Acetylene (C₂H₂), as a key raw petrochemical material, is widely applied in manufacturing worthy chemical commodities such as acetaldehyde, acetic acid, and 1,4-butynediol [1–3]. However, carbon dioxide (CO₂) is the inevitable by-product of the acetylene industrial production process, and the existence of CO₂ will cut down energy conversion efficiency [4,5]. Owing to the identical molecular kinetic diameter and similar physical properties between CO₂ and C₂H₂ (Table S1 in Supporting information) [6–8], obtaining high-purity acetylene (>99%) to manufacture valuable chemicals is relatively energy-intensive in the industry [9,10]. Consequently, developing energy-saving, cost-efficient, and environmentally friendly substitutive approaches is essential for C₂H₂ purification.

Porous physical adsorbents-based gas adsorptive separation can realize energy-efficient economy and operability, which has attracted the immense attention of researchers [11–13]. Metal–organic frameworks (MOFs) exhibit tunable pore structure, shape, and size [14], which provide numerous opportunities to create unique conditions for gas adsorption and separation. Pillar–layered MOFs (PL-MOFs) constructed from two-dimensional (2D) layers and adjustable pillars have great potential for C₂H₂ and CO₂ separation via delicate pore environment regulation [15–17]. To date, a pillar-layered strategy [18,19] utilizing the selection of different pillars with distinct chemical properties, has been applied in pore environment control of PL-MOFs to obtain the anticipatory pore environment for the optimization of C₂H₂/CO₂ separation [18]. Previous studies have proven that tuning the pillar length and derivatization is capable of manipulating the pore size and pore environment for task-specific gas separation [19,20]. However, most of the pillars are confined to aromatic linkers due to the characteristics of structural stability and easy functionalization [21]. Compared to aromatic linkers, aliphatic linkers possess desirable features such as steric effect, nonaromaticity, and saturated C—H groups [22,23], are expected to expand the kinds of MOF and conducive to create a low-polar pore chemical environment, which generates distinct host-guest interactions for distinguishing specific gas molecules [21,24,25]. So far, the study on the kind of host-guest interactions and the mechanism of adsorption between gas and aliphatic linkers are still scarce.

In light of these considerations, based on the tunability of PL-MOFs, we chose Ni(btc)(pyz) (named Ni-MOF-1; H₃btc = 1,3,5-benzene tricarboxylic acid; pyz = pyrazine) as a prototype MOF [26], which has verified the ability to remove trace CO₂ from C₂H₂/CO₂ mixture [27]. Inspired by abundant saturated C—H groups of the aliphatic ligand 1,4-diazabicyclo[2.2.2]octane (dabco), we speculate that tuning pillars from pyz to dabco may create an unusual aliphatic chemical environment, pore size, structural rigidity, and enhanced adsorption performance toward C₂H₂/CO₂ separation (Scheme 1) [25,28]. In this regard, Ni(btc)(dabco) (termed Ni-MOF-2) is successfully obtained based on reticular chemistry [29]. As expected, Ni-MOF-2 not only exhibits the relatively high C₂H₂ capacity over CO₂ and 3.5-fold strengthened selectivity but also realizes a complete splitting of equimolar C₂H₂/CO₂ mixture (Scheme 1b). In addition, separation performance and regeneration stability for equimolar C₂H₂/CO₂ mixture of Ni-MOF-2 were testified by the dynamic breakthrough experiments. Theoretical calculations reveal that host-guest interaction is converted from hydrogen bonding interaction to van der Waals (vdW) interaction and the saturated C—H groups of dabco serve as primary adsorption sites to interact with C₂H₂. The enhanced C₂H₂/CO₂ selectivity, low regeneration energy, and cyclic stability suggest that Ni-MOF-2 is a promising porous adsorbent for C₂H₂/CO₂ separation.

