Dear Rajdeep Banerjee,
Some smearing schemes lead to negative occupation numbers. If you are
using this only for the DOS in a semi-conductor, you could as well use the
Fermi-Dirac broadening scheme (or Gaussian).
To understand why, well please refer to the mathematical form of the
functions in the original literature and apply them (analytically) to
semi-conductors. :)
Greetings,
apsi
-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
Institut f?r Chemie der Universit?t Z?rich, CH-8057 Z?rich
Tel: +41 44 63 54 497 / Mobile: +41 79 71 90 935
On Thu, 8 May 2014, Rajdeep Banerjee wrote:
> Dear all,
> ?????????? I was calculating band gap with hybrid functionals for a
> semiconductor. If I don't use smearing (i.e. occupations = 'fixed') the dos?
> seems ok but whenever I use smearing and plot dos, at some energies I get
> negative density of states ! I can't understand why this is happening.
> Can/should one use smearing for hybrid functionals? Following is the
> important part of the input:
>
> ?&system
> ??? ibrav = 4,
> ??? celldm(1) = 6.474424, celldm(3) = 5.837554613 ,
> ??? nat = 10, ntyp = 3,
> ??? ecutwfc = 40,
> ??? london = .true.,
> ??? nbnd = 25,
> ??? occupations = 'smearing', smearing = 'mv', degauss = 0.001,
> ??? input_dft = 'hse',
> ??? exxdiv_treatment = 'gygi-baldereschi',
> ??? x_gamma_extrapolation = .true.,
> ??? nqx1 = 4, nqx2 = 4, nqx3 = 1
>
> ...
>
> ??? Another problem is that whenever I use smearing the output doesn't give
> HOMO-LUMO as it considers it as metal and prints only the fermi energy. So
> if I'm studying semiconductor to metal transition there's no way one can
> check the band gap change.
> ????????????????
> ??? I'd be greatful if anyone can shine some light on above problems.
>
> Thnaks,
> Rajdeep Banerjee
> Ph. D. student
> JNCASR
> Bangalore, India
>
>
