Influence of the Amide Groups in the CO2/N2 Selectivity of a Series of Isoreticular, Interpenetrated Metal−Organic Frameworks

Here we report the use of a pillaring strategy for the design and synthesis of three novel amide-functionalized metal−organic frameworks (MOFs), TMUs-22/-23/-24, isoreticular to the recently reported imine-functionalized TMU-6 and TMU-21 MOFs. An extensive study of their CO2 sorption properties and selectivity for CO2 over N2, from single gas sorption isotherms to breakthrough measurements, revealed that not only the incorporation of amide groups but also their accessibility is crucial to obtain enhanced CO2 sorption and CO2/N2 selectivity. Therefore, the MOF with more accessible amide groups (TMU-24) shows a CO2/N2 selectivity value of ca. 10 (as revealed by breakthrough experiments), which is ca. 500% and 700% of the selectivity values observed for the other amide-containing (TMU-22 and TMU-23) and imine-containing (TMU-6 and TMU-21) MOFs. ■ INTRODUCTION Microporous metal−organic frameworks (MOFs), constructed from metal-based nodes and polytopic organic ligands, are a promising class of porous materials for carbon capture, separation, and storage owing to their very high surface areas, tunable pore sizes and shapes, and adjustable pore surface functionality. For these specific applications, it is widely accepted that certain chemical functionalities located at the pore surfaces of MOFs can increase their CO2 uptake and selectivity. Among these functionalities, CO2 has been found to bind at unsaturated metal centers via the electron-rich terminal oxygen atom. Also, CO2 can interact with electron rich Lewis basic groups via electron deficient carbon atom. For this purpose, most efforts have been made on the direct and post-synthetic functionalization of MOFs with amine groups. More recently, another approach based on the combined optimization of pore size and high charge density has anticipated very promising results. Besides the use of amine groups, the intrinsic characteristics of the amide groups make them also very attractive for establishing cooperative intermolecular interactions with CO2. 21 Amide groups possess two types of donor−acceptor affinity sites: the −NH moiety that can act as an electron acceptor (Lewis acid), and the −CO group that can act as an electron donor (Lewis base); both allowing establishment of NH···OCO and NCO···CO2 interactions. In the past few years, the study of some amide-functionalized MOFs has confirmed their potential for selectively adsorb CO2. 5,22,24−30 For example, Chen, Wang et al. have reported that pores decorated with acylamide groups in the Cu(II)-based UTSA-48 MOF promote strong interactions with CO2 and high selectivity for CO2 over CH4. 5 Recently, Wang et al. reported a threefold interpenetrated acylamide-MOF that shows higher selective adsorption of CO2 rather than other gases with a significant isosteric heat. Similarly, Zaworotko et al. found that the decoration of a rht-type MOF also with acylamide functions can significantly enhance the CO2 uptake and the selective adsorption of CO2 over N2 in comparison to PCN-61, an isoreticular analogue with acetylene moieties instead of amide groups. Remarkably, comparable trends of selectivity have also been observed in other MOFs decorated with acylamide and oxamide groups, as well as in some recent Received: July 15, 2016 Revised: August 30, 2016 Published: September 6, 2016 Article pubs.acs.org/crystal © 2016 American Chemical Society 6016 DOI: 10.1021/acs.cgd.6b01054 Cryst. Growth Des. 2016, 16, 6016−6023 porous covalent organic frameworks also containing acylamide groups. Recent studies reported by us have shown that a series of porous interpenetrated MOFs can be designed by pillaring 2D layers, comprising Zn(II) ions and the V-shaped dicarboxylate ligand 4,4′-oxybisbenzoic acid (H2oba), with linear dipyridylbased ligands (Figure 1). Following this approach, we