journal of

MEMBRANE
SCIENCE
ELSEVIER

Journal of Membrane Science 9 I,( 1994) 77-86

Gas permeation characteristics of polymer-zeolite

membranes

mixed matrix

Murat G. Siier, Nurcan BaC, Levent Yilmaz*

Department of Chemical Engineering, Middle East Technical University, 06531 Ankara, Turkey

(Received February 15, 1993; accepted in revised form January IO, 1994)

Abstract
Mixed matrix membranes of polyethersulfone
(PES), a glassy polymer, and hydrophilic zeolites 13X and 4A
were prepared by using different membrane preparation procedures. Using selected procedure (c), the permeation
rates of N2, Oz, Ar, CO2 and Hz were measured with a variety of membranes prepared at different zeolite loadings.
Significant differences in measured permeability and calculated selectivity values demonstrated the importance of
the type and percentage of zeolite. For both zeolitic additives, permeabilities and selectivities are enhanced at high
zeolite loadings.
In order to understand the permeation mechanism taking into account the polymer-zeolite
interactions, macropositioning of zeolites and matrix distribution,
the heterogeneous membrane morphologies were investigated by
scanning electron microscopy (SEM ). Significant changes in the membrane morphologies of PE- 13X and PES4A matrices were observed, implying the importance of zeolite type.
Key words: Gas separations;

Mixed matrix membrane;

Polyethersulfone

1. Introduction

The development of membranes for the separation and purification of gases based on the selective permeation of one or more components
of a mixture has attracted a great deal of interest
during the last decade, and modifications in the
physical and chemical structures of polymer films
have been made to achieve better separation
characteristics [ 11. The latest emerging membrane morphology with potential for the future
is a matrix membrane composed of two interpenetrating matrices of different materials [ 2-61.
Tsujita [ 71 investigated the effects of addi*Corresponding

membrane;

Zeolite; Membrane morphology

tives on polymeric membranes. His conclusion

was that the polymer-filler system often has an
elevated glass transition temperature ( Tg ) indicative of the restricted segmental motion of the
polymer itself because of strong polymer-filler
interaction. Therefore the coefficients of permeability and diffusion decrease with filler content.
In the case of weak polymer-filler interaction, the
filler may form a void in the interface between
the polymer and the filler. The permeability
coeffkient then increases considerably, indicating hydrodynamic permeation through the void
or pore in the membrane matrix. These findings
point to the need for the investigation of different polymer-filler combinations in order to cre-

author.

0376-7388/94/%01.00
0 1994 Elsevier Science B.V. All rights reserved
SSDI0376-7388(94)00018-T

et al. /Journal

78

of Membrane Science 91 (I 994) 77-86

ate membranes with widely differing permeabilities and selectivities.

Recently, interest has been focused on novel
polymer-zeolite
mixed matrix membranes, because the interaction of materials in the membrane matrix and the shape-selective
catalytic
properties of zeolites can favor permselective
separations. Kulprathipanja et al. [ 21 found that
membranes composed of cellulose acetate and
silicalite- 1 have improved characteristics as the
silicalite content increases. Te Hennepe [ 3 ] obtained improved results in the separation of various alcohols from water by pervaporation using
silicalite-tilled silicone rubber membranes. Jia et
al. [6] studied the permeation of various gases
composed
of
membrane
using
a
poly (dimethylsiloxane)
(PDMS), which is a
rubbery polymer, and silicalite-1, a hydrophobic
zeolite. In their study only a couple of very high
zeolite loadings were investigated, and they concluded that silicalite played the role of a molecular sieve in the membrane by facilitating the
permeation of smaller molecules but hindering
the permeation of larger ones. Okumus et al. [ 5 ]
studied the separation of water-alcohol
mixtures by pervaporation
using a mixed matrix
membrane composed of cellulose acetate and
zeolite 13X or 3A. Their results showed that the
addition of zeolite to the matrix improved the
fluxes substantially, with some loss in selectivity.
Gurkan et al. [ 4 ] focused on the separation of
02-Nz and HZ/N2 gas pairs using a zeolite 13Xfilled polysulfone membrane made by extrusion.
They found substantial increases in selectivities

