Environ. Sci. Technol. 2008, 42, 1643–1650
Examination of Arsenic Speciation in Sulfidic Solutions Using X-ray Absorption Spectroscopy D O U G L A S G . B E A K , * ,† RICHARD T. WILKIN,† ROBERT G. FORD,† AND SHELLY D. KELLY‡ National Risk Management Research Laboratory, U.S. Environmental Protection Agency, 919 Kerr Research Drive, Ada, Oklahoma 74820, and Biosciences Division, Argonne National Laboratory, Bldg 203, RM E113, 9700 South Cass Ave, Argonne, Illinois 60439
Received July 26, 2007. Revised manuscript received November 26, 2007. Accepted December 4, 2007.
Both thioarsenites and thioarsenates have been demonstrated to exist in sulfidic waters, yet there is uncertainty regarding the geochemical conditions that govern the formation of these arsenic species. The purpose of this research was to use advanced spectroscopy techniques, speciation modeling, and chromatography to elucidate the chemical speciation of arsenic in sulfidic solutions initially containing arsenite and sulfide. Results of X-ray absorption spectroscopy (XAS) show that experimental solutions contained mixtures of arsenite and thioarsenites with increasing substitution of sulfur for oxygen on arsenic as the sulfide concentration increased. Experimental samples showed no evidence of polymeric arsenic species, or transformation of thioarsenites to thioarsenates. The arsenic speciation measured using XAS was similar to predictions obtained from a thermodynamic model for arsenic speciation, excluding thioarsenate species in sulfidic systems. Our data cast some doubt on the application of chromatographic methods for determining thioarsenates and thioarsenites (or mixtures) in natural waters in cases where the arsenic oxidation state cannot be independently verified. The same chromatographic peak positions proposed for thioarsenates can be explained by thioarsenite species. Furthermore, sample dilution was shown to change the species distribution and care should be taken to avoid sample dilution prior to chromatographic analysis.
Introduction Determining the chemical speciation of arsenic (As) in natural water is vital to understanding As reactivity, fate, mobility, and toxicity in the environment. In the absence of sulfide, the speciation of As generally depends on redox conditions and pH. Arsenate species dominate in oxic conditions. When conditions become more reducing, thermodynamic reasoning predicts a mixture of arsenate and arsenite until conditions become sufficiently reduced such that arsenite dominates. However, in sulfidic solutions As speciation is further complicated by the potential formation of thioarsenic species (thioarsenite and/or thioarsenate species). * Corresponding author phone: (580) 436-8665; fax (580) 436-8703; e-mail:
[email protected]. † U.S. Environmental Protection Agency. ‡ Argonne National Laboratory. 10.1021/es071858s CCC: $40.75
Published on Web 01/31/2008
2008 American Chemical Society
The existence of thioarsenic species historically is based on measurements of solutions in equilibrium with a solid phase such as orpiment (As2S3) (1–7). These previous macroscopic studies have suggested a wide array of thioarsenic species ranging from monomeric to polymeric moieties that could explain the pH dependent solubility behavior of As in sulfidic solutions. More recently, thioarsenate species have been suggested as the dominant species in sulfidic waters (8–10); however, their solubility with respect to an As solid has not been demonstrated. The formation, identity, and stoichiometry of thioarsenic species continue to be subjects of focused research. Although the majority of the hypothesized species can account for the observed As solubility data, still lacking are direct measurements of the thioarsenic species present in experimental and natural solutions. Recent advances in analytical techniques have made possible direct measurements of thioarsenic speciation. These analytical methods generally fall into two categories: spectroscopic methods and chromatographic methods. Spectroscopic methods (11–14) can be further divided into UV/vis, Raman, and X-ray absorption spectroscopy (XAS). XAS includes X-ray absorption near edge spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS) spectroscopy. Chromatographic methods (9, 15–18) typically employ ion chromatography in conjunction with inductively coupled plasma-mass spectrometry (ICP-MS) to determine the speciation of As in solution. Spectroscopic and chromatographic methods have demonstrated that thioarsenic speciation in aqueous solutions depends on the ratio of S:As and the pH of the solution (11–14). Spectroscopic and chromatographic methods have generally confirmed the existence of monomeric species, but the existence of polymeric species and the identity(ies) of the polymeric species is still not agreed upon. Most spectroscopic studies of the As-S-H2O system have been performed at As and sulfide concentrations much higher than what is typically found in natural environments. Elevated As and sulfide concentrations were used to enhance the spectroscopic response and the signal of minor components. Arsenic speciation at these elevated concentration ranges may not be reflective of natural systems. The primary weakness of the spectroscopic methods is that these techniques reveal the composite speciation of the solution, since individual species are not analytically isolated prior to detection. To our knowledge, the chromatographic methods have only addressed the existence of monomeric thioarsenic species. This fact indicates that polymeric species are presumed not present, unimportant, or are difficult to detect using the current chromatographic methods. Differing viewpoints have been expressed regarding thioarsenic speciation determined using chromatographic methods, i.e., whether speciation is as thioarsenites (18) or thioarsenates (9). This issue highlights the limitation of the chromatographic technique in that the method does not probe the oxidation state of As directly. The oxidation state of arsenic is inferred rather than measured. The strength of the method is in the ability to separate individual species. In this study we used XAS to examine the speciation of As in solutions initially containing arsenite plus sulfide and used this information to re-evaluate previous speciation studies in light of the new findings. The XAS speciation data were also compared to a previous speciation model proposed by Wilkin et al. (18). In addition the pH-dependent stability of thioarsenite species is evaluated using XAS spectroscopy, VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1643
TABLE 1. Sample Identification and Solution Conditions Used for Experimental XAS Samples sample
pH
[As(III)] (mM)
[H2S] (mM)
S:As
saturation indexa
AsIII50 AsIIIMS50 AsIIIS50 AsIIIS250 AsIIIHS250
9.52 8.03 7.52 8.16 10.52
0.72 0.79 0.73 3.50 3.39
0.02 0.93 10.4 10.4 104
0.03 1.18 14.3 2.97 30.7
-6.82 0.36 -4.16 -1.41 -15.61
a Saturation index calculated orpiment solubility (2).
based
on
disordered
and the speciation data are used to address the question over which group of species, thioarsenites or thioarsenates, are detected using chromatographic methods.
Materials and Methods Stock Solutions. All stock and reagent solutions were made with deoxygenated water. Deoxygenated water was produced by bubbling nitrogen (N2) through deionized water using a gas washing bottle (Fisher). The deoxygenated water was stored in a glovebox containing an inert atmosphere (high purity N2). Stock sulfide solution was made by bubbling 100% hydrogen sulfide (H2S) (Matheson) through deoxygenated NaOH (certified Fisher 1N) for 10 min. The resulting solution concentration was 1.1 M H2S. Stock arsenite solutions were made by adding sodium arsenite (NaAsO2, J.T. Baker) to deoxygenated water. Stock arsenate solutions were prepared by adding sodium arsenate (Na2HAsO4 · 7H2O, Alfa Aesar) to deoxygenated water. All stock reagents were stored in the glovebox. X-ray Absorption Spectroscopy (XAS). Solutions for XAS analysis were prepared using the previously described stock solutions (see Table 1) just prior to analysis and kept in a purged glovebag during analysis. Stock solutions were mixed together at appropriate ratios to yield experimental samples. In some cases, pH was adjusted by addition of concentrated HCl or NaOH. Characteristics of the samples analyzed using XAS are shown in Figure 1 and Table 1. To prevent redox transformations of the solutions during analysis, samples were kept in a glovebag with a N2 atmosphere and a specially designed flow cell was used (Supporting Information Figure S1). The solutions were pumped using a peristaltic pump (Ismatec, flow was 0.5 mL min-1) from the sample reservoir in the glovebag to the flow cell placed in the X-ray fluorescence detector. Once the sample passed through the flow cell it was collected in a waste container. This flow-through system was used to decrease the likelihood of samples undergoing redox transformations during analysis. The flow cell was machined from acrylic, and the windows were made from Kapton tape to minimize X-ray attenuation. No oxidation of arsenite or reduction of arsenate was detected in trial runs which indicated that this design was sufficient to prevent redox transformations of the experimental solutions. The As K-edge XAS measurements were made at the highphoton-flux insertion device (10-ID) beamline, MR-CAT (19), at the Advanced Photon Source (Argonne National Laboratory). The insertion device was tapered to reduce the variation in the X-ray intensity to less than 15% throughout the scanned energy range (11 700 to 12 700 eV). A silicon (111) doublecrystal monochromator was used to select the X-ray energy. A rhodium mirror was used to eliminate X-rays with higher harmonic energies and to focus the X-ray beam vertically. The incident X-ray intensity was monitored through use of a nitrogen-filled ionization chamber. The fluorescent X-ray signal was monitored through the use of a Stern-Heald detector (20) with a Ge filter of three absorption lengths and 1644
