E-fermi for band structure

[user]

Dear All,

I am wondering how VASP sets the Fermi level (E-fermi) in the band structure calculations.

For semiconductors, it appears to me that VASP sets the Fermi level in the middle of the band gap; not at the valence-band top.

For semimetals and metals, does VASP set the Fermi level at the highest occupied state(s)? If so, does it depend on the way we choose different high symmetry directions in the BZ?

In all cases: Since we do not scan the whole BZ, how precise is the position of the Fermi level given by VASP (in the non-selfconsistent run to get band structure)?

I would be very glad if someone helps me clarify this issue.

Thanks.

[admin]

vasp sets the Fermi level on the highest occupied Kohn-Sham state, therefore, you are right, it may be inaccurate if the k-mesh does not include the k-point at which the VBM lies.
if there is a gap, EF may be slightly higher than the VBM, if SIGMA is chosen very large.
Please note however that for all systems using 3D periodicty the zero-level of the E-scale is only given within an additive constant, therefore the absolute values of the eigenvalues (and of course EF) have no physical meaning.
If you do a slab calculation, the E=0 level can be set to the constant vacuum level, and all energies can be taken with respect to this level..

[user]

Thanks, Admin. That's very helpful.