SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 6 January 2006
Published in Soil Sci Soc Am J 70:192-203 (2006)
DOI: 10.2136/sssaj2005.0054
© 2006 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Peak, D.
Right arrow Articles by Huang, P. M.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Peak, D.
Right arrow Articles by Huang, P. M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Peak, D.
Right arrow Articles by Huang, P. M.
Related Collections
Right arrow Soil Chemistry

Soil Chemistry

Selenite Adsorption Mechanisms on Pure and Coated Montmorillonite: An EXAFS and XANES Spectroscopic Study

Derek Peak*, U. K. Saha and P. M. Huang

Dep. of Soil Science, Univ. of Saskatchewan, 51 Campus Dr., Saskatoon SK S7N 5A8 Canada

* Corresponding author (derek.peak{at}usask.ca)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
Selenite (SeO32–) is an oxyanion of environmental importance due to its toxicity to animals at higher concentrations, notably waterfowl and grazing animals. Sorption of SeO32– with mineral phases typically controls the movement and bioaccessibility of SeO32– in soils and sediments. Previous studies have successfully utilized synchrotron-based Extended X-ray Absorption Fine Structure (EXAFS) and X-ray Absorption Near Edge Structure (XANES) spectroscopy to determine SeO32– bonding mechanisms on Fe and Mn oxides, but the direct evidence of SeO32– surface complexation mechanisms on important mineral phases such as Al hydroxide and aluminosilicate minerals is still lacking. In this study both EXAFS and XANES spectroscopy was conducted on aqueous SeO32– solutions and on a variety of Al-bearing sorption samples at pH 4.5. The sorbents chosen were a hydroxyaluminosilicate (HAS) polymer, a hydroxyaluminum (HYA) polymer, montmorillonite, and both HYA and HAS coated montmorillonite. For SeO32– sorption on montmorillonite, only bidentate binuclear inner-sphere complexation was observed. For the hydroxyaluminum and hydroxyaluminosilicate polymers, a mixture of outer-sphere and bidentate binuclear inner-sphere was observed. When montmorillonite was coated with either HYA or HAS polymers then adsorption behavior was intermediate between that of the mineral and the pure polymer. Since temperate soils often contain aluminum-hydroxy and aluminosilicate coated minerals rather than discrete Al hydroxide minerals and pristine clay surfaces, the adsorption mechanisms observed on these coated surfaces are more realistic of the natural environment than sorption on pure minerals.

Abbreviations: EXAFS, Extended x-ray absorption fine structure spectroscopy • HAS, hydroxyaluminosilicate • HAS-Mt, hydroxyaluminosilicate-coated montmorillonite • HYA, hydroxyaluminum • HYA-Mt, hydroxyaluminum-coated montmorillonite • pKa, acid dissociation constant • Mt, montmorillonite • PZC, point of zero charge • PZSE, point of zero salt effect • RSF, radial structure function • XANES, X-ray Absorption Near Edge Structure spectroscopy • XAS, X-ray Absorption Spectroscopy


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
SELENIUM is a common metalloid in the soil environment due to dissolution of Se-bearing minerals and to anthropogenic sources such as wastewater from coal burning power plants. Selenium is also a micronutrient required for human and animal health. Selenocysteine is important because it is a component of glutathione peroxidase, an enzyme that catalyzes the removal of toxic peroxides that are commonly formed during aerobic metabolic processes (Lehninger et al., 1993). Therefore Se uptake by forages is linked to proper health in grazing animals. However, if soil Se levels are too high, then it tends to accumulate in forages and toxicity effects are often observed in grazing animals (Rosenfeld and Beath, 1964). Selenium toxicity is also commonly encountered in aquatic ecosystems, as Se has a smaller range between deficiency and toxicity than any other element in aquatic systems (Chapman, 1999). Bioaccumulation is often responsible for Se toxicity to waterfowl, as Se levels increase from low concentrations in the water itself to extremely high levels in plants and fish. In waterfowl, Se is both embryotoxic and teratogenic in high concentrations (Hoffman, 2002).

Soil Se undergoes a variety of redox reactions, and can be found in oxidation states ranging from -2 (selenide) to +6 (selenate) (Shriver et al., 1994) with the form found in the environment being dependent on soil redox status (Huang and Fuji, 1996). One important intermediate form of selenium in soils and surface waters is SeO32–. Selenium in SeO32– has an oxidation state of +4, and SeO32– is a weak diprotic acid that can exist as H2SeO3, HSeO3–, or SeO32– depending on solution pH (pKa1 is 2.64 and pKa2 is 8.4) (Shriver et al., 1994). Selenite is considered to be one of the more toxic forms of Se, and as such its chemistry in the soil environment has been researched quite extensively. Sorption processes are extremely important for SeO32– in soils, as adsorption between SeO32– and soil mineral phases is often quite strong, and sorption complexes can also explain the slow reduction of oxidized Se forms in the presence of soils (Neal and Sposito, 1991). Sorption of SeO32– is also an important reaction step in the reduction of SeO32– to elemental Se in the presence of green rusts (Myneni et al., 1997).

Several researchers have studied the sorption of SeO32– on inorganic soil components. Hingston and coworkers (Hingston et al., 1971, 1974) found that SeO32– adsorption on Fe and Al oxides is pH dependent but does not vary with ionic strength. Additionally, they observed that the majority of SeO32– was irreversibly bound to the oxide surface. They concluded that SeO32– reacts to form covalent chemical bonds with metal oxide surfaces. Zhang and Sparks (Zhang and Sparks, 1990) utilized a pressure jump chemical relaxation technique and observed that the reaction between SeO32– and goethite consisted of two reaction steps. The fast step was attributed to outer-sphere complex formation, and the subsequent slower step was assigned to formation of an inner-sphere surface complex.

