Binding energy

[lamia]

Dear Yambo devoloppers,

I've calculated the absorption coefficient using GW+BSE. However, it shows that the first absorption peak occurs at around 3 eV, which is larger than the minimum band gap of 2.63 eV!!! Therefore, the exciton binding energy is negative ??? Is this physically reasonable (Since the GW+BSE calculation take the electron-hole interaction into consideration)? Is there any explanation? Especially, the exciton binding energy for 2D materials is much larger than in their 3D counterpart, such as, the exciton binding energy for monolayer MoS2 ranges from 0.6 to 1.0 eV.

Thanks,

Best,

[Daniele Varsano]

Dear Taouli,

Therefore, the exciton binding energy is negative ??? Is this physically reasonable (Since the GW+BSE calculation take the electron-hole interaction into consideration)? Is there any explanation?

If there are not bound excitons it is possible. Anyway check carefully your convergence parameters.
Best,
Daniele

[lamia]

Dear Daniele,

Thank you for your answer.
" If there are not bound excitons it is possible." as you said. could you give me a reference, please?
How could I check if the excitons are bounded or not ?
What's the influence of unbounded excitons on optical properties and on material properties in general ?

Best,

Lamia

[Daniele Varsano]

Dear Lamia,
the exciton is unbound if its energy is larger than the gap.
Please check if you have bound dark excitons looking at the excitation energy using ypp.
If you want, you can post report/output file and I'll have a look.

Best,
Daniele

[lamia]

Dear Daniele,

You'll find in the attachments the files.
Thank you

Best,

Lamia

[Daniele Varsano]

Dear Lamia,

from the output files, you sent it is very likely that something went wrong.
The problem I can see is not about the negative binding energy, but because you have an extremely large negative excitation ( -49.46 !!!).
Here are two suggestions:
1) You are using a very ancient version of the code (3.4.2)of about 4 years ago which is not supported anymore, I strongly suggest you update to a newer release (5.0). You can download it at this link http://www.yambo-code.org/download/
2) The cutoff size you are using is totally off: the box side should be slightly shorter than your supercell size, about 19.5 bohr. The parameter you are considering can provide a totally wrong interaction potential.
3) It seems that you are using smearing in your nscf calculation; this is not needed for a semiconductor. It should be harmless but I'm not totally 100% sure that it can cause some inconsistency when a scissor operator is applied, probably not, but I would remove it for safety.

Best,
Daniele

[lamia]

Dear Daniele,

You have said before that the exciton could be bounded or unbounded? What's the influence on the absorption of material in the case of unbounded? is it necessary to be bounded to get a good absorption material?

Could you provide me with some references about absorption in the case of unbouded exciton, please?

Best,

[Daniele Varsano]

Dear Taouil,

unbound excitons are excitations at energies larger than the quasiparticle gap, this happens in all materials for large enough energies (resonance in the continuum).
First excitations instead are usually bound as there is always some residual electron-hole interaction unless you have a complete screening e.g. in metals.
But also in the case of metals, the definition of bound/unbound is not so strict. Excitations in the continuum can be also classified as bound as their energy is smaller than its independent particle counterpart (see e.g. Spataru et al Nano Lett. 7, 1626 (2007).

As I told in my previous post, there was something odd and unpysical in your results, I hope you solved it.

Best,
Daniele