Chemical Bonds in Molecules Are Covalent but Not Ionic

Abstract We report a comprehensive bonding analysis of the nonpolar bonds H–H, Li–Li, and F–F and the polar bonds Li–H, H–F, and Li–F as well as the isoelectronic dications HeNe2+, BeHe2+ and BeNe2+. We also analyze the chemical bonds in the heavier halogen systems EF and ECl (E = Na–Cs). It is shown that the bonding interactions in all neutral molecules in the electronic ground state originate from the constructive interference (resonance) of the wave functions of the neutral fragments, resulting in the formation of covalent bonds of different polarities. Methods of bonding analysis that only describe the finally formed bonds provide interesting information about the electronic structure of the molecules, but do not reveal insight into the inter- and intra-atomic mechanisms that lead to these bonds. The fact that a covalent interaction, caused by the quantum interference, has evolved to be polar should not be mistakenly considered as an argument for classifying it as caused by classical electrostatic ionic interaction. The latter applies to some excited states at large internuclear distances in vacuum, or in ionic solutions when the bare ions are stabilized by solvent molecules, or in ionic solids when the electrostatic interactions are enhanced in crystal lattices by the Madelung factor. But the chemical bond in molecules in their electronic ground state comes from the constructive interference of the wave functions when they are occupied by at least one electron. This also holds true for highly polarized bonds such as Cs–F. A second electron enhances, in most cases but not always, the bonding interactions. The electron pair suggested by Lewis is a useful model for a chemical bond, but the formation of an electron pair is not the physical driving force for a chemical bond. A detailed examination of the real-space presentation is made, with particular emphasis on the understanding of reality in quantum theory. The results of the QTAIM methods are critically examined, concluding that the Laplacian distribution ∇2ρ(r) is not a reliable feature for the formation of chemical bonds in polar molecules. Critical remarks are also made on some representations of the Valence Bond method.