[Pw_forum] Determining Bandgap with DOS

[Guido Fratesi]

Dear Izaak,

a few preliminary questions you might find helpful answering before 
going on:

- have you used a sufficient number of k points?

- have you used a sufficiently dense energy grid in the plot?

- how is it possible that the curve "without smearing" is more smeared 
than the one "with smearing"? Check what default values are.

- why do you indicate by arrows not the state found in the DOS, but the 
energy point at which your gaussian (of arbitrary width indeed) looks 
different from zero?

- why not looking at eigenvalues printed by pw.x on output to get the 
real (computed) numbers? (Yet the DOS plot is helpful to get an overview)

Hope this helps.
Guido


On 05/17/2012 11:43 PM, Izaak Williamson wrote:
> Hello all,
>
> I am trying to calculate the electronic bandgap for wurtzite ZnO using
> the attached plot of DoS (E), where the Fermi level is represented by a
> vertical
> dashed-line and EV represents the valence band maximum.
>
> My question is, should I use EC1 or EC2 as my conduction band minimum?
> If I use EC1, the bandgap is ~0.5eV but if I consider those states at
> that point negligible and use EC2 then the bandgap is ~1.5eV. (NOTE:
> the experimental bandgap is 3.36eV.)
>
> Thanks for the help.

-- 
Guido Fratesi

Dipartimento di Scienza dei Materiali
Universita` degli Studi di Milano-Bicocca
via Cozzi 53, 20125 Milano, Italy