Simulating vapor–liquid nucleation of n-alkanes

Abstract

A combination of the aggregation-volume-bias Monte Carlo algorithm, the configurational-bias Monte Carlo algorithm, and the umbrella sampling technique was applied to investigate homogeneous vapor–liquid nucleation in ethane, n-butane, and n-heptane. The simple transferable potentials for phase equilibria-united atom (TraPPE-UA) force field was used in this investigation. It was found that for the n-heptane case, the TraPPE-UA force field predicted a nucleation rate that is about three to four orders of magnitude higher than that measured by an upward thermal diffusion cloud chamber experiment. Comparison of the simulation results to the classical nucleation theory (CNT) shows that CNT consistently overestimates the barrier heights for all chain lengths investigated. The offset on the barrier heights was found nearly independent of the supersaturation for both ethane and n-butane, similar to a Lennard-Jones system previously studied. This also directly leads to a good agreement on the cluster sizes between the simulation and the CNT calculated from the nucleation theorem. For n-heptane, however, the offset was found to depend on the supersaturation. It appears that CNT predicts a slightly weaker dependence of the nucleation rate on supersaturation, which agrees with both density functional calculations and the experiments. Structural analysis demonstrates that the orientational order near the surface differs significantly between the critical nucleus and the bulk planar liquid–vapor interface for n-heptane systems, whereas the density in the interior of the critical nucleus is in good agreement with the bulk liquid density. The different surface ordering offers a microscopic explanation for the differences observed for n-heptane between the CNT on one side and experimental observations and simulations on the other side.

Unravelling the Peculiar Nucleation Mechanisms for Non-Ideal Binary Mixtures with Atomistic Simulations

“…However, neither chainlike molecules of n-nonane nor the orientationally dependent hydrogen-bonding interactions (involved by ethanol) can be modeled through the use of a spherical LJ bead. 39 Thus, it is difficult to draw connections between the systems examined by previous theoretical studies employing LJ models and the n-nonane/ethanol mixture studied here with atom-based force fields.…”

Improving the efficiency of configurational-bias Monte Carlo: A density-guided method for generating bending angle trials for linear and branched molecules

“…For example, we repeated the phase equilibrium calculation reported for an nheptane system (with 300 molecules and a liquid box close to 40 Å) in Ref. 30 and found that more than 60% of the computer time was spent on the generation of the intramolecular angles. For an isolated molecule in a gas-phase, the angle generation consumed over 99% of the computer time.…”

Determination of surface tension in binary mixtures using transition-matrix Monte Carlo

“…A number of other studies have found that twobody molecular models fail to predict accurately and simultaneously the phase behavior and surface tension of real systems. 3,4,8,13,60 For example, the pure Lennard-Jones fluid, a potential commonly used in molecular modeling, reproduces the phase behavior of argon, but is unable to reproduce simultaneously and quantitatively its surface tension. 3,4,13,60 It is generally agreed upon that many-body effects need to be included in order to accurately reproduce the surface tension of argon.…”

“…The most straightforward approach is to simulate directly the interfacial region using molecular-dynamics or Monte Carlo simulation. [1][2][3][4][5][6][7][8][9][10] In this case, one simply juxtaposes the coexisting phases of interest and waits for an explicit interface to form and stabilize, which can require a significant amount of time. The interfacial tension is given by the mean difference between the normal and transverse components of the stress tensor calculated over the interfacial region.…”