On the effect of morphological features on the properties of MFI zeolite membranes

On the effect of morphological features on theproperties of MFI zeolite membranes

M.P. Bernal, J. Coronas, M. Men�eendez, J. Santamar�ııa *

Department of Chemical and Environmental Engineering, Faculty of Science, University of Zaragoza, 50009 Zaragoza, Spain

Received 19 June 2002; received in revised form 10 March 2003; accepted 12 March 2003

Abstract

MFI type zeolite membranes were prepared on alumina and stainless steel tubular supports by different synthesis

procedures, giving rise to different zeolite layer structures and distributions of zeolite material with respect to the

support. This was used as a base to establish a classification of zeolite membranes. SEM and EPMA analysis showed

clear differences among different types of MFI zeolite membranes, concerning the morphology and location of the

zeolite deposits. Three types of membranes were identified, namely: type-A membranes, in which the zeolitic material

was located mainly inside the support pores; type-B membranes, with a thin layer of randomly oriented crystals on top

of the support; and type-C corresponding to MFI c-oriented membranes. These morphological differences translated

into diverse qualitative and quantitative behavior of the corresponding membranes, as shown by the evolution of single-

gas permeances and separation selectivity with temperature, and by adsorption and temperature-programmed de-

sorption experiments.

� 2003 Elsevier Science Inc. All rights reserved.

Keywords: Zeolite membrane; MFI; Membrane characterization; Adsorption properties; Butanes separation

1. Introduction

Zeolites are silicate or alumino-silicate crystal-

line materials with a very regular and microporous

structure (pore size from 0.3 to 1 nm) [1,2]. Their

small pore size and the possibility of having dif-

ferent chemical compositions (Si/Al ratio, ex-

changed cations) in their framework can give rise

to a very selective interaction with adsorbed mol-

ecules depending on their size, shape and chemi-

cal characteristics. During the last ten years the

interest in developing thin zeolitic films or zeolitemembranes has grown enormously [3,4] driven

mainly by the fact that they can combine sepa-

ration [5–9] and catalysis [10] over a broad tem-

perature range. Most of the zeolite membranes

reported in the literature are of the MFI type,

which includes silicalite-1 and ZSM-5.

A wide variety of synthesis methods for MFI

zeolite membranes has been reported. The meth-ods do not only differ with respect to the synthe-

sis gel compositions and the supports employed,

but also with respect to the procedures used to

bring into contact the support and the synthesis

solution. It is clear that the preparation method

* Corresponding author. Tel.: +34-976-76-11-53; fax: +34-

976-76-21-42.

E-mail address: [email protected] (J. Santamar�ııa).

1387-1811/03/$ - see front matter � 2003 Elsevier Science Inc. All rights reserved.

doi:10.1016/S1387-1811(03)00331-7

www.elsevier.com/locate/micromeso

Microporous and Mesoporous Materials 60 (2003) 99–110

mail to: [email protected]

dramatically influences the characteristics of the

zeolite membrane formed. Ramsay and Kallus

[11], as well as other authors [12], pointed out that

the distribution of the zeolite on the support, apart

from influencing the mechanical strength and de-

fect properties of the membrane, may also causedifferences in gas separation behavior. Such effects

have been ascribed to differences in the micro-

structure and inter-connectivity of the crystalline

zeolite layers. Coronas et al. [12] observed a

maximum in the permeation of n-butane as a

function of temperature on ZSM-5 membranes

when zeolite crystals were present inside the sup-

port pores. However, they did not find a maximumwhen a layer of the same zeolite was formed on top

of the support, and there were no crystals inside.

Also, the temperature at which the maximum in

the n/i-butane separation was reached shifted to

higher temperatures (135–137 �C) for the mem-

branes that had their material mainly inside the

support pores, while it was at 92–110 �C for those

with a zeolite layer on top of the support. Pieraet al. [13] reported similar results for the gas phase

separation of alcohol/O2 mixtures using both sili-

calite-1 and ZSM-5 membranes.

The references in the previous paragraph [9–13]

belong to the so-called directly synthesized mem-

branes, in which hydrothermal synthesis takes

place with the support immersed into the synthesis

gel, without a previous seeding step. Conversely, inseeded syntheses of zeolite membranes (also

termed secondary synthesis) the aim is to separate

the crystal nucleation and growth steps, in order to

obtain a better control of the film microstructure.

The zeolite membranes prepared by secondary

(seeded) growth typically exhibit a structure where

the constituent crystals are preferentially oriented

in a crystallographic direction (in general, the ‘‘c’’direction). The behavior of these membranes is

clearly different from those prepared by the direct

synthesis method. Thus, with oriented membranes

it is generally found [14,15] that the temperature

at which the maximum selectivity occurs for a

given separation is lower than for the randomly-

oriented directly-synthesized membranes. Further-

more, thin membrane films grown on top of aporous support are capable of high permeation

fluxes, as shown recently by Hedlund et al. [16].

These authors blocked the support pores to avoid

penetration of the synthesis gel, and then pro-

ceeded to synthesize a thin (0.5 lm) silicalite-1 film

with an exceptionally high gas permeation using

secondary synthesis on 50 nm seeds. This allowed

them to achieve high-temperature, high-pressureseparation of xylene isomers, something that had

not been achieved with directly synthesized mem-

branes.

