Abstract
The adequate choice of the interaction model is essential to reproduce qualitatively and estimate quantitatively the experimentally observed characteristics of materials or phenomena in computer simulations. Here we present the results of a benchmarking of density-functional theory calculations of rigid and flexible metal-organic frameworks (MOFs). The stability of these systems depends on the dispersion interactions. We compare the performance of two functionals, Perdew-Burke-Ernzerhof (PBE) and PBE designed for solids, with and without the dispersion corrections (D2 and TS), in reproducing the high-accuracy low-temperature X-ray and neutron diffraction data for both groups of MOFs. We focus our analysis on the key structural parameters: the lattice parameters, bond lengths, and angles. We show that the dispersion long range correction is essential to stabilize the structures and, in some cases, to converge the system to a geometry that is in line with the experimentally observed structure, especially for breathing MIL-53 structures or zeolitic imidazolate frameworks. We find that for all structures and all analyzed parameters, the D2-corrected PBE functional performs the best, except for bonds involving the metal ions; however, even for these bonds the difference between the experimentally observed and calculated lengths is small. Therefore, we recommend the use of the PBE-D2 functional in further numerical analyses of rigid and flexible nanoporous MOFs.