Work flow for DOS and band structure with hybrid functionals

[reynaldo.putra]

Hello everyone,

I would like to confirm whether the following work flow is sufficient:
1. Geometric optimization using PBE
2. SCF calculation from the PBE-optimized geometry using HSE06
3. DOS and band structure calculation from the charges in the SCF calculation using HSE06

From my understanding, geometric optimizations using either PBE and HSE06 will produce sufficiently similar results; only the electronic properties (DOS and band structure) are underestimated with pure PBE. With this in mind, is the aforementioned workflow sufficient?

[michael_wolloch]

Dear Reynaldo Putra,

forces, and thus geometry, are dependent on the electronic structure, which is different for each functional. If PBE and HSE06 result in a "sufficiently similar" equilibrium geometry depends on the system and your definition of "sufficiently similar".

J. Chem. Phys. 153, 164710 (2020) e.g. reports lattice parameters for rare-earth oxides for PBE, PBE+U, and HSE06. For La2O3, there is a volume difference of 2%. This might be acceptable for your use case, or not. As always, it depends on what you want to ultimately calculate.

Be advised that for hybrid functionals, there are a few pitfalls for band structure calculations. Read the section on Band-structure calculation using hybrid functionals, and check out the KPOINTS_OPT file if you are using 6.3.0 or newer.

Cheers, Michael

[michael_wolloch]

Dear Reynaldo,

another thought: You can use your second step (HSE static calculation) to check if the forces are reasonably low. In other words, if the geometric optimization with PBE resulted in a relaxed structure for the HSE functional.