One-step ethylene separation from ternary C2 hydrocarbon mixture with a robust zirconium metal–organic framework

Yuan Liu, Hanting Xiong, Jingwen Chen*, Shixia Chen Zhenyu Zhou Zheling ZengShuguang Deng, Jun Wang*

1 School of Chemistry and Chemical Engineering, Nanchang University, Nanchang 330031, China

2 School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ 85287, United States

Keywords:

ABSTRACT

1. Introduction

Ethylene (C2H4) is a fundamental feedstock in the petrochemical industry and extensively employed to produce polymers and various fine chemicals [1]. Generally, C2H4is produced from the dehydrogenation of ethane (C2H6) and steam cracking of naphtha,which inevitably coexists with certain amounts of acetylene(C2H2)and C2H6impurities[2].The presence of C2H2will poison the C2H4polymerization catalysts even at μl?L–1level (1000–5000 μl?L–1);meanwhile,the existence of C2H6will reduce production efficiency[3]. Generally, high-purity C2H4is obtained through a two-step separation process, in which C2H2is eliminated by partial hydrogenation or solvent extraction, and C2H6is removed by energyintensive cryogenic distillation under cryogenic temperature(248 K) and high pressure (2 MPa) with more than 150 trays [4].The energy input for olefin purification accounts for about 0.3%of global energy consumption[5]. Therefore,it is urgent to exploit one-step and energy-efficient separation technologies for C2H4separation from C2H2/C2H4/C2H6gas-mixtures [6–8].

Adsorptive separations are possibly the only established and proven method for one-step C2H4purification [9,10]. The critical factor lies in the development of high-performance adsorbents[9]. Metal–organic frameworks (MOFs) are promising platforms for one-step C2H4purification because of their almost infinite combinations of organic linkers and metal ions/clusters [11,12]. The aperture sizes of MOFs can be precisely tuned to realize molecular sieving separation [13], and specific binding sites can be introduced to enhance the host–guest interactions based on thermodynamic mechanisms [14]. Currently, the development in MOFs mainly focuses on separating binary gas-mixtures and has made significant processes such as for C2H2/C2H4[15–18], C2H4/C2H6[19,20], and C2H2/CO2[21,22] separations. Nevertheless, designing and developing an efficient MOF adsorbent for one-step C2H4separation from C2H2/C2H4/C2H6gas-mixture are still highly challenging. Because, as shown in Table S1 (in Supplementary Material), the kinetic diameter, polarizability, and quadrupole moment of C2H4(0.42 nm, 42.52 × 10–25cm3, and 0.5 × 10–35C?cm2) are just between C2H2(0.33 nm, 39.3 × 10–25cm3, and 2.4 × 10–35C?cm2)and C2H6(0.44 nm, 44.7 × 10–25cm3, and 0.2 × 10–35(C?cm2). In 2019, Zaworotko et al. [23] reported a synergistic sorbent separation technology (SSST) using a combination of three MOFs for one-step C2H4purification from C2H6/C2H2/C2H4and C2H6/C2H4/C2H2/CO2gas-mixtures. Apparently, developing a single adsorbent that can simultaneously adsorb both C2H2and C2H6over C2H4will significantly improve energy- and capital-efficiency.

Generally,MOF adsorbents with strong electron polar sites,e.g.,nitrogen and fluorine, usually demonstrate a high binding affinity for C2H2[24],whereas the preferential adsorption of C2H6requires inert and non-polar pore surfaces [25]. Therefore, rare adsorbents can realize the one-step C2H4separation from three-component C2hydrocarbon mixtures [11,15,18,26]. For example, Hao et al.[27] reported a MOF adsorbent, TJT-100, with methyl-rich nonpolar pores for selective C2H2and C2H6adsorption from C2H2/C2H6/C2H4gas-mixture.Li and co-workers introduced Lewis basic amine groups into the organic linkers of MOFs to construct C2H2/C2H6-selective adsorbent UiO-67–(NH2)2[18]. The suitable pore size and functional groups synergistically lead to stronger van der Waals interactions toward C2H2and C2H6over C2H4. Recently,our group has developed a hybrid ultramicroporous material,CuTiF6–TPPY, in which the anion pillars (TiF62-) can provide strong binding affinity for C2H2and the porphyrin moieties with large πsurfaces can selectively interact with C2H6[11]. The synergistic interaction system enabled the one-step purification of C2H4from C2H2/C2H6/C2H4gas-mixtures. Despite these achievements, stable MOFs showing both high gas uptakes and selectivity for one-step C2H4purification are urgently wanted.

