What's the definition of bias voltage?

[lknife]

Hi, everyone,

When I calculate the IV curve, what's the definition of the bias voltage, V_R-V_L or V_L-V_R? Here V_R and V_L represent the chemical potential of the right electrode and the left electrode, respectively.

Thank you very much for your attention!

[Petr Khomyakov]

It is explained in a detailed note in the main tutorial on ATK transport calculations, see https://docs.quantumwise.com/tutorials/atk_transport_calculations/atk_transport_calculations.html.

[lknife]

Thank you very much for your reply. After carefully reading that tutorial, it seems that the definition of the bias would be V_L-V_R. "" However, according to my data, it looks like the opposite case. Please see the attached files:

It is shown in the .txt(.log) file that some of the SCF calculations for an IV simulation did not converge. According to the definition above, the bias should be -0.24V, -0.27V and -0.30V. However, according to the IV plot, they are mostly occurred at 0.24V, 0.27V and 0.30V.

Could you please help me with this problem?

Thanks a lot again for your help!

[Petr Khomyakov]

It would be helpful to see the python script related to this calculation.

[lknife]

The script was posted on another topic:
https://quantumwise.com/forum/index.php?topic=5412.msg23316#new

[Petr Khomyakov]

Which version of ATK are you using? Please post the actual, full log file for your script.

[lknife]

I have a permanent license for ATK2015 and another license for ATK2016, which was expired a few days ago and its renewing is on the way.

Because the full .log file for the IV calculation is too large, I cannot post it here.

I have known what's the bias is defined in ATK. Now my question has been changed to: how to improve the convergence for the calculation of transmission spectrum, could you please give me some advices?

When I tried to calculate the IV curve, it did not converge under several bias. So I want to redo the calculation bias by bias. I used the previously convergent device configuration as an initial guess to calculate the transmission spectrum under the next bias. I have successfully done the calculation under bias of 0.18V, however, it failed to converge under 0.21V.  Attached are the .py and .log file for this calculation, as well as the transmission spectrum under bias 0.18V.

I appreciate you if you can kindly give me your valuable advice on how to improve the convergence of the calculation.

Thank you very much again for you help!

[Ulrik G. Vej-Hansen]

Convergence under bias can be tricky, but I suggest you try the following (in arbitrary order):

* Increase the electron temperature to 1000 K.

* Increase the k-point sampling in the c-direction.

* Decrease the damping factor to e.g. 0.05.

* Increase the number of history steps to e.g. 20 or 25. Note that this requires more memory.

* Change the electrode voltages to 0.105 V and -0.105 V. Same absolute difference, but different shifts.

[lknife]

I do thank you very much for your valuable advises!

For the dampling factor and the history steps, due to my poor knowledge about ATK and misunderstanding of a post in the forum, I used to do the wrong operation: to increase the damping factor and decrease the history steps, in order to improve the convergence. Thank you again for your guidance.

[Ulrik G. Vej-Hansen]

You are welcome, and when it comes to those parameters it is somewhat system-dependent what the right approach is. So in some cases, the opposite would be the best way to go