There is a wealth of materials that are beam sensitive and many of them only exist in nanometric crystals, because the growth of bigger crystals is either impossible or so complicated that it is not reasonable to spend enough time and resources to grow big crystals before knowing their potential for research or applications. This difficulty is encountered in minerals, zeolites, metal-organic frameworks or molecular crystals, including pharmaceuticals and biological crystals.

In order to study these crystals and potentially discover highly interesting materials a structure determination method that can deal with beam sensitive crystals of nanometric size is needed. The nanometric size makes them destined for electron diffraction, since electrons interact much more strongly with matter than X-rays or neutrons. In addition, for the same amount of beam damage, electron diffraction yields more information than X-rays [1].

The recently developed low-dose electron diffraction tomography (LD-EDT) [2] not only combines the advantages inherent in electron diffraction, but is also optimized for minimizing the electron dose used for the data collection. While using only minimal dose, the data quality is still high, allowing not only the solution of complex unknown structures, but also their refinement taking into account dynamical diffraction effects.

In this contribution we present several examples of crystals solved and refined by this method. The range of the crystals presented includes a synthetic oxide (Sr₅CuGe₉O₂₄), a natural mineral (bulachite) and a metal organic framework (Mn-formiate). The dynamical refinement can be successfully performed on data sets that needed less than 0.1 e^-/Ų for the entire data set.

[1] R. Henderson, Quarterly Reviews of Biophysics 28, 2 (1995), 171-193

[2] S. Kodjikian and H. Klein, Ultramicroscopy, 2019, 200, 12-19

Electron diffraction experiments and electron crystallography are experiencing a major leap in the nanocrystallography revolution [1]. Moreover, the use of the continuous rotation method (as in X-ray crystallography) for the collection of electron diffraction data is surpassing all previous data collection methods available in the past decade [2].

Unfortunately, there is no dedicated device for performing such experiments. All experiments found in the literature are done in (modified)-Transmission Electron Microscopes, as these are the only sources of electron beams. Though these devices are not optimal for performing such kind of experiments. Many factors play an important role here. In fact, scientist interested in using this technique, have even suggested on guidelines to use a (S)TEM as an electron diffractometer [3].

Therefore, there is a huge necessity that fully integrated electron diffractometers are available for the scientific community and industrial facilities. This necessity is now a reality. Here we present a novel device optimized and dedicated for electron diffraction experiments that uses the continuous rotation method. We will showcase this device and its advantages against electron microscopes that can perform ED measurements. We will show experimental evidence on the improvement of the data quality compared to data sets collected elsewhere. We will highlight the ease of use of this device.

The study of the structure of single crystals has typically been achieved with X-ray diffraction while many decades of progress and research have led to hardware improvements which have pushed the limits of X-ray diffraction. The current generation of home lab instruments allow the study of crystals down to about 1 micron in size with sources such as the FR-X, a high-power rotating anode¹.

In the quest to study even smaller samples than this, microED has become increasingly popular in recent years². As electrons interact more strongly with a crystalline sample than X-rays do, the study of samples smaller than 1 micron becomes possible and, in fact, necessary. We would like to introduce our solution for microED, the Synergy ED, along with results we have obtained using it, and efforts we have made to improve the quality of results.

1. Matsumoto, T., Yamano, A., Sato, T. et al. "What is This?" A Structure Analysis Tool for Rapid and Automated Solution of Small Molecule Structures. J Chem Crystallogr (2020).
2. Nannenga, B.L., MicroED methodology and development. Struct Dyn. (2020) 7(1).
3. Gruene, T. et al Rapid Structure Determination of Microcrystalline Molecular Compounds Using Electron Diffraction Angew. Chem. Int. Ed. (2018) 57(50): 16313–16317

Electron Diffraction (ED) as such has been around since the early days of Electron Microscopy. However, only since Transmission Electron Microscopes (TEMs) are available with accelerating powers of 200 to 300 kV and 2D detectors have become fast enough, Electron Crystallography really took off.

So far, ED could only been done in modified TEMs, resulting in challenging experiments and limited datasets, yet, structures could be obtained from samples in the range of merely tens of nanometers, that were unsolvable with either conventional or even synchrotron X-ray radiation.

