Adsorption of Antimonate by Gibbsite : Reversibility and the Competitive Effects of Phosphate and Sulfate Soil Chemistry

Soil Sci. Soc. Am. J. 80:1197–1207 doi:10.2136/sssaj2016.04.0129 Open access. Received 29 Apr. 2016. Accepted 16 Aug. 2016. *Corresponding author (messington@utk.edu). © Soil Science Society of America. This is an open access article distributed under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Adsorption of Antimonate by Gibbsite: Reversibility and the Competitive Effects of Phosphate and Sulfate Soil Chemistry

Antimony has been identified as a contaminant of emerging concern due to its suspected acute toxicity and carcinogenic properties in humans (Gebel et al., 1997;Schnorr et al., 1995).Antimony has no known biological function, and both the USEPA (Flynn et al., 2003;USEPA, 2004) and the US Department of Defense (Salatas et al., 2004) have identified Sb as a priority metal pollutant.Further, drinking water guidelines of 20 and 6 mg L −1 have been set by the World Health Organization (WHO, 2006) and the USEPA (USEPA, 2004).Antimony commonly exists in two oxidation states in soil, Sb(III) and Sb(V), with the Sb(III) state displaying greater toxicity, but lower solubility and mobility, than the Sb(V) form (Bencze, 1994;Filella et al., 2002).In the presence of O 2 , Fe(III), and Mn(IV) (as in oxic and suboxic soils), Sb(V) is the thermodynamically stable state (Andreae et al., 1981;Takayanagi and Cossa, 1997).Several studies have reported the rapid oxidation of Sb(III) to Sb(V) in oxic systems and in the presence of Fe(III) and Mn(IV) oxyhydroxides (Belzile et al., 2001;Leuz et al., 2006;Thanabalasingam and Pickering, 1990;Watkins et al., 2006;Xu et al., 2011).Indeed, in oxic soils and sediments, Sb is almost exclusively found in the Sb(V) oxidation state (Mitsunobu et al., 2006;Ritchie et al., 2013;Scheinost et al., 2006;Takaoka et al., 2005).
The aqueous speciation of Sb(V) is dominated by the antimonate species [Sb(OH) 6 -].Antimonate is derived through the hydrolysis of Sb(OH) 5 0 , with a pK a of 2.85.Additional aqueous species, such as ion pairs and complexes with metal cations, are not known to occur, or their occurrence has not been established.Leuz et al. (2006) hypothesized the formation of KSb(OH) 6 0 , but Rakshit et al. (2011) labeled the occurrence of this species unlikely.The retention of Sb(V) in soils has been attributed to adsorption by oxyhydroxides of Al, Fe, and Mn and clay minerals (Filella and Williams, 2012).An important surface functional group in this process is the aluminol group on Al oxyhydroxides and at edge locations on phyllosilicate minerals.Xu et al. (2001) examined Sb(V) adsorption by an activated alumina and observed maximum Sb(V) adsorption to occur in the pH 2.8 to 4.3 range, and, with increasing solution pH, Sb(V) adsorption decreased until insignificant at pH 8.These authors also investigated Sb(V) retention in the presence of competing ligands.They reported that 1 mol L −1 initial concentrations of NO 3 , acetate, and Cl reduced the retention of Sb(V) by <10% from a 0.16 mol L −1 Sb(V) solution, whereas PO 4 and SO 4 reduced Sb(V) adsorption by >60%.Xi et al. (2010a) observed Sb(V) retention by kaolinite to decrease from an adsorption maximum in pH 3.6 to a minimum in pH 9.2 suspensions.They also observed Sb(V) retention to decrease with increasing ionic strength [controlled by Ca(NO 3 ) 2 at pH 6].Xi et al. (2010b) found PO 4 and SO 4 to reduce the retention of Sb(V) by montmorillonite, whereas Biver et al. (2011) found Sb(V) retention by hydrous Al oxide, kaolin-ite, and montmorillonite to decrease with increasing suspension pH and for carbonate to compete with Sb(V) for retention sites.Rakshit et al. (2011Rakshit et al. ( , 2015) ) examined Sb(V) retention by gibbsite and kaolinite as a function of pH and ionic strength and in the presence of PO 4 .As observed by other investigators, Sb(V) retention was strongly dependent on pH, with adsorption decreasing with increasing pH.However, the influence of ionic strength on Sb(V) retention was method dependent.Adsorption isotherms illustrated decreasing Sb(V) retention with increasing ionic strength at pH 6.1, but Sb(V) adsorption envelope studies showed no ionic strength dependence.Excess suspension PO 4 concentrations [100 mmol L −1 PO 4 to 4.11 or 41.1 mmol L −1 Sb(V)] decreased Sb(V) retention throughout the 3 to 9 pH range studied.Rakshit et al. (2011Rakshit et al. ( , 2015) ) interpreted their experimental findings to indicate the inner-sphere surface complexation of Sb(V) by aluminol functional groups.These authors also modeled the Sb(V) adsorption edges using the diffuse layer surface complexation model, which only consider the formation of inner-sphere surface complexes for all ions (including protons and hydroxyls).The inner-sphere surface complexation of Sb(V) is also supported by the spectroscopy findings of Ilgen and Trainor (2012), which suggested the occurrence of both mono-and bidentate inner-sphere Sb(V) surface complexes on hydrous Al oxide, kaolinite, and nontronite aluminol groups.The inner-sphere complexation of Sb(V) by hydrate alumina was also supported by density function theory modeling performed by Mason et al. (2012).
