Science in China Series B: Chemistry © 2007

Science in China Press Springer-Verlag

Preparation and pervaporation performance of high-quality silicalite-1 membranes CHEN HongLiang, LI YanShuo, LIU Jie & YANG WeiShen† State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China

High-performance silicalite-1 membranes were successfully synthesized on novel porous silica tubes by two-step in-situ hydrothermal synthesis. The flux and separation factor towards ethanol/water mixture at 60℃ were 0.56 kg/(m2·h) and 84, respectively. The as-synthesized silicalite-1 membranes were characterized by scanning electron microscopy (SEM). The influence of different synthesis conditions on the separation performance of the silicalite-1 membranes was investigated. It was found that the average flux of silicalite-1 membranes was improved by about 26% after filling the silica tubes with mixed solution containing glycerol and water. After calcinating at 400℃ for 5 h repeatedly, membrane synthesized on silica tube still showed high pervaporation performance towards ethanol/water mixture even at a calcination rate of 4℃/min, which suggested that silica support was more suitable for preparing high-performance silicalite-1 membranes. silica support, silicalite-1 membrane, mixed solution filling, pervaporation

In recent years, great interest has been focused on the － synthesis of zeolite membranes on various supports[1 6] because of the potential applications of zeolite membranes in a wide range of industrial processes, and significant progress in preparation and characterization of － zeolite membranes has been seen[7 10]. MFI membranes (silicalite-1 and ZSM-5) have been the subject of the majority of these studies, mainly because of their suitability for the separation of commercially important species, such as p-xylene from its isomers[11,12]. For silicalite-1 membrane, which is more hydrophobic than that of ZSM-5 membrane, another important application is to extract organics from low concentration aqueous solu－ tions[2 6], such as extracting ethanol from fermentation broth[2]. For the separation of alcohol/water mixtures, only alcohol and a small quantity of water can pass the silicalite-1 membrane, and energy consuming is lower than that of conventional separation including vacuum distillation, solution extraction, and gas stripping by CO2, which is in favor of solving the extraction problem of ethanol from fermentation broth. www.scichina.com

www.springerlink.com

In order to prepare high-performance silicalite-1 membranes, the influence of different supports is important. The hydrophobicity of silicalite-1 membranes will decrease as a result of the dissolution of alumina supports during hydrothermal synthesis, so the alumina supports are unsuitable for preparing silicalite-1 membranes. Geus[13] found that improving the separation performance of zeolite membranes could be achieved by using the porous intermediate layers like silica between the ceramic or metal support and the zeolite layer, and the porous intermediate layers were recommended to be used for suppressing the influence of supports. In 2003, Lai et al.[14] reported that the presence of the silica layer eliminated the formation of stress-induced cracks during calcinations, and high-performance zeolite membranes were prepared. In the current work, aiming at eliminating the influence of alumina supports, highReceived October 17, 2005; accepted May 18, 2006 doi: 10.1007/s11426-007-0011-2 † Corresponding author (email: [email protected]) Supported by the National Advanced Materials Committee of China (Grant No. 2003AA328010) and the Ministry of Science and Technology of China (Grant No. G2003CB615802)

Sci China Ser B-Chem | February 2007 | vol. 50 | no. 1 |70-74

1 Experimental

ture was evaluated by pervaporation apparatus. The permeate and feed concentrations of ethanol and water were measured by off-line GC (HP 5890). The total flux was calculated by weighting the condensed permeate. The separation factor was determined as αA/W= (YA/YW)/(XA/XW), where XA, XW, YA, and YW denote the mass fractions of components A (ethanol) and W (water) on the feed and permeate sides, respectively.

