[Truncated abstract] Although exchange bias was discovered more than four decades ago, a satisfactory understanding of every instance of exchange bias observed in experiment has not yet emerged. Understanding exchange bias is complicated by many factors. For example, details of the antiferromagnet interface structure set up during the initial field cooling, thermal activation processes in the ferromagnet and antiferromagnet, and domain formation and domain wall movement in the antiferromagnet are all important in determining features associated with exchange bias. Two exchange bias systems are investigated in this thesis. One is a disordered system: a single layer Co/CoO film with random interfaces prepared by a reactive RF sputtering technique. The other is a ‘model’ system of Fe/KFeF 3 bilayers with compensated interfaces prepared by molecular beam epitaxy (MBE). The central theme of this work is to understand exchange bias and other related magnetic properties in these two very different systems. The Co/CoO exchange bias system studied here is different in structure from conventional exchange bias systems such as bilayer and multilayer systems where interfaces between ferromagnet and antiferromagnet are reasonably well defined. In this Co/CoO system, the Co and CoO is in the form of particles distributed randomly in a sputtered film. The interfaces between the Co and CoO are randomly distributed and may not be continuous over a large length scale. More importantly, the interface area is dependent on the shape and size of the particles and on their distribution. Many unique magnetic properties are related to the random interface in this system. For example, exchange bias and coercivity obtained at low temperatures are very large due to the large interface area between Co and CoO particles. The interface area can be controlled by changing the Co/CoO mass ratio in the film. Unlike in bilayer systems, film thickness in this single layer Co/CoO system turns out not to be critical for exchange bias and coercivity. The independence of film thickness may be technically important. More interestingly, because the interface is random, exchange bias can be setup by field cooling in any direction. Both training and magnetic viscosity effects were studied and provided evidence of thermal activation processes in this Co/CoO system. Training is explained as formation of a domain wall in the CoO with motion limited locally due to limited continuity of Synthesis and Characterization of Magnetic Thin Films - Exchange Bias Systems interfaces between the Co and CoO. Specific magnetization measurements over time were made and studied using viscosity theory. The magnetic viscosity was found to be strongly temperature dependent. There is a broad distribution of blocking temperatures which might be due to a broad distribution of Co particle sizes

This work investigates the growth dynamic of the reactive molecular beam epitaxy of Fe3O4 films, and its impact on the cation distribution as well as on the magnetic and structural properties at the surface and the interfaces. In order to study the structure and composition of Fe3O4 films during growth, time-resolved high-energy x-ray diffraction (tr-HEXRD) and time-resolved hard x-ray photoelectron spectroscopy (tr-HAXPES) measurements are used to monitor the deposition process of Fe3O4 ultrathin films on SrTiO3(001), MgO(001) and NiO/MgO(001). For Fe3O4\SrTiO3(001) is found that the film first grows in a disordered island structure, between thicknesses of 1.5nm to 3nm in FeO islands and finally in the inverse spinel structure of Fe3O4, displaying (111) nanofacets on the surface. The films on MgO(001) and NiO/MgO(001) show a similar result, with the exception that the films are not disordered in the early growth stage, but form islands which immediately exhibit a crystalline FeO phase up to a thickness of 1nm. After that, the films grown in the inverse spinel structure on both MgO(001) and NiO/MgO(001). Additionally, the tr-HAXPES measurements of Fe3O4/SrTiO3(001) demonstrate that the FeO phase is only stable during the deposition process, but turns into a Fe3O4 phase when the deposition is interrupted. This suggests that this FeO layer is a strictly dynamic property of the growth process, and might not be retained in the as-grown films. In order to characterize the as-grown films, a technique is introduced to extract the cation depth distribution of Fe3O4 films from magnetooptical depth profiles obtained by fitting x-ray resonant magnetic reflectivity (XRMR) curves. To this end, x-ray absorption (XAS) and x-ray magnetic circular dichroism (XMCD) spectra are recorded as well as XRMR curves to obtain magnetooptical depth profiles. To attribute these magnetooptical depth profiles to the depth distribution of the cations, multiplet calculations are fitted to the XMCD data. From these calculations, the cation contributions at the three resonant energies of the XMCD spectrum can be evaluated. Recording XRMR curves at those energies allows to resolve the magnetooptical depth profiles of the three iron cation species in Fe3O4. This technique is used to resolve the cation stoichiometry at the surface of Fe3O4/MgO(001) films and at the interfaces of Fe3O4/MgO(001) and Fe3O4/NiO. The first unit cell of the Fe3O4(001) surface shows an excess of Fe3+ cations, likely related to a subsurface cation-vacancy reconstruction of the Fe3O4(001) surface, but the magnetic order of the different cation species appears to be not disturbed in this reconstructed layer. Beyond this layer, the magnetic order of all three iron cation species in Fe3O4/MgO(001) is stable for the entire film with no interlayer or magnetic dead layer at the interface. For Fe3O4/NiO films, we unexpectedly observe a magnetooptical absorption at the Ni L3 edge in the NiO film corresponding to a ferromagnetic order throughout the entire NiO film, which is antiferromagnetic in the bulk. Additionally, the magnetooptical profiles indicate a single intermixed layer containing both Fe2+ and Ni2+ cations.