Linker desymmetrisation directs low-polar cages in an anion-pillared MOF for acetylene and ethylene purification from ternary mixtures

Abstract

Anion-pillared metal–organic frameworks (APMOFs) constitute a promising class of porous adsorbents, yet precise organization of anionic groups to maximize their strong electronegative character while circumventing highly polar pore environments and high adsorption enthalpies remains a substantial challenge. Here, we report a novel APMOF, CTGU-45, featuring an atypical low-polar cage surface distinct from reported APMOFs and then explore its C₂H₂ and C₂H₄ purification from ternary C₂H₂/CO₂/CH₄ and C₂H₂/C₂H₄/C₂H₆ mixtures. Theoretical calculation results revealed the mechanisms of simultaneous recognition of C₂H₂ and C₂H₆ molecules from C₂ hydrocarbons: the SiF₆²⁻ pillars provide strong binding sites for C₂H₂ via C–H⋯F hydrogen bonds, while the TIB-derived low-polar surface offers an optimal environment for C₂H₆ adsorption through synergistic C–H⋯π and C–H⋯N interactions. Dynamic breakthrough experiments further confirm the efficient separation of C₂H₂/CO₂, C₂H₂/CO₂/CH₄, and notably one-step production of polymer-grade C₂H₄ (≥99.95%, 9.57 L kg⁻¹) from ternary C₂H₂/C₂H₄/C₂H₆ mixtures with low regeneration energy. Crucially, CTGU-45 can be synthesized on a 500 mL scale with retained performance, underscoring its practical feasibility and laying a solid foundation for future industrial applications.

---

Introduction

Acetylene (C₂H₂) and ethylene (C₂H₄) are pivotal feedstocks in the chemical industry.¹ C₂H₂ is typically generated by partial oxidation of methane (CH₄) or thermal cracking of hydrocarbons, which co-produces significant amounts of carbon dioxide (CO₂).^2–5 Similarly, C₂H₄ is obtained via steam cracking or ethane (C₂H₆) dehydrogenation, where C₂H₆ and trace C₂H₂ will unavoidably be produced as by-products.^6,7 These impurities severely compromise the purity and productivity of polymer-grade C₂H₄ or high-purity C₂H₂, necessitating highly efficient separation processes.^8,9 However, the similar physicochemical properties of these gas molecules make separations (C₂H₂/CO₂/CH₄, C₂H₂/C₂H₄/C₂H₆) particularly difficult (Table S2). Conventional methods like catalytic hydrogenation and cryogenic distillation are complex and energy-intensive, thus driving the urgent need for efficient, alternative separation technologies.^10–12

Adsorptive separation using porous materials presents a promising, energy-efficient alternative to traditional cryogenic distillation.^13–16 Among various adsorbents, metal–organic frameworks (MOFs) have garnered immense attention due to their designable pore structures and tunable surface functionalities. In particular, anion-pillared MOFs (APMOFs) have emerged as an important subclass of MOF adsorbents. These materials utilize inorganic anions (e.g., SiF₆²⁻, GeF₆²⁻, NbOF₅²⁻, MoO₄²⁻) as structural pillars to construct robust frameworks with strong electrostatic potential fields and abundant electronegative atoms that act as hydrogen-bond (H-bond) acceptors, enabling highly selective recognition of C₂H₂ with strong positive H atoms and C₂H₆ with more H atoms. Initially, APMOFs are constructed from linear bidentate ligands to form one-dimensional (1D) channels.^17–20 The subsequent introduction of tri- and tetra-dentate ligands created more open APMOFs with a “cage + window” structure, which provides high capacity from large cages and high selectivity from narrow windows. Certainly, these APMOFs have been considered as valuable candidates capable of addressing the “trade-off” between capacity and selectivity in gas separation.

To date, only three highly symmetrical tridentate modules (Tripp, 2,4,6-tris(4-pyridyl)pyridine; TPA, tri(pyridin-4-yl)amine; and TPT, 2,4,6-tri(4-pyridyl)-1,3,5-triazine) have been successfully employed to construct APMOFs and used for gas separation, such as C₂H₂/CO₂, CO₂/N₂, C₃H₄/C₃H₆, Xe/Kr, and C₆ hydrocarbons. These ligands share critical structural features: their terminal coordinating units are all pyridine rings, which enforce a nearly ideal equilateral triangular arrangement of the three coordination N atoms. Although limited rotation of the pyridine rings is possible, the spatial positions of these key N donors remain essentially fixed, enabling a predictable assembly of frameworks. Interestingly, replacing the terminal pyridyl moiety with less symmetric imidazole rings would introduce significant conformational flexibility. The free rotation around the C–N single bonds linking the central benzene ring to the imidazole arms results in variable spatial orientations of the coordination N atoms, thereby breaking the inherent topological symmetry and creating opportunities for constructing APMOFs with unique pores. Moreover, certain challenging separations, e.g., one-step C₂H₄ purification from ternary C₂ mixtures, are seldom realized in APMOFs owing to intrinsic polarity mismatches, suggesting that a reconfiguration of functional sites within APMOFs is required. Since the coordination geometry and configuration of inorganic anions are relatively well-defined, the linker desymmetrisation offers a powerful means to systematically reorganize the functional sites within APMOFs. By changing the symmetry of the organic linkers, it becomes possible to create frameworks in which the chemical environment around each coordination site can be independently tuned. This controlled asymmetry allows for the introduction of complementary binding sites, modulation of local polarity, and the rational design of preferential adsorption domains, ultimately enhancing the selectivity and efficiency of challenging gas separations.

