Novel Methods to Increase Analytical Efficiency in X-ray Imaging

Identifying the mechanistic underpinnings of the function of materials, especially irreversible phenomena, at the level of single particles is crucial to improving their performance. While several techniques of nano-scale imaging of complex systems are available, their throughput and value are hindered by the dearth of coding packages directed to enhance experimental runtimes and data analysis. The following thesis details three pythonic codebases/modules to improve intra and post-experimental analysis for the following techniques: nano-Scanning X-Ray Diffraction Microscopy (nSXDM), Bragg Coherent Diffraction Imaging (BCDI), and nano-Computed Tomography (nCT). For nSXDM, the developed Python package allows for routine experimental analysis to be completed through a graphical user interface. We used this package to analyze LiNi0:33Mn0:33Co0:33O2 (NMC111) nano-particles during their first electrochemical cycle. For BCDI, we compared machine learning classification methods for defect identification of a single, 2D BCDI pattern. Our methods lay the groundwork for intra-experimental analysis of nanocrystals. Lastly, we compiled a machine learning suite of old and novel network implementations into a Python package called TomoSuitePY for low-dose projection and sparse-angle computed tomography (CT). Metrics of the sparse-angle network implementations were benchmarked with an experimental dataset containing LiNi0:8Mn0:1Co0:1O2 (NMC811) nano-particles. All aspects of this thesis aim to improve the analysis throughput of X-ray imaging techniques that are otherwise demanding in data volume and computation.