Carbon fiber reinforced plastics (CFRPs) are innovative structural materials because of their tunable mechanical properties, and excellent specific strength and rigidity owing to their fiber orientations. However, it is difficult to assess the failure progress because of the inhomogeneity of the material, which causes complicated fracture behaviors, such as fiber breakage and delamination1. In particular, although fatigue failure accounts for 80% of the causes of mechanical failures, the fatigue crack initiation has not been clarified for CFRP structures. Thus, fatigue tests must be conducted individually with different fiber orientations and applied stress levels, despite the huge costs of the experiments. Several failure assessments based on phenomenological approaches have been proposed2–5. However, they have not been specifically related to the mechanical properties of the carbon fiber, matrix resin, and their interface. Therefore, a general mechanism of the initiation and propagation of the fatigue crack in CFRP structures should be clarified to develop a versatile evaluation method.
One key aspect of fatigue failure is the transverse crack generated in the 90° layers, where carbon fibers do not directly affect the strength in the loading direction. The generation of the transverse crack is considered to be the first process of fatigue failure because it gradually increases during the early stage of the loading cycles6–9. Although transverse cracks do not immediately reduce the rigidity and strength of CFRP laminates, they eventually cause delamination along the interface between layers with different fiber orientations. Hosoi et al. proposed an estimation method for the remaining fatigue life based on the accumulation of transverse cracks in 90° layers10–12. However, the prediction of the initiation of a transverse crack remains a challenge because it strongly depends on the stacking sequence of the laminates.
The origin of transverse crack has been focused on the interfacial debonding between the carbon fiber and matrix resin13, as shown in Fig. 1. The scanning electron microscope observation at the site of the transverse crack implies that it propagates along the interface of the carbon fibers, and the remaining interfacial debonding is confirmed near the transverse crack10. However, it is extremely difficult to detect the nanoscale opening gap of the interfacial debonding among the innumerable carbon fibers in the CFRP laminate before propagating to be a transverse crack owing to the small diameter of the carbon fiber, which is only 7 µm. Even though finite element analysis has been utilized to investigate the failure progress in the 90° layer considering the interfacial debonding in addition to the yielding of the matrix resin, they usually require many parameters to represent the experimental results14–17.
To focus on the interfacial debonding of a specific carbon fiber, a single fiber fragmentation test has been widely used to estimate the interfacial shear strength by embedding a single carbon fiber along the loading direction18–20. The interfacial debonding in the transverse direction can be characterized by a cruciform specimen method21. However, the interfacial debonding initiated from the interior of the matrix resin may not correspond to those initiated from the CFRP laminates because the transverse cracks are generally initiated from the free surface. Martyniuk et al. prepared an epoxy sample embedding a relatively large single glass fiber with a diameter of 50 µm, and observed it using in situ synchrotron radiation (SR) X-ray computed tomography (CT)22. They succeeded in capturing the three-dimensional (3D) debonding behavior of the fiber from the free surface toward the interior of the sample. This implies that the interfacial debonding of a carbon fiber with a diameter of approximately 7 µm can be observed if X-ray CT can be realized with a higher magnification and resolution. The sensitivity to the difference in the densities needs to be also improved because the density of the matrix resin is closer to that of carbon fiber than that of the glass fiber.
In this study, an epoxy sample with a single carbon fiber embedded in the transverse direction was prepared to capture the interfacial debonding between the carbon fiber and epoxy matrix. Interfacial debonding was generated under static and cyclic loading, and was observed using SR X-ray CT at the large synchrotron radiation facility SPring-8 (Hyogo, Japan). To realize the in situ SR X-ray CT, a piezoelectric actuator-driven desktop fatigue testing machine was developed to apply static and cyclic loads directly along the beamline.