Abstract
Ionic liquids (ILs) and deep eutectic solvents (DESs) as innovative classes of solvents with excellent physical and chemical properties exhibit huge potential in numerous areas. Despite a great number of experimental works on those solvents, theoretical studies from the molecular level are still insufficient. In this dissertation, Ab initio molecular dynamics (AIMD) simulations for imidazolium-based ILs were performed to validate our two previously reported force fields (FF): ± 0.8 charge-scaled OPLS-2009IL FF and virtual site OPLS-VSIL FF. Then, those two FFs and another OPLS-DES FF were applied to the extractive desulfurization simulations. Five generally applied imidazolium-based ILs and two widely studied DESs were selected to investigate the interactions between solvent and solute, like thiophene, dibenzothiophene, and 4-methyldibenzothiophene. Besides the inorganic sulfur simulations, this work also focused on the studies of the conformational changes for two structurally and functionally similar monooxygenases in the process of alkanesulfonate catalysis, SsuD and MsuD, using accelerated MD simulations. Three distinct mobile loop conformations for SsuD systems, open, closed, and semiclosed, were identified by the simulations. New salt bridges and a novel π−π interaction were first reported. Unlike SsuD, the MsuD complex showed a tighter loop, demonstrating that a longer alkane chain was required for SsuD to maintain the appropriate architecture for desulfonation. In addition, machine learning models trained with a deep neural networks (DNNs) package were applied to reproduce the potential energy surface of various organic reactions in solution.