QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: perfetti on June 27, 2012, 18:01
-
Dear Everyone,
I have a big trouble now: I am optimizing the geometry of my two probe device. It's a very basic one, composed of one copper chain absorbed on carbon nanotube, and then I performed a full device configuration optimization of my setup. However, I found that my optimized geometry is much different from what people got in literature.
My initial geometry is like this: 1.jpeg
While my after geometry is like this:
2t.jpeg.
While in the one in literature, they usually get geometry like this:
3.jpeg.
I know there are surface layers in my structure, and in ATK they constrained the electrodes automatically, however, I didn't expect such a result. I am not sure if this optimization is meaningful or not, since it looks much different from others, and I don't know if the following properties calculation from this optimized structure are reliable.
Any one giving advice is very appreciated. Thank you very much.
-
Can you share a view from the top of your initial and final configuration?
-
The top views have been attatched.thx.
Can you share a view from the top of your initial and final configuration?
-
Seems to me that your tube is armchair and the article picture shows a zigzag tube.
-
Dear Dr. Blom,
That's True. But will that cause such a big difference?
I see in armchair CNT case, others also got a similar pattern after optimization, as the article pattern.
I am just confused.
Seems to me that your tube is armchair and the article picture shows a zigzag tube.
-
Dear Perfetti.
Since a device optimization comes with some constraints, you must first focus your optimization on the electrode where the wire is aligned perfectly on top of
the carbon nanotube. I haven't look into your exact structure, but I think the main problem is that the structure starts in a construction with such a symmetry
that all the forces cancels one another. This you can avoid by using the rattle tool in the builder just to shake the structure out of this high symmetry state.
The other thing is that I think that your metal chain has too few atoms in it, to make this "snaking" as reported in the paper. Perhaps if you add more atoms per length to the chain it will look better.
But still if I were you, I would focus on the electrode first as the problem is not related to device at first.
-
Dear Nordland,
Thank you for your advice.
So that you are meaning, I should optimize the electrode first.And then repeat the electrode to the central part?
There do have zero forces in the electrode part and the surface layers from the very beginning.
What do you mean by this: "This you can avoid by using the rattle tool in the builder just to shake the structure out of this high symmetry state." especially the rattle tool?
Thank you.
Dear Perfetti.
Since a device optimization comes with some constraints, you must first focus your optimization on the electrode where the wire is aligned perfectly on top of
the carbon nanotube. I haven't look into your exact structure, but I think the main problem is that the structure starts in a construction with such a symmetry
that all the forces cancels one another. This you can avoid by using the rattle tool in the builder just to shake the structure out of this high symmetry state.
The other thing is that I think that your metal chain has too few atoms in it, to make this "snaking" as reported in the paper. Perhaps if you add more atoms per length to the chain it will look better.
But still if I were you, I would focus on the electrode first as the problem is not related to device at first.
-
Let me try to set up something simple up for you. Be right back.
-
My assumption was correct. The problem is that your electrode is not optimized prior to building the system, and therefore your device ends up being incorrect.
I have tried to setup the electrode calculation and then I get the same structure as you present in the papers.
Please feel free to use my structure if you want.
-
Dear Dr. Nordland and Dr. Blom,
What you said are very true.
Dear Dr. Nordland, thanks for your patience and advice. It's very helpful.
On the other hand,could you tell me how did you define the distance between the cu chain and the carbon nanotube? Are there any rules?
Best regards.
-
I just added them on top of the two carbon atoms in some random distance, and then I did a optimization using DFT.
-
Dear Dr. Nordland,
I have two questions:
(1) It seems that the Cu atoms in your device are not exactly on top of the carbon atoms. It seems that you moved the Cu atoms' Z coordinates a little bit, and then get the snake result. Did you do that by intended, or just did that by chance? How did you decide how much should the Cu atoms been moved?
If I put the Cu atoms directly on top,without changing the coordinates, I would get a straight line instead of a snake pattern.
(2) If I moved my device's Z coordinates at the same amount as your device, I would get a snake line, but very ugly. I am not sure if I did all the things right, so I just attatch my script and optimized structure, hoping you could tell me if it's right.
