Chemical Signatures of TNT-Filled Land Mines

Talanta 54 (2001) 501–513 www.elsevier.com/locate/talanta

Chemical signatures of TNT-ﬁlled land mines

Thomas F. Jenkins a,*, Daniel C. Leggett a, Paul H. Miyares a, Marianne E. Walsh a, Thomas A. Ranney b, James H. Cragin c, Vivian George c

a US Army Cold Regions Research and Engineering Laboratory, Hano6er, NH 03755, USA b Science and Technology Corporation, Hano6er, NH 03755, USA c Draper Laboratory, Fort Bel6oir, VA 22060, USA

Abstract

The equilibrium headspace above several military-grade explosives was sampled using solid phase microextraction fibers and the sorbed analytes determined using gas chromatography with an electron capture detector (GC-ECD). The major vapors detected were the various isomers of dinitrotoluene (DNTs), dinitrobenzene (DNBs), and trinitrotoluene (TNTs), with 2,4-DNT and 1,3-DNB often predominating. Although 2,4,6-TNT made up from 50 to 99% of the solid explosive, it was only a minor component of the equilibrium vapor. The flux of chemical signatures from intact land mines is thought to originate from surface contamination and evolution of vapors via cracks in the casing and permeation through polymeric materials. The levels of external contamination were determined on a series of four types of Yugoslavian land mines (PMA-1A, PMA2, TMA5 and TMM1). The flux into air as a function of temperature was determined by placing several of these mines in Tedlar bags and measuring the mass accumulation on the walls of the bags after equilibrating the mine at one of five temperatures. TNT was a major component of the surface contamination on these mines, yet it accounted for less than 10% of the flux for the three plastic-cased mines, and about 33% from the metal antitank mine (TMM1). Either 2,4-DNT or 1,3-DNB produced the largest vapor flux from these four types of land mines. The environmental stability of the most important land mine signature chemicals was determined as a function of temperature by fortifying soils with low aqueous concentrations of a suite of these compounds and analyzing the remaining concentrations after various exposure times. The kinetics of loss was not of first order in analyte concentration, indicating that half-life is concentration dependent. At 23°C, the half life of 2,4,6-TNT, with an initial concentration of about 0.5 mg kg−1, was found to be only about 1 day. Under identical conditions, the half-life of 2,4-DNT was about 25 days. A research minefield was established and a number of these same four mine types were buried. Soil samples were collected around several of these mines at several time periods after burial and the concentration of signature chemicals determined by acetonitrile extraction and GC-ECD analysis. Relatively high concentrations of 2,4,6-TNT and 2,4-DNT were found to have accumulated beneath a TMA5 antitank mine, with lower concentrations in the soil layers between the mine and the surface. Signatures were distributed very heterogeneously in surface soils, and concentrations were very low (low mgkg−1 range). Lower, but detectable, concentrations of signatures were detectable irregularly in soils near the PMA-1A mines in contrast to the TMA5 mines. Concentrations of signature chemicals were generally below detection limits (B1 mgkg−1) near the TMM1 and PMA-2 mines, even 8 months after burial. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: TNT-ﬁlled land mines; Signature chemicals; GC-ECD

* Corresponding author.

1. Introduction various isomers of DNT and TNT. However, the largest component of the vapor was found to be Detection and elimination of buried land mines 2,4-DNT for all eight domestic TNTs, and all but remains an important and intractable problem in four of the foreign TNTs. In most cases, the many countries throughout the world. The use of concentration of 2,4-DNT in the vapor exceeded plastic-cased mines has reduced their detectability that for 2,4,6-TNT by an order of magnitude or using electromagnetic induction, and geophysical more. techniques such as ground penetrating radar suf- A preliminary study was also conducted to fer high levels of false positives. An approach that assess how quickly vapors from military-grade is under investigation currently is the chemical TNT could permeate soil barriers [3]. One gram detection of vapors that evolve from explosives of military-grade TNT was placed on top of por- and are transported to the air or soil in the tions of two moist test soils (a sand and a clay) immediate vicinity of buried land mines. Research and was then covered with 2.5 cm of the appro- is underway currently in our laboratory and else- priate soil in enclosed septum vials. The vials were where to determine the qualitative and quantita- stored at room temperature for 25 days and the tive nature of the chemical signatures from the headspace above the soil was sampled periodi- explosives within the land mines, that may be cally. The soil itself was also analyzed after 25 detectable at the surface above buried mines. days. The results indicated that 2,4-DNT vapors had permeated through the soil to the headspace within 4 days, while vapors of 2,4,6-TNT were 2. Background not detectable in the headspace until day 12. From these results, it was concluded that DNT In the early 1970s, initial research was con- and TNT vapors diffuse through soil barriers in a ducted to define the vapor signatures of buried relatively short period of time. The ratio of soil to land mines. The strategy employed was to define headspace concentration indicated that most of the vapor signature of military-grade explosives and then try to understand the attenuation and the vapors were retained within the soil barrier. modification of these signatures due to mine cas- Therefore, the surface soil directly above a buried ings and soil barriers. Utilizing gas chromatogra- mine should have considerably larger amounts of phy with an electron capture detector (GC-ECD), DNT and TNT than the headspace. These results an initial study indicated that although 2,4,6-trini- have been confirmed recently using several differ- trotoluene (2,4,6-TNT) accounted for over 99% of ent test soils [4]. the solid military-grade TNT sample, it provided The objectives of the work described here were only 58% of the vapor composition [1]. The man- to analyze the vapor signatures from various ufacturing impurity, 2,4-dinitrotoluene (2,4- types of military explosives, investigate the attenu- DNT), which was only 0.08% of the solid, ation effects of mine casings and soil barriers on accounted for 35% of the vapor composition be- signature compounds under a variety of environ- cause of its much higher vapor pressure. The mental conditions, evaluate their stability in soil, remainder of the vapor composition was found to and quantify the accumulation of signatures in be due to other isomers of DNT and TNT. It was surface soils above buried, TNT-filled land mines. concluded that 2,4-DNT could be as important as 2,4,6-TNT for purposes of detection. In a more thorough study [2], results were 3. Experimental reported for eight domestic and fourteen foreign military-grade TNT samples. Except for two un- 3.1. Analytical standards known compounds that responded to the GC- ECD, the results confirmed that the major portion Analytical standards for 2,4,6-TNT, 2,4-DNT, of the vapor composition was produced by the 2,6-DNT, 1,3-dinitrobenzene (1,3-DNB), RDX T.F. Jenkins et al. / Talanta 54 (2001) 501–513 503

