Colloids and Surfaces A: Physicochem. Eng. Aspects 379 (2011) 109–115
Contents lists available at ScienceDirect
Colloids and Surfaces A: Physicochemical andEngineering Aspects
journa l homepage: www.e lsev ier .com/ locate /co lsur fa
rsenate adsorption on an Fe–Ce bimetal oxide adsorbent: EXAFS study andurface complexation modeling
iaomin Doua,b,∗∗, Yu Zhangb,∗, Bei Zhaob, Xiaomei Wub, Ziyu Wuc, Min Yangb
College of Environmental Science and Engineering, Beijing Forestry University, 100083, Beijing, ChinaState Key Laboratory of Environmental Aquatic Chemistry, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, P.O. Box 2871, 100085, Beijing, ChinaBeijing Synchrotron Radiation Facility, Institute of High Energy Physics, Chinese Academy of Sciences, 100049, Beijing, China
r t i c l e i n f o
rticle history:eceived 30 June 2010eceived in revised form 17 October 2010ccepted 18 November 2010vailable online 25 November 2010
eywords:rsenatedsorption
a b s t r a c t
The mechanism of arsenate (As(V)) adsorption on an Fe–Ce bimetal (hydrous) oxide (Fe–Ce) was inves-tigated using complementary analysis techniques including extended X-ray absorption fine structure(EXAFS) and surface complexation modeling. The As K-edge EXAFS spectra showed that the second peakof Fe–Ce after As(V) adsorption was the As–Fe shell, which supported the finding that As(V) adsorptionoccurred mainly at the Fe surface active sites. Two As–Fe distances of 3.30 A and 3.55 A were observedfrom the EXAFS spectra of As(V) adsorbed samples at three pH levels (5.0, 7.6, and 9.0) and two ini-tial surface loadings of 70 and 130 mg/g, which indicated that monodentate mononuclear and bidentatebinuclear As surface complexes coexisted. When compared with the reported dominant species of biden-
e–Ce bimetal oxideurface complexesXAFSD-MUSIC model
tate binuclear complex for As existing on iron (hydro)oxides, the existence of Ce atoms in the bimetaloxide and the high surface loading were the likely reasons for the existence of the monodentate com-plex. A Charge Distribution-Multi-site Sites Complexation (CD-MUSIC) model showed that protonatedmonodentate (MH) and deprotonated bidentate (B) complexes preferred to exist on the Fe–Ce surfacein a high surface loading range (� = 5.11–14.4 �mol/m2). The MH complex was shown to be dominant
sultssurfa
at pH < 8. Based on the reAs(V) on Fe–Ce with high
. Introduction
Arsenic (As) occurs widely in groundwater through natural ornthropogenic sources, and people in many parts of the world arexposed to As via drinking water supplies [1]. Due to the increasedwareness of the health risks of As, a great deal of effort has beenevoted to the development of adsorbents for the removal of Asrom water [2]. For example, the use of iron-, manganese-, alumina-titanium-oxide, as well as their composite oxides, have been
nvestigated extensively for the removal of As from water [2–8].mong these, ferric oxide-based materials have received the great-st attention because they have promising binding affinity for Asnd are inexpensive [2].
The mechanism of As(V) adsorption on iron (hydro)oxides
as been extensively investigated using the modern surface andtructure analytical techniques, such as extended X-ray absorptionne structure (EXAFS), X-ray photoelectron spectroscopy (XPS),T-, DR-, ATR-) Fourier-transform infra-red spectroscopy (FTIR), or
their combinations [3,7,8]. Among them, EXAFS is a powerful in-situtechnique that can provide local atomic structural information suchas the coordination number, inter-atomic distance and the natureof the neighboring atoms [3,9–11]. It has been used to directlyidentify species of As adsorbed onto iron (hydro)oxides containingminerals and adsorbents [12–18]. The presence of a bidentate bin-uclear complex (2C) with an As–Fe inter-atomic distance of about3.2 A under relatively high surface loading conditions has beenreported in most studies conducted to date, including goethite[13,15,19], ferrihydrite [13] and lepidocrocite [13,18]. These find-ings have been supported by density functional theory calculations[13]. In addition, 2C was observed on other oxides such as gibbsite[20], synthetic birnessite [6] and titanium dioxide [3]. The othertwo complex forms of bidentate-mononuclear complex (2E) andmonodentate complex (1V), in which the anticipated As–Fe inter-atomic distances are 2.85 and 3.60 A, respectively, have also beensuggested, but controversy remains regarding their possible exis-tence [10,13,16,19,21,22]. Recently, a study based on IR and EXAFSreported that monodentate coordination was, in fact, the predom-
inant geometry of arsenate-goethite surface complexes, and thatthere is no evidence for bridging-bidentate coordination [23].
