The ABINIT Users Mailing List ( CLOSED )

[Fabien Bruneval <fabien.bruneval@cea.fr>]

Dear Manuel,

The effect of the *eigenvalue* of a semicore state onto the GW band gap 
is usually negligible.

1) The exchange part does not contain any reference to the eigenvalues.

2) The correlation part contains some denominators like 1/(E_valence - 
E_semicore + omega_plasma).
If the difference of eigenvalues E_valence - E_semicore is large, the 
correction arising from these terms will be vanishing.
Note that the important quantity is the relative position of the 
semicore states with respect to the band gap region.

Concerning ZnO, did you check the papers from Kotani and van Schilfgaarde?

Cheers,

Fabien



Manuel Pérez Jigato wrote:
> Dear all,
>
> i would like to ask you whether at the time of computing GW 
> quasiparticle band-gaps in semiconductors-insulators that contain 
> elements with d-electrons,
> the quantitative effect of deep semicore level band energies is 
> important.
> Seemingly, wavefunctions are very important for the exchange 
> self-energy, at
> the time of computing band gaps, but i have no idea on the exact effect
> of eigenenergies.
>
> The reason for me to ask is that for wurtzite-ZnO, i get a change
> in LDA gamma-point band energy of 5.7 eV, when two different Zn 
> pseudopotentials
> are employed (both include 3s 3p 3d 4s 4p as valence), and still i 
> cannot find
> full-potential all-electron tabulated published band energies for the 
> Zn 3s levels (usually people stop at -20 eV, and these are at ca -120 
> eV).
> One of the pseudos is Goedecker, and the other is a troullier-martins 
> pseudo.
>
> thanks
>
> Manuel Perez Jigato
> KU Leuven
>
>
> Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm