Comparison of Arsenic(V) and Arsenic(III) Sorption onto Iron Oxide Minerals: Implications for Arsenic Mobility S U V A S I S D I X I T A N D J A N E T G . H E R I N G * California Institute of Technology, 1200 East California Boulevard, Environmental Science and Engineering (138-78), Pasadena, California 91125 Arsenic derived from natural sources occurs in groundwater in many countries, affecting the health of millions of people. The combined effects of As(V) reduction and diagenesis of iron oxide minerals on arsenic mobility are investigated in this study by comparing As(V) and As(III) sorption onto amorphous iron oxide (HFO), goethite, and magnetite at varying solution compositions. Experimental data are modeled with a diffuse double layer surface complexation model, and the extracted model parameters are used to examine the consistency of our results with those previously reported. Sorption of As(V) onto HFO and goethite is more favorable than that of As(III) below pH 5-6, whereas, above pH 7-8, As(III) has a higher affinity for the solids. The pH at which As(V) and As(III) are equally sorbed depends on the solid-to-solution ratio and type and specific surface area of the minerals and is shifted to lower pH values in the presence of phosphate, which competes for sorption sites. The sorption data indicate that, under most of the chemical conditions investigated in this study, reduction of As(V) in the presence of HFO or goethite would have only minor effects on or even decrease its mobility in the environment at near-neutral pH conditions. As(V) and As(III) sorption isotherms indicate similar surface site densities on the three oxides. Intrinsic surface complexation constants for As(V) are higher for goethite than HFO, whereas As(III) binding is similar for both of these oxides and also for magnetite. However, decrease in specific surface area and hence sorption site density that accompanies transformation of amorphous iron oxides to more crystalline phases could increase arsenic mobility. Introduction Arsenic has been found to occur naturally at concentrations exceeding 10 ppb or 0.13 ��M, the United States and WHO drinking water standards (1, 2), in groundwater in many countries and has been implicated in human disease and mortality, most notably in Taiwan (3), India (4), Chile (5), and Bangladesh (6, 7). These occurrences of arsenic have been distinguished based on the prevailing redox conditions in the aquifers (7). Under oxidizing conditions, As(V) is predominant and is mobilized at high pH (7). In reducing environments, arsenic occurs mostly as As(III). Several mechanisms (singly or in combination) have been invoked to explain arsenic mobility. These include microbial reduction of As(V), reductive dissolution of iron oxyhydroxide phases, and competition of solutes for sorption sites on iron oxides (6-11). A close coupling between the biogeochemical cycles of iron and arsenic in both oxidizing and reducing environ- ments has been well established (for reviews see refs 7, 12- 14). Numerous studies have quantified and modeled As(V) and As(III) sorption onto amorphous iron oxides, goethite, lepidocrocite, and hematite (15-26). Both As(V) and As(III) sorb strongly to iron oxide however, the sorption behavior of arsenic is dependent on its oxidation state and the mineralogy of the iron oxides. Competition between arsenic and other sorbates (such as phosphate, silicic acid, and bicarbonate) has also been studied (22-26). Phosphate, whose concentrations in groundwater can exceed those of arsenic, is particularly effective at competing with arsenate for sorption sites on iron oxide minerals (24-26). A positive correlation has been found between arsenic and phosphate in groundwater in Bangladesh (27). In contrast, silicic acid (although present at concentrations 10 times higher than phosphate) shows no correlation with arsenic levels of the well waters (van Geen, personal communication). Iron oxide minerals often form in natural waters and sediments at oxic-anoxic boundaries. The solids formed initially are poorly crystalline and have high specific surface area but, over time, undergo transformation to more crystal- line forms, such as goethite or hematite. The time scale of these transformations varies depending on temperature, pH, and the presence of other co-occurring solutes (28-30). In natural sediments, both crystalline and amorphous iron oxide minerals can coexist crystalline solids can be 2 to 10 times more abundant (based on Fe content) than amorphous solids (31). Recent studies have also shown that amorphous iron oxides are transformed to magnetite during reductive dis- solution (32-33). Therefore, the mobilization of arsenic during the transformation of iron oxides will depend on the relative affinity of the original and transformed minerals for the arsenic species. The objectives of this study are to investigate the condi- tions under which reduction of arsenate would favor its mobilization and compare the relative affinity of arsenic for different iron oxide minerals. Previous studies quantifying arsenic sorption onto iron oxide minerals have focused either on one arsenic species or one type of solid. Therefore, the combined effect of arsenate reduction and diagenesis of iron minerals on arsenic mobility remains unclear. In this study, we compare the sorption of As(V) and As(III) onto HFO and goethite at different total arsenic concentrations and in the presence of phosphate. These results show that the effect of As(V) reduction on arsenic mobility will be strongly de- pendent on solution composition. Arsenic mobility during early diagenesis is evaluated by comparing the affinity and sorption densities of HFO, goethite, and (for the first time) magnetite for As(III). The different iron oxide minerals are found to exhibit, on a surface area normalized basis, similar sorption densities for arsenite and also similar binding constants. Materials and Methods Materials. All chemicals were reagent grade and used without further purification. Solutions were prepared with Milli-Q (18 M��-cm) water. Plastic volumetric flasks and reaction vessels (Polypropylene) were cleaned with 1% HNO3 and rinsed several times with deionized water before use. As(V) stock solution was prepared from reagent grade Na2HAsO4��� * Corresponding author phone: (626)395-3644 fax: (626)395-2940 e-mail: jhering@caltech.edu. Environ. Sci. Technol. 2003, 37, 4182-4189 4182 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003 10.1021/es030309t CCC: $25.00 ��� 2003 American Chemical Society Published on Web 08/13/2003

7H2O (Sigma) and As(III) stock from NaAsO2 (J. T. Baker). Phosphate stock solution was prepared from NaH2PO4���H2O (Merck). Mineral Sorbents. Sorption experiments were conducted with three iron oxides: HFO, goethite, and magnetite. HFO was synthesized by the method of Schwertmann and Cornell (34), as modified by Wilkie and Hering (20). Fresh HFO was prepared each day for the experiments. Goethite was prepared by mixing 180 mL of 5 M KOH with 100 mL of 1 M Fe(NO3)3���9H2O (34). The suspension was diluted to 2 L and aged for 60 h at 70 ��C. After the synthesis of the minerals, suspensions were washed repeatedly with deionized water and freeze-dried. Both oxides were prepared under ambient atmospheric conditions. Magnetite was synthesized by slow addition of a mixture of ferrous and ferric chloride solutions (10 mL of 2 M FeCl2 added to 40 mL of 1 M FeCl3 solution) to 400 mL of 0.9 M NH3 solution (35). Both the ammonia solution and aqueous Fe solutions were purged with N2 gas prior to their use and purging was continued during the mineral synthesis. The supernatant was then decanted, and solids were washed five times in the reaction vessel with N2-purged water. Solid suspensions were then transferred to 250 mL centrifuged bottles and washed repeatedly with deionized water. The solids were then freeze-dried. The multipoint BET surface areas of the solids determined by N2 adsorption using a Gemini 2360 surface area analyzer (Micromeritics) are given in Table 1. For HFO, a specific surface area of 600 m2 g-1 is assumed (36). The identity of the solids (except HFO) was confirmed with a Scintag Pad V X-ray diffractometer using a copper X-ray source. Arsenic Adsorption Edges. All the experiments were performed with a background electrolyte of 0.01 M NaClO4. Sorption of As(III) and As(V) onto iron oxides was initiated by the addition of solid suspensions at different pH values to arsenic stock solutions. The pH of the resulting suspensions drifted (by up to 1-2 pH units at near neutral pH) in the first 5-10 min and then stabilized. The final pH, measured at the end of the experiments, is reported. Suspensions were continuously mixed using an end-over-end shaker. Experi- mental conditions are summarized in Table 1. The solid concentrations chosen correspond to similar total concen- trations of surface sites for HFO and goethite and about 2-fold higher surface site concentrations for magnetite. Equilibra- tion times for experiments with HFO and goethite were based on previous studies (18, 16), and determined for magnetite in preliminary experiments. After the desired reaction time, the suspensions were centrifuged, and the supernatant was filtered through 0.2 ��m filters (Nalgene, cellulose acetate) and analyzed for arsenic. The amount of arsenic sorbed to the solids was calculated by difference. In experiments with HFO, the solid concentration was verified by dissolving the suspension in 1% HNO3 and analyzing for total iron. Sorption experiments with HFO and goethite were conducted in open atmospheric conditions. A few experiments performed under N2 in an anaerobic chamber (Vacuum Atmospheres Dri-Lab HE-63-P) did not reveal any difference in the sorption edges. All sorption experiments with magnetite were conducted inside the anaerobic chamber. Sorption Isotherms. Sorption isotherms on the iron oxides were conducted to estimate the maximum sorption density. The experiments were performed at pH 4.0 ( 0.1 for As(V) and at pH 8.0 ( 0.2 for As(III), corresponding to the pH values of maximum sorption of the arsenic species. The pH of the suspensions was adjusted with HClO4 and NaOH during the experiment. Arsenic Sorption in the Presence of Phosphate. The influence of phosphate on arsenic sorption was studied by simultaneously adding arsenic and phosphate stock solutions to sorbent suspensions prepared at different pH values. Both the phosphate and arsenic concentrations of filtered solutions and pH were measured at the end of the experiment. Analytical Methods. The pH of the solutions was mea- sured using an Orion model 720A pH meter, calibrated using commercial pH 4.0, 7.0, and 10.0 buffers. Filtered solutions from sorption experiments were acidified with 1% HNO3 and analyzed for total arsenic, phosphorus, and iron by ICP-MS. The relative standard deviation of these measurements was always better than 5%. Surface Complexation Model. A surface complexation model with the diffuse double layer (DDL) model for electrostatics was used to describe the arsenic sorption edges (36). Constants for protonation of the surface hydroxyl groups and aqueous species were taken from previous studies (Table 2). The stoichiometries of the surface complexes used to fit sorption data are listed in Table 2. Similar surface complexes have been used in previous studies (19, 24, 26). The computer program FITEQL (40) was used to obtain the intrinsic As(III) and As(V) surface complexation constants. The surface site densities were set to values obtained from sorption isotherms (Table 1), and only the surface complexation constants were optimized. Model predictions with fixed site densities and complexation constants were performed using MINEQL+ (37). For both fitting and predictions, activity coefficients of aqueous species were calculated using the Davies equation. Results and Discussion Arsenate Sorption Edges on HFO and Goethite. Sorption of arsenate on HFO and goethite for total arsenate concentra- tions ranging from 10 to 100 ��M is shown in Figure 1. In the pH range of the experiments, arsenate sorption on both the solids decreases with increasing pH (Figure 1). At pH 4.0, the amount of As(V) sorbed onto both HFO and goethite remains close to 100% for total As(V) concentrations below 50 ��M but decreases to 65% (with HFO) or 70% (with goethite) at 100 ��M total As(V). Substantially less As(V) is sorbed at higher pH values. At pH 10, the sorbed concentra- tions of As(V) on HFO converge to a value of about 300 ��mol g-1, corresponding to 70, 30, and 7% of the total As(V) concentrations of 10, 35, and 100 ��M, respectively. With goethite, the sorbed As(V) concentrations at pH 10 show less convergence and correspond to 85, 40, and 25% of the total As(V) concentrations of 10, 50, and 100 ��M, respectively. The higher maximum sorption densities observed with HFO (2,100 ��mol g-1) as compared to goethite (140 ��mol g-1) are consistent with similar densities of sorption sites per unit area for both the solids but a much greater specific surface area for HFO than goethite. Similar trends have been observed in several previous studies. However, the sorbent and sorbate concentrations were higher by as much as 2 orders of magnitude in those TABLE 1. Material Properties and Experimental Conditionsa HFO goethite magnetite specific surface areab (m2 g-1) 600c 54 90 solid concentration (g L-1) 0.03 0.5 0.5 equilibration time (h) 4 24 24 total As concentration (��M) 10-100 10-100 50-150 As-to-Fe ratio (�� 10-3) 30-300 2-18 8-23 sorption densityd As (V) mol As per mol Fe 0.24 0.016 sites per nm2 2.6 2.0 As (III) mol As per mol Fe 0.31 0.016 0.025 sites per nm2 3.5 2.0 2.2 a All experiments conducted at room temperature (23 ( 0.4 ��C) with 0.01 M NaClO4 background electrolyte medium. b Measured by N2 BET. c Assumed value (36). d Obtained from sorption isotherm. VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4183