Programmed fluorine binding engineering in anion-pillared metal-organic framework for record trace acetylene capture from ethylene

Abstract

Porous physisorbents are attractive candidates for selective capture of trace gas or volatile compounds due to their low energy footprints. However, many physisorbents suffer from insufficient sorbate-sorbent interactions, resulting in low uptake or inadequate selectivity when gases are present at trace levels. Here, we report a strategy of programmed fluorine binding engineering in anion-pillared metal-organic frameworks to maximize C₂H₂ binding affinity for benchmark trace C₂H₂ capture from C₂H₄. A robust material (ZJU-300a) was elaborately designed to provide multiple-site fluorine binding model, resulting in an ultrastrong C₂H₂ binding affinity. ZJU-300a exhibits a record-high C₂H₂ uptake of 3.23 millimoles per gram (at 0.01 bar and 296 kelvin) and one of the highest C₂H₂/C₂H₄ selectivity (1672). The adsorption binding of C₂H₂ and C₂H₄ was visualized by gas-loaded ZJU-300a structures. The separation capacity was confirmed by breakthrough experiments for 1/99 C₂H₂/C₂H₄ mixtures, affording the maximal dynamic selectivity (264) and C₂H₄ productivity of 436.7 millimoles per gram.

A robust fluorinated physisorbent exhibits the record acetylene uptake and selectivity for trace acetylene capture from ethylene.

INTRODUCTION

Selective capture and removal of trace gas or volatile organic compounds represent a major contribution to many environmental and energy consumption issues. For example, during the production of the most critical feedstock ethylene (C₂H₄) through thermal cracking of hydrocarbons, a trace amount of acetylene (C₂H₂) is inevitably coproduced as an impurity (ca. 1%), which would poison the ethylene polymerization catalysts (1). Therefore, the removal of trace C₂H₂ impurity from C₂H₄ to an acceptable level [<40 parts per million (ppm)] is of prime importance to obtain polymer-grade ethylene (2). Current industrial approaches include catalytic hydrogenation using noble metal catalysts or solvent extraction of cracked olefins by organic solvents; however, they suffer from poor selectivity, high cost, and environment unfriendliness (3). Adsorption separation by porous physisorbents has shown to be more attractive than traditional techniques in view of process economy, eco-friendliness, and product purity (4–21).

Ideal physisorbents should exhibit both high adsorption selectivity and large uptake capacity for trace C₂H₂ component. Owing to the highly designable and tunable pore size and chemistry, a number of microporous metal-organic frameworks (MOFs) have been developed as promising physisorbents for efficient C₂H₂/C₂H₄ separation (22–40). However, most of the reported materials still suffer from inadequate sorbate-sorbent interactions, leading to low C₂H₂ uptake or insufficient selectivity when C₂H₂ gas is present at trace concentrations. Among them, anion-pillared SIFSIX materials represent the most promising subclass of MOF adsorbents, which consist of square lattice layers of organic ligands and metal nodes, pillared by fluorinated anions to generate a three-dimensional (3D) network with primitive cubic (pcu) topology (41–49). SIFSIX materials with pcu topology are typically divided into two categories: non-interpenetration and twofold interpenetration. The non-interpenetrated materials (e.g., SIFSIX-1-Cu and SIFSIX-3, classified into the first generation) show the preferential binding of C₂H₂ over C₂H₄ because the fluorinated anions have a stronger interaction with C₂H₂. However, these materials exhibit the relatively large pore sizes (>4.2 Å) and insufficient C₂H₂ binding affinity with a moderate adsorption enthalpy (Q_st) of ca. 30 kJ mol⁻¹, delimiting their C₂H₂/C₂H₄ selectivity less than 10 (47). To further improve gas selectivity, the interpenetration strategy was used to fine-tune pore size/chemistry in SIFSIX platform. Some twofold interpenetrated SIFSIX materials (classified into the second generation) were developed to boost C₂H₂/C₂H₄ separation (47, 48). Such interpenetration architecture was found to enable each C₂H₂ molecule to simultaneously bind with two F atoms from different nets through C─H···F hydrogen bonding (Fig. 1), leading to much stronger C₂H₂ binding energy (ca. 40 kJ mol⁻¹) than the first-generation materials. For example, SIFSIX-2-Cu-i with aperture size of 4.4 Å exhibits both improved uptake capacity and gas selectivity (44) for trace C₂H₂ component (47). Fine-tuning of pore size to 3.4 Å in SIFSIX-14-Cu-i (also called UTSA-200a) resulted in the nearly molecular sieving of C₂H₂ from C₂H₄ with the record selectivity up to 6000 (48), surpassing all the previous benchmark materials. However, the two-site fluorine binding model of the second-generation materials limited their trace C₂H₂ uptake not so high (1.5 to 1.8 mmol g⁻¹ at 0.01 bar) due to the inadequate binding energy. Furthermore, the stability investigations of SIFSIX-14-Cu-i in a moist environment revealed a structural phase change, which resulted in the loss of C₂H₂ capture and thus precluded the practical applications (48). Therefore, the state-of-the-art SIFSIX materials still suffer from the insufficient low-pressure C₂H₂ uptake and poor moisture stability, and there is a lack of ideal and stable materials to have both ultrahigh C₂H₂ uptake and selectivity for trace C₂H₂ removal from C₂H₄.

