When pure polycrystalline Ag is heat-treated in a atmosphere including $O_2$ at 520℃, oxygen atoms adsorb on the Ag and diffuse into it. As a result of the diffusion of oxygen atoms, grain boundaries of Ag migrate. If it, however, is heat-treated in a atmosphere including no oxygen atoms such as $N_2$, grain boundaries do not migrate. Though composition analysis is not made on the specimen showing the grain boundary migration, it is indirectly shown that migration is caused by oxygen atoms. Since oxygen atoms solve interstitially into Ag, it is also shown that grain boundary migration by interstitial atoms is also possible. Because of high diffusivity of interstitial atoms, coherency strain energy, which is proved to be the driving force of diffusion induced grain boundary migration by substitutional atoms fails to explain the migration induced by interstitial atoms.
A new model is proposed assuming that misfit dislocations are generated to accomodate the lattice parameter change during oxygen diffusion into polycrstalline Ag and that density of misfit dislocations is dependent on the surface orientations. Since orientations of two grains consisting a grain boundary, densities of misfit dislocations generated in the two grains are different. Due to dislocation density difference between two grains driving force of migration exist and grain boundary migrate into a grain with higher dislocation density. Grain boundary migration induced by interstitial atoms is considered as similar phenomenon to strain induced grain boundary migration observed in Al when it is mechanically deformed.
Faceting of grain boundary is observed. Faceted plane of a grain boundary is often observed to be parellel to the twin plane of a retreating grain so {111}planes. The tendency may be explained by the step mechanism of grain boundary migration proposed by Gleiter. According to the step mechanism {111}planes have no steps and lowest mobility. If a grain boundary plane during migration become parellel to {111}planes, its velocity drops drastically. The grain boundary plane with the orientation remains due to the low velocity and become the part of the growth shape.
New grains nucleated and grow on the edge of the specimen as well as the migration of grain boundaries present in the interior. This phenomenon happens near the edge only so might not be induced by only the diffusion of oxygen atoms. Instead, nucleation of new grains may be caused by the exess dislocations generated during specimen preparation and have no relations with diffusion of oxygen atoms. Growth of nucleated grains continue over the region of excess dislocations. Therefore, growth of nucleated grains is thought to be driven by dislocations generated by diffusion of oxygen atoms.
Faceting of Ag surfaces is observed when it is exposed to atmosphere containing oxygen atoms at 820℃ . Oxygen atoms adsorb preferentially on the low index planes such as {111},{100} and {110} and reduce the surface energies. Original surfaces break up into low index planes to reduce the total surface energy. Fractions of faceted planes are calculated on the basis of Wollf theorem. Orientation of original surface is assumed to maintain during faceting. According to the calculation, stable regioin against the faceting exist. The region near the grain boundary grooved prior to the faceting so orientations of the region change continuusly. One of the faceted planes, which is not a crystalline plane is curved near the grain boundary.
Allotropic transformation of pure Fe proceeds by massive reaction irrespective of cooling rates. When it is cooled at 10℃/$\min$ through the transformation temperature following holding 1 h at 950℃ , transformation from $\gamma$ to α occur. α grains grow across the grain boundaries of $\gamma$. It is called boundary crossing and shows that transformation proceed by massive transformation. This implies that massive transformation does not need large driving force, which is contradict to that massive reaction is possible quite below the $T_0$. Instead, the growth rate of massive phase is very fast because no compjosition chang happen across the interface and growth is interface-controlled. If nucleation rate of massive phase is comparable to that of ewuilibrium phase, massive reaction become dominant due to higher growth rate. When it is cooled near the $T_s$ parent phase undergo massive reation in the two phase region.
