M. Rajib H. Mozumder*
Lamont-Doherty Earth Observatory of Columbia University, NY, 10964, USA
Benjamin C. Bostick
Lamont-Doherty Earth Observatory of Columbia University, NY, 10964, USA
Magdi Selim
School of Plant, Environmental, and Soil Sciences, Louisiana State University AgCenter, Baton Rouge, LA, 70803, USA
M. Atikul Islam
Department of Geology, University of Dhaka, Dhaka, 1000, Bangladesh
Elizabeth M. Shoenfelt
Lamont-Doherty Earth Observatory of Columbia University, NY, 10964, USA
Tyler Ellis
Lamont-Doherty Earth Observatory of Columbia University, NY, 10964, USA
Brian J. Mailloux
Environmental Science, Barnard College, New York, NY, 10027, USA
Imtiaz Choudhury
Department of Geology, University of Dhaka, Dhaka, 1000, Bangladesh
Kazi M. Ahmed
Department of Geology, University of Dhaka, Dhaka, 1000, Bangladesh
Alexander van Geen
Lamont-Doherty Earth Observatory of Columbia University, NY, 10964, USA
Arsenic, Groundwater, Transport, Column experiment, Kinetic model, Bangladesh
Risk Management in Agriculture
2.1. Sediment coring and column preparation Sediment cores containing Holocene gray sand and Pleistocene orange sand were collected in Araihazar, Bangladesh, immediately before the experiment. Intact cores were retrieved (30 cm long, 1.8 cm outer diameter) using a hammer-driven soil corer (AMS 424.45) from drilling depths between 12.2 and 18.3 m. Immediately after retrieval, the cores were refrigerated in nitrogen-flushed Mylar bags that were heat-sealed after adding oxygen absorbers (IMPAK sorbent systems). Within 24 h, and inside a nitrogen inflated glove chamber (Glas-Col04408-38), a total of 15 undisturbed sediment columns (8 grays and 7 orange sediment columns), 7.5 cm in length and 1.6 cm in diameter, were prepared from the central portion of the recovered cores using a precision tube cutter. The inlets and outlets of the columns were enclosed with custom-made plugs after inserting a thin layer of glass wool to prevent the loss of fine particles. A column packed with pure sand (ACROS Organics 370942500) was prepared in parallel as a control.
2.2. Experimental setup The gray and orange sediment columns along with a sand column containing 99.8% SiO2 (Quartz) and ~0.01% iron oxide (mesh size: 40e100) were eluted with groundwater directly at the wellhead from a shallow well screened from 18 to 20 m, a depth where As concentrations typically peak in the study area. The influent groundwater from the shallow well was pumped continuously at a rate of 8 L/min into a bag (50 L capacity) shaped like a pyramid that was kept overflowing through a narrow opening at the top to ensure a constant supply of anoxic groundwater (Fig. 2). The storage bag was placed at a higher elevation than the columns to ensure uninterrupted, steady flow in the event of a pump stoppage or electricity failure. The custom-made bag (Ready Containment LLC) facilitated the escape of gas bubbles that tend to cling to the corners of a regular container. Groundwater from the storage bag reached a manifold that divided the flow into the columns at different rates using peristaltic pumps (Gilson Minipuls 3) and various tubing diameters. The columns were housed in custom-made anoxic chambers (modified Becton-Dickinson#260672) with pouches that consume oxygen (Becton-Dickinson#260678) and anaerobic indicator strips (Becton-Dickinson#271051). The columns were placed inside the chamber in their natural orientation, with the groundwater entering the top of each column, and with sufficient backpressure to prevent gravity flow and degassing. Of the experiments performed, a total of 10 sediment columns (6 gray and 4 orange sand columns) were successfully completed, and 5 were compromised due to repeated flow interruption. We focus here on the 10 successful columns. Most of the columns (n = 8) were eluted either at an average pore water velocity (PWV) of 154 ± 10 cm/day or 75 ± 10 cm/day. In addition, one orange sand column was eluted at 40 cm/day and one column of gray sand was eluted at 230 cm/day.
2.3. Sampling and onsite measurements Column effluents were collected in a manually operated fraction collector in acid leached (10% HCl) 15 ml vials every 1-3 h for the first 10 days of the experiment, followed by every 12 h for 2 weeks. The sample volume was documented at the time of collection. Every other sample was acidified in the field with Trace Metal grade HCl. Some of the remaining samples remained unacidified for anion analysis while others were acidified later in the lab with Optima grade HCl for cation analysis. After 20 days, groundwater flows into the storage bag was interrupted and the retained water in the bag was spiked with sodium bromide (Fisher#S255-500) at about 120 mg/L and sodium phosphate monobasic dihydrate (Fisher #S381-500) at about 6 mg/L. The purpose of spiking with bromide and phosphate was to characterize sediment porosity and column dispersion and the effect of phosphate exchange, respectively. The influent groundwater was sampled daily at an outlet before it reached the storage bag and from the bottom of the storage bag (Fig. 2). Dissolved oxygen concentration in the influent water was tested daily with a visual kit (Chemetrics 0e40 ppb). The input water was also monitored daily for pH, ORP (oxidation-reduction potential), electrical conductivity, and temperature using Oakton probes (UX-35650-10 and UX-35634-30). A pH flow-through cell (UX-05662-48) was used to measure pH in a subset of column effluents. Arsenic speciation cartridges were used in the field to separate As(V) from As(III) in the influent water as well as in a subset of column effluents immediately after collection (Meng et al., 2001). Samples for dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) were collected in 22 ml clear glass vials. The DOC samples were acidified to 0.1% HCl while DIC samples were left unacidified.
2.4.1. Reflectance spectroscopy Sediment cuttings collected while drilling was packed in plastic Saran wrap. A Konica Minolta CM-600d spectrophotometer was used to measure the difference in diffuse spectral reflectance between 530 and 520 nm (Horneman et al., 2004) through the plastic wrap soon after the samples were collected, as a proxy for the redox state of surficial Fe oxides (Fig. S1a-b). Measurements were made in triplicate by the spectrophotometer and recorded for three different spots on each sample of cuttings.
2.4.2. X-ray fluorescence (XRF) and sediment phosphate extraction The Saran-wrapped sediment cuttings were also analyzed with a handheld X-ray fluorescence analyzer (InnovX Delta) in 3-beam soil mode for bulk As concentration. The internal calibration of the instrument was verified by analyzing reference NIST standards (SRM 2709, 2710, and 2711) in between samples. As a measure of the exchangeable As content of the sediment, 1 g of sand (n = 8) immersed in 55 ml N2-flushed 1 M Na2HPO4 solution was adjusted to a pH of 5 for 24 h (Zheng et al., 2005). The extraction was conducted in an anaerobic chamber (Coy#120001) and the solutions were filtered through 0.45 mm media.
2.4.3. X-ray absorption spectroscopy (XAS) Iron (Fe) K-edge X-ray absorption spectra were collected on pristine gray and orange sediment cuttings in fluorescence mode at the Stanford Synchrotron Radiation Lightsource (SSRL) on beamlines 4e1 and 11e2. Fe spectra were collected using either 32 or 100-element Ge detectors windowed on the Fe Ka fluorescence peak, using a 3mx Mn filter. Spectra were calibrated using a Fe metal foil (7112.0 eV). No photooxidation or reduction was observed during data collection.
Water Research 183 (2020) 116081
Journal