Performance evaluation of activated carbon cloth for fire debris analysis with artificial neural network approach

Forensic fire investigates the origin and cause of the fire. Detection and identification of ignitable liquid (IL) residue in fire debris may provide the vital clue of the fire cause, especially important to prove incendiary fires. The scope of this study is narrowed to the usage of activated carbon cloth (ACC) as the adsorbent material by passive headspace diffusion as the extraction technique of IL residue. This study investigates the detection of selected target compounds that represent wide IL residue range from the lightest compound of n-hexane (C6) to the heavier compound of eicosane (C20) at different extraction parameters especially the temperature setting (60 °C to 120 °C) and exposure period (2 hours to 24 hours). Data sets from the chromatographic pattern vary significantly with different parameters were chosen. Computational modelling of artificial neural network (ANN) based on the pattern was developed and utilised to evaluate the extraction performance of ACC for optimisation purposed. The resolution of chromatographic behaviour of 14 selected target compounds that represent the ignitable liquid was used as input for the ANN model. The ANN display a response model of (2:2-17-14:14) allows the optimum condition with the practical setting to be 4 hours at 100 °C for urgent sampling while 18 hours at 80 °C is intended for overnight sampling. Selected optimum condition and practical settings for effective extraction of volatile compounds are important knowledge to facilitate busy laboratory operation as well as the identification and interpretation of complex fire debris samples. Thus, the finding of this research has a relevant implication for the forensic analyst who performed fire evidence investigation