Unsupervised Learning Algorithm for the Molecular Characteristics of Organic Matter in Low-Rank Coals Based on Gc×Gc/Tof-Ms Data

Abstract

Two low-rank coals were sequentially dissolved in cyclohexane, acetone, and methanol at 300 °C, and a comprehensive two-dimensional gas chromatography/time-of-flight mass spectrometry (GC×GC/TOF-MS) was used to analyze the 6 thermal dissolution (TD) extracts to obtain the molecular characteristics of soluble organic matter. GC×GC/TOF-MS improved separating ability, overcame co-elution, and comprehensively separated isomers with similar structures. Low-polar compounds like aliphatic hydrocarbons and arenes tend to be extracted by cyclohexane. Acetone can extract monocyclic aromatic hydrocarbons (MAHs) with methyl, ethyl, propyl, and vinyl groups because it can break some of the long alkyl side chains and bridged bonds. Meanwhile, acetone can form N-H⋯O hydrogen bonds with nitrogen-containing organic compounds (NCOCs) to extract more NCOCs. Methanol has a high nucleophilicity and good hydrogen supply ability, which can destroy the ArCH2OArR structural unit in coal molecules and generate the dominant alkyl substituted phenols. Additionally, a series of species including some isomers of tricyclic aromatic hydrocarbons (TAHs) and alkyl phenols in the SPs were only identified by GC×GC/TOF-MS. Two chemometric methods, principal component analysis (PCA) and hierarchical clustering analysis (HCA) were applied to mine effective molecular information from the big MS data. The score plot and loading plot of PCA revealed the differences among the seven hydrocarbon categories in the six TD extracts. Meanwhile, the distribution and composition characteristics between the six TD extracts and the 23 types of compounds are also shown in the tree and heat maps of HCA.

Keywords: Low-rank coal, Thermal dissolution, GC×GC/TOF-MS, Principal component analysis, Hierarchical cluster analysis