Quantum chemical computation-based strategy for alternating least squares initialization in multivariate curve resolution analysis of spectral-pH data

The main motivation of this work is to provide initial estimates for the initialization of the iterative optimization within the multivariate curve resolution - alternating least squares (MCR-ALS) algorithm for the decomposition of second-order data. It is demonstrated that the combination of quantum chemical calculations with chemometrics constitutes a novel strategy for the ALS initialization in the MCR resolution of pH-modulated chemical data. In this work, the second-order data arise from acid-base experiments of p-nitrophenol (pNP) done under a pH-gradient generated by an automated flow injection (FI) system monitored by UV-vis spectroscopy. The absorption spectra of the species involved in the chemical equilibrium were simulated by means of time-dependent density functional theory (TD-DFT) methods and were utilized to start the ALS optimization. The new approach based on the Tamm-Dancoff-approximation (TDA) CAM-B3LYP method is recommended to obtain the simulated spectra to initialize MCR-ALS, as an alternative to the routinely methods used to generate initial estimates.