-

journal of MEMBRANE

SCIENCE

Journal of Membrane Science 95 ( 1994) 22 l-228

Separation of ethanol/water mixture by silicalite membrane on pervaporation

Tsuneji Sanoap*, Hiroshi Yanagishitab, Yoshimichi Kiyozumib, Fujio Mizukamib, Kenji Harayab

‘Japan Advanced Institute of Science and Technology, Tatsunokuchi, Ishikawa 923-12, Japan bNational Institute ofMaterials and Chemical Research, Tsukuba, Ibaraki 305, Japan

Received 1 October 1993; accepted 17 May 1994

Abstract

Pure silicalite membranes were prepared on porous supports of sintered stainless steel or alumina discs. The silicalite layer was characterized by X-ray diffraction, SEM and mercury porosimetry. Individual crystals were intergrown in three dimensions into the polycrystalline phase. The membranes were not disintegrated by thermal treatment in vacua or calcination for removing the organic amine occluded in the channels of silicalite, indicating high thermomechanical stability of the membrane. The liquid separation potential of the membrane was investi- gated by pervaporation of an aqueous ethanol solution. The high ethanol permselectivity with a separation factor a( EtOH/H*O) of more than 60 was achieved for a 5 ~01% aqueous ethanol solution at 3O”C, indicating no cracks and pores between the silicalite grains within the membrane. From adsorption experiments of ethanol and water on silicalite, it was found that the high permselectivity is attributable to the selective sorption of ethanol into the silicalite membrane.

Keywords: Separation; Ethanol; Silicalite membrane; Pervaporation

1. Introduction

Great interest has been focused on prepara- tion of zeolite membranes from the standpoint of application to gas or liquid separation. Most of the zeolite membranes have been made by embedding the zeolite crystals in polymer or ce- ramic matrixes, or by in situ crystallization on porous substrates. Recently, several researchers succeeded in preparing pure zeolite membranes of silicalite, ZSM-5,ZSM-35, gmelinite, NaA and SAP0 [l-9]. Although zeolite membranes are

* Corresponding author.

polycrystalline films and very fragile, the gas separation potential of zeolite membranes is in- dicated in Refs. [ 10-l 2 1. However, only few studies have been dedicated to the separation of liquid mixtures.

Te Hennepe et al. reported that the membrane performance of the silicon rubber membrane, which is effective for separation of ethanol/water mixtures, is considerably improved by addition of silicalite crystals to the silicon rubber matrix [ 13 1. At 70 wt% silicalite content a high separa- tion factor (Y ( EtOH/H20) of N 20 was achieved. This result suggests that if a pure silicalite mem-

0376-7388 / 94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSD10376-7388(94)00120-N

222 T. Sane et al. /Journal of Membrane Science 95 (1994) 221-228

brane were synthesized, a higher separation fac- tor would be obtained.

From these standpoints, we have investigated the liquid separation potential of the zeolite membrane on pervaporation. It was already re- ported that the silicalite membrane is very effec- tive for the separation of ethanol/water mix- tures [ 141. In this paper, the pervaporation performance of silicalite membrane for separa- tion of ethanol/water mixtures was studied in detail.

2. Experimental

The hydrothermal synthesis of silicalite mem- branes was performed as follows. Colloidal silica (Cataloid SI-30 from Shokubai Kasei Co.; 30.4 wt% SiOZ, 0.38 wt% NazO, 69.22 wt% water) was added to a stirred solution of tetrapropylam- monium bromide (TPABr ) and sodium hydrox- ide, to give a hydrogel with a composition of 0.1: 0.05 : 1: 80 TPABr-NazO-Si02-H20. Then the hydrogel was transferred to a 300 ml stainless steel autoclave. A porous support of sintered stainless steel or alumina disc (5 cm diameter) with an average pore diameter of 0.5-2 pm was placed on the bottom of the autoclave. The au- toclave was placed in an air-heated oven at 170’ C

Pirani gauge A

L- 3.5

Stirrer

To

for 48 h. After the completion of the crystalliza- tion under autogenous pressure without stirring, the autoclave was cooled down, and the support was recovered. The silicalite membrane on the support was washed with deionized water and dried at 100’ C. Then the silicalite membrane was heated at 300-380’ C under vacuum, or calcined at 500°C for 20 h in order to decompose the or- ganic amine occluded in the zeolite framework. The sihcalite membrane was not disintegrated by these processes.

