Environ. Sci. Technol. 2002, 36, 2060-2066
Use of Liquid Chromatography/ Tandem Mass Spectrometry To Detect Distinctive Indicators of In Situ RDX Transformation in Contaminated Groundwater H A R R Y R . B E L L E R * ,† A N D KEVIN TIEMEIER‡ Lawrence Livermore National Laboratory, P.O. Box 808, L-542, Livermore, California 94551, and U.S. Army Operations Support Command, SOSMA-ISE, 1 Rock Island Arsenal, Rock Island, Illinois 61299
An important element of monitored natural attenuation is the detection in groundwater of distinctive products of pollutant degradation or transformation. In this study, three distinctive products of the explosive RDX (hexahydro-1,3,5trinitro-1,3,5-triazine) were detected in contaminated groundwater from the Iowa Army Ammunition Plant; the products were MNX (hexahydro-1-nitroso-3,5-dinitro-1,3,5triazine), DNX (hexahydro-1,3-dinitroso-5-nitro-1,3,5triazine), and TNX (hexahydro-1,3,5-trinitroso-1,3,5-triazine). These compounds are powerful indicators of RDX transformation for several reasons: (a) they have unique chemical features that reveal their origin as RDX daughter products, (b) they have no known commercial, industrial, or natural sources, and (c) they are well documented as anaerobic RDX metabolites in laboratory studies. The products were analyzed by LC/MS/MS (liquid chromatography/ mass spectrometry/mass spectrometry) with selected reaction monitoring and internal standard quantification using [ring-U-15N]RDX. Validation tests showed the novel LC/ MS/MS method to be of favorable sensitivity (detection limits ca. 0.1 µg/L), accuracy, and precision. The products, which were detected in all groundwater samples with RDX concentrations of > ca. 1 µg/L (25 out of 55 samples analyzed), were present at concentrations ranging from near the detection limit to 430 µg/L. MNX was the typically the most abundant of the three nitroso-substituted products; concentrations of the products seldom exceeded 4 mol % of the RDX concentration, although they ranged as high as 26 mol % (TNX). Geographic and temporal distributions of RDX, MNX, DNX, and TNX were assessed. A degradation product resulting from RDX ring cleavage, methylenedinitramine, was not detected by LC/MS/MS in any sample (detection limit ca. 0.6-4 µg/L). This extensive field characterization of MNX, DNX, and TNX distributions in groundwater by a highly selective analytical method (LC/ MS/MS) is significant because very little is known about the occurrence of intrinsic RDX transformation in contaminated aquifers. * Corresponding author phone: (925)422-0081; fax: (925)423-7998; e-mail:
[email protected]. † Lawrence Livermore National Laboratory. ‡ U.S. Army Operations Support Command, 1 Rock Island Arsenal. 2060
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 9, 2002
Introduction RDX (hexahydro-1,3,5-trinitro-1,3,5-triazine; Figure 1) is a cyclic nitramine explosive that has contaminated groundwater, soil, and surface water at many military installations, prompting concerns about potential toxic effects (1-3). Our understanding of the occurrence of intrinsic RDX degradation is very limited, in part because published field studies on this topic are exceedingly rare. Furthermore, most laboratory studies of bacterial RDX metabolism have been carried out under conditions that are not representative of aquifer environments (for example, in the presence of yeast extract or nutrient broth). Nonetheless, laboratory studies have established that anaerobic RDX metabolism occurs more readily than aerobic metabolism (e.g., refs 1, 4-7), and that MNX, DNX, and TNX (Figure 1) are transient RDX intermediates under anaerobic conditions (4-6, 8-10). Other anaerobic RDX metabolites that have been observed include methylenedinitramine or MEDINA (Figure 1), bis(hydroxymethyl)nitramine [(HOCH2)2NNO2], hydrazine, 1,1-dimethyland 1,2-dimethylhydrazine, formaldehyde, methanol, and nitrous oxide (1, 4, 10). However, known metabolites other than MNX, DNX, TNX, and MEDINA are not well suited to be distinctive indicators of in situ RDX transformation either because they are labile in aqueous solution or because they could derive from compounds other than RDX. MNX, DNX, and TNX may be indicative of anaerobic (rather than aerobic) bacterial RDX metabolism, as studies with aerobic isolates have