Exploiting chemometrics for screening analysis by near-infrared spectroscopy without the support of reference analytical methods: Application to bulk sweeteners

Abstract

A novel analytical approach based on near-infrared (NIR) Fourier transform miniaturized spectrophotometer and chemometrics is proposed for screening commercial products. A guide for analysis by the proposed approach is provided and validated. The approach is more environmentally friendly and cost-effective, reducing the cost of implementation by avoiding the need for large data sets of samples analyzed by costly and laborious reference analytical methods. The combination of multivariate techniques (principal components analysis, PCA, alternating least-squares multivariate curve resolution, MCR-ALS, and partial least squares regression, PLS) with visual spectra inspection and spectral correlation yielded consistent qualitative and quantitative results. As a case study, the approach was applied to commercial bulk sweeteners commonly used in low carbohydrate diet (erythritol and xylitol) and non-conformities detection was feasible using few reference samples which were not analyzed by conventional analytical methods. The effect of particle size of commercial products and their mixtures on the NIR spectrum as well as the non-linear spectral response observed for binary mixtures of sweeteners are discussed. Mixtures of sweeteners, simulating product adulterations, were analyzed using PCA, non-linear MCR-ALS, and PLS, to evidence non-conformity of commercial products and to estimate the level of adulteration.