Arsenic mobility in groundwater surface water systems in carbonate rich pleistocene glacial drift aquife

Arsenic mobility in groundwater surface water systems in carbonate rich pleistocene glacial drift aquife

Arsenic mobility in groundwater/surface water systems in carbonate-rich Pleistocene glacial drift aquifers (Michigan) Kathryn Szramek a, *, Lynn M. Walter a , Patti McCall b a Department of Geological Sciences, 2534 C.C. Little Bldg., University of Michigan, Ann Arbor, MI 48109, USA b Insight Environmental Services, Inc., 5892 Sterling Drive, Howell, MI 48843, USA Abstract Within the Lower Peninsula of Michigan, groundwaters from the Marshall Formation (Mississippian) contain As derived from As-rich pyrites, often exceeding the World Heath Organization drinking water limit of 10 mg/L. Many Michigan watersheds, established on top of Pleistocene glacial drift derived from erosion of the underlying Marshall Formation, also have waters with elevated As. The Huron River watershed in southeastern Lower Michigan is a well characterized hydrogeochemical system of glacial drift deposits, proximate to the Marshall Fm. subcrop, which hosts carbonate-rich groundwaters, streams, and wetlands (fens), and well-developed soil profiles. Aqueous and solid phase geochemistry was determined for soils, soil waters, surface waters (streams and fens) and groundwaters from glacial drift aquifers to better understand the hydrogeologic and chemical controls on As mobility. Soil profiles established on the glacial drift exhibit enrichment in both Fe and As in the oxyhydroxide-rich zone of accumulation. The amounts of Fe and As present as oxyhydroxides are comparable to those reported from bulk Marshall Fm. core samples by pre- vious workers. However, the As host in core samples is largely unaltered pyrite and arsenopyrite. This suggests that the transformation of Fe sulfides to Fe oxyhydroxides largely retains As and Fe at the oxidative weathering site. Groundwaters have the highest As values of all the waters sampled, and many were at or above the World Health limit. Most groundwaters are anaerobic, within the zones of Fe 3+ and As(V) reduction. Although reduction of Fe(III) oxy- hydroxides is the probable source of As, there is no correlation between As and Fe concentrations. The As/Fe mole ratios in drift groundwaters are about an order of magnitude greater than those in soil profiles, suggesting that As is more mobile than Fe. This is consistent with the dominance of As(III) in these groundwaters and with the partitioning of Fe 2+ into carbonate cements. Soil waters have very low As and Fe contents, consistent with the stability of oxy- hydroxides under oxidizing vadose conditions. When CO 2 charged groundwaters discharge in streams and fens, dis- solved As is effectively removed by adsorption onto Fe-oxides or carbonate marls. Although Fe does not display conservative behavior with As in groundwaters, a strong positive correlation exists between As and Sr concentrations. As water–rock interactions proceed, the As/Fe and Sr/Ca ratios would be expected to increase because both As and Sr behave as incompatible elements. Comparisons with groundwater chemistries from other drift-hosted aquifers proximate to the Marshall sandstone are consistent with these relations. Thus, the Sr content of carbonate-rich groundwaters may provide useful constraints on the occurrence, origin and evolution of dissolved As in such systems. # 2004 Elsevier Ltd. All rights reserved. 0883-2927/$ - see front matter # 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.apgeochem.2004.01.012 Applied Geochemistry 19 (2004) 1137–1155 www.elsevier.com/locate/apgeochem * Corresponding author. E-mail addresses: kszramek@umich.edu (K. Szramek), lmwalter@umich.edu (L.M. Walter). 1. Introduction Elevated As levels in surface and groundwater systems can be derived from both anthropogenic and natural sources. Although anthropogenic As contamination from mining operations, fossil fuel processing, and pes- ticides/herbicides applications is typically local in extent, contamination can reach levels thousands of times of that from natural sources (e.g. Smedley and Kinniburgh, 2002). Natural As sources have recently received increas- ing attention due to the discovery of regional-scale As contamination of groundwaters, with As enrichment far above the World Health Organization (WHO) max- imum contamination limit (MCL) of 10 mg/L (0.13 mM) in drinking water (Welch et al., 1999, 2000; Nordstrom, 2002; Smedley and Kinniburgh, 2002). Perhaps the most widely known problem of naturally occurring As enrichment of groundwater occurs in unconsolidated deltaic sediments in Bangladesh. Here, potentially 30 Â 10 6 people have been exposed to levels of As up to 2500 mg/L in the groundwater (Nordstrom, 2002). Arsenic is not an abundant element in the earth’s continental crust (Wedepohl, 1995). It can, however, be concentrated in sulfide-bearing minerals such as pyrite (Savage et al., 2000). The most common sources of As in the natural environment include volcanic rocks, specifi- cally their weathering products and ash (Nicolli et al., 1989; Smedley et al., 2002); marine sedimentary rocks (Smedley and Kinniburgh, 2002); hydrothermal ore deposits and associated geothermal waters (Korte and Fernando, 1991); and fossil fuels, including coals and petroleum (Korte and Fernando, 1991; Smedley and Kinniburgh, 2002). Although igneous and metamorphic rocks contain As, the average concentrations (1.5 and 5 mg kg À1 , respectively) (Ure and Berrow, 1982; Smedley and Kinniburgh, 2002) are lower than the average range from sedimentary deposits (5–10 mg kg À1 )(Webster, 1999). Typically the As concentrations in sedimentary rocks increase with increasing amounts of sulfide minerals, oxides, organic matter and clays (Smedley and Kinniburgh, 2002). Arsenic has 4 oxidation states in aquatic systems, ÀIII, 0, +III and +V with the two main inorganic species found in water being arsenite (III) and arsenate (V) (e.g. Cullen and Reimer, 1989; Drever, 1997; Kim, 1999; Stollenwerk, 2003). Thermodynamics predicts that arsenite is stable under reduced conditions and arsenate is stable under oxidized conditions. However, both spe- cies can be found regardless of the redox conditions, suggesting that kinetic or microbial processes are impor- tant controls on speciation (Smedley and Kinniburgh, 2002; Stollenwerk, 2003). The geochemical behavior of arsenate is often compared to that of phosphate, while As acid is comparable with boric acid (e.g. Drever, 1997). Thus, arsenate is much less mobile under intermediate pH conditions. Arsenic contents in groundwaters depend on the source of As, the geochemical evolution along the flow path, and the redox state of the system. Many different mechanisms of As release have been observed in natural systems. Focusing on sedimentary occurrences, the two main pathways are the reductive dissolution of Fe oxyhydroxides (FeOOH) that releases adsorbed As (Nickson et al., 1998, 2000; McArthur et al., 2001; Dowling et al. 2002; Kolker et al., 2003) and the oxida- tive dissolution of As-rich pyrite (Mallick and Rajagopal, 1996; Mandal et al., 1996; Chowdhury et al., 1999). In anaerobic laboratory experiments, high HCO 3 À con- centrations promote release of As from sulfide minerals (Kim, 1999; Kim et al., 2000). Both McArthur et al. (2001) and Harvey et al. (2002) report that the reducing con- ditions associated with organic matter decomposition may increase As mobility. Similarly, Dowling et al. (2002) observe that high levels of dissolved As and Fe are positively correlated with NH 3 and CH 4 , suggesting that microbial breakdown of FeOOH releases As. Taken together, any or all of these processes could reasonably influence the mobilization and transport of As in groundwater systems. The formation of the source FeOOH material under- going oxidative dissolution in the subsurface commonly occurs in oxidizing soil profiles. Here, As released by oxidative weathering can be adsorbed to the product Fe oxyhdroxides in the zone of accumulation. Some researchers have examined soils developing on parent materials rich in As and Fe from both anthropogenic and natural As sources (Strawn et al. 2002; Courtin- Nomade et al., 2003; Ne ´ el et al., 2003). These researchers indicate that successive oxidation and re-precipitation processes can also occur, with progressive loss of As in the solid phase, i.e., Fe-oxides. This loss of As is a result of the incomplete sorption of As back onto the FeOOH as it is re-precipitated as a solid phase. There has been increasing concern about elevated As levels in the groundwaters of the glaciated mid- continent region (Fig. 1 A). Arsenic levels exceeding the WHO MCL have been observed in groundwaters from bedrock and glacial aquifers in the southeastern Lower Peninsula of the state of Michigan (Fig. 1B, C). The source of As in the region is thought to be oxic weath- ering of As-rich pyrite (as high as 8.5 wt.% As) from the Marshall Fm. and Coldwater Shale, both of Mis- sissippian age (Kolker et al., 2003). Iron oxyhydroxides found in glacial deposits that contain rock fragments of the Marshall Fm. and the Coldwater Shale have As concentrations up to 0.7 wt.