Estratégias analíticas não-destrutivas para monitoramento da qualidade de misturas biodiesel/diesel

Abstract

Biodiesel and ethanol are the most industrially produced biofuels due to their economic viability, as well as being important alternatives to conventional fossil fuels, offering a more ecological and sustainable perspective. In this work, near-infrared spectroscopy (NIR) combined with chemometric tools was used for monitoring the quality of biodiesel/diesel blends in terms of (i) simultaneous classification of the synthesis route and biodiesel feedstock, (ii) authentication of second-generation biodiesel, and (iii) biodiesel content quantification. Data-Driven Soft and Independent Modeling of Class Analogy (DD-SIMCA) achieved 100% sensitivity and specificity in the test set for authenticating ethyl blends, while the Successive Projections Algorithm for interval selection in Partial Least Squares Discriminant Analysis (iSPA-PLS-DA) correctly discriminated all ethyl blends containing cotton, sunflower, and soybean biodiesel. Additionally, only one misclassification occurred when ethyl and methyl blends of the same three oil feedstocks were included in the model. For the last two applications, two spectral regions (881-1651 and 1911-2203 nm) were investigated to avoid sample dilution with organic solvents. As a result, the first derivative of Savitzky-Golay with a second-order polynomial and a 15-point window proved to be the most promising preprocessing technique when applied to the spectral range of 1911-2203 nm using DD-SIMCA, with efficiencies of 97.6 and 99.2% for authenticating jatropha and tallow biodiesel/diesel blends, respectively. For biodiesel content quantification, a relative prediction error (REP) of only 2.85% was obtained. As advantages, the proposed analytical methodology contributes especially to United Nations Sustainable Development Goal (SDG/UN) No. 7 (affordable and clean energy), and aligns with the principles of Green Analytical Chemistry.