Macroscopic Studies of the Effects of Selenate and Selenite on Cobalt

University of Maine, Orono, Maine 04469. LYNN E. KATZ*. Department of Civil and Environmental Engineering, ECJ 8.6,. University of Texas, Austin, Texa...
0 downloads 0 Views 130KB Size
Environ. Sci. Technol. 2002, 36, 1212-1218

Macroscopic Studies of the Effects of Selenate and Selenite on Cobalt Sorption to γ-Al2O3 ERIC J. BOYLE-WIGHT Department of Civil and Environmental Engineering, University of Maine, Orono, Maine 04469 LYNN E. KATZ* Department of Civil and Environmental Engineering, ECJ 8.6, University of Texas, Austin, Texas 78712 KIM F. HAYES Department of Civil and Environmental Engineering, The University of Michigan, Ann Arbor, Michigan 48109

Metal ion sorption can be significantly impacted by the presence of other solutes or complexing species. In this research, macroscopic sorption studies were conducted to evaluate the effect of strongly sorbing Se(IV) and weakly sorbing Se(VI) oxyanions on cobalt(II) sorption to γ-Al2O3. Se(IV) was found to significantly alter Co(II) sorption as a function of Co(II) surface coverage, while Se(VI) was found to have no effect on Co(II) sorption. Under low Co(II) surface loadings (0.5 µmol/m2) where coprecipitation of Co(II) and Al(III) in the form of layered double hydroxides (LDH) is expected to be the dominant sorption mechanism for the single-sorbate case. The extent of the Co(II) sorption reduction in Co(II)/Se(IV) bisorbate systems compared to the corresponding single-sorbate systems increased with increasing Co(II) surface coverage. The rate of Co(II) desorption was reduced in the presence of Se(IV) compared to the single-sorbate case, indicating a direct interaction between Co(II) and Se(IV). A reaction between Co(II) and Se(IV) is further supported by an increase in Se(IV) sorption in the same bisorbate samples where Co(II) sorption is decreased. Thus, the macroscopic data indicates Se(IV) may be altering the mechanism of Co(II) sorption, potentially forming a ternary surface complex or different surface precipitate.

Introduction A mechanistic description of reactions occurring at mineral/ water interfaces has profound implications for both environmental assessment and technological development. Many environmental problems associated with raw material pro* Corresponding author phone: (512)471-4244; fax: (512)471-5870; e-mail: [email protected]. 1212

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

duction, agricultural methods, and waste management practices have resulted in unnaturally high concentrations of toxic metal and nonmetal ions in the environment. Management of these contaminated areas often requires reliable, quantitative prediction of the fate and transport of metal ion contaminants in surface and groundwaters. Ion partitioning (or sorption) to mineral surfaces typically reduces solute mobility and often controls the fate and transport of metal ions. Predictive models for sorption of metal cations and oxyanions in single-sorbate systems have developed significantly over the past two decades. However, metal ion sorption in multicomponent systems containing both cations and oxyanions has received much less attention, and our understanding of even bisorbate sorption is limited. To develop adequate understanding of multicomponent sorption, investigations need to be conducted at both macroscopic and molecular scales. Macroscopic data acquired from batch equilibrium studies can be used to describe specific ion sorption as a function of solution conditions. Molecular scale studies using techniques such as X-ray absorption spectroscopy (XAS) provide a means to elucidate mechanistic changes underlying observed trends in the macroscopic data. The goal of this study was to determine the effect of two oxyanions of selenium, selenite (Se(IV)), and selenate (Se(VI)) on the sorption of cobalt(II) to γ-Al2O3. The effect of Se(IV) on Co(II) sorption was studied over a range of cobalt surface coverages. The studies with Se(VI) focused on high cobalt surface coverages. The overall research plan was comprised of three parts: (1) macroscopic sorption experiments at various cobalt coverages and anion concentrations, (2) XAS analysis of single and multisorbate sorption samples, and (3) preparation and spectroscopic investigation of various precipitates and coprecipitates. This paper focuses on the results of the macroscopic sorption studies.

Background Iron, silica, manganese, and aluminum oxide minerals are prevalent in the environment either as individual soil components or as surface coatings on clays and other soil phases. An aluminum oxide, γ-Al2O3, was chosen for this research as a well-characterized representative of oxide minerals and oxide coatings that form on particle surfaces. The relatively high surface area of γ-Al2O3 facilitates a wide range of sorption experiments. Additionally, aluminum oxide samples are well suited to surface spectroscopic studies of sorbed transition metals due to the low absorption energy of aluminum. This property allows for collection of cobalt XAS spectra that would be significantly more difficult in the presence of other oxide minerals such as those of iron oxide. Therefore, even though γ-Al2O3 is not commonly found in natural environments, it does serve as an analogue to more commonly found surfaces such as aluminum hydroxides and aluminol layers in clays. In this regard, studying reactions at the alumina/water interface provides a clearer understanding of reaction phenomena occurring in natural systems while maintaining optimal experimental conditions for molecular scale studies. Cobalt Sorption. Cobalt(II) sorption has been studied on many types of mineral surfaces including amorphous iron oxyhydroxide (1), goethite (2), γ-Al2O3 (3), R- Al2O3 (4-7), TiO2 (3, 5, 8), kaolinite (9-11), and montmorillonite (11, 12). Macroscopic sorption studies on aluminum oxides demonstrate that cobalt sorption is highly pH dependent (5, 6, 8, 13) and is not affected significantly by changes in ionic strength, a characteristic typical of strongly sorbing metal ions (6, 8). Molecular scale investigations of cobalt removal 10.1021/es001775a CCC: $22.00

