Automatic detection and classification of ignitable liquids from GC–MS data of casework samples in forensic fire-debris analysis

Abstract

Investigation of fire scenes, in order to determine whether arson was involved, presents major challenges to both the forensic investigator who collects fire debris samples and the forensic scientist who analyzes them. A gas chromatography – mass spectrometry (GC–MS) chromatogram from the sample is compared to chromatograms from known ignitable liquid (IL) standards, and the analyst determines whether there are sufficient similarities to justify a positive detection (the fire debris contains IL residue) and classification (which IL it is). We have employed the chemometric software OpenChrom to accomplish real-time automatic detection and classification of 374 fire debris samples from casework fire scenes. In the tested samples, full agreement was achieved in IL detection between the software and the forensic experts: in no case did the software classify an IL-containing sample as “no IL” (false negative) or a “no IL” sample as containing IL (false positive). Moreover, of the 236 samples designated by the experts as containing IL, all were similarly classified by the software either as “gasoline” or “petroleum distillates” (diesel fuel, lighter fluid etc.). This study demonstrates the potential of an off-the-shelf fire debris automatic IL-detection tool in a forensic laboratory setting using hundreds of casework samples. Employment of this tool may help increase the efficiency and objectivity of the forensic laboratory.