disagreement between the vibrational levels of it and those of the Bowman potential and for remaining uncertainties in the assignment of the observed levels. We have suggested in the above analysis a corrected well depth of >1075 cm'1 11for the Bowman potential. This is to be compared with the ab initio well depth of -1109 cm'1. There is clearly reasonable agreement among these two values and the 1062-cm"1 well depth of the Bowman potential. There is, however, a large discrepancy between the zero-point levels of the two surfaces that arises primarily from what appears to be an overly large contribution (~440 cm"1) from the bending vibration on the ab initio surface. This leads not only to a systematic shift in the computed vdW stretching levels compared to experiment, but to the prediction of fewer excited bending states.The fact that experimental estimates of the dissociation energy, which range from 718 to 742 cm"1, are so closely in accord with the present calculation (724 cm"1) and estimation (~739 cm"1) on the Bowman surface strongly suggests that the bending po-tential is better described by that surface. This inference is independent of the manner in which the non-vdW stretching levels are assigned. The well depth of the ab initio surface would have to be in error by enough to allow a 126-cm'1 increase in the dissociation energy. While Chakravarty and Clary10 are inclined to accept a 15% error in the ab initio well depth, one must expect such an error also to affect the bending potential.

No abstract

No abstract