Accelerated ab initio Molecular Dynamics of III/V Semiconductor Thin-Film Epitaxy

技术总结该奖项支持关于材料模拟中持续挑战的计算和理论研究和教育：在长时间和长度尺度上进行结构演变的模拟，同时保留原子水平的准确细节。PI旨在提高罕见事件模拟的能力，以实现对广泛范围的凝聚态系统的模拟。PI的目标是开发方法来提高当前加速分子动力学的能力。一个基于Bond-Boost方法的加速分子动力学协议将被开发用于经典模拟和从头计算。在该算法中，利用并行计算动态地对Bond-Boost势进行参数化，并将这种基于超动力学的算法与并行复制动力学相结合，以获得高加速和高效的并行缩放。为了将加速的分子动力学模拟外推到更大的长度和时间尺度，它们将被整合到动力学蒙特卡罗模型中。加速分子动力学和动力学蒙特卡罗的结合将使这项技术能够定量地应用于从原子尺度到介观尺度的多尺度问题。自适应Bond-Boost方法将应用于基于第一性原理的GaAs(001)上薄膜生长的模拟--包括同质外延和InAs异质外延。这些系统中的薄膜生长在技术上具有重要意义；在电子、光电子和自旋电子器件中都有应用。从根本上讲，最近的实验工作表明，在同质外延过程中，GaAs(001)衬底可以转变并对扩散和形成的形貌起到积极的作用。该系统打破了衬底是静态模板的传统范式。在这样的动态环境中，吸附原子扩散、岛核和多层生长的方式可能与传统的薄膜生长图景有很大的不同。PI旨在解决这些现象及其在该化合物半导体系统中薄膜生长中的作用。对于在GaAs(001)衬底上InAs异质外延的研究，现有的半经验势将与实验和第一性原理密度泛函理论进行比较，以获得一系列广泛的性质。采用经典和从头计算相结合的分子动力学模拟方法，研究了在GaAs(001)表面InAs润湿层上的沉积和扩散过程。InAs通过STranski-Krastanov生长模式在GaAs衬底上形成自组装量子点，该研究将是第一个解决润湿层中量子点成核的实空间、原子尺度的研究。该项目包括研究生和博士后培训以及本科生的参与。这项研究的各个方面将包括在研究生和本科生的课程中。通过在会议和国际研讨会上组织研讨会，将推动理解多尺度建模的最新水平。非技术总结该奖项支持关于材料模拟中持续挑战的计算和理论研究和教育：使用最新的理论和算法发展，这些发展只能在很短的时间内描述少量原子，大约百万分之一秒，以描述具有多个原子的材料在制造过程中如何在通常的几秒到几小时的时间尺度上演变。PI旨在提高加速分子动力学的能力。在普通的分子动力学中，每个原子的运动都是在计算机上模拟的。计算机逐步处理非常小的时间增量，这些时间增量是由需要包括单个原子相互作用产生的过程所设置的。在这种速度下，模拟速度太慢，相对来说，无法达到描述一种材料在另一种材料表面上生长的薄膜所需的较长时间。加速分子动力学应用了统计力学的思想，使计算机模拟能够访问更长的时间尺度，并包括不经常发生但在确定例如薄膜结构方面发挥重要作用的物理和化学过程。PI将使用开发的新模拟方法来模拟半导体材料砷化镓表面上材料薄膜的生长。这将作为理解在其他材料上生长薄膜的模型，其中许多对于电子、光电子和自旋电子器件的应用非常重要，例如激光、太阳能电池和未来的计算机。技术瓶颈是由于缺乏对分子束外延生长薄膜结构如何演变的了解，分子束外延是指原子从蒸汽沉积到表面。通过模拟这种材料制造技术，PI旨在开发新的洞察力，了解如何更有效地制造适用于应用的材料薄膜。这项研究的各个方面将包括在研究生和本科生的课程中，并将通过在会议和国际研讨会上组织研讨会来提高理解多尺度建模的最新水平。学生将接受模拟材料和材料生长的先进方法的培训。

TECHNICAL SUMMARYThis award supports computational and theoretical research and education on a continuing challenge in materials simulation: Conducting simulations of structural evolution over long time and length scales, while retaining accurate detail at the atomic level. The PI intends to advance capabilities of rare-event simulations to enable simulations of a wide range of condensed-matter systems. The PI aims to develop methods to advance the current capabilities of accelerated molecular dynamics. An accelerated molecular dynamics protocol will be developed for both classical and ab initio simulations based on the Bond-Boost method. In this algorithm, parallel computing is exploited to parameterize the Bond-Boost potential on the fly, as well as to combine this hyperdynamics-based algorithm with parallel replica dynamics for large acceleration with efficient parallel scaling. To extrapolate accelerated molecular dynamics simulations to larger length and time scales, they will be integrated into kinetic Monte Carlo models. The combination of accelerated molecular dynamics with kinetic Monte Carlo will allow the quantitative application of this technique to multi-scale problems, where the length and time scales range from atomic scales to mesoscopic scales.The Adaptive Bond-Boost method will be applied in first-principles-based simulations of thin-film growth on GaAs(001) -- both homoepitaxy and InAs heteroepitaxy will be studied. Thin-film growth in these systems is technologically significant; there are applications in electronic, optoelectronic, and spintronic devices. From a fundamental perspective, recent experimental work has shown that GaAs(001) substrates can transform and play an active role in diffusion and the morphologies that form during homoepitaxy. This system breaks the conventional paradigm that the substrate is a static template. The way that adatom diffusion, island nucleation, and multi-layer growth occur in such a dynamical