The (H2)2 potential surface and the interaction between hydrogen molecules at low temperatures

Abstract
The (H2)2 rigid rotor interaction potential has been calculated for the intermolecular distance range R = 3.0–11.0 a.u. for six relative orientations to estimate both the isotropic and anisotropic components of the full intermolecular potential. A partially optimized basis set limited in size to 78 independent Gaussian functions was used throughout the energy calculations, which required only very small corrections for basis set unsaturation effects. Correlation effects were computed both at the variational (single and) double excitation PNOCI level and using the CEPA2‐PNO approximation to estimate higher order excitation effects. While the latter rigidrotor surface may overestimate the strength of the H2–H2 interaction in the vicinity of the well by a few wave numbers in the rigidrotor PNOCI surface from the present study, which we regard as an upper bound to the true rigidrotor surface, is also slightly deeper than almost all previous theoretical and empirical ’’fit’’ potentials in the well region. Since low energy scattering experiments lack sensitivity to the precise well depth because of the long de Broglie wavelengths appropriate to scattering at energies comparable with the well depth, the significance of the current disagreement on the well depth between the rigid‐rotor level of the theory (which puts it at least 27.5 cm−1) and experiment is uncertain. However, coupled with the present uncertainty stemming from the CI formalism of ∼±3 cm−1, is a further uncertainty due to the expected increase in intramolecular zero‐point energies due to hindered H2 vibration within the dimer. A zero‐point increase within each H2 at Re of 0.1% would effectively raise the rigid‐rotor potentials by some 4 cm−1, bringing the well depth into line with experimental estimates. In the low energy part of the repulsive wall of the potential (where the relative importance of these small effects diminishes), which has been probed rather sensitively by recent rotationally inelastic scattering experiments and which strongly influences solid H2 properties, we find that both the present PNOCI and CEPA2 surfaces lie in between the respective semiempirical fit surfaces for the H2–H2 ’’pair’’ interaction. At the zero crossing of the isotropic component of the potential, we find excellent agreement between the present theoretical estimates of R0(2.99±0.03 Å) and the most recent experimental evaluation of the parameter (2.98±0.03 Å) by Buck et al. [J. Chem. Phys. 74, 535 (1981)].