(hexahydro-1,3,5-trinitro-1,3,5-triazine), 2-amino- 3.3. SPME analysis 4,6-dinitrotoluene (2-ADNT), 4-amino-2,6-dini- trotolene (4-ADNT), and HMX (octahydro- Polydimethysiloxane/divinylbenzene SPME 1,3,5,7-tetranitro-1,3,5,7-tetrazocine) were Stan- fibers (65-mm film coating from Supelco) were dard Analytical Reference Materials obtained used in this study because of the enhanced pre- from the US Army Environmental Center, Ab- concentration of 2,4,6-TNT, relative to polyacry- erdeen Proving Ground, Maryland. Standards for late or polydimethylsiloxane fibers. These fibers 2,5-DNT, 2,3,4-TNT, 3,4,5-TNT, and 2,4,5-TNT were used in both laboratory and field holders. were obtained from Dr Stanley Caulder, Naval These fibers were conditioned following the man- Warfare Station, US Navy, Indian Head, Mary- ufacturer’s recommendations and just before use land. Standards for 1,2-DNB, 1,4-DNB, 2,3- were cleaned by baking in a 250°C injector with a DNT, 3,4-DNT, 3,5-DNT, and 3,5-dinitroaniline flow of helium carrier gas. (3,5-DNA) were obtained commercially from Aldrich Chemical Company. Standards for 2,3,5- 3.4. Vapor sampling abo6e military-grade TNT TNT and 2,3,6-TNT were obtained from C. Rib- using SPME fibers audo, Picatinny Arsenal, New Jersey. About 100 mg of military-grade explosive was 3.2. Gas chromatographic analysis placed in either a 40-ml amber vial (Supelco) or a 150-ml glass jar. The containers were closed with Gas chromatographic analyses were obtained aluminum foil, and the foil was held in place using a Hewlett Packard 6890 GC equipped with with a plastic screw cap with an opening to a micro cell 63Ni ECD. Samples were either in- allow sampling through the foil. Samples jected directly or desorbed thermally into a were stored in the dark and equilibrated for packed injection port modified for megabore cap- at least a week before sampling at any of five illary columns and lined with a deactivated Restek different temperatures (−12, −4, 4, 20.5 and Uniliner. Analytical columns were 6 m×0.53 mm 3192°C). Prior to sampling, the SPME device ID fused silica, 1.0 mm film thickness. Quantita- was equilibrated at the same temperature as the tive results were obtained on an HP-5 (5%- sample. diphenyl dimethyl siloxane) column; analyte To sample the headspace, the aluminum foil confirmation was obtained on either a Restek was punctured by the needle housing of the fiber, RTX-200 (crossbond trifluoropropylmethyl and the fiber was extended into the headspace for polysiloxane) or a Restek RTX-225 (50% sorption periods ranging from 0.1 to 20 min. The cyanopropylmethyl–50% phenylmethyl polysilox- opening in the screw cap was just the size of the ane) [5]. outer diameter of the SPME holder and helped The GC oven was held at 100°C for 2 min, then hold the assembly while sampling. After the programmed at 10°C min−1 to 200°C, followed sampling was complete, the fiber was withdrawn by a 20°C min−1 ramp to 250°C (220°C for into the protective needle and the needle RTX-225). For analysis of acetonitrile soil ex- was removed from the vial. The outside of the tracts, the carrier gas was either hydrogen or needle was wiped with a tissue wetted with helium. For solid phase microextraction (SPME), acetone to eliminate any vapors sorbed to the desorption and analysis, only helium could be metal, and the needle was inserted into the injec- used because polynitroaromatic compounds were tion port of the gas chromatograph. The fiber reduced partially to the corresponding was extended and vapors desorbed for 2 min. A monoamino compounds by hydrogen, catalyzed new piece of aluminum foil was placed over the apparently by the metal needle on the SPME vial to maintain equilibrium and enable additional device. sampling. 504 T.F. Jenkins et al. / Talanta 54 (2001) 501–513

