An investigation of spin-valves and related films by TEM

The work presented in this thesis is a study of the reversal mechanisms of the magnetic layers within spin-valve materials and related films. Spin-valves display the phenomenon of giant magnetoresistance (GMR) and are now being utilised as magnetoresistive read heads in commercial applications.

A spin-valve typically consists of two ferromagnetic layers separated by a thin spacer layer (eg Cu in the range 2-10nm). One of the ferromagnetic layers is "pinned" or fixed in direction by exchange coupling to an antiferromagentic layer such as FeMn or IrMn. This shifts the hysteresis loop by a few hundred Oersted (Oe), and so only the other "free" ferromagnetic layer can reverse under the influence of the comparatively small magnetic field from a passing tape or disk. This reversal corresponds to a change between a parallel ‘low’ resistance state to an antiparallel ‘high’ resistance state under the application of a magnetic field of » 10-15Oe.

The reversal mechanisms that take place in the "free" and "pinned" magnetic layers in a range of spin-valve materials have been studied using Lorentz modes of transmission electron microscopy. These results form the bulk of this thesis. The effect of Molybdenum impurity on the magnetoresistance of Ni80Fe20 is considered as a secondary topic, and only chapter 6 is given over to these results.

The first chapter introduces the basic concepts of ferromagnetism, magnetoresistance, and its application in magnetic storage technology. Particular emphasis is given to the various energy contributions that are present in thin magnetic films. This leads to the concept of domains and domain walls. Anisotropic magnetoresistance (AMR) and giant magnetoresistance (GMR) are introduced. These phenomena have enabled the production of many thin film sensors for applications including magnetic storage technology. Spin-valves are then discussed, reference being given to their application as a sensor in magnetic read-head assemblies. The possible commercial benefits of using GMR based devices for magnetic storage applications are highlighted.

Transmission electron microscopy was the primary tool used to investigate the materials discussed in this thesis. Thus chapter 2 is devoted to discussion of the instrumentation and techniques employed. An overview of the important parts of a TEM is introduced, including the electron gun and microscope column. The aberrations which limit the resolution of the microscope are mentioned before the techniques used to image structural properties are presented. Particular attention is then paid to the Fresnel, Foucault and Low angle diffraction (LAD) imaging modes of Lorentz electron microscopy which are used to investigate the magnetic structures in thin films. Finally, the methods used to apply magnetic fields in-situ are discussed.

Chapter 3 begins with a review of spin-valves detailing parameters such as the magnetic layer configuration, anisotropy arrangement, free layer reversal and deposition technique used for the samples investigated in this thesis.

Structural properties such as the average grain size of the spin-valves are investigated by bright and dark field imaging. Diffraction studies allow some compositional data and the level of texture to be evaluated.

The main body of results in this chapter concentrates on the magnetisaton reversal mechanism of the free layer in a range of FeMn-biased spin-valves with parallel anisotropy. The thickness of the copper spacer layer (and hence the strength of the interlayer coupling strength), is varied in the range 2-10nm, as is the angle of applied field, q , with respect to the biasing direction. Fresnel imaging and LAD studies reveal there to be 3 modes of reversal that are possible in the free layer depending on the interlayer coupling strength and field orientation. Two of the observed modes involve a combination of magnetisation rotation and domain assisted processes while the third involves coherent rotation of magnetisation alone. The boundary between the modes of reversal was shown to be indistinct, and involved a free layer reversal that proceeded by an increasing amount of magnetisation rotation as q was decreased. This was accompanied by the formation of an increasing density of low angle, low mobility domain walls.

The observed modes of reversal are presented on a phase diagram and shown to be in many ways consistent with a coherent rotation model by Labrune. The main discrepancy between the model and observed modes of reversal is that domains are forbidden in the model. The chapter concludes with a brief study of a spin valve in a crossed anisotropy arrangement.

