# Resolving Dispersion and Induction Components for Polarisable Molecular Simulations of Ionic Liquids

## TL;DR

This paper develops a method to incorporate polarisation into molecular simulations of ionic liquids by adjusting Lennard-Jones parameters, resulting in more accurate predictions of transport properties and densities.

## Contribution

It introduces a strategy to adapt existing non-polarisable force fields for polarisation by subtracting a fraction of van der Waals energy using SAPT, improving simulation accuracy.

## Key findings

- Adding Drude dipoles enhances transport property predictions.
- Scaling Lennard-Jones interactions speeds up dynamics and matches experimental densities.
- Polarisation weakens spatial correlations beyond the first shell.

## Abstract

One important development in interaction potential models, or atomistic force fields, for molecular simulation is the inclusion of explicit polarisation, which represents the induction effects of charged or polar molecules on polarisable electron clouds. Polarisation can be included through fluctuating charges, induced multipoles or Drude dipoles. This work uses Drude dipoles and is focused on room-temperature ionic liquids, for which fixed-charge models predict too slow dynamics. The aim of this study is to devise a strategy to adapt existing non-polarisable force fields upon addition of polarisation, because induction was already contained to an extent, implicitly, due to parametrisation against empirical data. Therefore, a fraction of the van der Waals interaction energy should be subtracted so that the Lennard-Jones terms only account for dispersion and the Drude dipoles for induction. Symmetry-adapted perturbation theory (SAPT) is used to resolve the dispersion and induction terms in dimers and to calculate scaling factors to reduce the Lennard-Jones terms from the non-polarisable model. Simply adding Drude dipoles to an existing fixed-charge model already improves the prediction of transport properties, increasing diffusion coefficients and lowering the viscosity. Scaling down the Lennard-Jones terms leads to still faster dynamics and to densities that match experiment extremely well. The concept developed here improves the prediction of density and transport properties and can be adapted to other models and systems. The inclusion of polarisation and the down-scaling of Lennard-Jones terms affects onyl slightly the ordering of the first shell of counterions, leading to small decreases in coordination numbers. The effect of polarisation is major beyond first neighbours, weakening spatial correlations, a structural effect related to the faster dynamics of polarisable models.