Abstract

Molecular adsorption is a fundamental factor in gas storage applications and boosting the adsorption of molecular hydrogen would greatly improve hydrogen technology. However, adsorption interactions can be challenging to predict accurately. While density functional methods are routinely used to model adsorption, it is known that the choice of density functional approximation (DFA) can have a considerable impact on the predicted adsorption energies and mechanisms. Fixed-node diffusion Monte Carlo (FN-DMC) is a many-body wavefunction based method that provides highly accurate predictions in non-covalently bound systems, where the uncontrolled error from the fixed-node approximation is typically small. As such FN-DMC can be used to guide the use of DFAs and provide useful mechanistic insights. However, it has been previously shown that hydrogen adsorption can have a small degree of covalency on metal doped carbon allotropes. To gauge the impact of the fixed-nodes in FN-DMC for weakly covalent interactions, we predict small analogous molecular complexes using coupled cluster theory for reference. We find that coupled cluster theory and FN-DMC predict a consistent picture of molecular hydrogen adsorption, despite the fixed-node approximation. Thus, we build our understanding from the small molecular complexes to the large extended materials, even as we tentatively go beyond non-covalent interactions, towards charge transfer and weak covalent bonding. Ultimately, we find that boosting the hydrogen molecule adsorption energy is not as straightforward as predictions from DFAs indicate.