首 页
滚动信息 更多 >>
本刊2021年SCI影响因子12.256 (2022-07-07)
npj Computational Materials 2019年影响因子达到9... (2020-07-04)
npj Computational Materials获得第一个SCI影响因... (2018-09-07)
英文刊《npj Computational Materials(计算材料学... (2017-05-15)
开放的数据库 (2015-12-17)
快捷服务
最新文章 研究综述
过刊浏览 作者须知
期刊编辑 审稿须知
相关链接
· 在线投稿
会议信息
友情链接
  中国科学院上海硅酸盐研究所
  无机材料学报
  OQMD数据库
mm.jpg
期刊介绍

  《npj 计算材料学》是在线出版、完全开放获取的国际学术期刊。发表结合计算模拟与设计的材料学一流的研究成果。本刊由中国科学院上海硅酸盐研究所与英国自然出版集团(Nature Publishing Group,NPG)以伙伴关系合作出版。
  主编为陈龙庆博士,美国宾州大学材料科学与工程系、工程科学与力学系、数学系的杰出教授。共同主编为陈立东研究员,中国科学院上海硅酸盐研究所研究员高性能陶瓷与超微结构国家重点实验室主任。
  办刊目的与报道范围
  《npj 计算材料学》是在线出版、完全开放获取的国际学术...
【查看详细】
近期文章 更多 >>

Experimental reconstructions of 3D atomic structures from electron microscopy images using a Bayesian genetic algorithm        
Annick De Backer, Sandra Van Aert, Christel Faes, Ece Arslan Irmak, Peter D. Nellist & Lewys Jones   
npj Computational Materials 8:216 (2022)
doi.org/10.1038/s41524-022-00900-w
Published online: 12 October  2022
Abstract| Full Text | PDF OPEN

Abstract: We introduce a Bayesian genetic algorithm for reconstructing atomic models of monotype crystalline nanoparticles from a single projection using Z-contrast imaging. The number of atoms in a projected atomic column obtained from annular dark field scanning transmission electron microscopy images serves as an input for the initial three-dimensional model. The algorithm minimizes the energy of the structure while utilizing a priori information about the finite precision of the atom-counting results and neighbor-mass relations. The results show promising prospects for obtaining reliable reconstructions of beam-sensitive nanoparticles during dynamical processes from images acquired with sufficiently low incident electron doses.

摘要: 本文介绍了一种贝叶斯遗传算法。这种方法可以使用实验所得的Z-衬度图像,一次投影就可以重建单质纳米晶粒的原子模型。以环形暗场扫描模式所得的透射电子显微图像可以预测某个被投影的原子列所含有的原子数目,并进一步用来搭建初始三维模型。这种有限精度的原子计数结果与近邻原子列之间的质量关系(neighbor-mass ralations)组成了本算法中贝叶斯定理所需的先验信息,以便用于后继的结构能量最小化过程。研究结果表明这种算法在可靠地重建处于动态变化中,对电子束敏感,从而其图像所用的入射电子剂量相当低的纳米晶结构上具有广阔的应用前景。 

Editorial Summary

How to build structures using priori knowledge and experience ? 

Monotype metal nanocrystals are important catalysts. Their structures as well as the dynamic structure change are closely related to their catalytic performance. Atomic resolution annular dark field scanning transmission electron microscopy (ADF STEM) can be used to characterize the structures of such materials by combining modeling methods. Although some modeling methods based on atom-counting and energy minimization have been developed, it is still a challenge to find the truly experimental structure, not just the global energy minimum from a purely computational calculation. In this study, a new statistical modeling algorithm is proposed. This method makes full use of single projection image, which is benefit to reconstruct surface structures of nanoparticles more reliably. By using the ADF STEM images of Pt nanocrystals from simulation and experiment, respectively, Professor Sandra's team from EMAT and NANOlab Center of Excellence at University of Antwerp in Belgium, together with the team of Professor Lewys from Advanced Microscopy Laboratory of Center for Research on Adaptive Nanostructures and Nanodevices (CRANN), and School of Physics of Trinity College Dublin at The University College Dublin in Ireland, have introduced Bayes' theorem of probability and statistics into the processing and structural reconstruction of the images. The number of atoms in finite precision for a given atomic column can be inferred by the scattering cross-section distribution based on the images. And then the atom-counting values, together with the neighbor-mass relation extracted from the known crystal structure knowledge, are used as prior information required by Bayes' theorem to calculate the probability of the assumed number of atoms. Looping through all atomic columns in this way yields an initial model used for the followed genetic algorithm search, which tries for the best matching model based on energy minimization. The Bayesian genetic algorithm can overcome the problem of obtaining theoretical rather than experimental structures by using genetic algorithms alone. The introduction of neighbor-mass relations can also improve the accuracy of modeling in comparison to only considering atom-counting. The study not only contributes to reliable reconstructions of experimental structures for beam-sensitive monotype nanoparticles, but also can be beneficial to beam-insensitive nanoparticles difficult for reliable structure reconstruction before, because their images are obtained under high electron dose or multiple projections but containing too much noise during experiments and environmental distortions. 

编辑概述

如何在结构建模中利用已有的知识与经验?

单质金属纳米晶是重要的催化剂。它们的结构及其动态变化与催化性能密切相关。联合建模的手段,分辨率在原子级别的环形暗场扫描透射电子显微成像可用于表征这类材料的结构。虽然目前已经发展了基于原子计数与能量最小化的建模方法,但是如何求取实验结构,而不只是理论计算的全局最稳定结构仍然是一个挑战。本研究提出了一种新的基于统计学的建模算法。这种方法充分利用实验得到的单轴投影图像,有助于真实地重建纳米颗粒的表面结构。来自比利时安特卫普大学材料科学电子显微中心与卓越纳米实验中心的Sandra教授团队同来自爱尔兰自适应纳米结构与纳米机器研究中心先进显微实验室和都柏林大学三一学院物理系的Lewys教授团队合作,分别以Pt纳米晶的模拟和实验数据为例,将概率统计学的贝叶斯定理引入到电子显微图像的处理与结构重建中,根据图像推测散射交叉截面分布,进而获得给定原子列所含的、仍带有偏差的原子个数。这些原子个数值与基于晶体结构提取的近邻质量关系一起作为应用贝叶斯定理所需的先验知识,用来计算该原子列含有所假定原子数的概率。如此遍历所有原子列就得到可用于遗传算法搜索的初始模型,随后仍然基于能量最小化进一步搜索最佳匹配的模型。这种贝叶斯遗传算法避免了单独使用遗传算法时经常给出理论的而不是实验结构的问题,而且近邻质量关系的引入也提高了原先单独考虑原子计数时建模的准确性。该研究不但有助于电子束敏感单质纳米晶粒实验结构的重建,而且也可以用于对电子束不敏感,比如可以在强电子束和多方向投影下获得图像,但是图像包含过多的实验与环境噪声的纳米晶粒结构的可靠表征。

Switchable half-metallicity in A-type antiferromagnetic NiI2 bilayer coupled with ferroelectric In2Se3        
Yaping Wang, Xinguang Xu, Xian Zhao, Weixiao Ji, Qiang Cao, Shengshi Li & Yanlu Li    
npj Computational Materials 8:218 (2022)
doi.org/10.1038/s41524-022-00894-5
Published online: 23 October  2022
Abstract| Full Text | PDF OPEN

