| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
imunekc
a Dep. of Chemical Engineering, Univ. of Mississippi, Oxford, MS. K.M. Dontsova, duty station: U.S. Army Engineer Research and Development Center, 3909 Halls Ferry Rd., Vicksburg, MS 39180
b Computer Sciences Corporation, Vicksburg, MS
c Univ. of California, Riverside
d Environmental Processes Branch, Environmental Laboratory, U.S. Army Engineer Research and Development Center, Vicksburg, MS
* Corresponding author (Kateryna.Dontsova{at}gmail.com)
Received for publication January 5, 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: ADNTs, amino-dinitrotoluenes (2ADNT, 2-amino-4,6-dinitrotoluene and 4ADNT, 4-amino-2,6-dinitrotoluene) • ARAMS, Army Risk Assessment Modeling System • CEC, cation exchange capacity • Comp B, Composition B • DNX, hexahydro-1,3-dinitroso-5-nitro-1,3,5-triazine • HMX, octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine • HPLC, high performance liquid chromatography • kd, adsorption coefficient • MMR, Massachusetts Military Reservation • MNX, hexahydro-1-nitroso-3,5-dinitro-1,3,5-triazine • OC, organic carbon • OM, organic matter • RDX, hexahydro-1,3,5-trinitro-1,3,5-triazine • TNT, 2,4,6-trinitrotoluene • TNX, hexahydro-1,3,5-trinitroso-1,3,5-triazine
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Batch-determined fate and transport parameters for explosives commonly used by the military were reviewed by Brannon and Pennington (2002). Existing models for predicting environmental fate of explosives rely on information obtained in batch tests with pure explosive compounds (Pennington and Patrick, 1990; Comfort et al., 1995; Xue et al., 1995; Price et al., 1997; Brannon et al., 1998; Price et al., 2001a; Price et al., 2001b; Brannon et al., 2002; Brannon et al., 2005). However, explosives are typically formulated in combinations of two or more compounds with waxes, stabilizers, and binders. Binders and waxes have been shown to decrease dissolution rates of individual explosive compounds from formulations (Lynch et al., 2002b; Phelan et al., 2002). The effects of formulation on other fate and transport processes are unknown.
Solid particles ranging in size from small (<2mm) to large (up to the diameter of the projectile) may be deposited on the soil surface when projectiles fail to detonate completely (a low-order detonation) (Pennington et al., 2005; Jenkins et al., 2006). Similar residues can be deposited when duds or unexploded ordnance that are disposed of by detonation in place fail to detonate completely. These failures result in locally scattered chunks of Comp B on the soil surface (Jenkins et al., 2006). An understanding of transport behavior of the components of this residue is necessary because of detections of one component of Comp B, RDX, in groundwater of military training ranges (USEPA, 1997).
This study focused on Comp B, one of the primary explosive formulations currently used in artillery projectiles and many other munitions. To accurately represent field conditions, residues from deliberate low-order detonations of Comp B-filled artillery shells were used. The objective was to determine transport parameters for TNT and RDX from solid phase Comp B under saturated conditions. The parameters for solid Comp B were compared with those for pure RDX and TNT; RDX and TNT in dissolved Comp B; and with adsorption coefficients determined in batch tests.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
To ensure even distribution of the soil and to minimize preferential flow, soils were poured into columns in small amounts at a time and manually pressed. Average packed bulk density (BD) was 1.27 g cm–3 for Adler soil, and 1.60 g cm–3 for Plymouth soil. Bulk density was determined from the dry mass of soil used to pack a column of known volume.
Saturated Flow Experiments
Columns were saturated from the bottom with 0.005 M CaBr2 solution for 4 h. Pore volume was determined during saturation as the volume of solution necessary to fill the packed column. Columns were then connected to the pump on the top and flow was started. Outflow was monitored to determine volumetric flow rate. Once a constant flow rate was established (target flow rate was 1.18 mL min–1, or solution flux of 0.87 cm h–1; measured average flow rate was 1.09 mL min–1, or solution flux of 0.8 cm h–1), one pore volume of solution was pumped through the column to assure uniform flow (Gaber et al., 1995). The selected water flux was lower than the saturated hydraulic conductivities for both soils and about 20% of the average 1-yr rainfall intensity (4 cm h–1) for the eastern United States. Flow then was switched to solutions of RDX, TNT, or Comp B; solid Comp B was placed directly on the soil surface. Target concentrations of military grade RDX and TNT were 10 mg L–1; measured concentrations were 10.03 mg L–1 and 9.4 mg L–1, respectively. Composition B solution had an RDX concentration of 6.87 mg L–1, TNT of 3.28 mg L–1, and octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) of 0.58 mg L–1. Octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine is an impurity in Comp B typically at 1 to 10% (3.5% in our Comp B).
