Vacuum space for surface energy comparison

[tak]

Dear VASP Forum:

Here is another basic question I have.

I want to calculate Fe(100) surface energies with several layers (3 to 6) to find where it converges. My question is if the vacuum spaces have to be same for each calculation. In other words, do I have to increase lattice vectors (unit cell) as layer grows in order to keep vacuum space? Please see my POSCAR examples for 3 and 4 layers. I increased z-lattice vector as layer increases.

Thank you very much

POSCAR for 3 layers:
bcc Fe(100) Surface Relaxation
2.866
1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 3.0
3
Selective Dynamics
Cartesian
0.0 0.0 0.0 F F T
0.5 0.5 0.5 F F T
0.00.0 1.0 F F T

POSCAR for 4 layers:
bcc Fe(100) Surface Relaxation
2.866
1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 3.5
4
Selective Dynamics
Cartesian
0.0 0.0 0.0 F F T
0.5 0.5 0.5 F F T
0.0 0.0 1.0 F F T
0.50.5 1.5 F F T

INCAR for both calculations:
SYSTEM = Fe bcc
ISTART=0
ICHARGE=2
ISMEAR = 2
SIGMA = 0.2
ENCUT = 300

ISPIN =2
NBANDS = 25

IBRION = 2
NSW = 15
POTIM = 0.2
EDIFF = 1E-5

[admin]

the vacuum thickness for slabs of varying thickness should be kept constant, therefore youhave to re-scale the lattice vector if you insert more layers in the slab. If you keep 3.0*2.866 Ã… for the 4-layers calculation as well, your vacuum will certainly not be thick enough (just about 4.3 Ã…)
INCAR, NBANDS : please mind that the number of bands should also be increased (re-scaled) with the number of valence electrons in your system, so please adjust this parameter as well.

[forsdan]

If you want to compare your results, you can look at the work by P. Blonski and A. Kiejna in

Surface Science 601 (2007), 123-133, "Structural, electronic, and magnetic properties of bcc iron surfaces"

where they have calculated surface energies for Fe (including 100) with VASP and present convergence results with respect to the number of layers.

Best regards,
/Dan Fors
<span class='smallblacktext'>[ Edited Thu Feb 28 2008, 12:02PM ]</span>