BSE and exciton analysis

[manaswitakar]

Hi,

Firstly, I apologise for maybe posting in the wrong forum, but I do not have access/permission to the BSE forum.

I am trying to perform BSE calculations and exciton analysis on double perovskites (Cs2PbSnI6). I followed the tutorials on BSE and exciton analysis, however I do not understand fully and have a few questions:

1. When I plot the the absorption spectrum fro the BSE kernel, I can change the broadening of the peak by changing the BdmRange keyword. What is the physical meaning of this keyword? How do I know which value is correct? Does changing the BDmRange affect the exciton amplitude/energies?

2. Once I figure out which excitons I wish to analyse from the "o-3D.....E_sorted" file, I create the ypp_AMPL.in file with certain range of excitonic states. After this, when I run app, it shows that certain states are merged with certain other states. What is this merging based on?

3. For a particular exciton analysis, I get two files: namely, "o-3D.....amplitude_at_#" and "o-3D....weights_at_#". In the amplitude file, there are amplitude of the particular exciton at various energies. How is this related to the corresponding exciton in the E_sorted file?

4. Can I apply broadening to the individual excitons peaks I wish to analyse? How do I plot all the bright excitons in the system?

5. Finally, in the "o-3D....weights_at_#" file, what does "Symm_Kv" and Symm_Kc" signify? If the Kc and Kv are of the same index, this means there is a direct (optical) transition?

I am sorry if these are lots of questions, please let me know as convenient.

Thanks and Best Regards,
Manaswita

[Daniele Varsano]

Dear Manaswita,

I do not have access/permission to the BSE forum.

That's strange, what kind of message do you receive?

1. What is the physical meaning of this keyword? How do I know which value is correct? Does changing the BDmRange affect the exciton amplitude/energies?

The physical meaning is the lifetime of the excitation, in principle this could be calculated at different level of sophistication, but in a standard calculation at 0K, you introduce as a parameter. This adds a broadening to your excitation (the solution is a delta peak) with a Lorentzian shape. You can change it adjusting for instance to compare your spectra with experiments. Excitation energies are not changed as well as the oscillator strengths, ie the area beneath the peak. Of course for large damping the peaks are broader and the maximum eight is reduced.

2. What is this merging based on?

Degenerate excitation are merged according to a threshold, you can avoid this by setting the threshold to zero (Degen_Step=0.0 eV). The default is 10meV.

How is this related to the corresponding exciton in the E_sorted file

No relations, or better in the input file you select the index of the exciton you want to analyze. In the weight you have the single particle contribution participating to that exciton (BSE eigenvector |A_cvk|^2). In the ampl file you have essentially peaks at the energies of the cv transitions participating in that exciton vs energy.

4. Can I apply broadening to the individual excitons peaks I wish to analyse? How do I plot all the bright excitons in the system?

Boradening is meant to be applied to the spectrum, you can consider a fixed value, or a linear function from the minum to the maximum energy you are plotting (the two number in the input indicates min et max broadening). Birght excitons, being bright they are the peaks in the absorption spectrum. You can also plot the spectrum and the I_sorted file as impulse which reporting excitations energy and intensities.

If the Kc and Kv are of the same index, this means there is a direct (optical) transition?

You have the label for the K point and the associated symmetry of the transition participating in the exciton. If you are looking at optics you are in the long wavelenght limit q~0, so that the exciton is formed by direct excitation (same k between valence and conduction). This is done by construction as you selected
BSEQptR=1, ie the q=0 limit. You can also look at finite momentum excitation as seen e.g. in electron energy loss spectroscopy, but as far as you are interested in optical excitation, you need to do the calculatio in the q=0 limit and only vk-->ck (same k) transitions are allowed.

Best,

Daniele

[manaswitakar]

Hi Daniele,

Thanks for your responses.

I do not get any message. I just cannot enter the writing mode on the forum.

Another query. If I want to analyse specific excitons from the E_sorted file (let's hypothetically say all bright excitons with 1-5 index). Then, I will get 5 amplitude files and 5 weight files. In order to get the whole spectrum of the bright excitons, I just add the amplitude files and divide by 5 ?

Thanks again and Best Regards,
Manaswita

[Daniele Varsano]

Dear Manasawita,
I'm not sure if I understand your question.
The spectrum is found in the output eps/alpha/eels files, amplitude files contains the spectral composition of each exciton which is *not* weighted with their dipole strengths.

Best,
Daniele

[manaswitakar]

Hi Danele,

Thanks for your response and sorry for delay in coming back. I think I could understand now. However, I realise, in the optical absorption spectrum, the IPA energies are the DFT band gaps. even when I set the :

KfnQPdb= "E < SAVE/ndb.QP"
% KfnQP_E
0.000000 | 1.000000 | 1.000000 | # [EXTQP BSK BSS] E parameters (c/v) eV|adim|adim
%

I attach the input file and the report file for your reference.

Please let me know where am I going wrong.

Regards,
Manaswita

[Daniele Varsano]

Dear Manaswita,
for some reason, the QP energies are not read. At the end of the report you can see the mirroring of the input actually taken into account for hte calculation and you can see that the QP correction string is not present. Moreover, you have BSENGBlk= 1 RL.
The latter is because it seems that the static screening databases are not present in your SAVE director and the em1s keyword is missing from your input as well as the parameters needed to calculate it. To add it you need to generate your input adding the keyword "-X s".
A complete bse input generation reads:
yambo -o b -k sex -X s -y d -V all
Also, I can see that input and report belongs to a different version of the code.

About the QP file not read, it sounds strange. You can try to use:
KfnQPdb= "E<./SAVE/ndb.QP"
and see if it takes effect.

Best,
Daniele

[manaswitakar]

Hi Daniele,

Thanks for your response. I already perform static screening and BSE kernel calculations before the diagonalisation. Do I still need to add the em1s flag in the 3rd step? I attach the input and report files of the previous 2 steps for your reference.

Regarding the QP energies, it still gives no change.

Regards,
Manaswita

[manaswitakar]

Sorry for not attaching all files previously.

Regards,
Manaswita

[manaswitakar]

Hi Daniele,

I could figure out the QP energies. I was repeating the 'KfnQPdb' keyword; first time "none" and after a few lines "SAVE/ndb.QP". So the code read the first entry and was printing the DFT energies. It works fine now.

Thanks and Regards,
Manaswita