EUROPEAN CONGRESS AND EXHIBITION ON ADVANCED MATERIALS AND PROCESSES - EUROMAT 2021

Computational models of metal-organic frameworks (MOFs) have been mostly limited to infinite periodic systems with unit cells of just a few nanometers in diameter because these are, from a computational perspective, relatively easy to simulate through the use of periodic boundary conditions. However, recent experimental evidence on MOFs has demonstrated how their physical properties are profoundly influenced by the widespread presence of correlated disorder at various length scales [1, 2]. An accurate description of these materials therefore requires modelling techniques that go beyond the nanometer/nanosecond scale in order to incorporate such spatial heterogeneities. Because explicit atomistic simulations of e.g. micrometer-sized MOF crystals are not yet feasible, we may alternatively choose to adopt a coarse-grained (CG) representation.

While preliminary work has demonstrated the potential of such CG models, their widespread application to disordered finite MOF crystals is still faced with major challenges [3]. From a methodological point of view, we here propose a systematic coarse-graining procedure, and identify (i) the automated design of optimal coarse-grained mappings and (ii) the construction of transferable potentials as two of the most important hurdles to overcome. By using a rigorous statistical mechanical framework and basic linear algebra, we elucidate the intrinsic connection between both problems and are able to propose new solution strategies based on the generalized eigendecomposition of the mapping operator and enhanced sampling techniques. The proposed algorithms are universally applicable, and we establish their effectiveness by constructing CG models for a number of MOFs including UiO-66(Zr) and HKUST-1(Cu), and show their correspondence with the underlying atomistic description in terms of structural, mechanical and thermal properties.

[1] Angew. Chem., Int. Ed., 54: 3417, 2015.

[2] Nat. Commun., 9: 1573, 2018.

[3] Dalton Trans., 4510, 4370-4379, 2016.