Effective Isotopic Fractionation Factors for Solute Removal by Reactive

Oct 8, 2008 - heavier isotopes relative to lighter isotopes. This enrichment ... delta notation, which expresses a measured isotope ratio as the per m...
0 downloads 0 Views 178KB Size
Environ. Sci. Technol. 2008, 42, 7850–7855

Effective Isotopic Fractionation Factors for Solute Removal by Reactive Sediments: A Laboratory Microcosm and Slurry Study SCOTT K. CLARK* AND THOMAS M. JOHNSON Department of Geology, University of Illinois at Urbana-Champaign, 245 Natural History Building, 1301 West Green Street, Urbana, Illinois 61801

Received June 30, 2008. Revised manuscript received September 08, 2008. Accepted September 16, 2008.

reaction rates than heavier isotopes. As reduction proceeds, the remaining reactant becomes progressively enriched in heavier isotopes relative to lighter isotopes. This enrichment is quantified by measuring the ratio, R, of the abundance of a heavier isotope to that of a lighter isotope. The Rayleigh distillation equation, which relates R to progress of chemical reactions (6-8) can be expressed ln

( ) ( )

R(t) 1 ) - 1 × ln(f(t)) R0 R

where R(t) and R0 are the isotope ratios of the reactant at time t and before any reduction, respectively, f(t) is the fraction of reactant remaining; and R is the fractionation factor that expresses the magnitude of the kinetic isotope effect: R)

Wetlands remove many dissolved pollutants from surface waters by various mechanisms. Stable isotope ratio measurements may provide a means of detecting and possibly quantifying certain removal processes, such as reduction of SeO42-, Cr(VI), NO3-, and HClO4-, that fractionate isotopes. However, the magnitude of the isotopic fractionation for a given reaction depends on the setting in which it occurs. We explore the case where isotope ratio shifts in surface waters are used to detect or quantify reactions occurring in pore waters of underlying sediments. A series of SeO42- reduction experiments reveals that the effective isotopic fractionation, observed in the water column as a result of SeO42- diffusion into underlying, Se-reducing sediments, is weaker than the intrinsic fractionation induced by the same reduction reactions in well-mixed systems in which reaction sites are not separated from measured SeO42-. An intact sediment core yielded an effective ε (≈ δreact - δinstantaneous prod) of 0.20‰, whereas the intrinsic ε was 0.61‰. These results are consistent with previously published reactive transport models. Isotopic studies of sediment-hosted reactions in wetlands and other surface water systems should use the smaller effective fractionation values, which can be estimated using the models.

Introduction Wetlands can remove many pollutants from waters (e.g., Se(VI), Cr(VI), NO3-, U(VI), HClO4-, TCE, MTBE, and nitroaromatic compounds); redox reactions may cause part or most of the removal. Se, Cr, and U are less mobile in their reduced states (1-3), whereas NO3- and HClO4- are transformed into innocuous species under reducing conditions (4, 5). Concentration measurements are often useful in quantifying reduction rates in these environments. However, mixing, diffusion, and advection can affect concentrations, and thus decreases in concentrations do not reliably indicate reduction of pollutants. Isotope fractionation studies offer an alternative method of estimating reduction reactions that is independent of concentration analyses. Reaction rates of an element’s isotopes differ, with lighter isotopes typically having greater * Corresponding author phone: (517) 353-4524; fax: (517) 3538787; e-mail: [email protected]. Current address: Department of Geological Sciences, Michigan State University, 206 Natural Science Building, East Lansing, MI. 7850

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008

(1)

Rreact Rinst prod

(2)

where Rreact and Rinst prod are the isotope ratios of the reactant pool and the instantaneous reaction product, respectively, at one instant in time (9). “Inst prod” is used to distinguish between an incremental reaction product and a cumulative product. The Rayleigh equation is strictly applicable only in a well-mixed, closed system with no back-reaction of products and a constant R (8, 10, 11). Isotope ratio measurements are commonly reported in delta notation, which expresses a measured isotope ratio as the per mil deviation from an interlaboratory standard, Rstd: δsample )

(

)

