Self-consistent solution of Dyson's equation up to second order for closed- and open-shell atomic systems

K. Peirs, D. Van Neck, M. Waroquier
Journal of Chemical Physics
117(9), 4095-4105
2002
A1

Abstract 

Green’s function techniques are powerful tools for studying interacting many-fermion systems in a structural and diagrammatical way. The central equation in this method is the Dyson equation which determines, through an approximation for the self-energy, the Green’s function of the system. In a previous paper [J. Chem. Phys. 115, 15 (2001)] a self-consistent solution scheme of the Dyson equation up to second order in the interaction, the Dyson(2) scheme, has been presented for closed-shell atoms. In this context, self-consistency means that the electron propagators appearing in a conserving approximation for the self-energy are the same as the solutions of the Dyson equation, i.e., they are fully dressed. In the present paper this scheme is extended to open-shell atoms. The extension is not trivial, due to the loss of spherical symmetry as a result of the partially occupied shells, but can be simplified by applying an appropriate angular averaging procedure. The scheme is validated by studying the second-row atomic systems B, C, N, O, and F. Results for the total binding energy, ionization energy and single-particle levels are discussed in detail and compared with other computational tools and with experiment. In open-valence-shell atoms a new quantity—the electron affinity—appears which was not relevant in closed-shell atoms. The electron affinities are very sensitive to the treatment of electron correlations, and their theoretical estimate is a stringent test for the adequacy of the applied scheme. The theoretical predictions are in good agreement with experiment. Also, the Dyson(2) scheme confirms the nonexistence of a stable negative ion of N. The overall effect of the self-consistent Dyson(2) scheme with regard to the Dyson(1) (i.e., Hartree–Fock) concept, is a systematic shift of all quantities, bringing them closer to the experimental values. The second-order effects turn out to be indispensable for a reasonable reproduction of the electron affinity. © 2002 American Institute of Physics.