Benchmarking of the R Anisotropic Dispersion Energy Term on the S22 Dimer Test Set.

The effects of including the anisotropic E term to the dispersion energy in addition to the leading E term are examined by using the effective fragment potential (EFP) method on the S22 test set. In this study, the full anisotropic E term is computed whereas the isotropic and spherical approximations are used for the E term. It is found that the E term is positive for hydrogen-bonded complexes and has a magnitude that can be as large as 50% of E, giving rise to larger intermolecular distances than those obtained with E alone. The large positive value of E is analyzed for the hydrogen-bonded uracil dimer; it is found to originate from the large magnitude of the dynamic polarizability tensors as well as the proximity of the LMOs involved in hydrogen bonding. Conversely, E tends to be negative for dispersion-dominant complexes, and it has a very small magnitude for such complexes. The optimized geometries for these systems are therefore not greatly affected by the presence of the E term. For the mixed systems in the S22 test set, an intermediate behavior is observed. Overall, the E term is most important for systems with hydrogen bonding interactions and mixed systems. A full anisotropic treatment of the E term and higher order terms may need to be included to obtain more accurate interaction energies and geometries.