Blank values, adsorption, pre concentration, and sample preservation for arsenic speciation of environme

Analytica Chimica Acta 512 (2004) 1–10
Blank values, adsorption, pre-concentration, and sample preservation
for arsenic speciation of environmental water samples
Jen-How Huang
a,∗
, Gunter Ilgen
b
a
Department of Soil Ecology, Bayreuth Institute for Terrestrial Ecosystem Research (BITÖK), University of Bayreuth, D-95440 Bayreuth, Germany
b
Central Analytics, Bayreuth Institute for Terrestrial Ecosystem Research (BITÖK), University of Bayreuth, D-95440 Bayreuth, Germany
Received 15 December 2003; received in revised form 26 January 2004; accepted 2 February 2004
Available online 14 April 2004
Abstract
Arsenic is the focus of public attention because of its toxicity. Arsenic analysis, its toxicity, and its fate in the environment have been
broadly studied, still its blank values, adsorption to sampling materials and pre-concentration of water samples as well as stabilization of
arsenic compounds in water samples under field conditions have been very little investigated. In this study, we investigate the blank values
and adsorption of arsenic compounds for different laboratory materials. We focused our work onto pre-concentration of water samples and
how to stabilize arsenic compounds under field conditions. When using glassware for arsenic analysis, we suggest testing arsenic blank values
due to the potential release of arsenic from the glass. Adsorption of arsenic compounds on different laboratory materials (<10%) showed
little influence on the arsenic speciation. Pre-concentration of methanol–water solutions could result in potential overestimation of arsenic
compounds concentrations. Successful pre-concentration of water samples by nitrogen-purge provides an analytical possibility for arsenic
compounds with high recoveries (>80%) and low transformation of arsenic compounds. Thus, concentrations as low as 1ng As l
−1
can be
determined. Addition of ethylenediaminetetraacetic acid (EDTA) and storage in the dark can decrease the transformation among arsenic
compounds in rainwater and soil-pore water for at least a week under field conditions.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Arsenic; Pre-concentration; Blank value; Adsorption; Stability; Speciation
1. Introduction
Today, arsenic is the focus of public attention. The WHO

guidelines for arsenic in drinking water standard decreased
with time, and in the last edition, it was 10 ␮gl
−1
(1993).
Because of the limitations in analytical techniques, 10 ␮gl
−1
is regarded as a provisional guideline value. However, this
value would be less than 10 ␮gl
−1
if based on health criteria
[1]. In the past decades, analytical techniques for arsenic spe-
ciation have developed [2]. Toxicity of arsenic compounds
to human beings and creatures, and the fate and behaviors
of arsenic compounds in the environment have been broadly
studied [3,4]. Nevertheless, some basic and relevant infor-
mation for arsenic analysis is rare, such as the arsenic blank
values and the adsorption behaviors to sampling materials.
Since arsenic in the form of As
2
O
3
is used in the glass in-
dustry [5,6], release of arsenic from glassware during exper-
∗
Corresponding author. Tel.: +49-921-555761; fax: +49-921-555799.
E-mail address: (J H. Huang).
iments cannot be excluded. Use of different flasks and vials
for arsenic compounds analysis is inevitable for sampling,
sample storage, pre-concentration, and further analytical
steps. Procedural blank values of arsenic released from con-

tainer walls may leadto an overestimation of the arsenic con-
centrations and a false impression of arsenic speciation. This
effect is especially serious for less contaminated samples.
Although arsenic in different waters has been much in-
vestigated [3], the materials of containers for sampling were
usually not well documented. Koch et al. [7] and Zheng
et al. [8] used polypropylene bottles and Guerin et al. [9]
used polycarbonate bottles for the water samples. Adsorp-
tion of organotin compounds to the container walls leads to
underestimates of their concentrations in water [10]. Similar
to organotin compounds, organic arsenic compounds may
have high affinity to polymer materials, but no information
is available.
In less contaminated water, the concentrations of arsenic
compounds are usually low. The baseline concentrations of
arsenic in rainwater (including snow) and river water are
0003-2670/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2004.02.043
2 J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10
less than 0.03 and 0.1−0.8␮gl
−1
, respectively [3]. Because
the toxicity of arsenic compounds depends on their chem-
ical forms, a false impression of speciation may lead to
a false risk assessment of the environmental water sam-
ples. Thus, a reliable pre-concentration method is a basis
for a more precise speciation analysis for the less contam-
inated samples and a sequential risk assessment. Hydride
generation enhances the sensitivity normally up to 100-fold
over the commonly used liquid sample nebulization pro-