Table 1
Permeabilities

of various gases through homogeneous

Gas

Measurement
(C)

co*

35
35
35
25
25
25

He
CH,
02

Ar
N*

temp.

when compared with pure polysulfone. However, glassy polymer-hydrophilic

zeolite pairs
have not been investigated extensively in a systematic manner. As the resultant morphology is
heterogeneous,
the preparation
and pretreatment of membranes may affect the structure and
performance.
In the membrane literature, this
aspect is generally disregarded and very few details of membrane preparation and its effects are
reported.
The high performance
engineering thermoplastic polyethersulfone
(PES) is here selected
as the membrane polymer. The repeat unit structure of polyethersulfone
possesses a certain degree of rigidity. PES has a glass transition temperature of 225 C [ 81 and is thus a glassy
polymer at preparation and application temperatures. These properties have made PES a popular membrane material for various applications
[ 8,9 1. The permeabilities
of various gases measured under different conditions by using dense
homogeneous PES films are summarized in Table 1. In this study, mixed matrix membranes of
PES and zeolite 13X or 4A have been developed
and the permeation rates of N2, 02, Ar, COZ and
Hz have been measured, having regard to the potential effects of zeolite type and loading and
membrane preparation methodology on permeability and separation factors. In order to understand the morphology of heterogeneous
membranes, an investigation of the physical structure
of the developed membranes has been carried out
by scanning electron microscopy (SEM).

PES membranes

Pressure gradient
(atm)

Permeability
(barrer )

Membrane preparation

technique

10
10
10
7
7
7

2.8
8.0
0.1
0.629
0.229
0.0974

Compression molding
Compression molding
Compression molding
Solution casting in DMF
Solution casting in DMF
Solution casting in DMF

Ref.

8
8
8
15
15
15

M.G. Siier et al. /Journal ofMembrane Science 91 (I 994) 77-86

2. Experimental
2.1. Materials

The polymeric material used was the commercial polyethersulfone Victrex 41OOP, which was
kindly supplied by ICI. The solvent was dimethylformamide
(DMF), which was purchased from Merck. As adsorptive fillers, zeolite
13X and zeolite 4A were used; they were supplied by Alfa Products. As the zeolites are highly
hydrophilic, they were dried at 350-400C for
2-3 h before each membrane preparation.

79

measurements. No change in membrane thickness was observed due to gas permeation . Measured thicknesses were in the range 70-80 pm.
The measured thickness for each individual
membrane was used in the permeability
calculations.
2.3. Gas permeability measurements
The permeability coefficients were determined in an apparatus conforming to ASTM
D 1435-82, as shown in Fig. 1.
Measurements were made by a constant vol-

2.2. Membrane preparation methodologies

During this study, three different methods for
membrane preparation were developed. The
membranes were prepared by dissolving PES and
dispersing zeolite 13X or 4A in DMF at the desired percentages with thorough mixing to ensure homogeneity. The mixture was then mechanically cast onto a glass plate as a film 500 pm
thick.
Procedure (a)

The cast film is allowed to dry at room temperature for one day. Further annealing is carried out at 80-100C for 6-8 h.
Procedure (b)

The cast film is allowed to dry at barometric

pressure with nitrogen circulation at 60-80 C for
8-10 h. It is then quenched by water, followed
by annealing in a vacuum at 80-l 00C for 1624 h.
Procedure (c)

The cast film is allowed to dry under a partial

vacuum with nitrogen circulation at 60-80 C for
8-10 h. Subsequent quenching by water is followed by annealing at 80-100C for 16-24 h at
barometric pressure.
All the membranes used in this study, unless
otherwise specified in Table 5, were produced by
procedure (c). The final thicknesses of the dry
membranes were measured with a digital micrometer before and after the gas permeation

A:
0:

c:
D:
E:
F :

G:
Ii:

Membrane
Pressure Transducer
Digital Meter
Gas Chamber
High Pressure Purge
Permeability
Cell
Vacuum Part
Constant Temperature

Fig. 1. Permeability

Both

apparatus.

