Analysis of Strain Relaxation, Ion Beam Damage and Instrument Imperfections for Quantitative STEM Characterizations

It is illustrated that the preparation of thin specimens from bulk materials can have significant influence on the interpretability of (S)TEM data. The results of the presented measurements show that and the elastic strain relaxation in low dimensional structures alters the overall strain state of the material – and hence affects strain measurements – as well as the contrast of STEM measurements and is generally needed to be incorporated in comparative simulation studies that involve strained structures. Furthermore, the ion beam thinning process itself can introduce – even with relatively low energies – a serious alteration of the surface which can affect the contrast of STEM measurements. Hence, the correlation to thickness measurements is complicated due to the distinct difference in scattering behaviour between (partially) amorphized surface layers in comparison with crystalline material. Although parts of these effects cannot be avoided the inclusion of amorphous pseudo-oxide layers in simulations has been shown to provide reasonable agreement with the experimental data. Furthermore, the impact of a finite electron source with limited coherence has been investigated. It can be shown that a reproduction of experimental contrast by simulation can only be achieved by the inclusion of an additional focus spread as well as a lateral point spread due to partial spatial coherence. Finally, the previous results are combined to reconstruct the three-dimensional shape of several antiphase domains within gallium phosphide grown on silicon-(001). At first the concept was demonstrated for a simple but highly strained interface and second for large structures with thousands of atomic columns. It is shown that although the contrast mechanism for annular dark-field imaging is in principle straight forward and mathematically simple, the details of atomic resolution microscopy are still very challenging. Realistic assumptions about the specimen properties and the electron optics have been shown to be of great relevance for data evaluation. It is clear that the research should be extended to the regime of low angular dark-field imaging where strain and inelastic scattering play a even more relevant role. Furthermore, it is of great importance to investigate the aforementioned practical aspects of damage layers and optical imperfections for other advanced imaging techniques like diffraction imaging. In addition, it is worth investigating in how far through focus depth section can be utilized to increase the reliability of structure restoration along the transmission direction. It is expected that the improvement of accuracy and robustness of atomic counting techniques will greatly increase the power of a (S)TEM by providing simultaneously lateral and depth information about arrangement and composition. Furthermore, it is clear that the role of high performance simulations will have an even more important role in the future