Discussion

The simplest model desribing NiO/NiFe exchange coupling indicates the highest HE should be observed when the NiO surface is oriented to maximize the number of uncompensated spins, i. e. the (111) planes [10] . We have shown that this model clearly does not describe NiO/NiFe exchange couples. The data represented in Figure 2 indicate the bulk texture of the NiO layer has very little influence on the interface exchange coupling energy since the exchange field does not correlate with the texture of the NiO films. Further, Figure 1b shows a strong H_E for NiFe films grown on the (001) thus should produce zero exchange field. The substrate thus should produce zero exchange field. The substrate properties influence H_E, however, since H_E for the bilayer on MgO is roughly one third that of the polycrystalline film deposited simultaneously on an amorphous buffer layer.

Our results agree with [5] who use an MOCVD technique to grow epitaxial (001) oriented NiO buffer layers. They also find non-zero HE values for NiFe grown on (001) oriented NiO, which are roughly half what they measure for NiFe grown on polycrystalline NiO buffers. However, they measure isotropic coercivities which are nearly an order of magnitude larger than those of reactively sputtered or IBS NiO films. This difference in coercivity may be due to the elevated surface roughness of the epitaxial MOCVD NiO layers relative to those deposited using the other techniques. [11] describes NiFe layers deposited on single crystal CoO substrates and shows that rougher interfaces or interfaces with more crystalline disorder produce higher exchange fields (CoO has structural and magnetic properties similar to NiO). Growth studies of IBS films indicate the broad distribution of adatom energies present in IBS deposition can produce very smooth surfaces and interfaces [12]. Low angle XRD data on the IBS grown NiO/NiFe bilayers show the surface has approximately 5Å of roughness.

In summary, models for the bilayer magnetic response should divide the NiO into two layers. We can estimate the thickness of a surface layer in the NiO, whose spins are dynamic during the NiFe reversal and so are responsible for the large NiFe coercivity, is equal to the critical NiO thickness needed to produce the unidirectional exchange anisotropy. A static NiO layer below the dynamic layer establishes the direction of HE. Large H_E are observed independent of the texture of the NiO layer. Thus the interfacial interaction of the NiFe and the NiO is not influenced by the average NiO morphology. Unfortunately it is difficult to probe the properties of the interface independently of the rest of the NiO film and observe the magnetic structures that form. Variations in H_c and H_E in coupled bilayers as well as the presence of a critical thickness, and the training effect, indicate that the dynamics of the interfacial antiferromagnetic domain structure of the NiO during the NiFe magnetization reversal is the key to understanding the magnetic response of oxide based exchange couples.[8]

The antiferromagnetic order of the NiO and the presence of non-zero H_E is unaffected by bulk morphological variations. In this sense, the NiO is a more forgiving exchange bias layer than FeMn/NiFe or NiMn/NiFe exchange couples, since these compounds require the specific FCC structure throughout the buffer layer to produce the antiferromagnetic phase and achieve a non-zero H_E.[13] This difference may be due to the more robust antiferromagnetism associated with the super-exchange interaction and ionic bonding in the oxide materials.

We would like to thank Keith Wilfinger for helpful discussions and R. G. Musket for RBS measurements. Part of this work was performed under the auspices of the U. S. Department of Energy (DOE) by LLNL under contract No. W-7405-ENG- 48, and part was supported by DOE's Tailored Microstructures in Hard Magnets Initiative. LEJ supported by DOE's Science and Engineering Research Semester, and Partners in Industry and Education programs.