As they advise everywhere, SKF should be considered without SOC, and already in NSCF, spin-orbit interaction should be included. But what is the actual question, as I know, band splitting occurs with SOC calculations and then the DOS that I take from SCF gives another band gap. How to deal with it?
$\begingroup$
$\endgroup$
3
-
3$\begingroup$ What k-point mesh do you use for the DOS calculation and at which k-point do you find the valence band maximum in the band structure calculation? $\endgroup$– Gregor MichalicekCommented May 14, 2022 at 18:17
-
$\begingroup$ @GregorMichalicek Hello I use 28x28x28 k-mesh (I think that's enough :)))) VBM in band structure calc equal 3.6500 eV $\endgroup$– namelessenkoCommented May 23, 2022 at 11:19
-
$\begingroup$ @namelessenko 28x28x28 doesn't include the gamma-point, and your VBM looks suspiciously like it might be at the gamma-point. Try a k-point grid with an odd number of k-points in each dimension, and see whether it improves it (e.g. 27x27x27). $\endgroup$– Phil HasnipCommented Jun 27 at 21:58
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
The DOS and Band Structure normally use a different set of k-points and I will assume that is what happened here. There are two possibilities for what we are seeing right now.
The first is that your lines for the band gap on both plots is not actually drawn correctly or aligned. You should use what looks like Sumo to actually get a numerical value for both.
The second is that the DOS may not have hit the correct k-points to actually see the correct valence band minimum or conductance band maximum. This would effect the DOS gap as well, showing it to be artificially too big.
answered May 16, 2022 at 14:42
-
$\begingroup$ I think it's some problem with aligned of graphics. Because accuracy of step in k-mesh is enough. $\endgroup$ Commented May 23, 2022 at 12:01