Questions about Geometry Optimization of Defective Graphene with Al atoms

[Fishpack]

Hi, everybody.
I've tried to simulate the atomic structure of the defective graphene with Al atoms to obtain the phonon dispersion curve of the defective graphene with Al atoms. The Al atoms will be located in the middle of the defects.
(I think the figure that is attached will be helpful.)

However, whenever I optimized the geometry of the defective graphene with Al atoms, the result is different  with the atomic structure of the attached figure.

How can I realize the atomic structure of the defective graphene with Al atoms? and Is
it OK that I use the potential combined with tersoff and the Lennard-Jones potential for this case? Please help me!
In addition, sometimes the atom are blown up. I mean that the atom are not binded.
Please let me know!

[Petr Khomyakov]

Could you give more computational details of how exactly you have been doing these calculations? It might be helpful if you enclose a python script related to geometry optimization of the structure.

[Fishpack]

Sure, Thank you!
Please check the file attatched.

[Petr Khomyakov]

The target structure seems to have a single phosphorous atom (i.e., not multiple Al atoms) as a point defect acting as dopant. So, you have no target structure.

I do not know if Al atoms can go substitutional in the graphene lattice. The Al atoms might rather prefer to be on top of the graphene sheet.  In any way, you may also consider using other force fields available in ATK-2016.4 or 2017.b1 to study this problem. 

[Fishpack]

Thank you for your reply.
I could obtain the structure what I wanted to.
However, now I'm trying to make a atomic structure of defective graphene with PtO2.
Based on the defective graphene layer, I want to attach a PtO2 to C atoms that has a dangling bond.
(I think the attached file will be helpful to understand what I want. The final structure will be C-O-Pt-O).
I hope you tell me about it!

[Petr Khomyakov]

What is your actual question?

[Fishpack]

Sorry for my post.
I'm not a native English speaker. So, I feel hardness to express.
Anyway, my purpose is to make a atomic structure as shown in the attached file.
At the C atom that has a broken bonding, I want to arrange the O, Pt, and O atoms with bondings and connections.
I'm not accustomed to use the interface of the  Builder in the VNL.
Please let me know and I hope you understand my purpose.

[Petr Khomyakov]

I would suggest having a look at http://docs.quantumwise.com/tutorials/builder_manual/builder_manual.html, especially, reading the section on 'Move Tool'.

As a general guideline:

1. Add PtO2 crystal from the Database to the Stash.
2. Convert it from bulk to molecular configuration, using the corresponding plugin in the Builder.
3. Select and delete all the atoms except the O-Pt-O atoms aligned in a row. This is how you get your O-Pt-O molecule.
4. Rotate the row of atoms such that it is aligned along the z-axis, using the Rotate plugin in the Coordinate Tools.
5. Click on the actual graphene structure in the Stash.
6. Drag-and-drop the O-Pt-O molecule from the Stash to this structure to merge the molecule and graphene structures.
7. Select the O-Pt-O atoms.
8. Use Move Tool to position the molecule where you want on the graphene sheet.

The detail information on the Builder plugins is given in the tutorial I have mentioned. 
 

[Fishpack]

Thank you so much!

[Fishpack]

I've tried to perform the molecular dynamics with the structure of the upper post (C-O-Pt-O)
(Not DFT calculation)

Well, the oxygen and platinum atoms are blown without binding.

The used potential was the LJ potential.

I need your help.

[Julian Schneider]

The Lennard-Jones potential is probably not a suitable potential to model bonding between atoms, as it is a non-bonded potential which only accounts for dispersion interactions (plus a repulsive part).
The best potential and parameters depends on what you want to simulate.
If you already know that the atoms will bond and how they will bond and you want to maintain the bonding in an MD simulation, a Morse-potential (http://docs.quantumwise.com/manuals/Types/MorsePotential/MorsePotential.html#morsepotential-) might be a better choice than LJ to keep the bonds stable.
If you don't know if and how the atoms are bonding then it is more difficult to select a  suitable potential. In this case it might make more sense to run a DFT calculation (at least on a small part of the system that contains your C-Pt-O2 group) to understand how that part behaves.