Zhoqing Zhng,Xili Cui,Xioming Jing,Qi Ding,Jiyu Cui,Yunin Zhng,Youssef Belmkhout,Krim Adil, Mohmed Eddoudi, Huin Xing,*

a Key Laboratory of Biomass Chemical Engineering of Ministry of Education, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China

b State Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002, China

c Advanced Membranes and Porous Materials Center, King Abdullah University of Science and Technology, Thuwal 23955, Saudi Arabia

Keywords:Adsorption and separation Trans-/cis-butene Ultramicroporous metal–organic frameworks Pore engineering Guest-adaptive

ABSTRACT Trans-/cis-olefin isomers play a vital role in the petrochemical industry. The paucity of energy-efficient technologies for their splitting is mainly due to the similarities of their physicochemical properties.Herein, two new tailor-made anion-pillared ultramicroporous metal–organic frameworks (MOFs), ZU-36-Ni and ZU-36-Fe(GeFSIX-3-Ni and GeFSIX-3-Fe)are reported for the first time for the efficient trans-/cis-2-butene (trans-/cis-C4H8) mixture splitting by enhanced molecular exclusion. Notably, ZU-36-Ni unexpectedly exhibited smart guest-adaptive pore channels for trapping trans-C4H8 with a remarkable adsorption capacity (2.45 mmol?g-1) while effectively rejecting cis-C4H8 with a high purity of 99.99%.The dispersion-corrected density functional theory (DFT-D) calculation suggested that the guestadaptive behavior of ZU-36-Ni in response to trans-C4H8 is derived from the organic linker rotation and the optimal pore dimensions, which not only improve the favorable adsorption/diffusion of trans-C4H8 with optimal host–guest interactions, but also enhance the size-exclusion of cis-C4H8. This work opens a new avenue for pore engineering in advanced smart or adaptive porous materials for specific applications involving guest molecular recognition.

1. Introduction

Trans-/cis-isomers differ only in the spatial arrangement of the atoms.Remarkably,such a minor variation imparts significant differences in their reactivity in organic synthesis and pharmacological activity. Trans-/cis-olefins have important applications in chemical research and processing. For example, trans- and cis-2-butenes (C4H8), the simplest olefins displaying trans-/cisisomerism,are the basic raw materials for producing various types of polymers and organic chemicals. Notably, cis-C4H8is a crucial feedstock for the production of maleic acid, butadiene, and polymers. However, the inevitable presence of trans-C4H8as an impurity in cis-C4H8adversely impacts the quality of the products [1–4].High-purity trans-C4H8(>95%)is significant for several applications, such as the production of propylene via the metathesis of trans-C4H8and ethylene [5,6]. Therefore, it is highly necessary to separate cis-C4H8and trans-C4H8. The similarity in the molecular structures and boiling points (Fig. 1(a) and Table S1 in Appendix A)of trans-/cis-olefin isomers poses great challenges in their separation [7–10]. Furthermore, 2-C4H8is highly reactive and tends to undergo copolymerization or dimerization at elevated temperatures. This characteristic renders the isolation of the high purity individual 2-C4H8isomers highly challenging via the traditional energy-intensive extractive distillation [9,11–12]. Size-selective physisorption using ultramicroporous materials is a promising energy-efficient alternative and has been demonstrated as a promising candidate for the efficient separation of light hydrocarbon mixtures[13–16]. However, to the best of our knowledge,the efficient separation of trans-/cis-olefin isomers by porous materials has rarely been reported.

Fig. 1. Schematic of representative robust porous materials with (a) rigid pore structures and (b) a typical Langmuir adsorption isotherm for microporous materials. Novel porous materials with (c) guest-adaptive pore channels and (d) corresponding desired stepped adsorption isotherm for increasing the working capacity. 1 bar = 105 Pa.

