The fast growth of memory-intensive applications in machine learning, neuromorphic computing, and autonomous systems, combined with the slowing pace of traditional CMOS scaling, have created new demands for both embedded and discrete memories in electronic systems. Spintronic devices – which combine data storage in a nano-magnet with electrical readout using spin-dependent electronic transport phenomena – are prime contenders to address this need. They combine nonvolatile operation and nanosecond speed with high endurance required for embedded and random access memory applications. We will review recent progress and perspectives of spintronics focusing on two areas: (i) How to build spintronic memory devices with unprecedented energy efficiency, speed, and integration density, with an eye on applications in brain-inspired computing, (ii) How emerging device concepts in spintronics, which frequently emerge from a desire to build better computing elements, can be adapted to other types of devices that are important for IoT, with examples being spintronic oscillators, microwave detectors, and random number generators.
Pedram Khalili is Associate Professor of EECS at Northwestern University. He works on developing the computing systems of the future, starting from novel nano-scale devices/materials that enable systems with unprecedented performance and energy efficiency. He was an adjunct assistant professor at UCLA EE in 2013-2017, where he co-led the memory program in the NSF TANMS center, focusing on electric-field-controlled magnetic memory with unprecedented energy efficiency. In 2009-2014, he was the project manager of two DARPA multi-institution programs, focusing on the development of STT-MRAM and non-volatile logic. In 2012 he co-founded Inston, Inc., which pioneers voltage-controlled MRAM. He has published >90 journal publications with over 3,000 citations and an h-index of 29, and has 15 issued US patents.