Determinação de teor de sebo em óleos para produção de biodiesel usando espectroscopia no infravermelho médio

Abstract

Biodiesel is a fuel composed of long-chain fatty acid esters that serves as a sustainable alternative to petroleum diesel. The quality of biodiesel is directly linked to the physicochemical characteristics of the raw materials, whether from vegetable oils or animal fats. Mixing vegetable oils and animal fats is common in the biofuel industry to alter the saturation and unsaturation levels for better product performance. These fats are low-cost because they are easy-to-obtain byproducts of the food industry. Using them in biodiesel production benefits the environment by preventing the improper disposal of animal waste in rivers and streams, thus preventing environmental issues. In this context, a model was developed in this work to determine bovine fat in vegetable oils from soybean, palm, cotton, and corn using mid-infrared spectroscopy (FTIR-ATR) and chemometrics. Spectra were obtained in the range of 4000 to 660 cm⁻¹ from 32 samples of vegetable oils with different concentrations of bovine fat. Subsequently, principal component analysis (PCA) was performed to explore possible groupings. Partial least squares (PLS) regression was carried out, evaluating the performance of Kennard-Stone (KS) and SPXY sampling, and normalizations such as mean centering, autoscaling, Pareto, and Poisson. The quality of the regression was assessed by analytical figures of merit. PCA showed clustering according to the degree of saturation of the mixtures. The model with KS and Pareto showed better performance, with calibration and prediction R² figures of merit of 0.9995 and 0.9871, respectively, RMSEP = 4%, LOD = 1.61%, LOQ = 4.83%, and inverse analytical sensitivity of 0.4207%. Based on these results, the use of mid-infrared spectroscopy combined with chemometrics proved effective for determining the fat content in biodiesel raw materials.