Ni-MOF-1 and Ni-MOF-2 are both obtained by the post-synthetic ligand exchange (PSLE) method [26,30,31]. They are two isostructural PL-MOFs constructed from the same 2D layers and different pillars (Figs. 1a–e). In the 2D layers, each Ni²⁺ ion coordinates with four O atoms from three btc ligands to form hexagon pores (Figs. S1 and S2 in Supporting information). Then, pyz and dabco as pillars (Figs. 1b and c) connect adjacent 2D layers (Fig. 1a) stacking up and down to generate three-dimensional (3D) frameworks of Ni-MOF with the same hms topology (Figs. 1d and e, Table S2 in Supporting information). Ni-MOF-1 and Ni-MOF-2 show hexagonal-prism cages with pore apertures of 6.9 Å × 7.3 Å [32] and 7.0 Å × 8.0 Å, respectively (Figs. 1f and g), which is consistent with pore size distribution (Fig. S7b in Supporting information). The high phase purity of Ni-MOF-2 was verified by the bulk phase powder X-ray diffraction (PXRD) patterns (Fig S6 in Supporting information). The crystallinity of Ni-MOF-2 can remain after being soaked in common organic solvents for 24 h (Fig. S8 in Supporting information). The whole structure and crystallinity of Ni-MOF-2 survive up to 300 ℃, as verified by thermogravimetric analysis (TGA) (Fig. S9b in Supporting information) and variable temperature PXRD (VT-PXRD) (Fig. S10 in Supporting information). Compared to the thermal stability of Ni-MOF-1, the thermal robustness of Ni-MOF-2 is still preserved, implying the tiny impact of aliphatic pillars on the structural stability. The hexagonal-prism cages, chemical, and thermal stabilities of Ni-MOF-2 pave the way for gas separation.

The activated Ni-MOF-2 was prepared by heating at 150 ℃ for 12 h under a dynamic vacuum. Nitrogen (N₂) gas sorption isotherms at 77 K Ni-MOF-2 show a typical type-Ⅰ sorption curve, indicating the microporous structure of Ni-MOF-2. As shown in Fig. S7a (Supporting information), the saturated N₂ capacity of Ni-MOF-2 increased from 282.4 cm³/g [32] in Ni-MOF-1 to 351.6 cm³/g, aligning with the pore size derived from the single crystal structure. The calculated Brunauer–Emmett–Teller (BET) surface area of Ni-MOF-2 is 636.0 m²/g, slightly lower than the value of Ni-MOF-1 (686.0 m²/g) (Table S3 in Supporting information) [32]. Additionally, the effective free volume is also narrowed down from 52.8% in Ni-MOF-1 to 44% in Ni-MOF-2. The microporosity, high porosity, appropriate pore size, and additional saturated C—H groups of Ni-MOF-2 exhibit the potential to address the challenge of C₂H₂/CO₂ separation.

To verify the effect of pillar modification on the adsorption of gas molecules, single-component adsorption isotherms of C₂H₂and CO₂ for Ni-MOF were examined. As shown in Figs. 2a and b and Fig. S11 (Supporting information), at both 273 K and 298 K, Ni-MOF-1 prefers adsorbing C₂H₂ at low pressure but the higher CO₂ capacity upon increasing the pressure. However, after pillar variation of Ni-MOF-1, Ni-MOF-2 demonstrates a higher C₂H₂ adsorption capacity than CO₂ within the full pressure region, indicating the potential of Ni-MOF-2 for separating equimolar C₂H₂/CO₂. Moreover, for the diminution of surface area and effective free volume, the saturated capacities of C₂H₂ and CO₂ on Ni-MOF-2 at 298 K are reduced from 106.4 cm³/g and 128.2 cm³/g to 85.4 cm³/g and 75.1 cm³/g, respectively (Fig. 2b and Table S3), demonstrating the greater reduction of CO₂ capacity and reversal of the interactions between frameworks and gas molecules after pore environment regulation. And the adsorption trend at 273 K of Ni-MOF is the same (Fig. 2a). Notably, the capacity of C₂H₂ of Ni-MOF-2 at 298 K and 1 bar is higher than ZJU-74a (76.0 cm³/g) [33], NKMOF-1-Ni (61.0 cm³/g) [34], and Cu(bpy)NP (50.7 cm³/g) [35], comparable to CuSnF₆-dpds-cds (80.9 cm³/g) [1], and CAU-10-pydc (88.3 cm³/g) [36], and lower than NCU-100 (102.4 cm³/g) [5], and HIAM-111 (108.9 cm³/g) [37] under the same condition (Fig. 2e and Table S4 in Supporting information).

