Synthesis, reproducibility, characterization, pervaporation and technical feasibility of preferentially b-oriented mordenite membranes for dehydration of acetic acid solution

Journal of Membrane Science 385–386 (2011) 20–29

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Synthesis, reproducibility, characterization, pervaporation and technical feasibility of preferentially b-oriented mordenite membranes for dehydration of acetic acid solution Kiminori Sato ∗ , Kazunori Sugimoto, Tomohiro Kyotani, Naoto Shimotsuma, Tsunehiko Kurata Separation Material Department, Mitsubishi Chemical Corporation, 8-3-1 Chuo, Ami, Inashiki, Ibaraki, Japan

a r t i c l e

i n f o

Article history: Received 15 July 2011 Received in revised form 29 August 2011 Accepted 2 September 2011 Available online 21 September 2011 Keywords: Zeolite Mordenite Acetic acid Membrane–distillation hybrid system Pervaporation

a b s t r a c t This paper describes the development of mordenite membranes for industrial purpose to apply in dehydration processes of acetic acid (AAc) aqueous solution under high-temperature and high-pressure conditions. The membrane formation process is clarified as a function of synthesis time by investigation of the properties of permselective, crystalline state and surface morphology in the synthesized polycrystalline thin film. The crystallographic development from the random orientation to the preferred orientation of b-axis was observed in the synthesized membranes with increasing of crystallization times. The membrane performance in pervaporation was measured in a feed mixture of water (50 wt.%)/AAc (50 wt.%) to be 10.9 kg m−2 h−1 for permeate flux and to be 0.77 × 10−6 mol m−2 s−1 Pa−1 for water permeance with separation factor (˛) of 500 at 130 ◦ C. A technical feasibility study was undertaken for a hybrid pervaporation–distillation system to produce acetic acid with 1 wt.% of water content from a feed flow with water (55 wt.%)/AAc (45 wt.%) for 1000 kg h−1 . It was a case in which dehydration from 55 wt.% to 40 wt.% water content was processed in a feed flow before distillation by pervaporation at 110 ◦ C. Two values of (1) the energy saving ratio compared with a case of distillation alone and (2) the required membrane area were calculated to be 34% and 55 m2 , respectively, using the experimentally determined values of membrane performance. This size of the membrane module for the 55 m2 membrane area could be a realistic scale for industrial utilizations. Therefore, the present fabrication method for mordenite membrane deserves to further technological developments of large-scale membranes and mass-production. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The recent economical and political situation related with global environmental problems such as the global warming issue induces the industry to reduce the energy-consumption not only because of the point of cost-reduction but also because of reduction of CO2 emission. It is a case in Japan that the chemical industry consumes 15% of the energy-consumption in the entire industrial field, and 40% of such energy is consumed in separating operations by distillation processes [1]. Similar situation should be found in the almost industrial nations. Membrane separation is expected to be an alternative separation process to conventional energy-intensive separating processes because the utilization of membrane technology will contribute to the achievement of significant energy savings. Dehydration is the most important and prevailing separating process in the chemical industry. Membrane separation can

∗ Corresponding author. Tel.: +81 29 887 4819; fax: +81 29 887 7291. E-mail address: [email protected] (K. Sato). 0376-7388/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2011.09.001

be utilized for dehydration of azeotropic systems, because those processes are applied by conventional distillation which has a typical energy-consuming feature. Many studies are reported on polymeric membranes for separation of water and solvents [e.g., 2–6]. Almost utilization of polymeric membranes, however, is limited in rather lower temperatures than 100 ◦ C because of their insufficient thermal, mechanical and chemical resistance. For practical utilizations in industrial processes, the membrane separation unit should be installed in a series of process flow operated at high-temperature and high-pressure conditions. The properties of thermal- and mechanical-resistance are required for membrane materials that should be used for the chemical processes. Zeolite membranes have been studied and developed for 20 years for utilizations, in particular, NaA type zeolite membranes have been studied [e.g.,7–13] and commercially developed [14,15]. It has been demonstrated in a plant for ethanol dehydration that a hybrid distillation–membrane system can replace the conventional azeotropic distillation processes [15]. In the case of hybrid system for purification of hydrous ethanol, conventional distillation column and membrane unit are integrated (Fig. 1a).

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

21

Fig. 1. The configuration of membrane–distillation hybrid systems: (a) a distillation–vapor permeation hybrid system for ethanol dehydration and (b) a pervaporation membrane–distillation hybrid system for acetic acid dehydration.

The NaA type zeolite membranes exhibit higher dehydrating performance, however, those have limits to particular processing conditions. It was reported that the membrane performance of NaA type zeolite membranes could be unstable under higher water feed concentrations [16] or under acidity conditions [17]. The NaA zeolite membranes cannot cover such conditions with acidic or high-water contents. However, a potential demand for membrane separation units working under acid conditions is recognized in the chemical industry. Dehydration of acetic acid aqueous solution is an example of the membrane separation for energy savings. Because separation of acetic acid from water in the distillation is an energyconsuming process due to that the evaporating heats for both of acetic acid and water are high. A certain degree of energy saving is expected in dehydration of acetic acid aqueous solution with a hybrid membrane–distillation system [18]. It is a case of our concern that the membrane process would be placed before distillation in a case of acetic acid dehydration where the feed contains high-water content of 55 wt.% is purified into the product of acetic acid with 1 wt.% water content (Fig. 1b). This configuration of the pervaporation–distillation system can take advantage of benefit of membrane separation because such the higher driving force by the higher water vapor pressure through membranes is available for the permeate flux. To realize these dehydration processes, we have to prepare an appropriate membrane that can be operated under such higher temperatures as over 100 ◦ C, strong acidic conditions as pH < 3, and high-water content conditions. Mordenite membrane is a candidate for dehydration of acetic acid aqueous solution because of the property of acid-resistance. Many studies on fabrication methods and permeating properties of mordenite membrane have been reported [19–25]. In particular, the research group of Matsukata and coworkers reported in the series of studies [26–29] on highly water-selective mordenite membranes. Those mordenite membranes [26] exhibit higher permeate fluxes than other mordenite membranes in other previous studies, suggesting that this synthesis method should be developed for our purpose of the practical utilization of acid-proof membrane.

