Geometry relaxation of large system

Question on input files/tags, interpreting output, etc.

Please check whether the answer to your question is given in the VASP online manual or has been discussed in this forum previously!

Moderators: Global Moderator, Moderator

Post Reply
Message
Author
User avatar
SKM
Full Member
Full Member
Posts: 125
Joined: Wed Oct 30, 2019 5:39 am
License Nr.: 5-516

Geometry relaxation of large system

#1 Post by SKM » Thu Jan 13, 2022 3:04 pm

Hi
i have system with 182 atoms, surface slab with ligand attached. its not converging with default NELM=60 (explicitly set).
Earlier on some occasions, though the system not converged electronic step, but after few such steps, then it used get electronic and eventually ionic convergence also used happen.


In present case also electronic convergence not happening, and also the ions get distort a lot and very randomly. The system seems to be metallic. [not sure though??]

attaching the input files. Kindly help to resolve the issue.
Regards
You do not have the required permissions to view the files attached to this post.
Regards
SKM

alexey.tal
Global Moderator
Global Moderator
Posts: 228
Joined: Mon Sep 13, 2021 12:45 pm

Re: Geometry relaxation of large system

#2 Post by alexey.tal » Fri Jan 14, 2022 12:32 pm

Dear SKM,

If I understood you correctly, the issue is that the electronic iterations do not converge for your system within 60 iterations. I see that in the first ionic iteration the total energy is slowly but steadily converging, so I would suggest that you try to increase the number of iterations (NELM) or use the default mixer (IMIX = 4 and AMIX = 0.4) to speed up the convergence.

User avatar
SKM
Full Member
Full Member
Posts: 125
Joined: Wed Oct 30, 2019 5:39 am
License Nr.: 5-516

Re: Geometry relaxation of large system

#3 Post by SKM » Fri Jan 14, 2022 2:39 pm

hi
thanks for quick reply
i changed the tags as you suggested.
but only first electronic iterations step got converged, but again the next step on wards, i cant see any gradual reduced value of dE even if the NELM is kept at 860 steps. anyway. i stopped the run in the middle as no point in continuing.
attached the revised files. Kindly suggest.
You do not have the required permissions to view the files attached to this post.
Regards
SKM

alexey.tal
Global Moderator
Global Moderator
Posts: 228
Joined: Mon Sep 13, 2021 12:45 pm

Re: Geometry relaxation of large system

#4 Post by alexey.tal » Fri Jan 14, 2022 4:54 pm

It looks like some of the atoms in your structure are far away from their optimal positions and they get displaced too much in the first ionic iteration. You can limit the step width of the ionic relaxations with POTIM, for example POTIM = 0.1.

User avatar
SKM
Full Member
Full Member
Posts: 125
Joined: Wed Oct 30, 2019 5:39 am
License Nr.: 5-516

Re: Geometry relaxation of large system

#5 Post by SKM » Sat Jan 15, 2022 12:04 pm

Thank you.
it worked for that system.
exactly same system with one addition of molecule start giving issue on non-ionic convergence this time.
even in continuing run, the total energy just fluctuating within 1 eV and the total force/max force also fluctuating.
as the system is slightly bigger now in number of atoms, is POTIM need to be reduced further to restrict ionic movement?
The input files and outcar attached, please. kindly help.
Regards
You do not have the required permissions to view the files attached to this post.
Regards
SKM

alexey.tal
Global Moderator
Global Moderator
Posts: 228
Joined: Mon Sep 13, 2021 12:45 pm

Re: Geometry relaxation of large system

#6 Post by alexey.tal » Mon Jan 17, 2022 5:04 pm

The problem with this structure is essentially the same, i.e. some atoms are too close which causes huge forces in the system. Reducing POTIM should work but it will also slow down the convergence. You should try to use a better initial structure for your calculation.

User avatar
SKM
Full Member
Full Member
Posts: 125
Joined: Wed Oct 30, 2019 5:39 am
License Nr.: 5-516

Re: Geometry relaxation of large system

#7 Post by SKM » Tue Jan 18, 2022 9:25 am

Hi
thanks for the reply.
but my system the closest distance between two atoms is 1.10 Angstroms. I run many systems earlier with even as close as 0.5 angstroms. Do you think i still have to increase the distances, in the initial geometry?

Regards
Regards
SKM

User avatar
SKM
Full Member
Full Member
Posts: 125
Joined: Wed Oct 30, 2019 5:39 am
License Nr.: 5-516

Re: Geometry relaxation of large system

#8 Post by SKM » Fri Jan 21, 2022 12:15 pm

Thank you.
with slight change in shift of atoms, its running ok.
Thanks again.
Regards
Regards
SKM

User avatar
SKM
Full Member
Full Member
Posts: 125
Joined: Wed Oct 30, 2019 5:39 am
License Nr.: 5-516

Re: Geometry relaxation of large system

#9 Post by SKM » Tue Jan 25, 2022 4:03 pm

hi
this seems very tricky, for a slight change by adding a ligand to the system, the relaxation trend got changed and now not converging. The dE in electronic step keeps fluctuating but not converging.
May i know, why these type of relaxation issues, when we not changed except adding atoms? or does this depend on adding any metallic ions? As this system i replaced Copper with Manganese.
kindly see the attached input and log files.
Regards
You do not have the required permissions to view the files attached to this post.
Regards
SKM

chandan_kumarvishwakarma1
Newbie
Newbie
Posts: 26
Joined: Mon Oct 26, 2020 7:56 am

Re: Geometry relaxation of large system

#10 Post by chandan_kumarvishwakarma1 » Sat Jan 29, 2022 2:51 pm

Hello,

The problem in relaxation of the large supercell is common, to solve this problem you can take help from the
attached presentation.
" wiki/images/b/b6/VASP_lecture_Basics2.pdf "

---
Chandan

Post Reply