Methylcellulose-assisted synthesis of a compact and thin NaA zeolite membrane

Abstract

A compact and thin NaA zeolite membrane was prepared by utilizing methylcellulose (MC) as space confinement additive under microwave heating. MC played the roles of patching the defects and reducing the thickness of membrane, which improved obviously the permselectivity and, particularly, ensured an above 10⁻⁶ mol m⁻² × s⁻¹ × Pa⁻¹ of H₂ permeance.

---

Gas separation with membranes is an energy-efficient and environment friendly alternative to cryogenic and adsorptive or absorptive gas separation processes.¹ Among various types of organic and inorganic membranes, zeolite membranes have been paid much more attentions, due to their adjustable and uniform pore size at molecular level, better thermal stability, large chemical inertness, and high mechanical strength.² Of different zeolite membranes, NaA zeolite membrane possesses a smaller pore size (4.2 Å), and hence, it behaves as the most attractive candidate for the separation of H₂ from other small molecule gases, e.g., N₂, CO, CO₂ and CH₄, via molecule sieving.³ Up till now, various methods had been reported in the literature to prepare NaA zeolite membrane, such as, directly hydrothermal crystallization,⁴ adding seeds⁵ or organic cations,⁶ microwave heating⁷ and gas-phase transformation.⁸ However, the membranes obtained usually exhibited a relatively low permselectivity, due to the presence of inter-crystalline defects that led to the permeations of gases through the membranes via Knudsen diffusion rather than by molecular sieving.

In order to improve the membrane integrity, mainly two strategies were adopted by researchers. One strategy involved repeat synthesis,⁹ which was expected to repair the defects in the previously obtained membrane layer via the covering by the subsequently grown/deposited zeolite layers. Unfortunately, the results obtained were usually far beyond satisfactory, because of the issues associated with the repeat synthesis, for examples, the introduction of new defects due to the dissolution of the previously deposited NaA zeolite layer,¹⁰ the transformation of NaA zeolite into thermodynamically more stable dense zeolite phases,¹¹ and particularly, the increase in the membrane thickness that led to an obvious decrease in the permeance of H₂.^9a Another strategy involved the employment of organic coupling agents as covalent linkers,^4,6a,12 which was expected to promote the anchoring and adhesion of seed crystals with high surface coverage on support surface or to introduce the polymeric intermediate buffer layers between zeolite layers and support. While an appreciably high permselectivity could be achieved, the obtained zeolite membrane usually possessed a very low permeance (<10⁻⁷ mol m⁻² s⁻¹ Pa⁻¹). In addition, this method was also associated with a few issues, e.g., the generation of organics-containing waste water, the complexity of synthesis procedure, and the instability of organic layers. Thus, the synthesis of high-performanced NaA zeolite membrane with both large permeance and high permselectivity is still a great challenge. Basing on this consideration, oriented growth of the uniform small crystals of zeolite onto the defects appears to be a better approach. It was reported that methylcellulose (MC)¹³ played a role of space confinement additive for the synthesis of small crystal-sized zeolites, since MC hydrogel with three-dimensional adjustable pores controlled the growth of zeolite crystals. MC is environmentally friendly, and most interestingly, the gelation and dissolution of MC aqueous solution is thermoreversible.¹⁴ This unique characteristic of MC manifests itself a very promising profound in the synthesis of zeolite membranes.

In the present work, a novel method for the preparation of NaA zeolite membrane is developed, for the first time, by adding methylcellulose (MC) into the synthesis solution. The general procedure, being illustrated in Fig. 1, is as follows: a NaA-seeded support is first prepared via hot dip-coating method,¹⁵ by employing α-Al₂O₃ porous ceramic disc (22 mm × 1.5 mm) and NaA (ca. 100 nm) sol. Then, the seeded support is immersed into a MC-free synthesis solution and subjected to NaA growth under microwave heating. After that, the α-Al₂O₃ ceramic disc is thoroughly washed with cold water, resulting in the NaA zeolite membrane, namely M1. The same procedure as above is carried out to prepare the NaA zeolite membrane, namely M2, from the MC-free synthesis solution, and that, namely M3, from another synthesis solution containing MC, by using the M1 membrane as support.

Fig. 1 Schematic representation of the synthesis of NaA zeolite membrane by adding methylcellulose (MC) as space confinement additive.