 2002 American Chemical Society Published on Web 02/16/2002

FIGURE 1. Representative ternary complexes and corresponding surface complexation reactions for Co(II) and Se(IV) (tSOH represents a surface hydroxyl site). indicate that Co(II) forms mono- or bidentate innersphere (i.e., in which cobalt adsorbs directly to surface oxygen atoms) surface complexes with aluminum oxides at low surface loadings (8) (less than 0.1 µmol/m2 BET surface area). Increasing the surface loading to greater than 0.25 µmol/m2 results in polymerization of cobalt regardless of the fact that (1) at this coverage less than 10% of the surface contains cobalt (8) and (2) Co(II) is undersaturated with respect to bulk solution Co(OH)2 solubility (14). In the study by Chisholm-Brause (8), extended X-ray absorption fine structure (EXAFS) analysis indicated that increasing the surface coverage from 0.25 to 9.92 µmol/m2 resulted in an increase in the size or number of polymeric surface complexes. Thus, the structure of Co(II) at or near the surface changes as a function of surface coverage. Our understanding of the nature of the polymer phase has evolved over the past several years. Recent studies suggest the formation of a coprecipitated phase containing Co(II) or Ni(II) and Al(III) and having a hydrotalcite-like structure (4, 15). Thus, Co(II) sorption to aluminum oxide follows a mononuclear adsorption process at low surface coverage (less than 0.1 µmol/m2) and a coprecipitation process at high surface coverage (greater than ca. 0.5 µmol/m2). Selenium Sorption. Selenium is a concern in agricultural areas because it is a necessary nutrient for many crops but is toxic at high concentrations, requiring a critical balance in agricultural soils (16). Sorption of selenium oxyanions is well characterized for various oxides (1, 17-19). The two oxidation states of Se in the oxyanions SeO42- (Se(VI)) and SeO32- (Se(IV)) exhibit very different sorption behavior. Se(VI) sorption is strongly affected by changes in ionic strength (17) and exists exclusively as SeO42- in solutions with a pH greater than 4. The ionic strength behavior is typical of weakly sorbing anions that predominantly form outersphere ionic surface complexes (in which a water of hydration separates selenate from the surface). Selenous acid is weaker than selenic acid with first and second acidity constants of 2.35 and 7.94 (28). Thus, at near neutral pH, Se(IV) exists as both SeO3) and HSeO3- anions, both of which may participate in sorption. Se(IV) sorbs at a higher pH compared to Se(VI), and spectroscopic data suggest the formation of mono- or bidentate innersphere surface complexes (19, 20). While the presence of ionic strength effects for Se(VI) sorption is consistent with outersphere surface complexes, iron-selenate inner sphere surface complexes or surface precipitates have been proposed based on EXAFS data (21). In any case, Se(VI)

surface complexes on oxides and soils are considered to be much weaker than Se(IV) surface complexes (16, 22). The different sorption behavior of these two oxidation states of selenium provides an excellent opportunity for studying competitive sorption with cobalt. Based on the specific sorption of several oxyanions (1), selenite also serves as an analogue to other strongly sorbing oxyanions such as arsenate, phosphate, and thiosulfate. In contrast, selenate serves as an analogue for weakly sorbing oxyanions such as sulfate. Multisorbate Sorption. Previous research examining cation/anion multisorbate systems has focused primarily on macroscopic scale investigations. One broad anion/cation study demonstrated that many strongly sorbing oxyanions such as SeO32-, AsO42-, and PO43- increased the removal of Zn, Co, and Cd on amorphous iron oxyhydroxide (1) whereas weakly sorbing anions (SO42-, CrO42-, and AsO32-) had no effect on metal ion removal. While metal ion surface coverages in this study were less than 0.4 µmol/m2, PO43- has also been shown to increase Cd(II) sorption on goethite at surface coverages as high as 1.5 µmol/m2 (23). One mechanism proposed in these studies is enhanced electrostatic attraction (23) which is supported by electrophoteric mobility studies showing that SeO32- and PO43- alter the net surface charge while SO42- has no effect (24). A second proposed mechanism is the formation of a secondary phase such as FeAsO4 (1). In their work, Benjamin and Bloom also found increases in Cd(II) removal through addition of FeAsO4(s). Increased removal was considered to result from either a greater affinity of the sorbing metal for the new surface phase or an increase in availability of sorption sites. More direct evidence was provided by Waychunas (25) in a wide-angle X-ray scattering (WAXS) study showing that growth of iron oxyhydroxide crystals is poisoned by the presence of coprecipitated FeAsO4. These studies demonstrate that oxyanion interactions with oxide surfaces can strongly influence additional surface reactions. Ternary surface complexes have also been proposed for several metal-anion systems. In general, complexes can sorb to the surface in three possible configurations as shown in Figure 1: (1) through a metal-surface bond, (2) through a ligand-surface bond, or (3) both metal and ligand can bind simultaneously to the surface (bridge bonded) (26, 27). Direct evidence for ternary surface complexes has been demonstrated by Bargar et al. (27) in an XAFS investigation of leadchloro complexes. Evidence was presented for lead-chloro VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1213

