We establish an end-to-end framework that generalizes across alloy families and, applied here to Co-Cu-Fe-Mo-Ni, maps alloy composition to B5 step ensembles (B5 sites) on fcc(211), *N-adsorption energies ΔE(*N) and rates for ammonia decomposition reaction (ADR). A DFT-trained cluster-expansion (CE) model, combined with Metropolis Monte Carlo (MMC) and microkinetics, predicts site-resolved ΔE(*N) and enables composition-wide predictions of site-specific turnover frequencies (TOFs) and surface-averaged activities (⟨TOF⟩). MMC reveals temperature-driven Cu enrichment in the outermost layer, shifting ΔE(*N) toward weaker binding relative to statistically random surfaces and suppresses ⟨TOF⟩. Reducing Cu content systematically enhances activity, whereas Cu-free Co-Fe-Mo-Ni medium-entropy alloys cluster near the volcano maximum and deliver high, composition-robust rates. Site-level analysis shows that the most active B5 sites are Cu-lean and typically multimetallic, consistent with surface-averaged trends. DFT validation on 40 CE-screened high-activity B5 sites confirms predictive fidelity. The framework provides practical, testable design rules─minimize Cu participation at B5 and preserve configurational disorder─and is readily extensible to other alloy families and to both thermochemical and electrochemical reactions.
Wang et al. (Wed,) studied this question.