The above discussion strongly suggests that the

location and morphology (including orientation)

of the crystalline deposits play a key role in de-

termining the performance of a zeolite membrane.

In a previous review work [17] two categories weredistinguished regarding the distribution of the

zeolitic material on directly synthesized MFI

membranes: type-A membranes, with the zeolitic

material preferentially deposited inside the porous

structure, and type-B membranes, where most of

the zeolite exists as a thin layer on top of the po-

rous support. In this work a third category of MFI

zeolite membranes is added, the so-called type-Cmembranes, where the zeolite material forms a

columnar oriented layer [14,15,18], without zeolite

in the pores of the support. Our hypothesis is that

membranes belonging to the different types will

display a different behavior, even if they are of

comparable quality and contain the same zeolitic

material. The aim of this work was to find addi-

tional evidence to support this statement, leadingto the establishment of a morphology-based clas-

sification of MFI zeolite membranes.

2. Experimental

A variety of synthesis methods and supports has

been used in this work in order to obtain a widevariation of membrane features. The ZSM-5

membranes used in this work were prepared on

commercial supports of alumina (SCTe, 5 or 200

nm pores in the separation layer) and stainless

steel (Motte, 500 nm effective filtration size, 8–10

lm actual size of pores), respectively. The tubes

used had a permeable length between 4 and 5 cm,

and internal and external diameters of 7 and 10mm, respectively. Na-ZSM-5 membranes were

prepared using the gel described in [12]. Its molar

100 M.P. Bernal et al. / Microporous and Mesoporous Materials 60 (2003) 99–110

composition was: 21SiO2:987H2O:3NaOH:

1TPAOH:0.105Na2Al2O4, where TPAOH stands

for tetrapropyl ammonium hydroxide, the struc-

ture-directing agent. ZSM-5 membranes without

Na in their structure (H-ZSM-5) were synthesized

using the following gel [19]: 19.46SiO2:438H2O:1TPAOH:0.0162Al2O3. A third type of ZSM-5

membranes, namely boron-substituted ZSM-5

membranes were prepared with the following

gel [20]: 19.46SiO2:438H2O:1.55TPAOH:0.195B-

(OH)3. For the silicalite-1 membranes prepared by

direct hydrothermal synthesis, a gel without alu-

minum but otherwise with a similar composition

as for Na-ZSM-5 membranes was used [21].Four synthesis methods were used in this work.

Three of these procedures were employed to pre-

pare membranes by direct hydrothermal synthesis:

(1) A support tube, with both ends wrapped with

Teflon tape, was introduced into the autoclave

that was then filled with the gel. The synthesis

was then carried out with the synthesis gel incontact with both the inside and outside of

the support tube.

(2) Before the introduction of the support in the

autoclave, one end of the wet tube was

wrapped with Teflon tape, plugged with a Tef-

lon cap and filled with the synthesis gel; the

other end was then wrapped with tape and

plugged with another Teflon cap. The synthe-sis was then started with the inside of the tube

full of synthesis gel. This method was em-

ployed with c-Al2O3 supports.

(3) A combination of the two previous methods

was used for membranes supported on stain-

less steel tubes. The first synthesis was similar

to procedure (1) but with the autoclave rotat-

ing horizontally and the support wrapped withTeflon tape on the outside. After the first syn-

thesis, the following synthesis cycles were car-

ried out with procedure (2).

Alternatively, silicalite-1 membranes were pre-

pared by the seeded synthesis method [15], which is

the fourth method (4). In this case, silicalite-1

nanocrystals (ca. 150 nm) were first obtained in aseparate hydrothermal synthesis at 125 �C for 8 h

using a solution with the following molar compo-

sition: 10SiO2:110H2O:1NaOH:2.4TPAOH. The

inner side of the tube was dip-coated three times

with a 20 g/l suspension of these crystals, and then

secondary growth was performed using procedure

(1) but after wrapping the tube outside with Teflon

tape. In this case, the gel used had the followingmolar composition: 10SiO2:280H2O:0.44NaOH:

0.5TPABr.

In all cases, to prepare the membranes the au-

toclave was placed in a convection oven at 170 �Cbetween 8 and 72 h. The synthesis was repeated

until a membrane impermeable to N2 under a

pressure gradient of 1.5 bar was obtained. Then

the template was removed by heating up to either440 or 480 �C under a ramp of 1 �C/min and

maintaining at 480 �C for 8 h. Table 1 shows some

of the main properties and the synthesis procedure

for each of the membranes used in this work.

The membranes were tested for the separation

of n/i-butane binary mixtures. Mass flow control-

lers (Brooks Instruments B.V.) were used to feed

continuously a 100 cm3 (STP)/min of an equimolarmixture of the butanes into the tube side (reten-

tate) of the membrane, while the other side (per-

meate) was swept with a 100 cm3 (STP)/min stream

of He. Atmospheric pressure was maintained on

both sides of the membrane. Samples at the exit of

both the permeate and retentate streams were an-

alyzed by on-line gas chromatography (Varian

3400). The permeances were calculated using thelog mean partial pressure difference, and the sep-

aration selectivity as the ratio of permeances.

