Automatic generation of density fitting auxiliary basis sets for all electron Dirac-Kohn-Sham calculations

TL;DR

This paper introduces an automated workflow for generating auxiliary density basis sets applicable to all elements, enhancing the accuracy of relativistic Dirac-Kohn-Sham calculations for molecules with heavy elements.

Contribution

The study presents a novel automated algorithm for creating auxiliary basis sets for all elements, specifically tailored for relativistic calculations including spin-orbit effects.

Findings

Achieved high accuracy with Coulomb energy errors of a few micro-hartree.

Validated the basis sets on a large molecular dataset covering all periodic table groups.

Established a general workflow for future optimization including exact exchange.

Abstract

In this study, we present a general workflow that enables the automatic generation of auxiliary density basis sets for all elements of the periodic table (from H to Og) to facilitate the general applicability of relativistic Dirac-Kohn-Sham calculations. It is an important tool for the accurate description of relativistic effects, including spin-orbit coupling, in molecules containing heavy elements. The latter are very important in various fields ranging from catalysis to quantum technologies. The automatic generation algorithm is based on an even-tempered scheme inspired by a previous work by P. Calamicini et al. J. Chem. Phys. 2007, 126, 7077, in which the auxiliary basis sets were generated for non-relativistic DFT calculations within the GGA approximation. Here, the algorithm uses basic information from the principal relativistic spinor basis set (exponents and angular momentum values) and includes a simple strategy to account for the high angular momentum of electrons in heavy and superheavy elements. The workflow developed here allows us to perform extensive automated tests aimed at verifying the accuracy of the auxiliary basis sets in a large molecular data set of about 300 molecules representing all groups and periods of the periodic table. The results show that our auxiliary basis sets achieve high accuracy, with errors in the Coulomb energies of a few {\mu}-hartree, which are of the same order of magnitude as in the non-relativistic density fitting. The automatic workflow developed here is general and will be applied in the future for optimization of auxiliary basis sets to include of exact exchange to relativistic approaches. The latter will be a crucial step for the accurate description of spectroscopic properties and the spin-dynamics in molecular systems containing heavy elements.