You are on page 1of 10

Journal of Membrane Science 244 (2004) 141–150

A semi-continuous method for the synthesis of NaA zeolite


membranes on tubular supports
M.P. Pina, M. Arruebo, M. Felipe, F. Fleta, M.P. Bernal,
J. Coronas, M. Menéndez, J. Santamarı́a∗
Department of Chemical Engineering, University of Zaragoza, 50009 Zaragoza, Spain

Received 26 March 2004; received in revised form 23 June 2004; accepted 25 June 2004
Available online 11 September 2004

Abstract

Zeolite NaA membranes have been synthesized by secondary growth on the external surface of ␣-alumina tubular supports using a semi-
continuous system in which fresh gel was periodically supplied to the synthesis vessel. Compared to traditional batch methods, the procedure
developed in this work provides a better control of the synthesis and crystallization conditions and is easier to implement at an industrial scale.
The influence of the renewal rate during synthesis on the pervaporation and vapor permeation performance of the membranes was studied,
with the best performance corresponding to the case where the gel was renewed every 20 min of synthesis time. The membranes obtained by
the semi-continuous method displayed an excellent separation performance in the pervaporation of ethanol/water mixtures (e.g., a separation
factor of 3600 at a water permeation flux of 3.8 kg/h m2 ).
© 2004 Elsevier B.V. All rights reserved.

Keywords: Zeolite NaA membranes; Semi-continuous synthesis; Renewal rate; Pervaporation; Ethanol dehydration

1. Introduction able membrane. Recently, many investigations have reported


on the PV properties of microporous inorganic membranes
Supported zeolite membranes have been extensively including silica [4] and hydrophilic zeolite membranes such
developed during the last decade in view of their poten- as zeolite A [5–10], faujasite-type zeolites (X and Y) [11],
tial applications in membrane separators, membrane reac- MFI-type zeolites [12] and mordenite [13]. It is clear that
tors and selective sensors [1,2]. Zeolite membranes combine zeolite membranes are interesting for industrial separation
the intrinsic properties of zeolites and the advantages of a processes on account of their chemical and thermal stabil-
membrane-type configuration, making them excellent can- ity, and of their highly specific interactions with permeating
didates for the separation applications based on preferential molecules. However, the problem of reproducibly achieving
adsorption, differences in diffusion rates or molecular sieving large membrane areas in suitable modules is still far from
effects. fully solved, and this is likely one of the main reasons for
Pervaporation (PV) processes have been advocated as an the much wider industrial development of non-zeolitic in-
effective and energy-efficient technique for the separation of organic membranes. Thus, at least several dozen industrial
azeotropic and close-boiling liquid mixtures, and for the de- manufacturers nowadays prepare a wide variety of inorganic
hydration of organic solvents [3]. On the other hand, vapor membranes (glass, silica, metallic, ceramic and carbon [14]),
permeation (VP) is a separation process in which the compo- while only one type of zeolite membrane (zeolite A) is indus-
nents of a vapor-phase feed are selectively separated on a suit- trially manufactured on a relatively large scale, and by one
company (Mitsui Engineering).
∗ Corresponding author. Tel.: +34 976 761153; fax: +34 976 762142. A survey of recent literature on zeolite membrane prepa-
E-mail address: iqcatal@posta.unizar.es (J. Santamarı́a). ration reveals that synthesis processes, even for well-known

0376-7388/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2004.06.049
142 M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150