Several scientists have used spectroscopic techniques to determine precisely how SeO32– bonds on mineral surfaces. Hayes and coworkers (Hayes et al., 1987) used EXAFS spectroscopy to determine bonding mechanisms of SeO42– and SeO32– on goethite. They determined that SeO32– forms inner-sphere surface complexes with goethite, and that these complexes are bidentate binuclear (bridging two adjacent Fe octahedral). Manceau and Charlet (Manceau and Charlet, 1994) published results from EXAFS studies of SeO32– adsorbed on hydrous ferric oxide. They also found bonding mechanisms consistent with a bidentate binuclear bonding mechanism, although a bidentate mononuclear SeO32– surface complex was also observed. This second, shorter Se-Fe distance was not observed on more crystalline Fe oxides such as goethite. More recently, Foster and coworkers studied SeO32– sorption on Mn oxides. (Foster et al., 2003) They observed that SeO32– forms both bidentate mononuclear (Se-Mn 3.07 Å) and monodentate surface complexes (Se-Mn 3.49 Å) on hydrous Mn oxides. MnSeO3·H2O precipitation could not completely be ruled out in this study, but if it does occur then the precipitate is extremely disordered and amorphous. Only one researcher has reported Se EXAFS data for SeO32– adsorbed on an Al oxide (Boyle-Wight et al., 2002b). Their study was conducted on {gamma}-Al2O3 at pH 7.5 in the presence or absence of Co2+. The researchers concluded that SeO32– formed inner-sphere complexes based on macroscopic data (Boyle-Wight et al., 2002a), but they were not able to conclusively observe these complexes with EXAFS. They attributed this to Al being an extremely weak backscatterer. However, there do appear to be some second shell contributions present in the radial structure functions (RSFs) of their SeO32– samples (Fig. 4b of Boyle-Wight et al., 2002b). This highlights the need for additional studies of SeO32– reactivity with aluminum-bearing mineral phases.

While previous EXAFS studies have concentrated on Fe and Mn oxides, many different mineral phases composed of Al (e.g., aluminosilicates and aluminum oxides) are potentially important sorbents for SeO32– in soils. To further complicate matters, it is well known that soil minerals are often associated with both inorganic and organic coatings in natural systems. These conglomerate sorbent phases may have reactivity much different from their individual components. Previous research (Saha et al., 2004) used HYA and HAS coated montmorillonite as a model system to investigate the importance of coatings on SeO32– adsorption. It was found that these coatings greatly enhanced the sorption capacity of montmorillonite and affect adsorption rates as well.

The goal of this study was to utilize synchrotron-based XAS (both EXAFS and XANES spectroscopy) to elucidate the adsorption mechanisms of SeO32– on HYA and HAS polymers, HYA and HAS-coated montmorillonite, and pure montmorillonite sorbent phases at pH 4.5. This pH was chosen for XAS studies because previous researchers (Saha et al., 2004) have extensively studied the reaction kinetics and sorption behavior of SeO32– with these surfaces and inferred that different adsorption mechanisms are responsible for differences in macroscopic reactivity. The spectroscopic experiments conducted in this research are therefore useful to validate the kinetic models and thermodynamic results from previous laboratory experiments (Saha et al., 2004). EXAFS spectroscopy was chosen because it is a well-established method for probing oxyanion-bonding mechanisms on mineral surfaces, and it is expected that possible inner-sphere complexes will be easily differentiated in our samples based on Se-Al distance. A large amount of specific chemical information is also contained in the XANES region, and we believe that XANES analysis will also be useful in probing SeO32– adsorption mechanisms on our sorbents. For example, researchers have found that XANES spectroscopy can distinguish outer-sphere and inner-sphere complexation of As(III) on aluminum oxides (Arai et al., 2001), outer-sphere and inner-sphere bonding mechanisms of lead on montmorillonite (Strawn and Sparks, 1999), and can distinguish the protonation states of As(III) and As(V) in aqueous solutions (Myneni et al., 1999). The XANES technique can be used to obtain detailed information about precipitate geometry (including cluster size) in surface precipitates (Waychunas et al., 2003), and can determine the coordination of sulfate in Fe minerals (Myneni et al., 1997). Based on the above studies, if outer-sphere complexation is occurring simultaneously with inner-sphere complexation, then it should be visible in the XANES spectra.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Mineral Synthesis
Hydroxyaluminum precipitate was prepared using a method based on that of Sims and Bingham (Sims and Bingham, 1968) for the synthesis of hydroxy-Fe precipitate. A 200-mL aliquot of 1.5 M AlCl3 was placed in a 1-L beaker and neutralized slowly at a rate of 0.5 mL min–1 with 2.0 M NaOH up to a OH/Al ratio of 2.7 under continuously stirred condition. A HAS precipitate was prepared following a method based on Su and Suarez (Su and Suarez, 1997) for the synthesis of allophone. A 100 mL of 3.0 M AlCl3 solution was mixed well with 100 mL of 1.5 M Na2SiO3 and placed in a 1-L beaker. The resulting solution, containing a Si/Al molar ratio of 0.5, was then neutralized at the rate of 0.5 mL min–1 with 2.0 M NaOH up to a OH/Al ratio of 2.7 under continuously stirred condition. The resulting precipitates were aged for 1 h, resuspended in 1 L of distilled water, centrifuged at 1000 x g, and the supernatant was discarded. The precipitates were then washed three times with 600 mL of 95% ethanol followed by washing with distilled water until the leachate did not form a precipitate with AgNO3. The washed precipitates were kept as suspension in distilled water with a density of solid as 20 g L–1 and used immediately for SeO32– adsorption experiments. Unused precipitate was freeze-dried and analyzed for physical and chemical properties.