Keep this in mind, herein, by utilizing a tridentate TIB linker (1,3,5-tri(1H-imidazol-1-yl)benzene) with C₃ symmetry, a novel SiF₆²⁻ anion hybrid Cu-MOF (SIFSIX-Cu-TIB, CTGU-45, CTGU = China Three Gorges University) has been successfully constructed. Single-crystal X-ray diffraction (SCXRD) analysis reveals that the activated CTGU-45 features uniform cage-like pores with an internal cage diameter of approximately 7.6 Å and a pore window size of about 6.5 Å. In contrast to most reported cage-structured APMOFs, the main pore surface of CTGU-45 is defined solely by TIB ligands and Cu(II) nodes, resulting in a low-polar pore environment that provides optimized adsorption space for C₂H₆ molecules. This establishes a foundation for the one-step production of high-purity C₂H₄ from ternary C₂ hydrocarbon mixtures (C₂H₂/C₂H₄/C₂H₆). Meanwhile, the marginalized F atoms of the inorganic SiF₆²⁻ pillars serve as strong basic sites, engaging in pronounced interactions with the acidic H atoms of C₂H₂, thereby facilitating efficient C₂H₂/CO₂, C₂H₂/CO₂/CH₄, and C₂H₂/C₂H₄ separations (Scheme 1). The outstanding separation performances, as well as the underlying adsorbate–adsorbent interactions, have been comprehensively verified by a combination of experimental static adsorption/dynamic breakthrough tests and theoretical calculations. This work not only creates CTGU-45 as an efficient adsorbent for challenging C₂ hydrocarbons separation but also demonstrates the great potential of ligand engineering to break the inherent limitations of conventional APMOFs, opening up new avenues for developing advanced adsorbents.

Scheme 1 Comparison of cage polarity and using CTGU-45 for the separation of ternary gas mixtures. Unlike the high-polarity cages in typical APMOFs (left), CTGU-45 (right) features a low-polarity cage surface constructed exclusively from Cu²⁺ nodes and TIB ligands. The SiF₆²⁻ pillars served as strong basic sites for C₂H₂, and non-polar cage surfaces provided the interaction for C₂H₆, which promotes CTGU-45 to realize the effective C₂H₂ and C₂H₄ purification from C₂H₂/CO₂/CH₄ and C₂H₂/C₂H₄/C₂H₆ mixtures, respectively.

Results and discussion

Blue octahedral single crystals of CTGU-45 were successfully synthesized by reacting Cu(OAc)₂·H₂O and TIB ligand in a mixed solvent of N,N-dimethylacetamide (DMA)/H₂O, with HBF₄ added in a sealed glass tube (Fig. S3). SCXRD analysis revealed that CTGU-45 crystallizes in a three-dimensional (3D) framework with the cubic system space group Pm3̄n. Each Cu(II) cation is coordinated in a square-planar geometry by four N atoms from imidazole groups of four distinct tridentate TIB ligands (Fig. S2). The SiF₆²⁻ anions as 2-connected nodes occupy the two axial positions to complete the octahedral coordination. Notably, the SiF₆²⁻ anions originated from the in situ etching of the borosilicate glass vessel by HF (from HBF₄ hydrolysis). Due to the free rotation of the C–N bonds between the central benzene and imidazole moieties, the TIB ligands in CTGU-45 exhibit substantial disorder. Upon removal of the minor disordered components, the real structure shows a uniform cage-like architecture with a pore diameter of approximately 7.6 Å (Fig. 1f). Each incomplete octahedral cage displays four identical triangular windows consisting of three TIB linkers and three Cu²⁺ ions. The effective aperture of these windows, after considering the van der Waals radii of H atoms, is only about 6.5 Å, allowing the entry of C₁–C₂ hydrocarbon molecules (Fig. S1). These cages are further linked by SiF₆²⁻ anion, whose F atoms act as potential Lewis basic sites, which may work synergistically with the confined space and aromatic surface to enhance the binding of gas molecules. In contrast to most reported APMOFs, which typically consist of bidentate ligands and anion pillars forming 1D linear channels (e.g., SIFSIX-1-Cu,²¹ SIFSIX-3-Ni²²), the structure presented here features a well-defined windowed cage architecture. Furthermore, APMOFs constructed from tridentate ligands are exceptionally rare. To our knowledge, only three tridentate ligands (Tripp, TPA, and TPT) have been reported for constructing APMOFs, such as Tripp-Cu-SIFSIX²³ (Fig. 1b) and SIFSIX-Cu-TPA (Fig. 1d),²⁴ both of which exhibit two distinct types of cages. Tripp-Cu-SIFSIX displays octahedra cages (∼7.2 Å) and icositetrahedron cages (∼8.9 Å), while SIFSIX-Cu-TPA possesses icosahedral cages (∼6.0 Å) and tetrahedral cages (∼4.7 Å). Though the pyridine rings of Tripp (D_3h, excluding the non-coordinating pyridine N atom) and TPA (C_3v) ligands could rotate around C(pyridine)–C(pyridine) or N–C(benzene) single bonds, their three pyridyl N donors were locked into a fixed, equilateral triangular geometry. In contrast, the rotation of C(benzene)–N(imidazole) bonds in the TIB (C₃) linkers results in dynamic positions of coordination imidazole N atoms, breaking the perfect symmetry, as presented in Fig. 1a and S4. Such ligand asymmetry easily leads to a distorted coordination geometry and complicates the crystallization process, hindering the predictable assembly of conventional networks. Concretely, in the Tripp-Cu-SIFSIX, SIFSIX-Cu-TPA, and CTGU-45, all the organic linkers act as 3-connected topological nodes with D_3h (Tripp/TPA-node) and C₃ (TIB-node) symmetry, respectively. Meanwhile, the coordination numbers of Cu²⁺ are either 5 or 6, exhibiting square pyramidal (C_4v) or octahedral (D_4h) coordination geometries. The combination of organic nodes and metal nodes ultimately gives rise to (3,5)-c or (3,6)-c topological frameworks in three APMOFs (Fig. 1c/g/e). It is noteworthy that although Cu²⁺ acts as a 6-c node in both SIFSIX-Cu-TPA and CTGU-45, the four pyridine rings coordinated to the Cu²⁺ center in the former exhibit relative dihedral angles close to 90°, whereas the four imidazole rings coordinated to Cu²⁺ in the latter are severely twisted. This causes the 6-c nodes in CTGU-45 to link adjacent nodes in a more distorted fashion, breaking the D_4h extension symmetry of the 6-c nodes observed in SIFSIX-Cu-TPA. This, together with the 3-c nodes with C₃ symmetry, facilitates the formation of a geometrically self-consistent fundamental cage-like network in the absence of SiF₆²⁻ anions (Fig. S4). The foregoing analysis has validated the critical influence of ligand symmetry on the formation of anionic hybrid frameworks. Overall, compared to Tripp-Cu-SIFSIX and SIFSIX-Cu-TPA, CTGU-45 shares the common feature of incorporating SiF₆²⁻ anions, whose electronegative F atoms serve as Lewis-basic sites for interacting with acidic H atoms of C₂H₂. However, a fundamental distinction lies in their pore architecture: in conventional APMOFs, the SiF₆²⁻ anions are integral components that participate directly in cage formation, resulting in highly polar pore surfaces dominated by F atoms (Fig. 1h/k and j/m). In contrast, in CTGU-45, the pore cages are constructed exclusively by the TIB ligands and Cu(II) nodes, while the SiF₆²⁻ anions merely serve as inter-cage pillars (determined by the topology, Fig. 1i/l). This unique arrangement creates a low-polar pore surface rich in aromatic rings and N atoms, which not only provides optimal van der Waals interactions with highly polarizable C₂H₆ molecules but also enables synergistic binding through multiple C–H⋯π and C–H⋯N interactions. This distinctive pore environment, combined with the selective C–H⋯F binding sites at the SiF₆²⁻ pillars, allows CTGU-45 to achieve cooperative adsorption of both C₂H₂ and C₂H₆, thereby providing a robust foundation for one-step C₂H₄ purification from ternary mixtures. The phase purity of the as-synthesized CTGU-45 was confirmed by powder X-ray diffraction (PXRD), as the experimental pattern matches well with simulated ones (Fig. S5). The thermogravimetric analysis (TGA) revealed the high thermal stability of activated CTGU-45, which remains stable up to 350 °C (Fig. S6). The initial weight loss below 100 °C is ascribed to solvent removal, while the subsequent major loss above 350 °C corresponds to the decomposition of the MOF framework. The N₂ adsorption isotherm of CTGU-45 at 77 K displayed a typical Type I profile, and the corresponding HK pore size distribution suggested ultra-micropores of 6.1 Å, collectively confirming its microporous characteristic. The N₂ adsorption uptake of 125.5 cm³ g⁻¹ at 1 kPa and the calculated BET surface area of 544.8 m² g⁻¹ further corroborated its porous structure (Fig. 2a). Moreover, the experimental pore volume of CTGU-45 was calculated to be 0.21 cm³ g⁻¹, which is consistent with the theoretical value from crystal structure (0.24 cm³ g⁻¹). Combined with its excellent thermal stability, these integrated properties collectively endorse CTGU-45 as a highly promising adsorbent for gas separation applications.