Abstract: Electrically controlled half-metallicity in antiferromagnets is of great significance for both fundamental research and practical application. Here, by constructing van der Waals heterostructures composed of two-dimensional (2D) A-type antiferromagnetic NiI2 bilayer (bi-NiI2) and ferroelectric In2Se3 with different thickness, we propose that the half-metallicity is realizable and switchable in the bi-NiI2 proximate to In2Se3 bilayer (bi-In2Se3). The polarization flipping of the bi-In2Se3 successfully drives transition between half-metal and semiconductor for the bi-NiI2. This intriguing phenomenon is attributed to the joint effect of polarization field induced energy band shift and interfacial charge transfer. Besides, the easy magnetization axis of the bi-NiI2 is also dependent on the polarization direction of the bi-In2Se3. The half-metallicity and magnetic anisotropy energy of the bi-NiI2 in heterostructure can be effectively manipulated by strain. These findings provide not only a feasible strategy to achieve and control half-metallicity in 2D antiferromagnets, but also a promising candidate to design advanced nanodevices.

摘要: 反铁磁体中的电控半金属性对基础研究和实际应用都具有重要意义。本文将二维双层A型反铁磁NiI2 材料(bi-NiI2)与不同厚度的铁电In2Se3 材料进行组合,构建了多种范德瓦尔斯异质结,最终在靠近双层In2Se3 (bi-In2Se3)的bi-NiI2上实现了可切换的半金属性。bi-In2Se3的极化翻转成功地驱使bi-NiI2在半金属和半导体之间进行转变,这种有趣的现象是极化场诱导的能带位移和界面电荷转移共同作用的结果。此外,bi-NiI2的易磁化轴也依赖于bi-In2Se3的极化方向。应变可以有效地调控异质结中bi-NiI2的半金属性和磁各向异性能。这些发现不仅为实现和控制二维反铁磁体的半金属性提供一种可行的策略,而且为设计先进的纳米器件提供了一种极具潜力的选择。 

Editorial Summary

The half-metallicity is achievable for the combination of antiferromagnetism and ferroelectricity 

Complex oxides exhibit rich physical phenomena such as Mott insulator, multiferroics and high-temperature superconductivity. And searching for topological states have become one of the most active projects in condensed matter physics. Along (111) direction of perovskite oxide, because of the transition metal atom resides on a buckled honeycomb lattice, these systems are predicted to realize the quantum spin Hall. However, it is very difficult to prepare perovskite oxide grown in (111) direction experimentally, so there has been no significant progress.  A team led by Prof. Hanghui Chen from NYU Shanghai and Prof. Gang Li form ShanghaiTech University proposed a stacking method for the construction of (SrMO3)1/(SrM’O3)1 oxide superlattice in the (001) direction by using a variety of different transition metal perovskite oxides. They found strong topological insulators and Dirac semi-metals in the (001) oxide superlattice through first principles calculations and model analysis. The design principle of this study is to achieve non-banal topological properties through the band inversion of the d orbitals of two different transition metal atoms and a particular parity property of (001) superlattice geometry. Through calculation and analysis, it is found that the superlattice represented by (SrTaO3)1/(SrIrO3)1 has Z2 index of (1; 001) are strong topological insulators. The (SrMoO3)1/(SrIrO3)1 superlattice exhibits multiple coexisting topological insulator and topological Dirac semi-metal states. This study provides a novel and feasible direction for finding topological states in complex oxides. 

编辑概述

当反铁磁遇到铁电:让半金属性成为可能

反铁磁材料由于具有强抗干扰能力、无杂散场、超快动力学等优点,有望为下一代自旋电子学器件带来革命性的进步,调控实现反铁磁材料的半金属性或产生完全自旋极化的电流是将反铁磁材料应用于自旋电子学中的关键。目前,通过电场调控能够实现上述目标,但是由于完全自旋极化的传导电子会随着电场的撤销随之消散,由此得到的半金属性是易失的,这对存储和逻辑器件而言并非理想选择。该研究提出了一种可行的调控方案,通过耦合材料固有的铁电性调控反铁磁材料的电子结构,能够获得非易失性的半金属性。来自山东大学晶体材料国家重点实验室的赵显教授、李妍璐教授研究团队和济南大学自旋电子学研究所的李胜世团队,设计构建了由二维A型反铁磁NiI2双层材料(bi-NiI2)与不同厚度的二维铁电In2Se3材料组成的范德瓦尔斯异质结,通过第一性原理计算方法预测了与双层In2Se3耦合的bi-NiI2能够实现半金属性的产生与切换。当bi-In2Se3的铁电极化由向上变为向下时,bi-NiI2会经历从半金属到半导体的转变,且bi-NiI2的易磁轴会由面内转变为面外,这一现象是极化场诱导的能带移动与层间电荷转移共同作用的结果。此外,异质结中bi-NiI2的半金属性和磁各向异性可以通过应变来进行有效的调控。基于该异质结,作者提出了一种铁电存储器件,其数据读取过程是将铁电层的极化态转变为反铁磁的导电态来进行检测的。该研究为二维反铁磁材料中非易失性电控半金属性提供了一种切实有效的方法,将极大地推动反铁磁自旋电子学的发展。

Emergent topological states via digital (001) oxide superlattices         
Zhiwei Liu, Hongquan Liu, Jiaji Ma, Xiaoxuan Wang, Gang Li & Hanghui Chen     
npj Computational Materials 8:208 (2022)
doi.org/10.1038/s41524-022-00894-5
Published online: 29 September  2022
Abstract| Full Text | PDF OPEN

Abstract: Oxide heterostructures exhibit many intriguing properties. Here we provide design principles for inducing multiple topological states in (001) (AMO3)1/(AM’O3)1 oxide superlattices. Aided by first-principles calculations and model analysis, we show that a (SrMO3)1/(SrM’O3)1 superlattice (M = Nb, Ta and M’ = Rh, Ir) is a strong topological insulator with Z2 index (1;001). More remarkably, a (SrMoO3)1/(SrIrO3)1 superlattice exhibits multiple coexisting topological insulator (TI) and topological Dirac semi-metal (TDS) states. The TDS state has a pair of type-II Dirac points near the Fermi level and symmetry-protected Dirac node lines. The surface TDS Dirac cone is sandwiched by two surface TI Dirac cones in the energy-momentum space. The non-trivial topological properties arise from the band inversion between d orbitals of two dissimilar transition metal atoms and a particular parity property of (001) superlattice geometry. Our work demonstrates how to induce non-trivial topological states in (001) perovskite oxide heterostructures by rational design.

