wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 5

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

Geochemical changes in individual sediment grains duringsequential arsenic extractions

Elisabeth Eiche a,*, Utz Kramar a, Michael Berg b, Zsolt Berner a, Stefan Norra a,Thomas Neumann a

a Institute of Mineralogy & Geochemistry, Karlsruhe Institute of Technology, 76131 Karlsruhe, GermanybEawag, Swiss Federal Institute of Aquatic Science and Technology, 8600 Dubendorf, Switzerland

a r t i c l e i n f o

Article history:

Received 15 January 2010

Received in revised form

4 May 2010

Accepted 1 June 2010

Available online 9 June 2010

Keywords:

Groundwater arsenic contamination

Mobility

Sediment leaching

Re-precipitation

Micro-synchrotron X-ray

fluorescence analysis (mS-XRF)

Vietnam

* Corresponding author. Tel.: þ49 721 608 33E-mail address: elisabeth.eiche@kit.edu (

0043-1354/$ e see front matter ª 2010 Elsevdoi:10.1016/j.watres.2010.06.002

a b s t r a c t

High concentrations of As in groundwater frequently occur throughout the world. The

dissolved concentration, however, is not necessarily determined by the amount of As in

the ambient sediment but rather by the partitioning of As between different minerals and

the type of fixation. Sequential extractions are commonly applied to determine associa-

tions and binding forms of As in sediments. Due to the operational nature of the extracted

fractions, however, the results do not provide insight into how and where precisely As is

bound within mineral grains and no information about elemental associations or involved

mineral phases can be gained. Furthermore, little is known about possible geochemical

alterations that actually occur within a single grain during sequential extraction. There-

fore, micro-synchrotron X-ray fluorescence analysis was applied to study the micro-scale

distribution of As and other elements in single sediment grains.

Arsenic was found to be mainly enriched in Fe oxy-hydroxide coatings along with other

heavy metals resulting in high correlations. Phosphate leached 34e66% of As from the

studied grains. The release of As in this leaching step was accompanied by the disap-

pearance of correlations between As and Fe as well as by a higher Fe/As ratio compared to

untreated samples. During the Fe-leaching step the coatings were largely dissolved leading

to much lower concentrations of As and Fe. The correlation between As and Fe was

preserved only in association with K, indicating the presence of both elements in silicate

structures.

Several distinctive features were observed such as the release of Fe, Mn and Cr during

phosphate leaching as well as the lowering of mean K concentrations due to the Fe-

leaching which indicates that not only target mineral phases were dissolved in these

extraction steps. The importance of re-precipitation processes during sequential extraction

was indicated by a consistently observed increase of the Fe/As ratio from the untreated to

the Fe-leached samples.

ª 2010 Elsevier Ltd. All rights reserved.

27; fax: þ49 721 608 4170.E. Eiche).ier Ltd. All rights reserved.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 55546

1. Introduction As sequentially from soils and sediments (Keon et al., 2001;

High arsenic (As) concentrations in many groundwater

systems around the world have received increasing attention

in the past decade (e.g. Smedley and Kinniburgh, 2002; Berg

et al., 2007; Winkel et al., 2008). Much is known about the

fixation and release of As by now, even though there is still

some dispute about the biogeochemical nature and sequence

of the mechanisms involved (e.g. Das et al., 1996; BGS and

DPHE, 2001; Dowling et al., 2002; Stuben et al., 2003; Swartz

et al., 2004; Polizzotto et al., 2006; Stute et al., 2007; Winkel

et al., 2008; Buschmann and Berg, 2009). What seems to be

clear is that released As results mainly from within the

sediment itself. Known As containing constituents are sili-

cates (e.g. biotite (26.5e49.6 mg/kg, Seddique et al., 2008),

chlorite, (5e50 mg/kg, Pal et al., 2002; Sengupta et al., 2004),

amphibole), Fe/Mn oxy-hydroxides (e.g. goethite, hematite

(As� 160 mg As/kg, Bauer and Onishi, 1969), magnetite

(8e40 mg/kg, Sengupta et al., 2004)) and carbonates (e.g.

siderite (5e20 mg/kg, Bhattacharya et al., 2003)). Fe oxy-

hydroxides which precipitate during diagenetic processes

frequently occur as coatings on host grains (mainly on quartz

and feldspars). As a consequence the average As concentra-

tions of the grains can get enhanced to 30 mg As/kg or more

(Reynolds et al., 1999; Pal et al., 2002; Horneman et al., 2004;

Sengupta et al., 2004). Arsenic is either retained on the

surfaces of these minerals, e.g. through outer- and inner-

sphere complexes, or incorporated into the mineral structure

itself (Filgueiras et al., 2002; Smedley and Kinniburgh, 2002).

There is a need to know more about where and in which

associations As is bound to or incorporated in mineral grains.

In order to be able to predict the mobility and transport of As

with some level of confidence, however, it is important to

evaluate its partitioning and speciation apart from the quan-

tification of bulk concentrations (Gruebel et al., 1988; Keon

et al., 2001; van Herreweghe et al., 2003). A small change in

thepartitioning ofAs betweensediment andwater can already

give rise toa significant increase in theAsconcentrationsof the

groundwater (Radu et al., 2005; Meharg et al., 2006).

There are several instrumental possibilities to directly

determine associations and binding forms of trace elements

in sediments such as synchrotron-based X-ray fluorescence

(mS-XRF), X-ray absorption near edge structure (XANES),

extended X-ray absorption fine structure (EXAFS) or particle-

induced X-ray emission (PIXE) (Bacon and Davidson, 2008).

These methods are still not widely available which is why

sequential extraction schemes, though time consuming but

with low instrumental requirements, are frequently used

(Keon et al., 2001; Filgueiras et al., 2002; Wenzel et al., 2001;

McArthur et al., 2004; Berg et al., 2008). Nowadays sequential

extractions are commonly accepted and adapted to the cor-

responding needs but they are often criticized for their well

known weaknesses such as the lack of selectivity of extrac-

tants, possible re-adsorption and re-precipitation processes

during extraction or the lack of quality control (Tipping et al.,

1985; Kheboian and Bauer, 1987; Keon et al., 2001; Gleyzes

et al., 2002; van Herreweghe et al., 2003; Wright et al., 2003;

Hudson-Edwards et al., 2004; Bacon and Davidson, 2008).