High membrane performance and high reproducibility are required in fabrication of zeolite membranes for industrial purpose. It was indicated that zeolite membranes with high performance could be fabricated by a secondary growth method [e.g., 7–9,12,13,26,30,31]. Furthermore, the advantage of this method has been verified by the studies that up-scaled zeolite membranes were fabricated with the seeded growth method for NaA type and faujasite with higher performance for industrial purpose [8,15,32]. Addition to the secondary growth method, an employment porous substrate with a larger average pore size (∼1.0 ␮m) to ensure higher fluxes was a significant feature in the industrial zeolite membrane fabrication [15,31]. The both of seeded growth method and utilization of substrate with a larger average pore size ∼1.0 ␮m will be employed in mordenite membrane fabrication method in this study. Membrane formation is caused by crystal growth of zeolite on the substrate. The crystal growth is sequential results of crystallization with synthesis time. The crystallization time is a key factor in determining properties in synthesized membranes. Hence, it is required to investigate changes of properties of permselective, crystallographic, morphological and microstructure as a function of synthesis time. Particularly, the relation of reproducibility and synthesis time has to be clarified. The series of synthesis of membranes with time would contribute not only to optimize the synthetic condition but also to give information about formation mechanism of the membrane. A study of technical feasibility on the fabricated mordenite membrane has to be undertaken before developments on the largescale fabrication and mass-production of the membranes. Namely, it is required to estimate the required membrane area and the energy-saving efficiency in a case of dehydration of acetic acid aqueous solution for the present mordenite membrane. Then, we have to judge that the required membrane area could be realistic for industrial cases. The purpose of this study is (1) to synthesize mordenite membrane in a lab-scale with higher performance and reproducibility as a function of synthesis time, (2) to observe the development of microstructure in membranes, (3) to measure performance in

22

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

Fig. 2. Schematic diagram of apparatus for pervaporation at higher temperatures (a) and the sectional schematic for membrane module (b).

pervaporation mode for water–acetic acid separation at temperatures conditions up to 130 ◦ C and (4) to undertake a technical feasibility study to estimate the required membrane area and the saved energy for a case of membrane–distillation hybrid system (Fig. 1b).

The quality of synthesized membrane samples was examined by measurements of permselective properties of permeate flux (J) and separation factor (˛) by conducting pervaporation experiments at 75 ◦ C in a feed mixture of water (10 wt.%)/isopropanol (IPA 90 wt.%) for comparison with results from previous studies. The separation factor (˛) is defined as follows:

2. Experimental

˛=

2.1. Membrane fabrication and characterization Samples of mordenite membrane were synthesized hydrothermally at 180 ◦ C and for 3–8 h by a secondary-growth method on the outer surface of tubular ␣-alumina monolayer-type substrate (i.d. = 9 mm, o.d. = 12 mm, length = 10 cm, NORITAKE Ltd.) with the average pore size of 0.9 ␮m. A dip-coating technique was employed for seeding. The substrate was dipped vertically into an aqueous suspension with concentration of 0.5–1.0 wt.% of commercially available mordenite crystalline powder (Tosoh) which was crashed to be sizes smaller than 1 ␮m. The details of this seeding process have been reported in our previous study [12]. The synthesis mixture with a molar composition of 10Na2 O:0.15Al2 O3 :36SiO2 :960H2 O was used [26]. Hydrothermal treatment was performed at 180 ◦ C in a Teflon-lined stainless autoclave with 300 cm3 volume for one sample. After the hydrothermal reaction, the autoclave was quenched with water and the sample was recovered to be washed with deionized water and dried at room temperature.

[Ca /Cb ]permeate [Ca /Cb ]feed

where Ca and Cb are the weight fraction of components a and b, respectively. In the water/IPA separation, Ca and Cb represent that of water and IPA, respectively. The details of pervaporation experiments were reported in the previous work [12]. The crystallographic characterization of synthesized membranes was carried out with an X-ray diffractometor (Rigaku, RINT-Ultima III) using Cu K␣ radiation with 40 kV and 30 mA. The surface of membrane samples was observed with scanning electron microscopy (SEM; HITACHI, S-4800). The cross section of a membrane was observed with transmission electron microscopy (TEM; HITACHI, HF-2000) with an accelerating voltage of 200 kV. 2.2. Pervaporation in water–acetic acid mixture Measurements of membrane performance of permeate flux (J) and separation factor (˛) were carried out in a feed mixture of water (50 wt.%)/acetic acid (50 wt.%) at 60–130 ◦ C with a feed circulation-type pervaporation apparatus (Fig. 2a). The

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

23

Table 1 Sizes for molecules and zeolitic pores.

