Numerous problems in molecular and interfacial physics involve processes spanning many different length scales, a feature that has motivated efforts in recent years to develop hybrid simulation techniques that allow for atomistic treatment of select regions, and a coarse-grained (e.g. continuum) description for other parts of the problem where a high level of detail is not necessary. Previously, we developed a multiscale simulation strategy using a stochastic particle-based technique called “smoothed dissipative particle dynamics” (SDPD). SDPD is a thermodynamically consistent formulation of continuum solvers such as smoothed particle hydrodynamics that features scale-dependent thermal fluctuations. Using our multiscale approach, it is possible to couple a molecular dynamics (MD) region to a hierarchy of SDPD domains, where each one is characterized by a different length scale. This allows for simulations that simultaneously include regions spanning from the atomistic to the non-fluctuating continuum limit.
While there have been a number of hybrid simulation strategies for single-component fluids proposed in the last few years, extending these types of approaches to multicomponent systems remains a major challenge. In this talk, we describe a novel generalization of our multiscale methods to systems that involve one or more dissolved species. First, we develop a new multicomponent formulation of SDPD through a particle discretization of the diffusion equation with fluctuations in the concentration field. Using this approach, we consider several simple systems and describe how to bridge multicomponent SDPD with current techniques for multiscale SDPD simulation. This is achieved by carefully constructing an interface between SDPD fluids with a different resolution such that momentum, mass, and the amount of solute in the system are globally conserved. In addition, this multicomponent SDPD model can be reconciled with our multiscale MD-SDPD coupling methodology. We discuss the construction of interfaces for coupled MD-SDPD simulations, which requires Monte Carlo trials to match chemical potentials of domains with different resolution. Finally, we consider several simple case studies and demonstrate that these approaches reproduce correct thermodynamic properties at equilibrium.
See more of this Group/Topical: Computational Molecular Science and Engineering Forum