Yin
Rao
a,
Xinhao
Li
b,
Haozhi
Xi
a,
Zhongwen
Jiang
a,
Wangzhi
Li
a,
Han
Zhou
c,
Yanping
Zhang
c,
Chuixiong
Wu
c,
Yue-Biao
Zhang
b and
Qiaowei
Li
*a
aDepartment of Chemistry, State Key Laboratory of Porous Materials for Separation and Conversion, Collaborative Innovation Center of Chemistry for Energy Materials, Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, Fudan University, Shanghai 200433, P. R. China. E-mail: qwli@fudan.edu.cn
bSchool of Physical Science and Technology, Shanghai Key Laboratory of High-Resolution Electron Microscopy, State Key Laboratory of Advanced Medical Materials and Devices, ShanghaiTech University, Shanghai 201210, P. R. China
cSuzhou MiQro Era Quantum Technology Ltd, Jiangsu 215011, P. R. China
First published on 10th March 2025
Separating propylene from ethylene in the product mixture derived from the methanol-to-olefin conversion process in an energy-efficient way is challenging, and adsorptive separation utilizing porous materials such as metal–organic frameworks (MOFs) provides a potential solution. Herein, we constructed a highly connected MOF structure with “stretched” [Zn8SiO4] as metal nodes and mixed binding groups (carboxylate and pyrazolate) as linkers. The mixed binding groups adapted well to the increase in the inter-metal distance caused by the insertion of SiO44− in the nodes, enabling the formation of a highly symmetrical 12-connected net with high porosity and suitable pore apertures. This framework exhibited an impressive C3H6 adsorption capacity of 180.5 cm3 g−1 and a calculated C3H6/C2H4 selectivity of 8.6 at 298 K and 100 kPa, which was attributed to the abundant N/O sites on the pore surface that generated stronger interactions with C3H6. Furthermore, dynamic breakthrough experiments demonstrated that the framework was capable of effectively separating C3H6/C2H4 under various conditions, making it a promising benchmark adsorbent for industrial applications.
MOFs offer significant advantages in terms of tunable pore sizes and functionalities,11–19 making them particularly suitable for adsorptive separation and purification.20–26 While MOFs have been widely used to separate various gas mixtures, the separation of C3H6 and C2H4 is still a great challenge. Their kinetic diameters are very similar (C3H6, 4.68 Å; C2H4, 4.16 Å),22 making their size-/shape-selective separation difficult. Therefore, designing differentiated interaction sites in the porous MOFs is a prerequisite for achieving a high separation selectivity.27,28 On the one hand, higher polarizability of C3H6 (62.6 × 10−25 cm3) suggests stronger dispersion forces with the framework than C2H4 (42.5 × 10−25 cm3).22 On the other hand, the interaction difference can be further amplified by introducing more weak-to-medium interactions, such as C–H⋯O/N hydrogen bonds and C–H⋯π interactions, between the framework and the guests.29,30
Herein, we synthesized an MOF structure by the reticulation between a relatively rare [Zn8SiO4]-based secondary building unit (SBU) and 4-(1H-pyrazol-4-yl)benzoic acid (H2PyBC). Compared with the traditional Zn–carboxylate MOFs, the SiO44− in the SBU centre “stretched” the SBU size, and the mixed binding groups (carboxylate and pyrazolate) prompted its pointing direction to take on a cuboctahedron geometry, which further guided the overall framework into an fcu net with 12-connected (12-c) nodes. The framework exhibited a Brunauer–Emmett–Teller (BET) surface area of 1965 m2 g−1, with the pore size concentrated at 7.9 and 13.0 Å. Interestingly, the C3H6 adsorption capacity of the MOF reached 180.5 cm3 g−1 at 298 K and 100 kPa, while the C2H4 uptake was merely 68.2 cm3 g−1, indicating a stronger interaction between C3H6 and the framework. Grand canonical Monte Carlo (GCMC) simulations revealed a binding energy difference of 10.11 kJ mol−1 for C3H6 and C2H4 when they interacted with the strongest binding sites on the backbone. Furthermore, dynamic breakthrough experiments concluded that for an equimolar C3H6/C2H4 mixture, the breakthrough time interval reaches 41.3 min at 298 K and maintains 35.3 min at 328 K, suggesting the potential of applying this MOF for C3H6/C2H4 separation.