Fig. 3. The influence of synthesis time on permeate flux (J), separation factor (˛) and IPA concentration into the permeate in pervaporation at 75 ◦ C in a feed mixture of water (10 wt.%)/IPA (90 wt.%) with data from previous studies of (+) [27], () [26] and () [22], for comparison.

pervaporation apparatus is equipped with a heater of 14 kW and a feed tank of 40 L to supply feed liquid stably. The recirculation feed flow rate was maintained at 10 L min−1 . A sealed membrane sample set in the module of a stainless tube with 15 mm inner diameter (Fig. 2b). The effective membrane area was ca. 20 cm2 . The fluctuation in feed temperature was within 1 ◦ C. The permeate pressure was kept below 1 kPa with a vacuum pump and permeate vapor was collected in a trap chilled with liquid nitrogen. 3. Results and discussion 3.1. Membrane synthesis 3.1.1. Effect of synthetic times on membrane quality and reproducibility We synthesized mordenite membrane samples at conditions of synthetic time from 3 to 8 h to clarify the effect of synthetic time on membrane quality. Six samples were fabricated at an individual condition to investigate reproducibility in the synthesized membrane samples. Fig. 3 shows the results of membrane performance of permeate flux and separation factor (˛) as a function of the synthesis time. The synthesized membranes in this study exhibit the higher permeate fluxes than those in the previous studies. One of reason for the higher permeate fluxes in our membranes could be the employment of substrate having a wider average pore size of 0.9 ␮m. The substrate with an average pore size of ∼0.1 ␮m was utilized in the previous study [26]. These results imply that the employment of substrates with a wider average pore size is effective to high-performance membranes for industrial purpose. The separation factor (˛) in average are increased from ˛ = 200 in samples synthesized for 4 h to ˛ = 3100 in those for 8 h. The longer synthetic time enhances the feature of IPA molecule sieving through the membrane. It is generally suggested that transport pathway for molecules in zeolite membranes can be classified into two types of pores: zeolitic pores and nonzeolitic pores. The main nonzeolitic pores for transport pathway should be contained in grain boundary [12,33,34]. Therefore, the drastic increasing of separation factor from ˛ = 200 in membranes synthesized for 3 h to ˛ = 1400 in those for 4 h should be caused by minimizing of the grain boundary in which undesired non-selective pathway should be contained. The separation factors at 5–8 h for synthesis time are ˛ > 3000 and the leakage of IPA into the permeate are less than 0.6 wt.%. To elucidate the permeate behavior of synthesized membranes in detail, partial fluxes of water and IPA with the function of

Molecule

˚ Kinetic diameter (A)

Water 2-Propanol Acetic acid

2.96 4.70 4.36

Mordenite Pore type

˚ Pore size (A)

12-ring 8-ring

6.5 × 7.0 2.6 × 5.7

synthesis time are investigated (Fig. 4). The partial IPA fluxes through the samples synthesized for 5–8 h are rather constant and those are smaller than 0.003 kg m−2 h−1 . On the other hand, there is a drastic decreasing of partial IPA flux between 4 h (jIPA = 0.017 kg m−2 h−1 ) and 5 h (jIPA = 0.014 kg m−2 h−1 ) for synthetic time. The reduction of IPA flux could be mainly due to the further decreasing of nonzeolitic pores for permeate of IPA molecules. The reduction of IPA leakage through the membranes is not found in the samples synthesized for 6–8 h. These results imply that the minimizing of grain boundary was completed until 5 h of crystallization time. The IPA molecule with size of 0.47 nm in kinetic diameter [35] could permeate through a zeolitic pore of the 12-membered ring (12-MR) in mordenite crystal (Table 1). The very slight permeate flux of IPA could be resulted from this type of zeolitic pore. The pathway of 12-MR channel can be realized in the specific crystalline microstructure in the membrane, which is represented by the preferred orientation of c-axis of mordenite crystals [26–28]. To investigate the development of pathway of 12-MR, X-ray diffraction (XRD) experiments are needed. The results of XRD study for synthesized membranes in this study will be referred in the later section. Membrane samples synthesized for 4–8 h exhibit so high separating performance that the separation factors (˛) are over 1000. A value of separation factor (˛) of 1000 in a feed composition of water (10 wt.%)/IPA (90 wt.%) could be a tentative criteria for practical utilization. The separating ability represented by the value of ˛ = 1000 at this feed composition indicates that the IPA concentration into the permeate is less than 1.0 wt.%. This separating ability could be sufficient for some practical applications. Based on the criterion, it is found that the sufficient separation ability was achieved in membranes synthesized for over 4 h for certain of industrial application. Reproducibility is critical to the zeolite membrane manufacturing for industrial purpose. Fig. 5 shows the experimental results for reproducibility represented by the dispersion of membrane performance synthesized for 3–8 h. The extent of dispersion for permeate fluxes is decreasing with the increasing of synthetic duration. Those

Fig. 4. The influence of synthesis time on average partial IPA flux (jIPA ) and partial water flux (jw ) through mordenite membranes in pervaporation at 75 ◦ C in a feed mixture of water (10 wt.%)/IPA (90 wt.%).

24

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

Fig. 5. The influence of synthesis time on reproducibility of synthesized six membrane samples qualified by pervaporation in pervaporation at 75 ◦ C in a feed mixture of water (10 wt.%)/IPA (90 wt.%).

are small extent of dispersion for samples synthesized for 6–8 h, while the permeate fluxes are lower. The relation of dispersion and permeate fluxes is in a trade-off relationship. The condition for 5 h synthesis time could be optimized, since the condition for 4 h indicates a large extent of dispersion of fluxes. On the other hand, when the membrane with higher separating ability is required for a certain separating process, or membrane samples with higher reproducibility are required, those membranes can be synthesized for longer synthetic durations over 6 h (Figs. 3 and 5). These results indicate that we can exploit various types of the membrane quality that can be controlled by synthesis time. 3.1.2. Observations of crystallization of membranes with SEM, TEM and XRD Fig. 6 shows the XRD patterns of (a) mordenite powder for seed, (b) seeded substrate before synthesis, and (c and d) synthesized mordenite membrane samples for 3 and 8 h. Other phase than mordenite and ␣-alumina for the substrate is not observed. The XRD pattern for mordenite powder should be caused by sufficient

Fig. 6. X-ray diffraction profiles for (a) mordenite seed powder, (b) seeded ␣alumina substrate, (c) synthesized membrane for 3 h and (d) synthesized membrane for 8 h.

