Introduction

Ethylene (C2H4), the largest feedstock in petrochemical industries with a global production capacity of more than 170 million tons in 2018, exceeds any other chemical feedstock molecules for its wide application in polyethylene manufacture1. The industrial separation of C2H4 from ethane (C2H6) is typically through cryogenic distillation at high pressure and very low temperature, using very high towers, because they have similar sizes and volatilities2. Such a process is highly energy consuming, and thus exploring of other effective C2H4 separation methods at ambient conditions is highly demanded. Selective adsorption by porous materials for producing high-purity C2H4 has been determined to be one of the most desired methods for its low energy consumption3,4. Of particular recent interest in this regard is using C2H6-selective porous materials enriched with nonpolar/inert surfaces (e.g., introducing aromatic or aliphatic entities), because they prefer to capture more polarizable C2H6, and thus can directly afford high-purity C2H4 in a single adsorption step, avoiding an additional high-energy-consuming desorption step5,6. So far, most of the developed C2H6-selective adsorbents are metal-containing materials6,7,8,9,10. However, the polar metal centers in these materials can interact with unsaturated C2H4 molecules via strong π-complexation interactions, and thus weaken their C2H6/C2H4 separation abilities11. Therefore, developing other C2H6-selective adsorbents such as metal-free organic porous materials is highly required12,13, because they cannot only avoid the abovementioned problem, but also are easier to construct inert surfaces. However, the research of such materials is still in its initial stage14,15.

Porous organic cages (POCs), as a type of porous materials, are intrinsically porous given their hollow cavities16,17,18,19,20,21. The discrete inherent nature of POCs makes them possessing distinct benefits in solution processing, regeneration as well as post-synthesis modification21,22,23,24. Following the first elegant research reported in 2009 by Cooper et al.25, the number of POCs with different shapes, sizes and properties has increased substantially26,27,28,29,30,31,32,33,34,35,36, but robust POCs with high surface areas are still few, which highly hinder their practical use in gas storage as well as separation. Over the past decade, although Brunauer–Emmett–Teller (BET) surface areas of POCs have increased from the initial 624 to 3758 m2 g–1, most of POCs have BET values <1000 m2/g37,38. As for binary gas mixture separation, Cooper et al. first reported that imine-linked tetrahedral POC (CC3) exhibits selectivity of 20.4 for noble gas Xe/Kr separations at low concentrations in 201439. Two years later, they showed that CC3 could also separate sulfur hexafluoride (SF6), a much more potent greenhouse gas than carbon dioxide, from nitrogen (N2) with the highest SF6/N2 selectivity reported for any material under ambient conditions40. The selectivity for the abovementioned two cases arises from the precise size matching between Xe and SF6 with organic cage cavity than Kr and N2, respectively. In 2019, their team utilized post modification synthesis method to alter internal cavities of CC3 to produce hybrid cocrystal material (6ET-RCC3-R/CC3-S), which are excellent quantum sieves for hydrogen isotope separation with excellent deuterium/hydrogen selectivity (8.0)41. In the same year, Zhang’s and Zaworotko’s groups presented a soft imide-based POC (NKPOC-1) with gate opening behavior and could efficiently separate binary or ternary C3 hydrocarbon mixtures42. Up to now, POC adsorbents have only been limited to separate the above gas components by mixed gas experiments, and thus there is still much room to explore POC materials for real gas separation applications, especially for industrially important gases.

Calix[4]resorcinarenes, a subset of calixarenes, were derived from acid-catalyzed condensation between resorcinol and various kinds of aldehydes43,44,45. They possess electron-rich π cavities and eight upper-rim phenolic groups, and are effective hosts for inclusion of various guests ranging from small gases to large organic molecules45,46,47,48,49. Notably, their upper rims can be easily functionalized, making them good molecular building blocks for constructing self-assembled cages as well as porous polymers50,51,52,53,54,55,56. Very recently, our group employed predesigned concave-shaped tetraformylresorcin[4]arene (RC4ACHO) as secondary building blocks and different diamines as linkers to systematically design and synthesize several porous POCs with structural diversity from [2 + 4] dimeric lanterns, [3 + 6] trimeric triangular prisms, to [6 + 12] hexameric octahedra57. However, utilizing calix[4]resorcinarene-based POCs as solid adsorbents for gas separation applications remains unexploited. Herein, we present that the robust and highly porous [6 + 12] octahedral calix[4]resorcinarene-based POC (CPOC-301) is an excellent C2H6-selective material, and can be used as a robust absorbent to directly afford high-purity C2H4 from C2H6/C2H4 mixture.

