Core-level calculation in charged supercell

[ashwin_r]

I have a supercell (metal oxide; 3D bulk) that has been fully optimized with an oxygen vacancy in a 2+ charge state. More precisely, I created an O vacancy, removed 2 electrons from the supercell and relaxed the entire structure. I now want to see how this vacancy modifies the core-level spectra of the neighboring metal atoms.

My question is whether running a core-level calculation in the final-state approximation (ICORELEVEL = 2) is compatible with charged supercells. The VASP documentation mentions that the core electron is moved to the valence, effectively increasing NELECT by 1 (or CLZ), which would require a neutralizing background. Given that the supercell already requires a neutralizing background from the outset, does the core-level calculation take this into account properly, i.e., use a net neutralizing background that effectively compensates 1 missing electron (2 electrons originally removed plus the 1 electron added from core level)?

I don't expect that the initial-state approximation should pose any such issues but please let me know if that is also potentially problematic for charged supercells.

[marie-therese.huebsch]

Hi, thank you for your question. This is certainly good to clarify. I will investigate the issue and ask a team member that knows the method better than me. Once I have some insight, I will let you know.

Best regards,
Marie-Therese

[marie-therese.huebsch]

Yes, core-level calculations are fully compatible with a charged supercell. You just have to be aware of how NELECT and the compensating background interact, because the NELECT set in INCAR takes precedence.

The usual input for a final-state (ICORELEVEL=2) calculation is:

CLNT = 1 # species index
CLN = 1 # main quantum number n of the core electron
CLL = 0 # l quantum number of the core electron
CLZ = 1 # number of electrons excited from the core
By default, VASP takes the core electron out of the core according to CLZ and adds the corresponding charge to the neutralizing background and it automatically increases NELECT by CLZ. In an otherwise neutral cell you don't need to touch NELECT at all.

The important subtlety for your case: NELECT in the INCAR takes precedence over this automatic adjustment. So in a charged cell, where you are already setting NELECT by hand for the 2+ state, you must set NELECT explicitly in every step yourself. VASP will not add the +1 on top of your manual value.

Best regards,
Marie-Therese