Ni atoms deposit to the surface of graphene

[qiuweicheng]

Dear Sir,

I want to simulate the Ni thin film growth on graphene via Vapor Deposition. Before this, I read carefully the  SiC tutorial  "Simulating Thin Film Growth via Vapor". I write the Ni-growth-on-graphene.py file by refering to the Script_deposit_SiC.py file.  But, the results show that the substrate_atoms diffuse into the vapor_atoms, and diffuse out of the crystal lattice. How can I solve this problem?

The main difference between my py file and Script_deposit_SiC.py is as follows:
1. the SiC structure is replaced by Ni/Graphene supercell
2. potentialSet = Tersoff_SiC_2005() is replaced by  potentialSet = ReaxFF_CHONi_2015()
3. possible_elements = [Silicon, Carbon] is replaced by possible_elements = [Nickel, Nickel]
4. vapor_temperature=2400.0*Kelvin is replaced by vapor_temperature=300.0*Kelvin

The results and py file are shown in attachments.
Thanks

[Anders Blom]

Seems the potential is not capable of describing bulk Ni, so then there is not much you can do except find a better potential, or use DFT (don't!).

[qiuweicheng]

Hi, Anders Blom
Thank you for your reply!
1. For the question that the substrate_atoms diffuse into the vapor_atoms, if I add the  constraints = [RigidBody(bottom),               FixAtomConstraints(substrate, fix_cartesian_direction=[False, False, True])] for the deposition, Can it solve the problem? Would that be proper?
2. I am a new studier for the MD,  So, Could you give me some suggestion for a better potential for Ni growth on graphene?

[Anders Blom]

1. No, the issue probably lies with the potential itself. You need  to read the original paper defining the potential and check what it's designed for. You can also run a test by making a slab of Ni, constrain a few layers, and just run an optimization or a few hundred MD steps and see if the surface remains stable.

2. I assume you have performed an extensive literature study of Ni/graphene, and how to simulate growth? Did anyone use an empirical potential already? If not, it's unlikely that one exists.

[qiuweicheng]

Thank you for your good suggestions.
Here, there is another problem. The SiC example introduces the method of BulkConfiguration growth by MD.  But,  How to use the BulkConfiguration growth by MD to  build a DeviceConfiguration? or How can we use the MD method to grow a DeviceConfiguration?
There is very little in the way of examples and documentation out there to show this.

[Anders Blom]

A device is made up of 3 bulk structures: two electrodes and a central region

[qiuweicheng]

Yes, it is. But, if the central region is grow by MD method, it will not be periodic.  That will result in the fail of  extended electrode.
In other words, we don't know how to bulid the DeviceConfiguration using the MD growth BulkConfiguration?

[Anders Blom]

Often you can finish the ends with pure crystalline material, which you don't grow but just add manually, but I agree it's a bit challenging in practice.