CAREER: Understanding the Role of Spin-Dynamics in the Formation of Magnetic Microstructure

该奖项的部分资金来自2021年美国救援计划法案(公法117-2)。非技术总结这个职业奖项支持基础研究和教育，旨在通过计算机模拟在原子长度尺度上模拟材料的性质来了解材料的磁性行为。磁铁是实现从电机、发电机到耳机等现代技术的关键材料。然而，即使是最常见的磁铁，也很难使用现有的计算建模技术在原子尺度上理解它的行为。在这个项目中，Pi和他的学生将开发新的计算方法，使用应用于材料建模的机器学习工具，准确描述原子自旋及其在材料中的排列的耦合演化。这些方法将用于理解元素铁和钴中的相变及其与磁性的关系。从这些研究中获得的理解和新知识可以通过帮助选择材料和确定工艺来指导新磁体的开发，从而使最有效的磁体成为可能。该奖项还支持国际和平协会在高中、本科生和研究生层面的教育活动，以及社区外联活动，特别是在代表性不足的群体中。除了在计算材料建模领域对研究生和本科生进行培训外，PI还将开发一门机器学习理论和方法课程。虽然大多数人都知道机器学习和磁性很重要，但对于它们是什么以及它们是如何工作的，人们存在一些误解。国际和平协会将与当地中学和成人教育团体合作，提供机会更多地了解这些主题，并帮助消除围绕这些主题的一些谜团。技术总结这个职业奖项支持理论和计算研究，旨在了解磁矩和晶体配位在确定材料的磁性和微结构属性方面的相互作用。尽管尝试了许多次，但经典的分子动力学(MD)模拟未能充分地捕捉到磁性材料的行为。即使是众所周知的铁从铁素体到奥氏体的马氏体相变也超出了最先进的MD方法的范围，因为它们不能准确地将磁矩描述为环境相关的动态变量。在这个项目中，PI和他的学生将开发出在MD尺度上精确描述耦合自旋和晶格动力学的方法，并将其应用于铁和钴，从而在理解、建模和开发磁性材料方面取得重大进展。这将通过使用基于机器学习的人工神经网络的原子间势来实现，该势明确地将单个原子的自旋视为一个自由度。这些势能将使用密度泛函理论结果的广泛数据库进行训练，并用于模拟铁和钴的相界，监测局部磁矩在这些边界的能量学和稳定性中的作用，以及这些边界在材料内发生相变时的演变。该奖项还支持PI在高中、本科生和研究生水平的教育活动，以及社区推广活动，特别是在代表性不足的群体中。除了在计算材料建模领域对研究生和本科生进行培训外，PI还将开发一门机器学习理论和方法课程。虽然大多数人都知道机器学习和磁性很重要，但对于它们是什么以及它们是如何工作的，人们存在一些误解。PI将与当地中学和成人教育团体合作，提供机会更多地了解这些话题，并帮助解开围绕这些话题的一些谜团。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This award is funded in part under the American Rescue Plan Act of 2021 (Public Law 117-2).NONTECHNICAL SUMMARYThis CAREER award supports basic research and education with an aim to understand the magnetic behavior of materials using computer simulations that model their properties at atomic length scales. Magnets are crucial materials that enable modern technologies from electric motors and generators to headphones. However, even the behavior of the most common magnet, iron, is difficult to understand at the atomic scale using existing computational modeling techniques. In this project, the PI and his students will develop new computational methods to accurately describe the coupled evolution of atomic spins and their arrangement within the material using tools of machine learning as applied to materials modeling. The methods will be applied to understanding phase transformations and their connection to magnetic properties in elemental iron and cobalt. The understanding and new knowledge gained from these studies can then be used to guide the development of new magnets by helping to select materials and determine the processes to make the most efficient magnets possible. This award also supports the PI's educational activities at the high school, undergraduate, and graduate levels, as well as community outreach, particularly among under-represented groups. In addition to training graduate and undergraduate students in the area of computational materials modeling, the PI will develop a course in machine learning theory and methods. While most people know that machine learning and magnetism are important, there are a number of misconceptions about what they are and how they work. The PI will work with local secondary schools and adult education groups to provide opportunities to learn more about these topics and help dispel some of the mystery around them. TECHNICAL SUMMARYThis CAREER award supports theoretical and computational research with an aim to understand the interplay between magnetic moment and crystal coordination in determining the magnetic and microstructural properties of materials. Despite numerous attempts, classical molecular dynamics (MD) simulations have failed to adequately capture the behavior of magnetic materials. Even the well-known martensitic transformation of iron from ferrite to austinite is beyond state-of-the-art MD methods, as they are not able to accurately describe the magnetic moment as an environmentally dependent dynamic variable. In this project, the PI and his students will develop the methodology to accurately describe coupled spin and lattice dynamics at the MD scale and apply it to iron and cobalt, enabling a significant advance in the understanding, modeling, and development of magnetic materials. This will be accomplished by using machine learned artificial neural network based interatomic potentials which explicitly consider the spin of individual atoms as a degree of freedom. These potentials will be trained using extensive databases of density functional theory results and used to model phase boundaries in both iron and cobalt, monitoring the role of local magnetic moments in the energetics and stability of such boundaries, and the evolution of these boundaries as phase transformations occur within the material.This award also supports the PI's educational activities at the high school, undergraduate, and graduate levels, as well as community outreach, particularly among under-represented groups. In addition to training graduate and undergraduate students in the area of computational materials modeling, the PI will develop a course in machine learning theory and methods. While most people know that machine learning and magnetism are important, there are a number of misconceptions about what they are and how they work. The PI will work with local secondary schools and adult education groups to provide opportunities to learn more about these topics and help dispel some of the mystery around them.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.