Predicting vapor-liquid phase equilibria with augmented ab initio interatomic potentials.

Area of Science:

• Computational chemistry
• Physical chemistry
• Materials science

Background:

• Accurate prediction of vapor-liquid phase equilibria is crucial for chemical engineering and materials science.
• Traditional empirical models often require extensive parameterization.
• Ab initio methods offer a theoretically grounded approach to developing interatomic potentials.

Purpose of the Study:

• To investigate the accuracy of ab initio interatomic potentials for predicting vapor-liquid equilibria.
• To assess the impact of including three-body interactions on predictive accuracy.
• To compare the performance of ab initio potentials against empirical models.

Main Methods:

• Monte Carlo simulations were performed for argon and krypton.
• Seventeen different ab initio interatomic potentials, combining two-body and three-body terms, were evaluated.
• Three-body interactions were modeled using Axilrod-Teller-Muto and effective Marcelli-Wang-Sadus potentials.
• Results were validated against experimental and reference data across a wide range of conditions.

Main Results:

• Ab initio potentials, particularly those incorporating theoretically based three-body terms, demonstrate systematic improvements in predicting vapor-liquid equilibria.
• The predictive accuracy of recent ab initio potentials is comparable to that of established empirical models.
• The Marcelli-Wang-Sadus potential, when combined with two-body ab initio models, offers a computationally efficient and accurate method for VLE prediction.

Conclusions:

• Ab initio interatomic potentials are highly effective for predicting vapor-liquid equilibria.
• The inclusion of three-body interactions significantly enhances the accuracy of these predictions.
• The Marcelli-Wang-Sadus potential provides a promising avenue for cost-effective and accurate VLE estimations.