Corrugated graphene exposes the limits of a widely used ab initio van der Waals DFT functional

Theoretical formulations capable of modeling chemical interactions over 3-4 orders of magnitude of bond strength, from covalent to van der Waals (vdW) forces, are one of the primary goals in materials physics, and chemistry. Development of vdW corrections for density-functional theory has thus been a major research field for two decades. While many of these corrections are semiempirical, more theoretically rigorous ab initio functionals have been developed. The ab initio functional vdW-DF2, when coupled with the reoptimized B86 exchange function (vdW-DF2-rB86), has typically performed as well, if not better than most semiempirical formulations. Here we present a system, Co intercalation of graphene on Ir(111), for which a semiempirical correction predicts local corrugation maxima in locations at which the vdW-DF2-rB86 functional predicts global minima. Sub-angstrom precision quantitative structural measurements show better agreement with the semiempirical correction. We posit that it is balancing the weak vdW interaction with the stronger, even covalent, interactions that proves a challenge for the vdW-DF2-rB86 functional.