Bilayer-skyrmion-based design of neuron and synapse for spiking neural network

Debasis Das,Yunuo Cen,Jianze Wang,Xuanyao Fong

DOI: https://doi.org/10.1103/PhysRevApplied.19.024063

2023-02-03

Abstract:Magnetic skyrmion technology is promising for the next-generation spintronics-based memory and neuromorphic computing due to their small size, non-volatility and low depinning current density. However, the Magnus force originating from the skyrmion Hall effect causes the skyrmion to move along a curved trajectory, which may lead to the annihilation of the skyrmion in a nanotrack during current-induced skyrmion motion. Consequently, circuits utilizing skyrmionic motion need to be designed to limit the impact of the skyrmion Hall effect. In this work, we propose a design of an artificial neuron, and a synapse using the bilayer device consisting of two antiferromagnetically exchange coupled ferromagnetic layers, which achieves robustness against the skyrmion Hall effect by nullifying the Magnus force. Using micromagnetic simulations, we show that the bilayer device can work as an artificial neuron and also as a synapse by modifying its uniaxial anisotropy. We also demonstrate that our proposed skyrmionic synapse has an intrinsic property of perfectly linear and symmetric weight update, which is highly desirable for the synapse operation. A spiking neural network implemented using our proposed synapse and neuron was simulated and showed to achieve 96.23\% accuracy in performing classification on the MNIST handwritten digit dataset.

Mesoscale and Nanoscale Physics

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to use magnetic skyrmion technology to design efficient and stable neuromorphic computing systems in the field of spintronics, especially for neurons and synapses in Spiking Neural Network (SNN). Specifically, the researchers proposed a double - layer - structure - based scheme to overcome the challenges brought by the skyrmion Hall effect.

### Problem Background

Magnetic skyrmions have great potential in next - generation spintronics memories and neuromorphic computing due to their small size, non - volatility, and low depinning current density. However, due to the Magnus force caused by the skyrmion Hall effect, skyrmions will move along a curved trajectory when moving in a nano - track, which may lead to skyrmion annihilation and thus affect the stability of the circuit.

### Research Objectives

To solve the above problems, this paper proposes a design based on a double - layer device, which consists of two antiferromagnetically exchange - coupled ferromagnetic layers. Through this design, the Magnus force can be effectively cancelled out, enabling skyrmions to move in a straight line, thereby improving the stability and operation speed of the system. In addition, the research also shows how to achieve the functions of artificial neurons and synapses by modifying the uniaxial anisotropy and proves that the proposed skyrmion synapses have perfect linear and symmetric weight - update characteristics.

### Main Contributions

1. **Cancelling the Magnus Force**: By using a double - layer structure, the mutual cancellation of the Magnus force is achieved, ensuring that skyrmions move in a straight line.
2. **Artificial Neuron and Synapse Design**: It shows how to use a double - layer device to achieve the functions of artificial neurons and synapses, especially in the presence of a uniaxial anisotropy gradient, simulating the Leaky Integrate - and - Fire (LIF) model.
3. **Linear and Symmetric Weight Update**: It is proved that the proposed skyrmion synapses have ideal linear and symmetric weight - update characteristics, which are very important for synaptic operations.
4. **Performance Verification**: Through simulation, it is shown that the SNN constructed based on the proposed neurons and synapses achieves an accuracy rate of 96.23% in the MNIST handwritten digit classification task.

### Summary

Through the innovative double - layer structure design, this paper successfully solves the influence of the skyrmion Hall effect on neuromorphic computing systems, providing new ideas and technical means for future high - efficiency and low - power - consumption neuromorphic computing.