Elastic constant measurement by energy with spin polarization

[terencelz]

Hi all,

I'm doing some transition metal nitride calculation, and some of them are ferro-magnetic. I understand spin polarization should be considered (ISPIN = 2) to determine the lattice constant, but when one fits the equation of state, he also gets bulk modulus. Further, if he uses this lattice constant and ISPIN = 2 to get the total energy-strain curve (hopefully 2nd order polynomial), spin polarization would be a bit different, thus contributing to the total energy difference.

Question is, should this contribution be considered to determine elastic constants? Does spin polarization physically change as fast as one distort the cell? If not, what method is more physically reasonable?

I also looked at the stress tensor, and surprisingly, the relevant matrix element is proportionate to the energy difference. i.e. when ISPIN = 1 and energy difference is small, stress is small at the same strain level. When ISPIN = 2, energy difference is large, stress is large as well. This is interesting, because magnetization is directly reflected on stress.

Am I correct to make such an analysis?
<span class='smallblacktext'>[ Edited ]</span>

[admin]

if a material is magnetic, spin polarization has to be taken into account in the calculations as well of course, not just for the calculation of the lattice constant