found that the pores of the synthesized MOFs can be easily functionalized with the two groups incorporated in the pillar ligands, since the pores of these MOFs, which run perpendicular to the {Zn(oba)} layers, are also delimited by these ligands. For example, the use of the pillar ligands N,N′-bis(4-pyridylmethylene)-1,4-benzenediamine (bpmb) and its naphthalene analogue N,N′-bis(4-pyridylmethylene)-1,5naphthalenediamine (bpmn) recently allowed the formation of two isoreticular threefold interpenetrated MOFs showing pores functionalized with imine groups [called TMU-6 and TMU-21 (TMU = Tarbiat Modares University)]. Herein, inspired by this approach and the potential of the amide groups for enhancing the selectivity toward CO2, we report the synthesis of three newly acylamide-containing MOFs. The selected pillar ligands are N,N′-bis(4-pyridinyl)terephthalamide (bpta), N,N′-bis-(4-pyridylformamide)-1,4-benzenediamine (bpfb), and N,N′-bis(4-pyridylformamide)-1,5-naphthalenediamine (bpfn) (Figures 1 and S1), which lead to the assembly of: [Zn2(oba)2(bpta)]·(DMF)3 (TMU-22; where DMF is N,Ndimethylformamide); [Zn2(oba)2(bpfb)]·(DMF)5 (TMU-23), in which the acylamide group is inverted in comparison to TMU-22; and [Zn2(oba)2(bpfn)]·(DMF)2 (TMU-24), in which the use of the naphthalene analogue also induces the acylamide groups to be more directed toward the pores than in TMU-22 and TMU-23 (Figure 2). Importantly, the resulting threefold interpenetrated TMUs-22/-23/-24 are isoreticular to TMU-6 and TMU-21. The similarities (structure and stability) and differences (functional group and accessibility) of these MOFs allow study of the influence of the amide and the imine groups on their N2 and CO2 sorption properties as well as on their selective sorption of CO2 over N2. More specifically, this latter selectivity was evaluated by performing kinetics and breakthrough experiments for a CO2/N2 gas mixture. ■ EXPERIMENTAL SECTION Materials and Characterization. The commercially available reagents were used without further purification. Ligands, TMU-6 and TMU-21, were synthesized following reported procedures. See Supporting Information and Figure S1 for synthetic details. Elemental analyses were collected on a CHNS Thermo Scientific Flash 2000 elemental analyzer. Powder X-ray diffraction (PXRD) measurements were collected on a Philips X’pert diffractometer with monochromatic Cu Kα radiation (λCu = 1.5406 Å). Thermogravimetric analyses (TGA) were carried out in a PerkinElmer Pyris 1 under N2 atmosphere and heating rate of 10 °C·min−1. Fourier transform infrared (FT-IR) spectra were recorded on a Bruker Tensor 27FTIR spectrometer equipped with a Golden Gate diamond attenuated total reflection (ATR) cell, in absorption mode at room temperature. General Synthesis of TMU-22, TMU-23, and TMU-24. A mixture of Zn(NO3)2.6H2O (0.297 g, 1 mmol), H2oba (0.258 g, 1 mmol), and the corresponding amide ligand (0.5 mmol) and DMF (50 mL) was sonicated until all solids were uniformly dispersed (∼3 min) and divided into 7 glass vials. The vials were heated at 120 °C for 3 days and then cooled to room temperature. Colorless (TMU-22) and red-brown (TMU-23 and TMU-24) crystals were obtained as pure phases, washed by DMF, and dried at room temperature. Data for TMU-22. Yield: 44%. FT-IR (cm−1): 1691 (m), 1595 (vs), 1504 (s), 1395 (vs), 1331 (m), 1298 (m), 1238 (vs), 1160 (vs), 1098 (m), 777 (m), 659 (m), 524 (m). EA on solvent-free sample: calcd. Figure 1. 