Magnetic modelling of the free layer reversal of FeMn-biased spin-valves is presented in Chapter 4. This uses the modified Stoner-Wohlfarth coherent rotation model of Labrune. The model makes 3 assumptions, namely that i) the presence of domains and domain processes are forbidden; ii) there can be no twist of the magnetisation vector within a ferromagnetic layer; iii) the biased layer magnetisation remains fixed.

Using this model, the spin-valve free layer energy is modelled as a function of magnetisation for several of the samples studied experimentally in chapter 3. It is found that for all cases, except that for which coherent rotation of magnetisation alone is observed, an energy barrier is overcome at the switching field.

The form of the energy curves allows for a quantitative description of the observed domain processes to be made. For reversals that are seen to take place by a low number of domain walls rapidly sweeping through the free layer, the energy curves reveals that a large energy barrier with a deep minimum is overcome at the switching field. On the other hand, for cases where the magnetisation reversal proceeds by a higher density of low angle, low mobility domain walls, the energy curves display small energy barriers and shallow minima. For the case of coherent rotation of magnetisation alone, a single energy minimum is present throughout the reversal process.

For successful operation of a spin-valve, one of the magnetic layers must remain fixed while the other reverses under the influence of an external magnetic field. One way of achieving this is by exchange biasing one of the ferromagnetic layers with an antiferromagnetic pinning layer such as FeMn or IrMn. Chapter 5 investigates both the structural and magnetic properties of such an arrangement, emphasizing the influence of temperature on the exchange biasing mechanism. This is of interest because spin-valves incorporated into commercial devices often have to operate at elevated temperatures.

Reversals of the pinned ferromagnetic layer in the range -150° C to 300° C are conducted on samples of Ni80Fe20 biased by Fe50Mn50 and Ir20Mn80. Initially, at room temperature (where good alignment between the direction of applied field and the biasing direction is achieved), the reversal is observed to take place by complex domain assisted process and involved large numbers of 360° wall structures which are stable to high valves of applied field strength. The reversal is markedly different from that observed for free layer spin-valve reversals. Exchange biasing densities of 0.098mJ/m2 and 0.115mJ/m2 are obtained for FeMn and IrMn at room temperature respectively. A linear decrease in the strength of the exchange biasing occurs as the temperature is increased, and the amount of magnetisation rotation gradually increases as the temperature is raised. This is partly due to a slight misalignment between the biasing direction and applied field direction in the variable temperature rods, but is also due to the decreasing strength of the exchange biasing at elevated temperatures.

When the temperature is increased above the blocking temperature of the antiferromagnetic pinning layers, the reversal mechanism takes place at low applied field strengths, and an increase in the magnetisation ripple followed by a single domain wall sweeping rapidly through the ferromagnetic layer is observed. This corresponds to the exchange biasing effect disappearing as all magnetic order is lost in the antiferromagnetic layer above its blocking temperature (» 140-150° C for FeMn and » 285° -300° for IrMn. The pinned layer then behaves like an isolated ferromagnetic layer. After cooling back to room temperature in the presence of a magnetic field along the biasing direction back to room temperature, the exchange biasing effect returns. However its strength is diminished when compared to the initial room temperature measurement.

The effect of Molybdenum impurity on Ni80Fe20 is investigated in chapter 6. Magnetic heads exploiting the AMR effect which were produced in Philips manufacturing displayed a reduced AMR ratio than was achievable in research. This was thought to be due to the diffusion of Mo into the permalloy during the manufacturing process. Annealing experiments carried out on samples of Ni80Fe20 with a Mo layer deposited on top are carried out. It is found that for annealing temperatures £ 275° C that a slight improvement in the AMR could be achieved. This is thought to be due to grain growth which reduced the sheet resistance of the samples. For annealing at 300° C and above a decrease in the AMR ratio is observed. Rutherford Backscattering Spectrometry (RBS) revealed that Mo diffusion into the Ni80Fe20 layer is taking place. The amount of diffusion was low (» 1%), and did not increase with temperature. It is likely that the Mo which diffused into the Ni80Fe20 did so by diffusing down the grain boundaries as opposed to an interdiffusion process between the two layers.