Fig. 7. The influence of synthesis time on development of the preferred orientation of b-axis by a value of  defined as [peak intensity of (0 2 0) reflection/peak intensity of (0 0 2) reflection].

random orientation of crystals. The XRD pattern for sample synthesized for 3 h is similar to patterns for the mordenite powder used for seeding. The random orientation was observed in the sample synthesized for 3 h. In contrast, it is observed that the peak intensities for (0 2 0) and (1 5 0) planes which are perpendicular or nearly perpendicular to the b-axis are heightened in the membrane synthesized for 8 h. The difference in XRD patterns for samples synthesized for 3 and 8 h is due to a preferential orientation. These results indicate that a preferential orientation of b-axis (POb) being parallel to the membrane surface was developed at 8 h for synthesis time. Coincidentally, the peak intensity for (0 0 2) that was perpendicular to the c-axis is decreased with increasing of synthesis time from 3 h to 8 h. To clarify the details of development of the preferred orientation in the fabricated membranes with the increasing of crystallization time, a degree of POb is evaluated qualitatively using a value obtained from the XRD patterns on four samples at an individual synthetic time from 3 h to 8 h (Fig. 7). The value of  b–c is defined as [peak intensity of (0 2 0) reflection/peak intensity of (0 0 2) reflection] or  b–c = I (0 2 0)/I (0 0 2). The  b–c -value for random orientation is determined to be 1.1 in the XRD pattern for powder sample of mordenite used as seed crystals in this study. The average value of  b–c factor in samples synthesized for 4 h was 1.2, whereas, the average value of  b–c factor in samples synthesized for 8 h was increased to 5.7. The same consideration on the POb over the preferentially orientation of a-axis (POa) using the peak intensities of reflection of (0 2 0) and (2 0 0) gave the results that the average value of  b–a was increased from 0.18 at 3 h for synthesis time to 1.27 at 8 h. The predominant of POb was confirmed in the wellcrystallized membranes. These results indicate that the degree of POb was developed with the sequence of crystallization from the random orientation in the seeded substrate before hydrothermal treatment. The membranes synthesized for 3 h exhibit the random orientation and the membranes synthesized for 4–5 h show small degrees of POb. The influence of crystallization time on the development of POb is revealed. The POb started at the time between 3 and 4 h for synthesis time and the POb developed drastically after 6 h for synthesis time. To clarify the relations between morphological developments on the membrane surface and synthesis time, SEM observation was carried out on samples synthesized for 3–8 h (Fig. 8). In the SEM observation, following items were particularly examined on the membrane surface: (i) roughness and (ii) morphological features of crystalline shape, sizes and a state of interaction of adjacent crystalline grains (Fig. 9). The SEM images show that the roughness of membrane surface is changed from rough morphology to smooth morphology with increasing of crystallization duration. The flat and smooth surface morphology from a wide scope is observed in

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

25

Fig. 8. SEM (scanning electron microscopy) photographs of the surface of synthesized mordenite membrane for 3–8 h.

samples synthesized for 6–8 h. Significant morphological change is not observed in the samples synthesized between 6 and 8 h (Fig. 8d–f). The surface is covered with well-intersected grains of uniform sizes of ∼2 ␮m. Here, the definition of well-intersected grains is shown in Fig. 8d . On the other hand, the surface of samples for 4 and 5 h is not smooth but rough, uneven. In the SEM image of membrane surface for 3–5 h, some areas have massive crystalline aggregates of framboidal-like texture (“island” in Fig. 9) and some areas exhibit lower height of crystalline surface (“basin” in Fig. 9). No part of the surface is covered with intersected crystals in the samples for 3–4 h, whereas, a part of the surface of sample for 5 h is covered with intersected faceted crystals. The homogeneous morphology has not formed at this stage for 3–5 h. This morphological

feature for 5 h can be considered to be a transitional feature to the matured membrane morphology. In the SEM images from the samples synthesized for 4–5 h, it is found not only that the surface is rough but also some of mordenite crystalline grains exhibit the crystalline surface with hexagonal shape (Fig. 8c ). These hexagonal shaped faces are perpendicular to the c-axis [26–28], indicating that those grains are c-oriented mordenite. On the other hand, the grains exhibiting hexagonal face are diminished in the samples synthesized for 6 and 8 h (Fig. 8d–f). These observations are consistent with the results observed in XRD experiments that the peak intensity for (0 0 2) is decreased and the peak intensity for (0 2 0) is increased between 6 and 8 h.

Fig. 9. Schematic for development of mordenite membrane to observe and interpret the morphology and microstructure.

26

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

Fig. 10. TEM (transmission electron scopy) observation on a cross section of synthesized mordenite membrane for 4 h.

