Weakly Bound Materials: Competing Energy Scales, Quantum Effects, and Machine Learning

Weakly bound systems, comprising molecular materials, vdW heterostructures, and interfaces between different material components, give rise to a rich variety of nuclear motion and tuneable nuclear structures that are tightly connected to diverse electronic properties in these systems. In this contribution, I will discuss how we push the limits of density-functional theory and different ab initio techniques that capture nuclear motion to unravel the properties of realistic interfaces. I will give an overview of simulation methods that are applicable for large system sizes and that can capture the quantum nature of nuclei in anharmonic potential energy landscapes [1], as well as what the inclusion of non-adiabatic effects in these simulations entails [5]. I will discuss how machine-learning techniques can speed up the computation of nuclear and electronic properties [2, 3], allowing larger and more accurate simulations that can reach thermodynamic limits of first-principles accuracy. I will finish by showing a couple of applications where the quantum nature of the nuclei becomes indispensable to assess the structural and electronic properties of 2D materials and interfaces [4, 5, 6].