Predicting the properties of complex materials from molecular simulations

计算机模拟已经成为科学研究中实验和理论的宝贵合作伙伴。它有望高精度地预测凝聚态的复杂行为，探索实验无法获得的参数，并指导新功能材料的有针对性的设计。这项研究使用分子水平模拟作为计算显微镜，以更好地了解无定形固体如何在应力下流动。重要的例子包括许多日常应用中熟悉的聚合物和金属玻璃，但我们对这些材料中塑性的基本机制的了解远远比不上我们对金属等结晶固体的了解。无序固体中的形变物理学为检验和探索非平衡统计物理学中的新概念提供了一个极具挑战性的机会。为了提高工程实践的可靠性，我们发展了一种多尺度方法，其中在原子水平上塑性活动的时空组织反映在本构模型中，该本构模型高保真地预测了大范围参数的力学行为。 纳米物理对宏观材料行为有很大影响的第二类材料是高电荷软材料，如胶体大离子和包括蛋白质和DNA在内的所有生物分子。它们的结构和功能受生物分子、周围离子和溶剂化水分子之间的静电相互作用的微妙相互作用所支配。完全原子化的模拟非常详细地解释了所有这些成分，但只能描述几纳秒的生物物理过程。我们将开发新的模拟方法，极大地简化计算，同时保留有关溶液中原子级介电响应的关键信息。这些方法将为研究人工和生物纳米粒子的集体行为、DNA链的结构和功能打开大门，并将为许多生物技术应用中至关重要的病毒自组装和药物输送机制提供新的见解。

Computer simulation has become an invaluable partner of experiment and theory in scientific research. It holds the promise of predicting the complex behaviour of condensed matter with high accuracy, exploring parameters inaccessible to experiments, and guiding the targeted design of new functional materials. This research employs molecular level simulations as computational microscope to better understand how amorphous solids flow under stress. Important examples include polymer and metallic glasses that are familiar from many everyday applications, but our understanding of the elementary mechanisms of plasticity in these materials is far less advanced than our knowledge of crystalline solids such as metals. The physics of deformation in disordered solids provides a challenging opportunity to test and explore new concepts in nonequilibrium statistical physics. In order to improve the reliability of engineering practices, we develop a multiscale approach, where the spatio- temporal organization of plastic activity at the atomic level is reflected in constitutive models that predict the mechanical behaviour for a wide range of parameters with high fidelity. A second class of materials in which nanoscale physics greatly impacts the macroscopic material behaviour are highly charged soft materials such as collodial macroions and all biomolecules including proteins and DNA. Their structure and function is governed by a delicate interplay of electrostatic interactions between biomolecules, surrounding ions, and the solvating water molecules. Fully atomistic simulations account for all these constituents in great detail, but can only describe biophysical processes for a few nanoseconds. We will develop novel simulation methodologies that greatly simplify the computation while retaining critical information about the atomic level dielectric response in solution. These methods will open the door to studying the collective behaviour of artificial as well as biological nanoparticles, the structure and function of DNA strands and will provide new insight into the mechanisms of viral self-assembly and drug delivery that are critical in many biotechnology applications.