Impact of high-order anharmonicity and nitrogen vacancy defects on the thermal conductivity of MoSi2N4

Accurately determining the thermal transport properties of the two-dimensional material MoSi₂N₄ is essential for its application in micro- and nanoscale semiconductor devices. To date, substantial discrepancy persists between theoretical prediction and experimental measurement of the thermal conductivity κL of MoSi₂N₄, with the underlying physical mechanisms remaining unclear. In this work, the intrinsic κL of MoSi₂N₄ is accurately determined by using the neuroevolutionary potential combined with the phonon Boltzmann transport equation (BTE). Taking the temperature-dependent interatomic force constants extracted from effective harmonic model as the inputs, the room-temperature κL of MoSi₂N₄ calculated by the BTE method (approximately 351.844pt0exW0.16em0exm^-10.16em0exK^-1) exhibits remarkable consistency with results obtained from a homogeneous nonequilibrium molecular-dynamics method (approximately 343.044pt0exW0.16em0exm^-10.16em0exK^-1). These results underscore the critical influence of higher-order phonon scattering and temperature renormalization of force constants on the intrinsic κ_ι of MoSi₂N₄. Our calculations also demonstrate that the thermal transport performance of MoSi₂N₄ could be evidently suppressed by nitrogen vacancy. As for the MoSi₂N₄ sample with 0.02% randomly distributed N vacancy, the κL could be decreased to approximately 175±8.964pt0exW0.16em0exm^-10.16em0exK^-1, which provides a plausible explanation for the discrepancy between theoretical predictions and experimental measurements (approximately 1734pt0exW0.16em0exm^-10.16em0exK^-1). Interestingly, compared to the internal N vacancy, the surface N vacancy exhibits particularly pronounced effect in suppressing the κL of MoSi₂N₄. Such a phenomenon is primarily attributed to significant modifications of the symmetry of the charge density distribution and the geometric environment induced by the surface N vacancy, which weaken Si-N bond interactions and amplify the vibrational motion of Si atoms, thereby enhancing the phonon anharmonicity. This work offers valuable insights into the thermal transport properties of the MoSi₂N₄ family and serves as a case study for exemplifying the use of machine-learning interatomic potentials to investigate the impact of low-concentration disordered vacancy defects on the thermal conductivity of materials.