Amides Do Not Always Work: Observation of Guest Binding in an Amide-Functionalized Porous Metal–Organic Framework

An amide-functionalized metal organic framework (MOF) material, MFM-136, shows a high CO2 uptake of 12.6 mmol g–1 at 20 bar and 298 K. MFM-136 is the first example of an acylamide pyrimidyl isophthalate MOF without open metal sites and, thus, provides a unique platform to study guest binding, particularly the role of free amides. Neutron diffraction reveals that, surprisingly, there is no direct binding between the adsorbed CO2/CH4 molecules and the pendant amide group in the pore. This observation has been confirmed unambiguously by inelastic neutron spectroscopy. This suggests that introduction of functional groups solely may not necessarily induce specific guest–host binding in porous materials, but it is a combination of pore size, geometry, and functional group that leads to enhanced gas adsorption properties.


Experimental Methods and Equipment
4-Carboxyboronic acid 1 was purchased from Frontier Scientific. All other chemicals and reagents used in this experiment were purchased from Fischer Scientific or Sigma Aldrich and used as received without further purification. 1 H NMR and 13 C NMR spectra were measured on a Bruker AV400 spectrometer. High-resolution electrospray mass spectra were measured on a Bruker MicroTOF spectrometer with samples dissolved in MeOH as the solvent. Scanning was done in both positive and negative mode from m/z. Infrared (IR) spectra were recorded in the 400-4000 cm -1 range in ATR sampling mode with a Thermo Scientific iD5 diamond ATR on a Nicolet iS5 FT-IR spectrometer. Elemental analysis of ligands was carried out on a CE-440 elemental analyser (EAI Company). TGA measurements were performed using a Perkin Elmer TGA 7 Gravimetric Analyser under a flow of N 2 (20 ml min 1 ) at a heating rate of 5 °C min -1 . Powder X-ray diffraction (PXRD) patterns were obtained using PANalytical X'Pert Pro MPD diffractometer in Bragg-Brentano geometry with Cu-Kα1 radiation (λ = 1.5406 Å). Samples were evenly dispersed on zero-background silicon plates with a cavity depth of 0.3 mm.

Synthesis of MFM-136
H 2 L (10 mg, 0.03 mmol), Cu(NO 3 ) 2 .3H 2 O (22 mg, 0.09 mmol) and hydrochloric acid (2M, 0.1 cm 3 ) was dissolved in DMF (3 cm 3 ) and heated in a sealed pressure tube at 80 o C for 18 h. The resulting green hexagonal crystal plates were sequentially rinsed with DMF and acetone before being filtered and dried to give MFM-136 (yield 7.0 mg). Elemental analysis calc. for [Cu(C 19 H 11     for Lorentz and polarisation effects using CrysAlisPro; 1 corrections for the effects of adsorption were applied using a numerical absorption correction based on Gaussian integration over a multifaceted crystal model. The structure was solved by direct methods and refined by full-matrix least-squares using the SHELXTL software package. 2 Structure solutions in a variety of space groups including R-3m and R3m were explored; R32 was the space group which presented a sensible solution with the most tractable disorder in the amide portion of the ligand.

Refinement Experimental Details
Phenyl ring C21-C26 and amide oxygen atom O28 were found to be disordered over two orientations. The occupancies of these disorder components were refined before being fixed at values of [0.5]. Geometric similarity restraints were applied to the chemically identical bond distances of the disorder components of the phenyl rings (SADI) and their geometries were restrained to be approximately planar (FLAT). The carbonyl C-O bond distances of S8 the disorder components were restrained to be the same (SADI). Rigid bond and similarity restraints were applied to the anisotropic thermal displacement parameters of the disordered atoms (RIGU and SIMU). Hydrogen atoms were geometrically placed and refined using a riding model. The crystal was found to be an inversion twin; the contributions of the two components were refined and converged to a ratio of 0.52(1):0.48(1). Several disordered solvent molecules could not be sensibly modelled, and so the structure was treated with PLATON SQUEEZE. 3 A total of 2113 electrons were accounted from the P1 cell, equating to two diethylformamide molecules per asymmetric unit, which have been included in the unit cell contents and calculation of derived parameters. direction. Colours: copper, turquoise; oxygen, red; carbon, black; nitrogen, blue; hydrogen, white. S10

