Although chemical warfare agents (CWAs) are prohibited under international law, there have been numerous crimes that violates the 1997 Chemical Weapons Convention (CWC) during the last decade, especially in the civil war in Syria where sarin, mustard gas and chlorine have all been used. CWAs have also been used in political assassinations and attempts thereof. In such situations, it is important to identify the deployed CWA and to find information on how it was produced, as this information is potentially of considerable value for any ensuing judicial process. The development and use of advanced analytical methods and multivariate data analysis methods are required to produce this kind of robust forensic evidence and intelligence.   

This thesis describes conducted research that aims at retrospectively tracing the synthesis methods applied in the production of CWAs. In three studies, methods for the determination of the employed synthetic route have been assessed. The relative distribution of the impurities gave a unique profile – in effect a “chemical fingerprint” - that was used for retrospective determination of the production method of a specific CWA. The study in paper I was done on the nerve agent Russian VX, S-[2-(diethylamino)ethyl] O-isobutyl metylphosphonothioate, while paper II and III focused on sulfur mustard, bis(2-chloroethyl)sulfide. This thesis discusses the study set up, the choice of analytical methods, methods for data processing and the manner in which classification methods have been employed. The studies shows that the classification models could clearly separate the six production routes used in paper I and the five routes used in paper II. In paper III, a novel non-targeted approach in combination with high-resolution mass spectrometry allowed detection of additional low-concentration compounds in the sulfur mustard samples. This method produced data with sufficient information for classifying samples according to the production method of the precursor thiodiglycol (TDG).   

The performance of the classification models was successfully validated with test set samples. All test set samples were correctly assigned in paper I and paper II. The classification of TDG in paper III was more demanding, but still as much as 56-89% of the test set samples were correctly assigned. In addition to the established classification models, compounds with importance for route differentiation were identified, which gave enhanced information on the chemicals formed during the employed synthesis conditions. Their stability has also been investigated, and the results showed that the majority of the chemical attribution signatures (CASs) were stable at room temperature.  

The fourth study in this thesis (paper IV) is an international inter-laboratory comparison jointly conducted by eight defence research laboratories based in Europe, North America, Asia and Australia respectively. All participating laboratories analysed the same samples prepared at the Swedish Defence Research Agency (FOI). The impurity profiles in nerve agent precursor metylphosphonyl dichloride samples were compared by a gas chromatography mass spectrometry (GC/MS) method using a retention index to facilitate data comparison. Retention indices of 16 CASs were calculated and compared, and this showed that the between-laboratory variation was low. This work is a first step towards a harmonised laboratory method for the profiling of CWA samples. The methods developed in this thesis will enhance accurate source attribution of CWAs and could potentially be used when alleged use of a CWA is being investigated.