Performance of SCAN density functional for a set of ionic liquid ion pairs

TL;DR

This study evaluates the accuracy of the SCAN density functional in predicting the electronic structure of ionic liquid ion pairs, demonstrating its near-CCSD(T) level performance at a lower computational cost.

Contribution

It provides a comprehensive assessment of the SCAN functional's effectiveness for ionic liquids, comparing it with other popular functionals and high-level coupled-cluster methods.

Findings

SCAN closely matches coupled-cluster results for ion pairs

SCAN offers a cost-effective alternative to high-level methods

Other functionals show varying degrees of accuracy

Abstract

Computational chemistry is a powerful tool for the discovery of novel materials. In particular, it is used to simulate ionic liquids in search of electrolytes for electrochemical applications. Herein, the choice of the computational method is not trivial, as it has to be both efficient and accurate. Density functional theory (DFT) methods with appropriate corrections for the systematic weaknesses can give precision close to that of the post-Hartree--Fock coupled cluster methods with a fraction of their cost. Thence, we have evaluated the performance of a recently developed non-empirical Strongly Constrained and Appropriately Normed (SCAN) density functional on electronic structure calculations of ionic liquid ion pairs. The performance of SCAN and other popular functionals (PBE, M06-L, B2PLYP) among with Grimme's dispersion correction and Boys-Bernardi basis set superposition error correction was compared to DLPNO-CCSD(T)/CBS. We show that SCAN reproduces coupled-cluster results for describing the employed dataset of 48 ion pairs.