cedures. However, several organic arsenic compounds do
not form volatile hydrides under the tetrahydroborate treat-
ment. Methods have been developed to convert them to
hydride-forming species, but are usually complicated and
troublesome [2]. Elizalde-Gonzalez et al. [11] used natural
zeolites for pre-concentration of arsenic species in the wa-
ter samples, but this method was only tested with inorganic
arsenic. Arsenic enrichment by solvent evaporation has of-
ten been applied for methanol–water extracts of biological
samples, but little is reported about the validity of these
pre-concentration procedures.
Many efforts have been made to preserve arsenic com-
pounds in their original forms in the environmental samples
[2]. Unlike other environmental waters, the sampling of
rainwater and soil-pore water normally requires long-term
collection in the field for a certain period or for sampling
certain amounts. Although the storage of the samples in a
frozen state is the most highly recommended procedure [12],
these conditions are not very practical for preserving rain-
water and soil-pore water in the field. Therefore, a practical
stabilization of the arsenic compounds under field conditions
for several days is essential for such samples. Acidification
of samples with nitric acid, hydrochloric acid or acetic acid
has been applied to decrease microbial activity [13,14].
However, interference with the arsenic speciation may oc-
cur and Hall et al. [15] reported an immediate oxidation of
As(III) after addition of either nitric acid or hydrochloric
acid in spiked river water. Oxidation of As(III) by oxygen is
increased by several order of magnitudes due to the presence
of Fe(II) and UV radiation light [16]. Therefore, storage of

the samples to protect against UV radiation does not reduce
only the microbial activity, but also prevents photo-oxidation
due to UV exposure [17]. Addition of ethylenediaminete-
traacetic acid (EDTA) and the storage of the samples in
opaque bottles can stabilize arsenate and arsenite in ground-
water at 20
◦
C for more than 3 months, and is superior to the
addition of hydrochloric acid, nitric acid, and sulfuric acid
[18].
The objectives of this study are: (1) to identify the poten-
tial sources of the arsenic blank values as caused by arsenic
released from different laboratory materials, (2) to investi-
gate the adsorption behavior of different arsenic compounds
on different laboratory materials, (3) to evaluate different
pre-concentration methods for aqueous samples to provide
a more precise speciation of arsenic compounds in less con-
taminated water samples, (4) to evaluate the validation of
the pre-concentration method for methanol–water solutions,
and (5) to test the stabilization effect of arsenic compounds
provided by EDTA in the dark using different rainwater and
soil-pore water samples under conditions similar to those in
the field.
2. Experimental
2.1. Instrumentation
A liquid chromatograph (LC) (BIOTEK Instruments,
USA), consisting of a gradient pump (System 525), cap-
illary PEEK tubing (0.25 mm i.d.) and a 200-␮l injection
loop (Stainless Steel), and a LC autosampler 465 (Kontron
Instruments, Germany) was connected to an anion-exchange

column (IonPak AG7 and AS7, both Dionex) and cou-
pled to an inductively coupled plasma-mass spectrometer
(ICP-MS) (Agilent 7500c, Japan), equipped with a concen-
tric nebulizer (Glass Expansion, Australia) and a Scott-type
glass spray chamber.
The separation was performed at a flow rate of 1 ml min
−1
,
using a nitric acid gradient between pH 3.4 and 1.8. The
dipotassium salt of benzene-1,2-disulfonic acid (0.05mM)
was added to the eluent as an ion-pairing reagent. At the
outlet of the separation column, an internal standard (10␮g
Ge l
−1
in 0.01M nitric acid) was added by means of a
Y-connector.
Determination of total arsenic in liquid samples was con-
ducted directly with the ICP-MS, using germanium (10 ␮g
Ge l
−1
) as an internal standard.
Detection limits for arsenic species and the total arsenic
were calculated as three times the standard deviation from
instrumental blank values.
2.2. Reagents
Arsenate (As(V)), arsenite (As(III)), and dimethylarsinic
acid (DMA) were purchased from Merck (Darmstadt,
Germany). Arsenbetaine (AsB) was obtained from Fluka
and monomethylarsonic acid (MMA) and arsenocholine
(AsC) from Argus Chemicals, Italy. De-ionized water, used

throughout the work, was purified in a Milli-Q system
(Millipore Corp., Milford, MA).
Individual stock solutions (50mg As l
−1
) of As(III),
As(V), MMA, DMA, AsB, and AsC were prepared
in Milli-Q water, and stored at 4
◦
C in the dark. A
multi-compound working solution with a total concentration
of 5 ␮gAsl
−1
was prepared before each use by diluting the
stock solutions with Milli-Q water.
2.3. Release of arsenic from different materials
The tested bottles of different materials were filled with
Milli-Q water or 10% nitric acid, and incubated in the dark
at room temperature for 24 h. After incubation, 1 ml of the
solution was analyzed for total arsenic using the ICP-MS.
J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10 3
2.4. Adsorption of arsenic compounds to different materials
To 50ml artificial rainwater (containing 11.6 mg l
−1
NH
4
NO
3
, 7.85 mg l
−1
K