2D layers formed by the association between Zn(II) ions and oba linkers are further pillared by amide/imine-functionalized dipyridylbased ligands yielding to threefold interpenetrated porous pcu-MOFs. Crystal Growth & Design Article DOI: 10.1021/acs.cgd.6b01054 Cryst. Growth Des. 2016, 16, 6016−6023 6017 (%) for C46H30N4O12Zn2: C, 57.46; H, 3.15; N, 5.83; found: C, 57.50; H, 3.11; N, 5.74. Data for TMU-23. Yield: 36%. FT-IR (cm−1): 1673 (vs), 1597 (vs), 1540 (m), 1510 (m), 1399 (vs), 1309 (m), 1220 (vs), 1159 (s), 1088 (m), 1067 (m), 1014 (m), 801 (m), 658 (m), 524 (m). EA on solvent free sample: calcd. (%) for C46H30N4O12Zn2: C, 57.46; H, 3.15; N, 5.83; found: C, 57.00; H, 3.10; N, 5.65. Data for TMU-24. Yield: 38%. IR (cm−1): 1667 (vs), 1595 (vs), 1570 (m), 1505 (s), 1386 (vs), 1235(s), 1158 (vs), 1089 (m), 1065 (m), 1015 (m), 878 (m), 659 (m), 522 (m). EA on solvent free sample: calcd. (%) for C50H32N4O12Zn2: C, 59.37; H, 3.19; N, 5.54; found: C, 57.10; H, 3.12; N, 5.01. Crystallography. Crystallographic data for TMU-22 and TMU-23 were collected at 100 K at XALOC beamline at ALBA synchrotron (λ = 0.79474 and 0.82653 Å, respectively). Data were indexed, integrated, and scaled using the XDS and IMOSFLM programs. Absorption correction was not applied. SCXRD for TMU-24 was collected at 293 K on a Bruker AXS SMART Apex diffractometer using graphite monochromated Mo Kα radiation (λ = 0.71073 Å) and was corrected for Lorentz and polarization effects. The frames were integrated with the Bruker SAINT software package. Absorption corrections were applied using the program SADABS giving max/ min transmission factors of 1.000/0.270. The structures were solved by direct methods and subsequently refined by correction of F against all reflections, using SHELXS2013 and SHELXL2013 within the WinGX package. All non-hydrogen atoms were refined with anisotropic thermal parameters by full-matrix least-squares calculations on F using the program SHELXL2013. Hydrogen atoms were inserted at calculated positions and constrained with isotropic thermal parameters. In TMU-24, the carbonyl group of the N,N-ligand is distorted in two positions with 55% and 45% of occupancy. The structures contain some disordered molecules. Attempts to adequately model of disordered DMF molecules (40 electrons) were unsatisfactory; therefore, the PLATON/SQUEEZE routine was applied to mask out the disordered electron density. The residual electron density was assigned to 3, 5, and 1 molecules of DMF in TMU-22 (509 electrons per unit cell), TMU-23 (876 electrons located per unit cell), and TMU-24 (384 electrons per unit cell). Crystallographic and refinement data, and the main bond dist


■ INTRODUCTION
Microporous metal−organic frameworks (MOFs), constructed from metal-based nodes and polytopic organic ligands, are a promising class of porous materials for carbon capture, separation, and storage owing to their very high surface areas, tunable pore sizes and shapes, and adjustable pore surface functionality. 1−3 For these specific applications, it is widely accepted that certain chemical functionalities located at the pore surfaces of MOFs can increase their CO 2 uptake and selectivity. 4−10 Among these functionalities, CO 2 has been found to bind at unsaturated metal centers via the electron-rich terminal oxygen atom. 11−13 Also, CO 2 can interact with electron rich Lewis basic groups via electron deficient carbon atom. 14 For this purpose, most efforts have been made on the direct and post-synthetic functionalization of MOFs with amine groups. 15−18 More recently, another approach based on the combined optimization of pore size and high charge density has anticipated very promising results. 4,19,20 Besides the use of amine groups, the intrinsic characteristics of the amide groups make them also very attractive for establishing cooperative intermolecular interactions with CO 2 . 21 Amide groups possess two types of donor−acceptor affinity sites: the −NH moiety that can act as an electron acceptor (Lewis acid), and the −CO group that can act as an electron donor (Lewis base); both allowing establishment of NH···OCO and NCO···CO 2 interactions. 22 −24 In the past few years, the study of some amide-functionalized MOFs has confirmed their potential for selectively adsorb CO 2 . 