Some adsorption experiments were carried out

on silicalite-l and ZSM-5 membranes and on zeo-

lite powder to investigate preferential adsorption

on the different membrane types. To this end, the

freshly calcined membrane (or the zeolite powder)

was placed in a glass module and saturated at 27.5�C under either pure n-butane or an equimolar n/i-

butane mixture. Both sides of the membrane were

exposed simultaneously to the same adsorbable

mixture. Once steady state was reached in the

composition of the gas streams in contact with

both membrane sides (the gas composition was

continuously monitored by means of a data ac-

quisition system based on a TCD cell), the butanesflux was stopped and the membrane was flushed

for 200 s with an N2 stream (50 cm3 (STP)/min) to

M.P. Bernal et al. / Microporous and Mesoporous Materials 60 (2003) 99–110 101

remove gas phase hydrocarbons from the mem-brane environment. After these first 200 s, the

temperature of the membrane module was in-

creased to 500 �C at either 1 or 2 �C/min, while

keeping the N2 feed. All gases exiting the mem-

brane module (N2 sweep plus desorbed hydrocar-

bons) were collected in a sealed container and

analyzed by gas chromatography in order to de-

termine the amount of each of the butane isomersdesorbed from the membrane.

A JEOL JSM-6400 electron microscope was

used for taking the SEM photographs, and EPMA

analysis was used to obtain radial Si/Al profiles for

membranes of different types. A temperature-pro-

grammed permeance device described elsewhere

[22] was also used for the measurement of propane

permeance.

3. Results

3.1. Morphology and composition

Fig. 1 shows a scheme of the structure and SEM

photographs (top view and cross-section) of the

three different types of MFI membranes consid-ered in this work. The type-A and type-B mem-

branes come from direct hydrothermal synthesis

and, consequently, the top view of these films is

rougher, showing the presence of non-integrated

material that nucleates homogeneously and then

precipitates onto the growing zeolite layer [23].

Heterogeneous nucleation occurs on the support,

and the crystals grow initially in all directions.Because of this and also due to the material pre-

cipitated from the bulk of the solution it is often

impossible to observe the existence of a preferen-

tial orientation in the layers. This contrasts sharply

with the appearance of the type-C membranes,

where the preferential c-orientation is evident in

the zeolite layer. The preferential c-orientation of

the seeded, type-C membranes was confirmed byXRD analysis, as shown in Fig. 2, where the XRD

patterns of type-B and type-C membranes are

compared. As already mentioned the origin of the

columnar, c-oriented layer is the seeding of the

support with zeolite nanocrystals. These grow

preferentially towards the bulk of the solution

(where the nutrients are), giving rise to a well-

inter-grown layer, that can be appreciated in the

Table 1

Some properties of the membranes used in this work

Membrane Support Synthesis time

[h]

Synthesis

procedure

N2 Permeance

[mol/(m2 s Pa)]

Ideal N2/SF6

selectivity

SIL-1 c-Al2O3 8 2 7.2� 10�7 5.5

SIL-2 c-Al2O3 8/15 2 7.2� 10�7 99.4

SIL-3 a-Al2O3 72 1 1.1� 10�6 12.0

SIL-4 a-Al2O3 72 1 9.3� 10�8 –

SIL-5 Stainless steel 15/8/15/8 3 3.3� 10�7 19.1

SIL-6 c-Al2O3 8/15/8/15/8 2 1.8� 10�6 36.7

Sec-SIL-1 a-Al2O3 20/20 2ary growth 1.1� 10�6 8.9


Sec-SIL-3 a-Al2O3 20 2ary growth 1.7� 10�7 8.0



Na-ZSM-5-1 a-Al2O3 15/8/15 1 1.5� 10�7 30.4

Na-ZSM-5-2 c-Al2O3 15/8/15 1 2.7� 10�7 4.0

Na-ZSM-5-3 Stainless steel 8/15/8 1 1.8� 10�7 32.6

H-ZSM-5-1 a-Al2O3 72 1 1.8� 10�7 6.0

H-ZSM-5-2 a-Al2O3 24/24 1 1.6� 10�7 21.7

H-ZSM-5-3 c-Al2O3 15/8 2 6.8� 10�7 59.8

B-ZSM-5-1 Stainless steel 24/24/24 3 5.2� 10�8 45.0


top view for this membrane. Note that for the

type-A membrane there is an important amount of

zeolite crystals inside the pores of the support (see

insert in the microphotograph corresponding to

the cross-section of type-A membranes). In con-trast, these were absent from type-B and type-C

membranes (details of the support pores not

shown for these membranes).

The distribution of zeolitic material in the sup-

port obtained with the different synthesis proce-

dures was further investigated by EPMA analysis

on cross-sections of the membranes used. Fig. 3

compares the evolution of the Si/Al atomic ratioacross the membrane for an H-ZSM-5 membrane

synthesized by procedure (1), a silicalite-1 mem-

brane prepared by procedure (2) and a silicalite-1

membrane prepared by secondary (seeded) growth.

