Environ. Sci. Technol. 2006, 40, 2018-2024
Early Breakthrough of Molybdenum and Uranium in a Permeable Reactive Barrier S T A N J . M O R R I S O N , * ,† PAUL S. MUSHOVIC,‡ AND PRESTON L. NIESEN§ S.M. Stoller Corporation, 2597 B 3/4 Road, Grand Junction, Colorado 81503, U.S. Environmental Protection Agency Region 8, Federal Facilities Program, Office of Environmental Protection and Remediation, 999 18th Street, Suite 300, Denver, Colorado 80202, and Cotter Corporation, 7800 East Dorado Place, Englewood, Colorado 80111
A permeable reactive barrier (PRB) using zerovalent iron (ZVI) was installed at a site near Can˜on City, CO, to treat molybdenum (Mo) and uranium (U) in groundwater. The PRB initially decreased Mo concentrations from about 4.8 to less than 0.1 mg/L; however, Mo concentrations in the ZVI increased to 2.0 mg/L after about 250 days and continued to increase until concentrations in the ZVI were about 4 times higher than in the influent groundwater. Concentrations of U were reduced from 1.0 to less than 0.02 mg/L during the same period. Investigations of solid-phase samples indicate that (1) calcium carbonate, iron oxide, and sulfide minerals had precipitated in pores of the ZVI; (2) U and Mo were concentrated in the upgradient 5.1 cm of the ZVI; and (3) calcium was present throughout the ZVI accounting for up to 20.5% of the initial porosity. Results of a column test indicated that the ZVI from the PRB was still reactive for removing Mo and that removal rates were dependent on residence time and pH. The chemical evolution of the PRB is explained in four stages that present a progression from porous media flow through preferential flow and, finally, complete bypass of the ZVI.
Introduction Molybdenum (Mo) and uranium (U) contaminate streams and groundwater associated with Mo- and U- ore and tailings deposits (1-4). The U.S. Environmental Protection Agency (EPA) promulgated a groundwater standard of 0.1 mg/L for Mo at inactive uranium processing sites because of potential toxicity to humans and cattle (5). A standard of 0.044 mg/L for U was considered protective of cancer risks in humans (5). Under near-earth-surface conditions, Mo is typically in the +6 oxidation state and, at pH values more than 5, forms a water-soluble molybdate (MoO42-) complex (2). Kaback and Runnells (1) found that Mo concentrations in contaminated streams near the Climax Colorado molybdenum mill were controlled by adsorption to ferric oxyhydroxides rather than by mineral solubility. Under oxidized conditions, U(VI) is typically mobile as a carbonate complex, and under reducing conditions it forms stable U(IV) minerals (6). * Corresponding author phone: (970)248-6373; fax: (970)248-6040; e-mail:
[email protected]. † S.M. Stoller Corporation. ‡ U.S. Environmental Protection Agency Region 8. § Cotter Corporation. 2018
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 6, 2006
Permeable reactive barriers (PRBs) are zones of reactive materials placed in the subsurface to passively treat groundwater (7, 8). Contaminants are removed from groundwater as it passes through the PRB. PRBs containing zerovalent iron (ZVI) as the reactive material are effective in treating groundwater contaminated by chromium (8), nitrate (9), organic solvents (7, 10, 11), arsenic (12, 13), and U (3, 9, 14, 15). PRB remediation technology using ZVI is relatively new; the first commercial field installation was completed in 1994 (11). Because the passive nature of PRBs offers cost advantages over other groundwater remediation technologies, more than 80 full-scale ZVI-based PRBs have been installed worldwide (16). Few studies have investigated mechanisms responsible for the removal of Mo from solution by the ZVI. Using batch tests with the ZVI, Cantrell et al. (14) determined that removal of Mo(VI) from solution was slower than for U(VI), chromium, or technetium. On the basis of modeled redox values, they suggest that aqueous Mo(VI) was reduced to solid-phase Mo(III) by the ZVI. Chemical equilibrium modeling of the ZVI treatment of uranium mill tailing seepage indicated that aqueous Mo concentrations could be controlled by mineralization of ferrous molybdate (FeMoO4) (17). A variety of mechanisms have been proposed for the removal of U(VI) by the ZVI. Cantrell et al. (14) suggested a mechanism of reductive precipitation of U(IV) minerals based on thermodynamic mineral stability. Reductive precipitation is supported by fluorescence spectroscopy analysis of product solid phases from batch tests (18) and from a column test that produced