Rsample - 1 × 1000(‰) Rstd

(3)

The fractionation factor can similarly be expressed in per mil units by defining another parameter, the ε value (8): ε ≡ (R - 1) × 1000(‰)

(4)

This parameter is particularly useful because, to a close approximation, ε ≈ δreact - δinst prod. With knowledge of the initial isotopic composition of the reactant and R, one need measure only δreact to determine the extent of reduction, provided R is correct for the conditions being studied. Because kinetic isotope effects are sensitive to reaction mechanisms and rates (12-15), experimentally determined fractionation factors are expected to vary depending on the reaction studied (e.g., bacterial versus abiotic reaction) and the reaction conditions. Fractionation factors for microbially mediated reactions have been determined for various elements and compounds, such as reduction of Se oxyanions (16) and perchlorate (4) by pure microbial cultures, and the break-down of complex organics such as TCE (17) and MTBE (18-20) by microbially enriched microcosms. Research on abiotically induced fractionation factors includes reduction of Cr(VI) by continuously mixed suspensions of magnetite and sediments (21-23), SeO42- by green rust (24), U(VI) by zerovalent iron (25), and nitroaromatic compounds by Fe(II)/goethite and juglone/H2S (26). Our goal is to extend the Rayleigh distillation model to quantify removal of contaminants from surface water bodies when removal is driven by reactions occurring in pore waters of underlying sediments. As a contaminant is removed from water moving through a wetland or stream, isotopic fractionation induced by reactions in pore waters will be manifested in the surface water. Measured isotope ratios in the surface water should change as a function of the fraction of contaminant removed. As we show below, that function 10.1021/es801814v CCC: $40.75

 2008 American Chemical Society

Published on Web 10/08/2008

is the Rayleigh model, but the observed isotopic fractionation is diminished relative to that for the same reaction when reaction sites are in intimate contact with the surface water in a well-mixed system. We refer to this latter factor as the intrinsic fractionation factor, Rint (27, 28). The effective fractionation factor, Reff (29), expresses the weaker isotopic fractionation observed when the studied system is not homogeneous and the measured reactant pool is not in intimate contact with reaction sites. In this paper, we present results of several laboratory batch experiments that determine Reff for selenate reduction in sediments of a littoral wetland. We incubated unamended sediments as fully mixed suspensions, and as intact or homogenized static sediment beds underlying water columns. We show that theoretical expressions for Reff, developed in studies of O2 reduction and denitrification in marine environments (27-30), agree with the experimental results, and we discuss implications for studies of contaminant transformation in surface waterssediment systems.

Experimental Methods An overview of the field and experimental methods is provided below. A more complete description is presented in the Supporting Information. Collection Methods. An intact sediment core for a microcosm study and a grab sample for sediment slurries were collected from under approximately 30 cm of water in a littoral wetland area of Sweitzer Lake. The lake is a 70-ha man-made reservoir with a history of elevated Se levels (31-33) located south of Delta, CO. Inflow at Sweitzer Lake is dominated by a managed allotment from an irrigation canal, which is fed by the Uncompahgre River during the growing season. Experiments. The intact sediment core experiment was initiated by injecting a Na2SeO4 solution into the surface water to achieve an initial Se(VI) concentration of approximately 960 ng/mL. By the 15th day, the Se(VI) concentration decreased to less than 1% of the initial loading. The experiment was then repeated using the same initial Se(VI) concentration. Five simultaneous reduction experiments were performed using homogenized sediments from the grab sample: three anoxic slurry incubations, an anoxic sediment bed, and a sediment bed with the overlying waters exposed to the air. Each of these experiments consisted of 4 g of sediment and 100 mL of Sweitzer Lake water inside 120 mL serum bottles. One anoxic bottle was autoclaved prior to adding Se. Anoxic slurries were continuously mixed on an orbital shaker table at 120 rpm, following the method described in Ellis (34). To create a nonreactive, well-mixed water column (approximately 4 cm deep) over a relatively thin (approximately 0.4 cm thick), homogenized sediment bed, bottles for the two sediment bed experiments were homogenized overnight on a shaker table and then sediments were allowed to settle. The headspace in one of these bottles was equilibrated with the atmosphere and remained open to the atmosphere during the experiment. Initial Se(VI) concentrations in the mixed slurries and sediment bed experiments were approximately 2500 ng/mL. Loadings with Se(VI) concentrations 100-250× greater than the typical dissolved Se(VI) concentrations at Sweitzer Lake enabled collection of measurable quantities of Se(VI) throughout the duration of the experiments. High Se concentrations may have affected the microbial community and thereby affected the reduction rates and Rint (12, 35). Otherwise, experiments were designed to minimize introduced variables. Concentration Measurements. Dissolved Se concentrations were first determined by hydride-generation atomic absorption spectrometry (HG-AAS). Using isotope dilution