Results

Crystal structure and characterization of CPOC-301

As shown in Fig. 1a, CPOC-301 is successfully prepared via self-assembly of RC4ACHO (1 equiv.) and p-phenylenediamine (2 equiv.) under mild condition. Single-crystal X-ray diffraction reveals that CPOC-301 has a truncated octahedron structure, with eight trigonal ports having edge length reaching about 12 Å, and a large cavity with inner diameter as well as volume that respectively reach 16.8 Å and 4270 Å3. The solid-state packing of CPOC-301 suggests that it possesses a one-dimensional channel, with a diameter of ~7 Å, viewed from [001] direction (Fig. 1b). Notably, quantitative tautomerization of imines in CPOC-301 to their keto-enamine forms has been observed from 1H NMR and FT-IR spectroscopy (Supplementary Figs. 1–3). Such a transformation makes CPOC-301 more stable even when it is stored in air for over half a year, as has been confirmed by FT-IR spectra as well as 1H NMR studies (Supplementary Figs. 3 and 4). Thermal gravimetric analysis (TGA) suggests that CPOC-301 exhibits thermal stability up to 300 °C under nitrogen atmosphere (Supplementary Fig. 5). PXRD reveals that desolvated CPOC-301 retains its crystal packing (Supplementary Fig. 6), which is very uncommon in large POC systems.

Fig. 1: Schematic illustration for assembly of CPOC-301.
figure 1

a The X-ray crystal structure of CPOC-301. b The solid-state molecular packing of CPOC-301 viewed from [001] direction, where H atoms are omitted for clarity. color codes: phenyl ring; orange, carbon; gray, oxygen; red, nitrogen; blue, and hydrogen; light turquoise.

Full size image

Gas adsorption and separation performances

The channel-type structure of CPOC-301 can potentially lead to interesting solid-state gas sorption properties. The permanent porosity of the activated CPOC-301 was confirmed by N2 gas sorption experiments at 77 K (Fig. 2). Its N2 adsorption isotherm shows a typical type I curve with a small fraction of a type IV adsorption behavior. The maximum N2 adsorption is 670 cm3 g−1, and the calculated BET of CPOC-301 is up to 1962 m2 g−1. The pore volume as well as the micropore volume of CPOC-301 respectively reach 1.03 and 0.46 cm3 g−1. The pore size distribution (PSD) profile exhibits a relatively sharp distribution of micropores from 1.59 to 1.77 nm (Fig. 2 inset), consistent with the cavity size of CPOC-301 in the crystal structure.

Fig. 2: Porosity measurements.
figure 2

N2 gas sorption isotherm at 77 K for CPOC-301, inset: the calculated PSD of CPOC-301.

Full size image

The high surface area and high density of nonpolar calix[4]resorcinarene cavities in CPOC-301, which may favor the preferential adsorption toward more polarizable C2H6 over C2H4, prompted us to investigate the sorption of these gases. Single-component adsorption isotherms of CPOC-301 for C2H6 and C2H4 were measured at 273, 283, and 293 K at 1 bar. Notably, CPOC-301 exhibits preferential adsorption of C2H6 (87 cm3 g−1) over C2H4 (75 cm3 g−1) at 293 K (Fig. 3a), and also the other two temperatures (Supplementary Figs. 7 and 8). The corresponding isosteric heat of adsorption (Qst) at zero coverage for C2H6 and C2H4 was calculated to be 32.4 and 24.2 kJ mol−1, respectively (Fig. 3b and Supplementary Figs. 9 and 10). Such a difference suggests that the host-guest interactions between CPOC-301 and C2H6 are much stronger than that of C2H4. Notably, the C2H6 uptake capacity of CPOC-301 at 293 K and 1 bar exceeds most of the reported porous organic materials14 and C2H6-selective MOF materials5.