zeolite structures (e.g., MFI, LTA) are still carried out batch- zeolite flat membranes on PTFE and glass supports using
wise, using a hydrothermal route. Successful membrane for- a circulated solution system 50 cm3 /min and heating only
mation in one-process (direct membrane synthesis) requires the substrate area in the range of 353–403 K for crystallisa-
nucleation and growth of zeolite crystals on the support sur- tion periods of 24–48 h. Under these conditions, the forma-
face, a process that competes with solution events. To reduce tion of phases concomitant to NaA zeolite was avoided and
crystal nucleation and growth in the bulk of the synthesis solu- the amount of crystals forming in the liquid phase could be
tion the so-called secondary (seeded) growth methods can be greatly reduced. Kumakiri [20] proposed a continuous sys-
used, involving a first step in which a closely packed layer of tem for cross-flow filtration of a suspension of zeolite A not
zeolite seed crystals is deposited onto the surface of a support. only for the formation of a seed layer but also for the subse-
In this way, subsequent hydrothermal synthesis can be car- quent membrane growth under a constant feed flow rate. The
ried out under less concentrated conditions that are sufficient tubular zeolite membranes obtained showed modest pervapo-
for crystal growth but hinder nucleation in the bulk. Both di- ration selectivity, with H2 O/ethanol separation factors of 25
rect and secondary synthesis methods may be acceptable for for a zeolite layer thickness of 2 ␮m. The author pointed out
small-scale applications (a few cm2 of membrane area), but that the use of longer synthesis times should make the zeolite
present significant problems of the synthesis of larger struc- membrane denser and increase selectivity. Richter et al. [21]
tures. Generally speaking, increasing the zeolite membrane have prepared 30 ␮m thick ZSM-5 membranes on the inner
area in industrial modules involves the use of longer tubu- surface of ceramic tubes and capillaries by seeding with nano-
lar or monolith supports, and/or reduction of tubular diame- sized MFI particles and pumping a clear synthesis solution
ters to capillary dimensions to increase the membrane area 0.25 cm/min through the support tubes. The membranes so
per unit volume. In addition to the variations of composition obtained showed H2 /SF6 permselectivities above the Knud-
with time that necessarily take place during batch process- sen factor (up to 30). Finally, Tiscareño et al. [22] used a
ing, the use of longer and narrower supports will likely cause centrifugal system for the synthesis of zeolite A membranes
the development of concentration gradients, leading to non- with the double objective of achieving solution renewal and
homogeneous membrane characteristics, and possibly also at the same time driving the crystal nuclei formed in the ho-
to the formation of defects and non-desired phases on the mogeneous phase towards the support. In this way, crystal
membrane. nuclei were deposited before they reached a size that would
A number of studies have attempted to solve the prob- be difficult to accommodate into the growing membrane, thus
lems that prevent the synthesis of large area membranes. promoting the formation of a higher quality zeolite layer.
Regarding hydrothermal synthesis, the works of Morigami In this work, we propose a new semi-continuous procedure
et al. [9] and Li et al. [15], who prepared NaA zeolite and to prepare NaA zeolite membranes over tubular supports in
ZSM-5 zeolite membranes, respectively, on tubular alumina a single step, in a reaction vessel where the synthesis solu-
supports are noteworthy, since the surface of the membranes tion is rapidly removed from the membrane environment at
prepared approached 300 cm2 per tube. Although batch pro- regular intervals, and simultaneously replaced by the same
cessing was used in both cases, successful membrane forma- volume of fresh gel. This represents a compromise between
tion was achieved either by using a suitable excess of reactants a batch process, in which there is a significant depletion of
[9] or by repeated synthesis cycles [15]. Another possibility reactants over time, and a process with a continuous feed of
that could be explored to circumvent the problem of reactant low flow rate, where a certain variation of concentration along
depletion during synthesis is the use of the so-called dry gel the support is inevitable, likely leading to inhomogeneous
method, involving deposition of a layer of dry aluminosil- growth of the membrane. The aim of this work is to improve
icate gel on the support, followed by the transformation to the synthesis concerning the above-mentioned problems of
the zeolite under the presence of vapors [16]. Other authors unwanted synthesis in the bulk of the liquid phase and poor
propose the direct heating of the substrate while the reaction crystal intergrowth due to local exhaustion of reactants in the
mixture is kept at lower temperature [17,18]. In this manner, autoclave. At the same time, this procedure allows a better
the reaction is suppressed in the bulk and promoted on the sur- control of the composition of the reacting mixture and avoids
face, and the phase transformations of metastable materials heating and cooling steps between synthesis, thus improving
to other zeolite types (P zeolite, X zeolite, hydroxi-sodalite) the feasibility of the scale up.
can be delayed.
Even if the problems of reactant depletion and synthesis of
zeolite material in the bulk of the solution were solved, an im- 2. Experimental
portant problem remains, related to the fact that several batch
synthesis cycles (with their associated heating and cooling 2.1. Membrane preparation
processes) are often required to achieve a zeolite membrane
of good quality. Thus, a synthesis procedure in which reac- Zeolite NaA membranes were hydrothermally synthe-
tants are continuously supplied to the synthesis vessel while sized on the external surface of ␣-alumina symmetric tubes
this is maintained at a constant temperature would clearly of 7 mm ID and 10 mm OD, with a nominal pore size of
be desirable. Yamazaki and Tsutsumi [19] prepared A-type ∼1900 nm. These supports were first enamelled at both ends
M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150 143