Metal–organic frameworks (MOFs) or porous coordination polymers (PCPs) are custom-tailored porous crystalline materials with tunable pore chemistry. This kind of material has recently been successfully used as systems for separating mixtures of varying degrees of complexity [16], such as paraffins and olefins[13,17–21], olefins and alkynes [23–31], n-isomer and iso-isomer mixtures [12,32–34], and other analogous molecules. However,achieving high efficiency for the high-level separation complexity using MOFs is still exceedingly challenging [16] when the differences in size and shape between the probes, such as trans-/cisisomers, are subtle. The trans-/cis-isomer separation selectivity,and particularly the diffusivity, achieved using the current stateof-the-art MOFs is still not suitable for use in industrial processes in comparison to distillation [9]. For example, zeolitic imidazolate framework-7 (ZIF-7) with its narrow pore window size and structural flexibility exhibits gate-opening phenomenon in response to the external stimuli of trans-/cis-C4H8. However, both the isomers exert the same gate-opening pressure of 2 kPa, which leads to a poor separation performance [35]. Zeolites with rigid frameworks have been used to separate trans-/cis-C4H8mixtures by sieving effect, however, their trans-C4H8uptake capacity is very low at 1.05 and 0.83 mmol?g-1on ITQ-32(ITQ stands for Instituto de Tecnología Química)[36] and deca-dodecasil 3 rhombohedral (DD3R)[37], respectively, owing to the limited space available for the gas uptake in the rigid pore structures within zeolites (Fig. 1(a)). Similar low trans-C4H8capacity was also observed for metal-gallates(Ni,Mg,and Co)owing to their robust nature[8].In general,robust zeolites exhibit Langmuir-type adsorption isotherms for trans-C4H8which is the cause for their limited loading capacity (Fig. 1(b)) in swing adsorption processes driven by pressure. This, in turn, presents major bottlenecks in their practical application owing to recyclability concerns. Furthermore, porous materials with large pore sizes (> 5.0 ?, 1 ? = 10-10m) usually exhibit high capacity but almost no separation selectivity for trans-/cis-C4H8, such as Y-fum-fcu-MOF (fum stands for fumarate; fcu stands for facecentered cubic) [10] and ZJNU-30 (ZJNU stands for Zhejiang Normal University).Thus,the discovery of a porous material with optimal pore dimensions, functionality, and energetics, that could discriminate or sieve particular trans-/cis-olefin isomers without sacrificing high gas uptake capacity, is a significantly profound challenge.

Anion-pillared ultramicroporous MOFs featuring electronegative inorganic and contracted pore surface [38–41] have unveiled outstanding separation performance for several important industrial gases such as C2H2/C2H4[28]and C3H6/C3H8[21].The variable combination of inorganic anions and metal ions enables the ultrafine-tuning of the pore apertures within the 0.1–0.5 ? scale[41–46].Herein,we report the results from the further exploration of this fluorinated ultramicroporous platform that allowed us to unveil ZU-36-Ni (GeFSIX-3-Ni, Ni(GeF6)(pyz)2, GeFSIX = GeF62-,3 = pyrazine = pyz), which displayed an unprecedented efficiency in trapping significant amounts of trans-C4H8while achieving effective exclusion of the cis-isomers (Fig. 1(c)). Importantly, ZU-36-Ni displayed an interesting step-wise adsorption isotherm that indicates an enhanced adsorption capacity and regeneration process with less energy input. Moreover, the adaptive pore channels for separating trans-C4H8, derived from the organic linker rotation for the guest molecule,conferred an increased sorption capacity to ZU-36-Ni (2.45 mmol?g-1) while the contracted pore window enhanced the cis-C4H8exclusion effect,leading to improved trans-/cis-C4H8separation selectivity (Fig. 1(d)).

2. Material and methods

2.1. Materials

Nickel(II) tetrafluoroborate hexahydrate (Ni(BF4)2?6H2O, 99%,J&K Scientific, China), ammonium hexafluorogermanate ((NH4)2-GeF6, 99.99%, J&K Scientific), ammonium hexafluorosilicate((NH4)2SiF6,99.99%,Sigma–Aldrich,USA),iron(II)tetrafluoroborate hexahydrate (Fe(BF4)2?6H2O, 97%, Sigma–Aldrich), and methanol(CH3OH, anhydrous, 99.8%, Sigma–Aldrich) were purchased and used without further purification.

