Machine learning-driven generation and screening of potential ionic liquids for cellulose dissolution

Abstract

Cellulose, a highly versatile material, faces challenges in processing due to its limited solubility in common solvents. Ionic liquids have been found to possess high solvating capacities for cellulose. However, the experimental development of ionic liquids with optimal cellulose solubilities remains a time-consuming trial-and-error process. In this work, a virtual molecular library containing billions of potentially de novo ionic liquid candidates has been generated utilizing Monte Carlo tree search and recurrent neural network techniques. The library is subsequently screened through two predictive machine learning models, which have been pre-trained for predicting cellulose solubility and melting point of ionic liquids. The promising candidates were further validated and screened using the Conductor-like Screening Model for Real Solvents (COSMO-RS) model. Our work offers an efficient workflow and virtual molecular library, which should facilitate theoretical and experimental development of novel ionic liquids.