Anal Bioanal Chem (2012) 403:28252834

DOI 10.1007/s00216-012-6006-7

ORIGINAL PAPER

Development and validation of an SPE HG-AAS method

for determination of inorganic arsenic in samples
of marine origin
Rie R. Rasmussen & Rikke V. Hedegaard &
Erik H. Larsen & Jens J. Sloth

Received: 14 December 2011 / Revised: 27 March 2012 / Accepted: 29 March 2012 / Published online: 21 April 2012
# Springer-Verlag 2012

Abstract The present paper describes a novel method for the

quantitative determination of inorganic arsenic (iAs) in food
and feed of marine origin. The samples were subjected to
microwave-assisted extraction using diluted hydrochloric acid
and hydrogen peroxide, which solubilised the analytes and
oxidised arsenite (AsIII) to arsenate (AsV). Subsequently, a pH
buffering of the sample extract at pH 6 enabled selective
elution of AsV from a strong anion exchange solid-phase
extraction (SPE) cartridge. Hydride generation atomic absorption spectrometry (HG-AAS) was applied to quantify the
concentration of iAs (sum of AsIII and AsV) as the total arsenic
(As) in the SPE eluate. The results of the in-house validation
showed that mean recoveries of 101104% were achieved for
samples spiked with iAs at 0.5, 1.0 and 1.5 mgkg1, respectively. The limit of detection was 0.08 mg kg1, and the
repeatability (RSDr) and intra-laboratory reproducibility
(RSDIR) were less than 8% and 13%, respectively, for samples
containing 0.2 to 1.5 mg kg1 iAs. The trueness of the SPE
HG-AAS method was verified by confirming results obtained
Published in the special paper collection Recent Advances in Food
Analysis with guest editors J. Hajslova, R. Krska, M. Nielen.
Electronic supplementary material The online version of this article
(doi:10.1007/s00216-012-6006-7) contains supplementary material,
which is available to authorized users.
R. R. Rasmussen : R. V. Hedegaard : E. H. Larsen : J. J. Sloth (*)
National Food Institute, Division of Food Chemistry,
Technical University of Denmark,
Mrkhj Bygade 19,
2860 Sborg, Denmark
e-mail: jjsl@food.dtu.dk
Present Address:
R. V. Hedegaard
Faculty of Life Sciences, Department of Food Science,
University of Copenhagen,
1958 Frederiksberg, Denmark

by parallel analysis using high-performance liquid chromatography coupled to inductively coupled plasma mass spectrometry. It was demonstrated that the two sets of results were
not significantly different (P<0.05). The SPE HG-AAS method was applied to 20 marine food and feed samples, and
concentrations of up to 0.14 mg kg1 of iAs were detected.
Keywords Inorganic arsenic . Marine samples . SPE HGAAS . Microwave-assisted extraction . Food and feed
control . Speciation

Introduction
Arsenic (As) is a ubiquitous metalloid found in soils,
groundwater, surface water, air, and consequently also in
various food items. It is released into the environment from
natural sources such as volcanic activity and weathering of
minerals as well as from anthropogenic sources including
ore smelting, burning of coal, and the use of As-containing
pesticides and growth promoters [1]. Arsenic has a complex
chemistry, especially in the marine environment, where
more than 50 different naturally occurring As-containing
compounds have been identified, comprising both organic
and inorganic forms [2]. Examples of arsenic compounds
found in the marine environment are shown in Table 1. In
fish and other seafood, As is bioaccumulated and is present
predominantly as the organoarsenical arsenobetaine (AB),
which is generally assumed to be of no toxicological concern to humans [3]. Other organoarsenic species include,
e.g. methylarsonate (MA), dimethylarsinate (DMA), arsenocholine and tetramethylarsonium ion, which are usually
found only as minor constituents in marine samples and
generally believed to be of minor toxicological concern
[3]. Furthermore, arsenosugars are the predominant As

2826

R.R. Rasmussen et al.

Table 1 Examples of arsenic compounds found in the marine environment

For simplicity, the compounds are depicted in their fully deprotonated form. Names and acronyms are as proposed by Francesconi and Kuehnelt [2]
and Sele et al. [5]

compounds in algae [4], and in fatty fish, lipid-soluble As

compounds (arsenolipids) have been reported as major
forms of As [5]. The toxicity data on arsenosugars are
limited but indicate a low human toxicity, and for arsenolipids, data on their toxicity are still lacking [6]. Inorganic
arsenic consists of arsenite in oxidation stage three (AsIII)
and arsenate in oxidation stage five (AsV). The content of
iAs in seafood is generally low, especially in fish where
concentrations below detectable levels (<0.001 mg kg1) are
often reported [7, 8]. In shellfish, somewhat higher levels
are usually found, and in a few cases, very high levels
(>0.5 mg kg1) have been reported in bivalves [9, 10].
Several studies have demonstrated that iAs exposure may
lead to a range of adverse effects, including various cancers
and skin lesions. Consequently, when dietary exposure to
As was recently evaluated by the European Food Safety
Authority (EFSA) [3] and by the FAO/WHO Joint Expert
Committee on Food Additives (JECFA) [11], emphasis was
on iAs exposure, whereas exposure to organoarsenic species
was not considered. Both evaluations provided estimates of
toxicological intake limits for iAs as a Benchmark Dose
Level (BMDL), 0.38 g(kgbw)1day1 for cancers of
the lung, skin and bladder as well as skin lesions (EFSA

BMDL01) and 27 g(kgbw)1day1 for lung cancer

(JECFA BMDL0,5). Due to limited specific data on the
content of iAs in food, several assumptions were made on
the basis of the total As content in the foodstuff. Hence, in
both reports, an urgent need for further data on iAs in food
commodities was called upon in order to improve the background data for future risk assessment analysis.
Arsenic, either as total or inorganic, in foodstuffs is
currently not regulated by the European Union (EU) [12],
whereas Australia and New Zealand have established national maximum levels (MLs) of 12 mg kg1 for the content of iAs in marine foods [13]. In contrast, the EU
regulation on undesirable substances in animal feed sets
maximum contents for total arsenic in a number of feed
commodities (240 mg kg1 at 12% moisture content) [14].
Interestingly, in a footnote in this EU directive, it is stated
that, upon request from the competent authorities, the responsible operator must perform an analysis to demonstrate
that the content of iAs is less than 2 mg kg1 in the feed and
feed-ingredients of marine origin. One of the main reasons
for not having implemented MLs for iAs directly in the
directive instead of MLs for total arsenic is a lack of standardised methods to enable reinforcement of the regulation.

