Henrik Eliasson

Indium–zirconium oxides rank among the most selective and stable catalysts for CO ₂ hydrogenation to methanol. Yet, despite extensive research, the mechanistic origin of the exceptional role of monoclinic zirconia remains unresolved and continues to set the benchmark in the field. Here we show that monoclinic hafnia, a wide-bandgap oxide rarely explored in catalysis, can outperform this benchmark. Nanostructured indium–hafnium oxides synthesized via flame spray pyrolysis achieve up to 70% higher indium-specific methanol productivity than indium–zirconium oxides, with the largest gains observed for single atoms of indium. Experimental and theoretical analyses reveal that a combination of stable monoclinic support surfaces, flexible chemical potential of indium single atoms and the presence of a cooperative hydride–proton reservoir collectively enhance CO ₂ activation and intermediate hydrogenation. Crucially, the precise control of surface hydroxylation is required. These findings establish a new benchmark for green methanol synthesis and provide generalizable design principles for next-generation oxide supports in single-atom catalysis.

Supported reducible oxides, such as indium oxide on monoclinic zirconia (In₂O₃/m‐ZrO₂), are promising catalysts for green methanol synthesis via CO₂ hydrogenation. Growing evidence suggests that dynamic restructuring under reaction conditions plays a crucial but poorly understood role in catalytic performance. To address this, the direct visualization of the state‐of‐the‐art In₂O₃/m‐ZrO₂ catalyst under CO₂ hydrogenation conditions (T = 553 K, P = 1.9 bar, CO₂:H₂ = 1:4) is pioneered using in situ scanning transmission electron microscopy (STEM), comparing its behavior to In₂O₃ on supports with similar (tetragonal, t‐ZrO₂ or anatase TiO₂) or lower (LSm‐ZrO₂) surface areas. Complementary in situ infrared spectroscopy and catalytic tests confirm methanol formation under equivalent conditions. A machine‐learning‐based difference imaging approach differentiates and ranks restructuring patterns, revealing that partially reduced InO_x species on m‐ZrO₂ undergo cyclic aggregation‐redispersion via atomic surface migration, maintaining high active phase dispersion. High‐resolution ex situ STEM analysis further shows the epitaxial formation of In₂O₃ mono‐ and bilayers on (100) m‐ZrO₂ facets, highlighting strong oxide‐support interactions. In contrast, sintering prevails on t‐ZrO₂, a‐TiO₂, and low‐surface m‐ZrO₂, correlating with lower methanol productivity. This work underscores the pivotal role of oxide‐support interfacial interactions in the reaction‐induced restructuring of InO_x species and establishes a framework for tracking nanoscale catalyst dynamics.

Indium‐zirconium (InZrO _x ) and zinc‐zirconium oxides (ZnZrO _x ) have emerged as highly selective and stable catalysts for CO ₂ hydrogenation to methanol, a versatile energy carrier. However, the disparity in synthesis methods, catalyst formulations, and structures previously studied precludes quantitative comparisons between the two families. Herein, a rigorous framework is pioneered to benchmark InZrO _x and ZnZrO _x materials prepared by a standardized flame spray pyrolysis synthesis platform, enabling consistently high surface areas and tunable metal speciation ranging from isolated atoms (10 mol%). Isolated indium and zinc species are commonly identified to be optimal for activity and methanol selectivity in their respective families, maximizing CO ₂ and H ₂ activation abilities. InZrO _x outperforms ZnZrO _x across speciations and is less structure sensitive, as deviations from atomic dispersion is less detrimental on performance for the former. Focusing on representative catalysts featuring saturation of isolated species, the higher activity of 5 mol% InZrO _x over its ZnZrO _x counterpart is linked to differences in surface oxygen vacancy chemistry, a lower degree of product inhibition, and more facile hydrogenation of the formate intermediate to methoxy. The identification of reactivity descriptors governing both families facilitates the development of unified guidelines in designing reducible oxide catalysts.

The computational cost of simulating scanning transmission electron microscopy (STEM) images limits the curation of large enough data sets to train accurate and robust machine learning networks for deep feature extraction from atomically resolved STEM images. For nanoparticle size estimation in particular, a diverse data set is essential due to the large variations in size, shape, crystallinity, orientation, and dynamical diffraction effects in experimental data. To address this, we train a 3D convolutional neural network to predict STEM images from voxelized atomic models, achieving a 100x speed-up compared to traditional multislice simulations while maintaining high image quality. We then generate a data set of 100.000 synthetic multislice images and investigate the performance of different size-estimator architectures as a function of training set size. A ResNet18-based model trained on 4000 real and 100.000 synthetic images is found to perform the best, reducing the median size-estimation error from 9.89% without synthetic data to 5.26%.

Transmission electron microscopy (TEM) plays a crucial role in heterogeneous catalysis for assessing the size distribution of supported metal nanoparticles. Typically, nanoparticle size is quantified by measuring the diameter under the assumption of spherical geometry, a simplification that limits the precision needed for advancing synthesis‐structure‐performance relationships. Currently, there is a lack of techniques that can reliably extract more meaningful information from atomically resolved TEM images, like nuclearity or geometry. Here, cycle‐consistent generative adversarial networks (CycleGANs) are explored to bridge experimental and simulated images, directly linking experimental observations with information from their underlying atomic structure. Using the versatile Pt/CeO₂ (Pt particles centered ≈2 nm) catalyst synthesized by impregnation, large datasets of experimental scanning transmission electron micrographs and physical image simulations are created to train a CycleGAN. A subsequent size‐estimation network is developed to determine the nuclearity of imaged nanoparticles, providing plausible estimates for ≈70% of experimentally observed particles. This automatic approach enables precise size determination of supported nanoparticle‐based catalysts overcoming crystal orientation limitations of conventional techniques, promising high accuracy with sufficient training data. Tools like this are envisioned to be of great use in designing and characterizing catalytic materials with improved atomic precision.

We present a facile chemical method for converting HDPE waste into valuable long-chain dicarboxylic acids (LCDCAs), thus transforming polymer waste into monomers for potentially novel polyester synthesis.

Mechanochemical activation in a ball mill effectively stabilizes Fe single atoms on a defective boron nitride host, resulting in a stable and selective catalyst for propylene production via N ₂ O-mediated oxidative dehydrogenation of propane.

To accurately capture the dynamic behavior of small nanoparticles in scanning transmission electron microscopy, high-quality data and advanced data processing is needed. The fast scan rate required to observe structural dynamics inherently leads to very noisy data where machine learning tools are essential for unbiased analysis. In this study, we develop a workflow based on two U-Net architectures to automatically localize and classify atomic columns at particle-support interfaces. The model is trained on non-physical image simulations, achieves sub-pixel localization precision, high classification accuracy, and generalizes well to experimental data. We test our model on both in situ and ex situ experimental time series recorded at 5 frames per second of small Pt nanoparticles supported on CeO₂(111). The processed movies show sub-second dynamics of the nanoparticles and reveal site-specific movement patterns of individual atomic columns.

We explore small Pt nanoparticles on CeO₂ in an interdisciplinary study combining in-situ and ex-situ scanning transmission electron microscopy (STEM), size-selected cluster deposition, neural networks, and density functional theory (DFT), to provide fundamental insights into the structure and dynamics of supported nanoparticles under various conditions [ 1].