environment may be considerably different than that envisioned in the conventional picture of thin-film growth. The PI aims to resolve these phenomena and their role in thin-film growth in this compound semiconductor system. For studies of InAs heteroepitaxy on GaAs(001), existing semi-empirical potentials will be tested against experiment and first-principles density-functional theory for an extensive list of properties. A combination of classical and ab initio accelerated molecular dynamics simulations will be used to probe deposition and diffusion on an InAs wetting layer on GaAs(001). InAs forms self-assembled quantum dots via the Stranski-Krastanov growth mode on GaAs substrates and the proposed studies will be the first real-space, atomic scale studies to resolve the nucleation of quantum dots from the wetting layer.This project includes graduate student and postdoctoral training and involvement of undergraduate students. Aspects of the research will be included in graduate and undergraduate coursework. The state of the art in understanding multi-scale modeling will be advanced through organization of symposia at conferences, as well as international workshops.NON-TECHNICAL SUMMARYThis award supports computational and theoretical research and education on a continuing challenge in materials simulation: Using the newest theoretical and algorithmic developments, which can describe a small number of atoms for only a very short time, on the order of one millionth of one millionth of a second, to describe how materials with many atoms evolve over usual time scales of seconds to hours during their fabrication. The PI aims to advance capabilities of accelerated molecular dynamics. In ordinary molecular dynamics, the motion of each atom is simulated on a computer. The computer steps through very small time increments that are set by the need to include processes that occur from the interaction of individual atoms. At this rate, simulations are too slow to reach the, relatively speaking, long times needed to describe, for example the growth of a thin film of one material on the surface of another. Accelerated molecular dynamics applies ideas from statistical mechanics to enable computer simulations to access longer time scales and to include physical and chemical processes that happen infrequently but play an important role in determining, for example, the structure of a thin film. The PI will use new simulation methods that are developed to simulate the growth of thin-films of materials on the surfaces of the semiconductor material gallium arsenide. This will serve as model for understanding thin film growth on other materials, many being important for applications in electronic, optoelectronic, and spintronic devices, such as lasers, solar cells, and computers of the future. Technological bottle-necks arise from a lack of understanding of how the structure of thin films evolves as they are grown by molecular-beam epitaxy, in which atoms are deposited onto the surface from a vapor. By simulating this materials fabrication technique, the PI aims to develop new insight into how films of materials can be fabricated more effectively for applications. Aspects of the research will be included in graduate and undergraduate coursework and the state of the art in understanding multi-scale modeling will be advanced through organization of symposia at conferences, as well as international workshops. Students will be trained in advanced methods for simulating materials and materials growth.