Individual and Competitive Adsorption of Arsenate and Phosphate To

The School of Engineering and Applied Science at Washington Univeristy in St. Louis provided H.Z. with a first-year fellowship. H.Z. was also supporte...
2 downloads 0 Views 154KB Size
Download PDF
Environ. Sci. Technol. 2008, 42, 147–152

Individual and Competitive Adsorption of Arsenate and Phosphate To a High-Surface-Area Iron Oxide-Based Sorbent HUI ZENG,† BRIAN FISHER,‡ AND D A N I E L E . G I A M M A R * ,† Department of Energy, Environmental and Chemical Engineering and Center for Materials Innovation, Washington University in St. Louis, One Brookings Drive, St. Louis, Missouri 63130, and Department of Civil Engineering and Geological Sciences, University of Notre Dame, Notre Dame, Indiana 46556

Received June 25, 2007. Revised manuscript received October 16, 2007. Accepted October 26, 2007.

Individual and competitive adsorption of arsenate and phosphate were studied on a high-surface-area Fe/Mn(hydr)oxide sorbent with surface and bulk properties similar to those of two-line ferrihydrite. It has maximum adsorption densities of 0.42 µmol As m-2 at neutral pH and 1.24 µmol As m-2 at pH 3. A surface complexation model (SCM) that used the diffuse double layer model was developed that could simulate single and binary sorbate adsorption over pH 4–9. The predominant adsorbed arsenate and phosphate species were modeled as bidentate binuclear surface complexes at low pH and as monodentate complexes at high pH. The model initially overpredicted the inhibition of arsenate adsorption by the presence of phosphate. The overprediction was resolved by separating surface sites into two types: ones to which both arsenate and phosphate bind and a smaller number to which only phosphate binds. The modified model predicted the competitive adsorption of arsenate and phosphate over pH 4–9 at total As concentrations of 6.67 and 80.1 µM and a total P concentration of 129 and 323 µM. The model may be used to predict arsenic adsorption to the sorbent for a given water source based on solution chemistry.

Introduction Arsenic is a toxic element that occurs naturally in soils, rocks, and groundwater. Arsenic is present in the form of arsenite (As(III)) and arsenate (As(V)). As(V) dominates at oxidizing conditions and As(III) dominates in reducing environments. High arsenic concentrations in groundwater have been reported in countries around the world (1). Human exposure to arsenic through drinking water consumption can cause serious long-term health problems. In the United States, the maximum contaminant level (MCL) for arsenic in drinking water was recently lowered from 50 to 10 ppb. New technologies for removing arsenic can help communities comply with this standard. Sorbent technologies, which are often suitable for small systems, can have high arsenic removal efficiencies and * Corresponding author phone: 314-935-6849; fax: 314-935-7211; e-mail: [email protected]. † Washington University in St. Louis. ‡ University of Notre Dame. 10.1021/es071553d CCC: $40.75

