Since the discovery of mechanically exfoliated graphene in 2004, research on ultrathin two-dimensional (2D) nanomaterials has grown exponentially in the fields of condensed matter physics, material science, chemistry, and nanotechnology. Highlighting their compelling physical, chemical, electronic, and optical properties, as well as their various potential applications, in this Review, we summarize the state-of-art progress on the ultrathin 2D nanomaterials with a particular emphasis on their recent advances. First, we introduce the unique advances on ultrathin 2D nanomaterials, followed by the description of their composition and crystal structures. The assortments of their synthetic methods are then summarized, including insights on their advantages and limitations, alongside some recommendations on suitable characterization techniques. We also discuss in detail the utilization of these ultrathin 2D nanomaterials for wide ranges of potential applications among the electronics/optoelectronics, electrocat...

Recent Advances in Ultrathin Two-Dimensional Nanomaterials

Metal species with different size (single atoms, nanoclusters, and nanoparticles) show different catalytic behavior for various heterogeneous catalytic reactions. It has been shown in the literature that many factors including the particle size, shape, chemical composition, metal–support interaction, and metal–reactant/solvent interaction can have significant influences on the catalytic properties of metal catalysts. The recent developments of well-controlled synthesis methodologies and advanced characterization tools allow one to correlate the relationships at the molecular level. In this Review, the electronic and geometric structures of single atoms, nanoclusters, and nanoparticles will be discussed. Furthermore, we will summarize the catalytic applications of single atoms, nanoclusters, and nanoparticles for different types of reactions, including CO oxidation, selective oxidation, selective hydrogenation, organic reactions, electrocatalytic, and photocatalytic reactions. We will compare the results o...

Metal Catalysts for Heterogeneous Catalysis: From Single Atoms to Nanoclusters and Nanoparticles.

Single-Atom Catalysts: Synthetic Strategies and Electrochemical Applications

Electrochemical reduction of CO2 to chemical fuel offers a promising strategy for managing the global carbon balance, but presents challenges for chemistry due to the lack of effective electrocatalyst. Here we report atomically dispersed nickel on nitrogenated graphene as an efficient and durable electrocatalyst for CO2 reduction. Based on operando X-ray absorption and photoelectron spectroscopy measurements, the monovalent Ni(i) atomic center with a d9 electronic configuration was identified as the catalytically active site. The single-Ni-atom catalyst exhibits high intrinsic CO2 reduction activity, reaching a specific current of 350 A gcatalyst−1 and turnover frequency of 14,800 h−1 at a mild overpotential of 0.61 V for CO conversion with 97% Faradaic efficiency. The catalyst maintained 98% of its initial activity after 100 h of continuous reaction at CO formation current densities as high as 22 mA cm−2. Electrocatalysts with improved activity and stability for the conversion of CO2 to CO are being sought. Using operando spectroscopies, the authors identify atomically dispersed Ni(i) as the active site in a nitrogenated-graphene-supported catalyst with high intrinsic activity and stability over 100 hours.

Atomically dispersed Ni(i) as the active site for electrochemical CO2 reduction

Photoreduction of CO2 into sustainable and green solar fuels is generally believed to be an appealing solution to simultaneously overcome both environmental problems and energy crisis. The low selectivity of challenging multi-electron CO2 photoreduction reactions makes it one of the holy grails in heterogeneous photocatalysis. This Review highlights the important roles of cocatalysts in selective photocatalytic CO2 reduction into solar fuels using semiconductor catalysts. A special emphasis in this review is placed on the key role, design considerations and modification strategies of cocatalysts for CO2 photoreduction. Various cocatalysts, such as the biomimetic, metal-based, metal-free, and multifunctional ones, and their selectivity for CO2 photoreduction are summarized and discussed, along with the recent advances in this area. This Review provides useful information for the design of highly selective cocatalysts for photo(electro)reduction and electroreduction of CO2 and complements the existing reviews on various semiconductor photocatalysts.

Cocatalysts for Selective Photoreduction of CO2 into Solar Fuels.