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
FIGURE 1. Solubility diagram for As with respect to disordered orpiment (2). Area below the curves represents undersaturated conditions and points falling on the curves indicate equilibrium between the solid and solution at the specified pH. Black curve and symbols ) pH 7, red curve and symbols ) pH 8, green curve and symbols ) pH 9, and blue curve and symbols ) pH 10. The magenta symbols do not specify pH; for the present study (see Supporting Information Table 1) and Helz et al. (2005), the pH ranged from 8 to 12. The area inside the box indicates environmentally relevant conditions. Solution data for all points are presented in Table 1 and Supporting Information Table S3. Note filled symbols represent experimental studies; open symbols are from natural systems. with an argon-filled ionization chamber. The size of the X-ray beam at the sample was approximately 300 µm vertical by 800 µm horizontal. Linearity tests (21) were performed on the samples by monitoring the incident to fluorescent X-ray intensity with a 50% decrease in the incident X-ray intensity. These tests showed that the linearity in the samples was better than the average fluctuation in the incident to fluorescent X-ray intensity of 0.05%. All XAS measurements were at room temperature. The As K-edge EXAFS data were collected in quick scanning mode, requiring minutes for an EXAFS measurement with an energy scan range of 1000 eV. Also, the As K-edge XANES data were collected in quick scanning mode, requiring approximately 30 s for a XANES measurement with an energy scan range of 300 eV. The edge position of the arsenite and arsenate standards differed by approximately 3.9 eV, as determined by the energy value at the first derivative peak position of the normalized XANES data. The step accuracy of the monochromator at 12 keV is approximately (0.20 eV. For our experiments, the limiting factor in determining the valence state of As in the samples is the purity of the XANES standards and the signal-to-noise ratio. Raw XANES and EXAFS data were processed using the Athena software package (22) that is based on the IFEFFIT procedures (23). Multiple scans (2-10 scans depending on data quality and if static or kinetic data were collected) of each sample were collected, and each scan was aligned using a reference spectrum (sodium arsenate). The aligned spectra were averaged into a final spectrum for each sample. The averaged spectra were step height normalized and linear combination of fits (LCF) operations were processed using Athena. LCF were obtained using elemental As, aqueous sodium arsenite, aqueous sodium arsenate, orpiment, enargite (Cu3AsS4, Betts Minerals), and dimethyl thioarsenate (DMTA (24),) as standards and the collection of the standard spectra were the same as for the prepared solutions. The EXAFS, χ(k), spectra were produced using Athena with Rbkg
value of 1.0 Å (25). These spectra were modeled using FEFFIT with theoretical models constructed using Atoms (26) and FEFF6 (27) using reference structures (As2O3, claudite for As-O bonds and As2S3, orpiment for As-S bonds). The value for S02 ) 0.88 ( 0.12 was determined from a sodium arsenate standard. The error in S02 has been propagated to the error in the EXAFS coordination number (CN). The EXAFS models are described in Supporting Information Table S1. As Speciation Modeling. Thermodynamic As speciation modeling was accomplished using The Geochemist’s Workbench, Release 6.0 (RockWare, Golden, CO). Thermodynamic databases were modified to include the As species proposed in ref (18) and As2S3 solubility products reported in refs 2 and 6. The standard database (thermo.dat) was modified and used for freshwater samples. For hyperalkaline lakes and seawater, the Pitzer database (thermo_phrqpitz.dat) was also modified to include the thioarsenic species. Chemical Analysis. Preservation of sulfidic samples by acidification for analysis of total dissolved As has been shown to precipitate As sulfides (28). To eliminate the possibility of As precipitation in sulfidic solutions, samples collected for total As were preserved as follows. Sample pH was raised to pH g 10 by the addition of 1.0 M NaOH and followed by the addition 0.5 mL of 30% H2O2. After 30 min, HCl was used to acidify the sample to pH < 2. Addition of NaOH prevents the precipitation of elemental sulfur and the peroxide oxidizes all As to arsenate and all HS- to SO42-, which avoids precipitation upon acidification. Total As and S were measured using inductively coupled plasma-optical emission spectrometry (ICP-OES, Perkin-Elmer Optima 3300DV). Speciation of As and thioarsenic species was accomplished using ion chromatography (IC) coupled online to ICP-MS (IC-ICP-MS). Additional IC-ICP-MS method details can be found in the Supporting Information. Samples for speciation analysis were preserved by maintaining the samples at 4 °C, eliminating headspace, and limiting elapsed time prior to analysis to less than 48 h.
