Judith C. Yang

Shape and compositional control are widely used to tailor electrocatalyst properties, yet how these deliberately engineered features evolve under operating conditions remains poorly understood, obscuring the relationship between the as-synthesized catalyst and its active state. Here, we directly track the evolution of shape-controlled zigzag Pt₃Fe nanowires during the hydrogen evolution reaction under weakly alkaline conditions using in situ electrochemical liquid-cell transmission electron microscopy (EC-TEM), complemented by in situ X-ray absorption spectroscopy (XAS), elemental analysis, electron tomography, and electrostatic simulations. Operando EC-TEM reveals that the initially well-defined zigzag nanowires undergo spatially heterogeneous morphological evolution, including protrusion growth and nanoparticle formation preferentially at high-curvature regions. Correlated elemental mapping reveals selective Fe depletion and the formation of Pt-rich surface structures, while in situ XAS reveals increased Pt–Pt and decreased Pt–Fe coordination together with changes in the Fe chemical environment under HER conditions. Three-dimensional electron tomography resolves the zigzag nanowires as thin, plate-like structure rather than the previously proposed spiral-like geometry. Electrostatic simulations based on the experimentally determined geometry reveal electrostatic heterogeneity at high-curvature regions, providing a possible link between the nanowire geometry and the spatial preference of the observed morphological evolution. Together, these findings reveal coupled morphological evolution and elemental redistribution in shape-controlled alloy catalysts during electrochemical reactions. More broadly, these findings establish catalyst morphology as a dynamic parameter that not only defines the initial structure but also influences its subsequent evolution, highlighting dynamic structural evolution as an integral component of structure–property relationships in multicomponent electrocatalysts.

Autoencoder-transformer model applicable to both simulated and experimental XAS spectra is developed to predict the oxidation state of copper.

While in situ high-resolution transmission electron microscopy (HRTEM) allows atomic-resolution observation of dynamic processes such as chemical reactions and phase transformations, quantitative analysis of such data is often hindered by its complexity and volume that make manual labeling or traditional segmentation methods ineffective. Here, we demonstrate a data-efficient deep learning workflow using a standard U-Net architecture to segment atomic positions from in situ HRTEM images of copper oxides on copper substrates and gas interfaces—a system with challenging heterogeneous backgrounds and complex interfaces. The workflow is optimized for small datasets, achieving high accuracy after being trained on fewer than 50 manually labeled images augmented with rotation, scaling, and Gaussian filtering. The resulting U-Net model achieved test accuracies of approximately 92% for bulk atoms and 63% for interfacial atoms on images similar to the training dataset, with most misclassifications occurring at the complex interfacial regions. Moreover, our approach offers an approximately 300,000-fold improvement in labeling speed compared to manual labeling. This work presents a practical and adaptable framework for accelerating quantitative analysis of experimental systems for which acquiring large, annotated datasets is prohibitive, demonstrating the effective application of U-Net for a common challenge in electron microscopy.

High-resolution transmission electron microscopy (HRTEM) is crucial for observing material’s structural and morphological evolution at Angstrom scales, but the electron beam can alter these processes. Devices such as CMOS-based direct-electron detectors operating in electron-counting mode can be utilized to substantially reduce the electron dosage. However, the resulting images often lead to a low signal-to-noise ratio, which requires frame integration that sacrifices temporal resolution. Several machine learning (ML) models have been recently developed to successfully denoise HRTEM images. Yet, these models are often computationally expensive, and their inference speeds on GPUs are outpaced by the imaging speed of advanced detectors, precluding in situ analysis. Furthermore, the performance of these denoising models on datasets with imaging conditions that deviate from the training datasets has not been evaluated. To mitigate these gaps, we propose a new self-supervised ML denoising pipeline specifically designed for time-series HRTEM images. This pipeline integrates a blind-spot convolution neural network with pre-processing and post-processing steps, including drift correction and low-pass filtering. Results demonstrate that our model outperforms various other ML and non-ML denoising methods in noise reduction and contrast enhancement, leading to improved visual clarity of atomic features. Additionally, the model is drastically faster than U-Net-based ML models and demonstrates excellent out-of-distribution generalization. The model’s computational inference speed is in the order of milliseconds per image, rendering it suitable for application in in-situ HRTEM experiments.

The negatively charged boron vacancy (VB−) in hexagonal boron nitride (hBN) is a promising quantum defect that can be used to sense pressure, temperature, and magnetic field with high spatial resolution. hBN enriched with the boron-10 and nitrogen-15 isotopes, denoted h10B15N, has good contrast and coherence for quantum sensing because nitrogen-15 has nuclear spin (1/2), reducing hyperfine interactions. Boron vacancies can be generated by neutron irradiation, which causes the transmutation of the boron-10 isotope to lithium-7. In this study, the ancillary structural, compositional, and mechanical properties of h10B15N crystals that have been subjected to neutron irradiation fluences from 1.4 × 1016 to 8.4 × 1017 n/cm2 were thoroughly characterized. Besides creating VB−, the process also induces other defects that generate strain in the crystal lattice. In turn, the mechanical properties of these crystals change drastically. Investigated here are the visual changes, lattice integrity, composition, crystal strain, and elastic constants (C33 and C66) to assess how these characteristics change as a function of neutron fluence.

