| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Civil Engineering, University of Nebraska, Lincoln, NE 68588-0531
b School of Natural Resources, University of Nebraska, Lincoln, NE 68583-0915
c Water Science Laboratory, University of Nebraska, Lincoln, NE 68583-0844
* Corresponding author (scomfort{at}unl.edu)
Received for publication March 16, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Abbreviations: HPLC, high performance liquid chromatography • LC–MS, liquid chromatography–mass spectrometry • RDX, hexahydro-1,3,5-trinitro-1,3,5 triazine • USDOE, United States Department of Energy
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The USDOE Innovative Treatment and Remediation Demonstration (ITRD) program was initiated to evaluate emerging technologies that may potentially replace inefficient or ineffective technologies. The ITRD process for the Pantex Plant recommended three candidate technologies for further testing: (i) oxidation by KMnO4, (ii) anaerobic biodegradation, and (iii) chemical reduction by dithionite-treated (reduced) aquifer material. This research discusses in situ chemical oxidation of RDX by permanganate.
In situ chemical oxidation involves the addition of a chemical oxidant to destroy contaminants in-place. Potassium permanganate (KMnO4) is an oxidizing agent with a strong affinity for organic compounds containing carbon–carbon double bonds, aldehyde groups, or hydroxyl groups. Considerable research with chlorinated solvents has shown that permanganate is attracted to the negative charge associated with the 
electrons of chlorinated alkenes such as tetrachloroethene, trichloroethene, dichloroethene, and vinyl chloride (Oberle and Schroder, 2000). Although the chemical structure of RDX does not readily lend itself to reaction with permanganate, IT Corporation and SM Stoller Corporation (2000) initially demonstrated effective RDX loss by KMnO4 treatment. By using 14C-labeled RDX and samples of the Pantex perched aquifer sand, our objectives were to measure RDX transformation and mineralization using varying KMnO4 concentrations and determine the effects of initial pH and RDX concentration on reaction rates. Because mineralization was incomplete (<90%) when lower KMnO4 concentrations were used, we also attempted to identify transformation products and determined the ability of native biota to mineralize degradates produced by permanganate in aquifer microcosms.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Initial tests treated replicated experimental units (n = 4) containing 150 mL of aqueous 14C-RDX (2.5 mg L–1, 30000 dpm mL–1, uniformly ring-labeled) with 20000 mg KMnO4 L–1. Because no 14C-labeled volatile organics were detected in initial trapping tests, subsequent experiments concentrated on measuring loss of parent RDX and 14C activity from solution and production of 14CO2. Solution samples from the reaction flask and CO2 traps were taken daily for the first 5 d and at 48-h intervals in the following weeks. Changes in solution RDX concentrations were determined by removing 1.2-mL aliquots, quenching the reaction with 120 µL of 0.5 mg MnSO4·H2O mL–1, centrifuging at 12000 x g for 10 min, and transferring 1.0 mL of supernatant to a high performance liquid chromatography (HPLC) vial for analysis. RDX was quantified at 220 nm by HPLC using a Keystone NA column (Keystone Scientific, Bellefonte, PA) with an isocratic mixture of methanol and H2O (30:70) at a flow rate of 1.5 mL min–1 (Bier et al., 1999). Standards were prepared with analytical RDX obtained from AccuStandard (New Haven, CT). Changes in solution 14C activity were determined by removing 0.5 mL of sample, mixing with 18 mL of Ultima Gold scintillation cocktail (Packard, Meriden, CT), storing for 24 h, and then determining activity by liquid scintillation (LS) analysis. At the end of the experiment, we filtered the solution from the reaction flask through Whatman (Maidstone, UK) #1 filters, and allowed the precipitate to dry. We then mixed 0.5 g of precipitate with 400 mL of Combustaid (Packard), and determined precipitated 14C by combusting the precipitate in a Packard biological oxidizer. The oxidizer gas stream was trapped in a 3:2 (v/v) mixture of Carbosorb/Permafluor (Packard) and counted by LS analysis. RDX destruction kinetics were fit to a pseudo first-order rate equation by nonlinear regression analysis using the computer software SigmaPlot 2000 (SPSS, 2000).
Batch Experiments
Additional batch experiments were performed in 250-mL Erlenmeyer flasks and agitated with an orbital shaker. Initially, we duplicated the 20000 mg KMnO4 L–1 treatment used in the flow-through system (with and without aquifer solids), but subsequent treatments used lower KMnO4 concentrations to treat RDX-contaminated slurries. For slurry experiments, we used fresh aquifer sediment obtained during drilling of the perched aquifer (approximately 90 m). Aquifer material was sent to our laboratory in ice-packed coolers and immediately transferred to a cold room (4.5°C) for storage. Soil analyses (Midwest Labs, Omaha, NE) determined that the aquifer material was approximately 91% (w/w) sand, 3% silt, and 6% clay with 0.1% organic material and had an alkaline pH (2:1 solution to soil ratio, pH = 9). Although mostly sand, our observations during handling of the Pantex material revealed that small pebble-sized occlusions of clay were also present.