Radiolabeled 14C-RDX and 14C-TNT were added as tracers to the bulk solutions (specific activity of RDX was 7.68 mCi mmol–1 and TNT 4.08 mCi mmol–1) at 0.145% for RDX in solution and 0.278% for TNT. The radiotracer allowed easy monitoring of explosives concentrations in outflow as the experiment progressed.
For the solid Comp B treatment, steady-state flow was briefly stopped so that 10.0 g of low-order detonation residues containing solid Comp B (65% Comp B and 35% nonexplosive debris, primarily soil particles) between 2 layers of glass wool could be placed on the surface of the soil. Solid Comp B was a composite of the 0.25- to 2-mm size fractions of residue from low-order detonation experiments conducted for another project (Pennington et al., 2005). In that project, detonations were performed on a large tarp so that residue could be recovered by sweeping. Therefore, residues were relatively clean except for uncontaminated soil introduced via occasional shrapnel tears in the tarp. The explosives component included 57.6% RDX, 38.9% TNT, and 3.5% HMX. Flow was resumed at the same rate as soon as solid Comp B residues were in place.
After four to six pore volumes of explosive solutions, flow was switched back to the background solution (or solid Comp B was removed), which was applied for another four to six pore volumes. For every soil and explosive treatment two experiments were conducted. In one, flow was uninterrupted, whereas in the other, flow was interrupted for 12 to 24 h to allow explosives to equilibrate with the soil. During flow interruption both inflow and outflow from the column were stopped and the column remained saturated. Flow interruption techniques are often used to elucidate rate-limited sorption processes (Murali and Aylmore, 1980; Fortin et al., 1997; imunek et al., 2002).
Outflow samples were analyzed using high performance liquid chromatography (HPLC) (Waters HPLC, GenTech Scientific, Arcade, NY) and liquid scintillation (Tri-Carb 2500TX Liquid Scintillation Analyzer, PerkinElmer Life and Analytical Sciences, Boston, MA). Every fifth sample (approximately every 0.1 L) was analyzed by scintillation counting and every 20th (approximately every 0.4 L) by HPLC. Standard USEPA Method 8330 (HPLC) (USEPA, 1994) was used with the addition of HMX and the following degradation products: mono- and diamino-nitrotoluenes and azoxy dimers of TNT, and mono-, di-, and trinitroso derivatives of RDX. Detection limits were 20 µg L–1 for TNT, HMX, 2-amino-4,6-dinitrotoluene (2ADNT), and 4-amino-2,6,-dinitrotoluene (4ADNT); 200 µg L–1 for RDX and 50 µg L–1 for RDX degradation products. Concentrations of explosives in solid Comp B residue were determined by shaking 5 g of residue with 150 mL acetonitrile in 3 sequential extractions followed by filtering (45-µm GF filter) and HPLC analysis.
To determine the longitudinal dispersivity (
) for each soil and monitor for signs of preferential flow, all solutions were prepared with 3H2O tracer (Selim et al., 1995). Specific activity of 3H2O was 2.18 mCi mmol–1 with 1.239 x 10–6% of 3H2O in solution. Liquid scintillation was used to measure tritium activity.
Numerical Analysis
Experiments were analyzed using the HYDRUS-1D code for simulating the one-dimensional movement of water, heat, and multiple solutes in variably saturated porous media (imunek et al., 2005). Minimization of the objective function that includes the sum of squared deviations between measured and simulated concentrations was performed by the Levenberg-Marquardt nonlinear minimization algorithm (Marquardt, 1963). The following models were used in analysis: convection-dispersion equation for 3H2O tracer and two-site sorption model (with decay) (chemical nonequilibrium model) for explosives and their degradation products.
