Abstract
SCF MOs for azulene have been obtained by the semiempirical Pariser, Parr, Pople procedure using the Nishimoto–Mataga method of calculating repulsion integrals and the assumption that nearest neighbor resonance integrals are independent of interatomic distance. Excited states calculated from these MOs by a CI calculation are in very good agreement with experiment. Ground state charge densities, bond orders, and the dipole moment are similar to other SCFMO calculations and reveal no disadvantage in adopting a constant resonance integral for all bonds. It is shown that estimates of the π-electron charge density by n.m.r. methods are not compatible with direct dipole moment measurements and it is suggested that the interpretation of the n.m.r. measurements suffers from inaccuracies in estimating ring currents. Doubt is also thrown on the use of simple relationships between calculated π-bond orders and bond lengths obtained by X-ray crystallographic measurements on the solid state, particularly since all the bond lengths in azulene are predicted to be longer than in benzene whereas experiment shows some to be shorter. Calculations on spin densities and charge densities of the singly charged azulene anion and cation have been performed by a restricted Hartree–Fock perturbation method in which the matrix elements for the interaction between singly excited states and the ground state are calculated using the closed shell SCFMOs of azulene as the basis set. Agreement with experiment for the anion is fairly good. For the cation our results are in severe disagreement with recent VB calculations, but there are no experimental results available to decide between the two methods.