Stability and structure of pentavalent antimony complexes with aqueous organic ligands

Chemical Geology 292–293 (2012) 57–68

Contents lists available at SciVerse ScienceDirect

Chemical Geology journal homepage: www.elsevier.com/locate/chemgeo

Research paper

Stability and structure of pentavalent antimony complexes with aqueous organic ligands Marie Tella a, b,⁎, Gleb S. Pokrovski b a b

CIRAD, UPR Recyclage et risque, F-34398, Montpellier, France Géosciences-Environnement-Toulouse, GET, UMR 5563 of CNRS, University of Toulouse III, 14 avenue Edouard Belin, F-31400, Toulouse, France

a r t i c l e

i n f o

Article history: Received 15 February 2011 Received in revised form 4 November 2011 Accepted 6 November 2011 Available online 18 November 2011 Editor: J. Fein Keywords: Sb Humic Complexation XAS Potentiometry Dialysis

a b s t r a c t Despite the rapidly growing concern about antimony pollution of waters and soils, the effect of organic matter on the behavior of this toxic trace element is poorly understood because of a lack of data on Sb V-organic ligand interactions in aqueous solution. We used in situ potentiometry and X-ray absorption spectroscopy (XAS) to measure in aqueous solution at ambient conditions the stability and structure of aqueous complexes formed by pentavalent antimony (Sb V) with low molecular weight organic ligands, such as carboxylic acids (acetic, adipic, malonic, lactic, oxalic, citric and salicylic), phenols (catechol), polyols (xylitol and mannitol), and amines (glycine), which have O- and N-functional groups typical of natural organic matter. Potentiometric titrations from pH 2 to 10 demonstrate negligible Sb V complexing with amine and carboxylic acids with single functional group (acetic acid) or non-adjacent functional groups (adipic acid). In contrast, Sb V forms stable complexes with poly-carboxylic, hydroxy-carboxylic acids, and with aliphatic and aromatic polyol ligands in the pH range typical of natural waters. XAS measurements show that in these species the Sb V atom has a distorted octahedral geometry composed of 6 oxygen atoms forming a five- or six-membered bidendate cycle. Stability constants of Sb V-organic complexes, generated for the first time in this study, were used to model Sb V binding with natural humic acids containing the same functional groups as those used in this work. Our predictions of Sb V binding with natural humic acids indicate that in an aqueous organic-rich solution of 1 μg L− 1 Sb and 20 mg L − 1 dissolved organic carbon (DOC) up to 40% of total Sb binds to aqueous organic matter via carboxyl and hydroxy-carboxyl functional groups at pH ≤ 4, whereas at neutral-to-basic pH this amount does not exceed 5%. These estimations are in agreement with direct dialysis measurements conducted with a purified commercial humic acid. The low affinity of Sb V to organic matter at near-neutral pH contrasts with that of Sb III whose organic complexes may account up to 80% of total Sb in DOC-rich waters. The large differences in Sb III versus Sb V binding to organic matter may be used for tracing in organic-rich sediments and waters the two main Sb oxidation states, which have different toxicities for aquatic organisms. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Antimony is a trace element on Earth (300 μg kg− 1 in the continental crust, Wedepohl, 1995) with concentrations in unpolluted natural waters typically below 1 μg L − 1 (e.g., Filella et al., 2002). However, natural antimony abundances are perturbed by anthropogenic factors, which largely exceed natural fluxes (Nriagu, 1989). Pollution from industrial waste and mining activities may yield aqueous Sb concentrations up to ~100 μg L − 1 (e.g., Ulrich, 1998). Such concentrations are ~20 times greater than the drinking water limits for Sb recommended by European and American Agencies (5 μg L− 1, European Union Council, 1998; 6 μg L − 1, USEPA, 1999). Despite this growing pollution ⁎ Corresponding author at: CEREGE, Université Aix Marseille, CNRS, CDF, IRD, BP 80, 13545 Aix en Provence, France. Tel.: +33 4 42 97 15 64; fax: +33 4 42 97 15 59. E-mail address: [email protected] (M. Tella). 0009-2541/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.chemgeo.2011.11.004

concern about antimony, studies of its behavior in natural aquatic environments are very scarce, and the effects on Sb mobility of different ligands, mineral and organic surfaces or particles and live organisms in waters are insufficiently known. The present contribution is aimed at better assessing the role of dissolved organic matter on the fate of pentavalent Sb in aquatic systems. Antimony occurs in surface environments mainly in two oxidation states, +3 and +5. Pentavalent antimony, Sb V, the less toxic species (Gurnani et al., 1994), prevails in oxygenated natural waters but may also be persistent under reducing conditions where Sb III is the thermodynamically stable form (e.g., Pokrovski et al., 2006). The acid–base equilibrium of Sb V major hydroxide species, SbðOHÞ05 and SbðOHÞ− 6 , are well documented (Fig. 1; Baes and Mesmer, 1976; Accornero et al., 2008), as well as their adsorption on Mn, Fe and Al oxy-hydroxide mineral surfaces (e.g., Thanabalasingam and Pickering, 1990; Belzile et al., 2001; Johnson et al., 2005; Leuz et al.,

58

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

X-ray absorption spectroscopy (XAS) measurements, provided first direct evidence that Sb V is capable of complexing with polyfunctional organic ligands like (oxy-)carboxylic acids and polyols in aqueous solution. In the present work, we significantly extend these first data by conducting systematic potentiometric pH experiments and XAS measurements, which allowed us to refine the existing data and to determine the stability and structure of aqueous Sb V complexes with a large number of representative low molecular weight organic ligands possessing functional groups similar to those of NOM (Table 1). We report here a first set of values of formation constants for a large number of Sb V-organic complexes, which allow the effect of dissolved NOM on Sb transport in natural waters and within polluted sites to be estimated. 2. Material and methods 2.1. Potentiometric pH measurements

Fig. 1. Distribution of inorganic species of SbIII and SbV in aqueous solution as a function of pH at 25 °C and 1 bar according to Baes and Mesmer (1976) and Accornero et al. (2008).

2006). In contrast, only few studies have explored Sb associations with natural organic matter (NOM) mostly from soil environments (e.g., Ettler et al., 2007). In aquatic systems, a part of total antimony was found as complexes with low-weight organic ligands in the dissolved fraction (e.g., Deng et al., 2001; Pokrovsky and Schott, 2002). A recent systematic study of the stability and molecular structure of Sb III complexes with various low molecular weight hydroxycarboxylic ligands in aqueous solution (Tella and Pokrovski, 2009) has significantly extended the existing databases of stability constants of Sb III-organic species (Martell et al., 2004; Filella and May, 2005) and provided the first structural data on Sb III in its organic complexes. These new results indicate that poly-functional carboxylic acids may be important binders for trivalent antimony in the natural pH range, and thus affect both the mobility and toxicity of Sb III in the environment. In contrast, published data on stabilities and structures of Sb V-organic complexes exist only for oxalic acid (Tella and Pokrovski, 2008), to the best of our knowledge. As a result, the existing estimations of Sb V binding by NOM in natural systems are inconsistent and controversial. For example, no SbV retention by natural humus was found by Pilarski et al. (1995), and desorption of SbV from humic acid colloids upon oxidation of adsorbed Sb III was observed by Buschmann and Sigg (2004). On the other hand, a major part of both Sb III and Sb V was found to be bound to humic acids in contaminated soils (Steely et al., 2007). It is clear that SbV complexing with organic matter requires further investigation because the identity of binding sites and stoichiometry and structure of complexes remain largely uncertain. Such work is however complicated by the fact that NOM, dominated by humic and fulvic acids, presents extremely complex poly-functional structure that precludes assessing unambiguously the sorption mechanisms or nature of aqueous complexes and identifying chemical sites (potentially numerous) responsible for SbV binding. As a first step, antimony interactions with NOM can be assessed from accurate data on simple organic ligands that possess moieties representative of humic substances (e.g., Linder and Murray, 1987; Pokrovski and Schott, 1998; Tella and Pokrovski, 2009). Our recent preliminary work of Sb V-organic complexing (Tella and Pokrovski, 2008), using a small number of ligands and limited pH and

