Desenvolvimento de um método para determinação da adulteração do azeite de oliva com óleos de soja e canola por espectroscopia no infravermelho médio e calibração multivariada

Abstract

Olive oil (AO) has been marketed in several countries, including Brazil, mixed with other vegetable oils. The purpose of this work was to determine the AO content in mixtures with soybean and canola oil using infrared spectroscopy and partial least squares (PLS) multivariate calibration. 128 mixtures of AO, soybean oil and canola oil (0.00 to 100.00 g of AO 100 g-1) were prepared. The spectra were obtained in an infrared spectrophotometer, in the range of 3500 to 650 cm -1. The mixtures were divided into calibration sets (n = 95) and validation (n = 33) by the Kennard-Stone algorithm. The spectrum was divided into segments ranging from 1 to 100, and the PLS calibration was applied for each segmentation interval. The selected region (1219-1030 cm-1) presented the lowest value for the square root of the mean cross-validation error (RMSECV), which was 3.35. After the identification and exclusion of the outliers, the RMSECV value was 1.44, with r = 0.998. The accuracy was estimated as repeatability, whose value was 0.39 g 100 g-1. The limits of detection (LOD) and quantification (LOQ), lower and higher, were respectively 1.6 and 3.1 g 100 g-1 of mixture and 4.8 and 9.3 g 100 g-1 The obtained model Presented adequate predictive capacity and can be used in the quantification of AO in mixed oils with soybean and canola oils and in AO adulterated with these oils.