Specifically, Zn(NO3)2·6H2O (2340 mg, 7.87 mmol), H2PyBC (168 mg, 0.89 mmol), and 1H-benzotriazole (HBTA, 220 mg, 1.85 mmol) are dissolved in 34.0 mL N,N-dimethylformamide (DMF) in a 100 mL pressure-resistant tube with a Teflon cap. 6.0 mL aqueous Na2SiO3 solution (0.015 M) is further added. After sonication, the tube with the clear solution is placed in an oven at 115 °C for 12 h, and colorless octahedral crystals are obtained. Yield: 73.0 mg, i.e., 47% based on Na2SiO3.
Although two MOFs with the same [Zn8SiO4] nodes have been previously reported,31,32 each node in the FDM-201 is connected to twelve neighbouring nodes, while formerly it tended to connect with only 6 or 8 nodes through the linker pairs (Fig. 2),31 and the binding groups were pure carboxylate rather than mixed carboxylate/pyrazolate. Compared with the conventional Zn4O(COO)6 (ref. 18) and Zn(NN)2 (ref. 33 and 34) SBUs (Fig. 2a), the inter-metal distances in the [Zn8SiO4] nodes are lengthened because of the insertion of SiO44− in the SBU centre, making the SBU “stretched”. Moreover, in order to cope with this SBU stretch, the metal–linker coordination needs to be distorted. Compared to the metal–pyrazolate linkage in FDM-201, the metal–carboxylate linkage in the previous example31 shows a higher degree of flexibility (Fig. 2b and c), causing the O–Zn⋯SiO⋯Zn–O moiety to deviate from a planar configuration when the Zn⋯Zn distance is lengthened. As a result, the “pairing” of neighbouring ligands in parallel directions in the pure carboxylate version is evident, making the SBU six-connected. With less distortion in the N–Zn⋯SiO⋯Zn–N, the configuration of the SBU with mixed carboxylate/pyrazolate adopts a cuboctahedral geometry in FDM-201. Overall, the binding group difference (carboxylate vs. pyrazolate) imposes different distortion extents in the metal coordination sphere and consequently influences the connecting directions of the stretched SBUs.
![]() | ||
Fig. 2 Control of distinct coordination linkages giving rise to [Zn8(SiO4)(C8H4O4)6]n31 and FDM-201 with different topologies. (a) Typical Zn⋯Zn distance in the conventional Zn4O(COO)6 and Zn(NN)2 units. (b) SBU in [Zn8(SiO4)(C8H4O4)6]n can be viewed as 6-c because the twelve linkers are organized into six pairs owing to the distortion in the O–Zn⋯SiO⋯Zn–O moiety. (c) SBU in FDM-201 showing cuboctahedron geometry, in which the O–Zn⋯SiO⋯Zn–O moiety is coplanar. |
The powder X-ray diffraction (PXRD) pattern of the as-synthesized FDM-201 matches with the corresponding simulated pattern, confirming its crystallinity and phase purity (Fig. 3a). The as-synthesized sample can be successfully activated by solvent exchange with acetone, followed by heating at 80 °C under vacuum. The sharp peaks in the PXRD pattern match well with those from the corresponding as-synthesized sample, indicating that it maintains high crystallinity after activation (Fig. 3a). Furthermore, FDM-201 possesses outstanding thermal stability, with no significant weight loss observed in the thermogravimetric analysis (TGA) until it reaches 450 °C in an air atmosphere (Fig. S6†). The crystals retain sharp PXRD peaks after being immersed in water for 6 h (Fig. S7†), although soaking for a longer time would induce a structural transformation. A diffraction peak (2θ = 6.9°) corresponding to a new condensed structure with a formula of Zn2PyBC(OH)2 (Fig. S2, S8, and S9, see the discussion in the ESI†) emerges in the PXRD after being immersed in H2O for 12 h, and the peak becomes more evident after 1 day (Fig. S8†). Immersing the crystals in boiling water would accelerate the structural transformation, as indicated by the completely new phase after 1 day (Fig. S8†). However, after soaking the FDM-201 crystals in a mixed H2O/DMF solvent (v/v, 4:
1) for 7 days, we found that the structure is retained (Fig. S7†).
