Electronic contribution to thermal conductance

Asked by David Guzman on 2017-11-12

Hello Nick,
Is there a simple way with Transiesta/TBTran to compute the electronic thermal conductance?
Do you have any suggestions?


Question information

English Edit question
Siesta Edit question
No assignee Edit question
Solved by:
David Guzman
Last query:
Last reply:
Nick Papior (nickpapior) said : #1

If you are thinking about Eq. 31 in 10.1016/j.cpc.2016.09.022 then yes, TBtrans calculates it (and writes it out in the output).

You simply need to recompile TBtrans with netcdf-4 support (see the manual).

David Guzman (guzmand) said : #2

Hello Nick,

Sorry to bother you again with this. Two more questions if I may:

1- I am not sure how Eq. 31 of the paper relates to the standard way of computing the kappa_e:
κel = (2/hT)(K_2-(K_1^2/K_0)
Kn=int (E − μ)^n τ_{el}(E)(−∂f (E,T )/∂E)dE.
These equations 1 and 2 in https://arxiv.org/pdf/1709.00220

2- I am using TBtrans with netcdf-4 support but I do not see any information about Qee' written in the standard output. Is there a flags that needs to be set in the input.fdf?

Thanks for your help,

Nick Papior (nickpapior) said : #3

Ok, so not eq 31. Eq. 31 is the heat flow between two leads due to the applied bias and temperature gradient. In the output of tbtrans it is denoted P (for power) and is always written just below the current (I). However, they are based on the difference between two fermi-functions.

You can use sisl (https://github.com/zerothi/sisl) to calculate the specific quantities in the paper you mention. In that one you can easily extract the energy, chemical potential, electronic temperature and transmission (\tau). Then the rest goes easy.

Currently those equations are not implemented, but should be quite trivial.

David Guzman (guzmand) said : #4

Hello Nick,

Thanks for you suggestions. I was able to work the equations out using SISL.

Thanks again,

Nick Papior (nickpapior) said : #5

Would you mind sharing the implementations such that they could be incorporated into sisl?

If you would, please mail me, or open a pull request or issue on github :)

David Guzman (guzmand) said : #6

Sure. Let me clean this up and will share it.

Nick Papior (nickpapior) said : #7

Thanks, that would be very welcome! :)