# [CP2K:3583] Re: general question on calculating metals using cp2k

carlo antonio pignedoli c.pig... at gmail.com
Thu Nov 3 10:43:33 UTC 2011

```Dear Jun form the experience I have for systems I worked on,
a slab 4 layer thick could be the minimum thickness for a first investigation,
be careful that you would have, in such a thin slab, a spurious interaction
among surface states.
To reduce this, a "empirical" way is hydrogenation of one side of the slab.

Dont' forget to consider the relevance of van der Waals corrections.

Usually the smallest slab I consider in xy is in the order of 6x6
compared to a primitve cell
but this is just an indication and the result depend really on the
particular system you are dealing with.

Magnetic systems are always complex.. I have not enough experience on
this to give
you suggestions.

Ciao

Carlo

On Thu, Nov 3, 2011 at 11:34 AM, Jun <chen... at gmail.com> wrote:
> Thanks, Carlo.
>
> Yes, I want to do slab calculations. The properties I am interested in
> are, for example, adsorption energies of some atoms or molecules, work
> function, DOS et al. For start, I guess the thickness of four layers
> should be enough, I mean for Pt(111). What I am not so sure is the xy
> dimensions. Would MxM, say M=5, be enough? I read some discussions on
> the magnetic state of Ni that Ni slab of M=6 and 6 layers is necessary
> to get a reasonable spin state. I hope it is less demanding for non-
> magnetic metals like Pt.
>
> Cheers,
> Jun
>
> On Nov 3, 8:53 am, carlo antonio pignedoli <c.pig... at gmail.com>
> wrote:
>> Dear Jun,
>> regarding the size of the cell, it depends on which property of the
>> system you would like to reproduce.
>> There is a simple rule:
>> if your property is well converged with a k-point cacluation for a
>> unit cell on a MxM regular grid not shifted (I mean
>> including Gamma point..)
>> then a supercell as big as MxM the unit cell will give you "exactly"
>> the same results.
>>
>> Of course overriding symmetries could bring the MxM system to a
>> geomoetry not compatible
>> with the unit cell (this does not mean that the result is wrong...)
>> that's in part why I was saying "exactly".
>>
>> If you are interested in slab calculations, no k-points in the
>> direction perpendicular to the slab surface,
>> then the thickness of the slab, again, depends "solely" on the
>> particular properties of the system that you
>> would like to reproduce.
>>
>> For the behavior of the optimization algorithms i let somebody more
>> expert to reply.
>>
>> Ciao
>>
>> Carlo
>>
>>
>>
>>
>>
>>
>>
>> On Wed, Nov 2, 2011 at 6:35 PM, Jun <chen... at gmail.com> wrote:
>> > Hi,
>>
>> > I intend to run some calculations on metals. I am wondering if experts
>> > could give me some ideas on the performance of cp2k. For example, to
>> > model a metal surface, say Pt(111), what size of the system is big
>> > enough for compensating the lack of k-point sampling? Usually, how
>> > many iterations are needed to reach SCF convergence? How long it
>> > roughly takes for each SCF step? How slow is the matrix
>> > diagonalisation compared to OT? ...
>>
>> > It would be nice if someone shows me an input example for metals. I
>> > know there was one given by Prof. Hutter about two years ago. Any up-
>> > to-date one?
>>
>> > Thanks a lot,
>>
>> > Jun
>>
>> > --
>> > You received this message because you are subscribed to the Google Groups "cp2k" group.
>> > To post to this group, send email to cp... at googlegroups.com.
>> > To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
>> > For more options, visit this group athttp://groups.google.com/group/cp2k?hl=en.
>
> --
> You received this message because you are subscribed to the Google Groups "cp2k" group.
> To post to this group, send email to cp... at googlegroups.com.
> To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/cp2k?hl=en.
>
>

```