To clarify the vertical microstructure of membrane, TEM observation was performed on a sample synthesized for 4 h (Fig. 10). It is revealed that the synthesized membrane is polycrystalline film composed by continuous alignment of vertically elongated grains. The thickness of membrane above the substrate is ∼2 ␮m. A significant layer of fine particle layer or voids just above the substrate is not observed in this study. Sawamura et al. reported the layer composed of fine crystals above the substrate in TEM observation on a mordenite membrane sample synthesized for 6 h using same chemical composition to this study [29]. The absence of fine particle layer in our sample might be due to difference in synthesis conditions of seeding and/or of properties of substrate. The fine particle layer above the substrate should be formed in the early stage where the reaction between seed crystals, substrate and solution could be critical. It was reported that the fine particle layer might not have critical role to the permeate behavior [29], and we agree with that the top layer of continuous alignment crystals governs selective properties of the membrane. 3.1.3. The influence of microstructure of the membrane on permselective properties Mordenite consists of 12-membered ring (MR) channels ˚ running along to the c-axis and 8-MR (2.6 × 5.7 A) ˚ (6.5 × 7.0 A) running along to the b-axis (Table 1). Fig. 7 shows that the POb was developed in the membranes synthesized for 6–8 h. These results suggest that the zeolitic pore of 8-MR running along to the b-axis is predominant in the upper most part of developed membranes synthesized for 6–8 h. However, the leakage of IPA molecule through the membranes is decreased with increasing of synthetic time to 5 h (Figs. 3 and 4). The partial IPA fluxes through the membranes synthesized for 6–8 h are rather constant and small (jIPA < 0.0028 kg m−2 h−1 ). Namely, the drastic decreasing of partial IPA flux occurred between 3 and 5 h, whereas the drastic development of preferential orientation of b-axis occurred at the synthesis time of 5–6 h. The both features are not always coincident. Therefore, it is concluded that the decreasing of partial permeate flux for IPA through the membranes synthesized for 3–5 h could not always be caused by only the sieving effect with the 8-MR channels. The decreasing of partial permeate flux for IPA could be caused mainly by the closing grain boundary or ineffective of nonzeolitic pores. 3.1.4. Implication to the membrane formation mechanism The formation mechanism of present mordenite membrane is implied by experimental results of properties of permselective, morphology, microstructure, crystallography in the fabricated membranes as the function of synthesis time. It has been generally reported that the polycrystalline thin films of various kinds of

functional materials such as of diamond [36] and nitrides [37] are constructed with preferred columnar crystals formed by the crystal growth mechanism of geometric selection [38] or evolutional selection [39]. This growth mechanism is usually referred to as the van der Drift model or the mechanism of evolutional selection. In this mechanism, grains having their faster growth direction vertical to the substrate surface survive as the polycrystalline film forms, whereas other grains are eliminated by overgrowth of the survived grains. This mechanism is also suggested to be worked in formation of zeolite membranes [34,40–42]. As for mordenite membrane, Matsukata and coworkers [28,42] proposed that the mechanism of evolutional selection should be operated in the formation of mordenite membrane with a preferentially oriented c-axis. The present experimental results reveal (1) the columnar structure of the polycrystalline thin film in the TEM observation; (2) the developments of POb from the random orientation with increasing of crystallization time in the XRD observation; (3) the elimination of grains exhibiting the hexagonal crystalline faces up to surface which are grown with c-axis oriented until 6 h for synthesis time in the SEM observation. These features are the typical evidence for the evolutional selection mechanism. The mordenite grains having their lattice direction of c-axis outward are eliminated by adjacent crystals having their lattice direction of b-axis outward, because the planes of (0 2 0) and (1 5 0) which are perpendicular or nearly normal to b-axis have faster growth rates. Namely, the crystals with b-axis direction to the surface are tend to survive to the ultimate membrane surface. It is interpreted that these results are the evidence for the mechanism of evolutional selection. In the synthesis time at 4–5 h, small degree of POb was observed. These results indicate that the competitive growth between crystalline grains with b-axis and c-axis directions was not so severe that adjacent crystals did not interfere each others extensively and many crystals with c-axis direction to the surface were not eliminated completely. This interpretation is supported by the SEM observation that the crystals with c-axis direction to the surface were confirmed in the membrane surface synthesized for 3–5 h. Furthermore, from the results of TEM observation, the establishment of columnar structure of the membrane at 4 h for synthesis time indicates the competitive growth by the mechanism of evolutional selection should be operated without significant developments of the POb. Namely, the mechanism of evolutional selection plays a key role in the formation of closely packed polycrystalline thin film composed tightly by columnar grains growing perpendicular to the membrane surface without voids and other deteriorate space than grain boundary in all synthesis duration. 3.2. Pervaporation performance in water–acetic acid mixture Pervaporation performance of permeate flux (J) and separation ability in a feed mixture of water (50 wt.%)/acetic acid (50 wt.%) through a membrane sample synthesized for 5 h were measured at 60–130 ◦ C (Fig. 11). The permeation fluxes are increased from 0.89 kg m−2 h−1 at 60 ◦ C to 10.9 kg m−2 h−1 at 130 ◦ C. The apparent activation energy for permeate flux is determined to be 36 kJ mol−1 at a temperature range of 60–90 ◦ C and to be 41 kJ mol−1 at range of 100–130 ◦ C. The separation factor (˛) ranged from ˛ = 350 at 120 ◦ C to ˛ = 640 at 60 ◦ C, indicating a tendency of decreasing of the separation factor with increasing temperature. The leakage of acetic acid into the permeate is small for all temperature range, particularly those are less than 0.3 wt.% above 80 ◦ C. This separating behavior exhibits a sufficient ability for practical utilizations in dehydration of acetic acid aqueous solution. Fig. 12 shows the relations of permeate flux (J), water permeance (˘ w ) and the difference of water activity between feed and permeate sides. An approximate linear correlation is found between the permeate flux and the difference of water activity

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

27

stability of membrane. The reproducibility of membrane performance was confirmed and no instability was recognized within the laboratory time-scale in this study. 3.3. A study of technical feasibility on pervaporation–distillation hybrid system

Fig. 11. Pervaporation performance of mordenite membrane in a feed mixture of water (50 wt.%)/acetic acid (50 wt.%) at 60–130 ◦ C.

