Fig. 1 Flow of uncertainty-based active learning (US)92Obtaining a fine approximation of a black box function (BBF) is important in the field of material design. Active learning aims to obtain a better BBF with less training data. In this study, we investigate whether Active learning based on uncertainty sampling (US, Fig. 1) can obtain a BBF in regression tasks using material databases. The liquidus surfaces of the ternary systems (the case of uniform and small dimensional inputs) are compared with the material databases of inorganic materials, small molecules and polymers (the case of non-uniform and large dimensional inputs). In the first case, US works. In the latter case, US is sometimes inefficient. And it works when the dimensions of the inputs are small. This study was supported by a project subsidized by the Core Research for Evolutional Science and Technology (CREST) (Grant number JPMJCR2234).Photocatalytic technology is appealing in producing H2O2 due to its environmentally friendly aspects of low energy consumption and no pollution. The photocatalytic H2O2 production as value-added chemicals and sustainable solar fuel has attracted enormous attention globally attributed to its sustainability process, where water and oxygen are abundant and ecologically natural resources, and solar light is applied to initiate the chemical reaction. In this study, kapok microtubules (t-KF) is utilized for the first time as an electron donor to boost the photocatalytic activity of carbon-doped g-C3N4 (CCN) through ligand-to-metal charge transfer (LMCT), significantly improving H2O2 production. Cellulose, a major component of kapok fiber, plays a promising role in photocatalysis due to its electron-rich hydroxyl groups, which can enhance photocatalytic activity. A simple hydrothermal grafting of CCN onto t-KF was accomplished by employing succinic acid (SA) as a cross-linker. The increase in the H2O2 production rate of CCN-SA/t-KF was 1.80-fold as compared to the pristine g-C3N4. This improvement is ascribed to the formation of the LMCT complex and the presence of carbon doping, which induced a charge transfer from the t-KF ligand to the metal CCN. This study highlights the potential of microtubular kapok fiber as a cost-effective, high-performance cellulose-based LMCT photocatalyst. [1] A. Koizumi et al., submitted.Poster Award NomineePoster Award NomineeKapok Microtubules Incorporated Carbon-Doped Graphitic Carbon Nitride for Photocatalytic H2O2 Production P5-23Does Uncertainty-based Active Learning Work in Materials Science?Ai Koizumi1, Guillaume Deffrennes2, Kei Terayama3,4,5, and Ryo Tamura1,5,61 Center for Basic Research on Materials, National Institute for Materials Science (NIMS)2 Univ. Grenoble Alpes, CNRS3 Graduate School of Medical Life Science, Yokohama City University4 MDX Research Center for Element Strategy, Tokyo Institute of Technology 5 RIKEN Center for Advanced Intelligence Project6 Graduate School of Frontier Sciences, The University of TokyoP5-24Nur Shamimie Nadzwin Hasnan1,2, Mohamad Azuwa Mohamed1, and Lok Kumar Shrestha2,3 1 Universiti Kebangsaan Malaysia2 Research Center for Materials Nanoarchitectonics, National Institute for Materials Science (NIMS)3 University of Tsukuba
元のページ ../index.html#92