Theoretical predictions for magnetic anisotropy of superlattice defects (abstract)

Abstract

Most experimental superlattices contain numerous defects that substantially affect the anisotropy. Calculations based on two different theoretical approaches are presented for a variety of substitutional defects at the interface. The first method involves electronic structure calculations employing the local density approximation within the layer Korringa–Kohn–Rostoker technique. Defects are treated as periodic at a third nearest-neighbor spacing. Large in-plane contributions to the anisotropy are found for a substituted atom within the (111) Co layer of both the Co/Pd and Co/Pt systems. For example, −203 uRy is obtained for a single defect in the 1Co/1Pt superlattice: this is −4.2 times the interface anisotropy. This is believed to be the first ab initio prediction of the anisotropy energy of defects. The second method sums pair interactions using the potential (M⋅R)2, where M is the magnetization and R is the vector connecting the two atoms. This method has been previously demonstrated1 to be accurate in comparison both to electronic structure calculation and to experiment. For the fcc (111) interface, it is predicted that the anisotropy (in units of interface anisotropy) of a substitution in a monolayer is −4, the anisotropy of an adatom is −2, and the anisotropy of a one-atom recess that does not penetrate the magnetic layer is −2. These predictions are found to aproximately match the electronic structure theory in those cases we have tested. Finally, the summation of a variety of randomly chosen defects to form a diffuse interface yields an anisotropy equal to that of a perfect interface times the sum over all layers i of [P(i)−P(i+1)]2, where P(i) is the probability of a magnetic atom in layer i. This suggests that the highest anisotropy is obtained at the sharpest interface.