The photodynamics along the main decay paths of thymine after excitation to the lowest pi,pi* state have been studied with MS-CASPT2 calculations and semiclassical CASSCF dynamics calculations including a surface hopping algorithm. The static calculations show that there are two decay paths from the Franck-Condon structure that lead to a conical intersection with the ground state. The first path goes directly to the intersection, while the second one is indirect and involves a minimum of the pi,pi* state, a small barrier, and a crossing between the pi,pi* and n,pi* states. From the static calculations, both paths have similar slopes. The dynamics calculations along the indirect path show that, after the barrier, part of the trajectories are funneled to the intersection with the ground state, where they are efficiently quenched to the ground state. The remaining trajectories populate the n,pi* state. They are also quenched to the ground state in less than 1 ps, but the static calculations show that the decay rate of the n,pi* state is largely overestimated at the CASSCF level used for the dynamics. Overall, these results suggest that both direct and indirect paths contribute to the subpicosecond decay components found experimentally. The indirect path also provides a way for fast population of the n,pi* state, which will account for the experimental picosecond decay component.

The role of the seam of intersection between the lowest (pi,pi*) and (n,pi*) excited states in the decay of electronically excited singlet thymine has been investigated with ab initio complete active space self-consistent field (CASSCF) calculations and direct dynamics variational multiconfiguration Gaussian (DD-vMCG) quantum dynamics on the full-dimensional CASSCF surface, with 39 degrees of freedom. The seam has a sloped-to-peaked topography, and the dynamics at the different segments of the seam have been studied by varying the initial conditions of the propagation. When the wave packet is directed to the peaked segments, part of it traverses the seam, stays on the (pi,pi*) state and heads towards decay to the ground state. In contrast to this, when the wave packet is driven to sloped seam segments it bounces back to the minimum of the (pi,pi*) state. Significant population transfer to the (n,pi*) state is observed in both cases. The results suggest that a sloped-to-peaked topography can be used to control photochemical reactivity, by driving the wave packet to different regions of the seam where a different outcome of the propagation can be expected.

Recently, the surprising result that ab initio calculations on benzene and other planar arenes at correlated MP2, MP3, configuration interaction with singles and doubles ͑CISD͒, and coupled cluster with singles and doubles levels of theory using standard Pople's basis sets yield nonplanar minima has been reported. The planar optimized structures turn out to be transition states presenting one or more large imaginary frequencies, whereas single-determinant-based methods lead to the expected planar minima and no imaginary frequencies. It has been suggested that such anomalous behavior can be originated by two-electron basis set incompleteness error. In this work, we show that the reported pitfalls can be interpreted in terms of intramolecular basis set superposition error ͑BSSE͒ effects, mostly between the C-H moieties constituting the arenes. We have carried out counterpoise-corrected optimizations and frequency calculations at the Hartree-Fock, B3LYP, MP2, and CISD levels of theory with several basis sets for a number of arenes. In all cases, correcting for intramolecular BSSE fixes the anomalous behavior of the correlated methods, whereas no significant differences are observed in the single-determinant case. Consequently, all systems studied are planar at all levels of theory. The effect of different intramolecular fragment definitions and the particular case of charged species, namely, cyclopentadienyl and indenyl anions, respectively, are also discussed.