Data-Driven Many-Body Models with Chemical Accuracy for CH₄/H₂O Mixtures

Area of Science:

• Computational chemistry and molecular modeling.
• Physical chemistry of liquids and mixtures.
• Development of accurate potential energy functions.

Background:

• Many-body potential energy functions (PEFs) are essential for accurate molecular simulations.
• Existing models like TTM-nrg provide a basis, but improvements are needed for complex systems.
• Coupled cluster reference data offers high accuracy for developing PEFs.

Purpose of the Study:

• To develop and validate many-body PEFs (MB-nrg) for neat methane and methane-water mixtures.
• To assess the accuracy of MB-nrg PEFs in representing many-body effects across different phases.
• To investigate the impact of PEF accuracy on the solvation structure of liquid mixtures.

Main Methods:

• Development of PEFs using the TTM-nrg and MB-nrg theoretical frameworks.
• Utilizing coupled cluster reference data for PEF parameterization.
• Performing molecular dynamics simulations of liquid methane and methane/water mixtures.

Main Results:

• MB-nrg PEFs achieve subchemical accuracy for many-body effects in small clusters.
• MB-nrg PEFs enable predictive simulations from the gas to the liquid phase.
• Accurate short-range interactions, provided by MB-nrg, are critical for quantitative solvation structure in mixtures.

Conclusions:

• MB-nrg PEFs represent a significant advancement in simulating methane and its mixtures.
• Accurate treatment of both polarization and short-range interactions is vital for liquid-phase simulations.
• These PEFs facilitate more reliable predictions of local solvation structures.