My Community

It is well known that DFT gets the adsorption site wrong for CO on Pt(111). On using DFT+U (PRB 68(2003) 73401) for this system, this error has been shown to be corrected.
I am trying to repeat this calculation but even with U=1.25 eV, the fcc site continues to be the more favorable than atop site. In addition to using PAW PP and ISPIN=2, the INCAR file is as follows. Could someone point out what is missing here?

NWRITE=2
PREC = HIGH
GGA = 91
VOSKOWN = 1
LREAL = .FALSE.
IALGO = 48
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 2 -1 -1 # atom order Pt C O
LDAUU = 1.25 0 0
LDAUJ = 0 0 0
LDAUPRINT= 2

EDIFF = 1e-4
ISPIN = 2
NELM = 400
LMAXMIX = 6

ISIF = 2
EDIFFG = 1e-3
NSW = 150
IBRION = 2
POTIM = 0.5

ISMEAR = 2
SIGMA = 0.20


NSIM = 4
NPAR = 4
LPLANE = .TRUE.
LSCALU = .FALSE.



<span class='smallblacktext'>[ Edited ]</span>

If I read the paper correctly, the on site correction is applied to the CO molecule so as to shift the position of the CO LUMO. You're applying the correction to Pt, which, if anything, might cause a correction to the site energies opposite to what you want (or make Pt insulating...).

Cheers,

Matt

In that case, the following input should give the correct answer:
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = -1 1 1# atom order Pt C O
LDAUU = 0 0.75 0.75
LDAUJ = 0 0 0
LDAUPRINT= 2

However, this also results in fcc site being more favorable than atop site. Any other suggestions?
<span class='smallblacktext'>[ Edited Thu Oct 26 2006, 05:19PM ]</span>

vasp.5 (using hybrid functionals) will solve that problem. It will be released next year.