Virtual dark‐field and Virtual high angle angular dark field images reconstructed from electron diffraction patterns

Abstract

Dark‐field images are routinely produced in transmission electron microscopy (TEM) by selecting a specific diffracted beam with the objective aperture while the incident beam illuminates a large area of the thin specimen. Similar pictures are produced in scanning‐transmission electron microscopy (STEM) mode by scanning the focused beam over the region of interest and reconstructing the microstructure thanks to the signal collected with an annular detector. When using high angle annular dark field (HAADF) or low camera length, the signal is then sensitive to chemical composition, and the resulting image illustrates phase composition. In both cases TEM or STEM, the information carried by the diffracted electron beam is filtered thanks to a given technical component that provides a limited range of possible settings. Typically, few apertures, which differ by their diameter, are available in conventional TEM. In STEM mode, the camera length is the only practical parameter that may be adapted for sorting the signals. In some cases it could be of interest to extend these capabilities to non‐standard situations. For example, small precipitates that promote faint diffracting beams could be highlighted by collecting the intensities of not only one but several of them. To that respect, acquiring the entire diffracted signal with a spatially resolved detector ‐ e.g.: a CCD camera ‐ and sorting numerically the information out of this complete set of signal may reveal refined features and produce uncommon views of the microstructural features. ACOM‐TEM technique [1] allows such image treatment. It is this approach that is used to construct so‐called virtual bright field (VBF) or dark field (VDF) images. The construction of such images has been described in [2]. The set of diffraction patterns (DP's) collected during a scan is a digital data that may be post‐processed in non‐restricted way for sorting the relevant information about material's microstructure. An example of the filtering capability of the present approach concerns materials in which phases were not easily recognise by template matching technique, because of their diffraction patterns being too similar. Such a case is illustrated with sintered Diamond/Colbalt material (fig 1 a). Indeed due to dynamical diffraction effects, diffraction patterns from Diamond (crystal structure fd3m, a=0.3566nm) and Cobalt (crystal structure fm3m, a=0.3544nm) are geometrically similar, and cannot be interpreted in terms of two different crystal structure. The resulting crystal orientation map is correct, but the Phase Map is irrelevant (fig 1. b). Limitation of template matching technique (TM) used by ACOM‐TEM technique is illustrated figure 1.c) and d), when a Diamond's DP can be better recognize using Cobalt templates (green) with index quality (IQ) of 1723 than diamond templates (red, IQ=1700). By contrast, the background of the diffraction patterns for both materials are different, depending essentially of the atomic number Z of the compound. Therefore, creating, as post data treatment, VDF using information's contained in the background of DP will help to discriminate chemical composition (fig 2 a) and b)), and generate correct phase map (fig2.c)), without help of further complexed chemical analysis.