Determination of inorganic arsenic by SPE HGAAS

The same need for development of validated methods of

analysis for specific determination of iAs has also been
emphasised by EFSA (2009) [3] and WHO (2011) [11] in
their recent evaluations. The reason for absence of regulatory
limits for iAs has been described in the context of a so-called
vicious circle, where the various activities in the process
towards legislation on species (i.e. iAs) level are highly interlinked (Fig. 1). The legislators on one hand rely on reliable
risk assessment to base their risk management decisions
concerning MLs on. This, on the other hand, requires reliable
analytical data for the iAs content in food- and feeding-stuffs
that the risk assessors can use in their evaluation process.
Development of validated methods of analysis for iAs is a
prerequisite for the generation of reliable analytical data, and
legislation is often an encouraging factor for method development and validation. Hence, the vicious circle is completed,
and the challenge is how to break it.
Determination of arsenic species has been comprehensively reviewed by Francesconi and Kuehnelt in 2004 [2]. A
number of different extraction solvents have been applied
for arsenic speciation analysis in marine samples with water
or methanol or mixtures thereof as the most commonly used.
Other approaches include enzymatic [15] and ionic extractants [16] in order to improve the extraction of iAs by
breaking bonds to thiol groups in the proteins, which have
an affinity to AsIII [17]. The extraction procedures typically
apply mixing/shaking, sonication or microwave-assisted extraction (e.g. microwave-assisted heating in closed chambers). Currently, high-performance liquid chromatography
coupled to inductively coupled plasma mass spectrometry
(HPLC-ICP-MS) is the most commonly used technique for
arsenic speciation analysis, due to low limits of detection
and high selectivity of this detector even when analyzing
real sample matrices. The challenge in chromatographic
separation of arsenic species lies within their chemical nature, including different charges and pKa values, molecular
sizes and functional groups. Several liquid chromatography
approaches have been applied for the separation of arsenic
compounds in combination with inductively coupled plasma
mass spectrometry (ICP-MS) as an arsenic-specific detector,
with anion exchange chromatography as the predominant
technique of choice [1820].
Toxicological evaluation
(Risk assessment)
Legislation
(Risk management)

Analytical data

Method development
Fig. 1 The circle of progress towards accurate and reliable methods
for the determination of inorganic arsenic. Reprinted from [20]

2827

The recent increased emphasis on dietary iAs exposure,

partly encouraged by the recent EFSA [3] and WHO [11]
evaluations, have initiated several studies on specific determination of iAs levels in foodstuffs [7, 19, 21], in all of which
iAs is determined by HPLC-ICP-MS. However, many laboratories involved in feed and food control do not have the ICPMS technique available, mainly due to the relatively high cost
of investment for this technology compared with other elemental detection techniques like, e.g. AAS and atomic fluorescence spectrometry. Consequently, in an enquiry by the
European Committee for Standardization, the majority of
experts in the working group on contaminant analysis in
animal feed preferred a future standardised method for determination of iAs in animal feed to be based on AAS detection
(following an off-line separation step) rather than a method
based on HPLC-ICP-MS [22].
As a consequence, the present study was initiated with
the aim to develop an analytical method for the selective
determination of the toxicologically relevant iAs in food and
feed of marine origin based on off-line separation by solidphase extraction (SPE) followed by AAS detection. The
current study presents the method development as well as
the method performance characteristics obtained from an inhouse validation study on marine food and feed samples. To
our knowledge, this is the first report describing the combined use of SPE and AAS as a novel speciation approach
for specific iAs analysis in marine food and feed samples.

Experimental
Chemicals and reagents
The following chemicals and reagents were used to prepare
the solutions used for the experimental work, including extraction of samples, the SPE and HPLC separations as well as
the hydride generation atomic absorption spectrometry (HGAAS) and ICP-MS detection procedures. Methanol (MeOH)
was HPLC-grade (Rathburn, Walkerburn, Scotland, UK) and
acetic acid (CH 3 COOH) pro-analysis quality (Merck,
Darmstadt, Germany). The 6769% nitric acid (HNO3) and
3437% hydrochloric acid (HCl) were Plasma Pure-grade
(SCP Science, The Netherlands); 30% hydrogen peroxide
(H2O2) and 25% ammonia (NH3) were Suprapure grade
(Merck); sodium hydroxide (NaOH) was Ph Eur grade
(Merck), and ammonium carbonate ((NH4) 2CO 3)) was
Puratronic grade (Alfa Aesar, Karlsruhe, Germany). Sodium
tetrahydroborate (NaBH4) (Fluka, Buchs, Switzerland), Lascorbic acid (Fluka) and potassium iodide (KI) (Merck) were
all pro-analysis quality. The 30% silicone antifoam emulsion
in water was obtained from Sigma-Aldrich (Steinheim,
Germany). Water was ultra-purified using a Millipore system
(Molsheim, France).