Herein, we report a stable Zr-based MOFs, Zr–TCA (TCA = 4,4′,4′′-tricarboxytriphenylamine), for efficient one-step separation of C2H4from C2H2/C2H6/C2H4gas-mixtures. The suitable pore size of 0.6 nm×0.7 nm and abundant oxygen atoms on the pore channels render preferential adsorption toward C2H2and C2H6over C2H4. Specifically, the separation selectivity of C2H2/C2H4and C2H6/C2H4reaches 5.64 and 2.72 at 100 kPa and 298 K, respectively. Computational simulations reveal that the oxygen atoms of Zr6(μ3-O)6(COO)6clusters and phenyl rings of organic ligands can provide multiple C—H???O, C—H???π, and van der Waals interactions for selective C2H2and C2H6adsorptions. Furthermore,dynamic breakthrough experiments confirm the feasibility of yielding high-purity C2H4(>99.9%) from binary C2H6/C2H4(50/50 and 10/90, volume ratio) and C2H2/C2H4(1/99 and 50/50, volume ratio) and ternary C2H2/C2H6/C2H4(1/9/90, volume ratio) gasmixtures in a single step. The structural and chemical stability of Zr–TCA have been systematically evaluated as well.

2. Experimental

2.1. Synthesis of Zr–TCA

4,4′,4′′-Tricarboxytriphenylamine (H3TCA) (0.377 g, 1 mmol)and zirconium(IV)nitrate[Zr(NO3)4?5H2O](0.429 g,1 mmol)were introduced into a mixed solvent of N,N-dimethylformamide(DMF,5 ml)and acetic acid(HAc,5 ml).After fully dissolving under continuous stirring at room temperature,the solution was transferred to a Teflon vessel (25 ml). Subsequently, the Teflon vessel was sealed and heated at 393 K in an oven for 72 h. After cooling to room temperature, the solid products were separated by centrifugation and washed thoroughly with DMF and methanol. The obtained products were dried at 333 K under vacuum for 14 h.The solvent-free sample for gas adsorption experiments was obtained after activation at 373 K till no mass change was observed. Yield: ca. 60.8% based on Zr(NO3)4?5H2O.

2.2. Gas adsorption tests

The adsorption–desorption isotherms of C2H2,C2H4,and C2H6at 273, 298, and 323 K were measured on a 3Flex adsorption device(Micromeritics, USA). The time-dependent adsorption profiles of C2H4and C2H6were collected on intelligent gravimetric analyzer(IGA-100, HIDEN, UK). Zr–TCA was treated under high vacuum(<9.2 × 10–4kPa) at 373 K for 12 h before gas adsorption tests.

3. Results and Discussion

3.1. Characterizations

Despite extensive attempts, high-quality single crystals of Zr–TCA was failed to be obtained for single-crystal X-ray diffraction analysis, therefore Rietveld refinement of powder diffractometry was used to determine the crystal structure (Fig. S1 and Tables S2 and S3). As shown in Fig. 1(a), six Zr atoms were linked by six carboxylate ligands and six μ3-O to form Zr6(μ3-O)6(COO)6clusters,which were further connected by μ2-O groups to generate onedimensional (1D) chains across xy-plane. These 1D chains were connected by TCA3-linkers to generate 1D hexagonal channels along the z plane and extended to form three-dimensional porous networks,depicted as the yellow hexagonal prism in Fig.1(b).The Zr6(μ3-O)6clusters were capped by terminal μ3-O and μ2-O groups,thus no open metal sites (OMSs) existed on the pore walls of Zr–TCA and the π-complexation between C2H2/C2H4and OMSs could be eliminated. Meanwhile, the oxygen atoms in Zr—O clusters can form C—H???O interactions with hydrocarbons, and the abundant aromatic rings can create appropriate pore environments for selective C2H6adsorption through C—H???π interactions [25].

