Premium
The evaluation of molecular electron affinities
Author(s) -
Baker Jon,
Nobes Ross H.,
Radom Leo
Publication year - 1986
Publication title -
journal of computational chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.907
H-Index - 188
eISSN - 1096-987X
pISSN - 0192-8651
DOI - 10.1002/jcc.540070312
Subject(s) - chemistry , electronic correlation , basis set , perturbation theory (quantum mechanics) , wave function , møller–plesset perturbation theory , residual , basis (linear algebra) , electron , atomic physics , computational chemistry , molecule , density functional theory , physics , quantum mechanics , mathematics , algorithm , geometry , organic chemistry
Abstract The performance of a variety of levels of theory in evaluating molecular electron affinities (EAs) has been systematically examined. Calculations have been carried out for six different basis sets and for nine theoretical procedures including unrestricted (UHF) and restricted (RHF) Hartree‐Fock theory, Møler‐Plesset perturbation theory (UMP2, UMP3, UMP4), configuration interaction (UCISD, RCISD, RCISD(Q)) and equations‐of‐motion (EOM) approaches. Electron affinities were evaluated for CH 3 , NH 2 , OH, F, C 2 H, CN, BO, N 3 , OCN, and NO 2 . Very poor results are generally obtained unless diffuse functions are included in the basis set and electron correlation is incorporated. Even with the largest basis set used in the present study (6‐311 + + G(2 d , 2 p )), there are still residual errors greater than 0.2 eV (UMP4) or 0.6 eV (CISD) in the absolute EAs. However, better results are obtained under certain circumstances for relative EAs. The results appear to be significantly affected by spin contamination in the UHF wave‐functions. For those systems for which spin contamination is small, best absolute values of the EAs generally come from the EOM and UMP2 calculations, whereas the most constant errors (thereby allowing systematic correction) are found at the UMP4, CISD, and RCISD(Q) levels. For the systems for which spin contamination is larger, best results are obtained with the CI‐based procedures (CISD and RCISD(Q)). The errors in calculated EAs for the molecules with differing electronic characteristics can vary quite widely. Caution must therefore be exercised before applying schemes which rely on a constancy of errors to estimate electron affinities. The UMP procedures appear particularly suspect in this regard if spin contamination is significant. The RCISD(Q) approach is recommended under such circumstances.