Phonon-Mediated Heat Transport in CO₂ Hydrate

Area of Science:

• Materials Science
• Computational Chemistry
• Thermodynamics

Background:

• Gas hydrates, particularly CO2 hydrates, show potential for carbon sequestration and refrigeration.
• Understanding their thermal transport properties is crucial for technological applications.
• Current knowledge of hydrate thermal mechanisms is limited.

Purpose of the Study:

• To develop a machine learning force field for sI-type CO2 hydrate.
• To accurately predict the thermal transport properties of CO2 hydrates.
• To elucidate the mechanisms governing heat transfer in CO2 hydrates.

Main Methods:

• Neuroevolution potential (NEP)-based machine learning force field development.
• Training on first-principles data (DFT).
• Homogeneous nonequilibrium molecular dynamics (HNEMD) simulations.

Main Results:

• The NEP model accurately reproduced structural characteristics (RDFs, lattice expansion) and thermal conductivity.
• Observed an anomalous increase in thermal conductivity (0.58 to 0.72 W·m-1·K-1) between 230-300 K.
• Identified low-frequency phonons (<10 THz) as dominant for heat transfer, with scattering and localized vibrations suppressing conductivity.

Conclusions:

• The developed NEP framework provides accurate predictions of CO2 hydrate thermal properties.
• Mechanistic insights into heat transfer resolve previous controversies in hydrate thermal physics.
• The NEP model serves as a robust tool for advancing hydrate-based carbon capture and thermal energy applications.