AI-Driven Coverage-Dependent Kinetics for NH ₃ Synthesis on Fe(110)

Adsorbate-adsorbate interaction serves as a critical bridge between macroscopic mass transfer and microscopic adsorption-reaction kinetics in heterogeneous catalysis, yet its dynamic characterization remains underexplored. Herein, we integrate density functional theory (DFT), artificial intelligence (AI)-driven structural screening incorporating a deduplication algorithm and the equivariant graph neural network (NequIP), and coverage-dependent kinetic modeling to establish a quantitative framework for ammonia synthesis over Fe(110). The AI screening framework enables high-precision energy prediction (mean absolute error = 0.028 eV) to efficiently identify the lowest-energy adsorption configuration. The coverage-dependent model corrects significant deviations in traditional models, predicting a turnover frequency (TOF) of 4.4 × 10^-7 s^-1 at 673.15 K and 300 mbar and revealing atomic hydrogen dominance (69.4%) on the interface due to intermediate repulsive interactions. Coverage under temperature adjustment critically regulates the rate-determining steps: N₂ dissociation dominates below 673.15 K, while NH₂ hydrogenation emerges as rate determining at temperatures ≥773.15 K. This work advances the connection between macroscopic conditions and microscopic surface dynamics via coverage, providing a paradigm for catalytic mechanism studies and rational catalyst design, particularly for low-pressure ammonia synthesis.