FIGURE 2. Co(II) sorption to 2 g/L γ-Al2O3: 430 day rate study (inset shows data for initial 120 h).

TABLE 1. Bisorbate Sample Conditions aluminum oxide concn, g/L

Co(II) total concn added, mM

Se(IV) [Se(VI)] total concn added, mM

1 2 2 2 5 25 25

0.24 0.24 0.24 0.24 0.24 0.1 0.1

0.4 0.4 1.2 [2.0] 0.4 1.0 10

ionic strength, M 0.1 0. 0.01 0.01 0.1 0.1 0.1 and 0.05

max Co(II) surface coverage, µmol/m2 3.5 1.5 1.5 1.5 0.6 0.05 0.05

complexes on goethite surfaces with Pb(II) bonded to surface oxygen atoms and Cl- bonded to surface iron atoms. However, no evidence was found for lead-chloro complexes on γ-Al2O3. This is partially ascribed to the inability of Clto bond to surface Al atoms, underscoring the potential importance of the ability of the ligand to form aqueous complexes with both surface and sorbate metals. The double bonding arrangement described by Bargar et al. (27) supports the observed increase in Pb removal due to the formation of a stronger binding complex (PbCl+).

Methods Batch equilibrium studies of Co(II) sorption to γ-Al2O3 were conducted with and without Se(IV) or Se(VI) present for a range of surface loading conditions (surface coverages) as shown in Table 1. The maximum surface coverage for each experiment was calculated assuming 100 percent of the cobalt initially added to the system is sorbed to the surface. The actual surface coverage of the samples varied as a function of pH from zero to the maximum value. The maximum surface coverage was varied by changing the initial sorbate and sorbent concentrations. In all cases, the solution conditions were undersaturated with respect to Co(OH)2 and CoSeO3 solubilities (28). The degree of undersaturation varies depending on solution conditions but typically is less than 70% of saturation for Co(OH)2 and less than 30% of saturation for CoSeO3. Because Se(IV) and Se(VI) sorb over a much broader pH range than Co(II), it was possible to achieve a relatively constant anion coverage for cobalt sorption over a range of pH values. The surface area of the commercially available γ-Al2O3 (Buehler, Lake Bluff, IL) used in these experiments was 1214

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

determined to be 80 m2/g by the N2 BET method (29). Oxide suspensions were prepared as stock solutions by hydrating the dry powder in boiled (CO2 free) double deionized water (NANOpure, Barnstead) for a minimum of 2 days. Sorption samples were prepared by adding appropriate amounts of 0.1 or 0.01 M stock solutions of Co(NO3)2, Na2SeO3 or Na2SeO4, and NaNO3 prepared from purified salts (Puratronic, Alfa Aesar) to polypropylene (PP) centrifuge tubes containing a predetermined volume of boiled (CO2 free) NANOpure water and oxide suspension and sealed while under a nitrogen atmosphere to prevent recarbonation. An appropriate amount of sodium nitrate was added to maintain the ionic strength. The pH was adjusted to a different value in each tube after solute addition by adding incremental volumes of carbonate free 0.1 M NaOH or HNO3. Samples, in sets of cobalt only (single sorbate) and cobalt plus selenium (bisorbate samples) were rotated end over end for a minimum of 48 h. Blanks containing no oxide were run at the high end of the pH range, exhibiting no precipitate formation (carbonate or hydroxide) or loss of solute. The reaction time for the experiments was determined by conducting a rate study of single sorbate samples under similar experimental conditions. A set of identical samples were prepared and sampled periodically over several months. Cobalt sorption was very rapid during the first few hours and then gradually approached apparent equilibrium after 24 h (Figure 2). No significant additional removal occurred after 48 h and up to 100 days. After reaction, the pH was measured using an Orion 8303 electrode with an error of ( 0.01 pH units before centrifuging at 10-12 000 rpm for 40-45 min. Aliquots of supernatant were acidified in 0.1 M ultrapure HNO3 to preserve the redox state and analyzed for solute concentration by flame atomic absorption with an error of ( 2%. Errors on the data plots are typically smaller than the symbols used in the figures. The percent sorbed was determined as the difference between the solute concentration and a blank sample containing no oxide material. Blanks were prepared for each sample set: one at low pH for solution concentrations and one at the highest pH in the samples as a control for the formation of solution precipitates.

