect that altering the pore surface chemistry has on
adsorption of various contaminants. This type of information, combined with
the knowledge generated on the effect of pore size, is invaluable in designing
highly efficient systems for removal of trace contaminants.

31.3.1.4 Application: waterpurification

As part of a government sponsored project to develop a man-portable water puri-
fication system for military use, ACF were designed as a compact filter with
low-pressure drops coupled with high adsorption capacity toward chemical con-
taminants [19]. Experiments measuring adsorption isotherms of BTEX (ben-
zene, toluene, ethyl benzene, and xylene which are the residual components of
spilled fuel) and two chemical warfare simulants on tailored ACF showed much
higher equilibrium capacities as compared to an industry standard GAC (Calgon
Filtrasorb 200). The next phase involved design and testing of equal bed weights
of ACF filters versus GAC beds in breakthrough mode. The results demon-
strated that the good contact efficiency of the fibers allowed for realization of
these higher capacities thus yielding longer breakthrough times. In addition,
comparison of the breakthrough curves showed two very important differences
between ACF and GAC. In all cases, an immediate breakthrough of the contami-
nant was noted with GAC, due to short circuiting through the relatively large
interstitial sites of the bed. Using benzene as an example (Fig. 31.7), the first
effluent concentration recorded was 130 ppb for GAC in contrast to below 1 ppb
for the ACF. Since the EPA maximum contaminant level for benzene is 5 ppb,
the danger level was immediately exceeded when using GAC. Secondly, after this

E
0

C
c

0 5 10 15 20 25 30 35 40
Volume (L)

Fig. 31.7. Breakthrough curves for benzene.

1013
initial breakthrough, the GAC effluent concentration steadily climbed while the
ACF stayed level (close to zero) until a sharp breakthrough curve occurred. This
was indicative of fast adsorption kinetics and a very short mass transfer zone
giving the ACF an added advantage under dynamic conditions. This was in con-
trast to the GAC where the much longer diffusion path lengths and buried
micropores resulted in premature breakthrough. These same benefits were also
obtained with the chemical warfare agents where 3-fold adsorption superiority
was measured. In addition, the ACF could be easily regenerated multiple times
at low temperatures to yield 100% working capacity. Regeneration of GAC would
require much higher temperatures thus altering the pore structure and reducing
the adsorption capacity.

31.3.2 Ion exchange fibers to remove inorganic contaminants

The concept of developing ion exchange materials in the form of fibers was first
reported by Economy et al. 30 years ago [20]. Fibrous ion exchange materials
have several advantages over the conventional ion exchange units. These advan-
tages include the ability to be fabricated in the form of felts or fabrics where the
contact efficiency is greatly improved. A fibrous form eliminates the need for
currently used retainers such as canisters, screens, etc. Recently, two new
concepts have been developed to yield fiber systems with a range of capabilities.
The first involves the concept of designing ion exchange materials by coating
glass fiber substrates with a polymeric ion exchange resin [21]. In the second
system, modifications of the aforementioned ACF where the pores have been
chemically treated to enhance their use for ion exchange as well as their ability
to chelate and remove specific contaminants from water [22].

31.3.2.1 Coated ion exchange fibers

In this approach, cationic exchange fibers were prepared by (1) coating glass
fiber substrates with a polystyrene/divinylbenzene oligomer, (2) curing, and (3)
sulfonating (see Fig. 31.8). Through improved polymerization techniques, the
use of solvents prior to functionalization and end-use was eliminated thus
greatly simplifying the overall preparation (Fig. 31.9). Use of relatively thin
coatings on glass fiber substrates appeared to insure longer life by eliminating
long-term fatigue and cracking caused by residual strains and swelling of the
beads (Fig. 31.1). A range of cation exchange capacities (CEC) was achieved as a
function of the sulfonation time and temperature with values as high as 5 meq/g
based on resin loading (see Fig. 31.10). This is comparable to the best CEC values
for the commercial beads. However, kinetic experiments showed the contact
efficiencies of the new systems were greatly improved over the traditional beads
due to greater surface/volume ratio associated with the film morphology and
shorter diffusion lengths. This translated into an order of magnitude increase in
ion exchange rate as illustrated by Fig. 31.11. The rate of exchange was also
increased when the ion concentration was higher. The film of solution immedi-

1014
 HC:CH 2

S Coatig
*CiR5
4h Fiberglass Substrate
with Resin Coating

t 2SO
V
Diivinylbenzene Oligomerization Reaction H H H
-c-c-c-c-C-C-

HS3'
-C-C--C-C--C-
H2 H2 | H2

SO3 H SO3-H

Sulfonation Reaction
Styrenic Sulfonic Acid Resin on
Glass
Fig. 31.8. Synthesis of polymeric ion exchange fibers.

Ion Exchange
Beads

Suspension
Poymerization
(organic
solvents
usedformacroporoity)
Ion Exchange Fibers
Ion Exchange
Fibers Distallation
forSolvent
Removal
toAllow
forPorosity
Oligomer Coating &Curing Stage
ofBeads
Preswelling Using Solvents
Organic
Functionalization Stage toAllow
Core
Functionallization

Stepped Dilution Rinse Functinalizatin

I Stepped
Dilution
Rinse
I~~~~~~~~~~~~~~~~~~
Beads
Shipped
&Stored
with~50%
Mdsture
Content

IPreswelling
ofBeads
Overnight
Prior
toEnd-use

Fig. 31.9. Comparison of fiber and bead preparation demonstrating that far fewer steps are
required with IEF.

ately in contact with the resin was rapidly depleted of ions by proton exchange
from the resin. Further exchange was delayed until more ions from the bulk
solution could diffuse into the surface film. Diffusion was easier with increasing
ionic concentration in bulk solution, making the exchange faster. This supports
the view that at low concentrations, the rate was controlled by "film" diffusion

1015
 Low Temperature 25C)
5 HighTemperature (95 C}

4-

0
4 8 12 16 19,5 2 4 5.5

Sulfonation Time (hrs)

Fig. 31.10. Capacities resulting from varied sulfonation treatments demonstrating the ease of
functionalization of the IEF.

4
10 T ,

- Fiber- 0.01 M NaCI

| Beads- 0.01 M NaCI

--e--Fiber- 0.005 M NaCI

Ia Bea ds- 0.005 M NaCl I

1000 I

e~~~~~~~~~~~~~
, .
P~~~~~~~~~~
100
E--!

F i i

I I I . . - --
1u
0 0.5 1 1.5 2 2.5 3
Total Volume 0.1 M NaOH Added, (ml)

Fig. 31.11. Batch rates of exchange for two saline concentrations illustrating an order of
magnitude faster reaction rates for the IEF.

1016
 90 -

80-
A Cationic Beads (5% DVB)
O Cationic Fiber (8% DVB) O
70-

60
E
50- A A A A A A A A A
0

(- 40-

- 30
Conditions:
n, 20- 100 ml/min flow rate
o pH = 4.65 at 27.9 C
Lowest residual (-1.2g resin each)
10- level <1 ppb
0
O- r 0 omm 0
. I I I I I I I I I I
0 500 1000 1500 2000 2500 3000 3500

Cumulative Effluent Volume/g, (ml/g)

2+
Fig. 31.12. Dynamic mode breakthrough of Pb demonstrating the IEF's superior mass transfer
kinetics (influent conc = 150 ppm).

and at higher concentrations; the rate was controlled by "particle" diffusion. In

addition, the usefulness of the ion exchange process resides with the ability to
quickly regenerate with no loss in capacity. The IEF were successfully
regenerated for ten cycles with no observable changes in capacity or stability.

31.3.2.2 Application: heavy metal removal

Although water softening is of important commercial use for ion exchange mate-
rials, a much more serious problem is the removal of toxic inorganics such as
heavy metals that remain in wastewaters from the plating industry, tap water
that contains excessive lead and copper owing to corrosion, well water that con-
tain species such as arsenic, etc. Recent testing on both cations and anions (uti-
lizing an ionic coating based on vinyl benzylchloride co-divinylbenzene) has
proven the superiority of the fibrous substrate over the beads. Figure 31.12 dis-
plays the dynamic mode breakthrough of lead and Fig. 31.13 of arsenate. Both
results display similar trends as observed with the ACF where breakthrough for
the beads was immediate while the IEF had an effluent concentration below
detectable limits until a distinct breakthrough occurred.

31.3.2.3 Carbon-BasedIEFs
Carbon-based IEF provide a different mechanism of accessing the ionic species
via a microporous network in comparison to the swollen networks of the beads or
fiber coatings. This allows the capability to utilize both pore surface chemistry
(oxidation and/or sulfonation) and molecular sieving properties to selectively

1017
 35-
Conditions:
100ml/min flow rate A
30-
pH 3.1 at 28.5C
(-2.0 g resin each) A
25
A
A v
A *
A Anionic Bead Filter
A O Anionic Fiber Filter
A
< 15-

a 10-
a' 00 0

5-
o lowest residual
0 00
O0 level <0.1ppm
0- 0 o oo o o o o o O
, i * . i . . I . I . i
20 40 60 80 100 120 140 160
Cumulative Effluent Volume/g resin, (ml/g)

Fig. 31.13. Dynamic mode breakthrough of Arsenate (V) on anionic fiber versus beads (influent
cone = 104.7 ppm).

remove ions. Utilization of ACF greatly increases the surface area available for
incorporation of ion exchange groups. In addition, these materials exhibit
excellent chemical and thermal resistance. This is required for many industrial
applications including the removal of cations from corrosive media, hot solutions
or melts.
Oxidation of carbons to prepare weak acid cationic (WAC) exchangers has
been extensively studied [16,23,24]. The ability to functionalize carbons with
strong acid cationic (SAC) groups would lead to systems with ion exchange
capability over the entire pH range. The preparation of high surface area
carbons typically involves treating precursors to high temperature (700-1100°C)
in activating environments (e.g. CO2steam) [25]. It has been reported that
degradation of phenolic resins at temperatures above 700°C was associated with
the loss of aromatic hydrogen [26]. The presence of aromatic hydrogen is
essential for electrophilic substitution reactions such as sulfonation. Due to the
poor reactivity of carbons activated at temperatures above 700°C, reports of
sulfonation in the literature are scarce. Instances where it has been
accomplished include coals and carbon blacks, albeit with low ion exchange
capacities [27,28]. Thus recent synthetic routes utilizing chemical activation at
much lower temperatures (allowing for retention of the aromatic hydrogen)
were examined to prepare SAC exchange fibers derived from phenolic resins.
Principal variables investigated to maximize the degree of sulfonation included
the carbonization and sulfonation temperatures.
After sulfonation, the increase in oxygen was significantly higher than
expected as calculated from addition of only sulfonic acid groups. Thus, it was
concluded that oxidation of the microporous carbon accompanied the high
temperature sulfonation of these systems; this result is consistent with reports

1018
 .50

WX'
'0'

3
cat to
1A

Fig. 31.14. Strong acid ion exchange capacities of ACF series as a function of carbonization and
sulfonation temperatures.

in the literature [29]. It is recognized that three groups (carboxylic acids,

quinones, and phenolic hydroxyls) typically dominate an oxidized pore surface
[30]. Figures 31.14 and 31.15 display the strong and weak ion exchange
capacities of the carbon-based IEF as a function of carbonization and sulfonation
temperatures. The data show that at higher carbonization temperatures the
reactivity of the fiber for sulfonation decreases, likely due to deactivation of the
ring structure. On the other hand, capacity increased significantly as
sulfonation temperature increased. Also, it was found that the presence of
carboxylic acid and phenolic hydroxyl functionalities further enhanced the
exchange capacity of these materials at higher pH. Total ion exchange capacities
were measured as high as 20 meq/g, much greater than any reported values.
Under the same experimental conditions, the exchange capacity of commercially
available beads (Purolite C100X10H) was measured to be 4.7 meq/g. This
particular type of bead was selected for comparative purposes in this study
because it was designed for high flow rate systems and enhanced thermal
stability. The ion exchange capacity of the fibrous systems is competitive with
cationic exchange beads at low pH and substantially better with higher pH
influents. In all cases, the increased oxygen contents were consistent with the
amount of acidic groups necessary to explain the increased WAC. This character-
istic can be very useful since hard water is naturally buffered at pH > 7. Material
durability is equally as important as capacity in ion exchange processes. Use of
carbonaceous ion exchangers eliminates several problems encountered with ion
exchange beads. For example, these fibers will not swell and therefore will not
undergo osmotic shock. Secondly, the fiber is more thermally stable and does not
lose significant weight until greater than 250°C, which is the degradation

1019
 8 a

i'

Fig. 31.15. Weak acid ion exchange capacities of ACF series as a function of carbonization and
sulfonation temperatures.

temperature of sulfonic acid groups. This is in contrast to the relatively low

thermal stability of commercially available cationic exchange beads; even beads
specifically designed with high thermal stability have a maximum use tempera-
ture of only 130°C. Atmospheric oxygen, particularly at higher temperatures,
causes resin degradation and a significant reduction in performance [31].
Finally, contact efficiency is greatly enhanced with fiber vs. bead configurations
providing narrow mass transfer zones and extended time to breakthrough.

31.4 SUMMARY

In this chapter it has been shown that preparing activated carbons or ion
exchange resins in the form of fibers has greatly improved their contact
efficiency and ease of regeneration. Because of these improvements, it is now
possible to remove a wide range of trace contaminants (both organic and
inorganic) to below one ppb. This represents a several fold improvement over
commercially available GAC and IEB. In the case of ACF, additional improve-
ments in efficiency are possible through further tailoring the pore dimensions
and surface chemistry. It is noteworthy that use of a glass fiber substrate for
preparing the ACF and IEF has greatly simplified the synthesis as well as
improving the wear resistance.
New directions for preparing activated organic fibers at relatively modest
temperatures of 200-400°C are indicated along with the preparation of an ACF
containing sulfonic acid groups. It is anticipated that both of these systems will
open up new and important pathways for removal of trace contaminants from
the environment.

1020
REFERENCES

1 P.L. Walker, Carbon,28 (1990) 261.

2 P.N. Cheremisinoff and F. Ellerbusch, Carbon Adsorption Handbook. Ann Arbor
Science Publishers, Ann Arbor, Michigan, 1980.
3 R.C. Bansal, J.B. Donnet and F. Stoeckli, Active Carbon.Marcel Dekker, New York,
1988.
4 V.L. Snoeyink. In: F.W. Pontius (Ed.), Water Quality and Treatment. McGraw-Hill,
New York, 1990, pp. 781-875.
5 B. McEnaney. Carbon, 26 (1988) 267.
6 S.J. Gregg and K.S.W. Sing, Adsorption, Surface Area, and Porosity, 2nd edn.
Academic Press, London, 1982.
7 K. Dorfner. Ion Exchangers, Ann Arbor Science Publishers, Ann Arbor, Michigan,
1972.
8 Z. Yue, C.L. Mangun and J. Economy, Carbon, 40 (2002) 1181.
9 M. Daley, D. Tandon, J. Economy and E.J. Hippo, Carbon, 34 (1996) 1191.
10 J. Economy, M. Daley, E.J. Hippo and D. Tandon, Carbon, 33 (1995) 344.
11 K.L. Foster, R.G. Fuerman, J. Economy, S.M. Larson and M.J. Rood, Chem.
Materials, 4 (1992) 1068.
12 C. Mangun, M. Daley, R. Braatz and J. Economy, Carbon, 36 (1998) 123.
13 C.L. Mangun, M.A. Daley and J. Economy, Proc. of the Annual AWMA Conference,
San Antonio, TX, 1995.
14 C.L. Mangun, R.D. Braatz, J. Economy and A.J. Hall, Ind. Eng. Chem. Res., 38 (1999)
3499.
15 C.L. Mangun, K.R. Benak, J. Economy and K.L. Foster, Carbon, 39 (2001) 1809.
16 C.L. Mangun, K.R. Benak, M.A. Daley and J. Economy, Chem. Mater., 11 (1999)
3476.
17 M.A. Daley, C.L. Mangun, J.A. DeBarr, S. Riha, A.A. Lizzio, G.L. Donnals and J.
Economy. Carbon, 35 (1997) 411.
18 C.L. Mangun, J.A. DeBarr and J. Economy, Carbon, 39 (2001) 1689.
19 Z.Yue, C. Mangun, J. Economy, P. Kemme, D. Cropek and S. Maloney, Environ. Sci.
Technol., 35 (2001) 2844.
20 J. Economy, L.C. Wohrer and F.J. Frechette, Ion Exchange Fiber. Ind. Res. I.R.-100
Award Ann., Sept., 1972.
21 L. Dominguez, K.R. Benak and J. Economy, Polym. Adv. Technol., 12 (2001) 197.
22 K. Benak, L. Dominguez, J. Economy, C. Mangun and Z. Yue, Proc. of the Annual
AWWA Conference, Denver, CO, 2000.
23 S.S. Barton, M.J.B. Evans and J.A.F. MacDonald, Adsorp. Sci. Technol., 10 (1993) 75.
24 Y.F. Jia and K.M. Thomas, Langmuir, 16 (2000) 1114.
25 D.M. Ruthven, Principlesof Adsorption andAdsorption Processes.Wiley, New York,
1984.
26. M.C. Roman-Martinez, D. Cazorla-Amoros, A. Linares-Solano, C. Salinas-Martinez de
Lecea and F. Atamny, Carbon, 34 (1996) 719.
27 D. Rivin, J. Aron and L.B. Richards, U.S. Patent 3442679 (1969).
28 A.K. Mittal and C. Venkobachar, Ind. Eng. Chem. Res., 35 (1996) 1472.
29 I.N. Ermolenko, I.P. Lyubliner and N.V. Gulko, Chemically Modified Carbon Fibers
and Their Applications. VCH, New York, 1990.
30 B.R. Puri, In: P.L. Walker (Ed.), Chemistry and Physics of Carbon. Marcel Dekker,
New York, 1970. Vol. 6.
31 J.R. Dale and J. Irving, in: J.A. Greig (Ed.), Ion Exchange Developments and
Applications, Proceedings of IEX '96. SCI, Great Britain, 1996.

1021
Chapter 32

New polymeric extraction materials

Abdul Malik

32.1 INTRODUCTION

Sample preparation is an important step in chemical analysis, especially when

dealing with traces of target analytes dispersed in complex matrices. Such matri-
ces are commonplace in environmental, petrochemical, and biological samples.
Such samples are not generally suitable for direct introduction into analytical
instruments for their quantitative or qualitative determination. This incompati-
bility is related to two factors. First, the complex matrices may have a detrimen-
tal effect on the performance of the analytical system, or they may interfere with
the analysis of the target analytes. Second, target analyte concentration may be
so low in the sample that it may go beyond the detection limit of the analytical
instrument. In both cases a sample preparation is necessary to make the sample
detectable or compatible with analytical instrumentation. This is achieved
through sample clean-up and sample preconcentration. Sample derivatization is
also sometimes necessary to facilitate analysis and detection of target com-
pounds.
Sample preparation is the most time-consuming and error-prone step in the
analytical cycle [1], and has long been a neglected area in analytical chemistry
[2]. As a result, sample preparation techniques have not developed at a pace con-
sistent with the rapid growth and sophistication of the analytical techniques [3].
For example, the Soxhlet extraction technique [4], invented more than a hun-
dred years ago, is widely utilized in current analytical practice for the extraction
of solid samples. Liquid-liquid extraction is another classical extraction tech-
nique that is still extensively used in various formats. In recent years, however,
the importance of sample preparation in the overall analytical process is becom-
ing increasingly recognized by analytical chemists and scientists using analytical
chemistry in their research [5-9]. Another factor that had a positive impact on
further development of extraction techniques is the realization of harmful envi-
ronmental and health effects of large volumes of organic solvents used in classi-
cal sample preparation techniques [9-11]. The positive stand of US Environ-
mental Protection Agency (EPA) and other regulatory agencies worldwide
towards reduction in the use of harmful organic solvents has created a favorable
environment for the emergence and/or further development of environmentally
Comprehensive Analytical ChemistryXXXVII
J. Pawliszyn (Ed.)
© 2002 Elsevier Science B.V. All rights reserved 1023
benign sample preparation techniques including solid-phase extraction (SPE)
[12], supercritical fluid extraction (SFE) [13-17], accelerated solvent extraction
(ASE) [18,19], microwave-assisted extraction (MASE) [20-251, subcritical water
extraction [26-28], liquid-liquid microextraction (LLME) [29-34], extraction on
a single solvent drop [35-39], and membrane extraction [2,7,40-45].
Recently, much effort has been made to bring sample preparation techniques
to the right level of speed and sophistication compatible with those of modern
analytical techniques. This is evident from numerous recent developments in
the area of sample preparation with emphasis on miniaturization, high through-
put, and automation. Especially notable are the explosive growths in the area of
SPE [1,12,46-49] and SPME [50-571.
Polymers are an important group of materials that often provide superior
performance in analytical sample preparation by allowing effective extraction
and preconcentration of trace analytes in an environmentally benign way. In
solid-phase extraction, polymeric materials provide a number of advantages over
their conventional bonded silica counterparts. Among others, these include
performance stability over a wide pH range allowing for the extraction of acidic,
basic, neutral, and ionic species [1], ease of surface modification through graft-
ing [58] to achieve desired levels of sorbent polarity, better recovery, and
reduced dependence of the extracted amount on pre-extraction wetting period.
Development of solid-phase microextraction by Pawliszyn and coworkers [59]
demonstrated the possibility of complete elimination of the use of organic
solvents during extraction and/or preconcentration of target analytes. In this
technique a polymeric coating on a fused silica fiber efficiently performs extrac-
tion and preconcentration of analytes from various matrices.
Besides their use in traditional SPE and SPME formats (i.e., cartridge-,
disk-, or packed precolumn-based SPE and fiber-based SPME), polymeric
materials are being introduced in new, non-traditional extraction formats
including (1) open tubular microextraction (OTME) (also called in-tube SPME)
which uses a polymeric coating on the inner surface of a piece of capillary to
perform extraction, (2) monolithic capillary microextraction (MCME) which
utilizes a porous monolithic bed as the extraction medium, and (3) polymeric
solution-mediated extraction.

32.2 SCOPE OF THIS CHAPTER

This chapter focuses primarily on new synthetic polymeric materials in analyti-

cal extraction techniques. The text is divided into three major thematic areas
where new polymeric materials have been developed and/or utilized as extrac-
tion media: (a) SPE, (b) SPME, and (c) polymeric solution-mediated extraction.
A brief introduction to the conventional sorbents used in these areas is given to
highlight the historical background that dictated the development of new
polymeric sorbents. The text also includes brief descriptions of the working
principles and specificities of these techniques in various (especially newly

1024
developed) formats. In cases where recent extensive reviews are available for
particular types of polymeric extraction materials, the main attention is focused
on the most recent developments not covered by existing reviews. Molecularly
imprinted polymers are now well-established extraction materials for SPE, and
two very recent reviews [47,48] have thoroughly covered the existing literature.
However, in the SPME area, MIPs have been introduced only very recently. For
this reason, emphasis in this chapter has been placed on molecularly imprinted
materials as extraction media for SPME. Extensive reviews have also been
published on chemically modified polystyrene-divinylbenzene materials for SPE
[1,49,60]. Taking this into consideration, only a brief discussion is provided here
of these topics, mainly covering the general principles and latest developments
not covered by those recent reviews.
Special attention is focused on sol-gel polymers that advantageously combine
organic and inorganic components in their structures. A general scheme of
sol-gel chemistry is presented that allows for the in situ creation of these poly-
mers in a simple way under mild thermal conditions (often at room temperature)
using appropriately designed sol solutions. The sol-gel polymeric coatings and
monolithic beds prepared using such sol solutions get chemically bonded to the
surface of the substrate and, therefore, possess significantly higher thermal and
solvent stabilities-attributes that are important for effective extraction and
subsequent release of the analytes from the coatings for further analysis. The
organic-inorganic hybrid nature of these polymers provides advanced material
properties allowing for the extraction of both polar and nonpolar analytes pres-
ent in the matrix. A wide variety of functional groups can be incorporated in
these polymeric materials by properly designing the sol solution composition.
The preparation and extraction performance of sol-gel polymeric coatings and
monolithic beds with chemically incorporated polymers, dendrimers, macro-
cycles, and other organic ligands are presented. Future directions in the develop-
ment of sol-gel polymers for analytical extraction are pointed out. To our
knowledge, sol-gel polymers have not yet been covered by any of the existing
reviews on polymeric extraction materials.

32.3 NEW POLYMERIC MATERIALS IN SOLUTION-MEDIATED

EXTRACTION

Liquid-liquid extraction (LLE) [61] is a classical extraction technique for effect-

ive isolation of dissolved organics from aqueous media. The fact that LLE often
employs large volumes of organic solvents to perform extraction makes it
hazardous to the environment and to the human health. Because of recent
regulatory pressure to reduce the use of hazardous organic solvents, much
research work is being devoted to finding an effective alternative to LLE.
Aqueous polymeric solutions have the potential to replace hazardous organic
solvents used in LLE to perform liquid-phase extraction in an environmentally
benign way [62]. Extraction techniques that employ polymeric and related

1025
solutions include (a) cloud-point extraction (CPE) [63,64], (b) micellar extrac-
tion (ME) [65,66], (c) extraction based on aqueous biphasic systems (ABS) [67],
and (d) extraction using thermoseparating polymer systems (TPS) [68]. Rogers
and coworkers [62] have published an extensive review covering the existing
literature on all four of these polymer solution mediated extraction techniques
up to and including 1999. Here we will focus on new developments (not covered
by the above-mentioned review) with emphasis on future trends in this area.

32.3.1 Dendrimers in solution-mediated extraction

A new trend in solution-mediated extraction technology is to use dendrimer

solutions to perform the extraction. Goetheer et al. [69] described the use of
functionalized poly(propylene imine) (PPI) dendrimers as reactive extractants
for anionic species from an aqueous phase to a supercritical carbon dioxide
phase. The same dendrimers were found to be efficient phase transfer catalysts.
Baars et al. [70] showed that poly(propylene imine) dendrimers with apolar end
groups provide highly selective extractants for basic analytes. They used palmi-
toyl chloride-modified PPI dendrimers (1-5 generations) to extract anionic
xanthene dyes from water to an organic phase. The modified dendrimer provided
inverted unimolecular dendritic micellar structures with a polar interior of
tertiary amines and an apolar periphery. Excellent extraction efficiency was
reported: 50 dye molecules per dendrimer molecule (fifth generation). The
authors concluded that the extraction of anions is related to the tertiary amine
interior of the dendritic micelle, and provides a very useful tool in selective
purification of anions.
Powell et al. [71] synthesized carbon dioxide-soluble polymers with multiple
ligand sites for CO2-based metal extraction. The CO2-soluble polymers were
prepared via free-radical polymerization of a fluorinated acrylate and a series of
acrylate- and styrene-based monomers functionalized with the desired ligand
sites or ligand precursors. They successfully extracted copper and europium
using CO2-solutions of the prepared polymers. Based on europium luminescence,
the authors established that in this extraction system europium was bound
(along with four molecules of water) to the polymer in the inner coordination
sphere. Extraction techniques based on CO2-mediated polymeric solutions are
appealing due to the environmental friendliness of CO2, the high purity of
carbon dioxide as a solvent, and ease of solvent removal from the extracted
materials.

32.3.2 Temperature-responsive polymers (TRP) in solution mediated

extraction

Temperature-responsive (or thermoresponsive) polymers belong to stimuli-

responsive polymeric materials-a group of polymers whose molecular
conformations and certain physical properties are sensitive to external stimulus

1026
factors. In the case of temperature-sensitive polymers, the stimulus factor is
temperature. Temperature-responsive polymers respond to the temperature
change in the environment by a corresponding change in their molecular
structure and properties including phase transition, solubility, hydrophobicity,
hydrophilicity, etc. [72]. Examples of other types of stimuli-responsive polymers
are pH-responsive polymers, photo-responsive polymers, electric field-respons-
ive polymers, and chemical species-responsive polymers. In the latter cases, the
external stimuli are pH, light, electric field, and chemical species.
Temperature-responsive polymers are soluble in water. However, above a
certain temperature (a characteristic value for each TRP), called lower critical
solution temperature(LCST), a temperature-responsive polymer becomes insol-
uble in water due to temperature-induced phase transition. This physical change
is reversible, and the polymer goes back into solution when its temperature is
lowered below its LCST. Thanks to this property, TRPs have been extensively
used in diverse areas of science and technology: as an effective medium for con-
trolled drug delivery [73-77], as controlled permeability membranes [78,79], as
bioconjugates [80,81], as nontoxic media for cell culture [82,83], as chromato-
graphic stationary phases [72,84-89], and as environmentally benign extraction
media [90-99].
Poly(N-Isopropylacrylamide) (PNIPAAm) is the most widely used and best
studied among temperature-responsive polymers known to date. Scheme 32.1
illustrates synthesis of PNIPAAm through radical teleomerization of N-iso-
propylacrylamide using 3-mercaptopropionic acid as the chain transfer reagent.
It also shows the possibility of chemical attachment of PNIPAAm to silica
particles that can be used either as a chromatographic stationary phase or as a
solid-phase extraction media [72,80,100]. Cole et al. [101] described an alterna-
tive synthetic scheme via aqueous redox polymerization of N-isopropyl-
acrylamide.
One important factor responsible for the wide use and popularity of poly(N-
isopropylacrylamide) as a temperature-responsive polymer is its moderate value
for the lower critical solution temperature (-32°C) which, being just above the
room temperature, is very convenient for experimentation [97,102]. Below
LCST, the polymer chains are hydrated to an expanded, water-soluble form.
Above LCST, the polymer chains dehydrate and shrink to a water-insoluble form
that separates out of the solution as a gummy precipitate that can be easily iso-
lated.
The temperature-responsive phase-transition property of these polymers
provides a powerful analytical tool for the preconcentration of trace analytes in
aqueous media. When a temperature-responsive polymer is solubilized in an
aqueous medium (below LCST) containing trace of organic contaminants, the
polymer chains are dispersed throughout the entire volume of the solution, and
can bind with individual molecules of the trace contaminant through hydropho-
bic and/or other forms of molecular level interactions. If, after an appropriate
contact period, the system is warmed above LCST, the temperature-responsive

1027
 H2C=CH HooCCH 2 CH 2 S-CH 2 CH -H
C =O AIBN C=O
NH + HOOCCH 2CH 2 SH - NH
CH DMF CH
/
H3C CH3 H3C CH

[PAAm MPA Semitlechelic PIPAAmr

0O
N-OH O
_________ IjN- 2 sSCHS
-CCH CH-CH

DCC 0 NH
CH
/
H3C \CH3

Silica O CHH 2 CHc2NH

2 I

Silica SIO CH2 CHCtN-CCH 2CH2SCHCHH

*- x,._) I C=O
1,4-dioxane
NH
CH
/
HC \CH3

Scheme 32.1. The synthesis of poly(N-isopropylacrylamide) (PIPAAm) and the coupling to

aminopropyl silica surface. (Reproduced from Ref. [72], with permission of Marcel Dekker).

polymer chains precipitate out together with the contaminant molecules bound
to them. This translates into an effective capture of the contaminant molecules
(that were originally dispersed throughout the entire volume of the aqueous
sample) and their transfer to the precipitated polymeric matrix. Since the
volume of the precipitate is exceedingly small compared with the total volume of
the solution, the temperature-responsive precipitation results in a remarkable
enrichment of the target trace analytes.
The hydration/dehydration phenomena associated with the temperature-
responsive polymers also provide a mechanism to fine tune the selectivity of
these polymers based on the temperature to be used. At temperatures below
LCST, these polymers are hydrophilic while above LCST they are hydrophobic
[72]. This opens up new possibilities for liquid-phase separations in an environ-
mentally benign way by using aqueous mobile phase (no organic modifier) and
TRP-bonded stationary phases, and fine tuning the hydrophobicity of the sta-
tionary phase by adjusting the temperature. Essentially, this gives the possibil-
ity to replace mobile phase composition programming (gradient elution) with
temperature-programmed elution.
Temperature-responsive polymers have been used for the extraction of
hydrophobic and amphiphilic analytes. Figure 32.1 illustrates the preconcentra-
tion effect obtained for hydrophobic pesticides in ground water through poly-
mer-mediated extraction using PNIPAAm as the TRP.

1028
 (A) (B)

8
c
X a.
0
0

IL

10min
ii
d 2
I , .
F ff 'O
i-I I II
Ii I - II -. .- 0 10 20 30 0 10 20 30
Lni ch I Time (min) Time (min)
Left: Fig. 32.1. Chromatograms of some pesticides (0.1 gM) in groundwater. Mobile phase, 70%
(v/v) acetonitrile; flow rate, 1 ml/min; detection wavelength, 254 nm. (A) without concentration,
(B) with polymer-mediated extraction. The polymer phase formed from a 10-ml sample solution
containing 0.1% (w/v) PNIPAAm was dissolved by 80 l of acetonitrile. A 20-a1l portion of the
resulting solution was injected. (Reproduced from Ref. [97], with permission of American
Chemical Society).
Right: Fig. 32.2. Chromatograms of river water: (A) without preconcentration, (B) with precon-
centration via polymer-mediated extraction. Conditions: 250 x 4.6 mm i.d. Intersil ODS column;
30% (v/v) aqueous acetonitrile mobile phase; 1.0 ml min' flow rate; UV detection at 254 nm.
(Reproduced from Ref. [98], with permission of Elsevier.)

An example of temperature-responsive preconcentration of amphiphilic

analytes is presented in Fig. 32.2. Here nonionic surfactant, nonylphenol (NP)
and its monoethoxylated derivative were pereconcentrated using PNIPAAm as
the temperature responsive preconcentration medium.
Analyte preconcentration with temperature-responsive polymers does not
need the elution step commonly required by other extraction techniques (e.g.,
SPE). This simplifies the analytical procedure. In temperature responsive
extraction, the extracted analytes are dissolved in a very small volume of water
or aqueous/organic medium and the solution is directly injected into the HPLC
system. Saitoh and coworkers [97,98] explained that since in aqueous solution
PNIPAAm is hydrophilic, practically it does not stick to the hydrophobic
reversed-phase HPLC column, neither does it appreciably interfere with the UV
detection since it lacks a strong UV-absorbing chromophore.

32.4 NEW POLYMERIC MATERIALS IN SOLID-PHASE EXTRACTION

32.4.1 Working principle of solid-phase extraction

Solid-phase extraction (SPE) is based on preferential accumulation of the target

analyte(s) from a liquid phase into a solid adsorbent bed (usually within a cylin-
drical confinement) through which the liquid sample is percolated. This can be
materialized in a variety of formats: (a) particle-packed bed in the form of a car-

1029
tridge or a precolumn, (b) sorbent disk, (c) monolithic bed (a single piece of
porous solid within a tubular confinement), and (d) flow-through membranes.
The ultimate aim is to obtain an enriched sample of the target analyte(s) and to
free it from any interferences that might be present in the original liquid sample.
To this end, the solid-phase should be carefully selected so that it provides high
selectivity for the target analyte(s). The solid-phase selection should be based on
physicochemical and structural characteristics of the analytes to facilitate vari-
ous types of molecular level interactions with the analytes present in the sur-
rounding environment. In order for SPE to be effective, such interactions
between the target analyte(s) and the solid sorbent should dominate over analo-
gous interactions of the target molecules with the original liquid matrix.
Once the solid phase has been selected, a number of steps needs to be
performed in sequence to obtain a clean, preconcentrated sample of the target
analyte(s) free of interferences. These are: (a) activation (conditioning)- pre-
treatment of the solid phase with an appropriate solvent to create a favorable
environment on the solid surface for intimate interaction between the target
analyte(s) and the solid sorbent; (b) sample application-passing the liquid
sample through the solid bed to provide sorbent/analyte interaction to achieve
retention of the target analyte(s); (c) washing-passing an appropriate liquid
through the solid bed to selectively remove the interferences (but not the target
analyte(s)); and (d) elution-recovery of the extracted and purified target
analyte(s) by passing through the sorbent bed a small volume of a solvent that
has higher affinity for the target analyte(s) than the affinity of sorbent toward
the analyte(s). The practical aspects of performing SPE on a cartridge are
illustrated in Fig. 32.3.
Solid-phase extraction is a more environmentally benign extraction tech-
nique than liquid-liquid extraction (LLE). SPE drastically reduces the consump-
tion of hazardous organic solvents. It is simple and easy to operate. Thanks to its
simplicity, reliability, and applicability to a diverse range of samples, SPE has
undergone explosive growth over the last few decades, and has now become the
most popular sample preconcentration technique [46].
It is evident from the above discussion that the heart of SPE is the solid
sorbent used for extraction. Future advances in SPE are strongly dependent on
the development of new types of sorbents with advanced sorption characteris-
tics. Novel synthetic polymeric sorbents have great promise for future develop-
ments in SPE. Before addressing this central topic of the chapter, a brief
discussion highlights the advantages and drawbacks of conventional solid-phase
extraction media with emphasis on how polymeric materials can overcome the
limitations of conventional extraction media and expand the scope of SPE.