2
SO
4
, 1.11 mg l
−1
Na
2
SO
4
,
1.31 mg l
−1
MgSO
4
·7H
2
O, 4.32mgl
−1
CaCl
2
) was added
50 ␮l the stock solution of mixed arsenic compounds solu-
tion with each 5mg As l
−1
to adjust the concentrations of
each arsenic compounds to 5 ␮gAsl
−1
. Fifty milliliters of
the artificial rainwater was used to fill the bottles of different
materials. The bottles were shaken in the dark at 5 ± 1

◦
C
and 20±1
◦
C for 24 h, and 1 ml of the solution was taken for
the analysis of the arsenic compounds using LC–ICP-MS.
High-density polyethylene (translucent, 250ml), polypro-
pylene (translucent, 250 ml), Teflon FEP (transparent,
250 ml), and polycarbonate (transparent, 125 ml) bottles
were purchased from Nalgene, USA. Glass bottles (trans-
parent, 250ml) were purchased from Schott, Germany.
Teflon PFA bottles (translucent, 250 ml) and polyethylene
bottles (translucent, 250 ml) were bought from Vitlab, Ger-
many and VWR, Germany, respectively. These bottles were
either new or used very rarely. For the adsorption experi-
ments, the bottles were kept incompletely filled to prevent
the solution from making contact with the cap, which is
made of different materials from the bottle, in the cases of
glass, polycarbonate, and Teflon FEP bottles. All the bottles
were cleaned with detergents and distilled water at 70
◦
C.
2.5. Freeze-dry pre-concentration
Five and ten milliliters of mixed arsenic compound solu-
tions in the polyethylene tubes each with 5 ␮gAsl
−1
were
frozen at −40
◦
C in the dark. The frozen samples were then

freeze-dried to dryness or until ca. 1 ml samples were left.
For the dry samples, 1 ml of Milli-Q water was used to rinse
the tubes, and then, for determining the arsenic compounds
using LC–ICP-MS. For samples with ca. 1 ml remaining,
the sample thawed in the dark at room temperature, and the
arsenic compounds were determined using LC–ICP-MS.
2.6. Nitrogen-purge pre-concentration
Nitrogen-purge pre-concentration was conducted using a
Turbo Vap II (Zymark, USA). In principle, nitrogen-purging
at 1bar evaporates solvents such as methanol and water
either from the methanol–water solutions or as just aque-
ous samples. A water bath was used to control the sample
temperature during pre-concentration at maximum 60
◦
C.
For the recovery tests in methanol–water solutions, a 10ml
methanol–water solution (90% methanol, v/v) of mixed
arsenic compound solutions each with 5 ␮gAsl
−1
were
pre-concentrated at 25, 30, 40, 50, and 60
◦
C. For the re-
covery tests in the aqueous samples, 5 and 10ml mixed
arsenic compound solutions each with 5 ␮gAsl
−1
were
pre-concentrated at 25, 30, and 40
◦
C, preventing the degra-

dation and transformation of the organic arsenic compounds
at high temperatures. Pre-concentration was stopped auto-
matically when the solution volume had reached 0.7 ml.
An additional 0.3ml of Milli-Q water was added to rinse
the tube wall to desorb residual arsenic compounds. After
pre-concentration, the residual solutions were determined
by LC–ICP-MS as mentioned above.
2.7. Speciation analysis of arsenic compounds in
environmental water samples
Rainwater, soil-pore water, and river water were collected
in November 2003 from a remote site, the Lehstenbach
catchment in NE Bavaria, Germany. Polyethylene samplers
were placed 1m above the ground in opaque polyethylene
tubes to exclude sunlight. Samplers were installed under the
canopy for through-fall sampling and at an open site for bulk
precipitation. Soil-pore water from the forest floor and the
mineral soil were collected by lysimeters at 20 and 90cm
depth, respectively. The water samples were afterwards fil-
tered through a membrane filter and pre-concentrated imme-
diately, stored at 4
◦
C, and analyzed by the with LC–ICP-MS
within 48 h.
2.8. Stability of arsenic compounds in rainwater and
soil-pore water
Rainwater (bulk precipitation and through-fall) and
soil-pore water (20 and 90cm) were collected at the same
event, but not filtered. Through-fall usually has a higher
concentrations of cations, anions, and dissolved organic
carbon (DOC) than bulk precipitation as a result of canopy

washout. Soil-pore water usually has a higher ionic strength
and is more strongly buffered than rainwater. Soil-pore
water sampled from forest floors contains higher concen-
trations of cations, anions, and DOC as compared to that
sampled from mineral soil [19].
The sampled rainwater and soil-pore water were spiked
with different arsenic compounds immediately after sam-
pling. For this, we used an amount of each 5␮gAsl
−1
and
1.25 mM EDTA, and incubated the samples in the dark at
20 ± 1
◦
C for seven days. A set of control samples were
spiked in parallel with the same amount of arsenic com-
pounds, and incubated under the same conditions without
addition of EDTA.After incubation, all samples were filtered
with membrane filter and then, analyzedby the LC–ICP-MS.
3. Results and discussion
3.1. Some sources of blank values
We found significant amounts of total arsenic in the tap
water. In contrast, total arsenic in the Milli-Q water was
below the detection limit (Table 1). Although the concen-
tration of total arsenic in the tap water was low (0.13␮g
As l
−1
) compared to the WHO limit of drinking water stan-
dard (10 ␮gAsl
−1
), using tap water for rinsing bottles and

