An analytical ab initio potential surface and the calculated tunneling energies for the HCl dimer

Abstract

The six-dimensional potential energy surface of the HCl dimer has been calculated ab initio at 1654 nuclear geometries [A. Karpfen, P. R. Bunker and P. Jensen, Chem. Phys., in press]. In the present paper we have fitted an analytical function to these points; the analytical function is similar to that used previously by us for the potential surface of the HF dimer. The fitted function has 38 adjustable parameters and the standard deviation of the weighted fit is 19.0 cm −1. We have determined the minimum energy path for the trans-bending tunneling motion on this surface, and have calculated the tunneling and K-rotation energies and wavefunctions. Around equilibrium the path is qualitatively similar to that for the HF dimer in that there are two equivalent hydrogen-bonded structures of C s symmetry (which are approximately L-shaped with a “bound” and a “free” H-atom) that can tunnel through a C 2 h saddle point (the “closed” C 2 h saddle point). However, away from equilibrium the path is qualitatively different from that found for the HF dimer since the HCl dimer never becomes linear along the path; in fact it passes through a second C 2 h saddle point (the “open” C 2 h saddle point). As a result the A-rotational constant only varies slightly along the path, and this explains the experimental observation that the tunneling splitting varies little with K-type rotation for the HCl dimer, in contrast to the situation for the HF dimer. Quantitatively it is clear that errors in the ab initio calculation, errors in the fitting of an analytic function to the points, the correction to the path that is caused by the zero point motion in the other vibrations, and the coupling between the four low-frequency modes, will all be relatively more significant than they were for the HF dimer because the full six-dimensional potential is much flatter; the ab initio dissociation energy is only ∼600 cm −1, and the ab initio tunneling barrier is only ∼70 cm −1. Therefore, we modify the tunneling potential, by fitting to data, in order to improve the reliability of the predictions.