Copyright 2007, International Petroleum Technology Conference

This paper was prepared for presentation at the International Petroleum Technology
Conference held in Dubai, U.A.E., 46 December 2007.

This paper was selected for presentation by an IPTC Programme Committee following review
of information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the International Petroleum Technology Conference
and are subject to correction by the author(s). The material, as presented, does not
necessarily reflect any position of the International Petroleum Technology Conference, its
officers, or members. Papers presented at IPTC are subject to publication review by Sponsor
Society Committees of IPTC. Electronic reproduction, distribution, or storage of any part of this
paper for commercial purposes without the written consent of the International Petroleum
Technology Conference is prohibited. Permission to reproduce in print is restricted to an
abstract of not more than 300 words; illustrations may not be copied. The abstract must
contain conspicuous acknowledgment of where and by whom the paper was presented. Write
Librarian, IPTC, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
Membranes have been of continuing interest to the petroleum
and chemical industries for gas separations. While glassy,
polymeric membranes have provided efficient performance to
date, significant improvements over current membrane
technology will likely require novel materials. This paper will
review the development and status of a new technology based
on composite membranes, termed mixed matrix membranes,
of polymer matrices in which molecular sieves are dispersed
to give enhanced separation of natural gas from its impurities
compared to membranes of the polymer alone.
The technology is particularly of interest to the separation
of natural gas from its impurities, such as CO
2
and H
2
S.
In laboratory testing, the composite membranes, composed
of molecular sieves in commercial membrane polymer
matrices, showed significant improvement in both selectivity
and flux compared to membranes of the polymers alone for
the separation of CO
2
from natural gas. Enhancements were
obtained using both carbon molecular sieves and small pore
zeolites.
While membranes have been of interest due to their
compactness, light weight, and ease of operation, there has not
been widespread application due to low selectivity and flux,
and limited robustness. This has led researchers to study
molecular sieve membranes, including carbon molecular sieve
and zeolitic materials. While these membranes offer very
attractive properties, their cost, difficulty of commercial scale
manufacture, and brittleness remain major challenges. Mixed
matrix membrane technology, which combines the benefits of
molecular sieves with the ease and low cost of processing
polymer membranes, offers a potential solution to these
challenges.

Introduction
With worldwide natural gas production at around 50 trillion
standard cubic feet per year, there is continuing interest in
improved technology for natural gas purification. A large
fraction of natural gas resources is high in impurities (N
2
, H
2
S,
and especially CO
2
), making production difficult or
uneconomic. In the lower 48 United States, for example, 550
trillion SCF cannot be processed due to high CO
2
content [1].
Since much of worldwide gas resources are offshore or in
remote locations, membranes are becoming attractive for gas
separations due to their light weight, compactness, and ease of
operation. This makes membranes potentially more attractive
than much larger and heavier amine units, especially for gas
resources high in CO
2
(>10%), where regeneration of the
spent amine solutions is energy intensive and disposal of spent
amine has environmental issues. Nevertheless, of the $5
billion/year market for natural gas separation equipment,
membrane processing represents only about 1% [2].
Affecting the ability of membranes to compete against
amine treating for natural gas has been low selectivity for
removing the impurity gases and the low flux of current
commercial membranes. Also detrimental to the greater use
of membranes is poor reliability due to fouling and loss of
selectivity (e.g., through plasticization), and inability to
operate effectively at high temperature and high hydrocarbon
partial pressure. The sensitivity of polymer membranes to
impurities and high temperature has necessitated the use of
extensive pretreatment equipment, in part offsetting the weight
and space advantages of compact membranes. While
improvements have been made in polymer membranes, further
advances are needed to successfully commercialize a number
of high impurity offshore and remote natural gas fields.
For membranes, the intrinsic productivity can be measured
by its permeability, where the permeability of a penetrant gas
A can be defined as the flux (n
A
), normalized by the
membrane thickness (L) and the transmembrane partial
pressure (p
A
):

P
A
= n
A
L/ p
A
(1)

The thinner the membrane, the higher the productivity.
For modern polymeric membranes, the thickness of the
separating layer is less than 0.5 micron, and sometimes even
less than 0.1 micron. This thin, dense, nonporous skin layer is
often on a microporous substrate of a material which provides
mechanical strength in sheet or hollow fiber form (50-500

IPTC 11465
Enhancing Natural Gas Purification With Advanced Polymer/Molecular Sieve
Composite Membranes
S.J. Miller and D.Q. Vu, Chevron Engineering Technology Company, and W.J. Koros, School of Chemical Engineering,
Georgia Institute of Technology
2 IPTC 11465
microns in diameter), enabling operation at high
transmembrane operating pressure. Permeability can also be
defined as the product of the average diffusion coefficient, D
A
,
and the solubility coefficient, S
A
, where permeability is
usually given in units of Barrers, where 1 Barrer = 1 x 10
-10

[cm
3
(STP)cm]/(scm
2
cm Hg).

