CAREER: First Principles-Enabled Prediction of Thermal Conductivity and Radiative Properties of Solids

项目负责人：阮秀林，普渡大学。项目编号：cbet -1150948。该项目将能够从第一性原理预测固体的导热和辐射特性。固体的热导率和远红外热辐射特性是许多现代和新兴应用中的关键问题，如热管理、电子、光伏和热电。这两种性质虽然看似无关，但在原子尺度上是由称为声子的同一热能载体的色散关系和弛豫时间决定的。为了指导这些材料的设计和合成，非常希望从第一性原理预测它们的热性能，即从它们的原子结构中预测它们的热性能，而不使用可调参数。然而，现有的经典原子间势甚至对于诸如硅和碳等标准材料也是不准确的，因为这些势不是用于热传递建模的目的。对于大多数其他固体，经典势还没有发展，使其热输运性质的预测是不可能的。因此，本提案的目标是制定新的方法，可以开发准确的原子间势或可以完全绕过经典势的使用，用于热性质预测。为了实现这一目标，将并行开发两种多尺度多物理场方法。在第一种方法中，第一性原理计算将用于开发精确的经典原子间势，这些势将被有意地优化用于热传递建模，然后这些势将被用于经典MD来预测热导率。为了绕过具有挑战性和繁琐的潜在开发过程，第二种方法将引入一种新的紧密结合分子动力学（TBMD）方法来产生原子的轨迹，然后将其用于声子光谱分析，以获得光谱声子弛化时间以及热导率和辐射特性。这种预测能力将首先在硅等标准材料上进行演示，然后在一系列重要但复杂的热电和光伏材料上进行演示，包括Bi2Te3和GaAs体和纳米材料。该建议的智力价值集中在基于热传导和辐射性质的第一性原理的全新预测方法上。紧密结合的MD与声子光谱分析将彻底改变许多具有重要技术意义的材料的热输运性质预测，而在此之前，由于缺乏经验的原子间相互作用势，原子尺度的预测是不可能的。这些方法还将首次用于实际重要的热电和光伏纳米材料，包括Bi2Te3和GaAs，以指导实验合成。这项研究工作将影响热科学和教育/推广项目。新的预测方法由于其通用性将引起广泛的兴趣。包括热管理、热电、电子和光伏在内的重要应用将受益于使用这些方法产生的新见解。代表性不足的本科生将继续参与研究。一个关键的教育/外联组成部分将是向nanoHUB和thermalHUB传播该项目产生的研究和教育规范，供公众使用。将提供全面的文档、在线讲座和教程来解释这些代码。考虑到这些材料提供的新功能，它们将在PI领域引起广泛的兴趣。

PI: Xiulin Ruan, Purdue UniversityProposal Number: CBET-1150948The proposed effort will enable the prediction of thermal conductive and radiative properties of solids from first principles. Thermal conductivity and far-infrared thermal radiative properties of solids are critical issues in many modern and emerging applications such as thermal management, electronics, photovoltaics, and thermoelectrics. Both properties, although seemingly unrelated, are governed in the atomic scale by the dispersion relation and relaxation time of the same thermal energy carrier called phonon. To guide the design and synthesis of these materials, it is highly desirable to predict their thermal properties from first principles, i.e., from their atomic structures without the use of adjustable parameters. However, existing classical interatomic potentials are inaccurate even for standard materials such as silicon and carbon since the potentials were not intended for the purpose of thermal transport modeling. For most other solids, the classical potentials haven't been developed yet, making the prediction of their thermal transport properties impossible. Therefore, it is the objective of this proposal to formulate new methodologies that can develop accurate interatomic potentials or can completely bypass the use of classical potentials, for thermal property prediction. Toward this goal, two multiscale multiphysics methods will be developed in parallel. In the first method, first principles calculations will be used to develop accurate classical interatomic potentials that are intentionally optimized for thermal transport modeling, and the potentials will then be employed in classical MD to predict thermal conductivity. In order to bypass the challenging and tedious potential development process, the second method will introduce a new tight-binding molecular dynamics (TBMD) method to produce the trajectory of atoms, which will then be used in phonon spectral analysis to obtain spectral phonon relaxation time as well as thermal conductivity and radiative properties. The predictive power will be demonstrated first on standard materials such as silicon, and then on a range of important but complex thermoelectric and photovoltaic materials, including Bi2Te3 and GaAs bulk and nanomaterials.The intellectual merit of the proposal centers around fundamentally new prediction methods based on first principles for both thermal conductive and radiative properties. The tight binding MD together with phonon spectral analysis will revolutionize thermal transport property prediction of a wide range of materials of technological importance, on which atomic scale prediction was not possible before due to the lack of empirical interatomic potentials. The methods will also be used on practically important thermoelectric and photovoltaic nanomaterials, including Bi2Te3 and GaAs, for the first time to guide experimental synthesis.The research effort will impact thermal science and education/outreach programs. The new prediction methods will be of broad interest due to their generality. Important applications, including thermal management, thermoelectrics, electronics, and photovoltaics will benefit from the new insights generated using these methods. Under-represented and undergraduate students will continue to be involved in research. A key education/outreach component would be the dissemination of the research and education codes resulted from this project to nanoHUB and thermalHUB for general public use. Comprehensive documentation, online lectures, and tutorials explaining the codes will be provided. These materials will be of wide interest in the PI's field given the new capabilities they provide.