Neural-Network-Based Molecular Dynamics Unravel Generic Multitransition States in CVD Growth of Planar Honeycomb 2D Lattices
A chemistry-aware active-learning distillation framework to construct computationally efficient, high-fidelity machine-learning force fields (MLFFs) with near quantum-chemical accuracy and reveals a universal, nonclassical nucleation pathway governed by a generic multitransition states (MTSs) mechanism.