Talanta 45 (1998) 531 – 542

On-line cryogenic trapping with microwave heating for the

determination and speciation of arsenic by flow
injection/hydride generation/atomic absorption spectrometry
J.L. Burguera *, M. Burguera, C. Rivas, P. Carrero
Venezuelan Andean Institute for Chemical Research (IVAIQUIM), Faculty of Sciences, Uni6ersity of Los Andes, P.O. Box 542,
Mérida, Venezuela
Received 4 February 1997; received in revised form 12 June 1997; accepted 13 June 1997

Abstract

An on-line flow injection-hydride generation/atomic absorption spectrometry method was developed for the
preconcentration and selective determination of inorganic arsenic [As(III) and As(V)] and its methylated species. The
separation of the arsenic species was performed by an automated pH-selective arsines generation technique, using
sodium tetrahydroborate(III) as reductant. Each arsine was cryogenically trapped in a PTFE coil, knotted and sealed
inside another wider diameter tube, through which liquid nitrogen was suctioned by negative pressure. Then, based
on their different boiling points, the arsine species were selectively liberated by using a heating cycle of microwave
radiation, followed by atomic absorption detection. A sample solution aliquot mixed with 1% citric acid was used for
the determination of As(III) alone, while a second sample aliquot mixed with 2 mol l − 1 nitric acid was used for the
quantitative determination of total inorganic arsenic, monomethylarsonic acid and dimethylarsinic acid. Based on 10
ml sample, the detection limits lie within the range 20 – 60 ng As l − 1, which are sufficiently low to detect the
arsines-forming species in natural waters. These values are negatively affected by the reagents purity and background
noise due to flame flickering, but the sensitivity can substantially be improved by increasing sample size or running
several consecutive reactions. © 1998 Elsevier Science B.V.

Keywords: Arsenic speciation; Cryogenic trapping; Microwave heating; Flow injection; Hydride generation; Atomic
absorption spectrometry; Automated analysis

1. Introduction for scientists working in different fields. It is of

Arsenic is one of the elements which, because of
its abundance in the environment, its known toxi-
city for living organisms and its complex bio- and
geochemistry is of particular concern and interest
* Corresponding author. Fax: +58 74 401286.
special importance to study some aquatic ecosys-
tems, where, depending on the specific conditions
of the water bodies (pH, salinity, redox potential,
temperature, presence of microorganisms, etc.),
arsenic may exist in different oxidation states
(As(III) and As(V)) or organic forms [1,2]. It is
thought that As(V) as a phosphate analog is

0039-9140/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved.

PII S 0 0 3 9 - 9 1 4 0 ( 9 7 ) 0 0 1 8 9 - 6
532 J.L. Burguera et al. / Talanta 45 (1998) 531–542
Table 1
Some physical constantsa of the arsenicals studied
Compound pK
Arsenous acid [As(III)] 9.23
(HAsO2)
Arsenic acid [As(V)] 2.25
(H3AsO4)
Monomethylarsonic acid (MMAA) 2.60
[(CH3AsO(OH)2]
Dimethylarsinic acid (DMAA) 6.19
[(CH3)2AsO(OH)]
a
Values taken from references.
b
a0 = [H+]/[H+]+K1
is the fraction of undissociated arsenic species at equilibrium, where K1 is the first dissociation constant of the acids and [H+] is the
hydrogen ion concentration.
readily taken by aquatic organisms and trans-
formed into simple methylated species:
monomethylarsonic acid (MMAA) and dimethy-
larsinic acid (DMAA). Due to biotransformation
processes, these species are frequently detected
not only in the aquatic biota, but also in marine
and freshwater systems [1]. Thus, although the
total concentration of the element is useful to
know, the individual determination of the differ-
ent species is more relevant. However, when such
a problem has to be solved, the analytical
chemists are faced with very difficult tasks in
assuring the acquisition of accurate data, since
spontaneous changes in species distribution may
occur in time, leading to erroneous results and of
course to misleading conclusions.
For the determination of arsenic species in wa-
ter samples, where the concentrations are usually
below 1 mg l − 1, whatever the method, one or
more preconcentration steps may be always neces-
sary, while to differentiate between species it is
more common, though possibly more time con-
suming to use a separation technique. Most work-
ers have successfully used high performance liquid
chromatography (HPLC) [3 – 5] and ion exchange
[6–8], followed by arsines generation with induc-
tively coupled plasma atomic emission spectrome-
try (ICP) or atomic absorption spectrometry
(AAS) detection [2]. In general, these techniques
are time consuming and involve tedious proce-
a0 at pH 1b a0 at pH 4b Volatile product Boiling point (°C)
1.00 1.00 AsH3 −55
0.95 0.017 AsH3 −55
0.98 0.038 CH3AsH2 2
1.00 1.00 (CH3)2AsH 35.6
dures to separate the individual species. Selective
generation of the arsines followed by their freeze
trapping and sequential volatilization prior to the
determination of the analyte by AAS [9,10] or
other methods [11], have proved to be more sim-
ple and sensitive. This is the bases of the specia-
tion procedure developed in this study.
It is well documented that trivalent and pen-
tavalent arsenic show different behavior in the
generation process due to kinetic discrimination.
In the batch arrangements this difference is not
very pronounced and provided selected experi-
mental conditions in flowing systems, mainly in
flow injection (FI) only As(III) is reduced, proba-
bly because the As(V) reaction is too slow [12,13].
This effect could be reduced or even eliminated by
using long reaction coils [14]. This approach is not
generally recommended since it increases the total
time of analysis and could result in increased
interferences [15]. Differentiation between the two
valence states is made possible using pH selective
arsine generation technique and carefully con-
trolling the experimental conditions for the gener-
ation of the volatile species. Not only inorganic,
but also some organic containing arsenic com-
pounds, namely MMAA and DMAA are reduced
to the corresponding arsines (Table 1) by sodium
tetrahydroborate in acidic media. Thus, they may
contribute to the apparent yield of arsine if the
generator is connected directly to the atomizer, so
 J.L. Burguera et al. / Talanta 45 (1998) 531–542 533

