Unravelling exceptional acetylene and carbon dioxide adsorption within a tetra-amide functionalized metal-organic framework

Understanding the mechanism of gas-sorbent interactions is of fundamental importance for the design of improved gas storage materials. Here we report the binding domains of carbon dioxide and acetylene in a tetra-amide functionalized metal-organic framework, MFM-188, at crystallographic resolution. Although exhibiting moderate porosity, desolvated MFM-188a exhibits exceptionally high carbon dioxide and acetylene adsorption uptakes with the latter (232 cm3 g−1 at 295 K and 1 bar) being the highest value observed for porous solids under these conditions to the best of our knowledge. Neutron diffraction and inelastic neutron scattering studies enable the direct observation of the role of amide groups in substrate binding, representing an example of probing gas-amide binding interactions by such experiments. This study reveals that the combination of polyamide groups, open metal sites, appropriate pore geometry and cooperative binding between guest molecules is responsible for the high uptakes of acetylene and carbon dioxide in MFM-188a.

T he emergence of metal-organic frameworks (MOFs) as multifunctional materials results from the combination of their high porosity and precise design of pore functionality. This allows fine tuning of pore size and geometry, framework topology and chemical functionality for targeted applications such as gas storage, separation, catalysis, sensing and proton conductivity [1][2][3][4][5][6] . Gaining in-depth understanding and control of the supramolecular host-guest interactions is crucial if practical materials are to be developed 7 . Two strategies have been commonly reported to enhance the host-guest interactions in MOFs: (i) incorporation of open metal sites (OMSs); (ii) functionalization of the organic linkers to provide binding sites for guest species. The favourable role of OMSs in guest binding has been widely confirmed. For example, in situ neutron diffraction experiments for D 2 -loaded NOTT-101 (ref. 8) and , CO 2 -loaded MOF-74Mg (ref. 10), C 2 D 2 -loaded HKUST-1 (ref. 11) and MOF-74Fe (ref. 12), CD 4 -loaded MOF-74Mg (ref. 13) and D 2 , N 2 and O 2 -loaded Cr-BTT 14 (BTT 3 À ¼ ¼ 1,3,5-benzenetristetrazolate) clearly conclude that OMSs are the primary binding domains for guest molecules. This observation has been further evidenced by neutron spectroscopic experiments, which confirmed the large translational and/or rotational hindrance of bound guest molecules at OMSs 15 . In contrast, experimental investigations of the role of organic functionalization in guest binding in MOFs are scarce [16][17][18] , with the vast majority of reported examples relying predominantly on computational modelling [19][20][21] .
The amide group (-OCNH-) is of particular interest because it offers dual-functionality from both CO-and NH-sites. Notably, the introduction of amide groups in MOFs has been reported to strengthen the host-guest interactions in MOFs 19,22,23 . However, experimental observation of the precise role of amides in guest binding in MOFs is still largely lacking. Here we report the synthesis, crystal structure and gas uptake properties of the tetra-amide tetra-isophthalate Cu II complex MFM-188 (MFM, Manchester Framework Material), which exhibits an exceptionally high CO 2 uptake (120 cm 3 g À 1 at 298 K and 1.0 bar), and significantly, a record high C 2 H 2 uptake (232 cm 3 g À 1 at 295 K and 1.0 bar) among porous MOFs under the same conditions. Gas binding is investigated by a combination of neutron powder diffraction (NPD) and inelastic neutron scattering (INS) for both CO 2 -and C 2 H 2 -loaded MFM-188. These crystallographic and dynamic experiments successfully establish a detailed molecular mechanism for guest binding at a crystallographic resolution.
