Piezo-strain induced nonvolatile control of magnetic skyrmion nonlinear dynamics for artificial synapse device applications

Abstract

High-performance artificial synaptic devices that emulate the functions of biological synapses are crucial for advancing energy-efficient brain-inspired computing systems. Current studies predominantly focus on memristive devices, which achieve synaptic functions through nonvolatile electric current-assisted carrier modulation. However, these methods often suffer from excessive energy consumption. Here, a type of low-energy-consumption artificial synapse based on strain-mediated electric-field control of magnetic skyrmion's radius is demonstrated, where the energy consumption is 10 fJ per state and the non-volatility is achieved by local ferroelectric domain switching under bipolar electric fields. The proposed skyrmion-based synaptic device can replicate essential synaptic behaviors, including long-term potentiation (LTP), long-term depression (LTD), paired-pulse facilitation, paired-pulse depression, and spiking-time-dependent plasticity, aligning it closely with the biological synaptic system. The synaptic weight change and non-linearity of the artificial synapse are emulated by modulating the magnetic skyrmion's radius through precisely engineering the applied electric-field pulses. Simulation using the Modified National Institute of Standards and Technology database reveals that the pattern recognition rate decreases exponentially with increasing LTP/LTD non-linearity, quantifying the effect of the LTP/LTD non-linearity on the pattern recognition rate. This work underscores the potential of strain-mediated electric-field control of single skyrmion's radius as a groundbreaking approach for developing high density and low-energy consumption artificial synaptic devices.