Montmorillonite (Mt), hydroxyaluminum-coated montmorillonite (HYA-Mt), and hydroxyaluminosilicate-coated montmorillonite (HAS-Mt) were prepared and characterized in a previous study (Saha et al., 2004). Surface area and point of zero salt effect (PZSE) measurements for all sorbents used in this study were previously reported by Saha and coworkers (Saha et al., 2004) and are compiled in Table 1. PZSE for the montmorillonite were not experimentally determined, but is typically below 3. It is important to note that PZSE is not necessarily equal to the point of zero charge (PZC) of mineral surfaces, and instead is the point at which titrations conducted at different ionic strength cross one another. Researchers have shown that the PZC may differ from the experimentally determined PZSE, but both PZC and PZSE follow the same trends. Literature PZC values for pure aluminum hydroxide minerals range from 8.5 to 9.5 (Sparks 1995), which is consistent with the experimentally determined PZSE results of Saha and coworkers (2004). Based on experimentally determined PZSE values, at pH 4.5 one sorbent used in this study (Mt) will be negatively charged, two sorbents (HYA-Mt and HAS-Mt) are close to neutral/have a small surface charge, and the remaining sorbents (HYA and HAS) will be strongly positively charged. Previous researchers have shown that outer-sphere complexation of other oxyanions such as arsenite (Arai et al., 2001; Goldberg and Johnston, 2001) and SeO42– (Peak and Sparks, 2002; Wijnja and Schulthess, 2000) is common when solution pH is below the PZC of metal oxides as is the case for the pure HYA and HAS samples. It is also known that inner-sphere complexation of oxyanions can result in a surface complex with an overall negative charge. This leads to a decreased PZC, and may be important in the HAS and HYA-coated samples which have PZSE near the experimental pH of 4.5.


View this table:
[in this window]
[in a new window]
  		Table 1. Characteristics of sorbents used in this study.

 
Aqueous Samples
Solutions containing 50 mM total SeO3 (as Na2SeO3) at pH 1.10, 3.32, 5.52, 8.50, and 11.02 were prepared in a glovebox under a nitrogen atmosphere to exclude oxygen from the samples. An oxygen-free environment was necessary to obtain spectra free of SeO32– oxidation at alkaline pH. Initially, 5 mL of 100 mM SeO32– solution pH was adjusted to the desired pH with either 1 M HCl or 1 M NaOH then enough deionized water (18.2 M{Omega} Barnstead Diamond NanoPure) was added to produce a 10-mL final volume. These samples were placed in a centrifuge tube, sealed with parafilm, placed into a nitrogen-filled Ziploc bag, and shipped to the Advanced Photon Source (Chicago, IL). Samples were analyzed in transmission mode by placing the sealed tube in the path of the beam.

Adsorption Sample Preparation
A 1 g L–1 suspension of sorbent was placed in 0.01 M NaNO3, adjusted to pH 4.5, equilibrated overnight, and then Na2SeO3 was added from a freshly prepared 1 mM stock solution to reach a final concentration of 25 µM SeO32–. Sample pH was readjusted to pH 4.5 after SeO32– addition and again if necessary as the reaction proceeded. After a 48-h reaction time, the suspensions were centrifuged at 1000 x g and the supernatants were analyzed with flame Atomic Absorption Spectrophotometry (AAS). HyA-Mt, HAS-Mt, Mt, and pure HYA and HAS adsorption studies were done open to the atmosphere at pH 4.5 as part of a separate research project (Saha et al., 2004). However, dissolved carbon dioxide concentrations in solutions at pH 4.5 are extremely low, and no oxidation of SeO32– was observed in the XANES spectrum of any samples. Selenite sorption amounts (expressed as µmol m–2) for all samples are included along with reaction conditions in Table 2. One complication with this surface coverage/surface loading calculation arises from uncertainty in what portion of the mineral surface is actually available for reaction with SeO32–. The µmol m–2 loading numbers in Table 2 are derived from the total (internal plus external) surface area measurements on the sorbents. However, it is not known whether the interlayer space of Mt remains accessible to SeO32– after coating with HYA or HAS. If only external surface area is considered, the surface loadings are fairly similar for Mt, HYA-Mt, and HAS-Mt: 0.08, 0.16, and 0.09 µmol m–2, respectively. Higher surface loadings were not attempted with these samples because we were concerned that unreacted SeO32– in solution at equilibrium would be detectable in our EXAFS experiments and could make detection of outer-sphere complexes impossible. With all the samples as prepared more than 95% of the signal was calculated as arising from the adsorbed SeO32– and not from entrained solution. The sorption capacity of the HYA and HAS sorbents was much larger so more SeO32– could be adsorbed without issues from entrained solution. This allowed us to collect higher quality EXAFS data than with the coated samples. There is some evidence that surface loading has an effect on sorption mechanism of oxyanions, especially when both outer-sphere and inner-sphere complexes form simultaneously. The major effect when pH is held constant is that slightly more inner-sphere complexes form at higher coverages rather than a dramatic shift in the configuration of inner-sphere complexes.


View this table:
[in this window]
[in a new window]
  		Table 2. Structural parameters of selenite in solution and sorbed on mineral surfaces.

