Query

[kca002]

Hi:

I have placed on top of C8 graphite surface unit cell three layers of Pt fcc(111) p(2x2) atoms (total 12 atoms with 4 atoms in each layer). Both the lattices matched quite well. I'm trying to simulate platinum layers on graphite surface and perform reactions on them. I performed an ISIF3 geometry optimization and after the optimization the graphite surface was pushed out of the unit cell with the three Pt layers inside the unit cell. Will I be able to perform a simulation with two different surfaces like graphite and Pt in VASP? If I can, could you tell me how? Can I just fix the coordinates of graphite and Pt without performing any geometry optimization and carry our reactions on them?

Thanks,

[admin]

if you do a surface calculation, you cannot use ISIF=3 for the surface runs, because it makes no sense: The vacuum which you need to insert between the surfaces cannot be accounted for properly. ISIF works on the whole cell, i.e. slab+adsorbate+vacuum)).
please
1) use ISIF=3 to optimize the structure of the substrate surface (the graphite if I understand you correctly).
2) use this equilibrium structure (lateral bond lengths, layer distances) as the input structure of your slab and keep the cell fixed (a,b,c,...)
3) then put the adsorbate (Pt) onto the slab and relax (ISIF need not be set explicitely in this case). If required, make use of the 'selective dynamics' option in POSCAR (i.e.you can keep some atoms fixed, and the others are relaxed). It is strongly recommended to do a 'unsymmetric slab' calculation to save computing time:
-- fix the bottom layer(s) of the graphite substrate (F F F)
-- put Pt on ONE side of the slab only, give IDIPOL=3 in INCAR to correct for the dipole moment due to the missing reflection symmetry along z (i.e.one side of the slab is terminated by C, the other one by Pt)
--- if you do a geometry optimization of this cell, the adsobate and the layers of the substrate which are free to relax will converge to their equilibrium geometry.