Results and Discussion Thioarsenic Speciation, Thermodynamic Considerations. Figure 1 shows the pH-dependent solubility behavior of disordered orpiment (2). For each pH, oxyanion species dominate on the left side of the plot and conversely thioarsenic species dominate on the right side of the v-shaped curve. The boxed area shown in Figure 1 is the range of total arsenic and total sulfide concentrations most typically encountered in natural sulfate-reducing systems. Data from several published studies are plotted on this figure and will be discussed later. Note that two previous studies employing XAS (Helz et al. (12) and Bostick et al. (11)) to determine As speciation are outside environmentally relevant conditions; consequently, the As speciation determined in these studies may be different from what would be typically observed in natural systems. As Speciation by XAS. To determine As speciation in the experimental solutions, XAS data were collected for standard reference materials. The white-line position and the edge position for each reference compound were determined from the normalized XANES and derivative XANES spectra (Figure 2A and B), respectively (see Supporting Information Table S2). The XANES data clearly demonstrate there are distinguishable absorption edge positions for each species of As. Important differences are the pure As(III)-S and As(III)-O bonds have a 1.4 eV shift; As(III)-O and As(V)-O bonds have a 3.9 eV shift. These energy shifts indicate changes in the covalency of As bonds for a given effective As oxidation state. For example, the As(III)-O core-electron has a higher binding energy than As(III)-S and requires more energy to excite the core-electron from the As atom.
The edge position information was applied to the experimental solutions. The XANES spectra are shown in Figure 2C and D (Supporting Information Table S2). The exact arsenic speciation cannot be determined from the edge position alone, but all sample adsorption edge positions as well as white-line positions (Supporting Information Table S2) fall between the orpiment (As(III)-S bonds) and arsenite (As(III)-O bonds) standards. These results indicate that the solutions could be mixtures of arsenite, thioarsenite, and possibly thioarsenates. Next LCF was employed to distinguish arsenite and thioarsenate and to determine the speciation of As in the experimental samples. The results of the LCF fitting and speciation modeling are shown in Table 2. The LCF procedure found no contribution of thioarsenate to the experimental spectra, and the forced inclusion of thioarsenate with thioarsenite did not improve the spectral fit. In addition, the use of thioarsenate alone without thioarsenite resulted in insignificant fits. In general, although there are differences between the results obtained from the LCF and thermodynamic modeling, the results of the XANES, LCF and thermodynamic modeling are in agreement that the solutions are mixtures of thioarsenites and arsenite (Table 2). These three methods all show that as the S:As ratio increases in solution the As(III)-O bonding contribution to the speciation becomes less significant. For example, at low sulfide concentrations the arsenite and 1:1 thioarsenite species dominate. As the sulfide concentration increases, the dominant solution species shift to the 2:1 thioarsenite species with the loss of the oxyanion species, and with further increases in sulfide the 3:1 thioarsenite species dominate. Similar