CAREER: Extending the lattice stability framework in ab initio alloy thermodynamics

技术摘要材料研究部和网络基础设施办公室为该职业奖提供资金。该奖项支持计算和理论研究和教育，旨在提高我们基于第一性原理量子力学计算预测固态合金相图的能力，而不依赖于实验输入。该项目提供了一个坚实的概念框架和相关的算法来模拟技术相关合金中常见的现象，但合金理论领域目前还无法处理这些现象。这项研究有两个重点：1)晶格不稳定性。完善的CALPHAD和簇展开形式基本上是基于一组定义良好的晶格（如bcc， fcc, hcp）在所有组成中至少保持亚稳的假设。然而，在许多合金体系中（例如Ti-Al， Cu-Fe等），一些晶格甚至不是机械稳定的，因此阻碍了任何计算适当自由能的尝试。该项目通过适当地限制传统上用于基于严格的几何划分技术计算自由能的积分域来处理这一基本概念问题，但不需要完全的非调和处理。2)新型化合物预测。虽然存在强大的方法（例如簇展开）来模拟共同底层晶格上的合金有序现象，但该项目设计了适用于违反该假设的补充方法（特别是在Lave相，sigma相等中）。其思想是将合金系统的能量分解为原子团簇贡献的总和。然后，提出的算法尝试使用组合技术重新组装最低能量簇，以寻找低能晶体结构。所提出的方法将在PI的合金理论自动化工具包（ATAT）中实施，这是一个已经建立用户基础的软件包。这项研究活动的教育组成部分包括通过一个配套的教育网站来增强ATAT，该网站将提供关于热力学和相图的教程和社交网络工具，以及在本科和研究生课程中使用ATAT，让学生亲身体验未来的材料设计工具。为了促进多样性，该项目将在加州理工学院MRSEC的努力基础上，从加州州立大学洛杉矶分校招募少数族裔本科生。非技术摘要材料研究部和网络基础设施办公室为该职业奖提供资金。该奖项支持旨在提高我们预测“相图”能力的计算和理论研究。相图通常被称为“材料科学的路线图”，因为它们表明当不同的化学元素被合金化时，哪些化合物会随着温度和压力的变化而形成，因此为材料设计提供了至关重要的指导。该项目侧重于基于量子力学计算的预测方法，而不需要实验输入，这是发现真正新材料所必需的能力。该项目提供了一个坚实的概念框架和相关算法来模拟技术相关合金中常见的现象，但合金理论领域目前还没有足够的能力来处理这些现象。所提出的方法将在PI的合金理论自动化工具包（ATAT）中实施，这是一个已经建立用户基础的软件包。这项研究活动的教育组成部分包括通过一个配套的教育网站来增强ATAT，该网站将提供关于热力学和相图的教程和社交网络工具，以及在本科和研究生课程中使用ATAT，让学生亲身体验未来的材料设计工具。为了促进多样性，该项目将在加州理工学院材料研究科学与工程中心的努力基础上，从加州州立大学洛杉矶分校招募少数族裔本科生。

TECHNICAL SUMMARYThe Division of Materials Research and the Office of Cyberinfrastructure contribute funding to this CAREER award. This award supports computational and theoretical research and education aimed at improving our ability to predict phase diagrams of solid state alloys, based upon first-principles quantum mechanical calculations, without relying on experimental input. This project provides a firm conceptual framework and associated algorithms to model phenomena that commonly occur in technologically-relevant alloys, but that the field of alloy theory is currently poorly equipped to handle. The research has two thrusts:1) Lattice instabilities. The well-established CALPHAD and cluster expansion formalisms are fundamentally based on the assumption that a set of well-defined lattices (e.g. bcc, fcc, hcp) remain at least metastable at all compositions. However, in numerous alloy systems (e.g. Ti-Al, Cu-Fe, etc.), some lattices are not even mechanically stable, thus impeding any attempts at calculating a proper free energy. This project handles this fundamental conceptual issue by suitably constraining the domain of integration traditionally employed to calculate free energies based on rigorous geometrical partitioning techniques, but without necessitating a full anharmonic treatment.2) Novel compound prediction. While powerful methods (e.g. the cluster expansion) exist to model alloy ordering phenomena on a common underlying lattice, this project devises complementary methods that are applicable when this assumption is violated (notably, in Lave phases, sigma phases, etc.). The idea is to decompose the energy of an alloy system as a sum of atom cluster contributions. The proposed algorithm then attempts to re-assemble the lowest energy clusters, using combinatorial techniques, in search of low-energy crystal structures.The proposed methods will be implemented in the PI's Alloy Theoretic Automated Toolkit (ATAT), a software package that already has an established user base. The educational components of this research activity include augmenting ATAT by a companion educational web site, which will provide tutorials and social networking tools focusing on thermodynamics and phase diagrams, as well as using ATAT in undergraduate and graduate classes to give students hands-on experience with materials design tools of the future. To promote diversity, this project will build upon the efforts of the MRSEC at Caltech in recruiting minority undergraduates from California State University, Los Angeles.NON-TECHNICAL SUMMARYThe Division of Materials Research and the Office of Cyberinfrastructure contribute funding to this CAREER award. This award supports computational and theoretical research aimed at improving our ability to predict "phase diagrams". Phase diagrams are often called the "road maps of materials science", as they indicate which compounds form as a function of temperature and pressure, when different chemical elements are alloyed, thus providing crucial guidance in material design. This project focuses on prediction methods based on quantum mechanical calculations, without necessitating experimental input, a capability that is essential to enable the discovery of truly novel materials. The project provides a firm conceptual framework and associated algorithms to model phenomena that commonly occur in technologically-relevant alloys, but that the field of alloy theory is currently poorly equipped to handle. The proposed methods will be implemented in the PI's Alloy Theoretic Automated Toolkit (ATAT), a software package that already has an established user base. The educational components of this research activity include augmenting ATAT by a companion educational web site, which will provide tutorials and social networking tools focusing on thermodynamics and phase diagrams, as well as using ATAT in undergraduate and graduate classes to give students hands-on experience with materials design tools of the future. To promote diversity, this project will build upon the efforts of the Materials Research Science and Engineering Center at Caltech in recruiting minority undergraduates from California State University, Los Angeles.