The perplexing structural diversity of metal-organic frameworks (MOFs) leads to astonishing unique properties and versatile potential technological applications of this porous materials class. However, whenever the organic linker molecules in a specific MOF compound may adopt multiple spatial orientations in-between the inorganic nodes they coordinate to, a near infinite number of possible MOF configurations arises through orientational isomerism. The latter is particularly challenging to account for in computational studies. In this work, we present a novel simulated annealing (SA) approach via Monte Carlo (MC) runs to gain insights into linker molecule orientations within such compounds, at the example of the three increasingly complex MOF systems MOF-5-OH, SNU-70 and UiO-66(Zr)-NH2. We thereby successfully identify reasonable approximations to the global minimum structures for all MOFs with regard to collective linker molecule orientations by performing random re-orientation attempts of a randomly chosen organic linker in each MC step. The efficiency of the simulated annealing procedure is improved and guaranteed by exploiting the accuracy and speed of the state-of-the-art neural network potential (NNP) MACE-MP-0a in all energy and force calculations. Critically, a substantial potential energy gain is observed during annealing, accompanied by systematic structural changes within the MOF compounds. The combined results of these energetic and structural characteristics highlight how it is indeed cooperative effects arising from the organic linker molecules that play a decisive role in determining the most favorable MOF configuration. Intriguingly, root-mean-square-deviation (RMSD) calculations based exclusively on atoms of the inorganic nodes are found to be sensitive structural descriptors for tracking linker molecule re-orientations, while powder X-ray diffraction (PXRD) patterns and radial distribution functions (RDFs) derived from molecular dynamics (MD) simulations at standard conditions show less distinctive trends. In summary, the utility of the presented simulated annealing strategy is justifiable by its broad applicability for investigating linker-related orientational isomerism in any relevant MOF as well as the possibility for studying even more complex host-guest systems in a similar manner.
Hörfarter et al. (Thu,) studied this question.