Aquifer slurries consisted of 75 g aquifer material (dry weight) mixed with 150 mL of 14C-RDX (2.5 mg L–1). Slurries were treated with 0, 1000, 2000, and 4000 mg KMnO4 L–1. In a second experiment, these same KMnO4 treatments were repeated but after 21 and 50 d, additional KMnO4 (dry solid) was added to the reaction flasks at the original concentrations to simulate repeated injections of KMnO4. Sampling for temporal changes in RDX and 14C was conducted as described above.
To determine permanganate consumption by the Pantex perched aquifer, five concentrations of KMnO4 (5–500 mg L–1, 2:1 solution to soil ratio) were mixed with the Pantex sand and shaken on an orbital shaker for 48 h. Permanganate concentrations were measured with a UV/vis spectrophotometer (UV-2101PC; Shimadzu, Kyoto, Japan) at 525 nm.
Effects of Initial RDX Concentration and pH
To quantify the effects of initial RDX concentration and pH on destruction kinetics, we repeated the slurry experiments with 4000 mg KMnO4 L–1 and varied the initial RDX concentrations (1.3, 2.6, 4.3, 7.2, and 10.4 mg L–1) and pH. The pH effects on RDX transformation and mineralization rates were determined by treating 5.0 mg RDX L–1 with permanganate in the following buffers: 10 mM NaH2PO4 (initial pH = 4.1), 10 mM Na2HPO4 (pH = 8.3), 10 mM Na3PO4 (pH = 11.3), 5 mM NaH2PO4 + 5 mM Na2HPO4 (pH = 7.5), and an unbuffered control (pH = 9.6). Temporal changes in RDX, 14C, and pH were monitored.
RDX Transformation Products
To identify RDX transformation products, we used permanganate concentrations between 500 and 4000 mg L–1 to treat 14C-labeled RDX solutions (150 mL) without aquifer material under batch conditions. Solutions were analyzed with an HPLC equipped with a photodiode array (Shimadzu) and radio-isotope detector (Packard). Several combinations of HPLC columns and mobile phases were employed in an attempt to separate and identify degradate peaks and included Keystone NA and Aquasil reverse phase (Keystone Scientific) and Supelcogel C-610H ion exclusion columns (Supelco). Typical mobile phases were varied and included 20:80 acetonitrile to water or 50:50 methanol to water. Nitrate was analyzed using cadmium reduction, following Lachat Method 12-107-01-1C (Lachat Instruments, Loveland, CO). Ammonium was analyzed colorimetrically with a Reflectoquant test kit (EM Science, Gibbstown, NJ).
Headspace above an aqueous RDX solution (150 mL, 12.5 mg L–1, 14C-labeled) and slurry (150 mL, 75 g) was monitored for N2O and 14CO2 evolution following treatment with 4000 KMnO4 L–1. Each reactor contained a 7-mL scintillation vial filled with 6 mL of 0.5 M NaOH to trap evolved 14CO2. Headspace, solution (RDX and 14C), and 14CO2 traps were sampled weekly for four weeks. We obtained N2O samples by removing 1000 µL from the headspace of the batch reactors with a gas tight syringe and transferring to a 12-mL N2–flushed sample vial. Headspaces were subsequently exposed to the atmosphere for approximately 5 min while we sampled for solution RDX and exchanged 14CO2 traps. Sample vials were transported to the University of Nebraska's Water Sciences Laboratory where 500 µL of the diluted gas was injected into a Hewlett-Packard (Palo Alto, CA) 5890 GC equipped with a Restek (Bellefonte, PA) 80/100 Porapak Q Column (2-m length x 2-mm diameter) and electron capture detector (ECD). Calibration standards were prepared by diluting N2O standards (Scott Specialty Gases, Plumsteadville, PA) and used to quantify N2O concentrations.
Two liquid chromatography–mass spectrometer (LC–MS) systems were used to identify RDX degradation products: an LCQ HPLC–MS system (Finnegan, Austin, TX), which employs an ion-trap mass analyzer, and a Quattro micro triple quadrupole LC–MS–MS (Waters, Milford, MA). Samples injected into the LCQ ion trap were separated with a BetaBasic C-18 column at 50°C using a mobile phase of 0.01 M NH4OOCH (in H2O) and isopropanol (80:20, pH = 8 with NH4OH) at a flow rate of 0.2 mL min–1 (Cassada et al., 1999). Data were collected in full scan mode (negative ion) from 50 to 600 amu. Samples injected into the triple quadrupole used the same column and flow rate, but the mobile phase consisted of 2% (v/v) formic acid–acetonitrile–methanol (60:24:16).
