Kramers-Kroning relation

[Dipankar Saha]

Hi,
Can we calculate (@ ATK) the real_part elements of dielectric tensor, utilizing  the "Kramers-Kroning" relation?

Best_
Dipankar

[Jess Wellendorff]

The OpticalSpectrum analysis utilizes Kubo-Greenwood theory, and the real part of the dielectric tensor should be obtained using the "evaluateDielectricConstant" method http://docs.quantumwise.com/manuals/Types/OpticalSpectrum/OpticalSpectrum.html That link gives an example of getting the real part at zero frequency as a 3x3 matrix. Hope this helps, otherwise I may have misunderstood your question.

[Dipankar Saha]

http://docs.quantumwise.com/manuals/Types/OpticalSpectrum/OpticalSpectrum.html
Why this page is not available?

Did you change the documentation?

[Petr Khomyakov]

Try this one: https://docs.quantumwise.com/manual/Types/OpticalSpectrum/OpticalSpectrum.html.

[Dipankar Saha]

How do you calculate the dipole matrix? What does E_nm represent ?

[Petr Khomyakov]

- Regarding the Kubo-Greenwood formula and related dipole matrix definition, e.g., see Walter A. Harrison. Solid State Theory. McGraw-Hill, 1970, or just google it.

- Regarding the energy term in the denominator in the KG formula, E_nm = E_n - E_m, which is a transition energy between the 2 states labeled with the indexes n and m (quantum numbers).

[Dipankar Saha]

@ Optical Spectrum calculation_
It seems difficult to get a convergence with the increase in no. of bands below & above Fermi level.
Why So?

[Petr Khomyakov]

What exactly do you mean by "difficult"? Which system are you studying? How many bands have you used for the optical spectrum calculation? Typically, the number of bands should be increased significantly (by an order of magnitude), compared to the default values.

The actual number of bands required to achieve convergence is very much system-dependent, as one should take into account all the essential optical transitions for a particular material system, and optical properties of materials vary a lot.   

[Dipankar Saha]

Yes, I agree this is very much system dependent. For large structures where the no. of bands (below & above the Fermi level ) >150, it is difficult to find convergence. Set 'bands_below /above _fermi_level=' to values e.g. 10, 25, 50 and so on, you will find the output changing significantly.

(However, for smaller structure this is not a problem. Say out of 10 bands if you select anything > 6, the output will not vary. )

Anyway, thank you  Petr for all the information! 

Best Regards_
Dipankar Saha

[Ulrik G. Vej-Hansen]

It is often useful to look at it as an energy interval instead of a fixed number of bands. You need to cover an adequate energy interval around the Fermi level, so the number of bands to include is, to a first approximation, proportional to the number of bands in the system.

[Dipankar Saha]

Okay... / Thank you Ulrik!