Investigating the enhancement mechanism of cellulose enzymatic hydrolysis: Machine learning-assisted acidic deep eutectic solvent pretreatment.

In this work, a series of acidic deep eutectic solvents (ADESs) were synthesized from organic acids combined with either choline chloride (Chcl) or betaine (Bet), and their effectiveness in the selective fractionation and enzymatic hydrolysis of corn stover was systematically evaluated. It was innovatively integrated the ADES physicochemical properties with machine learning revealed that Chcl-based ADESs identified as the first principal component via machine learning-driven principal component analysis (PCA), exhibited a significantly stronger dissociating effect on lignocellulose than Bet-based ADESs, due to their Cl- ions and hydroxyl groups. Critically, it was revealed that a suitable net hydrogen-bond donating capacity of ADESs, particularly with an appropriate carboxylic acid content (defined as the second principal component), was the key to enhancing the selective degradation of lignocellulose and its subsequent enzymatic hydrolysis. Notably, Chcl-glycolic acid (Chcl-GA) displayed particularly exceptional pretreatment performance, achieving 79.5% hemicellulose removal and significantly increasing enzymolysis efficiency to 85.4% under the optimized conditions. Kinetic analysis revealed that the degradation of cellulose, hemicellulose and lignin during the Chcl-GA pretreatment all followed the first-order kinetic model, with hemicellulose exhibited the highest degradation rate constant (k = 0.0197) and the lowest activation energy (Ea = 2576 kJ/mol), allowing it to be efficiently degraded at the initial stage. Mass balance analysis further revealed that the recovery yields of hemicellulose sugars and lignin from the pretreatment liquid were 93.6% and 80.3%, respectively, providing a novel strategy for the rational design of ADESs for efficient lignocellulose biorefinery.