1.1 Synthesis of silicalite-1 membranes

1.4 Membrane thermal stability

The silica tubes were prepared by casting method. These tubes were 11 mm in OD, 7 mm in ID, 90 mm in length, 0.3 μm in pore diameter, and of 45% porosity. Synthesis solution was prepared by mixing sodium hydroxide, tetrapropylammonium bromide (TPABr), silica sol, and deionized water at room temperature. The molar composition of the synthesis solution was TPABr:0.25Na2O:10SiO2:800H2O. Two-step in-situ hydrothermal synthesis was used to synthesize silicalite-1 membranes, and before each hydrothermal synthesis, the silica tubes were filled with the mixed solution containing water and glycerol to decrease the penetration of synthesis solution into support pores and the contrasted tubes were synhtesized without filling, then sealed with two teflon caps at both ends and placed vertically in teflon-lined autoclaves. After the synthesis solution was added, the autoclaves were put in an air oven and aged at 60℃ for 12 h (without aging for some tubes before the secondary crystallization), and then the oven temperature was increased to 180℃ and crystallization was carried out for a certain time. After crystallization, the autoclaves were taken out and cooled down to room temperature, and the as-synthesized membranes were washed with deionized water, dried at 80℃, then the silicalite-1 membranes were calcined at 500℃ for 12 h to remove the templates.

The membrane S-1 was calcined repeatedly in the air after removing the templates. Each calcination was carried out at 673 K for 5 h at a calcination rate of 0.2－5 ℃/min, and the thermal stability was evaluated by separating the ethanol/water mixture.

performance silicalite-1 membranes were successfully prepared by using silica tubes instead of alumina supports, and the influence, such as two-step synthesis conditions, filling and without filling the silica tubes, on the preparation of silicalite-1 membranes was also investigated in detail.

1.2 Membrane characterization The top view and the cross section of silicalite-1 membranes were characterized by scanning electron microscopy (SEM, JEM-1200), and SEM was also used to estimate the thickness of the silicalite-1 membranes (cross section). The membrane surface was sprayed with gold before taking pictures. 1.3 Pervaporation measurements The calcined membranes were sealed in a module, and the separation performance towards ethanol/water mix-

2 Results and discussion 2.1 Effects of filling solution Figure 1 shows the SEM images of silicalite-1 membranes synthesized on silica tubes. It can be seen that the zeolite crystals were well intergrown and defect-free membranes were formed, which was verified by pervaporation date shown in Table 1. Comparing Figure 1(a) with (c), it can be seen that the crystal size of membrane with solution filling (Figure 1(a)) was smaller than that of membrane without solution filling (Figure 1(c)). The possible reason was that the glycerol seeping from tubes may play the role of templates partly and enhance the number of nuclei on the support surface. Seen from cross section, two membranes showed similar membrane thickness, about 40 μm. Table 1 shows the influence of mixed filling solution on the separation performance of silicalite-1 membranes. The average flux and separation factor of membranes filled with mixed solution (S-1 and S-2) were 0.54 kg/(m2·h) and 74, respectively. While for the membranes without solution filling (S-3 and S-4), the average flux and separation factor were 0.43 kg/(m2·h) and 77, respectively. The average flux was improved by about 26%. The reason was that diffusion resistance could be decreased by using solution filling method, and this could be verified by experimental observation. After each hydrothermal synthesis, the tubes without solution filling were fully filled with synthesis solution and the zeolite layer was formed on the bottom of tubes, while no zeolite layer was formed for the tubes with solution filling. These results showed that the formation of zeo-

CHEN HongLiang et al. Sci China Ser B-Chem | February 2007 | vol. 50 | no. 1 | 70-74

71

Figure 1 SEM images of silicalite-1 membranes synthesized on silica supports. (a) Top view and (b) cross section of membrane with solution filling; (c) top view and (d) cross section of membrane without solution filling. Table 1 Influence of mixed filling solution on the separation performance of silicalite-1 membranesa) Membrane Aging temp(℃) Aging time (h) Synthesis time (h) Synthesis temp(℃) Filling Flux (kg/(m2·h)) S-1 60+60 12+12 22+10 180 yes 0.56 S-2 60+60 12+12 22+10 180 yes 0.52 S-3 60+60 12+12 22+10 180 no 0.37 S-4 60+60 12+12 22+10 180 no 0.48 no no 24+48 185 no 0.16 Ref. [3]b) Ref. [4]c) no no 170 no 0.40 24−144 a) Feed ethanol concentration: 3%, feed temperature: 60℃; b) feed ethanol concentration: 5%; c) feed ethanol concentration: 5%.

lite crystals in the supports pores would be restrained by using filling solution, which decreased the diffusion resistance of silicalite-1 membranes. Comparing the date shown in Table 1 with those reported in the literature[3,4], the silicalite-1 membranes synthesized on silica tubes showed higher separation performance, which suggested that silica supports were more suitable for preparing high-performance silicalite-1 membranes. 2.2 Influence of aging and crystallization time The influence of crystallization time on the separation performance of silicalite-1 membranes is shown in Table 2. It was found that the separation factors of all membranes were lower than 60, which showed that a shorter first crystallization time was unsuitable for the preparation of silicalite-1 membranes with high selectivity. The 72