trends were noted in Wilkin et al. (18) and Bostick et al. (11) for thioarsenite species and for thioarsenates by Stauder et al. (9) and Wallschläger et al. (10). Note, however, that thermodynamic modeling efforts are currently unable to predict the abundance and distribution of thioarsenate species because equilibrium data for these species are unavailable. EXAFS χ(k) spectra were also modeled to determine As speciation in the experimental samples. The χ(k) spectra and corresponding fits are shown in Figure 3, and fit results are listed in Table 3. The differences in speciation can be readily demonstrated by the phase shifts in the χ(k) spectra (Figure 3C). These speciation changes follow the increasing sulfide concentration trend reported earlier. AsIIIS50, AsIIIHS250, and AsIIIS250 samples (Table 1) are in phase and are dominantly 3:1 thioarsenite. The AsIIIMS50 sample (thioarsenite/ arsenite) shows a phase shift in the χ(k) spectrum and a further shift in the χ(k) spectrum was observed in sample AsIII50 (arsenite species). The coordination number (CN) around the central As in the thioarsenites would be expected to be approximately CN ) 3 for the 1:1, 2:1, and 3:1 thioarsenites. For the 4:1 thioarsenite or thioarsenate the CN is expected to be 4. For the experimental samples (Table 3) the CN ranged from 2.1 to 3.5, which indicates that the solutions are mixtures of As(III)-S and As(III)-O. This also agrees with the CN of the first shell as found in Bostick et al. (11), CN range 2.8-3.9 and Helz et al. (12), CN range 2.9-3.4. The As(III)-O bond distance (Table 3) in this study (1.77-1.79 Å) is in good agreement with previous work, 1.77-1.82 Å for the As-O bond in thioarsenites (11). Likewise, the As-S bound distances (Table 3) were found to be 2.23-2.24 Å which agree with previously published As-S values of Helz et al. (2.15-2.23 Å) (12) and Bostick et al. (2.12-2.31 Å) (11). Bostick et al. (11) found a significant As-As interaction at approximately 3.5 Å in their EXAFS, which was attributed to a possible polymeric As species. Our data could be modeled with an As-As interaction, but with no statistical improvement in spectral fit (R-factor or Reduced χ2). The lack of statistical improvement in the fit and the lack of a strong feature at ∼3.5 Å (as seen in realgar) leads us to conclude VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1645
FIGURE 2. Normalized (A) and derivative (B) XANES for selected standard reference materials and normalized (C) and derivative (D) XANES for experimental samples examined in this study. The key for (D): I ) elemental As [As(0)], II ) thioarsenite [As(III)-S], III ) arsenite [As(III)-O]/ enargite (thioarsenate) [4:1 As(V)-S], IV ) dimethylthioarsenate (DMTA, a thioarsenate) [2:1 As(V)-S], V ) arsenate [As(V)-O]. For standard edge and white line positions see Supporting Information Table S2 and Table 1 for experimental solution compositions.
TABLE 2. Sample Speciation Based on XANES Linear Combination of Fits (LCF) and Thermodynamic Modeling (18)a LCF
sample AsIII50 AsIIIMS50 AsIIIS50 AsIIIS250 AsIIIHS250 a
modeling
Σ 1:1 2:1 3:1 4:1 thioarsenate arsenite thioarsenite arsenite thioarsenites thio-arsenite thio-arsenite thio-arsenite thio-arsenite (%) (%) (%) (%) R-factor (%) (%) (%) (%) (%) 0 0 0 0 0
86 68 37 21 26
14 32 63 79 74
0.002 0.008 0.003 0.017 0.020
98 46
1 6
1 30 1 8
18 96 92 100
3
For solution data see Table 1. The ratio X:Y is the ratio of S to As in the thioarsenite species.
