Combining In Situ Techniques (XRD, IR, and 13C NMR) and Gas Adsorption Measurements Reveals CO2-Induced Structural Transitions and High CO2/CH4 Selectivity for a Flexible Metal-Organic Framework JUK-8.

Flexible metal-organic frameworks (MOFs) are promising materials in gas-related technologies. Adjusting the material to processes requires understanding of the flexibility mechanism and its influence on the adsorption properties. Herein, we present the mechanistic understanding of CO2-induced pore-opening transitions of the water-stable MOF JUK-8 ([Zn(oba)(pip)]n, oba2- = 4,4'-oxybis(benzenedicarboxylate), pip = 4-pyridyl-functionalized benzene-1,3-dicarbohydrazide) as well as its potential applicability in gas purification. Detailed insights into the global structural transformation and subtle local MOF-adsorbate interactions are obtained by three in situ techniques (XRD, IR, and 13CO2-NMR). These results are further supported by single-crystal X-ray diffraction (SC-XRD) analysis of the solvated and guest-free phases. High selectivity toward carbon dioxide derived from the single-gas adsorption experiments of CO2 (195 and 298 K), Ar (84 K), O2 (90 K), N2 (77 K), and CH4 (298 K) is confirmed by high-pressure coadsorption experiments of the CO2/CH4 (75:25 v/v) mixture at different temperatures (288, 293, and 298 K) and in situ NMR studies of the coadsorption of 13CO2/13CH4 (50:50 v/v; 195 K).

Introduction

According to the Intergovernmental Panel on Climate Change (IPCC),[1] the growing concentration of carbon dioxide in the atmosphere has enhanced the greenhouse effect. This has triggered environmental issues such as droughts, wildfires, flooding, and heatwaves.[2] Beyond many anthropogenic sources of CO2 emission, chemical separation is responsible for the release of 10–15% of global output,[3] and increasing demand for high-purity chemicals will enhance its impact.

Over 40 years of industrial separation and purification, microporous materials have played a crucial role in these processes,[4,5]e.g., zeolites, silica, alumina, and activated carbon are commercially used in paraffins/isoparaffins, N2/O2, O2/N2, C2H4/C2H2 separation, or CH4 purification.[4] To meet the demand and reduce the impact of the industry on the environment, cooperative efforts are necessary to develop materials and procedures for green technologies.[6] Recently, a novel class of porous compounds, metal–organic frameworks (MOFs), have emerged as a potential game-changer in gas-related technologies.[7−12] In 2006, Chen et al. have shown the first gas chromatographic separation of alkanes by using twofold interpenetrated MOF-508.[13] Since this time, many valuable reports stressed the usefulness of MOFs in purification and separation based on different technologies. For example, Long and co-workers have evaluated rigid MOF-177 and CPO-27 for postcombustion carbon dioxide capture via temperature swing adsorption.[14] Zaworotko et al. used synergetic MOF sorbents to make ethylene pure enough for producing polymers,[15] and the Eddaoudi group upgraded natural gas.[16] Among these reports, it is important to highlight the promising role of MOFs in the construction of semipermeable mixed-matrix membranes for separation processes of industrially important molecules.[17] Further progress in gas-related technology could result from flexible MOFs, which are a subgroup of metal–organic frameworks that respond to external stimuli like adsorbates with considerable structural transformation.[18−24] Intrinsic flexibility improves selectivity,[25−27] considerably increases the working capacity,[28,29] enables self-accelerating CO sorption,[30] and influences gas separation.[31] However, examination of the transition mechanism caused by the external stimulus requires sophisticated in situ techniques,[32]e.g., nuclear magnetic resonance (NMR),[33] powder X-ray diffraction (PXRD),[34] infrared spectroscopy[35] (IR), and others.[36] The obtained information serves as input for the understanding–tuning–developing cycle for adjusting the crucial features of the adsorbent for advanced applications.

The acylhydrazone MOFs[37] are a novel and developing group of metal–organic frameworks that bear the −C(O)=N–NH– motif on the pore surface. Members of this family show very interesting properties, such as extraordinary stability,[38] structural transformation,[39] excellent catalytic reactivity in CO2 fixation to epoxides,[40] proton conductivity,[41] as well as sensing activity.[42] Reports about guest–framework interactions with similar amide groups can be found in the literature.[43] However, to the best of our knowledge, there is still lack of comprehensive experimental study concerning the influence of the −C(O)=N–NH– moiety on those MOF properties. In this work, we have bridged this gap by utilization of three complementary in situ techniques (IR, PXRD, 13C NMR of CO2) corroborated with single-crystal X-ray diffraction (SC-XRD). By this approach, we characterize the CO2-induced transition mechanism of water-stable JUK-8 ([Zn(oba)(pip)] oba2– = 4,4′-oxybis(benzenedicarboxylate), pip = 4-pyridyl-functionalized benzene-1,3-dicarbohydrazide).[24]In situ PXRD provides global information about two-step structural transformation, while in situ IR and NMR shed light on the interaction between carbon dioxide and acylhydrazone group (−C(O)=N–NH−).[39 ,40] Moreover, one-component (CO2, CH4, Ar, O2, and N2) and multicomponent (CH4/CO2) equilibrium adsorption studies in a broad temperature range have shown high selectivity of JUK-8 toward carbon dioxide.