5,22,24−30 For example, Chen, Wang et al. have reported that pores decorated with acylamide groups in the Cu(II)-based UTSA-48 MOF promote strong interactions with CO 2 and high selectivity for CO 2 over CH 4 . 5 Recently, Wang et al. reported a threefold interpenetrated acylamide-MOF that shows higher selective adsorption of CO 2 rather than other gases with a significant isosteric heat. 31 Similarly, Zaworotko et al. found that the decoration of a rht-type MOF also with acylamide functions can significantly enhance the CO 2 uptake and the selective adsorption of CO 2 over N 2 in comparison to PCN-61, an isoreticular analogue with acetylene moieties instead of amide groups. 24 Remarkably, comparable trends of selectivity have also been observed in other MOFs decorated with acylamide 25−28 and oxamide 22 groups, as well as in some recent porous covalent organic frameworks also containing acylamide groups. 29,30 Recent studies reported by us have shown that a series of porous interpenetrated MOFs can be designed by pillaring 2D layers, comprising Zn(II) ions and the V-shaped dicarboxylate ligand 4,4′-oxybisbenzoic acid (H 2 oba), with linear dipyridylbased ligands ( Figure 1). 32−34 Following this approach, we found that the pores of the synthesized MOFs can be easily functionalized with the two groups incorporated in the pillar ligands, since the pores of these MOFs, which run perpendicular to the {Zn(oba)} layers, are also delimited by these ligands. 29,30,35 For example, the use of the pillar ligands N,N′-bis(4-pyridylmethylene)-1,4-benzenediamine (bpmb) and its naphthalene analogue N,N′-bis(4-pyridylmethylene)-1,5naphthalenediamine (bpmn) recently allowed the formation of two isoreticular threefold interpenetrated MOFs showing pores functionalized with imine groups [called TMU-6 and TMU-21 (TMU = Tarbiat Modares University)]. 33,34 Herein, inspired by this approach and the potential of the amide groups for enhancing the selectivity toward CO 2 , we report the synthesis of three newly acylamide-containing MOFs. The selected pillar ligands are N,N′-bis(4-pyridinyl)terephthalamide (bpta), N,N′-bis-(4-pyridylformamide)-1,4-benzenediamine (bpfb), and N,N′-bis(4-pyridylformamide)-1,5-naphthalenediamine (bpfn) (Figures 1 and S1), which lead to the assembly of: [Zn 2 (oba) 2 (bpta)]·(DMF) 3 (TMU-22; where DMF is N,Ndimethylformamide); [Zn 2 (oba) 2 (bpfb)]·(DMF) 5 (TMU-23), in which the acylamide group is inverted in comparison to TMU-22; and [Zn 2 (oba) 2 (bpfn)]·(DMF) 2 (TMU-24), in which the use of the naphthalene analogue also induces the acylamide groups to be more directed toward the pores than in TMU-22 and TMU-23 ( Figure 2). Importantly, the resulting threefold interpenetrated TMUs-22/-23/-24 are isoreticular to TMU-6 and TMU-21. The similarities (structure and stability) and differences (functional group and accessibility) of these MOFs allow study of the influence of the amide and the imine groups on their N 2 and CO 2 sorption properties as well as on their selective sorption of CO 2 over N 2 . More specifically, this latter selectivity was evaluated by performing kinetics and breakthrough experiments for a CO 2 /N 2 gas mixture.

■ EXPERIMENTAL SECTION
Materials and Characterization. The commercially available reagents were used without further purification. Ligands, TMU-6 and TMU-21, were synthesized following reported procedures. 33,34 See Supporting Information and Figure S1 for synthetic details. Elemental analyses were collected on a CHNS Thermo Scientific Flash 2000 elemental analyzer. Powder X-ray diffraction (PXRD) measurements were collected on a Philips X'pert diffractometer with monochromatic Cu Kα radiation (λ Cu = 1.5406 Å). Thermogravimetric analyses (TGA) were carried out in a PerkinElmer Pyris 1 under N 2 atmosphere and heating rate of 10°C·min −1 . Fourier transform infrared (FT-IR) spectra were recorded on a Bruker Tensor 27FTIR spectrometer equipped with a Golden Gate diamond attenuated total reflection (ATR) cell, in absorption mode at room temperature.