Gas Adsorption Measurements
Volumetric adsorption data for N 2 were recorded at 77K (liquid N 2 ) on an Autosorb-1c instrument under ultra-high vacuum in a clean system with diaphragm and turbo pumping system using ultra-pure research grade (99.9999%) N 2 .
The BET surface areas were calculated using the software integrated into the instrument. Vacuum dried powder samples were further degassed at 100 o C and 10 -10 bar for a minimum of 16 h to yield desolvated sample, which was then loaded in the instrument for N 2 adsorption measurements.
CO 2 and CH 4 gravimetric sorption isotherms were recorded at 273-298 K (maintained using a temperatureprogrammed water bath) on a Hiden Isochema IGA-003 system under ultra-high vacuum produced by a turbo pumping system. All gases used were ultra-pure research grade (99.999%) purchased from BOC or Air Liquide. In a typical gas adsorption experiment, around 65 mg of acetone-exchanged MFM-136 was loaded into the IGA system and outgassed at 120°C under dynamic high vacuum (10 -9 bar measured at pump) for 24 h to give fully desolvated MFM-136.

Calculation of Selectivities
The Henry's Law selectivity for gas component i over j at 273 K was calculated based on the equation below:

= /
Where is the selectivity ratio of gas i over gas j, and are Henry's Law constant for gas i and j, respectively. The Henry Law constants were calculated directly from the adsorption isotherms from fitting of the initial slope at low pressure (up to 150 mBar).   Figure S11. Linear least squares fit for the low pressure region of the N 2 isotherm for MFM-136 measured at 273 K.

Calculation of Isosteric Heats of Adsorption.
The temperature-dependent adsorption data were analysed using the virial equation: where the p is pressure, n is the amount adsorbed and A 0 , A 1 , etc. are Virial coefficients. The enthalpy of adsorption at zero coverage was determined from the relationship:

Neutron Powder Diffraction Studies
NPD experiments were undertaken using the WISH diffractometer at the ISIS facility. 10 Acetone-exchanged MFM-136 was loaded into a 6 mm diameter vanadium sample can and outgassed at 1 x 10 -7 mBar and 120 °C for 3 days. The sample was loaded into a liquid helium cryostat and cooled to 7 K for data collection of the empty framework. Guest gases CO 2 and CD 4 were volumetrically dosed from a calibrated volume after warming the sample to 293 K (CO 2 ) or 150 K (CD 4 ) with measurements made at loadings of 1.8 and 2.3 molecules of CO 2 gas per copper metal and 1.1 molecules of CD 4 gas per copper metal. The gas loading was calculated based upon the ideal gas equation (PV = nRT) for the known volume of the gas panel, and the pressure of the system was carefully controlled to ensure delivery of the targeted amount of gas into the system. The sample cell was isolated after reaching the target dosing amount to minimise the presence of "free gas" inside the can. The sample was then slowly cooled to 7 K (over ~ 3 hours) to ensure CO 2 and CD 4 were completely adsorbed with no condensation in the cell. Sufficient time was allowed to achieve thermal equilibrium before data collection.
The locations of adsorbed CO 2 and CD 4 molecules within MFM-136 were determined as a function of gas loading by sequential Fourier difference map analysis followed by Rietveld refinement using the Topas software package. [11] Analysis of the Fourier map of the outgassed data indicated no residual nuclear density in the voids. The structure from the single crystal X-ray diffraction experiment was used as a starting point for the framework model which was geometrically restrained and refined against the NPD data. The framework atom coordinates were subsequently fixed before the models of guest molecules were developed.
All binding sites were checked carefully for their unambiguous presence in the final structural model; i.e., a parallel refinement without each of the binding sites was carried out to confirm the presence of each site by comparing the R factors and the residual peaks. The CO 2 guests were constrained to have linear geometries; common C-O / C-D bond distances and isotropic thermal factors were included for the guest molecules.
Final refinements comprised all free structural variables from both the framework and guest molecules. S17  Figure S13. Observed (blue), calculated (red) and difference (grey) profiles of the Rietveld refinement of the neutron powder diffraction data (detector banks 2-5) for bare MFM-136.       * WALL indicates guest sites located in the apertures between cages A and B

