Reconfigurable Photoelectric Coaxial Fiber-Based Memristors for Neuromorphic Computing

作者：Keying Li, Jieru Song, Le Chen, Y Q He, Jialin Meng, Tianyu Wang, Hao Zhu, Qingqing Sun, Peining Chen, David Wei Zhang, Lin Chen · 发表于：ACS Nano · 年份：2026 · DOI：10.1021/acsnano.6c00539 · 被引用次数：1 · 研究领域：Advanced Memory and Neural Computing、Neural Networks and Reservoir Computing、Ferroelectric and Negative Capacitance Devices

Bioinspired optoelectronic neuromorphic devices have overcome the limitations of von Neumann architectures. However, most existing devices used in neuromorphic applications are unsuitable for highly flexible, wearable applications. To address this constraint, this work successfully demonstrated a Pt/TiO x /NiO x /Au memristor by depositing thin films on coaxial fibers and patterning via electron beam evaporation. Notably, this reconfigurable device can switch between a synapse and a neuron under a cross-modal signal without compromising its internal structure. Under electrical signals, the device emulates the integrate-and-fire function of volatile neurons. Under optical signals, it exhibits nonvolatile synaptic functionalities, including excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), the transition from short-term memory (STM) to long-term memory (LTM), and learning-experience behaviors. Using the short-term memory with decaying characteristics under optical pulse stimulation, fashion image recognition from the MNIST database and spoken-digit recognition from the NIST TI46 database were achieved through reservoir computing (RC) with recognition rates of 93.0% and 98.6%, respectively. Meanwhile, the device exhibits stable responses to optical and electrical inputs under different bending radii and bending cycles, indicating its robust stability and high flexibility under mechanical deformations. This work demonstrates significant potential for fiber-...