Journal of Colloid and Interface Science 255, 52–58 (2002)doi:10.1006/jcis.2002.8646

Arsenic Adsorption onto Pillared Clays and Iron Oxides

Veronique Lenoble, Omar Bouras, Veronique Deluchat, Bernard Serpaud, and Jean-Claude Bollinger1

Laboratoire des Sciences de l’Eau et de l’Environnement, Faculte des Sciences, 123 avenue Albert Thomas, 87 060 Limoges, France

Received April 18, 2002; accepted July 30, 2002

Arsenic adsorption was carried out on simple materials such asgoethite and amorphous iron hydroxide, and more complex ma-trices such as clay pillared with titanium(IV), iron(III), and alu-minum(III). These matrices were synthesized from a bentonitewhose montmorillonitic fraction was pillared according to opti-mized parameters. These sorbents were characterized by variousmethods: XRD, FTIR, BET, DTA/TGA, surface acidity, and ze-tametry. Elimination of arsenite and arsenate as a function ofpH was studied. Arsenate elimination was favored at acidic pH,whereas optimal arsenite elimination was obtained at 4 < pH < 9.For pH values above 10, the pillared clays were damaged and elimi-nation decreased. Equilibrium time and adsorption isotherms werealso determined for arsenite and arsenate at each matrix auto-equilibrium pH. Amorphous iron hydroxide had the highest ad-sorption capacities both towards arsenate and arsenite. Adsorp-tion capacities of goethite and iron- and titanium-pillared claystoward arsenate were similar, but those toward arsenite were differ-ent. Desorption experiments from the various matrices were carriedout. Iron- and titanium-pillared clays showed a desorption capac-ity above 95% and around 40% respectively, but no desorption ratecould be obtained for iron (hydr)oxides as they were damaged dur-ing the process. C© 2002 Elsevier Science (USA)

Key Words: arsenic; pillared clays; iron oxides; adsorption.

INTRODUCTION

Arsenic is a toxic trace element occurring in natural watersin a variety of forms including soluble, particulate, and organic-bound, but mainly as inorganic trivalent As(III) and pentavalentAs(V) oxidation states. In many parts of the world, groundwateris polluted with arsenic. This pollution can be caused by humanactivities (mining, pesticides, . . .) but usually, the main sourceof arsenic is geogenic. Epidemiological studies have demon-strated a significant increase in the risks of lung, skin, bladder,liver, and other cancers associated with high levels of arsenic indrinking water. It has also been shown that arsenic interferes withsome hormones by altering downstream receptor function (1, 2).Consequently, in the case of arsenic, the European standardlevel in drinking water has been lowered to 10 µg/L (Directive

1 To whom correspondence should be addressed. Fax: (+33) 555-457-459.E-mail: jcbollinger@unilim.fr.

520021-9797/02 $35.00C© 2002 Elsevier Science (USA)All rights reserved.

98/83/CE) and similar reductions in arsenic levels have beenadopted elsewhere, including the United States.

To remove arsenic from groundwater, classical methods suchas adsorption by activated carbon (3), membrane processes (4),or ion exchange (5, 6) can be used. However, the most commonarsenic removal method is coupled coagulation and softening(7), removing arsenic through processes of chemical sorptionand particulate removal. The prior oxidation of the trivalent formof arsenic into the less toxic pentavalent form is an importantstage in the treatment since arsenate (AsO3−

4 , As(V)) is removedmore effectively than arsenite (AsO3−

3 , As(III)) (8–10). But, thismethod is too costly in terms of investment and running coststo be applied in plants of low flows (<10 m3/h) in sparselyinhabited zones (e.g., the center of France or some areas aroundthe world where major environmental problems with arsenichave been identified (11)). Therefore, it is essential that methodsthat allow those installations to supply drinkable water at lowcost should be developed.

Clays or modified clays are used worldwide as adsorptive me-dia for heavy metals (12, 13), bacteria (14), or organic contami-nants (15). Experiments have shown that clay functionalizationcan be optimized by matching clay structure with a suitablereactive (16).

