A Molecular Dynamics Assisted Machine Learning Investigation on Property Dependent Transitions in a Mixture of Deep Eutectic Solvent and Water.

Deep eutectic solvents (DES) are highly viscous, which hinders their use in various applications. Their viscosities can be reduced by adding a small amount of water. However, when an excessive amount of water is added, the DES loses its inherent structure and becomes an aqueous solution of DES components. Various properties indicate structural changes from DES-dominant to water-dominant at different water concentrations. When hydrated DES is used for any application, it is very important that the original supramolecular structure is retained. Hence, it will be beneficial to have the property-dependent transition points in DES-water mixtures. We have proposed a method for determining those using classical molecular dynamics simulations and machine learning. Various structural and dynamical properties of DES-water mixtures at different water mole fractions are calculated and fitted using machine learning methods to locate their respective transition point. Transition points varied across properties, indicating a gradual disruption of the interactions present in the studied system.