Seed oil detection in extra virgin olive oil by differential scanning calorimetry

• Published in: Journal of thermal analysis and calorimetry 2023-07, Vol.148 (14), p.6833-6843
• Main Authors: Pereira, Lucas H., Pereira, Juliana, Garcia, Jerusa S., Trevisan, Marcello G.
• Format: Article
• Language: English
• Subjects: Adulterants; Aluminum; Analytical Chemistry; Atmospheric models; Binary mixtures; Calibration; Calorimetry; Chemistry; Chemistry and Materials Science; Cooling; Corn; Crystallization; Errors; Heating rate; Inorganic Chemistry; Measurement Science and Instrumentation; Multivariate analysis; Olive oil; Physical Chemistry; Polymer Sciences; Soybean; Soybean oil; Soybeans; Sunflower oil; Vegetable oils
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-023-12178-1

The aim of this work was to develop and apply an analytical method to detect adulterants in Extra Virgin Olive Oil (EVOO) using DSC with controlled cooling and partial least squares (PLS). For such a purpose, several binary mixtures were prepared using sunflower, corn and soybean oils as adulterants. These materials were weighed and prepared in proportions ranging from 5 to 95% (m/m). The samples were submitted to DSC analysis within the following parameters: dynamic atmosphere of N 2 (50 mL.min −1 ); temperatures ranging from 30 to − 80 °C and from − 80 to 30 °C at a cooling/heating rate of 5 °C min −1 ; about 8 mg of sample into a sealed aluminum crucible. The PLS models were constructed based on DSC data, and curves were normalized by the respective initial masses of samples so as to eliminate influence on mass variation. Data on sample were pre-processed, normalized by their respective standard variation and mean centered. Multivariate analysis results were also compared with the univariate calibration using T onset data (referring to the oil crystallization event). The PLS models were successfully constructed to quantify the adulteration level. Calibration errors of 2.34, 2.61 and 4.02% m/m were found for sunflower, corn and soybean oils, respectively, with 3–4 latent variables. Prediction errors of sunflower, corn and soybean oils were, respectively, 3.36, 5.62 and 9.55% m/m. Therefore, the univariate model demonstrates lower calibration errors of 0.59, 0.88 and 0.66% m/m for sunflower, corn and soybean oils, respectively, but using a smaller concentration range (30 to 80% m/m for sunflower oil and 10 to 65% m/m for corn and soybean oil). Furthermore, a DSC-based strategy is quite successful in detecting seed oils in EVOO.