What alternatives to optical properties with HSE06

[pascal_boulet1]

Hello,

We would like to calculate the optical properties of a slab mimiking a surface using HSE06 but we are facing computer limits problems, in particular the time limit of the queuing system.

The point of using the HSE06 functional is tho have a correct, opened gap (it is closed with PBE).

Are there alternatives to using HSE06?

We thought that perhaps the use of the scissor operator could help us (in that case we would use PBE) but we have two questions: 1) Is it implemented with optical properties? 2) It seems that VASP reoptimizes the electron density at the beginning of the optical properties calculations, even though one restarts from an old density, so is the scissor operator useful at all in that case?

Thank you for your help
Best regards
Pascal
University of Aix-Marseille, France

[merzuk.kaltak]

Dear Pascal,

The scissor operator can be used to determine optical properties.
Avoiding changes in the charge density can be achieved by setting ICHARG to 11.

In case your system has localized electrons (often the case in transition metal oxides) a DFT+U approach might "open" the band gap in the system at the same cost than a PBE/GGA calculation.

[pascal_boulet1]

Dear Merkuk,

Thank you for your reply. That will help us a lot.

Best regards
Pascal

[pascal_boulet1]

Hello again,

I have tried with ICHARG=11, it works fine, thank you. Suprisingly (to me...), it necessitates more memory than with ICHARG=1, so I have to request more computer nodes. Is there a reason for this memory increase?

Best regards,
Pascal

[merzuk.kaltak]

Could you post the corresponding input and OUTCAR files for both runs showing this memory increase?