Investigation of magnetostrictive Fe1−x Gax bilayer films and devices

Magnetic memory technology, especially hard disk drives are the leading technology for storing data. Increasing demand for improved storage density, along with faster processing times and low power consumption, have led to the invention and exploration of more sophisticated technologies based on magnetism. In these state of the art technologies, the primary aim is to manipulate the magnetisation state of the material in order to store information. In the next generation of magnetic memory technologies, the manipulation of the magnetisation state by an external magnetic field has been replaced with electric current or electric field. More recently, the use of strain mediated magnetoelastic coupling to change the magnetisation has attracted a lot of interest for development of energy efficient logical processing and information storage devices. One method to demonstrate this is to use a hybrid piezoelectric/ferromagnetic device. A voltage across the piezoelectric transducer induces a mechanical strain in the ferromagnetic layer and results in the manipulation of the magnetic anisotropy via the inverse magnetostriction effect. In this thesis, hybrid structures of a piezoelectric transducer and a magnetostrictive ferromagnet alloy of Fe 1−x Ga x have been used to investigate the strain manipulation and control of magnetisation. The Fe rich Fe-Ga alloy has demonstrated enhanced values of magnetostriction and has been shown to be very magnetically sensitive to strain both in bulk crystals and in thin film cases. Due to a very high magnetostriction value of (3/2) λ100 = 395 ppm, and no rare earth constituents, the material is a competitive candidate for strain mediated magnetic storage devices. The investigations described in the thesis are on 5 nm bilayer films of Fe 1−x Ga x deposited on GaAs (001) substrates by the magnetron sputter deposition technique. The ferromagnetic layers were separated by either a Cu or Al spacer layer of thickness 5 nm or 10 nm. The grown ferromagnetic layers had different Ga concentration so that they demonstrate different magnetostriction value and the magnetisation reversal process in each layer will be unique and independent. SQUID magnetometry along with ferromagnetic resonance experiments and mathematical modelling of minimising the free magnetic energy, revealed that there is a strong cubic magnetocrystalline anisotropy in the individual layers which was approximately equal for all the samples. The uniaxial anisotropy varied in each of the grown samples due to variation in the interface bonds between the substrate and the metallic stack. By modelling each of the layers to be independent, and solving at the switching field regions, the domain wall depinning energies for each layers have also been estimated. It is revealed that the domain wall depinning energies for the layers grown on the substrate is weaker than the layer grown on the metallic stacks. Ferromagnetic resonance experiments along with mathematical modelling were also used to investigate the dynamic properties of these bilayer films. The role of magnetic anisotropies and spacer type and thickness on the magnetisation precession in terms of the resonance frequency, Gilbert damping and linewidth have been investigated. A narrow linewidth of 3.8 mT and 4.7 mT for the top and bottom layers with a low Gilbert damping value of approximately 0.015 and 0.019 have been obtained which makes these films a competitive candidate for applications of microwave spintronic devices. An investigation of the effects of strain on the magnetisation reversal is described in chapter 4, by employing magnetotransport measurements on processed Hall bar devices mounted on piezoelectric transducers. The measured transport data containing contributions from the anisotropic magnetoresistance and giant magnetoresistance effects arising from distinct magnetisation reversal processes of each layer which were independent for each layer and dependent on the voltage induced strain. This strain-mediated modification of the measured resistances was different for all the samples. The induced strain changed the switching fields of the individual layers and was found to be higher for the 5 nm Al spacer samples than the 5 nm Cu spacer samples. However, the 5 nm Cu sample demonstrated a higher giant magnetoresistance contribution to the measured longitudinal resistance. Finally, the working parameters for a multi-level memory cell operated by voltage-induced strain and based on the layers studied in this thesis have been determined. The conceptualised device is an attractive candidate for high density magnetic information storage. The extension to more than one layer would increase the possible storage density by utilising the third spatial dimension to stack storage elements