scf convergence issue: isolated systems in PBE + halogen elements

Total energy, DFT+U, BigDFT,...

Moderator: amadon

Post Reply
marco.digennaro
Posts: 13
Joined: Thu Jun 16, 2016 8:47 am
Location: Bruxelles, Be
Contact:

scf convergence issue: isolated systems in PBE + halogen elements

Post by marco.digennaro » Thu Nov 26, 2020 1:31 pm

Good morning,

I am testing the convergence of a few isolated systems and I find problems in converging systems containing very electronegative elements. I am varying ecut and acell only, pseudo is PBE and kpt is always 1.

Most of the molecules converge fine, but it becomes more difficult the more I move right in the periodic table.

BF4 and PF6 are the worst molecules, never converging.
BF3 and BCl3 show some non-converging case, but taking an high ecut (>20) and an high acell (>30) is enough.
N2 also presents some problems.

So, what is wrong with PBE+halogen atoms?

Here's an example of non converging system (BF4):

Code: Select all

#Definition of the unit cell
acell 3*30

#Definition of the atom types
ntypat 2
znucl  5 9

#Definition of the atoms
natom 5
typat 1 2 2 2 2
xcart

   0.0001148660   0.0000001274   0.0000948037
   0.0000254663   0.0000002928   1.4168343567
   1.3357270137   0.0000003076  -0.4722566817
  -0.6679098939  -1.1565412448  -0.4723165605
  -0.6679091494   1.1565405706  -0.4723160520

#Definition of the planewave basis set
ecut 30

#Definition of the k-point grid
kptopt 0          # Enter the k points manually
nkpt 1            # Only one k point is needed for isolated system,
kpt 0.0 0.0 0.0   # taken by default to be 0.0 0.0 0.0

#Definition of the SCF procedure
nstep  300        # Maximal number of SCF cycles
toldfe 1.0d-6
diemac 2.0
Thank you
Marco Di Gennaro
Toyota Motor Europe (Be)

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: scf convergence issue: isolated systems in PBE + halogen elements

Post by ebousquet » Tue Dec 01, 2020 4:15 pm

Hi Marco,
Hard to say... I guess you are running with PAW, right? JTH vs GBRV?
Did you test with increasing nline, or playing with diemac and diemix?
Cheers,
Eric

Post Reply