Two types of pH measurements were performed. ‘Classical’ titrations of organic-bearing solutions with standardized NaOH were conducted with and without Sb. These measurements permit the coverage of a wide range of H+ concentrations, enabling determination of pH intervals in which complexation occurs. The second type of measurements consisted of titrations of organic-bearing solutions by a concentrated KSb(OH)6-bearing solution in a narrow pH range. This technique, more sensitive than a NaOH titration, enables the detection of very small pH changes (≤0.01 pH units) induced by the formation of Sb V-organic species. Titrations by NaOH were carried out on experimental solutions of acetic, adipic, lactic, malonic, oxalic, citric, salicylic acids, catechol, xylitol, mannitol, and glycine (0.002–0.01 m1 of organic ligand and ~0.0015 m Sb). Titrations by KSb(OH)6 were performed on aqueous acetate, adipate, citrate and oxalate-bearing solutions (0.002, 0.01 and 0.05 m of organic ligand and up to 0.0015 m Sb). Solutions were prepared from the corresponding organic reagents and potassium hexahydroxoantimonate(V) salt, KSb(OH)6. Ionic strength was fixed to 0.1 using KNO3. Measurements of pH were performed at 25 °C with a high resolution (0.1 mV) pH-meter equipped with a combination glass electrode that was fixed in the Teflon cover of a ~60 cm3 water-thermostated (±0.2 °C) Pyrex-glass potentiometric cell. Before each titration the electrode was calibrated on the activity scale with 0.01 M HCl (pH25 °C = 2.04) and the NIST phthalate (pH25 °C = 4.01), phosphate (pH25 °C = 6.86) and borate (pH25 °C = 9.18) buffer solutions. For NaOH titrations, a DL70ES Mettler Toledo automatic titrator, equipped with an automatic stirrer, was used to inject NaOH aliquots (5–50 μL) into the experimental solution, whereas in KSb(OH)6 titrations weighed injections were performed manually with a syringe (~0.3 mL per aliquot). After each injection, values of pH were recorded once the steady potential was achieved (driftb 0.5 mV min− 1). Before and after each experiment, solution aliquots were taken from the cell using a plastic syringe, filtered through a 0.45 μm membrane, diluted, and analyzed for total antimony by Atomic Absorption Spectroscopy, and for SbIII by titration with iodine (see Tella and Pokrovski, 2009 for details). No precipitation or reduction of Sb V was observed in all experiments, and the absence of trivalent Sb was also confirmed by X-ray absorption spectroscopy (see below). Titrations of organic-bearing solutions without Sb were performed on different hydroxy-carboxylic acids, polyols and simple N-bearing ligands including glycine and amide, in order to estimate experimental uncertainties and check the validity of organic-acid dissociation constants. The experimental titration curves of pH versus mNaOH were found to be within 0.05–0.1 pH units far from the equivalence 1 m denotes molality (i.e., the number of moles of a solute per one kg of water in solution) through the entire article.

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

59

Table 1 Schematic structures of the organic compounds used in this study and stepwise deprotonation constants (pK = − log10 K) of their functional groups at 25 °C and 1 bar at zero ionic strength (according to Martell et al., 2004). Acetic acid (HAce)

pK (CO2H) = 4.76

Adipic acid (H2Adi)

pK1 (CO2H) = 4.42 pK2 (CO2H) = 5.42

Lactic acid (H2Lac)

pK (CO2H) = 3.86

Oxalic acid (H2Oxa)

pK1 (CO2H) = 1.40 pK2 (CO2H) = 4.26

Citric acid (H4Cit)

pK1 pK2 pK3 pK4

Salicylic acid (H2Sal)

pK1 (CO2H) = 4.37 pK2 (OH) = 13.3

Malonic acid (H2Mal)

pK1 (CO2H) = 2.80 pK2 (CO2H) = 5.69

Catechol (H2Cat)

pK1 (OH) = 9.45 pK2 (OH) = 13.0

Xylitol (Xyl)

pK = 13.8

Mannitol (Man)

pK = 13.6

Glycine (HGly)

pK1 (CO2H) = 2.34 pK2 (NH3) = 9.58

(CO2H) = 3.13 (CO2H) = 4.76 (CO2H) = 6.40 (OH) ~ 16

pK1 (CO2H) = 3.63 pK2 (SH) = 10.24

Thiolactic acid (H2ThLac)

points of the organic acid calculated, using the dissociation constants from the CRITICAL database (Martell et al., 2004). Consequently, these values were adopted in modeling of Sb V-organic complexing. Although at the proximity of the equivalence point the absolute errors between calculated and measured pH may attain 0.5–1.0 pH units, the good reproducibility of titration curves themselves (within ±0.1 pH units) enables accurate detection of Sb-organic complexing in these pH regions by comparison between Sb-free and Sb-bearing solutions having similar concentrations of organic compound.

2.2. X-ray absorption spectra acquisition and data reduction X-ray absorption spectroscopy (XAS) measurements were performed of both organic-free and organic-bearing solutions of Sb V. Two organic-free solutions of Sb V analyzed at pH ~ 7 and pH ~ 12 were prepared by dissolution of KSb(OH)6 in pure water and ~0.01 m NaOH, respectively. Because the low solubility of Sb2O5 and KSb(OH)6 at acidic conditions at room temperature (b10 − 4 m) does not allow accurate spectra acquisition within a reasonable time

60

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

from solutions of such low concentrations, an organic-free solution at pH ~ 1 was prepared by dissolution of Sb2O5 at 50 °C (~ 1 m HClO4, ~ 7 × 10 − 4 m Sb V) for XAS analyses. Sb V-bearing organic solutions were prepared from the corresponding reagents (oxalic and citric acids, catechol, xylitol and thiolactic acid) and KSb(OH)6 salt. Organic ligand concentrations were chosen on the basis of potentiometric results to insure the large predominance of Sb V-organic species over the uncomplexed antimonic acid in XAS experimental solutions. Final Sb concentrations (0.001–0.01 m Sb) and pH values were measured as described above; these concentrations were sufficient to obtain a good X-ray absorption spectrum (to at least 10–11 Å − 1) at ambient temperature in fluorescence mode after 1–2 h of acquisition. X-ray absorption spectra (including near-edge structure region or XANES, and extended X-ray absorption fine-structure region or EXAFS) were collected from aqueous solutions at ambient temperature both in transmission and fluorescence modes at the Sb K-edge (~30.5 keV) over the energy range 30–32 keV at FAME CRG beamline (Proux et al., 2005) of the European Synchrotron Radiation Facility (ESRF, Grenoble, France). The storage ring was operated at 6 GeV in uniform mode with a 160–200 mA current. The beam energy was selected using a Si (220) double crystal monochromator with sagittal focusing. Silicon diodes were employed for measuring the intensity of incident and transmitted X-ray beam. Fluorescence spectra were collected in a 90° geometry using a Canberra 30-element Ge detector. Acquisition time per point varied from 0.5 s before the absorption edge to 10 s at the spectrum end. Data analysis was performed with Athena and Artemis package (Ravel and Newville, 2005) based on the IFEFFIT program (Newville, 2001). More details about spectra acquisition and treatment are given in Tella and Pokrovski (2009). Two to ten XAS scans (depending on the signal-to-noise ratio; 40 min per scan acquisition time) were recorded for each sample, carefully inspected and if found reasonably free from beam fluctuations and other flaws, were added together. No SbIII was detected in all spectra except one (see below), demonstrating that SbV was stable in the presence of most O-bearing organic ligands and was not affected by the high beam flux. The only exception was a Sb V-thiolactic acid solution (0.007 m Sb, 0.1 m H2ThLac, pH~ 6), which exhibited an energy shift of 5 eV toward lower energies characteristic of SbV reduction to SbIII. Linear combination fits of the XANES spectrum yielded 16% of III SbðOHÞ− 6 and 84% of Sb -thiolactic complex. It was thus concluded that SbV is unlikely to be stable in the presence of ligands containing reduced sulfur; these data were not considered in further analysis. 2.3. Chemical speciation calculations Calculations of species distribution and equilibrium constants in solution were performed using the computer code BALANCE based on the Gibbs free energy minimization of the system (Akinfiev, 1986). This is realized by iterations on solution ionic strength, and using standard Gibbs free energies of aqueous species, and mass-balance and electroneutrality constraints for the aqueous solution. The standard states for solid phases and water are unit activity at all temperatures. For aqueous species, the standard state convention corresponds to unit activity for a hypothetical 1 molal solution whose behavior is ideal. The activity of aqueous species is expressed as ai ¼ mi  γi

