CDS&E/Collaborative Research: Genetic Algorithm Driven Hybrid Computational/Experimental Engineering of Defects in Designer Materials

CDS

• 批准号: 1404967
• 负责人: Alexander Balandin
• 金额: $ 16.8万
• 依托单位: University of California-Riverside
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-07-15 至 2018-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1404967&HistoricalAwards=false
• 关键词: CDS& Collaborative; Research; Genetic; Algorithm

The goal of this collaborative research is to build a computational framework with experimental validation to systematically engineer designer materials that provide targeted properties. Defects in materials significantly influence their properties, for instance in energy transfer. The current state-of-the-art techniques allow one to induce defects in materials controllably and subsequently predict and correlate relevant properties to the defect concentration. In this research, computational and experimental approaches will be integrated to construct designer defect-engineered materials that will provide desired properties. In this approach, first a targeted property will be ascertained, and then the defect concentration and distribution will be predicted to synthesize novel material structures that provide the predetermined property. The hybrid computational framework will motivate the general scientific community to leverage advanced computing infrastructures. The efforts will establish new research and learning communities for computational-experimental data-enabled science and engineering through promotion of diversity and undergraduate research experience, outreach to community college students and work-force development, outreach to the general public (through education about the power of simulation-based engineering with an online gaming tool and hands-on activities at science centers) and integration of research with education. This project is a comprehensive effort to rigorously integrate developments in high throughput synthesis and characterization with improving access and availability to high performance computing resources towards solving an inverse design of materials problem. A computational framework that sweeps through millions of possible structural permutations and combinations for property prediction will be developed and validated against experiments. The framework will (1) employ massively parallel molecular dynamics simulations to predict transport properties of nanomaterials, (2) synthesize defect engineered nanostructures and measure the corresponding transport properties, (3) integrate the above nanoscale computations and experiments through a genetic algorithm based hybrid optimization scheme to predict the optimal material structure for specified transport properties and formulate a hierarchy of increasingly complex duals of cost-functionals and design parameters, and finally (4) close the loop by validating the inverse design results experimentally.

这项合作研究的目标是建立一个具有实验验证的计算框架，以系统地设计提供目标特性的设计材料。材料中的缺陷会显著影响其性能，例如能量传递。目前最先进的技术允许人们可控地诱导材料中的缺陷，并随后预测和关联相关属性的缺陷浓度。在这项研究中，计算和实验的方法将被集成到构建设计师的缺陷工程材料，将提供所需的性能。在这种方法中，首先将确定目标性质，然后将预测缺陷浓度和分布以合成提供预定性质的新型材料结构。混合计算框架将激励一般科学界利用先进的计算基础设施。这些努力将通过促进多样性和本科生研究经验、与社区大学生的联系和劳动力发展，为计算实验数据支持的科学和工程建立新的研究和学习社区，外联一般公众（通过在线游戏工具和科学中心的实践活动，了解基于模拟的工程的力量）以及研究与教育的结合。该项目是一个全面的努力，严格整合高通量合成和表征的发展，提高访问和高性能计算资源的可用性，以解决材料的逆向设计问题。将开发一个计算框架，扫描数百万种可能的结构排列和组合，以进行属性预测，并通过实验进行验证。该框架将（1）采用大规模并行分子动力学模拟来预测纳米材料的输运性质，（2）合成缺陷工程纳米结构并测量相应的输运性质，（三）通过基于遗传算法的混合优化方案来整合上述纳米级计算和实验，以预测针对指定输运性质的最佳材料结构，并制定一个层次结构，最后（4）通过实验验证逆设计结果来闭合环路。