Practical Pathways to Efficient MRAM: Spin-Orbit Torques, Low-Damping, and Anomalous Hall Conductivity in Polycrystalline Materials

Abstract

There are multiple pathways forward to next generation magnetic random access memory. In this thesis we explore two simple solutions with industry implementation in mind. The first is a low-damping (α< 5×10⁻³) ferromagnetic single-layer with modest anti-damping spin-orbit torque (SOT), $θ_{DL} ≈ 0.05. Here, we investigate an alternative approach to the traditional heavy metal/ferromagnet bilayer to produce SOTs, which suffers from high-damping that is detrimental to energy-efficiency. Instead, of breaking inversion symmetry at the interface we continually break symmetry along the thickness axis by creating an intentional compositional gradient that is purely ferromagnetic and maintains low damping. Crucially, we find that a compositional gradient is not necessary to achieve large damping-like SOTs, instead finding direct evidence from grazing-incidence x-ray diffraction for a strain gradient.

The next pathway investigated is an easy-to-grow, polycrystalline alternative to non-collinear antiferromagnets which require high temperature growth (>400°C). We find that sputter-grown γ-FeMn with no post-annealing, has a small non-zero net magnetization (≈(0.02-0.07)μB/atom) and perpendicular magnetic anisotropy only slightly larger than those found in non-collinear antiferromagnets like Mn₃Sn while still exhibiting a large anomalous Hall conductivity of 14 S/cm at room temperature. We show that these unique magnetic and transport properties are the result of pinning at the grain boundaries which can be tuned to enhance the anomalous Hall conductivity.