Towards Polymer‐Based Hydrogen Storage Materials: Engineering Ultramicroporous Cavities within Polymers of Intrinsic Microporosity

Three structurally diverse polymers of intrinsic microporosity reversibly adsorb significant quantities of hydrogen (1.4–1.7 % by mass at 77 K) and represent the first examples of a new type of purely organic hydrogen storage material, which can be tailored to meet the specific requirements of hydrogen physisorption. A major technical obstacle to the widespread use of hydrogen (H2) as a nonpolluting fuel for cars is the lack of a safe and efficient system for on-board storage.1 Of the many potential solutions being investigated,2 an attractive possibility is a system based on the reversible adsorption of H2 on the internal surface of a microporous material such as a zeolite,3 carbon,4 or metal-organic framework (MOF).5, 6 At present, the quantity of H2 that can be adsorbed onto any type of microporous material falls below the requirements of a practical H2 storage system. Hence, there is an urgency to develop materials which can be tailored to provide a structure7, 8 and chemical composition9 suitable for the specific demands of H2 physisorption. Previously, organic polymers have not been considered as materials for the storage of hydrogen because polymers generally have enough conformational and rotational freedom to pack space efficiently and thus do not offer high surface areas. However, the recently reported polymers of intrinsic microporosity (PIMs) are composed wholly of fused-ring subunits designed to provide highly rigid and contorted macromolecular structures that pack space inefficiently. Hence they form solids with large amounts of interconnected free volume, providing accessible internal surface areas in the range 500–900 m2 g−1.10, 11 PIMs are prepared by using a benzodioxane formation reaction between suitable monomers, one of which must contain a site of contortion such as a spiro-center or a rigid nonplanar unit. A PIM can be prepared either as an insoluble network or as a soluble polymer, suitable for solution-based processing, depending upon the number of catechol and aromatic ortho-dihalide groups possessed by the monomers. For example, a soluble PIM (PIM-1) is prepared from the reaction between 5,5′,6,6′-tetrahydroxy-3,3,3′,3′-tetramethyl-1,1′-spirobisindane and tetrafluoroterephthalonitrile, whereas, a network-PIM (HATN-network-PIM) is formed from the reaction between the same readily available spiro-cyclic bis(catechol) monomer and hexachlorohexaazatrinaphthylene (Scheme 1).11, 12 PIM-1 can be cast from solution to form robust microporous self-standing films that show excellent promise for gas13 and solution phase membrane separations.11 The HATN-network-PIM has in-built ligands for metal complexation and the PdII-loaded material is proving an excellent heterogeneous catalyst for Suzuki aryl–aryl coupling reactions.12 Structure of the PIMs used in this study: a) PIM-1, b) HATN-network-PIM, and c) CTC-network-PIM. In the context of hydrogen adsorption, we believe that PIMs may offer an attractive combination of properties including low intrinsic density (they are composed of only light elements—C, H, N, O—a real advantage over MOF materials),14 chemical homogeneity (an advantage over carbons), thermal and chemical stability, and synthetic reproducibility. Of particular interest is the potential to tailor the micropore structure by choice of monomer precursors, for example, by the use of monomers that contain pre-formed cavities to provide sites of an appropriately small size for hydrogen adsorption. To investigate this possibility, the bowl-shaped receptor monomer, cyclotricatechylene (CTC),15 was incorporated within a network-PIM by using the benzodioxane-forming reaction between CTC and tetrafluoroterephthalonitrile. The gas-adsorption properties of the resulting material, designated CTC-network-PIM (Scheme 1), were compared to those of the previously prepared PIM-1 and HATN-network-PIM. For each material used in the study, the structure was confirmed by elemental analysis, IR spectroscopy, and, in the case of PIM-1, solution 1H and 13C NMR spectroscopy. The weight-average molar