Spin–Orbit Torque‐Induced Domain Nucleation for Neuromorphic Computing

• Published in: Advanced materials (Weinheim) Vol. 33; no. 36; pp. e2103672 - n/a
• Main Authors: Zhou, Jing, Zhao, Tieyang, Shu, Xinyu, Liu, Liang, Lin, Weinan, Chen, Shaohai, Shi, Shu, Yan, Xiaobing, Liu, Xiaogang, Chen, Jingsheng
• Format: Journal Article
• Language: English
• Published: Germany Wiley Subscription Services, Inc 01.09.2021
• Subjects: Algorithms; Artificial intelligence; Artificial neural networks; Cobalt - chemistry; Cobalt base alloys; Cobalt compounds; Cognitive tasks; Computer Simulation; Computer Systems; Copper - chemistry; crystal symmetry; Deep Learning; deep learning network; domain nucleation; Hardware; Intermetallic compounds; Magnetic domains; Magnetization reversal; Materials science; Neural networks; Neural Networks, Computer; Neuromorphic computing; Neurons; Nucleation; Platinum - chemistry; Platinum compounds; spin–orbit torques; Torque; crystal symmetry; spin-orbit torques; domain nucleation; deep learning network; neuromorphic computing
• ISSN: 0935-9648; 1521-4095; 1521-4095
• DOI: 10.1002/adma.202103672

Neuromorphic computing has become an increasingly popular approach for artificial intelligence because it can perform cognitive tasks more efficiently than conventional computers. However, it remains challenging to develop dedicated hardware for artificial neural networks. Here, a simple bilayer spintronic device for hardware implementation of neuromorphic computing is demonstrated. In L11‐CuPt/CoPt bilayer, current‐inducted field‐free magnetization switching by symmetry‐dependent spin–orbit torques shows a unique domain nucleation‐dominated magnetization reversal, which is not accessible in conventional bilayers. Gradual domain nucleation creates multiple intermediate magnetization states which form the basis of a sigmoidal neuron. Using the L11‐CuPt/CoPt bilayer as a sigmoidal neuron, the training of a deep learning network to recognize written digits, with a high recognition rate (87.5%) comparable to simulation (87.8%) is further demonstrated. This work offers a new scheme of implementing artificial neural networks by magnetic domain nucleation. The dominant mode of field‐free magnetization switching in L11‐CuPt/CoPt bilayer is demonstrated to be domain nucleation. The stable intermediate states from domain nucleation grant excellent memristive plasticity, which allows the spintronic device to be used as an artificial sigmoidal neuron. A proof‐of‐concept live training is performed to recognize written digits, with a high recognition rate comparable to simulation.