Fig. 1 (a) The frameworks of Tripp-Cu-SIFSIX, SIFSIX-Cu-TPA, and CTGU-45 are based on the assembly of SiF₆²⁻, Cu²⁺, and Tripp/TPA/TIB ligands, respectively. (b and c) The crystal structure and (3,5)-connected pto topology of Tripp-Cu-SIFSIX. (d and e) Crystal structure and (3,6)-connected ith topology of SIFSIX-Cu-TPA. (f and g) The crystal structure and (3,6)-connected topology of CTGU-45. Cage structures within (h)/(k) Tripp-Cu-SIFSIX, (i)/(l) CTGU-45, and (j)/(m) SIFSIX-Cu-TPA. H atoms are omitted for clarity.

Fig. 2 (a) N₂ adsorption–desorption isotherms recorded at 77 K. The inset represents pore size distribution analyzed based on the H–K method. The single-component adsorption isotherms of (b) C₂H₂, CO₂, CH₄ and (c) C₂H₂, C₂H₄, C₂H₆ on CTGU-45 at 298 K. (d) Adsorption heats of CH₄, CO₂, and C₂ hydrocarbons. IAST selectivity of (e) C₂H₂/CO₂ (v/v, 50/50), C₂H₂/CH₄ (v/v, 50/50) mixtures and (f) C₂H₂/C₂H₄ (v/v, 50/50), C₂H₆/C₂H₄ (v/v, 50/50) mixtures at 298 K. (g) Adsorption kinetics profiles of CTGU-45 for C₂H₂, C₂H₄, C₂H₆, CO₂, and CH₄. (h) Comprehensive comparison of C₂H₂/CO₂ selectivity and C₂H₂-Q_st value of CTGU-45 with the reported APMOFs. (i) A comparison of near-zero coverage isosteric heats for C₂H₆ among representative porous materials employed for one-step C₂H₄ purification from C₂ hydrocarbon mixtures.

Single-component adsorption isotherms of CH₄, CO₂ and C₂–C₃ hydrocarbons on CTGU-45 were measured at 273, 283 and 298 K (Fig. 2b, c and S7–S18). CTGU-45 exhibited an obviously preferential adsorption of C₂H₂ over CO₂ and CH₄, with an uptake of 2.83 mmol g⁻¹ at 298 K and 101 kPa, significantly exceeding those of CO₂ (1.64 mmol g⁻¹) and CH₄ (0.74 mmol g⁻¹). Notably, the uptakes of C₂H₂ (2.00 mmol g⁻¹) and C₂H₆ (1.88 mmol g⁻¹) are significantly higher than that of C₂H₄ (1.75 mmol g⁻¹) at 30 kPa, thus showing great potential for one-step C₂H₄ purification. The C₂H₂ uptake capacity of CTGU-45 is higher than that of many reported APMOFs, such as NBOFFIVE-dps-Cu (1.65 mmol g⁻¹),²⁵ sql-SIFSIX-bpe-Zn (1.79 mmol g⁻¹),²⁶ DICRO-4-Ni-I (1.92 mmol g⁻¹),²⁷ ZNU-15 (1.61 mmol g⁻¹),²⁸ BSF-2 (1.85 mmol g⁻¹),²⁹ and BSF-4 (2.38 mmol g⁻¹).³⁰

