Mobility of arsenic in a Bangladesh aquifer: Inferences from geochemical profiles, leaching data, and mineralogical characterization

Abstract

Aquifer geochemistry was characterized at a field site in the Munshiganj district of Bangladesh where the groundwater is severely contaminated by As. Vertical profiles of aqueous and solid phase parameters were measured in a sandy deep aquifer (depth >150 m) below a thick confining clay (119 to 150 m), a sandy upper aquifer (3.5 to 119 m) above this confining layer, and a surficial clay layer (<3.5 m). In the deep aquifer and near the top of the upper aquifer, aqueous As levels are low (<10 μg/L), but aqueous As approaches a maximum of 640 μg/L at a depth of 30 to 40 m and falls to 58 μg/L near the base (107 m) of the upper aquifer. In contrast, solid phase As concentrations are uniformly low, rarely exceeding 2 μg/g in the two sandy aquifers and never exceeding 10 μg/g in the clay layers. Solid phase As is also similarly distributed among a variety of reservoirs in the deep and upper aquifer, including adsorbed As, As coprecipitated in solids leachable by mild acids and reductants, and As incorporated in silicates and other more recalcitrant phases. One notable difference among depths is that sorbed As loads, considered with respect to solid phase Fe extractable with 1 N HCl, 0.2 M oxalic acid, and a 0.5 M Ti(III)-citrate-EDTA solution, appear to be at capacity at depths where aqueous As is highest; this suggests that sorption limitations may, in part, explain the aqueous As depth profile at this site. Competition for sorption sites by silicate, phosphate, and carbonate oxyanions appear to sustain elevated aqueous As levels in the upper aquifer. Furthermore, geochemical profiles are consistent with the hypothesis that past or ongoing reductive dissolution of Fe(III) oxyhydroxides acts synergistically with competitive sorption to maintain elevated dissolved As levels in the upper aquifer. Microprobe data indicate substantial spatial comapping between As and Fe in both the upper and deep aquifer sediments, and microscopic observations reveal ubiquitous Fe coatings on most solid phases, including quartz, feldspars, and aluminosilicates. Extraction results and XRD analysis of density/magnetic separates suggest that these coatings may comprise predominantly Fe(II) and mixed valence Fe solids, although the presence of Fe(III) oxyhydroxides can not be ruled out. These data suggest As release may continue to be linked to dissolution processes targeting Fe, or Fe-rich, phases in these aquifers.

Introduction

Arsenic contamination in the groundwaters of Bangladesh is widespread and poses a significant health risk to the millions who depend on this water (Chowdhury et al., 2000; Karim et al., 2000). Beginning in the 1960s, shallow (10–75 m) wells, or tubewells, were drilled throughout Bangladesh to provide drinking water. Surface water consumption had been a primary source of waterborne disease in Bangladesh, and groundwater was thought to be a safe alternative. Unfortunately, much of the water extracted from the alluvial aquifers contains naturally occurring As from the alluvium deposited on the Ganges, Brahmaputra, and Megna river floodplains during the Holocene (BGS & DPHE, 2001). The high concentrations of As in well water came to public attention in 1993, when the Bangladesh Department of Public Health Engineering tested wells in western Bangladesh after extensive As contamination was discovered in West Bengal nearly a decade before (Chadha and Ray, 1999; BGS & DPHE, 2001). Since then, high levels of As have also been found in northwestern Bangladesh (Badruzzaman et al., 1998) and in many of the wells in central and southern Bangladesh (BGS & DPHE, 2001). In a national survey of 3534 wells less than 150 m deep, As concentrations exceeded the Bangladesh standard of 50 μg/L in 27% of the wells, and As concentrations greater than 1000 μg/L were found in some (BGS & DPHE, 2001). It is unclear what proportion of the population (>120 million) will eventually suffer health effects as a result of exposure (Anawar et al., 2002), but the implication of existing dose-response data (Mazumder et al., 1998) is that the prevalence of arsenocosis will be more than one million people (Yu et al., 2003).

Observations of low oxidation/reduction potential, high levels of aqueous Fe and bicarbonate, and low levels of sulfate, have led researchers to hypothesize that microbially mediated reduction of Fe(III) oxyhydroxides has been the primary mechanism mobilizing As from these shallow sediments (Bhattacharya et al., 1997; Nickson et al 1998, Nickson et al 2000; BGS & DPHE, 2001; Harvey et al., 2002). Although release of As through oxidation of pyrite has also been advanced as a mobilization mechanism (Chowdhury et al., 1999), regional (Nickson et al., 2000; BGS & DPHE, 2001) and local geochemical surveys (Harvey et al., 2002) indicate an inverse relationship typically exists between dissolved sulfate and As in the pore waters of the Holocene aquifers, inconsistent with patterns expected from dissolution of arsenical pyrite. In addition, acid volatile sulfides have been detected in sediments cored throughout the depth range of peak aqueous As concentrations at one site within a highly impacted area south of Dhaka (Harvey et al., 2002), suggesting that formation, rather than oxidative dissolution, of sulfides is more likely in these deposits.

