Density functional theory for reaction energies: Test of meta and hybrid meta functionals, range-separated functionals, and other high-performance functionals

Research output: Contribution to journalArticlepeer-review

148 Scopus citations

Abstract

The present study compares the accuracy of 30 density functionals for four databases of reaction energies studied recently by Grimme and co-workers. For 20 of the density functionals, the calculations are new, and the calculations are compared to previous work for the other 10. We present the results in detail for 11 of the functionals and as mean unsigned errors for the others. The results presented in detail are for the seven most recent Minnesota functionals (M05-2X, M06-L, M06-HF, M06, M06-2X, M08-HX, and M08-SO), three range-separated functionals (HSE, LC-ωPBE, and ωB97X-D), and one dispersion-corrected global hybrid generalized gradient approximation (B97-D); the other functionals include five dispersion-corrected functionals and their uncorrected analogs, eight high-performing functionals on a recent catalytic-energies test, and the TPSSh functional because it is of special interest to compare its performance to that of M08-SO. Three of the four databases contain a total of 21 rearrangement reaction energies and 13 diverse dissociation or association energies, and the fourth contains three dissociation reaction energies of alkali metal clusters and three dissociation reaction energies of alkali-metal-cation-benzene complexes. The results are especially promising for the Minnesota hybrid meta-GGA functionals and the ωB97X-D, B2PLYP-D, and HSE functionals.

Original languageEnglish (US)
Pages (from-to)669-676
Number of pages8
JournalJournal of Chemical Theory and Computation
Volume7
Issue number3
DOIs
StatePublished - Mar 8 2011

Fingerprint Dive into the research topics of 'Density functional theory for reaction energies: Test of meta and hybrid meta functionals, range-separated functionals, and other high-performance functionals'. Together they form a unique fingerprint.

Cite this