Apparent inconsistency in the magnetic moment of Li2NiO2


I am currently exploring battery materials, and I am a bit puzzled by the information provided on magnetism.

For the compound Li2NiO2 (, the final magnetic moment is 2.000 Bohr magnetons, but the magnetic ordering is NM. Does NM mean “non-magnetic” ? If so, how is it possible that the material would be non-magnetic, but still have a finite magnetic moment ?



This entry’s structure has no site property of magmom. so, this entry becomes NM.

>>> s = m.get_structure_by_material_id('mp-19308')
>>> print(s.site_properties)

This entry should have magmom because it has the final magnetic moment of 2 Bohr magnetons.

1 Like

Thanks for finding this. This is bug from when we started directly incorporating the dielectric calculations into our infrastructure. The total magnetic moment from the dielectric calculation is being chosen over the structure relaxation hence this issue. Our structure optimization calculation is run without magnetic moments here so this material should be labeled Non-magnetic for now. We’ve also debated labeling the magnetic ordering as “Unknown” when we don’t have a calculation with a specific magnetic ordering.

We’ll try and fix this for this upcoming release. If not, it will definitely happen at the end of March.

EDIT: Actually, scratch that. This bug is a bit more subtle. We have a number of calculations from early MP ~ 2012, where there was an issue with what VASP wasn’t properly printing out the final magnetic moments on a per atom basis. We will schedule this calculation to be rerun so we can get the proper final magnetic moments.

1 Like

Yes, the correct reported ordering here should be “Unknown.”

Clearly the material is magnetic given that the total magnetization is positive, but this specific calculation did not have site-projected magnetic moments, so we do not have enough information to say whether it is ferromagnetic, ferrimagnetic, etc. (it would probably be reasonable to assume in this case that it’s actually ferromagnetic, but we can’t prove that algorithmically).

For anyone unfamiliar with the kind of DFT calculations that the Materials Project does, the net magnetization of any one given spin-polarized calculation is unambiguous and known exactly from the output of the calculation, but assigning a magnetic moment on a per-site (per-atom) basis is not an exact science and is an additional step, and has historically omitted from certain types of calculations we perform.

Since we did not have site-projected magnetic moments for this calculation, the ordering was incorrectly labelled as “non-magnetic” since the ordering analysis is done on the structure, did not find a structure annotated with magnetic moments, and therefore concluded it was not magnetic.

(Also thanks @shyamd for investigating!)

1 Like

Thank you very much for the detailed explanations !

1 Like