Fig. 2 shows the XRD patterns for various membranes. One can see that, besides those of α-A1₂O₃, the diffraction peaks of NaA zeolite are identified for all the membranes, which indicates clearly the presence of NaA zeolite crystals on the support. The peak intensity of α-A1₂O₃ is lower for both the M2 and M3 membranes than for the M1 membrane, indicating that the NaA zeolite layer is thicker or compacter over the former two membranes, relative to the latter membrane. The peak intensity of NaA zeolite is remarkably lower and the width is broader (see Fig. S1†) for the M3 membrane than for the M1 and M2 membranes. This can be due to the fact that the average crystal size of NaA in the M3 membrane is reduced (see Fig. 3), because of the addition of MC and its space confinement effect,^13b relative to those in both the M1 and M2 membranes. Besides, no diffraction peak of MC is identified for the M3 membrane. It suggests that MC has been removed from the M3 membrane, which can be also evidenced by FT-IR characterization (see Fig. S2†).

Fig. 2 XRD patterns of (a) M1; (b) M2 and (c) M3 membranes. (The peaks for α-alumina and NaA zeolite are respectively marked by the symbols ● and ■. The standard diffraction pattern for NaA zeolite (JCPDS#39-0223) is shown by the vertical lines above x axis.)

Fig. 3 SEM micrographs of NaA zeolite membranes. (a), (c) and (e) Top view for the M1, M2 and M3 membranes; (b), (d) and (f), cross-sectional view for the M1, M2 and M3 membranes.

Fig. 3 shows the SEM micrographs for various membranes. The top view of micrographs (Fig. 3a, c and e) reveals that the NaA zeolite membranes consist of spherical crystals of NaA zeolite with undefined crystal facets, agreeing with the reports on microwave synthesis of zeolite membrane.^11,16 The aggregation of NaA zeolite crystals constructs a continuous zeolite layer over support for all the membranes. In both the M1 and M2 membranes, a lot of defects in the forms of pinholes and voids between NaA zeolite crystals are present. Compared to those in the M1 membrane, both the amount and dimension of defect are larger in the M2 membrane. Besides, the NaA zeolite in the M2 membrane possesses a larger average size (ca.1200 nm) than in the M1 membrane (ca. 700 nm), with the crystal size distribution in the former case being also broader than in the latter case (see Fig. S3†). Contrastively, no obvious defect can be identified in the M3 membrane, and the NaA zeolite crystals in this membrane are relatively small (ca. 350 nm) and their size distribution is also narrow (see Fig. S3†). The reduction in the average crystal size of NaA by the addition of MC, as manifested by SEM, is consistent with that by XRD (see Fig. S1†). From the cross-sectional view of micrographs (Fig. 3b, d and f), one can see that all the membranes possess a two-layer structure, with the top layer of NaA zeolite being tightly grown over the bottom layer of α-A1₂O₃ support. The thickness of NaA zeolite layer is identified as 4.12, 9.75 and 4.85 μm for the M1, M2 and M3 membranes, respectively. It shows that, while the thickness of NaA zeolite layer via twice synthesis enlarges remarkably without the addition of MC, it increases only slightly with the addition of MC.

The above results may be due to the fact that, for the preparation of M2 membrane, while the additional NaA zeolite crystals are deposited onto the support, relative to the M1 membrane, the original crystals of NaA zeolite in the M1 membrane are also subjected to further growth. This leads to a larger thickness of M2 membrane than the M1 membrane. In addition, a proportion of NaA zeolite crystals can be dissolved by the caustic synthesis solution, leading to the appearance of more defects with larger dimension in the M2 membrane, relative to the M1 membrane. For the preparation of M3 membrane, an appropriate amount of Si/Al active ingredients have been entrapped into the pores of MC hydrogel at room temperature, and this MC hydrogel can be adsorbed not only on the surface but also in the defect sites of the M1 membrane. Upon microwave heating, the adsorbed MC hydrogel is gelled, generating a protective layer to avoid the contact of Si/Al active ingredients in the bulk synthesis solution with the surface of M1 membrane. Therefore, on the one hand, NaA zeolite crystals can be gradually generated from the Si/Al active ingredients in the pores of MC gel, however, their overgrowth is inhibited due to the space confinement effect. This enables that the M1 membrane as support is covered by a layer of smaller-sized NaA zeolite crystals, and thus, the resultant M3 membrane possesses a relatively small thickness, which is only very slightly larger than that for the M1 membrane. On the other hand, no more the Si/Al active ingredients can be available for the crystallization of NaA zeolite between the MC gel and M1 membrane. This enables that the M3 membrane is free from the destruction by the dissolution of synthesis solution, and the defects in the M1 membrane is patched due to the crystallization of NaA zeolite from the Si/Al active ingredients entrapped in the MC gel that have been adsorbed in the defect sites. It should be addressed that the MC gel becomes dissolvable at room temperature, and therefore, it is completely removed from the M3 membrane via sufficient washing by cold water, without affecting the integrity of membrane. As a result, the M3 membrane possesses much less defects and also a smaller thickness, relative to the M2 membrane.