Results and Discussion

2.1 Guest-Dependent Structural Transformations Elucidated by Ex Situ SC-XRD

JUK-8 (Jagiellonian University in Kraków-8) is a microporous MOF assembled from eight interpenetrated subnetworks held together by hydrogen bonds and π···π stacking interactions (Figures S1 and S2; Tables S1 and S2).[24] Despite the high level of interpenetration, fully solvated JUK-8op {[Zn(oba)(pip)]·DMF·3H2O}n (CSD code: ZUFXIK) has one-dimensional zig-zag channels propagating along the [001] direction (Figure ). Upon thermal removal of guest molecules (443 K, 10–3 mbar), further denoted as activation, all eight diamondoid subnetworks collectively breathe to reach a new closed phase JUK-8cp ([Zn(oba)(pip)]; Figures S3 and S4), whose structure was elucidated in this work by SC-XRD.

Figure 1

Comparison of three guest-dependent crystalline phases of JUK-8 elucidated by SC-XRD: (a) packing diagrams showing the entrance to the [001] channel, H atoms are omitted; (b) single subnetworks viewed along the c axes; (c) configuration of the oba2– linker, H atoms are omitted; (d) Zn coordination sphere; selected bond lengths (Å) and angles (o) in JUK-8op, HO@JUK-8ip, and JUK-8cp (for more information, see Table S4). Zn, green; C, gray; O, red; N, blue; and H, pale gray.

Due to the high affinity of JUK-8cp toward water, the desolvated phase exposed to a trace amount of moisture immediately transforms to the previously described intermediate phase {[Zn(oba)(pip)]·2H2O} (HO@JUK-8ip; Figures S3–S5, CSD code: ZUFXOQ).[24] To prevent water adsorption by JUK-8cp, a suitable single crystal of the unknown cp phase was placed under an inert atmosphere into a preheated capillary, which was sealed and transferred for synchrotron SC-XRD measurements (Figures S6 and S7).

During the activation of JUK-8op, the monoclinic symmetry of the crystal structure (space group 2/) remains unchanged; however, the one-dimensional channels transform to zero-dimensional cages (Figure S8). Shrinking of the unit cell volume (8050.1 Å3 → 6531 Å3) upon transition from the op to cp phase is accompanied by a considerable contraction of the a-axis and moderate changes of c, b, and β cell parameters (Tables and S3). The meticulous comparison of the three structures reveals that the distances between the nearest symmetry equivalent zinc atoms (Zn···Zn) from different subnetworks are 7.63 Å (op), 7.53 Å (ip), and 7.43 Å (cp). The relatively small difference of Zn···Zn distances (Δ = 0.20 Å) between op and cp phases proves that the breathing motion practically does not change the relative positions of subnetworks and the observed changes mostly rely on rearrangements around Zn2+ cations including slight bending of the oba2– linkers (Figure and Table S4). Thermal removal of guest molecules from JUK-8op also causes reinforcement of hydrogen bonds between adjacent subnetworks (N(4)-H(4)···O(3); Table S1), as well as it is responsible for considerable shortening (by 0.25 Å) of the Zn1-O7 bond (Figure ).

Table 1

Unit Cell Parameters of Investigated Phases Found by SC-XRD and Obtained from PXRD Patterns, Juxtaposed with Corresponding Geometric Porosity Parametersa

	V /Å3	a /Å	b /Å	c /Å	β/deg	space group	mpd/Å	pws/Å	Vpt/cm3·g–1	Vpe/cm3·g–1
JUK-8opb	8050	16.98	18.51	26.16	101.65	monoclinic C2/c	4.67	4.13	0.241
H2O@JUK-8ipb	6727	13.87	17.54	27.72	94.34		3.53		0.035
JUK-8cpb	6531	12.98	18.06	27.94	94.01		3.43		0.012
JUK-8cpc	6518	13.05	17.98	27.84	93.30
CO2@JUK-8ipc	6631	13.28	18.00	27.82	94.46					0.037
CO2@JUK-8opc	7817	16.04	18.13	26.96	98.87					0.220

mpd, maximum pore diameter (Zeo++);[44]pws, pore window size (Zeo++); Vpt, theoretical pore volume data from single-crystal structure (Mercury 4.3.1; probe radius =1.3 Å); and Vpe, experimental pore volume for CO2 adsorption (195 K) calculated according to the Gurvich rule (Figure ).

Data from single-crystal structure.

Data from in situ powder X-ray diffraction analysis (195 K) at p/p0 = 0.12 and 0.85 for CO and CO, respectively.