Meanwhile, the CO₂ adsorption amount is markedly lower than that of top-performing APMOFs, including SIFSIX-Cu-TPA (4.79 mmol g⁻¹),²⁴ QDU-MOF-1 (5.70 mmol g⁻¹),³¹ TIFSIX-2-Cu-I (4.3 mmol g⁻¹),³² ZNU-9 (4.32 mmol g⁻¹),³³ and ZNU-12 (4.33 mmol g⁻¹).³⁴ According to the virial equation, the isosteric heat of adsorption (Q_st) values at near-zero loading for C₂H₂, C₂H₄, C₂H₆, CO₂, and CH₄ on CTGU-45 are 30.9, 25.8, 27.3, 28.0, and 20.9 kJ mol⁻¹, respectively (Fig. 2d and S24–S28). The higher Q_st value for C₂H₂ and C₂H₆, consistent with the steeper slope of its single-component adsorption isotherm, indicates a stronger binding affinity between the framework and C₂H₂ and C₂H₆ molecules. Overall, except for ZNU-8 (27.2 kJ mol⁻¹)³³ and MPM-1-TIFSIX (30.1 kJ mol⁻¹),³⁵ CTGU-45 exhibits the lowest Q_st value for C₂H₂ among reported APMOFs, which is significantly lower than that of QDU-MOF-1 (72.52 kJ mol⁻¹),³¹ TIFSIX-17-Ni (48.3 kJ mol⁻¹),³⁶ NBOFFIVE-dps-Cu (53.6 kJ mol⁻¹),³⁷ SIFSIX-Cu-TPA (39.1 kJ mol⁻¹)²⁴ and so on (Fig. 2h). Given that only one APMOF (ZUL-100)³³ has been reported for C₂H₄ purification from C₂ mixtures, the C₂H₆ Q_st value of CTGU-45 (27.3 kJ mol⁻¹) is not only lower than that of ZUL-100 (34.2 kJ mol⁻¹),³³ but also lower than those of most previously reported C₂H₂/C₂H₆-selective MOFs for one-step C₂H₄ purification, such as Zn-trz-ox (38.7 kJ mol⁻¹),³⁸ NUM-9 (35.8 kJ mol⁻¹),³⁹ Zn-atz-oba (30.0 kJ mol⁻¹),⁴ TJT-100 (29.0 kJ mol⁻¹),⁴⁰ and BUT-321 (28.5 kJ mol⁻¹)⁴¹ (Fig. 2i). This low Q_st value ensures easy desorption and low-energy regeneration, which might significantly reduce the operating cost in practical separation cycles.

In addition, since practical gas separation is a non-equilibrium process, the kinetic adsorption behavior should be evaluated synchronously. Therefore, we studied the kinetic diffusion rate of C₂H₂, C₂H₄, C₂H₆, CO₂, and CH₄ in CTGU-45. As shown in Fig. 2g, the time required to reach adsorption equilibrium follows the order: T_ads (CH₄) < T_ads (C₂H₂) < T_ads (CO₂) < T_ads (C₂H₆) < T_ads (C₂H₄). In detail, the calculated diffusion time constants (D_c/r_c²) of these gases on CTGU-45 are 0.0380 (CH₄), 0.2628 (C₂H₂), 0.0284 (CO₂), 0.0141 (C₂H₆), and 0.0104 s⁻¹ (C₂H₄), respectively (Fig. S29). Interestingly, in the desorption kinetics tests (Fig. S30), while CH₄ and CO₂ still exhibited the fastest desorption, the desorption rate sequence of the C₂ gases (T_des (C₂H₄) < T_des (C₂H₆) < T_des (C₂H₂)) was exactly opposite to their respective adsorption rate sequence. This observation directly reflects the relative strength of their interactions with the framework, consistent with the assessments discussed earlier.

To further assess the separation potential of CTGU-45, the adsorption selectivity for equimolar C₂H₂/CO₂, C₂H₂/CH₄, C₂H₂/C₂H₄, and C₂H₆/C₂H₄ mixtures at 298 K was calculated based on the Ideal Adsorbed Solution Theory (IAST) after fitting isotherms to the single-site/dual-site Langmuir–Freundlich equation with excellent accuracy (Fig. S19–S23, R² > 0.99999). As illustrated in Fig. 2e and f, the IAST selectivity of CTGU-45 at 298 K and 101 kPa reached 6.99 for C₂H₂/CO₂ and 38.72 for C₂H₂/CH₄; notably, the selectivity for C₂H₂/C₂H₄ (50/50, v/v) (1.63) and C₂H₆/C₂H₄ (50/50, v/v) (1.40) further underscore its potential for one-step C₂H₄ purification. The C₂H₂/CO₂ selectivity is moderate among benchmark APMOFs, being comparable to those of NBOFFIVE-3-Ni (6.0) and TIFSIX-2-Cu-I (6.5), and higher than SIFSIX-Cu-TPA (5.3), QDU-MOF-1 (2.95), and ZNU-8 (3.74) (Fig. 2h). Meanwhile, the moderate C₂H₂/CO₂ selectivity, combined with the high C₂H₂/CH₄ selectivity, suggests that CTGU-45 is a viable candidate for separating C₂H₂/CO₂ and C₂H₂/CH₄ mixtures. Furthermore, the calculated C₂H₂ and CO₂ uptake (2.20 and 0.32 mmol g⁻¹) in the equimolar mixture closely matches the corresponding single-component adsorption capacities at 50 kPa (2.37 and 1.02 mmol g⁻¹), confirming the consistency between mixture prediction and gas adsorption measurement.

