Compressive Sensing Methods and Applications for Electron Microscopy



Nicholls, Daniel ORCID: 0000-0003-1677-701X
(2023) Compressive Sensing Methods and Applications for Electron Microscopy. Doctor of Philosophy thesis, University of Liverpool.

[thumbnail of 200962918_Feb2023.pdf] Text
200962918_Feb2023.pdf - Author Accepted Manuscript

Download (30MB) | Preview

Abstract

Scanning Transmission Electron Microscopy (STEM) is at the forefront of high resolution material characterisation. Modern STEMs can be outfitted with a wide variety of ancillary equipment that each provides additional support alongside regular imaging capability. A proficient microscopist can use these tools to characterise a number of materials, though many prove difficult and most are incapable of being characterised with the current techniques available - biological, battery, and catalyst materials are, for example, often much too sensitive to image using standard techniques. The limiting factor for most of these materials is that they suffer from beam damage during characterisation - they are destroyed or changed under the electron beam. Many techniques have arisen to aid with imaging these materials, either by reducing the electron dose rate required to perform the necessary characterisation or increasing the material's tolerance for exposure to the electron beam. A secondary limiting factor to STEM analysis is the acquisition speed. Many transient phenomena, such as atoms moving across the surface of a material, happen at such speeds that exceed the current capabilities of STEM. Transient phenomena such as drift also act as a resolution limiter which plagues every microscope aiming to perform high resolution imaging. Compressive sensing (CS) STEM is a low-dose imaging technique that reduces the number of pixels required to form STEM images. It accomplishes this by combining subsampling, the act of deliberately forming incomplete images by manipulating the electron beam path, and inpainting, a method of infilling gaps in incomplete images. This technique, however, is still in its infancy with regards to electron imaging. The initial attempts to apply CS to scanning electron microscope modalities has done so successfully, but that success has not been explained or justified in great detail. Validation of compressive sensing for electron microscopy is vital at this stage, as while the technique has shown promise, it has not been rigorously examined. The goals of this thesis are then to develop and use compressive sensing as a low-dose STEM imaging technique, investigate and explain why subsampling and inpainting is beneficial, and apply it to various microscopy techniques with a focus on maximising the output from these techniques.

Item Type: Thesis (Doctor of Philosophy)
Divisions: Faculty of Science and Engineering > School of Engineering
Depositing User: Symplectic Admin
Date Deposited: 11 Aug 2023 15:11
Last Modified: 11 Aug 2023 15:12
DOI: 10.17638/03171305
Supervisors:
  • Browning, Nigel
  • Mehdi, Beata
URI: https://livrepository.liverpool.ac.uk/id/eprint/3171305