Machine Learning-Assisted Analysis of Graphene Configurational Tuning and Atomic Patterning

The remarkable electronic, topological, and structural properties of graphene can be tuned by modifying configurations such as stacking, twist angle, defects, dopants, and strain. Using atomistic modeling and scanning transmission electron microscopy, we have studied the top-down atomic patterning of different dopants in graphene. Our results indicate that vacancy motion in bilayer graphene is constrained, which is supported by both experimental data and first-principles/molecular dynamics studies. Consequently, bilayer graphene emerges as an optimal template for property manipulation via doping. Understanding the relationship between different tuning parameters and desired properties is critical for practical applications of graphene. We used machine learning to predict these properties using high-throughput training datasets. Our study enhances our understanding of graphene’s subtle responses to different modifications.