Additionally, the thermodynamically based surface complex-ation model (SCM), based on spectroscopic observations, is apowerful tool for simulation of the surface reaction behavior and
or providing the protonation state of the adsorbed species [24].he most widely used SCM is the one-pK/multi-site model, whichonsiders both the charge distribution (CD) over the surface lig-nds and the diverse and heterogeneous surface sites. This modelses the Pauling concept of the CD and is an extension of theulti-site complexation (MUSIC) approach [14]. Based on EXAFS
nalysis of As(V)/As(III) adsorption on goethite, the protonationtates of surface complexes and their distribution were simulatedsing the CD-MUSIC model [16,25]. Similarly, monomethylar-onic acid (MMA)/dimethylarsinic acid (DMA) adsorption on TiO2as investigated, and non-protonated bidentate and monodentate
omplexes were suggested, respectively [11].As discussed above, most studies conducted to evaluate As
dsorption mechanisms have focused on single metal oxides suchs ferric-, alumina-, manganese-oxides or TiO2 [3–7,16,26,27].owever, there is little information available regarding bimetalxide adsorbents such as Fe–Mn, Fe–Zr, and Ce–Ti oxides, whichave been reported for As adsorption [28–30]. Recently, an
ron–cerium bimetal oxide (Fe–Ce) adsorbent was successfullyeveloped for As(V) removal [31–33], which showed a significantlyigher As(V) adsorption capacity than the referenced cerium and
erric oxides (CeO2 and Fe3O4) prepared by same procedure andome other adsorbents reported recently [31]. Also, it has good per-ormance characteristic of both cerium oxide (e.g. high affinity) anderric oxide (e.g. resistance to acids and bases, low solubility andheap). Additionally, batch and long-term column studies of arsen-te adsorption performance using groundwater samples from Innerongolia and the suburbs of Beijing, respectively, have shown that
he Fe–Ce adsorbent was promising [32]. Moreover, XPS resultshowed the existence of abundant metal hydroxyl groups on theurface of Fe–Ce, among which Fe-OH mainly reacted with As. Theonodentate species was assumed to be formed on Fe–Ce at pHafter As(V) absorption based on the triplet splitting of the As-O
3 vibration in T-FTIR and XPS analysis [31]. However, direct in-itu evidence explicating the adsorption mechanism has yet to beealized for the Fe–Ce bimetal adsorbent.
In the present study, a combination of EXAFS and the CD-MUSICodel was used to investigate the As(V) adsorption mechanism on
he Fe–Ce surface as a function of pH and surface loading. Specifi-ally, the objectives of this study were as follows: (1) to investigatehe forms of surface complex of As(V) on Fe–Ce and their relativeistribution at different pH and surface loadings by using a combi-ation of EXAFS and the CD-MUSIC model; and (2) to simulate thedsorption behavior of As(V) on Fe–Ce using the CD-MUSIC model.
. Materials and methods
.1. Materials
All chemicals used were of analytical reagent grade. 1000 mg L−1
s(V) stock solutions were prepared by dissolving 4.1653 ga2HAsO4·7H2O in 1 L of distilled water. As(V)-bearing solutionas prepared by diluting As(V) stock solution to the given concen-
rations with distilled water. Fe–Ce bimetal oxide adsorbents wererepared using a previously described co-precipitation method33].
.2. Methods
.2.1. Zeta potential measurements
The zeta potential (�) was measured for a 0.05 g/L Fe–Ce sus-
ension with 0.1 and 1.0 mM As(V), and without As(V) in the pHange 3 to 10 using a Zetasizer 2000 (Malvern Instruments Inc.,nited Kingdom), with 0.01 M NaClO4 added as the backgroundlectrolyte. Samples for testing were prepared according to the fol-
chem. Eng. Aspects 379 (2011) 109–115
lowing procedures. Fe–Ce suspensions with or without As(V) atthe desired pH and ion strength were shaken at 25 ◦C and 180 rpmfor 24 h. The equilibrium pH was then measured and the suspen-sion was injected into the electrophoretic cell for Zeta potentialmeasurement in triplicate, after which the average reading wasrecorded. The pH at the point of zero charge (PZC) was obtainedby interpolating the zeta potential data to the zero potential.
2.2.2. EXAFS data collection and analysisSix Fe–Ce samples were prepared as a function of surface load-
ings (70 and 130 mg As(V)-adsorbed/g-Fe–Ce) and pH values (5.0,7.6, and 9.0). The samples were prepared by shaking the As(V) andFe–Ce suspension at predefined conditions, after which the suspen-sion was centrifuged. The soluble As in the filtrate was analysed,and the wet paste was used to take the spectra. In addition to theFe–Ce samples, scorodite from the Geological Museum of China(GMC) and synthesized CeAsO4 were also scanned as referenceAs-containing compounds.
Transmission XAS spectra of As K-edge for arsenate adsorbedFe–Ce and references were collected at the 4W1B beamline of theBeijing Synchrotron Radiation Facility (BSRF). The storage ring wasoperated at 2.2 GeV with a beam current of 80 mA. A Si(1 1 1) doublecrystal monochromator was used to provide an energy resolution of1.5 eV. To suppress the unwanted harmonics, the monochromaticcrystal faces were detuned, reducing the incident beam by 30%.