3.5. Calibration of sampling rate for SPME sampling rates found for the 2,4- and 2,6-DNTs, fibers because we were unable to locate reliable vapor pressure data for these compounds. SPME fibers have generally been used for collection and analysis of volatile organic com- 3.6. Land mines used for e6aluation of surface pounds where the fiber comes into equilibrium contamination, signature flux rates, and burial at with the concentration in the vapor or aqueous the research minefield solution within a short period of time (seconds to minutes). We found that SPME fibers did not Four types of land mines from Yugoslavia come to equilibrium with vapors of polyni- were studied. The TMA-5 is a plastic encased troaromatics even after many hours. In fact, the antitank mine containing 5.3 kg of TNT. It is mass of individual compounds collected was re- 27.5×30 cm and about 11-cm thick. The lated linearly to sampling time for periods of at TMM1 is a metal-cased antitank mine contain- least 1 h. Thus, to convert mass collected to concentration in the headspace, we developed a ing 5.6 kg of TNT. It is cylindrical in shape, nonequilibrium calibration procedure [4]. 32.6 cm in diameter, and 9.0-cm thick. The This calibration was obtained as follows. The PMA-1A is a shoebox-shaped antipersonnel mass of each compound sorbed by the SPME mine. It contains a 200-g block of TNT and is fiber in a given exposure time was determined 14.4×6.8- and 3.5-cm thick. The case on the using response factors obtained from injection PMA-1A is not vapor tight. The PMA-2 is a of liquid standards. The headspace vapor con- hockey-puck-shaped antipersonnel mine contain- centrations in units of mass per ml above each ing 100 g of TNT. It is 6.8 cm in diameter and pure compound was computed from the vapor 3.0-cm thick. The fuses and detonators were not pressures of 2,4,6-TNT, 2,4-DNT and 2,6-DNT present in these mines, but otherwise they were as a function of temperature from literature val- intact. ues [6]. An equivalent sampling rate (ml min−1) for a given sampling time was then obtained by 3.7. Sampling of exterior contamination le6els on dividing the mass sorbed by the fiber in that intact land mines exposure time (mass per min) by the headspace concentration (mass per ml). Because the mass The concentrations of signature chemicals on collected is linear with sampling time, a sam- −1 the exterior of the Yugoslavian PMA-1A, PMA- pling rate in ml min can be obtained. We 2, TMA-5, and TMM1 land mines were sam- point out that this is not an actual sampled vol- pled prior to burial of these mines at a research ume because the calculation ignores the fiber– minefield. These mines were unused, but had air partition coefficient; that is, it assumes 100% been in storage for up to 50 years. Sampling uptake. This makes no difference in the com- took place at a research minefield in July 1998. puted sample concentrations, however, as long as the partition coefficient is the same for sam- Paper filter disks (1.5 cm in diameter) were ples as for vapors from the pure substances, saturated with methanol and placed randomly which seems to be a reasonable assumption. For over each mine surface. Upon drying, they were 2,4,6-TNT the sampling rate for this type of picked up with forceps and transferred to an fiber was found to be 8.2 ml min−1 at 23°C. amber vial containing 3.0 ml of acetonitrile. The Similarly, the sampling rates for 2,4-DNT and number of disks per mine was keyed roughly to 2,6-DNT were estimated at 4.2 and 4.1 ml the size of its surface. Except for the PMA2, no min−1, respectively, at 23°C. Sampling rates for more than 2% of the surface was sampled by other trinitroaromatics were assumed to be the disks. The vials were transported at 4°C by equivalent to that for 2,4,6-TNT, and other overnight carrier to our laboratory and kept at dinitroaromatics were assumed to have mean −4°C until analyzed. Analyses were conducted T.F. Jenkins et al. / Talanta 54 (2001) 501–513 505 by GC-ECD using the conditions described 3.10. Stability of indi6idual components of land above. This experiment is described in more detail mine signatures in soil elsewhere [7]. The stability of five of the major signature 3.8. Estimation of flux rate of signature chemicals chemicals in soil was evaluated using soil from the into air for intact land mines research minefield. Replicates of air-dried soil were moistened and allowed to stand for 3 days to The flux of signature chemicals from intact re-establish microbial activity, and then they were mines into air was estimated for these four fortified with an aqueous solution of explosive-re- Yugoslavian land mines at temperatures ranging lated analytes (2,4,6-TNT, 1,3-DNB, 2,4-DNT, from 3 to 32°C. These estimates were obtained by 2,6-DNT, and RDX) at soil concentrations of placing individual mines of each type in appropri- approximately 0.5 mg kg−1 for the nitroaromatics ately sized Tedlar bags with one end removed, and 0.14 mg kg−1 for RDX. The final moisture reclosing the bags with plastic tongue-in-groove content of the soil was 40% on a dry weight basis. strips, and allowing the mines to remain for peri- Experimental protocols for this evaluation are ods ranging from 1 to 4 days depending on the reported elsewhere [9,10]. Fortified samples were temperature. After the exposure period, the bags stored in the dark at either 22, +4or−4°C. were opened, the mines removed, and the bags Individual samples were extracted and analyzed rinsed with 5 to 25 ml of acetonitrile depending by reversed-phase high performance liquid chro- on the size of the bag used. The concentration of matography (RP-HPLC) after time periods rang- each signature component was determined by RP- ing from 4 h to 30 days to determine the HPLC as described below for soil extracts. This concentrations of the fortified analytes and their experiment is described in more detail elsewhere environmental transformation products as a func- [8]. tion of time.