In this study, a montmorillonite was pillared (17): a polyca-tion was intercalated between clay layers in order to obtain amicroporous material with increased interlamellar spaces andpore volumes. This allows greater thermal stability and makesthe material a potential catalyst and adsorbent (17) as favor-able adsorption sites are created. Titanium(IV), iron(III), andaluminum(III) polycations were used for intercalation. Arseniteand arsenate adsorption was then carried out on these materi-als. Two iron oxyhydroxides (goethite and amorphous iron hy-droxide) were synthesized and used in arsenic adsorption as acomparative test. The desorption efficiency of the investigatedmatrices was also tested.

MATERIALS AND METHODS

Reagents

All chemicals were of analytical grade and used without fur-ther purification. All solutions were prepared with high puritydeionized water (resistivity 18.2 M� · cm, TOC < 10 µg/L) ob-tained with a Milli-Q water purification system. All glassware

O

ARSENIC ADSORPTION

was cleaned by soaking in 10% HNO3 and rinsed three timeswith deionized water. The arsenate stock solution was preparedfrom sodium salt heptahydrate Na2HAsO4 · 7H2O (Fluka, purity>98.5%). The arsenite stock solution was prepared from sodium(meta)arsenite NaAsO2 (Fluka, purity >99%).

Iron Oxides

Amorphous Iron Hydroxide

Amorphous iron hydroxide, designated by HFO (according toDzombak and Morel (18)), was prepared according to a slightlymodified Schwertmann and Cornell (19) protocol. To 500 mL of0.5 M Fe(NO3)3 · 9H2O (Prolabo, purity >98%), 60 g of NaOHpellets (Prolabo, purity >98%) were added slowly. The solutionwas continuously bubbled with nitrogen during the synthesisto prevent carbonate formation. The HFO precipitate was cen-trifuged and washed three times with deionized water. The solidobtained was dried at 25◦C for 24 h, ground for homogenizationand sheltered from light.

Goethite

Goethite was also prepared according to Schwertmannand Cornell (19) by thoroughly mixing 100 mL of 1 MFe(NO3)3 · 9H2O and 180 mL of 5 M NaOH under nitrogen bub-bling. The solution was then diluted to 2 L and put in an oven at70◦C for 60 h. The precipitate was centrifuged and washed threetimes with deionized water. The solid obtained was dried at 60◦Cfor 24 h, ground for homogenization, and sheltered from light.

Pillared Clays

These syntheses were already optimized in a previousstudy (16).

Clay Preparation

The starting material was a sample of bentonite from Maghnia(Algeria) supplied by ENOF (Entreprise Nationale des Sub-stances Utiles et Materiaux non Ferreux), whose chemical com-position was known (20). The preparation was performed bydispersing rough bentonite in a 1 M NaCl solution to replace allexchangeable cations with Na+, washing with deionized water,separation by centrifugation to eliminate all other solid phases,and recovery of the montmorillonitic fraction (<2 µm) by de-cantation. The final solid was designated by Na-Montm.

Pillared Solutions

Three pillared solutions of titanium(IV), iron(III), and alu-minum(III) called respectively PCBT, PCBF, and PCBA wereprepared for use in pillaring Na-Montm. PCBT: 27.5 mL oftitanium chloride (TiCl4, Acros, purity 99.9%) and 50 mL of6 M HCl (Prolabo) were mixed with as much stirring as nec-essary to dissolve TiCl4 vapor. To this solution, 227 mL of

deionized water was added slowly. PCBF: 100 mL of 0.43 MFe(NO3)3 · 9H2O (Prolabo), and 114.67 mL of 0.75 M NaOH

NTO PILLARED CLAYS 53

(Prolabo) were mixed with a vigorous stirring. PCBA: 50 mLof 0.5 M AlCl3 · 6H2O (Acros), and 220 mL of 0.225 M NaOHwere also mixed with vigorous stirring. The final solutions sup-plied polycations, symbolized respectively by Tix Hy , Fex (OH)y ,and Alx (OH)y . Solutions were kept in darkness for ageing.