General Synthesis of TMU-22, TMU-23, and TMU-24. A mixture of Zn(NO 3 ) 2 .6H 2 O (0.297 g, 1 mmol), H 2 oba (0.258 g, 1 mmol), and the corresponding amide ligand (0.5 mmol) and DMF (50 mL) was sonicated until all solids were uniformly dispersed (∼3 min) and divided into 7 glass vials. The vials were heated at 120°C for 3 days and then cooled to room temperature. Colorless (TMU-22) and red-brown (TMU-23 and TMU-24) crystals were obtained as pure phases, washed by DMF, and dried at room temperature.  Crystallography. Crystallographic data for TMU-22 and TMU-23 were collected at 100 K at XALOC beamline at ALBA synchrotron 36 (λ = 0.79474 and 0.82653 Å, respectively). Data were indexed, integrated, and scaled using the XDS 37 and IMOSFLM 38 programs. Absorption correction was not applied. SCXRD for TMU-24 was collected at 293 K on a Bruker AXS SMART Apex diffractometer using graphite monochromated Mo Kα radiation (λ = 0.71073 Å) and was corrected for Lorentz and polarization effects. The frames were integrated with the Bruker SAINT 39 software package. Absorption corrections were applied using the program SADABS 40 giving max/ min transmission factors of 1.000/0.270. The structures were solved by direct methods and subsequently refined by correction of F 2 against all reflections, using SHELXS2013 41 and SHELXL2013 42 within the WinGX package. 43 All non-hydrogen atoms were refined with anisotropic thermal parameters by full-matrix least-squares calculations on F 2 using the program SHELXL2013. Hydrogen atoms were inserted at calculated positions and constrained with isotropic thermal parameters. In TMU-24, the carbonyl group of the N,N-ligand is distorted in two positions with 55% and 45% of occupancy. The structures contain some disordered molecules. Attempts to adequately model of disordered DMF molecules (40 electrons) were unsatisfactory; therefore, the PLATON/SQUEEZE 44 routine was applied to mask out the disordered electron density. The residual electron density was assigned to 3, 5, and 1 molecules of DMF in TMU-22 (509 electrons per unit cell), TMU-23 (876 electrons located per unit cell), and TMU-24 (384 electrons per unit cell). Crystallographic and refinement data, and the main bond distances and angles, are listed in Table 1 and Tables S1−3.
Kinetics of Adsorption. Kinetic evaluation of the sorbents was performed at 298 K in homemade glass manometric adsorption equipment using pure gas components (CO 2 and N 2 ). This equipment consists of a manifold of a well-known volume (to define the initial gas dose), a sample holder containing the adsorbent, and a pressure transducer. Prior to the adsorption experiment, the samples were degassed under vacuum at the activation temperature used for single gas sorption experiments. The initial pressure in the manifold was defined at 500 mbar for all adsorbates. Once equilibrated, the manifold was expanded to the sample holder, pressure readings being recorded every second and lasting for 10 min. Moreover, the sample holder was immersed in a temperature-controlled water bath at 298 K to minimize any heat effect due to the adsorption process itself.
Breakthrough Measurements. Breakthrough curve experiments for different mixtures of gases were carried out using a column. The sample powders were packed in the middle part of the column. Here, the sample masses used were 0.30 g. The flow rates of all gases were controlled by mass flow controllers. Before the measurement, the samples were activated at the desired temperature for 2 h using a He total flow of 50 mL·min −1 .  The gas stream from the outlet of the column was analyzed online with a mass spectrometer.
The CO 2 /N 2 selectivities (σ) of samples were evaluated by the following equation: where q i is the adsorption capacity of i component, and p i is the partial pressure of i component. The amount adsorbed for each gas was calculated by the integration of the breakthrough curves. For this purpose, breakthrough experiments with H 2 (not adsorbed in these materials) have been performed using the same bed and the same flows. Then, the H 2 breakthrough curve has been used to calculate the time zero and more importantly to consider the shape of the breakthrough in the integration. The amount of gas adsorbed was the difference between the shape of the CO 2 or N 2 breakthrough and the H 2 breakthrough. Thus, the dispersion and the very small pressure drop can be considered.

