Electrode Calculation not converging with DFT

[ruyam]

I am trying to analyze the Ni-graphene interface. My Left electrode calculation is getting converged but right electrode calculation is not converging when I use DFT for zero bias. When I do the calculations with Extended Huckel calculator instead of DFT, its getting converged and calculates transmission spectrum. PFA my python file. Kindly suggest where am I going wrong in my approach because Extended Huckel went fine.

[Anders Blom]

You are by accident missing an atom both in the left and right electrodes. This will lead to dangling bonds and bad convergence. Also note that you should set initial spins of the magnetic part of the system.

[ruyam]

Dear Anders Sir,

Thank You for your reply. I'll try to correct my errors and analyse the system again.

[ruyam]

Dear Anders Sir,

I rectified the errors you mentioned. But, now when I try to optimize the structure with max. Force of 0.05eV/Ang (as suggested in your paper in Phys.Rev.B on Ni-Graphene interface), I fail to get the structure optimized even after 67 Force steps. The structure gets distorted alot at the interface.
PFA my python file and also a image depicting the device structure after 67 steps.
Kindly, let me know where am I going wrong.


Thanks in advance.

regards,
ruyam

[Umberto Martinez]

You are using a bulk configuration, meaning that your system is periodic in the three directions.
try to repeat you structure in the C and big direction and you will find:

1) you need a bigger vacuum region such that the you have an isolated system.

2) you have two graphene-nickel interfaces.
I guess you want to use a device configuration instead, as in you original post.

take a look to this tutorial
http://quantumwise.com/publications/tutorials/builder/174-graphene-nickel-interface

[ruyam]

Dear Umberto Sir,

What I understand is that I should use the device configuration instead of bulk and a bigger vacuum region. But I fail to understand what you mean by :

" 2) you have two graphene-nickel interfaces. "

The structure I am trying to analyze has only 1 Ni-Graphene interface. Do you mean that I should have a symmetric device configuration so that I can have 2 interfaces ? And if it is so, how does having 2 interfaces help optimize the structure compared to having 1 Ni-Graphene interface?   

[Umberto Martinez]

no, I do not mean you should have two interfaces.
You do have two "interfaces" (two regions where Ni and C atoms are at a very short distance) in your previous example and this is because the bulk configuration is periodic in C direction as well.
see attached figure.

sorry for the confusion and I hope it is more clear now.

yes, use the device configuration.