Currently I'm using Optimized Norm-Conserving Vanderbilt Pseudopotential from the link https://github.com/pipidog/ONCVPSP.
I'm remember this discussion in the past, and to get the correct value we need to add more orbitals to the pseudo-potential, if I remember correctly. Is there a reference that discusses this precision issue in Quantum ESPRESSO?
Is there an openly available pseudo-potential I can use to improve my data?
I've found part of the answer to my question in this reference (sec II.A): Article Colloquium : Spintronics in graphene and other two-dimension...
It explains the requirement of d orbitals.
But there's still the question: is there QE pseudos with d orbitals for carbon easily available?
Gerson J Ferreira,
Apparently, there are not pseudopotentials for the C atom with information about the d orbitals. This is admittedly not completely surprising as the d orbitals in carbon are empty. Possibly, you can take the tool to generate pseudos in Q-E espresso (I believe it is called ld) and its input files for C and include the d-orbitals on the pseudo (plus this is relativistic pseudo as you want SOC). You will need to test it carefully, though, by comparing some calculations with the standard pseudo and the one you produced.
Sorry I cannot be of more help,
Roberto
Thanks, I've never tried to generate my own pseudo, but I'll read about it and maybe I'll try it.
I understand the motivation to include only occupied orbitals, but hybridizations with low energy unoccupied orbitals certainly are relevant. Wouldn't it be a better approach to always include the first few (not sure how many) unoccupied orbitals when building pseudos?
DFT Is not a theory for unoccupied states. As such those are in principle completely arbitrary. Moreover, the Hamiltonian is defined only by the occupied states. Therefore, one should be able to reproduce the unoccupied states.
- Badges
- Science topic