Significance of Iron(II,III) Hydroxycarbonate Green Rust in Arsenic

Aug 24, 2004 - hydroxycarbonate green rust (or simply, carbonate green rust) and magnetite were the major iron corrosion products identified with X-ra...
0 downloads 0 Views 181KB Size
Environ. Sci. Technol. 2004, 38, 5224-5231

Significance of Iron(II,III) Hydroxycarbonate Green Rust in Arsenic Remediation Using Zerovalent Iron in Laboratory Column Tests CHUNMING SU* AND ROBERT W. PULS Ground Water and Ecosystems Restoration Division, National Risk Management Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, 919 Kerr Research Drive, Ada, Oklahoma 74820

We examined the corrosion products of zerovalent iron used in three column tests for removing arsenic from water under dynamic flow conditions. Each column test lasted 3-4 months using columns consisting of a 10.3-cm depth of 50:50 (w:w, Peerless iron:sand) in the middle and a 10.3cm depth of a sediment from Elizabeth City, NC, in both upper and lower portions of the 31-cm-long glass column (2.5 cm in diameter). The feeding solutions were 1 mg of As(V) L-1 + 1 mg of As(III) L-1 in 7 mM NaCl + 0.86 mM CaSO4 with or without added phosphate (0.5 or 1 mg of P L-1) and silicate (10 or 20 mg of Si L-1) at pH 6.5. Iron(II,III) hydroxycarbonate green rust (or simply, carbonate green rust) and magnetite were the major iron corrosion products identified with X-ray diffraction for the separated fractions (5 and 1 min sedimentation and residual). The presence of carbonate green rust was confirmed by scanning electron microscopy (hexagonal morphology) and FTIR-photoacoustic spectroscopy (interlayer carbonate stretching mode at 1352-1365 cm-1). X-ray photoelectron spectroscopy investigation revealed the presence of predominantly As(V) at the surface of corroded iron particles despite the fact that the feeding solution in contact with Peerless iron contained more As(III) than As(V) as a result of a preferential uptake of As(V) over As(III) by the Elizabeth City sediment. Extraction of separated corrosion products with 1.0 M HCl showed that from 86 to 96% of the total extractable As (6.9-14.6 g kg-1) was in the form of As(V) in agreement with the XPS results. Combined microscopic and macroscopic wet chemistry results suggest that sorbed As(III) was partially oxidized by the carbonate green rust at the early stage of iron corrosion. The column experiments suggest that either carbonate green rust is kinetically favored or is thermodynamically more stable than sulfate green rust in the studied Peerless iron corrosion systems.

Introduction Arsenic (As) was ranked first on the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) Priority List of Hazardous Substances in the United States in 1999 and 2001. The priority list was prepared by the * Corresponding author phone: (580)436-8638; fax: (580)436-8703; e-mail: [email protected]. 5224 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004

Agency for Toxic Substances and Disease Registry of the Centers for Disease Control and Prevention, U.S. Department of Health and Human Services. Arsenic contamination of groundwater as a source of drinking water has been a health risk for many parts of the world. Consequently, there is a urgent need to develop cost-effective methods for As removal from groundwater and wastewater. Recent research efforts have focused on using zerovalent iron (Fe0) to remove As from acid rock drainage (1, 2), NaNO3 background solution in laboratory column test and retrieved groundwater from a landfill in a field column test (3), groundwater of Bangladesh (4), water with added background electrolytes (NaCl and/or CaSO4) with or without added phosphate or silicate (5-12), landfill leachate in large-scale field pilot experiments (13), and tap water (14). These studies have demonstrated that Fe0 is an effective sorbent for As and Fe0 may be used as a medium in passive chemical permeable reactive barrier (PRB) technologies to immobilize As from groundwater via surface adsorption/complexation or coprecipitation. Limitations of using Fe0 to remediate As in water include competition from other oxyanions such as phosphate and silicate for sorption sites at corroded iron surfaces and possible later release of sequestered As due to subsequent mineralogical transformation of initially formed corrosion products. The presence of phosphate and/or silicate has been shown to decrease the effectiveness of Fe0 for removing As in both batch (8) and column tests (12). Removal of As from synthetic groundwaters high in dissolved phosphate (3 mg of P L-1) and silicate (30 mg of Si L-1) required greater doses of Fe(II) or Fe(III) via coprecipitation than without P and Si (15). Because in the presence of silicate the adverse effect of phosphate on As(V) adsorption by iron hydroxides had been shown to be magnified (16), we conducted column tests on the combined effect of phosphate and silicate tests on As removal using Fe0 (12). These column tests provided macroscale evidence for the continued corrosion of Peerless Fe0 as the source of sorbents for As and confirmed previous batch tests (8) showing the adverse effects of phosphate and silicate on As removal by Fe0; however, the mineralogical properties of iron corrosion products and redox status of sorbed As at the surface of Fe0 had not been examined. Here we provide microscopic evidence for the formation of new solid phases and their importance in As removal and transformation during iron corrosion in laboratory columns. These findings should have important implications for the long-term effectiveness and life span of iron-based PRBs for groundwater As remediation.

