Dependence of the electrostatic molecular potential upon the basis set and the method of calculation of the wave function. Case of the ground, ^3A_1(\pi\to\pi*) and ^1A_1(\pi\to\pi*) states of formaldehyde

The dependence of the molecular electrostatic potential V on the method of calculating the wave function and on the basis set has been examined for three electronic states of H2CO. The calculations analysed here refer to two different basis sets: a minimal set and a split-valence shell set, both supplemented with diffuse p orbitals. The methods of calculation include the SCF procedure (for the ground state), CI procedures of increasing complexity, which practically reach their asymptotic limits, and, for the excited states, simpler methods (rigid orbital excitation, complete CI of single excitations, electron-hole potential methods). It appears that in the ground state the two bases give equivalent descriptions of V when the CI process reaches the asymptotic limit and that the SCF descriptions approximate fairly well the more accurate ones. For the 3A1(π → π*) state the conclusions on the two basis sets are similar and, in addition, it is shown that a simple method (EHP) gives an approximation of V that reproduces the essential features of the complete CI calculations. The inadequacy of both basis sets for the representation of the 1A1(π → π*) state turns out to be evident. A discussion is presented concerning the possibility of adopting the calculation of V as an auxiliary tool to compare the accuracy of different descriptions of the same electronic state.