Fig. 12. The relations between permeate flux (J), water permeance (˘ w ) and the driving force of water activity across the membrane.

across the membrane. The permeation is principally governed by the driving force of the difference of water activity across the membrane. The influence of temperature on water permeance is examined. The water permeance is slightly increased of 10% from 8.6 × 10−7 mol m−2 s−1 Pa−1 at 60 ◦ C to 9.5 × 10−7 mol m−2 s−1 Pa−1 at 90 ◦ C, on the other hand, the water permeance is decreased of 23% to 7.7 × 10−7 mol m−2 s−1 Pa−1 at 130 ◦ C. The pervaporation experiments were repeated at the same conditions of temperatures up to 130 ◦ C from 60 ◦ C to observe the

A study of technical feasibility was undertaken on a hybrid pervaporation–distillation system producing dehydrated acetic acid with water (1 wt.%)/acetic acid (99 wt.%). The aim of technical feasibility is (1) to calculate the required membrane area and (2) to calculate the energy savings by comparison with a base case of distillation alone. The calculation was undertaken for a given configuration of the hybrid membrane–distillation system in which a separation unit of pervaporation with the mordenite membranes is placed before a conventional distillation tower with a feed flow rate of 1000 kg h−1 with a feed composition of water (55 wt.%)/acetic acid (45 wt.%) (Fig. 13a). In the technical feasibility for hybrid system, we examined three cases of the water content (Xw ) in the flow from the membrane unit into the distillation (Table 2). In three cases, the burden of dehydration with membrane unit (Xw = Xf0 − Xw , Fig. 13b) was changed. The base case of dehydration by only distillation was investigated for comparison (Table 2). A commercial program Aspen Plus was used to calculate the required energy. Table 2 shows the results including the determined theoretical stage and the reflux ratio. The energy of 1.4 G cal h−1 is required for the base case of distillation process without membrane separation. On the other hand, the total required energy of the pervaporation–distillation hybrid system is decreased with increasing of the burden by membrane separation for dehydration before distillation. Compared with the base case without membrane separation, the reduction ratio of the total energy change from 34% at Xw = 60% to 56% at Xw = 80%. These results indicate that the employment of membrane separation in front of distillation contribute drastically to energy savings. The required membrane areas were calculated using the values of water permeance (˘ w ) that was measured at the operating temperature condition of 110 ◦ C in a feed composition of water (55 wt.%)/acetic acid (45 wt.%). The influence of water content in the feed on the water permeance was not taken in this calculation. It is calculated that membrane are of 55 m2 is required for dehydration of 1000 kg h−1 from 55 wt.% water content into 40 wt.% water content. The larger membrane areas of 95 m2 and 140 m2 are required for the conditions of Xw = 70 wt.% and Xw = 80 wt.%,

Fig. 13. The configuration of pervaporation membrane–distillation hybrid system (a) and distillation alone for technical feasible study.

28

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

Table 2 Results of calculation of the required energy and membrane area with feed rate of 1000 kg h−1 . Preheat for feed [kcal h−1 ]

Productenergy ratio [kcal kg−1 Aac]

Required membrane area [m2 ]

Dehydration by membrane unit water content [wt.%]

Theoretical stage number

Reflux ratio

Base case

Distillation only XF0 → Xw 45 → 60 45 → 70 45 → 80

30

3.63

1,418,695

0

0

1,418,695

3,185

–

30 28 25

3.65 3.86 4.27

745,856 499,501 312,532

134,097 191,151 234,054

51,144 51,144 51,144

931,096 741,795 597,729

2,090 1,665 1,342

55 95 140

Case1 Case2 Case3

Reboiler [kcal h−1 ]

Heat for membrane separation [kcal h−1 ]

Condition

Total [kcal h−1 ]

Table 3 Required membrane area and modules for dehydration mordenite membrane for 1000 kg h−1 feed rate. Zeolite type

Mode

Processing temperature [◦ C]

Mordenite A

PV VP

110 120

Water content in/out membrane

Input [wt.%]

Output [wt.%]

55 10

40 0.5

Feed rate [kg h−1 ]

Extracted water weight [kg h−1 ]

1000 1000

150 99.5

Water permeance [mol m−2 s−1 Pa−1 ]

8.5 × 10−7 25 × 10−7

Required membrane area [m2 ]

Required modules with 12 m2 membrane area

55 36

5 3

respectively. Fig. 14 shows the relation of required membrane area and required energy for the product of acetic acid in the hybrid pervaporation–distillation system. With increasing of the dehydrating load with the membrane separation before distillation, the required energy for production of acetic acid is increasing and the required membrane area is decreasing. A sensitive analysis on the required energy for product (E) to the dehydrating load with the membrane separation (Xw ) indicates that the efficiency of membrane employment indicated by a gradient of dE/dXw becomes to be rather gentle. These results indicate that (1) an employment of membrane separation is effective for wide range of water concentration in the feed and (2) the efficiency is particularly effective in placement of membrane separation at conditions with higher water content in the feed such Xw ∼20 wt.% before distillation. To understand the scale of membrane area for the present case of the pervaporation–distillation hybrid system dealing with 1000 kg for an hour, a comparative examination would be undertaken. The estimated scale of module unit with the mordenite membranes could be evaluated by a comparison with a previously installed case of dehydration plant using with NaA zeolite membrane for dehydration of hydrous ethanol. Table 3 shows the