Since the support is alumina in all cases and Si is

only introduced during the synthesis, the increase

of the Si/Al ratio corresponds to the presence of

zeolite deposits.For the membrane H-ZSM-5-1 (type-A mem-

brane) the first part of the profile in Fig. 3 (Si/Al

ratio around 100) corresponds to the continuous

5 10 15 20 25 30 35 40

(200)

(101)

Type C membrane

Type B membrane

Inte

nsity

[a.u

.]

2Θ [°]

Fig. 2. Comparison of XRD patterns for type-B and type-C

membranes.

Fig. 1. Scheme of the different types of MFI zeolite membranes. Samples of cross-sections and top view of each type of membrane.

Type-A membrane: H-ZSM-5-2, type-B membrane: Na-ZSM-5-2, type-C membrane. Sec-SIL-2. The picture at the left for type-A

membrane contains an insert showing part of the cross-section in greater detail.


layer of zeolite on top of the porous support. Inthe next section the Si/Al ratio decreases sharply,

passing through a minimum (Si/Al¼ 0.027 at

x ¼ 70 lm), then increasing to an approximately

constant level (Si/Al¼ 1.6). The variation of the Si/

Al atomic ratio inside the support (i.e., beyond the

continuous layer where the Si/Al ratio of 100 was

observed) can be explained by taking into account

the porous structure of the support and the modeof contact used during synthesis. In procedure (1),

the synthesis gel penetrates from both sides of

the membrane. However, the inner side has the

smallest pores (200 nm, followed by an interme-

diate layer with pores of 800 nm) and penetration

of the viscous synthesis gel from this side of the

support is limited, causing the minimum observed

in the Si/Al ratio. Conversely, the outer side haspores of 12 lm that facilitate the penetration of the

synthesis gel, filling the pores of the support with

H-ZSM-5 crystals and giving rise to the high Si/Al

ratio observed.

The evolution of the Si/Al ratio is different for

the silicalite-1 membrane (type-B membrane

termed SIL-1 in Fig. 3). Again, a rapid drop can be

observed after the zeolite layer, which is consid-erably thinner in this case compared to membrane

H-ZSM-5-1. However, after the drop the Si/Al

ratio stays at a low level, comparable to the min-

imum value observed for the previous membrane.

This is due to the synthesis method used (proce-

dure (2)), which only puts into contact the inner

support side (with small pores at the interface)

with the synthesis gel. This is effective in prevent-

ing a deep penetration of the synthesis gel inside

the membrane pores, and explains why zeolite

deposits were not observed deep inside the support

pores. A similar pattern of the Si/Al profile wasfound for the c-oriented silicalite-1 membrane

(type-C), also because the synthesis procedure

prevented contact between the outside of the

membrane tube and the synthesis gel.

3.2. Permeation measurements

Fig. 4 shows the propane permeance for differ-ent MFI membranes: Na-ZSM-5-1, SIL-2, B-

ZSM-5-1 and Sec-SIL-2, which were synthesized

by procedures (1)–(3) and seeded secondary

growth, respectively. Because synthesis procedures

(2) and (3) give rise to membranes with a contin-

uous zeolite layer on the porous support mem-

branes SIL-2 and B-ZSM-5-1 can be considered as

type-B. On the other hand, Sec-SIL-2 is a type-Cmembrane, while Na-ZSM-5-1 prepared by pro-

cedure (1) is a type-A membrane, with a large

amount of zeolite material inside the support

pores.

The permeance of adsorptive gases on zeolite

membranes follows the trend with increasing

-100 -50 0 50 100 150 250 500 750 1000

0.01

0.1

1

10

100

Sec-SIL-1, type C

SIL-1, type B

H-ZSM-5-1, typeA

Si/A

l Ato

mic

Rat

io

Depth from the Inner Surface [µm]

Fig. 3. EPMA diagram of three different MFI zeolite mem-

branes.

50 100 150 200 250

10-7

10-6

Sec-SI_2,type C

B-ZSM-5-1,type B (on SS support)

SIL-2, type B (on alumina support)

Na-ZSM-5-1, type A

Pro

pane

Per

mea

nce

[mol

/(m2 ·s

·Pa)

]

Temperature [°C]

Fig. 4. Temperature-programmed permeation (TPP) of pro-

pane for several MFI zeolite membranes. Feed: 100 cm3 (STP)/

min. Permeate at room pressure. DP ¼ 1 bar. No sweep gas was

used.


temperature described elsewhere [17,24,25]: in the

range of low temperatures, the permeance in-

creases with temperature, driven by the increased

mobility of adsorbed species, even though the

amount of adsorbed material starts decreasing.

Eventually, a maximum in permeance is reached,and above this temperature the decline in occu-

pancy prevails which gives rise to a decrease in

permeance. At a certain temperature level a local

minimum is observed in the permeation curve.

This corresponds to the point where adsorption

effects become small enough so that the permeance

is controlled by activated transport through the

micropores, increasing monotonically with tem-perature. In a recent work [22], it was shown that

the position of the maximum and minimum in the

permeance–temperature diagram can be used to

rank adsorption effects in MFI membranes: a

stronger adsorption would shift both the maxi-

mum and the minimum in the diagram towards

higher temperatures.

