It was recalled that, in many cases, alumina scale was assumed to grow predominantly via O diffusion, although it had also been reported that such growth could be controlled by the diffusion of Al. These mechanisms could be modified by active elements. Unfortunately, there was a lack of data on self-diffusion in alumina and, due to its stoichiometry, the diffusion data corresponded to an extrinsic diffusion mechanism so that it was not possible to compare O and Al diffusion coefficients. In order to obtain information about the growth mechanism of alumina scale, 18O and 26Al self-diffusion coefficients in Al2O3 were determined by using the same samples under the same experimental conditions so as to permit direct comparison. Bulk and sub-boundary diffusion coefficients for both isotopes were determined in single crystals of undoped alumina. Grain-boundary diffusion coefficients were estimated only for O diffusion in polycrystals. The diffusion of O in the lattice, sub-boundaries and grain boundaries of yttria-doped -alumina was also studied. It was found that the O and Al bulk diffusion coefficients were of the same order of magnitude. In the case of sub-boundaries, the Al diffusion was slightly faster than O diffusion. Doping with yttria caused a slight increase in the bulk O diffusivity, but decreased the grain-boundary diffusion coefficients; due to segregation effects. The results were compared with the oxidation constants of alloys which developed an alumina scale upon oxidation. It was concluded that neither lattice self-diffusion nor grain boundary self-diffusion could explain the growth rate of alumina scales.

M.Le Gall, A.M.Huntz, B.Lesage, C.Monty, J.Bernardini: Journal of Materials Science, 1995, 30[1], 201-11