ReportsChemistry

Ordered macro-microporous metal-organic framework single crystals

See allHide authors and affiliations

Science  12 Jan 2018:
Vol. 359, Issue 6372, pp. 206-210
DOI: 10.1126/science.aao3403

Mesoporous metal-organic frameworks

The diffusion limitations on gas storage and catalytic reaction of microporous materials can often be overcome if they are incorporated into a mesoporous structure with much larger pores. Shen et al. grew ordered arrays of microcrystals of the ZIF-8 metal-organic framework, in which zinc ions are bridged by 2-methylimidazole linkers, inside a porous polystyrene template. These materials showed higher reaction rates for the Knoevenagel reaction between benzaldehydes and malononitriles and better catalyst recyclability.

Science, this issue p. 206

Abstract

We constructed highly oriented and ordered macropores within metal-organic framework (MOF) single crystals, opening up the area of three-dimensional–ordered macro-microporous materials (that is, materials containing both macro- and micropores) in single-crystalline form. Our methodology relies on the strong shaping effects of a polystyrene nanosphere monolith template and a double-solvent–induced heterogeneous nucleation approach. This process synergistically enabled the in situ growth of MOFs within ordered voids, rendering a single crystal with oriented and ordered macro-microporous structure. The improved mass diffusion properties of such hierarchical frameworks, together with their robust single-crystalline nature, endow them with superior catalytic activity and recyclability for bulky-molecule reactions, as compared with conventional, polycrystalline hollow, and disordered macroporous ZIF-8.

Porous materials have attracted considerable scientific attention because of their wide industrial applications ranging from gas separation to catalysis (1). Topological features—particularly pore sizes at each dimension and the uniformity of these pores—in porous materials are key factors that determine their preferred function in a particular application (2, 3). The construction of hierarchical porous structures that maintain their overall crystalline order is desirable because the high structural regularity may, in turn, afford markedly improved properties. A few advantageous examples include substantially improved conductivities and electron mobilities for mesoporous TiO2 single crystals (4) as compared with nanocrystalline TiO2 and much stronger framework acidities and stabilities for mesoporous crystalline zeolites with respect to amorphous molecular sieves (5, 6).

The synthesis of highly ordered meso- or macroporous crystalline materials at a wide range of length scales with different structural coherencies remains an important goal (7, 8), particularly with respect to enlarging the pores of metal-organic frameworks (MOFs). Most reported MOFs exhibit porosities restricted to the microporous regime, which limits their applications to diffusion-limited processes (9, 10). Synthetic strategies to generate controllable porosities within MOFs mainly include templating, etching, and template-free and defect-induced methods (1114). However, these approaches cannot generate ordered macro-microporous MOF single crystals. Alternatively, extendable ligands and three-dimensional (3D)–ordered templates have been used to design periodic meso- and macropores (1519). The mesopores built from long ligands often suffer from rapid collapse after guest removal, whereas template-duplicated macropores are generally derived from the assembly of intergrown MOF polycrystals. Until now, MOFs featuring both macropore ordering (pores with diameter > 50 nm) and single crystallinity have not been achieved. On the basis of these premises, the development of reliable methods to design ordered meso- or macroporous MOF single crystals with robust framework structures is imperative.

