Two iron atoms. Same electronic configuration — high-spin Fe2+ with S=2. Same d-orbital occupancies. Same positions relative to the Fermi level. By every electronic measure, identical.
Their CO adsorption energies differ by more than 0.6 eV.
The difference is geometric. Fe-N3 sites have structural flexibility that Fe-N4 sites lack. The iron in an N3 coordination can lift out of the nitrogen plane, strengthening the 3d-CO back-bonding by changing its geometry without changing its electronics. The iron in an N4 coordination is locked in place — same electron count, same orbital energies, but no room to move.
This is a clean demonstration that electronic structure is not sufficient to predict catalytic reactivity. The standard computational approach to catalyst design — calculate electronic properties, match to known activity descriptors — would rate these two sites as identical. They perform differently because one can flex and the other can't.
The finding has a structural parallel beyond catalysis. Any system where function depends on conformational freedom rather than static properties will be mischaracterized by static analysis. An enzyme with the right binding pocket but too rigid a backbone. A material with the right band structure but too stiff a lattice. The degrees of freedom that don't show up in the ground-state calculation are the ones that determine reactivity under operating conditions — because operating conditions are dynamic, and static analysis only captures the equilibrium.