Interactions of common synthesis solvents with MOFs studied via free energies of solvation: Implications on stability and polymorph selection

DOI
10.1021/acs.chemmater.5c01410
Graphic
Abstract

Solvents are central to metal–organic framework (MOF) solvothermal synthesis. However, how solvent–MOF interplay impacts MOF stabilization individually and across polymorphs is not well understood. To address this knowledge gap, here we perform data-driven analysis of 20,532 heats of adsorption at dilute conditions (ΔHads) and 447 free energies of solvation (ΔFsol) for four solvents, namely, dimethylformamide (DMF), water (H2O), methanol (MeOH), and n-hexane (C6) (which was used as a control). To accelerate data collection, we developed a protocol to extrapolate ΔFsol from calculations with the MOFs only partially solvated. Free energies were obtained via thermodynamic integration. We found ΔFsol and ΔHads to be only moderately correlated due to solvent–solvent interactions coming into play when the MOF is solvated. In any case, trends in ΔFsol were ultimately explained on the basis of solvent kinetic diameter and polarity, as well as MOF void fraction (Vf) and functionalization polarity. For instance, the correlation between ΔFsol and Vf was one of the strongest correlations presented in this study (more so as the solvent size increases), indicating that small-pore MOFs are more easily stabilized by solvation than large-pore MOFs. We also found that solvation-induced MOF stabilization became more pronounced as solvent kinetic diameter (polarity) decreased (increased). We found differences in this solvation-induced stabilization between polymorphs capable of overcoming inherent (i.e., in vacuum) differences in polymorph stability, causing the most stable polymorph to “switch.” We found the probability to cause “switches” to increase as solvent kinetic diameter (polarity) decreased (increased). Inspection of multivariate linear regression coefficients suggested that differences in solvation-induced stabilization in polymorphs can be primarily explained by their differences in density, Vf, and, to a lesser extent, volumetric surface area.

Themes and Expertise
Porous CrystalsMachine Learning Free Energy Calculations