Determination of the structure of the core of grain boundaries is the key to a fundamental understanding of properties such as grain boundary migration. Whilst theory, especially computer simulation, offered valuable insights, nothing can supplant experimental data. High resolution electron microscopy has been used to reveal the dislocation structure of high angle boundaries, some elements of the structural units which are thought to provide the characteristic building blocks of grain boundaries, rigid translations and a remarkable multiplicity of structure in neighboring regions of particular grain boundaries. The understanding which results from these studies gives an insight to the way in which polycrystalline and polyphase materials evolve towards a state where the excess energy due to the interface is minimized providing the necessary atomic mobility is possible. This evolution occurs at the nucleation and growth stages of microstructural development. In practice nucleation selects low energy orientations which may be perpetuated during growth. For the important case of thin films interfacial processes control (a) the formation of grain structure in single component films and (b) the development of preferred orientations in heteroepitaxial deposits.

High-resolution electron microscopy at 400 kV has been used to study the core structure of large-angle <001> tilt grain boundaries in NiO. The deduced atomic structures show considerable variety and complexity, even for boundaries with the same tilt axis and misorientation. The coexistence of asymmetric and symmetric boundaries which is observed at each misorientation suggests that asymmetric grain boundaries containing densely packed planes are associated with relatively low free energies. A multiplicity of core structures is found for the ∑=5 and the ∑=13 grain boundaries. The normal component of the rigid-body translation in these boundaries is quite small in view of the relatively open structure that is expected for grain boundaries in ionic solids. Substantial concentrations of Schottky pairs near the boundary cores, as well as relaxations parallel to the misorientation axis, are indicated.

GaAs has been grown on (100) and (111) Si and GaP substrates by MOCVD during the same run at a substrate temperature of 550°C. The defect structure of the interface and the epitaxial GaAs layers has been analysed by transmission electron microscopy (TEM). Depending on the substrate type and orientation different defects and defect orientations were observed which can be attributed to different types of misfit dislocations and the thermal mismatch between the two materials.

Quantitative X-ray diffraction techniques have been used to determine the atomic structures of the ∑ = 5 and 13 [001] twist boundaries in Au with a resolution of 0.09 Å or better. The reciprocal lattices of these boundaries were mapped out using synchrotron radiation. The atomic structures were obtained by testing model structures against the intensity observations with a chi square analysis. The boundary structures were modeled using polyhedra, including octahedra, special configurations of tetrahedra and Archimedian anti-prisms, interwoven together by the boundary symmetry. The results of this work point to the possibility of obtaining general rules for grain boundary structures based on X-ray diffraction observations that give the atomic positions with high resolution.

The structure of the ∑=19, (331)/(331), [110] boundary is investigated by high-resolution transmission electron microscopy. Atomic models are considered for this observed boundary taking particular account of the polarity of grains on either side of the boundary. The Burgers vector of a grain boundary dislocation associated with a step observed at this boundary is examined. A lattice translation across (110) APB is investigated through image matching.

Transmission Electron Microscopy (TEM) was used to analyze the interface between RF sputtered alumina thin films and (NiZn)Fe2O4 substrates. Bright field and dark field TEM images, High Resolution Electron Microscopy (HREM), and Selected Area Electron Diffraction (SAED) patterns provided direct evidence of eta-alumina crystalline regions near the filmsubstrate interface. The orientation and existence of the crystalline alumina phase was dependent on the substrate crystal orientation.

Ag-Cd alloys have been internally oxidized to produce CdO precipitates which have a cube-on-cube orientation relationship with respect to the Ag matrix. Electron microscopy studies revealed that the Ag/CdO interfaces are incoherent which can be explainedby the presumably weak bonding forces across the interface and the large lattice misfit. A comparison of experimentally observed lattice images with computer simulated ones based on different interfacial structure models shows that an oxygen plane forms the terminating layer on the oxide side even at interfacial steps. This result is supported by thermodynamic considerations.

The crystallographic theory of interfacial dislocations is outlined, and shown to encompass the case of defects in single crystals. It is demonstrated that four distinctive types of dislocations can arise in interfaces. Defects in two of these categories arise ubiquitously and have been studied extensively in the past, but defects in the remaining categories have only been observed relatively recently. The significance of these latter defects is assessed by discussing the type of interfaces in which they can occur, and their possible contribution to static and dynamic interfacial properties. Specific examples of their roles in accommodating misfit in epitaxial interfaces, and shuffle-less twinning in hcp metals are discussed.

Copper wires annealed at 1000°C to produce a ‘bamboo’ grain structure were used to measure the grain boundary energy, γB, and its dependence on boundary plane orientation, δγB/δϕ. Data for boundaries between crystals showing the ∑11 orientation relation were used to integrate the δγB/δϕ values to give a ‘γB-plot’ for ‑11 boundaries. The ‑11 and ‑33 γB-plots were added together. Comparison with a ‑33 YB-plots indicates the conditions under which a ‑33 boundary dissociates into separate ‑11 and ‑33 boundaries.