The intercalation of Na-Montm by the pillaring solutionswas obtained with the following parameters: final concentra-tion [Ti]f : 0.82 M, [Fe]f : 0.2 M, and [Al]f : 0.1 M, molar ratioH+/Ti = 1.2; OH/Fe = 2, and OH/Al = 2, and solution agePCBT : 3 hours, PCBF : 10 days, and PCBA : 2 days.

Preparation of Pillared Montmorillonites

The given pillaring solution was added drop by drop to Na-Montm suspensions (concentration between 0.1 and 0.5% w/w).Different ratios of pillaring solutions/Na-Montm were studied:PCBT/Na-Montm = 1, 5, 10, 20, and 40 mmol/g; PCBF/Na-Montm = 2, 4, 5, 7, and 10 mmol/g. Studies had already beencarried out for Al-pillared clay (20). After filtration and washingthree times with deionized water, the solid obtained was driedat 40◦C for at least 72 h, ground, and sheltered from light.

According to procedures adopted elsewhere (16, 20), thepillared compounds obtained with pillaring solutions of tita-nium, iron, and aluminum were designated by Tix Hy-Montm,Fex (OH)y-Montm, and Alx (OH)y-Montm respectively.

Solid Characterization

Iron oxides and modified clay sorbents were characterizedusing different methods, in order to define their structure as wellas possible. Crystallization states of solids and, in the case of thethree pillared clays, optimization of the ratios and of the basalspacing between adjacent layers, were determined by X-raydiffraction (XRD) analysis on a Siemens D5000 diffractome-ter using filtered copper Kα1 radiations. Oriented clay filmswere prepared by spreading a few drops of the pillared clay sus-pension on a glass plate (30 × 45 mm2). The film was dried for24 h at room temperature before analysis. This method is rec-ommended in order to obtain a high basal spacing signal (20).

Infrared spectra (FTIR) were obtained using KBr pellets(Acros, IR grade 99%, 1 mg of solid with 150 mg of KBr)with a Perkin–Elmer Fourier transform spectrometer at 2 cm−1

resolution. Cationic exchange capacity (CEC) was measuredby the ammonium acetate method (ISO Norm 11260, 1994).Specific surface areas (SSA) were determined according to theBrunauer–Emmet–Teller (BET) protocol on a MicromeriticsASAP 2000 apparatus. Surface charges were determined frompotentiometric titrations of 1 g/L solid suspension with NaOH(0.001 M) and HNO3 (0.001 M) in a 0.01 M NaNO3 mediumas the supporting electrolyte (21, 22), the pH meter was a PHM250 (METERLAB). This allows pHzpt, the pH of zero point oftitration to be determined. Zetametric experiments were carriedout on a Zetaphoremeter II, model Z3000, to back up the resultsobtained by titration.

Any chemical reaction, heat release or absorption, or changeof state was detected by differential thermal analysis (DTA).

E

54 LENOBL

Determination of any volatile compound emitted or absorbedby the sample was underscored as a function of temperature bythermogravimetry (TGA). Both DTA and TGA measurementswere performed on Setaram Labsys apparatus.

Arsenic Analysis

Arsenic analysis was carried out using a Varian SpectrAA800 graphite furnace atomic absorption spectrometer (GFAAS),with Zeeman background correction. All measurements werebased on integrated absorbance and performed at 193.7 nm byusing a hollow cathod lamp (Varian). The modifier used was apalladium–magnesium mixture. Pretreatment temperature was1400◦C; atomization temperature was 2500◦C. The calibrationrange was 20–100 µg/L of arsenic. The accuracy was 5%; theRSD was ±7% (repeatability tests, n > 100).

Adsorption Experiments

The adsorption studies were performed separately on arsen-ite or arsenate solutions. Experiments were carried out with asorbent concentration of 1.6 g solid/L. Each solid was mixedwith arsenic solution in flasks which were closed and put on anorbital shaker (IKALABORTECHNIK KS 501) at 200 rpm.

The solids are aimed to be applied in small drinking waterplants; therefore experiments were conducted without adjustingthe pH of the solutions. Equilibration time was determined forarsenite and arsenate. At regular time intervals, solutions werefiltered on Whatman ashless filters (Prolabo, porosity 8 µm,enough to prevent matrices getting through), their arsenic con-centration was measured by GFAAS. Adsorption isotherms wererealized in order to work out arsenic adsorption onto each matrixas a function of matrix autoequilibrium pH and surface charge,and according to arsenite/arsenate speciation (23).

