S. Hope et al., THICKNESS DEPENDENCE OF THE TOTAL MAGNETIC-MOMENT PER ATOM IN THE CU NI/CU/SI(001) SYSTEM/, Physical review. B, Condensed matter, 55(17), 1997, pp. 11422-11431
Systematic measurements of the magnetic moment per Ni atom in Cu/Ni/Cu
/Si(001) structures have been made using polarized neutron reflection
(PNR) for Ni thicknesses in the range 30 Angstrom < t < 400 Angstrom a
t room temperature. We find a dramatic reduction in the magnetic momen
t per atom for t < 100 Angstrom and near bulk values above 100 Angstro
m. These results are corroborated by alternating gradient magnetometer
measurements on the same samples. A Cu/Ni-wedge/Cu/Si(001) structure
with 30 Angstrom < t < 150 Angstrom was studied using magnetic circula
r x-ray dichroism (MCXD), polar magneto-optical Kerr effect (MOKE), an
d reflection high-energy electron diffraction (RHEED) in order to esti
mate the variation in the values of (L-z), (S-z), perpendicular anisot
ropy strength, and surface in-plane Ni lattice constant, respectively,
during epitaxial growth. RHEED measurements show that the in-plane la
ttice constant falls by 1.7% in the Ni thickness range 30 Angstrom < t
< 90 Angstrom. The MCXD measurements reveal the same trend for (L-z),
(S-z), and total moment per atom versus Ni thickness as found for the
total moment by PNR. Polar MOKE measurements confirmed the transition
from a perpendicular easy axis towards an in-plane magnetic easy axis
as has already been extensively studied in the literature. Comparison
of the PNR results with RHEED measurements reveal a striking correlat
ion between the increase of in-plane strain and reduction in magnetic
moment per atom with decreasing Ni thickness. While a direct strain-in
duced variation of the moment based on bulk phase calculations cannot
account for the magnitude of the moment variations we observe, we show
that the results cannot be attributed to sample contamination, interd
iffusion, or a reduction of the Curie temperature with decreasing Ni t
hickness. Furthermore, the presence of a magnetically dead layer in th
e samples is not consistent with the PNR results. The strong moment va
riation partially explains the large thickness range for which perpend
icular anisotropy is observed in this system.