Selenium Speciation in Whole Sediment using X-ray Absorption

Jun 24, 2010 - ... Abingdon, UK) to minimize thermal disorder and radiation damage. .... EXAFS contribution, combined with poor signal-to-noise from d...
2 downloads 0 Views 1MB Size
Environ. Sci. Technol. 2010, 44, 5389–5394

Selenium Speciation in Whole Sediment using X-ray Absorption Spectroscopy and Micro X-ray Fluorescence Imaging C H E R Y L I . E . W I R A M A N A D E N , †,§ KARSTEN LIBER,† AND I N G R I D J . P I C K E R I N G * ,‡ Toxicology Centre, University of Saskatchewan, 44 Campus Drive, Saskatoon, Saskatchewan S7N 5B3, Canada and Department of Geological Sciences, University of Saskatchewan, 114 Science Place, Saskatoon, Saskatchewan S7N 5E2, Canada

Received March 13, 2010. Revised manuscript received May 26, 2010. Accepted June 2, 2010.

A field survey was conducted in a freshwater lake system in the Athabasca Basin, northern Saskatchewan, Canada that receives treated metal mining and milling process effluent containing elevated levels of selenium. Whole sediment, pore water, surface water, and chironomid larvae were analyzed in an attempt to link whole sediment selenium speciation to various environmental factors, including selenium availability to benthic macro-invertebrates, a trophic level through which selenium can enter the diet of higher trophic level organisms. Speciation was measured using synchrotron-based selenium K-edge X-ray absorption spectroscopy (XAS). All lake averages of sediment samples (reference or exposure sites) contained a significant proportion (approximately 50%) of elemental selenium which is relatively insoluble in water, immobile, and not considered to be bioavailable. The presence of elemental selenium was confirmed by extended X-ray absorption fine structure (EXAFS) analysis of select samples. Inorganic metal selenides were also found in whole sediment samples and confirmed using micro X-ray fluorescence imaging. Dissolved selenium concentrations in pore water were correlated to the amount of selenite in whole sediments provided that the sites were classified according to whole sediment sand content. Sand content itself is likely inversely correlated to sediment organic matter content, adsorption sites, and redox potential.

Introduction Although essential for the health of most organisms, selenium (Se) has the propensity to bioaccumulate, and therefore lower, seemingly innocuous selenium concentrations can have toxic effects at higher trophic levels. The Se water quality guideline set by the Canadian Council of Ministers of the Environment is 1 µg/L (total Se). However, there is growing acceptance of the need for site-specific understanding of Se transfer into the food chain because Se accumulation in aquatic ecosys* Corresponding author phone: (306) 966-5706; fax: (306) 9668593; e-mail: [email protected]. † Toxicology Centre. ‡ Department of Geological Sciences. § Current address: Minnow Environmental, Georgetown, Ontario L7G 3M9, Canada. 10.1021/es100822z

 2010 American Chemical Society

Published on Web 06/24/2010

tems is largely through the dietary pathway (1-3). The continued release of selenium into the aquatic environment can result in its accumulation in sediment. This accumulation has been found to positively correlate with sediment total organic carbon (TOC) content in a study site in northern Saskatchewan (r 2 range 0.38-0.98, p < 0.05 (4)) and elsewhere (5-7). Accumulation of selenium in benthic macro-invertebrates likely occurs predominantly via a sedimentary (and dietary) uptake route, as opposed to direct uptake from the water column (3, 4, 8, 9). However, the amount of selenium available for uptake from sediment also depends on the selenium chemical formsor speciationsin sediment and associated pore water. Thus, total selenium concentration in water or sediment does not necessarily indicate selenium availability to organisms (10). Therefore the speciation of selenium is an important consideration in undertaking risk assessments, predicting selenium uptake and bioaccumulation, and understanding selenium distribution and transfer into the food chain in any ecosystem. The David Creek drainage basin in the Athabasca Basin, northern Saskatchewan, has experienced decades of exposure to treated uranium mining and milling effluent. Thus, from a preventive as well as a remediative perspective, it is necessary that the exposure routes and toxicity to higher organisms, such as fish, be well understood. Selenium concentrations at this location are high enough in predatory fish to be of concern and to have effects (11, 12). Understanding mobility, availability, and accumulation within the environment is therefore essential. While information exists regarding the total selenium concentrations in the environment, there is a lack of knowledge regarding the selenium speciation, largely due to difficulties in measuring speciation in complex, heterogeneous samples. The selenium species in treated mine/mill effluent is transformed by either biological or physicochemical processes, neither of which is fully understood. Selenium can enter the diet of higher trophic level organisms at the level of their benthic macro-invertebrate prey, most of which are less sensitive than their predators to elevated selenium levels. It is therefore important to understand the mechanisms of Se speciation transformation within their surface sediment habitat and the resulting bioavailability and accumulation in resident benthic invertebrates. Determining selenium chemical speciation in a complicated matrix such as lake sediment using synchrotron based X-ray absorption spectroscopy (XAS) is well established (13-15). This technique allows speciation measurements without intrusive methods that perturb the matrix chemistry and selenium speciation itself. The selenium K near-edge spectrum is sensitive to selenium oxidation state and chemical environment. Therefore, XAS allows the measurement of selenium species in challenging matrices, thus providing information that would otherwise be difficult to obtain. In this paper we present selenium speciation measurements of lake sediments from a freshwater lake system in northern Saskatchewan that contains elevated levels of selenium. Sediment is the site where selenium enters the base of the food chain and transfers to higher trophic level species (3). The combination of bulk whole sediment speciation with imaging of selenium-bearing sediment particles thus allows for detailed empirical investigations of selenium geochemistry and the external environmental factors affecting selenium bioavailability. As there is increasing interest in selenium in a variety of environments and VOL. 44, NO. 14, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5389

