The fracture of a thin metal foil constrained between alumina or sapphire blocks has been studied by a number of investigators. The systems that have been investigated include Al [1,2], Au , Nb , and Cu . Except for Al/ Al2O3 interfaces, these systems exhibit a common fracture mechanism: pores form at the metal/ceramic interface several foil thicknesses ahead of the crack which, under increasing load, grow and link with the initial crack. This mechanism leaves metal on one side of the fracture surface and clean ceramic on the other. This has not been the observation in Al/ A12O3 bonds where at appropriate thicknesses of Al, the fracture appears to proceed as a ductile rupture through the metal.
The failure of sandwich geometry samples has been considered in several published models, e.g., [6,71. The predictions of these models depend on the micromechanic mechanism of crack extension. For example, Varias et al. proposed four possible fracture mechanisms: (i) near-tip void growth at second phase particles or interfacial pores and coalescence with the main crack, (ii) high-triaxiality cavitation, i.e., nucleation and rapid void growth at highly stressed sites at distances of several layer thicknesses from the crack tip, (iii) interfacial debonding at the site of highest normal interfacial traction, and (iv) cleavage fracture of the ceramic. Competition among the operative mechanisms determines which path will be favored.
This paper addresses the question of why the fracture of the A1/A12O3 system appears to be different from other systems by probing the fracture mechanism using X-ray tomographic microscopy (XTM). We have experimentally duplicated the simplified geometry of the micromechanics models and subjected the specimens to a well defined stress state in bending. The bend tests were interrupted and XTM was performed to reveal the mechanism of crack extension.