文献类型：期刊文献

英文题名：Knowledge Graph for Methane Selective Conversion: Revisiting and Predicting Product Selectivity and Methane Conversion

作者：Xu, Boyu[1];Li, Gaoyang[2];Wang, Bohan[3,4];Bian, Jiawei[4,5];Pan, Hui[2];Min, Yulin[2];Qi, Guodong[6];Xu, Jun[6];Deng, Feng[6];Ju, Feng[4,5];Ling, Hao[4,5];Wang, Zhendong[7]

机构：[1]Univ Utrecht, Dept Informat & Comp Sci, Princetonpl 5, NL-3584 CC Utrecht, Netherlands;[2]Shanghai Univ Elect Power, Shanghai Key Lab Mat Protect & Adv Mat Elect Power, Shanghai 200090, Peoples R China;[3]East China Univ Sci & Technol, Sch Mat Sci & Engn, Shanghai 200237, Peoples R China;[4]East China Univ Sci & Technol, East China Univ Sci & Technol Utrecht Univ Joint R, Shanghai 200237, Peoples R China;[5]East China Univ Sci & Technol, Sch Chem Engn, Shanghai 200237, Peoples R China;[6]Chinese Acad Sci, Innovat Acad Precis Measurement Sci & Technol, Natl Ctr Magnet Resonance Wuhan, State Key Lab Magnet Resonance Spect & Imaging, Wuhan 430071, Peoples R China;[7]Sinopec Shanghai Res Inst Petrochem Technol Co Ltd, State Key Lab Green Chem Engn & Ind Catalysis, Shanghai 201208, Peoples R China

年份：2025

卷号：12

期号：48

外文期刊名：ADVANCED SCIENCE

收录：;EI(收录号：20254119316344);WOS:【SCI-EXPANDED(收录号:WOS:001587217500001)】；

基金：This work was supported by Sinopec Foundation (224250), the National Key R&D Program of China (2022YFA1504500 and 2023YFA1509102),the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB0540000), the National Natural Science Foundation of China(22378131, 22478239 and 22272185), Shanghai Pujiang Programme(23PJD021), and Science and Technology Commission of Shanghai Municipality (No.19DZ2271100).

语种：英文

外文关键词：deep neural networks; knowledge graph; large language models; methane selective conversion; predicting catalytic performance

摘要：Selective conversion of methane to carbon-based compounds is promising but currently limited by issues related to scale. Unraveling the interconnected relationships between products selectivity and methane conversion plays a pivotal role in understanding of this complex process. In this work, a knowledge graph (KG) is constructed for methane conversion with the help of a robust large language model basing on the literature reported methane conversion over different catalysts under various conditions. This KG, structured around 11 entity types and 32 relationship types, allows to effectively analyze advancements in methane conversion - including identifying optimal catalytic processes, evaluating reaction conditions, and tracking development trends. A deep neural network analysis of the KG highlighted catalysts with metal active sites and multifunctional supports as particularly effective for methanol production under conditions suitable for industrial-scale applications. These findings provide valuable insights for targeted catalyst development and industrial applications.