Predicting Solid-Liquid Interfaces with Joint Density Functional Theory (JDFT)

Modeling the electrochemical interface offers a significant challenge due to the complexity of the liquid and the long length and time scales. Joint Density Functional Theory (JDFT), which is implemented in the open source software JDFTx, combines a classical continuum liquid model with quantum-mechanical density functional theory for an accurate and efficient description of solvent-solute interactions. While JDFT is a promising approach for describing the electrochemical interface, additional benchmarking has been required to demonstrate the robustness of the approximate functionals and determine areas of improvement to be addressed by development of novel functionals. Here we further test the accuracy of a universal solvent-electron coupling functional which was trained on three different solvents and multiple solute molecules to capture nonlocal effects such as van der Waals interactions. By computing the solvation energies of 240 solutes in water using both the universal and solvent-specific coupling functionals, we demonstrate that near-chemical accuracy is maintained for the test set as well as the training set. Furthermore, we explore the accuracy of JDFT for water at interfaces relative to X-ray reflectivity measurements and molecular dynamics simulations.