Dear All,
I am trying to calculate the diffusion path of an atom through a Zeolite type structure. I am using 3 images, each image having 120 atoms. Yet, after more than 300 iterations the path is not converged. The system is going on the proper route, but too slow. I have run the neb.ef script and the output is:
0 0.00000000 -842.53654800 0.00000000
1 0.00000000 -841.02064900 1.51589900
2 0.00000000 -842.37074500 0.16580300
The force on the first and last image seems to be small. The force on the middle image seems to be yet relatively high.
For the last 100 iterations I have used the INCAR:
###################################################################
ISTART = 0
ICHARG = 2
ENCUT = 550
SPRING=-5
IMAGES=3
ICHAIN=0
LCLIMB=.TRUE.
PREC = High
ISPIN = 1
ISIF = 2
ISYM = 0
EDIFF = 5e-7
NELM = 120
NELMIN = 5
ALGO = Normal
EDIFFG = -0.025711
NSW = 30
POTIM = 0.2950000
IBRION = 1
ISMEAR = 0
SIGMA = 0.11
LCHARG = FALSE
LWAVE = FALSE
LREAL = Auto
LPLANE = TRUE
NPAR = 48
################################################################################
I would like just to ask you to confirm me if my INCAR is not wrong (am I using too rigid parameters ? should I still use IBRION=3 ?). I have very little experience with NEB, and I would largely appreciate any answer.
Wish you a good end of the week,
Ed
Slow convergence
Moderator: moderators
Re: Slow convergence
It's a little hard to tell from this information. But is does look like you do not have our code linked into vasp, since I do not see the forces printed in the output from nebef.pl (column 2). The last column, which are are interpreting as force, is the energy difference with respect to the first image. If you put OUTCARs in your 00 and 04 directories, it will include those endpoint energies as well.