ð1Þ

where ai is the activity of the ith species, mi is its molality, and γi is its activity coefficient. Activity coefficients for neutral species were assumed to be unity and those for charged species were calculated using the extended Debye–Hückel equation:

log10 γi ¼

pffiffi −Az2i I pffiffi þ Γ γ 1 þ B a˚ i I

ð2Þ

where A and B refer to the Debye–Hückel electrostatic parameters and were taken from Helgeson and Kirkham (1974); I is the effective molal ionic strength (I= 0.5 Σzi2 mi); zi and åi represent the ionic charge and the distance of the closest approach of the ith species, respectively; and Γγ designates the mole fraction to molality conversion factor Γγ = log (1+ 0.018 m*), where m* is the sum of the molalities of all solute species. We adopted a value for åi of 4.5 Å for all charged species. Equilibrium constants derived in this study are expressed on the activity scale. 2.4. Dialysis experiments A few dialysis experiments were conducted to determine the extent of Sb V binding with humic acid. The equilibrium concentrations of dissolved Sb V were measured inside and outside the dialysis tubing (Spectra/Por Biotech Cellulose Ester membrane, 200 Da) introduced into a ~ 100 mL polypropylene vial. The detailed procedure is described elsewhere (Tella and Pokrovski, 2009). The dialysis tubing was filled with an aqueous solution of purified humic acid (Fluka) and the external solution in the vial contained only inorganic Sb V in the form of SbðOHÞ05 and SbðOHÞ− 6 depending on pH (see Fig. 1). Amounts of humic acid or Sb-humic acid complexes passed through the membrane were negligible, as shown by analyses of dissolved organic carbon in the external solution. The oxidation state of antimony was checked at the end of the experiment by HPLC–ICP-MS (Hydrosciences, Montpellier, France) and indicated that no reduction of Sb V took place during the experiments. 3. Results and discussion 3.1. Results from X-ray absorption spectroscopy 3.1.1. Sb solid phases and organic-free aqueous solutions Modeling of the EXAFS spectrum of antimony pentoxide (Sb2O5, Fig. 2) yielded the first shell Sb\O (1.95 Å) and next-nearest Sb\Sb (3.61 Å) distances in agreement within 0.04 Å with X-ray diffraction data (Table 2; Jansen, 1979). For potassium hexahydroxo-antimonate, KSb(OH)6, reasonable models were obtained for the first shell around Sb with 6 O atoms at 1.99 Å. No K and Sb atoms were detected in the outer shells. Although we could not find the crystallographic structure of KSb(OH)6, the comparison with the analogous compound NaSb (OH)6 reveals good agreement with its crystallographic data, which indicate 6 O atoms at an average distance of 1.98 Å, and heavier atoms of Na and Sb at much greater distances from the Sb absorber (4.0 Å, Table 2, Asai, 1975), which would be difficult to detect by the EXAFS method poorly sensitive to the long-order interactions. XANES spectra of Sb V aqueous organic-free solutions from pH 1 to 12 are similar to those of solid KSb(OH)6 and Sb2O5 (Fig. 3) where antimony presents a 6-fold coordination. Consequently, in its inorganic hydroxide species, Sb V is likely to be six-coordinated in an octahedral geometry. EXAFS spectra and their corresponding Fourier transforms of Sb V organic-free solutions at pH ~ 1 and pH ~ 7 are shown in Fig. 2. The derived structural parameters are reported in Table 2. Similar parameters for the first atomic shell of Sb with ~6.6 ± 1.0 O atoms at 1.97 ± 0.01 Å were found for all organic-free solutions (Table 2). However, the DW factor for the solution at pH = 1 is significantly higher than those for the near neutral and basic solutions (0.006 vs 0.002 Å 2, Table 2). The elevated DW factor would indicate a larger structural disorder around Sb at acidic conditions. The presence of water molecules in the first shell coordination of Sb V in its hydroxide complexes at acidic pH could explain these observed differences due to the presence of 2 Sb\O subshells in the Sb(OH)6 − n(H2O)n complex, Sb\(OH) and Sb\(OH2), leading to higher structural disorder. However, the resolution of our EXAFS spectra is insufficient to be able to obtain a robust 1st shell model with 2 distinct subshells. Water molecules have also been observed for Sb III in its hydroxide

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

61

O

C 7×10-3 m Sb, 0.3 m HXyl, pH=8 7×10-3 m Sb, 0.3 m HXyl pH=8 8×10-3 m Sb, 0.3 m H2Cat, pH=9

8×10-3 m Sb, 0.3 m H2Cat pH=9

7×10-3 m Sb, 0.3 m H4Cit, pH=4

7×10-3 m Sb, 0.3 m H4Cit pH=4 7×10-3 m Sb, 0.25 m H2Oxa pH=3

| χ(R)| (Å- 2)

k²χ χ (k)(Å-2)

7×10-3 m Sb, 0.25 m H2Oxa, pH=3

7×10-4 m Sb, 1 m HClO4, pH=1

7×10-4 m Sb, 1 m HClO4 pH=1

4×10-3 m Sb, pH=7

4×10-3 m Sb, pH=7

KSb(OH)6 (s)

KSb(OH)6 (s)

Sb

Sb2O5 (s)

2

4

6

8

10

12

Sb2O5 (s)

14

0

1

k (Å-1)

2

3

4

5

6

R(Å)

Fig. 2. Normalized k2-weighted EXAFS spectra of organic-free and organic-bearing solutions, potassium hexahydroxo-antimonate (KSb(OH)6) and diantimonate pentoxide (Sb2O5), and their corresponding Fourier Transform magnitudes (uncorrected for phase shift). The vertical dashed lines indicate the position of different neighbors around Sb. Thin dashed curves represent fits of experimental data.

complexes at acidic pH (Tella and Pokrovski, 2009). Our XAS results are in good agreement with the Sb V aqueous speciation in dilute solution, dominated by the neutral antimonic acid species SbðOHÞ05 ⋅H2 O at pH b 2, and its anionic counterpart SbðOHÞ− 6 at higher pH values, according to existing potentiometric data (Fig. 1,

Baes and Mesmer, 1976; Accornero et al., 2008; references therein). No outer-shell feature was observed in inorganic Sb aqueous solutions at all pH, indicating the absence of polymerization, and/or significant hydratation of the Sb hydroxide complexes by outer sphere water molecules.

Table 2 Structural parameters of SbV atomic environment obtained from fitting Sb K-edge EXAFS spectra of potassium hexahydroxoantimonate KSb(OH)6, antimony pentoxide Sb2O5, and aqueous organic-free solutions of SbV. Sample

Scatterer

N (atoms)

R (Å)

σ2 (Å2)

R-factor

χ2v

R (Å) (XRD)a

solid Sb2O5

O Sb

6 (f) 8 (f)

1.95 ± 0.01 3.61 ± 0.02

0.004 ± 0.001 0.011 ± 0.002

0.06

335

1.99 3.59

solid KSb(OH)6

O

6.0 ± 0.8

1.99 ± 0.01

0.003 ± 0.001

0.02

964

1.98

0.0007 m Sb by Sb2O5 dissolution, 1 m HClO4, pH = 1

O

6.5 ± 0.9

1.96 ± 0.01

0.006 ± 0.001

0.01

110

NA

0.004 m KSb(OH)6, pH = 7

O

6.8 ± 0.6

1.98 ± 0.01

0.003 ± 0.001

0.01

164

NA

0.004 m KSb(OH)6, 0.01 m NaOH, pH = 12

O

6.6 ± 0.6

1.98±0.01

0.002 ± 0.001

0.002

201

NA

R = antimony-scatterer mean distance; N = number of scatterers; σ2 = squared Debye–Waller factor (DW, relative to σ2 = 0 adopted in the calculation of reference amplitude and phases functions by FEFF6); the goodness of the total fit is defined by R-factor and χ2v as described in IFEFFIT (Newville, 2001). Fitted k range = 2.5–13 Å− 1; R range = 1.2–4.0 Å. (f) indicates that number of scatterers was fixed to the crystallographic value. Theoretical ab initio backscattering amplitude and phase-shift functions for Sb\O, Sb\C, Sb\H and Sb\Sb single and multiple scattering paths were computed by the FEFF6 code (Zabinsky et al., 1995) using the crystal structures of antimonate oxide (Sb2O5, Jansen, 1979) and potassium antimonyl tartrate (K2[Sb2(C4H2O6)2] × 3H2O, Kamenar et al., 1969). The validity of these functions was checked by fitting experimental EXAFS spectra of these solids. The amplitude reduction factor, S02, was found to be 0.95 ± 0.05 from the fits of these compounds and was fixed to 0.95 in the fits of experimental solutions. NA = not applicable. a XRD data are from Asai (1975) for the analogous sodium hexahydroxo-antimonate and from Jansen (1979) for antimony pentoxide.