The permeance of the Na-ZSM-5-1 membrane(type-A) exhibits a wide maximum in the 175–185

�C interval, and the minimum of the diagram is

not even reached within the range explored (up to

240 �C) indicating a strong influence of adsorp-

tion. In contrast, membrane B-ZSM-5-1 (type-B)

presents only a shallow maximum at a much lower

temperature (ca. 100 �C) and a minimum at about

175 �C, both facts indicating weaker adsorptioneffects. The same situation is observed for SIL-2

(also a type-B membrane), where the maximum is

not observed, and the minimum appears in the

100–120 �C interval. Finally, the oriented mem-

brane (Sec-SIL-2, type-C) presents a smooth

minimum in its permeation–temperature diagram,

followed by a strong activation from 150 �C on-

wards.The above results seem to indicate that ad-

sorption effects increase for membranes with zeo-

litic material inside the support pores, as observed

for the type-A membrane of Fig. 4. Interestingly,

in this figure the membrane that occupies the next

position in terms of adsorption effects is membrane

B-ZSM-5-1. While this was prepared using pro-

cedure (3) and therefore it is nominally a type-Bmembrane, it might be noted that a porous stain-

less steel support was used. In this case, the sup-

port pores measured by Hg porosimetry are close

to 10 lm [26], i.e., 50 times larger than those of the

SIL-2 membrane. This means that penetration of

the synthesis gel inside the support pores cannot be

avoided, and a significant amount of zeolite ma-

terial is to be expected there, increasing the ad-sorption effects.

Figs. 5 and 6 show the variation of n-butane and

i-butane permeances and of the n/i-butane sepa-

ration selectivity with temperature for different

membranes: type-A (H-ZSM-5-2), type-B (H-

ZSM-5-3) and type-C (Sec-SIL-3). Krishna and

co-workers [27] demonstrated that linear and

branched alkanes present a different adsorptionpattern on the silicalite-1 pore network, with

branched alkanes predominantly located at the

channel intersections and linear alkanes distrib-

uted more evenly. This makes it possible for n-

butane to effectively hinder the access of i-butane

0 50 100 150 200 250

10-9

10-8

10-7

H-ZSM-5-2, type A

H-ZSM-5-2, type A

H-ZSM-5-3, type B

H-ZSM-5-3, type B

Sec-SIL-3, type C

Sec-SIL-3, type C

i-butane

Per

mea

nce

[mol

/(m2 ·s·

Pa)]

Temperature [°C]

10-8

10-7

n-butane

Fig. 5. n-Butane and i-butane permeances as a function of the

temperature for each type of MFI zeolite membrane. Feed: 100

cm3 (STP)/min of an equimolar n-butane/isobutane mixture.

Permeate: 100 cm3 (STP)/min of He. Pressure at retentate and

permeate side: 1 bar.


to the zeolite pore network and, as a consequence,

the separation of these butane isomers at low to

moderate temperatures is often quoted as an ex-

ample of adsorption-governed separation [17,27].

At low temperatures, the membrane micropores

are mainly occupied by the preferentially adsorbed

n-butane, blocking the entrance of i-butane, and ahigh selectivity is reached (the maximum selectivity

for membranes H-ZSM-5-3 and Sec-SIL-3 in Fig.

6 was reached at about 25 and 50 �C, respectively).

As the temperature increases, the diffusivity of

both components increases, while the pore occu-

pancy diminishes due to the weakening of ad-

sorption at higher temperatures. Above a certain

temperature level, the blockage of the pores by n-butane will no longer be significant, and a sharp

drop in selectivity can be expected. This is clearly

observed in Fig. 6 for type-B and type-C mem-

branes, where a rapid drop in selectivity above

100–120 �C can be observed. The same behavior

was reported for the type-B membranes of Tuan

et al. [19,20] and Coronas et al. [12] prepared by

identical procedures. In contrast, the type-Amembrane H-ZSM-5-2 in Fig. 5 reached its max-

imum n/i-butane selectivity at about 130 �C, (i.e.,

80–100 �C higher than the B/C membranes) and

the selectivity at 185 �C was still around six, i.e.,

two to three times the value obtained for type-B

and type-C membranes at this temperature. The

fourth curve included in the graph (SAC-SIL)

corresponds to data previously obtained in our

laboratory by Alfaro et al. [28] with silicalite-1

membranes synthesized by steam-assisted crystal-

lization. Electron microscopy observations con-

firmed that, while a continuous zeolite layer on topof the support was not formed, the preparation

method used gave rise to a large quantity of

crystals within the support pores, i.e., a true type-

A membrane had been obtained. It can be seen

that the range of selective operation for this

membrane is even broader, being able to maintain

an n/i-butane selectivity of 8.9 at 230 �C.