■ RESULTS AND DISCUSSION
Synthesis and Crystal Structures. TMUs-22/-23/-24 were synthesized following the same pillaring strategy reported for TMU-6 and TMU-21. 33,34 Solvothermal reactions between Zn(NO 3 ) 2 ·6H 2 O, H 2 oba, and the corresponding pillar ligand in DMF at 120°C for 3 days produced prismatic crystals of TMUs-22/-23/-24 suitable for single-crystal X-ray diffraction analyses. The simulated (derived from the single crystal structures) and experimental powder X-ray diffraction (PXRD) patterns were consistent ( Figure S2), confirming that all synthesized MOFs can be obtained as pure phases; as also corroborated by Fourier transformed infrared spectroscopy and elemental analysis (performed on the activated samples, vide infra).
TMU-22 and TMU-23 crystallize in the monoclinic P2 1 /n and P2 1 /c space groups, respectively (Table 1). They are neutral threefold interpenetrated MOFs formed by {Zn(oba)} layers connected through the N,N′-pillar ligands. As in TMU-6, their basic unit is a dinuclear zinc cluster, in which both Zn(II) centers are penta-coordinated to four carboxylate O atoms from three fully deprotonated oba ligands and to one N atom from the N,N′-donor ligand (Figure 1). The geometry around the metal centers can be described as a distorted square pyramid, with Addison parameter 45 τ values in the range of 0.35−0.49. In these structures, each nonlinear dicarboxylate oba ligand is coordinated to three Zn(II) centers forming the {Zn(oba)} layers along the ab plane. Because of the dinuclear nature of the metal cluster, these layers are formed by two sheets of parallel oba ligands that form π−π stacking interactions between them ( Figure 1; Tables S4, S5). These {Zn(oba)} layers are then pillared by the linear acylamide-functionalized ligands, extending the structure in three dimensions. It is important to mention here that, due to the conformational flexibility of both oba and bpta ligands, TMU-22 is a supramolecular conformational isomer of a reported MOF. 31 Although both compounds are threefold interpenetrated, they still possess apparent 1D pore channels (∼7.0 × 6.5 Å 2 and ∼6.8 × 6.4 Å 2 including van der Waals (vdW) radiifor TMU-22 and TMU-23, respectively) running along the ⟨101⟩ direction and noticeable free space (33% for TMU-22 and 32% for TMU-23), comparable to the values found for TMU-6 (∼8.1 × 7.4 Å 2 ; and 33% free space) 46 (Figures 1 and S3−5). The channels are filled with disordered guest solvent molecules, squeezed due to thermal disorder.
Here it is very important to note that, even though TMUs-22/-23/-24 are isoreticular, their acylamide groups do not show the same orientation on the pore surfaces (Figure 2b,c,e) as well as do not participate in the same number of supramolecular interactions between the interpenetrated networks (Tables S4−6). In TMU-23, all donor −N−H and acceptor −CO groups of the two acylamide groups per pillar ligand establish H-bonds and N−H···π interactions with the other two networks forming the threefold interpenetrated structure. In TMU-22, each donor −NH moiety of the two acylamide groups per pillar ligand establishes H-bonds with carboxylate groups of the other networks, whereas the two acceptor −CO groups do not participate in H-bonds. In TMU-24, only one of the two acylamide groups forms Hbonds. Interestingly, the other acylamide group is pointing toward the pores so that it does not participate in direct Hbonds with the other networks (Figure 2e). Altogether this acylamide group is made easily accessible to potential interactions with CO 2 guest molecules. Contrariwise, even though all the other H-bonded acylamide groups of TMUs-22/-23/-24 are located on the surfaces of the pores, they are not directly pointing toward the channels, and therefore they are in principle less accessible to guest molecules.