% (Kolker et al., 2003). The highest value reported for Michigan groundwater comes from the Marshall Fm. and is 220 mg/L (2.94 mM) (Kim, 1999, 2002; Kolker et al., 2003). Most prior studies of As in Michigan groundwaters (Kim, 1999; Kim et al., 2000, 2002; Welch et al., 2000; Kolker et al., 2003) have focused on watersheds in the eastern-most part of the 1138 K. Szramek et al. /Applied Geochemistry 19 (2004) 1137–1155 state, locally known as the ‘‘Thumb’’ region (Fig. 1B). Heterogeneity in As levels is to be expected, given the complex interplay between Pleistocene glacial history, erosion, deposition, and fluctuations in recharge rates to drift and bedrock aquifers. In this contribution, the authors explore the patterns and causes of elevated As concentrations in ground- waters from unconfined glacial drift aquifers in the Lower Peninsula of Michigan (see Fig. 1) in proximity to the Marshall Fm. To gain a fuller understanding of Fig. 1. Hydrogeological framework of the upper Midwest (US) and Lower Peninsula of Michigan. (A) Average Pleistocene drift thickness in the Midwest Region. The Lower Peninsula of the state of Michigan is marked with a box. (B) Middle to Upper Paleozoic bedrock geology of the Lower Peninsula of Michigan. The Marshall Formation sub-crop forms a circle around the state. The two main bedrock aquifers are in Devonian carbonates and sands of the Marshall Fm. (C) Shaded relief map of Michigan marked with outlines of the 4 USGS reference watersheds identified by name. Each watershed is marked with a line of section referred to in Fig. 2. K. Szramek et al. /Applied Geochemistry 19 (2004) 1137–1155 1139 the processes regulating the As contents of glacial drift groundwater systems in Michigan, the authors investi- gated the geochemical relations between As and other geochemical variables in groundwater, surface water, and soils in a well constrained portion of the Huron River watershed in southeastern Michigan. This is part of a larger study of C cycling and transformations in the Huron watershed (Szramek, 2002). As shown in the following section, the Huron River watershed is estab- lished on top of heterogeneous glacial deposits and has groundwaters which exhibit a large range of As con- centrations, many above the WHO MCL of 0.13 mM. 2. Hydrogeologic framework of arsenic occurrence in lower Michigan groundwaters The Michigan Basin is a cratonic depression filled with mainly Paleozoic era sedimentary bedrock and mantled by Pleistocene glacial deposits (Dorr and Eschman, 1970). As shown in Fig. 1B, the principal bedrock aquifers in the basin are Devonian carbonates, Mississippian and Pennsylvanian units, including the Marshall Fm., and the glacial deposits (Rheaume, 1991). A major aquitard, the Coldwater Shale, underlies the Marshall Fm. sandstones. Due to the bowl-like shape of the basin the sub-crops of the Marshall Fm. and the Coldwater Shale form a nearly concentric ring within the Michigan Basin. Bedrock is mantled by a sequence of Pleistocene glacial deposits up to 300 m thick which show the record of 2 Ma of Pleistocene ice sheet advances and retreats (e.g. Dorr and Eschman, 1970). These glacial sequences exhibit a range of hydrologic properties and include permeable sands and gravels in outwash deposits, less permeable tills, and highly impermeable lakebed clays. The glacial deposits are also the primary control on the topography of the state, and glacial depositional features commonly define watersheds (Fig. 1C). Of special interest in framing this study were areas where glacial drift aquifers overlie the Marshall Fm. In Fig. 1C, the locations of 4 watersheds (Thumb region, Huron River, Kalamazoo River, and Manistee River) established on top of the Marshall Fm. subcrop are displayed. Groundwater chemical data including As concentrations are available for each of these 4 water- sheds from the United States Geological Survey NWIS Web database (2001) and Kim (1999). Schematic hydrogeologic cross-sections (Fig. 2 A–D) show that the watersheds fall along a continuum between highly permeable open systems to those with significant per- meability contrasts within the drift aquifer materials to those with virtually no permeability. The Thumb area is primarily covered with lakebed clay deposits, reducing contact between surface flow systems and the underlying bedrock aquifer. The Thumb area is part of a regional groundwater discharge system and saline water is commonly encountered within 60 m of the surface (Rheaume, 1991). This situation is unusual for Michigan because most other areas have significant communi- cation between surface waters and groundwater flow systems as exemplified by the Huron, Manistee and Kalamazoo watersheds. The groundwater As concentrations in each of the 4 watersheds are displayed in Fig. 3. Although the Marshall Fm. is located within each of these reference areas, As concentrations vary widely. Differences in the permeability and transmissivity of the glacial drift deposits would be expected to play a large role in the variability of As levels. These factors encourage or inhibit the oxidization and reduction processes known to mobilize As from sulfide and oxide minerals. Groundwater from the Thumb region has the max- imum As value reported in the Lower Michigan area (2.94 mM As) and 70% of the 100 wells sampled have As concentrations in excess of the WHO MCL of 0.13 mM (Kolker et al., 2003). The median As concentrations in groundwaters from the 4 watersheds are, from lowest to highest: 0.010 mM in the Manistee watershed, 0.0134 mM in the Kalamazoo, 0.029 mM in the Huron, and 0.121 mM in the Thumb region. The hydrogeology of the Huron watershed study site is similar in many ways to the Manistee and Kalamazoo watersheds and offers an interesting counterpoint to prior geochemical investigations of the Thumb region. 