Functional data analysis, a new approach to aligning three-way liquid chromatographic with fluorescence detection data

Abstract

Functional data analysis (FDA) arises as a promissory auxiliary methodology designed to help the analytical chemists, especially chemometricians. However, although the innovative ideas of this approach have barely spread into the chemical research field. In this work, a novel approach for aligning three-way chromatographic-spectral data based on FDA methodology is proposed. Unlike most of the available algorithms, this novel method allows performing data alignment when the test data matrix contains unexpected chemical interferences. Simulated and experimental analytical systems composed of calibrated analytes and potential interferents in the test samples are studied. The experimental system corresponds to the determination of the four polycyclic aromatic hydrocarbons (benzo[a]pyrene, dibenzo[a,h]anthracene, benzo[b]fluoranthene and benzo [k]fluoranthene) in the presence of benzo[j]fluoranthene and benzo[e]pyrene as potential interferences. These compounds are priority pollutants; hence, its reliable quantification in environmental samples is an analytical challenge. The method consists in decomposing the three-way data in each sensor mode and in making a functional alignment of the pure vectors obtained from the individual chromatograms of each analyte and interferent. The functional preprocessing step enables the analysis of the data set with second-order multivariate calibration algorithms, such as PARAFAC. The results illustrate that the proposed method restores the trilinearity of the three-way data, thus being able to successfully quantify the analytes in the presence of the interferences, that is, exploiting the second-order advantage. MCR-ALS is also applied to both simulated and experimental data to evaluate the performance of the PARAFAC second-order calibration model proposed. The performance of the PARAFAC and MCR-ALS models are compared and discussed.