High-throughput phase elucidation of polycrystalline materials using serial rotation electron diffraction

The development of new materials with specific properties is a major focus in materials science. Many crystalline materials find applications due to their diverse optical, electrical, thermal, mechanical, and magnetic properties. While innovative synthetic routes are constantly being developed, the resulting products are often polycrystalline materials with crystals too small or too complex for routine analysis using X-ray diffraction (XRD). This poses a significant bottleneck in materials development, particularly in high-throughput synthesis screening where numerous polycrystalline products are generated rapidly. XRD, while a well-established technique for phase analysis and structure determination, has limitations when dealing with polycrystalline materials containing multiple phases, ultra-low content phases (<1%), phases with similar unit cell parameters, or structures with large unit cells or low symmetries. These limitations lead to peak overlaps in PXRD patterns, making identification difficult and potentially causing interesting materials to be overlooked. Single-crystal XRD (SCXRD) requires crystals of a certain minimum size, which is often not the case in polycrystalline materials. Powder XRD (PXRD), while routinely used, suffers from peak overlap, making accurate phase identification challenging. This need for more effective techniques is amplified by the increasing use of high-throughput synthesis methods, generating many samples requiring rapid and reliable phase elucidation. Electron diffraction, owing to the significantly stronger interaction of electrons with matter compared to X-rays, offers a potential solution. Three-dimensional electron diffraction (3D ED) is analogous to SCXRD but can analyze much smaller crystals (down to 50 nm). While 3D ED techniques have advanced, the process of selecting crystals for analysis remains largely manual and time-consuming, prone to human bias, potentially leading to missed phases. Serial electron diffraction (SerialED) methods automate crystal searching and data collection, enabling high-throughput 3D ED data acquisition. The high-throughput SerialRED method, which builds on SerialED, automatically screens and analyzes hundreds of crystals, combining phase analysis and structure determination in a single workflow. Combined with hierarchical cluster analysis (HCA), SerialRED provides an objective, high-throughput approach for phase analysis and structure determination, making it suitable for high-throughput synthesis screening.

The literature highlights the challenges of phase identification in polycrystalline materials using conventional XRD techniques. Many research articles demonstrate the limitations of PXRD in resolving complex mixtures or identifying minor phases. Several studies have explored the use of electron diffraction for structure determination and phase analysis of materials that are challenging to study using XRD. The development of 3D ED techniques, including continuous-rotation electron diffraction (cRED), has been extensively reported, showing its potential for resolving complex structures. However, these methods often require manual intervention, limiting throughput. The development and application of SerialED and SerialRED methods have been previously documented, showcasing their advantages in automation and high-throughput analysis of polycrystalline materials. The application of SerialRED, in particular, has been successful in analyzing protein nanocrystals, further supporting its potential. This study builds on these advancements to demonstrate the method's effectiveness in characterizing complex zeolite materials.

This study used SerialRED for high-throughput phase identification of zeolite synthesis products. Zeolites, being typically metastable polycrystalline microporous materials with complex structures, often form multiphase mixtures making phase identification challenging. The researchers synthesized zeolites by combining multiple tetrahedrally coordinated framework T atoms (T = Si, Ge, Al, or B) and a simple organic structure-directing agent (OSDA), 4-dimethylaminopyridine (DMAP). The Si/Ge molar ratios were varied (5-15), along with (Si + Ge)/T (T = Al or B) molar ratios (5-100). PXRD was initially used to characterize the phase compositions; products not identifiable by PXRD or conventional 3D ED were further investigated using SerialRED. The most complex sample (product A: Si/Ge = 10, (Si + Ge)/Al = 15) was chosen to demonstrate SerialRED's capabilities. The SerialRED protocol, implemented in Instamatic software, automatically screened crystals and collected 3D ED data from hundreds of crystals over 6 hours, yielding 321 3D ED datasets. DIALS was used for on-the-fly unit cell determination. High-throughput phase analysis was performed using HCA in edtools, employing the Euclidean distance between unit cell parameters as the metric. Datasets within each cluster were further analyzed using intensity-based HCA to select high-correlation datasets, removing poor-quality data and outliers. These were then merged for structure analysis. Another sample (product B: Si/Ge = 5, (Si + Ge)/Al = 12.5) was used to demonstrate identifying phases with similar morphologies and unit cell parameters and to explore the potential for quantitative phase analysis. Conventional cRED was also used for comparison. PXRD patterns were analyzed using the SVD-index method in TOPAS v.6, and Pawley profile fitting was performed to refine phase compositions. Synchrotron powder X-ray diffraction (SPXRD) was used for product B to compare quantitative phase analysis results.

