This paper results from the 3-month visit of PhD student Jonas Mikkelsen at the Hylleraas Centre in Oslo in the fall of 2024.
Conclusion of the paper
The ability to calculate atom-atom polarizabilities at different levels of theory is of interest for parameterizing force fields as well as for analyzing chemical bonding. Using high-level coupled-cluster and MCSCF response theory, our results of roughly equal contribution of charge-flow and induced dipole components to the molecular dipole polarizability are at variance with the prevailing approach of modelling electric polarization in force fields entirely in terms of either charge-flow or induced dipoles.
Furthermore, charge-flow models often assume that only contributions from directly bonded atoms are non-zero, but the present work shows that the contribution from atom pairs separated by more than one bond may be greater than for directly bonded atom pairs. A direct transfer of the calculated rank-rank atom-atom polarizability parameters into a force field environment thus requires a new framework.
The present work indicates that the charge-flow component to non-nuclear sites for modelling out-of-plane polarization is insignificant compared to the induced dipole term. The calculated rank-rank atom polarizabilities are geometry dependent and the methods developed in the present work allows probing the dependence, including the behavior for bond dissociation. While the dependence on geometry may be negligible for simulating molecules near their equilibrium geometry, it will be essential for developing reactive force fields and for modelling spectroscopic properties.
Suitable parameterization of the geometry dependence requires access to accurate reference values, as can be provided by the methods developed in the present work. The ability to calculate frequency dependent atom-atom polarizabilities furthermore provides access to calculate atom-atom dispersion parameters.
Published: 10th July 2026;