Slaters beat Gaussians for Fe-complexes

In a recent paper, J. Phys. Chem. A, 112 (28), p. 6384, 2008, Güell, Luis, Sola, and Swart, compare ADF’s Slater Type Orbital (STO) basis sets to Gaussian Type Orbitals (GTO) basis sets in a systematic study on spin-state energies of iron complexes. Their findings are summarized in the graph below.

Slow convergence with Gaussian basis sets

The abstract states that “STO basis sets give consistent and rapidly converging results, while the convergence with respect to basis set size is much slower for the GTO basis sets.” They conjecture that this may be related to “the cusps at the nuclei (correctly described in STO, incorrectly described in GTO)”. The GTO basis sets of standard size in particular have problems with the high-spin state.

A similar conclusion was drawn for reaction and activation energies by transition metal catalysts in a 2013 paper by Kazaryan & Baerends: “The smaller GTO basis sets give really large discrepancies compared to the converged results. It is worth noting that the STO basis set convergence is significantly faster.”

No convergence with ECP basis sets

In case GTO basis sets are combined with Effective Core Potentials (ECPs), the results worsen in the sense that the common basis set limit for STO’s and GTO’s does not seem to be reached even for quite large ECP basis sets.

Comparison of Slater and Gaussian basis set performance for spin-state energies in Fe-compounds