Grand Canonical Monte Carlo (GCMC) simulations and density functional theory with dispersion corrections (DFT-D) calculations were employed to elucidate the gas adsorption mechanisms in CTGU-45, revealing two primary binding sites within the framework: one near the SiF₆²⁻ anion pillars and another inside the pore cavity (Fig. 3). In general, the region adjacent to the SiF₆²⁻ pillars serves as a strong adsorption site for CO₂, CH₄, C₂H₂, and C₂H₄, whereas the pore cavity functions as the preferred strong binding region for C₂H₆. Taking C₂H₂ as a representative case, the first corresponds to a strong adsorption site, where the C₂H₂ molecule was confined by two SiF₆²⁻ with a binding energy of −53.42 kJ mol⁻¹. Specifically, each H atom of C₂H₂ forms two C–H⋯F hydrogen bonds (H-bonds) with F atoms from two adjacent SiF₆²⁻ pillars, resulting in a total of four H-bonds with distances ranging from 2.34 to 3.69 Å. These are complemented by C≡C⋯H interactions (distances: 2.58–3.45 Å) between the π-electrons of C₂H₂ and H atoms of imidazole rings. The second site corresponds to a weaker interaction environment within the pore cavity, involving C≡C⋯H (distances: 3.85–3.94 Å), C–H (C₂H₂)⋯N(imidazole) (2.91–3.82 Å), and C–H (C₂H₂)⋯π (imidazole/benzene) interactions with the distances of 3.10–3.92 Å. The binding energy for this weak-interaction site is −19.92 kJ mol⁻¹, substantially lower than that of the above-mentioned site. A similar distribution of adsorption sites was observed for CO₂, CH₄, and C₂H₄ with strong binding near the SiF₆²⁻ pillars and weaker interactions in the pore cavity. For CO₂, the electronegative O atoms participated in C=O⋯H and C=O⋯N contacts with H and N atoms of the imidazole rings, as well as C=O⋯π interactions with the aromatic systems. Simultaneously, its electropositive C atom interacts with F atoms of the SiF₆²⁻ pillars via O=C⋯F interactions. In the case of CH₄, C₂H₄, and C₂H₆, adsorption was governed by C–H⋯F and C–H⋯N interactions with the pillar F and imidazole N atoms, respectively, along with C–H⋯π contacts involving the aromatic rings of the TIB ligands. Notably, C₂H₆ exhibits its strongest adsorption within the pore cavity rather than near the anion pillars. Consequently, the binding energies of CO₂ and CH₄ at both types of adsorption sites are substantially lower than those of C₂H₂, while those of C₂H₂ and C₂H₆ are higher than those of C₂H₄. This energetic trend (C₂H₂ > CO₂ > CH₄; C₂H₂ > C₂H₆ > C₂H₄) further rationalizes the preferential adsorption of C₂H₂ over CO₂ and CH₄, and of C₂H₂ and C₂H₆ over C₂H₄ by CTGU-45. The stronger intermolecular interactions between the framework and C₂H₂/C₂H₆ directly correlate with its higher adsorption affinity, which is consistent with the observation on Q_st and single-component adsorption data.

Fig. 3 Optimal binding sites and adsorption configurations of (a and b) CO₂, (c) CH₄, (d and e) C₂H₂, (f and g) C₂H₄, and (h and i) C₂H₆ molecules within CTGU-45 defined by DFT-D calculations.

Dynamic breakthrough experiments were conducted to further evaluate the practical separation performance of CTGU-45 for C₂H₂/CO₂, C₂H₂/CO₂/CH₄, and C₂ hydrocarbon mixtures. As depicted in Fig. 4a, CTGU-45 can achieve efficient separation of a C₂H₂/CO₂ (50/50, v/v) mixture at a total gas flow rate of 4.0 mL min⁻¹, wherein CO₂ gases were detected in ∼9 min g⁻¹ while C₂H₂ breakthrough occurred at ∼22 min g⁻¹. Based on the integration of the curve, the dynamic CO₂ and C₂H₂ capture amounts of CTGU-45 were calculated to be 6.10 and 47.68 cm³ g⁻¹, corresponding to approximately 26 and 90% of their respective static uptake values at 50 kPa (23.04 and 53.12 cm³ g⁻¹), respectively. After adsorption saturation, the column was regenerated by purging with helium at 10 mL min⁻¹ under ambient temperature. The desorption profiles indicated rapid release of both gases, with CO₂ completely desorbed within 23 min g⁻¹. Beyond this point, high-purity C₂H₂ (>95.2%) was obtained at a productivity of 22.1 L kg⁻¹ (Fig. 4a). Subsequently, we further investigated the separation capacity of CTGU-45 for actual ternary C₂H₂/CO₂/CH₄ mixtures (33/33/33, v/v/v) with flow rates of 6 and 3 mL min⁻¹. Owing to its higher C₂H₂ uptakes and selectivity over CO₂ and CH₄, efficient separation of C₂H₂ from C₂H₂/CO₂/CH₄ mixtures can be fulfilled by CTGU-45 under different temperatures (298 and 308 K, Fig. S31).

Fig. 4 Experimental column breakthrough curves and the desorption curves for (a) C₂H₂/CO₂ (50/50, v/v) and (b) C₂H₂/CO₂/CH₄ (33/33/33, v/v/v) on CTGU-45 at 298 K with a total flow of (a) 4 and (b) 6 mL min⁻¹. The dynamic breakthrough curves for (c) C₂H₂/C₂H₄ (50/50 and 10/90, v/v), (d) C₂H₆/C₂H₄ (50/50 and 10/90, v/v), and (e) C₂H₂/C₂H₆/C₂H₄ (1/10/89, v/v/v) on CTGU-45 at 298 K with a total flow of (c) 1, (d) 1, and (e) 15 mL min⁻¹. (f) Breakthrough cycling test of CTGU-45 for C₂H₂/C₂H₆/C₂H₄ (1/10/89, v/v/v, 298 K, 15 mL min⁻¹).

As illustrated in Fig. 4b, CH₄ first eluted through the fixed bed at 3 min g⁻¹, followed by CO₂ at 4 min g⁻¹, whereas the C₂H₂ was retained in the packed column over 18 min g⁻¹. Despite CO₂ and CH₄ exhibiting a faster adsorption rate than C₂H₂ in CTGU-45, the stronger host–guest interactions between the framework and C₂H₂ (C–H⋯F, C–H⋯N, and C–H⋯π interactions) confer thermodynamic dominance in the actual separation process, leading to preferential C₂H₂ capture and the longest breakthrough time. The desorption curve further confirmed that polymer-grade C₂H₂ can be recovered at a productivity (>95.2%) of 19.7 L kg⁻¹.

