A Self-Powered Multimodal Sen-Memory System

The development of multimodal systems with fused sensing-and-memory (sen–memory) functionalities is drawing growing attention for artificial intelligence applications. However, conventional systems typically record the information based on a coinput of sensory signal and the electrical signal. Self-powered multimodal sen–memory system has the potential to promote highly efficient neuromorphic applications and remains to be studied. Herein, we demonstrate the fabrication of pectin nanowire-based moisture generator and pectin film-based “quantized” memristor. The generator can serve as both a power supply and a humidity sensor. Electric energy harvested by the generator from ambient humidity can drive the memristor. Integrating the generator, the memristor, a capacitor, and a photoresistor yields self-powered system that can record the humidity and optical information. Benefit from the voltage-dependent multilevel quantized conductance (QC) of the memristor, the magnitude of humidity and optical stimulus can be distinguished. In addition, taking advantage of the multilevel QC switching behavior, image pattern recognition are successfully realized by constructing an artificial neuromorphic network simulator with a
${238}\times {242}$
memristor array. The present sen-memory system can be applied to emulate multimode-fused sensing and memory behavior of human visual memory for future efficient bio-realistic applications.