Topology. Topological analysis performed with TOPOS 47 revealed that all the studied MOFs show the same threefold interpenetrated, 6-connected net with pcu topology (Figure 1). The pcu topology is ubiquitous in MOFs, including the iconic MOF-5. 48 Here, it is worth mentioning that the pcu net is edge-transitive with cubic symmetry and Pm3̅ m space group. However, in the present case, the overall symmetry of TMUs-22/-23/-24 decreased to monoclinic (Table 1). This symmetry decrease is attributed to the fact that the points of extension of the Zn(II) dimer do not match with the vertices of the perfect octahedron required to produce highly symmetrical pcu-MOFs (Figure 1). In addition, the use of a pillaring strategy with mixed ligands (including the flexible oba and pillar N,N′ligands) also forces the symmetry to differ from that of MOF-5.
Thermal Stability. The TGA diagrams of all as-synthesized materials revealed a first weight loss of 16 In all cases, this weight loss was attributed to the removal of all guest solvent molecules. Decomposition of all materials starts at around 400°C, confirming their high thermal stability ( Figure  S8).
Activation Process. All compounds were activated using a very similar protocol that started with a solvent-exchange step. TMUs-6/-21 and TMUs-22/-23/-24 were immersed in acetonitrile and dichloromethane, respectively, for 3 days, during which the solvent was exchanged once daily. Note here that previous stability tests showed that all synthesized MOFs were stable in these solvents ( Figure S9) for at least 24 h at r.t. prior to gas sorption measurements. It is noteworthy that PXRD diagrams collected after the sorption studies confirm that the frameworks retain their crystallinity and remained unaltered after both processes ( Figure S10). These observations were further corroborated by elemental analysis and TGA, in which only the weight losses around 400°C corresponding to their decomposition were observed ( Figure S8).

Crystal Growth & Design
Nitrogen Sorption at 77 K. N 2 isotherm at 77 K on TMU-22 showed a characteristic type I behavior of microporous materials, as expected from its crystal structure (BET area (A BET ) = 680 m 2 .g −1 ; pore volume (V t ) = 0.28 cm 3 .g −1 ). Contrariwise, TMU-21, TMU-23, and TMU-24 were found to be nonporous materials (negligible N 2 uptake). In the case of TMU-6, it showed much lower porosity than that expected from its crystal structure (A BET = 150 m 2 .g −1 ; V t = 0.08 cm 3 .g −1 ) (Table 2, Figure S11). Thus, all activated compounds, except TMU-22, reveal a "N 2 -phobic" behavior. 49−52 We reasoned that this behavior could be due to the existence of different structural rearrangements depending on the MOF during sorption or when exposed to cryogenic temperatures and/or under vacuum, reducing/preventing access to the porosity. Indeed, similar temperature dependent behaviors are already well-known in soft materials such as in MIL-53 (Al). 53 However, other factors cannot be fully excluded, such as the existence of strong interactions between N 2 and the channel walls at 77 K that hinder diffusion into the material. 54 CO 2 Sorption at Various Temperatures and Heats of Adsorption. Contrariwise to the adsorption of N 2 , the CO 2 sorption isotherms at various temperatures revealed that all of them are porous (Figures 3a and S12a−16a). CO 2 uptake at 203 K was calculated to be 9.0, 4.5, 7.2, 7.2, and 6.3 mmol.g −1 for TMU-6, TMU-21, TMU-22, TMU-23, and TMU-24, respectively ( Table 2). The observed pore volumes (at 760 Torr) were 0.32 cm 3 .g −1 (TMU-6), 0.16 cm 3 .g −1 (TMU-21), 0.25 cm 3 ·g −1 (TMU-22), 0.27 cm 3 .g −1 (TMU-23), and 0.22 cm 3 .g −1 (TMU-24). These pore volumes calculated from the CO 2 adsorption data are in agreement with the theoretical values extracted from the crystal structures. The variations observed can be attributed to small structural changes widely observed in such interpenetrated sorbents. 35,55−57 Isosteric heats of adsorption of CO 2 (Q st ) were calculated from the isotherms collected at different temperatures (Table 2 and Figures S12−16). The two acylamide-functionalized TMU-24 and TMU-22 exhibited the highest Q st for CO 2 , from 24 to 26 kJ·mol −1 and 23 to 24 kJ·mol −1 , respectively. The different Q st values for CO 2 between TMU-24, TMU-22, and TMU-23 can be attributable to the different accessibility of their acylamide groups and to the different number of supramolecular interactions in which they are involved (vide supra). CO 2 /N 2 Separation Properties. In light of the porosity of the materials for CO 2 and their apparent low affinity for N 2 , we further investigated their potential for CO 2 /N 2 separation by performing kinetics experiments (Figures 4 and S17). As expected, the highest selectivity (σ) for CO 2 over N 2 was observed for TMU-24 (σ = 14.1, Table 3), which is the material exhibiting the highest Q st of CO 2 and with the most accessible acylamide groups pointing toward the porous channels. Overall, Table 2. Summary of BET Areas, Total Pore Volumes, and CO 2 Uptake Observed for TMUs-6/-21-24   the CO 2 /N 2 selectivity of TMU-24 estimated from kinetics experiment is ca. 70% higher than for the other materials of the family.