Figure 2

In situ monitoring of CO2 adsorption by JUK-8cp at 195 K: (a) adsorption/desorption isotherms; (b) PXRD patterns (λ = 1.540599 Å) measured in parallel to CO2 physisorption; (c) unit cell volume changes during adsorption/desorption; and (d) 13C NMR of adsorbed 13CO2 as a function of pressure increase (adsorption) and subsequent pressure release (desorption) at 195 ± 3 K. Purple and red dashed lines visualize the changes of signals I and II, respectively. A narrow signal at 127.8 ppm of a pure 13CO2 gas at 1 bar (195 K) is indicated by an orange rectangle.

CO2-Induced Transformation Monitored by In Situ Techniques (PXRD, IR, and 13C NMR)

By combining three complementary in situ techniques (IR, PXRD, and NMR) during CO2 adsorption (195 K), supported by the single-crystal investigation, we shed light on the mechanism of flexibility in JUK-8 and its influence on the framework properties. PXRD provides global information about CO2-induced JUK-8 breathing (Figure S9), whereas IR and NMR spectroscopies probe interactions between CO2 and functional groups.

The measured CO2 adsorption isotherm (195 K), followed by in situ PXRD, on a ground sample of JUK-8 demonstrates a good agreement with the ex situ data (Figures S10 and S11) and the previously published isotherm.[24] Unit cell parameters derived from the PXRD pattern of the activated sample match the calculated parameters from the single-crystal structure of the desolvated phase (JUK-8cp), and it indicates that the used model is correct (Table ).

From a structural point of view, the mechanism of CO2-induced transition involves two steps. (1) In the first step (pressure range p = 0.00–0.12 bar), JUK-8cp adsorbs ∼ 0.5 CO2 molecules per Zn (CO@JUK-8ip, [Zn(oba)(pip)]·1/2CO2) and unit cell volume slightly swells (by 1.7%). It indicates that carbon dioxide molecules occupy 0-D cages, each between two zinc atoms (Figure S8). (2) Exceeding p = 0.12 bar causes the second opening step, characterized by the change of pore dimensionality (0-D → 1-D; CO@JUK-8ip → CO@JUK-8op), and the unit cell volume of JUK-8 abruptly increases to 7640 Å3 (by 17.2%). Further adsorption of CO2 leads to slight swelling, and the highest unit cell volume is observed at p = 0.85 bar on a desorption branch (7812 Å3; increase by 19.9%), which is 223 Å3 lower than for the H2O/DMF-loaded JUK-8op.

A comparable mechanism of CO2 adsorption is found for the SNU-9 material.[45] However, in its desolvated phase, SNU-9 has one-dimensional channels that enable diffusion. For JUK-8, the CO2 transport mechanism to 0-D cages is still unknown and will be the subject of further detailed investigations.

Assuming that the mutual position of the eight subnetworks during CO2 adsorption does not change, the transition of JUK-8cp to CO@JUK-8op, similarly to the activation process (Figure ), involves collective breathing of all subnetworks with the rearrangement of zinc coordination spheres (Figure and Table S4). The comparison of the IR spectra collected for CO@JUK-8op under CO2-rich atmosphere (∼0.99 bar; 195 K) and for the evacuated sample JUK-8cp proves this hypothesis (Figures and S12).

Figure 3

In situ IR spectra recorded during carbon dioxide adsorption: OCO asymmetric stretching region for the oba2– carboxylates (left) and characteristic (N–H)pip, (C=O)pip, and CO2 regions (right). Black curve, JUK-8cp (195±3 K); red curve, CO2-loaded JUK-8op (195±3 K; p/p0 = 0.99).

The structural transformation entails alteration of the OCO angles and the Zn–O bond lengths, which is observable by considerable changes in the OCO asymmetric stretching region of the oba2– carboxylate linkers. Additionally, we have observed two signals from the adsorbed CO2 (at 2340 and 2376 cm–1), as compared to gaseous CO2 (2345 cm–1). The broad and intense signal at 2340 cm–1 can be ascribed to the weakly interacting CO2 inside the pore; the second one at a higher wavenumber is associated with a more directional interaction between the framework and carbon dioxide. The meticulous comparison of the bands from acylhydrazone groups (−C(O)=N–NH–; i.e., ν(N–H)pip and ν(C=O)pip) in JUK-8cp and CO@JUK-8op shows that electron density is exchanged between CO2 and those groups. In the spectra, it is manifested by changes of band position and their intensity. Furthermore, the described observations clearly demonstrate the role of an acylhydrazone in the proposed catalytic mechanism for CO2 cycloaddition that leads to terminal/internal epoxides, as described by Suresh and co-workers (Figure S13).[40]

To get a deeper insight into the mechanism of CO2 adsorption, we have supported in situ IR and PXRD investigations by in situ NMR measurements. The latter technique is known for high sensitivity to any changes in the electron structure of the investigated species. It can be used for studying host–guest interactions as well as to distinguish between different adsorbates inside the framework and the nonadsorbed free gas (see, for example, refs (46−53) and references therein). 13C NMR spectroscopy of adsorbed CO2 is frequently used to characterize porous materials such as MOFs (see, e.g., the review articles[53,54,59]). For the investigation of the single- and mixture gas adsorption, the previously reported homebuilt in situ apparatus was used.[50] Pure 13CO2 gas at 1 bar (195 K) yields a narrow signal at 127.8 ppm in agreement with the literature[51,52] (Figure ).