The crystal structure of FDM-201 reveals that it features two types of cages: an octahedral cage with an inscribed sphere diameter of 13.4 Å and a tetrahedral cage with 5.4 Å in diameter, with the assumption that the linkers do not rotate. Based on Zeo++ calculation,35 the simulated surface area of FDM-201 is 1898 m2 g−1, and its simulated pore volume is 0.70 cm3 g−1. Five parallel 87 K Ar adsorption isotherms are measured to experimentally verify the consistent and permanent porosity of FDM-201 (see Fig. 3b for one isotherm and Fig. S10† for all isotherms). The isotherms exhibit typical type I adsorptions, with a maximum Ar adsorption capacity of 583 ± 6 cm3 g−1 at P/P0 = 0.99. The BET surface area and pore volume of FDM-201 are calculated to be 1965 ± 37 m2 g−1 and 0.74 ± 0.01 cm3 g−1, respectively. The pore size distribution of FDM-201 is further analysed using the isotherm in Fig. 3b and a kernel based on the nonlocal density functional theory (NLDFT). The analysis reveals that the pore size distribution spans two intervals, from 6.1 to 10.5 Å and from 10.5 to 16.2 Å, with predominant concentrations at 7.9 and 13.0 Å, respectively (Fig. 3c). These values align approximately with theoretical predictions, indicating that the pores within the periodic and robust framework remain highly accessible after the activation.
The high BET surface area and pore volume of FDM-201, coupled with the abundant O/N interaction sites on the pore surfaces, intrigued us to explore its potential application in gas adsorption and separation. For gas separation, not only the pore size but also their apertures are critical. Analysis of the FDM-201 crystal structure further reveals that the octahedral and tetrahedral cages have relatively small apertures (triangles with an incircle of 3.9 Å) and make the pores inaccessible to C3H6 (kinetic diameter, 4.68 Å) and C2H4 (kinetic diameter, 4.16 Å). However, it should be noted that the benzene ring in the PyBC linker can easily rotate at ambient temperature,36 leading to a “widening” of the pore aperture and an increase in the pore size. After considering the linker rotation, the sizes of the octahedral cage, tetrahedral cage, and pore aperture become 13.4, 8.3, and 6.2 Å. As a result, C3H6 and C2H4 can indeed be adsorbed by the FDM-201 framework.
To investigate the C3H6 and C2H4 adsorption capability of FDM-201 in experiments, single-component adsorption isotherms at 273, 298, and 328 K are collected (Fig. 4a and b). FDM-201 exhibits an adsorption uptake of 180.5 cm3 g−1 for C3H6 at 298 K and 1 bar, much higher than that of the C2H4 case (68.2 cm3 g−1). In addition, the adsorption capacity of C3H6/C2H4 is 200.5/108.4 cm3 g−1 at 273 K and 137.4/31.6 cm3 g−1 at 328 K. Remarkably, its C3H6 adsorption capacity at 298 K (180.5 cm3 g−1) is among the many high-performance MOFs, such as spe-MOF (236.9 cm3 g−1),37 FJI-H8-Me (211.0 cm3 g−1),28 NEM-4 (197.4 cm3 g−1),38 MAC-4 (127.0 cm3 g−1),39 and CoV-bco-tpt (110.0 cm3 g−1)40 (see Table S7 in the ESI† for a summary of the C3H6 adsorption capacity of benchmark MOFs). The higher uptake of C3H6 than that of C2H4 under the tested temperature range suggests that FDM-201 is a potential candidate for C3H6/C2H4 separation.
Isosteric heat of adsorption (Qst) is calculated based on the single-component gas adsorption isotherms at 273 and 298 K, employing the Clausius–Clapeyron equation (Table S4 and Fig. S11 and S12, see the ESI†). At zero coverage, the Qst values of FDM-201 for C3H6 and C2H4 are 27.1 and 19.1 kJ mol−1, respectively (Fig. 4c). The notably higher Qst for C3H6 confirms stronger interaction between C3H6 and the framework, and it is preferably adsorbed (see Table S5† for a summary of Qst values of benchmark MOFs). It is worth noting that the Qst value for C3H6 at high coverage is apparently higher than the value at zero coverage, indicating that at a high-pressure range, the C3H6–C3H6 interaction is stronger than the C3H6–framework interaction.