Fig. 1. Schematic diagram of the FI-Hg-AAS system used for the determination of arsenic species with on-line cryogenic trapping
and volatilization with MW heating. The system is shown in the sample and nitric acid loading position for the determination of
total inorganic arsenic, MMAA and DMAA.

that inorganic arsenic concentration will be over-
estimated.
This paper describes an automated method that
utilizes an ingenious trapping procedure for the
separation and preconcentration of four arsenic
species by first trapping and then selectively
volatilizing them prior to AAS analysis. For the
first time microwave (MW) energy was used to
selectively volatilize the trapped species.
2. Experimental
2.1. Apparatus
The experimental assembly used in this work is
schematically shown in Fig. 1. It consists on three
Ismatec IPC peristaltic pumps (P1 – P3) and two
Latek TMW injectors, one 4-way, (V1), and the
other 6-way, (V2), used to choose the acidic solu-
tions for HG and to inject the acidic mixture,
respectively. P1, P2 and V2 are synchronized in
such a way that to avoid excess NaBH4 and
acidified sample being carried to waste. It also
includes a time based solenoid injector (TBSI) [16]
used to distribute the He gas in the system, a gas
phase separator (GPS) recently described else-
where [17], a drying tube for water vapors (DT),
an NaOH trap for acid vapors and CO2 (AT), a
cryogenic trap (CT) located inside a domestic
Panasonic programmable microwave oven
(MWO) and a quartz cell (QC) aligned in the light
path of a Varian arsenic lamp. A Varian, Model
1475 atomic absorption spectrometer linked to a
TDK 286 personal computer, which controls the
whole system, was capable of measuring both
peak height and peak area values. All determina-
tions were made at least in triplicate under the
optimized experimental operating parameters
given in Table 2.
To remove water vapors and reagents mist, a
drying tube (DT) made of polypropylene, 5 cm
534 J.L. Burguera et al. / Talanta 45 (1998) 531–542