Transport of Nonreactive Solute
Transport of the 3H2O tracer can be described using the convection-dispersion equation for constant water content, flux density, and dispersion coefficient:
 | [1] |
is the water content (cm3 cm–3) equal to porosity for saturated experiments, z is the spatial coordinate (cm), t is time (h), and D is the dispersion coefficient (cm2 h–1) assumed to be the product of the longitudinal dispersivity, (cm), and q divided by .
Two-Site Sorption Model (with Decay)
The concept of two-site sorption (Selim et al., 1976; van Genuchten and Wagenet, 1989) that permits consideration of nonequilibrium adsorption–desorption reactions was used to describe the transport of explosives. The two-site sorption concept assumes that the sorption can be divided into two parts:
 | [2] |
Sorption on the Type-2 nonequilibrium sites is assumed to be a first-order kinetic rate process. The mass-balance equation for the Type-2 sites in the presence of degradation is given by:
 | [3] |
is a first-order kinetic rate constant for sorption to kinetic sites (h–1). A linear distribution coefficient was used because the majority of explosive compounds exhibit linear sorption by soils (Pennington et al., 1999) and our preliminary batch studies showed that 3 out of 4 studied isotherms were linear. The governing transport equation is then as follows for constant water content, flux density, dispersion coefficient, bulk density, and distribution coefficient:
 | [4] |
is bulk density (g cm–3), µw is first-order rate coefficient in the liquid phase (h–1), and 
is the zero-order rate coefficient that we used to account for dissolution of solid Comp B (µg cm–3 h–1). In Eq. [4] the first-order rate coefficients µs and µw lump all removal mechanisms of the parental compound, such as irreversible sorption and decay. Complicated attenuation processes of TNT that include covalent bonding of decomposition products to soil organic matter functional groups make separation of irreversible sorption and decay difficult. Coefficients µw and µs were set to be equal for this experiment because adsorbed explosives can decay (Achtnich and Lenke, 2001; Balakrishnan et al., 2004) and setting decay coefficients equal decreased the number of fitted parameters. Only µw is reported further in the paper. Dissolution was modeled as a zero-order production process of limited duration. It was assumed that during the time when Comp B was present at the top of the profile, there was a zero-order production process in a small layer close to the top boundary.
Parameter Estimation
The numerical analysis of experimental data was performed as follows. First, the conservative tracer breakthrough curves were used to estimate the longitudinal dispersivity, (cm). The chemical nonequilibrium model was then used to analyze explosives breakthrough curves. Longitudinal dispersivity was fixed at a value determined for tracer. The following parameters were estimated for explosives: fraction of sites with instantaneous adsorption, f, adsorption coefficient, kd (g–1cm3), first-order rate coefficient for dissolved phase (degradation rate), µw (h–1), and first-order rate coefficient for two-site nonequilibrium adsorption, (h–1). Radiotracer results gave an estimate of the movement of the total mixture of explosives without considering individual parent and/or daughter products, whereas HPLC measurements gave an estimate of the movement of TNT, RDX, and their transformation products. In 14C radiotracer experiments the rate coefficients (µw and µs) corresponded to irreversible attenuation, whereas in HPLC experiments they corresponded to the sum of attenuation and degradation. In this paper, reversible adsorption was defined operationally as the process that allowed explosives to be transported further as a result of the desorption, whereas irreversible attenuation was defined as the process that made explosives unavailable for further transport. The production coefficient, , was fitted in HYDRUS-1D simulations for experiments with solid Comp B.
Mass-balance calculations were performed for the 14C radiotracer and HPLC results (Table 2) by integration of each breakthrough curve. The accuracy of mass-balance estimates was evaluated using recovery of tritiated water as a conservative tracer.
|
imunek and Hopmans, 2002). Parameter estimates were considered significant if they were different from zero (confidence intervals did not intersect with zero). If parameter estimates were consistently nonsignificant, parameters were removed from the model. To improve the confidence in adsorption coefficients and dissolution rates, kinetic parameters (f and ) were removed from the model for solid Comp B experiments. However, if only a few parameter estimates were nonsignificant, these parameters were kept in the model, but not used to draw conclusions. Interrupted and uninterrupted flow experiments were treated as replicates for the purposes of statistical analysis of transport parameters. For comparison between treatments, differences were considered significant when greater than the sum of standard errors of the means multiplied by 1.96 (for 95% probability).