INS Experiments
INS experiments were carried out on the TOSCA spectrometer at the first target station of the ISIS Facility at the STFC Rutherford Appleton Laboratory (UK). 12 TOSCA is a general purpose inelastic neutron spectrometer which is able to cover the whole range of molecular vibrations from 0-4000 cm 1 . The instrument is comprised of 130 3 He detectors in the forward and backscattering geometry located 17 m downstream of a 300 K Gd poisoned water moderator. A temperature of 10 ± 0.2 K was maintained during data collection by two He closed cycle refrigerators with 30 mbar He as an exchange gas.
A sample of desolvated MFM-136 was loaded into an 11 mm cylindrical vanadium sample container, sealed with an indium vacuum seal and connected to a gas handling system. The sample was degassed at 10 -7 mbar and 373 K for 2 days prior to the experiment to remove any remaining trace guest molecules. Gas loading was performed by a volumetric method at 293 K in order to ensure that the adsorbent was available in the gas phase and to ensure sufficient mobility within the crystalline structure of MFM-136. The sample was then slowly cooled to 10 K to ensure the guest molecule of interest was completely adsorbed with no condensation in the cell. Sufficient time was allowed to achieve thermal equilibrium before inelastic neutron spectra were collected to allow for full thermal equilibrium before data collection. In order to remove the adsorbed gas, the temperature of the sample cell was increased to 373 K and the gas dosed volumetrically back into dosing volume, when 95 % of the dosed gas was returned to the dosing volume, the sample was connected to a turbomolecular pump and degassed at 10 -7 mbar and 373 K for 2 hours to ensure all of the gas molecules had been removed.

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. 13,14 Exchange and correlation were approximated using the PBE functional. 15 For MFM-136 the structure determined by NPD was used as the initial structure. The plane-wave cut-off energy was 750 eV using on-the-fly generated pseudopotentials and spin polarised calculations were carried out. The equilibrium structure, an essential prerequisite for lattice dynamics calculations was obtained by BFGS geometry optimization after which the residual forces were S27 converged to zero within 0.007 eV/ Å -1 . Phonon frequencies were obtained by diagonalisation of dynamical matrices computed using density-functional perturbation theory. 16 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. 17 It was found that even with these stringent convergence conditions, there were a number of imaginary modes. To simplify the problem, a DFT calculation was carried out for the deprotonated linker (L 2bound to two Na + ions) as an isolated molecule with the PBE functional, norm-conserving pseudopotentials, a plane-wave cut-off of 830 eV and Γ-point sampling of the electronic states. Cu and Na have negligible contribution to the INS intensity in these calculations and the calculated INS spectrum for Na 2 L shows good agreement with the experimental INS spectrum for MFM-136, Figure S20. The vibrational modes for the ligand can therefore be identified. This approach is justified because the INS spectrum is dominated by modes that involve hydrogen motion and these occur exclusively in the linker. It is emphasised that for all the calculated spectra shown the transition energies have not been scaled. The adsorption of CO 2 at 273 and 298 K and N 2 at 77 K was simulated using grand canonical Monte Carlo (GCMC) simulations implemented in the MuSiC software package 18 and using translation, rotation and energy-biased insertion and deletion moves. All simulations were allowed at least 8 x 10 6 equilibration steps, followed by 12 x 10 6 production steps for each pressure point. The framework was treated as rigid, with atoms kept fixed at their crystallographic positions. Both Lennard-Jones and Coulombic terms were considered for both sorbate-framework and sorbate-sorbate interactions. Lennard-Jones parameters for the framework atoms were taken from the OPLS 19 force field with the exception of copper, for which UFF parameters 20 were used. Partial charges for the MOF atoms were calculated using the extended charge equilibration method. 21 Nitrogen and CO 2 were simulated as rigid molecules using the TraPPE model. 22 Adsorption sites with stronger CO 2 -MOF interaction energies will be preferentially used at low loading. It should be noted that there are a number of combinations of molecular location/orientation which result in CO 2 -MOF interaction energies between the extrema (-35 and -20 kJ/mol). As there are more than two distinct adsorption sites and there is sufficient overlap in energy between these sites, no steps are observed in the simulated or experimental isotherms.