摘要: 氧化物异质结构具有许多有趣的特性。在此,我们提供了在(001) (AMO3)1/(AM’O3)1氧化物超晶格中产生多拓扑态的设计原理。通过第一性原理计算和模型分析,我们得出的结果显示出(SrMO3)1/(SrM’O3)1超晶格(M = Nb, Ta and M’ = Rh, Ir)是Z2指数为(1;001)的强拓扑绝缘体。更特别的是,(SrMoO3)1/(SrIrO3)1超晶格表现出拓扑绝缘体(TI)和狄拉克拓扑半金属(TDS)多态共存现象。TDS态在费米能级附近具有一对II型狄拉克点和受对称性保护的狄拉克“节线”。在能量-动量空间中表面TDS狄拉克锥被两个表面TI狄拉克锥夹在中间。非平庸的拓扑性质是由两个不同过渡金属原子的d轨道能带反转以及(001)超晶格几何构造特别的宇称性质产生的。我们的工作展示了如何通过合理的设计在(001)钙钛矿氧化物异质结构中产生非平庸的拓扑态。 

Editorial Summary

Searching for topological states in complex oxides 

Complex oxides exhibit rich physical phenomena such as Mott insulator, multiferroics and high-temperature superconductivity. And searching for topological states have become one of the most active projects in condensed matter physics. Along (111) direction of perovskite oxide, because of the transition metal atom resides on a buckled honeycomb lattice, these systems are predicted to realize the quantum spin Hall. However, it is very difficult to prepare perovskite oxide grown in (111) direction experimentally, so there has been no significant progress. 
A team led by Prof. Hanghui Chen from NYU Shanghai and Prof. Gang Li form ShanghaiTech University proposed a stacking method for the construction of (SrMO3)1/(SrM’O3)1 oxide superlattice in the (001) direction by using a variety of different transition metal perovskite oxides. They found strong topological insulators and Dirac semi-metals in the (001) oxide superlattice through first principles calculations and model analysis. The design principle of this study is to achieve non-banal topological properties through the band inversion of the d orbitals of two different transition metal atoms and a particular parity property of (001) superlattice geometry. Through calculation and analysis, it is found that the superlattice represented by (SrTaO3)1/(SrIrO3)1 has Z2 index of (1; 001) are strong topological insulators. The (SrMoO3)1/(SrIrO3)1 superlattice exhibits multiple coexisting topological insulator and topological Dirac semi-metal states. This study provides a novel and feasible direction for finding topological states in complex oxides. 

编辑概述

氧化物中的拓扑态:要如何寻觅?

复杂氧化物表现出非常丰富的物理现象,包括莫特绝缘性、多铁性和高温超导。另一方面,寻找非平庸拓扑态已经成为凝聚态物理领域最热门的课题之一。因过渡金属原子在(111)方向特有的蜂巢结构,此类钙钛矿氧化物体系被预言可实现量子自旋霍尔态。但由于在实验上制备(111)方向生长的钙钛矿氧化物非常困难,因此一直没有明显进展。来自上海纽约大学的陈航晖教授团队和来自上海科技大学的李刚教授,基于多种不同过渡金属钙钛矿氧化物,提出了一种在(001)方向堆叠构建(SrMO3)1/(SrM’O3)1氧化物超晶格的方法。他们通过第一性原理计算以及模型分析,在氧化物超晶格(001)晶向上发现了强拓扑绝缘体和狄拉克半金属。作者的设计原则是,通过两种不同的过渡金属原子d轨道的能带反转,以及(001)氧化物超晶格几何构造特别的宇称性质,来实现非平庸的拓扑性质。经过计算和分析发现,以(SrTaO3)1/(SrIrO3)1为代表的超晶格是具有Z2指数为(1;001)的强拓扑绝缘体。(SrMoO3)1/(SrIrO3)1超晶格更是存在强拓扑绝缘体和拓扑狄拉克半金属的共存相。作者的研究为在复杂氧化物中寻找拓扑态提供了一种新颖可行的方向。

Linking atomic structural defects to mesoscale properties in crystalline solids using graph neural networks         
Zhenze Yang & Markus J. Buehler     
npj Computational Materials 8:198 (2022)
doi.org/10.1038/s41524-022-00879-4
Published online: 17 September  2022
Abstract| Full Text | PDF OPEN

Abstract: Structural defects are abundant in solids, and vital to the macroscopic materials properties. However, a defect-property linkage typically requires significant efforts from experiments or simulations, and often contains limited information due to the breadth of nanoscopic design space. Here we report a graph neural network (GNN)-based approach to achieve direct translation between mesoscale crystalline structures and atom-level properties, emphasizing the effects of structural defects. Our end-to-end method offers great performance and generality in predicting both atomic stress and potential energy of multiple systems with different defects. Furthermore, the approach also precisely captures derivative properties which strictly observe physical laws and reproduces evolution of properties with varying boundary conditions. By incorporating a genetic algorithm, we then design de novo atomic structures with optimum global properties and target local patterns. The method would significantly enhance the efficiency of evaluating atomic behaviors given structural imperfections and accelerates the design process at the meso-level.

摘要: 结构缺陷在固体中很丰富,对宏观材料性能至关重要。然而,缺陷-属性链接通常需要大量的实验或模拟工作,并且由于纳米级设计空间的广度,通常包含有限的信息。在本文红,我们报告了一种基于图像神经网络 (GNN) 的方法来实现介观尺度晶体结构和原子级特性之间的直接转换,强调结构缺陷的影响。我们的端到端方法在预测具有不同缺陷的多个系统的原子应力和势能方面提供了出色的性能和通用性。此外,该方法还精确地捕获了严格遵守物理定律的衍生属性,并再现了具有不同边界条件的属性演变。通过结合遗传算法,我们随后重新设计具有最佳全局属性和目标局部模式的原子结构。该方法将显着提高在给定结构缺陷的情况下评估原子行为的效率,并加速介观层次的设计过程。 

Editorial Summary

Graph neural network: bridging the gap between atomic structural defects and mesoscale properties

The structural defects of materials are inevitable, and are crucial to the performance of a variety of materials. Not only the defects themselves, but also their atomic level distribution will affect the local and global properties of the crystal. At present, multi-scale modeling methods from quantum level to continuum level have been developed to calculate the effects of structural defects and reveal the mechanism behind experimental observations. However, due to the heterogeneity introduced by material defects, the design space of defect entities usually contains a large number of possible structures, and simulation may be expensive and time-consuming, especially when the system size increases dramatically. The emergence of machine learning (ML) methods, especially deep learning (DL), is a possible solution. However, at present, these machine learning based models either focus on small crystal structures or only predict single attributes. In order to overcome these difficulties, Professor Markus J. Buehler's team from the Atomic and Molecular Mechanics Laboratory of Massachusetts Institute of Technology introduced a general method to directly convert the crystal structure represented by a graph with spatial information into atomic level attributes, such as atomic stress field or potential energy distribution. The performance of this method is demonstrated by testing models on several large crystal systems, including 2D graphene and 3D aluminum systems with different types of structural defects and target atomic properties. The proposed method achieves high accuracy, and captures the physical information extracted from atomic prediction in all data sets studied as a potential alternative to expensive molecular simulation. The model is further combined with an optimization algorithm to screen designs with low stress concentrations and specific local stress patterns. This method shows the high precision, versatility and diversity of the transformation between structure and attribute on the atomic scale. The ideas presented here can also be applied to other applications in scientific and engineering problems, such as the magnetic field of a spin system, the electronic density in molecules, and the mechanical state of the structure, etc.

编辑概述

微观介观交流忙,图像神经网络来帮忙

材料的结构缺陷是不可避免的,并且对多种材料的性能至关重要。不仅缺陷本身,而且它们的原子级分布都会影响晶体的局部和全局性质。目前,已经开发了从量子水平到连续体水平的多尺度建模方法,以计算结构缺陷的影响并揭示实验观察背后的机制。然而,由于材料缺陷引入的异质性,缺陷实体的设计空间通常包含大量可能的结构,模拟都可能既昂贵又耗时,尤其是在系统规模激增时。机器学习 (ML) 方法的出现,尤其是深度学习 (DL),是一个可能的解决方案,然而,目前这些基于机器学习的模型要么专注于小晶体结构,要么只预测单一属性。为了克服这些困难,来自美国麻省理工学院原子与分子力学实验室的 Markus J. Buehler教授团队引入了一种通用方法,将由具有空间信息的图形表示的晶体结构直接转换为原子级属性,例如原子应力场或势能分布。通过在多个大型晶体系统上测试模型来展示该方法的性能,包括具有不同类型结构缺陷和目标原子特性的 2D 石墨烯和 3D 铝系统。所提出的方法实现了高精度,并在研究的所有数据集中捕获了从原子预测中提取的物理信息,作为昂贵的分子模拟的潜在替代方案。该模型进一步与优化算法相结合,以筛选具有低应力集中和特定局部应力模式的设计。本方法在原子尺度上显示了结构和属性之间转换的高精度、通用性和多样性。这里提出的想法也可以应用于科学和工程问题中的其他应用,例如自旋系统的磁场、分子中的电子密度以及架构结构的机械状态等。

Intrinsic hard magnetism and thermal stability of a ThMn12-type permanent magnet         
Tumentsereg Ochirkhuyag, Soon Cheol Hong & Dorj Odkhuu     
npj Computational Materials 8:193 (2022)
doi.org/10.1038/s41524-022-00821-8
Published online: 09 September  2022
Abstract| Full Text | PDF OPEN

Abstract: Herein, we theoretically demonstrate that simple metal (Ga and Al) substitutional atoms, rather than the conventional transition metal substitutional elements, not only stabilize the ThMn12-type SmFe12 and Sm(Fe,Co)12 phases thermodynamically but also further improve their intrinsic magnetic properties such that they are superior to those of the widely investigated SmFe11Ti and Sm(Fe,Co)11Ti magnets, and even to the state-of-the-art permanent magnet Nd2Fe14B. More specifically, the quaternary Sm(Fe,Co,Al)12 phase has the highest uniaxial magnetocrystalline anisotropy (MCA) of about 8?MJ?m?3, anisotropy field of 18.2?T, and hardness parameter of 2.8 at room temperature and a Curie temperature of 764?K. Simultaneously, the Al and Ga substitutional atoms improve the single-domain size of the Sm(Fe,Co)12 grains by nearly a factor of two. Numerical results of MCA and MCA-driven hard magnetic properties can be described by the strong spin-orbit coupling and orbital angular momentum of the Sm 4f-electron orbitals.

摘要: 在本文中,我们从理论上证明,简单的金属(Ga 和 Al)取代原子,而不是传统的过渡金属取代元素,不仅在热力学上稳定了 ThMn12 型 SmFe12 和 Sm(Fe,Co)12 相,而且进一步提高了它们的固有磁性。性能使其优于广泛研究的 SmFe11Ti 和 Sm(Fe,Co)11Ti 磁体,甚至优于最先进的永磁体 Nd2Fe14B。具体而言,四元 Sm(Fe,Co,Al)12 相具有最高的单轴磁晶各向异性 (MCA),约为 8?MJ?m-3,各向异性场为 18.2?T,室温下的硬度参数为 2.8,居里温度为 764?K。同时,Al 和 Ga 取代原子将 Sm(Fe,Co)12 晶粒的单畴尺寸提高了近两倍。 MCA 和 MCA 驱动的硬磁特性的数值结果可以通过 Sm 4f 电子轨道的强自旋轨道耦合和轨道角动量来描述。

Editorial Summary

Simplicity is beauty: simple metal stronger than transition metal

ThMn12 alloy is a kind of potential high performance permanent magnet with intrinsic hard magnetic properties, and has a broad application prospect. In particular, the thermodynamically stable large-scale production of SmFe12 single crystal has a large demand for industrial applications (such as motors and generators), but it is very difficult to obtain single crystal phase of ThMn12 structure at present. In order to stabilize the structure of ThMn12, a third alternative metal element, including Ti or V, is essential. However, the doping of these transition metal (TM) elements seriously reduces the intrinsic magnetism. On the other hand, in order to maximize the permanent hard magnetic properties, the grain size of SmFe12 based magnets must be close to the single domain (SD) size (~51-54 nm). However, it is quite difficult to prepare nanometer sized ThMn12 type SmFe12 in actual samples. This problem must be solved to make full use of SmFe12 as a practical high-performance permanent magnet. Professor Dorj Odkhuu from Incheon University in South Korea and the collaborators of Ulsan University proposed a possible solution. Through the first principle density functional theory (DFT), density functional perturbation theory (DFPT) and Monte Carlo (MC) simulation of the system, it was found that simple metal (SM) Al and Ga replacement atoms, compared with traditional TM replacement elements, thermodynamically stabilized the ThMn12 type Sm(Fe, Co)12 structure, At the same time, the grain size of SD was improved and the magnetism was enhanced. The intrinsic hard magnetic properties of the quaternary Sm (Fe, Co, Al)12 and Sm (Fe, Co, Ga)12 compounds proposed in this study at high temperatures are superior to the widely studied SmFe11Ti and Sm (Fe, Co)11Ti compounds. This work still solves the main problems of the structure and thermal instability of ThMn12 from the mechanism, thus providing theoretical guidance for the practical application of SmFe12 based high-performance permanent magnets.

编辑概述

主族胜于过渡:既稳定又细微的高性能永磁体

ThMn12型合金具有固有硬磁特性,是一类潜在的高性能永磁体,应用前景广阔。目前热力学稳定的ThMn12型SmFe12 单晶在电机工业方面需求量极大,但样品难以稳定。为此此人们对其掺杂了包括 Ti 或 V在内的第三种金属元素,可掺杂如Ti或V这样的过渡金属元素又会严重降低ThMn12型合金的固有磁性。同时,为最大限度地提高永久硬磁性能,SmFe12基磁体的晶粒尺寸还必须接近纳米级的单畴(SD)尺寸(~51-54 nm)。遗憾的是,目前制备纳米级ThMn12型SmFe12单晶还相当困难。SmFe12用作高性能永磁体似乎道阻且长。来自韩国仁川大学的Dorj Odkhuu教授等,通过第一性原理密度泛函理论 (DFT)、密度泛函微扰理论 (DFPT) 和蒙特卡罗 (MC) 模拟发现,与传统的 过渡金属置换元素相比,简单的金属Al 和 Ga 置换原子在热力学上不仅稳定了 ThMn12 型 Sm(Fe,Co)12 结构,同时还改善了 SD 晶粒尺寸、增强了磁性。作者提出的四元 Sm(Fe,Co,Al)12 和 Sm(Fe,Co,Ga)12 化合物在高温下的固有硬磁性能优于正被广泛探索的 SmFe11Ti 和Sm(Fe,Co)11Ti,从机理上解决了 ThMn12的结构和热不稳定性的主要问题,从而为SmFe12基高性能永磁体的实际应用提供了理论指导。