Published on Web 11/22/2007

 2008 American Chemical Society

relatively small footprints. Iron (hydr)oxides are promising sorbent materials because they have strong chemical affinities for As and large specific surface areas. Previous research has observed strong adsorption of As to pure phases of amorphous iron hydroxide (2), goethite (2–4), and magnetite (2). Sorbents that incorporate iron oxides, including activatedcarbon-based iron oxide-containing sorbents (5, 6), ironoxide-coated sand (7), and nanostructured hydrous ironoxide-coated polymeric beads (8), also have high arsenic adsorption capacities. Unlike most pure forms of iron oxides and oxyhydroxides, these sorbents are composed of particles that can be packed in columns to provide porosity and permeability values that are useful in arsenic removal treatment systems. Equilibrium adsorption can be described by isotherm equations (e.g., Langmuir or Freundlich) and surface complexation models (SCM). SCMs have the advantages of explicitly considering pH effects, electrostatic contributions to adsorption, and competition with other adsorbates. SCMs for arsenic adsorption to iron oxides and oxyhydroxides have been developed in past research (2, 3, 9). Despite the previous work, SCMs have not been significantly applied to predict the performance of specific iron-oxide-based sorbents proposed for use in water treatment. Adsorption of arsenate can be affected by competition from phosphate, which is chemically similar to arsenate and strongly adsorbs to iron oxyhydroxides. Phosphate species are often present in groundwater at concentrations comparable to or greater than those of arsenic. A few studies have modeled the competitive adsorption of arsenate and phosphate to iron oxyhydroxide minerals. Manning and Goldberg (10) modeled the competitive adsorption of arsenate and phosphate to goethite using the constant capacitance model (CCM) with six complexation reactions that included both monodentate and bidentate surface complexes for arsenate and phosphate. Gao and Mucci (3) modeled the competitive adsorption to goethite in artificial seawater using the CCM with six complexation reactions that included only monodentate surface complexes. They found that arsenate adsorption was reproduced well, but that phosphate adsorption was underpredicted at pH < 6.5. Hiemstra and van Riemsdijk (9) modeled competitive adsorption to goethite using the charge-distributed multisite complexation model (CD-MUSIC) with six surface complexation reactions that included both monodentate and bidentate surface complexes for arsenate and phosphate. To extend the application of surface complexation modeling of arsenate adsorption to iron oxide materials other than goethite and other pure iron oxy(hydroxide) minerals, more information on adsorption reactions on engineered sorbents and the associated equilibrium constants is needed for developing SCMs that can predict arsenate adsorption over a wide range of pH values and phosphate concentrations. The objectives of this study were to quantify arsenate and phosphate adsorption to a novel iron-oxide-based sorbent and to develop a surface complexation model that can account for single adsorbate and competitive adsorption.

Materials and Methods Characterization of Sorbent. A novel ferric-hydroxide-based sorbent was supplied by the Enviroscrub Technologies Corporation (Minneapolis, MN). The sorbent is composed of black particles with an average diameter of 300 µm. It has a specific surface area of 297 m2/g, which indicates that the particles possess very large amounts of surface area in internal pores. The powder X-ray diffraction (XRD) pattern shows a VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

147

few broad peaks (Figure S1 in the Supporting Information), which are consistent with two-line ferrihydrite. As determined by X-ray fluorescence (XRF), the overall Fe/Mn ratio is 3.8: 1 (mol/mol) and the sorbent contains traces of carbon, potassium and silicon (Table S1 in the Supporting Information). Although the manganese content may decrease the point of zero charge (pHpzc) of iron oxyhydroxides, the measured zeta-potential of this sorbent still indicates a pHpzc of approximately 8 (Figure S2 in the Supporting Information), which is very close to the reported value of pure ferrihydrite (11). In addition, as determined by X-ray photoelectron spectroscopy (XPS), the surface Fe/Mn ratio is 13.7:1 (mol mol-1), which suggests that most of the manganese atoms are incorporated in the bulk of the sorbent and not at the surface (Figure S3 in the Supporting Information). Thus the surface mineralogy of the sorbent is considered as a poorly crystalline two-line ferrihydrite. Adsorption Experiments. Batch adsorption experiments were conducted at room temperature (20 ( 2 °C) over the pH range 3–9. Experiments examined conditions of (i) arsenate-only, (ii) phosphate-only, and (iii) arsenatephosphate competition. In noncompetitive adsorption experiments, total As concentrations were varied from 6.67 to 400 µM and total P concentrations were varied from 64.5 to 774 µM. In competitive adsorption experiments, total As concentrations ranged from 6.67 to 80.1 µM and total P concentrations of either 129 or 323 µM were used. Although these As concentrations are above the 10 ppb standard (0.13 µM) and values encountered in most treatment applications, they were selected to allow investigation of the sorption properties over a broad range of As loadings. A sorbent concentration of 1 g L-1 was chosen for all experiments. Buffer solutions of 1 mM were used to keep the pH at 6 (MES), 7 and 8 (HEPES), and 9 (TAPS) (Sigma, St. Louis, MO). These buffers were selected because of their low affinity toward metals (12), which indicates a low affinity for adsorption to iron-oxide-based sorbents. Because these buffers are organic sulfonic acids that will exist as either the uncharged acid or the anionic conjugate base species, they will not form complexes with the anionic arsenate and phosphate species in the system. No buffer was used for pH 3, 4, and 5. The ionic strength was fixed at 0.01 M using NaNO3. Capped 125-mL HDPE bottles were used as batch reactors. Arsenate and phosphate stock solutions were made from Na2HAsO4 · 7 H2O (98.0-102%, Alfa Aesar, MA) and NaH2PO4 · 7 H2O (Fisher Scientific). All solutions were prepared with ultrapure water (resistivity > 18.2 MΩ-cm). To start each experiment, sorbent, buffer, and ultrapure water were mixed. The suspension was adjusted to the desired pH with 1 M NaOH before arsenate and/or phosphate stock solutions of the same pH were added. The batch reactors were then shaken on a platform shaker (New Brunswick Scientific, NJ) at 50 rpm for 24 h. Preliminary studies using a total arsenic concentration of 13.3 µM showed that adsorption equilibrium was achieved within 30 min. After 24 h, samples were collected, filtered with 0.2 µm polycarbonate membranes (Millipore, MA), and acidified to 1% HNO3 for preservation prior to analysis. The pH of the batch suspension was measured at the end of each experiment. Analytical Methods. The multipoint BET surface area of the sorbent was determined by N2 adsorption (Autosorb1-C, Quantachrome). Elemental composition was determined by X-ray fluorescence (SRS-200, Siemens, Germany). Surface elemental composition was determined by X-ray photoelectron spectroscopy (Quantera SXM, Physical Electronics). X-ray powder diffraction (XRD) patterns were collected using Cu KR radiation (D-MAX/A, Rigaku, Japan). Zeta potential was measured by a nanoparticle characterization instrument with zeta potential capability (Nanoseries ZS, Malvern Instruments, U.K.). 148

