Estimation of interfacial properties from hysteretic energy loss in unidirectional ceramic matrix composites

Abstract

When ceramic matrix composites are subjected to fatigue loading levels sufficient to initiate microstructural damage to the constituents, the mechanical response of the laminate, e.g. the residual strength, stiffness and life of the composite, is governed by the physical state of the fiber/matrix interface. During loading, the chemical bonds, which develop between fiber and matrix during processing, are broken. This 'debonding' results in a significant decline in load transfer between the two constituents and leads to a measurable increase in laminate compliance. With continued cyclic loading, the interface debonds grow in length which further degenerates the composite strength. Moreover, within the debonded regions, frictional sliding between the fiber and matrix is permitted and leads to surface wear of the constituents [1]. Ultimately, the progression of this damage mode leads to a further decline in the interfacial shear stress and load transfer between the constituents. Knowledge of the progression of both damage mechanisms, debonding and the reduction in interfacial shear strength, is critical to characterize ceramic composites since these mechanisms govern, in large part, the degradation in laminate properties. Unfortunately, experimental observation of these kinds of damage is not an easy task. However, attempts to measure these properties experimentally using single fiber and microcomposite tests have been conducted [2, 3]. Moreover, several techniques for estimating interfacial properties computationally using various models have been presented in the literature [4-8]. As in this study, several micromechanics models use hysteresis measurements to gain insight into the state of the fiber/matrix interface [4-6, 8]. The authors use the hysteresis data for a myriad of purposes ranging from the derivation of empirical constants to validation of simplified failure criterion. The current study is unique in that assumptions are not made regarding either the failure of the interface (debonding), nor the associated degradation in shear resistance during fatigue. The present study attempts to infer a logical progression of both mechanisms without specific failure criteria. Rather, the analysis is a 'what must they be' comparison between the experimental measurements of hysteretic energy loss within a given fatigue cycle and the numerical predictions from the one-dimensional shear-lag analysis. The strength of the model and its application as presented herein resides in its simplicity allowing the validated approach to be incorporated into more rigorous micromechanics models which more accurately model the instantaneous state of stress within the laminate as has been the evolution of the early shear-lag models.