Identical location scanning transmission electron microscopy is a method for materials characterization which enables imaging the same regions of interest in a material before and after a certain process. Utilizing this method for observing changes of hydrogen fuel cell catalysts in the form of metallic nanoparticles on carbon support enables a reliable assessment of their structure-stability relationship as changes can be observed at the atomic scale. For a faster, more objective and reproducible analysis of such images at atomic resolution, image analysis algorithms can be used. This master's thesis focuses on the development of such algorithms, particularly on taking advantage of concepts already in use in other fields and transferring them to electrocatalysis. After an introduction into the problem, identical location microscopy, image simulation, and image analysis algorithm development are described. The latter represents the core of this work and covers various preprocessing methods, determination of atomic column positions, segmentation of images into structurally different parts, and alignment of identical location images. The algorithms were developed on experimental images of a catalyst featuring platinum-cobalt alloy nanoparticles on carbon support, and tested on simulated images of idealized models of platinum nanoparticles on carbon support. After demonstrating the algorithms' performance on both types of images, their application on an actual example of a catalyst with platinum-cobalt alloy nanoparticles is shown and a discussion of advantages and disadvantages of this approach follows thereafter.
|
