Fast Methodology for Identification of Olive Oil Adulterated with a Mix of Different Vegetable Oils

• Published in: Food analytical methods 2019-01, Vol.12 (1), p.293-304
• Main Authors: Filoda, Paula Freitas, Fetter, Lucas Flores, Fornasier, Franccesca, Schneider, Rosana de Cassia de Souza, Helfer, Gilson Augusto, Tischer, Bruna, Teichmann, Aline, da Costa, Adilson Ben
• Format: Article
• Language: English
• Subjects: Analytical Chemistry; Calibration; Canola oil; Chemistry; Chemistry and Materials Science; Chemistry/Food Science; Chromatography; Corn; Fatty acids; Food Science; Fourier transforms; Identification methods; Infrared spectra; Infrared spectroscopy; Mathematical models; Microbiology; Oils & fats; Olive oil; Predictions; Regression analysis; Soybeans; Spectrum analysis; Vegetable oils; Vegetables
• ISSN: 1936-9751; 1936-976X
• DOI: 10.1007/s12161-018-1360-5

This paper investigated the application of Fourier transform infrared spectroscopy (FTIR) with partial least squares regression algorithms (PLS) and variable selection methods for the rapid identification of extra virgin olive oil (EVOO) adulterated with different vegetable oils. For this purpose, a unique calibration model was proposed for the identification and quantification of adulteration independent of the adulterating oil. Calibration models were developed for simultaneous determination of the concentration of oleic, linoleic and linolenic fatty acids. Robust models were also developed for quantification of the percentage of adulteration in the samples, independent of the adulterating oil. The calibration set consisted of 68 adulterated EVOO samples, prepared by the addition of vegetable oils (soybean, sunflower, corn, and canola oil) at different levels (1 to 80%, v/v ), and ten commercial samples of EVOO were used to validate the models. The mid-infrared spectra were recorded in the wave number range 3200 to 650 cm −1 for all samples. Chromatographic analysis was performed to determine the fatty acid profile of the samples from the calibration and prediction sets. PLS models were developed and different strategies were investigated during the preprocessing of the IR spectra. The results of the prediction were compared with those obtained from the conventional analysis, and RMSEP values of 4.44, 1.92, and 0.62% were obtained for oleic, linoleic, and linolenic acids, respectively. The proposed methodology allowed us to quantify the acids simultaneously in the samples and also demonstrated a good predictive power to identify the adulterated samples.