Symposium CT02—In Situ TEM Characterization of Dynamic Processes During Materials Synthesis and Processing

Nanoporous gold obtained through gold-silver dealloying exhibits unparalleled performance in catalysis, sensing, and battery electrodes^1–3. For example, combining metal oxides with nanoporous gold enables highly reversible cathodic reactions with significantly lower overpotentials than state-of-the-art electrodes for lithium-ion batteries⁴. The functionality and performance of nanoporous gold varies strongly with the synthesis conditions¹. However, the relationship between synthesis conditions and the final product's structural and morphological properties is still poorly understood. This is because it is difficult to obtain a time-resolved analysis of the dealloying process in real-world conditions^5,6.
We performed in situ liquid-phase Scanning/Transmission Electron Microscopy (S/TEM) to obtain a time-resolved analysis of the structure and chemistry of AuAg at the atomic scale during electrochemical dealloying at variable electrochemical and hydrodynamic conditions. The results indicate that pore formation occurs through a priming step of phase separation, where silver and gold form intertwined but separated domains and dealloying then proceeds through silver dissolution. The dealloying process is surface limited, and for that reason, thinner areas de-alloy faster than thicker ones. These results suggest that the previously observed inverse correlation between electrochemical overpotential and pore sizes is caused by an initiation of the second step before phase separation is complete when the separated domains are not fully developed. These insights allow direct correlations between processing parameters and the final properties, enabling rational optimization of properties such as crystallite size, grain boundary density and edge/terrace ratio for various technologic applications.


References
1. Seker, E., Reed, M. L. & Begley, M. R. Nanoporous gold: fabrication, characterization, and applications. Materials (Basel). 2, 2188–2215 (2009).
2. Wittstock, A., Biener, J. & Bäumer, M. Nanoporous gold: a new material for catalytic and sensor applications. Phys. Chem. Chem. Phys. 12, 12919–12930 (2010).
3. Kim, S. H. Nanoporous gold: Preparation and applications to catalysis and sensors. Curr. Appl. Phys. 18, 810–818 (2018).
4. Chen, L. Y. et al. Nanoporous metal/oxide hybrid materials for rechargeable lithium–oxygen batteries. J. Mater. Chem. A 3, 3620–3626 (2015).
5. de Jonge, N. Theory of the spatial resolution of (scanning) transmission electron microscopy in liquid water or ice layers. Ultramicroscopy 187, 113–125 (2018).
6. Jungjohann, K., Evans, J., Aguiar, J., Arslan, I. & Browning, N. Atomic-Scale Imaging and Spectroscopy for In Situ Liquid Scanning Transmission Electron Microscopy. Microsc. Microanal. 18, 621–627 (2012).

Hafnia based resistive random-access memory (RRAM) devices are promising candidates as next generation non-volatile memories and are appealing because of the compatibility with back-end-of-line processes in the current semiconductor fabrication process. Thus, improving device reliability is crucial for their use in future memory applications. Understanding the influence of the microstructure and the atomistic processes involved will help enhance device performance. As shown in recent work (1), grain boundary engineering the hafnia layer in TiN/HfO₂ /Pt RRAM can yield forming free devices with threading grain boundaries that act as preferential pathways for conductive filament formation (2). The techniques that will be presented provide examples of how to correlate the local structure to the device properties. In situ experiments that involve heating as well as biasing have been carried out on devices mounted on MEMS-based chips by FIB in situ lift-out techniques (3) and are key towards providing a component specific understanding of the physical/chemical mechanisms involved. Furthermore, we have performed a temperature-dependent study to directly monitor the microstructure evolution of the hafnia layer directly inside the microscope, thus giving a unique insight into the growth mechanism and formation of grain boundaries. The transformation from an amorphous/nanocrystalline to a polycrystalline state lowered forming voltages and device-to-device variability of the memory devices and served as a basis for a direct structure-property correlation. To further analyse the complex texture transfer mechanisms involved, phase-determination 4D-STEM experiments were carried out to locally correlate HfO₂ grain orientations with the underlying TiN electrode. A Machine Learning approach was implemented for analysing the generated 4D-STEM data sets (4).
References:
1. S. Petzold et al., Advanced Electronic Materials. 5, 1900484 (2019).
2. M. Lanza et al., Appl. Phys. Lett. 100, 123508 (2012).
3. A. Zintler et al., Ultramicroscopy. 181, 144–149 (2017).
4. A. Zintler et al., Microsc Microanal, 1–3 (2020).

To date, visualizing the friction process between nanocontacts at an atomic scale is rarely accomplished. Here, through designing the nanocontact and performing controlled motion between asperities under high-resolution transmission electron microscopy (HRTEM), the real-time atomic-friction process is captured. Moreover, the interface dynamics and friction mechanism are illustrated by combining in-situ TEM observation with molecular dynamics simulation. For the first time under in-situ TEM, the atomic-friction between single-asperity nanocontacts is revealed to display a discrete stick-slip behavior and an asynchronous process for the accumulation and dissipation of the strain energy together with the nonuniform motion of interface atoms. This work provides a study approach to realize in-situ atomic-friction research and attains a fundamental understanding of friction phenomena at the atomic-scale.

The synthesis of bimetallic CoPt nanoparticles aims to optimize the use of expensive platinum for critical chemical reactions such as the oxygen reduction reaction (ORR). The crystallographic configuration of the intermetallic solution is directly related to its catalytic performance. For instance, the valence state influences the platinum oxygen bond strength, so it is necessary to know and control the crystal's electronic configuration at the atomic level to enhance catalytic properties. Hence, it is crucial to have a precise knowledge of the electronic configuration of Co and Pt and the atomic level to guide the future design of more efficient and active nanostructures.