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

The pH was measured with a glass electrode and an Accumet pH meter. Dissolved arsenic concentrations were determined by inductively coupled plasma-mass spectrometry (ICP-MS) (Thermo, Germany) or inductively coupled plasma-optical emission spectrometry with a hydride vapor generation accessory (ICP-OES-HVGA) (Varian, Australia). For ICP-OES-HVGA, samples were prepared in 1.0 M HCl and 1% (w/v) KI to reduce As (V) to As (III) so that the gaseous hydride AsH3 would form in the vapor generation accessory upon addition of 0.6% (w/v) NaBH4 and 0.5% (w/v) NaOH. Dissolved phosphate was analyzed colorimetrically (Standard method 4500-P E) (13). Surface Complexation Modeling. The first step of the SCM approach was optimization of the adsorption equilibrium constants of arsenate and phosphate using data from noncompetitive batch adsorption experiments. The ability to model the competitive adsorption of arsenate and phosphate using the parameters and equilibrium constants derived from the noncompetitive experiments was then evaluated. The diffuse double layer model (DLM) was used in this study to account for the relationship between surface charge and surface potential. The software program FITEQL 4.0 (14) was used to determine the adsorption equilibrium constants that provided the optimal simulation of the experimental data. The surface sites (≡FeOH) were considered to be similar to those of two-line ferrihydrite. Surface sites were divided into two types: sites accessible to both arsenate and phosphate (≡FeOH), and sites to which only phosphate could bind (≡FeOHp). The choice of this two-site model, which was necessary to explain the adsorption results, is discussed in Results and Discussion. The acid–base titration of the sorbent was interpreted as the protonation and deprotonation of these surface sites (reactions 1 and 2 in Table 1). It was assumed that the two types of sites have the same surface acidity constants. Surface acidity constants were fixed at values taken from the literature (15). The density of surface sites was determined from the maximum adsorption density of arsenate or phosphate at pH 3. Previous studies, which included direct spectroscopic characterization of adsorbed As with extended X-ray absorption fine structure spectroscopy (EXAFS), reported that arsenate adsorbs to the surface of iron (hydr)oxides primarily as bidentate binuclear surface complexes at both pH 4 and 8 at high As loadings (9, 16, 17). Monodentate complexes were also observed at low As loading at pH 8 (17). The SCM includes monodentate surface complexes as dominant species at high pH and bidentate binuclear surface complexes as dominant species at low pH for adsorbed arsenate and phosphate. In contrast to models with only monodentate surface complexes, the inclusion of bidentate surface complexes requires (1) the selection of an appropriate mass action expression for bidentate adsorption and (2) the identification of the relationship between concentrations of monodentate and bidentate adsorption sites. For arsenate adsorption to a bidentate site (≡(FeOH)2) to form a bidentate binuclear surface complex (≡(FeO)2AsO2-) (reaction 4 in Table 1), the equilibrium constant (K) is given as eq 1 with respect to the activities of the species. K )

