Fermi level in VASP6

[duccoi]

Hi, I do a spin-polarized test of a LiCoO2 defect structure with a Li vacancy in VASP version 5.4 and 6. The results obtained from VASP5.4 show that the Fermi level is at the highest occupied states of the valence band. However, based on the result estimated from VASP6, the Fermi level for the defect structure is at the bottom of the conduction band. Is it changed?

[marie-therese.huebsch]

Hi, the situation you observe is the consequence of a bisection algorithm. In the current implementation (VASP 6.3.0) a gaped system may yield a Fermi energy at the top of the highest occupied band or at the bottom of the unoccupied bands. In principle, one can argue the Fermi level should be in the center of the band gap and this will be implemented in a future version.

See these related posts:
Fermi level with tetrahedral integration [v6.1.0]
Information requested on tetrahedral integration bug fix in v6.1.0

Cheers,
Marie-Therese

[duccoi]

Dear Marie-Therese,
Thank you for your information.
I agree that the Fermi level can be set in the gap from the top of the highest occupied band to the bottom of the unoccupied bands. However, for some defects or doped structures, the Fermi-level shifting can provide information about the change of chemical potentials. In my opinion, it is more convenient to set the Fermi level at the top of the highest occupied band for all calculations; or at the bottom of the unoccupied bands for all calculations. The inconsistent situation, which in some cases, it is at the highest occupied band, and in some other cases, it is at the bottom of the conduction band, can make users confused.
Best regards!