Accelerating Lattice Thermal Conductivity Calculations in MXenes: A Machine Learning Force Field Approach.

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Materials Science

Background:

• Lattice thermal conductivity is crucial for materials performance.
• Traditional methods like the phonon Boltzmann transport equation (PBTE) with density functional theory (DFT) are computationally expensive.
• MXenes, particularly Ti2C and Ti3C2, are promising 2D materials with tunable thermal properties.

Purpose of the Study:

• To predict the lattice thermal conductivity of Ti2C and Ti3C2 MXenes and their functionalized variants.
• To evaluate the efficiency of active DFT-based on-the-fly machine learning force fields (MLFF) for thermal transport calculations.
• To investigate the impact of surface functionalization (O, F, OH) on the thermal conductivity of MXenes.

Main Methods:

• Utilized active DFT-based on-the-fly machine learning force fields (MLFF) for calculations.
• Predicted lattice thermal conductivity for pristine Ti2C and Ti3C2 MXenes.
• Investigated the effect of oxygen (O), fluorine (F), and hydroxyl (OH) surface terminations on thermal conductivity.

Main Results:

• Predicted lattice thermal conductivities of 73.10 W m⁻¹ K⁻¹ for Ti2C and 101.15 W m⁻¹ K⁻¹ for Ti3C2.
• Observed a significant reduction in lattice thermal conductivity upon introduction of surface functional groups.
• MLFF predictions were tens to thousands of times faster than conventional DFT calculations.

Conclusions:

• MLFF provides a highly efficient alternative to traditional DFT methods for calculating lattice thermal conductivity in MXenes.
• Surface functionalization is an effective strategy to tune the thermal properties of 2D MXenes.
• MLFF shows great potential for accelerating the exploration and optimization of thermal transport in advanced materials.