Abstract

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.

摘要

由于样品风化、石油成分的复杂性以及收集材料的质量或数量等许多混杂变量，溢油法医调查通常具有挑战性，但在处理混合油时，难度会进一步加大。在这种情况下，成熟的油品指纹识别技术变得不够用，包括气相色谱-火焰离子化检测 (GC/FID) 和气相色谱-质谱 (GC/MS) 诊断比率分析。在处理高度精炼润滑油（润滑油）的混合物时，GC/FID 分析通常会产生不确定的结果，而诊断比率分析可能会因缺失或低响应生物标记化合物而受到影响。本研究探讨了通过多层分析方法解决混合润滑油分析挑战的可行性。该分析用多元统计补充了传统的 GC/FID 和 GC/MS 诊断比率分析，以快速筛选大型数据集。主成分分析 (PCA) 和线性判别分析 (LDA) 被证明是一种有效且直观的定性方法，可用于可视化、区分和表征四种高度相似的润滑油混合物。在诊断比率分析中重要的非线性混合模式通过 LDA 远不明显。总的来说，这些发现为有前途的未来研究奠定了基础，该研究涉及复杂混合油法医案件的多元统计方法。主成分分析 (PCA) 和线性判别分析 (LDA) 被证明是一种有效且直观的定性方法，可用于可视化、区分和表征四种高度相似的润滑油混合物。在诊断比率分析中重要的非线性混合模式通过 LDA 远不明显。总的来说，这些发现为有前途的未来研究奠定了基础，该研究涉及复杂混合油法医案件的多元统计方法。主成分分析 (PCA) 和线性判别分析 (LDA) 被证明是一种有效且直观的定性方法，可用于可视化、区分和表征四种高度相似的润滑油混合物。在诊断比率分析中重要的非线性混合模式通过 LDA 远不明显。总的来说，这些发现为有前途的未来研究奠定了基础，该研究涉及复杂混合油法医案件的多元统计方法。