{ ≡ (FeO)2AsO2-} { ≡ (FeOH)2}{AsO43-}{H+}2

(1)

In this approach, bidentate sites are considered to be specific combinations of two adjacent monodentate sites; hence, the site is included as {≡(FeOH)2} and not as {≡FeOH}2. The assumption of a specific rather than a random combination of two ≡FeOH sites was used because the sites are fixed at the surface, which is different from the status of free ions in

TABLE 1. Reactions and Parameters Used in Surface Complexation Modeling no. 1 2 3 4 5 6 7 8

reaction

Log K

ref

surface acidity reaction ≡FeOH2+ ) ≡FeOH + H+ ≡FeOH)≡FeO- + H+ adsorption reaction ≡FeOH + AsO43- + H+ ) ≡FeOAsO32- + H2O ≡(FeOH)2 + AsO43- + 2H+ ) ≡(FeO)2AsO2- + 2H2O ≡FeOH + PO43- + H+ ) ≡FeOPO32- + H2O ≡(FeOH)2 + PO43- + 2H+ ) ≡(FeO)2PO2- + 2H2O ≡FeOHp + PO43- + H+ ) ≡FeOpPO32- + H2O ≡(FeOHp)2 + PO43-+2H+) ≡(FeOp)2PO2-+ 2H2O

6.51 -9.13 Log K (Log K*) 16.6 (23.2) 16.9 (23.4) 16.9 (23.4)

(15) (15) (this study)

Surface Site Density (sites per nm2) generally accessible

phosphate only

1.50

0.63

aqueous solution (18, 19). Based on previous work (20), a linear relationship (eq 2) is assumed between the concentrations of bidentate and monodentate sites. { ≡ (FeOH)2} ) k{ ≡ FeOH}

(2)

A modified adsorption equilibrium constant (K*) can be calculated from K and k. K/ ) K × k )

{ ≡ (FeO)2AsO2-} { ≡ FeOH}{AsO43-}{H+}2

(3)

K* is the equilibrium constant determined for binuclear bidentate surface complexes in this model. A binuclear bidentate surface complex (e.g., ≡(FeO)2AsO2-) contains two iron atoms coming from two surface sites of ≡FeOH. Consequently, when accounting for the mole balance of surface sites (eq 4 with [] denoting molar concentrations), each binuclear bidentate complex contains two ≡FeOH groups. TOT ≡ FeOH ) [ ≡ FeOH] + [ ≡ FeOH+ 2] + [ ≡ FeO-] + [ ≡ FeOAsO23 ] + 2[ ≡ (FeO)2AsO2 ] + ( ) [ ≡ FeOPO23 ] + 2[ ≡ FeO 2PO2 ]

(4)

This is in contrast to the mass action expression (eq 3), in which {≡FeOH} is only raised to the first power. Therefore, in an equilibrium model, different coefficients must be used for the surface sites in mass action and mass balance matrices prepared as part of the numerical solution of chemical equilibrium problems. The software application FITEQL 4.0 allows the use of different coefficients in these two matrices. In the model, generally accessible sites (≡FeOH) and phosphate-only sites (≡FeOHp) have similar mass action and mass balance equations.