Using the sphere rotation method, low energy orientations of Ag-Ni boundaries have been determined by sintering Ag spheres on a (011)Ni-plate [1]. The pole figures reveal that approximately 70% of all spheres have rotated into the orientation (011)Ag [01]Ag ∥ (011)Ni [01Ni. In the following, this orientation is called “parallel cube” (PC)-because all equally indexed directions and planes are parallel. This orientation was also observed in earlier epitaxy investigations [2]. In addition, results from sphere rotation experiments on other fcc materials (Au/LiF, Au/KCl, Au/NaCl, Ag/NaCl, Au/MgO, and Cu/MgO [3,4]) likewise show that the PC orientation is by far the most intensely populated and must be assumed to have particularly low energy. From these observations the question arises: why is the PC orientation energetically favoured in so many systems, despite widely varying misfit and bonding character? As a representative system with large misfit (approx. 15%), heterophase boundaries in the system Ag-Ni are being examined by TEM. An important aspect is the possible existence of misfit dislocations whose expected arrangement can be calculated by the O-lattice theory if they are assumed comparable with bulk edge dislocations. The expected misfit dislocations are located midway between the O-lattice points (Fig. 1). For the compensation of the misfit in the PC orientation, two different dislocation types are necessary:

Transmission electron microscopy techniques were applied to investigate the structure of interfaces between noble metals and NiO. Interfaces between NiO and Pt or Cu were formed by diffusion bonding. NiO and Cu form aligned lattices, while no unique topotactical relationship was found for the NiO-Pt interfaces. Internal reduction of (Ni,Cu)O was also used to investigate NiO-Cu interfaces at polyhedral Cu precipitates in a NiO matrix. The morphology and compositions of precipitates was studied for several reduction treatments. The formation of largely (111) interfaces during solid state reduction suggests that these planes of densest packing are associated with the lowest interfacial free energy. The quite large lattice mismatch is accommodated by misfit dislocations as observed by HREM.

In a polycrystal of silicon, dissociated grain boundaries with orientations close to either the ∑=3{111}, ∑=3{112}, ∑=9{122} and ∑=27{255} coincident site lattices (CSL) were observed both optically and in the transmission electron microscope (TEM). Macroscopically ∑=3{111} and ∑=27{255} appeared stable whereas ∑=9{122} dissociated frequently into two ∑=3{111} boundaries. TEM observations revealed ∑=27{255} to dissociate into one or more triangular grains in several regions. These grains consisted of a ∑=3{111}, a ∑=9{122} and a ∑=9{111}/{115} boundary. The latter ∑=9 boundary dissociated further to form twin triangular grains. The driving forces for various dissociations are discussed.

Recent tight binding and Car-Parrinello simulations of grain boundaries in semiconductors are reviewed. A critique is given of some models of embrittlement that are based on electronic structure considerations. The structural unit model of grain boundary structure is critically assessed using some results for mixed tilt and twist grain boundaries. A new method of characterizing interfacial structure in terms of bond angle distribution functions is described. A new formulation of thermodynamic properties of interfaces is presented which focusses on the role of the local atomic environment. Effective, temperature dependent N-body atomic interactions are derived for studying grain boundary structure at elevated temperatures.

A parametrized tight-binding (TB) method based on the TB-UMTO approach in the atomic sphere approximation (ASA) [26] has been developed. The Hamiltonian is written in terms of the canonical structure matrix and potential parameters. The former is for a given configuration of atoms evaluated using a Dyson-type equation and the latter are those found self-consistently for the ideal lattice. A warping correction has been added to the scheme to be able to account for the effects of local straining which can not be included in the ASA. This is essential for applications in defect studies. Using this method the structure and energy of the ∑ = 5 [001] twist boundary in copper has been calculated.

The bond-bending and bond-stretching three-body potential of Stillinger and Weber is used to study the energy and structure of grain boundaries in silicon.

An atomic model of the structure of a <001> ∑ = 13 grain boundary is proposed. The segregation of cobalt at the same grain boundary is investigated by M∑ϐbauer spectroscopy, autoradiography and the I6/I14 method quantitatively. The results, which are consistent with the assumption that CoSi2 precipitates are formed, are discussed in view of the atomic structure.

A large number of different lattice clusters has been analyzed according to the occurrence of low index atomic planes in the clusters. A relationship of this analysis to the classification of symmetrical grain boundaries is discussed. The classification is based on the combinations of structural units that describe the structure of different grain boundaries. The order of plane types following from the analysis can be considered as a measure of exceptionality of grain boundary types.

In materials with cubic crystal lattices, the structures of grain boundaries are believed to change in a continuous manner as misorientation changes. They have been shown to change discontinuously as grain boundary phase transformations occur with changes of composition or temperature. In this paper we explore the possibilities of continuous and discontinuous structure changes with rnisorientation and temperature, for materials with hexagonal lattices. It is shown that the irrational c/;a ratio of most hexagonal metals gives rise to the likelihood of such transformations.

A constant pressure molecular-dynamics technique is used to simulate aluminum bicrystals. The ion-ion pair potential is obtained from second order pseudopotential theory and is explicitly electron density dependent. The total cohesive energy of a metal also includes a structurally independent term which is electron density dependent. The density dependence of the total cohesive energy is carried through in the lagrangian equations of motion, so the dynamics of the simulation explicitly contains global density dependence. The simulation is at constant pressure and temperature. Temperatures up to the bulk melting point are explored in a ∑=5 symmetric tilt aluminum bicrystal.

A new method for the molecular dynamics simulation of bulk planar interfaces at high temperatures is presented. The method uses the basic Parrinallo-Rahman (constant stress) scheme modified, however, for the application to inhomogeneous systems. Since our computational cell contains only one interface with 2-d periodic boundary conditions we are able to study isolated interfaces all the way up to melting. The interaction between boundaries which may lead to their annihilation at higher temperatures, which is a problem when 3-d periodic boundaries are applied, is thus avoided. As an application, the method is used to study the stability of a grain boundary at high temperatures. Observations on a possible connection between grain boundary migration and “premelting” are discussed.