Efficient and Accurate Prediction of Adsorption Energies via a Hybrid PBE/CCSD(T) Scheme Using Hydrogen Saturated Clusters

Accurate prediction of adsorption energies remains a central challenge in heterogeneous catalysis, hindered by the trade-off between computational accuracy and efficiency. Conventional DFT with GGA functionals exhibits systematic overbinding errors, while high-accuracy methods like CCSD(T) remain prohibitively expensive for extended surfaces. To bridge this gap, we introduce a novel hybrid PBE/CCSD(T) embedding scheme. Our key innovations are twofold: (i) a robust hydrogen-saturation protocol for clusters that effectively mitigates boundary effects from truncation, and (ii) a linear interpolation strategy that leverages the strong correlation between PBE and CCSD(T) energies on clusters. This approach requires CCSD(T) calculations only for minimal clusters containing 1-3 metal atoms for CO adsorption on metal surfaces. Validated across a series of CO adsorption systems on transition metal surfaces (Pt, Pd, Rh, Cu, Ir, Ru), our method achieves corrected adsorption energies very close to experimental benchmarks. This work provides a general, efficient, and accurate framework for predictive modeling in surface science and catalysis.