that polymeric species were not present (see Supporting Information Figure S2). The rejection of polymeric solution species when solutions are undersaturated with respect to orpiment was also proposed by Helz et al. (12). However, their study suggested that a thioarsenic trimer may exist at conditions near orpiment saturation. The existence of the trimeric species is not based on XAS, but rather that near saturation the trimeric species are more thermodynamically favorable. The Helz et al. (12) and Bostick et al. (11) studies were carried out at sulfide concentrations higher than would normally be encountered in natural environments (Figure 1). This could possibly explain the presence of polymeric species in the previous experimental studies. 1646
2 54 99 100 100
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
In summary, all samples prepared in this study were arsenite, thioarsenite, or contained a mixture of arsenite and thioarsenite species and with increasing sulfide concentration the number of S atoms surrounding the central As atom increased. There was no evidence of thioarsenate species or polymeric species present in the solutions. The experimental data demonstrate that there was reasonable agreement in arsenic speciation between XANES, EXAFS, thermodynamic modeling, and the LCF analysis. As Speciation by Chromatography. Chromatography methods (9, 15, 18) have also been used to investigate As speciation in sulfidic environments. Some chromatographic lines of evidence have suggested that thioarsenate species
FIGURE 3. Magnitude (A) and the real part (B) of the Fourier transform of the EXAFS χ(k) spectra (open circles) and model (solid line) for samples analyzed (See Table 3 for fitting parameters). The oxygen (red) and sulfur (blue) components of the model are shown beneath the spectra. The EXAFS χ(k) k2 spectra is shown in C. The open circles are the EXAFS χ(k) k2 data and the lines represent the model for the aqueous As species.
TABLE 3. EXAFS Fitting Results for Experimental Samplesa sample
path
CN
R (Å)
σ2 (Å2)
∆E (eV)
R-factor
AsIII50 AsIIIMS50
As-O As-O As-S As-S As-S As-S
3.5 ( 0.2 2.1 ( 0.2 1.4 ( 0.4 3.2 ( 0.2 2.1 ( 0.2 2.3 ( 0.2
1.770 ( 0.004 1.793 ( 0.006 2.243 ( 0.012 2.228 ( 0.003 2.228 ( 0.003 2.228 ( 0.003
0.0054 ( 0.0006 0.0054 ( 0.0006 0.013 ( 0.005 0.0061 ( 0.0005 0.0061 ( 0.0005 0.0061 ( 0.0005
-0.3 ( 0.5 -0.3 ( 0.5 -0.3 ( 0.5 0.5 ( 0.4 0.5 ( 0.4 0.5 ( 0.4
0.002 0.004
AsIIIS50 AsIIIS250 AsIIIHS250
0.011 0.022 0.002
a Path ) neighboring atom type, CN ) coordination number, R ) interatomic distance, σ2 ) mean-square displacement of the atoms within the path, and ∆E ) energy phase shift.
are dominant in sulfidic environments (9). However, this conclusion has not been reached by all researchers. Wilkin et al. (18) and Wood et al. (14) suggest that thioarsenites are the dominant species and Hollibaugh et al. (15) proposed an undetermined (valance) thioarsenic species. Stauder et al. (9) suggested that a disproportionation reaction occurs between dissolved arsenite and sulfide to produce a thioarsenate species as shown in eq 1. 5H2As(III)O3 + 3H2S ) 2As(0) + 3H2As(V)O3S- + 6H2O + 3H+ (1) The monothioarsenate formed in eq 1 can further react to form the di, tri, tetrathioarsenates, and arsenate or decompose to arsenite and elemental sulfur. Stauder et al. (9) demonstrated chromatographic separation of the proposed thioarsenates, but it should be noted that there was no direct measurement of the As oxidation state in the isolated As species. Stauder et al. (9), Wallschläger and Stadey (10), and
Planer-Friedrich et al. (8) have synthesized thioarsenate compounds as standards. We have collected preliminary characterization and XAS data on synthesized thioarsenate compounds, and results indicate that these synthesized compounds, although not pure, have a thioarsenate component (data not shown). Our experimental results indicate that rapid formation of thioarsenates is unlikely in sulfidic environments where As(III) and sulfide are initially present. The first line of evidence comes from the XAS speciation. From eq 1, the reaction product As(0) would constitute 40% of the initial As and 60% of the initial As would react to form As(V) species. Assuming that the reaction yield is as suggested in Stauder et al. (9) (1 mol of As(III) would yield 0.6 mol of thioarsenic products and 0.4 mol of As(V) and As(0)), then for a 670 µM solution (our lowest concentration) both As(0) (110 µM) and thioarsenate (160 µM) would be detectable in our XANES analysis (detection limit ∼70 µM). As can be seen in Figure 2 neither As(0) nor As(V) species were detectable in any of VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1647