Separation factor 84 65 84 70 31 47

total crystallization time and synthesis solution of membrane S-8 were the same as those of membranes S-1 and S-2 shown in Table 1, but the separation factor of the former was much lower than that of the latter. The reason was that larger and more defects were easily formed for a shorter first crystallization time, which could be verified by the seeping rate of water from tubes. Furthermore, the large defects could not be eliminated effectively even after a longer secondary crystallization time. Therefore, high-performance zeolite membranes could only be prepared with a suitable crystallization time. Figure 2 shows the SEM images of silicalite-1 membrane without aging before the secondary crystallization. Compared Figure 1(a) and (b) with Figure 2(a) and (b), it was found that the size of zeolite crystals increased

CHEN HongLiang et al. Sci China Ser B-Chem | February 2007 | vol. 50 | no. 1 | 70-74

Figure 2 SEM images of silicalite-1 membranes synthesized on silica supports (S-9). (a) Top view; (b) cross section. Table 2 Influence of crystallization time on the separation performance of silicalite-1 membranesa) Membrane Aging time (h) Synthesis time (h) Synthesis temp(℃) Filling Aging temp(℃) S-5 60+60 12+12 10+10 180 yes S-6 60+60 12+12 10+14 180 yes S-7 60+60 12+12 10+18 180 yes S-8 60+60 12+12 10+22 180 yes a) All the results were the average of two tubes; feed ethanol concentration: 3%; feed temperature: 60℃.

obviously for membrane (Figure 2(a)) without aging before the secondary crystallization, and the thickness of the as-synthesized membrane (Figure 2(b)) increased by about 10 μm. The reason was that aging was useful for the formation of nuclei on the tube surface and in the solution, which could consume the nutrition at the same time and result in the decrease of membrane growth rate. If no aging was carried out before the secondary crystallization, the zeolite crystals would prefer to growing on the membrane surface during the secondary crystallization because the small crystals, which were formed on the membrane surface during the first crystallization, played the role of nuclei, which resulted in the formation of silicalite-1 membrane with thickness of 50 μm. Table 3 shows the influence of aging and crystallization time on the separation performance of silicalite-1 membranes. The silicalite-1 membranes with high separation factor were obtained without aging before the secondary crystallization, but the flux was low. No improvement of separation factor was achieved and the

Flux (kg/(m2·h)) 0.65 0.59 0.63 0.47

Separation factor 51 56 48 45

flux decreased by only prolonging the secondary crystallization time. The reason was that the as-synthesized membrane showed high diffusion resistance, and the resistance came from the following two aspects: one was that longer pathway was needed for ethanol molecules through thicker membrane, which resulted in the enhancement of diffusion resistance; the other resistance came from the formation of zeolite crystals in the support pores. Therefore, in order to improve the separation performance of zeolite membranes, the membrane thickness should be decreased while keeping defect-free, and the formation of zeolite crystals in the support pores should also be suppressed or avoided. 2.3 Thermal stability of silicalite-1 membrane The influence of calcination rate on the separation performance of silicalite-1 membrane (S-1) is shown in Figure 3. It was found that the silicalite-1 membrane synthesized on silica tubes showed high thermal stability, and the membrane still kept high separation performance at a calcination rate from 0.2 to 4℃/min, but the flux

Table 3 Influence of aging and crystallization time on the separation performance of silicalite-1 membranesa) Membrane Aging time (h) Synthesis time (h) Synthesis temp(℃) Filling Aging temp(℃) S-9 60+no 12+0 22+10 180 yes S-10 60+no 12+0 22+12 180 yes S-11 60+no 12+0 22+14 180 yes S-12 60+no 12+0 22+16 180 yes a) All the results were the average of two tubes; feed ethanol concentration: 3%; feed temperature: 60℃.