defining a permeation length of approximately 5 cm. A seed- gel volume from the autoclave (10 cm3 ) corresponding to the
ing step was ensured, by rubbing the outer support surface length of pipe between electro-pneumatic valves E1 and E2.
with zeolite A crystals with a mean size of 1 ␮m. Sec- The length and diameter of this piece of pipe could be changed
ondary growth of the seed layer deposited on the support in order to adjust the renewal volume to a desired value. Af-
tubes was carried out for 5 h at approximately 90 ◦ C us- ter each renewal step, compressed air was admitted through
ing a synthesis gel with the following molar composition: valve E3 for the complete removal of the discharged fluid.
1.8SiO2 :1.0Al2 O3 :3.9Na2 O:273H2 O. To avoid corrosion in case the electrovalves were reached by
SiO2 was supplied as SiO2 ·NaOH (27 wt.% SiO2 , 14 wt.% the highly alkaline synthesis solution, an intermediate vessel
NaOH and 59 wt.% H2 O; from Aldrich) and Al2 O3 as filled with cold paraffin was used as a buffer. The renovation
Al2 O3 ·3H2 O (from I.Q.E). Once the Al precursor was com- rate was set at the desired value by the electrovalves control
pletely dissolved in NaOH solution (50 wt.%, prepared from unit.
NaOH pellets 99% pure supplied from Merck) under vig- A first series of synthesis revealed the presence of sig-
orous stirring at 60 ◦ C, the Si precursor was slowly added, nificant temperature variations in the synthesis vessel due
keeping constant the temperature to avoid the formation of to the low temperature of the feed. This was corrected by
aggregates. The synthesis gel was then poured onto a ther- the use of a heating tape wrapped around the gel container
mally insulated stainless steel vessel. and pipelines connected to the autoclave inlet, which pre-
The experimental set-up employed for the semi- heated the feed to a constant temperature of around 60–65 ◦ C.
continuous synthesis of NaA zeolite membranes, is shown This temperature was high enough to reduce thermal upset
in Fig. 1A. Fresh gel was supplied from a vessel pressur- upon feed introduction, and low enough to avoid synthesis in
ized with dry nitrogen (around 10 bar), by removing a known the inlet pipe. A more detailed scheme of the Teflon-lined

Fig. 1. (A) Experimental set-up for the semi-continuous synthesis of NaA zeolite membranes; (B) scheme of the Teflon-lined autoclave used.
144 M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150

Fig. 2. Experimental set-up for the pervaporation and vapor permeation system.

autoclave immersed in the thermostatic bath is shown in


Fig. 1B. The reaction temperature was monitored by means
of two thermocouples positioned in the upper (top) and lower
sections (bottom) of the autoclave in order to register temper-
ature fluctuations with gel introduction. The membrane was
fixed in vertical position by means of a Teflon plug in such a
way that the points for temperature measurement were close
to the zeolitic layer growing on the external seeded surface
of the alumina support.

2.2. Membrane characterization and testing

The experimental set-up for the PV and VP of wa-


ter/ethanol mixtures is shown in Fig. 2. In both cases, the
membrane module was immersed in a temperature-controlled
bath (SELECTA, PRECISTERM) of silicone oil (Rhodosil).
The temperature was measured with a thermocouple located
in the membrane tube side (permeate). An HPLC pump
(SHIMADZU, LC-10AT VP) was used to feed 0.083 ml/s
(5 ml/min) of a water/ethanol (10/90 wt.%) liquid mixture to
the outer side of the membrane, through a heating coil im-
mersed in the thermostatic bath. For PV measurements, the
pressure at the retentate side was adjusted to values higher
than the mixture vapor pressure (around 6 bar at 125 ◦ C),
whereas for VP measurements a pressure of 1.1 bar at 140 ◦ C
was used. In either case, the vacuum (around 0.5 mbar) at the
permeate side achieved by a rotary pump (TELSTAR, G6)
provided the driving force for diffusion and pervaporation.
Both the feed and the liquid collected in the liquid-N2 traps
were analysed in a gas chromatograph (VARIAN, 3400) with
TCD and FID detectors. The H2 O/ethanol separation factor Fig. 3. Temperature profiles vs. synthesis time at 1 renewal every 16 min for
was calculated as the quotient between the ratios of the weight feed gel introduced at: (A) room temperature; (B) 60 ◦ C.
fractions of water and alcohol in the permeate and feed sides,
respectively.
The phases present in the NaA membranes were charac- autoclave after synthesis and of the membrane top lay-
terized by X-Ray diffraction (XRD) analysis (Rigaku/Max ers. The membranes were also examined by scanning
diffractometer Cu K␣ radiation and graphite monochro- electron microscopy, SEM (JEOL JSM-6400 operating at
mator) of the powders collected at the bottom of the 20 kV).
M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150 145