Fig. 1. Programmed fluorine binding engineering in anion-pillared SIFSIX materials for C₂H₂/C₂H₄ separation, indicating the proposed strategy of designing multiple fluorine binding interactions to target both ultrahigh C₂H₂ capture and selectivity.

To address the above drawbacks, we sought to develop a new generation of robust SIFSIX material that affords a superior C₂H₂ uptake capacity at 0.01 bar while maintaining the record-high gas selectivity. As we know, when gas selectivity reaches a very high level, the separation performance for 1/99 C₂H₂/C₂H₄ mixtures would be mainly dominated by trace C₂H₂ uptake at 0.01 bar (40, 48). To maximize the low-pressure C₂H₂ uptake capacity, it is important to construct stronger binding sites in SIFSIX materials than that in the second generation. However, SIFSIX materials with pcu topology have shown the limitation to achieve the above targets due to the insufficient fluorine binding model. We envisioned that, if we can construct a unique type of multiple-site fluorine binding model that enables more fluorine atoms to simultaneously interact with C₂H₂ (Fig. 1), then it is highly potential to maximize C₂H₂ binding affinity and thus boost trace C₂H₂ removal for C₂H₄ purification. By programmed fluorine binding engineering, we here designed and synthesized a stable SIFSIX material, [Zn(TPE)SiF₆]_n [termed as ZJU-300, 1,1,2,2-tetra(pyridin-4-yl)ethene (TPE)]. This material features a non-interpenetrated framework with fsc (four, six type c) topology (50) and appropriate channel pore aperture size of 3.8 Å (Fig. 2). Such structural architecture enables SiF₆²⁻ anions to be parallelly adjacent in close proximity with a F···F distance of 3.8 Å, offering the proposed multiple-site fluorine binding model for trace C₂H₂ capture. Single-crystal x-ray diffraction (SCXRD) experiments on gas-loaded ZJU-300a visually unveiled that such multiple-site fluorine binding model allows eight fluorine atoms to concurrently and strongly interact with C₂H₂ molecule, resulting in an ultrastrong C₂H₂ binding affinity with the highest Q_st value (61.1 kJ mol⁻¹) among SIFSIX materials. The activated ZJU-300a thus exhibits the record-high C₂H₂ uptake of 3.23 mmol g⁻¹ (at 0.01 bar and 296 K) and one of the highest C₂H₂/C₂H₄ selectivity (1672), setting up a benchmark for this separation. Dynamic breakthrough experiments on 1/99 C₂H₂/C₂H₄ mixtures provide both record-high dynamic separation selectivity (264) and C₂H₄ purification capacity of 436.7 mmol g⁻¹, several times higher than the reported state-of-the-art SIFSIX-14-Cu-i (91 and 101.3 mmol g⁻¹) (48) and zeolite Ni@faujasite (Ni@FAU) (97 and 116.8 mmol g⁻¹) (32).

Fig. 2. Description of the crystal structure of ZJU-300.

(A) Scheme illustration of the first-generation and second-generation SIFSIX materials with pcu topology. (B) Scheme illustration of a type of SIFSIX materials, built from four-connected N-donor ligands to construct the fsc topology and rhombic pores containing parallelly adjacent SiF₆²⁻ anions. (C) Single-crystal structure of ZJU-300 viewed along with the b axis, exhibiting three types of pore cavities served as potential C₂H₂ binding sites (H atoms have been omitted for clarity). (D) The staggered arrangement feature of pore cavities. Color code: Zn (purple), Si (yellow), F (red), N (blue), and C (gray).

RESULTS

Synthesis and characterization

The organic ligand of TPE was simply synthesized through a multistep reaction procedure (fig. S1). The slow diffusion of a methanol solution of ZnSiF₆·6H₂O into a TPE solution for 1 week produced ZJU-300 as colorless block crystals suitable for single-crystal x-ray analysis. The powder samples can be facilely produced by an alternative fast mixing method. A methanol solution (5 ml) of ZnSiF₆·6H₂O (0.125 mmol) was added to a methanol solution (15 ml) of TPE (0.15 mmol) at 80°C, and the mixtures were kept at 80°C for as short as 10 min to obtain a white powder. This facile synthesis indicates its potential to be produced at large scale. The purity of the bulk sample was confirmed by powder x-ray diffraction patterns (PXRDs) and gas adsorption measurements (fig. S3). Scanning electron microscopy (SEM) images revealed that the single crystals and rapidly synthesized particles of ZJU-300 show the same cubic morphology but with different dimensions (fig. S4).

