The development of analytic-gradient methodology for excited states withinconventional time-dependent density-functional theory (TDDFT) would seem tooffer a relatively inexpensive alternative to better establishedquantum-chemical approaches for the modeling of photochemical reactions.However, even though TDDFT is formally exact, practical calculations involvethe use of approximate functionals, in particular the TDDFT adiabaticapproximation, whose use in photochemical applications must be furthervalidated. Here, we investigate the prototypical case of the symmetric CC ringopening of oxirane. We demonstrate by direct comparison with the results ofhigh-quality quantum Monte Carlo calculations that, far from being anapproximation on TDDFT, the Tamm-Dancoff approximation (TDA) is a practicalnecessity for avoiding triplet instabilities and singlet near instabilities,thus helping maintain energetically reasonable excited-state potential energysurfaces during bond breaking. Other difficulties one would encounter inmodeling oxirane photodynamics are pointed out but none of these is likely toprevent a qualitatively correct TDDFT/TDA description of photochemistry in thisprototypical molecule.
展开▼
