Researches On Chemometric Methods And Their Applications For The Analysis Of Two-Dimensional Data Arrays

The scientific analysis of chemical data sets is to discover the hidden relationships between the chemical properties and their constitutions. The recent development of high-dimensional chemometrics resolution methods provides a more powerful tool for us to explore the complex chemical systems as opposed to conventional analytical methods. The primary purpose of this work is to develop new methodology in herbal fingerprinting research and spectral analysis and to test the adequacy of chemometrics resolution techniques and multivariate analysis in areas related to the chemical sciences. Hence, a fair part of the thesis describes methodological developments related to herbal fingerprinting and chemometrics methods, as well as their application to chemical systems. In all, this thesis covers four aspects as below.1. Fingerprinting research of herbal medicines (chapter 2 to chapter 4). In herbal fingerprints, there are commonly two main encountered disturbances: 1) disturbances introduced by the measurement device itself and 2) disturbances due to the diversity of the investigated samples. Concerning these two disturbances, spectral correlative chromatography is first proposed to determine the presence or absence of some spectral correlative peaks among a set of spectrochromatograms from herbal medicines under some experimental conditions in chapter 2. In chapter 3, local least squares regression model issuitably constructed for modeling fingerprint data with factors that shift. This is relevant for handling retention time shifts of some interested peaks in chromatograms of various samples. It facilitates the subsequent quality evaluation of herbal fingerprints. In chapter 4, a multivariate calibration model inspired by secured principal component regression is described for detecting unexpected chromatographic features among herbal fingerprints. Meanwhile, it presents a viable approach to qualitatively evaluate diverse herbal objects with a regular class of fingerprints.2. Hyphenated chromatographic resolution methods (chapters 5 and 8). Chemometrics resolution methods, such as subwindow factor analysis, heuristic evolving latent projections and so on, are utilized to well resolve the acidic fraction of petroleum ether extract of Radix Rehmanniae Preparata and non-hydrocarbon fractions of crude oil into pure spectra and chromatographic profiles of individual components, which makes the qualitative identification and relative quantitative analysis of each compound more reliable.3. Near-infrared spectroscopic analysis of human serum albumin, y-globulin, and glucose in buffer solutions (chapter 6). Partial least squares regression, secured principal component regression and modified secured principal component regression are performed for the quantitative determination of human serum albumin, y-globulin, and glucose in phosphate buffer solutions by near-infrared spectroscopy. The discussion focuses on the selection of calibration models relating the multicomponent outputs ofspectrometers to the chemical concentrations, the extraction of informative spectral features, the correction of measurement background and error disturbances, and the determination of critical model parameters, etc.4. Dynamic investigation of isotactic polystyrene during the cold crystallization process (chapter 7). The structural evolutions of isotactic polystyrene are explored. Fourier transform-infrared spectral variations are due to the formation of ordered local chains, the conformational changes in helical structures and the ordering of phenyl rings. Chemometrics multivariate curve resolution technique is promising to analyze and visualize spectral data and to integrate them with other information, making spectral intensity variations more amenable to interpretation to explore molecular dynamics of polymers.