A significant loss in the giant magnetoresistive signal of magnetic stacks with antiferromagnetic coupling across nonmagnetic intermediate layers is caused by regions with a ferro- rather than an antiferromagnetic coupling. The impact of these ferromagnetic coupling regions extends itself into the lateral direction due to the bulk exchange coupling. The present micromagnetic model provides a tool by which a detailed quantitative evaluation of the impact of periodic arrays of parallel line defects is possible. These defects have deviating exchange-coupling constants, and/or anisotropy constants or directions, bulk exchange constants, saturation magnetization, etc., in specific regions. Previously, we developed a phenomenological model of trilayers with two magnetic films separated by a nonmagnetic interlayer that contains one such defect. This model, with a relatively small number of free parameters, allows one to trace complete hysteresis curves. A large number of mode branches reveal themselves and jumpwise transitions between these modes frequently occur along the hysteresis loops. The present micromagnetic model requires a sufficiently accurate assessment of the starting magnetization configuration in order to get a convergence of the code. In general, the micromagnetic code is not capable of overcoming the above irreversible mode conversions. The mode branches evaluated by the phenomenological model are applied to provide the micromagnetic model with appropriate starting configurations after meeting a situation of nonstability. The micromagnetic theory of Brown constitutes the basis of the present approach. The micromagnetic effective field is calculated at grid points and the torque exerted by it on the magnetic dipole is made zero at each grid side by an iteration scheme. The long ranging magnetostatic fields are given by convolution integrals and are evaluated in the Fourier space by using two-dimensional fast Fourier transforms. The single defect is micromagnetically studied by zero padding techniques. Depending on the course of the external field, two different wall regions reveal themselves, to wit, the wall core and the so-called Néel tails. These tails were not incorporated into the phenomenological model. Provided that the defects are sufficiently wide spaced, the agreement between both models is rather good in the core regions. The impact on the GMR signal, in particular of the Néel tails, will be discussed with emphasis on systems with weak interlayer coupling, e.g., the decoupled systems.
Read full abstract