Fig. 3: C2H6/C2H4 separation performances.
figure 3

a Experimental C2H6 and C2H4 adsorption isotherms of CPOC-301 at 293 K. b Isosteric heat of adsorption plots for the adsorption of C2H6 and C2H4 by CPOC-301. c Experimental breakthrough curves for equimolar mixture of C2H6/C2H4 at 298 K and 1 bar over a packed bed of CPOC-301. d The recyclability of CPOC-301 under multiple mixed gas column breakthrough tests.

Full size image

Motivated by the high uptake capacity and C2H6-selective behavior of CPOC-301, the ideal adsorbed solution theory (IAST) is used to assess its separation selectivity for C2H6/C2H4 (50:50) (Supplementary Figs. 11–13). IAST calculation results demonstrate that C2H6/C2H4 selectivity range is from 1.3 to 1.4 at 293 K. Compared to other C2H6-selective porous organic molecular materials, the selectivity value of C2H6 over C2H4 for CPOC-301 is comparable to [4 + 6] boronic ester cage (1.29) reported by Mastalerz et al.58, but lower than the recently reported hydrogen-bonded organic frameworks including HOF-76 (2.0)14 and ZJU-HOF-1 (2.25) reported by Li and Chen (Supplementary Fig. 14)15. To evaluate the actual separation performance of CPOC-301, the experimental breakthrough studies were conducted in a packed column of the activated CPOC-301 sample under an equimolar C2H6/C2H4 mixture at ambient conditions. The breakthrough curves depicted in Fig. 3c prove that CPOC-301 can efficiently realize the complete separation of C2H4 from C2H6/C2H4 mixtures. Notably, C2H4 gas breaks through the adsorption bed first to produce an outflow of pure gas containing no detectable C2H6. Conversely, C2H6 gas breaks through column following a substantial time-lapse, because the C2H6 molecule is more preferentially adsorbed in CPOC-301 than C2H4. From dynamic breakthrough experiment, the calculated separation factor for an equimolar mixture of C2H6/C2H4 was 1.3, consistent with the predicted IAST result. For practical industrial applications, the ideal adsorbent should also have good recycling performance. We performed multiple C2H6/C2H4 mixed-gas dynamic breakthrough experiments under similar operating conditions. The separation performance of C2H6/C2H4 does not obviously change within seven continuous cycles (Fig. 3d). Besides, NMR and PXRD data after breakthrough experiments in addition to N2 gas adsorption and PXRD of the sample after being exposed to air and soaked in water (Supplementary Figs. 15–19), indicate that CPOC-301 is robust enough to be a promising candidate for C2H4 purification.