 
X-Ray Absorption Spectroscopy
EXAFS spectra were collected at the Se K-edge (12.658 keV) at the PNC-CAT Bending Magnet beamline (20-BM) at the Advanced Photon Source at Argonne National Laboratory. The electron storage ring was operating at 4.5 GeV and in top up mode (100 mA) for all experiments. The beamline was calibrated using a Pt foil placed in the path of the beam downfield from the sample chamber between I and Iref. Solution samples were analyzed in transmission mode to avoid self absorption effects while sorption samples were analyzed in fluorescence mode using a solid-state 13-element Ge detector (Canberra). Solution samples were analyzed at room temperature while a cryostat was utilized to lower sorption samples (analyzed as moist pastes) to 25 K to avoid any beam-induced oxidation over long scan times (up to 4 h for the most dilute samples). An aqueous SeO32– sample at pH 9.8 was analyzed at both 25 K and 298 K to compare with the room temperature data and no significant differences were observed.

XAS Data Analysis
WinXAS version 2.3 (T. Ressler, 1997) was used for all data reduction and analysis. All scans were first checked for edge shifts, which were corrected if observed. Multiple scans (from 5 for solutions to 25 scans for the Mt sorption sample) were averaged and then background subtracted using a linear equation for the pre-edge region and a second-order polynomial for the post-edge region. After baseline correction, then every spectrum was normalized to an edge jump of 1.0. EXAFS fitting was done by comparing experimental spectra to theoretical single scattering paths for model compounds. ATOMS 3.0 (Ravel, 2001) was used to construct mineral phases from crystallographic data, and FEFF7 (Zabinsky et al., 1995) was then used to calculate theoretical single scattering paths. Fits were conducted in R space with the amplitude reduction factor (S02) fixed at 0.86 while coordination number, bond distance, Debye-Waller, and e0 shift were all allowed to vary. Data was then also fit in k space to verify that optimized parameters were similar in both cases. Linear combination XANES (LC-XANES) fitting was conducted on samples using the methods recommended by WinXAS and aqueous and adsorbed SeO32– reference spectra. The fitting with reference spectra was conducted in two runs: an initial run where E0 was allowed to vary to optimize the relative contributions of all components, and then a second run where E0 shift was fixed at 0.00.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
EXAFS of Selenite Aqueous Solutions
To investigate how protonation state may affect EXAFS spectra, solutions containing 50 mM Se were analyzed at three different pHs (1.1, 5.5, and 11.0) to produce solutions containing only H2SeO3, HSeO3–, or SeO32–. The k3–weighted chi data and the RSFs obtained from performing a Fourier transformation on the chi data are shown in Fig. 1 (raw data denoted with solid lines). Structural parameters obtained from EXAFS fitting (fitted spectra are shown with open circles in Fig. 1) are tabulated in Table 2. In all protonation states, ~2.5 oxygen atoms surrounds the central Se atom at 1.70 ± 0.01 Å. This bond distance is in good agreement with previous studies (Hayes et al., 1987; Manceau and Charlet, 1994, etc.), and the coordination number differs from the true value of 3.0 by only 20% (the estimated error for the fitting routine). Interestingly, the amplitude of oscillations in the chi data and the intensity of the Se-O shell in the RSF decrease as SeO32– becomes protonated (Fig. 1). This is most likely caused by a shift from three equivalent Se-O distances in SeO32– (with bond delocalization) to single and double Se-O bonds in both biselenite and selenous acid. It was possible (data not shown) to fit the H2SeO3 data with two separate Se-O distances (at 1.61 and 1.74 Å), but the r2 was similar to the single shell fit that is presented in Fig. 1.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 1. Se K-Edge EXAFS results for aqueous selenite solutions. Solid lines denote the raw data, and open squares are the fit to theoretical standards.

 
EXAFS of Selenite Adsorbed on Mineral Samples
Samples containing SeO32– adsorbed on HYA and HAS polymers, Mt, and HYA-Mt or HAS-Mt polymers at pH 4.5 were analyzed with EXAFS to determine bonding mechanisms on these sorbents. The k3–weighted chi data and the Radial Structure Functions obtained from performing a Fourier transformation on the chi data are shown in Fig. 2 for all sorption samples (raw data denoted with solid lines). Structural parameters obtained from EXAFS fitting (fitted spectra are shown with circles in Fig. 2) are tabulated in Table 2. In all samples there are features (denoted by arrows in Fig. 2) in both the raw chi data and the Fourier transforms that suggest inner-sphere complexation of SeO32–. There are clearly some constructive and destructive interferences in the chi data that result in deviations from a single Se-O shell (as seen in the aqueous samples), and when a Fourier transform is performed then a second shell Se-Al contribution is observed between 3.16 and 3.22 Å in all samples. Fitting routines suggest that this Se-Al bond distance is shortest (3.16 Å) when SeO32– is sorbed on montmorillonite, longest (3.22–3.24 Å) when SeO32– is adsorbed on hydroxyaluminum and hydroxyaluminosilicate polymers, and of an intermediate distance (3.16–3.18 Å) on the polymer-coated montmorillonite samples. This bond distance is consistent with those calculated for bidentate binuclear (also termed bridging or corner sharing) SeO32– complexes. While the change in bond distance with pure polymer sorbents is larger than the error associated with the fits (estimated at 0.02 Å), it is not significant enough to represent a change in bonding configuration from monodentate to bidentate. One would predict a monodentate complex to have a Se-Al distance of ~3.42 Å and a bidentate mononuclear (edge-sharing) complex is expected to have a Se-Al bond distance of ~2.57 Å. These numbers are based on simple geometrical calculations based on an Al-O distance of 1.72 Å [as calculated for octahedral Al(H2O)63+] and an Se-O distance of 1.70 Å (as observed with EXAFS in this study). Instead, it is more likely that there may be some slight deviation in bond lengths and coordination number for a bidentate binuclear bonding environment that is possible as the sorbent structure is changed.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 2. Se K-Edge EXAFS results for selenite adsorbed on a range of sorbents at pH 4.5 and 0.01 M I. Solid lines denote the raw data, and open squares are the fit to theoretical standards. Arrows denote spectral features consistent with Se-Al backscattering due to inner-sphere complexation.