80

M.G. Siier et al. /Journal of Membrane Science 9 I (1994) 77-86

ume, variable pressure technique in a permeation cell. These high pressure side was at 100 psig
and the permeate side was initially at barometric
pressure. The permeation rates of N2, 02, Ar, COZ
and H2 were evaluated by measuring the increase in pressure on the permeate side using a
pressure transducer (Data Instruments, Model
SA).
The individual gas permeations were evaluated using Eq. ( 1 ), which is a combined form of
Ficks law of diffusion and Henrys law.
Ji = PiAP/G

(1)

where Ji is the flux, Pi is the permeability, AP is

the pressure difference across the membrane and
6 is the membrane thickness. For comparison
purposes, the separation factor (Y is determined
by ratioing the individual permeation rates:
a=P,/Pj

(2)

2.4. Membrane characterization

For the characterization
of membrane morphology, electron micrographs of membranes,
with different zeolite percentages were obtained
by scanning electron microscopy on a Cambridge Stereo-Scan, Model S4-10 and a Leitz
AMR 1000.

3. Results and discussion

Reproducibility could be a serious problem in
preparation and testing of membranes with complex heterogeneous structures [ 10 1. A series of
reproducibility
experiments was therefore carried out in order to check the gas permeation
setup, the methodology and the membrane preparation techniques. The permeability of each gas
through each membrane was measured three
times, and the observed relative error between
successive runs was ca. 5 l-3%. Further measurements were made with further portions from
the same casting and the observed relative error
was ca. 52-59/o. Finally, the permeabilities
of
membranes with the same zeolite composition
but cast at different times were measured, and

the observed relative error was ca. + 3-6%. These

results showed that our testing system and methodology were reliable, and that the membrane
preparation system can reproducibly yield heterogeneous mixed matrix membranes with the
same performance.
It is believed that this reproducibility is achieved by using long preparation and annealing times to produce a stable
membrane structure.
The permeabilities of gases through PES-zeolite 13X and PES-zeolite
4A mixed matrix
membranes prepared by procedure (c ) are given
in Table 2. The maximum zeolite loading at
which a workable membrane could be produced
is 50%.
The effects of zeolite type and percentage on
the permeabilities of different gases can be evaluated with the help of Table 2 and Figs. 2-5.
From the table and figures it can be observed that
the permeabilities of N2, 02, Ar, CO2 and Hz decreased up to a loading of 8 wt% for zeolite 13X
and 25 wt% for zeolite 4A. Above these loadings,
a recovery in permeabilities started, favoring increasing permselectivity.
Although the magnitude of the permeability
Table 2
Permeabilities of gases through PES and zeolite 13X or 4Afilled PES membranes at T= 25C, AP= 100 psig
Zeolite type Zeolite content
(wt%)

N,

0,

Ar

None

0.0

0.14

0.52

0.18

2.6

6.5

13x

4.0
8.3
16.6
33.3
42.0
50.0

NM
0.077
0.088
0.097
0.12
0.12

0.38
0.29
0.33
0.37
0.51
0.50

NM
0.11
0.13
0.15
0.19
0.20

2.4
1.5
1.8
2.7
4.2
5.2

4.9
3.8
4.7
6.0
7.5
8.5

4A

8.3
16.6
25.0
33.3
42.0
50.0

0.12
0.12
NM
0.097
0.17
0.25

0.47
0.47
NM
0.41
0.74
1.10

0.17
0.14
NM
0.12
0.26
0.35

2.3
2.3
1.6
2.0
5.8
10.7

6.0
5.8
4.5
4.8
8.9
14.1

CQ

Hz

Permeabilities are given in terms of barrer: 1 barrer = 1O- cm3

(STP) cm/cm2 cmHg s.
NM = not measured.

