RESUMO
This paper describes the feasibility of synchronous fluorescence (SyF) spectroscopy combined with multivariate data analysis for qualitative and quantitative determination of adulteration of cold pressed grape seed oil (GSO) with refined soybean oil (SBO). SyF spectroscopy data of oil samples were collected in the region of 250-800â¯nm at excitation-emission wavelength differences (∆λ) of 10, 20, 30, 40, 50, 60, 70 and 80â¯nm. Three different multivariate methods, namely principal component analysis (PCA), soft independent modeling of class analogies (SIMCA) and partial least square regression (PLSR) analysis were used for data analysis. To simulate the adulteration of cold pressed GSO with refined SBO, ninety-six adulterated samples were prepared at adulterant levels from 5% to 50%. HPLC-FLD method was used as reference in order to authenticate pure oils and binary oil mixtures. The SIMCA models provided an excellent classification for pure cold pressed GSO versus other vegetable oil samples, with a 95% significance level. The classification error rate of SyF spectroscopy for detecting SBO added to GSO was also better than 5%. PLSR calibration models constructed for the evaluation of GSO purity and for the adulterants SBOs were internally validated by the leave-one-out procedure (cross-validation) and their predictive ability was assessed by independent external validation sets. Under the optimum conditions, the plots of observed versus predicted values exhibited a good linearity (R2 > 0.99). The root mean square errors of calibration (RMSEC) and cross-validation (RMSECV) were in the range 0.55-4.46% and 2.14-5.35%, respectively. Excellent predicting capabilities were also obtained using an external validation set consisting of GSO adulterated with SBOs from different brands.