4 J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10
Table 1
Blank values (ng As l
−1
) of total arsenic in Milli-Q water and tap water,
and release of total arsenic from different laboratory materials
Sample
Milli-Q water <DL
Tap water 130 ± 11
Polyethylene
a
<DL
Polycarbonate
a
<DL
Glass 1
a
<DL
Glass 2
a
16 000 ± 1300
Glass
b
<DL
Glass
HNO
3
c
14 ± 5.9
Mean values and S.Ds. of three replicates are shown; Detection limit

(DL) of total arsenic: 10ng As l
−1
.
a
Incubated with Milli-Q water for 24h.
b
Rinsing with Milli-Q water after 10% nitric acid bath, and incubated
with Milli-Q water for 24h.
c
Incubated with 10% nitric acid for 24h.
as a solvent for arsenic trace analysis should be avoided to
prevent probable contamination.
Release of arsenic from plastic bottles (e.g. polyethylene
and polycarbonate) and from most of the glassware was not
detected when Milli-Q water was used. However, a large
amount of arsenic was released as As(V) (16 ␮gAsl
−1
)in
the Milli-Q water in one particular case. Generally, when
glassware was first incubated in a 10% nitric acid bath and
then rinsed several times with Milli-Q water, no release of
arsenic was found in the Milli-Q water from the glassware.
Thus, although the use of As
2
O
3
in the glass industry [5,6]
suggests the potential occurrence of arsenic in glassware,
only one case of release of arsenic from glassware was iden-
tified. For arsenic analysis, special care, such as testing ar-

senic blank values, should be taken when using glassware.
Small amounts of arsenic were detected when the glass-
ware was incubated with 10% nitric acid. In this case, there
are two possible sources to the arsenic; leaching from glass-
ware by nitric acid or the blank value originated from nitric
acid itself. Therefore, use of glassware and cleaning glass-
ware with nitric acid is not recommended.
3.2. Adsorption of arsenic compounds on different
materials
Most of the spiked arsenic compounds showed an adsorp-
tion of <5% on glass and the different polymer materials at
Fig. 1. Recoveries of arsenic compounds from bottles of different materials at 5
◦
C. Mean values and S.Ds. of three replicates are shown. (᭿) Polyethylene;
(
) high-density polyethylene; ( ) polypropylene; (ᮀ) glass; ( ) Teflon FEP; ( ) Teflon PFA; and ( ) polycarbonate.
0
50
100
150
200
As(III) MMA DMA As(V) AsB AsC
Recoveries (%)
Fig. 2. Recoveries of arsenic compounds from the bottle of polyethylene
(
᭿
) and glass (ᮀ)at20
◦
C. Mean values and S.Ds. of three replicates
are shown.

5
◦
C (see Fig. 1). Only MMA had a slightly higher adsorp-
tion on all materials, but it was still below 10%. We have
also tested the adsorption of arsenic compounds on glass
and polyethylene at 20
◦
C, and obtained similar results as
at 5
◦
C (see Fig. 2). However, a transformation of As(V) to
As(III) during the batch experiment was observed.
The adsorption experiments indicated negligible adsorp-
tion of both inorganic and organic arsenicals to different
materials. The temperature variation seemed to have little
influence on the adsorption of arsenic compounds to differ-
ent materials. Underestimation of concentrations of arsenic
compounds, caused by adsorption to container walls, should
be low.
3.3. Pre-concentration of water samples by freeze-drying
We tried to pre-concentrate the spiked water samples by
freeze-drying. However, recoveries of arsenic compounds
in the aqueous solutions after freeze-drying were low, and
the loss of arsenic compounds related directly to the time
used to freeze-dry the samples (Table 2). In all cases, re-
coveries of the total arsenic in the solution decreased after
freeze-drying, suggesting the loss of arsenic compounds in
the freeze-drying process.
As(III) always had the lowest recoveries and As(V)
had the highest recoveries compared to the other arsenic