The characterization of the membrane was achieved by X-ray diffraction (Mac Sci. MXP 18 ) with the incident beam at a glancing angle of 1 .OO" . The surface and the cross section of membrane were characterized by scanning electron microscopy (SEM, Hitachi H-800) with an energy-dispersive X-ray analysis system (EDX). The pore size distribution of the mem- brane was measured by a mercury porosimeter (Micromeritics AutoPore 9200).

The pervaporation measurements using an aqueous ethanol solution as a feed were per- formed on a standard pervaporation apparatus as shown in Fig. 1. Liquid nitrogen was used as a cooling agent for the cold trap. The composi- tions of the feed and the permeate were deter- mined by gas or liquid chromatography. Flux and

Thermometer

Stirrer Silicdite membrane

Pervaporation cell

Fig. 1. Schematic diagram of pervaporation apparatus.

T. Sano et al. /Journal of Membrane Science 95 (1994) 221-228 223

separation factor cy ( EtOH/H20) were calcu- lated from:

Flux ( kg/m2 h)

(weight of permeate, kg)

= (membrane area, m2) x (permeation time, h)

and

Separation factor (Y ( EtOH/H2 0)

= [G,O”/G,O IPermeate t GtOHICH20lFeed

where C,,,, and CH20 are the volume fractions of ethanol and water, respectively.

To evaluate the adsorption performance of the silicalite membrane, vapor-sorption experi- ments of water and ethanol on silicalite were car- ried out with a Belsorp 18 adsorption apparatus.

3. Results and discussion

3.1. Characterization of silicalite membrane

Fig. 2 shows X-ray diffraction diagrams of sil- icalite membranes supported on porous stainless steel and alumina discs. The reflection peaks corresponding to silicalite were observed in the X-ray diffraction diagrams. The peaks derived from the porous supports were not observed. Fig.

(4 silicalite/sintered stainless steel support

sintered stainless steel support

5 10 20 30 40 50

28 (degrees)

3 shows the scanning electron micrographs of the surface and the cross section of the membranes. The surface was formed by an aggregate of crys- tals of 1 O-25 pm. The growth on the porous sup- port led to a randomly grown crystalline layer. The average thickness of the silicalite layer (400- 500 pm) was confirmed by Si line analysis with EDX. These results indicate that the surface of the support was covered with densely packed sil- icalite crystals.

If there are cracks or pores between the silical- ite grains within the membrane, separation per- formance cannot be expected. Therefore, the pore size distribution of the membrane was measured by mercury porosimetry. Fig. 4 shows the pore size distribution of the pure silicalite membrane detached from the stainless steel support by a mechanical method. One peak was observed at N 500 nm. To clarify where the peak originated from, a cross section of the membrane was mea- sured by SEM at higher magnification (Fig. 5 ). The support side of the membrane consists of small crystals 2-4 pm in size. Pores or interstices between the silicalite grains were visible. The crystal sizes in the middle part of the membrane were larger than those on the support side. Pores or interstices were also observed. On the other hand, the solution side (the top layer) of the membrane was composed of a layer of highly in- tergrown zeolite crystals. Although the shape of

I (B) silicalitekalumina support

20 30 40

28 (degrees)

Fig. 2. X-ray diffraction diagrams of silicalite membranes on porous supports.

224 T. Sano et al. /Journal ofMembrane Science 95 (1994) 221-228

(A) silicalitekintered stainless steel support

(B) silicalite/alumina support

Surface Cross section Fig. 3. SEM images and Si line analysis for surface and cross section of silicalite membranes on porous supports.