mass of the PIM-1 sample was estimated as 230 000 g mol−1 by using gel permeation chromatography as calibrated against polystyrene standards. The BET surface area of each PIM is in the region of 800 m2 g−1 (Table 1) as measured by nitrogen adsorption at 77 K (Figure 1 a). Analysis of the low-pressure nitrogen adsorption data by the Horvath–Kawazoe method16 indicates that in each case the pore size distribution is strongly biased towards pores in the range 0.6–0.7 nm (Figure 1 b)—that is, they are predominantly ultramicroporous.17 The particularly marked concentration of micropores of 0.6 nm diameter for the CTC-network-PIM appears related to the internal dimensions of the bowl-shaped CTC subunit and suggests that pore size distribution within PIMs can be tuned by the choice of monomer precursor. a) Nitrogen adsorption (filled symbols) and desorption isotherms (open symbols) at 77 K; b) apparent micropore size distributions (dV/dw; the unit is mL g−1 nm−1) calculated by the Horvath–Kawazoe method (carbon slit-pore model) for PIM-1 (○), HATN-network-PIM (▵), and CTC-network-PIM (□). PIM BET area m2 g−1a % mass H2 (1 bar, 77 K)b % mass H2 (10 bar, 77 K)c H2 per fused ringc PIM-1 760 1.04 1.44 0.5 HATN 820 1.37 1.56 0.43 CTC 830 1.43 1.70 0.56 The measurement of H2 sorption is technically demanding and can be prone to errors.18 The classical volumetric method can give false high values due to H2 leaks, whereas gravimetric analysis may exaggerate H2 adsorption greatly due to contamination with molecules of greater mass (e.g. H2O). Hence, complementary sorption isotherms for each of the PIMs were obtained for ultrapure, dry H2 at 77 K by using both gravimetric analysis (Hiden IGA-1), over the pressure range 0 to 20 bar, and volumetric analysis (Micromeritics ASAP 2020), over the range 0 to 1 bar. For the gravimetric analysis, the buoyancy correction required to take into account the mass of hydrogen displaced by the sample at a given pressure was calculated using a density of 1.4 g mL−1, which was based upon helium pyconometry measurements. For both methods of analysis, consistent results were obtained as shown in Figure 2 and listed in Table 1. In addition, for both types of measurement the adsorption is rapid and completely reversible, which is consistent behavior with that expected for the physisorption of H2 on a microporous material. The isotherms show that the three PIMs each adsorb significant quantities of H2 (maximum 1.4–1.7 % by mass) at relatively low pressures with saturation being reached at less than 10 bar pressure and with most of the adsorption taking place below 1 bar. The H2 adsorption (filled symbols) and desorption isotherms (open symbols) at 77 K for PIM-1 (○), HATN-network-PIM (▵) and CTC-network PIM (□), obtained using a) volumetric and b) gravimetric analysis. It is of interest that the order of PIMs based on the efficiency of H2 uptake at 77 K and 1 bar is: CTC-network > HATN-network > PIM-1, which is the reverse of that for nitrogen adsorption under the same conditions (Figure 1 a). Calculations show that the maximum number of hydrogen molecules adsorbed per fused-ring of the polymeric repeat unit (Scheme 1) is approximately 0.5 H2 per ring. However, the spread of values from 0.43 (HATN-network), 0.5 (PIM-1) to 0.56 (CTC-network) suggests that the greater predominance of ultramicropores, resulting from the bowl-shaped CTC subunit, enhances H2 adsorption. The results show that these PIMs adsorb comparable amounts of H2 to that of the best examples of zeolites3 and MOFs5, 19 and, although their performance falls short of some very high surface area carbons, they adsorb similar quantities of H2 to carbons of similar surface area.4, 7 However, to attain practical hydrogen storage materials from PIMs, it will be necessary to engineer examples with larger accessible surface areas (>2000 m2 g−1) whilst maintaining a predominately ultramicroporous structure to retain the beneficial multi-wall interactions with H2 molecules. This is a similar goal to that of many researchers investigating nanoporous carbons7 and MOFs.8 For PIMs, microporosity results from the association of the macromolecules during precipitation from