Lessons Learned on Obtaining Reliable Conductivity Estimates From Molecular Dynamics Simulations

The calculation of reliable ionic conductivities from molecular dynamics simulations is not a straightforward task, especially for strongly correlated systems, such as ionic liquids or highly concentrated electrolytes, where the Nernst-Einstein approach tends to fail. In this manuscript, we present the newly implemented conduct module for TRAVIS. It allows the calculation of the ionic conductivity using the Einstein-Helfand and Green-Kubo approaches, which explicitly include ionic correlations in their formalism. We provide a broad overview of accessible transport properties and compare methods and best practices for obtaining statistically reliable estimates of ionic conductivity and other physicochemical properties derived from electrolyte molecular dynamics simulations, including transport numbers and the inverse Haven ratio. To validate our implementation and demonstrate the conduct module's capabilities, we simulated the ionic liquid 1-ethyl-3-methylimidazolium dicyanamide ([EMIm][DCA]) as well as the ether-based electrolyte lithium bis(fluorosulfonyl)imide in ethylene glycol dimethyl ether (LiFSI/DME).