• Rational screening of optimal ChCl-Gly for resveratrol extraction via COSMO-RS predictions. • Establishment of an efficient UAE-DES process achieving a comparable yield to conventional reflux in one-third of the time (20.5 min vs. 60 min). • SEM visualization reveals synergistic cell wall disruption by DES and ultrasound, explaining the enhanced extraction mechanism. Resveratrol is a phenolic compound renowned for its anti-inflammatory, antioxidant, and anti-apoptotic properties. To develop a greener alternative to conventional organic solvents, this research established an effective ultrasound-assisted extraction (UAE) approach utilizing deep eutectic solvents (DES) to isolate resveratrol from Polygonum cuspidatum. The optimal DES, choline chloride–glycerol (ChCl-Gly) with a molar ratio of 1:2 and a 20% water content, was rationally screened using the COSMO-RS model. Subsequent optimization of the extraction parameters was conducted using response surface methodology (RSM). The maximum resveratrol content reached 4.45 ± 0.25 mg/g, which proved similar to that obtained by conventional ethanol reflux extraction (ERE), under the following parameters: 64 °C bath temperature, 200 W ultrasonic power, 20.5 min duration, and 33.3 mL of liquid-to-solid ratio. Notably, the DES-UAE method achieved this efficiency in a significantly shorter time (20.5 min vs. 60 min for ERE). Scanning electron microscopy (SEM) indicated that the combined action of DES and ultrasound caused more severe disruption of plant cell walls than ERE, providing a mechanistic explanation for the enhanced extraction efficiency. This work demonstrates that COSMO-RS-guided screening of DES enables the rapid identification of an optimal solvent system—ChCl-Gly—for the efficient ultrasound-assisted extraction of resveratrol from P. cuspidatum . The integration of computational prediction with experimental validation not only accelerated solvent selection but also yielded a fast, effective, and environmentally friendly alternative to conventional extraction methods.
Lu et al. (Sun,) studied this question.