Computationally Guided Design of Natural Deep Eutectic Solvents for Sustainable Extraction of Polyphenol Compounds: Process Intensification, Molecular Mechanisms, and Solvent Circularity

Abstract The sustainable recovery of plant-derived polyphenols requires extraction systems that can simultaneously reduce organic solvent use, improve process efficiency, and provide mechanistic clarity. Herein, a computationally guided natural deep eutectic solvent (NADES) strategy was developed for efficient and sustainable polyphenol recovery from Dioscorea cirrhosa Lour. The conductor-like screening model for realistic solvents (COSMO-RS) was first employed to predict solvent-solute affinities among eight NADES candidates, identifying betaine-maleic acid (Bet-MA) as the most promising green solvent due to its favorable polarity distribution and hydrogen-bonding complementarity. The ultrasound-assisted extraction process was subsequently optimized using an artificial neural network coupled with a genetic algorithm. Under identical initial extraction conditions, Bet-MA achieved a 1.63-fold higher TPC yield than 70% ethanol. After ANN-GA optimization, the UABM process reached 38.14 mg GAE/g DW, corresponding to a 1.73-fold increase compared with the conventional UAE control. Mechanistic investigations combining FT-IR, SEM, and 100 ns molecular dynamics simulations revealed a synergistic dual mechanism: ultrasonic cavitation disrupted the compact plant cell wall architecture, while Bet-MA provided stronger thermodynamic stabilization of released phenolics through electrostatic and van der Waals interactions. UPLC-Q Exactive/MS profiling confirmed the enrichment of diverse phenolic acids and flavonoids, and antioxidant assays verified that the intensified process preserved the bioactivity of the recovered extract. The Bet-MA solvent retained approximately 62 and 49% of its initial extraction capacity after the fifth and sixth reuse cycles, respectively, suggesting its reuse potential under the tested conditions. An AGREEprep-based greenness assessment gave a score of 0.80 for the Bet-MA-based process, compared with 0.59 for conventional ethanol extraction, indicating a more favorable greenness profile. This work provides a rational framework for sustainable biomass valorization through computational solvent design, process intensification, molecular-level mechanism elucidation, and solvent reuse.