EXAFS data reduction and analysis were performed withWinXAS 2.3 [34] using the following procedures: (1) Two scansper sample were aligned and then averaged; (2) First- and second-order polynomial functions were used to fit the pre-edge regionfor background removal and the post-edge region for normal-ization, respectively; (3) The spectra were then converted tophotoelectron wave vector (k) space with respect to E0 determinedfrom the second derivative of the raw spectra; (4) �(k) spectrafor Fe–Ce after arsenate adsorption and the reference samples(CeAsO4 and FeAsO4·2H2O) were extracted using a cubic splinefunction consisting of ≤7 knots over the range k = 3.5–14.7 A−1.Fourier transformation (FT) of the raw k3�(k) function was con-ducted over a consistent region in K space (3.5–14.7 A−1) to obtainthe radial structure function (RSF) using a Bessel window func-tion and a smoothing parameter (ˇ) of 3 to minimize the effectsfrom truncation in the RSFs; (5) The experimental spectra werefitted with single-scattering theoretical phase-shift and ampli-tude functions calculated with the ab initio computer code FEFF7 [35] using atomic clusters generated from the crystal struc-ture of gasparite–(Ce) (CeAsO4) for synthesized CeAsO4, and usingscorodite (FeAsO4·2H2O) for As(V) adsorbed Fe–Ce and referenceAs compounds, respectively. The final nonlinear least squares fitwas conducted on the raw k3 weighted �(k) function. The many-body amplitude reduction factor (S2
0) was fixed at 0.9. For the AsK-edge, each spectrum was first fit roughly to estimate �E0, whichis the difference in threshold energy between theory and the exper-iment, by fixing coordination numbers (CN) and the Debye-Wallerparameter (o’2) as the same values as those of the related referencemodel oxide (e.g. butlerite, scorodite), after which �E0 was fixedaccording to the best fit. Finally, the spectrum was fitted using esti-mated values for CN, R, and o’2 as starting values. The fitting resultswere then evaluated based on the residual value, with a good fitbeing considered as having a residual value of less than 20. Errorestimates of the fitted parameters were CN, ±20%; R, ±0.02 A; ando’2, ±20–30%.
2.2.3. Surface complexation modelingThe CD-MUSIC model with the triple plane option was used to
describe As(V) adsorptive behavior on Fe–Ce. The basic principles ofthe model have been well documented elsewhere [36–38]. Modelcalculation was conducted using the FITEQL 4.0 program [39], with
X. Dou et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 379 (2011) 109–115 111
Table 1Structural parameters for As(V) adsorbed Fe–Ce samples prepared at pH values of 5.0, 7.6 and 9.0 and two As(V) surface loadings of 70 mg As(V)/g Fe–Ce and130 mg As(V)/g Fe–Ce at each pH value.
Samples As-O shell CN R o’ As–Fe shell CN R o’ �E0
odified executable files from Gustafsson [40]. A description ofhe CD-MUSIC formulation for phosphate adsorption on goethitesing a modified version of the FITEQL program was presented byadanier and Eick [41], and for molybdate and tungstate adsorptionn ferrihydrite by Gustafsson [40]. For Fe–Ce, since the Fe–OH siteas probably the main active site for As(V), three types of active
urface oxygens on Fe–Ce may be distinguished as singly (SOH1/2)−,oubly (S2OH0) and triply (S3O1/2−) coordinated. Based on Paul-
ng’s rules and bond valence, S2OH0 was deduced as inert becauset cannot become either protonated or deprotonated in a normal pHange. Thus, only the singly and triply coordinated surface groupsere considered in surface complexation reactions (Table 1). The
riply coordinated surface group could not form inner sphere com-lexes [25]; therefore, only its protonation reactions and ion-paireactions were listed (Eqs. (5) and (8)–(9) in Table 2). The proton
1/2− 1/2−
ffinity constants of SOH and S3O were set to be the samend equal to the pHpzc value of Fe–Ce for symmetrical adsorp-ion based on previous reports [11,36]. The site densities of theOH1/2− and S3O1/2− groups were optimized as 18.01 sites/nm2
nd 5.2 sites/nm2, respectively, by fitting the surface titration data
able 2urface reactions and related parameters used in the CD-MUSIC model for As(V) adsorpti
of Fe–Ce (Fig. S1); further details of the surface titration and datareduction are provided in the supplementary materials. In addition,the outer layer capacitance C2 was set as 5.0 F/m2, which was thesame as that used for goethite [36]. The electrolyte affinity con-stants, KClO4− and KNa+ , and the inner layer capacitance, C1, weredetermined by fitting the surface titration data for Fe–Ce (Fig. S1and Table 2).
An important aspect of the CD model is that adsorbed ionsshould not be treated as point charges on the scale of the com-pact portion of the interface. Inner sphere surface complexes ofoxyanions as well as cations have a spatial distribution of chargethat can be attributed to two different electrostatic planes. A cer-tain fraction f (CD factor) of the charge of the central ions in thecomplex will be attributed to the surface plane (0-plane), while theremaining part (1 − f) is attributed to other ligands in the complex
that are located in the 1-plane. For the species listed in Table 2,when the charge of central As ions was assumed to be symmet-rically neutralized by all surrounding ligands in accordance withPauling’s rules, values of f = 0.25 and f = 0.50 were gained for mon-odentate and bidentate species, respectively. It should be noted that
112 X. Dou et al. / Colloids and Surfaces A: Physico
F0
nbawtfvcmtsrCbaaI(awaarTd
3
3
atfAtrcuattaaa
ig. 1. Zeta potential of Fe–Ce suspensions reacted with 0, 0.1 and 1.0 mM As(V) in.01 M NaClO4 solution.