The aperture size of the 1D channel of Zr–TCA is measured to be 0.6 nm × 0.7 nm (Fig. 1(c)). The powder X-ray diffraction (PXRD)pattern of Zr–TCA is consistent with the calculated one (Fig. 1(d)), demonstrating the high phase purity of the as-synthesized microcrystals. After activation, the unchanged PXRD pattern revealed the rigidness of Zr–TCA. Thermogravimetric analysis(TGA) result indicated that Zr–TCA was stable up to 723 K(Fig. S2). The permanent porosity of Zr–TCA was determined by N2adsorption isotherm at 77 K (Fig. 1(e)). The rapid increase of adsorption amount at the relative low-pressure region (P/P0< 0.01) disclosed its microporous nature. The Brunauer–Emme tt–Teller (BET) specific surface area and t-plot micropore volume of Zr–TCA were determined to be 350.9 m2?g-1and 0.18 cm3?g-1,respectively. The pore size distribution of Zr–TCA estimated by the non-local density functional theory (NLDFT) model showed a centered pore size at 0.65 nm, which agreed well with the pore sizes derived from crystal analysis (Fig. 1(f)).

3.2. Adsorption and separation selectivity of Zr–TCA

The single-component gas adsorption isotherms of C2H2, C2H4,and C2H6were collected at 273, 298, and 323 K (Figs. 2(a) and S3). Specifically, Zr–TCA displayed an inverse adsorption trend for C2hydrocarbons,and the adsorption capacity was in the following order: C2H2(2.78 mmol?g-1) > C2H6(2.28 mmol?g-1) > C2H4(2.02 mmol?g-1) at 100 kPa and 298 K, indicating the selective adsorption of C2H6and C2H2over C2H4(Fig. 2(a)). Notably, the C2H6uptake (2.28 mmol?g-1) surpassed most of the benchmark adsorbents for one-step C2H4purification such as Al–NDC(2.23 mmol?g-1) [26], MUV-11 (1.83 mmol?g-1) [14], Zn–ATA(1.10 mmol?g-1) [28], and Zn-atz-oba (2.04 mmol?g-1) [29](Table S4).

Fig.1. (a)The structure of Zr6 cluster.(b)View of the 3D structure of Zr–TCA with 1D hexagonal channels across xy-plane.Color code:Zr,cyan;C,gray;N,blue;O,red;H,light gray.(c)A schematic illustration of the pore sizes of Zr–TCA view along the z-axis;unit:nm.(d)PXRD patterns,(e)N2 adsorption–desorption isotherms at 77 K,and(f)pore size distribution based on the NLDFT model of Zr–TCA.

Fig.2. (a)C2H2,C2H4,and C2H6 adsorption isotherms and(b)C2H2/C2H4(1/99)and C2H6/C2H4(50/50)IAST selectivity of Zr–TCA at 298 K.(c)Comparison of IAST selectivity and C2H6 uptake performances at 298 K and 100 kPa.(d)Comparison of IAST selectivity for C2H2/C2H4(1/99)on one-step C2H4 purification adsorbents.(e)Comparison of IAST selectivity for C2H2/C2H4 (50/50) and C2H6/C2H4 (50/50) on one-step C2H4 purification adsorbents. (f) Qst of C2H2, C2H6, and C2H4 on Zr–TCA.