The structure of MFM-188 can be viewed as 3D alternate packing of three types of metal-ligand cages (denoted as A, B and C) (Fig. 2a). The smallest cage A is comprised of four {Cu 2 } paddle-wheels and two ligands, and has an elongated, distorted octahedral geometry (internal size of B 9.4 Â 9.4 Â 13.4 Å). The {Cu 2 } paddle-wheels occupy the four equatorial vertices with two biphenyl cores from the ligands occupying the apical vertices. Corner-sharing of cages A with biphenyl groups running along the c axis and with paddle-wheels along the a/b axis extends the structure (Fig. 2b). Two further distorted cuboctahedral cages (B and C) result from this assembly, each comprising of eight {Cu 2 } paddle-wheels and four ligands. The length of cage C, defined by the separation between the centroids of the two closest {Cu 2 } paddle-wheels along the c direction, is 17.4 Å. Each Cu II site has a coordinated water molecule at the axial position and points to the centre of Cage C. Cages B and C are connected through shared lozenge-shaped windows on the ab plane and through their apical square-shaped windows along the c direction. Taking the van der Waals radii into account, cages B and C have internal pore size of B 11 Â 11 Â 17 Å and B 17 Â 17 Â 17 Å, respectively. Each cage B and C is surrounded by eight cages A, while each cage A is surrounded by four cages B and four cages C, with cages A, B and C present in a 2:1:1 ratio. The total accessible volume of MFM-188 upon removal of guest solvents is 73% using the PLATON/VOID routine 25 . Within the void space of MFM-188, there is a high concentration of amide (-OCNH-) groups pointing into cage B (O-centres) and into cage C (N-H centres), thus endowing these cages with a combination of open metal sites and multiple functional groups. In contrast, none of the amide groups protrude into cage A, which is functionalized solely with phenyl rings.
Characterization of MFM-188. The purity of bulk MFM-188 material was confirmed by powder X-ray diffraction ( Supplementary Fig. 3) and elemental analysis. Activated MFM-188a was prepared by solvent exchange with methanol and then flowing supercritical CO 2 through the sample for 12 h followed by heating at 100°C under dynamic vacuum for 20 h. N 2 adsorption at 77 K shows a reversible type I isotherm    Supplementary  Fig. 5). The pore size distribution, calculated using a non-local density functional theory model, reveal two broad peaks centred at 12.5 and 14.7 Å, in good agreement with the pore size measured for cages A, B and C. Importantly, the total pore volume measured from the N 2 isotherm (1.12 cm 3 g À 1 ) compares favourably with that (1.07 cm 3 g À 1 ) calculated from the X-ray crystal structure, confirming complete activation of the material. The overall porosity of MFM-188a is moderate in comparison with the most porous MOFs reported to date 26 .
The Q st values at zero coverage determined by virial analysis of adsorption isotherms (Supplementary Figs 6 and 7) are moderate but comparable to typical Cu II MOFs where adsorption of gas molecules occurs primarily on OMSs [21][22][23]28,30 , with Q st (CO 2 ) ¼ 21.0 kJ mol À 1 and Q st (C 2 H 2 ) ¼ 32.5 kJ mol À 1 . Interestingly, the Q st plots do not display any major decrease across the whole adsorption range (Fig. 3b), suggesting the presence of other favourable binding sites in addition to OMSs. We were thus motivated to investigate further the role of amides and OMSs in CO 2 and CH 2 adsorption in MFM-188a.
Determination of adsorption domains for CO 2 and C 2 H 2 . The locations of adsorbed CO 2 and C 2 D 2 molecules within desolvated MFM-188a were determined by in situ NPD as a function of gas loading. NPD patterns were recorded at 7 K for the desolvated material ( Supplementary Fig. 8) and at loadings of 1.75 CO 2 /Cu (127 cm 3 g À 1 ) (Supplementary Fig. 9) and 3.2 C 2 D 2 /Cu (234 cm 3 g À 1 ) ( Supplementary Fig. 10), which correspond to the adsorption uptakes at 1 bar and room temperature for the respective gas species. Fourier difference map analysis of the NPD data of the desolvated MFM-188a indicates no residual nuclear density peak in the pore, thus confirming the complete activation and structural integrity of the desolvated sample (Supplementary Data 2). Loading of CO 2 or C 2 D 2 in MFM-188a was accompanied by significant changes in the relative peak intensities of the diffraction patterns. Successive Fourier difference map analysis followed by Rietveld refinement allowed identification of the position, occupancy and orientation   for adsorbed gas molecules within the framework cages ( Supplementary Data 3 and 4). In both cases, six different sites (1, 2, 3, 4, 5 and 6 in the order of decreasing occupancy) were observed for adsorbed guest molecules: one in cage A, one in cage B and four in cage C, all of them having the same crystallographic multiplicity (Fig. 4). Notably, 83% of loaded CO 2 and 72% of loaded C 2 D 2 molecules were found in cage C, which contains open Cu II sites and the N-H sites of amides pointing to the centre of the cage. The smallest cage A plays a different role in CO 2 and C 2 D 2 uptakes with a minimum contribution to the CO 2 adsorption, but considerable effect to the C 2 D 2 binding.