We report a single-crystalline MOF with 3D ordering of macro-micropores, and we used the ZIF-8 structure as a proof of concept. This structure’s 3D framework is built up from connecting ZnII ions through 2-methylimidazole linkers, showing a periodically microporous topology featuring large cavities (11.6 Å) and small apertures (3.4 Å) (20, 21). First, monodisperse polystyrene spheres (PSs) were assembled into highly ordered 3D PS monoliths (Fig. 1A and figs. S1 and S2). ZIF-8 precursors—that is, 2-methylimidazole and Zn(NO3)2—were then filled into the PS monolith interstices to form “precursor@PS” monoliths, which were subsequently soaked in CH3OH/NH3·H2O mixed solutions. We chose the mixed solutions as the solvent, with the idea that NH3·H2O can induce rapid crystallization of the precursors, whereas CH3OH can effectively stabilize such precursors and adjust the balance between nucleation versus growth of ZIF-8. As a result, the precursors smoothly turned into oriented and 3D-ordered ZIF-8 single crystals (fig. S3) (22). Obviously, the proposed double-solvent–assisted method could overcome the energetic barrier to homogeneous nucleation, which is usually unavoidable in conventional nanocasting processes (23, 24). This new type of macro–microporous MOF was denoted as SOM-ZIF-8(x, y) (SOM stands for single-crystal ordered macropore), in which x and y represented the feeding molar ratios of 2-methylimidazole/Zn(NO3)2 and the volume fraction of NH3·H2O in the solvent used for synthesis, respectively.

Fig. 1 In situ nanocasting synthesis of SOM-ZIF-8 and its structure confirmation.

(A) Schematic diagram of our strategy to design SOM-ZIF-8. THF, tetrahydrofuran. (B) Representative SEM image of SOM-ZIF-8(3, 0.5). (C) SEM images of individual crystals taken from four different directions. (D) SEM image of the ZIF-8 grown on the external surface of the PS template. (E and F) SEM images of two selected semifinished SOM-ZIF-8(3, 0.5) crystals, confirming their oriented growth within ordered voids. Yellow areas, {100} planes of macropore arrangements; red areas, {111} planes of macropore arrangements. (G) Illustration of the shaping and templating effects of the 3D-ordered PS template on the morphologic evolution of SOM-ZIF-8.

Low-resolution scanning electron microscopy (SEM) images (Fig. 1B and fig. S4) revealed that SOM-ZIF-8(3, 0.5) displayed a tetrakaidecahedron morphology, although some crystals looked different as various directions were viewed. The oriented arrangement of 3D-ordered macropores could be identified in the representative SEM images of individual crystals taken from four different directions (Fig. 1C). On inspection, we identified that the ordered arrangement of macropores on the surface of six square planes and eight triangle planes inherited {100} and {111} planes of the colloidal PS template, respectively.

The crystal growth of ZIF-8 is known to start with the formation of cubes composed of 6 (100) facets, which can subsequently grow into truncated rhombic dodecahedra composed of 12 (110) and 6 (100) facets (25, 26). In our study, ZIF-8 growing on the external surface of a PS template also displayed a truncated rhombic dodecahedra morphology with 6 (100) and 12 (110) well-developed facets (Fig. 1D and fig. S5). Given that the difference between the two crystal morphologies was the existence of the PS template, we assumed that the PS template was the controlling agent that allowed for the preferential formation of tetrakaidecahedron SOM-ZIF-8 with oriented macropore arrangement.

To validate this assumption, we conducted a controlled experiment in which the precursor@PS monolith was first broken down to very small particles before soaking with CH3OH/NH3·H2O for further crystallization (under otherwise identical synthetic conditions). As expected, almost all of the obtained crystals also exhibited an obvious truncated rhombic dodecahedra morphology (fig. S6). A closer inspection revealed that the (100) facets of many poorly developed crystals also inherited {100} planes of the colloidal PS template. A selected crystal (Fig. 1E) provided direct insights into the effects of ordered PSs as template on the morphological features of SOM-ZIF-8(3, 0.5). In one part, the 3D PS template guided the crystal growth into a macroporous tetrakaidecahedron with its (100) and (111) facets matching well with the {100} and {111} planes of macropore arrangements, respectively. However, the crystal had grown into conventional macropore-free truncated rhombic dodecahedra in the remaining part, owing to the absence of PSs. This oriented growth model was further confirmed by a broken crystal composed of an internal oriented 3D-ordered macroporous system and an external compact faceted surface, consistent with conventional ZIF-8 morphology (Fig. 1F). Thus, the shaping and templating effects of the 3D-ordered PS template could directly guide the morphological evolution of SOM-ZIF-8 into a tetrakaidecahedron morphology and match its crystal face to the oriented 3D-ordered macroporous arrangement (Fig. 1G).

