k-points for transport calculation

[smd]

Why we need large k-points for transport calculation as the semi-infinite electrodes contain self-energies? How many k-points in the transport direction will be required for a defected CNT with pristine CNT as an electrode?

[Petr Khomyakov]

It is required to accurately compute the Fermi energy in the electrodes. It is impossible to say exactly how many points you need, as it is a matter of convergence tests. Typically, if you have convergence issues with SCF device calculation, one of the possible reasons for that is inaccurately-computed Fermi level. In general, one should also converge the k-point sampling in the lateral directions as well - for CNT, that would just mean that you should have sufficiently thick vacuum padding around the tube to avoid spurious interaction between periodic images of CNT.

[smd]

Thank you sir. I am getting the same results for transport calculations even if I am increasing the k-points in the lateral direction or keeping it 1.

[Petr Khomyakov]

For modeling a single CNT, you should use a single (Gamma) k-point in the lateral directions, and check if the results converge with respect to vacuum padding thickness.

[Nayab Shiraz]

can you please explain what you mean by vacuum padding thickness?

[Petr Khomyakov]

When doing atomistic simulations, one has to set a finite simulation cell with the corresponding boundaries, meaning that for any systems such nanotubes or 2D materials one should assume vacuum or any other dielectric medium around these systems. Separation distance between the nanotube and simulation cell boundaries can be seen as vacuum/dielectric padding thickness.