3.9. Soil sampling at the research mine ﬁeld 3.11. Soil extractions

Four different types of land mines were buried Soils were extracted using acetonitrile according at a research minefield in July 1998. Soil samples to the ultrasonic bath method specified in SW846 were collected adjacent to some of these mines in Method 8330 [5,11]. Soils from the research mi- September 1998, November 1998, and April 1999. nefield were extracted without air-drying to mini- The top several millimeters of soil was scraped mize volatilization losses, using 2 g soil and 5 ml from the surface using a steel paint scraper. Sam- of acetonitrile. Soils from the stability study were ples of subsoil were obtained with a 3/4 in. stain- also extracted without air drying; 5 g soil and 10 less steel corer in several centimeter increments ml of acetonitrile. from just below the surface sample to the depth of burial. Samples were collected over the mine, 3.12. RP-HPLC analysis along the edges of the mines, and at several distances away from the mines. In several cases, Soil extracts from the stability study, and some mines were excavated totally and soil samples extracts from the soil samples collected at the collected from under the mines as well. All sam- research minefield that were determined to con- ples were placed in 250-ml amber glass bottles and tain explosives by GC-ECD, as well as Tedlar bag stored in a cooler. After each day’s collection, rinses, were analyzed using RP-HPLC as de- samples were packed in ice and shipped by scribed in EPA SW846 Method 8330 [11]. A 15 overnight carrier to the laboratory for analysis. cm×3.9 mm LC-8 column (Nova-Pak from Wa- Samples were maintained cold and in the dark ters, Corp.) was used with an eluent composed of until extraction was conducted, always within a 14/86 isopropanol/water at a flow rate of 1.4 ml week of collection. min−1. 506 T.F. Jenkins et al. / Talanta 54 (2001) 501–513

4. Results and discussion 4.2. Headspace concentration as a function of temperature 4.1. Equilibrium headspace concentrations abo6e military-grade TNT Headspace concentrations for the seven vapors with the highest concentration above Yugoslavian Three samples of military-grade explosive were PMA-1A TNT were obtained at ﬁve temperatures studied — 1966 US military-grade TNT obtained ranging from −12 to 31°C (90.5°C). Because from Picatinny Arsenal, TNT taken from a the sampling rate (ml sampled per min) is a Yugoslavian PMA-1A antipersonnel land mine, function of temperature, the volume sampled for and TNT taken from a Yugoslavian PMA-2 an- a given sampling time was corrected for tempera- tipersonnel land mine. Equilibrium headspace ture. A plot of the logarithm of the concentration concentrations at 22°C were determined using of each vapor versus temperature is presented in SPME collection and GC-ECD. The results for Fig. 1. three of the major vapors detected are presented For all seven compounds, the log of the vapor in Table 1. concentrations appears to be related linearly to For these explosives, either 2,4-DNT or 1,3- temperature demonstrating the exponential rela- DNB were the vapors found at the highest con- tionship between partial pressure and tempera- centration in the equilibrium headspace; other ture. The slope for the six compounds present at isomers of DNT and DNB were also detected, but low concentration (impurities) in the solid are at lower concentrations. The concentration of quite similar, but different than the slope of the 2,4,6-TNT was very similar for the three TNT samples, as expected, since all three of the mili- line for 2,4,6-TNT. Thus, as the temperature is tary-grade TNTs were relatively pure (99%+). reduced, the percentage of the vapor due to the Thus, 2,4,6-TNT should establish its vapor pres- volatile impurities becomes larger, whereas at sure at a given temperature. higher temperature, the percentage from 2,4,6- The ratio of 2,4-DNT to 2,4,6-TNT in the TNT becomes greater. Nevertheless, at all envi- headspace ranged from about 4 to 20. These ronmental temperatures, the largest portion of the ratios are consistent with those reported earlier equilibrium vapor is due to 1,3-DNB and 2,4- for a series of eight US and 16 foreign TNT DNT. Thus, to the degree that the signature from samples [2]. The concentration of 1,3-DNB in the buried land mines is controlled by evolution of headspace was also higher than that for 2,4,6- volatiles from the solid, these results, together TNT; its concentration may be a reﬂection of the with those reported elsewhere, [2] indicate that concentration of benzene in the toluene used for 2,4-DNT and 1,3-DNB are likely to be major the production of TNT and thus may be more compounds of interest for nearly all TNT-based variable for other military-grade TNTs. explosives.

