Computational studies of earth abundant and organo- catalytic systems

Abstract

This thesis focuses on the application of Density Functional Theory (DFT) to manganese catalysed hydrogenation reactions (Part I) and organocatalytic reactivity (Part II).

Part I: A DFT benchmarking study was performed to reproduce experimentally determined values of hydricity, the heterolytic metal hydride bond strength of 3d transition metal (TM) complexes (Chapter 4) to ensure an accurate methodology for work on these systems. This methodology was employed for the modelling of manganese catalysed enantioselective ketone hydrogenation (Chapter 5). Rational design led to a catalyst with improved stereocontrol by introducing steric bulk in the active site (Chapter 5). Further modification of the ligand backbone led to the identification of routes to improve selectivity and activity by varying sterics and electronics on the ligand (Chapter 5). Using the dataset generated, Machine Learning was performed to predict DFT level barrier heights from GFN2-xTB level descriptors, though the predictive power was limited by the distribution and size of the dataset (Chapter 6).

Part II: DFT has been applied to a number of asymmetric reactions involving both isothiourea catalysis (Sections 9.1–9.5) and hydrogen bonding catalysis (Section 9.6). Isothiourea catalysts involve a chalcogen bonding interaction that locks the conformation of a key acyl ammonium intermediate, promoting facially-selective reactivity. The strength of the interaction can be modulated through decoration of the catalyst backbone (Section 9.1), though the nucleophilicity is key to retain catalytic activity over background reactions. DFT can be used to identify the stereodetermining transition states in reactivity (Sections 9.2–9.5) though some tailoring of methodology is required with large dipole moments (Section 9.2) and to adjust entropic contributions (Sec tion 9.3). GFN2-xTB performs well as a preoptimisation technique and for some transition states, but leads to overbinding of the chalcogen interaction, shortening the distance into the repulsive regime of DFT. This remains a powerful technique for conformational sampling, identifying representative low-lying conformations to accurately model diastereoselective rearrangement reactions (Section 9.6).