Point charges derived from electrostatic potentials and from distributed multipole expansions are compared critically throughout a full ab initio self-consistent-field study of several saturated hydrocarbons, using the split-valence type 6-31G, 6-31G*, and 6-31G** basis sets. Potential-derived net atomic charges failed to be transferable from one molecule to another structurally similar one, especially for aliphatic groups, hence constituting a major drawback for their use in force-field parametrization. Consideration of a limited number of extra-atomic charges is necessary to reach transferability and mimic the effects of higher order moments of the electron charge distribution. However, it has been shown that another method, recently proposed and based on the distributed multipole expansion of the molecular electrostatic potential, provides a reasonable compromise for both chemical transferability and accuracy to reproduce the electrostatic potential. Nevertheless, charges obtained from both methods have a tendency to reproduce the SCF potential with conspicuous errors, therefore indicating that the molecular charge distribution of nonpolar systems cannot be represented accurately by only atom-centered point charge models.
ASJC Scopus subject areas
- Physical and Theoretical Chemistry