3. Materials and methods 3.1. Study location The Portage Creek catchment is located in the wes- tern portion of the Huron watershed (Fig. 4A), where the Marshall Formation sub-crops beneath glacial drift (Fig. 1B). The groundwater in the Huron watershed is mainly hosted in glacial drift aquifers (Twenter et al., 1976). The area has high topographic gradients as it is one of the headwater catchments of the Huron River. The mean annual temperature for the region is 10  C and the average annual precipitation is 80 cm. The drainage area for Portage Creek is approximately 205 km 2 and is mainly comprised of hardwood forests, with limited urban development. Portage Creek flows through lakes and wetlands on its course toward the main stem of the Huron River. The work focused on the Hell Fen area (Fig. 4B). The fen is located along Tiplady Road near Hell, Michigan (W83  59 0 08 00 and N42  26 00 36 000 ). Fens are ground- water-fed wetlands that have high concentrations of Ca 2+ and HCO 3 À and circum-neutral pH (Glaser et al., 1990, Komor, 1994; Almendinger and Leete, 1998). Fig. 4A shows the surface drainage in the fen with small 1140 K. Szramek et al. /Applied Geochemistry 19 (2004) 1137–1155 creeks flowing along the surface. The fen is surrounded by topographic highs allowing for the discharge of shallow groundwaters. The groundwater discharges at this point because of the drift impermeability and heterogeneity in the location of the fen (Fig. 5). 3.2. Sample sites Groundwater wells are all producing from glacial drift. Well sites were chosen based on their proximity to Hell Fen (shown in Fig. 4B) and on the availability of well driller’s information. As seen in Fig. 5, the hetero- geneity of the drift allowed for waters being drawn from different drift types. Sampled wells were mainly unconfined and varied from 16 to 60 m deep. A sequence of shallow groundwaters that discharge into the fen were sampled using 5 PVC piezometers transecting the fen to cover aerial variability (Fig. 4C). Care was taken to prevent surface water contamination by packing swelling clay around the outside of the upper half of the set pipe. The piezometers sampled water at a depth of approximately 90 cm. Fig. 2. Schematic bedrock geologic and glacial drift cross-sections for the four reference watersheds in the Lower Peninsula of Michigan. In each case, the Marshall Formation aquifer is confined on either side by shale aquitards. However, glacial drift thickness, lithology and permeability differ markedly among the 4 watersheds. (A) Manistee River watershed has the thickest and most permeable drift section comprised of outwash sands and gravels, with till in moraine deposits. (B) ‘‘Thumb’’ area has the lowest drift permeability and the thinnest cover over bedrock, as lakebed clays comprise most of the Pleistocene section. (C) Kalamazoo River watershed has drift comprised of permeable sands and gravels but is a relatively thin cover such that the eastern side of the area has Paleozoic bedrock very near the surface. (D) Huron River watershed has a relatively thick drift section characterized by impermeable lenses of till spread throughout sandy outwash and moraine deposits. K. Szramek et al. /Applied Geochemistry 19 (2004) 1137–1155 1141 Surface water samples were collected in conjunction with a larger study on the C systematics of the Huron River watershed. The surface sampling locations (Fig. 4A) were selected based on their relationship to confluences with the main stream, at points before and after the stream passed through a lake system. Care was taken to collect upstream of large roads and develop- ments to limit potential contamination from local runoff. The soil water sample sites are all located in upland areas that surround the fen. The sites H-1, H-2, and H-3 are shown in Fig. 4B. Ceramic-cup tension lysimeters (Soil Moisture Corp.) were installed in these sites for the collection of waters at depths ranging from 23 to 100 cm. Soil samples were collected from two locations, H-1 and H-2. The soils were sampled every 10 cm to a depth of 1.5 m using a large-diameter auger. Samples were taken from the center of the augured material to limit contamination from adjacent soil layers. All samples were bagged in air tight Bitran bags and then frozen. A representative subsample of this material was ground to pass a 63 mm mesh prior to geochemical analyses. 