Grain boundary (GB) solute segregation has long been associated with premature failure of engineering materials. Local changes in composition at the GBs can result in such phenomena as intergranular corrosion, temper embrittlement, intergranular fracture and fatigue as well as changes in electrical properties. The relationship between GB structure and solute segregation is not well understood.

The transformation interfaces of some well known products of austenite decomposition in steels, namely pearlite, and proeutectoid grain boundary allotriomorphic cementite and Widmanstatten cementite, have been examined by TEM. This has been accomplished by using high- Mn high- C alloys that decompose at intermediate ageing temperatures but in which the untransformed parent austenite phase remains stable at room temperature, thus preserving the transformation interface. In conventional steels the residual austenite would undergo the martensite reaction, destroying the interface in the process. Defects in the interface have been analysed and related to intersections with stacking faults in the parent austenite phase. The nature and occurrence of these stacking faults, and their effect on the transformation, is considered.

Theories of late-stage phase separation are discussed from the perspective of statistical mean-field approaches, whereby the material interfaces are assumed to interact with the appropriate average microstructural environment. For coarsening of two- and three-dimensional phases, recent progress is presented for selecting the most appropriate averages that characterize the microstructural environment surrounding a domain. The effective mean field, as well as the average interaction distance over which transport occurs, may be determined self-consistently by imposing global constraints that reflect the microstructural phase fractions and by using explicit spatial and ensemble averages of the matrix transport fields. Comparison of new theoretical coarsening rates with liquid-phase sintering experiments show good agreement over a wide range of phase fractions. Extension of this approach to grain growth, which involves more complex topological interactions, is also discussed.

We present a review of systematic studies of diffusion induced grain boundary migration (DIGM). The results are compared with structural models for the grain boundaries in order to assess the effects of structure upon DIGM. The nucleation of DIGM is also assessed in the light of grain boundary structure and it is demonstrated that changes of grain boundary solute concentration can induce large enough energy changes to drive novel grain boundary dissociation reactions.