These results were further confirmed by breakthrough experiments with CO 2 /N 2 gas mixture ( Figures 5 and S18).
Once again, TMU-24 performed much better than its isoreticular variants, with a CO 2 /N 2 selectivity of 10.0 at r.t. for an equimolar CO 2 /N 2 gas mixture ( Figure 5). This value represents 700% of the selectivity found in similar conditions for its imine analogue TMU-21 (σ = 1.4). It also performs better than TMU-22 and TMU-23, the other acylamide containing MOFs with less accessible functional groups. Indeed, the selectivity of TMU-24 is 500% and 580% the selectivity of TMU-22 and TMU-23, respectively. Therefore, not only the incorporation of amide functional groups in the frameworks but also their accessibility for the CO 2 guest molecules are crucial to enhance the CO 2 /N 2 selectivity. This observation was further confirmed in more realistic gas mixture conditions (CO 2 :N 2 = 1:9), in which TMU-24 was still found to perform well with a selectivity of 4.6 ( Figure 5). It is important to remark that the CO 2 adsorption capacity found in the breakthrough experiment at r.t. (∼0.65 mmol.g −1 ) for TMU-24 is slightly lower than the one extracted from the isotherm at 500 mbar and r.t. (∼0.85 mmol.g −1 ). This feature can be attributed to a small amount of N 2 adsorbed in these conditions (298 K and gas mixture).

■ CONCLUSIONS
We successfully synthesized three novel, threefold interpenetrated pcu-MOFs containing acylamide functional groups (TMU-22, TMU-23, and TMU-24), isoreticular to the recently reported imine-containing MOFs, TMU-6 and TMU-21. The study of their sorption properties revealed that the insertion of amide functional groups in MOFs does not necessarily induce a noticeable improvement of the CO 2 /N 2 separation properties. Indeed, another aspect that needs to be addressed is the accessibility of the functional groups, especially in interpenetrated frameworks. If neglected, it could jeopardize the efficiency of MOFs functionalization, as illustrated by the performance of TMU-22 and TMU-23, two acylamidefunctionalized analogues of the parent imine-functionalized TMU-6; that all three of these MOFs show negligible difference of the selectivity. This demonstrates the needs for smart materials design approaches, further confirmed by the deliberate alteration of the pore shape of TMU-24 using a bulky, central naphthalene core. This resulted in better accessibility of the acylamide groups, leading to a 70% improvement of CO 2 /N 2 selectivity according to kinetics studies (from 8.2 to 14.1 for TMU-21 and TMU-24, respectively). Moreover, a dramatic increase of the selectivity was obtained from breakthrough experiments. The CO 2 /N 2 selectivity for TMU-24 corresponds to ca. 700% of this iminecontaining analogue, TMU-21. Moreover, its selectivity is ca. 500% and 580% of TMU-22 and TMU-23, two isoreticular, amide containing MOFs with less accessible functional groups.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.cgd.6b01054.
Synthetic procedure for ligand synthesis, crystallographic data, PXRD diagrams, TGA plots, gas sorption and separation data (PDF)

Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.