how to compare with experiment results?

[yanghuang]

Dear Developers and Users,
I want to calculate exciton binding energy, firstly, the calculated optical absorption spectrum (given by the imaginary part of the dielectric function) should be matched with the experiment results as much as possible. But for my system, there is no the imaginary part of the dielectric function in experiment, only the absorbance spectrum. So, how to compare with experiment results?
My system is strongly anisotropic.

Thanks in advance!

[Daniele Varsano]

Dear Yang,

You can calculate exciton binding energies by comparing the interacting absorption spectrum (Bethe Salpeter, TD Hartee Fock etc..) with the independent particle spectrum, as it is exactly the quasi particle gap minus the excitation energy.
The fact that a calculation match or not with the experiments depends on the level of the theory and from how sophisticate theory are need for the material.

If you have the absorbance you can compare it with the calculated absorption, as the two are related.
If your system is anisotropic, you will need to calculate the absorption along the different axis.

Best,
Daniele

[yanghuang]

Dear Daniele,
Thank you so much for your detailed reply.
The following is the calculated absorption spectrum, it is not well matched with the experiment results (see firs post). Specifically, light absorption starts at 1.0 eV in my results (see red circle), but it starts at 2.0 eV in experiments. The calculated GW band gap is 2.0 eV. What parameters can be used to make light absorption starts at 2.0 eV?
Another question, there are more peaks in my results (peak-2,peak-3), these peaks can be eliminated?


Best,
Yang

[Daniele Varsano]

Dear Yang Huang,

comparison with experiments it is not always straightforward, as it depends on the experimental conditions and the approximation you did (remember that there are several approximations in the GW/BSE approach as plasmon=pole approximations, static screening in BSE etc..).
The best you can do is to be sure you carefully converged your results and analyze you calculations at the level of theory you considered.
Coming to your specific spectra, please consider that in the experiment the width beneath the peaks is given by the lifetime of your excitations, possibly vibrionic effects, resolution of the experiment etc...
In your calculation you set a damping by hand:

Code: Select all

%BDmRange				
0.100000|0.100000|	eV	#	(BSS)	Damping range
%

Just looking at your figure I sould say that you excitation is higher than 1eV, try to reduce your damping range.

Another question, there are more peaks in my results (peak-2,peak-3), these peaks can be eliminated?

Well, why do you want to eliminate them? It is clear from the exp that there are several excitations above 3eV even if they are not clearly resolved.

My suggestion is to carefully check the convergences of the calculations, next you can play a bit with the artificial damping in order to mimic
the peak width of the exps.

Best,

Daniele

[yanghuang]

You can calculate exciton binding energies by comparing the interacting absorption spectrum (Bethe Salpeter, TD Hartee Fock etc..) with the independent particle spectrum, as it is exactly the quasi particle gap minus the excitation energy.

Dear Daniele,

Just want to confirm that the GW+BSE spectrum and the GW+RPA spectrum both are from the same file o.eps_q1_haydock_bse, the GW+BSE spectrum (col 2), the GW+RPA spectrum (Col 4), right?

For instance, the binding energy = 4.1 (GW+RPA) - 3.5 (GW+BSE) = 0.6 eV, right?

Best,
Yang Huang

[Daniele Varsano]

Dear Yang Huang,
yes this is right.

Anyway instead of looking at absorption plots, you can be a bit more precise by looking at the GW direct gap if you calculated it at the right K point, and the excitation energies. In order to look at the excitation energies, if you diagonalize it instead of using Haydock by using the ypp post processing (see ypp -e s option). Depending on the size of the BSE matrix the diagonalisation can be more or less feasible.
Please note that you can also have "dark excitons", i.e. excitation of the system having zero, or very low oscillator strengths and these are hidden when looking to the absorption.

Best,
Daniele

[yanghuang]

Dear Daniele,

Thank you so much for your advice.

Best,

Yang Huang

[yanghuang]

Dear Daniele,

Sorry to bother you again.
Attachment is the results of GW+BSE.
Comparing Figure (a) and (b), light absorption starts at a position larger than 2.0 eV when BDmRange 0.03 | 0.03.
Comparing Figure (b) and (c), light absorption starts at a position less than 2.0 eV when BDmRange 0.03 | 0.03.
Therefore, the parameter BDmRange does not change the position of light absorption, the excition energy always 2.6 eV, right?

Best,

Yang Huang

[Daniele Varsano]

Dear Yang Huang,

Therefore, the parameter BDmRange does not change the position of light absorption, the excition energy always 2.6 eV, right?

Of course BDmRange does not change the excitation peak, the calculation provides you the excitation energy, next the spectrum is built by considering Lorentzian centered in the excitation peaks having a width given by BDmRange.

Best,

Daniele

[haseebphysics1]

Dear Daniele,

I have pasted the plot of BSE (most converged result), RPA and comparison with the experimental curve. From this plot the difference of sharp peaks of RPA and the corresponding peak of BSE gives more than 1 eV difference for this semiconductor (exciton B.E.)! (KS-RPA - KS-BSE)
But the more important point of worry for me is that the sorted energy file from ypp gives (only the below the bandgap are shown below)


# E [ev] Strength Index
#
2.00291776657 0.04534414038 1.00000000000
2.027220249 0.4988466972E-3 2.000000000
2.18450331688 0.04390767217 3.00000000000
2.192421675 0.1276613475E-3 4.000000000
2.201964140 0.1689653378E-2 5.000000000
2.210659027100 0.104321807623 6.000000000000
2.228118658 0.2731759334E-2 7.000000000
2.2380900383 0.0089330776 8.0000000000
2.250730038 0.1531953302E-6 9.000000000
2.279653072 0.5574001465E-2 10.00000000
2.298375607 0.2712711692E-2 11.00000000
2.306696653 0.5385429133E-2 12.00000000
2.309972048 0.3514389973E-2 13.00000000
2.311466455 0.8024664596E-2 14.00000000
2.311852455 0.2226549434E-2 15.00000000
2.313919783 0.5148362252E-3 16.00000000
2.319422007 0.7972204003E-5 17.00000000



keeping in mind that 2.33 is the KS bandgap,

1: Does the above data not show that "strongest bound exciton" has B.E equal to 2.32 - 2.0029 ~ 0.32 eV?

2: Can you please shed some light on the previous argument when I was getting exciton BE from the absorption experiment and why that was coming out to be so much negative (as compared to 0.32 eV)?

3: Can exciton exist after the bandgap energies?

Thanking you,
Take care and stay safe!