CAREER: EMERGENT BEHAVIOR OF VAN DER WAALS MAGNETIC MATERIALS: THE GENESIS OF MATERIALS INTELLIGENCE

该项目将量子计算和人工智能应用于理解和发现新的范德华（vdW）材料，这些材料是原子薄片，可以在层与层之间堆叠弱键。石墨就是一个熟悉的例子。我们现在知道如何在可控组合中分离和测量一个、两个或任意数量的这样的薄片。2017年才被发现的磁性vdW层将成为具有许多潜在技术应用的新型量子材料的基石，从数据存储开始。在不久的将来，这些材料可能会让工程师们控制被称为自旋的磁性量子，从而在新兴的二维自旋电子学领域带来更多的应用。它们也可能表现出“拓扑”秩序，这种秩序对杂质和缺陷具有鲁棒性，使它们适合在量子计算机中的应用。vdW（即二维）材料的磁性与体（三维）材料的磁性不同，为物理探索创造了新的途径。可以调整vdW磁体的单层、多层和异质结构（不同类型层的堆叠）的化学成分，以设计具有理想和潜在惊人性能的新材料。由此产生的候选vdW单层、多层及其异质结构的数量非常大，估计远远超过数万亿个vdW材料。因此，使用实验或直接计算模拟来检查所有可能的组合是不可能的。然而，在人工智能（AI）的帮助下，寻找具有理想性能的新材料的挑战可以减轻。该项目将利用人工智能来寻找具有新颖自旋和拓扑特性的新型vdW材料。此外，人工智能将被利用来提供对这些材料中自旋秩序和拓扑秩序的微观起源的物理洞察。这项跨学科研究将训练处于职业生涯不同阶段的科学家将人工智能工具应用于自己的研究。该项目的教育部分将侧重于通过精心设计的课程和实践研究经验，对本科生和研究生进行教学和指导。此外，学生将学习有效的科学传播，并通过公开讲座和网络研讨会分享他们的研究成果。PI还将成立一个公平和包容委员会，并创建一个指导网络，以解决STEM领域的社会不公正现象和代表性不足的少数群体（urm）的需求。这些活动将提供一个支持性的环境，使urm能够实现目标。技术概述：具有本征磁序的二维材料是研究降维奇异自旋自由度的一个平台。范德华材料及其异质结构中的新型自旋现象是凝聚态物理研究的前沿。特别是，在vdW异质结构中，磁性和拓扑顺序之间的相互作用可能会产生新的行为。众所周知，当电子自旋局限于二维空间时，拓扑态就会出现。当两个或多个不同的单层结合成异质结构时，可能会出现令人惊讶的行为，例如增强的拓扑秩序或自旋织构的形成。vdW材料、多层材料和异质结构的总数估计为~10^{21}。这个巨大的搜索空间对于第一原理计算和实验探测来说是压倒性的。该项目将密度函数理论计算与新的人工智能（AI）工具相结合，以促进在这个大型材料空间中的有效导航。目标是发现新的材料，并在电子水平上获得对vdW材料自旋特性和新现象的物理见解。这一见解将为基本理解晶体结构如何影响二维材料（包括单层、多层和异质结构）的磁性铺平道路。研究团队开发的人工智能架构将在凝聚态物理的其他领域和更广泛的科学界得到应用。这项研究是跨学科的，将把跨越不同领域的科学家和主题联系起来。来自物理学的概念可能推动人工智能的进步，反之亦然。此外，通过这项工作发现的二维材料将为实验学家提供一个研究新颖和涌现自旋特性的材料游乐场。此外，拟议的研究将对工业产生潜在影响-例如，数据存储和其他自旋电子学应用，促进未来的存储技术。这项研究还可能有助于在未来的计算机体系结构中取得进展，比如量子计算。该项目将培训处于职业生涯不同阶段的科学家将人工智能工具应用于自己的研究。教育部分侧重于通过精心设计的课程和实践研究经验，教学和指导本科生和研究生。此外，学生将学习有效的科学传播，并通过公开讲座和网络研讨会分享他们的研究成果。PI将成立一个公平和包容委员会，并创建一个导师网络，以解决STEM领域的社会不公正现象和代表性不足的少数群体（urm）的需求。这些活动将提供一个支持性的环境，使urm能够实现目标。该奖项反映了美国国家科学基金会的法定使命，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。

NON-TECHNICAL SUMMARYThis project applies quantum calculations and artificial intelligence to the understanding and discovery of new van der Waals (vdW) materials, atomically thin sheets that can be stacked with weak bonds between layers. Graphite is a familiar example. We now know how to isolate and measure one, two, or any number of such sheets in controllable combinations. Discovered only in 2017, magnetic vdW layers will become the building blocks for novel quantum materials with numerous potential technological applications, starting with data storage. In the near future, these materials may give engineers control over the quantum of magnetism, called spin, leading to additional applications in the nascent field of two-dimensional spintronics. They may also exhibit "topological" order, which is robust against impurities and imperfections, making them suitable for applications in quantum computers.The magnetic properties of vdW (i.e. two-dimensional) materials are different from those of bulk (three-dimensional) materials, creating new avenues for physics exploration. The chemical composition of monolayers, multilayers and heterostructures (stacks of different types of layers) of vdW magnets can be tuned to design novel materials with desirable and potentially surprising properties. The resulting number of candidate vdW monolayers, multilayers and their heterostructures is combinatorially large, with estimates far exceeding trillions of vdW materials. Consequently, inspecting the entire set of possible combinations using experiments or direct computational simulations is impossible. Nevertheless, the challenge of searching for new materials with desirable properties can be mitigated with the help of artificial intelligence (AI). This project will leverage AI to search for new vdW materials with novel spin and topological properties. In addition, AI will be harnessed to provide physical insight into the microscopic