32.4.2 Traditional solid-phase extraction materials

Traditional SPE sorbents include silica-based bonded HPLC packings [46],

carbon-based extraction media [104,105], and polystyrene-divinylbenzene

1030
 (A) Specimen
reservoir
Medical-grade
polypropylene

Polyethylene
Sorbent bed _ ? fritted disk (20 p)

Fritted disk
(20 )
Luer lip

..f~,

Fig. 32.3. (A) Typical design of a (B) Conditioning A Sample ilution

solid-phase extraction car- 6 solvent solvent
tridge. (B) Steps in the opera-
tion of a solid-phase extraction
experiment. I = compound of (typical manual
interest, * = interference. (Re- injection volume)
produced from R.E. Majors, /
LC-GC, 4 (1986) 972, by
permission of Advanstar
6
Waste Sample Wash
50 Ipl

Elution
Communications). application

(PS-DVB) [1]. The following discussion on traditional SPE sorbents will be brief
since the focus of this review is on new polymeric sorbents.

32.4.2.1 Silica-basedextraction materials

Although silica is commonly represented by a simple molecular formula (SiO,),
in reality, it is an inorganic polymeric material [106]. In silica, silicon and oxygen
atoms form tetrahedral structures with the silicon atom at the center, and the
four oxygen atoms on four corners of each tetrahedron. In amorphous silica
(most commonly used in chromatographic separations and sample prepara-
tions), each of these tetrahedra may be chemically attached to up to four
neighboring tetrahedra through silicon-oxygen (siloxane) bonds in a random
way [107]. Underivatized silica materials are used as an easily available media
for normal-phase HPLC separations and sample purifications. Functionalized
silica materials are widely used in sample preparation techniques including SPE
[108], and as chromatographic stationary phases for HPLC [109,110], SFC
[111], capillary electrochromatography [112,113], and GC [114-116].
Silica-based extraction materials are characterized by high mechanical
strength that makes them well-suited for high-pressure and high-voltage opera-
tions often required in liquid-phase separations and sample preparations. How-
ever, silica-based materials inherently possess a narrow pH stability range
which excludes their operation under extreme pH conditions. Besides, most
silica-based materials (including alkyl-bonded silicas, the most widely used types

1031
of silica-based stationary phases and extraction media) are characterized by low
retention for polar analytes. In solid-phase extraction applications, this results
in low breakthrough volumes that make such materials often unsuitable for the
preconcentration of polar compounds. Because of these limitations, much
research effort is directed towards developing alternative materials that possess
a wider range of pH stability. These include the development of polymeric
materials [1], zirconia- [117-122] and titania-based materials [123-127], and
polymer-encapsulated silica [128-132]. Recently, Kirkland and coworkers [133]
reported the possibility of creating silica materials with a wide pH stability range
(up to pH 11) via sol-gel technology.

32.4.2.2 Carbon-basedextraction materials

Carbon-based sorbents have found useful applications as extraction media [104,
105] and as stationary phases for chromatographic separations [134-136]. The
main applications of carbon-based materials in analytical sample preparation
involve trapping and preconcentration of volatile organic compounds (VOCs)
from air and water. They are also used for the preconcentration of semi-volatile
and non-volatile compounds, although to a much lesser extent.
Four different types of carbon-based materials have found industrial, envi-
ronmental, and analytical applications. These are: (a) activated carbon [137,
138], (b) graphitized carbon black [139-145], (c) carbon molecular sieve [146,
147], and (d) porous graphitic carbon [148-155].
(a) Activated carbon: The use of activated carbon can be traced back to the
ancient Egyptians who used activated charcoal for medicinal purposes. More
recently, activated carbon has found a range of applications thanks to its high
surface area and strong sorptive properties. This especially concerns applica-
tions involving (a) sampling of volatile compounds with boiling points higher
than 0°C (activated charcoal is not a strong enough adsorbent for the retention
of gases with boiling points less than 0°C [156]) and (b) isolation and removal of
certain undesirable and/or hazardous components from gaseous or liquid media.
The first class of applications includes the use of activated carbon as a cheap and
efficient sorbent for use in preconcentration tubes for sampling industrial and
ambient air. The trapped volatile analytes are then recovered from the sampling
cartridge by desorbing with a suitable solvent-most frequently carbon disulfide
(for nonpolar analytes) or binary mixture of CS2 with water, methanol, 2-propa-
nol, 2-butanol, or dimethylformamide (for polar analytes) [156,157]-and fur-
ther analyzed by gas chromatography. The second class of applications include
use of activated carbon in water purification, air filtration, manufacture of respi-
rators and protective gas masks, recovery of solvents from process streams, etc.
Activated carbon is also a promising candidate for use as an adsorbent in minia-
turized traps [158-162].
The strong interaction of activated carbon with adsorbed analytes makes it
unsuitable for many analytical applications (especially those involving semi-
volatile-, low-volatile-, and/or polar compounds) where the extracted analytes

1032
need to be quantitatively recovered. Because of this strong sorbate/sorbent
interaction, it is difficult to desorb such analytes from activated carbon materi-
als resulting in poor recovery. Besides, activated carbon materials are character-
ized by complicated surface and micropore structures containing various polar
groups [163] that are difficult to reproduce, resulting in significant batch-to-
batch variations in their properties. The adsorptive properties of activated car-
bons are greatly affected by the atmospheric humidity. The adsorptive capacity
of activated carbon strongly depends on its micropore structure. The huge
surface area (-1000 m2 /g) of activated carbon may promote certain catalytic
reactions. In such a case, the composition of desorbed analytes may not be repre-
sentative of those originally adsorbed.
(b) Graphitized carbon black (GCB): The term "graphitized carbon black"
refers to the forms of carbon prepared by thermal treatment of ordinary carbon
black to 2700-3000°C in an inert gas environment, usually in a graphitizing
furnace [164,165]. This high-temperature treatment leads to the breakdown of
various functional groups of the original carbon black surface, and results in a
material with a homogeneous surface that is practically free from unsaturated
bonds, ions, free radicals, and lone electron pairs. Graphitized carbons are also
free of micropores, and possess a typical surface area of 100 m2 /g [142]. Solute
adsorption takes place on the external surface of GCB predominantly due to
dispersive interaction between the solute molecules and the nonspecific adsorp-
tion sites on the GCB surface containing hexagonal array of carbon atoms. GCB
may partially regain surface heterogeneity during cooling of the carbonaceous
surface after graphitization [140] and the small amounts of polar adsorption
sites present on the surface are capable of offering specific interactions with
polar solutes. From a crystallographic point of view, three distinct types of
graphitized carbon structures can be differentiated: (a) Bernal perfect three-
dimensional hexagonal structure of graphite [166]; (b) Lipson-Stokes
three-dimensional rhombohedral structure of graphite [167]; and (c) Warren
two-dimensional turbostratic structure of graphite [168].
The use of graphitized carbon blacks and carbon-based extraction media in
solid-phase extraction have recently been reviewed [104,105]. Recent SPE appli-
cations of GCB include the preconcentration of phenols [140,142,144,169],
pesticides and herbicides [142,145,169-176], estrogens [177], alkylbenzene-
sulfonates and metabolites in fish [178], and surfactants in water [142]. Non-
polar nature of graphitized carbon and a wide range of pH stability makes this
material attractive for use as an effective sorbent for extraction and precon-
centration of analytes under extreme pH conditions not suitable for silica-based
extraction media. In this respect, GCB enjoys the advantages inherent in poly-
meric extraction materials. However, GCB has some significant shortcomings
that include excessive solute retention and poor mechanical stability. Attempts
have been made to improve mechanical strength of graphitized carbon
materials. To this end, a new type of sorbent, porous graphitic carbon (PGC),
was developed by immobilizing graphitized carbon on a silica structure.

1033
 (c) Porous graphitic carbon (PGC): Graphitized carbon black is practically
nonporous, and has a homogeneous hydrophobic surface. Activated carbon, on
the other hand, is porous but suffers from surface heterogeneity. Therefore,
carbon-based porous materials with homogeneous hydrophobic surfaces present
considerable theoretical and practical interests [148-152,155]. PGCs are such
materials; like GCB, they are characterized by a wide pH stability range. They
are prepared by impregnating a porous template material (e.g., silica) with
organic polymerizing mixture (e.g., phenol-formaldehyde mixture, phenol-
hexamine mixture, etc.) [148,153,154] and the polymerization reaction is
conducted within the pores of the template material. After the polymerization
reaction, the in situ created polymer is converted to glassy carbon by heating to
-1000°C in an inert environment. At this point, the silica template is removed
by treating with an alkali to produce porous graphitic carbon which is further
heated to 2000-28000°C in an inert environment. This thermal treatment
removes the micropores, improves surface homogeneity, and produces some
degree of graphitization determined by the temperature at which the thermal
treatment is carried out. The particle and pore size characteristics of the
resulting PGC is determined by the chosen template material, while the surface
characteristics of the PGC is determined by the final thermal treatment and any
other chemical treatment that may follow.
A study was undertaken by Puig and Barcelo [144] to compare the perform-
ances of PGC and polymeric sorbents in SPE for the preconcentration of
phenolic compounds in environmental waters. The results showed that except
for one compound, 4-chloro-2-aminophenol, PGC provided inferior performance
both in terms of breakthrough volumes and detection limits.
(d) Carbonmolecularsieve (CMS): Carbon molecular sieves are carbon-based
microporous sorbents with very uniform pores [146,147]. They are characterized
by high surface area and strong retention of organic molecules. They are
prepared by controlled pyrolysis of suitable organic materials (e.g., polyvinyl
chloride, petroleum pitch, etc) at temperatures usually exceeding 400°C. They
possess disordered cavity-aperture structure resulting from the cross-linking of
crystallites. Important factors that determine the pore structure, pore distribu-
tion, and properties of CMS include (a) identity of the pyrolyzed polymer, (b)
purity of the polymer, (c) the used method of carbonization, and (d) presence of
impurities. Carbon molecular sieves are commercially available (e.g., Carbosieve
and Carboxen types from Supelco).

32.4.3 Polymeric extraction materials in SPE

32.4.3.1 Poly(styrene-divinylbenzene)-basedmaterialsfor SPE

Polymeric materials have found wide applications in solid-phase extraction
technique [1,46-49] as well as chromatographic column packings [179]. Among
polymers that have been used as extraction media are (a) polystyrene-divinyl-
benzene (PS-DVB) [1,46,49] and polysiloxane-based materials [180,181]. Porous

1034
PS-DVB-based sorbents have gained wide popularity among various polymers
that have been evaluated as extraction media. This is explained by the fact that
PS-DVB-based sorbents are characterized by a number of advantageous proper-
ties compared with silica-based surface-bonded extraction media, originally
developed for use as stationary phases in HPLC. Although both the silica-based
reversed-phase extraction materials and PS-DVB-based sorbents have hydro-
phobic surfaces, thanks to the presence of aromatic benzene moieties on the
surface, PS-DVB-based extraction media allow for selective extraction through
n-n: interactions with the analyte molecules.
By far the most outstanding advantage of PS-DVB-based polymeric sorbents
is their stability over a wide range of pH. This makes them suitable for extrac-
tion even under extreme pH conditions that are not compatible silica-based
sorbents with surface-bonded organic ligands. A second advantage of the PS-
DVB-based extraction materials is their ability to extract analytes from aqueous
media without requiring prolonged preconditioning with a polar organic solvent.
Silica-based surface-bonded extraction media often require such a precondition-
ing step because of the poor wettability of the bonded hydrophobic ligands with
water.
Further improvement in the wettability of PS-DVB-based sorbents with
water, and hence their ability to extract analytes from aqueous media, was
achieved through derivatization of PS-DVB with various polar groups including
acetyl [182-184], sulfonate [185], hydroxymethyl [182,183], benzoyl [186],
o-carboxybenzoyl [187], 2,4-dicarboxybenzoyl [188], 2-carboxy-3/4-nitrobenzoyl
[188], and tetrakis(p-carboxyphenyl)porphyrin [189,190]. The chemical basis of
all these derivatization reactions is presented in Scheme 38.2 [49]. PS-DVB has
already become a widely used SPE material. Various aspects of synthesis and use
of PS-DVB and functionalized PS-DVB have been covered in recent reviews
[1,49,188b]. Details on the development of PS-DVB-based extraction materials
and their use in analytical extractions can be found in these and other publica-
tions [182-190].

32.4.3.2 Molecularly imprintedpolymers in SPE

Recent years have witnessed a rapid growth in the development and utilization
of molecularly imprinted polymers [47,48,191-198] especially as highly selective
extraction materials for SPE and SPME [47,48,199-208] and as chromato-
graphic stationary phases [192,193,209-212]. The affinity of the sorbent for a
particular solute (or class of solutes) can be greatly enhanced by creating binding
sites (on the sorbent) that have complementary structures relative to the target
analyte(s). The process involved in creating such binding sites is termed "molec-
ular imprinting". To create molecularly imprinted extraction materials, a temp-
late (usually the target analyte or a closely related compound) is used during the
sorbent preparation. The template molecules are temporarily included into the
structure of the polymeric sorbent during its synthesis from monomers. Subse-
quently, the template molecules are removed from the sorbent leaving behind

1035
 AOH

N., __
% -T

CH3COCI H
A10 3

Cx/C' 0
\Cb
c0t
CO(t

IN(CHAJ

0-

Scheme 32.2. Scheme of chemical modifications of polystyrene-divinylbenzene resin.

(Reproduced from Ref. [49], with permission of Elsevier).

binding sites on the sorbent with structural features complementary to those of

the target solute(s). The principle is illustrated in Fig. 32.4 [191].
The above-illustrated principle is materialized in a versatile way through
self-assembly of the template and a functionalized monomers undergoing poly-
merization and cross-linking reactions [47]. As the polymerization proceeds, the
template molecules remain incorporated in the growing polymer. Subsequent
removal of the template molecules leaves behind template-selective binding sites
in the polymeric structure.

1036
 (a) fsH
+-OH
+2 I -

Cor 0 H
synthesise
adduct

0:)- I polymerise

(b)

n mamix
in excess

Fig. 32.4. Schematic representation of the molecular imprinting process for: (a) covalent
imprinting of phenyl -D-mannoside using vinylphenylboronic acid; and (b) non-covalent
imprinting of Boc-phenylalanine using methacrylic acid. (Reproduced from Ref. [191] by
permission of Elsevier).

Molecularly imprinted extraction materials show high solute specificity

comparable with or even better than that of immunoaffinity extraction
materials [214] prepared through immobilization of antibodies using time-
consuming multiple-step procedures. A significant difference in the extraction

1037
 A B C

Naproxen
Nap roxen
Ibuprofcn Ibuprofen

Naproxen

6
I

5A tob1
r 20 o 5 lo i 2 -o 6 i- i15 20
Time (min) Time (min) Time (min)

Fig. 32.5. HPLC Chromatograms of a standard Ibuprofen sample (A), control plasma sample (B),
and control plasma sample spiked with Ibuprofen (C) using column switching techniques.
Conditions: 20 gl injection volume; precolumn, RAM-MIP-1 (10 x 4 mm i.d.); analytical column,
Cosmosil 5C-18-MS (150 x 4.6 mm i.d.); eluent for pretreatment, 20 mM phosphoric acid-
acetonitrile (78:22 (v/v), pH 2.24) at 1.0 ml min/min for 5 min; mobile phase, 20 mM sodium
phosphate buffer-acetonitrile (75:25 (v/v), pH 7.34) at 1.0 ml/min; detection, UV absorbance at
223 nm; Ibuprofen concentration at 5.0 )g/ml in (A) and (C). (Reproduced from Ref. [215] by
permission of American Chemical Society).

properties of the MIPs and immunoaffinity sorbents is that immunoaffinity

sorbents generally show high affinity for the solutes in aqueous media while
MIPs typically exhibit their optimum solute recognition in organic media [213].
Complete removal of the template molecules from MIPs after their
preparation is of prime importance. However, it may be difficult to achieve
complete elimination of the template molecules even after repeated washing of
the prepared MIP using various solvents. In such cases, template bleed during
analysis (i.e., gradual release of the residual template molecules that could not
be removed from the MIP after its preparation and cleanup) may adversely
affect the accuracy and precision in the trace-level determination of the target
analyte that was used to imprint the polymer. This problem was overcome by
using a template, which is structurally related to (but not identical with) the
target analyte. For example, to prepare an ibuprofen-specific MIP, Haginaka
and Sanbe [215] used naproxen (which is structurally similar to ibuprofen) as
the template. Use of this MIP should not affect the trace level determination of
ibuprofen even in the event of template bleed, since these two compounds elute
wide apart from each other in HPLC as illustrated in Fig. 32.5.
Another problem encountered with the use of MIPs for the isolation of target
analytes from biological samples is the adsorption of biopolymers (proteins,
nucleic acids, etc.) on the surface of the MIP leading to a number of undesirable
effects including, blockage of the sorbent pores, reduction of retention and sam-
ple capacity, and template recognition ability of the MIP. This also affects the
reproducibility of the analytical procedure. A possible solution to this problem is

1038
to develop MIPs with restricted-access material (RAM) characteristics as will be
discussed in the next section.

32.4.3.3 Restricted-access media in SPE

Restricted-access materials (RAMs) are specially designed extraction media that
are widely used for the extraction and purification of small molecules (e.g., drugs
and their metabolites) from biological fluids containing proteins, nucleic acids or
other biopolymers. Direct introduction of such samples into a chromatographic
column is undesirable, since the presence of adsorptive biopolymers in the
sample may lead to a number of undesirable effects including retention time
shift for the analytes, reduced sample capacity, low analytical reproducibility,
column clogging, increased column back pressure, reduction of column lifetime,
and column failure. Conventional solid-phase extraction materials are difficult
to use for the purification of such biological fluids since proteins and other
biopolymers get adsorbed on the surface of the extraction media blocking the
pores. RAMs are designed in such a way that the biopolymers cannot enter into
their pore structure due to small pore diameters, charge repulsion, and/or other
molecular barriers placed on their way to the pores. The internal surface (i.e.,
surface within the pores) of such a material is derivatized with hydrophobic
(and/or ion exchange) ligand(s) while the outer surface is provided with a bonded
hydrophilic layer presenting an inert environment for the biopolymers.
When a biological sample is passed through a RAM extraction bed, small
target molecules are retained by the ligands anchored inside the pores, while
proteins and other biopolymers get excluded from the pores and move down the
bed with the liquid flow through interparticular channels. Further, biopolymers
have little chances for adsorption on the RAM outer surface that presents a
hydrophilic inert environment for biopolymers.
Historically, restricted-access materials have been differentiated into four
major types [216]: (a) internal surface reversed-phase (ISRP) [217], (b) semi-
permeable surface (SPS) [218,219], (c) shielded hydrophobic phase (SHP) [220],
and (d) mixed functional phase (MFP) [221,222].
An ISRP restricted access material consists of high-performance silica mate-
rial with two types of surface-bonded moieties: (a) hydrophobic group(s) bonded
to the internal surface residing within the pore structure, and (b) hydrophilic
group(s) bonded to the external surface located outside the pores. This type of
RAM can be prepared by using silica containing immobilized polypeptide groups.
The hydrophilic external layer is created by selective enzymatic hydrolysis of
those polypeptide groups that are located on the RAM surface outside the pores
[223-225]. ISRP RAMs have been prepared using silica with various immobi-
lized surface moieties including glycine-L-phenylalanine-L-phenylalanine
(GFF) [217,226] and N-octanoylaminopropyl [227,228] groups. ISRP RAMs
have been reported to have a typical pore size of 5.2 nm [229] and they exclude
proteins with a molecular weight higher than 15,000 dalton [230]. Protein exclu-
sion by such RAMs is assumed to be based on a combination of size exclusion and

1039
charge repulsion mechanisms. In GFF ISRP, for example, negatively charged
glycine groups are present at the pore entrance and can take part in the exclu-
sion of negatively charged proteins.
A semipermeable surface (SPS) RAM particle normally consists of a core
region of functionalized silica and a covalently bonded hydrophilic layer of
poly(oxyethylene) on the external surface. Practically, any derivatized silica
support (or even non-functionalized, bare silica) can be used to prepare this type
of RAM. The external coating of poly(oxyethylene) serves as a hydrophilic
barrier to proteins and other biopolymers. This hydrophilic polymeric barrier,
effectively prevents macromolecules from entering into the porous structure of
the RAM resulting in their exclusion. Small molecules, however, can diffuse
through the hydrophilic barrier and interact with the core material resulting in
their retention. Compared with ISRP RAMs, SPS restricted-access materials
possess greater analytical flexibility since they can be used for the extraction of
both polar and nonpolar analytes. This advantage of SPS RAMs stems from the
flexibility in using cores with a wide variety of functional groups-polar or
nonpolar. It should be mentioned that only nonpolar functional groups are used
in ISRP RAMs allowing for the extraction of only nonpolar analytes.
Shielded hydrophobicphase (SHP) RAM consists of silica particles contain-
ing a bonded hydrophilic polyethylene oxide) network with imbedded hydropho-
bic phenyl moieties. In this type of RAM the network covers the entire surface of
the silica support-both inside and outside the pores. Small molecules can
diffuse through the network and interact with the imbedded phenyl groups
resulting in their retention. Hydrated macromolecules (e.g., proteins), on the
other hand, are excluded because of the hydrophilic shield provided by the
poly(ethylene oxide) network.
Mixed functionalphase (MFP) RAM consists of silica particles functionalized
with a hydrophilic diol group and a hydrophobic group that can be chosen from a
wide selection of hydrophobic moieties including phenyl [231,232], butyl [232],
octyl [232], and cyclodextrin [221,233].
A general drawback of RAMs is the limited flexibility in manipulating the
retention characteristics. This is explained by the fact that to avoid protein
precipitation during extraction, the mobile phase organic modifier content needs
to be kept at a low value (< 25% for acetonitrile, < 20% for 2-propanol, and < 10%
for THF) [223]. Since silica-based RAMs are predominantly used in current
analytical practice, a second inherent disadvantage of using RAMs is the narrow
pH range over which these materials can be operated.
Development of organic polymer-based RAMs could effectively overcome
this difficulty. However, to date, there are very few reports on the development
and use of totally organic polymer-based restricted-access materials, although
polymers have found applications in silica-based RAM technology as a semi-
permeable hydrophilic barrier 218,2191 and as a hydrophilic shielding network
[220]. Haginaka et al. [215,234-235] have recently described a procedure to
prepare molecularly imprinted sorbents with restricted-access properties

1040
 (S)-naproxen
V-65 4-vinylpyridine
dibutyl phthalate toluene EDMA
swelling _ swelling swelling
seed particle
hydrophi
GMMA
GCMA
KS2 0,
polymerization hydrophilization

Scheme 32.3. Synthetic scheme of surface-modified MIP for (S)-naproxen (Reproduced from Ref.
[235], with permission of Elsevier).

(RAM-MIPs) using polystyrene-based seed particles. For this, they used a

multi-step swelling and polymerization procedure as illustrated in Scheme 32.3.
The (s)-naproxen-specific MIP was created on -1 ,um swollen polystyrene
particles using 4-vinylpyridine as the functional monomer and ethylene glycol
dimethacrylate as the cross-linker. Next, the outer surface of the prepared mate-
rial was hydrophilized by using a 1:1 mixture of glycerol monomethacrylate
(GMMA) and glycerol dimethacrylate (GDMA). The technical details of the
whole procedure can be found in Refs. [235,236]
Figure 32.6 shows scanning electron micrographs of MIP and RAM-MIP
particles prepared by the above-mentioned procedure. RAMs have traditionally
been used in the reversed-phase mode. However, normal-phase mode operation
is also possible. Recently, Gustavson et al. [237] demonstrated such a possibility.
Using a RAM with a bonded polyethylene oxide-based external coating and an
underivatized silica core, those authors selectively extracted methyl oleate from
a nonpolar hexane environment containing 16 EPA priority PAH pollutants. A
second trend is the development of RAM substitutes. In this direction, attempts
have been made to use variously functionalized porous polymers as a substitute
for RAMs. However, protein adsorption is still a problem for those polymer-
based materials.
Development of RAMs has had a significant impact on the speed and automa-
tion in biological sample cleanup and analysis. It opened new possibilities for
direct sample introduction into a RAM precolumn for sample cleanup followed
by on-line transfer of the extracted analytes into the HPLC system using a col-
umn switching technique [238-241]. Further developments in the area of
restricted-access media are likely to be focused on the development of polymeric
materials that have the potential to overcome the shortcomings of silica-based
RAMs, particularly the problems associated with their narrow pH stability
range.

32.4.3.4 Sol-gel polymeric materialsfor SPE

Sol-gel chemistry [106,242] provides an elegant pathway for the synthesis of
organic-inorganic hybrid polymers with advanced material properties. A signifi-
cant advantage of the sol-gel route is that sol-gel reactions take place under

1041
Fig. 32.6. Scanning electron micrographs of the MIP (A, C) and RAM-MIP (B, D) for
(S)-naproxen. Magnifications: 2000x (A, B) and 10000x (C, D). (Reproduced from Ref. [215] by
permission of American Chemical Society).

extraordinarily mild thermal conditions-often at room temperature. Besides,

the organic/inorganic composition and the structure of the synthesized materi-
als (and hence their properties) can be controlled by adjusting the composition of
the sol solution. A typical sol solution contains one or more sol-gel precursors
(metal alkoxides), an organic ligand with a sol-gel active functional group, a
sol-gel catalyst (an acid, a base, or a fluoride), dissolved in a suitable solvent sys-
tem containing controlled amounts of water. In the presence of water, the sol-gel
precursors undergo hydrolysis, and the hydrolyzed products take part in poly-
condensation reactions in a variety of ways leading to polymeric chain growth
through chemical bonding of the hydrolyzed species among themselves and/or
with various sol-gel-active species present in the sol solution. Ultimately, these
reactions result in the formation of a hybrid organic-inorganic sol-gel polymer.
Although the hydrolytic polycondensation route is most frequently used for
the synthesis of sol-gel materials, nonhydrolytic pathways for such syntheses
have also been reported [243-247]. The use of sol-gel polymers in analytical
extractions began relatively recently. Niessner and coworkers [248-252] intro-
duced sol-gel immunosorbents (ISs) to perform immunoaffinity extraction of
various polycyclic aromatic hydrocarbons and their derivatives from aqueous
media. They prepared stable immunoaffinity sorbents by encapsulating appro-

1042
 34

29

- 24
E
19

14

-1
0 5 10 15 20 25 30 35
Time (min)

Fig. 32.7. HPLC chromatograms of urine samples obtained after (a) C-18 SPE (upper plot) and
(b) immunoextraction (lower plot) of hydroxyphenanthrenes (OH-PHEs) and hydroxypyrenes
(OH-PYRs). Conditions: 10-ml volumes ofurine sample from the same subject, final volume 3 ml,
100 lp injection volume. Peaks: (1) 2-/3-OH-PHEs, (2) 9-OH-PHE, (3) 1-OH-PHE, (4)
4-OH-PHE, and (5) 1-OH-PYR. (Reproduced from Ref. [252] by permission of American
Chemical Society).

priate antibodies in the sol-gel matrix. The sol-gel immunosorbents effectively

retained the target analytes and provided sample cleanup by eliminating inter-
ferences. Figure 32.7 presents two LC chromatograms obtained after extraction
of hydroxyphenanthrenes in a human urine sample using (a) sol-gel immuno-
sorbent (lower trace) and (b) a conventional ODS bonded phase (upper trace).
The advantage of the sol-gel immunosorbent over conventional ODS-bonded
solid-phase extraction material is evident from a comparison of the two HPLC
traces. Compared with the ODS phase, the sol-gel immunosorbent provided a
much cleaner extract by efficiently eliminating the matrix interferences. The
flat baseline of the lower HPLC trace (Sol-gel IS) compared with the elevated
baseline of the upper trace (ODS phase) makes this point very clear.
Seneviratne and Cox [253] prepared dimethylglyoxime (DMG)-doped micro-
porous sol-gel silica material for the solid-phase extraction of nickel (II) from
aqueous samples. They also prepared diethylenetriamine (DTA)-doped meso-
porous sol-gel silica to extract copper (II). An interesting phenomenon observed
by those authors was that in micropores of the sol-gel material, DMG forms 1:1
complex with nickel (II) compared with the characteristic 1:2 ratio observed in
free solution. For sol-gel materials with a liquid phase confined in the porous
structure, such apparent deviation from "normal" behavior has been observed in
others cases too. An example is the unexpected extraction of aromatic hydrocar-
bons into polar sol-gel silica monolith from hydrophobic solvents [254]. In the
confined environment of liquid-filled silica pores, the hydrogen bonding between
the silanol groups and other chemical species may play a decisive role in the

1043
TABLE 32.1
Extraction recyclability through four extraction/stripping cycles with 33 mg of sol-gel in 5.0 ml of
SrCl 2 solutions. Extraction was conducted were conducted at 23°C. (Reproduced from Ref. [255],
with permission of American Chemical Society).
Extraction cycle [Sr2 + ] (ppm) Capacity Kd %Sr2+ removed
Initial Final
1 25.0 2.50±0.18 3.9x10- 2 1.4x10 3 90.0+0.2
2 32.25 5.2±0.4 4.7x10 2 7.8x10 2 83.8+0.4
3 24.7 0.29±0.02 2.8x10 -2 1.3x10 4 98.8+0.2
4 27.6 1.44+0.11 4.2x10 2 2.8x10 3 94.8-0.1

occurrence of unexpected phenomena like extraction of aromatic hydrocarbons

by polar silica matrix from a hydrophobic solvent medium.
Yost and coworkers [255] described a crown ether-encapsulated sol-gel mate-
rial for selective extraction of Sr (II) from an aqueous solution containing excess
of competing ionic species such as Ca (II). The crown ether ligand, 1,4,10,13-
tetraoxa-7,16-diazacyclooctadecane-7,16-bis(malonate), is known for high affin-
ity toward Sr(II) ions and was encapsulated in the hydrophilic silica through
sol-gel process. The extracted Sr (II) ions were easily recovered from the sol-gel
matrix by washing with an acid or ethylenediaminetetraacetic acid (EDTA)
disodium salt. This washing also regenerates the sorbent, and extraction proce-
dure can be repeated with the regenerated sorbent. The used crown ether encap-
sulated sol-gel material was found to remove 91.4 ± 1.3% of the target Sr (II)
ions from aqueous solution in presence of excess Ca (II) ions (Table 32.1).
The simplicity of the sorbent preparation, extraction, and sorbent regenera-
tion together with high recovery of Sr (II) ions in the presence of excess calcium
ions, gives the sol-gel material a competitive edge over traditional sorbents with
chemically grafted crown ether ligands.
Sol-gel hybrid organic-inorganic materials have tremendous potential for
being used as extraction media as well as chromatographic stationary phases.
Section 32.6.3.3 will present the applications of sol-gel polymeric materials in
SPME, and capillary microextraction using sol-gel surface coatings or mono-
lithic beds.

32.5 NEW POLYMERIC MATERIALS FOR MEMBRANE-BASED

EXTRACTION

Membrane-based extraction techniques [256-261] are attracting increasing

interest in modern analytical practice. A membrane represents a selective
barrier between two phases [262] through which preferential mass transfer can
take place from one phase into another. The phase from which the mass transfer
takes place is called a donor phase or feeding phase, while the phase on the other

1044
Fig. 32.8. Schematic representation of transport through mem- Donor Membrane Acceptor
branes. (Reproduced from Ref. [263] by permission ofElsevier). phase phase

side of the membrane that receives the transferred mass is called the acceptor
phase or permeate phase. Figure 32.8 illustrates the selective mass transfer
through a membrane [263].
Membrane-based extraction and separation processes are based on differen-
tial transport rates of various chemical species through the interface. In mem-
brane-based systems, the mass transport process can be divided into three
categories [263,264]: (a) passive, (b) facilitated, and (c) active. Passive transport
is commonly used in applications dealing with analytical extractions or separa-
tions. In this mode, the membrane serves merely as a physical barrier. Here, the
driving force for the transport process is the electrochemical potential gradient
of various molecular or ionic species between the donor and acceptor phases
located on the two sides of the membrane. In the facilitated mode of transport, in
addition to the electrochemical potential gradient, a carrier is used in the mem-
brane to enhance the permeability during the transport process. Active trans-
port is typical of living cellular membrane where the transport takes place
against the electrochemical potential gradient, and is caused by a chemical reac-
tion taking place inside the membrane.
Based on the nature of the used membrane and the trapping media (if any),
membrane extraction techniques can be differentiated into: (a) supported liquid
membrane extraction (SLME) [40,42], (b) microporous membrane liquid-liquid
extraction (MMLLE) [42,265], (c) membrane extraction with sorbent interface
(MESI) [257,266-271], and (d) polymeric membrane extraction (PME)
[272-278]. Here the discussion will be confined primarily to the use of polymeric
materials in membranes extraction.
Luo et al. [273] have studied the effect of polymer swelling on the mass trans-
fer performance in membrane extraction. Fu et al. have reported the use of
potentiometry to study extraction thermodynamics of polyanions into plasticized
polymer membrane doped with lipophilic ion exchangers [275]. Melcher and
coworkers [276-279] developed and studied membrane extraction systems with
silicone rubber membranes and various combinations of aqueous/organic liquids.
Recently, Hauser and Popp [261,280] described a silicone rubber hollow
fiber-based dynamic membrane extraction device (Fig. 32.9) to enrich and
analyze volatile organic compounds from water (or air). Being battery-operated,
portable, and easy-to-interface with a portable GC, the device appears to be

1045
 Peristaltic
Ambient air Water sample - pump l

1 Silicone hollow
fibre
I
Activated
charcoal -+

Tosample inlet
of AirmoBTX

Fig. 32.9. Schematic of a flow cell for dynamic extraction with hollow fibers. (Reproduced from
Ref. [280] by permission of Elsevier).

perfectly suitable for field operation. Volatile organic compounds in water are
allowed to diffuse through the silicone hollow fiber, and are enriched in the
integrated activated charcoal-based sorption tube. Typical enrichment time is 10
min. The enriched analytes are then thermally desorbed and analyzed by GC. A
typical example of BTEX analysis in water using dynamic membrane extraction
using a silicone hollow tube membrane is presented in Fig. 32.10. Low /zg/l
extraction sensitivity was achieved. Li and Zhang [281] developed poly(vinyl-
chloride)-based membranes to extract zinc ions from aqueous media. Mayer et
al. [108] compared the performance of octadecylsiloxane-bonded silica- and
porous polymer particle-loaded membranes in solid-phase extraction.
Recently, Jonsson and Mathiasson published extensive reviews on supported
liquid membrane extraction and related techniques [44,282]. Cordero et al. [263]
reviewed the interfacing of membrane extraction techniques with GC, HPLC,
and CE. Van de Merbel published an elaborate review on on-line coupled mem-
brane-based extraction techniques [2] covering principles, instrumentation,

Benzene Toluene
60000

50000

Ethylbenzene
40000
m-Xylene

30000
o-Xylene

20000

10000

I,
'Y-PL------
1, , m

0t fWY-lea
_- PLW s, i XI ]AI ,1

0.00 50.00 100.00 150.00 200.00 250.00

Fig. 32.10. Gas chromatogram obtained after extraction of 5 .og/l BTEX in water with a 0.3 m
silicone hollow fiber 0.7 x 0.9 mm. Temperature, 20°C; extraction time, 10 min. (Reproduced
from Ref. [280] by permission of Elsevier).

