[Wien] mBJ potential and LDA+U

[F. Tran]

Dear Kamil Klier,

The purpose of the calculation done in the first step is just to
create the files case.vrespsum (which contains a part of the kinetic
energy density) and case.r2v (the xc potential), which are necessary
to start a MBJ calculation. If these files are not present in the
directory, the MBJ calculation will crash. Then, during the MBJ
calculation, these two files will be updated at each iteration,
so, the MBJ results are independant of the functional used in the
first step. It is only for magnetic systems, for which several different
states can sometimes easily be stabilized, that the choice of the
functional used to generate the starting case.vrespsum and case.r2v
files can eventually have some influence on the final results. 


On Thu, 17 Jun 2010, Kamil Klier wrote:

> 
> Dear Colleagues,
> 
> Please comment on the following question regarding the mBJ calculations
> (section 4.5.8 of the Wien2k 10.1 user manual):
> 
> Is it acceptable, on theoretical grounds, to calculate the scf cycle using
> LDA+U, not just LDA or PBE as stated in the User Manual, before the
> calculation using the mBJ potential?
> 
> This query is based on some calculations I have done for ZnO, and am now
> running for anion-doped ZnO.
> 
> I have read the PRL 2009 vol. 102 paper by Fabian Tran and Peter Blaha, and am
> getting a comparable bandgap for ZnO using Wien2k 10.1, as expected. The mBJ
> bandgap does not seem to depend on the functional in the scf calculation done
> in the first step (User Manual p. 49, section 4.5.8).
> I have tried the 'regular' LDA, PBE, WC, -eece, and LDA+U and all of these
> yielded the same bandgaps.
> 
> However, the separation of the filled VB levels, O2p and Zn3d below it, was
> found significantly greater with LDA+U than by the other methods, appx. by 1 -
> 1.2 eV.  And this LDA+U result is in a better agreement with XPS experiment
> than the rest. So the 'best' result is obtained by LDA+U followed by mBJ.
> 
> If there are no conflicts between the potentials in LDA+U and mBJ, it would
> seem that this combination will yield better relative energies of the d-levels
> and the bandgap.  Please correct if this is wrong.
> 
> Greetings, Kamil Klier
> 
> 
> 
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>