Swelling and creep damage accumulation in hot-pressed alumina

A. Gordon Robertson


This thesis consists of two parts. The first describes work on swelling - the second, on creep damage articulation.

The first part develops a models for pressure-driven diffusional dedensification (or swelling) in a polycrystal containing isolated pores. The model is sufficiently general to be directly applicable also to pressure sintering. It is tested against experimental data for hot-pressed alumina (HPA) which swells during high-temperature, pressureless anneals in air.

The model includes two independent, coupled, pore growth rate components: (1) diffusional (de)densification, driven by gas pressures in closed pores, and (2) coalescence of pores attached to grain boundaries, driven by grain growth against pore drag. Internal pores pressures may be due to inert trapped gas or to gas generated by chemical reactions.

In a polycrystal with inert trapped gas, the trapped gas drives early swelling, but is self-dissipating. Pore coalescence, by progressively releasing capillarity constraint, becomes the dominant cause of swelling at longer times.

The model predicts what magnitudes of chemically-generated pore pressures will increase swelling rates beyond those arising from inert trapped gas alone. And, for research on creep damage accumulation, it predicts what precreep anneals are required to dissipate an initial trapped-gas pore pressure, and the range of distinct, predamaged porous microstructures available.

The second part of the thesis presents experimental data from flexural and uniaxial tensile creep tests. The data demonstrate that HPA's cavitation damage tolerance and failure strains are high enough that creep fracture tests should be done in uniaxial tension, rather than bending. A high temperature tensile test system, with optical extensometry was designed, built and tested. The uniaxial tensile data suggest design improvements for testing ductile ceramics.

The combined mechanical and microstructural data indicate that, at 1250 C, failures at high stress are controlled by slow crack growth from microstructural heterogeneities. An abrupt transition at about 200 MPa separates rapid failures from failures at strains on the order of 1.0. At about 50 MPa, a more gradual transition occurs, to fracture at higher strains, controlled by strain-driven damage. The internal microstructural data show internal creep damage at levels which cavity coalescence generates macrocracks in HPA.