how accurately can VASP calculate the cohesive energy of Diamond and graphite?

[tracy]

Dear all:
Could you please tell me how accurately we can calculate the cohesive energy of Diamond and graphite? According to my calculated results, the cohesive energy of Diamond is about 9.05eV/atom , and the cohesive energy of graphite is about 9.20eV/atom((I used PAW-GGA potential for both cases). But compared to the experimental values, the results is too large. Can we calculate the cohesive energy of Diamond and graphite more accurately? Any comments would be welcomed!
The INCAR and KPOINTS for Diamond are attached below:
INCAR:
SYSTEM=Diamond C
PREC=High
EDIFF=1e-06
ISMEAR=1
SIGMA=0.1
VOSKOWN=1
ISPIN=1
ENCUT=350.0

KPOINTS:
K-Points
0
Monkhorst Pack
11 11 11
0 0 0

For graphite, I used the geometry which forsdan told me(http://cst-www.nrl.navy.mil/lattice/struk/a9.html ), the INCAR file is similar to the Diamond one, and I used a 8*8*4 k-mesh.

Thanks a lot!

[forsdan]

Well, even if we assume that your energies for both the bulk and atomic system are well converged with respect to all settings (k-points, cutoff energy, etc) and that you have accounted for the possibility of a spin-polarized C atom there still remains the issue with how well the specific GGA describes the bulk system and the atomic system.

In order to obtain accurate cohesive energies both the bulk system and atom must be described accurately by the specific xc-functional. For instance, GGA-PBE in many cases gives an acceptable description of the atomic energies and together with the good description for the bulk system it can provide reasonable cohesive energies. However, this is very system dependent and the results can be quite off in other cases.

The description of bulk graphite is an issue here since the forces between the graphite layers have large contributions from van der Waals forces which are non-local in nature. These forces are are not described accurately by the semi-local GGA functionals which leads to wrong bond distance between the layers.


So these issues determine how well you can calculate the cohesive energies for your systems. I'm sure that there are articles that give a more elaborated investigation on this issue for graphite so it's recommended to look up those papers first.

Please notice that this is just my limited knowledge about the problem. It's far from complete so if anyone else with more experience wants to add or correct something please do.

Best regards,
/Dan Fors







<span class='smallblacktext'>[ Edited Tue Apr 07 2009, 06:16PM ]</span>

[tracy]

Thanks Dan! Your comment is very helpful!!! Have a nice weekend!