 
For the pure polymer samples, it appears that the single Se-Al scattering path may not fully describe the second shell; the coordination number for both HYA (2.55) and HAS(2.80) samples is quite large and the width of the second shell of the Fourier transforms is consistent with a disordered bonding environment. This could be due to either a large static disorder in the bonding environment (meaning that a single type of surface complex has a large range of bond distances) or two distinct complexes with slightly different Se-Al distances in HYA and HAS samples. It is not possible to distinguish these possible explanations with the existing EXAFS data. In the coated montmorillonite samples this effect is less pronounced, coordination numbers decreases to 1.78 for HYA-Mt and 1.80 for HAS-Mt, and bond distances also shorten slightly to 3.16 Å (HYA-Mt) and 3.18 Å (HAS-Mt). In the pure montmorillonite adsorption sample, the Se-Al distance (3.16 Å) is significantly shorter than the interatomic Se-Al distances observed in any of the pure polymer samples (3.22–3.24 Å). When considered together, it appears that there a bidentate binuclear complex forms on all surfaces, but that there is a slight difference in bonding environment depending on sorbent. This is reasonable given the nature of montmorillonite and the polymers. On montmorillonite, the only available Al groups for reaction would occur at the fairly structured edge sites of the mineral. On the HYA and HAS polymers, however, essentially all of the Al on the surface of the polymer is available for reaction. Additionally, these polymers are of short-range order rather than crystalline. This results in longer bond distances and larger coordination number as SeO32– is essentially enveloped by the polymer. On polymer-coated montmorillonite surfaces, a portion of the polymer's reactive surface is tied up by interaction with the planar tetrahedral sheet and with pillaring in the montmorillonite interlayer. When reacted with SeO32–, these polymer-coated montmorillonite sorbents have Se-Al coordination numbers and bond distances somewhat intermediate between what is observed on a pure montmorillonite surface and on a pure polymer phase.

Some of the more qualitative differences in the EXAFS spectra also deserve discussion. One interesting observation in the raw chi data is that the amplitude of the montmorillonite sorption sample is much less than the other samples, and the amplitude of the coated montmorillonite samples is also somewhat decreased relative to the pure polymer sorption samples. This is also visible in the form of decreased Se-O shell height in the RSFs. This is very similar to the effects of protonation on aqueous SeO32–. Although coordination numbers and Se-O distance was similar for all protonation states of aqueous SeO32–, H2SeO3 had a decrease in EXAFS amplitude (in the chi data) and Se-O shell height (in the RSF). This may be a general effect that is caused by full localization of bonds to form a formal Se = O double bond and two formal Se-O single bonds. The fact that the amplitude of the chi data and the height of the Se-O shell in the RSFs are larger in the polymer-coated and pure polymer sorption samples implies that there may be some contributions to the spectra from SeO32– in a different coordination with more delocalized Se-O bonding. This could possibly be explained with the presence of some additional outer-sphere complexation, which occurs simultaneously with inner-sphere complex formation. However it is difficult to conclusively assign outer-sphere complex formation with EXAFS when inner-sphere complexes are also present. Previous researchers (Arai et al., 2001) have found that XANES spectroscopy is extremely useful tool in determining outer-sphere complexation in such systems, and for this reason XANES spectra of aqueous and adsorbed SeO32– were next examined in detail.

XANES of Selenite in Aqueous Solutions
By changing solution pH to produce H2SeO3, HSeO3–, and SeO32–, it is possible to produce standards for XANES analysis that represent the full range of molecular symmetry possible for SeO32–. When SeO3 is protonated, either partially or fully, then there are changes in molecular structure that may affect symmetry. First of all, the proton may disrupt vertical mirror planes or it may be placed in a position that preserves them. This will change symmetry from CS to C1. Second, bond delocalization may no longer occur, which would reduce symmetry of the structure further. Diagrams of molecular symmetry and Lewis structures for all protonation states of SeO32– are compiled in Fig. 3 to aid in the following discussion.



View larger version (13K):
[in this window]
[in a new window]
  		Fig. 3. Relationship between protonation state, molecular symmetry, and bond delocalization for selenite species.

 
There has not been a rigorous assignment of symmetry for all protonation states of SeO32– based on either molecular modeling or vibrational spectroscopy. However, Nakamoto (Nakamoto, 1997) includes SeO32– in the discussion of pyramidal XO3 molecules, and researchers (Wang and Zhang, 2002) have recently optimized the structure of H2SO3, HSO3, and SO3 using both theoretical modeling and experimental data. The optimization of SO3 is not directly applicable to SeO32– because SO3 is trigonal planar unlike SO32– or SeO32–, but the effects of protonation on bond delocalization are still relevant for SeO32–.

Infrared and Raman spectroscopy verify that SeO32– has C3v symmetry, meaning that all three oxygens are interchangeable (Nakamoto, 1997). In other words, the bonding described by a Se = O double bond in Lewis structures is actually fully delocalized (hence the three resonance structures drawn in Fig. 3). This is also the case in the molecular modeling of SO3, which is reported as having D3H symmetry (Wang and Zhang, 2002).