Table 1
Autoclave temperatures measured during hydrothermal synthesis of NaA zeolite membranes using the semi-continuous system
Preheating Renewal rate (min−1 ) Top temperatures (◦ C) Bottom temperatures (◦ C)
Minimum Maximum Average Minimum Maximum Average
NO 1/16 61 94 89 37 99 75
YES 1/16 61 96 93 72 100 92
NO 1/20 61 98 90 41 103 82
YES 1/20 59 96 93 67 99 90
NO 1/25 61 94 89 43 101 92
YES 1/25 61 97 90 76 105 96

3. Results and discussion autoclave. With this modification it can be seen that, while a
drop of temperature at the bottom of the autoclave still occurs
3.1. Semi-continuous synthesis of NaA zeolite upon gel renewal, the range of temperature variation within
membranes the autoclave has been notably narrowed. Interestingly, the
effect of gel renewal upon the temperature at the top of the
All the membranes prepared in this work have been syn- autoclave is now more marked, and a clear coupling between
thesized at a temperature of approximately 90 ◦ C, measured the temperature oscillations of both thermocouples can be
at the top of the autoclave. After initial batch synthesis ex- observed. This is probably due to the fact that in the first case
periments with the above-described gel composition, XRD (cold gel supply) mixing in the vessel is more limited due to
analysis (not shown here) of the powders collected at the the higher density of the cold feed, which produces a layer
bottom of the autoclave after synthesis, revealed that an in- of denser solution that tends to stay near the bottom of the
duction period of 3 h was necessary for the development of autoclave.
LTA structure. Therefore, a synthesis period of 5 h was se- Table 1 summarizes the average autoclave temperatures
lected as the duration of the synthesis in this work. as well as the minimum and maximum temperature values at
Fig. 3A shows the evolution of the top and bottom temper- the top and bottom of the autoclave for different experimen-
atures in the autoclave for a given renewal rate (1/16 min−1 ) tal conditions (renewal rate, gel preheating). Again, a clear
when the feed was not preheated. A sharp temperature de- improvement of temperature control, especially at high re-
crease is observed at the bottom of the autoclave after every newal rates, is obtained with a preheated feed. Moreover, gel
fresh gel addition, reaching values as low as 37 ◦ C. A period preheating gives rise to membranes of a significantly higher
close to 15 min is needed for the temperature to recover its quality, as will be shown below by the results of pervapo-
original value at this location, and almost immediately a new ration experiments with water–ethanol mixtures. In spite of
addition of fresh gel takes place, meaning that a stable temper- this, inlet temperatures higher than 65 ◦ C were not used, to
ature is never reached at the bottom of the autoclave. Fig. 3B avoid synthesis occurring before the gel enters the autoclave,
shows the results of an analogous experiment with preheat- with the associated problems of reagent exhausting, and crys-
ing of the gel (to approximately 65 ◦ C) before entering the tallization and possibly plugging in the inlet duct.

Table 2
Some details of the synthesis procedure and characteristics of the NaA zeolite membranes prepared in this work
Sample Gel preheated Renewal rate (min−1 ) Zeolite loading (mg/g) N2 permeance (mol/m2 s Pa)
MC13 NO 1/13 5.2 6.8 × 10−8
MC16 NO 1/16 6.8 5.9 × 10−8
MC20 NO 1/20 6.6 6.6 × 10−8
MC25 NO 1/25 6.3 5.0 × 10−8
MC38 NO 1/38 5.6 1.4 × 10−8
MC75 NO 1/75 5.5 2.0 × 10−8
MH13 YES 1/13 8.1 1.2 × 10−8
MH16A YES 1/16 7.6 6.7 × 10−10
MH16B YES 1/16 9.0 1.5 × 10−9
MH20A YES 1/20 8.1 1.4 × 10−9
MH20B YES 1/20 6.9 3.8 × 10−10
MH20C YES 1/20 8.1 7.8 × 10−10
MH25 YES 1/25 8.0 1.3 × 10−9
MH38 YES 1/38 6.4 2.0 × 10−9
MH300a YES – 4.8 4.0 × 10−8
a Sample synthesized using the discontinuous approach, i.e., no renewals during the 5 h synthesis.
146 M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150

Fig. 4. SEM micrographs of sample: (A) MC20 top view and cross-section; (B) MH20 top view and cross-section; (C) MH13 top view and cross-section.