1046
interfacing (to GC, LC, and CE), and application aspects of dialysis, electro-
dialysis, and filtration techniques. Bauer [259] reviewed application of mem-
brane extraction in conjunction with mass spectrometry (membrane
introduction mass spectrometry, MIMS).
One limitation of liquid membrane-based extraction systems is the short
membrane lifetime. This can be overcome by using a polymeric membrane
instead of liquid membranes. Silicone rubber-based membranes can be a good
choice since they possess extended lifetimes. Another significant advantage of
polymeric membranes is that they are practically insoluble in aqueous or many
organic media, and allow for the use of extraction systems with various combina-
tions of aqueous and organic liquids. On the other hand, the solute diffusion
through polymeric membranes is slower than that in the liquid membranes.
This results in slower extraction by polymeric membranes. Another disadvan-
tage of using polymeric membranes is that the composition of the membrane
remains fixed during the extraction process, which translates into reduced
flexibility in fine-tuning the extraction selectivity. Since the silicone rubber-
based membranes are hydrophobic, their use as the extraction medium is
primarily limited to the extraction of nonpolar analytes, and may not be suitable
for the extraction of polar compounds. To overcome this difficulty, hydrophobic
polymeric membranes have been surface-modified with a thin layer of some a
hydrophilic polymer (e.g., polyacrylate) [283,284].
Polymeric membrane extraction is still an emerging area; an important
trend is the flow-through solid-phase extraction using polymeric membranes
(membrane SPE). This is consistent with the general trend of miniaturization in
separation and extraction technologies. Materialization of membrane extraction
in a flow-though mode should make this extraction technology much faster
compared with the traditional passive or facilitated modes commonly employed
in liquid membrane extraction techniques (SLME or MMLLE).
In a recent report, Fritz and Masso [285] described a miniaturized flow-
thorough membrane extraction device using a miniaturized (1.2 mm in height
and 0.7 mm in diameter) sulfonated polystyrene-divinylbenzene resin disk
housed inside the removable needle of a 50 Al Hamilton gas-tight syringe as
illustrated in Fig. 32.11. The membrane rested on a stainless steel screen. All
these parts were assembled in a Teflon ferrule-a standard part of the used
Hamilton syringe. Aqueous samples containing eight substituted benzene
compounds (at 10 ppb level concentration) were automatically passed through
the membrane using a single-syringe infusion pump. It required only 5 ul
volume of acetonitrile to elute the extracted compounds. Analyte recovery was
higher than 90% with an average relative standard deviation of 4.6%.
Another significant advancement in the area of membrane SPE is the recent
development of polymeric membranes with a surface-grafted molecularly im-
printed layer [272,287,288]. Wang et al. [287] prepared membranes with surf-
ace-grafted MIPs. For this, they used a special photoreactive polymer, poly-
(acrylonitrile-co-(diethylamino)dithiocarbamoylmethylstyrene). Using theo-

1047
 <_ Miniaturized
membrane

<_ Teflon
ferrule

<(- 55 pm Mesh screen

It

Fig. 32.11. Miniaturized-SPE assembly inside

removable needle chamber. (Reproduced from Ref.
[286] by permission of Elsevier).

phylline as the template, these authors conducted graft copolymerization of

acrylic acid and N,N-methylenebis(acrylamide) (MBA) under UV initiation that
resulted in a theophylline-specific membrane. However, such a procedure was
lengthy and complicated that required a special polymer for membrane forma-
tion. The resulting membranes were characterized by low permeability, large
macrovoids, and strongly asymmetric pore morphology.
The most recent developments in the area of photografted MIP membranes
were reported by Ulbricht and coworkers [272,288] who developed a general
method for molecular imprinting and surface functionalization of porous
synthetic polymeric membranes. Those authors developed a procedure for MIP
synthesis in aqueous environment using desmetrin (a triazine herbicide) as the
imprint molecule [288]. Their aim was to achieve affinity membrane extraction
of the template molecules from aqueous solutions. The authors optimized the
synthesis, extraction, and elution conditions. Extraction was performed by
filtering the aqueous solution through the MIP membranes at high flow rates,
typically 10 ml/min. A 200-fold enrichment and a low nanomolar extraction
sensitivity (2 x 10 - 9 M) were achieved.
Further development in the area of photografted MIP membranes was
reported in ajoint publication by researchers from Germany (Wulff and cowork-
ers) and Ukraine (Sergeyeva and coworkers) [272]. In this joint work, the
authors described a surface functionalization procedure to create a thin molecu-
larly imprinted polymeric layer on microfiltration polymer (polyvinylidene
fluoride) membranes with a hydrophilic polyacrylate surface layer. The main
focus of this work was to create a surface-grafted MIP layer that will minimize

1048
 50
MP
40-

330-

Fig. 32.12. Selectivity of terbumeton- 20-

imprinted thin layer MIP composite mem- i
brane to other herbicides of related chemical I10-

permission of Elsevier).
-5
structure. Herbicide concentrations, 10 M
in water. (Reproduced from Ref. [272] by o LL~~E
terbumeton atrazine desmertyn metribuzine

the nonspecific interaction between the membrane and the analytes (thereby
maximize the effect of template-specific interaction), at the same time retaining
the high permeability of the membranes without blocking the membrane pores.
They photopolymerized a methanolic solution containing a functional monomer
(2-acrylamido-2-methyl-l-propane sulfonic acid (AMPS), methacrylic acid
(MAA), or acrylic acid (AA)) and a cross-linker (N,N-methylene-bis-acrylamide)
(MBAA) using UV radiation and benzophenone as the photoinitiator. A rela-
tively hydrophobic triazine herbicide, terbumeton, was selected as the template
molecule. In contrast to their previous work on the development of a surface-
grafted MIP synthesis procedure using aqueous solutions [288], the focus here
was on the synthesis of MIP grafted surface layer using polymerizing solutions
prepared in an organic solvent. Such a method should greatly increase the
number of templates that can be used for the synthesis.
The prepared MIP-grafted membranes showed high specificity for the
template molecule (terbumeton), as is evident from Fig. 32.12. It should be noted
that flow-through membrane SPE is very attractive from an analytical point of
view, and new developments can be expected in this area in the near future.

32.6 NEW POLYMERIC MATERIALS FOR SOLID-PHASE

MICROEXTRACTION

32.6.1 Working principle and variants of solid-phase

microextraction

Solid-phase microextraction (SPME) [50,51] is based on the principle of equilib-

rium extraction where a stationary phase coating created on the outer surface of
a micro fiber (typically 100 m in diameter) directly extracts analytes from the
surrounding medium. In the conventional format of SPME, a polymeric coating
(typically 7-100 m in thickness) is applied to the SPME fiber covering the outer
surface of an approximately 1-cm segment of the fiber at one of its ends. Such a
fiber is installed in a specially designed syringe (SPME syringe). For this, the
fiber is glued or mechanically mounted into a piece of stainless steel holder
connected with the syringe plunger. During handling and/or operation, the outer

1049
 (A) 2

(B .- I
(B) 1 2--'

Left: Fig. 32.13. Modes of SPME operation: (a) direct extraction, (b) headspace SPME.
(Reproduced from J. Pawliszyn, B. Pawliszyn and M. Pawliszyn, The Chem. Educator,2 (1997) 1,
with permission of Springer).
Right: Fig. 32.14. Schematic of a sol-gel fiber (A) and a sol-gel open tubular microextraction
capillary (B). 1, sol-gel coating; 2, fused silica fiber (A) or capillary (B).

extraction coating is protected from mechanical damage by retracting the fiber

into the needle of the SPME syringe. To carry out extraction, the fiber is brought
in contact with the analyte by exposing the coated segment of the fiber into the
sample. This is accomplished by simply depressing the plunger of the syringe.
Analytes in the surrounding medium are extracted by the stationary phase
coating on the external surface of the fiber. Agitation of the sample (e.g., using a
stir bar) is often used to assist the extraction process. The working principle of
SPME is briefly illustrated in Fig. 32.13.

32.6.2 Polymeric surface coatings for SPME

Polymeric surface coatings are predominantly used as extraction media in

conventional SPME [50-57,289-291] as well as in the newly emerging extraction
techniques like in-tube SPME [292-307] or capillary microextraction (CME)
[308]. Besides polymeric coatings, in a few instances the use of nonpolymeric
porous coatings have also been reported. These include SPME coatings created
by using carbon-based materials [143,309,310] or by gluing reversed-phase
HPLC particles onto the surface of an SPME fiber [311,312]. Surface coating-
based extraction systems are usually rapid, easy-to-automate, and suitable for
on-line operation. They are also compact and suitable for field applications.
However, the most fascinating aspect of surface coating-based extraction
systems is that they do not require exhaustive extraction of the analytes for
quantitation [313].
To create a polymeric coating on a fused silica SPME fiber, the outer
protective coating of the fiber is first removed from the designated end-segment,
and then a coating of the selected extraction polymer is applied to the bare outer
surface of that end-segment. This can be accomplished by simply dipping the
bare end segment of the fiber into a liquid containing the coating polymer.
In-tube SPME typically uses a polymeric coating on the inner surface of fused

1050
silica capillary. A small segment of a wall-coated GC capillary column is often
used for this purpose. Figure 32.14 illustrates the construction of an SPME fiber
(Fig. 32.14A) and a wall-coated capillary for in-tube SPME (Fig. 32.14B).
Both in conventional and in-tube operation modes of SPME, the polymeric
coating is exposed to the environment containing the target analytes, and time is
allowed for an equilibrium to establish. Typically, it takes less than an hour to
reach the extraction equilibrium. The extraction process is aided either by
mechanically agitating the sample with a stir bar (as in conventional SPME) or
by flowing the sample over the coated surface (as in in-tube SPME or capillary
microextraction). After this, the fiber or the capillary is withdrawn from the
sample, and the extracted analytes are rapidly desorbed from the coating for
introduction into an analytical instrument-most commonly a GC, HPLC, SFC
or a CE system. The analyte desorption can be carried out either thermally by
introducing the fiber (with the extracted analytes on its coating) into a heated
GC injection port using an SPME syringe specially designed for SPME-GC anal-
ysis, or by rinsing the coated capillary with an appropriate solvent as is per-
formed in in-tube SPME/HPLC analysis.
SPME has gained wide popularity because of its simplicity, speed, solventless
extraction capability, portability, and suitability for automation. A fascinating
attribute of SPME techniques is that they simultaneously accomplish a number
of important analytical operations in a very simple way. These include sampling,
sample preparation, and analyte preconcentration. They also provide a robust
mechanism for the introduction of the extracted analytes into an analytical
instrument for quantitation. In addition, the fact that the extracted amount of
the analyte in SPME is practically independent of the sample volume [291]
makes the techniques especially valuable for direct field sampling and analysis.
In SPME techniques, a vital analytical role is played by the polymeric coating
that serves as the extraction medium, thus making the extraction process
environmentally benign by eliminating the need for hazardous organic solvents
that are used in large volumes by conventional extraction techniques.
Conventional sorbents for SPME include two types of extraction media
[314]: (a) homogeneous polymers and (b) composite extraction media consisting
of a partially cross-linked polymer embedded with porous particles of a second
component. New polymers for SPME coatings include: (1) Nafion, (2) Poly-
pyrrole, (3) molecularly imprinted polymers, and (4) sol-gel hybrid organic-inor-
ganic polymers (sol-gel PDMS, sol-gel PEG, sol-gel-crown ether, sol-gel
dendrimer, etc.). From the polarity point of view, conventional SPME coatings
can be classified into four categories: (1) nonpolar (e.g., PDMS), (2) moderately
polar (e.g., PDMS/DVB composite), (3) polar (e.g., polyacrylate), and (4) highly
polar (e.g., Nafion).
Since PDMS is nonpolar in nature, it can effectively extract nonpolar
analytes from a wide range of matrices. However, PDMS coatings are not
efficient for the extraction of polar analytes. More polar SPME coatings are
needed for the extraction of polar analytes. Polyacrylate (PA)-based coatings

1051
[317-322] are more suitable for this type of applications. Polyethylene glycol
(PEG)-based coatings (e.g., Carbowax 20M and Carbowax/divinylbenze) are also
used [323-325]. It should be mentioned that relatively thick coatings are to be
used in SPME to provide reasonable sample capacity and extraction sensitivity.

32.6.2.1 Homogeneouspolymeric SPME coatings

Polydimethylsiloxanne (PDMS) and polyacrylate (PA) belong to the category of
homogeneous SPME coatings. Such coatings are usually applied to the external
surface of the solid SPME fiber via an automated coating procedure immediately
following the fiber drawing process. The conditions can be optimized to obtain
coatings of desired thicknesses. Commercial polyacrylate coated fibers are avail-
able with a coating thickness of 85 Am. This coating is only partially cross-linked.
PDMS fibers are most widely used in SPME, and are commercially available in
three different thicknesses: 7, 30 and 100 Am. The 7-Am thick PDMS coating is
cross-linked, while PDMS coatings with the other two thicknesses (30 and 100
Am) are not. It should be pointed out that the degree of cross-linking is difficult
to control, and a high degree of cross-linking may significantly change the sorp-
tion properties of the polymeric coating. Cross-linking makes the polymeric coat-
ings more rigid, and may change their diffusion and mass transfer
characteristics. Besides, it becomes increasingly difficult to stabilize coatings of
higher thickness through cross-linking reactions. This might explain, why com-
mercial fibers with PDMS coatings of higher thicknesses (30 and 100 am) are not
cross-linked. The smaller coating thickness and the cross-linked structure pro-
vide higher thermal and solvent stability to the 7-tm thick PDMS coating. How-
ever, because of lower coating thickness it possesses smaller sample capacity.
Non-cross-linked and partially cross-linked fibres are characterized by rela-
tively lower thermal stability compared with their cross-linked counterparts.
This lower thermal stability is often responsible for the incomplete desorption of
analytes (especially the high-boiling ones) that require higher desorption temp-
erature than the allowable upper temperature for such fibers, and leads to the
carry-over problem (memory effect) [291,315,316]. Non-cross-linked or partially
cross-linked coatings also often show low solvent stability, and SPME fibers with
such coatings are not recommended to bring into direct contact with organic
solvents that may dissolve the coatings. Because of this solvent stability
problem, SPME devices with such coatings may not be suitable for use in
hyphenation with liquid-phase separation techniques (e.g., HPLC, CEC) where
the extracted sample is often desorbed using liquid mobile phases rich in the
organic solvents.

32.6.2.2 Polymeric composite coatings

A polymeric coating of this category consists of two components: a porous
particulate component, and a nonparticulate component. SPME fibers of this
type are prepared by coating the fiber surface with a blend of the two compo-
nents in which the particulate component is embedded in the partially cross-

1052
linked polymeric component. Unlike the preparation of homogeneous polymeric
coatings, such a process is difficult to automate and is usually done manually.
Because of the presence of two sorbents in the coating, these coatings provide
improved selectivity. In addition, the fiber selectivity can be fine tuned for the
analytical problem at hand by proper selection of the two coating components. It
should be mentioned that such composite coatings possess lower mechanical
stability than the homogeneous polymer coatings. The following commercial
SPME coatings belong to this category: (1) polydimethysiloxane/divinylbenzene
(PDMS/DVB), (2) polydimethylsiloxane/Carboxen (PDMS/Carboxen), (c) Carbo-
wax/divinylbenzene (CW/DVB), and (d) Carbowax/templated resin (CW/TPR).
In written notation of these composite coatings, the non-particulate component
is usually put first, followed by a "/" sign that separates it from the particulate
component that follows. Small pieces (-2 cm) of these externally coated fibers
are then installed on individual fiber assembly to be used for extraction.

32.6.3 Newly emerging polymeric materials for solid-phase

microextraction

32.6.3.1 Molecularly imprintedpolymers for SPME

Molecularly imprinted polymers are now widely used in solid-phase extraction
(SPE), and a wealth of literature (including extensive reviews) is available on the
subject [47,48]. However, it is only very recently that MIPs have been introduced
in SPME [207,208]. High specificity of MIPs for the template molecules makes
them effective media for selective extraction of target analytes.
The first report on the use of MIPs as the stationary phase coating for
in-tube SPME was made by Pawliszyn and coworkers [208]. They packed a fused
silica capillary with particles of an MIP that they prepared by polymerizing a
mixture containing methacrylic acid (monomer), ethylene glycol dimethacrylate
(cross-linker), and racemic propranolol (template), following a procedure
described by Andersson [326]. The particle-packed capillary was used for
selective on-line extraction of propranolol and related 5-blockers from biological
fluids. For a standard mixture, the authors achieved a detection limit of 0.32
Ag/ml with an RSD value of less than 5%.
Koster et al. [207] were the first to report on the use of MIP coatings on
SPME fibers. They prepared a molecularly imprinted methacrylate polymer
using clenbuterol as a template, and demonstrated the possibility of selective
extraction of brombuterol from spiked human urine (Fig. 32.15).
Molecularly imprinted polymers have great potential for selective extraction
of target analytes from various complex matrices. However, adsorption of bio-
polymers or humic substances may serve as a barrier to full utilization of this
analytical potential. Development of MIPs with restricted-assess characteristics
might be one possibility to address this problem. Of course, we need to bear in
mind that molecularly imprinted polymeric materials constitute a newly emerg-
ing area for SPME, and a lot of research activities can be expected in this area.

1053
 100000

80000

9 60000

40000 a)

b)

Fig. 32.15. LC-ECD chromatograms of the 20000

extraction (45 min) of human urine with MIP-
coated fiber: (a) blank urine, (b) spiked urine
(100 ng/ml brombuterol), and washing of the
fiber, and (c) blank urine and washing of the 0
fiber. (Reproduced from Ref. [207] by perm-
ission of American Chemical Society). Time (min)

32.6.3.2 Polypyrrole-basedcoatingsfor SPME

Although SPME has made remarkable achievements over the last decade in
extracting nonpolar, hydrophobic compounds from aqueous media, still there
are two major areas where the progress has been only moderate. These two areas
are: (a) extraction of polar and ionic analytes, and (b) reliable hyphenation of
SPME with liquid-phase separation techniques (e.g., HPLC, CEC, etc.). The
limiting factor in the development of both of these areas is the absence of
polymeric SPME coatings with desirable sorbent properties.
A closer look at these problems indicates that an effective solution to the first
problem will require development of highly polar surface-bonded SPME coatings
that will be able to effectively compete with water to extract polar compounds
from aqueous media. If the extracted polar compounds need to be analyzed by
GC, then the polar polymeric coatings must also withstand high desorption
temperature of the GC injection port. To ensure reasonable sample capacity the
coatings should also be relatively thick (several tens of micrometer in thickness).
Furthermore, if the polar analytes are to be extracted from an organic solvent
media, then the coating must also possess high solvent stability. Development of
coatings with high solvent stability is also the key to the solution of the second
problem mentioned above.
One of the main reason for limited progress in the above two areas is that
commercially available coatings that have been mainly used to solve those
problems often fail to simultaneously satisfy all these requirements. Develop-
ment of new polymeric coatings with advanced material properties will be a
prerequisite not only for the solution of these and other outstanding problems,
but also for the further development of SPME as an extraction technique.

1054
 To this end, Pawliszyn and coworkers [304-307,327] recently developed
poly(pyrrole)-based coatings for SPME. Their main emphasis was to develop
polypyrrole (PPY) and poly(N-phenylpyrrole) (PPPY) coatings for reliable
hyphenation of in-tube SPME with HPLC for automated on-line extraction and
HPLC analysis. Outstanding solvent stability of these polymers made them an
appropriate choice for this job, since in such an interface the desorption of the
extracted analytes is normally carried out using organic-rich solvents. Another
important attribute of polypyrrole-based coatings is their outstanding pH stabil-
ity. For example, PPY coatings provided stable performance within a wide pH
range (1.5-10) [307]. PPY coatings possess a porous structure with an extraction
mechanism primarily governed by surface adsorption of the analytes. Table 32.2
compares the SPME performance of PPY coatings with that of commercially
available capillaries with the following coatings: (a) polar silica (host), (b)
Omegawax (Omeg), (c) SPB-1, and (d) SPB-5. The data clearly demonstrate the
superiority of the PPY coating over its commercial counterparts in the extrac-
tion of PAHs. The authors explained this enhanced extraction capability of PPY

TABLE 32.2
Comparison of the extraction efficiencies for PAHs on different capillary coatings (adapted from
Ref. [307], with permission of American Chemical Society)
PAH Compound Detector Amount of analyte extracted (ng)c or extraction yield
responseb (%)d
F
Host Omeg SPB-1 SPB-5 PPY
Naphthalene 0.058 0.9 1.7 1.9 3.0 5.9
Acenaphthylene 0.046 0.6 1.8 2.5 3.7 7.6
Acenaphthene 0.027 0.7 1.8 4.1 4.6 6.7
Fluorene 0.213 1.2 2.1 3.8 6.3 8.1
Phenanthrene 0.050 0.8 2.5 4.4 5.4 11.1
Anthracene 0.024 0.8 2.6 4.9 7.3 11.8
Fluoranthene 0.090 1.0 3.2 5.8 10.3 17.5
Pyrene 0.115 1.1 3.1 6.1 12.0 18.0
Benza[a]anthracene 0.082 1.5 4.8 6.1 9.9 20.7
Chrysene 0.052 1.4 4.4 5.2 9.2 18.7
Benzo[b]fluoranthene 0.072 1.9 6.3 6.4 8.6 23.3
Benzo[k]fluoranthene 0.107 1.5 5.8 5.7 8.4 15.7
Benzo[a]pyrene 0.079 1.8 6.0 6.0 8.8 18.7
Dibenz(a.h)anthracene 0.098 1.4 4.9 2.9 6.1 10.7
Benzo(ghi)perylene 0.099 1.6 5.2 3.1 6.7 10.8
Indeno(1,2,3-cd)pyrene 0.088 19 7.3 6.3 8.8 16.0
aCompound concentrations and experimental conditions (see notes b, c, and d).
bF (detector response factor) was obtained by injecting 10 l of 1,g/ml solution (for each PAH 10
ng was injected).
CA 1-ml sample (100 ng/ml for each PAH) was analyzed by in-tube SPME, and the amount of
analyte extracted (nA) was calculated by equation: nA = FA = (m/Ad)A; the extraction yield (%)=
nA x 100/M where is the total amount of analyte in the 1 ml solution (see Ref. [307] for details).
dSelectivity factor was calculated only for 15 cycles on the basis of nA relative to n (naphthalene).

1055
 44.0

PPPY Film

19.5

14.5

9.5

Fig. 32.16. SPME-GC analysis of

40 g/ml methanol (1), ethanol (2), 4.5
and 2-propanol in hexane using
PPPY-coated fiber. (Reproduced
from Ref. [306] by permission of -0.5 6
Elsevier). Time [min]

coatings based on effective T-Ti and hydrophobic interactions between the PPY
coatings and PAH solutes. Enhanced extraction by PPY-coated capillaries was
also observed for polar compounds, such as phenol and its derivatives. However,
PPY-coated capillaries showed reduced extraction efficiency for monocyclic aro-
matic hydrocarbons compared with commercial capillaries. PPY-coated capillar-
ies also showed enhanced extraction efficiency for organoarsenic compounds and
for anionic species [305]. This may be explained based on the multifunctional
nature of the polymer and the positive charge that it carries.
Another significant advantage of PPY coatings was found to be their very lit-
tle affinity for nonpolar solvents like hexane. This fact, combined with high sol-
vent stability of PPY coatings, opens new possibility of extracting polar
impurities in nonpolar organic solvents. Figure 32.16 shows a chromatogram,
illustrating SPME (followed by GC analysis) of lower aliphatic alcohols from
hexane using PPY coating. It should be pointed out that PPY coatings are held
on the substrate by physical forces. Their high solvent stability is due to poor sol-
ubility of these polymers in a wide range of common organic solvents.

32.6.3.3 Sol-gel polymeric extraction media for solid-phase microextraction

In its traditional format, SPME has a number of drawbacks. First, since the
stationary phase coating is applied to the outer surface of the fiber, it is more
vulnerable to mechanical damage. Second, due to the lack of chemical bonding
between the coatings and the substrate, traditional coatings do not possess
adequate thermal and solvent stability. Third, free radical cross-linking proce-
dures [328,329] that are commonly used to immobilize thin sub-micrometer
coatings in GC columns, do not work very well for thick (several tens of
micrometers in thickness) stationary phase coatings needed for SPME. This is
especially true for polar stationary phase coatings [330,331].
In a new variant of SPME, so-called in-tube SPME [208,292-308] short seg-
ments of commercially available capillary GC columns are used to extract
analytes. However, the commercial GC capillary columns have thin stationary
phase films (which translates into low sample capacity) as well as low thermal

1056
and solvent stabilities. In most part, this is due to the lack of chemical bonding
between the stationary phase coating and the inner walls of the capillary to
which these coatings are applied. Sol-gel coating technology [332-336] easily
overcomes these problems by providing surface-bonded organic-inorganic
hybrid stationary phase coatings. The sol-gel coating technology is applicable for
creating stationary phase coatings both on the outer surface of conventional
SPME fibers and on the inner walls of the capillary used for in-tube SPME (cap-
illary microextraction). In addition, sol-gel technology can be used for the cre-
ation of both thin and thick coatings. Finally, sol-gel technique is very much
suitable for in situ creation of porous monolithic beds [337-352] within a fused
silica capillary using a properly designed sol solution that solidifies and turns
into a single piece of porous structure chemically bonded to the inner walls of the
capillary. Monolithic beds are analogs to particle-packed sorption beds. How-
ever, because of the facts that the porous sol-gel monolithic bed is created with-
out packing sorbent particles, and that the bed is directly bonded to the inner
walls of the capillary, sol-gel monolithic beds are characterized by enhanced
mechanical stability.

Preparationof sol-gel polymeric coatings for SPME

Preparation of a sol-gel coating involves the following operations: (1) prepara-
tion of the sol solution, (2) bringing the surface of the substrate (fiber or
capillary) in contact with the sol solution, and (3) thermal treatment of the
coated surface.
Preparationof the sol solution. The preparation of sol solution has been
previously described [333,335]. The sol solution consists of the following ingredi-
ents: (a) at least one sol-gel precursor, (b) a surface deactivation reagent, (c) an
organic polymer (or ligand) with sol-gel-active functional group(s), (d) a sol-gel
catalyst, and (d) a suitable solvent system. Typical composition of a sol solution,
and roles played by its components in the surface-bonded coating creation
process are summarized in Table 32.3.
The following is an example of preparing a sol solution that can be used to
create a sol-gel PDMS coating, either on the outer surface of an SPME fiber or on
the inner surface of a fused silica capillary: 0.1 g of hydroxy-terminated PDMS is
dissolved in 100 Al of methylene chloride placed in a micro centrifuge tube (1.5
ml); PMHS (40 Al) and MTMS (100 Al) are added to this solution, and the con-
tents of the centrifuge tube are thoroughly vortexed. Finally, 100 Al of TFA con-
taining 5% water (sol-gel catalyst and source of water) is added to the centrifuge
tube and vortexed again. The contents of the centrifuge tube are then centri-
fuged for 4 min at 13,000 rpm (15,682 G) to separate out any precipitate that
might have formed during the mixing process. The clear sol solution from the top
part of the centrifuge tube is then transferred to a clean vial for further use.
Application of sol-gel coating to the outer surface of SPME fiber. The first
report on the preparation of fibers with sol-gel coatings was made by Malik and
coworkers [333]. The protective polyimide coating was burnt off from an approx-

1057
TABLE 32.3
Chemical ingredients of a sol solution designed for the creation of a sol-gel PDMS coating.
(Reproduced from Ref. [335], with permission of the Royal Society of Chemistry).
Sol-gel solution Chemical name Function
ingredient
Sol-gel precursor Methyltrimethoxy- Serves as a source for the inorganic component
silane of the sol-gel organic-inorganic hybrid coating
and delivers it through hydrolytic polycondens-
ation reactions. Also provides active silanol
groups on the sol-gel coating to facilitate its
chemical bonding to the fiber/capillary surface.
Organic polymer 1) Hydroxy-terminated Provides organic component for the sol-gel
with sol-gel-active polydimethylsiloxane organic-inorganic hybrid coating through
functional groups polycondensation reactions with the hydrolyzed
products of the precursor.
Deactivation Poly(methylhydro- Provides desired level of high-temperature
reagent siloxane) derivatization of silanol groups on the sol-gel
coating and thus allows for the control of the
organic-inorganic hybrid coating polarity. The
derivatization reaction takes place during
thermal conditioning of the fiber after
completion of the sol-gel coating procedure.
Sol-gel catalyst Trifluoroacetic acid Plays a dual role int he sol-gel process: (a) as an
(TFA) acid-catalyst for the sol-gel process; (b) as a
source of water for the hydrolysis of the pre-
cursor (the commercial TFA used in this work
contained approx. 5% of water). It thus provides
a mechanism for the sol-gel process to be con-
ducted in organic solvent-rich environment with
controlled amount of water.

imately 1 cm end segment of a fused silica fiber using a cigarette lighter. The
bare end of the fiber was then carefully cleaned by using a piece of Kimwipe
tissue paper soaked with methanol, making sure that no soot particles remained
on the surface. The exposed end of the fiber was further treated with a 0.1 M
NaOH solution for 30 min to enhance the concentration of surface silanol groups
that would facilitate chemical bonding of the sol-gel coating to the exposed
surface. The treated end of the fiber was thoroughly washed with deionized
water. At this point, the fiber was ready for coating. To coat, the treated end of
the fiber was vertically dipped into the previously prepared sol solution placed in
a vial. The fiber was allowed to stay in this position for -20 min. Following this,
the fiber was withdrawn from the solution and subjected to a stepwise (or
programmed) thermal conditioning in a stream of helium. For a sol-gel PDMS-
coated fiber, the final conditioning temperature was as high as 360°C. If
necessary, the entire process of dipping, withdrawing, and thermal conditioning
of the fiber should be repeated to obtain the desired coating thickness. This
procedure was used and adapted by Wang et al. [353] to prepare sol-gel polyeth-

1058
Fig. 32.17. Scanning electron micrographs of a sol-gel PDMS-coated SPME fiber: (A) cross-
sectional view, 342x magnification, and (B) surface view, 3420x magnification. (Reproduced
from Ref. [333] by permission of American Chemical Society).

ylene glycol (PEG) coated SPME fibers, and by Zeng et al. [354-356] to prepare
SPME fibers with sol-gel crown ether coatings.
Scanning electron micrographs in Fig. 32.17 illustrate the cross-sectional
view (Fig. 32.17A) and surface view (Fig. 32.17B) of a sol-gel PDMS-coated
SPME fiber. They reveal that the sol-gel PDMS coating created on the SPME
fiber is remarkably uniform (Fig. 32.17A) and has a porous structure (Fig.
32.17B) that provides enhanced surface area, and facilitates fast mass transfer
and improved sample capacity.
Application of sol-gel coatings to the innersurface of the capillaryfor in-tube
SPME. A general procedure for the creation of sol-gel stationary phase coating
on the inner walls of a fused silica capillary was first described by Malik and
coworkers [332]. Briefly, to prepare a sol-gel PDMS-coated capillary, a hydro-
thermally treated fused silica tubing (e.g., 10 m x 250 gm i.d.) was filled with the
sol-gel solution using a home-made filling/purging device [334] under 100 psi
helium pressure. The solution was allowed to stay inside the capillary for 10-30
min, after which the solution was expelled from the capillary under the same
helium pressure. Following this, the capillary was dried by purging with helium
for 30 min at room temperature. The capillary was then thermally conditioned
under a continuous flow of helium: from 40°C to 350°C at 1°C min', holding the
column at the final temperature for 5 h. Finally, the capillary was rinsed with
methylene chloride/methanol (50:50 v/v) mixture and dried under helium purge.
At this point, the capillary was ready for use in capillary microextraction. The
sol-gel coated capillary was cut into small segments (3.5 cm and longer) for use in
capillary microextraction. Following the same procedure, longer sol-gel coated

1059
Fig. 32.18. Scanning electron micrograph of a
sol-gel PDMS-coated tubing for capillary
microextraction. Magnification 10000x.
(Reproduced from Ref. [308] by permission of
American Chemical Society).

capillaries (typically 10-30 m) can also be prepared for use as a GC separation

column. Figure 32.18 is a scanning electron microscopic image of a sol-gel PDMS
coating on the inner surface of a 250-,um i.d. fused silica tubing used for capillary
microextraction [308].
Scheme 32.4 illustrates sol-gel reactions that take place in the sol solution
during in situ creation of a surface-bonded sol-gel PDMS coating, either on the
outer surface of an SPME fiber [333] or on the inner surface of a microextraction
capillary [308]. Similar reaction schemes were described by Zeng et al. for the
preparation of sol-gel crown ether coated SPME fibers [354-356] and by Wang et
al. [353] to prepare SPME fibers with sol-gel polyethylene glycol (PEG) coatings.

Advantages of sol-gel polymeric coatings in analytical microextraction

An outstanding attribute of sol-gel polymeric coatings is that they are chemically
bonded to the silica substrates to which they are applied. This gives sol-gel
coatings a number of significant advantages over conventional coatings used in
microextraction. First, because of chemical bonding to the substrate, sol-gel
coatings are characterized by high thermal stability [333]. For example, sol-gel
coated PDMS fibers can easily withstand a desorption temperature of 340°C
(Fig. 32.19). More recent experimental data showed that sol-gel PDMS coated
fibers can be routinely operated using a desorption temperature as high as
360°C. By comparison, for a conventionally coated commercial PDMS fiber, the
upper temperature limit is 250-270°C [357] which gives sol-gel coatings about
100°C desorption temperature advantage over their conventional counterparts.
This higher temperature stability, in turn, translates into more complete
desorption of the extracted analytes and higher peak area precision as illus-
trated in Table 32.4. Higher thermal stability of sol-gel coatings also provides a
reliable solution to sample carry-over problems [291,315,316] caused by
incomplete release of the extracted analytes at the desorption temperature.
Another significant consequence of higher thermal stability of sol-gel coatings is
the possibility of expanding the application range of SPME to higher boiling
compounds that will require enhanced thermal stability of the coating for their
effective desorption from the fiber.

1060
 I. Hydrolysis of tilhesol-gel precursor: (1 alkyl or alk.xy groups, and R' = alkyl or hydroxy functionalities)

R R'
I Ca(alyst I
C1130-Si-OCII 3 + 20
0 HO-Si-O
O-Si- + C 3OII

OC113 OIl

II. Polycondensation of Hydrolyzed products:

i I

0 CH3 CH3 H3
_O it O-+iiOf o--+ li-OS --- oi-O-rSi-OH
? ? ClH3 CHj CH
3

I P VH3 ICHJ CH3

-o O-it O-Si O-Si-O-oi-Ot-o i-OH
O O CH CH 3 CH 3

UI I
-OH 0° 0° ICH
3 ICH
3 CH
3
-O. + HO- SiO-ito-i-O
. ,i_-oiii-
Column Wall ? ? CH 3 CH 3 CH 3

-o.I ? ?? ¶H3 H3 FH3

rIO-lit O--it O- OS (i-i- ti-1- OH

-O 1 C113 C11 3 CH 3

Wall-bonded PDMS Coating

Scheme 32.4. Chemical reactions involved in sol-gel coating with hydroxy-terminated PDMS
stationary phase. (Reproduced from Ref. [332] with permission of American Chemical Society).

Direct chemical bonding of sol-gel coatings to the substrate surface (the

outer surface of SPME fibers or the inner surface of the microextraction capil-
lary) provides higher solvent stability to the coatings. This advantage is espe-
cially significant in those applications where the coating needs to be exposed to
organic solvents. Such applications include analyses done through hyphenation
of SPME or CME to liquid- or fluid-phase separation techniques, e.g., HPLC
[292-294,302,358-361], SFC [362] or CE [363-365]).
Solvent stability of the coating is a prerequisite for these applications since
the extracted analytes have to be desorbed from the extraction capillary using an

1061
 s

3 4
I

Fig. 32.19. SPME-GC analysis of PAHs. Conditions: 200

,am fused silica fiber, sol-gel PDMS coating (10 m thick-
ness); direct 30 min extraction with stirring; desorption
temperature, 340°C; carrier gas, helium; split ratio, 100:1;
GC column, 10 m x 250 pm i.d., sol-gel PDMS stationary 1
phase; temperature programming, from 40°C (5 min) at
6°C/min; detector, FID (360°C). Peaks: (1) naphthalene, (2)
acenaphthylene, (3) fluorene, (4) phenanthrene, (5) fluor-
anthene. (Reproduced from Ref. [335] by permission of I __
:
Y __I t
z .

