Metallopolymeric Memristor Based Artificial Optoelectronic Synapse for Neuromorphic Computing

Abstract

Mimicking the human brain to achieve neuromorphic computing holds promise in the field of artificial intelligence (AI). Optoelectronic synapses are regarded as the crucial foundation stone in neuromorphic computing due to their capability to intelligently process optoelectronic input signals. Herein, two donor–acceptor (D–A)-type metallopolymers, P-Cu and P-Zn, containing porphyrin moieties are designed and synthesized, which are utilized as a resistive switching layer for preparation of memristors. The resulting memristors exhibit significantly enhanced electrical characteristics, displaying a high ON/OFF ratio, a low threshold voltage (Vth), and superior cycle-to-cycle reproducibility. This enhancement is attributed to the formation and dissociation of charge transfer (CT) states induced by inserted metal ions. Importantly, the P-Cu-based memristor demonstrates the capability to co-modulate optoelectronic signals, effectively emulating versatile synaptic functions of the nervous system. These functions include excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), short-term plasticity (STP), long-term plasticity (LTP), transition from short-term memory (STM) to long-term memory (LTM), and learning-experience behavior. Moreover, multiple Boolean logical functions were successfully implemented using the paired stimuli of electrical pulses. The neuromorphic computing function was also proven through pattern recognition, achieving a recognition rate of up to 86.08% for handwritten digits. This study offers a potent approach for developing multifunctional artificial synaptic devices and artificial neural network platforms and opens up the innovative application of metallopolymers in the fields of optoelectronics and AI.