At the same time, arsenite and arsenate elimination by eachmatrix as a function of pH was determined. The pH of arsenicand solid suspensions was adjusted with concentrated HCl orNaOH.

Desorption Procedure

As(V) and As(III) desorption was carried out by immersingthe dry As-loaded matrix in 1 M HCl (Prolabo) for a periodequal to the sorption time. The volumes used for adsorption andfor desorption were respectively Va = 50 mL and Vd = 25 mLand the ratio Vd/Va equaled 0.5 (24, 25). The quantity of arsenicgoing into solution was then determined by GFAAS.

RESULTS AND DISCUSSION

Characterization

Iron Oxides

XRD pattern and FTIR spectra showed that the synthesizediron (hydr)oxides were well-defined mineral phases: goethite

and HFO (results not shown). This is backed up by the com-

ET AL.

0

10

20

30

40

50

60

0 200 400 600 800 1000 1200

Temperature (°C)

Hea

tfl

ow(a

rbit

rary

unit

s)

-35

-30

-25

-20

-15

-10

-5

0

5A

B

C

D

del

tam

(%)

exo

FIG. 1. Thermal analysis of iron oxides: DTA and TGA studies, respec-tively, of goethite (A, C) and of HFO (B, D).

parison with literature data (26–29). DTA/TGA studies (Fig. 1)showed a thermal stability up to 350◦C for HFO (as spectrashowed an exothermic peak, therefore a structural change, nearthis temperature), and up to 900◦C for goethite (as there was nostructural change). Mass losses corresponding to dehydration (–H2O) and dehydroxylation (–OH) were respectively of 24% at140◦C and 4% at 370◦C for HFO, and of 7% at 110◦C and 11%at 300◦C for goethite. Zetametry and surface acidity measure-ments allowed pHzpt values, i.e., 6.7 and 5.0, for goethite andHFO respectively, to be determined. HFO pHzpt was quite lowcompared to Dzombak and Morel values (18), but this can beexplained by HFO synthesis: NaOH pellets were used insteadof a solution, pH variations within the solution were thereforewider and this may be responsible for the low pHzpt. BET proto-col gave a SSA of 39 m2/g for goethite and 200 m2/g for HFO, inagreement with earlier data from Davranche and Bollinger (30)and Raven et al. (31).

Pillared Clays

Results are reported in Table 1. Figures 2a and 2b present theXRD lines obtained for different ratios of PCBF/Na-Montmand PCBT/Na-Montm. For PCBF/Na-Montm ratios of 7and 10 mmol/g and for PCBT/Na-Montm ratios of 10 and20 mmol/g, there were two peaks, therefore two basal spacingvalues. Depending on the pillaring agent used, the basalspacing between adjacent layers changes and creates favorable

TABLE 1Characterization of Pillared Clays: Results from the Present

Study and Bouras et al. (16)

Sample d001 (A)a SSA (m2/g) CEC (meq/100 g)

Raw bentonite 14 54 65Na-Montm 15 91 78

(9.6 at 200◦C)Tix Hy -Montm 26 249 10Fex (OH)y -Montm 22 165 12Alx (OH)y -Montm 20 229 14

a After thermal treatment.

O

ARSENIC ADSORPTION

FIG. 2. XRD of different PCBF/Na-Montm (a) and PCBT/Na-Montm(b) for various ratios r (mmol/g) and the corresponding basal spacing d (A):A r = 2 d = 14, B r = 4 d = 14, C r = 10 d = 20.3 and d = 14, D r = 7 d = 22.6and d = 14, E r = 40 d = 20, F r = 5 d = 14, G r = 20 d = 26.7 and d = 14,H r = 10 d = 25 and d = 14, I r = 1 d = 16. Step = 0.04◦ and step time = 32 s.

adsorption sites in the microporous system. The convenient ratiois the one giving the larger basal spacing; as SSA increased, ad-sorption is favored. Consequently, the ratios used as a referencefor PCBF/Na-Montm and PCBT/Na-Montm were respectivelyr = 7 mmol/g and r = 20 mmol/g. The ratio PCBA/Na-Montmused was 4 mmol/g, as previously shown (20).