To that end, we studied carbon-supported CoPt nanoparticles under reaction conditions with in situ XFM and in situ EELS. The sample was enclosed in an X-ray and electron-transparent closed-cell reactor, capable of providing both elevated temperatures and exposure to O2 and H2. The 1 μm XFM beam is a bulk approach that provides spectroscopic information over a large set of nanoparticles. In contrast, the EELS was performed in scanning transmission electron microscopy (STEM) mode with a probe size lower than 1 Å, allowing us to characterize the sample at the atomic level. This combination of approaches allows us to determine the chemical properties of individual particles and the overall system.

Results showed the progressive change of the oxidation state of cobalt upon reduction or oxidation. The valence state of cobalt could be determined, and we observe that it is spatially heterogeneous. Particle migration and coalescence was observed for a temperature above 400 °C and we determined the temperature where the reduction of cobalt is triggered. We also identified localized segregation of elements within the bimetallic structure, providing crucial information about the nanocatalysts' degradation mechanisms. This combined in situ approach can be applied to a large range of nanostructures and opens new opportunities to understand catalysts under reaction conditions.

We discuss challenges and possibilities using energy dispersive X-ray analysis (EDS) in in-situ/operando experiments. EDS in TEM/STEM, SEM and T-SEM (electron transparent specimen in SEM) is a powerful and fast technique and well suitable for analyzing the whereabouts of particular species, be that static or during processes in liquids or gases, the application of force, changing temperature and involving specimen beam interaction. The time scale for useful EDS results lies in the range of minutes to seconds, depending mainly on spatial resolution (a 2D map or a spectrum of just one atom [1]) and characteristic X-ray yield. The latter again has various dependencies, which include probe current, radiation damage, collection angle, fluorescence yield and detector efficiency.
The dedicated sample holders and reaction cell designs, which are used for in situ experiments, often present a challenge for EDS analysis. It is important to understand how the holder design influences the spectroscopic efficiency and quantitative result. We discuss effects, such as spurious signals e.g. from holder materials, absorption e.g. in reaction cell windows, the influence of heat radiation on EDS spectra [2], specimen drift etc. Approaches and tools available so far to keep such effects under control and correct for respective analysis errors are shown. One simple step to take for example, is optimizing the geometry of the holder and EDS detector position. Adjusting the materials of holders and reaction cell windows to increase EDS efficiency and avoid parasitic signals would be a more complex task.
Furthermore, a stream of data from a changing specimen needs specific analysis strategies. Suitable data acquisition and processing tools, not only for EDS, but including complementary techniques, such as diffraction-based analysis options in SEM, are being developed.
[1] R. M. Stroud et al. Appl. Phys. Lett. 108 (2016) 163101
[2] T. T. van Omme et al., Ultramicroscopy 129 (2018) 14

Innovations in liquid-phase electron microscopy (LP-EM) are discussed that have enabled experiments at the optimized conditions needed to examine soft matter [1]. Soft matter is referred to as a state of condensed matter that undergoes dynamic changes already at low energy, comparable to kT. Examples are polymers, weakly bound self-assembled structures, and biological systems. The main difficulty in such experiments is the occurrence of electron beam radiation effects, and the resulting complex experimental design and data interpretation that is inevitably influenced by those effects. By at least partially resolving the involved experimental difficulties, LP-EM is now capable of providing nanometer spatial resolution and sub-second temporal resolution for quantitative microscopy of soft matter in materials science and biology. This presentation will provide an overview of the different LP-EM systems available including graphene liquid cells, explain how the resolution does not depend much on the electron optical system but rather on the type of material, the available electron dose, and the sample thickness [2, 3]. Strategies for designing experiments are described as well including the usage of quantum dots [4]. An exciting new discovery is the ability to mitigate radiation damage so that the experiments can be carried out at much higher doses than what was previously expected [5]. Two examples will be given of the application of LP-EM, one in cancer research involving the epidermal growth factor receptor in breast cancer [6], and one in the observation of quasi-crystals self-assembled in liquid from gold nanoparticles [7]. Finally, a perspective of directly imaging dynamic soft matter processes is given including a discussion on sparse imaging and artificial intelligence.

References
[1] H. Wu, H. Friedrich, J.P. Patterson, N. Sommerdijk, N. de Jonge, Adv. Mater. 32, 2001582 (2020).
[2] N. de Jonge, Ultramicroscopy 187, 113-125 (2018).
[3] N. de Jonge, L. Houben, R.E. Dunin-Borkowski, F.M. Ross, Nat. Rev. Mater. 4, 61 (2019).
[4] D.B. Peckys, C. Quint, N. Jonge, Nano Lett. 20, 7948 (2020).
[5] S. Keskin, N. de Jonge, Nano Lett. 18, 7435 (2018).
[6] D.B. Peckys, U. Korf, S. Wiemann, N. de Jonge, Mol. Biol. Cell 28, 3193 (2017).
[7] E. Cepeda-Perez, D. Doblas, T. Kraus, N. de Jonge, Sci. Adv. 6, 1404 (2020).

