Search

Abstract Ni-based catalysts are highly reactive for dry reforming of methane (DRM) but they are prone to rapid deactivation due to sintering and/or coking. In this study, we present a straightforward approach for anchoring dispersed Ni sites with strengthened metal-support interactions, which leads to Ni active sites embedded in dealuminated Beta zeolite with superior stability and rates for DRM. The process involves solid-state grinding of dealuminated Beta zeolites and nickel nitrate, followed by calcination under finely controlled gas flow conditions. By combining in situ X-ray absorption spectroscopy and ab initio simulations, it is elucidated that the efficient removal of byproducts during catalyst synthesis is conducted to strengthen Ni–Si interactions that suppress coking and sintering after 100 h of time-on-stream. Transient isotopic kinetic experiments shed light on the differences in intrinsic turnover frequency of Ni species and explain performance trends. This work constructs a fundamental understanding regarding the implication of facile synthesis protocols on metal-support interaction in zeolite-supported Ni sites, and it lays the needed foundations on how these interactions can be tuned for outstanding DRM performance.

The joint of multisource remote sensing (RS) data for land cover classification has become a popular research topic. Although studies have shown that the fusion of multisource data can improve the accuracy of classification, the current limitation lies in the inadequate exploitation of information, resulting in spectral confusion categories overlapping, and varying visual differences within the same category. To address these problems, this article proposes a multiscale spectral–spatial attention residual fusion network (MSSARFNet) that aims to enhance the classification performance of multisource RS data through effective spectral–spatial feature extraction and fusion. Specifically, three modules are designed: the multiscale spectral attention residual module (MSpeARM), the multiscale spatial attention residual module (MSpaARM), and multiscale convolution fusion (MCF) module. First, we divide the channels of the features into multiple paths and apply spectral–spatial attention mechanisms on each path. To further enhance connectivity, convolutional operation is utilized to establish connections between different paths, thus forming MSpeARM and MSpaARM. The MSpeARM suppresses redundant features and enhances effective features along the channel dimension to better differentiate spectral-confused categories. The MSpaARM highlights that different visual patterns of objects in the same category can mutually reinforce each other by weighting all positional features, regardless of their spatial differences. Second, to fuse these two sets of features, the MCF module is designed to learn multilevel semantic features and enhance fusion at a granular level. Experimental evaluations on three RS datasets demonstrate that the proposed method achieves excellent classification performance, indicating the effectiveness of MSSARFNet.

The vulnerability of modern neural networks to random noise and deliberate attacks has raised concerns about their robustness, particularly as they are increasingly utilized in safety- and security-critical applications. Although recent research efforts were made to enhance robustness through retraining with adversarial examples or employing data augmentation techniques, a comprehensive investigation into the effects of training data perturbations on model robustness remains lacking. This paper presents the first extensive empirical study investigating the influence of data perturbations during model retraining. The experimental analysis focuses on both random and adversarial robustness, following established practices in the field of robustness analysis. Various types of perturbations in different aspects of the dataset are explored, including input, label, and sampling distribution. Single-factor and multi-factor experiments are conducted to assess individual perturbations and their combinations. The findings provide insights into constructing high-quality training datasets for optimizing robustness and recommend the appropriate degree of training set perturbations that balance robustness and correctness, and contribute to understanding model robustness in deep learning and offer practical guidance for enhancing model performance through perturbed retraining, promoting the development of more reliable and trustworthy deep learning systems for safety-critical applications.

Spiking neural network (SNN) is a brain-inspired model with more spatio-temporal information processing capacity and computational energy efficiency. However, with the increasing depth of SNNs, the memory problem caused by the weights of SNNs has gradually attracted attention. In this study, we propose an ultra-low latency adaptive local binary spiking neural network (ALBSNN) with accuracy loss estimators, which dynamically selects the network layers to be binarized to ensure a balance between quantization degree and classification accuracy by evaluating the error caused by the binarized weights during the network learning process. At the same time, to accelerate the training speed of the network, the global average pooling (GAP) layer is introduced to replace the fully connected layers by combining convolution and pooling. Finally, to further reduce the error caused by the binary weight, we propose binary weight optimization (BWO), which updates the overall weight by directly adjusting the binary weight. This method further reduces the loss of the network that reaches the training bottleneck. The combination of the above methods balances the network's quantization and recognition ability, enabling the network to maintain the recognition capability equivalent to the full precision network and reduce the storage space by more than 20%. So, SNNs can use a small number of time steps to obtain better recognition accuracy. In the extreme case of using only a one-time step, we still can achieve 93.39, 92.12, and 69.55% testing accuracy on three traditional static datasets, Fashion- MNIST, CIFAR-10, and CIFAR-100, respectively. At the same time, we evaluate our method on neuromorphic N-MNIST, CIFAR10-DVS, and IBM DVS128 Gesture datasets and achieve advanced accuracy in SNN with binary weights. Our network has greater advantages in terms of storage resources and training time.

Precise control of charge transfer between catalyst nanoparticles and supports presents a unique opportunity to enhance catalytic performance. Here the authors demonstrate a scanning transmission electron microscopy method to visualize atomic-scale structure and sub-nanometer-scale charge distribution in heterogeneous catalysts, shedding light on the atomic-scale mechanisms behind their highly active perimeter sites.

Styrene is an important commodity chemical that is highly energy and CO2 intensive to produce. Here, authors report a redox-oxidative dehydrogenation scheme and a tailored core-shell redox catalyst to convert ethylbenzene to styrene with up to 91.4% single-pass yield and 82% energy savings.