The SCXRD studies indicated that the framework of ZJU-300 crystallizes in a tetragonal P4/mcc space group (No. 124). As depicted in Fig. 2 (A and B), the use of different types of organic linkers to construct SIFSIX materials could result in different structure network and pore characteristic. In general, the first- and second-generation SIFSIX materials are built from two-connected N-donor ligands to construct the pcu topology and square pores (41–49). However, the use of four-connected organic linkers can lead to a different structure topology (51–53). As shown in Fig. 2C, four terminal nitrogen atoms of TPE linkers coordinate with zinc(II) atoms to construct a 2D [Zn(TPE)]_n layer. The 2D layers are further pillared by SiF₆²⁻ anions in the third dimension to form a 3D network without interpenetration. Unlike the pcu topology in square grid SIFSIX materials, the resulting ZJU-300 holds a 4,6-connected fsc network (Fig. 2B). ZJU-300 exhibits two types of small rhombic pore channels with the same window size of 3.8 Å by 3.1 Å, viewed along the c axis. In addition, another type of pore channels with the aperture size of 3.8 Å by 2.7 Å was observed along the a and b axes. The aperture sizes of all the pore channels are larger than the kinetic diameter of C₂H₂ (3.3 Å) but smaller than that of C₂H₄ (4.2 Å). Because of the non-interpenetration feature, the SiF₆²⁻ anions in ZJU-300 are parallelly adjacent and in close proximity to each other in the same layer, with a short F···F distance of 3.8 Å after subtracting van der Waals radius. Such adjacent SiF₆²⁻ anions combined with TPE linkers can form three types of pore cavities with small size of 3.8 Å. In each cavity, eight F atoms from adjacent SiF₆²⁻ anions point toward the center of the cavity with the average distance of 3.2 and 4.0 Å between these F atoms and center of the cavity, which creates multiple-site fluorine binding model to provide very strong adsorption sites for C₂H₂ capture. In comparison, the SiF₆²⁻ anions in the benchmark SIFSIX-14-Cu-i are stagger disposed along pore channels due to the twofold interpenetration, restricting only two F atoms to interact with C₂H₂ molecule (fig. S16) (48). Evidently, the unique multiple fluorine binding sites observed in ZJU-300 enable more F atoms to interact with C₂H₂ compared to SIFSIX-14-Cu-i, probably leading to much stronger C₂H₂ adsorption affinity. Furthermore, there exist a large amount of pore cavities densely and orderly arranged in ZJU-300, in which cavity I and cavity III are staggered with each other along the b axis, and cavity II and cavity III show a staggered arrangement along the a axis (Fig. 2D). Therefore, the dense pore cavities with multiple fluorine binding sites not only can provide an ultrastrong binding affinity for trace C₂H₂ capture but also can enable the dense packing of C₂H₂ molecules within the pores, potentially affording both ultrahigh C₂H₂ capture capacity and selectivity for trace C₂H₂ removal.

The permanent porosity of activated ZJU-300a was determined by nitrogen (N₂) sorption isotherms at 77 K. As revealed in Fig. 3A, ZJU-300a exhibits a typical type I sorption behavior, and the saturated N₂ uptake is up to 244 cm³ g⁻¹. The Brunauer-Emmett-Teller surface area and pore volume of ZJU-300a were calculated to be 796.2 m² g⁻¹ and 0.38 cm³ g⁻¹, respectively, which are 30.1 and 40.7% higher than those of SIFSIX-14-Cu-i (612.0 m² g⁻¹ and 0.27 cm³ g⁻¹) (fig. S5) (48). Such larger surface area/pore volume observed in ZJU-300a can provide more pore spaces to capture more C₂H₂ molecules and thus improve C₂H₂ adsorption amount. The pore size distribution, determined by the Horvath-Kawazoe model based on N₂ adsorption at 77 K, shows a pore size of 4.1 Å (Fig. 3A), close to the result observed from the crystal structure.

Fig. 3. Gas sorption properties.

(A) N₂ adsorption isotherm of ZJU-300a at 77 K. Inset shows pore size distribution for ZJU-300a based on Horvath-Kawazoe model. BET, Brunauer-Emmett-Teller. STP, standard temperature and pressure. P/P₀, the ratio of actual preesure to saturated vapor pressure. (B) Adsorption isotherms of ZJU-300a for C₂H₂ (red) and C₂H₄ (blue) at 296 K. (ads, adsorption; des, desorption). (C) Comparison of C₂H₂ uptake (0 to 0.04 bar) among the representative SIFSIX materials at room temperature. (D) Comparison of C₂H₂ capture capacity for ZJU-300a with SIFSIX materials and other best-performing materials at 0.01 bar. (E) Comparison of C₂H₂ uptake at 0.01 bar and 1/99 C₂H₂/C₂H₄ selectivity for ZJU-300a with the top-performing materials reported.

