Ce(20Å)/Fe($x$Å)

Figure: Normalised magnetisation vs temperature scans for Ce(20Å)/Fe($x$Å) multilayers. Cerium layer thickness is constant.

Figure 7.1 shows the scans for Ce(20Å)/Fe($x$Å) multilayers. The sample with the thinnest iron layer, 20/10, shows either ferromagnetic or no interlayer coupling. There is a small reduction in magnetisation as the temperature increases, as expected in the ferromagnetic layers below $T_{c}$. This is illustrated in Figure 3.8(a). Although coupling will be seen in other samples with iron layers of this thickness, Mössbauer results[3] show this to be an anomalous sample, most probably because of the thinness of both the cerium and iron layers. The Mössbauer data for this sample shows iron moments lying $51^{\circ}$ out of plane and with a hyperfine field $12\%$ less than that of other samples with $10\ensuremath{\unskip\,\mathrm{\AA{}}}$ iron layers. Hysteresis measurements also show this anomalous behaviour (see Section 7.1.3).

As the iron layer thickness is increased to $15\ensuremath{\unskip\,\mathrm{\AA{}}}$ in the 20/15 sample we see coupling between the iron layers. The data for this sample shows an antiferromagnetic temperature dependence around $T_{N}$, as illustrated in Figure 3.7(a) (note the y-axis is $\nicefrac{1}{\chi}$), with a $T_{N} = 152\pm10\ensuremath{\unskip\,\mathrm{K}}$. The Mössbauer data for this sample shows a $12\%$ reduction in magnetic hyperfine field and spin moments lying $31^{\circ}$ out of plane. This is one possible explanation for the small effect compared to the samples with thicker iron layers.

The flat value of $M$ up to $T_{N} = 152\pm10\ensuremath{\unskip\,\mathrm{K}}$ corresponds to a system of layers antiferromagnetically coupled and, due to small in plane anisotropy, spin flopped with the antiferromagnetic axis normal to the applied field. The $M$ response is then due to the canting of the layer moments until their coupling to the applied field overcomes the antiferromagnetic coupling at $T_{N}$. The increase in $M$ with temperature up to $T_{N}$ for the 20/17 and 20/20 samples is likely to arise from incomplete spin flop of the antiferromagnetic system which allows some $\chi_{\parallel}$ combination to the $\chi_{tot} = \chi_{\parallel} + \chi_{\perp}$.

At an iron layer thickness of $17\ensuremath{\unskip\,\mathrm{\AA{}}}$ the characteristic change in $\chi_{tot}$ for an antiferromagnet is much more pronounced. The Mössbauer data show this sample to have only a $9\%$ reduction in hyperfine field and spin moments fully in plane. The increased in plane magnetisation in the iron layers may be giving an increased variation in $\chi_{tot}$ compared to the 20/15 sample, assuming the same moment orientation. This sample has a $T_{N} = 160\pm5\ensuremath{\unskip\,\mathrm{K}}$.

As the iron layer thickness is increased further to $20\ensuremath{\unskip\,\mathrm{\AA{}}}$ the effect becomes stronger still. For this sample $T_{N} = 96\pm5\ensuremath{\unskip\,\mathrm{K}}$. A larger change in $\chi$ between $1.8\ensuremath{\unskip\,\mathrm{K}}$ and $T_{N}$ is tending to give a lower $T_{N}$, consistent with a less uniform alignment of spin moments to the field producing a weaker coupling across the cerium layers (the 20/15 sample cannot be included in this comparison due to its undetermined moment orientation).

From this set of results it can be seen that at this cerium layer thickness where the iron moments are not significantly out of plane the coupling is antiferromagnetic.