[Pw_forum] Phonon Dispersion with LO-TO splitting

Srijan Kumar Saha srijan.india at gmail.com
Mon Mar 14 15:42:12 CET 2011


  Dear Eyvaz,

  Thanks a lot for your helpful reply.

> Z* and \epsilon, therefore, LO-TO splitting  are  for
semiconductors/insulators.
> As you know, all valence bands are occupied, and conductivity bands are
empty. Thus,
> you can omit the smearing keyword to let  pw.x know that you deal with a
semiconductor.

I agree with you.
However, in practice, sometime smearing keyword needs to be used for a
complicated
semiconductor or insulator in order get a faster scf convergence. And hence,

it is better to identify a system is non-metal or metal by its band-gap
instead of smearing
keyword. In particular, phonon code in QE complain a lot if the system is a
spin-polarized  insulator.
For example,    "fixed occupations and lsda need tot_magnetization",
 "The phonon code with constrained magnetization is not yet available",
 " IF (epsil.AND.lgauss) &  CALL errore ('phq_readin', 'no elec. field with
metals', 1)", etc. etc.
Atleast for the last error, I guess, it is better if we write,
 " IF (epsil.AND.gap_zero)  &  CALL errore ('phq_readin', 'no elec. field
with metals', 1)"
in place of the current existing statement. Then it will give a chance to
compute \epsil also for
an insulating system which uses "smearing" keyword just for faster
convergence.

>  See example06

For magnetic insulator, is there any further example which shows
a phonon-dispersion with LO-TO splitting ?

 > P.S. So, they are not related with postprocessing

I meant, postprocessing after ph.x.
I thought, q2r.x and matdyn.x, plotband.x as post-processing of phonon code
(ph.x) .


Thanking you and with my best regards,

Srijan Kumar
R&D Assistant
JNCASR
Bangalore 560064

On Mon, Mar 14, 2011 at 1:54 PM, Eyvaz Isaev <eyvaz_isaev at yahoo.com> wrote:

> Dear Srijan,
>
> Z* and \epsilon, therefore, LO-TO splitting  are  for
> semiconductors/insulators.
> As you know, all valence bands are occupied, and conductivity bands are
> empty. Thus,
> you can omit the smearing keyword to let  pw.x know that you deal with a
> semiconductor.
> Then you will got dyn1-file  with Z* and \epsion, q2r.x and matdyn.x will
> take care about these
> parameters and LO-TO splitting.
>
> Having IFC (*.fc) file you can edit Z* and  \epsilon.
>
> See example06  and /Doc/PH.txt.
>
> Best regards,
> Eyvaz.
> P.S. So, they are not related with postprocessing
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20110314/93a3669d/attachment.html>


More information about the users mailing list