Gas adsorption measurements

Single-component adsorption isotherms of ZJU-300a for C₂H₂ and C₂H₄ were collected at different temperatures up to 1 bar (Fig. 3B and figs. S6 to S8), respectively. As presented in Fig. 3 (B and C), ZJU-300a exhibits an extremely steep C₂H₂ adsorption increase at the low-pressure region and 296 K, much faster than that of the first- and second-generation materials (e.g., SIFSIX-1-Cu, SIFSIX-14-Cu-i, and SIFSIX-2-Cu-i). This indicates that ZJU-300a shows a much stronger binding affinity toward C₂H₂. At a partial pressure of 0.01 bar, as an indicator of the C₂H₂ capture ability of adsorbents from 1/99 C₂H₂/C₂H₄ mixture, ZJU-300a shows a record-high C₂H₂ uptake amount of 3.23 mmol g⁻¹, which is substantially higher than all the SIFSIX materials and almost two times higher than that of the benchmark SIFSIX-14-Cu-i (1.83 mmol g⁻¹) (48). As shown in Fig. 3D, this C₂H₂ uptake at 0.01 bar notably surpasses the other state-of-the-art materials reported so far, including Ni@FAU (1.72 mmol g⁻¹) (32), Cu(4,4'-dipyridylsulfone)₂(TiF₆) (ZUL-100) (2.96 mmol g⁻¹) (40), and Cu₂(1,3,5,7-adaman-tane tetracarboxylic acid) (ATC-Cu) (2.54 mmol g⁻¹) (54). Under 296 K and 1 bar, the C₂H₂ uptake of ZJU-300a gradually increases to 5.37 mmol g⁻¹, which is also much higher than that of SIFSIX-14-Cu-i (3.65 mmol g⁻¹) due to its increased surface area (fig. S9) (48). Compared with C₂H₂ adsorption, the C₂H₄ uptake in ZJU-300a increases much slowly in the whole pressure ranges with a much lower uptake of 2.39 mmol g⁻¹ at 296 K and 1 bar (Fig. 3B), affording an obviously preferential adsorption of C₂H₂ over C₂H₄. This large adsorption difference between C₂H₂ and C₂H₄ can be well supported by the experimental isosteric heat of adsorption (Q_st), wherein the initial Q_st value of C₂H₂ (61.1 kJ mol⁻¹) is notably higher than that of C₂H₄ (39.7 kJ mol⁻¹) found in ZJU-300a (fig. S10). The C₂H₂ Q_st value of ZJU-300a is also higher than that of SIFSIX-1-Cu (30.0 kJ mol⁻¹) (47), SIFSIX-3-Zn (21.0 kJ mol⁻¹) (47), and SIFSIX-14-Cu-i (40.0 kJ mol⁻¹) (48), further confirming its strongest C₂H₂ binding affinity among SIFSIX materials (fig. S13). In addition, we also investigated the time-dependent adsorption kinetics profiles of rapidly synthesized ZJU-300a at 296 K. As shown in fig. S14, ZJU-300a shows a notably faster initial rate for C₂H₂ adsorption than C₂H₄, and both gas adsorption reaches equilibrium rapidly within 10 min. Fitting the adsorption data with micropore diffusion model (55) affords a high kinetic selectivity of 19.2, which can further promote its high separation performance for C₂H₂/C₂H₄ mixtures.

Next, we used ideal adsorbed solution theory (IAST) to calculate the adsorption selectivity of ZJU-300a for 1/99 C₂H₂/C₂H₄ mixture (56). As shown in fig. S15, owing to the record-high C₂H₂ capture capacity at 0.01 bar and low C₂H₄ uptake, ZJU-300a exhibits one of the highest IAST selectivity of 1672 at 296 K and 1 bar, far surpassing that of the first- and second-generation SIFSIX materials except SIFSIX-14-Cu-i (Fig. 3E). We note that this selectivity is also notably higher than the other benchmark materials, such as Ni@FAU (97) (32), ZUL-100 (175) (40), and Ni₃(pzdc)₂(7Hade)₂ (168) (31). As we know, adsorption selectivity and uptake capacity have been deemed as the two most important criteria to evaluate the separation performance of adsorbents. As shown in Fig. 3E, the first-generation SIFSIX materials show both low C₂H₂ uptake at 0.01 bar and poor gas selectivity due to their relatively large pore size and insufficient binding affinity. Fine-tuning of pore size and interpenetration in the second-generation leads to both notably enhanced C₂H₂ uptake and selectivity, wherein the representative SIFSIX-14-Cu-i exhibits the record gas selectivity (48). By optimizing the ultrastrong multiple-site fluorine binding model, ZJU-300a displays the markedly improved C₂H₂ uptake (3.23 versus 1.83 mmol g⁻¹) while maintaining the comparably high selectivity in comparison to SIFSIX-14-Cu-i (Fig. 3E). When compared with other best-performing materials, ZJU-300a also achieves both the highest C₂H₂ uptake capacity and gas selectivity so far, making it as the current benchmark for C₂H₂/C₂H₄ separation.