long, 1.0 cm i.d., was filled with Drierite® as drides to the QC, introduced in the system by
desiccant; the carbon dioxide and acid vapors alternating the positions of the TBSI, was from
absorber (AT) was a 5 cm long, 1.0 cm i.d., AGA, which certifies a purity of 99.99%. The
polypropylene tubing partly filled with small removal of the liquid from the GPS, as well as the
sodium hydroxide beads, Becker analyzed® washing procedure were controlled by P3 main-
reagent of 0.4 – 0.5 cm of diameter. Both were tained in operation during the entire process.
incorporated to the experimental arrangement as
shown in Fig. 1. 2.2. Reagents
The CT, made from PTFE 1.0 m long, 0.8 mm
i.d., was sealed in a knotted form inside a Standard solutions (1000 mg As l − 1) of arsen-
polystyrene tube 1.0 m long, 0.5 cm i.d., through ite, arsenate, MMAA and DMAA as well as
which liquid nitrogen was suctioned by negative reductant solutions were prepared as indicated
pressure produced by an IWAKI air pump (AP), elsewhere [8]. Freshly prepared solution of sodium
Model AP-115 operated at a maximum flow of 15 tetrahydroborate(III) [2.0% (w/v) in sodium hy-
l min − 1. This produces a vacuum of 450 mm Hg, droxide 0.2% (w/v)] was used throughout. High
enough to almost fill the polystyrene tubing with purity mineral acids (HCl, HClO4, H2SO4, HNO3
liquid nitrogen. and H3PO4), commercially available from Merck,
In order to assure proper active sites on the QC were submitted to a purification process [19]. The
surface during the atomization step, the cell was organic acids (formic, acetic, oxalic, citric, tartaric
periodically cleaned with concentrated hy- and trichloracetic), tested for the generation of the
hydrides were also analytical grade reagents from
drofluoric acid as recommended in the literature
Merck. High purity deionized water (18 M cm − 1)
[18]. The pH was measured with a Metrohm
was used throughout to prepare all the solutions.
Herisau Titriskop, Model E 516. Helium used to
The glass- and plastic-ware used for the prepara-
maintain a stable baseline and to carry the hy-
tion and storage of solutions was precleaned by
Table 2
soaking in nitric acid and rinsed with water. Solu-
Operating conditions of the on-line arrangements tions containing potential interferents were pre-
pared from chloride or nitrate salts. All chemicals,
Wavelength 193.7 nm after preparation were degassed by bubbling
Band width 0.5 nm through a flow of nitrogen and then, were kept in
Lamp Varian (5 mA)
Quarts cell (QC) 15 cm path length×8 mm i.d.
closed dispenser bottles to avoid absorption of
Flame Air/acetylene (12.0/1.5 l min−1) carbon dioxide from air.
Hellium flow rate The buffers considered in this study were: acetic
Direct to QC 250 ml min−1 acid-acetate (1 mol l − 1) (pH range 1.0–5.0), citric
Through GPS 250 ml min−1 acid-citrate (5+ 1 sodium citrate 1 mol l − 1 +
Reductant
Concentration 2% (w/v) in 0.1 mol l−1 NaOH
citric acid 10% (m/v)) for a pH range 5.5–6.0 and
Flow rate 1 ml min−1 phosphoric acid-phosphate (pH range 5.5–8.5),
Acids prepared as indicated in the literature [20].
Concentration
Citric acida 1% (w/v)
2.3. Samples
Nitric acidb 2 mol l−1
Flow rate 5 ml min−1
Sample flow rate 5 ml min−1 For the analysis of arsenic species in water
P1 activation time 1 min samples, special care must be taken on the design
P2 activation time 2 min of the sampling protocol to avoid losses or
Trapping coil length 1m
changes between species. The sampling scheme
MWO power 700 W
adopted in this study was as follows: river and
a
For the selective determination of As(III).btFor MMAA, trout fishery waters were collected at each sam-
DMAA and total inorganic arsenic determination. pling site using a 1l high density polyethylene
 J.L. Burguera et al. / Talanta 45 (1998) 531–542
bottle, which was rinsed several times with the
sample in an attempt to saturate adsorption sites
on the vessel walls. The representative sample was
taken at last from that site; where possible, the
sample was taken from below the surface. The
bottles were filled to the brim with the samples to
avoid any air space, were transported to the labo-
ratory without any additives and were kept refrig-
erated (4°C) until the analysis was carried out.
2.4. Procedure
The procedure consists on various steps con-
cerning arsines generation and cryogenic trapping
followed by selective volatilization and detection,
as described below.
2.4.1. Procedure 1. Determination of total
inorganic arsenic, MMAA and DMAA.
Arsines generation starts when valve V1 and
pump P2 are activated for 160 s to deliver sample
and nitric acid solutions to fill the loop (L) of V2.
At the same time, P1 is activated to deliver a
volume of reductant. The reaction takes place in
R1 and is completed in the GPS. The volume of
NaBH4 is controlled by the P1 tubing diameter
and its time of activation. Concomitantly with the
activation of V1 and P2, the air pump is turned on
to fill in the polystyrene tubing inside the MWO
with liquid nitrogen. As the reaction takes place,
the TBSI changes position to deviate the flow of
helium through the GPS in order to help the
sweeping of the volatile products (AsH3 from
As(III) and As(V) and the corresponding arsines
of MMAA and DMAA) to CT. The H2 evolved
as by-product of the reaction, passes directly to
the QC (its boiling point is −295°C), water va-
pors are retained in DT, CO2 and acid vapors are
absorbed in AT and the arsines are freezed in CT.
The selective volatilization of the arsines trapped
is the most critical step of all. It is carried out by
activating the MWO for 5 min at its maximum
power setting (700 W), while the carrier gas flows
continuously and the signal is recorded. During
the whole process, P3 is activated and consecu-
tively it performs several functions: it removes
liquid waste from the GPS, allows the transport
of the acidified sample from the loop towards the
535
GPS, and also performs the washing procedure.
The latest is always carried out between samples
and consists of flowing water through the sample
channel for 30 s to flush away the last traces of
the previous sample, thus avoiding inter-samples
contamination and/or memory effects.
2.4.2. Procedure 2. Determination of As(III)
Arsine from As(III) was selectively generated
from a solution aliquot mixed with 1% citric acid,
which was introduced by activating V1 to an
alternate position, and the signal was obtained
following the procedure 1.
The amount of As(V) is determined by arith-
metic difference between the responses of total
inorganic arsenic obtained in procedure 1 and
As(III) obtained in procedure 2.
3. Results and discussion
3.1. Preliminary considerations