Materials and Methods Column Tests. Detailed column tests were described in an earlier publication (12). Briefly, the column consisted of a 10.3-cm depth of 50:50 (w:w, Peerless iron:sand) in the middle and a 10.3-cm depth of a sediment from Elizabeth City, NC, in both upper and lower portions of the 31-cm-long glass column (2.5 cm in diameter) with three side sampling ports. The flow velocity (upflow mode) was maintained at 4.3 m d-1 during the 3-4-month experiments. Three treatments were employed: As without added P and Si, As with low P and Si, and As with high P and Si. A total of six columns were used for duplicate run of each treatment (1A, 1B, 2A, 2B, 3A, and 3B). For all the treatments, added As was from a stock solution of 1 mg of As(V) L-1 + 1 mg of As(III) L-1 using Na2HAsO4‚7H2O (Sigma) and NaAsO2 (Baker) in simulated Elizabeth City groundwater (ECGW) with a chemical composition of 7 mM NaCl + 0.86 mM CaSO4 (pH 6.5, adjusted with 1.0 M NaOH). Low P and Si treatment had concentrations 10.1021/es0495462 Not subject to U.S. copyright. Publ. 2004 Am. Chem.Soc. Published on Web 08/24/2004

of 0.5 mg of P L-1 + 10 mg of Si L-1 and high P and Si treatment of 1 mg of P L-1 + 20 mg of Si L-1 using NaH2PO4‚H2O (Sigma) and Na2SiO3‚9H2O (Sigma). All chemicals used were ACS reagent or analytical grade. Separation of Fines. Preliminary examination of whole zerovelent iron particles using a variety of analytical instruments such as scanning electron microscope and X-ray diffractometer did not yield detailed enough mineralogical information on the iron corrosion products due to dilution by the bulk uncorroded iron metal. Consequently, we separated the fines from zerovalent iron to facilitate better resolution of instrumental analysis. The fines from asreceived Peerless Fe0 were obtained by ultrasonication of a slurry of 50 g of iron filings in 250 mL of water for 10 min followed by collection on a 0.45-µm filter membrane and air-drying of the fines in the laboratory atmosphere. After completion of the column tests, the glass columns were transferred into an anaerobic glovebox (3-6% H2 in N2), and all material manipulations prior to redox-sensitive analytical work were performed in the glovebox to avoid re-oxidation. Filings mixed with sand were removed from the column center section. The bottom 2-3 cm of the iron-filings zone of each column was highly cemented and was difficult to loosen from the column. The filings/sand mixture recovered from the column was crushed and mixed. About three-fourths of the mixture was then transferred to a 250-mL mixing glass cylinder, diluted to ca. 230 mL with N2-purged deionized water, shaken thoroughly, and allowed to settle for 5 min. The material suspended after 5 min was transferred to a filter apparatus and filtered through a 0.45-µm membrane filter. The fines were then transferred to a vacuum desiccator and dried over Drierite. This step was designed especially to eliminate quartz sand grains from the iron corrosion products to be used for microscopic examination. The separation step was repeated to collect the 1-min sedimentation fraction, the remaining fines were also collected as residual fraction. Subsamples of separated fractions were examined with scanning electron microscopy (SEM) equipped with energydispersive X-ray (EDX) spectroscopy and FTIR-photoacoustic spectroscopy (PAS). For the remaining filings/sand mixture, a stirring bar retriever inside a plastic tube was used to transfer the strongly magnetic fraction onto paper towels for preliminary drying. After the filings were patted dry on the towels, they were transferred to a vacuum desiccator containing Drierite and completely dried overnight. The dried filings were mixed by stirring in an evaporating dish and stored under anaerobic conditions before being used for the X-ray photoelectron spectroscopy (XPS) analysis. Wet Chemical Analysis and As Speciation. Samples of 0.025 g of the above separated fractions were extracted in 15 mL of 1.0 M HCl for 24 h in the anaerobic glovebox. The extractants were filtered through 0.22-µm membranes and then analyzed with inductively coupled plasma-optical emission spectrometry (ICP-OES) for As, Ca, Fe, Mn, Si, S, and P. Arsenic speciation analysis for the HCl extractants were accomplished with ion chromatography-hydride generation-atomic fluorescence spectrometry (IC-HG-AFS) (12). As(V) and As(III) were first separated by IC, then reacted to form arsine gas through a hydride generation apparatus and finally detected by an atomic fluorescence spectrometer. The quantification limit of the system was 5 µg of As L-1 for each As species. SEM Analysis. Particle morphology and elemental composition of separated fractions were determined using a JEOL JSM-6360 SEM equipped with an Oxford Instruments EDX system. The weight and atomic percentage of the elements present on the iron particles were semiquantitatively determined using EDX. X-ray Diffraction. Pristine Peerless Fe0 particles were mounted under ambient conditions into a cavity on a zero-

