Ultra-Low-Power Reservoir Computing Based on Synthetic Antiferromagnetic Skyrmion Pairs
Zhizhong Zhang, Jinyi Sun, Zhenyi Zheng, Kelian Lin, Kun Zhang, Jinkai Wang, Youguang Zhang, Weisheng Zhao, Yue Zhang · IEEE Electron Device Letters · 2022
Reservoir computing (RC) is a promising method to realize the universal learning system. However, the complexity and power consumption limit its application. In this letter, we propose an ultra-low power RC device based on Synthetic antiferromagnetic skyrmion pairs (SAFSPs). By leveraging the Ruderman-Kittel-Kasuya- Yosida (RKKY) effect and spin-orbital torque, skyrmion pendulum motion can be realized. Furthermore, periodic oscillations can be achieved through magnetic anisotropy engineering. Based on this nonlinear motion of magnetic skyrmions, a RC system can be built. Our results show that the cognitive task of nonlinear time series can be carried out by a network of the proposed device and the power consumption can be reduced to 466nW.