Layered two-dimensional (2D) black phosphorus (BP) exhibits novel semiconducting properties including a tunable bandgap and high electron mobility. However, the poor stability of BP in ambient environment severely limits potential for application in future electronic and optoelectronic devices. While passivation or
encapsulation of BP using inert materials/polymers has emerged as a plausible solution, a detailed fundamental understanding of BP’s reaction with oxygen is imperative to rationally advance its use in applications. Here, we use in situ environmental transmission electron microscopy to elucidate atomistic
structural changes in mechanically exfoliated few-layered BP during exposure to varying partial pressures of oxygen. An amorphous oxide layer is seen on the actively etching BP edges, and the thickness of this layer increases with increasing oxygen partial pressure, indicating that oxidation proceeds via initial formation of
amorphous PxOy species which sublime to result in the etching of the BP crystal. We observe that while fewlayered BP is stable under the 80 kV electron beam (e-beam) in vacuum, the lattice oxidizes and degrades at room temperature in the presence of oxygen only in the region under the e-beam. The oxidative etch rate also increases with increasing e-beam dosage, suggesting the presence of an energy barrier for the oxidation reaction. Preferential oxidative etching along the [0 0 1] and [0 0 1] crystallographic directions is observed, in good agreement with density functional theory calculations showing favorable thermodynamic stability of the oxidized BP (0 0 1) planes compared to the (1 0 0) planes. We expect the atomistic insights and fundamental understanding obtained here to aid in the development of novel approaches to integrate BP in future applications.
References:
Naclerio et al. ACS Appl. Mater. Interfaces 2020, 12, 13, 15844-15854, DOI:10.1021/acsami.9b21116

Processing pure metals by employing severe shear deformation can result in introduction of defects and thereby microstructural refinement. However, the atomic scale mechanisms of defect evolution and microstructural refinement during extreme shear deformation of pure metals is still at its infancy. Here, we report our observations of dynamic shear-induced low-angle grain boundary(GB) formation processes in nanocrystalline Au via in-situ shear deformation inside a high-resolution transmission electron microscopy. It is found that nanotwins nucleate from free surfaces and propagate inwards. This is accompanied by dislocations glide as well. The alignment of the gliding dislocations leads to the formation of low-angle GB. The collaborative migration of nanotwins and dislocations result in the material being able to accommodate large plastic strain during shear deformation and fully recover under reversed loading with negligible damage accumulation. This nanotwin-mediated dislocation slip mechanism offers new insights for understanding low-angle GB formation during shear deformation of face centered cubic metals.

Using controlled heat to control the structure of materials during processing is the most common in-situ transmission electron microscope (TEM) technique, as temperature causes atomic diffusion and thus allows structural re-ordering. Combining this with the high-resolution imaging and spectroscopy ability of the TEM, these experiments can provide detailed insights into what mechanism governs the relation between materials processing and structure.

Thin-film technology and its application to substrates for in-situ TEM has allowed heating systems with very localized heating of the specimen in TEM heating holders, enabling more stable imaging during in-situ TEM heating experiments. More recently, direct temperature measurements on these thin-film heating systems have made temperature measurements at the sample become closer to becoming a reality. Although double-tilt TEM heating holders have been available, they suffer from some of the same limitations as standard double-tilt TEM holders do, often to a greater extent because of the added electrical connections to operate the on-chip heater and sensor. Specifically, backlash in the tilting mechanism and lack of repeatability of the tilt makes it difficult to get the sample in exactly the right orientation and to know the exact angle of rotation. This work will use an optimized single and double-tilt in-situ TEM thin-film heating platform that minimizes mechanical artifacts in the tilting mechanism and provides very stable image performance when heated to temperatures up to 1000°C, where it can be run for more than 150 hours. Outside of standard temperature characterization of these thin-film heating TEM systems using different high resolution optical thermal imaging techniques, we will also report on experiments using melting standards to confirm the on-chip temperature sensor's accurate temperature response. We can identify very local and discrete melting events of samples with known melting points and correlate this to the sensor response, where local melting happens at very discrete moments (typically within a few frames). Surface diffusion occurs in matters of seconds following that. This results in in-situ TEM heating specimen holders that allow high-temperature imaging combined with double-tilt sample position control and accurate temperature indication.

This work will present how we used these new in-situ TEM sample heating systems to characterize two nanomaterial systems. Starting with gold nanorods to characterize particle motion and stability under the beam and then the phase transformation and (rapid) shape change at the melting point. We will present the rapid frame-to-frame phase transformation process of gold nanorods as we ramp the sample temperature to 1000°C and show that we can use the system to track phenomena in nanostructures over the full temperature range of the TEM heating system with good image stability.

We also show experimental data on few-layer 2D MoS₂ samples that were transferred onto the open viewing area of the heating chip. Depending on the heating rate, heating the MoS₂ sample in-situ resulted in either ordered crystalline hexagonal islands (<20 nm) that consist of a mixture of 2H and 3R phases or yielded a mix of nanocrystalline and amorphous regions [1]. Energy-dispersive X-ray spectroscopy (EDS) confirmed the chemical content of each segregated area.

These results illustrate this newly developed in-situ TEM double-tilt heating platform's optimized overall mechanical stability and temperature performance.

Reference:
[1] Pawan Kumar et al., npj 2D Materials and Applications 4,16 (2020).

In situ (scanning) transmission electron microscopy [(S)TEM] in a gas atmosphere has proven to be an outstanding technique for investigating dynamic processes in material structures under high resolution conditions. New innovations in-situ gas handling systems and in-situ software have enabled researchers to re-create complex environmental conditions inside the TEM by precisely controlling temperature, gas pressure, flow, and the introduction of gas and vapor mixtures to the sample during the course of the experiment. These advances have made it possible to investigate dynamic catalytic processes, such as the degradation catalytic material, one example being the strong metal support interaction (SMSI). These studies showed in real-time that an amorphous reduced titania layer is formed at low temperatures, and that the crystallization of the layer into either mono- or bilayer structures is dictated by the reaction environment and in accordance with the theory. Investigation these processes under high resolution conditions further allowed the correlation between the formation of the SMSI and a dramatic reshaping of the metallic surface facets.
Another field were in situ research can help to overcome the current limitations is in the field of semiconductor research. Compound semiconductors are promising candidates for optoelectronic applications on Si. Ga(N,As,P) for example could be capable of overcoming the efficiency limitations of gallium indium arsenide phosphide ((Ga,In)(As,P)) based structures. Unfortunately, the realization of good quality epitaxial layers of highly metastable materials is indeed very challenging. A successful approach to improve the layer quality and therefore increase the optical output of these structures is post-growth annealing. A robust mechanistic understanding cluster formation, desorption processes, and material distribution within the Ga(N,As,P) and Ga(P,Bi) layer due to thermal treatment is desirable, but replicating these experiments under in-situ TEM conditions is complicated due to the handling requirements of the necessary precursor gases. To overcome the issues handling toxic and pyrophoric group III and group V precursor gases, it was necessary to develop an in situ system which fulfills all of the technical- and safety-requirements of a modern MOVPE machine, the results of which will improve our understanding of the factors necessary to optimize the efficiency of semiconductor materials.

1.Zhang, S. et al. Dynamical Observation and Detailed Description of Catalysts under Strong Metal–Support Interaction. Nano Lett. 16, 4528–4534 (2016).

2.Straubinger, R., Beyer, A., Ochs, T., Stolz, W. & Volz, K. In Situ Thermal Annealing Transmission Electron Microscopy (TEM) Investigation of III/V Semiconductor Heterostructures Using a Setup for Safe Usage of Toxic and Pyrophoric Gases. Microsc. Microanal. 23, 751–757 (2017).

3.Straubinger, R. et al. Thermally Introduced Bismuth Clustering in Ga(P,Bi) Layers under Group V Stabilised Conditions Investigated by Atomic Resolution In Situ (S)TEM. Scientific Reports 8, 1–7 (2018).

Polymorphs widely exist in nature and synthetic systems and are well known to determine material properties. Understanding phase transformation mechanisms among polymorphs enables the design of structures and tuning of phases to tailor material properties. However, current understanding is limited due to the lack of direct observations of the structural evolution at the atomic scale. Here, integrating (semi) in situ transmission electron microscopy and density functional theory, we report atomic structural evolutions of phase transformation from anatase (A) to rutile (R), brookite (B), R-phase, and TiO. Understanding the atomic structural evolution sheds light on interpreting and controlling TiO₂ polymorphs and interface structures for various applications. Here, we reveal the transitional atomic gradient structures, that form during phase transformation processes, alter electronic structures in 3D across the bulk of the crystals and thus substantially increase the active volume to separate electrons and holes and the resulting photoactivity. These new insights suggest that interphase matter based on gradient structures can be designed to induce new functions not achievable using abrupt interfaces. These findings enable a new materials design paradigm that can potentially be harnessed for a broad range of applications.
We also reveal the anisotropic nature of the electron-beam effect: dependence of crystallographic orientation with respect to electron-beam irradiation direction. The revealed electron-beam effects in our work provide guidance for in situ transmission electron microscopy studies.

Strain offers the possibility to tune the structure and properties of materials. The dimensionality of the material structures determines to what degree they can be strained prior to fracture. While bulk semiconductors fracture at 0.5-1.5 % strain and corresponding nanowires fracture at about 3-5 %, 2D structures can withstand up to 10% strain before fracture. This presentation will illustrate how in situ electron microscopy can reveal site specific information about the correlation between strain and catalytic activity, electrical, optical and thermal properties. We also show how strain can be introduced in nanostructures during their growth in solution and thereby tune the properties.

We have used high spatial resolution and precision using scanning transmission electron microscopy (STEM) to image the influence of atomic site-specific strain on catalytic activity of supported nanoparticles [1]. The data includes high precision information about the position of individual atoms as well as atom columns with varying number of atoms. We have also used in situ STEM to correlate nanoscale strain to electrical transport properties in III-V semiconducting nanowires [2]. The study shows an inhomogeneous strain distribution within nanowires where a uniaxial strain is applied. We have mapped the energy band gap and bulk plasmon as a function of tensile and compressive strain [3] in III-V nanowires. The work illustrates the potential of using combined in situ electrical and mechanical experiments where we have developed methods for in situ experiments and illumination with light for both scanning electron microscopes [4] and TEMs. We have performed in situ measurements of the resistance and thermal handling capabilities of wrinkled reduced graphene oxide [5].

[1] T. Nilsson Pingel, M. Jørgensen, A.B. Yankovich, H. Grönbeck and E. Olsson, “Influence of atomic site specific
strain on catalytic activity of supported nanoparticles”, Nature Communications 9, 2722 (2018).
[2] L. Zeng, C. Gammer, B. Ozdol, T. Nordqvist, J. Nygård, P. Krogstrup, A.M. Minor, W. Jäger and E. Olsson, “Correlation between electrical transport and nanoscale strain in InAs/In_0.6Ga_0.4As core-shell nanowires”, Nano Lett. 18, 4949 (2018).
[3] L. Zeng, T. Kanne, J. Nygård, P. Krogstrup, W. Jäger and E. Olsson, “The effect of bending deformation on charge transport and electron effective mass of p-doped GaAs nanowires”, Phys. Status Solidi RRL 13, 1900134 (2019).
[4] J. Holmér, L. Zeng, T. Kanne, P. Krogstrup, J. Nygård, L. De Knoop and E. Olsson, “An STM-SEM setup for characterizing photon and electron induced effects in single photovoltaic nanowires”, Nano Energy 53, 175 (2018).
[5] H.M. Nilsson, L. de Knoop, J. Cuming and E. Olsson, “Localized resistance measurements of wrinkled reduced graphene oxide using in situ transmission electron microscopy”, Carbon 113, 340 (2017).