FIGURE 4. (A) Derivative XANES time series. Solution conditions were S:As ) 13.7 and pH 12 (I ) elemental As, II ) thioarsenite, III ) arsenite/thioarsenate, and IV ) arsenate). (B) Real part of the Fourier transform of the EXAFS χ(k) spectra (Data ) black line, fit ) open red circles, contribution to the fit of the As(III)-O path ) green line, and contribution to the fit of the As(III)-S path ) blue line). From the derivative XANES there is no indication of either As(V) species or As(0) species in the XANES and there is no indication of sample oxidation. The EXAFS analysis indicates thioarsenite speciation. our experimental samples (based on spectroscopic comparisons to elemental As, enargite, and DMTA); consequently, we were unable to provide evidence to support the reaction in eq 1 for our experimental systems. All solutions have an As oxidation state of As(III). A second line of evidence that supports the initial rapid formation and stability of thioarsenite species is shown in Figure 4. Time-resolved XANES data show that when arsenite is added to a solution containing sulfide (pH 12) the oxidation state of As remains As(III) in our experimental system for time periods up to 35 min. Additionally, stability of thioarsenite mixtures has been tracked using UV/vis spectroscopy and examined over longer periods of time (data not shown). The third line of evidence is provided by chromatographic analysis. A typical chromatogram of a solution that contains a mixture of thioarsenic species is shown in (Supporting Information Figure S3). From the XAS spectroscopy we suspect that the solution injected into the IC contained As in the As(III) oxidation state and furthermore that the solution is a mixture of arsenite and thioarsenites. It should be noted that a small amount of arsenate is typically detected in the chromatography. The arsenate could be due to partial oxidation prior to preservation of the sample or an oxidation 1648
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
product during introduction to and separation within the chromatographic system. A question arises as to whether thioarsenates and thioarsenites have similar chromatography. One possibility is that the thioarsenite and thioarsenate species coelute, i.e., the thioarsenite and thioarsenate species have the same retention times. If this is true then the employed chromatographic method would be unable to resolve the speciation of As in sulfidic environments, since thioarsenite and thioarsenate species could not be directly distinguished. This possibility is also postulated by PlanerFriedrich et al. (8). Work is currently underway to address the coelution possibility. Another explanation is that the chromatography itself causes oxidation of thioarsenite species to thioarsenate species. This could also explain the existence of trace amounts of arsenate (below quantitation limits) in some samples examined herein and would indicate that the As speciation changed during analysis. Stauder et al. (9), Wallschläger and Stadey (10) and Planer-Friedrich et al. (8) did not explore the formal oxidation state of As in the thioarsenic species in their solutions, although their conclusions clearly favored the presence of thioarsenates. Yet the oxidation of samples during chromatographic separation was hypothesized by Planer-Friedrich et al. (8) as a potential limitation of this method. Wallschläger and Stadey (10) also indicated that samples could be oxidized during electrospray analysis, even though they argued that this was likely not occurring because solutions containing arsenite did not show significant oxidation to arsenate. If the IC-ICP-MS method is indeed leading to partial oxidation of samples then this will limit chromatography applications for analysis of sulfidic systems. Note that experiments at pH 12 (Figure 4) indicate no oxidation of thioarsenite species over time periods 2× longer than typical of chromatographic separations, suggesting that species transformation is not factor in chromatographic analysis. Although studies employing chromatographic methods have proposed that thioarsenate species are the dominant species in natural waters, these methods have not directly measured the oxidation state of As. Our direct measurements of As oxidation state in controlled mixtures of arsenite and sulfide and the subsequent use of chromatography on similarly synthesized samples indicates that As can be present as As(III) (thioarsenites and oxyanion species). This, in itself, is compelling evidence that thioarsenite species are present in solutions that initially contain only arsenite and sulfide as was the case in Wallschläger and Stadey (10) and our systems. We have no doubt that thioarsenates can form and persist in natural and synthetic solutions. In fact, natural sulfidic systems may be characterized by complex nonequilibrium distributions of arsenic oxyanions, thioarsenites, and thioarsenates. Yet, our results clearly show that thioarsenites are the initial products of reactions between arsenite and sulfide. Reaction mechanisms and the environmental conditions that result in thioarsenate formation remain unresolved and these topics are the focus of continuing studies. One factor that complicates direct comparisons between XAS and chromatographic