81

h4.G. Siier et al. /Journal of Membrane Science 91(1994) 77-86

0.6
L-

IO

30

20

40

F/, ZEOLITE

50

60

LOADING

Fig. 2. Effect of zeolite loading on permeabilities

and Ar for PES- 13X mixed matrix membranes.

20

o/uZEOLITE

of N2, O2

50

611

LOADING

Fig. 4. Effect of zeolite loading on permeabilities

and Ar for PES-4A mixed matrix membranes.

of NZ, 02,

1
CI

10

10

20

% ZEOLlTE

40

JO

60

04

LOADING

Fig. 3. Effect of zeolite loading on permeabilities

HZ for PES-13X mixed matrix membranes.

of CO2 and

and its rate of change with zeolite loading were

different for different gases, the general trend of
permeability as a function of zeolite loading was
the same for all the studied gases, suggesting a
similar permeation mechanism for each. An important observation
is that, at the maximum
zeolite loading, gas permeabilities are either not
altered or, in most cases, significantly improved.
This effect is especially noticeable when zeolite
4A is used.
The ideal separation factors of PES- 13X and
PES-4A mixed matrix membranes for some gas
pairs are tabulated in Table 3. Table 3 indicates
that PES-13X membranes possess better separation characteristics than PES-4A membranes.

10

30

R ZEOLITE

40

50

LOADING

Fig. 5. Effect of zeolite loading on permeabilities

H2 for PES-4A mixed matrix membranes.

of CO2 and

On the other hand, gases permeate more readily

through PES-4A membranes. The selectivities
for the economically important gas pairs Hz/N*,
C02/N2, and C02/02 [ 111 are substantially increased, owing to the faster permeation of H2 and
COZ through the membrane matrix (Figs. 6 and
7 ) . However, for the pairs 0z/N2 and Ar/N2 the
increase is only slight.
Table 2 and Figs. 6 and 7 clearly show that at
high zeolite loadings both gas permeabilities and
selectivities increase, indicating the possible potential of this type of mixed matrix membrane
for commercial applications.
The increase in separation factors can not be

M.G. Siier et al. /Journal ofMembrane Science 91 (I 994) 77-86

82
Table 3
Selectivities
membranes
Zeolite
type

of PES and zeolite

Zeolite content
(wt%)

None

13X or 4A-filled

PES

X/N,
O2

Ar

COz

Hz

3.71

1.29

18.8

46.4

5.6

CWO2

13x

8.3
16.6
33.3
42.0
50.0

3.70
3.75
3.81
4.29
4.18

1.43
1.47
1.54
1.61
1.68

19.5
20.5
21.9
35.7
43.1

49.4
53.4
61.3
63.1
70.8

5.26
5.45
7.32
8.30
10.4

4A

8.3
16.6
33.3
42.0
50.0

3.92
3.92
4.22
4.35
4.40

1.40
1.16
1.23
1.50
1.40

19.2
19.2
20.2
34.4
42.8

48.6
48.6
49.5
52.4
56.4

4.89
4.89
4.78
7.89
9.73

Table 4
Kinetic diameter of experimental

Kinetic diameter
(A)

lites varies for the different gases, and this variation may contribute
to the increase
in
permeabilities
and selectivities. This chemical
interaction effect plays an important role especially in CO2 permeability and selectivity at high
zeolite loadings, because specific interactions
with highly charged sites are most significant for
gases (i.e. CO*) that have the ability to interact
with the polar surfaces of zeolites A and X during permeation. The adsorption affinity of the
studied gases on zeolite 4A is in the following sequence [ 131:
COz>N2>CH4>02>Ar>H,
For an uptake of one CO1 molecule per cavity
the affinity sequence of zeolites is [ 14 ]
A>X>Y
3. I. Comparison of membrane preparation
procedures

gases

Nz

Ar

O2

CO2

H2

3.64

3.4

3.46

3.3

2.89

solely explained by using the molecular sieving

effect of zeolites, because all the gas molecules
are able to pass through the windows of zeolites 13X and 4A [ 121. On the other hand, shape
selectivity may play a role if there is a homogeneous distribution in the matrix. The kinetic diameters of the gas molecules are given in Table 4