Several extraction procedures have been developed to leach

Wenzel et al., 2001; van Herreweghe et al., 2003) and have

been reported in numerous papers. Due to the operational

nature of the fractions, however, the results of all these

methods do not provide insight into howandwhere exactly As

is bound within mineral grains and no information about

elemental associations can be gained. Furthermore, limited

information can be obtained about which minerals are actu-

ally involved in the release or re-adsorption of As depending

on the extractant used. To be able to compare results pre-

sented in different publications, there is need for better

understanding, from a geochemical and mineralogical point

of view, which alterations occur within the sediment (Bacon

and Davidson, 2008). Furthermore, a detailed knowledge of

the processes involved in each extraction step would enhance

the transferability of the results to As mobilization in nature.

Several studies have already tested the feasibility of

sequential extractions, have tried to characterize mineralog-

ical changes during these extractions, and/or have investi-

gated the selectivity of extractant solutions (Gruebel et al.,

1988; Whalley and Grant, 1994; Dodd et al., 2000; La Force

and Fendorf, 2000; Wright et al., 2003; Paul et al., 2009). To

our knowledge, however, only a few papers have focused on

extraction schemes that are specialized on As. Furthermore,

the impact of extractants on the geochemistry andmineralogy

of single grains has not been clarified so far. Synchrotron-

based XRF analysis seems to be an ideal tool to map trace

elemental distributions and, therefore, to investigate

geochemical changes on a micrometer scale within single

grains. Applying this analytical method, the focus of the

present study lies (1) on the assessment of the variability and

distribution of As species within single sediment grains, (2)

the determination of As-bearing phases, and, (3) on the

changes occurring with regard to points (1) and (2) during

different sequential extraction steps.

2. Methods

2.1. Sequential extraction

Sequential extraction experiments were carried out with

sediment samples, which were in contact with high and low

As concentrations in the respective groundwater bodies to

investigatewhether As is bound to similar or differentmineral

phases within these sediments. Therefore, sediment cores

were drilled at two sites in Van Phuc, Red River Delta, Viet-

nam, which differ considerably in the dissolved As concen-

tration in the respective groundwater (Berg et al., 2008; Eiche

et al., 2008). One site shows dissolved As concentrations of

less than 10 mg/L (site L (low)), whereas up to 600 mg/L occur at

the other site (site H (high)). The samples were taken from

sandy delta sediments (depth: w20e50 m), which differed in

their colour (site L: orange, site H: grey) indicating differences

with regard to the redox-conditions (formore details see Eiche

et al., 2008).

The sediments used for sequential extractions and subse-

quent mS-XRFmeasurements were flushed with nitrogen after

sampling and frozen until preparation in order to avoid

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 5 5547

oxidation and alteration of the sediment during transport and

storage. For the synchrotron-based mS-XRF analysis one

sediment sample with relatively high bulk As concentration

from each site was chosen. Sample 1 (S1) taken at site L had

20 mg As/kg and sample 2 (S2) originating from site H, 35 mg

As/kg. During preparation, the samples were split into three

parts of w0.5 g each. Two aliquots were transferred into

centrifuge tubes in order to leach the sediments with the

respective extractants (Table 1) of the extraction scheme of

Keon et al. (2001), (slightly modified); (step 1: P-leaching (LP);

step 2: dissolution of Fe oxy-hydroxides (LFe)). The third

aliquot was left untreated. After each step the samples were

centrifuged at 4500 rpm for 20 min and the solution was dec-

anted. Then, sedimentmaterial waswashedwithMili-Qwater

and dried at room temperature.

The untreated sediment as well as the leachedmaterial (LP

and LFe) was embedded in Epotek resin, and then sliced and

polished to prepare thin sections of 60e100 mm thickness.

Subsequently, the thin sections were attached to a piece of

Mylar film as carrier material.

2.2. Micro-synchrotron-XRF (mS-XRF)

The mS-XRF measurements were carried out at the FLUO-

beamline of the Ǻngstromsource (ANKA) of the For-

schungszentrum Karlsruhe (FZK), Germany. The emitted

primary beam was monochromatized for the analysis by

a double multilayer monochromator (W-BC4, 3 nm period). In

order to eliminate possible interferences of As with lead (Pb)

the beam energy was adjusted to 12.5 keV, which is just below

the adsorption edge of Pb. The beamwas focused to a spot size

of 2� 8 mm by means of refractive lenses. It penetrated the

sample at an incident angle of 45� and, therefore, the char-

acteristic X-rays emitted by the elements had a nearly cylin-

drical formwith a diameter of the spot size. Consequently, the

signal corresponding to ameasuring point always consisted of

an average value of the characteristic X-rays emitted from the

penetrated area. The signal was collected by a Vortex Silicon

Multicathode Detector with a resolution of 142 eV at 5.9 keV.

The measuring time in each spot was 10e20 s.

Table 1 e Adapted sequential extraction scheme used forthe sediment leaching. Thin sections were prepared afterleaching with the respective solutions.

Step Target phase Extractant Ref.

Step 1

LP

Strongly adsorbed 0.5 M NaH2PO4 Keon et al.

(2001)

Step 2

LFe

Coprecipitated with

acid volatile sulphides,

carbonates, Mn-oxides,

very amorphous Fe

oxy-hydroxides

1 M HCl Keon et al.

(2001)

Coprecipitated with

amorphous Fe

oxy-hydroxides

0.2 M NH4þ-

oxalate/oxalic acid

Keon et al.

(2001)

Coprecipitated with

crystalline Fe

oxy-hydroxides

0.5 M Na-citrateþ1 M

NaHCO3; 0.5 g

Na2S2O4xH2O

van

Herreweghe

et al. (2003)

The absorption of primary X-rays by the sample and of the

emitted characteristic X-rays (fluorescence) depends on the

thickness, density and major element composition of the

sample. Certified reference samples (STHS6-G; Jochum et al.,

2000) were used for calibration and quantification. The

measurements were not carried out under vacuum; therefore,

the Si content could not be quantified. However, its relative

abundance within a grain is correct and can be used for

statistical analysis.

The mineral grains were chosen for analysis according to

their appearance under the optical microscope. In order to do

justice to the sediment sample heterogeneity, grains with

obvious differences in appearance were selected. They can be

grouped as follows: (1) coated mineral grains and (2) mafic

minerals, including black, reddish and greenish minerals.

Either a transect through a selected grain was analysed or

a complete elemental mapping was carried out.

2.3. Statistical analysis

To investigate preferential associations of As and to identify

the changes occurring in elemental associations during

sequential extractions within sediment grains Pearson’s

correlation coefficients for all elements were calculated. Due

to the fact that some of the analysed minerals and coatings

were thinner (nm to several mm) than the mS beam (2� 8 mm),

each measuring point consisted of a mixed signal. It was,

therefore, difficult to identify minerals potentially involved

into the As fixation based on elemental associations alone.

