Environ. Sci. Technol. 2001, 35, 1266-1275
Geochemical Reactions Resulting from In Situ Oxidation of PCE-DNAPL by KMnO4 in a Sandy Aquifer MATTHEW D. NELSON,† B E T H L . P A R K E R , * ,† T O M A . A L , ‡ JOHN A. CHERRY,† AND DIANA LOOMER‡ Department of Earth Sciences, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1, and Department of Geology, University of New Brunswick, Fredericton, New Brunswick, Canada E3B 5A3
Although the potential for KMnO4 to destroy chlorinated ethenes in situ was first recognized more than a decade ago, the geochemical processes that accompany the oxidation have not previously been examined. In this study, aqueous KMnO4 solutions (10-30 g/L) were injected into an unconfined sand aquifer contaminated by the dense non-aqueousphase liquid (DNAPL) tetrachloroethylene (PCE). The effects of the injections were monitored using depth-specific, multilevel groundwater samplers, and continuous cores. Two distinct geochemical zones evolved within several days after injection. In one zone where DNAPL is present, reactions between KMnO4 and dissolved PCE resulted in the release of abundant chloride and hydrogen ions to the water. Calcite and dolomite dissolved, buffering the pH in the range of 5.8-6.5, releasing Ca, Mg, and CO2 to the pore water. In this zone, the aqueous Ca/Cl concentration ratio is close to 5:12, consistent with the following reaction for the oxidation of PCE in a carbonate-rich aquifer: 3C2Cl4 + 5CaCO3(s) + 4KMnO4 + 2H+ f 11CO2 + 4MnO2(s) + H2O + 12Cl- + 5Ca2+ + 4K+. In addition to Mg from dolomite dissolution, increases in the concentration of Mg as well as Na may result from exchange with K at cationexchange sites. In the second zone, where lesser amounts of PCE were present, KMnO4 persisted in the aquifer for more than 14 months, and the porewater pH increased gradually to between 9 and 10 as a result of reaction between KMnO4 and H2O. A small increase in SO4 concentrations in the zones invaded by KMnO4 suggests that KMnO4 injections caused oxidation of sulfide minerals. There are important benefits of carbonate mineral buffering during DNAPL remediation by in situ oxidation. In a carbonate-buffered system, Mn(VII) is reduced to Mn(IV) and is immobilized in the groundwater by precipitating as insoluble manganese oxide. Energy-dispersive X-ray spectroscopy analyses of the manganese oxide coatings on aquifer mineral grains have detected the impurities Al, Ca, Cl, Cu, Pb, P, K, Si, S, Ti, U, and Zn indicating that, similar to natural systems, precipitation of manganese oxide is accompanied by coprecipitation of other elements. In addition, the consumption of excess KMnO4 by reaction with reduced minerals such as magnetite will be minimized because the rates of these reactions increase with decreasing pH. Aquifer cores collected after the KMnO4 injections exhibit dark brown to black bands of manganese oxide reaction products in sand layers where DNAPL was originally present. 1266
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001
Mineralogical investigations indicate that the manganese oxide coatings are uniformly distributed over the mineral grains. Observations of the coatings using transmission electron microscopy indicate that they are on the order of 1 µm thick, and consequently, the decrease in porosity through the formation of the coatings is negligible.
Introduction Persistent plumes of dissolved chlorinated solvent contamination in sandy aquifers are common due to subsurface accumulations of dense non-aqueous-phase liquids (DNAPLs). For permanent aquifer restoration, the DNAPL must be removed or destroyed in situ. One approach for in situ destruction of chlorinated ethenes is chemical oxidation using potassium permanganate (KMnO4), which was first advocated by Schnarr and Farquhar (1). Schnarr et al. (2) showed the effectiveness of KMnO4 for destruction of tetrachloroethylene (PCE) and trichloroethylene (TCE) in batch and column experiments using DNAPL mixed in sand and in field experiments in the sandy aquifer at Canadian Forces Base Borden, Ontario. Other investigators have also reported encouraging results based on field trials (3-5). The reactions presented by Schnarr et al. (2) for the oxidation of PCE and TCE, respectively, by KMnO4:
C2Cl4 + 2MnO4- f 2CO2 + 2MnO2(s) + Cl2 + 2Cl- (1) C2Cl3H + 2MnO4- f 2CO2 + 2MnO2(s) + 3Cl- + H+ (2) indicate that oxidation is accompanied by production of manganese oxide, CO2(g), HCl, and Cl2(g). Although the capability for KMnO4 to oxidize chlorinated ethenes has been reported, the geochemical processes associated with the oxidation process have not previously been examined. In this study, aqueous solutions of KMnO4 (10-30 g/L) were injected into a PCE-DNAPL contaminated sand aquifer. The geochemical processes resulting from the addition of KMnO4 to the aquifer were investigated through the collection and analysis of groundwater samples from a detailed array of multi-level bundle samplers and cores of aquifer material. For comparison, KMnO4 was injected into zones containing PCE-DNAPL and also into zones with aqueous PCE but no DNAPL. The goal of this study was to identify the geochemical processes that control the chemistry of the groundwater and the abundance and distributions of secondary minerals in the treated aquifer.