With power conversion efficiencies surpassing 25% in 2019, hybrid organic-inorganic lead-halide perovskites solar cells attract ever more interest in the photovoltaics community. Photo-induced phase separation has been reported for a range of mixed-halide perovskites, which limits the available band-gap energies for photovoltaic applications. An enhanced understanding of the phase separation mechanism is essential to rationalize limitations and design stable perovskite semiconductors. During electron microscope experiments, the electron beam may cause, in the same way as photons when using a laser beam for irradiation, phase transformations in halide perovskites. In the present work, we report about the in-situ monitoring of electron-beam-induced phase segregation in CsPb(Br,I)₃ crystallites by means of high-resolution imaging in a transmission electron microscope. The acquired time series were evaluated with multivariate analysis to classify the structural change over time. An algorithm “scans” the TEM image creating a diffractogram for each patch. The diffractogram patterns of all patches of the whole time series are collected into a two-dimensional image stack, which was then classified using principal component and the independent component analysis blind source separation. By this approach, it was possible to directly monitor and provide an atomistic picture of the in-grain phase segregation. The authors will provide insight into the various aspects of this in-situ approach for the study of nanoparticles.

Solid-state batteries are considered to be one of the most promising battery configurations for future energy storage with desirable properties including safety, capacity, and longevity. However, recent studies found that interfaces and grain boundaries in solid state batteries, regardless of the type of solid electrolyte, are the primary feature limiting their commercialization. One major advantage of solid-state batteries over conventional liquid electrolyte-based batteries was expected to be reduced dendrite growth, however, dendritic growth was recently observed and was revealed to be associated to grain boundaries. Interfacial resistances between electrolyte and electrodes are higher than projected and the majority of solid electrolytes are predicted to be unstable to lithium metal. Mitigating these issues can’t be achieved without a fundamental understanding of the atomic and electronic structures of these interfaces and, more importantly, how they evolve upon electrochemical cycling. Here, we highlight several recent in situ microscopy studies that unveil the interface stability of model solid electrolytes, e.g. LiPON and (Li₇La₃Zr₂O₁₂ )LLZO, with lithium metal and compare their responses to different electrochemical cycling conditions. The role of grain boundaries in dendrite nucleation and propagation within LLZO is investigated by combining monochromated EELS with in situ microscopy imaging. The root cause of dendrite network formation in polycrystalline solid electrolytes will be discussed. We further provide a perspective as to how these issues, i.e. resistivity and dendrite growth at grain boundaries, could possibly be mitigated by controlling critical steps during synthesis.

Acknowledgement
Research sponsored by Oak Ridge National Laboratory’s Center for Nanophase Materials Sciences (CNMS), which is a U.S. Department of Energy Office of Science User Facility.

The critical area for understanding and progressing battery technology is the interface between the electrode and the electrolyte. The electrochemistry occurring at this interface, including the formation and evolution of an intermediate solid-electrolyte interphase (SEI) layer, is notoriously complex. Applying in-situ characterisation techniques to illuminate these dynamics will help with the diagnosis of the specific interfacial processes, and thus facilitate the design of better electrolytes. In this talk I will discuss some of our recent work using in-situ liquid-cell TEM to probe the evolving electrode interface in real time.[1]

Multivalent electrolytes, using chemistries based on Ca or Mg ions, are promising candidates for next-generation batteries. Recent breakthroughs are beginning to overcome the longstanding problem of devising effective multivalent electrolytes, yet a major challenge remains in designing electrolytes that yield a stable SEI that does not inhibit cycling. We performed extensive in-situ TEM characterisation of one of the more promising new electrolytes, a calcium borohydride salt in tetrahydrofuran (THF),[2] capturing the real-time nucleation, growth, and dissolution of calcium under various current density conditions. In-situ TEM imaging demonstrates the existence of a critical current density, beyond which adverse dendritic calcium plating morphologies dominate, and yielding detached isolated calcium deposits on stripping.[3]

Several other next-generation batteries are based on Li ion chemistries, and to attain peak performance desire a lithium metal anode. Yet ensuring that they remain stable over many cycles has proven challenging. One of the most effective avenues to combat this has been tailoring the SEI by adjusting the electrolyte composition with additives. These fluoride-rich interphases significantly improve cycling efficiency, yet diagnosing how they alter the structural dynamics of Li electroplating and stripping is difficult without in-situ imaging. I will discuss how using in-situ TEM can provide unique insights into the distinct morphological changes that occur to Li plated from electrolytes tailored to form a fluoride-rich interphase. Our observations reveal that the fluoride-rich SEI favours the formation of a densely interwoven Li deposit, as opposed to the more dendritic structures formed from standard electrolyte. This denser structure proved more amenable to uniform dissolution, leaving behind fewer isolated dead Li fragments, and thus yielding superior efficiency.

[1] Pu, S.; Gong, C.; Robertson, A. W. Liquid Cell Transmission Electron Microscopy and its Applications. Royal Society Open Science, 2020, 7, 191204.
[2] Wang, D.; Gao, X.; Chen, Y.; Jin, L.; Kuss, C.; Bruce, P. G. Plating and Stripping Calcium in Organic Electrolyte. Nature Materials, 2018, 17, 16-20.
[3] Pu, S.; Gong, C.; Gao, X.; Ning, Z.; Yang, S.; Marie, J. J.; Liu, B.; House, R. A.; Hartley, G. O.; Luo, J.; Bruce, P. G.; Robertson, A. W. Current-Density-Dependent Electroplating in Ca Electrolytes; From Globules to Dendrites. ACS Energy Letters, 2020, 5, 2283-2290.