Although the generation and atomization con-
ditions for the determination of these arsenic spe-
cies have been previously studied [8], the
peculiarity of the system proposed here demand
for an exhaustive optimization of all experimental
variables. The parameters controlling each step of
the analytical process have been optimized under
the criteria of providing the best sensitivity, reso-
lution and reproducibility as well as good recover-
ies for the analyte measurement in the samples.
During the development of the analytical method
the concentration of the arsenic species in the test
solutions was maintained at 1 g l − 1, which based
on 10 ml sample volume, makes 10 ng of arsenic
injected.
It was clearly demonstrated some time ago that
the reduction reactions for arsines generation are
not only pH dependent, but also related to the
pKa values of each arsenic acid [21]. From the
physical constants values given in Table 1 it is
obvious that As(III) and DMAA will be reduced
over a wide pH range, while the other species over
a narrower range. This behavior, added to the
fact that each compound form a different arsine,
each with a different boiling point, makes possible
536 J.L. Burguera et al. / Talanta 45 (1998) 531–542
their selective volatilization under controlled reac-
tion conditions.
3.2. Effect of acids type and concentration
Different types of acids as well as concentra-
tions are described in the literature to reduce
arsenic species with NaBH4. The conclusions are
sometimes contradictory depending on the scope
of research being carried out as well as on the
experimental assembly and reaction conditions,
although kinetic factors or mixing dynamics ap-
pears to be the most reasonable explanation.
In order to find the appropriate acidic system,
which will avoid strong pH variations during the
mixing of samples with the alkaline reductant, a
series of mineral (HCl, H2SO4, HNO3, HClO4,
and H3PO4) and organic (formic, acetic, oxalic,
citric, tartaric and trichloracetic) acids were
tested. Some of these solutions have shown
promising selectivity for the arsenic species stud-
ied, but independent of the acid type, we found
that when the pH was higher than 8.5 there was
no vigorous hydrogen evolution such as occurs at
lower pH values, and consequently no response
was obtained from any of the arsenic species.
The effect of varying the mineral acids concen-
tration on the responses from the four As species
was studied. All but phosphoric acid affect simi-
larly the responses on changing acid type and
concentration in the range studied. The results
obtained, clearly show that concentrations of 2
mol l − 1 and lower, allow the evolution of arsines
from all species to almost identical extents. At
higher concentrations, the responses from As(III)
and As(V) increased, reaching rapidly a constant
value, while the other species (MMAA and
DMAA) experienced a marked decay. Further-
more, in highly acidic solutions the effervescence
due to hydrogen generation also increases causing
splashing of solution droplets on the GPS walls
and cap. Water vapor and/or reagents mist might
condense on the transfer lines and consequently
can trap the analyte gases and then release them
slowly causing loss in sensitivity and memory
effects. To reduce the amount of moisture into the
drying tube and thus avoiding its frequent chang-
ing, the acid concentration was kept as low as the
experiment requirements allowed it. In orthophos-
phoric acid the two organic species showed a
slight increase in sensitivity, although never
reached the values obtained in 2 mol l − 1 of the
other mineral acids. Therefore, this acid was not
used for further studies. Below 2 mol l − 1 of any
other mineral acid studied, it is possible to deter-
mine total arsenic in direct transfer systems, with-
out the need of sample digestion prior to analysis.
It is also possible to speciate MMAA, DMAA
and total inorganic arsenic in arrangements like
the one described in this paper. The literature
abounds with papers where HCl is generally used,
although it was repeatedly reported that its va-
pors cause devitrification of the QC [17]. There is
also some controversy on the use of H2SO4. An-
derson and co-workers [21] found that at concen-
trations above 0.1 mol l − 1 the signals for all
species decayed considerably, behavior attributed
to a series of factors including rapid reductant
decomposition at higher acid concentrations, sul-
fate interference or unknown characteristics of
degassing. This explanation is not at all convinc-
ing because it is not based on any experimental
data. However, some other authors proposed its
use below 2 mol l − 1 [22]. Above that concentra-
tion it was found to produce disproportionation
in the analysis of DMAA. It is also well known
that perchloric acid is not recommended unless
special fume hoods are available. Based on the
results shown in Fig. 2(a), and on the consider-
ations made above, we are entitled to recommend
the use of nitric acid for the speciation and quan-
titation of MMAA, DMAA and total inorganic
arsenic, although in our system, sulfuric acid also
provides good recoveries for all species.
Since both inorganic As species yield the same
hydride, it was necessary to search for a different
acidic media which would differentiate between
As(III) and As(V), thus allowing their speciation.
It was interesting to find that in all organic acids
tasted, there was a great difference between the
As(III) and As(V) responses. The generation of
the arsines from As(III) and DMAA is essentially
independent on the acidity of the reaction media
in formic, tartaric and acetic acids, while the
signals from As(V) and MMAA become negligi-
ble especially in acetic, oxalic and citric acids. It
 J.L. Burguera et al. / Talanta 45 (1998) 531–542
Fig. 2. Effect of inorganic acids (a–e), organic acids (f – j) and
buffer systems (k–n) on the absorbance signals from 10 ng As
of each arsenic species; other conditions as specified in Table
2.
appears that in these media, the rate of the reduc-
tion reaction of As(V) to As(III) decreases until it
eventually becomes much slower than the rate of
hydrolysis of NaBH4 under the same conditions
[23]. To avoid any positive contribution from
As(V) in the determination of As(III), 1% citric
acid (Fig. 2b) is recommended for further studies
in order to differentiate As(III) from As(V).
The instrumental parameters and the acids type
and concentration used for further work are listed
in Table 2.
3.3. Effect of reductant
Sodium tetrahydroborate(III) has been chosen
as the obvious reducing agent for the synthesis of
the arsines in the FI system described in this
paper. The decomposition reaction of this reagent
