Machine learning-guided discovery of multifunctional Nb2CX2 MXenes (X = S, Se, Te): insights into mechanical properties and thermal conductivity

Abstract

The lattice thermal conductivity of Nb₂CX₂ (X = S, Se, Te) MXenes has been investigated with machine learning force fields (MLFF). To validate the precision of the MLFF method, structural parameters and elastic properties were analyzed and compared between density functional theory (DFT) and MLFF results. The results showed minimal deviations, demonstrating that the MLFF method provides reliable accuracy at the DFT level. The calculated Young's modulus is high, indicating the strong mechanical properties of these materials and their potential for energy storage applications. Interestingly, functionalization with heavier terminal groups reduces the lattice thermal conductivity (κ_L), with Nb₂CS₂, Nb₂CSe₂, and Nb₂CTe₂ exhibiting values of 16.93, 12.05, and 6.01 W m⁻¹ K⁻¹ at 300 K, respectively. This trend is further explained through phonon group velocities, phonon lifetimes, and Grüneisen parameters, which provide deeper insight into the underlying thermal transport mechanisms. Our findings highlight the key role of surface functionalization in tailoring thermal transport, paving the way for optimized MXenes in thermal management. This study also establishes MLFF as a highly efficient, accurate, and cost-effective tool for accelerating materials discovery in energy and thermal applications.