PNNL

Abstract

The illicit chemical alteration of petroleum fuels is of scientific interest, particularly to regulatory agencies which set fuel specifications, or excises based on those specifications. One type of alteration is the reaction of diesel fuel with concentrated sulfuric acid. Such reactions are known to subtly alter the chemical composition of the fuel, particularly the aromatic species native to the fuel. Comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC × GC–TOFMS) is ideally suited for the analysis of diesel fuel, but may provide the analyst with an overwhelming amount of data, particularly in sample-class comparison experiments comprised of many samples. The tile-based Fisher-ratio (F-ratio) method reduces the abundance of data in a GC × GC–TOFMS experiment to only the peaks which significantly distinguish the unaltered and acid altered sample classes. Three samples of diesel fuel from different filling stations were each altered to discover chemical features, i.e., analyte peaks, which were consistently changed by the acid reaction. Using different fuels prioritizes the discovery of features which are likely to be robust to the variation present between fuel samples and which will consequently be useful in determining whether an unknown sample has been acid altered. The subsequent analysis confirmed that aromatic species are removed by the acid alteration, with the degree of removal consistent with predicted reactivity toward electrophilic aromatic sulfonation. Additionally, we observed that alkenes and alkynes were also removed from the fuel, and that sulfur dioxide or compounds that degrade to sulfur dioxide are generated by the acid alteration. In addition to applying the previously reported tile-based F-ratio method, this report also expands null distribution analysis to algorithmically determine an F-ratio threshold to confidently select only the features which are sufficiently class-distinguishing. When applied to the acid alteration of diesel fuel, the suggested per-hit F-ratio threshold was 12.4, which is predicted to maintain the false discovery rate (FDR) below 0.1%. Using this F-ratio threshold, 107 of the 3362 preliminary hits were deemed significantly changing due to the acid alteration, with the number of false positives estimated to be about 3.