The sandwiches used in this study were grown by molecular beam epitaxy using elemental sources. First, a ZnSe buffer layer a few thousand angstroms thick was deposited on a semi-insulating (100) GaAs substrate. Then the sandwich layers of Fe, Ag, and CoFe alloy were sequentially deposited on the ZnSe layer, in each case maintaining good epitaxial registry with the substrate. Finally a thin Ag or Al cap layer was deposited on each sample to prevent oxidation. The first Fe layer was always grown with a slightly elevated substrate temperature of 175C in order to optimize crystalline perfection,[17] but subsequent layers were grown with the substrate cooled to -10 to -20°C to minimize interdiffusion. Auger electron spectroscopy at each stage of growth showed that C and O contamination was less than 1%.
Reflection high-energy electron diffraction (RHEED) camera photographs of the surface at various stages of the growth are shown in Fig. 1. The photographs show that the first Fe layer goes down in the (001) orientation, as has been demonstrated previously.[17] The Ag and CoxFe1-x layers also go down in the (001) orientation, with the alloy layer in the bcc phase for all compositions despite the fact that at room temperature bulk CoxFe1-x is mixed-phase for Co fractions greater than x = 0.75.[18] The RHEED pattern for the pure Co film suggest that the surface is rough, and that there are crystalline defects. However, the symmetry of this film is clearly cubic, as demonstrated both by the RHEED pattern and by angle-dependent ferromagnetic resonance measurements which are described further below. A total of five Fe/Ag/CoxFe1-x sandwiches were grown with alloy compositions ranging from pure Fe (x = 0) to pure Co (x = 1).
Flux rates during growth were monitored using a calibrated UTI 100C quadropole mass analyzer. The total thickness of the deposited metal was measured by x-ray fluorescence (XRF) spectroscopy. The nominal thicknesses of the corresponding layers were kept constant from sample to sample at (Fe 80Å/Ag 60Å/CoxFe1-x 80Å/Ag 20Å). Actual layer thicknesses and alloy compositions were calculated by combining XRF and integrated Auger intensities measurements.
Using well-established analysis methods,[19] the angular dependence of the 35 GHz ferromagnetic resonance field was used to determine the magnetocrystalline anisotropy of each magnetic layer in each sample. Measured layer thicknesses and anisotropies are given in Table I. Results for the plain Fe films were similar to those previously observed for Fe on ZnSe.[19] Magnetic hysteresis loops were taken using a vibrating-sample magnetometer (VSM). Magnetoresistance measurements were made using Ag paint or pressed In contacts placed at the two ends of 2mm by 1cm strips cleaved off the original 1 cm by 1 cm wafer. Several samples were measured with both types of contacts, and no difference was observed. The measurement current was always along the GaAs [110] direction and the applied field always in-plane. All measurements were performed at room temperature. The sandwiches had resistivities in the range 15 ± 5 microOhm-cm with no systematic trend versus alloy composition.