Collaborative Research: DMREF: Symmetry-Guided Machine Learning for the Discovery of Topological Phononic Materials

合作研究：DMREF：用于发现拓扑声子材料的对称引导机器学习

• 批准号: 2118448
• 负责人: Mingda Li
• 金额: $ 56万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-10-01 至 2025-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2118448&HistoricalAwards=false
• 关键词: Collaborative; Research; DMREF; Symmetry; Guided

Non-technical Description: Fundamental understanding and control of heat conduction processes in materials are important for energy infrastructure, electronic devices, and renewable energy generation systems. This project focuses on a novel property of phonons – vibrations of atoms that carry the heat in materials - called "topology". This property may allow new phenomena, such as heat conduction perpendicular to the temperature gradient direction and more efficient transport of heat waves on the material surfaces. To discover topological phonons, the research team will exploit a Materials Genome approach to search for materials hosting these special heat carriers. Once candidates are identified, the research team will synthesize and characterize them, and the results will be used to refine the search algorithm. The research team plans to establish a public database storing the heat conduction properties of a large number of materials. This research will not only advance the fundamental understanding of how topology affects heat conduction in real materials, but also provide new routes to realizing unusual functionalities such as heat conductors that can be switched on and off. This project also supports educational activities to teach basic materials physics concepts to K-12 and undergraduate students through hands-on class projects and short courses. To promote diversity in the materials science workforce, the team also provides research opportunities to high school and undergraduate students from underrepresented minority communities. Technical Description: While the topology of electronic states has been a central theme in condensed matter physics for the past decade, topological phononic states have received much less attention. Unlike their fermionic counterparts, topological states in the entire phonon spectrum can contribute to observable material properties, making topological phononic materials ideal testbeds for emerging new physics in topological bosonic systems, including phonon thermal Hall effects, novel topological phonon-electron interactions and the resulting phenomena, such as unusual superconducting states. This project aims to systematically identify materials hosting intrinsic topological phonons in the thermal regime, where the topological phononic states explicitly modify intrinsic material properties, including thermal transport, electron-phonon interactions, and surface phonon modes. The research team will seek to accelerate material discovery by incorporating symmetry-guided machine learning based on Euclidean neural networks. Machine learning predictions will be verified using first-principles phonon simulation and topological invariance analysis. Promising candidate materials will be synthesized as thin films and bulk single crystals and characterized using inelastic neutron and x-ray scattering, thermal transport, and surface-sensitive spectroscopy and scanning probe measurements. This research will advance fundamental understanding of topological bosonic systems and examine new thermal functionalities enabled by topological phonons.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

非技术描述：对材料热传导过程的基本理解和控制对于能源基础设施、电子设备和可再生能源发电系统非常重要。该项目的重点是声子的一种新特性-在材料中携带热量的原子振动-称为“拓扑”。这种性质可以允许新的现象，例如垂直于温度梯度方向的热传导和在材料表面上更有效地传输热波。为了发现拓扑声子，研究小组将利用材料基因组方法来寻找承载这些特殊热载体的材料。一旦确定了候选人，研究小组将对其进行综合和定性，结果将用于改进搜索算法。研究小组计划建立一个公共数据库，存储大量材料的导热性能。这项研究不仅将推进对拓扑结构如何影响真实的材料中热传导的基本理解，而且还为实现不寻常的功能提供了新的途径，例如可以打开和关闭的热导体。该项目还支持教育活动，通过动手课堂项目和短期课程向K-12和本科生教授基本材料物理概念。为了促进材料科学劳动力的多样性，该团队还为来自代表性不足的少数民族社区的高中和本科生提供研究机会。 技术说明：在过去的十年里，电子态的拓扑结构一直是凝聚态物理学的一个中心主题，而拓扑声子态却很少受到关注。与费米子对应物不同，整个声子谱中的拓扑态可以对可观察的材料性质做出贡献，使拓扑声子材料成为拓扑玻色子系统中新兴物理学的理想试验平台，包括声子热霍尔效应，新颖的拓扑声子-电子相互作用以及由此产生的现象，如不寻常的超导态。该项目旨在系统地识别在热状态下承载固有拓扑声子的材料，其中拓扑声子状态明确地修改固有材料特性，包括热输运，电子-声子相互作用和表面声子模式。研究团队将寻求通过结合基于欧几里得神经网络的机器学习来加速材料发现。机器学习预测将使用第一性原理声子模拟和拓扑不变性分析进行验证。有前途的候选材料将被合成为薄膜和大块单晶，并使用非弹性中子和x射线散射，热传输，表面敏感光谱和扫描探针测量的特点。这项研究将推进对拓扑玻色子系统的基本理解，并研究拓扑声子实现的新的热功能。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Panoramic Mapping of Phonon Transport from Ultrafast Electron Diffraction and Scientific Machine Learning

超快电子衍射和科学机器学习的声子传输全景图

• DOI: 10.1002/adma.202206997
• 发表时间: 2022
• 期刊: Advanced Materials
• 影响因子: 29.4
• 作者: Chen, Zhantao;Shen, Xiaozhe;Andrejevic, Nina;Liu, Tongtong;Luo, Duan;Nguyen, Thanh;Drucker, Nathan C.;Kozina, Michael E.;Song, Qichen;Hua, Chengyun
• 通讯作者: Hua, Chengyun