an insufficient mass of ferric oxyhydroxide to account for the uptake of U by adsorption (19). In contrast, Fiedor et al. (20) used X-ray photoelectron spectroscopy (XPS) to determine that most of the U present in samples collected from a ZVI PRB was adsorbed as U(VI) on hydrous ferric oxide. Noubactep et al. (15) studied the effects of the availability and the reactivity of corrosion products on the process of U removal by the ZVI (in long-term batch experiments) and concluded that initial deposition of U(VI) was by coprecipitation with iron oxide corrosion products. By submerging ZVI plates in U-bearing solutions and analyzing their surfaces with XPS and ion mass spectroscopy, Scott et al. (21) concluded that U(VI) precipitation was caused by a combination of reduction by structural Fe(II) and incorporation of U(VI) into ferric oxides. Many researchers have noted precipitation of minerals in PRBs caused by corrosion of the ZVI including goethite, magnetite, ferrihydrite, green rust, FeS, aragonite, calcite, and siderite (3, 9, 22-26). Some have noted that precipitation of these minerals has led to loss of porosity and in some cases decreases in hydraulic conductivity (23, 25, 26). Lai et al. (27) observed preferential pathways in a PRB that were likely caused by mineral precipitants. Birke et al. (28) report that at least two of three full-scale PRBs in Germany have hydraulic problems, such as groundwater bypassing the ZVI because of mineral precipitation. In some PRBs, precipitation of corrosion products has apparently caused negligible effects on groundwater flow (10, 26). Models that couple mineral precipitation to hydraulic properties suggest that typical PRBs will show only subtle hydraulic changes over 10 years, and significant changes will not occur for 30 years (29). In June 2000, construction was completed on a PRB at a uranium-ore milling site operated by Cotter Corporation (Cotter) near Can ˜ on City, CO. The PRB was used to remove Mo and U from groundwater that flows into an aquifer beneath a residential area. The Cotter PRB uses the ZVI as the reactive medium and is the first PRB that was constructed 10.1021/es052128s CCC: $33.50
2006 American Chemical Society Published on Web 02/14/2006
treat Mo contamination, after 1200 days of operation Cotter initiated a pumping operation to remove contaminated groundwater from the upgradient sand.
Experimental Section
FIGURE 1. Schematic of Can˜on City PRB showing monitor well locations. to mitigate Mo. Based on results of laboratory tests prior to construction, the PRB was expected to last for at least 10 years. The Cotter PRB performed well by removing Mo and U from the groundwater for about 250 days. After that time, effluent Mo concentrations began to increase, and by about 1000 days concentrations of Mo in the ZVI were higher than influent concentrations. This study was undertaken primarily to understand the causes for the failure of the PRB to treat Mo for longer time periods and particularly why concentrations were higher in the ZVI than in upgradient groundwater. During the course of the investigation it became apparent that mineral precipitation in the PRB caused hydraulic changes that led to eventual groundwater bypass. Field and laboratory data were used to develop a conceptual model that explains the evolution of groundwater chemistry and mineral plugging of the PRB.
Field Site Description Geochemistry/Hydrology of the Alluvial Aquifer. The source of Mo and U in the groundwater is an active uranium-ore processing mill located hydraulically upgradient of the PRB. Ore-processing activities also caused increases in concentrations of sulfate and carbonate in the groundwater. Mean concentrations of Mo and U in the alluvial groundwater system hydraulically upgradient of the PRB are 4.8 mg/L and 1.0 mg/L, respectively. Contaminated groundwater flows through alluvium composed of unconsolidated sand, gravel, and silt. The alluvium overlies bedrock composed of sandstone, coal, and claystone of the Vermejo Formation. Saturated thickness of the alluvium varies from less than 0.2 m to more than 1.2 m and depth to groundwater is about 5.5 m in the vicinity of the PRB. On the basis of pumping tests, groundwater flux in the alluvial aquifer is estimated at 3.8 to 11.4 L/min. Permeable Reactive Barrier. The PRB consists of a 1.52-m thick (parallel to groundwater flow) zone of granular ZVI (-8 +50 mesh, Peerless Metal Powders and Abrasives, Detroit, MI; initial porosity approximately 50%) with a 0.61-m thick zone of silica sand (-8 +12 mesh) on each side (Figure 1). Peerless ZVI was selected because it had been shown to be effective in treating U in other PRBs (3, 4, 22). The ZVI zone is 9.1 m long (perpendicular to groundwater flow) and 2.1 m high. Vertical concrete walls support both ends of the PRB. The ZVI zone contains approximately 81 t of the ZVI and is keyed into bedrock composed of claystone. Funnel walls constructed of 36-mil Hypalon extend 86.9 m west and 25.9 m east of the PRB. The PRB and funnel walls capture much of the contaminated groundwater in the alluvium. Three monitor wells were placed in each of the two sand units and in the ZVI (Figure 1). Because the PRB failed to