P
A
= D
A
S
A
(2)

The ability of a membrane to separate A and B, or the
permselectivity,
A/B
, is then the ratio of their permeabilities.


A/B
= P
A
/P
B
(3)

For mixed gas feeds, the selectivity can be calculated from


A/B
= (y
A
/y
B
)

/

(x
A
/x
B
)

(4)

where x
A
is the mole fraction of component i on the
upstream side and y
i
is the mole fraction of that component on
the downstream side. For a typical conventional polymeric
membrane used commercially, such as cellulose acetate, the
CO
2
/CH
4
selectivity at 35C is about 12-15.
Despite advances in polymeric membranes, limits exist in
the tradeoff between permeability and selectivity, first
described by Robeson, and generally referred to as the upper
bound (Fig. 1). These limits led to development of molecular
sieve membranes, particularly zeolitic [3] and carbon [4].
Although carbon molecular sieves and zeolites offer very
attractive permeation properties with permeabilities and
selectivities significantly higher than polymeric materials,
processing challenges and high costs hinder their industrial
application.


Fig. 1. Permeation properties of polymeric membrane materials
() in comparison to the superior properties achievable with
various molecular sieves (carbons and zeolites) [10].

For example, it is estimated that a zeolite membrane
module would cost around US $3000/m
2
of active membrane
area compared to US $20 to $200/m
2
for existing gas-
separation polymeric membrane modules [3]. Even though
the processing challenges to produce defect-free membranes
under acceptable processing conditions may be overcome with
modifications to existing technologies, handling issues (e.g.,
inherent brittleness) may limit their cost-effective use as
continuous membranes in module form due to expensive
fabrication and replacement costs.

Mixed Matrix Membranes
Recently, an alternative approach to achieving superior
membrane performance, which utilizes a composite of
molecular sieve particles within a polymer matrix, has been
proposed. These membranes, termed mixed matrix exploit
the best of both materials, i.e., the high selectivity of
molecular sieves coupled with the processability, low
manufacturing cost, and mechanical properties of polymeric
membranes. Ideally, such materials can be packaged as
hollow fiber modules to maximize the surface-to-volume ratio.
The mixed matrix or heterogeneous material concept is not
new. Incorporation of inorganic particles or fillers into
polymeric materials has been examined for many applications.
These applications use such fillers or resins to enhance the
mechanical properties of the original polymer matrix and more
recently to improve the separation properties over those of the
polymer material for liquid separations or pervaporation. For
example, inorganic fillers or fibers, such as metal or silica
particles, have been integrated into polymers to increase
mechanical strength and toughness for industrial materials and
for liquid separation (reverse osmosis, ion exchange, etc.)
membranes. In membrane applications, activated carbon or
other adsorbents have been dispersed within a polymeric
matrix, resulting in enhanced filtering capabilities for
microscopic, submicron particles, such as enzymes or other
macromolecules. Also, more recently, nonporous, fumed
silica particles have been incorporated into glassy polymers
and have provided enhanced membrane permeability and
selectivity for large organic molecules over small permanent
gases (i.e., reverse-selective membranes) [5].
Beginning in the 1980s, polymeric mixed matrix
membranes using zeolites or carbon molecular sieves or
activated carbon have been actively pursued for pervaporation
applications (e.g., alcohol-water separations). Initially, there
was also some limited research focused specifically on using
mixed matrix membranes in gas separations. Paul and Kemp
[6] are cited for discovery of a delayed diffusional time lag
effect for CO
2
and CH
4
when they added zeolite 5A in PDMS.
They observed that there was an immobilizing adsorption of
CO
2
and CH
4
by the zeolite 5A that significantly increased the
time lag, but had minor effects on the steady-state permeation.
Several researchers [7-9] also examined gas separations with
membranes incorporating zeolites or commercial carbon
molecular sieves (CMS) into rubbery polymers (e.g., PDMS,
ethylene-propylene rubber, etc.).
The performance of a mixed-matrix membrane can be
modeled using the following equation:

+
+
=
) P P ( P 2 P
) P P ( 2 P 2 P
P P
d c d c d
d c d c d
c eff

(5)