Single-crystal diffraction studies on gas-loaded ZJU-300a

To gain better insight into the record C₂H₂ capture capacity and selectivity of ZJU-300a, we performed the SCXRD experiments on both C₂H₂- and C₂H₄-loaded ZJU-300a crystals to directly visualize the locations of the adsorbed C₂H₂ and C₂H₄ molecules. According to the SCXRD analysis, C₂H₂@ZJU-300a was found to exhibit three types of binding sites (site 1, site 2, and site 3) for C₂H₂ molecules (Fig. 4, A and B), which are preferentially located at two rhombic pore channels (cavity I and cavity II in Fig. 2) and at the middle of two adjacent TPE ligands (cavity III). As shown in Fig. 4 (C and D), the adsorbed C₂H₂ molecules on site 1 and site 2 exhibit the same binding interactions with the host framework. Each C₂H₂ molecule at these two sites strongly interacts with eight F atoms originated from two adjacent SiF₆²⁻ anions through cooperative eight C─H···F hydrogen bonds, with the distances of 2.11 to 3.84 Å. Furthermore, the C₂H₂ molecule is also bound with eight C─H groups from four surrounding pyridine rings of TPE linkers through supramolecular interactions (C_C2H2···H = 3.03 to 3.36 Å). The site 3 C₂H₂ molecule located at the middle of two adjacent TPE ligands surrounded by four SiF₆²⁻ anions (Fig. 4E). Similarly, each C₂H₂ molecule at this site is cooperatively bound with two SiF₆²⁻ anions through two C─H···F hydrogen bonding and with eight pyridine rings of TPE linkers through eight supramolecular interactions (C_C2H2···H = 2.84 to 3.33 Å). In comparison, the two-interpenetrated SIFSIX materials (e.g., SIFSIX-14-Cu-i) have been revealed to show one type of binding sites, in which each C₂H₂ molecule can interact with two SiF₆²⁻ sites from different nets through two C─H···F hydrogen bonds (fig. S16) (48). Evidently, the non-interpenetration framework with fsc topology in ZJU-300a can create a unique multiple fluorine binding site to offer more numbers of C─H···F hydrogen bonds and additional supramolecular interactions with C₂H₂, thus resulting in the much stronger C₂H₂ binding affinity. Furthermore, because of the dense distribution of these three binding sites, the site 1 adsorbed C₂H₂ interacts with two adjacent C₂H₂ molecules in site 3 through the C─H···C interactions (2.71 and 3.84 Å), forming the interlaced C₂H₂ chains within the channels along the b axis (Fig. 4A). Similarly, each C₂H₂ molecule in site 2 is also bounded to two site 3 C₂H₂ molecules (4.33 and 4.35 Å) along the a axis. Therefore, the dense and strong multiple fluorine binding sites observed in ZJU-300a not only provide the ultrastrong binding affinity for C₂H₂ capture but also enable the dense packing of C₂H₂ within the pores, thus contributing to its record-high uptake capacity at 0.01 bar. The SCXRD data indicate that the adsorbed C₂H₂ amount in site 1, site 2, and site 3 corresponds to 1.84, 1.84, and 1.84 mmol g⁻¹ gas uptake (fig. S17 and more detailed explanation in the Supplementary Materials), and the total value of 5.52 mmol g⁻¹ is very close to the experimental C₂H₂ uptake (5.37 mmol g⁻¹) at 296 K and 1 bar.

Fig. 4. Adsorption binding sites of C₂H₂ and C₂H₄ in ZJU-300a.

(A) The SCXRD structure of C₂H₂-loaded ZJU-300a viewed along the c axis, indicating three types of binding sites for C₂H₂ and the intermolecular interactions between the adsorbed C₂H₂ molecules. (B) The dense packing of C₂H₂ in the whole structure of ZJU-300a, viewed along the b axis. (C) Illustration of the binding interactions of C₂H₂ molecules in ZJU-300a at site 1, (D) site 2, and (E) site 3, as determined by SCXRD studies. (F) Illustration of the binding interactions of C₂H₄ molecule in ZJU-300a obtained by SCXRD studies.