In contrast, H2SeO3 represents another extreme in symmetry. When the structure of H2SO3 (analogous to H2SeO3) was optimized (Wang and Zhang, 2002), it was found to possess CS symmetry, meaning that there is one Se = O localized double bond and two Se-OH bonds with protons arranged so that there is a reflection plane in the molecule.

Biselenite (HSeO3–) is somewhat more difficult to assess, since it is possible that it may or may not have bond delocalization among the two bare oxygen atoms, and the single proton may or may not lie in the mirror plane. Wang and Zhang (Wang and Zhang, 2002) reported an optimized C1 symmetry for the HOSO2 molecule, but they do not mention whether this is due to lack of delocalization or if it is caused by proton position in the molecule. One would expect that since SO3 and SeO32– exhibit bond delocalization, HSeO3– would as well.

Given the above information about symmetry, XANES spectra of aqueous SeO32– species may now be more rigorously interpreted. Transmission mode XANES spectra of 50 mM SeO32– solutions adjusted with acid or base were collected and presented in Fig. 4a . An enlargement of the 12.66- to 12.69-keV range of these samples is also shown in Fig. 4b to highlight differences in the XANES region. The two most important observations that can be made from the XANES spectra of aqueous SeO32– solutions are: (1) H2SeO3 has features that are different in position and shape from the other samples and (2) The position of peaks in HSeO3– and SeO32– is fairly similar but the intensity of the features is different. These observations are completely consistent with the symmetry discussion above if there is bond delocalization in the unprotonated oxygens of HSeO3–. In this case, both HSeO3– and SeO32– would have shared pi electron density and intermediate Se-O bond lengths. Such delocalization cannot occur with H2SeO3, and distinct long (single bond) and short (double bond) Se-O distances must occur in this molecule. One important consequence of this interpretation of the data is that molecules with different symmetry (SeO32– is C3V while HSeO3– is CS or C1) have similar spectra and not molecules with the same symmetry (both H2SeO3 and HSeO3– are either C1 or CS depending on the position of protons). This suggests that bond delocalization is the dominant force in affecting XANES spectra of SeO32–. This is not too surprising since the Se-O bond lengths in the first coordination shell will certainly affect the geometry of the molecule, and therefore the intensity and position of multiple scattering effects in the XANES spectra.



View larger version (18K):
[in this window]
[in a new window]
  		Figure 4. (a) XANES spectra of H2SeO3, HSeO3–, and SeO32– (b) 12.665 to 12.69 keV energy range expanded to more clearly show differences in the near edge spectra.

 
LC-XANES of Aqueous Solutions
Aqueous samples were also analyzed that contained two SeO32– protonation states (pH 3.3 and 8.5). These spectra were fit with a linear combination approach using WinXAS 2.3 and the H2SeO3, HSeO3–, and SeO32– solutions as standards. Error in LC analysis was estimated at 3% by comparing the fit of all individual scans to the fit results for the overall averaged spectrum. Results are shown in Fig. 5 . The pH 3.3 spectrum could be fit with a mixture of 82% HSeO3– and 18% H2SeO3, and the pH 8.5 spectrum could be fit with a mixture of 44% HSeO3– and 56% SeO32–. These distributions are in agreement (within 0.5%) with the expected speciation of SeO32– using published pKa values of 2.63 and 8.4. Using those pKas, the Henderson–Hasselbalch equation predicts 82.4% HSeO3– and 17.6% H2SeO3 for the pH 3.3 sample and 55.7% SeO32– and 44.3% HSeO3– for the pH 8.5 sample. This excellent agreement between solution speciation models and LC-XANES fitting suggests that changes in the XANES spectra as a result of Se-O bond delocalization are fairly quantitative in nature.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 5. Linear combination XANES fits of aqueous samples where more than one selenite protonation state is present. LC-XANES predicted distributions are within 0.5% of actual speciation values, and the estimated error of the LC-XANES fitting was 3%.

 
XANES of Selenite Adsorbed on Mineral Samples
XANES spectra of SeO32– adsorbed on a variety of mineral soil components are shown in Fig. 6 . The spectra of HSeO3– (aq) and H2SeO3(aq) are also included for reference. HSeO3– is the dominant solution species at pH 4.5. XANES spectra of all sorption samples contain features that are clearly different from HSeO3–, which suggests that some inner-sphere complexation is occurring. This is in good agreement with the EXAFS studies. However, while the EXAFS spectra all had generally similar features, the XANES spectra for different sorbents show some important differences. In particular, the XANES spectrum of SeO32– sorbed on montmorillonite has a well-defined 12.67 keV peak in the XANES region (noted by the arrow in Fig. 6a) that is similar in both energy and shape to H2SeO3(aq). As discussed previously, H2SeO3 has a fully localized double bond and two Se-O-H bonds. The fact that adsorbed SeO32– has features in similar positions therefore suggests that it has similar coordination chemistry and is most likely bound as a bidentate complex. This is also consistent with the EXAFS fit results for the montmorillonite sample. It has also been observed that SeO32– adsorbed on FeOOH via a bidentate binuclear inner-sphere complexation mechanism has a similar XANES spectrum over all pH range (data not shown). One other type of surface complex would result in a fully localized Se = O bond: monodentate adsorption of HSeO3– would produce a monodentate biselenite complex with one of selenite's oxygens coordinated to Al, one to a proton, and the third present as the double-bonded O. As mentioned in the EXAFS analysis section, this complex would produce a Se-Al bond distance of ~3.4 Å, which was not observed in any of the sorption samples. Instead, the EXAFS data showed evidence for the bidentate binuclear SeO32– species. Based on that, the monodentate biselenite species can be ruled out.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 6. (a):Se K-Edge XANES spectra of selenite sorbed onto Mt, HYA,HAS, HYA-Mt, and HAS-Mt samples. HSeO3 and H2SeO3 solutions are included for comparison. (b) 12.665 to 12.69 keV energy range expanded to more clearly show differences in the near edge spectra for HAS, HYA, HYA-Mt, and Mt samples.