Separation mechanism

To further understand the role of CPOC-301 in the mechanism of selective C2H6/C2H4 adsorption, modeling studies based on an efficient conformer search algorithm (CREST)59,60 and first-principles dispersion-corrected density functional theory (DFT-D) calculations were performed14,15. The primary binding sites for C2H6 and C2H4 molecules were found to be located at the calix[4]resorcinarene cavities. The lowest-energy gas binding configurations are displayed in Fig. 4a, b. For more clarity, only one adsorbed gas molecule at calix[4]resorcinarene cavity site is shown, because the remaining five sites within CPOC-301 are crystallographically identical. Notably, we believed that C2 hydrocarbon guests occupy the six primary binding sites of CPOC-301, because one CPOC-301 molecule was found to adsorb about 22.5 C2H6 and 19.5 C2H4 molecules as calculated from adsorption isotherms at 293 K. The corresponding calculated static binding energies (ΔE, ΔE = EPOC + gas − EPOC − Egas) of C2H6 and C2H4 are around −42.7 and −41.2 kJ mol−1, respectively. Such binding energy variation suggests the stronger host-guest interactions between C2H6 and calix[4]resorcinarene cavity than that of C2H4, consistent with our aforementioned experimental observation. The higher binding energy of C2H6 is mainly ascribed to nonplanar C2H6 molecule sterically “matches” better to the electron-rich cavity of calix[4]resorcinarene than planar C2H4 molecule14,15, because C–H···π bonds of guest C2H6 is more numerous than that of C2H4 (12 for C2H6 and 7 for C2H4; and speciffic H···π distances are listed in Supplementary Fig. 20). Moreover, the Hirshfeld surface was performed to further reveal the intermolecular contacts present within the corresponding simulated crystal structures61,62. Notably, the strong intermolecular interactions between C2 hydrocarbons and calix[4]resorcinarene cavity are indicated as bright red spots on the Hirshfeld surface (Supplementary Fig. 21), which are close to the abovementioned interaction regions of C–H···π bonds between the C2 guests and calix[4]resorcinarene host. Moreover, C2H6 and C2H4 gas sorption as well as C2H6/C2H4 mixture breakthrough experiments by other types of calix[4]resorcinarene-based POCs comprising one hexameric octahedron with functional methyl groups (CPOC-301-Me) and a trimeric triangular prism (CPOC-201) were measured to confirm the aforementioned host-guest interaction results57. Both of them show the same behavior in C2H6 and C2H4 gas sorption and separation compared to CPOC-301 (Supplementary Figs. 22–27). To further authenticate our result, host-guest interactions between CPOC-301 and C2H2 (acetylene), another guest molecule of the C2 hydrocarbons family, has also been simulated (Fig. 4c). The calculated ΔE value of C2H2 is −36.7 kJ mol−1, indicating that its host-guest interaction is weakest in the C2 hydrocarbons. This is consistent with the experimental Qst result with value of 20.2 kJ/mol for C2H2 (Supplementary Figs. 28 and 29), and has been confirmed by C2H6/C2H2 IAST calculation (Supplementary Fig. 30), C2H4/C2H2 and C2H6/C2H4/C2H2 breakthrough experiments (Fig. 5 and Supplementary Fig. 31).

Fig. 4: Mechanism study.
figure 4

A comparison of preferential a C2H6, b C2H4, and c C2H2 adsorption sites and close C–H···π interactions within the calix[4]resorcinarene cavities observed by DFT-D calculations. Carbon is gray, oxygen red, and hydrogen yellow. Dashed bonds highlight C–H···π interactions.

Full size image
Fig. 5: C2H4/C2H2 and C2H6/C2H4/C2H2 separation performances.
figure 5

a Isosteric heat of adsorption plots for the adsorption of C2H2 by CPOC-301. b The experimental column breakthrough curve of CPOC-301 with an equimolar C2H2/C2H4 mixture. c The experimental column breakthrough curve of CPOC-301 with C2H6/C2H4/C2H2 mixture.

Full size image

Discussion

We report a pioneering work using shape-persistent amine-linked CPOC-301 as POC adsorbent for C2H4 purification. Specifically, the highly porous CPOC-301 can efficiently trap C2H6 from C2H6/C2H4 mixture and therefore directly produce high-purity C2H4. The preferential interactions with C2H6 over C2H4 in CPOC-301 are because C2H6 form more multiple C–H···π hydrogen bonds with resorcin[4]arene cavities than C2H4 guests as indicated by results of DFT-D calculations. This finding could shed some light on the design and synthesis of POCs based on supramolecular cavitands as “porous additives” in column and membrane separation applications for industrially important gases in the future. Efforts to explore these possibilities are ongoing.

Method

Characterization

Under ambient temperature, proton nuclear magnetic resonance (1H NMR) data were collected by means of a Burker AVANCE 400 (400 MHz) spectrometer. Fourier-transformed infrared spectroscopy (FT-IR) spectra were taken on a Magna 750 FT-IR spectrometer utilizing KBr pellets in the 400–4000 cm−1 region. High-resolution electrospray ionization mass spectrometry (ESI-TOF-MS) was collected using a Bruker MaXis™ 4G instrument. The TGA was collected at 10 °C/min ramp rate in dynamic N2 flow within a temperature range of 30–900 °C by a NETZSCH STA 449C thermal analyzer. The pattern of powder X-ray diffraction (PXRD) was collected on a Rigaku Mini 600 X-ray diffractometer for CuKα radiation (λ = 0.154 Å), with 0.5°/min scan speed of and 0.02° in 2θ step size.