Batch Adsorption Studies
Linear distribution coefficients determined from HYDRUS-1D simulations were compared to kds determined by five-point batch studies using RDX and TNT with 14C radiolabel. Adsorption partitioning was conducted at a 4:1 solution/soil mass ratio in 25-mL glass centrifuge tubes. Triplicate samples and a blank were spiked at concentrations of 0.001, 0.0025, 0.005, 0.0075, and 0.01 µCi mL–1 (37, 92.5, 185, 277.5, and 370 Bq mL–1). Total explosive concentrations were 1 to 10 mg L–1 RDX and 1 to 10 mg L–1 TNT. The samples were shaken for 24 h on a reciprocating shaker at 180 excursions per minute, centrifuged at 8288 relative centrifugal force (RCF) for 60 min, and the aqueous phase removed and analyzed by liquid scintillation counting. Linear adsorption isotherms were determined from the solution concentrations and calculated sediment concentrations. Equilibration time was determined from kinetic adsorption curves established by shaking 10 mg L–1 radiolabeled RDX and TNT solutions with soils at 4:1 solution/soil mass ratio for 0, 0.5, 1, 2, 6, 12, and 24 h, and 5 d.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The RDX sorption isotherms were linear (Fig. 1) in agreement with published reports (Xue et al., 1995; Singh et al., 1998; Tucker et al., 2002; Yamamoto et al., 2004). The kd values determined from the isotherms (0.48 cm3 g–1 for Adler soil and 0.65 cm3 g–1 for Plymouth soil) were similar to values reported from shake tests with soils of comparable texture, OM content, and CEC (0.16 to 2.2 cm3 g–1, average of 0.93 cm3 g–1, median 0.92 cm3 g–1) (Brannon et al., 1992; Ainsworth et al., 1993; Xue et al., 1995; Myers et al., 1998; Pennington et al., 1999; Price et al., 2000).
|
Conservative Tracer
Hollow triangles in Fig. 2 and 3 represent measured values for outflow concentrations of the conservative tracer, 3H2O, labeled on figures as "measured water." Dispersivity was larger for the coarser Plymouth soil (0.62 to 2.29 cm) than for the finer Adler soil (0.07 to 0.17 cm) (Table 3), but generally small as expected for short repacked columns.
|
|
|
RDX
Solid diamonds represent measured values for the 14C-RDX radiotracer concentration for the two soils with and without interrupted flow (Fig. 2). For 14C-RDX, breakthrough was observed later than for the conservative tracer indicating adsorption to the solid phase. The average kd value determined from the curves was slightly larger for Plymouth than Adler soil, but the difference was not significant.
The 14C-RDX radiotracer exhibited little retardation, as is generally reported in the literature (Selim et al., 1995; Brannon and Pennington, 2002; Tucker et al., 2002). The kd values for 14C-RDX determined from HYDRUS-1D (Table 3) were smaller than values determined in the batch experiments (Fig. 1). Batch adsorption coefficient values for Plymouth soil were similar to values reported by Speitel et al. (2002) for this soil. High mobility of 14C-RDX was supported by the mass balance (Table 2). For Adler soil, sorption was reversible; recovery of 14C-RDX varied between 98 and 105%. Furthermore, irreversible attenuation was not statistically significant when modeled by HYDRUS-1D. However, for Plymouth soil, some irreversible attenuation was observed, as indicated by lower 14C-RDX recovery in outflow (81 to 91%) and significant numbers for irreversible attenuation rate coefficient (0.012 ± 0.002 h–1). Breakthrough curves also indicated that the outflow concentration was less than inflow for Plymouth soil, but similar to inflow concentration for Adler soil. Incomplete recovery of 14C-RDX in Plymouth soil is consistent with 8% unextractable 14C-RDX after 168 d observed by Singh et al. (1998) and irreversible attenuation of RDX in Plymouth (7.4%) reported by Speitel et al. (2002).
The asymmetric shape of the breakthrough curves, particularly for Plymouth soil, and results of HYDRUS-1D simulations suggest rate-controlled adsorption and desorption. An estimated 30% of adsorption sites in Adler soil and 33% in Plymouth soil exhibited kinetic adsorption, with the rate of exchange in 0.12 to 0.16 h–1 range (Table 3). The decrease in concentration following flow interruption also indicated chemical nonequilibrium.
