折り畳み可能な電界紡糸CuAg@炭素ナノ繊維膜を用いたインテリジェントリアルタイム材料認識用高性能トライボエレクトリックナノジェネレータ

Electrospinning is a versatile technique for fabricating triboelectric materials. Although carbonization can significantly improve the electrical conductivity of fiber membranes, their application as triboelectric layers remains underexplored. In this work, a copper-silver (CuAg) alloy-doped carbon nanofibers (CuAg@CNFs) membrane was designed and employed as a triboelectric layer to enhance the accuracy and sensitivity of self-powered tactile sensors based on triboelectric nanogenerators (TENGs) for material identification. The CuAg@CNFs membrane, obtained by electrospinning and subsequent thermal conversion, exhibits a three-dimensional conductive network with in situ-generated alloy nanoparticles. Such a structural configuration promotes efficient charge migration across the interface, thereby enhancing electrical output. The fabricated CuAg@CNFs based-TENG delivers an output voltage of 4.3 V and a charge density of 40 nC m^-2, maintaining stable operation over repeated cycles. By correlating triboelectric signals with material surface characteristics, an intelligent sensing framework integrating a convolutional neural network (CNN) was constructed. The system achieved a recognition accuracy of 99.6% across ten materials and maintained high discriminative capability in complex environments. During real-time operation, the recognition accuracy for four types of material categories (aluminum, board, glass, and plastic) reached 100, 93, 99, and 97.5%, respectively. This study demonstrates a feasible strategy for constructing high-performance self-powered sensing systems through the synergistic combination of material design and machine learning algorithms.