The ideal adsorption solution theory (IAST) was employed to evaluate the selectivity for C2H2/C2H4(1/99) and C2H6/C2H4(50/50) gas-mixtures by fitting the adsorption isotherms of C2H2,C2H4, and C2H6through the dual-site Langmuir–Freundlich model(Figs. S4–S6 and Table S5). The IAST selectivity of C2H6/C2H4(50/50) on Zr–TCA was calculated to be 2.72 at 298 K and 100 kPa (Fig. 2(b)), which is the second highest in C2H4one-step purification adsorbents, including TJT-100 (1.2) [27], and MUV-11(1.53)[14],just inferior to NPU-3(3.21)[15](Table S4).This property enabled the potential for one-step C2H4purification from C2H2/C2H6/C2H4gas-mixtures. Moreover, Zr–TCA broke the‘‘trade-off” effect by exhibiting a good balance between C2H6/C2H4separation selectivity and C2H6adsorption capacity (Fig. 2(c)and Table S6).Meanwhile,the C2H2/C2H4(1/99)IAST selectivity of 5.64 for Zr–TCA was only lower than MUV-11 (6.92) [14] and CuTiF6–TPPY (5.47, 50/50) [11] among one-step C2H4separation adsorbents (Fig. 2(b) and (d)). Noticeably, the superior IAST selectivity for both C2H2/C2H4(5.21)and C2H6/C2H4(2.72)rendered Zr–TCA as a new benchmark for one-step C2H4separation from ternary C2H2/C2H4/C2H6mixtures (Fig. 2(e)).

To evaluate the binding affinity of Zr–TCA for C2hydrocarbons,the Clausius–Clapeyron equation was applied to calculate the isosteric heat of adsorption (Qst) for C2H2, C2H6, and C2H4(Fig. S9 and Table S7). At near zero-coverage, the Qstincreased in the sequence of C2H4(23.9 kJ?mol-1) < C2H6(35.3 kJ?mol-1) < C2H2(43.8 kJ?mol-1) (Fig. 2(f)), which is consistent with the trend of thermodynamic gas uptakes. The Qstof C2H6was lower than most of the best-performing MOFs, such as Zn-atz-ipa (45.8 kJ?mol-1)[23], ZSTU-1 (43 kJ?mol-1) [30], NUM-7a (35.8 kJ?mol-1) [31],Fe2(O2)(dobdc) (66.8 kJ?mol-1) [32], ZJU-121a (47.1 kJ?mol-1)[25], and MAF-49 (61 kJ?mol-1) [33] (Table S6), illustrating facile and energy-efficient adsorbent regenerations. Furthermore, the kinetic adsorption tests on Zr–TCA showed an equilibrium time of 5 min for C2H6and 12 min for C2H4at 298 K and 50 kPa. The kinetic selectivity of C2H6/C2H4was calculated to be 1.61(Fig.S10).

3.3.GCMC and DFT-D simulations for C2H2/C2H6/C2H4 gas-mixtures on Zr–TCA

Fig.3. Theoretical simulations for C2 distribution density and binding sites in Zr–TCA.GCMC-simulated (a)C2H6 and(b)C2H4 distribution density on Zr–TCA at 10 kPa.The density of gas molecules was depicted according to the color scale.(c)GCMC simulations for the distribution density of ternary C2H2/C2H6/C2H4(1/9/90)at 298 K.The number of each colored dot represents the proportion of different C2 gas molecules. The DFT-calculated binding sites of (d, f) C2H2, (g) C2H6, and (h) C2H4 in Zr–TCA. (e) Synergistic interaction of C2H2 molecules adsorbed in neighboring site I and site II.Hydrogen atoms were omitted for clarity except in Fig(f).Framework:C,black;O,red;H,white;Zr,cyan. Gas: C, orange; H, white. Interaction: C—H???C, red; C—H???π, light blue; C—H???O, black broken lines; unit, nm.

