Evaluation of butter oil adulteration with soybean oil by FT-MIR and FT-NIR spectroscopies and multivariate analyses

• Published in: Food science & technology 2019-06, Vol.107, p.1-8
• Main Authors: Pereira, Cristina Guimarães, Leite, Antonio Iranaldo Nunes, Andrade, Jonathan, Bell, Maria José Valenzuela, Anjos, Virgílio
• Format: Article
• Language: English
• Subjects: Butter oil; Chemometrics; Fraud; Infrared spectroscopy; Vegetable oil
• ISSN: 0023-6438; 1096-1127
• DOI: 10.1016/j.lwt.2019.02.072

The use of Fourier-transform mid-infrared (FT-MIR) and near-infrared (FT-NIR) spectroscopies associated with multivariate modeling was explored to obtain a rapid method that can detect adulteration of butter oil (BO) with soybean oil (SO). A total of 11 treatments comprising original BO, SO and their mixtures (10–90%, w/w) were analyzed. The substitution of BO by SO in the samples resulted in clear changes in the spectra intensities and positioning. The second and first derivative of MIR and NIR spectra, respectively, pointed out differentiations mainly in regions associated to unsaturated fatty acids. Principal component analyses could explore the similarities and differences among the samples, validating the spectral observations. Partial least squares models using both MIR and NIR spectral data allowed the prediction of the percentage of BO adulteration with SO with high accuracy, low relative errors and determination coefficients of the global fit above 0.9 for all calibration and validation datasets. Therefore, the spectroscopic tools applied in our work, when associated with chemometric approaches, showed good performance in the detection of BO adulterations. They can be of great importance in the development of new methodologies applied to quality control and authenticity of foods. •Adulterations of butter oil by soybean oil (0–100%) were made.•MIR and NIR spectroscopy techniques were used to study adulterations.•MIR and NIR spectra revealed bands associated with fatty acids within the samples.•PCA was able to discriminate the differences in the spectra.•PLS models resulted in low errors and high accuracy predicting % of adulteration.