CAREER: Advancing Atomic-Level Understanding of Kinetically Driven Solid-Solid Phase Transitions from First Principles and Machine Learning

职业：从第一原理和机器学习推进对动力学驱动的固-固相变的原子级理解

• 批准号: 2238516
• 负责人: Liping Yu
• 金额: $ 52.83万
• 依托单位: University of Maine
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-03-01 至 2023-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2238516&HistoricalAwards=false
• 关键词: CAREER; Advancing; Atomic; Level; Understanding

NONTECHNICAL SUMMARYThis CAREER award supports theoretical and computational research to advance the fundamental understanding of solid-solid phase transitions. Most materials have several different stable crystalline structures, each with a characteristic set of physical, chemical, and mechanical properties. Carbon, which can form graphite (flaky, black material used in pencils) or diamond (hard, colorless gemstone) structures, is a well-known example. Solid-solid transitions that occur between different crystalline forms of the same compound are ubiquitous and important phenomena. They can lead to a wide variety of technologically important applications such as diamond and steel production, synthesis of ceramic materials, thermal energy harvesting and storage, rewritable optical data storage, and nonvolatile electronic memories. Historically, considerable progress has been made in understanding solid-solid transitions from thermodynamics concerning the relative phase stability (the “driving force” for the phase transition), regardless of transition paths between the initial and final structures. However, the kinetics that dictates whether or not the transition can occur in practice under given environmental conditions and which path the transition likely takes place remain poorly understood. This project will advance the atomic-level understanding of kinetics underlying solid-solid transitions without using empirical data and develop an advanced artificial intelligence method for the fast and accurate prediction of kinetic barriers that control solid-solid transition in various environments. The data and methods acquired will be broadly disseminated to the scientific community and the general public through open-source distributions and publications.Education and outreach activities are integrated in this project with the goal to inspire and develop a diverse, globally competitive next generation STEM workforce in computational materials science that will benefit the State of Maine as well as the nation. The research team will (i) develop a “kinetics-driven phase-change materials by design” module for high school students in collaboration with the Maine Center for Research in STEM Education, (ii) develop an advanced courses in “computational materials physics and modeling” for seniors and graduate students in science and engineering departments at the University of Maine, (iii) expand the partnership between the University of Maine and Oak Ridge National Laboratory to provide students the opportunity to take advantage of facilities and computational resources in the national laboratory to expand their experiences beyond the traditional university setting, and (iv) create a summer research fellowship program to provide opportunities for talented undergraduates majoring in science, engineering, and mathematics to conduct computational materials research.TECHNICAL SUMMARYThis CAREER award supports theoretical and computational research to advance atomic level understanding of solid-solid phase transitions. Solid-solid phase transitions are ubiquitous phenomena that play key roles in diverse technologies across physics, chemistry, biology, materials science and engineering. Despite having been studied for over a century, the fundamental understanding of phase transition kinetics remains largely qualitative or phenomenological; the atomistic mechanism of such transition processes and design rules for controlling kinetics are still crucially missing. This project will advance atomic-level understanding of kinetics of solid-solid phase transitions using a combined method of modern first-principles electronic structure theory calculations, quantitative chemical bond analysis, and machine learning. The specific objectives are to (i) identify physical principles and structural motifs that control kinetic barriers of polymorphic transitions from first principles, and (ii) develop a bottom-up physics-driven machine learning method for the fast and accurate prediction of transition barriers. The study will be carried on a set of select well-known phase-transition materials that are technologically important for energy and electronic applications. The research will accelerate the design and discovery of new functional phase-change materials where kinetics is essential.Education and outreach activities are integrated in this project with the goal to inspire and develop a diverse, globally competitive next-generation STEM workforce in computational materials science that will benefit the State of Maine as well as the nation. The research team will (i) develop a “kinetics-driven phase-change materials by design” module for high school students in collaboration with the Maine Center for Research in STEM Education, (ii) develop an advanced courses in “computational materials physics and modeling” for seniors and graduate students in science and engineering departments at the University of Maine, (iii) expand the partnership between the University of Maine and Oak Ridge National Laboratory to provide students the opportunity to take advantage of facilities and computational resources in the national laboratory to expand their experiences beyond the traditional university setting, and (iv) create a summer research fellowship program to provide opportunities for talented undergraduates majoring in science, engineering, and mathematics to conduct computational materials research.This project is jointly funded by the Division of Materials Research through the Condensed Matter and Materials Theory program, and the Established Program to Stimulate Competitive Research (EPSCoR).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.