Site Description The experiments were conducted at Canadian Forces Base Borden, Ontario, Canada. The geology of the area consists of an unconfined, carbonate mineral-containing sand aquifer that is approximately 4 m thick and overlies a clay aquitard. The sand was deposited in a prograding beach environment on the shores of the glacial lake “Main Lake Algonquin” approximately 10 500 yr ago (6). The sand in the upper aquifer is fine to medium grained with horizontal bedding (7), and cores collected in the area show distinct horizontal and subhorizontal layering ranging from several millimeters to * Corresponding author telephone: (519)888-4567, ext 5371; fax: (519)883-0220; e-mail:
[email protected]. † University of Waterloo. ‡ University of New Brunswick. 10.1021/es001207v CCC: $20.00
2001 American Chemical Society Published on Web 02/16/2001
FIGURE 1. Sheet pile cell in which 771 L of PCE-DNAPL was spilled in 1991: (a) locations of continuous cores, KMnO4 injection points, and multi-level sampling devices; (b) cross section of aquifer stratigraphy and schematic representation of PCE-DNAPL distribution in the cell prior to the experiment; (c) area within the cell where the KMnO4 injection experiment was conducted. several centimeters thick. Mackay et al. (8), describe the texture of individual beds and laminae as varying from silt to medium sand with occasional pebbles. Falling head permeameter tests performed on samples from the upper Borden Aquifer indicated spatial variations in permeability on the scale of centimeters (9-11). The contact between the sand and the clay is thought to be erosional (12), and cores from the base of the aquifer display a series of complex silt and sand beds overlying a sandy to pebbly silt with some cobbles. The portion of the aquifer in which the KMnO4 injections were performed is isolated by a 9 × 9 m sheet pile cell that is keyed into the upper 60 cm of the aquitard. In 1991, 770 L of PCE was released into the cell for geophysical experiments (13). Following the release, an estimated 177 L was removed by pumping PCE-DNAPL (12), and approximately 248 L was removed from the shallow part of the cell by vacuum extraction and air sparging (14). Morrison (12) estimates that 82-158 L of DNAPL has infiltrated into the aquitard and that 188-264 L of DNAPL remains in the cell. The KMnO4 injections were conducted in a small experimental volume of the cell, approximately 1.5 m wide by 3 m long within the lower 2 m of the aquifer. The area that was monitored following the injections is shown in Figure 1c.
Experimental Methods Injection Experiments. Two permanganate injection episodes were conducted inside the 9 × 9 m cell, one in
November 1997 and the second in July 1998. The injections were conducted using the Drive-Point Delivery System described by Nelson et al. (15). For each episode there were two injection locations (Figure 1), with two injection depths at each location for the first episode, and three injection depths for the second episode. During the first episode, 7 kg of KMnO4 was injected in 427 L of solution. The average injection rate was 3.8 L/min at an average pressure of 470 kPa. The injection pressures varied between 380 and 515 kPa, with injection rates between 3.5 and 4.5 L/min. The injection took approximately 8 h including equipment setup and disassembly. In the first hole, INJ-1, 99 L was released at 2.70 m depth, and 103 L was released at 3.30 m depth. In the second hole, INJ-2, 84 L was released at 2.70 m depth, and 98 L was released at 3.30 m depth. The remaining 43 L of fluid was injected while pushing the injection tip to desired depths. During the second injection episode, 9 kg of KMnO4 was delivered in 429 L at an average injection rate of 3.7 L/min, with pressures between 430 and 590 kPa. This injection episode took approximately 7 h. As in the first episode, two injection holes were used. In the first hole, INJ-3, 104 L was released at 2.63 m depth, 90 L was released at 3.15 m depth, and 21 L was released at 3.35 m depth. In the second hole, INJ-4, 105 L was released at 2.50 m depth, 82 L was released at 3.15 m depth, and 39 L was released at 3.30 m depth. Groundwater Chemistry. After the initial 8-h period of hydraulic disturbance created by the injections, the groundVOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1267
TABLE 1. Comparison of Borden Background Groundwater Chemistry with Groundwater Chemistry in PCE and Non-PCE Containing Zones following Permanganate Treatmenta analyte
backgroundb
PCE-DNAPL containing zonec
non-PCE-DNAPL containing zoned
pH Ca Mg Na K Cl MnO4SO4 NO3 Fe Mn Ni Cu Zn H2S
6.8-8.0 50-110 2.5-6 1-2 0.1-1.2 1-2.8 10-30