Results and Discussion Single Sorbate Sorption. While electrolyte components are not necessarily inert and may be competitive sorbates, for the purpose of this discussion single sorbate refers to a system containing only Co(II), Se(IV), or Se(VI) in conjunction with a sodium nitrate background electrolyte. Likewise the term

FIGURE 3. Single-sorbate pH sorption edges for (a) Co(II) and (b) Se(IV) and Se(VI) on 5 g/L γ-Al2O3 (M.S.C. refers to the maximum surface coverage assuming 100% removal in units of µmol/m2). bisorbate will be used to represent systems containing Co(II) with Se(IV) or Se(VI). Single sorbate sorption data exhibit pH sorption trends characteristic of cations (Figure 3a) and anions (Figure 3b) on metal oxides (30). For example, sorption increases with increasing pH for Co(II) and decreases with increasing pH for both of the anions studied. Sorption of Se(VI) (the weaker sorbing anion) is strongly influenced by ionic strength and sorbs at a lower pH than Se(IV). Co(II) sorption in these systems is representative of a surface-sorbate interaction where the ion is sorbing against an unfavorably charged surface, indicative of strong surface affinity for the sorbate and the formation of metal ion-surface oxygen complexes. Indeed, reported values of the pH of the point of zero charge (PZC) for γ-Al2O3 range between 8.1 and 9.8 (31). Thus, the surface without adions present has a net positive charge for all data shown. The shift in the Co(II) pH sorption edge to higher pH for decreasing γ-Al2O3 concentrations from 25 g/L to 5 g/L is also consistent with surface complexation reactions involving a finite number of surface hydroxyl sites (32, 33). The large pH shift observed for the Co(II) sorption edge between 25 g/L and 5 g/L data is also consistent with XAS results (3, 8, 34) and surface complexation modeling studies that include low surface coverages (6), suggesting that Co(II) sorption to aluminum oxides is dominated by chemiadsorbed mononuclear surface complexes at surface coverages less than 0.1 µmol/m2. In contrast to low sorption coverage, comparison of data collected at the same initial Co(II) concentration and γ-Al2O3 concentrations of 5 and 1 g/L (leading to higher cobalt surface coverage under similar bulk solution conditions) show little

FIGURE 4. Single- and bisorbate pH sorption edges for Co(II) and Se(IV) on 25 g/L γ-Al2O3 for low cobalt surface coverage and (a) moderate Se(IV) surface coverage and (b) high Se(IV) surface coverage (selenite coverage at pH ) 7.5). change in the pH of the sorption edge. Previous research has shown that Co-Al coprecipitates with hyrotalcite like structures (4, 34) dominate at coverages greater than ca. 0.5 µmol/m2. Hydrotalcites are mixed Me(II), Me(III) hydroxides that have a significantly lower solubility than pure Me(II) hydroxide phases (35). Cobalt hydrotalcites have been synthesized with aluminum under conditions slightly over (36, 37) and slightly under the cobalt hydroxide solubility (4, 38), similar to conditions in this study. The maximum surface coverage (0.6 µmol/m2 at 100% removal) shown for the 5 g/L data is consistent with the range of surface coverages where coprecipitation begins to dominate. The steepening of the sorption edge at lower solid concentration and higher surface coverage is also consistent with a change in partitioning mechanism from adsorption to precipitation (34). Bisorbate Sorption. The goal of the bisorbate sorption experiments was to evaluate the effect of a strongly competing oxyanion on Co(II) sorption at low surface coverage (less than 0.1 µmol/m2) and the effect of strongly and weakly competing oxyanions on Co(II) sorption at high (greater than 0.5 µmol/m2) surface coverage. Low Coverage Bisorbate Sorption. Bisorbate systems were conducted for low Co(II) surface coverages over a range of Se(IV) coverage conditions (Figure 4a). The presence of Se(IV) did not affect the extent of Co(II) removal at Se(IV) surface coverages less than 0.4 µmol/m2 (corresponding to selenite solute concentrations less than 1.0 mM). Increasing the Se(IV) surface coverage to 2.0 µmol/m2 using an initial concentration of 0.01 M Se(IV) resulted in a dramatic increase VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1215