For some reason, no dedicated Electron Diffractometer has been available commercially so far. Data quality would greatly benefit from a setup that focuses on the diffraction capability over imaging and allowing for faster and more complete datasets through proper 3D electron diffraction (3D-ED).

We will present a possible Electron Diffractometer design for Electron Crystallography from the point-of-view of X-ray Crystallography and indicate improvements over present TEM-based as well as X-ray instruments.

The class of transition metal phosphates (TMPs) shows a wide range of chemical compositions, variations of valence states and respective crystal structures. Among TMPs, VO(P₂O₇) and LiFePO₄ are of special interest as the only commercially used heterogeneous catalyst for the selective oxidation of butane to maleic anhydride [1] and cathode material in rechargeable batteries [2]. Due to their structural features, TMPs are considered as proton exchange-membranes in fuel cells, working in the intermediate-temperature range [2, 3]. We report on the successful ab initio structure determination of two novel titanium pyrophosphates, NH₄Ti(III)P₂O₇ and Ti(IV)P₂O₇, from X-ray powder diffraction data. Both compounds were synthesized via a new molten salt synthesis route. The low symmetry space groups P2₁/c (NH₄TiP₂O₇) and P-1 (TiP₂O₇) complicate the structure determination, making the combination of spectroscopic, diffraction, and computation techniques mandatory. In NH₄TiP₂O₇, titanium ions (Ti³⁺) occupy the TiO₆ polyhedron, coordinated by five pyrophosphate groups, one as a bi-dentate ligand. This secondary coordination causes the formation of one-dimensional six-membered ring channels with a diameter d_max of 514(2) pm, stabilized by ammonium ions. Annealing NH₄TiP₂O₇ in inert atmospheres results in the formation of the new TiP₂O₇, showing a similar framework consisting of [P₂O₇]^4- units and TiO₆ octahedra as well as an empty one-dimensional channel (d_max = 628(1) pm). The structures can be related to the high-voltage pyrophosphate cathode material Li₂FeP₂O₇ also crystallizing in P2₁/c [4]. Li₂FeP₂O₇ consists of a three-dimensional arrangement of undulating [Fe₄P₈O₃₂]_∞ layers [4] building a channel system that is occupied by Li⁺ ions. The structural relation to Li₂FeP₂O₇ implies a good proton conductivity of NH₄Ti(III)P₂O₇ and Ti(IV)P₂O₇. Both newly synthesized phosphates, NH₄Ti(III)P₂O₇ and Ti(IV)P₂O₇, show a proton conductivity based on the Grotthus mechanism. The activation energy of the proton migration of NH₄Ti(III)P₂O₇ belongs to the lowest which has been reported for this class of materials and indicates its potential application as a proton electrolyte in the intermediate temperature range. In situ X-ray diffraction study of the transformation of NH₄TiP₂O₇ to TiP₂O₇ reveals a two-step mechanism, the decomposition of ammonium ions coupled with the oxidation of Ti³⁺ to Ti⁴⁺ and subsequent structural relaxation.

Since its discovery, MgB₂ is presented as a candidate to replace conventional superconductors in several applications such as SMES and Magnetic Resonators. This possibility is based on its critical temperature (T_c), its low cost, and the possibility of forming cables by the “Powder In Tube” (PIT) method. Considerable research has been carried out to optimize the material and the cable conformation [1]. It was determined that the superconducting properties can be improved by inducing changes in the microstructure, with different synthesis methods [2], and/or the crystalline structure, by doping [3]. In this work, two different samples were prepared and characterized to find convenient methods to improve the material and the cables superconducting properties without increasing the cost.

The Mg (B_1-xC_x) ₂ samples were prepared starting from Mg and C-doped nano-B powders, with and without the addition of nano-SiC. The compounds were mixed in an agate ball mill inside a glove box. Then, they were compacted into pellets and sintered in a tube furnace with a circulating controlled Argon atmosphere. The samples were heat treated at two temperatures, 700ºC and 900ºC, for different periods. The specimens were characterized by X-ray diffraction (XRD), transmission electron microscopy (TEM) (Fig. 1), and SQUID magnetometry. XRD data were refined by the Rietveld method with FullProf [4] as can be seen in Fig. 2. Finally, 700ºC in-situ syntheses were performed on a nano-SiC doped sample at the P61A line in DESY, to study the reaction development.

