The creep rate (ė) predicted by the boundary diffusion (Db) model is ė≃150σDbWΩ/(GS)3kT, where σ is the stress, W is the boundary width, (GS) is the average grain size, and Ω is vacancy volume. The stress dependence is the same as the lattice diffusion model, given by C. Herring, while the grain size dependence and the numerical constant are greater for boundary diffusion. Discussion of the mechanism of creep in polycrystalline alumina is based on the differences between the lattice and boundary diffusion models.

Photomicrographs of pore and grain boundary structures in

Porous structures having a continuous solid phase with isolated pores were prepared by the addition of different amounts of crushed naphthalene to an alumina casting slip. Samples of from 5 to 500/, porosity were fired together for comparable grain development, eliminating structural variables except porosity. Eff ects of porosity and temperature on strength, elastic modulus, modulus of rigidity, and coefficient of thermal expansion were investigated. Effects of porosity on thermal stress resistance and torsional creep properties were studied at constant temperature.

Experimental measurements of the rate of shrinkage of pressed A1203 compacts, the neck growth between single-crystal A1203 spheres and plates, and the effect of particle size on neck growth between single-crystal AlzOa spheres and plates are mutually consistent with the bulk diffusion sintering model. Tbe temperature dependence of the rate of shrinkage and neck growth in AizOJ is characterized by an activation energy of 165 kcal. per mole. Apparent diffusion coefficients and temperature dependence calculated from the shrinkage of pressed compacts of Fez03 agree with measured diffusion coefficients for the diffusion of Fe in Fez03.

Models for initial-, intermediate-, and final-stage densification under pressure have been developed, which explicitly include both the surface energy and applied pressure as driving forces. For the initial stage, the time dependences and size effects given by the integrated equations are identical to those reported earlier for surface energy (alone) as the driving force. The only modification is that the surface energy (γ) is expanded into (γ+PaR/π), where Pa is the applied pressure and R is the particle radius. For the intermediate stage of the process, the Nabarro-Herring and Coble creep models may be adapted to give approximate (∼4×) densification rates for lattice and boundary diffusion models, respectively. In these cases the complex driving force is written as: (Pa/D+γk), where D is the relative density, and k is the pore surface curvature. At the final stage of the process those models are invalid; an alternate model is developed based on diffusive transport between concentric spherical shells which will give a better assessment of the time dependence of densification high density (>95%); the driving force is (Pa/D+γk) in this case also. Because of the fact that the pore size is some unknown function of density, the rate equations cannot be integrated without further information. It is shown that of the various relations which have been assumed in development of models for hot pressing, for the effective stress in relationship to the applied stress and the porosity, (Pa/D) is the only form which satifies the criteria demanded by self-consistency in generation of steady-state diffusion models.

During sintering in alumina powder compacts, the density has been found to increase linearly with the logarithm of time

Effects of chemical inhomogeneities and single-crystal seeds on normal and discontinuous grain growth were investigated in both undoped and MgO-doped A1203. The chemical impurities in the samples were exsolved at a lower temperature than the sintering temperature and measured by SEM/EDS to determine the correlation between the distribution of impurities and the microstructure in A1,03. A feature of this study was the use of clean-room processing and firing procedures to maintain sample composition at the levels initially present in the starting powders. As the local concentrations of chemical impurities (i.e., Si, Ca) increased, the grain boundary-grain boundary dihedral angle distribution widened, with many angles at MOO, the grain-size distribution widened, and pore-boundary separation was enhanced. Discontinuous grain growth was observed in regions of undoped AI2O3 containing the largest Ca and Si concentrations. It is suggested that doping with MgO solute reduces the effects of impurities on grain growth by increasing the bulk solubility and decreasing interfacial segregation of impurities, especially Si, and by narrowing the distribution of grain boundary-grain boundary dihedral angles. [Key words: alumina, grain growth, microstructure, sintering, inhomogeneity.] I. lntroductionHE influence of MgO on the sintering of A1203 has been T studied extensively. It is known that MgO suppresses poreboundary separation, eliminates discontinuous grain growth, and decreases the average grain growth rate in dense Al2O3.'-l2 The detailed mechanism of how MgO acts to change the microstructure evolution of A1203 is still not understood.The objective of this study was to investigate the effects of large seed crystals and chemical inhomogeneities on local microstructure, especially their role in the initiation of discontinuous grain growth in A1203 of typical purity (~2 0 0 ppm impurities). In this study, sintering and grain growth in undoped and MgO-doped AI2O3 have been examined using compacts seeded with single crystals. Clean-room processing and firing techniques have been used to reduce uncontrolled additional contamination so that effects of chemical inhomogeneities in the starting materials could be isolated. The single-crystal seeds were placed in the samples to determine the importance of a grain which is large relative to the average matrix grain size in poreboundary separation and in discontinuous grain growth in A1203 and to study the orientation dependence of growth. The impurities in the undoped and MgO-doped A1203 samples were exsolved at a lower temperature than the sintering temperature and the resulting second phases were examined qualitatively by scan-

The grain-growth behavior of A1203 compacts with small contents (C10 wt%) of various liquid-forming dopants was studied. Equiaxed andlor elongated grains were observed for the following dopants: MgO, CaO, S i 0 2 , or CaO + TiOz. The platelike grains, defined as the abnormal grains larger than 100 pm with an aspect ratio 2 5 and with flat boundaries along the long axis, were observed when the boundaries were wet with the liquid phase and the codoping satisfied two conditions of size and valence. These dopings were NazO + SiOz, CaO + S O 2 , SrO + S i 0 2 , or BaO + SiOz. However, an addition of MgO to the A1203 doped with CaO + SiOz resulted in the change of grain shape from platelike to equiaxial. Equiaxed grains were also observed for the MgO + Si02 doping, indicating that two conditions were necessary but not sufficient to develop the platelike grains. The fast growth rate of the platelike grains was explained by an increased interfacial reaction rate due to the codopants. At the same time the codopants made the basal plane, which appeared a s the flat boundaries, the lowest energy plane. The appearance of the platelike grains was favored in compacts with a small grain size and with a narrow size distribution at the onset of abnormal grain growth. Accordingly, the use of starting powders with a small particle size and narrow size distribution, smaller amounts of dopings, and high sintering temperature resulted in an increased number of the platelike grains.[