speciation analysis is sample dilution. Spectroscopic methods typically require elevated arsenic concentrations, whereas chromatographic methods are better suited for low analyte concentrations and for environmental studies. Wallschläger and Stadey (10) reported that chromatographic speciation analysis of As in natural waters requires sample dilution to overcome matrix effects caused by high concentrations of matrix anions in the samples. Sample dilution might change the As speciation, but to what extent? The experiment depicted in Figure 5was conducted to examine the changes in As speciation with subsequent dilution of a prepared sample using the IC-ICPMS method. The primary solution was made using 10 µM As (as arsenite), 0.3 mM sulfide (∑H2S), and pH was fixed at 9.7
FIGURE 5. Changes in As speciation produced by the dilution of samples using the IC-ICP-MS chromatographic method. Original solution parameters are arsenite ) 10 µM, ∑H2S ) 0.30 mM, and pH 9.7. The sample dilution factors were 2× (0.5), 5× (0.2), and 20× (0.05). (A) The effect of dilution on arsenite and the individual thioarsenite species measured by IC-ICP-MS. (B) The relationship between the measured speciation of arsenite and ∑thioarsenite species and the thermodynamic model results. using a bicarbonate/NaOH buffer. This solution was diluted 2×, 5×, and 20× just prior to analysis using the deoxygenated buffer solution to maintain the pH at 9.7 for all dilutions. The As mass balance was calculated and the recovered As ranged from 90.9 to 100% for the diluted samples. The As speciation clearly changes with increasing dilution (Figure 5A). The primary solution is dominated by thioarsenic species but the abundance of thioarsenic species decreases favoring arsenite with increasing dilution. This dilution series was compared with the thermodynamic model as shown in Figure 5B. Although there are differences between the model and the actual data in the diluted and undiluted samples, the model and measurements generally agree as a function of dilution. Speciation Modeling. As discussed earlier the XAS (LCF) trends broadly agree with the thermodynamic speciation model that was proposed in ref (18). The major difference between the LCF and modeling results is that at high sulfide concentrations the model over predicts the proportion of thioarsenite species and under predicts the arsenite species based on LCF analysis (Table 2). Although improvements and additions to the thermodynamic database are needed, particularly with respect to thioarsenate species and the protonation states of the various thioarsenic anions, the available data can be used to estimate the significance of thioarsenic species in natural systems. To evaluate the importance of thioarsenite species in natural sulfidic environments we applied a geochemical model (18) to published data from the Black Sea (29), Lake Pavin (30), Mono Lake (15), Owens Dry Lake (31), Sannich Inlet (32), and a groundwater well near East Helena, MT. These environments encompass a wide range of natural As
and sulfide concentrations (Figure 1) as well as other parameters such as salinity and pH. For locations where the salinity was greater than or equal to seawater, Pitzer calculations were used to calculate the activities of the major cations and anions. Although Pitzer constants are not available for the thioarsenic species, it is desirable to accurately represent the activities of the major cations and anions because these can influence and modify the As speciation in solution and ion activity products for predicting relative saturation. It should be noted that we are unaware of any thermodynamic data for the thioarsenate species and if these species are present our model will not be reflective of the true speciation present in these systems. Even with this limitation the model is still useful in showing the overall importance of thioarsenic species in natural sulfidic environments. The model results for all published data chosen for analysis are shown in Supporting Information Table S3. Although there are no strong correlations between S:As, sulfide, and pH with the abundance of thioarsenite species, several general trends can be noted. First, thioarsenite species become dominant as sulfide concentrations increase to values >100 µM. Second, pH is important in determining the prevalence of the thioarsenite species. At low pH the thioarsenite species tend to dominate at lower As and/or sulfide concentrations. The thioarsenite speciation can be defined in terms of the ratio of thioarsenite to total As(III) as a percentage (ranges: high (50-100%), medium (25-50%), and low (0-25%)). Forty-eight percent of all the published sulfidic samples modeled here have high range, 18% medium range and 34% show a low range. The sum of the high and medium range indicates that for most of the data modeled, thioarsenite species were significant or dominant species. Data obtained from Lake Pavin where sulfide concentrations are
