Environ. Sci. Technol. 2004, 38, 1866-1876
Effects of Iron Purity and Groundwater Characteristics on Rates and Products in the Degradation of Carbon Tetrachloride by Iron Metal M A R IÄ A L . T AÄ M A R A A N D ELIZABETH C. BUTLER* School of Civil Engineering and Environmental Science, University of Oklahoma, Norman, Oklahoma 73019
Carbon tetrachloride (CT) batch degradation experiments by four commercial irons at neutral pH indicated that iron metal (Fe0) purity affected both rates and products of CT transformation in anaerobic systems. Surface-area-normalized rate constants and elemental composition analysis of the untreated metals indicate that the highest-purity, leastoxidized Fe0 was the most reactive on a surface-areanormalized basis in transforming CT. There was also a trend of increasing yield of the hydrogenolysis product chloroform (CF) with increasing Fe0 purity. Impurities such as graphite in the lower purity irons could favor the alternate CT reaction pathway, dichloroelimination, which leads to completely dechlorinated products. High pH values slowed the rates of CT disappearance by Peerless Fe0 and led to a pattern of decreasing CF yields as the pH increased from 7 to 12.9. The Fe/O atomic ratio vs depth for Peerless Fe0 filings equilibrated at pH 7 and 9.3, obtained by depth profiling analysis with X-ray photoelectron spectroscopy, indicated differences in the average oxide layer composition as a function of pH, which may explain the pH dependence of rate constants and product yields. Groundwater constituents such as HS-, HCO3-, and Mn2+ had a slight effect on the rates of CT degradation by a highpurity Fe0 at pH 7, but did not strongly influence product distribution, except for the HS- amended Fe0 where less CF was produced, possibly due to the formation of carbon disulfide (CS2).
Introduction Carbon tetrachloride (CT) is a groundwater pollutant currently regulated under the Safe Drinking Water Act with a maximum contaminant level (MCL) of 5 µg/L (1). In situ transformation of CT occurs by natural degradation in the subsurface or in iron permeable reactive barriers (PRBs) (2, 3). Previous studies have addressed the influence of surface area, groundwater constituents (i.e., carbonates, sulfides, organic matter content, anthropogenic substances), and pH on the rates of chlorinated contaminant transformation by iron metal (Fe0) (4-12). Under reducing conditions, CT can be transformed by two main pathways: (1) hydrogenolysis, or replacement of * Corresponding author phone: (405) 325-3606; fax: (405) 3254217; e-mail:
[email protected]; address: 202 W. Boyd, Room 334, Norman, OK 73019. 1866
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 6, 2004
chlorine by hydrogen to produce chloroform (CF), and (2) dichloroelimination, which forms a dichlorocarbene intermediate that rapidly undergoes hydrolysis to form completely dechlorinated products such as carbon monoxide and formate (13-23). Direct transformation of CT to methane and other hydrocarbons via an initial dichloroelimination step, and without formation of CF, dichloromethane (DCM), or chloromethane, has also been proposed (24). CF, which is toxic, has been reported as the only major product (yields > 50%) of CT reduction by Fe0 in aqueous systems (1, 4, 25). Completely dechlorinated dichloroelimination products likely account for the incomplete mass balance in these systems. Different commercial iron metals (i.e., electrolytic vs cast) that vary in composition and types of impurities have been used to study the kinetics of degradation of chlorinated compounds (1, 3-12, 25-32). Cast irons produced by hightemperature reduction of ore (i.e., hematite (R-Fe2O3) and limonite (Fe2O3‚3H2O)) by coke, a coal-derived carbon, contain substantial amounts of carbon (C) and silicon (Si), and smaller amounts of transition metals and other impurities such as sulfur (S). Irons produced by electrolysis are typically more pure. Because of their lower cost, cast irons are typically used in PRBs. We hypothesized that Fe0 impurities such as those present in cast irons could influence the rates or products of CT reduction by Fe0 by acting as catalysts (e.g., Cu, Cr, and Mn) or as additional reactive species and/or adsorption sinks for chlorinated contaminants (e.g., S, Si, and C). For example, studies have reported faster rates of CT (5) and TCE (6, 30) degradation after treatment of Fe0 with dissolved sulfide, and one study proposed a relationship between the S content of Fe0 and its reactivity with respect to TCE reductive dechlorination (30). Impurities are also expected to influence the extent of Fe0 corrosion, which in turn may affect the rates of contaminant reduction. Localized corrosion can occur when two different metals are in contact. For example, at a bimetallic Mn-Fe0 spot on an Fe0 particle, corrosion of Fe0 can be hindered because Mn is more electropositive than Fe0, making Mn sites preferential anodes, and protecting Fe0 against oxidation (33). Groundwater chemical characteristics (including pH, alkalinity, and the type and abundance of ionic species) and microbial activity (e.g., the presence of sulfate-reducing bacteria) also influence