(301g) Obtaining Homogeneous Fluid Properties From An Inhomogeneous Simulation

The chemical potential is a very useful and important property. It can be related to many different phenomena, such as phase equilibria, transport processes, and chemical reaction rates. Better methods are needed to predict chemical potentials from computer simulations. For instance, several of the current methods, such as Grand Canonical Monte Carlo and Widom's method, become inefficient or may even fail for dense liquids. A new method, called the chemical potential perturbation (CPP) method, was presented at the 2009 AIChE Annual Meeting. The CPP method doesn't require difficult particle insertions. Instead, an external force-field is applied to the simulation, causing the composition and density to depend upon position in the simulation cell. Because the system is allowed to equilibrate, the chemical potential (relative to some reference state) as a function of composition and density can be determined from the applied field, after correcting for effects of the inhomogeneity of the system. The CPP method has been further developed and refined to predict differences in the chemical potential, free energy, and pressure of a homogeneous (bulk) fluid at each density in the simulation of an inhomogeneous fluid. We describe three methods to obtain the contribution of the inhomegeneity of the system. The first uses van der Waals density gradient theory, the second uses a Taylor series expansion in density gradients of the normal and tangential components of the local pressure tensor, and the third uses the Triezenberg-Zwanzig definition of surface tension and the direct correlation function. We present results for a Lennard-Jones fluids at supercritical (shown in the figure), two-phase, liquid, and vapor conditions. In particular, the CPP method works well for dense fluids where other methods become inefficient or even fail.