Abstract
A potential energy surface is calculated for the Li+F2→LiF+F reaction using an ab initio multistructure valence-bond approach. The orthogonalized Moffit (OM) method is used to apply a correction for the large errors made by the ab initio calculation in representing the F- ion relative to the F atom. The lowest potential energy surface is predicted by the OM method to be of the highly ‘attractive’ or ‘early downhill’ type and possesses a substantial well with respect to dissociation to the products LiF+F. The ground state wavefunction and the charge distribution corresponding to it are analysed and the results contrasted with past expectations. The attributes of the surface predicted by the OM method are compared with those suggested by experiment and used in trajectory calculations for analogous systems. The surface is found to be in qualitative agreement with all the features deduced from experiments. The electron-jump region of the surface is examined and the electron-jump distance is plotted as a function of F-F separation. A crude ‘harpooning’ model is used to estimate the total reactive cross section, which is predicted to be a sharply increasing function of the F2 vibrational quantum number. The excited states of the system, in various nuclear configurations, are examined, and the contribution of non-adiabatic transitions to the reactive scattering is considered. The possible outcome of non-reactive Li+F2 collisions at hyperthermal energies is also discussed. Surfaces are calculated using Moffit's atoms-in-molecules method and Hurley's ICC method to correct for the atomic errors in the ab initio calculation. These surfaces are compared with those calculated by the ab initio and OM methods.