Neuromorphic Computing with Interfacial Dynamics in Electrolytic Memristive Synapse

Abstract

ABSTRACT Brain‐inspired computing demands devices that operate at biological voltages and embody biological principles. Existing liquid‐based memristors offer ion‐mediated computation yet exceed physiological levels and require rigid matrices. Here, we report a hydrogel‐confined electrolytic memristive synapse (EMS) that leverages reversible Ag/AgCl interfacial electrochemistry for synaptic modulation and reservoir‐state encoding at only 100 mV. The device exhibits chloride‐ion‐dependent memristive hysteresis and emulates fundamental synaptic learning rules. Such dynamics provide the short‐term memory and nonlinearity used in a simulated reservoir‐computing framework, achieving 95.05% accuracy in handwritten‐digit classification and exhibiting good performance under noise and mechanical deformation. The device conformability and the design flexibility of the electrochemical interface offer new opportunities toward bio‐integrated systems, promising for adaptive human‐machine computing architectures.