Convergence problem in electron transport for silicon electrode system

[coraline24]

Hi all,

I'm calculating the transmission spectrum of electron transport through a pi-conjugated molecule between 2 Si electrodes (attached structure below).
The transmission spectrum was calculated for bias -2V to 2V with 0.5 increment.

My problem is the transmission spectrum calculation of every bias except -2V converged.
I have already tried changing the "Damping factor" and "History step" in the Pulay Mixer as suggested by the QuantumATK manual. It worked with most of my calculations except for bias -2V.

For bias -2V I have tried changing damping factors: 0.1, 0.05, 0.03, 0.011, 0.01, 0.009, 0.008, 0.007, 0.005; history steps: 5, 10, 15, 20, 25, 30; mixture: damping factor 0.01 + history step 10.

But none of those converged. The energy is stuck at lower energy compared to zero bias energy or converged bias calculation energy.

So,
1. Could you give me any suggestions on what I can do next to make it converge?
2. Can you explain what is the meaning of the phenomenon that energy keeps fluctuating around a value in transport calculation?
3. Can you explain in more detail than the manual on the effect of changing damping factor and history step on the convergence?

Thank you,
Cora

[AsifShah]

Hi,

Did u try damping 0.01 with history steps 20? Try it, if you haven't yet!
It may work.

Also, what is ur electrode extension length? Is it doped or undoped?

[coraline24]

Hi,

Thank you for your reply.

Yes I have tried damping 0.1 with step 20. I have tried damping factors: 0.1, 0.05, 0.03, 0.011, 0.01, 0.009, 0.008, 0.007, 0.005 with step 20, nothing worked.

The electrode is undoped Si, you can check the electrode extension length from my attached model

[Anders Blom]

It might have something to do with the odd k-point shift seen in the input file. Try to get rid of that, and have the central region and electrodes have matching k-point sampling in X/Y.