Hydrogeochemical evolution of arsenic in groundwater: Sources and sinks in the Mississippi River Valley alluvial aquifer, southeastern Arkansas, United States

Twenty one of 118 irrigation water wells completed in the shallow (25-30 m thick) Mississippi River Valley alluvial aquifer in the Bayou Bartholomew watershed, southeastern Arkansas had arsenic (As) concentrations (<0.5 to 77 μg/L) exceeding 10 μg/L. Two nested monitoring wells (10 m and 36 m deep) were installed in the vicinity of the highest, median, and lowest concentrations of As at three sites in Jefferson County, Arkansas. Sediment and groundwater samples were collected to characterize the mobilization, transport, and distribution of As in aquifers. A traditional five-step sequential extraction was performed to differentiate the exchangeable, carbonate, amorphous Fe and Mn oxide, organic, and hot HNO3-leachable fraction of As and other compounds in sediments. The Chao extraction removes amorphous Fe and Mn oxides by reductive dissolution and is a measure of reducible Fe and Mn in sediments. The hot HNO3 extraction removes mostly crystalline metal oxides. Significant total As (20%) is complexed with amorphous Fe and Mn oxides in sediments. Arsenic abundance is not significant in carbonates or organic matter. Significant (40-70 μg/Kg) exchangeable As is only present at shallow depth (0-1 m). Arsenic is positively correlated to Fe extracted by Chao reagent (r=0.83) and HNO3 (r=0.85). Increasing depth has a positive relationship (r=0.56) to Fe (II)/Fe (the ratio of Fe concentration in the extracts of Chao reagent and hot HNO3), but it has a negative relationship (r=-0.45) to As extracted by Chao reagent. Fe (II)/Fe is positively correlated (r=0.76) to As extracted from Chao reagent. Although Fe (II)/Fe increases with depth, the relative amount of reducible Fe decreases noticeably with depth. The amount of reducible hydrous Fe oxides (HFO), as well as its complexed As decreases with depth. Possible explanations for the decrease in reducible Fe and its complexed As with depth include historic flushing of As and Fe from HFO and aging of HFO to crystalline phases. As+5, the dominant As-species in groundwater, has positive relations (r=0.84) to decreasing redox (RmV).;Inverse geochemical modeling (PHREEQC) was used to identify the evolution of groundwater with emphasis on As in the research area. The modeling was based on flow paths defined by high-precision (±2 cm) water level contour map; x-ray diffraction (XRD), scanning electron microscopic (SEM), and chemical analysis of boring-sediments for minerals; and detailed chemical analysis of groundwater along the flow paths. Potential phases were constrained using general trends in chemical analyses data of groundwater and sediments, and saturation indices data (MINTEQA2) of minerals in groundwater. Modeling results show that calcite, gypsum, halite, fluorite, Fe oxyhydroxide, organic matter, CO2 (gas), H2S (gas) are dissolving, whereas sphalerite, FeS, siderite, barite, and vivianite are mostly precipitating. Significant amount of sediment sulfide was detected in the aquifer sediments below the water table.;The redox environment, chemical data of sediments and groundwater, and the results of geochemical modeling indicate that reductive dissolution of Fe oxyhydroxide is the dominant process of As release in the groundwater. Gypsum solubility and SO42- reduction with co-precipitation of As and sulfide is an important limiting process controlling the concentration of dissolved As in groundwater. Spatial and temporal variability of As is controlled by spatial distribution and redox status of different redox zones at various depths in the aquifer.;Application of surface complexation models (SCM) to predict the sorption behavior of As and HFO in the laboratory has increased in the last decade. However, the application of SCM to predict the sorption of As or other elements in natural sediments has not been often reported, and such applications are greatly constrained by the lack of site specific model parameters. Attempts have been made to use SCM considering a component-additivity approach which accounts for relative abundances of pure phases in natural sediments, followed by the addition of SCM parameters individually for each phase. Although, few reliable and internally consistent sorption databases related to HFO exist, the use of SCM using laboratory-derived sorption databases to predict the mobility of As or other elements in natural sediments has increased. This study evaluated the ability of the SCM using the geochemical code PHREEQC to predict solid phase As in the sediments of the research area. The SCM option of double layer model (DLM) was simulated using ferrihydrite and goethite as sorbents quantified from chemical extractions; calculated surface site concentrations, published surface properties, and normalized published laboratory-derived sorption constants for the sorbents. The model over predicts extracted As in deeper (21-36.6 m) reduced coarse-sediments 4 to 24-fold. The model predicts 57-92% of extracted As in shallow (0-17 m) relatively oxidized fine-sediments. The model is very sensitive to the speciation of As, and the presence of competitive ions in groundwater.