However, multivariate statistical analysis (factor analysis)

made it possible to delineate these mixed signals. In order to

apply this multivariate approach the data set was checked for

normal distribution and cleaned of outliers.

3. Results

The results of mS-XRF measurements of transects through

grains and complete mappings of whole grains of the

untreated as well as the leached samples are listed in Table 2.

3.1. Untreated grains

The elemental mapping of the untreated material resulted in

solid-bound As concentrations in the samples from <1

(detection limit) to >500 mg/kg. In both samples high As

concentrations were present in coated minerals (S1:

<100 mg/kg, S2: <206 mg/kg) and mafic minerals (S1: <58 mg/

kg, S2: <514 mg/kg). These values indicate strong punctual

enrichments of As compared to the bulk concentrations of

20 mg/kg (S1) and 35 mg/kg (S2), respectively, quantified

previously by XRF analysis (Eiche et al., 2008). The mean As

concentrations (Table 2), however, were slightly lower

compared to the bulk As concentrations with 12 (S1) and 23

(S2) mg/kg, respectively. Mean Fe concentrations were 12 (S1)

and 7.6 (S2) g/kg (Table 2), respectively, with strong enrich-

ments in coatings (S1: <64 g/kg, S2: <56 g/kg) as well as in

mafic minerals (S1: <54 g/kg; S2: <52 g/kg). Elements like Mn,

Ni, Cu, Zn were also enriched in the coatings along with As

and Fe which is expressed in relatively good correlations

Table 2 eMean elemental concentrations measured withmS-XRF analysis and calculated mean Fe/As ratios ofseveral grains of sample 1 (S1) and 2 (S2). LP: leachedwithphosphate solution; LFe: after dissolution of Fe oxy-hydroxide and carbonate phases.

Element S1 S2

untreated LP LFe untreated LP LFe

As mg/kg 12 8 2.9 23 7.9 1.3

Fe g/kg 12 10 1.5 7.6 5 0.9

Fe/As 1020 1180 400 320 530 720

K g/kg 5.1 4.7 1.5 8.5 8 2.8

Ti g/kg 1.1 1.0 1.0 1.6 1.2 0.4

Cr mg/kg 44 28 7.3 33 11 7.7

Mn mg/kg 584 373 8.4 54 28 9.2

Ni mg/kg 16 8 8.6 8.3 5.2 2.9

Cu mg/kg 6.2 6 5.1 8.9 6.0 4.2

Zn mg/kg 19 12 14 27 24 15

No. of

grains

11 12 17 16 11 13

Fig. 1 e Microscopic image and distribution of Si (not quantified

a transect (white line: transect of focus). In (a) and (b) coated grai

a mafic mineral (S1-m) is illustrated.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 55548

between these elements in coated minerals (r> 0.7). Sample 1

had a higher median Fe/As ratio (1020) compared to S2 (320)

which is due to the higher median As and lower median Fe

concentrations in S2 (Table 2). In general, mafic minerals

have much higher median Fe/As ratios compared to coatings,

which leads to the assumption that the precipitation of Fe

oxy-hydroxide coatings leads to a relative enrichment of As

compared to Fe.

The results of the transect measurement along the

example shown in Fig. 1a (S1) revealed a high correlation

between Fe and As (r¼ 0.99, n¼ 41). Two factors could be

extracted which explain 91% of the variance. The first factor

(F1) comprised Ca, Cr, Cu, Fe, Ni, As,Mn, Zn. Apart fromCa and

Cr, which are lithophile elements but low in concentration

(Ca: median 58 mg/kg, Cr: 6.5 mg/kg), all other elements of F1

could be classified as siderophile or chalcophile. In the second

factor (F2) mainly lithophile elements such as K, V and Ga,

and, Mn had high loadings (Table 3). The mineral phase rep-

resented by F1 was mainly present as coating as indicated by

the factor scores (data not shown).

In the grain displayed in Fig. 1b (S2), As and Fe were also

highly correlated (r¼ 0.96, n¼ 51). However, the factor anal-

ysis based on this data set led to different elemental

), Fe and As concentrations in analysed grains along

ns are displayed (S1, S2). In (c) the elemental distribution of

Table 3 e Factor loadings resulting from transectmeasurements of two coated (S1, S2) and onemafic grain(S1-m) of the untreated samples. Elements with lowcommunalities (<0.7) were excluded from the analysis(l.c.[ low communality).

Untreated samples

S1 S2 S1-m

F1 F2 F1 F2 F1 F2

Ca 0.97 0.10 0.19 0.91 l.c. l.c.

Cr 0.96 0.00 0.37 0.85 0.22 0.88

Fe 0.93 0.30 0.67 0.73 0.74 0.59

Ni 0.93 0.29 l.c. l.c. 0.68 0.53

As 0.90 0.39 0.85 0.44 0.71 0.59

Cu 0.86 0.18 0.84 0.41 l.c. l.c.

Mn 0.70 0.70 0.41 0.87 0.98 0.00

Zn 0.69 0.53 0.92 0.22 0.11 0.91

Ga 0.25 0.94 0.85 0.35 0.12 0.86

V 0.21 0.93 l.c. l.c. l.c. l.c.

K 0.08 0.98 l.c. l.c. 0.93 0.00

Si �0.85 �0.37 l.c. l.c. l.c. l.c.

Ti l.c. l.c. 0.48 0.75 l.c. l.c.

Explained variance [%] 58 33 45 44 43 41

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 5 5549

associations compared to the previous example. In F1 chal-

cophile elements like Fe, As, Cu, Zn and Ga were grouped

together whereas in F2 Ca, Cr, Fe, Mn and Ti had high loadings

indicating a lithophile origin (Table 3). Both factors explained

89% of the variance. The factor scores indicated that the

mineral phase represented by F1 was mainly present in the

coating area (data not shown).

The results of the factor analysis of mafic minerals in

general, mostly led to numerous factors and low communali-

ties for several elements and were, therefore, difficult to

interpret. In the mafic mineral S1-m (Fig. 1c) Fe and As were

also highly correlated (r¼ 0.87, n¼ 51). For the factor analysis,

Fig. 2 e Microscopic image and distribution of Si (not quantified)

leaching with 0.5 M Na-phosphate solution.

however, several elements (Si, V, Ca, Ti, Cu) had to be excluded

due to their low communality. Two factors could be extracted

which explained 84% of the total variance. Both factors had

relatively high loadings for As (Table 3). In F1, K grouped

together with Fe, As, Mn and Ni. The second factor (F2), which

was mainly dominant in the core of the grain according to the

factor scores, was stamped by high loadings for Zn, Ga, Cr.