Outflow was not analyzed for RDX by HPLC because it was assumed that 14C-RDX accurately represents RDX in soil as there is little RDX degradation under aerobic conditions (Pennington and Brannon, 2002; Speitel et al., 2002).
TNT
The 14C-TNT radiotracer exhibited greater attenuation than 14C-RDX as indicated by a delay in the breakthrough curve and smaller outflow concentrations (Fig. 3). Part of the attenuation was attributed to reversible adsorption, and part to irreversible attenuation.
Average kds were 3.2 times greater for 14C-TNT than for 14C-RDX in both soils (Table 3) suggesting a greater affinity of TNT for adsorption sites. This difference was close to the 2.4-fold difference in log Kow of the compounds (2.06 and 0.87 for TNT and RDX, respectively) suggesting that partitioning of TNT to the OM may be a significant mechanism for reversible adsorption.
For both soils, 14C-TNT kds were small compared to values reported in the literature (Brannon and Pennington, 2002) (0.53 and 0.63 cm3 g–1 for Adler and Plymouth, respectively, difference not statistically significant). However, the irreversible attenuation contribution was considerable with the attenuation rate significantly smaller in Adler (0.023 ± 0.005 h–1) than in Plymouth (0.101 ± 0.022 h–1) soil. A 4.4-fold difference in rate between soils was consistent with difference in OC content (0.2 and 0.78% for Adler and Plymouth soils, respectively), which is a matrix for irreversible attenuation (Thorn et al., 2002). The mechanism for irreversible attenuation of TNT involves reduction to amino transformation products, followed by covalent bonding to the functional groups, e.g., carboxylic acid, of OM (Thorn et al., 2002).
The 14C-TNT kds determined from batch shake tests (2.4 and 1.6 cm3 g–1 for Adler and Plymouth, respectively) and column experiments (0.53 and 0.63 cm3 g–1 for Adler and Plymouth, respectively) differed. Batch kds were probably larger because they included irreversible attenuation and TNT is known to have a contribution of covalent bonding to total retention by soils (Thorn et al., 2002). Selim et al. (1995) found that adsorption coefficients for TNT obtained from batch experiments were larger than those obtained from column experiments and explained it by more complete mixing in batch experiments.
Asymmetric breakthrough curve, as well as the decrease in concentration when flow was interrupted, indicated rate-controlled sorption of 14C-TNT in agreement with Selim et al. (1995). Similar to 14C-RDX, 14C-TNT adsorption was instantaneous to a greater fraction of sites in Adler (0.63) than in Plymouth (0.38) soils. Both kinetic sorption and retarded nonlinear intra-particle diffusion (Fesch et al., 1998) may be causing nonequilibrium effects in TNT transport in soils. Rate of sorption () accounts for both forms of the nonequilibrium.
Outflow concentrations of HPLC-measured TNT in Plymouth soil were small and inconsistent and, therefore, were not used for the simulation. However, for Adler soil TNT kds were similar to the values for the 14C-TNT radiotracer, whereas degradation was considerably greater, because HPLC-measured TNT included both irreversible attenuation and degradation (Table 3). Estimates for kinetic parameters (f and ) in HPLC-measured TNT breakthroughs were nonsignificant. The limited number of samples analyzed by HPLC combined with the selection of samples with non-zero TNT concentrations (and error) resulted in larger confidence intervals and lower R2.
Mass-balance calculations for 14C-TNT were consistent between similar treatments (Table 2). Seventy-three to seventy-five percent of 14C-TNT was recovered in Adler and 41 to 42% in Plymouth soil. Speitel et al. (2002) reported that 74 to 85% of 14C-TNT radiotracer was recovered for the subsoil coming from MMR, which is the same soil as Plymouth. For other soils, 20 to 50% of added TNT was reported to be unextractable (Comfort et al., 1995; Kaplan and Kaplan, 1982b; Selim et al., 1995). Since degradation cannot be estimated from radiotracer data, unrecovered 14C-TNT is assumed to be covalently bound to the soil OM by reactions reported by Thorn et al. (2002). Also, TNT may form coplanar complexes with K-exchanged phyllosilicate clays in soils (Weissmahr et al., 1999); however, in studied soils only 0.5 to 1.2% of exchange sites were occupied by K. Selim et al. (1995) reported that irreversible attenuation was a dominant retention mechanism of TNT.
