Results

At the chosen Ag spacer thickness of about 60, ferromagnetic resonance shows no measurable antiferromagnetic interlayer coupling between the Fe layer and the alloy layer, consistent with previous results on Ag-spacer systems.[20,21] The qualitative features of the magnetization curves are expected to evolve in a complex way as a function of composition since the four-fold magnetocrystalline anisotropy (K₁) of bulk Co_xFe_1-x alloys changes from positive at x = 0 (pure Fe) to zero near x = 0.4 to negative near the bulk limit of the bcc phase at x = 0.7. 22 The values of K1 for the Fe/Ag/Co_xFe_1-x sandwiches are gathered in Table I. They are in generally good agreement with bulk values. The anisotropy of Co_xFe-_1-x films as a function of composition is discussed in more detail in Ref. 16.

A VSM loop of the sample with a pure Co layer is shown in Fig. 2a). Previous experience with films of bcc Co has shown[23] it to have a large negative four-fold anisotropy K₁ = -6.6x105 erg/cm³, so one anticipates that a Co film will have a <11> easy axis in the (001) plane. In fact, the Co film in the Fe/Ag/Co sandwich has a <10> easy axis and a large positive K₁. The reason for this discrepancy is not clear. The Co film does have a bigger coercivity than the Fe, as would be expected given its much larger anisotropy.[22]

Due to the difference in their coercivities, the moments of the Fe and Co layers are antiparallel for the field range (Hc_Fe = 0.015 kOe) < |H| < (Hc_Co = 0.470 kOe). As has been shown previously,[13,24,25] the antiparallel orientation of the magnetic moments in different layers is sufficient to cause the occurrence of the giant or spin-valve magnetoresistance. Figure 2b) shows the magnetoresistance data for the same Fe/Ag/Co sandwich whose magnetization curve is shown in Fig 2a). As expected, the resistance is maximum in the applied field region where |H| is less than the Co coercive field but greater than the Fe coercive field. The origin of the small shoulders in the MR near H = 0.38 kOe is harder to explain. These features may indicate that the Co film is reversing in two sequential 90° steps since the shoulders are near the halfway point. However, the easy-axis magnetization reversal of a thin film with cubic anisotropy usually occurs by the motion of 180° walls, not 90° domain walls.[22]

The Fe/Ag/Co71Fe29 sandwich whose magnetization data is displayed in Figure 3a) is quite different. Since the Co-rich alloy layer has a negative K1, its easy direction in the (001) plane is a <11> axis. Therefore the <10> magnetization curve in Fig. 3a) is an easy loop for the Fe film, but a hard loop for the alloy film. Knowing this we can easily see that the low-field square reversal with H_C = 0.04 kOe is attributable to Fe, while the higher-field linear magnetization characteristic corresponds to the Co_xFe_1-x layer. Because there is no significant interlayer magnetic coupling, the moments in each of the layers should each lie along their own easy axes at zero applied field, meaning that at H=0 there is a 45° angle between the Fe and Co_xFe_1-x moments.

The low-field magnetization behavior of the same sample is shown in more detail in Fig. 3b). As was mentioned previously, easy axis magnetization reversal in cubic-anisotropy materials generally proceeds by the motion of 180° domain walls.[22] The Fe magnetization curve therefore has a feature only at Hc, the coercive field. Hard axis reversal, on the other hand, should have the following steps as the applied field is decreased from a high value: first, reversible domain rotation of 45° from the hard axis to the easy axis; next, after the applied field reverses, motion of 90° domain walls or irreversible 90° rotation; lastly, reversible rotation 45° from the reversed easy axis to the hard direction along the applied field. All these features are observed in the hard-axis reversal of the alloy film in Fig. 3b). The moment of the alloy film jumps 90 degrees from one (110) axis to the other at a larger absolute field value than that at which the Fe film's moment reverses by 180° from one (100) direction to the other. Since the magnetometer measures only the component of the moment along the applied field, the 90° jump is positively identifiable as the smaller of the two low-field discontinuities of the moment in Fig. 3b). Fig. 3c) shows that, as expected, there is a magnetization peak in the field region where there is a maximum angle of 135° between the alloy-film moment and the Fe film moment.

As Figure 4 shows, data for the Fe/Ag/Fe sandwich (x = 0) is surprising. There is only one ferromagnetic resonance line for this sandwich, in contrast to the two resonances observed when x­ 0, so that the magnetocrystalline anisotropies of the two films appear to be identical. Previous work has shown that K1 in Fe films on ZnSe and Ag is not very sensitive to disorder, but that the coercive field Hc may be affected.[19] Broader RHEED streaks from the top Fe film indicate it has more defects than the lower Fe film, so a different coercivity for this film seems reasonable. From the general considerations described above, one can then attribute the lower-field peak in the magnetoresistance in Fig. 4b) to the film's moment irreversibly rotating 90° before the other film's. We therefore expect the angular separation of the Fe moments at the MR peak to be 90 degrees.

Clearly the higher-field magnetoresistance "2nd peak" in Fig. 4b) cannot be so easily explained. In order to understand the origin of the second peak, which does not correspond to any discernible feature in Fig. 4a), it is necessary to recall that the vibrating sample magnetometer only measures the component of the magnetization along the applied field. Thus the reversal of a component of the magnetization which is orthogonal to the applied field will not show up in a VSM loop. Florczak and Dahlberg have suggested that in thin (001) Fe films a small misalignment of the applied field from the hard <11> axis causes the magnetization reversal to occur as two 90° irreversible rotations.[26] The reason is that the moment then has two hard-axis energy barriers to surmount in order to complete a reversal. The second of the two transitions is unobservable in a VSM experiment because the component of magnetization perpendicular to the applied field reverses.[26] In keeping with this hypothesis, the higher-field magnetoresistance peak of Fig. 4b) may well coincide with a second 90° irreversible rotation by one of the Fe moments. The field at which the second MR jump occurs is in good agreement with the field at which Florczak and Dahlberg observed the reversal of the perpendicular component of the Fe magnetization using the Kerr effect.[26] In this picture, the higher-field MR peak is smaller than the lower-field peak because both moments have already rotated further toward the hard axis at the field where the second jump occurs. This leads to an angular separation of the moments less than 90° at the second peak, and thus a smaller resistance change.