0.01 1 10
2
10
4
10
6
CO
2
permeability (Barrers)
Robeson 1991
Upper Bound
IPTC 11465 3
In this case, P
eff
is the effective permeability of the
membrane,
d
refers to the volume fraction of dispersed phase,
while the subscripts on the respective permeabilities d and c
refer to the dispersed (sieve) and continuous (polymer) phases.
The ratio of permeabilities [see Eq. (3)] for (P
eff
)
A
/(P
eff
)
B
for
component A vs. B can be increased by using sieves of high
selectivity as the dispersed phase, in this case for separating
CO
2
from CH
4
. By implementing this methodology, the upper
bound can be approached and even exceeded as sieve loading
is increased.



Fig. 2. The 6FDA polyimide 6FDA-6FpDA.

Polymers of most interest for mixed matrix membrane
applications are those of high intrinsic selectivity. These
glassy polymers, primarily polyimides, have a degree of
rigidity, resulting in high selectivity and permeability.
Especially attractive are the 6FDA polyimides, derived from
the polymerization of 4,4-hexafluoroisopropylidene
diphthalic anhydride and a diamine (Fig. 2), in which the
polymer chains have limited rotational ability [10].
This high rigidity also complicates achieving adhesion
between polymer and sieve. Poor adhesion leads to sieve-in-
a-cage membranes (Fig. 3), in which the sieve is bypassed,
resulting in no enhancement in selectivity. In general,
polymers of greatest selectivity are the most rigid, and
consequently present the greatest challenge in getting good
polymer/sieve adhesion. This is especially true with hollow
fiber membranes, where the constraints in spinning and
stresses involved are greater than in casting dense films.
Researchers have explored surface and interfacial
chemistry approaches and annealing techniques to improve
polymer-sieve contact and to achieve adequate sieve
dispersion in mixed matrix membranes for gas separations.
Recent work by Mahajan and Koros [11] identified some key
issues for mixed matrix membrane formation with rigid,
glassy polymers. They emphasize appropriate selection of the
sieve and matrix phases as important considerations to match
the fast-gas permeabilities. Having a highly permeable
polymer matrix phase can result in bypass of the selective
sieve phase, while a low-permeability polymer underutilizes
the sieves and results in poor productivity. So, the polymer
matrix phase must have sufficient permeability for gas
molecules to have continuous pathways through the sieves.
For example, rubbery polymers (e.g., silicone rubber) of high
permeability and poor selectivity, even at high zeolite content,
still have poor selectivity due to extensive bypassing of the
zeolite particles. On the other hand, using glassy polymers of
very low permeability results in improved but still unattractive
permeability in the mixed matrix membrane.
Considerable work has focused on improving the bond
between sieve and polymer, by tailoring the sieve surface.
Treatments with chemical reagents such as silanes or, more
recently, methylating the surface with Grignard reagents
typify such approaches [12]. Fig. 4 shows excellent adhesion
between a polyetherimide polymer (Ultem

) and zeolite A,
after treatment with Grignard reagent. These treatments can
add considerable cost to the membrane, and researchers are
continuing to seek lower cost methods.
Research in mixed matrix membranes have encompassed
three main types of molecular sieves: zeolites, carbon sieves,
and aluminophosphate sieves.

Zeolites.
A number of small pore zeolites have been studied for
CO
2
/CH
4
separation in mixed matrix membranes. While early
studies were on zeolite A, that zeolite has proven less
attractive for gas separations from humid feeds, due to the
tendency to selectively absorb water, effectively closing the
pores to other penetrants [13]. Recent studies have focused on
more hydrophobic zeolites, particularly those of high
SiO
2
/Al
2
O
3
molar ratio. Among these are the CHA structure
(3.8 pore) having a SiO
2
/Al
2
O
3
ratio of about 30 or greater.
These zeolites can be synthesized with crystallite sizes below
0.5 micron, preferred for membranes with a separating layer of
less than one micron. Fig. 5 shows a high silica CHA zeolite
in an Ultem

matrix. Table 1 shows permeation results with a

90/10 CH
4
/CO
2
feed at 35 C and 50 psia for Ultem

films
and films of the same polymer containing 15% CHA. This
data shows a significant enhancement in selectivity and
permeability for separating CO
2
.



Fig. 3. SEM of zeolite A in a matrix of the polyimide Matrimid

[11].

N N
O
O O
O
* *
n
CF
3
F
CF
3
F
CF
3
F
3
C
4 IPTC 11465


Fig. 4. Grignard-treated zeolite A in an Ultem

matrix.



