The ultimate goal of this project was to assist fire investigations by providing a stronger scientific basis for the observed weathering of ignitable liquid residues in casework samples.
In fire debris analysis, ignitable liquid residues are commonly identified using gas chromatography-mass spectrometry (GC-MS). The detection and identification of an ignitable liquid can help arson investigators determine whether a fire was intentional or accidental. To assist with the identification of ignitable liquid residues, chromatograms of questioned samples are typically compared to those of known ignitable liquids that have been weathered (evaporated) to different extents. Practitioners typically perform such weathering at room temperature and to a limited number of extents of weathering, so their database of weathered residues is likely to deviate markedly from casework residues. The current experiments were designed to elucidate the effects of three different weathering factors on the distribution of residues: 1) the temperature at which weathering occurs, and 2) the porosity of the substrate, and 3) the penetration depth of an ignitable liquid in the substrate. A nine-component synthetic gasoline simulant was experimentally evaporated to different extents at different temperatures on four different substrates, including cotton fabric, nylon carpet, plywood and pine wood. Additional experiments were conducted in which 30-s or 30-min delays were implemented between spiking the gasoline simulant on each substrate and initiating the weathering at 210℃. The weathered residues in the different substrates were collected using solid-liquid extraction in pentane and analyzed using gas chromatography-mass spectrometry (GC-MS). The distribution of experimentally weathered residues were compared to a thermodynamic model that was previously developed by our group. In the absence of a substrate, the model provides accurate predictions of the relative peak areas for weathering conducted from 30-210℃ and up to 95% weathering. For example, the root mean squared error of predictions (RMSEPs) of the model was on the order of 2% for N=180 predicted peak areas at 210℃. Although the weathering temperature significantly alters the relative distribution of volatiles remaining in the weathered residues, the model accounts for the effect of temperature and the accuracy remains quite constant at ~2% between 30-210℃. The presence of a relatively non-porous substrate like cotton fabric had a small effect on the accuracy of the model; the RMSEPs increased to ~2.9% for N=117 predictions; however, in the presence of more porous substrates, like pine wood and plywood, the RMSEPs increased to a low of 3.7% for pine wood with no delay between spiking and weathering to 7.8% for plywood with a 30-minute delay. The prediction errors indicate that porous substrates like untreated wood prevent the volatile components from evaporating at their normal rates, and that the relative distribution of weathered residues hardly changes beyond 50% weathering. The results indicate that the entrapment of ignitable liquids in porous substrates make them appear less weathered than one would expect for a thermodynamic model. (Publisher Abstract}