62

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

A

B

Fig. 3. Normalized XANES spectra of (A) antimonic acid solutions, potassium hexahydroxoantimonate (KSb(OH)6), diantimonate pentoxide (Sb2O5) and senarmontite (Sb2O3 cub.), and (B) SbV organic-bearing solutions. Spectra are offset vertically for clarity in figure B.

3.1.2. Organic-bearing aqueous solutions XANES spectra of Sb V aqueous solutions in the presence of all organic ligands investigated show intensities of the white line and post-edge features similar to that of KSb(OH)6(s) suggesting a similar octahedral environment around Sb (Fig. 3). The first-shell EXAFS spectra of all studied organic-bearing solutions are similar to those of hydroxide species and yielded robust fits with ~6 O atoms at ~ 1.98 Å (Table 3). However, significant differences are observed for DW factors, which are higher for oxalate and citrate-bearing solutions

(σ 2 = 0.008) than for catechol and xylitol ones (σ 2 = 0.004). Like for the hydroxides species of Sb V, these observations could be attributed to larger structural disorder of O atoms around Sb in citrate and oxalate species than in complexes with catechol and xylitol. Because most Sb in the studied solutions is expected to be complexed with catechol or xylitol ligands according to potentiometric data (see below), the high symmetry of the 1st shell may indicate that all 6 O atoms come from similar Sb\O\C bonds and that free Sb\OH or Sb\OH2 bonds, expected to have different lengths and disorder

Table 3 Structural parameters of SbV atomic environment obtained from fitting Sb K-edge EXAFS spectra of aqueous organic-bearing solutions in the presence of oxalic and citric acids, catechol and xylitol. Sample

Scatterer

N (atoms)

R (Å)

σ2 (Å2)

R-factor

χv2

0.007 m Sb, 0.25 m H2Oxa, pH = 1

O C1

6.5 ± 0.7 2a

1.96 ± 0.01 2.81 ± 0.01

0.008 ± 0.001 0.013 ± 0.010

0.004

63

0.007 m Sb, 0.25 m H2Oxa, pH = 3

O C1

6.4 ± 0.5 2a

1.97 ± 0.01 2.89 ± 0.03

0.008 ± 0.001 0.012 ± 0.002

0.004

82

0.007 m Sb, 0.3 m H4Cit, pH = 4

O C1

6.6 ± 0.9 2a

1.97 ± 0.02 2.82 ± 0.05

0.008 ± 0.001 0.010 ± 0.007

0.004

31

0.008 m Sb, 0.3 m H2Cat, pH = 3

O C1

6.5 ± 0.7 6a

1.98 ± 0.01 2.81 ± 0.02

0.004 ± 0.001 0.006 ± 0.005

0.002

56

0.008 m Sb, 0.3 m H2Cat, pH = 5

O C1

6.4 ± 0.3 6a

1.99 ± 0.01 2.82 ± 0.02

0.004 ± 0.001 0.006 ± 0.005

0.002

51

0.008 m Sb, 0.3 m H2Cat, pH = 9

O C1

6.6 ± 0.6 6a

1.98 ± 0.01 2.82 ± 0.02

0.004 ± 0.001 0.010 ± 0.007

0.002

30

0.007 m Sb, 0.3 m HXyl, pH = 8

O C1

6.5 ± 0.4 6a

1.98 ± 0.01 2.89 ± 0.02

0.003 ± 0.001 0.007 ± 0.004

0.003

50

See Table 2 for explanation of EXAFS parameters. Fitted k range = 2.5–13.0 Å− 1; R range = 1.2–4.0 Å. In italic: structural parameters for the second shells are tentative (see Section 3.1 for details). a Number of scatterers was fixed according to the stoichiometry found from potentiometric experiments and the organic ligand structure (Table 1).

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

than Sb\O(\C), are absent in the studied polyol solutions. The presence of Sb\O\C linkages is also indicated by the small but distinct second-shell contribution apparent on the FT of these spectra at ~ 2.4 Å (Fig. 2, uncorrected for phase shift), which is likely to stem from carbon atoms constituting the Sb\O\C bonds. The presence of 2nd shell signal in organic-bearing solutions is further revealed by comparison of their filtered EXAFS signals with that of inorganic Sb solutions, which do not display any detectable 2nd shell (Fig. EA1). The 2nd shells of EXAFS spectra of all polyol-bearing solutions have larger amplitudes and different phase shifts than inorganic Sb solutions. This provides a qualitative support for the presence of Sb\O\C bonds in the organic bearing solutions. Because of the weak signal from this feature does not allow accurate extraction of the number of C atoms from least-squares EXAFS fits, these numbers were fixed in the modeling assuming the formation of 1:3 Sb:Cat/Xyl bidendate complexes based on the potentiometric results below (Section 3.2). The derived Sb\C6 and Sb\C12 distances of 2.8–2.9 and 3.3–3.4 Å (Table 3) are consistent with the catechol and xylitol molecular structures. As for catechol and xylitol samples, the amplitudes of EXAFS contributions of outer shells in the presence of citrate and oxalate are slightly greater and de-phased compared to inorganic Sb species (Fig. EA1), likely indicating a weak contribution of Sb\O\C bonds from organic complexes. Unfortunately, the Sb 2nd shell signal is too weak to derive from EXAFS fits the number of C atoms of the carboxylic ligands bound to Sb. This number was thus tentatively fixed at 2 in the fit based on potentiometric data that suggest bidentate 1:1 (Sb:ligand) stoichiometries for these solutions (see Section 3.2 below). Note that such stoichiometries imply 4 O atoms coming from unbounded \OH groups of SbðOHÞ− 6 and 2 others oxygen atoms from \O\C bonds of the organic ligand, in agreement with the structural disorder shown by DW factors. The derived 2nd shell Sb\C distances of 2.9 Å (Table 3) are in agreement with the carboxylic ligand structures and are similar to those found for analogous complexes of trivalent Sb (Tella and Pokrovski, 2009). 3.2. Results from potentiometric experiments 3.2.1. Organic-free solutions Titrations of an organic-free solution of Sb V (1.5 × 10 − 3 m) by NaOH from pH 2 to 11 are in agreement with the following dominant reaction of Sb V hydrolysis with an equilibrium constant identical within errors to that from Baes and Mesmer (1976) and Accornero et al. (2008): 0

−

SbðOHÞ5 ðaqÞ þ H 2 O ¼ SbðOHÞ6 þ H

þ

logK ¼ −2:5  0:4

ð3Þ

This suggests the absence in our solution of significant amounts of polymeric Sb hydroxide species which were reported to form at mSb > 0.001 m in some old studies (see Baes and Mesmer, 1976; references therein). Our potentiometric results are also in good agreement with the XAS analysis of more concentrated solutions (mSb ~ 0.004 m) showing the change of the dominant Sb monomeric hydroxide species between pH 1 and 7 and the absence of Sb\Sb contributions that might arise from polymeric species. 3.2.2. Hydroxy-carboxylic acids Titrations curves, from pH 2 to 10, of acetic and adipic acid solutions (0.01 m) are identical with and without Sb V (2 × 10 − 3 m), demonstrating that Sb-acetate and adipate complexes are likely to be negligible at these conditions. In contrast, with similar concentrations of lactic, citric, oxalic, salicylic and malonic acids, having two or more adjacent functional groups, we observed systematic pH differences between Sb-free and Sb-bearing solutions in the pH range 2.5–4.5 (Fig. 4 and Fig. EA2). These findings clearly demonstrate the formation of stable aqueous Sb V complexes via Sb\O\C bonds with

63

A

B

Fig. 4. Titration curves of (A) SbV-bearing oxalic acid solution (1.0 × 10− 3 m Sb, 1.5 × 10− 3 m H2Oxa) and (B) SbV-bearing catechol solution (7.4 × 10− 3 m Sb, 0.3 m H2Cat, 0.01 m HCl) with NaOH. The shaded areas represent the uncertainties on measured pH (± 0.1 pH) and NaOH concentrations.

organic ligands having two or more adjacent functional groups; this results in the release of OH − groups from the inorganic Sb hydroxide species with consequent pH changes. The same type of complexes Table 4 Statistical criteria allowing to choose the best SbV speciation model in the presence of oxalic, citric, lactic and salicylic acids in solutions of 0.01 m of organic ligand from pH 2 to 10 with ~ 4 × 10− 3 m SbV. Δlog K

ΔpH

Oxalic acid

(1:1)− 1 (1:2)− 1 (1:3)− 1

0.4 2.0 3.1

0.03 0.14 0.19

Citric acid

(1:1)− 1 + (1:1)− 2 + (1:1)− 3 (1:2)− 1 + (1:2)− 2 + (1:2)− 3 (1:3)− 1 + (1:3)− 2 + (1:3)− 3

0.2 1.3 2.5

0.02 0.13 0.23

Lactic acid

(1:1)− 1 (1:2)− 1 (1:3)− 1

0.2 0.8 1.0

0.06 0.21 0.83

Salicylic acid

(1:1)− 1 (1:2)− 1 (1:3)− 1

0.6 3.0 3.0

0.10 0.32 0.44

Ligand

ΔpH represents the average difference between measured and modeled pH, Δlog K represents the scatter of the log K value in the whole pH range (see text). The organic complexes adopted in this study are shown in bold.

