Condensed Matter Physics Seminar

Exchange bias refers to a shift of the center of the ferromagnetic hysteresis loop away from zero-field effectively resulting in a unidirectional anisotropy. This phenomenon is observed as a result of the exchange interaction at the interface between the ferromagnet, which is usually a thin film or nanoparticle, and an antiferromagnetic material.  Despite the fact that this effect was discovered approximately fifty years ago and is used in magnetic sensors found in commercial hard drives, the fundamental mechanism responsible for this interaction was poorly understood until recently. Ideal antiferromagnets (antiferromagnets with relatively simple magnetic interactions) can be used, in principle, to assess or validate theories that explain exchange bias. Many of the proposed theories rely on the existence of a net magnetization at the antiferromagnet's surface and some predict that the exchange bias should increase as the size of antiferromagnetic domains is decreased. We have used transition metal difluoride epitaxial thin films, such as Fe_xZn_1-xF₂ and Fe_xNi_1-xF₂, to test these theories. These materials have a simple rutile, body-centered tetragonal crystal structure where next-nearest neighbor magnetic exchange interactions dominate. For the case of Fe_xZn_1-xF₂/Co bilayers, the dilution of the FeF₂ antiferromagnet with increasing non-magnetic ZnF₂ concentration leads to smaller domains and a frozen magnetization in the antiferromagnet. On the other hand, Fe_xNi_1-xF₂/Co is a system where the anisotropy of the antiferromagnet can be changed from being parallel to the [001] axis, for the case of pure FeF₂, to being in the (001) plane, for the case of pure NiF₂, without significant changes in the structure. For intermediate concentrations the anisotropy is weakened, allowing a study of exchange bias as a function of the magnetic anisotropy in the antiferromagnet. I will describe recent experiments in these systems which demonstrate that aspects of the proposed theories (such as the existence of net magnetic moments in the antiferromagnet at the interface with the ferromagnet) are indeed correct, whereas other aspects (such as the dependence of the exchange bias on the antiferromagnetic domain size) are not verified by the data. The critical importance of the antiferromagnetic anisotropy in determining the magnitude and dependence of the exchange bias will also be discussed.

This work was supported by the National Science Foundation