That Fe films can be grown epitaxially on MgO and Al2O3 substrates is well-known. One might therefore expect to be able to grow high-quality Fe/Si superlattices on these substrates. Figure 13a) shows high-angle x-ray diffraction spectra for a purely (001)-oriented (Fe40Å/SiÅ)x60 multilayer grown on MgO(001). The spectrum in Figure 13b) is data for a highly (011)-oriented (Fe40Å/Si14)x46 multilayer grown on Al2O3. Both multilayers were deposited at +200°C. Figure 13c) shows a ø scan for the MgO (110) and Fe (110) peaks for the film on the MgO substrate. These sets of peaks are offset from one another by 45° in ø, confirming the well-known epitaxial relation Fe(001) || MgO(001) and Fe(110) || MgO(100). The ø scans for the Al2O3 substrate show that this film is only weakly oriented in-plane. Mattson et al. have previously grown Fe/FeSi multilayers on Al2O3, but they did not comment on the orientation of the multilayer. Rocking curves widths for both films are about 1° wide, indicating a considerably smaller mosaic than for the multilayers grown on glass. SAXS data for the multilayers on single-crystal substrates are comparable to the data for films grown on glass.
The films grown on MgO are the only purely (001)-textured Fe/Si multilayers produced by IBS so far. Dekoster et al. have grown epitaxial Fe/FeSi multilayers on MgO(001) by MBE, but they do not present any x-ray diffraction data or magnetization curves. Magnetization curves of films grown on single-crystal substrates (not shown) are qualitatively similar to those grown on glass or oxidized Si substrates. The only differences are that the saturation fields are higher for the epitaxial samples and that magnetocrystalline anisotropy effects are observed. The magnetocrystalline anisotropy energies of epitaxial trilayers grown on MgO and Ge are similar to bulk Fe.
The shape of the high-angle peaks plus superlattice satellites are described by a theory due to Fullerton et al. Application of this theory to the Fe/Si multilayers is difficult because the silicide lattice constant, the thickness of the remaining pure Fe and the thickness of the silicide spacer can be estimated only roughly. A precise determination of the silicide lattice constant should make a quantitative analysis of these satellite features possible.