Fermi energy not in midgap

[ashish dabral]

Hi,

I have calculated PDOS for semiconducting systems. While plotting dos I realized that for the semiconductor, 

the fermi energy is not lying in the midgap. No spin polarization is used. I just take the fermi energy and 

check if the mid of CB min. and VB maximum give me the fermi energy(mid-gap). There is a slight offset towards

conduction band. The system is definitely semiconducting. I can calculate the fermi energy by taking mid value

from CB and VB but why is fermi energy from cp2k  calculation is not matching? Only Gamma point is used as

the systems are large (100-200 atoms)

Any pointers?

Thanks

[Matt W]

Hi,

can you define 'slight offset'? The Fermi level is determined numerically, so it is bound to not be precisely at the centre of the gap.

(Curiosity - should the Fermi level be precisely mid gap for a real material, or is the result just for simple 1D bands and no degeneracy?)

Matt

[ashish dabral]

I have monolayers of various materials which have been simulated multiple times and with different software.

There is a vacuum in z direction 15 Ang. PBE, GGA is used.

Let's take the below scenario where the different levels are given in eV .

VBM     = -3.88        Valence band

CBM     = -2.255     conduction band

Ef         = -1.44       Fermi energy from cp2k output

Ef         = -2.66       (VBM + (CBM-VBM)/2) mid-gap

The fermi level is in the bandgap.

The error in the two fermi levels above :

delta Ef = 0.40

All my systems are monolayers but I have simulated these monolayers earlier using supercells. I know just using Gamma points doesn't give

a good resolution but still there shouldn't be 0.4 eV difference. Would you suggest just taking an average of VBM and CBM in that case.

VBM and CBM are identified by filling total electrons with 2 per level. so basically nelectrons/2 th energy level is my valence band and the next

one with 0 occupation my conduction band.

Thanks for any suggestions.