not reported these compounds as metabolites (11, 12). However, traces of MNX were reported during aerobic mineralization of RDX by the fungus Phanerochaete chrysosporium (13). Analytical methods to detect MNX, DNX, TNX, and MEDINA in groundwater require high sensitivity and confidence in compound identification, as the mere detection of these products, even at low concentrations, can have significant regulatory implications. The conventional U.S. EPA method for RDX and other explosives, Method 8330 (14), is not well-suited to analysis of RDX products, as it relies on high-performance liquid chromatography with UV spectrophotometric detection (LC/UV) and is thus susceptible to false positive results (or high detection limits) resulting from coeluting interferences. LC/UV was used to analyze nitroso-substituted RDX products in groundwater at the Louisiana Army Ammunition Plant (LAAP) and Joliet Army Ammunition Plant (JAAP) (15). MNX, DNX, and TNX were never detected in JAAP groundwater, and MNX and TNX were detected rarely in LAAP groundwater. Notably, all MNX and TNX detections in the 15 LAAP wells occurred only on the last sampling date, even though wells had been sampled frequently (typically, 5 times in 11 months), suggesting potential difficulties with the LC/UV technique. Cassada et al. (16) used an electrospray LC/MS (liquid chromatography/ mass spectrometry) method to determine RDX, MNX, and TNX in groundwater at the Cornhusker AAP (Nebraska); the method involved solid-phase extraction, full-scan data acquisition, and internal standard quantification. In that study, which reported results for groundwater collected from two off-site wells near the Cornhusker AAP, MNX was observed in both wells at ∼0.05-10 µg/L, and TNX was found in one well at ca. 1 µg/L (25 out of 55 samples analyzed) and were present at concentrations between 0.03 and 430 µg/L. Average detected concentrations of MNX, DNX, and TNX were 65, 24, and 39 µg/L, respectively. As a result of the sensitivity and specificity afforded by SRM, signal/noise ratios were generally high for all analytes (e.g., Figure 3). In addition, retention times of the analytes in groundwater samples agreed well with those of authentic standards (Figure 3). To illustrate, the average retention time for MNX in samples was 3.49 ( 0.01 min (mean ( SD) vs 3.49 min for the standard, the average retention time for DNX in samples was 3.19 ( 0.01 min vs 3.19 min for the standard, and the average retention time for TNX in samples was 3.02 ( 0.01 min vs 2.97 min for the standard. MEDINA was not detected in any sample; although peaks with the appropriate retention time were detected in some samples, these peaks were revealed not to be MEDINA when the samples were reanalyzed using an LC column with a
FIGURE 3. Mass chromatograms for LC/MS/MS (selected reaction monitoring) analysis of [M + 75]- adducts of MNX (m/z 281 f 46 transition), DNX (m/z 265 f 46 transition), and TNX (m/z 249 f 113 transition) in a standard and in a groundwater sample from the IAAAP. different (cyanopropyl) stationary phase. The lack of MEDINA detection is noteworthy in light of the relatively high concentrations of MNX, DNX, and TNX in some samples. For example, in five samples, the lowest concentrations of the three nitroso products ranged from 8 to 190 µg/L (G-18, G-20, G-58, 800-MW-25, JAW-72). Even if the MEDINA detection limit is corrected for the low recovery observed during spiking tests with groundwater, it would still be in the range of 4 µg/L. Thus, either the ring cleavage pathway leading to MEDINA formation is not as important as the reductive pathways leading to nitroso product formation under in situ conditions, or MEDINA is much less stable than MNX, DNX, and TNX under site conditions. Recent studies of anaerobic RDX degradation suggest different views of MEDINA’s stability; it can occur as a transient metabolite (10) or a persistent transformation product that appears at substantial concentrations relative to RDX (25). The geographic distributions of MNX and RDX concentrations in the Line 800/Pink Water Lagoon area are presented in Figure 4B (Spring, 2001 sampling). Several aspects of MNX distributions are clear from Figure 4B: (a) concentrations of MNX generally correspond well with those of its parent compound, RDX, (b) the