While country-wide surveys have mapped the extent of dissolved As contamination in tubewells, only a few studies have been devoted to the analysis and characterization of the sediments to elucidate what role solid phases play in As mobility in these aquifers (Foster et al., 2000; Nickson et al., 2000; BGS & DPHE, 2001; Breit et al., 2001). Nickson et al. (2000) extracted “diagenetically available” metals from late Pleistocene-Holocene sediments taken from two shallow cores collected from the Gopalganj district using hot concentrated HCl and found Fe and As in these extracts to be correlated. BGS & DPHE (2001) analyzed sediments taken from up to 150 m in depth in several districts of Bangladesh. In addition to mineralogical characterization by scanning electron microscopy and X-ray diffraction, they used selective dissolution with ammonium oxalate to liberate solid phase As. They found ammonium oxalate-extractable As and Fe to be highly correlated. Extraction methods used by Nickson et al. (2000) and BGS & DPHE (2001) did not differentiate sorbed versus coprecipitated As. Breit et al. (2001) and Foster et al. (2000) analyzed shallow (<50 m) sediments using X-ray absorption spectroscopy, finding evidence for As associated with aluminum hydroxides and the weathered, Fe-rich edges of phyllosilicates in gray, reduced sediments of the Holocene aquifer.

Data characterizing deeper aquifer materials are particularly sparse, and little work has been done with the orange-brown Pleistocene sediments typically underlying a confining clay layer present at up to 50 m in thickness in some areas of Bangladesh. The pore water extracted from wells deeper than 150 m typically has much lower As concentrations (van Geen et al., 2003a), with only 5% of wells sampled in the national survey exceeding 10 ug/L (BGS & DPHE, 2001). For this reason, deeper wells, some penetrating the Pleistocene sediments below the confining clay layer, are currently being installed in some villages as a presumably safe alternative to the As-contaminated shallower wells (van Geen et al., 2003b; Yu et al., 2003). These installation programs are in addition to the deep well installations that have occurred in past years in the coastal areas to remedy saline intrusion. However, it is not known whether these deeper wells will remain free of As once extraction of water from the wells commences. Deeper wells, especially those completed where a confining unit is absent, might become contaminated with time if As moves downward from the shallower aquifer, induced by deep well pumping (BGS & DPHE, 2001). Shifts in geochemical conditions resulting from pumping may also facilitate As mobilization to pore water. To our knowledge, no analyses of the deeper aquifer below this clay confining layer have been conducted to determine at what concentrations, and in what forms, As is present in these sediments. These issues need to be addressed before future water use from the deeper aquifer can be planned.

The objectives of the present study are therefore to contribute to understanding the geochemical mechanisms that control partitioning of naturally occurring As between aquifer solids and groundwater in the aquifers of Bangladesh. We provide a detailed geochemical characterization of aqueous and solid phases along a depth profile extending to 165 m below ground surface at a site in a village within the Munshiganj district south of Dhaka, Bangladesh. Our investigation also includes one of the first analyses of sediments taken from the deeper aquifer underlying the clay aquiclude. These data complement an earlier presentation of aqueous and solid phase profiles from this site by Harvey et al. (2002), who showed that aqueous Ca and NH₄ ⁺ closely follow the profile of aqueous As throughout the upper aquifer, which peaks at a depth range of 30 to 40 m, and aqueous inorganic carbon and organic carbon (DOC) also increase with increasing As up to the peak depth. They inferred from these observations that microbial respiration and associated ammonification of organic matter resulted in the reductive dissolution of Fe(III) oxyhydroxides and possibly dissolution of calcite as well, the former being a likely cause of the elevated aqueous As levels in the upper aquifer (Harvey et al., 2002). Furthermore, they demonstrated the susceptibility of upper aquifer sediments to further release of As to pore water after introduction of labile organic matter to stimulate microbial respiration (Harvey et al., 2002), suggesting that some As may yet be associated with solid phases susceptible to reductive dissolution.

In this paper, we provide additional analyses of dissolved constituents from this site as well as characterization of solid phase As. We use sequential extraction techniques to differentiate between adsorbed and coprecipitated As in the sediments and to quantify readily extractable Fe versus Fe incorporated in more recalcitrant phases such as silicate minerals and pyrite. Although the utility of extraction methods can be limited by nonselectivity (Ryan and Gschwend, 1991) and readsorption of target elements (Gruebel et al., 1988; Ostergren et al., 1999), their application can aid in characterizing relative leachability of elements from sediment matrices. We also use spectroscopic methods to elucidate spatial associations between As and Fe in sediment matrices. In addition, we perform geochemical equilibrium modeling, with input parameters provided from the aqueous and solid phase analyses, to explore whether amorphous Fe(III) oxyhydroxide, even as only a relatively small amount of the total pool of readily extractable Fe, could explain dissolved and solid phase levels of As in the upper aquifer, as suggested by the in-situ mobilization experiments conducted by Harvey et al. (2002). We use a depth where aqueous As is highest as an example for this exercise. With the intent to inform more reliable evaluation of alternative water supply options for Bangladesh, these analytical methods are used in an effort to understand the mobility of As in sediments in the upper and lower aquifers and, in particular, what types of geochemical alterations, possibly induced by pumping, may further mobilize any As present in these aquifers.