locations, this speciation study should have broad relevance to other studies of selenium in freshwater systems.

Materials and Methods Study Site. The study site in northern Saskatchewan (57°12′N, 105°41′W), approximately 700 km north of Saskatoon, comprises a group of lakes within the David Creek drainage system receiving treated uranium ore milling effluent (Figure S1 in the Supporting Information (SI)). Wolf Lake receives the effluent directly. From September 19 to 23, 2006, we sampled six sites each at three lakes downstream of Wolf Lake: Fox, Unknown, and Delta Lakes, as well as three sites at David Lake, a reference location upstream of the effluent discharge point. Delta Lake is approximately 10 km downstream of the effluent discharge point. Sampling and Laboratory Analyses. Sampling of surface water, sediment cores (whole sediment and centrifuged pore water; top 2.5 cm of the cores for both), and benthic organisms, and analysis of surface water, and total selenium using inductively coupled plasma-mass spectrometry (ICPMS) were carried out as previously described (4) with the modifications described in SI. Sediment cores (i.d. 4.8 cm) were collected in acrylic core tubes and refrigerated (4 °C) until they could be processed (within 4-10 days). Potential speciation changes due to this storage time and temperature are likely minimal as demonstrated by the long half-lives (decades) of selenite and particulate organo-selenide species in deep ocean waters (16). The top 2.5 cm of sediment in each core tube was later extruded in the laboratory at room temperature and was gently homogenized in a glass beaker so that all subsamplessfor grain size, TOC, total selenium, pore water extraction, and speciationswere not different due to localized differences in sediment character. The subsample for whole sediment selenium speciation analysis was stored in a cryo-vial, in a freezer at -80 °C. Additional sediment characteristics, including sediment percentage sand, silt, clay, and TOC and pore water pH and TOC have been reported elsewhere (4). Speciation Analysis using X-ray Absorption Spectroscopy (XAS). Whole sediment and chironomid samples, stored at -80 °C since isolation, were handled entirely under liquid nitrogen, a temperature known to prevent organic selenides from volatilizing (17). Samples were ground using an agate pestle and mortar and carefully packed into a frozen cuvette ensuring that there were no air spaces. The custom 2-mm path length cuvette was made of a polymer that has no elemental interferences below the bromine K-edge, with an optical window made of metal-free Mylar tape. Once tightly packed, the cuvette was sealed with a drop of liquid glycerol (anhydrous) which quickly froze in liquid nitrogen, and was stored in liquid nitrogen until data acquisition. XAS was performed on beamline 9-3 at the Stanford Synchrotron Radiation Lightsource (SSRL; Menlo Park CA) and on the Hard X-ray Micro Analysis (HXMA) beamline at the Canadian Light Source (CLS; Saskatoon SK, Canada). Source and beamline configurations are given in the SI. All spectra were calibrated at 12658.0 eV, the inflection point of the selenium K-edge, using a foil of hexagonal elemental selenium placed downstream of the sample and simultaneously measured in X-ray transmission. Whole sediment and chironomid spectra were collected in X-ray fluorescence mode using a 30-element germanium detector (Canberra, Meriden CT) equipped with arsenic filters and Soller slits to maintain the total incoming count rate of each element in the pseudolinear regime. Spectra of soluble aqueous dilute (1-5 mM) standards were also collected using fluorescence. Solid powder standards were diluted with boron nitride (99.5%, Alfa Aesar, Ward Hill, MA) and were collected in X-ray transmission mode using nitrogen-filled ion chambers. Nearedge spectra (also called X-ray absorption near-edge spec5390

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 14, 2010

trum, XANES) were collected for all whole sediment and whole body chironomid samples. During data collection, samples were maintained at approximately 10 K in a liquid helium cryostat (Oxford Instruments, Abingdon, UK) to minimize thermal disorder and radiation damage. Selenium’s several stable oxidation states make it susceptible to photoreduction which would appear as changes in the near-edge spectrum with time in the beam. To monitor this, all samples and standards were collected with at least two sweeps. Selenate showed slight reduction to selenite, but no effects were seen on the +4 or lower oxidation states at 10 K. Several Se K near-edge standards were collected at both beamlines and all samples collected at the CLS were adjusted to account for the difference in calibration between beamlines 9-3 and HXMA. Background subtraction, normalization, and data analysis were carried out using the EXAFSPAK program suite. Linear combinations of near-edge spectra of environmentally appropriate standards were fit to the spectra of whole sediments and whole body chironomids. A standard whose fractional contribution to the fit was less than 3 times its estimated standard deviation in the fit was excluded from subsequent fits. Extended X-ray Absorption Fine Structure (EXAFS). EXAFS spectra were collected for whole sediments from Unknown Lake sites 2 and 3. Due to insufficient time for EXAFS collection on all samples, these were chosen for their relatively high total selenium concentrations (45 and 27 µg/g dry weight, respectively) and substantial elemental selenium fractions observed from their near-edges. EXAFS spectra were quantitatively curve fit using the feff phase and amplitude functions (18). Elemental Selenium Fluorescence Spectra. Elemental selenium exhibits several allotropes, including metallic gray elemental selenium and red elemental selenium which forms Se8 rings. Only red elemental selenium was considered here, recognizing the potential for bacteria to respire selenate or selenite to precipitate red elemental selenium (19, 20). As elemental selenium is highly concentrated, it is prone to X-ray fluorescence self-absorption which, for sufficiently large particles, results in distorted near-edge spectra (21). To account for this, fluorescence spectra for a range of elemental selenium spheres were simulated according to Pickering et al. (21) (Figure S2) using a transmittance spectrum of elemental selenium and assuming the density of red monoclinic elemental selenium (4.39 g · cm-3) (22). Additionally, as such elemental selenium fractions deduced from nearedge fitting are under-represented due to beam attenuation by the particle, as previously described (21) these were corrected using the ratio of the simulated spectrum edge jump to that calculated for the same sphere without absorption effects. Micro X-ray Fluorescence Imaging. Synchrotron micro X-ray fluorescence imaging was employed to confirm the presence of selenium species in whole sediments. Fox Lake site 2 was chosen for its higher total selenium concentration. A frozen subsample of the whole sediment sample used for bulk XAS speciation was thawed just prior to acquisition, smeared thinly between two polypropylene films, and placed at 45° to the incident beam. To prevent drying, humid nitrogen gas flowed over the sample continuously throughout data acquisition. Data were acquired at SSRL beamline 2-3 using a Si(220) double crystal monochromator with incident X-rays at 13 keV. A Kirkpatrick-Baez mirror pair (Xradia Inc.) provided a beam spot of 2.5 × 2.5 µm2. A silicon drift Vortex detector (SII nanotechnology USA, Inc.) measured fluorescence from selenium and from other elements including sulfur, iron, nickel, and arsenic. Data were processed using SMAK v0.37 microtoolkit.

FIGURE 1. Linear combination of selenium standards fit to whole sediment spectra from (i) Unknown Lake site 3 and (ii) Fox Lake site 3. Data (filled points), best fit (solid), residual (difference between data and fit, offset dotted), and standards (identified in legend): (a) elemental selenium; (b) selenobis-glutathione; (c) selenite; (d) trimethyl selenonium iodide; and (e) selenomethionine. Standards are scaled according to their contributions to the fits. Whole sediment selenium concentrations were (i) 26.8 ( 10.0 µg/g dry weight (dw); (ii) 3.8 ( 1.1 µg/g dw.

