This thesis presents and explains the effect of structure, dilution, anisotropy and temperature on the exchange bias effect in metallic ferromagnetic/antiferromagnetic (FM/AFM) thin films. The exchange bias effect appears at the interface between a FM and an AFM when the FM/AFM bilayer is either deposited or cooled below the Néel temperature of the AFM, in the presence of a magnetic field. The effect manifests itself in a shifted and broadened hysteresis loop of the FM/AFM bilayer along the field axis direction. This is the consequence of the unidirectional anisotropy induced in the FM layer which results from the exchange coupling across the FM/AFM interface. Despite intensive research activities in this field, many aspects of exchange bias are not yet fully understood. This refers also to the role of structure, dilution by nonmagnetic defects, anisotropy and temperature on the exchange bias of metallic FM/AFM bilayers. However, the exchange bias is used in many applications such as spin valves, read heads, magnetic random access memories, etc, for pinning the magnetization of a FM layer in a certain direction. The thesis is organized in 8 chapters. First, an introduction is given in chapter 1. This is followed in chapter 2 by a review of the existing exchange bias theories from the discovery to the current models, with the emphasis on the domain state model. A brief description of some of the experimental techniques that were used is given in chapter 3. The central part of the thesis, chapter 4, addresses the domain state model for metallic FM/AFM bilayers, both from an experimental and Monte Carlo simulation point of view. Experimentally, we tested the domain state model for two metallic FM/AFM systems: NiFe/FeMn and CoFe/IrMn. The metallic AFMs FeMn and IrMn have low and intermediate magnetic anisotropy, respectively, polycrystalline structure, and Néel temperatures above room temperature. Lattice matched Cu dilution was used both for FeMn and IrMn. Our experimental results showed that the presence of Cu dilution throughout the volume of the AFM gives rise to a domain state and to an enhanced exchange bias field. This is in agreement with the initially proposed domain state model for high-anisotropy and epitaxial AFM CoO. The thermoremanent magnetization of the diluted AFM (without FM in contact) showed qualitatively similar dilution and temperature dependencies to those of the exchange bias field of the FM/AFM bilayers. These observations are also in agreement with the domain state model and indicate that the domain state magnetization is at the origin of exchange bias in the systems under consideration. The initially proposed domain state model was extended for describing the EB properties of metallic, polycrystalline AFMs with low and intermediate anisotropies, such as FeMn and IrMn. For this purpose Monte Carlo simulations were performed using a Heisenberg type AFM which accounts for the three-dimensional rotation of the magnetic moments. The granular structure of the AFM, the energy barriers upon reversal of the AFM moments and the thermal relaxation in the AFM were included in the model. The influence of various parameters, such as temperature, dilution, AFM grain size, etc. on exchange bias and thermoremanent magnetization was tested within the adapted domain state model. We found a good agreement between the experimental data and the Monte Carlo simulations which suggests that the adapted domain state model can be successfully applied to metallic, low and intermediate anisotropy AFMs. In chapter 5 the thermally activated reversal and the blocking temperature distribution in CoFe/IrMn bilayers are investigated. It is shown that the blocking temperature distribution measured at 5 K is mainly given by the contribution of the isolated uncompensated moments in the AFM. This is in agreement with our Monte Carlo simulations from chapter 4 where an average blocking temperature of about 10-2 K was estimated for the isolated uncompensated moments in the AFM IrMn. The influence of the AFM grain size on the blocking temperature distribution is also investigated. We found that larger crystalline FM/AFM grains give rise to a larger median blocking temperature and to a shifted blocking temperature distribution towards higher temperatures as compared to that of smaller FM/AFM grains. A novel study of the training effect and the temperature dependence of the exchange bias field and coercivity in epitaxial and polycrystalline NiFe/FeMn metallic thin films is presented in chapter 6 of this thesis. We show that the training effect depends on the crystalline structure and epitaxial quality of the AFM FeMn, being larger for the polycrystalline sample and decreasing in magnitude with increasing growth quality of the epitaxial bilayers. The vertical shift of the hysteresis loop (also referred to as the pinned AFM magnetization) is used to estimate the fraction of pinned uncompensated moments per AFM monolayer. We find a very good qualitative agreement between the decrease of the fraction of pinned uncompensated moments of an AFM monolayer and the decrease of the exchange bias field as a function of the hysteresis cycle number. Our estimated maximum values of the fraction of pinned uncompensated moments of an AFM FeMn monolayer (2-3%) are in a good agreement with those determined in literature for other AFMs. Moreover, the magnitude of the exchange bias field and of the coercive field depends strongly on the crystalline structure and epitaxial quality of the bilayers. In chapter 7 we discuss the exchange coupling between an amorphous FM (CoFeB) and a crystalline AFM (IrMn). We show that when the AFM IrMn is deposited on top of the FM CoFeB no exchange coupling appears. On insertion of a thin crystalline FM layer of NiFe between the amorphous CoFeB and the IrMn, exchange bias appears and it is dependent on the thickness of the NiFe layer. Moreover, an enhancement in the blocking temperature of the CoFeB/NiFe/IrMn trilayers is observed on increasing the thickness of the NiFe layer. X-Ray diffractometry measurements show that these effects are directly correlated with the (111) texture in the AFM phase of the IrMn layer, which develops progressively with increasing thickness of the NiFe layer. A thin nonmagnetic spacer layer of Cu or Ru at the NiFe/IrMn interface is found to reduce the exchange coupling and the coercive field of the exchange biased system. The exchange bias field vanishes for nonmagnetic spacers thicker than about 1 nm, indicating that exchange bias is a short-range exchange interaction. We also show and explain the decrease of blocking temperature with increasing thickness of the nonmagnetic spacer. The exchange coupled trilayer CoFeB/NiFe/IrMn is used for inducing an additional source of anisotropy in giant magnetoresistance sensors and magnetic tunnel junctions, by setting its anisotropy in an orthogonal direction with respect to that of the pinned FM layer. We show that the resistance of such a device depends linearly with the external field over a range of more than 200 Oe. This characteristic could be beneficial in different applications which require a linear resistance dependence on the external magnetic field. In chapter 8 the main conclusions of the thesis and some suggestions for further work are presented.
展开▼
