Problems with convergence of wide ribbons

[CHILL]

Hello,

I have been calculating decorated graphene ribbons of different width. Until some certain width the relaxation process goes smooth but after I hit this limit the behaviour of the scf cycles becomes really erratic.
I attached a screenshot of one of those calculations and it can be seen there that everything is not converging at all.
I used around 20 Angstrom of vacuum and tried to increase it without any success. My ecut was 45 Hartree.
Another thing, for the widths with which I had problems when I remove the decorating atoms (so I end up with just wide graphene ribbons) the calculations are stable. It suggests that there shouldn't have been a small mistake in the input (like vacuum, ecut etc.) and it probably has something to do with the adatom in combination with large width.

Could anyone give me a hint of what might be wrong here?
If necessary I can share my input/output.

Thank you!

Best regards, Mikhail

[ebousquet]

Dear Mikhail,
If everything works fine up to a given thickness, it is possible that you reach a transition in the electronic state/structure, that can makes the calculation convergence more troublesome. It is also possible that you have oscillating electronic states if you have an internal electric field between the surfaces too, etc. You might have to change your convergence parameters like diemac, diemix, nline, etc. Did you play with these different parameters?
Best wishes,
Eric

[CHILL]

Dear Eric,

Thank you for your response!
Following your advice I tried to change variables you mentioned and diemac did the trick!
I always have problems with diemac/diemix variables because I do not fully understand the physical meaning of those.
Do you know where I can find detailed mechanisms behind those variables reviewed or explained ? (couldn't find anything on the description page)

Best regards, Mikhail

[ebousquet]

Dear Mikhail,
Great that it works
You can certainly find some explanations in the book of R. Martin "Electronic Structure Basic Theory and Practical Methods".
These references could also be helpful:
https://journals.aps.org/prb/abstract/10.1103/PhysRevB.28.5462
https://www.sciencedirect.com/science/article/abs/pii/0009261480803964?via%3Dihub
https://journals.aps.org/prb/abstract/10.1103/PhysRevB.23.3082
https://journals.aps.org/prb/abstract/10.1103/PhysRevB.25.4260 --> related to diemac
https://www.sciencedirect.com/science/article/pii/S0021999196900595?via%3Dihub
But if somebody has a better ref, please cite them here!
Best wishes,
Eric