background quartz slide to make a flat surface for XRD examination. For separated fines from pristine Peerless Fe0, a flat quartz slide without cavity was used. The separated fractions from column tests were mounted on the quartz slide inside the anaerobic glovebox. Approximately 20 mg of material was mixed with a drop of glycerol to form a smooth paste to prevent oxidation by air. The slide was taken out of the glovebox and examined with a Rigaku Miniflex X-ray diffractometer at a scan rate of 0.5° 2θ min-1 (Fe KR radiation; operated at 30 keV and 15 mA). FTIR-PAS Analysis. The fines recovered from the sedimentation and filtration process were used for FTIR-PAS examination using the small PAS cups (6 mm i.d.). Nonmagnetic forceps were required for handling the loaded sample cups because the samples were all strongly magnetic, both before and after air oxidation. After being loaded in the photoacoustic accessory, samples were purged with cryogenically dried He gas for 30-60 min prior to sealing and running. This process helps remove water vapor from the photoacoustic cell and prevents the strong rotational-vibrational fine structure of water vapor from obscuring the spectrum of the solid. After that, samples were run using the MTEC photoacoustic accessory and the Digilab FTS-45 FTIR spectrometer. All spectra were obtained by coaddition of 64 scans at a resolution of 4 cm-1 over the entire 4000-400 cm-1 mid-IR range. The single-beam spectra were ratioed to a single-beam spectrum of a carbon standard to convert to “photoacoustic units”, which is an absorbancelike unit. After obtaining the spectrum of the separated material from column 1A, the sample was allowed to oxidize in the sample cup; subsequently, spectra were taken at days 1 and 2 after oxidation. There was little difference between spectra at day 1 and those at day 2. All other FTIR samples were transferred to vials at the conclusion of the experiments and were allowed to oxidize completely. Spectra were also obtained for all the oxidized materials. XPS Analysis. XPS measurements were made for corroded iron filings from column tests on a Physical Electronics 5500 X-ray photoelectron spectrometer, with a monochromatized Al KR X-ray source, operated at a power of 400 W with a beam size of 1.2 mm × 1 mm. The analytical depth was 5.0-7.5 nm using a 800-µm X-ray spot. The ion current was approximately 7 A at an angle of incidence of 45°. An analyzer pass energy of 187 eV was used for survey scans (0-1400 eV), and 12 eV was used for high-resolution scans of individual core levels. The analyzer was calibrated by measuring and correcting for the binding energies peak for the Au 4f 7/2 peak at 83.98 eV, the Cu 2p 3/2 peak at 932.68 eV, and the Ag 3d 5/2 peak at 368.0 eV. Surface charging was controlled using a PHI 04-090 specimen neutralizer. The neutralizer energy was adjusted to obtain a C 1s (hydrocarbon) signal at 284.6 eV with the narrowest peak width. Samples were transferred under an inert nitrogen atmosphere using a glovebox and a sealed container, and individual corroded iron filings were selected if they had a relatively flat surface to analyze and they represented the general characteristic of the other filings in that particular batch. High-resolution spectra were collected for the elements that were present in the survey scans. For the most part, these included the C 1s, O 1s, As 3d, As 3p3/2, Fe 2p, Si 2p, Al 2p, N 1s, and Ca 2p peaks when present. Sample surfaces were briefly sputtered with Ar ion beam for 0.1 min to remove any surface species, especially the surface carbon. The samples were then re-analyzed as above. Finally the surface was sputtered again for 1.0 min to determine if the composition had changed. The ion gun was operated at 4 kV, and the beam was rastered in a 2 × 2 mm area. Data were processed using PHI PC ACESS Software V.7.1A and PHI sensitivity factors. Gaussian line shapes were VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5225