3.3. Adsorption of n/i-butane mixtures

Adsorption experiments under a 50/50 kPa n/i-

butane mixture were performed on membranes of

different types and on zeolite powder, as described

in the experimental section. Table 2 shows that for

type-A membranes (SIL-3 and SIL-4) the ratio

between the amounts of n-butane and i-butaneadsorbed was in the 1.64–2.03 interval. When the

inner side of these membranes was scrubbed with

sand paper to remove the zeolitic material existing

on top of the support, the n/i-butane adsorbed

ratio clearly increased (from 2.03 to 3.43 and from

1.64 to 2.14, respectively). This indicates a higher

n/i-butane adsorption selectivity of the zeolitic

material inside the support pores. The type-Bmembranes (SIL-5, Na-ZSM-5-2 and Na-ZSM-5-

3) showed a considerably lower n/i-butane ratio in

the material adsorbed (in the 1.20–1.49 range). In

fact, the ratio for these membranes was close to the

values obtained for silicalite-1 and ZSM-5 pow-

ders (1.19–1.25). Finally, the oriented membranes

Sec-SIL-4 and Sec-SIL-5 prepared by secondary

growth showed n/i-butane ratios of 1.58 and 1.71,respectively, which is intermediate between the

values found for type-A and type-B membranes.

3.4. Temperature-programmed desorption experi-

ments

To have a direct assessment of the strength of n-

butane adsorption in the different types of mem-branes some temperature-programmed desorption

(TPD) experiments were carried out after satura-

0 50 100 150 200 2500

5

10

15

20

25

30

35

SAC-SIL, type A

H-ZSM-5-3, type B

H-ZSM-5-2, type A

Sec-SIL-3, type C

n/i-

Bu

tan

e S

elec

tivi

ty

Temperature [°C]

Fig. 6. n/i-Butane separation selectivity as a function of the

temperature for each type of MFI zeolite membrane. Same

conditions as in Fig. 5 except for the curve (SAC-SIL) corre-

sponding to data previously obtained in our laboratory by

Alfaro et al. [28] with type-A silicalite-1 membranes.


tion under a pure n-butane stream. The results are

shown in Fig. 7. It can be seen that n-butane de-

sorption starts at a similar temperature in all

membranes (ca. 40 �C). However, while in all theother membranes the maximum in the desorption

curve appears at temperatures of 64–65 �C, for the

type-A membrane it appears at a clearly higher

temperature (80.5 �C) indicating a stronger ad-

sorptive interaction.

4. Discussion

The above results show several distinctive fea-

tures of the diverse types of membranes investi-gated. First of all, the distribution of zeolitic

material is clearly different, with most of the zeolite

deposits inside the support pores for type-A

membranes, while these exist as a layer on top of

the support for type-B and type-C membranes.

Regarding adsorption, the behavior is also differ-

ent, with type-A membranes apparently showing

the strongest specific interaction, as indicated bythe position of the maxima and minima in the

temperature-programmed permeation curves and

in the temperature-programmed desorption ex-

periments shown in Fig. 7. Furthermore, the in-

fluence of adsorption phenomena in type-A

membranes extends into the higher temperature

range, and these membranes are able to perform

selective n/i-butane separations even at tempera-tures in the vicinity of 200 �C. Finally, while the

adsorption of linear paraffins in MFI zeolites is

generally favored over that of branched hydro-

carbons of the same carbon number, this process is

enhanced in type-A membranes. Thus, it has been

demonstrated that the ratio between the amounts

of n- and i-butane adsorbed on type-A membranes

50 75 100 1250.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

TP

D N

orm

aliz

ed S

ign

al

Temperature [°C]

SIL-3, type A

SIL-7, type Bon alumina support

SIL-5, type B on SS support

Sec-SIL-4,type C

Fig. 7. Temperature-programmed desorption (TPD) of n-

butane normalized to the zeolite weight and to the maximum

signal value for membranes SIL-3 (type-A), SIL-5 (type-B),

SIL-7 (type-B) and Sec-SIL-4.

Table 2

Results of desorption experiments carried out after adsorption of an equimolar n/i-butane mixture at room temperature and atmo-

spheric pressure on different types of MFI membranes

Membrane Type n/i-Butane molar ratio

in the desorbed material

Estimation of zeolite weight

on the membrane [mg]a

SIL-3 A 2.03 545

SIL-3b A 3.43 –

SIL-4 A 1.64 425

SIL-4b A 2.14 –

SIL-5 B 1.49 230

Na-ZSM-5-2 B 1.23 160

Na-ZSM-5-3 B 1.20 146

Sec-SIL-4 C 1.58 215

Sec-SIL-5 C 1.71 234

Sec-SIL-5b C 1.55 –

Silicalite-1c – 1.25 500

ZSM-5c – 1.19 500

a Calculated as the weight increase after the synthesis. There could be two main sources of error in this estimation of zeolite content:

the partial dissolution of the support during the synthesis and the presence of amorphous material.b After sandpapering the inner side of the membrane to eliminate zeolitic material outside the support pores.c 0.5 g of silicalite-1 or ZSM-5 powder prepared with the same procedures and gels used for membrane synthesis.


is higher than on type-B membranes and even than

on the pure MFI crystals.

Several factors might contribute to the above

experimental results.

(i) It seems clear that the different preparation

methods may produce different relative amountsof amorphous and crystalline material in the zeo-

lite membranes prepared. Thus, in principle,

membranes with a thin layer on top of the support

or membranes prepared by a seeded synthesis

would be expected to have a large proportion of

crystalline material. On the other hand, in type-A

membranes it is conceivable to expect a slower

reaction of the material inside the support poresand therefore a larger proportion of amorphous

material at the end of a given synthesis period.

