Information processing applications like big data, internet of things, deep learning and autonomous systems are fast becoming mainstream concepts and leading to ever increasing energy requirements. Spintronics –where information is carried by spin instead of charge– is a key enabling technology for energy-efficient nanoelectronics. It has already had enormous impact on information storage and sensor technologies, giving us low-power non-volatile memories and unprecedented sensing versatility. Magnetic RAM technologies (MRAM), an example of non-volatile memories, encode the memory state in the magnetisation of a ferromagnetic layer. Current-generation MRAM draw on the spin transfer torque (STT), a physical phenomenon in spintronics that achieves magnetisation switching using a spin-polarised current from a ferromagnet. However, obstacles to uptake and widespread implementation remain in the form of reliability, material degradation and write speed. These derive from material and architectural limitations. The solution to these obstacles may come via a device architecture recently proposed and demonstrated by the spintronics community based on the spin-orbit torque (SOT). In SOT-MRAM the magnetisation of the free layer is controlled by injecting an in-plane current in an adjacent metallic layer, typically tungsten or platinum, which induces spin torque via either the spin Hall effect or the Rashba-Edelstein effect. SOT-MEM used novel quantum materials, known as topological insulators, to achieve and validate the generation of large SOTs for magnetisation switching in future memory technologies. Parallel efforts were devoted to evaluate adequate intellectual protection and valorisation routes and subsequent steps on the path to market for MRAM based on novel materials, building a tentative roadmap for applications and scaling.
