A typical H perpendicular to I magnetoresistance trace for an
MBE-grown specimen is shown in Figure
1. At zero applied field, the moments of the two Fe films
are antialigned due to the exchange coupling
A_{12}.[1] As has been explained previously,6 at the
field indicated in Figure 1 as
H_{J}, the magnetic moment of one of the two Fe
layers jumps by about 90 degrees toward the applied field
direction. Magnetic saturation is finally achieved at the
field H_{s}. In the analysis of the data,
H_{J} is defined as the applied field where |dr/dH|,
the slope of the resistivity, has a maximum. H_{s} is
determined by fitting the linear portion of the
magnetoresistance data for H_{J} < H <
H_{s}, and taking the field at which the linear fit
reaches the saturation resistance. The fields H_{J}
and H_{s} are shown in Figure
2 as a function of temperature for an Fe-Cr-Fe sandwich
with a chromium thickness, t_{Cr}, of 16Å.
Clearly the temperature variation of H_{s} is well-
described by a straight line, but H_{J} shows no
temperature dependence within experimental error. The
temperature variation of H_{s}, though small, is
significant since H_{s} is a function of the
antiferromagnetic exchange. When H is along a hard axis, as
in Figure 2, the relation is:

H_{s} = 4A_{12}/tFeM_{s} +
2K_{1}/M_{s} (1)

where A_{12} is the interlayer exchange coupling per
unit area, K_{1} is the fourth-order cubic
magnetocrystalline anisotropy energy characteristic of Fe,
t_{Fe} is the thickness of a single Fe layer, and
M_{s} is the saturation magnetization.[2,6] There is
at the moment no closed-form formula for H_{J}, but
it is also a function of K_{1} and A_{12}.

The saturation field H_{s} was determined for the
evaporated samples in a similar fashion as for the MBE-grown
samples. The slope and zero-intercepts of the linear fits to
H_{s}(T) for both kinds of sandwiches are presented
in Table I. Also included are
parameters derived from a linear fit to H_{s}(T) data
in Ref. 2. This data was taken on an MBE-grown Fe-Cr
superlattice with twenty repeats of t_{Fe} =
16Å and tCr=12 Å.[2]

The magnetoresistance itself as a function of temperature was also determined from individual field sweeps like that shown in Figure 1. Figure 3(a) shows the temperature variation of the H perpendicular to I magnetoresistance for two MBE-grown samples, while Figure 3(b) displays the temperature dependence of both the H perpendicular to I and H||I magnetoresistance for a polycrystalline sandwich. The data for the other evaporated polycrystalline sandwich (not shown) are nearly identical to those in Figure 3(b). From these plots one can see that Dr/r(T) in all samples is nearly constant up to about 70 K, above which it decreases linearly with temperature.

The values of the exchange constant A_{12} in Table I were determined using Equation
(1) along with (K_{1}/M_{s}) = 0.25 kOe for
the MBE-grown samples, a value determined by ferromagnetic
resonance (FMR) measurements.[6] For the polycrystalline
sandwiches, which have a small observed in-plane magnetic
anisotropy, H_{s} =
(4A_{12}/tFeM_{s}). The linearity of
H_{s}(T) then implies that the antiferromagnetic
exchange A_{12} is also approximately linear with
temperature. In the Fe-Cr superlattices, where the exchange
is very large,[2] the linearity of H_{s}(T) similarly
requires a linear temperature dependence of the exchange. The
linearity of A_{12}(T) is a potentially useful clue
about the fundamental nature of the interlayer exchange
interaction.

Wed Oct 11 09:49:01 PDT 1995