Table 1 Equilibrium headspace concentrations of major vapors above military-grade explosives

Source of explosive (composition) Headspace concentration (pg ml−1) at 22°C

1,3-DNB 2,4-DNT 2,4,6-TNT

US military 1966 (TNT) 0.35a 0.0700.55 Yugoslavian PMA-1A (TNT) 4.6 1.4 0.078 Yugoslavian PMA-2 (TNT) 0.289.7 0.077

a No estimates of uncertainty are provided for headspace concentrations because only one sample of each type of explosive was available. T.F. Jenkins et al. / Talanta 54 (2001) 501–513 507

Fig. 1. Mass of analytes in headspace vapor over PMA-1A TNT sorbed on PDMS/DVB ﬁber.

4.3. Analyte surface concentrations on the TNT. For TMA-5, the mean surface concentra- exterior of land mines tion of TNT is much higher (about 90 ng cm−2) than on the other mines and is an order of When land mines are buried, surface contami- magnitude higher than the mean surface concen- nation is an immediate source of chemical signa- trations of 1,3-DNB and 2,4-DNT. Both the con- ture that will contaminate quickly the soil centration and absolute mass of surface surrounding the mine. The levels of surface con- contamination for the small PMA-2 antipersonnel tamination on four different types of land mines mine are very low compared with the other mines were estimated using surface swabs before the studied, indicating that it may provide a signature mines were buried at a research minefield. The that is less detectable than some of the other number of swabs used per mine was a compro- mines. mise between, (1) providing as good an estimate A larger selection of landmines from seven of the surface contamination as possible; and (2) countries was sampled in 1992 [12]; many of these keeping the level of signature essentially unal- contained RDX/TNT compositions as the main tered. This was important since these mines were charge. A wider range of surface concentrations being buried to study the generation of chemical was indicated by their results. TNT, DNT, and signature in the soil. RDX were found on most mines. No analysis was Estimates of surface contamination with the done for DNB. three most prevalent signature chemicals for four types of land mines are presented in Table 2. The 4.4. Estimation of flux rate of signature chemicals most important explosive-related contaminants into air for intact land mines are 1,3-DNB, 2,4-DNT and 2,4,6-TNT, as expected. The levels of surface contamination for The flux of chemical signatures from land mines each differ from mine type to mine type, and from into air is thought to arise from surface contami- individual to individual within mine types. For nation plus that coming from the interior of the the PMA-1A, PMA-2 and TMM1, the mean sur- mines through openings or cracks in the casing or face concentrations of 1,3-DNB and 2,4-DNT permeation through polymeric materials. The flux range from about 1 to 11 ng cm−2 and are within rates for the four Yugoslavian mines was esti- a factor of two of the concentrations of 2,4,6- mated by measuring the mass of signatures de- 508 T.F. Jenkins et al. / Talanta 54 (2001) 501–513 posited on the inner surfaces of Tedlar bags after through the polymeric casings is a more signifi- equilibration of the mines at temperatures ranging cant contributor to the flux for the plastic-cased from 4 to 33°C for measured time periods ranging mines. from 1 to 4 days [8]. Before and after each experiment, the mines were swabbed to estimate 4.5. Stability of explosi6e-related chemicals in the surface concentrations for each mine used in these soil flux experiments. Two TMA-5, two TMM1, four PMA-1A, and The stability of five of the major explosive-re- four PMA-2 mines were studied individually. The lated signature chemicals in soil was evaluated flux from each was computed from the mass using soil from the research mine field that had deposited in the bag, an estimate of the surface been fortified with an aqueous solution containing area for that type of mine, and the residence time 2,4,6-TNT, 1,3-DNB, 2,4-DNT, 2,6-DNT and for the experiment. It is possible that some explo- RDX. RP-HPLC chromatograms for samples sive vapors also permeated through the Tedlar held at 22°C and analyzed at day 0, 1, and 7 are plastic during these experiments, or were other- presented in Fig. 2. The concentration of 2,4,6- wise not recovered by acetonitrile extraction. TNT declines rapidly; dropping from an initial concentration of 0.56 to 0.31 mg kg−1 after only Thus, these flux values probably represent mini- 1 day, and to 0.17 and 0.13 mg kg−1 after 7 and mum estimates of the true flux. 13 days, respectively. Two transformation prod- Mean flux rates and standard deviation (S.D.) ucts of 2,4,6-TNT (2-ADNT and 4-ADNT), and a for each of the four mine types, computed on a fg transformation product of 1,3-DNB (3-nitroani- cm−2 s−1 basis at 23°C, are presented in Table 3. line) can be identified clearly in the chro- The mean flux of 2,4,6-TNT is very similar for all matograms after day 1. The instability of TNT in four mine types, ranging from 4.4 to 7.7 fg cm−2 −1 soil at this temperature is consistent with results s . Flux rates for 2,4-DNT were higher than for reported elsewhere for several other soils [9,13]. TNT and varied over a factor of ten from mine The concentrations of TNT, RDX, 2,4-DNT, type to mine type; the mean flux for the TMA-5 2,6-DNT and 1,3-DNB are plotted as a function −2 −1 was 123 fg cm s while that for the TMM1 of time after fortification in Fig. 3 for samples −2 −1 was 11.9 fg cm s . The flux for 1,3-DNB was held at 22°C. The stability of 2,4-DNT in this soil also higher than for TNT and varied over a factor is much greater than that for TNT; its concentra- of 80; the mean flux for the PMA-1A was 158 fg tion declined from 0.51 to 0.35 mg kg−1 after 13 cm−2 s−1 while that for the TMM1 was 2.0 fg days. The stability of 2,6-DNT is similar to 2,4- cm−2 s−1. It is interesting that the lowest flux on DNT; the stability of 1,3-DNB is intermediate a surface area basis for all three chemicals was between that for 2,4,6-TNT and 2,4-DNT. No from the metal-cased TMM1. The reason for this substantial loss of RDX as a function of time was is uncertain, but it may be that surface contami- observed at any temperature, and this agrees with nation is the main contributor to flux for the the results reported elsewhere for aerobic soil metal-cased mine, but permeation of the vapors [9,13]. Table 2 Concentration of major signature chemical on exterior of land mine casings

