Atomic Charges and Electron Density Partitioning

Abstract

Aust. J. Phys., 1985, 38, 273-87 Atomic charges are derived from two dissimilar methods of partitioning the electron density of diatomic molecules. The results given by both methods are similar, with the exception of those for molecules containing lithium; factors responsible for this discrepancy are explored. The charges derived are correlated closely with electronegativity differences and with dipole moments. They follow chemically sensible trends and have reasonable magnitudes. The partitioning methods used in the derivation can also be applied to the analysis of diffraction data for crystalline solids.

A model study of the κ-refinement procedure for fitting valence electron densities

“…Maslen & Spackman (1985) have shown that other quite disparate methods of partitioning the deformation electron density yield essentially the same results. The Hirshfeld partitioning is simple to apply and yields a smooth continuous Apa(r).…”

“…It was found that, for all the diatomics studied, the atomic charges determined by numerical integration in this range were insignificantly different from those obtained by Maslen & Spackman (1985), who performed a numerical integration over r from zero to infinity by Gaussian quadrature. Values of the integrand in (6) for each of the diatomics were effectively zero at distances greater than 10a.u.…”

“…This result stems from the partitioning of the deformation density via (2) in the construction of an 'observed' valence density. Maslen & Spackman (1985) have discussed how the Hirshfeld scheme must yield charge transfers which are lower than expected for cation-anion pairs, and note that there is no partitioning scheme which can retrieve all of the expected charge transfer from the deformation density. What is important here is that, given a particular charge transfer, the K-refinement procedure is able to give a very good estimate of that charge transfer.…”

“…Kappas and charge transfers, Aqcat¢ (in electrons), from K refinement of Apa(r) spherically averaged about the nuclear position of atom a 'Observed' charge transfer values, Aqobs, are from integration of the partitioned deformation density (Maslen & Spackman, 1985), and are given only for the second atom in each diatomic. density.…”

“…This represents a compromise between the desire to use a procedure which is capable of generalization to polyatomic molecules [ruling out (i)], computationally straightforward [favouring (iii)], and as close as possible in principle to the analysis of X-ray diffraction data [favouring (i) or (ii)]. Our choice of (iii) introduces a particularly subtle bias: we are interested in modelling electrostatic properties, yet the Hirshfeld partitioning at the monopole level yields atomic charges which systematically underestimate molecular dipole moments by ~35% (Maslen & Spackman, 1985). We believe this is of minor consequence, as our aim is to analyse the monopole deformations associated with each atom, particularly those within ---2 A. of each nucleus.…”