According to in situ PXRD, the CO2-induced transition (195 K), from cp → CO → CO phases, occurs between p/p0 = 0.00 and 0.12 (Figure a). However, in the case of in situ NMR, possibly due to a minor temperature difference, the gate opening pressure (gop) shifts to p/p0 = 0.16–0.19. Furthermore, coexistence of small amounts of different phase impurities visible in in situ PXRD collected during CO2 adsorption can also have a minor impact (Figures a and S11).

In the intermediate phase (CO@ip), the CO2 molecules are confined, and their mobility is restricted. Consequently, at p/p0 = 0.09, only a very broad signal (signal I) ranging from ca. 70 to 180 ppm with an intensity maximum at 178 ppm is observed (Figure ). The line shape is typical for CO2 and resembles the line shape observed for signals dominated by chemical shift anisotropy with rotational symmetry in powder samples. The chemical shift tensor then exhibits the two principal values δ⊥ (perpendicular to the symmetry axis) and δ|| (parallel to the symmetry axis). However, the measured chemical shift anisotropy (CSA) Δσ = δ⊥ – δ|| = 110 ppm is considerably lower than the value of 355 ppm, which would be expected for fully immobilized CO2 molecules in powder samples.[32,53] This indicates a restricted mobility in the pores resulting in partial averaging of the CSA. A similar property, the so-called residual dipolar couplings, is a well-known phenomenon in liquid-state NMR spectroscopy.[54,55] In the case of CO2 in MOFs, the molecules rapidly travel through the pores. For spatially anisotropic pore systems, the described averaged line shape results. NMR spectra, SC-XRD data, and in situ PXRD analysis indicate that CO2 molecules in CO@JUK-8ip are confined in 0-D pores (Figure S8). Notably, the CO2 adsorption mechanism corresponds to that previously reported for water vapor;[24] H2O molecules in the ip phase are localized in the vicinity of −C(O)=N–NH– and −COO– groups (Figure S5). During stepwise pressure increase (p/p0 = 0.09–0.99), the unit cell parameters rapidly change and the intensity maximum, i.e., the principal value δ⊥ of signal I (p/p0 =1), shifts from ∼178 to 170 ppm and the effective chemical shift anisotropy Δσ narrows from 110 ppm down to about 80 ppm (Figure ). Furthermore, at the gop of p/p0 = 0.19, an additional signal (signal II) appears. Its effective CSA has the opposite sign as signal I. The intensity maximum δ⊥ occurs initially at 120 ppm and shifts to lower values at increasing pressures. This second signal becomes more intense during the adsorption, and both signals have comparable intensity at p/p0 = 0.35. In the low-pressure regime (p/p0 = 0.09–0.30), signal II is relatively narrow. At further increasing pressure, the line becomes broader and transforms into the above-described CO2 tensor spectrum with an effective Δσ of −70 ppm at p/p0 ∼ 0.99.

To understand subtle local MOF–adsorbate interactions, we also calculated isotropic chemical shift (ICS) for both signals. In the case of signal I (p/p0 = 0.24), an ICS of 144 ppm is determined. This is higher than the value of 127 ppm measured for free bulk gas. The ICS of 121 ppm is obtained for signal II (p/p0 ∼ 0.99), which is close to the value for free gaseous CO2.[51] This observation proves the existence of two chemically different CO2 states inside the pore system and supports the IR data described above (Figure ). The considerably increased isotropic chemical shift of the first species of CO2 (signal I) compared to that of the second (signal II) indicates that the gas molecules in the MOF exhibit different chemical environments. Taking into consideration the IR, PXRD, and SC-XRD studies, CO2 molecules are expected to be adsorbed in the vicinity of acylhydrazone pockets (Figure S5). Two factors have a simultaneous impact on the chemical environment: the pore confinement and amount of adsorbed CO2 molecules.

After framework opening, further uptake causes an intensity increase of signal II. These species exhibit weaker CO2 interactions with the framework. Increasing gas uptake influences the width of both signals in the opposite direction. Signal I becomes narrower and signal II becomes broader (Figure ; dashed lines). These data indicate that the mobility of the weaker adsorbed CO2 (signal II) is more restricted at high pressure (p/p0 = 0.99), in contrast to that of the strongly interacting species (signal I). This can be explained by the fact that CO2 molecules causing signal II inside the partly filled, open pores of CO@JUK-8op are more mobile than at a higher degree of pore filling. On the other hand, higher pressure increases the distance between the adsorbate and functional groups in pockets, enhancing the motional freedom of the confined CO2 molecules causing signal I.

