The CO Puzzle is Resolvable Using Self-Consistent DFT+U Calculations

Reliable calculation of adsorbate binding is central to understand catalysts using simulations as errors propagate to phase diagrams, microkinetic models, and catalyst design predictions. The most well-known reliability issue is the “CO puzzle” as standard density functional theory (DFT) often predicts the wrong adsorption site and overbinds CO on transition-metal surfaces that affect modeling CO oxidation, water-gas shift, Fischer-Tropsch synthesis, and methanol synthesis among others. Here we show that a simple, self-consistent DFT+U scheme applied to the C and O p orbitals provides a practical route to resolve the CO puzzle. We find that a single DFT+U fit generally restores the experimental site preference, vibrational frequency, and adsorption energies within ±0.1 eV across Ag(111), Au(111), Co(0001), Cu(111), Ir(111), Pd(111), Pt(111), Ni(111), Rh(111), and Ru(0001) surfaces for several exchange-correlation functionals, with revPBE-D3(0)+U agreeing best with experiments. In addition, the same DFT+U fit allows to correct the O₂ gas-phase atomization energy, another great challenge in catalyst modeling. Partial density of states analysis confirms that the +U correction moves the 2π* orbital levels of CO favorably toward experimental photoemission measurements providing a physical motivation why the correction works in general. DFT+U is an ideal solution to retain a fully variational and geometry-optimizable framework that can be deployed easily to study CO binding to transition metal surfaces in established computational workflows.