In the structure of MFM-188Á7CO 2 , the CO 2 molecules at site 1 bind at OMSs in cage C (occupancy ¼ 0.52). The linear body of CO 2 (1) is perpendicular to the Cu-Cu axis (Fig. 5a). The second most populated site 2 (occupancy ¼ 0.42) is also located in cage C where adsorbed CO 2 molecules form H-bonds with both the free amide groups (-NH) and the adjacent isophthalate -CH groups Fig. 5b). CO 2 (3) and CO 2 (4) are also found in cage C (occupancy ¼ 0.25 for both) at the centre of the cage and forming weak supramolecular interaction to isophthalate -CH groups Fig. 5c,d). The remaining CO 2 (5) and CO 2 (6) sites are found in cage B and A (occupancy ¼ 0.19 and 0.11, respectively). Given the long distances between CO 2 (5)/CO 2 (6) and the framework surface (above 5 Å), no specific binding interaction could be identified, thus excluding the presence of dipole interactions between the amide -C ¼ O and adsorbed CO 2 molecules in cage B.
MFM-188 Á 12.8C 2 D 2 exhibits a different distribution of adsorbed C 2 D 2 molecules within the cages, indicating the presence of different gas-sorbent binding mechanisms. The OMSs are fully occupied by C 2 D 2 (1) molecules with a side-on interaction between the CC bond and the Cu II centre, Cu Á Á Á C ¼ 2.60(3), 2.37(3)Å (Fig. 5e), similar to those determined for C 2 D 2 -loaded HKUST-1 (ref. 28). C 2 D 2 (2) and C 2 D 2 (3) are found in cage C and A, respectively, (same site occupancy of 0.67). C 2 D 2 (2) forms H-bonds via its CC(d-) p electrons to the N-H groups pointing into the pore [N Á Á Á C C2H2 ¼ 3.24 (5) (Fig. 5f). In addition, the combination of C 2 D 2 (1) and C 2 D 2 (2) generates another adsorption site (4) with an occupancy of 0.49 in cage C, stabilized via the intermolecular C 2 D 2 Á Á Á C 2 D 2 dipole interactions (Fig. 5e). Specifically, C 2 D 2 (4) is perpendicular to C 2 D 2 (1) and strong intermolecular H-bonding is observed between these two sites [ (4) is reinforced by C 2 D 2 (2) through p Á Á Á p interactions of 3.76(1)Å and a dihedral angle of 84°. C 2 D 2 (5) and C 2 D 2 (6) are found in cage B and C, respectively, (occupancy of 0.22 for both) and without specific binding interaction to the MOF host. Thus, C 2 D 2 displays a dual role behaving as an H-acceptor from amides N-H and aromatic C-H groups, but a H-donor to the carboxylate oxygen, thus enhancing the C 2 D 2 -MOF-binding strength. In contrast, CO 2 participates in H-bond formation as an acceptor only. The highly cooperative binding of acetylene at both the open Cu II sites and free amides, coupled with the well-defined micropore windows, leads to the record high C 2 H 2 adsorption capacity in MFM-188a. Inelastic neutron scattering of gas-sorbent dynamics.