Subsequently, the ideal adsorbed solution theory (IAST) was employed to assess the adsorptive selectivity and evaluate the separation performance for equimolar C₂H₂/CO₂ mixture on Ni-MOF-1 and Ni-MOF-2. As depicted in Fig. 2c, the equimolar C₂H₂/CO₂ selectivity of Ni-MOF-1 is calculated as 1.0 and 1.3 at 273 K and 298 K and 100 kPa, respectively. Nevertheless, by virtue of switching pillars, Ni-MOF-2 exhibits a relatively high selectivity of 3.3 and 4.5 for equimolar C₂H₂/CO₂ mixture, respectively, 3.3- and 3.5-fold higher than that of Ni-MOF-1, respectively, implying the enhanced separation potential of Ni-MOF-2 for equimolar C₂H₂/CO₂. Meanwhile, the equimolar C₂H₂/CO₂ selectivity of Ni-MOF-2 at 298 K outperforms that of some reported MOFs for C₂H₂/CO₂, such as SNNU-98-Ni (3.3) [38], HIAM-111 (2.4) [37], and FJU-90 (4.3) (Table S4) [39].

To compare the binding affinity of Ni-MOF toward C₂H₂ and CO₂, the isosteric heats of adsorption (Q_st) were calculated (Fig. S13 in Supporting information). For Ni-MOF-1, Q_st of C₂H₂ and CO₂ were calculated to be 38.6 kJ/mol and 32.8 kJ/mol at zero coverage, respectively (Fig. 2d). And the Q_st of C₂H₂ on Ni-MOF-2 is 42.2 kJ/mol, surpassing the Q_st of CO₂ on Ni-MOF-2 (32.2 kJ/mol) (Fig. 2d), exceeding that of ZNU-12 (36.1 kJ/mol) [2], and MOF-303 (30.8 kJ/mol) [36], inferior to SOFOUR-TEPE-Zn (45.6 kJ/mol) (Fig. 2f and Table S4) [8]. The appropriate Q_st of C₂H₂ facilitates regeneration using a low energy penalty. Increased adsorption disparity of C₂H₂ and CO₂, enhanced equimolar C₂H₂/CO₂ selectivity, and moderate isosteric heat of adsorption make Ni-MOF-2 a prospective candidate for C₂H₂ purification.

To further validate the effect of pillar modification for equimolar C₂H₂/CO₂ separation, dynamic breakthrough experiments with different flow rates were conducted on Ni-MOF-1 and Ni-MOF-2 at 298 K and 1.0 bar. As shown in Figs. 3a and b, equimolar C₂H₂/CO₂ can be successfully separated on Ni-MOF-2 with a total inlet flow rate of 1 and 2 mL/min, whereas Ni-MOF-1 could not remove C₂H₂ from equimolar C₂H₂/CO₂ mixture. For Ni-MOF-2 with a flow rate of 1 mL/min, CO₂ was first eluted through the column at 99.5 min/g, followed by C₂H₂ at 118.6 min/g (Fig. 3a). When the flow rates increased to 2 mL/min and 3 mL/min, Ni-MOF-2 still achieved a clean separation of equimolar C₂H₂/CO₂ mixture, and the breakthrough point of CO₂ reduced to 53.6 min/g and 34.4 min/g (Figs. 3b and c). And Ni-MOF-2 can be regenerated by purging argon. The separation performance and regeneration of Ni-MOF-2 can be retained with a flow rate of 3 mL/min, proved by cycling breakthrough tests and PXRD (Fig. 3d and Fig. S17 in Supporting information). The practical separation performances and facile reproducibility pave the way for Ni-MOF-2 as the available adsorbent to reduce the energy footprint.

