2
$\begingroup$

I am trying to perform electron transport calculations. At the level of the scattering region calculations, I have noticed that the values of energies are no longer changing:

ts-scf:  123  -103750.909586  -103750.909586  -103750.909586  0.000000 -3.126529  0.000000  
ts-q:         D        E1        C1        E2        C2        dQ  
ts-q:   756.236   376.068     7.932   376.068     7.932 -.383E+01  
ts-Vha:  0.70752E+00 eV
..
.. 
..

ts-scf:  250  -103750.909586  -103750.909586  -103750.909586  0.000000 -3.126529  0.000000
ts-q:         D        E1        C1        E2        C2        dQ
ts-q:   756.236   376.068     7.932   376.068     7.932 -.383E+01
ts-Vha:  0.70752E+00 eV

as you can see, from step 123 to 250 all the values are the same and the same goes for the steps ahead. The values are no longer changing !!

Does this mean that the calculations are already converged and must already stop or is it normal and I should wait?

Or is there something wrong with my calculations?

$\endgroup$

1 Answer 1

3
$\begingroup$

In SIESTA the only conditions for convergence are that the Hamiltonian and the Density matrix are self-consistent. I.e. dDMax and dHMax are below some threshold. These are printed as the 5th and 7th column of the ts-scf line. So that condition is completely fulfilled, as they are both 0.000000.

However, TranSIESTA has an extra requirement. dQ must be below some threshold. dQ is the extra charge in your system, and it should be 0 or very close to it. In your case you have -3.8 electrons less than what you should have, and since that is not acceptable the SCF cycle will never end. Since the Hamiltonian and Density Matrix are already converged, dQ will not change anymore.

Assuming your electrode is a metal, the typical reason for having extra charge in transiesta is that there is not a long enough section of electrode material between the electrode and the scattering region. This makes it impossible for the charge differences to be screened before reaching the electrode, which should have exactly the bulk electronic density. Try to add more layers of electrode material and see if it solves your issue.

$\endgroup$
4
  • $\begingroup$ Thank you @Pol Febrer for your answer. The electrodes are also semiconductors, could this be why I have extra charges? Must I use metal electrodes instead? I have already added more layers and I still have extra charges. $\endgroup$ Commented Feb 5 at 12:06
  • 1
    $\begingroup$ Yes, the electrodes must be metallic. I think you could make it work with a semiconductor, but you would need a huge number of screening layers. Did the dQ decrease when you added the layers? $\endgroup$
    – Pol Febrer
    Commented Feb 5 at 19:37
  • 1
    $\begingroup$ By the way when I say you should add more layers I don't mean to increase the number of layers on the electrode bulk cell (the one that you compute before the TranSIESTA calculation). You can use the same electrode cell, but when you compute your full device with TranSIESTA you should add more layers between the electrode and the scattering region (e.g. maybe a surface?) to screen the charge effects. I don't know if this is what you understood with my first suggestion. $\endgroup$
    – Pol Febrer
    Commented Feb 5 at 19:41
  • $\begingroup$ Thank you, I added more layers in the scattering region and now the charge is small and stable. $\endgroup$ Commented 1 hour ago

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .