Chemometric Analysis of First Order Chemical Data

This thesis considers issues related to the multivariate analysis and calibration of chemical data collected in a vector format (one–way data). The work is designed to facilitate the task of the analytical chemist, focusing on methods of maximizing the amount of information that can be obtained by multivariate modeling from full sets of data generated by instruments or experiments. It is shown that more complex problems can be solved using a multivariate modeling methodology than by using the traditional univariate approach. The thesis is based on five papers that address different aspects of the multivariate analysis of one-way data, organized in a two-way matrix, with the aid of various chemometric techniques. The methodologies used cover variance analysis, predictive modeling, compression and alignment of data expressed as responses from various types of analytical instruments. The instrumental techniques used to acquire the chemical data considered in the thesis include Near Infrared (NIR), Ultra Violet-Visible (UV-VIS) and Nuclear Magnetic Resonance (NMR) spectrometry. Other types of data (e.g. FT-IR and GC signals) are also used for testing different solution spaces to different problems that has been encountered. Paper I. In the work described in Paper I, NIR reflectance data and multivariate calibration are used for building calibration models predicting yield, kappa number, Klason lignin, glucose, xylose and uronic acid contents in birch chips sampled during controlled Kraft digestion. The combination of NIR reflectance and multivariate calibration models predicts the descriptors well. Paper II. In the work described in Paper II, UV-VIS spectrometry and multivariate calibration are used to determine the nitrate ( 3 NO ) concentration in municipal wastewater. The method is based on scanned spectra of calibration samples measured as raw, unfiltered wastewater. The proposed method has a working range of 0.5-13.7 mg/l with a relative error of 3.4%. The method can also be used for the determination of total phosphorous, total nitrogen, ammonium nitrogen and iron.