Unconventional Exchange Bias and Multiferroics

Exchange bias is conventionally induced between a ferromagnet and an antiferromagnet where the Curie temperature (TC) considerably exceeds the Neel temperature (TN). Under these circumstances, cooling the bilayer in sufficient field to saturate the ferromagnet is sufficient to induce a finite bias below the blocking temperature (TB) which is close to T N. However, there are many systems of technological interest in which T C is not significantly greater than T N or in which it is not possible to field-cool the bilayer through T N; a highly relevant example is that of systems incorporating the multiferroic antiferromagnet BiFeO₃which has a T N above the temperature at which interfacial reactions appear to occur with conventional ferromagnets. This paper reports work on several systems which show unconventional bias, including detailed measurements of FeMn/CuNi bilayers in which it is demonstrated that bias can be induced despite TC = TN. The reason for this is that, at least for thin antiferromagnet layers, there are always likely to be regions of the interface in which the interfacial coupling energy exceeds the anisotropy energy of the antiferromagnet. This means that field-cooling the bilayer enables partial micromagnetic reordering of the antiferromagnet and provided that reversal of the ferromagnet during measurement at sufficiently low temperatures (i.e., below TB) cannot remove this then a small but permanent bias is induced. We have previously shown that this can be described in terms of an activation energy model which predicts the correct form of the dependence of bias on antiferromagnet thickness and temperature. This result is vital in understanding the problems associated with attempts to apply the high temperature antiferromagnet multiferroic BiFeO₃ in exchange bias systems. Here, a change in the electrical polarization o- - f the antiferromagnet is expected to lead to a corresponding change in the micromagnetic order and hence a change in the exchange interaction. However, this can only lead to the goal of an electrically-controllable exchange bias if the bias can be established in the first place, and the energy changes induced by the electrical reversal exceed the exchange energy at the interface. This paper will examine the limitations which will need to be overcome for successful application of this effect.