Flux (kg/(m2·h)) 0.43 0.39 0.37 0.44

CHEN HongLiang et al. Sci China Ser B-Chem | February 2007 | vol. 50 | no. 1 | 70-74

Separation factor 79 66 65 78

73

supports at a heating rate of above 0.25℃/min, so a lower calcination rate was performed. Increasing the calcination rate to 5℃/min, both the flux and separation factor decreased, and the reason was not clear. One possibility was that it may be the formation of small cracks due to the high calcination rate, which decreased the separation performance of the silicalite-1 membrane.

3 Conclusion

Figure 3 Influence of calcinations rate on the separation performance of silicalite-1 membrane (S-1).

changed slightly, which suggested that silica support was more suitable to prepare silicalite-1 membranes with thermal stability. In order to avoid the formation of cracks during calcination, the synthesized membranes were generally calcined at a low calcination rate because of different thermal expansion coefficients between supports and zeolite layers. For example, work by Holmes et al.[15] showed that the cracks would be formed for membranes synthesized on stainless steel 1

Noack M, Kolsch P, Schafer R, et al. Preparation of MFI membranes

A series of experiments showed that the synthesis conditions, such as mixed filling solution, aging, and crystallization time, exhibited important influence on the preparation of silicalite-1 membranes. Improving the separation performance of silicalite-1 membranes was achieved by filling supports with mixed solution and aging before crystallization. At the same time, the selection of two crystallization times was also one of the important factors to prepare high-performance silicalite-1 membranes. After repeated calcinations, silicalite-1 membrane synthesized on silica support showed thermal stability, which suggested that silica supports were more suitable to prepare high-performance silicalite-1 membranes. 9

of enlarged area with high reproducibility. Micropor Mesopor Mat,

development and perspective. Micropor Mesopor Mat, 2000, 38(1):

2002, 49(1-3): 25－37 2

3－24

Nomura M, Bin T, Nakao S. Selective ethanol extraction from fer-

10

mentation broth using a silicalite-1 membrane. Sep Purif Technol,

245(1-2): 1－33

Bowen T C, Kalipcilar H, Falconer J L, et al. Pervaporation of or-

11

ganics/water mixtures through B-ZSM-5 zeolite membranes on Matsusuda H, Yanagishita H, Negishi H. Improvement of ethanol

Chem Soc, 2004, 126(15): 4776－4777 12

selectivity of silicalite membrane in pervaporation by silicone rubber

6

Lin X, Kita H, Okamoto K. Silicalite membrane preparation, charac-

Matsufuji T, Nishiyama N, Matsukata M, et al. Separation of butane and xylene isomers with MFI-type zeolite membrane synthesized by a

coating. J Membr Sci, 2002, 210(2): 433－437 5

Yuan W, Lin Y S, Yang W S. Molecular sieving MFI-type zeolite membranes for pervaporation separation of xylene isomers. J Am

monolith supports. J Membr Sci, 2003, 215(1-2): 235－247 4

Bowen T C, Noble R D, Falconer J L. Fundamentals and applications of pervaporation through zeolite membranes. J Membr Sci, 2004,

2002, 27(1): 59－66 3

Caro J, Noack M, Kolsch P, et al. Zeolite membranes-state of their

vapor-phase transport method. J Membr Sci, 2000, 178(1-2): 25－34 13

Geus E R, Bekkum H V. Calcination of large MFI-type single crystals.

terization, and separation performance. Ind Eng Chem Res, 2001,

Part 2: Crack formation and thermomechanical properties in view of

40(19): 4069－4078

the preparation of zeolite membranes. Zeolites, 1995, 15(4):

Lin X, Kita H, Okamoto K. A novel method for the synthesis of high performance silicalite membranes. Chem Commun, 2000, (19):

333－341 14

Lai Z, Bonilla G, Diaz I, et al. Microstructural optimization of a zeo-

1889－1890

lite membrane for organic vapor separation. Science, 2003, 300(18):

7

Bein T. Synthesis and applications of molecular sieve layers and

456－460

8

Tavolaro A, Drioli E. Zeolite membranes. Adv Mater, 1999, 11(12):

growth of silicalite membranes on stainless steel supports. Chem

975－996

Mater, 1999, 11(11): 3329－3332

membranes. Chem Mater, 1996, 8(8): 1636－1653

74

15

Holmes S M, Markert C, Plaisted R J, et al. A novel method for the

CHEN HongLiang et al. Sci China Ser B-Chem | February 2007 | vol. 50 | no. 1 | 70-74