Mine Surface area (cm2)Disks/mine Surface area sampled (%) Surface concentration (ng cm−2, mean9S.D.)

2,4,6-TNT2,4-DNT1,3-DNB

345 4PMA-1A 2 9.093.8 4.892.6 5.093.5 PMA-2 118 1.390.943 0.9590.43 1.390.5 TMM1 102240 96.2111.096.57.3 13.2919.7 2720TMA-5 15 1 3.191 6.193.9 89.9987.5 T.F. Jenkins et al. / Talanta 54 (2001) 501–513 509

Table 3 Flux of three explosive-related components from intact land mines

Replicates Surface area (cm2)MINE Flux (mean9S.D.) at 23°C (fg cm−2 s−1)

1,3-DNB 2,4-DNT 2,4,6-TNT

PMA-1A 4 345 158952 63928 7.792.4 PMA-2 118 569314 48921 4.892.4 2TMM1 2240 2.090.4 11.992.2 4.490.8 TMA-5 27202 27.791.1 123914 6.890.1

At lower temperatures, the stability of all the highest concentrations detected in the soils were analytes increased substantially. For example, the obtained in the soil segments near the mines with TNT concentration declined from 0.56 to 0.31 mg concentrations as high as 4270 mgkg−1 for 2,4,6- kg−1 in 13 days at +4°C, but only from 0.56 to TNT, 3510 mgkg−1 for 2,4-DNT, 2690 mgkg−1 0.48 mg kg−1 for the same time period at −4°C. for 2ADNT and 2760 mgkg−1 for 4ADNT. We plotted the concentration of these analytes The frequency of detection and the concentra- with time on a log C/C0 (concentration/initial tions of these four compounds declined rapidly concentration) basis to assess whether the kinetics for samples collected near the surface for TMA-5. of loss was first order in analyte concentration. In For soils collected at the surface, 90% were below all cases, these plots were clearly nonlinear indi- a detection limit of 1 mgkg−1. The highest surface cating that the loss was not first order. Thus, the soil concentrations obtained for 2,4-DNT, half-life for each analyte is concentration depen- 2ADNT, 4ADNT and 2,4,6-TNT were 60, 90, 70 dent. For the initial concentrations we studied and 5.2 mgkg−1, respectively. The distribution of (about 0.5 mg kg−1), the half life was about 1 day signatures in surface and subsurface soils about for 2,4,6-TNT, about 10 days for 1,3-DNB, about the TMA-5 mines was quite heterogeneous. 20 days for 2,6-DNT, about 25 days for 2,4-DNT, Explosive signature chemicals were detected in and much greater than 30 days for RDX at 22°C. about 20% of the soil samples collected near PMA-1A mines. Signatures (2ADNT and 4.6. Concentrations of signature chemicals in the 4ADNT) were detected in only one of the surface soil near buried land mines soils for this type of mine. Qualitatively, the chemicals detected in the subsurface soil samples Soil samples adjacent to a number of buried were 2,4-DNT, 2ADNT, 4ADNT, 2,4,6-TNT, land mines were collected and analyzed 2, 4, and 1,3-DNB and 2,6-DNT. The highest concentra- 9 months after burial. Over 600 samples were tions observed were much lower than those found analyzed, and it is not possible to discuss the near the TMA-5 mines; for 2,4-DNT, the maxi- results of all these analyses in detail here. Instead, mum concentration was 77 mgkg−1; for 2ADNT, examples of the levels of signature accumulation it was 314 mgkg−1; for 4ADNT, it was 317 mg for various individual mines will be discussed, as kg−1; for 1,3-DNB, it was 32 mgkg−1; and for they related to differences for various mine types. 2,4,6-TNT, it was only 1.5 mgkg−1. Chemical For the soil samples collected after 2 months of signatures were detected in only one soil sample burial, there was a substantial difference in the collected near a PMA-2 mine and were not de- frequency of detection of explosive-related com- tected in any soil from areas with TMM1 mines. pounds near the different types of mines. The Greater concentrations of signature chemicals highest frequency of detection (24%) was for the were found in soils when sampled 4 months after TMA-5 mines where the chemicals present most burial. In addition to more time for accumulation often, and at the highest concentration, were 2,4- of signatures, higher rainfall over this period in- DNT, 2ADNT, 4ADNT and 2,4,6-TNT. The creased soil moisture levels compared with that 510 T.F. Jenkins et al. / Talanta 54 (2001) 501–513 for the 2-month samples. We have also done formation products of TNT (2-ADNT and 4- experiments that demonstrate that, under given ADNT). The lower concentration of 2,4,6-TNT conditions, chemical flux from plastic mine cas- relative to its transformation products confirms ings is consistently greater into water than into that 2,4,6-TNT is relatively unstable at warm air, suggesting that fluxes may also have been environmental conditions. Lower concentrations higher into wet soil. Here again, much higher of other DNT isomers were detected in some levels of signatures were found in soil samples soils, as were transformation products of 1,3- collected near the TMA-5 and PMA-1A mines, DNB (3-nitroaniline) and 2,4-DNT (4-amino-2-ni- and evidence was found that these signatures had trotoluene). Only very low concentrations of migrated upwards; signatures were detected in 2-ADNT and 4-ADNT were detected in several several surface samples (Table 4). The signature isolated samples collected near the TMM1 (less chemicals present at the highest concentrations than 1 mgkg−1) and PMA-2 (8 mgkg−1) mines. were 2,4-DNT and the two environmental trans- Otherwise, soil concentrations of signature chemi-