Except for Cr, these elements have a chalcophile character.

3.2. Grains leached with phosphate (LP)

After leaching with the 0.5 MNa-phosphate solution,mean As

concentrations were lower (S1: 8.0 mg/kg; S2: 7.9 mg/kg)

compared to the untreated sediment material (Table 2),

especially when taking only the coated grains into consider-

ation (S1: 5.5 mg/kg; S2: 4.0 mg/kg). A considerable enrich-

ment of As was still detectable in some coated mineral grains

(Figs. 2 and 3a), indicating that a substantial amount of As was

not released by P-leaching. Mean Fe concentrations were also

slightly lower (S1: 10 mg/kg; S2: 5 mg/kg) in comparison to the

untreated sediment grains in both samples (Table 2).

However, the median Fe/As ratios were higher (S1: 1180; S2:

530, Table 2) which points towards preferential release of As

compared to Fe in this extraction step. Potassium concentra-

tions remained more or less unchanged (S1: 4.7 g/kg; S2: 8 g/

kg, Table 2). Elemental concentrations of Cr, Mn and Ni were

also decreased after P-leaching (Table 2).

In the coated grain from S1 LP-a (Fig. 2), which was entirely

mapped for elemental concentrations as well as in the transect

S1 LP-b (Fig. 3a), As was highly enriched mainly in the coating

(S1 LP-a: <540 mg/kg; S1 LP-b: <58mg/kg). Arsenic showed

a high correlation with Fe in each of the samples (rS1 LP-a¼ 0.79,

n¼ 2808; rS1 LP-b¼ 0.99, n¼ 51). Two factors, which explained 84

and 88% of the variance, were extracted by factor analysis in S1

LP-a and S1 LP-b, respectively. Comparable to the examples of

untreated grains chalcophile and siderophile elements were

, Fe and As concentrations in a coated grain of S1 LP-a after

Fig. 3 e Microscopic image and distribution of Si (not quantified), Fe and As concentrations of two coated grains ((a) S1 LP-b,

(b) S2 LP) along a transect (white line: transect of focus) which were analysed after leaching with 0.5 M Na-phosphate

solution.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 55550

grouped in F1 (Fe, As, Zn, Ni, Mn, Ti, Cu; see Table 4), whichwas

dominant in the coating area. The lithophile elements repre-

sented by F2 (K, V, Si, Ga)were dominating in themineral phase

of the grain core.

In S2 LP (Fig. 3b) As concentrations were mainly below

detection limit (<1 mg/kg) with only slightly higher

Table 4 e Factor loadings resulting from a completemapping of a coated grain (S1 LP-a) and transectmeasurements of two coated grains (S1 LP-b, S2 LP) afterP-leaching. Elements with low communalities (<0.7)were excluded from the analysis (l.c.[ lowcommunality).

Phosphate leached samples

S1 LP-a S1 LP-b S2 LP

F1 F2 F1 F2 F1 F2

Zn 0.95 0.01 0.96 0.04 0.85 0.47

Mn 0.94 0.11 0.99 0.04 0.87 0.43

Fe 0.94 �0.20 0.99 0.03 0.83 0.51

As 0.88 0.11 0.99 0.01 �0.03 0.85

Ni 0.88 0.23 0.97 0.05 0.95 0.13

Cu 0.82 0.31 0.96 0.02 0.66 0.64

Ti 0.71 0.53 0.84 0.19 0.26 0.81

V 0.27 0.85 0.44 0.82 0.85 0.51

K 0.09 0.95 �0.33 0.91 0.20 0.93

Si 0.09 0.92 �0.07 0.93 �0.59 0.02

Ga �0.04 0.89 0.23 0.84 0.96 0.09

Cr l.c. l.c. 0.83 0.01 0.75 0.49

Ca l.c. l.c. l.c. l.c. 0.41 0.82

Explained

variance [%]

50 34 62 26 49 33

concentrations in the coating (<1.3 mg/kg). In contrast to the

previous grain samples, As and Fe did not show a correlation

(r¼ 0.38, n¼ 41). The two factors extracted from the recorded

mS-XRF data set explained 82% of the total variance. Both

siderophile as well as chalcophile elements were grouped in

F1 (Ga, Ni, Mn, Zn, Cr, V, Fe). Arsenic was included in F2

together with the lithophile elements K, Ti and Ca (Table 4).

The factor scores indicated that the mineral phase which was

represented by F2 was mainly present in the lower part of the

coating (data not shown).

3.3. Grains after dissolution of Fe oxy-hydroxides andcarbonates (LFe)

After dissolution of Fe oxy-hydroxides and carbonates, median

As concentrations were low in general (S1: 2.9 mg/kg, S2:

1.3 mg/kg, Table 2) as well as in the coatings (<9 mg/kg, Fig. 4b),

with occasional readings below detection limit (<1 mg/kg,

Fig. 4a). Very high As concentrations were still present inmafic

minerals with up to 140 mg/kg (Fig. 4c). Median Fe concentra-

tions were considerably reduced in both samples (S1: 1.5 g/kg;

S2: 0.9 g/kg, Table 2) with high median concentrations occur-

ring only in mafic minerals (<80 g/kg). However, in some

coatings a certain Fe enrichment (S1: <13 g/kg; S2: <36 g/kg)

was still detectable. Higher concentrations of As or Fe in the

coatings were mainly associated with elevated concentrations

of K which was expressed by good correlations among these

elements (r> 0.7). Median elemental concentrations of Mn and

Cr were also clearly lowered after the dissolution of Fe oxy-

hydroxides and carbonates whereas a decrease in Ni and Zn

concentration was only detectable in S2 (Table 2). As expected,

no considerable correlations between As and Fe remained in

coated grains after the LFe extraction step. The median Fe/As

Fig. 4 e Microscopic image and distribution of Si (not quantified), Fe and As concentrations along a transect (white line:

transect of focus) of two coated grains ((a) S2 LFe, (b) S1 LFe) and a mafic grain (S1 LFe-m) after the dissolution of Fe oxy-

hydroxides and carbonates. In (a) relatively high As concentrations remain, whereas in (b) the concentrations are mainly

below detection limit of 1 mg/kg.

Table 5 e Factor loadings resulting from transectmeasurements of two coated grains (S1 LFe, S2 LFe) afterdissolution of Fe oxy-hydroxides and carbonates.Elements with low communalities (<0.7) were excludedfrom the analysis (l.c.[ low communality).

After dissolution of Fe oxy-hydroxides and carbonates

S1 LFe S2 LFe S1 LFe-m

F1 F2 F1 F1 F2 F3

Ca 0.98 0.17 0.86 l.c. l.c. l.c.

