Z-Contrast Imaging of Grain-Boundary Core Structures in Semiconductors
- PDF / 3,873,940 Bytes
- 5 Pages / 576 x 777.6 pts Page_size
- 55 Downloads / 201 Views
determined by Lord Rayleigh, the incoherent mode of image formation has double the resolving power of the coherent mode.2 Incoherent imaging with electrons is most easily implemented in a scanning transmission electron microscope. The Z-contrast image is obtained by scanning an electron probe of atomic dimensions across the specimen and collecting electrons scattered to high angles. The 100-kV and 300-kV microscopes used for the work described in this article are capable of producing electron probes of 0.22 nm and 0.13 nm, respectively. These probe diameters determine the resolution of the technique. The resulting map of the scattering power for the specimen is highly local, for it is mostly scattering produced close to the atomic nuclei that gets out to the high-angle annular detector. The annular dark-field detector collects a substantial fraction of the scattered electrons and thus effectively averages over the phase relationships between diffracted beams leading to an image with incoherent characteristics. A direct correspondence exists between the object and its image, and there are no contrast reversals with sample thickness or objective lens defocus.3 As long as the electron probe is smaller than the atomic column spacing and the inner detector angle is much larger than the beam convergence, the imaged atomic columns can be treated as independent scatterers. With this technique, unanticipated atomic arrangements present at defect cores will be directly revealed. Additionally as the inner detector angle is increased beyond the minimum angle for transverse incoherence, the role
of thermal diffuse scattering in the imaging process increases, decreasing the coherence length along the atomic columns.4 Thus the scattered intensity can approach the compositional sensitivity of the atomic-number-squared dependence of Rutherford scattering and allows detection of compositional inhomogeneities—for example, impurities segregated in the defect cores. We have used this technique to explore the atomic arrangements present in silicon tilt grain boundaries. These examples illustrate the power of the technique to provide a unique and unambiguous image of defect cores, including several structures not previously anticipated. Models of Grain Boundaries There are two general categories of grain-boundary models. The first was that of Quincke, who suggested the presence of an amorphous cement between grains. 5 The other view is that grain boundaries are best described as a crystalline zone between the lattices of two adjacent grains. Low-angle grain boundaries composed of an array of lattice dislocations became the defining structures of these models since the original suggestions by Taylor6 and Burgers.7 However at misorientations greater than about 15°, the separation of the required dislocations becomes so small that the description was thought to be unphysical. Instead most high-angle grainboundary models at present incorporate more complicated but still ordered transition zones to accommodate the misorientation. The most co
Data Loading...
