> Dear siesta user
>
> I got graphene phonon dispersion with vibra. without
> kgrid_Monkhorst_Pack(%block), means with just single k-point (gamma)
> and   superr cell 2*2*0  got to a accurate graphene dispersion.
>
> I think that it's not true physically but result is correct.
> my question is :
>  Could a single k-point along the  gamma direction be enough to
> describe phonon dispersion?
>
> Best
> Drogar
>

Dear Drogar,
it seems that you mix up two issues:

Monkhorst_Pack, single k-point or not define the quality
(accuracy) of your electronic structure calculation. I'd say,
single k-point for graphene is not good enough, but you'll certainly
get some levels. However, you'll have no dispersion of electron bands,
no Dirac point etc.

The phonon dispersion comes from how frequency varies with wavevector
of phonon, say q. It has nothing to do with Monkhorst-Pack;
technically in Siesta it is given by supercell size. In your case
supercell has more than one cell in the (X,Y) plane, so you ought
to get SOME phonon dispersion. The question of how good it is can be
answered by testing the results vs. increased supercell size
(I bet there is already enough publications dealing with this issue).
A priori 5*5*1 supercell is not obviously too small,
but it depends on which effects you are after.

Your "but result is correct" may mean that you've got luck and
the force constants are not too sensitive to your insufficient
k-mesh integration. Usually they are. Look more carefully;
maybe you results are not that correct after all.

Best regards

Andrei Postnikov

Responder a