can one adjust Fermi level by setting the electrodes have same voltages?

[bubble]

Hi,
Now I have a question. If I want to adjust the Fermi level of a device. For example, if I set VL=VR=0 V, then I can obtain the transmission spectrum at zero bias. However, If I set VL=VR=0.5 V, can I obtain the transmission spectrum (at equilibrium) corresponds to the whole Fermi level is shifted by 0.5 eV?

Thank you.

[Petr Khomyakov]

Could you clarify your question? If you have a transmission spectrum calculated for zero bias, you may still calculate the current for a given bias voltage, but this is not a self-consistent procedure, meaning that  you may miss some important nonlinear effects.

I also note that if the bias voltage is not zero, the system is then out of equilibrium, i.e., it cannot be described with a single Fermi level. 

[bubble]

  Sometimes we need to change the Fermi level of  a system. Of course, there are many different methods for this. For example, by changing gate voltages or by doping atoms etc.

  Now, my question is, for the device configuration in ATK, whether I can apply the SAME voltages to the left and right electrodes to shift the Fermi level of the device?  For example, if I set VL=VR=0.5 V (VL, VR are the voltages of the two electrodes). Then,  in present case,  the device is still in equilibrium (there is NO current). After SCF loops, I will obtain many transport properties. e.g. transmission spectra.  I want to know whether the obtained spectra are the result  after the Fermi level is shifted by 0.5 eV?

[Petr Khomyakov]

No, the Fermi level is not supposed to shift because there is no potential gradient across the system, but just a trivial constant shift of the electrostatic potential, which is in any way defined up to a constant.   

[bubble]

Thank you for your suggestion.

I calculated an example, i.e. the Fe-MgO-Fe MTJ in the database. I set the VL=VR=0 or VL=VR=0.5 V for the device (see the attachment for the two input py files, here SE method is employed, version=ATK2017.2). But I found the spectra are very different (see the third attachment file). How to explain this?

[bubble]

 

[Petr Khomyakov]

I do not know if the SE method is an appropriate approach to this problem. In any way, it seems that all your calculations are done non-self-consistently. I would guess that doing a self-consistent calculation (for that you should untick 'No SCF iteration' in the Device Calculator settings) is more adequate and should supposedly give you identical transmission spectra for the two cases you have mentioned.