Superconductivity and topological aspects of two-dimensional transition-metal monohalides         
Wen-Han Dong, Yu-Yang Zhang, Yan-Fang Zhang, Jia-Tao Sun, Feng Liu & Shixuan Du    
npj Computational Materials 8:185 (2022)
doi.org/10.1038/s41524-022-00871-y
Published online: 30 August  2022
Abstract| Full Text | PDF OPEN

Abstract: Two-dimensional (2D) superconducting states have attracted much recent interest, especially when they coexist with nontrivial band topology which affords a promising approach towards Majorana fermions. Using first-principles calculations, we predict van der Waals monolayered transition-metal monohalides MX (M = Zr, Mo; X = F, Cl) as a class of 2D superconductors with remarkable transition temperature (5.9–12.4 K). Anisotropic Migdal-Eliashberg theory reveals that ZrCl have a single superconducting gap Δ ~ 2.14 meV, while MoCl is a two-gap superconductor with Δ ~ 1.96 and 1.37 meV. The Z2 band topology of 2D MX is further demonstrated that MoF and MoCl are candidates for realizing topological superconductivity. Moreover, the Dirac phonons of ZrCl and MoCl contribute w-shape phononic edge states, which are potential for an edge-enhanced electron-phonon coupling. These findings demonstrate that 2D MX offers an attractive platform for exploring the interplay between superconductivity, nontrivial electronic and phononic topology.

摘要: 二维超导态最近引起了人们的极大兴趣,而二维超导与非平庸拓扑态的共存为研究马约拉那费米子提供了一种有前景的方法。通过第一性原理计算,我们预测范德华单层的过渡金属单卤化物MX(M = Zr, Mo; X = F, Cl)是一类超导转变温度为5.9–12.4 K的二维超导体。基于各向异性米格达尔-埃利亚什伯格理论的研究表明,ZrCl具有单超导能隙Δ ~ 2.14 meV,而MoCl是具有Δ ~ 1.96和1.37 meV的双能隙超导体。进一步计算表明二维MX家族具有Z2能带拓扑,且MoF和MoCl是拓扑超导体的候选材料。此外,ZrCl和MoCl的狄拉克声子贡献了w型的声子边界态,可能导致边界增强的电子-声子耦合。这些发现表明二维MX家族为探索超导性、非平庸电子和声子拓扑之间的相互作用提供了一个有吸引力的平台。

Editorial Summary

Two-dimensional transition-metal monohalides: rich superconducting and topological states

Due to the development of thin film fabrication technics, 2D superconductors have received continuous attention from the scientific community in recent years. 2D superconducting materials have revealed rich physics, such as Ising pairing, quantum critical effect, and interface-induced high superconducting transition temperature Tc. On one hand, van der Waals materials have great advantages in practical applications due to their weak interaction with the substrate and transferability, but the known intrinsic 2D van der Waals superconductors have limited types and most of them exhibit low Tc. On the other hand, the coexistence of superconductivity and topology conduces to the exploration of topological superconductivity, boundary-enhanced electron phonon coupling, and topological phonon mediated superconductivity, etc. It is fundamentally interesting and important to study 2D van der Waals superconductors with high Tc and topological properties.
Based on first principles calculations, Professor Shixuan Du's team from the Institute of Physics, Chinese Academy of Sciences and Professor Feng Liu from the University of Utah have coordinated to predict a class of 2D van der Waals materials with rich superconducting and topological properties, i.e., transition-metal monohalides MX (M = Zr, Mo; X = F, Cl). The authors discovered that the strong electron-phonon coupling and Tc (5.9-12.4 K) of MX family are caused by acoustic soft modes, and these soft phonon modes originate from mechanism of either Fermi surface nesting or latent lattice instability. The differences of MX family in Fermi surface compositions lead to the characteristics of either single superconducting gap or two superconducting gaps. With respect to the electronic and phononic topologies, MX family contains both candidates of intrinsic topological superconductors and 2D Dirac phonon contributed w-shaped edge states. In addition, the Janus structure with breaking inversion symmetry is expected to reveal chiral phonon related enhancement of superconductivity. This work provides a new idea for the study of 2D superconductivity, topological states and chiral phonons in a single material platform. 

编辑概述

二维过渡金属单卤化物:丰富的超导和拓扑态

由于薄层材料制备技术的发展,二维超导体近年来受到科学界的持续关注,为揭示伊辛配对、量子临界效应以及界面产生的高温超导机制等提供了研究平台。一方面,范德华材料由于层间相互作用弱且易转移而在实际应用中极具优势,但目前已知的本征二维范德华超导体种类有限且大多Tc较低。另一方面,超导性和拓扑性的共存有助于探究拓扑超导、边界显著增强的电子-声子耦合以及拓扑声子介导的超导等方面。研究同时具备较高超导转变温度(Tc)和拓扑性质的新型二维范德华超导体在凝聚态物理与材料科学领域都具有重要性。来自中国科学院物理研究所的杜世萱教授团队与美国犹他大学的刘锋教授合作,通过第一性原理计算预言了一类具有丰富超导和拓扑性质的二维范德华材料:过渡金属单卤化物MX(M = Zr, Mo; X = F, Cl)。研究表明,MX家族由声学支软模导致较强的电声耦合和Tc(5.9-12.4 K),而这些声子软模来源于费米面嵌套或潜在晶格不稳定性。MX家族费米面构成的差异使得其呈现出单超导能隙或双超导能隙特征。此外,MX家族都具有非平庸的电子拓扑不变量Z2 = 1,且MoF 和MoCl为拓扑超导体的候选材料。ZrCl和MoCl在布里渊区边界存在二维狄拉克声子,相应的声子边界态呈现出特殊的w型色散,表明在一维锯齿形纳米带中存在潜在的边界增强电声耦合。有趣的是,打破空间反演的Janus结构Zr2FCl(M2XY)显示出类旋子式(roton-like)的声子软化和超导增强。该工作为研究单一材料平台中的二维超导、拓扑态以及手性声子提供了新思路。

Excitation and detection of coherent sub-terahertz magnons in ferromagnetic and antiferromagnetic heterostructures         
Shihao Zhuang and Jia-Mian Hu.     
npj Computational Materials 8:167 (2022)
doi.org/10.1038/s41524-022-00851-2
Published online: 11 August  2022
Abstract| Full Text | PDF OPEN

Abstract: Excitation of coherent high-frequency magnons (quanta of spin waves) is critical to the development of high-speed magnonic devices. Here we computationally demonstrate the excitation of coherent sub-terahertz (THz) magnons in ferromagnetic (FM) and antiferromagnetic (AFM) thin films by a photoinduced picosecond acoustic pulse. Analytical calculations are also performed to reveal the magnon excitation mechanism. Through spin pumping and spin-charge conversion, these magnons can inject sub-THz charge current into an adjacent heavy-metal film which in turn emits electromagnetic (EM) waves. Using a dynamical phase-field model that considers the coupled dynamics of acoustic waves, spin waves, and EM waves, we show that the emitted EM wave retains the spectral information of all the sub-THz magnon modes and has a sufficiently large amplitude for near-field detection. These predictions indicate that the excitation and detection of sub-THz magnons can be realized in rationally designed FM or AFM thin-film heterostructures via ultrafast optical-pump THz-emission-probe spectroscopy.