Trans-2-butene (trans-C4H8, 99.9%), cis-2-butene (cis-C4H8,99.9%), and helium (He, 99.99%) were purchased from Hangzhou Jingong material Co., Ltd. (China). The mixture of 1,3-butadiene/trans-2-butene/1-butene/cis-2-butene/iso-butene/n-butane/isobutane (45/6.5/13/5.5/24/5/1, v/v) was purchased from Shanghai Weichuang Standard Gas Co., Ltd. (China).

2.2. Material syntheses

ZU-36-Ni (GeFSIX-3-Ni) was prepared using a literature report[41]. In a typical process, 1 mmol of Ni(BF4)2?6H2O (340 mg),1 mmol of(NH4)2GeF6(223 mg),and 1 g of pyrazine were dissolved in 2 mL of CH3OH and 2 mL of H2O, and stirred at ambient conditions for 2 d, which yielded a blue powder. The blue powder was then heated to 140 °C at 5 °C?min-1and was maintained for 24 h under vacuum to obtain the ZU-36-Ni material. SIFSIX-3-Ni was synthesized with the same method except that (NH4)2GeF6was substituted by (NH4)2SiF6. For ZU-36-Fe, the synthesis procedure is the same as that for GeFSIX-3-Ni,except that Ni(BF4)2?6H2-O was replaced by Fe(BF4)2?6H2O.

2.3. Characterization

Powder X-ray diffraction (PXRD) was conducted at room temperature on a Bruker D8 Advance diffractometer (Bruker AXS,Germany) using Cu-Kα radiation (λ = 1.5418 ?). PXRD data treatment and the structural determination were performed using the JANA2006.FullProf.98 program was applied for the Rietveld refinements. The background was refined with a polynomial function.The thermal stability of the obtained materials was investigated via thermalgravimetric analysis (TGA, TA Instruments SDT 600,USA) under N2atmosphere with a flow rate of 20 mL?min-1.

2.4. Gas adsorption

The sorption isotherms of C4hydrocarbons at low pressures up to 1 bar(1 bar=105Pa)were collected on a fully automated ASAP 2050 adsorption analyzer (Micromeritics Instruments, USA). The temperature was controlled with a water circulation bath.

2.5. Breakthrough test of C4 isomers

The fixed-bed breakthrough tests were conducted on a selfmade dynamic gas breakthrough equipment [30]. The test was conducted using a stainless-steel chromatographic column with an inner diameter of 4.6 mm and length of 50 mm. Samples of ZU-36-Ni, SIFSIX-3-Ni, and ZU-36-Fe were packed in three of the same columns which weighed 0.62, 0.64, and 0.67 g, respectively.The column packed with the sample powders was first activated with a flow of He (10 mL?min-1) at 100 °C for 12 h. After the activation, a cis-C4H8/trans-C4H8(50/50, v/v) mixture with a flow rate of 0.5 mL?min-1was introduced. After the breakthrough test, the fixed-bed was regenerated under He flow (5 mL?min-1) at 100 °C for 12 h. The actual separation performance of the as-synthesized material for C4mixtures including 1,3-butadiene, trans-2-butene,1-butene, cis-2-butene, iso-butene, n-butane, and iso-butane(1,3-C4H6/trans-C4H8/n-C4H8/cis-C4H8/iso-C4H8/n-C4H10/iso-C4H10,45/6.5/13/5.5/24/5/1,v/v)was further investigated with a flow rate of 0.75 mL?min-1. The real-time outlet gas eluted from the fixedbed was monitored using a gas chromatography (Micro GC-490,Agilent,USA).For studying the effect of humidity on the separation performance, the cis-C4H8/trans-C4H8(50/50, v/v) mixture with a flow rate of 1 mL?min-1was introduced into a water tank at 298 K, and the outflow gas was then flowed through a sorption column. The outlet gas from the column was monitored using a GC-2010 (Shimadzu, Japan) equipped with a flame ionization detector (FID) and a thermal conductivity detector (TCD).

A correction for the dead time was applied by He breakthrough experiments, and the He retention time (He is regarded as nonadsorbed) was applied as the dead time.