For C₂H₂/C₂H₄ and C₂H₆/C₂H₄ mixtures with different compositions (50/50 or 10/90, v/v), C₂H₄ first eluted from the packed column while C₂H₂ or C₂H₆ remained strongly retained within the CTGU-45, demonstrating its ability to directly produce polymer-grade C₂H₄ products. In detail, C₂H₄ molecules broke through the fixed bed and were detected at 4.2 (3.7) and 42.9 (15.5) min g⁻¹ for the 50/50 (10/90) C₂H₂/C₂H₄ and C₂H₆/C₂H₄ mixtures, respectively, whereas C₂H₂ and C₂H₆ were adsorbed in the column over 51.7 (27.6) and 49.9 (51.4) min g⁻¹. As depicted in Fig. 4c and d, the productivity of 99.9% pure C₂H₄ from the 50/50 (10/90) C₂H₂/C₂H₄ and C₂H₆/C₂H₄ mixtures reached 28.6 (47.7) and 6.0 (33.3) L kg⁻¹, respectively. Building on the excellent separation performance observed for these binary mixtures, breakthrough experiments were further conducted using a ternary C₂H₂/C₂H₆/C₂H₄ (1/10/89, v/v/v) mixture. Notably, C₂H₄ eluted (0.9 min g⁻¹) prior to both C₂H₂ (1.8 min g⁻¹) and C₂H₆ (1.7 min g⁻¹), enabling the direct production of high-purity C₂H₄ (Fig. 4e). After calculation, 9.57 L kg⁻¹ of C₂H₄ with ≥99.95% purity can be directly obtained in a single cycle (298 K, 15 mL min⁻¹). For both ternary mixtures, CTGU-45 maintains reliable efficiency in C₂H₂ and C₂H₄ purification with steady gas productivities when the operating temperature is raised from 298 to 308 K, or when the flow rate is decreased from 6 to 3 mL min⁻¹ (15 to 14 mL min⁻¹, Fig. S31). Moreover, repeated breakthrough experiments over five consecutive cycles showed no discernible decay in either C₂H₄ purity or productivity (9.24–11.0 L kg⁻¹), confirming the stable one-step C₂H₄ purification capability of CTGU-45 (Fig. 4f and S32). Finally, PXRD measurements reveal that CTGU-45 displays moderate chemical stability across a wide pH range (1–13) and in aqueous environments (Fig. S33). The combination of high separation efficiency, excellent adaptability, good thermal/chemical stability, and mild regeneration requirements (room-temperature He purging) underscores the potential of CTGU-45 for practical C₂H₂ and C₂H₄ purification.

In addition to the impressive separation properties, CTGU-45 crystals can be successfully synthesized at scale-up levels using glass containers with volumes ranging from 50 to 500 mL (Fig. 5a). PXRD measurements confirm the basic phase purity of the CTGU-45 sample obtained via scaled-up synthesis (Fig. 5b). Although the N₂ adsorption capacity of samples synthesized in 50, 100, and 200 mL vessels were only slightly lower than that of the 20 mL reference (Fig. 5c), and the sample prepared in a 500 mL bottle exhibits a 25.2% reduction in N₂ uptake. To further assess the impact on functional performance, C₂H₂ adsorption isotherms were measured at 298 K (Fig. 5d). The results indicate that the C₂H₂ uptake of samples from 20 and 50 mL vessels is nearly identical, while those from 100 and 200 mL vessels show an approximate 8.8% decrease. In contrast, the sample synthesized in the 500 mL bottle exhibits a 25.4% reduction in C₂H₂ adsorption capacity, consistent with the trend observed in N₂ adsorption. Through additional analysis of the N₂ ads/des–orption isotherms, pore size distributions (Fig. S34), and PXRD patterns of samples synthesized at different scales, we found that the decrease in adsorption capacity may arise not only from undesired structural defects but also from the presence of by-products (amorphous, non-porous, and non-adsorbing). Since MOF synthesis is a complex self-assembly process governed by both thermodynamic and kinetic factors, changes in the scale of the reaction system—that is, the microscopic reaction conditions—can perturb the reaction pathway, leading to the formation of unknown by-products. To mitigate the formation of amorphous by-products and structural defects during industrial production for CTGU-45, future efforts might focus on two key strategies: (i) the implementation of continuous-flow synthesis or microwave-assisted heating can ensure uniform nucleation and growth conditions, thereby suppressing kinetic traps that lead to amorphous phases (ii) dynamic adjustment of reaction parameters (e.g., reactant concentration, temperature gradients, local mass transfer); to steer the self-assembly toward the desired crystalline product.

Fig. 5 (a) Scale-up synthesis of CTGU-45 at scales of 50, 100, 200, and 500 mL. (b) PXRD patterns, (c) N₂ adsorption–desorption isotherms, and (d) C₂H₂ adsorption isotherms of CTGU-45 synthesized at different scales via a solvothermal method.

Conclusions

In summary, we reported a SiF₆²⁻ embedded MOF, CTGU-45, featuring abundant electronegative F atoms and a low-polar cage environment, which enables one-step C₂H₄ purification via the simultaneous removal of C₂H₂ and C₂H₆. CTGU-45 exhibits high C₂H₂ and C₂H₆ uptake, markedly exceeding those for CO₂, CH₄, and C₂H₄. The low-polar pore environment effectively addresses the critical challenge of achieving C₂H₆-selective adsorption in C₂H₆/C₂H₄ separation, a behavior that is exceedingly rare among reported APMOFs. As a result, ∼9.57 L kg⁻¹ of polymer grade C₂H₄ (≥99.95%) can be directly produced from ternary C₂H₂/C₂H₆/C₂H₄ (1/10/89, v/v/v) mixtures in a single step. DFT calculations reveal that strong C–H⋯F interactions between the inorganic SiF₆²⁻ pillars and C₂H₂ enable CTGU-45 to not only efficiently purify C₂H₂ from C₂H₂/CO₂/CH₄ mixtures but also, in synergy with its C₂H₆-trapping capability, to achieve one-step production of high-purity C₂H₄ from ternary C₂ hydrocarbon mixtures. In addition, the low C₂H₆ Q_st (27.3 kJ mol⁻¹) ensures facile regeneration with low energy input. Notably, CTGU-45 can be synthesized on a scalable level (up to 500 mL) with well-retained adsorption capacity. Overall, this work demonstrates that breaking ligand symmetry in APMOFs can fundamentally alter the spatial role of anion pillars, thereby reshaping pore surface chemistry and enabling rare one-step C₂H₄ purification from C₂ hydrocarbon mixtures.