3.3. Water collection Well water was generally collected from the well owner’s outdoor tap. A laminar flow of water was allowed to run into the collecting vessel until temper- ature and dissolved O 2 were stabilized, typically taking around 20 min, depending on distance to the well-head and presence and size of the holding tank for the household. Samples were only taken after temperature and dissolved O 2 levels stabililized, indicating that the water sample was representative of in situ conditions. Aliquots were immediately taken and transferred into crimp sealed glass bottles filled with no headspace to limit O 2 contamination. Shallow groundwater was collected from the piezo- meters using a peristaltic pump. Pumping was main- tained on the wells for approximately 15 min to allow for the removal of stagnant water. Stream and fen water samples were collected over 3 seasons (10/00, 5/01, and 6/01) to capture variability in the system. Base flow of the streams in the Huron watershed is during the summer months, however, fre- quent thunderstorms can interfere with capturing the stream at base flow. Soil waters were collected from 3 nests of ceramic-cup tension lysimeters (Soil Moisture Corp.). Approximately 48 h before sample collection, tension was pulled on the lysimeters to 30 cbars to draw water into the ceramic- cup. If soil water was present, it was extracted using acid washed syringes. 3.4. Field measurements and sample preservation Temperature, conductivity, dissolved O 2 (DO), and pH were determined at the field location. Temperature and DO was measured using a YSI model 58 meter and Fig. 3. Range of groundwater As concentrations in the 4 reference watersheds (Kim, 1999; USGS, 2001). Number of samples and median As concentrations are as follows: n=13; 0.01 mM (Manistee); n=24, 0.0134 (Kalamazoo); n=18, 0.029 (Huron); and n=25, 0.121 (‘‘Thumb’’). The minimum As concentration is constrained at 0.01 mM, the As detection limit for these data sets. The World Health Organization MCL of 0.13 mM is indicated. 1142 K. Szramek et al. /Applied Geochemistry 19 (2004) 1137–1155 Fig. 4. Location of the Huron Watershed field study sites showing elevation and physiography. (A) Portage Creek catchment at the northeastern edge of the Huron Watershed (shown as small inset map). Surface water sampling locations are shown in the gray circles. The indicated sample numbers correspond to those in Table 1. The Huron River was sampled after the Portage Creek confluence at the USGS gage station 4173000 (Huron R. near Dexter, MI). (B) Topography of the study area at the town of Hell, MI. An extensive wetland area studied is indicated as ‘‘Hell Fen’’ on the map. Groundwater and surface water locations are indicated by black and white circles, respectively. Soil water and soil profile sampling sites are located at H-1 and H-2. The line shown from H-1 to H-2 sites indicates the location of the driller log lithologic sections shown in Fig. 5. (C) Expanded scale view of Hell Fen with fen surface drainage sample locations indicated by the open circles. The dashed line across the fen is the location of the piezometers, locations are numbered from 1 through 5, spaced roughly equally, going from west to east. K. Szramek et al. /Applied Geochemistry 19 (2004) 1137–1155 1143 a YSI 5239 DO probe with high sensitivity membrane, directly at the source, either in the stream or at the groundwater well. Conductivity was observed in the field using a Corning 316 meter with a two point cali- bration 0 and 1413 mS, mostly to provide a rapid geo- chemical reference point. Dissolved O 2 measurements were precise to Æ 5% saturation and conductivity measurements to within Æ 5%. A Corning 315 high sensitivity pH meter with an Orion Ross combination pH electrode calibrated with low ionic strength buffers of 4.1 and 6.97 were used to measure pH in the field as close to the water temper- ature as possible. The pH of a sample can change due to degassing and warming; therefore, the samples were placed in a large volume airtight container and mea- sured at least twice to ascertain electrode stability. The precision of pH determinations is Æ0.01 pH units. Samples for later chemical analysis in the lab were collected in HDPE bottles. The bottles and filters used for the As samples underwent a 3-step acid wash proce- dure and were dried in a laminar flow hood. Aliquots for analyses were filtered through a 0.45-mm nylon filter into their respective bottles while still in the field and refrigerated until analyzed. Samples collected for total dissolved As analyses were acidified down to approxi- mately a pH of 2 with optima grade HNO 3 (Fisher Scientific). Samples intended for As speciation (As III/As V) determinations were filtered into dark glass bottles, filled with no headspace, and were not acidified. Dissolved inorganic C (DIC) and ICP–AES aliquots were preserved in the field with CuCl 2 , and HNO 3 , respectively. The DIC aliquots were placed in serum vials, filled with no headspace, and then crimp-capped using Teflon-lined septa. Aliquots for titration alkalinity and ion chromatographic analyses were placed in HDPE bottles filled with no headspace without any acid treatment. Refrigeration on site and rapid analysis back at the University laboratory was essential for As speciation and for alkalinity titrations to prevent oxidation and carbonate/hydroxide precipitation. 