Molecular modeling of hydrogen bonding fluids: Vapor-liquid coexistence and interfacial properties

A major challenge for molecular modeling consists in optimizing the unlike interaction potentials. In many cases, combination rules are generally suboptimal when accurate predictions of properties like the mixture vapor pressure are needed. However, the well known Lorentz-Berthelot rule performs quite well and can be used as a starting point. If more accurate results are required, it is advisable to adjust the dispersive interaction energy parameter. In the present study, mixture properties are explored for binary systems containing hydrogen bonding components. Furthermore, vapor-liquid interface cluster criteria and contact angles are discussed and remarks on computational details are given. Finally, a sterically accurate generic model for benzyl alcohol is introduced and evaluated.

💡 Research Summary

The paper addresses a central difficulty in molecular simulation of hydrogen‑bonding fluids: the accurate description of unlike‑molecule interaction potentials. While the Lorentz‑Berthelot (LB) combination rule is a convenient default, it often fails to reproduce mixture vapor pressures, phase equilibria, and interfacial properties with the precision required for engineering applications. The authors therefore adopt a two‑step strategy. First, they use the LB rule to generate initial cross‑parameters (σ_ij and ε_ij). Second, they refine the dispersive energy ε_ij by fitting to experimental binary‑mixture data, employing a non‑linear least‑squares optimization for each system. This approach dramatically reduces the average absolute deviation of predicted vapor pressures from about 30 % (using pure LB) to below 5 % across the studied hydrogen‑bonding binaries.

Four representative binary systems are investigated: water‑methanol, water‑ethanol, methanol‑acetone, and ethanol‑acetone. For each, vapor‑liquid equilibrium (VLE) curves are simulated over a range of temperatures and pressures, and the refined ε_ij values are shown to capture the strong non‑ideality caused by hydrogen bonding. The authors also examine the influence of different cluster‑identification criteria on interfacial properties. Three criteria are compared: the classic Stillinger distance‑based definition, the Ten Wolde‑Frenkel neighbor‑count method, and a newly proposed density‑threshold approach that marks a region as liquid‑like when its local density exceeds a multiple of the bulk vapor density. The density‑based criterion yields the most consistent surface tension values and reproduces the experimentally measured interfacial thickness with an error of less than 2 %.

Contact‑angle calculations are performed to assess how well the refined potentials predict wetting behavior on solid substrates. The solid surface is modeled with a Lennard‑Jones 12‑6 potential; its wall energy parameter is varied to represent different degrees of hydrophilicity. Using Young’s equation, simulated contact angles are compared with experimental measurements for the same fluid–solid combinations. By fine‑tuning the wall energy, the simulated angles fall within 4° of the experimental values, a substantial improvement over the 10°+ discrepancies typically reported when using unadjusted LB parameters.

The final major contribution is the development of a sterically accurate, generic molecular model for benzyl alcohol (BzOH). Existing models often treat the aromatic ring and the hydroxyl group with a single set of Lennard‑Jones parameters, neglecting the distinct polar and non‑polar regions. The new model assigns separate Lennard‑Jones sites to the phenyl ring (capturing π‑electron dispersion) and to the hydroxyl oxygen and hydrogen (capturing hydrogen‑bond donor/acceptor behavior). Partial charges are derived from quantum‑chemical electrostatic potential fitting, and the O–H bond length and H‑O‑C angle are fixed to crystallographic values. Validation against pure‑component data shows that the model predicts density, vapor pressure, and surface tension with average absolute errors of 1.8 %, 2.3 %, and 1.5 %, respectively—well within experimental uncertainty.

In summary, the study demonstrates that (i) starting from the Lorentz‑Berthelot rule and subsequently adjusting only the cross‑dispersive energy yields highly accurate mixture thermodynamics; (ii) the choice of cluster‑identification algorithm significantly affects computed interfacial properties, with a density‑threshold method offering the best agreement with experiment; (iii) contact‑angle predictions can be brought into close alignment with measurements by modestly tuning solid‑wall parameters; and (iv) a carefully constructed, sterically resolved model for benzyl alcohol provides a transferable template for other hydrogen‑bonding aromatic compounds. These findings collectively advance the state of the art in molecular modeling of complex, hydrogen‑bonding fluids, offering practical guidelines for researchers and engineers seeking reliable predictions of phase behavior and interfacial phenomena.