Fig. 5. CHA zeolite-Ultem

mixed matrix membrane [12].

Table 1. Permeation results for CO
2
separation with
mixed matrix membrane using Ultem

polymer.
1 Barrer = 10
-10
cm
3
(STP) cm/(cm
2
sec cm Hg)

Sample P
CO2, Barrer

CO2/CH4

Ultem

1.5 39
Ultem

/ 15% CHA 2.8 51

Carbon Molecular Sieves (CMS).
High-temperature pyrolysis of polyimide films and particles
has been found to lead to carbon molecular sieves of high
permeability and selectivity for separating CO
2
from methane.
With the polyimide Matrimid

pyrolyzed to 800C for 2 hours,

the resulting CMS had a selectivity for CO
2
/CH
4
of 200 [14].
These films and particles could then be ground to particles
suitable for incorporation into mixed matrix membranes.
Following priming the sieves with a small amount of polymer
to aid in minimizing sieve aggregation and enhance the
compatibility of the sieve surface and the polymer, mixed
matrix membranes could then be formed which showed good
polymer-sieve contact by scanning electron microscopy with
both the polyetherimide Ultem

and with Matrimid

. Results
for dense (non-asymmetric) mixed matrix films containing up
to about 35 wt % CMS are shown in Fig. 6, showing enhanced
CO
2
/CH
4
selectivity and permeability, with selectivities at
35C of over 50 at the highest loading [15]. In these studies,
enhancements by as much as 45% in CO
2
/CH
4
selectivity and
200% in CO
2
permeability over the corresponding intrinsic
permeation properties of the pure polymer matrix phases were
observed. These performance enhancements increased
incrementally as the loading of CMS particles (up to 35% by
weight) being dispersed within the two polymer matrices
(Matrimid

5218 and Ultem

1000) increased. The

experimental selectivity enhancements shown in Fig. 6
compare reasonably well with model predictions, such as the
simple mixed matrix model proposed in Equation (5) [16].
That silanization or other chemical treatment of the sieve
was not required with these polymers to obtain good adhesion
is particularly advantageous for carbon sieves. The CMS
particles appear to have better affinity to glassy polymers,
achieving good adhesion and polymer-sieve contact with
minimal preparation and casting modifications, as shown in
Fig. 7 with fine CMS particles (mostly <1 m) and good
distribution. Another advantage of CMS particles is that the
permeation properties (permeability and selectivity) of the
CMS particles are tailorable by modifying the pyrolysis
protocol [14]. Therefore, such adaptability may be
advantageous when other gas separations are considered.
Continuing work is evaluating other higher-performance
polymers as the continuous matrix phase



Fig. 6. Permeation results for CO
2
/CH
4
separation with Ultem

and
Matrimid

membranes containing different loadings of CMS [15].

IPTC 11465 5



Fig. 7. Scanning electron micrograph (cross-section) of
successful mixed matrix film containing carbon molecular sieve
(CMS) inserts. Continuous polymer matrix is Matrimid

5218
with CMS loading of 19 vol %. Good polymer-sieve contact is
observed [15].

Aluminophosphate Molecular Sieves.
Hydrophobic aluminophosphate sieves, like carbon
molecular sieves, can be used to form mixed matrix
membranes with minimal surface treatment required except
for the most rigid polymers. Among those sieves proven
effective for CO
2
/CH
4
separation are CHA and ERI (5.1A X
3.6 ). The latter has been shown to be especially effective,
resulting in a CO
2
/CH
4
selectivity at 35C of 63 for a
membrane of 18 wt % sieve in an Ultem

matrix [17].

Improving Membrane Durability
While mixed matrix membranes have demonstrated improved
performance in selectivity and productivity, the sensitivity of
the polymer matrix phase to impurities and harsh conditions
continues to be a drawback. These impurities, such as C
4
+
hydrocarbons, as well as a high partial pressure of CO
2
, due to
its enhanced solubility in glassy polymers, can lead to
swelling (plasticization) of the polymer, and increased
permeation of all gas components, resulting in a loss of
selectivity.
This has led to considerable research in polyimides
containing crosslinkable groups within the chains which, upon
crosslinking, have shown significantly more resilience to
degradation by impurities or high CO
2
pressure. Such
crosslinkable groups include carboxylic acids which can then
form chemical crosslinked bonds with multifunctional
alcohols [18]. One particularly promising polymer is that
derived from 6FDA and 2,4,6-trimethyl-1,3-phenylenediamine
(DAM), which also contains 3,5-diaminobenzoic acid
(DABA), where crosslinking can occur via esterification of the
acid group with a diol such as 1,3-propanediol (Fig. 8). These
membranes have shown excellent resistance to plasticization
due to high CO
2
partial pressure [19].

