Reaction Analysis and Process Optimization with Online Infrared Data Based on Kinetic Modeling and Partial Least Squares Quantitation

Abstract

In-situ Fourier transform infrared (FT-IR) spectroscopy has been recognized as an important technology for online monitoring of chemical reactions. However, analysis of the real-time IR data for identification and quantification of uncertain reactants or intermediates is often ambiguous and difficult. Here, we propose an analysis algorithm based on reaction kinetic modeling and the chemometric method of partial least squares (PLS) to comprehensively and quantitatively study reaction processes. Concentration profiles and apparent kinetic parameters can be simultaneously calculated from the spectral data, without the demand of complicated analysis on characteristic absorbance peaks or tedious sampling efforts for multivariate modeling. Paal–Knorr reactions and glyoxylic acid synthesis reactions were selected as typical reactions to validate the algorithm. A lack of fit of the Paal–Knorr reaction spectra was less than 2.5% at various conditions, and the absolute errors between the predicted values and HPLC measurement of glyoxylic acid synthesis were less than 6% during the reaction process. Moreover, the reaction kinetic models extracted from FT-IR data were used to simulate reaction processes and optimize the conditions in order to maximize product yields, which proved that this analysis method could be used for process optimization.