Arsenic Redistribution between Sediments and Water near a Highly

A solution 1 M in sodium phosphate, 0.1 M in ascorbic acid, and adjusted with .... suboxic from 0 to 3 cm, anoxic from 3 to 11 cm, and sulfidic from 1...
0 downloads 0 Views 796KB Size
Environ. Sci. Technol. 2005, 39, 8606-8613

Arsenic Redistribution between Sediments and Water near a Highly Contaminated Source A L I S O N R . K E I M O W I T Z , * ,†,‡ Y A N Z H E N G , †,§ S T E V E N N . C H I L L R U D , † B R I A N M A I L L O U X , †,| H U N B O K J U N G , § M A R T I N S T U T E , †,| A N D H . J A M E S S I M P S O N †,‡ Lamont-Doherty Earth Observatory of Columbia University, 61 Route 9W, Palisades, New York 10964, Department of Earth and Environmental Sciences, Columbia University, New York, New York 11367, School of Earth and Environmental Sciences, Queens College, Flushing, New York 11367, and Department of Environmental Science, Barnard College, Columbia University, New York, New York 10027

Mechanisms controlling arsenic partitioning between sediment, groundwater, porewaters, and surface waters were investigated at the Vineland Chemical Company Superfund site in southern New Jersey. Extensive inorganic and organic arsenic contamination at this site (historical total arsenic >10 000 µg L-1 or >130 µM in groundwater) has spread downstream to the Blackwater Branch, Maurice River, and Union Lake. Stream discharge was measured in the Blackwater Branch, and water samples and sediment cores were obtained from both the stream and the lake. Porewaters and sediments were analyzed for arsenic speciation as well as total arsenic, iron, manganese, and sulfur, and they indicate that geochemical processes controlling mobility of arsenic were different in these two locations. Arsenic partitioning in the Blackwater Branch was consistent with arsenic primarily being controlled by sulfur, whereas in Union Lake, the data were consistent with arsenic being controlled largely by iron. Stream discharge and arsenic concentrations indicate that despite large-scale groundwater extraction and treatment, >99% of arsenic transport away from the site results from continued discharge of high arsenic groundwater to the stream, rather than remobilization of arsenic in stream sediments. Changing redox conditions would be expected to change arsenic retention on sediments. In sulfur-controlled stream sediments, more oxic conditions could oxidize arsenicbearing sulfide minerals, thereby releasing arsenic to porewaters and streamwaters; in iron-controlled lake sediments, more reducing conditions could release arsenic from sediments via reductive dissolution of arsenicbearing iron oxides.

Introduction Arsenic is a toxic metalloid occurring naturally in soils and sediments throughout the world (1). Its mobility is controlled * Corresponding author phone: (854)365-8793; e-mail: [email protected]. † Lamont-Doherty Earth Observatory. ‡ Department of Earth and Environmental Sciences, Columbia University. § School of Earth and Environmental Sciences, Queens College. | Department of Environmental Science, Barnard College. 8606

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 22, 2005

largely by pH and redox changes, and can be present at high concentrations in both natural and contaminated environments under reduced, circumneutral conditions as well as oxidized and reduced alkaline environments such as Mono Lake (2). Arsenic has been widely used in agricultural and manufacturing applications (3) and has thus become a common anthropogenic pollutant. In the subsurface, arsenic mobility is strongly influenced by redox conditions. Release of arsenic from sediments into associated waters has been attributed to reductive dissolution of iron oxyhydroxides to which arsenic had been sorbed (2, 4), although arsenic and iron solubility have been shown to be decoupled in some environments (5-7), indicating that this mechanism is not solely responsible for high arsenic concentrations in reducing groundwaters. Sulfur species have also been shown to influence dissolved arsenic concentrations, particularly under strongly reducing conditions (8) where sulfide can complex arsenic and thereby increase arsenic solubility (9, 10). Sulfide can also remove arsenic from solution by precipitation of sulfide mineral phases (11) or sorption to iron sulfide minerals (12). O’Day et al. (13) recently proposed a conceptual model for arsenic mobility in reducing shallow sediments, which characterizes the behavior of arsenic as either sulfurcontrolled, as observed in this study in the Blackwater Branch (BWB), or iron-controlled, as observed in Union Lake (UL). This published model (13) indicates that under high-iron conditions, aqueous sulfide is rapidly consumed in the formation of iron sulfide minerals (14, 15), while under lowiron conditions, aqueous sulfide is available to complex aqueous arsenic and to form solid arsenic-sulfide minerals. The control of arsenic mobility by either iron or sulfur is determined by the iron/sulfur ratio. This work expands upon this conceptual model. Under low-iron, sulfur-controlled conditions, removal of arsenic from solution occurs primarily through formation of solid arsenic sulfides (16), with aqueous thioarsenites as potential intermediates (10). Under high-iron, iron-controlled conditions, the consumption of sulfide by available iron prevents formation of aqueous thioarsenites (17). Arsenic oxyanions sorb to the iron sulfides that have formed (18), and further pyritization may subsequently occur (19-21) wherein strong As-Fe bonds form (12, 22). The Blackwater Branch and Union Lake (both contaminated by the Vineland Chemical Company) are classified within this system, as are other environments (4, 8). Little is currently known about the geochemical behavior of thioarsenites under natural conditions, although the implied presence of these compounds in both aqueous and sedimentary phases in BWB sediments indicates a need for a better understanding of their behavior.

Site Overview The Vineland Chemical Company manufactured arsenical biocides in Vineland, NJ from 1950 to 1994. During this time arsenic salts were stored improperly, thereby introducing large-scale arsenic contamination to soils and groundwater. A groundwater extraction and treatment plant operating since 2000 removes 6000-7500 m3 day-1 of highly contaminated groundwater and subsequently discharge treated effluent to the BWB, a small freshwater stream that runs along one side of the site (Figure 1); there are no tributaries to the BWB between the Vineland Chemical site and the Maurice River. Since manufacture of arsenicals commenced, the BWB has delivered arsenic downstream into the Maurice River, Union Lake, and Delaware Bay (23). The associated sediments in these surface water bodies are now noteworthy reservoirs of 10.1021/es050727t CCC: $30.25

 2005 American Chemical Society Published on Web 10/12/2005

FIGURE 1. Map of the Vineland Chemical Co. site, showing the location of Vineland within New Jersey, the general region including coring locations marked with stars, and the NPL site with surface water sampling locations numbered. Bar charts show BWB discharge and arsenic concentrations with water sampling locations marked. The EPA standard shown is the 0.13 µmol L-1 (10 µg L-1) standard effective in 2006. Flow in the Maurice River (MR) was obtained from the USGS gauging station in nearby Norma, New Jersey. contaminant arsenic, providing potential for human exposure and ecological damage. Projected total costs for remediation of the highly contaminated Vineland Chemical site are ∼$100 million over three decades, making it one of the most expensive interventions at a NPL site (24).

Methods Field Methods. Stream discharges were measured on July 11, 2004, at four locations along the BWB (Figure 1) using a USGS-type AA current meter. Two cross-channel transects were taken per location. For each transect, velocity measurements were made twice at points spaced 0.5-1 m apart across the stream at 0.6 times the local depth. At these and other locations, surface waters were passed through 0.45 µm filters, acidified to 1% with Optima grade nitric acid, and returned to the laboratory for high resolution inductively coupled plasma mass spectrometry (ICPMS) analysis. All concentrations described herein as dissolved were passed through 0.45 µm filters, although this fraction may contain colloids. On July 11, 2004, two hand piston cores were collected from fine-grained bottom sediments of the BWB adjacent to a swampy area (Figure 1). These fine-grained, highly reduced sediments represent one of two major types of sediments found on the BWB bottom; the other type is primarily sand. The proportions of these two types are not known. On July 12, 2004, three push cores were collected in one of the deepest parts of UL; this ∼5 m deep area was located via bathymetric map and depth finder. The ∼30 cm cores collected reached