2828

Serial dilutions of arsenite (AsIII) standard stock solution

(1,000 mg l1) from Plasma Cal (SCP Science, France) were
prepared daily in water. Standard solutions of the following
arsenic species were prepared in water/sodium DMA trihydrate (Merck, Hohenbrunn, Germany), disodium MA hexahydrate (Chem Service, West Chester, PA, USA) and AB
(BCR CRM626, Institute for Reference Materials and
Measurements; IRMM, Geel, Belgium).
Samples, sample preparation and extraction
Trout samples and oyster samples were sampled at Danish fish
farms by the Danish Veterinary and Food Administration as
part of their ongoing food control and monitoring programme.
For both sample types, each sample consisted of approximately 1 kg in total. The samples were homogenised, and subsamples were freeze-dried, homogenised and ground to a fine
powder using a commercial coffee-grinder (Braun Aromatic
KSM 2, Germany). Fish feed and fish meal samples were
kindly provided by a feed company (Nutreco, Boxmeer, The
Netherlands) from their normal sample flow. Two natural
incurred marine-based reference materials TORT-2 (Lobster
Hepatopancreas) and DORM-3 (Dogfish muscle) certified for
total As at 21.61.8 and 6.880.30 mg kg1, respectively,
were obtained from the National Research Council of Canada
(Ontario, Canada).
In acid-cleaned quartz vessels, 0.20 g subsamples (dry
weight) were extracted with 10 ml weak acidic aqueous
mixture (0.06 M HCl and 3% H2O2) assisted by microwave
energy. The programme of the microwave oven (Multiwave,
Anton Paar GmbH, Austria) was adjusted to maintain temperature of 90 C for 25 min followed by a 7-min cooling
step. The extracts were transferred to 15-ml polypropylene
tubes. After 10 min centrifugation (2,500g, 10 C), the
sample extracts were transferred to clean plastic tubes,
which were stored at 5 C for up to 7 days prior to analysis.
SPE HG-AAS detection
Inorganic arsenic was selectively separated from other arsenic compounds using strong anion exchange SPE. Optimal
performance was obtained using Strata SAX, silica-based
SPE cartridge (Phenomenex, 6 ml, 500 mg 55 m, 70 )
using a flow rate of ~1 ml min1. The procedure required
first a pre-conditioning of the cartridge with 2 ml MeOH and
equilibration with 2 ml buffer (20 mM (NH4)2CO3, 0.03 M
HCl and 1.5% H2O2), before loading 4.0 ml of the buffered
sample (1:1 mixture of the centrifuged sample extract and
40 mM (NH4)2CO3 buffer) adjusted to a pH in the range
5.07.5. Subsequently, the SPE cartridges were washed with
3.0 ml of 0.5 M acetic acid and finally eluted by 1.25 ml of
0.5 M HCl. The final eluates were pre-reduced before the
(total) arsenic concentration was determined by HG-AAS.

R.R. Rasmussen et al.

Pre-reduction included mixing of 1 ml SPE eluate with 7 ml

reduction solution (30 mM KI, 28 mM L-ascorbic acid,
0.1% (v/v) silicone, 3 M HCl), which was allowed to react
for 1 h at room temperature. Furthermore, 6 ml of 3 M HCl
was added, and it was left to react for another hour at room
temperature prior to HG-AAS measurement. The AsIII
standards and the SPE sample eluates were pre-reduced
simultaneously.
The HG-AAS system (Thermo Fisher Scientific,
Cambridge, UK) was operated according to the manufacturers instructions (instrumental parameters are summarised
in Table 2). The autosampler (Cetac ASX-260) introduced
the pre-reduced sample to the vapour generator (VP100)
where it reacted with freshly prepared NaOH/NaBH 4
(0.5%w/v) and HCl (4.7 M) solutions to generate gaseous
arsenic hydrides (AsH3) from AsIII. The hydrides were
swept to a gasliquid separator using argon gas and transported through a dryer tube (61022 mm id, Perma pure
model MD-110-24F-4, Toms River, USA) to the electrically
heated open-ended T-shaped silica cell (Vapour Furnace
EC90) of the atomic absorption spectrometer (iCE 3000).
Data were collected by the SOLAAR software.
For SPE development, two other strong anion exchange
cartridges were tested: the HyperSep Retain-AX Polymeric
SPE and the HyperSep Strong Anion Exchanger (SAX)
both from Thermo Fischer.

Table 2 Typical operating conditions of the HG-AAS system

Instrument mode

Absorbance

Element specific lamp

Arsenic

Wave length (nm)

Slit width (nm)
Electrical heated cell (C)

193.7
0.5
900

Burner height (mm)

17

Delay time (s)

Measurement time (s)
Replicates
Signal
Background correction
Sample introduction
Pump speed (rpm)
Sample flow rate (ml min1)
NaBH4 flow rate (ml min1)
HCl flow rate (ml min1)
Argon gas pressure (psi/bar)
Carrier gas flow (ml min1)
Wash time (s)
HG-AAS gas dryer connection tube

50
4
3
Continuous
On D2 QuadLine
Autosampler
34
8
1.9
1
5/0.34
125
6.6
Perma Pure MD-series,
fluorocarbon
Compressed air