In Adler soil, the amount of TNT recovered by HPLC was much lower than 14C-TNT recovered since 14C-TNT elutes as both TNT and its decomposition products. HPLC showed recovery of 7 to 15% as TNT and 29 to 35% as a sum of TNT and its measured products. In Plymouth, HPLC TNT recovery was even lower than in Adler soil with less than 5% of inflow. Measured products included ADNTs and azoxy compounds in agreement with Comfort et al. (1995) who observed that a significant fraction of TNT was recovered as amino degradates of TNT.
Forty to forty four percent (14C-TNT recovery minus HPLC TNT and its measured products) in Adler soil and 36 to 39% in Plymouth soil were unaccounted for, likely as undefined or unmeasured degradation products. TNT mineralization to CO2 is unlikely (Kaplan and Kaplan, 1982a; Speitel et al., 2002).
Solution Phase Composition B
TNT and RDX from dissolved Comp B behaved like pure TNT and RDX (Fig. 4) with kds, irreversible attenuation rate, degradation rate, fraction of sites with instantaneous adsorption, and rate of exchange for kinetic sites for 14C-RDX radiotracer, RDX, and TNT in Comp B not significantly different from the ones obtained for pure explosives (Table 3).
|
The HMX, which was not studied separately, was stable with no significant degradation or irreversible attenuation and 100% recovery (96 to 103%). The HMX kds (0.43 to 0.48 cm3 g–1) were in general agreement with batch values summarized by Brannon and Pennington (2002). Close agreement between reported batch kds for HMX (0.086 to 5.02 cm3 g–1, median 0.77 cm3 g–1, average 1.62 cm3 g–1) (Myers et al., 1998; Brannon et al., 1999; Pennington et al., 1999) and column-determined values can be explained by the absence of irreversible attenuation.
Degradation products of TNT accounted for 65% of outflow in Adler soil and 25% in Plymouth soil. They included ADNTs, which predominated, in agreement with Selim et al. (1995), and azoxy compounds. There was more 4ADNT than 2ADNT in agreement with Comfort et al. (1995). Furthermore, 4ADNT exhibited an earlier breakthrough. This can be explained by higher partition coefficients for 2ADNT (Pennington and Patrick, 1990), and greater production of 4ADNT because biotic reduction of nitro group in para position is more prevalent. Kaplan and Kaplan (1982a) reported that 4ADNT accounted for the majority of the two amines in compost spiked with TNT.
Degradation products of RDX were only a small part of the outflow (Table 2): 0.4 to 4% of RDX input (0.4 to 5% of the RDX outflow). The fraction of products doubled in interrupted flow experiments, as additional time allowed degradation to proceed further. Hexahydro-1-nitroso-3,5-dinitro-1,3,5-triazine (MNX) was the only detected degradation product in Plymouth soil. In Adler soil, DNX (hexahydro-1,3-dinitroso-5-nitro-1,3,5-triazine) and TNX (hexahydro-1,3,5-trinitroso-1,3,5-triazine) were detected. Concentrations of all degradation products increased in the outflow after flow interruption. Degradation pathways of RDX and TNT were modeled, but are not reported, since parameter estimates were nonsignificant.
Solid Composition B
Breakthrough curves from solid Comp B indicated that despite the small range of particle sizes, variability in dissolution rate of Comp B among the experiments was significant (Fig. 5). For two experiments outflow concentrations decreased in time, possibly due to a decrease in dissolution rate as smaller particles were dissolved. Smaller particles with a larger surface area dissolve faster due to greater contact with solution (Lynch et al., 2002a). The sharp initial peak observed for the interrupted flow experiment in Adler soil can be explained by the presence of microscopic Comp B particles on the surface of the principle particles as observed by Phelan et al. (2003). As dissolution is strongly affected by particle size, microscopic particles will dissolve fastest and result in a sharp increase in solution concentration. Another prominent feature of the breakthrough curves was similarity in outflow concentrations of TNT and RDX, contrary to what was observed for dissolved Comp B (Fig. 4) and to observations of groundwater concentrations in the field (Clausen et al., 2004). Variance from the field results was explained by the small length of the column and limited residence time of solution. Longer residence would cause a greater change in solution concentration of TNT than RDX due to differences in degradation rates. Longer soil profiles enhance the contribution of reversible sorption and irreversible attenuation. Difference between solid and dissolved Comp B experiments indicated that TNT has a greater dissolution rate than RDX.
