通过人工智能赋能的数字孪生技术解析压缩态碳纤维纸的微观结构复杂性

Abstract In the decarbonized society based on the renewable sources, the carbon fiber papers (CFPs) are regarded as key porous materials for the electrochemical energy conversion and storage devices. Searching the optimum microstructure of assembled carbon fiber paper under compression is one of the central challenges in this uprising technology. Herein, we present a tomography-based analytical approach to correlate CFP microstructures and transport parameters under the compressed state. For the sake of artificial intelligence, the prediction accuracy on the pore and solid structures is dramatically improved up to 98 % consistency when compared with the analytical solution, by identifying the true shape of cylindrical carbon fibers. The three-dimensional U-net algorithm was incorporated into the conventional X-ray computed tomography, to gain a complete separation of carbon fiber and binder. Subsequently, the origin of two different microstructures in the through-plane direction, i.e. transitional surface region and core region, are investigated as a function of compression ratio (CR). Finally, the structure-property relationship of CFP is thoroughly evaluated over a wide range of the paper thicknesses, PTFE contents and CRs. We demonstrate that the microstructural three-dimensionality, which is one decisive factor determining the transport and electrochemical properties in energy devices, can be further analysed by exploring the formation factors of solid and pore structures with increasing CRs. The insights gained from this work not only enhance the fundamental understanding of CFP microstructures but also pave the way for optimizing the design and operation of next-generation energy devices, promising a more efficient and sustainable energy landscape.