Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - F. Fuchs

Pages: [1] 2
1
I just tried to reproduce the piezoelectric tensor from the manual (https://docs.quantumatk.com/manual/Types/PiezoelectricTensor/PiezoelectricTensor.html) as well. I am using version Q-2019.12. The result I got also disagrees with the manual. There is also different to what arlonne has posted earlier.
And yes, I just copied and pasted the input script from the manual.

Maybe some defaults have been changed from version to version? An update of the manual would be nice in this case.

The result I got is also wrong in the sense that the xx, yy, zz components are not zero. Thus, I suppose that the input script in the manual requires more precise DFT parameters.

Regards,
Florian


Result with Q-2019.12 using the input file from https://docs.quantumatk.com/manual/Types/PiezoelectricTensor/PiezoelectricTensor.html
+------------------------------------------------------------------------------+
| Piezoelectric Tensor Report                                                  |
+------------------------------------------------------------------------------+
|                                                                              |
| Tensor in units of [C/m**2]:                                                 |
|                                                                              |
|           x              y              z                                    |
| xx    7.44784e-01    9.67739e-01   -7.44377e-02                              |
| yy   -7.44414e-02    1.83366e-01   -7.44846e-02                              |
| zz   -7.44414e-02    9.67739e-01   -8.62789e-02                              |
| yz   -1.87937e+00   -7.51877e-04   -2.50626e-04                              |
| xz   -7.51875e-04   -1.33056e+00   -7.51877e-04                              |
| xy    9.77514e-03    9.26680e-03    2.69444e+00                              |
|                                                                              |
+------------------------------------------------------------------------------+

2
My student made a test calculation and set the keyword to False in 2019.12. The resulting transmission spectrum agrees with the result from the earlier QuantumATK version.

This gives us a viable work around. Thank you!

3
Dear QuantumATK developers,

we recently observed an artefact when switching from QuantumATK version 19.03 to 19.12. The transmission spectra, which we have calculated for two overlapping graphene sheets, shows a transmission of zero in a certain interval around the Fermi energy when using the newer version. You can find the comparison in the attachment including also the input scripts. There is no physical explanation for this artefact and we therefore assume a bug since the recent QuantumATK update (the results using QuantumATK 19.03 make perfect sense).

Thank you in advance.

4
Thank you for your answer.

I tried to increase the number of k-points further (up to 3001). In the attached image you can find a plot of the location of band edges versus the number of k-points in z-direction (red squares mean that an even number of k-points was used). One can see that this oscillatory behavior is also present at the highest k-point number under study.

It looks like it will never converge...

What can cause this strange behavior?

5
Dear ATK-experts,

I calculated the band structure of a couple of silicon nanowires (SiNWs) using DFT and the meta-GGA potential by Tran and Blaha with ATK 15.1.
For one single SiNW I observed that the Fermi energy E_F is not located in the center of the band gap. I tried to change some settings and finally realized that this depends on the k-point sampling.
Attached you can find the band structures for 51, 52, and 53 k-points in the direction of the wire (1 k-point in the other directions). For 52 k-points, E_F is located in the center. For 51 k-points it is close to the valence band and for 53 k-points it is close to the conduction band. I tried 31 k-points and 61 k-points and for those values, E_F is also in the middle of the band gap. (*)
I did not observe this behavior for the GGA functional (51 k-points). But when using the meta-GGA potential, E_F is also not centered for different tolerance values in the IterationControlParameters or different c-parameters (I varied these values for 51 k-points). The non-centered E_F also appeared before and after the relaxation of the SiNW. For all other SiNW under study, E_F was centered in the band gap.

Any idea what could be the reason of this? It is probably not very critical, because the band structure itself does not change at all for 50+ k-points. But it is still very puzzling.

The corresponding python-script is attached.

Thank you very much.

(*) I similar problem was discussed in http://quantumwise.com/forum/index.php?topic=3006.0 (even though they used ATK 14) and the problem could be solved with an odd number of k-points. In my case however, the problem seems to appear mainly for odd k-points.

6
Thank you very much for your helpful reply! Indeed, I achieve almost identical results with the 2015 version when using a different value for the c-parameter.
(Btw.: I observed a slight shift of the band structure with respect to the Fermi energy when looking at SiNW structures. This shift is however rather small; approx. 0.01 eV.)

Is there some more documentation of the used approximations, which you have mentioned, available? I couldn't find any, yet.

7
Hello modelling experts,

I made a couple of DFT calculations using meta-GGA functional by Tran and Blaha. I used ATK 12.8 before and now switched to ATK 15. I realized that the results changed. Two example observations:
1) In order to achieve the same band gap in bulk silicon, I now need to use a different value for the c-parameter.
2) I did some two-probe simulations using version 12.8 and looked at the local density of states. Near the electrodes I observed some deviation from bulk properties in the electrodes (it was already bulk-like further away from the electrodes, so it was not a convergence problem with respect to the device length) . This problem however,  disappeared when using ATK 15 (which is very nice!).

My question is: What changed between ATK 12.8 and ATK 15 related to the Tran and Blaha functional? In the release of ATK 14 I read "New MGGA library and implementation - now also with TPSS functional" and I suppose that this causes the deviations. I would like to know some more details. How meaningful are the results calculated with the older implementation in ATK 12.8?