the predominant minerals that can precipitate on Fe0 PRBs, which in turn can affect reaction rates and/or products (5-11). Analysis of iron samples taken from operating PRBs indicates that precipitates of siderite (FeCO3), ferrous sulfide (FeS), and iron oxyhydroxides (e.g., goethite (R- FeOOH)) coexist on iron surfaces (34). In addition, pH values as high as 10 have been measured in the vicinity of Fe0 PRBs due to the corrosion of iron (35). Batch and field studies on the transformation of chlorinated and nonchlorinated compounds by Fe0 have found that the rate constants decrease as pH increases (4, 31, 36) possibly due to the precipitation of iron oxyhydroxides that block reactive sites or decrease Fe0 corrosion rates by forming a protective layer that inhibits the reduction reaction (37). Consistent with this, one study found that the thickness of the iron oxide layer, rather than the amount of ferrous iron (Fe(II)) in the oxide, was the predominant factor affecting the rates of CT and nitrobenzene degradation at an iron-oxide-coated gold electrode (38). Our research hypothesis was that Fe0 purity, the concentrations of specific impurities such as S, C, and transition metals, the iron surface properties (i.e., presence of oxides, 10.1021/es0305508 CCC: $27.50
2004 American Chemical Society Published on Web 02/03/2004
sulfides, and carbonates), pH, and the adsorption of cations such as Mn2+ have the potential to affect the rates and products in the transformation of CT. To test our hypothesis, we studied the effects of iron type and impurities, groundwater characteristics ([HS-], [HCO3-], [Mn(II)], and pH), and Fe0 acid-washing pretreatment in the anaerobic transformation of CT by Fe0 in batch systems. Four commercial iron metals, two electrolytic and two cast irons, were chosen because of their differences in types and content of impurities arising from their manufacturing methods. CT was chosen as a model compound because of its widespread presence in contaminated aquifers, well-known transformation pathways under reducing conditions, and relatively fast transformation rates by Fe0.
Materials and Methods Iron Metals and Chemicals. Electrolytic iron powders and 40-mesh cast iron filings were obtained from Fisher Scientific (Fairlawn, NJ). Peerless cast iron aggregate (size ETI 8/50) was supplied by Peerless Metal Powders & Abrasive Company (Detroit, MI). For brevity, we use the following abbreviations: Fisher electrolytic iron powder (certified by Fisher as >99% Fe), “FE 99%”; Fisher electrolytic iron powder (certified by Fisher as >93%), “FE 93%”; Fisher 40-mesh iron filings, “FF-40 mesh”; and Peerless cast iron aggregate, “Peerless”. Average specific surface areas (SSAs) as measured by single-point BET with a Flowsorb II 2300 (Micromeritics, GA) with N2 adsorption were (in m2/g) as follows: 0.089 ( 0.027 (FE 99%); 0.195 ( 0.020 (FE 93%); 7.42 ( 0.20 (FF-40 mesh); and 1.785 ( 0.069 (Peerless). Uncertainties represent 95% confidence intervals calculated from four replicate measurements. The values of SSA for FE 99% and Peerless are within the range of previously reported values (12, 29, 32, 36, 39, 40). We are not aware of a previously reported SSA for FE 93%. The SSA for FF-40 mesh, although higher than many previously reported values (36, 40, 41), is similar to the value of 5.69 m2/g reported by Moore et al. (42), who suggest that differences between manufactured lots of iron could account for differences in reported SSA values. Differences in porosity due to variations in the cast iron manufacturing process (e.g., temperature or partial pressure of oxygen), or due to differences in the composition of the coke and/or iron ore used to manufacture the cast iron, might explain the differences in SSA between lots. To facilitate comparison between SSAs in future studies, we report the lot numbers of the irons we used, where available: Lot #010531 (FE 99%); Lot #996991 (FE 93%); and Lot #980675 (FF-40 mesh). The Peerless iron had no lot number, but it was shipped in May 2000. Analysis of the four iron metals with electron probe X-ray microanalysis-wavelength dispersive spectrometry (EPMAWDS) was conducted on a Cameca SX50 electron microprobe. EPMA-WDS was used to identify the elemental impurities present in each iron sample, which were then quantitatively assayed by bulk chemical analysis. The EPMA-WDS electron beam operated at an accelerating voltage of 20 kV and a current of 100 nA. Scanning of a 20-µm diameter spot in each sample was performed over nearly the full range of WDS motion using LiF, PET, TAP, and PC1 diffraction crystals, which together cover essentially the complete wavelength from C to U. The bulk elemental composition of each iron was assayed by Arrow Laboratory, Inc., Wichita, KS, using several methods: Mn, P, Cu, Ni, Cr, Mo, Mg, Al, V, Ti, and Fe were assayed by inductively coupled plasma/atomic absorption spectroscopy (ICP/AA), C and S were assayed by combustion IR, and Si was assayed using the perchloric gravimetric method. Samples were analyzed for both graphitic C and Austenitic C. Austenitic C is a solid solution of C in Fe formed at temperatures between 900 and 1400 °C.
