Knowing the atomic crystal structures of ordered porous solids is essential in understanding their behaviors and properties, developing new applications, and designing new porous materials. Electrons have a much shorter wavelength and much stronger interaction with atoms in a crystal compared with X-ray. Therefore, electron crystallography can effectively determine the structures of nano- and micro-sized crystals. Three-dimensional electron diffraction (3D ED) methods have been developed for the structure determination of various types of complex crystal structures. Continuous rotation electron diffraction (cRED) has unique aspects in both fast data collection and accurate structure determination. 

This thesis focused on the accuracy of crystal structure analysis and the potential of unraveling structural details by cRED. The cRED method was first applied for the ab initio structure determination of a beam-sensitive biocomposite metal-organic framework (MOF), BSA@ZIF-CO₃-1. The atomic structure of BSA@ZIF-CO₃-1 obtained by cRED was the same compared to that obtained by single crystal X-ray diffraction (SCXRD). Accurate atomic structures could be obtained by cRED. The sample of BSA@ZIF-CO₃-1 was initially regarded as a pure new phase, however, during the cRED data collection and processing procedure, two distinct crystal systems and unit cells were revealed. BSA@ZIF-CO₃-1  was identified as the major phase in the sample, and a new MOF, denoted ZIF-EC1, as the minor phase. ZIF-EC1 has a dense 3D framework with high N and Zn densities, which is a promising candidate for electrocatalysis. The discovery of ZIF-EC1 was followed by investigating the effects of improving 3D ED data completeness on the structural analysis. I successfully solved the structures of ZIF-EC1 from each individual dataset with the lowest completeness of 44.5% and refined to a high precession (better than 0.04 Å). Then I merged ten datasets to obtain a high data completeness, the structural model is improved, peaks appear more spherical in the electrostatic potential maps. 

The next part of this thesis was focused on unraveling structural details. By applying cRED, each non-Hydrogen atom from guest molecules can be separately localized from the difference Fourier map for two open framework germanates, SU-8 and SU-68. The atomic structure of both the framework and the guest molecules obtained by cRED is as reliable and accurate as that obtained by SCXRD. In the last part, the application of cRED into determining structures for new materials are highlighted. The structure of two new MOFs, Cd-MOF and Pb-MOF are successfully determined by cRED.