Synthetic procedure

All reagents and solvents were supplied by Sinopharm Chemical Reagent Co., Ltd with analytical grade, and utilized without further purification. CPOC-301 was synthesized as follows57: 162 mg (0.20 mmol) RC4ACHO63 and 43 mg (0.4 mmol) p-phenylenediamine were added into 5 mL nitrobenzene and 15 mL CHCl3. After sealing the mixture in a 48 mL pressure vial, heating to 65 °C with stirring for 2 days, and cooling down naturally. Red block single crystals of CPOC-301 with ~78% yield were obtained by slow methanol vapor diffusion into the above mixture. 1H NMR (400 MHz, CDCl3, 298 K): δ 1.04 (d, 144H), 1.60 (m, 24H), 2.10 (t, 48H), 4.65 (t, 24H), 7.37 (s, 24H), 7.40 (s, 48H), 9.18 (s, 24H), 10.38 (s, 24H), 16.21 (s, 24H). p.p.m. ESI-TOF-MS calculated for CPOC-301, C360H384N24O48 [M-2H]2− 2905.4178, found 2905.3963.

Gas adsorption measurements

Automatic volumetric adsorption equipment (Micromeritics, ASAP 2020) was utilized to conduct all gas adsorption-desorption measurements of CPOC-301. The data of PSD were obtained from the N2 sorption isotherm at 77 K based on the DFT model in the Micromeritics ASAP 2020 software package (assuming cylinder pore geometry). Before measurements, samples’ degassing was accomplished at 100 °C for 10 h under dynamic vacuum (below 10 μmHg) for 10 h for removing the adsorbed impurities. The calculated pore volume and micropore volume are based on built-in software of ASAP 2020 physisorption analyzer. The isosteric heat of sorption for C2 hydrocarbons was regarded as a function of gas uptake using compared the adsorption isotherms at 273, 283, and 293 K. After data being modeled with a virial-type expression comprising ai and bi parameters (Eq. (1)), the heat of adsorption (Qst) is determined through fitting parameters by means of Eq. (2), in which P refers to the pressure, N refers to the adsorbed amount, T refers to the temperature, R refers to universal gas constant, while m and n determine the number of terms required to describe the isotherm adequately. The parameters were obtained from fitting of C2 hydrocarbons adsorption isotherms fitted with R2 > 0.999. To assess C2H6/C2H4 separation performance, IAST of Myers and Prausnitz64 and pure component isotherm fits by dual-site Langmuir–Freundlich equation were employed for calculating molar loadings in the mixture for specific partial pressures of bulk gas phase (Eq. (3)), where N refers to molar loading of species (mmol g−1), A refers to saturation capacity of species (mmol g−1), B refers to Langmuir constant (kPa−c), C refers to Freundlich constant and P refers to bulk gas phase pressure of species (kPa). The adsorption selectivity based on IAST for C2H6/C2H4 mixture is identified using Eq. (4):

$${\rm{ln}}p={{\rm{ln}}}N+\frac{1}{T}\mathop{\sum }\limits_{i=0}^{m}{a}_{i}{N}^{i}+\mathop{\sum }\limits_{i=0}^{n}{b}_{i}{N}^{i}$$
(1)
$${Q}_{{{\rm{st}}}}=-R\mathop{\sum }\limits_{i=0}^{m}{a}_{i}{N}^{i}$$
(2)
$$N={A}_{1}\tfrac{{B}_{1}\times {P}^{{C}_{1}}}{1+{B}_{1}\times {P}^{{C}_{1}}}+{A}_{2}\tfrac{{B}_{2}\times {P}^{{C}_{2}}}{1+{B}_{2}\times {P}^{{C}_{2}}}$$
(3)
$${S}_{A/B}=\frac{{x}_{A}{y}_{B}}{{x}_{B}{y}_{A}}$$
(4)

