Further benchmarks of a composite, convergent, statistically calibrated coupled-cluster-based approach for thermochemical and spectroscopic studies

Abstract

A flexible, high-level, composite approach based on coupled cluster theory has been used to predict the atomization energies and equilibrium structures of 13 small, first-row compounds. Each of the major components in this approach can be systematically improved, thereby providing a practical measure of the inherent uncertainty (or degree of convergence) in the final results. Comparison with Active Thermochemical Table data, which relies on a network of experimental and theoretical data, showed excellent agreement for the atomization energies. With the addition of the latest molecular systems to the Computational Results Database, the composite approach was found to yield a mean absolute deviation of 0.19 kcal mol−1 and a root-mean-square deviation of 0.31 kcal mol−1 across 142 comparisons. If the analysis is limited to experimental data with estimated uncertainties of 0.2 kcal mol−1 or less, the error metrics are cut in half. Similar good agreement is found with experimental structures, but the relative scarcity of accurate equilibrium structures limits the significance of the statistical analysis. Unavoidably, many of the comparisons could not be made with r e structural parameters. Explicitly correlated methods were found to be effective at replicating results obtained from the standard method with large basis sets, thereby reducing the high computational cost for several of the components.