Modeling Long-Time and Macroscopic Behavior of Complex Atomistic Systems with Application to Silicon-based Lithium Batteries

在许多应用领域中，系统的行为敏感地取决于与原子尺度有关的性质，即埃和飞秒尺度。然而，感兴趣的属性和行为通常是宏观的，发生在厘米到米的尺度上，并且具有在几分钟到几年的尺度上缓慢演变的特征。到目前为止，似乎还没有计算上容易处理的基于原子的模型来研究这种从几分钟到几年的时间尺度上的缓慢现象，以及在宏观样本中，同时保持对材料的严格原子描述。该项目解决了预测科学中的这一长期鸿沟。这种方法为在原子尺度上研究由缓慢的、耦合的、热-机械-化学过程所介导的器件级特性提供了前所未有的能力。因此，除了锂离子电池的这种应用外，这种方法作为一种使能工具，在需要仔细计算原子水平的过程的同时，还需要阐明长时间尺度上的宏观性质，例如合金和辐照材料的稳定性、互连中的电迁移、腐蚀和环境辅助破裂等，预计将产生深远的影响。该项目开发的经验原子级动力学模型、变分平均场近似格式、变分时间离散化算法和空间粗粒化算法将被实现到一个经过验证和验证的高性能计算求解器--扩展拟连续体(XQC)求解器中，以在社区中广泛传播。它结合了非平衡统计力学和动力学与近似理论的原理。这种方法为研究复杂的多组分原子体系的长期宏观行为提供了前所未有的能力，该体系在原子尺度上受缓慢的、耦合的热-力-化学过程的调节。将这一新的方法应用于体硅和纳米线(SiNW)中的硅锂离子的研究。硅作为高能量密度的负极材料在锂离子电池中的潜在用途受到锂离子电池中广泛的机械退化的阻碍。目前仅用于汽车应用的此类存储的全球市场规模估计为15亿美元，预计到本十年结束时将增长到3000%以上。能够在器件级别和大量充放电循环中预测模拟硅锂的能力有望使识别和评估用于锂电池应用的新型纳米工程材料成为可能。

In a number of areas of application, the behavior of systems depends sensitively on properties that pertain to the atomistic scale, i.e., the angstrom and femtosecond scales. However, often the properties and behaviors of interest are macroscopic and take place on the scale of centimeters to meters, and are characterized by slow evolution on the scale of minutes to years. No computationally-tractable atomistically-based models appear to be as yet available to study such slow phenomena over time scales of the order of minutes to years and in macroscopic samples while maintaining a strictly atomistic description of the material. This project addresses this chronic gap in predictive science. This approach offers unprecedented capability for the study of device-level properties mediated by slow, coupled, thermal-mechanical-chemical processes at the atomistic scale. Thus, beyond this application to Li-ion batteries, this methodology may be expected to have far-reaching impact as an enabling tool in applications requiring the careful accounting of atomic-level processes simultaneously with the elucidation of macroscopic properties over long time scales, e.g., stability of alloys and irradiated materials, electromigration in interconnects, corrosion and environmentally-assisted cracking, among others. The empirical atomic-level kinetic models, variational meanfield approximation schemes, variational time-discretization algorithms and spatial coarse-graining schemes developed under the project will be implemented into a verified and validated high-performance computing solver, the Extended Quasicontinuum (XQC) solver, for broad dissemination in the community.This work is concerned with the further development and implementation of a novel multiscale analysis methodology. It combines elements of non-equilibrium statistical mechanics and kinetic and approximation theory. This approach offers unprecedented capability for the study of the long-term macroscopic behavior of complex multi-species atomistic systems mediated by slow, coupled, thermal-mechanical-chemical processes at atomistic scales. Application of the novel methodology to the investigation of silicon lithiation, both in bulk and in nanowires (SiNW) is considered. The potential use of silicon as a high energy-density anode material in Li-ion based batteries is hampered by the extensive mechanical degradation that occurs during lithiation. The current global market size for such storage, just for vehicle applications, is estimated at $1.5 billion, and is expected to grow to by more than 3,000% by the end of the decade. The ability to simulate silicon lithiation predictively at the device level and over large numbers of charge/discharge cycles is expected to enable the identification and assessment of novel nano-engineered materials for Li battery applications.