摘要: 高频磁振子(自旋波的量子)的激发对于高速磁振子器件的发展至关重要。在此工作中,我们通过计算证明了光致皮秒声波脉冲对铁磁和反铁磁薄膜中亚太赫兹磁振子的激发,并分析了磁振子的激发机制。通过自旋泵浦效应和自旋电荷转换,这些磁振子可以将亚太赫兹电荷电流注入到相邻的重金属薄膜,进而发射电磁波。利用考虑了声波、自旋波和电磁波互相耦合的动态相场模型,我们证明了所发射的电磁波保留了所有亚太赫兹磁振子模式的频谱信息,并且其近场强度足以被检测到。这些预测表明,亚太赫兹磁振子的激发和检测可以通过超快光泵太赫兹频谱仪在铁磁或反铁磁薄膜异质结构中实现。

Editorial Summary

How can we see nanometer spin waves?

In experiments, ultrafast time-resolved magneto-optical Kerr (TR-MOKE) microscopy is commonly used to probe the time-dependent change of magnetization in ferromagnets. However, because the wavelength of the sub-terahertz (0.1-1×1012 Hz) spin wave is close to the penetration depth of the probe laser pulse, it is difficult to detect by this commonly used method. This work demonstrates that, using spin-charge current conversion in heavy metal films, sub-terahertz spin waves can be detected via electromagnetic waves emitted by the alternating charge currents. Prof. Jiamian Hu and his PhD student Shihao Zhuang from the Department of Materials Science and Engineering, University of Wisconsin, USA, obtained the frequencies of the standing spin waves in ferromagnetic and antiferromagnetic thin films using analytical calculation. Through dynamical phase-field simulations, it is demonstrated that a single picosecond acoustic pulse can excite sub-terahertz spin waves in ferromagnetic and antiferromagnetic thin films. The excited magnetic moments pump spin currents into the adjacent heavy metal thin film. Via the inverse spin Hall effect, the spin currents in the heavy metal films are converted into alternating charge currents which emit electromagnetic waves. The study found that the emitted electromagnetic waves retain the spectral information of all excited sub-terahertz spin waves and are strong enough to be detected. This will provide a basis for the excitation and detection of terahertz spin waves and their applications in high-speed magnon devices. In addition, the computational model used in this study considers the full coupling between acoustic, spin, and electromagnetic waves for the first time, and can be used to accurately model physics of ultrafast magnon-phonon-photon coupling in more complex ferromagnetic or antiferromagnetic thin-film-based heterostructures (such as superlattices) as well as other physical processes and devices that involve phonon-magnon-photon coupling such as cavity magnonics and mechanical antennas.

编辑概述

怎样才能”看见”纳米自旋波?

在实验中,人们通常利用时间分辨的磁光克尔效应(TR-MOKE)显微镜来探测铁磁体中磁化强度随时间的变化。然而,因为亚太赫兹(0.1-1×1012 Hz)自旋波的波长与探测激光脉冲的穿透深度相近, 其难以被此常用的方法所探测到。该研究证明了,利用重金属薄膜中的自旋电流转换,亚太赫兹自旋波可以通过交变电流所产生的电磁波间接地被探测到。来自美国威斯康星大学材料科学与工程系的胡嘉冕教授及其博士生庄世豪利用理论解析得到了在铁磁和反铁磁薄膜中自旋波驻波的频率。并通过模拟计算,证明了单个皮秒声波脉冲可以在铁磁和反铁磁薄膜中激发出亚太赫兹的自旋波。被激发的磁矩会向相邻重金属薄膜泵入自旋流。由于逆自旋霍尔效应,重金属薄膜中的自旋流会被转化成交变的电荷电流并发射电磁波。该研究发现所发射的电磁波保留了所有被激发的亚太赫兹自旋波的频谱信息并且其强度足够被检测到,这将为太赫兹自旋波的激发和探测,及其在高速磁振子器件中的应用提供理论依据。另外,该研究所用的计算模型首次考虑了声波、自旋波和电磁波之间的互相耦合,并且可被用于更复杂的铁磁或反铁磁薄膜基异质结构(例如超晶格)以及其他涉及声子-磁振子-光子的物理过程和设备中(例如谐振腔和天线),以准确模拟超快磁振子-声子-光子耦合物理过程。

Superior printed parts using history and augmented machine learning         
Meng Jiang, Tuhin Mukherjee, Yang Du & Tarasankar DebRoy    
npj Computational Materials 8:184 (2022)
doi.org/10.1038/s41524-022-00866-9
Published online: 23 August  2022
Abstract| Full Text | PDF OPEN

Abstract: Machine learning algorithms are a natural fit for printing fully dense superior metallic parts since 3D printing embodies digital technology like no other manufacturing process. Since traditional machine learning needs a large volume of reliable historical data to optimize many printing variables, the algorithm is augmented with human intelligence derived from the rich knowledge base of metallurgy and physics-based models. The augmentation improves the computational efficiency and makes the problem tractable by enabling the algorithm to use a small set of data. We provide a verifiable quantitative index for achieving fully dense superior parts, facilitate material selection, uncover the hierarchy of important variables that affect the density, and present easy-to-use visual process maps. These findings can improve the quality consistency of 3D printed parts that now limit their greater industrial adaptation. The approach used here can be applied to solve other problems of 3D printing and beyond.

摘要: 因为 3D 打印体现了与其他制造工艺不同的数字技术,所以机器学习算法非常适合打印完全致密的优质金属零件。传统的机器学习需要大量可靠的历史数据来优化很多打印变量,本工作算法被人类智能增强,这些智能源自丰富的冶金知识库和基于物理模型。该增强通过使算法能够运用小数据集,提高了计算效率并使问题易于处理。我们提供了一个可验证的量化指标,来获得完全致密的优质零件,加速材料选择,揭示影响密度的重要变量的层次,并提供易于使用的可视化工艺图。这些发现可以提高 3D 打印零件的质量一致性,打印零件的质量不一致性限制了它们更大的工业适应性。本工作使用的方法可用于解决 3D 打印及其他方面的其他问题。

Editorial Summary

Superior printed parts:augmented machine learning with human intelligence

The quality inconsistency of 3D printed parts limits their greater industrial applications. Machine learning can uncover the correlation between various process variables and lack of fusion, but traditional machine learning needs a large volume of reliable historical data to optimize many printing variables. This work employed the augmented machine learning with human intelligence to tackle a small set of data to achieve superior printed parts by reducing the lack of fusion voids.  A research group led by Meng Jiang from State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, China, Department of Materials Science and Engineering, The Pennsylvania State University, USA, implemented the augmented machine learning strategy and synergistically combined a mechanistic model and historical experimental data to reveal the conditions necessary to reduce the lack of fusion void formation in laser powder bed fusion (PBF-L). They identified five important mechanistic variables according to the rich knowledge base of metallurgy and physics-based models. Based on these variables, decision tree and linear regression predicted the lack of fusion with 93% and 90% accuracy, respectively. In addition, they derived the same hierarchical importance of the mechanistic variables on the lack of fusion by using three feature selection indexes, information gain, information gain ratio, and Gini index. Especially, they provided a verifiable lack of fusion index for achieving fully dense superior parts and presented easy-to-use visual process maps. This strategy can improve the quality consistency of 3D printed parts, facilitate materials selection, support the discovery of new printable alloys, and is equally attractive to solve important problems of other manufacturing processes. 