Results and Discussion Non-Competitive Adsorption. Arsenate. Arsenate adsorption was pH-dependent and increased with dissolved arsenate concentration up to a maximum adsorption density (Figure 1a). Arsenate adsorption decreased with increasing pH from 4 to 9, and adsorption at pH 3 was nearly identical to that at pH 4 (Figure 1b). At neutral pH, the maximum adsorption density was 0.42 µmol As m-2 (9.3 mg g-1 or 0.01 mol As mol Fe-1) (not shown here) and at pH 3 and 4 the maximum density was 1.24 µmol As m-2 (27 mg g-1 or 0.04 mol As mol Fe-1). The maximum density at pH 3 and 4 is less than the 0.24 mol As mol Fe-1 measured by Dixit and Hering (2) for amorphous iron hydroxide, but higher than the 0.016 and 0.025 mol As mol Fe-1 that they measured for goethite and

(this study)

magnetite. For total arsenic of 6.67 and 80.1 µM, a 1 g L-1 sorbent suspension removed nearly 100% of the arsenate over pH 4–7 and significantly less at pH > 7 (Figure 1b). The maximum density of 1.24 µmol As m-2 at pH 3 was calculated as the total generally accessible site density, which is equal to 1.50 sites per nm2 assuming bidentate binuclear surface complexes at low pH. Arsenate is assumed to adsorb as monodentate and bidentate binuclear surface complexes to ≡FeOH sites (reactions 3 and 4 in Table 1) in SCM. Protonated bidentate surface complexes may be present (21) and were considered in initial model optimization; however, their inclusion did not significantly improve the model fit to the experimental data. Therefore, to simplify the model, only nonprotonated bidentate surface complexes were considered. The equilibrium constants (K* and K) for arsenate adsorption that were optimized with SCM are shown in Table 1. The SCM fits the experimental data for percent adsorption within 10% over the range of pH 4–9 and total As concentrations from 6.67 to 160 µM (Figure 1b), except at pH 9 and total As 6.67 µM where the model is only within 25% of the experimental data. At a total As loading of 160 µM, the binuclear bidentate surface complex ≡(FeO)2AsO2- is predicted to be the dominant form of adsorbed arsenate over pH 4–6.8 and the monodentate complex ≡FeOAsO32- is predicted to dominate above pH 6.8 (Figure 1c). At a lower As loading of 6.67 µM, the relative contribution of the monodentate complex ≡FeOAsO32- to the adsorbed arsenate is even higher, and this species becomes dominant at a lower pH. Thus monodentate complexes may be more significant at low As loadings. The formation of phosphate or arsenate surface complexes introduces negative surface charge. At low pH with the pristine surface predominantly positively charged, adsorption of bidentate complexes that occupy two surface sites may be favorable. Further, as adsorption densities increase, the surface can become negatively charged even below the pHpzc of the sorbent and adsorption of anionic species can have a repulsive electrostic component. Bidentate complexes introduce less negative charge than do monodentate complexes, and at higher adsorption densities their introduction of less negative charge can make their formation favorable. Phosphate. Phosphate adsorbed strongly to the sorbent, with adsorption decreasing with increasing pH (Figure 2b). Phosphate adsorption increased with dissolved P concentration up to a relatively stable denisty (Figure 2a). The highest adsorption density at pH 7 was 0.74 µmol P m-2 (6.8 mg g-1) (not shown here). The maximum adsorption density at pH 3 was 1.77 µmol P m-2, which is higher than the value of 1.24 µmol m-2 for arsenate. Phosphate adsorbs as monodentate and bidentate binuclear surface complexes to ≡FeOH sites VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

