Three-dimensional in-situ observation and cohesive zone modeling of tension-induced delamination of two-dimensional C/SiC composites via deep learning-based damage identification

Abstract

The initial defects in two-dimensional (2D) carbon fiber-reinforced silicon carbide matrix (C/SiC) composites directly influence their mechanical properties. Structural discrepancies between the layers resulting from the layer stacking relationship can lead to delamination and cracking in 2D C/SiC composites. In this study, in-situ scanning electron microscopy and in-situ computed tomography (CT) tensile tests were performed to investigate the interlayer adhesion and load transfer and their influence on the tensile behavior of 2D C/SiC composites. Deep learning-based image segmentation approach was used for quantitative analysis of damage, and the delamination mechanism was expounded. The results revealed that the cracks on the surface coating exhibited periodic cracking, and the interfacial sliding stress between the coating and the laminated preform C/SiC composite was 1.88 MPa. Combined results of in-situ CT tensile testing and deep learning, revealed the existence of at least two interlayer bonding states within the 2D C/SiC composite in a random phase state. One of them featured more interlayer pores and stronger adhesion, while the other exhibited smaller pores and weaker adhesion. During the tensile process, the tunnel crack underwent deflection and formed delamination cracks due to uneven interlayer bonding strength. The delamination cracks propagated until they merged with the main crack, eventually leading to the material fracture. Finally, based on the damage analysis of 2D C/SiC composite by in-situ CT, a simulation method for the cohesive fracture between layers was proposed, which indicated that weak interlayer bonding would trigger the tensile delamination phenomenon.