T. Sano et al. /Journal ofMembrane Science 95 (1994) 221-228 225

0.0 ----+----- ,./f L.. _ d-l. ___.__. .-

10 102 103 104 105

RP (nm)

Fig. 4. Pore size distribution of silicalite membrane detached from stainless steel support determined by mercury porosimetry.

the crystal is not clear due to intergrowth, the crystal size appears to become larger. No pores originated from silicalite grains were observed. This indicates that the peak at 500 nm in the pore

size distribution as shown in Fig. 4 is attribut- able to the pores between the silicalite grains in the middle and the support side of the mem- brane. These characteristics of the silicalite membrane prepared are consistent with those re- ported in the literature [ lo- 12 1.

3.2. Pervaporation performance

As mentioned in the experimental section, the organic amine (TPABr) is used in the silicalite synthesis as a template. The organic amine re- mains in the channels of the silicalite crystals. In order to use the membrane as a pervaporation membrane, the amine must be removed from the channels by certain procedures. As the silicalite membrane experiences irregular stresses that arise from a difference in the thermal expansion between the support and the silicalite crystals and from removal of volatile materials from the zeo- lite framework, cracks are easily formed within the membrane during the treatment process.

Solution side

Fig. 5. SEM images of cross section of silicalite membrane detached from stainiess steel support at high magnification.

226 T. Sano et al. /Journal of Membrane Science 95 (1994) 221-228

Therefore, the influence of pretreatment condi- tions of the membrane on the pervaporation per- formance was studied. The pervaporation mea- surements were conducted at 60°C using an aqueous ethanol solution of 5 ~01% as the feed. The results obtained are summarized in Table 1. The silicalite membrane after air-drying at 100 ’ C (containing the template still) showed a very low flux combined with a separation factor smaller than 1, indicating no cracks and pores between the silicalite grains within the membrane before the pretreatment. However, in the case of the membrane after thermal treatment under a vac- uum, the flux and the separation factor increased with an increase in treatment temperature and time. The membrane calcined at 500°C to re- move the template completely showed a high flux combined with a high separation factor of N 60. This separation factor is the most favorable one among organic and inorganic membranes which have been published so far. These results suggest that the membrane changes from a water-selec- tive membrane to an ethanol-selective one by decreasing the amount of template occluded in the zeolite framework, and that the separation of ethanol/water takes place by transport through the zeolite channels.

concentrations of the permeates for the alumina supports are slightly lower than those for the stainless steel ones. Taking into account that the hydrophobicity of the ZSM-5 type zeolite de- creases when the content of framework alumi- num increases [ 15,161, this suggests that the top layer of the membrane on the alumina support is not composed of silicalite crystals but composed of ZSM-5 ones. The difference between ZSM-5 and silicalite is the content of framework alumi- num, which is nil for the latter. Although the syn- thesis mixture does not contain an aluminum source, it is reasonable to consider that a part of the surface of the alumina support is dissolved in the high-pH solution. Actually, the surface Si0JA120, ratio of the membrane on the alu- mina support measured by EDX was 392.

Fig. 6 shows the pervaporation performance of several silicalite membranes, which were pre- pared by the same method. It is noted that each membrane shows a slight different performance. We have considered that the difference in mem- brane performance is attributable to a slight dif- ference in membrane thickness or to very small cracks formed during the thermal treatment of the membrane. It is also noted that the ethanol

Next, the influence of the feed temperature on the pervaporation performance was studied us- ing the silicalite membrane supported on a sin- tered stainless steel disc (Fig. 7). The flux in- creased linearly with an increase in the feed temperature due to the higher vapor pressure of the components of the feed. The separation fac- tor a(EtOH/H,O) decreased slightly with the temperature. A few introductory experiments were carried out to investigate the influence of the feed ethanol concentration on the pervapor- ation performance. Fig. 8 illustrates the relation- ship between flux and separation factor as func- tion of the feed ethanol concentration. The flux decreased rapidly by adding ethanol to water and then increased slightly with the ethanol concen- tration. On the other hand, the separation factor a!(EtOH/H20) reached a maximum value at a feed ethanol concentration of w 3 ~01% and then