N N
CF
3 CF
3 O
O
O
O
2
N N
CF
3 CF
3 O
O
O
CH
3
CH
3
O C H
3
O O
3
O
N N
CF
3
CF
3
O
O
O
O
2
N N
CF
3
CF
3
O
O
O
CH
3
CH
3
O C H
3
O
3
n

n



Fig. 8. 6FDA-DAM:DABA polymer crosslinked with 1,3-
propanediol.

Table 2 gives permeation results with a 90/10 CH
4
/CO
2

feed at 35C and 65 psia for films of crosslinked 6FDA-
DAM:DABA polymer and the same crosslinked polymer
containing 15% CHA. This data shows a significant
enhancement in selectivity for separating CO
2
.

Table 2. Permeation results for CO
2
separation with
mixed matrix crosslinked polymer.

Sample P
CO2, Barrer

CO2/CH4

Crosslinked 6FDA-DAM:DABA 57.52.9 37.10.7
Crosslinked 6FDA-DAM:DABA / 15% CHA
56.52.8 43.80.7

Conclusions
Mixed matrix membrane technology, which combines the
benefits of molecular sieves with the ease and low cost of
processing polymer membranes, has been shown to be a
potential solution to the challenges of improved selectivity,
flux, and robustness in membrane purification of natural gas.
The incorporation of zeolites, carbon molecular sieves, and
aluminophosphate molecular sieves into the polymer have all
been shown to offer improvements in selectivity and
permeability. Considerable progress has been made in getting
a good bond between the polymer phase and the dispersed
phase. Mixed matrix membranes under development with
new crosslinked polymers are expected to introduce a new
generation of highly selective and robust membranes for
purification of natural gas.

References

[1] H.S. Meyer, Volume and Distribution of Subquality Natural Gas
in the United States; Gas Research Institute: Chicago, IL
(2000).
[2] R.W. Baker, Ind. Eng. Chem. Res., 41 (2002) 1393.
[3] J. Caro, M. Noack, P. Kolsch, and R. Schaefer, Micropor.
Mesopor. Mater. 38 (2000) 3.
[4] D.Q. Vu, W.J. Koros, and S.J. Miller, Ind. Eng. Chem. Res. 41
(2002) 367.
[5] T.C. Merkel, B.D. Freeman, R.J. Spontak, Z. He, I. Pinnau, P.
Meakin and A.J. Hill, Science 296 (2002) 519.
[6] D.R. Paul and D.R. Kemp, J. Polym. Sci. 41 (1973) 79.
[7] M. Jia, K.V. Peinemann, and R. D. Behling, J. Membr. Sci. 57
(1991) 289.
[8] J.M. Duval, B. Folkers, M.H.V. Mulder, G. Desgrandchamps,
and C.A. Smolders, J. Membr. Sci. 80 (1993) 189.
12 m
6 IPTC 11465
[9] S.B. Tantekin-Ersolmaz, L. Senorkyan, N. Kalaonra, M. Tatlier
and A. Erdem-Senatalar, J. Membr. Sci. 189 (2001) 59.
[10] D.Q. Vu, Ph.D. Dissertation, University of Texas at Austin,
Austin, TX (2001).
[11] R. Mahajan and W.J. Koros, Ind. Eng. Chem. Res., 39 (2000)
2692.
[12] S. Husain and W.J. Koros, J. Memb. Sci., 288 (2006) 195.
[13] T.T. Moore, T.Vo, R. Mahajan, S. Kulkarni, D. Hasse, and W.J.
Koros, J. Appl. Poly. Sci., 90 (2003) 1574.
[14] K. Steel, Ph.D. Dissertation, University of Texas at Austin,
Austin, TX (2000).
[15] D.Q. Vu, W.J. Koros, and S.J. Miller, J. Memb. Sci., 211 (2003)
311.
[16] D.Q. Vu, W.J. Koros, and S.J. Miller, J. Membr. Sci., 211
(2003) 335.
[17] S.J. Miller, A. Kuperman, and D.Q. Vu, US Patent No. 7 138
006 (2006).
[18] J.D. Wind, C. Staudt-Bickel, D.R. Paul, and W.J. Koros, Ind.
Eng. Chem. Res., 41 (2002) 6139.
[19] A.M.W. Hillock, Ph.D. Dissertation, Georgia Institute of
Technology, Atlanta, GA (2005).