Royal Society of Chemistry). 0 5 10 15 20 25 30 35 min

TABLE 32.4
Effects of desorption temperature on the peak area repeatability data for PAHs and dimethyl-
phenols in SPME-GC using sol-gel PDMS-coated fibers conditioned at different temperatures.
Conditions: extraction time for each sample (room temperature), 30 min; split-splitless injection
(splitless desorption for 5 min), injection temperature is the same as the fiber conditioning
temperature; 10 m x 250 pm i.d. sol-gel PDMS capillary GC column; column temperature
programmed from 40°C (hold for 5 min) to 230°C at 6°C/min; detector: FID, 350°C. (Adapted
from Ref. [333] with permission of American Chemical Society).
Compound %RSD (n = 3)
Fiber conditioning temperature
250°C 300°C 310°C 320°C
(A) Polycyclic aromatic hydrocarbons
Naphthalene 1.67 1.04 1.01 0.98
Acenaphthylene 3.86 3.72 2.65 2.01
Fluorene 9.25 3.33 3.20 3.00
Phenanthrene 5.93 0.65 0.52 0.50
Anthracene 3.52 1.08 1.05 1.01
Fluoranthene 2.61 1.02 0.80 0.64
(B) Dimethylphenol isomers
2,5-Dimethylphenol 7.14 3.74 0.84 0.58
3,4-Dimethylphenol 4,72 3.11 2.84 1.79
2,6-Dimethylphenol 5.91 4.84 1.03 1.00
2,3-Dimethylphenol 8.73 4.78 2.65 1.23

1062
TABLE 32.5
The extraction precision of the sol-gel PEG coated fibers conditioned at 280°C. The evaluation
was made over an extended period of operation consisting of 150 runs. M.P. stands for mean peak
area. (Reproduced from Ref. [353] with permission of Elsevier).
Compound Extraction period
10th 50th 100th 150th
M.P. RSD M.P. RSD M.P. RSD M.P. RSD
( (%) (%) (%)
Benzene 14,355 6.2 14,210 5.3 14,307 5.7 14,345 5.8
Toluene 18,301 5.9 18,219 5.7 18,269 6.1 18,247 6.0
Ethylbenzene 25,220 4.7 25 341 5.1 25,274 5.3 25,297
p-Xylene 26,691 3.9 26,536 4.4 26,598 3.7 26,674 4.1
o-Xylene 29,732 4.4 29,775 4.1 29,826 4.7 29,793 3.9

organic or aqueous-organic solvent system. Another application area where

high solvent stability of the coating is essential, involves extraction of analytes
from a nonaqueous environment. It should be mentioned that the surface
coatings of most conventionally prepared SPME fibers and/or microextraction
capillaries is not chemically bonded to the substrate surface. Immobilization of
the coating through cross-linking reactions provides some degree of solvent
stability, but direct chemical bonding gives sol-gel coatings a significant sol-
vent-stability advantage over conventional coatings.
Because of direct chemical bonding to the substrate, sol-gel coatings possess
extended lifetimes as is illustrated by the data presented in Table 32.5. These
data illustrate the high degree of run-to-run repeatability of the mean peak area
in sol-gel SPME-GC analysis over a course of 150 runs. This, in turn, translates
into consistent long-term performance of sol-gel PEG-coated fibers in SPME.
Sol-gel coatings have porous structures [333,353] resulting in an enhanced
surface area. This facilitates fast mass transfer and rapid establishment of the
extraction equilibrium, as is illustrated in Fig. 32.20 showing that the extraction
equilibrium on a sol-gel PEG coated fiber was reached within 30 min. This trans-
lates into the possibility for speedy analysis using sol-gel coated fibers or capil-
laries.
Jinno et al. [360] compared five different SPME coatings for the extraction of
benzodiazepines from human urine. The SPME coatings under investigation
were (a) sol-gel C poly(dimethylsiloxane), (b) polyacrylate, (c) Carbowax/
templated resin, (d) methyl-octyl poly(dimethylsiloxane), and (e) poly(dimethyl-
siloxane). The authors found that sol-gel Cl poly(dimethylsiloxane) coated fiber
provided the highest extraction efficiency, as is illustrated in Fig. 32.21. Both
PDMS and C,, (undecyl) moieties on the used sol-gel coating are nonpolar in
nature, while the extracted benzodiazepines are relatively polar compounds.
This higher extraction efficiency of the "nonpolar" sol-gel fiber for benzo-

1063
 300

250

200
x

2 150
,-A
1t00 *-B
'-C
50 -D
-E
0

Extraction Time (s)

Fig. 32.20. Extraction time profile for 10 ng/ml BTEX (constant stirring at room temperature).
(A) Benzene; (B) toluene; (C) ethylbenzene; (D)p-xylene; (E) o-xylene. GC conditions: 40°C for 2
min, then programmed at 7°C/min to 110°C, and then 20°C/min to 130°C for 1 min; injection
temperature, 280°C; FID, 300°C; desorption time 20 s. (Reproduced from Ref. [353] by
permission of Elsevier).

diazepines over other polar coatings (including polyacrylate and Carbowax/

templated resin coatings) may seem paradoxical. However, if one considers that
the sol-gel coating also contains polar moieties like silanol groups, this apparent
contradiction easily gets resolved. Thus, because of the presence of polar and
nonpolar sorption sites, sol-gel coatings are capable of extracting both polar and
nonpolar analytes.

Fig. 32.21. Extracted amounts

estimated by peak area counts in LC
chromatograms with various fiber
coatings. Concentration of each
analyte 0.1 /g/ml; extraction time, 60
min; extraction temperature, 60°C;
desorption time, 30 min (room temp-
erature); desorption solvent, mobile
phase (acetonitrile:5 mM phosphate
buffer, 40:60). (Reproduced from Ref:
I---i I I . . . I D- 7 Q. I
LObU] Dy permission or Koyal noclety HA LWIrM Sol-get UM
lU Us
of Chemistry). Fiber coatings

1064
TABLE 32.6
Names and chemical structures of sol-gel reagents used to prepared a sol-gel ODS monolithic bed
in a fused silica capillary. (Reproduced from Ref. [337] with permission of American Chemical
Society).
Reagent function and reagent name Reagent structure

1. Sol-gel coprecursor: tetramethoxysilane (TMOS) OCH3

H3CO-iOCH 3
OCH 3
2. Sol-gel coprecursor: N-octadecyl- dimenthyl CliCr 9 CH3
[3-(trimethoxy- silyl)propyl] ammonium chloride (C,,-TMS) C c2 CHr2 )3 - SiOC

C13] OCH3

3. Deactivation reagent: phenyldimethylsilane (PheDMS) i-H

4. Catalyst: trifluoroacetic acid (TFA) F3C-C-OH

Sol-gel monolithic beds for extraction

Preparationof Monolithic beds for Microextraction.A number of procedures has
been described for the preparation of sol-gel monolithic beds [337,338,340,
345,351,366-368]. The reported monolithic beds have been exclusively used in
separation columns for capillary electrochromatography or HPLC. Perhaps, the
simplest of the developed procedures for the preparation of sol-gel ODS mono-
lithic beds, is that described by Hayes and Malik [337]. Various ingredients used
to prepare the sol solution for the preparation of sol-gel ODS monolithic columns
are presented in Table 32.6. According to this method, a hydrothermally treated
fused silica capillary is filled with the sol solution that is allowed to stay inside
the capillary for an extended period (a few hours). During this time sol-gel
polymerization reactions proceed inside the capillary to transform the sol
solution into a porous solid matrix. With both ends sealed, the capillary is then
heated (from 30°C to the final temperature which may vary depending on the
monolith material) using a program rate of 0.2°C min ', and holding it at the
final temperature for 2 h. After this, the monolithic capillary is rinsed with a
series of appropriate solvents (e.g., methylene chloride, methanol and water).
Before use for extraction, the monolithic capillary is thermally conditioned (e.g.,
30-300°C, at 1C/min) under continuous purge with an inert gas. Scanning
electron micrographs representing the cross-sectional view of a fused silica
capillary with an in situ prepared porous sol-gel monolithic bed and its direct

1065
Fig. 32.22. Scanning electron micrographs of a sol-gel ODS monolithic capillary: (A)
cross-sectional view, magnification 1800x; (B) longitudinal view of the wall region showing
anchorage of the monolithic structure to the capillary inner surface, magnification 1000x.
(Reproduced from Ref. [337] by permission of American Chemical Society).

anchorage to the capillary inner walls are presented in Figs. 32.22A and B,
respectively.
Scheme 32.5 presents the chemical reactions involved in the preparation of
sol-gel ODS monolithic bed that can be used either in capillary electrochroma-
tography (CEC), capillary LC, or in capillary microextraction.
Advantages of sol-gel monoliths as extraction media. Like sol-gel coatings,
sol-gel monolithic beds are also chemically bonded to the silica substrates to
which they are applied (e.g., to the inner surface of a fused silica capillary) (Fig.
32.22B). This chemical bonding gives mechanical stability to the extraction bed.
Unlike particle-packed extraction beds (as in SPE cartridges or HPLC columns),
in monolithic beds there are no free-to-move particles in the bed. The bed rather
represents a single piece of porous solid anchored to the capillary inner walls
through chemical bonding.
Sol-gel monolithic beds are characterized by enhanced permeability. Sol-gel
reaction conditions can be easily optimized to obtain monolithic beds with an
order of magnitude (or more) higher permeability than the conventional parti-
cle-packed beds. The presence of flow-though channels makes the monolithic
bed an effective medium for fast extraction. Like sol-gel polymeric coatings,
sol-gel monolithic beds are also characterized by enhanced thermal and solvent
stability thanks to their direct chemical bonding to the capillary inner walls. As a
result, capillary microextraction with monolithic beds can be easily interfaced,
either with a liquid-phase or a gas-phase separation technique.
Sol-gel monolithic capillary contains a significantly higher amount of the
sol-gel extraction medium compared with that on an SPME fiber or an open
tubular microextraction capillary. This translates into higher sample capacity,
and consequently, higher sensitivity of sol-gel monolithic microextraction com-
pared with conventional or in-tube SPME (open tubular capillary microextrac-
tion). Figure 32.23 presents two gas chromatograms illustrating extraction

1066
Complete Hydrolysis of N-Octadecyldimethyl[3-(trihydroxysilyl)propyl]ammonium Chloride
CH HCH o
CH(CH2 ) . t )Si-oCH
(C11 2 ?cH3 + 3 H20 c C(CH
o .
3 2)ir I (CH) 3 -Si-SI-OH + 3 CH 3OH
CH 3 OCH 3 CH3 OH

Condensation of Tetrahydroxysilane with

IV-Octadecyldimethyl[3-(trihydroxysilyl)propyllammonium Chloride

MCCH- OH H H3 ,

CH3 OH Ca, 0 0

Bonding of the Sol-Gel ODS Coating Deactivation of the Sol-Gel

with the Fused-Silica Surface Mediated Fused-Silica Coated Surface with
A. Phenyldimethylsilane (PheDMS)
(~--
+ i? _..- cH
Si-OH + i-(CHz). i (Hzt)fCI 2
.HO- I HJ
H
CH3I 3 C--CH3
OH CH3
0si-o-si- o-
OH
Si-(CH2)N-(CH)fCH
O C113
+ a
CH,
OH
-. O Si-- i-(CH 2"N(CH2).CH
OH H3
H3 C-I,-CH3
B.
&
-2 TH + lc
Hs$H CH3
+ O-SI-O-Si--(CH2 ) N(CH2 ),7-CH3
0 OH CH3
.,d
$-d
- 2°Si-O-S-0-Si(CH~t:)yCH )TCHa

OH O l13

Scheme 32.5. Chemical reactions involved in the preparation of sol-gel ODS stationary phase.
(Adapted from Ref. [336] with permission of American Chemical Society).

sensitivities of two sol-gel microextraction capillaries with (A) a sol-gel ODS

monolithic bed and (B) a sol-gel ODS coating.
An aqueous sample containing a PAH (fluorene) and an alcohol (hexa-
decanol) was extracted under identical conditions using these two sol-gel capil-
laries. These results suggest that the sample capacity (and hence extraction
sensitivity) of the used sol-gel monolithic capillary is at least two orders of
magnitude higher than that of the capillary with the surface-bonded sol-gel
coating.

32.6.3.4 Sol-gel dendrimer-basedextraction media for SPME

Dendrimers constitute a unique class of molecular species that possess highly
branched, tree-like architecture [369]. Thanks to their branched architecture,
dendrimers have outstanding material properties which have been exploited in a
wide range of scientific disciplines [370-374]. Because of this unique molecular

1067
 1nnn
(B)
900 900 Open Tubular

800 800 -

700 700 -

o 600 600 -

M. 500 500 -

E
400 400 -

300 300 -

200 200 - o
O
100 100- : x
LL I

0 0-
0 2I I
I 10 20 30 min

Fig. 32.23. Comparison of the extraction sensitivity in capillary microextraction using: (A) a
sol-gel ODS monolithic capillary (50 Am i.d.) and (B) a sol-gel ODS-coated fused silica capillary
(250 Am i.d., -1 gm coating thickness). (From Ref. [368]).

architecture, dendrimer-based materials also have great potential for being used
in analytical chemistry (e.g., as effective extraction media, as selective stationary
phases or mobile phase additives for various chromatographic and electro-
migration separation techniques). However, only on rare occasions have these
analytical potentials of dendrimers been utilized [70,375-380]. Recently,
Newkome et al. [381] described a method for the synthesis of phenyl-terminated
dendrimers with a trimethylsilyl derivatized sol-gel-active root (Fig. 32.24).
Using sol-gel coating technology [232], the authors chemically bonded these den-
dritic moieties to the fused silica capillary inner surface in the form of a station-
ary phase coating. These sol-gel dendrimer-coated capillaries were further used
as GC columns that showed unique selectivity. In an ongoing project, these
authors have made use of these sol-gel dendrimer-coated capillaries to perform
microextraction [382].
Figure 32.25 shows a gas chromatogram of polycyclic aromatic hydrocarbons
that were extracted from an aqueous matrix using a sol-gel dendrimer-coated
capillary. In principle, a wide variety of dendrimers can be prepared and easily
bonded to silica particles for use as solid-phase extraction media or HPLC sta-
tionary phase. It should also be possible to prepared sol-gel monolithic beds with
dendritic ligands. Such monolithic beds can also be used as extraction media or

1068
 - f)
(EtO) 3SiCH 2CH 2CH 2NH

(EtO) 3SiCH 2 CH 2CH 2N

Fig. 32.24. Chemical structures of phenyl-terminated dendrimers with a triethoxy-derivatized

sol-gel active root. (A) First generation dendrimer; (B) second generation dendrimer; and (C)
third generation dendrimer. (Adapted from Ref. [381] with permission of Elsevier).

as stationary phases for liquid- or gas-phase separation techniques. Because of

extraordinary selectivity and advanced material properties offered by dendri-
mers, a significant growth in research activities directed towards analytical
applications of these uniquely designed molecules can be expected in the near
future.

32.7 CONCLUSION

The use of polymeric extraction materials with advanced material properties is a

growing trend in analytical extraction, sample cleanup, and preconcentration.
Polymeric materials are being used in a wide variety of extraction formats,
including solution-mediated extraction, solid-phase extraction, and solid-phase
microextraction. A wide range of pH stability is an attribute of many polymers,
and this inherent property of polymeric materials is being exploited to solve ana-
lytical extraction problems that involve operations under extreme pH conditions
where silica-based extraction materials fail to provide stable performance. Sig-

1069
Fig. 32.25. CME-GC analysis of PAHs using a 4
I
sol-gel dendrimer-coated fused silica extraction
capillary. Extraction conditions: gravity-fed
extraction, 30 min; post extraction purge with
helium, 2 min. GC conditions: sol-gel PDMS
capillary column (10 m x 0.25 mm i.d.); desorp-
tion temperature, from 30 to 300°C; column
temperature programming, from 30°C (5 min) at
'IAs'
15°C/ min; detector, FID (300°C). (From Ref. 5 10 15 20 25 30
[382]). Minutes

nificant attention is being paid to the development of molecularly imprinted

polymers, temperature-responsive materials, restricted-access materials, den-
dritic materials, conductive polymers, and sol-gel polymeric materials. Although
molecularly imprinted polymers are currently finding a wide range of applica-
tions in solid-phase extraction, they have just been introduced in the solid-phase
microextraction area. Molecularly imprinted polymers with restricted-access
properties show great promise in selective preconcentration, cleanup, and auto-
mated analysis of biological samples. Sol-gel polymeric materials constitute
another group of advanced material systems that have great interest for analyti-
cal extraction technology. Thanks to the organic-inorganic hybrid nature of
sol-gel materials, they often provide selectivity that is difficult to achieve by
using either purely inorganic or purely organic materials. Moreover, this selec-
tivity can be easily fine-tuned by changing the proportions of the organic/inor-
ganic ingredients of the sol solution used to create these materials. Sol-gel
reactions can take place under extraordinarily mild thermal conditions (nor-
mally at room temperature) that allow their application on various substrates
without requiring the use of high temperature. Sol-gel polymeric materials get
chemically bonded to the silica substrates they are applied to. This chemical
bonding is responsible for high thermal and solvent stability of sol-gel materials
that can be effectively utilized in analytical extractions. It can be assumed with
confidence that future progress in analytical extraction technology will be
closely related to the development of newer polymeric materials with advanced
material properties.

1070
REFERENCES

1 C.W. Huck, and G.K. Bonn, J. Chromatogr.A, 885 (2000) 51.

2 N.C. van de Merbel, J. Chromatogr.A, 856 (1999) 55.
3 R.E. Majors, in: Settle (Ed.), Handbook of Instrumental Techniques for Analytical
Chemistry. Prentice Hall, Upper Saddle River, NJ, USA, 1997, Ch. 2, pp. 17-54.
4 F. Soxhlet, Dingers Polytech J., 232 (1879) 461.
5 Z. Penton, Advances in Chromatography.Marcel Dekker: New York, 1997.
6 M. Stoepppler, Sampling and Sample Preparation for Analytical Chemists.
Springer-Verlag, Berlin, 1997.
7 T.H. Lee and M.I. Aguilar, Adv. Chromatogr., 41 (2001) 175.
8 R.E. Majors, LC-GC, 13(1995) 542.
9 J.D. Thomas, Trends Anal. Chem., 14 (1995) 186.
10 D. Noble, Anal. Chem., 65 (1993) 693A.
11 Chem. Eng. News, 5 (1992).
12 E.M. Thurman and M.S. Mills, Solid-Phase Extraction: Principles and Practice.
Wiley, New York, 1998.
13 S.B. Hawthorne, Anal. Chem., 62 (1990) 633A.
14 M. McHugh and V. Krukonis, Supercritical Fluid Extraction. Principles and
Practice.Butterworths, Stoneham, MA, 1986.
15 L.T. Taylor, Supercritical Fluid Extraction. Wiley, New York, 1996.
16 S.A. Westwood, Supercritical Fluid Extraction and Its Use in Chromatographic
Sample Preparation.CRC Press, Boca Raton, FL, USA, 1993.
17 M.L. Lee and K.E. Markides, Analytical Supercritical Fluid Chromatography and
Extraction. Chromatography Conferences, Provo, UT, USA, 1991.
18 B.E. Richter, B.A. Jones, J.L. Ezzell, N.L. Porter, N. Avdalovic and C. Pohl, Anal.
Chem., 68 (1996) 1033.
19 0. Zuloaga, N. Etxebarria, L.A. Fernandez and J.M. Madariaga, TrendsAnal. Chem.,
17 (1999) 642.
20 A. Abu-Samra, J.S. Morris and S.R. Koirtyohann, Anal.Chem., 47 (1975) 1475.
21 A. Zlotorzynski, Crit. Rev. Anal. Chem., 25 (1995) 43.
22 H.M. Kingston and S.J. Haswell, Microwave-Enhanced Chemistry. fundamentals,
Sample preparationsand Applications. ACS, Washington, DC, 1997.
23 Q. Jin, F. Liang, H. Zhang, L. Zhao, Y. Huan and D. Song, Trends Anal Chem., 18
(1999) 479.
24 R.C. Richter, D. Link and H.M. Kingston, Anal. Chem., 73 (2001) 31A.
25 D. Lorian and G. Knapp, Anal. Chem., 73 (2001) 1515.
26 K.J. Hageman, L. Mazeas, C.B. Grabanski, D.J. Miller and S.B. Hawthorne, Anal.
Chem., 68 (1996) 3892.
27 Y. Yang and B. Li, Anal. Chem., 71 (1999) 1491.
28 S.B. Hawthorne, S. Trembley, C.L. Moniot, C.B. Grabanski and D.J. Miller, J.
Chromatogr.A, 886 (2000) 237.
29 B.O. Keller and L. Li, Anal. Chem., 73 (2001) 2929.
30 E.K. Rasmussen, K.M. Pedersen-Bjergaard, H.G. Ugland and T. Gronhaug, J.
Chromatogr.A, 873 (2000) 3.
31 L. Zhao and H.K. Lee, J. Chromatogr.A, 919 (2001) 381.
32 T.G. Halvorsen, S. Pedersen-Bjergaard and K.E. Rasmussen, J. Chromatogr.A, 909
(2001) 87.
33 L. Zhu, C. Tu and H.K. Lee, Anal. Chem., 73 (2001) 5655.
34 S. Pederson-Bjergaard and K.E. Rusmussen, Anal. Chem., 71 (1999) 1650.
35 W. Liu and H.K. Lee, Anal. Chem., 72 (2000) 4462.
36 Y. He and H.K. Lee, Anal. Chem., 69 (1997) 4634.

1071
37 M. Ma and F.F. Cantwell, Anal. Chemn., 71 (1999) 388.
38 H. Liu and P.K. Dasgupta, Trends Anal. Chem., 15 (1996) 468.
39 A.L. Theis, A.J. Waldack, S.M. Hansen and M.A. Jeannot, Anal. Chem., 73 (2001)
5651.
40 G.A. Audunsson, Anal. Chem., 58 (1986) 2714.
41 R.A. Bartsch and J.D. Way, Chemical Separation with Liquid Membrane. ACS,
Washington, D.C., 1996.
42 J.A. Jonsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 318.
43 J.A. Jonsson and L. Mathiasson, Trends Anal. Chem., 18 (1999) 325.
44 J.A. Jonsson and L. Mathiasson, Adv. Chromatogr., 41 (2001) 53.
45 D. Kou, A.S. Juan and S. Mitra, Anal. Chem., 73 (2001) 5462.
46 M.C. Hennion, J. Chromatogr. A, 856 (1999) 3.
47 F. Lanza and B. Sellergren, Adv. Chromatogr.(2001) 41, 53.
48 N. Masque, R.M. Marce and F. Borrull, Trends .Anal. Chem., 20 (2001) 477.
49 M.E. Leon-Gonzales and L.V. Perez-Arribas, J. Chromatogr. A, 902 (2000) 3.
50 J. Pawliszyn, Solid-phase Microextraction. Theory and Practice. Wiley, New York,
(1997.
51 J. Pawliszyn (Ed.), Applications of Solid-Phase Microextraction. Royal Society of
Chemistry, Cambridge, UK, 1999.
52 M. de Fatima Alpendurada, J. Chromatogr.A, 889 (2000) 3.
53 K.G. Furton, J. Wang, Y.L. Hsu, J. Walton and J.R. Almirall, J. Chromatogr.Sci., 38
(2000) 297.
54 N.H. Snow, J. Chromatogr.A, 885 (2000) 445.
55 G. Theodoridis, E.H. Koster and G.J. de Jong, J. Chromatogr. B, 745 (2000) 49.
56 G.A. Mills and V. Walker, J. Chromatogr.A, 902 (2000) 267.
57 A.D. Harmon, Food. Sci. Technol., 79 (1997) 81.
58 Y. Ikada, Biomaterials, 15 (1994) 725.
59 R.P. Belardi and J. Pawliszyn, Water Pollut. Res. J. Canada, 24 (1989) 179.
60 M.R. Buchmeiser, Chem. Rev., 100 (2000) 1565.
61 T. Sekine and Y. Hasegawa, Solvent Extraction Chemistry. Fundamentals and
Applications, Marcel Dekker, New York, (1997.
62 J.G. Huddleston, H.D. Willauer, S.T. Griffin and R.D. Rogers, Ind. Eng. Chem. Res.,
38 (1999) 2523.
63 H. Ishii, M. Junichiro and H. Watanabe, Bunseki Kagaku, 26 (1977) 252.
64 W.L. Hinze and E. Pramauro, Crit. Rev. Anal. Chem., 24 (1993) 133.
65 H. Tani, T. Kamidate and H. Watanabe, J. Chromatogr. A, 780 (1997) 229.
66 J.F. Scamehorn and J.H. Harwell, Surfactants in Chemical/Process Engineering:
Surfactant Science Series. Marcel Dekker, New York, 1988.
67 J.G. Huddleston, A. Veide, K. Kohler, J. Flanagan, S.O. Enfors and A. Lyddiatt,
Trends Biotechnol., 9 (1991) 381.
68 H.O. Johansson, G. Karlstrom, B. Mattiasson and F. Tjerneld, Bioseparation, 5
(1995) 269.
69 E.L. Goetheer, M.W. Baars, L.J. van den Broeke, E.W. Meijer and J.T. Keurentjes,
Ind. Eng. Chem. Res., 39 (2000) 4634.
70 M.W. Baars, P.E. Froehling and E.W. Meijer, Chem. Commun., 1997 (20) (1997)
1959.
71 K.R. Powell, T.M. McCleskey, W. Tumas and J.M. DeSimone, Ind. Eng. Chem. Res.,
40 (2001) 1301.
72 H. Kanazawa and Y. Matsushima, Adv. Chromatogr., 41 (2001) 311.
73 T. Okano, Y.H. Bae and S. Kim, in: J. Kost (Ed.), Pulsed and Self-Regulated Drug
Delivery. CRC Press, Boca Raton, FL, 1990, p. 17.
74 N. Ogata, Polym. Prepr., 37 (1996) 113.

1072
 75 C. Ganorkar, F. Liu, M. Baudys, et al. Polym. Prepr., 38 (1997) 560.
76 N.Y. Rapoport, W.G. Pitr and J. Pruitr, Polym. Prepr., 41 (2000) 716.
77 J.E. Chung, M. Yamato, M. Yokoyama, et al. Polym. Prepr., 39 (2000) 178.
78 N. Kubota, T. Matsubara and Y. Eguchi, J. Appl. Polym. Sci., 70 (1998) 1027.
79 T. Peng and Y.L. Cheng, J. Appl. Polym. Sci., 70 (1998) 2133.
80 Y.G. Takei, T. Aoki, K. Sanui, N. Ogata, T. Okano and Y. Sakurai, Bioconjugate
Chem., 4 (1993) 341.
81 Y.G. Takei, T. Aoki, K. Sanui, et al. Bioconjugate Chem., 5 (1994) 577.
82 T. Okano, N. Yamada, H. Sakai and Y. Sakurai, J. Biomed. Mater. Res., 27 (1993)
1243.
83 T. Takezawa, Y. Mori, T. Yonaha and K. Yoshizato, Exp. Cell Res., 208 (1993) 430.
84 K. Hosoya, K. Kimata, T. Araki, N. Tanaka and J.M. Frechet, Anal. Chem., 67 (1995)
1907.
85 H. Kanazawa, Y. Kashiwase and K. Yamamoto, Anal. Chem., 69 (1997) 823.
86 H. Kanazawa, Y. Matsushima and T. Okano, Trends Anal. Chem., 17 (1998) 435.
87 H. Lakhhirai, T. Okano, N. Nurdin, et al. Biochim. Biophys. Acta, 1379 (1998) 303.
88 H. Kanazawa, T. Sunamoto, Y. Matsushima, A. Kikuchi and T. Okano, Anal. Chem.,
72 (2000) 5961.
89 J. Kobayashi, A. Kikuchi, K. Sakai and T. Okano, Anal. Chem., 73 (2001) 2027.
90 C. Matsubara, S. Izumi, K. Takamura, S. Yoshioka and Y. Mori, Analyst, 118 (1993)
553.
91 C. Matsubara, N. Kikuchi and K. Takamura, Bunseki Kagaku, 44 (1995) 311.
92 M. Matsukata, T. Aoki, K. Sanui, N. Ogata, A. Kikuchi, Y. Sakurai and T. Okano,
Bioconjugate Chem., 7 (1996) 96.
93 T. Saitoh, T. Ohyama, T. Sakurai, T. Kaise and C. Matsubara,Anal. Sci., 13 (1997) 1.
94 T. Saitoh, T. Sakurai, T. Kaise and C. Matsubara, Anal. Sci., 13 (1997) 181.
95 T. Saitoh, T. Ohyama, K. Takamura, T. Sakurai, T. Kaise and C. Matsubara,
Talanta, 46 (1998) 541.
96 T. Saitoh, M. Haga, T. Sakurai, T. Kaise and C. Matsubara, Anal. Sci., 14 (1998) 929.
97 T. Saitoh, Y. Yoshida, T. Matsudo, S. Fujiwara, D. Akira, K. Iwaki, Y. Suzuki and C.
Matsubara, Anal. Chem., 71 (1999) 4506.
98 T. Saitoh, Y. Yoshida and C. Matsubara, J. Chromatogr.A, 891 (2000) 69.
99 T. Ohyama, K. Arai, T. Nakagawa, C. Matsubara and K. Takamura, Bunseki
Kagaku, 46 (1997) 59.
100 G.C. Chen and A.S. Hoffman, J. Biomater. Sci. Polym. Edn., 5 (1994) 371.
101 C. Cole, S.M. Schreiner, J.H. Priest, N. Monji and A.S. Hoffman, ACS Symposium
Series 350. ACS, Washington, DC, 1987, p. 245.
102 M. Heskins, J.E. Guillet and E. James, J. Macromol. Sci. Chem. A, 2 (1968) 1441.
103 M.C. Hennion and V. Pichon, Environ. Sci. Technol., 28 (1994) 576A.
104 M.C. Hennion, J. Chromatogr.A, 885 (2000) 73.
105 E. Matisova and S. Skrabakova, J. Chromatogr.A, 707 (1995) 145.
106 R.K. Iler, The Chemistry of Silica. Wiley, New York, 1979.
107 M.L. Lee, F.J. Yang and K.D. Bartle, Opem Tubular Column Gas Chromatograhy.
Wiley, Ner York, 1984.
108 M.L. Mayer, S.K. Poole and C.F. Poole, J. Chromatogr.A, 697 (1995) 89.
109 K.K. Unger, Packings and stationary phases in Chromatographic Techniques.
Marcel Dekker, NY, 1990.
110 S.O. Akapo and C.F. Simpson, J. Chromatogr.Sci., 28 (1990) 186.
111 M. Ashraf-Khorassani, L.T. Taylor and R.A. Henry, Anal. Chem., 60 (1988) 1529.
112 K.D. Altria, N.W. Smith and C.H. Turnbull, Chromatographia,46, (1997) 664.
113 M. Pursch and L.C. Sander, J. Chromatogr.A, 887 (2000) 313.
114 V.G. Berezkin, A. Malik and V.S. Gavrichev, J. High Resolut. Chromatogr./

1073
 Chromatogr. Commun., 6 (1983) 388.
115 A. Malik, V.G. Berezkin and V.S. Gavrichev, Chromatographia,22 (1986) 117.
116 T. Takeuchi, M. Watanabe, T. Hamanaka, H. Haraguchi and D. Ishii, J. High
Resolut. Chromatogr., 13, (1990) 606.
117 J. Zhao and P.W. Carr, Anal. Chem., 72 (2000) 4413.
118 B. Yan, J. Zhao, J.S. Brown, J. Blackwell and P.W. Carr, Anal. Chem., 72 (2000) 1253.
119 J. Nawrocki, M.P. Rigney, A. McCormick and P.W. Carr, J. Chromatogr., 657 (1993)
229.
120 Y. Hu and P.W. Carr, Anal. Chem., 70 (1998) 1934.
121 J. Yu and Z. El Rassi, J. Chromatogr., 631 (1993) 91.
122 J. Yu and Z. El Rassi, J. High. Resolut. Chromatogr., 17 (1994) 705.
123 Z.T. Jiang and Y.M. Zuo, Anal. Chem., 73 (2001) 686.
124 M. Kawahara, H. Nakamura and T. Nakajima, J. Chromatogr., 515 (1990) 149.
125 A. Kurganov, U. Trudinger, T. Isaeva and K. Unger, Chromatographia,42 (1996)
217.
126 K. Tani and Y. Suzuki, J. Liq. Chromatogr. &. Rel. Technol., 19 (1996) 3037.
127 J. Winkler and S. Marme, J. Chromatogr. A, 888 (2000) 51.
128 C.H. Cars, L.M. van Elven, A.M. van Herk and A.L. German, Br. Polym. J., 21 (1989)
133.
129 H. Engelhardt and W.M. Cunat, Chromatographia,40 (1995) 657.
130 H. Figge, A. Deege, J. Kohler and G. Schomburg, J. Chromatogr., 351 (1986) 393.
131 Y. Shen, X. Shao, K. O'Neill, J.S. Bradshaw and M.L. Lee, J. Chromatogr. A, 866
(2000) 1.
132 G. Zhang and B. Xu, J. Chromatogr.A, 912 (2001) 335.
133 J.J. Kirkland, J.W. Henderson, J.J. Destefano, M.A. Vanstraten and H.A. Claessens,
J. Chromatogr.A, 762 (1997) 97.
134 A.V. Kiselev and J.P. Yashin, Gas Adsorption Chromatography.Plenum Publishing,
New York, 1969.
135 I. Halasz and C. Horvath, Anal. Chem., 34 (1969) 409.
136 J.H. Knox and P. Ross, Advances in Chromatography, Volume 37. Marcel Dekker,
New York, 1997.
137 F.D. Shire and B. McEnaney, Energeia, 2 (1991) 1.
138 J. Shurrock, M. Bauer, M. Holmes and A. Rachwal, Disinfection. By. Products. in.
Drinking. Water. Current.Issues. The Royal Society of Chemistry, Cambridge, UK,
(1999.
139 A. Di Corcia and R. Samperi, Anal. Chem., 62 (1990) 1490.
140 A. Di Corcia, S. Marchese and R. Samperi, J. Chromatogr., 642 (1993) 163.
141 B. Altenbach and W. Giger, Anal. Chem., 67 (1995) 2325.
142 C. Crescenzi, A. Di Corcia, G. Passariello, M.I. Turnes Carou and R. Samperi, J.
Chromatogr.A, 733 (1996) 41.
143 F. Mangani and R. Cenciarini, Chromatographia,41 (1995) 678.
144 D. Puig and D. Barcelo, J. Chromatogr. A, 733 (1996) 371.
145 J. Slobodnik, O. Oztezkizan, H. Lingeman and U.A. Brinkman, J. Chromatogr. A,
750 (1996) 227.
146 I.A. Dolova, A.V. Kiselev and Y.I. Yashin, Zh. Fiz. Khim., 48 (1974) 1285.
147 S.P. Nandi and P.L. Walker, Sep. Sci., 11 (1976) 441.
148 M.T. Gilbert, J.H. Knox and B. Kaur, Chromatographia,16 (1982) 138.
149 J.H. Knox, K.K. Unger and H. Mueller, J. Liq. Chromatogr., 6 (1983) 1.
150 K.K. Unger, Anal. Chem., 55 (1983) 361A.
151 E. Skuchtanova, L. Feltl and E. Smolkova-Keulemansova, J. Chromatogr., 292
(1984) 233.
152 J.H. Knox, B. Kaur and G.R. Millward, J. Chromatogr., 352 (1986) 3.

