Fourier-transform near infrared spectroscopy and chemometrics for discrimination of cold-pressed oils and determination of their chemical parameters

Abstract

Principal component analysis was performed to discriminate commercial cold-pressed cosmetic oils based on their Fourier-transform near infrared spectroscopy spectra and chemical parameters such as the composition of fatty acids, content of tocopherols, total carotenoids, polyphenols, and chlorophylls, as well as calculated oxidizability and iodine values. It was found that the oils analyzed differed significantly in the chemical composition. The level of total unsaturated fatty acids ranged from 74.0 to 93.4%. The content of carotenoids in oils ranged from 3.1 to 197.1 mg/kg, total chlorophylls from 0.04 to 46.3 mg/kg, and total phenolics from 36 to 596 mg/kg. The oils tested differed also in the content of tocopherols (from 11 to 3836 mg/kg). Principal component analysis based on Fourier-transform near infrared spectroscopy spectra revealed a different pattern of discrimination of the oils compared to principal component analysis based on the chemical parameters. However, using partial least squares regression, good correlations were found between Fourier-transform near infrared spectroscopy spectra and the contribution of linoleic acid (18:2), monounsaturated fatty acids, polyunsaturated fatty acids, unsaturated fatty acids, calculated oxidizability, or calculated iodine values. Good models with coefficients of determination not lower than 0.989 and with low root-mean-square error for cross-validation were obtained when the range from 4800 to 4500 cm−1was applied. Values of residual predictive deviation for these models were higher than 3.0 indicating very good prediction accuracy. The models obtained were successfully used to predict these parameters for new selected oils.