The CEMS fit results are shown in Table 8.3 and the spectra are shown in Figures 8.3 and 8.4.

Table: Best fit parameters for magnetite thin film samples. The magnetic hyperfine field for the Fe$ _3$O$ _4$/Al$ _2$O$ _3$B sample is a distribution, shown in Figure 8.4(b).
Sample IS QS Field Area
(mms) (mms) (kG) ($ \%$)
Fe$ _3$O$ _4$/Pt $ 0.24$ $ -0.06$ $ 479.8$ $ 28$
$ 0.65$ $ +0.00$ $ 450.6$ $ 62$
$ 0.35$ $ +0.00$ 0 $ 10$
Fe$ _3$O$ _4$/Al$ _2$O$ _3$A $ 0.32$ $ +0.00$ $ 490.2$ $ 21$
$ 0.66$ $ +0.04$ $ 451.2$ $ 79$
Fe$ _3$O$ _4$/Al$ _2$O$ _3$B $ 0.45$ $ -0.08$ (Fig. 8.4) $ 80$
$ 0.33$ $ +0.89$ 0 $ 20$

The spectrum for Fe$ _3$O$ _4$/Pt shows good crystalline growth of magnetite. The spectrum can be fitted well with three components: two magnetic sextets and one non-magnetic singlet. The two sextets show an exact match with the bulk hyperfine parameters for Fe$ _3$O$ _4$ A and B sites.[57,58]

The non-magnetic singlet, with an isomer shift of $ 0.35\ensuremath{\unskip\,\mathrm{\nicefrac{mm}{s}}}$, matches exactly that for iron in a platinum matrix.[7] Thus we conclude that this component is obtained from iron diffused into the platinum substrate.

The ratio of A sites to B sites for this spectrum is 0.45:1 which is slightly less than that of bulk samples where this ratio is 0.53:1. In this system there are a limited number of layers. With A and B layers stacking one on the other it is possible that there is an uneven number of A sites, giving a smaller fraction of A sites than in an infinite lattice. If B sites were formed on the upper and lower extent of the thin film this would give a ratio of A:B of 0.45:1 for a $ 54\ensuremath{\unskip\,\mathrm{\AA{}}}$ film assuming the bulk lattice constant of $ 8.39\ensuremath{\unskip\,\mathrm{\AA{}}}$.[59]

The spectrum for the Fe$ _3$O$ _4$/Al$ _2$O$ _3$A sample also shows good crystalline growth. This spectrum requires only two components to give a good fit: two magnetic sextets. The hyperfine parameters for the two sextet components again are consistent with those for bulk samples.

The statistics of the Fe$ _3$O$ _4$/Al$ _2$O$ _3$A spectrum are too poor to give accurate information about the relative intensity of the A and B sites: the A site linewidth needs to be fixed to give a sensible result.

Figure 8.3: Fe$ _3$O$ _4$ thin films on (a) platinum and (b) sapphire substrates. Sample (b) was grown using an oxygen plasma source.

The spectrum for the Fe$ _3$O$ _4$/Al$ _2$O$ _3$B sample is shown in Figure 8.4(a). This has been fitted with two components: a distribution of magnetic sextets and a doublet. The population of hyperfine fields in the distribution is shown in Figure 8.4(b).

Figure 8.4: Fe$ _3$O$ _4$ thin film on a sapphire substrate, grown with a standard sputtering source. (a) shows the fitted spectrum and (b) shows the hyperfine field distribution.

The distribution has an average isomer shift of $ 0.45\ensuremath{\unskip\,\mathrm{\nicefrac{mm}{s}}}$. This is inbetween the values for A and B site isomer shifts in magnetite and is due to the single distribution component combining both the A and B site components as would be expected in a spectrum from a more crystalline sample. The distribution population reaches a peak and then ends at $ \sim470\ensuremath{\unskip\,\mathrm{kG}}$, consistent with an iron oxide, most probably magnetite as evidenced by the isomer shift value.

The hyperfine field depends sensitively upon the local environment of the Mössbauer atom. A distribution of hyperfine fields indicates a distribution of inequivalent iron sites. The distribution shows that the most likely environment is that of normal, crystalline magnetite. There is also, however, a substantial area of iron at a reduced hyperfine field, showing a reduced number of magnetic neighbours or an increased spacing between them. The majority of the reduced hyperfine field component occurs between $ \sim280\ensuremath{\unskip\,\mathrm{kG}}$ and $ \sim380\ensuremath{\unskip\,\mathrm{kG}}$. A small peak at $ 330\ensuremath{\unskip\,\mathrm{kG}}$ may indicate the presence of some metallic bcc iron. Both effects point to substantial defects in the thin film and that the normal sputtering source is not capable of producing films on Al$ _2$O$ _3$ substrates of the quality necessary for MR applications.

The doublet has an isomer of shift of $ 0.33\ensuremath{\unskip\,\mathrm{\nicefrac{mm}{s}}}$, indicative of an Fe$ ^{3+}$ charge state. A possible cause is a small amount of iron pentrating into the substrate and substituting for aluminium in the Al$ _2$O$ _3$ compound. At low concentrations in the Al$ _2$O$ _3$ substrate the Fe$ ^{3+}$ ions would not be magnetically ordered and at room temperature the spin-lattice relaxation times would be much shorter than the Mössbauer sensing time (see Section 2.4) resulting in a doublet component.

Dr John Bland, 15/03/2003