Dryer gas

Determination of inorganic arsenic by SPE HGAAS

HPLC-ICP-MS reference method

A method based on HPLC-ICP-MS was used for comparison of the results obtained by the SPE HG-AAS approach. The sample extracts (0.06 M HCl, 3% H2O2)
were filtered through 0.45 m polytetrafluoroethylene filters in Mini-UniPrep HPLC vials (Whatman International,
Maidstone, Kent, UK) prior to analysis. Aliquots of the
extract (25 l) were injected onto the HPLC-ICP-MS
system. Separation of AsV from other arsenic species
was obtained on a polymer-based strong anion exchange
column (ICSep ION-120; 10 m, 4.6120 mm) equipped
with a guard column (Transgenomics, San Jose, CA,
USA) by isocratic elution (1.0 ml min1 at 30 C) using
an Agilent 1100 series HPLC system with a quaternary
pump and autosampler (Agilent Technologies, Waldbronn,
Germany). The mobile phase was prepared daily by dissolving ammonium carbonate (40 mM) in 3% (v/v) aqueous methanol solution followed by adjustment of pH to
10.3 with 25% (v/v) aqueous ammonia and subsequently
filtration through a 0.45 m polyvinylidene difluoride
filter (Millipore, Denmark) prior to use. The HPLC run
time was 15 min. The method has previously demonstrated its ability for separation of arsenate from other organoarsenic compounds and used for selective determination of
inorganic arsenic in seafood samples [8].
The HPLC-ICP-MS (Agilent 1100 HPLC and Agilent
7500ce ICP-MS, Agilent Technologies, Waldbronn,
Germany) was equipped with a MicroMist concentric nebuliser (Agilent) and a Scott type double-pass water-cooled
spray chamber (Agilent). The outlet of the HPLC column
was connected to the nebuliser by a short length of PEEK
tubing (0.13 mm id). Data was collected using the Agilent
Chemstation ICP-MS chromatographic software. Typical
plasma conditions were 1,500 W RF power, 15 lmin1
plasma gas, 0.97 lmin1 carrier gas and 0.17 lmin1 makeup
gas. Analysis was performed in the time-resolved analysis
mode monitoring ion traces at m/z 75 (for 75As) and m/z 35
(for 35Cl) with 1 and 0.01 s integration time per data point,
respectively. Any drift in sensitivity of the HPLC-ICP-MS
system was adjusted by injecting an AsV standard for every
six to eight samples. Chloride and and other major organoarsenic species were chromatographically resolved from
AsV and did not interfere with the analysis [8].
Sample analysis
The sample extracts were analysed by both the SPE HGAAS and HPLC-ICP-MS approaches as described above.
Quantification was done by the use of external calibration
standards matching the final sample solvent composition.
Eight standards 0; 0.239.15 g l1 for HG-AAS and 0; 2
60 g l1 for HPLC-ICP-MS (both corresponding to 0; 0.1

2829

4 mg kg1) were analysed twice: in the beginning and at the

end of each sequence. Calibration curves were obtained by
plotting the response of the analyte against the concentration. The method of least squares was applied to draw the
best straight line though the data points. The reagent blank
and background level of the spiked sample material were
subtracted the relevant samples before the recovery was
calculated.
Method validation
Validation data were collected in three series (one for each of
trout, oyster and fish feed), which were performed on separate
days by two different laboratory technicians. The spiked samples had a natural iAs content below 0.14 mg kg1. Each
series comprised 24 samples, i.e. marine samples spiked
with AsIII at three levels (0.5, 1.0 and 1.5 mg kg1), each
analysed in triplicates, one reagent blank, two reference
materials (TORT-2 and DORM-3) analysed in duplicates,
single determinations of seven unspiked samples (of either
trout, oyster and fish feed) and three other marine
samples.
From the results obtained, the relative standard deviation
under repeatability conditions (RSDr), intra-laboratory reproducibility conditions (RSDIR) and recovery was calculated for each compound according to ISO guidelines [23]
and as described by the Nordic Committee on Food
Analysis (NMKL) [24]. RSDr and RSDIR represent the
variation between repeated extractions and analysis within
days and between days, respectively. The obtained RSDIR
values were compared with the maximum values recommended by the European Commission [25], which can be
calculated by the Horwitz equation (2(10,5 log C)) for the
specific concentration level (C in milligrammes per kilogrammes). The limit of detection (LOD) and limit of quantification (LOQ) were determined as three and six times the
standard deviation at intra-laboratory conditions (SDIR) divided by the recovery, both based on results from the lowest
spike level.

Results and discussion

The SPE HG-AAS method procedure can be divided into
three separate steps, comprising extraction, separation and
detection. In the following, the outcome of the experiments
conducted for each step in the method development phase is
described. The final developed method was subsequently inhouse-validated, and the method performance characteristics are described as well as comparison of obtained data
with the reference method based on HPLC-ICP-MS. Finally,
results from the application of the method on a range of
marine food and feed samples are presented.

2830

R.R. Rasmussen et al.

Extraction
The aim of the extraction procedure was to achieve a high
extraction efficiency without compromising species integrity, i.e. no interconversions between iAs and organoAs species. Furthermore, the extraction solvent should be
compatible with the SPE cartridges.
In a previous study, an alkalinealcoholic extraction solvent (0.9 mol l1 sodium hydroxide in 50% (v/v) aqueous
ethanol) and microwaves were used for the extraction of iAs
in marine samples [26]. However, the alkaline extracts had a
poor compatibility with the SPE cartridges, due to the high
pH (see the SPE separation section below). Furthermore,
this extraction solvent extracted less iAs for TORT-2 than
the weak acidic aqueous solvent (0.06 M HCl, 3% H2O2)
applied in the present study (Table 3). This may be due to a
higher extraction efficiency of iAs by water compared with
organic solvents as previously reported [2]. To facilitate
oxidation of AsIII to AsV H2O2 was added. The efficiency
of the oxidation was verified by HPLC-ICP-MS analysis,
which revealed that only AsV was present in sample extracts
spiked with AsIII.
Average recoveries of 101104% were achieved for marine samples spiked with AsIII (Table 3). Hence, no irreversible protein binding was formed as reported by Larsen et al.
[26] and Munoz et al. [27], despite the high affinity of AsIII
for thiol-groups in proteins [17]. However, since no certified
reference materials are available for iAs, the extraction
efficiency of iAs from natural incurred samples is difficult
to evaluate [28]. Several studies have determined the iAs
concentration in the CRM TORT-2 (certified for total arsenic), however, no consensus value (concentration range,
0.091.1 mg kg1) is reported as recently compiled by
Lefroy et al. [19, 26]. In contrast, the determination of iAs
in rice seems not to depend on the extraction approach
applied as recently demonstrated in a large interlaboratory

