Exceptional negative thermal expansion in isoreticular metal-organic frameworks.

The thermal-expansion properties of substances are very important in materials design; for example, cracks form when joined materials expand or contract by different amounts upon heating. The most famous example of a substance that contracts when heated is ice: it transforms into water, which has a higher density than ice. Negative thermal expansion (NTE) in solids is relatively rare, although examples have been found in zeolites. The underlying physics of NTE remains poorly understood. Herein, we show, on the basis of molecular simulations, that the recently synthesized isoreticular metal–organic frameworks (IRMOFs) consistently have negative thermal-expansion coefficients and are by far the most contracting materials known. Our simulations point to two competing effects: a local effect, where all bond lengths increase with temperature, and a second long-range effect, where the thermal movement of the linker molecules leads to a shorter average distance between corners upon heating. MOFs are a new class of nanoporous materials that have good stability, large void volumes, and well-defined tailorable cavities of uniform size. Their potential appears great, because these are precisely the properties needed for catalysis, separations, and storage/release applications. MOFs generally consist of metal or metal–oxygen vertices interconnected by rigid or semirigid organic molecules. A large variety of MOFs, featuring different linker molecules and different types of bonding between the vertices with the linkers, have been produced by various research groups. The specific examples shown in Figure 1 are IRMOFs developed by Yaghi and co-workers. In general, the IRMOFs consist of zinc–oxygen complexes connected by carboxylate-terminated linkers, forming a three-dimensional lattice of cubic cavities. Molecular simulations of adsorption in MOFs have shown very good agreement with experiment, and it is interesting to note that simulations of diffusion in MOFs preceded experiments by almost two years. In addition to predicting macroscopic observables, simulations can also provide useful molecular-level insights. To systematically investigate the thermal properties of MOFs, we herein simulate the (cubic) structures of several IRMOFs of varying linker length. We obtain information about the unit-cell length (L) as a function of temperature and about adsorbate loadings (q) as a function of pressure. We show that the experimental data scattered in the literature (for different adsorbate loadings and temperatures) are, in fact, consistent, and we elucidate the different effects of temperature and loading on L at the microscopic level. Various models for MOF flexibility have recently appeared. Our flexible framework model for IRMOF-1, IRMOF-10, and IRMOF-16 is described in the Supporting Information. It is similar in spirit to the model of Greathouse and Allendorf, but differs in the treatment of the carboxylate group and has the advantage of being calibrated to experimental data. It reproduces the experimental unit-cell lengths (L ; Table 1), bond lengths, and bond angles of the frameworks (see the Supporting Information), as well as the adsorption of molecules by the frameworks, not only qualitatively, but also quantitatively (Figure 2). The Zn4O cluster is modeled using only Lennard-Jones and Coulombic potentials between the individual atoms, while the linker molecule is simulated using a combination of general force-field parameters (DREIDING/CVFF parameters) for bond, bend, and (improper) torsion constants. Quantum-mechanical calculations were performed to obtain partial charges for all atoms. The starting parameters for the Zn4O cluster are the parameters of the core-shell models used with considerable success for metal oxides. However, MOFs do not have the local periodicity of metal oxides, and further refinement was, therefore, unavoidable. The fitting process not only involved the unit-cell length, and the bond lengths and angles within the structure, but also the adsorption isotherms of methane and CO2 at several temperatures. All simulations were carried out for one unit cell using high precision for the electrostatics (Ewald summation). Test simulations using 2 E 2E 2 unit cells gave equivalent results, but were deemed too expensive for production runs. [*] Dr. D. Dubbeldam, Prof. R. Q. Snurr Chemical and Biological Engineering Department Northwestern University 2145 Sheridan Road, Evanston, IL 60208 (USA) Fax: (+1)847-467-1018 E-mail: [email protected] Homepage: http://zeolites.cqe.northwestern.edu