149

FIGURE 1. (a) Arsenate adsorption isotherms at four pH values; (b) arsenate adsorption edges at three total arsenic loadings; (c) speciation of adsorbed arsenate at total As ) 160 µM. I ) 0.01 M NaNO3. Data are shown as symbols and model simulations are shown as lines. (reactions 5 and 6) and ≡FeOHp sites (reactions 7 and 8) in SCM (Table 1). The concentration of generally accessible sites ≡FeOH was equal to 1.50 sites nm-2. The concentration of phosphate-only sites ≡FeOHp was calculated to be 0.63 sites nm-2 from the difference between the maximum adsorption densities of phosphate and arsenate. It was assumed that the two sites have the same phosphate adsorption equilibrium constants. The optimization of the model for phosphate adsorption resulted in slightly larger adsorption equilibrium constants for phosphate than for arsenate (Table 1). Comparing the experimental data in Figures 1b and 2b, phosphate showed a higher adsorption density than did As at the same total adsorbate loading of 80 µM between pH 7 and 9. Stronger adsorption of phosphate (larger Log K and Log K*) than of arsenate in this pH range is consistent with the results of Hiemstra and van Riemsdijk (9) and of Manning and Goldberg (10) but contrary to those 150

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

FIGURE 2. (a) Phosphate adsorption isotherms at four pH values; (b) phosphate adsorption edges at four total phosphorus loadings; (c) speciation of adsorbed phosphate at total P ) 258 µM. I ) 0.01 M NaNO3. Data are shown as symbols and model simulations are shown as lines. ≡(FeO)2PO2- denotes ≡(FeO)2PO2- + ≡(FeOp)2PO2- and ≡FeOPO32- denotes ≡FeOPO32- + ≡FeOpPO32-. of Gao and Mucci (3) for goethite. The model fit phosphate adsorption within 10% for total P loadings from 80.6 to 387 µM over most pH conditions. However, the model underestimated phosphate adsorption at pH 9 and the lowest P loading (Figure 2b) with a predicted adsorption of only 50% of the actual adsorption. The predicted speciation of adsorbed phosphate is similar to that of adsorbed arsenate (Figure 2c). A binuclear bidentate surface complex ≡(FeO)2PO2- is the dominant form of adsorbed phosphate over pH 4–6.3 and the monodentate complex ≡FeOPO32- is dominant above pH 6.3. Competitive Adsorption. Phosphate inhibition of arsenate adsorption in competition was not as significant as initially expected based on individual adsorption. In arsenate-only experiments (Figure 1b), the percentage of arsenate adsorbed

FIGURE 4. Phosphate adsorption edges with a total P loading of 129 µM in the presence of arsenate. I ) 0.01 M NaNO3. Data are shown as symbols and model simulations are shown as lines.

FIGURE 3. Arsenate adsorption edges with total As loading of (a) 6.67 µM and (b) 80.1 µM in the presence of phosphate. I ) 0.01 M NaNO3. Data are shown as symbols and model simulations are shown as lines. dropped 20–60% in the pH range 6–9 with total As loading increasing from 6.67 to 80.1 µM. Due to the higher adsorption affinity of phosphate, phosphate should take up more surface sites than arsenate at the same total adsorbate loadings so that the percentage of arsenate adsorbed should decrease more when total phosphate is added than when total arsentate is increased by a comparable amount. However, our experimental results showed that in competitive adsorption (Figure 3a) arsenate adsorption only dropped 7% when adding 129 µM total phosphate to a suspension with 6.67 µM arsenate. Roy et al. (22) and Hingston et al. (23) also observed that phosphate did not inhibit arsenate as much as would have been expected based on strict competition for the same sites. Therefore, phosphate inhibition of arsenate adsorption would be significantly overpredicted if adsorption is assumed to occur at only one type of site. In a one-site model, the overprediction of phosphate inhibition of arsenate adsorption could be resolved by lowering the adsorption constants for phosphate, but then the model could not simulate the adsorption of phosphate in single-sorbate experiments. Therefore, another approach of adjusting adsorption capacity was proposed. In this study, the simulation of competitive adsorption of arsenate and phosphate was improved by adding a site type that is accessible to phosphate but not to arsenate. In previous work, Hingston et al. (23) included adsorbate-specific sites in their competitive Langmuir isotherm model to account for the competitive adsorption of phosphate and arsenate to goethite and gibbsite. Roy et al. (22) included competitive coefficients in a modified Freundlich isotherm model for arsenate and phosphate adsorption to soils to account for the greater inhibition of arsenate adsorption by phosphate than of phosphate adsorption by arsenate. In our noncompetitive adsorption experiments, phosphate has a larger