Mn 0.95 0.28 0.85 0.36 �0.29 0.73

Ti 0.95 0.29 0.87 0.17 0.26 0.63

Ni 0.86 0.50 l.c. �0.01 0.96 0.11

V 0.68 0.56 l.c. 0.41 0.56 0.31

Fe 0.55 0.78 0.96 0.36 0.36 0.80

Zn 0.42 0.83 0.89 0.44 0.12 0.64

Ga 0.30 0.91 0.91 0.93 0.02 0.22

As 0.27 0.86 l.c. �0.09 0.94 0.11

K 0.13 0.95 0.95 0.88 0.10 0.34

Cr l.c. l.c. 0.85 0.88 �0.06 0.15

Si l.c. l.c. l.c. �0.49 0.41 0.68

Explained variance [%] 46 45 80 30 24 25

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 5 5551

ratio in S1 (400) was much lower compared to untreated sedi-

ment in all samples whereas in S2 (720), the ratios were higher

compared to the previous two leaching steps (Table 2).

Due to the fact that As concentrations were very low in

most of the Fe-leached coatings as in S2 LFe (<1.9 mg/kg;

Fig. 4a), As was excluded from most of the factor analysis due

to low communality. Only one factor was extracted from the

data set of S2 LFe which explained 80% of the variance (Table

5). In this factor, which was dominant in the coating area of

the analysed grain, mainly lithophile elements were grouped

(K, Ca, Ga, Fe, Mn, Ti, Zn, Cr).

In one grain of S1 LFe (Fig. 4b) As concentrations were high

enough (<8.4 mg/kg) to be considered in the statistical anal-

ysis. In this grain, As and Fe showed good correlations

(r¼ 0.74, n¼ 30). The two extracted factors explained 91% of

the variance. In both F1 (Ca, Mn, Ti, Ni, V) and F2 (K, Fe, As, Zn,

Ga), which were dominant at the rim of the measured grain,

lithophile elements had high loadings (Table 5).

In mafic minerals like S1 LFe-m (Fig. 4c), high concentra-

tions of As (<146 mg/kg) were still present after dissolution of

Fe oxy-hydroxides and carbonates, but mainly in the core of

the grain and not in the coating. However, as expected, there

was no considerable correlation between As and Fe (r¼ 0.37,

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 55552

n¼ 26). The three factors derived from factor analysis

explained 79% of the variance. Factor 1 (K, Cr, Ga) and F3 (Mn,

Fe, Si, �Ti, �Zn), which dominated at the rim of the grain,

consistedmainly of lithophile elements (Table 5). Themineral

phase represented by F2 (As, Ni) consisted of chalcophile

elements and dominated in the core of the grain.

4. Discussion

4.1. Arsenic bearing phases

In order to be able to know which mineral phases might be

affected during sequential extraction steps it is important to

understand themineralogy of the sediment (Dodd et al., 2000).

Since we did not have access to mS-XRD analysis during this

study the mineralogical nature of As-bearing phases was

assessed for the analysed grains through statistical relation-

ships. For interpretation of the leaching results it is also

important to localize the mineral phases in order to be able to

evaluate if they were altered by the extraction solutions.

Typical primary As-bearing minerals are probably mafic

silicates or phyllosilicates as indicated by the grouping of As

with K and other lithophile elements in one factor (S1-m,

Table 3; S2 LP, Table 4; S1 LFe, Table 5). Mainly, Fe is included

in the factor suggesting the presence of Fe bearing mafic

phyllosilicates like biotite or chlorite which are known As-

bearing phases (Anawar et al., 2004; Sengupta et al., 2004; Nath

et al., 2005; Seddique et al., 2008). In the case of S1 LFe (Fig. 4a)

the phyllosilicate is located at the rim of a core grain.

However, it is not a coating in the actual sense, but rather

attached to the surface of a host mineral possibly resulting

from weathering of the central silicate core grain.

Further As-bearing minerals of importance seem to be

sulphides, which mainly occur as distinct authigenic mineral

grains. For example in a mafic grain (S1 LFe-m) a NiAs

sulphide seems to be present as a small inclusion with As

concentrations of up to 140 mg/kg (Fig. 4c). Due to the fact that

this phase is enclosed by a silicate phase it cannot get into

contact with extracting solutions. This, however, will not

distort the transferability of the interpretation because it is

also not likely that this mineral phase will be dissolved easily

in nature. There are also indications that sulphides can occur

as surface precipitates (S1, S2, Table 3; S1 LP-a/b, Table 4).

Biogenic pyrite formation is considered to be very effective in

removing As together with other elements like Ni or Cu from

groundwater (Saunders et al., 1997; Anawar et al., 2002;

Smedley and Kinniburgh, 2002; Lowers et al., 2007). These

surface precipitates are likely to get dissolved in oxic envi-

ronments and, therefore, can release As into the groundwater.

Iron oxy-hydroxides, frequently considered as one of the

most important groups of minerals in controlling the As

release (e.g. Nickson et al., 2000; Swartz et al., 2004), are also

important As hosts in the analysed samples (<140 mg/kg; S1,

S1 LP-a). However, these oxy-hydroxide phases could only be

detected as coatings and not as discrete minerals. Within the

coatings As is very heterogeneously distributed. In addition to

As, Fe oxy-hydroxides do incorporate or adsorb other heavy

metals like Zn, Cu, Mn and Ni.

High As concentrations can occur in mafic and coated

minerals, but the highest measured As concentrations were

generally present in mafic minerals. It is assumed that As and

Fe are closely associated in solids (Harvey et al., 2002). In this

study, clear correlations were found between As and Fe

mainly if As was enriched in the coatings but only rarely if

present in mafic minerals. Based on this fact, it can be

concluded that significant correlations between As and Fe are

mainly observable after dissolution or weathering of discrete

primary minerals followed by the precipitation of Fe oxy-

hydroxide coatings. The precipitation of Fe oxy-hydroxides

further seems to lead to the enrichment of As relative to Fe as

indicated by the much lower Fe/As ratio in coatings in

comparison to mafic mineral phases.

