Vasp energy for core level binding energy calculation

[sawantk]

Hello,

I am trying to calculate core level binding energy using "ICORELEVEL = 2" tag on Vasp 5.4.1(final state approximation for valence electrons). I understand that ECL = E(nc-1) − E(nc) (Köhler and Kresse), where E(nc-1) is (Etot-E_Fermi) with the following tags:

"ICORELEVEL = 2
CLNT = 2 #POSCAR species
CLN = 2 #2
CLL = 1 #p
CLZ = 1 # exciting 1 electron"

After running the calculation I found that the total energy is more negative for the excited electron calculation than the standard calculation. I observed similar trends for data published in literature using this method. Is there a more direct explanation for the sign of the relative energies ? If vasp reports only valence electron energies, how does one reference the core energies between the two calculations ?

Thank you and regards,
Kaustubh Sawant

[ferenc_karsai]

Can you please upload your calculation with the important files according to the VASP forum guidelines.

[sawantk]

Sorry for not uploading the files, since I was more interested in the interpretation. Nevertheless, I have attached an example ZnO bulk calculation, where Zn 2p electron is excited for one atom. The total energy in OSZICAR for this calculation is -346.93 eV. While the total energy with ICORELEVEL = 0 (default) is -17.86 eV. As I understand the difference in TOTEN wrt to E_fermi for these two calculations is the energy required to to excite a core electron from the core to a conduction band. This was answered in an earlier topic : forum/viewtopic.php?t=3153.

Is there a more direct explanation for the sign of the relative energies ? If vasp reports only valence electron energies, how does one reference the core energies between the two calculations ?

Thank you and regards,
Kaustubh Sawant

[ferenc_karsai]

I talked to Georg Kresse and he cleared the matter:

Only relative energies between different structures are meaningful. For example you have a surface and could have three different possible sites. Then you would have to calculate the difference for each of them with and without a core hole. Now only these differences can be compared with each other. This is also written in the tutorial (wiki/index.php/ICORELEVEL), but not deeply explained:
"Third, absolute energies are not meaningful, since VASP usually reports valence energies only. Only relative shifts of the core electron binding energies are relevant (in some cases, the VASP total energies might become even positive). "

You should also check the literature, where this is even further explained:
L. Köhler and G. Kresse, Phys. Rev. B 70, 165405 (2004).

[sawantk]

Thank you for the explanation. As I understand, the method should be used only when all inputs are same except a binding site as shown in Phys. Rev. B 70, 165405 (2004). Secondly, this method should not be used to compare BE say between Zn 2p of ZnO and Zn metal. All Vasp users should make a note of this.

Thank you and regards,
Kaustubh Sawant