a gravel layer underlying this man-made lake (Figure 1). The entire UL water column was oxic (dissolved oxygen >7 mg L-1) at the time of core collection. One core from each location was sectioned within 1 h of collection into 2 cm slices (the depth of which is reported as the central depth) under a nitrogen atmosphere. Each section was centrifuged for 20 min at 5000 rpm, separating porewater and sediment. Centrifuge tubes were opened in a nitrogen atmosphere, and porewaters were filtered with 0.45 µm filters. Residual sediments were sealed and placed on dry ice, and a subsample of the porewaters was immediately analyzed by differential pulse cathodic stripping voltammetry (DPCSV) for As(III) and total inorganic As (25). Residual porewaters were acidified to 1% with Optima grade nitric acid and returned to the laboratory for ICPMS analysis. Laboratory Methods. Sediments were frozen until laboratory analysis. Selected samples were thawed in a nitrogen atmosphere, and aliquots were separated for gamma counting, sequential extraction, and X-ray absorption near edge spectroscopy (XANES). The gamma counting aliquot was dried, ground to a fine powder, and sealed in airtight, 100mL aluminum cans or 4-mL plastic vials for radionuclide analysis. Measurement of 137Cs and other gamma-emitting radionuclides were made using either an intrinsic germanium or a lithium-drifted germanium detector; sediment cores obtained immediately adjacent to cores collected for porewater analyses were also gamma counted. Gamma counting data were used to elucidate sedimentation rate. VOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8607

The XANES aliquot (consisting of damp solids) remained frozen under nitrogen until analysis at the National Synchrotron Light Source (Brookhaven National Laboratory, Upton, NY) on beamline X-11A with an unfocused spot size of approximately 10 mm H by 0.5 mm V and a Si(111) crystal. Standard spectra were collected in transmission mode with an ionization chamber, and sample spectra were collected in fluorescence mode with a Lytle detector after a Ge filter. Samples were wet packed into custom-made polycarbonate sample holders and then sealed with Kapton tape. Next, 6-10 scans were averaged for each sample. A sodium arsenate standard in Kapton tape was placed between the 2 and 3 in. ionization chamber to account for any drift during each run. Normalized and derivative XANES spectra were fit with a linear combination of the following sodium salt standards: arsenate, arsenite, monomethyl arsenate (MMA), and dimethyl arsenate (DMA) as well as arsenopyrite and thioarsenites. The thioarsenite category included both realgar (AsS or As4S4) and orpiment (As2S3) standards; data were not sufficient to distinguish between these minerals. In the samples, peaks attributed to thioarsenites could be due to these minerals, sorption of aqueous thioarsenites (10), or amorphous arsenic sulfides, so these species will be referred to here as thioarsenites; these species all show coordination between As-S, not between As-Fe (12). Optimal fits were determined by minimizing the square of the error between the observed and calculated spectra between 11.85 and 11.96 keV. Because of the relatively low concentrations of MMA and DMA in the samples, the data for these species are combined. For the sequential extraction, 0.2-0.3 g wet sediment, equivalent to 0.02 g-0.07 g dry sediment, was weighed into amber vials under nitrogen. A solution 1 M in sodium phosphate, 0.1 M in ascorbic acid, and adjusted with sodium hydroxide to pH 5 was degassed with nitrogen; 10 mL was added to each sediment sample. The vials were sealed with gastight septa and mixed for 24 h on a wrist-action shaker. The samples were then centrifuged for 10 min at 5000 rpm, and aqueous phases were analyzed for As(III) by DPCSV. Residual extract was retained for ICPMS analysis (26). The sediments were washed with 15 mL of deionized water, centrifuged, and the rinse was discarded, possibly leading to a small but systematic underestimation of total elemental concentrations in sediments. Fifteen milliliters of 1.2 M Optima grade HCl was then added to each vial, and the samples were shaken for 1 h and centrifuged. The HCl extracts were retained for ICPMS analysis (27). Finally, previously leached sediments were digested completely using nitric acid followed by perchloric and hydrofluoric acids (28). Some volatilization of arsenic could have occurred during this step (29), which would lead to systematic underestimation of arsenic in these sediments, but this is unlikely because the sediments were oxidized by this point in the digestion procedure. These latter digests were refluxed repeatedly on a hotplate until no visible sediment remained, dried down, and then diluted for ICPMS analysis. A laboratory blank showed all elements below the analytical limits of detection, two sets of duplicate digests gave results consistent within 10%, and duplicate analyses of one digest gave variability 70% of arsenic in all samples examined. Although the XANES and phosphate extraction examine somewhat different reservoirs of arsenic, both indicate predominantly reduced arsenic in the sediment (Figure 2).

