Why does the IVCharacteristics script produce different results compared to manu

[zwh]

Hello, I encountered an inconsistency when using QuantumATK with the same device model. The data calculated using the IVCharacteristics script shows significant discrepancies compared to results obtained by individually setting each gate voltage with the same basis set. Attached are my IVC script, the loop script for separate gate voltage configurations, and the computational result graphs. Could you please help me identify what might be causing this issue? Thank you.
Additionally, my alternative approach involves first using the test script to calculate the SCF (self-consistent field) of the device, then employing the TE script (Transmission Eigenchannel) to extract the current information of the device.

[Anders Blom]

In principle there should not be a difference, however I am thinking perhaps this is due to numerics.

In the IVCharateristics workflow, a previous point is used to start the next one, whereas in your script each bias point is started from scratch. So the convergence patterns might be slightly different. It is therefore possible that the discrepancy goes away if a tighter convergence criterion is used.

In your results you have normalized it to A/m, but if there currents themselves are very small, the difference will be more pronounced.

[zwh]

Dear Mr. Anders Blom,

Thank you for your prompt response. Following your recommendations, I have carefully adjusted the convergence criteria in the test script to 1e-6 and conducted renewed validations. The updated script file is attached to this email. However, the test results remain largely consistent with previous data and did not yield the expected improvements.

I would be most grateful if you could investigate the root cause of this issue when convenient. Should you require any supplementary materials or additional test data, please do not hesitate to let me know.

Wishing you continued success in your work and good health.

[Anders Blom]

Can you share the HDF5 files so I don't have to rerun all calculations?

[zwh]

Dear Anders Blom,

The HDF5 file I need to share is quite large in size, making it difficult to upload via the forum. Would it be possible for you to provide an email address where I can send the compressed file directly? Your assistance would be greatly appreciated.

Thank you for your support.