Growth strains and creep in thermally grown alumina: Oxide growth mechanisms

In situ measurements of growth strains and creep relaxation in alpha-Al2O3 films, isothermally grown on beta-NiAl alloys at 1100 degrees C, are reported and analyzed. Samples containing the reactive element Zr, and Zr-free samples, are examined. For Zr-free samples, steady state growth strains are compressive, whereas the growth strains are tensile when the reactive element (RE) is added to the alloy. This behavior is attributed to the counterflow of oxygen and aluminum interstitials, and to simultaneous counterflow of oxygen and aluminum vacancies, all moving through the grain boundaries. Cross diffusing oxygen and aluminum interstitials may merge and combine within the film, forming new oxide along grain boundary walls, a mechanism that leads to an in-plane compressive stress. Cross diffusing oxygen and aluminum vacancies will also merge and combine within the film; in this case material is removed from grain boundary walls, a mechanism that leads to an in-plane tensile stress. When no RE is present, the interstitial mechanism dominates and the resultant stress is compressive. Consistent with the dynamic segregation model, the RE slows the outdiffusion of A1 interstitials permitting the tensile mechanism to dominate. This interpretation invokes the unconventional view that oxygen and aluminum interstitials and vacancies, created in and driven by the strong chemical gradient, all participate meaningfully in the scale growth process. Grain boundary diffusion measurements were obtained from low stress creep data, interpreted using the Coble model of grain boundary diffusion. Reported diffusion measurements of oxygen through grain boundaries of alpha-Al2O3, which are known to be inconsistent with oxide scale growth, are critically examined. A simple picture, a balanced defect model, emerges that is consistent with the dynamic segregation model, observed growth stresses and their dependence on the presence of a reactive element, sequential oxidation experiments, and our best knowledge about grain boundary diffusion coefficients. (C) 2008 American Institute of Physics. [DOI: 10.1063/1.3009973]