[Pw_forum] why I can not get the correct band gap of TiO2?

[Gerardo Ballabio]

On 10/26/2005 04:19:07 PM, #TAO JUNGUANG# wrote:
> End of band structure calculation
>           k = 0.0000 0.0000 0.0000     band energies (ev):
>    -48.0496 -48.0314 -24.6221 -24.6221 -24.5729 -24.5342 -24.4355
> -24.4355
>    -10.8942  -9.9348  -9.1108  -9.1108   1.4602   1.7866   2.9518
> 3.3254
>      3.3254   4.5197   4.5197   4.9432   5.9659   6.2973   6.2973
> 6.9316
> from the output file, I think the band gap is
> (1.4602-(-9.1108)=10.571eV),first, am I right getting band gap like
> this?

No.

You aren't getting any band gap unless you calculate at least one  
empty band. In your input "nbnd" isn't specified, thus the default  
number of bands is computed -- that is, only occupied ones.

Indeed, O has 6 valence electrons, Ti has 12 (assuming that semicore  
3s and 3p are included). You have 4 O + 2 Ti = 48 electrons in your  
cell, at 2 electrons per band, this makes exactly 24 bands.

You must specify an "nbnd" of at least 25.

Gerardo