APPLICATION OF COSMO-SAC MODELING TO PREDICT LIPID EXTRACTION FROM SEMI-DEFATTED BRAZIL NUT PRESS CAKE USING DL-MENTHOL-BASED NATURAL DEEP EUTECTIC SOLVENTS AND ETHANOL AS A SUSTAINABLE ALTERNATIVE

The Brazil nut (Bertholletia excelsa HBK), native to the Amazon Rainforest, is highly valued for its nutritional quality and economic importance to Norte region in Brazil. With around 70% lipids, one of its main by-products is oil, typically obtained by cold pressing. This process produces a semi-defatted press cake as a residue, containing approximately 45% lipids, including important nutraceuticals such as tocopherols and phytosterols that can be recovered using solvents. DL-menthol-based Natural Deep Eutectic Solvents (NADES) has been studied as sustainable alternatives for extracting hydrophobic compounds like lipids. This study was aimed at evaluating, through thermodynamic computational simulation, the potential of these solvents combined with ethanol to extract both major and minor lipids. The screening was conducted using the COSMO-SAC model (Conductor-like Screening Model - Segment Activity Coefficient) to calculate sigma profiles and activity coefficients at infinite dilution for the triacylglycerols POLi and OOLi, and minor lipids β-sitosterol and γ-tocopherol, which are predominant in the lipid fraction of the Brazil nut. The solvents tested included DL-menthol combined with acetic acid, citric acid, lactic acid, and glycerol in molar ratios from 1:1 to 4:1, with or without ethanol in concentrations of 50-95% mol at a temperature of 60°C. The results were compared to pure ethanol. Computational simulation shows that ethanol enhances the extraction capacity of eutectic solvents for POLi and OOLi, especially in the DL-menthol:acetic acid combination, with coefficients ranging from 0.8 to 0.04. For β-sitosterol and γ-tocopherol, the best combinations were DL-menthol:acetic acid and DL-menthol:lactic acid in the presence of ethanol, with values lower than those of pure ethanol, closely resembling the interaction with n-hexane. Thus, the screening enabled the identification of the combinations with the best solute-solvent interactions, showing that COSMO-SAC modeling can be an interesting tool for predictive analysis.