highest concentrations are in localized areas on the perimeter of the Pink Water Lagoon, and (c) the highest concentrations are in the Upper Till, and, on the western side of the lagoon, Lower Till, hydrostratigraphic units. Distributions of MNX, DNX, and TNX relative to RDX are presented in Figure 5 for all samples with RDX concentrations exceeding 10 µg/L. In Figure 5, the Pink Water Lagoon wells are arranged geographically, starting at the south end of the lagoon (G-20) and moving counterclockwise around the lagoon perimeter; the three wells from Lines 2 and 3 are on the far right portion of the x-axis. MNX was the most abundant of the three nitroso-substituted products in 12/14 samples represented in Figure 5; MNX concentrations typically ranged from 1.5 to 4 mol % of RDX concentrations. However, TNX
was the dominant nitroso-substituted product in wells 800MW-25 and G-17 (Figure 5), both of which are located on the northeastern perimeter of the lagoon and screened in the Upper Till unit (Figure 4). In these two wells, the relative concentrations of MNX, DNX, and TNX were all greater than in any of the other wells represented in Figure 5 and ranged from 3.4 to 26 mol % of RDX. Clearly, relative distributions of RDX products were not a strong function of absolute concentration, as the samples from these two wells had similar relative compositions despite having RDX concentrations that differed by a factor of >70 (Figure 5). In addition, relative distributions of RDX products in these two samples differed from distributions typically observed in batch laboratory studies of anaerobic RDX biotransformation, namely, MNX > DNX > TNX (4, 5, 8). Temporal trends in MNX, DNX, TNX, and RDX concentrations were examined for eight monitoring wells that were sampled in Fall, 2000 and Spring, 2001; RDX concentrations in these wells ranged from 370 to 12 500 µg/L during these two sampling periods. The trends are presented in Figure 6 as [(Spring, 2001)/(Fall, 2000)] concentration ratios. In general, no dramatic shifts in concentration were apparent for any of these samples; most Spring, 2001 and Fall, 2000 concentrations agreed within ca. 30%. Overall, the data suggest a slight decrease in concentration over time, as the median ratios for MNX, DNX, TNX, and RDX ranged from 0.73 (TNX) to 0.97 (MNX). One trend of interest is an RDX ratio of 1 (G-20, G-19, and JAW-72; Figure 6). Such a trend would be consistent with product formation at the expense of RDX. However, such interpretations are confounded by several factors, including whether there is an ongoing RDX source and the kinetics of the formation and consumption of MNX, DNX, and TNX. In some samples (800-MW-6, JAW-70, JAW-54), RDX and its products all behaved very similarly over time. Implications of RDX Product Distributions and in Situ Geochemistry. In general, the power of detecting distinctive VOL. 36, NO. 9, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2063
FIGURE 4. Monitoring well locations and potentiometric surface contours (A) and RDX and MNX concentrations [represented as RDX (µg/L)/MNX (µg/L)] (B) in the Line 800/Pink Water Lagoon area of the IAAAP (Spring, 2000 sampling). The information in (A) is based on maps produced by Harza Engineering (26). A single “ 10 µg/L (Spring, 2000 sampling). Pink Water Lagoon wells (G-20-G-57) are arranged geographically (see text). Note that the relative concentration of TNX in well G-17 is off scale. RDX concentrations (µg/L) are noted in italics above the bars for each station.
FIGURE 6. Time trends for RDX, MNX, DNX, and TNX concentrations for eight IAAAP wells sampled in Fall, 2000 and Spring, 2001. Data are expressed as [(Spring, 2001)/(Fall, 2000)] concentration ratios. including the transient formation of MNX, DNX, and TNX, does not necessarily lead to mineralization but rather to the production of other nonvolatile metabolites, possibly as a result of cometabolic reduction reactions and subsequent abiotic hydrolysis reactions (e.g., refs 4 and 8). In addition to the presence of MNX, DNX, and TNX in certain wells, data indicating oxygen depletion in those wells suggest that subsurface conditions supported microbial activity and were conducive to anaerobic RDX metabolism. All but two of those wells had minimum dissolved oxygen
(DO) concentrations of
