H2/CO2 separations in multicomponent metal-adeninate MOFs with multiple chemically distinct pore environments†

Metal–organic frameworks constructed from multiple (≥3) components often exhibit dramatically increased structural complexity compared to their 2 component (1 metal, 1 linker) counterparts, such as multiple chemically unique pore environments and a plurality of diverse molecular diffusion pathways. This inherent complexity can be advantageous for gas separation applications. Here, we report two isoreticular multicomponent MOFs, bMOF-200 (4 components; Cu, Zn, adeninate, pyrazolate) and bMOF-201 (3 components; Zn, adeninate, pyrazolate). We describe their structures, which contain 3 unique interconnected pore environments, and we use Kohn–Sham density functional theory (DFT) along with the climbing image nudged elastic band (CI-NEB) method to predict potential H2/CO2 separation ability of bMOF-200. We examine the H2/CO2 separation performance using both column breakthrough and membrane permeation studies. bMOF-200 membranes exhibit a H2/CO2 separation factor of 7.9. The pore space of bMOF-201 is significantly different than bMOF-200, and one molecular diffusion pathway is occluded by coordinating charge-balancing formate and acetate anions. A consequence of this structural difference is reduced permeability to both H2 and CO2 and a significantly improved H2/CO2 separation factor of 22.2 compared to bMOF-200, which makes bMOF-201 membranes competitive with some of the best performing MOF membranes in terms of H2/CO2 separations.


S9
Determination of Composition: Elemental analysis was used to determine the abundance of C, H, N, and Zn in a dried sample of as-synthesized bMOF-200. The data revealed a molecular formula of Zn2Cu(4-pyz)2(ad)(DMF) • (DMF)2,(H2O)3 which closely matched the ratios of ligands, metals, and solvents observed in the refined crystal structure. 1 H-NMR was conducted to confirm the relative amounts of the organic components and any solvent within the framework ( Figure S1). The resulting spectra yielded a 4-pyz : adenine ratio of approximately 2 : 1 and three DMF molecules, as well as some water. TGA analysis was then used to determine the mass percent of solvents (DMF and water) in the dried material by measuring the change in mass with increasing temperature ( Figure S2). The solvent molecules (3 water and 3 DMF) account for approximately 33% of the total mass. The TGA plot shows a 33% weight loss by 250°C, well above the boiling point of both compounds. We note that the small pore windows within bMOF-200 may inhibit solvent loss. All three characterization methods agree well with the determined molecular formula. XPS was used to determine the oxidation state of the Cu in a MeOH exchanged sample (vide infra) of bMOF-200 ( Figure S3). The Cu spectrum was collected before and after a 10 second sputtering from an Ar ion beam to etch the surface of the crystals. Before any etching occurred, the Cu spectrum shows Cu(II) character as evidenced by a satellite peak centred at 943 eV. The loss of this satellite peak and a concurrent shift in the 2p3/2 signal to lower binding energies in the post-etching spectrum indicate an increase in Cu (I) character, which suggests that surface oxidation occurs on bMOF-200 crystals, converting Cu (I) to Cu (II). We further investigated the Cu sites using EPR spectroscopy to gain insight into their coordination geometry ( Figure S4). The spectrum for as-synthesized bMOF-200 yielded an AII value S10 of 144 which is within the range of values assigned to a square planar geometry. Finally, PXRD patterns showed a nearly identical match to the simulated powder pattern derived from the single crystal structure, indicating the phase purity of the compound ( Figure S5). Figure S1. 1 H-NMR spectrum of as-synthesized bMOF-200 after washing with DMF and drying on a Schlenk line for 1 hour. The peaks at 8.57 ppm corresponds to adenine protons. The broad peak at 8.114 ppm is assigned to 4-pyz protons. The small peak at 8.147 ppm is due to a small amount of formate from the decomposition of DMF. Peaks at 7.95, 2.89, and 2.73 ppm correspond to DMF. The integration shows approximately three DMF molecules per adenine molecule. Figure S2. TGA of as-synthesized bMOF-200. The black curve shows the recorded weight loss with increasing temperature and the red curve is the first derivative weight loss with respect to temperature. ~33% of the sample weight is lost before 250 °C, which corresponds to 3 DMF and 3 H2O. Subsequent weight loss steps are ascribed to sample decomposition. Figure S3. XPS of methanol-washed bMOF-200 before (red) and after (black) surface etching with an Ar + beam. A slight shift in both Cu 2p1/2 and Cu 2p3/2 peaks to lower binding energies and the loss of the satellite peak between 945 and 940 eV indicate the presence of more Cu(I) character after etching. Therefore, we attribute the Cu(II) character primarily to surface oxidation of the MOF crystals in air. Figure S4. CW-EPR of as-synthesized bMOF-200 with respective g and A tensor values. The simulated curve (red) is a combination of two separate simulated components: a broad, featureless signal (blue) and a signal with hyperfine splitting (green). Figure S5. PXRD patterns of simulated (black) and as-synthesized bMOF-200 (red). Simulated pattern was calculated from SC-XRD data.

