[Pw_forum] CuBTC open metal organic frameworks CONVERGENCE

[Ari P Seitsonen]

Hello Francesco,

   Adding to the previous posts, I found a better convergence in systems 
with vacuum using

     mixing_mode     = 'local-TF'

   Personally I do not find using 'conv_thr = 1e-8' as very strict, this is 
what I usually use for relaxation (and a bit tighter still when going for 
phonons etc), but indeed one might save some iterations, in particular at 
the beginning of the relaxation. Did you check that your electronic 
structure (density of states for example) looks as expected, for example 
having a band gap if such is expected?

   And personal opinion again, I would be surprised if the bad convergence 
is due to 50 Ry being too low, as those elements, in particular C, H and 
O, should be fine already at 35-40 Ry, not to mention 50. I can just 
imagine that the vdW-DF is extremely sensitive to some noise that arises 
from the augmentation charges, that was maybe not an issue when those 
pseudo potentials were first adapted in calculation (ie using "plain GGA" 
or LDA).

     Greetings,

        apsi

-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
     Ecole Normale Supérieure (ENS), Département de Chimie, Paris
     Mobile (F) : +33 789 37 24 25    (CH) : +41 79 71 90 935


On Mon, 23 Jan 2017, Francesco Pelizza wrote:

> Hello dear community,
>
>
> Here in strathclyde some colleagues are using QE to relax and than study
> some MOF's.
>
> The problem starts from the beginning, no convergence in the first SCF
> cycle before a fixed cell relaxation.
>
>
> We took care of spin polarized calculation, switched to non polarized
> calculation and many other things, WFC from 50 to 80 Ry and RHO cutoffs
> 10 times the WFC, with gamma sampling and 3 x 3 x 3.
>
> Also checked the smearing, degauss values, nothing seemed to make the
> convergence better.
>
> Here you can find some input files and output files.
>
>
> We used ultrasoft pseudopotentials applying PBE and then vdW-DF functionals.
>
>
> No success at all.
>
>
> Please some of you have better experience of us in these systems to tell
> us what we do wrong?
>
>
> BW
>
> Francesco Pelizza and Chris Campbell
>
>
> Strathclyde University
>
>
>