Band energies and partial/fractional occupation in OUTCAR?

[reynaldo.putra]

Hello,

I do not know what to make out of fractional occupation in the OUTCAR file.

fractional-occupation.png

I have been working with semiconductors, so I understand that when the occupancy goes 1.0 to 0.0 from one band energy to the next, these can be considered the energies of the VBM and CBM. But I do not understand what partial/fractional occupation means.
I plotted the total DOS using the VASPKIT module, and found different DOS intensities for the spin up and down states.

BTON-surface-dos.png

So far I have been working with semiconductor materials that have equal DOS on the spin up and down states, hence I am still unfamiliar with this concept and would like to gain more understanding and further references when possible.

What does partial occupation of the band energy mean? What are the typical materials with unequal spin up and down DOS? What are the implications of having different band gaps in the spin up and spin down states?

Thank you for your time, and have a good day.

[christopher_sheldon1]

Dear Reynaldo,

Thank you for your question and sorry for the slow response.

What does partial occupation of the band energy mean?

At 0K, the electron occupancy goes rapidly from 1 to 0 between occupied and unoccupied energy levels. For insulators, this is not an issue as there is a large band gap. However, for metals which have a zero band gap, there is an immediate change from occupied to unoccupied at the Fermi level, i.e., a step function. This would be difficult to calculate, creating a lot of numerical noise and resulting in poor SCF convergence. Instead, the smearing is used to make the shift from occupied to unoccupied more gradual (the occupancy goes from 1.0 --> 0.0 over a few 100 meV, rather than instant). This improves convergence. You can read more about smearing on our wiki (Smearing).

Semiconductors have a small bandgap so, while they are less likely to experience the same problems of metals, the smearing that is used may be large enough to result in the electron occupancy being smeared across a few bands that lie near one another in energy. The smearing is effectively a trick to improve convergence by including an electronic temperature (depending on the smearing method that is used).

What are the typical materials with unequal spin up and down DOS?

Materials that have unequal spin up and spin down tend to be magnetic systems, e.g., ferromagnets, antiferromagnets (see our wiki and tutorials).

What are the implications of having different band gaps in the spin up and spin down states?

This I am unfamiliar with. So long as the bands gaps are both large, they would both be insulators; if small, both semiconductors/ metals. The properties of the system could change but I am unsure how.

Best wishes,

Chris