To preliminarily evaluate the C3H6/C2H4 separation performance of FDM-201, the ideal adsorption solution theory (IAST) is employed to calculate the selectivity (Table S6 and Fig. S13†). For an equimolar C3H6/C2H4 mixture at 100 kPa, the calculated selectivities at 273, 298, and 328 K are 11.9, 8.6, and 7.7, respectively (Fig. 4d). The selectivity at 298 K is comparable with that of the current outstanding materials, such as HOF-NBDA (11.1),41 MAC-4 (9.5),39 and JLU-MOF132 (9.2),42 (Table S7†). Combined with the high adsorption capacity (C3H6, 180.5 cm3 g−1), FDM-201 appears to be a fairly good option among the known materials for C3H6/C2H4 separation (Fig. 4e and Table S7†). Impressively, the selectivity only drops slightly to 7.7 at 328 K, suggesting a high selectivity even at a relatively elevated temperature proximate to the actual operating condition.43 The IAST selectivity of FDM-201 for the C3H6/C2H4 mixtures at 20/50 and 10/90 (v/v) ratios are further calculated,37,44 and the framework maintains a high selectivity of 14.1/9.4/7.3 for C3H6/C2H4 (v/v, 20/50) at 273/298/328 K, and 14.8/9.4/7.2 for C3H6/C2H4 (v/v, 10/90) at these three temperatures (Fig. S14 and Table S8†).
Separation potential (Δq) is a comprehensive metric that combines both adsorption capacity and selectivity to evaluate the separation ability of materials.45 For C3H6-selective adsorbents, Δq presents a reliable prediction of the maximum amount of pure C2H4 recovered during a fixed-bed separation process. According to the IAST results on equimolar mixtures, Δq is calculated to be 7.42, 6.00, and 3.73 mmol g−1 at 273, 298, and 328 K, respectively (Fig. 4f). The Δq value for FDM-201 at 298 K is much higher than some benchmark C3H6/C2H4 separating-MOFs, such as MAC-4 (4.4 mmol g−1),39 Zn-BPZ-TATB (3.74 mmol g−1),30 Zn-BPZ-SA (1.92 mmol g−1),46 and Zn2(oba)2(dmimpym) (3.04 mmol g−1).47
To gain a deep understanding of the interactions between the framework and the adsorbates, GCMC simulations are employed to investigate the preferential binding sites within the octahedral cage of FDM-201 for C2H4 and C3H6. For both gases, the strongest adsorption sites, with multiple supramolecular interactions between the backbone and the guests, are located near the [Zn8SiO4] core (Fig. 5). Specifically, three C–H⋯O/N hydrogen bonds (C⋯O/N bond length: 3.41–4.02 Å), two C–H⋯π interactions (3.72–3.79 Å) and one π⋯π interaction (3.85 Å) exist between one C2H4 and this site in FDM-201. The calculated binding energy is 38.61 kJ mol−1. On the other hand, stronger interaction with four C–H⋯O/N hydrogen bonds (3.45–4.03 Å), two C–H⋯π interactions (3.63–3.73 Å), and one π⋯π interaction (3.82 Å) enabled the binding energy of C3H6 to this site to be 48.72 kJ mol−1. Based on the simulation, the difference in the static binding energies between C3H6 and C2H4 is 10.11 kJ mol−1, which is roughly in agreement with the difference in the Qst calculated from the adsorption isotherms (8.0 kJ mol−1) and corroborates with the preferred C3H6 adsorption in FDM-201.
![]() | ||
Fig. 5 Interactions between the preferential binding site and (a) C2H4 and (b) C3H6. Atom colour code: Zn, light blue; Si, yellow; O, red; N, dark blue; C, grey; H, pink. |
To verify the actual C3H6/C2H4 separation performance, dynamic breakthrough experiments for binary C3H6/C2H4 (v/v, 50/50) mixtures are conducted at 298 and 328 K. Before the experiments, the acetone-exchanged sample is activated under vacuum at 80 °C for 12 h. Subsequently, it is packed into a stainless-steel column (l = 15 cm, d = 0.42 cm). The packed column is firstly purged with He at a flow rate of 20 mL min−1 for 30 min. At 298 K, under a mixed gas flow (flow rate, 5 mL min−1), C2H4 first elutes from the column at 23.5 min, whereas C3H6 is under detection limit until 64.8 min (Fig. 6a). The breakthrough time interval, during which high-grade C2H4 can be collected, is calculated to be 41.3 min. At 328 K, FDM-201 can still efficiently separate C3H6 and C2H4, with an interval of 35.3 min (Fig. 6b). Furthermore, when the feed gas composition is adjusted to 20:
50
:
30 (v/v/v, C3H6 to C2H4 to He, with the C3H6/C2H4 ratio close to the application scenario), FDM-201 continues to exhibit effective separation performance with the interval time being 82.4 and 53.0 min at 298 and 328 K, respectively (Fig. 6c and d). The breakthrough experiments have proved that FDM-201 is a promising candidate for highly efficient propylene/ethylene separation across different composition ratios and different temperatures.
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2418870 and 2418871. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d5ta00771b |
This journal is © The Royal Society of Chemistry 2025 |