Small-Molecule Adsorption in Open-Site Metal–Organic Frameworks: A Systematic Density Functional Theory Study for Rational Design

Abstract

Using density functional theory, we systematically compute and investigate the binding enthalpies of 14 different small molecules in a series of isostructural metal–organic frameworks, M-MOF-74, with M = Mg, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn. The small molecules we consider include major flue-gas components, trace gases, and small hydrocarbons, i.e., H2, CO, CO2, H2O, H2S, N2, NH3, SO2, CH4, C2H2, C2H4, C2H6, C3H6, and C3H8. In total, the adsorption energetics of 140 unique systems are presented and discussed. Dispersion interactions are included by employing a nonlocal van der Waals density functional, vdW-DF2. Hubbard U corrections are applied to the localized d electrons of transition metal atoms, and the impact of such corrections is assessed quantitatively. For systems for which measured binding enthalpies have been reported, our calculations lead to excellent overall agreement with experimentally determined structures and isosteric heats of adsorption. For systems that have yet to be realized or characterized, this study provides quantitative predictions, establishes a better understanding of the role of different transition-metal cations in small-molecule binding at open-metal sites, and identifies routes for predicting potential candidates for energy-related gas-separation applications. For example, we predict that Cu-MOF-74 will exhibit selectivity of CO2 over H2O and that Mn-MOF-74 can be used to separate trace flue-gas impurities and toxic gases from gas mixtures.

Experimentalists and theorists need to talk

“…Each type of imperfection, as well as the application the material is being used for, needs to be modelled using a different approach. This complicates computational screening greatly and means that models cannot truly capture the properties of a real material or predict its properties accurately 8 .…”

Understanding Small‐Molecule Interactions in Metal–Organic Frameworks: Coupling Experiment with Theory

“…[ 58,59 ] While it was hypothesized from high initial isosteric heats (−47 kJ mol −1 ), [ 40 ] derived from gas-adsorption measurements, that CO 2 molecules preferentially bind at the open metal site, other methods such as diffraction, IR and Raman spectroscopy, and density functional theory (DFT) have been used to afford direct evidence of the location and orientation of CO 2 molecules binding within the pore. [60][61][62][63][64][65] From neutron powder diffraction (NPD) data, it has been found that CO 2 molecules bind in an "end-on" orientation with Mg-O(CO 2 ) distances and angles that range from 2.24 to 2.39 Å and 125 to 144°, respectively, depending on the CO 2 loading level. [ 60 ] These results agree well with DFT-derived Mg-O(CO 2 ) distances and angles computed at the B3LYP-D level to be 2.31 Å and 129°, respectively.…”

“…[ 65,76,78,79 ] Geier et al [ 79 ] demonstrated from adsorption-isotherm data and breakthrough experiments collected on the Mg-, Mn-, Fe-, Co-, Ni-, and Zncontaining analogs that the highest achievable separation selectivity for ethane/ethylene and propane/propylene could be realized with the Fe 2+ and Mn 2+ analogs, respectively (from 318 to 358 K). In a recent study, Lee et al [ 65 ] utilized vdW-DF2 with Hubbard U corrections to assess 140 unique systems; they studied 10 metal-substituted M 2 (dobdc) analogs, both hypothetical and known, and their interactions with 14 different small molecules, including C1-C3 hydrocarbons. Compared with experimental results, the theoretically predicted binding geometries and enthalpies indicated good agreement across all the hydrocarbon systems studied, with the exception of C 3 H 8 , which has more internal degrees of freedom relative to other small molecules, making it diffi cult to resolve the global minimum.…”

“…For example, the M 2 (dobdc) series (M = Mg, Mn, Fe, Co, Ni, Cu, and Zn) optimized with PBE +U resulted in mean absolute errors (MAEs) in the metal-oxygen bonds ranging from 0.9-2.1% and lattice constants in good agreement with experiment. [ 65 ] Finally, we should note that the current efforts in generating hypothetical structures have many limitations. For example, trade-offs in computational costs between evaluating structures for synthetic accessibility and exploring additional degrees of structural freedom make it diffi cult to develop a high-throughput algorithm that still yields an accurate, yet exhaustive list of predicted structures.…”

“…While these QM methods have not been used in a truly high-throughput manner for screening materials, they have been applied to families of materials to understand, for example, how metal substitution affects adsorption properties. [ 65 ] For molecular simulations, a classical force fi eld that properly describes the host-guest and guest-guest interactions must be chosen. For other classes of porous materials (e.g., zeolites and the isoreticular MOF series), well-established force fi elds in the literature have been shown to perform well.…”