Discussion Blackwater Branch Core. This core can be subdivided qualitatively into three zones: suboxic from 0 to 3 cm, anoxic from 3 to 11 cm, and sulfidic from 11 to 31 cm. Neither Eh nor dissolved oxygen (DO) data are available to confirm these

classifications; however, these divisions are descriptive of and supported by the geochemical conditions observed. In the suboxic zone, the geochemistry is probably controlled by diffusion of DO from the oxygen-saturated streamwater and by sediment deposition on the stream bottom. The relatively oxic nature of this zone is supported by the relatively large percentage of sediment iron extracted by HCl, containing primarily iron oxides (27), although no orange-colored sediments were observed. The low dissolved arsenic concentrations in this depth interval are consistent with effective sorption onto these oxides (34) (Figure 2). Sediment deposition probably provides a large portion of sediment sulfur, manganese, iron, and arsenic to this uppermost zone of the core. Particle deposition could account for local maxima of sediment sulfur and arsenic at 3 cm if upstream, reduced, sulfidic sediments were resuspended and subsequently redeposited at this location more rapidly than oxidation could occur. The local maximum of porewater sulfur at 1 cm could be explained by oxidation of redeposited sulfidic sediments. The maxima of iron and manganese in the sediment at 3 cm are probably due to a combination of particle deposition and redox trapping (i.e., capture of redoxsensitive metals at a redox boundary in the sediment) of diffusive porewater fluxes of these elements from the anoxic zone below. In the anoxic zone from 3 to 11 cm, geochemical conditions appear to reflect a reducing but not strongly sulfidic environment. There are broad maxima of dissolved iron and manganese consistent with these elements being in primarily soluble (+2) oxidation states. Dissolved manVOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8609

