Thesis presented June 15, 2016

Abstract:
aIn this thesis we study the diffusive transport of the charge, spin and heat in metallic structures involving ferromagnets. In particular, we focused on the part of the transport which polarization is not collinear to the surrounding magnetization.
For example, a spin-polarized current arriving on a magnetic layer with a magnetization pointing in another direction will have its transverse part (i.e. non-collinear to the magnetization) precess and be absorbed by the magnetization, over a distance of up to a few nanometers. We present a state-of-the-art collection of values for those two characteristic lengths, of precession and transverse absorption. We also show that this behavior as a tremendous impact over the dynamics, notably that of "long" magnetic domain walls (over ten nanometers).
We also study the spin-transfer torque in those magnetic structures, and focus on two major aspects. First the amplitude of the torque, to know if it is strong enough to start one of the known dynamics: magnetic switching or steady-state precession. Second, the dependence of the torque with the relative angle between the magnetizations: in some cases, a non-collinear configuration may be stabilized. Two driving forces have been considered, a voltage bias or a temperature difference (by including spin-dependent thermoelectric effects).
This whole study is performed within the framework of our theory, the Continuous Random Matrix Theory, that we present in its entirety, from its origin with the scattering theory, to the diffusion differential equations, one of the main results of this thesis.
We also present the numerical tool we developed, based on this theory, which we used to perform all of our simulations. This tool allows for the evaluation of the diffusive transport in three-dimensional metallic structures, using (mostly) readily available material parameters.

Keywords:
Unified theory of spin, heat and charge transport

On-line thesis.