In summary, breathing and swelling change the spatial arrangement of the framework, thus considerably influencing the number and mobility of the adsorbed CO2 species in JUK-8.

During desorption, the reverse transition mechanism is observed. However, hysteresis occurs due to MOF···CO2 and CO2···CO2 interactions. Below p/p0 = 0.08, the framework transforms in the CO@ip state (Figure ); signal I dominates the spectrum and becomes stepwise broader, finally reaching the initial value of 110 ppm at p/p0 = 0.03. Unit cell volume contraction of JUK-8 during desorption decreases the distance between CO2 molecules to −C(O)=N–NH– functional groups and again immobilizes the adsorbate causing the appearance of signal I. The described in situ NMR signals are fully reproduced even after 3 cycles, which proves that the CO2 environment during the adsorption is independent of the cycling experiment (Figure S14).

Figure 4

Single-gas adsorption measurements for JUK-8: CO2 (195 and 298 K; red), Ar (84 K; green), O2 (90 K; gray), N2 (77 K; blue), and CH4 (298 K; black). Full symbols, adsorption and open symbols, desorption.

Single- and Mixed-Gas Adsorption Properties in a Broad Range of Temperature

Single-gas adsorption isotherms of Ar, N2, O2, and CO2 gases, measured in the low-temperature regime, suggest high selectivity toward carbon dioxide (Figures and S15). Thermally activated MOF, JUK-8cp, does not adsorb nitrogen (77 K), argon (84 K), and oxygen (90 K). The total uptake at p/p0 = 0.99 is equal to 6, 6, and 12 cm3 for N2, Ar, and O2, respectively. On the other hand, the activated pristine material adsorbs 144 cm3/g CO2 (195 K) at 0.99 bar (p/p0 = 0.99) with gate opening pressure at p = 0.08 bar (p/p0 = 0.08). Low-temperature adsorption studies indicate the potential applicability of JUK-8 in gas-related technologies. To assess this, we measured high-pressure single-component CH4 and CO2 isotherms at an ambient temperature (298 K). Methane, similarly to O2, N2, and Ar at low temperature, does not open JUK-8cp, and the total uptake of CH4 at a very high pressure (∼32 bar) is considerably low (7 cm3/g) compared to that of CO2 at similar conditions (108 cm3/g at ∼32 bar and 298 K; Figure ). On the other hand, due to the different thermodynamic conditions, the absolute value of gop shifts from 0.08 bar (195 K, Figure S15) to 11.56 bar (298 K).

From a thermodynamic point, the flexible MOFs are not inert to adsorbates and their adsorbing specific areas change during adsorption.[56] Thus, to prove the preferable adsorption of CO2, we used CO2/CH4 selectivity factor S (eq ) instead of the ideal adsorbed solution theory (IAST), recommended for rigid materials. For the 27.7% content of CH4 in a single-component experiment, we have obtained a very good S = 17.1 at 298 K (eq )Here, nCO and nCH denote the number of moles adsorbed under the specified gas composition (nCO and nCH; pCO =15.42 bar and pCH = 5.92 bar). It is noteworthy, however, that the calculation based on one-component isotherms may not correspond to real multicomponent adsorption. Recently, a few reports have investigated flexibility during multicomponent experiments;[25,26,57] however, this type of investigation is still rare and sought after. There are still open questions. Does selectivity for a gas pair arise from the weak affinity of a MOF toward one component? Does the open phase coadsorb a gas that normally does not interact with the closed-pore phase? And what is the influence of temperature? Taking these issues into consideration, we carried out mixed-gas coadsorption experiments for JUK-8 and CO2/CH4 (75:25 v/v) gas mixtures at different (288, 293, and 298 K) temperatures (Figure ).

Figure 5

Isothermal multicomponent adsorption experiments for CO2/CH4 (75:25 v/v) mixtures in JUK-8 material for three different temperatures. Blue triangles, total adsorbed volume; red circles, partial CO2 adsorbed volume; and black squares, partial CH4 adsorption.

The narrowest pore diameter of JUK-8op is approx. 4.1 Å, which indicates that methane, whose kinetic diameter is 3.8 Å, could be coadsorbed. However, regardless of temperature, CH4 does not enter one-dimensional channels and at pCH ∼ 5 bar, JUK-8 adsorbs only 3–7 cm3 of methane. Furthermore, S values calculated from coadsorption of the CO2/CH4 (75:25 v/v) at ∼20 bar are equal to 8.83, 9.48, and 1.99 for 288, 293, and 298 K, respectively (Figure S16); the obtained values are consistent with one-component isotherms for CH4 and CO2 at 298 K (S = 17.1). This indicates that JUK-8 shows selectivity factor higher than zeolites and activated carbon and is comparable to MIL-125 and its derivatives (Table S5).