The available literature suggests that the adsorption of Sb(V) by the aluminol functional group occurs in part via the formation of inner-sphere surface complexes.Adsorption envelopes illustrate strong Sb(V) retention in acidic, relative to alkaline, suspensions.Common exchangeable anions, such as NO 3 − and Cl − , have a negligible influence on Sb(V) adsorption, whereas ligands that display inner-sphere surface complexation, such as PO 4 and CO 3 species, reduce Sb(V) adsorption.However, outer-sphere surface complexation is suggested by the dependence of Sb(V) retention on ionic strength.Evaluations by Essington and Stewart (2015) indicate that Sb(V) retention by gibbsite and kaolinite may be dominated by inner-sphere complexation in acidic (pH 5.5) systems, where retention increased with increasing temperature, and outer-sphere complexation in alkaline (pH 8) systems, where retention either decreased or was invariant with increasing temperature.The rationale that underlies this research is the need to understand the processes affecting Sb(V) adsorption by soil minerals and to establish the ability to predict Sb(V) adsorption in the absence or presence of competing ligands.The specific objectives of this research are to (i) characterize the adsorption of Sb(V) by gibbsite as a function of pH and ionic strength, (ii) examine the reversibility of the Sb(V) adsorption process, (iii) quantify the competitive effects of PO 4 and SO 4 on Sb(V) adsorption, and (iv) develop surface complexation models to predict Sb(V) adsorption and to evaluate the ability of these models to predict Sb(V) retention alone and in the presence of competing ligands.

MATERIALs AND METHODs
All chemicals used were obtained from Fisher Scientific (unless noted otherwise) and were of analytical grade or better.All solutions were prepared using CO 2 -free distilled and deionized water (>18W).Stock solutions of 10 mmol L −1 potassium antimonate, phosphate, and sulfate were prepared from KSb(OH) 6 •6H 2 O (Sigma Aldrich), KH 2 PO 4 , and K 2 SO 4 .Potassium nitrate background electrolyte solutions (10 and 100 mmol L −1 ) were prepared from KNO 3 .All pH adjustments were made using certified 0.5, 0.1, or 0.01 mol L −1 solutions of nitric acid (HNO 3 ) or potassium hydroxide (KOH).The solution concentrations of Sb, P, and S were determined with a Spectro CIROS inductively coupled argon plasma-atomic emission spectrometer (ICP-AES) calibrated with commercially available ICP standards.The method detection limits for these elements ranged from 0.01 to 0.1 mg L −1 , depending on element and background electrolyte composition.The solutions were also analyzed for Al to evaluate gibbsite dissolution during the course of the experiments.The concentrations of Al in the equilibrium solutions were either at or below the method detection limit (0.01 mg L −1 ).All pH measurements were performed under CO 2 -free conditions in a N 2 glove box using a calibrated (pH 4, 7, and 10) combination pH electrode.
Gibbsite (alumina hydrate, SF-4) was obtained from Alcan Chemicals.The gibbsite was treated with 0.01 mol L −1 NaOH for 30 min to remove poorly crystalline Al(OH) 3 (Sarkar et al., 1999).The treated gibbsite was repeatedly centrifuge-washed with distilled-deionized water until a suspension pH of 7 was attained.The gibbsite was then freeze-dried and stored until needed.The surface area of the gibbsite was determined by five-point Brunauer-Emmett-Teller (BET) N 2 adsorption isotherms using a Beckman Coulter SA3100 surface area analyzer.The BET surface area of the gibbsite was 5.82 m 2 g −1 .X-ray diffraction (XRD) was used to verify the mineralogy of the SF-4.The XRD patterns were generated using a Bruker AXS D8 Advance with K760 generator with Cu Ka radiation and a Ni filter.The XRD patterns were compared with Joint Committee on Powder Diffraction Standards files for mineral identification.The alumina hydrate was determined by XRD to be monoclinic gibbsite without any detectable impurities with distinct diffraction lines at 0.485 nm (100% intensity) and 0.437 nm (30% intensity).
The zeta potential of gibbsite suspensions was determined by microelectrophoresis as a function of pH and swamping electrolyte using the Zeta-Meter System 4.0 (Zeta Meter).Gibbsite suspensions having a solid/solution ratio of 0.2 g L −1 were prepared in 50-mL polypropylene tubes with the following swamping electrolytes: 10 or 100 mmol L −1 KNO 3 or 10 mmol L −1 of either KSb(OH) 6 , KH 2 PO 4 , or K 2 SO 4 .The suspension pH was adjusted in each tube individually with HNO 3 or KOH to achieve a pH range between 3 and 12 for each swamping electrolyte.After a 24-h equilibration, suspension pH was determined, and the contents of each tube were transferred into a GT-2 electrophoresis cell.A minimum of 10 particles were tracked across a single scale of division for each suspension, and an average zeta potential (mV) reading was obtained.
A series of adsorption edge studies was performed to investigate ligand adsorption reversibility and the competitive effects of PO 4 and SO 4 on Sb(V) retention by gibbsite.The adsorption-desorption studies were performed in duplicate and in 2-L, flat-bottomed, water-jacketed glass reaction vessels with recirculating water held at a constant temperature of 25°C.All experiments were conducted in a CO 2 -free environment in a N 2 -filled glove box to eliminate the influence of CO 2 .Gibbsite was suspended, using a Teflon-coated magnetic stir bar and mechanical stir plate, in a background electrolyte solution of either 10 or 100 mmol L −1 KNO 3 at a solid/solution ratio of 10 g L −1 .Nitric acid was used to lower the suspension pH to 3.5.The suspension was allowed to fully hydrate overnight in the N 2 environment to ensure CO 2 removal.The suspension pH was then increased to pH 9.5 with additions of 0.5 mol L −1 KOH.When the solution pH was stable, a volume of 10 mmol L −1 KSb(OH) 6 was added to yield a 50 mmol L −1 solution of Sb(V).The suspension was allowed to equilibrate for a minimum of 1 h after Sb(V) introduction (preliminary kinetic studies were used to confirm the 1-h adsorption equilibration period).Varying the adsorption equilibration period (0.67-2 h) did not influence antimony retention.Although longer equilibration periods (from several hours to days) have been reported for antimonate adsorption by aluminol functional groups (Ilgen and Trainor, 2012;Xi et al., 2010a), the batch system findings suggest that antimonate adsorption equilibrium is achieved at a period of less than 1 h (consistent with the findings of Kameda et al. [2012]).After the equilibration period, the solution pH was recorded, and a 15-mL aliquot of suspension was removed with a polypropylene syringe.The sample was passed through a 0.45-mm membrane syringe filter and stored under refrigeration for analysis by ICP-AES.Once the pH was recorded and the sample removed, an aliquot of a HNO 3 solution was added to lower the suspension pH by approximately 0.2 pH units.The suspension was again equilibrated at the new pH value and sampled, and the pH was lowered.This process was repeated until a suspension pH of approximately 3.5 was achieved.
Upon the completion of the adsorption titration, the reversibility of the adsorption process (desorption) was investigated by incrementally increasing the suspension pH.As with the adsorption portion of the experiment, the pH of the suspension was increased using aliquots of a KOH solution to achieve stepped pH changes of approximately 0.2 pH units.Small changes in the composition of the equilibrium state are a necessary condition for reversibility to occur (Essington, 2015;Sposito, 1981).At each step, the pH was recorded, and a 15-mL suspension aliquot was removed with a polypropylene syringe.The suspension was passed through a 0.45-mm nylon syringe filter and stored under refrigeration for analysis.Preliminary kinetic studies indicated that an 8-h period was required to achieve desorption equilibrium after each incremental pH change.The adsorption-desorption titration experiments were repeated to examine SO 4 and PO 4 retention behavior, starting with an aliquot of 10 mmol L −1 KH 2 PO 4 or K 2 SO 4 to achieve an initial 50 mmol L −1 solution PO 4 or SO 4 concentration.The concentration of adsorbed Sb(V), SO 4 , or PO 4 was computed as the difference between the mass of ligand added and the mass in solution at equilibrium divided by the mass of solid.
Competitive ligand adsorption was investigated in triplicate using a batch method.Baseline, noncompetitive ligand retention was examined in a N 2 -filled glove box by combining 0.25 g of gibbsite and 25 mL of 10 mmol L −1 KNO 3 in 50-mL polypropylene tubes, resulting in a solid/solution ratio of 10 g L −1 .Aliquots of HNO 3 or KOH were added to each tube to achieve a pH range between 3.5 and 9.5.After pH stabilization of the suspensions, which occurred after shaking for approximately 16 h, a volume of 10 mmol L −1 KSb(OH) 6 , 10 mmol L −1 KH 2 PO 4 , or 10 mmol L −1 K 2 SO 4 was added to achieve an initial ligand concentration of 50 mmol L −1 .The suspensions were then equilibrated on a platform shaker for a minimum of 12 h at ambient temperature (20-22°C).After equilibration, the solid and solution phases were separated by centrifugation filtration through a 0.45-mm nylon syringe filter.The solutions were analyzed for pH and refrigerated until analyzed by ICP-AES.Noncompetitive ligand adsorption in 100 mmol L −1 KNO 3 was also examined using the above procedure.
Competitive ligand adsorption was determined using methods similar to the noncompetitive batch procedure for the 10 mmol L −1 KNO 3 systems.Three scenarios of ligand addition to the pH stabilized gibbsite suspensions were used, and all scenarios involved equimolar initial ligand concentrations.In the first scenario, gibbsite suspensions were equilibrated with Sb(V) (described above), followed by the additions of either SO 4 or PO 4 and an additional 12-h equilibration.The second scenario involved the initial adsorption of either SO 4 or PO 4 , followed by the addition of Sb(V).The third scenario involved the simultaneous additions of either Sb(V) and SO 4 or Sb(V) and PO 4 and a 12-h equilibration.The initial concentrations of Sb(V), PO 4 , and SO 4 in the competitive systems were designed to be 50 mmol L −1 (through the addition of the appropriate 10 mmol L −1 ligand salt solution).Actual concentrations were determined through the analysis of control samples (no solid).The difference between the initial mass of ligand added to each tube and the equilibrium suspension mass of the ligands was defined as the adsorbed mass.