solvent. Therefore, control over the resulting pore structure will be achieved by the design of the macromolecular structure by choice of appropriate monomer precursors; for example, by the addition of larger substituents to frustrate polymer chain packing further, by the optimization of the distance between the sites of contortion (e.g. spiro-centers) or by the use of monomers with predefined cavities as illustrated by the CTC-network-PIM. In addition, unlike all other types of microporous materials, the structures of PIMs are not necessarily constrained by a fixed network structure and may be dissolved in suitable solvents and swollen by suitable non-solvents. This will allow established methods of polymer structural manipulation to be applied to help optimize H2 adsorption; for example, the rapid removal of solvent from a swollen PIM could generate additional free volume and thus provide greater accessible surface area.20 With many synthetic and processing aspects to explore, we believe that the optimization of PIMs for H2 adsorption presents additional opportunities towards achieving the US Department of Energy's ambitious 2010 target of a practical storage system that holds 6.0 % H2 by mass.21 The preparations of PIM-111 and HATN-network-PIM12 were as previously described. CTC-network-PIM: To a stirred solution of tetrafluoroterephthalonitrile (0.47 g, 2.349 mmol) and cyclotricatechylene (0.574 g, 1.566 mmol) in dry DMF (50 mL) was added finely ground anhydrous potassium carbonate (2.6 g, 18.84 mmol). The reaction mixture was then heated at 120 °C for 12 h under nitrogen. On cooling, the reaction mixture was added to 300 mL of stirred distilled water and the solid product collected by filtration and washed with water and MeOH. Purification was achieved by refluxing the product with MeOH, acetone, and THF. The yellow network polymer was then ground and dried in a vacuum oven at 120 °C (0.8 g, 93 % yield). IR (KBr cm−1): ν=3060, 3010, 2956, 2870 (C-H), 2235 (CN), 1610 (C-CAr), 1170, 1250 cm−1 (C-O). Elemental analysis (%) found: C 61.64, H 3.78, N 7.29, F 0.91 %; ideal polymer structure C33H12N3O6 requires C 72.53, H 2.21, N 7.69, F 0.00 %). TGA shows loss of approximately 5 % water below 100 °C with significant ash residue obtained during combustion analysis. Surface area (BET)=830 m2 g−1. Gas adsorption studies: Volumetric N2 and H2 sorption studies were undertaken using a Micromeritics Instrument Corporation (Norcross, Georgia) Accelerated Surface Area and Porosimetry (ASAP) 2020 system. Before sorption analysis, the sample was subjected to the degas vacuum system under ultra high vacuum (10−9 bar) at a temperature of 120 °C overnight. The sample was back-filled with nitrogen and transferred to the analysis system. The sample was then again degassed under ultrahigh vacuum (10−9 bar) at a temperature of 100 °C for a period of at least 6 h, and kept at ultra high vacuum until analysis. Sorption analysis was carried out at liquid-nitrogen temperature (77 K). Helium was used for the freespace determination, after sorption analysis, both at ambient temperature and at 77 K. Apparent surface areas were calculated from N2 adsorption data by multi-point BET analysis. Apparent micropore distributions were calculated from N2 adsorption data by the Horvath–Kawazoe method, assuming a slit-pore geometry and the original H–K carbon–graphite interaction potential. Similar results are obtained using cylinder pore geometry with a Ross–Olivier carbon–graphite interaction potential. Gravimetric H2 sorption studies were undertaken by using a Hiden Isochema (Warrington, England) Intelligent Gravimetric Analyser (IGA), which incorporates a microbalance capable of measuring weights with a resolution of ±0.2 μg, and can operate over a pressure range from ultrahigh vacuum to 20 bar. Before sorption analysis, the sample was degassed under ultrahigh vacuum (10−9 bar) at a temperature of 120 °C for a period of at least 6 h. Sorption and desorption analysis was carried out at liquid nitrogen temperature (77 K). Measured masses were corrected for buoyancy. The density values for buoyancy corrections (ca. 1.4 g mL−1) was obtained by helium pycnometry in the IGA.