Gy. Guo, STRAIN, INTERDIFFUSION, MAGNETISM AND MAGNETIC-ANISOTROPY IN CU NI/CU(001) SANDWICHES/, Journal of magnetism and magnetic materials, 176(2-3), 1997, pp. 97-110
The effects of interface, strain and interdiffusion on magnetic moment
s and magnetocrystalline anisotropy in Cu/Ni/Cu(0 0 1) trilayers have
been studied by first-principles spin-polarized relativistic band-stru
cture calculations. Ideal Cu/Ni-n/Cu(0 0 1) trilayers are modelled by
NinCu5(0 0 1) superlattices (n = 1, 3, 5) whilst Ni/Cu interdiffusion
is simulated by (NiCu)(2)Ni-n(NiCu)(2)Cu-3(0 0 1) superlattices (n = 3
, 5, 7) containing two Ni/Cu mixed monolayers (ML) at each Ni/Cu inter
face. The main results include that ideal Cu/Ni-n/Cu(0 0 1) trilayers
(n greater than or equal to 5) exhibit magnetoelastic properties very
similar to that of bulk Ni and the flat Ni/Cu interface anisotropy is
in-plane but small (-0.01 meV/surface atom), that the perpendicular ma
gnetocrystalline anisotropy (PMA) of ideal Cu/Ni-n/Cu(0 0 1) trilayers
(n greater than or equal to 5) predominantly comes from the magnetoel
astic anisotropy due to Ni/Cu lattice mismatch strain, and that one si
ngle Ni monolayer embedded in Cu is nonmagnetic and so are the Ni atom
s in the Ni/Cu mixed layers two Mt away from the Ni slab. The first sp
in-orientation transition from an in-plane orientation to the perpendi
cular magnetization in the ideal Cu/Ni/Cu(0 0 1) trilayers is predicte
d to occur at Ni film thickness of about 3 ML. Introducing (-0.29 meV/
surface atom), thus postponing the first spin-orientation transition t
ill 6 ML. Ni film thickness of the second spin-orientation transition
from the perpendicular direction back to an in-plane magnetization is
estimated to be around 40-60 ML. The calculated bulk Ni magnetostricti
on coefficient is in reasonable agreement with experiments. The calcul
ated PMA is related to the Ni orbital moment anisotropy and also to th
e Ni d-orbital-decomposed density of states near the Fermi level. (C)
1997 Elsevier Science B.V. All rights reserved.