A comparative study of Cr/Au and Ti/Au ohmic contacts on (001) β-Ga2O3 was conducted. The electrical behavior from current-voltage measurements and the interfacial composition and microstructure as determined from high-resolution transmission electron microscopy (TEM) with energy dispersive x-ray analysis were compared for the different contacts at selected points in an annealing series (300–700 °C, 1 min. anneals in N2). Cr/Au contacts became ohmic at temperatures (300–350 °C) approximately 50–100 °C lower than Ti/Au contacts (400–450 °C). Cr/Au and Ti/Au contacts demonstrated optimal ohmic behavior (lowest resistance) when annealed to 450–500 and 500–600 °C, respectively, with Ti/Au contacts yielding a lower total resistance than Cr/Au. Cross-sectional TEM images of Cr/Au contacts annealed at 450 °C revealed the presence of Au nanoclusters at the Ga2O3 interface and CrOx layers at both the top of the contact and the Ga2O3 interface. Whereas TiOx also formed at the top and bottom interfaces in 450 °C-annealed Ti/Au contacts, the TiOx surface layer appeared to be variable in thickness and/or discontinuous, unlike the CrOx surface layer. Au nanoclusters were not detected at the interface in the Ti/Au contacts. The interdiffusion and oxidation observed in both contact metallizations point to the need for diffusion barriers that may allow these contacts to be used in future Ga2O3-based devices that operate at elevated temperatures.

At low temperatures (< 400°C), single atoms of Rh supported on rutile TiO2 (rTiO2) are responsible for the formation of CO during the reverse water gas shift (RWGS), while methane...

High-resolution transmission electron microscopy (HRTEM) is crucial for observing material's structural and morphological evolution at Angstrom scales, but the electron beam can alter these processes. Devices such as CMOS-based direct-electron detectors operating in electron-counting mode can be utilized to substantially reduce the electron dosage. However, the resulting images often lead to low signal-to-noise ratio, which requires frame integration that sacrifices temporal resolution. Several machine learning (ML) models have been recently developed to successfully denoise HRTEM images. Yet, these models are often computationally expensive and their inference speeds on GPUs are outpaced by the imaging speed of advanced detectors, precluding in situ analysis. Furthermore, the performance of these denoising models on datasets with imaging conditions that deviate from the training datasets have not been evaluated. To mitigate these gaps, we propose a new self-supervised ML denoising pipeline specifically designed for time-series HRTEM images. This pipeline integrates a blind-spot convolution neural network with pre-processing and post-processing steps including drift correction and low-pass filtering. Results demonstrate that our model outperforms various other ML and non-ML denoising methods in noise reduction and contrast enhancement, leading to improved visual clarity of atomic features. Additionally, the model is drastically faster than U-Net-based ML models and demonstrates excellent out-of-distribution generalization. The model's computational inference speed is in the order of milliseconds per image, rendering it suitable for application in in-situ HRTEM experiments.

Understanding oxide reduction is critical for advancing metal production^1,2, catalysis^3,4 and energy technologies⁵. Although carbon monoxide (CO) and hydrogen (H₂) are widely used reductants, the mechanisms by which they work are often presumed to be similar, both involving lattice oxygen removal^6,7,8,9. However, because of growing interest in replacing CO with H₂ to lower CO₂ emissions, distinguishing gas-specific reduction pathways is critical. Yet, capturing these atomic-scale processes under reactive gas and high-temperature conditions remains challenging. Here we use environmental transmission electron microscopy, which is capable of real-time, atomic-resolution imaging of gas–solid redox reactions^{10,11,12,13,14,15,16}, to directly visualize the gas-dependent oxide reduction dynamics in NiO. We show that CO drives surface nucleation and the growth of metallic Ni islands, leading to self-limiting surface metallization. Conversely, H₂ activates a coupled surface-to-bulk transformation, where protons from dissociated H₂ infiltrate the oxide lattice to promote the inward migration of surface-generated oxygen vacancies and enabling bulk metallization. By contrast, oxygen vacancies formed by CO remain confined near the surface, where they rapidly form a metallic Ni layer that inhibits further reduction. These results reveal distinct atomistic pathways for CO and H₂ and provide insights that may guide metallurgical processes and catalyst design.

Titanium dioxide (TiO₂) is one of the most extensively studied oxides as an active catalyst or catalyst support, particularly in energy and environmental applications, but the atomistic mechanisms governing its dynamic response to reactive environments and their correlation to reactivity remain largely elusive. Using in situ environmental transmission electron microscopy (ETEM), synchrotron X-ray diffraction (XRD), ambient-pressure X-ray photoelectron spectroscopy (AP-XPS), temperature-programmed reduction (TPR), reactivity measurements, and theoretical modeling, we reveal the dynamic interplay between oxygen loss and replenishment of anatase TiO₂ under varying reactive conditions. Under H₂ exposure, anatase TiO₂ undergoes surface reduction via lattice oxygen loss, forming Ti₃O₅. In contrast, CO₂ exposure induces oxygen replenishment, reversing stoichiometry. In mixed H₂/CO₂ environments, the reverse water-gas shift (RWGS) reaction proceeds selectively on stepped and high-indexed TiO₂ surfaces, whereas the thermodynamically stable TiO₂(101) surface remains inactive and intact. Critically, H₂ pretreatment generates oxygen vacancies on TiO₂(101), transforming it into an active Ti₃O₅ or defect-rich surface that catalyzes RWGS. By correlating surface structure, defect dynamics, and gas-phase interactions, this work deciphers the competition between H₂-driven reduction and CO₂-driven oxidation pathways at the atomic scale. These insights establish defect engineering as a strategic lever to activate inert TiO₂ facets, advancing the design of adaptive catalysts for sustainable fuel synthesis technologies.