CRITIQUE OF INTERATOMIC POTENTIALS OBTAINED FROM NEUTRON-DIFFRACTION

Fredrikze et al. [Phys. Rev. Lett. 62, 2612 (1989)] constructed an interatomic potential for argon directly from a single physical property, namely, by inversion of neutron scattering data. The well depth of the potential has an assigned error of +/- 3%. To appraise its quality, we fit analytic functions to the experimental potential and use them to calculate certain spectroscopic values associated with the ultraviolet absorption spectrum of the argon dimer. These values are the vibrational and rotational constants of the five lowest vibrational levels of the ground state of Ar2, values which, together with a knowledge of the second virial coefficient and long-range forces, completely define the potential well. The predictions based on these potentials and a highly accurate multiproperty potential are then compared with the experimental data. It is found that, unlike the multiproperty potential, the single-property neutron diffraction potential cannot predict these experimental spectroscopic data within experimental error. Other properties which are significantly determined by the potential well are also analyzed based on this potential. The inability of the potential to predict these additional properties reinforces our conclusion about its inadequacy as a proper characterization of the interaction.