Rational Design of Mixed Solvent Systems for Acid-Catalyzed Biomass Conversion Processes Using a Combined Experimental, Molecular Dynamics and Machine Learning Approach

摘要

Mixtures of water and organic cosolvents (mixed solvent systems) play an important role in mediating acid-catalyzed biomass conversion reactions. A minimum amount of water is typically required to dissolve biomass-derived materials, while adding an organic cosolvent can enhance the rates and selectivities of the desirable, catalytic reaction steps. Understanding the molecular-level bases underlying these solvent effects would provide a powerful measure of control over the reaction environment for biomass conversion processes, whereby the rates of desired reaction steps could be preferentially enhanced over the undesirable ones by modulating the composition of the solvent system. However, a quantitative basis to anticipate these solvent effects is currently lacking, and optimizing the composition of the liquid phase for new biomass conversion reactions typically requires laborious screening of the continuous space of possible mixed solvent systems. Herein, we summarize our efforts to estimate solvent effects on the rates and selectivities of liquid-phase, acid-catalyzed biomass conversions reactions using experiments, classical molecular dynamics simulations, and machine learning tools. We then synthesize these insights into a workflow that allows for the rational design of mixed solvent systems for acid-catalyzed biomass conversion processes using computationally efficient methods and minimal experiments. We demonstrate this design framework by analyzing two case studies: the acid-catalyzed dehydration of cyclohexanol to cyclohexene, and the partial dehydration of fructose to 5-hydroxymethylfurfural.