New compact density matrix averaged ANO basis sets for relativistic calculations.

Abstract

When including relativistic effects in quantum chemical calculations, basis sets optimized for relativistic Hamiltonians such as the atomic natural orbital-relativistic core-correlated (ANO-RCC) basis set have to be used to avoid large errors that appear upon contraction of the basis set. While the large size of the ANO-RCC basis set in terms of primitive basis functions allows for highly accurate calculations, it also hinders its applicability to large sized systems due to the computational costs. To tackle this problem, a new compact relativistic ANO basis set, the ANO-eXtra Small (XS) basis set, is introduced for elements H-Ca. The number of primitive basis functions in ANO-XS is about half that of the ANO-RCC basis set. This greatly reduces the computational costs in the integral calculations especially when used in combination with Cholesky decomposition. At the same time, the ANO-XS basis set is able to predict molecular properties such as bond lengths and excitation energies with reasonable errors compared to the larger ANO-RCC basis set. The main intention for the ANO-XS basis set is to be used in conjunction with the ANO-RCC basis set for large systems that can be divided with regions demanding different qualities of basis sets. This is exemplified in CASPT2 calculations for an Ir(C3H4N)3 complex, where substituting the larger ANO-RCC for the compact ANO-XS basis set at the ligand atoms yields only minor differences for a large number of excited states compared to calculations employing the ANO-RCC basis set on all atoms. Thus, accurate calculations including relativistic effects for large systems become more affordable with the new ANO-XS basis set.