1074
153 0. Chiantore, I. Novak and D. Berek, Anal. Chem., 60 (1988) 638.
154 B.J. Bassler and R.A. Hartwick, J. Chromatogr. Sci., 27 (1989) 162.
155 D. Berek and I. Novak, Chromatographia,30 (1990) 582.
156 V.G. Berezkin and Y.S. Drugov, Gas Chromatography in Air Pollution Analysis.
Elsevier, Amsterdam, (1991.
157 Annual Book of ASTM Standards. Philadelphia, USA, 1991.
158 E. Atlas and S. Schauffler, Environ. Sci. Technol., 25 (1991) 61.
159 E. Atlas, S. Schauffler, J.T. Merril, C.J. Hahn, B. Ridley, J. Walega, J. Greenberg, L.
Heidt and P. Zimmerman, J. Geophys. Res., 97 (1992) 331.
160 K. Grob, J. Chromatogr., 321 (1985) 45.
161 B.V. Burger and Z. Munro, J. Chromatogr., 370 (1986) 449.
162 B.V. Burger and Z. Munro, J. Chromatogr., 394 (1987) 443.
163 C. Pierce, R.N. Smith, J.W. Wiley and H. Cordes, J.Am. Chem. Soc., 75 (1951) 4551.
164 W. Engewald, J. Porschmann and T. Welsh, Chromatographia,30 (1990) 537.
165 A. Di Corcia and A. Liberti, Adv. Chromatogr., 14 (1976) 305.
166 J.D. Bernal, Proc. Royal Soc. London, Ser. A, 100 (1924) 749.
167 H. Lipson and A.R. Stokes, Proc. Royal Soc. London, Se. A, 181 (1942) 93.
168 B.E. Warren, Phys. Rev., 59 (1941) 693.
169 N. Masque, R.M. Marce and F. Borrull, J. Chromatogr.A, 793 (1998) 257.
170 A. Lagana, G. D'Ascenzo, G. Fago and A. Marino, Chromatographia,46 (1997) 256.
171 R. Curini, A. Gentili, S. Marchese, et al. J. Chromatogr.A, 874 (2000) 187.
172 T. Potter, L. Martin, S. Belflower, et al. J. Agric. Food Chem., 48 (2000) 4103.
173 H. Sabnik and R. Jeannot, J. Chromatogr.A, 879 (2000) 73.
174 E. Majzik-Solymos, E. Visi, G. Karoly, et al. J. Chromatogr. Sci., 39 (2001) 325.
175 F.J. Schenk and D.J. Donoghue, J. Agric. Food Chem., 48 (2001) 6412.
176 K.N. Norman and S.H. Panton, J. Chromatogr.A, 907 (2001) 247.
177 C. Bartoni, R. Curini and G. D'Ascenzo, Environ. Sci. Technol., 34 (2000) 5059.
178 J. Tolls, M. Heller and D. Sijm, Anal. Chem., 71 (1999) 5242.
179 Q.C. Wang, W.K. Hosoya, F. Svec and J.M.J. Frechet, Anal. Chem. (1992) 64, 1232.
180 E. Baltussen, F. David, P. Sandra, H. Janssen and C. Cramers, J. High Resol.
Chromatogr., 21 (1998) 645.
181 B. Tienpont, F. David, C. Bicchi and P. Sandra, J. Micro Sep., 12 (2000) 577.
182 J.J. Sun and J.S. Fritz, J. Chromatogr., 522 (1990) 95.
183 J.J. Sun and J.S. Fritz, J. Chromatogr., 590 (1992) 197.
184 N. Masque, M. Galia, R.M. Marce and F. Borrull, J. Chromatogr.A, 771 (1997) 55.
185 P.J. Dumont and J.S. Fritz, J. Chromatogr.A, 691 (1995) 123.
186 N. Masque, M. Galia, R.M. Marce and F. Borrull, Analyst, 122, (1997) 425.
187 N. Masque, M. Galia, R.M. Marce and F. Borrull, J. Chromatogr. A, 803 (1998) 147.
188 (a) N. Masque, M. Galia, R.M. Marce and F. Borrull, Chromatographia,50 (1999) 21.,
(b) J. High Resolut. Chromatogr., 22 (1999) 547
189 A.D. Alder, F.R. Longo and J.D. Finerelli, J. Org. Chem. (1967) 32, 476.
190 D.G. Kim, M.W. Jung, I.R. Paeng, J.S. Rhee and K.J. Paeng, Microchem. J., 63 (1999)
134.
191 A. Mayes and K. Mosbach, Trends Anal. Chem., 16 (1997) 321.
192 L. Schewitz, L.I. Anderson and S. Nilsson, J. Chromatogr. A, 817 (1998) 5.
193 P.T. Vallano and V.T. Remcho, J. Chromatogr.A, 887 (2000) 125.
194 E. Reid, H.M. Hill and I.D. Wilson, Drug Development Assay Approaches, Including
Molecular Imprinting and Biomarkers. Methodol. Surv. Bioanal. Drugs 25. Royal
Society of Chemistry, Cambridge, UK, 1998.
195 S. Pleasance and R.A. Biddlecombe, DrugDevelopment Assay Approaches, Including
Molecular Imprinting and Biomarkers. Methodol. Surv. Bioanal. Drugs 25. Royal
Society of Chemistry, Cambridge, UK, (1998.

1075
196 R.A. Bartsch and M. Maeda, Molecular and Ionic Recognition with Imprinted
Polymers. ACS Symposium Series 703. ACS, Washington, D.C., (1998.
197 D. Kriz and K. Mosbach, Anal. Chim. Acta, 300 (1997) 71.
198 D. Kriz, O. Ramstrom and K. Mosbach, Anal. Chem., 69 (1997) 345A.
199 C. Berggren, S. Bayoudh, D. Sherrington and K. Ensing, J. Chromatogr. A, 889
(2000) 105.
200 B. Bjarnason, L. Chimuka and 0. Ramstrom, Anal. Chem., 91 (1999) 2152.
201 C. Crescenzi, S. Bayoudh, P.A. Cormack, T. Klein and K. Ensing, Anal. Chem., 73
(2001) 2171.
202 R. Koeber, C. Fleischer, F. Lanza, K.S. Boos, B. Sellergren and D. Bercello, Anal.
Chem. (2001) 73, 2437.
203 N. Masque, R.M. Marce, F. Borrull, P.A. Cormack and D.C. Sherington, Anal. Chem.,
72 (2000) 4122.
204 M.T. Muldoon and L.H. Stanker, Anal. Chem., 69 (1997) 803.
205 W.M. Mullett and E.P. Lai, Anal. Chem., 70 (1998) 3636.
206 A. Zander, P. Findlay, T. Renner and B. Sellergren, Anal. Chem., 70 (1998) 3304.
207 E.H. Koster, C. Crescenzi, W. den Hoedt, K. Ensing and G.J. de Jong, Anal. Chem., 73
(2001) 3140.
208 W.M. Mullett, P. Martin and J. Pawliszyn, Anal. Chem., 73 (2001) 2383.
209 C. Baggiani, G. Giraudi, C. Giovannoli, F. Trotta and A. Vanni, J. Chromatogr.A, 883
(2000) 119.
210 Y. Chen, M. Kele, S.B. Sajonz and G. Guiochon, Anal. Chem., 71 (1999) 928.
211 S. Vydyasankar, M. Ru and F.H. Arnold, J. Chromatogr.A, 775 (1997) 51.
212 K. Yoshizako, K. Hosoya, Y. Iwakoshi, K. Kimata and N. Tanaka, Anal. Chem., 70
(1998) 386.
213 B. Sellergren, Trends Anal. Chem., 18 (1999) 164.
214 E. Schoenzetter, V. Pichon, D. Thiebaut, A. Fernandez-Alba and M.C. Hennion, J.
Micro Sep., 12 (2000) 316.
215 J. Haginaka and H. Sanbe, Anal. Chem., 72 (2000) 5206.
216 D.J. Anderson, Anal. Chem., 65 (1993) 434R.
217 I.H. Hagestam and T.C. Pinkerton, Anal. Chem., 57 (1985) 1757.
218 C.P. Desilets, M.A. Rounds and F.E. Regnier, J. Chromatogr., 544 (1991) 25.
219 L.J. Glunz, J.A. Perry, B. Invergo, H. Wagner, T.J. Szczerba, J.D. Reteike and P.W.
Gunz, J. Liq. Chromatogr., 15 (1992) 1361.
220 D.J. Gisch, B.T. Hunter and B. Feibush, J. Chromatogr., 433 (1988) 264.
221 J. Haginaka and J. Wakai, Anal. Chem., 62 (1990) 997.
222 J. Haginaka and J. Wakai, Chromatographia,29 (1990) 223.
223 T.C. Pinkerton, T.D. Miller, S.E. Cook, J.A. Perry, J.D. Rateike and T.J. Szczerba,
Biochromatography, 1 (1986) 96.
224 T.C. Pinkerton, Am. Lab., 20 (1888) 70.
225 T. Nakagawa, A. Shibukawa, N. Shimono, T. Kawashima, H. Tanaka and J.
Haginaka, J. Chromatogr., 420 (1987) 297.
226 J.A. Perry, B. Invergo, H. Wagner and T.J. Szczerba, J. Liq. Chromatogr., 15 (1992)
3343.
227 J. Haginaka, H. Yasuda and Y. Kimura, Anal. Chem., 61 (1989) 2445.
228 J. Haginaka, J. Wakai, N. Yasuda, H. Yasuda and Y. Kimura, J. Chromatogr., 515
(1990) 59.
229 S.E. Cook and T.C. Pinkerton, J. Chromatogr., 368 (1986) 233.
230 J.A. Perry, J. Liq. Chromatogr., 13 (1990) 1047.
231 J. Haginaka and J. Wakai, Anal. Sci., 8 (1992) 141.
232 J. Haginaka and J. Wakai, J. Chromatogr., 596 (1992) 151.
233 J. Haginaka and J. Wakai, Anal. Sci., 8 (1992) 137.

1076
234 J. Haginaka, H. Sanbe and H. Takehira, J. Chromatogr.A, 857 (1999) 117.
235 J. Haginaka, H. Takehira, K. Hosoya and N. Tanaka, J. Chromatogr. A, 849 (1999)
331.
236 J. Haginaka and H. Sanbe, J. Chromatogr. A, 913 (2001) 141.
237 K.E. Gustavson, W.M. DeVita, A. Revis and J.M. Harkin, J. Chromatogr. A, 883
(2000) 143.
238 C. Schafer and D. Lubda, J. Chromatogr.A, 909 (2001) 73.
239 K.S. Boos and C.H. Grimm, Trends Anal. Chem., 18 (1999) 175.
240 K. Racaityte, E.S. Lutz, K.K. Unger, D. Lubda and K.S. Boos, J. Chromatogr.A, 890
(2000) 135.
241 R.A. van der Hoeven, A.J. Hofte, M. Frenay, H. Ikrth, U.R. Tjaden, J. van der Greef,
A. Rudolphi, K.S. Boos, G. Marko Varga and L.E. Edholm, J. Chromatogr.A, 762
(1997) 193.
242 C. Brinker and G. Scherer, Sol-Gel Science. Academic pPess, Boston, MA, 1990.
243 S. Acosta, P. Arnal, R.J. Corriu and D. LeClercq, in: A.K. Cheetham (Ed.), Better
Ceramics Through Chemistry VI (Mat. Res. Soc. Symp. Proc.). Materials Research
43
Society, Pittsburgh, 1994 p.
244 M. Andrianainarivelo, R.J. Corriu, D. Leclerq, P.H. Mutin and A. Vioux, J. Sol Gel
Sci. Technol., 8 (1997) 89.
245 G. Guerrero, P.H. Mutin and A. Vioux, Chem. Mater., 12 (2000) 1268.
246 J.N. Hay and H.M. Raval, Chem. Mater., 13 (2001) 3396.
247 P. Innocenzi, A. Sassi, G. Brusatin, M. Guglielmi, D. Favretto, R. Bertani, A. Venzo
and F. Bobanneau, Chem. Mater., 13 (2001) 3635.
248 J. Zuhlke, D. Knopp and R. Niessner, Fresen. J. Anal. Chem., 352 (1995) 654.
249 M. Cichna, P. Markl, D. Knopp and R. Niessner, Chem. Mater., 9 (1997) 2640.
250 B. Spitzer, M. Cichna, P. Markl, G. Sontag, D. Knopp and R. Niessner, J.
Chromatogr. A, 880 (2000) 113.
251 M. Cichna, P. Markl, D. Knopp and R. Niessner, J. Chromatogr.A, 919 (2001) 51.
252 M. Schedl, G. Wilharm, S. Achatz, A. Ketrup, R. Niessner and D. Knopp, Anal.
Chem., 73 (2001) 5669.
253 J. Seneviratne and J.A. Cox, Talanta, 52 (2000) 801.
254 J.D. Badjic and N.M. Kostic, J. Phys. Chem., 104 (2000) 11081.
255 T.L. Yost, B.C. Fagan, L.R. Allain, C.E. Barnes, S. Dai, M. Sepaniak and Z. Xue, Anal.
Chem., 72 (2000) 5516.
256 R.D. Blanchard and J.K. Hardy, Anal. Chem., 56 (1984) 1621.
257 M.J. Yang and J. Pawliszyn, Anal. Chem., 65 (1993) 2538.
258 N.C. van de Merbel, J.J. Hegeman and U.A.Th. Brinkman, J. Chromatogr.A, 634
(1993) 1.
259 S.J. Bauer and R.G. Cooks, Am. Lab., 25 (1993) 36.
260 G. Matz, G. Kibelka, J. Dahl and F. Lenneman, J. Chromatogr.A, 830 (1999) 365.
261 B. Hauser, P. Popp and A. Paschke, Int. J. Environ. Anal. Chem., 74 (1999) 107.
262 H. Startman, in: P.M. Bungay, H.K. Lonsdale and M.N. de Pinho (Eds.), Synthetic
Membranes-Science, Engineering,and Applications. Reidel, Dordecht, 1986.
263 B.M. Cordero, J.L. Pavon, C.G. Pinto, M.E. Laespada, R.C. Martinez and E.R.
Gonzalo, J. Chromatogr.A, 902 (2000) 195.
264 P. Meares, in: P.M. Bungay, H.K. Lonsdale and M.N. de Pinho (Eds.), Synthetic
Membranes-Science, Engineering, and Applications. Reidel, Dordecht, 1986.
265 Y. Shen, J.A. Jonsson and L. Mathiasson, Anal. Chem., 70 (1998) 946.
266 K.F. Pratt and J. Pawliszyn, Anal. Chem., 64 (1992) 2107.
267 M.J. Yang, S. Harms, Y.Z. Lu and J. Pawliszyn, Anal. Chem., 66 (1994) 1339.
268 Y.Z. Luo, M. Adams and J. Pawliszyn, Analyst, 122 (1997) 1461.
269 Y.Z. Luo, Anal. Chem., 72 (2000) 1058.

1077
270 A. Segal, T. Gorecki, M. Phillippe, J. Lips and J. Pawliszyn, J. Chromatogr. A, 873
(2000)
271 M.J. Yang and J. Pawliszyn, LC-GC, 14 (1996) 364.
272 T.A. Sergeyeva, H. Matuschewski, S.A. Piletsky, J. Bendig, U. Schedler and M.
Ulbricht, J. Chromatogr.A, 907 (2000) 89.
273 G.S. Luo, F. Xia and Y.Y. Dai, J. Appl. Polym. Sci., 73 (1999) 1555.
274 L.W. Schmidt, J.J. Sun, J.S. Fritz, D.F. Hagen, C.G. Markell and E.E. Wisted, J.
Chromatogr., 641 (1993) 57.
275 B. Fu, E. Bakker, V.C. Yang and M.E. Meyerhoff, Macromolecules, 28 (1995) 5834.
276 R.G. Melcher, Anal. Chim. Acta, 214 (1988) 299.
277 R.G. Melcher and S.A. Bouyoucos, Process Control Qual., 1 (1990) 63.
278 P.L. Morabito and R.G. Melcher, Process Control Qual., 3 (1992) 35.
279 R.G. Melcher and P.L. Marbito, Anal. Chem., 62 (1990) 2183.
280 B. Hauser and P. Popp, J. Chromatogr.A, 909 (2001) 3.
281 F. Li and X. Zhang, Am. Lab., 33 (2001) 36.
282 J.A. Jonsson anmd L, Mathiasson, J. Chromatogr.A, 902 (2000) 205.
283 M. Ulbricht, K. Richau and H. Kamusewitz, Colloids Surfaces A, 138 (1998) 353.
284 M.J. Steuck, US Patent 4618533, 1986.
285 J.S. Fritz and J.J. Masso, J. Chromatogr.A, 907 (2001) 89.
286 J.S. Fritz and J.J. Masso, J. Chromatogr.A, 909 (2001) 79.
287 H.Y. Wang, T. Kobayashi and N. Fuji, J. Chem. Technol. Biotechnol., 70 (1997) 355.
288 S.A. Piletsky, H. Matuschewski, U. Schedler, A. Wilpert, E.V. Piletskaya, T.A. Thiele
and M. Ulbricht, Macromolecules, 33 (2000) 3092.
289 H. Kataoka, S. Narimatsu, H.L. Lord and J. Pawliszyn, Chromatography (Japanese
Review), 20 (1999) 237.
290 A. Penalver, E. Pocurull, F. Borrull and R.M. Marce, Trends Anal. Chem., 18 (1999)
557.
291 Z. Zhang, M.J. Yang and J. Pawliszyn, Anal. Chem., 66 (1994) 844A.
292 R. Eisert and J. Pawliszyn, Anal. Chem., 69 (1997) 3140.
293 Y. Gou, R. Eisert and J. Pawliszyn, J. Chromatogr.A, 873 (2000) 137.
294 Y. Gou, C. Tragas, H. Lord and J. Pawliszyn, J. Micro Sep., 12 (2000) 125.
295 Y. Guo and J. Pawliszyn, Anal. Chem., 72 (2000) 2774.
296 H. Hartmann, J. Burhenne and M. Spiteller, Fresen. Enuiron. Bull., 7 (1998) 96.
297 H. Kataoka, H.L. Lord and J. Pawliszyn, J. Chromatogr.B, 731 (1999) 353.
298 H. Kataoka, H.L. Lord and J. Pawliszyn, J. Anal. Toxicol., 24 (2000) 257.
299 H. Kataoka, H.L. Lord and J. Pawliszyn, J. Chromatogr.A, 880 (2000) 35.
300 H. Kataoka, H.L. Lord, S. Yamamoto, S. Narimatsu and J. Pawliszyn, J. Micro Sep.,
12 (2000) 493.
301 H. Kataoka, S. Narimatsu, H. Lord and J. Pawliszyn, Anal. Chem., 71 (1999) 4237.
302 H. Kataoka and J. Pawliszyn, Chromatographia,50 (1999) 532.
303 M.E. McComb, R.D. Oleschuk, E. Giller and H.D. Gesser, Talanta, 44 (1997) 2137.
304 J. Wu, H.L. Lord, J. Pawliszyn and H. Kataoka, J. Micro. Sep., 12 (2000) 255.
305 J. Wu, Z. Mester and J. Pawliszyn, Anal. Chim. Acta, 424 (2000) 211.
306 J. Wu and J. Pawliszyn, J. Chromatogr. A, 909 (2001) 37.
307 J. Wu and J. Pawliszyn, Anal. Chem., 73 (2001) 55.
308 S. Bigham, J. Medlar, A. Kabir, C. Shende, A. Alli and A. Malik, Anal. Chem., 74
(2002) 752.
309 D. Djozan and Y. Assadi, Chromatographia,(1997) 45 (suppl), 183.
310 H.B. Wan, H. Chi, M.K. Wong and C.Y. Mok, Anal. Chim. Acta (1994) 298, 219.
311 Y. Liu, M.L. Lee, K.J. Hageman, Y. Yang and S.B. Hawthorne, Anal. Chem., 69 (1997)
5001.
312 Y. Liu, Y. Shen and M.L. Lee, Anal. Chem., 69 (1997) 190.

1078
313 J. Pawliszyn, J. Chromatogr.Sci., 38 (2000) 270.
314 V. Mani, Applications of Solid-phase Microextraction. Royal Society of Chemistry,
Cambridge, UK, 1999.
315 D.W. Potter and J. Pawliszyn, Environ. Sci. Technol., 28 (1994) 298.
316 K.D. Buchholz and J. Pawlyszyn, Anal. Chem., 66 (1994) 160.
317 W. Vaes, C. Hamwijk, E.U. Ramos, H.J. Verhaar and J.L. Hermens, Anal. Chem., 68
(1996) 4458.
318 A. Podgornik, M. Barut, J. Jancar and A. Strancar, J. Chromatogr.A, 848 (1999) 51.
319 R. Swart, S. Brouwer, J.C. Kraak and H. Poppe, J. Chromatogr.A, 732 (1996) 201.
320 R. Swart, J.C. Kraak and H. Poppe, J. Chromatogr.A., 689 (1995) 177.
321 R. Swart, J.C. Kraak and H. Poppe, Chromatographia,40 (1995) 587.
322 Z.X. Tan and V.T. Remcho, Anal. Chem., 69 (1997) 581.
323 B.J. Hall and J.S. Brodbelt, J. Chromatogr.A., 777 (1997) 275.
324 A. Penalver, E. Pocurull, F. Borrull and R.M. Marce, J. Chromatogr.A, 922 (2001)
377.
325 Y.C. Wu and S.D. Huang, Anal. Chem., 71 (1999) 310.
326 L.I. Andersson, Anal. Chem., 68 (1996) 111.
327 J. Wu, X. Yu, H. Lord and J. Pawliszyn, Analyst, 125 (2000) 391.
328 B.W. Wright, P.A. Peaden, M.L. Lee and T.J. Stark, J. Chromatogr., 248 (1982) 17.
329 L.G. Blomberg, J. Microcol. Sep., 2 (1990) 62.
330 K. Janak, M. Horka and M. Krejci, J. Microcol. Sep., 3 (1991) 115.
331 V.G. Berezkin, V.E. Shiryaeva and T.P. Popova, Zh. Analit. Khim., 47 (1992) 825.
332 D.X. Wang, S.L. Chong and A. Malik, Anal. Chem., 69 (1997) 4566.
333 S.L. Chong, D. Wang, J.D. Hayes, B.W. Wilhite and A. Malik, Anal. Chem., 69 (1997)
3889.
334 J.D. Hayes and A. Malik, J. Chromatogr.B, 695 (1997) 3.
335 A. Malik and S.L. Chong, in: J. Pawliszyn (Ed.), Applications of Solid-phase
Microextraction. Royal Society of Chemistry, Cambridge, UK, 1999, pp. 73-91.
336 J.D. Hayes and A. Malik, Anal. Chem., 73 (2001) 987.
337 J.D. Hayes and A. Malik, Anal. Chem., 72 (2000) 4090.
338 Z. Chen and T. Hobo, Anal. Chem., 73 (2001) 3348.
339 D.C. Collins, Q. Tang, N. Wu and M.L. Lee, J. Micro Sep., 12 (2000) 442.
340 M.T. Dulay, J.P. Quirino, B.D. Bennett, M. Kato and R.N. Zare, Anal. Chem., 73
(2001) 3921.
341 L. Roed, E. Lundanes and T. Greibrokk, J. Chromatogr.A, 890 (2000) 347.
342 L. Roed, E. Lundanes and T. Greibrokk, J. Micro Sep., 12 (2000) 561.
343 Q. Tang, N. Wu and M.L. Lee, J. Microcol. Sep., 11 (1999) 550.
344 Q. Tang, N. Wu and M.L. Lee, J. Micro. Sep., 12 (2000) 6.
345 Q. Tang, B. Xin and M.L. Lee, J. Chromatogr.A, 837 (1999) 35.
346 N. Ishizuka, H. Minakuchi, K. Nakanishi, N. Soga, K. Hosoya and N. Tanaka, J.
High Resol. Chromatogr., 21 (1998) 477.
347 H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka and N. Tanaka, J. Chromatogr.A,
762 (1997) 135.
348 H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka and N. Tanaka, J. Chromatogr.A,
797 (1998) 121.
349 N. Ishizuka, H. Minakuchi, K. Nakanishi, N. Soga, H. Nagayama, K. Hosoya and N.
Tanaka, Anal. Chem., 72 (2000) 1275.
350 N. Ishizuka, H. Minakuchi, K. Nakanishi, N. Soga and N. Tanaka, J. Chromatogr.A,
797 (1998) 133.
351 H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka and N. Tanaka, Anal. Chem., 68
(1996) 3498.
352 K. Nakanishi, H. Minakuchi, N. Soga and N. Tanaka, J. Sol Gel Sci. Technol., 8

1079
 (1997) 547.
353 Z.Y. Wang, C.H. Xiao, C.Y. Wu, C. Wu and H. Han, J. Chromatogr.A, 893 (2000) 157.
354 Z. Zeng, W.L. Qiu and H. Xing, et al. Anal. Sci., 16 (2000) 851.
355 Z. Zeng, W. Qiu and Z. Huang, Anal. Chem., 73 (2001) 2429.
356 Z. Zeng, W. Qiu, M. Yang, X. Wei, Z. Huang and F. Li, J. Chromatogr.A, 934 (2001) 51.
357 SUPELCO Catalog 2000: ChromatographyProducts for Analysis and Purification.
Aldrich-Sigma Co., USA, 2000, p. 259.
358 K. Jinno, M. Taniguchi and M. Hayashida, J. Pharm.Biomed. Anal., 17 (1998) 1081.
359 D.A. Volmer and J.P. Hui, Rapid. Commun. Mass. Spectrom., 12 (1998) 123.
360 K. Jinno, M. Taniguchi, H. Sawada and M. Hayashida, Applications of Solid-Phase
Microextraction. Royal Society of Chemistry, Cambridge, UK, 1999, pp. 527-539.
361 M. Moder and P. Popp, in: J. Pawliszyn (Ed.), Applications of Solid-Phase Micro-
extraction, Royal Society of Chemistry, Cambridge, UK, 1999, pp. 311-326.
362 Y. Hirata and J. Pawliszyn, J. Microcol. Sep., 6 (1994) 443.
363 D. Figeys, A. Ducret, J.R. Yates and R. Aebersold, Nat. Biotechnol., 14 (1996) 1579.
364 D. Figeys, Y. Zhang and R. Aebersold, Electrophoresis, 19 (1998) 2338.
365 R. Carabias-Martinez, E. Rodriguez-Gonzalo, J. Dominguez-Alvarez and J.
Hernandez-Mendez, J. Chromatogr.A, 869 (2000) 451.
366 S.M. Fields, Anal. Chem., 68 (1996) 2709.
367 G. Chirica and V.T. Remcho, Electrophoresis,20 (1999) 50.
368 S. Bigham, J. Medlar, K. Hamlet, A. Kabir and A. Malik. Manuscript in preparation.
369 G.R. Newkome, C.N. Moorefield and F. Vogtle, Dendrimersand Dendrons: Concepts,
Syntheses, Applications. Wiley-VCH, Weinheim, Germany, 2001.
370 F. Zeng and S.CI Zimmerman, Chem. Rev., 97 (1997) 1681.
371 R.S. Roy and K. Das, Chem. Commun. 2000 (7) (2000) 519.
372 D. Astruc, J.C. Blais, E, Cloutet, L. Djakovitch, S. Rigaut, J. Ruiz, V. Sartor and C.
Valerio, Top. Curr. Chem., 210 (2000) 229.
373 M.T. Reetz, Heterocycl. Chem., 35 (1998) 1065.
374 A. Hierlemann, J.K. Campbell, L.A. Baker, R.M. Crooks and A.J. Ricco, J. Am. Chem.
Soc., 120 (1998) 5323.
375 A.I. Cooper, J.D. Londono, G. Wignall, et al. Nature (London), 389 (1997) 368.
376 S.A. Kuzdzal, C.A. Monnig, G.R. Newkome and C.N. Moorefield, J. Chem. Soc.,
Chem. Commun, (1994) 2139.
377 S.A. Kuzdzal, C.A. Monnig, GR. Newkome and C.N. Moorefield, J. Am. Chem. Soc.,
119 (1997) 2255.
378 P.C. Muijselaar, C. Cramers, J.M. Jansen, E. van de Barbander-van den Berg, J.
High Resolut. Chromatogr., 18 (1995) 121.
379 M. Castagnola, L. Cassiano, A. Lupi, I. Messana, M. Patamia, R. Rabino, D.V.
Rossetti and B. Giardina, J. Chromatogr.A, 694 (1995) 463.
380 C. Stathakis, E.A. Arriaga and N.J. Dovichi, J. Chromatogr.A, 817 (1998) 233.
381 G.R. Newkome, K.S. Yoo, A. Kabir and A. Malik, TetrahedronLett., 42 (2001) 7537.
382 A. Kabir, C. Hamlet, S.Y. Yoo, G.R. Newkome and A. Malik, Manuscript in
preparation.

1080
Chapter 33

Immunosorbents in sample preparation

Valerie Pichon, Nathalie Delaunay-Bertoncini and
Marie-Claire Hennion

33.1 INTRODUCTION: INTEREST FOR SELECTIVE SOLID-PHASE

EXTRACTION METHODS

There is, nowadays, an increasing effort to monitor the environment for com-
pounds that may pose a risk to human and ecosystem health. This requires ana-
lytical methods that can be applied to different types of matrix (waters, soils,
sediments, foodstuff, biological fluids, etc.). Despite advances in the develop-
ment of highly sensitive analytical instrumentation for the final determination
of analytes, a sample pretreatment is usually necessary in order to extract, con-
centrate and isolate the analytes of interest in complex matrices. Solid-phase
extraction is, nowadays, widely accepted as a powerful technique for sample
preparation. Recent years have witnessed important developments in SPE for-
mats and sorbents with improved n-alkylsilicas, new highly cross-linked copoly-
mers and carbonaceous sorbents. These sorbents now have the capability for
extracting organic compounds over a wide range of polarities for screening pur-
poses [1]. However, analyte retention is based on hydrophobic interactions and
their selectivity is often too low for trace analysis in complex samples. Co-
extraction of other analytes and matrix interferences generally occurs, and this
can even become a major problem when analytes of interest are at the trace level
and interferences are present at high concentrations. Additional clean-up proce-
dures are then required, but then the sample pretreatment involves several
steps, with both an increase in the risk of loss and contamination and a decrease
of the reliability of the results.
Selectivity can be greatly enhanced by using materials involving antigen-
antibody interactions, thus providing selective extraction methods based on
molecular recognition. The antibodies are covalently bonded onto an appropri-
ate sorbent to form a so-called immunosorbent (IS), to be packed in a SPE car-
tridge or precolumn. Thanks to the high affinity and high selectivity of the
antigen-antibody interactions, they allow a high degree of molecular selectivity
and have been shown to be a unique tool in the sample preparation area in recent
years [2-6]. Extraction, concentration and clean up from complex liquid samples
ComprehensiveAnalytical Chemisty XXXVII
J. Pawliszyn (Ed.)
0 2002 Elsevier Science B.V. All rights reserved 1081
are achieved in the same step from large volumes when required. Their applica-
tion to extracts from solid matrixes is solvent-free and simpler than any other
clean-up procedure.
First, ISs have been described in the biological field because of the availabil-
ity of antibodies, which can be very selective for large molecules and easily
obtained. Many examples have been described for the immunoextraction of anti-
bodies, hormones, peptides, enzymes, recombinant proteins, viruses and sub-
cellular components. Obtaining selective antibodies for small size molecules was
more difficult and the development of immunochemical methods in the SPE field
targeting low molecular weight analytes is rather recent. The binding of analyte
to antibody is the result of a good spatial complementarity and is a function of
the sum of intermolecular interactions. Therefore, an antibody can also bind one
or more analytes with a structure similar to the analyte that has induced the
immune response, and this is the so-called cross-reactivity of antibodies. It is
usually considered a negative feature for immunoassays, but it was exploited in
extraction, so that ISs have been developed for single analytes, single analytes
and metabolites or a class of structurally related analytes. During the last
decade, commercial IS have been introduced for the clean up of samples for the
analysis of toxins (aflatoxins, ochratoxins, fumonisins) and veterinary drugs
(clenbuterol) in food. ISs for many contaminants or class of contaminants have
been described in the literature [4].Several class-selective immunosorbents
have been optimized for trapping groups of pesticides and priority industrial
organic pollutants [6]. Immunosorbents are robust and validation studies using
certified reference materials have demonstrated their reliability.
The objective of this chapter is to describe the potential of immunoextrac-
tion. When synthesizing an immunosorbent (IS) and developing an immuno-
extraction procedure, several parameters have to be carefully controlled and
optimized for obtaining a robust and selective SPE method. These parameters
first include the nature and the amount of the antibodies and the sorbent
selected for their immobilization. The importance of the capacity and its rela-
tionship with the analyte recovery and breakthrough volume are highlighted.
Emphasis is also given to the optimization of the extraction sequence, especially
to the desorption step.

33.2 CHARACTERISTICS OF IMMUNOSORBENTS

33.2.1 Antibodies

The overall process for the preparation of an immunosorbent first involves the
production of antibodies. Because compounds of low molecular weight (<1000
Dalton) are unable to evoke an immune response, they have to be modified by
binding to a larger carrier molecule, usually a protein (e.g., bovin serum
albumin, P-lactoglobuline) to form an immunoconjugate that can be used for the
immunization. It is often necessary to introduce a functional group into the

1082
selected molecule in order to make possible this coupling. The design of the
so-called hapten is the most crucial step in the development of an immunochemi-
cal technique for small molecules because it determines the features of the
resulting antibodies. The optimum hapten for a selected target analyte has to be
a near-perfect mimic of this molecule in structure, geometry, electronic and
H-bonding capabilities and hydrophobic properties [7]. Moreover, the strategy
will be different whether a single compound is targeted or a whole class of
structurally related compounds. For the class-selective assay, the handle will be
best located at a position that leaves the common sites exposed to the immune
system. For small molecules, the use of a spacer arm in the linker is required in
order to favor recognition by the immune system. The length and the chemical
structure of the linker should be selected in order to maximize the exposure of
the target molecules and the spacer itself should not be selectively recognized.
The length of the spacer from three to six atoms has been shown to be optimal. A
good immune response seems to be obtained with 10-30 haptenic groups per 100
kDa of carrier protein [8].
New technologies for developing specific antibodies (Abs) are an active area,
using sometimes very sophisticated methods, but one has to realize that to date,
commercial Abs use mainly polyclonal or monoclonal antibodies. In both cases,
the different steps, required to produce Abs, are similar up to the immunization
of the host animal with the immunogen. After a period of a few weeks, the serum
is collected and the G-type immunoglobulins (IgG) fraction is isolated an puri-
fied. This fraction contains polyclonal antibodies (pAbs) made of an heteroge-
neous mixture of antibodies that can bind the antigen with a variety of
strengths, because they are directed against the various antigenic determinants
(epitopes) on the antigen. When a small molecule is targeted, one can expect the
number of epitopes to be low. It is difficult to measure the concentration of
specific antibodies in the whole IgG fraction. It is commonly found that it
contains about 15% of active antibodies specific of the target molecule. If
polyclonal antibodies have proven to be powerful in immunoassay techniques,
they have some limitations because polyclonal antiserum varies from one animal
to another and the supply of Abs ends when the animal dies. This can be
overcome by the use of the hybridoma technique based on the fusion of spleen
lymphocytes (which produce Abs) after isolation from immunized mice with
myeloma cells. The advantage of this technology is the capability of an unlimited
production of monoclonal Abs (mAbs) with constant affinities. However, in
contrast to pAbs production, this technique is difficult, laborious, time-
consuming and expensive.

33.2.2 Antibody immobilization

The immunosorbents (ISs) are produced by linking the antibodies to a solid

support. Immobilization of antibodies can be done in different ways, such as
physical or affinity adsorption or covalent binding. Physical adsorption, which

1083
has been widely used for the coating of microtiter plates in immunoassays, is
based on hydrophobic interactions and cannot be used when the desorption of
the trapped compounds requires an organic modifier (see immunoextraction
procedure). The affinity adsorption, using protein A or G immobilized on a
stationary phase, allows a good orientation of the antibodies, but this process
also suffers from the risk of lingand leakage and steric problems due to the size of
the immobilized proteins. Covalent binding appears to be the best process of
immobilization to obtain a good stability between the sorbent and the antibod-
ies. In this case, some specific properties are required for the sorbents since they
should be porous and possess large-size pores, contain appropriate functional
groups for the bonding and have a high capacity. They should also be hydrophilic
in order to avoid additional non-specific interactions. Most of the studies report
the use of silica and agarose gels for the covalent immobilization of antibodies.
Another method-the sol-gel method-consists in encapsulating antibodies
in the pores of a hydrophilic glass matrix. This method was applied to the synthe-
sis of immunosorbents for PAHs [9,10]. Many authors have also attempted to
use hydrophilic organic polymers because they allow an oriented bonding of the
antibodies and they have higher capacities. However, hydrophilic polymers all
contain 7r-bonds that give rise to non-specific hydrophobic interactions and
decrease dramatically the selectivity of the ISs, which is their primary interest in
sample preparation methods.

33.3 PROPERTIES OF SOLID-PHASE EXTRACTION

IMMUNOSORBENT

33.3.1 The immunoextraction sequence

A typical SPE sequence using an immunosorbent packed in a disposable

cartridge is very similar to that using a conventional C18 silica cartridge. One
difference is that the IS should be kept in a wet media, usually a PBS (phosphate
buffer saline) solution. A typical sequence of immunoextraction is presented in
Fig. 33.1. It includes a conditioning step with a few ml of pure water, the sample
(a) (b) (c)

<

J -< Antibodies
o 4 Analyte-antigen
011 1o: Interferences

Fig. 33.1. Off-line procedure used for the immunosample pretreatment on immunosorbents. (a)
Percolation of the sample; (b) washing to eliminate the nonretained compounds; (c) elution of
compounds retained by the immobilized antibodies.