该职业奖支持理论和计算研究，以促进对固-固相变的基本理解。大多数材料都有几种不同的稳定晶体结构，每种结构都有一套独特的物理、化学和机械性能。碳就是一个众所周知的例子，它可以形成石墨（铅笔用的片状黑色材料）或钻石（坚硬、无色的宝石）结构。在同一化合物的不同晶型之间发生的固-固转变是普遍存在的重要现象。它们可以导致各种各样的技术上重要的应用，如金刚石和钢铁生产，陶瓷材料的合成，热能收集和存储，可重写的光学数据存储和非易失性电子存储器。从历史上看，无论初始结构和最终结构之间的过渡路径如何，从热力学的相对相稳定性（相变的“驱动力”）来理解固-固相变已经取得了相当大的进展。然而，在给定的环境条件下，决定这种转变是否能在实践中发生以及这种转变可能发生的途径的动力学仍然知之甚少。该项目将在不使用经验数据的情况下推进对固体-固体转变背后动力学的原子水平理解，并开发一种先进的人工智能方法，用于快速准确地预测各种环境中控制固体-固体转变的动力学障碍。所获得的数据和方法将通过开源分发和出版物广泛传播给科学界和一般公众。教育和推广活动被整合到这个项目中，目的是激发和发展一个多样化的，具有全球竞争力的下一代计算材料科学STEM劳动力，这将有利于缅因州和国家。该研究团队将(i)与缅因州STEM教育研究中心合作，为高中生开发“动力学驱动相变材料设计”模块，（ii）为缅因大学理工科的大四学生和研究生开发“计算材料物理和建模”高级课程。（iii）扩大缅因大学和橡树岭国家实验室之间的合作关系，为学生提供利用国家实验室的设施和计算资源的机会，以扩展他们在传统大学环境之外的经验；（iv）创建一个夏季研究奖学金计划，为科学、工程和数学专业的有才华的本科生提供进行计算材料研究的机会。该职业奖支持理论和计算研究，以推进对固体-固体相变的原子水平理解。固-固相变是一种普遍存在的现象，在物理、化学、生物、材料科学和工程等多种技术中发挥着关键作用。尽管已经研究了一个多世纪，但对相变动力学的基本理解仍然主要是定性的或现象学的；这种转变过程的原子机制和控制动力学的设计规则仍然非常缺乏。该项目将利用现代第一性原理电子结构理论计算、定量化学键分析和机器学习相结合的方法，推进对固-固相变动力学的原子水平理解。具体目标是(i)确定从第一原理控制多态过渡的动力学障碍的物理原理和结构基序，以及（ii）开发一种自下而上的物理驱动的机器学习方法，用于快速准确地预测过渡障碍。这项研究将对一组众所周知的相变材料进行选择，这些材料在技术上对能源和电子应用具有重要意义。这项研究将加速设计和发现新的功能相变材料，其中动力学是必不可少的。教育和推广活动被整合到这个项目中，目的是激发和发展一个多样化的，具有全球竞争力的计算材料科学下一代STEM劳动力，这将有利于缅因州和国家。该研究团队将(i)与缅因州STEM教育研究中心合作，为高中生开发“动力学驱动相变材料设计”模块，（ii）为缅因大学理工科的大四学生和研究生开发“计算材料物理和建模”高级课程。（iii）扩大缅因大学和橡树岭国家实验室之间的合作关系，为学生提供利用国家实验室的设施和计算资源的机会，以扩展他们在传统大学环境之外的经验；（iv）创建一个夏季研究奖学金计划，为科学、工程和数学专业的有才华的本科生提供进行计算材料研究的机会。该项目由材料研究部通过凝聚态物质和材料理论项目以及促进竞争研究的既定项目（EPSCoR）共同资助。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。