Quantitative atom detection from atomic-resolution transmission electron microscopy images
18 september 2019
Campus Groenenborger, T.129 - Groenenborgerlaan 171 - 2020 Antwerpen (route: UAntwerpen, Campus Groenenborger
Organisatie / co-organisatie:
Department of Physics
Sandra Van Aert & Arjan den Dekker
PhD defence Jarmo Fatermans - Faculty of Science, Department of Physics
Nanomaterials have attracted increasing scientific interest, because their exact atomic structure may lead to interesting and unexpected physical and chemical properties. Due to several important developments in aberration correction technology, transmission electron microscopy imaging has become an excellent technique to visualise nanomaterials down to sub-angstrom resolution in order to understand their properties.
However, a merely visual interpretation of electron microscopy images is inadequate to obtain precise structure information. Therefore, a quantitative approach is required. Hereby, an important assumption is that the number of atomic columns in the image is known. This number can be determined visually from atomic-resolution images of beam-stable materials for which a high incoming electron dose can be used resulting in a sufficiently high signal-to-noise ratio. However, beam-sensitive and light-element materials should be imaged with a sufficiently low electron dose to avoid beam damage, which causes images of such materials to exhibit low signal-to-noise ratio and low contrast. In these cases, a visual determination of the number of atomic columns is not straightforward and may lead to biased structure information.
To overcome this problem, an alternative, quantitative method is proposed which determines the number of atomic columns for which there is most evidence in the image data. This method allows detecting atomic columns and even single atoms from atomic-resolution electron microscopy images in an automatic and objective manner. The validity and usefulness of this method have been demonstrated by analysing images of samples of different shape, size, and atom type. Moreover, besides detecting atomic columns from electron microscopy images, the proposed methodology also offers a way to evaluate the relation between image-quality measures. In addition, a superior performance to detect the correct number of atomic columns is observed as compared to alternative detection techniques.
In conclusion, the development of a new quantitative method in this thesis has pushed quantitative electron microscopy towards a more objective interpretation. The method generalises the characterisation of nanomaterials at the atomic scale in electron microscopy in order to obtain accurate and precise structure information about a material.