3.5. Geochemical analyses Arsenic was measured on a Thermo-Finnigan Ele- ment 2 mass spectrometer using a modified method of hydride generation (Klaue and Blum, 1999). Most ana- lyses were for total As concentrations. Here, all species of As in aqueous solutions are oxidized to As(V) using 10% (v/v) HNO 3 and ultraviolet oxidation in a con- tinuous-flow reaction vessel. A small suite of samples were collected for determination of As speciation. The same method of hydride generation was used (e.g. Klaue and Blum, 1999), but the aqueous sample is passed through column pretreatment to separate the As (III) from As(V) prior to the oxidation step. As(V) is then reduced with 1% (w/v) NaBH 4 in 0.1 M NaOH to form AsH 3 gas. The AsH 3 gas is then swept with Ar into the mass spectrometer after passing through a liquid/vapor separator (Klaue and Blum, 1999). A few of the samples were run without hydride generation on a Finnigan Element 1 ICP–MS. The detection limit for As run on high resolution is about 0.004 mM. Major element chemistry on waters was measured by ICP-AES for cations and ion chromatography for anions. A Leeman Labs, Inc., Plasma-Spec ICP-AES 2.5 was used to analyze for Ca, Mg, Na, Sr, and Fe with a precision of Æ 2% for major and Æ 5% for minor ele- ments. Anions (Cl À and SO 4 2À ) were analyzed on a Dionex 4000I series ion chromatograph (IC) with an AS14 column to a precision of Æ 2%. Aliqouts of soil leaches were analyzed using a Finnigan Element 1 ICP– MS at a precision of Æ 1.5 to 2%. Total alkalinity was measured within 24 h of sample collection by electrometric endpoint titration using a Radiometer TitraLab automated titration system with a TIM900 titration manager and ABU91 or ABU93 autoburette. Due to the given measurement precision (Æ 0.01 meq/kg), the pH range of the samples, and the ionic composition of the solutions, HCO 3 À was calcu- lated as equivalent to total alkalinity. Charge balance calculations performed on water chemistry data to check for internal analytical consistency were within 5%. 3.6. Solid soil collection and analysis Soils were extracted for hydroxide and carbonate bound metals using a modified strong acid leach descri- bed by Hossner (1996). In a study by Chen and Ma Fig. 5. Lithologic heterogeneity of the drift is shown in this schematic cross-section of the Hell Fen area. Driller well log records for private wells were used to construct the cross- section. The locations of the two soil profiles and lysimeter sites are shown as H-1 and H-2. 1144 K. Szramek et al. / Applied Geochemistry 19 (2004) 1137–1155 (2001), a similar method was tested on 20 different soils and shown to be an effective way to determine total As. The extraction method uses approximately 0.5–0.8 g of soil ground to finer than 63 mm which is treated with 5 ml of ‘‘aqua regia’’ (3 HCl:1 HNO 3 ) in an acid-cleaned 125- ml polypropylene bottle. The soils were reacted for 3 h on a shaker table at room temperature. After that time, the reaction was stopped by the addition 95 ml of H 2 O to form a 5% acid solution to prevent cation precipita- tion. The solutions were then filtered through a 0.45-mm polypropylene filter into acid cleaned vials. Blanks (same procedure without soil) were carried out in the same manner and subtracted from the final calculations. This cold acid extraction technique primarily dis- solves the most reactive fractions in the soil (hydroxides and carbonates) and does not significantly attack silicate or sulfide minerals. The effectiveness of the modified technique was confirmed via repeat extractions on the solid residue. No additional As was recovered in repeat digests. Additionally, several samples from the base of the soil column were analyzed by S-coulometry to determine if sulfides were present in the bulk parent material. Results of S analyses were below detection, consistent with the maturity of the weathering zones in these well developed soils. 