hackingmaterials
Default POTCAR settings for Lanthanides not compatible with Hybrid DFT
Trying to run a hybrid calculation on a systems containing Lanthanide elements (specifically Dy and Er) I ran into an error:
internal error in SET_CORE_WF: core electrons incorrect
Apparently, this is because some of the lanthanide potentials are not compatible with hybrid DFT (https://cms.mpi.univie.ac.at/vasp-forum/viewtopic.php?t=11307). The default Dy and Er POTCARs are Dy_3 and Er_3, instead we should be using just Dy and Er.
Interestingly, my calculation with Pr ran fine even though the POTCAR used was Pr_3. Some testing to see which POTCARs are compatible with hybrid DFT is needed.
I'm a bit confused by that thread -- if you use, say, the Gd_3 POTCARs are your calculations charged-balanced? It says there are 18 electrons for the Gd and 9 electrons for the Gd_3. Either way that seems short an electron? Also the Gd POTCAR we're using is now almost 20 years old . . . (I understand your question is about Dy/Er)
I didn't understand the part about charge-balanced (is part of the thread missing)?
But for Gd I've had problems in the past: Gd_3 doesn't play well with hybrid, but Gd is very difficult to converge for systems I've tried.
I'm just very confused by what these pseudopotentials are doing. It doesn't help that there's no documentation on how they were generated.
@utf did you find out any additional info on this? Hoping to start a conversation here about modernizing our pseudopotential choices, so this might be helpful :)
After some testing, I've found this isn't an issue when using the latest pseudo potentials.