3. Results and discussion

3.1. Fine-tuned pore structure

Two ultramicroporous MOFs, ZU-36-Ni and ZU-36-Fe(Fe(GeF6)(pyz)2), were prepared by the reaction of ammonium hexafluorogermanate ((NH4)2GeF6), pyrazine, and Ni(BF4)2or Fe(BF4)2in a CH3OH and H2O mixture,followed by heating the isolated solid at 140 °C for 24 h in vacuo (Figs. 2(a) and (b)). The refined unit cell parameters of ZU-36-Ni were a = b = 6.984 ?,and c = 7.587 ? (also termed as the pore dimension of ZU-36-Ni,Table S2 in Appendix A), which is in accordance with the threedimensional scales of trans-C4H8(7.4 ? × 5.35 ? × 4.16 ?) and favors the preferential binding of trans-C4H8in the unit cells of ZU-36-Ni. In contrast, ZU-36-Fe showed a longer pore cell with c = 7.73 ?, resulting from the weak coordination affinity between Fe2+and the N atoms in the organic linker. Such a different pore dimension may lead to different sorption behaviors and host–guest interaction modes in limited pore space [27]. The introduction of GeF62-with increased Ge–F distance (1.83 ?) results in onedimensional (1D) contracted pore channels compared with SIFSIX-3-Ni (Si–F distance: 1.67 ?). The abundant electronegative F atoms protruding into the 1D pore channels can bind the guest molecule via strong H-bonding[42–45].The quasi-maximum pore sizes(upper limit of the pore size,Fig.S1 in Appendix A)of ZU-36-Ni and ZU-36-Fe(blue break lines in Fig.2(c))are 4.75 and 4.85 ?,respectively. Such ultra-micro pores could efficiently exclude cis-C4H8(4.94 ?, kinetic diameter), but allow the trapping of trans-C4H8(4.31 ?, kinetic diameter) (Fig. 2(d)). The purity of the assynthesized ZU-36-Ni and ZU-36-Fe was confirmed by comparing the PXRD patterns with the calculated patterns of ZU-36-Ni and ZU-36-Fe (Fig. S2 in Appendix A). The Brunauer–Emmett–Teller(BET) surface areas calculated by CO2adsorption isotherms at 273 K were 313 and 295 m2?g-1for ZU-36-Ni and ZU-36-Fe,respectively(Fig. S3 in Appendix A). Thermostability is a key metric that reflects certain aspects of the framework stability.The TGA results demonstrated that ZU-36-Ni is stable up to 340 °C (Fig. S4 in Appendix A), which is relatively superior to the other reported ultramicroporous MOFs such as NbOFFIVE-1-Ni (310 °C) [43] and SIFSIX-3-Ni (210 °C). The improved thermal stability of ZU-36-Ni compared with the analogous MOFs may be attributed to the short and strong bonds between Ni2+and the organic linkers, and the strong binding affinity of GeF62-with Ni2+,which leads to the contracted framework.Furthermore,the structure and adsorption performance of both the anion-pillared MOFs could be well retained after exposure to humid air, indicating their high tolerance to humid air (Figs. S2 and S3).

