Polymer Optoelectronic Synapse with Tunable Negative Photoconductance Memory for Sequential Signal Processing

Abstract

With the development of artificial intelligence and neuromorphic computing, bioinspired computational models have gained widespread attention. This paper presents a light synapse device based on polyvinylidene fluoride (PVDF), which integrates the negative photoconductivity (NPC) effect and analog switching memory (ASM) effect within the same pixel. The NPC effect of this memristor enables high-performance short/long-term synaptic plasticity that can be modularly adjusted through optical pulse parameters. Under 405 nm laser illumination, the light synapse device exhibits a stable negative light response and achieves short-term depression (STD) and long-term depression (LTD). Furthermore, the device demonstrates excellent performance in simulating paired-pulse facilitation (PPF), pulse count dependence, pulse width dependence, and pulse height dependence. Additionally, we constructed a neuromorphic visual system with a memory computing architecture, which achieved an accuracy of 96.50% in handwritten digit pattern recognition, showcasing its immense potential in artificial intelligence and neuromorphic computing. This research provides an approach for developing efficient and flexible neuromorphic computing devices and lays the foundation for the design of future intelligent visual systems.