Accelerated atomistic simulation of dislocations in nuclear materials

这项研究计划旨在改进和使用计算机模拟来研究加拿大核电反应堆中使用的金属和合金的关键机械性能。这些模拟将集中在几个原子到几百万个原子的尺度上的材料的物理性质。更具体地说，位错动力学将以前所未有的现实主义水平进行建模，无论是通过精确地建模原子间相互作用，还是通过达到与实验时间标度相当的时间标度。 在核动力应用中，位错及其与辐射缺陷的相互作用在决定结构材料性能方面起着至关重要的作用。当材料暴露在中子辐射下时，它们会导致有害的影响，如辐射诱导的硬化和脆化。 这些效应产生于在原子(纳米)水平上发生的过程，并扩大到宏观尺度。为它们建模的典型模拟工作有两个主要问题。首先，描述原子间相互作用的物理模型在位错动力学方面的预测能力有限，特别是在合金中。其次，传统模拟方法几十纳秒的时间尺度太短，无法模拟真实应变率下的位错运动(典型的模拟比实验快10个数量级)。 本研究项目的目的就是解决这两个问题。原子间相互作用的问题将通过开发一种进化算法来解决，其中不仅有可调整的参数，而且函数形式本身也是优化的。与时间尺度相关的问题将通过扩展动力学激活松弛技术来处理位错，动力学激活松弛技术是加速动力学的一种形式。 从技术角度来看，这是一个雄心勃勃的研究计划。它需要最先进的方法，如进化算法、机器学习和加速动力学。这将要求参与该项目的研究生掌握和改进这些方法，这本身就应该引起原子模拟社区的极大兴趣。此外，在原子水平上准确模拟位错动力学的能力不仅对加拿大核电应用，而且对其他金属和合金的研究，特别是在建筑或运输的背景下，将产生深远的影响。

This research program aims to improve and use computer simulations to investigate crucial mechanical properties of metals and alloys used in Canadian nuclear power reactors. These simulations will focus on physical properties of the materials on the scale of a few atoms to a few million atoms. More specifically, the kinetics of dislocations will be modeled, with an unprecedented level of realism, both by modeling inter-atomic interactions accurately and by reaching timescales that are comparable with experimental timescales. Dislocations and their interactions with radiation-induced defects play a critical role in determining the properties of structural materials in nuclear power applications. They lead to deleterious effects such as radiation-induced hardening and embrittlement, as the materials are exposed to neutron radiation. These effects arise from processes that take place at the atomic (nano-) level and scale up to the macro-scale. Typical simulations efforts to model them have two major issues. First, the physical models to express interatomic interactions have limited predictive ability in the context of dislocation kinetics, especially in alloys. Second, the timescales accessible to conventional simulation methodsa few tens of nanosecondsare too short to simulate dislocation motion at realistic strain rates (typical simulations are 10 orders of magnitude faster than experiments). The objective of this research project is to address these two issues. The interatomic interactions issue will be addressed by developing an evolutionary algorithm, where not only are there adjustable parameters, but where the functional form is itself optimized. The issue related to timescales will be addressed by extending the kinetic Activation Relaxation Technique, a form of accelerated dynamics, to handle dislocations. This is a ambitious research plan, from a technical point of view. It requires state-of-the-art methods, such as evolutionary algorithms, machine learning and accelerated dynamics. This will require the graduate students involved in the project to master and improve these methods, which should in itself be of great interest for the atomistic simulations community. Furthermore, the ability to accurately model dislocation dynamics at the atomic level will have a profound impact not only for Canadian nuclear power applications, but also for the study of other metals and alloys, notably in the context of construction or transportation.