3.2. Adsorption performances

When used as sorbents for the separation of trans-/cis-C4H8,ZU-36-Fe exhibited a typical Langmuir-type adsorption isotherm for trans-C4H8with strong binding affinity and high uptake at low pressures. The trans-C4H8uptake amount on ZU-36-Fe is 1.81 mmol?g-1at 1 bar and 298 K. On the other hand, ZU-36-Ni(Fig. 3(a)) exhibited a stepped-adsorption isotherm for trans-C4H8. At the low-pressure range (< 0.01 bar), the less steep slope of the adsorption isotherm indicated that trans-C4H8interacts less strongly with ZU-36-Ni, which caused the low capture uptake of trans-C4H8at such low pressures. With the pressure increasing,the slope increased, indicating that ZU-36-Ni shows increased and homogeneous binding affinity for trans-C4H8. Finally, ZU-36-Ni showed a remarkable trans-C4H8capacity of 2.45 mmol?g-1(equals to one molecule per cell), which is significantly higher than that on ZU-36-Fe although the pore size is relatively smaller(Fig. 3(b)). Such reversal in adsorption behavior is attributed to the adaptivity of the pore structure of ZU-36-Ni, which allowed the enhanced accommodation of trans-C4H8molecules. A desorption pressure(Pdesor)of 0.01 bar was selected according to the purity and yield requirements of the product. The working capacity(Fig.S5 in Appendix A)of ZU-36-Ni,2.25 mmol?g-1,is much higher than that for ZU-36-Fe (0.77 mmol?g-1). Notably, ZU-36-Ni exhibited much higher uptake for trans-C4H8(2.45 mmol?g-1)than other reported size-sieving materials (Table S3 in Appendix A), such as ITQ-32 (1.1 mmol?g-1) [36] and DD3R [37] (0.832 mmol?g-1at 303 K).In contrast,both ZU-36-Ni and ZU-36-Fe showed relatively negligible adsorption of cis-C4H8because of the molecular exclusion effect. Owing to its relatively smaller aperture size, ZU-36-Ni (4.75 ? vs 4.85 ? for ZU-36-Fe) exhibited a lower cis-C4H8uptake (0.35 mmol?g-1) than ZU-36-Fe (0.5 mmol?g-1) and SIFSIX-3-Ni (0.8 mmol?g-1, Fig. S6 in Appendix A) at 1 bar and 298 K. Such a low cis-C4H8uptake and high trans-C4H8capacity endowed ZU-36-Ni with a benchmark trans-/cis-C4H8uptake ratio of 7, which is much higher than that of ZU-36-Fe (3.6) and the other previously reported materials such as Mg-gallate (3.2) [8],Y-fum-fcu-MOF (0.94) [10], and ZJNU-30 (1.13).

Fig. 2. Schematic illustration of (a) synthesis and (b) pore structure of ZU-36 material. (c) Quasi-maximum and empirical pore size are defined by paralleled F–F distance(blue break lines)and diagonal F–F distance(pink break lines),respectively,and the unit cell of ZU-36 viewed from a direction with c axis controlled.(d)Molecular structures and sizes of trans-C4H8 and cis-C4H8.

Fig.3. (a)Stepped sorption isotherms of trans-C4H8 on ZU-36-Ni compared with(b)typical Langmuir adsorption isotherms of trans-C4H8 on ZU-36-Fe(298 K).(c)Trans-/cis-C4H8 adsorption isotherms on other ultramicroporous materials at 298 K.(d)Ideal adsorbed solution theory(IAST)selectivities of various MOFs for trans-/cis-C4H8(50/50,v/v)mixture.

Other ultramicroporous MOFs were also investigated for comparison.Interpenetrated anion-pillared MOFs with larger pore size only exhibit moderate uptake ratios for trans-/cis-C4H8(Fig. 3(c),Fig. S7 in Appendix A, and Table S3). For example, ZU-32(GeFSIX-2-Cu-i) with a pore window size of 4.5 ? × 4.5 ? exhibits high trans-C4H8and cis-C4H8uptake capacity (3.55 and 2.85 mmol?g-1, respectively) at 1 bar and 298 K but a low uptake ratio of 1.37 (Fig. 3(c)), and moderate separation potential.SIFSIX-1-Cu and ZIF-8-Zn exhibit high but almost the same uptake for both trans- and cis-C4H8, indicating the negligible separation selectivity for trans-/cis-C4H8mixtures (Fig. 3(c)).

3.3. Separation selectivities

The feasible separation selectivity of anion-pillared ultramicroporous MOFs for trans-/cis-C4H8(50/50,v/v)mixture were qualitatively evaluated using calculations of the ideal adsorbed solution theory (IAST) (Fig. 3(d), Table S4 in Appendix A) [47]. ZU-36-Ni and ZU-36-Fe displayed separation selectivities of 191 and 170,respectively, which were much higher than that for ZU-32 (7.6),ZIF-8-Zn (1.2), and ZJNU-30 (1.5). Furthermore, the initial slope ratios (Figs. S8–S13 and Table S5 in Appendix A) also suggest that ZU-36-Ni (18.7) exhibits excellent separation performance compared with other materials, such as Ni-gallate (7.9) [8] and ZU-32(7), and can be a promising physical adsorbent for trans-/cis-C4H8separation.

