Absorption with light polarized in an arbitrary direction

[luonn]

Dear Yambo developers and users:
I want to know if yambo can calculate the light absorbance of a 2D material when the light propagates in an arbitrary direction, i.e. neither parallel nor perpendicular to the surface of the 2D material?

We know that by adjusting the BLongDir parameter, the dielectric function, epsilon_xx (1 0 0), epsilon_yy (0 1 0), epsilon_zz ( 0 0 1), can be obtained.
% BLongDir
1.000000 | 0.000000 | 0.000000 | # [BSS] [cc] Electric Field
%
if BLongDir is set as follows:
% BLongDir
1.000000 | 0.000000 | 1.000000 | # [BSS] [cc] Electric Field
%
then the obtained result is epsilon_xx+epsilon_yy, or other epsilon that discribe the absorption with light propagates in the xz plane with an angle of pi/4 ?


Thanks a lot

Nannan Luo

[Daniele Varsano]

Dear Nannan Luo,

or other epsilon that discribe the absorption with light propagates in the xz plane with an angle of pi/4 ?

Yes, the BLongDir is a versor of module 1, then it is multiplicated for the small q0.

Code: Select all

% BLongDir
1.000000 | 0.000000 | 1.000000 | # [BSS] [cc] Electric Field
%

is a vector 1/sqrt(2)[1,0,1]

Best,
Daniele

[Jing Liu]

Dear Daniele,
I want to do a BSE calculation with different directions of polarized light (along x, y, z) by reading the gw data.
I have found the related parameters as LongDrXp in the gw calculation and BLongDir, LongDrXs, LongDrXp in BSE calculation. My questions are as follows:
1.
% LongDrXp
1.000000 | 0.000000 | 0.000000 | # [Xp] [cc] Electric Field
Does it means the incidence direction is along x or the vibration direction is x ?
2.
If I want to change the direction to y, should I change the LongDrXp in gw calculation as 0.000000 | 1.000000 | 0.000000 |? or just change the BLongDir in BSE calculation?
3.
Should I change all the three parameters or just the BLongDir ?
Thank you very much !

Best,
Jing Liu

[Daniele Varsano]

Dear Jing,

LongDrXs / LongDrXp are the field direction for the calculation of the screening. Unless the system is highly anisotropic, it should not change much the entity of the screening, hence the GW gap or the W kernel in BSE:

If you want to look at different polarisation in absorption you need to change BLongDir

Best,
Daniele

[Jing Liu]

Thanks so much!

[pyadav]

Dear Daniele.

Suppose I want to investigate the anisotropy in the BSE absorption spectra, would change just the

Code: Select all

% BLongDir
 1.000000 | 0.000000 | 0.000000 |        # [BSS] [cc] Electric Field

will do the job or I need to change

Code: Select all

% LongDrXp
 0.100000E-4 |  0.00000    |  0.00000    # [Xp] [cc] Electric Field

?

Thank you,
Pushpendra

[Daniele Varsano]

Dear Pushendra,

you do not need to specify the modulus, just the direction, then the code will take care of the small q limit.

BLongDir indicates the polarization for the absorption.

LongDrXp is meant for the screening, if the system is highly anysotropic the eps_oo(q) may be affected, but usually, the effect is small. You can set that variable to
(1,1,0) to have a sort of average for the calculation of the screening.

Best,
Daniele

[pyadav]

Dear Daniele,

Thank you so much for the clarifications.

I have another issue related to the same problem.

1. I'm calculating absorption spectra at the G0W0+BSE level for the two different polarization directions i.e.
Along X

Code: Select all

BLongDir
 1.000000 | 0.000000 | 0.000000 |        # [BSS] [cc] Electric Field

Along Z

Code: Select all

BLongDir
 0.000000 | 0.000000 | 1.000000 |        # [BSS] [cc] Electric Field

In both the cases the exciton energies are same but I'm observing a huge difference in the absorption spectra and also in their corresponding oscillator strengths.

When the polarization direction is along X then the exciton peaks below the QP gap are too small to identify while in the case of polarization along Z the exciton peaks have prominent peak heights but in both the cases the exciton energies are same. I'm attaching the input and the corresponding output files.

2. Why the absorption/oscillator strength data are very small (in order of 10^(-8))

Kindly if can have a look and identify the issue that would be very helpful.

Thanks in advance!
--
Pushpendra

[Daniele Varsano]

Dear Pushendra,

2) this is explained in the FAQ:
https://www.yambo-code.eu/faq/

see:
When calculating absorption spectra using the truncated coulomb cutoff, I notice very small number in the imaginary macroscopic dielectric function, and the static real part is one, is that a problem?,

You have a 2D material and you are using a Coulomb cutoff, in this case, the epsilon is not a well-defined quantity and you should look at the polarizability alpha.


1) When changing the polarization of light you are looking at dipoles in different directions: q.<c|r|v> and they can have very different values if the system is anysotropic. Here <c|x|v> is very different from <c|z|v>.
Moreover, in your case you are looking in the direction orthogonal to the system so this is expected as z is the direction where you have confinement of the wfs.

Best,
Daniele

[pyadav]

Dear Daniele,

Thank you so much for the looking into the data and writing your response.

I see that there are some problem in 2D epsilon and an alternate choice to represent the absorption physics is using the Im(alpha).

The second thing is that I should not compare the absorption data for (1 0 0) and (0 0 1) because its 2D system and (0 0 1) will definitely give a different result.

I still have a question that why the exciton peak for

Code: Select all

BLongDir
 1.000000 | 0.000000 | 0.000000 |        # [BSS] [cc] Electric Field

case is too small even when I'm plotting the Im(alpha) for the absorption curve?

The 'alpha' data is attached below.

Regards,
Pushpendra