Pure CO2 opens the framework at p ≈ 11.56 bar (298 K), while the presence of CH4 considerably increases gop to 15.70 bar (298 K). This effect is caused by gas competition. In the studied temperature range, we have observed a strong linear relationship (pgop =0.59–160T; R2 = 0.999) between gop and temperature (Figure S17). In the methane-rich atmosphere, CO2 opens the framework at 9.75, 12.66, and 15.70 bar for 288, 293, and 298 K, respectively. On average, each increase of temperature by 5 K causes the increase of gop by 3 bar. Furthermore, independent of temperature and type of experiment, the amount of adsorbed CO2 before gop is almost constant (18–22 cm3·g–1; [Zn(oba)(pip)]·1/2CO2), which indicates that the mechanism of the transition does not depend on the studied conditions.

To characterize the selectivity in the low-temperature regime, we performed an in situ13C NMR coadsorption experiment using a 13CO2/13CH4 mixture (molar ratio 1:1) at 195 K (Figures and S18). At a total pressure of 0.73 bar, we observed two narrow gas-phase signals resulting from gaseous CH4 and CO2 at approx. −10 and 127 ppm, respectively.[58,59] There is no evidence for adsorbed CH4 at this pressure. However, the gas-phase signal of CO2 is weaker than that of CH4, which indicates the presence of adsorbed CO2. However, the broad signals of adsorbed CO2 are not detectable at such low pressures. After stepwise increasing the total pressure, signal II of adsorbed CO2 with an effective CSA of ca. −70 ppm finally dominates the spectrum in analogy to the single-component adsorption experiment described above. Since the CO2 signal shapes are not significantly different from the single-phase adsorption studies, it is concluded that the adsorption and switching mechanism are not significantly influenced by the presence of CH4. At a maximum pressure of 5.50 bar, a relatively weak signal at −4 ppm becomes detectable, which is caused by the adsorbed methane. It means that only minor amounts of methane coadsorb on JUK-8op@CO even at low temperature and high pressure. The intensity of the signal due to adsorbed methane only corresponds to 0.6% of the signal II of adsorbed CO2. This is 1 order of magnitude less than the amount measured by mixed-gas coadsorption experiments (2.2–8.3%). The selectivity factor (nCH·nCO)/(nCH·nCO) calculated from these in situ NMR data at 5.50 bar amounts to 160. During pressure release, the MOF first releases CH4. At 1.92 bar, no adsorbed methane and only small amounts of free gas are observable. Further pressure reduction results in a decreasing intensity of the signals of free and adsorbed CO2 molecules and the MOF switches back into the intermediate phase.

Figure 6

In situ13C NMR spectra of JUK-8 during adsorption of the gas mixture (CO2 and CH4, molar ratio 1:1, 195 K) at selected pressures (top) and 13C spectrum of the 13CO2/13CH4 gas mixture at 5.50 bar (bottom). The signal at −10 ppm (gray bar) corresponds to gaseous methane, at −4 ppm to the adsorbed methane, at 127 ppm to the gaseous carbon dioxide (orange bar), and the broad signal II between 90 and 180 ppm to the adsorbed carbon dioxide. Signal II gradually transforms into the even broader signal I during pressure release, as can be seen in the spectra shown on top as well as in Figure .

The selectivity factor of 160 calculated from 13C in situ NMR data during adsorption is considerably higher than the value obtained from single- and multicomponent volumetric adsorption measurements (1.99–17.1). The observed difference can be explained by the following aspects influencing eq . (i) The maximum uptake of CO2 at 298 K (108 cm3/g at STP) is considerably lower than at 195 K (144 cm3/g at STP), i.e., nCO2ads at 195 K is higher compared to those at 288, 293, and 298 K (Figure and 5). (ii) The accuracy of the in situ NMR, compared to the volumetric experiment, is higher, thus enabling a lower detection limit for adsorbed methane. In the NMR experiments, we detected very low amounts of the labeled methane (nCH), as can be seen in Figure . (iii) The initial CO2/CH4 ratio, equal to 1:1 (50:50 v/v) in the 13C NMR study, is different from 3:1 (75:25 v/v) in the multicomponent volumetric studies.

Stability in Water

Apart from a structural flexibility, a promising adsorbent for gas-related technologies has to meet other criteria such as stability to impurities,[60,61] selectivity, endurance for repeatable adsorption–desorption stress,[62] and lack of adsorption hysteresis. In the previous work, it has been shown that JUK-8 is chemically stable, e.g., immersed in water for 24 h does not change its adsorption properties (shape of isotherm and total capacity) and withstands repeatable vapor adsorption/desorption.[24] Herein, we repeatedly (hundred times) immersed JUK-8 in water that was subsequently evaporated by heating at a lowered pressure. We monitored the process by powder X-ray diffraction and infrared spectroscopy, and both techniques provided evidence that the structure of JUK-8 remains intact (Figure S19). In summary, JUK-8 is very stable, highly selective toward CO2, and resistant to repeatable adsorption/desorption stress. However, it has disadvantages such as high gate opening pressure at a working temperature, moderate CO2 uptake, and hysteresis.