4. Results and discussion 4.1. General water chemistry Major element chemistry of waters from the Portage catchment [soil water (lysimeter), groundwater (well and piezometer), and surface water (Portage Creek and Hell Fen)] is dominated by Ca 2+ , Mg 2+ and HCO 3 À (Fig. 6A). The stoichiometry of the dissolution reaction for carbonate minerals with CO 2 yields 2 mol HCO 3 À for each mole of divalent cations (Ca 2+ +Mg 2+ ) and most waters are close to this ideal value. The Mg 2+ /Ca 2+ ratio of the waters (see Table 1) falls very close to 0.5, suggesting that 1 mol of dolomite dissolves per 1 mol of calcite. The glacial drift contains fragments of Paleozoic carbonates (calcite and dolomite), and this is evident from the soil extract data presented later in this section. Aqueous speciation and carbonate mineral saturation state calculations indicate that groundwaters are all near equilibrium with respect to dolomite and approximately twice saturated with respect to calcite (Szramek, 2002). Given the average groundwater temperature around 10  C, dolomite is more soluble than calcite, permitting dolomite dissolution concurrently with calcite precipita- tion. As will be discussed later in this section, carbonate mineral recrystallization would be expected to occur along the groundwater flow path, and is evidenced by the significantly elevated Sr 2+ /Ca 2+ ratios in many of the groundwaters (see Table 1). Given the high topographic gradients in the Hell fen area, it is common for groundwaters to discharge into surface flow systems with attendant degassing of dis- solved CO 2 , especially in the summer when there are large temperature increases during discharge. Under these conditions, calcite supersaturation (IAP/K) can increase to values as great as 16, which produces the CaCO 3 marl of the Hell fen surface sediments (Szramek, 2002). Car- bonate precipitation has an important regulating effect on the nutrient cycling in fens (e.g. Boyer and Wheeler, 1989) because phosphate has a very strong affinity for adsorption on carbonate mineral surfaces (e.g. DeKanel and Morse, 1978, Walter and Burton, 1988). Fig. 6A shows that surface waters commonly have Ca 2+ +Mg 2+ concentrations greater than those in the groundwaters. A plot of Na + vs. Cl À (Fig. 6B) shows that groundwaters and soil waters tend to have very low Cl À concentrations, but the surface waters can be extre- mely enriched in Cl À . Approximately 20,000 t of salt are added each year by the Washtenaw County transporta- tion department (Mulcahy, 2003) and CaCl 2 is also commonly used to deice walks and driveways. Two fen water samples shown in Fig. 6B have Cl À in excess of all other samples. These two samples are taken from a location close to the road and experienced larger input of salt as a result. All the surface waters have Cl À in excess of Na + suggesting both salts contribute to the overall solute load of the surface waters. Thus, water chemistry in the Portage catchment is dominated by inputs from carbonate mineral dissolution and anthropogenic salt sources. 4.2. Arsenic in soil profiles The geochemistry of soil extracts (Al, Fe, As, Ca, Mg) for the two profiles H-1 and H-2 is reported in Table 2. The trace metal, As and Al relations vs. depth for the two soil profiles are shown Fig. 7 A–D. The zone of accumu- lation (B horizon) for the soil is evident by the increased concentrations of Al, Fe and As between 50 cm and 125 cm (Fig. 7A–C). In this zone the Ca and Mg concentra- tions indicate that the carbonates have been selectively weathered out of the soil column until at least 130 cm in H-1 and 160 cm in H-2. Below the zone of accumula- tion, the Ca and Mg contents rapidly increase towards relatively unaltered parent glacial drift values (Table 2). The ultimate source of the As in the soil is from As- rich pyrite from the Marshall Fm. that was incorporated into the drift and then oxidized and re-precipitated as FeOOH within the drift (Kolker et al., 2003). Soil pro- files H-1 and H-2 have an average As/Fe ratio within the zone of accumulation of 0.5 (mol 10 À3 ). This value is similar to values reported by Kolker et al. (2003) in the Thumb area for the bulk Marshall Fm., taken from core cuttings that range from 0.9 to 1.8 (mol 10 À3 ) and till derived from the Marshall Fm approximately 1.5–2.7 K. Szramek et al. / Applied Geochemistry 19 (2004) 1137–1155 1145 (mol 10 À3 ). The bulk Marshall Fm. hosts As in unaltered As-rich pyrite and arsenopyrite, whereas the till and soil profiles host As in Fe oxyhydroxides. The similar values indicate that the oxidation of Fe sulfides to Fe oxy- hydroxides appears to closely follow the value of the As and Fe of the precursor phase. 4.3. Relations between arsenic and iron in waters Arsenic and Fe concentrations in the water samples from the Portage catchment are reported in Table 1.In a plot of As vs. Fe (Fig. 8A), each water type appears clustered in composition space with respect to Fe and Fig. 6. General major element geochemistry of the water samples. (A) Carbonate geochemistry: All waters fall near the 1:2 mol ratio of Ca 2+ +Mg 2+ :HCO 3 À indicating that dissolution of carbonate minerals is the major process controlling the water chemistry. This pattern is typical of surface waters and shallow groundwaters in the carbonate-rich drift deposits of the upper Midwest (e.g. Rheaume, 1991; Szramek, 2002). (B) Anthropogenic salt inputs into the Portage Creek catchment include NaCl and CaCl 2 . Surface waters are most influenced by additions of these two salts, explaining why many plot above the 1:2 stoichiometric line in Fig. 6A. Most groundwaters and soil waters have low Cl À contents, with most Na + derived from plagioclase feldspar dissolution. 