Configuration interaction study of the oscillator strenghts for the 1A11A1 and 1A11A1 transitions of the water molecule

Abstract

Abstract Oscillator strengths for transitions between the X 1 A 1 ground state of water and its B 1 A 1 and D 1 A 1 excited states are computed employing two different theoretical approaches. In one series of calculations a common orthonormal one-electron basis is employed for all of the above states, while in the other type of treatment two different, mutually non-orthogonal, sets are used; the multireference single- and double-excitation (MRD CI) method is employed in each case, with configuration selection, to generate the various electronic wavefunctions. It is found that the use of ground-state SCF MOs leads to poor convergence in the wavefunctions of the (Rydberg-type) B 1 A 1 and D 1 A 1 excited states and consequently also for the corresponding B  X and D  X f -values; this behavior is seen to be closely related to the near degeneracy of the two excited states, each of which is a mixture of the 3a 1 → 3sa 1 and 1b 1 → 3pb 1 configurations. Analogous computations with the A 1 B 1 1 b 1 → 3sa 1 MOs show much better convergence properties, and the resulting f -values compare well with what is obtained when state-specific orbital sets are employed separately for ground and excited states and non-orthonormal techniques are applied to compute the desired transition moments. These results tend to confirm previous findings which indicate conceptual and computational advantages for the calculation of excited-state wavefunctions and properties within the context of a state-specific theory. They also show that although the goal of eliminating the dependence of MRD CI calculations on the choice of MO basis is very nearly approached for energy quantities, it is less satisfactorily achieved for other properties, especially when the existence of nearly degenerate electronic states is a critical factor.