Electroreduction of CO2 into useful fuels, especially if driven by renewable energy, represents a potentially 'clean' strategy for replacing fossil feedstocks and dealing with increasing CO2 emissions and their adverse effects on climate. The critical bottleneck lies in activating CO2 into the CO2(•-) radical anion or other intermediates that can be converted further, as the activation usually requires impractically high overpotentials. Recently, electrocatalysts based on oxide-derived metal nanostructures have been shown to enable CO2 reduction at low overpotentials. However, it remains unclear how the electrocatalytic activity of these metals is influenced by their native oxides, mainly because microstructural features such as interfaces and defects influence CO2 reduction activity yet are difficult to control. To evaluate the role of the two different catalytic sites, here we fabricate two kinds of four-atom-thick layers: pure cobalt metal, and co-existing domains of cobalt metal and cobalt oxide. Cobalt mainly produces formate (HCOO(-)) during CO2 electroreduction; we find that surface cobalt atoms of the atomically thin layers have higher intrinsic activity and selectivity towards formate production, at lower overpotentials, than do surface cobalt atoms on bulk samples. Partial oxidation of the atomic layers further increases their intrinsic activity, allowing us to realize stable current densities of about 10 milliamperes per square centimetre over 40 hours, with approximately 90 per cent formate selectivity at an overpotential of only 0.24 volts, which outperforms previously reported metal or metal oxide electrodes evaluated under comparable conditions. The correct morphology and oxidation state can thus transform a material from one considered nearly non-catalytic for the CO2 electroreduction reaction into an active catalyst. These findings point to new opportunities for manipulating and improving the CO2 electroreduction properties of metal systems, especially once the influence of both the atomic-scale structure and the presence of oxide are mechanistically better understood.

Partially oxidized atomic cobalt layers for carbon dioxide electroreduction to liquid fuel

Generative adversarial networks (GANs) have great successes on synthesizing data However, the existing GANs restrict the discriminator to be a binary classifier, and thus limit their learning capacity for tasks that need to synthesize output with rich structures such as natural language descriptions In this paper, we propose a novel generative adversarial network, RankGAN, for generating high-quality language descriptions Rather than training the discriminator to learn and assign absolute binary predicate for individual data sample, the proposed RankGAN is able to analyze and rank a collection of human-written and machine-written sentences by giving a reference group By viewing a set of data samples collectively and evaluating their quality through relative ranking scores, the discriminator is able to make better assessment which in turn helps to learn a better generator The proposed RankGAN is optimized through the policy gradient technique Experimental results on multiple public datasets clearly demonstrate the effectiveness of the proposed approach

/pdf/adversarial-ranking-for-language-generation-570bakb1n8.pdf

Adversarial ranking for language generation

Adversarial examples expose the vulnerabilities of natural language processing (NLP) models, and can be used to evaluate and improve their robustness. Existing techniques of generating such examples are typically driven by local heuristic rules that are agnostic to the context, often resulting in unnatural and ungrammatical outputs. This paper presents CLARE, a ContextuaLized AdversaRial Example generation model that produces fluent and grammatical outputs through a mask-then-infill procedure. CLARE builds on a pre-trained masked language model and modifies the inputs in a context-aware manner. We propose three contextualized perturbations, Replace, Insert and Merge, that allow for generating outputs of varied lengths. CLARE can flexibly combine these perturbations and apply them at any position in the inputs, and is thus able to attack the victim model more effectively with fewer edits. Extensive experiments and human evaluation demonstrate that CLARE outperforms the baselines in terms of attack success rate, textual similarity, fluency and grammaticality.

Contextualized Perturbation for Textual Adversarial Attack

Generative adversarial networks (GANs) have great successes on synthesizing data. However, the existing GANs restrict the discriminator to be a binary classifier, and thus limit their learning capacity for tasks that need to synthesize output with rich structures such as natural language descriptions. In this paper, we propose a novel generative adversarial network, RankGAN, for generating high-quality language descriptions. Rather than training the discriminator to learn and assign absolute binary predicate for individual data sample, the proposed RankGAN is able to analyze and rank a collection of human-written and machine-written sentences by giving a reference group. By viewing a set of data samples collectively and evaluating their quality through relative ranking scores, the discriminator is able to make better assessment which in turn helps to learn a better generator. The proposed RankGAN is optimized through the policy gradient technique. Experimental results on multiple public datasets clearly demonstrate the effectiveness of the proposed approach.

Adversarial Ranking for Language Generation.

We study how to generate captions that are not only accurate in describing an image but also discriminative across different images. The problem is both fundamental and interesting, as most machine-generated captions, despite phenomenal research progresses in the past several years, are expressed in a very monotonic and featureless format. While such captions are normally accurate, they often lack important characteristics in human languages - distinctiveness for each caption and diversity for different images. To address this problem, we propose a novel conditional generative adversarial network for generating diverse captions across images. Instead of estimating the quality of a caption solely on one image, the proposed comparative adversarial learning framework better assesses the quality of captions by comparing a set of captions within the image-caption joint space. By contrasting with human-written captions and image-mismatched captions, the caption generator effectively exploits the inherent characteristics of human languages, and generates more discriminative captions. We show that our proposed network is capable of producing accurate and diverse captions across images.

Dianqi Li

Papers

Partially oxidized atomic cobalt layers for carbon dioxide electroreduction to liquid fuel

Adversarial ranking for language generation

Contextualized Perturbation for Textual Adversarial Attack

Adversarial Ranking for Language Generation.

Generating Diverse and Accurate Visual Captions by Comparative Adversarial Learning.