[121.
The tables indicate that, although the argon
molecule has a smaller kinetic diameter than oxygen, it has a lower permeability.
Indeed, the
permeability should have been greater if diffusional aspects are assumed to be significant. The
reason for the result may be the non-spherical
shape of the oxygen molecule. On the other hand,
marked increases of CO2 permeability at high
zeolite loading may be attributed to the polarity
and adsorption of gases within the membrane
matrix in addition to the diffusional aspects
which imply faster permeation of smaller molecules (i.e., HZ). The chemical interaction of zeo-

Membranes prepared by different procedures

are compared by measuring the permeation rates
of gases through each at a specified zeolite loading. Membranes prepared by procedure (a) have
a filter paper-like structure which is a highly
permeable, non-selective film that allows passage of all gases. In terms of physical appearance,
membranes prepared by procedure (b ) are much
more brittle than those prepared by procedure
(c), so higher zeolite loadings are not possible
Table 5
Comparison

of membrane preparation

Method

Membrane

PES
PES

Fr
C

:
C

Zeolite
(wt%)
0
0

procedures

N2

O2

CO2

H2

0.14
0.14

0.50
0.52

2.6
2.6

6.5
6.5

PES-13X
PES-13X
PES-13X

33.3
33.3
33.3

AP
AP
0.076 0.31
0.097 0.37

AP
1.0
2.7

AP
2.3
6.0

PES-4A
PES-4A
PES-4A

33.3
33.3
33.3

AP
AP
0.11 0.38
0.097 0.41

AP
1.8
2.0

AP
3.5
4.8

AP: all pass.

83

M.G. Siier et al. /Journal ofMembrane Science 91(1994) 77-86

with membranes prepared by procedure (b).

The results tabulated in Table 5 demonstrate
that, for mixed matrix membranes, the membrane preparation procedure strongly affects the
permeabilities of the gases. As the permeability
values for homogeneous membranes prepared by
two different procedures were almost the same,
it can be concluded that the change in the procedure mainly affects the polymer-zeolite
interactions.
In our view, it can be claimed that the major
difference between procedures (b) and (c) is the
drying rate, and one can speculate on the effect
of difference in drying rate as follows. In procedure (b ), initially a slow release of solvent takes
place for the distribution of the matrix, and then
rapid drying begins in the vacuum, which may
inhibit the relaxation of intersegmental packing
and cause substantial increases in the density of
the matrix. This densified matrix may then
show resistance to the permeation of gases and
may have lost the homogeneous distribution of
zeolite within the matrix. In contrast, in procedure (c) a quick release of solvent is followed by
slow drying, which may allow relaxation of the
packing and enhance homogeneity. The matrix
is thus less dense and permeation is much higher,
with an increase in separation factors.
As shown in Table 5 permeability values for
homogeneous membranes prepared by procedures (b) and (c) were almost the same, implying that both procedures produces stable and satisfactory structures. It can therefore be concluded
that the change in procedure mainly affected the
polymer-zeolite interactions and the resultant
micromorphology
of
the
heterogeneous
membranes.
3.2. General character of membranes
The SEM images of PES-13X and PES-4A
membranes are shown in Figs. 8- 11.. These micrographs indicate a homogeneous distribution
of zeolites in the matrix, where zeolite particles
create cave-like porous structures into which they
fit. The cave-like porous structure is believed to
arise from the partial incompatibility of polymer
chains and zeolite crystals, implying that the

51 -

I l&N21

p5

HUN2

60

15

10

20
30
40
96 ZEOLITE LOADING

50

Fig. 6. Effect of zeolite loading on selectivities of C02/N2

and HZ/N2 for PES- 13X mixed matrix membranes.

