An intensive study is carried out to investigate the veracity of an elasticity model called the axisymmetric damage model (ADM) to predict the micromechanical failure modes in a number of ideal unidirectional composites made from silicon carbide fibers and various glass matrices with uniform fiber spacing. In Part 1, the specimens contain two fiber coatings. Detailed experiments are conducted to study the fracture process by observation of surface cracking and through SEM micrographic analysis. Failure scenarios postulated through the ADM are consistent with the observed behavior using the measured constituent properties. Some of the features of the composites predicted in this work are: (1) Numerous matrix microcracks, approximately normal to and bridging the fibers, are observed. For the present materials, matrix crack initiation precedes the onset of nonlinearity of the composite stress-strain curve. A semiempirical model derived earlier and a mechanistic crack growth model are used to study this behavior. (2) Carbon mounds (or mountains), which are roughly triangular in cross-section and have a base width of approximately 5 μm, occur in the specimens. Their location displays a sensitive dependence on the properties of the glass matrix. The ADM suggests the origin of this phenomenon and the loading condition leading to it. (3) A massive degree of debonding is observed within the interfacial region. Again, the nature of this debonding is quite sensitive to the properties of the glass matrix, among other things. The model results are consistent with debonding initiated by radial tension and having continued growth under shear stress. (4) The progressive matrix cracking process is statistical in nature as the distance between successive microcracks varies considerably with position at their initiation. Calculation of the matrix stress field ahead of the interfacial debond leads to bounds, which surround the experimental observations, on the average spacing as a function of applied and residual stresses. (5) Multiple fiber breaks suggest stress transfer has taken place in some regions such that high values of the redistributed fiber axial stress have developed. It is inferred that such failures can only take place in regions where the interface is continuous, while the model results are consistent with the observed failure stress in the fibers. We conclude with an evaluation of the quality of the model and the definition of future research and data needs required to further solidify our findings. We also offer some comments on the implications with regard to our understanding of the behavior of ceramic matrix composites. The well-controlled microgeometry of these composites and the overall consistency of their behavior with analytical models suggest that they can be regarded as truly pedigreed materials.
Read full abstract