[Pw_forum] Graphene structure

[Hande Ustunel]

Dear Ettore,

As with any structure including metals, you would need some sort of
smearing.

Try something like

    occupations='smearing'
    smearing='mv'
    degauss=0.01

in the &system portion of your input. You must however be informed about
the usage of smearing since the number of k-points that will be sufficient
for convergence will be degauss-dependent. I would read the papers refered
to in the Doc/INPUT_PW.txt. As a rule of thumb soemthing like 0.01 or 0.02
are good smearing widths. 

Incidentally, our vacuum is a tad too large. This will ill-condition your
problem (and consequently mess up your convergence) as well as increase the
run-time. I would decrease it to something like 15 A.

If these fail (which I doubt they will in your case) you can try one of the
following remedies which sometimes work :

1) Increase number of kpoints.
2) Change the location of Ti slightly.
3) Change pseudo (if you can) 
4) Change mixing mode and mixing beta.

Hope this helps,
Hande
    

On Mon, 22 Nov 2010, Ettore Baldini Neto wrote:

> Dear QE users,
> 
> I'm kinda new with quantum espresso. I'm trying to relax a Graphene structure with  32C atoms plus one Ti atom at a given position and the structure do not converge.
> In what follows I copy the input. Could anyone give-me some hint? Is there anything wrong or that can be improved? I'm using the pp's from the QE data base. I'm planning to generate my own pp's soon.
> I'm using a k-point grid of 6x6x1 since with SIESTA (dzp basis) the energy and force convergence is achieved without problem.
> 
> With best regards,
> 
> ************************************************************
> Dr. Ettore Baldini-Neto
> Wernher von Braun Center for Advanced Research
> Campinas, Brazil.
> ************************************************************

-- 
Hande Toffoli
Department of Physics
Office 439
Middle East Technical University
Ankara 06531, Turkey
Tel : +90 312 210 3264
http://www.physics.metu.edu.tr/~hande