Development of a tiered analytical method for forensic investigation of mixed lubricating oil samples

Oil spill forensic investigations are often challenging due to many confounding variables such as sample weathering, oil composition complexities, and the quality or quantity of collected materials, but the difficulty is further compounded when dealing with mixed oils. In this case, well-established oil fingerprinting techniques become inadequate, including gas chromatography-flame ionization detection (GC/FID) and gas chromatography-mass spectrometry (GC/MS) diagnostic ratio analysis. In dealing with mixtures of highly refined lubricating (lube) oils, GC/FID analysis often yields inconclusive results, while diagnostic ratio analysis can be compromised by missing or low response biomarker compounds. The present study explored the feasibility of addressing the challenges of mixed lube oil analysis through a multi-tiered analytical approach. This analysis supplemented traditional GC/FID and GC/MS diagnostic ratio analyses with multivariate statistics to rapidly screen large data sets. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) proved to be effective and intuitive qualitative methods for visualizing, differentiating, and characterizing four highly similar lube oil mixtures. Non-linear mixing patterns that were significant in the diagnostic ratio analysis were far less evident through LDA. Overall, these findings lay the groundwork for promising future study involving multivariate statistical approaches to complex mixed oil forensic cases.