537
in acidic media is completed within a fraction of a
second and its rate increases when the pH de-
creases. This means that the formation of the
hydrides is even faster and apparently NaBH4 has
sufficient life to effect the reduction of the arsenic
compounds [24]. However, the separation of the
hydrides from the liquid phase takes much more
time. In our system, the mixing process is carried
out in such a way that there is no decomposition
of NaBH4 prior to its reaction with the analyte.
There is a limitation on the optimization of the
reductant flow rate and concentration due to the
generation of excess hydrogen which causes varia-
tions on the signal-to-background ratios (S/B).
Above a concentration of 3% (w/v) of reductant,
the background absorption suddenly increased,
possibly owing to the burning H2 inside the QC,
while a concentration lower than 0.5% (w/v) pro-
duces small S/B ratios for all species. The opti-
mum NaBH4 concentration was thus chosen to be
2% (w/v).
3.4. Effect of 6olumes and flow rates
The volumes of sample, acids and reductant are
controlled by the flow rates delivered by P1 and P2
and by their time of operation. The system allows
equivalent standard or sample and acid volumes
to be mixed in the loop. The volume of the
acidified sample is thus limited by the loop capac-
ity which it was chosen to allow a maximum
volume of 10 ml to be carried to the GPS each
time. By varying the flow rate of P2 from 1 to 15
ml min − 1, and operating it for 1 min, the re-
sponse increased continuously as a consequence
of increasing the amount of analyte in the GPS.
However, if an adequate control of the time is
carried out, in order to introduce the same
amount of analyte, the flow rates of sample and
acids had no effect at all, up to 7 ml min − 1.
However, for higher flow rates, the reproducibility
deteriorated, probably due to an inefficient mixing
of both solutions with the reductant in R1 and in
the GPS. For the reasons discussed above, we
have chosen a flow rate of 5 ml min − 1 for P2.
Also a variation in the reductant volume up to 4
ml, had no effect on the absorbance values, but a
bigger volume produced unpaired results due to
538 J.L. Burguera et al. / Talanta 45 (1998) 531–542
increased effervescence caused by violent hydro-
gen production. A volume of NaBH4 of 1 ml was
considered appropriate for complete reaction.
3.5. Effect of carrier gas
Helium was used in two parts of the system: (a)
to maintain a stable baseline and (b) to carry the
generated arsines from the GPS to the cryogenic
trap and to transfer the collected hydrides, once
they have been revolatilized, to the atomizer.
The baseline was stabilized by continually pass-
ing a flow of 250 ml min − 1 of He to the QC
through the TBSI. The flow of He meant to
transport the arsines to the trap is also aided by
the hydrogen by-product. Once the acidity of the
reaction media and the reductant concentration
were optimized, the carrier flow rate was no
longer critical, but, both flows have to be equal to
avoid problems of the gas reverse movement due
to pressure-built in the system.
The flow of He used to transfer the volatilized
arsines to the QC is more critical since it affects
the residence time of the analyte in the atomizer
and also the separation effectiveness. The effect of
altering the carrier gas flow rate in a range as
wide as possible (20 – 1000 ml min − 1) shows that
for high flow rates ( \ 500 ml min − 1) the hydrides
are carried out too quickly, resulting in a loss of
resolution and decrease in sensitivity. This is logi-
cal because the larger is the flow rate the lesser is
the residence time of the analyte in the atomizer.
At low flow rates ( B 50 ml min − 1) the signals
gave good reproducibility, but distorted and
broadened peaks were obtained with consequent
loss in resolution.
A compromise had to be made in choosing 250
ml min − 1 as optimum carrier gas flow rate to
transport the arsines to the QC.
3.6. Microwa6e-aided selecti6e 6olatilization
As the generated hydrides are collected in the
freezing trap, the precision of the analysis will
only be affected by an efficient energy absorption
as MW pass into the trapped arsines and their
proper transfer to the QC. Polystyrene and PTFE
tubing used for housing the liquid nitrogen and to
trap the hydrides, respectively, are good insulators
and effectively transparent to microwaves. In this
case, heat through the coils walls becomes in-
significant. As liquid nitrogen is also transparent
to the radiation, only the trapped, frozen products
will be heated and selectively volatilized from the
coil. In order to obtain good reproducibility and
reasonable resolution of the results, the different
experimental sequences have to be strictly coordi-
nated and carried out at specific time intervals.
Typical output signals obtained from the HG-
AAS system are shown in Fig. 3. The selective
volatilization profiles were similar when the ar-
senic compounds were present individually or in
admixture. The volatilization times were 90, 180
and 235 s for dimethylarsine, methylarsine and
arsine, respectively. The higher sensitivity for all
species was obtained at maximum power setting
(700 W) of the MWO, because the signals rapidly
decay at lower power settings; no signals were
obtained for none of the species at 75% of the
MWO power, probably because of an insufficient
sample heating, thus resulting in no vaporization
of the arsines. Therefore, this parameter was ex-
cluded from those experimentally optimized in
this work.
In previous works, cryogenically trapped arsi-
nes were evolved upon worming, in order of their
boiling points or according to their volatility
[10,11]. However, this was not the case in this
work, probably because the separation is carried
out using MW energy at a given frequency, which
heats the trapped hydrides by dipole rotation,
being quite different from the conventional con-
ductive heating. The sequence of arsines evolution
could be due to the contribution of two mecha-
nisms: (1) to the absorption of MW energy as it
passes into the sample, depending mainly on the
dielectric constant of each compound and (2) to
the dipole size which affects the penetration depth
of the MW energy; the larger the molecular size of
the dipole, the more it is penetrated by the MW
energy. It is obvious that the major contribution
of these two mechanisms of heating is dominated
by dipole rotation, and the heating time is shorter
for those molecules with higher dielectric con-
stant. However, the efficacy of heating by dipole
rotation depends here upon the samples charac-
 J.L. Burguera et al. / Talanta 45 (1998) 531–542 539