onsymmetrical neutralization of the central ion charge is possi-le, for instance, in response to a shift in electron density in thedsorbed species caused by protonation of an oxygen atom, whichould lead to a change in the f value. These findings indicate that
he use of the Pauling concept to estimate the CD value of sur-ace species may be insufficiently accurate [36,42]. Recently, CDalues derived from geometries optimized with molecular orbitalalculations using density functional theory (MO/DFT) were used inodeling As(III)/As(V) [25], carbonate [43] and silicate [42] adsorp-
ion on goethite. Because the Fe–Ce was amorphous, the exacttructural molecular cluster could not be extracted from model fer-ic oxides; therefore, the Pauling concept was used to calculate theD value in this study. For the modified FITEQL program developedy Gustafsson [40], the f value was restricted to being only manu-lly adjustable. The f values for species SOAsO3
2.5−, SOAsO3H1.5−
nd SOAsO(OH)20.5− were set as 0.25 following Antelo et al. [44].
n addition, preselected parameter sets of f values of 0.5 and 0.5as in the phosphate complex on goethite [41]), 0.5 and 0.6 [36],nd 0.35 and 0.65 [44] for species S2O2AsO2
2− and S2O2AsO2H−
ere attempted in the fitting process. The fitting was initiated byssuming that only monobentate and bidentate species existed,fter which a trial and error procedure was pursued. If the fittingesults were poor, combinations of two or more species were tried.he procedures were repeated until a good fit to the experimentalata was achieved.
. Results and discussion
.1. Changes in Zeta potential before and after As loading
The zeta potential results of the Fe–Ce adsorbent in the absencend presence of As(V) under different concentrations as a func-ion of pH are shown in Fig. 1. pHpzc of Fe–Ce occurred at pH 5.8or Fe–Ce in 0.01 M NaClO4. In the presence of 0.1 mM and 1 mMs(V), pHpzc shifted to low pH values (about 4.8 and 3.6, respec-
ively), and the zeta potential became more negative in the pHange of 3–10. These results indicated that the higher the aniononcentration, the lower the zeta potential and corresponding val-es of pHpzc, which was similar to results obtained for As(V) and Pdsorption on ferric oxides [44,45]. Shifts in the pHpzc and reversal
rends of the zeta potential values with increasing ion concen-rations were used as evidence of strong specific ion adsorptionnd inner-sphere surface complex formation [45]. Therefore, it wasssumed that inner-sphere complexes formed on Fe–Ce after As(V)dsorption.
chem. Eng. Aspects 379 (2011) 109–115
3.2. Arsenic K-edge EXAFS analysis
EXAFS spectra were employed to further determine the localcoordination environments of the As adsorbed complexes. Fig. 2(a)and (b) showed the k3 weighted As K-edge EXAFS spectra andthe RSFs as Fourier transforms (FT) versus radial distance for theAs adsorbed Fe–Ce and reference As compounds (CeAsO4 andFeAsO4·2H2O). The resolved structural parameters obtained byfitting the theoretical paths to the experimental spectra for Asadsorption on Fe–Ce are shown in Table 1. FT of the �(k) functionisolates the contributions of different coordination shells, in whichthe peak position corresponds to the interatomic distances. Thesepeak positions in Fig. 2(b) are uncorrected for the phase shift, sothey deviate from the true distance by 0.3–0.5 A.
Under different As(V) loads (70 and 135 mg/g) and pH values(5.0, 7.6, and 9.0), the RSF spectra of As(V) adsorbed Fe–Ce yieldedpatterns very similar to those of the first and second shells cen-tered at about 1.40 and 2.99 A (Fig. 2(b)), respectively. As shownin Fig. 2(b) and Table 1, the first peak in the RSF was the resultof backscattering from the nearest neighbor As–O shell for Fe–Ceafter As(V) adsorption. The average As–O distance was about 1.68 A,which is in good agreement with the results of previous stud-ies [14,17,46]. The average CN of O was calculated to be 4.8. Theposition of the second shell of Fe–Ce was consistent with that ofscorodite, and relatively lower than that of CeAsO4 (Fig. 2(b)). Inresolving the structural parameters of the surface complexes, thetheoretical paths of As–Fe, As-Ce or a combination of As–Fe and As-Ce were attempted when fitting the raw k3 weighted �(k) functionin the data reduction process. The fits were not successful when As-Ce or a combination of As-Ce and As–Fe were used. Consequently,As–Fe was finally adopted and the best fit results are shown inTable 1. Based on these findings, it was inferred that the second shellwas contributed by As–Fe and that As(V) might primarily react withFe active sites. Combined with the results of XPS [31], these find-ings suggested that Ce sites were not the main active sites involvedin the reaction with As(V) for Fe–Ce bimetal oxide. However, theintroduction of Ce into Fe oxide might play an important role inthe modification of the surface characterization of the Fe oxide.The main roles of Ce were to break the magnetite structure of theFe(II)/Fe(III) system through oxidation of Fe2+, and to activate theFe atoms to acquire more abundant Fe-OH on the surface of Fe–Ce[31].
The As atom was coordinated by 1.8–2.25 Fe atoms withRAs–Fe = 3.30–3.32 A and 0.8–1.25 Fe atoms at RAs–Fe = 3.58–3.65 A,respectively (Table 1). The As–Fe distance of 3.30–3.32 A is in goodagreement with the distance of 3.31 A for As(V) on goethite, lep-idocrocite, maghematite, hematite, and ferrihydrite lepidocrocite[13,15,18], 3.30 A for As(V) on goethite [17], 3.37 A for As(V) on arg-onaut mine [14], and 3.35 A for As(V) on magnetite and ferrihydrite[12,16]. It has been reported that the distance RAs–Fe = 3.20–3.32 Aindicated the existence of bidentate binuclear complex on adsorbedferric oxides [10,13,19,21]. In the present study, the distanceRAs–Fe = 3.58–3.65 A indicated that the monodentate mononuclearcomplex was also formed on the surface of Fe–Ce, which is con-sistent with the distance of about 3.6 A for As(V) on ferrihydrite[10,21], goethite [19], iron incorporated mine tailings [14] and cor-rosion products of zero valent iron [18].