To elucidate the improved separation performance of Ni-MOF-1 and Ni-MOF-2, grand canonical Monte Carlo (GCMC) simulations and density functional theory (DFT) calculations (Fig. 4) were performed on Ni-MOF-2. The aliphatic rings of btc ligands and saturated C—H groups of dabco ligands are the primary adsorption sites of C₂H₂ and CO₂ revealed by GCMC (Fig. S15 in Supporting information). As for DFT calculations, comparing with the C–H···O hydrogen bonding interactions as dominant host-guest interactions between Ni-MOF-1 and C₂H₂ or CO₂ [27], C–H···C vdW interactions and C–H···O hydrogen bonding interactions play the main role between frameworks and C₂H₂ and CO₂, respectively. As shown in Figs. 4a and b, after pillars regulation, in the site Ⅰ of Ni-MOF-2, two C–H···O (H–O, 2.52–2.53 Å) hydrogen bonding interactions and four C–H···C (3.11–3.47 Å) vdW interactions between C₂H₂ and btc/dabco ligands were observed; as for CO₂, five C–H···O (H–O, 2.78–3.43 Å) hydrogen bonding interactions and four C–H···C (2.55–3.28 Å) vdW interactions were formed. In site Ⅱ, C₂H₂ was firmly captured by twelve C–H···C (H–C, 2.63–3.50 Å) vdW interactions; however, eight C–H···O (H–O, 2.82–3.30 Å) hydrogen bonding interactions were formed between Ni-MOF-2 and CO₂ (Figs. 4c and d, Table S5 in Supporting information). The binding pattern of site Ⅲ is similar to site Ⅱ (Figs. 4e and f, Table S5). The type of host-guest interactions of C₂H₂ underwent the variation from hydrogen bonding interaction to vdW interaction after pillar modification, invariable for CO₂; and the number of host-guest interactions plainly clarifies the reason for the improved separation performance of Ni-MOF-2 compared with prototype Ni-MOF-1.

In summary, we reported two isostructural Ni-MOF-1 and Ni-MOF-2, whose molecular recognition abilities for C₂H₂ and CO₂ are diverse. The challenge lies in equimolar C₂H₂/CO₂ mixture separation in Ni-MOF-1 can be addressed via pillar variation from pyz to dabco by virtue of delicate pore environment regulation. Owing to supplemental vdW interactions with C₂H₂, pillar-modified Ni-MOF-2 achieved efficient separation of equimolar C₂H₂/CO₂ and the 3.5-fold selectivity increased than that of Ni-MOF-1. The practical separation performance, cycle stability, and facile regeneration of Ni-MOF-2 were validated by cyclic experimental breakthrough tests. DFT calculations proved that the extra C—H groups of dabco contribute to more C–H···C vdW interactions between C₂H₂ and the framework, guaranteeing preferential adsorption of C₂H₂ over CO₂ in Ni-MOF-2. This work sheds new light on delicate pore environment regulation for gas molecular recognition and suggests the potential for other challenging gas separations.

CRediT authorship contribution statement

Nan Lu: Writing – original draft, Visualization, Investigation, Formal analysis, Data curation. Lan Lan: Investigation, Formal analysis. Qiang Gao: Resources, Investigation, Formal analysis, Data curation. Ming Liu: Investigation, Funding acquisition, Formal analysis. Tong-Liang Hu: Visualization, Software, Formal analysis, Data curation. Na Li: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Formal analysis, Data curation. Xian-He Bu: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (Nos. 22035003, 22371139, 22305130), the Haihe Laboratory of Sustainable Chemical Transformations, and the Programme of Introducing Talents of Discipline to Universities (No. B18030).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.110887.