Environ. Sci. Technol. 2005, 39, 8038-8044
Removal of Arsenic from Groundwater by Zerovalent Iron and the Role of Sulfide R. KO ¨ B E R , * ,† E . W E L T E R , ‡ M. EBERT,† AND A. DAHMKE† Institute for Geosciences, Christian-Albrechts-University of Kiel, Ohlshausenstrasse 40, 24098 Kiel, Germany, and HASYLAB/DESY, Notkestrasse 85, D-22603 Hamburg, Germany
Several recent investigations have shown encouraging potential for the removal of arsenic (As) from groundwater by granular zerovalent iron (Fe0). In contrast to previous studies conducted, we have investigated the applicability of this method and the nature of As bonding under conditions with dissolved sulfide. Three column tests were performed over the period of 1 year using solutions with either As(V) or As(III) (2-200 mg/L) in the input solution. Arsenic outflow concentrations decreased from initially 30-100 µg/L to concentrations of below 1 µg/ L with time. XANES (Xray absorptions near edge structure) and EXAFS (expanded X-ray absorption fine structure) spectra indicated that As in the solid phase is not only directly coordinated with oxygen, as is the case in adsorbed or coprecipitated arsenite and arsenate. Samples with high sulfur content showed additional bonding, for which Fourier transformations of EXAFS data exhibited a peak between 2.2 and 2.4 Å. This bonding most likely originated from the direct coordination of sulfur or iron with As, which was incorporated in iron sulfides or from adsorbed thioarsenites. The formation of this sulfide bonding supports the removal of As by Fe0 because sulfide production by microbial sulfate reduction is ubiquitous in permeable reactive barriers composed of Fe0.
Introduction Groundwater contaminated by As poses a health risk to millions of persons worldwide (1). Numerous research activities have demonstrated the capability of Fe0 to remove As from groundwater within permeable reactive barriers (PRBs), on-site reactors, or household filtration devices, which can be used for the treatment of anthropogenic and geogenic contaminations (2-13). The main mechanism for the removal of As by Fe0 in experiments without microbial activity was shown to be sorption to iron (hydr)oxides, such as magnetite or green rust, which results from corrosion of metallic iron and cover the Fe0 surface (2, 5, 14). Auger mappings of Fe0 samples from a field test where sulfate was reduced to sulfide by bacteria showed that the spatial distribution of As was closely related to that of sulfur (10). This can be the result of various processes, such as the formation of arsenic sulfides or arsenopyrite, the incorporation of As in iron sulfides, the sorption of As to sulfate green rust, or the sorption of * Corresponding author phone: +49-431-8802860; fax: +49-4318807606; e-mail:
[email protected]. † Christian-Albrechts-University of Kiel. ‡ HASYLAB/DESY. 8038
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005
FIGURE 1. Setup of sequential column experiments (hAsS, sediment with high Ag content; IAsS, sediment with low As content; OM, organic matter).
thioarsenic species (As-S compounds) to any surface. Even though microbial sulfate reduction (MSR) was reported in numerous Fe0 applications (15-18) and can be regarded as an omnipresent process after a certain adaptation time in Fe0 reactors, there is no information about the effect of MSR or sulfide on As removal by Fe0. The goals of this contribution were therefore to examine the basic feasibility of Fe0 technology for As removal under conditions with moderate to high sulfide concentrations and to investigate As removal mechanisms in the presence of sulfide.
Methods Column Experiments. Three sets of column experiments (Ø, 10 cm; L, 112 cm), each composed of two or three sequential columns connected by viton tubing, were performed using artificial groundwater (Figure 1). In system 1, a column filled with As-contaminated aquifer sediment from a site in Wiesbaden (Germany) was set before a column filled with Fe0 (iron sponge, ISPAT, Hamburg, Germany) to achieve most realistic conditions for the solution that enters Fe0. A third column was added upstream of the contaminated sediment column in systems 2 and 3. In system 2, this column contained organic matter (compost, chaff, and wood chips), and in system 3 it contained a mixture of organic matter and gypsum. The upstream materials were each followed by two columns of which the first was filled with As-contaminated sediment and the second with Fe0. A section with low Ascontent sediment was assembled in the front area of the sediment-filled columns of systems 2 and 3 to investigate sulfide consumption by iron vis-a`-vis precipitation as arsenic sulfide. This aspect of study is not covered in this paper. This difference in system 1 had no observable impact on the questions discussed here. The first 10 cm of each of the columns were filled with quartz gravel to distribute the flow across the cross-sectional area of the columns. Liquid samples were taken with syringes at a three-way valve before the solution entered the columns (inflow) and at nine sampling ports, which were located at distances of 3, 8, 13, 22, 32, 47, 62, 77, and 92 cm from the end of the gravel. The test solution was pumped by peristaltic pumps from the bottom to the top for 420 days through systems 1 and 10.1021/es0504439 CCC: $30.25
2005 American Chemical Society Published on Web 09/13/2005
