Thermodynamics-informed neural network predicts solvent behavior for biomass refining
Thermodynamics-Informed Learning Captures Hierarchical Hydrogen-Bond Networks in Deep Eutectic Solvents
Machine Learning
Summary
Deep eutectic solvents (DESs) are special liquids that help break down plant materials, but their effectiveness depends on complex hydrogen bonds between ingredients. The authors developed a neural network called TiNN that uses thermodynamics to understand these bonds and predict how well different mixtures dissolve substances like sugars and lignin. Trained on many data points, TiNN predicts solubility much more accurately than previous methods and can identify solvent formulas that remove specific plant components well. This tool helps guide the design of better solvents for converting plants into useful products.
What this means in practice
- •For chemical engineers: Predict effective solvent mixtures for refining plant biomass based on hydrogen-bond networks to improve fractionation processes.
- •For industrial chemists: Screen and select deep eutectic solvent formulations that optimize removal of lignin and hemicellulose from biomass feedstocks.
Authors
Yanpeng Ye, Chaoran Duan, Mian Xu, S. Mani Sarathy
Abstract
Deep eutectic solvents (DESs) offer a tunable platform for lignocellulose refining, but their solvation behavior depends on multicomponent hydrogen-bond networks rather than individual constituent properties. We develop TiNN, a thermodynamics-informed neural network that hierarchically represents HBA-HBD association and solute accommodation through a learned free-energy pathway. Trained on 938 records, TiNN achieves a test-set $R^2$ of 0.917 and generalizes to new HBA-HBD combinations and sparse conditions. Ablation identifies solute-specific stabilization as the largest contributor to prediction accuracy, while the importance of hydrogen-bond compatibility depends on the dissolution mechanism. Experimental measurements of glucose, xylan, and Kraft lignin in acidic, basic, and neutral DESs show that TiNN reduces the geometric mean solubility error from more than 23-fold with COSMO-RS to approximately 1.5-fold. Without training on fractionation data, TiNN also distinguishes DES formulations with high and low lignin and hemicellulose removal from corn stover. These results support chemically grounded solvent screening for lignocellulose fractionation.