Oxide-supported noble metal catalysts have been extensively studied for decades for the water gas shift (WGS) reaction, a catalytic transformation central to a host of large volume processes that variously utilize or produce hydrogen. There remains considerable uncertainty as to how the specific features of the active metal-support interfacial bonding—perhaps most importantly the temporal dynamic changes occurring therein—serve to enable high activity and selectivity. Here we report the dynamic characteristics of a Pt/CeO2 system at the atomic level for the WGS reaction and specifically reveal the synergistic effects of metal-support bonding at the perimeter region. We find that the perimeter Pt0 − O vacancy−Ce3+ sites are formed in the active structure, transformed at working temperatures and their appearance regulates the adsorbate behaviors. We find that the dynamic nature of this site is a key mechanistic step for the WGS reaction.

Reducible oxides are widely used catalyst supports that can increase oxidation reaction rates by transferring lattice oxygen at the metal-support interface. There are many outstanding questions regarding the atomic-scale dynamic meta-stability (i.e., fluxional behavior) of the interface during catalysis. Here, we employ aberration-corrected operando electron microscopy to visualize the structural dynamics occurring at and near Pt/CeO2 interfaces during CO oxidation. We show that the catalytic turnover frequency correlates with fluxional behavior that (a) destabilizes the supported Pt particle, (b) marks an enhanced rate of oxygen vacancy creation and annihilation, and (c) leads to increased strain and reduction in the CeO2 support surface. Overall, the results implicate the interfacial Pt-O-Ce bonds anchoring the Pt to the support as being involved also in the catalytically-driven oxygen transfer process, and they suggest that oxygen reduction takes place on the highly reduced CeO2 surface before migrating to the interfacial perimeter for reaction with CO.

In many materials systems, such as catalytic nanoparticles, the ability to characterize dynamic atomic structural changes is important for developing a more fundamental understanding of functionality. Recent developments in direct electron detection now allow image series to be acquired at frame rates on the order of 1000 frames per second in bright-field transmission electron microscopy (BF TEM), which could potentially allow dynamic changes in the atomic structure of individual nanoparticles to be characterized with millisecond temporal resolution in favourable cases. However, extracting such data from TEM image series requires the development of computational methods that can be applied to very large datasets and are robust in the presence of noise and in the non-ideal imaging conditions of some types of environmental TEM experiments.Here, we present a two-dimensional Gaussian fitting algorithm to track the position and intensities of atomic columns in temporally resolved BF TEM image series. We have tested our algorithm on experimental image series of Ce atomic columns near the surface of a ceria (CeO2) nanoparticle with electron beam doses of ~125-5000 e -Å -2 per frame. The accuracy of the algorithm for locating atomic column positions is compared to that of the more traditional centroid fitting technique, and the accuracy of intensity measurements is evaluated as a function of dose per frame. The code developed here, and the methodology used to explore the errors and limitations of the 2 measurements, could be applied more broadly to any temporally resolved TEM image series to track dynamic atomic column motion. IntroductionAberration-corrected transmission electron microscopy (TEM) is a powerful tool for characterizing atomic structures with sub-angstrom spatial resolution [1][2][3]. Atomic resolution TEM images generally have an acquisition time on the order of seconds, and analysis of the images typically treats atomic structures as static. However, in many systems, such as catalytic nanoparticles, the atomic structure may undergo dynamic changes, particularly at the particle surface [3][4][5][6][7]. The ability to characterize these dynamic atomic structural changes is important for developing a more fundamental understanding of catalytic functionality and requires imaging techniques with both atomic-scale spatial resolution and improved temporal resolution.Recent advances in direct detection technology now allow TEM image series to be acquired at frame rates on the order of 1000 frames per second (fps), with high sensitivity at low electron fluences, or low electron doses, (using terminology more common in the electron microscopy community) [8][9][10][11][12][13]. This technology has the potential to allow atomic structures to be characterized with millisecond temporal resolution, which can potentially reveal new information about atomic structural dynamics. To extract quantitative information about structural dynamics from temporally resolved image series of a nanoparticle, computational methods are ...

No abstract

A deep convolutional neural network has been developed to denoise atomic-resolution transmission electron microscope image datasets of nanoparticles acquired using direct electron counting detectors, for applications where the image signal is severely limited by shot noise. The network was applied to a model system of CeO2-supported Pt nanoparticles. We leverage multislice image simulations to generate a large and flexible dataset for training the network. The proposed network outperforms state-of-the-art denoising methods on both simulated and experimental test data. Factors contributing to the performance are identified, including (a) the geometry of the images used during training and (b) the size of the network's receptive field. Through a gradient-based analysis, we investigate the mechanisms learned by the network to denoise experimental images. This shows that the network exploits both extended and local information in the noisy measurements, for example, by adapting its filtering approach when it encounters atomic-level defects at the nanoparticle surface. Extensive analysis has been done to characterize the network's ability to correctly predict the exact atomic structure at the nanoparticle surface. Finally, we develop an approach based on the log-likelihood ratio test that provides a quantitative measure of the agreement between the noisy observation and the atomic-level structure in the network-denoised image.