compounds. Ellwood and Maher [20] reported that As(III)
concentrations determined in freeze-dried and air-exposed
sediments were much lower than in sediments that were not
freeze-dried with minimum air exposure. Exposure of the
J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10 5
Table 2
Recoveries (%) of arsenic compounds in aqueous solutions pre-concentrated by freeze-drying
As(III) MMA DMA As(V) AsB AsC
Freeze-dry 24 h, starting with 10 ml 26.1 ± 17.9 41.0 ± 32.9 57.7 ± 17.5 79.4 ± 11.9 57.2 ± 16.7 51.9 ± 16.0
Freeze-dry until ca. 1 ml solution
left, starting with 10ml
a
54.4 ± 1.10 67.4 ± 6.28 68.5 ± 6.60 71.0 ± 10.9 67.4 ± 6.09 68.3 ± 6.08
Freeze-dry until ca. 1 ml solution
left, starting with 5ml
b
74.1 ± 26.8 81.1 ± 17.9 82.8 ± 18.1 83.9 ± 19.5 80.3 ± 17.3 79.5 ± 16.4
Mean values and S.Ds. of three replicates are shown.
a
Listing for ca. 13h.
b
Listing for ca. 8h.
Fig. 3. Recoveries of arsenic compounds after pre-concentration starting with methanol–water solution (90% methanol, v/v) using nitrogen-purge method.
(
᭿) Arsenite; (
) monomethylarsonic acid; ( ) dimethylarsinic acid; (ᮀ
) arsenate; (
) arsenbetaine; and ( ) arsenocholine. Mean values and S.Ds. of
three replicates are shown.
samples to the air prior to analysis seems to oxidize As(III)

into As(V). This effect may lead to a false impression of
the true speciation within the sample.
3.4. Pre-concentration of methanol–water solutions by
nitrogen-purge
After pre-concentration of methanol–water solution (90%
methanol, v/v) by nitrogen-purge, the arsenic compounds,
especially As(III) and MMA, had recoveries of >100% ac-
companied by large standard deviations (see Fig. 3). This
phenomenon was more apparent when the pre-concentration
was conducted at lower temperatures (25 and 30
◦
C) than
that at higher temperatures (50 and 60
◦
C).
It is well established that addition of carbon (as methanol)
to aqueous solutions improves the ionization efficiency
of arsenic in the plasma [21,22]. Kohlmeyer et al. [22]
demonstrated that adding methanol could enhance the ar-
senic signals in the LC–ICP-MS. We tested the influence
of methanol concentration on signals for the different ar-
senic compounds. The signals increased generally with the
increase in methanol concentrations (see Fig. 4). However,
As(III) and MMA signals were much more enhanced com-
pared to the other arsenic compounds when the methanol
concentrations were less than ca. 50% but leveled-off when
the methanol concentrations were between 50 and 100%.
Besides, the As(III) and MMA peaks were more close to
each other, broadened, and then, overlapped as the methanol
concentration in solution increased (see Fig. 5). Since the

retention time of methanol in our LC program was very
close to the As(III) and MMA peaks (see Fig. 6), especially
As(III), the enhancement of arsenic signals by addition
of methanol was thereby in the order: As(III)  MMA
 the other arsenic compounds. We also suspected that
large amounts of methanol in the eluent interfered with the
separation of As(III) and MMA.
There seems to be significant amounts of methanol left as
a residue in the pre-concentrated methanol–water extracts.
The effect was only slightly reduced with the increasing
pre-concentration temperature. According to our results in
this section, pre-concentration of methanol–water solutions
at all temperatures may result in overestimating the concen-
trations of As(III) and MMA in the samples. Therefore, we
suggest either to avoid using methanol for extraction or spe-
cial care should be taken when calibrating the arsenic com-
pounds concentrations.
300
600
900
1200
020406080100
Methanol(%v/v)
Intensity
As (III) MMA
DMA As(V)
AsB AsC
Fig. 4. Response of arsenic compounds as a function of methanol con-
centration. Mean values and S.Ds. of three replicates are shown.
6 J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10

0
2000
4000
6000
8000
10000
12000
14000
0 100 200 300 400 500 600 700 800 900
Retention time (s)
Intensity (cps)
0%
30%
60%
100%
0
2000
4000
6000
8000
80 100 120 140 160
Retention time (s)
Intensity (cps)
As(III)
aMMA
DMA
As(V)
AsC
aAsB
As(III)

aMMA
Fig. 5. Liquid chromatogram of arsenic compounds with each 5 ␮gAsl
−1
in methanol–water solutions (0, 30, 60, and 100% methanol, v/v).
0
5000
10000
15000
20000
25000
0 100 200 300 400 500 600 700 800 900
Retention time (s)
Intensity (cps)
4000
0
8000
12000
16000
As75 (left axis)
Ar40 C13 (right axis)
As(V)
DMA
AsB
MMA
As(III) AsC
Fig. 6. Liquid chromatogram of 5␮gAsl
−1
arsenic standard and trace monitor of m/z 53.
3.5. Pre-concentration of water samples by nitrogen-purge
Pre-concentration using a nitrogen-purge showed recov-

eries of >80% of all arsenic compounds at different tem-
peratures and different pre-concentration ratios (see Figs. 7
Fig. 7. Recoveries of arsenic compounds after pre-concentration using nitrogen-purge method. (a) Starting with 5 ml solution; and (b) starting with 10 ml.
(
᭿) Arsenite; (
) monomethylarsonic acid; (
) dimethylarsinic acid; (ᮀ) arsenate; ( ) arsenbetaine; and ( ) arsenocholine. Mean values and S.Ds. of
three replicates are shown.
and 8). As(III) and AsC always had the lowest recover-
ies among the arsenic compounds and under all conditions
tested. In contrast, As(V) and AsB were enriched in most
cases, reflecting the probable oxidation of As(III) to As(V)
and AsC to AsB [12]. Even though the evaporation of water
J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10 7
0
50000
100000
150000
200000
250000
0 100 200 300 400 500 600 700 800 900
Retention time (s)
Intensity (cps)
original
5 times concentrated
10 times concentrated
As(III)
aMMA
aDMA
AsB