comparison of dehydration for acetic acid aqueous solution with the mordenite membrane and for the hydrous ethanol with the NaA zeolite membrane. In the installed plant for dehydration of hydrous ethanol from water (10 wt.%)/ethanol (90 wt.%) to anhydrous ethanol >99.7 wt.% with feed rate of ca.1000 kg h−1 , three membrane modules have been set in dehydration. The utilized membrane module has membrane area of 12 m2 (Fig. 15). If we prepare up-scaled mordenite membranes with membrane performance reported in this study for lab-scale samples, we will be able to install the mordenite membrane in the membrane module with membrane area of 12 m2 . This result indicates that the suggested value of membrane area 55 m2 in the case of this technical feasibility can be realized by 5 membrane modules. The unit scale of 5 membrane modules is realistic scale for applications in the industry. We expect from a point view of cost for raw material that the cost of mordenite membrane with an industrial scale would not be so different from the cost for present commercialized NaA zeolite membrane. Because the most expensive material for zeolite membranes is substrate material, and the similar substrate material would be used for both membranes of NaA and mordenite. Noted that other factors such as a manufacturing apparatus and

Fig. 14. The influence of the water weight ratio (Xw ) in the flow into distillation after pervaporation on the required membrane area and the required energy.

Fig. 15. An available membrane module with membrane area of 12 m2 for dehydration.

K. Sato et al. / Journal of Membrane Science 385–386 (2011) 20–29

reproducibility could influence the cost of mordenite membrane manufacturing. 4. Conclusions We developed a fabrication method for mordenite membrane with higher permeate flux and selectivity than those previously reported and measured the membrane performance of flux and separation factor under a wide range of temperatures containing the practical process conditions of higher temperatures up to 130 ◦ C. Higher permeate fluxes up to 10.9 kg m−2 h−1 were observed in a feed mixture of water (50 wt.%)/acetic acid (50 wt.%) and the leakage of acetic acid concentration into the permeate was less than 0.3 wt.% at temperatures over 100 ◦ C. A technical feasibility was carried out for the cases in which a pervaporation unit dehydrates acetic acid aqueous solution with a feed rate of 1000 kg h−1 from 55 wt.% of water content to 40, 30 and 20 wt.% of water contents at 110 ◦ C before the solution is purified into acetic acid with 1 wt.% of water content by distillation. The results suggest that the developed mordenite membrane deserves to the further technically developments of up-scale and mass-production for realization of practical utilizations in the industry. Acknowledgments This study was financially supported in part by a project of “Green and sustainable chemistry/fundamental development of ordered-nanoporous membranes for highly-refined separation technology” in the New Energy and Industrial Technology Development Organization’s (NEDO), Japan. We thank Prof. M. Matsukata, the director of the project, for his support and encouragement. References [1] Y. Iwamoto, H. Kawamoto, Trends in research and development of nanoporous ceramic separation membranes—saving energy by applying the technology to the chemical synthesis process, Sci. Technol. Trend 32 (2009) 43. [2] R.Y.M. Huang, Pervaporation Membrane Separation Processes, 1991. [3] R.W. Baker, Membrane Technology and Applications, McGraw-Hill, New York, 2004. [4] P. Shao, R.Y.M. Huang, Polymeric membrane pervaporation, J. Membr. Sci. 287 (2007) 162. [5] P.D. Chapman, T. Oliveira, A.G. Livingston, K. Li, Membranes for the dehydration of solvents by pervaporation, J. Membr. Sci. 318 (2008) 5. [6] D. Gorri, A. Uritaga, I. Ortiz, Pervaporation recovery of acetic acid from an acetylation industrial effluent using commercial membranes, Ind. Eng. Chem. Res. 44 (2005) 977. [7] H. Kita, K. Horii, Y. Ohtoshi, K. Tanaka, K. Okamoto, Synthesis of a zeolite NaA membrane for pervaporation of water/organic mixtures, J. Mater. Sci. Lett. 14 (1995) 206. [8] M. Kondo, M. Komori, H. Kita, K. Okamoto, Tubular-type pervaporation module with zeolite NaA membrane, J. Membr. Sci. 133 (1997) 133. [9] M. Pera-Titus, J. Liorens, J. Tejero, F. Cunill, Description of the pervaporation performance of A-type zeolite membranes: a modeling approach based on the Maxwell–Stefan theory, Catal. Today 118 (2006) 73–84. [10] Y. Li, H. Chen, J. Liu, W. Yang, Microwave synthesis of LTA zeolite membranes without seeding, J. Membr. Sci. 277 (2006) 230. [11] J. Zah, H.M. Krieg, J.C. Breytenbach, Pervaporation and related properties of time-dependent growth layers of zeolite NaA on structured ceramic supports, J. Membr. Sci. 284 (2006) 276–290. [12] K. Sato, T. Nakane, A high reproducible fabrication method for industrial production of high flux NaA zeolite membrane, J. Membr. Sci. 301 (2007) 151. [13] K. Sato, K. Sugimoto, T. Nakane, Preparation of higher flux NaA zeolite membrane on asymmetric porous support and permeation behavior at higher temperatures up to 145 ◦ C in vapor permeation, J. Membr. Sci. 301 (2008) 181.