study on iAs determination in rice, where consistent results

were obtained among different approaches, including the
procedure described in the present study [29].
The stability of three organoarsenic species: AB, MA
and DMA during the extraction step was evaluated by
subjecting standard solutions of the compounds to the
microwave procedure. AB is the predominant arsenical is
most seafood and usually constitutes more than 8090%
of the total arsenic present in, e.g. fish. MA and DMA
are typical minor arsenicals in seafood but structurally
very similar to arsenate (AsV), differing only by one and
two methyl groups, respectively (Table 1). No degradation or conversion of the AB, MA and DMA standards
were observed during the sample extraction. Other arsenic compounds of interest in marine samples include
arsenolipids, arsenosugars and thio-arsenic compounds.
The arsenolipids are not extracted by the aqueous extractant used in the present study and will not appear in the
extracts. The stability of (oxo-) arsenosugar compounds
in a comparable acidic environment (0.078 M HCl) has
previously been systematically investigated by Gamble et
al. [29], and no degradation to inorganic arsenic was
observed. The same accounts for the thio-arsenosugars,
which presence have been reported in algae and bivalves
[30], since they will convert to their oxo-analogues when
exposed to an oxidising environment [31] like the presence of
the strong oxidant H2O2 in the extractant solution, and
hence not be present in the extracts. Additionally, no
significant change in the iAs levels during extraction of
real samples was found, since the HPLC-ICP-MS analysis
of extracts gave ~100% recovery for the marine samples
spiked with iAs (unpublished data). The insurance of no
interspecies transformation to inorganic arsenic is of
course highly important when assessing the extraction
efficiency, so a higher value is not misinterpreted as
increased extraction efficiency.

Table 3 Results of the SPE HG-AAS validation

Spike

Reference material

Low

Medium

High

TORT-2

DORM-3

Level (mgkg1)
Observations (n)

0.5
9

1.0
9

1.5
9

0.840.03a
6

0.190.01a
6

Mean recovery (%)

Repeatability RSDr (%)
Reproducibility RSDIR (%)
RSDHorwitz (%)

101
4
5
18

103
8
9
16

104
5
6
15

107
3
9
16

94
7
13
20

Determination of inorganic arsenic in two reference materials from fish (TORT-2 and DORM-3) and in trout, oyster and fish feed spiked at three
levels. Number of spiked samples (n), mean recovery and the relative standard deviation of the repeatability (RSDr) and the in-house reproducibility
(RSDIR) are reported. RSDHorwitz is the Horwitz relative standard deviation
a

Reference value determined by HPLC-ICP-MS (meanstandard deviation, n06)

Determination of inorganic arsenic by SPE HGAAS

2831
100

SPE separation

90
80

pH 6
pH 8
pH 10

70

Retained on SPE (%)

From a previous study, it was known that AsV is retained on

a strong anion exchange HPLC column at pH 10.3 and
thereby separated from other coextracted organoarsenic
compounds present in fish and seafood [8]. The separation
mechanism in anion exchange chromatography is based on
different charges (pKa values) of the arsenic species at a
specific pH (Electronic supplementary material, Table S1).
The AsV has a more pronounced anionic character compared with the other arsenic compounds at the same pH
and is thus more efficiently retained on anion exchange
sorbent. Since the two inorganic arsenic species (AsIII to
AsV) have different anionic properties, the oxidation incorporated in the extraction step was necessary for separation
of inorganic arsenic from organoarsenic compounds using
strong anion exchange SPE.
The SPE cartridges applied should ideally be (1) compatible with the extract solvent, (2) able to retain arsenate on the
stationary phase, (3) release organoarsenic in the washing step
and (4) commercially available. In the initial steps of the SPE
method development, three different strong anion exchange
SPE cartridges were tested for their ability to retain arsenate.
The pH in the extracts clearly affected SPE retention of
arsenate. The polymer-based HyperSep Retain-AX retained
on average 84% AsV (5% standard deviation, n04) at a pH
of 6 and 10, whereas at higher and lower pH, the retention was
inferior. For the silica-based Strata SAX, SPE quantitative
retention was obtained at pH 6 and 8, but, at higher pH, no
retention was observed, probably due to incompatibility of the
silica-material with alkaline solvent. At pH 4, the Strata SAX
cartridge only partly retained AsV (Fig. 2). From these experiments, it was decided to select the Strata SAX cartridges and a
pH value of 6 for further optimisation work. HyperSep SAX
performed similarly to the Strata SAX for AsV retention at the
different pH values. Our results are in line with previous
studies [32, 33] where SPE was applied for the selective
retention and elution of arsenic species in various water and
urine samples.
Application of aqueous standard solutions of the organoarsenic species AB, MA and DMA on the SPE cartridges
resulted in partial retainment on the SPE cartridge (Fig. 2)
with MA having the highest retention (80%) at pH 6 due to
its anionic properties. The three organoarsenic species were
quantitatively eluted by acetic acid in the subsequent washing step, without simultaneous elution of AsV. Finally, a
small volume of hydrochloric acid was used for the elution
of AsV (elution step) and collected for subsequent analysis
of As. The arsenic content in the SPE eluates (load, wash
and sample) was also evaluated for real samples using
HPLC-ICP-MS. The separation procedure in a real sample
is illustrated in Fig. 3, which shows overlaid HPLC-ICP-MS
chromatograms from the loading-, washing- and elution

pH 4

60
50
40
30
20
10
0
AB

MA

DMA

As (V)