Conclusions

In this work, we have determined the crystal structure of a desolvated closed-pore phase (JUK-8cp) of a flexible water-stable metal–organic framework JUK-8, which enables a detailed insight into the mechanism of its phase transitions caused by thermal and pressure stimuli. The removal of solvent molecules from JUK-8 practically does not change the relative positions of its component subnetworks, whereas the structural transition involves rearrangements around zinc cations with slight bending of the carboxylate linkers. The structural analysis carried out by in situ PXRD during the adsorption of CO2 indirectly indicates the existence of two minima in the free-energy profile of the investigated MOF, while in situ IR and NMR spectroscopies uncover preferential positions of the adsorbed CO2 molecules. The detailed analysis of one- and multicomponent equilibrium adsorptions at a broad temperature range demonstrates that JUK-8cp is a highly selective adsorbent of CO2 from CO2/CH4 mixtures. In summary, this work provides versatile insights toward the understanding of adsorbate–flexible MOF interactions, which is essential for further development of high-performance materials that could meet the expectations of energy-efficient industry.

Experimental Section

Synthesis

JUK-8 and 4-pyridyl-functionalized benzene-1,3-dicarbohydrazide (pip) were prepared according to the published method.[24] All other reagents and solvents were of analytical grade (Sigma-Aldrich, POCH, Polmos) and were used without further purification.

Single-Crystal X-ray Diffraction

Due to high affinity of JUK-8cp toward water,[24] a suitably sized single crystal of desolvated JUK-8cp was picked up in a glovebox (MBRAUN) equipped with the Leica microscope (Figure S6). The crystal was closed in a priori activated (453 K for ∼2 h) borosilicate glass capillary (d = 0.3 mm). The data set was collected at the BESSY MX BL14.3 beamline of Helmholtz-Zentrum Berlin für Materialien und Energie.[63] Monochromatic X-ray radiation with a wavelength of λ = 0.8939 Å (E = 13870 eV) was used in experiments. The data set was collected at 100 K. The crystal symmetry and scan range were determined in each particular case using the iMosflm program.[64] The φ-scans with an oscillation range of 1° were used for data collection. For each data set, 180 images were collected to reach the maximal possible completeness. The data set was processed in the automatic regime using XDSAPP 2.0 software.[65] The crystal structure was solved by direct methods and refined by full-matrix least squares on F2 using the SHELX-2018/3 program package.[66] All nonhydrogen atoms were refined in anisotropic approximation. Hydrogen atoms were refined in geometrically calculated positions using “riding model” with Uiso(H) = 1.2Uiso(C). CCDC2072669 contains the supplementary crystallographic data for JUK-8cp. Experimental data on single-crystal X-ray experiments are summarized in Table S1.

IR Spectra

IR spectra were recorded on a Thermo Scientific Nicolet iS10 FT-IR spectrophotometer equipped with an iD7 diamond ATR attachment.

IR

In situ IR spectra were recorded on a Bruker Tensor 27 spectrometer equipped with an MCT detector and working with a spectral resolution of 2 cm–1. Prior to the adsorption of CO2 at 195 K (Linde Gas Polska, 99.95% used without further purification), the samples were ground and activated in the form of self-supporting wafers for 2 h at 473 K.

Powder X-ray Diffraction

PXRD patterns were recorded at room temperature (295 K) on a Rigaku Miniflex 600 diffractometer with Cu Kα radiation (λ = 1.5418 Å) in a 2θ range from 3 to 45° with a 0.05° step at a scan speed of 2.5° min–1.

In Situ Powder X-ray Diffraction

PXRD patterns during the CO2 adsorption were measured at Helmholtz-Zentrum Berlin für Materialien und Energie on KMC-2 beamline. The detailed description of the measuring setup is provided in the literature.[34] Prior to experiments, the as-synthesized JUK-8op was ground (∼5 min in a mortar and pestle) and evacuated at 443 K for ∼16 h. For the part of JUK-8cp prepared in this way, a CO2 isotherm at 195 K was recorded (Figure S10). Due to the heterogeneous distribution of crystallographic orientations (texture) of a polycrystalline material, the degassed JUK-8cp was again ground before in situ experiments. PXRD patterns, measured during the adsorption and desorption of CO2 at 195 K, were indexed using the DICVOL program, integrated into the FullProf.2k V.6.30. Further, the Le Bail fit was performed to refine the unit cell and profile parameters (Figure S11).

