Messages

781

Scripts, Tutorials and Applications / Re: How to apply a force on the bulk system

« on: December 27, 2008, 09:20 »

Hey Rosen.

I found this script that I have made some time ago. It is a module for creating a strain on bulk configuration
in order to model the behavior of the bandgap under strain for GaN.

I have uploaded the functionality here, and the results I got from my studies on GaN.

Perhaps you can use it for you work as well.

782

General Questions and Answers / Re: fail to diagonalize matrix

« on: December 26, 2008, 16:27 »

Hey!

I have looked a little into your system, but I haven't figured out the troubles yet. However I spotted a fatal error in your script.
The electrode cell of the nanotube must be repeated at least 3-4 times more or else you will get incorret transport results!
I have written in another post the reason and the way it is seen (Incorrect transport for short electrodes)

783

General Questions and Answers / Re: bias voltages

« on: December 26, 2008, 11:23 »

I found this test example for Li-H2-Li, where I compare the Linear Response Current in ATK with the fully selfconsistent current from ATK.

I have attached the graph.

784

General Questions and Answers / Re: how to choose energy range

« on: December 26, 2008, 10:59 »

You are welcome

785

General Questions and Answers / Re: how to choose the eigenchannel index

« on: December 26, 2008, 10:59 »

Inside VNL you do it by dropping the twoprobe on the NanoLanguage  scripter, and selecting the analysis tab. Select the transmission eigenstates,
and the dialog below will show all the input data. In the field Eigenchannel index you simply enter the number (starting from zero) of the eigenchannel you want to calculate. Where 0 being the one with the highest transmission, and 1 then one with the second highest transmission and so on.

786

General Questions and Answers / Re: fail to diagonalize matrix

« on: December 25, 2008, 12:19 »

Looking at the electrode by themself, they look correct.

1) When does the error happen exactly ? In the electrode calculation, in the equvialent bulk calculation or twoprobe calculation?
    Or is it first when you are trying to do the MPSH calculation the error happens?

2) If there is anything that could be wrong judging from the file, since I don't have the scattering region, it could be the equvialent atoms that are incorrect.
    However if the systems looks correct in VNL, it is properly not the case, and must be something different.

787

General Questions and Answers / Re: how to choose energy range

« on: December 25, 2008, 12:07 »

It is hard to setup a general guideline for matters like these, however when this is transport calculation, then most interesting things happen around the fermi level.
Therefore the energy range should always included the zero energy, and therefore I think that renren12313 suggestion about the energy window should be from -2 ev to +2 eV is good. I usally do it from -5 eV to 5 eV just to get all the details of the spectrum, even if the range from 2-5 does not provided alot of information.

Regarding the enery spacing (aka the resolution) it depends on the system you are working with. If you are studying the transmission in ideal systems, the resolution could be quite low (~0.1 eV), however if the systems involves both spin-polarized metals and high band-gap semiconductors interfaces etc, then you should go with a better resolution and use something like ~0.01-0.05 eV. But again it is question of what kind of details you want, since increasing the resolution (lowering the energy spacing) will also make the calculation more time consuming.

So for me it always boils down to: Do I want to see the trend (higher energy spacing) or do I want to see the details? (low energy spacing)
For example I have posted a series of transmission spectrums on strain of atomic wires ,
where the goal was to understand the what happend when a wire go strained more and more. Therefore I went for the trend and used a higher energy resolution.

788

General Questions and Answers / Re: fail to diagonalize matrix

« on: December 24, 2008, 12:39 »

Alternative you can try inspecting the system in VNL Nanoscope to see if there is anything wrong with the geometry.

789

General Questions and Answers / Re: a spin problem

« on: December 24, 2008, 12:38 »

Hej Renren123123.

I am not sure exactly what you mean by your question, but I will try to answer it anyway.

When dealing with periodic systems (and two-probes) you can set the initial spin for
the system atom by atom. This is qualified guess, however if you don't have any idea what the initial spin
should be, you set the initial scaled spin to 1 for all the atoms.

In VNL under the option "initial scaled spin", if you enter a single value there, it will use this value for all the atoms
in the system, and hence you can just enter 1 here at it should work.
The results will not depend on this value since it is just a qualified guess, but you will perhaps have longer calculation time
if the self-consistent density is far away from your guess.

I hope this answered your question.

790

General Questions and Answers / Re: bias voltages

« on: December 24, 2008, 12:32 »

I guess it is matter of opinion and to what degree one can accept the approximation of the linear-response approximation.
I have done two cases studies of it once, I will try to see if I can find them, however I remember a couple of things for sure.

If the system you are calculating on, is typical standard system without spin polarization and perhaps have ohmic resistance,
then it goes something like this.

0-0.2 V Technical no deviation.
0.2V-0.4V Small deviation, but still within the model precision.
0.4V-0.7V Slightly deviation, but still shows the same behavior
0.7V-1.0V The deviation becomes more than 10-20%, and this could be considered high bias.
1.0V -      Typical large errors.

However if the target systems is a MTJ's (Magneto-Tunnel-Junction) like FeMgOFe, where you want to understand the details of the spin current
in order to calculate the spin-torque etc, then I would set 0.3 V as a "high voltage".

But there are no need to despair, since one can always use the fully selfconsistent calculate current functionality to get the right results.
Now I will go look, if I can find my case studies on this matter.

791

General Questions and Answers / Re: fail to diagonalize matrix

« on: December 24, 2008, 11:17 »

Usally this problem happens when two atoms are placed on top of one another. Perhaps if you post the script, I can try to see if I can see the error.

792

General Questions and Answers / Re: How to make a model from other soft

« on: December 22, 2008, 21:20 »

I tried to do a little case study with the script and I tryed to simulated how the transmission spectrum depends on the strain of a lithium wire.

I tried to strain a perfect lithium wire by straining it by 1 Ang to 3 Ang over at distance of 5 atoms, and then took a look at how this effect the transmission spectrum.
I have attached the result in a little graph below.

793

Scripts, Tutorials and Applications / Atomic Chainbuilder.

« on: December 22, 2008, 14:13 »

I have also posted this script under the Q&A, but I think I would share it in here as well.

The script should be adjusted to your needs, but no worry the script should be self-explained,
and it can be edited straight forward. When you are done, you can either drag and drop into the VNL
in order to visualize it or perform a calculation on the wire.

The script can build any mono-atomic wires, it will also allow your create defects/donor atoms into the
wire. It will also make you able to create a buckling of the wire. Futhermore you can also put a strain on
your wire, if it is that you are desired.

Please feel free to use - I would like ideas to improvements!

794

General Questions and Answers / Re: How to make a model from other soft

« on: December 22, 2008, 14:09 »

I have reworked a old atomic chain builder, I found.

I have attached it below, feel free to use.

The script should be adjusted to your needs, but no worry the script should be self-explained,
and it can be edited straight forward. When you are done, you can either drag and drop into the VNL
in order to visualize it or perform a calculation on the wire.

Below is a picture from VNL of a very strained gold wire with a Hydrogen defect causing some buckling in the middle.

795

Scripts, Tutorials and Applications / Re: How to apply a force on the bulk system

« on: December 21, 2008, 23:26 »

If you are looking into structures under stress/strain, you might want to be interested in trying out
my improved relaxation routines for bulk-crystals , that takes into account the stress and strain of the crystal.
You can find the script and details here (Improved optimizers).