The Impact of Larger Basis Sets and Explicitly Correlated Coupled Cluster Theory on the Feller–Peterson–Dixon Composite Method

Abstract

Abstract The degree of completeness in the one-particle expansion, which in most standard orbital–based electronic structure methods involves atom-centered Gaussian basis functions, is central to the prediction of accurate atomic and molecular properties. While there are hundreds of basis sets in the literature capable of yielding low-to-medium quality results, high accuracy requires the use of a convergent and computationally efficient sequence of basis sets capable of systematically addressing the painfully slow one-particle expansion. This issue will be discussed in the context of the Feller–Peterson–Dixon composite method which is intended for well-converged thermochemical and spectroscopic properties across a large portion of the periodic table. Two strategies, one involving very large basis sets and the other involving newly developed, explicitly correlated methods, will be contrasted. Comparison with well-established experimental data covering hundreds of chemical systems (neutral molecules, anions, and cations in both ground and excited states) demonstrates that this admittedly computationally expensive approach is capable of achieving root–mean–square deviation agreement for atomization energies, electron affinities, and ionization potentials within ±0.5 kcal/mol and maximum errors within ±1 kcal/mol.