In contrast, the crystal structure of C₂H₄@ZJU-300a revealed that the adsorbed C₂H₄ molecules show one type of binding sites, primarily located at the corner of the rhombic pore channels (fig. S18). As shown in Fig. 4F, each C₂H₄ molecule can interact with two SiF₆²⁻ anions through three C─H···F hydrogen bonding (2.29 to 3.48 Å) and with six pyridine rings of TPE linkers through supramolecular interactions (C_C2H2···H = 2.85 to 3.50 Å). Compared with that of C₂H₂, C₂H₄ molecule shows a much weaker interaction with the framework because of its much less and weaker C─H···F hydrogen bonds. This can be further confirmed by the higher experimental Q_st value of C₂H₂ (61.1 kJ mol⁻¹) than that of C₂H₄ (39.7 kJ mol⁻¹). Thus, the more number of binding sites and much higher binding affinity toward C₂H₂ afford the ultrahigh C₂H₂/C₂H₄ selectivity of ZJU-300a. All of these results can visually elucidate the mechanisms of both ultrahigh C₂H₂ adsorption and selectivity observed in ZJU-300a.

Column breakthrough experiments

To validate the excellent separation performance of ZJU-300a for actual 1/99 C₂H₂/C₂H₄ mixtures, the dynamic breakthrough experiments were conducted on a packed column of activated ZJU-300a at 296 K under a total gas flow of 5 ml min⁻¹. For a comparison, the breakthrough experiments on the fixed bed adsorbers packed with the first-generation SIFSIX-1-Cu and the second-generation SIFSIX-14-Cu-i were also measured under the same conditions. The experimental breakthrough curves for all the selected SIFSIX materials are depicted in Fig. 5A. Efficient separation can be accomplished by all these materials for 1/99 C₂H₂/C₂H₄ mixtures. The C₂H₂ breakthrough time was observed to be 128 and 512 min g⁻¹ for SIFSIX-1-Cu and SIFSIX-14-Cu-i, affording a polymer-grade C₂H₄ productivity of 20.6 and 101.3 mmol g⁻¹, respectively. These C₂H₄ productivities are slightly higher than the values reported in the literatures (17.7 and 85.7 mmol g⁻¹ for SIFSIX-1-Cu and SIFSIX-14-Cu-i) (47, 48). Similarly, a highly efficient separation for C₂H₂/C₂H₄ mixture was also achieved by ZJU-300a: The C₂H₄ gas eluted through the adsorption bed immediately in a high-purity grade, whereas C₂H₂ was retained in the packed column over 2058 min g⁻¹. This C₂H₂ breakthrough time of ZJU-300a is even 4.1 times longer than that observed in SIFSIX-14-Cu-i. According to the breakthrough curves, the C₂H₂/C₂H₄ dynamic selectivity of ZJU-300a was calculated to be 264 (Fig. 5C and fig. S19), notably higher than all the reported materials including the previous state-of-the-art Ni@FAU (97) (32) and SIFSIX-14-Cu-i (91). During the breakthrough process, ultrapure C₂H₄ (C₂H₂ < 1 ppm) productivity from the outlet effluent for ZJU-300a was determined to be record high of 436.7 mmol g⁻¹. As revealed in Fig. 5B, this polymer-grade C₂H₄ productivity far surpasses all the SIFSIX materials, including four times larger than SIFSIX-14-Cu-i (101.3 mmol g⁻¹). Note that, although ZJU-300a shows a slightly lower IAST selectivity than SIFSIX-14-Cu-i, the much higher C₂H₂ uptake at 0.01 bar (3.23 versus 1.83 mmol g⁻¹) led to its four times higher C₂H₄ productivity in breakthrough experiments. This result verifies that, when gas selectivity reaches a very high level, C₂H₂ uptake capacity at 0.01 bar would play the most important role to determine the final C₂H₄ productivity. When compared with the other best-performing materials, ZJU-300a also shows the highest C₂H₄ productivity (Fig. 5C), much larger than that of ZUL-100 (134.1 mmol g⁻¹) (40) and Ni@FAU (116.8 mmol g⁻¹) (32). Subsequently, multiple breakthrough experiments on C₂H₂/C₂H₄ mixed gas indicate that the separation capacity of ZJU-300a can be recycled at least 15 continuous cycles (Fig. 5D), confirming its feasible recyclability and regeneration capability. As inferred from the PXRD and gas adsorption studies on associated samples, ZJU-300a can maintain its structural integrity after multiple breakthrough experiments (figs. S20 to S22).

Fig. 5. Breakthrough experiments and stability characterization.

(A) Experimental column breakthrough curves for 1/99 C₂H₂/C₂H₄ separation with ZJU-300a, SIFSIX-14-Cu-i, and SIFSIX-1-Cu at a flow rate of 5 ml min⁻¹ under ambient conditions, respectively. (B) Comparison of experimental C₂H₄ productivity from C₂H₂/C₂H₄ (1/99) mixture through fixed bed adsorbers packed with ZJU-300a and representative SIFSIX materials. (C) Comparison of C₂H₄ productivity and dynamic selectivity of ZJU-300a with other top-performing materials, obtained from the breakthrough curves. (D) The cycling tests of 1/99 C₂H₂/C₂H₄ mixture with a flow rate of 10 ml min⁻¹ in an absorber bed packed with ZJU-300a under ambient conditions. C/C₀, the ratio of gas concentration at the outlet and inlet. (E) PXRD patterns of ZJU-300 samples after treatment under different conditions. (F) C₂H₂ adsorption isotherms at 296 K treated under various conditions.

