optimisation of geometry of large molecular junction

[kavita]

HI.........
I am a new user of VNL. I trying to use this from past 8 months. I am working on molecular junctions. My central molecules are Quite large like 40- 200 atoms. I am confused that how many layers of electrodes to be attached for Different types of molecules. i.e what is the relation between no of atoms in the molecule with the no. of layers of electrodes.
Like when I am trying for molecule with 46 atoms.
1. I have attached 5 layers on each side, then it give error that no. of atoms are not equal on both sides of electrodes. But they are same in script.
2. If I put less no layers then molecule breaks.
3. If i take 8 layers then total no. of molecules is very large and it will not converge easily.
These are problems with such a small system of 46 atoms.
 then what about larger molecules like 129 atoms
for this I have put 11 layers on each side still it gives same error.
I don't know what to do.
please help me if anyone knows...
I am attaching XYZ file, inp file and error messages with this post

[Anders Blom]

There is no immediate relation between the molecule and the number of layers you need. The central region does however need to contain at minimum as many layers as the electrode (the electrode copy). Moreover, those atoms are no allowed to move in relaxations. Now, they are automatically constrained if you run the relaxation as a device system, but what seems to happen is that you have large forces on the atoms, which cause some atoms to move inside the region monitored by ATK (the electrode copy, which can't contain any other atoms either).

So, if you make the molecule larger or not (in Z, at least) will not have any influcence on the surfaces (only the repetitions in XY, in case it's larger in those directions).

The error will most likely go away if you add just one layer of Si on each surface.

Finally, it may be a good idea to do a pre-optimization of the system as periodic in Z. You can either try to set it "non-self-consistent" in the GUI, or extract the central region and constrain all the Si atoms. Then relax this, also the strain in Z, and use this to construct a new device (which you can relax or keep as it is, it should be an ok geometry to work with).

[kavita]

Thanx for the reply
I am a new user so I dont know how to keep Z constant. in other words i didnt understand your solution.
secondly my molecule contain Oxygen can I still do non consistent calculations.
Please tell me the procedure to solve this problem.
Like in steps or in script format.

[kstokbro]

The problem with you setup is that the outermost Si layer is also in the  equivalent electrode, i.e. you have 4 Si layers in the electrode and the same number in the central region.  So when you relax the structure the molecule atoms move into the equivalent electrode region.

The solution is that you have 5 electrode layers in the central region. This can easily be obtained using the extend electrode facility if you take you structure into the builder.

Another problem with you structure is that you have the wrong (111) terminating layer of the silicon electrode. Silicon has 2 different types of terminating layers, one has 1 dangling bond, the other has 3 dangling bonds. You have chosen the 3 dangling bond terminating layer, which is not observed experimentally.

[kavita]

how to change the dangling bond using vnl.
I have tried with 5 layers also there is same problem.
how can i convert 4 layers to 5 layers in builder please tell if u are talking about custom builder then i know, otherwise please tell me

[kavita]

PLEASE REPLY IF U ARE ONLINE

[Anders Blom]

Transfer the structure to the Builder (not Custom Builder) and use the + and - buttons to add a length corresponding to 1 layer to the surface.