Production of chemicals for antagonistic use can in many cases be done using several different synthetic methods. Chemical analysis of such material can provide not only the identity of the agent itself, but also information on the production method. Impurities, such as precursors and impurities therein, byproducts, degradation products and additives give rise to a chemical attribution profile. These signatures (peak areas) were used in this study to construct multivariate data analysis models with the aim to identify the production method of unknown samples. We chose O-(iso-butyl-(S)-(2-diethylaminoethyl) methylphosphonothiolate (Russian VX, RVX, VR) and bis(2-chloroethyl)sulfide (mustard gas, HD) as model agents. They were prepared using various synthetic methods to emulate clandestine lab conditions, and the products were analyzed by GC-MS (EI and CI). The detected compounds were identified where possible and their presence and relative abundance in the various samples was used to construct a PLS-DA model. The datasets for the two agents differed substantially. For RVX, the samples had high levels of attribution signature compounds; however, the same compounds were produced in many of the synthetic routes. Samples from HD synthesis had a considerably lower content of signature compounds. The models correctly predicted the production methods of an external test set. Patterns in the profiles were the key to distinguish between different production methods, rather than the presence of specific compounds. This work demonstrates how PLS-DA modelling of attribution signatures can increase forensic analytical capability for chemical warfare agents, but also for a number of other toxic agents, for instance industrial chemicals, pesticides and other substances that in view of their accessibility also may represent a considerable threat.

Chemical forensics using multivariate data analysis of chemical attribution signatures from production of chemical warfare agents

253rd National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.