Author contributions

J.-J. Wu, P.-D. Zhang, X.-Q. Wu, and D.-S. Li conceived this project and designed the experiments. J.-J. Wu, S.-H. Huang, W.-W. Dong and Y.-P. Wu conducted the experiments. J.-J. Wu, P.-D. Zhang and X.-Q. Wu drafted the work. D.-S. Li revised and finalised the manuscript. All authors approved the submitted version.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support the findings of this study are openly available in the supporting information (SI). Supplementary information: experimental procedures, TGA curves, gas adsorption/separation isotherms, crystallographic data, and theoretical computational details. See DOI: https://doi.org/10.1039/d6sc00819d.

CCDC 2522768 contains the supplementary crystallographic data for this paper.⁴²

Acknowledgements

We acknowledge the financial support from the National Natural Science Foundation of China (No. 22401172, No. 22371165), the 111 Project (D20015), the project of Hubei Three Gorges Laboratory (Z2022078, SK213002, SC232018, SC240013), Natural Science Foundation of Hubei Province (2024AFB021, 2022CFB326, Z2023203), Natural Science Foundation Innovation and Development Joint Fund of Hubei Province (2024AFD197).

Notes and references

1. K. M. Carsch, H. Z. H. Jiang, R. A. Klein, A. S. Rosen, P. S. Summerhill, J. L. Peltier, A. J. Huang, R. A. Murphy, M. N. Dods, H. A. Silva, Z. Hasanbasri, H. Kwon, S. L. Karstens, Y. Yabuuchi, J. Börgel, J. W. Taylor, K. R. Meihaus, K. C. Bustillo, A. M. Minor, K. A. Persson, C. M. Brown, R. D. Britt, N. P. Stadie and J. R. Long, Science, 2025, 390, 808–812.
2. I. J. Vitórica-Yrezábal, C. A. McAnally, M. P. Snelgrove, M. R. Warren, A. H. Hill, S. P. Thompson, M. Quinn, S. Mottley, S. Mottley, A. J. Fletcher and L. Brammer, Nat. Chem., 2025, 17, 1705–1711.
3. Z. Zhou, T. Ma, H. Zhang, S. Chheda, H. Li, K. Wang, S. Ehrling, R. Giovine, C. Li, A. H. Alawadhi, M. M. Abduljawad, M. O. Alawad, L. Gagliardi, J. Sauer and O. M. Yaghi, Nature, 2024, 635, 96–101.
4. J.-W. Cao, S. Mukherjee, T. Pham, Y. Wang, T. Wang, T. Zhang, X. Jiang, H.-J. Tang, K. A. Forrest, B. Space, M. J. Zaworotko and K.-J. Chen, Nat. Commun., 2021, 12, 6507.
5. C. Lu, S. Liu, Z. Wang, X. Wei, X. Chen, X. Wang, J. Pang, S. Geng, X. Lu, J. Duan, F. Dai and X.-H. Bu, Adv. Mater., 2025, 38, e14488.
6. L. Li, R.-B. Lin, R. Krishna, H. Li, S. Xiang, H. Wu, J. Li, W. Zhou and B. Chen, Science, 2018, 362, 443–446.
7. Y. Yang, L. Li, R.-B. Lin, Y. Ye, Z. Yao, L. Yang, F. Xiang, S. Chen, Z. Zhang, S. Xiang and B. Chen, Nat. Chem., 2021, 13, 933–939.
8. P. Ajayan, W. Wang, Y. Chen, X. Bu and P. Feng, Adv. Mater., 2024, 36, 2408042.
9. B. E. R. Snyder, A. B. Turkiewicz, H. Furukawa, M. V. Paley, E. O. Velasquez, M. N. Dods and J. R. Long, Nature, 2023, 613, 287–291.
10. L. Li, L. Guo, D. H. Olson, S. Xian, Z. Zhang, Q. Yang, K. Wu, Y. Yang, Z. Bao, Q. Ren and J. Li, Science, 2022, 377, 335–339.
11. G.-R. Si, X.-J. Kong, L.-H. Xie, Y.-T. Zhang, T. He and J.-R. Li, Sci. Bull., 2025, 70, 3475–3478.
12. Z.-H. Guo, L.-Q. Yang, Q.-G. Zhai, G.-P. Yang and Y.-Y. Wang, Angew. Chem., Int. Ed., 2025, 64, e202519278.
13. L. Wang, H. Chen, C. Lou, G. Xiong, Y. Jiang, B. Chen and Y. Zhang, Angew. Chem., Int. Ed., 2025, 64, e202519160.
14. Z.-S. Wang, M.-Y. Zhou, D.-Y. Hu, X.-X. Chen, J.-X. Chen, X.-W. Zhang, D.-D. Zhou, J.-P. Zhang and X.-M. Chen, J. Am. Chem. Soc., 2025, 147, 47977–47984.
15. Y. Xiao, A. N. Hong, Z. Jia, X. Bu and P. Feng, J. Am. Chem. Soc., 2025, 147, 42786–42795.
16. T. Wu, D. Zhou, Z. Yang, Y. Liu, P. Yin, K. Lu, F. Jiang, X. Suo, L. Yang, X. Cui and H. Xing, J. Am. Chem. Soc., 2025, 148, 5299–5306.
17. Z. Ji, Y. Zhou, C. Chen, D. Yuan, M. Wu and M. Hong, Angew. Chem., Int. Ed., 2024, 63, e202319674.
