Analysis of automotive lubricating grease residue: A forensic investigation using ATR-FTIR spectroscopic and chemometric interpretation

Abstract

Automotive lubricating greases are widely utilized in automobiles, making them frequently encountered trace evidence in incidents involving a vehicle's collision with another vehicle, person, animal, or stationary object. Due to their high transferability, lubricating grease can act as valuable corroborative evidence during crime scene investigations. They can assist in establishing a link between the vehicle and the individual (victim or accused) to the crime scene, while also confirming contact between the vehicle and the victim. In this study, 21 brands of automotive lubricating grease samples were analyzed for their identification and differentiation, employing a rapid and non-destructive ATR-FTIR spectroscopic technique combined with chemometrics. ATR-FTIR spectra of all samples were visually examined and categorized into six groups based on their similarities and differences. After visual examination, the training dataset was subjected to chemometric analysis using PCA and SVM tools. PCA was employed to explore trends within the dataset, while SVM classified samples, achieving 97.62 % training accuracy and 88.09 % external validation accuracy. To ensure an unbiased validation of the SVM model, the training and validation datasets were comprised of a distinct set of spectra. A blind test validated the SVM model, resulting in 100 % prediction accuracy. Additionally, a study was performed to evaluate how substrate type, storage conditions, and storage duration could affect the linking of the substrate grease samples to their source. The findings revealed that the above-mentioned factors, particularly sunlight exposure and the washing of the substrate grease samples, significantly influence the SVM prediction accuracy.