Fig. 2. RP-HPLC chromatograms for stability study; extracts from soils held at 22°C on days 0, 1 and 7. T.F. Jenkins et al. / Talanta 54 (2001) 501–513 511

also collected 8 months after burial. These samples were collected in the spring after the soils had a prolonged period of low temperatures. Mean concentrations under these two mines were 952, 2420, 7490, 605, and 837 mgkg−1, respectively, for 1,3-DNB, 2,4-DNT, 2,4,6-TNT, 4-ADNT, and 2-ADNT. The 2,4,6-TNT and 1,3-DNB concentrations increased substantially, probably due to their increased stability at the lower soil temperatures during winter. Concentrations of signature chemicals in the surface soils were low (generally less than 20 mgkg−1 for all the five major components) for soils from these two TMA-5 mines. Concentrations of signature chemicals in surface soils near the PMA-1A mines were also low, generally below 10 mgkg−1; many samples were below detection limits. The concentration of the two TNT transformation products (2-ADNT and 4-ADNT) were generally higher than that for 1,3-DNB, 2,4-DNT or 2,4,6-TNT with concentra- Fig. 3. Soil from research minefield concentration (mg kg−1) tions in the surface soil as high as 357 mgkg−1. vs. time (days) at 22°C. Concentrations of all the signature chemicals in soils near TMM1 and PMA-2 mines remained cals were generally below a detection limit of very low (generally less than a detection limit of 1 m −1 about 1 gkg near these two mines. mgkg−1) even after mines have been buried for 8 At the 4-month sampling event, one TMA-5 months. Even below the mine, soil concentrations mine was selected and completely excavated; a were less than 10 mgkg−1 for both these types of total of 87 soil samples were collected during mines. excavation including several samples from below the mine. The concentrations of signature chemicals were distributed very heterogeneously in the 5. Conclusions soil around the perimeter of the mine. Concentra- tions in surface soil directly above the center of The three major source chemicals for chemical the mine were very low (10 mgkg−1 for 2,4-DNT detection of TNT-filled land mines are 1,3-DNB, and less than 1 mgkg−1 for 2,4,6-TNT). Concen- 2,4-DNT, and 2,4,6-TNT. Surface concentrations trations in the surface soil above the perimeter of and flux for these signature chemicals vary among the mine were very spatially heterogeneous, but the various types of land mines and for individu- often higher than that over the center of the mine. als within these types. Flux for the plastic-cased Signature concentrations under the mine, where mines, and for those types that are not well soils were wet, were much larger with concentra- sealed, are much greater than that for metal-cased tions for 1,3-DNB, 2,4-DNT, 2,4,6-TNT, 2- mines and those with good seals. ADNT, and 4-ADNT of 34, 2070, 266, 616, and The stabilities of the three signature chemicals 641 mgkg−1, respectively. These high concentra- in soils follow the order 2,4-DNT more stable tions under the mine may serve as a source of than 1,3-DNB and 2,4,6-TNT, all three being signature moving upward toward the surface as more stable as temperature declines. Environmen- soils dry during prolonged drought periods. Sam- tal transformation converts 2,4,6-TNT to two ples of soil from below two TMA-5 mines were monoamino products (2-ADNT and 4-ADNT) 512 T.F. Jenkins et al. / Talanta 54 (2001) 501–513 and these chemicals appear to have increased Acknowledgements stability in soil relative to TNT, becoming part of the land mine signature in soils near buried mines. The authors acknowledge the financial support In many cases, the signatures most prevalent in and encouragement provided by Dr Regina Du- surface soils are 2,4-DNT, 2-ADNT, and 4- gan from the Defense Advanced Research ADNT. Projects Agency (DARPA), Arlington, VA, USA. Precipitation appears to wash signature chemi- They also acknowledge Tommy Berry from the cals underneath land mines where relatively high US Army Engineer Waterways Experiment Sta- concentrations can accumulate for some mines. tion for providing mine locations, and together This reservoir of signature can then migrate up- with Charlotte Hayes and Glenn Myrick from ward, forming a halo of low concentration signa- AsCII Corporation, for collection of soil samples ture in surface soils. This signature is distributed and mine surface swipe samples. Philip Thorne, heterogeneously and does not extend very far Louise Parker and Jessica Kopczynski of the US beyond the edges of the mine. Army Cold Regions Research and Engineering Additional work is underway to improve our Laboratory are acknowledged for their assistance understanding of the dynamics of signature accu- in analysis of soils from Ft. Leonard Wood. Dr mulation as a function of burial time and season. Stanley Caulder of the Naval Warfare Station, Additional types of mines are being studied to Indian Head, MD, USA is acknowledged for include some containing RDX as well. supplying several reference standards that are not

