Correct Band Gap in sumo-bandstats?

[jfrmhrz]

I have done a non-spin polarized band structure calculation of a TMD material, by checking the band gap type and value using sumo-bandstats, there seems to be Indirect and Direct band gap as follows:
2 | Indirect band gap: 0.924 eV
3 │ Direct band gap: 0.989 eV
4 │ k-point: [0.28, 0.00, 0.00]
5 │ k-point indices: 26
6 │ Band indices: 152, 153
7 │
8 │ Valence band maximum:
9 │ Energy: -2.089 eV
10 │ k-point: [-0.00, -0.29, 0.00]
11 │ k-point location: between X-Y
12 │ k-point indices: 7
13 │ Band indices: 152
14 │
15 │ Conduction band minimum:
16 │ Energy: -1.165 eV
17 │ k-point: [-0.36, -0.00, 0.36]
18 │ k-point location: between L-Z
19 │ k-point indices: 64
20 │ Band indices: 153

I would like to know what is the meaning of this direct band gap here? as I am expecting my material to be an Indirect band gap semiconductor according to the difference between VBM and CBM.

Could you please explain that @ajjackson
Thank You.

[ajjackson]

The "direct bandgap" reported here is smallest gap in the bandstructure without any difference in k (i.e. momentum transfer). In this case it is larger than the indirect bandgap, which is the difference between VBM and CBM, so it is technically correct to describe this system as an "indirect bandgap material".

However, I would point out that for this system the difference between these transitions is just 0.065 eV, so this material may show a lot of "direct bandgap" character in practice. You'd need to do some optics calculations to get more information about that.

[jfrmhrz]

Thank you very much for your explanation, this material is monolayer ReSe2, which is been reported to have this conflicts of being direct/indirect band gap semiconducting material. Could you please suggest what possible optic calculations to follow in order to confirm the nature of the band structure? Thanks

…

Sent from my iPhone

On Dec 5, 2022, at 8:50 PM, Adam J. Jackson ***@***.***> wrote:  The "direct bandgap" reported here is smallest gap in the bandstructure without any difference in k (i.e. momentum transfer). In this case it is larger than the indirect bandgap, which is the difference between VBM and CBM, so it is technically correct to describe this system as an "indirect bandgap material". However, I would point out that for this system the difference between these transitions is just 0.065 eV, so this material may show a lot of "direct bandgap" character in practice. You'd need to do some optics calculations to get more information about that. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.

[ajjackson]

The first step is simple enough: you can use an LOPTICS=.TRUE. calculation with sumo-optplot to have a look at the predicted absorption spectrum. Make sure to use plenty of k-points; you can use the same self-consistency tricks as a band structure calculation. But this plot will probably not be enough to distinguish which transitions are leading to more absorption, when the difference is so small.

To get a better look at the optical transition matrix it is possible to uncomment a line of VASP source and recompile in order to have it write out a plain-text "WAVEDERF" file when calculating optics. This can be inspected to determine which transitions are relevant; sometimes the calculated bandgap is really "dipole-forbidden" and the optical gap is driven by other states. (E.g. see Fig 2 of https://pubs.acs.org/doi/10.1021/acsenergylett.2c01961)

I'm afraid I don't have any handy tools for working with WAVEDERF files and haven't looked at one for a while. Maybe if you ask around you can find someone who will help with that analysis!

I would be very wary of e.g. Tauc plot analysis for this system making a strong assumption that the gap is purely direct or indirect. Until you've looked at the transition matrix it seems safest to assume both are present.

[JaafarMehrez]

Thank you very much, that was very informative. Happy Computing :) Jaafar Mehrez

…

On Tue, Dec 6, 2022 at 12:52 AM Adam J. Jackson ***@***.***> wrote: The first step is simple enough: you can use an LOPTICS=.TRUE. calculation with sumo-optplot <https://smtg-ucl.github.io/sumo/sumo-optplot.html> to have a look at the predicted absorption spectrum. Make sure to use plenty of k-points; you can use the same self-consistency tricks as a band structure calculation. But this plot will probably not be enough to distinguish which transitions are leading to more absorption, when the difference is so small. To get a better look at the optical transition matrix it is possible to uncomment a line of VASP source and recompile in order to have it write out a plain-text "WAVEDERF" file when calculating optics. This can be inspected to determine which transitions are relevant; sometimes the calculated bandgap is really "dipole-forbidden" and the optical gap is driven by other states. (E.g. see Fig 2 of https://pubs.acs.org/doi/10.1021/acsenergylett.2c01961) I'm afraid I don't have any handy tools for working with WAVEDERF files and haven't looked at one for a while. Maybe if you ask around you can find someone who will help with that analysis! I would be very wary of e.g. Tauc plot analysis for this system making a strong assumption that the gap is purely direct or indirect. Until you've looked at the transition matrix it seems safest to assume both are present. — Reply to this email directly, view it on GitHub <#191 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH3XWURY7MVSEV3QXD7CTA3WLYMVJANCNFSM6AAAAAASQR7JUM> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[ajjackson]

Indeed, good luck with the research!