The 2-pK a formulation of the triple-layer model (TLM) was used to describe ligand adsorption.Davis and Kent (1990) and Goldberg (1992) offer a detailed description of the TLM, and recent examples of the application of the 2-pK a TLM to describe ligand adsorption may be found in Goldberg (2014b) and Essington and Vergeer (2015).The goal of the modeling was to identify surface complexation reactions for each ligand with the least number of surface species (simplest model) that describes the adsorption data (i.e., generates the lowest residual error values).The model must be applicable to both ionic strength conditions and be consistent with the mechanistic interpretations of the experimental data.The gibbsite surface properties (e.g., specific surface, site density, capacitances) and the protonation and deprotonation (formation of ºAlOH 2 + and ºAlO − ), outersphere complexation of counter ions (formation of ºAlO − -K + and ºAlOH 2 + -NO 3 − ), and aqueous speciation reactions used in the chemical modeling are given in Tables 1 and 2. Inner-and outer-sphere surface complexation reactions that describe the adsorption of Sb(V), PO 4 , and SO 4 by gibbsite were developed and examined in the manner described by Essington and Vergeer (2015).The computer software FITEQL 4.0 (Herbelin and Westall, 1999) was used to fit the adsorption data and to optimize the adsorption equilibrium constants for the proposed surface complexation reactions.For the titration systems, chemical models were optimized simultaneously using the adsorption data from both the 10 and 100 mmol L −1 ionic strength.The models that generated the lowest goodness-of-fit parameter (weighted sum of squares of residuals divided by the degrees of freedom, V Y ) were then reoptimized for the 10 mmol L −1 batch systems and used to describe competitive ligand adsorption.

REsULTs AND DIsCUssION
Zeta potential curves for gibbsite in the swamping electrolyte (KNO 3 ) indicate that the isoelectric point (IEP) (and the point of zero charge, pH pzc ) of gibbsite is 10.55 (Fig. 1).This is also the common intersection point for the zeta potential measurements in 10 and 100 mmol L −1 KNO 3 .The measured gibbsite IEP is slightly higher than the 7.8 to 10.4 pH pzc range compiled by Karamalidis and Dzombak (2010) but is within the 8.7 to 11 range reported by Adekola et al. (2011).An increase in the swamping electrolyte concentration results in a decreasing zeta potential as the pH is decreased below the IEP.Increased concentrations of electrolyte in the diffuse layer tend to shield the particle charge, decreasing the extent of charge influence in the solid-solution interface and decreasing the response (particle movement) in an electric field (Yu, 1997).
The presence of 10 mmol L −1 K 2 SO 4 does not appreciably change the IEP of gibbsite relative to the KNO 3 systems, which 4 ‡ S T , total site concentration, mmol L −1 0.3866 § C 1 , inner-layer capacitance, F m −2 0.9 ‡ C 2 , outer-layer capacitance, F m −2 0.2 ¶ a, suspension density, g L −1 10 Background electrolyte, mmol L −1 KNO 3 10 or 100 † BET-N 2 adsorption.‡ An average of 8 nm −2 singly coordinated ºAlOH sites on the edge surface and 0 nm −2 on the planar surface, after Hiemstra and Van Riemsdijk (1999).§ The total concentration of singly-coordinated sites for gibbsite computed using S T = (n s 10 18 aS A )/A N , where A N is the Avogadro constant.¶ Sahai and Sverjensky (1997).
may be due to the absence of SO 4 adsorption at the IEP.However, there is a negative shift in the zeta potential at pH values below the IEP in SO 4 relative to NO 3 systems.The less positive zeta potential values may indicate that the electrostatic surface complexes of SO 4 provide greater negative charge to the near-surface region bounded by the particle shear plane, relative to the background electrolyte (divalent SO 4 2− vs. monovalent NO 3 − ), and that the electrostatic interaction is sufficiently strong for SO 4 to be carried inside the shear plane ( Juang and Wu, 2002).This result has also been interpreted to indicate the covalent bonding of SO 4 , even though a charge reversal is not observed (Pommerenk and Schafran, 2005).The IEP of gibbsite shifts to the 4.6 to 4.9 range when reacted with PO 4 .This is macroscopic evidence of the inner-sphere complexation of PO 4 by gibbsite (or surface precipitation), as is often reported in the literature (Karamalidis and Dzombak, 2010).This finding is also consistent with the strong PO 4 adsorption behavior and with the lack of an ionic strength effect on PO 4 retention described below.A similar result is observed in the Sb(V) system.Within the pH range studied (pH >5), the zeta potential is negative (averaging −51 mV), but unlike the PO 4 system a charge reversal is not observed.As with PO 4 , strong inner-sphere surface complexation, or the formation of a surface precipitate, may be inferred.
