Comparison of spin Hall magnetoresistance temperature dependence in YIG/metal systems

The spin Hall magnetoresistance (SHMR) is a recently discovered effect that occurs in a normal metal (NM) that is in contact with a magnetic material. An insulating magnetic material is ideal as it prevents other magnetic effects by confining the electrons to the NM. To study this a new method for making yttrium iron garnet (YIG) has been developed using sputtering techniques so that devices are quicker and cheaper to produce than by existing liquid phase epitaxy (LPE). The YIG material magnetic properties were extensively characterised up to 280 nm thick. The magnetisation drops below the bulk value in thin samples believed to be caused by a contaminated layer at the substrate interface. As the SHMR is an interface effect the roughness of the films is important and was investigated by X ray reflectivity (XRR) and atomic force microscopy (AFM). The crystal structure was investigated using transmission electron microscopy (TEM). The SHMR is also significantly improved by a factor of 3 after treating YIG with piranha acid before metals are deposited on top.

For SHMR measurements platinum was used first and provided similar results to what is found in the literature. The angular dependence has been measured in a cryostat using four probe resistance measurements with a split pair magnet that allows the sample to be rotated in any direction relative to the applied field. This shows excellent agreement with the theory and was studied for the temperature range of 1.5 K - 300 K. This data has then been fitted using various models for the spin diffusion length and the conclusion reached is that a variation of the Elliot-Yafet mechanism (EY) is the most likely explanation for the observed temperature dependence. The same effect was found to occur in tungsten but with the important difference that it has a negative temperature coefficient of resistivity. This has been successfully fitted with the same spin relaxation models as the platinum. The tungsten is not believed to be in the beta phase, something which if achieved in further work could allow the size of the effect be improved by an order of magnitude up to several percent and so potentially allow the effect to have applications in a device.