Computational Strategy in Catalyst Design

Citation

Nielsen, Robert J. (2005) Computational Strategy in Catalyst Design. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/60VH-AQ40. https://resolver.caltech.edu/CaltechETD:etd-06052005-221223

Abstract

The strategy and efficacy of applying computational tools to the development of new catalytic cycles is discussed using the enantioselective palladium-catalyzed aerobic oxidation of secondary alcohols as a model case. The key interactions responsible for the unique reactivity of ((–)-sparteine)PdX₂ complexes (X = chloride, acetate) in kinetic resolutions of secondary alcohols are elucidated using density functional theory with the Poisson-Boltzmann polarizable continuum solvent model. Enantioselectivities in these reactions are found to follow directly from calculated energies of diastereomeric beta-hydride elimination transition states incorporating (R) and (S) substrates. This relationship reveals an important role of the anion, namely to communicate the steric interaction of the ligand on one side of the Pd^II square plane and the substrate on the other side. When no anion is included, no enantioselectivity is predicted. Locating these transition states in different solvents shows that higher dielectrics stabilize the charge separation between the anion and metal and draw the anion farther into solution. Thus the solvent influences the barrier height (rate) and selectivity of the oxidation.

Based on this understanding, computational assays for selectivity, reaction rate and stability are developed and used to screen possible mimics of the natural product (–)-sparteine which could be synthesized in both antipodes. Derivatives of the bispidine and bispidinone structures are predicted to have high selectivity but poor stability on palladium. Experimental results verify that catalytically active (bispidine)PdX₂ complexes do not form.

Mechanisms by which palladium diacetate complexes of N-heterocyclic carbenes may oxidize alcohols (a reaction known to occur with no enantioselectivity) are examined computationally. The strong trans effect of the carbene distinguishes the behavior of these complexes from other palladium catalysts. No traditional beta-hydride elimination is predicted to be capable of generating the high deuterium kinetic isotope effect measured using this catalyst. Instead, the low-energy pathway consistent with previous experimental observations (KIE, activation parameters, kinetics) is a "reductive" beta-hydride elimination, in which the beta-hydrogen of the alcohol is transferred directly to a bound acetate ligand. Assuming that relative energies of transition states of this type will determine enantioselectivity, new, chiral carbene ligands are hypothesized and screened. Careful placement of stereocenters and steric bulk has led to ligands with high predicted enantioselectivity and stability.

Recurring factors in the induction of selectivity by asymmetric ligands are observed. Strengths and weaknesses of quantum chemistry as applied to catalytic cycles are discussed, along with the synergy of theory and experiment. Common pitfalls and areas in need of improvement are highlighted.

Thesis Files