2, and for 342 days through system 3. Systems 2 and 3 were run without the upstream materials during the first 80 days. The average velocities in Fe0-1, Fe0-2, and Fe0-3 were 39, 41, and 36 cm/d, respectively, resulting in residence times until the ninth sampling port of 56, 54, and 61 h. The test solution was made by adding salts and acid to tap water to achieve an approximate pH value of 7.5, concentrations (in mg/L) of 40 Na, 4 K, 150 Ca, 20 Mg, 13 Si, 70 Cl-, 23 NO3-, 140 SO42-, and an alkalinity of 5 mmol/L. Adding cis-dichloroethylene (cis-DCE), which was present at the contaminated site in addition to As, within a mixing chamber to this solution resulted in input concentrations between 0.2 and 7.4 mg/L of cis-DCE. Degradation of cis-DCE showed no effects on As removal and is not further discussed in this contribution. Analysis of the Liquid Phase. As, Fe, Mn, Ca, Mg, K, Na, and Si concentrations were determined by ICP-OES (ICPOES Vista, Varian), and SO42-, NO3-, and Cl- concentrations were determined by ion chromatography with conductivity detection (DX 500, Dionex). Arsenic concentrations lower than 10 µg/L were measured by ICP-MS (Agilent 7500c). Sulfide was analyzed by photometry (PM2DL, Zeiss) using the methylene blue method, and alkalinity was measured by HCl titration. For the measurement of pH and EH, electrodes (InLab 422, Metler Toledo; PT 8280, Schott) were used. In addition to standard analyses, some samples were analyzed for the As species by coupling an ion chromatographic separation (system Gold, Beckman) and ICP-MS (PQ ExCell, Thermo). IC-ICP-MS measurements were done at the UFZ - Umweltforschungszentrum Halle-Leipzig GmbH using methods described by Daus et al. (19) and Mattusch et al. (20). All samples were filtrated by 0.2 µm regenerated cellulose membrane filters (Minisart RC 25, Sartorius) to focus the investigations on dissolved concentrations. This implies that particulate transport of contaminants was not considered in this study. Arsenic transport by colloidal iron in Fe0 reactors is the subject of further planned investigations. Samples for ICP-OES and ICP-MS (cations and Astot) were acidified with HNO3 (65% supra pure) to pH < 2. Acidification of cation samples was done immediately after sampling and filtration, whereas samples for total As measurements were acidified, at the earliest, 3 days after sampling and at least 2 days before measurement. The latter was done to avoid precipitation of arsenic sulfides in solutions containing high As and sulfide concentrations. The time before acidification was necessary for the oxidation of sulfide in contact with the atmosphere. Acidifying the samples 2 days before measurement assured redissolution of iron hydroxides, if such phases had formed. Samples for As species analysis were stored at most for 2 days in headspace-free glass vials before measurement. Preparation of Solid Samples from Column Experiments. Columns were cut open under argon atmosphere, and samples were taken at the positions of the sampling ports. For the examination of precipitates that had formed on Fe0 and to reduce the iron background, precipitates were detached from the Fe0 grains. Fe0 grains, together with dried acetone (max. 0.01% H2O), were put in 125 mL glass flasks within an argon-filled glovebox. The flasks were fastened on a linear shaker and agitated at 140 min-1 for 45 min. The flasks were then opened in a second argon box to separate the detached precipitates from the grains and to dry the samples without contact to atmospheric oxygen. The samples were stored in sealed and argon-filled glass vials until further use. X-ray Diffraction (XRD). Detached precipitate samples were taken out of the argon-filled vials, ground in an agate mortar, and mounted on a quartz slide immediately before measurement. No efforts were made to exclude atmospheric oxygen during measurements. Samples were examined with a D-5000 Siemens diffractometer at a scan rate of 0.01 or
0.02° 2θ with 2-18 s measurement per step. The instrument was operated with Cu KR radiation at 40 kV and 30 mA. Digestion. 30-200 mg of the samples and HNO3 (65%, subboiled) were heated to 120 °C in Teflon crucibles with loosely attached caps. After possible gas development had ceased, caps were closed and heat was increased to 165 °C within 1 h. This temperature was maintained for 4-5 h. This procedure did not digest the solids completely. Minor quantities of pristine Fe0 samples and samples from the experiments were not digested. These phases were identified as pyroxenes and quartz in comparable quantities by XRD and were not accounted for in the calculation of elemental concentrations. X-ray Absorption Spectroscopy (XAS). X-ray absorption fine structure (XAFS) spectra were taken in fluorescence mode at beamline A1 at the Hamburger Synchrotron Strahlungslabor (HASYLAB). The synchrotron radiation from a bending magnet was monochromatized by use of a Si(111) channel cut double crystal monochromator. Higher harmonic radiation was suppressed by detuning of the second crystal to 60% of maximum transmission. The θ angle of the monochromator crystals is measured by a highly reproducible angle encoder, so that no energy standard was necessary for the precise determination of the relative edge positions. Typical deviations of the edge position are 97%) and sulfate. Sulfate input into Fe0-2 was