Thank you!

8
Dear VNL developers,

When exporting a .png or .jpeg from the VNL viewer, bonds are drawn in front of the atoms. This seems useless. An image in favor of better illustration is appended.

Thank you for fixing this!

Details:
Build: 2015.1.5b36b25
Platform: Linux
Python 2.7.2
Qt: 4.8.5
PyQt: 4.10.3

9
You could make some test calculations with a longer electrode cell to see whether it influences your results or not (you must adjust your charge appropriately).

I suppose that 1 nm should be sufficient.
See also http://quantumwise.com/publications/tutorials/item/115-how-to-set-up-the-electrodes-properly-in-a-two-probe-system

10
General Questions and Answers / Re: Silicon Nanowire FET
« on: August 25, 2014, 10:22 »
Looks like your electrodes are semiconducting and therefore no noteworthy current is possible.

You should consider doping your electrodes. Either by using real dopants or by using the charge keyword available in ATK (if you have a new version of ATK, you should also use http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.atomiccompensationcharge.html as pointed out in this forum somewhere else.). The latter is also used in the Si-NW-FET tutorial.

11
Thank you for the information!

We now tried to surround the electrodes by a metallic region in order to shield the electrostatics from the boundaries (I basically added a cylindrical gate to the electrodes. This is easy because we have carbon nanotubes as electrodes). This seems to work pretty good.
However, we are now wondering if it is necessary to add a metallic region to the electrode copies inside the central region as well.
What do you think?

12
General Questions and Answers / Re: About the charge option
« on: April 10, 2014, 14:51 »
Thank you for the information!

One final question (related to ATK 12.8.2, since ATK 13.8 is unfortunately not available yet):
When I add additional charge to the electrodes, does this also increase the net charge of the central region?
To put it in another way: Can the additional charge of the electrodes flow into the the central region?

(I would think so because the system is an open system. However, this would make the additional charge to an unsuitable method for simulating a system with doped electrodes. But obviously it works and it is also used for the SiFET tutorial.)

13
General Questions and Answers / Re: About the charge option
« on: April 9, 2014, 14:26 »
I agree that only the position of the bands relative to the Fermi level should be important.

However, I observe the following:

When I look at the same system and increase the size of the unit cell, the slope of the curves in the previous picture increases. This means that the shift of the band structure values due to charge addition is increased by using bigger cells (only the values are shifted, the look of the band structure is not influenced). The Fermi level changes as a result, too.
By looking at the Fermi level versus the system size for different doping levels, I obtain the attached picture ("fermi_dependency.png"). Only for big cells, the Fermi level is independent of the cell size.

The observed effect is interesting for us because it correlates with the observed shift of the transfer characteristic of our CNTFET (see http://quantumwise.com/forum/index.php?topic=2670.msg12714#msg12714). The charged CNT is used as electrodes for the transistor.

We think that the observed effect (shift of the transfer characteristic) comes from electrostatics inside the cell, because the additional charge increases the electrostatic effects inside the cell. Therefore, a bigger cell is needed so that the manipulation of the field by the Neumann boundaries is small.

We know what to do against this effect (using small amounts of charge and big cells), but we don't understand the effect yet.
Maybe someone can enlighten us.

14
General Questions and Answers / About the charge option
« on: April 8, 2014, 16:01 »
Hello everyone,

I want to clarify myself about the meaning of the charge option for electrodes (ATK version 12.8.2).
It is clear that the Fermi level gets shifted when additional charge is added.

To estimate the influence of the electrode charge in a device configuration, I looked at one single (7,0) CNT (bulk configuration) and varied the charge (periodic boundary conditions along the CNT, Neumann boundary conditions perpendicular to the CNT (as this is needed for the device configuration)).
Afterwards, I subtracted the Fermi level from the resulting band structures (to see the influence on the band structure and the Fermi level separately).
 
The result is attached.
You can see the energy values at the Gamma point versus different amounts of additional charge. The red curve is the Fermi level, the black curves correspond to the band structure. The inset shows the band structure of the (7,0) CNT for 0.0 additional charge.

The movement of the Fermi level relative to the bands is clear. However, there is also a shift of the whole band structure in addition, which I don't understand yet (shouldn't the black curves be constant?).

Can someone explain this behavior?
I would also be interested in further information on how exactly the additional charge is added to the system.

Thanks in advance!


15
I want to illustrate what ziand said in the previous post.
I have therefore attached an image, where you can see transfer characteristics for various cell sizes (the size of the cell can be found in the legend).
Neumann boundaries were used perpendicular to the device to obtain these data. I have also attached one structure as an example.
Even for a cell of 166 Angstrom, the transfer characteristic is further shifted towards smaller gate voltages.

This effect may be related with the doping of the electrodes (which was set to 0.5 using the charge keyword in ATK 12.8.2) and resulting electrostatics. However, I would not expect such a drastic shift.

Any ideas what could be the problem?
Should I maybe also adjust the value for the charge keyword when increasing the cell extend?
(since an increasing doping leads to a shift towards smaller gate voltages during our simulations, too)

Pages: [1] 2