3.4. Dispersion-corrected density functional theory (DFT-D)calculations

To better understand the origin of the guest-adaptivity, the binding sites of trans-C4H8were systematically investigated through DFT-D calculations (Figs. 4 and S14 in Appendix A). The initial ZU-36-Ni exhibited a primitive cubic (pcu) network with vicinal pyrazine rings in one cell perpendicular to each other and parallel with the inorganic pillars (Fig. 4(a)). When trans-C4H8was trapped into the pore channels, an obvious rotation of pyrazine was observed to adapt the trans-C4H8molecules (Figs. 4(b)and S14). Trans-C4H8preferentially resides at the middle of the cavity because of the suitable pore dimension and π–π interactions between its sp2carbons and the aromatic ring of pyrazine. After saturation, one trans-C4H8molecule is grasped by eight F atoms from the two planes with C–H???F H-bonding (distances: 2.50–2.59, 3.41, and 3.47 ?) accompanied with the pyrazine rotation by 9.5° (Fig. 4(b)), with a calculated binding energy (ΔE) of 49.6 kJ?mol-1. Such effective binding configuration of trans-C4H8in ZU-36-Ni results from the combination of suitable c-axis length(7.587 ?)and pore size of ZU-36-Ni,which affords full immobilization of one trans-C4H8in one cell. In summary, the guest-adaptive behavior of ZU-36-Ni is realized by the rotation of organic linkers to maximize the host–guest interactions with optimal conformation. Additionally, the transport of trans-C4H8from one cell to another in the 1D pore channels requires co-operative rotation of the pyrazines to accelerate this process owing to the limited pore space [20]. Such adaptive configuration transformation for guest molecules makes a great contribution to enhancing the recognition ability of trans-C4H8and increasing the uptake capacity.

The calculated binding sites of trans-C4H8in ZU-36-Fe were quite different (Fig. 4(c)). Trans-C4H8is bound only by the four F atoms from the same plane via strong H-bonding,which indicated the availability of a large space unoccupied by the guest molecules in one unit cell. This is consistent with the adsorption isotherm of trans-C4H8on ZU-36-Fe, and only 0.8 molecule of trans-C4H8trapped in each unit cell of ZU-36-Fe, thus leading to a reduced uptake amount of trans-C4H8at saturation. Such a different optimized binding configuration of trans-C4H8in ZU-36-Fe, compared with that in ZU-36-Ni, is due to the fact that the longer c-axis(7.73 ?) in ZU-36-Fe could not fully match the scale or dimension of trans-C4H8. The calculated ΔE of trans-C4H8on ZU-36-Fe was 60.5 kJ?mol-1, which is much higher than that of ZU-36-Ni(49.6 kJ?mol-1), implying the stronger host–guest interactions between trans-C4H8with ZU-36-Fe at low trans-C4H8loading. The lower ΔE on ZU-36-Ni can be ascribed to the compensation by the deformation of the framework (11.0 kJ?mol-1) to adapt the guest molecule. Simultaneously, coverage-dependent adsorption enthalpy (Qst) calculated based on Clausius–Clapeyron equation using the isotherms at different temperatures (Figs. S15 and S16 in Appendix A) shows that the Qstfor trans-C4H8at zero loading on ZU-36-Ni is 42.0 kJ?mol-1(Fig.S17 in Appendix A),which is also lower than that on ZU-36-Fe (61.8 kJ?mol-1), signifying that much milder regeneration conditions are required for ZU-36-Ni compared with those for ZU-36-Fe. To confirm the easier regeneration of ZU-36-Ni,cyclic adsorption tests were conducted with the materials regenerated using the room temperature and vacuum condition (Fig. S18 in Appendix A). Indeed, the results confirmed that ZU-36-Ni can be more easily regenerated with the trans-C4H8uptake well retained, whereas the trans-C4H8uptake on ZU-36-Fe slightly declined under the same conditions, which may be attributed to the insufficient regeneration of ZU-36-Fe resulting from the strong binding affinity for trans-C4H8.

3.5. Breakthrough experiments

Fig. 4. (a) Initial framework of ZU-36. Binding configurations of trans-C4H8 in (b) ZU-36-Ni and (c) ZU-36-Fe, respectively, obtained by DFT-D calculations. Color code: H,gray-25%; C, gray; N, blue; Ni, turquoise; Ge, light blue; F, peak green; Fe, lime. Bond length unit: ?.

