A PZT/SiC floating gate transistor for high-accuracy artificial neuromuscular synapse

Abstract The neuromuscular junction (NMJ) is vital for transmitting nerve impulses to muscles. Artificial NMJs, aiming to replicate biological synapses, are promising for biomimetic robotics, prosthetics, and brain-inspired computing. However, current technologies struggle with plasticity, output power, and amplification. This paper presents an innovative artificial NMJ using a silicon carbide (SiC) combined with Pb(Zr, Ti)O 3 (PZT) ferroelectric field-effect transistor. By combining a ferroelectric gate with a SiC transistor, we achieved extensive non-volatile conductance modulation with current dynamic range of 0.1–10 μA, mimicking long-term synaptic plasticity. The PZT/SiC transistor offers a higher threshold voltage and reduced operational current, enhancing energy efficiency. Its synaptic plasticity allows for adjustable strength through five consecutive cycles testing and stimulation. The PZT/SiC application in convolutional image processing shows its high accuracy exceeded 95% only via 9 epochs, demonstrating its potential, paving the way for efficient, scalable artificial NMJs in bio-hybrid systems and neuromorphic robots.