3.2. NaA zeolite membrane characterization ter of the NaA zeolite membrane, variable amounts of water
could remain adsorbed in the micropores when the perme-
Table 2 summarizes some of the main characteristics ance measurements were carried out, affecting the values
(gel preheating, renewal rate, zeolite loadings and nitrogen given. Therefore, the permeances given in Table 2 have to
permeance) for the membranes used in this work. For iden- be taken as lower limits (i.e., the nitrogen permeance with a
tification purposes, the samples synthesized without gel pre- completely water-free membrane could be higher), but they
heating were denoted using “MC”, whereas “MH” was used can still be used as a first indication of membrane quality
when the gel was preheated. Both, weight gain and nitro- since a significant concentration of defects would translate
gen permeance, were measured after drying the membrane into larger permeation fluxes. In general terms, higher zeo-
at 100 ◦ C overnight. This is only a mild drying treatment lite weight gains (6.0 mg/g versus 7.4 mg/g on average) and
and therefore it is likely that, due to the hydrophilic charac- lower values of nitrogen permeance (4.6 × 10−8 mol/Pa s m2
M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150 147

versus 7.4 × 10−9 mol/Pa s m2 on average) were detected Table 3


Water/ethanol separation performance in VP and PV of some of the NaA
when the gel was preheated. Additionally, sample MH300, zeolite membranes prepared in this work
which corresponds to a synthesis carried out batchwise (i.e.,
Sample Vapor permeation Pervaporation
with the same gel composition but without gel renewal), ex-
hibits permeance and weight gain values similar to those of Water flux Separation Water flux Separation
the “MC” series. The zeolite weight gains for the MH mem- (kg/h m2 ) factor (kg/h m2 ) factor
branes in Table 2 were on average 30% higher than those for MC38 0.89 362 2.9 371
the MC series, a result expected in view of the higher average MC20 0.81 1052 2.2 94
MH38 0.80 2863 3.8 163
temperatures in the autoclave. On the other hand, the nitro- MH25 0.80 839 3.1 1307
gen permeances measured for the MH membranes were one MH20B N/A N/A 3.8 3603
order of magnitude lower, indicating a lower concentration MH20C 1.0 3162 N/A N/A
of defects (a higher membrane quality) for the membranes MH16A 0.65 2736 2.5 1046
prepared with a preheated gel.
These observations are in good agreement with SEM anal-
ysis as can be seen in Fig. 4A and B, where MC20 and seeded surface seems to be less ordered, and the membrane
MH20B, both prepared with one gel renewal every 20 min, thickness achieved is only around 8 ␮m (8 microns), giving
are compared. The MC20 micrographs reveal the presence a higher permeation flux with respect to MH20B.
of amorphous material among cubic crystals of NaA (with Fig. 5 shows XRD patterns obtained from the analy-
a size around 5 ␮m (5 microns)) forming an irregular layer sis of the external surface of MH13, MH20B and MC20
with a thickness of ca. 15 ␮m (15 microns) on the external
surface of the alumina support. On the contrary, amorphous
material seems to be absent from the membrane MH20B top
layer, where two types of crystals can be distinguished: larger
cubic crystals 5–10 ␮m (5–10 microns) in size, and smaller
units (3 ␮m (3 microns) edge) covering the inter-boundary
gaps. These crystals form a homogeneous, well-intergrown
layer with a thickness around 10 ␮m (10 microns). Increas-
ing the renewal rate to 1/13 min−1 , i.e., MH13, renders a
base layer of cubic crystals around 2 ␮m (2 microns) in size
(see left part of Fig. 4C) coexisting with larger units (around
5 ␮m (5 microns) edge). Some of these larger crystals seem
almost detached from the continuous zeolite layer, suggesting
that they may have nucleated in the bulk of the liquid phase
(aided by the higher average concentration of reactants that
is obtained with a higher renewal rate) and then precipitated
onto the zeolite film. The growth of zeolite crystals over the

Fig. 6. (A) Water permeation fluxes and (B) water/ethanol separation factors
Fig. 5. XRD patterns of zeolite NaA membranes synthesized with the semi- as a function of gel renewal rate for NaA zeolite membranes synthesized by
continuous method under different conditions. The stars denote the reflec- the semi-continuous method, using preheated gel. PV conditions: 6 bar and
tions corresponding to the ␣-Al2 O3 commercial support. 125 ◦ C on the retentate side.
148 M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150