Column breakthrough experiments

The mixed-gas breakthrough separation experiment was performed through a home-built setup equipped with a mass spectrometer (Pfeiffer GSD320). For instance, in a typically conducted breakthrough experiment for C2H6/C2H4/He (10:10:80, v/v/v), C2H4/C2H2/He (10:10:80, v/v/v), and C2H6/C2H4/C2H2/He (9:9:2:80, v/v/v/v) gas mixtures, CPOC-301 powder was subjected to packing into a custom-made stainless-steel column (3.0 mm I.D. × 120 mm) having void space filled with silica wool. Activating sample was accomplished through heating the packed column at 100 °C for 12 h under a constant He flow (10 mL min−1 at 298 K and 1 bar). The He flow was then turned off and the C2 hydrocarbon gas mixture was permitted to flow into the column (2 mL min−1). The mass spectrometer was employed to continuously monitor the outlet effluent from column. After the breakthrough experiment, the sample was regenerated in situ in the column at 100 °C for 12 h. The complete breakthrough of C2H6 was determined using downstream gas composition reaching that of feed gas. On the basis of the mass balance, the gas adsorption capacities can be attained from the following equation65:

$${q}_{i}=\frac{{C}_{i}V}{22.4\times m}\times {\int }_{0}^{t}\left(1-\frac{F}{{F}_{0}}\right){dt}$$

where qi refers to the equilibrium adsorption capacity of gas i (mmol g−1), Ci represents the feed gas concentration, V refers to the volumetric feed flow rate (cm3 min−1), t represents the adsorption time (min), F0 and F, respectively, refer to the inlet and outlet gas molar flow rates, and m represents adsorbent mass of (g). The separation factor (α) of breakthrough experiment can be calculated as follows:

$${\rm{\alpha }}=\frac{{q}_{A}{y}_{B}}{{q}_{B}{y}_{A}}$$

In which yi is molar fraction of gas i (i = A, B) in gas mixture.

Binding energy calculations

The molecular input of CPOC-301 was generated starting from its crystal structure, in which one of two disordered motifs was selected and the isobutyl group was reduced to a methyl group to simplify the simulation. The initial structure of CPOC-301 was then optimized by the semiempirical extended tight-binding (xtb) program package developed by Grimme group66. The initial binding sites for C2H6, C2H4, and C2H2 were determined using noncovalent interaction (NCI)/iMTD algorithm in CREST according the procedure in the literature59,60. During binding site screening, CPOC-301 was free of any constraints, allowing structural relaxation as well as adaption to the C2 hydrocarbon guest. According to CREST calculation, a structure ensemble of NCI complexes within a 6 kcal mol−1 energy window is obtained, and the energetically lowest conformation was selected as the first binding site.

The free CPOC-301, C2 hydrocarbon guests, and C2@CPOC-301 were further optimized by DFT-D method by means of Dmol3 module as implemented in Accelrys Materials Studio package67. The PBE-type exchange-correlation functional with a generalized gradient approximation, the double numerical plus polarization basis sets that include a d-type polarization function on all non-hydrogen atoms and a p-type polarization function on all hydrogen atoms and Grimme method for DFT-D correction were employed for all calculations. Besides, FINE quality mesh size was employed in the calculations. The energy, force, and displacement convergence criteria were respectively set as 1 × 10−5 Ha, 2 × 10−3 Ha, and 5 × 10−3 Å. The binding energies (ΔE bind in kJ mol−1) were calculated as the differences in total energies E between fully optimized C2@CPOC-301 and the CPOC-301 and C2 hydrocarbon guests in terms of the following equation:

$${{\Delta}} E={E}_{{\rm{POC}}+{\rm{gas}}}-{E}_{{\rm{POC}}}-{E}_{{\rm{gas}}}$$

where EPOC + gas stands for the energy of the fully optimized C2@CPOC-301 structure, while EPOC and Egas respectively represent energies of bare CPOC-301 structure and isolated C2 hydrocarbon molecule, respectively. Based on such an equation, more negative binding energy means more favorable binding. The detailed calculated results for energy of the C2@CPOC-301 are summarized in Supplementary Table 1.