-
1) Regarding the small displacement, it was not intended, but needed ( so i was lucky ). It goes back to the point of my initial statement about the rattle tool.
If you look at the structure with the Copper atoms being placed exactly just above, you can from a purely geometrical point of view, see that any forces between a Copper atom and a carbon atom in the x-y plane, would be mirrored in this plane in the center of nanotube, making this configuration to a saddle point for the forces, and therefore you would never seen the copper atoms moving out of this line. This is why I, personally, always when building configurations of this type, hits the rattle button in the Builder a couple of times to ensure that the configuration is not locked in some geometrical, but artificial, sub-space due to a too symmetric start configuration. This ensures that if the structure is strictly symmetric, then I know it is the true, because my starting guess was a random asymmetrical configuration.
2) If I repeat my system, I get the following periodic structure, and I think it is pretty :)
-
Dear Nordland and everyone,
Thanks for the advice, and I did it successfully for the zigzag case. However, not for the armchair one. I tried to optimize one metal atom one unit cell(the same as the literatureļ¼ , and I tried two metal atoms in two unit cells, however, both I got finally are the same case: The metal atoms are on the right top of the structure, without a snake sign.
I am attatching the ideal picture that I want to achieve; it's using one metal atom per unit cell for an armchair CNT.
Thanks for any one who could give me advice.
-
I must have a golden touch for building these systems, because I have gotten it to work for the armchair as well. :)
There is one Copper atom per unitcell, but the desired structure is not periodic with the unitcell for the armchair carbon nanotube,
so you will need the double cell and two copper atoms in your structure to get the correct result.
Recipe for building the system:
- Build a CNT (7,7)
- Repeat it twice in the Z-axises
- Select a Carbon atom, open the translate dialog, tick of copy, and move it in the direction away from the center. ( In my case this was
[-1.5,1.5,0] Ang - Select a neighboring ring to the carbon atom selected before, and press the geometrical center tool, and move this new atom in the same direction as before.
- Select the two carbon atoms outside that are now outside the tube, and using the periodic table tool, change them to Copper.
- Send-to the Scripter and optimize it.
I tried relaxing it with crap parameters and this is what I got.
-
Dear Dr. Nordland,
I really need your help.
1) I used the structure.py you gave to me, and I got a straight line.
2) I followed your recipe and moved the atoms at 1.5Angstrom, and then relaxed. And I got another straight line.
The structure in your structure.py is not exactly 1.5Angstrom away from x,y, so I used both coordinates. However, both I got are straight lines. So I really don't know what to do next.
I am attatching the scripts I used as well as the images I got.
Plz give me some advice.Thanks.
-
Okay I tried a couple of different things and ended up with the straight line and then I said to myself: What is the problem with this structure?
The bond between the two copper atoms must be stretched more than 20%, and therefore the normal FHI pseudo-potential is known for not being really super in the long range limit, and therefore I tried using the new HGH pseudo-potential and the tier 5 basis set.
And using the Copper basis set which is really excellent for Copper also in the long range, you can get the chain to snake like desired.
I am just redoing the calculation now, because I managed to delete the output, but you can have the input file.
-
Thanks. Dear Dr. Nordland.
This time the shape after relaxation seems better, however, there's no bond between Cu atoms. I tried to move one Cu atom almost 0.2 A closer to another Cu atom so they'll have bonds at first, but there's still no bonds after relaxation.
The picture of relaxed structure is attatched.
-
Allow me to quote myself :)
The bond between the two copper atoms must be stretched more than 20%, ....................
The bonds in VNL is purely visual, and if you open the properties window in the viewer, you can see that it will draw bonds between atoms that are within by default 0.5 to 1.1 of the covalent radius of other atoms. You can increase this maximum range, and if you increase it to 1.20 I think you will see the bonds between the copper atoms as well. Tune this parameter and you can get the plot you want :)
-
Thanks, Dr. Nordland.
I really learned a lot from you.
Have a great weekend.
:D