1146 K. Szramek et al. / Applied Geochemistry 19 (2004) 1137–1155 [...]... dissolved As was obtained on a suite of groundwater and surface water samples (Portage Creek) and indicated that As(III/V) ratio for the surface waters is $0.5 and the groundwaters is $10 This indicates that the groundwater has less As(V) than As(III), which is expected because of the low dissolved O2 in the groundwaters and the reduction of As(V) to As(III) in anoxic conditions The surface waters of Portage... piezometers from this study The Sr2+ of the groundwaters shows a positive trend with As (R2=0.8186) The Mg/Ca of the groundwaters also tends to increase with increasing Sr, however the relationship has an R2 below 0.6 The increase in Sr2+ can indicate increased contact time, or residence time of waters in carbonate bearing aquifers An increased residence time of groundwater may allow for a higher As concentration... glacial drift may play a large role in forming the precursor material that can then be mobilized under reducing condition in deeper portions of the groundwater flow system Strontium proved to be the most diagnostic indicator of As concentrations in the groundwaters The increase in Sr2+ with the accompanying increase in As concentrations indicates that recrystallization in the aquifer is an important process... overlie the intact bedrock The oxic open system weathering conditions in unsaturated zones leads to formation of Fe-oxyhydroxides, that very effectively retain As in the soil profiles Currently the sulfide minerals in the Marshall Fm are described as pristine (Kolker et al., 2003) This indicates that reducing conditions are maintained in the Marshall Fm Thus, the fluctuating conditions in the glacial drift may... al., 2003) Arsenic rich Fe-oxides are found in glacial material associated with the Marshall Sandstone and Coldwater Shale in Michigan (Kolker et al., 2003) At one time all the As in the region was incorporated in sulfide minerals The erosion of source rock and the incorporation into glacial drift sequences allows for a complex interstratification of multiple exposure horizons and weathering zones that... Hydrogeochemistry of arsenic and other inorganic constituents in groundwaters from La Pampa, Argentina Appl Geochem 17, 259–284 Stollenwerk, K.G., 2003 Geochemical processes controlling transport of arsneic in groundwater: a review of adsorption In: Welch, A.H., Stollenwerk, K.G (Eds.), Arsenic in Ground Water Kluwer, Netherlands, pp 67–100 Strawn, D., Donner, H., Zavarin, M., McHugo, S., 2002 Microscale investigation... likely due to the close connection between surface waters and groundwaters in this area and the rapid groundwater discharge rates Variations in the As/Fe ratios may be used to understand the chemical evolution paths of the different groundwater samples An increase in the As/Fe ratio can indicate a loss of Fe relative to the As or a gain of As relative to Fe In a natural system Fe is controlled by oxidation... predicted by the amount of water rock interaction that has occurred with drift carbonate minerals (e.g increase in Sr/Ca with dissolution–reprecipitation reactions) (B) As (mM) vs Sr(mM) for groundwaters in glacial drift wells from the Manistee and Kalamazoo watersheds (USGS, 2001 data set) Note the relatively low Sr and As contents, comparable with the lowest values observed in the Huron watershed This is... Fe as the water discharges into the fen The fen water is the result of shallow groundwater being discharged due to permeability differences within the local glacial deposits and the steep slopes generally Fig 8 Arsenic and iron concentration relations (A) As (mM) vs Fe (mM) for waters in the Portage catchment The groundwaters have higher As and Fe concentrations than all other water samples in the system... Chakraborti, D., 1996 Arsenic in groundwater in seven districts of West Bengal, India-the biggest As calamity in the world Curr Sci 70, 976–986 Mallick, S., Rajagopal, N.R., 1996 Groundwater development in the As affected alluvial belt of West Bengal-some questions Curr Sci 70, 956–958 McArthur, J.M., Ravenscroft, P., Safiulla, S., Thirlwall, M.F., 2001 Arsenic in groundwater: testing pollution mechanisms . Arsenic mobility in groundwater/surface water systems in carbonate-rich Pleistocene glacial drift aquifers (Michigan) Kathryn. reducing conditions are maintained in the Marshall Fm. Thus, the fluctuating conditions in the glacial drift may play a large role in forming the precursor material that