Fig. 3. Analytical signals obtained from inorganic arsenic (5 ng each) and its methylated species (10 ng each) using (a) 2 mol l-1 nitric
acid and (b) 1% w/v citric acid. (c) This shows the responses from a trout fishery sample. The experimental conditions were as
specified in Table 2.

teristics, in particular the inherent mobility of 3.7. Analytical figures of merit

each trapped species without ionic migration as it
is the case of the heating pattern of any ionic
solution.
Other aspects of the proposed system, which
have been studied, are the trapping coil length
and its arrangement inside the freezing tube into
the MWO. Many coil length were assessed (0.7–
2.0 m), but no difference in signal and arsines
separation performance were observed. Trapping
coils longer than 3 m caused substantial disper-
sion, resulting in a decrease in sensitivity and
deterioration of the resolution. On the other
hand, for tubing length shorter than 0.7 m, the
signals decreased due to a poor trapping effi-
ciency. The efficiency of the hydrides collected
and the resolution were better in a knitted tube
than in a straight one of the same length and
inner diameter. Thus, it was proposed to use a 1
m long trapping coil which assured a relatively
high sensitivity and reasonable good resolution
for all the arsenic species studied.
Under the optimal experimental conditions out-
lined in Table 2, the AAS signals were measured and
the concentrations of As(III), As(V), MMAA and
DMAA were calculated against the individual
arsenic species standard curves, which in all cases
were linear up to 50 ng. The sensitivity values (slopes
of the calibration curves), the lower limits of the
linear range, the detection limits (DL, expressed as
twice the background standard deviation) and the
reproducibility values are given in Table 3.
The relative standard deviation (R.S.D.%) was
calculated from the responses obtained by consec-
utively injecting 10-times a solution containing the
four species at concentrations corresponding ap-
proximately to 5-times the lower limit. The precision
of measurements at the 1 g l − 1 level was around
1.5% for all species.
The accuracy of the proposed procedure was
tasted on the bases of % recovery calculated for a
trout fishery water sample spiked with the four
540 J.L. Burguera et al. / Talanta 45 (1998) 531–542

Table 3
Analytical characteristics

Compound Slope (10−2) Lower limita (ng) Detection limita (ng) R.S.D.b (%)

As(III) 0.99 0.50 0.20 2.5

As(V) 0.98 0.60 0.24 2.8
MMAA 0.90 0.80 0.35 3.5
DMAA 0.80 1.50 0.60 3.8

a
Calculated for 10 ml total sample volume.
b
Obtained for a concentration 5-times higher than the lower limit (n = 10).