in Co(II) removal (Figure 4b). The increase in cobalt removal cannot be attributed to precipitation of cobalt selenite based on a reported formation constant for cobalt selenite of 107.08 (28). Furthermore, the control samples not containing γ-Al2O3 did not show any indication of precipitation at similar pH values. The increase in cobalt removal is consistent with Co(II) sorption data on iron oxyhydroxides in the presence of selenite, phosphate, and arsenate (1) and on aluminum oxides in the presence of thiosulfate (26). In the former study, comparison of Co(II) sorption to iron oxyhydroxide in the presence of 10-4 M selenite and 10-3 M selenite showed an increase in cobalt removal at the higher anion concentration. Thus, while specific anion removals were not reported in these previous studies, their data demonstrate sorption enhancement with increasing anion coverage to an even greater extent then observed in our study. The increase in cobalt removal as a function of Se(IV) surface coverage is consistent with an electrostatic enhancement mechanism. Sorbed Se(IV) has been shown to contribute to the net surface charge of goethite, decreasing the electrophoretic mobility with increasing Se(IV) concentration (24). While actual Se(IV) removals were not reported, an estimate of the Se(IV) surface coverage can be calculated based on reported solids characteristics and assuming 100% selenite removal. The data of Hansmann and Anderson (24) show the isoelectric point decreasing from pH ) 9 to pH ) 8.7 to pH ) 5 for 2, 4, and 15 µM selenite concentrations, respectively. These solute concentrations would correspond to surface coverages of 0.8, 1.6, and 6.2 µmol/m2, respectively, assuming 100% removal. For comparison, Se(IV) surface coverages in this study based on measured 100% removal were 0.5 and 5 µmol/m2 for data presented in Figure 4a,b, respectively, or approximately the same order of magnitude as the those reported by Hansmann and Anderson (24). Therefore, the increase in Co(II) removal at high Se(IV) coverage in this study is consistent with an electrostatic effect in which Se(IV) sorption decreases the net positive surface charge at low pH. While the decrease in the positive surface charge may facilitate enhanced Co(II) adsorption, evidence of surface precipitate and ternary complex formation suggest anion/ cation interactions can be much more complicated than expected based solely on surface electrostatic considerations. Benjamin and Bloom (1) showed that Cd sorption to FeAsO4 was very similar to Cd(II) sorption to amorphous iron oxide with arsenate present. Both cases showed a similar increase in Cd(II) removal compared to single sorbate sorption to amorphous iron oxide. They proposed that a secondary ironanion surface phase formed at high anion coverage, providing a new surface for metal sorption. However, their data could also be interpreted assuming formation of a Cd-AsO4 ternary complex. Cadmium has been hypothesized to form ternary complexes with thiosulfate on aluminum and iron oxide surfaces based on the stability of Cd-S2O3 complexes and increased removal at pH 6 (26). The effect of metal-ligand complexation can be rather complicated. In the same study described above, chloro and sulfato ligands decreased Cd(II) sorption due to competitive effects between aqueous complexes and sorption. In contrast, lead-chloride ternary complexes have been proposed to account for an observed increase in lead sorption (39) and have recently been identified spectroscopically on goethite (27). This indicates that the same ligand can have opposite effects depending on surface type and sorbates. Indeed, formation of ternary complexes is not ligand specific but is correlated to the ability of a ligand to form solution complexes with both surface and sorbate metals (27). In this regard, the lack of aqueous complex formation data for metal-selenite complexes makes evaluation of ternary complexes difficult. 1216