Field Methods. Filtered groundwater samples were collected from the nine monitoring wells with a peristaltic pump. Values of pH were measured in the field using an electrode in a flow-through cell without filtering. In October 2004, after about 1600 days of operation, samples of solids were collected during excavation of 0.3-m lifts starting at the top of the ZVI. Samples used for bulk chemical analysis were placed in plastic bags. Samples used for electron microprobe (EM) and scanning electron microscopy (SEM) examinations were collected within 1 h of exposing the lift, placed in glass jars, flooded with nitrogen gas, tightly sealed, and placed under a nitrogen atmosphere until preparation for analysis. Laboratory Analysis. Groundwater samples were analyzed for calcium (Ca) by inductively coupled plasma emission, for Mo and U by inductively coupled plasma mass spectrometry, and for sulfate (SO42-) by ion chromatography. Detection levels for Ca, Mo, and U are 5, 0.005, and 0.0001 mg/L, respectively. Groundwater concentrations of Mo and U were measured starting with the installation of the PRB. Measurements of Ca and SO42- concentrations were initiated after about 500 days of operation to help evaluate causes of Mo breakthrough. To determine chemical concentrations in bulk solids, samples were digested in a microwave oven at 180 °C in concentrated nitric acid. Digestates were analyzed for Ca, Mo, and U by the same methods as those used for groundwater analysis. Solid samples used for EM examination were vacuumdried, mounted in epoxy, and cut by a diamond saw without water. The cut face was epoxied again, recut, polished with alumina compounds, and coated with a carbon film for presentation to the EM beam. Care was taken to minimize exposure to the atmosphere. Samples were examined with a JEOL Model 7333 EM operated at an electron beam energy of 15 keV and a current of 25-30 nA. Quantitative analyses were performed with wavelength-dispersive spectrometers using a beam diameter of about 1 µm. Column Test. The test used a friable ZVI sample collected 0.7 m from the base of the ZVI and 0.8 m from the upgradient face. With minimal exposure to the atmosphere, the moist ZVI sample weighing 2082 g (1914 g on dry weight basis) was placed in a 5.1-cm-diameter, 45.7-cm-high clear acrylic column. Synthetic groundwater (SGW) was pumped at various rates through the column from bottom to top. SGW had major ion chemistry similar to that in upgradient wells, pH near 7.26, and less than 0.2 mg/L dissolved oxygen concentration to approximate groundwater chemistry in the mid-portion of the ZVI. Dissolved oxygen and pH were controlled by injecting argon and CO2, respectively, into the influent solution. The influent pH value was increased at times to about 10.5 to evaluate the effect on Mo partitioning. Effluent samples were collected with a fraction collector and were preserved with 1% concentrated nitric acid prior to analysis. Molybdenum concentrations were analyzed colorimetrically by the ternary complex method (30). Sulfate and chloride were measured by ion chromatography, iron (Fe) and Ca by flame atomic absorption, and alkalinity by titration with sulfuric acid. Oxidation-reduction potential and pH data were recorded every 0.5 min with in-line electrodes.
Results Groundwater Chemistry. Groundwater-chemical concentration trends are presented for the west transect of monitor wells (Figure 1; wells 1451, 1453, and 1455); these wells are VOL. 40, NO. 6, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2019
FIGURE 3. Solid-phase concentrations (after 1500 days of operations) of (a) Mo and (b) Ca in the ZVI relative to the upgradient face and base of the ZVI zone. Porosity loss is the percent of original porosity that is occupied if all Ca resides in calcite; it is based on the molar volume of calcite and the mass of Ca measured in the core samples.
