Constrained density functional theory based configuration interaction improves the prediction of reaction barrier heights

Publisher Website
Google Scholar
Abstract
In this work, a constrained density functional theory based configuration interaction approach (CDFT-CI) is applied to calculating transition state energies of chemical reactions that involve bond forming and breaking at the same time. At a given point along the reaction path, the configuration space is spanned by two diabaticlike configurations: reactant and product. Each configuration is constructed self-consistently with spin and charge constraints to maximally retain the identities of the reactants or the products. Finally, the total energy is obtained by diagonalizing an effective Hamiltonian constructed in the basis spanned by these two configurations. By design, this prescription does not affect the energies of the reactant or product species but will affect the energy at intermediate points along the reaction coordinate, most notably by modifying the reaction barrier height. When tested with a large set of reactions that include hydrogen transfer, heavy atom transfer, and nucleophilic substitution, CDFT-CI is found to improve the prediction of barrier heights by a factor of 2–3 for some commonly used local, semilocal, and hybrid functionals. Thus, just as CDFT can be used to cure energy errors in charge localized states, CDFT-CI can recover the correct energy for charge delocalized states by approximating the true wave function as a linear combination of localized configurations (e.g., reactant and product). The well-defined procedure and the promising results of CDFT-CI suggest that it could broaden the applicability of traditional DFT methods for reaction barrier heights.