Spin Decomposed Charge Density (PARCHG)

[jeffrey_zom]

Hi,

I am doing calculations on electron polarons.
I would like to visualise the electron density belonging to the electron polaron.
I've attempted to do so by "Band-decomposed Charge Density" (BCD) calculations, with either EINT or IBAND.

For EINT I select an energy interval that correspond to the energy range of the polaronic state, obtained from inspecting the plotted Density of States (EINT around 1.5 eV, see DOS.png). This works when the energy range of the band belonging to the electron polaron is nicely isolated from the other bands, which is not the case for all my calculations (not shown here), and can therefore not be used.

For IBAND I select the single band that belongs to the highest occupied state. As in a neutral cell there are 640 electrons, the band of the 641th electron (NELECT = 640+1) is 321. However, as you can also see in the figure, the 321th band is for 1 spin state the polaron state, whereas for the other spin state, it is the conduction band edge, such that obtained electron density is incorrect with IBAND = 321.

My question: Is there a way I can obtain the spin decomposed charge density, without the need of using EINT as there is overlap in the energy range for some calculations?

Kind regards,

Jeffrey

Attachments:
- All input files for the BCD calculation except WAVECAR (using IBAND).
- DOS.png

[jeffrey_zom]

Hi,

Update on my own question, I have solved my problem:
As stated in the wiki, the PARCHG contains two parts for spin-polarised calculations.
The total electron density (spin up density + spin down density), and the magnetisation density (spin up density - spin down density).
I've created a small (python) script which can extract these 2 parts, and recalculate the separate spin up and spin down density.
This allows for the use of IBAND, while also selecting a certain spin state.

For those of you interested in the script, I've provided it in the attachment.
It might not be the most optimised code, but for a file of 140 MB it completes in ~10 seconds.
Just make sure to adjust the string to the location of the PARCHG file.

It will output 4 files (in the same directory): PARCHG_electron_density, PARCHG_magnetisation_density, PARCHG_spin_up, PARCHG_spin_down. These can be visualised with for instance VESTA.

It might be interesting for a future VASP version to implement additional input parameters which allows for obtaining these spin-decomposed electron densities without the need of custom postprocessing scripts.

Kind regards,

Jeffrey

[jonathan_lahnsteiner2]

Dear jeffrey_zom,

Great that you were able to answer your question. You are completely right that PARCHG contains two data sets for spin-polarized calculations. This information is also available in the vasp-wiki.

Thank you for providing your script which might be useful for other users.

All the best Jonathan