ganese and iron concentrations gradually decrease below ∼7 cm. In the sulfidic zone below 11 cm, sulfur concentrations and speciation dominate the geochemical conditions. The source of sulfur at these depths may be groundwater influx: the hydraulic gradient between groundwater and streamwater in this section of the BWB indicates potential for groundwater discharge. Additionally, total sulfur concentrations in the two deepest porewater samples, 105 ( 4 µM, are in relatively good agreement with total sulfur concentrations in groundwater from four nearby wells, 168 ( 85 µM; these concentrations are sufficient to support significant sulfate reduction. At 31 cm, the largest fraction of arsenate was observed in the porewaters and on the sediments, consistent with an influx of suboxic or oxic groundwaters. However, the high fractions of As(III) in the porewaters and of thioarsenites detected by XANES in the sediments above 31 cm are consistent with reducing conditions in this zone of the core. The decrease in porewater sulfur concentrations between 31 and 11 cm is consistent with reduction of incoming groundwater sulfate to sulfide in this depth interval and precipitation of some fraction of this sulfide (14). Union Lake Core. The Union Lake core can be subdivided into two depth intervals: the suboxic zone from 0 to 3 cm and the anoxic zone from 3 to 29 cm. In the suboxic zone, oxygen should diffuse into the core from the overlying oxygenated lake water, although no orange iron-oxide type sediments were observed. This is consistent with lower dissolved iron and manganese concentrations observed at depths less than 5 cm and with sediment maxima for these elements at 3 cm caused by trapping at a redox boundary, especially in the HCl extractable iron fraction, that is, amorphous iron oxides (27). A major fraction (>35%) of sediment arsenic in this zone is arsenate, also consistent with relatively oxic conditions in this zone of the core. Sulfur (probably in the form of sulfate) appears to be diffusing into the core from the lake water above; in uppermost porewaters and lake bottom waters concentrations of total S measured by ICPMS are similar (77 and 87 µM, respectively) (31). The geochemical environment below ∼3 cm in the UL core indicates anoxic conditions. In this anoxic zone, dissolved total sulfur concentrations decrease downward and sediment sulfur concentrations increase, consistent with diffusion of sulfate from above, reduction of sulfate to sulfide, and subsequent precipitation of sulfide minerals in deeper sediments (22). In this anoxic depth zone, dissolved iron concentrations are constant and high (∼300 µM), consistent with strongly reducing conditions and insufficient sulfide to remove a large fraction of iron via iron sulfide precipitation. Amorphous FeS, a relevant phase under low-temperature reducing conditions (14), is very undersaturated when the saturation index is calculated assuming all dissolved sulfur is sulfide and using the most recent data of Benning et al. for the Ksp (35). Mechanisms Controlling Arsenic Mobility. Arsenic mobility in the BWB and UL was expected to be either ironcontrolled or sulfur-controlled (13). In the BWB, Fe/S ratios were low in sediments and porewaters implying sulfurcontrolled arsenic mobility; in UL, Fe/S ratios were high implying iron-controlled arsenic mobility (Figure 3). To more quantitatively examine controls on arsenic mobility, porewater arsenic data from each sediment core were examined for covariance with porewater manganese, iron, and sulfur; with these elements and with arsenic from each step of the sediment extractions; and with linear combinations of two of these variables. In the Blackwater Branch sediment core, the covariance (Table 1) between porewater and arsenic phosphate-extractable As(III) (but not total phosphate-extractable arsenic, R ) 0.43, p > 0.05) implies that this is a primary reservoir of 8610

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 22, 2005

FIGURE 3. Comparison of Fe/S ratios from the Blackwater Branch, the Clark Fork River, Montana (8), Union Lake, and Balmer Lake, Ontario (4). Sediment data are displayed in the upper panel and porewater data in the lower panel; only data from below the depth of oxygen penetration are shown. The BWB and the Clark Fork River are sulfur-controlled environments, and UL and Balmer Lake are iron-controlled environments. The horizontal line across each box represents the median value, the upper and lower limits of each box represent the 25th and 75th percentiles, error bars represent 10th and 90th percentiles, and outlier points are shown.

TABLE 1. Significant (p < 0.01) Correlations between Porewater Arsenic and Other Variables Correlations with BWB [As]aq [As(III)]PO4 [S]aq [∑S]sed

R ) 0.67 R ) 0.70 R ) 0.69

Correlations with UL [As]aq [S]aq [Fe]aq [Fe]PO4

R ) -0.65 R ) 0.67 R ) 0.64

sediment-water exchange. The correlation of porewater arsenic with both porewater and sediment sulfur supported the designation of the geochemical conditions within this core as sulfur-controlled (Table 1). Correlations between porewater arsenic and linear combinations of two variables, as shown below for the UL core, were all weaker (R < 0.4) than these univariate correlations. Below 3 cm in the BWB core, that is, if the porewater sulfur peak at 1 cm is ignored, the R value between porewater arsenic and sulfur increases to 0.84 (p < 10-4). The UL core, by comparison, showed an inverse correlation between dissolved arsenic and sulfur (Table 1) consistent with dissolved sulfur and arsenic chemistry in this core being controlled by different factors. The positive correlations between dissolved arsenic and dissolved and phosphate-extractable iron in the UL core are consistent with iron-controlled arsenic mobility in this core. Sulfur-controlled arsenic mobility in the BWB core and iron-controlled arsenic mobility in the UL core can also be explained by examining geochemical conditions in each. In the BWB sediment core, arsenic concentrations are consistently high (>10 µM) only in the depth interval below 11 cm, where total dissolved sulfur concentrations are elevated, and may support sulfate reduction. At sulfide concentrations above 50 µM, thioarsenite species dominate, and the total solubility of arsenic is increased (10). Both manganese and

