Transition states for chemical reactions I. Geometry and classical barrier height

A new computational procedure for the characterization of transition states for chemical reactions is proposed and tested. Previous calculations have frequently employed a single point high-level energy calculation at a transition state geometry obtained with a less expensive computational method, Energy[Method(1)]parallel to Geom[Method(2)]. If we instead search the "inexpensive" intrinsic reaction coordinate (IRC) for the maximum of Energy[Method(1)] along this reaction path, the resulting "IRCMax method", Max{Energy[Method(1)]}parallel to IRC(Geom[Metho(2)]}, reduces errors in transition state geometries by a factor of 4 to 5, and reduces errors in classical barrier heights by as much as a factor of 10. When applied to the CBS-4, G2(MP2), G2, CBS-Q, and CBS-QCI/APNO model chemistries, the IRCMax method reduces to the standard model for the reactants and products, and gives rms errors in the classical barrier heights for ten atom exchange reactions of 1.3, 1.2, 1.0, 0.6, and 0.3 kcal/mol, respectively. (C) 1998 American Institute of Physics.