Fig. 5. (a) Breakthrough experiments for trans-/cis-C4H8 (50/50, v/v) mixture separation on ZU-36-Ni and ZU-36-Fe (with dead volume excluded; CA/C0: the relative concentration in outlet stream compared with that in feed gas). (b) Cycling breakthrough experiments for trans-/cis-C4H8 (50/50, v/v) separation on ZU-36-Ni.

The actual separation performances of the trans-/cis-C4H8(50/50, v/v) mixture on ZU-36-Ni and ZU-36-Fe were evaluated using experimental fixed-bed breakthrough tests at 1 bar and 298 K (Fig. 5(a)). Both materials exhibit excellent trans-/cis-C4H8separation performances. Cis-C4H8elutes out of the column of ZU-36-Ni or ZU-36-Fe almost simultaneously with high purity(>99.99%),indicating the excellent sieving effect of both materials for cis-C4H8.Trans-C4H8could be trapped in the ZU-36-Ni fixed bed for about 58 min (93.5 min?g-1) with the corresponding capture amount of 1.15 mmol?g-1, which is better than that of ZU-36-Fe(37 min, 55.2 min?g-1) with a capture amount of 0.72 mmol?g-1.Additionally,a sharp molecular cut-off behavior for the separation of trans-/cis-C4H8mixture was not observed when using SIFSIX-3-Ni(Fig.S19 in Appendix A),which is consistent with the isotherms of trans-/cis-C4H8on the material (Fig. S6). More importantly, for ZU-36-Ni, there was no noticeable loss in trans-C4H8adsorption and separation capacity even after 10 cycles of breakthrough experiments (Fig. 5(b)), illustrating the excellent structural and cycling stability of ZU-36-Ni for trans-/cis-C4H8mixtures separation. Furthermore, the separation performance is unimpeded by humidity(Fig.S20 in Appendix A)showcasing the strong potential of ZU-36-Ni for trans-/cis-C4H8mixture separation for industrial applications. Last but not least, ZU-36-Ni also exhibited good separation performance for the C4mixture (1,3-C4H6/trans-C4H8/n-C4H8/cis-C4H8/iso-C4H8/n-C4H10/iso-C4H10, 45/6.5/13/5.5/24/5/1, v/v, Fig. S21 in Appendix A) indicating that ZU-36-Ni is a promising material for C4hydrocarbon separation.

4. Conclusions

In summary, two anion-pillared ultramicroporous MOFs, ZU-36-Ni (GeFSIX-3-Ni) and ZU-36-Fe (GeFSIX-3-Fe) are reported for the first time and used for highly efficient trans-/cis-C4H8splitting.ZU-36-Ni with its guest-adaptive pore channels coming from the rotation of organic linkers, exhibited an interesting step-wise adsorption isotherm for trans-C4H8. This attribute confers ZU-36-Ni with an increased capacity (2.45 mmol?g-1) compared to ZU-36-Fe (1.81 mmol?g-1) that does not possess adaptive pore channels. In addition, ZU-36-Ni adsorbed less cis-C4H8than ZU-36-Fe,as ZU-36-Ni with the contracted pore window size excluded cis-C4H8with a higher efficiency.The excellent trans-/cis-C4H8separation selectivity(191)and high-purity cis-C4H8(99.99%)observed in the breakthrough tests present ZU-36-Ni as an ideal adsorbent for trans-/cis-C4H8separation. This work provides new insights into the structural property–adsorption relationships necessary for anticipating the discovery of smart and efficient porous materials for the separation of hydrocarbon isomers of different dimensions and shapes.

Acknowledgments

This work was supported by the Zhejiang Provincial Natural Science Foundation of China (LZ18B060001), and the National Natural Science Foundation of China (21725603, 21476192, and U1862110).

Compliance with ethics guidelines

Zhaoqiang Zhang, Xili Cui, Xiaoming Jiang, Qi Ding, Jiyu Cui,Yuanbin Zhang, Youssef Belmabkhout, Karim Adil, Mohamed Eddaoudi, and Huabin Xing declare that they have no conflict of interest or financial conflicts to disclose.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.eng.2021.10.013.