Stability characterization

Given that the feed gases in the practical C₂H₂ removal unit often contain a small amount of water and acidic gases (57), the extreme separation conditions require adsorbents with high and long-term water and pH stability. Unfortunately, most of square grid SIFSIX materials with pcu topology were observed to decompose or undergo phase transformations under low levels of water and moisture conditions (58). As shown in figs. S23 to S28 and table S5, our PXRD experiments confirmed that the crystallinity of most representative SIFSIX materials (e.g., SIFSIX-1-Cu, SIFSIX-2-Cu-i, and SIFSIX-14-Cu-i) was lost after exposure to the moisture or water solution, severely dampening their industrial applications. However, ZJU-300 with fsc topology exhibits an extremely high and long-term stability toward water and acid-base environments. As depicted in Fig. 5E, the framework of ZJU-300 can retain its structural integrity without phase change or loss of crystallinity observed after exposed to air for 3 months or water for 10 days, as revealed by PXRD patterns. Furthermore, the ZJU-300 samples were immersed in HCl (pH 3) and NaOH (pH 11) solutions for 1 day, respectively. The PXRD patterns indicate that the framework also can maintain structural stability with no phase transformation observed. The C₂H₂ uptake capacities at 296 K after the above treatments are very close to those of the pristine material (Fig. 5F), further confirming its ultrahigh chemical stability. Variable temperature PXRD patterns revealed that ZJU-300 shows a high thermal stability up to 200°C (fig. S30). The high stability encouraged us to further evaluate its separation performance for a 1/99 C₂H₂/C₂H₄ mixture under 40% humidity conditions. As shown in fig. S31, ZJU-300a can almost retain its exceptional separation capacity with slightly decreased dynamic selectivity (252) and C₂H₄ productivity (430.3 mmol g⁻¹), indicating that the presence of water vapor has a slight effect on the separation capacity. Given that the feed streams in industry typically contain a small amount of acidic gas (e.g., H₂S) (59), the breakthrough experiments for a 1/99 mixture with 1000 ppm H₂S were also performed on ZJU-300a to investigate the effect of acidic gases. As shown in fig. S32, the largely overlapped breakthrough curves reveal that ZJU-300a can retain its high separation capacity under this condition. Overall, compared with the previously best SIFSIX-14-Cu-i, ZJU-300 exhibits both the substantially improved 0.01-bar C₂H₂ capture capacity and water/pH stability while maintaining the ultrahigh gas selectivity, making it as the current benchmark for the important C₂H₂/C₂H₄ separation. In comparison to the state-of-the-art zeolites (e.g., Ni@FAU and Na-X) (32, 35), although ZJU-300a holds some superiorities in gravimetric C₂H₂ uptake and C₂H₄ productivity, these zeolites exhibit comparable or higher volumetric gas capacities along with lower adsorption enthalpy (table S3). Combined with their high selectivity, notable stability, and low cost, these zeolites are still placed among the best choices for this separation by far, especially from the point of view of practical application.

DISCUSSION

We have successfully proposed and demonstrated a strategy of engineering multiple fluorine binding sites in anion-pillared materials to maximize trace C₂H₂ capture for C₂H₄ purification. The resulting ZJU-300a features the unique multiple-site fluorine binding model, providing one of the strongest binding affinity to afford the benchmark C₂H₂ capture capacity at 0.01 bar. This material thus demonstrates a record high C₂H₂ uptake of 3.23 mmol g⁻¹ (at 296 K and 0.01 bar) and one of the highest C₂H₂/C₂H₄ selectivity reported to date. The fundamental C₂H₂ binding and separation mechanisms have been illustrated by gas-loaded SCXRD studies. Breakthrough experiments on ZJU-300a revealed both unprecedented dynamic selectivity (264) and C₂H₄ productivity of 436.7 mmol g⁻¹ for actual C₂H₂/C₂H₄ mixtures, far exceeding all the reported materials including SIFSIX-14-Cu-i (91 and 101.3 mmol g⁻¹) and Ni@FAU (97 and 116.8 mmol g⁻¹). Combined with its facile synthesis and ultrahigh water stability, this material represents by far the benchmark adsorbent for trace C₂H₂ removal from C₂H₂/C₂H₄ mixtures. This programmed gas binding engineering revealed in this work may provide some guidance to facilitate the design of ideal physisorbents with ultrahigh binding affinity for highly efficient capture of trace gas or volatile organic compounds.

MATERIALS AND METHODS

Chemicals

All raw chemicals and reagents were commercially available and used directly without further purification. TPE was simply synthesized through a multistep reaction procedure (fig. S1 and more details in the Supplementary Materials). N₂ (99.999%), C₂H₂ (99.6%), C₂H₄ (99.9%), He (99.999%), and mixed gases of C₂H₂/C₂H₄ = 1/99 (v/v) were purchased from JinGong Company (China).

