Dolgounitcheva O, Díaz-Tinoco Manuel, Zakrzewski V G, Richard Ryan M, Marom Noa, Sherrill C David, Ortiz J V
Department of Chemistry and Biochemistry, Auburn University , Auburn, Alabama 36849-5312, United States.
Center for Computational Molecular Science and Techology, School of Chemistry and Biochemistry and School of Computational Science and Engineering, Georgia Institute of Technology , Atlanta, Georgia 30332-0400, United States.
J Chem Theory Comput. 2016 Feb 9;12(2):627-37. doi: 10.1021/acs.jctc.5b00872. Epub 2016 Jan 25.
Comparison of ab initio electron-propagator predictions of vertical ionization potentials and electron affinities of organic, acceptor molecules with benchmark calculations based on the basis set-extrapolated, coupled cluster single, double, and perturbative triple substitution method has enabled identification of self-energy approximations with mean, unsigned errors between 0.1 and 0.2 eV. Among the self-energy approximations that neglect off-diagonal elements in the canonical, Hartree-Fock orbital basis, the P3 method for electron affinities, and the P3+ method for ionization potentials provide the best combination of accuracy and computational efficiency. For approximations that consider the full self-energy matrix, the NR2 methods offer the best performance. The P3+ and NR2 methods successfully identify the correct symmetry label of the lowest cationic state in two cases, naphthalenedione and benzoquinone, where some other methods fail.
