Search

Images with dimensions of 3-5 A were observed in the tilted beam dark field mode electron microscope using the diffuse scattering from b.c.c. Zr-Nb alloys. In an attempt to identify the origin of the image contrast, computer calculations were made of the dark field images from various crystalline defects with atomic dimensions. For the experimental conditions of this study, the calculated image profile for a vacancy had a width at half maximum of ∼ 3 Å and a peak intensity of 6 × 10-4 of the incident intensity. The measured intensities from the observed images were in the range of 4 × 10-3 to 1·2 × 10-2 and a comparison of the calculated and experimental values showed that the observed images could not have resulted from the simple defects used for the contrast calculations. The present status of this work is that the observed contrast results either from another type of crystalline defect, possibly associated with a phase change in this alloy system, or from a surface film on the thin foil specimen. Some experimental evidence exists to reduce the possibility of the latter, while preliminary image calculations demonstrate that higher intensities can occur for a newly proposed defect structure. It is suggested that the best chance for success in the direct imaging of point defects would be achieved by using the dark field mode with a heavy element specimen having a defect with a large strain field.

High-resolution electron microscopy was used to image directly the 0-lattice of 〈001〉 twist boundaries with misorientations up to 25° and 0-lattice spacings as small as 7 Å. The results were entirely consistent with the conclusions reached on the basis of the diffraction experiments described in Part I. An experimental method was devised for determining whether the 0-lattice was imaged by conventional diffraction strain contrast or by interference strain contrast which arises from interference between f.c.c. and 0-reflections resulting from the periodic structure of the boundary (Part I). It was demonstrated that spacings down to about 34 Å were imaged mainly by means of diffraction strain contrast, while networks with spacings of about 16 Å or less were imaged mainly by means of interference strain contrast. Efforts were also made to detect secondary grain boundary dislocation networks, and such networks were observed in the vicinity of the Σ = 5 boundary over a slightly larger range of misorientations than had been detected previously.

An electron diffraction technique was developed to study the structure of high-angle grain boundaries. Diffraction patterns were obtained from the grain boundary region, and it was demonstrated for the first time that high-anglo 〈001〉 twist boundaries have a periodic nature over the entire range of possible misorientations. Relatively strong grain boundary reflections corresponding to reflections from the 0-lattice wore obtained from all boundaries studied. These results indicate that atomic relaxations occur around the 0-lattice points and those relaxations appear to be qualitatively similar to those which are well known to occur in low-angle twist boundaries.

An electron microscope imaging technique is described which is sensitive to the change in mean inner potential associated with the core region of dislocations in NiO. Through-focus series of end-on dislocations in grain boundaries in NiO imaged under kinematical conditions show characteristic contrast behaviour related to the presence of a change in mean inner potential. Theoretical contrast calculations confirmed that the image contrast is caused by a decrease in mean inner potential. The origin of this decrease could be a change in density at, a charge on, or local segregation at the dislocation core. By fitting the calculated images to the observations a semi-quantitative estimate can bo made of the decrease in mean inner potential. If this decrease is related entirely to the decrease in density, there is poor agreement \v ith the predictions of theoretical models of the dislocation core structure in NiO.

The structures of selected small- and large-angle [001] tilt boundaries in gold bicrystals were studied using X-ray diffraction techniques. All the boundaries examined gave rise to extra reflections. It was shown that these reflections could not result from (1) double diffraction, (2) surface diffraction or (3) diffraction of radiation with one-half or one-third of the characteristic CuKα wavelength and, therefore, must come from the grain boundary. The presence of these reflections is direct evidence for the periodic nature of all the grain boundaries studied. The grain boundary reflections were frequently split into pairs which are present on the high- and low-angle side of the tail of the f.c.c. reflection. This splitting was related to the presence of a small twist component superimposed on the tilt boundary. In general the low-angle reflections were observed to be higher intensity than the high-angle reflections. The analysis of these intensity observations indicates that the tilt-boundary structure in the plane of the boundary consists of a large region with a small expansion followed by a small region with a large contraction. The diffraction technique is unique in being able to obtain this type of detailed information on the structure of grain boundaries.

The diffraction effects expected from the periodic structure of twist boundaries in Si are determined by an examination of the reciprocal lattice of these boundaries. Methods of analysis are developed to distinguish between the real diffraction spots due to the periodic boundary structure and those due to double diffraction effects. The electron microscope images from the boundaries studied in Si bicrystals were frequently found to be complex and contained moiré fringes which provided no information on the actual boundary structure. By analysing the electron diffraction patterns from these boundaries for the presence of new diffraction spots it is possible to show that all the Σ1 [001], Σ1 [111] and Σ3 [111] twist boundaries examined have a periodic structure.

An electron diffraction technique has been developed to study the thickness of grain boundaries in polycrystalline specimens. Grain boundaries with a periodic structure in Al2O3, Cu-Bi and Au, were examined in an edge-on orientation using electron microscopy and diffraction. Each boundary gave rise to streaks in the diffraction pattern, the length of which was inversely related to the thickness of the boundary. For each boundary studied the magnitude of the boundary thickness was similar to the periodicity of the boundary, as determined from the image and the diffraction pattern.

An important characteristic of a grain boundary is the depth to which its displacement field penetrates into the neighbouring single crystals. This boundary thickness, which is called the 'structural width' in this paper, may be studied by the use of an electron diffraction technique which is sensitive to the width of the region of large strain at the boundary. The periodic structure of a grain boundary gives rise to extra reflections and, since these are associated only with the thin boundary region, they are elongated into relrods along the direction normal to the boundary plane. By selecting boundaries which are parallel to the electron beam, these relrods may be cut along their length by the Ewald sphere; the length of the resulting streaks in the diffraction pattern is inversely related to the grain boundary structural width. Edge-on tilt boundaries with an [001] tilt axis have been studied in manufactured gold bicrystals for misorientation angles from 3.5° to 36.9°, and for (100) and (110) boundary planes. The value obtained for structural width for all the boundaries is shown to be related in almost all cases to the magnitude of the O-lattice periodicity associated with the misorientation.

The influences of bonding type and misorientation angle on the structure of [001] twist grain boundaries along the direction normal to the interface have been studied using an electron diffraction approach in which 'streaks' of diffuse scattering are recorded from the grain boundary core region. The boundaries studied were in NiO (with largely ionic bonding), Au (metallic bonding) and Ge (covalent bonding), in some cases with varying misorientation angles. The electron diffraction patterns have been analysed in terms of changes of plane spacing normal to the interface using the theoretical analysis in the preceding paper (Vaudin, Burkel and Sass 1986). Qualitative conclusions are drawn on the influence of bonding type and misorientation angle on the structure of [001] twist grain boundaries. The results are discussed with particular reference to the limitations of the experimental technique, dynamical diffraction effects, grain boundary segregation and reconstruction of the grain boundary core.

The three-dimensional reciprocal lattice of a large-angle [001] twist boundary in a gold bicrystal was examined using X-ray diffraction techniques. The intensity profiles of the relrods due to the thin grain boundary region were measured and the grain boundary thickness was determined from the inverse of the width of the peaks. The boundary thickness was estimated from several intensity profiles to be ∼ 8 å, which agrees well with the O-lattice period in the plane of the boundary. The detailed shape of the intensity profile for a particular relrod depends in a complicated fashion on the atomic displacements associated with the boundary region. As an indication of the complications that can arise it is noted that one peak gave a thickness of approximately one-half that measured from the other peaks. At present the variation in peak width for different relrods is not understood.