Strong metal–support interactions (SMSI) are effective in tuning the structures and catalytic performances of catalysts but limited by the poor exposure of active sites. Here, the authors develop a strategy to engineer SMSI via a reverse route, which is in favor of metal site exposure while embracing the SMSI.

Single-atom catalysts have been studied for CO oxidation, but experimental in situ investigations are limited. Here, the authors use a suite of in situ/operando spectroscopy to identify key intermediates and define design principles to enhance the CO oxidation activity of atomically dispersed Rh on heteropoly acids.

The large-scale interconnection of the power grid has brought great benefits to social development, but simultaneously, the frequency of large-scale fault accidents caused by extreme weather is also rocketing. The power grid is regarded as a representative complex network in this paper to analyze its functional vulnerability. First, the actual power grid topology is modeled on the basis of the complex network theory, which is transformed into a directed-weighted topology model after introducing the node voltage together with line reactance. Then, the algorithm of weighted reactance betweenness is proposed by analyzing the characteristic parameters of the power grid topology model. The product of unit reliability and topology model’s characteristic parameters under extreme weather is used as the index to measure the functional vulnerability of the power grid, which considers the extreme weather of freezing and gale and quantifies the functional vulnerability of lines under wind load, ice load, and their synergistic effects. Finally, a simulation using the IEEE-30 node system is implemented. The result shows that the proposed method can effectively measure the short-term vulnerability of power grid units under extreme weather. Meanwhile, the example analysis verifies the different effects of normal and extreme weather on the power grid and identifies the nodes and lines with high vulnerability under extreme weather, which provides theoretical support for preventing and reducing the impact of extreme weather on the power grid.

Abstract The ab initio density functional theoretical studies show that energetics favor CO oxidation on single Pd atoms supported on θ-alumina. The diffuse reflectance infra-red spectroscopy (DRIFTS) results show that carbonates are formed as intermediates when single supported Pd atoms are exposed to a gaseous mixture of CO + O2. The rapid agglomeration of Pd atoms under CO oxidation conditions even at 6 °C leads to the presence of Pd particles along with single atoms during CO oxidation experiments. Thus, the observed CO oxidation has contributions from both single Pd atoms and Pd particles.

Tuning electronic properties of metallic catalysts is a useful way to improve their activity, however control over metal-support interactions is still challenging. Here the authors report a vacancy-induced interfacial electronic effect for Pt assembled on vacancy-abundanth-BN nanosheets leading to superior CO oxidation catalysis.

Abstract New convenient wet-chemistry synthetic routes have made it possible to explore catalytic activities of a variety of single supported atoms, however, the single supported atoms on inert substrates (e.g. alumina) are limited to adatoms and cations of Pt, Pd, and Ru. Previously, we have found that single supported Pt atoms are remarkable NO oxidation catalysts. In contrast, we report that Pd single atoms are completely inactive for NO oxidation. The diffuse reflectance infra-red spectroscopy (DRIFTS) results show the absence of nitrate formation on catalyst. To explain these results, we explored modified Langmuir-Hinshelwood type pathways that have been proposed for oxidation reactions on single supported atom. In the first pathway, we find that there is energy barrier for the release of NO2 which prevent NO oxidation. In the second pathway, our results show that there is no driving force for the formation of O=N-O-O intermediate or nitrate on single supported Pd atoms. The decomposition of nitrate, if formed, is an endothermic event.

With a recent result of Suzuki (2001) we extend Caristi-Kirk's fixed point theorem, Ekeland's variational principle, and Takahashi's minimization theorem in a complete metric space by replacing the distance with a τ-distance. In addition, these extensions are shown to be equivalent. When the τ-distance is l.s.c. in its second variable, they are applicable to establish more equivalent results about the generalized weak sharp minima and error bounds, which are in turn useful for extending some existing results such as the petal theorem.

For functions f1, . . ., fn on a set D, we characterize their linear independence with an invertible matrix from their values at n distinct points in D. With the matrix, the pointwise convergence of a sequence {gk} of functions in the span{f1, · · ·, fn} is shown to be equivalent to those of the sequences of the coordinates of gks in the span. When fis are bounded, a pointwise convergent sequence {gk} must uniformly converge to a function in the span. It turns out that the limit of a convergent sequence {gk} inherits the continuity, differentiability, and integrability of fis. Furthermore the (pointwise or uniform) convergence of a sequence of solutions of an n-th order constant coefficients linear differential equation is completely determined by that of the sequence of relevant initial conditions. [ABSTRACT FROM AUTHOR]

Polyvinyl chloride (PVC) is ubiquitous in everyday life; however, it is not recycled because it degrades uncontrollably into toxic products above 250°C. Therefore, it is of interest to controllably dechlorinate PVC at mild temperatures to generate narrowly distributed carbon materials. We present a catalytic route to dechlorinate PVC (~90% reduction of Cl content) at mild temperature (200°C) to produce gas H2 (with negligible coproduction of corrosive gas HCl) and carbon materials using Ga as a liquid metal (LM) catalyst. A LM was used to promote intimate contact between PVC and the catalytic sites. During dechlorination of PVC, Cl is sequestrated in the carbonaceous solid product. Later, chlorine is easily removed with an acetone wash at room temperature. The Ga LM catalyst is reusable, outperforms a traditional supported metal catalyst, and successfully converts (untreated) discarded PVC pipe. [ABSTRACT FROM AUTHOR]