I am working on the thermal conductivity from the EPI for doped materials like the example of B-doped diamond.
I would like to know the reasonable minimum degaussw. That’s because I probably need a very small value since the doping move the Fermi level below VBM by only 3-100 meV. On the other hand, the value cannot be too small because of the native phonon effect. For scf and nscf, I use 1 mRay degauss. Please also comment on this value.
I look forward to your comments. Thank you very much!
Tianshi Wang
Ph.D. candidate, MSEG
University of Delaware, DE USA
You will need to run convergence tests for various smearing values as well as k- and q- fine meshes. In principle you can use a smearing value as small as 1 meV but you will need very dense meshes. I would say a values of 10 meV is good in most cases, but even in this case you will need to check convergence with respect to the k- and q-meshes.
So a smearing (degaussw) less than 1 meV is usually not needed. The degaussw and k-mesh size are correlated during the convergence test. Therefore, is it ok that I set degaussw as 1 meV and only test the size of k and q meshes?
Do you have any experiences on how large the mesh should be? I am using a [i:218zm8ps]k[/i:218zm8ps]-mesh of 100x100x100 and q-mesh 35x35x35 which doesn’t seems very well.
Unfortunately, without testing there is no way to know if a value is converged or not. You can do smaller meshes a see how the value of interest changes.