aAsC
As(V)
Fig. 8. Liquid chromatogram of arsenic compounds with each 5 ␮gAsl
−1
in an aqueous solution, after pre-concentration (5–1 ml and 10–1 ml, respectively)
using nitrogen-purge at 30
◦
C.
was done by nitrogen-purge, we can not completely exclude
the exposure of arsenic compounds to O
2
in the atmosphere
during pre-concentration. Five times pre-concentration (see
Fig. 7a) showed less transformation of As(III) to As(V) than
10 times pre-concentration (see Fig. 7b). This may support
the exposure of arsenic compounds to air, because the 10
times pre-concentration procedure (ca. 9 h) lasts twice as
long as that for the five times (ca. 4 h).
Comparing the results from freeze-drying and nitrogen-
purge, the use of the latter is concluded to be superior to
freeze-drying to pre-concentrate water samples, because it
is more convenient, has higher recoveries, and less arsenic
transformation. Although the recoveries of arsenic com-
pounds pre-concentrated with nitrogen-purging varied little
with temperature, we recommend pre-concentrating water
samples at low temperatures. Pre-concentration at higher
temperatures may enhance the risk of transformation among
arsenic compounds.
Combining pre-concentration with nitrogen-purge and
LC–ICP-MS allows the detection of arsenic compounds in

the water samples at the 1ng Asl
−1
level. It is particu-
larly suitable for the environmental water samples with low
arsenic concentrations.
0
200
400
600
800
1000
0 100 200 300 400 500 600 700 800 900
Retention time (s)
Intensity (cps)
10 times concentrated
original
As(III)
DMA
As(V)
AsC
Fig. 9. Liquid chromatogram of arsenic compounds in the river water before and after 10 times pre-concentration using nitrogen-purge method.
3.6. Determination of arsenic compounds in less
contaminated water samples
We determined the concentrations of arsenic compounds
in the less contaminated rainwater, soil-pore water, and river
water. The concentrations of total arsenic in these water sam-
ples were all below 1 ␮gAsl
−1
with the dominance of inor-
ganic arsenic, As(III), and As(V). Remarkable amounts of

organic arsenic were observed in through-fall and soil-pore
water from the forest floor with MMA, DMA, AsB or AsC
up to 290 ng Asl
−1
(Table 3). Organic arsenic compounds
were found only in trace amounts in bulk precipitation, min-
eral soil-pore water, and river water.
The pre-concentration was most effective in the case of
mineral soil-pore water and river water. No organic arsenic
compounds could be detected in mineral soil-pore water and
river water without pre-concentration, due to the their ex-
tremely low concentrations (see Fig. 9). The concentrations
of arsenic compounds determined in pre-concentrated sam-
ples were not far from those determined in the original sam-
ples (recoveries >80%), demonstrating the validation of the
pre-concentration; however, transformation of small amount
among arsenic compounds seems inevitable. Except prob-
8 J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10
Table 3
Concentrations of arsenic compounds in rainwater, river water, and soil-pore water, and recoveries (%) of arsenic compounds as determined on original
samples and on samples pre-concentrated 10 times
As(III) MMA DMA As(V) AsB AsC Total As Organic As (%)
Bulk precipitation
a
(ng As l
−1
) 82.8 <DL 13.9 242 <DL <DL 339 4.11
Bulk precipitation
b
(ng As l

−1
) 76.2 <DL 15.3 219 <DL <DL 312 4.92
Recovery (%) 94.0 – 110 83.5 −− 92.0
Through-fall
a
(ng As l
−1
) 90.7 <DL 291 452 34.9 58.5 928 41.4
Through-fall
b
(ng As l
−1
) 88.4 <DL 285 440 35.4 52.3 901 41.3
Recovery (%) 97.4 – 97.8 97.2 102 89.5 97.1
Soil-pore water 20cm
a
(ng As l
−1
) 88.3 20.9 130 387 35.2 41.5 702 32.4
Soil-pore water 20cm
b
(ng As l
−1
) 77.7 21.0 127 422 34.1 38.3 719 30.6
Recovery (%) 87.9 101 97.7 109 96.7 92.2 102
Soil-pore water 90cm
a
(ng As l
−1
) 11.4 <DL <DL 114 <DL <DL 125 0.00