X-ray diffraction (XRD) patterns of SOM-ZIF-8(3, 0.5) are compared with conventional (denoted C-ZIF-8, fig. S7) and simulated ZIF-8 in Fig. 2A. SOM-ZIF-8(3, 0.5) exhibited only XRD reflections ascribed to ZIF-8 (20, 21), confirming the formation of phase-pure ZIF-8 with good crystallinity. Nitrogen-adsorption experiments were further performed to investigate the porosity in the material (Fig. 2B). Both SOM-ZIF-8(3, 0.5) and C-ZIF-8 showed similar type I isotherms with a high nitrogen adsorption capacity at very low relative pressures, indicative of their microporous structures (20, 21), as also confirmed by the corresponding micropore size distribution curves (Fig. 2B, inset). The Brunauer-Emmett-Teller surface area and micropore volume of SOM-ZIF-8(3, 0.5) were 1540 m2/g and 0.59 cm3/g, respectively—values higher than those of C-ZIF-8 (1397 m2/g and 0.55 cm3/g). The formation of a 3D-ordered macroporous system did not influence its inherent microporous structure. The presence of macropores in SOM-ZIF-8(3, 0.5) was fully confirmed by means of mercury intrusion porosimetry. As shown in fig. S8, SOM-ZIF-8(3, 0.5) displays a regular macroporous distribution with an average size of ~80 nm, which could be assigned to the throat size of “ink-bottle” pores between neighboring macropores.

Fig. 2 Structure characterization of SOM-ZIF-8.

(A) XRD patterns and (B) nitrogen adsorption-desorption isotherms of various samples. The inset of (B) shows the corresponding micropore size distribution from the Horvath-Kawazoe model, where SOM-ZIF-8(3, 0.5) has been shifted up by 0.3 cm3 g−1 for clarity. a.u., arbitrary units. (C) HAADF-STEM image of an individual SOM-ZIF-8(3, 0.5) crystal. (D to F) TEM images of SOM-ZIF-8 at different magnifications taken along the <011> zone axis. The inset of (D) shows the corresponding SAED patterns, and the inset of (E) shows the indexed FT patterns. (E) and (F) are magnified views of the areas outlined in (D) and (E), respectively.

SOM-ZIF-8(3, 0.5) was subsequently characterized in detail by transmission electron microscopy (TEM) to further confirm its macroporous structure and composition. A representative high-angle annular dark-field–scanning transmission electron microscopy (HAADF-STEM) image revealed that closely packed ordered PSs were imprinted into entire SOM-ZIF-8(3, 0.5) crystals (Fig. 2C). The average diameter of the interconnected macropores was ~270 nm, similar to that of PS latex, because the ordered macroporous structure was a negative replica of the PS template. The highly ordered macropore arrangement in SOM-ZIF-8(3, 0.5) was clearly evidenced from low-magnification TEM images of different directions (fig. S9). Selected-area electron diffraction (SAED) patterns taken from an entire crystal (Fig. 2D, inset) revealed the single-crystalline nature of SOM-ZIF-8(3, 0.5). High-resolution TEM (HRTEM) images at different magnifications along the <011> zone axis showed uniform lattice fringes with consistent orientations over the entire image region, without domain boundaries or interfaces observed (Fig. 2, D to F). Fourier transform (FT) patterns derived from HRTEM further confirmed the single crystalline nature of this material (Fig. 2E, inset). The elemental mapping images (fig. S10) pointed to a homogeneous distribution of elements (including C, N, and Zn) in whole SOM-ZIF-8(3, 0.5) crystals.