Liquid cell transmission electron microscopy provides exciting opportunities for imaging processes in liquid with good spatial and temporal resolution. The quality and quantification of liquid cell data continue to improve, driven by advances in liquid cell equipment, control of local conditions, and theoretical understanding of electron beam effects. Liquid cell studies that combine imaging with electrochemical measurements have proved particularly useful in studies of energy materials, corrosion processes and crystal growth. The ability to control the temperature is important in extending the capabilities of liquid cell electrochemistry in probing these materials reactions. The electrode material is also important as it determines the reliability and consistency of the electrochemical measurements and can also affect the image resolution. Here, we explore the opportunities for liquid cell electrochemistry that arise from the use of heating as well as new electrode materials, specifically graphene and other two-dimensional (2D) materials. Heating is implemented by patterning electrodes over microfabricated liquid cell chips that already include a Pt resistance heater. 2D electrodes are implemented by transfer onto liquid cells that already include electrical contacts. We will describe the performance of combined heating and biasing chips and evaluate several 2D materials for their properties as electrodes in liquid cell electrochemistry, in comparison with conventional electrode choices such as Pt, Au or carbon. The minimal electron scattering from the 2D layer is especially helpful for preserving image resolution in electrochemical experiments carried out in plan view, where the imaging takes place through the electrode. Of the 2D materials we have tested we will focus on graphene as it appears to be particularly promising. We discuss its mechanical and electrochemical properties within the liquid cell, as well as the negligible voltage drop expected even in large area electrodes tens of micrometers across. We also discuss its stability under electron beam irradiation, even at the higher accelerating voltages often used in liquid cell electron microscopy. To demonstrate the capabilities of heating and graphene electrodes we deposit various metals from acidified electrolytes. Heating changes the kinetics, increasing the deposition rate at regions of elevated temperature. For graphene electrodes, the improved resolution allows us to explore nucleation and growth processes and demonstrate an unexpected coarsening process that occurs during deposition. We finally consider some future opportunities that will be enabled by the use of new chip designs in liquid cell electrochemistry. Particularly exciting are the possibilities of incorporating heating with novel electrode designs, and the performance of 2D materials that show favorable scavenging properties. The use of stacked 2D materials, as well as 2D layers patterned to control nucleation sites, will expand the opportunities further. We anticipate that both temperature control and incorporation of 2D materials into liquid cell design will open up new opportunities for investigating a wide range of problems relevant to energy storage and electrocatalysis as well as fundamentals of electrochemical reactions.

Al₂O₃-supported Pt/Pd bimetallic catalysts were studied using in-situ atmospheric pressure and ex-situ transmission electron microscopy. Significant metal vaporization was observed at temperatures above 600°C, both in pure oxygen and in air. This behavior implies that material transport through the vapor during typical catalyst aging processes for oxidation can play a more significant role in catalyst structural evolution than previously thought. Concomitantly, Pd diffusion away from metallic nanoparticles on the surface of Al₂O₃ can also contribute to the disappearance of metal particles. Electron micrographs from in-situ oxidation experiments were mined for data, including particle number, size, and aspect ratio using machine learning image segmentation. Under oxidizing conditions, we observe not only a decrease in the number of metal particles and an increase in average size but also a decrease in the surface area to volume ratio. Some of the metal that had dissolved into the support can be regenerated and reappear back on the catalyst support surface under reducing conditions. Real-time observation of particles during separate oxidation and reduction processes provides nanometer-scale structural details of supported Pt/Pd particles at intermediate states not observable through typical ex-situ experiments. These observations represent a first step towards understanding how rapid cycling between oxidative and reductive catalytic operating conditions affects catalyst structure.

I will discuss my group’s recent progress on capturing and understanding the intermediates and fluctuation dynamics at the nanometer resolution using liquid-phase TEM. The specific systems concern the nucleation and growth pathways of nanosized colloids into superlattices and moving membrane proteins in their native lipid and liquid environment. We find that prenucleation precursors and a layer-by-layer growth mode exist universally in diverse nanoparticle shapes. Single particle tracking, machine learning and simulations combined unravel the energetic and kinetic characteristics generic to the unexplored nanoscale, enabling advanced crystal engineering. For membrane proteins, we find that they exhibit real-time “fingering” fluctuations, which we attribute to dynamic rearrangement of lipid molecules wrapping the proteins. The conformational coordinates of protein transformation obtained from the real-space movies are used as inputs in our molecular dynamics simulations, to verify the driving force underpinning the function-relevant fluctuation dynamics. This platform invites an emergent theme of structural biophysics as we foresee.

We use liquid-phase transmission electron microscopy (TEM) to study the self-assembly of charged gold nanocubes, which showed a unique reconfigurable superlattice. The lattice can adopt two different states with same energy but different orientations, and our analysis showed the size-dependence of this reconfiguration process. Combining interaction modeling with simulation, we revealed the entropic nature of reconfiguration which dampens reconfiguration as the crystal size increases. We expect this tunable reconfiguration behavior to shed light on the design of responsive mechanical metamaterial.

The self-assembly structure of a general amphiphilic block copolymer is carried out by direct dissolution into a selective solvent or through the gradual exchange from common solvents to selective solvent for one block. In this process, the self-assembled structure was either dynamically equilibrated in a solvent to have the lowest free energy or kinetically trapped to exhibit a uniform morphology. In the case of kinetically trapped assemblies, morphological reconstruction and dynamic deformation occurred, implying the possible Brownian motion-based particle growth by collision and fusion. The self-assembly accompanying the dynamic nanoscale process of these organic molecules has not been directly observed, and basic mechanisms such as fusion, fragmentation, and growth are still subject to much research. The block copolymer self-assembly used in this study is a crew-cut micelle, not a star-shape micelle, and it can be confirmed that the volume fraction due to the ratio of the block length is controlled and the self-assembly structure transitions. Here, using the amphiphilic block copolymer, the self-assembly behavior of nanoparticles in real-time according to the molecular weight of the block in an aqueous solution was to be observed using a transmission electron microscope(TEM) and the dynamic behavior of the nanostructure was quantitatively analyzed by multiple object tracking analysis. This study provides a useful strategy for observing mechanisms of amphiphilic block copolymer self-assembly and the growth behavior of organic molecules based on Brownian motion.

