Characterization of Biodiesel from Animal Fat, Vegetable Oil, and Adulterants by Infrared Spectroscopy Combined with Chemometric Methods

Abstract

Second- and third-generation feedstocks, combined with first-generation feedstocks, have been showing a tendency to meet the supply and quality standards of biodiesel, a less polluting renewable energy source that has been gaining prominence year after year. The objective of the study was to quantify the weight proportion (%) that each biodiesel contributed to the composition of the blend, in addition to the presence of the adulterant soybean oil. Fourier-transform infrared spectroscopy combined with chemometric techniques allowed distinguishing between feedstocks of the animal and vegetable origin and adulterants and determining their ratios in 45 blends. Multivariate control charts based on the net analyte signal allowed obtaining most data on biodiesel from beef tallow and lard, whereas the interference chart was used for matrix variability. Partial least squares discriminant analysis (PLS-DA) was applied to the same dataset, and the coefficient of determination was 0.9999 for the calibration and prediction sets, proving to be a robust method for the identification of B100 biodiesel blends from different feedstocks and adulterants. PLS was used as a qualitative and quantitative tool, showing optimal correlation between calibration and prediction, with R cv 2 = 0.9999 and R p 2 = 0.9999, and low root-mean-square error of cross-validation and root-mean-square error of prediction values. Orthogonal signal correction was employed for data processing in the PLS-DA and PLS models, rendering them more cohesive and eliminating the variability of the dataset that is orthogonal to the parameter of interest. The full spectral range and only one latent variable were used. PLS-DA and PLS models proved to be 100% reliable for the identification and quantification of adulterants in biodiesel and biodiesel blends.