The prerequisite to achieve uniform SOM-ZIF-8 materials was the use of methanol/ammonia water mixtures as suitable solvents. Methanol is largely employed in the preparation of uniform ZIF-8 materials but generally does not provide high synthetic yields (fig. S11). Pure ammonia water could effectively increase the yield of ZIF-8, but with too high of a crystallization rate to prepare uniform ZIF-8 crystals (fig. S11). We successfully realized the controlled crystallization of SOM-ZIF-8(3, 0.5) within template-confined spaces by combining the advantages of the two types of solvents, as stated above. In the absence of NH3·H2O, all ZIF-8 crystals nucleated and grew outside of the PS template, producing tiny solid ZIF-8 nanocrystals (Fig. 3A and figs. S12 to S14). Comparably, a homogeneous crystallization of the precursors occurred when using pure NH3·H2O as solvent, affording a monolithic SOM-ZIF-8 (figs. S12, S13, S15, and S16). Uniform SOM-ZIF-8 with single-crystalline nature could be efficiently obtained within a wide volume fraction of NH3·H2O in the solvent (Fig. 3, B and C, and figs. S17 and S18). The particle size grew from ~1.3 to ~4.3 μm when the volume fraction of NH3·H2O was increased from 0.33 to 0.67 (fig. S19), which demonstrates that crystal sizes can be flexibly tuned by simply changing the solvent’s NH3·H2O content. Additionally, variations in the feeding molar ratio of 2-methylimidazole/Zn(NO3)2 from 2 to 4 indicated that uniform SOM-ZIF-8 with similar sizes could be prepared in a rather wide composition range (figs. S20 and S21). Lower temperatures appeared to favor the tetrakaidecahedron morphology; syntheses at a slightly higher temperature of 55°C produced near-sphere SOM-ZIF-8(3, 0.5) with larger particles (3.7 μm) (fig. S22). More importantly, a series of SOM-ZIF-8 with macropore sizes ranging from ~190 to ~470 nm could be prepared by carefully controlling the diameter of PS templates, indicating the general applicability of the proposed strategy (Fig. 3, D to K, figs. S23 to S27, and table S1).

Fig. 3 Controlled growth of SOM-ZIF-8 by regulating solvents and PS sizes.

(A to C) SEM images of various samples prepared with different volume fractions of NH3·H2O: (A) 0, (B) 0.33, and (C) 0.67. (D to K) HAADF-STEM images of various SOM-ZIF-8 crystals with macropore sizes ranging from 190 to 470 nm, as determined by carefully controlling the diameter of the PS templates: (D and H) 190 nm, (E and I) 340 nm, (F and J) 400 nm, and (G and K) 470 nm. The images in (H) to (K) are magnified views of the areas outlined by the squares in (D) to (G).

The catalytic properties of SOM-ZIF-8 were investigated and compared to those of bulk C-ZIF-8, polycrystal hollow ZIF-8 (denoted as PH-ZIF-8, figs. S28 and S29), and macroporous ZIF-8 synthesized by disordered PSs as the template (denoted as M-ZIF-8, fig. S30) in the Knoevenagel reaction between benzaldehydes and malononitriles as a model (27). As expected, SOM-ZIF-8 exhibited substantially improved activities with respect to C-ZIF-8, as the conversion of benzaldehyde could be completed after 2 hours for the optimal SOM-ZIF-8(3, 0.33) versus 8 hours for C-ZIF-8 (Fig. 4A). SOM-ZIF-8(3, 0.33) also outperformed various state-of-the-art heterogeneous catalysts, including amino-functionalized molecular sieves and UiO-66-NH2, under identical reaction conditions (figs. S31 to S37). Despite featuring similar macropore sizes, SOM-ZIF-8 exhibited much improved catalytic activities as compared with PH-ZIF-8 and M-ZIF-8. The interconnected macropores in SOM-ZIF-8 can be entered from the external crystalline surface, facilitating the accessibility to reactive sites. In contrast, macropores in PH-ZIF-8 and M-ZIF-8 were completely or partly occluded in the microporous matrix and thus were much less accessible to bulky reactants. To further support these hypotheses, the diffusion process of green fluorescent protein (GFP) from the outside to the interior of individual SOM-ZIF-8, M-ZIF-8, and C-ZIF-8 crystals was subsequently monitored in situ (fig. S38). Compared with M-ZIF-8 and C-ZIF-8, SOM-ZIF-8 clearly allowed a much faster GFP diffusion throughout the entire crystal, confirming the enhanced diffusion efficiency of 3D-ordered macropores in SOM-ZIF-8 for bulky-molecule–involved applications. Furthermore, the reaction time for complete conversion of benzaldehyde was consistently 2 to 3 hours when the average particle size was decreased from 4.3 μm for SOM-ZIF-8(3, 0.67) to 1.3 μm for SOM-ZIF-8(3, 0.33).