With these techniques, we were able to determine the lattice parameters, the existing phase percentages, the grain sizes, the stress state, and the T_c. These data together allow for a better understanding of the synthesis parameters and the effect of doping on the phase formation as a way to improve the superconducting properties of the material.

[1] Buzea, C. Yamashita, T., (2001). Supercond. Sci. Technol., vol. 14, no. 11, p. R115, 2001.

[2] Silva, L. B. S. Da, Serquis, A., Hellstrom, E. E., & Rodrigues, D. (2020). Superconductor Science and Technology, 33(4), 45013.

[3] Serrano, G., Serquis, A., Dou, S. X., Soltanian, S., Civale, L., Maiorov, B., Holesinger, T. G., Balakirev, F., Jaime, M. (2008). J. Appl. Phys., vol. 103, no. 2, pp. 1–5, 2008.

[4] J. Rodriguez-Carvajal (1990). Abstracts of the Satellite Meeting on Powder Diffraction of the XV Congress of the IUCr, p. 127.

Ruddlesden-Popper manganites La_xSr_2-xMnO_4-δ recently gained interest as promising electrode materials for solid oxid fuels cells. For 0.25 ≤ x ≤ 0.6, their stability under reducing atmosphere –along with the preservation of their K₂NiF₄-type I4/mmm symmetry –has been demonstrated using in situ high-temperature neutron and X-ray powder diffraction. [1] However, abnormally large anisotropic displacement parameters and complex changes in cell parameters point to the presence of disorder, which might explain the material’s increased electrical conductivity in diluted hydrogen. Submicron sized crystals are sufficient for electron diffraction (ED) to obtain two-dimensional single-crystal diffraction patterns, which can be interpreted in a more straightforward way than powder data. Therefore, single-crystal ED might pick up features which were missed during X-ray and powder diffraction. Using a dedicated environmental holder in a transmission electron microscope, we performed several series of in situ ED experiments to track structure transformations of La_0.5Sr_1.5MnO₄ upon heating in a 5% H₂ /He atmosphere. As the current state-of-the-art in situ equipment only permits tilting of the holder along one axis, conventional in-zone patterns cannot be obtained, and 3D ED is the optimal method to acquire sufficient diffraction data for structure analysis. We also performed the same experiments on Sr₂MnO₄ as a reference, since this material is known to undergo a space group transformation to a monoclinic P2₁/c supercell when reduced to Sr₂MnO_3.55 [2]. For La_0.5Sr_1.5MnO₄ a coexistence of both the tetragonal Ruddlesden-Popper phase and a perovskite phase has been noted upon heating to 750°C in reducing atmosphere, which has not been reported before. However, apart from the diluted hydrogen, the electron beam might possess some reductive power too, and the high temperatures can lead to decomposition. Therefore, we systematically examined the influence of different external factors, repeating the experiment with i.a. varying beam exposures, while heating in vacuum and reducing ex situ in 5% H₂ /He.

[1] Sandoval, M., Pirovano C., Capoen, E., Jooris, R., Porcher, F., Roussel, P., Gauthier, G. (2017). Int. J. Hydrog. Energy. 42 (34), 21930-21943.[2] Broux, T., Bahout, M., Hernandez, O., Tonus, F., Paofai, S., Hansen, T., Greaves, C. (2013). Inorg.Chem. 52 (2), 1009-1017.

Keywords: TEM, in situ, 3DED, Ruddlesden-Popper manganite, LSMO

This work was supported by BOF 38689 - New method to acquire in situ information on crystal structures changed by chemical reactions.

Metallic nanoparticles have been shown to have a wide variety of applications from catalysis to plasmonics and medicinal drug delivery [1-3]. The onset of functional properties not seen in bulk can be attributed to and finely controlled by particle size, shape, homogeneity, and chemistry [4-6]. Given nanoparticles’ small size – typically sub-100 nm – their properties can be strongly influenced by crystallographic defects such as twinning, interfaces, and surfaces. As such, a fundamental study of the energetics of atomic mobility at the surface of a nanoparticle can provide an invaluable understanding of the relationship between crystal structure and functional properties. Moreover, surface energetics play a key role in controlling nanocrystal growth and shape. In this study, Au nanoparticles with different organic ligands, and therefore different surface stabilities, are investigated in real space using quantitative atomic resolution aberration corrected scanning transmission electron microscopy (STEM).

