You can estimate polarization by integrating the charge density difference. Basically, you take the charge density from CHGCAR files (with and without polarization, ISPIN =2 and SPIN =1, respectively) and use VASPKIT (I believe it has this functionality), or custom scripts to obtain the polarization (perform the integration).

Hope that helps,

Good luck~!

Thanks for your answer. Can you please elaborate this-(with and without polarization, ISPIN =2 and SPIN =1, respectively); do you mean from spin component 1 and spin component 2?

Sure thing. ISPIN = 2 is for spin-polarized calc., and ISPIN = 1 is for non-polarized calc. Whatever calculation you're running, run it independently twice: one with ISPIN = 2 set in your INCAR file and the other with ISPIN = 1. Once both have found their ground states, you do what I suggested by taking the CHGCAR files generated by each run.

See if it helps,

Rgds

The Berry phase method is for *electric* polarisation, it can tell you things like the dipole moment. However, for half-metals you almost certainly are wanting the *spin* polarisation at the Fermi-level. This requires you to calculate the density-of-states using a good k-point integration mesh, and then to calculate the fraction of up and down states at the Fermi energy.

If the Fermi energy is close to a band-edge, you will need a very fine k-point mesh or your results will be dominated by whatever smearing you use as your integration kernel. Depending on the material, you may also need a Hubbard U term in the Hamiltonian to counteract the self-interaction error.