Microporous membranes with pore apertures below the nanolevel can exhibit size selectivity by serving as a molecular sieve, which is promising for overcoming Robesons "upperbound" limits in membrane-based gas separation. [1] Zeolites, polymers of intrinsic microporosity (PIMs), metal oxides, and active carbon [2a] are the typical materials used for this purpose. Metal-organic frameworks (MOFs) have attracted much research interest in recent years, and are emerging as a new family of molecular sieves. [2b,3-5] MOFs are novel porous crystalline materials consisting of metal ions or clusters interconnected by a variety of organic linkers. In addition to promising applications in adsorptive gas separation and storage or in catalysis, their unique properties, such as their highly diversified structures, large range in pore sizes, very high surface areas, and specific adsorption affinities, make MOFs excellent candidates for use in the construction of molecular sieve membranes with superior performance. [6,7] The preparation of MOF membranes for gas separation is rapidly becoming a research focus. A number of attempts have been made to prepare supported-MOF membranes; [8][9][10][11][12] however, progress is very limited and so far there are only very few reports of continuous MOF films on porous supports being used as separating membranes. Recently, Guo et al. reported a copper-net-supported HKUST-1 (Cu 3 (BTC) 2 ; BTC = benzene-1,3,5-tricarboxylate) membrane exhibiting a H 2 /N 2 selectivity of 7 [13] (separation factor of H 2 over N 2 is calculated as the permeate-to-retentate composition ratio of H 2 , divided by the same ratio for N 2 as proposed by IUPAC [28] ); this is the first MOF membrane to show gasseparation performance beyond Knudsen diffusion behavior.
Very recently, Ranjan and Tsapatsis prepared a microporous metal-organic framework [MMOF, Cu(hfipbb)(H 2 hfipbb) 0.5 ; hfipbb = 4,4'-(hexafluoroisopropylidene)bis(benzoic acid)] membrane by seeded growth on an alumina support. [14] The ideal selectivity for H 2 /N 2 , based on single permeation tests, was 23 at 190 8C. This higher selectivity, compared to the report from Guo et al., might be a result of the smaller effective pore size (ca. 0.32 nm of MMOF versus 0.9 nm of HKUTS-1), [15] which results in a relatively low H 2 permeance of this MMOF membrane (10 À9 mol m À2 s À1 Pa À1 at 190 8C). The authors attributed this finding to the blockage of the onedimensional (1D) straight-pore channels in the membrane. Therefore, with regard to H 2 separation, small-pore MOFs having three-dimensional (3D) channel structures are considered to be ideal membrane materials. Zeolitic imidazolate frameworks (ZIFs), a subfamily of MOFs, consist of transition metals (Zn, Co) and imidazolate linkers which form 3D tetrahedral frameworks and frequently resemble zeolite topologies. [16][17][18] A number of ZIFs exhibit exceptional thermal and chemical stability. [16] Another important feature of ZIFs is their hydrophobic surfaces, which give ZIF membranes certain advantages over zeolite membranes and sol-gel-derived silica membranes in the separation of H 2 in the presence of steam. [19] Very recently we reported the first result from permeation measurements on a ZIF-8 membrane. [20] The ZIF-8 membrane showed a H 2 /CH 4 separation factor greater than 10. Whereas the ZIF-8 pores (0.34 nm) are slightly larger than the kinetic diameter of CO 2 (0.33 nm), and are very flexible, the H 2 /CO 2 separation on this ZIF-8 membrane showed Knudsen selectivity. In the current work, we therefore chose ZIF-7 as a promising candidate for the development of a H 2 -selective membrane to satisfy the above requirements. ZIF-7 (Zn-(bim) 2 ) is formed by bridging benzimidazolate (bim) anions and zinc cations with soladite (SOD) topology. [16,18] The pore size of ZIF-7 (the hexagonal window size in the SOD cage) estimated from crystallographic data is about 0.3 nm, which is just in between the size of H 2 (0.29 nm) and CO 2 (0.33 nm). We could therefore expect a ZIF-7 membrane to achieve a high selectivity of H 2 over CO 2 and other gases through a molecular sieving effect.
In many cases, it was reported that the heterogeneous nucleation density of MOF crystals on ceramic supports is very low, [8,9,14] which makes it extremely difficult to prepare supported-MOF membranes by an in situ synthesis route. Chemical modifications of substrate surfaces have been proposed to direct the nucleation and orientation of the deposited MOF layers. [21,22] Based on our knowledge in the development of zeolite membranes, [23,24] we adopted a seeded secondary growth method for the ZIF-7 membrane prepara-