iron are elevated in porewaters well above the depth at which arsenic concentrations are elevated; this suggests that the decrease in arsenic concentrations at above 11 cm is associated with the decrease in sulfur concentrations at this same depth, and not with redox conditions that control concentrations of iron and manganese. The correlation of porewater arsenic and sediment sulfur (Table 1) implies some exchange between dissolved arsenic and sulfur-bearing sediment phases, most likely sorbed or amorphous thioarsenites. Large fractions of solid thioarsenites observed by XANES (Figure 2) support this conclusion and are consistent with previous predictions (13). If arsenic mobility in the UL core is iron-controlled, as indicated by the data examined thus far, then arsenic oxyanions would be expected to be sorbed to minerals (22). Phosphate and HCl extractable iron are plausible phases to which arsenic might be sorbed, and which therefore could control arsenic levels in UL porewaters. Data for porewater arsenic in UL correlated best with the following linear combination of two sediment variables: [As]aq ) A[Fe]PO4 + B[Fe]HCl, where A and B are adjustable empirical constants derived by optimizing the equation to maximize correlation and minimize residuals between predicted and measured porewater arsenic (A ) 0.057, B ) -0.012). This equation improved the correlation with observed porewater arsenic concentrations to an R value of 0.85 (p ) 0.007). The signs of the coefficients also agree with expectations: phosphate extractable iron phases, that is, those containing weakly bound arsenic, are a source of arsenic to porewaters (thus A > 0); while HCl extractable iron phases, that is, iron oxides known to tightly sorb arsenic oxyanions (34, 36), are a sink of arsenic from porewaters (27), and B < 0. XANES data show a high proportion of sediment arsenite throughout this Union Lake core and an increasing proportion of thioarsenites with depth, supporting the conclusions that arsenic is primarily present as sorbed oxyanions in this iron-controlled core, possibly with some transformations to arsenic- and sulfurcontaining minerals with depth. Data indicate that arsenic is sequestered in this core via sorption, which may allow redox-controlled movement of arsenic within the sediment column, similar to that seen in Coeur d’Alene Lake (37). This mobility would be inhibited if Fe-As-S phases that are intermediates to pyritization (12, 38) were forming, but XANES data do not indicate formation of Fe-As bonds (12), consistent with no formation of these phases. Comparison of the Blackwater Branch and Union Lake Cores. In the sulfur-controlled BWB core, 57% (median value) of sediment arsenic was phosphate extractable as compared to 82% of sediment arsenic in the UL core. Phosphate is itself an oxyanion, which releases arsenic by competition for surface sorption sites (27); the greater proportion of arsenic released from the UL core by phosphate extraction is consistent with phosphate more effectively displacing arsenic oxyanions than thioarsenites. This difference may also indicate that the arsenic-sulfur species formed in the BWB are more recalcitrant than the sorbed arsenic oxyanions primarily present in the UL core. A fraction of arsenic sulfides was observed by XANES on both cores: 50-90% in the BWB core and 10-60% in the UL core. Within thioarsenites, realgar (AsS) may be more prevalent in the UL core because this system is so iron-rich, and, conversely, orpiment (As2S3) may be more prevalent in the BWB core because this system appears to be sulfurcontrolled and (relatively) iron poor (13), although the XANES data do not allow us to confirm these predictions. XANES data are consistent with precipitation of thioarsenites in the BWB (not with sorption of arsenic to iron sulfides, e.g., (12)) and weaker sorption of arsenic oxyanions without subsequent mineralization in UL.

