{"title":"机器学习辅助金基单原子合金催化剂的发现,用于电化学 NO 还原 NH3 反应","authors":"Hui-Long Jin, Qian-Nan Li, Yun-Yan Tian, Shuo-Ao Wang, Xing Chen, Jie-Yu Liu, Chang-Hong Wang","doi":"10.1007/s12598-024-02833-3","DOIUrl":null,"url":null,"abstract":"<p>Direct electrochemical conversion of NO to NH<sub>3</sub> has attracted widespread interest as a green and sustainable strategy for both ammonia synthesis and nitric oxide removal. However, designing efficient catalysts remains challenging due to the complex reaction mechanism and competing side reactions. Single-atom alloy (SAA) catalysts, which increase the atomic efficiency and the chance to tailor the electronic properties of the active center, have become a frontier in this field. Here, we performed a systematic screening of transition metal-doped Au SAAs (denoted as TM/Au, TM = Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ru, Rh, Pd, Ag and Pt) to find potential catalysts for electrochemical NO reduction reaction (NORR) to NH<sub>3</sub>. By employing a four-step screening strategy based on density functional theory (DFT) calculations, Zn/Au SAA has been identified as a promising NORR catalyst due to its superior structural stability, reaction activity and NH<sub>3</sub> selectivity. The electron-involved steps on Zn/Au are thermodynamically spontaneous, which results in a positive limiting potential (<i>U</i><sub>L</sub>) of 0.15 V. The preferred NO affinity compared to H adatom demonstrates that Zn/Au can effectively suppress the hydrogen evolution reaction. Machine-learning (ML) investigations were adopted to address the uncertainty between the physicochemical properties of SAAs and the NORR performance. We applied an extreme gradient boosting regression (XGBR) algorithm to predict the limiting potentials in terms of the intrinsic features of the reaction site. The coefficient of determination (<i>R</i><sup>2</sup>) is 0.97 for the training set and 0.96 for the test set. The electronic structure analysis combined with a compressed-sensing data-analytics approach further quantitatively verifies the coeffect of d-band center, charge transfer and the radius of doped TM atoms, i.e., features with the highest level of importance determined by the XGBR algorithm. This work provides a theoretical understanding of the complex NORR to NH<sub>3</sub> mechanisms and sheds light on the rational design of SAA catalysts by combining DFT and ML investigations.</p><h3 data-test=\"abstract-sub-heading\">Graphical abstract</h3>\n","PeriodicalId":749,"journal":{"name":"Rare Metals","volume":null,"pages":null},"PeriodicalIF":9.6000,"publicationDate":"2024-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Machine-learning-aided Au-based single-atom alloy catalysts discovery for electrochemical NO reduction reaction to NH3\",\"authors\":\"Hui-Long Jin, Qian-Nan Li, Yun-Yan Tian, Shuo-Ao Wang, Xing Chen, Jie-Yu Liu, Chang-Hong Wang\",\"doi\":\"10.1007/s12598-024-02833-3\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Direct electrochemical conversion of NO to NH<sub>3</sub> has attracted widespread interest as a green and sustainable strategy for both ammonia synthesis and nitric oxide removal. However, designing efficient catalysts remains challenging due to the complex reaction mechanism and competing side reactions. Single-atom alloy (SAA) catalysts, which increase the atomic efficiency and the chance to tailor the electronic properties of the active center, have become a frontier in this field. Here, we performed a systematic screening of transition metal-doped Au SAAs (denoted as TM/Au, TM = Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ru, Rh, Pd, Ag and Pt) to find potential catalysts for electrochemical NO reduction reaction (NORR) to NH<sub>3</sub>. By employing a four-step screening strategy based on density functional theory (DFT) calculations, Zn/Au SAA has been identified as a promising NORR catalyst due to its superior structural stability, reaction activity and NH<sub>3</sub> selectivity. The electron-involved steps on Zn/Au are thermodynamically spontaneous, which results in a positive limiting potential (<i>U</i><sub>L</sub>) of 0.15 V. The preferred NO affinity compared to H adatom demonstrates that Zn/Au can effectively suppress the hydrogen evolution reaction. Machine-learning (ML) investigations were adopted to address the uncertainty between the physicochemical properties of SAAs and the NORR performance. We applied an extreme gradient boosting regression (XGBR) algorithm to predict the limiting potentials in terms of the intrinsic features of the reaction site. The coefficient of determination (<i>R</i><sup>2</sup>) is 0.97 for the training set and 0.96 for the test set. The electronic structure analysis combined with a compressed-sensing data-analytics approach further quantitatively verifies the coeffect of d-band center, charge transfer and the radius of doped TM atoms, i.e., features with the highest level of importance determined by the XGBR algorithm. This work provides a theoretical understanding of the complex NORR to NH<sub>3</sub> mechanisms and sheds light on the rational design of SAA catalysts by combining DFT and ML investigations.