Computing dielectric spectra in molecular dynamics simulations: using a cavity to disentangle self and cross correlations

TL;DR

This paper introduces a cavity-based protocol in molecular dynamics simulations to accurately compute dielectric spectra by isolating short-range correlations, reducing noise, and aligning with experimental boundary conditions.

Contribution

The paper presents a novel cavity approach for dielectric spectrum calculation that distinguishes self and cross correlations, improving accuracy and noise robustness over standard methods.

Findings

Cavity protocol matches standard dielectric spectra results.

Cavity method reduces noise sensitivity.

Self-spectrum aligns with experimental boundary conditions.

Abstract

Dielectric spectra are typically obtained in molecular dynamics (MD) simulations by analyzing the fluctuations, in the absence of an applied electric field, of the total dipole moment of the simulation box. We compare this standard method with a protocol that focuses on a virtual cavity whose size is chosen to include short-range dipolar cross-correlations, while excluding long-range correlations that are affected by the choice of electrostatic boundary conditions. We tested this protocol on three non-polarizable systems with different dielectric permittivities. We showed that it produces the same dielectric spectra as the standard method while being less sensitive to noise. The question of the decomposition of a dielectric spectrum into self and cross contributions is discussed in the context of both methods. We propose that, for a liquid with a sufficiently high dielectric permittivity, the cavity protocol yields a self-spectrum consistent with the electrostatic boundary conditions applicable to the experimental situation.