On the use of gradual transformation methods in computing phase behavior

Edward Maginn
University of Notre Dame

Many of the atomistic simulation techniques used to compute the equilibrium between phases (vapor-liquid, solid-liquid and even liquid-liquid) rely upon exchanges of molecules between the different phases. When the molecules in question are large, complex and/or have directional interactions with other species, there are enthalpic barriers that make such exchanges very difficult. When one or more of the phases is dense or ordered, entropic barriers also exist which lower the probability of successful exchanges. Here we provide an overview of three different techniques we have used in recent years to overcome these sampling difficulties within a Monte Carlo or molecular dynamics framework. The first method, which we called “continuous fractional component” Monte Carlo relies upon the gradual creation and destruction of molecules in different phases in a coupled manner. By adding biasing weights to the changes, significant free energy barriers can be overcome. We show how this method has been used to compute isotherms of gas absorption in ionic liquids. The second method, which we call pseudo-supercritical path sampling, provides a means for computing the free energy difference between a crystal material and its liquid at a given temperature by gradually transforming the crystal to a liquid along a suitable path. We show how this method can be used to compute melting points and solubility limits of crystalline materials. Finally, free energies of transfer / solvation can be computed using so-called “expanded ensemble” methods in which a random walk is performed between states where a solute is fully coupled and fully decoupled with the solvent. We show how this method can be combined with other biased sampling methods to efficiently compute the solvation free energy of very complex molecules.

Presentation (PDF File)

Back to Workshop IV: Physical Frameworks for Sampling Chemical Compound Space