STEM imaging has become one of the leading methods of materials characterization, and aberration correction has made atomic resolution imaging widely available. Advanced imaging and image processing techniques have been developed to quantify specimen features including thickness projected along the path of the beam [7-12]. Such quantitative methods generally rely on high angle annular dark field (HAADF) STEM, whereby images are formed using electrons that have been scattered to high angle with a strong atomic number dependence (so-called Z contrast). While these techniques have been applied to great effect under specific controlled conditions, most of the beam-specimen interaction is discarded both from lower scattering angles and via the angular and azimuthal integration of HAADF detectors. Such information is rich with detail about specimen morphology and can be used to improve quantitative precision when paired with image simulation.

Recently, the development of fast, high-dynamic range, direct electron detectors has made it possible to record the scattering distribution (k_x, k_y) as a function of probe position (r_x, r_y), generally referred to as 4D-STEM [13-15]. Such detectors have noise levels well below that of single-electron strikes, making them ideal for quantitative imaging. By collecting the whole scattering distribution at each probe position rather than integrating over an annulus, images can be formed using specific regions of the diffraction pattern that are most strongly affected by variations in thickness. Additionally, several images can be formed from the same dataset using different scattering regimes to further constrain thickness measurements.

Here we report on advances in quantitative thickness determination using 4D-STEM paired with multislice simulations. A detailed comparison of the advantages and challenges of using 4D-STEM opposed to conventional HAADF-STEM will be covered. Furthermore, we apply these 4D-STEM atom counting techniques to metallic nanoparticle systems to probe the energetics of beam-induced surface atom motion. The implications of such surface energetics on nanoparticle properties will be discussed.

1. Dreaden,et al. (2012). Chem. Soc. Rev. 41, 2740-2279.
2. Sau, et al. (2010). Adv. Mater. 22, 1805-1825.
3. Zhou, et al. (2011). Chem. Soc. Rev. 40, 4167-4185.
4. Balbuena, P. and Seminario, J. (2006). Nanomaterials: Design and Simulation, Elsevier: Amsterdam.
5. Yin, Y. and Alivisatos, A.P. (2005). Nature 437, 664-670.
6. Henry (2005). Prog. Surf. Sci. 80, 92-116.
7. LeBeau, et al. (2010). Nano Lett. 10, 4405-4408.
8. Van Aert, et al. (2013). Phys. Rev. B 87, 064107.
9. De Backer, et al. (2013). Ultramicroscopy 134, 23-33.
10. Katz-Boon, et al. (2013). Ultramicroscopy 124, 61-70.
11. Katz-Boon, et al. (2011). Nano Lett. 11, 273-278.
12. Dwyer, et al. (2012). Appl. Phys. Lett. 100, 191915.
13. Tate, et al. (2016). Microsc. Microanal. 22, 237-249.
14. Ballabriga, et al. (2011). Nucl. Instrum. Methods Phys. Res. 633, S15-S18.
15. Johnson, et al. (2018). Microsc. Microanal. 24, 166-167.

The authors acknowledge the use of the instruments and scientific and technical assistance at the Monash Centre for Electron Microscopy, a Node of Microscopy Australia. This research used equipment funded by Australian Research Council grant LE0454166.