 
It could also be possible that the XANES spectrum of Mt resembles aqueous H2SeO3 because H2SeO3 adsorption into the interlayer of the mineral occurs. This has been proposed by previous researchers (Saha et al., 2004), who predicted that H2SeO3 will be preferentially adsorbed in the interlayer of montmorillonite because HSeO3– and SeO32– are repelled by the negatively charged interlayer. The mechanism of this reaction is unknown, but conceivably could involve hydrogen bonding between H2SeO3 and O from silica tetrahedral layers of the interlayer space. Once H2SeO3 diffuses to the interlayer space, a ligand exchange reaction would also be possible. While the relative importance of interlayer H2SeO3 to the overall XANES spectrum is difficult to assess, the XANES do seem to suggest that such complexation cannot be disregarded.

The important changes in XANES features for SeO32– adsorbed on HYA, HYA-Mt, and Mt are enlarged in Fig. 6b for clarity; HAS, HAS-Mt, and Mt spectra show the same trends. The broadening and weakening of the 12.67 keV XANES feature to a varying extent in the HYA, HAS, HYA-Mt, and HAS-Mt samples suggests that a fraction of the SeO32– is in a surface complex that retains the molecular structure of SeO32– in solution (note the similarity to the HSeO3– spectrum). The two possible sources of such a bonding environment are the presence of unbound (entrained) or weakly bound biselenite in these samples. Since the entrained solution SeO32– levels were <3% of the total Se signal in all samples, weakly bound SeO32– is the most likely source of these spectral features. This also makes sense given the nature of HYA and HAS polymers. Aluminum hydroxide is strongly positively charged at pH 4.5 and so outer-sphere complexation is possible. It has previously been shown that arsenite forms both outer-sphere and inner-sphere complexes on aluminum hydroxide below the mineral's PZC (Arai et al., 2001) (Goldberg and Johnston, 2001), so this type of bonding mechanism is certainly not unprecedented for HYA samples. Since aluminosilicates have a lower PZC than aluminum hydroxides, one would expect relatively more inner-sphere complexes and relatively less outer-sphere complexes on HAS. When the XANES data are examined then it can indeed be observed that the features of HAS are somewhat sharper and better defined than those of HYA. Montmorillonite, on the other hand, has a permanent negative charge from isomorphic substitution. This negative surface charge would make outer-sphere complexation of negatively charged oxyanions quite unlikely. When HYA or HAS polymers are used to pillar the montmorillonite interlayer and coat the planar sites of the smectite then the resulting colloid has intermediate surface behavior. Some of the positive surface charge of the polymers is neutralized by the association with the negatively charged mineral surface, which results in an overall PZSE of pH 4.4 to 4.7 for the polymer-coated montmorillonite sorbents (Saha et al., 2004). Based on that fact alone, one would expect (and can observe) a relatively small amount of outer-sphere complexation in the polymer-coated mineral samples compared to the pure polymers (Fig. 6).

When the EXAFS results and XANES results are considered together, a fairly complete picture of SeO32– adsorption on HYA, HAS, and montmorillonite becomes clear. The proposed bonding mechanisms (consistent with both techniques) are summarized in Fig. 7 . These bonding mechanisms are also completely consistent with the kinetics experiments previously conducted by Saha and coworkers (Saha et al., 2004). In those experiments, they used a fast and a slow reaction to model the kinetics of SeO32– adsorption on pure and coated montmorillonite. They determined that activation energy for SeO32– adsorption on montmorillonite was much greater than the energy of activation for SeO32– adsorption on either HYA or HAS coated montmorillonite. This was the case on Mt, HYA-Mt, and HAS-Mt for both the fast reaction (39, 11, and 19 kJ mol–1, respectively) and the slow reaction (53, 32, and 27 kJ mol–1, respectively) (Saha et al., 2004). This decrease in activation energy was attributed to the presence of positively charged Al-OH2+ functional groups on both HYA and HAS polymers at pH 4.5. These minimize repulsion between the negatively charged biselenite and montmorillonite (Saha et al., 2004). The observation (from XAS) of outer-sphere complex formation on the polymers and the polymer-coated clay is also consistent with the observed decrease in activation energy. Saha and coworkers (Saha et al, 2004) also observed that the rates of reaction were significantly increased on the HYA and HAS coated clay sorbents. This is also consistent with the presence of outer-sphere complexes, which are known (via many pressure-jump chemical relaxation experiments) to have rates of formation that are more rapid than those of inner-sphere complex formation on mineral surfaces.



View larger version (11K):
[in this window]
[in a new window]
  		Fig. 7. Selenite adsorption mechanisms for pH 4.5 samples consistent with EXAFS and XANES spectroscopy.