However, the comparison of XRD patterns (not

shown) did not indicate significant differences in

the amorphous halo between the membranes pre-

pared, and therefore clear evidence regarding a

higher proportion of amorphous material in type-

A membranes does not exist. Furthermore, itseems unlikely that a zeolite with a larger pro-

portion of amorphous material would give a

stronger adsorption, or show an enhanced prefer-

ence for n-butane versus i-butane.

(ii) It could also be speculated that thermal

stress (due to the different expansion coefficients of

the support and the zeolite top layer) gives rise to

the appearance of some inter-crystalline defectsthat are at least partly responsible for the fast

decrease of selectivity above 130 �C in type-B and

type-C membranes. In this case, type-A mem-

branes where the zeolite material exists mainly as

aggregates within the support pores rather than as

a continuous top layer, would be at an advantage,

and a more robust behavior with temperature

could be expected. However, while a higher ther-mal robustness for type-A membranes is a rea-

sonable assumption, this would not explain the

displacement of maxima and minima in tempera-

ture-programmed permeation, or the enhanced

preference for n-butane adsorption of the zeolite

material in these membranes.

(iii) Third, it is also possible, that the different

behavior is related to differences in the molar Si/Alratio of the membranes prepared. It is well known

that when MFI membranes are prepared on alu-

mina supports, Al can be leached and incorpo-

rated into the MFI framework, due to the highly

alkaline conditions prevailing during synthesis.

Different preparation methods could therefore

lead to different amounts of Al leached [29], and

thus to a different chemical composition of thezeolite film. This would be particularly important

for type-A membranes, in which the deep pene-

tration of the gel inside the support would provide

a larger leaching interface: thus, a higher Al con-

tent could be invoked to explain the differences

observed for type-A membranes in the adsorption

of C4 hydrocarbons. This hypothesis, however,

also fails to explain some of the results obtained.For instance, the reference silicalite-1 and ZSM-5

samples in Table 2 show very similar ratios of

adsorbed n- and i-butanes. The same applies to the

type-B samples Na-ZSM-5-2 and Na-ZSM-5-3, in

spite of the fact that one was prepared on c-alu-

mina as support and the other on stainless steel.

Similarly, SIL-5 and SIL-7 are type-B membranes

prepared, respectively, on stainless steel and alu-mina supports, and they also display the same

temperature for the maximum of the TPD diagram

(Fig. 7), indicating a similar adsorption strength.

(iv) A fourth possible explanation relates to the

microstructure of type-A membranes. The prepa-

ration method involves nucleation and growth

of zeolite crystals inside the support pores, leading

to simultaneous growth in multiple directionswithin a confined environment. It seems reason-

able to expect that this would lead to a different

density and distribution of non-zeolitic pores (i.e.,

inter-crystalline voids) and/or grain boundaries,

which seem to play an important role in the sep-

aration. The fact that the internal and external

surfaces of zeolite crystals may have different ad-

sorption (and catalytic) properties has been rec-ognized for a long time. Derouane [30,31] showed

that the physisorption energies of sorbates on zeo-

lites are directly related to confinement effects, and

also concluded that molecular shape selective

effects are not restricted to the intra-crystalline

volume of zeolites, but occur significantly on their

external surfaces. Given the significant differences

in the degree and type of confinement betweenintra-crystalline and inter-crystalline pores, it seems

reasonable to expect a different adsorption be-


havior for type-A membranes, where the synthesisprocedure gives rise to a large proportion of inter-

crystalline pores. Additionally, the presence of

very small and tortuous inter-crystalline voids in

the support pores would help to dissipate part of

the thermal stress generated upon heating, and to

maintain the selectivity at higher temperatures. It

is interesting to note that the oriented type-C MFI

membranes exhibit intermediate n/i-butane ad-sorption ratios. As it can be inferred from the

SEM photographs of these oriented membranes

and from the scheme of Fig. 1, the columnar

crystals grow from the seeds and eventually inter-

grow to form a continuous layer. However, close

to the bottom (interface support-zeolite) inter-

crystalline voids clearly exist. The cross-section of

these voids decreases in the vertical direction, untila size is reached that could also contribute to the

preferential adsorption of n-butane, in the same

way as for type-A membranes.

5. Conclusions

SEM observations and EPMA analysis clearlyshow that a different distribution and morphology

of the zeolite material can be obtained depending

on the preparation method used. This is the basis

for the classification of zeolite membranes pro-

posed in this study, as type-A, type-B and type-C

membranes, depending on whether the zeolite

material is mainly inside the support pores, or as a

film (non-oriented/oriented) on top of it, respec-tively. The study of the adsorption and separation

behavior of these membranes as a function of

temperature has shown significant differences be-tween type-A, type-B and type-C MFI membranes

that are summarized in Table 3. While a definitive

explanation of this different behavior cannot be

proposed with the evidence available at this stage,

it seems reasonable to assume that the distribution

of zeolitic material in type-A (and to a lesser extent

in type-C membranes), with the creation of inter-

crystalline micropores plays a major role for theadsorption differences observed, and the better

performance of type-A membranes at high tem-

peratures.

Acknowledgements

Financial support from DGICYT (QUI97-1085)and DGA (P041/2000) is gratefully acknowledged.

