All lectron Carbon PseadoPotential

[ronald_redmer1]

Dear Developers,

I am currently planning to investigate hot, dense hydrocarbons at temperatures exceeding 100,000 K. For this, I require an all-electron carbon pseudopotential. I would appreciate any information on whether such a pseudopotential can be provided.

Thank you in advance for your assistance.

Best regards,
Armin Bergermann

[martin.schlipf]

I am not sure whether such a potential can be generated easily. One could probably construct a 1/r potential but I am not sure what kind of grids you would need to really converge that.

Could you elaborate why you need to include the 1s state into the valence window? Even if you have 100,000 K that is just 8.6 eV which is a small fraction of the binding energy of the 1s state. So I would still expect that these electrons are mostly inert.

I am unsure how VASP would perform under these extreme conditions. I could imagine that many default choices from potentials, number of bands, cutoff, ... would not be appropriate. If you do run calculations like these, you need to make sure that you are carefully testing the parameters.

[ronald_redmer1]

Hi,

I should have been more precise: The calculations are between 100000 and 500000 K. At these temperatures the core electrons become important. We use VASP regularly for temperatures up to 1.000.000 K. (see e.g. https://doi.org/10.1103/PhysRevResearch.2.023260) I agree, many default choices might not be correct, e.g. the number of bands must be increased drastically, Pseudos must be checked carefully, ....

The 1/r potential does not help for mixtures - at least I was not able to set it up correctly. It did work though for carbon only simulations. Years ago, you constructed an all electron potential for Oxygen ( PAW_PBE O_sv 05Jul2007 8.00000000000000 ). That's why I assumed something similar should be possible for Carbon.

Thank you very much!

[ronald_redmer1]

Hi,

are there any updates?

Thank you very much for your advise/help,

Armin Bergermann

[martin.schlipf]

Sorry, I did not reply to your last message.

We discussed this in our group meeting and Georg believes it should be possible to construct a C_sv potential. I cannot promise a fast solution for this issue, it depends on when Georg finds time to look at it.