Using the Effective Diffusivity of Polycrystals to Infer a Complete 5D Structure-Property Model for Hydrogen Diffusivity in Iron Grain Boundaries

晶界是一种缺陷，它强烈地影响--在某些情况下控制--工程材料的性能，包括强度、脆化、腐蚀、太阳能电池效率等。 例如，结构金属中的这些缺陷可以作为腐蚀和脆化的优先位置，这可能导致工程结构的灾难性故障，例如最近旧金山-奥克兰海湾大桥地震安全杆的故障。尽管这些缺陷具有深刻的重要性，但它们的结构是如此复杂，以至于很难预测它们的性质。此外，晶界的类型如此之多，以至于逐个隔离和测试每种晶界是不可行的。该项目提出了一种称为“属性本地化”的新的高效框架，而不是逐个测试它们，以使用少量样本同时推断属性。由于其相关的腐蚀和脆化问题，这个框架将被用来开发一个模型来预测铁中的氢的影响。通过将来自不同领域（医学，计算机科学和材料科学）的理论工具与先进的计算和实验方法以及新兴的显微镜技术相结合，允许对该项目进行高通量探测，将产生新的见解并能够预测属性。这种新的理解将导致设备和工程结构的发展，提高安全性和性能。该项目是在研究生和本科生联合辅导方案的框架内进行的，旨在通过两个主要战略增加工科女生的代表性和保留率：在学术生涯早期参与指导研究，与同行合作，通过社交媒体平台建立同伴指导网络。技术摘要本项目旨在利用三大科学进步的汇合点，以前难以解决的问题：预测作为其五维晶体结构的函数的晶界（GB）的性质。3D表面EBSD显微镜，多晶原子研究和大型数据集反卷积的逆方法为这一转型机会提供了关键。GB结构-性质模型发展的一个重要障碍是5D GB晶体学状态空间的维数，以及相应的大量双晶体实验，这些实验需要充分探索这个空间（~10^5）。拟议的工作将使用层析重建技术来反转均匀化的概念，从而确定从已知的微观结构和有效的性能测量的结构-性能关系。这种去卷积方法被称为属性定位，其主要优点是将所需实验的数量和复杂性降低了几个数量级。新的框架适用于特定的问题，构建一个定量模型的氢扩散率在铁GB作为其五个晶体学参数的函数。这项工作对损坏和退化（如氢脆和腐蚀）有影响，从而产生更耐用的材料，并实现清洁能源经济所需的氢燃料的安全，可靠和有效的储存和输送。将研究与辅导和外联活动相结合，以提高女学生的保留率，将有助于发展一支直接造福社会的多样化科学和工程队伍。

Non-technical AbstractGrain boundaries are defects that strongly influence-and in some cases govern-the properties of engineering materials including strength, embrittlement, corrosion, solar cell efficiency, among others. For example, these defects in structural metals can act as preferential sites for corrosion and embrittlement, which can lead to catastrophic failure of engineering structures like the recent failure of the seismic safety rods in the San Francisco-Oakland Bay Bridge. In spite of their profound importance, the structure of these defects is so complex that it is extremely difficult to predict their properties. Furthermore, there are so many types of grain boundaries that isolating and testing each kind one-by-one is not feasible. Instead of testing them one-by-one, this project presents a new efficient framework called "property localization" to infer the properties simultaneously using a small number of samples. Because of its relevance to the problem of corrosion and embrittlement, this framework will be used to develop a model to predict the effects of hydrogen in iron. By combining theoretical tools from diverse fields (medicine, computer science, and materials science) with advanced computational and experimental methods and emerging microscopy techniques that permit high-throughput probing of this project will yield new insight and enable the prediction of properties. This new understanding will lead to the development of devices and engineering structures with improved safety and performance. This project is being conducted within the framework of a combined graduate and undergraduate mentoring program that is designed to increase the representation and retention of female engineering students through two primary strategies: involvement in mentored research early in their academic career, and engagement with peer-to-peer mentoring networks via a social media platform.Technical AbstractThis project aims to exploit the confluence of three scientific advances to tackle a previously intractable problem: predicting the properties of grain boundaries (GBs) as a function of their five-dimensional crystallographic structure. 3D surface EBSD microscopy, polycrystalline atomistic studies, and inverse methods for deconvolution of large datasets provide the keys for this transformational opportunity. A significant obstacle to the development of GB structure-property models has been the dimensionality of the 5D GB crystallography state space and the correspondingly large number of bicrystal experiments that is required to adequately probe this space (~10^5). The proposed work will use tomographic reconstruction techniques to invert the concept of homogenization, and thereby determine structure-property relationships from known microstructures and effective property measurements. This deconvolution method is referred to as property localization and has the primary advantage of reducing the number and complexity of the required experiments by orders of magnitude. The new framework is applied to the specific problem of constructing a quantitative model for hydrogen diffusivity in iron GBs as a function of their five crystallographic parameters. This work has implications for damage and degradation, such as hydrogen embrittlement and corrosion, leading to more resistant materials and enabling the safe, reliable and efficient storage and delivery of hydrogen fuel necessary for a clean energy economy. The integration of the research with the mentoring and outreach activities to increase retention of women students will support the development of a diverse science and engineering workforce that will directly benefit society.