Rough Mob – Roughness and Mobility of Coarse-Grained Molecular Models

软物质系统的分子模拟越来越需要多尺度方法，以保证其准确性和计算可行性。粗粒度模型可以从原子参考模拟中以几种系统的方式生成。他们很好地再现了母体原子模拟的静态结构分布。然而，总的来说，它们无法捕获分子迁移率和随后的输运量，如扩散系数、粘度或某些特征时间。它们的流动性往往高出几个数量级。我们提出了一种方法（“Rough Mob”）来预测这种人为的分子迁移率加速，因为系统的原子模型被粗粒度模型所取代。它的前提是，通过将化学成分的几个真实原子合并成一个超原子，它的一些表面结构会丢失：一个粗糙的原子群被平滑成一个没有结构的球形超原子。我们进一步断言，正是这种表面粗糙度的降低，使得光滑的超原子能够比粗糙的原子部分更快地相互滑动。这个想法已经在我们自己发表的一组脂肪族和芳香烃溶剂的初步研究中得到了验证，这些溶剂都被粗粒化成一个单一的超原子。结果表明，适当定义的原子模型和粗粒度模型之间的纯几何粗糙度差异与观测到的迁移率增加之间存在良好的线性关系。这是非常令人鼓舞的，因为它开辟了一种预测传输系数的方法，通过在粗粒度模拟中计算它们，然后使用Rough Mob加速因子将它们重新缩放到现实值，而不需要昂贵或不可行的原子计算。在提出的工作中，这种方法将推广到更大的系统和条件集合：(i)系统，在粗粒度水平上，需要多个超原子，包括聚合物。（ii）不同化学物质的混合物。（iii）不同的状态点（温度、密度、成分）。要对所有这些情况进行验证。最后，将Rough Mob概念与另外两种预测粗粒度上的先验加速的尝试进行比较：过度熵缩放和近似Mori-Zwanzig方法。本建议的最终目的是促进分子模拟方法的探索，这些方法能够可靠地计算复杂软物质系统的流变和输运性质。它超越了纯粹的学术兴趣：有了这些方法，聚合物挤压和注射成型等工艺可以被设计成更具成本效益和节能的。

The molecular simulation of soft-matter systems increasingly needs multiscale methods in order to be both accurate and computationally feasible. Coarse-grained models can be generated in several systematic ways from atomistic reference simulations. They reproduce static structural distributions of the parent atomistic simulations very well. They collectively fail, however, to capture the molecular mobility and the ensuing transport quantities like the diffusion coefficient, the viscosity or certain characteristic times. Their mobility is often several orders of magnitude too high. We propose a method (“Rough Mob”) to predict this artificial acceleration of the molecular mobility, as the atomistic model of a system is replaced by a coarse-grained model. Its premise is that, by uniting several real atoms of a chemical moiety into one superatom, some of its surface structure is lost: A rough atomistic group of atoms is smoothened into a structure-less spherical superatom. We further assert that it is this reduction in surface roughness, which enables smooth superatoms to glide past one another faster than rough atomistic moieties. This idea has been tested in our own published preliminary work on a set of aliphatic and aromatic hydrocarbon solvents, which have all been coarse-grained into a single superatom. It was shown that there is a well-defined linear relation between a suitably defined purely geometric roughness difference between atomistic and coarse-grained models and the observed mobility increase. This is very encouraging, as it opens a way of predicting transport coefficients by calculating them in a coarse-grained simulation and then using the Rough Mob acceleration factor to rescale them to realistic values without the need for costly or infeasible atomistic calculations. In the proposed work, this approach is to be generalised to a larger set of systems and conditions: (i) Systems which, at the coarse-grained level, need multiple superatoms, including polymers. (ii) Mixtures of different chemical species. (iii) Different state points (temperature, density, composition). Validations are to be carried out for all these cases. Finally, the Rough Mob concept is to be compared to two other attempts to predict a priori the acceleration upon coarse-graining: excess-entropy scaling and approximate Mori-Zwanzig approaches. The ultimate aim of this proposal is to contribute to the quest for molecular simulation methods, which are able to calculate reliably the rheological and transport properties of complex soft-matter systems. It goes beyond purely academic interest: With such methods available, processes such as polymer extrusion and injection moulding could be designed to be both more cost-effective and energy-effective.