(ID), dissolved Se(VI) concentrations were later determined to better precision on the subset of samples analyzed for isotope ratios. Total Se and Se(VI) concentrations were measured by HG-AAS on the intact sediment core samples, with Se(IV) measured on the second loading of the intact sediment core. Total Se, but not Se(VI), concentrations were measured by HG-AAS for slurry and sediment bed samples. For consistency, only Se(VI) ID results are used in data interpretation. Isotope Ratio Measurements. This is the first Se isotope study applying a double spike method on a MC-ICP-MS. Isotope ratios were obtained from filtered sample aliquots containing approximately 50 ng of Se(VI), which were mixed with a Se(VI) double isotope spike consisting primarily of 74Se and 82Se (36). The spike was added prior to any processing steps, other than filtering. Once added, the double spike provided an internal correction for any fractionation that might occur during sample processing (37, 38), and corrected for mass discrimination during MC-ICP-MS analysis (36). Selenium (VI) was then isolated from the sample matrix via anion-exchange resin columns and a hydride generation step. Isotope ratio analyses were performed on a Nu Plasma HR multicollector inductively coupled plasma mass spectrometer (MC-ICP-MS) located in the Department of Geology at the University of Illinois. To make 78Se/76Se results measured here directly comparable to other published Se isotope values, we also report 80Se/76Se in Table S1 in the Supporting Information. (Details of MC-ICP-MS analysis and data reduction procedures are given in the Supporting Information). Quality control data on isotope measurements from this study and a concurrent field study from Sweitzer Lake include 21 replicate samples, 274 measurements of unprocessed SRM 3149 standard (δ78/76Se(‰) ) 0.00 ( 0.13‰, 2σ), 16 processed SRM 3149 measurements (δ78/76Se(‰) )0.00‰ ( 0.12‰, 2σ), and 19 measurements of a provisional interlaboratory standard, MH495 (δ78/76Se(‰)SRM ) -1.01‰ ( 0.11‰, 2σ). Based on the root-mean-square (rms) difference of 21 replicate sample preparations, sample uncertainty in δ78/76Se(‰) is ( 0.15‰. Three samples (continuously mixed slurry at 60 h, anoxic sediment bed at 48 h and 120 h) were assigned a 2σ uncertainty of ( 0.27‰ due to low signal intensities (see Supporting Information). In calculating the uncertainty in the ε value for the anoxic sediment bed experiment, a 0.22‰, 2σ uncertainty was applied to the 84 h sample because the concentration and isotope ratio for the replicate analysis were outside the stated 2σ uncertainty range. The 78Se/76Se ratio in the autoclaved sample collected at 168 h was elevated by 0.26‰ relative to the initial isotope ratio without a concentration change relative to the initial loading. However, precision based on rms differences from all replicates and because concentration and isotope trends can be fitted to a Rayleigh distillation model for each experiment suggest that the overall concentration and isotope results are robust.

Results and Discussion Concentrations. Selenium (VI) concentrations (Figure 1; Table S1) decreased by at least 86% in all experiments except for the autoclaved slurry, in which no Se(VI) concentration change was observed. This indicates that Se(VI) concentrations decreased due to microbial respiration of Se(VI) in sediments. Se(IV) is assumed to be the immediate product of Se(VI) reduction. Dissolved Se(IV) concentrations, calculated in the second loading of the intact sediment core, were