1084
percolation, a washing step using a few mL of pure water or water containing a
small amount of organic solvent in order to eliminate compounds that do not
react with antibodies by specific interactions, and, finally, the desorption step
with a few ml of an organic solvent-water mixture. Regeneration is possible
using a PBS solution and storage at 4°C.

32.3.2 Affinity for the analyte-antigen and the structurally related

analytes: class specificity

The capabilities of ISs for trapping the antigen and related analytes depends on
the cross-reactivity of the antibodies. The easiest way to measure the potential of
an IS for trapping a group of related analytes is to measure the recoveries (ratio
between the extracted amount and the percolated amount) that are obtained
when samples which are spiked with all the analytes are percolated. Column A of
Table 33.1 shows this potential for ISs obtained for phenylureas when bonding
polyclonal anti-isoproturon and anti-chlortoluron antibodies [11-14]. It was
difficult to predict that the anti-isoproturon IS should be class-specific because
its structure is slightly different from those of other phenylureas (see Fig. 33.2).
This was easier for anti-chlortoluron antibodies and the IS show good recoveries
for almost all the phenylureas. Figure 33.3 shows the chromatograms
corresponding to the on-line analysis of 50 ml of drinking water (Fig. 33.3a) and
surface-water (Fig. 33.3b) samples spiked with 0.5 rg/1l of a mixture of phenyl-
ureas using a precolumn packed with the anti-chlortoluron IS [12,15]. The
spiking mixture contained 13 phenylureas and only fenuron and fluometuron

TABLE 33.1
Potential for class selective immunoextraction sorbents. A: Comparison of two IS prepared using
anti-isoproturon and anti-chlortoluron polyclonal antibodies; B: comparison of two IS prepared
with monoclonal and polyclonal anti-isoproturon antibodies. Average recoveries obtained for
nine phenylureas after the percolation of water, spiked with 1 g/l of each analyte (from Refs.
[11-14])
A B
Polyclonal Polyclonal Monoclonal Polyclonal
anti-isoproturon anti-chlortoluron anti-isoproturon anti-isoproturon
(V = 50 ml) (V = 50 ml) (V = 25 ml) (V = 25 ml)
Fenuron - - 0 <10
Metoxuron 21 (3) 80 (5) 19 (5) 61 (4)
Monuron 98 (4) 78 (8) 28 (5) 101 (5)
Chlortoluron 95 (4) 95 (5) 100 (2) 102 (8)
Isoproturon 99 (3) 90 (8) 99 (1) 99 (3)
Difenoxuron 37 (4) 17 (9) 6 (6) 67 (6)
Linuron 61 (6) 85 (8) 83 (2) 96 (5)
Chlorbromuron 95 (5) 102 (5) 100 (9) 101 (5)
Diflubenzuron 101 (3) 76 (9) 59 (7) 99 (6)

1085
 H3 C' \' No OH3
,%~cH H3C -- N /CH 3
H3C - C--
CH 3 Cl 0 CH
isoproturon chlortoluron
H

H3 C,C N< -¢ N/OH3

CH3
demethyl-isoproturon fenuron

H3C\ N5H
CI~ N N /CH3
HC > -N <

O H 01 Cr>I-N\OMe
di-demethyl-isoproturon linuron
H
N H H3

CI - N N<CH
CH3
O CF 3 CH 3
monuron
fluometuron
H
C0 /\ N N<OH
Br N
H
OMe

Cl 0 CH3
CI O~N
\CH 3
diuron chlorbromuron
CH,O / \ N CH3
H
CiN CH 3
CI O CH3
methoxuron C CneburonH,
neburon
Fig. 33.2. Structure of commonly used phenylureas.

were not recovered. The chromatograms for these two different water matrices
are similar, since no interfering analytes are showing up in the River Seine
sample, thus pointing out the selectivity of the immunoextraction.
The values reported in Table 33.1 indicate recoveries of 80-100% for several
analytes and lower recoveries for other analytes. This is explained by the differ-
ent affinity constants that exist between the structurally related analytes and
the polyclonal antibodies bonded onto silica. The affinity order can be made clas-
sifying analytes with increasing recoveries, but even for analytes having total
recovery, the affinity can be different.

1086
 (b)

Fig. 33.3. Class-selective extraction of phenylurea

herbicides: on-line solid-phase immunoextraction-
LC analysis-UV diode array detection of 50 ml of
river Seine (Paris city) water sample spiked with a
mixture of 13 phenylureas at a concentration level
of 0.5 /g/l for each compound, using a precolumn
packed with an anti-chlortoluron IS. (1) Fenuron, (a)
(2) metoxuron, (3) monuron, (4) metabenzthia-
zuron, (5) chlortolurdn, (6) fluometuron, (7) iso-
proturon, (8) difenoxuron, (9) buturon, (10)
Imuron, (11) cnlororomuron, 1Z) alIluzoenzuron,
it -- ---- T 1=
"'
r"-- I' -'
I ;- / A
t10) ILDuuron. U aV ecCllL snow1n aL z44 nm1
20 40 60' 80 T@
(from Ref. [6]).

ISs have been developed for single analytes and their metabolites. The list
includes drugs such as clenbuterol, chloramphenicol, diethylstilbestrol, oest-
radiol-17-3, thiobendazole, tetracyclins or aflatoxins [16-26], algal toxins [27-
29], pesticides and their transformation products [11-13,15,17,30-41]. Class-
selective IS have been described for the extraction of triazines, phenylureas,
polyaromatic hydrocarbons, BTEX (benzene, toluene, ethylbenzene and xylene)
isomers, benzidine and azodyes [9-15,36,37,42-50].

33.3.3 Polyclonal vs. monoclonal antibodies for class-selective ISs

Many laboratory-made class-selective ISs have been made using polyclonal

antibodies. However, the trend is to use monoclonal antibodies because the IS
reproducibility can only be achieved by a long-term supply of reproducible
antibodies which can be better guaranteed by mAbs. Moreover, further mAb
large-scale production no longer requires animals. The third column of Table
33.1 shows the recoveries obtained for phenylureas using an IS obtained when
bonding monoclonal anti-isoproturon antibodies. It can be seen that cross-
reactivity and class-selective trapping are obtained when using mAbs. Direct

1087
comparison of individual recoveries obtained with mAbs and pAbs requires the
knowledge of the amount of bonded antibodies in each case and will be discussed
later. The fact that mAbs show almost the same cross-reactivity as pAbs can be
easily explained by the small size of the molecules. When the molecules of
interest are very small, the polyclonal antibody mixture cannot contain a large
number of specific antibodies for the different parts of the small molecule as it
can contain when the molecule is very large and has several epitopes. Then, the
probability is high for having similar properties between monoclonal and poly-
clonal antibodies.

32.3.4 Immunosorbent capacity

The capacity of an immunosorbent is defined as the maximal amount of the

analyte-antigen that can be bound onto the sorbent during the percolation of the
sample. If a larger amount of compound is percolated through the IS, recoveries
of extraction will decrease due to its overloading. The capacity first depends on
the total number of immobilized active antibodies (specific for the target ana-
lyte). Knowing the amount of antibodies bonded during the immobilization pro-
cedure, it is possible to estimate the theoretical capacity of an immunosorbent.
However, this estimation is based on the first hypothesis that all the immobi-
lized antibodies are specific of the analyte antigen. Such an evaluation is possible
when using monoclonal antibodies but impossible when using polyclonal ones
since the amount of specific antibodies in the IgG fraction is unknown. Although
it is theoretically possible to isolate specific antibodies in the IgG fraction, it is
not often described because this can only be performed using a structurally
related analyte within the targeted group which shows enough affinity for
retaining antibodies but not too much for allowing their desorption without
organic solvent. The second hypothesis is that the immobilization process allows
a good accessibility of the antigen for the active sites of the antibodies, which is
certainly not ensured in the case of random orientation. The accessibility
depends on the bonding density, i.e. the number of antibodies per g of sorbent (or
per ml of bed sorbent) [14]. In the literature, different bonding densities are
encountered, depending whether monoclonal or polyclonal antibodies are used.
Because a theoretical approach of the capacity is difficult, it is necessary to
have experimental methods for capacity determination. A common method
consists of measuring the amount of analyte adsorbed as a function of the
analyte concentration in the percolated sample (for a fixed percolated volume).
Figure 33.4 corresponds to the curve obtained after the percolation of 25 ml of
pure water spiked with an increasing amount of atrazine on an anti-atrazine IS.
The peak areas reported on the y-axis are related to the amount of atrazine
bonded on the IS, that was desorbed and analyzed by LC with a UV detection.
The resulting curve has a Langmuir shape with a linear part, corresponding to
the range of concentration that can be exploited for quantitative analysis, and a
plateau corresponding to the capacity.

1088
 S
2
TI

Concentration
(pg/ll)

Fig. 33.4. Binding-capacity curve obtained with an anti-atrazine IS by measuring the amount of
atrazine adsorbed onto the IS as a function of the analyte concentration in the percolated
samples (from Ref. [61).

The objective of a recent study was to optimize the production of ISs using
well defined monoclonal antibodies at different bonding densities [14]. First, the
capacities obtained onto silica and Sepharose have been compared. When anti-
isoproturon mAbs were bonded to silica (glutardialdehyde-activated silica with
30 nm pore size) at a density of 1 mg per 100 mg silica, the capacity of 100 mg of
the bonded support was 1.35 ± 0.1 Ag. When the same antibodies were bonded to
Sepharose 4B at a density of 1 mg per 0.5 ml gel, the capacity of the 0.5 ml of
bonded support was 1.34 ± 0.1/ g. These similar capacities for the two ISs show
that the accessibility of the antibodies is equivalent and high. Up to the bonding
density of 1 mg of antibodies per 100 mg silica, the relation between capacity and
amount of antibodies was linear. Above this value, the capacity was lower than
expected, certainly because of steric hindrance between antibodies.
When an IS targets several structurally related analytes, the definition of the
capacity is more complex. The competition process between the compounds in a
group during the percolation can be illustrated by the measurements of capaci-
ties using a mixture of these structural analogs. An example of measurements of
capacity obtained after the percolation of a mixture of compounds on an anti-
atrazine IS is illustrated in Fig. 33.5. The results show that a higher capacity is
obtained for the analyte-antigen, i.e. atrazine, when it is alone in the sample
than when the percolated samples were spiked with atrazine and several tri-
azines. The capacities measured for related compounds depend on the affinity of
the antibodies for the compounds. In this example, a stronger affinity for the
propazine than for atrazine induces a larger apparent capacity for propazine
than for the analyte-antigen. Then, the capacity measured with the analyte-
antigen cannot be transposed to the other compounds within the group and
depends strongly on the competition process, i.e. on the number of compounds
and their respective concentrations in the sample.

1089
 -a
2
I
a.

0.0 1.0 2.0 3.0

Concentration (pgn)

Fig. 33.5. Calibration curves of triazines in anti-atrazine IS. Percolation of 25 ml of water spiked
at concentration from 0.1 to 3 g/l for each compound (from Ref. [12]).

Quantitative analysis can be made only if recoveries are constant over the
whole linear calibration range and if the calibration curves do not depend on this
competition process. In other words, recoveries and calibration curves should be
independent of whether the analyte-antigen or any other related analytes is on
its own or in a mixture of related compounds. Therefore, the calibration range
should be defined by experiments based on the percolation of samples simulta-
neously spiked with a mixture of compounds. In this case, the more restricted
linear range will be obtained for the compounds for which the affinity of the anti-
bodies is the lowest (terbutylazine in the example of Fig. 33.5). The upper limit
of this linear range defines the concentration range of the method for all the
compounds of the group for quantitative analyses of real samples. Increasing the
amount of antibodies immobilized on the sorbent can extend this range of con-
centration. However, the amount of sorbent has to be adapted in order to keep a
bonding density that ensures a good accessibility of the analytes for the
antibodies.

33.4 OPTIMIZATION OF THE IMMUNOEXTRACTION PROCEDURE

33.4.1 Sample volume and extraction recovery

One important parameter in the SPE sequence is the sample volume which can
be percolated without loss in recovery. Two parameters can affect the recovery
which are the IS capacity and/or the affinity of antibodies towards compounds.
Insufficient retention induces a low breakthrough volume and an incomplete
recovery when overcome. In general, immunoextraction is used to extract
compounds at trace levels, so, the greater the breakthrough volume and the
capacity are, the better the limit of detection is. As seen in the section above,

1090
capacities are rather low compared to that of other reversed-phase sorbents for
which overloading of the capacity is unlikely to occur. So, it is essential to verify
if experimental conditions do not induce overloading of the IS capacity or of the
breakthrough volume, because then calibrations are no longer linear.
The high affinity for the analyte-antigen is seen by the high sample volume
that can be handled without breakthrough. One litre of water containing 100 ng
of isoproturon was percolated with 100% recovery through an IS obtained with 1
g of silica and 200 Al of crude anti-isoproturon serum [51]. High breakthrough
volumes were also reported using anti-phenylurea or anti-triazine ISs when
water was spiked with the analyte-antigen alone [13,42] allowing detection
limits in water in the range 5-50 ng/l.
Comparison of recoveries provided by different ISs requires knowledge of the
nature of the bonded antibodies, which can be made by the capacity measure-
ments. Columns B of Table 33.1 gives extraction recoveries obtained after the
percolation of 25 ml of water spiked with nine phenylureas on anti-isoproturon
ISs based on monoclonal and polyclonal antibodies [11]. For these experiments,
the amount of the respective antibodies was adapted in order to have the same
capacity (4200 ng). This capacity was larger than the sum of the amount of
compounds in the sample (25 ng of each compound) in order to avoid a decrease
of the recoveries caused by the overloading of the IS. As expected, the recoveries
of the analyte-antigen, i.e. isoproturon, were close to 100% for both -ISs. They
were also similar for fenuron, chlortoluron, linuron and chlorbromuron. For the
other compounds, the recoveries obtained with monoclonal antibodies were
lower than with the polyclonal ones. Nevertheless, the cross-reactivity of the
monoclonal antibodies remains wide ensuring their application to the selective
extraction of class of related compounds with the advantage to provide more
reproducible ISs.
This comparison was carrying out using ISs characterized by the same
capacity and by percolating the same volume of sample. The modification of
these two parameters can modify the recoveries. Table 33.2 shows the recoveries
obtained after the percolation of 25 and 50 ml of water samples spiked with a
single phenylurea on 100 mg and 350 mg of an anti-isoproturon IS (bonding
density of 1 mg of monoclonal Abs for 100 mg of silica). Results obtained with
100 mg of IS show that a constant recovery about 100% is obtained for the
analyte-antigen when increasing the sample volume due to the strong affinity of
the antibodies for this compound. For other compounds, the increase of the
percolated volume causes a decrease of the recoveries due to the lower affinity of
the antibodies for these compounds. The breakthrough of these analytes can be a
problem for real application when low levels of concentration have to be
detected. High enrichment factors necessitate the percolation of high volumes of
sample with high recoveries. The increase of the amount of IS causes an increase
of the capacity of the IS. Then, the breakthrough of analytes occurs for larger
volumes of sample. This is demonstrated by the results in Table 33.2 obtained
with 350 mg of IS, which contains 3.5 times more antibodies than 100 mg of IS

1091
TABLE 33.2
Average recoveries (%) and standard deviation (n = 3), obtained after the percolation of 10, 25, or
50 ml LC grade water spiked with phenyureas through 100 or 350 mg of IS bonded with 1 mg and
3.5 mg of monoclonal anti-isoproturon antibodies, respectively
Amount of IS: 100 mg 350 mg
Sample volume: 25 ml 50 ml 25 ml 50 ml
Isoproturon (ISO) 84 (2) 94 (10) 95 (8) 100 (1)
Demethyl-ISO 72 (4) 52 (15) 100 (1) 102 (3)
Didemethy-ISO 66 (21) 24 (8) 99 (2) 100 (2)
Chlorbromuron 32 (3) 14 (7) 102 (3) 59 (8)
Diuron 24 (13) 15 (6) 95 (8) 64 (9)
Fluometuron 11 (0) 8 (2) 51 (13) 25 (5)
Chlortoluron 19 (2) 11 (5) 97 (20) 49 (3)
Linuron 5 (1) 9 (4) 51 (5) 35 (7)
Neburon 15 (0) 13 (7) 60 (9) 27 (4)
Monuron 8 (3) 2 (2) 28 (5) 17 (6)

(with the same bonding density). The increase of the capacity allows the simulta-
neous extraction of a larger number of compounds with higher recoveries.

33.4.2 Elution conditions

For desorption or elution from the IS, the analyte-antibody complex formed dur-
ing the percolation of the sample has to be disrupted. As the biochemical interac-
tion energies are high, desorption will occur only by notably modifying the
experimental conditions. Different approaches are possible: displacer agents,
chaotropic agents, pH variations or water-organic modifier mixtures. Solutions
of chaotropic ions disrupt the water structure around the antibody. In fact, vari-
ous aqueous solutions that have been successfully applied for many years for the
desorption of proteins were shown to be unable to desorb small molecules. Prob-
ably, the desorption of proteins is mainly based on the structure change of the
bound protein and not only on the antibody structure change. Elution with low
pH solutions have been described (acetic acid 2%, formic acid 0.1%, citric acid 0.1
M, 0.2 M glycine-HCl), but one inconvenience is that large volumes are required
for complete desorption [30,40]. Another solution is to reconcentrate this vol-
ume using classical SPE sorbent before analysis by LC or GC.
A water-organic mixture is often required to elute small molecules from the
IS. Common organic solvents are acetonitrile [10,12,13,15,20,31,33,34,42-47],
ethanol [35,52] or methanol [11-13,15,36,37,48]. The desorption volume is then
optimized to a few ml or less, thus leading to a concentrated extract. Pure
organic solvents such as methanol are often recommended in the procedure
supplied with commercialized ISs. The use of a mixture of water with organic

1092
solvent currently used in reversed-phase LC also induces an easy on-line coup-
ling of immunoextraction with LC (see below). The choice of the elution condi-
tions are important for reusability of the IS.

33.4.3 Regeneration and re-usability

Reusability involves a possible regeneration and this greatly depends on the

conditions used for elution which occurs by dissociation of the antibody-antigen
complex. However, the chemical bond between the antibody and the sorbent
solid surface greatly increases the stability of antibodies even in contact with a
solution containing high amounts of organic solvents, so that many examples of
IS reusability have been described (see Ref. [4]). The best parameter to follow
the IS evolution is its capacity because it is a measurement of the amount of
active antibodies. Using too low pH improved the elution but induced a rapid
decrease in capacity [41]. The temperature was also shown to have an effect on
regeneration. It has been observed that an IS maintained at 4C could be
recycled approx. 200 times before a detectable loss in capacity, whereas only 50
cycles could be applied when operated at room temperature [53]. PBS is the
common buffer used for regeneration in combination with a storage
temperature at 4°C. When elution is made with glycine or urea solution, a large
volume is needed for elution but, on the other hand, refolding of the antibodies is
easily obtained in PBS, whereas several studies pointed out some reduction in IS
effectiveness induced by organic solvents. However, using organic solvents does
not mean that regeneration is irreversible, but kinetics of the regeneration is
quite long and two days at 4°C can be necessary to re-obtain the initial capacity
[54]. The IS can be used several times without apparent loss in recovery when
the initial capacity is high enough. Several studies using commercial IS have
reported the regeneration conditions for their reusability. An anti-clenbuterol IS
could be regenerated at least 18 times for the clean-up of liver extracts using 2 ml
of 80% ethanol in water [16]. After regeneration using PBS, an aqueous solution
containing a known amount of clenbuterol was passed through the IS cartridge
to ensure that the cartridge functioned properly.
However, the reusability of disposable off-line cartridges is never
recommended when real complex samples are analyzed. Regeneration is more
important using on-line procedures.

33.4.4 Flow-rate effect during the sample percolation step

The effect of the flow-rate on the binding of analytes on the IS has been studied
[25,50,55,56]. When the application of the sample on the IS is performed by a
pump as in on-line systems, this parameter can easily be controlled. An increase
in recovery from 25 to 95% was observed as the flow-rate decreased from 2.0 to
0.2 ml/min when 1 ml of 50 g/l toluene in water was percolated through a 2 cm
x 1 mm i.d. precolumn packed with a silica-based anti-toluene IS [55]. Using a

1093
1.5 cm x 4.6 mm i.d. precolumn packed with a silica-based IS containing
anti-carbendazim antibodies, a similar effect of the flow-rate was observed
between 0.2 and 10 ml/min but the recovery decrease was from 75 to 60% [47].
Other studies did not notice any loss when increasing the sample flow rate up to
2-5 ml/min using 4.6-mm i.d. precolumn[31]. The effect of the flow-rate is
certainly more important with 1-mm i.d. precolumns due to the increase of the
linear velocity when reducing the internal diameter of the column.

33.4.5 How to discriminate between specific and non specific

interactions

Specific interactions occur when the analyte is only retained by interactions

provided by the recognition sites of the antibodies. Non-specific interactions can
occur with the other part of the antibody protein or with the sorbent used for
antibody immobilization. Most commonly, hydrophobic and ionic forces are
responsible for undesirable sorption. Silica- and agarose-based sorbent for
immobilizing antibodies have been shown to minimize the interactions gener-
ated by the solid support. Non-specific interactions occur with the handling of
highly hydrophobic analytes which have a tendency to adsorb everywhere. These
non-specific interactions were reduced by adding a small amount of organic
solvent (methanol or acetronitrile) or a detergent such as Tween 20 or Triton
X-100 to the sample before percolation [9,27,29,33,34,42,45,46,56] Discrimina-
tion between specific and non-specific interaction can be easily investigated by
using three different columns, one constituted by the non-bonded sorbent,
another by the sorbent bonded with non-specific antibodies, and the last by the
sorbent bonded with the specific antibodies of interest. Then, it was easy to show
that PAHs were partially retained by an anti-atrazine IS, whereas atrazine,
which is less hydrophobic than PAHs, was not retained by the anti-fluorene IS.
Non-specific interactions were also involved in the retention of PAHs on an
anti-fluorene IS but it was shown that the retention was better due to the
addition of the specific antigen-antibody interactions [42].

33.5 APPLICATIONS

33.5.1 Off-line immunoextraction.

The main advantage of using IS for the handling of liquid samples is that
extraction and clean-up are performed using a single cartridge which, although
the price is higher, shouldd be compared to the use of two cartridges and the
laborious and sometimes difficult use of Florisil sorbents for clean-up. Liquid
samples include serum, plasma, urine, milk and water. Urine is often diluted in
an equal volume of PBS.
Many off-line techniques are in fact used as clean-up extracts from various
matrices. After elution and further evaporation, GC or LC are applied for the

1094
final determination of analytes. The selectivity is the most important feature of
ISs and has been demonstrated for a variety of complex matrixes such as soil,
sediments, sludges, biological and plant tissues or food. The analysis of phenyl-
urea and triazine herbicides in several food samples was shown to be highly
simplified compared to conventional methods [36,37]. PAHs could be deter-
mined in waste sludges and sediments [45]. The method was validated using
certified reference sludges and sediments and the clean-up provided by the
anti-fluorene IS was shown to be better than that provided by conventional silica
clean-up [43].

33.5.2 On-line coupling with chromatographic methods

The first examples of on-line preconcentration used antibodies immobilized on

Sepharose [23,26,57-60]. Because Sepharose was not pressure-resistant it was
necessary to use a second C18 precolumn for the desorption and the re-concen-
tration of analytes from the immunocolumn. This two-precolumn set-up was
used when the desorption step from the IS was achieved with large volumes of
acidified aqueous solutions containing glycin/HC1, formic acid, acetic acid, phos-
phate buffer [18,30,41,50,55,56,61] or a low amount of organic solvent. Relevant
examples have been described in the biological field such as trace determination
of LSD drug and its metabolites in urine [40]. Raw human urine (25 ml) diluted
1:1 with PBS was pumped through a 2.1 mm i.d. protein G immunoaffinity
precolumn with non-covalently immobilized antibodies against LSD and was
analyzed by electrospray-MS after reconcentration onto a packed capillary
trapping column. Since the elimination half-lives of LSD metabolites are longer
than those of the parent LSD, their presence may be detectable for much longer
after the level of LSD has dropped below the limit of detection. Five metabolites
could be detected at concentrations as low as 2.5 ng/l.
When desorption cannot be obtained by aqueous solution without a high
proportion of organic solvent, direct desorption and re-concentration are impos-
sible without modification of the basic on-line set-up. A pump for adding water to
the desorption solution from the IS is then required [59].
The on-line set-up using a single silica-based IS has been widely applied for
the handling of various aqueous samples [12,13,33,38,42-45,61]. In this direct
on-line coupling of the IS with the LC analytical column, the desorption step is
achieved with a solution corresponding to the composition of the mobile phase
used for the LC separation, usually a methanol or acetonitrile-water gradient.
The on-line analysis of phenylureas represented in Fig. 33.3 allowed quantifica-
tion at the 0.1 pg/l level for a sample volume of 50 ml surface water and using a
simple UV diode array detector [13]. No interferences from humic acids and
other analytes were observed. After washing with water, the immunoprecolumn
was directly connected to the analytical column and a water-acetonitrile
gradient allowing the analytical separation was percolated through the system
precolumn-analytical column. In order that the IS should not be in contact with

1095
 (a)

(b)

9 10 20 T(min)

Fig. 33.6. On-line analysis of 2.5 ml of textile effluent diluted with 47.5 ml of PBS using a
precolum packed with (a) the non-selective PRP-1 and (b) the anti-benzidine IS. The 50-ml
sample was spiked with 1 gfl of each analyte: (1) benzidine, (2) 3,3'-dichlorobenzidine, (3)
4-aminobenzene (from Ref. [44]).

too high amounts of acetonitrile for a long time, the connecting valve was
switched when the percentage of acetonitrile was 50%. At each cycle, the IS was
washed with 5 ml of a solution containing 70% methanol and then regenerated
with 6 ml of PBS. The same precolumn could be re-used for more than 50 cycles
including a half of dirty samples. Thanks to this high degree of extraction
selectivity, the sample volume could be reduced. When combining the selectivity
of the IS with the high sensitivity achieved by LC-atmospheric pressure chemical
ionization (APCI)-MS in positive operation mode, detection and confirmation
could be obtained at the low ng/l from preconcentration volume as low as 20 ml
[33,34]. The method was validated through an interlaboratory study with
Aquacheck certified samples.
The on-line coupling of IS to LC is particularly appropriate for the trace
analysis of polar and/or volatile analytes because no evaporation of samples
occur. When sample matrices are very complex such as some industrial effluents,
the detection of polar analytes is hindered by many polar interferences. Owing to
their carcinogenic nature, benzidine and dichlorobenzidine are included in the
US EPA lists and in the European restricted list of "very toxic substances for the
environment". These analytes are still produced in large amounts as
intermediary compounds in the manufacture of dyes and pigments. An
anti-benzidine IS was designed which was able to trap aminobenzene and
related azodyes [26]. Figure 33.6 shows the chromatogram obtained for the
on-line analyses of 2.5 ml of a textile effluent diluted with 47.5 ml of PBS
solution, using a polylmeric precolumn (a) or using the IS precolumn (b). The

1096
selectivity provided by the IS is very obvious for such an industrial waste which
contains many polar interferences co-extracted by the polymer and co-eluted at
the beginning of the chromatogram. Another example deals with the 16 priority
PAHs of the US EPA list which contains some 2-3 rings PAHs that are both
volatile and hydrophobic [46]. Due to their hydrophobicity, the addition of an
organic solvent in the sample before percolation was necessary to avoid
adsorption on containers or connection tubes. Using fluorescence detection, it
was possible to quantify PAHs at 20 ng/l from 20 ml of surface water (regulatory
level for some PAHs in Europe for surface water taken for drinking water
plants).
In order to recover a whole class of compounds, two antibodies have some-
times been mixed in the cartridge bed. Mixed-bed sorbents using antibodies
against chlortoluron and isoproturon have been obtained for recovering the
group of phenylureas [15]. A mixed-bed IS was also tailored for trapping
ampicillin and cloxacillin [62].
Immunoextraction was also coupled to GC although the interfacing between
the immunoextraction and the GC part is not as straightforward as it is with LC.
After trace-enrichment of analytes on a 10 x 3 mm immunoprecolumn packed
with monoclonal anti-atrazine antibodies immobilized onto cellulose, they were
desorbed and re-concentrated on a reversed-phase precolumn packed with a
copolymer by means of an acidic buffer [42]. After clean-up and drying with
nitrogen, desorption and transfer to the GC was done with ethylacetate via an
on-column interface. The selectivity of the system was such that non-selective
flame ionization detection (FID) could be used to detect several triazines in river,
waste water and orange juice. The detection limits for 10 ml samples were 15-25
ng/l for FID and about 1.5 ng/l for nitrogen-phosphorus detection.

33.5.3 Potential for rapid sample throughput with coupling to

micro-LC

The potential for rapid sample handling was demonstrated with the direct
on-line coupling of IS with microLC-UV DAD. The direct on-line set-up was
performed using a 10 x 1 mm precolumn prepacked with a silica-based anti-
phenylureas IS and a 250 x 1 mm analytical column. Quantification limits of 10
ng/l have been obtained from sample volumes of 5 ml with UV-DAD detection
[61]. Figure 33.7 illustrates the high selectivity of the immunoextraction as
compared with the use of the same precolumn packed with C18 silica. Concen-
trations were around 50 ng/l for both chlortoluron and linuron, 150 ng/l for
isoproturon and 200 ng/l for diuron. Since there is no co-extraction of humic
materials, the analysis can be more rapid because usually the first eluted peak is
delayed in order to be after the humic acids peaks. The analysis of diuron used in
antifouling paints could be done in sea water in less than 10 min. This was
possible because the 4 0-um silica particles used for preparing the IS which
allowed a rapid sample flow-rate and because the sample volume is low.

1097
 mAbs

30

20

10

0 10 mi 20 30

Fig. 33.7. Comparison of chromatograms obtained after on-line preconcentration of 5 ml of River

Seine (Paris city) water using a non-selective C, 8 precolumn (top), and a selective anti-phenyl-
ureas immunosorbent (bottom). 3, Chlortoluron; 5, isoproturon; 6, linuron; 9, diuron (from Ref.
[61]).

On-line coupling of immunoextraction with micro-LC has the advantage of

providing high sensitivity. The potential of on-line IS-coupled column packed
capillary LC-ion spray tandem MS was demonstrated for the multi-residue
determination of five f-agonists in bovine urine [17]. Trace-enrichment and pre-
liminary sample clean-up was performed on-line using 25 ml of bovine urine
diluted with 25 ml of PBS and percolated through a 10 x 2 mm i.d. IS precolumn.
The column switching process involved elution and reconcentration onto a mini-
bore 20 x I mm precolumn which was further on-line coupled to a microbore (320
.tm or 1 mm i.d.) analytical column for optimum sensitivity in LC-MS-MS. The
advantage of LC-MS-MS is its high sensitivity and the fact that a complete sepa-
ration is not required so that it can be as rapid as 8 min. Lower levels for quanti-
fication were in the range 10-50 ng/l. The advantage of immunoextraction as a
good sample preparation method for LC/MS/MS was recently emphasized [63].

33.6 CONCLUSION

The coupling of immunoaffinity SPE in off-line or on-line procedures is a

powerful technique for the determination and easy quantification at low level of
many organic compounds in various complex samples. The sample handling step
is greatly simplified. The selectivity in the extraction step also allows easier
identification of analytes of interest and their confirmation is reinforced since

1098
trapping occurs on the basis of a structure recognition. Moreover, a clear
baseline enhances the overall sensitivity of the analysis and allows the handling
of a lower sample in comparison with non-selective extraction procedures. The
on-line combination of this selective preconcentration with a separation using a
very short micro-analytical column and MS detection is a good solution for rapid
analysis.

REFERENCES
1 M.-C. Hennion, J. Chromatogr. A, 856 (1999) 3.
2 E. Hogendoorn and P. Van Zoonen, J. Chromatogr.A, 892 (2000) 435.
3 N. Delaunay-Bertoncini, V. Pichon and M.-C. Hennion LC-GC Europe, 14/3 (2001)
162.
4 N. Delaunay, V. Pichon and M. C. Hennion, J. Chromatogr.B, 745 (2000) 15.
5 D.S. Hage and N.A. Nelson, Anal. Chem., 71 (2001) 199A.
6 V. Pichon, M. Bouzige, C. Miege and M.-C. Hennion, Trends Anal. Chem., 18 (3)
(1999) 219.
7 M.P. Marco, S.G. Gee and B.D. Hammock, Trends Anal. Chem., 14 (7) (1995) 415.
8 M. Worterg, K. Camman, K. Strupat and F. Hillenkamp, Fresenius' J. Anal. Chem.,
348 (1994) 240.
9 M. Cichna, P. Markl, D. D. Knopp, R. Niessner, Chem. Mater., 9 (1997) 2640.
10 M. Cichna, D.D. Knopp and R. Niessner, Anal. Chim. Acta, 339 (1997) 241.
11 V. Pichon, L. Chen, M.-C. Hennion, R. Daniel, A. Martel, F. Le Goffic, J. Abian and D.
Barcelo, Anal. Chem., 67 (1995) 2451.
12 V. Pichon, L. Chen, N. Durand, F. Le Goffic and M.-C. Hennion, J. Chromatogr. A,
725 (1996) 107.
13 V. Pichon, H. Rogniaux, N. Fisher-Durand, S. Ben Rejeb, F. Le Goffic and M.-C.
Hennion, Chromatographia,45 (1997) 289.
14 N. Delaunay-Bertoncini, V. Pichon and M.-C. Hennion, Chromatographia,53 (2001)
S-224.
15 V. Pichon, M. Bouzige and M.-C. Hennion, Anal. Chim. Acta, 376 (1998) 21.
16 J.F. Lawrence and C. Menard, J. Chromatogr.B, 696 (1997) 291.
17 J. Cai and J. Henion, J. Chromatogr.B, 691 (1997) 357.
18 H. Hooijerink, R. Schilt, E.O. van Bennekom and F.A. Huf, J. Chromatogr. B, 660
(1994) 303.
19 Th. Gude, A. Preiss and K. Rubach, J. Chromatogr. B, 673 (1995) 197.
20 R. Bagnati, M.G. Castelli, L. Airoldi, M. Paleologo-Oriundi, A. Ubaldi and R. Fanelli,
J. Chromatogr., 527 (1990) 267.
21 A.L. Savage, S.H. Sarijo and J. Baird, Anal. Chim. Acta, 375 (1998) 1.
22 E. Horne, T. Coyle, M. O'Keefe and D.L. Brandon, Analyst, 124 (1999) 87.
23 A. Kussak, B. Andersson and K. Andersson, J. Chromatogr.A, 708 (1995) 55.
24 A. Kussak, B. Andersson and K. Andersson, J. Chromatogr. B, 672 (1995) 253.
25 A. Farjam, G.J. De Jong, R.W. Frei, U.A.Th. Brinkman, W. Haasnoot, A.R.M.
Hamers, R. Schilt and F.A. Huf, J. Chromatogr., 452 (1988) 419.
26 W. Haasnoot, R. Schilt, A.R.M. Hamer, F.A. Huf, Ai. Farjam, R.W. Frei and U.A.Th.
Brinkman, J. Chromatogr., 489 (1989) 157.
27 C. Rivasseau and M.-C. Hennion, Anal. Chim. Acta, 394 (1999) 243.
28 L. Ten-Hage, N. Delaunay, V. Pichon, A. Coute, S. Puiseux-Dao and J. Turquet,
Toxicon, 38 (2000) 1043
29 N. Delaunay, V. Pichon, J.-P. Le Caer and M.-C. Hennion, Anal. Chim. Acta, 407
(2000) 173.

1099
30 G.S. Rule, V. Mordehal and J. Henion, Anal. Chem., 66 (1994) 230.
31 V. Pichon, L. Chen and M.-C. Hennion, Anal. Chim. Acta, 311 (1995) 429.
32 A. Martin-Esteban, P. Fernandez, D. Stevenson and C. Camara, Analyst, 122 (1997)
1113.
33 I. Ferrer, M.-C. Hennion and D. Barcelo, Anal. Chem., 69 (1997) 4508.
34 I. Ferrer, V. Pichon, M.-C. Hennion and D. Barcelo, J. Chromatogr.A, 777 (1997) 91.
35 S.J. Shahtaheri, M.F. Katmeh, P. Kwasowski and D. Stevenson, J. Chromatogr. A,
697 (1995) 131.
36 J.F. Lawrence, C. Menard, M.-C. Hennion, V. Pichon, F. Le Goffic and N. Durand, J.
Chromatogr.A, 732 (1996) 277.
37 J.F. Lawrence, C. Menard, M.-C. Hennion, V. Pichon, F. Le Goffic and N. Durand, J.
Chromatogr.A, 752 (1996) 147.
38 V. Pichon, E. Aulard-Macler, H. Oubihi, P. Sassiat, M.-C. Hennion and M. Caude,
Chromatographia,46, 9/10 (1997) 529.
39 J. Dalluge, T. Hankemeier, R.J.J. Vreuls and U.A.Th. Brinkman, J. Chromatogr. A,
830 (1998) 267.
40 J. Cai and J.D. Henion, Anal. Chem., 68 (1996) 72.
41 D.H. Thomas, M. Beck-Westermeyer and D.S. Hage, Anal. Chem., 66 (1994) 3823.
42 M. Bouzige, V. Pichon and M.-C. Hennion, J. Chromatogr.A, 823 (1998) 197.
43 S. Perez, I. Ferrer, M.-C. Hennion and D. Barcelo, Anal. Chem., 70 (1998) 4996.
44 M. Bouzige, P. Legeay, V. Pichon and M.-C. Hennion, J. Chromatogr.A, 846 (1999)
317.
45 C. Miege, M. Bouzige, S. Nicol, J. Dugay, V. Pichon and M.-C. Hennion, J. Chroma-
togr. A, 859 (1999) 29.
46 M. Bouzige, V. Pichon and M.-C. Hennion, Environ. Sci. Technol., 33 (1999) 1916.
47 S. Perez and D. Barcelo, Analyst, 125 (2000) 1273.
48 R.K. Betsen-Farmen, I.V. Botnen, H. Noto, J. Jacob and S. Ovrebo, Int. Arch. Occup.
Environ. Health, 72 (1999) 161.
49 S.D. Thomas and Q.X. Li, Environ. Sci. Technol., 34 (2000) 2649.
50 S. Ouyang, Y. Xu and Y.H. Chen, Anal. Chem., 70 (1998) 931.
51 S.J. Shahtaheri, P. Kwasowski and D. Stevenson, Chromatographia,47, (1998) 453.
52 A. Martin-Esteban, P. Kwasowski and D. Stevenson, Chromatographia,45 (1997)
364.
53 T.M. Phillips and J.M. Krum, J. Chromatogr.B, 715 (1998) 55.
54 N. Delaunay-Bertoncini, V. Pichon and M.-C. Hennion, submitted.
55 D.H. Thomas, V. Lopez-Avila, L.D. Betowski and J. Van Emon, J. Chromatogr.A,
724 (1996) 207.
56 J.G. Rollag, M. Beck-Westermeyer and D.S. Hage, Anal. Chem., 68 (1996) 3631.
57 A. Farjam, A.E. Brugman, A. Soldaat, P. Timmerman, H. Lingeman, G.J. De Jong,
R.W. Frei and U.A.Th. Brinkman, Chromatographia,31 (1991) 469.
58 A. Farjam, R. de Vries, H. Lingeman, R.W. Frei and U.A.Th. Brinkman, Int. J.
Environ. Anal. Chem., 44 (1991) 175.
59 A. Farjam, C. van der Merbel, H. Lingeman, R.W. Frei and U.A.Th. Brinkman, Int. J.
Enuiron.Anal. Chem., 45 (1991) 73.
60 A. Farjam, N.C. van der Merbel, A.A. Nieman, H. Lingeman and U.A.Th. Brinkman,
J. Chromatogr., 589 (1992) 141.
61 E. Schoenzetter, V. Pichon, D. Thi6baut, A. Fernandez-Alba and M.-C. Hennion, J.
Microcolumn Sep., 12/5 (2000) 316.
62 R. Dietrich, E. Usleber and E. Martlbauer, Analyst, 123 (1998) 2749.
63 J. Henion, E. Brewer and G. Rule, Anal. Chem., 70 (1998) 650A.

1100
Abbreviations
AA acrylic acid
ABS aqueous biphasic system
ACN acetonitrile
AFM atomic force microscopy
AMANDA ammonia measurement by annular denuder sampling with
on-line analysis
AMPS 2-acrylamido-2-methyl-1- propane sulfonic acid
AOT di-2-ethylhexyl sodium sulfosuccinate
APHA American Public Health Association
ASE accelerated solvent extraction
BC black carbon
BOSS Brigham Young University Organic Sampling System
BSA bovine serum albumin
BTEX benzene, toluene, ethylbenzene, xylenes
CBA ethylcarboxylic acid
CBG Coomassie Blue G
CE capillary electrophoresis
CEC capillary electrochromatography
CEPPWD circular end parallel plate wet denuder
CH cyclohexyl
CHD cyclohexanedione
CIF charcoal impregnated filter
CL chemiluminescence
CME capillary microextraction
CMS carbon molecular sieves
CPE cloud-point extraction
CVD chemical vapor deposition
CW Carbowax
DAD diode array detector
DBT dibenzothiophene
DCB 1,4-dichlorobenzene
DCCC droplet countercurrent chromatography
DI direct immersion
DM dialysis membrane
DME dynamic membrane extraction
DMG dimethylglyoxime

1101
DNPH 2,4-dinitrophenylhydrazine
DS diffusion scrubber
DS- dodecylsulfate anion
DTA diethylenetriamine
DVB divinylbenzene
EC elemental carbon
ECD electron capture detector
ECN Energieonderzoek Centrum Nederland (Netherlands Energy
Research Institute) (Ch. 5)
ED electrochemical detector
EDMA ethylene glycol dimethacrylate
EDTA ethylenediaminetetraacetic acid
ELISA enzyme-linked immunosorbent assay
EPA Environmental Protection Agency (of the U.S.A.)
ePTFE expanded poly(tetrafluoroethylene)
ESR electron spin resonance
FIA flow injection analysis
FID flame ionization detector
FL fluorescence detector
FMA fluorescein mercuric acetate
FP filter pack
FPD flame photometric detector
FSA Fluorosililic acid
FTIR Fourier transform infra red
FV flash volatilization
GAMS gas and aerosol monitoring system
GC gas chromatography; also graphitic carbon (Chap. 6)
GC/MS gas chromatography/mass spectrometry
GC-FID gas chromatography-flame ionization detection
GCB graphitized carbon black
GDMA glycerol dimethacrylate
GFF glycine-L-phenylalanine-L-phenylalanine
GMMA glycerol monomethacrylate
HF: Hydrofluoric acid
HMDS hexamethyldisilazane
HMHP hydroxymethyl hydroperoxide
HMX 1,3,5- 7-tetrahydro -1,3,5, 7-tetranitrotetrazine
HPLC high performance liquid chromatography
HPLC/UV high performance liquid chromatography/ultraviolet detector
HRP horseradish peroxidase
HS headspace
H.S. hydrophobic surroundings
HV high voltage
i.d. inside diameter

1102
IAE immunoaffinity extraction
IC ion chromatography
IC-CD-UV-MS ion chromatography-suppressed conductivity-ultraviolet
detection-mass spectrometry
ICP-MS induction coupled plasma mass spectrometer
IMPROVE Interagency Monitoring of Protected Visual Environments
IR infra red
ISRP internal surface reversed-phase
LC liquid chromatography
LC/MS liquid chromatography/mass spectrometry
LCST lower critical solution temperature
LCW liquid core waveguide
LED light emitting diode
LLE liquid-liquid extraction
LOD limit of detection
LPM liters per minute
LPME liquid-phase microextraction
LSD lysergic acid diethylamide
LSE liquidsolid extraction
MAA methacrylic acid
MAE microwave assisted extraction
MALDI-MS matrix-assisted laser desorption ionization mass spectrometry
4-MAP 4-methylacetophenone
MASE microwave-assisted solvent extraction
MB + methylene blue cation
MB.DS methylene blue-dodecylsulfate ion pair (Ch
MBAA N,N-methylenebis(acrylamide)
MBTH 3-methyl-2-benzothiazoline hydrazone
MCME monolithic capillary microextraction
MDA methylenedioxy amphetamine
MDEA methylenedioxy ethylamphetamine
MDMA methylenedioxy methamphetamine
ME micellar extraction
MECC micellar electrokinetic capillary chromatography
MEKC micellar electrokinetic chromatography
MESI membrane extraction with a sorbent interface
MF microfiltration
MFP mixed-functional phase
MFV methanol-fueled vehicle
MHP methyl hydroperoxide
MI molecular imprinting
MIMS membrane introduction mass spectrometry
MIP molecularly imprinted polymer
MMAD mass median aerodynamic diameter

1103
MMLLE microporous membrane liquid-liquid extraction
MS mass spectrometry or mass spectrometer
MTBE methyl tert-butyl ether
MTMS methyltrimethoxysilane
MW molecular weight
NAAQS National Ambient Air Quality Standard
NCAR National Center for Atmospheric Research
NMDS nation membrane diffusion scrubber
NMR nuclear magnetic resonance
NPD nitrogen-phosphorus detector
o.d. outside diameter
OC organic carbon
ODS octadecylsilane
OTCME open tubular capillary microextraction
PA polyacrylate
PAH polynuclear aromatic hydrocarbons; also polycyclic aromatic
hydrocarbons
PAR 4-(2-pyridylazo resorcinol)
PBA phenylboronic acid
PCBs polychlorobiphenyls
PCE perchloroethylne
PCS particle collection system
PDMS poly(dimethylsiloxane)
PEEK polyether ether ketone
PEG poly(ethylene glycol)
PES polyethersulfone
PFA Teflon perfluoroalkoxy Teflon
PFE pressurized fluid extraction
PGC porous graphitic carbon
PHCs petroleum hydrocarbons
PIPAAm poly(N-isopropylacrylamide)
PMDS porous membrane diffusion scrubber
PME polymeric membrane extraction
PMHS poly(methylhydrosiloxane)
PMMA poly(methyl methacrylate)
PMT photomultiplier tube
PNIPAAm poly(N-isopropylacrylamide)
PO-CL peroxyoxalate chemiluminescence
ppb parts per billion
ppbv parts per billion by volume
PPD parallel plate denuder
ppm parts per million
ppmv parts per million by volume
PPO poly(2,6-dimethyl-1,4-phenylene) oxide

1104
ppq parts per quadrillion
ppt parts per trillion
pptv parts per trillion by volume
PPWD parallel plate wet denuder
PPY polypyrrole
(PPPY) poly(N-phenylpyrrole)
PS poly(styrene)
PS-DVB Poly(styrene-divinylbenzene)
psi pounds per square inch
PTFE poly(tetrafluoroethylene)
PVC polyvinyl chloride
PVDF polyvinylidene fluoride
RAM restricted access material
RAMS real-time ambient mass sampler
RDX 1,3,5-hexahydro-1,3,5-trinitrotriazine
RH relative humidity
RO reverse osmosis
ROMP ring opening metathesis polymerization
rpm revolutions per minute
RSD relative standard deviation
RWAD rotating wet annular denuder
SIN signal to noise ratio
SAX trimethyl aminopropyl
SC-CO2 supercritical carbon dioxide
SCE saturated Calomel electrode
SCX benzenesulfonic acid
SEM scanning electron micrograph also microscopy
SFC supercritical fluid chromatography
SFE supercritical fluid extraction
SID surface ionization detector
SJAC steam jet aerosol collector
SLM supported liquid membrane extraction (Chap. 23)
SLME supported liquid membrane extraction (Chap. 32)
SLPM standard liters per minute
SMM surface modifying macromolecules
SPE solid phase extraction
SPME solid phase microextraction
SPS semipermeable surface
SS stainless steel
STEM scanning transmission electron microscopy
TC total carbon
TCE trichloroethylene
TCP 2,4-trichlorophenol
TDLAS tunable diode laser absorption spectrometer

1105
TDM therapeutic drug monitoring
Teflon AF Teflon amorphous fluoropolymer
TEOM tapered element oscillating microbalance
TEOS tetraethoxysilane
THF tetrahydrofuran
TMB tetramethylbenzidine
TMOS tetramethoxysilane
TNT 2,4,6-trinitrotoluene
TOFMS time-of-flight mass spectrometry
TPH total petroleum hydrocarbon
TPR templated resin
TPS thermoseparating polymer system
TRP temperature-responsive polymer
TSP total suspended particulate matter
UF ultrafiltration
UV ultraviolet
UPW ultrapure water
V-65 2,2'-azobis(2,4-dimethylvaleronitrile)
VC vinyl chloride
VOCs volatile organic compounds
WAD wet annular denuder
WEDD wet effluent diffusion denuders
XPS X-ray photoelectron spectroscopy

1106
Index
abietic acids, 648
absorption, 20
absorptive passive methods, 54
accelerated solvent extraction, 885, 945
acceptor flow rate influence, 519
acceptor phase, 754, 757-761
Accurel, 105, 113
acetaldehyde, 9, 124, 294
acetaminophen, 622
acetic acid, 121, 134, 147, 151, 294, 647
acetone, 294, 932
acetonitrile, 123, 292, 618, 1092, 1094
acetyl groups, 739
acetylacetone method, 11
American Conference of Governmental
Industrial Hygienists (ACGIH), 2, 4
acid aerosols, 161, 205
- neutralization of, 205
acid chlorides, 610
acid gases, 148
acid sulfates, 168
acidic compounds, 724
acidic herbicides, 747
acidic/basic gases, 205
acoustic levitation, 325
acrolein, 9, 632
actinides, 209
activated carbon, 342, 1005, 1032
activated carbon fibers, 1006, 1007
activated charcoal, 679, 681, 922
activation energy, 616
activity coefficients, 284, 295
acyl chlorides, 638
adenosine, 632
adiabatic expansion, 193
adsorbents, 551
adsorption behavior, 1009, 1011
adsorption chromatography, 738
adsorption isotherm, 752
adsorptive passive methods, 54
aeration chamber, 678, 680
aeration extracts, 678
aerosol acidity measurement, 205
aerosol carbon, 167
- measurement of, 172
aerosol collection, applications, 200
aerosol composition measurements, 163,
208
aerosol mass spectrometry, 169-171
aerosol metal concentrations by real-
time emission spectrometry, 171
aerosol sampling, 169, 180
aerosols, 21, 97, 98
- link with gases, 162
- particles, 161
- trace metals in, 171
aethalometer, 172, 174
affinity chromatography, 409
affinity sorbent extraction, 814
aflatoxins, 26
agglomeration, 193
aggregation pheromones, 672
agitation
- perfect, 439
- practical, 440
- selection of technique, 445
agrochemicals, 906
AHMT method, 11
Air & Waste Management Association, 5
air bubbles, 206
air/liquid separators, 185, 203
air monitoring, 538
1107
air pollution, 162 4-aminosulfonyl-7-fluoro-2,1,3-benzoxad
air sampling pumps, personal, 681 iazole, 625
air segmented liquid, 123 ammonia, 98, 103, 115, 120, 136, 140,
airborne particles, 163 141, 148, 153, 1011
alarm pheromones, 672 ammonium, 196, 198
alcohol oxidase, 124 ammonium acetate, 106
alcohols, 635, 650, 656, 753 ammonium iodide, 644
aldehydes, 9, 26 ammonium molybdate, 166
algal pheromones, 691 ammonium nitrate, 97
Aliquat-336, 508 ammonium sulfate, 97, 164
alkali metal ion adducts, 170 amperometry, 621
n-alkanes, 9 amphetamine, 650, 927
alkyl bromides, 610 analyte binding, 241, 242
alkyl chloroformates, 648 analyte collection, 592
alkyl hydroperoxide, 112 analyte distribution in natural systems,
alkylating anticancer drugs, 641 471
alkylation, 647 analyte enrichment, 757
alkylboronylation, 652 analyte hydrophobicity, 514
alkylchloroformates, 610 analyte trapping, 512
1-alkylthiol 2-alkyl substituted analyte-matrix interactions, 560
isoindoles (AAI), 612 analytical distillation procedures, 702
alkyltin, 941 analytical instrumentation,
allergens, 1, 24 interferences, 422
alpha track detectors, 17 analytical process steps, 253
alumina, 347, 738, 763 anandamide, 635
alveolar region, 205 anion exchangers, 739
Alzheimer's disease, 614 ANSYL hydrazones., 628
ambient aerosol, 205 antibodies, 741
- analysis of, 183 - affinity of, 1090
ambient air measurement, 200 - cross-reactivity, 741
ambient outdoor air, 204 - immobilization, 1083
American Public Health Association - mimics, 409
methods, 899 antidepressants, 928
American Society for Testing and antifouling compounds, 747
Materials (ASTM), 924 antigen-antibody interaction, 741
amidation, 610 antioxidants, 622
amines, 115, 120, 284, 612, 630, 638, 643 ants, 689
amino acids, 612, 614, 616, 622, 625, aphrodisiac pheromones, 672
638, 642, 653 aquatic colloids, 228
aminoacridines, 631 aqueous and organic liquids, 532
2-aminobenzoic acids, 630 aqueous environmental matrices,
2-aminonaphthalene trisulphone, 655 definition, 219
aminoglycan, 629 aqueous matrix, 468
aminoglycoside, 639 aqueous phase, 541
aminopropyl silica, 738 arable layers, 74

1108
arachadonyl ethanolamide, 635 azodyes, 741
arginine, 648
Aroclor, 903, 913 back extraction, 299
aromas, 699 bacteria, 24
aromatic dialdehydes, 612 balanced pressure sampling, 730
aromatic hydrocarbons, 752, 754 basic compounds, 724
aromatic sulfonates, 747 batch equilibrium, 733
aromatic sulfonic acids, 740 batch techniques, 267
aromatics, 294 bathometers, 36
arsenic speciation, 232 Beer's law behavior, 131
arson investigation, 921 benzaldehyde, 294
asbestos, 17 benzene, 23, 128, 293
ascending/descending drop, 314 benzo(a)pyrene, 13
ASHRAE, 22 benzodiazepines, 928
asthma, 161 benzoic acid, 294, 630
ASTM methods, 13 benzophenones, 928
asymmetric membranes, 105, 126 Berthelot reaction, 129
atmospheric aerosols, 163 BET method, 236
atmospheric chemistry, 162 binding energies, 736
atomic absorption, 16 bioaccumulation, 761
atomic force microscopy, 994 bioaerosols, 1, 24, 202, 203, 205
attractants, 670 bioanalyses, 244, 845
auto exhaust, 125 bioavailability, 241
autoanalyzer, 307 biodesulfurization of dibenzothiophene,
automated evaporative concentration 913
system, 728 biological activity of insect pheromones,
automated headspace, 290 669
automated in-tube solid-phase biological agents, 24
microextraction, 428 biological fluids, 284, 286, 293, 659
automated liquid-liquid extraction, 727 - volatiles in, 292
automated monitoring, 727 biological matrices, 293, 532, 548
automated processes, future trends, 865 biomedical analyses, 247, 524
automated sequential trace enrichment biotransformation, 647
of dialysates, 861 - pathways, 913
automated solid-phase extraction, 742 bisphenols, 635
automated SPE-GC coupling, 381 bisphosphonates, 618
automatic sampling systems, 52 bitumen, 913
automation, 377, 464, 729, 760 black carbon, 172
- applications, 844 blood, 926, 929-931
- for bioanalysis, 845 - alcohol, 279, 287, 289, 930
- for clinical chemistry, 861 - bone and tissue, 548
- strategies, 839 - gases, 531, 548
automobile exhaust, 162 Bodman's sampler, 38
autosamplers, 454 bonding density, 1088, 1090
- for GC, 430 boreholes, 42

1109
bound residues, 907
boundary layer, 268, 270
bovine serum albumin, 204
brain dialysate, 617
breakthrough curves, 736
breakthrough volumes, 359, 724, 734,
736, 743, 745, 1090
- determination, 360
- methods for estimating, 361
- models for prediction, 361
breath, 792
bromide, 196
bromoacetonitrile, 652
bronchoconstriction, 205
brown algae, 688
brown-field sites, 897
bubble gas, 84
bubbler/impinger, 98
buffers, 798
bulk liquid membranes, 505
bulk materials, 796
busulfan, 641
butanal, 294
butyltin, 660
butyric acid, 121, 147
C18 silica, 739
cadaverine, 621
caffeine, 930
Calberla stain, 26
calcium, 198
calibration, 497, 751
- selection of method, 458
cannabinoids, 635, 932
capillary, 130
capillary electrophoresis, 308, 967
capillary gas chromatography-mass
spectrometry, 625
capillary isoelectric focusing, 978
capillary isotachophoresis is, 973
capillary scale analyzers, 114
capillary zone electrophoresis (CZE),
609, 611,613, 617, 621, 631, 638
capsules, 797
carbamate pesticides, 747
1110
carbon-based extraction materials, 1032
carbon-based ion exchange fibers, 1017
carbon-based sorbents, 739
carbon content percentage, 737
carbon dioxide (C02), 22
carbon measurement, 174
carbon molecular sieve, 1034
carbon monoxide (CO), 22
carbonaceous particles, 172
carbonization in a soot-forming flame,
170
carbonyls, 610, 632, 639, 643
- as analytes and reagents, 650
carboxene, 751
carboxylic acids, 420, 610, 634, 635, 640,
643, 653
carboxylic groups, 739
cardiopulmonary and lung cancer
related mortality, 161
carrier-mediated transport, 507
carryover, 300, 743
cartridges, 343, 346, 360, 362, 732, 733,
738, 741, 746, 851
cation exchange capacity, 230
cation exchangers, 739, 1014
Celgard, 105, 123
CEN methods, 733
centrifugation, 163
ceramic honeycomb denuder, 164
certified reference material, 459, 940
charcoal filter, 236, 680
charged particle, 176
chemical activation, 1008
chemical cytometry, 617, 636
chemically bonded reversed phase silica,
737
chemically bonded sorbents, 349
chemically modified resins, 739
chemically nonspecific particle
measurements, 161
chemiluminescence, 113, 166, 635
chemoluminescence, 19-21
chiral separations, 658
chloracetaldehyde, 632
chloride, 198
chlorinated solvents, 895
chloroacetaldehyde, 632
chlorodifluoroacetic anhydride, 649
chloromethylanthracene, 634
chlorophenoxy acetic acid herbicides, 647
cholesterol, 124
Chromasil, 119
Chromosorb, 13
chromotropic acid method, 11
chrysene, 13
chrysolite, 17
cigarette smoke, 102
circular end parallel plate denuder, 101,
146
citric acid, 147
class-selective trapping, 1087
clay, particle size distribution of, 233
cleanup, 297, 723, 738, 742, 756, 758,
762, 763, 817, 878, 968
clinical chemistry, automation for, 861
closed loop stripping, 691
cloud chamber, 193
coagulation, 162
coated ion exchange fibers, 1014
coatings, 797
cocaine odor, 932
co-current chromatography, 314
coefficient of haze, 174
co-elution, 762
co-extractives, 874
coiled configuration, 100
coke oven gas, 118
cold finger, 416
cold trap, 285
collection efficiencies, 105, 121
collection efficiency, 106, 129, 178
colony forming units, 27
colorimetric analysis, 20
colorimetry, 19, 898
colorings, 222, 798
column switching methods, 763
combined hot water extraction-SPME,
578
T M
CombiPal autosampler, 430
commercial devices, 429
competition process, 1089, 1090
composite membranes, 990
composite sample, 75
compost, 219
computer-aided strategies for SPE
method development, 375
concentration enrichment, 516
condensation, 162
condensation nuclei counters, 193
conditioning step, 373
conductivity of an aqueous solution, 225
constant heating time, 290
contact SPME, 689
contamination, 97
continuous collections, 176, 680
continuous LLE, 726
continuous phase, 298
controlled-release binding agents, 798
controlled toxicological studies, 162
convection, 305
convective-diffusive mass transfer, 298,
302
convolution model of extraction, 264
cooking bags, 681
cooling, 494
Coomassie Blue G, 204
copper acetate, 126
copper diethyldithiocarbamate, 126
corers, 81
corona discharge, 178
coupled-column techniques, 378
Cour traps, 26
creams/ointments/gels, 795
p-cresol, 111
critical limits, 724
cross-reactivity, 1087
cryogenic condensation, 730
cryogenic focusing, 685, 731
cryogenic oven, 285
cryogenic sampler, 82
cryogenic trapping, 692, 730, 953
CTC Analytics, 430
cumene hydroperoxide, 135
cuticular pheromones, 689
cyanide, 118, 932

1111
cyanopropyl, 738 1,4-dichlorobenzene), 137
1-cyano-2-substituted-benz[f]isoindoles, dichlorofluoroacetyl, 649
615 dienophiles, 655
cyclohexanedione, 106 diethyl sulphate, 647
cyclohexyl, 738 diethyldithiocarbamate, 641
cyclone, 183 diethylhexyl phosphoric acid, 509
cyclophosphamide, 632 diffusion, 181, 759
cystine, 623 diffusion-based calibration, 269
Czapla-1 sampler, 41 diffusion coefficient, 98, 271, 412
diffusion constant, 534
dansyl chloride, 621 diffusion denuders, 98, 99, 102, 163,
dansyl derivatives, 621 164, 169, 205
dansyl hydrazine, 627, 632 diffusion scrubbers, 104
dansyl hydrazone, 632 diffusiophoresis, 144
decreased pulmonary function, 161 diffusive mass transfer, 302
defensive compounds, 677 dihydrorhodamine, 616
defensive secretions, 672 dimethamphetamine, 927
deisopropylatrazine, 737, 738 1,3-dimethyl-2-imidazolidinone, 293
demethylation, 618 dimethyl mercury, 659
dendrimers in solution-mediated dimethyl sulphate, 647
extraction, 1026 dimethylamine, 116
density, 221 DIN methods, 733
- of solids, 229 2,4-dinitrofluorobenzene, 653
denuder liquid considerations for IC 2,4-dinitrophenylhydrazine, 124, 639
coupling, 151 diol groups, 740
denuders, 3, 16 dioxane, 283
deproteinization, 800 direct extraction procedures, 712
derivatization, 10, 419, 686, 727, 752, 954 direct immersion SPME, 249, 810
- combined with extraction, 274 direct insertion membrane probe, 537,
- selection of reagent, 443 759
desorption, 1092, 1095 direct on-line coupling, 1095
- conditions, optimisation, 447 direct SPME, 750
desorption chemical direct trapping, 512
ionization-membrane inlet mass directional heterogeneity, 65
spectrometry, 541 directional variability, 66
detection, 943 discrete sample, 34
detergent, 1094 discs, 346, 360, 362, 733
determination of D and L-Leucine, 651 dispersed phase, 298
developmental biology, 617 displacement, 751
dextran ladder, 631 displacement chromatography, 734
dialysis, 483, 503, 761, 818, 825, 878, dissolution, initial rate of, 319
882, 978 dissolved organic carbon, 226
1,2-diamino-4,5-methylene-dioxy- dissolved organic matter, 249
benzene, 628 distribution coefficient, 224, 227, 749,
diazepam, 929 750, 751, 757, 759

1112
distribution constant
- calculation of, 454
- prediction of, 438
distribution ratio, 297, 533, 726
disulphide, 627
dithiocarbamates, 295
dithiothreitol, 623
DNA analysis, 920
DNPH, 10
DNPH impregnated cartridges, 91
Donnan exclusion, 106
donor-controlled extraction, 511
donor phase, 754, 757, 759
dosimeter tubes, 5
double-tube samplers, 69
doughy materials, 67, 68
dredged material, 219
dredges, 81
drinking water, 219
drop dissolution in the aqueous phase,
318
drop-bottle system, 40
droplet countercurrent
chromatography, 314
drops, 129, 314
drug binding studies, 472
drug-receptor binding, 243
drugs, 926, 930
- illicit samples, 792
drying step, 375
dual channel ion chromatograph, 198
dual wavelength absorbance detection,
132
Dufours gland, 689
Durham sampler, 26
dynamic (flowing) extractions, 592
dynamic headspace, 279, 287, 730
- analysis, 705
- sampling, 670
dynamic LPME, 322
dynamic systems, 685
dynamics of release, 684
eicosanoids, 654
Einstein diffusion equation, 316
electret detectors, 17
electric field, 176
electrochemical detection, 642
electrodialysis, 485
electron capture atmospheric pressure
chemical ionization, 653
electron capture detector, 6, 13
electron energy loss spectroscopy, 170
electron microscopy, 168
electron spin resonance, 996
electronic interactions, 742
electrophoresis, 114, 609
electropolished canisters, 87
electrospray ionization, 639, 646, 656
electrostatic aerosol analyzer, 178
electrostatic and thermal precipitation,
163
electrostatic attraction, 181
electrostatic collection, 176
electrostatic precipitators, 176
elixirs, 795
elution, 369, 735, 736, 745, 1092, 1093
emulsifying agents, 799
emulsion liquid membrane, 505
emulsions, 300, 726, 727
enantiomer, 651
end-capping, 737
endocrine disruption, 14
endotoxins, 24
enrichment factor, 299, 511, 723
enrichment techniques, 950
environmental analysis, 244, 294, 525
environmental hazards, 727
environmental samples, 532
environmental tobacco smoke, 1, 23,
201
enzymatic reactions, 615
ePTFE, 113, 126
equilibration, 685
equilibration time, 460, 683
equilibrium conditions, 395
equilibrium dialysis, 244
equilibrium partitioning, 247
equilibrium principles, 256
essential oil, 702
1113
esterification, 610 FeCl 3, 129, 130
estradiol, 649 fenfluramine, 650
estrogens, 650, 741, 747, 748 fermentation, 540, 545
estrone, 649 - volatiles, 686
ethanethiol, 644 fermenting fruit, 691
ethanol, 293, 647, 930, 1092 fermentors, 531
- in blood, 292, 293 ferrocene, 656
ethyl acetate, 287, 745 fiber SPME, 810
ethylamine, 116 fibre assemblies, 429
ethylation, 957 fibre coatings, 408, 441
ethylene oxide, 287 fibre conditioners, 431
European Pharmacopoeia, 293 fibre preparation, 414
evaporation, 162, 532, 534 field analysis, 722, 723, 727, 732, 746,
excimer, 621 753, 760
exhaustive electromigration injection, 134 field portable SPME-FastGC, 431
exhaustive extraction, 255, 257, 487, field TWA sampling using the modified
733, 748 SPME device, 407
expanded PTFE, 105 field-flow fractionation, 222, 234
experimental determination of retention film/drop sampler, 135
factors, 366 films, 129, 132
experimental parameters affecting filter-based automated particle
extraction efficiency, 415 collection system, 183
explosives, 895, 899, 905, 924 filter collection, 173
extraction, 255, 723, 944 filters, 98
- combined with derivatisation, 274 filtration, 98, 163, 180, 482, 722, 733,
- conditions for heterogeneous 848
samples, 458 fine particles, 98, 161
- conditions, optimization of, 265, 456 fingerprinting, 901
- convolution model, 264 flame ionization detector, 6, 9, 175
- efficiency, 224, 511 flame photometric detector, 13, 164
- kinetics, 750 flash vaporization-flame photometric
- mode, selection of, 444 detection (FV-FPD), 164
- of solids, 260 flat sheet membranes, 481
- parameters, 439, 724 flattened glass coil, 203
- phases, 751 flavor, 699
- rate law, 303 flavorings, 797
- syringe, 522 flophemesyl, 646
- time, 452, 453 Florisil, 738, 763
- with coated fibre, 397 flow injection analysis, 116, 307
- with in-tube SPME, 401 flow proportional composite sample, 34
exuvia, 677 flow rate, 515, 1093
flow-through equilibrium, 733
fabric denuders, 138 flow-through techniques, 266, 446, 683
faeces, 792 9-fluorenylmethylchloroformate, 617
falling drops, 130, 132 fluorescamine, 205

1114
fluorescein mercuric acetate, 119
fluorescence, 12, 14, 20, 109, 119, 612,
621, 631, 635
- quantum yield, 616
- tagging, 205
fluoresceneisothiocyanate, 619
1-fluoro-2,4-dinitrophenyl-5-L-alanine,
658
fluorogenic reactions between
nucleosides and aldehydes, 633
fluorogenic reagent, 616
4-fluoro-7-nitro-2,1,3-benzoxadiazole,
620
fluorophore, 619
fluorophoric reagent, 616
FMA, 130
fodder plants, 79
food analysis, 609, 869
- and industrial analysis, 526
- and packaging industries, 294
forensic science, 919
forest soil, 75
formaldehyde, 1, 9, 10, 14, 106, 124,
421, 651
formic acid, 121, 134, 147, 151, 646, 647
fossil fuel, 913
Fourier transform infrared
spectroscopy, 9, 168
fractional distillation, 682
fragrance analysis, 699
frass, 677
freely dissolved concentrations, 241
freons, 284
fritted glass filter, 181
frontal chromatography, 361, 724, 734
Frossling's equation, 129
full evaporative technique (FET), 289
fulvic acid, 762
fungal volatiles, 686
fungi, 24, 26, 688
5-furoylquinoline-3-carboxaldehyde
(FQ), 616
fusible materials, 68
GAC, 1006
galactose, 652
galactosemia, 652
gamma-hydroxybutyric acid, 932
Garret's sampler, 47
gas chromatography, 102, 321, 447, 643,
728, 941
gas chromatography-photoionization
detection, 164
gas chromatography-mass spectrometry
(GC-MS), 6, 7, 27, 92, 103
gas chromatography-flame photometric
detection (GC-FPD), 21
gas diffusion detection device, 117
gas diffusion flow injection analysis, 125
gas mixtures, 464
gas separation membranes, progress in,
987
gaseous matrices, 464
gaseous pollutants, 98
gaseous samples, 532
gases, interference from, 199
gases, sampling from, 538
gas-liquid equilibrator, 135
gasoline, 899, 903, 922
gas-phase reactions, 541
gas-tight syringe, 290
Gatling gun design, 100
gene expression, 617
gentamicin, 639
gentrisone, 644
glass honeycomb denuders, 100
glass annular denuders, 101
glass denuder, 103
glass fiber filter, 183
glass parallel plate wet denuder, 176
glucose, 124
glutamate, 614
glutaraldehyde, 9
glutathione, 616, 641
glycerol/phosphoric acid, 104
glycine, 1093
glycols, 137
glycoproteins, 630
gold film, 130
Gore-Tex, 105, 113

1115
Gormley-Kennedy equation, 105 - removal, 1017
grab samplers, 81 helium ionization detector, 23
gradient elution, 122 helium-purge membrane inlet, 537
granular activated carbon (GAC), 1005 hematin, 111, 112
graphitized carbon black, 349, 352, 739, Henry's constant, 295, 729
744, 1033 Henry's law, 729
gravel, particle size distribution of, 233 heptafluorobutyl, 650
gravitational settling, 163, 181 herbicides, 619, 741, 906
grease, 67, 901 heterogeneity, 63, 65
green leaf volatiles, 677 hexavalent, 183
Griess-Saltzman reagent, 20, 128, 133 hexylchloroformate, 649
Grignard reaction, 958 high performance liquid
gross sample, 61 chromatography see HPLC
groundwater, 219 high-resolution transmission electron
guanine, 633 microscopy, 170
high specific surface area, 739
H202, 103, 124 Hildebrand solubility parameter, 584
H2S, 117, 130 Hirst-type samplers, 26
H2SO4, 102, 103, 129, 137, 164 HNO3, 137
hair, 789 hold-up volume, 365
halide generation, 958 hollow fiber, 481, 520, 759, 762, 975, 977
haloacetic acids, 647 homocysteine, 622, 623, 625
halobimanes, 624 homogeneous polymeric SPME coatings,
halocarbons, 727 1052
halogenated hydrocarbons, 294, 752, horizontal pipelines, 49
754 horseradish peroxidase, 110
Hantzsch synthesis, 12 hot water extraction, 587
hazardous air pollutants, 6 hot water reverse phase LC, 601
hazardous nuclear material, 208 house-flies, 687
HCHO, 103, 106, 107 HPLC, 10, 13, 27, 611, 613, 633
headspace, 245, 272, 320, 329, 444, 461, HPLC-FLU, 631
491,678, 689, 690, 704, 728, 759 HPLC-MIMS, 548
- configuration, 397 HPLC-UV, 14
- gas chromatography, 279, 295 human breath, 500, 932
- solid phase microextraction, 248, human colon adenocarcinoma, 617
250, 398, 580, 709, 750, 810, 922, human exposure to pollutants, 201
931 humic acid, 249, 762
headspace vial, 280, 282, 284, 285, 287, humic matter, 237, 721, 740, 753, 758,
289, 291, 295 761
- multiple sampling from, 285 humidifier, 164
health effects, 2 hybrid systems for water treatment, 999
health hazards, 727 hydrazines, 632, 638
heavy metals, 162, 237, 895, 908, 909 hydrazones, 632, 640, 650
- effect of pH and redox potential on, hydride generation, 956
225 hydrodistillation, 702