Our goal is to clarify what happens immediately before and after a nucleation event and how the nucleation pathway is determined. In the last decade, in situ observations of the crystallization processes in a solution have been performed for several materials and non-classical nucleation pathways have been reported based on a liquid-cell transmission electron microscope (TEM). Examples include oriented attachment growth of nanocrystals [1], and two-step nucleation of calcium carbonate (CaCO3) and lysozyme protein crystals from amorphous/dense liquid particles [2, 3]. The unidirectional formation and growth of particles triggered by electron irradiation has also been observed. However, it is still difficult to directly observe crystals that follows the classical nucleation: grow beyond a certain large critical nucleus after fluctuations due to the competition between attachment and detachment of growth units. The difficulty is originating from their nanometric scale, rapidity, stochastic nature, and less ability to control the solution environment. We may be able to observe a crystallization process in lower magnification condition.
Then, unfortunately, crystalline nuclei cannot be observed separately as well as a conventional optical microscope.
To overcome the difficulties of direct observation, we have tried several approaches based on machine learning. One is a real-time prediction/detection at the moments of nucleation and the other one is enhancing the contrast of live images under very weak electron intensity. Both have been achieved based on machine learning.
Real-time prediction allows us to notify a nucleus before growth and, therefore, enable to observe and magnify it from a very beginning of nucleation. Enhances the contrast of TEM images allow us to magnify the
nucleus at a constant electron intensity. This does not eliminate the
effect of the beam completely, but it does minimize the change of a solution condition at higher magnification conditions. These have been evaluated by several nucleation and growth experiments. In the presentation, I will introduce more about our recent challenges and resent result of the early stages of crystallization.

[1] D. Li et al., Science 336 (2012) 1014.
[2] M. H. Nielsen et al., Science 345 (2014) 1158.
[3] T. Yamazaki et al. PNAS 114 (2017) 2154.

Crystallization by particle attachment (CPA), which is a common mechanism of colloidal crystallization resulting in hierarchical morphologies, has been both exploited to create nanomaterials with unique, emergent properties and implicated in the development of complex mineral textures. Oriented attachment (OA), a form of CPA in which crystalline primary particles align and attach along specific crystallographic directions, produces structures — typically referred to as mesocrystals — that diffract like single crystals, even though the constituent particle domains are still discernable. While the existence of mesocrystals has been well documented in a wide range of crystal systems and individual particle attachment events have been directly visualized, the mechanism by which these seemingly random events lead to well-defined, self-similar morphologies remains a mystery, as does the role of organic ligands, which are ubiquitous in nanoparticle systems. Combining in situ liquid phase TEM at 80°C with “freeze-and-look” TEM using indexed grids, we tracked formation of hematite (Hm) mesocrystals in the presence of oxalate and interpreted the results using classical density functional theory. The results show that formation of isolated Hm particles rarely occurs. However, once formed, interfacial gradients created by hematite-bound oxalate drive new hematite particles to repeatedly nucleate about 2 nm away from the new interface and then immediately undergo OA. We suspect our discovery reveals a widespread phenomenon with significant implications both for synthesis of materials and for understanding mineralization in natural environments. In addition, our discovery on the phenomenon close to solid interface highlights the complexity of material/solution interface, which lays the grounds for future research on the chemical and structural nature of the material/solution interface.

Phase change materials (PCMs) are semi-conducting alloys with distinct optical and electrical properties in the amorphous and crystalline phases that make them useful for memory applications. In memory devices, amorphous bits are crystallized in nanoseconds by either laser or Joule heating, but the amorphous phase must also be stable against crystallization for long-term data retention. Crystal growth rates relevant to memory devices span orders of magnitude and fundamental questions regarding PCM crystallization mechanisms remain open, partly due to the difficulty in measuring crystallization kinetics in certain temperature regimes.

Ag-In-doped Sb-Te (AIST) alloys are an important class of PCMs that have been widely studied. Experiments using time-resolved reflectivity [1] and ultra-fast differential scanning calorimetry [2] have broadened the temperature range for which the crystallization kinetics of AIST have been characterized, but these indirect methods of measuring crystal growth have not yet resolved the kinetic behavior just above the glass transition temperature, T_g, which is of especial interest in the characterization of glass-forming materials. Direct methods of measuring crystal growth using microscopic methods which could more completely characterize crystallization behavior have emerged in recent years.

In this presentation, the application of multiple imaging techniques used to directly quantify crystal growth rates in PCMs over a broad range of temperatures will be described, with a focus on recent experiments on an AIST alloy with a nominal composition of Ag₃In₄Sb₇₄Te₁₇. The measurable growth rates from the different techniques span from ~10^-9 to >10 m/s and the in-situ imaging techniques applied include optical microscopy, conventional transmission electron microscopy (TEM), and dynamic TEM, a photo-emission TEM technique with nanosecond-scale time resolution [3]. It will be shown that crystal growth just above T_g could be imaged during in situ TEM through the use of sub-framing and a high-frame-rate direct electron detection camera [4]. The use of complementary in situ experimental techniques allow the crystal growth rates to be mapped over a large temperature range and can give insights into the crystallization kinetics of PCMs. Challenges associated with integrating results from different microscopic techniques, the issues with the determination of temperature during in situ TEM experiments, and the incorporation of nanocalorimetry with simultaneous TEM imaging [5,6] during crystallization will be discussed.

References
[1] M. Salinga et al. Nat Comms 4 (2013) 1.
[2] J. Orava et al. Adv Funct Mater 25 (2015) 4851.
[3] V.L. Bird et al. Microsc Microanal, 24(S1) (2018) 1868.
[4] B.W. Reed et al. Microsc Microanal 23(S1) (2017), 84.
[5] M.D. Grapes et al. Rev Sci Instrum 85 (2014) 084902
[6] M.D. Grapes et al. Thermochimica Acta 658 (2017) 72.