The adsorption of Sb(V) by gibbsite is dependent on solution pH and ionic strength (Fig. 2a).Antimonate adsorption is at a relative maximum in strongly acidic environments (pH <5) and decreases with increasing pH to a minimum in strongly alkaline conditions (pH >9).Increasing the ionic strength (KNO 3 concentration) depresses Sb(V) retention by gibbsite, primarily in the pH <7 range.The pH dependence of Sb(V) adsorption by the aluminol group is similar to that observed by other investigators (Biver et al., 2011;Ilgen and Trainor, 2012;Rakshit et al., 2011Rakshit et al., , 2015;;Xi et al., 2010a;Xu et al., 2001).The decrease in Sb(V) adsorption with increasing ionic strength is not consistent with the recent findings of Rakshit et al. (2011Rakshit et al. ( , 2015)), who observed that the Sb(V) adsorption envelope of gibbsite and kaolinite was unaffected by a 100-fold change (from 0.001 to 0.1 mol L −1 KCl) in ionic strength and concluded that Sb(V) retention occurred via inner-sphere mechanisms.However, the adsorption isotherm results of Rakshit et al. (2011) did illustrate an ionic strength dependence of Sb(V) adsorption by gibbsite at pH 6.1.Further, Xi et al. (2010a) observed Sb(V) adsorption by kaolinite to decrease with increasing ionic strength [from 0.01 to 0.06 mol L −1 Ca(NO 3 ) 2 ] at pH 6.Our findings suggest that there is weak electrostatic character (outer-sphere adsorption) to the Sb(V) retention mechanism.This is evidenced by the reduction in retention when the background electrolyte concentration is increased from 0.01 to 0.1 mol L −1 KNO 3 .The findings are also consistent with the adsorption isotherm results of Essington and Stewart (2015).They observed Sb(V) adsorption by gibbsite to increase with increasing temperature in pH 5.5 systems, suggesting the predominance of an inner-sphere surface complexation mechanism.However, adsorption decreased with increasing temperature in pH 8 systems, suggesting the predominance of an outer-sphere mechanism.
The desorption data indicate that there is also a strong bonding component (inner-sphere adsorption) to the Sb(V) retention mechanism, particularly in acidic systems, consistent with the spectroscopic findings of Ilgen and Trainor (2012) (Fig. 2a).As pH is increases from the minimum value of 3.5, Sb(V) does not readily desorb from the gibbsite surface until the solution pH exceeds approximately 6.5 (desorption is hysteretic).At pH values above 6.5, the slopes of the adsorption and desorption envelopes are similar, indicating that desorption becomes reversible as solution alkalinity increases (nonhysteretic).The reversibility of Sb(V) adsorption-desorption in the alkaline pH range  was examined further by truncating the adsorption edge study to begin desorption at a pH of approximately 7 (Fig. 2a, insert).In this case, the adsorption and desorption envelopes are superimposed, indicating that the Sb(V) retention process is reversible in the alkaline range.
The adsorption-desorption data suggest that Sb(V) is retained by both strong and weak reaction mechanisms at the gibbsite surface.Strong, inner-sphere complexation mechanisms (ligand exchange) appear to predominate in acidic environments, whereas a weak outer-sphere mechanism (anion exchange) predominates in alkaline (although adsorption reversibility does not necessarily connote an outer-sphere mechanism).It is not uncommon for the predominant mechanism of ligand retention (anion vs. ligand exchange) to differ as a function of soil solution properties, such as pH.For example, ligand exchange mechanisms may predominate in neutral to acidic systems, whereas anion exchange predominates in alkaline soils.This is the case for arsenate, molybdate, and sulfate retention by Fe-and Aloxyhydroxides (Catalano et al., 2008;Goldberg, 2014aGoldberg, , 2014b;;Mansour et al., 2009) and for selenite and selenate by gibbsite and phyllosilicates (Goldberg, 2014b(Goldberg, , 2014c)).
The retention of SO 4 and PO 4 by gibbsite was examined for comparison to Sb(V) retention because the mechanisms involved in the adsorption of SO 4 and PO 4 by hydrous Al oxyhydroxides have been well established.Phosphate is retained by inner-sphere surface complexation mechanisms throughout a broad pH range (Karamalidis and Dzombak, 2010).Phosphate is strongly retained by gibbsite (Fig. 2b) and shows no dependence on ionic strength, an observation that is consistent with the ligand exchange adsorption mechanism.Further, PO 4 retention exceeds that of both Sb(V) and SO 4 .Phosphate adsorption is reversible in strongly alkaline systems, owing to the competitive effects of the hydroxide ion.The reversibility of PO 4 adsorption in slightly alkaline to acidic systems could not be evaluated due to the strong and complete retention in this pH range.
Except in strongly acidic systems (pH <4), SO 4 is retained predominantly by outer-sphere mechanisms by the aluminol functional group (He et al., 1997;Karamalidis and Dzombak, 2010).Sulfate adsorption by gibbsite increases with decreasing pH, with a strong dependence on ionic strength (Fig. 2c).In addition, SO 4 adsorption is reversible throughout the entire pH range studied (nonhysteretic; the adsorption and desorption envelopes overlap), differing from the desorption behavior of Sb(V).These findings are consistent with an anion exchange mechanism and the weak, electrostatic retention of SO 4 .Sulfate retention is also depressed, relative to that of Sb(V), throughout the entire pH range studied and particularly in the higher ionic strength systems.Despite their similar acid pK a values (1.99 for HSO 4 − dissociation and 2.85 for Sb(OH) 5 0 hydrolysis) (Table 2), the retention of Sb(V) by gibbsite is greater than that of SO 4 , is less affected by ionic strength, and is hysteretic.
