IP absorption

[nrbsh]

Dear all,
I have a question about independent particle absorption.
I calculate the absorption spectra in RPA and IP approximations.

The IP absorption spectra is zero in the gap energy while rpa spectra has a peak at this energy.

how can we explain this behavior?

WBR
-Nbsh
Teran uni

[Daniele Varsano]

Dear Nbsh (is this your name?),
Hard to say anything without more information,
can you please post the report file and the output files of your calculations?
Best,
Daniele

[nrbsh]

Dear Daniele,
Thanks for your reply.

Actually the IP absorption is not exactly zero at the gap point but its value is very small in compare with rpa absorption
(I've attached the diagram here, the energy gap is ~ 2.6 eV)

absorption coefficient has two main term, the dipole matrix and delta(E_c - E_v - Omega) to guarantee the energy conservation.

I think that in independent-particle approximation it only takes the last term (delta function); am I right?

Best regards,
Nbsh
Teran uni

[Daniele Varsano]

Dear Nbsh,
can you please post your inputs, outputs, and reports of your calculations.
Otherwise, again, it is impossible to say anything as we do not know the system and set up you are dealing with.

Best,
Daniele

[nrbsh]

Dear Daniele,
I attached them here.

Otherwise, again, it is impossible to say anything as we do not know the system and set up you are dealing with.

I think there isn't anything special.

WBR
Nbsh

[Daniele Varsano]

Dear Nbsh,

the direct band gap is found at 2.29 eV as reported in the report file.
This is indeed more or less the onset of both IP and RPA. Local field effects are quite strong and redistribute the oscillator strength toward lower energies.
Not sure this is what is usually found...

Please note that also in the IP you have the dipole moments between occupied and unoccupied states.
The RPA (local filed effects) is obtained starting from the independent response function \X^0 by solving the dyson equation

Code: Select all

X=X^0+X^0VX

and calculating the macroscopic dielectric function

Code: Select all

\eps_M=1/\eps_00^-1

see section 2.B of Onida, Rening, Rubio, REVIEWS OF MODERN PHYSICS, VOLUME 74, APRIL 2002, p. 601.
So many component of the dielectric matrix contribute in the (0,0) component of the inverse matrix.

Note that you are using a coulomb cutoff potential, when using the box I suggest you to set ycut and zcut just slightly smaller than your supercell.

Daniele

[pyadav]

Dear Daniele,

For the IP absorption spectra do I need to change mods in the following two settings to IP ?

Chimod= "IP" # [X] IP/Hartree/ALDA/LRC/PF/BSfxc
BSKmod= "IP" # [BSE] IP/Hartree/HF/ALDA/SEX

As I understand I also need to keep the following tag uncommented, please confirm.

KfnQPdb= "E < ./SAVE/ndb.QP" # [EXTQP BSK BSS] Database

Thanks,
Pushpendra

[Daniele Varsano]

Dear Pushendra,
it depends on what you need and in which way you perform the calculation:
Chimod= "IP" ! this is a linear response in G space
BSKmod= "IP" ! this is to calculate in the transition space

you have also Chimod in the BSE input, but it is meant to calculate the screening needed for the BSE kernel, if BSKmod is set to IP, you do not need to build the kernel.


KfnQPdb it depends if you want an IP spectrum using the GW eigenvalues (IP-GW), if you do not include it you will have the Kohn-Sham independent particle spectrum IP-KS.

Best,
Daniele

[pyadav]

Dear Daniele,

Thank you for the clarifications.

Best,
Pushpendra

[Fadil]

Dear all,

I calculated the RPA macroscopic dielectric function from dipole matrix elements and compared the results with yambo.
I follow the equations 6-8, 14 in 'yambo: An ab initio tool for excited state calculations'.
Even though the results show similar trend they are not exactly the same with the yambo results.
You can find input, output files and results for imaginary and real part of the macroscopic dielectric function in the attachment.
I changed 0+ but still no succeed.
Any idea what I am missing?

Best