Electrospinning-Driven IGZO Nanofiber Phototransistors Bridging Adaptive Processing and Recognition Learning

en This link goes to a English section zh This link goes to a Chinese section The development of neuromorphic electronics with visual perception and adaptive capability is highly desirable for advancing artificial vision systems. Herein, we have demonstrated a dual-plasticity adaptive phototransistor based on IGZO nanofibers that exhibits the perception and dynamic adaptation behavior of rod and cone cells to varying light environments. Benefiting from the coexistence of oxygen vacancies and trap states in nanofibers, the separation and injection of photogenerated carriers are significantly improved, thereby enabling the light-intensity-dependent dynamic adaptability and the gate-modulated photosensitivity with a narrower adaptation timescale than the bio-systems (< 2 min). Moreover, the phototransistor array replicates both photopic and scotopic adaptation, and achieves adaptive contrast enhancement for the overexposed images. Finally, the system enables real-time neuromorphic encoding and recognition of digital signals under varying adaptive processes, and the pattern recognition accuracy is significantly improved from 10% to 95.8%. These results not only demonstrate a facile route for bridging visual sensing, adaptive processing, and neuromorphic computing in a single transistor device but also lay the groundwork for future applications in machine vision and next-generation neuromorphic sensory systems. 摘要 en This link goes to a English section zh This link goes to a Chinese section 发展具备视觉感知与自适应能力的神经形态电子器件对于推动人工视觉系统具有重要意义。本文报道了一种基于IGZO纳米纤维的双重可塑性自适应光电晶体管,该器件能够模拟视网膜中视杆细胞与视锥细胞对不同光照环境的感知与动态自适应行为。得益于纳米纤维中氧空位与陷阱态的共存,光生载流子的分离与注入效率得到显著提升,从而实现了光强依赖的动态自适应特性以及栅压可调的光敏响应,其自适应时间尺度较生物系统(< 2分钟)更短。此外,该光电晶体管阵列成功复现了明视觉与暗视觉自适应机制,并对过曝光图像实现了自适应对比度增强。最终,该系统能够在不同自适应过程中完成数字信号的实时神经形态编码与识别,将模式识别准确率从10%显著提升至95.8%。这些研究成果不仅为在单一晶体管器件中实现视觉感知、自适应处理与神经形态计算的融合提供了一条创新路径,也为机器视觉与下一代神经形态传感系统的应用奠定了坚实基础.