Gradient-Driven Diffusion Using Dual Control Volume Grand Canonical Molecular Dynamics (DCV-GCMD)
- PDF / 432,732 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 19 Downloads / 142 Views
GRAND CANONICAL MOLECULAR DYNAMICS (DC V-GCMD)
FRANK VAN SWOL and GRANT S. HEFFELFINGER Parallel Computational Sciences Department, Sandia National Laboratories, Albuquerque, New Mexico 87185-1111 ABSTRACT Recently we developed a new nonequilibrium molecular simulation method [1] that allows the direct study of interdiffusion in multicomponent mixtures. The method combines stochastic insertion and deletion moves characteristic of grand canonical (GC) simulations with molecular dynamics (MD) to control the chemical potential p~t of a species i. Restricting the insertions and deletions to two separate control volumes (CV's) one can apply different •ti's in distinct locations, and thus create chemical potential gradients. DCV-GCMD can be used to study transient phenomena such as the filling of micropores or used in steady-state mode to determine the diffusion coefficients in multicomponent fluid mixtures. We report on the effects of molecular interactions and demonstrate how in a sufficiently nonideal ternary mixture this can lead to up-hill or reverse diffusion. In addition we introduce a novel extension of DCV-GCMD that is specifically designed for the study of gradient-driven diffusion of molecules that are simply too large to be inserted and deleted. INTRODUCTION Molecular diffusion is an essential part of many technologically important processes including crystal growth, membrane separations, and reaction kinetics. Theoretical approaches aimed at predicting diffusion behavior in a multicomponent system generally require detailed knowledge of both the thermodynamic behavior of the mixture and the diffusion coefficient matrix, [D], of mutual diffusion coefficients. Whereas simulations and molecular theory have both beeni very successfut at providing the equilibrium thermodynamic properties, determining transport properties has presented more serious difficulties. Recently we introduced a general and direct method for studying diffusion in mixtures [1]. The method combines grand canonical Monte Carlo (GCMC) techniques with molecular dynamics (MD) yielding a simulation that employs two control volumes. The control volumes are the only regions in space where we use GCMC moves (i.e. atom insertions and deletions). The atoms are moved through space by using MD exclusively, irrespective of the region. In both control volumes we have the freedom to select the chemical potential for each species. By setting different chemical potentials in distinct control volumes we can force the system to exhibit chemical potential gradients that will result in a nonequilibrium simulation of steady state diffusion. By measuring the atom fluxes and the steady-state density profile, we can use Fick's Law to determine the diffusion coefficients. To apply DCV-GCMD one must ensure that a sufficient number of successful atom insertions/deletions are performed in order that the requested chemical potentials are indeed reached inside each control volume. In practice, this implies a large number of attempted atom insertions/deletions per MD time
Data Loading...
