(609d) Implications of Material Functionalization and Solvent Identity on the Synthesizability and Polymorph Selection of Metal-Organic Frameworks

Metal-Organic Frameworks (MOFs) are a class of porous materials with highly tunable, modular structures promising for numerous applications. MOFs are formed by interconnected organic and inorganic building blocks, and different MOFs can be obtained by choosing a different combination of building blocks. Also, sometimes the same combination of building blocks can plausibly interconnect in arrangement of different topology (i.e., polymorphism), which can have drastically different properties. Considering the overwhelming combinatorics of building blocks and topologies, the “design space” for MOFs is practically infinite. Thus, high throughput computational screening established itself as the go-to approach to identify promising MOF designs for a given application. However, the “conversion” of promising computationally identified MOF design to actual synthesis of new MOF is dramatically low due to uncertainty on whether a given design can actually be made.

In earlier work, upon calculation of free energies on a database of 8,000+ MOFs, we found MOFs known to have been synthesized to fall below a 4.4 kJ/mol per MOF atom free energy threshold. Among synthesized MOFs where polymorphism was possible, we found that the structure known to have been synthesized corresponded to the lowest-free-energy polymorph in 80% of the experimentally observed cases. Thus, a standing question was what could explain the observed polymorph selection in the remaining 20% of cases. Considering that solvent selection may play a role on the relative stability of MOFs, in this work, we set out to evaluate the interactions of solvent molecules with a database of ~5,300 MOFs both through the calculation of their heat of adsorption and the MOF free energy of solvation. This MOF database ensures the presence of polymorphic families where polymorph selection was not explained by free energy of the MOF structure alone, while also featuring 400 topologically diverse “parent” MOFs and their functionalized versions (-OH, -SH, -F, -Br, -CN, -CF3, -CH₃, -NO₂, -NH₂), to allow us study the impact of functionalization on overall synthetic likelihood and polymorph selection. The studied solvents were DMF, methanol, and water (common solvents used in MOF synthesis), and n-hexane (as a control solvent).

In this presentation, we first discuss the fine methodological details of establishing a molecular simulation pipeline to calculate free energy of solvation in high throughput fashion, and the suitability of two common free energy methods (Finite Differences Thermodynamic Integration (FDTI) and Free Energy Perturbation (FEP) to calculate this quantity. Then, we discuss the effect of MOF functionalization on absolute free energy (and thus the effect of functionalization on overall synthetic likelihood) and free energy across polymorphic families (and thus the effect on polymorph selection). We follow with a discussion on structure-property relationships revealing the nature of solvent-MOF interactions with respect to MOF structure (for instance, we find that compared to other solvents, DMF enables stronger MOF-solvent interactions relative to solvent-solvent interactions than H₂O, MeOH and n-hexane). We also used these solvent heat of adsorption calculations to select a subset of MOFs (and thus polymorphic families) on which to perform the more intensive free energy of solvation calculations. We finish our presentation with a discussion of how MOF structure influences free energy of solvation, to what extent this quantity can alter polymorph selection, and what level of MOF-solvent interaction is needed to alter such selection.