Thermal and phase transition behavior of 2D quantum materials, enabled by machine-learned interatomic potentials

Two-dimensional (2D) quantum materials are expected to transform conventional electronics for a wide spectrum of applications. Here, we explore the combination of density functional theory (DFT) and machine-learned algorithmic training for the generation of moment-tensor potentials (MTPs) to model single-layer (1L) or bi-layer (2L) transition metal dichalcogenides (TMDs). First, we use the trained MTPs for predicting the thermal transport properties of 1L-MoS₂/WS₂ lateral heterostructures, showing that the thermal conductivity of 2D alloys is highly resilient to sulfur vacancies, and enabling the fine-tuning of material's thermal properties for heat management and energy storage and conversion applications. Furthermore, we employ our trained MTPs for studying the temperature-dependent phase transition dynamics of R-stacked 2L-TMDs, aiming to understand their paraelectric switching behavior. This is useful for modeling the ferroelectric properties of quantum systems that will be crucial components in the design and implementation of advanced electronic circuitry.