Advanced authenticity screening of commercial berry seed oils using full FTIR spectra and DSC curves coupled with chemometrics

Abstract

The feasibility of Differential Scanning Calorimetry (DSC) and Fourier Transform Infrared (FTIR) spectroscopy coupled with Partial Least Square Discriminant Analysis (PLS-DA) for the authenticity screening of blackcurrant-, strawberry-, and raspberry seed oils was evaluated. Authentic DSC and FTIR fingerprints of berry seed oils were reported and used as training data, reflecting their natural variability. In DSC analysis, full curves of melting phase transition were utilised instead of individual DSC parameters (peak temperature, peak height, enthalpy). Each berry seed oil was characterised by distinctive endothermic and exothermic peaks. A FTIR model was built based on the absorbance intensity at three spectral ranges of 3550–3200 cm −1 , 3010–2854 cm −1 , and 1720–720 cm −1 . With supervised learning and personalised data processing, the classification performance of both DSC and FTIR were improved. During external validation using commercial oils, the DSC models demonstrated good predictive performance in classifying blackcurrant and strawberry oils, while FTIR models were favoured for predicting the class of blackcurrant and raspberry oils. This work provides industry practitioners with practical insights into processing the thermal and spectral fingerprint data for the authenticity screening of berry seed oils. • Authenticity study on the 106 samples of cold-pressed and commercial berry seed oils. • Full DSC and FTIR spectra analysed by PLS-DA instead of individual parameters. • Multidata from genuine cold-pressed oils used as training data to build the model. • Authentication of commercial oils based on the model built from DSC, FTIR and composition. • DSC model excelled for blackcurrant, strawberry oils, while FTIR for blackcurrant, raspberry.

Topics & Concepts