CDS&E: Multiscale Dynamics Simulation of Self-Assembly and Transport in Nanoparticulate Systems

密苏里大学的CBET 1404482Zhang和湿纳米颗粒系统在包括制药、生物医学和化学加工在内的各种应用中都会遇到。在理解和优化这些系统时，准确的预测能力至关重要。由于这类系统中相互作用力场的物理性质很复杂，因此需要多尺度的方法，需要大规模的模拟。该研究旨在为纳米粒子系统的多尺度计算开发经济有效和准确的预测方法。通过使用数据使能方法整合不同尺度上的子模型，将对纳米粒子在干和湿纳米粒子系统中的自组装和传输进行多尺度动力学模拟。干颗粒系统的独特之处在于，它们既有固体的特征，也有液体的特征，这导致了许多迷人的特征。通过多尺度计算了解纳米颗粒在湿颗粒系统中的输运过程，将通过揭示主要参数来获得理想的最终纳米结构，从而显著改善这一过程。通过将原子间作用力从头计算和参数蒙特卡罗优化相结合，将“目标驱动”方法用于重新生成适当的参数集。将通过数据流支持方法进行涵盖原子-原子、原子-集群和集群-集群尺度的多尺度计算。通过原子-原子水平的模拟，将产生关键的力场参数集；原子-团簇水平的计算将主要研究单个或一组团簇(纳米颗粒)与浸入流动的液体分子之间的相互作用。一旦通过重建创新的笼状粒子和移植到可扩展的LAMMPS来建立所有的基础设施，许多与纳米粒子系统中的自组装和传输相关的现象可以在这个框架下进行严格的计算。通用的“目标驱动”方法将被应用到软件中，以满足许多优化工作的需要，包括获得所需的FF参数。最新开发的原子式笼状颗粒等子模型将进一步赋能在全球拥有非常大量用户的开源分子动力学程序LAMMPS。由于代码也兼容LIGGGHTS(基于LAMMPS)和CFDEM(LAMMPS和OpenFOAM的组合)，他们的用户也肯定会从这个项目获得的功能和数据中受益。研究成果将通过在档案期刊上发表的出版物、在国内和国际会议上的发言以及学生的论文/学位论文来传播。拟议的外展活动还将有助于提高公众对科学和工程的认识和认可，提升科学家和工程师的公众形象，这将在各级教育中发挥非常重要的作用，并可能对技术、经济和社会产生更广泛的影响。

CBET 1404482Zhang, University of MissouriDry and wet nanoparticulate systems are encountered in a variety of applications including pharmaceutics, biomedical and chemical processing. In understanding and optimization of these systems, an accurate predictive capacity is critical. Since the physics of the interacting force fields in such systems are complex, a multi-scale approach is necessary, requiring large-scale simulations. The proposed research aims to develop cost-effective and accurate predictive approaches for multi-scale computations of nanoparticulate systems.Multiscale dynamics simulation of self-assembly and transport of nanoparticles in both dry and wet nanoparticulate systems will be performed by integrating sub-models at different scales using data-enabled approaches. The unique behavior of the dry particulate systems is that they present both solid-like and liquid-like features leading to a number of fascinating characters. Understanding the transport of nanoparticles in wet particulate systems through multi-scale computation will significantly improve the process by revealing dominant parameters to obtain desirable final nanostructure. An "objective-driven" method will be utilized to regenerate appropriate parameter sets by combining Ab initio calculation for interatomic force construction and Monte Carlo optimization of parameters. A multiscale computation, which covers atom-atom, atom-cluster, and cluster-cluster scales, will be carried out through a data-stream support methodology. Through atom-atom level simulation, crucial sets of force field (FF) parameters will be generated; the atom-cluster level computation will mainly investigate interactions between a single or a group of clusters (nanoparticles) and immersing flowing liquid molecules. Once all the infrastructures are well established via reconstructing innovative cage particle and porting to extensible LAMMPS, many phenomena associated with self-assembly and transport in nanoparticulate systems can be rigorously computed under this framework. The general "objective-driven" method will be implemented into software to meet needs of many optimizations works, including obtaining desired FF parameters. The newly developed atomic style-cage granule and other sub-models will further empower the open-source molecular dynamics code LAMMPS, which has a very large number of users around the world. Since the code is also compatible to LIGGGHTS (based on LAMMPS) and CFDEM (a combination of LAMMPS and OpenFOAM), their users will also definitely benefit from the functionalities and data obtained from this project. The outcome of the research will be disseminated through publications in archival journals, presentations at national and international conferences, and students' thesis/dissertations. The proposed outreach activities will also help promoting the public awareness and recognition of science and engineering and promote the public image of scientists and engineers, which will play a very important role in education in all levels and may also have broader impacts on technology, economy, and society.