FIGURE 2. Changes over time of (a) Mo concentration, (b) pH, and (c) groundwater elevations for upgradient sand, downgradient sand, and ZVI portions of the western transect. MSL ) mean sea level. in the same area that was excavated for solid-phase sampling. The chemical composition and elevation trends in the central and east transects were similar to those in the west transect. The oxidation-reduction potential varied considerably but was typically 100-200 mV lower in the ZVI than in the upgradient sand. For the first 250 days of operation, concentrations of Mo were reduced to less than 0.1 mg/L in the ZVI, and concentrations in the downgradient sand were similar to those in the ZVI (Figure 2a). Between days 250 and 800, Mo concentrations in the ZVI and the downgradient sand averaged about 2 mg/L. After 800 days, Mo concentrations in the downgradient sand deviated significantly from those 2020
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 6, 2006
in the ZVI. After 1200 days, Mo concentrations in the ZVI increased rapidly to about 22 mg/L, greatly exceeding upgradient and downgradient values. Uranium concentrations in the ZVI remained less than 0.020 mg/L throughout the operation of the PRB, but U concentrations in the downgradient sand began to increase after 1000 days and were about 1.0 mg/L by 1500 days. Values of pH in the ZVI and downgradient sand varied significantly during operation of the PRB. From 107 to 268 days of operation, pH values in the ZVI and the downgradient sand exceeded 9.0 (Figure 2b). After 268 days, pH values in the ZVI decreased to nearly the same as those in the upgradient sand (about 7.5). After 900 days, pH values in the ZVI increased rapidly to more than 10.5. Calcium concentrations in the ZVI decreased from about 120 mg/L at 518 days to less than 10 mg/L after 1000 days. Despite low values in the ZVI after 1000 days, Ca concentrations in the downgradient sand were increasing. Sulfate concentrations of about 1200 mg/L were similar in the ZVI, upgradient, and downgradient zones through 800 days, after which concentrations in the ZVI decreased to as low as 400 mg/L. Groundwater Elevations. Groundwater elevations were relatively flat across the ZVI and the two sand zones for about
FIGURE 4. Electron images: (a) SEM photomicrograph of the ZVI collected 0.3 m from base and 1.5 m from upgradient face of PRB. Bladed calcium carbonate crystals (CC) precipitated on iron oxide (FeO) coatings. (b) EM backscatter electron image of the ZVI sample collected 0.3 m from the base and 2.5 cm from the upgradient face. ZVI grains with pseudomorphic replacement by iron oxide (a), altered rims containing calcium carbonate (CC), and late-stage iron sulfide (d, e, f, i). Letters refer to chemical analyses. Dark areas are pores. Scale bars are 100 µm. 365 days of operation, after which groundwater began to mound in the upgradient sand (Figure 2c). Groundwater elevations in the ZVI and downgradient sand zone were nearly the same through 890 days of operation, after which groundwater elevations increased in the ZVI. The initiation of pumping after 1250 days decreased groundwater elevations in the upgradient sand zone. Groundwater remained below the top of the ZVI at all times (Figure 2c). Solid-Phase Chemistry. An excavation provided direct observation of the PRB. The ZVI was indurated and would not break apart by striking it with a backhoe bucket; thus, a hydraulic hammer was used. However, some areas of friable material were present in the midportion of the ZVI. The sand zones were friable and sometimes flowed into the excavation. Nearly all hand samples effervesced and emitted a sulfurous odor when contacted with 10% HCl. Widespread translucent crystals, believed to be calcite, were observed using a binocular microscope in the field. The ZVI was pitch black except for a zone containing reddish brown iron oxides approximately 2 cm thick in about the upper 1 m of the upgradient face. Solid-phase concentrations of Mo in the ZVI ranged from 130 to 4050 µg/g (Figure 3a). Some Mo (background) was present in the ZVI before emplacing it in the PRB. Molybdenum is added to cast iron for hardness control in concentrations ranging from about 250 to 450 µg/g (31). Cast iron used in a PRB at Durango, CO, had a background Mo concentration of 326 µg/g (4). The Mo background concentration in the Cotter ZVI was not measured but is estimated at 130-300 µg/g based on these typical values, and values in portions of the PRB that were likely not affected by contaminant Mo. The highest Mo concentrations occurred within about 2 cm of the upgradient face and 1.5 m or less from the base (Figure 3a). Concentrations of Mo were above
background in samples collected 0.3-0.6 m above the base at all distances from the upgradient face. Calcium concentrations were also highest within several centimeters of the upgradient face and within 0.6 m of the ZVI base (Figure 3b). Unlike the Mo distribution, Ca concentrations were also high in samples collected as far as 1.52 m from the upgradient face. The Ca occurs as calcium carbonate minerals that are intergrown with or grown over iron oxide corrosion products (Figure 4). Assuming Ca is contained in the mineral calcite, precipitation of Ca accounts for as much as 20.5% of the available pore space (Figure 3b). Oxidized Fe deposits on ZVI surfaces (Figure 4a), and sulfur-bearing material coats calcium-bearing phases and lines pores (point f, Figure 4b), indicating that the last mineral to precipitate contained sulfur. Molybdenum and U are typically associated with Fe and S in ZVI alteration rims; calcium carbonate crystals are devoid of Mo and U. Column Test. A column test was conducted to simulate chemistry in the ZVI under controlled conditions. Specifically, we sought to determine (1) the reactivity of residual Cotter ZVI for removal of dissolved Ca, Mo, and U and (2) the effect of long residence time and high pH (up to 10.5) on partitioning of Mo to the solution phase. A total of 193 pore volumes of SGW (influent Ca, Fe, and Mo concentrations were 220,
