An explosive solution

By their very nature explosives are difficult to analyse in a laboratory setting. But as Christopher Borton and Loren Olson explain – there is a fast and accurate way of analysing trace level explosive compounds in ground water and soil

Both military installations and their local municipalities are increasingly concerned about hazardous explosive compounds from military weaponry entering water supplies^{1, 2, 3,} and there is clearly an important need for a fast, accurate and definitive detection method for these compounds and their degradation products. Energetic chemicals used as military explosives are mostly organic compounds containing nitro groups, such as nitroaromatic, nitroamine and nitrate ester compounds. However, they can be difficult to analyse in a laboratory setting because, by design, they are unstable, thermally labile molecules. The standard technique used for the analysis of these compounds has been high performance liquid chromatography (HPLC) with UV detection, following the guidelines set by the United States Environmental Protection Agency (USEPA) method 8330⁴. Although this method has been successful, the lack of sensitivity of UV detection requires a costly and time-consuming concentration step during sample preparation. More importantly, UV detection is not selective and requires a second HPLC analysis and, even after this step, there is still the possibility of false positive detection.

For this reason a selective and sensitive method for the detection of a wide range of nitroaromatic, nitroamine and nitrate ester compounds, which require a special set of conditions for sensitive and stable ionisation, has been developed using LC/MS/MS with negative ion atmospheric pressure chemical ionisation (APCI), for the analysis of explosives residues in ground water and soil samples. By using MS/MS detection, the concern of false positive detections has been virtually eliminated.

Instrumentation consisted of a HPLC system (Shimadzu SCL-10Avp integrated HPLC with two LC10ADvp pumps) equipped with an autosampler and coupled with a triple quadrupole mass spectrometer (Applied Biosystems API 3200 LC/MS/MS system) using negative ion APCI. The mass spectrometer was tuned and optimised to use two multiple reaction monitoring (MRM) transitions per analyte, so that the most sensitive MRM transition was used for quantitation (quantifier) while the second most sensitive MRM was used for detection confirmation (qualifier).

Water and soil samples for analysis were extracted in acetonitrile prior to LC/MS/MS. Water extracts were prepared by solid phase extraction, by loading 1l of the water sample onto a Waters Porapak RDX phase cartridge, dried with nitrogen and eluted with 5.0ml of acetonitrile, to which an equal volume of 1.0% acetic acid in water was added. Soil extracts were prepared from 2.0 ±0.1g of dry soil by sonication at 6.0 ±2.0°C for 18 hours, in 5.0ml of acetonitrile, after which the supernatant was removed and combined with an equal volume of 1.0% acetic acid. The low concentration of acetic acid was used in the final extracts to preserve any 2,4,6-trinitrophenyl-N-methylnitramine (Tetryl) present in the sample, because Tetryl quickly degrades at high pH and its detection becomes inconsistent. A calibration curve was prepared in 50:50 water:acetonitrile over a range of 0.49μg/l – 1.0mg/l.

As shown in table 1, the most intense precursor ion for each compound varied in nature. As expected, some of them produced a [M-H]- quasi-molecular ion, while others produced acetate adducts, ie. [M+CH3COO]-, or an intense M- molecular anion. The latter species can be produced in negative ion APCI by electron capture or by charge exchange with ionised ambient gas and/or solvent.^{5, 6}

Because of possible thermal degradation of compounds with APCI, chromatographic separation of most of the analytes was particularly important with this method. For example, 2,4,6-trinitrotoluene (TNT) can thermally degrade into 2,6- and 2,4-dinitrotoluene so, if these compounds were to co-elute, TNT would show up in the MRM transition of the dinitrotoluene compounds as a potential false positive. By chromatographically separating the compounds and monitoring two MRM transitions, the possibility of false positives is essentially eliminated.

Ratios of the measured area counts of both the quantifier and qualifier MRM transitions were monitored. Table 1 shows the transitions used for each analyte along with the observed ion ratio. The expected ion ratio for each explosive analyte was determined from the calibration curve. All analytes had a calculated correlation coefficient of 0.99 or greater for both quantifier and qualifier MRM transition using a linear regression fit for all analytes with a weighting factor of 1/x. If a weighting factor was not used the curve was then forced through the origin.