Crystallization of [Zn(TPE)SiF₆]_n (ZJU-300)

A solution of ZnSiF₆·6H₂O (0.518 mg, 0.0025 mmol) in 0.1 ml of methanol was added to a narrow glass tube. Methanol (2 ml) was carefully layered over this to act as a buffer layer. Last, a solution of TPE (1.01 mg, 0.003 mmol) in 0.3 ml of methanol was layered over the buffer layer, and the tube was left undisturbed for 1 week. Colorless block crystals of ZJU-300 were obtained.

Bulk synthesis of [Zn(TPE)SiF₆]_n (ZJU-300)

ZJU-300 was synthesized by slowly mixing a methanol solution (5 ml) of ZnSiF₆·6H₂O (25.9 mg, 0.125 mmol) with a methanol solution (15 ml) of TPE (50.5 mg, 0.15 mmol) at 80°C in a 25-ml glass vial. The mixtures were then kept in an oven at 80°C for 10 min, followed by slowly cooling to room temperature. A white powder was obtained by filtration and washed with methanol.

Sample characterization

¹H nuclear magnetic resonance spectra were recorded on a Bruker Advance DMX500 spectrometer using tetramethylsilane as the internal standard. The PXRD patterns were measured in a range of 2θ = 2° to 45° on an X’Pert PRO diffractometer using a Cu-Kα (λ = 1.54184 Å) radiation source at room temperature. Thermogravimetric analysis was performed on a Netzsch TG209F3 instrument, and the sample was heated under N₂ atmosphere with a heating rate of 5 K min⁻¹. The SEM images were observed by Hitachi S-4800 field emission SEM.

Gas sorption measurements

Before the gas sorption test, the fresh sample was first solvent-exchanged with dry methanol at least eight times within 3 days to completely exchange the guest solvent molecules in the framework. The solvent-exchanged sample was evacuated at room temperature for 12 hours and further at 363 K for 12 hours until the outgas rate was 4 μmHg min⁻¹ before measurements. Single-component gas adsorption isotherms of C₂H₂ and C₂H₄ were recorded by using a Micromeritics ASAP 2020 surface area analyzer, and a Julabo water bath was used to keep the adsorption tube at a constant temperature of 273, 283, 296, and 313 K, respectively. N₂ sorption isotherms were obtained on a Micromeritics ASAP 2460 instrument, and the sorption measurement was maintained at 77 K under liquid nitrogen bath. Kinetic sorption measurements were conducted by the Intelligent Gravimetric Analyzer (IGA001, Hiden, UK) under 296 K and 1 bar.

Single-crystal x-ray diffraction

SCXRD data were collected at 284 K for ZJU-300 and at 200 K for C₂H₂@ZJU-300a and C₂H₄@ZJU-300a on an Agilent Supernova charge-coupled device diffractometer equipped with graphite-monochromatic enhanced Mo-Kα radiation (λ = 0.71073 Å). A single crystal of solvent-exchanged ZJU-300 was selected and put into a capillary glass tube with inner diameter of 0.2 mm. This crystal was evacuated at 363 K for 12 hours, and the capillary glass tube was filled by pure C₂H₂ or C₂H₄ gas up to 1 bar and then sealed to obtain C₂H₂-loaded or C₂H₄-loaded ZJU-300a crystal. The datasets were corrected by empirical absorption correction using spherical harmonics, implemented in the SCALE3 ABSPACK scaling algorithm. The structure was solved by direct methods and refined by full-matrix least-squares methods with the SHELX-97 program package (60). The crystal data are summarized in table S1.

Breakthrough experiments

The breakthrough experiments were performed using a dynamic gas breakthrough apparatus equipped with stainless steel column (4.0 mm inner diameter by 120 mm). The mixed gas flows of C₂H₂/C₂H₄ (1/99, v/v) were introduced into breakthrough apparatus with the rate of 5 and 10 ml min⁻¹ at 296 K and 1 bar, respectively. The weight of activated ZJU-300a sample packed in the column was 0.352 g. For comparison, the breakthrough experiments were further carried out in a packed column of activated SIFSIX-14-Cu-i/UTSA-200 (0.205 g) and SIFSIX-1-Cu (0.380 g) for 1/99 C₂H₂/C₂H₄ gas mixture under a total flow of 5 ml min⁻¹ and ambient conditions, respectively. The outlet gas from the column was monitored using gas chromatography (GC-2014C, SHIMADZU) equipped with a thermal conductivity detector (detection limit, 0.1 ppm). The standard gas mixtures were detected to calibrate the concentration of the outlet gas. After every breakthrough experiment, the sample can be generated under a purging He gas with a flow of 10 ml min⁻¹ at 363 K for 4 hours.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S32

Tables S1 to S5

References