Use of analog spintronics device in performing neuro-morphic computing functions

Since spintronics devices are capable of retaining digital information as their magnetization direction, development of nonvolatile memories, so-called magnetoresistive random access memories (MRAMs), to realize low-power integrated circuits with the von Neumann architecture has been one of the mainstream outlets of spintronics research pursued in the last several decades. Meanwhile, neuromorphic-computing hardware with non-von Neumann architecture has started to attract a great deal of attention in the field of microelectronics. Neuromorphic computing allows for completion of complex tasks at high speeds and at low power consumption levels that conventional von Neumann computers struggle with [1,2]. Recent researches point out that the spintronics devices also have the capable characteristics to model the human brain [3-5]. In this presentation, we describe a proof-of-concept demonstration of an associative memory operation like the human brain using a spintronics device [6]. For this purpose, we employ a recently-found spin-orbit torque (SOT) induced switching [7-9] device consisting of an antiferromagnet (AFM)/ ferromagnet (FM) stack structure [10-12], which shows an analogue-like resistance switching and thus serves as an artificial synapse in artificial neural networks.