Curvature-induced integrate-and-fire and burst firing in an antiferromagnetic skyrmion neuron

Antiferromagnetic skyrmions, owing to their zero skyrmion Hall effect, high stability, and low driving current, are promising information carriers for neuromorphic devices. Here, we propose an antiferromagnetic skyrmion artificial neuron based on a U-shaped curved nanotrack that emulates both integrate-and-fire and burst-firing neuronal functions through curvature-regulated skyrmion dynamics. The curvature-induced energy barrier traps antiferromagnetic skyrmions near the cutoff line, while the accumulating inter-skyrmion repulsive interactions collectively overcome this barrier, enabling a biological-like integrate-and-fire behavior. Further tuning of the driving current modulates the interplay between the curvature-induced barrier and inter-skyrmion repulsion, which allows multiple skyrmions to successively cross the cutoff line and gives rise to burst firing. By tuning the material and geometric parameters, both firing modes and threshold currents can be effectively regulated. Moreover, burst firing exhibits lower energy consumption per spike, while burst-based encoding achieves lower bit error rates and enhanced noise robustness. These findings establish antiferromagnetic skyrmion neurons as promising building blocks for energy-efficient and reliable neuromorphic computing.