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

Co(II) is known to form complexes with several oxyanions including selenate, sulfate, and nitrate; however, formation constants for cobalt selenite complexes have not been reported. Aluminum selenite complex formation data are equally scarce, but aluminum is known to form complexes with other oxyanions. Selenite is also considered to form strong innersphere complexes with aluminum oxides (19) indicating an ability to form coordinative bonds with aluminum. Thus, while more work is needed in the area of selenite solution chemistry, the existence of selenite complexes and subsequent formation of ternary surface complexes cannot be excluded based on known metal ligand chemistry and surface complexes. The greater effect of Se(IV) on Co(II) sorption to iron oxyhydroxide (1) compared to γ-alumina in this study under similar surface loading conditions suggests a stronger affinity of Se(IV) for iron oxides compared to aluminum. The formation of stronger Co-SeO3 ternary complexes on iron surfaces than aluminum surfaces is reasonable. Formation constants presented in Smith and Martell (28) for aluminum and iron complexes are consistently higher for iron. Bargar et al. (27) identified ternary lead-chloro complexes on goethite but not on γ-alumina. In contrast, an electrostatic effect should be more similar for the two minerals since their PZC values are similar. Nevertheless, both an electrostatic effect and ternary complex formation support the observation that Co(II) sorption enhancement is a function of Se(IV) coverage. High Coverage Bisorbate Sorption. Bisorbate experiments at high cobalt surface loadings were examined for a range of cobalt and selenium coverages by changing the oxide concentration while maintaining similar solute concentrations. In these systems, Se(VI) had no effect on Co(II) removal (Figure 5a). In contrast to the Se(VI) high surface coverage data and the Se(IV) low coverage data, Co(II) removal was significantly decreased in the presence of Se(IV) as shown by the shift in the bisorbate pH sorption edge to higher pH in Figure 5a. In this figure, the Se(IV) and Se(VI) surface coverages were similar. Similar data were collected using the same cobalt concentration, similar Se(IV) surface coverages, and oxide concentrations of 1 and 5 g/L (edge plots not shown) to examine the effect of Se(IV) on Co(II) sorption at different cobalt surface coverage ranges. Table 2 summarizes the results of the three Co(II) surface coverage ranges by comparing changes in cobalt removals at pH 7.5. The data show that the effect of selenite on cobalt removal is greater at higher cobalt surface coverage where coprecipitation is considered to predominate over adsorption (4, 8, 34). The difference between the effect of Se(IV) and Se(VI) on Co(II) removal is consistent with the weaker sorbing characteristics of Se(VI). Because the weakly sorbing Se(VI) is more susceptible to competition with sorbing cobalt for surface sites, the lack of effect of Se(VI) on cobalt sorption could be attributed to competitive effects. However, the decrease in Co(II) removal in the presence of Se(IV) cannot be explained strictly by competition for sites. In bisorbate systems for which cobalt reduction was observed, Se(IV) removal increased in the presence of Co(II) relative to single sorbate Se(IV) samples (Figure 5a). Furthermore, competitive adsorption of Se(IV) for mononuclear sorption sites should have only a minor impact on Co(II) removal when coprecipitation dominates. In our studies, the decrease in Co(II) removal only occurred for surface coverage conditions for which hydrotalcite like coprecipitates or surface precipitates are known to be dominant surface species, and the effect of Se(IV) is greater as conditions for coprecipitation become more favorable. Se(IV) is potentially limiting hydrotalcite growth by competing with cobalt for available aluminum. Arsenate has

FIGURE 5. Single- and bisorbate pH (a) sorption and (b) desorption edges for Co(II) and selenium on 2 g/L γ-Al2O3 for high cobalt surface coverage (Co(II) is sorbed in the absence of Se(IV) in the bisorbate desorption edges).

TABLE 2. Effect of Se(IV) on Co(II) Sorption at pH ) 7.5a oxide concn (g/L)

decrease in Co(II) removal with respect to single sorbate Co(II) (%)

single sorbate Co(II) coverage (µmol/m2)

1 2 5

50 42 20

2.8 1.5 0.5

a Se(IV) surface coverages ranged from 1.1 to 1.3 µmol/m2 in these experiments.

been shown to poison growth of amorphous iron oxyhydroxide at high As/Fe ratios (25) due to favorable coprecipitation with iron. Similarly, high Se(IV) coverages may complex or precipitate with aluminum required for cobaltaluminum hydrotalcite formation. While the existence of aluminum selenite complexes or precipitates under solution conditions in this study is only speculative due to the lack of formation constant data, selenite hydrates of both aluminum, Al2(SeO3)3-6H2O (40), and iron Fe2(SeO3)2-6H2O (41, 42) have been synthesized. Under the hypothesis that aluminum-selenite complexes or precipitates form in the bisorbate systems, the solubility condition of the cobalt-aluminum solid solution would be expected to decrease, requiring a higher pH in order to reach the mixed metal hydroxide solubility limit at the same aluminum, selenite, and cobalt total concentrations. Furthermore, adsorption may become a more significant factor with the repression of coprecipitation, resulting in a shift in