The adsorption of Sb(V), SO 4 , and PO 4 by gibbsite in the batch systems (Fig. 3) is similar to that observed in the continuous titration systems (Fig. 2).Ligand adsorption increases with decreasing pH, the retention of Sb(V) and SO 4 decreased with increasing ionic strength, and the adsorption of PO 4 is not influenced by the ionic media.Sulfate adsorption is less than that of Sb(V) throughout the pH 3 to 10 range irrespective of ionic strength, and PO 4 adsorption is complete at pH values below approximately 7.
The inclusion of PO 4 as a competing ligand reduces the retention of Sb(V) by gibbsite in 10 mmol L −1 KNO 3 systems (Fig. 3a).The addition of PO 4 shifts the Sb(V) adsorption envelope to lower pH values, similar to the response observed by Rakshit et al. (2015).In addition, this shift is not particularly sensitive to the order of ligand addition.However, Sb(V) is slightly more competitive with PO 4 when preadsorbed, as indicated by the greater amount of Sb(V) retained relative to the other competitive systems.Phosphate adsorption by gibbsite is not influenced by Sb(V) (Fig. 3b).
The inclusion of SO 4 as a competing ligand reduces the retention of Sb(V) by gibbsite in the 10 mmol L −1 KNO 3 systems (Fig. 3c).The addition of SO 4 after the preadsorption of Sb(V) and the addition of SO 4 and Sb(V) simultaneously result in similar Sb(V) adsorption envelopes, with an approximate 10% reduction in Sb(V) adsorption by gibbsite at pH <6.5 [relative to noncompetitive Sb(V) adsorption].However, the Sb(V) adsorption envelope (pH > 6.9) is not affected by SO 4 .Preadsorbed SO 4 has a strong influence on Sb(V) adsorption throughout the pH range studied, with an approximate 20% reduction in Sb(V) retention in acidic systems.Sulfate adsorption is generally reduced in all competitive systems relative to adsorption in the absence of Sb(V) (Fig. 3d).However, the order of ligand addition did not significantly affect SO 4 retention.
The Sb(V) adsorption envelope and desorption hysteresis, coupled with the effects of Sb(V) on gibbsite surface charging, indicate that Sb(V) retention by gibbsite occurs via a combination of inner-and outer-sphere complexation mechanisms.Specifically, Sb(V) retention predominantly occurs via an outersphere mechanism [formation of ºAlOH 2 + -Sb(OH) 6 − ] in alkaline systems, with increasing inner-sphere complexation character as pH decreases into the acidic range.Spectroscopic studies indicate that retention of Sb(V) by hydrous aluminum oxide, and other adsorbents with ºAlOH surface functionality (e.g., aluminosilicates), proceeds via mono-and bidentate inner-sphere surface complexation processes [formation of ºAlOSb(OH) 5 − and (ºAlO) 2 Sb(OH) 4 − ] under acidic (pH 5) conditions (Ilgen and Trainor, 2012).These studies also suggest that the bidentate complex is the predominant inner-sphere species.
The ligand adsorption envelope data generated from the continuous titration beaker systems was used to develop the chemical models for ligand adsorption by gibbsite.Several models were evaluated for their ability to describe the Sb(V) adsorption edge: outer-sphere surface complexation [ºAlOH sphere/inner-sphere monodentate or outer-sphere/inner-sphere bidentate mechanisms.Using the triple-layer surface complexation model (SCM), coupled with the 2-pK a approach, the adsorption of Sb(V) in 10 and 100 mmol L −1 KNO 3 is well pre-dicted (V Y = 1.49) in the pH 3 to 10 range (Fig. 2a) using either the outer-sphere/inner-sphere monodentate or outer-sphere/ inner-sphere bidentate models (Table 3).The outer-sphere model, which considers only ºAlOH 2 + -Sb(OH) 6 − formation, also predicts the adsorption edge data (log K int = 11.61;V Y = 1.71).However, Sb(V) adsorption is slightly under-predicted at pH <4 (data not shown).Models involving only the formation of inner-sphere surface complexes could not predict the influence of ionic strength on Sb(V) adsorption and were not considered further.The SCM-predicted, adsorbed Sb(V) surface speciation is dominated by the ºAlOH 2 + -Sb(OH) 6 − complex throughout the pH 3 to 10 range (Fig. 4a and 4b).Only in strongly acidic systems (pH <6) is the inner-sphere ºAlOSb(OH) 5 − [or (ºAlO) 2 Sb(OH) 4 − )] species predicted to occur in significant, but minor, concentrations [10-15% of the total adsorbed Sb(V) at pH 3].This value is lower than the ~30% that may be assigned to the irreversible adsorbed fraction of Sb(V) that is estimated from the desorption envelope (Fig. 2a) and is assumed to represent inner-sphere complexation.
