GGA+U and adsorption site preference

[dcurulla]

Hi everybody,

I am doing some calculations on CO dissociation on Co surfaces and as expected I have the well-known problems about CO adsorption site preference. VASP calculations say that CO adsorbs preferentially on hcp sites for the Co(0001) while experiments clearly show that top sites are preferred.

I would be much interested in performing GGA+U calculations with Co to "predict" the correct adsorption site for CO. I have checked the VASP web site, but I have not been able to find how to perform such GGA+U calculations (only L(S)DA+U). Can somebody please explain me how to do such calculations?

I hope to hear from you soon...

[Veronika]

Closer description of LDA+U and description of the formalism can be found under:
http://cms.mpi.univie.ac.at/vasp/vasp/node158.html
(description of the tags is at the bottom of the page)
greetings

PS: In general LDA+U is only done for (highly) correlated systems, in principle this is a method to add additional onsite Coulomb Energy.
Are you sure that this is the appropriate method in this case?
<span class='smallblacktext'>[ Edited Tue Mar 08 2005, 04:27PM ]</span>

[admin]

In case you really insist on using GGA+U:
GGA + U is automatically used if a GGA-type pseudopotential is used in POTCAR,
the INCAR-tags are the same as for LDA+U

In addition, have you kept the following in mind:
IDIPOL = 3 (if only one side of the slab has CO adsorbed)
VOSKOVN = 1
may change the relative stability of adsorption sites on magnetic systems
(towards non-magnetic sites, usually)
1) please be aware that all calculations are T=0 data, therefore a comparison
may be difficult if a metastable adsorption state is frozen in experiment
--> check the relative stabilities of t,h, b adsorption energies
and the possibility of lateral diffusion