(152g) A Statistical Analysis of Computational Hydrogen Bond Features in Hydrophobic Non-Ion Deep Eutectic Solvents and Non-Deep Eutectic Solvents

Hydrophobic non-ion deep eutectic solvents have shown promise as advanced extractants for separate pollutants from aqueous solutions because of their potential in low toxicity, extremely low water solubility, and economic availability. However, the current discovery of hydrophobic non-ion deep eutectic solvents mainly relies on the intuition of the researchers or a trial-and-error method, which may have a low success rate and neglect many promising candidates. Hydrogen bonds between the compounds are considered the primary reason for the deep eutectic effect. Thus, the hydrogen bonds are likely to be the source of proper features that can be used to design new deep eutectic solvents and predict their properties. To guide the development of hydrogen bond-based deep eutectic solvent design, we analyze the structural and dynamic properties of hydrogen bonds in 50 hydrophobic deep eutectic solvents and 100 non deep eutectic solvent systems formed by various organic hydrogen bond donors and acceptors using molecular dynamics simulations. The structural and dynamic properties of the hydrogen bonds are analyzed using the distributions of average numbers and lifetimes of the inter- and intra- hydrogen bonds between compounds. The statistical analysis shows that the hydrophobic deep eutectic solvents possess a distinct gap between the compounds’ average intra-hydrogen bond numbers that the non-deep eutectic solvent systems do not present. This statistical analysis indicates that the intra-component and inter-component hydrogen bond numbers could be used to develop features that differentiate deep eutectic solvents from the non-deep eutectic solvent systems. Based on this observation, we developed a machine learning model to classify if two organic molecules can form deep eutectic solvents.