Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Deals with issues related to computation of optical spectra in reciprocal space: RPA, TDDFT, local field effects.

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano, Conor Hogan

Post Reply
WEIW
Posts: 34
Joined: Fri Nov 24, 2017 9:48 am

Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by WEIW » Mon Mar 20, 2023 10:27 am

Dear developers,
When I performed BSE calculation in 2D system, I set BEnRange from -2 to 10 eV to get its absorption spectrum. What confuses me is that the value of absorption is surprisingly nagetive at -2 to 0 eV and the dispersion is approximately centrosymmtric. I have tested this system many times, the same results is obtained that makes me very puzzled.
How can I solve this problem? The related statistics are attached.
Looking for your kindly reply!

Best,
Liu
You do not have the required permissions to view the files attached to this post.
Dr. LIU Hongling
School of Physics, Shandong University
Jinan, CHINA

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by Daniele Varsano » Mon Mar 20, 2023 10:56 am

Dear Liu,

this happens because you set BSEmod= "retarded".
In a retarded response function, the Im part of the response function is "odd" wrt the origin of the frequencies, while it is "even" in a resonant approximation. You can recognize it looking at the independent particle expression in the two approximations.


Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

WEIW
Posts: 34
Joined: Fri Nov 24, 2017 9:48 am

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by WEIW » Mon Mar 20, 2023 1:12 pm

Dear Daniele,
Thanks for you quickly reply! Could you please so kind give me some suggestions to solve this problem. That is, how can I get a more accurate absorption spectrum in the range from -2 to 10 eV? Reset BSEmod as 'resonant' or 'coupling'? In this system, low energy excitation seems significant.

Best,
Liu
Dr. LIU Hongling
School of Physics, Shandong University
Jinan, CHINA

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by Daniele Varsano » Tue Mar 21, 2023 11:07 am

Dear Liu,

this is not a problem!
Negative excitations do not indicate absorption, as they come from c-->v transitions and not v-->c transitions. You can interpret as emission.
Negative excitation energies, in Tamm-Dancoff approximations, ie when the coupling between resonant and antiresonant part of the excitonic Hamiltonian is neglected are by definition the opposite of the ones found at positive energies (resonant part): E_antiresonant=-E_resonant

Passing from resonant to retarted does not change the excitation energies, it just changes the way you build the response function (ie positive or negative intensity for E < 0).

If you are interested in low energy excitation, you can analyze them using the ypp utility (ypp -e s b 1): in resonant, or retarded mode only positive energy are reported (see above).
You can be then interested in seeing if the coupling has an effect, in this case also negative excitations are reported and you can see how much they deviate from the condition E_anti=-E_res.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

WEIW
Posts: 34
Joined: Fri Nov 24, 2017 9:48 am

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by WEIW » Wed Mar 22, 2023 7:49 am

Dear Daniele,
Thanks for your explanation! :D
One more thing I want to understand is that if I set BSEmod as 'coupling', the resultant absorption is no more 'odd' or 'even', so can this result will indicate a exciton insultor?
the absorption picture is attached below.
Best,
Liu
You do not have the required permissions to view the files attached to this post.
Dr. LIU Hongling
School of Physics, Shandong University
Jinan, CHINA

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by Daniele Varsano » Wed Mar 22, 2023 2:27 pm

Dear Liu,
in order to see if you have an exciton instability, you need to look if you are in presence of negative eigenergies. The spectrum is not the best way to spot them, as if there are excitation with small or zero (dark) oscillator strength you will not see them.
The way to proceed is to look directly at the excitations using the ypp utility (ypp -e s -b 1 -J ...).
This should be done not including the coupling, otherwise you will have all the antiresonant energies which are negative by default.


Best,

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

WEIW
Posts: 34
Joined: Fri Nov 24, 2017 9:48 am

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by WEIW » Thu Mar 23, 2023 1:03 pm

Dear Daniele,
I very appreciate your suggestions! And I have calculated using yambo as you recommand.
However, there is a huge discrepancy after analyzing the statistics I got.

When BSEmod was set as 'resonant' or 'retarded', I obtained o-allbz.exc_qpt1_E(I)_sorted in which there is no negative eigenergies.
When BSEmod was set as 'coupling', o-allbz.exc_qpt1_E(I)_sorted shows indeed negative eigenergies that can be understood as c to v transition (emission) as you said.

Interestingly, in coupling BSEmod, when I check the contribution (in o-allbz.exc_qpt1_weights_at_2406 attached) for negative eigenergies, there are both v to c transition and c to v transition, which make me confused.

To verify no mistakes in my calculation, I also calculation MoS2 monolayer, in 'coupling' BSEmod, as comparison. In negative eigenergies for MoS2, there is noly c to v transition contribution.

So how to explain this strange phenomenon? :cry:

The related datas are attached.
You do not have the required permissions to view the files attached to this post.
Dr. LIU Hongling
School of Physics, Shandong University
Jinan, CHINA

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Appearing negative absorption peak in o-bse.alpha_q1_diago_bse

Post by Daniele Varsano » Tue Mar 28, 2023 10:14 am

Dear Liu,

I would say that's nothing strange is happening here.
When going beyond the Tamm-Dancoff (TD) approximation (ie considering coupling) the |vc> and |cv> spaces are coupled so an eigenstate can have components from both of them.
In case of MoS2, the coupling is small, so the (TD) approximation is a good approximation, meaning that the coupling is negligible and resonant part and antiresonant parts provide you eigenstates with pure |vc> and |cv> components (associated to energies opposite in sign).

In addition, please note that you are analyzing degenerate eigenstates (as you can see the weight do not sum-up to 1). In presence of degeneracy ypp sum up
states near in energies up to a threshold in order to respect the symmetry. If you want to analyze a single state you need to set "Degen_Step=0. eV" in your ypp input.

Best,

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Post Reply