Soil-pore water 90cm
b
(ng As l
−1
) 12.6 <DL 5.3 107 <DL <DL 124 4.28
Recovery (%) 110 – − 93.9 −− 99.7
River water
a
(ng As l
−1
) 176 <DL <DL 811 <DL <DL 987 0.00
River water
b
(ng As l
−1
) 181 <DL 5.6 850 <DL 5.4 1040 1.06
Recovery (%) 103 – − 105 −− 106
Mean values of three replicates are shown and all S.Ds. are <5%. (−): Recovery not applicable.
a
Determining on original sample, DL: 10ng As l
−1
.
b
Determining on 10 times pre-concentration solution, DL: 1ng As l
−1
.
able transformation of the arsenic compounds during the
pre-concentration process, transformation of arsenic com-
pounds may occur before analysis due to the contact of the
samples with air or reduction caused by DOC in the water

samples [18]. Since the concentrations of arsenic compounds
in these samples were very low, a slight transformation may
already cause remarkable inaccuracy. Thus, storage of the
treated samples at low temperature (e.g. 4
◦
C) before anal-
ysis or immediate analysis of the samples is required.
Usually, DMA and MMA were the most common organic
arsenic compounds in the rainwater and soil-pore water, but
in small proportions [23,24]. Similarly, only trace amounts
of DMA were found in the bulk precipitation. However, con-
siderable amounts of organic arsenic compounds were ob-
served in the through-fall and soil-pore water from the forest
floor (40 and 30%, respectively), including trace amounts
of the more complicated AsB and AsC. The microbial ac-
tivity in the phyllosphere and forest floor is usually high
[19,25], suggesting these organic arsenic compounds may
0
4000
8000
12000
16000
0 100 200 300 400 500 600 700 800 900
Retention time (s)
Intensity (cps)
original
after incubation
As(III)
aMMA
aAs(V)

aAsB
DMA
a?
a?
aAsC
Fig. 10. Liquid chromatogram of arsenic compounds spiked with each 5 ␮gAsl
−1
in through-fall before and after 7 days incubation without addition
of EDTA.
be formed by in situ methylation. The proportion of organic
arsenic decreased to trace amounts in the mineral soil-pore
water and inorganic arsenic compounds were enriched in
the river water. Arsenic compounds may undergo different
transformations and transport processes in the soils. How-
ever, these can not be explained only using our results and
available knowledge. More investigations about the trans-
formation and transportation of arsenic in the forest ecosys-
tems are necessary, to gain a more clear sight of the biogeo-
chemical fate and the behavior of arsenic in the terrestrial
environments.
3.7. Stability of arsenic compounds in rainwater and
soil-pore water
The chromatogram in Fig. 10 showed a typical transfor-
mation of arsenic compounds in unfiltered rainwater and
soil-pore water without any treatment. Large amounts of
As(V) were converted into As(III), which might be caused
J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10 9
Table 4
Recoveries (%) of spiked arsenic compounds in rainwater and soil-pore water with addition of 1.25mM EDTA, and incubated at 20
◦

C in the dark for 7
days
As(III) MMA DMA As(V) AsB AsC
Bulk precipitation 108 ± 17.9 101 ± 2.93 88.8 ± 1.77 81.9 ± 26.8 85.9 ± 1.78 98.9 ± 3.10
Through-fall 105 ± 10.9 85.2 ± 0.88 89.4 ± 1.60 91.6 ± 7.10 92.1 ± 4.09 93.6 ± 1.08
Soil-pore water 20cm 101 ± 2.17 106 ± 3.15 94.1 ± 0.89 96.9 ± 5.32 101 ± 2.33 106 ± 4.21
Soil-pore water 90cm 106 ± 4.65 86.8 ± 6.31 94.5 ± 2.57 95.2 ± 2.49 95.2 ± 1.73 90.2 ± 0.79
Mean values and S.Ds. of three replicates are shown.
by reduction by DOC [18]. The sum of inorganic arsenic
compounds was higher than the original amount after 7
days incubation, suggesting the degradation of organic ar-
senic species. Although MMA and DMA were suggested to
be much more stable [26], we have also observe decreased
amounts of MMA. Nevertheless, we found that DMA is also
much more enriched after incubation. Since DMA was the
degradation intermediate of AsB and AsC [12], the abun-
dance of DMA after incubation was caused by the degrada-
tion of AsB and AsC. The results from the control experi-
ment indicate the essential steps to reduce transformation of
arsenic compounds during storage in the field.
With the addition of 1.25 mM EDTA to rainwater and to
soil-pore water, transformation of arsenic compounds and
degradation of organic arsenic compounds during the in-
cubation was successfully reduced (Table 4). A loss of or-
ganic arsenic compounds (<15%) and transformation be-
tween As(V) and As(III) (<20%) were observed, especially
in the case of rainwater. There may be some variation among
rainwater and soil-pore water samples; of course, this we
have not accounted for here.
EDTA chelates metal cations, buffers the sample pH, and

