[CP2K-user] [CP2K:14238] Re: Geometry optimisation of organic molecule on fcc Au(111) slab

Anton S. Lytvynenko anton.s.... at gmail.com
Tue Nov 17 12:25:59 UTC 2020


Dear Dmitrii,

17.11.20 03:37, DMITRII Drugov пише:
> I built Au(111) slab with a= 4.07825 

a= 4.07825 is the cell dimension of the gold structure itself. The
symmetry of the (111) slab is hexagonal with a = b = 5.76752
(4.07825*sqrt(2)) and gamma = 120. You can use Avogadro to build the
slab and calculate the dimensions of the new "cell", but better consider
this topic to choose
https://groups.google.com/d/msgid/cp2k/9d5922dc-c114-4bf1-bbc8-4b576728d29bn%40googlegroups.com?utm_medium=email&utm_source=footer
, as Avogadro could be a bit tricky in this regard.

So finally you have:

ABC 17.30256 17.30256 50.0

ALPHA_BETA_GAMMA 90.0 90.0 120.0

You should leave enough layers of gold atoms ("thickness") in the slab
and enough space for vacuum. Ideally, you should start from 4--6 layers
and 30-50 A for the third dimension and check convergence of the results
in respect to these parameters. Use that paper to see what to expect
(however, you should be aware of the metallic nature of your system).

In principle, I see no reason to fix any atom in the DFT optimization
(it could be reasonable for MM or semiempirics, but Au-Au and
Au-whatever interactions are well described via DFT). If the geometry is
built correctly, no devastation of the metal layer will happen. If you
see any notable disorder at the equilibrium, I would double-check the
setup rather than try to fix the atoms.

Periodicity in x and y dimensions (within the slab plane) is a must in
your case (if you want to deal with 2D layer rather than nanoparticle).
Regarding Z direction, it depends on your study, you can try both cases
and compare the results. With PBC along z axis, you should provide
really large z to exclude spurious self-interactions. Without PBC along
z, you might be unable to provide some studies reasonably, face
unexpected issues or just have scarce literature base for comparison.

Hope that helps.

Yours,

Anton





More information about the CP2K-user mailing list