Please suggest me, how to treat pseudopotential and basis for f block elements like Sm in simulating Sm2O3. I used the psml pseudopotential from pseudo-dojo database, which gives quite accurate lattice constant, and non-magnetic ground state. But, it doesn't have 4f orbitals in valence configuration leading to vanishing DOS of 4f orbitals. However, the moment I used self-generated pseudopotential with valence 5s25p65d04f6 configuration, it gives a ferromagnetic ground state (not found in PW calculation), and also error in lattice constant exceeds 5%. So, what should be the correct way to do these calculations? Am, I making some mistake?

  • $\begingroup$ Leaving as a comment because it's not a complete answer: the simple answer is that DFT is not particularly good for f electron systems; the f electrons are typically fairly tightly bound with strong electron-electron interaction. I also have to give the disclaimer that I'm not an expert on SIESTA. I've had some success with Samarium using pslibrary for pbe pseudos and generating my own with oncvpsp for nc pseudos, but this is for plane wave systems (with QE). Samarium often exists as mixed valence so you could maybe try 5d1 4f5 or some fractional occupations? $\endgroup$
    – MSteg
    Jan 15 at 16:52


You must log in to answer this question.

Browse other questions tagged .