|
When the above approach was applied to the data, simulated dissolution rates (Table 3) varied between the experiments, reflecting variability in outflow. However, a highly significant correlation (P > F < 0.01) existed between HYDRUS-1D-estimated dissolution rates of HMX and RDX and a significant correlation (P > F < 0.05) between RDX and TNT. This supported the accuracy of HYDRUS estimates of dissolution rates of Comp B and indicated that dissolution rates of explosives in a mix are linked. Dissolution rate was the highest for TNT and lowest for HMX. According to Brannon and Pennington (2002) when tested alone, TNT has the highest dissolution rate (4164 µg cm–2 h–1), followed by HMX (702 µg cm–2 h–1) and RDX (361 µg cm–2 h–1). Several studies indicated that the dissolution rate of individual explosives is decreased by being present in formulations, whereas solubility is not affected (Lynch et al., 2002b; Phelan et al., 2002; Taylor et al., 2004). In this study, the link between dissolution rates of different components of Comp B was analyzed by comparing slopes of regressions between dissolution rates of RDX, TNT, and HMX (Table 4) with other relevant properties of studied explosives normalized to RDX: individual dissolution rate, measured content in Comp B, and solubility. None of the listed properties gave a good explanation for the dissolution rates; however, a product of individual dissolution rate and explosives content in Comp B was in very good agreement with HYDRUS-1D estimates of dissolution rates in the formulation. This can be explained by dissolution being related to the surface area, that for each individual compound decreased proportionally to its content in the formulation.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Combining column flow-through studies with numerical analysis of outflow concentrations of explosives and an inert tracer allowed separation of the contributions of physical and chemical nonequilibrium processes. Whereas for studied soils with uniform particle-size physical nonequilibrium was negligible, chemical nonequilibrium (kinetic sorption) contributed to transport and distribution of both RDX and TNT. The contribution of chemical nonequilibrium was confirmed by interrupted flow experiments.
The behavior of dissolved Comp B was similar to that of pure RDX and TNT. Estimated adsorption and degradation rates for RDX, TNT, and HMX were generally smaller for solid Comp B than for dissolved explosives in column and batch studies. Therefore, using distribution coefficients from batch studies and transport parameters determined for pure explosives in solution can underestimate transport of formulations. However, great variability in outflow concentrations in experiments with solid phase Comp B indicated that dissolution was controlling transport processes. The dissolution rate of Comp B was affected by the dissolution rate of pure compounds and by their fraction in Comp B.
|
ACKNOWLEDGMENTS |
|---|
imunek was supported by the Terrestrial Sciences Program of the Army Research Office (Terrestrial Processes and Landscape Dynamics and Terrestrial System Modeling and Model Integration). Authors are grateful to T. Myers of ERDC, Vicksburg, MS for his review of the manuscript. Disclaimer: Permission has been granted by the Chief of Engineers to publish this report. |
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
imunek, J., and J.W. Hopmans. 2002. Parameter optimization and nonlinear fitting. p. 139–157. In J.H. Dane and G.C. Topp (ed.) Methods of soil analysis. Part 1. Physical methods. Third ed. SSSA, Madison, WI.imunek, J., D. Jacques, J.W. Hopmans, M. Inoue, M. Flury, and M.T. van Genuchten. 2002. Solute transport during variably-saturated flow: Inverse methods. p. 1435–1449. In J.H. Dane and G.C. Topp (ed.) Methods of soil analysis. Part 1. Physical methods. Third ed. SSSA, Madison, WI.imunek, J., M.T. van Genuchten, and M. ejna. 2005. The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat, and multiple solutes in variably-saturated media. Version 3.0. Department of Environmental Sciences, University of California Riverside, Riverside, CA.This article has been cited by other articles:
|
|
 |
|
  J. Simunek, M. Th. van Genuchten, and M. Sejna Development and Applications of the HYDRUS and STANMOD Software Packages and Related Codes Vadose Zone J., May 27, 2008; 7(2): 587 - 600. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