The performance of membrane is evaluated by the experiment of gas permeation. Fig. 4 shows the permeances of various gases and ideal permselectivities of H₂ over various other gases for both the M1 and M3 membranes. One can see that the permeance of various gases for the two membranes both displays an order H₂ > N₂ > CO₂ > CO > CH₄ > C₃H₈, being approximately inverse with that for the kinetic diameter of these gases. The unexpected smaller permeance of CO₂, relative to N₂, can be due to the fact that the former gas is more easily condensed and adsorbed in the pores of zeolite than the latter one.¹⁷ Both the permeances of H₂ through the M1 and M3 membranes are appreciably high, being in a level of 10⁻⁵ to 10⁻⁶ mol m⁻² s⁻¹ Pa⁻¹, while the permeance of H₂ through the M1 membrane (10.48 × 10⁻⁶ mol m⁻² s⁻¹ Pa⁻¹) is ca. three times more than that through the M3 membrane (3.65 × 10⁻⁶ mol m⁻² s⁻¹ Pa⁻¹). The ideal permselectivities of H₂ over various gases for both the M1 and M3 membranes, shown in the inset of Fig. 4, exhibits an order α(H₂/N₂) < α(H₂/CO₂) < α(H₂/CO) < α(H₂/CH₄) < α(H₂/C₃H₈), which closes to the order of the kinetic diameters of these gases. The abnormally larger permselectivity of H₂ over CO₂, relative to that over N₂, is due to the smaller permeation of CO₂ than N₂ as mentioned above. In addition, the permselectivities of H₂ over other gases through the M3 membrane are all obviously larger than those through the M1 membrane and also those through Knudsen diffusion. For examples, the permselectivities of H₂ over N₂ and C₃H₈, i.e., α(H₂/N₂) and α(H₂/C₃H₈), through the M3 membrane are 5.62 and 15.87 and those through the M1 membrane only 3.52 and 4.68, while the corresponding Knudsen diffusion selectivities are 3.74 and 4.69, respectively. Considering that the M3 membrane (4.85 μm) possesses the similar thickness as the M1 membrane (4.12 μm), the fewer defects in the former membrane contribute predominantly its relatively lower permeance but larger ideal permselectivity, relative to the latter membrane. Fig. 5 shows the real permselectivities of H₂ over various other gases from equimolar mixtures for both the M1 and M3 membranes. One can see that the real permselectivity shown in Fig. 5 exhibits the same regularity with the corresponding ideal permselectivity shown in Fig. 4. Compared to the ideal permselectivity, the real permselectivity is decreased obviously for the M1 membrane, however, only slightly for the M3 membrane. In the case of M3 membrane, the real permselectivities of H₂ over various other gases are determined as 5.87, 5.03, 6.96, 7.59, 14.56 for α(H₂/CO₂), α(H₂/N₂), α(H₂/CO), α(H₂/CH₄) and α(H₂/C₃H₈), respectively, being higher remarkably than those of the corresponding Knudsen diffusion. It is indicated that the gas permeation through the M3 membrane can be contributed by zeolitic fluxes, which is referred as surface diffusion, intra-crystalline flux or zeolitic diffusion, while the gas permeation through the M1 membrane is only related to nonzeolitic fluxes, which are governed by Knudsen and viscous flux mechanisms.¹⁸ The above results confirm further that the M3 membrane possesses fewer defects than the M1 membrane.

Fig. 4 Permeation of various small molecule gases through the M1 and M3 membranes at 298 K and under 0.1 MPa pressure difference. The inset shows the ideal permselectivity of H₂ over other gases.

Fig. 5 Real permselectivity of H₂ over other gases from equimolar mixtures on M1 and M3 membranes at 298 K and under 0.1 MPa pressure difference.

