We theoretically investigate condensed matter systems where the spin and magnetic moment of electrons can be used to process or store information. This field of study, known as spintronics, ushered in a revolution in magnetic storage capacity by vastly improving the sensitivity of hard disk read heads. Spintronic devices now promise a wide range of applications, from storing binary information in next-generation magnetic memories to mimicking properties of neurons and synapses for artificial intelligence applications.
We are currently identifying and characterizing the physical mechanisms underlying spin-orbit torque, which is a potentially low-power way to write magnetic memories. Spin-orbit torques occur when electrons carrying current take angular momentum from atomic nuclei and transfer that angular momentum to electrons carrying magnetic order. In addition to investigating spin-orbit torque in “traditional” materials like heavy metals and transition metal ferromagnets, we are exploring how “quantum” materials like topological insulators and low-dimensional materials can enhance spin-orbit torque efficiency and create new functionality. Finally, we are applying what we learn about these systems to design and simulate neural networks driven by spin-orbit torque, which could operate at lower power than traditional, fully CMOS-based neural networks.