The ABINIT Users Mailing List ( CLOSED )

[James Johns <jej7x@xenon.cchem.berkeley.edu>]

I have done ML convergence studies, from 15-4ML, and the trend is still 
the same, serious underestimation of the WF

On Mon, 23 Apr 2007, Pawel Scharoch wrote:

>       Dear James,
>
>    just one remark: calculations of any quantity with the use of
> a metallic slab models require convergence tests with respect to the
> number of monolayers, because of the quantum size effect (QSE)
> (see e.g.:   Kiejna A. et al. Quantum-size effect in thin Al(110) slabs,
> Surf Sci 432 (1-2): 54-60 Jul 9 1999). Of course I am not sure
> whether this effect is responsible for the discrepancy you get.
>
> With best regards,
>
>
>
> ----- Oryginalna wiadomo?? -----
> Od: James Johns <jej7x@xenon.cchem.berkeley.edu>
> Data: Niedziela, Kwiecie? 22, 2007 6:49 am
> Temat: [abinit-forum] Mg workfunction
> Do: forum@abinit.org
>
> > Ok, so I'm going partially off of the tutorials here, but I'm
> > attempting
> > to approximate the workfunction of Magnesium(0001) surfaces.  I've
> > attatched a sample input file.  My problem is that despite all
> > attempts to
> > better optimize the convergence critea based on nband, ecut, the
> > kpt grid,
> > or anything, it's underestimating the workfunction by roughly a
> > factor of
> > two.  I'm using the fermi level energy level from a 10 atom slab
> > calculation as the workfunction.  Is there something wrong with
> > this?
> > When I do this for the Ag(111) surface, my estimation is roughly
> > 100-200meV higher than the experimental version.  The fact that
> > it's only
> > slightly more ( and seems to converge based on typical paramters
> > such as
> > size of vaccuum spacing etc), while on Mg it is severely less
> > makes me
> > think that it's not something inherent in the LDA (or GGA)
> > approximation.
> > Does anyone have any thoughts, or alterations to the input file to
> > correct
> > for something I'm doing stupidly?
> > Thanks a lot,
> > --James Johns
> > UC Berkeley
> > Charles Harris group member