Solvent exchange procedures
As-synthesized bMOF-200 crystals soaking in DMF were exchanged with 10 mL of dry methanol three times a day for three days. Then, the crystals were dried under a N2 stream until they became a free-flowing powder. The composition of the dried samples water molecules were also observed. TGA analysis revealed an approximate loss 11% by mass by 230°C which correspond to three water molecules and one MeOH (calculated 12.9%) ( Figure S7). The loss of the final MeOH molecule occurs at the small step between 230-280°C and accounts for an additional mass loss of ~5% (calculated 4.8%).
We attribute this step to the loss of coordinated methanol, the displacement of which requires temperatures above the normal boiling point at atmospheric pressure. A S16 coordinated MeOH is indeed resolved in the single crystal structure of the methanolexchanged crystal (see page S54). FT-IR spectroscopy was used to monitor the solvent exchange in bMOF-200 ( Figure S8).  Figure S9. PXRD pattern of simulated (black) and MeOH-bMOF-200 (red) (A). A peak at 4 2q degrees becomes much more pronounced after solvent exchange with MeOH and matches a peak at the same position in the simulated PXRD pattern (B). The range between 3.8 and 4.2 2q degrees in the simulated pattern in Figure S9B is multiplied by 100 to make this peak more apparent. This peak corresponds to the (200), which bisects the crystal along any of the equivalent faces and in which the coordinated DMF lies. The increase in intensity may result from replacing the coordinated DMF with MeOH.
A B S20

Synthesis of bMOF-201
Unlike bMOF-200, bMOF-201 syntheses did not require the use of evacuated tubes to obtain a pure product. In a typical reaction, 0.125 mmol (16.    The black curve shows the recorded weight loss with increasing temperature and the red curve is the first derivative weight loss with respect to temperature. The loss of a portion of the uncoordinated water molecules (~11%) and MeOH occurs before 100°C. The weight loss between 200°C and 250°C is attributed to the removal of coordinated monocarboxylates formate or acetate, which is approximately 5% of the total mass. Subsequent weight losses are attributed to framework decomposition and release of any trapped solvent. Figure S12. PXRD patterns of simulated (black) and synthesized bMOF-201 (red). Simulated pattern was calculated from SC-XRD data.
Approximately 50 mg of sample was added to a pre-weighed sample tube, which was installed onto an instrument port. The sample was then heated at 60°C for 1 hour, 100°C for 2 hours, then finally at 150°C for 24 hours under vacuum. For each temperature ramp, the rate was maintained at 10°C/min. After cooling back to room temperature, the sample was weighed again, then installed onto a Micromeritics 3-Flex gas adsorption analyzer.

Gas adsorption isotherms
The N2 gas adsorption isotherm was collected on a Micromeritics 3-flex gas adsorption

Computational Methodology
Density Functional Theory (DFT) calculations were used to examine physisorption of H2 and CO2 inside the MOF and the diffusion barrier of H2 and CO2 through the two distinctive openings on the Zn-pyrazolate cage. The structure of the MOF was obtained through SC-XRD. The base unit cell contains a total of 2448 atoms, which is computationally demanding, if not infeasible. Thus, we have generated a reduced unit cell with 888 atoms, by replacing some of the adenine ligands with their mirrored images. We believe this necessary simplification is valid, because one would expect stereochemistry at the substituted locations to play a minimum role in the adsorption and diffusion of linear molecules. In each physisorption calculation, we have placed one guest molecule inside the MOF and the entire structure was optimized with no geometric constraints. We have tested 15 unique initial placements in each set of H2 and CO2 physisorption calculations.
The shortest distance between the adsorbate and the MOF at the optimized structure is reported in Table S1. The positioning of the CO2 with the largest binding energy with bMOF-200 is shown in Figure S15.