Static crystallographic experiments are unable to gain dynamic insight into the gas-sorbent systems. To directly visualize the binding dynamics of adsorbed CO 2 and C 2 H 2 molecules with accessible functional groups, INS was measured for MFM-188a as a function of CO 2 and C 2 H 2 loading at 11 K (Fig. 6). The INS spectra for the bare MOF show multiple features, fully modelled via DFT calculations (Fig. 6c). The peak at 65 meV corresponds to the out-of-plane wagging modes of the N-H group, and peaks at 84 and 112-156 meV originate from various deformational motions of the phenyl rings and wagging modes of aromatic C-H groups. Comparison of the INS spectra for bare and CO 2 -loaded MFM-188a shows an overall stiffening effect as evidenced by a global shift of peaks to slightly higher energy (Fig. 6a). In addition, a noticeable change in the peak at 65 meV and several small changes to the peaks between 84 and 156 meV were observed, indicating the reduction of the motion of both amide N-H and aromatic C-H groups (Fig. 6d). This suggests the formation of H-bonds between CO 2 molecules and these functional groups, particularly with amide N-H sites. In comparison, the INS spectra of C 2 H 2 -loaded MFM-188a shows significant increase in intensity due to the recoil motion from the H atoms on adsorbed C 2 H 2 (Fig. 6b). Comparison of the difference INS spectra (i.e., signals from adsorbed C 2 H 2 and changes of the local MOF modes) and that of the solid C 2 H 2 shows a number of interesting observations. Firstly, the low-energy INS peaks (below 25 meV, assigned as the translational modes of C 2 H 2 molecules) of the difference spectra shift slightly to the lower energy region but maintain the resolved threefold peak profiles as observed in free solid C 2 H 2 . This indicates that the C 2 H 2 molecules are ordered with restricted translational motions within MFM-188a. Secondly, the INS peaks at 80 and 95 meV (assigned as the acetylene asymmetric and symmetric C-H bending mode, respectively) split from a single-peak profile to a double-peak profile upon adsorption, indicating the presence of adsorbed species resulting from slightly different binding energies to the MOF host (Fig. 6f). This is in excellent agreement with the presence of the strongly bound C 2 H 2 molecules to the OMSs and free amides, and weakly bound C 2 H 2 molecules to the phenyl rings and in the centre of the cages, as observed by NPD. Thirdly and most importantly, the INS peak at 65 meV (assigned as the out-of-plane wagging modes of the N-H group) disappears completely upon inclusion of C 2 H 2 molecules in the pore, indicating loss of this mode and confirming unambiguously the direct formation of strong H-bonds between the free amides (N-H) and C 2 H 2 molecules, as found in the NPD model. This observation confirms the first example of H-bonding between an amide group (N-H) and C 2 H 2 in the solid state. Overall, the INS study is in excellent agreement with NPD results and confirms the crucial role of the free amides in the pore for gas uptakes.

Discussion
MFM-188a shows excellent performance for potential applications for C 2 H 2 storage (27.0 wt%) and exhibits high adsorption uptake of CO 2 (23.7 wt%) under ambient conditions. A 3D plot summarising the relationship between BET surface area, density of 'OMSs þ functional groups' in the structure, and uptake of C 2 H 2 is shown in Fig. 3c for a number of the best MOFs reported to date. It is clear that to achieve high C 2 H 2 uptake capacity, integration of high BET surface area (and porosity to a wider extent) and high density of binding sites in the pore structure are required. However, this is often inherently contradictory since high surface area/porosity will naturally dilute the binding sites in the pore, leading to an inevitable trade-off between these two factors. MFM-188a displays a suitably high BET surface area and high density of binding sites owing to its framework topology and pore geometry, and therefore, shows a record high C 2 H 2 uptake.
Although it is widely believed that the amide groups in pores will actively participate in gas adsorption via H-bonding, no cogent experimental evidence has been reported to date. The present study represents a unique example of a comprehensive investigation of the gas-sorbent binding interaction in a tetraamide functionalized MOF via a combination of neutron diffraction and spectroscopic techniques. These experiments offer  key insights into the molecular details of this host-guest system from both crystallographic and dynamic perspectives. A highly cooperative binding mechanism of CO 2 and C 2 H 2 was found on the OMSs and free amide groups in the pores of MFM-188a, with one site enhancing binding at the other. This work not only offers a new porous material for high-capacity CO 2 and C 2 H 2 storage, but also, more importantly, facilitates the design and structural optimization for new porous materials with improved performance in gas adsorption.