the pH sorption edge to higher pH. Thus, the decrease in Co(II) removal may be a result of a change in the relative contribution of coprecipitation and adsorption. The hypothesis that ternary complexes are forming at low cobalt coverage should also hold true at high surface coverages. Formation of ternary complexes is supported by the increase in Se(IV) removal in the presence of Co(II) (Figure 5a) and should result in less cobalt removal than a cobalt/ aluminum coprecipitate phase due to surface site constraints similar to single sorbate Co(II) sorption at low surface coverages. Ternary complexes have been described as either metal-like, sorbing with an increase in pH and bonding through the metal, ligand-like, sorbing with a decrease in pH and bonding through the anion, or bridge bonded through both the metal and the ligand (26, 27). In this case, the ternary complex would bond through the metal or through both the metal and the oxyanion since cobalt removal increases with increasing pH. This is supported by the fact that Se(IV) removal increases to a greater extent at higher cobalt coverages. Desorption Studies. The general procedure for constructing bisorbate experiments involved adding acidified Co(II) and Se(IV) stock solutions concurrently. This procedure resulted in Se(IV) sorbing prior to Co(II) in the course of sample preparation because the initial pH of the samples was maintained below the Co(II) pH sorption edge. To investigate the effect of sorbing Co(II) prior to addition of Se(IV), a similar sample set was prepared by first equilibrating cobalt at 95% removal and then adding Se(IV) and adjusting the pH to lower values. Thus, in this experiment Se(IV) sorption occurs while cobalt is desorbing. The results of this experiment are presented in Figure 5b. Single sorbate Co(II) desorption data coincide with the sorption data, indicating that cobalt removal under these experimental conditions is completely reversible. A separate cobalt desorption rate study showed cobalt sorption in the absence of Se(IV) to be completely reversible within hours (data not presented) verifying that the Se(IV) effect is not a result of an irreversible sorption process attributable to Co(II) sorption. Apparently, the cobalt-aluminum coprecipitates that form in singlesorbate systems during the short reaction times (approximately 48 h) and surface conditions of these experiments are quickly dissolved. In contrast to the single-sorbate data shown in Figure 5b and the bisorbate data in which Se(IV) was sorbed first (Figure 5a), Co(II) desorption in the presence of Se(IV) (Figure 5b) shows an enhancement in Co(II) removal compared to the single sorbate sorption data. These data suggest that the addition of Se(IV) after Co/Al coprecipitate formation does not disrupt the hydrotalcite phase within the reaction times shown. In fact, the addition of Se(IV) appears to prevent the dissolution of the precipitate phase over the time period studied. Regardless of whether the effect is kinetic or thermodynamic irreversibility, the change indicates that Se(IV) interacts with Co(II) at the interface. Two possible mechanisms are (1) formation of ternary complexes or (2) formation of a new coprecipitate phase that includes Se(IV). The formation of ternary complexes could reduce desorption of Co(II) by binding more strongly when complexed with Se(IV). Bargar et al. (27) proposed that leadchloride complexes form bridge-bonded ternary complexes on goethite where both chloride and lead have bonds with surface iron. Their hypothesis is partly based on the fact that PbCl+ and FeCl2+ have similar formation constants while Pb(II) bidentate surface complexes are known to form on goethite. Again, following the hypothesis presented above that Se(IV) can form strong complexes with aluminum, a similar ternary complex bonding arrangement could exist for the cobalt-selenite system. Potentially, bridged ternary complexes form at the surface after cobalt is sorbed. In this VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1217

case, the apparent greater sorption at lower pH shown in Figure 5b may be reflecting the fact that only the adsorbed cobalt is retained on the surface, while coprecipitated cobalt is dissolved. Another possibility is that selenite exchanges for nitrate in the hydrotalcite structure. Hydrotalcite materials are composed of sheet like layers of mixed metal hydroxides that carry a slight overall positive charge counterbalanced by anions in the interlayers. Selenite is a much stronger binding ligand than nitrate and generally forms slightly soluble compounds with metals. In this regard, a hydrotalcite with selenite as the counterion may be more insoluble or slower to dissolve than a hydrotalcite with nitrate. Macroscopic data alone are not capable of determining the true mechanism underlying the trends in bisorbate data. Direct evidence at a molecular level is needed to differentiate between surface precipitation and coprecipitation and adsorption mechanisms. Clearly, the presence of selenite alters the sorption mechanism of cobalt at both low and high cobalt coverages, potentially in different ways. In this regard, spectroscopic investigation of these systems is warranted. A companion paper describes such a study (43).

Acknowledgments The authors thank Howard Liljestrand, Samuel Traina, and three anonymous reviewers for providing valuable comments on this manuscript. This research was supported by the National Science Foundation, Grant Nos. BES-9625047 and BES-9896214.

Literature Cited (1) Benjamin, M. M.; Bloom, N. S. In Adsorption from Aqueous Solutions; Tewari, P. H., Ed.; Plenum Press: New York, 1981. (2) Schenck, C. V.; Dillard, J. G.; Murray, J. W. J. Coll. Int. Sci. 1983, 95(2), 398-409. (3) Chisholm-Brause, C.; O’Day, P. A.; Brown, G. E., Jr.; Parks, G. A. Nature 1990, 348(6), 528-530. (4) Towle, S. N.; Bargar, J. R.; Brown, G. E., Jr.; Parks, G. A. J. Coll. Int. Sci. 1997, 187, 62-82. (5) Tewari, P. H.; Lee, W. J. Coll. Int. Sci. 1975, 52, 77-88. (6) Katz, L. E.; Hayes, K. F. J. Coll. Int. Sci. 1995, 170, 477-490. (7) Katz, L. E.; Hayes, K. F. J. Coll. Int. Sci. 1995, 170, 491-501. (8) Chisholm-Brause, C. J. Ph.D. Dissertation, Stanford University, 1991. (9) O’Day, P. A.; Brown, G. E.; Parks, G. A. J. Coll. Int. Sci. 1994, 165, 269-289. (10) O’Day, P. A.; Parks, G. A.; Brown, G. E. Clays Clay Miner. 1994, 42, 337-355. (11) Papelis, C.; Hayes, K. F. Colloids Surfaces A 1995, 107, 89-96. (12) Chen, C. C.; Hayes, K. F. Geochim. Cosmochim. Acta 1999, 63(19/ 20), 3205-3215. (13) Huang, C. P.; Lin, T. T. In Adsorption from Aqueous Solutions; Tewari, P. H., Ed.; Plenum Press: New York, 1981. (14) Bayes, C. F.; Mesmer, R. E. Robert E. Krieger Publishing Company, Inc.: FL, 1995.