Table 4
Some examples extracted from the literature of pervaporation performance for different zeolite membranes
Zeolite Mixture (A/B) T (◦ C) Concentration of A (wt.%) Separation factor ␣(A/B) Flux (kg H2 O/m2 h) Support Reference
NaA H2 O/i-propanol 70 10 8500 0.82 Zirconia sheets [6]
NaA H2 O/ethanol 50 10 46 000 0.77 Mullite tubesa [7]
NaA H2 O/ethanol 75 10 42 000 2.08 Mullite tubesa [7]
NaA H2 O/ethanol 120 10 47 000 8.37 Mullite tubesa [7]
NaA H2 O/ethanol 105 10 30 000 4.53 Mullite tubesa [9]
NaA H2 O/ethanol 75 10 10 000 2.15 ␣-Al2 O3 tubes [10]
NaX H2 O/ethanol 75 10 410 0.98 Mullite tubes [11]
NaY H2 O/ethanol 75 10 130 1.59 Mullite tubes [11]
Silicalite Ethanol/H2 O 60 5 58 0.760 S.S. disks [12]
NaX H2 O/ethanol 75 10 400 0.5 ␣-Al2 O3 disks [21]
NaA H2 O/ethanol 75 10 23 000 2.5 ␣-Al2 O3 disks [21]
Mordenite H2 O/i-propanol 75 10 3360 0.1 ␣-Al2 O3 tubes [25]
Silicalite i-Propanol/H2 O 80 5 25 0.11 ␣-Al2 O3 disks [26]
ZSM-5 H2 O/I-propanol 80 5 501 0.14 ␣-Al2 O3 disks [26]
Mordenite H2 O/ethanol 150 10 210 1.1 ␣-Al2 O3 tubes [27]
NaA H2 O/ethanol 125 10 3603 3.80 ␣-Al2 O3 tubes This work
a Al2 O3 :SiO2 ratio equal to 65:35.

membranes. The comparison with the NaA zeolite pattern dehydration of solvents by PV can be explained by the so-
[23] shows that pure zeolite with the LTA-type structure is called adsorption-diffusion model [3,8,24]. This process in-
the only crystalline phase present in all the samples analyzed. volves three-steps: (i) adsorption of the components onto the
When the crystals collected at the bottom of the autoclave af- membrane, (ii) diffusion of the components across the mem-
ter synthesis were examined by XRD, NaA zeolite was again brane, and (iii), finally, desorption of the components into the
the only phase found, independent of the renewal rate em- vapor phase at the permeate side. Separation takes place be-
ployed. Finally, when the peak intensities of characteristic cause selective adsorption of water, both on the zeolite pores
diffractions of ␣-alumina and NaA (at 35.3◦ and 24.2◦ , re- and also on small defects between neighboring crystals, hin-
spectively) were compared, the ratio was around three times ders the permeation of the alcohol. This mechanism is consis-
higher for samples MH13 and MH20B than for MC20 (with- tent with the separation properties found for the membranes
out preheating). This is in agreement with the higher amount prepared in this work.
of crystalline material evidenced by the SEM micrographs Fig. 6A and B illustrate the influence of the renewal rate
for the membranes prepared with preheated gel. on the pervaporation performance (water flux and separation
factor) of membranes prepared with the preheated gel. A con-
3.3. Water-ethanol separation performance tinuous line was drawn across the average values obtained in
each case, while the dotted lines join the maximum and mini-
Table 3 shows the water/ethanol separation performance mum values registered for a given renewal rate. The beneficial
in VP and PV experiments of some of the membranes pre- effect of gel renewal is clear from the results of Fig. 6B: under
pared in this work. After an initial stabilization period of otherwise identical conditions, an H2 O/ethanol separation
approximately 2 h, the membranes showed a stable perfor- factor of only 54 was obtained in 5-hour batch syntheses with-
mance for the duration of the experiments (up to 10 h). In VP out gel renewal. Also, it can be seen that an optimum value
tests, H2 O/ethanol separation factors ranged between 362 and exists for the separation factor: at renewal rate of 1/20 min−1 ,
3162, and water fluxes between 0.65 and 1 kg/h m2 , depend- a clear maximum with a value of 3600 was obtained in the
ing on the gel preheating and the renewal rate employed. On H2 O/ethanol separation factor. The trend regarding the per-
the other hand, in PV experiments water fluxes at least double meation flux is less clear, but nevertheless the maximum av-
than those in VP tests were measured, as could be expected, erage value for the permeation flux (3.8 kg H2 O/h m2 ) was
given the higher driving force in pervaporation. Despite the also obtained at this renewal rate. SEM observations also in-
differences in nitrogen permeance shown in Table 2, the PV dicate that the best intergrowth and crystallinity correspond
and VP fluxes are in a relatively narrow range (similar results to the membranes prepared at 1/20 min−1 renewal rate.
were observed by Okamoto et al. [10]). The nucleation and growth of crystals in solution is of-
Table 3 also shows that samples prepared with a preheated ten put forward as an example of a complex phenomenon
gel usually give considerably higher separation factors, com- under the influence of many factors, such as variations of
pared to their cold-gel counterparts. This effect is clearer with reactant concentration in the solution around the crystals, lo-
VP measurements (3162 versus 1052 for M20 membranes, cal variations of temperature, degree of mixing, presence of
and 2863 versus 362 for M38 membranes, respectively). The impurities and existence of other surfaces capable of provid-
separation properties of hydrophilic zeolite membranes for ing additional nucleation sites. Growing zeolite crystals on a
M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150 149

