Accelerant classification by gas chromatography/mass spectrometry and multivariate pattern recognition Original Research Article

Petroleum-based accelerants are commonly associated with arson-related fire. Detection and correct classification of accelerant is crucial to arson investigation. The application of gas chromatography/mass spectrometry (GC/MS) and multivariate pattern recognition techniques for petroleum-based accelerant detection and classification is presented. The method feasibility and matrix effects on accelerant classification were studied using principal component analysis (PCA). A soft independent model classification analogy (SIMCA) model was then developed to evaluate evaporation, sample size, and sample charring. Depending on the sample class, the detection limits of correct classification were in the range of 5–50 μl. The detection limits of closest classification could be as low as 1.25 μl in the charred samples.