Analysis Of Molecular Features Of Eight Lignites With Chemometrics

Baiyinhua lignite?BYH?,Dongming lignite?DM?,Dayan lignite?DY?,Hami lignite?HM?,Shengli lignite?SL?,Xiheishan lignite?XHS?,Xiaolongtan lignite?XLT?and Zhaotong lignite?ZT?were selected as the coal samples,which were sequentially dissolved at 300 ^oC by cyclohexane,methanol and ethanol to afford soluble portions?SPs?.SPs were analyzed by a gas chromatograthy/mass spectrometry?GC/MS?and an orbitrap mass spectrometry?orbitrap-MS?.Chemometric methods were introduced to the analysis of MS data for in-depth understanding of molecular structures of organic species in coals based on the statistical work applied in petroleomics and metabolomics.From the GC/MS data,alkanes,arenes and arenols can be dissolved out efficiently with cyclohexane.Methanol is an efficient solvent to obtain arenols,ethers and ketones.In ethanol SPs,high relative abundances of esters,arenols and alcohols were detected.Fold change and t-test analysis were performed on two clusters of cyclohexane,methanol and ethanol SPs except three outlier samples?XLT-CYH,DY-MT and XHS-EA?separately.The information of the inter-cluster differential compounds was obtained.There were 59 differential compounds including alkanes and arenes found between two clusters of cyclohexane SPs.Two clusters of methanol SPs had 39 differential compounds including arenols,ketones and arenes.There were66 differential compounds including arenols,esters,alcohols,and ethers found between two clusters of ethanol SPs.Partial least squares discriminant analysis?PLS-DA?was performed to establish classification prediction models and the prediction ability of the models was verified with the known samples.The models had predictive ability although models were built with fewer samples.The accuracy of model prediction can be improved by increasing training samples.Atmospheric pressure chemical ionization?APCI?-Orbitrap MS with high resolution can detect more compounds compared with GC/MS.The analytical results of SPs indicate that more N₁,O₁N₁,O₂N₁,O₁,O₂ and HC class species are detected by APCI in positive mode.O₁,O₂,N₁,O₁N₁ and a small amount of HC class species were rich in cyclohexane SPs.In most of the methanol and ethanol SPs,higher relative abundances of O₁N₁,O₂N₁ and N₁ class species were detected.Chemometric methods like longest distance method and Euclidean distance of HCA?hierarchical clustering analysis?were introduced to figure out the difference between 24SPs and Nbclust function were used to obtain the optimal clustering number of 7?A,B,C,D,E,F and G?.The differential compounds between A and B,C and D,E and F as well as G were obtained by using fold change and t-test analysis,respectively.For example,in cluster D,the molecular weight distribution,Double bond equivalent?DBE?vs the distributions of carbon number,and relative intensity vs the distribution of DBE were employed to explore differences in SPs of the same cluster and get the differences and similarities of these SPs based on O₁N₁,N₁,O₂N₁ and O₂ class species,which were the main class species of cluster D.Such results provide a complete,fast and effective analytical process for the analysis of the high resolution MS data of coal samples.There are 26 figures,21 tables,1 attached figure,10 attached tables and 90 references in this thesis.