Towards more realistic molecular modeling of catalysis with density functional theory: Combined QM/MM and ab initio molecular dynamics methods

In recent years, Kohn-Sham density functional theory (DFT) has emerged as the premier method to perform practical 'first principles' quantum mechanical calculations due to the method's favourable balance of efficiency and accuracy in dealing with chemical systems. Despite the efficiency of the method, DFT calculations often neglect large ligand effects, solvent effects and finite (non-zero) temperature effects because of the high computational expense of including them. In some cases, these neglected effects are not trivial, but critical to the chemistry of the molecular systems. This thesis involves the development and application of more realistic computational models that account for these often neglected effects, using the ab initio molecular dynamics method and the combined quantum mechanics and molecular mechanics (QM/MM) approach. For this purpose, the combined QM/MM methodology of Singh and Kollman has been implemented into both a conventional density functional electronic structure package (ADF) and a Car-Parrinello ab initio molecular dynamics package (PAW). The IMOMM coupling scheme of Maseras and Morokuma has been adapted as to allow for energy conserving dynamics and harmonic normal-mode frequency calculations to be performed thereby allowing free energy surfaces to be mapped out with the method. Additionally, a multiple time-step QM/MM molecular dynamics technique has been developed which allows the MM region to be sampled at a faster rate than the QM region. This provides faster equilibration and better ensemble averaging during the calculation of the free energy barriers, without increasing the expense of QM calculation. These implemented methodologies have been applied to examine recently developed transition metal based homogenous catalysts for olefin polymerization. Most notably we have examined the Ni-diimine catalyst of Brookhart in which the extended ligand structure plays a critical role in controlling the polymerization chemistry. We have also evaluated the applicability of using ab initio molecular dynamics as a practical tool in studying transition metal based catalysis. Finally, features have been implemented into the Car-Parrinello QM/MM method to allow for solvent simulations in the near future. The impact of this thesis has been to extend the envelope of high level mechanistic computational modeling to include large ligand effects, finite temperature effects and eventually reactions in condensed phases.