神经形态工程学
磷光
材料科学
系统间交叉
光电子学
纳米技术
光热治疗
人工神经网络
噪音(视频)
超短脉冲
计算机科学
有机发光二极管
紫外线
能量(信号处理)
作者
Zihan Ying,Yi Zhang,Jiahao Fan,Mengqi Li,Siqi Liu,Jun Zhang,Fangjian Xing,Cihui Liu,朱丹,Yingchun Cheng,Yunsong Di,Chaofeng Gao,Xiangxing Xu,Zhixing Gan
摘要
ABSTRACT Room‐temperature phosphorescence (RTP) materials hold immense potential for neuromorphic vision, yet achieving system‐level cognitive capabilities within a single material remains challenging. Here, we design a carbon dots (CDs)–epoxy composite wherein hydrogen‐bond‐mediated interfacial assembly between electron‐donating CDs surface groups and electron‐accepting epoxy moieties deliberately generates interfacial exciplexes. This engineered interface achieves a minimized singlet–triplet energy gap and enhanced spin–orbit coupling (SOC), enabling efficient intersystem crossing (ISC) into a long‐lived triplet reservoir that emits green phosphorescence (∼468 ms). We obtain multiple neuromorphic functions by exploiting the exciplex kinetics. Ultraviolet (UV)‐dose‐dependent triplet filling enables optical learning and noisy pattern capture. The cooperative effect in densely populated exciplex networks establishes an intensity‐gated threshold for spontaneous physical noise reduction at the sensory terminal. Alternating UV‐induced hydrogen‐bond strengthening and near infrared‐induced photothermal relaxation bidirectionally modulate exciplex populations, realizing all‐optical plasticity with high symmetry (AR = 0.33) that supports a fully connected neural network achieving >95% recognition accuracy on noisy digits. The millisecond‐lived triplet reservoir further transcribes dynamic motion trajectories into static 2D afterglow patterns, enabling contactless trajectory recognition and real‐time control without high‐speed cameras. This work provides a unified material strategy integrating sensing, memory, and computation within a single neuromorphic visual terminal.
科研通智能强力驱动
Strongly Powered by AbleSci AI