64

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

was found between Sb III and these organic ligands (Tella and Pokrovski, 2009). The following rules were applied to derive the stoichiometry and stability constants of Sb V-organic complexes: (i) a minimal number of species were used to fit the measured pH values; (ii) based on XAS results, we adopted a coordination of 6 for Sb in its organic species; and (iii) based on pH measurements showing no pH changes with mono-dentate ligands, only bidendate complexes were considered. These three rules allow us to constrain both the metal/ligand ratio and charge of Sb-organic species, so that only complexes with 1:1, 1:2 and 1:3 Sb:L ratios and charges |z| ≤ 3, imposed by the number of dissociating protons in the ligands, were adopted. These stoichiometries are similar to those found for complexes of boron, germanium, and trivalent antimony with these organic compounds (Tella and Pokrovski, 2009; references therein). The selection, among Sb\L species obeying the above rules, of complexes formed in our experimental solutions was performed using the computer code BALANCE (Akinfiev, 1986). Statistical parameters for selected speciation models are reported in Table 4. A set of species and their corresponding formation constants (Ki) were assumed to be adequate when all experimental pH data points were reproduced within error (ΔpH ≤ 0.1 units) using their Ki values averaged over the studied pH range. If two or more sets satisfy the ΔpH criterion above, the best speciation model for each organic ligand among these sets, shown in bold characters in Table 4, is one that yields the least variation of log Ki for each aqueous complex over the titrated pH range. The stoichiometries and stabilities of Sb V-organic species derived in this study are briefly discussed below. 3.2.2.1. Oxalic and lactic acids. Two NaOH titration experiments in the pH range 2–12 and three KSb(OH)6 titrations at pH ~ 2, 4 and 6 were realized on oxalic-acid solutions of 0.002–0.01 m. It can be seen in Fig. 4A and 5 that addition of Sb V results in significant pH changes in the range ~2.5–5.0. These changes are accurately described by the formation of a single (1:1) − 1 complex according to

−

−

SbðOHÞ6 þ HOxa

−

¼ SbðOHÞ4 Oxa þ H2 O þ OH

−

A similar (1:1) − 1 complex was found with lactic acid as inferred from two NaOH titrations of a 0.009 m H2Lac–0.0009 m Sb solution and three KSb(OH)6 titrations at pH ~3, 4 and 6 on 0.002–0.01 m lactic acid solutions: −

−

SbðOH Þ6 þ HLac

−

¼ SbðOHÞ4 Lac þ H2 O þ OH

−

ð5Þ

logK 5 ¼ −6:6  0:3

3.2.2.2. Citric acid. Deprotonation of citric acid in the near-neutral pH range yields 3 citrate anions of charges −1, −2 and −3, potentially capable of complexing Sb. Two NaOH titration experiments in the pH range 3–10 and four KSb(OH)6 titrations at pH ~3, 4, 6 and 10 were realized on 0.002–0.01 m citric acid solutions. All these experiments could be best modeled using a set of three 1:1 complexes having charges of −1, −2 and −3, reflecting the stepwise dissociation of citric acid: −

SbðOH Þ6 þ H 3 Cit

−

−

−

ð6Þ

þ 2H2 O

ð7Þ

¼ SbðOHÞ4 H 2 Cit þ H 2 O þ OH

logK 6 ¼ −6:5  0:4 −

SbðOHÞ6 þ H 2 Cit

2−

þ

þ H ¼ SbðOHÞ4 HCit

2−

logK 7 ¼ 7:8  0:2 −

SbðOH Þ6 þ HCit

3−

þ

þ H ¼ SbðOHÞ4 Cit

3−

þ 2H 2 O

ð8Þ

logK 8 ¼ 8:0  0:4 This speciation scheme is similar to that of boron-citrate complexes found by Lemarchand (2005).

ð4Þ

logK 4 ¼ −6:2  0:4 The refined stability constant is in good agreement with the recent value (log K = −6.4 ± 0.1) obtained in a preliminary study on the basis of a limited dataset (Tella and Pokrovski, 2008).

Fig. 5. Titration curve of a 2 × 10− 3 m oxalic acid solution at pH ~ 4 by a 0.01 m KSb (OH)6 solution. Note that titration with a Sb-free 0; 1 m KNO3 solution yields no significant change of pH (b0.02 units).

Fig. 6. Stylized structures of SbV complexes formed with (a) lactic acid via reaction 5; (b) oxalic acid via reaction 4; (c) citric acid via reactions 6 to 8; (d) salicylic acid via reaction 9; (e) catechol via reaction 11; (f) xylitol via reaction 12.

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68 Table 5 Stability constants of aqueous SbV complexes with lactic, oxalic, citric and salicylic acids, catechol, xylitol and mannitol derived in this study at 25 °C/1 bar. Reactiona

log K

Oxalic acid (H2Oxa0) − − − SbðOHÞ− 6 + HOxa = Sb(OH)4 Oxa + H2 O + OH

(K4)

− 6.2 ± 0.4

Lactic acid (H2Lac ) − − SbðOHÞ− 6 + HLac = Sb(OH)4 Lac + H2O + OH

(K5)

− 6.6 ± 0.3

Citric acid (H4Cit0) − − − SbðOHÞ− 6 + H Cit = Sb(OH)4 H2 Cit + H2O + OH 2− SbðOHÞ− + H+ = Sb(OH)4 HCit2– + 2H2 O 6 H2Cit 3− + SbðOHÞ− H = Sb(OH)4 Cit3– + 2H2O 6 HCit

(K6) (K7) (K8)

− 6.5 ± 0.4 7.8 ± 0.2 8.0 ± 0.4

Salicylic acid (H2Sal0) − − − SbðOHÞ− 6 + HSal = Sb(OH)4 Sal + H2O + OH

(K9)

− 2.7 ± 0.6

Malonic acid (H2Mal0) − − − SbðOHÞ− 6 + HMal = Sb(OH)4 Mal + H2O + OH

(K10)

≤−6.9

Catechol (H2Cat ) 0 − SbðOHÞ− 6 + 3H2 Cat = SbCat3 + 6H2O

(K11)

5.5 ± 0.5

Xylitol (Xyl0) − 0 SbðOHÞ− 6 + 3Xyl = Sb(H− 2Xyl)3 + 6H2O

(K12)

5.6 ± 0.5

Mannitol (Man0) 0 − SbðOHÞ− 6 ± 3Man = Sb(H− 2Man)3 + 6H2O

(K13)

5.6 ± 0.5

(K14)

≤−7.3

0

0

0

Glycine (HGly ) 0 0 − SbðOHÞ− 6 + HGly = Sb(OH)4 Gly + H2O + OH

65

angular distortion and thus low strain energy (see Martell and Hancook, 1996; Pokrovski and Schott, 1998 for details). In the case of salicylate, the presence of an aromatic cycle (Table 1) increases the reactivity of hydroxyl groups (Martell and Hancook, 1996) and thus allows the formation of more stable six-membered bidentate complexes than those formed by the non-cyclic malonate ligand. 3.2.3. Polyols Titration curves of 0.3 m catechol, 0.3 m xylitol and 0.3 m mannitol solutions in the presence of 0.007 m Sb V yielded similar and systematic shifts in the pH range 2–4 attaining 0.5 pH units, compared with Sb-free solutions (e.g., Figs. 4B and EA2). To account for such shifts all dissolved Sb should be complexed with these ligands in the studied solutions. Additional KSb(OH)6 titrations at pH ~1.5 and 3.0 were realized on catechol, xylitol and mannitol solutions of 0.05 m. Such ligand concentrations, much higher than those of carboxylic acids above, were dictated by the necessity to obtain measurable pH shifts in titrations with polyols, which are weaker complexants for Sb than carboxylates. It was found that different stoichiometries of Sb:polyol complexes (1:1, 1:2 or 1:3) may all account for the measured pH. The tentative choice between these possible species was done based on the XAS results that show a regular symmetrical environment of Sb in such species (low DW factors), consistent with similar six Sb\O interatomic distances, suggesting a 1:3 stoichiometry for the dominant Sb complex with the three ligands formed according to the reactions: −

0

−

SbðOH Þ6 þ 3H2 Cat ¼ SbCat 3 þ 6H 2 O logK 11 ¼ 5:5  0:5

ð11Þ

a

Reactions are expressed with dominant species of organic acids in the investigated pH range.

