Abstract
Are the vertical ionization energies from a bound electronic system, initially in its
ground state, equal to minus the corresponding exact Kohn-Sham orbital energies of
density functional theory (DFT)? This is known to be true for the first or lowest vertical
ionization energy. We show that the correction from time-dependent DFT arises from
the continuum and need not vanish. Recent work compared the experimental photoemission
thresholds of the molecules Cu2O−, CuO−, CuO−
2 , and CuO3− with minus
the corresponding orbital energies from a generalized gradient approximation (GGA)
and its global and range-separated hybrids with exact exchange, finding striking differences
which were attributed to self-interaction error, strong correlation, or both.
Here we extend that work to include the local spin density approximation (LSDA), its
Perdew-Zunger self-interaction correction with Fermi-L¨owdin localized orbitals (LSDASIC),
a quasi-self-consistent locally scaled-down version of LSDA-SIC (QLSIC), and the Quantum Theory Project QTP02 range-separated hybrid functional, all but LSDA implemented
in a generalized Kohn-Sham approach. QTP02 impressively yields a near
equality for many sp-bonded molecules. But, for the copper oxide anions studied here,
none of the tested methods reproduces the experimental photoemission thresholds.