Magnetic adatoms on symmetric slab

[mrm]

Dear head admin and all,

I apologize in advance if this has been discussed in the forums before but I spent almost an hour searching for it and could not find a similar situtation.

I am calculating the magnetic behavior of magnetic adatoms on a slab. I am using a symmetric slab to eliminate dipole effects. (I would prefer to avoid using the dipole correction in VASP.) Since I am using a symmetric slab I assumed that the initial MAGMOM directions for the adatoms should be positive for those on top of slab and negative for the ones on bottom of slab (or vice versa) in order to have a truly symmetric supercell. Could this result in a *zero* total magnetization of the unit cell (read from OSZICAR)? Obviously this is not what I want but will the symmetric slab model enforce such a result?

Assuming I actually obtain a reasonable nonzero answer, how should I interpret the total supercell magnetization? Should I divide it by two in order to obtain the magnetization for each side of the slab containing the adatom?

I am aware that setting LORBIT in INCAR will give me the magnetization projected onto individual atoms (as integrated over RWIGS spheres) but this may not be a quantitatively accurate result due to non-space-filling spheres for RWIGS.

I would really appreciate your help or suggestions. Thank you.

[alex]

Hi there,

if you feel running into that much trouble I'd suggest you give a one-sided adsorption a serious thought: is the effect you are interested in really that sensitive to a small dipol? Does the size of a potential error drop below a threshhold, where it does not matter anymore, if the vacuum is large enough?

Just my €0.02

Alex

[mrm]

Dear Alex,

Thank you for your suggestion. I have thought about the issue further and believe that I now understand my own question better in the original post and can answer it myself. However, I appreciate your suggestion and have gone the route that you suggest. I was already running an asymmetric slab before but had encountered trouble relaxing the structure beyond a certain point but your suggestion encouraged me to give it another try. After turning on spin polarization (for asymmetric slab) the structure has relaxed much better. Previously I thought that perhaps the dipole was preventing relaxation or possibly it was stuck in a local minimum. Thank you.

[alex]

Hi again,

the question is: how do you discriminate between a 'good' and a 'bad' structure? Did you do a frequency calculation?

Cheers,

Alex