References

[1] R.M. Barrer, Hydrothermal Chemistry of Zeolites, Aca-

demic Press, London, 1982, 360 pp.

[2] D.W. Breck, Zeolite Molecular Sieves: Structure, Chemis-

try and Use, Wiley, New York, 1974, 771 pp.

[3] J. Caro, M. Noack, P. K€oolsch, R. Schafer, Micropor.

Mesopor. Mater. 38 (2000) 3.

[4] A. Tavolaro, E. Drioli, Adv. Mater. 11 (1999) 975.

[5] H.H. Funke, M.G. Kovalchick, J.L. Falconer, R.D. Noble,

Ind. Eng. Chem. Res. 35 (1996) 1575.

[6] H. Kita, T. Inoue, H. Asamura, K. Tanaka, K. Okamoto,

Chem. Commun. (1997) 45.

[7] K. Kusakabe, T. Kuroda, S. Morooka, J. Membrane Sci.

148 (1998) 13.

[8] G. Xomeritakis, M. Tsapatsis, Chem. Mater. 11 (1999)

875.

Table 3

Summary of the different properties of type-A, type-B and type-C MFI membranes

Membrane property Type

A B C

Zeolitic material Mainly inside the

support pores

Mainly as a layer

on top of the support

Oriented layer

on top of the support

Temperature of the maximum

n/i-butane separation selectivity

>125 �C ca. room temperature ca. 50 �C

Molar ratio of n/i-butane adsorbed

on the whole membrane

1.64–3.43 1.20–1.49 1.58–1.71

Temperature of maximum during

desorption of n-butane

ca. 80 �C ca. 65 �C ca. 65 �C


[9] J. Coronas, J.L. Falconer, R.D. Noble, Ind. Eng. Chem.

Res. 37 (1998) 166.

[10] M.P. Bernal, J. Coronas, M. Men�eendez, J. Santamar�ııa,

Chem. Eng. Sci. 57 (2002) 1557.

[11] J.D.F. Ramsay, S. Kallus, in: N.K. Kanellopoulos (Ed.),

Recent Advances in Gas Separation by Microporous

Ceramic Membranes, Elsevier Science BV, Amsterdam,

2000, p. 373.

[12] J. Coronas, J.L. Falconer, R.D. Noble, AIChE J. 43 (1997)

1797.

[13] E. Piera, A. Giroir-Fendler, J.A. Dalmon, H. Moueddeb,

J. Coronas, M. Men�eendez, J. Santamar�ııa, J. Membrane

Sci. 142 (1998) 97.

[14] G. Xomeritakis, S. Nair, M. Tsapatsis, Micropor. Meso-

por. Mater. 38 (2000) 61.

[15] M.P. Bernal, G. Xomeritakis, M. Tsapatsis, Catal. Today

67 (2001) 221.

[16] J. Hedlund, J. Sterte, M. Anthonis, A.J. Bons, B. Carsten-

sen, N. Corcoran, D. Cox, H. Deckman, W. De Gijnst, P.P.

de Moor, F. Lai, J. McHenry, W. Mortier, J. Reinoso, J.

Peters, Micropor. Mesopor. Mater. 52 (2002) 179.

[17] J. Coronas, J. Santamar�ııa, Separ. Purif. Method. 28 (1999)

127.

[18] L.T.Y. Au, W.Y. Mui, P.S. Lau, C. T�eellez-Ariso, K.L.

Yeung, Micropor. Mesopor. Mater. 47 (2001) 203.

[19] V.A. Tuan, J.L. Falconer, R.D. Noble, Ind. Eng. Chem.

Res. 38 (1999) 3635.

[20] V.A. Tuan, J.L. Falconer, R.D. Noble, AIChE J. 46 (2000)

1201.

[21] M.D. Jia, K.V. Peinemann, R.D. Behling, J. Membrane

Sci. 82 (1993) 15.

[22] M.P. Bernal, J. Coronas, M. Men�eendez, J. Santamar�ııa, J.

Membrane Sci. 195 (2002) 125.

[23] E. Piera, M.A. Salom�oon, J. Coronas, M. Men�eendez, J.

Santamar�ııa, J. Membrane Sci. 149 (1998) 99.

[24] W.J.W. Bakker, F. Kapteijn, J. Poppe, J.A. Moulijn,

AIChE J. 43 (1997) 2203.

[25] N. Nishiyama, K. Ueyama, M. Matsukata, AIChE J. 43

(1997) 2724.

[26] M. Arruebo, Ph.D. Thesis, University of Zaragoza (Spain),

2001.

[27] T.J.H. Vlugt, R. Krishna, B. Smit, J. Phys. Chem. B 103

(1999) 1102.

[28] S. Alfaro, M. Arruebo, J. Coronas, M. Men�eendez,

J. Santamar�ııa, Micropor. Mesopor. Mater. 50 (2002)

195.

[29] E.R. Geus, M.J. den Exter, H. van Bekkum, J. Chem. Soc.

Faraday Trans. 88 (1992) 3101.

[30] E.G. Derouane, J. Catal. 100 (1986) 541.

[31] E.G. Derouane, J. Molec. Catal. 134 (1998) 29.


Documents

On the effect of morphological features on the properties of MFI zeolite membranes