origins of spin order and topological order in these materials.This interdisciplinary research will train scientists at different stages of their careers to apply AI tools to their own research. The educational component of the project will focus on teaching and mentoring undergraduate and graduate students through carefully designed coursework and hands-on research experiences. In addition, students will learn effective science communication and share their research via public lectures and webinars. The PI will also form an Equity and Inclusion Committee and create a mentoring network to address social injustice in STEM and the needs of underrepresented minorities (URMs). The activities will provide a supportive environment that will empower URMs to achieve.TECHNICAL SUMMARYTwo-dimensional (2D) materials with intrinsic magnetic order are a platform for studying exotic spin degrees of freedom in reduced dimensions. Novel spin phenomena in van der Waals (vdW) materials and their heterostructures are at the forefront of condensed-matter-physics research. In particular, new behavior may emerge from an interplay between magnetic and topological order in vdW heterostructures. It is well known that topological states emerge when electron spins are confined to two dimensions. When two or more distinct monolayers are combined into a heterostructure, surprising behavior may arise, such as enhanced topological order or the formation of spin textures. An estimate for the total number of vdW materials, multilayers and heterostructures is ~10^{21}. This vast search space is overwhelming for first-principles calculations and experimental probes alone. The project combines density-functional-theory calculations with new artificial-intelligence (AI) tools to facilitate efficient navigation through this large materials space. The goals are to discover novel materials and to gain physical insight into spin properties and emerging phenomena of vdW materials at the electronic level. This insight will pave the way to a fundamental understanding of how crystal structure influences magnetic properties in 2D materials, including monolayers, multilayers and heterostructures. The AI architectures the research team develops will find use in other areas of condensed matter physics and the broader scientific community.The research is interdisciplinary and will connect scientists and topics that span diverse areas. Concepts from physics may enable advances in AI and vice versa. Furthermore, the 2D materials discovered through this work will provide experimentalists with a materials playground in which to study novel and emergent spin properties. In addition, the proposed research will have potential impact in industry - e.g., data storage and other spintronics applications, facilitating future memory technologies. This research may also help create advances in future computer architectures, such as those for quantum computing.This project will train scientists at different stages of their careers to apply AI tools to their own research. The educational component focuses on teaching and mentoring undergraduate and graduate students through carefully designed coursework and hands-on research experiences. In addition, students will learn effective science communication and share their research via public lectures and webinars. The PI will form an Equity and Inclusion Committee and create a mentor network to address social injustice in STEM and the needs of underrepresented minorities (URMs). The activities will provide a supportive environment that will empower URMs to achieve.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.