maximum adsorption density at pH 3 than arsenate, which suggests that some sites are only accessible to phosphate. As noted earlier, this extra portion of sites is the difference between the maximum adsorption density of phosphate and arsenate at pH 3 and is equal to 0.63 sites nm-2. By applying the equilibrium constants obtained from noncompetitive adsorption experiments and SCM optimization, the two-site model successfully predicted inhibition of arsenate adsorption by phosphate. The model provided good simulations (e15% error) of the experimental data at total As loadings of 6.67 and 80.1 µM and total P loadings of 129 and 323 µM (Figures 3a). The model did overpredict the inhibition by 45–60% at pH 9 and the low As loading of 6.67 µM; this overprediction is also observed for the arsenate-only experiments and is not primarily caused by factors introduced by the competitive adsorption model. Phosphate adsorption was also affected by the presence of arsenate. The model fit phosphate adsorption data within 25% at 129 µM phosphate with either 6.67 or 80.1 µM arsenate for pH less than 9 (Figure 4). Arsenate inhibition of phosphate adsorption was limited because of the large ratio of P to As loading at a total As loading of 6.67 µM; most source waters will have an excess of P relative to As. Inhibition was most apparent at the higher total As loading of 80.1 µM. Environmental Implications. A surface complexation model involving a small number of adsorption reactions successfully predicted arsenate adsorption and competitive adsorption of arsenate and phosphate over a wide pH range and variable concentrations of adsorbates. The model was least successful in predicting competition at pH 9. The model may be used to predict arsenic adsorption for a given water source based on solution chemistry instead of on the results of time-consuming and costly experiments specific to a water source. The model can be expanded to account for effects of other solutes that can interfere with arsenate adsorption such as H4SiO4, which usually occurs at higher concentrations than phosphate. With slight modifications, the model may also be applicable to arsenic adsorption to other iron oxidebased sorbents used in water treatment, such as granular ferric hydroxide. The sorbent studied is potentially useful for drinking water treatment. It has a high capacity for arsenate adsorption with a maximum adsorption density of 9.3 mg g-1 at neutral pH. The sorbent has arsenic adsorption capacities comparable to those observed for amorphous iron oxides and goethite, and it has the advantage of being composed of physically stable sand-size particles that can be packed in columns. VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

151

Acknowledgments Partial funding and sorbent for this project were provided by EnviroScrub Technologies Corporation. Charles Hammel of Enviroscrub provided additional information about the sorbent. The School of Engineering and Applied Science at Washington Univeristy in St. Louis provided H.Z. with a firstyear fellowship. H.Z. was also supported by the Charles and Marlene Buescher and Cecil Lue-Hing & Bertha Winingham Lue-Hing schorlarships. B.F. was supported through a National Science Foundation Research Experiences for Undergraduates program (Award ID: 0139400). We appreciate the helpful suggestions of two anonymous reviewers, which aided in our improvement of the manuscript.

Supporting Information Available Table of elemental composition of the sorbent and three figures. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Vaughan, D. J. Arsenic. Elements 2006, 2, 71–75. (2) Dixit, S.; Hering, J. G. Comparison of arsenic(V) and arsenic(III) sorption onto iron oxide minerals: Implications for arsenic mobility. Environ. Sci. Technol. 2003, 37, 4182–4189. (3) Gao, Y.; Mucci, A. Individual and competitive adsorption of phosphate and arsenate on goethite in artificial seawater. Chem. Geol. 2003, 199, 91–109. (4) Sun, X. H.; Doner, H. E. An investigation of arsenate and arsenite bonding structures on goethite by FTIR. Soil Sci. 1996, 161, 865–872. (5) Gu, Z. M.; Fang, J.; Deng, B. L. Preparation and evaluation of GAC-based iron-containing adsorbents for arsenic removal. Environ. Sci. Technol. 2005, 39, 3833–3843. (6) Vaughan, R. L.; Reed, B. E. Modeling As(V) removal by an iron oxide impregnated activated carbon using the surface complexation approach. Water Res. 2005, 39, 1005–1014. (7) Benjamin, M. M.; Sletten, R. S.; Bailey, R. P.; Bennett, T. Sorption and filtration of metals using iron-oxide-coated sand. Water Res. 1996, 30, 2609–2620. (8) Sylvester, P.; Westerhoff, P.; Möller, T.; Badruzzaman, M.; Boyd, O. A hybrid sorbent utilizing nanoparticles of hydrous iron oxide for arsenic removal from drinking water. Environ. Eng. Sci. 2007, 24, 104–112.