If iron carbide (Fe3C) were present in an iron sample, it would have been assayed as Austenitic C. All reagents were ACS grade and all aqueous solutions were prepared with Nanopure water (18 MΩ‚cm resistivity, Barnstead Ultrapure Water System, IA). The following Good’s buffers (43) (0.05 M) were used for pH control (7 < pH < 10.5): N-[2-hydroxyethyl]piperazine-N′-[2-ethanesulfonic acid] (HEPES) for 7 < pH < 8.4, N-cyclohexylaminoethanesulfonic acid (CHES) for 8.4 < pH < 9.3, and 3-cyclohexylamino-1-propanesulfonic acid (CAPS) for pH ) 10.2. Sodium hydroxide solutions were used for pH values greater than or equal to 12. The ionic strengths of these solutions were as follows: 0.012 M (pH 7 with HEPES), 0.04 M (pH 8.4 with HEPES), 0.006 M (pH 8.4 with CHES), 0.025 M (pH 9.3 with CHES), 0.02 M (pH 10.2 with CAPS), 0.02 M (pH 12 with 0.02 M NaOH), and 0.1 M (pH 12.9 with 0.1 M NaOH). CT stock solutions (33-37 mM) were prepared in methanol sparged with 99.998% purity nitrogen. Aqueous solutions were deoxygenated with 99.998% purity nitrogen and then placed inside an anaerobic glovebox (Coy Laboratory Products, Inc., Grass Lake, MI) with a catalytic oxygen removal system (atmosphere 97% N2, 3% H2). Kinetic Experiments. Batch experiments were conducted in 8-mL vials containing 152 g/L of Fe0 and 7.8 mL of pH buffer (or NaOH solution for pH g 12). This liquid volume resulted in essentially no headspace. Vials were spiked with the CT stock solution to give an initial aqueous concentration of 0.33-0.37 mM and 1% v/v in methanol. Selected CT degradation experiments were also done in a saturated solution of CT (4 × 10-3 M) prepared in Nanopure water. Experiments with this higher CT concentration were done to facilitate detection of formate that might have been produced at relatively low concentrations in our experiments. Low levels of formate, or a compound that coeluted with formate on the ion chromatograph, were present even in samples containing only Fe0 and no CT, so it was not possible to determine whether formate was a CT reaction product without starting with a significantly higher initial concentration of CT. Background formate in Fe0 blanks may have come from the reduction of dissolved CO2 via radical reactions (44). CT degradation experiments were also done with FE 99% equilibrated overnight in 1 mM solutions of NaHS, NaHCO3, and MnBr2, and with all four Fe0 samples after acid washing. Fe0 samples were acid-washed for 20 min with 1 M HCl, and Cl- residuals were removed by rinsing 10× with deoxygenated Nanopure water. Vials containing Fe0 and no CT (Fe0 blanks) were prepared so that possible anionic products of reaction between Fe0 and water, or between Fe0 and the pH buffer, which might interfere with Cl- analysis, could be measured. Vials containing CT but no Fe0 (CT blanks) were also prepared to ensure that significant CT loss due to volatilization was not occurring. All oxygen-sensitive procedures were carried out inside the anaerobic glovebox. Spiked vials were crimp-sealed with a Teflon-coated stopper and aluminum crimp seal, placed in an incubator at 25 °C in the dark, and mixed on a rocking platform shaker until sampling. At regular time intervals, vials were centrifuged and then sacrificed to measure CT, CF, Cl-, and, for certain experiments (see above), formate. For some experiments, samples were also prepared for analysis of CO and CH4 to identify whether these compounds were reaction products. These samples were prepared in 60-mL vials with the same Fe0 mass loading (152 g Fe0/L solution) as the 8-mL vials, and zero headspace. At sampling times, an 18-mL aliquot of the aqueous supernatant from these vials was transferred to a 22-mL vial that was rapidly capped with a Teflon-coated septum, crimpsealed, and allowed to equilibrate at room temperature for VOL. 38, NO. 6, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1867
SCHEME 1. Possible Pathways for CT Degradation (based on information from 2, 13-23, 46). Letters a through j Identify Different Pathways. Compounds Measured in Our Experimental Systems Are Enclosed in Boxes
