Question about XAS calculation code in VASP6.1

[zhaoru_sun1]

Dear everyone,

The XAS calculaion method of VASP6.1 described in VASPwiki is

And the relevant code in src/core_con_mat.F is

Code: Select all

                                    EPSILON_HELP = PREFACTOR/FREQUENCY *                  &
                                                   WDES%WTKPT(KPOINT) * DELTAFUN *          &
                                                   MECC1 * CONJG(MECC2) * CH_AMPLIFICATION

"FREQUENCY" in the code should correspond to "ω" in the Wiki, but I am confused why not "FREQUENCY**2" like "ω^2" in wiki? What's wrong?

[ferenc_karsai]

Dear Zhaoru,

thank you very much for pointing out that problem. The equation is correct and unfortunately that was a bug in the code. It will be corrected in the next release.

Theoretically this could be more than a constant shift, but we tested the results for all of our calculations and fortunately only the overall intensities change visibly. Since one reports only arbitrary intensities and scales them all calculations until now should be correct.

[zhaoru_sun1]

thank you very much for pointing out that problem. The equation is correct and unfortunately that was a bug in the code. It will be corrected in the next release.

Thank you. I am reviewing the XAS theory. There is a problem with the relationship between the dielectric function and XAS that puzzles me.
In F. Karsai et al. Phys. Rev. B 98, 235205 (2018), the imaginary part of the dielectric function is:





While from EQ(3.10) in C. Brouder, J. Phys.: Condens. Matter 2, 701 (1990), the absorption cross section is:



So I can roughly get:



But this is not consistent with F. Karsai et al. Phys. Rev. B 98, 235205 (2018):

the transversal expression for the imaginary part of the dielectric function in the long-wavelength limit (q=0) is directly proportional to the absorption spectrum

This makes me confused.

[ferenc_karsai]

Your assessment look correct and it would have been better to write without the word "directly":

"the transversal expression for the imaginary part of the dielectric function in the long-wavelength limit (q=0) is proportional to the absorption spectrum"

[zhaoru_sun1]

Your assessment look correct and it would have been better to write without the word "directly":

"the transversal expression for the imaginary part of the dielectric function in the long-wavelength limit (q=0) is proportional to the absorption spectrum"

Thank you. If so, for clarify I think the description in XANES_in_Diamond:

The XAS spectrum is the imaginary part of the frequency-dependent dielectric function

may need some modification and also the corresponding script plot_core_imdiel.sh.
Although the additional frequency factor has little effect on XAS within ~10 eV of the X-ray energy region.

[ferenc_karsai]

Thank you very much. Indeed the description of the tutorial is very sloppy. I will correct the tutorial and it's description.

[zhaoru_sun1]

Thank you.
Is there any way for VASP to output the contributions of different unoccupied orbits in XAS calculation, that is, the discrete transition amplitudes before broadening. Just like what you did in Figure 2 and Figure 3 in your paper F. Karsai et al. Phys. Rev. B 98, 235205 (2018).

[ferenc_karsai]

Figure 2 is just an electron density plot, with the density calculated only in a given energy region (or for bands). Of course using the core-hole in the calculation. Please see the following for band decomposed charge densities:
wiki/index.php/Band_decomposed_charge_densities

Figure 3 is a BSE calculation and just shows the different directions of the diagonal elements of the dielectric function. Note that this calculation was done using WIEN2k.

There is no feature in VASP selecting only specific contributions from selected orbital quantum numbers. VASP always calculates all contributions in a given range.