4.2. Geochemical alterations caused by sequentialextraction

The P-leaching step aims at removing adsorbed As from

sediment particles (Keon et al., 2001; Wenzel et al., 2001). The

much higher decrease of As concentrations in S2 (66%)

compared to S1 (34%) indicates that a higher proportion of As

was adsorbed in S2. This finding is consistent with the high

groundwater As concentrations and the reducing conditions

prevailing at the study site where S2 was sampled (Berg et al.,

2008; Eiche et al., 2008). These conditions are typically related

to a high proportion of P-leachable As (Swartz et al., 2004;

Zheng et al., 2005; van Geen et al., 2006; Berg et al., 2008; van

Geen et al., 2008). Sample 1, on the other hand, originates

from an only moderately reducing study site with low

groundwater As concentrations (Eiche et al., 2008). At both

sites, high As concentrations were occasionally detected in

coatings (S1 LP-a/b) after the P-leaching stepwhich, on the one

hand, confirms that a certain amount of As is not only present

in an adsorptive way. On the other hand, remaining high

concentrations in the coatings could also indicate that

adsorbed As was not fully released during the P-leaching step,

an assumption that could be tested by X-ray absorption

analysis in further studies.

The lowered mean Fe concentrations after P-leaching

would imply that some Fe bearing minerals were already

dissolved in this extraction step which, in fact, is not intended

(Keon et al., 2001). The higher Fe/As ratios compared to the

untreated samples, however, indicate that more As relative to

Fe was released. The fact that As and Fe persisted in being

highly correlated in P-leached coatings as in the untreated

samples indicates that no major alterations in the mineral

structure took place. Apart from As and Fe, further elements

seem to be affected by the P-leaching such as Mn, Cr or Ni,

whose mean concentrations were reduced by approximately

50% (Table 2). All these elements showed high correlations

with As and Fe in the untreated samples and are probably

associated with As in the Fe oxy-hydroxide coatings. There is

a need for further investigation as to whether and by what

means Fe and other elements are released during P-leaching

or whether the decrease in concentration is due to sample

heterogeneity. The possibility that sample heterogeneity

could explain the sometimes highly variable results in

sequential extractionwas also argued by Keon et al. (2001) and

van Herreweghe et al. (2003). As expected, silicates were not

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 5 5553

affected by P-leaching, which is corroborated by comparable

concentration levels of the lithophile elements K and Ti in the

untreated samples.

After dissolution of Fe oxy-hydroxides and carbonates high

As and Fe concentrations were only present in mafic minerals

(S1-m) or in sulphides (S1 LFe-m) which were obviously not

affected by these extraction steps. Mean As and Fe concentra-

tions, however, were considerably lower in coated minerals

compared to theuntreated and theP-leached samples (Table 2).

This indicates, firstly, that this leaching step strongly affects Fe

phases as intended, and secondly, that As is predominantly

associated with these phases. As a consequence, no consider-

able correlations remain between Fe and As in formerly coated

minerals. The correlation between As and Fe is preserved only

in association with K. Therefore, it can be assumed that both

elements are mainly present in silicate structures after the

dissolution of Fe oxy-hydroxides which implies that the

extractant, as intended, effectively removesFeoxy-hydroxides.

Arsenicwill onlybe released fromtheremainingsilicatesdue to

strong long-term weathering (Filgueiras et al., 2002). Further

elements like Mn or Cr which were associated with Fe oxy-

hydroxide coatings in the untreated samples were also

considerably released during the LFe extraction steps.

More released Fe compared to As during the LFe extraction

step should lead to low Fe/As ratios, at least lower compared to

the P-leaching step as in S1 (Table 2). In S2, however, the ratio

wasconstantly increasing fromtheuntreatedsample to theFe-

leached sample. This increase in the Fe/As ratio implies that

more As than Fewas released, which is not very likely. Instead

it seems more feasible that a noticeable re-precipitation of Fe

mineralphases tookplace, apossibilitywhichhasalreadybeen

argued by several authors (Tipping et al., 1985; Kheboian and

Bauer, 1987; Gruebel et al., 1988; Filgueiras et al., 2002).

Finally, the much lower median concentrations of K in both

samples and of Ti in S2 compared to the untreated and P-

leached samples indicate that some silicatesmight actually be

affected by the Fe-leaching step. Further investigation is

needed to determine whether this is really happening or

whether it is simply related to sample heterogeneity.

5. Conclusions

This study demonstrated that mS-XRF analysis is a powerful

tool to obtain a better picture of the distribution of As within

sediment grains and its associations with other elements.

This detailed knowledge of the distribution and concentration

gradients of As sediment particles as well as its dominant

associations can already help to make qualitative predictions

about the availability and mobility of As in soils, groundwater

and sediments.

Furthermore, mS-XRF analysis helped to provide a better

understanding of geochemical alterations that occur during

sequential sediment extraction. The significant decrease of As

concentrations in coatings due to P- and Fe-leaching was

expected and confirms again that leached As was mainly

associated with Fe oxy-hydroxide coatings. Less anticipated,

however, was the release of Fe, Mn and Cr during P-leaching

as well as the lowering of mean K concentrations due to the

Fe-leaching. The consistently observed increase of the Fe/As

ratio indicates that re-precipitation processes are likely to

occur during the sequential extraction.

This study shows that elemental mappings enable the

identification of significant alterations associated with

sequential extractions. This is particularly important because

conclusions about the mobility and/or availability of As in soil

and groundwater environments are frequently based on

results derived from sequential extraction.

Acknowledgements

The authors appreciate the opportunity to use the FLUO-

beamline at ANKA, Research Center Karlsruhe (FZK). In this

context, we would like to thank Rolf Simon, Institute of

Synchrotron Radiation, beamline-scientist at the FLUO-

Beamline at ANKA, for his important help with the adjust-

ment of the experimental settings and the performance of the

experiment. We are also very grateful to all colleagues from

the CETASD and HUMG in Hanoi, Vietnam, particularly to

Pham Hung Viet, Pham Thi Kim Trang, Vi Mai Lan, Bui Hong

Nhat, Dao Manh Phu, Pham Qui Nhan for their help in

organising and realizing the field campaign. Thanks also to

the LGFG Baden-Wurttemberg for financially supporting this

work.

r e f e r e n c e s

Anawar, H.M., Komaki, K., Akai, J., Takada, J., Ishizuka, T.,Takahashi, T., Yoshioka, Y., Kato, K., 2002. Diagenetic controlon arsenic partitioning in sediments of the Meghna Riverdelta, Bangladesh. Environ. Geol. 41, 816e825.

Anawar, H.M., Akai, J., Sakugawa, H., 2004. Mobilization of arsenicfrom subsurface sediments by effect of bicarbonate ions ingroundwater. Chemosphere 54, 753e762.

Bacon, J.R., Davidson, C.M., 2008. Is there a future for sequentialchemical extraction? Analyst 133, 25e46.

Bauer, W.O., Onishi, B.M.H., 1969. Arsenic. In: Wedepohl, K.H. (Ed.),Handbook of Geochemistry. Springer, Berlin.