compounds at two concentration levels: 5- and tive, precise and accurate. The DL were as good as,
10-times the corresponding DL. The mean values or better than those previously obtained in manu-
obtained are given in Table 4. Substantial amounts ally operated systems using arsines trapping tech-
of the species were recovered, thus enabling this niques [9–11]. However, these fitures (Table 3)
method to be used for the analysis of real samples. could be further improved due to the possibility of
The accuracy was further tested by determining: preconcentration of the products by using a bigger
(1) the different species of arsenic in a certified volume or running several consecutive reactions.
water sample and (2) the sum of all arsenic species Several consecutive reactions have been carried
(As) were compared with those obtained by out, thus achieving preconcentration of the species
ETAAS [25], where owing to the analytical for those samples with extremely low arsenic spe-
difficulties in quantifying low arsenic concentra- cies content (e.g. MMAA concentration in the
tions, the technique of standard additions was sample from the trout fishery). For samples with
adopted. The As agreed well with the As(T) and arsenic content over the linear working range,
with the certified values of two water samples dilution is avoided by reducing the operating time
(Table 5), which is another indication of satisfac- of the P2, thus introducing less amount of analyte
tory accuracy of the proposed method. However, in the system.
the lack of materials which certify the arsenic
species and the poor sensitivity of available alter- 3.8. Applications
native methods for detection of such low concen-
trations of arsenic species, made further studies on The concentrations of total arsenic in most
the accuracy of our procedure impossible. natural water systems are reported [26,27] to be in
The analytical parameters obtained in this work the range 0.1–1.0 mg l − 1. However, in different
show that the methodology here proposed is sensi- parts of the world, the levels of arsenic are two
and even three orders of magnitude higher than
Table 4
Mean recovery values (%) for trout fishery water samplesa that of most natural waters consumed by man
spiked with the four arsenic species [28,29]. Such is the case, for example, of the
drinking water consumed in Cordoba, Argentina
Compound Concentration levelb [30], where the arsenic levels are as high as 1400 g
l − 1, while in the shallow tubewells in South Ben-
DL×5 DL×10 Total mean recovery
(%) gal, India [7,28,29], the levels are from 60 to
58/000 g l − 1. It is also reported [31] that mineral
As(III) 83.5 9 6 93.4 9 4 88.5 waters may contain up to 50-times and hot
As(V) 100.99 4 96.893 98.9 springs up to 300-times more arsenic than the
MMAA 102.095 99.0 93 100.5
normal background level. In an attempt to protect
DMAA 98.99 5 100.4 94 99.7
aquatic life as well as to establish a control over
a
Endogenous values as in Table 5. the implications of arsenic concentrations for
b
Analysis in triplicate for the species at each level. public health, different international organiza-
 J.L. Burguera et al. / Talanta 45 (1998) 531–542
Table 5
Arsenic content in different types of water
Sample As(III) Arsenic concentration (mg l-1)a
As(V)
River water
1 1.209 0.02 4.20 90.09
2 1.359 0.03 4.909 0.10
Trout fishery 1.90 9 0.4 8.50 90.20
HPS certified samplec 57.2790.43 24.73 90.30
NIST certified sampled 34.769 0.30 16.38 9 0.20
a
Results expressed as X 9St/ n (n= 5).
b
Determined by ETAAS.
c
Certified value for HPS certified reference material. Trace metals in drinking water = 80.0 94.0 mg l−1.
dt
NIST SRM 1643b, Trace elements in water. Reference value =49 ng g−1.
tions recommended the permissible dissolved total
arsenic concentration in natural waters, between 8
and 50 g l − 1.
The analytical procedures optimized in this
work have been used for the quantitation of four
arsenic species in different type of water samples:
river waters and samples from a trout fishery. The
range of arsenic concentrations of these water
samples provided an opportunity to apply our
procedure to the analysis of real samples.
The results shown in Table 5 are in general
agreement with those reported in the literature
[27,32]. In the aquatic environment, arsenic is
found predominantly in the inorganic form with
As(V) being thermodynamically stable. This is the
case for the turbulent high mountains river waters
analyzed here, where the oxidation processes are
favored under aerobic conditions. None of the
methylated species was detected in river waters,
despite the reports that these species can represent
a considerable fraction of the total arsenic present
[27,32]. The low temperatures of these waters may
inhibit the biological formation of such species. In
the sample from the trout fishery the dominant
species were As(V) and DMAA. The presence of
methylated species in this sample might be at-
tributed to unidentified biological mediation;
some authors have concluded that arsenate is
taken up by phytoplankton and subsequently con-
verted to MMAA and DMAA, probably as part
of a detoxification mechanism, and replaced back
into the water column [33 – 35]. Both, MMAA and
541
As As(T)b
MMAA DMAA
— — 5.40 5.41 90.20
— — 6.25 6.22 90.20
