QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: LocalSpin on July 29, 2022, 04:45

Title: SpinTrasnferTorque vs calculateLinearResponseSpinTransferTorkance
Post by: LocalSpin on July 29, 2022, 04:45
Dear ATK Experts

I am learning the calculation of  spin transfer torque using ATK. I reproduced the tutorial magnetic carbon-chain device using the parameter "SpinTransferTorque", which evaluates the torkance in real space (denoted as tau_R here). I got the results the same as the tutorial, as shown in the attached figure (top pannels).

And I also try to use the parameter "calculateLinearResponseSpinTransferTorkance", which returns the atom-resolved torkance (denoted as tau_atom here). The results are shown in the bottom pannels of the attached figure.

The red dashed lines denote some atomic positions in the device. I found there are differences between the results from these two parameters, for examples:  The maximums of tau_R and tau_atom have the opposite sign.  And in some case, espacially for tau_y in the middle pannels, some peaks of tau_atom appear at the positions where the tau_R are not large.

So I have some questions about the two parameters

1.  Why there are large differences between tau_R and tau_atom  (the opposite sign, the different positons of maximums)? Which one should I trust?

2.  I thought the maximums of tau_R should appear exactly at atomic positions. But at atomic positions tau_R curve shows a valley between two peaks appear above and below the atomic positions. Why do the peaks not appear at the atomic positions?

3. Can I qualitatively estimate the layer(atomic)-resolved torkance by averaging the peak values around the layer(atomic)-position?

I am using QuantumATK 2022-SP1.

Thank you very much!