Abstract
Current memristor technologies remain limited by instability, high operating voltage, and low switching ratio, primarily due to stochastic filament formation and defect migration. Here, a fundamentally different electrochemical mechanism is proposed through the development of a plating/stripping memristor (PSM) featuring stable, low-voltage, and bio-inspired conductance switching. Constructed with Zn/Cu electrodes and a deep eutectic gel electrolyte (DEGE), the PSM accurately emulates spike-rate-dependent plasticity and long-term synaptic dynamics. The DEGE matrix offers a corrosion-resistant, dendrite-free, and ionically homogeneous environment, facilitating gradual and programmable conductance evolution. Remarkably, the Zn/DEGE/Cu PSM exhibits switching behavior with a low-resistance state centered at 15.3 µV and dual high-resistance states at -10.0 mV and +11.1 mV, governed by electrochemical equilibrium, highlighting its sub-millivolt-level operation and energy-efficient switching characteristics. Furthermore, the Zn/DEGE/Cu PSMs are integrated into a reservoir computing framework using 4-bit pulse-encoded conductance states. When applied to pattern recognition tasks, the DEGE-based PSM system demonstrates a reliable classification accuracy of 89.3%, driven by device-derived temporal dynamics. Overall, this study establishes a new materials and mechanistic foundation for energy-efficient neuromorphic computing, bridging electrochemical reactions with biologically plausible information processing.