Although the predicted surface speciation of adsorbed Sb(V) is generally supported by the macroscopic evidence, the SCM model-predicted minor importance of the inner-sphere Sb(V) surface complex is inconsistent with the spectroscopic findings of Ilgen and Trainor (2012).Surface spectroscopy indicates the average bonding environment; thus, findings reported by Ilgen and Trainor (2012) support the predominance of the inner-sphere surface complexation of Sb(V) at pH 5.5 by a hydrous aluminum oxide.Our macroscopic findings also support this but further illustrate the importance of the outer-sphere complexation of Sb(V), particularly in slightly acidic to alkaline systems.The surface characteristics of the hydrous aluminum oxide and the surface loading of Sb(V) used by Ilgen and Trainor (2012) [S A of 196 m 2 g −1 , solid/solution ratio of 2.5 g L −1 , and an initial Sb(V) concentration of 100 mmol L −1 ] differ from those of gibbsite (Table 1).However, the Sb(V) adsorption density on gibbsite at pH 5.5 of 0.26 mmol m −2 is nearly identical to that on the hydrous aluminum oxide (0.29 mmol m −2 ).The only area of contradiction occurs in the surface complexation modeling of the adsorption envelopes in acidic systems.The strong influence of Sb(V) adsorption on ionic strength yields a modeled adsorption envelope that is dominated by the outer-sphere complexation of Sb(V).In addition to the assumed nature of the solidsolution interface embodied in the TLM, the TLM results are dependent on a number of literature-derived and fixed param-eters (e.g., surface site density, inner-and outer-layer capacitance, and counter-ion surface association constants).The FITEQL convergence problems only allowed for minor modifications to these parameters.However, within those limits, the model-predicted predominance of the outer-sphere Sb(V) surface complex was not altered.
Phosphate displays strong adsorption by gibbsite, with nearly complete removal of PO 4 at pH values less than approximately 7 (Fig. 2b).Phosphate adsorption is independent of ionic strength and shifts the gibbsite IEP from 10.55 in KNO 3 systems to 4.8 in KH 2 PO 4 (Fig. 1).These results indicate that PO 4 is retained primarily by inner-sphere surface complexation mechanisms.However, SCMs that involved only inner-sphere surface species did not adequately describe the adsorption envelopes or cause convergence problems in the FITEQL program.Johnson et al. (2002) and Van Emmerik et al. (2007) reported that PO 4 formed monodentate inner-sphere surface complexes on g-Al 2 O 3 and gibbsite and that outer-sphere complexation contributed to adsorption in alkaline systems (pH >8).Thus, the adsorption of PO 4 by gibbsite was described by considering an outer-sphere ºAlOH 2 + -HPO 4 2− species and a monodentate inner-sphere ºAlOPO 2 OH − species (Table 3).Using this model, the adsorption of PO 4 in 10 and 100 mmol L −1 KNO 3 is well predicted (V Y = 0.84) in the pH 3 to 10 range (Fig. 2b).The predicted surface speciation of adsorbed PO 4 is dominated by the outer-sphere ºAlOH 2 + -HPO 4 2− species throughout the pH 3 to 10 range (Fig. 4b).Although the concentration of the inner-sphere ºAlOPO 2 OH − species increases with decreasing pH, this complex generally does not dominate at any pH.The adsorption of SO 4 by gibbsite is strongly influenced by solution ionic strength and is reversible throughout the pH Table 3. surface complexation models used to describe the adsorption of ligands by gibbsite as a function of pH and ionic strength using the triple-layer model formulation for batch and titration systems.Common logarithms of the intrinsic surface complexation constants (log K int values) ± sD optimized using FITEQL and suspension parameters described in Tables 1 and 2 and the ligand adsorption edge data presented in Fig. 2 and 3. 1.528 2.249 † For titration systems, the log K int values were optimized simultaneously using ligand adsorption edge data for both the 10 and 100 mmol L −1 KNO 3 systems.‡ Weighted sum of squares of residuals divided by the degrees of freedom.range studied (Fig. 2c).Sulfate adsorption does not alter the IEP of gibbsite, although zeta potentials are less positive relative to the indifferent KNO 3 electrolyte alone (Fig. 1).These experimental findings indicate that SO 4 retention proceeds primarily through an outer-sphere surface complexation mechanism, an interpretation that is supported by the available literature (Goldberg, 2010;He et al., 1996He et al., , 1997)).Sulfate adsorption by gibbsite was successfully modeled by considering only outersphere complexation and the formation of ºAlOH 2 + -SO 4 2− data (log K int = 9.48; V Y = 1.53) (Table 3; Fig. 2c).
The competitive Sb(V)-PO 4 and Sb(V)-SO 4 adsorption envelope data were modeled using the SCMs for the single-ligand systems described above and in Table 3.The intrinsic constants obtained from the single-adsorbate titration 10 and 100 mmol L −1 KNO 3 systems were reoptimized for the batch 10 mmol L −1 KNO 3 systems before they were applied to predict competitive ligand retention (Table 3).These reoptimized constants differed slightly [log K int = 11.28 vs. 11.44 for ºAlOH 2 + −Sb(OH) 6 − formation and −2.56 vs. −1.04 for ºAlOSb(OH) 5 − ] due to the differences in the input data sets [Sb(V) adsorption under two ionic strength conditions vs. one ionic strength].In general, the intrinsic constants in SCMs developed for single-adsorbate systems require reoptimization to adequate describe ligand retention in competitive systems (Essington and Anderson, 2008;Goldberg, 2010).Further, reoptimization was required to describe competitive Sb(V) and PO 4 adsorption by birnessite (Essington and Vergeer, 2015).The adsorption envelopes for Sb(V) and PO 4 on gibbsite are well predicted (V Y = 6.234) without reoptimization (Fig. 3a  and 3b).However, the envelopes for Sb(V) and SO 4 are underpredicted (Fig. 3c and 3d), although the V Y = 16.05 is within the V Y <20 range cited by Herbelin and Westall (1999) to represent an adequate description of the adsorption envelopes when using the default error parameters in FITEQL.

