Forensic discrimination of automotive paint samples using pyrolysis-gas chromatography-mass spectrometry with multivariate statistics

Analytical pyrolysis-gas chromatography (Py-GC) has been a standard method for the forensic analysis of automotive paint for a number of decades. Automotive paints are often identified by visual comparison of pyrograms for peak presence and intensities; however, such analyses can be subjective and time consuming. A preliminary investigation based on Py-GC-mass spectrometric analysis of 100 automobile paint samples of five different colors is presented. Designed experiments are employed to select pyrolysis conditions for adequate discrimination. Pattern recognition techniques including principal component analysis and canonical variates analysis are used to visualize clustering of pyrograms to validate comparisons between different automotive paint pyrograms. These methods have the potential to ease the interpretation task for data sets involving a large number of comparisons.