Automated serial rotation electron diffraction combined with cluster analysis: an efficient multi-crystal workflow for structure determination

Serial rotation electron diffraction (SerialRED) has been developed as a fully automated technique for three-dimensional electron diffraction data collection that can run autonomously without human intervention. It builds on the previously established serial electron diffraction technique, in which submicrometre-sized crystals are detected using image processing algorithms. Continuous rotation electron diffraction (cRED) data are collected on each crystal while dynamically tracking the movement of the crystal during rotation using defocused diffraction patterns and applying a set of deflector changes. A typical data collection screens up to 500 crystals per hour, and cRED data are collected from suitable crystals. A data processing pipeline is developed to process the SerialRED data sets. Hierarchical cluster analysis is implemented to group and identify the different phases present in the sample and to find the best matching data sets to be merged for subsequent structure analysis. This method has been successfully applied to a series of zeolites and a beam-sensitive metal-organic framework sample to study its capability for structure determination and refinement. Two multi-phase samples were tested to show that the individual crystal phases can be identified and their structures determined. The results show that refined structures obtained using automatically collected SerialRED data are indistinguishable from those collected manually using the cRED technique. At the same time, SerialRED has lower requirements of expertise in transmission electron microscopy and is less labor intensive, making it a promising high-throughput crystal screening and structure analysis tool.