Ab initio study of the n-π* electronic transition in acetone: Symmetry-forbidden vibronic spectra

Abstract

Ab initio calculations of geometry and vibrational frequencies of the first singlet excited A21(1A″) state of acetone corresponding to the n-π* electronic transition have been carried out at the CASSCF/6-311G** level. The major geometry changes in this state as compared to the ground state involve CO out-of-plane wagging, CO stretch and torsion of the methyl groups, and the molecular symmetry changes from C2v to Cs. The most pronounced frequency changes in the A″1 state are the decrease of the CO stretch frequency v3 by almost 500 cm−1 and the increase of the CH3 torsion frequency v12 from 22 to 170 cm−1. The optimized geometries and normal modes are used to compute the normal mode displacements which are applied for calculations of Franck–Condon factors. Transition matrix elements over the one-electron electric field operator at various atomic centers calculated at the state-average CASSCF/6-311+G** level are used to compute vibronic couplings between the ground A11, A21, and Rydberg B21(n-3s), 2 A11(n-3py), 2 A21(n-3px), 2 B21(n-3pz), and B11(n-3dxy) electronic states, and the Herzberg–Teller expansion of the electronic wave function is applied to derive the transition dipole moment for A11→A21 as a function of normal coordinates. The results show that the intensity for this transition is mostly borrowed from the allowed A11-B21(n-3s) transition due to vibronic coupling between A21 and B21 through normal modes Q20, Q22, and Q23 and, to some extent, from the A11-B11 transition due to Q19 (CO in-plane bend) which couples A21 with B11(n-3dxy). The calculated total oscillator strength for the n-π* transition through the intensity-borrowing mechanism, 3.62×10−4, is in close agreement with the experimental value of 4.14×10−4. Ninety-four percent of the oscillator strength comes from the perpendicular component (b1 inducing modes) and 6% from the parallel component (b2 modes). Calculated spectral origin, 30 115 cm−1 at the MRCI/6-311G** level, underestimates the experimental value by ∼300 cm−1. Calculated positions of the most intense peaks in the spectra also reasonably agree with the experimental band maximum. The presence of numerous weak vibronic peaks densely covering a broad energy range (∼12 000 cm−1) explains the diffuse character of the experimental n-π* band. Most of the bands observed in fluorescence excitation spectra [Baba and Hanazaki, Chem. Phys. Lett. 103, 93 (1983); Baba, Hanazaki, and Nagashima, J. Chem. Phys. 82, 3938 (1985)] can be assigned based on the computed spectrum.