29

[14] Y. Morigami, M. Kondo, J. Abe, H. Kita, K. Okamoto, The first large-scale pervaporation plant using tubular-type module with zeolite NaA membrane, Sep. Purif. Technol. 25 (2001) 251. [15] K. Sato, K. Aoki, K. Sugimoto, K. Izumi, S. Inoue, J. Saito, S. Ikeda, T. Nakane, Dehydrating performance of commercial LTA zeolite membranes and application to fuel grade bio-ethanol production by hybrid distillation/vapor permeation process, Micropor. Mesopor. Mater. 115 (2008) 184. [16] Y. Li, H. Zhou, G. Zhu, J. Liu, W. Yang, Hydrothermal stability of LTA zeolite membranes in pervaporation, J. Membr. Sci. 297 (2007) 10. [17] Y. Hasegawa, T. Nagase, Y. Kiyozumi, T. Hanaoka, F. Mizukami, Influence of acid on the permeation properties of NaA-type zeolite membranes, J. Membr. Sci. 349 (2010) 189. [18] A. Verhoef, J. Degreve, B. Huybrechs, H. van Veen, P. Pex, B. Van der Bruggen, Simulation of a hybrid pervaporation–distillation process, Comput. Chem. Eng. 32 (2008) 1135. [19] K. Suzuki, Y. Kiyozumi, T. Seike, K. Obata, Y. Shindo, S. Shin, Preparation and characterization of a zeolite layer, Chem. Express 5 (1990) 793. [20] N. Nishiyama, N. Ueyama, M. Matsukata, Gas permeation through zeolite–alumina composite membrane, AiChE J. 43 (1997) 2724. [21] A. Tavolaro, A. Julbe, C. Guizard, A. Basile, L. Cot, E. Drioli, Synthesis and characterization of a mordenite membrane on an ␣-Al2 O3 tubular support, J. Mater. Chem. 10 (2000) 1131. [22] X. Lin, E. Kikuchi, M. Matsukata, Preparation of mordenite membranes on ␣-alumina tubular support for pervaporation of water–isopropyl alcohol mixtures, Chem. Commun. (2000) 957. [23] Y. Zhang, Z. Xu, Q. Chen, Synthesis of small crystal polycrystalline mordenite membrane, J. Membr. Sci. 210 (2002) 361. [24] L. Casado, R. Mallada, C. Tellez, J. Coronas, M. Menendez, J. Santamaria, Preparation, characterization and pervaporation performance of mordenite membranes, J. Membr. Sci. 216 (2003) 135. [25] M. Asghari, T. Mohammadi, R.F.F. Alamdari, F. Agend, Thin-layer template-free polystalline mordenite membranes on cylindrical mullite supports, Micropor. Mesopor. Mater. 114 (2008) 148. [26] G. Li, E. Kikuchi, M. Matsukata, The control of phase and orientation in zeolite membranes by the secondary growth method, Micropor. Mesopor. Mater. 62 (2003) 211. [27] G. Li, E. Kikuchi, M. Matsukata, Separation of water–acetic acid mixtures by pervaporation using a thin mordenite membrane, Sep. Purif. Technol. 32 (2003) 199. [28] M. Matsukata, K.-I. Sawamura, T. Shirai, M. Takada, Y. Sekine, E. Kikuchi, Controlled growth for synthesizing a compact mordenite membrane, J. Membr. Sci. 316 (2008) 18. [29] K.-I. Sawamura, T. Shirai, T. Ohsuna, T. Hagino, M. Takada, Y. Sekine, E. Kikuchi, M. Matsukata, Separation behavior of steam from hydrogen and methanol through membrane, J. Chem. Eng. Jpn. 41 (2008) 870. [30] J. Hedlund, J. Sterte, M. Anthonis, A.-J. Bons, B. Carstensen, N. Corcoran, D. Cox, H. Deckman, W.D. Gijnt, P.-P. de Moor, F. Lai, J. McHenry, W. Mortier, J. Reinoso, J. Peters, High-flux MFI membranes, Micropor. Mesopor. Mater. 52 (2002) 179. [31] K. Sato, K. Sugimoto, T. Nakane, Synthesis of industrial scale NaY zeolite membranes and ethanol permeating performance in pervaporation and vapor permeation up to 130 ◦ C and 570 kPa, J. Membr. Sci. 310 (2008) 161. [32] K. Sato, K. Sugimoto, T. Nakane, Mass-production of tubular NaY zeolite membranes for industrial purpose and their application to ethanol dehydration by vapor permeation, J. Membr. Sci. 319 (2008) 244. [33] M. Nomura, T. Yamaguchi, S. Nakao, Ethanol/water transport through silicalite membranes, J. Membr. Sci. 144 (1998) 161. [34] Z. Liu, T. Ohsuna, K. Sato, T. Mizuno, T. Kyotani, T. Nakane, O. Terasaki, Transmission electron microscopy observation on fine structure of zeolite NaA membrane, Chem. Mater. 18 (2006) 922. [35] M.E. van Leeuwen, Derivation of Stockmayer potential parameters for polar fluids, Fluid Phase Equilibr. 99 (1994) 1. [36] Ch. Wild, N. Herres, P. Koidl, Texture formation in polycrystalline diamond films, J. Appl. Phys. 68 (1990) 973. [37] Y. Kajiwara, S. Noda, H. Komiyama, Comprehensive perspective on the mechanism of preferred orientation in reactive-sputter-deposited nitrides, J. Vac. Sci. Technol. A 21 (2003) 1943. [38] P.G. Grigoriev, Ontogeny of minerals, IPST, Jerusalem, 1965. [39] A. van der Drift, Evolutionary selection: a principle governing growth orientation in vapour-deposited layers, Philips Res. Rep. 22 (1967) 267. [40] G. Bonilla, D.G. Vlachos, M. Tsapatis, Simulations and experiments on the growth and microstructure of zeolite MFI films and membranes made by secondary growth, Micropor. Mesopor. Mater. 42 (2001) 191. [41] A.-J. Bons, P.D. Bons, The development of oblique preferred orientations in zeolite films and membranes, Micropor. Mesopor. Mater. 62 (2003) 9. [42] G. Li, R. Lin, E. Kikuchi, M. Matsukata, Growth mechanism of a preferentially oriented mordenite membrane, J. Zhengjiang Univ. SCI 6B (2005) 369.