Arsenic species

Fig. 2 Retention of arsenic species on Strata-SAX SPE when loaded

as standard solutions in 5 mM NH4, 3% H2O2. iAsV was tested at
pH 410 and AB, MA and DMA at pH 4 and 6. Mean value with
standard deviation for duplicate experiments presented

steps, respectively, of the SPE separation of a TORT-2

extract. Here, it can be seen that AsV eluted exclusively in
elution-step, whereas the other arsenic species eluted in the
fractions from the loading- and washing steps.
HG-AAS
The content of iAs was determined (as total As) in the SPE
eluates by HG-AAS following the same principles applied
in a European standardised method for the determination of
total arsenic in animal feeding stuffs [34]. HG-AAS detects
volatile active arsenicals (e.g. hydrides). Since AsV has a
Signal intensity 75As (cps)

Organic arsenic

SPE fractions
Load
Wash
Eluate

75,000
60,000
45,000

Inorganic arsenic (V)

30,000
15,000
0

Time (min)

Fig. 3 Overlaid HPLC-ICP-MS chromatogram of three SPE fractions

(load, wash and eluate) of a seafood sample (TORT-2) containing both
inorganic and organoarsenic species

2832

Method performance
The SPE HG-AAS method was in-house-validated using
spiked and naturally incurred marine samples. The spike
levels in the samples were chosen near the maximum levels
for iAs in marine foods in Australia and New Zealand at 1
2 mg kg1 [13], currently the only countries with regulation
on iAs in marine foods. This concentration range is also
relevant for the animal feeding-stuff regulation in the EU
where a maximum level at 2 mg kg1 is indicated in a
footnote in the EU directive on undesirable substances in
animal feeding-stuffs [14]. Hence, the samples were spiked
at 0.5, 1 and 1.5 mg kg1, respectively, in order to validate
the method at levels, which would be relevant for future
food and feed control purposes.
A summary of the validation results are shown in Table 3.
With the current SPE HG-AAS method, detected were 0.90
0.07 mg kg1 in TORT-2 and 0.180.02 mg kg1 in DORM-3
(meanstandard deviation, n06). The method performance
was evaluated by comparison with the criteria in the EU
Commission decision concerning the performance of analytical methods and the interpretation of results for detection of
residues in products of animal origin [25], and the SPE HGAAS method complied with all the performance criteria listed.
The recovery was estimated from the spike experiments at
three different levels, and mean recoveries at 101104% were
obtained, which are within the range of 80% to 110% for
trueness requested by the European Commission [25]. The
repeatability was calculated from replicate measurements,
and low values were obtained for RSDr at 38%. The
intra-laboratory reproducibility (RSDIR) values were less
than 13% for samples containing 0.2 to 1.5 mg kg1
iAs. This level of precision was satisfactory since the RSDIR
values did not exceed the Horwitz equation values
(RSDHorwitz) [36] as requested by the European Commission
[25] (Table 3). The recoveries and RSDs were within the same
range for low, medium and high spike and for trout, oyster and

fish feed, which were analysed in three different analytical

series.
A linear response was obtained up to 4.6 g l 1
(corresponding to 2 mg kg1), with correlation coefficient
(R2) values0.997 for the external calibration curves made
of standards containing the same concentration of potassium
iodide, ascorbic acid and hydrochloric acid as that of the
samples. The spiking experiments with AsIII resulted in
quantitative recoveries (101104%, Table 3) indicating no
matrix effects. The iAs concentration in reagent blanks was
low showing that no interfering compounds were introduced
during the extraction and separation procedures. Plastic
containers were also applied whenever possible in order to
avoid potential contamination with arsenate as previously
reported [37, 38].
The LOD and LOQ were 0.08 and 0.16 mg kg1, respectively, and were calculated as three and six times, respectively,
the standard deviation at intra-laboratory reproducibility conditions (SDIR) divided by the average recovery both at the
lowest spike level (0.5 mg kg1). At LOQ level, acceptable
values for trueness (94%) and reproducibility (13%) were
obtained as illustrated by analysis of the natural incurred
DORM-3 sample containing 0.19 mg kg1 (Table 3). If the
LOD instead was calculated as three times the standard deviation at repeatability conditions (SDr) based on results from
repeated injections of a standard solution with a low concentration (0.1 g l1 corresponding to 0.04 mg kg1) as described by the Nordic Committee on Food Analysis [24],
LOD of 0.02 mg kg1 was obtained. However, this LOD
3.5

y= x
3.0

HPLC-ICPMS (mgkg-1)

lower volatile formation hydride generation efficiency (and

consequently lower sensitivity) than AsIII [35] the sample
eluates were pre-reduced to AsIII prior to analysis. The
volatile species generating organoarsenicals include arsenosugars [35], MA, DMA and trimethylarsine oxide [2], all
having lower-generation efficiency compared with AsIII
[35]. However, in the present procedure, the organoarsenic
compounds eluted in the loading and washing step of the
SPE separation procedure and are consequently not present
in the elution fraction to interfere with the iAs determination. Furthermore, results from spiking experiments with
AsIII in samples containing naturally incurred (organo-)
arsenic species showed quantitative recoveries verifying
the absence of other volatile active species in the detection
step (Table 3).

R.R. Rasmussen et al.