To gain insights into the mechanism of preferential C2H2and C2H6adsorption over C2H4, the grand canonical Monte Carlo(GCMC) and first-principles dispersion-corrected density functional theory (DFT-D) calculations were carried out to probe the binding sites of C2H2, C2H6, and C2H4in Zr–TCA. The calculated distribution density at 298 K showed that C2H6and C2H4molecules were mainly adsorbed near the corner region along with the phenyl ring in dumbbell shaped areas (Fig. 3(a) and (b)), and the uptake increased with the increase of pressure (Fig. S11). The distribution density of C2H4was sparser than that of C2H6. In contrast, C2H2was preferentially adsorbed near the Zr–O cluster at a low pressure of 10 kPa (Fig. S12(a)), and more C2H2molecules gathered in the middle of the elliptical channel with increased pressure (Fig. S12(b)). Furthermore, the distribution densities of binary C2H6/C2H4(50/50) and ternary C2H2/C2H6/C2H4(1/9/90)gas-mixtures in Zr–TCA were determined by GCMC simulations(Figs.S13 and 3(c)).Denser distribution of C2H6was observed than that of C2H4for C2H6/C2H4(50/50)gas-mixture(Fig.S13).For ternary C2H2/C2H6/C2H4(1/9/90) mixture, the trace amount of C2H2and small proportion of C2H6(9%) can still be efficiently captured(Fig. 3(c)).

The adsorption sites of C2H2/C2H6/C2H4were further investigated by DFT-D calculations. The C2H2molecules were trapped by two distinct adsorption sites: (i) the C—H???O (0.208–0.291 nm) interaction with Zr6(μ3-O)6(COO)6clusters (site I,Fig. 3(d)), (ii) van der Waals (0.304–0.328 nm) interactions between C2H2and TCA ligands(site II,Fig.3(f)).Moreover,the suitable packing pattern of adsorbed C2H2molecules in neighboring site I and site II enabled the interactions between each C2H2molecules through Cδ–???Hδ+dipole–dipole interactions (Fig. 3(e)), further strengthening the C2H2adsorption. The binding energy (ΔE)of C2H2was calculated to be 54.72 kJ?mol-1(site I) and 33.19 kJ?mol-1(site II). Meanwhile, C2H6was adsorbed through C—H???O interactions (0.270–0.295 nm) by the oxygen atom from carboxylates and two oxygen atoms from Zr–O clusters as well as C—H???π interactions (0.321 nm) by phenyl groups (Fig. 3(g)),yielding a C2H6binding energy of 50.27 kJ?mol-1. The adsorption position of C2H4is similar to that of C2H2in site I, but showing longer distances of C—H???O interactions (0.290–0.335 nm, Fig. 3(h)). The binding energy (ΔE) of C2H4is calculated to be 36.53 kJ?mol-1, lower than the values of C2H2and C2H6, demonstrating the higher binding affinity toward C2H6and C2H2over C2H4.

3.4. Dynamic breakthrough experiments

Dynamic breakthrough tests for binary C2H2/C2H4(1/99 and 50/50), C2H6/C2H4(10/90 and 50/50), and ternary C2H2/C2H6/C2H4(1/9/90) gas-mixtures were conducted at 298 K to evaluate the practical separation performances of Zr–TCA. Clean separation of C2H2/C2H4(1/99) was achieved by Zr–TCA, high-purity C2H4(>99.9%) was first eluted from the column at 22.0 min?g-1, while C2H2was trapped and then eluted at 84.1 min?g-1with a highpurity C2H4(>99.9%)productivity of 60.68 L?kg-1(Fig.4(a)).When increasing the flow rate from 1.0 to 2.0 ml?min-1,high-purity C2H4can still be efficiently obtained from C2H2/C2H4mixture with a separation time of 29.3 min?g-1and a high-purity C2H4(>99.9%)productivity of 54.72 L?kg-1(Fig. 4(a)). Furthermore, C2H2/C2H4(50/50) gas-mixture could also be separated by Zr–TCA at 1.0 and 2.0 ml?min-1(Fig. S14). For C2H6/C2H4(10/90) gas-mixture,high-purity C2H4can be collected at 26.3 min?g-1, and the breakthrough point for C2H6was 35.0 min?g-1at 1.0 ml?min-1(Fig.4(b)).The dynamic adsorption capacity of C2H6and C2H4were calculated to be 0.17 and 1.13 mmol?g-1,respectively,and the productivity of high-purity C2H4(>99.9%) was 6.51 L?kg-1. Similarly, the dynamic breakthrough experiment of C2H6/C2H4(50/50) indicated that high-purity C2H4(>99.9%)can be obtained from a lower C2H4concentration with a productivity of 2.79 L?kg-1, and C2H6was eluted from the adsorption bed at 33.3 min?g-1(Fig.4(c)).The separation performance of Zr–TCA for a more challenging ternary C2H2/C2H6/C2H4(1/9/90)gas-mixture was further evaluated.High-purity C2H4first broke through the column at 26.8 min?g-1, followed by C2H6(34.5 min?g-1)and C2H2(61.3 min?g-1),which was consistent with their thermodynamic uptakes(Fig.4(d)).The productivity of highpurity (>99.9%) C2H4was calculated to be 5.61 L?kg-1.

Fig.4. Experimental breakthrough curves for(a)C2H2/C2H4(1/99)(mixed gas flow:1.0 and 2.0 ml?min-1),(b)C2H6/C2H4(10/90),(c)C2H6/C2H4(50/50),and(d)C2H2/C2H6/C2H4 (1/9/90) gas-mixtures on Zr–TCA at 100 kPa and 298 K (mixed gas flow: 1.0 ml?min-1).

Fig.5. (a)Cycling adsorption isotherms for C2H6 on Zr–TCA at 100 kPa and 298 K.(b,c)Cycling column breakthrough curves for C2H6/C2H4(10/90 and 50/50)gas-mixtures.PXRD patterns of Zr–TCA after immersing in different solvents (d) and acid/basic solutions (e) for one week. (f) PXRD patterns of Zr–TCA after heating at different temperatures.

3.5. Structural stability tests

The stability and cyclability of adsorbents are critical in practical applications, therefore Zr–TCA was systematically evaluated under various conditions. The adsorption capacities of C2H6could be well maintained in three repeated single-component adsorption cycles (Fig. 5(a)). Furthermore, no obvious deterioration in breakthrough time and working capacity was detected in three breakthrough cycles for C2H6/C2H4(10/90 and 50/50) (Fig. 5(b) and(c)).The PXRD patterns of Zr–TCA after immersing in different solvents for one week were almost unchanged, and the high crystallinity of Zr–TCA was maintained in acidic and medium basic solutions (pH < 11), indicating its high chemical stability (Fig. 5(d)and(e)).Moreover,the PXRD patterns of Zr–TCA were identical to the fresh one after treatment in the temperature range of 100–300 °C, whereas the intensity slightly decreased after heating at 400 °C for 1 h, demonstrating its potential practical applications(Fig. 5(f)).

4. Conclusions

In summary, we have successfully synthesized a thermal and chemical stable adsorbent, Zr–TCA, for one-step C2H4purification from C2hydrocarbon mixture. The appropriate pore sizes (0.6 nm× 0.7 nm) and suitable pore environment of Zr–TCA led to preferential adsorption toward C2H2and C2H6over C2H4. Zr–TCA thus exhibited high IAST selectivity for C2H2/C2H4(5.64) and C2H6/C2H4(2.72),outperforming most C2H4one-step purification adsorbents.Computational simulations revealed that the abundant oxygen atoms and low-polarity phenyl rings serve as recognition sites for the selective adsorption of C2H2and C2H6, respectively.Dynamic breakthrough tests confirmed the practical feasibility of Zr–TCA for one-step separation of high-purity C2H4(>99.9%) from a three-component C2H2/C2H6/C2H4gas-mixture.

Data Availability

Data will be made available on request.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research work was supported by the National Natural Science Foundation of China (21908090, 22008099, 22108243,and 22168023)and Natural Science Foundation of Jiangxi Province(20224ACB204003).

Supplementary Material

Supplementary material to this article can be found online at https://doi.org/10.1016/j.cjche.2023.01.001.