(630a) Walking through Chemical Reaction Space

Two key aspects to the rational design of catalysts are the ability to assess potential reaction mechanisms and the understanding of complex chemistry that might involve many competing reactions. The first problem can be summarized as the need to be able to compute reaction rates from first principles atomistic models. Electronic structure theory (i.e. quantum chemistry) provides the tools for this, but has been divided into two camps – density-based and wavefunction-based methods. Traditionally, it has been thought that the latter were computationally prohibitive for realistic chemical problems because of their poor scaling with molecular size. We discuss recent advances that question this conclusion and suggest that the intrinsic computational complexity of density-based and wavefunction-based methods may not be very different. The second problem concerns the identification of all the potential reactions that might compete. Traditional hypothesis-based approaches proceed by creative identification of possible reactions that can then be assessed using computational methods. An alternative method would use computational methods to discover these reactions – akin to Monte Carlo search of the space of chemical reactions. We show how this is now possible and how it can be combined with traditional hypothesis-based approaches.