The most obvious geochemical differences between these two cores are the high sulfur in the porewaters of the BWB core and the very high iron in the porewaters and sediment of the UL core (Figures 2 and 3). The total sulfur (sediment plus porewaters) in the BWB core is only ∼1.5 times higher than that in the UL core. The total iron in the UL core, however, is 3-10 times higher than that in the BWB core (Figure 3). This implies that the high iron in the UL core removes sulfide from solution, preventing build-up of high aqueous sulfur concentrations and resultant formation of aqueous thioarsenites. The data in this study do not address the cause of the high iron in UL sediments, but one possibility is that particle fluxes in the BWB have lower iron abundances than those delivered via the Maurice River to UL. Comparison with Previously Published Data. The data presented herein are consistent with predictions made by O’Day et al. (13) that with lower Fe/S ratios, arsenic would be sulfur-controlled, e.g., the BWB, and with relatively high Fe/S ratios, it would be iron-controlled, e.g., UL. For the data in this study, both porewater and sediment Fe/S ratios fit this prediction (Figure 3). Iron and sulfur (sulfate + sulfide) concentrations in sediment and porewaters were estimated. These estimates were made from figures published for the Clark Fork River, Montana (8), and from the August 1999 data (the only dataset with sulfide measurements) from Balmer Lake, Ontario (4). Conditions in porewaters of the Clark Fork River were described as sulfidic (8), and arsenic mobility in the sediments beneath Balmer Lake was attributed to dissolution of iron phases (4). Sediment Fe/S ratios fit the paradigm described above; the two sites that may be classified as iron-controlled, Balmer Lake and UL, have similar and high sediment Fe/S ratios, while the sulfur-controlled Clark Fork River and BWB have much lower sediment Fe/S ratios. The porewater Fe/S ratios for the Clark Fork River and the BWB are also similar to one another and relatively low. However, the porewater Fe/S ratios of Balmer Lake do not fit the paradigm of high Fe/S ratios at iron-controlled sites. This is likely due to mining impacts on this lake and resultant very high porewater sulfate concentrations of up to ∼350 mg L-1 (4). The geochemical environments at the Clark Fork River (8), Balmer Lake (4), and East Palo Alto (13) are all appreciably different from those near the Vineland Chemical site. However, classifications based on Fe/S ratios appear to be instructive as to mechanisms influencing arsenic mobility. These mechanisms are significant because the type of events that might mobilize arsenic would be somewhat different; in iron-controlled environments, anoxia of overlying waters would release arsenic by altering speciation of sorbed arsenic species or dissolution of host minerals, whereas in sulfurcontrolled environments, exposure of sediments to oxic conditions could release arsenic by dissolution of authigenic arsenic sulfide phases. Sources of Offsite Arsenic Transport. The BWB serves as an integrator of most of the arsenic transported downstream from the Vineland site. There are four plausible sources of arsenic transported downstream in the BWB: (a) erosion and overland transport of arsenic-bearing soil minerals, (b) effluent from the pump and treat plant which is discharged to the stream, (c) remobilization of sediment-bound arsenic in the BWB, and (d) continued high-arsenic groundwater discharge from the Vineland Chemical Company site into the BWB. Total arsenic transported downstream was estimated from the streamflow and arsenic concentration measurements (Figure 1) to be ∼22 ( 3 mol (1.64 kg) day-1. Overland transport can be eliminated as a primary source of arsenic to the BWB due to the dry conditions on the day of stream sampling and the absence of surface water tributaries to the BWB between the site and the Maurice VOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8611

River (30). Plant effluent contributes ∼0.64 mol day-1, less than ∼3% of the total downstream transport. An upper bound on remobilization of sediment-bound arsenic can be estimated by calculating the diffusive flux of arsenic from the porewaters using Fick’s Law (39). This diffusive flux is ∼4.5 µmol arsenic m-2 day-1 or 35 mmol arsenic day-1 (using stream length of 1.4 km and width of 5.5 ( 0.2 m). Even if the diffusive flux from porewaters were an order of magnitude greater than this estimate, it would represent only ∼1.6% of total downstream arsenic flux. Similar stream and porewater measurements were made in the summer of 2003 (40), and the diffusive flux was estimated as ∼200 mmol day-1,