In summary, the M3 membrane synthesized basing on the M1 membrane as support possesses the obviously improved permselectivities of H₂ over other gases, being larger than the corresponding Knudsen diffusion, relative to the M1 membrane, while the permeance of H₂ can be as high as above 10⁻⁶ mol m⁻² s⁻¹ Pa⁻¹. The much better performance of permeation for the M3 membrane than for the M1 membrane can be attributed to the introduction of MC in the synthesis solution, which enables not only the defects in the substrate membrane to be patched up but also the thickening of the synthesized zeolite membrane to be retarded. Therefore, the present work provides a new way to the synthesis of high performance zeolite membranes for gas separation.

Acknowledgements

This work was financially supports from the Project #21376068 for National Natural Science Foundation (NSFC), Program for Lotus Scholar in Hunan Province.

Notes and references

1. P. Bernardo, E. Drioli and G. Golemme, Ind. Eng. Chem. Res., 2009, 48, 4638.
2. (a) N. W. Ockwig and T. M. Nenoff, Chem. Rev., 2007, 107, 4078; (b) N. Rangnekar, N. Mittal, B. Elyassi, J. Caro and M. Tsapatsis, Chem. Soc. Rev., 2015, 44, 7128.
3. (a) A. S. Huang and J. Caro, J. Mater. Chem., 2011, 21, 11424; (b) A. Huang, Q. Liu, N. Wang, X. Tong, B. Huang, M. Wang and J. Caro, J. Membr. Sci., 2013, 437, 57; (c) C. Yuan, Q. Liu, H. Chen and A. Huang, RSC Adv., 2014, 4, 41982.
4. H. Li, J. Xu, J. Wang, J. Yang, K. Bai, J. Lu, Y. Zhang and D. Yin, J. Membr. Sci., 2014, 471, 84.
5. (a) J. Shao, Z. Y. Zhan, J. G. Li, Z. B. Wang, K. Li and Y. S. Yan, J. Membr. Sci., 2014, 451, 10; (b) M. Moheb Shahrestani, A. Moheb and M. Ghiaci, Vacuum, 2013, 92, 70.
6. (a) K. P. Dey, D. Kundu, M. Chatterjee, M. K. Naskar and W. Suchanek, J. Am. Ceram. Soc., 2013, 96, 68; (b) A. S. Huang and J. Caro, Chem. Mater., 2010, 22, 4353.
7. D. Kunnakorn, T. Rirksomboon, P. Aungkavattana, N. Kuanchertchoo, D. Atong, S. Kulprathipanja and S. Wongkasemjit, Desalination, 2011, 269, 78.
8. Z. L. Cheng, Z. S. Chao, H. Q. Lin and H. L. Wan, Chin. J. Chem., 2003, 21, 1430.
9. (a) X. C. Xu, W. S. Yang, J. Liu, L. W. Lin, N. Stroh and H. Brunner, J. Membr. Sci., 2004, 229, 81; (b) M. K. Naskar, D. Kundu and M. Chatterjee, J. Sol-Gel Sci. Technol., 2012, 64, 117.
10. J. Caro, D. Albrecht and M. Noack, Sep. Purif. Technol., 2009, 66, 143.
11. X. Xu, W. Yang, J. Liu and L. Lin, Sep. Purif. Technol., 2001, 25, 241.
12. C. Yuan, Q. Liu, H. Chen and A. Huang, RSC Adv., 2014, 4, 41982.
13. (a) X. Liu, X. Zhang, Z. Chen and X. Tan, Ceram. Int., 2013, 39, 5453; (b) H. Wang, B. A. Holmberg and Y. Yan, J. Am. Chem. Soc., 2003, 125, 9928.
14. L. Li, P. M. Thangamathesvaran, C. Y. Yue, K. C. Tam, X. Hu and Y. C. Lam, Langmuir, 2001, 17, 8062.
15. H. Li, J. Wang, J. Xu, X. Meng, B. Xu, J. Yang, S. Li, J. Lu, Y. Zhang and X. He, J. Membr. Sci., 2013, 444, 513.
16. (a) Y. S. Li, J. Liu, H. L. Chen, W. S. Yang and L. W. Lin, Chin. J. Catal., 2006, 27, 544; (b) N. Hu, Y. Zheng, Z. Yang, R. Zhou and X. Chen, RSC Adv., 2015, 5, 87556.
17. Y. S. Lin, I. Kumakiri, B. N. Nair and H. Alsyouri, Sep. Purif. Rev., 2007, 31, 229.
18. S. A. Hosseinzadeh Hejazi, A. M. Avila, T. M. Kuznicki, A. Weizhu and S. M. Kuznicki, Ind. Eng. Chem. Res., 2011, 50, 12717.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11716c

---