a 2-hour period to allow partitioning of CO and CH4 between the aqueous and gas phases. Then, 500 µL of the headspace was manually withdrawn with a gastight syringe and injected into the GC-TCD (see below). Duplicate 22-mL vials were sampled at each reaction time. Kinetic data for the transformation of CT were collected over the course of 2-3 halflives. Analytical Methods. CT and CF were quantified by GCECD on a Shimadzu GC-17A equipped with a J&W Scientific DB-624 capillary column (30 m × 0.53 mm × 3 µm), after extraction of 25 µL of aqueous sample into 1000 µL of isooctane. The GC method used direct injection of 1 µL of the isooctane extract. The oven temperature was isothermal at 40 °C for 2 min, ramped at 5 °C/min to 55 °C, and isothermal at 55 °C for 2 min. External calibration standards for the GC-ECD were prepared in isooctane. Cl- and formate were measured by ion chromatography using a Dionex LC20 instrument with a Dionex AS11 column and an ED50 conductivity detector. The analytical method used the following NaOH gradient elution: 0.5 mM NaOH for 2 min, increased NaOH concentration from 0.5 to 5.0 mM over 3.5 min, then increased NaOH concentration from 5 mM to 38 mM over 12 min. The eluent flow rate was 2 mL/min and the sample loop volume was 10 µL (45). IC standards were prepared in the same aqueous buffer solution used to prepare samples. The concentrations of CO and CH4 were measured on a GC-TCD with a J&W HP-molesieve column (30 m × 0.321 mm × 12.00 µm). Oven and detector temperatures were 40 °C and 110 °C, respectively, the split ratio was 30:1, and Helium was the carrier gas. Gas mixtures of 4.5% CO (v/v) in N2, and 1000 ppm and 10% of CH4 (v/v) in N2 were purchased from Scott Specialty gases (Plumsteadville, PA) and used to prepare calibration standards. These standards were prepared by injecting known volumes of the gases into the headspace of the 22-mL vials that had been crimp-sealed after addition of 18 mL of buffer solution, allowing the samples to equilibrate, and then sampling the headspace. At least five standards for each compound of interest were used to prepare a calibration curve. X-ray Photoelectron Spectroscopy (XPS) and Scanning Electron Microscopy (SEM) Analyses. Untreated Peerless 1868
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 6, 2004
Fe0 filings were equilibrated in pH 7 and pH 9.3 anoxic buffers for a 24-h period, and subsequently dried without further treatment by mild heating on a hot plate inside an anaerobic glovebox. These samples were then analyzed by SEM to characterize their morphology. We performed these analyses using Peerless Fe0 filings because their coarser size made it possible to prepare cross-sections for SEM analysis. For SEM analysis, the iron filings were embedded in a commercial resin (Embed 812-DER 736, Electron Microscopy Sciences, Fort Washington, PA), machined on a lathe to have a flat base surface, and polished with diamond lapping films (Allied, High Tech Products, Inc., Rancho Dominguez, CA) of different grit sizes ranging from 30 to 1 µm to ensure a smooth finish for adequate optical reflection. Once polished, the cross sections were covered with a conductive layer of either carbon or gold-palladium and analyzed with SEM (ETEC Autoscan) in both the secondary and backscattered modes. Images were collected with a beam potential of 15 or 20 kV. Peerless samples were also analyzed with a Physical Electronics PHI 5800 X-ray photoelectron spectrometer operating under vacuum (2 × 10-9 Torr). Fe0 samples were mounted on a sample holder by pressing them against adhesive graphite tape inside the glovebox. An airtight transfer device was then used to transfer the samples from the glovebox to the XPS vacuum chamber. Once inside the XPS vacuum chamber, surfaces were irradiated with monochromatic Al KR X-rays (1486.6 eV) of 350 W and analyzed at an electron takeoff angle of 45° with respect to the plane of the sample. Survey scans in the binding energy range from 50 to 1150 eV were used to identify all detectable elements on the samples: C, Fe, O, S, Si, Mn, and N. (EPMA-WDS detected additional elements, present at smaller concentrations, that were assayed by bulk chemical analysis.) Quantification of the atomic composition of C, Fe, O, S, Si, Mn, and N at each depth was carried out by integrating the peaks corresponding to each element with the aid of the Shirley background subtraction algorithm, and then converting these peak areas to atomic percent by using the sensitivity factors for each element in the PHI 5800 system software. An 800-µm spot size and 23 eV pass energy were used for XPS analyses. Binding energies were corrected by reference to the C1s line at 284.8 eV for hydrocarbon. Depth profiling analysis by Argon ion
TABLE 1. Product Distribution in Selected Experiments products (%)a type of Fe0 Peerless Peerless FF-40 mesh Peerless FE 99%
elapsed time (h) 2.4 20 9 26 26
pH 7 9 unbufferedd unbufferedd unbufferedd
CT0 (mM)
buffer HEPES CHES none none none
CT remaining
0.37 0.35 0.26 4.0e 4.0e
16 1 9 0 0
CF
CO
CH4
HCOO-
72 59 40 63 77
NDb
