Figure 2 shows that H_{s} for
the t_{Cr} =16Å MBE-grown sandwich increases by
0.48 kOe between 300 K and 77 K. A previous FMR study of
single Fe films grown on GaAs found that
K_{1}/M_{s} increases by about 0.06 kOe when
T decreases from 300 to 77 K.[7] Assuming that
K_{1}(T) is similar for the Fe-Cr sandwiches grown on
ZnSe epilayers, the change of K_{1}/M_{s}
with temperature can account for only a 0.12 kOe increase in
H_{s} from 300 K to 77 K. Therefore the increase of
H_{s}(T) in the MBE-grown samples, while coming
partly from the temperature dependence of
K_{1}/M_{s}, is dominated by
A_{12}(T), just as in the polycrystalline and
superlattice samples. Currently, efforts are underway to
determine A_{12}(T) and K_{1}(T)
independently by fitting the H_{J}(T) and
H_{s}(T) data of Figure 2.

The temperature dependence of the magnetoresistance in the
Fe-Cr sandwiches is more complex. Like H_{s}, Dr/r
has two terms, one arising from the intrinsic anisotropy of
the Fe layers, and another arising from effects related to
the interlayer electron transport.[8] The Fe-layer
contribution, which is known as the anisotropic
magnetoresistance (AMR), is about the same size as in bulk
Fe.[8] The AMR is a sinusoidal function of the angle between
H and I. It is the second term, however, which is responsible
for the anomalously large magnetoresistance reported in Ref.
2. This anomalous contribution, which has been called the
spin-valve magnetoresistance, depends only on the relative
orientation of the Fe moments, and not on their absolute
orientation.[9] Therefore, the spin-valve magnetoresistance
should be isotropic with respect to the applied field
orientation. This fact allows one to separate the AMR and
spin-valve contributions to the H||I and H perpendicular to I
data of Figure 3(b): the AMR is the
difference between the H||I and H perpendicular to I traces,
while the spin-valve component is half the sum of the two
data sets. The temperature dependences of the separate AMR
and spin-valve terms are shown in Figure
4. The AMR is nearly constant with temperature, but the
spin-valve part of the magnetoresistance, while constant
below about 70 K, has a linear decrease at higher
temperatures.

Barthelemy et al. have suggested that the decrease of the
magnetoresistance with increasing temperature is due in part
to spin-wave scattering.[2] Interlayer spin fluctuations
could also be responsible for the linear decrease of
A_{12} with T. In this regard it is relevant to point
out that Dr, the field-induced change in resistivity, is
itself a function of temperature. The temperature dependence
of the spin-valve effect is therefore not just a trivial
consequence of the temperature variation of the resistivity.
The data shown in Figure 4 thus
contain real information about the scattering processes
responsible for the spin-valve magnetoresistance. At the
present time, however, the true nature of these scattering
processes remains unknown.

There are still many unanswered questions about the Fe-Cr
system. For example, Table I shows
that the magnitude of the magnetoresistance in the Fe-Cr
sandwiches does not appear to be correlated with either the
residual resistivity ratio or A_{12}. Nonetheless,
significant progress has been made towards understanding the
antiferromagnetic exchange and giant magnetoresistance of the
Fe-Cr superlattices and sandwiches.[3,4,5] The linearity of
the temperature dependence of the antiferromagnetic exchange
and spin-valve magnetoresistance should prove helpful in
understanding the role of spin waves and spin-flip scattering
in the Fe-Cr system.

The authors would like to thank J.R. Cullen, K.B. Hathaway, and A.C. Ehrlich for useful discussions, W.S. Rupprecht for help with data analysis, and D.R. King and F. Kovanic for technical assistance.

Wed Oct 11 09:49:01 PDT 1995