152

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

(9) Hiemstra, T.; Van Riemsdijk, W. H. Surface structural ion adsorption modeling of competitive binding of oxyanions by metal (hydr)oxides. J. Colloid Interface Sci. 1999, 210, 182–193. (10) Manning, B. A.; Goldberg, S. Modeling competitive adsortpion of arsenate with phosphate and molybate on oxide minerals. Soil Sci. Soc. Am. J. 1996, 60, 121–131. (11) Schwertmann, U.; Fechter, H. The point of zero charge of natural and synthetic ferrihydrites and its relation to adsorbed silicate. Clay Miner. 1982, 17, 471–476. (12) Good, N. E.; Winget, G. D.; Winter, W.; Connolly, T. N.; Izawa, S.; Singh, R. M. M. Hydrogen ion buffers for biological research. Biochemistry 1966, 5, 467–477. (13) Clesceri, L. S., Greenberg, A. E., Eaton, A. D., Eds. Standard Methods for the Examination of Water and Wastewater, 20th ed.; American Public Health Association, American Water Works Association, and Water Environment Federation: Washington, DC, 1999. (14) Herbelin, A. L.; Westall, J. C. FITEQL - A Computer Program for Determination of Chemical Equilibrium Constants from Experimental Data; OR State University, 1999. (15) Waite, T. D.; Davis, J. A.; Payne, T. E.; Waychunas, G. A.; Xu, N. Uranium(VI) adsorption to ferrihydrite: application of a surface complexation model. Geochim. Cosmochim. Acta 1994, 58, 5465– 5478. (16) Sherman, D. M.; Randall, S. R. Surface complexation of arsenic(V) to iron(III) (hydr)oxides: Structural mechanism from ab initio molecular geometries and EXAFS spectroscopy. Geochim. Cosmochim. Acta 2003, 67, 4223–4230. (17) Waychunas, G. A.; Rea, B. A.; Fuller, C. C.; Davis, J. A. SurfaceChemistry of Ferrihydrite 0.1. EXAFS Studies of the Geometry of Coprecipitated and Adsorbed Arsenate. Geochim. Cosmochim. Acta 1993, 57, 2251–2269. (18) Benjamin, M. M. Modeling the mass-action expression for bidentate adsorption. Environ. Sci. Technol. 2002, 36, 307–313. (19) Benjamin, M. M. Water Chemistry, 1st ed.; McGraw-Hill: New York, 2002. (20) Morel, F. M. M.; Hering, J. G. Principles and Applications of Aquatic Chemistry; Wiley Interscience: New York, 1993. (21) Tejedor-Tejedor, M. I.; Anderson, M. A. The protonation of phosphate on the surface of goethite as studied by CIR-FTIR and electrophoretic mobility. Langmuir 1990, 6, 602–611. (22) Roy, W. R.; Hassett, J. J.; Griffin, R. A. Competitive coefficients for the adsorption of arsenate, molybdate and phosphate mixtures by soils. Soil Sci. Soc. Am. J. 1986, 50, 1176–1182. (23) Hingston, F. J.; Posner, A. M.; Quirk, J. P. Competitive adsorption of negatively charged ligands on oxide surfaces. Discuss. Faraday Soc. 1971, 52, 334–342.

ES071553D