 
LC-XANES of Selenite Adsorbed on Polymer-Coated Montmorillonite
To estimate the relative distribution of SeO32– on the components of polymer-coated clays, LC-XANES fitting was conducted on the polymer-coated samples from Fig. 6 using the pure polymers and the montmorillonite spectra as standards. The results are shown in Fig. 8 . Error in LC analysis was estimated at 5% by comparing the fit of individual scans to fit results for the overall averaged spectrum. It was observed that ~54% of the adsorbed SeO32– on HYA-Mt was bound to the montmorillonite surface, while on HAS-Mt approximately 42% was bound to the clay mineral phase. The most likely explanation for this difference in sorption is the difference in surface area for the two-coated sorbents. HAS-coated Mt has 487 m2 g–1 surface area while HYA has only 450 m2 g–1; PZSE for the two coated clays in both cases is similar: 4.7 for HYA and 4.4 for HAS (Saha et al., 2004). The surface area differences suggest that in the HAS-Mt sample there is relatively more polymer available for reaction with SeO32– than in the case of the HYA-Mt. It was reported by Saha and coworkers that more Al (1.21 mol Al3+kg–1 Mt) for HAS versus 1.16 mol Al3+kg–1 Mt for HYA-Mt) sorbed on Mt from HAS solutions than HYA solutions during the sorbent synthesis.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 8. Linear combination XANES fits of selenite on HYA and HAS coated montmorillonite. Estimated error for the fitting is ± 5%.

 
There is an important assumption made in the choice of standards used in the above analysis. Because there is both outer-sphere and inner-sphere adsorption in the HYA and HAS samples, using these mixed bonding environment samples as standards assumes that the sorption mechanisms on the polymers is unchanged by interaction with montmorillonite. It is also possible that the interaction with the Mt surface could change the ratio of outer-sphere to inner-sphere on the polymer coating. However, one would expect that if these polymer-Mt interactions were dramatically different for the HAS and HYA samples then it would be reflected in the PZSE of the coated minerals. Instead, the PZSE values are quite similar (4.7 for HYA-Mt and 4.4 for HAS-Mt). In view of their PZSE values (Saha et al., 2004), HAS-Mt should form slightly more inner-sphere complexes and slightly fewer outer-sphere complexes compared with HYA-Mt. This surface charge effect coupled with the increased total amount of polymer on the HAS-Mt sample (via surface area and via mol Al3+ adsorbed) is consistent with the data presented in Figs. 6 and 8.

Larger Significance of this Study
From an environmental standpoint, these results are the first to clearly demonstrate the mechanism of SeO32– adsorption on aluminum-bearing mineral phases. On montmorillonite, only bidentate binuclear inner-sphere complexation was observed. For the hydroxyaluminum and hydroxyaluminosilicate polymers, a mixture of outer-sphere and bidentate binuclear inner-sphere was observed. When montmorillonite was coated with either HYA or HAS polymers then adsorption behavior was intermediate between that of the mineral and the pure polymer. The adsorption behavior of these polymer-coated minerals is especially important to natural systems. Temperate soils often contain aluminum-hydroxy and aluminosilicate coated minerals rather than discrete aluminum hydroxide minerals and pristine clay surfaces. More XAS studies are planned for contaminant sorption on more realistic and complicated sorbents such as organic and inorganic polymer coated clay minerals.

The finding (from XANES) that outer-sphere complexes form on the HAS, HYA, HYA-Mt, and HAS-Mt samples but not on pure Mt is consistent with the macroscopic kinetics experiments previously conducted by Saha and coworkers (Saha et al., 2004). They determined that activation energy for SeO32– adsorption on montmorillonite was much greater than the energy of activation on HYA-Mt or HAS-Mt. They also observed that reaction rates were more rapid on HYA-Mt and HAS-Mt, which is also consistent with the presence of outer-sphere complexes. This effect on kinetics parameters demonstrates that shifts in complexation from outer-sphere to inner-sphere can have a large effect on macroscopic adsorption experiments, and therefore on larger-scale transport of SeO32– in the environment.

The XAS results also show that it is possible to infer bonding mechanisms of SeO32– on mineral phases reliably with XANES as well as with EXAFS spectroscopy. The EXAFS results also suggest that it may be possible to infer the presence of outer-sphere oxyanion complexation in samples where inner-sphere complexes are also present by the amplitude of the EXAFS spectra and the height and shape of the first (Me-O) shell in RSFs. However, XANES spectra remain more conclusive and reliable for observing outer-sphere complex formation. The interpretation of Se XANES data relies on changes in molecular geometry and Se-O bond delocalization that occur as SeO32– changes protonation state in solution or forms inner-sphere complexes on surfaces.

Careful use of XANES spectral features in determination of bonding configuration can extend the utility of X-ray absorption spectroscopy in environmental samples for two reasons. First is the fact that concentrations in natural samples are often too low for EXAFS spectroscopy to be conducted. Additionally, many synchrotron-based studies now utilized microprobe beamlines where XANES spectroscopy is far more practical to conduct than is EXAFS.


   ACKNOWLEDGMENTS
 
This research was supported by funding from Natural Sciences and Engineering Research Council of Canada Discovery grants #261432 and 2383 and by the Saskatchewan Synchrotron Institute. Additionally, assistance in beamline configuration and optimization were provided by Drs. J. Cross, R. Gordon, and S. Heald of PNC-CAT at the Advanced Photon Source at Argonne National Laboratory. PNC-CAT facilities at the Advanced Photon Source, and research at these facilities, are supported by the US DOE Office of Science Grant No. DEFG03-97ER45628, the University of Washington, a major facilities access grant from NSERC, Simon Fraser University and the Advanced Photon Source. Use of the Advanced Photon Source is also supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. W-31-109-Eng-38. D.P. also thanks Dr. S.D. Siciliano (University of Saskatchewan) for his careful review of the work and suggestions.

Received for publication February 21, 2005.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Peak, D.
Right arrow Articles by Huang, P. M.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Peak, D.
Right arrow Articles by Huang, P. M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Peak, D.
Right arrow Articles by Huang, P. M.
Related Collections
Right arrow Soil Chemistry

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome