Efficient purification of ethene by an ethane-trapping metal-organic framework

Abstract

Separating ethene (C₂H₄) from ethane (C₂H₆) is of paramount importance and difficulty. Here we show that C₂H₄ can be efficiently purified by trapping the inert C₂H₆ in a judiciously designed metal-organic framework. Under ambient conditions, passing a typical cracked gas mixture (15:1 C₂H₄/C₂H₆) through 1 litre of this C₂H₆ selective adsorbent directly produces 56 litres of C₂H₄ with 99.95%+ purity (required by the C₂H₄ polymerization reactor) at the outlet, with a single breakthrough operation, while other C₂H₆ selective materials can only produce ca. ⩽ litre, and conventional C₂H₄ selective adsorbents require at least four adsorption–desorption cycles to achieve the same C₂H₄ purity. Single-crystal X-ray diffraction and computational simulation studies showed that the exceptional C₂H₆ selectivity arises from the proper positioning of multiple electronegative and electropositive functional groups on the ultramicroporous pore surface, which form multiple C–H···N hydrogen bonds with C₂H₆ instead of the more polar competitor C₂H₄.

The separation of high purity ethene from the mixed gaseous products of cracking poses significant obstacles. Here, the authors present a metal-organic framework which, in contrast to most absorbents, selectively binds the less polar ethane thus allowing the efficient collection of the target product.

As the most important chemical product, ethene (C₂H₄) is generally obtained through steam cracking and thermal decomposition of naphtha or ethane (C₂H₆) (ref. ¹). Besides being obtained as a byproduct of petroleum refining, C₂H₆ is also isolated on an industrial scale from natural gas (CH₄ 70∼96%, C₂H₆ 1∼14% and CO₂ 0∼8%) (ref. ²). As a result of their similar physical properties, it is difficult to separate C₂H₆, C₂H₄ and CO₂ (refs ^{3, 4, 5}). In industry, C₂H₆ and C₂H₄ are separated by cryogenic high-pressure distillation, typically at 7–28 bar and 183–258 K using very high towers consisting of over 150 trays, which is very energy consuming (7 GJ t⁻¹) and constitutes a notable portion of the ethylene cost⁶,⁷. To save energy, separation methods effective at the ambient temperature and pressure are highly demanded⁸,⁹,¹⁰,¹¹. Passing the gas mixture through a fixed-bed adsorber can be a very simple and promising approach to afford low energy consumption and high product purity.

Because unsaturated hydrocarbons like to coordinate with metal ions, C₂H₄ can be selectively bound and separated from its saturated counterpart C₂H₆ (refs ^{12, 13, 14, 15, 16}). Compared with other types of porous materials, porous metal-organic frameworks (MOFs) are unique for their diversified/designable framework structures and pore surfaces, including the ease of introducing open metal sites (OMSs), which have shown great potentials for C₂H₄/C₂H₆ separation¹⁷,¹⁸,¹⁹,²⁰,²¹,²². In the fix-bed separation process, the un-adsorbed C₂H₆ first breakthrough, and C₂H₄ enriched in the stationary phase is later obtained by heating and/or inert-gas blowing. Because the un-adsorbed C₂H₆ residing in the mobile phase contaminates the desired product C₂H₄ during the desorption stage, the highest C₂H₄ purity produced by a full adsorption–desorption cycle can just reach 99%+ (refs ^{13, 17, 23, 24}), and at least four such cycles are necessary to achieve 99.95%+ (ref. ²⁵), the lower limit required by the C₂H₄ polymerization reactor²⁶,²⁷,²⁸. Obviously, this problem can be solved by using a C₂H₆ selective adsorbent, which not only improves the C₂H₄ purity but also reduces energy consumption. The simple separation operation and device (just one adsorption process in a single breakthrough operation) are also necessary for onsite supply of purified C₂H₄. However, such an unusual adsorption behaviour has been only reported for a few low-polarity or hydrophobic MOFs²⁹,³⁰,³¹,³²,³³,³⁴,³⁵,³⁶, and their C₂H₄/C₂H₆ separation performances (that is, C₂H₆/C₂H₄ selectivities) are poor, because the polarities of C₂H₄ and C₂H₆ are very similar and can be hardly distinguished by hydrophobic adsorbents.

As C₂H₆ possesses the lowest polarity (quadrupole moment) compared with similar molecules such as C₂H₄ and CO₂ (Supplementary Table 1)³⁷; polar adsorbents are generally selective for the latter gases. However, considering that the electropositive and electronegative portions locate quite differently among these gas molecules, we speculated that by rational utilization of polar functional groups, it is still possible to design a MOF with optimized pore size/shape and surface electrostatic distribution that can bind C₂H₆ much stronger than for C₂H₄. Herein, we report the design, structure and gas adsorption/separation properties of such a C₂H₆-trapping MOF, which is useful for not only direct producing highly pure C₂H₄ from C₂H₄/C₂H₆ mixtures, but also efficient separation of four-component CH₄/C₂H₄/C₂H₆/CO₂ mixtures and extraction of C₂H₆ from natural gas.

Results

Synthesis, structure and stability

Bis(5-amino-1H-1,2,4-triazol-3-yl)methane (H₂batz) with two 3-amino-1,2,4-triazole rings bridged by a methylene group was designed as a new ligand combining multiple nitrogen atoms as hydrogen-bonding acceptors and methylene groups as dipole repulsion groups, as well as short bridging lengths for construction of an ultramicroporous framework. Reaction of H₂batz and Zn(OH)₂ in dilute aqueous ammonia produced a porous metal-azolate framework [Zn(batz)]·0.5H₂O (MAF-49·H₂O). Single-crystal X-ray diffraction (SCXRD) analysis of MAF-49·H₂O (Supplementary Table 2 and Supplementary Data 1) showed that each Zn(II) is tetrahedrally coordinated by four triazolate nitrogen atoms from three batz^2– ligands (Supplementary Fig. 1), and each batz^2– ligand coordinates to three Zn(II) ions in a bisimidazolate mode, giving a three-dimensional (3D) coordination framework with narrow 1D zigzag channels (Fig. 1a). Since only four of the eight nitrogen donors of batz^2– are utilized according to the coordination requirement of Zn(II), the pore surface of MAF-49 is rich with electronegative nitrogen atoms, although some of them form intra-framework N–H···N hydrogen bonds to reduce their abilities as hydrogen-bonding acceptors. Notably, the narrowest section of the 1D channel (3.3 × 3.0 Ų) is approximately a folded four-membered ring defined by a pair of free amino groups (with their lone electron pairs) and a pair of methylene groups with a cis-configuration, which is occupied by a guest H₂O molecule with two O–H···N and two C–H···O hydrogen bonds (Fig. 1b).

Figure 1. X-ray crystal structure of MAF-49·H₂O.

(a) Framework (Zn purple, C dark grey, H light grey, N blue) and pore surface (yellow/grey curved surface) structures. Guest molecules are omitted for clarity. (b) Local environment and hydrogen-bonding interactions of the narrowest channel neck (highlighted by green dashed lines).

Thermogravimetry and powder XRD showed that MAF-49·H₂O can be readily activated and is stable to 450 °C in nitrogen (Supplementary Fig. 2), in boiling water for at least 1 month and in aqueous acid/base (4≤pH≤12) at room temperature for at least 1 week (Supplementary Fig. 3), which is extraordinary among MOFs and can be partly explained by the strong metal-azolate coordination bonds³⁸. SCXRD showed that complete dehydration leads to a slight framework expansion (0.17% in volume, Supplementary Table 2 and Supplementary Data 2).

Gas adsorption property and mechanism

Single-component adsorption isotherms for CH₄, C₂H₆, C₂H₄ and CO₂ were measured for guest-free MAF-49 at 298 K, 307 K and 316 K (Fig. 2a and Supplementary Fig. 4). According to their different isotherm shapes, it can be judged that the host–guest binding follows C₂H₆>C₂H₄>CO₂>CH₄. The gas adsorption enthalpies were calculated quantitatively by Virial analyses (Fig. 2b and Supplementary Fig. 5), which are 60 kJ mol⁻¹, 48 kJ mol⁻¹, 30 kJ mol⁻¹ and 25 kJ mol⁻¹ for C₂H₆, C₂H₄, CO₂, and CH₄, respectively, at zero-coverage. The mixed gas adsorption isotherms for equimolar C₂H₆/C₂H₄, C₂H₆/CO₂ and C₂H₆/CH₄ mixtures were simulated by the ideal adsorbed solution theory³⁹, in which the single-component adsorption isotherms were fitted by the Langmuir−Freundlich model (Supplementary Fig. 6). At total pressure of 100 kPa and a temperature of 316 K, the C₂H₆/C₂H₄, C₂H₆/CO₂ and C₂H₆/CH₄ selectivities of these mixtures were calculated as ca. 9, 40 and 170, respectively (Supplementary Fig. 7). Notably, the C₂H₆/C₂H₄ selectivity of MAF-49 is much higher than the highest value in the literature (2.4 for IRMOF-8 at 318 K) (ref. ³¹). Except for CH₄ with obviously lower molecular weight and boiling point, which interacts weakly with all adsorbents, the binding strength order of MAF-49 for other three heavier gases is unusual. Among a variety of physical properties of the four gases, only the polarizability trend is consistent with the binding trend (Supplementary Table 1). Nevertheless, the small differences of their polarizabilities are not enough to explain the large variation of their adsorption enthalpies, especially for C₂H₆ and C₂H₄. Notably, the C₂H₆ adsorption enthalpy is significantly higher than reported values, while the C₂H₄ one is moderate¹⁸.

Figure 2. Single-component gas adsorption properties.

(a) Gas adsorption isotherms for C₂H₆, C₂H₄, CO₂ and CH₄ in MAF-49 at 316 K. (b) The coverage-dependent C₂H₆, C₂H₄, CO₂ and CH₄ adsorption enthalpy obtained by the Virial method. (c) C₂H₆ adsorption isotherms of MAF-49, MAF-3, MAF-4 and IRMOF-8 measured at 316 K. (d) Coverage-dependent C₂H₆ adsorption enthalpy of MAF-49, MAF-3, MAF-4 and IRMOF-8.

To elucidate the very different C₂H₆, C₂H₄ and CO₂ affinities of MAF-49, their preferential host–guest structures and energy changes were calculated by grand canonical Monte Carlo simulation and further periodic density functional theory optimization. The obtained binding energies of the final host–guest structures are −56.7, −45.5 and −41.3 kJ mol^–1 for C₂H₆, C₂H₄ and CO₂, respectively. However, to adsorb these gas molecules, the host framework undergoes different structural distortions from the guest-free form and consumes energies of +0.2 kJ mol^–1, +0.3 kJ mol^–1 and +5.6 kJ mol^–1, respectively. Taking both the host–guest binding and host-framework distortion into consideration, the total energies or adsorption enthalpies can be calculated as −56.5 kJ mol^–1, −45.2 kJ mol^–1 and −35.7 kJ mol^–1 for C₂H₆, C₂H₄, and CO₂, respectively, which are consistent with the experimental values (Supplementary Table 3). In the density functional theory optimized host–guest structures, it can be seen that C₂H₆, C₂H₄ and CO₂ are all adsorbed in or very close to the narrowest channel neck, but they interact very differently with the pore surface.

C₂H₆ forms three strong C–H···N hydrogen bonds and three weak C–H···N electrostatic interactions with MAF-49 (Fig. 3a,d, Supplementary Table 4). Specifically, one methyl group interacts with two amino groups and an coordinated triazolate nitrogen atom of the narrowest channel neck, forming one very short and directional (C6-H61···N8) and one unsymmetrical-bifurcated/three-centred (C6-H62···N8A/N3A) hydrogen bonds, in which the H···N separations (2.15 Å) are much shorter than the sum of van der Waals radii of nitrogen (1.55 Å) and hydrogen (1.20 Å) atoms. The third strong hydrogen bond involves the hydrogen atom (H71) of another methyl group and a coordinated triazolate nitrogen atom (C8-H71···N1A), which is approximately centro-symmetric with the strongest one (C6-H61···N8) about the molecular centre and fits well with the most stable stagger conformation of C₂H₆. Besides, the less polar part of the pore surface, that is, two methylene groups of the batz^2– ligand (C3), fits well with the guest C₂H₆ molecule in the context of both molecular shape and electrostatic potential.

Figure 3. Host–guest fittings and interactions.

Preferential adsorption sites for (a) C₂H₆, (b) C₂H₄ and (c) CO₂ in MAF-49 revealed by computational simulations (Zn purple, C dark grey, H light grey, N blue). Schematic representation of the corresponding host–guest interactions for (d) C₂H₆, (e) C₂H₄ and (f) CO₂. Strong (H···N/O<2.3 Å), weak (2.3 Å<H···N/O<2.8 Å) and almost negligible (H···N/O>2.8 Å) C–H···N interactions are displayed as red, blue and black dashed lines, respectively. 3D electron density maps (F_o–F_c contoured at 0.80 e Å⁻³ in yellow) of MAF-49 loaded with trace amounts of (g) C₂H₆, (h) C₂H₄ and (i) CO₂.

For C₂H₄, two less strong C–H···N hydrogen bonds and two very weak C–H···N electrostatic interactions were observed (Fig. 3b,e and Supplementary Table 4). The strongest one involves one methylene group and one amino group at the narrowest channel neck (C6-H61···N8), while the secondary one involves another methylene group and an uncoordinated triazolate nitrogen atom (C7-H71···N2A), which are also approximately centro-symmetric about the molecular centre. These two C–H···N hydrogen bonds are similar in geometry with the first and third strongest ones for C₂H₆. However, their H···N separations (2.54–2.65 Å) are obviously longer, albeit C₂H₄ is more polar (Supplementary Table 4). The cis-configuration of the two electronegative amino groups and two electropositive methylene groups of the narrow channel neck is crucial for the very different host–guest interactions. Obviously, the molecular geometry of C₂H₄ prevents the two hydrogen atoms of a methylene group to form two strong hydrogen bonds with the narrow channel neck like H₂O and C₂H₆. Furthermore, there is significant steric hindrance and electrostatic repulsion between the two C–H moieties of the two methylene groups from the host channel neck and the guest C₂H₄ (C3···C6=3.88 Å, Supplementary Fig. 8), which pushes the guest away from the best position for forming a strong C–H···N hydrogen bond with the p-position amino group. Conversely, the methylene group of the host channel neck fits well with the threefold symmetric methyl group of C₂H₆ (Fig. 3a). For the less strong C–H···N hydrogen bonds and other weak electrostatic attractive interactions, C₂H₆ also fits much better with the locations of the electronegative nitrogen atoms, as compared with those for C₂H₄ (Fig. 3 and Supplementary Table 4). These observations indicate that the proper locations of both the electronegative nitrogen atoms and the electropositive methylene groups play critical roles in distinguishing C₂H₆ and C₂H₄ with large adsorption enthalpy difference.

In the simulated host–guest structure for CO₂, the guest carbon atom locates exactly at the centre of host channel neck, forming short contacts with two free amino groups simultaneously (N···C=2.91 Å), while two oxygen atoms of CO₂ interact with two methylene groups, respectively, through weak C–H···O hydrogen bonds (C···O=3.33, H···N=2.45 Å, ∠C–H···N=135°) (Fig. 3c,f, Supplementary Table 4). Although these host–guest interactions seem relatively strong, the channel neck diameter (measured by the separation of the p-position amino and methylene groups, N8···C3 3.60 Å) significantly expanded from the guest-free state (3.13 Å), while it changes little after loading C₂H₆ (3.18 Å) and C₂H₄ (3.31 Å), indicating that there is significant steric hindrance and repulsive effect between the CO₂ molecule and the host framework, and the very short C···N separation is actually the result enforced by the contraction action of the channel neck (Supplementary Fig. 9). It should be noted that all carbon atoms of C₂H₆ and C₂H₄ reside on one side of the quadrangular channel neck, resulting in much smaller steric hindrance effects compared with CO₂ (Supplementary Fig. 10).

To confirm the simulation results and directly visualize the host–guest interactions, we carried out SCXRD analyses for MAF-49 loaded with trace amounts of C₂H₆, C₂H₄ and CO₂ (denoted as MAF-49·C₂H₆, MAF-49·C₂H₄ and MAF-49·CO₂, respectively, see Supplementary Table 2 and Supplementary Data 3,4,5). Compared with the unit-cell volume of guest-free MAF-49, those of MAF-49·C₂H₆ and MAF-49·C₂H₄ showed minor shrinkage (<0.2%), while that of MAF-49·CO₂ showed relatively large expansion (1.4%). Further, the N8···C3 separation order of MAF-49, MAF-49·C₂H₆, MAF-49·C₂H₄ and MAF-49·CO₂ is consistent with that predicted by computational simulations (Supplementary Fig. 10). In all host–guest crystal structures, the residue electron density peaks can be unambiguously found inside the narrow host channel neck (Fig. 3g–i). Furthermore, in the final crystal structures, all guest molecules locate very similar or identical with those predicted by computational simulations (Supplementary Fig. 10).

Mixed gas separation

To investigate the practical separation performance of MAF-49, breakthrough experiments were carried out at 313 K and 1 bar. To evaluate and compare the performances of the materials unambiguously, identical column and flow rate were used, and the parameters of each column were optimized (all columns have similar voidage, Supplementary Table 5). Besides, we used the specific injection amount (mmol g^–1) of the mixed gas as the abscissa, meaning that the breakthrough time (s) was not only divided by the adsorbent weight (g) but also multiplied by the flow rate of the injected mixed gas (mmol s^–1)⁴⁰.

To compare the gas adsorption and separation properties of MAF-49 with other protopytical MOFs, breakthrough experiments using an equimolar C₂H₆/C₂H₄/CO₂/CH₄ mixture injection were carried out (Fig. 4 and Supplementary Figs 11 and 12). For MAF-49, a clean and sharp separation of all four gases was observed, while other MOFs showed much poor separation performances and complicated effluent sequences dependent on their pore surface structures. With transition-metal OMSs, [Cu₃(btc)₂] (HKUST-1, H₃btc=benzene-1,3,5-tricarboxylic acid) and [Co₂(dobdc)] (MOF-74-Co/CPO-27-Co, H₄dobdc=2,5-dihydroxyl-1,4-benzenedicarboxylic acid) showed binding strength orders C₂H₄>C₂H₆>CO₂. Because the main-group-metal OMS tends to form strong interaction with the oxygen atom of CO₂, [Mg₂(dobdc)] (MOF-74-Mg/CPO-27-Mg) showed a binding strength order CO₂>C₂H₄>C₂H₆. Without pore surface active site, [Zr₆O₄(OH)₄(bdc)₁₂] (UiO-66, H₂bdc=1,4-benzenediarboxylic acid) and [Zn(mim)₂] (MAF-4 or ZIF-8, Hmim=2-methylimidazole) can barely distinguish the three heavier gases. Nevertheless, the low-polarity adsorbent MAF-4 exhibits slightly better performance compared with UiO-66, and exhibits a separation order similar with that of MAF-49. As expected from the analyses of adsorption isotherms, MAF-49 can also clearly separate two-component C₂H₄/C₂H₆, C₂H₆/CO₂ and C₂H₆/CH₄ mixtures (Supplementary Fig. 13). It should be noted that C₂H₆ could not be detected before its breakthrough points, meaning that C₂H₆ is efficiently extracted and high-purity C₂H₄/CO₂/CH₄ can be obtained directly.

Figure 4. Four-component gas mixture separation.

Breakthrough curves of CH₄/CO₂/C₂H₄/C₂H₆ mixture (1:1:1:1 (vol)) for (a) CPO-27-Mg, (b) CPO-27-Co, (c) MAF-4, (d) UiO-66, (e) HKUST-1 and (f) MAF-49 measured at 313 K and 1 bar. Lines are drawn to guide eyes. C_i and C_o are the concentrations of each gas at the inlet and outlet, respectively.

Considering that selective adsorption of C₂H₆ over C₂H₄ could be beneficial for purification of C₂H₄ under fixed-bed adsorption/breakthrough processes, and some hydrophobic MOFs²⁹,³⁰,³¹,³²,³³,³⁴,³⁵,³⁶, such as [Zn(bim)₂] (MAF-3 or ZIF-7, Hbim=benzimidazole), MAF-4 and [Zn₄O(ndc)₃] (IRMOF-8, H₂ndc=naphthalene-2,6-dicarboxylic acid), were recently reported to exhibit such a property, we compared the C₂H₄/C₂H₆ adsorption and separation properties of these MOFs with MAF-49 in detail. Single-component C₂H₆ adsorption isotherms were measured for MAF-3, MAF-4 and IRMOF-8, which show adsorption enthalpies of 25 kJ mol⁻¹, 18 kJ mol⁻¹ and 30 kJ mol⁻¹, respectively (Fig. 2c and Supplementary Fig. 14), at zero loading, being much lower than that of MAF-49. Although the C₂H₆ uptake at 1 bar for MAF-49 (38 cm³ g⁻¹) is lower than that of the more porous adsorbents IRMOF-8 (91 cm³ g⁻¹), MAF-4 (48 cm³ g⁻¹) and MAF-3 (41 cm³ g⁻¹), its C₂H₆ uptake at 0.06 bar (36 cm³ g⁻¹) is ca. 4 times that of IRMOF-8 (9 cm³ g⁻¹), 19 times that of MAF-4 (1.9 cm³ g⁻¹) and 45 times that of MAF-3 (0.8 cm³ g⁻¹) (Fig. 2d). Considering that a purity of 100% is impossible and the C₂H₆ concentration before its breakthrough point is lower than the detection limit of the conventionally used thermal conductivity detector, the gas stream at the column outlet was analysed with a mass spectrometer (MS). For a 1:1 C₂H₄/C₂H₆ mixture injection (Fig. 5a and Supplementary Fig. 15a), the breakthrough points of C₂H₄ and C₂H₆ for MAF-49 were observed by thermal conductivity detector at 1.09 and 1.44 mmol g⁻¹, respectively, during which the C₂H₆ concentration was determined as 0.014–0.016% by MS, corresponding to a C₂H₄ purity of 99.986–99.984% (Fig. 5a and Supplementary Figs 15a and 16). Under identical conditions, the highest C₂H₄ purities achieved by MAF-3, MAF-4 and IRMOF-8 are only 99.5%, 99.6% and 99.9% (C₂H₆ concentrations of 0.5%, 0.4% and 0.1%), respectively, reflecting their much lower C₂H₆/C₂H₄ selectivity compared with MAF-49 (Fig. 5b–d, Supplementary Table 6 and Supplementary Figs 15–19). Nevertheless, such C₂H₄ purities are obviously higher than those reported for C₂H₄ selective adsorbent materials (99%+)¹³,¹⁷,²³,²⁴, which exemplify the feasibility of using C₂H₆ selective adsorbents for purifying C₂H₄, because the desired gas can be continuously purified by passing through the column and directly obtained from the first effluents. Indeed, desorbing the MAF-49 column saturated by 1:1 C₂H₄/C₂H₆ mixture can give C₂H₆ with 99%+ purity with a peak value of only 99.7% (Supplementary Fig. 20). A realistic comparison for the C₂H₄ purification performance of different adsorbents, of relevance to industrial operations, can be obtained by comparing the breakthrough amount of C₂H₄ (denoted as productivity) with the desired purity in a single breakthrough operation (for the calculation method see Supplementary Methods). For the MAF-49 column, 0.28 mmol g⁻¹ or 0.44 mol l⁻¹ of C₂H₄ with 99.95%+ purity can be recovered from a 1:1 C₂H₄/C₂H₆ mixture injection. For the MAF-3, MAF-4 and IRMOF-8 columns, their productivities are zero because the C₂H₄ effluents are not pure enough. Even for a C₂H₄ purity of 99%+, the productivity of the MAF-49 column (0.32 mmol g⁻¹ or 0.47 mol l⁻¹) is still much higher than the others (the largest value is 0.11 mmol g⁻¹ or 0.10 mol l⁻¹) (Supplementary Table 6).

Figure 5. C₂H₄/C₂H₆ separation performances.

C₂H₄/C₂H₆ (1:1) mixture breakthrough curves of (a) MAF-49, (b) MAF-3, (c) MAF-4 and (d) IRMOF-8, and C₂H₄/C₂H₆ (15:1) mixture breakthrough curves of (e) MAF-49, (f) MAF-3, (g) MAF-4 and (h) IRMOF-8 measured at 313 K and 1 bar. Solid symbols: C₂H₄, Open symbols: C₂H₆. Lines are drawn to guide eyes. C_i and C_o are the concentrations of each gas at the inlet and outlet, respectively. Horizontal red dashed lines highlight C₂H₆ composition at outlet of 0.05%, that is, C₂H₄ purity of 99.95%.

Since the C₂H₆ concentration in C₂H₄/C₂H₆ mixtures produced by hydrocarbon cracking is just ca. 5–9% (refs ^{41, 42, 43}), breakthrough experiments using a 15:1 C₂H₄/C₂H₆ mixture injection were carried out. The lowest C₂H₆ impurity or highest C₂H₄ purities achieved by the MAF-49, MAF-3, MAF-4 and IRMOF-8 columns are decreased to 0.005%, 0.4%, 0.1% and 0.04% or improved to 99.995%, 99.6%, 99.9% and 99.96%, respectively (Fig. 5e–h and Supplementary Figs 21–24). Obviously, using C₂H₆ selective adsorbents, the C₂H₄ purity can be increased by lengthening the adsorbent bed (increasing adsorbent amount), which is simpler and more convenient than the C₂H₄ selective adsorbents¹³,¹⁷,²⁵. For the 15:1 C₂H₄/C₂H₆ mixture injection and the C₂H₄ output purity of 99.95%+, the MAF-49 column gave a C₂H₄ productivity of 1.68 mmol g⁻¹ or 2.48 mol l⁻¹, which is about 30 or 50 times that of IRMOF-8 (0.06 mmol g⁻¹ or 0.05 mol l⁻¹), in the gravimetric or volumetric point-of-view, respectively (Supplementary Table 7). Note that for C₂H₄ purification, the adsorbent volume is more practical than its weight because the fixed-bed equipment does not need to move during operation. For lower C₂H₄ purities such as 99.5%+ and 99%+, the C₂H₄ productivities of MAF-49 and IRMOF-8 were also increased (Supplementary Tables 6 and 7), because the adsorber needs more time to reach adsorption saturation for the mixture gas containing low-concentration C₂H₆. Nevertheless, the C₂H₄ productivity of MAF-49 improved more significantly than for IRMOF-8 at all purity standards (Supplementary Tables 6 and 7), because the former material exhibits much higher C₂H₆ uptakes at the low pressure region (Fig. 2c). In contrast, the MAF-3 and MAF-4 columns only showed slightly increased C₂H₄ purities (did not reach 99.95%+) at a C₂H₄/C₂H₆ feeding ratio of 15:1 (Fig. 2c), because lengthening the adsorber is not so effective to improve the effluent purity by using adsorbents with weak impurity affinity. For C₂H₄ purities of 99.5%+ and 99%+, the C₂H₄ productivities of the MAF-3 and MAF-4 columns obtained by using a 15:1 C₂H₄/C₂H₆ input were unexpectedly lower than for the 1:1 C₂H₄/C₂H₆ mixture (Supplementary Tables 6 and 7), which can be attributed to the extremely low C₂H₆ adsorption ability of the adsorbents at the low pressure region. Also, the partial pressures of C₂H₄ and C₂H₆ in the 15:1 C₂H₄/C₂H₆ mixture are not beneficial for utilizing the differential gate-opening effect of MAF-3 (ref. ⁴⁴).

Discussion

In summary, we reported a unique adsorbent material showing selective adsorption of C₂H₆ over more polar analogous molecules such as C₂H₄ and CO₂, which can be useful for extraction of C₂H₆ from natural gases and particularly valuable for direct production of high-purity C₂H₄ from C₂H₄/C₂H₆ mixtures. The key to this C₂H₆ selectivity is a combination of multiple hydrogen-bonding acceptors and dipole repulsion groups locating at appropriate positions on the pore surface of a very narrow channel, which not only allows multiple attractive interactions for C₂H₆ but also enforces C₂H₄ to adopt a position that can only form fewer and weaker attractive interactions. In short, this work provides not only a new MOF with exceptionally high C₂H₄ separation/purification performances, but also a new molecular design strategy for developing next-generation adsorbents.

Methods

Materials and general methods

Reagents and solvents were commercially available and were used without further purification, H₂batz (ref. ⁴⁵), MAF-3 (ref. ³⁰), MAF-4 (ref. ⁴⁶), IRMOF-8 (ref. ⁴⁷), HKUST-1 (ref. ⁴⁸), CPO-27-Mg (ref. ⁴⁹), CPO-27-Co (ref. ⁴⁹) and UiO-66 (ref. ⁵⁰) were synthesized according to the literature methods. Elemental analyses (C, H, N) were performed with a Vario EL elemental analyzer. Thermogravimetry analysis was performed under N₂ with temperature increased with 5 °C min^–1 using a TA-Q50 system. Powder XRD patterns were collected on a Bruker D8 Advance diffractometer (Cu Kα) at room temperature.

Synthesis of [Zn(batz)]·0.5H₂O (MAF-49·H₂O)

A mixture of Zn(OH)₂ (0.100 g, 1.0 mmol), H₂batz (0.180 g, 1.0 mmol), aqueous ammonia (25%, 4 ml) and water (4 ml) was stirred for 15 min in air, then transferred and sealed in a 15 ml Teflon reactor, which was heated in an oven at 160 °C for 72 h. The oven was cooled to room temperature at a rate of 5 °C h⁻¹. The resulting colourless block crystals were filtered, washed and dried in air (yield ca. 86%). Anal. Calcd (%) for C₅H₇N₈O_0.5Zn: C, 23.77; H, 2.79; N, 44.36. Found: C, 23.97; H, 2.82; N, 44.13. Guest-free MAF-49 was obtained by heating the as-synthesized sample under high vacuum at 150 °C for 12 h.

Single-crystal X-ray crystallography

Diffraction intensities were collected on a Pilatus XtaLAB P300DS diffractometer with graphite-monochromated Mo Kα radiation. Absorption corrections were applied by using the multi-scan programme REQAB. The structures were solved by the direct method and refined with the full-matrix least-squares technique using the SHELXTL programme package. It should be noted that, because the molecular centres of the very short C₂H₆ and C₂H₄ molecules do not locate at the centre of the two-fold symmetric host channel neck as predicted by computational simulations, their molecular geometries have to be restricted during refinement of the crystal structures. Also, the positions of their hydrogen atoms were added according to the computational simulation result. Because of disorder and low occupancies of the gas molecules, anisotropic thermal parameters were only applied to all non-hydrogen atoms of the host framework. Crystal data for the compounds were summarized in Supplementary Table 2. Electron density maps were generated using the output from standard SHELXL refinements in a number of ways using WinGX and VESTA 3.0.8.

Gas sorption measurement

The sorption isotherms were measured with an automatic volumetric adsorption apparatus (BELSORP-max). The as-synthesized sample (about 200−300 mg) was placed in the sample tube and dried for 12 h at 320 °C to remove the remnant solvent molecules prior to measurement. CO₂ (99.999%), C₂H₄ (99.95%), CH₄ (99.999%) and C₂H₆ (99.99%) were used for all adsorption isotherm and breakthrough experiments (Supplementary Fig. 25). The temperatures were controlled by a water bath (298, 307 and 316 K).

Additional information

Accession codes: The X-ray crystallographic coordinates for structures reported in this Article have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number CCDC 1421354-1421358. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

How to cite this article: Liao, P.-Q. et al. Efficient purification of ethene by an ethane-trapping metal-organic framework. Nat. Commun. 6:8697 doi: 10.1038/ncomms9697 (2015).

Supplementary Material

Supplementary Information

Supplementary Figures 1-25, Supplementary Tables 1-7, Supplementary Methods and Supplementary References

Supplementary Data 1

X-ray single-crystal structures of as-synthesized MAF-49 in CIF format

Supplementary Data 2

X-ray single-crystal structures of guest-free MAF-49 in CIF format

Supplementary Data 3

X-ray single-crystal structures of C2H6 included MAF-49 in CIF format

Supplementary Data 4

X-ray single-crystal structures of C2H4 included MAF-49 in CIF format

Supplementary Data 5

X-ray single-crystal structures of CO2 included MAF-49 in CIF format