Initiation and Display of Delamination Failure During Fatigue of Graphite Fiber-Epoxy Composites

Laminar microdiscontinuities which lead to the delamination of two unidirectional layup graphite fiber-epoxy composite materials during fatigue have been characterized, and discontinuity geometry and type related to the fatigue lifetime of specimens in which they occur. Two types of discontinuities were commonly observed: voids and organic inclusions consisting of regions of excess hardener. A stress intensity range based analysis applied to discontinuities found on the fatigue fracture place demonstrates that the several orders of magnitude scatter in fatigue life found for a series of test specimens is a result of a variation in the type and size of discontinuities present from specimen to specimen. A nondestructive acoustic pulse echo system used to study the progress of crack propagation from the discontinuities during fatigue is also described. Scatter in the values of mechanical properties of graphite fiber-epoxy composites is substantially larger than that typically found for metallic structural materials. This has required the use of large safety factors in the design of structural components using such composite materials. A purpose of the present study was to determine the cause for scatter in the lifetimes of two graphite fiber-epoxy materials subjected to fatigue loading. The failure mode of particular interest was delamination, that is disbanding between layup plies. Delamination is the principal failure mode of fiber-epoxy composites loaded in shear, and is often a weak link in the failure sequence of such materials loaded in both tension and compression. The study was done using beam specimens loaded in 3-point bending, for an R = minimum load/maximum load = 0 loading condition. The materials characterized were 3501-5 epoxy/AS graphite and 934 epoxy/T300 graphite, a low and medium curing temperature material, respectively. Unidirectional fiber orientation 50 ply thick construction was used, providing a 0.63 em thick specimen. Specimens where load in fatigue normal to the layup plane, with th.e fibers positioned parallel to the specimen axis as shown in Fig. 1. The specimens failed due to subcritical crack growth along a plane between plies, followed by a final tensile failure of the outer fibers due to loss in specimen stiffness with the de 1 ami nat ion. Fatigue lifetime data for the materials are shown in Fig. 1, in comparison to the peak shear stress experienced at the neutral plane of the specimen, resulting from the 3-point loading condition. The several orders of magnitude scatter in fatigue lifetime is typical of such materials. Specimens were examined optically, after failure, to characterize areas on the delamination plane which might have acted as initiation sites for crack propagation. Principally, two types of microdiscontinuities were observed: voids and organic inclusions. The voids were apparently formed as the result of displaced fibers in the prepreg tape used to manufacture the composite panels. The organic inclusions were regions of exces' hardener and were easily visible owing to their bright yellow color. General observations were that, typically, there was only one discontinuity of any appreciable size on the fracture surface, and for specimens with longer fatigue lifetimes the discontinuities tended to be smaller. To quantify the relationship between discontinuity size and fatigue life a mode two stress intensity range, t:.K2, was associated with each discontinuity of any appreciable size on the fracture surface of a specimen, and the largest value of t:.K2 found was compared with the specimen fatigue life. The analysis entailed approximating the shape of each discontinuity as an ellipse, on the basis of the location of the preponderance of the area of the discontinuity. Major (2a) and minor (2b} axes of the ellipse were assigned by ignoring small appendages such as illustrated for the void in Fig. 2. A stress intensity solution by Kassir and Sih1 was then used to determine C.K2 for the discontinuity. The cyclic shear stress at the site of the discontinuity, needed to complete the calculation, was determined from the cyclic applied load and the distance of the discontinuity from the neutral plane. Results of such an analysis are shown in Fig. 3. For the 3501-5 epoxy/AS graphite, for instance, one finds that the lifetime data fall into two groups, one describing the effect of inclusions, the other of voids. The scatter in lifetime is substantially reduced by the analysis, demonstrating that it is a variation in type, size and location of the planar microdiscontinuities from specimen to specimen that is principally responsible for the lifetime scatter. The better fatigue properties of the 934 epoxy/T300 graphite is attributed to a smaller average size of discontinuity in the material tested, as also illustrated in Fig. 3. The longer lifetime of specimens containing inclusions, as compared to voids having the same t:.K2, is attributed to a longer interval of crack initiation at the inclusion sites. This was confirmed by a subsequent investigation in which the growth of cracks from the discontinuities was mapped, at increments in the fatigue life, using a computer contra lled acoustic pulse-echo facility. The technique used is illustrated schematically in Fig. 4. A specimen was emersed in a water bath and an acoustic C-scan made using a 15 MHz broadband focussed transducer, with a point-to-point resolution of better than 0.05 em. The X, Y position of the transducer over the specimen was computer controlled and echo samples were taken for a