18. Y. Jiang, Y. Hu, B. Luan, L. Wang, R. Krishna, H. Ni, X. Hu and Y. Zhang, Nat. Commun., 2023, 14, 401.
19. A. Cadiau, K. Adil, P. M. Bhatt, Y. Belmabkhout and M. Eddaoudi, Science, 2016, 353, 137–140.
20. F. Zheng, L. Guo, R. Chen, L. Chen, Z. Zhang, Q. Yang, Y. Yang, B. Su, Q. Ren and Z. Bao, Angew. Chem., Int. Ed., 2022, 61, e202116686.
21. X. Cui, K. Chen, H. Xing, Q. Yang, R. Krishna, Z. Bao, H. Wu, W. Zhou, X. Dong, Y. Han, B. Li, Q. Ren, M. J. Zaworotko and B. Chen, Science, 2016, 353, 141–144.
22. H.-M. Wen, C. Liao, L. Li, L. Yang, J. Wang, L. Huang, B. Li, B. Chen and J. Hu, Chem. Commun., 2019, 55, 11354–11357.
23. M. Lusi, P. B. A. Fechine, K.-J. Chen, J. J. Perry and M. J. Zaworotko, Chem. Commun., 2016, 52, 4160–4162.
24. H. Li, C. Liu, C. Chen, Z. Di, D. Yuan, J. Pang, W. Wei, M. Wu and M. Hong, Angew. Chem., Int. Ed., 2021, 60, 7547–7552.
25. M. Shivanna, K.-i. Otake, B.-Q. Song, L. M. van Wyk, Q.-Y. Yang, N. Kumar, W. K. Feldmann, T. Pham, S. Suepaul, B. Space, L. J. Barbour, S. Kitagawa and M. J. Zaworotko, Angew. Chem., Int. Ed., 2021, 60, 20383–20390.
26. M.-Y. Gao, A. A. Bezrukov, B.-Q. Song, M. He, S. J. Nikkhah, S.-Q. Wang, N. Kumar, S. Darwish, D. Sensharma, C. Deng, J. Li, L. Liu, R. Krishna, M. Vandichel, S. Yang and M. J. Zaworotko, J. Am. Chem. Soc., 2023, 145, 11837–11845.
27. H. S. Scott, M. Shivanna, A. Bajpai, D. G. Madden, K.-J. Chen, T. Pham, K. A. Forrest, A. Hogan, B. Space, J. J. Perry Iv and M. J. Zaworotko, ACS Appl. Mater. Interfaces, 2017, 9, 33395–33400.
28. H. Chen, Y. He, Y. Han, J. Hu, J. Li, Y. Jiang, B. Keshta, L. Wang and Y. Zhang, Chin. J. Struct. Chem., 2025, 44, 100508.
29. Y. Zhang, J. Hu, R. Krishna, L. Wang, L. Yang, X. Cui, S. Duttwyler and H. Xing, Angew. Chem., Int. Ed., 2020, 59, 17664–17669.
30. L. Wang, W. Sun, Y. Zhang, N. Xu, R. Krishna, J. Hu, Y. Jiang, Y. He and H. Xing, Angew. Chem., Int. Ed., 2021, 60, 22865–22870.
31. H.-Y. Li, Z.-Z. Xue, S.-D. Han, G.-M. Wang and T. He, Sep. Purif. Technol., 2025, 357, 130094.
32. K.-J. Chen, H. S. Scott, D. G. Madden, T. Pham, A. Kumar, A. Bajpai, M. Lusi, K. A. Forrest, B. Space, J. J. Perry and M. J. Zaworotko, Chem, 2016, 1, 753–765.
33. Y. Zhang, W. Sun, B. Luan, J. Li, D. Luo, Y. Jiang, L. Wang and B. Chen, Angew. Chem., Int. Ed., 2023, 62, e202309925.
34. Y. Zhang, Y. Han, B. Luan, L. Wang, W. Yang, Y. Jiang, T. Ben, Y. He and B. Chen, J. Am. Chem. Soc., 2024, 146, 17220–17229.
35. P. S. Nugent, V. L. Rhodus, T. Pham, K. Forrest, L. Wojtas, B. Space and M. J. Zaworotko, J. Am. Chem. Soc., 2013, 135, 10950–10953.
36. S. Mukherjee, N. Kumar, A. A. Bezrukov, K. Tan, T. Pham, K. A. Forrest, K. A. Oyekan, O. T. Qazvini, D. G. Madden, B. Space and M. J. Zaworotko, Angew. Chem., Int. Ed., 2021, 60, 10902–10909.
37. N. Kumar, S. Mukherjee, N. C. Harvey-Reid, A. A. Bezrukov, K. Tan, V. Martins, M. Vandichel, T. Pham, L. M. van Wyk, K. Oyekan, A. Kumar, K. A. Forrest, K. M. Patil, L. J. Barbour, B. Space, Y. Huang, P. E. Kruger and M. J. Zaworotko, Chem, 2021, 7, 3085–3098.
38. T. Wu, C. Yu, R. Krishna, Z. Qiu, H. Pan, P. Zhang, X. Suo, L. Yang, X. Cui and H. Xing, AIChE J., 2024, 70, e18312.
39. S.-Q. Yang, F.-Z. Sun, P. Liu, L. Li, R. Krishna, Y.-H. Zhang, Q. Li, L. Zhou and T.-L. Hu, ACS Appl. Mater. Interfaces, 2021, 13, 962–969.
40. H.-G. Hao, Y.-F. Zhao, D.-M. Chen, J.-M. Yu, K. Tan, S. Ma, Y. Chabal, Z.-M. Zhang, J.-M. Dou, Z.-H. Xiao, G. Day, H.-C. Zhou and T.-B. Lu, Angew. Chem., Int. Ed., 2018, 57, 16067–16071.
41. P.-D. Zhang, X.-Q. Wu, Q. Shuai, J. Yu, X. Zhang and J.-R. Li, ACS Mater. Lett., 2024, 6, 4632–4638.
42. CCDC 2522768: Experimental Crystal Structure Determination, 2026,  DOI:10.5517/ccdc.csd.cc2qp4m6.

---