The oxygen exchange and storage capacity of CeO₂ make it a common component in oxidation and reduction catalysts, with a wide range of applications such as automotive three-way catalysts, reforming, and fuel cells. These catalytic systems are typically based on a highly dispersed metal phase supported on CeO₂ nanoparticles, either pure or doped. In many cases, the fundamental catalytic mechanism usually involves reactant activation on the metallic component and oxygen transfer at or near the three-phase boundary that exists between the metal, oxide and gas phases. In the so-called Mars-van Krevelen type process, the reducible oxide support plays a key role by serving as a buffer to store and release oxygen to the reactant intermediates on the surface. The oxygen exchange process involves creation and annihilation of oxygen vacancies on the surface. Modelling with density functional theory suggests that the oxygen vacancy formation energy varies considerably depending on the characteristics of the surface site. To investigate the potential impact of surface heterogeneity during oxygen exchange, we have developed an in situ electron microscopy approach that allows us to directly observe the dynamic creation and annihilation of oxygen vacancies on CeO₂ surfaces and structural changes in Pt nanoparticles. We have employed this technique to explore the variation of lattice oxygen exchange with surface structural features such as nanofacets, step sites, adatom location and strain. In metal/oxide systems, the metal-support interface may be destabilized by these oxygen exchange events, giving rise to a surface that is constantly restructuring or fluxional behavior during catalysis. This presentation will discuss TEM based effort to determine in the oxygen exchange rate at different locations on CeO₂ nanoparticle surfaces and also at the metal/oxide interface of Pt/CeO₂ catalysts observed under operando conditions.
We gratefully acknowledge support of NSF grant NSF DMR (1840841, 1308085), CBET-1604971 and OAC-1940263.

Nanosecond pulsed laser melting using a frequency doubled Nd:YAG laser (l = 532 nm) was performed on implant amorphized Si and Si_1-xGe_x thin films. Undoped pseudomorphic Si_1-xGe_x thin films were grown to a thickness of 40 nm on (100) Si wafers and Ge⁺ implants produced ~20 nm amorphous layers. The dynamic evolution of the single pulse laser melt process was probed in detail by using in situ time resolved reflectometry (TRR) covering the sub-melt, partial and full amorphous melt, to full epi-layer melt regimes. TRR spectra were correlated as a function of energy density and Ge concentration with corresponding post-irradiation cross sectional TEM micrographs. It was shown that the resolution of the amorphous to crystalline conversion based on reflectivity differences allowed for the distinction between solid phase and liquid phase epitaxy regimes. Additionally, detection of the melt phase increased in sensitivity with increasing Ge concentration. Recovering the damage of implanted Si_1-xGe_x is a crucial step in CMOS manufacturing, and the metastable microstructures produced by pulsed laser melting of Si_1-xGe_x may be of interest for source/drain contact and channel strain engineering applications.

Lead halide perovskites have attracted great interest in recent years as promising materials for optoelectronics. Their ability to be solution processed is central to this interest, as it is a low-cost, scalable method with considerable flexibility. It has previously been observed that perovskite precursor inks are colloidal in nature, and that ink chemistry affects both the colloid distribution and the morphology of the deposited film. Yet, the mechanisms through which these effects arise are still poorly understood due in large part to the experimental difficulty of characterizing nanoscale colloids in solution. Here we demonstrate how cryo-electron microscopy can be used to overcome this challenge, presenting results from an interdisciplinary suite of methods that includes electron diffraction of vitrified inks and near-atomic resolution single particle averaging. Our results reveal for the first time that colloids in prototypical inks consist of a crystalline precursor phase, distinct from any of the starting materials. This shines light into the black box of perovskite solution processing and will ultimately enable better control of film morphology and optoelectronic properties.

The nucleation of solids plays an important role in crystallization applications. Various intermediates have been proposed, including a dense liquid phase formed via liquid-liquid phase separation, but little is known about the occurrence of liquid-liquid phase separations at the nanoscale and the evolution of the dense liquid phase into a solid phase. Herein, taking CaCO₃ as a model system, our in situ liquid cell transmission electron microscopy directly shows that the dense liquid phase first appears from highly supersaturated solution in the presence of proteins as a stabilizer, which has also been demonstrated by in situ liquid NMR and ATR-FTIR results. By analyzing the behavior of coalescing droplets, it is found that the dense liquid phase may be a high viscosity fluid. In addition, when highly charged proteins are added, a network of the dense liquid phase forms, reminiscent of spinodal patterns in polymer melts near the interior of a spinodal. Subsequent dehydration of liquid droplets drives the formation of less hydrated amorphous calcium carbonate. This direct observation of dense liquid phase formation dynamics helps provide an in-depth understanding of the liquid-liquid phase separation mediated nucleation model.

Cubic boron arsenide (BAs) is attracting greater attention owing to the recent experimental demonstration of ultrahigh thermal conductivity κ higher than 1000 W/m●K. However, its bandgap has not been settled and a simple yet effective method to probe its crystal quality is missing. Furthermore, traditional κ measurement methods are destructive and time-consuming, thus they cannot meet the urgent demand for fast screening of high κ materials. After we experimentally established 1.82 eV as the indirect bandgap of BAs and observed room-temperature band-edge photoluminescence, we developed two new optical techniques that can provide rapid and non-destructive characterization of κ with little sample preparation: photoluminescence mapping (PL-mapping) and time-domain thermo-photoluminescence (TDTP). PL-mapping provides a nearly real-time image of crystal quality and κ over mm-sized crystal surfaces; while TDTP allows us to pick up any spot on the sample surface and measure its κ using nanosecond laser pulses. These new techniques reveal that the apparent single crystals are not only nonuniform in κ but also are made of domains of very distinct κ. Because PL-mapping and TDTP are based on the band-edge PL and its dependence on temperature, they can be applied to other semiconductors, thus paving the way for rapid identification and development of high-κ semiconducting materials.