Density Functional Theory (DFT) Details
The DFT calculations performed here are done using the CP2K code. 2 The PBE functional was used to describe the exchange-correlations effects, and Grimme's D3 dispersion correction was included in order to describe van der Waals interactions inside the MOF. 3,4 Atomic species were described using the DZVP-MOLOPT basis set in combination with Geodecker, Teter and Hutter pseudopotentials, with a planewave cutoff of 360 Ry and relative cutoff of 60 Ry. 5 For the periodic system, the reduced unit cell has fixed angles of α = β = γ = 60°, and the optimize unit cell length is 31.061 Å, which is in close agreements with the experimental value of 30.864 Å. The freestanding calculations were performed in a periodic 30 × 30 × 30 Å 3 cubic cell. The diffusion barrier of H2 and CO2 through the square and triangular openings on the freestanding Zn-pyrazolate cage was investigated using nudged elastic band method implemented in CP2K, and the force acting on any atom is below 0.077 eV/Å at the optimized geometry. 6,7 The DFT functional used in this study has a mean absolute deviation (MAD) of 0.076 eV when benchmarked against the data set (S22x5) for interaction energies of noncovalently bonded complexes. 8,9 We believe the uncertainty in our calculated physisorption energy and diffusion barrier is similar in magnitude as the reported MAD. We do not expect systematic DFT errors, such as self-interaction, to be prominent for the weakly bound adsorbates investigated. 10

Breakthrough experiments
Breakthrough experiments were used to investigate the separation of H2 and CO2 by

Synthesis of UiO-66
UiO-66 was synthesized following a literature procedure. 11 A ratio of H2-BDC:Zr of 2:1 was used and the synthesis was performed at 200°C.

Synthesis of HKUST-1
HKUST-1 was synthesized following a literature procedure. 12

Synthesis of ZIF-8
ZIF-8 was synthesized following a literature procedure. 13 PXRD patterns were collected and compared to simulated patterns generated from the reported single crystal data in each respective reference ( Figure S16) (Figures 5c and 5d).  Table S1 and Table S2. The permeance of each membrane was measured for three consecutive days and both permeances and separation factor remained unchanged. The points for bMOF-200 and bMOF-201 reported in the Robeson plot (Figure 7) are the average values of the three different membranes listed in Tables S1 and Table S2.

As-synthesized-bMOF-200
Single crystal X-ray diffraction data for bMOF-200 was collected on a Bruker X8 Prospector Ultra diffractometer equipped with an Apex II CCD detector and an IμS microfocus CuK-α X-ray source (λ= 1.54178 Å). A purple cubic crystal of dimensions 0.07 x 0.07 x 0.07 mm 3 was mounted on a goniometer using MiTeGen MicroMesh tips. Data was collected under N2 stream at 230 K and processed using the Bruker APEX II software package.
Centrosymmetric space group 3 3 was determined based on intensity statistics and systematic absences. The data were collected and integrated to 0.89 Å by Bruker program SAINT. 14 Empirical absorption correction was applied using program SADABS. 14 The structure was solved with direct method using SHELXT and refined by full-matrix least-squares on F 2 using SHELXL in Olex2. [15][16][17] All the non-H atoms were refined anisotropically. All the H atoms were refined isotropically. Crystallographic data are summarized in Tables S5-S10.

MeOH-bMOF-200
Single crystal X-ray diffraction data for MeOH-bMOF-200 was collected on a Bruker X8 Prospector Ultra diffractometer equipped with an Apex II CCD detector and an IμS microfocus CuK-α X-ray source (λ= 1.54178 Å). A blue cubic crystal of dimensions 0.065 mm 3 was mounted on a goniometer using MiTeGen MicroMesh tips. Data was collected under N2 stream at 150 K and processed using the Bruker APEX II software package.
Centrosymmetric space group 3 3 was determined based on intensity statistics and systematic absences. The data were collected and integrated to 0.89 Å by Bruker program SAINT. 14 Empirical absorption correction was applied using program SADABS. 14 The structure was solved with direct method using SHELXT and refined by full-matrix least-squares on F 2 using SHELXL in Olex2. [15][16][17] All the non-H atoms were refined anisotropically. All the H atoms were refined isotropically. The asymmetric unit is shown in Figure S22. Crystallographic data are summarized in Tables S13-S20.
Data were collected under N2 stream at 150 K and processed using the Bruker APEX II software package.
Centrosymmetric space group 3 3 was determined based on intensity statistics and systematic absences. The data were collected and integrated to 0.90 Å by Bruker program SAINT. 14 Empirical absorption correction was applied using program SADABS. 14 The structure was solved with direct method using SHELXT and refined by full-matrix least-squares on F 2 using SHELXL in Olex2. [15][16][17] All the non-H atoms were refined anisotropically. All the H atoms were refined isotropically. Refinement details are included in the CIF. The asymmetric unit is shown in Figure S23. Crystallographic data are summarised in Tables S21-S29.