Biodegradation of Permanganate-Treated RDX
To determine the biodegradation potential of the RDX transformation products generated during KMnO4 treatment, 12 flasks containing 75 g aquifer material and 150 mL of aqueous RDX (5 mg L–1) were prepared. Eight of the flasks received 4000 mg KMnO4 L–1, while control flasks received none. We allowed KMnO4 to react with RDX until RDX was undetected by HPLC (<50 µg L–1). We then transferred 25 mL of solution from each of the flasks into 50-mL Teflon tubes, and quenched half (n = 4) with 400 µL MnSO4 (0.5 mg MnSO4·H2O per mL of H2O); all of the tubes were centrifuged at 5500 x g for 10 min. The supernatant containing 14C-RDX transformation products was then used as the stock solution (hereinafter referred to as 14C stock) for the aquifer microcosm experiments.
Aquifer microcosms were prepared in 250-mL glass jars with septa-containing screw-top lids. Each experimental unit contained 75 g (dry weight) of fresh aquifer sand, 15 mL of a 14C stock, and a 14CO2 trap consisting of a 7-mL glass vial with 6 mL of 0.5 M NaOH. The 15 mL of added 14C stock solution saturated the Pantex sand and provide a thin film of overlying solution. Radioactivity (dpm) of the unquenched and control treatments was diluted with H2O to provide the same initial radioactivity concentration as the quenched treatment. Carbon dioxide traps were sampled once per week to determine cumulative 14CO2 produced.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
|
|
One potential problem with permanganate is the production of insoluble manganese dioxide. Li (2000) warned that MnO2 can potentially cause plugging and flow diversion, and Lee et al. (2003) showed that low permeability zones can form along chlorinated solvent plumes, which decrease efficiency and lead to permanganate migrating away from the contaminant zone. Siegrist et al. (2002) showed that MnO2 can commingle with clay and silt particles in water and increase the amount of filterable solids generated in situ. Our batch experiments with the Pantex aquifer solids support this, as we noticed that the slurry became progressively more viscous the longer the reaction took place, and sampling with a pipette became more tedious. This problem occurred even with the 1000 mg KMnO4 L–1 treatment but may have been exacerbated by the fact that the batch reactors were agitated during treatment, which suspended the clay in solution, rather than keeping the clay in lenses as in the undisturbed material. Although our laboratory batch experiments differ from the miscible displacement that would occur in the field, the potential for plugging within the perched aquifer still needs to be taken into consideration.
Effects of Initial RDX Concentration and pH
Initial RDX concentrations between 1.3 and 10.4 mg L–1 did not greatly affect reaction rates when treated with 4000 mg KMnO4 L–1 (Fig. 4)
. RDX transformation rates (i.e., k = 0.08–0.108 d–1) and the extent of mineralization that occurred were similar among all RDX concentrations (Fig. 4). This indicates that slight differences in initial concentrations had negligible effects on destruction kinetics. The RDX concentrations observed in the Pantex perched aquifer (IT Corporation and SM Stoller Corporation, 2000) are similar to those used in our experiments. Monitoring solution 14C activity in the various RDX treatments revealed that roughly 40 to 50% of the activity was still present after 35 d (Fig. 4B). Because this was more than what we had observed in the previous experiment (10–30% 14C remaining after 70 d; Fig. 3), we acidified the slurry (pH < 3) at the end of the experiment and observed that an additional release of 35 to 40% of 14C activity (Fig. 4B). The retention of 14C activity by the permanganate-treated RDX solution (as dissolved CO2 or HCO–3) is problematic and related to the alkaline conditions that occur during treatment (see initial pH results below). Consequently, acidification is needed for an accurate assessment of mineralization kinetics.
|
The only pH treatment effect noted was the greater loss of solution 14C activity in the most acidic treatment (NaH2PO4) during the first few days (Fig. 4D). Because the NaH2PO4 treatment was significantly more acidic than the other treatments (pH 4–6), we again attributed this observation to the release of dissolved CO2 at the lower pH.
