Sensors allow an electronic device to become a gateway between the digital and physical worlds, and sensor materials with unprecedented performance can create new applications and new avenues for user interaction. Graphene oxide can be exploited in humidity and temperature sensors with a number of convenient features such as flexibility, transparency and suitability for large-scale manufacturing. Here we show that the two-dimensional nature of graphene oxide and its superpermeability to water combine to enable humidity sensors with unprecedented response speed (∼30 ms response and recovery times). This opens the door to various applications, such as touchless user interfaces, which we demonstrate with a 'whistling' recognition analysis.

Efficient solar-thermal energy conversion is essential for the harvesting and transformation of abundant solar energy, leading to the exploration and design of efficient solar-thermal materials. Carbon-based materials, especially graphene, have the advantages of broadband absorption and excellent photothermal properties, and hold promise for solar-thermal energy conversion. However, to date, graphene-based solar-thermal materials with superior omnidirectional light harvesting performances remain elusive. Herein, hierarchical graphene foam (h-G foam) with continuous porosity grown via plasma-enhanced chemical vapor deposition is reported, showing dramatic enhancement of broadband and omnidirectional absorption of sunlight, which thereby can enable a considerable elevation of temperature. Used as a heating material, the external solar-thermal energy conversion efficiency of the h-G foam impressively reaches up to ≈93.4%, and the solar-vapor conversion efficiency exceeds 90% for seawater desalination with high endurance.

No abstract

In image classification, visual separability between different object categories is highly uneven, and some categories are more difficult to distinguish than others. Such difficult categories demand more dedicated classifiers. However, existing deep convolutional neural networks (CNN) are trained as flat N-way classifiers, and few efforts have been made to leverage the hierarchical structure of categories. In this paper, we introduce hierarchical deep CNNs (HD-CNNs) by embedding deep CNNs into a category hierarchy. An HD-CNN separates easy classes using a coarse category classifier while distinguishing difficult classes using fine category classifiers. During HD-CNN training, component-wise pretraining is followed by global finetuning with a multinomial logistic loss regularized by a coarse category consistency term. In addition, conditional executions of fine category classifiers and layer parameter compression make HD-CNNs scalable for large-scale visual recognition. We achieve state-of-the-art results on both CI-FAR100 and large-scale ImageNet 1000-class benchmark datasets. In our experiments, we build up three different HD-CNNs and they lower the top-1 error of the standard CNNs by 2.65%, 3.1% and 1.1%, respectively.

A new method has been used to obtain pore size characteristics of MCM-41 catalyst supports and vanadiumsubstituted MCM-41 catalysts. The approach is based on the nonlocal density functional theory (NLDFT) model for nitrogen and argon adsorption in MCM-41, proposed recently. Samples with pore sizes varying from ca. 25 to 37 Å were prepared by hydrothermal synthesis. Two synthesis procedures employing different sources of V were used to prepare V/MCM-41 catalysts. The samples were characterized by X-ray diffraction (XRD). N 2 and Ar adsorption isotherms at 77 K were measured starting from the relative pressure P/P 0 ) 1 × 10 -5 . Analysis of adsorption isotherms was carried out in two stages. The first stage implies comparison of a given isotherm with a reference isotherm measured on a well-characterized sample of MCM-41 with uniform pores. From such a comparison, micropore volume, specific surface area of mesopores, and the point of the beginning of the capillary condensation are determined. In the second stage, pore size distributions are calculated from the NLDFT. Pore size distributions obtained from N 2 and Ar isotherms at 77 K were in perfect agreement. These results were compared with the traditional Barrett-Joyner-Halenda (BJH) method, and with the XRD data. It is shown that the BJH method underestimates an average pore size in MCM-41 materials by ca. 10 Å. Adsorption studies of V/MCM-41 catalysts revealed that the synthesis procedure with the direct addition of V 2 O 5 yields samples with a more uniform pore structure than the procedure with the use of VOSO 4 ‚3H 2 O solution.