Performance of Some Density Functional Methods in the Prediction of the Adiabatic Electron Affinities

Ding Xiu-Li,Wu Jian-Ming,Xu Xin
DOI: https://doi.org/10.3321/j.issn:0251-0790.2008.02.044
2008-01-01
Abstract:In this paper, we examined the performance of 12 density functionals(B3LYP, X3LYP, 03LYP, PBE0, B3PW91, BLYP, OLYP, OPBE, PBE, BPW91, VSXC and TPSS), combined with two basis sets DZP + + and 6-311 + G (3df, 2p)] in the prediction of the adiabatic electron affinities against a set of 91 well-established experimental values. We have reached a conclusion that all these modern functionals are generally good, giving a mean absolute deviation ( MAD) less than 0.24 eV. B3LYP and X3LYP are among the most reliable methods for this test set. When the DZP + + basis set is adopted, X3LYP is the best with a MAD of 0.14 eV, while B3LYP leads to a MAD of 0.16 eV. When the basis set is extended to 6-311 + G (3df, 2p), B3LYP improves its MAD to 0.13 eV, while the MAD of X3LYP remains to be 0.14 eV. Most hybrid functionals outperform the corresponding GGA functinals, with the exception of PBEO. The meta-GGA functionals (VSXC and TPSS) are not necessarily better than the GGA functionals.. Our calculations also reveal that there are flaws associated with the OPTX exchange functional in the calculations of EA.
What problem does this paper attempt to address?