On the reliability of atoms in molecules, noncovalent index, and natural bond orbital to identify and quantify noncovalent bonds.

Abstract

Atoms in molecules, noncovalent index, and natural bond orbital methods are commonly invoked to identify the presence of various noncovalent bonds and to measure their strength. However, there are numerous instances in the literature where these methods provide contradictory or apparently erroneous interpretations of the bonding. The range of reliability of these methods is assessed by calculations of a variety of systems, which include an H-bond, halogen bond, π-tetrel bond, CH··HC interaction, and a pairing of two anions. While the results appear to be meaningful for the equilibrium geometries, and those where the two subunits are progressively pulled apart, these techniques erroneously predict a progressively stronger bonding interaction as the two units are compressed and the interaction becomes clearly repulsive. The methods falsely indicate a bonding interaction in the CH··HC arrangement, and incorrectly mimic the behavior of the energy when two anions approach. These approaches are also unreliable for understanding angular deformations.