Berg, M., Stengel, C., Trang, P.T.K., Viet, P.H., Sampson, M.L.,Leng, M., Samreth, S., Fredericks, D., 2007. Magnitude ofarsenic pollution in the Mekong and Red River deltas e

Cambodia and Vietnam. Sci. Total Environ. 372, 413e425.Berg, M., Trang, P.T.K., Stengel, C., Buschmann, J., Viet, P.H.,

Dan, N.V., Giger, W., Stuben, D., 2008. Hydrological andsedimentary controls leading to arsenic contamination ofgroundwater in the Hanoi area, Vietnam: the impact ofironearsenic ratios, peat, river bank deposits, and excessivegroundwater abstraction. Chem. Geol. 249, 91e112.

BGS, DPHE, 2001. Arsenic contamination of groundwater inBangladesh. In: Kinniburg, D.G., Smedley, P.L. (Eds.), FinalReport, BGS Technical Report WC/00/19, vol. 2. BritishGeological Survey, Keyworth, UK.

Bhattacharya, R., Jana, J., Nath, B., Sahu, S.J., Chatterjee, D.,Jacks, G., 2003. Groundwater As mobilization in the BengalDelta Plain, the use of ferralite as a possible remedial measuree a case study. Appl. Geochem. 18, 1435e1451.

Buschmann, J., Berg, M., 2009. Impact of sulfate reduction on thescale of arsenic contamination in groundwater of the Mekong,Bengal and Red River deltas. Appl. Geochem. 24, 1278e1286.

Das, D., Samanta, G., Mandal, B.K., Chowdhury, T.R., Chanda, C.R., Chowdhury, P.P., Basu, G.K., Chakraborti, D., 1996. Arsenic ingroundwater in six districts of West-Bengal, India. Environ.Geochem. Health 18, 5e15.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 55554

Dodd, J., Large, D.J., Fortey, N.J., Milodowski, A.E., Kemp, S., 2000.A petrographic investigation of two sequential extractiontechniques applied to anaerobic canal bed mud. Environ.Geochem. Health 22, 281e296.

Dowling, C.B., Poreda, R.J., Basu, A.R., Peters, S.L., Aggarwal, P.K.,2002. Geochemical study of arsenic release mechanisms in theBengal Basin groundwater. Water Resour. Res. 38, 1173.

Eiche, E., Neumann, T., Berg, M., Weinman, B., van Geen, A.,Norra, S., Berner, Z., Trang, P.T.K., Viet, P.H., Stuben, D., 2008.Geochemical processes underlying a sharp contrast ingroundwater arsenic concentrations in a village on the RedRiver delta, Vietnam. Appl. Geochem. 23 (11), 3143e3154.

Filgueiras, A.V., Lavilla, I., Bendicho, C., 2002. Chemical sequentialextraction for metal partitioning in environmental solidsamples. J. Environ. Monit. 4, 823e857.

Gleyzes, C., Tellier, S., Astruc, M., 2002. Fractionation studies oftrace elements in contaminated soils and sediments: a reviewof sequential extraction procedures. Trends Anal. Chem. 21,451e467.

Gruebel, K.A., Daves, J.A., Leckie, J.O., 1988. The feasibility of usingsequential extraction techniques for arsenic and selenium insoils and sediments. Soil Sci. Soc. Am. J. 52, 390e397.

Harvey, C.F., Swartz, C.H., Badruzzaman, A.B.M., Keon-Blute, N.,Yu, W., Ashraf Ali, M., Jay, J., Beckie, R., Niedan, V.,Brabander, D., Oates, P.M., Ashfaque, K.N., Islam, S.,Hemond, H.F., Ahmed, M.F., 2002. Arsenic mobility andgroundwater extraction in Bangladesh. Science 298,1602e1606.

Horneman, A., Van Geen, A., Kent, D.V., Mathe, P.E., Zheng, Y.,Dhar, R.K., O’Connell, S.O., Hoque, M.A., Aziz, Z.,Shamsudduha, M., Seddique, A.A., Ahmed, K.M., 2004.Decoupling of As and Fe release to Bangladesh groundwaterunder reducing conditions. Part I: evidence from sedimentprofiles. Geochim. Cosmochim. Acta 68, 3459e3473.

Hudson-Edwards, K.A., Houghton, S.L., Osborn, A., 2004.Extraction and analysis of arsenic in soils and sediments.Trends Anal. Chem. 23, 745e752.

Jochum, K.P., Dingwell, D.B., Rocholl, A., Stoll, B., Hofmann, A.W.,et al., 2000. The preparation and preliminary characterizationof eight geological MPI-DING reference glasses for in-situmicroanalysis. Geostand. Newslett. 24, 87e133.

Keon, N.E., Swartz, C.H., Brabander, D.J., Harvey, C., Hemond, H.F.,2001. Validation of an arsenic sequential extraction method forevaluating mobility in sediments. Environ. Sci. Technol. 35,2778e2784.

Kheboian, C., Bauer, C.F., 1987. Accuracy of selective extractionprocedures for metal speciation in model aquatic sediments.Anal. Chem. 59, 1417e1423.

La Force, M.J., Fendorf, S., 2000. Solid-phase iron characterizationduring common selective sequential extractions. Soil Sci. Soc.Am. J. 64, 1608e1615.

Lowers, H.A., Breit, G.N., Forster, A.L., Whitney, J., Yount, J.,Uddin, M.N., Muneem, A.A., 2007. Arsenic incorporation intoauthigenic pyrite, Bengal Basin sediment, Bangladesh.Geochim. Cosmochim. Acta 71, 2699e2717.

McArthur, J.M., Banerjee, D.M., Hudson-Edwards, K.A., Mishra, R.,Purohit, R., Ravenscroft, P., Cronin, A., Howarth, R.J.,Chatterjee, A., Talukder, T., Lowry, D., Houghton, S.,Chadha, D.K., 2004. Natural organic matter in sedimentarybasins and its relation to arsenic in anoxic ground water: theexamples of West Bengal and its worldwide implications.Appl. Geochem. 19, 1255e1293.

Meharg, A.A., Scrimgeour, C., Hossain, S.A., Ruller, K.,Cruickshank, K., Williams, P.N., Kinniburg, D.G., 2006.Codeposition of organic carbon and arsenic in Bengal Deltaaquifers. Environ. Sci. Technol. 40, 4928e4935.

Nath, B., Berner, Z., Basu Malik, S., Chatterjee, D., Charlet, L.,Stueben, D., 2005. Characterization of aquifers conductinggroundwaters with low and high arsenic concentrations:

a comparative case study from West Bengal, India. Mineral.Mag. 69, 841e854.

