A novel and efficient chemometric approach to identifying oil families by saturate biomarker data and FTIR spectroscopy of asphaltene subfractions

Abstract The genetic classification is commonly performed by the utilization of a few interpretive biomarker ratios (e.g., 14 peak ratios) in chemometric methods. In this paper, a novel workflow with higher efficiency in revealing subtle differences between samples was suggested to solve the limitations of traditional methods. The peak areas of selected source related biomarker compounds were employed to generate 342 peak ratios. The use of higher peak ratios in the comparisons provides higher sensitivity regarding dissimilarities among studied samples. These peak ratios were utilized as an input in PCA and HCA techniques. Also, a new similarity coefficient measurement was suggested for the HCA technique. In this procedure, the moving window correlation coefficient technique was implemented on generated data series. The dissimilarity matrix for each pair of samples was constructed by the calculated minimum correlation coefficient in all moving windows. A complete linkage algorithm was then exerted on the dissimilarity matrix to relate different samples and define oil families. This new procedure provides higher efficiency regarding commonly used procedures in the PCA and HCA techniques for disclosing dissimilarities between samples. Furthermore, structural characteristics of asphaltenes due to their similarity with the precursor kerogen were applied to complement GC-MS data. Asphaltenes were fractionated into three subfractions for comprehensive comparison among studied samples. The proposed technique was carried out on the 18 oil samples from the Dezful Embayment, SW Iran. The studied oils comprise three distinct oil families, including group I (Rag-e Sefid and Shadegan samples), group II (Ahvaz samples), and group III (Qale Nar samples). Both FTIR and GC-MS corroborate these three oil families.