A question about transmission spectrum of spin transport

[Peng]

I started to run a spin  transport calculation of the molecules with 86 atoms. Due to the deficiency calculating resource in my department, I only used two different metal atom chains as electrodes, the one is Nickel, to inject the  spin, the other is Gold. It ran well. However it always happened to something strange in my spin transmission spectrum. No matter how I change my model, some of the real-up down and imag-up-down transmission  coefficient are nagative. I do not know if whether this phenomenon is caused by my asymmetry electrodes or other reasons.  I am a new beginner. Would you tell me what is the problem here?
Thanks!

[Petr Khomyakov]

Transmission coefficient is a positively defined quantity. If it turns out to be negative, there must be something wrong.  Please let us know what is the QuantumATK version you are using, and attach the python script and log files related to the trouble calculation.   

[Peng]

Thanks a lot for your timely reply
I am using ATK in 2017.0 version.  Here are my python script and log files. I also attach the part of data in transmission spectrum below.

[Petr Khomyakov]

You do electron transport calculations with spin-orbit interaction included. The diagonal elements of the transmission matrix have positive values, as it should be, and the total transmission, which is the trace of the transmission matrix will then be a real, positive number as well.

The off-diagonal elements are actually complex numbers, in general, as you may have noticed. And they become zeros when the spin-orbit coupling is turned off. So, I see no particular reason why the real and imaginary parts of these matrix elements have to be necessarily positive. For more information, take a look at this tutorial, https://docs.quantumwise.com/tutorials/noncollinear_intro/noncollinear_intro.html. You may also consider reading some literature on the transmission matrix properties when the  spin-orbit interaction is included in the electron transport calculations.

[Peng]

The supplements you gave to me is really hopeful. I had misunderstood the intrinsic meaning of elements in the transmission matrix before.   Thanks a lot.
As T(up) and T(down), which I put in the picture above, they represent diagonal elements exactly. The spin-orbit interaction is indepensable for my system.

I also would like to extract two off-diagonal elements from my results, how can I get T(real-down-up) and T(imag-down-up) in the using of ATK?

[Petr Khomyakov]

I think it is in the tutorial mentioned in my previous reply, see also https://docs.quantumwise.com/manual/Types/TransmissionSpectrum/TransmissionSpectrum.html. You also have these components in the  transmission spectrum report (in Text Representation)  that you had attached to your post as an image file.

[Peng]

 :)Thank you for your all replies again.