Triplet state Z/E-photoisomerizations of polyenes:: A comparison of ab initio and density functional methods

In the search for less CPU-costly methods for study of triplet state Z/E-photoisomerization of olefins, Kohn-Sham density functional theory (DFT) has been tested on 1,3-butadiene (BD) and 1,3,5-hexatriene (HT). Computed T-1 energies were compared to those from CASSCF, CASPT2, and spin-projected UMP4(SDTQ) calculations as well as experiments. For both molecules it is necessary that nonlocal gradient corrections are made to the exchange functional since usage of the local spindensity approximation for exchange in most cases leads to vertical and relaxed T-1 energies that are too high. Gradient-corrected DFT as well as hybrid functional methods lead to T-1 energies that are bracketed by the corresponding UMP4(SDTQ) and CASPT2 energies and lie at most 4 kcal/mol below measured values. The relaxed T-1 energies for planar geometnes are in slightly better agreement with experiment when calculated by pure nonlocal gradient-corrected DFT than by hybrid functional methods. However, T-1-state potential energy surfaces obtained by either type of method explain the experimental observations on triplet-state Z/E-photoisomerizations of BD and HT, and geometries of T-1 isomers of ED and HT compare well with those from UMP4(SDQ), UMP2, and CASSCF calculations. Finally, it should be noted that for both molecules UHF deviates from the higher computational levels in T-1 energies by 20-30 kcal/mol and should be avoided in all computations of T-1 states of olefins.