Molecular simulation of binary vapour-liquid equilibria with components differing largely in volatility

In this work the Gibbs-Duhem integration method is used in combination with molecular simulation in the semigrand ensemble to simulate vapour-liquid equilibria for the binary mixtures ethane-methane and carbon dioxide-dimethylsulphoxide (DMSO). In both binary mixtures, the two components have a large difference in relative volatility. The starting point for tracing the vapour-liquid phase envelope is either calculated in a pure species NVT Gibbs ensemble simulation or in a pure species NpT +test particle simulation. The chosen method depends on the vapour pressure of the pure species. An equation is presented that relates the initial slope of the Clapeyron equation to an ensemble average in the Gibbs ensemble. The NpT +test particle method is slightly modified in order to improve the convergence characteristics when one of the coexisting phases is a (very) dense phase. Configurational-bias methods are used in the NVT Gibbs ensemble to enhance the acceptance of particle exchanges, and in the NpT +test particle method to grow test particles in a dense liquid phase. Force-biased molecular translations and torque-biased molecular rotations are used to enhance the diffusion of molecules in state space. The work shows the capability of the combination of the NpT +test particle method and Gibbs-Duhem integration to predict vapour-liquid equilibria of binary systems with components differing largely in volatility.