CONCLUsIONs
The adsorption of Sb(V), PO 4 , and SO 4 by gibbsite was investigated as a function of pH and ionic strength using two experimental approaches: continuous pH titration and batch.Ligand desorption was also investigated using the continuous titration method.Both experimental approaches showed an Sb(V) adsorption envelope that was dependent on both pH and ionic strength.Antimonate adsorption decreased from a maximum in strongly acidic suspensions to a minimum in strongly alkaline; adsorption decreased with increasing ionic strength (controlled by KNO 3 ).Antimonate desorption was hysteretic in acidic suspensions but was reversible in pH >7 systems.The gibbsite zeta potential was negative throughout the pH 5 to 11 range in the presence of Sb(V), and Sb(V) adsorption shifted the IEP from 10.55 in KNO 3 to <5.The Sb(V) adsorption envelope, desorption, and zeta potential findings suggest that the retention of Sb(V) occurs by a combination pH-dependent inner-sphere and outer-sphere mechanisms.
The adsorption envelopes for PO 4 and SO 4 were also similar in both the titrations and batch systems.The adsorption of PO 4 was strong, particularly in the pH <7 range, independent of ionic strength, and reversible in alkaline systems.Phosphate adsorption decreased the IEP of the gibbsite to <5, indicating inner-sphere complexation.Sulfate adsorption decreased with increasing pH and ionic strength, and desorption was reversible throughout the pH 3 to 10 range.Further, the gibbsite IEP did not shift to a lower value in the presence of SO 4 .These experimental findings are consistent with the outer-sphere surface complexation of SO 4 by the aluminol group.
The competitive adsorption of Sb(V), PO 4 , and SO 4 by gibbsite was examined in batch systems using equimolar additions of ligand pairs [Sb(V)-PO 4 or Sb(V)-SO 4 ].The order or ligand addition on adsorption was also evaluated.Phosphate effectively competes with Sb(V) for the aluminol group, reducing Sb(V) retention throughout the pH range studied.However, Sb(V) did not influence PO 4 retention.Sulfate also competed with Sb(V) for adsorption sites, but only in acidic systems, whereas the inclusion of Sb(V) generally reduced SO 4 adsorption throughout the 3 to 10 pH range.In general, the order of ligand addition did not have a large impact on the adsorption envelopes.
An SCM was developed using the TLM formulation to predict Sb(V) adsorption by gibbsite.The Sb(V) adsorption envelopes at the two ionic strength levels were predicted by imposing a combination of inner-(mono-or bidentate) and outersphere mechanisms.The outer-sphere complex was predicted to dominate the surface speciation of Sb(V), whereas the innersphere complex was a minor species and only significant in acidic suspensions.The PO 4 adsorption envelopes were successfully modeled by using a combination of inner-and outer-sphere surface complexes, and the SO 4 envelope was predicted using only outer-sphere complexation.When applied to the competitive adsorption systems, the SCMs provided adequate descriptions of ligand retention without the need for reoptimization.

Fig. 1 .
Fig. 1.The zeta (z) potential of gibbsite particles as a function of pH and ionic environment.IEP, isoelectric point.

Fig. 2 .
Fig. 2. The adsorption (titration from high to low pH) and desorption (titration from low to high pH) of (a) sb(V), (b) PO 4 , and (c) sO 4 by gibbsite as a function of pH and ionic strength in the titration systems.The initial ligand concentrations are 50 mmol L −1 .The solid lines represent triple-layer model predicted adsorption.The insert in (a) illustrates sb(V) adsorption and desorption behavior when desorption was initiated at approximately pH 7. The sE about each data point is encompassed within the symbol.

Fig
Fig. 3. Noncompetitive ligand adsorption by gibbsite as a function of ionic strength (Is) (controlled by 10 and 100 mmol L −1 KNO 3 ) and competitive adsorption in 10 mmol L −1 KNO 3 of (a) sb(V) and (b) PO 4 in equimolar sb(V)-PO 4 and in (c) sb(V) and (d) sO 4 in equimolar sb(V)-sO 4 by gibbsite as a function of pH and order of ligand addition in batch systems.The initial ligand concentrations are 50 mmol L −1 .The lines represent the triple-layer model predicted adsorption in noncompetitive (solid line) and competitive (dashed line) in 10 mmol L −1 KNO 3 systems.The sE about each data point is encompassed within the symbol.

Fig. 4 .
Fig. 4. surface species for (a) sb(V) and (b) PO 4 predicted by the triple layer model for ligand adsorption by gibbsite as a function of pH and ionic strength in the titration systems.