0.70 90.01 1.80 90.05 12.90 13.0 9 0.30
— — 82.00 81.20 9 0.50
— — 51.14 48.70 9 0.20
DMAA are stable species in the aquatic environ-
ment, but their production and distribution in
oxic surface waters is a fluid process that is sea-
sonably dependent. A systematic, long-term sam-
pling should be undertaken in order to reach
solid, well-fundamented conclusions regarding the
natural processes in such water bodies.
4. Conclusions
The development of the selective preconcentra-
tion technique proved to be a powerful tool for
the determination of extremely low arsenic levels
in water samples. One of the attractions of our
procedure was the simplicity of the equipment
which allowed high efficiency of analyte introduc-
tion, ease of preconcentration and the possibility
of speciation.
Using the manifold described in this paper the
risk of contamination and volatilization losses are
drastically reduced since the system is completely
closed and all operations are performed on-line.
The analysis is carried out with optimum sensitiv-
ity due to the preconcentration in the cryogenic
trap, which allow the analyte to be conveyed to
the atomizer in a much higher yield than is ever
attained in conventional HG. This makes the
method attractive for application to matrices with
low arsenic content like the different types of
waters. Other benefits are the increased speed of
analysis compared to similar procedures, the use
542 J.L. Burguera et al. / Talanta 45 (1998) 531–542
of microwave energy to volatilize the hydrides
added to the facility to optimize the system since
fewer parameters are involved.
This system also affords the intimate mixing of
reagents with better pH control; it appears to be
much more tolerant to elements which normally
interfere in the hydride generation and provides
the opportunity to elegantly eliminate volatile re-
action by-products.
Automation of the whole procedure, from sam-
ple loading to data report leads to enhanced
sensitivity, precision and accuracy compared to
manual methods.
Acknowledgements
The authors appreciate financial support from
BID-CONICIT (Banco Interamericano de De-
sarrollo-Consejo Nacional de Investigaciones Ci-
entı́ficas y Tecnológicas), Project QF-46.
References
[1] A.C. Aurillo, R.P. Mason, H.F. Hemond, Environ. Sci.
Technol. 28 (1994) 577.
[2] M. Burguera, J.L. Burguera, Talanta (1997) in press.
[3] Ch-Jev Hwang, S-Jen Jiang, Anal. Chim. Acta 289 (1994)
205.
[4] X-Chun Le, W.R. Cullen, K.J. Reimer, Talanta 41 (1994)
495.
[5] M.A. Lòpez-Gonzalvez, M.M. Gòmez, C. Càmara, M.A.
Palacios, J. Anal. At. Spectrom. 9 (1994) 291.
[6] O. Jimenez de Blas, V. Gonzalez, R. Seisdedos Rodriguez,
J. Hernandez Méndez, J. AOAC Int. 77 (1994) 241.
[7] A. Chatterjee, D. Das, D.K. Mandal, T.R. Chowdhury,
G. Samanta, D. Chakraborti, Analyst 120 (1995) 643.
[8] J.L. Burguera, M. Burguera, C. Rivas, P. Carrero, C.
Rondón, M.R. Brunetto, Quim. Anal. (1997) in press.
[9] P.H. Masscheleyn, R.D. Delaune, W.H. Patrick Jr., J.
Environ. Qual. 20 (1991) 96.
[10] A.G. Howard, S.D.W. Comber, Mikrochim. Acta 109
(1992) 27.
[11] P. Michel, B. Averty, V. Colandini, Mikrochim. Acta 109
(1992) 35.
[12] M.S. Mohan, R.A. Zingaro, P. Micks, P.J. Clark, Int. J.
Environ. Anal. Chem. 11 (1982) 175.
[13] J. van Elteren, H.A. Das, C.L. De Ligny, J. Agterdenbos,
D. Bax, J. Radioanal. Nucl. Chem. Articles 179 (1994)
211.
[14] S.C. Pinillos, J. Sanz Asensio, J. Galbán Bernal, Anal.
Chim. Acta 300 (1995) 321.
[15] B. Welz, M. Sucmanová, Analyst 118 (1993) 1425.
[16] P. Carrero, J.L. Burguera, M. Burguera, C. Rivas, Ta-
lanta 40 (1993) 1967.
[17] M. Burguera, J.L. Burguera, C. Rivas, P. Carrero, M.R.
Brunetto, M. Gallignani, Anal. Chim. Acta 308 (1995)
339.
[18] S. McIntosh, Perkin-Elmer Technical Summary, Order
no. TSAA-17.
[19] Z. Li, N. Zhe-Ming, S. Xiao-Quan, Spectrochim. Acta
44B (1989) 339.
[20] C. Luca, The pH and Its Applications (Rom), 2nd ed.,
Editura Tecnica, Bucuresti, Romania, 1973.
[21] R.K. Anderson, M. Thompson, E. Culbard, Analyst 111
(1986) 1143.
[22] G. Rauret, R. Rubio, A. Padró, Fresenius J. Anal. Chem.
340 (1991) 157.
[23] J. Aggett, A.C. Aspell, Analyst 101 (1976) 341.
[24] J. Agterdenbos, D. Bax, Fresenius J. Anal. Chem. 323
(1986) 783.
[25] M.E. Soares, B.L. Bastos, M. Ferreira, Analyst 120
(1995) 2367.
[26] A. Leonard, in: E. Merian (Ed.), Metals and their Com-
pounds in the Environment: Occurrence, Analysis and
Biological Relevance, VCH, New York, 1991.
[27] W.R. Cullen, K.J. Reimer, Chem. Rev. 89 (1989) 713.
[28] A. Chatterjee, D. Das, D. Chakraborti, Environ. Pollut.
80 (1993) 57.
[29] A. Chatterjee, D. Das, D.K. Mandal, T.R. Chowdhury,
G. Samanta, D. Chakraborti, Analyst 120 (1995) 925.
[30] R.M. Bergoglio, Prensa Med. Argent. 51 (1964) 994.
[31] H.G. Seiler, H. Sigel (Eds.), Handbook on Toxicity of
Inorganic Compounds, Marcel Dekker, New York, 1988.
[32] L.E. Hunt, A.G. Howard, Marine Pollut. Bull. 28 (1994)
33.
[33] M.O. Andreae, in: P.J. Craig (Ed.), Organometallic Com-
pounds in the Environment, Wiley, New York, 1986.
[34] M.O. Andeae, Deep-Sea Res. 25 (1978) 391.
[35] M.O. Andeae, Arsenic: industrial, biomedical and envi-
ronmental perspectives., in: W.H. Lederer, R.J. Fenster-
heim (Eds.), Proceedings of the Arsenic Symposium,
1981, Van Nostrand, New York, 1983.