−

−

3.2.2.3. Salicylic and malonic acids. Modeling of titration curves of salicylic acid from pH 2 to 10 is consistent with a single (1:1) − 1 complex, similar to those formed with oxalic and lactic acids: −

−

SbðOHÞ6 þ HSal

−

¼ SbðOHÞ4 Sal þ H 2 O þ OH

−

ð9Þ

logK 9 ¼ −2:7  0:6 According to the salicylic acid structure (Table 1), this complex is a six-membered bidendate. Chain-like malonic acid (Table 1) is also expected to form 6-membered cycles with SbIII. Nevertheless, no effect of Sb V on acid–base titration curves was detected. The assumption of a similar (1:1)− 1 Sb V-malonate complex according to reaction (10) below with the same equilibrium constant as that of reaction (9) would have resulted in pH changes reaching one pH unit, which would be in marked disagreement with our experiments. Our titration data on malonic acid only permit the estimation of a maximum value of the stability constant for Sb-malonate complexes corresponding to the uncertainty of our pH measurements (b0.1 pH), and tentatively assuming a reaction analogous to Eq. (9) −

−

SbðOHÞ6 þ HMal

−

−

¼ SbðOHÞ4 Mal þ H2 O þ OH logK 10 ≤−6:9

ð10Þ

Measurements of more concentrated Sb and malonate solutions producing larger pH shifts would be necessary to better quantify the extent of Sb V complexing with malonate. Antimony (1:1)− 1 complexes with citric, oxalic and lactic acids present similar structures (Fig. 6) and stabilities (Table 5). Thus, the specific ligand geometry and chemistry (carboxyl vs hydroxyl–carboxyl) have little influence on the stability of SbV complexes. Our experiments indicate that the SbV affinity for malonate is much lower than for citrate, oxalate and lactate. The lower stability of SbV-malonate complexes is likely due to the presence of a six-membered cycle instead of the five-membered cycle in complexes with citrate, oxalate and lactate that presents the most stable configuration with a minimal

−

0

SbðOH Þ6 þ 3Xyl ¼ SbðH −2 XylÞ3 þ 6H2 O logK 12 ¼ 5:6  0:5

ð12Þ

−

0

SbðOH Þ6 þ 3Man ¼ SbðH−2 ManÞ3 þ 6H 2 O logK 13 ¼ 5:6  0:5 ð13Þ Caution should be taken, however, when applying these results to lower ligand concentrations, because of the possibility of formation of 1:1 and 1:2 complexes which could not be detected in the relatively concentrated solutions of this study (0.05–0.3 m of ligand). 3.2.4. N-bearing ligands Titrations of glycine solutions (4× 10− 3 m H2Gly) with and without V Sb (1× 10− 3 m Sb) from pH 3 to 10 do not yield detectable pH shifts. Assuming the following complexation reaction between SbðOHÞ− 6 and the glycine ligand with formation of a 1:1 bidendate complex −

0

0

SbðOH Þ6 þ HGly ¼ SbðOHÞ4 Gly þ H 2 O þ OH

−

ð14Þ

the maximal logK value of its equilibrium constant within the reproducibility of our titration experiments (±0.05 pH units) is estimated to be −7.3. This value implies much weaker complexing of Sb V with glycine than with the poly-functional carboxylic or phenolic ligands examined above. These results suggest that Sb V is likely to have very weak affinities to N-bearing organic ligands, in line with the general chemistry of Sb (e.g. Cotton and Wilkinson, 1988); however more measurements with different N-bearing ligands are necessary to make definitive conclusions. 3.3. Nature and stability of Sb V complexes with aqueous organic ligands and comparison with Sb III This study demonstrates for the first time that pentavalent antimony forms stable aqueous complexes with poly-functional organic ligands in aqueous solution. The complex formation occurs with organic ligands having aromatic or aliphatic hydroxyl and carboxyl functional groups via the establishment of pentagonal or hexagonal rings that incorporate two Sb\O\C bonds. Schematic structures of selected Sb V-organic complexes found in this study are displayed in Fig. 6. Our

66

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

results confirm the high stability of chelate organic species in which a metalloid is six-coordinated with oxygen atoms in an octahedral geometry (O\Sb\O angle ~90°) as it was found for germanium and silicon (Pokrovski and Schott, 1998; Pokrovski et al., 2000). For trivalent antimony, in contrast, because of its particular electronic configuration involving a lone non-bonding 5s2 pair of electrons, the formation of stable SbIII-chelate species requires a change of Sb coordination and geometry from 3-fold pyramidal (O\Sb\O angles ~110°) in its major inorganic species Sb(OH)3 (aq) to 4-fold pseudo-trigonal bipyramidal composed of the 5s2 electron pair of Sb and four oxygen atoms from two adjacent functional groups (O_C\OH and/or C\OH) of the organic ligand, forming a five-membered chelate cycle with O\Sb\O angles close to 90 °C (Tella and Pokrovski, 2009). It might thus be expected that the severe geometric constraints for Sb III-organic complexes leading to higher strain energy in comparison to their Sb V analogs would result in a lower stability of trivalent Sb complexes. Fig. 7 compares the fraction of dissolved Sb III and SbV complexed with organic ligands in three aqueous solutions, containing 1 × 10− 3 m total Sb and 5 × 10− 3 m of either oxalic or citric acid or catechol, calculated as a function of pH using the stability constants of Sb III and SbV complexes from Tella and Pokrovski (2009) and this study, respectively. It can be seen that at strongly acidic conditions (pH b 3), where the uncharged SbðOHÞ05 and SbðOHÞ03 are the major inorganic species (Fig. 1), complexes of Sb V with carboxylic acids account for a higher percentage of the dissolved metalloid than their SbIII analogs. Indeed, this is in agreement with the weaker structural constraints on octahedral Sb V coordination with bidendate organic ligands leading to higher complex stabilities (see above). In contrast, at near-neutral pH (4b pHb 9) SbV-carboxylic complexes are much weaker than those of SbIII. This may be explained by the change of Sb V inorganic speciation at pH> ~3, with the predominance of the negatively charged SbðOHÞ− 6 having weaker affinities for the negatively charged organic anions, whereas the Sb III speciation is dominated by the neutral Sb(OH)3 to pH ~11 (Fig. 1). The higher SbIII versus SbV affinity for catechol and polyol ligands is likely also to be controlled by the charge difference between SbIII and Sb V inorganic species. The main conclusions from our analysis are thus the following: (i) Sb affinity for organic ligands is controlled both by geometric constraints on the formation of chelate complexes and Sb inorganic speciation as a function of pH; and (ii) in the environmentally relevant pH range Sb III is likely to be much stronger bound with organic matter than Sb V. 4. Estimation of the effect of organic ligands on antimony mobility in surface waters The data generated in this study suggest the potential importance of Sb V complexation with natural organic matter (NOM) whose largest fraction is represented by humic and fulvic acids (Tipping, 2002). The two major limitations of our laboratory work with respect to the natural aquatic environments are i) the elevated Sb and organic ligands concentrations required by potentiometric and spectroscopic methods and ii) the complexity of NOM in comparison to the simple organic compound. Nevertheless, our results do provide first structural and stability data for SbV-organic complexes, which may serve as analogs for the different functional groups of humic and fulvic acids. As such, our data allow a first-order systematic assessment of the effect of NOM on Sb mobility in aquatic systems. The extent of SbV binding to a humic acid typical of continental waters (5 mmol g− 1 of carboxyl, 2 mmol g− 1 of aromatic hydroxyl, and 7 mmol g − 1 of aliphatic hydroxyl groups, see Thurman, 1985) may be estimated assuming that the stability constants of Sb complexes with different functional groups of the humic acid are the same as those with simple organic ligands having similar functions. Thus, the contribution of carboxylic functions of natural organic matter in Sb complexing is approximated using the stability constants derived in this study for each hydroxy-carboxylic acid (lactic, oxalic, citric and