reduces the microbial activity [27]. The stabilization effect of
EDTA was shown to be superior to that of mineral acids [18].
Our results demonstrated the stabilization effect of arsenic
compounds provided by EDTA and in the dark in unfiltered
rainwater and soil-pore water.
4. Conclusions
The use of laboratory glassware for arsenic analysis
should be careful because of potential release of arsenic.
Adsorption of arsenic compounds on different laboratory
materials had little influence on the arsenic speciation.
Pre-concentration of methanol–water solutions could re-
sult in potential overestimation of arsenic compounds
concentrations. Pre-concentration of aqueous samples by
nitrogen-purge provides an analytical possibility for arsenic
compounds at the level of 1 ng As l
−1
, and has high recov-
ery and low transformation. Addition of EDTA and storage
in the dark can reduce the transformation among arsenic
compounds in rainwater and soil-pore water under field
conditions.
Acknowledgements
The authors would like to thank Björn Berg for helpful
comments on this manuscript. Financial support was given
by the German Academic Exchange Program (DAAD) and
the German Ministry of Science and Education (BMBF,
Grant No.: BEO 0339476D).
References
[1] WHO, 2001. Environmental health criteria 224, arsenic and ar-
senic compounds (available at />ehc/ehc/ehc224.htm).

[2] Z. Gong, X. Lu, M. Ma, C. Watt, X.C. Le, Talanta 58 (2002) 77.
[3] P.L. Smedley, D.G. Kinniburgh, Appl. Geochem. 17 (2002) 517.
[4] D.C. Adriano, Trace Elements in Terrestrial Environments. Bio-
geochemistry, Bioavailability and Risk of Metals, second ed.,
Springer-Verlag, 2001.
[5] A.B. Mukherjee, R. Zevenhoven, P. Bhattacharya, in: Proceedings
of the Seventh International Conference on the Biogeochemistry
of Trace Elements, 15–19 June 2003, Uppsala, Sweden, vol. 2,
Symposia, pp. 98.
[6] M. Bissen, F.H. Frimmel, Acta Hydroch. Hydrob. 31 (2003) 9.
[7] I. Koch, J. Feldmann, L. Wang, P. Andrewes, K.J. Reimer, W.R.
Cullen, Sci. Total Environ. 236 (1999) 101.
[8] J. Zheng, H. Hintelmann, B. Dimock, M.S. Dzurko, Anal. Bioanal.
Chem. 377 (2003) 14.
[9] T. Guerin, N. Molenat, A. Astruc, K. Pinel, Appl. Organometal.
Chem. 14 (2000) 401.
[10] J H. Huang, Int. J. Environ. Anal. Chem. 84 (2004) 255.
[11] M.P. Elizalde-Gonzalez, J. Mattusch, R. Wennrich, J. Environ. Monit.
3 (2001) 22.
[12] M.A. Palacios, M. Cómez, C. Cámara, M.A. López, Anal. Chim.
Acta 340 (1997) 209.
[13] Y. Talmi, D.T. Bostick, J. Chromatogr. Sci. 13 (1975) 231.
[14] B. Daus, J. Mattusch, R. Wennrich, H. Weiss, Talanta 58 (2002) 57.
[15] G.E.M. Hall, J.C. Pelchat, G. Gauthier, J. Anal. Atom Spectrom. 14
(1999) 1157.
[16] M.T. Emitt, G.H. Khoe, Water Res. 35 (2001) 649.
[17] D.A. Polya, P.R. Lythgoe, F. Abou-Shakra, A.G. Gault, J.R. Brydie,
J.G. Webster, K.L. Brown, M.K. Nimfopouloss, K.M. Michailidis,
Miner. Mag. 67 (2003) 247.
[18] A.J. Bednar, J.R. Garbarino, J.F. Ranville, T.R. Wildeman, Environ.

Sci. Technol. 36 (2002) 2213.
[19] E. Matzner. Biogeochemistry of forested catchments in a chang-
ing environment: a case study in NE-Bavaria, Germany, Ecological
Studies, Springer-Verlag, in press.
[20] M.J. Ellwood, W.A. Maher, Anal. Chim. Acta 477 (2003) 279.
[21] U. Kohlmeyer, J. Kuballa, E. Jatnzen, Rapid Commun. Mass Spec-
trom. 16 (2002) 965.
10 J H. Huang, G. Ilgen /Analytica Chimica Acta 512 (2004) 1–10
[22] R. Ritsema, L. Dukan, T.R. Navarro, W. van Leeuwen, N. Oliveira,
P. Wolfs, E. Lebret, Appl. Organomet. Chem. 12 (1998) 591.
[23] R. Prongratz, Sci. Total Environ. 224 (1998) 133.
[24] R.J. Bowell, N.H. Morley, Appl. Geochem. 9 (1994) 15.
[25] B. Stadler, S. Solinger, B. Michalzik, Oecologia 126 (2001) 104.
[26] W.R. Cullen, K.J. Reimer, Chem. Rev. 89 (1989) 713.
[27] L. Xie, N.S. Hettiarachchy, M.E. Jane, M.G. Johnson, J. Food Sci.
68 (2003) 268.