On the performance of correlation consistent basis-sets for the calculation of total atomization energies, geometries, and harmonic frequencies

Abstract

The total atomization energies (Sigma D-e values), geometries, and harmonic frequencies for a number of experimentally well-described molecules have been calculated at the CCSD(T) (coupled cluster) level using Dunning's correlation-consistent cc-pVDZ([3s2p1d]), cc-pVTZ([4s3p2d1f]), and cc-pVQZ([5s4p3d2f1g]) basis sets. Additivity correction are proposed for binding energies and geometries. Using a three-term additive correction of the form proposed by Martin [J. Chem. Phys. 97, 5012 (1992)] mean absolute errors in Sigma D-e are 0.46 kcal/mol for the cc-pVQZ, 0.93 for the cc-pVTZ, and 2.59 for the c-pVDZ basis sets. The latter figure implies that, although unsuitable for quantitatively accurate work, three-term corrected CCSD(T)/cc-pVDZ binding energies can still be used for a rough estimate when the cost of larger basis set calculations would be prohibitive. CCSD(T)/cc-pVQZ calculations reproduce bond lengths to 0.001 Angstrom for single bonds, and 0.003 Angstrom for multiple bonds; remaining error is probably partly due to core-core and core-valence correlation. CCSD(T)/cc-pVTZ calculations result in additional overestimates of 0.001 Angstrom for single, 0.003 Angstrom for double, and 0.004 Angstrom for triple bonds. CCSD(T)/cc-pVDZ calculations result in further overestimates of 0.01 Angstrom for single bonds, and 0.02 Angstrom for multiple bonds. CCSD(T)/cc-pVDZ harmonic frequencies are in surprisingly good agreement with experiment, except for pathological cases like the umbrella mode in NH3. Both CCSD(T)/cc-pVTZ and CCSD(T)/cc-pVQZ harmonic frequencies generally agree with experiment to 10 cm(-1) or better; performance of cc-pVQZ is somewhat superior on multiple bonds or the umbrella mode in NH3. Again, a source of remaining error appears to be core correlation. The use of MP2/6-31G* reference geometries in the Sigma D-e calculation can result in fairly substantial errors in the uncorrected Sigma D-e values fbr systems with cumulated multiple bonds. These errors however appear to be largely absorbed by the three-term correction. Use of CCSD(T)/cc-pVDZ reference geometries appears to have no detrimental effect on computed Sigma D-e values and is recommended for cases where only single-point calculations in the cc-pVTZ basis set are possible.