porous support under conditions that favour their intergrowth is suitable for the scale up of zeolite membrane preparation
into a membrane adds an extra degree of complexity to the processes.
process. The procedure used in this work attempts to facili-
tate synthesis by minimizing the variations of concentration
in the bulk synthesis solution with time with respect to batch Acknowledgements
processing. Periodic gel renewal is also useful to improve
mixing in the synthesis vessel, reducing the concentration Financial support from MCYT, Spain, is gratefully ac-
gradients that always develop in the vicinity of growing crys- knowledged.
tals. However, the results obtained seem to indicate that there
is an optimum renewal rate, and increasing the renovation
rate beyond that value leads to a decrease in the quality of the References
membranes obtained. A major factor in this decrease may be
the nucleation and growth of crystals in the bulk of the syn- [1] J. Coronas, J. Santamarı́a, Separations using zeolite membranes, Sep.
Purif. Methods 28 (1999) 127.
thesis solution, aided by the increased gel concentration that [2] J. Caro, M. Noack, P. Kölsch, R. Schäfer, Zeolite membranes-state of
results from more frequent gel renewal. The crystals formed their development and perspective, Microporous Mesoporous Mater.
in the bulk grow to a certain size and eventually precipitate 38 (2000) 3.
onto the developing membrane, where they cannot be easily [3] X. Feng, R.Y.M. Huang, Liquid separation by membrane pervapo-
integrated, giving rise to the formation of defects. ration: a review, Ind. Eng. Chem. Res. 36 (1997) 1048.
[4] F.P. Cuperus, R.W. van Gemert, Dehydration using ceramic silica
Finally, Table 4 compares the PV separation properties pervaporation membranes-the influence of hydrodynamic conditions,
obtained for the membranes synthesized in this work (at Sep. Purif. Methods 27 (2002) 225.
1/20 min−1 renewal rate) with some of the best PV results [5] H. Kita, K. Korii, Y. Ohtoshi, K. Tanaka, K. Okamoto, Synthesis of
reported for zeolite membranes, extracted from a literature a zeolite NaA membrane for pervaporation of water/organic liquid
mixtures, J. Mater. Sci. Lett. 14 (1995) 206.
survey. While pervaporation results from different labora-
[6] J.J. Jafar, P.M. Budd, Separation of alcohol/water mixtures by per-
tories are often difficult to compare because of the differ- vaporation through zeolite A membranes, Microporous Mater. 12
ent experimental conditions used, it can be said that the (1997) 305.
membranes obtained with the semi-continuous system used [7] M. Kondo, M. Komori, H. Kita, K. Okamoto, Tubular-type perva-
in this work are among those with the best pervaporation poration module with zeolite NaA membrane, J. Membr. Sci. 133
(1997) 133.
performance.
[8] D. Shah, K. Kissick, A. Ghorpade, R. Hannah, D. Bhattacharyya,
Pervaporation of alcohol–water and dimethylformamide–water mix-
tures using hydrophilic zeolite NaA membranes: mechanisms and
4. Conclusions experimental results, J. Membr. Sci. 179 (2000) 185.
[9] Y. Morigami, M. Kondo, J. Abe, H. Kita, K. Okamoto, The first
large-scale pervaporation plant using tubular-type module with zeo-
A novel semi-continuous system for the preparation of
lite NaA membrane, Sep. Purif. Technol. 25 (2001) 251.
good-quality zeolite-A membranes has been developed, in [10] K. Okamoto, H. Kita, K. Korii, K. Tanaka, Zeolite NaA membrane:
which the synthesis gel is removed from the membrane en- preparation, single-gas permeation, and pervaporation and vapor per-
vironment at periodical intervals while at the same time an meation of water/organic liquid mixtures, Ind. Eng. Chem. Res. 40
equivalent amount of fresh gel is added to the synthesis ves- (2001) 163.
[11] H. Kita, H. Horita, K. Asamura, T. Tanaka, K. Okamoto, Pervapo-
sel. This provides a rational way to control the preparation
ration through faujasite membranes, Proceedings of the 5th Inter-
conditions by better approaching a constant concentration national Conference on Inorganic Membranes (ICIM’98), Nagoya,
during the synthesis process. The problem of temperature Japan, 1998, p. 536.
upset related to the introduction of a colder fluid in the syn- [12] T.C. Bowen, H. Kalipcilar, J.L. Falconer, R.D. Noble, Pervaporation
thesis vessel has been minimized by choosing a sufficiently of organic/water mixtures through B-ZSM-5 zeolite membranes on
monolith supports, J. Membr. Sci. 215 (1–2) (2003) 235.
high feed temperature that is nevertheless compatible with
[13] L. Casado, R. Mallada, C. Téllez, J. Coronas, M. Menéndez, J. San-
avoiding synthesis taking place outside the autoclave. tamarı́a, Preparation characterization and pervaporation performance
There is still considerable room for optimization of the of mordenite membranes, J. Membr. Sci. 216 (1–2) (2003) 135.
synthesis conditions. Thus, the feed concentration used has [14] H.P. Hsieh, Inorganic Membranes for Separation and Reaction, El-
been chosen within the interval of initial concentrations typi- sevier Science B.V, Amsterdam, 1996.
cally employed in successful batch syntheses of zeolite mem- [15] Y. Li, J. Wang, J. Shi, X. Zhang, J. Lu, Z. Bao, D. Yan, Synthesis
of ZSM-5 zeolite membranes with large area on porous, tubular
branes, but it must be taken into account that in these experi- ␣-Al2 O3 supports, Sep. Purif. Technol. 32 (1–3) (2003) 397.
ments a considerable depletion of reactants takes place. Keep- [16] M. Matsukata, E. Kikuchi, Zeolitic membranes-synthesis, properties
ing a nearly constant concentration in the vessel probably al- and prospects, Bull. Chem. Soc. Jpn. 70 (10) (1997) 2341.
lows the use of different (lower) feed concentrations and/or [17] A. Erdem-Senatalar, M. Tatlier, M. Ürgen, Preparation of zeolite
temperatures, leading to improved membranes. In spite of coatings by direct heating of the substrates, Microporous Mesoporous
Mater. 32 (1999) 331.
this, excellent results have been obtained in pervaporation ex- [18] T. Cetin, M. Tatlier, A. Erdem-Senatalar, U. Demirler, M. Ürgen,
periments (3.8 kg H2 O/h m2 with a H2 O/ethanol separation Lower temperatures for the preparation of thinner zeolite A coatings,
factor of 3603), indicating that the semi-continuous synthesis Microporous Mesoporous Mater. 47 (2001) 1.
150 M.P. Pina et al. / Journal of Membrane Science 244 (2004) 141–150

