Raman spectroscopy of automotive paints: Forensic analysis of variability and spectral quality

摘要

Raman spectroscopy was evaluated as a forensic technique for the analysis of automotive paints with the purpose to verify the measurement variability as well as to study the spectral quality by means of parameters and conditions. The objective lens, accumulation number, and laser (power, mode, and wavelength) were observed to significantly affect the spectral quality and thus should be taken into consideration during the development of a forensic database. On the other hand, exposure time, Raman shift range, and baseline correction had less influence on the features of the obtained spectra. Interestingly, samples collected from the bumper coating composed of a plastic substrate in comparison with fragments obtained from other parts of the same vehicle containing metallic substrate showed different absorption band patterns. In addition, less variability of data acquired directly onto the paint chip's surface has demonstrated that direct analysis should be preferred during spectra library development as opposed to cross-section measurements. Sample homogeneity and distinction power of the technique combined with Principal Components Analysis were also investigated. The preliminary results showed that although Raman spectroscopy is extremely accurate in the identification of vehicles, it is necessary to pay particular attention to the spectral variability in order to avoid false database matching and misleading of forensic investigations. Therefore, we suggest that separated spectra library shall be developed for each laser wavelength as well as for each sample substrate. Further studies with a higher number of paint chips will enable the establishment of prediction models aimed to identify unknown samples involved in hit-and-run cases.