CI

15
0

IO

20
30
40
cl0ZEOLITE LOADING

50

60

Fig. 7. Effect of zeolite loading on selectivities of C02/N2

and Hz/N2 for PES-4A mixed matrix membranes.

polymer interacts weakly with the zeolitic framework. As the filler content increases the void
spaces that are formed around the zeolite crystals may combine to give a channel network, so
increasing the permeation rates of gases (Fig. 9 ) .
On the other hand the increase in the free volume at the zeolite locations with the increase in
tiller content may cause an increase in the packing density of the unoccupied zones, which may
restrict the diffusion of gases. Therefore, for the
control of separation factors, it is difficult to
consider a mechanism that is affected by a molecular sieving mechanism alone.
At low zeolite loadings, the pores created by

84

M.G. Siier et al. /Journal

ofMembrane

Fig. 8. SEM micrograph of PES-4A (50 wt%) mixed matrix

membrane.

Science 91 (I 994) 77-86

Fig. 10. SEM micrograph of PES-4A (33.3 wtOY6)mixed matrix membrane.

Fig. 11. SEM micrograph

matrix membrane.

Fig. 9. SEM micrograph

trix membrane.

of PES-13X

(50 wt%) mixed ma-

of PES-13X

(33.3 wt%) mixed

zeolites cannot form a continuous channel network and gas molecules must cross polymer-void
interfaces alternately, hindering gas permeation.
Therefore the permeabilities
of gases in both

M.G. Siier et al. /Journal of Membrane Science 91(1994) 77-86

PES-13X
and PES-4A
membranes
are
decreased.
As the percentage of zeolite in the matrix increases, the channel network may mature and
connect the separate voids that provide an alternate path for gas molecules. This may lead to increases in the permeation rates of all gases. In addition to diffusional aspects, which imply faster
permeation of smaller molecules such as HZ, polar interaction of gases with zeolites may enhance the passage rates of polar gases such as
COZ. Moreover, shape selectivity is expected to
play a role by hindering the diffusion of large
molecules, causing a further increase in the separation factor.
The major difference between PES- 13X and
PES-4A mixed matrix membranes is believed to
be the macropositioning of zeolites in the matrix. Zeolite 13X crystals seemed to be more discrete (Fig. 9), whereas zeolite 4A crystals are
partly aggregated, forming wider cavities. This
could account for the different minima and recoveries in permeabilities in addition to the
chemical interaction of gases with the zeolites
[ 141. The agglomeration of 4A crystals in the
membrane matrix increases the free volume at
the zeolitic locations more markedly than is the
case for 13X, resulting in a larger permeability
increase. However, the separation factors are
much higher in PES- 13X membranes because
the positioning of zeolites is relatively separate
with smaller cavities. This enhances the role of
interaction of gas molecules with individual zeolite crystals.

85

( 3 ) The type of zeolite in the matrix is very

important. For PES-zeolite 13X membranes,
permeabilities show a recovery above 8 wt% zeolite loading. However, for PES-zeolite 4A membranes, permeabilities show a recovery only
above 25 wt% zeolite loading. The reason for this
may be the different chemical interactions of polar gases with the different zeolites and the macropositioning of zeolites in the matrix, which
imply the agglomeration of 4A crystals, whereas
13X crystals remain as discrete entities.
(4) The addition of zeolite induces a microporous cavity and channeling system, demonstrating the polymer-zeolite interactions and
incompatibility.
(5 ) The increasing selectivities with increasing filler content cannot be due solely to a molecular sieving mechanism, considering the kinetic
diameters of the gases. The shape-selective properties of zeolites, the polarity of gases and the microstructure of the membranes may play a role
in increasing permselectivities.
( 6 ) Changes in membrane performance are
not only due to zeolite crystals, but also depend
on the complex heterogeneous micromorphology, including the lack or presence of zeolite particle aggregation, and the cave-like voids created
around the zeolitic tiller. It should be noted that
this micromorphology and the resultant membrane performance depend strongly on zeolite
type and amount.