As shown in Table 1, after intercalation the d(001) basal spac-ing increased considerably up to 26 A for Tix Hy-Montm, 22 A

˚

for Fex (OH)y-Montm and 20 A for Alx (OH)y-Montm in com-parison with the precursor material, which presented only 9.6 A

NTO PILLARED CLAYS 55

4000 3000 2000 1500 1000 400cm-1

%T(arbitrary

units)

A

B

C

D

FIG. 3. FTIR spectra of Na-Montm (A), Fex (OH)y -Montm (B), Alx (OH)y -Montm (C), and Tix Hy -Montm (D).

for dehydrated Na-Montm. The (BET) SSA of the three studiedpillared clays were evaluated at −200◦C; prior to measurements,the samples were outgassed at 160◦C. The results were in therange 160–250 m2/g, which was much larger than those ob-tained for the precursor Na-Montm and consequently confirmedthe XRD results. They suggested the creation of microporoussystem in the interlayer spaces of these pillared clays. Theseobservations were reported also by our previous study (20).

The low values of CEC suggested the irreversibility ofcationic exchange: the intercalated metallic polycations werehardly exchanged. Thus, CEC represented only the exchangeof monomeric or dimeric species that coexisted with the poly-cations in the pillaring solution. FTIR of clay sorbents (Fig. 3)showed that Na-Montm had a strong hydrophilic effect provenby the band 2500–3800 cm−1, whereas the modified clay solidspresented a hydrophobic effect as this band no longer appeared.The thermal stability of all sorbents used was shown by theDTA/TGA results (Fig. 4). These curves confirmed good thermal

-60

-40

-20

0

20

40

60

0 200 400 600 800 1000 1200

Temperature (°C)

Hea

tfl

ow(a

rbit

rary

uni

ts)

exo

AB

C

D

E

FG

H

Del

tam

(arb

itra

ryu

nit

s)

FIG. 4. Thermal analysis of clay sorbents: DTA and TGA, respectively, of

Na-Montm (A, E), Fex (OH)y -Montm (B, F), Alx (OH)y -Montm (C, G), andTix Hy -Montm (D, H).

E

56 LENOBL

stability of the pillared clays up to 900◦C except for Fex (OH)y-Montm which presented a structural change at 350◦C. Theyshowed large endothermic peaks located at 70–160◦C and 450–530◦C respectively due to the dehydration and dehydroxyla-tion of pillared clay samples. Zetametric and pH-metric exper-iments showed a negatively charged surface for Tix Hy-Montmand Fex (OH)y-Montm for the pH values studied, and a pHzpt of5.0 for Alx (OH)y-Montm (results not shown).

Equilibrium Time

Results are shown in Figs. 5a and 5b (size of symbols waschosen to correspond to uncertainty). After 4 h, the adsorbedquantity was stable for all the solids and for both arsenite and ar-senate, except for Alx (OH)y-Montm. Therefore, an equilibriumtime of 4 h was chosen for the following experiments concern-ing all matrices but Alx (OH)y-Montm. This equilibrium timeis in agreement with Raven et al. (31) and Pierce and Moore(32) who observed that 99% of the maximum observed arsen-ite and arsenate adsorption on ferrihydrite was complete within4 h of reaction initiation. Alx (OH)y-Montm presented a loweradsorption capacity than the other matrices. The removal is notproportionally related to the surface area, so SSA is not the pri-mary factor controlling the interaction of arsenite and arsenatewith the different matrices.

0

0.5

1

0 4 8 12Time (h)

C/C

o

0

0.5

1

0 4 8 12

Time (h)

C/C

o

b

a

FIG. 5. Determination of arsenate (a) and arsenite (b) equilibrium timeusing Fex (OH)y -Montm (�), HFO (�), goethite (�), Tix Hy -Montm (�), andAlx (OH)y -Montm (×) at autoequilibrium pH. Co and C are respectively the

initial and final concentration in arsenite or arsenate. Symbol size correspondsto experimental uncertainty.