2.5

2.0

1.5

1.0

0.5

0.0
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

SPE HG-AAS (mgkg-1)

Fig. 4 Measurement of inorganic arsenic by two different methods;

HPLC-ICP-MS and SPE HG-AAS. In total, 72 blank, spiked and
natural incurred marine samples were analysed. The correlation is y0
x (95% confidence interval, regression analysis)

Determination of inorganic arsenic by SPE HGAAS

represents mainly the instrument performance and does only

to a lesser degree reflect the variances introduced in the full
analytical procedure. The measuring range was 0.16 to
2.00 mg kg1 and was defined as the concentration range from
the LOQ to the highest calibration standard, which was included in the linear response. The obtained LOD at 0.08 mg/
kg is more than an order of magnitude lower than the present
legislation levels for marine food and feed samples at 1
2 mg kg1, and the method will thus be suitable for control
purposes. On the other side, the obtained analytical limits may
be too high to enable monitoring of low level seafood/feed
products. Detection levels may be improved by increasing the
sample load on the SPE column or choosing alternative methodologies with higher detection power (e.g. HPLC-ICPMS) if
accurate determination at low levels is the purpose of the
study.
All sample extracts (from naturally incurred, spiked and
certified reference samples) were analysed by both the SPE
HG-AAS and the HPLC-ICP-MS reference methods (Fig. 4).
The results obtained by the two different methods were in
good agreement with each other and were not significantly
different (95% confidence level, using a paired t test and
regression analysis), verifying the high specificity and accuracy of the developed SPE HG-AAS technique. Although a
tendency towards lower results for the SPE HG-AAS method
is seen for samples in the low concentration range (up to
0.16 mg kg1), these results are below the LOQ of the method
and hence outside the validated measuring range.
Seafood and marine feed samples
In order to test the applicability of the method on real samples,
it was applied to the determination of inorganic arsenic in a
range of different seafood and feed samples, including oyster,
trout, fish meal and compound fish feed. In oyster samples
(N07), the iAs concentration was up to 0.14 mg kg1, and in
all samples of trout (N011), fish feed (N01) and fish meal
(N01), concentrations below the LOD (0.08 mg kg1) were
found. The contents were in accordance with previously
reported results for samples of marine origin, including fish
(<0.0020.006 mg kg1) [7], shellfish (0.0010.26 mg kg1)
[19] and fish feed (0.010.06 mg kg1) [39]. The iAs levels
detected in the oyster and fish samples were much lower than
the MLs for fish (2 mg kg1) and molluscs (1 mg kg1) in
Australia and New Zealand [13]. Similarly, the contents in the
fish meal and fish feed samples were much lower than the
2 mg kg1 ML in a footnote of the EU feed directive [14].

Conclusion
A novel method approach based on SPE HG-AAS for the
determination of inorganic arsenic in marine food and feed

2833

was developed and subsequently successfully singlelaboratory-validated. Results from the validation study gave
acceptable average recoveries in the range of 94107% and
repeatability and reproducibility RSD values at 38% and
513%, respectively, which were lower than the Horwitz
values at the corresponding concentration levels. The
obtained LOD at 0.08 mg/kg was more than an order of
magnitude lower than the present legislation levels for marine food and feed samples at 12 mg kg1. The method was
applied to the determination of iAs in 20 marine food and
feed samples where only low levels were detected (up to
0.14 mg kg1 in oysters). In conclusion, the SPE HG-AAS
method presents a novel speciation approach, which uses
inexpensive instrumentation (HG-AAS) and which is fit for
purpose as a candidate for future control analysis of marine
food and feed samples.
Acknowledgements Funding for this study was provided by the
European Communitys Seventh Framework Programme (FP7/20072013) under grant agreement n 211326 and the Committe Europenne
de Normalisation (CEN).

References
1. Smedley PL, Kinnibrugh DG (2002) A review of the source,
behaviour and distribution of arsenic in natural waters. Appl
Geochem 17:517568
2. Francesconi KA, Kuehnelt D (2004) Determination of arsenic
species: a critical review of methods and applications, 2000
2003. Analyst 129:373395
3. EFSA (European Food Safety Authority) (2009) Scientific opinion
on arsenic in food, EFSA Panel on Contaminants in the Food
Chain (CONTAM). EFSA J 7(10):1351:11351:199
4. Francesconi KA, Edmonds JS (1997) Arsenic and marine organisms. Adv Inorg Chem 44:147189
5. Sele V, Sloth JJ, Lundebye AK, Larsen EH, Berntssen MHG,
Amlund H (2012) Arsenolipids in marine oils and fats: a review
of occurrence, chemistry and future research needs. Food Chem
133:618630
6. Francesconi KA (2010) Arsenic species in seafood: origin and
human health implications. Pure Appl Chem 82:373381
7. Julshamn K, Nilsen BM, Frantzen S, Valdersnes S, Maage A,
Nedreaas K, Sloth JJ (2012) Survey on total and inorganic arsenic
in more than 900 fish samples from Norwegian waters. Food Addit
Contam B submitted
8. Sloth JJ, Larsen EH, Julshamn K (2005) Survey of inorganic
arsenic in marine animals and marine certified reference materials
by anion exchange high-performance liquid chromatographyinductively coupled plasma mass spectrometry. J Agric Food
Chem 53:60116018
9. Sloth JJ, Julshamn K (2008) Survey of total and inorganic arsenic
content in blue mussels (Mytilus edulis L.) from Norwegian fiords:
revelation of unusual high levels of inorganic arsenic. J Agric Food
Chem 56:12691273
10. ANZFA (Australia New Zealand Food Authority) (2001) The 19th
Australian Total Diet SurVey, Canberra, Australia. http://
www.foodstandards.gov.au/scienceandeducation/publications/
19thaustraliantotaldietsurveyapril2001. Accessed 6 Dec 2011