For each water and soil sample extracted a matrix spike was prepared to ensure extraction efficiency, and a continuing calibration verification (CCV) sample was analysed after the samples to verify that the instrument had maintained calibration. Matrix spike tests on all of the samples showed recoveries of 80 – 120% indicating good recoveries from the sample preparation procedure. Figure 1 shows the results of a matrix sample spiked with a 500μg/l standard mixture.

For each analyte, the lower limit of detection (LOD) was determined with a signal to noise ratio of 3:1 in both the quantitation ion and the confirmation ion, and the lower limit of quantitation (LOQ) was determined with a measured signal to noise ratio of 10:1. After sample preparation, all analytes could be detected at levels ranging from 0.05μg/l to 2.0μg/l in water samples and 25.0μg/kg to 1000 μg/kg in soil samples. Detailed detection limits for each analyte are given in table 2. Ion ratios were compared to those obtained from the calibration curve for confirmation and good correlation between standards and spiked matrices was found. Figure 2 shows a water sample spike with all target explosives.

Additionally, studies were performed to show that ion suppression is not present for these samples when using APCI. Although it is uncommon to see ion suppression in APCI negative ion mode, it is important to test for ion suppression, especially if inconsistent or unreliable results are noticed. To monitor possible ion suppressions, a post-column infusion of a 1.0μg/ml standard solution of all compounds of interest was performed during the injection of 20μl of matrix onto the column. The infusion of the standard solution creates a constant signal on the MRM traces so, if the sample matrix were to cause ion suppression, a dip in the signal should be recorded.

The method was applied to water and soil samples collected around the Denver and San Francisco Bay area. No explosive compounds were detected above the established LOD in any of the samples tested. The ion suppression test was performed on one water and two soil matrices and, in all cases, no significant ion suppression was observed.

This method has proven to be a sensitive and specific method that is a significant improvement from the traditional LC/UV technique. With the added specificity of MRM and the ion ratio confirmation using two transitions, the concern of false positive detection has virtually been eliminated. A matrix spike test on the matrices indicated that all analytes can be recovered with good sensitivity and show an accurate ion ratio for confirmation. It has also been shown that ion suppression in water and soil matrices is not a concern with APCI ionisation for the compounds tested here.

Future developments involve an Applied Biosystems API 5000 LC/MS/MS system with atmospheric pressure photo ionisation (APPI) to develop an ultra sensitive direct injection water analysis method, which would eliminate any sample preparation procedure and allow for higher throughput. These sensitive analytical methods for fast, unambiguous screening for trace levels of common explosive compounds also have potential applications outside environmental protection, such as in forensic and criminal laboratories and in combating the rise in terror threat.

Table 1. Monitored MRM transitions. *Ratio = quantifier MRM area/qualifier MRM area.

Table 2. Detection limits of tested analytes (LOD = lower limit of detection; LOQ = lower limit of quantitation).

Acknowledgements
The authors would like to thank Dr Paul Winkler of GEL Analytical in Golden, Colorado, for help with sample preparation during this study. We would also like to thank our colleagues at Applied Biosystems/MDS Sciex for their thoughts and contributions.

References
1. “Innovative Uses of Compost: Composting of Soils Contaminated by Explosives”, (1997). http:/www.epa.gov/epaoswer/non-hw/compost/explos.pdf
2. “Hazard Assessment for Munitions and Explosives of Concern (MEC) Workgroup”, (2007). http://ww.epa.gov.swerffrr/documents/hazard_assess_wrkgrp.htm
3. “6.3: Explosives”, (1983). http://www.epa.gov/ttn/chief/ap42/ch06/final/c06s03.pdf
4. “Explosives Residues: Standard Operating Procedure”, (1994). http://www.epa.gov/Region2/desa/hsw/sop8330.pdf
5. Harrison AG. (1983). Chemical Ionization Mass Spectrometry, C.R.C. Press, Inc., USA.
6. Kauppila TJ, Kotiaho T, Kostiainen R and Bruins AP. (2004), Journal of the American Society for Mass Spectrometry 15, 203-211.

By Christopher Borton (Staff Field Application Specialist) and Loren Olsen (Product Specialist) Applied Biosystems/MDS Sciex.