Tracking the origin of fuel oil and petrol in explosive mixtures of ANFO

Abstract

This research describes the analysis of petrol and diesel from different brands aiming at the identification and classification of ANFO explosives based on gas chromatography − mass spectrometry (GC − MS) and Fourier transform infrared spectroscopy (FTIR) using as an aid principal component analysis (PCA).

First, 91 petrol and diesel samples sold in the city of Lincoln (UK) during four seasons (winter, spring, summer, autumn) were analysed by GC − MS and FTIR in order to identify chemical composition and additives with the aim of obtaining a set of potential markers that will allow forensic scientists to identify fuels from different brands. The results demonstrated that each petrol and diesel samples chemically contains unique fuel compositions. MTBE and ETBE were identified as petrol additives that can be used to aid in the separation. The inconsistency in the amount of some petrol compounds, such as isooctane, was also observed and used to establish differences among petrol samples. In diesel, the distribution of FAME contents showed the effect of seasonal variation as these were found less used in winter. The selection of a reduced number of key fuel compounds and additives was also shown to be sufficient to allow a classification among the different fuel samples with a high accuracy (from 80–100%) using PCA-LDA.

Second, the classification of pre-blast ANFO explosives are excellently achieved using GC − MS and FTIR. Diesel can be extracted from ANFO mixtures and analysed by GC − MS. The chromatographic pattern of each extraction sample exhibited the characteristic peaks of FAME contents in some samples. PCA is effectively used to classify all diesel extracts. The selection in the specific diesel components considerably showed an effect to the classification among diesel samples and significantly provides different degree of discrimination of the ANFO samples. In addition, the use of mid − infrared spectral region combined with PCA and leave-one-out cross-validation is effectively able to discriminate different ANFO samples with high classification performance. Unfortunately, the determination and classification of ANFO from post blast samples was not successfully achieved.