Comment on ‘‘Effect of Interface Structure on the Microstructural Evolution of Ceramics’’

THE following comments are directed to the sections (2) ‘‘revisit to the grain growth problems in alumina’’ and (3) grain growth without a liquid phase in the article of concern. The purpose of this comment is to challenge the authors’ assertions and refute the following arguments, namely, first that ‘‘MgO can be considered a very effective grain growth promoter in the sintering of alumina,’’ second that abnormal grain growth (AGG) occurs only when the grain boundaries are facetted, and third that interface-controlled grain growth in alumina is pervasive even in aluminas containing impurities in the parts per million level. The authors reappraise the role of magnesia as a sintering aid in alumina in the context of interfaceversus diffusion-controlled growth. The appraisal of the literature provided in the article ignores several key pieces of data that contradict the authors’ arguments. Evidence of interface-controlled grain growth manifests itself in several readily observable phenomena, which are faceted grain boundaries, a grain growth exponent equal to two, a grain boundary mobility that is independent of liquidphase content or dopant concentration, and the presence of a liquid phase. While some abnormal grains in alumina display faceted grain boundaries, there have been numerous observations of curved and atomically rough abnormal grain boundaries (see Fig. 1). In fact, the highest mobility abnormal grains measured to date in alumina are reported to have been curved and atomically rough. Images provided in some of the previous publications by the current authors have also shown atomically rough grain boundaries in alumina. The article states, however, that the abnormal boundaries are ‘‘apparently straight without exception.’’ Often, abnormal grains in alumina show a plate-like morphology with a large slow-growing faceted basal plane, while the fast growth directions are often curved. A fair appraisal of the literature suggests that grain boundaries of abnormal grains may be either faceted or rough. The grain growth exponent for both doped and undoped alumina has been observed to be greater than two, which suggests diffusion-controlled grain growth. The grain growth exponent has been observed to be as large as six in some doped aluminas. All of these data suggest that grain growth and AGG in alumina are not always interface controlled. The article suggests that because all commercially available alumina is slightly impure that there must be intergranular films present at the grain boundaries of abnormal grains in alumina and in magnesia-doped alumina. However, there are no observations in the literature of an intergranular liquid or glass film present in magnesia-doped alumina showing normal grain growth when the dopant or impurity concentrations are in the parts per million levels. This supports Gavrilov et al.’s 4 assertion that magnesia acts as a liquid-phase ‘‘scavenger’’ or at least prevents the formation of intergranular films in some manner. The article references work showing intergranular glass in alumina that is doped with several atomic percent of dopant. The roughening of a boundary by magnesia containing an intergranular film has only been shown for thick wetting intergranular films rather than for the B1 nm intergranular films that are typically associated with AGG in alumina. Magnesia has also been shown to roughen boundaries that contain no film. It is likely, therefore, that magnesia prevents the formation of the 1–2 nm film that is typically associated with AGG in alumina. It is not fair, then, to suggest that the behavior observed when alumina is doped with several atomic percent of dopant should extend to dopant levels in the parts per million range, because wetting films and equilibrium thickness non-wetting films may behave differently. In addition, the kinetics of these systems have not been thoroughly quantified. It is often difficult to determine frommicrostructures of impinged abnormal grains whether the final grain size distribution is more dependent on the grain growth rate of the abnormal grains or the number density of abnormal grain ‘‘nuclei.’’ Assuming that AGG in alumina were interface controlled, the argument that magnesia prevents AGG by promoting diffusioncontrolled grain growth over interface-controlled grain growth contradicts itself. The grain boundary mobility of diffusion-controlled grain growth should be higher than for interface-controlled growth. The theory presented by the authors predicts that magnesia produces a greater number density of abnormal grains that grow faster and impinge early during growth. The grain growth rate should then be reduced by impingement. It is true that the grain growth rate should be reduced after impingement of abnormal grains because of the reduction in driving force for grain growth; however, the grain boundary mobility remains constantly high, whereas the evidence is to the contrary. It would be useful if the authors discussed the subject in defined quantitative terms such as grain boundary mobility and number density of abnormal grains rather than arbitrary terms such as ‘‘grain growth promoter,’’ which only confuse the arguments. It is instructive to recall the confusion that has arisen in the past from interpreting the observation that MgO can promote grain growth during sintering indirectly through its influence on pore mobility, whereas it actually directly inhibits grain growth in sufficiently dense samples of the same material due to lowering of the grain boundary mobility. Careful measurements of the grain boundary mobility of undoped versus magnesia-doped alumina have been made in both high purity and relatively impure aluminas. In both cases, the grain boundary mobility of the magnesia-doped alumina has been observed to be significantly reduced relative to the undoped alumina. In all cases, the grain size of the magnesia-doped alumina is smaller than the undoped one. This should provide a larger driving force for grain growth in both diffusionand interface-controlled growth. However, magnesia doping also reduces the grain boundary velocity of dense alumina. The results in the literature, therefore, indicate that magnesia doping in alumina reduces its grain size, grain boundary velocity, grain boundary mobility, and is not a ‘‘grain growth promoter.’’ There is yet to be any evidence that singularly doping magnesia into alumina promotes significantly enhanced grain D. Green—contributing editor