Fig. 4 Comparing the catalytic performance of SOM-ZIF-8 with C-ZIF-8, PH-ZIF-8, and M-ZIF-8.

(A) Benzaldehyde conversions over various samples as a function of reaction time. (B) Recyclability tests of SOM-ZIF-8(3, 0.33) and PH-ZIF-8.

Additionally, SOM-ZIF-8 exhibited superior structural stability and improved recyclability as compared with PH-ZIF-8 composed of intergrown polycrystals. SOM-ZIF-8(3, 0.33) could be reused at least seven times (Fig. 4B), with benzaldehyde conversion only decreasing from 94.6 to 87.0%. Comparably, a ≥5% decrease in catalytic activity was observed in every subsequent run for PH-ZIF-8. The TEM image of used PH-ZIF-8 indicates that the recycling operation may cause a severe collapse in its hollow structure, resulting in utrasmall ZIF-8 nanocrystals that are lost in the separation step (fig. S39). In sharp contrast, SOM-ZIF-8 experienced much less structural damage, with its 3D-ordered macro-microporous structure being well preserved after seven reaction cycles. The durability of SOM-ZIF-8 can be attributed to its robust single-crystalline framework with much lower defect density in comparison with PH-ZIF-8 (fig. S40).

The reaction of various substituted benzaldehydes with malononitrile was also performed with SOM-ZIF-8(3, 0.33) and C-ZIF-8 (table S2). For all investigated substrates, SOM-ZIF-8(3, 0.33) always showed much higher catalytic activities than C-ZIF-8 under identical reaction conditions. The activity enhancement was even notable for benzaldehydes with bulky groups (table S2, entries 11 to 16). In view of the improved catalytic efficiencies in the Knoevenagel reaction, SOM-ZIF-8 could be expected to show additionally enhanced catalytic activities in various bulky-molecule reactions, which cannot be promoted or achieved using purely microporous C-ZIF-8.

Supplementary Materials

www.sciencemag.org/content/359/6372/206/suppl/DC1

Materials and Methods

Figs. S1 to S40

Tables S1 and S2

MS and NMR Spectra

References (2832)

References and Notes

  1. Supplementary materials are available online.
Acknowledgments: We thank the National Natural Science Foundation of China (grants 21606087, 21436005, 51621001, 21576095, and 21606088), the Fundamental Research Funds for the Central Universities (grants 2017PY004 and 2017MS069), the Guangdong Natural Science Foundation (grants 2014A030310445, 2016A050502004, and 2017A030312005), and the Welch Foundation (grant AX-1730) for financial support. K.S. thanks D. Tong for help with the schematic design and Y. Zhao for GFP preparation. K.S., Y.L., and B.C. conceived the idea and designed the experiments. K.S. synthesized the materials and carried out most of the structural characterization. Y.H., L.L., and D.Z. performed the HRTEM images and analyzed the results. K.S., L.Z., X.C., J.C., and J.L. performed the catalytic tests and some structural characterization. K.S., Y.L., B.C., and R.L. cowrote the paper. All authors discussed and commented on the manuscript. All data are reported in the main text and supplementary materials. K.S. and Y.L. are inventors on a patent/patent application (201710819873.9) held/submitted by South China University of Technology that covers the method for preparing ordered macroporous MOF single crystals.
View Abstract

Navigate This Article