problem with total charge density and graphite calculations

Matt W MattWa... at
Tue Mar 10 13:43:50 UTC 2009

You're likely to have a rather significant basis set superposition
error with the rather small DZVP basis set you're using that will
cause an effective binding between the slabs when the basis functions
start to overlap.  This you can compensate for by using bigger basis
sets or calculating the correction.

The bigger problem is the metallicity, which is not well handled by
CP2k and is slowing down your electronic convergence by orders of


On Mar 10, 1:26 pm, Andrzej <kwa... at> wrote:
> Thanks for response. Well it is why I decided to multiply the unit
> cell.
> In my original crystal, which is an insulator, and the layers are
> binded by vdW interaction I also have an attraction between layers,
> when it is calculated using pure PBE - similarly as I observed for
> graphite. It worries me, because it looks as an artificial effect. If
> I have an artificial attraction using PBE, the binding energy obtained
> from PBE+D is oversetimated.
> On Mar 10, 1:05 pm, Matt W <MattWa... at> wrote:
> > Hi,
> > well one issue is when does the system become metallic?  Possibly in
> > some of the smaller systems there is enough of an artificial band gap
> > for you to reach convergence ... but it's not obvious to me why it
> > works at all.
> > Not sure why you're losing electrons though.
> > Matt

More information about the CP2K-user mailing list