Search

• Binary synapse based on the deterministic switching of the magnetization states by encoding high and low resistance states. • Sigmoid neuron based on the stochastic switching probability of the magnetization in the free layer. • All-spin BWN based on the multi-functional nanoscale SOT devices achieves ∼81.73 % recognition accuracy on the CIFAR-10 dataset. Spintronic devices driven by spin–orbit torque (SOT) have a lot of potential for future neuromorphic computing hardware platforms. Nevertheless, problems such as multistate loss, weight time-dependent variability, and output discontinuity ensue when the device size is reduced to the nanoscale, which are contradictory to the requirements of traditional high-precision neural networks. In this study, we propose a high-precision all-spin neural network based on magnetization switching of a simple nanoscale multi-functional SOT device. The all-spin neural network is implemented by the Binary Weight Network, in which the binary synapse is encoded by the two states formed by the device's deterministic switching and the practical neuron is realized by its stochastic switching probability. Furthermore, we use a difference derivation training algorithm for the general-purpose network in a novel way to be more compatible with the discontinuous neuron output. Using this strategy, our network can reach ∼81.73 % recognition accuracy on the universal CIFAR-10 dataset, paving the way for future practical hardware implementation of nanoscale SOT devices in high-precision compact neuromorphic computing. [ABSTRACT FROM AUTHOR]

• The Ta/GdFeCo/Ta heterostructures are utilized to achieve a spin–orbit torque-driven multistate magnetization switching. • The switching ratio is up to 47.7% and the β DL and β FL are estimated as 85.64 Oe/(107 A/cm−2) and 35.05 Oe/(107 A/cm−2), respectively. • The findings establish that a stable four-resistor state exists, proofing multilevel memory behaviors and synaptic plasticity under current drive. • Neuromorphic computing built from this SOT device can perform unsupervised handwritten digit recognition with an accuracy of 92.25%. Neuromorphic computation of ferrimagnetic materials as artificial synapses is essential for the development of ultra-low power spintronic devices. Here, Ta/GdFeCo/Ta heterostructures are utilized to achieve a spin–orbit torque-driven multistate magnetization switching, complying with the requirements of neuromorphic computation. For Ta/GdFeCo/Ta device, the switching ratio is up to 47.7 % when the critical current density value is 1.26 × 107 A/cm2. Moreover, the β DL and β FL are estimated as 85.64 Oe/(107 A/cm−2) and 35.05 Oe/(107 A/cm−2), respectively. Subsequently, the findings establish that a stable four-resistor state exists, proofing multilevel memory behaviors and synaptic plasticity under current drive. It is further demonstrated that neuromorphic computing built from this SOT device can perform unsupervised handwritten digit recognition with an accuracy of 92.25 %. This research highlights the potential of SOT-driven multi-energy stabilized residual states for neuromorphic computing and multi-energy storage in the future. [ABSTRACT FROM AUTHOR]