Weighted parallel factor analysis for calibration of HPLC-UV/Vis spectrometers in the presence of Beer's law deviations

An extension of the parallel factor analysis (PARAFAC) methodology is presented to allow accurate and reliable quantitative and qualitative analysis of nonlinear data collected from hyphenated instrumentation. The weighted PARAFAC method is applied to high-performance liquid chromatography-ultraviolet/visible (HPLC-UV/Vis) diode array spectrometry analysis. It is demonstrated that this method improves the quantitative errors when spectroscopic nonlinearities from solvent-solute interactions or detector saturation are introduced. As much as 50% improvements in the root mean squared errors of estimation are realized for test samples. This weighted PARAFAC algorithm implicitly treats nonlinear data as missing values. A method requiring no a priori information is presented, that facilitates determination of the nonlinear regions and optimal application of the weighted PARAFAC algorithm. (C) 1999 Elsevier Science B.V. All rights reserved.