Over the past few years, macromolecular structures have been successfully solved from Electron diffraction(ED) patterns using standard macromolecular X-ray crystallographic(MX) techniques[1]. This resulted in the emergence of a new field known as microED. MicroED is a very appealing technique as it allows for solving structures from nanocrystals thanks to the very strong electron-atom interaction. This is of great interest especially in protein crystallography where growing good quality macromolecular crystals up to micrometer sizes is often a challenge.
Besides, ED patterns provide information about the electrostatic potential which is a complementary information to the electron density maps provided by X-ray diffraction patterns. This can allow for example to determine accurately the location of positively charged ions. Although it is still necessary to grow nanocrystals with microED, as opposed to the popular cryo-EM imaging technique, there are evidence that microED should provide higher resolution than cryoEM[2]. On paper, the resolution with microED is in principle sufficient to resolve hydrogen atom positions.
Although microED has proven successful on an number of occurrences, theoretical works[3] suggest that dynamical diffraction corrupt the kinematic reflection intensities to the extent that solving macromolecular structures from standard MX techniques should not be possible for crystals larger than a few nanometres thick. In practice typical nanocrystals are a at least a few tens of nanometres which is an order of magnitude above the kinematic regime. This fact is partly reflected in the large R factors commonly found in microED which is usually in the 15-20% range. However, theoretical predictions usually offer much more pessimistic figures[4]. Moreover, even when Rfactor is improved using dynamical refinement technique, ED still fail to compete with R factors produced by X-ray techniques. As a result, there is still a great motivation in studying the effect of dynamical diffraction in electron diffraction experiments.
In this work, simulations of ED patterns have been performed on organic crystals with both the multislice algorithm(MS)[5,6] and the blochwave approach[6]. The differences between the 2 methods are presented with their advantages and modelling limitations.
Comparison with experimental patterns are presented for glycine and IRELOH with a discussion about differences between theory and experiment.
[1] Nannenga, B. L., & Gonen, T. (2019). The cryo-EM method microcrystal electron diffraction (MicroED). Nature Methods, 16(May), 369–379. https://doi.org/10.1038/s41592-019-0395-x
[2] Latychevskaia, T., & Abrahams, J. P. (2019). Inelastic scattering and solvent scattering reduce dynamical diffraction in biological crystals. Acta Crystallographica Section B: Structural Science, Crystal Engineering and Materials, 75, 523–531. https://doi.org/10.1107/S2052520619009661
[3] Glaeser, R. M., & Downing, K. H. (1993). High-resolution electron crystallography of protein molecules Robert. Ultramicroscopy, 52, 478–486.
[4] Oleynikov, P., Hovmöller, S., & Zou, X. D. (2007). Precession electron diffraction: Observed and calculated intensities. Ultramicroscopy, 107(6–7), 523–533. https://doi.org/10.1016/j.ultramic.2006.04.032
[5] Cowley, J. M., & Moodie, A. F. (1957). The scattering of electrons by atoms and crystals. I. A new theoretical approach. Acta Crystallographica, 10(10), 609–619. https://doi.org/10.1107/s0365110x57002194
[6] Kirkland, E. J. (2019). Advanced Computing in Electron Microscopy (Third Edit). Springer.
[7] Bethe, H. (1928). Theorie der Beugung von Elektronen an Kristallen. Annalen Der Physik, 392(17), 55–129. https://doi.org/10.1002/andp.19283921704

A well-known bottleneck in the structural characterization of macromolecules with X-ray diffraction is crystallization. Often the needed crystal size cannot be achieved despite extensive optimization of crystallization conditions. Nevertheless, the yield of sea urchin like needle clusters, microcrystals and almost two-dimensional platelets is a silver lining. Those crystals – too small for X-ray crystallography – could be applied to microcrystal electron diffraction (MicroED) methods.

3D electron diffraction (ED) is an uprising method for the structural characterization of nanocrystalline materials. ED has recently been applied to beam-sensitive materials like macromolecular crystals. Although there are also drawbacks for this method in protein crystallography, we used the spirit of the nanocrystallography revolution and started with first experiments on our transmission electron microscope.

So far we were able to acquire knowledge for the basic workflow for data collection with our instrument setup, a Zeiss Libra 120 plus TEM with an OMEGA energy filter and a TVIPS TemCam-XF416(ES) detector. We collected continuous rotation electron diffraction data for the zeolite ZSM-5 and protein nanocrystals (lysozyme) under cryo conditions.

With these fundamental achievements we are on track to apply MicroED to more challenging crystals and also solve novel protein structures in the future.