ET AL.

FIG. 6. Arsenate (a) and arsenite (b) elimination as a function of pH byadsorption onto Fex (OH)y -Montm (�), HFO (�), goethite (�), and Tix Hy -Montm (�).

pH Influence on Arsenic Removal

Adsorption experiments were carried out on each solid ex-cept Alx (OH)y-Montm, whose adsorption ability was previouslyproven to be very low, as seen above.

Results are shown in Figs. 6a and 6b. Under acidic conditions(3 < pH < 5), arsenate and arsenite removal was total, whereasat more neutral pH values (6 < pH < 8) arsenite was betteradsorbed. This agrees with data from Lombi et al. (23), andwith other results obtained on carbon-based adsorbents and onzerovalent iron (3, 33). Tix Hy-Montm was the least efficient ma-trix in both cases. As for HFO, the results were similar to thosedescribed by Wilkie and Hering (34).

For basic pH values (pH > 10), the pillared clays were dam-aged (changes in solutions colour and solids aspect) and arsenicelimination from the aqueous phase decreased.

Adsorption Isotherms

The solids are aimed to be used in small drinking water plants,therefore adsorption studies were carried out without pH ad-justment. Matrix autoequilibrium pH was measured as a func-tion of arsenic concentration. The trend was similar for arsenateand arsenite (Fig. 7). Arsenic concentration did not influencepH with goethite whereas with HFO, Fex (OH)y-Montm, andTix Hy-Montm the pH increased with the arsenic concentration.For pillared clays, pH was acidic at low concentrations and be-come circumneutral at high arsenic concentrations. The increaseof pH was not proportional to arsenic concentration and for Asconcentrations higher than 20 mg/L, pH reached a plateau.

According to the arsenic species stability diagram, triva-

lent arsenic is stable at pH 0–9 as uncharged H3AsO3 (pKa3

equals 9.29), while for pentavalent arsenic, the stable species are

57

Montm (pH. Sym

ARSENIC ADSORPTION ONTO PILLARED CLAYS

2

4

6

8

10

0 20 40 60 80

As concentration (mg/L)

fina

l pH

6

8

10

0 20 40 60 80As concentration (mg/L)

fina

l pH

2

4

6

8

10

0 20 40 60 80

As concentration (mg/L)

fina

l pH

2

4

6

8

10

0 20 40 60 80

As concentration (mg/L)

fina

l pH

Fex(OH)y- Montm Goethite

HFOTixHy-Montm

FIG. 7. Fex (OH)y -Montm, Tix Hy -Montm, goethite, and HFO autoeq

anionic: H2AsO−4 (pH 3–6) and HAsO2−

4 (pH 7–11), pKa2 andpKa3 values being 6.96 and 11.5, respectively (23).

Figures 8a and 8b present the adsorption isotherms obtained atautoequilibrium pH for each matrix. Symbol size corresponds toexperimental uncertainty (3% for arsenite and 5% for arsenate)calculated from duplicate isotherms. Table 2 presents matrixautoequilibrium pH, surface charge, and maximum capacities(obtained from a polynomial NLLS adjustment) for arsenite andarsenate species.

For Fex (OH)y-Montm and Tix Hy-Montm, surface charge andarsenic species explained the stronger arsenite adsorption. Max-imal capacities were the same for these two matrices.

0

2

4

6

8

10

0 20 40 60

As(V) concentration (mg/L)

adso

rbed

qu

anti

ty(

mg

As/

gm

at)

0

10

20

30

40

0 20 40 60

As(III) concentration ( mg/L)

adso

rbed

quan

tity

(m

gA

s/g

mat

)

b

a

FIG. 8. Arsenate (a) and arsenite (b) adsorption isotherms at Fex (OH)y -

�), HFO (�), goethite (�), and Tix Hy -Montm (�) autoequilibriumbol size corresponds to experimental uncertainty.

uilibrium pH as a function of As(III) (�) and As(V) (�) concentration.

For goethite, again, arsenite adsorption was stronger than ar-senate one. It can be noticed that Fex (OH)y-Montm, Tix Hy-Montm, and goethite presented the same maximal capacitiestoward arsenate: the adsorption process was not a strictly elec-trostatically controlled mechanism.

As for HFO, surface charge was positive for arsenite species,as the isotherm concentrations were below 30 mg/L. Arsenitespecies were neutral, so HFO maximal capacity toward arsen-ite was the highest. For arsenate isotherm, due to the pH val-ues, solid surface charge was mainly negative, so the maximalcapacity was lower for arsenate anionic species. Furthermore,for arsenate concentrations above 30 mg/L, the negative sur-face charge and the double negatively charged arsenate anionexplained the decrease in adsorption capacity. Clifford and Lin(5) obtained an adsorption capacity of 29.7 mg As(III)/g HFO,which agrees with our maximum capacity. Pierce and Moore(32) have suggested that the high adsorptive capacity of HFOcould be explained by its loose and highly hydrated structure.Ions could be free to diffuse and would not be restricted toexternal surface sites. Concerning kinetics, other studies haveshown that arsenate adsorption onto HFO was realized throughinner-sphere bidentate complexes (34, 35) and that the reactionsbetween arsenate and HFO were diffusion-controlled (31). Theyalso showed that As(III) adsorbed onto goethite was not stable:more than 20% of the As(III) was oxidized to As(V) over20 days (36).

Desorption Capacity

Clay regeneration could be done using 1M HCl (24, 25) orphosphate (33, 37). In our case, HCl seemed the most suitablereagent because as phosphate presents the same affinity as ar-senate, the phosphate-loaded matrix could no longer adsorb ar-senic. The HCl treatment can not be applied to iron (hydr)oxides

because these matrices are severely damaged during such aprocedure. The pillared clays presented various desorption

58 LENOBLE ET AL.

TABLE 2Matrix Autoequilibrium pH, Surface Charge, and Maximal Capacities for As(III) and As(V) Species

HFO Goethite Fex (OH)y -Montm Tix Hy -Montm

Autoequilibrium pH <9 =9 <9 <9pHzpt 5.0 6.7 — —

As(III) adsorption Solid surface charge >0 up to <0 <0 <0[As(III)] = 30 mg/L

As(III) species H3AsO3 H3AsO3/H2AsO−3 H3AsO3 H3AsO3

As(III) maximal capacity 28 22 13 13(mg As/g solid)

As(V) adsorption Solid surface charge <0 for <0 <0 <0[As(V)] > 17 mg/L

As(V) species H2AsO−4 /HAsO2−

4 HAsO2−4 H2AsO−

4 /HAsO2−4 H2AsO−

4 /HAsO2−4

As(V) maximal capacity 7 4 4 3

(mg As/g solid)

efficiencies. Iron-pillared clay showed a desorption capacityabove or equal to 95% towards arsenate, around 80% towardsarsenite. Titanium-pillared clay showed a desorption capacityaround 40% for As(V), and around 30% for As(III).

Thus, Fe-pillared clay seems to be a good matrix for arsenicremoval in water treatment. It corresponds to a compromisebetween an arsenate adsorption similar to that obtained withgoethite and an easy and effective desorption ability.

ACKNOWLEDGMENTS

This work was financially supported by the “Contrat de plan Etat-RegionLimousin.” The authors thank Professor H. Bril (Department of Geology, Uni-versity of Limoges) and Dr. B. Novack (Eidgenossische Technische Hochschule,Zurich) for useful discussions, and SPCTS (Department of Materials, Universityof Limoges) for access to some characterization techniques.

REFERENCES

1. Kaltreider, R. C., Davis, A. M., Lariviere, J. P., and Hamilton, J. W., Environ.Health Perspect. 109, 245 (2001).

2. Jain, C. K., and Ali, I., Water Res. 34, 4304 (2000).3. Pattanayak, J., Mondal, K., Mathew, S., and Lalvani, S. B., Carbon 38, 589

(2000).4. Sancha, A. M., O’Ryan, R., and Perez, O., in “Proceedings of the Santiago

Symposium, September 1998,” IAHS, Publ. No. 260, p. 17. 2000.5. Clifford, D., and Lin, C.-C., “Arsenic (III) and Arsenic (V) Removal from

Drinking Water in San Ysidro, New Mexico,” EPA/600/S2-91/011. U.S.EPA, Cincinnati, 1991.

6. Guenegou, T., Tambute, A., Jardy, A., and Caude, M., Analusis 26, 352(1998).

7. Wang, Y., and Reardon, E. J., Appl. Geochem. 16, 1241 (2001).8. Borho, M., and Wilderer, P., Water Sci. Technol. 34, 25 (1996).9. Meng, X., Bang, S., and Korfiatis, G. P., Water Res. 34, 1255 (2000).

10. Subramanian, K. S., Viraragavan, T., Phommavong, T., and Tanjore, S.,Water Qual. Res. J. Canada 32, 551 (1997).

11. Hug, S. J., Canonica, L., Wegelin, M., Gechter, D., and Von Gunten, U.,Environ. Sci. Technol. 35, 2114 (2001).

12. Celis, R., Hermosin, M. C., and Cornejo, J., Environ. Sci. Technol. 34, 4593(2000).

13. Lothenbach, B., Furrer, G., Scharli, H., and Schulin, R., Environ. Sci. Tech-nol. 33, 2945 (1999).

14. Herrera, P., Burghard, R. C., and Phillips, T. D., Vet. Microbiol. 74, 259(2000).

15. Dentel, S. K., Jamrah, A. I., and Sparks, D. L., Water Res. 32, 3689 (1998).16. Bouras, O., Chami, T., Houari, M., Khalaf, H., Bollinger, J. C., and Baudu,

M., Environ. Technol. 23, 405 (2002).17. Schooheydt, R. A., Pinnavaia, T., Lagaly, G., and Gangas, N., Pure Appl.

Chem. 71, 2367 (1999).18. Dzombak, D. A., and Morel, F. M. M., “Surface Complexation Modeling,

Hydrous Ferric Oxide.” Wiley–Interscience, New York (1990).19. Schwertmann, U., and Cornell, R. M., “Iron Oxides in the Laboratory,

Preparation and Characterization,” 2nd ed., VCH Verlag, Weinheim, 1996.20. Khalaf, H., Bouras, O., and Perrichon, V., Microporous Mater. 8, 141

(1997).21. Sigg, L., and Stumm, W., Colloids Surface 2, 101 (1981).22. Kummert, R., and Stumm ,W., J. Colloid Interface Sci. 75, 1170 (1980).23. Lombi, E., Wenzel, W. W., and Sletten, R. S., J. Plant. Nutr. Soil. Sci. 162,

451 (1999).24. Elizalde-Gonzalez, M. P., Mattusch, J., Einicke, W.-D., and Wennrich, R.,

Chem. Eng. J. 81, 187 (2001).25. Elizalde-Gonzalez, M. P., Mattusch, J., and Wennrich, R., J. Environ. Monit.

3, 22 (2001).26. Schwertmann, U., and Cornell, R. M., “The Iron Oxides, Structure, Prop-

erties, Reactions, Occurrence and Uses,” 2nd ed. VCH Verlag, Weinheim,1996.

27. Jambor, J. L., and Dutrizac, J. E., Chem. Rev. 98, 2549 (1998).28. Cambier, P., Clays Min. 21, 191 (1986).29. Towe, K. M., and Bradley, W. F., J. Colloid Interface Sci. 24, 384 (1967).30. Davranche, M., and Bollinger, J. C., J. Colloid Interface Sci. 232, 165

(2000).31. Raven, K. P., Jain, A., and Loeppert, R. H., Environ. Sci. Technol. 32, 344

(1998).32. Pierce, M. L., and Moore, C. B., Water Res. 16, 1247 (1982).33. Su, C., and Puls, R. W., Environ. Sci. Technol. 35, 1487 (2001).34. Wilkie, J. A., and Hering, J. G., Colloids Surfaces A 107, 97 (1996).35. Goldberg, S., and Johnston, C. T., J. Colloid Interface Sci. 234, 204 (2001).

36. Sun, X., and Doner, H. E., Soil Sci. 163, 278 (1998).37. Manning, B. A., and Goldberg, S., Environ. Sci. Technol. 31, 2005 (1997).