Short-Range Order in Group-IV Alloys: Insights from First-Principles and Large-Scale Atomistic Simulations Enabled by Machine-Learning Potentials

Group-IV semiconductor alloys are promising materials for next-generation electronics, photonics, and topological quantum technologies. A key challenge in realizing their potential lies in understanding their complex structural behavior, particularly short-range order (SRO), the subtle local atomic arrangements that can significantly influence electronic properties. Accurate modeling using density functional theory (DFT) is often limited by spatiotemporal constraints, creating a gap between theory and experiment. To bridge this gap, we develop machine-learning interatomic potentials based on a neuroevolution potential (NEP) approach trained on high-quality DFT datasets. These NEPs achieve near-DFT accuracy with the efficiency of empirical potentials, enabling large-scale atomistic simulations that can be directly compared with APT, 4D-STEM, and EXAFS measurements. Through theory-experiment synergy, we identify preferred atomic arrangements, confirming the presence of SRO, revealing its atomic-scale structure, and establishing atomic-order engineering as a powerful third degree of freedom for band-structure design, alongside composition and strain.