Table 1 Influence of pretreatment conditions on pervaporation performance

Treatment condition

Temp. (“C) Time (h)

FlUX

(kg/m’ h) Separation factor cu(EtOH/H,O)

air-drying 100 12 in vacua 300 6 in vacua 380 6 calcination 500 20

Feed temperature: 60°C; feed ethanol concentration: 5 ~01%.

0.00303 0.38 0.00840 0.58 0.0394 7.8 0.760 58

T 80 iz

& s 2

70

E

$ 60 B c 0 'E 50

5 i=? 8 40

8 ii

30

T. Sano et al. /Journal ofMembrane Science 95 (1994) 221-228 227

0 0

a 0

0

0

0

0 0.2 0.4 0.6 0.8 1 1.2

Flux (kg/m2h)

Fig. 6. Pervaporation performance of silicalite membranes prepared. Feed ethanol concentration: 5 voll; feed temper- ature: 30°C. (0) Alumina support; ( l ) sintered stainless steel support.

1

0.8 / 0

z I-+ NE 0.6 / -o----o _o_

& z 7 5 0.4 ii ---I

/ 0

0.2 7

0 20 40 60 80

Temperature (“C)

Fig. 7. Influence of feed temperature on pervaporation per- formance. Feed ethanol concentration: 5 voll.

Sorption property

It is well known that the transport mechanism in pervaporation through membranes can be de-

2 (A)

/

80

20 40 60 80 100

EtOH concentration (~01%)

0 ch

1 --I

OS._.-* . '0

io\r 0,

o- 0 20 40 60 80 100

EtOH concentration (~01%)

Fig. 8. Influence of feed ethanol concentration on pervapor- ation performance. Feed temperature: (A) 3O”C, (B) 60°C.

scribed by the sorption-diffusion model [ 171. The membrane performance depends upon ( 1) sorption of permeates in the feed side and (2) diffusion through the membrane. In the case of ethanol/water mixture, as the molecular sizes of ethanol and water are smaller than 5 8, ( z the pore size of silicalite), the sorption process de- termines the pervaporation performance mainly. Therefore, in order to clarify the sorption prop- erties of the silicalite membrane, adsorption iso- therms of ethanol and water on silicalite pow- ders were measured. Fig. 9 shows the adsorption isotherms of ethanol and water at 30°C. It is clearly shown that the amount of ethanol ad- sorbed on silicalite was N 3 times larger than that of water. It is indicated that the high hydropho- bic property of silicalite enhances the selective

228 T. Sano et al. /Journal of Membrane Science 95 (I 994) 221-228

8oI 60

s g40 >

0

EtOH

01 O----O -0-

d

::

8 H20

_.- 0-e -o--.

I I I

0 0.2 0.4 0.6 0.8 1 P/PO

Fig. 9. Adsorption isotherms of ethanol and water on silical- ite at 30°C.

sorption of ethanol into the membrane, resulting in the high separation factor.

4. Conclusions

( 1) The pure silicalite membranes exhibits high ethanol permselectivity with a separation factor a(EtOH/HzO) of more than 60 for a 5 ~01% aqueous ethanol solution at 30’ C.

(2) Cracks or pores between the silicalite grains, which affect the pervaporation perform- ance, do not exist within the membrane.

(3) The high ethanol permselectivity of the membrane is attributable to the high hydrophilic properties of silicalite.

References

[ 1 IT. Sano, Y. Kiyozumi, M. Kawamura, F. Mizukami, H. Takaya, T. Mouri, W. Inaoka, Y. Toida, M. Watanabe and K. Toyoda, Preparation and characterization of ZSM-5 zeolite film, Zeolites, 11 ( 199 1) 842.