Results

FIGURE 2. (i) Lake averages of dissolved selenium concentrations in surface water and pore water. Error bars represent the spread of total selenium data within each lake, averaged over 3 samples for David Lake, and 6 samples for all other lakes. (ii) Mean selenium species concentrations in whole sediment (green) and field-collected chironomids (blue) (see Tables S1 and S2 for details on individual samples). For lakes that do not have chironomid selenium speciation available, total selenium concentration is shown.

Whole Sediment Selenium Speciation. Selenium K nearedge spectra of whole sediments from the David Creek drainage basin were analyzed to determine the selenium speciation at the various sites. Sediment spectra were fit to spectra of environmentally relevant standards which show clear distinctions (Figure S3). Major peaks are due to dipoleallowed transitions of electrons to unoccupied bound states (1s to 4p). Generally, as valency increases, more energy is necessary to remove a core electron and the major peak shifts to higher energy; concomitantly, the 4p orbitals become less occupied and the peak gains more intensity (13). Therefore, spectral differences reflect different selenium chemical environments. Standards were categorized into five groups, discussed further below: (1) selenite, (2) organic selenides, (3) elemental selenium, (4) inorganic metal selenides, and (5) S-Se-S species modeled by seleno-bis-glutathione. Figure 1 shows near-edge fits for two sites which differ substantially in whole sediment sand content (4). Unknown Lake site 3 (5% sand) showed primarily elemental selenium whereas Fox Lake site 3 (95% sand) showed a high fraction of selenite compared to other sites, but the major selenium form was bound to sulfur (S-Se-S). Tables S1 and S2 present selenium speciation results of whole sediments and benthic macro-invertebrates, respectively. Lake averages of selenium species composition of whole sediments are presented in Figure S4. Selenium species concentration was calculated by multiplying total selenium concentration (4) with the selenium species fractional contribution in the least-squares fit. Figure 2 presents total selenium concentrations in the sampled environmental compartments (whole body chi-

ronomids, whole sediment, surface water, and pore water) with whole body chironomid and whole sediment selenium speciation. Extended X-ray Absorption Fine Structure (EXAFS). EXAFS of whole sediment samples from Unknown Lake sites 2 and 3 provided additional evidence for elemental selenium. EXAFS provides accurate interatomic distances and information on the number and type of atoms around the central absorbing atom. The Se K-edge EXAFS and Fourier transforms are shown in Figure 3. Both samples required a substantial Se-Se component to fit the first shell, at distances identical to that in elemental selenium (Table S3). Consistent with the near-edge fits, Unknown Lake site 3 showed a greater proportion of Se-Se and both samples needed a second ligand type in the average selenium coordination due to mixed selenium environments. This second ligand type fit well to Se-S for Unknown Lake site 3, consistent with nearedge observations of S-Se-S. The additional component for Unknown Lake site 2 could be fit to either Se-C or Se-S whereas the near-edge predicted Se-C and not Se-S species. This uncertainty arises from the predominant Se-Se EXAFS contribution, combined with poor signal-to-noise from dilute selenium and high scattered background. In both cases, however, EXAFS validated the near-edge fits by confirming the presence of elemental selenium and a mixture of selenium species. Micro X-ray Fluorescence Imaging. Micro X-ray fluorescence imaging was used to confirm selenium species fit to the whole sediment near-edge spectra. Several small regions within a Fox Lake site 2 sediment sample were imaged, of which one is shown (Figure 4). The panels show VOL. 44, NO. 14, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5391

FIGURE 3. Se K-edge EXAFS of standards and whole sediments. k3 weighted EXAFS (inset, ordinate scalebar ) 15 Å-3) and EXAFS Fourier transforms, phase corrected for first shell interactions, showing data (solid) and best fit (dashed) using parameters shown in Table S3. (a) Elemental selenium; (b) elemental selenium simulated for 7-µm radius spheres; (c) whole sediment from Unknown Lake site 3; (d) whole sediment from Unknown Lake site 2, with fit 1 (dashed) and fit 2 (dotted); (e) selenomethionine. two views of a high-selenium particle (upper left) with a very low-selenium particle (lower right) in which arsenic cooccurred with iron. Selenium in the upper left particle cooccurred with sulfur, consistent with the near-edge fitting of this site in which 49% of selenium was bound as S-Se-S (Table S1). Another Fox Lake site 2 region (image not shown) revealed a small (