The above direct in-situ EXAFS results further supported ourpreviously speculation that bidentate binuclear (C2� point group)and monodentate mononuclear complexes (C1 point group) existedon Fe–Ce under a high surface load (� = 22.2 �mol/m2) at pH 5.0
[31]. This speculation was based on the ex-situ techniques of tripletsplitting of As–O �3 (As–O FTIR peak) vibration via Gaussian profilefitting and XPS [31].
Most previous EXAFS studies have suggested that the domi-nant species of As existing on iron oxides as well as oxyhydroxides
X. Dou et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 379 (2011) 109–115 113
a for A
wotwpoiofa0pMpi0d[
e(fecFmawiAfpm
(� = 5.11–14.4 �mol/m2) than on naturally or synthesized environ-mental samples (� = 0.04–4.2 �mol/m2) might also be responsiblefor the differences in these complexes [16]. When comparedwith other complex forms (e.g. 2C and 2E), a linear, monodentate
Fig. 2. (a) Raw (solid line) and fitted (dotted line) k3 weighted �(k) spectr
ere bidentate binuclear complexes (2C) in a surface loading rangef 0.04–4.2 �mol/m2 [16–18,25,47]. It has also been suggestedhat the existence of monodentate mononuclear (1V) complexas favorable under low surface loading conditions. For exam-le, increased amounts of 1V complex were observed for As(V)n goethite, lepidocrocite and maghemite under low surface load-ng conditions (∼0.08 �mol/m2) [18], and also increased amountsf 1V were reported on goethite as surface loading decreasedrom 2.0 to 1.2 �mol/m2 [19]. However, an opposite trend waslso reported in which increasing surface loading of As(V) from.51 to 1.70 �mol/m2 was associated with increased amounts ofrotonated monodentate complex based on EXAFS results and CD-USIC modeling [25]. Most importantly, a recent rethinking report
ointed out that As(V) coordinates at the water-goethite interfacen a predominately monodentate fashion under surface loadings of.9–1.8 �mol/m2 and pH values of 3.3–8.3, and there was no evi-ence of bridging-bidentate coordination under these conditions23].
In the present study, the 1V complex was considered to co-xist with 2C under extremely high surface loading conditions� = 5.11–14.4 �mol/m2). The existence of Ce atoms on the sur-ace of bimetal oxides was probably one of the reasons for thexistence of monodentate. The pHpzc of ferric (hydro)oxides wereommonly observed to be near 9.0 [48]; however, the pHpzc ofe–Ce biometal oxide decreased to 5.8 [31]. In our previous study,ore amorphous structure, higher contents of hydroxyl groups
nd lower pHpzc for Fe–Ce oxide were observed in comparisonith the ferric oxide without Ce doping [31]. Therefore, the dop-
ng of Ce might influence the structure of the iron (hydro)oxides.ccordingly, when compared with iron (hydro)oxides, the sur-
ace complexes on Fe–Ce differed because of the chemical andhysical modifications of the iron oxides. On the other hand, theuch higher surface loading of As(V) on the composite adsorbent
circles), and 106.7 �M (� = 6.86–14.4 �mol/m2*, triangles). Fe–Ce addition was0.03 g/L. Solid lines show the fitting assuming two surface species, MH and B (optionI); dashed lines show the fitting assuming three surface species, MH, B and BH(option II). (*) Note, the values of surface loading were calculated from the adsorbedamounts of As(V) on Fe–Ce in the investigated pH range. The adsorbent dose was0.03 g/L in this study and the BET surface area was 90 m2/g.
114 X. Dou et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 379 (2011) 109–115
F or thre( , in 0.0
ahto
icv
3
naBrrucI(swst2agato
aapofMvoiwha
dation of China (No. 50508006, 50921064), the National High
ig. 4. The surface speciation of As(V) calculated using option I. The symbols stand f� = 5.11–12.0 �mol/m2, circles), and 106.7 �M (� = 6.86–14.4 �mol/m2, triangles)
rrangement was thought to save surface space and minimize stericindrance when there was too much arsenate molecular binding onhe Fe–Ce surface to reach the extremely high surface loading levelf 5.11–14.4 �mol/m2.
As discussed above, the EXAFS results showed that As(V) primar-ly reacted with Fe active sites and that the 2C and 1V complexeso-existed on Fe–Ce under the two surface loadings at the three pHalues investigated in this study.
.3. As(V) adsorption edge modeling
To describe the adsorption behavior and determine the proto-ation state of the adsorbed species of As(V) on Fe–Ce, the As(V)dsorption edge was simulated using the CD-MUSIC model (Fig. 3).ased on the species obtained from EXAFS, the possible surfaceeactions were constructed and further resolved (Table 2). Theesults showed that the adsorption edge could be well modeledsing two options (Table 2): protonated monodentate mononu-lear (MH, Fig. S2(a)) and bidentate binuclear (B, Fig. S2(b)) (option) or MH, B, and protonated bidentate binuclear (BH, Fig. S2(C))option II). The findings also demonstrated that the two surfacepecies, B and MH (option I), described the adsorption edge wellith a WSOS/DF value of 2.72 ± 0.2. WSOS/DF means the weighted
um of squares divided by degrees of freedom and is an indica-or of the goodness of fit (a value falling in the range of 0.1 to0 indicates a reasonably good fit) [49]. When combined withnother species of BH in option II, the quality of the fit remainedood (WSOS/DF = 2.52 ± 0.2), but no significant improvement waschieved. Accordingly, both options I and II were equally good,herefore, we were not able to distinguish between the two modelptions.
Option I was preferred because the minimum number ofdjustable parameters were used in the model simulation. Thebundance of the adsorbed surface species vs. pH based on thearameter sets of option I are shown in Fig. 4, while those basedn option II are shown in supplementary Fig. S3. Similar findingsor option I (Fig. 4) and II (Fig. S3) were observed in that both
H and B were the predominant species. MH was abundant at pHalues below 8, while B was abundant at pH values above 8. More-
ver, MH showed an increased trend with surface loading of Asncreasing from 5.11 to 14.4 �mol/m2, but for B, an inverse trend
as observed (Fig. 4 and Fig. S3). These findings indicated that theigher the surface loading of As(V), the more MH became dominantt neutral and acid pH (<8), which contradicted the previous conclu-
e initial As(V) concentrations, 34.6 �M (� = 5.54–10.5 �mol/m2, squares), 54.2 �M5 M NaClO4 solutions. The dose of Fe–Ce was 0.03 g/L.
sions that monodentate species were only present under relativelylow As surface loading conditions[19,21]. It has been reported thattwo similar options for surface complexes, MH and B (option I) orMH, B and BH (option II) were used to model As(V) on goethite(� = 0.5–1.7 �mol/m2), and that option II was preferred. These find-ings were similar to the finding that MH was present at lower pHvalues and shown to increase as As(V) surface loading increasedfrom 0.5 to 1.7 �mol/m2 [25]. The model results presented here fur-ther supported the results of EXAFS, which showed that two typesof surface complexes (monodentate mononuclear and bidentatebinuclear complex) coexisted on adsorbed Fe–Ce.
4. Conclusions
In this study, the adsorption mechanism of As(V) on a Fe–Cebimetal oxide adsorbent was investigated by using a combina-tion of EXAFS observation and the CD-MUSIC model. The resolvedresults from the As K-edge EXAFS spectra of As(V) adsorbed Fe–Ceunder three pH levels and two initial surface loadings showedthat As(V) primarily reacted with Fe-OH sites and monoden-tate mononuclear and bidentate binuclear As surface complexescoexisted. The existence of monodentate complex might be dueto the existence of Ce atoms in the bimetal oxide and thehigh surface loading observed in this investigation. Further, theAs(V) adsorption edge was well simulated using the CD-MUSICmodel with the TPM option. These results showed that proto-nated monodentate complex (MH) and deprotonated bidentatecomplex (B) preferred to co-exist at the Fe–Ce surface under asignificantly high surface loading range (� = 5.11–14.4 �mol/m2).MH was found to be dominant at pH values below 8, while Bwas dominant at pH values above 8. As As(V) surface loadingincreased, MH increased. Based on the above results, the adsorp-tive behavior of As(V) on Fe–Ce was satisfactorily interpreted andunderstood.
Acknowledgments
This work was supported by the National Natural Science Foun-
Technology Research and Development Program of China (No.2007AA06Z319), the Beijing Nova Program (No.2008A33) and theFundamental Research Funds for the Central Universities (YX2010-33). The authors are also thankful to Dr. K. Tanaka, honoraryprofessor of Tokyo University, Japan, for in-depth discussions.
[update, J. Colloid Interface Sci. 275 (2004) 214–224.
X. Dou et al. / Colloids and Surfaces A: Ph
ppendix A. Supplementary data
Supplementary data associated with this article can be found, inhe online version, at doi:10.1016/j.colsurfa.2010.11.043.
eferences
[1] B.K. Mandal, K.T. Suzuki, Arsenic round the world: a review, Talanta 58 (2002)201–235.
[2] D. Mohan, J.C.U. Pittman, Arsenic removal from water/wastewater using adsor-bents:a critical review, J. Hazard. Mater. 142 (2007) 1–53.
[3] M. Pena, X. Meng, G.P. Korfiatis, C. Jing, Adsorption mechanism of arsenicon nanocrystalline titanium dioxide, Environ. Sci. Technol. 40 (2006)1257–1262.
[4] S.R. Randall, D.M. Sherman, K.V. Ragnarsdottir, Sorption of As(V) on greenrust (Fe4(II)Fe2(III)(OH)12SO4·3H2O) and lepidocrocite (–FeOOH): surface com-plexes from EXAFS spectroscopy, Geochim. Cosmochim. Acta 65 (2001)1015–1023.
[5] Y. Kim, C. Kim, I. Choi, S. Rengaraj, J. Yi, Arsenic Removal using mesoporousalumina prepared via a templating method, Environ. Sci. Technol. 38 (2003)924–931.
[7] M. Ding, B.H.W.S. de Jong, S.J. Roosendaal, A. Vredenberg, XPS studies on theelectronic structure of bonding between solid and solutes: adsorption of arse-nate, chromate, phosphate, Pb2+, and Zn2+ ions on amorphous black ferricoxyhydroxide, Geochim. Cosmochim. Acta 64 (2000) 1209–1219.
[8] I. Carabante, M. Grahn, A. Holmgren, J. Kumpiene, J. Hedlund, Adsorption ofAs(V) on iron oxide nanoparticle films studied by in situ ATR-FTIR spectroscopy,Colloids Surf. A 346 (2009) 106–113.
[9] K. Amstaetter, T. Borch, P.L. Casanova, A. Kappler, Redox transformation ofarsenic by Fe(II)-activated goethite (�-FeOOH), Environ. Sci. Technol. 44 (2009)102–108.
10] G.A. Waychunas, J.A. Davis, C.C. Fuller, Geometry of sorbed arsenate on ferri-hydrite and crystalline FeOOH: re-evaluation of EXAFS results and topologicalfactors in predicting sorbate geometry and evidence for monodentate com-plexes, Geochim. Cosmochim. Acta 59 (1995) 3655–3661.
11] C. Jing, X. Meng, S. Liu, S. Baidas, R. Patraju, C. Christodoulatos, G.P. Korfiatis,Surface complexation of organic arsenic on nanocrystalline titanium oxide, J.Colloid Interface Sci. 290 (2005) 14–21.
12] G. Ona-Nguema, G. Morin, F. Juillot, G. Calas, G.E. Brown Jr., EXAFS analy-sis of arsenite adsorption onto two-line ferrihydrite, hematite, goethite, andlepidocrocite, Environ. Sci. Technol. 39 (2005) 9147–9155.
13] D.M. Sherman, S.R. Randall, Surface complexation of arsenic(V) to iron(III)(hydr)oxides: structural mechanism from ab initio molecular geome-tries and EXAFS spectroscopy, Geochim. Cosmochim. Acta 67 (2003)4223–4230.
15] M.L. Farquhar, J.M. Charnock, F.R. Livens, D.J. Vaughan, Mechanisms of arsenicuptake from aqueous solution by interaction with goethite, lepidocrocite,mackinawite, and pyrite: an X-ray absorption spectroscopy study, Environ. Sci.Technol. 36 (2002) 1757–1762.
16] G. Morin, G. Ona-Nguema, Y. Wang, N. Menguy, F. Juillot, O. Proux, F. Guyot, G.Calas, G.E. Brown Jr., Extended X-ray absorption fine structure analysis of arsen-ite and arsenate adsorption on maghemite, Environ. Sci. Technol. 42 (2008)2361–2366.
17] S.E. O’Reilly, D.G. Strawn, D.L. Sparks, Residence time effects on arsenateadsorption/desorption mechanisms on goethite, Soil Sci. Soc. Am. J. 65 (2001)67–77.
18] B.A. Manning, M.L. Hunt, C. Amrhein, J.A. Yarmoff, Arsenic(III) and arsenic(V)reactions with zerovalent iron corrosion products, Environ. Sci. Technol. 36(2002) 5455–5461.
19] S. Fendorf, M.J. Eick, P. Grossl, D.L. Sparks, Arsenate and chromate retentionmechanisms on goethite. 1. surface structure, Environ. Sci. Technol. 31 (1997)315–320.
20] A.C.Q. Ladeira, V.S.T. Ciminelli, H.A. Duarte, M.C.M. Alves, A.Y. Ramos, Mech-anism of anion retention from EXAFS and density functional calculations:arsenic (V) adsorbed on gibbsite, Geochim. Cosmochim. Acta 65 (2001)1211–1217.
21] G.A. Waychunas, B.A. Rea, C.C. Fuller, J.A. Davis, Surface chemistry of ferrihy-drite: part 1. EXAFS studies of the geometry of coprecipitated and adsorbedarsenate, Geochim. Cosmochim. Acta 57 (1993) 2251–2269.
22] M. Kanematsu, T.M. Young, K. Fukushi, P.G. Green, J.L. Darby, Extended triplelayer modeling of arsenate and phosphate adsorption on a goethite-based gran-ular porous adsorbent, Environ. Sci. Technol. 44 (2010) 3388–3394.
[
[
hem. Eng. Aspects 379 (2011) 109–115 115
23] J.S. Loring, M.H. Sandström, K. Norén, P. Persson, Rethinking arsenate coordi-nation at the surface of goethite, Chem. Eur. J. 15 (2009) 5063–5072.
24] M.S. Schaller, C.M. Koretsky, T.J. Lund, C.J. Landry, Surface complexation mod-eling of Cd(II) adsorption on mixtures of hydrous ferric oxide, quartz andkaolinite, J. Colloid Interface Sci. 339 (2009) 302–309.
25] M. Stachowicz, T. Hiemstra, W.H. van Riemsdijk, Surface speciation of As(III)and As(V) in relation to charge distribution, J. Colloid Interface Sci. 302 (2006)62–75.
26] C.Y. Jing, X. Meng, E. Calvache, G. Jiang, Remediation of organic and inorganicarsenic contaminated groundwater using a nanocrystalline TiO2-based adsor-bent, Environ. Pollut. 157 (2009) 2514–2519.
27] V. Lenoble, C. Laclautre, B. Serpaud, V. Deluchat, J.C. Bollinger, As(V) retentionand As(III) simultaneous oxidation and removal on a MnO2-loaded polystyreneresin, Sci. Total Environ. 326 (2004) 197–207.
28] Y.-M. Zheng, S.-F. Lim, J.P. Chen, Preparation and characterization of zirconium-based magnetic sorbent for arsenate removal, J. Colloid Interface Sci. 338 (2009)22–29.
29] G. Zhang, J. Qu, H. Liu, R. Liu, R. Wu, Preparation and evaluation of a novelFe–Mn binary oxide adsorbent for effective arsenite removal, Water Res. 41(2007) 1921–1928.
30] Z. Li, S. Deng, G. Yu, J. Huang, V.C. Lim, As(V) and As(III) removal from waterby a Ce-Ti oxide adsorbent: behavior and mechanism, Chem. Eng. J. 161 (2010)106–113.
31] Y. Zhang, M. Yang, X.M. Dou, H. He, D. Wang, Arsenate adsorption on an Fe–Cebimetal oxide adsorbent: role of surface properties, Environ. Sci. Technol. 39(2005) 7246–7253.
32] X.M. Dou, Y. Zhang, M. Yang, Y. Pei, X. Huang, T. Takayama, S. Kato, Occurrenceof arsenic in groundwater in the suburbs of Beijing and its removal using aniron–cerium bimetal oxide adsorbent, Water Qual. Res. J. Canada 41 (2006)140–146.
33] Y. Zhang, M. Yang, Y.X. Gao, F. Wang, X. Huang, Preparation and adsorptionmechanism of rare earth-doped adsorbent for arsenic(V) removal from ground-water, Sci. Chin. Ser. B. 46 (2003) 252–258.
34] T.J. Ressler, WinXAS: a program for X-ray absorption spectroscopy data analysisunder MS-Windows, J. Synchrotron Radiat. 5 (1998) 118–122.
35] A.L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Real space multiple-scattering calculation and interpretation of X-ray absorption near-edgestructure, Phys. Rev. B. 58 (1998) 7565–7576.
36] T. Hiemstra, W.H. Van Riemsdijk, A Surface structural approach to ion adsorp-tion: the charge distribution (CD) model, J. Colloid Interface Sci. 179 (1996)488–508.
37] T. Hiemstra, P. Venema, W.H.V. Riemsdijk, Intrinsic proton affinity of reactivesurface groups of metal (hydr)oxides: the bond valence principle, J. ColloidInterface Sci. 184 (1996) 680–692.
38] T. Hiemstra, W.H. Van Riemsdijk, On the relationship between charge distribu-tion, surface hydration, and the structure of the interface of metal hydroxides,J. Colloid Interface Sci. 301 (2006) 1–18.
39] A.L. Herbelin, J.C. Westall, FITEQL – A computer Program For Determination OfChemical Equilibrium Constants From Experimental Data, In Dept of chemistryRep. 99-01, Oregon State University, Corvallis, OR, 1999.
40] J.P. Gustafsson, Implementation of CD-MUSIC in FITEQL 4.0, 2003.http://www.lwr.kth.se/forskningsprojekt/MoW/fiteql.htm.
41] C.J. Tadanier, M.J. Eick, Formulating the charge-distribution multisite surfacecomplexation model using FITEQL, Soil Sci. Soc. Am. J. 66 (2002) 1505–1517.
42] T. Hiemstra, M.O. Barnett, W.H. van Riemsdijk, Interaction of silicic acid withgoethite, J. Colloid Interface Sci. 310 (2007) 8–17.
43] R. Rahnemaie, T. Hiemstra, W.H. van Riemsdijk, Carbonate adsorption ongoethite in competition with phosphate, J. Colloid Interface Sci. 315 (2007)415–425.
44] J. Antelo, M. Avena, S. Fiol, R. López, F. Arce, Effects of pH and ionic strengthon the adsorption of phosphate and arsenate at the goethite-water interface, J.Colloid Interface Sci. 285 (2005) 476–486.
45] S. Goldberg, C.T. Johnston, Mechanisms of arsenic adsorption on amorphousoxides evaluated using macroscopic measurements, vibrational spectroscopy,and surface complexation modeling, J. Colloid Interface Sci. 234 (2001)204–216.
46] D.M. Sherman, Electronic structures of iron(III) and manganese(IV) (hydr)oxideminerals: thermodynamics of photochemical reductive dissolution in aquaticenvironments, Geochim. Cosmochim. Acta 69 (2005) 3249–3255.
47] Y. Wang, G. Morin, G. Ona-Nguema, F. Juillot, F. Guyot, G. Calas, G.E. Brown,Evidence for different surface speciation of arsenite and arsenate on green rust:an EXAFS and XANES study, Environ. Sci. Technol. 44 (2009) 109–115.
48] M. Kosmulski, pH-dependent surface charging and points of zero charge II.
49] K.F. Hayes, G. Redden, W. Ela, J.O. Leckie, Surface complexation models: an eval-uation of model parameter estimation using FITEQL and oxide mineral titrationdata, J. Colloid Interface Sci. 142 (1991) 448–469.
50] D.A. Dzombak, F.M.M. Morel, Surface Complexation Modeling: Hydrous FerricOxide, Wiley-Interscience, New York, 1990.
