Interaction energies in noncovalent intermolecular complexes are strongly affected by electron correlation: for example, London dispersion forces are completely absent in an uncorrelated treatment. As a result, the most accurate methods to compute these energies, such as supermolecular coupled-cluster (CCSD(T)) approaches or symmetry-adapted perturbation theory (SAPT), suffer from slow convergence to the complete basis set (CBS) limit. The basis set convergence of interaction energies can be improved by an explicitly correlated (F12) algorithm as well as an addition of basis functions centered on the intermolecular bond. However, the multitude of possible combinations of the atom-centered basis set, the set of bond functions, and the specific F12 approach makes the selection of an optimal benchmark algorithm challenging. Using accurate CBS data for a selection of representative noncovalent complexes, we investigate (1) the optimal combinations of methods and basis sets for a precise recovery of the "gold-standard" CCSD(T)/CBS limit, (2) the best "silver-standard" variants providing highest accuracy when only double-zeta basis sets are feasible at the CCSD(T) level, and (3) the most precise CBS estimates of the leading-order dispersion and exchange-dispersion energies in SAPT using a recently developed F12 algorithm, with and without bond functions.

Converging noncovalent interaction energies to the basis set limit: Bond functions, explicit correlation, or both?

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.