Acknowledgments

4. Conclusions
( 1) Permeabilities first decrease and then increase with increasing zeolite loading. At high
zeolite loadings (42-50%) both permeabilities
and selectivities increase, indicating the potential of PES-13X and PES4A membranes for
commercial applications.
(2) It was observed that, for heterogeneous
membranes, the membrane preparation procedure strongly affects the performance of the
membrane.

The authors acknowledge financial support

from Turkish Scientific and Research Council
(TUBITAK) through grant number TBAG 1166
and from METU Research Fund, project number AFP 92-03-04-o 1.

References
[ 1 ] R.W. Spillman and M.B. Sherwin, Gas separation membranes: the first decade, Chem. Technol., 15 (1990) 378.

86

M.G.

Siier et al. /Journal

ofMembrane

[2]S. Kulprathipanja, R.W. Nousil and N.N. Li, Separation of fluids by means of mixed matrix membranes in
gas permeation, US Pat. 4,740,219 (1988).
[3]H.J.C. te Hennepe, Zeolite Filled Polymeric Membranes: A New Concept in Separation Science, Thesis,
University of Twente, 1988.
[4]T. Gurkan, N. Bat, G. Kiran and T. Gur, A new composite membrane for selective transport of gases, Proc.
6th Int. Symp. Synthetic Membranes in Science and Industry, Tiibingen, Germany, August 1989.
[ 5 ] E. Okumus, T. Gurkan and L.Yilmaz, Development of
a zeolite-polymer
membrane for pervaporation, paper
presented at Symp. Chromatographic
and Membrane
Processes in Biotechnology, Azores, Portugal, July 1990.
[6]M. Jia, K.V. Peinemann and R.D. Behling, Molecular
sieving effect of zeolite filled silicone rubber membranes, J. Membrane Sci., 57 ( 1991) 289.
[ 7]Y. Tsujita, Physical Chemistry of membranes, in Y. Ostada and T. Nakagawa (Eds. ), Membrane Science and
Technology, Marcel Dekker, New York, 1992, p. 3
[ 81 J.S. Chiou, Y. Maeda and D.R. Paul, Gas permeation in
polyethersulfone, J. App. Polym. Sci., 33 (1987) 1823.

Science

91 (1994) 77-86

[ 91 I. Pinnau, J. Wind and K.V. Peinemann, Ultrathin multicomponent polyethersulfone membranes for gas separation made by dry/wet phase inversion, Ind. Eng.
Chem. Res., 29 (1990) 2028.
[ 10 ] R.E. Kesting, Synthetic Polymeric Membranes: a Structural Perspective, 2nd ed., Wiley, New York, 1985.
[ 111 R.R. Zolandz and G.K. Fleming, in: W.S.W. Ho and
K.K. Sirkar (Eds.), Membrane Handbook, Van Nostrand Reinhold, New York, 1992, pp. 17-95.
[ 12 ] D.W. Breck, Zeolite Molecular Sieves, Wiley, New York,
1974.
[ 13lR.J. Harper, G.R. Stifel and R.B. Anderson, Adsorption
of gases on 4A synthetic zeolite, Can. J. Chem., 47
(1969) 4661.
[ 14 ] B. Coughlan and S. Killmartin, Zeolites X, Y and A enriched with trivalent cations: sorption of carbon dioxide
and ammonia, J. Chem. Sot. Faraday Trans. 1, 7 1
(1975) 1809.
[ 151 K. Haraya and S.-T. Hwang, Permeation of oxygen, argon and nitrogen through polymer membranes, J. Membrane Sci., 71 (1992) 13.