RDX Transformation Products
Past research has shown that permanganate is effective in attacking compounds with carbon–carbon double bonds, aldehyde groups, or hydroxyl groups and is especially attracted to the negative charge associated with the electrons of chlorinated alkenes (e.g., tetrachloroethene, trichloroethene, dichloroethene, and vinyl chloride) (Oberle and Schroder, 2000). RDX has none of the above characteristics, and its complete oxidation (mineralization) can take several days to weeks (e.g., Fig. 2 and 3). By comparison, Yan and Schwartz (1999) observed half-lives ranged from less than a minute to 4 h for chlorinated ethylenes treated with permanganate (1 mM MnO–4). By comparison, when we treated 1,3,5-trinitrobenzene (TNB; 2 mg L–1) with permanganate (4000 mg L–1), it was undetectable within 30 min. The large difference in destruction kinetics between a nitroaromatic (TNB) and a nitramine (RDX) indicates that different destructive mechanisms by permanganate are likely operative.
Considering that most of the reaction between KMnO4 and RDX occurred under alkaline conditions, one hypothesis is that permanganate is reacting with hydroxyl ions to form hydroxyl radicals (·OH) (Ladbury and Cullis, 1958; Wronska and Baranowska, 1964) and that the attack on RDX is indirect. Such an attack can occur at pH > 9 (Gates-Anderson et al., 2001). The limiting step would be the production of ·OH, which is responsible for the initial attack on the ring, and once that occurs, the permanganate concentration is high enough to quickly oxidize the RDX ring fragments. Although previous research supports ·OH formation under alkaline conditions, our pH experiments showed that pH had little influence on destruction kinetics (Fig. 4C). Moreover, earlier work by Bier et al. (1999) demonstrated that when ·OH attack on 14C-labeled RDX was the dominant mechanism (via the Fenton reaction), at least six radioisotope peaks were detected within 2 h during temporal sampling with HPLC-radioisotope detection. Using a similar analytical system, no major 14C-labeled degradates were detected during permanganate treatment of RDX (data not shown). The absence of detectable 14C-labeled peaks indicates that the initial attack of RDX by permanganate is likely the rate-limiting step and further degradation of the intermediates is relatively fast.
An alternative explanation is that permanganate is facilitating RDX hydrolysis, either directly or in concert with MnO2 production. The slow reaction kinetics observed (e.g., several weeks; Fig. 2) are more indicative of a hydrolysis reaction and as previously reported (Hoffsommer et al., 1977), once proton abstraction from the ß carbon occurs and a nitro group is released (E2 reaction), the double bond formed on the triazine ring would favor attack by MnO–4. Although ·OH attack could also cause a double bond to form, Balakrishnan et al. (2003) studied alkaline hydrolysis of RDX and identified 4-nitro-2,4-diaza-butanal (4-NDAB) and N2O as reaction products. We believe 4-NDAB is also a product of permanganate treatment of RDX. Liquid chromatographic analysis using both ion trap (Fig. 5) and triple quadrupole detection (data not shown) revealed the KMnO4–treated RDX contained ion peaks at m/z 118 (M-H). Ion trap also revealed a m/z 164 peak (Fig. 5). 4-NDAB has a mass of 119 and based on the mobile phase used, we believe the m/z 164 detection peak is a negative formate adduct of 4-NDAB ([M + HCOO]–). Adduct ion formation and detection is enhanced in electrospray using a buffer such as ammonium formate or formic acid (Cassada et al., 1999). Because 4-NDAB is not commercially available, we could not confirm it in our sample. However, Fournier et al. (2002) similarly observed a degradate peak of identical mass (m/z 118) during RDX biodegradation by Rhodococcus sp. Strain DN22 and matched their chromatograph to one produced by alkaline hydrolysis of RDX. Methylene dinitramine and other nitroso-RDX degradation products were not detected in the permanganate-treated RDX solutions analyzed by LC–MS.
|
Measurement of nitrous oxide release showed that N2O was a product of permanganate oxidation of RDX (solution and contaminated slurry) and constituted 20 to 30% of the N balance (Fig. 6) . Comparisons between the treated RDX solution and slurry showed that loss of parent RDX was similar in both medias but N2O production was greater from the pure solution (Fig. 6). This indicates that other reactions (i.e., biotic) may be occurring when the aquifer solids are present and altering the distribution of reaction products.
|
|
4000 mg L–1) produce slow but sustained rates of RDX destruction and mineralization in the presence of aquifer material from the Pantex Plant. These results support permanganate as an in situ chemical treatment for remediating the RDX-contaminated aquifer at the Pantex site. |
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Related articles in JEQ:
This article has been cited by other articles:
|
|
 |
|
  C. L. Duke, R. C. Roback, P. W. Reimus, R. S. Bowman, T. L. McLing, K. E. Baker, and L. C. Hull Elucidation of Flow and Transport Processes in a Variably Saturated System of Interlayered Sediment and Fractured Rock Using Tracer Tests Vadose Zone J., November 20, 2007; 6(4): 855 - 867. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Soil Science Society of America Journal | Journal of Plant Registrations | The Plant Genome | |||
