[Wien] Supercell calculation is not converging

[Laurence Marks]

Follow the science, not the DFT.

1) Unless you have a large grain size and synchotron data I really
wonder about the logic of removing 1 atom in a 2x1x1 supercell.
Particularly if this is an oxide, have you made it a metal whereas it
should be an insulator? Does the fit pass significance tests etc? Do
you have other data that implies vacancies?
2) In terms of the DFT, I expect that you have made it a metal which
is harder to converge than an insulator. Check in case.output2 in both
cases (a quick, pseudo DOOS). If it really does have vacancies, you
might need more than 7 kpts, and I suggest using TEMPS.

2011/9/5 shamik chakrabarti <shamikphy at gmail.com>:
> Dear wien2k users,
>         I have performed a 56 atom / unit cell calculation for a
> ferrimagnetic material. It was well converged both in energy and charge by
> 60 iterations. 14 K points were used for this calculation. Then I have
> generated a 2x1x1 supercell having 112 atoms / supercell. This was done to
> achieve better experimental model as obtained through Rietveld analysis.
> From Rietveld analysis it was also seen that for 112 atoms we can consider a
> cation (Fe) vacancy and we have generated it by deleting one Fe atom from
> the supercell. Final structure, hence, contains 111 atoms and all other
> occupancies at different lattice sites are well matched with Rietveld
> analysis of experimental XRD. I am using 2 k points for this 111 atom
> supercell. But even after 60 iterations the convergence achieved are as
> follows:
>    cc: 0.234...    cc was set to 0.001
>    ec: 0.063...    ec was set to 0.0001
> Sir, my question are:
> 1) As the supercell is merely an extension of the 56 atoms unit cell with
> more accurate values of atomic occupancies as in the experimental structure,
> why it should at all taking so much time to converge?
> 2) We have created one defect by creating a Fe vacancy...is it that the
> structures having defect take more time to converge?...but as far as our
> analysis until now..the structure should have one Fe vacancy per 112
> atoms...
> 3) As we have obtained converged solutions for 56 atom cell by using 14 k
> points..is it that we should use 7K points for 111 atom supercell
> calculation?..is 2 K points are actually small in order to achieve
> convergence of this structure?
> Any response in this regard will be very helpful for us. Thanks in advance.
> with regards,
>
> --
> Shamik Chakrabarti
> Senior Research Fellow
> Dept. of Physics & Meteorology
> Material Processing & Solid State Ionics Lab
> IIT Kharagpur
> Kharagpur 721302
> INDIA
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>



-- 
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
Research is to see what everybody else has seen, and to think what
nobody else has thought
Albert Szent-Gyorgi