salicylic). In organic-rich unpolluted waters, with typical Sb and DOC concentrations of 10− 8 mol L − 1 and 20 mg L − 1 respectively, the amount of Sb complexed with hydroxy-carboxyl groups of humic acid ranges between≤ 1% and 40% of total Sb V when using the stability constants of Sb-lactate and Sb-oxalate complexes, respectively (Fig. 8). The amount of Sb complexed with aromatic (phenolic) or aliphatichydroxyl functional groups of humic acid, predicted for the same solution using the stability constants between SbðOHÞ− 6 and catechol or xylitol, is less than 1% of Sb V in the pH range of natural waters (4b pHb 8). The effect of minor S- and N-bearing organic ligands on Sb V is, however, less constrained. Our XAS data on thiolactic acid indicate Sb V reduction to Sb III in the presence of thiol groups. From the soft–hard complexation rules (e.g., Cotton and Wilkinson, 1988) it is expected that the ‘hard’ SbV should not have significant affinities for the soft reduced sulfur ligands. An independent way to estimate the extent of Sb-humic acid interactions is to use the apparent complexation constants derived from dialysis experiments of this study. Our experiments showed that 75% of initial inorganic SbV was complexed by humic acid at pH~ 4 where the apparent complexation constant (Q) between Sb and the Dissolved Organic Carbon (DOC) has the highest value in the pH range investigated (4b pHb 7, Table 6). The obtained apparent complexation constant values indicate that the Sb V affinity for humic acid decreases with increasing pH. This is in agreement with the aqueous Sb inorganic speciation dominated by the antimonic acid anions SbðOHÞ− 6 having low affinities for the negatively charged organic ligands predominant at pH> 4–5. With these results we estimated Sb-humic acid interaction assuming that these values are independent of DOC concentration and Sb/DOC ratio, in agreement with previous findings for SbIII-humic complexing (Buschmann and Sigg, 2004). Calculations showed that Sb V amounts bound to organic matter in a 20 mg L− 1 DOC aqueous solution are less than 5% of total Sb at pH≥ 4 (Fig. 8). Thus, our rough estimations from dialysis experiments, though based on a limited dataset, are in agreement with the predictions using low-molecular weight organic ligands above, suggesting that on average ~5% of total Sb V may be complexed with aquatic organic matter via O-bearing functional groups at neutral pH in continental organic-rich waters. In contrast, at more acidic pH, up to 40% of total SbV may be complexed with aquatic organic matter in waters from acid mine drainage. The results of this study and our recent work (Tella and Pokrovski, 2009) help to better interpret Sb binding with natural organic matter and may provide insights into Sb oxidation state in environmental systems. Studies of natural non-contaminated organic-rich sediments show Sb fractions bound to organic matter of ≤10% (Crecelius et al., 1975), which is in rough agreement with our estimations for Sb V above. Other studies reported higher percentages of organically bound Sb, up to 70%, in organic-rich lake waters with DOC contents >15 mg L − 1 (Deng et al., 2001; Chen et al., 2003), which compares favorably with Sb III fractions complexed by organic matter (30–80% of Sb III in organic-rich waters, Tella and Pokrovski, 2009). The much higher affinity of Sb III than Sb V for organic ligands at slightly acidic

Table 6 Results of dialysis experiments of SbV with a purified commercial humic acid (Fluka), CDOC = 1.0 g L− 1 of carbon. pH

C iSbðOHÞ6 a

CfSb(OH)6

Ctot

niSb(OH)6/mDOC (mmol mg− 1)

Qa (L/g

3.9 5.8 6.7

4.9 × 10− 7 4.9 × 10− 7 4.9 × 10− 7

2.8 × 10− 7 3.8 × 10− 7 3.9 × 10− 7

1.1 × 10− 6 5.6 × 10− 6 4.9 × 10− 7

2.46 × 10− 6 2.46 × 10− 6 2.46 × 10− 6

2.9 ± 0.4 0.5 ± 0.1 0.3 ± 0.2

a

Q is the apparent complexation constant defined as V

concentration of Sb

C SbHum , C fSbðOHÞ6 C DOC −1

bound to humic acid (mol L

of DOC)

where CSbHum is the

) calculated as

C SbHum ¼

C tot −C fSbðOHÞ6 ; C tot ; Ctot is determinate from the difference between initial (CiSb(OH)6) and final (CfSb(OH)6) equilibrium concentrations (see for details Tella and Pokrovski, 2009).

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

to neutral pH inferred from our studies is confirmed by recent analyses of Sb speciation in organic-rich solutions from acidic soils showing that concentrations of Sb III extracted with solutions at pH 4 to 7 are positively correlated with organic carbon contents of the soil (Ettler et al., 2007). Note that direct determination of Sb oxidation state and its bonding environment in natural organic-rich waters, soils and sediments is difficult because most chemical treatments and analyses induce changes in Sb speciation compared to the original samples. Non-invasive direct spectroscopic methods are also hampered by low Sb natural concentrations. For example, a XAS study of Sb adsorbed on humic acids failed to report Sb-humic interactions (Scheinost et al., 2006), likely due to too high Sb inorganic/humic acid ratios that mask the signal from the minor Sb-organic compounds. In view of these limitations of the methods of soil and water analyses, our findings of contrasting Sb III and Sb V affinities for NOM may help to quantify in organic-rich natural samples the amount of the two main Sb redox forms having different toxicities to aquatic organisms.

5. Concluding remarks The stability and structure of complexes formed by Sb V with aqueous organic ligands having O-bearing functional groups typical of natural organic matter (carboxyl, aliphatic hydroxyl and aromatic hydroxyl) were determined at 25 °C/1 bar via potentiometric pH and XAS measurements. Stable complexes between Sb V and oxalic, citric, salicylic acids, catechol, xylitol and mannitol were found to form, whereas no complexing was detected with acetic, adipic and malonic acids and with glycine in the pH range of natural waters. Our data indicate that the formation of stable complexes between Sb V and organic ligands requires the establishment of bidendate structures, in which Sb is bound via oxygen atoms from two adjacent functional groups, forming 5- or 6-membered chelate species. These data allow generation of a first consistent set of stability constants for aqueous Sb V-organic complexes with simple organic ligands. These results, together with dialysis experiments, allow the estimation of the extent of Sb binding to DOC. Our calculations indicate that up to 40% of dissolved Sb V may be bound to humic compounds in acidic organic-rich waters typical of mine drainage (pH b 3–4, DOC > 10–20 mg L − 1), whereas this amount is reduced to less than 5% at nearneutral pH typical for most continental waters. Thus, SbV fate in most human-unaffected aquatic environments is unlikely to be related to organic matter, but rather controlled by adsorption/incorporation by mineral surfaces and oxy-hydroxide/clay colloidal particles. The Sb V complexing with NOM is slightly higher than that of Sb III in very acidic conditions (pH ≤ 2), but much lower at near-neutral conditions (pH ≥ 4). These differences are a direct function of the coordination and aqueous inorganic speciation of the two Sb redox forms. The contrasting affinities of Sb III and Sb V for organic matter may be used as a potential tracer of antimony redox forms in organic-rich waters and sediments, which is difficult to access unambiguously using common analytical techniques.

Acknowledgments This research has been supported by the French Ministère de l'Enseignement Supérieur et de la Recherche (awarding a Master thesis and Ph. D. fellowship to M. T.). We are grateful to the French CRG scientific committee of the ESRF for providing beam time and access to the synchrotron facility. We are indebted to Denis Testemale, Olivier Proux and Jean-Louis Hazemann for their assistance during XAS measurements, and Remi Freydier and Corinne Casiot for the HPLC/ICPMS analyses. Comments of Editor Jeremy Fein and two anonymous reviewers greatly improved this article.

