Abstract Acidic corrosion undermines infrastructure and energy systems, motivating quantitative models for corrosion prediction. However, existing models often lack accurate activation energies, neglect potential-dependent coverages, and oversimplify anodic dissolution as a concerted multi-electron event. Accordingly, we present a corrosion-prediction framework integrating first-principles calculations with machine-learning molecular dynamics. Our approach computes free-energy barriers at solid–liquid interfaces, incorporates competitive adsorption to determine potential-dependent surface coverages, and resolves Butler–Volmer kinetics of dissolution and hydrogen evolution. Applied to bcc-Fe(110), our model determines a 0.76 eV rate-determining barrier for anodic dissolution via an adsorbed FeOH intermediate, and identifies hydrogen evolution as Volmer-controlled. Predicted apparent activation energy, exchange current densities, corrosion potential, and corrosion current agree with experiment. Furthermore, the model successfully captures alloying effects, with Mn lowering both anodic and cathodic barriers and accelerating corrosion. Transferable and mechanism-based, our model offers a powerful tool for predicting corrosion across metals and guiding corrosion-resistant alloy design.
Bao et al. (Mon,) studied this question.