SerialRED successfully identified five zeolite framework types in product A: RTH, IWV, *CTH, *UOE, and POS. Three phases (RTH, IWV, and *CTH) were major components, while *UOE and POS were minor phases. The minor phases were not detectable by PXRD, highlighting SerialRED's ability to identify low-content phases. Two phases (IWV and *CTH) had similar crystal morphologies and unit cell parameters but were successfully distinguished. Structure determination was performed using SHELXT for each phase. The analysis of product B revealed a mixture of IWV and *CTH, with similar crystal size and morphology. The *CTH phase, not detected by conventional 3D ED, was identified by SerialRED, demonstrating its ability to overcome crystal selection bias and provide more robust phase information. A comparison of SerialRED results (crystal counts) with Rietveld refinement of SPXRD data (weight percent) for product B showed general agreement, suggesting SerialRED's potential for quantitative phase analysis. The detailed phase information revealed the roles of different framework T atoms (Si, Ge, Al, B) in the synthesis, showing how these atoms direct the formation of specific structural units. The synthesis of IWV and *CTH was shown to be promoted by the introduction of aluminum and germanium into the TON synthesis system. Similarly, the synthesis of SFE zeolite was promoted by the introduction of boron. This study also successfully synthesized SFE, IWV, and *CTH using simple, commercially available DMAP as the OSDA, providing economic viability for large-scale production. A catalysis test on product B (IWV and *CTH mixture) showed significantly higher catalytic efficiency in isopropylnaphthalene isomerization compared to MOR and SFE zeolites, attributed to the larger pore sizes of IWV and *CTH providing more accessible active sites.

The findings demonstrate SerialRED's significant advantages over conventional methods for phase elucidation in polycrystalline materials. Its ability to automatically analyze hundreds of crystals, overcoming human bias and limitations in crystal selection, provides a more comprehensive and objective assessment of phase composition. The identification of minor phases and phases with similar characteristics highlights the method's sensitivity and resolving power. The results validate the potential of SerialRED for quantitative phase analysis, although further investigation may be needed to refine the accuracy of quantitative results, especially when dealing with extremely similar phases. The identification of the different zeolite phases using SerialRED provided crucial insights into the role of different framework T atoms in the synthesis, enabling a more rational design of synthesis strategies for specific zeolite structures. The superior catalytic performance of product B in the isopropylnaphthalene isomerization reaction underscores the importance of accurate phase identification for understanding and optimizing catalytic properties. This information enables the design of better catalysts for industrial applications.

This study successfully demonstrated SerialRED's effectiveness in high-throughput phase identification of complex polycrystalline zeolite mixtures. The method's ability to identify minor and similar phases, coupled with its high-throughput nature, accelerates materials development and provides critical insights into synthesis-structure relationships. The potential for quantitative analysis was shown, and future research could explore optimizing the quantitative accuracy and expanding the applicability of this technique to other material systems. The integration of SerialRED into high-throughput synthesis workflows can revolutionize the discovery and development of polycrystalline materials.

While SerialRED demonstrates significant advantages, certain limitations exist. The accuracy of unit cell parameter determination from SerialRED datasets can be influenced by factors such as optical distortions and goniometer stability. Intensity-based clustering, while helpful, might be limited by the completeness of datasets, especially in cases with similar unit cell parameters. The study focused on zeolites; thus, further research is needed to evaluate the method's applicability across a wider range of materials. The quantitative accuracy of phase analysis using SerialRED remains to be further refined and validated through comparisons with various independent analytical techniques.