编辑概述

优质打印零件:人类智能增强机器学习

3D打印零件的质量存在不一致性,这限制其进一步工业应用。机器学习可以揭示工艺参数和未熔合关系,但是传统机器学习需要大量可靠的历史数据优化很多打印变量。该研究运用人类智能的增强机器学习处理小数据集,通过减少未熔合空隙获得优质打印零件。来自中国哈尔滨工业大学先进焊接与连接国家重点实验室和美国宾夕法尼亚州立大学材料科学与工程系的Jiang等,运用增强机器学习策略,协同结合机理模型和历史实验数据,揭示了减少激光粉末床熔合 (PBF-L) 中未熔合空隙形成的必要条件。他们根据冶金知识库和物理模型的人类智能,确定了5个影响未熔合缺陷的重要机理变量。基于这些变量,决策树和线性回归预测未熔合的准确率分别达到 93% 和 90%。此外,他们运用信息增益、信息增益比和基尼指数三个特征选择指标,得到相同的未熔合机制变量的层次重要性。尤为重要的是,他们提出一个可验证的未熔合指数来获得完全致密的优质零件,并提供易于使用的可视化工艺图。该研究策略可以提高 3D 打印零件的质量一致性,有助于材料选择,支持新的可印刷合金的研发,并且对于解决其他制造工艺的重要问题同样具有吸引力。

Generative design of stable semiconductor materials using deep learning and density functional theory         
Edirisuriya M. Dilanga Siriwardane, Yong Zhao, Indika Perera & Jianjun Hu     
npj Computational Materials 8:164 (2022)
doi.org/10.1038/s41524-022-00850-3
Published online: 4 August  2022
Abstract| Full Text | PDF OPEN

Abstract: Semiconductor device technology has greatly developed in complexity since discovering the bipolar transistor. In this work, we developed a computational pipeline to discover stable semiconductors by combining generative adversarial networks (GAN), classifiers, and high-throughput first-principles calculations. We used CubicGAN, a GAN-based algorithm for generating cubic materials and developed a classifier to screen the semiconductors and studied their stability using first principles. We found 12 stable AA′′MH6 semiconductors in the F-43m space group including BaNaRhH6, BaSrZnH6, BaCsAlH6, SrTlIrH6, KNaNiH6, NaYRuH6, CsKSiH6, CaScMnH6, YZnMnH6, NaZrMnH6, AgZrMnH6, and ScZnMnH6. Previous research reported that five AA′′IrH6 semiconductors with the same space group were synthesized. Our research shows that AA′′MnH6 and NaYRuH6 semiconductors have considerably different properties compared to the rest of the AA′′MH6 semiconductors. Based on the accurate hybrid functional calculations, AA′′MH6 semiconductors are found to be wide-bandgap semiconductors. Moreover, BaSrZnH6 and KNaNiH6 are direct-bandgap semiconductors, whereas others exhibit indirect bandgaps.

摘要: 双极晶体管问世以来,半导体器件技术的复杂性有了很大的发展。在本工作中,我们开发了一种计算管道,通过结合生成对抗网络 (GAN)、分类器和高通量第一性原理计算来发现稳定半导体。我们使用一种基于 GAN 的算法的CubicGAN算法生成立方材料,并开发一个分类器来筛选半导体,然后使用第一原理研究其稳定性。我们在 F-43m 空间群中发现了 12 种稳定的 AA′′MH6 半导体,包括 BaNaRhH6、BaSrZnH6、BaCsAlH6、SrTlIrH6、KNaNiH6、NaYRuH6、CsKSiH6、CaScMnH6、YZnMnH6、NaZrMnH6、AgZrMnH6 和 ScZnMnH6。前人研究报道合成了五种同一空间群的 AA′′IrH6 半导体。我们研究表明:AA′′MnH6 和 NaYRuH6 半导体与其他 AA′′MH6 半导体相比具有显著不同的性质。精确的杂化泛函计算发现 AA′′MH6 半导体是宽带隙半导体。其中,BaSrZnH6 和 KNaNiH6 是直接带隙半导体,其余是间接带隙半导体。

Editorial Summary

Stable semiconductor materials: generative design

Semiconductor materials are essential components of modern devices, such as electronic, photovoltaic and optoelectronic devices. However, semiconductors with various properties are required for different industrial applications. Therefore, computational approaches for exploring semiconductors are essential to enhance future technologies. This work developed a computational pipeline to discover stable semiconductors by combining generative adversarial networks (GAN), classifiers, and high-throughput first-principles calculations. A team led by Prof. Jianjun Hu from Department of Computer Science and Engineering, University of South Carolina, USA, developed a binary classifier to filter the semiconductors/Insulators (nonmetals) from the dynamically stable quaternary Cubic materials discovered using the CubicGAN model, and then studied the stability using first principles. This work found 12 stable cubic AA′′MH6 semiconductors in the F-43m space group. The DFT calculation results indicate that: 1) Compared with other AA′′MH6 materials, AA′′MnH6 and NaYRuH6 have higher Cii (i = 1, 2, 3) elastic constants, bulk modulus, shear modulus, and Young’s modulus. 2) At temperatures less than 200 K, AA′′MnH6 and NaYRuH6 have lower specific thermal capacity (Cv). 3)The highest Cv at 300 K found in this work is from BaSrZnH6 (127.96 JK?1mol?1). 4) All AA′′MH6 materials are wide-bandgap semiconductors. Among them, BaSrZnH6 and KNaNiH6 are direct-bandgap semiconductors, others exhibit indirect bandgaps. Moreover, the most important elemental and electronic properties were explored. This work will be useful in the development of optical and high-temperature power devices.