The theory of electron scattering by crystalline materials is well-established,with two main approaches to the calculation of diffracted intensities: the Bloch-wave (or scattering matrix) method, and the Multislice method [1, 2].Both have been implemented in numerous simulation programs and compute the required large-angle convergent beam electron diffraction (LACBED )patterns. While the accuracy of these simulations can be very high [3] they are also computationally intensive and relatively slow even with modern high-performance computing facilities and graphical processing units(GPUs). In Bloch-wave simulations the limiting step is the inversion of acomplex, non-Hermitian matrix, while for multislice the use of many configurations in the frozen-phonon approximation can increase simulation timesby several orders of magnitude [1, 2]. As a result, simulation times aregenerally several minutes at best. Machine learning (ML) as a computational approach has been gaining prominence in recent years and the implementation of neural networksas universal approximators holds great promise for the solution of inverse and/or computationally difficult problems. Importantly, once trained, a ML calculation is fast – typically a few milliseconds on a GPU. In the caseof modelling electron scattering, this may allow an increase in speed of 5–6 orders of magnitude. Here, we explore the simulation of electron scattering using a variational autoencoder (VAE) [4]. The VAE takes as an input a 128×128 pixel image of the projected potential of a unit cell of cubic material in the [001] orientationand gives an output of the 000 LACBED pattern of the same size. The VAE is trained end-to-end using 5527 Felix[5] Bloch-wave simulations of cubic materials taken from the inorganic crystal structure database (ICSD).[6] The simulations were split 85:10:5 into training, validation, and test sets. Similarity was quantified using a zero-mean normalised cross correlationloss function Z.[7] The position of features in reciprocal space was fixedby choosing an angular range that varied in inverse proportion with lattice parameter. Using 500 Bloch waves,each simulation typically required∼400 seconds to complete running on acluster of 160 cores.The VAE uses convolutional encoder and decoder sub-models, both 2 layers deep, to access a 12-dimensional latent space of encoded LACBED patterns. Calculations on an Nvidia GTX 1080Ti GPU are∼3.6×10⁵ times faster than a 160-core felix simulation. In this exploratory trial,only 5527 out of the 42879 cubic materials available on the ICSD were used for model training. Even with this very limited training data set, some VAE simulations approach the accuracy of felix, while others only produce a poor approximation. We estimate that training on > 12000 simulations would produce losses Z <5%, giving a similarity equivalent to that between felix and experiment.[8]

References

[1] B. G. Mendis,Electron beam-specimen interactions and simulation meth-ods in microscopy(John Wiley & Sons, 2018).

[2] E. J. Kirkland,Advanced computing in electron microscopy(Springer, 1998).

[3] L. J. Allen, S. Findlay,et al., Ultramicroscopy 151, 11 (2015)

[4] D. P. Kingma and M. Welling, in Auto-encoding variational bayes (International Conference on Learning Representations, 2014).

[5] R. Beanland, K. Evans, R. A. R ̈omer, and A. J. M. Hubert, Felix Bloch wave simulation: Source code, 2021.

[6] A. Belsky, M. Hellenbrandt, V. Karen, and P. Luksch, V. 58. N. 3.58,364 (2002).

[7] R. Beanland, K. Smith, P. Vanˇek, H. Zhang, A. Hubert, K. Evans, R. A.R ̈omer, and S. Kamba, Acta Crystallographica Section A77, (2021).

[8] A. J. M. Hubert, R. Romer, and R. Beanland, Ultramicroscopy 198, 1(2019)

The most conclusive and elucidating component of any small or macromolecular study is the proper structure determination. The two most commonly used tool for structure determination are being nuclear magnetic resonance spectroscopy (NMR) and X-ray diffraction. While both these techniques are extremely popular but have certain limitations. Recently a new technique 3D Electron Diffraction (3D ED) data collection for getting near to atomic resolution structures has taken a leap in last few years. In this method, once the intensities are extracted, the structures are obtained from the 3D ED data using similar tools as for X-ray diffraction structure determination like SHELX, olex2, etc. In general Independent atom model (IAM) is used for solving the structures, where a precomputed model of electrostatic potential is built using scattering factors from isolated, spherically averaged atoms or ions.¹ In reality the atoms in a molecule are not isolated and spherical, moreover, the usage of improper electron scattering factors in refinement may lead to physically unrealistic values. To overcome this, an aspherical TAAM refinement has been applied both for X-ray and ED refinement which largely improved the physical representation and refinement statistics of the structure.² We have chosen a model molecule β-glycine for this study for which 3D ED data is already available.³ Spherical and Aspherical TAAM refinement seemed to be possible in MoPro with the inclusion of electron scattering factors. Aspherical electron scattering TAAM model will be constructed using ELMAM2 and MATTS databank and refinement will be performed using MoPro. A comparison will be shown between reported data and spherical and aspherical TAAM refinement using MoPro and various statistics will be presented.