[ 21 J.G. Tsikoyiannis and W.O. Haag, Synthesis and char- acterization of a pure zeolitic membrane, Zeolites, 12 (1992) 126.

[ 3lM.W. Anderson, KS. Pachis, J. Shi and S.W. Carr, Syn- thesis of self-supporting zeolite films, J. Mater. Chem., 2 (1992) 255.

[4]J. Dong, T. Dou, X. Zhao and L. Gao, Synthesis of membranes of zeolites ZSM-5 and ZSM-35 by the va- pour phase method, J. Chem. Sot., Chem. Commun., (1992) 1056.

[ 51 G.J. Myatt, P.M. Budd, C. Price and S.W. Carr, Synthe- sis of a zeolite NaA membrane, J. Mater. Chem., 2 (1992) 1103.

[ 6 ] T. Sano, Y. Kiyozumi, K. Maeda, M. Toba, S. Niwa, F. Mizukami, I. Mukoyoshi and H. Shoji, Synthesis and characterization of polycrystalline SAPO-5 film, J. Mol. Catal., 77 (1992) L19.

[7]Y. Kiyozumi, I. Mukoyoshi, T. Sano and F. Mizukami, Synthesis of SAPO-n polycrystalline films, Nippon Ka- gaku Kaishi, (1992) 877.

[ 8]T. Sano, Y. Kiyozumi, K. Maeda, M. Toba, S. Niwa and F. Mizukami, Preparation of zeolite film using cellulose moulding, in R. von Ballmoos, J.B. Higgins and M.M.J. Treaty (Eds.), Proc. 9th Int. Zeolite Conf., Butter- worth-Heinemann, Boston, MA, 1993, p. 239.

[9]J.C. Jansen, W. Nugroho and H. van Bekkum, Con- trolled growth of thin films of molecular sieves on var- ious supports, in R. von Ballmoos, J.B. Higgins and M.M.J. Treaty (Eds.), Proc. 9th Int. Zeolite Conf., But- terworth-Heinemann, Boston, MA, 1993, p. 247.

[ 1OlE.R. Geus, M.J. den Exter and H. van Bekkum, Synthe- sis and characterization of zeolite (MFI) membranes on porous ceramic supports, J. Chem. Sot., Faraday Trans., 88 (1992) 3101.

[ 11 ]E.R. Geus, H. van Bekkum, W.J.W. Bakker and A. Moulijn, High-temperature stainless steel supported zeolite (MFI ) membranes: preparation, module con- struction, and permeation experiments, Microporous Mater., 1 (1993) 131.

[ 121 M.D. Jia, K.V. Peinemann and R.D. Behling, Ceramic zeolite composite membranes. Preparation, characteri- zation and gas permeation, J. Membrane Sci., 82 ( 1993) 15.

[ 13lH.J.C. te Hennepe, D. Bargeman, M.H.V. Mulder and C.A. Smolders, Zeolite-filled silicone rubber mem- branes. Part 1. Membrane preparation and pervapora- tion results, J. Membrane Sci., 35 ( 1987) 39.

[ 14]T. Sano, H. Yanagishita, Y. Kiyozumi, D. Kitamoto and F. Mizukami, Separation of ethanol/water mixture by silicalite membrane, Chem. Lett., ( 1992) 2413.

[ 15 ] R. Le Van Mao, 0. Pilati, A. Marzi, G. Leofanti, A. Villa and V. Ragaini, Adsorption characteristics of some zeo- lites, React. Kinet. Catal. Lett., 15 (1980) 293.

[ 161 D.H. Olson, W.O. Haag and R.M. Lago, Chemical and physical properties of the ZSM-5 substitutional series, J. Catal., 61 (1980) 390.

[ 17lM.H.V. Mulder and C.A. Smolders, On the mechanism of separation of ethanol/water mixtures by pervapora- tion. I. Calculation of concentration profiles, J. Mem- brane Sci., 17 (1984) 289.