COMPUTATIONAL SCREENING OF DEEP EUTECTIC SOLVENTS FOR PROCYANIDIN B1 AND B2 RECOVERY FROM BRAZIL NUT CAKE

Brazil nut (Bertholletia excelsa) is an important Amazonian resource with significant economic and nutritional relevance. However, the efficient utilization of by-products generated during its processing remains a challenge, particularly in the case of the cake resulting from mechanical oil extraction. This solid fraction contains compounds of nutritional and technological interest, representing a promising feedstock for the development of higher-value products. In a previous study, procyanidins B1 and B2 were identified as the major phenolic compounds present in this matrix and were therefore selected as target molecules in the present work. In this context, the objective was to investigate, through a computational approach, the potential of emerging green solvents, particularly Deep Eutectic Solvents (DES), for the recovery of these compounds, establishing a thermodynamic basis for their subsequent experimental evaluation. Computational solvent screening was performed using a multistep molecular modeling approach. Initially, multiple hydrogen-bond interactions were parametrized using the GMHB1808 software. Subsequently, molecular sigma profiles were obtained using the GAMESS Quantum Chemistry package and then employed in simulations performed with JCOSMO software to characterize solvent–solute interactions. A total of 500 DES formulations were investigated, considering four hydrogen-bond acceptors (HBAs), 25 hydrogen-bond donors (HBDs), and five HBA molar ratios (1:3, 1:2, 1:1, 2:1, and 3:1). The systems were evaluated at 333.15 K with the addition of 30% water. The predicted affinity of each solvent system toward procyanidins B1 and B2 was assessed using the activity coefficients at infinite dilution (γ_i^∞), with lower values considered indicative of more favorable solvent–solute interactions. The computational analysis enabled the selection of 20 DES systems for experimental validation at a 1:1 molar ratio. The selected systems comprised choline chloride and betaine as HBAs, combined with HBDs exhibiting favorable thermodynamic behavior, including 1,3-propanediol, γ-valerolactone, glycerol, 1,3-butanediol, formic acid, acetic acid, 1,2,4-butanetriol, and fructose. Tartaric acid and citric acid, which exhibited less favorable predicted interactions, were also selected as reference systems. The results demonstrate that the predicted affinity toward procyanidins B1 and B2 varied according to DES composition, highlighting the importance of solvent formulation for the recovery of these compounds. Overall, the computational screening provided a systematic strategy to narrow down the large number of possible DES combinations and select systems for experimental evaluation. This approach may contribute to the development of more sustainable extraction processes and to the valorization of Brazil nut cake as a source of high-value phenolic compounds.

Acknowledgments: CNPq - grant numbers 306753/2025-1; CAPES - Finance code 001 and FAPESP - grant numbers 2022/10368-4, 2024/02091-8 and 2025/21661-2.