Table 4 Concentrations of signature chemicals in soil near buried TMA-5 and PMA-1A land mines, November 1998a

PositionDepth Concentration (mgkg−1)

2,4,6-TNT2,4-DNT1,3-DNB 2-ADNT4-ADNT

Soil near TMA-5 land mine Sraevrmine Bd BdSurfaceOver Bd Bd Bd Bd BdSurface BdNortheast Bd Bd 0–6 cm Northeast Bd Bd Bd Bd Bd Bd Bd6–10cm BdNortheast Bd Bd 10–15 cm BdNortheast Bd Bd Bd Bd BdNorthwestSurface Bd Bd Bd Bd 0–6 cm Northwest Bd Bd Bd Bd Bd Northwest Bd Bd6–10cm Bd Bd Bd Surface BdSoutheast 220143780 BdSoutheast0–5cm 27917144 443 5–10 cm Southeast 10 236 10 147 239 10–15 cm Southeast 68 823 75 401 584 Southwest Bd BdSurface Bd Bd Bd 0–5 cm Southwest Bd 122 17 542 873 5–10 cm Southwest 14 441 64 528 651 10–15 cm Southwest Bd Bd Bd Bd Bd Soil near PMA-1A land mine Over mineSurface Bd Bd Bd Bd Bd South Bd BdSurface Bd44 0–2.5 cmSouth 9 52 84 6 2.5–5 cm South 1811142 95 88 5–8 cmSouth 7 31 trace 19 25

a Bd, Concentration below a detection limit of 1 mgkg−1. T.F. Jenkins et al. / Talanta 54 (2001) 501–513 513 available commercially. This publication reflects Development Center-Cold Regions Research and Engi- the personal views of the authors and does not neering Laboratory, ERDC-CRREL Technical Report 00-2, Hanover, NH, USA, in press. suggest or reflect the policy, practices, programs [8] D.C. Leggett, J.H. Cragin, T.F. Jenkins, T.A. Ranney, or doctrine of the US Army or Government of the Release of Explosive-Related Vapors from Landmines, United States. US Army Engineer Research and Development Center- Cold Regions Research and Engineering Laboratory, ERDC-CRREL Technical Report 01-6, Hanover, NH, USA, in press. References [9] C.L. Grant, T.F. Jenkins, S.M. Golden, Experimental Assessment of Analytical Holding Times for Nitroaro- [1] R.P. Murrmann, T.F. Jenkins, D.C. Leggett, US Army matic and Nitramine Explosives in Soil. US Army Cold Cold Regions Research and Engineering Laboratory, Regions Research and Engineering Laboratory, Special Special Report 158, Hanover, NH, USA, 1971. Report, Hanover, NH, USA, 1993, pp. 93–111. [2] D.C. Leggett, T.F. Jenkins, R.P. Murrmann, US Army [10] P.H. Miyares, T.F. Jenkins, Estimation of the half lives of Cold Regions Research and Engineering Laboratory, key components in the chemical vapor signature of land Special Report, Hanover, NH, USA, 1977, pp. 77–116. mines. US Army Engineer Research and Development [3] T.F. Jenkins, D.C. Leggett, R.P. Murrmann, Preliminary Center-Cold Regions Research and Engineering Labora- Investigation of the Permeability of Moist Soils to Explo- tory, ERDC-CRRELTechnical Report, Hanover, NH, sive Vapor. In Appendix of Spangler, G., 1974. USA, in press. [4] T.F. Jenkins, D.C. Leggett, T.A. Ranney. Vapor Signa- [11] US Environmental Protection Agency Office of Solid tures from Military Explosives Part 1. Vapor Transport Waste SW846 Method 8330, 1994. from Buried Military-Grade TNT, US Army Cold Re- [12] E. Bender, A. Hogan, D. Leggett, G. Miskolczy, S. gions Research and Engineering Laboratory, Special Re- MacDonald, Surface Contamination by TNT, J. Forensic port, Hanover, NH, USA, 1999, 99–121. Sci. (1992) 1673–1678. [5] M.E. Walsh, Talanta 54 (2001) 427–438. [13] M.P. Maskarinec, C.K. Bayne, L.H. Johnson, S.K. Hal- [6] P.A. Pella, Anal. Chem. 48 (1976) 1632–1637. lady, R.A. Jenkins, B.A., Tomkins, Stability of explosives [7] D.C. Leggett, T.F. Jenkins, A. Hogan, T.A. Ranney, P.H. in environmental water and soil samples. Oak Ridge Miyares, External Contamination on Landmines by National Laboratory Report ORNL/TM-11770, Oak Organo-Compounds. US Army Engineer Research and Ridge, TN, USA, 1991.