Methods
Materials and equipment. All chemical reagents and gases were obtained from commercial sources and unless otherwise noted were used without further purification. 1 H and 13 C NMR spectra were measured on a Bruker DPX 300, AV400 or AV(III)500 spectrometers. Residual protonated species in the deuterated solvents were used as internal references. Mass spectrometry was performed on a Bruker MicroTOF with the sample dissolved in MeOH or CH 3 CN. Elemental analyses were measured on a CE-440 Elemental Analyzer provided by Departmental Analytical Services at the Universities of Nottingham and Manchester. TGA was performed under a flow of air with a heating rate of 5°C min À 1 using a Perkin Elmer Pyris 1 thermogravimetric analyser. Powder X-ray diffraction measurements were carried out at room temperature on a PANalytical X 0 Pert PRO diffractometer using Cu-Ka radiation (l ¼ 1.5418 Å) at a scan speed of 0.02°s À 1 and a step size of 0.02°in 2y.
Preparation of H 8 L. Preparation of MFM-188. H 8 L (165 mg, 0.168 mmol) and CuCl 2 (0.225 g, 1.675 mmol) were dissolved in dimethyl sulfoxide (8 ml). N,N 0 -diethylformamide (16 ml), EtOH (16 ml) and aqueous HCl (8 ml, 0.1 M) were added to the resulting solution, which was placed in a tightly capped 250 ml Duran pressure plus laboratory bottle. The solution was heated at 80°C in an oven for 96 h, and a large amount of crystalline product precipitated. The blue crystal plates were isolated by filtration while the mother solution was still warm and washed sequentially with warm DMF and MeOH, and stored under MeOH (yield based on TGA of as synthesized material: 0.172 g, 64%, see Supplementary Fig. 11  GV1000 diffractometer (l ¼ 1.54056 Å). Details of data collection and processing procedures are included in the CIF files. Structures were solved by direct methods using SHELXS 31 and remaining atoms were localized from successive difference Fourier maps using SHELXL 32 . The hydrogen atoms from the linkers and the coordinated water molecules were placed geometrically and refined using a riding model. The refinement of the framework was performed by ignoring the contribution of the disordered solvent molecules. The region containing the disordered electron density was identified by considering the van der Waals radii of the atoms constituting the ordered framework. The contribution of this region to the total structure factor was calculated via a discrete Fourier transformation and subtracted in order to generate a new set of hkl reflections by means of the program SQUEEZE 33 . Acquisition and final refinement data are summarized in Supplementary Table 1.
Gas sorption measurements. Volumetric N 2 , CO 2 and C 2 H 4 isotherms for pressures in the range 0-1 bar were determined using a Micromeritics 3-Flex apparatus. Ultra-high purity He (99.999%), for void volume determination, N 2 (99.999%), CO 2 (99.999%) were purchased from BOC and used as received. C 2 H 2 gas was filtered through 4A molecular sieves and activated carbon to remove traces of acetone. Pore size distribution data for MFM-188a was determined by analysing the N 2 isotherm at 77K using a non-local density functional theory based on a carbon model containing cylindrical pores as implemented in the 3-Flex software package. The CO 2 adsorption isotherms at 273 and 298 K as well the C 2 H 2 isotherms at 273 and 295 K were fitted to the virial equation (1). P is the pressure expressed in bars, N is the amount expressed in mol g À 1 , T is the temperature in K, a i and b j are virial coefficients and m, n represent the number of coefficients. The values of the virial coefficients a 0 to a m were then used to calculate the isosteric heat of adsorption using equation (2). Q st is the coverage dependent isosteric heat of adsorption and R is the universal gas constant.
To evaluate the CO 2 /N 2 selectivity of MFM-188a, the N 2 isotherm was also measured at 298 K ( Supplementary Fig. 12) up to 1 bar. By determining the ratios of the Henry's law constants from single-component isotherms, the CO 2 /N 2 adsorption selectivity for MFM-188a is estimated as 22.5 at 298 K. It is worth noting that this selectivity carries large uncertainty because of the very low N 2 uptake measured under these conditions.