Nickson, R.T., McArthur, J.M., Ravenscroft, P., Burgess, W.G.,Ahmed, K.M., 2000. Mechanisms of arsenic release togroundwater, Bangladesh and West Bengal. Appl. Geochem.15, 403e413.

Pal, T., Mukherjee, P.K., Sengupte, S., 2002. Nature of arsenicpollutants in groundwater of Bengal Delta e a case study fromBaruipur area, West Bengal, India. Curr. Sci. 82, 554e561.

Paul, C.J., Ford, R.G., Wilkin, R.T., 2009. Assessing the selectivity ofextractant solutions for recovering labile arsenic associatedwith iron (hydr)oxides and sulfides in sediments. Geoderma152, 137e144.

Polizzotto, M.L., Harvey, C.F., Li, G., Badruzzman, B., Ali, A.,Newville, M., Sutton, S., Fendorf, S., 2006. Solid-phases anddesorption processes of arsenic within Bangladesh sediments.Chem. Geol. 228, 97e111.

Radu, T., Subacz, J., Phillippi, J.M., Barnett, M.O., 2005. Effects ofdissolved carbonate on arsenic adsorption and mobility.Environ. Sci. Technol. 39, 7875e7882.

Reynolds, J.G., Naylor, D.V., Fendorf, S.E., 1999. Arsenic sorptionin phosphate-amended soils during flooding and subsequentaeration. Soil Sci. Soc. Am. J. 63, 1149e1156.

Saunders, J.A., Pritchett, M.A., Cook, R.B., 1997. Geochemistry ofbiogenic pyrite and ferromanganese stream coatings:a bacterial connection? Geomicrobiol. J. 14, 203e217.

Seddique, A.A., Masuda, H., Mitamura, M., Shinoda, K.,Ymanaka, T., Ita, T., Maruoka, T., Uesugi, K., Ahmed, K.M.,Biswas, D.K., 2008. Arsenic release from biotite intoa Holocene groundwater aquifer in Bangladesh. Appl.Geochem. 23, 2236e2248.

Sengupta, S., Mukherjee, P.-K., Pal, T., Shome, S., 2004. Nature andorigin of arsenic carriers in shallow aquifer sediments ofBengal Delta, India. Environ. Geol. 45, 1071e1081.

Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source,behaviour and distribution of arsenic in natural waters. Appl.Geochem. 17, 517e568.

Stuben, D., Berner, Z., Chandrasekharam, D., Karmakar, J., 2003.Arsenic enrichment in groundwater of West Bengal, India:geochemical evidence for mobilization of As under reducingconditions. Appl. Geochem. 18, 1417e1434.

Stute, M., Zheng, Y., Schlosser, P., Horneman, A., Dhar, R.K.,Datta, S., Hoque, M.A., Seddique, A.A., Shamsudduha,Ahmed, K.M., Van Geen, A., 2007. Hydrological control of Asconcentrations in Bangladesh groundwater. Water Resour.Res. 43, W09417.

Swartz, C.H., Blute, N.K., Badruzzman, B., Ali, A., Barbander, D.,Jay, J., Besancon, J., Islam, S., Hemnond, H.F., Harvey, C.D.,2004. Mobility of arsenic in a Bangladesh aquifer: inferencesfrom geochemical profiles, leaching data and mineralogicalcharacterisation. Geochim. Cosmochim. Acta 68, 4539e4557.

Tipping, E., Hetherington, N.B., Hildon, J., Thompson, D.W.,Bowles, E., Hamilton-Taylor, J., 1985. Artefacts in the use ofselective chemical extraction to determine distributions ofmetals between oxides of manganese and iron. Anal. Chem.57, 1944e1946.

van Geen, A., Zheng, Y., Cheng, Z., Aziz, Z., Horneman, A.,Dhar, R.K., Mailloux, B., Stute, M., Weinman, B., Goodbred, S.,Seddique, A.A., Hoque, M.A., Ahmed, K.M., 2006. A transect ofgroundwater and sediment properties in Araihazar,Bangladesh: further evidence of decoupling between As andFe mobilization. Chem. Geol. 228, 85e96.

van Geen, A., Radloff, K., Aziz, Z., Cheng, Z., Huq, M.R.,Ahmed, K.M., Weinman, B., Goodbred, S., Jun, H.B.,Zheng, Y., Berg, M., Trang, P.T.K., Charlet, L., Metral, J.,Tisserand, D., Guillot, S., Chakraborty, S., Gajurel, A.P.,Upreti, B.N., 2008. Comparison of arsenic concentrations insimultaneously-collected groundwater and aquifer particles

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 5 4 5e5 5 5 5 5555

from Bangladesh, India, Vietnam, and Nepal. Appl.Geochem. 23, 3244e3251.

van Herreweghe, S., Swennen, R., Vandecasteele, C.,Cappuyns, V., 2003. Solid phase speciation of arsenic bysequential extraction in standard reference materials andindustrially contaminated soil samples. Environ. Pollut. 122,323e342.

Wenzel, W.W., Kirchbaum, N., Prohaska, T., Stingeder, G.,Lombic, E., Adriano, D.C., 2001. Arsenic fractionation in soilsusing an improved sequential extraction procedure. Anal.Chim. Acta 436, 309e323.

Whalley, C., Grant, A., 1994. Assessment of the phase selective ofthe European Community Bureau of Reference (BCR)sequential extraction procedure for metals in sediment. Anal.Chim. Acta 291, 287e295.

Winkel, L., Berg, M., Amini, M., Hug, S.J., Johnson, C.A., 2008.Predicting groundwater arsenic contamination in SoutheastAsia from surface parameters. Nat. Geosci. 1, 536e542.

Wright, M.T., Parker, D.R., Amrhein, C., 2003. Critical evaluationof the ability of sequential extraction procedures to quantifydiscrete forms of selenium in sediments and soils. Environ.Sci. Technol. 37, 4709e4716.

Zheng, Y., van Geen, A., Stute, M., Dhar, R., Mo, Z., Cheng, Z.,Horneman, A., Gavrieli, I., Simpson, H.J., Versteeg, R.,Steckler, M., Grazioli-Venier, A., Goodbred, S., Shahnewaz, M.,Shamsudduha, M., Hoque, M.A., Ahmed, K.M., 2005.Geochemical and hydrogeological contrasts between shallowand deeper aquifers in two villages of Araihazar, Bangladesh:implications for deeper aquifers as drinking water sources.Geochim. Cosmochim. Acta 69, 5203e5218.