Fe(II)4Fe(III)2(OH)12SO4 + CO32- a Fe(II)4Fe(III)2(OH)12CO3 + SO42- (1) An estimated equilibrium constant was obtained in a recent study (2):

K1 ) [SO42-/CO32-][Fe(II)4Fe(III)2(OH)12CO3/ Fe(II)4Fe(III)2(OH)12SO4] ) 103.1

FIGURE 1. X-ray diffraction patterns of the Peerless iron as received and its separated fines ( 3A. Each of the spectra contains a fairly broad band near 1640 cm-1 assigned to the scissors vibration of molecular water. This band is characteristic of many hydrated minerals. Additionally, each spectrum contains a large band at 1352 or 1365 cm-1 that is assigned to CGR (26, 27). The normal modes for carbonate anions in crystalline carbonates such as calcite (1448 cm-1) and siderite (1443 cm-1) are not particularly close to either of these two bands (28). Adsorbed carbonate onto amorphous iron oxide shows two bands at about 1480 and 1350 cm-1; whereas, free carbonate in solution shows a ν3 band at 1383 cm-1 (29). The two weak bands at 1451 and 1396 cm-1 for the pristine Peerless Fe0 fines may be attributed to adsorbed carbonate (Figure 5). The magnitude of the peak associated with CGR decreases in the order of the column treatment, 1A > 2A > 3A; hence, the treatment with increasing quantities of phosphate and silicate seems to interfere with the formation of the green rust. It is possible that part of the added phosphate formed X-ray amorphous ferric iron phosphates [Fe2(HPO4)3‚xH2O or FePO4‚xH2O] and silicate was sorbed to iron oxides, which probably diluted the concentrations of CGR in the corrosion products. Crystalline ferrous iron phosphate, vivianite [Fe3(PO4)2‚8H2O] was reported to form when sulfate GR was reacted with 20 mM Na2HPO4 (30). Vivianite was not detected in this study possibly because of lower phosphate concentration in our system. It is also noticeable that the IR band position for the interlayer carbonate shifted to a higher value (1365 cm-1) as the band height decreases for the high P and

FIGURE 2. X-ray diffraction patterns of the separated corrosion products of Peerless iron reacted with As in column tests that were collected after 5 min of sedimentation. Column influent contained 1 mg of As(V) L-1 + 1 mg of As(III) L-1 in simulated Elizabeth City groundwater (ECGW) with a composition of 7 mM NaCl + 0.86 mM CaSO4 (pH 6.5). Column 1A influent had no added P and Si, column 2A influent also contained 0.5 mg of P L-1 + 10 mg of Si L-1 in ECGW (pH 6.5), and column 3A influent also contained 1 mg of P L-1 + 20 mg of Si L-1 in ECGW (pH 6.5). Si treatment. Sorption of phosphate and silicate onto CGR may perturb the symmetry of carbonate ions that leads to band shift. The samples of fines from columns were allowed to oxidize following their initial study by FTIR, and then additional spectra were obtained of the oxidized material. The striking differences between spectra of as-is and oxidized fines is the disappearance of the bands at 1352-1365 cm-1 after oxidation, which is attributed to CGR. The carbonate green rusts have more abundant natural isostructural analogues in the minerals hydrotalcite [Mg6Al2(CO3)(OH)16‚4H2O] and pyroaurite [Mg6Fe(III)2(CO3)(OH)16‚ 4H2O]. A spectrum obtained in this study by FTIR-PAS for a hydrotalcite from Snarum, Norway, yielded a broad carbonate band at 1377 cm-1. A split ν3 carbonate mode at 1365 and 1400 cm-1 in synthetic hydrotalcite has been reported recently (31). Lattice mode and molecular anion region is the most complex region to assign in the spectra of the mixed corrosion products for bands at 1035, 1008, 912, 779, and 475 cm-1. Unfortunately, standards are not available to make definitive band assignments. Further studies are needed to identify these bands by comparing samples prepared with and without each of the added anions (arsenate, arsenite, sulfate, phosphate, and silicate). XPS Results. All the filings analyzed had “as-is” initial surfaces that contained C, O, Fe, Al, and Si, with trace levels (