CAREER:Predicting the Surface Structures of Crystalline Materials

该职业奖支持机器学习和数据挖掘方法的开发和应用，以预测各种化学环境中晶体材料的表面结构。 PI将开发一个三步流程，旨在通过最大限度地重复使用现有数据来最大限度地减少预测表面结构的计算费用。 在第一步中，进化算法将被用来开发一个可重复使用的图书馆可能的表面重建散装结构类型。 在第二步中，将开发进化算法和数据挖掘方法的组合，以确定特定材料表面最可能的表面结构。 在第三步中，从头计算和集群扩展将被用来确定具有最低能量的特定表面结构。 结构预测过程将被开发，验证，并应用于三个技术上重要的系统：钙钛矿结构的氧化物，金钯合金，尖晶石结构的氧化物。 这项研究将与一个教育推广计划相结合，该计划旨在加强那些有兴趣和能力通过计算研究发现和设计新材料的研究人员的渠道。 在小学一级，PI自愿与一位主要少数民族，低收入巴尔的摩市公立学校的大师级教师合作，分享科学知识，帮助构建有效的课程，并设计一个旨在教育和激发学生STEM活动的实践练习。 在中学阶段，PI将向参加VEX机器人比赛的巴尔的摩市学生教授计算机编程技能。 在高中一级，附近一所高中的一名女生将作为研究小组成员参加研究项目。 PI将与研究生合作开发一个在线教程，涵盖材料表面科学的基本主题，本教程的内容将被纳入约翰霍普金斯大学材料科学与工程系的核心课程。非技术性总结本职业奖支持先进的计算和数据挖掘方法的开发和应用，以预测原子在表面上的排列方式的材料。 使用计算机预测材料表面特性的能力将有助于为包括电池、催化剂和传感器在内的各种技术设计新材料。 然而，在预测表面的性质之前，首先需要预测表面上的原子结构或原子如何排列。 PI将通过开发一种方法来解决这个具有挑战性的问题，以低计算成本准确预测材料表面结构。 这将通过结合各种计算工具来实现，利用现有的表面结构知识来预测新材料的表面结构。 本研究开发的方法将用于预测三种代表性材料的表面结构，这些材料因其在电池和催化剂等技术中的重要性而被选择。这项研究将与一个教育推广计划相结合，该计划旨在加强那些有兴趣和能力使用计算机发现和设计新材料的研究人员的渠道。 在小学一级，PI自愿与一位主要少数民族，低收入巴尔的摩市公立学校的硕士教师合作，分享科学知识，帮助构建有效的课程，并设计一个旨在教育和激发学生对科学和工程的动手练习。 在中学阶段，PI将向参加机器人比赛的巴尔的摩市学生教授计算机编程技能。 在高中一级，附近一所高中的一名女生将作为研究小组成员参加研究项目。 PI将与研究生合作开发一个涵盖材料表面科学基本主题的在线教程，本教程的内容将纳入约翰霍普金斯大学材料科学与工程系的核心课程。

TECHNICAL SUMMARYThis CAREER award supports the development and application of machine learning and data mining methods to predict the surface structures of crystalline materials in a variety of chemical environments. The PI will develop a three-step process which is designed to minimize the computational expense of predicting surface structures by maximizing the re-use of existing data. In the first step, evolutionary algorithms will be used to develop a re-usable library of likely surface reconstructions for bulk structure types. In the second step a combination of evolutionary algorithms and data mining methods will be developed to determine the most likely surface structures for a particular material surface. In the third step, ab-initio calculations and cluster expansions will be used to identify the particular surface structures with the lowest energy. The structure prediction process will be developed, validated, and applied to three technologically important systems: perovskite-structured oxides, Au-Pd alloys, and spinel-structured oxides. The research will be integrated with an educational outreach program that is designed to strengthen the pipeline of researchers who have both the interest and ability to discover and design new materials through computational research. At the elementary school level, the PI has volunteered to partner with a master teacher at a majority-minority, low-income Baltimore City public school to share scientific knowledge, help construct an effective curriculum, and design a hands-on exercise intended to educate and excite students about STEM activities. At the middle school level, the PI will teach computer programming skills to Baltimore City students who are participating in a VEX robotics competition. At the high school level, a female student from a nearby high school will participate in the research project as member of the research team. The PI will work with the graduate student to develop an online tutorial that covers fundamental topics in materials surface science, and elements of this tutorial will be integrated into the core curriculum of the Department of Materials Science and Engineering at Johns Hopkins University.NONTECHNICAL SUMMARYThis CAREER award supports the development and application of advanced computational and data mining methods to predict how atoms are arranged on the surfaces of materials. The ability to use computers to predict the properties of material surfaces will facilitate the design of new materials for a wide range of technologies including batteries, catalysts, and sensors. However before a property of a surface can be predicted, it is first necessary to predict the atomic structure, or how the atoms are arranged, on the surface. The PI will address this challenging problem by developing a method to accurately predict material surface structures with low computational cost. This will be accomplished by combining a variety of computational tools in a way that leverages existing knowledge about the surface structures to predict the surface structure of a new material. The method developed in this research will be used to predict the surface structures of three representative classes of materials that were chosen for their importance in technologies such as batteries and catalysts. The research will be integrated with an educational outreach program that is designed to strengthen the pipeline of researchers who have both the interest and ability to use computers to discover and design new materials. At the elementary school level, the PI has volunteered to partner with a master teacher at a majority-minority, low-income Baltimore City public school to share scientific knowledge, help construct an effective curriculum, and design a hands-on exercise intended to educate and excite students about science and engineering. At the middle school level, the PI will teach computer programming skills to Baltimore City students who are participating in a robotics competition. At the high school level, a female student from a nearby high school will participate in the research project as member of the research team. The PI will work with the graduate student to develop an online tutorial that covers fundamental topics in materials surface science, and elements of this tutorial will be integrated into the core curriculum of the Department of Materials Science and Engineering at Johns Hopkins University.