Collaborative Research: Directed Design of Self-Assembling Materials: Enabling therModynamic Steering on Parallel computing clusters

自组装和定向组装被广泛应用，是创造先进功能、纳米结构或分层结构材料的有力途径。迄今为止，通过自组装或定向组装设计先进材料的进展在很大程度上是经验性的，对自组装过程的理论理解或控制有限。在最基本的层面上，人们希望了解中尺度或宏观尺度自组装的给定状态是如何在其分子成分的化学结构中编码的，并利用这种理解来设计具有特定所需性能的定制材料。本研究涉及热力学转向的发展，这是一套新的计算方法，用于先进材料的第一性原理设计，通过化学结构空间主动转向统计机械系统，达到具有规定集体性质的目标状态。热力学控制包括绘制自组装材料的自由能图，将其作为一系列热力学条件下化学结构的函数，以识别自组装成特定目标结构的微观成分。这是具有挑战性的，因为“搜索空间”（微观成分的所有可能的化学变化的空间）和“目标空间”（中尺度或宏观尺度自组装的所有竞争状态的空间）是非常高维的空间，排除了系统的探索。这些挑战是通过实施重要采样方法来优先探索搜索空间中最相关的区域，开发快速生成竞争热力学状态库的算法，应用现代方法在分子模拟中进行自由能计算，以及有效使用并行计算集群来解决的。这种方法被应用于球形胶体的设计，这些胶体可以自发地组装成不寻常的晶体结构（例如，金刚石晶格）。

Self-assembly and directed assembly are widely utilized and powerful routes to the creation of advanced functional, nanostructured, or hierarchically structured materials. To date, progress in the design of advanced materials via self-assembly or directed assembly has been largely empirical, with limited theoretical understanding or control of the self-assembly process. At the most fundamental level, one would like to understand how a given state of mesoscale or macroscale self-assembly is encoded in the chemical structure of its molecular constituents, and to exploit this understanding to design tailor-made materials with specific desired properties. This research involves the development of thermodynamic steering, a set of novel computational methods for the first-principles design of advanced materials by active steering of a statistical mechanical system through chemical structure space toward a target state having prescribed collective properties. Thermodynamic steering involves mapping the free-energy landscape of self-assembling materials as a function of chemical structure over a range of thermodynamic conditions to identify microscopic constituents that self-assemble into a specified target structure. This is challenging because the 'search space' (the space of all possible chemical variations of microscopic constituents) and the 'target space' (the space of all competing states of mesoscale or macroscale self-assembly) are very high-dimensional spaces, precluding systematic exploration. These challenges are addressed through implementation of importance sampling methods for preferentially exploring the most relevant regions of search space, development of algorithms for rapidly generating libraries of competing thermodynamic states, application of modern methods for free energy computation in molecular simulations, and efficient use of parallel computing clusters. This methodology is applied to the design of spherical colloids that spontaneously assemble into unusual crystal structures (e.g., the diamond lattice).