Concluding Remarks

Winding breeze, through paths will
Airs contain from course and journey,
Which stir the secret, resting still,
Picked anew by restless air to carry,
As fleck, a flight in feather 'til...

We have investigated some of the mechanisms working in nanoscale systems, particularly multilayers. The effects of strain and layer thicknesses have been observed in a number of multilayer and thin film samples. The composition and growth quality of thin film structures has been investigated.

The room temperature CEMS results for all of the REFe$_2$ Laves Phase samples show a low symmetry magnetic easy axis as expected from earlier studies on similar systems. The analysis of dipolar and contact hyperfine fields showed the magnetic easy axis to be close to the $\left[\bar{2}41\right]$ and $\left[\bar{3}51\right]$ directions. Mössbauer spectroscopy on these samples appears to lack the resolution necessary to determine the easy axis with great accuracy using this method. Persistent problems in determining angular data from these spectra have been the overlap of the components and the multidomain nature of the untrained samples. The analysis of the hyperfine field contributions is sensitive to small differences in hyperfine field and quadrupole splitting, differences which are smeared by the close proximity of the four components. There is ample room in the spectra for reducing the velocity range of the attenuator and thus increasing the energy resolution of the spectra. A field trained sample, where one of the equivalent easy axis directions is preferentially populated, would allow the use of peak intensities to be used for angular information as well as peak positions.

The spin moments in all REFe$_2$ Laves Phase samples have been seen to lie out of plane in all samples by approximately $22^{\circ}$, indicating a significant magneto-elastic anisotropy. The use of $2.5\ensuremath{\unskip\,\mathrm{kOe}}$ in plane applied field measurements indicated a much larger magnitude of magnetocrystalline anisotropy in the DyFe$_2$ system than in the YFe$_2$ system as expected from the large single ion anisotropy of the dysprosium moments.

In the DyFe$_2$/YFe$_2$ multilayer samples the anisotropy is seen to be dominated by the dysprosium single-ion anisotropy. The DyFe$_2$(50Å)/YFe$_2$(50Å) sample shows average hyperfine fields consistent with the DyFe$_2$ and YFe$_2$ thin film results, whilst samples with thinner layers show an enhanced hyperfine field of up to $17\%$. The current data do not provide any information on the basis of this effect. Although the calibration of these spectra has been checked, sequential runs of at least some of these samples would validate the findings of the data. More samples with thinner layers still, if possible, would investigate further this effect.

The DyFe$_2$/Dy multilayers have shown identical zero field properties to the DyFe$_2$ thin film system down to a DyFe$_2$ thickness of $50\ensuremath{\unskip\,\mathrm{\AA{}}}$. Although the results are rather uninteresting they do provide complementary information for the multilayers of two magnetic layers, helping to isolate the sources of structural effects in other samples.

In all samples studied under applied field the hyperfine fields were reduced. This is expected in YFe$_2$ where only the iron atoms possess a moment. It would be expected in all other systems containing dysprosium for the hyperfine fields to increase. Sample alignment with respect to applied fields is very important in systems with significant magnetic anisotropy and this alignment is not known and cannot be easily measured. In any future work care must be taken to record the positioning of the samples within equipment where possible.

Preliminary SQUID magnetometry data were taken from these samples but the rather large YFe$_2$ seed layer producing exchange springs and difficulties in sample alignment to the applied field precluded any further magnetometry work. Other groups have countered these problems by using samples with only a thin Fe seed layer to remove the unwanted exchange springs and Vector VSM equipment to allow precise angular positioning of the samples. Any further samples studied should therefore follow the same example, allowing magnetometry data to be taken to complement the Mössbauer data.

SQUID magnetometry results from the Ce/Fe multilayers show that most of the samples exhibit antiferromagnetic coupling, with a $T_{N}$ ranging between $125\ensuremath{\unskip\,\mathrm{K}}$ and $190\ensuremath{\unskip\,\mathrm{K}}$, dependent upon both cerium and iron layer thicknesses. The poor signal to noise ratio of much of the data has made some of the interpretation difficult and prevented definite conclusions to be drawn on behaviour trends with layer thickness. A lack of sample area measurements has also prevented a full comparison of some of the data recorded. Retaking some of the data with known sample sizes would enhance the analysis greatly.

The exchange coupling constant, $J(z)$, has been calculated for the antiferromagnetically coupled samples and shows an oscillatory $z$ dependence. The number of data points does not allow a real comparison with theoretical predictions for exchange coupling across nonmagnetic layers, although the data so far suggest an RKKY-like $z$ dependence.

CEMS results from the U/Fe multilayers shows that each iron layer is composed of BCC iron, a poorly-crystalline iron layer with a reduced hyperfine field of up to $3\%$ and a doublet from a paramagnetic UFe$_2$ layer. The hyperfine parameters for the paramagnetic doublet match best UFe$_2$ but other intermetallics cannot be totally discounted. Low temperature work below $T_{C} = 167\ensuremath{\unskip\,\mathrm{K}}$ for UFe$_2$ on these samples would help corroborate this finding.

The relative thicknesses of the constituent layers within the U/Fe iron layer scale nonlinearly with the thickness of the deposited iron layer below $60\ensuremath{\unskip\,\mathrm{\AA{}}}$. Above this thickness the disordered iron and UFe$_2$ layers reach maximum thicknesses of $20\ensuremath{\unskip\,\mathrm{\AA{}}}$ and $18\ensuremath{\unskip\,\mathrm{\AA{}}}$ respectively. Where the uranium layer has poor crystalline growth this is propagated into the iron layer and increases these thicknesses. This sets a layer thickness limit on the growth of good uranium layers of $22\ensuremath{\unskip\,\mathrm{\AA{}}}$.

Room temperature Mössbauer spectroscopy results from a selection of printer toner powders were used to produce the ratio of magnetite to maghemite in the powders.

CEMS results on magnetite thin films showed good crystal growth on a Pt(111) substrate, with some iron forming a non-magnetic layer diffused in the platinum. The amount of iron in such sites may have been exaggerated by the substrate preparation history. Magnetite deposited on Al$_2$O$_3$(0001) substrates showed good crystal growth when using an oxygen plasma source. The statistics on the spectrum obtained from this sample are very poor, preventing any quantitative analysis of site occupancy. A longer run on this sample would be beneficial, particularly as this sample is the most favourable type for technological applications. A sample produced from a normal sputtering source showed a distribution of hyperfine fields and a paramagnetic contribution from iron substituting for aluminium in the substrate. This shows the benefits of the oxygen plasma source for producing quality thin films on the Al$_2$O$_3$ substrate.