QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: hellboy on August 10, 2010, 17:41
-
Dear Sir,
Can you please check the geometry of the attachment. I want to see the I-V char of carbon nanotube with Cu electrodes.[APPLIED PHYSICS LETTERS 96, 102108, (2010)]. What should be the separation between CNT and electrodes? Is there a problem while extending the CNT into electrodes?
best regards...
-
Since the simulation takes a long time to complete, I am looking for the correct geometry from the experts.
I searched the forum but couldn't get anything closer to this geometry except Au electrode and Au wire example.
I want to know what will be the separation between CNT and Cu electrode.
-
There is no simple answer to that question. But you don't need a two-probe calculation to figure it out, just move the nanotube back and forth and compute the total energy of the equivalent bulk (probably a full relaxation is too complicated). Of course, this is still a very heavy calculation for such a system, but there aren't really any shortcuts. A good starting point is the common covalent radius of Cu and C, I guess. Or just relax a Cu-C dimer.
-
Dear Sir,
I want to see the I-V characteristics of a semiconducting (8, 0) CNT and also T(E,v) and DOS .
Can i use semiconducting (8, 0) CNT both as electrodes and central region, by cleaving then joining?
Will the I-V char show a semiconducting nature in this case or there will be no current at all ( order of 10-10 Ampere at high bias of 2 volts)?
Also, will there be differences in I-V char if i use metallic electrodes like copper ? (a metal-semiconductor junction may form)
-
Semiconducting leads will give no current. Certainly with metallic leads things will be different ;)
-
Thank you for clearing my doubt, this is certainly informative that semiconducting leads will give no current.
I used a semiconducting lead with a semiconducting central region and found current of the order of 10-10 amperes at 2 volts.
Then i radially compressed the central region and used radially compressed semiconducting electrodes similar to central region, and found current of the order of 10-6 amperes at 2 volts. Which may be due to semiconductor to metal transition, a common phenomenon observed in deformed CNTs. In case of lower bias voltages also there was an increase in current of the same order.
My question is that, Can't we comment on the conductivity of different semiconducting nanotube structures using semiconducting electrodes ?
Of course, the current obtained will be higher in case of metallic electrodes, and the ratio by which current increases may be equally valid for all CNT structures.
-
I would be careful to assume any degree of accuracy in the current for the semiconducting case. If it's 10e-11 or 10e-13 depends extremely strongly on the particular numerical parameters. This would correspond to two orders of magnitude if you compare the ratio to the conducting case, but the effect is not necessarily there in reality.
What you can say is certainly that there is a semiconducting-to-metal transition, and you can trust the absolute current in the conducting case. That should be enough for some nice conclusions, I hope :)
-
I have read that for semiconductor-to-metal transition, the transmission spectra should be above the Fermi level (0 eV).
But in this case, the transmisson spectra width widens at the Fermi level (at 2 volts bias), though the current obtained is high (micro-amperes). There are peaks observed in T (E) which could be the possible reason of high current.
So can we still say that there is a semiconductor-to-metal transition?
-
I'm sorry, I don't understand what you mean by "the transmission spectrum widens", or "the transmission spectrum is above the Fermi level"... ???
Also, the systems are quite different, it's hard to compare them directly. The s-to-m transition is geometry-induced, so you are radically changing the electronic structure.
-
Sorry, I wanted to say transmission near by 0 eV is zero.
Sir, please look at the pictures that are taken from "Appl. Phys. Lett. 94, 183506 (2009)"
In the top figure, the transmission width is wide (no transmission) in energy range (near by 0 eV)
In the bottom two pictures, there is transmission at 0 eV, which shows semiconductor-to-metal transition.
-
You are correct, these figures show s-to-m transition, and it seems your calculations do too?
-
Thank you very much sir for the discussion.
regards,
-
Dear Sir,
How do we provide electrode connection to metal-semiconductor CNT heterojunctions?
From [APPLIED PHYSICS LETTERS 96, 102108 (2010)], the separation distance between the a straight wire CNT and Cu electrode is 1.85 Å for end-contact and 1.20 Å for side-contact. These distances were determined by minimizing the interaction energy of the system to ensure a stable configuration.
Please look at the attached picture, the tube angle is troubling.
The atoms at the end of tube don't have same Z-coordinate.
regards,
-
This type of system (or rather, the kind you would like it to be) is not really possible to study with ATK, since the model as it is designed today relies on a perpendicular interface between the electrode and the central region.
-
Dear Dr. Blom,
I want to know how to skip the calculation for two probes, and only calculate the equivalent bulk? Thank you.
There is no simple answer to that question. But you don't need a two-probe calculation to figure it out, just move the nanotube back and forth and compute the total energy of the equivalent bulk (probably a full relaxation is too complicated). Of course, this is still a very heavy calculation for such a system, but there aren't really any shortcuts. A good starting point is the common covalent radius of Cu and C, I guess. Or just relax a Cu-C dimer.
-
The central region is a separate configuration by itself in the script (variable central_region), so you can just extract it (copy/paste the Python code, for instance) and use it like a regular bulk system.
-
1.I think I can simply use the Bulk mode to calculate?
2. I found that the Cu's covalent radii is 117 pm, while that for C is 77 pm. Does that mean the common covalent radii is 77 pm? So I should start at the distance of 0.77 Angstrom?
The central region is a separate configuration by itself in the script (variable central_region), so you can just extract it (copy/paste the Python code, for instance) and use it like a regular bulk system.
-
1. Well, it depends on what quantities you are looking for. You can't get any transport this way. But yes, if you want to compute it as bulk, then this is what you do.
2. Start...? Ah, this is for varying the distance from the Cu to the CNT? (Forum username not always obvious ;) ) Well, that's a bit too narrow, it might be hard to converge that close, but you can try.
-
Yeah I just want to get the distance that makes the system has lowest energy.
1. But I think here what you said by editing the script is equal to switch to the bulk mode, right?
Cause I have to edit the script to make it can be seen in vnl , and I have to edit it the way as bulk mode's script.
2. Is there any good standard way to decide the optimium distance? I mean, for the starting point?
1. Well, it depends on what quantities you are looking for. You can't get any transport this way. But yes, if you want to compute it as bulk, then this is what you do.
2. Start...? Ah, this is for varying the distance from the Cu to the CNT? (Forum username not always obvious ;) ) Well, that's a bit too narrow, it might be hard to converge that close, but you can try.
-
1. Correct
2. Well, you have it from the article already...