[19] S. Yamazaki, K. Tsutsumi, Synthesis of A-type zeolite membrane us- [24] T.C. Bowen, S. Li, R.D. Noble, J.L. Falconer, Driving force for
ing a plate heater and its formation mechanism, Microporous Meso- pervaporation through zeolite membranes, J. Membr. Sci. 225 (1–2)
porous Mater. 37 (2000) 67. (2003) 165.
[20] I. Kumakiri, Preparation and Permeation Mechanism of Zeolite [25] X. Lin, E. Kikuchi, M. Matsukata, Preparation of mordenite mem-
Membranes, Ph.D. Dissertation, The University of Tokyo (Japan), branes on ␣-alumina tubular supports for pervaporation of water-
2000. isopropyl alcohol mixtures, Chem. Commun. (2000) 957.
[21] H. Richter, I. Voight, P. Fischer, P. PuhlfürB, Preparation of zeolite [26] M. Noack, P. Kölsch, J. Caro, M. Schneider, P. Toussaint, I. Sieber,
membranes on the inner surface of ceramic tubes and capillaries, MFI membranes of different Si/Al ratios for pervaporation and
Sep. Purif. Technol. 32 (2003) 133. steam permeation, Microporous Mesoporous Mater. 35–36 (2000)
[22] F. Tiscareño-Lechuga, C. Téllez, M. Menéndez, J. Santamarı́a, A 253.
novel device for preparing zeolite-A membranes under a centrifugal [27] A. Navajas, R. Mallada, C. Téllez, J. Coronas, M. Menéndez, J.
force field, J. Membr. Sci. 212 (1–2) (2003) 135. Santamarı́a, Improvements on pervaporation performance using post-
[23] http://www.iza-structure.org. synthetic treatments on mordenite membranes, in preparation.

You might also like