</p><h3 data-test=\\\"abstract-sub-heading\\\">Graphical abstract</h3>\\n\",\"PeriodicalId\":749,\"journal\":{\"name\":\"Rare Metals\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":9.6000,\"publicationDate\":\"2024-06-22\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Rare Metals\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://doi.org/10.1007/s12598-024-02833-3\",\"RegionNum\":1,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"MATERIALS SCIENCE, MULTIDISCIPLINARY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Rare Metals","FirstCategoryId":"88","ListUrlMain":"https://doi.org/10.1007/s12598-024-02833-3","RegionNum":1,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATERIALS SCIENCE, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
摘要
将 NO 直接电化学转化为 NH3 作为一种绿色、可持续的合成氨和去除一氧化氮的策略引起了广泛关注。然而,由于复杂的反应机理和相互竞争的副反应,设计高效催化剂仍具有挑战性。单原子合金(SAA)催化剂提高了原子效率,并有机会定制活性中心的电子特性,已成为该领域的前沿技术。在此,我们对掺杂过渡金属的金 SAA(表示为 TM/Au,TM = Sc、Ti、V、Cr、Mn、Fe、Co、Ni、Cu、Zn、Ru、Rh、Pd、Ag 和 Pt)进行了系统筛选,以寻找将 NO 还原成 NH3 的电化学还原反应(NORR)的潜在催化剂。通过采用基于密度泛函理论(DFT)计算的四步筛选策略,Zn/Au SAA 因其优异的结构稳定性、反应活性和 NH3 选择性而被确定为一种有前途的 NORR 催化剂。Zn/Au 上的电子参与步骤在热力学上是自发的,这导致了 0.15 V 的正极限电位 (UL)。我们采用了机器学习(ML)研究来解决 SAAs 的物理化学特性与 NORR 性能之间的不确定性。我们采用极端梯度提升回归(XGBR)算法,根据反应场所的内在特征来预测极限电位。训练集的判定系数 (R2) 为 0.97,测试集的判定系数 (R2) 为 0.96。电子结构分析与压缩感应数据分析方法相结合,进一步定量验证了 d 波段中心、电荷转移和掺杂 TM 原子半径的协同效应,即 XGBR 算法确定的重要性最高的特征。这项工作从理论上理解了复杂的 NORR 转化为 NH3 的机理,并通过结合 DFT 和 ML 研究,为合理设计 SAA 催化剂提供了启示。
Machine-learning-aided Au-based single-atom alloy catalysts discovery for electrochemical NO reduction reaction to NH3
Direct electrochemical conversion of NO to NH3 has attracted widespread interest as a green and sustainable strategy for both ammonia synthesis and nitric oxide removal. However, designing efficient catalysts remains challenging due to the complex reaction mechanism and competing side reactions. Single-atom alloy (SAA) catalysts, which increase the atomic efficiency and the chance to tailor the electronic properties of the active center, have become a frontier in this field. Here, we performed a systematic screening of transition metal-doped Au SAAs (denoted as TM/Au, TM = Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ru, Rh, Pd, Ag and Pt) to find potential catalysts for electrochemical NO reduction reaction (NORR) to NH3. By employing a four-step screening strategy based on density functional theory (DFT) calculations, Zn/Au SAA has been identified as a promising NORR catalyst due to its superior structural stability, reaction activity and NH3 selectivity. The electron-involved steps on Zn/Au are thermodynamically spontaneous, which results in a positive limiting potential (UL) of 0.15 V. The preferred NO affinity compared to H adatom demonstrates that Zn/Au can effectively suppress the hydrogen evolution reaction. Machine-learning (ML) investigations were adopted to address the uncertainty between the physicochemical properties of SAAs and the NORR performance. We applied an extreme gradient boosting regression (XGBR) algorithm to predict the limiting potentials in terms of the intrinsic features of the reaction site. The coefficient of determination (R2) is 0.97 for the training set and 0.96 for the test set. The electronic structure analysis combined with a compressed-sensing data-analytics approach further quantitatively verifies the coeffect of d-band center, charge transfer and the radius of doped TM atoms, i.e., features with the highest level of importance determined by the XGBR algorithm. This work provides a theoretical understanding of the complex NORR to NH3 mechanisms and sheds light on the rational design of SAA catalysts by combining DFT and ML investigations.
期刊介绍:
Rare Metals is a monthly peer-reviewed journal published by the Nonferrous Metals Society of China. It serves as a platform for engineers and scientists to communicate and disseminate original research articles in the field of rare metals. The journal focuses on a wide range of topics including metallurgy, processing, and determination of rare metals. Additionally, it showcases the application of rare metals in advanced materials such as superconductors, semiconductors, composites, and ceramics.