2834
11. WHO (World Health Organization) (2011) Evaluation of certain
contaminants in food: seventy-second meeting of the Joint FAO/
WHO Expert Committee on Food Additives. WHO Technical
Report Series 959
12. Commission EC (2006) Commission regulation (EC) No 1881/
2006 of 19 December 2006 setting maximum levels for certain
contaminants in foodstuffs. Off J Eur Commun L364:524
13. ANZFA (Australia New Zealand Food Authority) (2011) Standard
1.4.1contaminants and natural toxicants. Federal Register of
Legislative Instruments F2011C00542: Issue 124
14. Commission EC (2009) Commission directive (2009/141/EC) of
23 November 2009 amending Annex I to Directive 2002/32/EC of
the European Parliament and of the Council as regards maximum
levels for arsenic, theobromine, Datura sp., Ricinus communis L.,
Croton tiglium L. and Abrus precatorius L. Off J Eur Commun
L308:2023
15. Dufailly V, Nicolas M, Richoz-Payot J, Poitevin E (2011)
Validation of a method for arsenic speciation in food by ion
chromatography-inductively coupled plasma/mass spectrometry
after ultrasonic-assisted enzymatic extraction. J AOAC Int
94:947958
16. Foster S, Maher W, Krikowa F, Apte S (2007) A microwaveassisted sequential extraction of water and dilute acid soluble
arsenic species from marine plant and animal tissues. Talanta
71:537549
17. Styblo M, Hughes MF, Thomas DJ (1996) Liberation and analysis
of protein-bound arsenicals. J Chrom B 677:161166
18. Grotti M, Lagomarsino C, Goessler W, Francesconi KA (2010)
Arsenic speciation in marine organisms from Antarctic coastal
environments. Environ Chem 7:207214
19. Leufroy A, Noel L, Dufailly V, Beauchemin D, Guerin T (2011)
Determination of seven arsenic species in seafood by ion exchange
chromatography coupled to inductively coupled plasma-mass
spectrometry following microwave assisted extraction: method
validation and occurrence data. Talanta 83:770779
20. Sloth JJ (2005) Speciation analysis of arsenicDevelopment of
selective methodologies for assessment of seafood safety. PhD scientific thesis, University of Bergen, Norway: ISBN 82-7744-128-2
21. Sirot V, Guerin T, Volatier JL, Leblanc JC (2009) Dietary exposure
and biomarkers of arsenic in consumers of fish and shellfish from
France. Sci Total Environ 407:18751885
22. CEN/TC327 WG1 (2007) Animal feeding stuffs-contaminants,
minerals and trace elements, Minutes from Meeting CEN/TC 327
Animal feeding stuffs (CEN document no. CEN/TC327 N598)
23. ISO (1994) 57252: Accuracy (truness and precision) of measurement methods and results. Part 2: basic method for the determination of repeatability and reproducibility of a standard measurement
method
24. NMKL (Nordic Committee on Food Analysis) (2009) Validation
of chemical analytical methods. NMKL procedure 4:ed. 3
25. Commission EC (2002) Commission decision of 12 August 2002
implementing Council Directive 96/23/EC concerning the performance of analytical methods and the interpretation of results. Off J
Eur Commun L221:836

R.R. Rasmussen et al.

26. Larsen EH, Engman J, Sloth JJ, Hansen M, Jorhem L (2005)
Determination of inorganic arsenic in white fish using microwaveassisted alkaline alcoholic sample dissolution and HPLC-ICP-MS.
Anal Bioanal Chem 381:339346
27. Munoz O, Velez D, Montoro R (1999) Optimization of the solubilization, extraction and determination of inorganic arsenic [As
(III) + (As(V)] in seafood products by acid digestion, solvent
extraction and hydride generation atomic absorption spectrometry.
Analyst 124:601607
28. Ptursdttir H, Gunnlaugsdttir H, Jrundsdttir H, Raab A,
Krupp EM, Feldmann J (2012) Determination of inorganic arsenic
in seafood: emphasizing the need for certified reference materials.
Pure Appl Chem 84:191202
29. Gamble BM, Gallagher PA, Shoemaker JA, Wei X, Schwegel
CA, Creed JT (2002) An investigation of the chemical stability of arsenosugars in simulated gastric juice and acidic environments using IC-ICP-MS and IC-ESI-MS/MS. Analyst
127:781785
30. Niegel C, Matysik FM (2010) Analytical methods for the determination of arsenosugarsa review of recent trends and developments. Anal Chim Acta 657:8399
31. Soeroes C, Goessler W, Francesconi KA, Schmeisser E, Raml R,
Kienzl N, Kahn M, Fodor P, Kuehnelt D (2005) Thio arsenosugars
in freshwater mussels from the Danube in Hungary. J Environ
Monit 7:688692
32. Yalcin S, Le XC (2001) Speciation of arsenic using solid phase
extraction cartridges. J Environ Monit 3:8185
33. Yu C, Caia Q, Guoa Z, Yanga Z, Khoob SB (2003) Inductively
coupled plasma mass spectrometry study of the retention behavior
of arsenic species on various solid phase extraction cartridges and
its application in arsenic speciation. Spectrochim Acta B 58:1335
1349
34. CEN (European Committee for Standardization) (2011)
Provisional European Norm: prEN16206Animal feeding
stuffsdetermination of arsenic by hydride generation atomic
absorption spectrometry (HGAAS) after microwave pressure
digestion (digestion with 65% nitric acid and 30% hydrogen
peroxide). prEN 16206
35. Schmeisser E, Goessler W, Kienzl N, Francesconi KA (2004)
Volatile analytes formed from arsenosugars: determination by
HPLC-HG-ICPMS and implications for arsenic speciation analyses. Anal Chem 76:418423
36. Horwitz W (1982) Evaluation of analytical methods used for
regulation. J Assoc Off Anal Chem 65:525530
37. Francesconi KA, Tanggaard R, McKenzie CJ, Goessler W (2002)
Arsenic metabolites in human urine after ingestion of an arsenosugar. Clin Chem 48:92101
38. Sloth JJ, Larsen EH, Julshamn K (2004) Selective arsenic speciation analysis of human urine reference materials using gradient
elution ion-exchange HPLC-ICP-MS. J Anal At Spectrom 19:973
978
39. Sloth JJ, Julshamn K, Lundebye AK (2005) Total arsenic and
inorganic arsenic content in Norwegian fish feed products.
Aquacult Nutr 11:6166