In Situ NMR

13C NMR measurements of 13C-enriched CO2 and CH4 were carried out using a BRUKER Avance 300 spectrometer at 195 K combined with a homemade in situ high-pressure gas adsorption apparatus. The apparatus is equipped with a gas mixing chamber to produce the desired CO2/CH4 gas mixture with a molecular ratio of 1:1 (Figure S18). It was adjusted by first filling the chamber with 13CH4 up to a certain pressure and afterward adding 13CO2 up to the final pressure. The pressures were always well below the critical pressure for both gases. This allows us to consider the gases as ideal, i.e., the gas pressure is assumed to be directly proportional to the gas concentration/number density of the molecules. Temperature calibration was carried out using the well-known temperature dependence of the 1H NMR signal of methanol.[67] The solvent-free samples were transferred into a 5 mm single-crystal sapphire tube in an argon-filled glovebox. The sample was activated again in the tube under high vacuum for 2 h before the measurements. After pressurization, the samples were equilibrated at least for 30 min to reach thermal equilibrium. The pressure was incremented stepwise by adding the required portion of the initial gas mixture to the sample tube. To ensure equilibrium state after each pressure increase, a 15 min equilibrium phase was allowed. The 13C NMR spectra were recorded at a resonance frequency of 75.47 MHz under 1H-decoupling using a 10 mm double resonance probe head, a 10.7 μs pulse length for 13C, and with a relaxation delay of 5 s. The chemical shifts were referenced relative to ethylbenzene. For rigid CO2, the chemical shift anisotropy (CSA) tensor has an overall width of 335 ppm. To ensure excitation of the full width of this, a sufficiently short pulse length must be chosen. We decided to choose a pulse flip angle of 60° to avoid the excitation problem and to decrease the relaxation delay to only 4 times T1 for quantitative measurements. Under our conditions, T1 values below 1 s are observed in full agreement with the literature.[67−70] The chosen recycle delay of 5 s is thus safely longer than 5 times of T1 and ensures full relaxation of the spin system, i.e., saturation effects can be excluded. Moreover, the comparison of the 13C NMR spectra with and without 1H-decoupling shows that NOE effects have only negligible influence upon the signal intensities under the applied conditions (Figure S18).

Single-Component Gas Adsorption Measurements

Prior to the physisorption measurements, the as-synthesized JUK-8 was evacuated at 443–453 K for ∼16 h. The Ar physisorption experiment was conducted with an AUTOSORB-iQ-C-XR from Quantachrome at 87 K (cryostat). Nitrogen (77 K), carbon dioxide (195 K), and oxygen (90 K) adsorption/desorption studies were performed on a BELSORP-max adsorption apparatus (MicrotracBEL Corp.), connected to the closed cycle helium cryostat DE-202AG (ARS). The adsorption temperature was set by a temperature controller LS-336 (LAKE SHORE), and the heat produced by the cryostat was removed from the system by a water-cooled helium compressor ARS-2HW. The sample was placed in a custom-made cell consisting of a 3 cm long rod-shaped copper cell of 1 cm diameter, sealed by a copper gasket from the exterior with a copper dome and insulated by dynamic vacuum (p < 10–4 kPa), and connected to the BELSORP-max adsorption instrument with a 1/8 inch stainless steel capillary.

High-Pressure Single-Gas Adsorption and Mixed-Gas Coadsorption Experiments

Volumetric high-pressure single-gas and mixed-gas adsorption experiments were conducted using the BELSORP-VC (Microtrac MRB) instrument. Helium gas (99.999% purity) was used for the dead volume measurement. Carbon dioxide (99.999% purity) and methane (99.999% purity) gases were used in adsorption experiments. All gases were purchased from Praxair.

All isotherms were measured on the same sample (m = 0.6321 g). The sample was degassed in dynamic vacuum over 24 h at 463 K in the measurement cell. Single-gas adsorption isotherms were measured at 298 K in a pressure range of 52–4262 kPa for CH4 and 0.5–4207 kPa for CO2. Mixed-gas adsorption was measured using the gas mixture of 75% CO2 and 25% CH4 (v/v) at 298, 293, and 288 K. The gas mixture composition and adsorption temperatures were chosen because of the pressure limitation of the instrument for gas mixtures. The gas mixture was prepared directly in the standard volume part of the instrument from the pure gases for each point of the isotherm separately. After dynamic mixing of the gases for 60 min, the composition of the mixture was determined by the gas chromatograph Agilent 490 Micro-GC-System (GC), coupled to the instrument manifold. The gas mixture was further purged through the sample cell over 60 min, and the overall adsorbed amount was determined from the pressure drop, taking the nonideality correction for each mixture component into account. The composition of the gas mixture after adsorption was analyzed by GC. To increase the reproducibility of the measurements, five GC measurements were done before and after adsorption. The adsorbed amount of mixture components was calculated from the difference in the mixture composition before and after adsorption. Before the measurement of each adsorption point, the sample was degassed in the ultrahigh vacuum for 60 min at 298 K. For each temperature, 5–6 points were measured reaching the maximal equilibrium pressure for the gas mixture of 2366 kPa at 288 K, 2443.4 kPa at 293 K, and 3000 kPa at 288 K.

JUK-8op (m = 200 mg) was repeatedly (100 times) immersed in approx. 1.2 mL of distilled water followed by evaporation at 393 K and 450 mbar. After a few evaporation cycles, IR spectra and/or PXRD patterns of the residue were recorded (Figure S19).