Neutron powder diffraction. NPD experiments were carried out at WISH, a long wavelength powder and single crystal neutron diffractometer at the ISIS Facility at the Rutherford Appleton Laboratory (UK) 34 . MFM-188 was loaded into a 6 mm diameter vanadium sample can and outgassed at 1 Â 10 À 7 mbar and 100°C for 1 day. The sample was loaded into a liquid helium cryostat and cooled to 7 K for data collection. CO 2 and C 2 D 2 gas was introduced by warming the samples to 290 K, and the gas dosed volumetrically from a calibrated volume. The gas-loaded sample was then cooled to 7 K over a period of 2 h to ensure good mobility of adsorbed CO 2 /C 2 D 2 within the crystalline structure of MFM-188a. The sample was kept at 7 K for an additional 30 mins before data collection to ensure the thermal equilibrium. The structure solution was initially established by considering the structure of the bare MFM-188a, and the residual nuclear density maps were further developed from subsequent difference Fourier analysis using TOPAS program Acquisition and refinement data are summarized in Supplementary Table 2.
Inelastic neutron scattering. INS spectra were recorded on the TOSCA spectrometer at the ISIS Facility at the Rutherford Appleton Laboratory (UK) for energy transfers between B À 2 and 500 meV. In this region TOSCA has a resolution of 1.5% DE/E. Before INS experiments, MeOH exchanged MFM-188 was activated by flowing supercritical CO 2 for 12 h, the desolvated sample was then transferred in a glovebox under inert atmosphere, and 0.5 g of activated MFM-188a were placed in a closed cylindrical vanadium sample cell with an indium seal. The sample was further degassed at 100°C under dynamic vacuum for 20 h to remove traces of adsorbed guest species. The temperature during data collection was controlled using a closed cycle refrigerator cryostat (11±0.1 K). The loading of 2C 2 H 2 /Cu and 2CO 2 /Cu was performed volumetrically at room temperature to ensure that C 2 H 2 and CO 2 were present in the gas phase when not adsorbed and also to ensure sufficient mobility of C 2 H 2 and CO 2 inside the MFM-188a framework. Subsequently, the temperature was reduced to 7 K for data collection. Background spectra (sample can plus MFM-188a) were subtracted to obtain the difference spectra. INS spectra for condensed C 2 H 2 and CO 2 in the solid state were measured by using a plate sample container. Approximately 1-2 l of each gas at room temperature was condensed slowly at temperatures below their melting points and cooled to temperature below 10 K for neutron scattering measurements.
DFT calculations and modelling of the INS spectra. Periodic density functional theory (periodic-DFT) calculations were carried out using the plane-wave pseudopotential method as implemented in the CASTEP code 35,36 . Exchange and correlation were approximated using the PBE functional 37 . For MFM-188 the structure determined by NPD was used as the initial structure. The plane-wave cutoff energy was 750 eV using on-the-fly generated pseudopotentials and spin polarized calculations were carried out. Owing to the size of the unit cell (12106 Å À 3 ), Brillouin zone sampling of electronic states was performed on 4 Â 4 Â 1 Monkhorst-Pack grid. The equilibrium structure, an essential prerequisite for lattice dynamics calculations was obtained by BFGS geometry optimization after which the residual forces were converged to zero within 0.007 eV Å À 1 . Phonon frequencies were obtained by diagonalization of dynamical matrices computed using density functional perturbation theory 38 . The atomic displacements in each mode that are part of the CASTEP output enable visualization of the modes to aid assignments and are also all that is required to generate the INS spectrum using the program ACLIMAX 39 . It was found that even with these stringent convergence conditions, there were five imaginary modes. It is emphasized that for all the calculated spectra shown the transition energies have not been scaled.
Data availability. The X-ray crystallographic coordinates for structures reported in this Article have been deposited at the Cambridge Crystallographic Data Center (CCDC), under deposition numbers 1496263 (as-synthesized MFM-188), 1504308 (desolvated MFM-188a), 1496264 (C 2 D 2 -loaded MFM-188) and 1496265 (CO 2 -loaded MFM-188). These data can be obtained free of charge from The Cambridge Crystallographic Data Center via www.ccdc.cam.ac.uk/data_request/cif, all other data are available from the corresponding authors upon request.