1218

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

(15) Scheidegger, A. M.; Lamble, G. M.; Sparks, D. L. J. Coll. Int. Sci. 1997, 186, 118-128. (16) Neal, R. H.; Sposito, G. Soil Sci. Soc. Am. J. 1989, 53, 70-74. (17) Hayes, K. F.; Papelis, L.; Leckie, J. O. J. Coll. Int. Sci. 1988, 125, 717-726. (18) Zhang, P.; Sparks, D. L. Environ. Sci. Technol. 1990, 24, 17481856. (19) Papelis, C.; Brown, G. E., Jr.; Parks, G. A.; Leckie, J. O. Langmuir 1995, 11, 2041-2048. (20) Hayes, K. F.; Roe, L. A.; Brown, G. E., Jr.; Hodgson, K. O.; Leckie, J. O.; Parks, G. A. Science 1987, 238, 783-786. (21) Manceau, A.; Charlet, L. J. Coll. Int. Sci. 1994, 168, 87-93. (22) Pardo, M. T.; Guadalix, M. E. Geoderma 1994, 63(1), 43. (23) Venema, P.; Hiemstra, T.; van Riemsdiik, W. H. J. Coll. Int. Sci. 1997, 192, 94-103. (24) Hansmann, D. D.; Anderson, M. A. Environ. Sci. Technol. 1985, 19(6), 544-551. (25) Waychunas, G. A.; Fuller, C. C.; Rea, B. A.; Davis, J. A. Geochim. Cosmochim. Acta 1996, 60(10), 1765-1781. (26) Benjamin, M. M.; Leckie, J. O. Environ. Sci. Technol. 1982, 16(3), 162-170. (27) Bargar, J. R.; Brown, G. E., Jr.; Parks, G. A. Geochim. Cosmochim. Acta 1998, 62(2), 193-207. (28) Smith, R. M.; Martell, A. E. Critical Stability Constants Vol. 4: Inorganic Complexes; Plenum Press: New York, 1989. (29) Gregg. S. J.; Sing, K. S. W. Adsorption, surface Area and Porosity, 2nd ed.; 1982. (30) Davis, James A.; Kent, D. B. Rev. Mineral. 1990, 23, 177-260. (31) Sposito, G. The Environmental Chemistry of Aluminum, 2nd ed.; Lewis Publishers: Boca Raton, FL, 1986. (32) Benjamin, M. M.; Leckie, J. O. J. Coll. Int. Sci. 1981, 79, 209-221. (33) Dzombak, D. A.; Morel, F. M. M. Surface Complexation Modeling: Hydrous Ferric Oxide; John Wiley: New York, 1990. (34) Hayes, K. F.; Katz, L. E. In Physics and Chemistry of Mineral Surfaces; Brady, P. V., Ed.; CRC Press: New York, 1996. (35) Thompson, H. A.; Parks, G. A.; Brown, G. E., Jr. Clays Clay Miner. 1999, 47, 425-438. (36) d’Espinose de la Caillerie, J.-B.; Kermarec, M.; Clause, O. J. Am. Chem. Soc. 1996, 117, 11471-11481. (37) Kannan, S.; Velu, S.; Ramkumar, V.; Swamy, C. S J. Mater. Sci. 1995, 30, 1462-1468. (38) Taylor, R. M. Clay Miner. 1984, 19, 591-603. (39) Gunneriusson, L.; Lo¨vgren, L.; Sjo¨berg, S. Geochim. Cosmochim. Acta 1994, 58(22), 4973-4983. (40) Morris, R. E.; Harrison, W. T. A.; Stucky, G. D.; Cheetham, A. K. J. Solid State Chem. 1991, 94, 227-235. (41) Rai, D.; Mattigod, S. V.; Moore, D. A. Mater. Res. Bull. 1988, 23, 1621-1628. (42) Rai, D.; Felmy, A. R.; Moore, D. A. J. Sol. Chem. 1995, 24(8), 735-752. (43) Boyle-Wight, E. J.; Katz, L. E.; Hayes, K. F. Environ. Sci. Technol. 2002, 36, 1219-1225.

Received for review October 16, 2000. Revised manuscript received November 27, 2001. Accepted November 27, 2001. ES001775A