编辑概述

稳定半导体材料: 生成设计

半导体材料是电子、光伏和光电子器件等现代设备的重要组成部分。然而,不同工业应用需要具有不同性质的半导体材料。因此,探索半导体的计算方法对于增强未来技术至关重要。本研究提出一种结合生成对抗网络 (GAN)、分类器和高通量第一性原理计算的计算通道,实现了稳定半导体材料的生成设计。来自美国南卡罗来纳大学计算机科学与工程系的Jianjun Hu教授团队开发了一个二元分类器,从CubicGAN模型生成立方材料过滤半导体/绝缘体(非金属),然后使用第一原理研究其稳定性。他们的研究在 F-43m 空间群中发现了 12 种稳定的 AA′′MH6 立方半导体材料。DFT计算结果表明:1)相对于其他 AA''MH6 材料的力学性能,AA''MnH6 和 NaYRuH6具有更高的 Cii (i = 1, 2, 3) 弹性常数、体积模量、剪切模量和杨氏模量;2)低于 200 K时, AA''MnH6 和 NaYRuH6 具有更低的比热容 (Cv) ;3)300 K 时,BaSrZnH6具有最高比热容 (127.96 JK-1mol-1)。4)所有AA′′MH6 半导体都是宽带隙半导体。其中,BaSrZnH6 和 KNaNiH6 是直接带隙半导体,其余是间接带隙半导体。此外,本工作还研究了最重要的元素和电子性质。该研究有助于开发光学和高温功率器件。

Materials structure–property factorization for identification of synergistic phase interactions in complex solar fuels photoanodes         
Dan Guevarra, Lan Zhou, Matthias H. Richter, Aniketa Shinde, Di Chen, Carla P. Gomes & John M. Gregoire     
npj Computational Materials 8:57 (2022)
doi.org/10.1038/s41524-022-00747-1
Published online: 5 April  2022
Abstract| Full Text | PDF OPEN

Abstract: Properties can be tailored by tuning composition in high-order composition spaces. For spaces with complex phase behavior, modeling the properties as a function of composition and phase distribution remains a formidable challenge. We present materials structure–property factorization (MSPF) as an approach to automate modeling of such data and identify synergistic phase interactions. MSPF is an interpretable machine learning algorithm that couples phase mapping via Deep Reasoning Networks (DRNets) to matrix factorization-based modeling of the representative properties of each phase in a dataset. MSPF is demonstrated for Bi–Cu–V oxide photoanodes for solar fuel generation, which contains 25 different phase combinations and correspondingly exhibits complex composition-structure-photoactivity relationships. Comparing the measured photoactivity to a learned model for non-interacting phases, synergistic phase interactions are identified to guide further photoactivity optimization and understanding. MSPF identifies synergistic interactions of a BiVO4-like phase with both Cu2V2O7-like and CuV2O6-like phases, creating avenues for understanding complex photoelectrocatalysts.

摘要: 材料的性能可以通过调节高阶组分空间中的组分来实现调控。对于具有复杂晶相行为的空间,将材料性能作为组分和相分布的函数来进行模拟仍然是一个巨大的挑战。我们提出了材料结构-性能分解 (MSPF) 作为一种自动建模此类数据并识别晶相协同相互作用的方法。 MSPF 是一种可解释的机器学习算法,它通过深度推理网络 (DRNet) 将晶相映射与数据集中基于矩阵分解模拟每个晶相的典型性能相结合。 MSPF 证明了用于太阳能燃料发电的 Bi-Cu-V 氧化物光阳极,它包含 25 种不同的晶相组合,并相应地表现出复杂的组分-结构-光活性关系。将测量的光活性与非相互作用相的学习模型进行比较,确定了晶相协同相互作用,以指导进一步的光活性优化和理解。MSPF 确定了类 BiVO4 相与类 Cu2V2O7 和类 CuV2O6 相的协同相互作用,为理解复杂的光电催化剂提供了新的途径。

Editorial Summary

Materials structure–property factorization for identification of synergistic phase interactions

Enhancing materials research via integration with artificial intelligence (AI) comprises a recent transformation in the evolution of materials science. Such efforts span the research lifecycle from experiment planning to data analysis, with early demonstrations of AI-assisted data processing naturally occurring in image analysis since many machine-learning (ML) algorithms were initially developed to automate pattern recognition in images. To further automate mapping of structure-dependent properties, the community has made a concerted effort in crystal structure phase mapping. While phase mapping is a route to accelerate generation of phase diagrams in high-order composition spaces, most immediately the results are needed to interpret variations in measured materials properties, i.e., the underlying composition–structure–property relationships. Moreover, phase mapping can seed a variety of further investigations. A team led by Prof. Gregoire from California Institute of Technology, USA, presented materials structure–property factorization (MSPF) as an approach to automate modeling of the properties as a function of composition and phase distribution, enabling identification of synergistic phase interactions. They used comprehensive high throughput experimentation to provide the performance data and demonstrated MSPF for the identification of optimal solar fuels photoanodes via measurement of photoelectrochemical (PEC) performance. In the example dataset, the Bi-Cu-V oxide composition library contains 25 unique combinations of phases with an average of about 13 samples per phase field. They introduced MSPF to model the average performance contribution of each phase as well as the interactions among phases, especially when the composition-structure-property relationships are too complex to be readily interpreted by manual analysis. These results demonstrate that the intuitive approach of adding small amounts of copper vanadates to the best known photoanode phase (BiVO4) is far less effective than the strategy discovered in the present work, wherein the relatively low-performance copper vanadate photoanodes are dramatically improved upon the addition of a small amount of BiVO4. MSPF identifies the phase combinations that merit further investigation to reveal the underlying mechanism of performance enhancement. This identification of emergent properties in complex, multi-phase materials is critical to accelerated exploration and understanding of high-performance materials in high-order composition spaces.

编辑概述

“庖丁解牛”— 结构-性能关系解析新方法

通过与人工智能 (AI) 集成来加强材料研究是材料科学发展的最新转变。这些努力跨越了从实验计划到数据分析的研究生命周期。早期的人工智能辅助数据处理首先出现在图像分析中,因为许多机器学习 (ML) 算法最初开发用来自动化图像模式识别。为了进一步自动化建立结构-性能之间的映射,这一领域的研究人员在晶体结构相映射方面做出了共同努力。虽然晶相映射可以加速高阶组分空间中生成相图的途径,但最直接的需求是解释测量材料特性中的一些变化,即潜在的组分-结构-性能关系。而且,晶相映射的研究可以像种子一样发展壮大,促进其它各种相关研究。自美国加州理工学院的Gregoire教授等人提出了一种材料结构-性能分解 (MSPF) 的方法,可以根据组分和晶相分布自动对性能进行建模,从而能够识别协同的晶相相互作用。他们通过全面的高通量实验来提供晶相识别的性能数据,证明了可以使用MSPF和光电化学 (PEC) 性能来识别最佳太阳能燃料光阳极。在示例数据集中,他们将Bi-Cu-V 氧化物组分库设置为 25 个独特的晶相组合,每个相平均约有 13 个样本,引入 MSPF 来模拟每个晶相的平均性能贡献以及晶相之间的相互作用,特别是当组分-结构-性能关系过于复杂而无法通过手动分析轻松解释时。结果表明,在已知最好的光阳极相 (BiVO4) 中添加少量钒酸铜的直观方法远不如目前工作中发现的策略有效,也就是在相对低性能的钒酸铜光阳极中添加少量BiVO4。MSPF 确定了值得进一步研究的晶相组合,有利于进一步揭示性能增强的潜在机制。这种对复杂多相材料中新性能的识别,以及加速探索、理解高阶成分空间中的高性能材料至关重要。

版权所有 © 中国科学院上海硅酸盐研究所  沪ICP备05005480号-1
地址:上海市长宁区定西路1295号 邮政编码:200050