1116
hydrogen bonding, 736, 737
hydrogen chloride, 151
hydrogen exchange rate, 170
hydrogen fluoride, 151
hydrogen interactions, 742
hydrogen peroxide, 109, 111, 135
hydrophobic chemicals in soil, 250
hydrophobic drugs, 294
hydrophobic filter, 192
hydrophobic interactions, 736, 737, 738,
740
hydrophobic membrane, 757, 759
hydrophobic polytetrafluoroethylene
filter, 192
hydrophobicity, 113
5-hydroxy-2-aminovaleric acid (HAVA),
648
hydroxyapatite, 139
hydroxyethyl starch, 644
hydroxyls, 643
hydroxymethyl hydroperoxide, 109
2-hydroxymethylpiperidine, 103
ifophosphamide, 632
ignitable liquid, 921
illicit drug samples, 792
Illicit pharmaceuticals, 793
immersed membrane, 999
immobilized antibodies, 410
immobilized fluorophore, 123
immunoaffinity extraction, 814, 861,
878
immunoaffinity sorbents, 741
immunoassay, 898
immunoconjugate, 1082
immunogenicity, 617
immunological trapping, 513
immunosorbents, 357, 744, 1088
impaction, 179, 193
impactor, 203
indirect trapping, 513
indoor air environment, 203
- kerosene-fueled space heaters, 201
indoor HONO from open flame sources,
137
indoors inhalable particulate matter
concentrations, 201
indophenol blue, 103, 104, 115, 129
induction coupled plasma-mass
spectrometers (ICP-MS), 209
inductively coupled argon plasmas, 171
inertial impaction, 180
inertial methods, 163
infinite-dilution activity coefficients, 471
infrared, 9
inhalable particulate matter
concentrations indoors, 201
inhalants, 796
injectables, 796
injection volume, 284, 285
inorganic amines, 545
inorganic analyses, 659
inorganic anions, 659
inorganic cations, 659
inorganic ions, 724, 740
inorganic oxide adsorbents, 347
inorganic salts, 721, 753
insect cuticles, 684
insect pheromones, collection and
isolation of, 670
insecticides, 906
insects, 24
insulin, 615
interception, 180
interfacial adsorption, 305
interferences from prolonged sample
collection, 168
inter-laboratory studies, 463
internal standards in headspace
quantitation, 289
internal surface, 236
interparticle velocity, 363
intramolecular condensation, 640
in-tube SPME, 266, 401, 810
in-tube SPME-HPLC, 426
in-vial desorption, 347
iodoacetamide, 642
ion chromatography, 16, 18, 19, 120, 757
- coupling, 151
ion exchange beads, 1006
1117
Ion exchange fibers to remove inorganic
contaminants, 1014
ion exchange groups, 740
ion exchange mechanisms, 737, 739
ion exchange membranes, 105, 125
ion-exchange sorbents, 353, 739
ion exchange preconcentration, 176
ion pair sorbents, 739
ion suppression, 722
ionic interactions, 736, 737
ionic strength, 286, 740
ionizable compounds, 752
ion-pair extraction, 509, 726, 727, 740
isoamyl alcohol, 292
isobutylchloroformate, 649
isobutyric acid, 121
isokinetic samplers, 36
isopeptides, 638
isoprene, 932
isopropanol, 282
isotachophoresis, 969
isotherm, 298
isothiocyanates, 619, 638
isotopic dilution, 458
isotopic enrichment, 614
isotopic ratios in sugars, 652
kaolinite, 230, 236
kerosene heaters produce acid aerosols,
206
ketones, 26
kinase, 657
K6nig reaction, 118
Kovats retention index, 685
Kuderna-Danish evaporator, 14, 728
lag time, 333
laminar discs, 344
large volume injector, 723, 745
lasalocid, 640
laser ablation, 919
laser desorption/fast gas
chromatography, 391
laser desorption/ionization, 169
laser induced fluorescence (LIF), 614
1118
leachates, 219
leaching techniques, 944
lead, 660
leafminer moths, 688
lepidopteran pheromones, 688
leucine, 620
light extinction, 172
limit of detection, 724, 743
Limulus amebocyte lysate (LAL) assay,
24, 25
linalool, 287
linear alkylbenzene-sulfonates, 740
linear dynamic range, 457, 460
linear sample capacity, 298
liquid chromatography/electrospray
ionisation mass spectrometry, 409
liquid chromatography/mass
spectrometry, 409LC-MS, 27
liquid core waveguide (LCW), 109
liquid drop, 130
liquid extraction system, 187
liquid filled fiber optics, 126
liquid matrices, 467
liquid membrane, 486, 504
liquid phase microextraction, 301, 523
liquid-liquid interface, 298
liquid-liquid extraction, 297, 701, 712,
725, 756, 757, 762, 801, 849, 878,
913, 926, 968, 969, 1025
liquid-phase microextraction, 821
liquids, sampling from, 541
living systems, 472
Los Angeles, fine particle mass in, 172
low-specificity sorbents, 349
lumpy materials, 65
lymphocytes, 617
lysimeters, 46
macrocyclic ligands, 356
magnetic stirring, 446
Makrofol polycarbonate plastic, 17
malachite green, 130
malathion, 931
MALDI (matrix assisted laser
desorption ionization), 325, 615
MALDI-TOF, 627, 631, 656
maleic acid, 147
maleimides, 626
malondialdehyde, 650
malonic acid, 147
malonyldialdehyde, 648
manual headspace, 290
Marfey's reagent, 658
marine pollution, 762
marking pheromones, 671
mass median aerodynamic diameter
(MMAD), 102
mass spectrometry, 13, 531
- for aerosol analysis, 169
mass transfer, 506
matrix-assisted laser desorption
ionization see MALDI
matrix effects, 306
matrix interferences, 721
matrix matching, 306
matrix modifier, 287, 294
matrix solid-phase dispersion, 875, 885
maximum rate extraction, 489
maze type impactor, 179
MBTH, 11, 129
- reaction with HCHO, 129
McIntyre-Day scoop, 81
measurement of the fate of emitted H2 S
in an oil field, 207
measuring cell, 537
MECC, 615, 619
medical air, 680
membrane-assisted solvent extraction,
523
membrane-based denuders, 104
membrane-based extraction, new
polymeric materials, 1044
membrane-based passive samplers, 54
membrane-based sample preparation,
818
membrane-based techniques, 754
membrane-controlled extraction, 511
membrane extraction, 275, 520, 819
- apparatus, 520
- progress in, 1000
- with sorbent interface (MESI), 493,
759
membrane inlet designs, 536
membrane inlet mass spectrometry
(MIMS), 531, 759
membrane matrix, 117
membrane probe, 531, 537
membrane separation, 479
membrane surface characterization
- by atomic force microscopy, 994
- by electron spin resonance, 996
- by measurement of pore sizes, 993
membrane techniques, 881
membrane temperature, 535
membrane transport, progress in, 998
membrane vesicles, 248
memory effects, 300
Meperidine, 928
mercaptans, 118
2-mercaptoethanesulphonic, 641
mercaptoethanol, 623
2-mercaptoethanol (2ME), 613
3-mercaptopropionic acid (MPA), 613
mercury, 659
MESI, 494
MESI-GC, 495, 500
metabolic cytometry, 636
metabolites, 741, 913
metal complexation, 356
metals, 908, 930
methacrylic acid, 121
methadone, 928, 929
methamphetamine, 650, 927
methanator, 175
methane, 9
methanesulfonic acid, 151
methanol, 113, 647, 930, 1092, 1094
methanol-fueled vehicle (MFV), 125
method detection limit, determination
of, 462
method selection, 372
method validation, 463
methyl-4-aminomethyl benzoate, 631
methylamine, 116, 620
methylarsonic acids, 660
1119
methyl benzoate, 932
3-methyl-2-benzothiazoline hydrazone,
129
N-methylcarbamates, 727
methyl ethyl ketone, 292
methyl formate, 647
N-methylglycamine, 620
methyl hydroperoxide, 109
methylenedioxymethamphetamine, 928
methylmercury, 931, 941
Metricel, 105
Mexican fruit flies, 686
micellar electrokinetic chromatography,
134
microcosms, 911
microdialysis, 614, 878, 882
microdrop-LPME, 315, 321
microenvironmental exposure monitor,
15
microextraction techniques, 880
micro-fibres, 473
microfilms of water, 46
microfiltration membranes, 992
microinjection-microspectroscop, 326
micro-LC, 1097
micromachining, 642
microplates, 854
microporous membrane liquid-liquid
extraction (MMLLE), 505, 518, 757,
763, 821
micro-ring electrode, 130
microscale glassware, 678
microsomes, 248
micro-SPE, 389
microsyringe, 321
microtiter plates, 746
microwave-assisted extraction, 566, 572,
822, 883, 945
microwave ovens, 568
mildew, 26
milk, 788
MIMS theory, 532
mineral insulating oils, 294
mineral oil, 901
mixed functional phase, 1040
1120
mixed-bed ion exchange column, 187
mixed-bed sorbents, 1097
mixed-mode sorbents, 354, 740
mixing, 284, 290
mixing chamber, 193, 203
mobile phase, 735
modifiers, addition of, 597
molar absorbtivity, 616
mold, 26, 204
molecular imprinted SPE, 742
molecular imprinting, 816
molecular recognition, 741
molecular sieve membranes, 987
molecularly imprinted polymers, 357,
741, 744, 816, 878, 1035, 1053
momentum transfer, 305
monensin, 640
monoamines, 621
monobromobimane, 623, 624
monochlorobenzene, 746
monoclonal antibodies, 1083, 1087, 1091
monocytes, 617
monodisperse aerosols, 169, 193
monofunctional silane, 737
monomers in polymers, 279, 284
monosaccharides, 644, 652
monosegmented flow extraction, 312
montmorillonite, 230, 236
morpholinoethanesulfonic acid, 152
mortality data, 161
Muencke-type absorbers, 14
multichannel membrane collector, 126
multichannel sampler, 38
multimodal size distribution, 162
multimode extractions, 354
multiple concentric tubes, 101
multiple headspace extraction, 285, 287
multiple PPD, 102
Na 2 B40, 133
Nafion, 105, 140, 164
Nafion dryer, 731
Nafion membrane DS (NMDS), 106
Nafion suppressor, 206
nanodrop-LPME, 315, 325
nanofiltration, 998
nanopore membranes, 113
naphthalene, 13
naphthalene sulfonates, 740, 748
naphthalene-2,3-dicarboxaldehyde
(NDA), 615
narasin, 640
National Ambient Air Quality
Standards (NAAQS), 18, 19, 22, 98
National Institute for Occupational
Safety and Health (NIOSH), 4, 5,
13, 23, 651
natural organic matter, 753, 762
needle vibration technique, 446
negligible depletion SPME, 245, 247
nephelometers, 16
Nernst's distribution law, 725
neurodegenerative disorders, 614
neutron activation analysis, 16
neutrophils, 617
NH3, 102, 129
N-hydroxysuccinimidyl esters, 610
N-hydroxysuccinimidyl
fluorescein-O-acetate , 622
nicotine, 23, 102, 928
Niemisto corer, 81
nitrate, 139, 166, 196, 198, 659
nitric acid, 151, 168
nitrite, 139, 200, 166, 659
nitroaromatics, 638
nitrogen blowdown, 728, 733
nitrogen dioxide, 1, 19, 104
nitrogen-phosphorus detector, 13
4-nitrophenol, 742
nitrous acid, 128, 151
NMR, 27
NO 2, 23, 101, 102
non specific interactions, 1094
non-dispersive infrared analyzer, 9, 22,
101
non-exhaustive extraction, 256, 277,
733, 748
non-hygroscopic aerosol, 196, 197
non-porous hydrophobic membrane DS
devices, 125
non-porous membranes, 504, 755, 756
non-target analysis, 722
non-volatile organic compounds, 532
non-wetting phase, 310
nuclear solid waste, 909
nucleation, 162
nucleophiles, 622
Nurnik-1 sampler, 41
nylon filters, 168
03, 102
OASIS column, 631, 656
octanol/water partition coefficient, 736,
897
octanol/water distribution constant, 361
odors, 699
off-flavors, 699
off-line automation of SPE, 746
off-line cleanup, 969
off-line equipment, 523
off-line immunoextraction, 1094
off-line preconcentration, 969
off-line sample clean-up, 968
off-site analysis, 722
oil film thickness on the surface of
water, 47
oligosaccharides, 630, 654, 655
on-column electrophoretic
concentration, 975
on-disc derivatization, 347
on-line concentration, 972
on-line connection to chromatography,
522
on-line coupling, 379, 969, 978, 1095
on-line electrophoretic concentration, 972
on-line LLE-GC, 727
on-line membrane cleanup, 977
on-line preconcentration, 1095
on-line solid-phase extraction, 743
on-line vs. off-line Analysis, 843
on-site analysis, 722
on-site monitoring, 544, 747, 753, 760,
897
OPA, 613
organic aerosols, 201
1121
organic-carbon normalized partition
coefficient, 237
organic chloramines , 545
organic compounds, 540
organic matrixes, 546
organic pollutants, 560
organic solvents, 259, 319, 753,
1093-1095
organic vapors, 137, 169
organic volatiles in drinking water, 287
organoarsenic, 941
organochlorine pesticides, 12, 727
organophosphate pesticides, 931
organophosphorous pesticides, 746
oscillometric conductometry, 114
OSHA (Occupational Safety and Health
Administration), 4
outer particle surface area, 233
overaspiration, 206
oxalic acid, 147, 151
oxidase enzymes, 124
oxidation, 174, 648
oximations, 654
ozone, 18, 124, 208, 646
- damage to the lung, 161
ozonides, 646
packing density, 363
palm weevils, 687
Palmes tubes, 19
parallel plate denuders, 101, 104, 141
parallel processing, 743, 746
Paraquat, 170
pararosaniline, 10, 21
Parkinson's disease, 614
partial pressure, 729
partially concurrent solvent
evaporation, 745, 746
particle collection, 179
- by vapor condensation, 192
particle deposition, 125, 140, 144
particle size distribution, 232
particle sizing
- centrifuging, 234
- electron microscopy, 234
1122
- optical microscopy, 234
- radiation scattering, 234
- sedimentation, 234
- sieving, 233
particle-embedded glass fiber discs, 344
particle-loaded membranes, 344
particle-particle reactions, 169
particles
- prehumidification of, 164
- thermal decomposition of, 164
particulate air pollution, 161
particulate matter, 1, 14, 23
particulate sulfate, 208
partition coefficient, 228, 280, 281-286,
289, 290, 294, 295
partition law, 725
partitioning, 725
passive dosimetry, 53
passive methods, 54
passive samplers, 4, 7, 54, 346, 761
passive time weighted average
sampling, 272
passive water sampling, 53
pattern-recognition techniques, 294
PDMS, 400, 408, 441, 442, 465, 494
peat samples, 549
Peltier cooled maze, 203
pentaacetylaldonitrile, 652
pentadecafluorooctanoyl, 650
pentafluorobenzaldehyde, 651
pentafluorobenzyl bromide, 653
pentafluorobenzyl hydroxyamine, 12
o-(2,3,4,5,6-pentafluorobenzyl)-
hydroxylamine, 421
pentafluorobenzylation, 646, 650, 654
pentafluorophenyldimethylsilyl, 646
pentanedione, 109
peptides, 615, 616, 622, 653
perfluorinated reagents, 649
perfluorotolyl, 650
periodic heterogeneity, 65, 66
periodical analyses, 49
permanent gases, 294
permeation, 117, 486, 489
- denuders, 117
- principles, 256 o-phthalic acid, 147
peroxidase substitute, 111 phosphatidylcholine, 634
peroxyacetyl nitrate, 101, 137 phosphines, 623
peroxyoxalate chemiluminescence photoacoustic spectroscopy, 9, 173
(PO-CL), 123 photoacoustic-IR, 22
perpendicular pipelines, 49 photoionization, 169
pertrimethysilylated, 643 photoionization detector, 9
perturbation, 336 photolytic activation, 636
pervaporation membranes, progress in, phthalates, 13
989 physicochemical applications, 470
pesticides, 12, 295, 609, 741, 742, 895, physicochemical constants, 276
906, 931 physicochemical properties of aqueous
Petri dish, 26 and solid environmental matrices,
petroleum, 895 219
petroleum hydrocarbons, 899, 909 phytoextraction, 909
pH, 416, 456, 724, 752 phytoremediation, 897, 909
- value of soil or sediment samples, picodrop-LPME, 315, 326
231 pipe sampler, 72
- value of water, 222 piped water, 52
- variations, 1092 planar diffusion scrubbers, 116
Phanaerocheatechrysosporium, 912 planar DS, 117
pharmaceuticals, 293 planar molecules, 739
pharmacokinetics, 642 plant volatiles, 691
phase ratio, 280, 285, 299, 725 plants, 688
phase separation, 307, 727 plasma, 209, 785, 929
phenanthrene, 13 plate number, 362
phencyclidine, 929 platelet activating factor, 650
phenolic compounds, 747 platinum-DDTC, 642
phenolic estrogens, 653 Plexiglas denuder, 153
phenolic substances, 134 plutonium, 209
phenols, 13, 653, 727, 737, 747 PM10, 98, 99
phenothiazines, 931 PM2.5, 98, 99, 162
phenoxy acid herbicides, 754 poisons, 926
phenyl silica, 738 polar compounds, 540
phenylboronic acids, 356 polarizable electrons, 739
phenylisothiocyanate, 638 pollen, 24, 25
phenylureas, 1091, 1095, 1097 polluted aqueous samples, 763
pheromones polluted wastewaters, 52
- biological contexts, 673 poly(dimethylsiloxane) (PDMS) coating,
- definition, 670 399
- gland extracts, 677 poly(styrene-divinylbenzene)
- methods for sampling and copolymers, 738
collection, 674 poly(styrene-divinylbenzene)-based
- types of, 670 materials for SPE, 1034
o-phthaldialdehyde, 115, 612 polyacrylate, 751

1123
polyamines, 621
polyaromatic hydrocarbons, 227, 899,
1096
polychlorinated biphenyls, 13, 100
polychlorobiphenyls, 903
polyclonal antibodies, 1083, 1087
polycyclic aromatic compounds, 13, 23,
92, 741, 747
polydimethylsiloxane, 751, 752
polyether antibiotics, 640
polymer coatings, 752
polymer sorbents, 738
polymerase chain reaction, 27
polymeric composite coatings, 1052
polymeric extraction materials in SPE,
1034
polymeric materials, 1024
polymeric membrane extraction, 505,
761
polymeric resins, 738
polymeric surface coatings for SPME,
1050
polymers, 552
polynuclear aromatic hydrocarbons,
100, 170
polypyrrole, 408, 409
polypyrrole-based coatings, 1054
polysaccharides, 627, 654
polystyrene-divinylbenzene, 732, 744,
751
polystyrene-divinylbenzene-N-vinylpyrr
olidone, 739
polyurethane foam, 12, 14, 92
polyvinylidene fluoride, 113
Porapak Q, 679, 681
pore blockage, 125
pore size, 113
pore structure, 1009, 1011
pore tortuosity, 113
Poreflon porous PTFE scrubbers, 123
Poreflon tubes, 114
Poropak, 13
porosity, 236
porous graphitic carbon, 352, 736, 738,
739, 1034
1124
porous hydrophobic membranes, 113
porous membrane diffusion scrubbers
(PMDS), 113, 114, 118
porous membranes, 117, 754, 975, 977
porous polymer beads, 343
porous polymers, 349
post-sampling aerosol analysis:, 167
potassium, 198
Prandtl boundary layer, 440
precipitation, 219
precision, 460, 744
precolumn, 743, 744
precolumn dimensions, 379
preconcentration, 183, 297, 968, 969
- of volatile organic compounds, 45
pre-equilibrium principles, 256
preservatives, 799
pressure drop, 362
pressure-balanced headspace sampling,
291
pressurised fluid extraction, 573
pressurized liquid extraction, 885, 886
primary samples, 61
- of wild plants, 79
primer pheromones, 672
probe for collecting samples from
pipelines, 49
profile sampling, 72
proline, 648
n-propanol, 292, 293
propionic acid, 121, 147
proportionality constant, 685
prostaglandins, 654
protein precipitation, 626, 847
protein removal, 800
proteins, 615, 616, 656
proteome analysis, 978
Pseudomonas pseudoalcaligenes,913
pulsed fluorescence analyzer, 164
purge-and-trap, 279, 280, 287, 294, 688,
690, 691, 730, 903
purge-and-membrane mass
spectrometry, 550
push-pull systems, 680
putrescine, 621
1-pyrenyldiazomethane, 420 response time, 533
pyrolysis, 552 restricted access material, 740, 744, 860,
pyrrolidone group, 739 1039
pyruvic acid, 147 restricted access sorbents, 355
retaining column, 745
quadrupole MS, 169 retention, 365
quantitation, 287, 289, 487 retention factors, 366, 735, 738
- techniques, 286 - methods for estimating, 366
- problems with, 289 retention gap, 745, 763
quantitative analysis by LLE, 306 retention volume, 735
quantitative extraction, 418 retinitispigmentosa, 656
quartz fiber filters, 168, 175 reverse osmosis membranes, 985
quaternary ammonium herbicides, 747 reverse osmosis, progress in, 983
quaternary ammonium salts, 740 reversed phase sorbents, 735
reversed-phase liquid chromatography,
racemization, 652 137
radioactive waste, 909 reversed-phase membrane inlet mass
radioimmunoassay, 25 spectrometry, 546
radon, 1, 16 rhinoceros beetles, 687
rainwater, 48 rhodamine, 616, 636
Raman microspectroscopy, 18 Rhodococcus, 913
random heterogeneity, 65 rhodopsin, 656, 657
random variability, 69 ribosides, 633
rate determining step, 302 rinse solvent, 369
reactions, monitoring, 545, 547 robots, 746
reactive oxygen species (ROS), 616 root layer soils, 75
real-time emission spectrometry, 171 rotary evaporator, 728
real-time single particle mass rotating wet annular denuders, 140
spectrometry, 171 Rotorod sampler, 26
receiving phase, 330 rubber, 24
recognition pheromones, 671 rumen fluid, 549
recovery, 511, 723, 735, 1090
redox potential, 224 saccharides, 636, 654
- in soils and sediments, 231 - determination via derivatization of
reductive amination, 629, 630, 636 the carbonyl, 627
reference material, definition, 940 safe sampling volume, 369, 734
refluxing with a hydrophobic filter, 192 salinomycin, 640
regeneration, 1093 saliva, 549, 787
relative response factors, 685 salt
repeated sampling, 689 - adding to aqueous samples, 282,
representative sample, 67 284, 457, 753
residual solvents in pharmaceuticals, - concentrations, 416, 984
279, 293 - salting out, 284, 312, 724, 726
respirable particulates, 15, 23 sample cleanup, 297, 723, 738, 742, 756,
respiratory ailments, 161 758, 762, 763, 817, 878, 968

1125
sample degradation, 722 scintillation cells, 17
sample handling, 289 seashore waters, 46
sample holding time, 92 secondary chemical equilibria, 298
sample preparation, 722 secondary flow, 311
- for analysis, 55 sedimentary matter, 219
- on-line ASTED, 861 sediments, 81, 549
- on-line immunoaffinity extraction, segmented flow, 183, 301, 727
861 segmentor, 307, 727
- on-line RAM columns, 860 segregation, 64
- on-line turbulent flow, 859 selective volatilization, 174
sample processing considerations, 373 selectivity, 517, 758
sample stability, 746 semi-infinite spherical diffusion, 318
sample stacking, 972 semiochemicals, 670
sample volume, 372, 744, 751, 753 semipermeable membrane devices), 54,
- and extraction recovery, 1090 761
- optimization, 448 semipermeable surface, 1040
samplers with a shutter closure, 82 semi-preparative column
samplers for loose and lumpy material, chromatography, 631
69 semi-volatile compounds, 540
samplers for loose materials, 71 semi-volatile organic compounds, 87, 91,
samples 759
- of undisturbed structure, 73 sensitivity, 460, 1097
- number necessary, 62 separating funnel, 726
- without structure preservation, 73 separation/detection technique,
sampling selection, 447
- biological plant materials, 78 Sepharose, 1089, 1095
- for speciation analysis, 944 sequential composite sample, 34
- from horizontal barrel boilers, 51 sequential injection analysis, 115, 129
- from streams, 36 sequential injection extraction, 312
- ground cover, 79 sequestration, 897
- leaves, 80 serial processing, 746
- groundwater, 42 serial sample processing, 742
- water and vapour from closed Serpentine-II, 106
circuits, 49 serum, 785, 926
sampling plan, 62 serum protein binding, 248
sampling port diameter, 685 serum proteins, 242
sampling protocol, 62 sewage sludge, 219
sampling site selection, 77 sex pheromones, 671
sampling soil from surface layers, 76 SFE see supercritical fluid extraction
sampling stick, 74 shielded hydrophobic phase, 1040
sand, particle size distribution of, 233 sialic acids, 627, 655
scanning electron microscopy (SEM), sick building syndrome, 24, 202, 203
994 silanol groups, 737
scanning probe microscopy, 170 silica, 347, 656, 732, 738, 763
Schiff Base, 629 silica-based extraction materials, 1031

1126
silicone membrane DS, 126
silicone rubber membrane, 118
silt, particle size distribution of, 233
silylation, 643, 654
simple composite sample, 34
simple permeation, 506
simultaneous distillation-extraction,
702
single-cell analysis, 980
single drop-LPME, 315, 318, 320
single jet inertial impactor, 198
single liquid drop techniques with two
phases, 301
single particle mass spectrometry, 170
single-tube denuders, 99
single-tube glass denuder, 104
single-tube samplers, 69
single-tube wet denuder, 138, 137, 181
size exclusion, 740
size exclusion chromatography, 234,
761, 763
Slocum's sampler, 39
sludge, 469
small-volume injector liners, 685
SO2, 129, 151
soda-lime trap, 187
sodium, 198
sodium acetate, 644
sodium borohydride, 623
sodium cyanoborohydride, 630, 631
sodium dodecylsulphate, 654
sodium perchlorate, 183
sodium tetraethylborate, 660
soil, 72, 219, 284, 286
- soil analysis, 77, 895, 899
- (bio)remediation, 911
- hydrophobic chemicals in, 250
- profile, 72
- samples, 532, 550
- slurries, 577
- surface pollution, 76
- texture, 233
- volatile organics in, 287
soil-water partition coefficients, 897
solanesol, 23
sol-gel monolithic beds for extraction,
1065
sol-gel polymeric extraction media, 1056
sol-gel polymers, 1025, 1041
solid adsorbents, 89
solid analysis, 532
solid environmental matrices,
definition, 219
solid environmental samples, 560
solid matrices, 469
solid phase analytical derivatization,
618
solid phase extraction see SPE
solid phase microextraction see SPME
solid sampling, 550
solid versus liquid sorbents, 268
solid-liquid extraction, 731
solid-solution partitioning of metals,
238
solid-supported LLE, 850
solids, extraction of, 260
solids injection syringe, 690
solubilities, 725
solubility parameter, 583
solvation parameter model, 367
solvent evaporation, 728
solvent extraction, 677
solvent-free standards, 413
solvent selection, 583, 726
solvent vapor exit, 745, 746
solventless gas chromatographic
injection, 647
solventless sample preparation, 898, 899
sonication, 901, 902, 906, 950
soot particles, 170
sorbent, 732
- with enhanced selectivity, 740
sorption of organic compounds, 238
source layer, 319
Soxhlet, 14, 582, 682, 899, 901, 902
SPE, 27, 713, 245, 731, 805, 878, 968
- cartridge, 266
- historical development, 342
- -liquid chromatography, 379
- off-line/cartridges, 851
1127
- off-line/microplates, 854
- on-line, 858
- new polymeric materials in, 1029
- theory, 359
- traditional materials, 1030
- working principle of, 1029
speciation, 335, 908
- definition, 939
- monitors, 161
specific attenuation cross section, 173
specific interactions, 1094
spectrophotometer, 21
spectroscopic methods for aerosol
analysis, 167
spermidine, 621
spermine, 621
spherical drop diffusion, 317
spin bar sorptive extraction (SBSE), 878
SPME, 4, 7, 12, 98, 102, 223, 227, 389,
683, 701, 807, 880, 906, 919, 951,
970
- applications, 464, 686
- devices, 749
- evolution of, 391
- for measuring freely dissolved
concentrations, 245
- -gas chromatography interface, 422
- -high performance liquid
chromatography interface, 425
- in biomedical analyses, 247
- in-tube, 266
- kinetics, 227
- method development and
optimization of, 435
- new polymeric materials, 1049, 1053
- optimization and calibration, 436
- principles of, 395
- time-weighted average sampling, 403
standard addition, 287, 306, 458
static (non-flowing) extractions, 595
static drops, 129
static headspace, 279, 280, 294, 704,
729, 903
static in-tube SPME time weighted
average sampling, 403
1128
stationary phase, 735
steady-state extraction, 490
steady-state flow, 532
steam distillation, 677
steam injection rates, 193
steric hindrance, 620
steroids, 930
stiffening agents, 799
stinkbugs, 688
stir bar sorptive extraction, 824, 880
stirring rate, 321
stomach fluid, 549
storage conditions, 94
storage stability, 88
subcritical water extraction, 587
submersible rectangular sampler, 47
succinic acid, 147
sugarcane weevils, 687
sulfamic acid, 104, 130
sulfate, 98, 102, 161, 164, 196, 198, 206,
207
sulfite, 124
sulfite oxidase, 124
sulfonation, 739, 1018
sulfonic acid groups, 739
sulfosalicylic acid, 626
sulfur, 641
sulfur dioxide, 21, 151, 164
sulfuric acid, 103, 162
sulfurous acid, 151
SUMMA canisters, 7
Super Q, 681
supercritical fluid extraction (SFE), 701,
714, 803, 878, 886, 902, 949
- of food samples, 888
- instrumentation, 562
- method development, 566
supercritical fluids, 561
superheated water extraction, 587
supersaturation, 193
supported liquid membrane (SLM), 301,
504, 506, 756, 820
suppositories, 796
surface-active impurities, 226
surface-active substances, 305
surface area, 236
surface-bound macrocyclic ligands, 356
surface layers of water, 46
surface modification by surface
modifying macromolecules (SMM),
990
surface porosity, 113
surface tension, 226
surface water, 36, 219, 753
surfactants, 740, 747, 748
suspended particles, 722, 733, 734, 761
suspended sediments, 83
suspending agents, 799
sweat, 790
swelling, 534
syringe-based automated system, 290
system map, 368
Taber trap, 26
tablets, 795
tamoxifen, 636
tandem mass spectrometry (MS/MS),
543
tapered element oscillating
microbalance (TEOM), 16, 102
target analysis, 722, 747
Tedlar® bags, 681
Teflon, 105, 204, 117, 126, 127, 169
temperature, 447, 448, 456, 498, 574,
589, 724, 752
- aqueous media, 221
- dependence of SLM extraction, 512
- solid samples, 228
temperature-programmed desorption
membrane inlet mass spectrometry,
540
temperature-responsive polymers in
solution mediated extraction, 1026
template, 741, 742
Tenax, 13, 679, 681,688, 730, 731
termites, 689
tert-butanol, 292
tertiary amine, 618
test sample, 61
tetrabutylammonium, 647
2,3,5,6-tetrafluorophenol, 641
tetramethylammonium hydroxide, 648
tetramethylbenzidine, 127
theoretical plates, 735
thermal decomposition of particles, 164
thermal desorption, 3, 8, 12, 102, 103,
683, 690, 730
thermal treatment, 174
thermodenuders, 102, 103, 164
thermodynamic constants, 283, 295
thermodynamic studies, 284
thermodynamics, 257, 436, 752
thermoelectrically cooled maze, 193
thermophoresis, 140, 193
thiamine, 112
thin layer chromatography, 27
thiochrome, 112
thioglycol methylate, 660
thiols, 622, 624
- fluorophoric tagging of, 624
throughfall waters, 48
Ti(IV)-4-(2-pyridylazo)-resorcinol, 123
time weighted average sampling, 272
time-of-flight mass spectrometry, 169,
615
time-weighted average, 3, 406
tin, 660
tissues, 792
toluene, 128
topicals, 795
tortuosity, 113
total metal burden, 238
total organic carbon, 226, 237
total suspended particulate matter, 99,
172
toxicological analysis, 926
trace analysis, 1096
trace element speciation, 939
- instrumentation, 941
trace gases, 97, 100, 129, 154
trace metals, 208
- in aerosols, 171
- mobilization of, 223
trail pheromones, 672
transacylation, 650
1129
transducin, 656 - Method IP-2B, 24
trap-and-release MIMS, 535, 541 - Method IP-5A, 19, 20
trap-in-needle system, 433 - Method IP-6A, 10
trapping efficiency, 686 - Method IP-6B, 10
trapping on glass, 691 - Method IP-7, 14
trapping time, 498 - Method IP-8, 12
triazines, 740, 741, 742, 747 - Method TO-13A, 92
trichloroethylene, 128 - Method 5021, 294
1,1,1-trichloromethane, 283 U.S. Pharmacopeia Method 467, 293
trifluoroacetyl, 650 UV absorption, 10
trifunctional silane, 737
trimethyl orthoacetate, 647 vacuum manifolds, 733
trimethylamine, 116 valeric acid, 147
Trimethyliodosilane, 643 validation, 463
trimethyllead, 941 valve and loop-based headspace
trimethylsilylation, 610, 643, 645 autosampler, 290
trimethylsilylimidazole, 644 valve and loop-based systems, 290
trimethyltrifluorotoluylammonium, 648 van Veen grab, 81
tri-octyl phosphine oxide (TOPO), 510 vapor condensation, 192, 208
triolein, 761 vapor distillation, 702
triphenylphosphine, 634 variability, 63
tunable diode laser absorption vial size, 285
spectrometer (TDLAS), 107 vinylacetic acid, 121
turbidity, water, 222 viruses, 24
turbulent flow, 859 visibility degradation due to aerosol
two-stage inlets, 537 light absorption, 173
tylenol, 622 visibility reduction, 174
visible light, 638
ultrafiltration, 244, 800 vitamin B1, 112
ultrasonic extraction, 904 vitamin D3, 655
ultraviolet, 638 volatile acids, 284
unconventional poisons, 930 volatile compound, 744
unforced water discharge, 52 volatile losses, 728
unsaturated soil water, 45 volatile organic compounds, 1, 5, 23, 26,
unsupported liquid membrane 40, 45, 72, 87, 294, 531, 534, 542,
techniques, 301, 329 727, 754, 759, 899, 903
urea, 1093 volatile organics in soil, 287
urine, 293, 786, 926, 927, 930, 931 volatiles in biological fluids, 292
U.S. EPA Methods, 5, 6, 15, 21, 92, 287, volatiles in high-boiling oils, 294
624, 731, 899 volatilization, 169
- Method IP-1OA, 15 volatilization loss, 221
- Method IP-1OB, 16 Vortex technique, 446
- Method IP-1A, 7
- Method IP-1B, 6 washing solution, 734
- Method IP-2A, 23 wasps, 689

1130
wastewater, 219, 750, 753 - glass, 137
water, 541 wet effluent systems, 183
water analysis, applications of SPE for, wetted denuders, 135, 136
747 wetting film, 310, 322
water analysis, procedures, 734 wetting phase, 310
water content of solid matrices, 228 Whitman two-film model, 303
water extraction, 578 whole blood, 785
water purification, 1013 whole-body extraction, 669
water salinity, 221 whole cell suspensions, 980
water sample storage, 56 whole-column cryotrapping, 731
water trap, 731 wines, 651, 881
water-organic modifier, 1092
water-soil partition coefficients, 910 XAD-2, 13, 14, 92
wet aerosol collectors, 199 XAD-4, 23
wet analysis, 175, 176, 200 xenobiotics, 879
wet annular denuders, 140, 141 X-ray diffraction, 18
wet denuders, 104 X-ray emission, 16, 170
- coupled IC analysis of acid gases X-ray fluorescence, 168
and ammonia, 148 xylene, 746
- which design to choose, 147 o-xylene, 289
wet diffusion denuder, 205
wet effluent diffusion denuders zeolite membranes, 989
(WEDD), 136 zone broadening, 307

1131