Multi-domain High-Resolution Platform for Integrated Spectroscopy and Microscopy Characterizations

In recent decades,materials science has experienced rapid development and posed increasingly high requirements for the characterizations of structures,properties,and performances.Herein,we report on our recent establishment of a multi-domain(energy,space,time)highresolution platform for integrated spectroscopy and microscopy characterizations,offering an unprecedented way to analyze materials in terms of spectral(energy)and spatial mapping as well as temporal evolution.We present several proof-of-principle results collected on this platform,including in-situ Raman imaging(high-resolution Raman,polarization Raman,low-wavenumber Raman),time-resolved photoluminescence imaging,and photoelectrical performance imaging.It can be envisioned that our newly established platform would be very powerful and effective in the multi-domain high-resolution characterizations of various materials of photoelectrochemical importance in the near future.

An improved theoretical procedure for the pore-size analysis of activated carbon by gas adsorption

Amorphous carbon materials play a vital role in adsorbed natural gas(ANG) storage. One of the key issues in the more prevalent use of ANG is the limited adsorption capacity, which is primarily determined by the porosity and surface characteristics of porous materials. To identify suitable adsorbents, we need a reliable computational tool for pore characterization and, subsequently, quantitative prediction of the adsorption behavior. Within the framework of adsorption integral equation(AIE), the pore-size distribution(PSD) is sensitive to the adopted theoretical models and numerical algorithms through isotherm fitting. In recent years, the classical density functional theory(DFT) has emerged as a common choice to describe adsorption isotherms for AIE kernel construction. However,rarely considered is the accuracy of the mean-field approximation(MFA) commonly used in commercial software. In this work, we calibrate four versions of DFT methods with grand canonical Monte Carlo(GCMC) molecular simulation for the adsorption of CH_4 and CO_2 gas in slit pores at 298 K with the pore width varying from 0.65 to 5.00 nm and pressure from 0.2 to 2.0 MPa. It is found that a weighted-density approximation proposed by Yu(WDA-Yu) is more accurate than MFA and other non-local DFT methods. In combination with the trapezoid discretization of AIE, the WDA-Yu method provides a faithful representation of experimental data, with the accuracy and stability improved by 90.0% and 91.2%, respectively, in comparison with the corresponding results from MFA for fitting CO_2 isotherms. In particular, those distributions in the feature pore width range(FPWR)are proved more representative for the pore-size analysis. The new theoretical procedure for pore characterization has also been tested with the methane adsorption capacity in seven activated carbon samples.

Scene word recognition from pieces to whole

Convolutional neural networks (CNNs) have had great success with regard to the object classification problem. For character classification, we found that training and testing using accurately segmented character regions with CNNs resulted in higher accuracy than when roughly segmented regions were used. Therefore, we expect to extract complete character regions from seene images. Text in natural scene images has an obvious contrast with its attachments. Many methods attempt to extract characters through different segmentation techniques. However, for blurred, occluded, and complex background cases, those methods may result in adjoined or over segmented characters. In this paper, we propose a scene word recognition model that integrates words from small pieces to entire after-cluster-based segmentation. The segmented connected components are classified as four types: background, in dividual character proposals, adjoined characters, and stroke proposals. Individual character proposals are directly inputted to a CNN that is trained using accurately segmented character images. The sliding window strategy is applied to adjoined character regions. Stroke proposals are considered as fragments of entire characters whose locations are estimated by a stroke spatial distribution system. Then、the estimated characters from adjoined characters and stroke proposals are classified by a CNN that is trained on roughly segmented character images. Finally, a lexicondriven integration method is performed to obtain the final word recognition results. Compared to other word recognition methods, our method achieves a comparable performance on Street View Text and the ICDAR 2003 and ICDAR 2013 benchmark databases. Moreover, our method can deal with recognizing text images of occlusion and improperly segmented text images.