67

Appendix A. Supplementary data Supplementary data to this article can be found online at doi:10. 1016/j.chemgeo.2011.11.004. References Accornero, M., Marini, L., Lelli, M., 2008. The dissociation constant of antimonic acid at 10–40 °C. Journal of Solution Chemistry 37, 785–800. Akinfiev, N.N., 1986. BALANCE: IBM computer code for calculating mineral–aqueous solution–gas equilibria. Geokhimia 6, 882–890. Asai, T., 1975. Refinement of the crystal structure of sodium hexahydroxoantimonate (V), NaSb(OH)6. Bulletin of the Chemical Society of Japan 48, 2677–2679. Baes, C.F., Mesmer, R.E., 1976. The Hydrolysis of Cations. Wiley, New York. Belzile, N., Chen, Y.-W., Wang, Z., 2001. Oxidation of antimony (III) by amorphous iron and manganese oxyhydroxides. Chemical Geology 174, 379–387. Buschmann, J., Sigg, L., 2004. Antimony(III) binding to humic substances: influence of pH and type of humic acid. Environmental Science and Technology 38, 4535–4541. Chen, Y.-W., Deng, T.L., Filella, M., Belzile, N., 2003. Distribution and early diagenesis of antimony species in sediments and pore waters of freshwater lakes. Environmental Science and Technology 37, 1163–1168. Cotton, F.A., Wilkinson, G., 1988. Advanced Inorganic Chemistry. Wiley. Crecelius, E.A., Bothner, M.H., Carpenter, R., 1975. Geochemistry of arsenic, antimony, mercury, and related elements in sediments of Puget Sound. Environmental Science and Technology 9, 325–333. Deng, T.-L., Chen, Y.-W., Belzile, N., 2001. Antimony speciation at ultra trace levels using hydride generation atomic fluorescence spectrometry and 8-hydroxyquinoline as an efficient masking agent. Analytica Chimica Acta 432, 293–302. Ettler, V., Mihaljevic, M., Sebek, O., Nechutný, Z., 2007. Antimony availability in highly polluted soils and sediments — a comparison of single extractions. Chemosphere 68, 455–463. European Union Council, 1998. Directive 98/83/EC on quality of water intended for human consumption. Official Journal L330, 32. Filella, M., May, P., 2005. Critical appraisal of available thermodynamic data for the complexation of antimony(III) and antimony(V) by low molecular mass organic ligands. Journal of Environmental Monitoring 7, 1226–1237. Filella, M., Belzile, N., Chen, Y.-W., 2002. Antimony in the environment: a review focused on natural waters: I. Occurrence. Earth-Science Reviews 57, 125–176. Gurnani, N., Sharma, A., Talukder, G., 1994. Effects of antimony on cellular systems in animals. Nucleus 37, 71–96. Helgeson, H.C., Kirkham, D.H., 1974. Theoretical prediction of the thermodynamic behavior of aqueous electrolytes at high pressures and temperatures: II. Debye– Hückel parameters for activity coefficients and relative partial molal properties. American Journal of Science 274, 1199–1261. Jansen, M., 1979. Die kristallstruktur von antimon(V)-oxid. Acta Crystallographica B35, 539–542. Johnson, A., Moench, H., Wersin, P., Kugler, P., Wenger, C., 2005. Solubility of antimony and other elements in samples taken from shooting ranges. Journal of Environmental Quality 34, 248–254. Kamenar, B., Grdenic, D., Prout, C.K., 1969. The crystal and molecular structure of racemic di-μ-tartrato-diantimonate(III) trihydrate (racemic ‘tartar emetic’). Acta Crystallographica B26, 181–188. Lemarchand, E., 2005. Etude des mécanismes de fractionnement isotopique du bore lors de son interaction avec les acides humiques et les oxydes de fer et de manganèse. PhD thesis, Université Paul-Sabatier Toulouse (in French). Leuz, A.-K., Monch, H., Johnson, C.A., 2006. Sorption of Sb(III) and Sb(V) to goethite: influence on Sb(III) oxidation and mobilization. Environmental Science and Technology 40, 7277–7282. Linder, P.W., Murray, K., 1987. Statistical determination of molecular structure and the metal binding sites of fulvic acids. Science of the Total Environment 64, 149–161. Martell, A.E., Hancook, R.D., 1996. Metal Complexes in Aqueous Solutions. Plenum Press. Martell, A.E., Smith, R.M., Motekaitis, R.J., 2004. Critical: Critically Selected Stability Constants of Metal Complexes Database. Version 8.0, NIST. Texas A & M University. Newville, M., 2001. IFEFFIT: interactive XAFS analysis and FEFF fitting. Journal of Synchrotron Radiation 8, 322–324. Nriagu, J., 1989. A global assessment of natural sources of atmospheric trace metals. Nature 338, 47–49. Pilarski, J., Waller, P., Pickering, W., 1995. Sorption of antimony species by humic acid. Water, Air, and Soil Pollution 84, 51–59. Pokrovski, G.S., Schott, J., 1998. Experimental study of the complexation of silicon and germanium with aqueous organic species: implications for germanium and silicon transport and Ge/Si ratio in natural waters. Geochimica et Cosmochimica Acta 62, 3413–3428. Pokrovski, G.S., Martin, F., Hazemann, J.-L., Schott, J., 2000. An X-ray absorption fine structure spectroscopy study of germanium-organic ligand complexes in aqueous solution. Chemical Geology 163, 151–165. Pokrovski, G.S., Borisova, A.Yu., Roux, J., Hazemann, J.-L., Petdang, A., Tella, M., Testemale, D., 2006. Antimony speciation in saline hydrothermal fluids: a combined X-ray absorption fine structure spectroscopy and solubility study. Geochimica et Cosmochimica Acta 70, 4196–4214. Pokrovsky, O.S., Schott, J., 2002. Iron colloids/organic matter associated transport of major and trace elements in small boreal rivers and their estuaries (NW Russia). Chemical Geology 190, 141–179.

68

M. Tella, G.S. Pokrovski / Chemical Geology 292–293 (2012) 57–68

Proux, O., et al., 2005. FAME: a new beamline for X-ray absorption investigations of very diluted systems of environmental, material and biological interests. Physica Scripta T115, 970–973. Ravel, B., Newville, M., 2005. Athena, Artemis, Hephaestus: data analysis for X-ray absorption spectroscopy using Ifeffit. Journal of Synchrotron Radiation 12, 537–541. Scheinost, A.C., et al., 2006. Quantitative antimony speciation in shooting-range soils by EXAFS spectroscopy. Geochimica et Cosmochimica Acta 70, 3299–3312. Steely, S., Amarasiriwardena, D., Xing, B., 2007. An investigation of inorganic antimony species and antimony associated with soil humic acid molar mass fractions in contaminated soils. Environmental Pollution 148, 590–598. Tella, M., Pokrovski, G.S., 2008. Antimony(V) complexing with O-bearing organic ligands in aqueous solution: an X-ray absorption fine structure spectroscopy and potentiometric study. Mineralogical Magazine 72, 205–209. Tella, M., Pokrovski, G.S., 2009. Antimony(III) complexing with O-bearing organic ligands in aqueous solution: an X-ray absorption fine structure spectroscopy and solubility study. Geochimica et Cosmochimica Acta 73, 268–290.

Thanabalasingam, P., Pickering, W.F., 1990. Specific sorption of antimony (III) by the hydrous oxides of Mn, Fe, and Al. Water, Air, and Soil Pollution 49, 175–185. Thurman, E.M., 1985. Organic Geochemistry of Natural Waters. Martinus Nijhoff, Dr W. Junk Publ. Tipping, E., 2002. Cation Binding by Humic Substances. Cambridge University Press. Ulrich, N., 1998. Speciation of antimony(III), antimony(V) and trimethylstiboxide by ion chromatography with inductively coupled plasma atomic emission spectrometric and mass spectrometric detection. Analytica Chimica Acta 359, 245–253. USEPA, 1999. United States Environmental Protection Agency. Toxic Release Inventory. Washington DC USA. Doc 810-F, 94-001. Wedepohl, H.K., 1995. The composition of the continental crust. Geochimica et Cosmochimica Acta 59, 1217–1232. Zabinsky, S.I., Rehr, J.J., Ankudinov, A.L., Albers, R.C., Eller, M.J., 1995. Multiple-scattering calculations of X-ray-absorption spectra. Physical Review B52, 2995–3009.